Biomaterials 34 (2013) 5036e5047

Contents lists available at SciVerse ScienceDirect

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

Hydroxyapatite-anchored dendrimer for in situ remineralization of

human tooth enamel
Duo Wu a, Jiaojiao Yang a, Jiyao Li b, c, Liang Chen b, c, Bei Tang b, c, Xingyu Chen a, Wei Wu a,
Jianshu Li a, *
a
College of Polymer Science and Engineering, Sichuan University, Yi Huan Road, South Section One, No. 24, Chengdu 610065, China
b
State Key Laboratory of Oral Diseases, Sichuan University, Chengdu 610065, China
c
West China School of Stomatology, Sichuan University, Chengdu 610065, China

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

Article history: In situ remineralization of hydroxyapatite (HA) on human tooth enamel surface induced by organic
Received 7 January 2013 matrices is of great interest in the fields of material science and stomatology. In order to mimic the
Accepted 15 March 2013 organic matrices induced biomineralization process in developing enamel and enhance the binding
Available online 8 April 2013
strength at the remineralization interface, carboxyl-terminated poly(amido amine) (PAMAMeCOOH)d
alendronate (ALN) conjugate (ALNePAMAMeCOOH) was synthesized and characterized. PAMAMeCOOH
Keywords:
has a highly ordered architecture and is capable of promoting the HA crystallization process. ALN is
Dendrimer
conjugated on PAMAMeCOOH due to its specific adsorption on HA (the main component of tooth
Specific adsorption
Hydroxyapatite
enamel), resulting in increased binding strength which is tight enough to resist phosphate buffered
In situ biomineralization saline (PBS) rinsing as compared with that of PAMAMeCOOH alone. While incubated in artificial saliva,
Dental restorative material ALNePAMAMeCOOH could induce in situ remineralization of HA on acid-etched enamel, and the re-
generated HA has the nanorod-like crystal structure similar to that of human tooth enamel. The hardness
of acid-etched enamel samples treated by ALNePAMAMeCOOH can recover up to 95.5% of the original
value with strong adhesion force. In vivo experiment also demonstrates that ALNePAMAMeCOOH
is effective in repairing acid-etched enamel in the oral cavity. Overall, these results suggest that
ALNePAMAMeCOOH is highly promising as a restorative biomaterial for in situ remineralization of
human tooth enamel.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction such as extreme acidic pH, high temperature, high pressure, or in

In human tooth enamel, nanorod-like hydroxyapatite (HA)
crystals are arranged into highly organized prism to form the main
unit. In the oral cavity, the tooth enamel can be damaged by the
local cariogenic bacteria in plaque (caries), non-bacterially derived
erosive challenges (such as acidic beverages) or mechanical force.
Traditional dental restorative materials involve metal, compound
resin and ceramics. However, since their structures, components
and properties are different from the natural material, they do not
fit well with natural tissues in the lesions interface. Many attempts,
including inorganic paste approaches [1e3], hydrothermal method
[4e6], self-assembly or agents-mediated methods [7e10], have
been taken to repair acid-etched enamel or to obtain enamel-like
structure. However, most of the above methods need conditions
* Corresponding author. Tel.: þ86 28 85466755; fax: þ86 28 85405402.
E-mail addresses: jianshu_li@scu.edu.cn, fredca2005@163.com (J. Li).
the presence of a concentrated solution of surfactant. Therefore, the
in situ regeneration or remineralization of HA under physiological
conditions as an alternative restorative material is very desirable in
stomatology.
The biomineralization process in nature, such as the construc-
tion of tooth enamel, is controlled by organic matrices (including
various proteins such as amelogenin for tooth enamel). By now,
quite a few natural or synthetic materials have been developed to
mimic the organic matrices to regenerate dental tissues, i.e.,
restoring the injured enamel by inducing HA remineralization on
dental surface. For example, amelogenin and its supramolecular
assembly have been directly used to obtain mineral layers con-
taining organized needle-like fluoridated HA crystals on the surface
of etched enamel [11e13]. Kirkham et al. utilized a self-assembled
anionic peptide to form scaffolds on caries-like lesions in dental
enamel under simulated intra-oral conditions, leading to a signifi-
cant net HA remineralization [14]. Peptide amphiphiles are also
used as artificial matrices for biological synthesis of tooth enamel

0142-9612/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biomaterials.2013.03.053
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
[15]. In addition, a glycerine-enriched gelatin gel containing
phosphate and fluoride ions [16], and a combination of glumatic
acid and nano apatite particles [17] have both been reported to be
capable of inducing enamel-like structure of reconstructed mineral
on dental surface. However, most of the above strategies still have
limitations, such as the difficulty in preparation of proteins/peptides,
the overuse of fluoride ions and the complicated multi-steps in clinic
applications. Thus, it is needed to develop a simple strategy to mimic
the functions of organic matrices to induce biomineralization on the
surface of tooth enamel.
It is widely known that dendrimer is a class of mono-dispersed
polymeric nanomaterials with plenty of branches radiating from
one central core and highly ordered architecture. It has been
referred as ‘artificial protein’ due to its biomimetic properties and
well-defined/easily tailored structure, i.e., its functional group,
generation and spatial structure are controllable [18]. Several kinds
of dendrimers or their derivatives have been applied in biominer-
alization field [19e22]. Specifically, poly(amido amine)-type
(PAMAM) dendrimer has been widely investigated in the crystal-
lization process of HA in recent years. The size and shape of HA
could be regulated by PAMAM dendrimer with different surface
groups, generations and concentrations [23,24]. Recently, an
amphiphilic PAMAM dendron has been synthesized with aspartic
acids on the periphery and an aliphatic chain at the focal point,
which exhibited a self-assembly behavior similar to that of ame-
logenin in the oriented growth of HA in vitro, i.e., initially aggre-
gating to nanospheres and further translating to linear chains [25].
However, because the free dendrimer or its derivatives in aqueous
solution could not contribute to the crystal growth on a specific
substrate, it is necessary to increase the binding capability between
PAMAM and the substrate of the human hard tissues (mainly HA) to
achieve in situ regeneration or remineralization.
It has been proved that alendronate (ALN) could easily adsorb on
the HA crystals. Palazzo et al. investigated the adsorption and
desorption kinetics of ALN towards synthetic HA nanocrystalline
materials. They found that the interaction between ALN and HA
surface takes place by ligand exchange in which the two phosphate
groups of the ALN molecule replace two surface phosphate groups
of HA [26]. This adsorption affinity is further proved to be a two-site
model, i.e., with two different binding sites, by using isothermal
titration calorimetry method [27]. Therefore, ALN has been conju-
gated with different drug carriers to provide them with the HA
binding specificity as bone- or tooth-targeted drug delivery sys-
tems [28e33]. In this work, we synthesize an HA-anchored ALNe
PAMAMeCOOH dendrimer and then evaluate its in situ biominer-
alization behavior on the surface of acid-etched human tooth
enamel (both in vitro and in vivo). It may mimic the role of natural
organic matrices in biomineralization. Anionic PAMAM dendrimer
with carboxyl groups on the periphery is linked with ALN to obtain
ALNePAMAMeCOOH conjugate. The ALNePAMAMeCOOH conju-
gate should adsorb on the surface of tooth enamel due to the HA
binding specificity of ALN, and then induce in situ HA regeneration
in artificial saliva (Scheme 1). Several instrumental methods have
been used to characterize the HA binding capability of ALNe
PAMAMeCOOH, the morphology and the mechanical properties
of newly generated HA on the surface of acid-etched enamel.
Ultimately, we hope to investigate the ALNePAMAMeCOOH as a
restorative material for in situ remineralization of tooth enamel.
2. Materials and methods
2.1. Materials
Dicyclohexyl carbodiimide (DCC), ethyl (dimethylaminopropyl) carbodiimide
(EDC), N-hydroxysuccinimide (NHS), alendronate (ALN), fluorescein isothiocyanate
(FITC) were purchased from Aldrich. Hydroxyapatite powder was purchased from
5037
National Engineering Research Center for Biomaterials, Sichuan University (medical
grade, spherical HA powder of 10 mm in diameter). Human tooth samples were
extracted following standard procedures for extraction at the Hospital of Stoma-
tology in Sichuan University, and handled with permission of Institutional Review
Board. All other reagents and solvents if not specified were purchased from Tianjin
Bodi Chemical Holding Company and were all of analytical grade except for chro-
matographic grade methanol (MeOH).
2.2. Synthesis of carboxyl-terminated PAMAM dendrimer (PAMAMeCOOH)
The divergent synthesis of PAMAM dendrimer includes two-step iterative
sequence to produce amine terminated structures. It involves alkylation with methyl
acrylate (MA) and followed by amidation with excessive 1, 2-ethylenediamine (EDA).
The alkylation step produces ester terminated dendrimer that is referred as ‘half-
generations’. The second step involves amidation of the ester terminated in-
termediates with a large excess of EDA to produce amine terminated dendrimer,
which is named as ‘full-generations’. PAMAM dendrimer was synthesized step by
step following the classical method reported by Tomalia [34,35]. G3.5 PAMAM
dendrimer was synthesized for further modification: 1H NMR (400 Hz, CDCl3)
d (ppm) ¼ 2.41e2.49 (eCH2eNCH2e), 2.28e2.38 (eCH2CONHe), 3.27e3.28
(eNHCH2), 2.75 (eCH2COOCH3), 3.66 (eCOOCH3), 7.27 (eNHe).
G3.5 PAMAM (1.681 g, 0.280 mmol) was dissolved in MeOH (10 mL) in a reflux
condenser, NaOH (0.430 g, 10.752 mmol) was added to the dendrimer solution and
then reacted at 60  C for 8 h. MeOH was evaporated and then the cold MeOH was
used to dissolve dendrimer and simultaneously precipitate unreacted NaOH. After
filtrating, the filtrate was evaporated again to obtain the alkaline hydrolyzed den-
drimer G3.5eCOONa, and then dissolved in water. Then pH value of the solution was
adjusted to 3 by adding 0.1 M HCl, followed by dialysis against water (molecular
weight cut-off: 3500) to remove NaCl and then lyophilized. Yield: 85%. 1H NMR
(400 Hz, D2O) d (ppm) ¼ 2.81e2.99 (eCH2eNCH2e), 2.62e2.68 (eCH2CONHe),
3.42e3.60 (eNHCH2e), 3.26e3.29 (eCH2COOH).
2.3. Synthesis of ALNePAMAMeCOOH dendrimer
The carboxyl groups of PAMAMeCOOH were activated by NHS (0.097 g,
0.840 mmol) in dimethyl sulfoxide (DMSO) for 0.5 h. After that, DCC (0.173 g,
0.840 mmol) was added under stirring and the reaction was continued for overnight
at 25  C. The reaction system was filtrated to remove the by-product 1,
3-dicyclohexylurea (DCU) and the filtrate containing NHSePAMAMeCOOH was
collected. The filtrate was then added into ALN (0.379 g, 1.400 mmol) aqueous so-
lution in three batches, 4 h apart. After the last batch the reaction was continued for
another 36 h at room temperature. The solution was then filtrated again and dia-
lyzed against water (molecular weight cut-off: 3500), followed by lyophilization to
obtain the final product ALNePAMAMeCOOH. Yield: 52%. 1H NMR (400 Hz, D2O)
d (ppm) ¼ 2.81e2.88 (eCH2eNCH2e), 2.69 (eCH2CONHe), 3.44e3.56 (eNHCH2e),
3.27e3.29 (eCH2COOH), 1.31e1.36 (eCH2CH2e).
2.4. Cytotoxicity assay
MTT assays were performed to measure the cytotoxicity of the synthesized
ALNePAMAMeCOOH. HepG2 cells were cultured in Dulbecco’s modified eagle
medium (DMEM), 10% heat-inactivated fetal bovine serum (FBS), 100 units/mL of
penicillin and 100 mg/mL of streptomycin at 37  C, with 5% CO2 and 95% relative
humidity. The cells were seeded in a 96-well microtiter plate (Nunc Co., Wiesba-
den, Germany) at a density of 104 cells/well and incubated in DMEM/well (100 mL)
for 24 h. The culture media was replaced with fresh culture media containing serial
dilutions of ALNePAMAMeCOOH, and the cells were incubated for 24 h. Then,
10 mL of sterile-filtered MTT stock solution in 5 mg/mL PBS was added to give a
final MTT concentration of 0.5 mg/mL. The unreacted dye was removed by aspi-
ration after 5 h. The formed formazan crystals were solubilized in DMSO (100
mL/well). The absorbance was measured using a microplate reader (Spectra Plus,
Tecan, Zurich, Switzerland) at a wavelength of 570 nm. The cell viability (%) was
calculated from 100  ([A]test  [A]PEI)/[A]control, where [A]test, [A]PEI and [A]control
represent the absorbance values of the wells with ALNePAMAMeCOOH, with PEI
(positive control) and without ALNePAMAMeCOOH (negative control), respec-
tively. The absorbance was the average value measured from six wells in parallel
for each sample.
2.5. Preparation of tooth enamel samples
Sound human third molars were obtained from and approved by the Hospital of
Stomatology in Sichuan University. After extraction, the organic contaminants on
the teeth were removed with a scalpel blade. Then, the teeth were further cleaned
with 3% sodium hypochlorite (to remove adhered bacteria) and rinsed with PBS. The
teeth were stored at 4  C in water containing 0.05% thymol before use. A diamond-
coated band saw was used to separate root from crowns and to cut sections longi-
tudinally, resulting in an approximately 5  5 mm2 square plate. They were ground
flat and polished with water-cooled carborundum discs. All the surfaces of sections
were protected with acrylic resin except the polished enamel surface. The sample
5038 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
Scheme 1. Schematic demonstration of the specific adsorption of ALNePAMAMeCOOH on the surface of tooth enamel and the subsequent in situ remineralization of HA.
surfaces were partly painted with two layers of acid-resistant nail varnish, which left
only an exposed 3  4 mm2 window. The tooth slices were stored at 4  C in PBS prior
to use.
2.6. Preparation of enamel erosion
37% phosphoric acid was used to prepare enamel erosion. All samples were
sonicated for 30 min before acid etching. Each enamel sample was exposed to
phosphoric acid (10 mL) for 45 s. PBS was used to wash sample surface, and another
5 min treatment of sonication was taken. The samples were stored at 4  C in PBS
prior to use.
2.7. HA-anchored property of ALNePAMAMeCOOH
In order to compare the HA-anchored property of ALNePAMAMeCOOH with
that of PAMAMeCOOH, excessive hydrophobic dye powder (dimethyl yellow)
was added into ALNePAMAMeCOOH and PAMAMeCOOH aqueous solution
(3 mg/mL, 1 mL) to obtain a supersaturated solution. After completely stirring,
the solution was centrifuged to remove undissolved dimethyl yellow and the
supernate was filtered through 0.22 mm filters. Then, HA (100 mg) was added to
the solution, and magnetically stirred at 37  C for 12 h. Finally, the HA powder
adsorbed with dye-loaded dendrimer was obtained by centrifugation
(10,000 rpm, 3 min), and then washed with water for three times before taking
photos and measured by FTIR.
In addition, to evaluate the binding capability of ALNePAMAMeCOOH and
PAMAMeCOOH on HA, HA (100 mg) was added to the aqueous solution of two
dendrimers with different concentrations (from 0.5 to 5.5 mg/mL, 1 mL). The
mixture was stirred at 37  C for 12 h and then centrifuged at 10,000 rpm for 3 min.
The supernate was taken out for UV (MAPADA 1800PC, China) absorbance analysis at
a wavelength of 282 nm. The amount of the dendrimer adsorbed on HA was
calculated by the decrease of dendrimer in the solution.
In the desorption test, 100 mg HA adsorbed with dendrimer (both ALNe
PAMAMeCOOH and PAMAMeCOOH, 4.5 mg/mL) was resuspended in 2.5 mL arti-
ficial saliva in an Eppendorf tube sealed with 8-kDa cut-off dialysis membrane. Then
the Eppendorf tube was suspended in 50 mL artificial saliva (37  C) with 100 rpm
stirring. At different time intervals, 5 mL dialysis solution was withdrawn for
measurement. Meanwhile, 50 mL fresh artificial saliva was replaced into the
container. The amount of released dendrimer was estimated by using UV detection
at 282 nm. 1, 2, 4, 8, 20, 30 and 45 h are chosen as the time intervals for this
experiment.
2.8. Characterization of the binding capability of ALNePAMAMeCOOH on human
tooth enamel
ALNePAMAMeCOOH and PAMAMeCOOH were diluted with deionized water
(4 mg/mL), separately. Each dendrimer solution (100 mL) was evenly pipeted on the
acid-etched tooth enamel surface in the window area. After drying at room tem-
perature, each sample was rinsed with 1 M PBS buffer for three times, deionized
water and dried again. ATR-IR (NICOLET iS10, Thermo Scientific, USA) character-
ization was taken and recorded before and after dendrimer coating, and also after
PBS washing.
ALNePAMAMeCOOH was also labeled with FITC by a procedure reported pre-
viously [36]. Briefly, equimolar of FITC dissolved in ethanol (with the addition of 1 M
sodium hydroxide and equimolar of EDC) was added to the aqueous solution of
ALNePAMAMeCOOH. The mixture was stirred for 3 days, and then filtered and
dialyzed against deionized water for 3 days. Finally the conjugate was lyophilized.
FITC labeled ALNePAMAMeCOOH solution (4 mg/mL, 100 mL) was pipeted onto the
surface of normal and acid-etched tooth enamel, separately. After 5 s, PBS buffer was
used to wash the sample surface for three times. After that, samples were dried and
then observed by confocal laser scanning microscopy (CLSM) (ZEISS LSM700,
Germany).
2.9. Biomineralization in artificial saliva
ALNePAMAMeCOOH or PAMAMeCOOH aqueous solution (4 mg/mL, 100 mL)
was evenly pipeted onto the window area of tooth enamel surface. After dried at
room temperature, samples were rinsed with deionized water and dried again
before incubation in artificial saliva.
The artificial saliva was used as a biomineralizing solution according to the
method reported by ten Cate and Duijsters [37]. It included CaCl2 (1.5 mM), KH2PO4
(0.9 mM), KCl (130 mM), NaN3 (1 mM) and HEPES (20 mM). The pH of solution was
then adjusted to 7.0 with 1 M KOH. As compared with the previous report [13], the
artificial saliva has a low supersaturation degree. The preparation of the solution is
under well-controlled conditions to avoid coagulation and precipitation that could
influence the in vitro test results. It could keep the stability at 37  C for at least 2
weeks once prepared in our lab. Each sample was vertically soaked in artificial saliva
(5 mL) in polystyrene tubes, ensuring that all the samples were under the surface of
artificial saliva solution. Artificial saliva was replaced by fresh solution every 24 h.
After the period of 1 day, 1 week, 2 weeks and 4 weeks, different sample groups
were removed out of artificial saliva and rinsed with deionized water for three times.
5 min sonication was taken before further investigation.
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
2.10. Structural and compositional analysis of in situ regenerated crystal
SEM (Hitachi S-450, 20 kV, Japan) was used to examine the enamel surface for
structural analysis after biomineralization. The samples were sputtered with Au
before observation. X-ray diffraction analysis (Dmax 1400, 40 kV, 110 mA, Japan) was
performed on the enamel surface to examine the type of the biomineralized crystal
and the orientation degree of crystallinity.
2.11. Mechanical property of remineralized tooth enamel
Knoop microhardness analyses of samples surface were performed using a
microhardness tester (Duramin-1/-2; Struers, Copenhafen, Denmark) with Knoop
diamond indenter under a 50 g load for 10 s. The mean values of all five measure-
ments for each sample taken at different time (original surface microhardness
(SMH), after acid erosion SMH1 and after biomineralization SMH2 of 2 weeks and
4 weeks) were recorded for comparison. The degree of hardness recovery was
calculated as %SMHR ¼ 100  (SMH2  SMH1)/(SMH1  SMH). The scratch tests
were performed under the standard conditions (diamond stylus R ¼ 0.2 mm;
scratching speed 3 mm/min; loading rate 50 N/min).
2.12. Animal experiments
The in vivo experiments were carried out with male SpragueeDawley (SD) rats
(8 weeks old, 210e240 g) and permitted by the Animal Research Committee of the
University. Each group contained three rats and each rat was used for two disk
samples, thus there were six samples in each group. The human enamel samples
(prepared following the same way as that for in vitro measurements) were cut into
1.5  2 mm2 plates and protected with acrylic resin to form disks (diameter: 4 mm;
thick: 1 mm). The disks were prepared with two holes on the acrylic resin section to
facilitate the implantation in oral cavity (see photos in discussion part). The enamel
area was treated with 37% phosphoric acid to make it demineralized (following the
same method as that for in vitro measurements). After that, the enamel area of each
disk was treated with 25 mL of deionized water (control group), PAMAMeCOOH or
ALNePAMAMeCOOH dendrimer (4 mg/mL), respectively, and then air-dried. Then,
the disks were fixed on the interior side of the rats’ cheek (one disk on each side of
cheek). Finally, the disks were taken out after 28 days. The disks were thoroughly
rinsed by deionized water, air-dried, and then investigated by SEM (Hitachi S-450,
20 kV, Japan).
3. Results
3.1. Characterization of synthesized compound
In this work, G3.5 PAMAM was synthesized following Tomalia’s
classical divergent synthesis strategy. After that, the alkaline
hydrolysized dendrimer (G3.5 PAMAMeCOONa) was activated by
NHS/DCC and then conjugated with alendronate in water to obtain
the ALNePAMAMeCOOH. The FTIR data is shown in Figure S1: the
peaks of 3295.16 and 3081.14 cm1 are due to amide vibration, and
the bands at 2939.92, 1650.07, 1432.11 and 1250.20 cm1 are
contributed by the NeH bending vibration, amide carbonyl, eCH2e
bending vibration and CeO stretch vibration, respectively. The
band at 1105 cm1 attributes to the PeO adsorption peak of PO3
4 in
ALN, which confirms the successful synthesis of ALN-PAMAMe
COOH. 1H NMR (Figure S2) was used to quantify the number of ALN
molecules per dendrimer by comparing the area of the dendrimer
protons between 3.27 and 3.29 ppm with the methylene protons of
ALN between 1.31 and 1.36 ppm, and it could be calculated that
each PAMAM molecular has 1.6 ALN molecules on average.
The cytotoxicity of ALNePAMAMeCOOH at various concentra-
tions (from 0.122 to 250 mg/mL) was also evaluated by MTT assay
using HepG2 cell lines. As can be seen in Fig. 1, ALNePAMAMe
COOH displays superior cell viability in the range of 81e106%,
which indicates that ALNePAMAMeCOOH has very low cytotox-
icity within a wide range of concentrations.
3.2. Binding capability of ALNePAMAMeCOOH to HA
Dendrimer has plenty of branches radiating from one central
core and the whole architecture is highly ordered, thus it has the
capability of including guest molecules due to the cavity in the
5039
Fig. 1. Cytotoxicity assay in HepG2 cells at various concentration of ALNePAMAMe
COOH by MTT assay.
spatial structure. Quite a few hydrophobic reagents could be
included in PAMAM dendrimer and their solubilities in aqueous
could be significantly improved [38]. In order to directly visualize
the HA-anchored behavior of ALNePAMAMeCOOH, a hydrophobic
dye (dimethyl yellow) was loaded into the cavity of ALNePAMAMe
COOH and PAMAMeCOOH and then dissolved in deionized water
to form a saturated solution. Then, 100 mg HA was added to the
solution with fully stirring, followed by centrifugation and sepa-
ration, and then thoroughly washed by deionized water for three
times. After incubated with HA, the supernate still remained yellow
color and the HA powder is only lightly stained by the dye in
PAMAMeCOOH group, while the HA powder was obviously stained
and the color of the supernate was fade in ALNePAMAMeCOOH
group (Figure S3a). It indicates that the PAMAMeCOOH loaded with
dimethyl yellow has a limited adsorption capability on HA, result-
ing in much more dimethyl yellow remaining in the supernate. On
the other hand, the ALNePAMAMeCOOH loaded with dimethyl
yellow can easily anchor on the surface of HA powder. The
adsorption ratio is much higher than that of PAMAMeCOOH and
the binding strength is strong enough to resist water rinsing. Since
the hydrophobic dimethyl yellow itself has been investigated not to
adsorb on HA under the same incubation conditions and it is loaded
in the cavity of dendrimers, the interaction between dimethyl
yellow and HA is neglectable in this study. As shown in Figure S4,
the adsorbed ALNePAMAMeCOOH on the HA surface could resist
the desorption by water.
In order to quantify the adsorption behaviors of ALNePAMAMe
COOH and PAMAMeCOOH on HA, the dendrimers were dissolved
in deionized water at different concentrations, fully stirred with HA
powder and then centrifuged. The amount of the dendrimer
adsorbed on HA was calculated by the decrease of dendrimer in
the solution. Figure S3b presents the adsorption isotherm of
two dendrimers on HA. It indicates that the saturated amount of
ALNePAMAMeCOOH on HA (100 mg) is 3.66 mg. And the
adsorption of ALNePAMAMeCOOH on 100 mg HA is always more
than that of PAMAMeCOOH within this concentration range.
Since the environment of biomineralization contains a lot of
flowing liquids, the binding strength to the substrate surface is very
important for a biomineralization-inducing material to fulfill its
function of in situ regeneration. To characterize this property of our
system, we applied both PAMAMeCOOH and ALNePAMAMeCOOH
onto the surfaces of tooth enamel samples, followed by sufficient
5040 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
washing with 1 M PBS solution. PBS buffer rather than deionized
water was chosen because PBS has a stronger washing off power
according to previous investigation [39]. Fig. 2 exhibits the ATR-IR
data of above samples before dendrimer coating, after dendrimer
coating and after PBS washing. As can be seen, the characteristic
peak of enamel (1040.24 cm1 for PO3 4 n1, n3 functional groups)
can be clearly observed. Then, after dendrimer coating, the char-
acteristic peaks of dendrimer (3500e3050 cm1 for amide vibra-
tion, 1650 cm1 for amide carbonyl) derived from the large sum of
amide bonds in its structure are detected in both groups. However,
after PBS rinsing, the former group sample does not exhibit obvious
characteristic peaks of dendrimer, indicating that there are few
PAMAMeCOOH on the surface of tooth enamel. In contrast, the
latter group sample still remained a large amount of ALNe
PAMAMeCOOH dendrimer on tooth enamel according to the
obvious amide characteristic peaks. Meanwhile, the peak at
Fig. 2. ATR-IR spectra of acid-etched tooth enamel, after dendrimer coating and after
PBS washing for PAMAMeCOOH treated group (a) and ALNePAMAMeCOOH treated
group (b).
1040.24 cm1 also verifies the exposure of the enamel surface. It is
noted that the phosphate peak is greatly enhanced for the sample
after PBS washing (Fig. 2b), which is due to the phosphate groups
both in ALNePAMAMeCOOH and residual PBS on the surface. The
residual PBS peak does not weaken the conclusion about the strong
specific adsorption of ALNePAMAMeCOOH since its characteristic
peak at 1650.07 cm1 is still obvious after PBS washing.
This result demonstrates that ALNePAMAMeCOOH can bind
tightly to enamel surface due to the HA-anchored capability of its
ALN functional unit. The binding capability between ALNe
PAMAMeCOOH and enamel surface is so strong that PBS buffer can
not wash it off in a large quantity, while the PAMAMeCOOH den-
drimer without ALN unit can be washed off from the enamel sur-
face by PBS buffer; Then, FITC labeled ALNePAMAMeCOOH
solution was dropped on the surface of normal tooth enamel and
acid-etched tooth enamel, separately. Samples were fully rinsed
with PBS buffer and dried at room temperature before CLSM
observation. Fig. 3 shows the fluorescence dispersion and sample
surface morphology. Meanwhile, the roughness of the enamel
surface is indicated in the line graphs at the bottom of Fig. 3 (a2),
(b2) and (c2). No fluorescence on the control sample can be
observed (Fig. 3, a1), and the enamel surface is not smooth. When
FITC labeled ALNePAMAMeCOOH solution was dropped on the
normal tooth enamel, the yellow-green fluorescence can be seen
and is sporadically dispersed on the surface. Thus, the FITC labeled
ALNePAMAMeCOOH is considered to be able to adsorb on the
normal tooth enamel. We can find that the surface roughness of the
sample obviously decreases after the coating of dendrimer (Fig. 3,
a2 and b2). When the tooth enamel was etched by phosphoric acid,
its surface roughness greatly increases, resulting in a significant
increase of total surface area (Fig. 3, c2). The enlarged enamel
surface provides more adsorption sites, thus there are much more
FITC labeled ALNePAMAMeCOOH adsorbed on the acid-etched
tooth enamel surface (Fig. 3, c1). With the CLSM investigation,
the fluorescence is so strong that the enamel prism is also clearly
visible.
3.3. Remineralization of tooth enamel in vitro
After being soaked in artificial saliva for different time periods,
the morphology of the acid-etched tooth enamel samples treated
with different dendrimers was observed by SEM. After the acid
etching, tooth enamel crystals at the surface become discontinuous
and broken, and the enamel prism can be clearly seen (Figure S5,
untreated). It is noted that the appearance of the chosen acid-
etched tooth enamel model is different from that of early enamel
caries, which appears as sub-surface lesions and the demineral-
ization is under the intact enamel surface. As shown in Fig. 4, new
crystals were formed on the tooth enamel surface of all three
groups after being soaked in the artificial saliva. Renewed crystals
after one day of biomineralization are different in size, shape and
growth direction, and the distribution on tooth enamel is not ho-
mogenous (Fig. 4, a1, b1, c1). After the biomineralization for one
day, PAMAMeCOOH group (Fig. 4, b1) and ALNePAMAMeCOOH
group (Fig. 4, c1) induce quite a few small crystals (see the enlarged
inserts), which are different from the original HA crystals of acid-
etched enamel (Figure S5) and should be the newly generated
crystals. Meanwhile, the profile of acid-etched tooth enamel is still
clearly visible although the original acid-etched surface is almost
covered by growing crystals.
Fig. 4 also exhibits the biomineralization behaviors of above
samples in artificial saliva for a continuous period of 4 weeks. As
can be seen, the biomineralization phenomenon of the control
group (Fig. 4, a1, a2, a3 and a4) is quite different from the other two
groups. It is noted that Fig. 4 a2 shows that the prism pattern is
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
Fig. 3. CLSM and the surface morphology images of normal tooth enamel surface (a1, a2), FITC-ALNePAMAMeCOOH treated normal tooth enamel surface (b1, b2) and FITC-ALNe
PAMAMeCOOH treated acid-etched tooth enamel surface (c1, c2).
completely covered by a layer of small crystals, similar to that
formed in Fig. 4 b2 of PAMAMeCOOH. As compared with the acid-
etched enamel (Figure. S5, untreated), the control group presents a
layer of rod-like crystals after one week, which should grow from
the original exposed HA crystals after acid-etching. This is because
that, even though there is no mediator on the enamel surface, the
exposed crystals still have certain capability to attract calcium and
phosphate to form small crystals [39]. However, this reminerali-
zation would be out of control if it lacks the regulation of organic
matrix. As shown in Figure S6, the green circles indicate that newly
formed crystals randomly aggregate on the surface of original
crystal, exhibiting the subsequent trend of irregular growth. At the
end of 4 weeks, the flake-like crystals are crowded on the enamel
surface and the original erosion surface is totally not visible.
However, this flak-like crystal is not consistent with the natural
human tooth enamel. The modulating effect of PAMAMeCOOH on
the regeneration of mineral looks has a very interesting time-
dependant phenomena as shown in Fig. 4 (b1, b2, b3 and b4). As
can be seen, the rod-like crystals cover almost all the tooth enamel
surface in 1 week, and the size and shape of these renewed min-
erals are relatively uniform (Fig. 4, b2). However, the growth of
these calcium phosphate minerals is out of control at the second
week as the renewed calcium phosphate layer is covered by flake-
like new crystals and distributed on the entire region. In the fourth
week, the flake-like new crystals grow bigger, like some irregular
blasting cluster. The middle part of the cluster is perpendicular to
the surface and its periphery bursts out. Finally, we can see that the
biomineralization behavior of the ALNePAMAMeCOOH group in
5041
artificial saliva is under good control in the whole 4-week period
(Fig. 4, c1, c2, c3 and c4). The nanorod-like renewed crystals have
almost uniform size and shape, which is very similar to that of the
natural tooth enamel. The renewed calcium phosphate coating is
also closely adhered to the original enamel surface and the original
acid-etched surface can not be observed.
In order to observe the cross section of the regenerated min-
erals, the tooth enamel samples which have been incubated in
artificial saliva for 4 weeks were cut and the long axes of the enamel
prisms were exposed. The cross section of acid-etched sample
(untreated) is presented in Figure S7 for comparison. It is shown
that the side views of the three groups are quite different (Fig. 5).
The control group (Fig. 5, a1 and a2) is loosen and accumulated
with many big sheet of calcium phosphate crystal on the coarse
surface. The morphology and size of new crystals are not similar to
that of the original enamel. Meanwhile, the biomineralization layer
of PAMAMeCOOH treated group (Fig. 5, b1 and b2) exhibits a
certain extent of orientation. However, the morphology of the new
crystals in PAMAMeCOOH treated group is heterogenous. Also, the
arrangement of new crystals becomes more loose from the bio-
mineralization interface to the top area. There is a clear line of
interface between the new crystals and original tooth enamel,
indicating that the new formed layer induced by PAMAMeCOOH
might not bind tightly. As for the ALNePAMAMeCOOH treated
group (Fig. 5, c1 and c2), there is a dense layer of biomineralized
crystals on the acid-etched enamel surface. The new generated
crystals are parallel bundles of nanorod-like HA with almost uni-
form size and shape, and mostly oriented perpendicular to the
5042 D. Wu et al. / Biomaterials 34 (2013) 5036e5047

Fig. 4. SEM images of the surface of acid-etched tooth enamel without treatment (a group), treated with PAMAMeCOOH (b group) and ALNePAMAMeCOOH (c group), after being
soaked in artificial saliva for 1 day (a1, b1 and c1), 1 week (a2, b2 and c2), 2 weeks (a3, b3 and c3) and 4 weeks (a4, b4 and c4). The inserts are enlarged details.

original tooth enamel surface. The morphology of the new regen- After 4 weeks, the regenerated crystals on the samples were
erated minerals (thicker than 10 mm) is very close to that of normal investigated by XRD, in terms of the type and the orientation de-
tissue and should be desirable for in situ biomineralization of hu- gree of crystallinity. As shown in Fig. 6, the diffraction peak (002) at
man tooth enamel. 2q ¼ 25.8, (211) at 2q ¼ 31.8 and (300) at 2q ¼ 32.8 are characteristic
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047 5043

Fig. 5. SEM micrographs of the cross section of acid-etched tooth enamel without treatment (a group), treated with PAMAMeCOOH (b group) and ALNePAMAMeCOOH (c group),
after being soaked in artificial saliva for 4 weeks. The a2, b2 and c2 are the enlarged photos of the red square area in a1, b1 and c1, respectively. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this article.)

XRD peaks of HA, which are presented in the original acid-etched
tooth enamel. The XRD patterns of the control, PAMAMeCOOH
and ALNePAMAMeCOOH group show that, with the help of den-
drimers, HA crystrals are formed after incubation in artificial saliva.
To describe the orientation degree of new crystals, the ratio of
diffraction intensity of z axis (002) to another direction is calculated
[12]. The (002)e(211) intensity ratio for acid-etched tooth enamel
is 1.659, while the values are 0.729, 1.391 and 1.159 for the control
group, PAMAMeCOOH treated group and ALNePAMAMeCOOH
treated group, respectively. This result indicates that the acid
erosion process leads to a high orientation degree of HA in acid-
etched tooth enamel. The orientation of the new formed crystals
coating is also maintained with the presence of dendrimers in the
biomineralization process [12].
The recovery degree of hardness is characterized by Knoop
microhardness test. As shown in Fig. 7, the surface microhardness
of all samples declines about 34e39% in general after acid erosion.
The control group with the biomineralized flake-like crystals can
only restore the microhardness to a very limited degree even at the
end of 4 weeks, which indicates that the renewed loose crystals
make little contribution to the hardness restoration. In contrast, the
surface microhardness recoveries (%SMHR) of the other two groups
treated by dendrimers are much more obvious after incubation in
artificial saliva. For the PAMAMeCOOH treated group, the new
generated crystals have a specific structure of dense layer at the

Fig. 7. The surface microhardness (SMH) of control group, PAMAMeCOOH treated

Fig. 6. XRD patterns of the original acid-etched tooth enamel, and the samples group, and ALNePAMAMeCOOH treated groups after being soaked in artificial saliva
without treatment (control), treated with PAMAMeCOOH and ALNePAMAMeCOOH for 2 and 4 weeks, comparing with the microhardness value of the original and acid-
after being soaked in artificial saliva for 4 weeks. etched tooth enamel.
5044 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
bottom and loose layer at the top (Fig. 5, b1 and b2), resulting in an
obvious microhardness improvement in the first 2 weeks but a
slight increase in the next 2 weeks. Finally, the dense arrangement
and the thickness (>10 mm as shown in Fig. 5, c1) of the new
generated HA layer in ALNePAMAMeCOOH treated group result in
significant %SMHR values: 91.1% after 2 weeks and 95.5% after 4
week, approaching to that of the normal tooth enamel; and the
value of the first 2 weeks is even higher than that of the PAMAMe
COOH treated group after 4 weeks. The results suggest that both
PAMAMeCOOH and ALNePAMAMeCOOH dendrimer can induce
biomineralization of HA on acid-etched tooth enamel and help to
recover the surface microhardness, but ALNePAMAMeCOOH is
much better due to its HA-anchored adsorption and well-
controlled regulation capability in the biomineralization process.
Meanwhile, by the scratch test, it is clear to see that the adhesion
force between newly formed HA layer and original enamel surface
has this sequence: ALNePAMAMeCOOH treated group > PAMAMe
COOH treated group >> control group (Fig. 8). The multiple signals
observed in the ALNePAMAMeCOOH group probably indicate that
there should be more than one interface from the bottom to top. It
should be the last signal that corresponds to the interface of newly
formed HA layer and original enamel surface.
3.4. In vivo experiments
To evaluate the performance of above materials in vivo, the tooth
enamel samples were cut into plates and protected with acrylic
resin to form disks (Fig. 9, a1). The disks were sutured to the cheek
of rat, by which the oral cavity of rat could provide an environment
with both ion and enzyme similar to that of human beings (Fig. 9,
a2). After 28 days, the samples were taken out and observed
by SEM. As can be seen, all the three groups can induce the
Fig. 8. Acoustic emission signal variation during the scratch test. The acid-etched tooth enamel (a), acid-etched tooth enamel treated with PAMAMeCOOH (b) and ALNePAMAMe
COOH (c) were tested after being soaked in artificial saliva for 4 weeks. This experiment is to reflect the adhesion force between newly formed HA layers and original enamel
surface.
regeneration of HA on the enamel surface (Fig. 9, b1, c1, d1). The
cracks on the surface should be caused by the strong and contin-
uous mechanical activity exerted on them during eating or grinding
of rats. As for the variation in the observed cracks, it should be
attributed to the different resistance of the newly generated crys-
tals to the mechanical activity, since all the samples were dried
under the same conditions before SEM observation both in vitro and
in vivo and we did not find this phenomenon in vitro. As compared
with the other two groups, the ALNePAMAMeCOOH group
exhibits a better surface morphology (Fig. 9, d1) and a thicker
regenerated HA layer (Fig. 9, d2). The better performance of ALNe
PAMAMeCOOH group should owe to its specific adsorption on the
HA surface as revealed/discussed in the section of in vitro experi-
ments. Meanwhile, the Ca/P ratios of regenerated minerals in the
three groups are 1.812, 1.506 and 1.670 (by EDS, Fig. 9, b3, c3, d3),
respectively. It reveals that the crystal induced by ALNePAMAMe
COOH is most similar to the HA (Ca/P 1.67) in natural tooth enamel.
Thus the in vivo experiments indicate that the ALNePAMAMe
COOH can effectively induce the in situ remineralization of HA on
the surface of acid-etched enamel in oral cavity.
4. Discussion
Figure S3b shows that the amount of PAMAMeCOOH adsorbed
on 100 mg HA is approximately linearly correlated with the increase
of concentration when it is less than 4.5 mg/mL. This stable increase
should be attributed to the physical adsorption of PAMAMeCOOH to
HA. As for ALNePAMAMeCOOH, the adsorption degree nearly ap-
proaches 100% within the range of 1.5e3.5 mg/mL, which means that
most of ALNePAMAMeCOOH in solution has adsorbed on HA
powder. This should be attributed to the specific chemical binding of
ALN on HA [26,27]. It is also concluded that the adsorption of ALN to
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047 5045

Fig. 9. In vivo experiments in rat oral cavity: acid-etched tooth enamel disks (a1); disks sutured to the cheek of rat (a2). SEM micrographs of the acid-etched tooth enamel disks
without treatment (b group), treated with PAMAMeCOOH (c group) and ALNePAMAMeCOOH (d group), after being implanted in rats’ oral cavity for 4 weeks. The b2, c2 and d2 are
the cross section of b1, c1 and d1, respectively. The b3, c3 and d3 are the EDS of the pink square area in b1, c1 and d1, respectively. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)

HA is relevant to the HA amount and the superficial area of tooth
enamel (HA). The saturated amount of ALNePAMAMeCOOH on
100 mg HA (3.66 mg) looks somewhat lower than the saturated
adsorption amount of other ALN conjugates on HA surface, i.e., the
quantity of ALNePAMAMeCOOH absorbing on 100 mg HA is less
than other ALN conjugates [28e31]. It is because that the steric
structure of dendrimer is pretty large and rigid, which might limit
the adsorbed quantity of ALNePAMAMeCOOH on each HA crystal
surface. According to this result, we choose a moderate concentra-
tion of 4 mg/mL of ALNePAMAMeCOOH for further experiments
5046 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
and we can see it is adequate for in situ biomineralization of HA in
later section. In addition, as shown in Fig. 3 (b1, b2), the adsorption
amount of FITC labeled ALNePAMAMeCOOH on the enamel surface
approaches to an equilibrium value, which should be due to the
adsorption capability on limited surface area. The acid-etched
enamel surface provides more adsorption sites for FITC labeled
ALNePAMAMeCOOH (Fig. 3, c1). This result is meaningful for clinic
application, since it indicates that the ALNePAMAMeCOOH can
easily adsorb on the hard tissue surface (especially the one after acid-
etching), and form the in situ HA regeneration sites for further
biomineralization.
In the in vitro remineralization, the control group without
dendrimer treatment (Fig. 4, a1) has some granules of new gener-
ated crystals on the acid-etched tooth enamel surface, which makes
it coarse and irregular. In the beginning, the acid-etched tooth
enamel has a certain degree of capable of inducing biomineraliza-
tion of HA without the help of mediators. It should be attributed to
the charge adsorption effect between the free mineral ions (calcium
and phosphate ions in the aqueous solution) and positive/negative
charge domains on the tooth enamel surface [39]. As for the
PAMAMeCOOH group, it is believed that it can act as a template in
the crystal growth process, whereas the physical binding strength
is not strong enough. Some PAMAMeCOOH may be released during
the everyday renewing process of the artificial saliva, especially in
the early stage. The free PAMAMeCOOH in artificial saliva can
restrain the biomineralization process of mineral crystals, similar to
the inhibition of CaCO3 formation by free dendrimer in aqueous
solution reported by Naka et al. [40]. As shown in Figure S8, the
desorption of PAMAMeCOOH from the dendrimer-adsorbed HA is
much faster than that of ALNePAMAMeCOOH, which may be an
additional evidence for this speculation. It is also noted that the
desorption test is an accelerated experiment, thus the time scale is
not consistent with that of the in vitro experiment. Therefore, the
crystal morphology on PAMAMeCOOH treated tooth enamel can be
the result of the cooperative action of both the free dendrimer in
solution and gradually decreased immobilized-dendrimer on the
enamel surface.
In contrast with the control group and PAMAMeCOOH group,
ALNePAMAMeCOOH regulates the growth of crystals, inducing
nanorod-like HA with high uniformity. Both the binding site and
the ordered architecture of dendrimer should contribute to the
well-controlled growth of new HA crystals. ALNePAMAMeCOOH
adsorbs tightly on the enamel surface by its ALN unit, the carboxyl
groups on the periphery of the PAMAMeCOOH unit can attract
Ca2þ, which may locate on the periphery of dendrimer. The specific
binding provides nucleation sites and mineralization template for
HA. Meanwhile, ALNePAMAMeCOOH mainly adsorbs on the side
surface of exposed nanorod-like HA of enamel, which may induce
crystal growth along the long axis of the rods. This specific longi-
tudinal growth leads to the uniformity of the growing HA. The tight
adsorption to enamel makes it a perfect template for the crystal-
lization of Ca2þ and PO3 4 , and the arrangement of the new HA
crystals could be in a well-organized order in the early stage. Then
the subsequent gathering of calcium and phosphate ions is well
controlled to generate new crystal layers.
In the whole period of 4-week remineralization, the flake-like
crystals are generally crowded on the enamel surface in the con-
trol group (Fig. 4, a1, a2, a3 and a4). According to the hardness test in
the following experiment, recrystallization of calcium and phosphate
in this way almost makes no contribution to the hardness recovery of
acid-etched tooth enamel. Thus, the biomineralization without
mediator is not ideal for restoring acid-etched enamel. The special
evolution process of the crystal morphology on PAMAMeCOOH
treated tooth enamel might be the result of the defects formed in the
early stage of mineralization. The defects are caused by the desorbed
free PAMAMeCOOH in artificial saliva as discussed above. Based on
above observations, we can conclude that the biomineralization
process regulated by PAMAMeCOOH in this way is also not ideal for
restoring acid-etched enamel. The loss of controllability at the latter
stage is not beneficial for reconstructing mechanical property of
further formed biomimetic structure. The good biomineralization
behavior of the ALNePAMAMeCOOH group in artificial saliva in the
whole 4-week period (Fig. 4, c1, c2, c3 and c4) should be attributed to
the strong binding strength between ALNePAMAMeCOOH and the
tooth enamel. The PAMAMeCOOH moiety in the molecule acts as a
biomineralization template since its carboxyl groups can induce
calcium and phosphate to assemble into new crystals. Also, the
binding strength is very strong, which is contributed by the ALN
moiety, thus there is very few free dendrimer in artificial saliva and
the biomineralization function is dominantly controlled by the
immobilized ALNePAMAMeCOOH on the surface of tooth enamel.
The significant improvements of microhardness and adhesion force
in ALNePAMAMeCOOH group also provide solid support to the
mechanism of HA-anchored biomineralization process as discussed.
5. Conclusions
In order to mimic the biomineralization process induced by
organic matrix in developing tooth enamel and enhance the
binding strength at the remineralization interface, ALN conjugated
PAMAMeCOOH dendrimer with low cytotoxicity was successfully
synthesized. The ALN moiety acts as the HA-anchored functional
unit in the molecular structure. The binding strength between
ALNePAMAMeCOOH and tooth enamel is so strong that the den-
drimer could not be washed off by PBS buffer. In artificial saliva,
ALNePAMAMeCOOH could induce in situ remineralization of HA
on acid-etched enamel. The new generated crystals are parallel
bundles of nanorod-like HA with almost uniform size and shape,
and mostly oriented perpendicular to the original surface. The
structure of remineralized crystals is similar to that of natural tooth
enamel, and its microhardness could recover up to 95.5% of the
original value with strong adhesion force. Animal experiment also
indicates that ALNePAMAMeCOOH could effectively induce the
remineralization of HA in the oral cavity of rat. Therefore, ALNe
PAMAMeCOOH could be a potential restorative nanomaterial for in
situ remineralization of tooth enamel.
Acknowledgments
Financial support from the National Natural Science Foundation
of China (51073102, 81170958), Fok Ying Tung Education Founda-
tion (122034), Program for New Century Excellent Talents in Uni-
versity (NCET-10-0592), Program for Changjiang Scholars and
Innovative Research Team in University (IRT1163), Foundations of
Sichuan Province (2012JQ0009, 2011SZ0130), Fundamental
Research Funds for the Central Universities (2010SCU22001,
2011SCU04A04) and Natural Science Foundation of Jiangsu Prov-
ince (BK2010248, BK2011340) are gratefully acknowledged.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.biomaterials.2013.03.053.
References
[1] Yamagishi K, Onuma K, Suzuki T, Okada F, Tagami J, Otsuki M, et al. A synthetic
enamel for rapid tooth repair. Nature 2005;433:819.
[2] Wang Z, Jiang T, Sauro S, Wang Y, Thompson I, Watson TF. Dentine remi-
neralization induced by two bioactive glasses developed for air abrasion
purposes. J Dent 2011;39:746e56.
 D. Wu et al. / Biomaterials 34 (2013) 5036e5047
[3] Dong Z, Chang J, Deng Y, Joiner A. In vitro remineralization of acid-etched
human enamel with Ca3SiO5. Appl Surf Sci 2010;256:2388e91.
[4] Chen H, Tang Z, Liu J, Sun K, Chang SR, Peters MC, et al. Acellular synthesis of a
human enamel-like microstructure. Adv Mater 2006;18:1846e51.
[5] Ashok M, Kalkura SN, Sundaram NM, Arivuoli D. Growth and characterization of
hydroxyapatite crystals by hydrothermal method. J Mater Med 2007;18:895e8.
[6] Zhan J, Tseng YH, Chan JC, Mou CY. Biomimetic formation of hydroxyapatite
nanorods by a single-crystal-to-single-crystal transformation. Adv Funct
Mater 2005;15:2005e10.
[7] Chen H, Clarkson BH, Sun K, Mansfield JF. Self-assembly of synthetic hy-
droxyapatite nanorods into an enamel prism-like structure. J Colloid Interface
Sci 2005;288:97e103.
[8] Fowler CE, Li M, Mann S, Margolis HC. Influence of surfactant assembly on the
formation of calcium phosphate materials-A model for dental enamel for-
mation. J Mater Chem 2005;15:3317e25.
[9] Zhang J, Jiang D, Zhang J, Lin Q, Huang Z. Synthesis of dental enamel-like
hydroxyapaptite through solution mediated solid-state conversion. Lang-
muir 2010;26:2989e94.
[10] Xie R, Feng Z, Li S, Xu B. EDTA-assisted self-assembly of fluoride-substituted
hydroxyapatite coating on enamel substrate. Cryst Growth Des 2011;11:
5206e14.
[11] Fan Y, Sun Z, Wang R, Abbott C, Moradian-Oldak J. Enamel inspired nano-
composite fabrication through amelogenin supramolecular assembly. Bio-
materials 2007;28:3034e42.
[12] Fan Y, Sun Z, Moradian-Oldak J. Controlled remineralization of enamel in the
presence of amelogenin and fluoride. Biomaterials 2009;30:478e83.
[13] Fan Y, Nelson JR, Alvarez JR, Hagan J, Berrier A, Xu X. Amelogenin-assisted
ex vivo remineralization of human enamel: effects of supersaturation degree
and fluoride concentration. Acta Biomaterialia 2011;7:2293e302.
[14] Kirkham J, Firth A, Vernals D, Boden N, Robinson C, Shore RC, et al. Self-
assembling peptide scaffolds promote enamel remineralization. J Dent Res
2007;86:426e30.
[15] Huang Z, Newcomb CJ, Bringas Jr PB, Stupp SI, Snead ML. Biological synthesis of
tooth enamel instructed by an artificial matrix. Biomaterials 2010;31:9202e11.
[16] Busch S. Regeneration of human tooth enamel. Angew Chem Int Ed 2004;43:
1428e31.
[17] Li L, Mao C, Wang J, Xu X, Pan H, Deng Y, et al. Bio-inspired enamel repair via
glu-directed assembly of apatite nanoparticles: an approach to biomaterials
with optimal characteristics. Adv Mater 2011;23:4695e701.
[18] Svenson S, Tomalia DA. Dendrimers in biomedical applications-reflections on
the field. Adv Drug Deliv Rev 2005;57:2106e29.
[19] Tsiourvas D, Tsetsekou A, Kammenou MI, Boukos N. Controlling the formation of
hydroxyapatite nanorods with dendrimers. J Am Ceram Soc 2011;7:2023e9.
[20] Wang L, Meng Z, Yu Y, Meng Q, Chen D. Synthesis of hybrid linear-dendritic
block copolymers with carboxylic functional groups for the biomimetic
mineralization of calcium carbonate. Polymer 2008;49:1199e210.
[21] Donners JJ, Nolte RJ, Sommerdijk NA. Dendrimer-based hydroxyapatite
composites with remarkable materials properties. Adv Mater 2003;15:313e6.
[22] Khopade AJ, Khopade S, Jain NK. Development of hemoglobin aquasomes
from spherical hydroxyapatite cores precipitated in the presence of half-
generation poly(amidoamine) dendrimer. Int J Pharm 2002;241:145e54.
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.
5047
[23] Yan SJ, Zhou ZH, Zhang F, Yang SP, Yang LZ, Yu XB. Effect of anionic PAMAM
with amido groups starburst dendrimers on the crystallization of
Ca10(PO4)6(OH)2 by hydrothermal method. Mater Chem Phys 2006;99:164e9.
[24] Xie L, Wang L, Jia X, Kuang G, Yang S, Feng H. Effects of glutamic acid shelled
PAMAM dendrimers on the crystallization of calcium phosphate in diffusion
systems. Polym Bull 2011;66:119e32.
[25] Yang S, He H, Wang L, Jia X, Feng H. Oriented crystallization of hydroxyapatite
by the biomimetic amelogenin nanospheres from self-assemblies of amphi-
philic dendrons. Chem Commun 2011;47:10100e2.
[26] Palazzo B, Iafisco M, Laforgia M, Margiotta N, Natile G, Bianchi CL, et al. Bio-
mimetic hydroxyapatiteedrug nanocrystals as potential bone substitutes with
antitumor drug delivery properties. Adv Funct Mater 2007;17:2180e8.
[27] Mukherjee S, Huang C, Guerra F, Wang K, Oldfield E. Thermodynamics of
bisphosphonates binding to human bone: a two-site model. J Am Chem Soc
2009;131:8374e5.
[28] Choi SW, Kim JH. Design of surface-modified poly(D, L-lactide-co-glycolide)
nanoparticles for targeted drug delivery to bone. J Control Release 2007;122:
24e30.
[29] Pignatello R, Cenni E, Micieli D, Fotia C, Salerno M, Granchi D, et al. A novel
biomaterial for osteotropic drug nanocarriers: synthesis and biocompatibility
evaluation of a PLGA-ALE conjugate. Nanomedicine 2009;4:161e75.
[30] Wang D, Miller S, Sima M, Kopeckova P, Kopecek J. Synthesis and evaluation of
water-soluble polymeric bone-targeted drug delivery systems. Bioconjug
Chem 2003;14:853e9.
[31] Liu XM, Wiswall AT, Rutledge JE, Akhter MP, Cullen DM, Reinhardt RA, et al.
Osteotropic b-cyclodextrin for local bone regeneration. Biomaterials 2008;29:
1686e92.
[32] Chen F, Liu XM, Rice KC, Li X, Yu F, Reinhardt RA, et al. Tooth-binding
micelles for dental caries prevention. Antimicrob Agents Chemother
2009;53:4898e902.
[33] Chen F, Rice KC, Liu XM, Reinhardt RA, Bayles KW, Wang D. Triclosan-loaded
tooth-binding micelles for prevention and treatment of dental biofilm. Pharm
Res 2010;27:2356e64.
[34] Tomalia DA, Baker H, Dewald JR, Hall M, Kallos G, Martin S, et al. A new class
of polymers: starburst-dendritic macromolecules. Polym J 1985;17:117e32.
[35] Tomalia DA, Naylor AM, Goddard III WA. Starburst dendrimers: molecular-
level control of size, shape, surface chemistry, topology, and flexibility from
atoms to macroscopic matter. Angew Chem Int Ed 1990;29:138e75.
[36] Perumal OP, Inapagolla R, Kannan S, Kannan RM. The effect of surface func-
tionality on cellular trafficking of dendrimers. Biomaterials 2008;29:3469e76.
[37] Ten Cate JM, Duijsters PP. Alternating demineralization and demineralization
and remineralization of artificial enamel lesions. Caries Res 1982;16:201e10.
[38] Milhem OM, Myles C, McKeown NB, Attwood D, DEmanuele A. Polyamido-
amine starburstÒ dendrimers as solubility enhancers. Int J Pharm 2000;197:
239e41.
[39] Chen H, Banaszak Holl M, Orr BG, Majoros I, Clarkson BH. Interaction of
dendrimers (artificial proteins) with biological hydroxyapatite crystals. J Dent
Res 2003;82:443e8.
[40] Naka K, Tanaka Y, Chujo Y. Effect of anionic dendrimer on the crystallization of
calcium carbonate in aqueous solution: size control of spherical vaterite
particles. Langmuir 2002;18:3655e8.