Hydrogel-beta-TCP scaffolds and stem cells for tissue engineering bone.

mg M
bu
ol, Departments of Orthopedic Surgery and Pathology,
collagen I and fibrin hydrogels specimens had superior bone tissue, although radiopacities were detected only in collagen I samples. VCT scan
revealed density values in all but the Pluronic F127 samples, with Houndsfield unit values comparable to native bone in collagen I and fibrin glue
pseudarthrosis, non-union of the fracture, and loss of function. not allow bone formation and growth, as it is not biodegradable
and may lower the threshold for infection at the surgical site [4].
Tissue engineering bone is a new alternative that has a
potential to overcome many of the drawbacks mentioned above
Bone 38 (2006) 555⁎ Corresponding author. Department of Pediatric Surgery, Massachusettssamples. Expression of bone-specific genes was significantly higher in the collagen I samples. Pluronic F127 hydrogel did not support formation
of bone tissue. All samples cultured in dynamic oscillating conditions had slightly higher mechanical strength than under rotating conditions. Bone
tissue can be successfully formed in vitro using constructs comprised of collagen I hydrogel, MSCs, and porous β-TCP scaffolds.
© 2005 Elsevier Inc. All rights reserved.
Keywords: Mesenchymal stem cells; Bone tissue engineering; Three-dimensional printing; β-TCP scaffolds; Hydrogels; Volumetric CT scanning
Introduction
Trabecular bone deficiency causes a dilemma in a variety of
clinical situations, including trauma, tumor resection, and
reconstruction. Bone tissue usually has the ability to repair
itself, but when a defect of critical size needs to be bridged, the
repair attempt fails in most cases, resulting in the formation of a
Autologous bone transplantation is the most effective
method for bone restoration because it provides three essential
elements: osteoconduction, osteoinduction, and osteogenic
cells. Allografts are believed to be osteoconductive [1] but
confer the risk of disease transmission and immune rejection
[1]. Bone cement [2,3] is readily available and does not cause
either rejection or disease transmission issues. However, it doesOrthopedic Research Laboratory, Boston, MA 02114, USA
e Skeletal Biology Research Center, Massachusetts General Hospital, Harvard School of Dental, Medicine, 15 Parkman Street, Boston, MA 02114, USA
Received 22 April 2005; revised 2 October 2005; accepted 3 October 2005
Available online 20 December 2005
Abstract
Trabecular bone is a material of choice for reconstruction after trauma and tumor resection and for correction of congenital defects. Autologous
bone grafts are available in limited shapes and sizes; significant donor site morbidity is another major disadvantage to this approach. To overcome
these limitations, we used a tissue engineering approach to create bone replacements in vitro, combining bone-marrow-derived differentiated
mesenchymal stem cells (MSCs) suspended in hydrogels and 3-dimensionally printed (3DP) porous scaffolds made of β-tricalcium-phosphate (β-
TCP). The scaffolds provided support for the formation of bone tissue in collagen I, fibrin, alginate, and pluronic F127 hydrogels during culturing
in oscillating and rotating dynamic conditions. Histological evaluation including toluidine blue, alkaline phosphatase, and von Kossa staining was
done at 1, 2, 4, and 6 weeks. Radiographic evaluation and high-resolution volumetric CT (VCT) scanning, expression of bone-specific genes and
biomechanical compression testing were performed at 6 weeks. Both culture conditions resulted in similar bone tissue formation. Histologicallyd ChildrenTs Hospital Boston, Harvard Medical SchoLaboratory for Tissue Engineering and Organ Fabrication, Warren 11-1157, Massachusetts General Hospital, Harvard,
Medical School, 55 Fruit Street, Boston, MA 02114, USA
b Massachusetts Institute of Technology, Cambridge, MA 02139, USA
c Department of Radiology, Massachusetts General Hospital, Charlestown, Navy Yard, MA 02129-4557, USAHydrogel-β-TCP scaffolds and ste
Christian Weinand a, Irina Pomerantseva a, Crai
Ijad Madisch c, Frederic Shapiro d, Harutsugi A
aGeneral Hospital, USA. Fax: +1 617 726 5057.
E-mail address: JVacanti@Partners.Org (J.P. Vacanti).
1 Full disclosure: Dr. Gupta has served as a consultant to Siemens
Corporation, Forchheim, Germany.
8756-3282/$ - see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.bone.2005.10.016cells for tissue engineering bone
. Neville a, Rajiv Gupta c,1, Eli Weinberg a,b,
kawa e, Maria J. Troulis e, Joseph P. Vacanti a,⁎
–563
www.elsevier.com/locate/bone[5]. It is possible to create various tissues including bone on
preformed scaffolds, using autologous cells [6,7]. Although
autologous osteocytes can be harvested from multiple sites in

onethe body, their supply is limited. As an alternative bone-
marrow-derived mesenchymal stem cells (MSCs) have high
proliferation capacity and multilineage potential; they can be
differentiated into osteoblasts [8] and have been used to tissue
engineer bone [5].
Additional temporary mechanical support is needed to
facilitate formation of weight-bearing new bone tissue by
isolated cells. Many synthetic and natural materials such as β-
tricalcium-phosphate (β-TCP) [9,10], poly(lactic-co-glycolic)
acid (PLGA), and magnesium possess adequate strength for
bone reconstruction and have proven to be osteoinductive and
osteoconductive [11–14]. Ultimately, for defects with substan-
tial curvature, the tissue-engineered constructs should also have
appropriate topography.
Here, 3D printing (3DP) can be used. From a CAD picture of
the desired part, a slicing algorithm draws detailed information
for every layer. Each layer begins with a thin distribution of
powder spread over the surface of a powder bed. Using a
technology similar to ink-jet printing, a binder material
selectively joins particles where the object is to be formed.
This layer-by-layer process repeats until the part is completed.
Following a heat treatment, unbound powder is removed,
leaving the fabricated part. The support gained from the powder
bed means that overhangs, undercuts, and internal volumes can
be created (as long as there is a hole for the loose powder to
escape). Material can be in a liquid carrier, or it can be applied
as molten matter. The proper placement of droplets can be used
to create surfaces of controlled texture and to control the internal
microstructure of the printed part [15].
Several materials, such as ceramics, metals, polymers, and
composites are suitable for 3DP [9,16]. However, elaborate
seeding and culture conditions are needed to achieve uniform
cellular distribution, sustain cell viability, and provide nutrients
for tissue formation in scaffolds. Several dynamic culture
methods such as shear stress-inducing perfusion [17] and
rotation in oxygen-permeable bioreactor (ROB) [18] have
been shown to promote bone formation [19]. However, the
new bone-like tissue is mainly found on the surface of the
scaffolds because of the limited nutrient and oxygen supply in
the center of the scaffold. In our experiment, we compare a novel
dynamic oscillating (DO) culture technique inducing longitudi-
nal linear shear stress onto a free-floating scaffold with the
established ROB technique, which applies rotational shear stress
[18].
To address these limitations, we used hydrogels to facilitate
delivery and distribution of cells in scaffolds made of β-TCP.
Several hydrogels including alginate, fibrin glue, and pluronic
F127 are able to promote bone formation by MSCs and
osteoblasts [16,20–23]. Collagen I hydrogel (Cellagen) is
rarely used in bone tissue engineering [24]; however, it is an
essential part of native bone and can be beneficial for the
formation of new bone tissue. To date, no comparison has
been made between these hydrogels with regard to their ability
to support formation of bone by differentiated MSCs.
556 C. Weinand et al. / BHydrogels lack the initial mechanical strength needed for
weight bearing, which is a serious disadvantage for implan-
tation and makes it impossible to use them alone as bonereplacements in vivo. Therefore, an additional support is
needed. Consequently, we examined the combination of
hydrogels as carriers for differentiated MSCs on osteoinduc-
tive scaffold materials.
In clinical practice, new bone formation is monitored using
conventional radiological techniques, but the image resolution
is limited. The volumetric CT (VCT) is a new ultra-high
isotropic spatial resolution (150 × 150 × 150 μm) scanner that
has not yet been evaluated in bone formation studies but would
prove to be an invaluable tool for this application. A parallel
development of digital flat-panel detectors for conventional X-
ray and mammography has provided ultra-high spatial resolu-
tion 2D images. A VCT scanner combines the advances in CT
with digital flat-panel detector technology. Unlike micro-CT,
VCT is suitable for in vivo imaging of large animal and eventual
human research. High-resolution, three-dimensional imaging
and quantitative capabilities of this new method can help
identify new developing bone in early stages, which does not
have high radiopacity and therefore cannot be detected using
conventional X-ray techniques.
The primary focus of this research is to optimize bone
formation in vitro by evaluating different dynamic culture
conditions, using a combination of osteoinductive material (β-
TCP) in the form of a 3DP scaffold and hydrogels as carriers for
differentiated MSCs.
Materials and methods
Mesenchymal stem cells
Bone marrow was aspirated from multiple sites of the iliac crests of 7-
month-old Yucatan mini pigs. The aspiration syringe contained Heparin (100 U/
ml bone marrow) to prevent clotting. Bone marrow was plated in cell culture
flasks (Corning Inc., Corning, NY), and adherent cells were expanded for 10–12
days, using growth medium consisting of DMEM supplemented with 100 U/ml
penicillin, 100 μg/ml streptomycin and 10% FBS. MSCs were passaged at 2:3
ratio when cells were 80% confluent. After a sufficient number of cells was
reached, they were cultured in osteogenic (OS) differentiation medium for 6
days. For osteogenic differentiation, growth medium was supplemented with
100 nM dexamethasone, 50 μg/ml ascorbic acid and 10 mM β-glycerolpho-
sphate (all from Sigma-Aldrich Co., St. Louis, MO).
Hydrogels in static conditions
Four hydrogels were evaluated for their ability to provide uniform
distribution of cells, support cell viability and growth and formation of bone-
like tissue. Differentiated MSCs were suspended in hydrogels at a concentration
of 2 × 106 cells/ml hydrogel [25] and plated in 6-well plates (BD Labware,
Franklin Lakes, NJ).
Fibrin glue
Differentiated MSCs were dispersed in 0.113 mg/ml porcine fibrinogen [26]
(Sigma-Aldrich Co.) in OS medium, and 1 ml of cell suspension was plated in
each well of a 6-well plate. Nearly instantaneous gelation occurred after 1 ml
(100 U) of thrombin (Sigma-Aldrich Co.) was added to each well.
Alginate hydrogel
3% Hydrogel was made by dissolving ultra pure alginate powder,
containing a high percentage of β-D mannuronic acid (Pronova Biomedi-
38 (2006) 555–563cals, Brakerǿya, Norway) in OS medium. MSCs were dispersed in the
hydrogel, and 2 ml of the suspension was placed into each well of a 6-well
plate. Instantaneous gelation was achieved after adding 1 ml of 100 mM

CaCl2 (Sigma-Aldrich Co.). Ungelled liquid was aspirated from the surface
of the hydrogel.
Pluronic F127 hydrogel
30% Hydrogel was made by stirring pluronic F127 (Sigma-Aldrich Co.)
with OS medium overnight at 4°C. MSCs were dispersed in liquid hydrogel on
ice, and 2 ml was placed in each well of a 6-well plate. Gelation occurred after
incubation at 37°C for 1 min.
Collagen I hydrogel
Collagen I hydrogel (Cellagen, ICN Biomedicals Inc., Aurora, OH) was
mixed on ice according to the manufacturer's specifications using 5× OS
medium. MSCs were mixed with liquid hydrogel on ice, and 2 ml was placed
cell morphology and evaluate newly formed tissue. Toluidine blue for
pericellular proteoglycan and alkaline phosphatase staining for expression on
cells was used to assess bone formation.
Radiological analysis
Two radiological imaging techniques were used to evaluate the newly
formed tissue at 6 weeks: X-ray images were taken from each specimen at 7 keV
for 7 s using conventional technique. For ultra-high-resolution VCT scanning,
the samples were imaged at 10 mA, 80 kV, 100 mA and 25 × 25 × 18 cm field-
of-view (Siemens, Germany). Radiological density of the samples was assessed
by measuring Houndsfield density units (HU). Mean value for four separate
557C. Weinand et al. / Bone 38 (2006) 555–563in each well of a 6-well plate. Gelation occurred after 1-min incubation at
37°C.
After differentiated MSCs were suspended in hydrogels and placed in 6-
well plates, 1 ml of OS medium was added to each well, and medium was
changed every second day thereafter. The hydrogel/cell suspensions were
cultured for 6 weeks in standard incubator conditions (37°C, 5% CO2). At 1, 2,
4 and 6 weeks, samples were fixed in 10% phosphate-buffered formalin and then
embedded in paraffin. Sections were stained with hematoxylin and eosin (H and
E) and von Kossa stain (Master Tech. Inc. Lodi, CA) to examine calcium
deposition.
β-TCP scaffolds
Microporous (100–250 μm) β-TCP scaffolds were generously provided by
Therics Inc. (Princeton, NJ) (Fig. 1). The scaffolds were made using 3DP
technique and were 2 × 2 × 0.7 cm (total volume of 2.8 cm3) with 2 cm × 2
mm × 2-mm channels running through the scaffold. The channels formed a
matrix of 16 cubicles, with each cubicle having a total volume of 28 mm3. This
scaffold design allows hydrogel penetration into the scaffold and may improve
delivery of nutrients and oxygen.
Suspensions of cells in hydrogels were prepared as described above,
and 2.8 cm3 of liquid hydrogel/cell suspension was placed onto each
scaffold. Scaffolds without hydrogels or cells served as controls. All
scaffolds were cultured free floating for 6 weeks in an incubator at 37°C
and 5% CO2 under either of two dynamic conditions: ROB and DO. For
ROB culture, 50-ml gas-permeable conical tubes [18] (BD Labware) were
filled with 40 ml OS medium and rotated at six RPM in a roller apparatus
[18]. For DO culture, gas-permeable 50-ml tubes with 35 ml OS medium
were secured on a Thermolyne RotoMix Type 50800 horizontal plate shaker
(Barnstead, Dubuque, IA) set at two RPM. Medium was changed every
second day.
Histological analysis
A 1 × 0.5 × 0.7-cm piece of construct was harvested sterilely at 1, 2, 4 and 6
weeks. Each sample contained at least two cubicles, which were evaluated
individually. The samples were fixed in 10% phosphate-buffered formalin and
embedded in paraffin. The sections were stained with H and E to demonstrateFig. 1. (a) Three-dimensional printed β-TCP scaffold, 2 × 2 × 0.7 cm with 4-mm2
microporous structure of the scaffold.cubicles in each sample was calculated, allowing for the comparison of the
radiological density of different samples.
Biomechanical compression testing
Compression testing of the specimens was performed using a Texture
Analyzer TA-XT Plus (Texture Technologies, Scarsdale, NY). Unconstrained
uniaxial compression was applied, while compressive force and displacement
were recorded after the probe tip contacted the sample. Experiments were run
until the 3.5-kg limit of the load cell was reached. The geometry of the samples
made measuring material moduli difficult. To simply compare samples, we
defined a relative stiffness measure as follows. We defined a uniaxial stiffness as
K = dF/dx, where F is the force and x is the displacement. The stiffness was
calculated at each data point, and an average stiffness Kave was taken as the
average of the stiffness values in the range 0 b x b 1.5 mm. To remove the effects
of geometry and simply compare material properties, we defined a relative
stiffness for each sampleKrel =Kave/Kcontrol, whereKcontrol is the average stiffness
of the control. Thus, Krel is a measure to simply compare the stiffnesses
between the various samples.
We compared samples by referring to one sample that required higher force
to compress as being ‘stiffer’ than a sample that required less force to compress
the same distance.
Gene transcription analysis
For transcription analysis, total RNA was purified from samples with
RNA STAT 60 according to the specifications of the manufacturer (Tel Test,
Friendswood, TX). RNA was quantified by spectrophotometric techniques
and integrity verified by visualization with ethidium bromide staining in a
formaldehyde gel [27]. One microgram of each sample was converted to
cDNA in a 30 μl reaction containing 1× PCR buffer, 5 mM dNTPs, 0.5 μg
random hexamer primers and 200 units MMLV RT (Promega, Madison,
WI). The reaction was subsequently heat-inactivated and diluted to 100 μl
with water. One-microliter aliquots were used in each 50 μl RT-PCR
reaction, using gene-specific primers and 1 mM dUTP–biotin. After 22
cycles, 10-μl aliquots were fractionated on a 3% agarose gel (1% agarose,
2% NuSieve GTG agarose), blotted onto Amersham Hybond N+ nitrocel-
lulose filter and probed with horseradish peroxidase-conjugated avidinchannels, forming 28-mm3 cubicles. (b) Scanning electron microscopy of the

(Sigma-Aldrich Co). The reaction was visualized using luminol substrate
(Pierce Chemical, Rockford, IL) and autoradiography. The following primers
were utilized:
ALP (ATGGCCTGAACCTCATTGAC, AAGCCTTTGGGGTTCTTGAT),
BMP-2 (AGGAGGGAGGTGTGTGAGC, GATGGGGACCTTACACT-
TGC),
Osteonectin (ON) (TTCCCCTCCTCCTGTTCTC, ACCCACCCGTCACT-
AAGACA),
Osteopontin (OPN) (GGGCCTCACAGTTGTTTGAT,CCGCAGGATTCA-
CTATGGTT),
GAPDH (CTCACTGGCATGGCCTTCCG, ACCACCCTGTTGCTGTAG-
CC).
Native swine bone was used as a control for biomechanical testing and
transcription analysis. The films were digitized, and transcription levels for
native swine bone were assigned a relative value of 1.
Statistical correlation analysis
Changes in the Houndsfield Units (HU) of each construct were correlated
with biomechanical strength and gene transcription levels. The Pearson, Kendall
and Spearman correlation coefficients were calculated, using MatLab 7.0 (The
MathWorks, Inc., Natick, MA).
Results
Four hydrogels were evaluated for their ability to support
bone formation by differentiated MSCs in static conditions and
on 3DP β-TCP scaffolds in two dynamic culture conditions.
Hydrogels in static conditions
The differentiated MSCs were dispersed evenly within all
hydrogels, and similar changes in cell morphology were
noted. During the first 2 weeks, cells had spindle-shaped
appearance with eosinophilic cytoplasm. After 2 weeks, cells
changed their morphology to large round osteoblast-like
cells with basophilic cytoplasm. Cells were embedded in
newly synthesized matrix where they formed new tissue.
The tissue had bone-like appearance, stained positively with
toluidine blue and cells positive for alkaline phosphatase.
Calcification of the new tissue was demonstrated by von
Kossa staining. In fibrin glue hydrogel, bone-like tissue was
present already at 4 weeks and most closely resembled
native bone histologically (Figs. 2a, b). By 6 weeks, there
was bone-like tissue present in all hydrogels. Newly formed
tissue in fibrin glue and collagen I most closely resembled
native bone. Collagen I hydrogel shrank significantly during
the observation period, leading to the compaction of new
bone-like tissue (Fig. 2c). In the alginate hydrogels,
chondron-like round cell doublets, similar to those in native
cartilage, were embedded in matrix. This matrix stained
positive with toluidine blue similar to cartilage (Fig. 3). The
cartilage-like tissue was surrounded by bone-like matrix.
Pluronic F127 was the only hydrogel that degraded quickly
after 1 week. New bone-like tissue was present at 6 weeks
but appeared only in islands in this specimen.
558 C. Weinand et al. / Bone 38 (2006) 555–563Fig. 2. (a) Fibrin glue hydrogel after 4 weeks in culture with embedded differentiatedM
Native pig bone, fibula. Magnification, 20 and 200×, H and E staining. (c) Collagen I
mass over time. Magnification, 20 and 200×, H and E staining.SCs, resembling native bone. Magnification, 20 and 200×, H and E staining. (b)
hydrogel after 4 weeks with dispersed differentiated MSCs; the hydrogel has lost

oneHydrogels in β-TCP scaffolds in dynamic conditions
β-TCP with fibrin glue
Similar histological results were found in fibrin glue
specimens cultured in DO and ROB conditions over 6
weeks. The hydrogel degraded faster in the channels of the
scaffold under DO conditions (2 weeks) than under ROB (4
weeks). We observed multilayers of elongated fibroblast-like
cells lining the channels at these time points. H and E and
toluidine blue staining showed deposition of extracellular
bone-like matrix on the surface of the scaffold first. By 6
weeks, the matrix had spread throughout the scaffold. The cells
changed their morphological appearance to spindle-shaped
basophilic osteocyte-like and cuboid large basophilic osteo-
blast-like cells, surrounded by new matrix. Calcium deposi-
tions in the new matrix around these cells were observed after
4 weeks in the channels and within the scaffold. Radiological
examination at 6 weeks did not show radiodensity in the
channels, whereas VCT scanning revealed an increase of 78.3
HU under the DO condition and 63.7 HU under the ROB
Fig. 3. Alginate hydrogel at 6 weeks in culture with dispersed differentiated
MSCs, showing bone-like tissue formation surrounding cartilage-like tissue.
Magnification, 20 and 200×, H and E staining.
C. Weinand et al. / Bcondition compared to the controls (Fig. 6). An absolute
average density of 130 HU within the channels was measured,
which corresponded well with the density in native bone (250
HU). Formation of bone-like structures was also confirmed by
the presence of gene expression of bone proteins. The results
were similar in both culture conditions (Fig. 7, Table 1). In
overall structure, the cellular samples were stiffer than the
controls. Specimens cultured in DO conditions had higher
mechanical stability (Fig. 8).
β-TCP with alginate
Formation of bone-like tissue was similar in both culture
conditions. Fibroblast-like cells with eosinophilic cytoplasm
were the primary cells dispersed within the entire scaffold at
1 and 2 weeks. At 6 weeks, the hydrogel was not
completely degraded, and extracellular matrix was formed
in the channels and on the surface of the scaffold,
surrounding large round basophilic osteoblast-like cells.
However, there was less bone-like tissue found than in the
scaffolds with fibrin glue. Multiple layers of cells werelining the surface of the scaffold and the channels.
Radiological examination at 6 weeks showed no radiopacity,
but VCT scanning detected HU values within the channels
similar to the densities of the hydrogel itself (results not
shown). DO culture conditions resulted in the formation of
more mature bone-like tissue according to VCT scanning
(Fig. 6), gene expression of bone-specific protein (Fig. 7,
Table 1) and mechanical testing (Fig. 8). Biomechanically,
alginate samples were stiffer than all other samples.
β-TCP with pluronic F127
Histological observations were similar for both culture
conditions. The hydrogel degraded quickly after 1 week in
culture. Within the channels, no hydrogel was present after
this time. Mainly spindle-shaped fibroblast-like cells and a
few large round basophilic osteoblast-like cells were present
in new bone-like matrix primarily on the surface of the
scaffold. These cells and matrix were present to a lesser
degree within the scaffold pores and channels at 6 weeks.
Toluidine blue and alkaline phosphatase staining were
positive. Radiological examination did not show radiopacity.
Additionally, VCT scanning did not show densities in the
range of bone as well (Fig. 6). Further expression of bone-
specific mRNA could not be detected (Fig. 7, Table 1). The
pluronic F127 sample cultured in ROB conditions had also
lower stiffness than both fibrin glue and collagen I specimens
(Fig. 8).
β-TCP with collagen I
There was no difference in formation of new bone-like
tissue in DO and ROB culture conditions. No loss of mass
of the hydrogel within the scaffold and compaction of cells
and tissue was observed during the 6-week period as
opposed to our observations in static culture. Collagen
hydrogel with dispersed cells was present throughout the
entire scaffold from week 1. Cells had fibroblast-like
spindle-shaped appearance at 2 weeks. The cellular mor-
phology changed to mainly large round basophilic osteo-
blast-like cells by 4 weeks. The cells surrounded themselves
with new matrix. Positive staining with toluidine blue,
alkaline phosphatase and von Kossa stains was detected at
this time point. At 6 weeks, new bone-like matrix was
mostly found on the surface of the entire scaffold. Within
the matrix, spindle-shaped basophilic osteocyte-like cells and
large round basophilic osteoblast-appearing cells stained
positively with toluidine blue and alkaline phosphatase
(Figs. 4a–d). New matrix was also found within the scaffold
and stained positively for calcification by von Kossa
staining. Radiological examination showed radiopacities
within the channels and bone-like tissue formation was
confirmed by VCT scanning (Figs. 5a, b). Houndsfield unit
densities were in the range of native bone (N250 unit
densities), and were higher in DO conditions (Fig. 6). In
overall structure, the density of the channels under the two
55938 (2006) 555–563different culture conditions increased significantly after 6
weeks compared to the controls. In case of DO culture
conditions, an increase of 157.3 HU and in case of ROB an

560 C. Weinand et al. / Bone 38 (2006) 555–563increase of 113.4 HU were observed. These findings were
strongly supported by the corresponding results of the
increased biomechanical strength of the scaffolds and the
Fig. 5. Collagen I hydrogel on β-TCP scaffold. Bone formation shown by
regular X-ray (a) and VCT scanning (b) after 6 weeks. Densities are visible
within the channels, indicated by VCT(lines).
Fig. 4. Bone formation on scaffold after 6 weeks, collagen I hydrogel under DO (
conditions. Staining toluidine blue, magnification, 40 and 200×.formation of bone-like structures by the presence of gene
expression of bone proteins.
Gene expression for bone proteins was similar in both
conditions and half that of native bone (Fig. 7, Table 1).
Fig. 6. Houndsfield units, measured in hydrogels under DO (dynamic
oscillating) and ROB (rotating bioreactor system) conditions. Collagen I
(Coll) has the highest values, while Pluronic F 127 (Pl) has negative density
values. FG—fibrin glue, Alg—alginate.
a) and ROB (b) conditions. Fibrin glue specimens under DO (c) and ROB (d)

component of native bone, and its abundance in hydrogels
surrounding differentiated MSCs proved to be beneficial for
bone formation in vitro.
It has previously been demonstrated that hydrogels support
bone formation in vitro [32–38], however, the lack of adequate
initial strength of hydrogel/cell constructs requires additional
mechanical support for implantation in weight-bearing tissue.
Scaffolds can support bone formation and possess adequate
strength. Some materials can be used for 3D printing, which
permits custom fabrication of scaffolds corresponding to the
561one 38 (2006) 555–563Mechanical compression showed slightly greater stiffness of
these samples than samples with Fibrin glue and acellular
controls. Samples cultured in DO conditions had higher
compression values (Fig. 8).
Statistical correlation analysis
Higher HU corresponded to increased biomechanical
strength of the scaffolds (Spearman correlation = 0.70, Pear-
son = 0.62, Kendall = 0.59). This association was robust across
the different correlation coefficients. A high correlation between
HU and gene transcription values was found (Spearman
correlation = 0.95, Pearson = 0.92, Kendall = 0.87). Also
robust associations were found between biomechanical stiffness
values and gene transcription (Spearman correlation = 0.51,
Pearson = 0.48, Kendall = 0.50). Bivariate correlations were
Fig. 7. Expression of genes for bone-specific proteins in hydrogels under two
different culture conditions: DO—dynamic oscillating, ROB—rotating biore-
actor system, Bone Proteins: ALP—alkaline phosphatase, BMP-2—bone
morphogenic protein 2, ON—osteonectin, OPN—osteopontin.
C. Weinand et al. / Btested non-parametrically using Spearman rank correlation and
Kendall tau b statistics. Spearman rank correlations showed:
biomechanical stiffness–HU = 0.69 (P b 0.06), biomechanical
stiffness–gene transcription values = 0.59 (P b 0.13), HU–gene
transcription values = 0.95 (P b 0.0003). Kendall tau b statistics
revealed biomechanical stiffness–HU = 0.57 (P b 0.06),
biomechanical stiffness–gene transcription values = 0.49
(P b 0.13), HU–gene transcription values = 0.87 (P b 0.005).
Discussion
We evaluated the ability of several hydrogels to support bone
formation by differentiated MSCs in static conditions and in two
dynamic culture conditions on 3DP β-TCP scaffolds. In our
experiments, we used MSCs because of their high proliferation
rate and ability to differentiate into osteocytes in culture
[5,8,28–31] and in hydrogels [32,33]. We first tested the ability
of collagen I hydrogel to support bone formation in static
conditions in comparison with fibrin glue, alginate and pluronic
F127, all currently used for bone tissue engineering. Bone-like
tissue was present in all hydrogels, however, it was best
supported by collagen I hydrogel. Collagen I is an essentialbone defect that needs to be bridged [9,16]. However, elaborate
seeding and culture conditions [17–19] are needed to achieve
uniform cellular distribution, sustain cell viability and provide
nutrients for tissue formation. Hydrogels can facilitate delivery
and distribution of cells within porous scaffolds. We therefore
evaluated the formation of bone tissue on 3DP β-TCP scaffolds
with differentiated MSCs suspended in four different hydrogels
cultured in two dynamic culture conditions. We chose β-TCP as
it is osteoinductive and osteoconductive and provides adequate
initial strength [9,10]. Additionally, it can be used in 3-
dimensional printing to achieve a custom-made shape.
We demonstrated that uniform cellular distribution within the
entire scaffold can be achieved with all four hydrogels and
confirmed that formation of new bone matrix depends on the
type of hydrogel and its degradation rate [34–38]. Fibrin glue
degraded within 2 to 3 weeks, which allowed for the earliest
formation of bone-like tissue at 4 weeks. Alginate provided a
substrate for suspended cells for more than 6 weeks, mostly on
the surface of the hydrogel [39], but leading to delayed
formation of bone-like tissue during 6 weeks. In addition, we
observed cartilage-like tissue, consistent with previous observa-
tions that cartilage growth is supported by alginate [37,40].
Pluronic F127 degraded rapidly within 1 week in culture,
resulting in very poor bone formation as demonstrated
histologically, radiologically and by gene transcription analysis.
Tissue most closely resembling native bone was detected in
collagen I hydrogel, confirming the advantage of this hydrogel.
A 6-week endpoint was chosen for radiological evaluation
and gene expression of bone proteins since bone fracture
healing is normally completed at this time point and the newly
formed matrix bone is radiologically visible. Our X-ray imaging
study revealed that bone formation was only present in collagen
I samples, demonstrating the advantage of this hydrogel. High-
resolution VCT scanning revealed up to 270 HU units in
Table 1
Quantitative analysis of gene expression
Collagen
DO
Collagen
ROB
Alginate
DO
Fibrin
DO
Fibrin
ROB
Bone
0.51 0.72 0.11 0.08 0.1 1
0.63 0.61 0.4 0.38 0.45 1
1.2 1.2 0.12 0.15 0.2 1
0.71 0.67 0.22 0.31 0.41 1
0.85 0.95 0.85 1.1 0.95 1Tissue engineered bone in collagen I hydrogel has values closest to native bone.
The values represent from top to bottom: alkaline phosphatase, bone
morphogenic protein 2, osteonectin, osteopontin, GADPH.

sam
alue
onecollagen I and fibrin glue samples, which is comparable to
native bone. No densities within the range of bone were found
in alginate and pluronic F127 samples.
Transcription analysis of genes important for bone formation
confirmed the histologic and radiologic results. Alginate and
pluronic F127 only weakly supported bone formation and had
the lowest transcription levels. The highest gene transcription
levels encoding bone proteins were found in the samples using
collagen I, followed by fibrin glue. These results were
supported by the robust association across the different
correlation coefficients. This might be a consequence of bone
consisting of collagen I, and that fibrin glue is a product in bone
formation after fractures.
Although collagen hydrogel showed the best bone
formation histologically, the highest biomechanical compres-
sion values were found in the samples made with alginate.
Fig. 8. Unconstrained uniaxial compression displacement graph, the values of the
for alginate under DO conditions, whereas fibrin glue and collagen I show lesser v
to have higher compressibility values than under ROB conditions.
562 C. Weinand et al. / BThis may be due to the longer degradation time of alginate
[40] and different biomechanical properties of this hydrogel.
High compression values were also found in fibrin glue and
collagen I samples. All cell-containing samples had higher
compression values than acellular samples at 6 weeks,
suggesting that the newly formed tissue contributed to
mechanical stability.
There was no difference in formation of bone-like tissue
histologically in both dynamic conditions; however, samples
cultured in DO had higher radiological densities. Both dynamic
culture conditions resulted in similar quantitative expression for
bone marker proteins in matching hydrogels, although samples
under DO conditions had slightly higher values. All samples in
DO conditions had higher stiffness values than corresponding
samples with the same hydrogel in ROB. This suggests that
linear fluid shear stress generated in DO culture may be more
beneficial for bone formation than rotational fluid shear stress in
ROB [41].
We demonstrated that higher HU corresponds to increased
biomechanical strength of the scaffolds (correlation between
0.57–0.61). This association was robust across the different
correlation coefficients. A high correlation between HU andgene transcription values was found (0.87–0.95, Spearman
0.95, P b 0.0003), demonstrating that an increase in HU is a
true marker of osteoblastic activity. In all specimens, one
could visually appreciate the change in HU in the VCT
images when the bone growth was successful. Higher
correlation between HU numbers and gene transcription
values indicate that VCT imaging is a valuable method for
detecting formation of new bone in-vitro. The imaging
results were concordant with biomechanical testing and
protein transcription and expression.
This study demonstrates that hydrogels in combination with
3DP β-TCP scaffolds support bone formation in vitro. The best
bone-like tissue was formed using collagen I hydrogel,
confirmed by histological and radiological examination and by
the level of gene expression important for bone formation. DO
culturing resulted in denser bone, as shown by higher stiffness
ples have been normalized to control groups. The highest compression values are
s under DO and ROB conditions. Samples cultured under DO conditions proved38 (2006) 555–563and volumetric CT. Although only one sample of each hydrogel
was tested biomechanically, the quality of new bone was
confirmed by histologic results, radiologic findings and
transcription of genes important for bone formation. In
conclusion, the combination of differentiated MSCs in collagen
hydrogel with 3DP β-TCP scaffolds can be used to engineer
tissue that meets individual needs for bone reconstruction.
Future work will concentrate on further evaluation of bone
formation in hydrogels combined with scaffolds in an in vivo
model.
Acknowledgments
We would like to thank Therics Inc., NJ, for generously
providing 3DP β-TCP scaffolds.
References
[1] Gazdag AR, Lane JM, Glaser D, Forster RA. Alternatives to autogenous
bone graft: efficacy and indications. J Am Acad Orthop Surg 1995;3
(1):1–8.

one[2] Chang DW, Weber KL. Segmental femur reconstruction using an
intercalary allograft with an intramedullary vascularized fibula bone flap.
J Reconstr Microsurg 2004;20(3):195–9 [Apr].
[3] Tai CY, Strande LF, Eydelman R, Sheng X, VanTran JL, Matthews
MS, et al. Absence of graft-versus-host disease in the isolated
vascularized bone marrow transplant. Transplantation 2004;77
(2):316–9 [Jan 27].
[4] Petty W, Spanier S, Shuster JJ. Prevention of infection after total joint
replacement. Experiments with a canine model. J Bone Jt Surg Am
1988;70(4):536–9.
[5] Montjovent MO, Burri N, Mark S, Federici E, Scaletta C, Zambelli
PY, et al. Fetal bone cells for tissue engineering. Bone 2004;35
(6):1323–33 [Dec].
[6] Langer R, Vacanti JP. Tissue engineering. Science 1993;260(5110):920–6
[May 14].
[7] Wald HL, Sarakinos G, Lyman MD, Mikos AG, Vacanti JP, Langer R. Cell
seeding in porous transplantation devices. Biomaterials 1993;14(4):270–8.
[8] Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca JD,
et al. Multilineage potential of adult human mesenchymal stem cells.
Science 1999:143–7.
[9] Hinz P, Wolf E, Schwesinger G, Hartelt E, Ekkernkamp A. A new
resorbable bone void filler in trauma: early clinical experience and
histologic evaluation. Orthopedics 2002;25(Suppl 5):s597–600.
[10] Porter BD, Oldham JB, He SL, Zobitz ME, Payne RG, An KN, et al.
Mechanical properties of a biodegradable bone regeneration scaffold.
J Biomech Eng 2000;122(3):286–8 [Jun].
[11] Dean D, Topham NS, Meneghetti SC, Wolfe MS, Jepsen K, He S, et al.
Poly(propylene fumarate) and poly(DL-lactic-co-glycolic acid) as scaffold
materials for solid and foam-coated composite tissue-engineered con-
structs for cranial reconstruction. Tissue Eng 2003;9(3):495–504.
[12] Lieb E, Tessmar J, Hacker M, Fischbach C, Rose D, Blunk T, et al. Poly(D,
L-lactic acid)-poly(ethylene glycol)-monomethyl ether diblock copolymers
control adhesion and osteoblastic differentiation of marrow stromal cells.
Tissue Eng 2003;9(1):71–84 [Feb].
[13] Dhert WJ, Klein CP, Jansen JA, van der Velde EA, Vriesde RC, Rozing
PM, et al. A histological and histomorphometrical investigation of
fluorapatite, magnesiumwhitlockite, and hydroxylapatite plasma-sprayed
coatings in goats. J Biomed Mater Res 1993;27(1):127–38.
[14] Witte F, Kaese V, Meyer-Lindenberg A, Switzer E, Haferkamp H,
Wirth CJ, et al. Degrading magnesium implants increase periosteal and
endosteal bone formation. ORS Paper Nr. 0256, ORS meeting in San
Francisco; 2004.
[15] Sherwood JK, Riley SL, Palazzolo R, Brown SC, Monkhouse DC, Coates
M, et al. A three-dimensional osteochondral composite scaffold for
articular cartilage repair. Biomaterials 2002;23(24):4739–51 [Dec].
[16] Schantz JT, Teoh SH, Lim TC, Endres M, Lam CX, Hutmacher DW.
Repair of calvarial defects with customized tissue-engineered bone grafts:
I. Evaluation of osteogenesis in a three-dimensional culture system. Tissue
Eng 2003;9(Suppl 1):S113–26.
[17] Meinel L, Karageorgiou V, Fajardo R, Snyder B, Shinde-Patil V, Zichner
L, et al. Bone tissue engineering using human mesenchymal stem cells:
effects of scaffold material and medium flow. Ann Biomed Eng 2004;32
(1):112–22 [Jan].
[18] Terai H, Hannouche D, Ochoa E, Yamano Y, Vacanti JP. In vitro
engineering of bone using a rotational oxygen-permeable bioreactor
system. Mater Sci Eng 2002;C20:3–8.
[19] Kreke MR, Goldstein AS. Hydrodynamic shear stimulates osteocalcin
expression but not proliferation of bone marrow stromal cells. Tissue Eng
2004;10(5–6):780–8 [May–Jun].
[20] Lawson MA, Barralet JE, Wang L, Shelton RM, Triffitt JT. Adhesion and
growth of bone marrow stromal cells on modified alginate hydrogels.
Tissue Eng 2004;10(9–10):1480–91 [Sept–Oct].
[21] Simmons CA, Alsberg E, Hsiong S, Kim WJ, Mooney DJ. Dual growth
factor delivery and controlled scaffold degradation enhance in vivo bone
formation by transplanted bone marrow stromal cells. Bone 2004;35
(2):562–9 [Aug].[22] Schantz JT, Hutmacher DW, Lam CX, Brinkmann M, Wong KM, Lim TC,
et al. Repair of calvarial defects with customised tissue-engineered bone
grafts: II. Evaluation of cellular efficiency and efficacy in vivo. Tissue Eng
2003;9(Suppl 1):S127–39.
[23] Fowler EB, Cuenin MF, Hokett SD, Peacock ME, McPherson III JC,
Dirksen TR, et al. Evaluation of pluronic polyols as carriers for grafting
materials: study in rat calvaria defects. J Periodontol 2002;73(2):191–7
[Feb].
[24] Chang SC, Wei FC, Chuang H, Chen YR, Chen JK, Lee KC, et al. Ex vivo
gene therapy in autologous critical-size craniofacial bone regeneration.
Plast Reconstr Surg 2003;112(7):1841–50 [Dec].
[25] Fuchs JR, Pomerantseva I, Ochoa ER, Vacanti JP, Fauza DO. Fetal tissue
engineering: in vitro analysis of muscle constructs. J Pediatr Surg 2003;38
(9):1348–53 [Sep].
[26] Silverman RP, Passaretti D, Huang W, Randolph MA, Yaremchuk MJ.
Injectable tissue-engineered cartilage using a fibrin glue polymer. Plast
Reconstr Surg 1999;103(7):1809–18 [Jun].
[27] Rosen KM, Villa-Komaroff L. An alternative method for the visualization
of RNA in formaldehyde agarose gels. Focus 1990;845(2):23–4.
[28] Bruder SP, Gazit D, Passi-Even L, Bab I, Caplan AI. Osteochondral
differentiation and the emergence of stage-specific osteogenic cell-surface
molecules by bone marrow cells in diffusion chambers. Bone Miner
1990;11(2):141–5 [Nov].
[29] Bruder SP, Fink DJ, Caplan AI. Mesenchymal stem cells in bone
development, bone repair, and skeletal regeneration therapy. J Cell
Biochem 1994;56(3):283–94 [Nov].
[30] Caplan AI. Mesenchymal stem cells. J Orthop Res 1991;9(5):641–50
[Sep].
[31] Nakahara H, Bruder SP, Haynesworth SE, Holecek JJ, Baber MA,
Goldberg VM, et al. Bone and cartilage formation in diffusion
chambers by subcultured cells derived from the periosteum. Bone
1990;11(3):181–8.
[32] Rutherford RB, Gu K, Racenis P, Krebsbach PH. Early events: the in vitro
conversion of BMP transduced fibroblasts to chondroblasts. Connect
Tissue Res 2003;44(Suppl 1):117–23.
[33] Rutherford RB, Moalli M, Franceschi RT, Wang D, Gu K, Krebsbach PH.
Bone morphogenetic protein-transduced human fibroblasts convert to
osteoblasts and form bone in vivo. Tissue Eng 2002;8(3):441–52 [Jul].
[34] Weng Y, Cao Y, Silva CA, Vacanti MP, Vacanti CA. Tissue-engineered
composites of bone and cartilage for mandible condylar reconstruction.
J Oral Maxillofac Surg 2001;59(2):185–90 [Feb].
[35] Alsberg E, Anderson KW, Albeiruti A, Franceschi RT, Mooney DJ. Cell-
interactive alginate hydrogels for bone tissue engineering. J Dent Res
2001;80(11):2025–9 [Nov].
[36] Burdick JA, Mason MN, Hinman AD, Thorne K, Anseth KS. Delivery of
osteoinductive growth factors from degradable PEG hydrogels influences
osteoblast differentiation and mineralization. Control Release 2002;83(1):
53–63 [Sep 18].
[37] Shin H, Zygourakis K, Farach-Carson MC, Yaszemski MJ, Mikos AG.
Modulation of differentiation and mineralization of marrow stromal cells
cultured on biomimetic hydrogels modified with Arg-Gly-Asp containing
peptides. J Biomed Mater Res 2004;69A(3):535–43 [Jun 1].
[38] Temenoff JS, Park H, Jabbari E, Conway DE, Sheffield TL, Ambrose
CG, et al. Thermally cross-linked oligo(poly(ethylene glycol) fumarate)
hydrogels support osteogenic differentiation of encapsulated marrow
stromal cells in vitro. Biomacromolecules 2004;5(1):5–10 [Jan–Feb].
[39] Chung TW, Yang J, Akaike T, Cho KY, Nah JW, Kim SI, et al. Preparation
of alginate/galactosylated chitosan scaffold for hepatocyte attachment.
Biomaterials 2002;23:2827–34.
[40] Wang L, Shelton RM, Cooper PR, Lawson M, Triffitt JT, Barralet JE.
Evaluation of sodium alginate for bone marrow cell tissue engineering.
Biomaterials 2003;24(20):3475–81 [Sep].
[41] Mow VC, Ratcliffe A, Rosenwasser MP, Buckwalter JA. Experimental
studies on repair of large osteochondral defects at a high weight bearing
area of the knee joint: a tissue engineering study. J Biomech Eng 1991;113
56338 (2006) 555–563C. Weinand et al. / B(2):198–207.