Rapid #: -10580189

CROSS REF ID: 404693

LENDER: FUG :: Electronic Journals

BORROWER: LYU :: Main Library

TYPE: Article CC:CCG

JOURNAL TITLE: Biomedical materials

USER JOURNAL TITLE: Biomedical materials

ARTICLE TITLE: Computer simulation of biomolecule–biomaterial interactions at surfaces and interfaces

ARTICLE AUTHOR: Wang, Qun

VOLUME: 10

ISSUE: 3

MONTH:

YEAR: 2015

PAGES: 032001-

ISSN: 1748-6041

OCLC #:

Processed by RapidX: 5/2/2016 6:10:54 AM

This material may be protected by copyright law (Title 17 U.S. Code)

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 128.227.193.61

This content was downloaded on 02/05/2016 at 13:07

Please note that terms and conditions apply.

Computer simulation of biomolecule–biomaterial interactions at surfaces and interfaces

View the table of contents for this issue, or go to the journal homepage for more

2015 Biomed. Mater. 10 032001

(http://iopscience.iop.org/1748-605X/10/3/032001)

Home Search Collections Journals About Contact us My IOPscience

© 2015 IOP Publishing Ltd

1. Introduction

Biomaterials are generally defined as artificial or natural materials used to replace, and repair body tissues and organs, which can perform, enhance or replace certain lost function [1]. After implantation into human bodies, an abundance of proteins adsorb on the surfaces of biomaterials instantaneously, which leads to a series of subsequent effects, such as complement activation, platelet activation, coagulation activities, and adherence of cells and bacteria [2, 3]. Therefore, understanding the interaction mechanism between biomaterial surfaces and biomolecules, such as proteins/peptides/amino acids, is crucial to the successful usage of biomaterials in clinics.

Recently, a large number of experimental researches have been carried out, such as protein interaction with HA and TiO2 [4, 5]. Although those experimental studies provide valuable results, the mechanism of the biomolecules interaction with biomaterials is still far from being fully understood. Most of the interac-tion happens at atomic and molecular levels, which

is quite difficult to be investigated by conventional experimental techniques. The complicated biological environments after biomaterials are implanted is a big obstacle for monitoring biomolecules’ interaction with biomaterial surfaces in situ [6, 7].

Computer simulation is an effective way to study the interactions between biomolecules and biomaterial surfaces, because it can provide information about the interaction at the atomic level that cannot be obtained directly from experiments [8]. Quantum mechanics (QM) and molecular dynamics (MD) are two widely used molecular modeling methods. QM mainly solves the Schrodinger equation. The higher the number of the electrons, the more difficult it is to solve the Schrodinger equation [9]. Utilizing a wide variety of approximations during solving the Schrodinger equation develops the Hartree–Fock equation [10], which is an effective and successful method of solving the Schrodinger equa-tion. Density functional theory (DFT) takes the total energy of the system as a function of electron density [11], which is not necessary to solve the complex Schro-dinger equation, and therefore DFT is considered to be

Q Wang et al

Printed in the UK

032001

bmm

© 2015 IOP Publishing Ltd

2015

10

biomed. mater.

bmm

1748-6041

10.1088/1748-6041/10/3/032001

3

00

00

biomedical materials

JW

11

June

2015

Computer simulation of biomolecule–biomaterial interactions at surfaces and interfaces

Qun Wang1,2, Meng-hao Wang1, Ke-feng Wang3, Yaling Liu4, Hong-ping Zhang5, Xiong Lu1 and Xing-dong Zhang3

1 Key Lab of Advanced Technologies of Materials, Ministry of Education, School of Materials Science and Engineering, Southwest Jiaotong University, Chengdu 610031, Sichuan, People’s Republic of China

2 College of Life Science and Technology, MianYang Normal University, Mianyang 621000, Sichuan, People’s Republic of China3 National Engineering Research Center for Biomaterials, Sichuan University, 610065 Chengdu, Sichuan, People’s Republic of China4 Department of Mechanical Engineering & Mechanics, Bioengineering Program, Lehigh University, USA5 Engineering Research Center of Biomass Materials, Ministry of Education, School of Materials Science and Engineering, Southwest

University of Science and Technology, Mianyang, 621010, Sichuan, People’s Republic of China

E-mail: luxiong_2004@163.com, carlcili@163.com, wangmenghao@gmail.com, fencal@163.com, yal310@lehigh.edu, zhp1006@126.com and zhangxd@scu.edu.cn

Keywords: biomaterials, biomolecules, computer simulation, hydroxyapatite, titanium oxide, graphene and graphene oxide

AbstractBiomaterial surfaces and interfaces are intrinsically complicated systems because they involve biomolecules, implanted biomaterials, and complex biological environments. It is difficult to understand the interaction mechanism between biomaterials and biomolecules through conventional experimental methods. Computer simulation is an effective way to study the interaction mechanism at the atomic and molecular levels. In this review, we summarized the recent studies on the interaction behaviors of biomolecules with three types of the most widely used biomaterials: hydroxyapatite (HA), titanium oxide (TiO2), and graphene(G)/graphene oxide(GO). The effects of crystal forms, crystallographic planes, surface defects, doping atoms, and water environments on biomolecules adsorption are discussed in detail. This review provides valuable theoretical guidance for biomaterial designing and surface modification.

Topical Review

Received 16 December 2014

Revised

accepTed foR publicaTion

26 January 2015

published 11 June 2015

doi:10.1088/1748-6041/10/3/032001Biomed. Mater. 10 (2015) 032001

2

Q Wang et al

more efficient than the Hartree–Fork equation. Due to the difficulty of solving the Schrodinger equation, QM generally only handles the models with ~100 atoms. QM is particularly suitable to investigate the physical properties based on the electron density of biomate-rials through investigating band structure, density of state, and electronic structures. MD is based on clas-sical Newton mechanics, which enables us to observe the whole time-dependent interaction process [12]. A meaningful MD simulation is highly dependent on the proper force field parameters, which are generally obtained through QM calculations and experimental results. MD simulation can investigate large models (~106 atoms) and therefore is suitable for studying the interaction between biomaterials and proteins or other large biomolecules.

The Materials Genome Initiative (MGI) project that was announced by American president Obama in 2011 emphasizes the importance of computer simulation in material development [13]. This project aims to greatly reduce the cost and the time cycle of new material devel-opment by integrating the high-throughput computa-tional screening and experimental data with established

material databases containing comprehensive mate-rial properties. The computational research could be applied to accelerate the development of biomaterials, because they generally have long experimental periods and high cost from laboratory to clinic applications. A large amount of high quality computational work on biomaterials has been accomplished during the preceding years, which provides considerable theoreti-cal guidance for the design of new biomaterials [8]. In this review, we provide an overview of the computer simulation of biomolecules interaction with three most popular biomaterials, including HA, TiO2, and G/GO,

as shown in tables 1–3.

2. Interaction of HA and biomolecules

Figuring out the interaction mechanisms of biomolecules with HA is important for further understanding biomineralizations, bone formation, and growth. In the following sections, we review the interaction mechanism of amino acids, peptides, proteins, and other bio-organic molecules with HA surfaces.

Table 1. The interaction of HA and biomolecules.

Biomaterials Biomolecules

Simulation

Methods References

HA (0 1 0) Amino

acids

zwitterionic Gly DFT Jimenez-Izal et al 2012 [22]

HA (1 0 0) zwitterionic Lys DFT Lou et al 2012 [23]

HA (0 0 1)and (0 1 0) zwitterionic Gly,Pro and Hyp DFT Barrios et al 2009 [24]

HA (1 0 0) and (0 0 1) Gly and Glu MD and steered

MD simulation

Pan et al 2007 [26]

HA (0 0 1) and (0 1 0) Peptides Pro-Hyp-Gly, Hyp-Pro-Gly, Pro-

Lys-Gly and Pro-Hyl-Gly

DFT and MD Almora-Barrios et al

2010 [28, 29]

HA (0 0 1) and step RGD and YIGSR MD Biswas et al 2013 [30]

HA (0 0 1) Proteins Bone morphogenetic protein-2

(BMP-2)

MD and QC Dong et al 2007 [34]

HA BMP-7 MD Zhou et al 2007 [36]

HA The 7–10th type III modules of

Fibronectin (FN-III7–10)

A molecular

docking

approach

Guo et al 2013 [39]

HA (1 0 0) Osteopontin (OPN) MD Lai et al 2014 [41]

Si-doped HA (1 0 0) Leucine-rich amelogenin protein

(LRAP)

MD, steered

MD and DFT

Chen et al 2008 [42]

HA (0 0 1) LRAP MD, steered

MD and DFT

Chen et al 2007 [43]

Ca-deficient HA FN-III7–10 and FN-III10 PTMC and MD Liao et al 2014 [40]

OCP Human serum albumin (HSA)

and lysozyme (LSZ)

MD Wang et al 2012 [52]

HA (0 0 1), (1 0 0) and

(1 1 0)

Polymers PE,PA,PLA MD Zhang et al 2009 [48]

HA(0 0 1) and (1 1 0) Chitosan MD Zhang et al 2008 [49]

HA (1 0 0) and (−1 0 0) PAAc MD Bhowmik et al 2007 [50]

HA (1 0 0) and (0 0 1) Other

oganic

molecules

Urea DFT Lu et al 2011 [27]

HA (1 0 0),(1 1 1) and

(0 0 1)

Citric acid Computer

simulations

Filgueiras et al 2006 [45]

HA (0 0 1) and (0 1 0) Citric acid MD de Leeuw et al 2007 [46]

HA (0 0 1) and (0 1 0) GalNAc,GlcA DFT Streeter et al 2011 [47]

Biomed. Mater. 10 (2015) 032001

3

Q Wang et al

2.1. The crystal structure of HAHA is the major mineral component of bone and teeth enamel and has excellent biocompatibility [14]. It has been widely applied in various fields, such as bone replacement biomaterials [15], drug carriers [16], and dental clinics [17]. HA is commonly considered to have a hexagonal crystal structure that belongs to the P63/m space group, as shown in figure 1(a). The lattice parameters of HA are a = b = 9.43 Å, c = 6.88 Å [18]. One unit cell contains ten Ca atoms, six phosphates (PO4), and two hydroxyl (OH) groups [19].There are two types of Ca ions inside the cell (labeled as Ca(I) and Ca(II)), (figures 1(b) and (c)). Four Ca(I) are surrounded by nine O atoms of six PO4 groups [three O(1), three O(2), and three O(3)]; six Ca(II) are coordinated by seven O atoms of five PO4 groups, and one O from OH (one O(1), one O(2), four O(3) and one O(4) of OH) [20, 21].

2.2. The interaction of HA and amino acidsThe principle interactions between amino acids and HA are electrostatic force and hydrogen bonds (H-bonds). The strong interaction frequently occurs between zwitterionic amino acids and HA due to the strong Ca–O interactions. Jimenez-Izal et al [22] investigated the adsorption of glycine (Gly) on the HA (0 1 0) surfaces using DFT at the B3LYP level. The results indicate that Gly is prone to adsorb as a zwitterion on the Ca-rich and P-rich nonstoichiometric HA (0 1 0) surfaces due to the electrostatic interaction between the –COO− and – +NH3 of Gly and the Ca2+ and −PO4

3 of HA. When Gly is adsorbed in a neutral form, the proton of the COOH group is shifted to the HA (0 1 0) surface and two O atoms of –COO− interacts with two different Ca2+ or the same Ca2+ to form the two Ca–O or O–Ca–O bidentate interactions. Lou et al [23] obtained similar results on the interaction of lysine (Lys) with HA (1 0 0)

Table 2. The interaction of TiO2 and biomolecules.

Biomaterials BiomoleculesSimulation

MethodsReferences

Perfect TiO2 Rutile(1 1 0) Amino

acids

Asp DFT Guo et al

2011 [56]

Anatase (1 0 1) Gly DFT Szieberth et

al 2010 [57]

Rutile (1 1 0) Cys DFT Muir et al

2013 [58]

Rutile (1 1 0) and anatase

(1 0 1)

Gly DFT Wang et al

2014 [54]

Rutile (1 1 0) Peptides Ala-Glu and Ala-Lys ab initio and

classical MD

Carravetta et

al 2006 [62]

Anatase (0 0 1) and rutile

(0 1 0)

RGD MD Zhang et al

2008 [63]

Rutile (1 1 0) RGD DFT and ab

initio MD

Muir et al

2014 [64]

Rutile (1 1 0) Other

oganic

molecules

four DNA bases

(adenine, guanine,

thymine, and

cytosine)

MD Monti et al

2011 [68]

Doped TiO2 Ca and N doped

rutile(1 1 0)

Amino acid Asp DFT Guo et al

2011 [56]

N-doped anatase (1 0 1) Small gas

molecules

CO DFT Liu et al

2013 [59]

Defective TiO2 Perfect rutile (1 1 0), and

surfaces with plane-O and

bridging-O deficiencies

Amino acid Arg DFT Zhang et al

2014 [60]

Perfect rutile (1 1 0) and

anatase (1 0 1) surfaces and

surfaces with step edge

defects

Peptides RGD MD Zhang et al

2011 [65]

Perfect rutile (1 1 0) and

three different types of

defects (O vacancies, steps,

and grooves)

Protein FN-III10 MD Wu et al

2010 [67]

Hydroxylated TiO2 Hydroxylated and non-

hydroxylated rutile (1 1 0)

Protein HSA MD Kang et al

2010 [5]

TiO2 in water

environments

Rutile (1 1 0) in water

environments

Amino acid Asp DFT Guo et al

2011 [56]

Rutile (0 0 1) surfaces in

water solution

Protein BMP-2 MD and

steered MD

Utesch et al

2011 [66]

Biomed. Mater. 10 (2015) 032001

4

Q Wang et al

surfaces by the DFT methods. They found that the O of carbonyl, H of amine, and H of methylene are active sites in Lys that can interact with the Ca2+ and O atoms of −PO4

3 on the HA (1 0 0) surface. The interaction is predominantly through the Ca–O ionic bonding, H-bonds and/or van der Waals force. The zwitterionic Lys is more stable on the HA (1 0 0) surface than its neutral form due to the two exposed carboxyl-O atoms interacting with the Ca2+. Likewise, Almora-Barrios et al [24] revealed that there are the strong interactions of the zwitterionic Gly, proline (Pro), and hydroxypro (Hyp) with HA (0 0 1) and (0 1 0) surfaces because of the H-bonds between H atoms of the amino groups and the O atoms of −PO4

3 groups, and electrostatic interactions between O atoms of carboxyl and Ca2+. However, these amino acids have larger adsorption energy on the (0 1 0) surface than on the (0 0 1) surface due to the

large number of undercoordinated surface species with dangling bonds on the (0 1 0) surface. Another reason is that the proton of the amino acids transfers to the HA surface, which creates strong ionic bonding between the carboxylic group and surface Ca2+.

Amino acids are also reported as effective regulators in the process of HA nucleation and growth through experiment and simulation methods [25, 26]. Pan et al [26] investigated the detailed adsorption mechanisms of Gly and glutamic (Glu) acids on HA (1 0 0) and (0 0 1) surfaces by MD and steered MD simulation. They ana-lyzed the departure forces, free energies, and surface cov-erage and discovered that negatively charged Glu is pre-ferred to adsorb strongly onto the HA (0 0 1) rather than HA (1 0 0), therefore the crystal growth along the a and b directions are more favorable than along the c direction, which leads to the formation of plate-like HA. However,

Table 3. The interaction of G/GO and biomolecules.

Biomaterials BiomoleculesSimulation

methodsReferences

G Amino acids Phe, His,Tyr and Try DFT and

MP2

Rajesh et al 2009 [78]

G and graphane Gly, Pro and Hyp DFT Cazorla 2010 [77]

G Peptides RGD DFT Guo et al 2013 [76]

G HSSYWYAFNNKT MD Akdim et al 2013 [80]

G EPLQLKM and HSSYWYAFNNKT MD Kim et al 2011 [81]

G Other

oganic

molecules

Pyrene–polyethylene glycol MD Xu et al 2014 [79]

G Bbisphenol A Zaib et al 2012 [82]

G Dioxin DFT and

MD

Kang 2005 [94]

A single vacancy G Peptides RGD DFT Guo et al 2013 [76]

Stone–Wales defects

and vacancies G

functional

groups

− COOH MD Al-Aqtash et al 2009 [83]

The single vacancy

and Stone–Wales

defective G

− OH DFT Ghaderi et al 2010 [86]

The single and

Stone–Wales defec-

tive G

H, F, and phenyl groups DFT Denis et al 2013 [87]

Au, Cr and Ni-

doped G

Amino acids Nitrated Tyr DFT Ding et al 2012 [88]

Ca-doped G Gly, Pro and Hyp DFT Cazorla 2010 [77]

Au-doped G Cys DFT Zhang et al 2011 [89]

pure, Ti-doped, or

N-doped G

Small gas

molecules

HCHO,CO, NO and SO2 DFT Zhang et al 2013 [90]

B, N, Si, Al, Cr, Mn,

and Au doped G

HCHO DFT Liu et al 2014 [91]

GO Amino acids Gly DFT Wang et al 2014 [54]

GO Peptides RGD DFT Guo et al 2013 [76]

Reduced GO Proteins A helical cytoplasmic protein MD Baweja et al 2013 [97]

GO Other

oganic

molecules

Bisphenol A DFT and

MD

Cortés-Arriagada et al

2013 [96]

GO DNA/RNA bases (adenine, guanine,

cytosine, thymine and uracil) and aromatic

amino acids (His, Phe, Try, and Tyr)

DFT Vovusha et al 2013 [95]

GO A single-stranded DNA MD He et al 2010 [98]

GO Camptothecin (CPT) DFT Saikia et al 2013 [99]

Biomed. Mater. 10 (2015) 032001

5

Q Wang et al

the neutral Gly is not inclined to be adsorbed towards the two HA surfaces. Thus, HA still forms rod-like crystal-lites along the c direction when Gly exists. These findings show that amino acids may regulate the growth of HA along certain crystal planes, like other small molecules, such as urea. Lu et al [27] indicated that the urea can control HA growth and regulate its morphology to form well-crystalline hexagonal HA. They discovered that urea can more severely suppress the growth of HA (1 0 0) surfaces than (0 0 1) surfaces due to the higher binding energy between the urea and HA (1 0 0) surfaces.

2.3. The interaction of HA and peptidesA peptide is a compound of α-amino acids that are linked together by peptide chains. They are generally referred to as oligopeptides and polypeptide when they consist of 2–10 amino acids and 10–50 amino acids, respectively. The interactions of peptides and HA are also mainly electrostatic interactions and H-bonds, which is in perfect line with the interaction of the individual amino acid and the HA surface. Almora-Barrios et al [28] also employed DFT to investigate the interaction of the HA and Pro-Hyp-Gly, Hyp-Pro-Gly, Pro-Lys-Gly, and Pro-hylysine (Hyl)-Gly. They found that O atoms of the terminal carboxyl and H atoms of amino groups interact with Ca2+ and −PO4

3 , forming Ca–O and H-bonds, and the Ca–O interaction between peptides with the HA surface are capital. The adsorption energies of Pro-Lys-Gly and Pro-Hyl-Gly

on HA surfaces are larger than that of Hyp-Pro-Gly and Pro-Hyp-Gly on HA surfaces because the positively charged – +NH3 of the Lys/Hyl and –OH of Hyl can bind to −PO4

3 and Ca2+, respectively.Different HA crystallographic planes have different

binding ability with peptides. Almora-Barrios et al [28, 29] studied the interaction of Pro-Hyp-Gly, Hyp-Pro-Gly, Pro-Lys-Gly, and Pro-Hyl-Gly peptides on the HA (0 0 1) and (0 1 0) surfaces by DFT and MD, and found that the stronger interaction occurs on the HA (0 1 0) surface rather than on the HA (0 0 1) surface, which is because protons easily transfer from the peptides to the reactive (0 1 0) surface.

Furthermore, the orientations of a peptide may affect the adsorption strength on the same crystal plane. Biswas et al [30] investigated the adsorption of argnine (Arg)-Gly-aspartic (Asp) (RGD) and tyrosine (Tyr)-isoleucine (Ile)-Gly-serine (Ser)-Arg (YIGSR) peptides on HA surfaces by MD methods. The results demon-strated that the adsorption strength is influenced by the initial peptide orientation. RGD has a large adsorption energy when it is at the step-edge of the HA surface, whereas YIGSR is the most stable when it is perpendicu-lar to the HA (0 0 1) surface.

2.4. The interaction of HA and proteinsThe interaction between HA and proteins plays an important part during tissue regeneration [31]. When HA is implanted into a body, proteins from the

Figure 1. (a) The 3D crystal structure of HA unit cell; (b) the Ca(I) and (c) Ca(II) sites of HA. Color codes: calcium, green; phosphorus, pink; oxygen, red; hydrogen, white.

Biomed. Mater. 10 (2015) 032001

6

Q Wang et al

surrounding body fluids rapidly adsorb onto their surfaces, and affect the subsequent cellular attachment, proliferation, and migration [32]. However, the interaction mechanism is not well understood due to the complex structure and configurations of proteins and various surface features of substrates. Therefore, only a few type of proteins’ interaction with HA have been studied by computer simulation as summarized in the subsequent sections.

Bone morphogenetic proteins (BMPs) are multi-functional growth factors that belong to the trans-forming growth factor beta superfamily. BMPs have the unique ectopic osteoinductive activity and play vital roles in the formation of bone [33]. Different BMP fam-ily members have different osteoinductive activity, in which BMP-2 and BMP-7 are the most widely studied. Dong et al [34] investigated the interaction BMP-2 on an HA (0 0 1) surface using MD and quantum chem-istry (QC) calculations. The results indicate that the –OH, –NH2, and –COO− of BMP-2 are the main groups that interact with HA. The electrostatic interaction exists between –COO− and the Ca2+. The water-bridged H-bonds also exist between BMP-2-H2O and HA. They also studied the effects of the orientation of BMP-2 on an HA (0 0 1) surface with six different orientations. The results showed that the strongest interaction of BMP-2 with the HA (0 0 1) surface is formed when the number of the adsorbed residues is the most [35]. Zhou et al [36] investigated the site-selective adhesion and the adsorption mechanism of BMP-7 on the HA surfaces in aqueous solution by steered MD simulations. They sug-gest that the adsorption sites are dependent on the func-tional groups of BMP-7, and therefore could be divided into two categories: COO− and NH2/ +NH3 . For COO−, the adsorption is induced by the electrostatic interac-tion between the negatively charged COO− and the pos-itively charged Ca2+ on the HA surface. For NH2/ +NH3 , the interaction happens between the NH2/ +NH3 and the

−PO43 on the HA surface by forming the intermolecular

H-bonds.Fibronectin (FN) is a major protein in extracellular

matrices (ECM) that involves in cellular adhesion and migration in a diverse range of physiological processes. Each monomer of FN is composed of types I, II, and III repeating modules, which are organized into func-tional domains [37]. FN intermediates its biological function through binding to the integrin, a transmem-brane heterodimer [38]. Guo et al [39] investigated the interaction of the 7–10th type III modules of FN (FN-III7–10) with HA by a molecular docking approach, and revealed that the FN-III10 is the most important module among FN-III7–10 in promoting FN binding to HA due to the RGD sequence. The Arg residues of RGD directly interact with the HA through electrostatic forces and H-bonds. Liao et al [40] employed parallel tempering Monte Carlo (PTMC) and MD methods to analyze the orientation and adsorption mechanism of the FN-III7–10 and FN-III10 on Ca-deficient HA surfaces. They indicated the adsorption of FN-III10 changed from

electrostatic interactions at the pre-adsorption stage to VDW interaction at the post-adsorption stage due to the mismatching of the charged groups of FN-III10 on the differently charged HA surface. However, the interaction of FN-III7–10 altered from the initial VDW interaction to the strong electrostatic interactions and H-bonds when the basic residue Arg enters into the Ca (I) vacancy of HA surfaces.

Osteopontin (OPN) consisting of many acidic amino acid residues is present in the extrafibrillar pro-tein matrix; it is non-collagenous protein and is natu-ral adhesive glue in bone [4]. Lai et al [41] using the MD method investigated the interfacial mechanical behavior between OPN and the HA (1 0 0) surface, and found the interfacial mechanical behavior is mainly dependent on the Ca–O interaction of acidic amino acid residues in OPN and Ca in HA. They discovered when pulling along the interface direction, some new bonds appeared between the acidic residues and HA surfaces, leading to a stick–slip shift of OPN on the HA surface. Therefore, they suggest that forming new bonds during loading is regarded as a capital mecha-nism for high fracture resistance in bone.

The surface texture and composition of biomate-rial surfaces can regulate the interaction of proteins with biomaterials. Chen et al [42] investigated the adsorption and desorption behaviors of leucine-rich amelogenin protein (LRAP), which is the major ingredient of the formation of the tooth enamel in promoting the HA growth, on a series of SiHA (1 0 0) surfaces (0%, 25%, 50%, 75%, 100%) using MD, SMD, and DFT methods. they discovered that the Si-doped into the HA (1 0 0) surface increases the sur-face negative charge due to inducing the −SiO4

4 , which leads to the – +NH3 , –NH2, and –OH moving towards the surface, and the negatively charged –COO− being rejected from the surface. Therefore, the interac-tion of −PO4

3 and Ca2+ are getting weaker gradually due to the effects of charge rejection and steric hin-drance large size −SiO4

4 , while the H-bonds are get-ting stronger. They also found that large content of Si substitution could greatly decrease the interaction, which suggests that the main interactions are the Ca–O electrostatic interactions between LRAP and SiHA, which is in line with their previous study [43]. However, they demonstrated that a small content of Si substitution in SiHA did not have the obvious hin-drance effects. Thereby, the proper content of the Si substitution can increase the bioactivity of SiHA, the redundant Si substitution would decrease the osteo-conductivity of the biomaterials. These results could provide a useful guidance in regulating the adsorp-tion efficiency of SiHA via changing the Si substitu-tion content.

2.5. The interaction of HA and other organic moleculesThe citric acid could be a good growth inhibitor for HA [44]. Filgueiras et al [45] studied the adsorption of

Biomed. Mater. 10 (2015) 032001

7

Q Wang et al

the citric acid molecule on HA surfaces, and revealed that the interactions are principally Ca–O interactions through O atoms of citric acid and Ca2+, and H-bonds between H atoms of OH and –CH2. The adsorption energy analysis indicated that the citric acid prefers to adsorb on the HA (1 0 0) and (1 1 1) surfaces rather than HA (0 0 1) surface. Thus, citric acid could inhibit growth of the former surfaces than the latter one because the strong interaction will inhibit the HA surface to further growth. The result implies that HA crystals would grow more rapidly along (0 0 1) direction when citric acid exists, leading to elongation in the c-direction. A similar HA crystal growth habit is reported by de Leeuw et al [46] when they investigated the adsorption of citric acid on the HA (0 0 1) and (0 1 0) surfaces in an aqueous environment by MD. They found citric acid adsorbs strongly on the HA (0 1 0) surface but not the HA (0 0 1) surface.

Saccharides N-acetylgalactosamine (GalNAc) and glucuronic acid (GlcA) are the monosaccharide components of chondroitin, which is a glycosami-noglycans (GAGs) found in bone and cartilage, and has been regarded as a regulating factor in the pro-cess of biomineralization. Streeter et al [47] studied the interactions of the GalNAc and GlcA on the HA (0 0 1) and (0 1 0) surfaces by DFT. The results showed that GalNAc interacts with HA mainly through its OH and acetyl amine functional groups, and deprotonated GlcA interacts with HA principally through its OH and carboxyl groups. The structural conformation of chondroitin chains and the strength of adsorption depend on the orientation of the saccharide on the HA surface. Both monosaccharides have stronger adsorp-tion on the (0 1 0) surface than on the (0 0 1) surface. By preferentially adsorbing on the (0 1 0) surface, GAGs could stabilize that surface and inhibit further crystal growth along the a- and b-axes, and lead to HA crystal growth along the c-axis.

2.6. The interaction of HA and polymersHA-Polymer composites are often widely studied as potential bone replacement materials because the composites exhibit both excellent bioactivity and enough mechanical strength. Due to the adjustability of HA content in the composite, the composite can be tailored to meet the ductility and toughness requirements for bone replacements. The interfacial interactions between the HA reinforcement and polymer matrix play an important role in improving the mechanical properties of composite systems. Zhang et al [48] investigated the interfacial interactions between HA and polyethylene (PE), polyamide (PA), and polylactic acid (PLA) by the MD method. The interactions of polymers on HA crystallographic planes (0 0 1), (1 0 0), and (1 1 0) were calculated and the results indicated that the HA (1 1 0) surface holds the highest binding energy with these polymers because HA (1 1 0) has a higher planar atom density than the HA (0 0 1) and (1 0 0)

surfaces. The binding energies of PA/HA and PLA/HA are much higher than that of PE/HA, which might be ascribed to large number of polar groups in the PA and PLA chains. The effects of a silane coupling agent (A174) on interfacial binding energies were also examined. The A174 increases the binding energy between PE and HA, but not for the PA/HA and PLA/HA systems. Meanwhile, Zhang et al [49] also investigated the interaction of chitosan with different HA crystallographic planes. It was found that the interaction of the chitosan chain on the HA (1 0 0) surface is stronger than that on the HA (0 0 1) and (1 1 0) surfaces and there are chemical interactions between N and Ca atoms and H-bonds between O atoms and OH groups. Bhowmik et al [50] evaluated the interfaces of HA-polyacrylic acid (PAAc) composites by the MD simulation. They found that the interaction is mainly a Ca–O interaction between the Ca and O atoms of –COOH of PAAC, and H-bonds between the H atoms of –COOH and the O atoms of PO4 and OH of HA. PAAc could favorably adsorb on the HA (1 0 0) or (1 0 0) surface, and therefore the most favorable orientation is the PAAc main chain is parallel to the c-axis of HA. These results of the MD simulation can guide the design of polymer/HA composites.

2.7. The interaction of Octacalcium phosphate (OCP) and biomoleculesOctacalcium phosphate (OCP) is a bioresorbable ceramic and could degrade gradually under a physiological environment. Similar to HA, the degraded products of OCP are calcium and phosphate ions, which are fully biocompatible and non-toxic to the biological system. Furthermore, OCP is not only a major ingredient of Ca–P precipitates in vivo but also is regarded as a precursor phase of new bone generation [51]. Wang et al [52] studied the absorption of acidic human serum albumin (HSA) and basic lysozyme (LSZ) on OCP surfaces, as shown in figure 2. The results showed that absorption energy of basic LSZ is higher than that of acidic HSA, which indicates that LSZ is easier to be absorbed onto OCP surfaces than HSA. This is because LSZ contains more basic residues, whereas HSA contains more acidic residues, and the basic residues have more affinity to the OCP surface than acidic residues. The interaction energies change with the OCP crystallographic planes, and the trend of changes for both proteins are similar, that is, OCP (0 0 1) > OCP (1 1 1) > OCP (1 1 0) > OCP (1 0 0). Note that the stronger interaction between material surfaces and proteins usually causes larger deformation, which is represented by the larger strain energy of the proteins. In Wang’s study, the strain energy strongly depends on the initiation orientations of the proteins rather than the OCP crystallographic planes. These results provide valuable guidance for understanding the mechanism of the osteoinductivity of OCP from an atomic level.

Biomed. Mater. 10 (2015) 032001

8

Q Wang et al

3. Interaction of TiO2/biomolecules

3.1. The characteristics of TiO2 surfaces

TiO2 is commonly formed on the surfaces of titanium

(Ti) and its alloys and is considered to be related

with their excellent corrosion resistance and the

biocompatibility of Ti alloys. The interactions of

biomolecules with Ti-based biomedical devices take

place mainly on the TiO2 surfaces. Several TiO2 surfaces

have been investigated, such as rutile (1 1 0), rutile

(1 0 1), anatase (1 0 1), and anatase (0 0 1) [53]. Among

them, rutile (1 1 0) and anatase (1 0 1) are the most

interested crystallographic planes of TiO2 because they

are thermodynamically stable. Rutile (1 1 0) surfaces

contain five-fold (5f) and six-fold (6f) Ti atoms and two

types of surface oxygen atoms, referred to as bridging

and in-plane O atoms, as indicated in figure 3(a) [54].

The 5f-Ti and bridging-O hang with their unpaired

bonds. Inside the bulk, each Ti atom is surrounded

with 6 O atoms and each O atom is surrounded with 3

Ti atoms. The imparities of the surface O and Ti atoms

offer a more reactive characteristic to the surfaces. On

Figure 2. Snapshots of simulation models (a) OCP (1 0 0) + HSA, (b) OCP (1 0 0) + LSZ after the MD simulations (side views), Proteins presented by ribbon in green [52]. Reproduced from K Wang et al 2012 J. Mater. Sci.: Mater. Med. 23 1045–53, © Springer Science + Business Media, LLC 2012, with permission of Springer Science + Business Media.

Figure 3. The 3D crystal structure of (a) rutitle (1 1 0) and (b) anatase (1 0 1) surfaces. Color codes: oxygen, red; titanium, grey. Reproduced from [54].

Biomed. Mater. 10 (2015) 032001

9

Q Wang et al

anatase (1 0 1) surfaces (figure 3(b)) [54], the bulk 6f-Ti and 3f-O have been substituted by 5f-Ti and 2f-O, respectively [55].

3.2. The interaction of TiO2 and amino acidsThe functional groups of amino acids, such as the amino and carboxyl group, play a key role when amino acids interact with TiO2 surfaces. Guo et al [56] studied Asp adsorption on the rutile (1 1 0) surfaces. It was found that the carbonyl O and the amino N interact with the two Ti atoms and form the most stable mode, so called ‘bridging bidentate mode’, as shown in figure 4. Some researches also found another bridging bidentate mode–the two O atoms in the carboxyl group adsorb on two Ti atoms, which are one of the most stable mode. Szieberth et al [57] studied the interaction of Gly with anatase (1 0 1), and also found that the most stable mode is the bridging bidentate mode. Muir et al [58] investigated cysteine (Cys) adsorption on rutile (1 1 0) surfaces and concluded that the adsorption is stable when bridging bidentate modes appear.

Doping a TiO2 surface with different elements might influence its interaction with amino acids. Guo et al [56] found that the doping of Ca and N are not conducive to the adsorption of Asp. The N-doping weakens the H-bonds between Asp and bridging-O due to the introdution of bridging-O vacancies. The Ca–doping also weakens the interaction of –NH2, –COOH and Ti atoms because both –NH2 and –COOH interact with the doped Ca atoms, while the interaction of Ca-N is weaker than that of Ti-N.

However, Liu et al [59] found that CO has stronger adsorption on the N-doped anatase (1 0 1) surface than on the undoped surface because the doped N site is also a chemical adsorption site besides the dan-gling O atom.

Surface defects always exist on those TiO2 surfaces including plane-O deficiencies, bridging-O defi-ciencies, and surface step edge defects, which might severely influence the physical and chemical proper-ties of material surfaces. Zhang et al [60] employed DFT to investigate the interaction Arg on the perfect rutile (1 1 0) surfaces, and surfaces with plane-O deficiency and bridging-O deficiency. They discov-ered that the adsorption of Arg on the surfaces with plane-O deficiency and bridging-O deficiency are less stable than that on the perfect rutile (1 1 0) surfaces.

A water environment should be considered when amino acids interact with TiO2 surfaces because the biological environment usually contains water [61]. Guo et al [56] investigated the Asp adsorption on the rutile (1 1 0) surfaces in water environments, and found that water weakened the Asp adsorption (figure 5). One possible reason is that water molecules are prone to trapping Asp, and therefore preventing further adsorp-tion of Asp. Another reason is that water molecules cap the Ti adsorption sites and hinder Asp interacts with rutile (1 1 0) surfaces directly.

3.3. The interaction of TiO2 and peptidesPeptide adsorption behavior on TiO2 surfaces is similar to that of amino acids. Carravetta et al [62] performed both ab initio and classical MD calculations to investigate the interactions between rutile (1 1 0) and Ala-Glu and Ala-Lys. They found that the bidentate mode between carboxyl O and two adjacent Ti is more stable than the monodentate mode between carboxyl O and one Ti. However, several researches found that the monodentate mode only appears when peptides adsord on TiO2 surfaces. Wang et al [54] investigate Gly on rutile (1 1 0) and anatase (1 0 1) surfaces, and suggest that the stable adsorption occurs when the O atom of carboxyl group interacts with the 5f-Ti atom of the TiO2 surface (figure 6). Zhang et al [63] used the MD method to study RGD on TiO2 surfaces, and found that RGD binds strongly with anatase (0 0 1) and rutile (0 1 0) surfaces through the main interaction of Ti and O of Asp of RGD. Muir et al [64] investigated the adsorption of RGD on the perfect rutile (1 1 0) surface and obtained similar results. They pointed out that the adsorption only occurs when the Asp end interacts with the surface Ti atoms.

The surface defects also affect peptide behaviors on material surfaces. Zhang et al [65] investigated the adsorption of RGD on the perfect rutile (1 1 0) and anatase (1 0 1) surfaces and surfaces with step edge defects (figure 7). They found that the interactions are weaker when RGD is on the top of surface step defects. However, the interactions between RGD and TiO2 sur-face are strengthened when the RGD falls down the grooves between the step edges.

Figure 4. (a) The molecular structure of Asp. (b) The bridging bidentate mode of Asp interaction with the rutitle (1 1 0). Reproduced from [56].

Biomed. Mater. 10 (2015) 032001

10

Q Wang et al

3.4. The interaction of TiO2 and proteinsProtein interaction with TiO2 surfaces is much more complicated than amino acid and peptides. Utesch et al [66] employed the MD and steered MD (SMD) methods to investigate the adsorption of BMP-2 on the hydrophilic rutile (0 0 1) surfaces in water solution. They

discovered that BMP-2 adsorbs loosely on TiO2 surfaces due to a two-layer water structure preventing the direct interaction of BMP-2 and TiO2. They also found that hydroxylated TiO2 surfaces may have a high affinity to proteins. Kang et al [5] studied HSA adsorption on hydroxylated and non-hydroxylated rutile (1 1 0) surfaces. They concluded that the hydroxylated surface has strong interaction with the protein because the hydroxyl groups diminish the H-bonds between water and the surface, which prevents the formation of a water layer and results in the direct protein-TiO2 contacts.

Introducing defects to TiO2 surfaces can influence the adsorption of proteins on TiO2 surfaces. Wu et al [67] investigated the adsorption behavior of FN-III10 on the perfect rutile (1 1 0) and three different types of defects (O vacancies, steps, and grooves) surfaces by MD meth-ods. They found that the adsorption could be stabilized when two O vacancies were substituted by the carbonyl O atoms of Gly40 and Pro64 in the O vacancy mode, while the secondary structure of the FN-III10 was not broken. In addition, introducing steps and grooves weakens the interatomic interaction of TiO2 atoms and leads to many unstable undercoordinated atoms, which results in the large structural deformation of FN-III10 when it is absorbed onto the step and groove surface. Their results suggest the defects could strengthen the interaction between TiO2 surfaces and the FN-III10 due to the high surface energy. It should also be noted that the too strong interaction induces the severe conformation change of proteins, and might damage the bioactivity of proteins.

Figure 5. The anionic Asp molecule adsorbs on the rutitle (1 1 0) surface with eight water molecules surrounding it. Reproduced from [56].

Figure 6. Gly interacts with TiO2 surfaces. (a) Rutitle (1 1 0)-Gly model; (b) Anatase (1 0 1)-Gly model. Reproduced from [54].

Biomed. Mater. 10 (2015) 032001

11

Q Wang et al

3.5. The interaction of TiO2 and DNA basesApart from the amino acids, peptides, and proteins, the adsorption of DNA onto the TiO2 surfaces are also investigated. Monti et al [68] employed a classical MD simulation to study the adsorption of four DNA bases (adenine, guanine, thymine, and cytosine) on the rutile (1 1 0) surfaces. The base molecules have a tendency to adsorb on the water layers and are prevented from directly contacting the TiO2. The base arrangements were regulated by the solvent, and the base rings were vertical to the rutile (1 1 0) surface to form the most H-bonds and minimized unfavorable interactions.

4. The interaction of biomolecues and G and GO

G is a flat monolayer of carbon atoms that are tightly packed into a 2D honeycomb lattice. As shown in figure 8(a), carbon atoms generate sp2 hybridization that form three σ bonds; non-hybridizated pz orbitals form two π bonds that are parallel to the G plane. The distinctive G structure leads to excellent mechanical, electrical, optical, and thermal properties, and a large specific surface area. GO could be regarded as G decorated with the oxygen-containing groups, such

Figure 7. RGD on TiO2 surfaces with step edges in vacuum after MD simulation with three initial configurations: (a) model I, lying on rutile (1 1 0); (b) model II, standing on N on rutile (1 1 0); (c) model III, standing on O on rutile (1 1 0); (d) model I, lying on anatase (1 0 1); (e) model II, standing on N on anatase (1 0 1); (f) model III, standing on O on anatase (1 0 1). Reproduced from [65].

Figure 8. The structures of (a) G and (b) GO.

Biomed. Mater. 10 (2015) 032001

12

Q Wang et al

as hydroxyl, carbonyl, carboxyl, and epoxy groups, as shown in figure 8(b). The hydroxyl and epoxy functional groups are mainly on the basal plane of GO, whereas the carbonyl and carboxyl groups are mainly at the edge of the GO [71]. GO has good hydrophilicity and also a high affinity with those biomolecules because of the abundant surface functional groups. Both G and GO have good biocompatibility and therefore have broad potential applications in biomedical fields such as tissue engineering [72], drug and gene delivery [73], biologic imaging, and biosensors [74, 75].

4.1. The interaction mechanism between biomolecules and G, defective G and doping GThe interaction between biomolecules with pristine G could be based on the non-covalent interaction of polar functional groups of biomolecules with the π bond of G. We studied the interaction between pristine G and RGD with five configurations [76], and revealed that the strongest adsorption is obtained when RGD is parallel to the basal plane of G surfaces. In this configuration, G interacts with all three functional groups of RGD and form − +NH3 … π,−COO− … π and guanidine …π interactions. The interaction of − +NH3 ··· π is stronger than that of guanidine − NH2··· π and −COO−···π. Similar results are reported by Cazorla [77], where DFT was used to study the interaction between Gly and G. It was reported that the strongest adsorption exists when forming the − +NH3 ···π interaction.

The interactions of biomolecules and G might also be based on π–π interactions if the biomolecules con-tain aromatic ring. Rajesh et al [78] studied the interac-tion of phenylalanine (Phe), histidine (His), tyrosine (Tyr), and tryptophan (Trp) molecules with G and revealed that π–π interactions are formed by the aro-matic rings of the amino acids. Xu et al [79] discovered that the π–π stacking interaction is the main force deter-mining the pyrene–polyethylene glycol adsorption on the G surface. Akdim et al [80] studied the peptide (HSSYWYAFNNKT) adsorption on the G surface and indicated that π–π stacking interactions form during the adsorption process. Kim et al [81] reported that peptides (EPLQLKM and HSSYWYAFNNKT) bond to the planar surface or edge of G through π–π stacking interaction. Zaib et al [82] also found that π–π stacking interaction exists between bisphenol A and G.

The defects in G can influence the chemical proper-ties of G, and also be the possible adsorption sites of the biomolecules [83–85]. There are various defects in G, such as single vacancies, 585 double vacancies, 555 − 777 reconstructed double vacancies, and Stone–Wales defects. Our group studied RGD adsorption on defective G with a single vacancy [76], as shown in fig-ure 9. The adsorption energies of RGD on defective G are found to be larger than that of RGD on pristine G. The − +NH3 and guanidine of RGD could stably inter-act with the vacancy of G because of the dangling C atoms in the vacancy. Al-Aqtash et al [83] revealed that the interaction of the −COOH with the Stone–Wales

defects, and vacancies G are stronger than that with pristine G. In Ghaderi’s et al [86] study, they tested the single vacancy and Stone–Wales defective sites, and dis-covered both defected G become more active for −OH adsorption and Stone–Wales defects are more reac-tive. −OH adsorption on the Stone–Wales defected G plane is more stable than on the single vacancy defected G plane. However, Denis et al [87] suggest that the reactivity of a single vacancy is higher than the Stone–Wales defect. They added one H, F, and phenyl groups to defective G and found the following reactivity: single vacancy > 585 double vacancy > 555 − 777 recon-structed double vacancy > Stone–Wales > perfect G.

Doping G with metal atoms could enhance the adsorption of biomolecules on G. Cazorla [77] found that doping G with Ca atoms dramatically enhances the interactions between collagen amino acids (Gly,

Figure 9. Top view of RGD on defective G surfaces with monovacancy: (a) V- +NH3 -RGD model, +NH3 interacts with the vacancy; (b) V-guanidine-RGD model, guanidine interacts with the vacancy; (c) V-COO−-RGD model, COO− interacts with the vacancy. The vacancies are highlighted in ball-and-stick style. Reproduced from [76].

Biomed. Mater. 10 (2015) 032001

13

Q Wang et al

Pro, Hyp) and G because Ca–O bonds are formed through the electronic charge transfers from Ca to the carboxylic group of the collagen amino acids. Ding et al [88] explored the adsorption of nitrated Tyr on pristine G and Au, Cr, and Ni-doped G by DFT, and found that the binding of nitrated Tyr is enhanced on mental-doped G because of the strong chemisorptions between metal and O atoms of –NO2. Zhang et al [89] also discovered that Cys has stronger adsorption on Au-doped G than on pristine G because of the strong Au–S, Au–N and Au–O interactions. Zhang et al [90] compared the adsorption of HCHO,CO, NO, SO2 on pure, Ti-doped, or N-doped G by DFT, and the results showed that only Ti doping greatly improved the interactions of gas molecules with G because electrons transfer from the Ti atom to the O atoms of the mol-ecules and form strong interactions. Liu et al [91] also employed DFT to investigate the adsorption HCHO on XG (X = B, N, Si, Al, Cr, Mn, and Au), and found that Si-, Au- especially Cr-, Al-, and Mn-doped G can strongly attract HCHO. Furthermore, they found the detection of HCHO with the Al-, and Mn-doped G are the most suitable due to the relatively big adsorption energies and charge transfers, the obvious spin density images, and the changes of density of states before and after the adsorption of HCHO.

Adsorption of biomolecules on G surfaces will break the symmetry of the G lattice, thereby affect the band gap, which could be employed to design G-based chemical sensors [92, 93]. Kang [94] reported that dioxin binding to G leads to the electronic density of states near the Fermi level increase, which suggests that G can be used for dioxin detection. Zhang et al [89] reported that metal-doped G is a good chemical sensor for Cys. They found that the adsorption of Cys on an Au-doped G surface causes larger charge transfer than on a pristine G surface, which induces dramatic changes in the electrical conductance of G. However, Cazorla [77] analyzed the density of states of G and graphene with Gly, Pro, and Hyp adsorption, and found that amino acids binding to graphene causes partial closure of its electronic band gap, whereas their bindings to G do not cause a similar phenomena.

4.2. The interaction mechanism between biomolecules and GOGO is more conducive to adsorbing biomolecules than G because of various O-containing groups on its surface, such as the OH, epoxy, and COOH groups, which could form H-bonds with biomolecules. We studied the interaction between GO and RGD, and revealed that the interaction between RGD and GO is stronger than that between RGD and G [76]. There are many H-bonds formed between the functional groups of RGD and O-containing groups, such as COO− with OH of GO, +NH3 with epoxy-O of GO, and guanidine with OH of GO, as shown in figure 10. Our group [54] also found Gly has a stronger interaction with GO than with G due to forming a H-bond between –OH in GO

and –NH2 in Gly. Vovusha et al [95] indicated that GO has larger binding energies with DNA/RNA bases (adenine, guanine, cytosine, thymine, and uracil), and aromatic amino acids (His, Phe, Trp, and Tyr) than with G because GO forms H-bonds with aromatic amino acids, whereas G only has π–π interactions with them

The π–π interactions appear between GO and aromatic molecules, like that in the case of aromatic molecules on G. Cortés-Arriagada et al [96] found that bisphenol A has the stronger interaction with GO than with G because of π–π coupling interaction. Baweja et al [97] found that the van der Waals and π–π stacking interactions are the major driving forces for the adsorp-tion of a helical cytoplasmic protein on reduced GO and G, and showed that the configurational of the protein on GO surface is stable due to the extensive hydra-tion of GO surface and not the existing π–π stacking interaction between Tyr residues of protein and π regions of GO. He et al [98] also found that the bind-ing between GO and single-stranded DNA has a π–π stacking interaction. Saikia et al [99] investigated the adsorption of anticancer drug camptothecin (CPT) on G, defect G and GO, and demonstrated that π–π stacking is the primary interaction between CPT and G, whereas H-bonds are formed between CPT and GO.

5. Future challenges

In summary, computer simulation has been one of the basic scientific research methodologies, together with theoretical and experimental methods. With the development of science and technology, computer simulation will provide more reliable theoretical guidance for the research in biomaterials and accomplish many great achievements. However, there are still challenges to be confronted in the area.

The first challenge is to accurately and rapidly sim-ulate the interaction between materials and proteins. These reactions play important roles on the cellular response in biological environments because proteins will be absorbed onto the biomaterial surfaces and therefore greatly influence the cell behavior [100]. How-ever, proteins possess four hierarchies that are generally much more complicated than amino acids or peptides, with sizes ranging from about 6000 atoms to a million atoms. It should also consider the water environment that is necessary to maintain the conformation of the protein, and other factors, such as the temperature and ionic strength of the body fluid, which make the simula-tion and analysis of proteins more complex. Thus, the interactions between materials and proteins are not eas-ily simulated by conventional routes. Therefore, only a few protein interactions with biomaterials have been explored by computer simulation, as reviewed in this paper. Nowadays, novel method-coarse-graining (CG) methods [101–105] have been developed to deal with large molecules, which takes the molecule or a part of molecule as a ‘particle’ without an internal structure. CG methods could greatly save the calculation time,

Biomed. Mater. 10 (2015) 032001

14

Q Wang et al

and make the long-time simulation of the larger and more complex system possible. It is necessary to develop the more advanced simulation approaches, which can explore not only the behavior of amino acid residues in proteins, but also the change of second to fourth structures during the interaction, which are helpful to understanding the biological functions of proteins.

Another challenge is the modeling of biomateri-als. Natural biomaterials are not perfect crystals. For example, carbonated HA is often found in human bone, in which carbonates incorporate into HA crystals and substitute for PO4 or OH in the crystal structure and subsequently change its properties. Calcium-deficient HA also exists in the biological systems, in which Ca vacancies appear in the HA crystal. The biological HA might also contain doping ions, such as Na, K. How-ever, there is a lack of experimental data that accurately describe the structures of carbonated, calcium-defi-cient, and ion-doped HA, which precludes further mod-eling of these nonstoichiometric materials. In addition, the surface topology/composition of the biomaterials might affect the configurations of the biomolecules. However, it is also a tough job to model the actual sur-face structures of materials because of the complexity of the surfaces. MGI projects could be a suitable route to face these challenges. By integrating high-throughput

experimental and computational studies, we might be able to obtain a giant amount of data in a relatively short period of time, and simulating complicated biological materials may also be possible.

Acknowledgments

This project was financially supported by 863 Program (2015AA034202), NSFC (51202152), Sichuan Youth Science-Technology Foundation (2011JQ0010), Construction Program for Innovative Research Team of University in Sichuan Province (14TD0050).

References

[1] Bauer S, Schmuki P, von der Mark K and Park J 2013 Engineering biocompatible implant surfaces: part I: materials and surfaces Prog. Mater Sci. 58 261–326

[2] Li M, Neoh K G, Kang E T, Lau T and Chiong E 2014 Surface modification of silicone with covalently immobilized and crosslinked agarose for potential application in the inhibition of infection and omental wrapping Adv. Funct. Mater. 24 1631–43

[3] Liu Y et al 2013 Engineering biomaterial-associated complement activation to improve vaccine efficacy Biomacromolecules 14 3321–8

[4] Azzopardi P V, O’Young J, Lajoie G, Karttunen M, Goldberg H A and Hunter G K 2010 Roles of electrostatics and conformation in protein-crystal interactions PLOS One 5 e9330

Figure 10. (a) Top view: the GO model with three C8O2(OH)2 units on the carbon plane. The dotted lines show the equilateral triange formed by three C8O2(OH)2 units. (b) Side view: a typical C8O2(OH)2 unit after geometric optimization, which contains two para-epoxy (para-O) groups at the opposite side of the carbon plane and two para-OH groups at the upside of the carbon plane. (c) Side view: RGD on GO surfaces with hydrogen bonds between RGD and O-containing groups of GO, as indicated by arrows . Reproduced from [76].

Biomed. Mater. 10 (2015) 032001

15

Q Wang et al

[5] Kang Y, Li X, Tu Y, Wang Q and Ågren H 2010 On the mechanism of protein adsorption onto hydroxylated and nonhydroxylated TiO2 surfaces J. Phys. Chem. C 114 14496–502

[6] Costa D, Garrain PA, Diawara B and Marcus P 2011 Biomolecule–Biomaterial interaction: a DFT-D study of glycine adsorption and self-assembly on hydroxylated Cr2O3 surfaces Langmuir 27 2747–60

[7] Rimola A, Corno M, Garza J and Ugliengo P 2012 Ab initio modelling of protein–biomaterial interactions: influence of amino acid polar side chains on adsorption at hydroxyapatite surfaces Phil Trans. R. Soc. A 370 1478–98

[8] Costa D, Garrain P A and Baaden M 2013 Understanding small biomolecule–biomaterial interactions: a review of fundamental theoretical and experimental approaches for biomolecule interactions with inorganic surfaces J. Biomed. Mater. Res. A 101 1210–22

[9] Friesner R A 2005 Ab initio quantum chemistry: methodology and applications PNAS 102 6648–53

[10] Hartree D R 1928 The wave mechanics of an atom with a non-Coulomb central field. Part I. Theory and methods Mathematical Proc. of the Cambridge Philosophical Society (Cambridge: Cambridge University Press) p 89–110

[11] Hohenberg P and Kohn W 1964 Inhomogeneous electron gas Phys. Rev. 136 864–71

[12] Zhang T, Xue Q, Zhang S and Dong M 2012 Theoretical approaches to graphene and graphene-based materials Nano Today 7 180–200

[13] de Pablo J J, Jones B, Lind C, Ozolins V and Ramirez A P 2014 The materials genome initiative, the interplay of experiment, theory and computation Curr. Opin. Solid State Mater. Sci. 18 99–117

[14] Shipman P, Foster G and Schoeninger M 1984 Burnt bones and teeth: an experimental study of color, morphology, crystal structure and shrinkage J. Archaeol. Sci. 11 307–25

[15] Yukna R A et al 1998 Multi-Center clinical evaluation of combination of anorganic bovine-derived hydroxyapatite matrix (ABM)/cell binding peptide (P-15) as a bone replacement graft material in human periodontal osseous defects. 6-month results J. Periodontol. 69 655–63

[16] Yang P P, Quan Z W, Li C X, Kang X J, Lian H Z and Lin J 2008 Bioactive, luminescent and mesoporous europium-doped hydroxyapatite as a drug carrier Biomaterials 29 4341–7

[17] Ogiso M 1998 Reassessment of long-term use of dense HA as dental implant: case report J. Biomed. Mater. Res. 43 318–20

[18] Posner A S, Perloff A and Diorio A F 1958 Refinement of the hydroxyapatite structure Acta Crystallogr. 11 308–9

[19] Kay M I and Young R A 1964 Crystal structure of hydroxyapatite Nature 204 1050–2

[20] Ren F, Leng Y, Xin R and Ge X 2010 Synthesis, characterization and ab initio simulation of magnesium-substituted hydroxyapatite Acta Biomater. 6 2787–96

[21] Ma X and Ellis DE 2008 Initial stages of hydration and Zn substitution/occupation on hydroxyapatite (0 0 0 1) surfaces Biomaterials 29 257–65

[22] Jimenez-Izal E, Chiatti F, Corno M, Rimola A and Ugliengo P 2012 Glycine adsorption at nonstoichiometric (0 1 0) hydroxyapatite surfaces: a B3LYP study J. Phys. Chem. C 116 14561–7

[23] Lou Z et al 2012 First-principles study of the adsorption of lysine on hydroxyapatite (1 0 0) surface Appl. Surf. Sci. 258 4911–6

[24] Almora-Barrios N, Austen K F and de Leeuw N H 2009 Density functional theory study of the binding of glycine, proline, and hydroxyproline to the hydroxyapatite (0 0 0 1) and (0 1 1 0) surfaces Langmuir 25 5018–25

[25] Landis W J and Silver F H 2008 Mineral deposition in the extracellular matrices of vertebrate tissues: identification of possible apatite nucleation sites on type I collagen Cells Tissues Organs 189 20–4

[26] Pan H, Tao J, Xu X and Tang R 2007 Adsorption processes of Gly and Glu amino acids on hydroxyapatite surfaces at the atomic level Langmuir 23 8972–81

[27] Lu X 2011 Hexagonal hydroxyapatite formation on TiO2 nanotubes under urea modulation CrystEngComm 13 3741–9

[28] Almora-Barrios N and de Leeuw N H 2010 A density functional theory study of the interaction of collagen peptides with hydroxyapatite surfaces Langmuir 26 14535–42

[29] Almora-Barrios N and de Leeuw N H 2010 Modelling the interaction of a Hyp-Pro-Gly peptide with hydroxyapatite surfaces in aqueous environment CrystEngComm 12 960–7

[30] Biswas S and Becker U 2013 Molecular modeling of cell adhesion peptides on hydroxyapatite and TiO2 surfaces: implication in biomedical implant devices J. Biomater. Nanobiotechnol. 4 351–6

[31] Sawyer A, Hennessy K and Bellis S 2005 Regulation of mesenchymal stem cell attachment and spreading on hydroxyapatite by RGD peptides and adsorbed serum proteins Biomaterials 26 1467–75

[32] Puleo D and Nanci A 1999 Understanding and controlling the bone-implant interface Biomaterials 20 2311–21

[33] Wozney J M 1992 The bone morphogenetic protein family and osteogenesis Mol. Reprod. Dev. 32 160–7

[34] Dong X L, Wang Q, Wu T and Pan H 2007 Understanding adsorption-desorption dynamics of BMP-2 on hydroxyapatite (0 0 1) surface Biophys. J. 93 750–9

[35] Dong X L, Wang Q, Wu T, Ma L Y and Fu C X 2011 The dynamic behaviours of protein BMP-2 on hydroxyapatite nanoparticles Mol. Simul. 37 1097–104

[36] Zhou H, Wu T, Dong X, Wang Q and Shen J 2007 Adsorption mechanism of BMP-7 on hydroxyapatite (0 0 1) surfaces Biochem. Biophys. Res. Commun. 361 91–6

[37] Potts J R and Campbell I D 1996 Structure and function of fibronectin modules Matrix Biol. 15 313–20

[38] Humphries J D, Byron A and Humphries M J 2006 Integrin ligands at a glance J. Cell Sci. 119 3901–3

[39] Guo T, Kang W, Xiao D, Duan R, Zhi W and Weng J 2013 Molecular docking characterization of a four-domain segment of human fibronectin encompassing the RGD loop with hydroxyapatite Molecules 19 149–58

[40] Liao C, Xie Y and Zhou J 2014 Computer simulations of fibronectin adsorption on hydroxyapatite surfaces RSC Adv. 4 15759–69

[41] Lai Z B, Wang M, Yan C and Oloyede A 2014 Molecular dynamics simulation of mechanical behavior of osteopontin-hydroxyapatite interfaces J. Mech. Behav. Biomed. Mater. 36 12–20

[42] Chen X, Wu T, Wang Q and Shen J-W 2008 Shield effect of silicate on adsorption of proteins onto silicon-doped hydroxyapatite (1 0 0) surface Biomaterials 29 2423–32

[43] Chen X, Wang Q, Shen J, Pan H and Wu T 2007 Adsorption of leucine-rich amelogenin protein on hydroxyapatite (0 0 1) surface through-COO-claws J. Phys. Chem. C 111 1284–90

[44] Rhee S H and Tanaka J 1999 Effect of citric acid on the nucleation of hydroxyapatite in a simulated body fluid Biomaterials 20 2155–60

[45] Filgueiras M, Mkhonto D and De Leeuw N 2006 Computer simulations of the adsorption of citric acid at hydroxyapatite surfaces J. Cryst. Growth 294 60–8

[46] De Leeuw N and Rabone J 2007 Molecular dynamics simulations of the interaction of citric acid with the hydroxyapatite (0 0 0 1) and (0 1 1 0) surfaces in an aqueous environment CrystEngComm 9 1178–86

[47] Streeter I and de Leeuw N H 2011 Binding of glycosaminoglycan saccharides to hydroxyapatite surfaces: a density functional theory study Proc. R. Soc. A 467 2084–101

[48] Zhang H P et al 2009 Molecular dynamics simulations on the interaction between polymers and hydroxyapatite with and without coupling agents Acta Biomater. 5 1169–81

[49] Zhang H P, Lu X, Fang L, Qu S, Feng B and Weng J 2008 Atomic-scale interactions at the interface of biopolymer/hydroxyapatite Biomed. Mater. 3 044110

[50] Bhowmik R, Katti K S and Katti D 2007 Molecular dynamics simulation of hydroxyapatite–polyacrylic acid interfaces Polymer 48 664–74

Biomed. Mater. 10 (2015) 032001

16

Q Wang et al

[51] Brown W, Eidelman N and Tomazic B 1987 Octacalcium phosphate as a precursor in biomineral formation Adv. Dent. Res. 1 306–13

[52] Wang K, Leng Y, Lu X, Ren F and Ge X 2012 Study of protein adsorption on octacalcium phosphate surfaces by molecular dynamics simulations J. Mater. Sci.: Mater. Med. 23 1045–53

[53] Kavathekar R S, Dev P, English N J and MacElroy J 2011 Molecular dynamics study of water in contact with the TiO2 rutile-1 1 0, 1 0 0, 1 0 1, 0 0 1 and anatase-1 0 1, 0 0 1 surface Mol. Phys. 109 1649–56

[54] Wang M H et al 2014 Density functional theory study of interactions between glycine and TiO2/graphene nanocomposites Chem. Phys. Lett. 599 86–91

[55] Diebold U 2003 Structure and properties of TiO2 surfaces: a brief review Appl. Phys. A 76 681–7

[56] Guo Yn, Lu X, Zhang Hp, Weng J, Watari F and Leng Y 2011 DFT study of the adsorption of aspartic acid on pure, N-doped, and Ca-doped rutile (1 1 0) surfaces J. Phys. Chem. C 115 18572–81

[57] Szieberth D, Ferrari A M and Dong X 2010 Adsorption of glycine on the anatase (1 0 1) surface: an ab initio study Phys. Chem. Chem. Phys. 12 11033–40

[58] Muir J M R and Idriss H 2013 Computational study of cysteine interaction with the rutile TiO2 (1 1 0) surface Surf. Sci. 617 60–7

[59] Liu J, Dong L, Guo W, Liang T and Lai W 2013 CO adsorption and oxidation on N-doped TiO2 nanoparticles J. Phys. Chem. C 117 13037–44

[60] Zhang H P, Lu X, Luo Xg, Lin Xy and Zhou Yf 2014 Effects of O-deficiency on the interaction between rutile and Arg: a density functional theory study Physica E 61 83–9

[61] Seitz R, Brings R and Geiger R 2005 Protein adsorption on solid–liquid interfaces monitored by laser-ellipsometry Appl. Surf. Sci. 252 154–7

[62] Carravetta V and Monti S 2006 Peptide-TiO2 surface interaction in solution by Ab Initio and molecular dynamics simulations J. Phys. Chem. B 110 6160–9

[63] Zhang H P, Lu X, Fang L M, Weng J, Huang N and Leng Y 2008 Molecular dynamics simulation of RGD peptide adsorption on titanium oxide surfaces J. Mater. Sci.: Mater. Med. 19 3437–41

[64] Muir J M R, Costa D and Idriss H 2014 DFT computational study of the RGD peptide interaction with the rutile TiO2 (1 1 0) surface Surf. Sci. 624 8–14

[65] Zhang HP et al 2011 Effects of aqueous environment and surface defects on Arg-Gly-Asp peptide adsorption on titanium oxide surfaces investigated by molecular dynamics simulation J. Biomed. Mater. Res. A 96A 466–76

[66] Utesch T, Daminelli G and Mroginski M A 2011 Molecular dynamics simulations of the adsorption of bone morphogenetic protein-2 on surfaces with medical relevance Langmuir 27 13144–53

[67] Wu C, Chen M and Xing C 2010 Molecular understanding of conformational dynamics of a fibronectin module on rutile (1 1 0) surface Langmuir 26 15972–81

[68] Monti S and Walsh T R 2011 Molecular dynamics simulations of the adsorption and dynamical behavior of single DNA components on TiO2 J. Phys. Chem. C 115 24238–46

[69] Novoselov K S et al 2004 Electric field effect in atomically thin carbon films Science 306 666–9

[70] Barinov A et al 2009 Initial stages of oxidation on graphitic surfaces: photoemission study and density functional theory calculations J. Phys. Chem. C 113 9009–13

[71] Fu Q and Bao X 2009 Progress in graphene chemistry Chin. Sci. Bull. 54 2657–66

[72] Sayyar S, Murray E, Thompson B C, Gambhir S, Officer D L and Wallace G G 2013 Covalently linked biocompatible graphene/polycaprolactone composites for tissue engineering Carbon 52 296–304

[73] Miao W et al 2013 Cholesteryl hyaluronic acid-coated, reduced graphene oxide nanosheets for anti-cancer drug delivery Biomaterials 34 9638–47

[74] Hui X et al 2013 Graphene-based nanoprobes and a prototype optical biosensing platform Biosens. Bioelectron. 50 251–5

[75] Jiang Y, Ji-Hyuk Y, Strickler J R, Woo-Jin C and Gunasekaran S 2013 Nickel nanoparticle-chitosan-reduced graphene oxide-modified screen-printed electrodes for enzyme-free glucose sensing in portable microfluidic devices Biosens. Bioelectron. 47 530–8

[76] Guo Y N, Lu X, Weng J and Leng Y 2013 Density functional theory study of the interaction of arginine-glycine-aspartic acid with graphene, defective graphene, and graphene oxide J. Phys. Chem. C 117 5708–17

[77] Cazorla C 2010 Ab initio study of the binding of collagen amino acids to graphene and A-doped (A = H, Ca) graphene Thin Solid Films 518 6951–61

[78] Rajesh C, Majumder C, Mizuseki H and Kawazoe Y 2009 A theoretical study on the interaction of aromatic amino acids with graphene and single walled carbon nanotube J. Chem. Phys. 130 124911

[79] Xu L and Yang X 2014 Molecular dynamics simulation of adsorption of pyrene–polyethylene glycol onto graphene J. Colloid Interface Sci. 418 66–73

[80] Akdim B et al 2013 Electronic properties of a graphene device with peptide adsorption: insight from simulation ACS Appl. Mat. Interfaces 5 7470–7

[81] Kim S N et al 2011 Preferential binding of peptides to graphene edges and planes J. Am. Chem. Soc. 133 14480–3

[82] Zaib Q, Khan I A, Saleh N B, Flora J R, Park Y G and Yoon Y 2012 Removal of bisphenol A and 17β-estradiol by single-walled carbon nanotubes in aqueous solution: adsorption and molecular modeling Water Air Soil Pollut. 223 3281–93

[83] Al-Aqtash N and Vasiliev I 2009 Ab initio study of carboxylated graphene J. Phys. Chem. C 113 12970–5

[84] Boukhvalov D and Katsnelson M 2008 Chemical functionalization of graphene with defects Nano Lett. 8 4373–9

[85] Wang C et al 2006 Chemical functionalization of carbon nanotubes by carboxyl groups on Stone–Wales defects: a density functional theory study J. Phys. Chem. B 110 10266–71

[86] Ghaderi N and Peressi M 2010 First-principle study of hydroxyl functional groups on pristine, defected graphene, and graphene epoxide J. Phys. Chem. C 114 21625–30

[87] Denis P A and Iribarne F 2013 Comparative study of defect reactivity in graphene J. Phys. Chem. C 117 19048–55

[88] Ding N, Lu X and Wu C-M L 2012 Nitrated tyrosine adsorption on metal-doped graphene: a DFT study Comput. Mater. Sci. 51 141–5

[89] Zhang Z, Jia H, Ma F, Han P, Liu X and Xu B 2011 First principle study of cysteine molecule on intrinsic and Au-doped graphene surface as a chemosensor device J. Mol. Model. 17 649–55

[90] Zhang H-P, Luo X-G, Lin X-Y, Lu X, Leng Y and Song H-T 2013 Density functional theory calculations on the adsorption of formaldehyde and other harmful gases on pure, Ti-doped, or N-doped graphene sheets Appl. Surf. Sci. 283 559–65

[91] Liu X Y and Zhang J M 2014 Formaldehyde molecule adsorbed on doped graphene: a first-principles study Appl. Surf. Sci. 293 216–9

[92] Abergel D, Apalkov V, Berashevich J, Ziegler K and Chakraborty T 2010 Properties of graphene: a theoretical perspective Adv. Phys. 59 261–482

[93] Becton M, Zhang L Y and Wang X Q 2013 Effects of surface dopants on graphene folding by molecular simulations Chem. Phys. Lett. 584 135–41

[94] Kang H S 2005 Theoretical study of binding of metal-doped graphene sheet and carbon nanotubes with dioxin J. Am. Chem. Soc. 127 9839–43

Biomed. Mater. 10 (2015) 032001

17

Q Wang et al

[95] Vovusha H, Sanyal S and Sanyal B 2013 Interaction of nucleobases and aromatic amino acids with graphene oxide and graphene flakes J. Phys. Chem. Lett. 4 3710–8

[96] Cortés-Arriagada D, Sanhueza L and Santander-Nelli M 2013 Modeling the physisorption of bisphenol A on graphene and graphene oxide J. Mol. Model. 19 3569–80

[97] Baweja L, Balamurugan K, Subramanian V and Dhawan A 2013 Hydration patterns of graphene-based nanomaterials (GBNMs) play a major role in the stability of a helical protein: a molecular dynamics simulation study Langmuir 29 14230–8

[98] He S 2010 A graphene nanoprobe for rapid, sensitive, and multicolor fluorescent DNA analysis Adv. Funct. Mater. 20 453–9

[99] Saikia N and Deka R C 2013 Ab initio study on the noncovalent adsorption of camptothecin anticancer drug onto graphene, defect modified graphene and graphene oxide J. Comput.-Aided Mol. Des. 27 807–21

[100] Kasemo B 2002 Biological surface science Surf. Sci. 500 656–77

[101] Dechang L, Baohua J, Kehchi H and Yoggang H 2009 Coarse grained modeling of biopolymers and proteins: methods and applications Int. J. Appl. Mech. 1 113–36

[102] Ingolfsson H I et al 2014 The power of coarse graining in biomolecular simulations Wiley Interdiscip. Rev.-Comput. Mol. Sci. 4 225–48

[103] Larson R G 2004 Principles for coarse-graining polymer molecules in simulations of polymer fluid mechanics Mol. Phys. 102 341–51

[104] Ramachandran A, Guo Q, Iqbal S M and Liu Y 2011 Coarse-grained molecular dynamics simulation of DNA trans location in chemically modified nanopores J. Phys. Chem. B 115 6138–48

[105] Taylor W R and Katsimitsoulia Z 2010 A coarse-grained molecular model for actin-myosin simulation J. Mol. Graph. Modelling 29 266–79

Biomed. Mater. 10 (2015) 032001