Oxidatively degradable poly(thioketal urethane)/ceramic composite bone cements with bone-like strength

Article abstract of DOI:10.1039/c6ra24642g

Synthetic bone cements are commonly used in orthopaedic procedures to aid in bone regeneration following trauma or disease. Polymeric cements like PMMA provide the mechanical strength necessary for orthopaedic applications, but they are not resorbable and do not integrate with host bone. Ceramic cements have a chemical composition similar to that of bone, but their brittle mechanical properties limit their use in weight-bearing applications. In this study, we designed oxidatively degradable, polymeric bone cements with mechanical properties suitable for bone tissue engineering applications. We synthesized a novel thioketal (TK) diol, which was crosslinked with a lysine triisocyanate (LTI) prepolymer to create hydrolytically stable poly(thioketal urethane)s (PTKUR) that degrade in the oxidative environment associated with bone defects. PTKUR films were hydrolytically stable for up to 6 months, but degraded rapidly (<1 week) under simulated oxidative conditions in vitro. When combined with ceramic micro- or nanoparticles, PTKUR cements exhibited working times comparable to calcium phosphate cements and strengths exceeding those of trabecular bone. PTKUR/ceramic composite cements supported appositional bone growth and integrated with host bone near the bone-cement interface at 6 and 12 weeks post-implantation in rabbit femoral condyle plug defects. Histological evidence of osteoclast-mediated resorption of the cements was observed at 6 and 12 weeks. These findings demonstrate that a PTKUR bone cement with bone-like strength can be selectively resorbed by cells involved in bone remodeling, and thus represent an important initial step toward the development of resorbable bone cements for weight-bearing applications.

Full text of DOI:10.1039/c6ra24642g

RSC Advances
View Article Online
View Journal | View Issue
PAPER
Oxidatively degradable poly(thioketal urethane)/
ceramic composite bone cements with bone-like
strength
Cite this: RSC Adv., 2016, 6, 109414
Madison A. P. McEnery,a Sichang Lu,b Mukesh K. Gupta,a Katarzyna J. Zienkiewicz,b
Joseph C. Wenke,c Kerem N. Kalpakci,d Daniel A. Shimko,d Craig L. Duvalla
abe
*
and Scott A. Guelcher
Synthetic bone cements are commonly used in orthopaedic procedures to aid in bone regeneration
following trauma or disease. Polymeric cements like PMMA provide the mechanical strength necessary
for orthopaedic applications, but they are not resorbable and do not integrate with host bone. Ceramic
cements have a chemical composition similar to that of bone, but their brittle mechanical properties
limit their use in weight-bearing applications. In this study, we designed oxidatively degradable,
polymeric bone cements with mechanical properties suitable for bone tissue engineering applications.
We synthesized a novel thioketal (TK) diol, which was crosslinked with a lysine triisocyanate (LTI)
prepolymer to create hydrolytically stable poly(thioketal urethane)s (PTKUR) that degrade in the oxidative
environment associated with bone defects. PTKUR films were hydrolytically stable for up to 6 months,
but degraded rapidly (<1 week) under simulated oxidative conditions in vitro. When combined with
ceramic micro- or nanoparticles, PTKUR cements exhibited working times comparable to calcium
phosphate cements and strengths exceeding those of trabecular bone. PTKUR/ceramic composite
cements supported appositional bone growth and integrated with host bone near the bone-cement
interface at 6 and 12 weeks post-implantation in rabbit femoral condyle plug defects. Histological
evidence of osteoclast-mediated resorption of the cements was observed at 6 and 12 weeks. These
findings demonstrate that a PTKUR bone cement with bone-like strength can be selectively resorbed by
cells involved in bone remodeling, and thus represent an important initial step toward the development
of resorbable bone cements for weight-bearing applications.
Received 3rd October 2016
Accepted 3rd November 2016
DOI: 10.1039/c6ra24642g
www.rsc.org/advances
resorbable polymers have emerged as an alternative approach
that combines the ductile mechanical properties of polymers
1. Introduction
with the osteoconductivity of ceramics to provide mechanical
stability and integration with host bone.3
Injectable and settable bone cements restore function to bone
damaged by trauma or disease in a number of orthopaedic
procedures, such as vertebroplasty, repair of tibial plateau
fractures, and screw augmentation. Poly(methyl methacrylate)
(PMMA) bone cements exhibit mechanical properties exceeding
those of trabecular bone, and therefore provide mechanical
stability to damaged bone.1 However, PMMA cements are non-
resorbable and do not integrate with host bone. Ceramic bone
cements are osteoconductive and integrate with host bone, but
their brittle mechanical properties preclude their use in weight-
bearing applications.2 Thus, composites of ceramics with
Poly(ester urethane)s (PEUR) have been investigated as
injectable bone gras due to their injectability, settability,
tunable mechanical properties, and resorption to breakdown
products easily cleared from the body. PEUR gras set within
clinically relevant working times and attain strengths in the
range of 10–80 MPa.4,5 Lysine-derived PEUR composites incor-
porating ceramic particles or allogra bone set in situ with no
surgical complications and support bone remodeling in sheep,
rats, and rabbits.4–8 Previous work has shown lysine triisocya-
nate (LTI)-derived PEURs undergo autocatalytic hydrolytic
degradation in which the acidic breakdown products accelerate
resorption.9,10 A degradation mechanism that allows for a more
controlled and predictable degradation rate is desired to ensure
that the gra degrades at a rate complementary to bone
formation and remodeling.
aDepartment of Biomedical Engineering, Vanderbilt University, Nashville, TN, USA.
E-mail: scott.guelcher@vanderbilt.edu
bDepartment of Chemical and Biomolecular Engineering, Vanderbilt University,
Nashville, TN, USA
cU.S. Army Institute of Surgical Research, San Antonio, TX, USA
dMedtronic Spinal and Biologics, Memphis, TN, USA
eCenter for Bone Biology, Vanderbilt University Medical Center, Nashville, TN, USA
Bone remodeling is commonly achieved by creeping substi-
tution, a process by which osteoclasts resorb residual gra and
109414 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
RSC Advances
osteoblasts deposit new mineralized matrix near the gra–bone microparticles were washed in 95% acetone, triple rinsed with
interface.11–13 The normal endogenous bone healing cascade water, and dried under vacuum.
involves an initial hematoma formation that induces the
MC3T3 cells were supplied by ATCC (Manassas, VA). Gibco™
immune response accompanied by release of pro- a-MEM medium, penicillin/streptomycin (P/S) and a Pierce™
a
inammatory factors.14 The inammatory phase is followed by bicinchoninic (BCA) Protein Assay kit were purchased from
a so callus formation that is rapidly replaced by woven Thermo Scientic™ (Waltham, MA). Sterile phosphate buffered
mineralized bone.15 As a result of the inammatory response, saline (PBS) and 0.25% trypsin were purchased from Corning
reactive oxygen species (ROS) are generated by inltrating cells Cellgro (Manassas, VA) and fetal bovine serum (FBS) from
at the defect site.16,17 Mature osteoclasts at sites of active bone HyClone (Pittsburgh, PA). Reagents for cell xation including
remodeling are also associated with an increase in ROS.17–19 glutaraldehyde and osmium tetroxide were purchased from
These ndings suggest that hydrolytically stable biomaterials Fisher Scientic and Sigma Aldrich, respectively.
that degrade in response to cell-secreted ROS may be a useful
new approach for the design of cell-degradable bone cements.
2.2 Thioketal diol synthesis
Thioketals (TK), the sulfur analogs of ketals, degrade in
response to cell-secreted ROS to thiol decomposition products The schematic for thioketal diol synthesis is illustrated in
with low cytotoxicity.20,21 Poly(thioketal urethane) (PTKUR) foams Fig. 1A. Bismuth(III) chloride was added to a dry boiling ask
synthesized from a TK macrodiol (1000 g molꢀ1) have been re- that was subsequently dried with a hot air gun under vacuum
ported to support ROS-mediated degradation and healing in for about 5 minutes to ensure completely dry catalyst condi-
cutaneous wounds.20 However, macrodiol-based PTKURs cannot tions. The ask was then purged with nitrogen and le under
achieve bone-like strength or the number of degradable units a positive pressure with nitrogen for the remainder of the
afforded by the new single TK-containing crosslinker. Further- reaction. Anhydrous acetonitrile was charged to the ask to
more, TK-based biomaterials have not been previously investi- dissolve the catalyst. 2,2-Dimethoxypropane and thioglycolic
gated in bone. In this study, a novel low molecular weight acid were added to the ask, and the reaction was allowed to
thioketal (TK) diol was synthesized and utilized to formulate proceed for 24 hours while stirring at room temperature. The
PTKUR bone cements that are hydrolytically stable but degradable carboxyl-terminated intermediate was ltered with a Buchner
by cell-secreted ROS. The TK diol was reacted with LTI to form funnel, rotary evaporated (Buchi Rotovap R-200, 35 ꢁC), and
a moldable and settable PTKUR cement with bone-like strength dried under vacuum overnight. The carboxyl groups were then
using a low-toxicity iron(III) acetylacetonate gelling catalyst. To reduced to produce a hydroxyl-terminated TK. A 3-neck boiling
enhance the osteoconductivity of the PTKUR, it was combined ask was tted to a 10 ꢁC condenser capped with a 1-way glass
with two different types of ceramics: (1) 85% b-tricalcium phos- stopcock, a constant pressure dropping funnel, and a rubber
phate (b-TCP)/15% hydroxyapatite (HA) ceramic mini-granules stopper. The reactor was heated with a heat gun under vacuum
(MASTERGRAFT®, MG), or (2) nanocrystalline hydroxyapatite for about 5 minutes to ensure completely dry reaction condi-
(nHA) particles.22,23 The reactivity, rheological properties, tions. The reactor was then placed in an ice bath, purged with
mechanical properties, degradation rate, and cell proliferation dry nitrogen, and maintained under positive pressure with
response of the cements were assessed in vitro. The biocompati- nitrogen throughout the functionalization. Lithium aluminum
bility and remodeling of PTKUR/ceramic composite cements were hydride (LiAlH4) was added to the 3-neck boiling ask and
investigated in a rabbit femoral condyle plug defect model to dissolved in diethyl ether. Using anhydrous techniques, anhy-
assess material resorption and integration with the host bone.
drous tetrahydrofuran was added to the boiling ask containing
the carboxyl-terminated TK. The resulting solution was then
transferred to the dropping funnel and added to the LiAlH4
solution dropwise at 0 ꢁC. Aer all of the TK solution was added,
the ice bath was replaced ꢁwith an oil bath and the reaction
mixture was reuxed at 52 C for 6–8 hours. Unreacted LiAlH4
2. Materials and methods
2.1 Materials
Thioglycolic acid, 2,2-dimethoxypropane, bismuth chloride, was quenched by adding DI water dropwise followed by 1 M
lithium aluminum hydride, 3-caprolactone, nanocrystalline sodium hydroxide to aid in product extraction. By-products of
hydroxyapatite (nHA, <200 nm), and anhydrous solvents were the reaction were ltered using a Buchner funnel and ltration
purchased from Sigma-Aldrich (St. Louis, MO). The 3-capro- ask, and a separation funnel and diethyl ether were used to
lactone was treated with magnesium sulfate, and nHA was dried extract and isolate the TK diol product. The solvent was
under vacuum at 80 ꢁC for at least 24 hours prior to use. Acros removed by rotary evaporation and the product dried under
Organics iron(III) acetylacetonate (FeAA) was purchased from vacuum overnight for a completely dry, solvent-free TK diol.
Fisher Scientic and used as received. Lysine triisocyanate (LTI) Nuclear magnetic resonance spectroscopy (1H NMR, Bruker 400
was purchased from Jinan Haohua Industry Co., LTD (Jinan, MHz NMR) in dimethylsulfoxide (DMSO) and attenuated total
China) and carbon-treated in methyl-tert-butyl ether 3 times for reectance Fourier transform infrared spectroscopy (ATR-FTIR)
24 hours at 70 ꢁC to remove impurities. MasterGra (MG) veried the chemical structure of the TK diol. Titration of
particles supplied by Medtronic (Memphis, TN) were ground to a sample reacted with excess p-toluenesulfonyl isocyanate with
100–300 mm diameter particles using a mortar and pestle and tetrabutylammonium hydroxide was used to determine the
ltered between 100 and 300 mm sieves. The resulting hydroxyl (OH) number of the TK diol according to ASTM E1899-
This journal is © The Royal Society of Chemistry 2016
RSC Adv., 2016, 6, 109414–109424 | 109415

View Article Online
RSC Advances
Paper
Fig. 1 Synthesis and characterization of low molecular weight thioketal diol. (A) Synthesis scheme. (B) Characterization by NMR indicates that the
targeted molecular structure was obtained. (C) Viscosity of the TK diol is independent of shear rate.
08.24 The molecular weight (Mn) was calculated from the OH Swelling of lms with indices of 110, 125, and 140 was calcu-
number using eqn (1) where f is the functionality of the diol:
lated according to eqn (2), where Ms is the swollen mass and M0
is the initial mass. This information was used to determine
effects of index on extent of crosslinking.
56100 f
Mn ¼
(1)
OH number
Ms ꢀ M0
% swelling ¼
ꢂ 100%
(2)
M0
2.3 Quasi-prepolymer synthesis and characterization
A quasi-prepolymer was prepared according to methods previ-
ously described.25 Briey, a 2.5 : 1 molar ratio of LTI : TK
(3.75 : 1 NCO : OH equivalent ratio) was charged to a 100 mL
boiling ask and purged with nitrogen while stirring in an oil
bath at 45 ꢁC. TK diol was added to LTI drop-wise from a syringe
through a 16 G needle inserted through the rubber stopper. The
reaction was allowed to proceed for 3 hours yielding an LTI-TK
quasi-prepolymer. The NCO number was determined by titra-
tion according to ASTM D2572-97.26
2.4.1 Reaction kinetics and working time. The reaction
kinetics of the composite were assessed using methods
described previously.8,27 ATR-FTIR was used to evaluate the
reaction rate of the isocyanate-terminated LTI-TK prepolymer
with the other components of the composite individually by
quantifying the disappearance of the isocyanate peak (around
2270 cmꢀ1). The isocyanate peaks were calibrated to a standard
curve of known NCO concentrations to nd an initial rate
constant for each reaction during the rst 6 minutes. These rate
constants along with the initial concentrations of each
component were input into a Matlab program to calculate the
number of isocyanate and hydroxyl equivalents versus time
assuming second-order chemical kinetics. Isocyanate and
hydroxyl conversion versus time were determined from the
calculated numbers of equivalents.
2.4 Polyurethane/ceramic composite synthesis and
characterization
PTKUR/ceramic composites were fabricated by reactive liquid
molding and catalyzed using a 5% FeAA solution in 3-capro-
lactone. The isocyanate index (NCO : OH equivalent ratio ꢂ
100) was 140 for all materials.8 TK diol, LTI-TK prepolymer, and
55 wt% MG or 60 wt% nHA particles were hand-mixed to yield
a reactive paste. These concentrations of the ceramic particles
were selected as the maximum values that could be added while
maintaining a cohesive reactive paste. Once homogeneous,
0.06 wt% FeAA (in solution) was added to catalyze the reaction
between the LTI-TK prepolymer and the TK diol. The
morphology of the composite was veried by scanning electron
microscopy (Hitachi S4200 SEM) following gold sputter coating
of thin sections of sample (Cressington Q108) for 45 seconds at
30 mA.
The tack-free time was measured by hand and dened as
when the material no longer stuck to a metal spatula.7
2.4.2 Compressive mechanical properties. Samples for
compressive studies were prepared by injecting composites into
6 mm diameter tubes and compressing under a 0.96 kg weight
to ensure cohesion throughout initial cure.28 Samples were cut
to a height equal to 2 times their diameter (12 mm) using
a Buehler IsoMet Low Speed Saw (Lake Bluff, Il). Modulus and
strength were measured at various time points over a 2 week
period to determine when the composites were completely
crosslinked. Specimens were preloaded to 12 N and compressed
at a rate of 25 mm minꢀ1 using an MTS 858 Bionix Servohy-
draulic Test System (Eden Prairie, MN). The engineering stress
PTKUR lms (without ceramic) synthesized using varying
isocyanate indices were submerged in water for 2 weeks and
water uptake measured periodically by weighing the samples.
109416 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
RSC Advances
was calculated by dividing the load by the platen-contacting
surface area and the engineering strain determined by
dividing the displacement by initial sample height. The slope of
the linear-elastic portion of the resulting stress–strain curve was
identied as the compressive modulus and the maximum stress
as the compressive strength. When a maximum stress could not
be identied, the stress at 10% strain was reported.29
2.4.3 Degradation. The degradation characteristics of
PTKUR were assessed in hydrolytic and oxidative conditions. An
accelerated degradation medium comprising 20 wt% hydrogen
peroxide in 0.1 M cobalt chloride in DI water simulated the
environment produced by reactive oxygen species at the implant
site.20,30,31 PTKUR lms (17 mg) were immersed in 350 mL (1 mL/
50 mg initial sample) degradation media and placed on a shaker
table at 37 ꢁC. PTKUR degradation was compared to lysine-
derived poly(caprolactone urethane) (PCLUR), which was ex-
pected to undergo minimal hydrolytic degradation. Oxidative
media was changed every 72 hours when time points exceeded 3
days to ensure the presence of oxidizing radicals. Samples were
washed 3ꢂ with 100 mL DI water, dried under vacuum for at
least 48 hours, and weighed at various time points to determine
the degradation rate. Samples were gold sputter-coated for 45
seconds and imaged using SEM to visualize the change in
architecture with degradation.
2.7 Implantation of PTKUR/ceramic composite cements in
rabbits
Animal experiments were performed in compliance with rele-
vant laws and the guidelines of IBEX Preclinical Research, Inc.
(Logan, UT). The animal protocol was approved by the Institu-
tional Animal Care and Use Committee (IACUC) at IBEX
Preclinical Research, and all surgical and care procedures were
carried out under aseptic conditions per the approved IACUC
protocol. PTKUR/ceramic composites were evaluated in cylin-
drical femoral condyle plug defects in eight New Zealand White
rabbits weighing 4–5 kg. The reactive components (TK diol,
FeAA catalyst, LTI-TK prepolymer, MasterGra, and nHA) were
gamma-irradiated using a dose of approximately 25 kGY prior to
use. Aer administration of anesthesia, bilateral defects 6–8
mm deep ꢂ 5 mm diameter were drilled in the femoral condyle
of the distal femurs of 8 rabbits. PTKUR/ceramic composites
incorporating either MG or nHA (n ¼ 3) were mixed on site,
injected into the defect, and allowed to cure for 10 minutes
prior to closing the wound. Animals were euthanized and
femurs harvested at 6 and 12 weeks to evaluate healing and
polymer degradation. Micro-computed tomography (Scanco
mCT 50) was performed with a voxel size of 17.2 mm and
a threshold of 237 (386 mg HA per cm3) to match the intensity of
the native trabecular bone surrounding the defect. Histology
preparation was performed by Histion. Calcied samples were
embedded in PMMA and sections taken from the center of the
defect area; the sections were stained with Stevenel's Blue or
hematoxylin and eosin (H & E) to identify new bone formation
and cellular activity at the defect site.
2.5 Rheology
Viscosity was characterized using a TA InstrumentsꢁAR 2000ex
rheometer tted with 25 mm parallel plates at 25 C. For the
starting materials (TK diol and LTI-TK prepolymer), a small
sample was injected between the plates which were subse-
quently depressed to a gap size of 500 mm. A frequency sweep
was applied at a constant strain in the linear viscoelastic region
(0.2 for the TK diol and 0.5 for the quasi prepolymer). A Cox-
Merz transformation related the dynamic data to viscosity as
a function of shear rate. The rheological properties of uncata-
lyzed (non-reactive) composites were found using a gap size of
1.5 mm. A constant strain of 1% was applied to the composite
through a frequency sweep and a Cox-Merz transformation
applied to characterize injectability.
2.8 Statistical analysis
Anova with post hoc comparisons using Tukey's multiple
comparisons test was applied to compression testing data to
compare statistical differences with cure time. The Holm–Sidak
multiple comparison test was used to evaluate signicance in
total protein over time for each composite individually, and the
plot shows standard error of the mean (SEM). All other data was
plotted with standard deviation, and p < 0.05 was considered
statistically signicant.
2.6 In vitro characterization
The surface chemistry of PTKUR polymer lms was observed by
3. Results and discussion
3.1 Thioketal diol and quasi-prepolymer characterization
´
water contact angle using a Rame-Hart Goniometer (Mountain
Lakes, NJ) to predict cellular behavior at the material interface.
Cellular attachment was veried using SEM and proliferation The TK diol was synthesized following the two-step reaction
was observed using a BCA Protein Assay kit. MC3T3 cells were scheme in Fig. 1A. The characteristic NMR peak for the methyl
seeded (2 ꢂ 104 cells per mL) onto thin sections of MG and nHA (1.59 ppm) and hydroxyl (4.8 ppm) groups of the TK diol indi-
composites that were conditioned in complete aMEM medium cate that the targeted product was achieved (Fig. 1B), and an
with 10% FBS and 1% P/S overnight. Samples were submerged ATR-FTIR absorbance peak around 3400 cmꢀ1 conrmed
in 5% glutaraldehyde followed by 2% osmium tetroxide and an hydroxyl functionalization.20 The OH number was found to be
ethanol dehydration ladder to x for SEM aer 24 hours incu- 574 mg KOH per g, which corresponds to a molecular weight of
bation. To measure proliferation, samples were taken from 196 g molꢀ1 (eqn (1)). These data conrm that the desired
culture at 1, 4, and 7 days. Samples were transferred to a new product with a theoretical molecular weight of 196.3 g molꢀ1
well, washed with PBS, and the cells trypsinized. Cell pellets was achieved. This low-molecular weight TK diol had a viscosity
were lysed using RIPA buffer to extract the cellular protein. The of 0.11 Pa s at a shear rate of 5 sꢀ1 and exhibited near Newtonian
BCA kit was used to quantify total protein at each time point.32 behavior at shear rates below 100 sꢀ1 (Fig. 1C).
This journal is © The Royal Society of Chemistry 2016
RSC Adv., 2016, 6, 109414–109424 | 109417

View Article Online
RSC Advances
Paper
A quasi-prepolymer was synthesized to improve handling by catalyst (25 : 1 gel : blow, Fig. 2F and G) compared to previously
increasing LTI viscosity, lowering the reaction exotherm, and investigated amine-based catalysts with high blowing power
minimizing phase separation during polymerization. TK diol (1 : 20 gel : blow).10,13,23,34
was reacted with a 2.5 molar excess of LTI to form an LTI-TK
PTKUR lms were made by mixing TK and LTI-TK prepol-
prepolymer (Fig. 2A). The excess of LTI greater than 2 renders ymer with iron catalyst without incorporating ceramic particles.
this component a quasi-prepolymer, although it will be referred The polymer lm exhibited a contact angle of 70.2ꢁ indicating
to as a prepolymer in this study.33 The LTI-TK prepolymer a moderately hydrophobic surface. Films of indices 110, 125,
exhibited Newtonian behavior, but the viscosity of 61 Pa s and 140 all swelled less than 3.5% aer soaking in water for 2
(measured at 5 sꢀ1, Fig. 2B) was considerably greater than that weeks and the differences between them were not signicant.
measured for TK diol or LTI (0.036–0.061 Pa s).25 The % NCO Since there was no difference in swelling and the swelling was
number of the prepolymer determined by titration was 25.1%, less than 5% for all samples, all of the indices were considered
which is slightly lower than the theoretical NCO number of suitable for use in vivo. An index of 140 was chosen for the
26.7% based on stoichiometry.
studies in this work to ensure complete crosslinking and a more
rigid composite as reported previously.8,28,35
The reactivity of the polymer was investigated using ATR-
FTIR. The second-order rate constant (ki, eqn (3)) of each
component was calculated based on the initial isocyanate
concentration (C0) and the disappearance of the isocyanate
peak (C).8,36
3.2 Composite characterization
Crosslinked PTKUR composites (Fig. 2C) incorporating either
MG or nHA particles were fabricated according to the schematic
in Fig. 2D. Fig. 2E shows the initial (e.g., uncatalyzed) dynamic
viscosities of both MG and nHA composites up to shear rates
of 100 sꢀ1. Both materials exhibit shear thinning behavior that
is more prominent at lower shear rates, which enhances
injectability, and have viscosities of 20–25 Pa s at a shear rate of
100 sꢀ1. SEM images of the composites showed minimal
porosity was achieved using a low-toxicity, iron-based gelling
1
1
¼ kit þ
(3)
C
C0
The catalyst was reduced by half (compared to the in vivo
studies) for the reactivity experiments to slow the reaction,
Fig. 2 Synthesis of poly(thioketal urethane) (PTKUR)/ceramic composites. (A) Synthesis scheme for LTI-TK prepolymer. (B) Viscosity of the LTI-
TK prepolymer is independent of shear rate. (C) Reaction of TK diol with LTI-TK prepolymer to form a crosslinked PTKUR network. (D) Fabrication
of PTKUR/ceramic composites by mixing LTI-TK prepolymer, TK diol, and ceramic particles (MG or nHA). (E) The viscosity of uncatalyzed (non-
reactive) LTI-TK/TK diol/ceramic mixtures decreases with increasing shear rate, providing evidence of shear-thinning behavior. (F–G) SEM
images of (F) MG and (G) nHA composites show lack of porosity. Due to their relatively large size (100–300 mm), MG particles (light grey and
labeled MG) can be distinguished from the PTKUR phase (dark grey).
109418 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
RSC Advances
Fig. 3 Kinetics of the setting reaction. (A) Second-order reaction kinetics plot of inverse NCO concentration (1/[NCO], g per equivalents NCO)
versus time. The specific reaction rate (k) was calculated from the slope of the 1/[NCO] versus time plot, which is anticipated to be linear for
second-order chemical reactions. The rate constant of the LTI-TK prepolymer-TK diol reaction (,) is substantially greater than that measured
for MG (O), nHA (B), or water (>). (B) Using the rate constant for the dominant reaction TK diol + LTI-TK prepolymer and the second-order
kinetics model, the conversions of the NCO and OH functional groups were calculated versus time.
Fig. 4 Mechanical properties of PTKUR/ceramic composites under static compressive loading. (A) Yield strength and (B) modulus of PTKUR/MG
composites measured versus time for up to two weeks. (C) Yield strength and (D) modulus of PTKUR/nHA composites measured versus time for
up to two weeks. Maximum compressive properties were achieved after 1 week cure time. The physical appearance of (E) MG and (F) nHA
composites after compressive testing supports this finding.
This journal is © The Royal Society of Chemistry 2016
RSC Adv., 2016, 6, 109414–109424 | 109419

View Article Online
RSC Advances
Paper
which was necessary to investigate the reaction mechanisms. anticipated to slowly react with the ceramic and environmental
Fig. 3A shows the calculation of the initial rate constant (ki) for water, as reported previously for allogra bone composites,9 due
each reaction from the slope of the 2nd order rate plot, in which to the substantially lower reactivity of the LTI-TK prepolymer with
the inverse concentration of NCO equivalents (g per equiv. NCO) these components. The tack-free time was determined by hand to
is plotted versus time. The plot is linear for the rst 6 minutes of be 6 minutes37 aer mixing, which is comparable to the setting
the reaction, which conrms that the reactions are second order times for calcium phosphate cements.2
as anticipated.8,36 Further, the very small slope for MG, nHA,
MG composites achieved a maximum compressive yield
and water with LTI-TK indicates these components have very strength of 40 ꢃ 7 MPa and modulus of 936 ꢃ 46 MPa aer 1
low reactivity, and thus they were not included in the conversion week of curing in air at RT (Fig. 4A and B). These composites had
calculations. The relatively high rate constant for the LTI-TK/TK an initial strength of 7.7 MPa and modulus of 36 MPa aer 16
gelling reaction compared to the LTI-TK/water blowing reaction hours curing at RT. nHA composites exhibited initial strength
(25 : 1 gel : blow ratio) conrms the preferential gelling activity and moduli much greater than MG composites as expected due
of the iron acetylacetonate (FeAA) catalyst compared to the tri- to the increased surface area-to-volume ratio of the mechanically
ethylene diamine (TEDA) catalyst investigated previously (1 : 20 robust nanoparticles.38,39 These cements had an initial
gel : blow).8,35 The concentration of LTI-TK prepolymer (I) and compressive yield strength and modulus of 31 ꢃ 3 MPa and 452
TK diol (D) were calculated as:
ꢃ 35 MPa, respectively. The composites reached a yield strength
of 90 ꢃ 6 MPa and modulus of 1267 ꢃ 277 MPa aer 1 week
(Fig. 4C and D). The mechanical properties of both composites
increased over the rst week, indicating that complete cross-
linking was achieved 1 week aer fabrication. The physical
appearance of the composites post-compression supports this
nding. MG composites up to 48 hours cure time experience
some elastic recovery to their original shape around 30 minutes
post-compression, where plastic deformation is more evident in
the 1 and 2 week samples (Fig. 4E). These changes in resilience
are less apparent in the stronger nHA samples (Fig. 4F).
Trabecular bone is reported to have a compressive strength of
5–10 MPa and modulus of 50–400 MPa.40–42 Therefore, the initial
compressive strength and modulus of MG composites are close
to the properties of trabecular bone and nHA composites exceed
dCD
dt
dCI
dt
¼
¼ ꢀkDCDCIM
(4)
where Cj is the concentration of each component (I or D, g
equiv.ꢀ1 minꢀ1) and M is the mass of the composite (g). The
conversions of LTI-TK prepolymer and TK diol were calculated
from the second-order kinetic model as:
Cj0 ꢀ Cj
xj ¼
(5)
Cj0
Conversion of NCO and OH groups are shown in Fig. 3B. The
hydroxyl groups in the TK diol are completely converted and an
excess of isocyanate functional groups remain, as anticipated
from the high isocyanate index of 140. The excess isocyanate is
Fig. 5 Degradation of PCLUR and PTKUR films. (A) PTKUR and PCLUR films were incubated in hydrolytic (open symbols) or oxidative (closed
symbols) conditions. After 4 months, PCLUR and PTKUR films substantially degraded in oxidative medium, while no degradation was observed in
hydrolytic medium. Furthermore, PKTUR films degraded after only 4 days incubation time in oxidative medium (inset). (B–D) SEM images show
the effects of oxidative degradation on the architecture of the PTKUR films after (B) 24 h, (C) 48 h, and (D) 72 h.
109420 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
RSC Advances
these properties. Both composites are mechanically stronger (Fig. 6B) composites aer 24 h culture. Cell proliferation on the
than trabecular bone aer 1 week. lms was assessed for up to 7 days post-seeding by measuring
The degradation rate of PTKUR lms under hydrolytic and the change in total protein with time. Fig. 6C shows that the cell
oxidative conditions was measured in vitro. PTKUR was population on MG composites increased with time, but the
compared to PCLUR as this material has been shown to degrade differences were not signicant. Cells proliferated on nHA
slowly in vivo.5 PTKUR degraded completely aer 4 days in vitro composites, as evidenced by the increase in total protein from
in oxidative media (Fig. 5A inset) but experienced minimal day 1 to day 7. Hydroxyapatite is the primary mineral compo-
hydrolytic degradation in PBS aer 4 months (Fig. 5A). SEM nent in bone, and therefore MC3T3 cells were expected to
images of PTKUR aer 24, 48, and 72 hours in oxidative media adhere and proliferate on scaffolds comprising 60 wt% nHA.43
show morphological changes in the lms in response to While MG contains only 15% HA, the beta-tricalcium phosphate
degradation, as evidenced by the formation of pores in the (b-TCP) component is also an osteoconductive ceramic.44,45 The
material (Fig. 5B–D). PCLUR degraded minimally in PBS as ex- slower proliferation rate of MC3T3 cells on MG composites
pected and did not completely degrade in oxidative media until could potentially be explained by the relatively large size (100–
about 5 months.
300 mm) of the MG microparticles, resulting in relatively large
areas of polymer that is less osteoconductive than the ceramic.
In contrast, phase-separation of the nHA and polymer compo-
nents was not observed in the nHA composites, suggesting that
the nHA is more uniformly distributed due to its smaller
particle size.
3.3 In vitro characterization
The osteoblast precursor MC3T3 cell line was used in all in vitro
studies to assess cell attachment and proliferation. SEM images
show that cells attached and spread on MG (Fig. 6A) and nHA
Fig. 6 MC3T3 cells (arrows) attached and spread on (A) MG and (B) nHA composites after 24 h incubation. Scale bar ¼ 50 mm. (C) Measurements
of total protein versus time indicate that cells proliferated faster on nHA composites.
Fig. 7 Images of transverse mCT sections of PTKUR/MG and PTKUR/nHA composite cements explanted at 6 and 12 weeks. Higher magnification
images of the defect periphery show evidence of trabecular infiltration (single white arrows) and trabecular densification (double white arrows).
Scale bar ¼ 1 mm.
This journal is © The Royal Society of Chemistry 2016
RSC Adv., 2016, 6, 109414–109424 | 109421

View Article Online
RSC Advances
Paper
Fig. 8 Images of transverse histological sections of PTKUR/MG and PTKUR/nHA composite cements. Low-magnification (2ꢂ) images of
cements at 12 weeks show appositional growth of dense trabecular bone near the host bone-cement interface. Higher magnification (20–40ꢂ)
images of PTKUR/MG cements at 6 and 12 weeks reveal evidence of residual MG (dark grey) particles, resorption of PTKUR (P, light grey), cellular
infiltration (blue), osteoid (arrows), and new bone (NB, red) formation. Similar observations were made for PTKUR/nHA cements, but the nHA
particles could not be distinguished due to their small size. Resorption of the cement (CM) was evident in the histological sections.
Low-magnication (2ꢂ) images of histological sections stained
3.4 Tissue and cellular response in the femoral condyle
defect model
with Stevenel's Blue stain show appositional growth of dense
trabecular bone near the host bone-cement interface at 12
weeks (Fig. 8). The materials were well-tolerated by the host
tissue and no adverse reactions were evident. Higher magni-
cation (20–40ꢂ) images show remodeling and integration of the
cements with host bone near the surface of the cements at 6 and
12 weeks. Due to the relatively large size of MG particles (100–
300 mm), the PKTUR (P) and MG particles (MG) could be
distinguished in the histological sections. PTKUR resorption
The composites were injected into femoral condyle plug defects
in rabbits to assess bone healing and cement resorption. In vivo
X-ray imaging immediately following the surgery indicated good
placement and complete ll of the defect with the materials.
mCT images of MG and nHA cements at 6 and 12 weeks are
shown in Fig. 7. Trabecular densication was evident at the
periphery of the defects, indicating that the material was inte-
grated with the host bone and initiating a healing response.
Fig. 9 Resorption of PTKUR/MG and PTKUR/nHA cements mediated by osteoclast-like cells at 6 and 12 weeks. Osteoclasts are identified as large
(>50 mm) multi-nucleated (nucleus stains dark blue) cells near the host bone-cement interface.
109422 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016

View Article Online
Paper
RSC Advances
near the interface was observed, resulting in cellular inltration U.S. Department of Energy and USAMRMC. This work was also
and new bone (NB, red) formation. Osteoid (arrows) was supported by the National Institutes of Health under award
observed near the surface of the residual PTKUR. While the nHA numbers AR064304 and EB019409 and the National Science
particles were too small to distinguish in the histological Foundation Graduate Research Fellowship Program under Grant
sections, similar phenomena were observed for nHA cements. No. 1445197. The rabbit study was funded by Medtronic Spinal
Resorption of the cement (CM) near the host bone interface and Biologics. Any opinions, ndings, and conclusions or
resulted in new bone formation and osteoid was evident near recommendations expressed in this material are those of the
the surface of the cement.
author(s) and do not necessarily reect the views of the National
Resorption appeared to be cell-mediated, as indicated by the Institutes of Health or the National Science Foundation.
irregular morphology of the cement (black arrows, Fig. 9) and
the presence of osteoclast-like cells, identied as large (>50 mm)
multi-nucleated (nuclei stained dark blue, Fig. 9) cells, near the
References
¨
bone-cement interface. In contrast, negligible degradation was
observed in the interior of the cement. These ndings are
consistent with the notion that resorption of the cements was
surface-mediated by osteoclasts and/or macrophages through
an ROS mechanism (Fig. 5) as we have reported previously for
PTKUR scaffolds implanted in cutaneous wounds.20 Due to their
relatively large size (100–300 mm), MG particles can be observed
in the SEM images as a distinct phase (“MG”, Fig. 2F). Since
osteoclasts are smaller than MG particles, resorption of the MG
and PTKUR phases is anticipated to proceed at different rates.
In contrast, the smaller nHA particles (100 nm) cannot be
distinguished from the PTKUR component (Fig. 2G). At the
length scale of an osteoclast, the nHA composites comprise
a single phase and are anticipated to resorb at a rate averaged
over the resorption rates of the individual nHA and PTKUR
components. Thus, differences in MG and nHA particle size
may affect gra resorption. Due to the low (<10%) porosity of
the cements, the rate of cellular inltration and remodeling was
slow. Increasing the porosity would be anticipated to accelerate
inltration of cells and consequent new bone formation.4
1 K.-D. Kuhn, Bone cements: up-to-date comparison of physical
and chemical properties of commercial materials, Springer,
2000.
2 M. Bohner, Eur. Cells Mater., 2010, 20, 3–10.
3 A. J. Wagoner Johnson and B. A. Herschler, Acta Biomater.,
2011, 7, 16–30.
4 R. Adhikari, P. A. Gunatillake, I. Griffiths, L. Tatai,
M.
Wickramaratna,
R. T. Mayadunne, J. Field, M. McGee and T. Carbon,
Biomaterials, 2008, 29, 3762–3770.
5 J. E. Dumas, T. Davis, G. E. Holt, T. Yoshii, D. S. Perrien,
J. S. Nyman, T. Boyce and S. A. Guelcher, Acta Biomater.,
2010, 6, 2394–2406.
6 I. C. Bonzani, R. Adhikari, S. Houshyar, R. Mayadunne,
P. Gunatillake and M. M. Stevens, Biomaterials, 2007, 28,
423–433.
7 J. E. Dumas, K. Zienkiewicz, S. A. Tanner, E. M. Prieto,
S. Bhattacharyya and S. A. Guelcher, Tissue Eng., Part A,
2010, 16, 2505–2518.
8 J. M. Page, E. M. Prieto, J. E. Dumas, K. J. Zienkiewicz,
J. C. Wenke, P. Brown-Baer and S. A. Guelcher, Acta
Biomater., 2012, 8, 4405–4416.
9 J. E. Dumas, E. M. Prieto, K. J. Zienkiewicz, T. Guda,
J. C. Wenke, J. Bible, G. E. Holt and S. A. Guelcher, Tissue
Eng., Part A, 2014, 20, 115–129.
S.
Houshyar,
T.
Moore,
4. Conclusion
In this study, a novel low-molecular weight thioketal diol
crosslinker was synthesized to prepare cell-degradable bone
cements with initial bone-like strength. The cements exhibited 10 A. E. Hafeman, K. J. Zienkiewicz, A. L. Zachman, H.-J. Sung,
initial compressive strength exceeding that of trabecular bone,
L. B. Nanney, J.
M. Davidson and S. A. Guelcher,
working times comparable to commercial bone cements (5–10
Biomaterials, 2011, 32, 419–429.
min), and degradation in response to reactive oxygen species 11 P. P. de Bruyn and W. T. Kabisch, Am. J. Anat., 1955, 96, 375–417.
secreted by cells. When implanted into femoral condyle plug 12 J. E. Dumas, P. B. BrownBaer, E. M. Prieto, T. Guda,
defects in rabbits, the cements supported appositional new
bone growth, osteoclast-mediated resorption, and integration
R. G. Hale, J. C. Wenke and S. A. Guelcher, Biomed. Mater.,
2012, 7, 024112.
with host bone. These ndings highlight the potential of pol- 13 E. M. Prieto, A. D. Talley, N. R. Gould, K. J. Zienkiewicz,
y(thioketal urethane)/ceramic composite bone cements for
repair of bone damaged by trauma or disease.
S. J. Drapeau, K. N. Kalpakci and S. A. Guelcher, J. Biomed.
Mater. Res., Part B, 2015, 103, 1641–1651.
´
14 F. Loi, L. A. Cordova, J. Pajarinen, T.-h. Lin, Z. Yao and
S. B. Goodman, Bone, 2016, 86, 119–130.
15 K. Schmidt-Bleek, B. J. Kwee, D. J. Mooney and G. N. Duda,
Tissue Eng., Part B, 2015, 21, 354–364.
Acknowledgements
The authors acknowledge Dr Mike Larson and his team at IBEX
Preclinical Research, Inc. for performing the rabbit surgeries 16 S. Nukavarapu, J. Freeman and C. Laurencin, Regenerative
and providing post-operative care. This research was supported
in part by an appointment to the Student Research Participation
Engineering of Musculoskeletal Tissues and Interfaces,
Woodhead Publishing, 2015.
Program at the U.S. Army Medical Research and Materiel 17 S. A. Sheweita, K. I. Khoshhal and H. H. Baghdadi, in Systems
Command administered by the Oak Ridge Institute for Science
and Education through an interagency agreement between the
Biology of Free Radicals and Antioxidants, Springer, 2014, pp.
2973–2995.
This journal is © The Royal Society of Chemistry 2016
RSC Adv., 2016, 6, 109414–109424 | 109423

View Article Online
RSC Advances
Paper
18 I. Garrett, B. Boyce, R. Oreffo, L. Bonewald, J. Poser and 31 E. M. Christenson, J. M. Anderson and A. Hiltner, J. Biomed.
G. Mundy, J. Clin. Invest., 1990, 85, 632. Mater. Res., Part A, 2004, 70, 245–255.
19 L. L. Key, W. C. Wolf, C. M. Gundberg and W. L. Ries, Bone, 32 R. Guo, S. Lu, J. M. Page, A. R. Merkel, S. Basu, J. A. Sterling
1994, 15, 431–436.
and S. A. Guelcher, Adv. Healthcare Mater., 2015, 4, 1826–
20 J. R. Martin, M. K. Gupta, J. M. Page, F. Yu, J. M. Davidson,
1832.
S. A. Guelcher and C. L. Duvall, Biomaterials, 2014, 35, 3766– 33 Dow polyurethanes
-
prepolymer denition, http://
3776.
dowac.custhelp.com/app/answers/detail/a_id/5082/ꢄ/dow-
polyurethanes-prepolymer-denition, (accessed 05/23/2016,
2016).
21 D. S. Wilson, G. Dalmasso, L. Wang, S. V. Sitaraman,
D. Merlin and N. Murthy, Nat. Mater., 2010, 9, 923–928.
22 R. Z. LeGeros, Clin. Orthop. Relat. Res., 2002, 395, 81–98.
23 A. D. Talley, M. A. McEnery, K. N. Kalpakci, K. J. Zienkiewicz,
D. A. Shimko and S. A. Guelcher, J. Biomed. Mater. Res., Part
B, 2016.
34 S. A. Guelcher, V. Patel, K. M. Gallagher, S. Connolly,
J. E. Didier, J. S. Doctor and J. O. Hollinger, Tissue Eng.,
2006, 12, 1247–1259.
35 A. J. Harmata, C. L. Ward, K. J. Zienkiewicz, J. C. Wenke and
S. A. Guelcher, J. Mater. Res., 2014, 29, 2398–2407.
24 ASTM International Standard E1899-08: Standard test
method for hydroxyl groups using reaction with p- 36 S. Parnell, K. Min and M. Cakmak, Polymer, 2003, 44, 5137–
toluenesulfonyl isocyanate (TSI) and potentiometric
titration with tetrabutylammonium hydroxide, 2016.
25 S. A. Guelcher, A. Srinivasan, J. E. Dumas, J. E. Didier,
5144.
37 K. Ashida, Polyurethane and related foams: chemistry and
technology, CRC press, 2006.
S. McBride and J. O. Hollinger, Biomaterials, 2008, 29, 38 H.-Y. Mi, X. Jing, M. R. Salick, T. M. Cordie, X.-F. Peng and
1762–1775. L.-S. Turng, J. Mater. Sci., 2014, 49, 2324–2337.
26 ASTM International Standard D2572-97: Standard test 39 K. Rezwan, Q. Chen, J. Blaker and A. R. Boccaccini,
method for isocyanate groups in urethane materials or
prepolymers, 2010.
27 R. Guo, C. L. Ward, J. M. Davidson, C. L. Duvall, J. C. Wenke
and S. A. Guelcher, Biomaterials, 2015, 54, 21–33.
28 A. J. Harmata, S. Uppuganti, M. Granke, S. A. Guelcher and
Biomaterials, 2006, 27, 3413–3431.
40 S. Fernando, M. McEnery and S. Guelcher, Advances in
Polyurethane Biomaterials, 2016, p. 481.
41 V. Karageorgiou and D. Kaplan, Biomaterials, 2005, 26, 5474–
5491.
J. S. Nyman, J. Mech. Behav. Biomed. Mater., 2015, 51, 345– 42 D. S. Muggli, A. K. Burkoth and K. S. Anseth, J. Biomed.
355. Mater. Res., 1999, 46, 271–278.
29 ASTM International Standard D1621-16: standard test 43 J. Buckwalter, M. Glimcher, R. Cooper and R. Recker, J. Bone
method for compressive properties of rigid cellular
plastics, 2016.
30 M. A. Schubert, M. J. Wiggins, J. M. Anderson and A. Hiltner,
J. Biomed. Mater. Res., 1997, 34, 519–530.
Jt. Surg., Am. Vol., 1995, 77, 1256–1275.
44 R. D. A. Gaasbeek, H. G. Toonen, R. J. van Heerwaarden and
P. Buma, Biomaterials, 2005, 26, 6713–6719.
45 P. V. Giannoudis, H. Dinopoulos and E. Tsiridis, Injury,
2005, 36, S20–S27.
109424 | RSC Adv., 2016, 6, 109414–109424
This journal is © The Royal Society of Chemistry 2016