TGF-β1-Modified Hyaluronic Acid/Poly(glycidol) Hydrogels for Chondrogenic Differentiation of Human Mesenchymal Stromal Cells

Article abstract of DOI:10.1002/mabi.201700390

In cartilage regeneration, the biomimetic functionalization of hydrogels with growth factors is a promising approach to improve the in vivo performance and furthermore the clinical potential of these materials. In order to achieve this without compromising network properties, multifunctional linear poly(glycidol) acrylate (PG-Acr) is synthesized and utilized as crosslinker for hydrogel formation with thiol-functionalized hyaluronic acid via Michael-type addition. As proof-of-principle for a bioactivation, transforming growth factor-beta 1 (TGF-β1) is covalently bound to PG-Acr via Traut's reagent which does not compromise the hydrogel gelation and swelling behavior. Human mesenchymal stromal cells (MSCs) embedded within these bioactive hydrogels show a distinct dose-dependent chondrogenesis. Covalent incorporation of TGF-β1 significantly enhances the chondrogenic differentiation of MSCs compared to hydrogels with supplemented noncovalently bound TGF-β1. The observed chondrogenic response is similar to standard cell culture with TGF-β1 addition with each medium change. In general, multifunctional PG-Acr offers the opportunity to introduce a range of biomimetic modifications (peptides, growth factors) into hydrogels and, thus, appears as an attractive potential material for various applications in regenerative medicine.

Full text of DOI:10.1002/mabi.201700390

Full PaPer
Biomimetic Hydrogel
www.mbs-journal.de
TGF-β1-Modified Hyaluronic Acid/Poly(glycidol) Hydrogels
for Chondrogenic Differentiation of Human Mesenchymal
Stromal Cells
Thomas Böck, Verena Schill, Martin Krähnke, Andre F. Steinert, Jörg Tessmar,
Torsten Blunk,* and Jürgen Groll*
fluent joint movement in weight bearing
areas;[1] however, the self-healing capactiy
In cartilage regeneration, the biomimetic functionalization of hydrogels with
growth factors is a promising approach to improve the in vivo performance
and furthermore the clinical potential of these materials. In order to achieve
this without compromising network properties, multifunctional linear
of cartilage is limited.[2] Thus, as a result
of a severe trauma or as a consequence
of high loads in the joint during lifetime
potentially associated with degenerative
poly(glycidol) acrylate (PG-Acr) is synthesized and utilized as crosslinker for
hydrogel formation with thiol-functionalized hyaluronic acid via Michael-
type addition. As proof-of-principle for a bioactivation, transforming growth
factor-beta 1 (TGF-β1) is covalently bound to PG-Acr via Traut’s reagent which
does not compromise the hydrogel gelation and swelling behavior. Human
mesenchymal stromal cells (MSCs) embedded within these bioactive hydrogels
show a distinct dose-dependent chondrogenesis. Covalent incorporation
of TGF-β1 significantly enhances the chondrogenic differentiation of MSCs
compared to hydrogels with supplemented noncovalently bound TGF-β1. The
observed chondrogenic response is similar to standard cell culture with TGF-β1
addition with each medium change. In general, multifunctional PG-Acr offers
the opportunity to introduce a range of biomimetic modifications (peptides,
growth factors) into hydrogels and, thus, appears as an attractive potential
material for various applications in regenerative medicine.
processes, the cartilage tissue becomes
damaged and loses its mechanical integ-
rity leading to the development of pain
and nonhealing cartilage defects. While
more than 20 million people in the US
suffer from osteoarthritis[3] and the World
Health Organization (WHO) expects oste-
oarthritis to be the fourth most common
cause leading to disability in 2020,[4] the
clinically available treatment options are
still not completely satisfying. Therefore,
in order to develop alternative cell-based
treatment options in trauma and degen-
erative disease, cartilage is a target for var-
ious tissue engineering approaches.[5]
Hydrogels consisting of hyaluronic acid
(HA) have been shown to be promising for
cartilage tissue engineering approaches,
since HA is amenable to chemical functionalization,[6] and its
presence enhances cartilage-specific extracellular matrix (ECM)
synthesis.[7] Prestwich and co-workers have used thiol-functional-
ized hyaluronic acid (HA-SH) and linear[8] or four-armed[9] end-
functionalized polyethylene glycol (PEG) acrylates for hydrogel
formation. The polymer precursors were crosslinked in situ via
pH-dependent Michael addition,[10] furnishing a thioether linkage.
Linear poly(glycidol) (PG) as a biocompatible and water-soluble
structural analog[11] of PEG exhibits a hydroxy methylene group at
each repeating unit, which itself can undergo multiple function-
alizations. In comparison to the widely used end-modified PEG,
side chain-modified PG can provide a much higher crosslinking
density and more versatile options for biomimetic functionaliza-
tion, without negatively affecting hydrogel formation. However,
despite the apparent potential advantages, PG has not yet been
commonly applied for cartilage engineering until now.
1. Introduction
Articular cartilage has important functions in the joint
including protection of bone surfaces and insuring smooth and
Dr. T. Böck, M. Krähnke, Prof. T. Blunk
Department of Trauma, Hand, Plastic and Reconstructive Surgery
University of Würzburg
Oberdürrbacher Str. 6, 97080 Würzburg, Germany
E-mail: blunk_t@ukw.de
V. Schill, Dr. J. Tessmar, Prof. J. Groll
Department of Functional Materials for Medicine and Dentistry
and Bavarian Polymer Institute
University of Würzburg
Pleicherwall 2, 97070 Würzburg, Germany
E-mail: juergen.groll@fmz.uni-wuerzburg.de
Prof. A. F. Steinert
Orthopedic Center for Musculoskeletal Research
University of Würzburg
Transforming growth factor-beta 1 (TGF-β1), a disulfide-
linked protein homodimer of ≈25.6 kDa, controls, together
with other growth factors, the chondrogenic differentiation of
mesenchymal stromal cells (MSCs).[12] The addition of TGF-
β1 as cell culture supplement facilitates MSC chondrogenesis
reproducibly and is well acknowledged as a crucial factor for
Brettreichstr. 11, 97074 Würzburg, Germany
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/mabi.201700390.
DOI: 10.1002/mabi.201700390
©
1700390 (1 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
the sustainability of chondrogenic differentiation.[13] Studies in
which TGF-β1 or its latent form has been covalently bound to
hydrogels led to improved cell performance in comparison to
exogenous supply of the growth factor.[14–16] However, so far,
hydrogels with covalently bound TGF-β1 have not been directly
compared to hydrogels with the same amount of noncovalently
bound TGF-β1, and thus just mixed into the hydrogel. Covalent
incorporation of TGF-β1 may omit the necessity for repeated
administration in vivo, possibly enhancing the clinical potential
of hydrogels for cartilage regeneration.
In this study, we evaluated multifunctional PG for crosslinking
HA-based hydrogels and covalent binding of TGF-β1 and subse-
quently assessed chondrogenic differentiation of MSCs. TGF-β1
was chosen as promising model substance, due to its crucial role
in in vitro chondrogenesis. A similar modification with TGF-β1
has been shown also in previous studies[15] to be suitable, due
to the distinct chondrogenic effects, to evaluate whether the bio-
functionalization was successful or not. Specifically, a multifunc-
tional PG acrylate (PG-Acr) was synthesized and HA-SH-based
hydrogels crosslinked with PG-Acr via Michael addition were
established. The hydrogels were characterized with regard to
rheological properties, swelling behavior, and degradation. Sub-
sequently, varying amounts of TGF-β1 were covalently bound to
PG-Acr and incorporated into the hydrogels; constructs with non-
covalently bound growth factor or TGF-β1 exogenously supple-
mented to the medium served as control. MSC chondrogenesis
was extensively evaluated by histology, immunohistochemistry
(IHC), quantitative biochemical assays, and quantitative real-
time polymerase chain reaction (qRT-PCR), with regard to quali-
tative and quantitative ECM deposition and gene expression.
CA, USA), silicone grease (medium viscous, Bayer, Leverkusen,
Germany), glass cylinders (inner diameter 6 mm, length 8 mm,
Hilgenberg) were used. Ultrapure water (H2O, resistance
>18.2 MΩ cm) was purified with Sartorius Arium Pro (Sartorius
AG, Goettingen, Germany).
2.1.3. Cell Culture
l-Ascorbic acid 2-phosphate sequimagnesium salt hydrate,
basic fibroblast growth factor (bFGF; BioLegend, London, UK),
dexamethasone, Dulbecco’s Modified Eagle’s Medium (DMEM)
high glucose 4.5 g L−1, Dulbecco’s Modified Eagle’s Medium/
Ham’s F-12 (DMEM/F12) (Thermo Scientific, Waltham, USA),
fetal bovine serum (FBS) (Thermo Scientific, Waltham, USA),
ITS+ Premix (Corning; NY, USA), Live/Dead cell staining kit
(PromoKine, Heidelberg, Germany), penicillin–streptomycin
(PS; 100 U mL−1 penicillin, 0.1 mg mL−1 streptomycin) (Thermo
Scientific, Waltham, USA), phosphate-buffered saline (PBS; Life
Technologies, Karlsruhe, Germany), l-proline, sodium pyru-
vate, and 0.25% trypsin-ethylenediaminetetraacetic acid (EDTA)
(Thermo Scientific, Waltham, USA) were used as received.
As primary antibodies, anti-aggrecan 969D4D11 (Thermo
Scientific, Waltham, USA), anti-collagen I ab34710 (Abcam,
Cambridge, UK), anti-collagen II II-4C11 (Acris, Herford,
Germany), and anti-collagen
Waltham, USA) were used.
X X53 (Thermo Scientific,
As secondary antibodies, donkey anti-mouse (Cy3; Dianova,
Hamburg, Germany), and goat anti-rabbit (Alexa Fluor 488;
Dianova, Hamburg, Germany) were used. Furthermore, anti-
body diluent Dako REAL (Dako, Hamburg, Germany), and
4′,6-diamidino-2-phenylindole (DAPI) mounting medium
ImmunoSelect (Dako, Hamburg, Germany) were used.
2. Experimental Section
Other reagents used in biological experiments: Brilliant
III Ultra-Fast SYBR Green QPCR Master Mix (Agilent, Santa
Clara, USA), bovine chondroitin sulfate (Sigma-Aldrich,
St. Louis, MO, USA), chloramine T (Carl Roth, Karlsruhe,
Germany), p-dimethylamino-benzaldehyde (DAB) (Carl Roth,
Karlsruhe, Germany), dimethylmethylene blue (DMMB), fast
green, Hoechst 33258, l-hydroxyproline, ImProm-II Reverse
Transcription System (Promega, Madison, USA), papain
(Worthington, Lakewood, USA), picric acid solution, Proteinase
K (Digest-All 4, Life Technologies Karlsruhe, Germany), Tissue
Tek Optimal Cutting Temperature compound (O.C.T., Sakura
Finetek, Tokyo, Japan), TRIzol Reagent (Thermo Scientific,
Waltham, USA), and safranin O were used as received.
2.1. Chemicals and Materials
Chemicals and materials were obtained from Sigma-Aldrich
(Sigma-Aldrich, St. Louis, MO, USA), if not stated otherwise.
2.1.1. PG-Acr Synthesis
Acrylic anhydride (Polysciences, Warrington, PA, USA), anhy-
drous N,N-dimethylformamide (>99.8%), benzoylated dialysis
tubing (molecular weight cut-off (MWCO) 2 kDa), calcium hydride
(92%, abcr GmbH, Karlsruhe, Germany), Biotech cellulose ester
dialysis membrane (MWCO 1 kDa, Spectrum Laboratories. Inc.,
Los Angeles, CA, USA), ethanol (>99.8%), ethyl vinyl ether (99%,
potassium hydroxide stabilized), glycidol (96%), potassium-tert-
butoxide (KOtBu, 1 m in tetrahydrofurane), pyridine (>99.8%, anhy-
drous), p-toluene sulfonic acid monohydrate (pTsOH, >98.5%),
sodium hydroxide, magnesium sulfate (p.a., Merck, Darmstadt,
Germany), and hydrochloric acid (39%, Merck) were used.
2.2. Analytical Instruments and Preparation
1H NMR spectra were recorded on a Bruker Fourier 300. Deu-
terated chloroform (CDCl3) and deuterated water (D2O) spectra
were recorded with nondeuterated solvent signals of CDCl3
(7.24 ppm) and 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid
sodium salt (0.00 ppm) as an internal reference, respectively.
Size exclusion chromatography (SEC) elugrams in dimeth-
ylformamid (DMF) (containing 1 g L−1 LiBr) were recorded
with an agilent 1260 infinity multi-detector suite from Polymer
Standards Service system (PSS, Mainz, Germany) with a
2.1.2. Hydrogel Formulation
2-Iminothiolane (Traut’s reagent), recombinant human TGF-
β1 (>98%, BioLegend), HA-SH (Glycosil, ESI BIO, Alameda,
©
1700390 (2 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
50 mm PFG precolumn and three 300 mm PFG columns (pore
size 7 mm) and a column oven at 40 °C. The elution rate was
1 mL min−1 and calibration was performed using poly(ethylene
glycol) standards (PSS, Mainz, Germany). Data was processed
with WinGPC software. Samples of the raw polymers and the
mercaptoethanol quenched acrylate derivatives were previously
filtered through 0.2 mm polytetrafluoroethylene (PTFE) syringe
filters (Roth, Karlsruhe, Germany). Rheological properties of
the hydrogels were evaluated using a Physica MCR 301 rheom-
eter from Anton Paar (Graz, Austria) equipped with a parallel-
plate geometry. The diameter of the plate was 25 mm and the
plate to plate distance was 0.5 mm. Freeze-drying of the poly-
mers after dialysis was performed with an freeze dryer (Alpha
1–2 LD, Christ, Osterode am Harz, Germany).
and acrylic acid were added (on free hydroxyl groups of PG) and
stirred overnight in the dark. For neutralization 1 mL of a 0.8 m
phosphate buffer (pH 8.0) was added and stirred for 0.5 h. The
final solution was dialyzed against ethanol (benzoylated dialysis
tubing, MWCO 2 kDa) for 2 d (10 × 100 mL) with light exclusion.
The ethanolic solution of PG-Acr was stored at 4 °C until further
use. Right before application, the solvent was removed by rotary
evaporation at room temperature after adding 10 mg phenothia-
zine as a stabilizer to about 5 mL of the PG-Acr solution.
t
1H NMR (300 MHz, D2O, δ) 1.238 (s, Bu), 3.330–4.041
(m, 5H, PG, CH2OH), 4.235–4.564 (m, 2H, COOCH2),
6.024–6.059 (m, 1H, OCOCHCH2, cis), 6.209–6.301 (m, 1H,
COCHCH2), 6.455–6.513 (m, 1H, OCOCHCH2, trans)
2.4. Hydrogel Formulation and Characterization
2.3. Poly(glycidol) Synthesis and Characterization
2.4.1. Hydrogel Preparation Procedure and Growth Factor
Incorporation
The monomer ethoxyethyl glycidyl ether (EEGE) was synthe-
sized according to an established procedure.[17] Glycidol and
ethyl vinyl ether were cooled to 0 °C. pTsOH was slowly added,
taking into account that the reaction temperature did not
exceed 20 °C. The obtained reaction mixture was subsequently
stirred for 3 h at RT. Afterward, the organic phase was washed
with 3 × 50 mL saturated NaHCO3 solution and dried with
MgSO4. Excess ethyl vinyl ether was removed using vacuum.
The resulting product was further dried with CaH2 and distilled
at about 1 mbar, and 60 °C. EEGE was obtained as a colorless
liquid and stored at 4 °C under argon atmosphere.
Before hydrogel formulation, HA-SH (Glycosil) stock solu-
tion (10 mg mL−1 in PBS, according to the manufacturer’s
instructions) and PG-Acr stock solution (125 mg mL−1 in PBS,
sparingly soluble phenothiazine was afterward removed by cen-
trifugation) were prepared. Three different kinds of hydrogels
were prepared: gels without TGF-β1, gels supplemented with
100 nm noncovalently bound TGF-β1, and gels that were cova-
lently modified with final concentrations of 10 nm, 50 nm, and
100 nm tethered TGF-β1.
In the case of the nonloaded hydrogels, PBS was added
to the PG-Acr stock solution to adjust its concentration to
29.5 mg mL−1. The hydrogels (n = 3 for each measurement)
were formulated by mixing the HA-SH stock solution with the
PG-Acr solution in a volume ratio of 4:1.
For covalent binding of TGF-β1 to the hydrogels, first, TGF-
β1 was thiol-modified using Traut’s reagent at a molar ratio of
4:1 of Traut’s reagent to TGF-β1 for 1 h at RT. Subsequently,
various doses of thiolated TGF-β1 were coupled to PG-Acr for
1 h at 37 °C, obtaining TGF-β1 containing PG-Acr solutions
(29.5 mg mL−1 in PBS). Finally, the hydrogels were formulated
by mixing the HA-SH stock solution with the differently TGF-
β1-modified PG-Acr solutions resulting in final concentrations
of 10 nm, 50 nm, and 100 nm tethered TGF-β1 (10 nm, 50 nm,
and 100 nm TGF-β1 + Traut).
For the preparation of hydrogels containing 100 nm of non-
covalently bound TGF-β1 (100 nm TGF-β1) a TGF-β1 con-
taining PG-Acr solution (29.5 mg mL−1 in PBS) was prepared as
described above, but without thiol-modification of TGF-β1 with
Traut’s reagent.
1H NMR (300 MHz, CDCl3, δ): 1.295 (t, 3H, CH3CH2,
3J = 7.0 Hz), 1.414 (m, 3H, CH3CH), 2.688–2.748 (m, 1H,
OCH2CHCH2), 2.894 (m, 1H, OCH2CHCH2), 3.188–3.292 (m,
1H, OCH), 3.480–3.932 (m, 4H, CH3CH2 and OCHCH2O),
4.821–4.884 (m, 1H, CHCH3) (Figure S1, Supporting
Information)
The linear poly(ethoxyethyl glycidyl ether) (PEEGE) was syn-
thesized in bulk using 50 equivalents of EEGE (10 mL, 9.62 g,
65.806 mmol) on one equivalent KOtBu initiator (1.316 mmol,
1.3158 mL 1 m KOtBu in tetrahydrofurane (THF)). For the reac-
tion, KOtBu was placed under argon atmosphere, and EEGE
was added afterward, the obtained mixture was subsequently
stirred for 1 d at 60 °C. The polymerization reaction was
stopped by addition of EtOH.
1H NMR (300 MHz, CDCl3, δ): 1.119 (t, 3H, CH3CH2,
3J = 6.95 Hz), 1.2165 (m, 3H, CH3CH), 3.320–3.658 (m, 7H,
CH2CH, CH2CH, CH2CH, CH3CH2), 4.582–4.671 (m, 1H,
CH3CH) (Figure S2, Supporting Information)
For deprotection, the linear polymer PEEGE was dissolved
in a minimal volume of ethanol and concentrated hydro-
chloric acid (37%) was added and stirred overnight at room
temperature. The solution was neutralized with 1.0 n NaOH,
dialyzed against H2O (MWCO 1 kDa) for 2 d (10 × 2 L), and
subsequently freeze-dried. The yield of the resulting PG over
two reaction steps (polymerization and deprotection) was about
70%, depending on the applied dialysis time.
2.4.2. Rheological Characterization
Oscillatory measurements with a standard plate-plate geom-
etry of 25 mm diameter were performed at 20 °C with 400 mL
of the hydrogel precursor solution. Time sweeps (angular
frequency = 1 rad s−1, strain = 1%) were performed over a
time period of 90 min. The gelation point was determined at
the intersect of the elastic and viscous modulus. Amplitude
t
1H NMR (300 MHz, D2O, δ): 1.221 (s, 9H, Bu), 3.577–3.784
(m, 5H)
For the preparation of functional polymers, the deprotected
linear PG was dissolved in dry DMF. 0.2 equivalents of pyridine
©
1700390 (3 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
(angular frequency = 10 rad s−1, strain 0.1–10%) and frequency
(angular frequency = 100–0.1 rad s−1, strain = 1%) sweeps were
performed immediately after the time sweeps to verify the cor-
rect measurements of the hydrogels. A water-soaked paper
towel and a cap were used to prevent drying-out of the hydrogel
solution during the measurements.
T175 cm2 flasks (Greiner Bio-One, Frickenhausen, Germany).
After 2 d the cells were washed with PBS to remove nonad-
herent cells, and the adherent cells were subsequently cul-
tured to a subconfluent level at 37 °C, 5% CO2 in proliferation
medium. For MSC passaging, cells were detached with 0.25%
trypsin-EDTA and seeded at a density of ≈5000 cells mL−1 into
T175 cm2 flasks.
2.4.3. Hydrogel Swelling, Sol Fraction, and Degradation
Measurements
2.4.6. MSC Hydrogel Encapsulation
For in vitro swelling measurements, hydrogels with 100 nm
covalently bound TGF-β1 and w/o TGF-β1 with a final volume
of 150 mL were mixed and the precursor solutions were filled
into glass cylinders with a diameter of 6 mm and a height of
8 mm, which were sealed to the bottom with silicone grease
and afterward incubated at 37 °C for 0.5 h. After 0.5 h the
premature hydrogels were removed from the glass cylinders,
weighed and put in microcentrifuge cups with 2 mL PBS and
incubated at 37 °C without shaking. Triplets of each hydrogel
type were produced for each time point (0.25 d and 1, 4, 7,
14, and 21 d). To determine the swelling ratio and sol frac-
tion, hydrogels and solvent were separated and excessive PBS
was removed carefully from the hydrogel surface using a filter
paper and the swollen weight (Ws) was evaluated. Subsequently,
the hydrogels were freeze-dried, the weight of the dry hydrogel
(Wd) was determined, and the swelling ratio (Qm) was calcu-
lated according to Equation (1).
The hydrogel precursor solutions were prepared as described
above (Section 2.4.1). Additionally, the PG-Acr stock solution
(29.5 mg mL−1 in PBS) was sterile-filtered through a 0.2 mm
syringe-filter prior utilization.
The MSCs were propagated up to passage 2 or 3 for hydrogel
encapsulation experiments. Prior to crosslinking, MSCs were
resuspended in the HA-SH solution, and either the differ-
ently TGF-β1-modified PG-Acr solutions (TGF-β1 + Traut), a
PG-Acr solution containing 100 nm TGF-β1 w/o Traut’s rea-
gent (100 nm TGF-β1), or unmodified PG-Acr were added
to the cell-laden HA-SH solutions at a final concentration of
20.0 × 106 MSC mL−1. Finally, 40 mL of the HA-SH hydrogel
solutions were filled into a glass ring (Ø 5 mm) and allowed to
gel for 30 min at 37 °C, 5% CO2. Afterward, hydrogel encap-
sulated MSCs were cultured for up to 21 d in chondrogenic
medium (DMEM high glucose 4.5 g L−1, supplemented with
1% ITS+ Premix, 40 mg mL−1 l-proline, 50 mg mL−1 l-ascorbic
acid 2-phosphate sequimagnesium salt hydrate, 0.1 × 10−6
m
dexamethasone, 1 × 10−3 m sodium pyruvate, and 1% PS), w/o
soluble TGF-β1. Growth factor-unmodified PG-Acr hydrogels
served as controls and were cultured in chondrogenic medium
either supplemented with TGF-β1 (addition with each medium
change, 10 ng mL−1), as standard control for in vitro chon-
drogenesis of MSC (TGF-β1 Medium), or in chondrogenic
medium w/o TGF-β1 (w/o TGF-β1), as a negative control.
Ws
Qm
=
(1)
Wd
2.4.4. Quantification of TGF-β1 Release
To compare the amount of total released TGF-β1 from hydrogels
loaded with 100 nm of tethered TGF-β1 (100 nm TGF-β1 + Traut)
and 100 nm of noncovalently bound TGF-β1 (100 nm TGF-β1),
cell-free hydrogels were prepared as described above (Section
2.4.1). The respective hydrogels (V = 40 mL) were maintained in
chondrogenic medium, w/o soluble TGF-β1, over a time course
of 21 d. The supernatants of the hydrogels were collected at dif-
ferent time points (0.25, 1, 2, 4, 9, 11, 14, 16, 18, and 21 d) and
stored at −20 °C until analysis. The amount of released TGF-β1
was measured using a LEGEND MAX Total TGF-β1 enzyme-
linked immunosorbent assay (ELISA) Kit (BioLegend, London,
UK) according to the manufacturer instructions.
2.4.7. Cell Viability Assay
The viability of encapsulated MSCs was assessed using a
Live/Dead cell staining kit. At day 2, 10, and 21 after encap-
sulation, cell-laden hydrogels were washed with PBS and incu-
bated in the Live/Dead staining solution (4 × 10−6 m ethidium
homodimer III (EthD-III), 2 × 10−6 m calcein acetoxymethyl
ester (Calcein-AM)) for 45 min. Following that, top view images
were captured immediately using a fluorescence microscope
(Olympus BX51/DP71, Olympus, Hamburg, Germany).
2.4.5. MSC Isolation and Expansion
2.4.8. Histological and Immunohistochemical Analyses
Human bone marrow-derived MSCs were isolated from the
surgically removed cancellous bone of patients undergoing
total hip replacement, with written informed consent from all
patients, and as approved by the local ethics committee. Briefly,
MSCs were collected by repeated washing of bone debris and
marrow in PBS, then centrifuged at 300 g for 10 min, resus-
pended in proliferation medium (DMEM/F12, supplemented
with 10% FBS, 1% PS and 5 ng mL−1 bFGF), and seeded into
The hydrogels were initially fixed in 3.7% PBS-buffered for-
malin for 60 min, washed two times in PBS for 15 min, and
incubated in O.C.T. overnight at 4 °C. On the next day, con-
structs were frozen in cryomolds containing fresh O.C.T.
using liquid nitrogen and stored at −80 °C until sectioned.[18]
Longitudinal sections at 8 mm were prepared and collected on
Super Frost plus glass slides (R. Langenbrinck, Emmendingen,
©
1700390 (4 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Germany). For histological evaluation the samples were stained
with either Weigert’s hematoxylin, fast green, and safranin O
for deposition of glycosaminoglycans (GAG)[19] or Weigert’s
hematoxylin and picrosirius red for collagen deposition.[20]
For immunohistochemical analyses, the cryosections were
rehydrated, and antigen retrieval was performed using Proteinase
K for 10 min at RT. Subsequently, sections were blocked with 1%
bovine serum albumin (BSA) for 30 min; primary antibodies
were diluted in Antibody diluent Dako REAL and incubated over-
night in a humidified chamber at RT. Antibodies for collagen
type I (ab34710, 1:800), collagen type II (II-4C11, 1:100), collagen
type X (X53, 1:200), and aggrecan (969D4D11, 1:300) were used.
Sections were washed three times in PBS for 3 min, and sec-
ondary antibodies were diluted in Antibody diluent Dako REAL
and applied in the dark for 1 h. A donkey anti-mouse (Cy3, 1:500)
and a goat anti-rabbit (Alexa Fluor 488, 1:400) secondary antibody
were used. Subsequently, the slides were washed three times
in PBS for 3 min and mounted with DAPI mounting medium
ImmunoSelect. Images were captured with a fluorescence micro-
scope (Olympus BX51, Olympus, Hamburg, Germany).
qRT-PCR analysis. qRT-PCR was carried out using PrimePCR
SYBR Green Assay (Bio-Rad, München, Germany) primer pairs
for aggrecan (ACAN, qHsaCID0008122), collagen I (COL1A1,
qHsaCED0043248), collagen II (COL2A1, qHsaCED0001057),
collagen
X (COL10A1, qHsaCED0043992), Sox9 (SOX9,
qHsaCED0021217), glyceraldehyde-3-phosphate dehydrogenase
(GAPDH, qHsaCED0038674), according to the manufacturer’s
cycle instructions. Expression of all genes was normalized to
the housekeeping gene GAPDH for each construct group and
time point. The x-fold increase in gene expression levels for
each gene was determined using the 2−∆∆CT method and further
normalized to cells from 2D samples at day 0.
2.4.11. Statistics
Statistics was performed using GraphPad Prism, Version 6.0
(GraphPad Software, La Jolla, USA). Results are presented as
mean values standard deviation (SD). Statistical significance
was assessed by either multiple t-tests followed by Holm–Sidak
post hoc test, or by two-way analysis of variance followed by
Tukey’s post hoc test, as appropriate. The statistical significance
level was set at p < 0.05.
2.4.9. Biochemical Analyses
For biochemical analyses, the constructs were digested in 1 mL
of a papain solution (3 U mL−1) for 16 h at 60 °C. Prior to
papain digestion the gels were homogenized at 25 Hz for 5 min
using a TissueLyser (Qiagen, Hilden, Germany).
3. Results and Discussion
3.1. Poly(glycidol) Synthesis and Characterization
DNA content of digested constructs was measured using
Hoechst 33258 DNA intercalating dye. DNA quantification
was carried out fluorometrically at 340 and 465 nm, using
salmon testis as standard.[21] The amount of GAG produced by
the encapsulated cells was measured using the DMMB assay.
The GAG amount was determined spectrophotometrically at
525 nm, using bovine chondroitin sulfate as standard.[22] The
content of hydroxyproline was measured, after acid hydrolysis
and reaction with DAB and chloramine T. The quantification
was adapted to 96-well format and carried out with a spectro-
photometer at 570 nm, using l-hydroxyproline as standard. The
amount of collagen was calculated using a hydroxyproline to
collagen ratio of 1:10.[23,24]
Linear homopolymeric PG with 1-ethoxyethyl protection
of the alcohol groups was obtained by anionic ring-opening
polymerization of ethyl glycidyl ether with potassium
tert-butanoate as initiator via solventless synthesis with a
monomer to initiator ratio of 50:1. 1H-NMR analysis con-
firmed complete monomer conversion and the structure of
the obtained polymer (Figure S2, Supporting Information).
After polymerization, the ethoxyethyl protecting groups were
completely removed by treatment with hydrochloric acid.
The PG alcohol groups were afterward partially converted to
PG-Acr by reaction with 0.2 equivalents of acrylic anhydride
catalyzed by pyridine (Figure 1).
To facilitate the work-up procedure, the resulting PG-Acr
was afterward treated with 0.8 m phosphate buffer (pH 8.0)
to decompose the formed organic pyridinium acrylate salt
(pKa = 5.25 in H2O),[25] which otherwise might have a high
binding affinity to the polyether structure of PG. Subsequently,
the obtained solution was dialyzed against EtOH to minimize
possible ester hydrolysis. The obtained ethanolic solution was
used to store the polymer in the dark for longer periods of time.
Before application, the solvent had to be removed, which was
performed by addition of phenothiazine to prevent an early
2.4.10. Gene Expression
Before RNA isolation, cell-laden hydrogels, and cells of 2D sam-
ples at day 0 were homogenized in TRIzol Reagent, and RNA
was subsequently isolated according to the manufacturer’s
instructions. First-strand cDNA was synthesized from total RNA
by using the ImProm-II Reverse Transcription System. Brilliant
III Ultra-Fast SYBR Green QPCR Master Mix was used for
Figure 1. Synthetic route to PG-Acr.
©
1700390 (5 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
1
Figure 2. Characterization of the multifunctional PG-Acr crosslinkers. A) H NMR spectra (D2O) of PG (top) and PG-Acr (bottom) to show acrylate
modification. B) Molecular weight distributions determined by SEC measurements (DMF) of PG and PG-Acr to show the molecular weight increase
after successful acrylation.
radical crosslinking of the reactive acrylate groups during the
procedure. Phenothiazine (soluble in water with ≈1.7 mg L−1)[26]
precipitated after the dissolution of the mixture in aqueous sol-
buffer-dependent Michael addition crosslinking reaction.[10,27]
A
possible side reaction is oxidative linking of thiols to disulfide
bridges. Due to the large differences in reaction kinetics in
favor of Michael addition, the contribution of disulfide forma-
tion between modified hyaluronic acid molecules is negligible
for the network formation. In order to prepare biomimetic
hydrogels, nucleophilic groups (for example free lysine amine
groups) of TGF-β1 were first reacted with Traut’s reagent
(Figure 3) in a 4:1 molar ratio of Traut’s reagent to TGF-β1,
as described in literature,[15] followed by reacting the resulting
conjugate with PG-Acr. The resulting PG-Acr-TGF-β1 conju-
gates were then used for hydrogel formation with the HA-SH
solution (and cells in cases of cell experiments) to yield an
overall concentration of 10 nm, 50 nm, or 100 nm TGF-β1 in the
hydrogel. The resulting overall polymer content of the hydro-
gels was 1.388% (w/v), with 0.8% (w/v) for HA-SH and 0.588%
(w/v) for PG-Acr, representing a low mass percentage hydrogel,
likely providing a good penetrability for cellularly produced
ECM.[28,29]
1
vent and could easily be removed by centrifugation. H NMR
and SEC analysis (Figure 2) demonstrated that, as desired, the
obtained degree of functionalization was 20% 2% (estimated
by comparison of the 1H NMR integrals of the CH2 group next
to the ester and the PG backbone groups), providing a sufficient
amount of acrylate functions for crosslinking and biomimetic
functionalization. In comparison to the used PG homopolymer
(Ð = 1.31), the molecular weight distribution of the PG-Acr
(Ð = 1.18) did not show enhanced dispersity, indicating the
effective protection of the acrylates by the added stabilizer.
3.2. Hydrogel Formation and Characterization
Hydrogels were formed by crosslinking of the linear poly-
meric precursors PG-Acr and HA-SH solution via the pH- and
Figure 3. In situ functionalization of PG-Acr with TGF-β1 via Traut’s reagent and crosslinking with the thiol-functionalized hyaluronic acid to bioactive
hydrogels.
©
1700390 (6 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 4. Characterization of the obtained hydrogels with covalently bound (100 nm TGF-β1) and without TGF-β1 (w/o TGF-β1). A) Representative
rheological time sweep indicating the similar hydrogel stiffness with and without growth factor (dashed line indicate the time of prematuration in
the glass cylinders). B) Magnification of the initial crosslinking phase indicating an early gel formation for both gels. C) Swelling ratio Qm based on
hydrogel mass.
By oscillatory rheometry (Figure 4A,B and Figure S3, Sup-
porting Information), information about the crosslinking
kinetics and fluid-solid properties during the hydrogel forma-
tion were gained. Hydrogels with and without 100 nm TGF-β1
were formulated by mixing the viscous HA-SH stock solution
with the low viscous PG-Acr or the PG-Acr/TGF-β1 stock solu-
tion. The resulting rheograms showed no significant differ-
ences between hydrogels with and without covalently bound
100 nm TGF-β1, both for the kinetics of gelation and the finally
achieved elastic moduli. The elastic modulus in both cases was
about 500 Pa after 30 min and 1100 Pa after 90 min of gelation.
These results demonstrate the suitability of side-chain modi-
fied PG for covalent binding of biomimetic cues such as growth
factors without compromising the gelation process. Even at the
highest investigated concentration of TGF-β1, the rheological
properties of the resulting hydrogels were not affected. Fur-
thermore, the gelation profile showed that the hydrogel system
provided a suitable time window between the gelation point at
about 2 min, after which embedded cells do not rapidly sink to
the bottom any more, and an ongoing gelation and crosslinking
for over 1 h, to result in well-distributed cells incorporated
within the hydrogels. The hydrogels were observed to be con-
siderably softer in comparison to native cartilage, for which an
elastic modulus of ≈0.4–0.6 MPa was reported.[30] However, a
hydrogel stiffness similar to that determined for the hydrogels
developed in this study has previously been shown to be suit-
able in cartilage engineering for robust chondrogenic differen-
tiation of MSCs.[31,32]
For further hydrogel characterization, the swelling ratios of
the hydrogels, with and without covalently incorporated 100 nm
TGF-β1, were evaluated at different incubation time points
(0.25, 1, 4, 7, 14, and 21 d). As shown in Figure 4C, in general
there was no significant difference between hydrogels without
(0.25 d: 14.0 0.7 and 21 d: 17.6 0.3) and those with 100 nm
TGF-β1 (0.25 d: 17.5 0.3 and 21 d: 16.9 0.8) at early and late
time points. No hydrogel dissolved within the time frame of
21 d, and swelling ratio only slightly increased over time, which
indicated a progressive loss of hydrogel integrity, possibly due
to ester hydrolysis and subsequent water gain. As ester bonds
close to the thioester are known to be prone to hydrolysis,[33]
hydrolysis experiments with PG-Acr and PG mercaptoethanol
as a model compound for the converted PG acrylate groups
were performed. After three weeks, around 90% of the ester
groups were still maintained. The weight of the released sol
fraction was evaluated additionally and showed no significant
increase within three weeks (data not shown), which one could
expect for a multianchored polymeric network, emphasizing
the benefits of employing multiple side chain-functionalized
PG-Acr in comparison to the commonly used bifunctional PEG
diacrylates or the tetrafunctional starPEG acrylates.[11,34]
©
1700390 (7 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
All groups exhibited high amounts of viable cells during
the 21 d of in vitro chondrogenesis, and we did not detect sig-
nificant differences between the different hydrogel groups
regarding cell viability (Figure S4, Supporting Information).
3.3.1. Determination of GAG Production
The determination of GAG production, either quantita-
tively using DMMB assay or histologically using safranin
O, showed that hydrogels cultured w/o TGF-β1 produced,
as expected, no significant amounts of GAG (Figure 6).
10 nm and 50 nm TGF-β1 + Traut-modified gels showed no, or
very little GAG deposition in the safranin O staining (Figure S5,
Supporting Information), and no significant GAG production
in the DMMB assay (Figure 6B). In contrast to that, the 100 nm
TGF-β1 + Traut group showed an intense GAG deposition in
histological images, which was distinctly stronger than that
determined for the group with noncovalently bound TGF-β1
(100 nm TGF-β1) (Figure 6A). While the safranin O staining
exhibited very strong pericellular signals in the 100 nm TGF-
β1 + Traut group, GAGs seemed to be more evenly distributed
throughout the gel in the TGF-β1 Medium group, but less
intense (Figure 6A). Reflecting the histological results, quanti-
fication of GAG demonstrated that the 100 nm TGF-β1 + Traut
group produced significantly higher amounts of GAG/DNA
(26.6 2.3 mg/mg), as compared to the 100 nm TGF-β1 group
Figure 5. Cumulative release of 100 nm covalently bound and noncova-
lently bound TGF-β1 from the hydrogels, determined by ELISA.
To further characterize the hydrogel systems, the amount
of TGF-β1 released from cell-free hydrogels was determined
via ELISA. Hydrogels loaded with covalently bound TGF-β1
(100 nm TGF-β1 + Traut) were compared to hydrogels loaded
with the same amount of noncovalently bound TGF-β1 (100 nm
TGF-β1). An ELISA of the initially incorporated TGF-β1 solu-
tion yielded correct TGF-β1 concentrations during growth
factor loading into the hydrogels (data not shown). The super-
natants of the respective hydrogels were collected over a time
course of 21 d. The resulting cumulative release curves showed
no significant differences between hydrogels with covalently
bound and noncovalently bound TGF-β1 (Figure 5). In both
cases no initial burst release occurred. Only a 7–8% overall
release of TGF-β1 was detectable after 21 d. Possible reasons
for the relatively low release, especially of the noncovalently
bound protein, may include electrostatic interactions of TGF-β1
with the hyaluronic acid component of the gels, or an effective
diffusion inhibition by the hydrogel network.
(13.0
2.7 mg/mg), indicating a beneficial effect of covalent
incorporation of TGF-β1 on chondrogenesis compared to just
mixing TGF-β1 into the gels (Figure 6B). In this regard, com-
parisons to previous studies employing covalently bound TGF-
β1 are difficult, as a group with incorporated, but not covalently
bound TGF-β1 was not investigated in these studies.[14–16] In
the present study, interestingly, the group with 100 nm teth-
ered TGF-β1 was superior to the TGF-β1 Medium control
group (17.2 3.1 mg/mg) after 21 d of chondrogenic differen-
tiation (Figure 6B). This result was in agreement with obser-
vations reported for purely PEG-based hydrogel systems with
tethered TGF-β1 compared to TGF-β1-supplemented medium,
using either MSC or chondrocytes.[14,15] In this study, in order
to covalently bind TGF-β1, the growth factor was thiol-modified
using Traut’s reagent at a molar ratio of 4:1 of Traut’s reagent
to TGF-β1 which, as demonstrated previously, does not impair
TGF-β1 bioactivity and functionality when compared to native
TGF-β1.[14,15]
3.3. Chondrogenic Differentiation of MSCs within
the HA/PG Hydrogels
3.3.2. Determination of Collagen Production
After the successful establishment and characterization of the
PG-Acr crosslinked HA-SH-based hydrogel system, we evalu-
ated the chondrogenic response of human MSCs to biomimetic
hydrogels incorporating TGF-β1 as model substance. Varying
concentrations of TGF-β1 were covalently bound to the hydro-
gels (TGF-β1 + Traut group). Additionally, the effect on MSC
chondrogenesis was compared to hydrogels with noncovalently
bound TGF-β1 (TGF-β1 group) and, as a standard control,
TGF-β1 supplemented with each medium change (10 ng mL−1;
TGF-β1 Medium group).
MSCs cultured in gels w/o TGF-β1 synthesized very little
amounts of collagens, as shown by hydroxyproline assay and pic-
rosirius red staining (Figure 7). Moreover, collagen staining for
the 10 nm and 50 nm TGF-β1 + Traut-modified groups resulted
in very weak signals (Figure S6, Supporting Information), and
substantially lower production of collagens compared to the
other chondrogenically induced groups (Figure 7B). Strong pic-
rosirius red staining was detected for the 100 nm TGF-β1 + Traut
group. Similar staining, but slightly more homogeneously dis-
tributed throughout the gel matrix, was observed for the TGF-β1
©
1700390 (8 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 6. A) Histological stainings of TGF-β1-laden hydrogels, seeded with 20.0 × 106 MSCs mL−1, after 10 and 21 d of chondrogenic differentiation.
Longitudinal sections were stained for deposition of glycosaminoglycans with safranin O; scale bars represent 100 mm. B) GAG production of MSCs
encapsulated in TGF-β1-modified hydrogels after 10 and 21 d of chondrogenic differentiation, shown for total GAG (GAG/Gel) and normalized to
DNA (GAG/DNA). Data are presented as means standard deviation (n = 3). (✶) Significantly different from control gels w/o TGF-β1 at the same
time point (p < 0.05). (a) Significantly different from the TGF-β1 Medium group at the same time point (p < 0.05). (b) Significantly different from the
10 nm TGF-β1 + Traut group at the same time point (p < 0.05). (c) Significantly different from the 50 nm TGF-β1 + Traut group at the same time point
(p < 0.05). (d) Significantly different from the 100 nm TGF-β1 + Traut group at the same time point (p < 0.05). (°) Significant differences of time points
within a group (p < 0.05). Representative results of one of three independent experiments are shown.
Medium group after 21 d (Figure 7A). In contrast, the collagen
signal in the 100 nm TGF-β1 group was distinctly weaker and
mainly limited to pericellular regions (Figure 7A). Quantifica-
tion of total collagen was well in agreement with the histological
results, as the determination of collagen/DNA content showed
that TGF-β1 Medium group (8.4 0.2 mg/mg) and 100 nm TGF-
β1 + Traut group (7.3 0.8 mg/mg) produced significantly higher
amounts than the 100 nm TGF-β1 group with nonthiol-function-
alized growth factor (3.2 1.9 mg/mg) (Figure 7B).
These results clearly demonstrated the superior effect of
covalent incorporation of 100 nm TGF-β1 into the HA hydro-
gels compared to the noncovalent incorporation of 100 nm
TGF-β1 with regard to GAG and collagen deposition. Cartilagi-
nous ECM was formed to a much higher extent in the TGF-β1
tethered group than in the nontethered group (Figures 6 and 7).
These distinct differential biological effects were observed
despite the fact that similar release profiles were detected for
hydrogels with covalently and noncovalently bound TGF-β1
(Figure 5). Possible explanations include potential differences in
TGF-β1 signaling. When free TGF-β1 is bound to its receptor,
the protein–receptor complex can undergo endocytosis with
subsequent intracellular degradation[35] leading to reduction
in concentration of free protein and receptor density. In hydro-
gels with covalently bound TGF-β1, the protein may bind to the
receptor triggering TGF-β1 signaling without consuming pro-
tein and receptor leading to prolonged signaling. Furthermore,
a modulating factor contributing to the differential biological
effects may be the way of growth factor immobilization and
presentation, respectively, which may have been advantageously
changed by the covalent binding.
In this study, PG was chosen as a structural analog for the
commonly used and terminal-modified PEG derivatives. PEG
hydrogels are biologically inert and need biological modifi-
cation with, e.g., peptide sequences in order to allow cells to
attach and recognize the hydrogel environment.[36] Additionally,
they can be combined with biological macromolecules in com-
posite hydrogels to increase their bioactivity.[8,9] PG is chemi-
cally closely related to PEG and biologically similarly inert but
provides the possibility of a higher functionalization density
due to side chain functionalization.[11,34] This is also true for
PG-based hydrogels,[37] as we were able to show recently by uti-
lizing thiol- and allyl-modified PG for the generation of thiol-
ene clickable PG hydrogels. By replacing the thiol-modified PG
component with bioactive thiol-modified HA, the deposition of
cartilage-specific matrix components secreted by encapsulated
MSCs, was significantly improved when compared to pure PG
hydrogels.[38] Because of these recent findings we decided in
the present study to combine the chondrosupportive effect of
HA with the possibility to introduce precise biomimetic func-
tionalization using acrylate-modified PG as multifunctional
crosslinker at the example of TGF-β1. In comparison to pre-
vious studies investigating covalently bound TGF-β1 which
were carried out using pure PEG hydrogels,[14,15] chondrogen-
esis of MSCs in the HA-based hydrogels presented in this study
©
1700390 (9 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 7. A) Histological stainings of TGF-β1-laden hydrogels, seeded with 20.0 × 106 MSCs mL−1, after 10 and 21 d of chondrogenic differentiation.
Longitudinal sections were stained for deposition of collagens with picrosirius red; scale bars represent 100 mm. B) Collagen production of MSCs
encapsulated in TGF-β1-modified hydrogels after 10 and 21 d of chondrogenic differentiation, shown for total collagen (Collagen/Gel) and normalized
to DNA (Collagen/DNA). Data are presented as means standard deviation (n = 3). (✶) Significantly different from control gels w/o TGF-β1 at the
same time point (p < 0.05). (a) Significantly different from the TGF-β1 Medium group at the same time point (p < 0.05). (b) Significantly different from
the 10 nm TGF-β1 + Traut group at the same time point (p < 0.05), (c) significantly different from the 50 nm TGF-β1 + Traut group at the same time
point (p < 0.05). (d) Significantly different from the 100 nm TGF-β1 + Traut group at the same time point (p < 0.05). (°) Significant differences of time
points within a group (p < 0.05). Representative results of one of three independent experiments are shown.
was much more pronounced, as shown on biochemical level as
well as in histological examination. This may be due to the fact
that HA inherently facilitates cellular attachment and migration
via the respective receptors[39,40] and showed chondrosupportive
effects in previous studies.[7]
was also clearly detectable, but slightly dimmer and pericellu-
larly less pronounced. In contrast to that, the group with non-
covalently bound 100 nm TGF-β1 showed only few pericellular
signals without deposition into the gel matrix (Figure 8). IHC
for collagen type I, undesired in articular cartilage, also showed
the highest amount of deposition for gels cultured in TGF-β1
containing medium, with strong pericellular signals. The col-
lagen type I signal in the 100 nm TGF-β1 + Traut was equally
distributed, but less pronounced, while the 100 nm TGF-β1
group again showed mostly weak pericelluar signals (Figure 8).
For collagen X, only very minor staining was observed in all
groups receiving TGF-β1 (Figure 8).
The hydrogels investigated in this study were permissive
for an overall even distribution of cell-derived ECM molecules
(Figures 6–8), which represents an advantageous feature of
the newly developed and relatively soft materials. This is well
in agreement with previous studies in which hydrogels with
different stiffness were compared. It was shown for cells that
were encapsulated in hydrogels with higher stiffness[7,28,29]
that the deposition of ECM molecules was mainly restricted
to pericellularly regions, in contrast to softer gels with a
more even ECM distribution. In previous studies employing
TGF-β1 covalently bound in PEG hydrogels, unfortunately
mechanical properties of the hydrogels were not determined;
however, the high amount of polymer applied (10 wt% PEG)
suggests relatively stiff hydrogels, which in turn explains
the observed uneven ECM distribution which was strongly
3.3.3. Immunohistochemistry
In order to examine ECM production in more detail, IHC was
employed to investigate the deposition of cartilage-specific ECM
components. As expected, the hydrogel constructs cultured in
medium w/o TGF-β1 showed no staining for cartilage-specific
ECM components, such as aggrecan or collagen II (Figure 8).
Cells encapsulated in the 10 nm and 50 nm TGF-β1 + Traut-
modified hydrogels also exhibited a very low deposition of
cartilage-specific ECM components (Figure S7, Supporting
Information). As already seen for deposition of GAGs, the
signal for aggrecan was most pronounced in the 100 nm TGF-
β1 + Traut group, exhibiting evenly distributed as well as strong
pericellular signals. Aggrecan also appeared to be distributed
throughout the hydrogel matrix in the TGF-β1 Medium group,
while the 100 nm TGF-β1 group revealed weaker, mainly pericel-
lular aggrecan signals (Figure 8). The most prominent staining
for collagen type II was detected in the TGF-β1 Medium group,
confirming the staining for total collagen by picrosirius red.
The type II collagen signal in the 100 nm TGF-β1 + Traut group
©
1700390 (10 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 8. Immunohistochemical staining of TGF-β1-laden hydrogels, seeded with 20.0 × 106 MSCs mL−1, after 21 d of chondrogenic differentiation.
Longitudinal sections were either stained for aggrecan, collagen II, and collagen X (all red) or collagen I (green) to show ECM development. Nuclei
(blue) were counterstained with DAPI; scale bars represent 100 mm.
pronounced in pericellular regions.[14,15] A comparative anal-
ysis focusing on the influence of HA hydrogel crosslinking
density on ECM distribution showed that an increase in
crosslinking density led to an overall decrease in cartilage-
specific ECM production and a more limited distribution
throughout the hydrogel.[29] Possible disadvantages resulting
from the relative softness of hydrogels such as the ones pre-
sented here with regard to load bearing may be overcome by
the utilization of reinforcement structures. Incorporation
of reinforcing fibers have been shown to be advantageous
for hydrogel-based cartilage engineering approaches[41] and
may improve the load resistance of the hydrogels in in vivo
applications.
group (Figure 9A,B,E). Thus, the gene expression profiles of
chondrogenic marker genes reflected the observations from
GAG and collagen determination and immunohistochem-
istry and supported the beneficial effect of tethered TGF-β1
on MSC chondrogenesis compared to nontethered growth
factor. COL1A1 expression increased initially in all chondro-
genically induced groups to a certain extent, but decreased
as chondrogenesis progressed (Figure 9C). The assessment
of the hypertrophic marker gene COL10A1 showed also in
the 100 nm TGF-β1 + Traut group a significant upregulation
after 21 d, in contrast to the other chondrogenically induced
groups (Figure 9D). Interestingly, this was not reflected in
the immunohistochemical results, showing only very minor
staining for collagen X after 21 d (Figure 8). Further investi-
gations with an extended time frame may elucidate the long-
term effects of covalently bound TGF-β1 with regard to MSC
hypertrophy. In general, qRT-PCR results reflected the pre-
vious findings and showed the beneficial effect of covalently
bound TGF-β1 in the 100 nm TGF-β1 group in comparison
to the noncovalently bound 100 nm TGF-β1 group on MSC
chondrogenesis.
3.3.4. Gene Expression
To further investigate possible differences between groups
with different modes of TGF-β1 application, mRNA expres-
sion profiles of chondrogenic marker genes were assessed
using qRT-PCR. In general, all chondrogenically differ-
entiated gel-encapsulated MSCs showed strongly elevated
expression of all analyzed genes, as compared to cells cul-
tured in hydrogels w/o TGF-β1 (Figure 9). In the 100 nm
TGF-β1 + Traut group, mRNA expression of the chondro-
genic markers COL2A1, ACAN, and SOX9 was increased
significantly, compared to the 100 nm TGF-β1 group (with
noncovalently bound growth factor) and the TGF-β1 Medium
4. Conclusions
PG-Acr was successfully synthesized and a hydrogel system
consisting of HA-SH and PG-Acr with crosslinking via
Michael addition was developed. By covalent binding of the
©
1700390 (11 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 9. Time course of gene expression of MSCs encapsulated in TGF-β1-modified hydrogels after 5, 10, 15, and 21 d of chondrogenic differentiation,
determined by qRT-PCR. Data are presented as means standard deviation (n = 3). (✶) Significantly different from control gels w/o TGF-β1 at the
same time point (p < 0.05). (a) Significantly different from the TGF-β1 Medium group at the same time point (p < 0.05). (b) Significantly different from
the 100 nm TGF-β1 + Traut group at the same time point (p < 0.05). (°) Significant differences of time points within a group (p < 0.05). Representative
results of one of two independent experiments are shown.
model substance TGF-β1 via Traut’s reagent to PG-Acr, the
hydrogel was successfully biofunctionalized without com-
promising the overall gelation process and swelling behavior
of the gels. With the exemplary covalent incorporation of
TGF-β1, the general suitability of PG-Acr for the generation
of hydrogels in cartilage tissue engineering approaches was
demonstrated. Chondrogenesis of human MSCs was sig-
nificantly improved in hydrogels with covalently bound TGF-
β1 compared to gels without covalent incorporation of the
growth factor. Possible effects on MSC hypertrophy should
be further investigated. In principle, because of its multifunc-
tionality, PG offers the opportunity to incorporate various bio-
logical cues, such as biomimetic peptides and growth factors
even at the same time, into hydrogels and to thereby enhance
the chondrogenic and eventually the clinical potential of the
developed hydrogels.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
T.B. and V.S. contributed equally to this work. The authors gratefully
acknowledge the received funding from the Interdisciplinary Center for
Clinical Research Würzburg (IZKF; Project No. D-218 and D-219) and from
the European Union’s Seventh Framework Programme (FP7/2007-2013)
under Grant Agreement No. 309962 (project HydroZONES). Furthermore,
this work was supported by the German Research Foundation (DFG),
collaborative research centre TRR225 (subproject A02).
Conflict of Interest
The authors declare no conflict of interest.
©
1700390 (12 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
[19] I. Martin, B. Obradovic, L. E. Freed, G. Vunjak-Novakovic, Ann.
Biomed. Eng. 1999, 27, 656.
[20] N. Schmitz, S. Laverty, V. B. Kraus, T. Aigner, Osteoarthritis Cartilage
2010, 18, S113.
Keywords
chondrogenesis, hyaluronic acid, mesenchymal stromal cells,
poly(glycidol), TGF-β1
[21] Y.-J. Kim, R. L. Y. Sah, J.-Y. H. Doong, A. J. Grodzinsky, Anal. Biochem.
1988, 174, 168.
[22] R. W. Farndale, D. J. Buttle, A. J. Barrett, Biochim. Biophys. Acta,
Gen. Subj. 1986, 883, 173.
Received: November 23, 2017
Revised: February 20, 2018
Published online:
[23] J. F. Woessner, Arch. Biochem. Biophys. 1961, 93, 440.
[24] A. P. Hollander, T. F. Heathfield, C. Webber, Y. Iwata, R. Bourne,
C. Rorabeck, A. R. Poole, J. Clin. Invest. 1994, 93, 1722.
[25] D. D. Perrin, Pure Appl. Chem. 1969, 20, 133.
[26] S. Ahmadian, V. Panahi-Azar, M. A. A. Fakhree, W. E. Acree,
A. Jouyban, J. Chem. Eng. Data 2011, 56, 4352.
[27] J. W. Chan, C. E. Hoyle, A. B. Lowe, M. Bowman, Macromolecules
2010, 43, 6381.
[28] I. E. Erickson, A. H. Huang, S. Sengupta, S. Kestle, J. A. Burdick,
R. L. Mauck, Osteoarthritis Cartilage 2009, 17, 1639.
[29] L. Bian, C. Hou, E. Tous, R. Rai, R. L. Mauck, J. A. Burdick, Biomate-
rials 2013, 34, 413.
[30] K. Boettcher, S. Grumbein, U. Winkler, J. Nachtsheim, O. Lieleg,
Rev. Sci. Instrum. 2014, 85, 93903.
[31] L.-S. Wang, C. Du, W. S. Toh, A. C. A. Wan, S. J. Gao, M. Kurisawa,
Biomaterials 2014, 35, 2207.
[32] J. S. Park, J. S. Chu, A. D. Tsou, R. Diop, Z. Tang, A. Wang, S. Li,
Biomaterials 2011, 32, 3921.
[33] A. Metters, J. Hubbell, Biomacromolecules 2005, 6, 290.
[34] A. Thomas, S. S. Müller, H. Frey, Biomacromolecules 2014, 15, 1935.
[35] F. Huang, Y.-G. Chen, Cell Biosci. 2012, 2, 1.
[36] J. Zhu, Biomaterials 2010, 31, 4639.
[1] M. Huber, S. Trattnig, F. Lintner, Invest. Radiol. 2000, 35, 573.
[2] D. J. Huey, J. C. Hu, K. A. Athanasiou, Science 2012, 338, 917.
[3] R. C. Lawrence, D. T. Felson, C. G. Helmick, L. M. Arnold, H. Choi,
R. A. Deyo, S. Gabriel, R. Hirsch, M. C. Hochberg, G. G. Hunder,
J. M. Jordan, J. N. Katz, H. M. Kremers, F. Wolfe, Arthritis Rheum.
2008, 58, 26.
[4] A. D. Woolf, B. Pfleger, Bull. W. H. O. 2003, 81, 646.
[5] B. Johnstone, M. Alini, M. Cucchiarini, G. R. Dodge, D. Eglin,
F. Guilak, H. Madry, A. Mata, R. L. Mauck, C. E. Semino,
M. J. Stoddart, Eur. Cells Mater. 2013, 25, 248.
[6] J. A. Burdick, G. D. Prestwich, Adv. Mater. 2011, 23, 41.
[7] C. Chung, J. A. Burdick, Tissue Eng., Part A 2009, 15, 243.
[8] X. Zheng Shu, Y. Liu, F. S. Palumbo, Y. Luo, G. D. Prestwich,
Biomaterials 2004, 25, 1339.
[9] A. Skardal, J. Zhang, G. D. Prestwich, Biomaterials 2010, 31, 6173.
[10] D. P. Nair, M. Podgórski, S. Chatani, T. Gong, W. Xi, C. R. Fenoli,
C. N. Bowman, Chem. Mater. 2014, 26, 724.
[11] R. K. Kainthan, J. Janzen, E. Levin, D. V. Devine, D. E. Brooks,
Biomacromolecules 2006, 7, 703.
[12] Y. Gañan, D. Macias, M. Duterque-Coquillaud, M. A. Ros,
J. M. Hurle, Development 1996, 122, 2349.
[13] B. Johnstone, T. M. Hering, A. I. Caplan, V. M. Goldberg, J. U. Yoo,
Exp. Cell Res. 1998, 238, 265.
[14] J. D. McCall, J. E. Luoma, K. S. Anseth, Drug Delivery Transl. Res. 2012,
2, 305.
[37] S. Stichler, T. Jungst, M. Schamel, I. Zilkowski, M. Kuhlmann,
T. Böck, T. Blunk, J. Teßmar, J. Groll, Ann. Biomed. Eng. 2017, 45,
273.
[38] S. Stichler, T. Böck, N. Paxton, S. Bertlein, R. Levato, V. Schill,
W. Smolan, J. Malda, J. Teßmar, T. Blunk, J. Groll, Biofabrication
2017, 9, 44108.
[15] B. V Sridhar, N. R. Doyle, M. A. Randolph, K. S. Anseth, J. Biomed.
Mater. Res. A 2014, 102, 4464.
[16] E. S. Place, R. Nair, H. N. Chia, G. Szulgit, E. Lim, M. M. Stevens,
Adv. Healthcare Mater. 2012, 1, 480.
[17] A. O. Fitton, J. Hill, D. E. Jane, R. Millar, Synthesis 1987, 1987,
1140.
[18] J.-L. Ruan, N. L. Tulloch, V. Muskheli, E. E. Genova, P. D. Mariner,
K. S. Anseth, C. E. Murry, Tissue Eng., Part C 2013, 19, 794.
[39] C. Hardwick, K. Hoare, R. Owens, H. P. Hohn, M. Hook, D. Moore,
V. Cripps, L. Austen, D. M. Nance, E. A. Turley, J. Cell Biol. 1992,
117, 1343.
[40] H. Zhu, N. Mitsuhashi, A. Klein, L. W. Barsky, K. Weinberg,
M. L. Barr, A. Demetriou, G. D. Wu, Stem Cells 2006, 24, 928.
[41] J. Visser, F. P. W. Melchels, J. E. Jeon, E. M. van Bussel, L. S. Kimpton,
H. M. Byrne, W. J. A. Dhert, P. D. Dalton, D. W. Hutmacher, J. Malda,
Nat. Commun. 2015, 6, 6933.
©
1700390 (13 of 13)
Macromol. Biosci. 2018, 1700390
2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim