Adsorption of Tripeptide RGD on Rutile TiO2 Nanotopography Surface in Aqueous Solution
ABSTRACT Molecular dynamics simulations were carried out to investigate the adsorption mechanisms of tripeptide Arg-Gly-Asp (RGD) on the nanotopography and perfect rutile TiO2 (1 1 0) surfaces in aqueous solution. It is shown that the amino groups (NH2 and NH3þ) and carboxyl group (COO�) of RGD are the main groups bonding to hydrophilic TiO2 surface by electrostatic and van der Waals interactions. It is also demonstrated
that RGD adsorbs much more rapidly and stably on the nanotopography surface than the perfect surface. On the hydrophilic TiO2 surface, the water molecules occupy the adsorption sites to form hydration layers, which have a significant influence on RGD adsorption. On the perfect surface, since the fivefold titanium atom is surrounded by surface bridging oxygen atoms above it and has a water molecule bonding to it, the amino group NH2 is the adsorption group. However, because the pit surface exposes
more adsorption sites and has higher surface energy, RGD can adsorb rapidly on the surfaces by amino groups NH2 and NH3þ, and the carboxyl group COO� may edge out the adsorbed water molecules and bond to the surface titanium atom. Moreover, the surface with higher surface energy has more adsorption energy of RGD.
- [Show abstract] [Hide abstract]
ABSTRACT: Immobilization of RGD peptides on titanium (Ti) surfaces enhances implant bone healing by promoting early osteoblastic cell attachment and subsequent differentiation by facilitating integrin binding. Our previous studies have demonstrated the efficacy of RGD peptide immobilization on Ti surfaces through the electrodeposition of poly(ethylene glycol) (PEG) (RGD/PEG/Ti), which exhibited good chemical stability and bonding. The RGD/PEG/Ti surface promoted differentiation and mineralization of pre-osteoblasts. This study investigated the in vivo bone healing capacity of the RGD/PEG/Ti surface for biomedical application as a more osteoconductive implant surface in dentistry. The RGD/PEG/Ti surface was produced on an osteoconductive implant surface, i.e. the grit blasted micro-rough surface of a commercial oral implant. The osteoconductivity of the RGD/PEG/Ti surface was compared by histomorphometric evaluation with an RGD peptide-coated surface obtained by simple adsorption in rabbit cancellous bone after 2 and 4 weeks healing. The RGD/PEG/Ti implants displayed a high degree of direct bone apposition in cancellous bone and achieved greater active bone apposition, even in areas of poor surrounding bone. Significant increases in the bone to implant contact percentage were observed for RGD/PEG/Ti implants compared with RGD-coated Ti implants obtained by simple adsorption both after 2 and 4 weeks healing (P<0.05). These results demonstrate that RGD peptide immobilization on a Ti surface through electrodeposited PEG may be an effective method for enhancing bone healing with commercial micro-rough surface oral implants in cancellous bone by achieving rapid bone apposition on the implant surface.Acta biomaterialia 04/2011; 7(8):3222-9. · 5.68 Impact Factor
Adsorption of tripeptide RGD on rutile TiO2nanotopography surface
in aqueous solution
Dai-Ping Song*, Ming-Jun Chen, Ying-Chun Liang, Qing-Shun Bai, Jia-Xuan Chen, Xiong-Fei Zheng
Precision Engineering Research Institute, Harbin Institute of Technology, Harbin 150001, China
a r t i c l ei n f o
Received 28 February 2009
Received in revised form 12 July 2009
Available online 28 July 2009
a b s t r a c t
Molecular dynamics simulations were carried out to investigate the adsorption mechanisms of tripeptide
Arg-Gly-Asp (RGD) on the nanotopography and perfect rutile TiO2(1 1 0) surfaces in aqueous solution. It
is shown that the amino groups (NH2and NH3
bonding to hydrophilic TiO2surface by electrostatic and van der Waals interactions. It is also demon-
strated that RGD adsorbs much more rapidly and stably on the nanotopography surface than the perfect
surface. On the hydrophilic TiO2surface, the water molecules occupy the adsorption sites to form hydra-
tion layers, which have a significant influence on RGD adsorption. On the perfect surface, since the five-
fold titanium atom is surrounded by surface bridging oxygen atoms above it and has a water molecule
bonding to it, the amino group NH2is the adsorption group. However, because the pit surface exposes
more adsorption sites and has higher surface energy, RGD can adsorb rapidly on the surfaces by amino
groups NH2and NH3
bond to the surface titanium atom. Moreover, the surface with higher surface energy has more adsorption
energy of RGD.
? 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
þ) and carboxyl group (COO?) of RGD are the main groups
þ, and the carboxyl group COO?may edge out the adsorbed water molecules and
Titanium materials are extensively used for biomedical pur-
poses , especially in medical implants and prostheses , be-
causeof their good biocompatibility,
osseointegration, excellent mechanical properties, corrosion resis-
tance and relatively low cost. As the TiO2layer on titanium bioma-
terials surface directly contacts biomolecules, proteins, cells and
solution it has a strong influence on the biological response, and
is therefore an important factor which directly influences the suc-
cess of implant [3,4]. TiO2has three crystalline phases: rutile, ana-
tase and brookite. Rutile is the only stable state, whereas anatase
and brookite are metastable phases, so they can be transformed
irreversibly into rutile at high temperature.
However, it is reported that surface topographic characters of
an implant at the micro/nanoscale are related to the host’s biolog-
ical response . A micro-rough surface, i.e. more than 10 lm in
surface roughness, will affect the mechanical characteristics, the
distribution and transfer of stress, the mechanical interlocking ef-
fect between implant and tissue, as well as the biocompatibility.
Due to its features being on the same size scale as a cell or a bio-
macromolecule, a micro-rough surface in the range from 10 nm
to 10 lm has a weak influence on the mechanical characteristics
of an interface, but a remarkable influence on the biocompatibility
. A micro-roughness of less than 10 nm will have a significant
influence on the interface structures, because the defects (such as
vacancy, grain boundary, step, pit) of the crystal structure within
this size range are active regions for adsorption and further affect
the integration of the implant .
Surface energy has direct influences on two important phenom-
ena of an efficient cell–biomaterial interaction : protein adsorp-
tion , and cell attachment . Increasing the surface energy
improves the adsorption of proteins and attachment of cells to al-
low the integration of tissues . A surface with higher surface
energy promotes the adhesion and spreading of cells [12,13]. Baier
et al. [14,15] showed the relationship between critical surface en-
ergy and cell adhesion: materials with low surface energy show a
low cell attachment. For example, good spreading occurs only
when surface energy is higher than approximately 0.057 J m?2
Cell adhesion is based on the interrecognition and interaction of
the integrin receptor and the protein ligand in extracellular matrix
(ECM) . Many ECM proteins regulating cell adhesion and
migration contain the Arg-Gly-Asp sequence (RGD); these proteins
include fibronectin, vitronectin, fibrinogen, platelet-binding pro-
tein, nestin and collagen. RGD-containing peptides regulate cell
adhesion by linking the cell surface integrin receptors a5b1,
axb1 and a3b1 . Some studies have found that peptide pat-
terning on biomaterial surface can enhance cell adhesion [19–
24]. For example, Verrier et al.  grafted RGD peptides onto
Ti6Al4V implants and their results indicate that the RGD sequence
1742-7061/$ - see front matter ? 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +86 451 86413840; fax: +86 451 86415244.
E-mail address: email@example.com (D.-P. Song).
Acta Biomaterialia 6 (2010) 684–694
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/actabiomat
is able to promote effectively the biointegration of human primary
osteoblastic cells on titanium alloy surface.
Many scholars have researched the adsorption of water and or-
ganic molecules on the TiO2(1 1 0) surface. Kornherr  simu-
lated the adsorption of multilayer water on the TiO2 (1 1 0)
surface and asserted that the adsorption energy of the oxygen-re-
duced surface is more than that of perfect surface. Borodin et al.
 studied the interface of poly(ethylene oxide) (PEO) and TiO2
by molecular dynamics (MD) and reported that surface structure
and electrostatic interactions between PEO and TiO2determine
the nature of PEO relaxation at the TiO2interface. Sushko et al.
 studied the interaction between rutile TiO2(1 1 0) and organic
molecules containing methyl, benzyl and carboxylic by ab initio
calculations and MD simulations. Carravetta and Monti  stud-
ied the interaction of TiO2and peptide (alanine, alanine–glutamic
acid and alanine–lysine) in solution by ab initio and MD simula-
tions. Monti et al.  investigated the effects of interadsorbate
interactions on the binding mechanism of dipeptide/TiO2surface
by MD simulations. However, these studies used the perfect or
the oxygen-reduced rutile TiO2(1 1 0) surface as the adsorption
substrate without considering complex micro/nanoscale surface
In this paper, based on our previous study on the surface energy
of rutile TiO2(1 1 0) , we use real pit surfaces as substrates to
study the effect of surface morphology and water on RGD adsorp-
tion by comparing with the perfect TiO2(1 1 0) surface through
MD. We also analyze the competitive mechanisms and conforma-
tions of RGD and water adsorbed on the surfaces.
2. Methods and modeling
Arg, Gly and Asp peptides have positive charge, zero charge and
negative charge, respectively, as shown in Fig. 1. To simulate the
RGD adsorption, the minimum distances between RGD and TiO2
surfaces are more than 4.6 Å in the initial simulation models; this
distance is much longer than the required distance for direct bond-
ing of RGD to TiO2surfaces.
In the case of rutile TiO2, (1 1 0) is the most stable surface. The
surface contains five- and sixfold titanium atoms and two types of
oxygen atoms (in-plane and bridging). The sixfold titanium atoms
are covered by the outermost bridging oxygen atoms (Obs) on the
surface, while the fivefold titanium atoms (Ti5c) are coordinated
to in-plane oxygen atoms. The (1 1 0) surface unit cell (1 ? 1) has
a dimension of a ? c which corresponds to 6.497 and 2.959 Å along
[?1 1 0] and [0 0 1] directions, respectively .
Since the substrate model with more than four layers away
from the fixed layer describes well the surface characteristics of ru-
tile TiO2(1 1 0) surface , in order to express sufficiently the
nanotopography, our substrate models contain 16 Ti–O layers
and the bottom two layers were fixed, as shown in Table 1. The
substrate of Model I is perfect TiO2(1 1 0) surface with unit cells
(10 ? 18 ? 16). The real rutile (1 1 0) surface has many defects,
such as oxygen vacancies and pits . Because the odd units of
atoms removed along [1 1 0] may result in higher surface energy
, substrates (nanotopography surfaces) of Models II and III
are given by removing some atoms in (5 ? 6 ? 1) and (5 ? 6 ? 3)
from substrate surface of Model I to create pits, respectively. Here
(5 ? 6 ? 1) denotes that the number of atom units cut out from
surface along the [?1 1 0], [0 0 1] and [1 1 0] directions is 5, 6 and
1, respectively. All surface substrates were relaxed for 250 ps,
and their average conformations of the last 50 ps were used as
the adsorption substrates of the perfect TiO2 (1 1 0) adsorption
Model I and of the nanoscale morphology TiO2(1 1 0) adsorption
Models II and III. The starting conformations of the tripeptide in
three models have the same distance, conformation and orienta-
tion, as shown in Fig. 1S and Fig. 2a and b. Surface bridging oxygen
and terminal oxygen around the pit rim are defined as surface oxy-
gen (Os), while surface fivefold titanium and terminal titanium
around the pit rim are defined as surface titanium (Tis).
All simulation models were carried out under the same simula-
tion conditions. Periodic boundary conditions were applied along x
and y directions of the models. The bottom two Ti–O layers were
fixed, while the remaining layers were allowed to relax. MD simu-
lations were carried out at 310.15 K with time step of 0.5 fs. Simu-
lations were fulfilled in the canonical ensemble (i.e. NVT ensemble)
and the Nose–Hoover thermostat  was used to maintain a
constant temperature. The velocity Verlet algorithm was used to
calculate the atomic motions and the particle–particle particle–mesh
(PPPM)  solver was applied to the calculate the electrostatic
interactions. A 12.0 Å cutoff distance was used for van der Waals
The interaction potential parameters necessary to carry out
MD simulations of the rutile/RGD/aqueous solution system were
taken from published data [26,28,35,36], existing force fields
 and parameters derived. TiO2 was calculated by using the
Buckingham potential , water molecules were described by
using the TIP3P model , and the AMBER force field 
was employed to adequately represent the peptide structures.
Their interactions were calculated by using the Lennard–Jones
potential and the Buckingham potential. The SHAKE algorithm
 was used to constrain all bond lengths and bond angles of
The following is the adsorption procedure:
1. Optimization of the solvent energy: while the RGD structure
(solute) and the TiO2surface were kept fixed, the energy of
water solvent was minimized to remove the abnormal VDW
Fig. 1. RGD (Arg-Gly-Asp). Color codes: oxygen, red; hydrogen, white; nitrogen,
blue; carbon, dark grey. (For interpretation of color mentioned in this figure the
reader is referred to the web version of the article.)
Items Model IModel IIModel III
Unit cells of substrate
Pit size of substrate
Surface energy of
Number of TiO2units
Number of water
Number of RGD
Number of total atoms
(10 ? 18 ? 16)
(10 ? 18 ? 16)
(5 ? 6 ? 1)
(10 ? 18 ? 16)
(5 ? 6 ? 3)
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
interaction by using the Polak–Ribiere conjugate gradient
2. The solvent was simulated by restraining molecular dynamics
(RMD). After the energy minimization, the RGD and substrate
were kept fixed, and the water molecules around RGD were
simulated for 20 ps by MD, i.e. the water solvent was equili-
brated for solvent molecules to fill empty space.
3. Special atoms (OCOO?) of RGD were simulated for 50 ps by RMD.
4. The adsorption simulations (production simulations) were run
for 8 ns for Model I and for 3 ns for Models II and III by MD.
The results were analyzed.
3. Results and discussion
3.1. Adsorption conformation of RGD
It can be seen from Fig. 3 that water molecules reach the hydro-
philic TiO2(1 1 0) surface earlier than RGD. This process is the
same as that reported by Kasemo . At the initial stage, water
molecules H2O?1and H2O?2adsorb to surface oxygen Os?1and
Os?2on the nanotopography surfaces (Fig. 5S), respectively, and
water molecules H2O?1, H2O?2and H2O?3adsorb to surface tita-
nium Tis?1(Fig. 6S); subsequently all of them are edged out by
RGD, except H2O?1in Fig. 6S. Water molecule H2O?3which ad-
sorbed to surface oxygen Os?2in Model II at about 0.46 ns is also
edged out by RGD at about 0.5 ns. Water molecules interact
and bind very differently on surfaces depending on the surface
properties. They form the surface water shells which influence
On the perfect surface, only guanido nitrogen atoms NNH2of the
side chain of Arg bond to surface oxygen atoms, and their distances
are shown in Fig. 4. After prolonged equilibration, guanido group
NH2adsorb on the surface at about 1.84 ns. The type of bond be-
tween RGD and the surface atoms does not change once RGD has
adsorbed on the surface, i.e. the amino nitrogen NNH2forms hydro-
gen bonds (HBs) with the surface bridging oxygen atoms, as shown
in the dashed lines in Fig. 3. Although one HB sometimes breaks up
in order to adjust the adsorption conformation of RGD, the other
HB remains and the broken HB is rapidly remade. Its average HB
number, the average bond length N–Osand the average bond angle
\NHOsis 0.98, 2.93 ± 0.15 Å and 151.96 ± 11.20?, respectively.
Nanotopography surfaces affect significantly the adsorption
conformation of RGD. The amino nitrogen atoms NNH2and NNH3þ
are close to surface oxygen in Model II (Fig. 5S). The former makes
HBs with Osfrom about 0.1 to 0.5 ns, and the latter makes HBs with
Osafter about 0.5 ns. The durations of water oxygen contacting
surface oxygen are very short (less than 30 ps), which are in good
agreement with their residence times shown in Tables 1S and 2S.
On the shallow pit surface, the amino nitrogen NNH2forms HBs
with the surface oxygen atoms Os?1in the pit at 0.4 ns (Fig. 5a).
The average bond length and the average bond angle is
2.94 ± 0.15 Å and 151.53 ± 9.76?, respectively. At 0.7 ns, the amino
Os?2exposed around the pit rim and with the water oxygen atoms
þof the N-terminus forms a HB with the surface oxygen atom
Fig. 3. Conformation of RGD adsorbed on the perfect TiO2(1 1 0) in Model I at 1.0 ns (a), 2.0 ns (b) and 4.0 ns (c). The green dashed lines represent the hydrogen bonds. TiO2,
RGD and water molecules are shown by CPK mode, ball-and-stick mode and line mode, respectively. Color codes: oxygen, red; hydrogen, white; nitrogen, blue; carbon, dark
grey; titanium, light grey. (For interpretation of color mentioned in this figure the reader is referred to the web version of the article.)
Fig. 2. Initial RGD adsorption models on nanoscale morphology TiO2in Model II (a) and Model III (b). To clearly display the deep pit, the water molecules are not displayed in
(b). TiO2, RGD and water molecules are shown by CPK mode, ball-and-stick mode and line mode, respectively. Color codes: oxygen, red; hydrogen, white; nitrogen, blue;
carbon, dark grey; titanium, light grey. (For interpretation of color mentioned in this figure the reader is referred to the web version of the article.)
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
bonding to the surface oxygen by HB (Fig. 5b). With the increase in
absorption time, the adsorption conformation becomes more and
more stable. As shown in Fig. 5c, the distances from the N atom
and from the C atoms of the Arg backbone to Os?2are slightly re-
duced at 2.5 ns. The amino NH3
HB with the oxygen atom and with the water oxygen atoms bond-
ing to the surface oxygen by HB, which forms a HB network. The
averagebond lengthand the
2.88 ± 0.13 Å and 145.96 ± 9.58?, respectively.
As shown in Fig. 6, on the deep pit surface in Model III, two car-
boxyl oxygen atoms of the side chain of Asp bond to the surface
titanium atom Tis?1 around the pit rim at 0.107 and 0.412 ns
(Fig. 6S), respectively. Their average distances are 2.03 ± 0.05 and
2.05 ± 0.06 Å, respectively. Their stable bonds will not break up
once formed, as shown in Table 1S and Fig. 6S. Strehle et al. 
demonstrated the adsorption of the protein on the TiO2surface
via the carboxylate groups of the protein side chains and C-termi-
nus by means of Raman spectroscopy. With increasing absorption
time, RGD orientates along the long axis of the pit. Moreover, the
amino nitrogen NNH2of the side chain of Arg bonds to surface oxy-
gen atom Os?1at 2.3 ns. At least one nitrogen atom remains within
the HB distance, and their average distances are 2.89 ± 0.15 Å.
A more detailed description of the conformational dynamics
and flexibility of the RGD tripeptide segments can be obtained by
þof the N-terminus still forms a
analyzing the distribution of backbone dihedral angles U (C–N–
Ca–C) and W (N–Ca–C–N) together with their time evolution dur-
ing the course of the simulation, as shown in Fig. 7 and Fig. 4S. In
Model II, the dihedral angle pair (U and W) of Gly turns at about
0.45 ns and about 0.8 ns (Fig. 7a), which results from the RGD
adsorbing on different sites of the pit surface. Although U of Gly
turns significantly at about 1.53 ns and about 1.98 ns, it returns
after only 0.18 and 0.015 ns, respectively. The backbone dihedrals
reach equilibration at about 0.7 ns in Model III which is less than
that in Models I (about 4.7 ns) and II (about 0.8 ns). The dihedral
anglepair ismainly distributed
(?75.16 ± 22.67?, ?60.44 ± 21.47?) at the earlier adsorption stage
and (?81.80 ± 27.35?, 70.63 ± 21.61?) at the later adsorption stage
in Model I (Ramachandran plot  in Fig. 7S). The pit has a signif-
icant influence on the dihedral angle pair distribution. The dihedral
angle pairs center on the allowed region (?81.67 ± 21.34?,
?67.07 ± 20.71?) and (72.73 ± 24.16?, 49.89 ± 16.27?) in Models II
and III, respectively. In addition, the backbone dihedral angles x
(Ca–C–N–Ca), identifying a trans (180?) or cis (0?) arrangement of
the peptide bond, oscillate around 180?, implying trans peptide
in theallowed regions
3.2. Radial distribution functions (RDFs), coordination number and
To clearly describe the adsorption conformation and adsorption
mechanism of RGD on rutile TiO2in aqueous solution, as well as
the competition effect of RGD with water molecules, we analyzed
the static and dynamic parameters of adsorption systems, such
as radial distribution functions, coordination number and resi-
dence time, as shown in Table 1S.
Coordination number is the total number of neighbors of a cen-
tral atom in a molecule or an ion. It represents a static property of
atom distribution. The coordination number ncoorof an atom is ob-
tained from the RDF g(r)4pr2dr by integrating from 0 up to the first
minimum rfminof g(r) .
Residence time represents the dynamic property of atom bond-
ing. To obtain a precise definition of the residence time, we used
the function pc,j(tk,t;t*) [43,44] which is a property of the molecule
j and is equal to either 0 or 1. It takes the value 1 if molecule j lies
within the first coordination shell of the atom c at both time steps t
and t + tk, and in the interim does not leave the coordination shell
for any continuous period longer than t*. Under all other circum-
stances, it takes the value 0. We defined an average quantity
Pc(t), characteristic of the atom c, by the expression:
Fig. 5. Conformation of RGD adsorbed on the shallow pit TiO2(1 1 0) in Model II at 0.4 ns (a), 0.7 ns (b) and 2.5 ns (c). The green dashed lines represent the hydrogen bonds.
TiO2, RGD and water molecules are shown by CPK mode, ball-and-stick mode and line mode, respectively. The waters bonding to RGD are also described in ball-and-stick
mode. Color codes: oxygen, red; hydrogen, white; nitrogen, blue; carbon, dark grey; titanium, light grey. (For interpretation of color mentioned in this figure the reader is
referred to the web version of the article.)
Fig. 4. The distances from the amino nitrogen NNH2of the side chain of Arg to
surface oxygen in Model I.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
In this definition, Pc(0) is the coordination number of atom c;
Pc(t) decays exponentially, i.e. Pc(t) = Pc(0)exp(?t/s), where s is
the residence time. When PcðtÞ=Pcð0Þ are all 1 during sampling
time, the residence time s is considered as infinity. According to
the literature , we set t*equal to 2 ps.
Os–Hw(water hydrogen), Os–Ow(water oxygen), Tis–Owand
Os–NNH2(peptide guanido nitrogen) in Model I, Os–NNHþ
amino nitrogen) RDFs in Model II, Os–NNH2and Tis–OCOO? (peptide
carboxyl oxygen) RDFs in Model III are shown in Fig. 8. RDFs of
RGD, water and surface atoms (except Os–NNH2, Os–NNHþ
OCOO?) in Models I, II and III are similar, as shown in Figs. 8S, 9S and 10S.
Tis–Owhas a sharp first neighbor peak centered at about 2.01 Å
that corresponds to the first layer of molecularly adsorbed water
molecules, whereas the well-separated lower second peak cen-
tered at about 3.58 Å corresponds to the second water layer, which
comprises water molecules interacting with the first layer of the
terminal and bridging oxygens. The data are in agreement with
experimental X-ray findings  which gave average positions of
2.1 and 3.8 Å for the first and second peaks, respectively. The pres-
ence of a very low amplitude of RDF between the first and the sec-
ond peaks indicates that water molecules do not diffuse but are
truly adsorbed onto the TiO2surface.
Os–Hwand Os–Owhave their first peaks centered at about 2.0
and 2.88 Å, respectively; their first minima are centered at about
3.15 and 3.86 Å, respectively. According to the definition of a HB,
the water molecules of the first water layer around Osbond to
the surface bridging oxygen atoms by HBs. Compared with the
water adsorbed on Tis, there is an obvious amplitude between
the first peak and the second peak of the Os–OwRDF, which indi-
cates that they interchange.
On the perfect surface, every fivefold titanium atom adsorbs a
water molecule and the coordination number is 1. On the nanoto-
pography surfaces, the titanium atoms with the dangling bonds
around the pit rim can adsorb two or three water molecules, so
that their average ncoorof Tis–Oware slightly more than 1. Simi-
larly, ncoorof Os–Owin Models II and III are more than that in Model
I. On highly hydrophilic TiO2 surface, s of Tis–Owis infinite in
Fig. 6. Conformation of RGD adsorbed on the deep pit TiO2in Model III at 1.0 ns (a) and 2.5 ns (b). The green dashed lines represent the hydrogen bonds. TiO2, RGD and water
molecules are shown by CPK mode, ball-and-stick mode and line mode, respectively. The waters bonding to RGD are also described in ball-and-stick mode. Color codes:
oxygen, red; hydrogen, white; nitrogen, blue; carbon, dark grey; titanium, light grey. (For interpretation of color mentioned in this figure the reader is referred to the web
version of the article.)
Fig. 7. Distributions of backbone dihedral angles of RGD adsorbed on the shallow pit surface in Model II (a) and the deep pit surface in Model III (b) as a function of simulation
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
Models I and II, which also indicates that the water molecule
hardly desorbs once it adsorbs on surface titanium. However, in
Model III, because the two water molecules adsorbed to Tiswere
edged out by Ocoo?, s of Tis–Owis 6.5 ? 106ps.
For water molecules more than 15 Å from substrate surface,
ncoor= 4.5 is obtained. This value is slightly lower than the value
of ncoor= 4.9 obtained by Obst and Bradaczek , who used a
TIP3P-water model, but is close to the experimental value of
ncoor= 4.5 given by Soper and Phillips  or ncoor= 4.4 given by
Narten and Levy  and Gaballa and Neilson . The residence
time is a function of temperature . For pure water, Bellissent-
Funel and Teixera  derived a residence time of 1.25 ps at
293 K and more than 10 ps at 253 K from neutron scattering exper-
iments. In addition, Bopp  obtained an experimental value of
2.5–10 ps in the bulk. Our result from the MD simulation is about
2.6 ps at 310 K.
The distances from NNH3þ (in Models I and III) and NNH2(in Mod-
el II) to Osare far away from the HB distance suggested by X-ray
and neutron diffraction experiments . However, in Models I
and III, RDFs of the amino nitrogen NNH2of the side chain of Arg
have sharp first neighbor peaks rfmax at about 2.85 and 2.87 Å
and the first minima rfminat about 3.64 and 3.75 Å, respectively.
In Model II, RDF of the amino nitrogen NNH3þ of N-terminus has a
sharp first neighbor peak at about 2.77 Å and a first minimum at
about 3.41 Å. These data are in agreement with the results from
Carravetta et al.  which are 2.77 and 3.6 Å for rfmaxand rfmin,
respectively. Their coordination numbers are more than 1. The
simulation results indicate that these amino nitrogen atoms bond
to surface oxygen.
Among the interactions between surface oxygen and other
atoms, the residence time of NNH2–Oson the perfect surface is max-
imum, up to 35.05 ps, and that of Os–Owtakes the second place, i.e.
15.4 ps. In Model II, the surface oxygen forms the stable bond with
NNH3þ and its residence time is infinite. On the deep pit surface
(Model III), the residence times of surface atoms and RGD are more
than those in the perfect surface and shallow pit surface, as shown
in Table 1S.
3.3. Hydrogen bond
For the analysis of HBs between the water molecules, the fol-
lowing common geometric definition of a HB was used: two water
molecules are said to be hydrogen bonded, if the distance between
their two oxygen atoms is less than 3.4 Å and the O–H...O angle is
larger than 135?. Similar definitions have been used by other
authors [44,46,53]. The number nHBof HBs is given as the time
average of the number of HBs both donated and accepted to the
number of water molecules at the same distance from the atom
or a randomly chosen water molecule, as shown in Table 2S.
Jorgensen et al.  reported a value of nHB= 3.50 at 298 K for
the TIP3P-water model. Sciortino and Fornili  found values of
nHB between 0.5 and 4.6 for minimum lifetimes between 3.95
and 0 ps, respectively. The average number and the average length
of HBs in our simulation are nHB= 3.09 and LHB= 2.88 Å, respec-
tively, corresponding to the minimum HB lifetime of 0.35 ps in
Sciortino’s work . However, our residence times of HBs are
more than Sciortino’s  minimum HB lifetimes and Bellissent-
Funel’s experimental results .
Due to the side-chain effect, nHBof carbonyl oxygen (OCO) of Arg
is about 1, which is about half of that of Gly. Peptide carboxyl oxy-
gen (OCOO?), especially carboxyl oxygen of C-terminus, forms the
most stable HB with water, and their residence times are 7.91,
8.5 and 33.88 ps on the perfect surface, the shallow pit surface
and the deep pit surface, respectively.
nHBof NNH2–Osin Model I are close to 1 and their residence
times are more than 7 ps. Although the distances from NNH3þ to
Osin Model II and from NNH2to Osin Model III are always within
the HB range, due to the adjustment of adsorption conformation
of RGD during the equilibrations, NNH3þ–Osand NNH2–Ossometimes
do not meet the angle condition of HB and their nHBare less than 1.
3.4. Adsorption mechanism of RGD
To study the adsorption mechanism of RGD, we analyzed the
function of the interaction energies of water and RGD with the sur-
face atoms during the course of the simulation, as shown in Fig. 9.
Adsorption of protein on various materials has been widely
studied and it has been found that factors such as electrostatic
interaction, hydrophobic interaction, hydrogen bonding, VDW
interaction and specific chemical interactions between protein
and the adsorbent play important roles [56–71]. Short peptide se-
quences interact with TiO2surface via electrostatic interaction also
[57–59]. Yang et al.  and Chen et al.  stated that electro-
static interaction may be the main interaction between fibrinogen
and TiO2surfaces. Our results also indicate that electrostatic inter-
action and VDW interaction (including HB interaction) may be the
Fig. 8. RDFs of TiO2surfaces, water and RGD.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
main interactions between RGD and hydrophilic TiO2surfaces. Sur-
face atoms and some hydroxyls retained on TiO2surface can inter-
act with RGD through these interactions, and the interaction
energies of water and RGD with TiO2surface atoms are shown in
Table 2. The amino groups (NH2and NH3
(COO?) are the main groups bonding to TiO2 surface, which is
the same as the adsorption of the 10th type III module of fibronec-
tin (FN-III10) on hydroxyapatite (HAP) (0 0 1) surface . Shen
et al.  found that the charged COO?and NH3
groups that interact with HAP (0 0 1) surface, while other groups,
such as charged guanido, neutral amino and hydroxyl, have consid-
erable interactions with the surface. The surface structures have
significant influence on the binding types of RGD–TiO2surface, as
shown in Table 3. On the nanotopography surface with higher sur-
face energy, the adsorption via carboxyl group is more realistic,
which corresponds to the experimental findings of Strehle et al.
As a direct characterization of surface adsorption, adsorption
energy is the energy difference between systems before and
after adsorption; the greater the adsorption energy, and the
more stable the adsorption. The adsorption energy of RGD on
the perfect surface is ?59.28 kcal mol?1, while the adsorption
energies are ?106.14 and ?237.78 kcal mol?1for the shallow
pit surface and the deep pit surface, respectively. These results
confirm that the adsorption of RGD on the nanotopography sur-
face with higher surface energy is more stable than that on the
It can be seen from Table 2 that there is a maximum interaction
energy of RGD with TiO2and a minimum interaction energy of
water with TiO2. On a perfect surface, however, surface fivefold
titanium atoms are surrounded by surface bridging oxygen atoms
þ) and carboxyl group
þare the strongest
above them. Once water adsorbs onto Ti5cand interacts with Obs
around it, they hardly desorb and RGD has no chance to approach
Ti5c. Although water forms HB with Obs, the residence time of its
HB and Ow?Obsis less than 3 and 20 ps, respectively, as shown
in Tables 2S and 1S. Therefore, the amino group of RGD has chance
to edge out the adsorbed water molecules and to form HB with Obs,
as shown in Fig. 9(a) and Fig. 5S.
However, the nanotopography surfaces expose more terminal
titanium and oxygen atoms around the pit rim. The strongest elec-
trostatic interaction occurs at surface peak regions and the weakest
interaction at surface valley regions. The pit weakens the interac-
tion of the atoms around the pit and many unstable undercoordi-
nated atoms remain. These atoms offset their initial positions
after relaxation, which leads to a significant increase in the surface
energy. The surface energy affects significantly the adsorption of
RGD: the higher the surface energy, the more stable the adsorption
is, which is in good agreement with the results from Baier et al.
[14,15] who stated that a surface with higher surface energy pro-
motes cell adhesion and spreading. Moreover, some studies also
found that TiO2polymorphs with higher surface energy bind water
more tightly  and the adsorption energies of hydrogen on the
surfaces with higher surface energies are larger . Higher sur-
face energy means that RGD can adsorb rapidly on the surfaces
via the amino groups NH2and NH3
group COO?may edge out the adsorbed water molecules and bond
to the surface titanium atom, as shown in Fig. 9b and Fig. 6S. How-
ever, the deeper pit is likely to restrain the movement of RGD. As
shown in Fig. 6b, both sides of RGD bond stably to the surface,
i.e. on one side Ocoo? bonds to titanium and on the other side
NNH2bonds to Os, which affects further cell recognition and
þ, and even that the carboxyl
?19.52 ± 1.44
?62.26 ± 2.74
?96.28 ± 4.19
?108.82 ± 5.18
?362.50 ± 7.35
Note: ‘...’ denotes hydrogen bond.
Binding types of RGD–TiO2surface and adsorption energy.
Binding types of RGD–TiO2surfaceAdsorption
Mode II NH2...Os, NH3
Mode III NH2...Os, COO?–Tis
Note: ‘...’ denotes hydrogen bond.
Fig. 9. Interaction energy of water and RGD with the surface atoms in Model II (a) and Model III (b) as a function of simulation time.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
Furthermore, the binding types of RGD–TiO2surface affect the
displacement of surface atom. Peptide carboxyl oxygen atoms
bonding to surface titanium atom result in the titanium atom being
greatly displaced, while peptide amino nitrogen atoms forming
HBs with surface oxygen atom do not affect the displacement of
the oxygen atom, which fluctuates around its initial position just
as a normal surface atom, as shown in Fig. 10.
3.5. Hydration layers
The water density q(z) has at least three peaks on the perfect
surface, while the densities of water in the pit have two and three
peaks in Models II and III, respectively, as shown in Fig. 11. How-
ever, the densities of water on the outermost surfaces in three
models have the same distributions. When z > 10 Å, water mole-
cules are distributed uniformly, as in bulk water. Water density
convergesto0.971 g cm?3, which
0.975 g cm?3from calculation  and 0.994 g cm?3from experi-
As a way of understanding the bonding mechanism of water
with the surface, we calculated the orientation order parameters
as a function of distance to the surface. The orientation order
parameters SU(z) , Sa(z) and Sb(z) were given by:
isin agreement with
where aidescribes the steric angle of the water molecule with the
surface by measuring the angle between the surface and the vector
(i.e. the rotational axis) of the O atom to the center of the two H
atoms of the molecule. Ranging from ?90? to +90?, a value of
+90? corresponds to an orthogonally orientated water molecule
with O down and both H atoms up. bi, on the other hand, is the tilt-
ing angle between the rutile surface and the plane defined by the
three atoms of the water molecule, with 0? corresponding to paral-
lel and 90? to orthogonal orientation of the water molecule with re-
spect to the surface plane. Nzis the number of water molecules at a
distance z from the surface. For SU(z), a value of ?0.5 means that the
water rotational axis is perpendicular to the surface, and a value of
1 indicates that it is parallel to the surface. Sa(z) and Sb(z) are the
mean values of aiand biof the water molecules at a distance z from
the surface, respectively.
It can be seen from Fig. 12 that there is a clear ordering of water
orientation in the first few angstroms from the perfect surface in
Model I. According to the peak-valley of surface water density
and the plus–minus alternation of Sa(z), there are at least four
water layers on the surface. In the first layer, a and b of the water
molecules are 60.1? and 71.2?, respectively, i.e. their conformations
are O-down and H-up. In the second layer,a of the water molecules
changes sharply to ?35.6?; the conformations also convert into the
O-up and H-down and form HBs with the water oxygen in the first
layer and the surface oxygen. The water orientation in the first
layer derived from our simulations is clearer than that from Kornh-
layer= ?33.2?, b2nd-layer= 70.0?. SU(z) also increases from ?0.17 to
0.58 and then sharply decreases to 0.37 at the boundary with the
second layer. This probably indicates that the water molecules ro-
tate as they move away from the surface to form HBs with other
water molecules and then quickly reorientate upon entering the
second layer. We can also see a similar sharp change in orientation
at the limit between the second and third layers. In both the third
and fourth layers, the water molecules still show some order in
their orientation, and further away from the surface, water orien-
tation seems to be completely random, Sa(z) ? 0?, Sb(z) ? 57.3?,
SU(z) ? 0.5.
The thicknesses of the first two hydration layers on the perfect
surface are 1.05 and 1.45 Å, respectively. The probability distribu-
tions of their positions on the surface are shown in Fig. 13. All
water molecules in the first hydration layer are located above
Ti5cand bond stably to Ti5c(Fig. 13a). The water molecules in the
second hydration layer are over Obs and form HBs with Obs
(Fig. 13b), while the water molecules in the third hydration layer
are over Ti5c(Fig. 11Sc) and form HBs with water molecules in
the first and second hydration layers. The water molecules in the
fourth hydration layer are distributed randomly (Fig. 11Sd). There
is a low-probability region in the bottom right corner of Fig. 11Sd,
which is the position of RGD adsorbed on the surface. Around the
pit rims, due to the relaxed displacement of the surface atoms with
the dangling bonds, water molecules in the first hydration layer lo-
cate around the surface titanium atoms (Figs. 12S and 13S), which
is different from the titanium atoms on the perfect surface.
To understand further the competition between RGD and water
for surface adsorption sites, we analyzed the residence times of
water molecules in the first two hydration layers. In the first
Fig. 10. Displacements of the surface atoms bonding to RGD in Models II and III.
Fig. 11. Surface water density as a function of distance from the outermost surfaces
in Models I, II and III.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
hydration layer, the water molecules that bond to Tison the perfect
surface have infinite residence time, which indicates that they will
not desorb once they adsorb on the surface. In the second hydra-
tion layer, the residence time of the water molecules that form
HBs with Osand Owin the first layer is 15.4 ps. When z > 15 Å, it
is 2.6 ps, which is in good agreement with the experimental value
2.5–10 ps of water in the bulk provided by Bopp . Therefore, on
the perfect surface, the first hydration layer is very stable and the
water in the second hydration layer diffuses and interchanges
those in the outermost layers, which makes it probable that the
RGD amino group NH2form HBs with Oswithout any external
force. However, the pit surfaces have more adsorption sites and
higher surface energy; hence RGD can adsorb rapidly on the sur-
faces via the amino groups NH2and NH3
COO?can even edge out the adsorbed water molecules and bond to
Tis. Furthermore, water molecules constitute a well-defined first
solvation shell coordinated with the carboxyl, carbonyl, guanido
þ, and the carboxyl group
and amino groups of RGD. The adsorbed water layers form HB
interactions with the hydrophilic groups of RGD, as shown in Fig. 5.
The adsorption mechanism of RGD on rutile TiO2(1 1 0) sur-
faces and the effects of nanotopography and water on RGD adsorp-
tion were investigated by MD simulations.
The water molecules occupy the adsorption sites on the hydro-
philic TiO2surfaces, i.e. the water oxygen atoms bond to the sur-
face titanium atoms to form the stable first hydration layer and
interact with the surface oxygen atoms to form the second hydra-
tion layer. Besides being in competition with RGD for the adsorp-
tion sites, the adsorbed water layers also play an intermediary
role, forming HB interactions with the hydrophilic groups of RGD.
The guanido (NH2), amino (NH3
of tripeptide RGD are the main groups bonding to TiO2surface
by electrostatic and VDW interactions. On the perfect surface, since
fivefold titanium atoms are surrounded by surface bridging oxygen
atoms above them and water molecules bonding to them, the
guanido group is the adsorption group. However, the nanotopogra-
phy surfaces have more adsorption sites and higher surface energy,
which results in RGD being able to adsorb rapidly on the surfaces
via the guanido and amino groups; moreover, the carboxyl group
may even edge out the adsorbed water molecules and bond to
the surface titanium atom. Although the adsorption of RGD on
the nanotopography surface with higher surface energy is most
stable and RGD orientates along the long axis of the pit, the surface
is likely to restrain the movement of RGD and affect further recog-
nition and adhesion of cell. In addition, the binding types of RGD–
TiO2surface may affect the displacement of surface atoms. The
bonding of peptide carboxyl oxygen atoms to surface titanium
atom causes the the titanium atom to be greatly displaced.
The adsorption of RGD on the rutile TiO2nanotopography sur-
faces is more stable and rapid than that on the perfect surface.
The surface with higher surface energy has more adsorption energy
of RGD. The adsorption energies of RGD on the shallow and deep
pit surfaces are ?106.14 and ?237.78 kcal mol?1, respectively,
which are at least 1.8 times as big as that on the perfect surface.
RGD adsorbs on the pit surfaces at about 0.1 ns, while the adsorp-
tion time on the perfect surface is at about 1.84 ns. Moreover, the
adsorption equilibration time is about 4.7 ns on the perfect surface,
which is much more than that on the shallow and deep pit sur-
faces, i.e. about 0.8 and 0.7 ns, respectively.
These results will help to investigate theoretically the patterned
peptide surface and the cell adhesion mechanism and to design the
surface structure of biomaterials.
þ) and carboxyl groups (COO–)
This work is funded by the National Natural Science Foundation
of China (No. 50675050 and 50875066) and the Multidiscipline
Scientific Research Foundation of Harbin Institute of Technology
(No. HIT. MD 2003. 10).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.actbio.2009.07.032.
Appendix B. Figures with essential color discrimination
Certain figures in this article, particularly Figures 1–13, are dif-
ficult to interpret in black and white. The full colour images can be
found in the on-line version, at doi: 10.1016/j.actbio.2009.07.032).
Fig. 12. Surface water density and orientation order parameter as a function of
distance from perfect surface in Model I.
Fig. 13. Probability distributions of water on the perfect TiO2(1 1 0) surface in
Model I: (a) first hydration layers; (b) second hydration layers. Color codes: surface
bridging oxygen, green asterisk; fivefold titanium, red plus-sign. (For interpretation
of color mentioned in this figure the reader is referred to the web version of the
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
 Leyens C, Peters M. Titanium and titanium alloys [Chen ZH et al. Trans.].
Beijing: Chemical Industry Press; 2005
 Diebold U.Thesurfacescience
 Hunter A, Archer CW, Walker PS, Blunn GW. Attachment and proliferation of
osteoblasts and fibroblasts on biomaterials for orthopaedic use. Biomaterials
 Boyan BD, Hummert TW, Dean DD, Schwartz Z. Role of material surfaces in
regulating bone and cartilage cell response. Biomaterials 1996;17:137–46.
 Liang YC, Song DP, Chen MJ, Bai QS. New development on effect of surface
roughness of Ti-material for biomedical applications on cell adhesion. Chin J
Mech Eng 2008;44:6–15.
 Kasemo B. Biocompatibility of titanium implants: surface science aspects. J
Prosth Dent 1983;49:832–7.
 Kasemo B, Lausmaa J. Material selection surface characteristics and chemical
processes at implant surface. Chicago, IL: Quintessence; 1985.
 dos Santos EA, Farina M, Soares GA, Anselme K. Surface energy of
hydroxyapatite and b-tricalcium phosphate ceramics driving serum protein
adsorption and osteoblast adhesion. J Mater Sci Mater Med 2008;19:2307–16.
 Hao L, Lawrence J. The adsorption of human serum albumin (HSA) on CO2laser
modified magnesia partially stabilised zirconia (MgO-PSZ). Colloids Surf B
 Kennedy SB, Washburn NR, Simon Jr CG, Amis EJ. Combinatorial screen of the
effect of surface energy on fibronectin-mediated osteoblast adhesion,
spreading and proliferation. Biomaterials 2006;27(20):3817–24.
 Poulsson AHC, Mitchell SA, Davidson MR, Johnstone AJ, Emmison N, Bradley
RH. Attachment of human primary osteoblast cells to modified polyethylene
surfaces. Langmuir 2009;25:3718–27.
 van Kooten TG, Schakenraad JM, van der Mei HC, Busscher HJ. Influence of
substratum wettability on the strength of adhesion of human fibroblasts.
 Redey SA, Razzouk S, Rey C, Bernache-Assollant D, Leroy G, Nardin M, et al.
Osteoclast adhesion and activity on synthetic hydroxyapatite, carbonated
hydroxyapatite, and natural calcium carbonate: relationship to surface
energies. J Biomed Mater Res 1999;45:140–7.
 BaierRE.Surface propertiesinfluencing
York: Academic Press; 1970.
 Baier RE, Meyer AE, Natiella JR, Natiella RR, Carter JM. Surface properties
determine bioadhesive outcomes: methods and results. J Biomed Mater Res
 Schakenraad JM, Busscher HJ, Wildevuur CRH, Arends J. The influence of
substratum surface free energy on growth and spreading of human fibroblasts
in the presence and absence of serum proteins. J Biomed Mater Res
 Singhvi R, Stephanopoulos G, Wang IIC. Review: effects of substratum
morphology on cell physiology. Biotech Bioeng 1994;43:764–71.
 Yao KD, Yin YJ. Biomaterials related to tissue engineering. Beijing: Chemical
Industry Press; 2003.
 Hasenbein HE, Andren TT, Bizios R. Micro-patterned surfaces modified select
peptides promote exclusive interactions with osteoblasts. Biomaterials
 Pierschbacher MD, Ruoslahti E. Cell attachment activity of fibronectin can be
duplicated by smallsynthetic
 Zreiqat H, Akin FA, Howlett CR, et al. Differentiation of human bone-derived
cells grown on GRGDSP-peptide bound titanium surfaces. J Biomed Mater Res
 Schliephake H, Scharnweber D, Dard M, et al. Effect of RGD peptide coating of
titanium implants on periimplant bone formation in the alveolar crest: an
experimental pilot study in dogs. Clin Oral Implan Res 2002;13(3):312–9.
 Schliephake H, Scharnweber D, Dard M, et al. Functionalization of dental
implant surfaces using adhesion molecules. J Biomed Mater Res B Appl
 Verrier S, Pallu S, Bareille R, et al. Function of linear and cyclic RGD containing
 Kornherr A. Multilayer adsorption of water at a rutile TiO2 (1 1 0) surface.
 Borodin O, Smith GD, Bandyopadhyaya R, Byutner O. Molecular dynamics
study of the influence of solid interfaces on poly(ethylene oxide) structure and
dynamics. Macromolecules 2003;36:7873–83.
 Sushko ML, Gal AY, Shluger AL. Interaction of organic molecules with the TiO2
(1 1 0) surface. Ab inito calculations and classical force fields. J Phys Chem B
 Carravetta V, Monti S. Peptide–TiO2surface interaction in solution by ab initio
and molecular dynamics simulations. J Phys Chem B 2006;110:6160–9.
 Monti S, Carravetta V, Zhang WH, Yang JL. Effects due to interadsorbate
interactions on the dipeptide/TiO2surface binding mechanism investigated by
molecular dynamics simulations. J Phys Chem C 2007;111:7765–71.
 Song DP, Liang YC, Chen MJ, Bai QS. Molecular dynamics study on surface
fragments of themolecule. Nature
(1 1 0).ApplSurf Sci
 Abrahams SC, Bernstein JL. Rutile-normal probability plot analysis and
accurate measurement ofcrystal
 Mezhenny S, Maksymovych P, Thompson TL, Diwald O, Stahl D, Walck SD. STM
studies of defect production on the TiO2(1 1 0)-(1 ? 1) and TiO2(1 1 0)-(1 ? 2)
surfaces induced by UV irradiation. Chem Phys Lett 2003;369:152–8.
 Hoover WG. Canonical dynamics: equilibrium phase-space distributions. Phys
Rev A 1985;31:1695–7.
 Eastwood H. Computer simulation using particles. New York: Adam Hilger;
 Matsui M, Akaogi M. Molecular dynamics simulation of the structural and
physical properties of the four polymorphs of TiO2. Mol Simul 1991;6:
 Jorgensen WL. Quantum and statistical mechanical studies of liquids. 10.
Transferable intermolecular potential functions for water, alcohols, and ethers
application to liquid water. J Am Chem Soc 1981;103:335–50.
 Cornell WD, Cieplak P, Bayly CI, Gould IR, Merz KM, Ferguson DM, et al. A
second generation force field for the simulation of proteins, nucleic acids, and
organic molecules. J Am Chem Soc 1995;117:5179–97.
 Ryckaert JP, Ciccotti G, Berendsen HJC. Numerical integration of the cartesian
equations of motion of a system with constraints: molecular dynamics of n-
alkanes. J Comp Phys 1977;23:327–41.
 Kasemo B. Biological surface science. Surf Sci 2002;500:656–77.
 Strehle MA, Rösch P, Petry R, Hauch R, Kiefer W, Popp J. A Raman spectroscopic
study of the adsorption of fibronectin and fibrinogen on titanuim dioxide
nanoparticles. Phys Chem Chem Phys 2004;6:5232–6.
 Koradi R, Billeter M, Wüthrich K. MOLMOL: a program for display and analysis
of macromolecular structures. J Mol Graphics 1996;14:51–5.
 Allen MP, Tildesley DJ. Computer simulation of liquids. Oxford: Oxford
University Press; 1991.
 Impey RW, Madden PA, McDonald IR. Hydration and mobility of ions in
solution. J Chem Phys 1983;87:5071–83.
 García AE, Stiller L. Computation of the mean residence time of water in the
hydration shells of biomolecules. J Comput Chem 1993;14:1396–406.
 Zhang Z, Fenter P, Cheng L, Sturchio NC, Bedzyk MJ, Pr ˇedota M, et al. Ion
adsorption at the rutile–water interface: linking molecular and macroscopic
properties. Langmuir 2004;20:4954–69.
 Obst S, Bradaczek H. Molecular dynamics study of the structure and dynamics
of the hydration shell of alkaline and alkaline-earth metal cations. J Phys Chem
 Soper AK, Phillips MG. A new determination of the structure of water at 25 ?C.
Chem Phys 1986;107:47–60.
 Narten AH, Levy HA. Liquid water: scattering of X-rays in water: a
comprehensive treatise. New York: Plenum Press; 1972.
 Gaballa GA, Neilson GW. The effect of pressure on the structure of light and
heavy water. Mol Phys 1983;50:97–111.
 Bellissent-Funel MC, Teixera J. Dynamics of water studied by coherent and
incoherent inelastic neutron scattering. J Mol Struct 1991;250:213–30.
 Bopp P. The physics and chemistry of aqueous ionic solutions. Amsterdam: D.
 Olovsonn I, Jonsonn PJ. The hydrogen bond. Amsterdam: North Holland; 1976.
 Krüger P, Straßburger W, Wollmer A, van Gunsteren WF. A comparison of the
structure and dynamics of avian pancreatic polypeptide hormone in solution
and in the crystal. Eur Biophys J 1985;13:77–88.
 Jorgensen WL, Chandrasekhar J, Madura JD. Comparison of simple potential
functions for simulating liquid water. J Chem Phys 1983;79:926–35.
 Sciortino F, Fornili SL. Hydrogen bond cooperativity in simulated water: time
dependence analysis of pair interactions. J Chem Phys 1989;90:2786–92.
 Hayashi T, Sano KI, Shiba K, Kumashiro Y, Iwahori K, Yamashita I, et al.
Mechanism underlying specificity of proteins targeting inorganic materials.
Nano Lett 2006;6(3):515–9.
 Chen HB, Su XD, Neoh KG, Choe WS. QCM-D analysis of binding mechanism of
phage particles displaying a constrained heptapeptide with specific affinity to
SiO2and TiO2. Anal Chem 2006;78(14):4872–9.
 Chen HB, Su XD, Neoh KG, Choe WS. Probing the interaction between peptides
and metal oxides using point mutants of a TiO2-binding peptide. Langmuir
 Gronewold TMA, Baumgartner A, Weckmann A, Knekties J, Egler C. Selection
process generating peptide aptamers and analysis of their binding to the TiO2
surface of a surface acoustic wave sensor. Acta Biomater 2009;5:794–800.
 Norde W. Adsorption of proteins from solution at the solid–liquid interface.
Adv Colloid Interface Sci 1986;25:267–340.
 Yang Q, Zhang YY, Liu ML, Ye M, Zhang YQ, Yao SZ. Study of fibrinogen
piezoelectric quartz crystal impedance and FTIR-ATR spectroscopy. Anal
Chim Acta 2007;597:58–66.
 Chen YL, Zhang XF, Gong YD, Zhao NM, Zeng TY, Song XQ. Conformational
changes of fibrinogen adsorption onto hydroxyapatite and titanium oxide
nanoparticles. J Colloid Interface Sci 1999;214:38–45.
 Patil S, Sandberg A, Heckert E, Self W, Seal S. Protein adsorption and cellular
uptake of cerium oxide nanoparticles as a function of zeta potential.
 Theodora S. Tsapikouni, Yannis F. Missirlis. Protein-material interactions: from
micro-to-nano scale. Mat Sci Eng B Solid 2008;152:2–7.
 Roth CM, Lenhoff AM. Electrostatic and van der Waals contributions to protein
adsorption: comparison of theory and experiment. Langmuir 1995;11:3500–9.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694
 Nakamura K, Matsumoto K. Properties of protein adsorption onto pore surface
during microfiltration: effects of solution environment and membrane
hydrophobicity. J Membr Sci 2006;280:363–74.
 Matsumura H, Saburi M. Protein adsorption and wetting of the protein
adsorbed surfaces studied by a new type of laser reflectometer. Colloids Surf B
 Hlady V, Buijs J. Protein adsorption on solid surfaces. Curr Opin Biotech
 Valenti LE, Fiorito PA, García CD, Giacomelli CE. The adsorption–desorption
process of bovine serum albumin on carbon nanotubes. J Colloid Interface Sci
 Williams DF. Biomaterials and tissue engineering in reconstructive surgery.
SA ¯DHANA ¯ 2003;28:563–74.
 To BCS, Lenhoff AM. Hydrophobic interaction chromatography of proteins I.
The effects of protein and adsorbent properties on retention and recovery. J
Chromatogr A 2007;1141:191–205.
 Shen JW, Wu T, Wang Q, Pan HH. Molecular simulation of protein adsorption
and desorption on hydroxyapatite surfaces. Biomaterials 2008;29:513–32.
 Levchenko AA, Li G, Boerio-Goates J, Woodfield BF, Navrotsky A. TiO2stability
landscape: polymorphism, surface energy, and bound water energetics. Chem
 Goclon J, Grybos R, Witko MG, Hafner J. Oxygen vacancy formation on clean
and hydroxylated low-index V2O5surfaces: a density functional investigation.
Phys Rev B 2009;79(7):075439.
 Hess B, van der Vegt NFA. Hydration thermodynamic properties of amino acid
analogues: a systematic comparison of biomolecular force fields and water
models. J Phys Chem B 2006;110:17616–26.
 Wagner W, Pruß A. The IAPWS formulation 1995 for the thermodynamic
properties of ordinary water substance for general and scientific use. J Phys
Chem Ref Data 2002;31:2.
 StöckelmannE,Hentschke R.A molecular-dynamics simulation study ofwater on
NaCl(1 0 0) using a polarizable water model. J Chem Phys 1999;110:12097–107.
D.-P. Song et al./Acta Biomaterialia 6 (2010) 684–694