Biomaterials 77 (2016) 207e215

Contents lists available at ScienceDirect

Biomaterials
journal homepage: www.elsevier.com/locate/biomaterials

Tissue engineered pre-vascularized buccal mucosa equivalents

utilizing a primary triculture of epithelial cells, endothelial cells and
fibroblasts*
M. Heller a, c, e, *, 1, E.V. Frerick-Ochs a, e, 1, H.-K. Bauer a, e, E. Schiegnitz b, e, D. Flesch a, e,
J. Brieger c, e, R. Stein a, B. Al-Nawas b, e, C. Brochhausen d, e, J.W. Thüroff a, R.E. Unger d, e,
W. Brenner a, e, *
a
Department of Urology, University Medical Center of the Johannes Gutenberg University Mainz, Langenbeckstr. 1, 55131 Mainz, Germany
b
Department of Maxillofacial Surgery, University Medical Center of the Johannes Gutenberg University Mainz, Langenbeckstr. 1, 55131 Mainz, Germany
c
Department of Otorhinolaryngology, University Medical Center of the Johannes Gutenberg University Mainz, Langenbeckstr. 1, 55131 Mainz, Germany
d
Department of Pathology, University Medical Center of the Johannes Gutenberg University Mainz, Langenbeckstr. 1, 55131 Mainz, Germany
e
For BiomaTiCS e Biomaterials, Tissues and Cells in Science, University Medical Center of the Johannes Gutenberg University Mainz, Langenbeckstr. 1, 55131
Mainz, Germany

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

Article history: Artificial generated buccal mucosa equivalents are a promising approach for the reconstruction of ure-
Received 8 September 2015 thral defects. Limiting in this approach is a poor blood vessel supply after transplantation, resulting in
Received in revised form increased morbidity and necrosis. We generated a pre-vascularized buccal mucosa equivalent in a tri-
26 October 2015
culture of primary buccal epithelial cells, fibroblasts and microvascular endothelial cells, using a native
Accepted 29 October 2015
Available online 30 October 2015
collagen membrane as a scaffold. A successful pre-vascularization and dense formation of capillary-like
structures at superficial areas was demonstrated. The lumen size of pre-formed blood vessels corre-
sponded to the capillary size in vivo (10e30 mm). Comparing native with a highly cross-linked collagen
Keywords:
Buccal mucosa equivalent
membrane we found a distinct higher formation of capillary-like structures on the native membrane,
Pre-vascularization apparently caused by higher secretion of angiogenic factors such as PDGF, IL-8 and angiopoietin by the
Tri-culture cells. These capillary-like structures became functional blood vessels through anastomosis with the host
Epithelial cells vasculature after implantation in nude mice. This in vitro method should result in an accelerated blood
Fibroblasts supply to the biomaterial with cells after transplantation and increase the succes rates of the implant
Microvascular endothelial cells material.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction
In the field of regenerative urology the reconstruction of
inherent or acquired urethra failures is an issue of great interest.
State of the art for the reconstruction of this kind of defect is the
utilization of autologous tissue. Many different approaches have
*
A portion of the work described herein was carried out by Heide-Katharina
Bauer and Elena Frerick-Ochs in partial fullfillment of the requirements for a
Doctoral and Masters' degree, respectively, at the Johannes Gutenberg University,
Mainz, Germany.
* Corresponding author. Department of Urology of the Johannes Gutenberg
University Mainz, Langenbeckstr. 1, D-55131 Mainz, Germany.
E-mail address: brenner@uni-mainz.de (W. Brenner).
1
These authors contributed equally to this study.
been developed to find suitable donor tissue such as skin from the
scrotum and penis or from extra genital areas such as the ureter,
appendix, rump, extremities as well as mucosa from the foreskin,
bladder or the gingiva [1e6]. Although autologous tissue trans-
plants are widely used in this field, it has been found that trans-
plants from certain donor sites were associated with complications
after transplantation such as transplant shrinkage, stricture and
formation of diverticula for example in the case of tissue from skin
or bladder [7].
In the case of the good accessible skin tissue it could be shown
that the transplantation resulted in post-operative contraction,
constriction or swelling [5]. For bladder mucosa one of the biggest
issues has been the difficult accessibility [6]. In contrast, buccal
tissue has been shown to be suitable for urethra reconstruction due
to its favourable morphological and mechanical properties. The

http://dx.doi.org/10.1016/j.biomaterials.2015.10.073
0142-9612/© 2015 Elsevier Ltd. All rights reserved.
208 M. Heller et al. / Biomaterials 77 (2016) 207e215
favorable morphology of buccal mucosa is basically due to the
epithelial cover layer and the fibrous layer underneath, the lamina
propria. In general, the epithelium is a multilayer of a basal layer,
spinous layer, granular layer and the superficial layer with a
thickness of approximately 500 mm [8]. In areas with higher me-
chanical stress such as the palate, the superficial layer consists of
keratinized cells. The lamina propria is a fibrous and good vascu-
larized layer, which is situated underneath the epithelium and is
crucial for the nutrient supply of the epithelium. It consists of type I
and type III collagen, elastin and fibroblasts and thus plays an
important role for the mechanical stability of the epithelial layer
above [9,10].
Since the success of an operative urethra reconstruction is
mainly determined by the size of adequate donor tissue for the
tissue transfer [11], one of the main issues is the limited quantity of
applicable donor tissue. Due to the described issues associated with
autologous transplants, one promising scientific approach is the
creation of artificial tissue equivalents in vitro via tissue engineer-
ing. It has been demonstrated that buccal tissue equivalents con-
sisting of keratinocytes alone or of a co-culture with keratinocytes
and fibroblasts could be cultivated on dermal matrices and then
implanted as “full thickness oral mucosa” [10e13].
However, a limitation in such artificial tissue transplants is the
missing vascularization which results in necrosis. This is one of the
main reasons for an insufficient in-growth of the host vasculature
and survival of the artificial transplants [14e16]. In order to over-
come the problem of insufficient vascularization after trans-
plantation, various strategies have been developed to promote
blood vessel formation: (1) a direct application of endothelial cells
on a biomaterial followed by the implantation, (2) addition of
growth factors to the biomaterial, (3) direct injection of endothelial
cells into the implantation site, in addition to others [17e19].
Nevertheless, a satisfactory result has as yet not been reached by
this approach. As a solution to this problem, the formation of a
capillary network in vitro appears to be the most advantageous at
this point to accelerate the vascularization of a tissue engineered
scaffold after transplantation [14].
The aim of the current study is the generation and cultivation of
pre-vascularized buccal tissue equivalents based on a tri-culture of
human gingival epithelial cells, fibroblasts and microvascular
endothelial cells and the evaluation of the tissue equivalent inte-
gration in vivo using a mouse model. As a scaffold a native collagen
membrane was utilized in order to generate a functional tissue
equivalent with pre-formed capillary-like structures. Furthermore,
the impact of collagen cross-linking on the formation of capillaries
has been evaluated in a comparative study with the native and the
highly cross-linked collagen membrane, and moreover the secre-
tion of angiogenic factors has been analyzed in vitro. The in vivo
experiments, using a nude mouse model, showed the functional
connection of the pre-formed capillary-like structures within the
generated tissue equivalents by the staining of a human specific
endothelial cell marker CD31 and by the perfusion of the pre-
formed vasculature with mouse erythrocytes 10 days after im-
plantation [20]. It is anticipated that such a pre-vascularized
construct will lead to an accelerated integration in the body and
may result in an improved blood vessel supply associated with
increased survival and higher success rates after transplantation in
humans.
2. Materials and methods
2.1. Sample preparation
Native collagen membrane, i.e. processed collagen derived from
natural source (Geistlich Bio-Gide®), and cross-linked collagen
membrane (Geistlich Bio-Gide® Pro, Wolhusen, Switzerland), both
porcine type I and III collagens, bilayered, including a cell occlusive
and a rough side [21], were used (Fig. 1A). The 4X cross-linking re-
action of the Geistlich Bio-Gide® Pro matrices was performed using
ethyl(dimethylaminopropyl)carbodiimide/N-hydroxysuccinimid
(EDC/NHS) chemistry. Prior to cell experiments, collagen mem-
branes were punched in 6 mm diameter, 28.3 mm2, using biopsy
punches and stored sterile until used. In order to distinguish be-
tween the cell occlusive and rough side, samples were carefully
marked. Immediately before cell seeding collagen membranes were
rehydrated with PBS.
2.2. Cell isolation and cell culture
Epithelial cells and fibroblasts were isolated from human
gingiva, obtained from patients who underwent surgery at the
Department of Maxillofacial Surgery, University Medical Center
Mainz, Germany. The study was performed in agreement with the
Declaration of Helsinki and approved by local ethics committee
(Landes€arztekammer Rheinland-Pfalz, Mainz, Germany:
837.439.05). Informed consent was obtained from each patient.
Epithelial cell isolation procedure: Prior to isolation, gingiva
samples were stepwise sterilized for 15 s in sterillium (Bode
Chemie GmbH, Hamburg, Germany), 15 s in 70% ethanol and finally
washed twice in sterile PBS. Then the tissue samples were punc-
tured on the epithelial side using a cannula and incubated for 4 h in
a 0.04% trypsin solution in DMEM at 37
C. Afterwards the trypsin
reaction was stopped using DMEM including 10% FCS (fetal calf
serum) and the epithelial layer was carefully separated from the
connective tissue. To isolate epithelial cells, the epithelial layer was
cut into small tissue fragments, placed in a 6-well plate and covered
with the epithelial cell-specific medium containing 60% DMEM
(Dulbecco's modified eagle medium, PAA, Pasching, Austria), 30%
DMEM/HAM's F12 (Gibco, Darmstadt, Germany), 10% FCS, 100 U/
100 mg/ml Penicillin/Streptomycin, 2.5 mg/ml Fungizone (Gibco,
Darmstadt, Germany), 22 mg/ml adenine, 7.4 ng/ml cholera toxin,
9 ng/ml EGF, 36 ng/ml hydrocortisone and 4.6 mg/ml insulin.
Fibroblast isolation procedure: To isolate fibroblasts, the con-
nective tissue of the buccal mucosa was cut into small fragments
(1  1 mm2), transferred and cultivated in 25 cm2 cell culture flasks
using DMEM medium with 10% FCS, 100 U/100 mg/ml Penicillin/
Streptomycin and 2.5 mg/ml Fungizone. After one week of cultiva-
tion cell media were changed at least every 3 days.
Microvascular endothelial cell isolation: Microvascular endo-
thelial cells (human dermal microvascular endothelial cells,
HDMEC) were isolated from human juvenile foreskin [22] and
cultured in PC-Medium (Customer Formulation Medium, Promo
Cell, Heidelberg, Germany) supplemented with 15% FCS, 10 mg/ml
Na-heparin, 0.2 ng/ml bFGF, 100 U/100 mg/ml Penicillin/Strepto-
mycin and 2.5 mg/ml Fungizone.
All cells were cultivated at 37
C, 5% CO2 and 95% relative
humidity.
2.3. Cell cultivation of the co- and triculture
Co- and tricultures were generated by placing the collagen
membrane (6 mm diameter, 28.3 mm2) in cell culture inserts for 24
well plates (Netwell Insert No. 3477 from Corning Costar Corpo-
ration, Amsterdam, Netherlands). In order to evaluate the influence
of the cell number on the formation of capillary-like structures
different cell numbers of endothelial cells and fibroblasts (ratio 1:1)
in coculture were seeded (2  105, 4  105, 8  105 each cell type/
scaffold). Epithelial cells were seeded with a cell number of 2  105.
For the generation of the triculture, the endothelial cells were
first seeded onto the rough side of the collagen membrane. After
 M. Heller et al. / Biomaterials 77 (2016) 207e215 209

Fig. 1. A) Scanning electron microscope images of the natural (Bio-Gide®) and the cross-linked (Bio-Gide® Pro) collagen membranes after 24 h incubation with cell culture medium,
each the cell occlusive and the rough side. B) Scheme of the pre-vascularised buccal mucosa equivalent of epithelial cells, fibroblasts and endothelial cells based on a native collagen
membrane. Human dermal microvascular endothelial cells and one day later fibroblasts were seeded on the porous side of the collagen membrane. After four days, buccal mucosa
epithelial cells were placed on the cell occlusive side. Seeded membranes were analysed after 3 weeks of cultivation.

1 day of incubation, the fibroblasts were added on top of the seeded
endothelial cells. After cultivation with endothelial cells and fi-
broblasts for three days the collagen membrane was carefully
turned upside-down (occlusive side up) and seeded with epithelial
cells (Fig. 1B). Seeded scaffolds were analysed after 3 weeks of
cultivation.
2.4. Confocal laser scanning microscopic (CLSM) analyses of
capillary-like structures
For PECAM/CD31 staining of endothelial cells seeded collagen
membranes were fixed in 3.7% Paraformaldehyde for 15 min and
treated in 0.2% Triton-X 100 for 10 min. The primary mouse-anti-
human-PECAM-1/CD31 (1:50 in 1% BSA) antibody, clone JC70A
(Dako, Hamburg, Germany) was incubated for 1 h. The secondary
anti-mouse antibody “Alexa 488” (Invitrogen, Gibco Life Technol-
ogies GmbH, Darmstadt, Germany) was used at a dilution of 1:1000
in 1% BSA for 1 h in the dark.
2.5. Masson Goldner staining
The Masson-Goldner staining kit from Merck (Darmstadt, Ger-
many) was used. First the cross-section samples were stained for
5 min with iron hematoxylin stain after Weigert and then
immersed in water for another 5 min. The specimens were incu-
bated for 30 s in 1% acetic acid, then 10 min in azophloxin-solution
for staining and afterwards in 1% acetic acid for 30 s. This was fol-
lowed by a staining with tungstophosphoric acid Orange G solution
for 1 min and a Light green stain for 2 min with a 30 s washing step
with 1% acetic acid after each staining step. Finally, specimens were
washed step-wise in 70% ethanol, 96% ethanol, 100% isopropanol
for 2e3 min and afterwards 3 times in xylene for 5 min.
2.6. Immunohistochemical staining
For the immunohistochemical staining procedures, seeded
collagen membranes were fixed in paraformaldehyde (3.7%),
paraffin embedded, cut in 5 mm sections and then deparaffinized.
For PECAM/CD31 staining the first antibody was used as described
above. For staining of Tenascin, collagen IV and cytokeratin 13 the
following antibodies were used: anti-tenascin, monoclonal anti-
body, mouse, clone BC-24 (1:2000); anti-collagen IV, monoclonal
antibody, mouse, clone COL-94 (1:250); anti-cytokeratin peptide
13, monoclonal antibody, mouse, clone KS-1A3 (1:100), all from
Sigma Aldrich (Steinheim, Germany). Five mm sections were rinsed
twice with aqua dest. and treated with proteinase K (DAKO,
Hamburg, Germany) for 3 min in a wet chamber. Samples were
rinsed twice with aqua dest. and incubated with peroxidase
blocking solution for 20 min. All further steps were conducted in
immunostaining chambers (coverplates, Thermo, Waltham, USA).
The first antibodies were incubated for 1 h. After washing, the
secondary antibody “polylink” from the Kitsystems LSABþ-System
HRP Rabbit/Mouse/Goat (DABþ) from Dako (Hamburg, Germany)
was added to the cover slides and incubated for 30 min. After
washing, samples were incubated at room temperature for 30 min
with horseradish peroxidase. The staining was developed using
100 ml chromogen DAB solution for 10 min. Counterstaining was
performed with haemalaun.
2.7. Analysis of the secretion of the angiogenic factors
The native and cross-linked collagen membranes were seeded
with endothelial cells and fibroblasts as described. After 3 weeks
culture medium was replaced with serum-free medium. Superna-
tant of the coculture was obtained after 24 h. Angiogenesis-specific
210 M. Heller et al. / Biomaterials 77 (2016) 207e215
cytokines were quantified using the Bio-Plex Pro human angio-
genesis assay (including PDGF-BB, IL-8, Angiopoietin-2 and VEGF)
(Bio-Rad Laboratories AG, Reinach, Switzerland) according to the
manufacturer's instructions. Briefly, ready-to-use antibody-bound
beads were added to the culture supernatant and incubated for
30 min. After washing, the beads were incubated with the bio-
tinylated secondary antibody, diluted in Bio-Plex Detection Anti-
body Diluent A, for 30 min. After washing again Strepavidin-PE
solution was added and incubated for 10 min. Beads were treated
with Bioplex Assay Puffer for 30 s and analysed in a Luminex-100
reader. A standard range was used for quantification. Values of
angiogenic factors secreted from cells seeded on the native collagen
membrane were normalized to those of angiogenic factors secreted
from of cells seeded on the cross-linked collagen membrane.
2.8. Analysis and evaluation of the tissue equivalent integration
in vivo
This study was approved by the Committee on the USE of Live
Animals in Teaching and Research, Rhineland-Platinate. For this
study, male CD-1 mice (6 weeks old, 20e25 g body weight Charles
River Laboratories, Sulzfeld, Germany) were used. For these ex-
periments mucosa equivalents were used that were cultivated
in vitro for 21 days with endothelial cells and fibroblasts as
described above. The in vivo integration of the pre-vascularized
buccal mucosa equivalents was analyzed by implanting the
in vitro generated equivalents subcutaneously in two prepared
subcutaneous pockets in the subscapular region and sutured at the
back fascia with the cell occlusive side upside down, in order to
keep the transplants in place. All surgical procedures were per-
formed under anaesthesia with ketamine (61,5 mg/kg b.w. i.p.,
Ketanest; Pfizer Pharma, Berlin Germany) and xylazine (2.3 mg/kg
b.w. i.p., Rompun; Bayer, Leverkusen, Germany).
2.9. Histology and immunohistochemical preparation and
evaluation
After 10 days the mice were euthanized by the dislocation of the
neck. Immediately after sacrifice the biomaterial was explanted
with surrounding tissue, fixed in 4.5% buffered formaldehyde
(Roti®-Histofix) for 24 h and then embedded in paraffin after
dehydration. After deparaffinization and rehydration, 5 mm sections
were stained with Mayer's haematoxylin and eosin (HE) to detect
cellular structures, in particular erythrocytes.
For detection of HDMEC the sections were deparaffinized and
rehydrated. After heat-induced epitope retrieval and blocking of
endogenous peroxidase, the sections were incubated with human
specific anti-PECAM-1 (CD31) antibody (Dako, Sweden) and the
Universal LSAB™ 2 Kit/HRP, Rabbit/Mouse (Dako, Sweden) was
used to detect the bound anti-PECAM-1 antibody. As negative
control, a similar tissue section was stained without the use of the
primary antibody. All negative controls remained unstained. The
slides were then evaluated by light microscopy.
3. Results
3.1. Optimization of culture conditions
In order to optimize the culture conditions concerning the for-
mation of capillary-like structures various cell numbers of endo-
thelial cells and fibroblasts in a 1:1 ratio were seeded on the rough
side of the native collagen membranes, cultured for three weeks
and stained for PECAM/CD31. It was found that on all samples
capillary-like structures were formed (Fig. 2). Samples seeded with
2  105 (Fig. 2AeC) and 4  105 cells of each type (Fig. 2DeF)
showed a higher amount of branched capillary-like structures and
elongated tubes in comparison to 8  105 cells/cm2 (Fig. 2GeI).
Furthermore, 4  105 and 8  105 cells/cm2 partly lead to a more
spread morphology of the endothelial cells (Fig. 2E and H, white
arrows), suggesting a lower degree of differentiation [23]. Taken
together the best results in formation of capillary-like structures
were reached using endothelial cells and fibroblasts in a concen-
tration of 4  105 cells/cm2 each.
Under this culture condition buccal epithelial cells, seeded on
the cell occlusive side of the native collagen membrane generated a
confluent multilayer, visualized by Phalloidin staining (Fig. 3). The
red stained epithelial cells in Fig. 3A on the cell occlusive side of the
collagen membrane revealed a spread morphology of the single
cells and a confluent layer as result of sufficient cell adhesion and
amplification.
Simultaneous staining of PECAM-1/CD31 and Phalloidin was
used to proof the successful cultivation of epithelial cells and
endothelial cells on the native collagen membrane under tri-
culture conditions (Fig. 3):endothelial cells (Fig. 3B) had formed
capillary-like structures forming a dense and confluent cell-layer.
3.2. Formation of capillary lumen within the buccal mucosa
equivalent
As additional quality evaluation of the formed capillary-like
structures within the native collagen matrices the lumina sizes
were analyzed using the immunohistological PECAM-1/CD31
stained specimens and CLSM-pictures (Fig. 4). Both methods
demonstrated that the lumen diameters were between 13 e 23 mm
and 10e30 mm and were thus in agreement. In the area of the
occlusive side of the membrane a positive staining indicated the
presence of endothelial cells. However, in this area no formation of
capillary-like structures and thus lumen could be observed.
3.3. Cell distribution and infiltration depth
The distribution of epithelial cells, fibroblasts and endothelial
cells seeded on the membranes was determine by immunohisto-
logically stainings (Fig. 5). The distinct red (in the web version)
staining of cytokeratin 13 on the upper cell occlusive side of the
membrane (Fig. 5A, arrows) showed a confluent multi-layer of
epithelial cells. Fibroblasts, visualized by the Tenascin-staining,
migrated completely into the membrane (Fig. 5B). The more
distinct staining on the rough bottom side of the collagen mem-
brane, indicated a higher amount of the fibroblast specific extra-
cellular matrix protein Tenascin and thus a higher fibroblast density
compared to deeper areas of the membrane. In contrast to the high
infiltration depth of the fibroblasts, microvascular endothelial cells
infiltrate the collagen membrane on the bottom side predomi-
nantly superficially (Fig. 5C, arrows).
3.4. Comparison of the native and cross-linked collagen matrices
Since native collagen matrices exhibit limited stability we
compared the native membrane to the more stable cross-linked
membrane by examining the formation of capillary-like struc-
tures. Both matrices consisted of the same basic material. While
endothelial cells formed a dense network of capillary-like struc-
tures on the native membrane, only a poor formation of these
structures was found on the cross-linked membrane (Fig. 6).
Although the cellular secretion of the angiogenic growth factor
VEGF was lower in the tri-culture on native collagen membranes,
all other investigated angiogenic factors PDGF, angiopoetin and IL-8
showed higher levels when compared to the cross-linked mem-
brane (Fig. 7).
 M. Heller et al. / Biomaterials 77 (2016) 207e215 211

Fig. 2. Influence of seeded endothelial cell numbers on formation of capillary-like structures. Endothelial cells and one day later fibroblasts in different concentrations were seeded
on the porous side of the collagen membrane. AeC: 2  105, DeE: 4  105, GeI: 8  105 cells per scaffold (28.3 mm2). After 3 weeks of cultivation the most successful formation of
capillary-like structures was observed by using 4  105 cells/scaffold. On samples seeded with higher cell numbers (DeI), cells showed a more spread morphology.

Fig. 3. A collagen membrane seeded with microvascular endothelial cells and one day later with fibroblasts, each 4  105 cells per scaffold (28.3 mm2) on the porous side of the
membrane and four days later with buccal mucosa epithelial cells on the cell occlusive surface. Visualization of A) epithelial cells on the cell occlusive membrane side stained with
phalloidin (red staining) B) and endothelial cells stained with a primary PECAM/CD31-and a secondary Alexa 488-labelled antibody (green staining) on the porous membrane side
in tri-culture after 3 weeks of cultivation. The uneven colouring of the epithelial cells is caused by unequal surface structures of the collagen membrane preventing focussing of the
whole area under observation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).
212 M. Heller et al. / Biomaterials 77 (2016) 207e215

Fig. 4. A native collagen membrane seeded with microvascular endothelial cells and one day later with fibroblasts, each 4  105 cells per scaffold (28.3 mm2) on the porous side of
the membrane. Immunohistological staining (5 mm sections of a paraffin embedded collagen membrane, DAB, left) and confocal laser scanning microscopic (CLSM) evaluation
(right) of PECAM/CD31 on tri-cultured membranes after 3 weeks cultivation. The brown structures in the immunohistological staining (left) show lumen of capillary-like structures
in superficial layers of the collagen membrane. The CLSM analysis (right) shows green stained capillary-like structures. The lumen sizes calculated by both methods were between
10 and 30 mm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).

Fig. 5. Five mm sections of a collagen membrane seeded with microvascular endothelial cells and one day later with fibroblasts, each 4  105 cells per scaffold (28.3 mm2) on the
porous side of the membrane and four days later with buccal mucosa epithelial cells on the cell occlusive surface. A) Epithelial cell layer on the cell occlusive side visualized by
Cytokeratin 13 staining (arrow), B) Infiltrated fibroblasts on the porous side and inside the membrane visualized by Tenascin staining (arrows), C) Endothelial cells in more su-
perficial areas of the collagen membrane visualized by CD31/PECAM staining (arrows).

Fig. 6. Collagen membrane seeded with microvascular endothelial cells and one day later with fibroblasts, each 4  105 cells per scaffold (28.3 mm2) on the porous side. Comparison
of capillary-like structures formed by endothelial cells on A) native collagen membrane and B) highly cross-linked collagen membrane after 3 weeks cultivation. On the native
collagen membrane the endothelial cells form a dense network of capillary like structures. In contrast, on the cross-linked collagen membrane a formation of capillary-like
structures could be observed, but the endothelial cells were incapable to develop a pronounced capillary network as observed on the native collagen membrane.

3.5. Pre-vascularized mucosa equivalent integration in vivo
The cell-seeded collagen membranes were implanted into the
animals after 21 days of cultivation with HDMEC and fibroblasts.
The implantation sites healed without complications and the
animals showed no signs of discomfort after surgery. After 10 days
the implants were removed and histollogically prepared as
described above. The examination of the histological samples show
that the blood vessels at superficial areas of the collagen membrane
are distinctly stained with human specific CD-31, indicating the
 M. Heller et al. / Biomaterials 77 (2016) 207e215
Fig. 7. Secretion of angiogenic factors in the co-culture of endothelial cells and fibroblasts, seeded on the porous side of the native or the cross-linked collagen membrane (each
4  105 cells per scaffold, 28.3 mm2), determined by luminex analyses. After 3 weeks of co-culture, medium was replaces by serum starved medium for 24 h and supernatant was
analyzed. Values of angiogenic factors secreted from cells seeded on the native collagen membrane were normalized to those of angiogenic factors secreted from cells seeded on the
cross-linked collagen membrane (dashed line).
formation of functional blood vessels from the pre-formed capil-
lary-like structures of the in vitro generated mucosa equivalent
(Fig. 8 A, black arrow). In contrast to these findings, the
erythrocyte-containing mouse vessels remained unstained. The
presence of erythrocytes inside the human-specific CD31 positive
vessels stained by haematoxylin and eosin demonstrated the
functional anastomosis of the human-derived vasculature to the
host vasculature (Fig. 8 B and C, black arrows).
4. Discussion
This study reports on the in vitro generation of a pre-
vascularized buccal mucosa equivalent consisting of a tri-culture
of epithelial cells, fibroblasts and endothelial cells on a native
collagen membrane sufficiently stable for handling for
transplantation.
Culture conditions for the formation of capillary-like structures
on the native collagen membrane were optimized by analyzing the
effect of different seeded cell numbers of fibroblasts and endo-
thelial cells. Afterwards, the effect of collagen cross-linking on the
tube formation of endothelial cells was investigated by comparing
the native with the cross-linked membranes under tri-culture
conditions. The complex morphology of a buccal mucosa [24] was
only observed on the native membrane as demonstrated by the
distribution and infiltration depth of the respective cells. The for-
mation of a higher number of capillary-like structures on the native
membrane compared to the cross-linked membrane could be
supported by an enhanced secretion of angiogenic factors of cells
that were cultivated on native collagen membranes. Our in vitro
data are in line with findings of Rothamel et al. [25] and Thoma
et al. [26] demonstrating decreased vascularisation on cross-linked
collagen membrane when compared to the respective native
collagen membrane in vivo.
The cultivation of primary human cells is demanding with
respect to the culture conditions. For a tri-culture, various condi-
tions such as culture medium and cell numbers seeded need to be
determined, adapted and optimized in order for all cell types to
grow and express their respective characteristics. In this study,
various cell numbers of fibroblasts and endothelial cells were
cultivated under tri-culture conditions to analyze the influence of
cell density on the formation of capillary-like structures. We could
demonstrate that a lower cell number of each cell type resulted in
an increased amount of branched capillary-like structures and thus
showed a direct effect on the microcapillary-like structure
213
formation of endothelial cells associated with a higher degree of
differentiation.
Based on previous studies it can be anticipated that cells seeded
in a high concentration show reduced intercellular communication,
resulting in reduced differentiation stages [27]. Taken together and
most notably the use of our special developed culture medium
supported the formation of a confluent epithelial cell layer simul-
taneously to the formation of capillary-like structures. The suc-
cessful development of an artificially generated tissue equivalent is
highly dependent on the choice of an adequate scaffolding material.
It has been shown that epithelial cells need a basement membrane
in order to grow and proliferate [28]. This was one of the criteria for
the selection of the native collagen membrane in that it needed to
exhibit a basement membrane like characteristic and function. It
was anticipated that the cell occlusive side of the of the native
collagen membrane with a pore size of 300 nm [29] that could
function as a natural barrier for epithelial cell penetration and
simultaneously provide an adequate base for the epithelial cells to
adhere and grow. Although the native basement membrane in
buccal mucosa consists of collagen type IV [30], and the native
collagen membrane used in these studies contains type I and III
collagens [31], our results demonstrated a good compatibility of the
membrane in the formation of a dense epithelial cell layer in the tri-
culture system. In addition, the mechanical stability of the mem-
brane for the 3 week cultivation period allowed sufficient time for
the cells to proliferate. Furthermore, in order to generate a well-
vascularized fibrous tissue, as is found in the lamina propria of
the native buccal mucosa in vivo, fibroblasts and endothelial cells
were seeded on the porous side of the collagen membrane. With a
pore size of 5 mm [29] the collagen membrane appeared to provide
suitable structures and conditions for the fibroblasts and endo-
thelial cells in order to migrate into the scaffold and to form a
vascularized, tissue-like structure.
For the evaluation of the successful pre-vascularisation, the
infiltration depth of endothelial cells and lumen formation of the
capillary-like structures was visualized using immuno-histological
CD31/PECAM staining. It could be observed that the endothelial
cells formed lumen only superficially, suggesting that endothelial
cells did not migrate into the deeper areas of the collagen mem-
brane. It is well known that for angiogenesis e. g. in embryogenesis
or during tumor growth, endothelial cells need chemotactical sig-
nals for a directed migration [32]. The absence of these signals may
prevent the endothelial cells from migrating into the deeper areas
of the collagen membrane. Alternatively, the gradually decreasing
214 M. Heller et al. / Biomaterials 77 (2016) 207e215
Fig. 8. Integration of the pre-vascularized mucosa equivalent after transplantation in
the neck area of mice. 10 days after surgery the mucosa equivalents were removed
again and stained immunohistochemically with a human specific anti-CD31 antibody.
A) The blood vessels in superficial areas of the collagen membrane (dashed line) are
distinctly stained, proving the origin from the pre-formed capillary-like structures of
the in vitro generated transplant. In contrast to these findings, the mouse vessels are
not stained (white arrow). B) The blood vessels inside the buccal mucosa equivalent at
superficial areas of the collagen membrane (dashed line) are stained with human
specific anti-CD31 antibody, showing that the pre-formed capillary-like structures
formed functional blood vessels in vivo (black arrows). C) The erythrocytes stained by
HE staining inside the blood vessels (black arrows) show the functional connection of
these blood vessels to the vessel system of the mouse.
pore size from the rough to the cell occlusive side of the membrane
may prevent a deep endothelial cell infiltration. In contrast, since it
was found in earlier works that fibroblasts tend to migrate and
infiltrate into collagen matrices [33], the more distinct and thus
complete infiltration inside the membrane was to be expected.
In order to evaluate the influence of cross-linking of collagen on
the formation of capillary-like structures, cell cultivation of the tri-
culture was performed in a comparative study on native and a
cross-linked collagen membrane. We demonstrated that there was
poor tube formation and pre-vascularisation of endothelial cells on
the cross-linked collagen membrane, whereas the capillary-like
network on the native collagen membranes was very dense and
showed highly branched and tubular structures. The cross-linking
of the collagen membrane potentially causes a physical barrier for
endothelial cell adhesion and thus for tube formation. Furthermore,
the chemical modification of the collagen may mask cellular
binding sites like the adhesion amino acid sequence RGD.
In this context, the angiogenic factors angiopoetin, IL-8 and
PDGF were secreted in higher amounts from cells that were seeded
on the native membrane compared to the cross-linked membrane.
The most significant difference between these two collagen
membranes was observed in the secretion of angiopoietin. Angio-
poietin is known to induce long un-leaky and stable vessels and
induces e like PDGF [34,35] e vessel maturation [36]. Furthermore,
it enhances interactions between endothelial cells and their
microenvironment in surrounding cells and extracellular matrix
[34], inducing a resistant capillary network. Also IL-8 induces the
formation of capillary tubes [37] and has been shown to lead to cell
survival and inhibition of apoptosis [38], supporting a stable
capillary network. Interestingly, VEGF, one of the most important
growth factors involved in angiogenesis was present in lower
concentrations on native collagen compared to the cross-linked
membrane. VEGF is relevant for endothelial cell differentiation
and early blood vessel formation [39]. The level of angiogenic fac-
tors was determined after 3 weeks of culture and it is possible that
at this time point the VEGF level was reduced since angiogenesis
was already proceeding. An in vitro generated pre-vascularized,
cell-containing equivalent of the buccal mucosa could be a prom-
ising alternative to autologous grafts. In transplantation studies
using collagen matrices for urethra reconstruction in animal
models a limited vascularization of the scaffolds is often observed
[40]. While focussing on the formation of capillary-like structures
in order to achieve an adequate oxygen and nutrient supply after
transplantation, the model described in this study also demon-
strated a sufficient distribution of the two other major cell types,
epithelial and fibroblasts, and the generation of tissue-like
structures.
The functional connection of the pre-formed capillary-like
structures to the vasculature of the host animals could be shown in
this study. According to the in vitro observations, the formation and
perfusion of the vasculature inside the buccal mucosa equivalent
from the already existing capillary-like structures took place only in
superficial areas of the collagen membrane. The absence of che-
motactical and angiogenic signals inside the collagen membrane
could have prevented the endothelial cells and/or blood vessels to
infiltrate deeper in the transplant as already discussed above. This
model demonstrated the formation of capillary-like structures,
however, additional optimization is required to achieve a deeper
infiltration of endothelial cells and the resulting formation of
capillary-like structures throughout the entire collagen membrane
and finally for a complete perfusion in depth of the implanted
tissue equivalent in vivo.
5. Conclusion
We have demonstrated that it is possible to create an artificially
generated pre-vascularized buccal mucosa equivalent by a tri-
culture of human epithelial cells, fibroblasts and endothelial cells
on a native collagen membrane. The results show an epithelial layer
on the upper side of the collagen membrane, a deep infiltration of
fibroblasts from the bottom side and superficially-formed capillary
structures by microvasclar endothelial cells exhibiting a lumen size
between 10 and 30 mm. Furthermore, the formation of functional
vasculature and anastomosis of the pre-formed capillary-like
structures to the host vasculature in vivo could be demonstrated,
supporting the promising approach of this study.
The pre-vascularized buccal mucosa equivalent may find use as
 M. Heller et al. / Biomaterials 77 (2016) 207e215
a beneficial replacement for commonly used autologous trans-
plants. The advantage is the pre-vascularization achieved in vitro
which may counteract transplant necrosis due to a more rapid and
improved nutrient and oxygen supply in vivo after implantation.
Acknowledgement
We thank Anne Sartoris for the expertise and support with all
technical questions. In addition, we would like to thank Geistlich
Pharma AG, Switzerland, especially Birgit Scha€fer, for generously
supplying the biomaterials, for training and assistance with the
analysis of the cell supernatants and for many helpful suggestions
and discussions.
References
[1] A. Dessanti, W. Rigamonti, V. Merulla, D. Falchetti, G. Caccia, Autologous
buccal mucosa graft for hypospadias repair: an initial report, J. Urol. 147
(1992) 1081e1883.
[2] D. Filipas, U. Wahlmann, R. Hohenfellner, History of oral mucosa, Eur. Urol. 34
(1998) 165e168.
[3] Y.M. Xu, Y. Qiao, Y.L. Sa, J. Zhang, H.Z. Zhang, X.R. Zhang, et al., One-stage
urethral reconstruction using colonic mucosa graft: an experimental and
clinical study, World J. Gastroenterol. 9 (2003) 381e384.
[4] S. Bhargava, C.R. Chapple, Buccal mucosal urethroplasty: is it the new gold
standard? BJU Int. 93 (2004) 1191e1193.
[5] G.D. Webster, M.W. Brown, R.B. Koefoot Jr., S. Sihelnick, Suboptimal results in
full thickness skin graft urethroplasty using an extrapenile skin donor site,
J. Urol. 131 (1984) 1082e1083.
[6] I.A. Mungadi, V.I. Mungadi, Oral mucosa grafts for urethral reconstruction,
Ann. Afr. Med. 8 (2009) 203e209.
[7] P.D.S. Corvin, G. Feil, A. Stenzl, Tissue engineering of the urethra and ureter,
Urol. A 43 (2004) 1213e1216.
[8] R. Stein, A. Schro €der, J.W. Thüroff, Use of buccal mucosa in urethral surgery,
Urol. A 46 (2007) 1657e1663.
[9] J. Liu, Z. Bian, A.M. Kuijpers-Jagtman, J.W. Von den Hoff, Skin and oral mucosa
equivalents: construction and performance, Orthod. Craniofac. Res. 13 (2010)
11e20.
[10] K. Moharamzadeh, I.M. Brook, R. Van Noort, A.M. Scutt, M.H. Thornhill, Tissue-
engineered oral mucosa: a review of the scientific literature, J. Dent. Res. 86
(2007) 115e124.
[11] M. Fossum, A. Nordenskjo € ld, Tissue-engineered transplants for the treatment
of severe hypospadias, Horm. Res. Paediatr. 73 (2010) 148e152.
[12] G. Lauer, R. Schimming, Tissue-engineered mucosa graft for reconstruction of
the intraoral lining after freeing of the tongue: a clinical and immunohisto-
logic study, J. Oral Maxillofac. Surg. 59 (2001) 169e177.
[13] K. Izumi, S.E. Feinberg, A. Iida, M. Yoshizawa, Intraoral grafting of an ex vivo
produced oral mucosa equivalent: a preliminary report, Int. J. Oral Maxillofac.
Surg. 32 (2003) 188e197.
[14] P.-L. Tremblay, V. Hudon, F. Berthod, L. Germain, F.A. Auger, Inosculation of
tissue-engineered capillaries with the host's vasculature in a reconstructed
skin transplanted on mice, Am. J. Transpl. 5 (2005) 1002e1010.
[15] S. Boyce, Cultured skin substitutes: a review, Tissue Eng. 2 (1996) 255e266.
[16] A.F. Black, F. Berthod, N. L'heureux, L. Germain, F.A. Auger, In vitro recon-
struction of a human capillary-like network in a tissue-engineered skin
equivalent, FASEB J. 12 (1998) 1331e1340.
[17] C.J. Kirkpatrick, S. Fuchs, R.E. Unger, Co-culture systems for vascularization-
learning from nature, Adv. Drug Deliv. Rev. 63 (2011) 291e299.
[18] H. Naderi, M.M. Matin, A.R. Bahrami, Review paper: Critical issues in tissue
engineering: biomaterials, cell sources, angiogenesis, and drug delivery sys-
tems, J. Biomater. Appl. 26 (2011) 383e417.
[19] J.M. Melero-Martin, Z.A. Khan, A. Picard, X. Wu, S. Paruchuri, J. Bischoff, In vivo
vasculogenic potential of human blood-derived endothelial progenitor cells,
215
Blood 109 (2007) 4761e4768.
[20] R.E. Unger, S. Ghanaati, C. Orth, A. Sartoris, M. Barbeck, S. Halstenberg,
A. Motta, C. Migliaresi, C.J. Kirkpatrick, The rapid anastomosis between pre-
vascularized networks on silk fibroin scaffolds generated in vitro with co-
cultures of human microvascular endothelial and osteoblast cells and the host
vasculature, Biomaterials 31 (27) (2010) 6959e6967.
[21] F. Schwarz, D. Rothamel, M. Herten, M. Wüstefeld, M. Sager, D. Ferrari,
J. Becker, Immunohistochemical characterization of guided bone regeneration
at a dehiscence-type defect using different barrier membranes: an experi-
mental study in dogs, Clin. Oral Implants Res. 19 (2008) 402e415.
[22] R.E. Unger, K. Peters, A. Sartoris, C. Freese, C.J. Kirkpatrick, Human endothelial
cell-based assay for endotoxin as sensitive as the conventional Limulus
Amebocyte Lysate assay, Biomaterials 35 (2014) 3180e3187.
[23] T.J. Fan, X.Z. Hu, J. Zhao, Y. Niu, W.Z. Zhao, M.M. Yu, Y. Ge, Establishment of an
untransfected human corneal stromal cell line and its biocompatibility to
acellular porcine corneal stroma, Int. J. Ophthalmol. 5 (2012) 286e292.
[24] S. Bhargava, J.M. Patterson, R.D. Inman, S. MacNeil, C.R. Chapple, Tissue-
engineered buccal mucosa urethroplasty-clinical outcomes, Eur. Urol. 53
(2008) 1263e1269.
[25] D. Rothamel, F. Schwarz, M. Sager, M. Herten, A. Sculean, J. Becker, Biodeg-
radation of differently cross-linked collagen membranes: an experimental
study in the rat, Clin. Oral Implants Res. 16 (2005) 369e378.
[26] D.S. Thoma, C.C. Villar, D.L. Cochran, C.H. Ha €mmerle, R.E. Jung, Tissue inte-
gration of collagen-based matrices: an experimental study in mice, Clin. Oral
Implants Res. 23 (2012) 1333e1339.
[27] J. Rouwkema, J. de Boer, C. van Blitterswijk, Endothelial cells assemble into a
3-dimensional prevascular network in a bone tissue engineering construct,
Tissue Eng. 12 (2006) 2685e2693.
[28] C.Q. Lin, M.J. Bissell, Multi-faceted regulation of cell differentiation by extra-
cellular matrix, FASEB J. 7 (1993) 737e743.
[29] W.J. Cho, J.H. Kim, S.H. Oh, H.H. Nam, J.M. Kim, J.H. Lee, PCL electrospun sheet-
embedded microporous PLGA membrane for effective guided bone regener-
ation, in: 3rd Kuala Lumpur International Conference on Biomedical Engi-
neering, Springer, Berlin Heidelberg, 2006, pp. 122e125.
[30] N.A. Kefalides, Isolation of a collagen from basement membranes containing
three identical a-chains, Biochem. Biophys. Res. Commun. 45 (1971) 226e234.
[31] M. Camelo, M.L. Nevins, R.K. Schenk, M. Simion, G. Rasperini, S.E. Lynch,
M. Nevins, Clinical, radiographic, and histologic evaluation of human peri-
odontal defects treated with Bio-Oss and Bio-Gide, Int. J. Periodontic Restor.
Dent. 18 (1998) 321e331.
[32] L. Lamalice, F. Le Boeuf, J. Huot, Endothelial cell migration during angiogen-
esis, Circ. Res. 100 (2007) 782e794.
[33] R.A. Clark, J.Q. An, D. Greiling, A. Khan, J.E. Schwarzbauer, Fibroblast migration
on fibronectin requires three distinct functional domains, J. Invest. Dermatol.
121 (2003) 695e705.
[34] R. Cao, E. Brakenhielm, R. Pawliuk, D. Wariaro, M.J. Post, E. Wahlberg, et al.,
Angiogenic synergism, vascular stability and improvement of hind-limb
ischemia by a combination of PDGF-BB and FGF-2, Nat. Med. 9 (2003)
604e613.
[35] L.J. Nissen, R. Cao, E.M. Hedlund, Z. Wang, X. Zhao, D. Wetterskog, et al.,
Angiogenic factors FGF2 and PDGF-BB synergistically promote murine tumor
neovascularization and metastasis, J. Clin. Invest. 117 (2007) 2766e2777.
[36] G. Thurston, Complementary actions of VEGF and angiopoietin-1 on blood
vessel growth and leakage, J. Anat. 200 (2002) 575e580.
[37] I.J. Fidler, L.M. Ellis, Chemotherapeutic drugs more really is not better, Nat.
Med. 6 (2000) 500e502.
[38] A. Li, S. Dubey, M.L. Varney, B.J. Dave, R.K. Singh, IL-8 directly enhanced
endothelial cell survival, proliferation, and matrix metalloproteinases pro-
duction and regulated angiogenesis, J. Immunol. 170 (2003) 3369e3376.
[39] M.B. Nourse, D.E. Halpin, M. Scatena, D.J. Mortisen, N.L. Tulloch, K.D. Hauch, et
al., VEGF induces differentiation of functional endothelium from human em-
bryonic stem cells: implications for tissue engineering, Arterioscler. Thromb.
Vasc. Biol. 30 (2010) 80e89.
[40] Villoldo GM1, M. Loresi, C. Giudice, O. Damia, J.M. Moldes, F. DeBadiola, et al.,
Histologic changes after urethroplasty using small intestinal submucosa un-
seeded with cells in rabbits with injured urethra, Urology 81 (2013)
1380e1e1380e5.