Acta Biomaterialia 92 (2019) 325–335

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Full length article

Mechanical and biological properties of ZnO, SiO2, and Ag2O doped

plasma sprayed hydroxyapatite coating for orthopaedic and dental
applications
Ashley A. Vu, Samuel Ford Robertson, Dongxu Ke, Amit Bandyopadhyay, Susmita Bose ⇑
W. M. Keck Biomedical Materials Research Laboratory, School of Mechanical and Materials Engineering, Washington State University, Pullman, WA 99164, United States

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

Article history: In this study, we explored a ternary dopant system utilizing 0.25 wt% ZnO to induce osteogenesis, 0.5 wt%
Received 6 January 2019 SiO2 to induce angiogenesis, and 2.0 wt% Ag2O to provide secondary infection control within a plasma
Received in revised form 23 April 2019 assisted hydroxyapatite coating for orthopaedic or dental applications. The objective of this study was
Accepted 7 May 2019
to understand the effects of ZnO, SiO2, and Ag2O dopants on the mechanical and biological properties
Available online 10 May 2019
of hydroxyapatite (HA) coatings on titanium (Ti). Coatings were deposited using a 30 kW plasma spray
system equipped with a supersonic nozzle to produce above standard coating bond strengths of
Keywords:
24 ± 2 MPa on Ti6Al4V and 22 ± 1 MPa on commercially pure Ti substrates. Antibacterial properties were
Hydroxyapatite
Plasma spray
revealed in vitro against E. coli and S. aureus. The ternary dopant system was implanted in 18 male
Titanium Sprague-Dawley rats with timepoints of 5 and 10 weeks. By week 5, ZnSiAg-HA produced 32% bone min-
Osseointegration eralization of 68% total bone formation compared to only 11% bone mineralization of 55% total bone for-
Dopants mation in the undoped coating. This system can be employed for replacement surgeries and revision
surgeries to reduce healing time and enhance osseointegration.

Statement of Significance

Total hip replacements increased 124% from 2000 to 2010 with an ever-increasing rate due to the rise in
average life span and an escalation in surgeries for young patients. Replacement surgeries come with the
risk of rejection, poor integration, and infection. This study incorporates biologically relevant metallic
oxides of ZnO, SiO2, and Ag2O within a hydroxyapatite coating on titanium deposited using a radio fre-
quency induction plasma spray. A ternary dopant system has not been explored in the current literature
and little is known about these particular dopants in vivo. This proposed system can be employed for
replacement surgeries to lower healing time and enhance osseointegration between implant and host
tissue.
Ó 2019 Published by Elsevier Ltd on behalf of Acta Materialia Inc.

1. Introduction done, implant integration still needs improvement to minimize

The number of total hip replacements increased 124% from
2000 to 2010 [1]. With any replacement of a once functioning body
part comes the risk of rejection, poor integration, and infection.
Patients with infected prosthesis have been found to stay in the
hospital for over 2 times longer which significantly increases the
overall cost of upwards of 3 times [2]. Implant success is an impor-
tant area of study and despite the tremendous research efforts
⇑ Corresponding author.
E-mail address: sbose@wsu.edu (S. Bose).
healing time, increase osseointegration, and reduce the risk of
infection.
Titanium (Ti) and its alloys are ideally suited for load bearing
bone implants due to their appropriate strength for application,
corrosion resistance, fatigue resistance, and biocompatibility [3–
5]. To address the bioinert nature of Ti, mechanical, chemical, or
physical surface modifications are implemented to modify and
improve implants while maintaining desirable properties of the
substrate material. This study utilized hydroxyapatite (HA) (Ca10
(PO4)6 (OH)2) coatings to modify the surface of Ti implants. HA is
a calcium phosphate with a comparable composition to natural

https://doi.org/10.1016/j.actbio.2019.05.020
1742-7061/Ó 2019 Published by Elsevier Ltd on behalf of Acta Materialia Inc.
326 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335
bone and has been reported to bond to living tissue as well as sup-
port bone ingrowth [6–9]. HA improves implant osteoconductivity
and bioactivity but lacks high osteoinductivity [10].
One way to improve osteoconductivity and potentially enhance
osteoinductivity is by adding dopants to a HA coating system [11].
These dopants are added in small amounts and are biologically rel-
evant metal oxides. Dopants, such as strontium, magnesium, sili-
con, and zinc, have been noted to have the ability to tailor
strength and strength degradation as well as enhance biological
responses in vitro and in vivo [12–15]. Important trace elements
are needed in osteogenesis whereby zinc and silicon are considered
two essential elements. Zinc is released during the skeletal break-
down process. Zinc has been reported to inhibit osteoclastic bone
resorption as well as induce osteoblastogenesis and osteoblastic
differentiation and mineralization [16–19]. Silicon has been
researched to have a strong stimulatory effect on cellular activities
including proliferation, differentiation, and mineralization of
osteoblast cells and inducing osteogenic differentiation of mes-
enchymal stem cells [20–22]. Silicon has also been noted to have
angiogenic capabilities. One study found that the presence of sili-
con induced VEFG expression in human dermal fibroblasts which
upregulated nitric oxide synthase and nitric oxide production in
human endothelial cells [23]. Another study demonstrated angio-
genic properties through employing a calcium silicate in a rabbit
femur defect [24].
Doping is not only beneficial for osteoconductivity but also for
preventative care. Implants are highly susceptible to infections
due to the localized host defense being compromised [25,26]. HA
is not an antibacterial material however infection control can be
achieved through doping. Silver has the ability to prevent bacterial
adhesion which is considered more advantageous than trying to
treat high resistive biofilm after its formation. Biofilm formation
will often lead to necessary prosthetic removal [27–30].
The objective of this study was to understand the effects of
ZnO, SiO2, and Ag2O dopants on the mechanical and biological
properties of hydroxyapatite coatings on titanium. Our group has
studied unary and binary systems employing optimized concen-
trations of various biologically relevant metal oxides but have
yet to explore a ternary dopant system [11,14,15,30]. We also have
not investigated a doped plasma coating system using either ZnO
or Ag2O in vivo. Others have also investigated binary systems using
Zn, Si, and Ag [31–34]. In this study, each dopant was added in
their oxide form with respective weight percentages of 0.25 wt%,
0.5 wt%, and 2 wt% for ZnO, SiO2, and Ag2O. The hypothesis was
the system will exhibit improved osseointegration in vivo. Surface
characterizations include Field Emission Scanning Electron Micro-
scopy (FESEM), X-ray diffraction (XRD), Fourier Transform Infra-
Red (FTIR) spectroscopy, surface roughness, and coating bond
strength. A silver ion release study was employed to ensure the
release of ions would be at nontoxic levels as well as a bacterial
study to showcase the antibacterial properties of the coating
in vitro. Finally, an in vivo rat distal femur model study was done
with 18 male Sprague Dawley rats. With the number of implants
needed worldwide, implant success is crucial. Improving implant
integration with surface modifications and incorporating antibi-
otic properties reduce the risk of implant failure and infection
which help reduce the need for future, costly, and painful revision
surgeries for patients.
2. Materials and methods
2.1. Coating preparation
Commercial HA powder (Monsanto, USA) was doped with
0.25 wt% ZnO, 0.5 wt% SiO2, and 2 wt% Ag2O. This dopant powder
and coating will be referenced as ZnSiAg-HA from now onward.
These values were chosen from previous optimization studies
based on cell interaction and physiochemical properties [30,35–
37]. The doped powder was prepared by milling the dopants with
HA in a 250 mL polypropylene Nalgene bottle containing a 2:1 wt
ratio of doped powder to 5 mm diameter zirconia milling med-
ium for 2 h at 80 rpm. The doped powder was then loaded into
the axial powder feed system of the 30 kW inductively coupled
radio frequency plasma spray system (Tekna Plasma System,
Canada). Ti-6Al-4V (Ti64) and commercially pure titanium (CpTi)
(President Titanium, MA) were cut into circular samples
(25.4 mm and 12.5 mm diameter, 2 mm thick) using a waterjet,
sandblasted using 80 grit garnet sand in a standard sandblast
cabinet, and washed ultrasonically in both DI water and acetone
prior to coating. The coatings were prepared at 25 kW plate
power with a 110 mm spray distance from the supersonic plasma
nozzle to the samples. Argon was utilized as central gas with a
flow rate of 25 standard 1 min 1 (s.L.p.m.) and as carrier gas with
a flow rate of 13 s.L.p.m. Sheath gas was a mixture of argon and
hydrogen at 60 s.L.p.m. and 6 s.L.p.m. respectively. Chamber pres-
sure was maintained at 5 lb-force per square inch gage (p.s.i.g.).
These plasma parameters were chosen mostly from a previous
optimization study wherein the supersonic plasma nozzle is fur-
ther explained [10]. The supersonic plasma nozzle employs a
velocity of 510 m s 1 when the powder is introduced in the lower
region of the plasma torch. These particles are introduced for
only 290 ls in the supersonic nozzle as opposed to 5 ms in a tra-
ditional plasma nozzle. This shortened time reduces heating of
the powder compared to a conventional plasma nozzle.
2.2. Phase analysis and surface characterization
A PANalytical X’Pert Pro MPD X-ray diffractometer was
employed to identify phase purity in the coating. The diffrac-
tometer used a Cu Ka radiation at 45 keV and 40 mA at room
temperature with a Ni-filter over the 2h range from 20° to
50°. A step size of 0.05° and a dwell of 10 s was utilized. An
Attenuated Total Reflection (ATR-FTIR) spectrophotometer (Nico-
let 6700, ThermoFisher, Madison, WI) was employed to obtain
the FTIR spectra in the 400 to 4000 cm 1 wave number range.
Both ZnSiAg-HA powder and coating were analyzed with FTIR.
A ZnSiAg-HA coated Ti64 sample was sectioned with a low
speed diamond blade then ground with progressive 250, 500,
1000, and 2000 grit silicon carbide paper and used for
microstructural characterization utilizing FESEM (DEI 200F, FEI,
OR). A Bruker Dektak XT profilometer was employed to deter-
mine the average surface roughness (Ra) of pure Ti64 samples,
HA coated Ti64, and ZnSiAg-HA coated samples. A force of
3 mg and a total elapsed time of 10 s was utilized. Contact angle
measurement (VCA Optima, AST Products, Inc., MA) using water
was also utilized to assess the hydrophilicity of the varying
implant composition surfaces.
2.3. Mechanical properties
To test bond strength of the doped coatings, a standard tensile
adhesion test was performed (ASTM C633). A total of 7 ZnSiAg-
HA samples were tested: 4 coated Ti64 samples and 3 coated CpTi
samples. Each sample was attached to posts using Armstrong A-12
epoxy resin as the adhesive and optimally cured at 93.3 °C (200 °F)
for 30 min then slowly cooled to room temperature. Each sample
was subjected to a tensile test with a constant cross-head speed
of 0.0013 cm s 1 until failure. Bond strength was calculated as:
failure load/sample area (A = 5.06 cm2).
 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335
2.4. Silver ion release
To ensure silver ion release was effective and not at toxic levels,
samples were immersed in both phosphate-buffer saline (PBS)
with a pH of 7.4 and acetate buffer solution (ABS) with a pH of 5.
PBS simulates the physiological environment within the body
and ABS simulates the acidic microenvironment post trauma or
surgery. The release study was done in an oven maintained at
36 °C ± 5 °C and subjected to constant shaking at 150 rpm. Aliquots
of 5 mL of each medium was collected at respective timepoints of
3, 6, 12 h, 1, 2, 3, 7, 10, 14 d and replaced with fresh solution. Con-
trols of both pure Ti64 and HA coated Ti64 were performed in trip-
licate simultaneously in both pH 7.4 and pH 5 alongside triplicates
of ZnSiAg-HA. The collected medium was analyzed for Ag+ concen-
tration using an Agilent 7700 ICP-MS. Gas flow conditions for
Argon consisted of 15 L/min plasma gas, 1 L/min auxiliary gas,
0.8 L/min nebulization gas, and 0.3 L/min make-up gas. The Ag+
standard used in this study was purchased from High-Purity Stan-
dards (Charleston, SC).
2.5. E. Coli and S. Aureus bacterial study
A disk diffusion sensitivity study was employed to compare the
antibacterial properties of the ZnSiAg-HA plasma samples com-
pared to pure HA discs. The bacterial cultures studied were Escher-
ichia coli (E. coli) and Staphylococcus aureus (S. aureus) (Carolina
Biological Supply company, Burlington, NC, USA) rehydrated from
MicroKwik CultureÒ pathogen vials. Cells were incubated for 48 h
at 37 °C prior to inoculation onto nutrient agar and brain heart
infusion agar plates, respectively. ZnSiAg-HA samples and HA discs
were placed onto agar plates after inoculation and left to incubate
for 24 h at 37 °C. Plates were imaged to compare the respective
zones of inhibition.
2.6. Rat distal femur model in vivo
Surgery procedures were followed from the Institutional Ani-
mal Care and Use Committee (IACUC) protocol from Washington
State University, Pullman, WA. Sprague-Dawley rats (Envigo,
Wilmington, MA, USA) underwent bilateral surgery in this study
and euthanized at 5 and 10 week timepoints. 18 skeletally mature
12-week old rats, weighing in between 280 and 320 g, were specif-
ically utilized to study a 3 mm defect size. Throughout the study,
rats were held in individual cages with humidity and temperature
controlled rooms with light and dark 12 h cycles. During surgery,
IsoFloÒ (Isoflurane, USP, Abbott Laboratories, North Chicago, IL,
USA) with oxygen (Oxygen USP, A-L Compressed Gases Inc., Spo-
kane, WA, USA) was used for anesthesia. 3 mm diameter defect size
was made with increasing drill bit sizes in the distal femurs of each
rat. Cylinder implants of pure Ti64, HA coated Ti64, and ZnSiAg-HA
coated Ti64 was randomized during surgery to mimic a statistically
randomized block design. After implantation, incisions were
closed using sutures and staples. Meloxicam was administered at
0.2 mg/kg following surgery for inflammation and pain control
for 3 days post-operatively. An overdose of CO2 was used for
euthanizing and preservation of removed femurs was done in
10% neutral buffered formalin solution.
2.7. Physical characterization in vivo
Pushout test was completed for each composition at 5 weeks
following femur harvesting to analyze shear modulus and pushout
strength between the growing tissue and the implant. A 50 lb
(22.7 kg) load cell was utilized in a servo-hydraulic controlled
Instron (Industrial Series DX Model, Instron, Norwood, MA, USA)
at a 0.33 mm/s cross-head speed. Load was applied in the Z direc-
327
tion and a clamping mechanism was used to hinder any motion
along the X and Y directions. Calculations of interfacial shear mod-
ulus derives from the linear region of the stress vs. strain plot gen-
erated from the Instron load-displacement values. Pushout
strength was calculated as the maximum force divided by the ini-
tial contact area.
2.8. Histology and histomorphometric analysis
Following 72 h of immersion in 10% formalin buffer solution,
femurs were dehydrated using consecutive ethanol concentrations
of 70%, 95%, and 100%, 1:1 ethanol-acetone solution, and conclud-
ing with 100% acetone. Thereafter, samples were embedded into
Spurs resin and a diamond saw blade was used to cut thin sec-
tioned slices of the femur with implant embedded. Sections were
rinsed in DI water, mounted onto glass slides using superglue,
and stained following a modified Masson Goldner’s trichrome
staining procedure. Implant interface was observed using a light
microscope (Olympus BH-2, Olympus America Inc., USA). Histo-
mophormetric analysis was performed with Gimp to quantify the
total osteoid formation as well as total bone formation. The region
around the implant assessed was 250 lm in radius using 3 differ-
ent images. All analysis was normalized over the radius of the
implant. Statistical significance was determined at a 95% confi-
dence level (p < 0.05) marked with * between compositions and
timepoints.
2.9. Statistical analysis
Characterizations reported as mean ± standard deviation were
performed in triplicate with a p value <0.05 considered significant
and indicated by *. These ANOVA with post-hoc Tukey statistical tests
include the mechanical testing, ion release, and histomorphometry.
3. Results
3.1. Phase analysis and surface characterization
Surface morphology of the ZnSiAg-HA showed surface rough-
ness and porosity (Fig. 1A). Coating thicknesses of about 75 lm
from cross-sections of both pure HA coating and ZnSiAg-HA coat-
ing were compared using FESEM (Fig. 1B). The ZnSiAg-HA coating
had comparable coating thickness to that of pure HA coating no
large gaps are seen between coating and substrate. The FESEM
images depict that the addition of the ternary dopant system did
not compromise an effective bonding zone between the coating
and the substrate as seen by the prominent interaction in the high
magnification images (Fig. 1B). Phase pure and crystalline HA was
identified in the XRD spectra (Fig. 2A) and FTIR spectra (Fig. 2B) of
the doped HA coating. Within the XRD spectra, HA (3 0 0) (ICDD
98-008-2289) peak contained Ag₂O (1 1 1) (ICDD 98-028-1041) in
the powder and absent after plasma coating. This Ag₂O phase
added to the HA (3 0 0) relative peak intensity in the powder. Once
absent, the peak intensity decreased to an average relative inten-
sity seen in typical HA peaks. Likewise, seen in the coating, the
addition of Ag (1 1 1) (ICDD 98-060-4630) phase increased the rel-
ative peak intensity in Ag₂O (0 0 2). Ag (0 0 2) peak can be seen in
the coating and absent in the powder. These peak changes from
powder to coating signify a shift from Ag₂O to Ag due to the plasma
treatment however phase purity and crystallinity of HA remain
unchanged. Within the FTIR spectra, between 970 and
1190 cm 1, antisymmetric (v3) P–O stretching modes of the phos-
phate group (PO4)3 were found. At 960 cm 1, symmetric (v1) P–O
stretching was found. Between 540 and 660 cm 1, antisymmetric
(v4) P–O bending modes were found. The plasma coating showed
328 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335

Fig. 1. (A) Surface morphology of the ZnSiAg-HA coating showing surface roughness and porosity. (B) The ZnSiAg-HA coating has comparable coating thickness to that of pure
HA coating and no large gaps are seen between coating and substrate. The FESEM images depict that the addition of the ternary dopant system does not compromise an
effective bonding zone between the coating and the substrate. Respective plasma coatings are shown between dashed lines.

the characteristic bands of (PO4)3 signifying unchanged phase
purity of HA. Crystalline HA was identified by the sharpness of
the peaks within these spectra. Neither XRD nor FTIR were able
to detect the low amounts of ZnO and SiO2 present. Two sizes of
samples were utilized in obtaining surface roughness measure-
ments using the profilometer: 12.5 mm and 25.4 mm diameter
(Tables 1 and 2). The results indicate that coating surface rough-
ness are similar when comparing HA and ZnSiAg-HA coated sam-
ples. Additionally, surface roughness can change based on the
plasma processing parameters. Increasing the powder gas feed rate
increased the surface roughness within the ZnSiAg-HA coating.
Contact angle with water was utilized to assess hydrophilicity.
Hydrophilicity increased with the smaller samples that have thin-
ner coatings compared to the larger samples with over 3 times the
coating thickness. This indicates that the thicker the coating, the
more hydrophobic the surface. Contact angles of less than 1 indi-
cate the droplet was unable to be measured due to the immediate
dispersion within the coating.
3.2. Mechanical properties
failure, but the delamination occurred from the top post at the
epoxy resin site rather than delamination between the coating
and substrate. The coating remained intact on the substrate itself
indicating strong bonding between the coating and substrate
otherwise no residue from the coating would be seen on the sub-
strate surface.
3.3. Silver ion release
Release of Ag+ ions was analyzed over the course of 14 days at
various timepoints in PBS and ABS (Fig. 3C). A total release of
1.71 ppm in PBS and 18.70 ppm in ABS of Ag+ ions was obtained.
The release kinetics in both pH 7.4 and pH 5 indicate a long-term
sustainable release as the kinetics became slower over the course
of time. No significant changes were seen in the coatings of HA
and ZnSiAg-HA in pH 7.4, however, in pH 5, pure HA coating was
completely removed by day 7 whereas some ZnSiAg-HA coating
still remained (Fig. 3D). The doped coating proved to be more resi-
lient in the acidic solution which indicates the doped coating was
more crystalline than the pure HA plasma coatings.

Plasma coating strength should be a minimum of 15 MPa (ISO
13779-4). HA control on Ti64 produced a tensile coating bond
strength of 26 ± 2 MPa (n = 3) whereas ZnSiAg-HA coated samples
produced strengths of 24 ± 2 MPa (n = 4) on Ti64 and 22 ± 1 MPa
(n = 3) on CpTi. The dopants did not affect bond strength below
standard and were within an acceptable range from an undoped
coating. The coating had a mainly cohesive failure as seen by the
coating residue on both posts (Fig. 3B). There was some adhesive
3.4. E. Coli and S. Aureus bacterial study
The ZnSiAg-HA plasma samples clearly show zones of inhibition
surrounding the samples compared to no zone of inhibition around
pure HA disc samples (Fig. 4). ZnSiAg-HA plasma samples pro-
duced zones of inhibition of 716 ± 1 mm2 and 330 ± 1 mm2 against
E. coli and S. aureus, respectively. Plasma samples showed a 54%
decrease in efficacy against S. aureus compared to E. coli.
 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335 329

Fig. 2. (A) XRD phase analysis of ZnSiAg-HA powder and coating showing an addition of AgO/Ag peak after plasma spraying with little to no change in all other characteristic
HA peaks. (B) FTIR spectra of ZnSiAg-HA powder and coating also indicating characteristic HA functional groups with no phase changes present.

Table 1
12.5 mm diameter sample characterizations with less than 100 mm coating thickness used for ion release study, bacterial testing, and in vivo implantation.

Composition Plasma Parameters Coating Thickness (lm) Surface Roughness (lm) Contact Angle (°)
Ti64 N/A N/A 2.1 82.70, 82.80
HA Coating Powder Gas – 13 70 ± 5 12.15 <1
Doped Coating 1 Powder Gas – 13 70 ± 5 11.11 <1
Doped Coating 2 Powder Gas – 13 70 ± 5 13.68 <1

Table 2
25.4 mm diameter sample characterizations used for tensile coating bond strength testing.

Composition Plasma Parameters Coating Thickness (lm) Surface Roughness (lm) Contact Angle (°)
Ti64 N/A N/A 3.5 82.70, 82.80
Doped Coating 1 Powder Gas – 10 165 (center) 8.31 109, 108.90
147 ± 7 (edges)
Doped Coating 2 Powder Gas – 13 147 (center) 11.23 59.90, 62.60
129 ± 6 (edges)
Doped Coating 3 Powder Gas – 13 183 (center) 11.92 44.50, 40
165 ± 3 (edges)

3.5. Pushout test
Pushout was performed to calculate the interfacial shear modu-
lus and pushout strength to assess the bonding between the
implant with bone tissue (Fig. 5B). Shear modulus values were
highest with ZnSiAg-HA coating and significantly higher than just
Ti64 implants. The HA coated implants showed less shear modulus
compared to doped coating however the difference was not signif-
icant with a 95% statistical confidence level. Pushout strength was
not statistically significant between ZnSiAg-HA and HA but both
330 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335

Fig. 3. (A) Schematic representation of disc samples used for coating bond strength testing with epoxy resin setup as well as silver ion release procedure. (B) ZnSiAg-HA
coating failure showing mostly cohesive failure. Adhesive failure seen from top post and epoxy resin, not from coating and substrate. ZnSiAg-HA coating bond strength on
Ti64 and commercially pure Ti showing increased coating bond strength on Ti64 samples with no significant difference compared to pure HA coating. (C) ZnSiAg-HA coating
release in pH 7.4 and pH 5 buffer solutions showing release within toxic levels of the average human body and significantly higher release in acidic buffer. (D) Coating images
at various release timepoints in ABS showing slower degradation in ZnSiAg-HA coated samples indicating higher crystallinity compared to HA coated samples.

were statistically higher than Ti64 alone. It should be noted that
both Ti64 implants were pushed out of the femurs whereas both
plasma coating compositions broke the femur before pushing the
implant completely out.
3.6. Histological evaluation
Optical microscopy images show new osteoid bone formation
(red/orange) and mineralization (green/blue) surrounding all
implant compositions (Fig. 6A). Osteoid formation can be seen in
all samples however a large gap and lowest osteoid formation
was seen with Ti64 implants. HA samples show highest osteoid
formation however ZnSiAg-HA coating showed highest total bone
formation (Fig. 6B and C). ZnSiAg-HA showed bone mineralization
even in the early timepoint indicating the doped coating enhanced
osteogenesis. Bone tissue can also be seen embedded within the
ZnSiAg-HA coating whereas the HA coating has bone tissue
attached just surface level indicating the doped coating has higher
 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335
Fig. 4. Bacterial disk diffusion test of ZnSiAg-HA plasma samples showing zones of
inhibition in both E. coli and S. aureus cultures compared to none seen with pure HA
discs.
osseointegration. By week 10, there was still a gap with Ti64
implants and osseous tissue as well as better osseointegration
within HA coatings. ZnSiAg-HA coatings still show the best bone
formation amongst all samples in both timepoints. No cytotoxic
effects can be seen by any implant.
SEM images of implant and osseous tissue interface confirm the
optical image observations (Fig. 7). Large gaps are consistently seen
between the Ti64 implants and osseous tissue. The gap did
decrease by week 10 compared to week 5. Better integration was
seen with HA coating and bone tissue however the best integration
was seen with the ZnSiAg-HA coating. In both timepoints, the
ZnSiAg-HA and bone tissue distinction becomes difficult to see
whereas the division between the pure HA coating and tissue
was seen clearly.
4. Discussion
The clinical significance of this study is to provide both
improved implant osseointegration and mitigate infections
Fig. 5. (A) Schematic representation of bilateral rat distal femur model. (B) Pushout test of implants from harvested femurs. ZnSiAg-HA coated implants showed statistically
higher shear modulus compared to bare Ti64 implants (*p < 0.05). ZnSiAg-HA showed higher average shear modulus compared to HA coated however the results were not
statistically significant at a 95% confidence level. Both plasma coated implants showed statistically higher pushout strength compared to bare Ti64.
331
through a ternary doped HA plasma sprayed coating. Plasma spray
deposition of HA on metal implants is a cost effective, widely used
system and can be further enhanced using dopants [11,14]. How-
ever, plasma spray deposition can induce phase decomposition of
HA due to the high temperature process however no extra phases
such as a/b tricalcium phosphate nor calcium oxide are seen
(Fig. 2B). Part of the phase purity can be attributed to the super-
sonic plasma nozzle used [10]. We have found in our previous
work that crystallinity can also be enhanced with dopants in
plasma spray. This is most likely due to the dopants having higher
melting temperatures than HA, therefore increasing the overall
melt temperature of the coating compared to control. This
increases the crystallinity by reducing the melt ratio (ratio of
non-melted to melted) of HA during the coating process, ultimately
leading to a lower dissolution rate of the coating as seen by the sil-
ver ion release study. A high melt ratio can result in decomposed
phases and amorphous phases [38,39]. Additionally, the XRD peaks
seen are sharp indicating the coating was crystalline. The sharp
bands within the FTIR spectrum also confirm crystallinity post
plasma deposition of the doped coating (Fig. 2B) [10,30,40,41].
Another indication of high phase purity and high crystallinity with
ZnSiAg-HA coating compared to HA coating can be seen during the
silver ion release study. The ZnSiAg-HA coating was clearly seen
degrading much slower compared to the HA coating (Fig. 3D). A
higher crystalline material will not degrade as quickly in an aque-
ous solution compared to an amorphous phase material [42,43]. By
day 7, HA coating was completely dissolved whereas ZnSiAg-HA
coating can still be seen on the sample surface. The silver ion
release done in PBS (pH 7.4) and ABS (pH 5) represent how the
coating degrades within a simulated body environment (Fig. 3C).
A toxic level within 24 h is exceeding 10 ppm [44] which neither
release exceeded this level. Toxicity within an entire body is also
dependent on bodyweight. A daily example of someone who is
198.4 lbs (90 kg) can handle 45 mL of 10 ppm silver in water
[45]. In this study, after 2 weeks in PBS and ABS, the release was
50 mL of 1.7 ppm and 18.7 ppm, respectively. Regarding in vitro
cell materials toxicity, our previous work has shown this amount
332 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335

Fig. 6. (A) Optical microscopy images showing osteoid formation using modified Masson-Goldner trichrome staining after 5 and 10 weeks for pure Ti64, HA coated Ti64, and
ZnSiAg-HA coated Ti64. Color coding: implant (black), osteoid tissue (red/orange), and mineralized bone tissue (green/blue). A gap is seen in both timepoints between the
pure Ti64 implant and tissue. Tissue integration within the coating can only be seen in week 10 for HA coated however tissue integration can be seen in both timepoints with
ZnSiAg-HA coated Ti64 indicating better osseointegration with ZnSiAg-HA compared to just HA. (B) Total osteoid formation in % around implant (250 lm radius) comparing
composition and timepoints. Higher osteoid formation is seen with a plasma coating compared to without in week 5. Higher osteoid formation is seen with HA however
ZnSiAg-HA has a higher total bone formation. (C) Total bone formation in % around implant (250 lm radius) comparing composition and timepoints. Statistically higher total
bone formation with a plasma coating compared to without in both timepoints. Statistically higher total bone formation with ZnSiAg-HA coating compared to HA coated in
week 5 indicating higher bone mineralization and osseointegration with the doped coating. (*p < 0.05). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)

of doped Ag2O within a plasma HA coating is not cytotoxic and
other dopant additions will simultaneously offset any cytotoxicity
concerns [30,39]. Release was significantly higher in ABS compared
to PBS which was expected due to the higher degradation proper-
ties of the acidic environment. The total amount of Zn and Si
embedded within the coating were much lower than toxic limits
for the body. Release of silver enabled zones of inhibition against
E. coli and S. aureus bacterial growth whereas pure HA samples
showed no antibacterial properties (Fig. 3). Previous work has
shown Ag-doped HA plasma coatings were effective against P.
aeruginosa and was assessed using a live/dead bacterial viability
kit with fluorescence staining and imaged using confocal micro-
scopy [30].
Coating bond strength is crucial for the clinical success of a
load bearing implant. In this study, the bond strengths on both
Ti64 and CpTi were higher than the ISO standard of 15 MPa or
higher (ISO 13779-2) (Fig. 3B). The addition of the dopants did
not hinder an effective coating bond strength on either titanium
substrates indicating this coating can be used in various applica-
tions such as Ti64 load bearing implants and CpTi dental applica-
tions. Coating bond strength largely depends on the substrate
material properties specifically their thermal conductivity and
coefficient of thermal expansion. CpTi (16–22 W/mK) has a
higher thermal conductivity compared to Ti64 (7.1–7.3 W/mK)
therefore during plasma processing, Ti64 maintains a hotter sur-
face relative to CpTi, providing better interaction between the
metal substrate with the pool melted HA which aids in increasing
the tensile coating bond strength comparatively. The coating and
substrate interface showed comparably no gaps nor cracks with
the doped coating compared to the HA coating indicating no
compromise to the bonding zone with the addition of the
dopants (Fig. 1B). The profilometer data shows higher surface
roughness with a plasma coating compared to as sandblasted
pure Ti64 sample. Appropriate surface roughness can facilitate
cell attachment compared to a smooth surface [46–48]. Higher
surface energy, borne from the hydrophilic nature of a surface,
has been shown to enhance cell attachment and even enhance
bioactivity of HA [49–51]. The surface of the thinner coating
was more hydrophilic compared to the thicker coating and there-
fore has higher surface energy. The difference in hydrophilicity
came from the differing processing parameters to produce thicker
coatings compared to thinner coatings such as longer plasma
exposure and more material addition which produced surface
or surface porosity differences.
All results for the rat distal femur model in vivo indicate
increased osseointegration with ZnSiAg-HA coatings compared to
 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335 333

Fig. 7. SEM images of implant interface for all compositions in both timepoints showing large gaps between implant and bone tissue with Ti64 as well as osseointegration
with both plasma coated samples with best integration seen with ZnSiAg-HA. Doped coating shows almost indistinguishable integration. Coatings are outlined with yellow
brackets. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

HA coatings and pure Ti64 implants. Pushout test was done as a
quantitative analysis of the bone tissue integration and results
showed highest integration with the doped coating (Fig. 5B). Sam-
ples without a coating pushed directly out of the femur whereas
coated samples broke the femur prior to full excitation indicating
higher integration with a plasma coating. Improved implant adhe-
sion to the surrounding bone tissue was due to the superior
osseointegration provided by the osteoconductive HA plasma coat-
ing and ternary dopant HA plasma coating compared to as sand-
blasted bare Ti64. These results reveal the effects of doping the
HA coating enhanced in vivo osseointegration through enabling
an early onset bone maturation process which was clearly seen
in the optical microscopy images (Fig. 6A). Optical microscopy
images may objectively show highest osteoid formation with HA
coatings (red/orange color) however highest total bone formation
was seen with the ZnSiAg-HA coating (red/orange and blue/green
color combined) (Fig. 6B and C). This finding indicated that the
doped coating enabled bone mineralization even in the early time-
point. By week 5, ZnSiAg-HA produced 32% bone mineralization of
68% total bone formation compared to only 11% bone mineraliza-
tion of 55% total bone formation with the HA coating. Both plasma
coatings showed significantly higher total bone formation in both
timepoints compared to pure Ti64 implants which was expected
due to Ti64 being bioinert. The ZnSiAg-HA coating also showed tis-
sue integration within the pores of the scaffold whereas bone tis-
sue was only attached to the HA coating surface. Integration
within the HA coated samples was not seen until week 10. The
SEM images confirmed the findings from the optical microscopy
images. The integration between the ZnSiAg-HA coating and bone
tissue was so high that the distinction was virtually indiscernible.
The HA coating has higher osseointegration in week 10 but in week
5 the integration to the coating was more surface level. Both time-
points show a gap with Ti64 implants with a slightly smaller gap in
week 10 (Fig. 7).
334 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335
5. Conclusion
A ternary dopant system utilizing ZnO, SiO2, and Ag2O was
deposited onto titanium samples via plasma spray deposition.
The doped coating had bond strengths of 24 ± 2 MPa on Ti6Al4V
and 22 ± 1 MPa on commercially pure Ti substrates as well as HA
phase preservation, and higher crystallinity than undoped coat-
ings. Silver was released below toxic levels and showed no signs
of cytotoxicity in vivo. Silver release provided antibacterial proper-
ties against E. coli and S. aureus strands of bacteria in vitro. A rat dis-
tal femur model was done to analyze the early stage
osseointegration of pure Ti64, HA coated, and ZnSiAg-HA coated
implants. By week 5, the doped coating showed 68% total bone for-
mation compared to only 55% with the HA coating. Of this total,
ZnSiAg-HA produced 32% bone mineralization compared to only
11% bone mineralization produced with the HA coating. The doped
coating showed the highest osseointegration of all compositions
and produced the highest shear modulus during pushout excita-
tion. This system can be implanted in orthopaedic and dental
replacement as well as revision surgeries to enhance early stage
implant osseointegration and reduce healing time.
Acknowledgements
Authors would like to acknowledge financial support from the
National Institutes of Health under the grant number R01
AR066361. A special thank you to the Peter Hooper GeoAnalytical
Lab, Franceschi Microscopy & Imaging Center, and John McCloy lab
at Washington State University for their help with data collection.
Thank you to Naboneeta Sarkar, Anish Shivaram, and Indranath
Mitra for their contributions to this work. This content is solely
the responsibility of the authors and does not necessarily represent
the official views of the National Institutes of Health.
References
[1] M.L. Wolford, K. Palso, A. Bercovitz, Hospitalization for total hip replacement
among inpatients aged 45 and over: United States, NCHS Data Brief. 2000
(2010) 1–8.
[2] H. Darwiche, W.K. Barsoum, A. Klika, V.E. Krebs, R. Molloy, Restrospective
analysis of infection rate after early reoperation in total hip arthroplasty, Clin.
Orthop. Relat. Res. 468 (2010) 2392–2396.
[3] G. Ryan, A. Pandit, D.P. Apatsidis, Fabrication methods of porous metals for use
in orthopaedic applications, Biomater 27 (2006) 2651–2670.
[4] R. Fraioli, P.M. Tsimbouri, L.E. Fisher, A.H. Nobbs, B. Su, S. Neubauer, F.
Rechenmacher, H. Kessler, M.P. Ginebra, M.J. Dalby, J.M. Manero, C. Mas-
Moruno, Towards the cell-instructive bactericidal substrate: exploring the
combination of nanotopographical features and integrin selective synthetic
ligands, Sci. Rep. 7 (2017) 16363.
[5] T.J. Webster, R.W. Siegel, R. Bizios, Osteoblast adhesion on nanophase
ceramics, Biomater 20 (1999) 1221–1227.
[6] S.B. Goodman, Z. Yao, M. Keeney, F. Yang, The future of biologic coatings for
orthopaedic implants, Biomater 34 (2013) 3174–3183.
[7] Y.C. Tsui, C. Doyle, T.W. Clyne, Plasma sprayed hydroxyapatite coatings on
titanium substrates Part 1: mechanical properties and residual stress levels,
Biomater 19 (1998) 2015–2029.
[8] S. Teixeira, M.A. Rodriguez, P. Pena, A.H. De Aza, S. De Aza, M.P. Ferraz, F.J.
Monteiro, Physical characterization of hydroxyapatite porous scaffolds for
tissue engineering, Mater. Sci. Eng. C. 29 (2009) 1510–1514.
[9] B. Leukers, H. Gülkan, S.H. Irsen, S. Milz, C. Tille, M. Schieker, H. Seitz,
Hydroxyapatite scaffolds for bone tissue engineering made by 3D printing, J.
Mater. Sci. Mater. Med. 16 (2005) 1121–1124.
[10] M. Roy, A. Bandyopadhyay, S. Bose, Induction plasma sprayed nano
hydroxyapatite coatings on titanium for orthopaedic and dental implants,
Surf. Coat. Technol. 205 (2011) 2785–2792.
[11] S. Bose, G. Fielding, S. Tarafder, A. Bandyopadhyay, Understanding of dopant-
induced osteogenesis and angiogenesis in calcium phosphate ceramics, Trends
Biotechnol. 31 (2013) 594–603.
[12] M. Bohner, Silicon-substituted calcium phosphates-a critical view, Biomater
30 (2009) 6403–6406.
[13] S. Tarafder, N.M. Davies, A. Bandyopadhyay, S. Bose, 3D printed tricalcium
phosphate bone tissue engineering scaffolds: effect of SrO and MgO doping on
in vivo osteogenesis in a rat distal femoral defect model, Biomater 1 (2013)
1250–1259.
[14] S. Bose, S. Tarafder, S.S. Banerjee, N.M. Davies, A. Bandyopadhyay,
Understanding in vivo response and mechanical property variation in MgO,
SrO and SiO2 doped b-TCP, Bone 48 (2011) 1282–1290.
[15] G. Fielding, S. Bose, SiO2 and ZnO dopants in three-dimensionally printed
tricalcium phosphate bone tissue engineering scaffolds enhance osteogenesis
and angiogenesis in vivo, Acta Biomater. 9 (2013) 9137–9148.
[16] M. Yamaguchi, Role of nutritional zinc in the prevention of osteoporosis, Mol.
Cell. Biochem. 338 (2009) 241–254.
[17] M. Yamaguchi, M.N. Weitzmann, Zinc stimulates osteoblastogenesis and
suppresses osteoclastogenesis by antagonizing NF-jB activation, Mol. Cell.
Biochem. 355 (2011) 179–186.
[18] I.S. Kwun, Y.E. Cho, R.A.R. Lomeda, H.I. Shin, J.Y. Choi, Y.H. Kang, J.H. Beattie,
Zinc deficiency suppresses matrix mineralization and retards osteogenesis
transiently with catch-up possibly through Runx 2 modulation, Bone 46
(2010) 732–741.
[19] M. Yamaguchi, M. Goto, S. Uchiyama, T. Nakagawa, Effect of zinc on gene
expression in osteoblastic MC3T3-E1 cells: enhancement of Runx2, OPG, and
regucalcin mRNA expressions, Mol. Cell. Biochem. 312 (2008) 157–166.
[20] A. Obata, S. Tokuda, T. Kasuga, Enhanced in vitro cell activity on silicon-doped
vaaterite/poly (lactic acid) composites, Acta Biomater. 5 (2009) 57–62.
[21] M. Mastrogiacomo, A. Papadimitropoulos, A. Cedola, F. Peyrin, P. Giannoni, S.G.
Pearce, M. Alini, C. Giannini, A. Guagliardi, R. Cancedda, Engineering of bone
using bone marrow stromal cells and a silicon-stabilized tricalcium phosphate
bioceramic: evidence for a coupling between bone formation and scaffold
resorption, Biomater 28 (2008) 1376–1384.
[22] A.M. Pietak, J.W. Reid, M.J. Stott, M. Sayer, Silicon substitution in the calcium
phosphate bioceramics, Biomater 28 (2007) 4023–4032.
[23] H. Li, J. Chang, Bioactive silicate materials stimulate angiogenesis in fibroblast
and endothelial cell co-culture system through paracrine effect, Acta Biomater.
9 (2013) 6981–6991.
[24] C. Wang, K. Lin, J. Chang, J. Sun, Osteogenesis and angiogenesis induced by
porous b-CaSiO3/PDLGA composite scaffold via activation of AMPK/ERK1/2 and
PI3K/Akt pathways, Biomater 34 (2013) 64–77.
[25] W. Zimmerli, C. Moser, Pathogenesis and treatment concepts of orthopaedic
biofilm infections, FEMS Immunol Med. Microbiol. 65 (2011) 158–168.
[26] W. Zimmerli, P. Sendi, Pathogenesis of implant-associated infection: the role of
the host, Semin. Immunopathol. 33 (2011) 295–306.
[27] L. Zhao, H. Wang, K. Huo, L. Cui, W. Zhang, H. Ni, Y. Zhang, Z. Wu, P.K. Chu,
Antibacterial nano-structured titania coating incorporated with silver
nanoparticles, Biomater 32 (2011) 5706–5716.
[28] A. Ewald, D. Hӧsel, S. Patel, L.M. Grover, J.E. Barralet, U. Gbureck, Silver-doped
calcium phosphate cements with antimicrobial activity, Acta Biomater. 7
(2011) 4064–4070.
[29] N. Matsumoto, K. Sato, K. Yoshida, K. Hashimoto, Y. Toda, Preparation and
characterization of b-tricalcium phosphate co-doped with monovalent and
divalent antibacterial metal ions, Acta Biomater. 5 (2009) 3157–3164.
[30] M. Roy, G. Fielding, H. Beyenal, A. Bandyopadhyay, S. Bose, Mechanical, In vitro
antimicrobial, and biological properties of plasma-sprayed silver-doped
hydroxyapatite coating, ACS Appl. Mater. Inter. 4 (2012) 1341–1349.
[31] N. Iqbal, M.R.A. Kadir, N.H. Mahmood, N. Salim, G.R. Froemming, H.R. Balaji, T.
Kamarul, Characterization, antibacterial and in vitro compatibility of zinc–
silver doped hydroxyapatite nanoparticles prepared through microwave
synthesis, Ceram. Inter. 40 (2014) 4507–4513.
[32] V. Stanić, S. Dimitrijević, J. Antić-Stanković, M. Mitrić, B. Jokić, I.B. Plećaš, S.
Raičević, Synthesis, characterization and antimicrobial activity of copper and
zinc-doped hydroxyapatite nanopowders, App. Surf. Sci. 256 (2010) 6083–
6089.
[33] Y. Huang, Z. Xu, X. Zhang, X. Chang, X. Zhang, Y. Li, T. Ye, R. Han, S. Han, Y. Gao,
X. Du, Nanotube-formed Ti substrates coated with silicate/silver co-doped
hydroxyapatite as prospective materials for bone implants, J. Alloys Comp. 697
(2017) 182–199.
[34] D.B. Jaroch, D.C. Clupper, Modulation of zinc release from bioactive sol–gel
derived SiO2-CaO-ZnO glasses and ceramics, J. Biomed. Mater. Res. Part A
Banner 82 (2007) 575–588.
[35] G.A. Fielding, A. Bandyopadhyay, S. Bose, Effects of SiO2 and ZnO doping on
mechanical and biological properties of 3D printed TCP scaffolds, Dent. Mater.
28 (2012) 113.
[36] A. Bandyopadhyay, S. Bernard, W. Xue, S. Bose, Calcium phosphate-based
resorbable ceramics: influence of MgO, ZnO, and SiO2 dopants, J. Amer. Ceram.
Soc. 89 (2006) 2675–2688.
[37] W. Xue, K. Dahlquist, A. Banerjee, A. Bandyopadhyay, S. Bose, Synthesis and
characterization of tricalcium phosphate with Zn and Mg based dopants, J.
Mater. Sci. Mater. Med. 19 (2008) 2669–2677.
[38] D. Ke, S.F. Robertson, W.S. Dernell, A. Bandyopadhyay, S. Bose, Effects of MgO
and SiO2 on plasma-sprayed hydroxyapatite coating: an in vivo study in rat
distal femoral defects, ACS Appl. Mater. Int. 9 (2017) 25731–25737.
[39] G.A. Fielding, M. Roy, A. Bandyopadhyay, S. Bose, Antibacterial and biological
characteristics of silver containing and strontium doped plasma sprayed
hydroxyapatite coatings, Acta Biomater. 8 (2012) 3144–3152.
[40] N. Pleshko, A. Boskey, R. Mendelsohn, Novel infrared spectroscopic method for
the determination of crystallinity of hydroxyapatite minerals, Biophys. J. 60
(1991) 786–793.
[41] M. Roy, A. Bandyopadhyay, S. Bose, Induction plasma sprayed Sr and Mg doped
nano hydroxyapatite coatings on Ti for bone implant, J. Biomed. Mater. Res. B.
Appl. Biomater. 99 (2011) 258–265.
 A.A. Vu et al. / Acta Biomaterialia 92 (2019) 325–335
[42] F. Fazan, P.M. Marquis, Dissolution behavior of plasma-sprayed
hydroxyapatite coatings, J. Mater. Sci. Mater. Med. 11 (2000) 787–792.
[43] G. Van den Mooter, The use of amorphous solid dispersions: a formulation
strategy to overcome poor solubility and dissolution rate, Drug. Discov. Today
Technol. 9 (2012) e79–e85.
[44] C. Greulich, D. Braun, A. Peetsch, J. Diendorf, B. Siebers, M. Epple, M. Kӧller, The
toxic effect of silver ions and silver nanoparticles towards bacteria and human
cells occurs in the same concentration range, RSC Adv. 2 (2012) 6981–6987.
[45] United States Environmental Protection Agency. Silver; CASRN 7440-22-4.
EPA, (1987).
[46] L.L. Hench, Bioceramics: from concept to clinic, J. Amer. Ceram. Soc. 74 (1991)
1487–1510.
[47] W. Xue, B.V. Krishna, A. Bandyopadhyay, S. Bose, Processing and
biocompatibility evaluation of laser processed porous titanium, Acta
Biomater. 3 (2007) 1007–1018.
[48] Y. Deng, X. Liu, A. Xu, L. Wang, Z. Luo, Y. Zheng, F. Deng, J. Wei, Z. Tang, S. Wei,
Effect of surface roughness on osteogenesis in vitro and osseointegration
335
in vivo of carbon fiber-reinforced polyetheretherketone–nanohydroxyapatite
composite, Inter. J. Nanomed. 10 (2015) 1425.
[49] M.A. Surmeneva, C. Leinhans, G. Vacun, P.J. Kluger, V. Schӧnhaar, M. Müller, S.
B. Hein, A. Wittmar, M. Ulbricht, O. Prymak, C. Oehr, R.A. Surmenev, Nano-
hydroxyapatite-coated metal-ceramic composite of iron-tricalcium
phosphate: improving the surface wettability, adhesion and proliferation of
mesenchymal stem cells in vitro, Coll. Surf. B. Biointer. 135 (2015) 386–393.
[50] N. Eliaz, S. Shmueli, I. Shur, D. Benayahu, D. Aronov, G. Rosenman, The effect of
surface treatment on the surface texture and contact angle of
electrochemically deposited hydroxyapatite coating and on its interaction
with bone-forming cells, Acta Biomater. 5 (2009) 3178–3191.
[51] R.A. Gittens, R. Olivares-Navarrete, A. Cheng, D.M. Anderson, T. McLachlan, I.
Stephan, J. Geis-Gerstorfer, K.H. Sandhage, A.G. Fedorov, F. Rupp, B.D. Boyan,
The roles of titanium surface micro/nanotopography and wettability on the
differential response of human osteoblast lineage cells, Acta Biomater. 9
(2013) 6268–6277.