Journal of Controlled Release 150 (2011) 272278

Contents lists available at ScienceDirect

Journal of Controlled Release

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j c o n r e l

3D systems delivering VEGF to promote angiogenesis for tissue engineering

Anne des Rieux a, Bernard Ucakar a, Billy Paul Kaishusha Mupendwa a, Didier Colau b, Olivier Feron c,
Peter Carmeliet d, Vronique Prat a,
a

Universit catholique de Louvain, Louvain Drug Research Institute, Unit de Pharmacie Galnique, 1200 Brussels, Belgium
Ludwig Institute for Cancer Research, Brussels Branch, 1200 Brussels, Belgium
Universit catholique de Louvain, Institut de recherche exprimentale et clinique, Pole of Pharmacology, 1200 Brussels, Belgium
d
Vesalius Research Center (VRC), K.U. Leuven, VIB, 3000 Leuven, Belgium
b
c

a r t i c l e

i n f o

Article history:
Received 15 July 2010
Accepted 28 November 2010
Available online 3 December 2010
Keywords:
VEGF encapsulation
Hydrogel
Scaffolds
Angiogenesis
Tissue engineering

a b s t r a c t
In most cases, vascularization is the rst requirement to achieve tissue regeneration. The delivery from
implants of angiogenic factors, like VEGF, has been widely investigated for establishing a vascular network
within the developing tissue. In this report, we investigated if encapsulation of VEGF in nanoparticles could
enhance angiogenesis in vivo as compared to free VEGF when incorporated into two different types of 3D
matrices: Matrigel hydrogels and PLGA scaffolds. Negatively charged nanoparticles encapsulating VEGF
were obtained with a high efciency by complex formation with dextran sulfate and coacervation by chitosan.
After 2 weeks, encapsulation reduced VEGF release from hydrogels from 30% to 1% and increased VEGF release
from scaffolds from 20% to 30% in comparison with free VEGF. VEGF encapsulation consistently improved
angiogenesis in vivo with both type of 3D matrices: up to 7.5- to 3.5-times more endothelial and red blood
cells were observed, respectively, into hydrogels and scaffolds. Hence, encapsulation in nanoparticles
enhanced VEGF efciency by protection and controlled release from 3D implants. Encapsulation and
incorporation of VEGF into 3D implants that, in addition to sustaining cell inltration and organization, will
stimulate blood vessel are a promising approach for tissue regeneration engineering.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Regardless of the tissue of interest, implants developed for tissue
engineering must be biocompatible and biodegradable. They should
serve as a three-dimensional template to provide structural support to
the newly formed tissue through an interpenetrating network of
pores (at least 100 m wide) to allow cell migration, tissue in-growth
and vascularization [1,2]. In addition, further enhancing of the
functionality of these matrices by loading them with growth factors,
acting on the surrounding tissues at therapeutic concentrations and
for an adequate period of time, is highly desirable [3]. The delivery
from implants of angiogenic factors, like vascular endothelial growth
factor (VEGF), has been widely investigated to promote new blood
vessel formation [4,5] which is a basic requirement for establishing a
vascular network within the developing tissue [6]. There are different
routes for VEGF administration that can be divided into local and
systemic delivery [7]. Intravenous delivery of VEGF is problematic.
The short circulation half-life, extraneous interactions with multiple
binding sites and VEGF degradation are incompatible with the
sustained local concentrations of VEGF required for the development
Corresponding author. Universit Catholique de Louvain, Louvain Drug Research
Institute, Unit de Pharmacie Galnique, Avenue E. Mounier, 73-20, 1200 Brussels,
Belgium. Tel.: + 32 2 7647320; fax: + 32 2 7647398.
E-mail address: veronique.preat@uclouvain.be (V. Prat).
0168-3659/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jconrel.2010.11.028

of mature blood vessels [8]. Incorporation of VEGF into poly(-lacticco-glycolic acid) (PLGA) scaffolds or into microspheres has shown the
potential to protect and locally deliver VEGF at a more constant rate,
leading to site-specic angiogenesis [8]. Still, VEGF released from the
implant remains susceptible to degradation. Current techniques to
encapsulate and protect VEGF often use harsh organic solvents and
submit proteins to mechanical stress, resulting usually in low protein
loading [9]. Huang et al. [8] developed a method to efciently
encapsulate VEGF based on the VEGF heparin binding domain afnity
for dextran sulfate, and subsequent coacervation by chitosan forming
nanoparticular polyelectrolyte complexes. Complexation and encapsulation preserve VEGF activity, provide a high loading efciency and
allow a controlled release, making this formulation suitable for VEGF
incorporation into implants.
A wide range of three-dimensional bioactive implants has been
developed with potential uses as delivery systems for therapeutic
drugs relevant for tissue repair processes [3], among which two main
types: hydrogels and porous polymeric scaffolds. Ideally, hydrogels
should be injectable in the lesion for a non-destructive delivery,
limiting then further damage. Several macromolecules have been
used to form injectable hydrogels [10], but Matrigel (composed of
structural proteins such as laminin, collagen IV, and heparan sulfate
proteoglycans [11]) has became the method of choice for many
studies involving in vivo testing of angiogenesis [12,13]. Depending
on the type of tissue, the use of polymeric scaffolds may be

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

advantageous to offer a physical support to the regenerating cells.

Previous studies have demonstrated that VEGF delivery from PLGA
scaffolds can achieve local concentrations of VEGF that induced blood
vessel growth and stimulated new tissue formation [1417].
In this report, we investigated if encapsulation of VEGF could
enhance angiogenesis in vivo when incorporated into two different
types of 3D matrices: Matrigel hydrogels and PLGA scaffolds. VEGF
was encapsulated into nanoparticular polyelectrolyte complexes by
rst binding to dextran sulfate and then by coacervation with
chitosan. VEGF incorporation efciency and release from implants
were characterized in vitro. The ability of these implants to induce
angiogenesis in vivo was quantied.

273

Healthcare) with PBS as an elution buffer. Purity and integrity of VEGF

were checked by SDS-Page and concentration was evaluated by UV
spectrophotometry (calculated dimer molar extinction coefcient:
e = 12,115 M1 cm1) and ELISA. The VEGF solution was ltersterilized and its concentration adjusted to 1 mg/ml. The activity of
VEGF was assessed by detection of VEGF receptor 2 (VEGFR2)
phosphorylation of HUVEC cells by Western blot [18]. It was
compared to a commercial VEGF (VEGF164, R&D Systems, produced
in murine myeloma cell line, NS0 derived). The activity of VEGF was
measured by its ability to phosphorylate the VEGF receptor 2 (R2) of
human umbilical vein endothelial cells (pre-screened HUVECs,
ECACC, # S200-05n). The detailed protocol can be found in
Supplementary material 1.

2. Materials and methods

2.1. Materials
Buffered sorbitol-complex medium (BMSY-medium) contained
glycerol as the carbon source and consisted per liter of: 200 mM
phosphate buffer, pH 6.8 and 13.4 g of yeast nitrogen extract (YNB)
(Invitrogen, Carlsbad, USA), 10 g sorbitol (Sigma, Saint Louis, USA),
0.4 mg biotin (Sigma), and 100 mg zeocin (Invivogen, Toulouse, FR).
Buffered methanol-complex medium (BMMY) is similar to BMGY
except that the carbon source is 1% (v/v) methanol (Merck,
Darmstadt, GE) and contains 1% (w/v) casamino acids (Becton
Dickinson, Bedford, USA). Dextran sulfate (500 kDa) was purchased
from Fluka (Buchs, CH) and chitosan (15 kDa) from Polysciences, Inc.
(Warrington, USA). VEGF quantication was performed by ELISA
using monoclonal anti-rat VEGF and biotinylated anti-rat VEGF (R&D
Systems Inc., Minneapolis, USA), as well as Pierce high sensitivity
streptavidinhorseradish peroxidase and TMB substrate kit (ThermoScientic, Rockford, USA). 75/25 poly(lactide-co-glycolide) (PLGA, RG
755 S) was kindly provided by Boehringer Ingelheim (Ingelheim, GE).
Anti phospho-VEGFR2 (TYR1175) was from Cell Signaling (Beverly,
USA). Zinc sulfate, mannitol, Triton X-100, bovine serum albumin
(BSA) and HUVEC (passage 1) were bought from Sigma, used at
passage 6 and cultivated in an endothelial cell growth medium kit
(Sigma). Growth Factor Reduced (GFR)-Matrigel was purchased
from Becton Dickinson.
2.2. Recombinant Rat VEGF164
2.2.1. Production
Pichia pastoris yeasts producing VEGF164 were kindly provided by
Prof. Carmeliet (Vesalius Research Center, KU Leuven, Belgium). Pichia
were inoculated in 50 ml of the BMSY-medium in 250 ml bafed ask,
24 h at 30 C in an orbital shaker. To induce production of VEGF164,
yeasts were centrifuged (1500 g for 10 min) and resuspended at 1 U
OD (600 nm) in 250 ml of the BMMY-medium (2 L bafed asks with
250 ml/ask). Induced cells were grown at 30 C under orbital shaking
(150 rpm). Every 24 h 1% (v/v) methanol was added. After 2 days, the
supernatant of the culture was collected after centrifugation (10,000 g,
20 min) and ltered on 0.45 and 0.2 m lters.
2.2.2. Purication
Recombinant VEGF164 was puried from the cell supernatant by
afnity chromatography, with heparin as ligand, followed by size
exclusion chromatography. Briey, 500 ml of the culture supernatant
was rst concentrated to 10 ml by ultraltration (MWCO 10,000) in
an Amicon chamber and then diluted to 50 ml with sodium phosphate
buffer 50 mM pH 6. The sample was loaded on a 5 ml Hitrap Heparin
HP column (GE Healthcare, Sweden) previously equilibrated with
sodium phosphate buffer 50 mM, NaCl 200 mM, pH 6. The column
was washed with the same buffer adjusted to 500 mM NaCl. VEGF was
eluted with 1 M NaCl. Dimeric VEGF (MW ~ 40,000) was separated
from aggregated forms on a Superdex75 HR10/300 column (GE

2.2.3. VEGF radioactive labeling and biotinylation

[3H] VEGF was prepared by reductive alkylation of amino groups as
previously described [19]. The specic activity of [3H] VEGF was
obtained by measuring the total protein concentration by the
microBCA test (Pierce, Thermo Scientic), according to the manufacturer's instructions, and the radioactivity by liquid scintillation. The
specic activity of [3H] VEGF was 6.13 10e4 Ci/g. [125I] VEGF was
obtained by incubation of VEGF with sodium iodine [125I] carrier free
(1 mCi, Perkin Elmer, Waltham, USA) and Iodination Beads (Pierce,
Thermo Scientic) according to the manufacturer's instructions. [125I]
VEGF specic activity was 11.2 Ci/g. VEGF integrity after labeling
was checked by SDS-Page electrophoresis.
Puried recombinant rat VEGF164 was biotinylated with the EZ-Link
Sulfo-N-hydroxysuccinimide ester of biotin (Thermo Scientic).
Unincorporated biotin was separated from biotinylated VEGF on a
HITRAP desalting column (GE Healthcare) with PBS as the elution buffer.
Biotin incorporation was quantitated by 4-hydroxyazonenzene-2carboxylic acid/avidin assay (Thermo Scientic). A mean ratio of 4.5
biotin per dimer of VEGF was obtained.
2.3. Formation of VEGF nanoparticles
VEGF nanoparticles (VEGF NP) were prepared as described by
Huang et al. [8]. Briey, 80 l of the VEGF solution (1 mg/ml, in PBS)
was added to 800 l of the dextran sulfate (DS) solution (1% w/v, in
water) under stirring (700 rpm) and were stirred for 30 min. Next,
1.6 ml of the chitosan (CS) solution (0.1% w/v, in acetic acid pH 6) was
added dropwise and stirred for 5 min. Finally 80 l of the zinc sulfate
solution (1 M, in water) was added and stirred for 30 min. The
particles were either incubated in 5% mannitol for 30 min, centrifuged, resuspended in 5% mannitol and freeze-dried or centrifuged
and resuspended at the desired concentration in water. The mean size
of nanoparticles with and without VEGF was determined by dynamic
light scattering and the zeta potential by phase analysis light
scattering using a NanoZS (Malvern, UK). The inuence of freezedrying on nanoparticle size and zeta potential was determined.
Incorporation efciency was evaluated by incorporating [3H] VEGF
in the formulation. Incorporation efciency was expressed as a
percentage of the total amount of [3H] VEGF found in the nanoparticles
over the radioactivity in the supernatant and in the nanoparticles.
2.4. Nanoparticle incorporation into 3D implants
2.4.1. Hydrogels
The free or encapsulated VEGF was mixed with 500 l of Growth
Factor Reduced (GFR)-Matrigel and was kept on ice. Matrigel
supplemented with PBS was used as a negative control. Hydrogels
gelied at 37 C forming a solid mass.
2.4.2. Polymeric scaffolds
Polymeric scaffolds were fabricated using a gas foaming/particulate leaching process as previously described [2022]. Briey, free

274

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

VEGF, CSDS or CSDSVEGF nanoparticles were freeze-dried and

mixed, rst with 3.6 mg of microspheres (PLGA 75/25) and then with
100 mg of NaCl (250 m425 m diameter). Scaffolds composed of
PLGA microspheres and salt only were used as a negative control.
Then the mixture was compressed in a 5 mm KBr die at 1500 psi using
a Carver press. The PLGA scaffolds were then equilibrated with high
pressure CO2 gas (800 psi) for 16 h in a custom-made pressure vessel
[23,24]. The pressure was released over a period of 30 min. To create
scaffold macroporosity, scaffolds were leached rst in 4 ml of sterile
water for 2 h and were then transferred in fresh sterile water for
30 min. PLGA scaffolds were then dried and kept under dry and sterile
conditions [4].
The maximal amount of lyophilized nanoparticles that can be
incorporated in the PLGA scaffolds without disrupting their integrity
was evaluated by adding 3 to 20 mg of freeze-dried nanoparticles
prepared without VEGF (CSDS) to the microsphere/salt mixture
(3.6 mg/100 mg). Scaffold integrity was evaluated visually. The
criteria were i) the general shape of the scaffold, ii) it resistance to
manual manipulation (crumbling), iii) the shape of the pores
evaluated by an optical microscope and compared to scaffolds without
nanoparticles, and iv) scaffold weight. The scaffolds were considered
to be non disrupted when no difference was observed with control
scaffolds (without nanoparticles).
The VEGF loss during leaching was evaluated by incorporating
H-VEGF (free or encapsulated) in the scaffolds and measuring
residual radioactivity in the scaffolds as well as in leaching supernatants. The loss during leaching was expressed as a percentage of the
total radioactivity found in the supernatants over the radioactivity in
the supernatants and in the scaffolds.

At designed time-points, implants were retrieved and xed in 4%

paraformaldehyde for 24 h before incubation in a 30% glucose solution
for 4 h and embedding in Tissue-Tek O.C.T. compound (Sakura
Finetek, Torrance, CA). Sections were cut at 12 m thickness for
Matrigel hydrogels and 14 m thickness for PLGA scaffolds using a
cryostat (Leica Microsystems, Wetzlar, GE). An antibody directed
against the murine endothelial cell surface marker (CD 31) was used
to determine the extent of endothelial cell colonization of the
implants. After permeabilization (Triton X-100 0.1% (v/v) in PBS)
and blocking (5% (w/v) BSA in PBS), the primary antibody (rat anti-CD
31 (1:50, BD Biosciences, San Jose, USA)) was applied for 1 h at 37 C.
Secondary antibody (Alexa Fluor 568 goat anti-rat (1:500, Invitrogen))
was used to visualize the antigen. Lastly, sections were incubated with
DAPI (Invitrogen) (50 ng/ml) for 5 min to allow visualization of cell
nuclei.
Image acquisition was performed by a MIRAX microscope (Zeiss,
Thornwood, USA), allowing the acquisition of entire sections, limiting
the information loss and experimenter inuence by eld selection.
In addition, high quality pictures were obtained allowing high
magnication enlargements. For each hydrogel condition, 8 sections
at different levels within hydrogels were imaged. For each PLGA
scaffold condition, 24 sections were imaged: 8 sections taken at one
extremity, 8 in the middle and 8 at the other extremity of scaffolds.
Quantication of CD 31 and red blood cells was then performed on
whole sections using a script of Axiovision (Zeiss). The implant
size was normalized by delimiting the implant borders and by
quantifying the percentage of red pixels over the total amount of
analyzed pixels within these borders. Results were expressed as
percentage of red pixels over the total amount of pixels within the
analyzed surface.

2.5. Release of VEGF from the 3D matrix

Nanoparticles were formed either with [125I] VEGF and incorporated into PLGA scaffolds (1 g VEGF/sample + 1 l of [125I] VEGF
(0.548 Ci)) or with biotinylated VEGF and incorporated into hydrogels (2 g of biotinylated VEGF/500 l hydrogel) (n = 3). Samples
were incubated in PBS at 37 C. After specic time periods, PBS was
removed and replaced by fresh PBS. Radioactivity of the supernatant
that contained the sample for that specic time period was measured
by gamma counter (Perkin Elmer) to quantify the amount of protein
that was released. The cumulative release of VEGF from the samples
was calculated by dividing the cumulative radioactivity released into
the buffer by the total amount of radioactivity released over the
course of the study plus the radioactivity that remained inside the
sample at the end of the study. The latter was analyzed by dissolving
the samples in 5 M NaOH to measure their radioactivity. Biotinylated
VEGF was quantied by ELISA. The cumulative release of VEGF from
the hydrogels was calculated by dividing the cumulative VEGF
released into the buffer by the total amount of VEGF incorporated
inside the hydrogels.
Activity of released VEGF was evaluated by detection of VEGF-R2
phosphorylation as described above.
2.6. Inuence of VEGF encapsulation on angiogenesis
Experimental protocols of animal studies were approved by the
Ethical Committee for Animal Care and Use of the Medicine Faculty of
the Universit Catholique de Louvain.
Inuence of NP of DS and CS alone on angiogenesis has been tested
by Huang et al. [8] with no effect reported on HUVEC proliferation, so
CSDS NP have not been tested here.
1 and 2 g of free or encapsulated VEGF were incorporated in
hydrogels and 1 g in scaffolds. Anesthetized Balb/C female mice
received a subcutaneous injection of 0.5 ml of GFR-Matrigel [12,25].
PLGA scaffolds were implanted into the intraperitoneal fat pad of
NMRI male mice, as previously described [26].

2.7. Statistics
Statistical analyses were done using PRISM. One-way and two-way
ANOVA with post-hoc Bonferroni's multiple comparison test were
performed with a p-level of 0.05. Error bars represent the standard
error of the mean in all gures.

3. Results
3.1. VEGF production
Pure 40 kDa VEGF164 was successfully produced. Its activity was
compared, as control, to a commercial VEGF164 (R&D systems).
Our VEGF induced VEGFR2 phosphorylation of HUVEC cells at the
same concentration than the control (100 ng/ml) (Supplementary
material 2).

3.2. Nanoparticle characterization

To protect and stabilize VEGF, it was complexed by dextran
sulfate (DS) and precipitated by coacervation with chitosan (CS) to
produce particular polyelectrolyte complexes encapsulating VEGF [8].
300 nm negatively charged nanoparticles were obtained (Table 1).
Encapsulation efciency was about 76% and the drug loading about
5 g VEGF/mg of nanoparticles. The size of CSDSVEGF nanoparticles
was slightly higher than CSDS nanoparticles and tends to increase,
but not signicantly, after freeze-drying. No loss of VEGF was detected
after freeze-drying.
The nanoparticle size was stable in PBS for 4 h and then increased
to micrometers, probably due to chitosan insolubility above pH 6.5,
while the nanoparticle size was stable in water for at least 12 h
(data not shown).

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

3.4. Nanoparticle incorporation into hydrogels: effect on angiogenesis

Table 1
Properties of CS/DS/VEGF nanoparticles.
Formulation

Diameter
(nm)

275

PDI

Zeta potential
(mV)

Encapsulation
efciency
(%)

a) Before freeze-drying
CSDSa
325 +/ 19
CSDSVEGFb
383 +/ 73

0.22 +/ 0.03
0.26 +/ 0.06

26 +/ 0.6
26 +/ 2

NA
76.5 +/ 6.5

b) After freeze-drying
319 +/ 17
CSDSa
CSDSVEGFb
457 +/ 24

0.20 +/ 0.01
0.29 +/ 0.01

25 +/ 0.3
34 +/ 1.5

NA
71 +/ 14.4

NA: Not Applicable.

a
n = 4.
b
n = 6.

3.3. VEGF release from 3D implants

VEGF release would likely be inuenced by its incorporation as
well as by the nature of the 3D matrix into which nanoparticles were
incorporated. Release of free or encapsulated VEGF was evaluated
over time after incorporation into hydrogels (Matrigel hydrogels)
and into PLGA scaffolds.
A 10 to 15% burst release was observed for VEGF whatever the
matrices or the formulation (2 h of incubation), except for encapsulated VEGF incorporated in hydrogel (0.3%), followed by a sustained
release over time (Fig. 1). When incorporated in scaffolds, release of
free VEGF was lower than encapsulated VEGF (24 versus 36%,
respectively, after 2 weeks). In hydrogels, encapsulation signicantly
slowed down VEGF release. Indeed, 20-times more free VEGF was
released after 2 weeks than encapsulated VEGF (32% and 1.6% for free
and encapsulated VEGF, respectively). Release of free VEGF from
hydrogels was similar to release of encapsulated VEGF from scaffolds
(2 weeks incubation time, 31% of encapsulated VEGF and 32% of free
VEGF, respectively). Concerning the activity of released VEGF, Huang
et al. [8] demonstrated a preserved bioactivity for released VEGF from
NP for 10 days by a HUVEC proliferation assay. VEGF released from
PLGA bridges [27] and from hydrogels was still able to induce
phosphorylation of HUVEC VEGF-R2 over a period of 42 days [28] and
of 2 weeks (Supplementary material 3), respectively.

Fig. 1. Inuence of VEGF encapsulation on its release from 3D implants. [125I] VEGF and
biotinylated VEGF were incorporated into 3D implants (hydrogels and PLGA scaffolds),
either free or encapsulated. Samples were incubated in PBS at 37 C. Radioactivity or
biotinylated VEGF in the supernatants was measured over time. (n = 3, * signicant,
P b 0.05).

The angiogenesis potential of free VEGF versus encapsulated VEGF

incorporated into an injectable hydrogel was studied. Matrigel was
chosen as a model for tissue regeneration due to its injectability and
proven ability to support cell and blood vessel growth [12,29,30].
Following injection, Matrigel solidies and permits subsequent
penetration by host cells that induce vascularization. Growth factor
reduced Matrigel was used to avoid interferences with the activity
of the incorporated VEGF. Free or encapsulated VEGF and, as control,
PBS, were mixed with Matrigel before subcutaneous injection.
Inuence of VEGF amount (1 g and 2 g) and of incubation times
(2 and 3 weeks) was evaluated. The effect on angiogenesis was
determined by quantication of endothelial and red blood cells inside
the Matrigel hydrogels after implantation.
Endothelial cells were visualized by immunostaining with an anti-CD
31 antibody and red blood cells were visible by autouoresence. The
percentage of red pixel represents both endothelial and red blood cells.
Entire sections of Matrigel hydrogels were analyzed. The benecial
effect of VEGF encapsulation on the stimulation of angiogenesis
was clearly demonstrated (Fig. 2). Indeed, 2 weeks after implantation,
endothelial cells, some of them organized as blood vessels, were observed within the samples containing the encapsulated VEGF (Fig. 2e
and f). In addition, red blood cells were present within these structures,
indicating their perfusion and their functionality. In contrast, for the
hydrogels loaded with PBS or free VEGF, endothelial cells were detected
only around the implants (Fig. 2a to d).
Quantitative analyses of the hydrogels loaded with VEGF nanoparticles showed that these implants contained 5.8 to 7.4-fold more
endothelial and red blood cells in comparison to the hydrogels loaded
with PBS or free VEGF, respectively (Fig. 3). No signicant differences

Fig. 2. Inuence of VEGF encapsulation on angiogenesis when incorporated into

hydrogels. VEGF (1 g), free (c, d) or encapsulated (e, f) was mixed with Matrigel and
injected s.c. in mice (2 weeks); negative control (a, b) consisted in PBS mixed with
Matrigel. Endothelial cells (red) were detected by immunouorescence (CD 31) and
red blood cells (red, arrow) by autouorescence. Nuclei (blue) were labeled with DAPI.
Pictures were acquired with a MIRAX microscope.

276

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

Fig. 3. Inuence of VEGF encapsulation on angiogenesis when incorporated in

hydrogels: quantication of endothelial and red blood cells. VEGF (1 or 2 g), free or
encapsulated was incorporated in Matrigel and injected s.c. in mice (2 or 3 weeks);
negative control consisted in PBS mixed with Matrigel. Endothelial cells were
detected by immunouorescence (CD 31) and red blood cells by autouorescence.
Pictures were acquired with a MIRAX microscope. 8 sections by condition (n = 3) were
imaged in their entirety and analyzed (Axiovision) (mean +/ SEM, *** P b 0.05).

were observed between 1 and 2 g of loaded VEGF, with implantation

duration of 2 and 3 weeks respectively. The implants loaded with PBS
or free VEGF showed no signicant difference.
3.5. Nanoparticle incorporation into PLGA scaffolds: effect
on angiogenesis
The objective was to evaluate the effect of VEGF encapsulation on
angiogenesis when incorporated into polymeric scaffolds developed
for tissue engineering [4,21,23,31,32]. Variable amounts of VEGF
nanoparticles were incorporated during scaffold processing and the
impact on scaffold integrity was determined. The potential improvement of encapsulated over free VEGF on VEGF loss during scaffold
leaching was estimated. Then, VEGF loaded scaffolds were implanted
in mice and angiogenesis was studied.
Freeze-dried CSDS nanoparticles were mixed with PLGA microspheres and NaCl crystals to form PLGA scaffolds. Physical integrity
was estimated visually. The physical integrity of scaffolds was
preserved up to 12 mg of incorporated nanoparticles per 3.6 mg
PLGA (n = 3). Incorporation of higher amounts induced scaffold
crumbling during the leaching phase. To measure VEGF loss after
scaffold leaching, lyophilized free or encapsulated [H] VEGF, was
used for the scaffold process. VEGF loss during salt leaching was 25%
for free VEGF and 4.4% for encapsulated VEGF (n = 3). Taking these
ratios into account, the amount of VEGF added to the scaffold process
was adapted to obtain 1 g of VEGF per implanted scaffold.
VEGF loaded scaffolds were implanted in the fat pad of mice, and
the inuence on angiogenesis of encapsulated versus free VEGF was
studied 2 weeks after implantation. Empty PLGA scaffolds were used
as control. Sections of the implanted scaffolds were immunostained.
Endothelial and red blood cells were identied within and around the
scaffolds for all the conditions (Fig. 4).
However, as observed for hydrogels, PLGA scaffolds containing
VEGF nanoparticles seemed to stimulate more efciently angiogenesis
(Fig. 4e and f).
In order to determine if VEGF encapsulation signicantly increased
the amount of endothelial and red blood cells within and around the
scaffolds, quantication of the percentage of red pixels over the total
amount of analyzed pixels was performed. Signicantly more
endothelial and red blood cells were detected with VEGF nanoparticles incorporated into PLGA scaffolds (Fig. 5) compared with free
VEGF or control, although the difference was less marked than
observed for the Matrigel hydrogels (3.5-folds). This could be

Fig. 4. Inuence of VEGF encapsulation on angiogenesis when incorporated in scaffolds.

VEGF (1 g), free (c and d) or encapsulated (e and f) was incorporated in PLGA scaffolds
and implanted in mouse fat pads (2 weeks); negative control (a and b) consisted in
empty scaffolds. Endothelial cells (red) were detected by immunouorescence (CD 31)
and red blood cells (red) by autouorescence. Nuclei (blue) were labeled with DAPI.
Pictures were acquired with a MIRAX microscope.

explained by the fact that PLGA scaffolds alone seemed to support cell
inltration and angiogenesis to a higher extent than Matrigel. No
signicant difference was observed between negative control and free
VEGF.
4. Discussion
When designing systems for tissue engineering, one of the rst
requirements is the colonization of the implant by blood vessels,
bringing oxygen and nutriments to support tissue reconstruction.

Fig. 5. Inuence of VEGF encapsulation on angiogenesis when incorporated in scaffolds:

quantication of endothelial and red blood cells. VEGF (1 g), free or encapsulated was
incorporated in PLGA scaffolds and implanted in mouse fat pads (2 weeks); negative
control consisted in empty scaffolds. Endothelial cells were detected by immunouorescence (CD 31) and red blood cells by autouorescence. Pictures were acquired with
a MIRAX microscope. 24 sections by condition (n = 3) were imaged in their entirety and
analyzed (Axiovision): 8 sections taken at one extremity, 8 in the middle and 8 at the
other extremity of scaffolds (mean +/ SEM, *** P b 0.05).

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

The major focus of this work was to determine if incorporation of

VEGFCSDS nanoparticles into two types of 3D implants (PLGA scaffold
or hydrogel) commonly used for tissue engineering (hydrogels and
polymeric scaffolds) improved angiogenesis within the implants, and
thus tissue regeneration. We investigated the effect of VEGF combined
release from nanoparticles and 3D matrices and the inuence of the
matrix nature on angiogenesis. The results indicate that implantation of
3D systems loaded with VEGF nanoparticles stimulated endothelial cell
colonization and blood vessel formation more efciently than matrices
loaded with free VEGF, regardless of the VEGF dose, implantation time
and implant type.
The improvement probably resulted from i) VEGF protection by
complexation of the heparin binding site of VEGF with dextran sulfate
combined with ii) protection and controlled release brought by
complex stabilization by chitosan. Indeed, Huang et al. [8] demonstrated in vitro, by ELISA and HUVEC proliferation test, that VEGF
encapsulation by DS and CS protected most of VEGF activity for at least
10 days and that VEGF was slowly released from nanoparticles.
We observed that encapsulation of VEGF, in comparison with free
VEGF, signicantly decreases burst release (0.3% versus 10%, respectively) from VEGF loaded Matrigel hydrogels and slowed down
diffusion from the implant about 20-fold after 2 weeks. It could be due
to higher size combined with electrostatic interactions due to the
negative surface charge of nanoparticles. Encapsulated VEGF was
released from PLGA scaffolds slightly faster than free VEGF (24%
versus 36%, respectively, after 2 weeks). Encapsulation did not affect
burst release but loss during leaching was taken into account for
scaffold preparation and the initial VEGF loading was adjusted to be
1 g of VEGF post-leaching for both conditions. Indeed, encapsulation
signicantly decreased VEGF loss during leaching (about 21%). In that
perspective, the burst release for free VEGF loaded scaffolds would be
approximately about 15% higher than for VEGF nanoparticle-loaded
scaffolds. Finally, compared to microspheres classically used to
protect and deliver proteins, complexation by DS and coacervation
by CS achieved a higher loading and did not compromise implant
integrity. Indeed, approximately 5 g of VEGF was incorporated in
1 mg of CSDS nanoparticles versus approximately 50 ng per mg of
microspheres [9]. This allows incorporation of a much higher amount
of VEGF into implants compared to microspheres.
Although encapsulated VEGF stimulated angiogenesis to a greater
extent than free VEGF in both implant types, differences between PLGA
scaffolds and Matrigel hydrogels were observed. The percentage of
endothelial and red blood cells quantied in nanoparticle-loaded
scaffolds was 7.5 times higher compared to Matrigel hydrogels, but
the scaffold itself, without VEGF, stimulated angiogenesis at the same
level than hydrogels loaded with encapsulated VEGF. This could be
explained by the fact that these PLGA scaffolds have been optimized for
tissue engineering [4] and thus promote cell inltration and
proliferation. However, encapsulation of VEGF did not affect release
from scaffolds as much as for hydrogels. Hence, the increase in
angiogenesis elicited by encapsulated versus free VEGF loaded into
Matrigel hydrogels was about 5 to 7-fold while it was only about 3.5fold for scaffolds. Considering these results it seems that it could be
interesting to combine the two systems, by lling PLGA scaffolds pores
with hydrogel, both loaded with VEGF nanoparticles.
5. Conclusion
We hypothesized that VEGF activity in vivo would be enhanced by its
encapsulation in CSDS nanoparticles and further incorporation into
implants used in tissue engineering, like injectable hydrogels and
polymeric scaffolds. Angiogenesis, the primary effect of VEGF, was
clearly improved by encapsulation: the extent of implant colonization
by endothelial cells and blood vessels was higher when both types of
implants were loaded with VEGF nanoparticles compared with free
VEGF (non-encapsulated). Nanoparticles composed of chitosan, dextran

277

sulfate and VEGF were obtained and incorporated in implants, without

impairing their structural properties and allowing a higher VEGF loading
compared to VEGF classical techniques of encapsulation (up to 5 g of
VEGF per mg of nanoparticles). VEGF encapsulation guarantees VEGF
efciency by protection and sustained release from 3D implants.
In the scope of the spinal cord regeneration, systems to deliver
active VEGF are attractive, since, in addition to its angiogenic activity,
VEGF has recently been shown to be a neurogenic factor [3335].
Then, encapsulation and incorporation of VEGF into 3D implants that,
in addition to support cell inltration and organization, will stimulate
blood vessel and axon growth, are a promising approach to stimulate
tissue regeneration.
Supplementary materials related to this article can be found online
at doi:10.1016/j.jconrel.2010.11.028.

Acknowledgments
The authors would like to thank Prof. Carmeliet (Vesalius Research
Center, KU Leuven, Belgium) for the gift of VEGF producing Pichia
pastoris yeast. The authors thank Sabine Cordi for her help with MIRAX
acquisition, as well as Christophe Pierreux for the RGB quantication
script. Sbastien Sart is thanked for assistance with VEGF production and
Laurenne Petit for the realization of VEGF activity tests. This work was
supported by grants from the Belgian Fonds National de la Recherche
Scientique (FNRS, Crdits aux chercheurs and FRSM 3.4617.09) and
CTB (Belgium). Anne des Rieux is FNRS Post-Doctoral Researcher.

References
[1] V. Guarino, F. Causa, L. Ambrosio, Bioactive scaffolds for bone and ligament tissue,
Expert Rev. Med. Devices 3 (2007) 405418.
[2] T.M. Freyman, I.V. Yannas, R. Yokoo, L.J. Gibson, Fibroblast contraction of a
collagen-GAG matrix, Biomaterials 21 (2001) 28832891.
[3] V. Mourino, A.R. Boccaccini, Bone tissue engineering therapeutics: controlled drug
delivery in three-dimensional scaffolds, J. R. Soc. Interface 43 (2010) 209227.
[4] C.B. Rives, A. des Rieux, M. Zelivyanskaya, S.R. Stock, W.L. Lowe Jr., L.D. Shea,
Layered PLG scaffolds for in vivo plasmid delivery, Biomaterials 3 (2009) 394401.
[5] A.H. Zisch, M.P. Lutolf, J.A. Hubbell, Biopolymeric delivery matrices for angiogenic
growth factors, Cardiovasc. Pathol. 6 (2003) 295310.
[6] Z.S. Patel, A.G. Mikos, Angiogenesis with biomaterial-based drug- and celldelivery systems, J. Biomater. Sci. Polym. Ed. 6 (2004) 701726.
[7] B.V. Somayaji, U. Jariwala, P. Jayachandran, K. Vidyalakshmi, R.V. Dudhani,
Evaluation of antimicrobial efcacy and release pattern of tetracycline and
metronidazole using a local delivery system, J. Periodontol. 4 (1998) 409413.
[8] M. Huang, S.N. Vitharana, L.J. Peek, T. Coop, C. Berkland, Polyelectrolyte complexes
stabilize and controllably release vascular endothelial growth factor, Biomacromolecules 5 (2007) 16071614.
[9] J.S. Golub, Y.T. Kim, C.L. Duvall, R.V. Bellamkonda, D. Gupta, A.S. Lin, D. Weiss, W. Robert
Taylor, R.E. Guldberg, Sustained VEGF delivery via PLGA nanoparticles promotes
vascular growth, Am. J. Physiol. Heart Circ. Physiol. 6 (2010) H1959H1965.
[10] C. He, S.W. Kim, D.S. Lee, In situ gelling stimuli-sensitive block copolymer
hydrogels for drug delivery, J. Control. Release 3 (2008) 189207.
[11] H.K. Kleinman, M.L. McGarvey, J.R. Hassell, V.L. Star, F.B. Cannon, G.W. Laurie,
G.R. Martin, Basement membrane complexes with biological activity, Biochemistry 2
(1986) 312318.
[12] N. Akhtar, E.B. Dickerson, R. Auerbach, The sponge/Matrigel angiogenesis assay,
Angiogenesis 12 (2002) 7580.
[13] D. Donovan, N.J. Brown, E.T. Bishop, C.E. Lewis, Comparison of three in vitro
human angiogenesis assays with capillaries formed in vivo, Angiogenesis 2
(2001) 113121.
[14] J.M. Kanczler, P.J. Ginty, J.J. Barry, N.M. Clarke, S.M. Howdle, K.M. Shakesheff,
R.O. Oreffo, The effect of mesenchymal populations and vascular endothelial growth
factor delivered from biodegradable polymer scaffolds on bone formation,
Biomaterials 12 (2008) 18921900.
[15] F.G. Rocha, C.A. Sundback, N.J. Krebs, J.K. Leach, D.J. Mooney, S.W. Ashley, J.P.
Vacanti, E.E. Whang, The effect of sustained delivery of vascular endothelial
growth factor on angiogenesis in tissue-engineered intestine, Biomaterials 19
(2008) 28842890.
[16] J.M. Kanczler, J. Barry, P. Ginty, S.M. Howdle, K.M. Shakesheff, R.O. Oreffo,
Supercritical carbon dioxide generated vascular endothelial growth factor
encapsulated poly(DL-lactic acid) scaffolds induce angiogenesis in vitro, Biochem.
Biophys. Res. Commun. 1 (2007) 135141.
[17] J.M. Kanczler, P.J. Ginty, L. White, N.M. Clarke, S.M. Howdle, K.M. Shakesheff, R.O.
Oreffo, The effect of the delivery of vascular endothelial growth factor and bone
morphogenic protein-2 to osteoprogenitor cell populations on bone formation,
Biomaterials 6 (2010) 12421250.

278

A. des Rieux et al. / Journal of Controlled Release 150 (2011) 272278

[18] T.T. Chen, A. Luque, S. Lee, S.M. Anderson, T. Segura, M.L. Iruela-Arispe, Anchorage
of VEGF to the extracellular matrix conveys differential signaling responses to
endothelial cells, J. Cell Biol. 4 (2010) 595609.
[19] G.E. Means, R.E. Feeney, Reductive alkylation of amino groups in proteins,
Biochemistry 6 (1968) 21922201.
[20] J.H. Jang, L.D. Shea, Controllable delivery of non-viral DNA from porous scaffolds,
J. Control. Release 1 (2003) 157168.
[21] L.D. Shea, E. Smiley, J. Bonadio, D.J. Mooney, DNA delivery from polymer matrices
for tissue engineering, Nat. Biotechnol. 6 (1999) 551554.
[22] D.J. Mooney, D.F. Baldwin, N.P. Suh, J.P. Vacanti, R. Langer, Novel approach to
fabricate porous sponges of poly(D,L-lactic-co-glycolic acid) without the use of
organic solvents, Biomaterials 14 (1996) 14171422.
[23] J.H. Jang, C.B. Rives, L.D. Shea, Plasmid delivery in vivo from porous tissueengineering scaffolds: transgene expression and cellular transfection, Mol. Ther. 3
(2005) 475483.
[24] M.O. Aviles, C.H. Lin, M. Zelivyanskaya, J.G. Graham, R.M. Boehler, P.B. Messersmith,
L.D. Shea, The contribution of plasmid design and release to in vivo gene expression
following delivery from cationic polymer modied scaffolds, Biomaterials 6 (2010)
11401147.
[25] P. Sonveaux, A. Brouet, X. Havaux, V. Gregoire, C. Dessy, J.L. Balligand, O. Feron,
Irradiation-induced angiogenesis through the up-regulation of the nitric oxide
pathway: implications for tumor radiotherapy, Cancer Res. 5 (2003) 10121019.
[26] H. Blomeier, X. Zhang, C. Rives, M. Brissova, E. Hughes, M. Baker, A.C. Powers, D.B.
Kaufman, L.D. Shea, W.L. Lowe Jr., Polymer scaffolds as synthetic microenvironments for extrahepatic islet transplantation, Transplantation 4 (2006) 452459.

[27] Y. Yang, L. De Laporte, C.B. Rives, J.H. Jang, W.C. Lin, K.R. Shull, L.D. Shea,
Neurotrophin releasing single and multiple lumen nerve conduits, J. Control.
Release 3 (2005) 433446.
[28] L. De Laporte, A. des Rieux, M.L. Zelivyanska, H.M. Tuinstra, N. De Clerck, A.A.
Postnov, L.D. Shea, VEGF and FGF-2 delivery from spinal cord bridges to enhance
angiogenesis following injury, J. Biomed. Mater. Res. Part A (in press).
[29] M.L. Ponce, Tube formation: an in vitro matrigel angiogenesis assay, Methods Mol.
Biol. (2009) 183188.
[30] A. Albini, R. Benelli, The chemoinvasion assay: a method to assess tumor and
endothelial cell invasion and its modulation, Nat. Protoc. 3 (2007) 504511.
[31] M.H. Sheridan, L.D. Shea, M.C. Peters, D.J. Mooney, Bioabsorbable polymer
scaffolds for tissue engineering capable of sustained growth factor delivery,
J. Control. Release 13 (2000) 91102.
[32] J.H. Jang, T.L. Houchin, L.D. Shea, Gene delivery from polymer scaffolds for tissue
engineering, Expert Rev. Med. Devices 1 (2004) 127138.
[33] P. Carmeliet, M. Tessier-Lavigne, Common mechanisms of nerve and blood vessel
wiring, Nature 7048 (2005) 193200.
[34] E. Storkebaum, D. Lambrechts, P. Carmeliet, VEGF: once regarded as a specic
angiogenic factor, now implicated in neuroprotection, Bioessays 9 (2004) 943954.
[35] E. Storkebaum, D. Lambrechts, M. Dewerchin, M.P. Moreno-Murciano, S. Appelmans,
H. Oh, P. Van Damme, B. Rutten, W.Y. Man, M. De Mol, S. Wyns, D. Manka, K.
Vermeulen, L. Van Den Bosch, N. Mertens, C. Schmitz, W. Robberecht, E.M. Conway,
D. Collen, L. Moons, P. Carmeliet, Treatment of motoneuron degeneration by
intracerebroventricular delivery of VEGF in a rat model of ALS, Nat. Neurosci. 1
(2005) 8592.