Article pubs.acs.org/Langmuir
Molecular Simulation of Fibronectin Adsorption onto Polyurethane Surfaces Melisa Panos,† Taner Z. Sen,‡ and M. Göktuğ Ahunbay*,† †
Department of Chemical Engineering, Istanbul Technical University, Istanbul, Turkey U.S. Department of Agriculture − Agricultural Research Service, Corn Insects and Crop Genetics Research Unit and Department of Genetics, Development and Cell Biology, Iowa State University, Ames, Iowa 50011, United States
‡
S Supporting Information *
ABSTRACT: Poly(ethylene glycol)-based polyurethanes have been widely used in biomedical applications; however, they are prone to swelling. A natural polyol, castor oil, can be incorporated into these polyurethanes to control the degree of the swelling, which alters mechanical properties and protein adsorption characteristic of the polymers. In this work, we modeled poly(ethylene glycol) and castor oil copolymers of hexamethylene diisocyanate-based polyurethanes (PEG-HDI and CO-HDI, respectively) and compared their mechanisms for fibronectin adsorption using molecular mechanics and molecular dynamics simulations. Results showed that the interplay between the hydrophobic residues concentrated at the Nterminal end of the protein, the surface roughness, and the hydrophilicity of the polymer surface determine the overall protein adsorption affinity. Incorporating explicit water molecules in the simulations results in higher affinity for fibronectin adsorption to more hydrophobic surface of CO-HDI surfaces, emphasizing the role that water molecules play during adsorption. We also observed that the strain energies that are indicative of flexibility and consequently entropy are significantly affected by the changes in the patterns of β-sheet formation/breaking. Our study lends supports to the view that while castor oil controls the degree of swelling, it increases the adsorption of fibronectin to a limited extent due to the interplay between its hydrophobicity and its surface roughness, which needs to be taken into account during the design of polyurethane-based biomaterials.
1. INTRODUCTION The surface adsorption of proteins onto materials is essential in various biomedical and biotechnological applications. An improved understanding of protein adsorption mechanism can therefore allow development of new materials for various applications, such as biocompatible materials for biomedical devices that are inserted in living bodies, biosensors that can detect the presence of target proteins, or purifiers that can absorb unwanted biological materials from a solution.1 For each type of application, different material properties are required. For example, once a biocompatible material is inserted into the body, the material surface will interface and interact with the blood proteins and its surface will be covered with a thin layer of vast arrays of proteins. Therefore, for such a material, enhanced binding specificity to a unique protein would be a highly desired property. Previous studies established that protein−material interactions depend on the protein characteristics, such as size, charge, hydrophobicity, conformation, and stability.1 Larger proteins are usually adsorbed more easily on surfaces because of their larger interaction surface, such as albumin vs fibrinogen on a silica surface. However, size is not always the main factor: e.g., hemoglobin has a higher surface affinity to the surface with respect to larger fibrinogen.2 The charge distribution and polarization on the protein surface also play an important role © 2012 American Chemical Society
in the adsorption process depending on the pH of the environment. During this complex process of adsorption, some amino acids can be reoriented outward from the protein surface and provide better contact with the material surface, which allows binding sites be accessible only in specific adsorbed conformations.1 Protein stability also plays a role, since less stable proteins spread more easily on the material surface, providing larger interaction areas.2 Similarly important is the surface characteristics of the materials. For example, higher surface roughness increases protein adsorption because of increased surface area.1 However, protein adsorption does depend on not only the characteristics of proteins and surfaces individually but also the complementarity of the protein and polymer properties. For example, most of the proteins being hydrophobic on the protein core have higher affinity to hydrophobic surfaces than hydrophilic surfaces when the protein can open up before being adsorbed, creating a larger binding interface. There is also a favorable entropic component of the system: on a hydrophobic surface, water−water interactions are much stronger than water− surface interactions, and water molecules form a nonwetting Received: April 17, 2012 Revised: July 26, 2012 Published: August 2, 2012 12619
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
mechanics calculations and molecular dynamics simulations to compare the affinity of these polyurethanes to the fibronectin protein in the presence of implicit and explicit water.
layer above the surface, which decreases the entropy of the system. During the protein adsorption process, this layer breaks apart and entropy increases instantaneously, acting as a driving force for adsorption. On the other hand, water molecules interact strongly with hydrophilic surfaces and compete with proteins for adsorption. In previous studies on human serum albumin (HSA) adsorption on copolymers of N-isopropylacrylamide/N-tert-butylacrylamide, it was shown that HSA adsorption is strongly favored on less hydrophilic (85:15) copolymer with respect to (50:50) copolymer.3,4 Another important characteristic is the electrostatic charges on the surface.5 Opposite charges favor adsorption and increase ionic strength, so considering that most of the proteins in the blood serum are negatively charged, negatively charged material surfaces would be less favorable for adsorption of these proteins (e.g., fibronectins). The complexity of the above-mentioned thermodynamic interactions makes adsorption a unique phenomenon to study. Thus, understanding the mechanism and energetics of protein− solid surface interactions is critical to efficiently design novel biomaterials. Atomistic-level computational simulations can provide valuable insight into the details of interactions of individual amino acids with a surface. For this purpose, different computational tools could be applied to understand the mechanism of protein adsorption, such as molecular dynamics (MD),6−15 molecular mechanics (MM),16−21 and Monte Carlo methods.22−24 These methods can provide crucial atomisticlevel details to understand and manipulate surface−protein interactions, which would be highly challenging to obtain via experimental methods. The information obtained using these computational methods can be very useful to tailor optimized surfaces to control adsorption of target proteins, e.g., by means of grafting with specific functional groups. In this work we focus on the adsorption characteristics of polyurethanes, which constitute a very appealing class of polymers because of their high biocompatibility and excellent physical and mechanical properties. These versatile polymers have been widely used in wound dressings as well as in cardiovascular and breast implants.25,26 Swelling of the poly(ethylene glycol) (PEG)-based polyurethanes is an important issue in biomedical applications and is needed to be controlled by addition of cross-linkers. Alternatively, castor oil (CO), which is attractive due its safety and effectiveness in biomedical applications, can be used as the polyol component for decreasing swelling degree of PEG-based polyurethanes due to its high content of ricinoleic acid, which has a hydroxyl functional group on the 12th carbon. Since the amount of castor oil in the polyurethane determines strongly polymer properties, the performance of the castor oil and PEG-based polyurethane films in protein adsorption has been studied in great detail.27,28 In order to better understand the alterations in the adsorption characteristics due to the incorporation of castor oil into PEG-based polyurethanes, we studied adsorption of type 1 module of the blood protein fibronectin (protein id: 1fbr29) on two model polyurethane systems: (1) crystalline poly(ethylene glycol)−hexamethylene diisocyanate (PEGHDI) and (2) amorphous castor oil−hexamethylene diisocyanate (CO-HDI) polymers, which constitute two limiting compositions. We constructed the atomistic models of these two polymers, calculated their mechanical properties driven by adsorption energetics, and compared these results with the available experimental data. Finally, we performed molecular
2. SIMULATION METHOD All the simulations in this study were performed using the Materials Studio 4.1 simulation package, and the secondary structure visualizations were obtained using Discovery Studio 3.0 software (Accelrys Inc., San Diego, CA). Construction of Bulk Polymer Models. In this part of the study, bulk polymer models were constructed and their structural and mechanical properties were estimated using the analysis tools in the simulation package. The intermolecular interactions of the polymer chains were modeled with the COMPASS force field,30 which was previously used to model polymeric systems.31 Polyurethanes used in this study are copolymers of polyol and diisocyanate monomers. The molecular structures of monomers used in this study are given in Figure 1. The experimental densities of the PEG-HDI and CO-HDI polyurethanes were reported as 1.23 and 0.96 g/cm3, respectively.28
Figure 1. Polyurethane monomers used in this study: (a) castor oil (CO), (b) poly(ethylene glycol) (PEG), and (c) hexamethylene diisocyanate (HDI) to create PEG-HDI and CO-HDI polyurethane copolymers.
In order to construct the crystalline PEG-HDI model, we first retrieved the structure of the PEG crystal unit cell that contains four chains of seven repeat units from the software library. These chains were then extended along their backbone, i.e., (100) direction, to obtain chains of 68 repeat units, which corresponds to the molecular weight of 3000 g/mol of the PEG polymer in accordance with the previous experimental work.28 The cell was further replicated 5 times along the (001) direction to create a surface large enough to accommodate the size of the protein. This PEG model was used in our study to identify the contribution of the HDI groups during protein adsorption by removing seven repeat units from each chain and replacing by the HDI groups, corresponding to a HDI content of 5 wt %. The geometry of the resulting PEG-HDI structure was optimized using the Discover module to obtain the final simulation cell of 214.28 × 13.04 × 96.0 Å3. Since the structure of the amorphous polymers does not form perfectly symmetrical structures like their crystalline counterparts, a different procedure was applied to construct the bulk model for the amorphous CO-HDI polymer. First, COHDI chains were constructed by connecting two CO and three HDI groups as shown in Figure 1, and its geometry was optimized based on energy minimization. Next, two of these chains were used to construct the amorphous polymer cell using the amorphous cell module of the software package and then equilibrated using the procedure proposed by Heuchel et al.31 to construct amorphous polymer models: The cell was first 12620
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
acids distributed on β-sheets and coils (Figure 3). In the presence of multiple interactions among the protein, water
compressed at high pressure (5000 bar) for 5 ps at 300 K. Next, NVT-MD runs were applied at 600 and 300 K, successively, each about for 20 ps. Subsequently, a 20 ps MD run was performed in the NPT ensemble at 1 bar, and the resulting density was compared to experimental value. If the density was lower than the experimental density, the first two steps were repeated. Finally, MD simulation was performed in NVT ensemble at 1 bar about 300 ps. The resulting CO-HDI simulation cell had dimensions of 34.26 × 34.26 × 34.26 Å3 with a density matching the experimental value of 0.98 g/cm3. Modeling Protein−Surface Interactions. In order to compare the affinity of the model surfaces to the protein, a twostep procedure was employed: First, energy minimization in dielectiric medium (implicit solvent) was performed in order to determine the optimal orientations for adsorption on each surface. Then, full atomistic MD simulations with explicit water were carried out using the structures determined in the minimization step on each of the crystalline and amorphous polyurethane surfaces to investigate the effect of protein−water competition on the adsorption behavior. This combination of energy minimization in implicit solvent and MD simulations in explicit water is useful in learning about different aspects of the protein−surface interactions and had been previously used in various studies.18−21
Figure 3. (a) Fibronectin type I module (protein id: 1fbr)29 (b) and its amino acid sequence and the secondary structure calculated using DSSP36 based on X-ray-determined crystal coordinates. Arrows indicate β-strands, the purple curves are hydrogen-bonded turns, and the sulfur bonds between cysteine residues are shown as dotted green lines. Figure 1b is taken from the Protein Data Bank.32
molecules, and surfaces, the choice of force field was based on the accurate representation of protein−water and protein− surface interactions, which were of the primary concern since the surfaces in this study were kept rigid. Hence, all the interactions were modeled with the consistent valence force field (CVFF),33 which was originally parametrized for peptide and protein structures and also applicable to polymers in aqueous environments.34 The CVFF force field was used previously in similar studies18,19,37 and was shown to reproduce the protein dynamics satisfactorily. In order to compare the affinity of the model surfaces to the protein, the adsorption energetics of the surfaces were compared by optimizing the geometry of the protein on each surface through energy minimization in the dielectric medium with ε = 78. This allowed direct evaluation of the surface− protein interactions without the interference of the water molecules. For this purpose the geometry of the isolated fibronectin domain was energy-minimized first, and then it was placed away from each of the model surfaces in six different orientations to remove the bias of approach angle. These orientations corresponded to the sides of a hypothetical rectangular box containing the protein domain as illustrated in Figure 4. For each starting orientation, energy minimization was carried out, while the polymer structure was kept rigid, to analyze changes in the interaction energy between the surface and polymer due to adsorption.
Figure 2. Bulk polymer matrices (a) PEG-HDI, (b) CO-HDI, and (c) PVA polymers. The atoms are colored in the following manner: gray for carbon atoms, red for oxygen atoms, blue for nitrogen atoms, and white for hydrogen atoms. The box around the molecules is shown to provide information about the matrix shapes used in the simulations.
The polymer surfaces were constructed to simulate surface adsorption of fibronectin by removing periodic boundaries of the simulation box to prevent edge effect. The surface of amorphous CO-HDI was extended by duplicating the initial simulation cell along the (100) direction, so that the surface is wide enough to adsorb the whole fibronectin protein. To compare our approach to that of Raffaini and Ganazzoli, we also simulated fibronectin adsorption to PVA. For this purpose, 50 PVA polymer chains, each of 20 repeat units, were used to construct the amorphous PVA model using the same procedure we used to generate the CO-HDI simulation cell. The resulting PVA simulation cell had dimensions of 65.29 × 39.53 × 23.93 Å3. The structure of the fibronectin type I module (1fbr) was retrieved from the Protein Data Bank32 and contains 93 amino
Figure 4. Fibronectin domain in six different initial orientations perpendicular to the polymer surfaces. The structures are rotated three dimensionally to obtain a sampling of possible approach angles. The perspective is given in such a way that the protein is approaching the polymer surface located below from the top. 12621
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
Figure 5. Absolute protein−surface interaction energies (first row) and strain energies (second row) (a) per protein amino acids and (b) per atoms within adsorption layers of different thickness. Note that the range for energy values in the y-axes differ for each figure.
function of protein atoms and amino acids being in contact with the polymer surface within arbitrary adsorption layers of thicknesses (δ) of 3, 5, and 7 Å were plotted. A linear model was fitted to the interaction energy as a function of the number of amino acids. The slope in the model, corresponding to the absolute interaction energies between the protein and the surface, was used as a basis for the comparison of the surface affinity to the protein, larger slope values indicating larger interaction strength. The calculation procedure was repeated for the strain energy, which is defined as
In the last part of this work, the orientations yielding minimum-energy configurations were chosen for the PEG-HDI and CO-HDI surfaces, and molecular dynamics were carried by replacing the implicit solvent by explicit water molecules in order to reveal the effect of water−protein−surface interactions. The protein was soaked with around 10 000 SPC water molecules35 at a density of 1 g/cm3. The starting configuration was selected after the energy minimization, and simulations were run up to 5 ns with a time step of 1 fs at 300 K until the system reached equilibrium.
Estrain = Efrozen − Efree
3. RESULTS AND DISCUSSION Analysis of Adsorption Energetics and Orientation. For each orientation, the interaction energy between the protein and the polymer surface (Eint) was calculated after the energy minimization: E int = (Eprot + Epoly ) − Etot
(2)
where Efrozen is the intramolecular energy of the protein after adsorption. The strain energy is indicative of broken H-bonds and thus breaking of β-sheets in the protein. The intramolecular energy (Eprot) of the free-standing fibronectin domain was calculated after geometry optimization by energy minimization. The resulting value of 3.19 MJ/mol agrees very well with the value of 3.13 MJ/mol calculated by Raffaini and Ganazzoli.18 Next, the interaction energy (Eint) and the strain (Estrain) energy of the protein−surface system were calculated for all six different protein orientations through energy minimization as described above. For each of the
(1)
where Eprot is the intramolecular energy of the fibronectin domain before adsorption, Etot is the total energy of the protein−polymer system after adsorption, and Epoly is the intramolecular energy of the polymer surface (Epoly = 0, as the polymer structure was kept rigid). Interaction energies as a 12622
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
Figure 6. Interaction energies with respect to protein amino acids and atoms within an adsorption layer of δ = 7 Å on the polymer surface. Numbers next to the data marks indicate the initial orientation of fibronectin above the surface as defined in Figure 4.
Figure 7. Configurational changes of fibronectin upon adsorption on the surfaces of (a) CO-HDI, (b) PEG-HDI, (c) PEG, and (d) PVA. The figures on the left show the configurations and positions in the beginning of the simulations and the figures on the right at the end of simulations with implicit water. 12623
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
the PVA surface. Similarly in our study, the interaction energy was the lowest for the PVA surface, where the intramolecular rearrangements were minor. Moreover, the strain energy accompanying these deformations was mainly due to the broken hydrogen bonds in the secondary structure and hence was much smaller than the interaction energy with the surface. Note that individual interactions between water molecules and the surface atoms were not considered in this part of the study since water molecules were treated implicitly, which allows comparison of our results with those of the above-mentioned study when applied the same approach. Therefore, the higher affinity of the protein to graphite cannot be attributed to the lack of competition between the protein and water molecules to bind onto the polymer surface. The extent of deformation of the protein upon adsorption was analyzed through the calculation of strain energies for each orientation, which were then plotted against both the number of amino acids and the number of protein atoms to obtain absolute strain energies. The relative ordering of the strain energies for the surfaces was the same for all three layer thicknesses. Absolute strain energies with respect to protein amino acids and atoms being in contact with the polymer surface within a distance of δ = 3, 5, and 7 Å are presented in Figure 5. It can be seen that the highest strain energy for the case of δ = 7 Å was obtained on the PEG surface, while for the rest the magnitudes of the strain energy were comparable. These results show that the deformation of the protein structure is similar for both polyurethane surfaces. Adsorption in the Presence of Explicit Water Molecules. In the final step, the minimum energy yielding orientations 6 and 3 for PEG-HDI and CO-HDI, respectively, was analyzed through MD simulations by replacing the implicit solvent by explicit water molecules as described in the Simulation Method section. The 5 ns simulations were sufficient for the stabilization of the protein on the surfaces that was monitored via the evolution of the protein−surface interaction energies (shown in the Supporting Information) as well as the trajectory of the protein on the surface. The resulting configurations on the PEG-HDI and CO-HDI surfaces are given in Figure 9. These additional simulations took into account the interactions among the protein, water molecules, and surface, thus providing a more realistic comparison of the protein affinity for the surfaces in the presence of competitive adsorption/desorption caused by water molecules. The resulting interaction energies for the hydrophilic PEG-HDI and hydrophobic CO-HDI surfaces (335 and 397 kJ/mol, respectively) shown in Table 1 indicate that the protein favors the hydrophobic surfaces more than the hydrophilic surface in the presence of explicit water. However, the difference in the interaction energies is smaller as compared to the results of molecular mechanics calculations performed with implicit water, where the PEG-HDI surface exhibited a significantly higher affinity to the protein. This difference may be explained by the competitive adsorption between the protein fragment and the water molecules on the hydrophilic surface. While 43 residues were adsorbed on the PEG-based hydrophilic surface in the absence of explicit water molecules, the number decreased to 16 in their presence, whereas the number for adsorbed residues decreased from 37 to 24 in the case of the CO-based hydrophobic surface. It should be noted that the presence of explicit water molecules led to weaker protein− surface interactions in the latter case, since one would expect that competitive adsorption between the water molecules and
orientations, the number of amino acids and the number of protein atoms adsorbed on the surface were determined. The resulting interaction energies were then plotted against the number of amino acids and the number of protein atoms within the three different adsorption layers. The slopes of the linear fits to the data, which correspond to absolute interaction energies indicating the adsorption strength of the polymeric surfaces, are compared in Figure 5 along with the strain energies. It can be seen that for the adsorption layers with thicknesses 5 and 7 Å the order of the absolute interaction energies remains unchanged. Furthermore, the best linear fit between the interaction energies and number of amino acids was obtained for δ = 7 Å in which case the highest interaction energies for PEG-HDI, CO-HDI, PEG, and PVA correspond to orientations 6, 3, 5, and 4, respectively, as shown in Figure 6. Note that absolute interaction energy of 13.3 kJ/mol for PVA surface for δ = 5 Å is in good agreement with the previously reported value of 14.5 kJ/mol.19 In correlation with the interaction energies, the smallest number of amino acids was adsorbed on the PVA surface. For the rest of the surfaces, the numbers are similar. Among the four polymers, PVA is the most hydrophobic one; stronger adsorption of fibronectin to the PVA surface is carried out by opening up to expose its hydrophobic core for adsorption. The PEG-HDI surface has a higher protein affinity than the COHDI surface when water is treated implicitly, although the latter is less hydrophilic. This can be explained by the favorable interactions between the hydrophilic amino acids on the fibronectin surface and surface atoms on the crystalline PEGHDI, allowing an energetically more efficient contact with the protein. The configurational changes of the protein upon adsorption on each of the four surfaces are shown in Figure 7. The comparison of interaction energies with respect to the percentage of polar amino acids on the polymer surfaces in the most favorable orientations in Figure 8 indicates interplay between the protein affinity and the hydrophilicity of the surfaces. In Raffaini et al.’s study on fibronectin adsorption of PVA and graphite surfaces,37 it was reported that interactions of the protein with both surfaces were favorable, but the hydrophobicity of the graphite surface induced a partial loss of the secondary structure, as opposed to the hydrophilicity of
Figure 8. Interaction energies with respect the percentage of polar amino acids within an adsorption layer of δ = 7 Å on the polymer surface in the most favorable orientations (simulations are done with implicit water). 12624
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
Figure 9. Conformational changes in fibronectin upon adsorption on (a) PEG-HDI and (b) CO-HDI surfaces: From top to bottom: secondary structure before adsorption, secondary structure after adsorption, atomistic detail on the polymer surface (explicit water molecules are not shown for clarity). Residues that are adsorbed on the surface are represented in red in before and after adsorption.
Table 1. The van der Waals and Electrostatic Contributions to the Interaction Energies and the Number (NAA) and List of the of Amino Acid Residues Adsorbed on the PEG-HDI and CO-HDI Surfaces in the Most Favorable Orientation in the Presence of Explicit Water Molecules (Polar Residues Are Underlined) Eint (kJ/mol)
EVdW int (kJ/mol)
EES int (kJ/mol)
PEG-HDI
335
257
77
CO-HDI
397
346
51
NAA
adsorbed residues Pro22 Gln24 Gly25 Trp26 Met27 Asp53 Thr54 Arg55 Thr56 Ser57 Tyr58 Arg59 Trp64 Lys67 Asp68 Arg70 Ala1 Glu2 Lys3 Phe5 Asp6 His7 Ala8 Ala9 Gln24 Glu36 Gly37 Ser38 Gly39 Arg40 Ile41 Gln52 Arg70 Asn72 Leu73 Lys87 Glu89 Arg90 His91 Thr92
16 24
Table 2. Change in the β-Sheets upon Adsorption
protein would not exist on a hydrophobic surface. For instance, in a recent study on BSA adsorption on a graphite surface it was concluded that an implicit model for the solvent is an efficient way to model the adsorption process.38 However, when the hydrophobic surface is not as smooth as a graphite surface, the interplay between the water, protein, and surface, which results in entrapment of water molecules between the protein and surface may lead to significantly different adsorption behavior based on the use of implicit or explicit solvent models. The protein-PEG-HDI interaction energy is as high as the protein-CO-HDI interaction energy considering the smaller number of adsorbed amino acids. This may be attributed to the stronger electrostatic interactions between the polar residues of the fibronectin fragment and the polar surface. The analyses of adsorbed residues listed in Table 2 show that for both surfaces
broken formed net change Σ absolute change
PEG-HDI
CO-HDI
−1 27 26 28
−4 17 13 21
the ratio of the polar residues to total number of adsorbed residues is approximately 2/3. On both surfaces, the positively charged linear Arg and Lys are the most strongly bound residues. The hydrophobic Trp and Phe have also high affinity to the surfaces (PEG-HDI and CO-HDI, respectively), possibly due to their large aromatic groups leading to strong van der Waals interactions. The difference in the protein affinity of the two surfaces is therefore due to their interactions with the 12625
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
Figure 10. Overall angular deformations of fibronectin residues after adsorption on PEG-HDI and CO-HDI.
angular deformation parameter (Δψ + ΔΦ)1/2 for each residue. Figure 10 compares overall deformation for both surfaces and shows that the deformations are slightly larger and more localized on the protein backbone when adsorbed on the PEGHDI surface with respect to the CO-HDI surface. This agrees with previous analyses based in Table 2 and Figure 9, indicating that the residues located at both ends of the protein fragment were adsorbed on the CO-HDI surface, providing better and stronger contact as compared to the PEG-HDI surface. Furthermore, the delocalized adsorption of the residues on the hydrophobic surface resulted in a decrease in the radius of gyration of the protein down to 14.4, which had a value of 17.1 before adsorption. On the other hand, the radius of gyration remained almost unchanged after adsorption on the hydrophilic surface with the value of 16.2.
hydrophobic Ala and Phe residues, which are concentrated at the N-terminal end of the protein fragment as shown in Figure 9. The three aliphatic Ala residues and Phe with the aromatic ring promote better adhesion to the hydrophobic surface with respect to the hydrophilic one. β-Sheet Formation/Breaking Affect Strain Energies. Figure 9 shows also the changes in the secondary structure due to surface adsorption where the formation of β-sheets upon adsorption can be clearly seen. This observation agrees with the analysis of the formation and breaking of individual β-sheets using the DSPP software.36 The changes of β-sheets of the protein after adsorption on both surfaces are given in Table 2. Total number of absolute (sheet formation + sheet breaking) changes was higher upon adsorption on the PEG-HDI surface, which correlates with the magnitude of the absolute strain energies, indicating that the structural changes of the fibronectin due to adsorption is more significant on the PEGHDI surfaces than the CO-HDI surface. The structure changes reflect also to the strain energies, which were calculated as 1.17 and 0.58 MJ/mol for the protein fragment after adsorption on PEG-HDI and CO-HDI surfaces, respectively. Similar to the case of the interaction energies, the strain energies differ significantly from the values calculated with the implicit water, where the strain energy for the PEG-HDI surface was lower than that for the CO-HDI surface: 100 and 123 kJ, respectively. Deformations of ψ and Φ angles of the amino acids upon adsorption were analyzed during the simulations. The magnitudes of the average deformations in ψ and Φ angles, Δψ and ΔΦ, with respect to the soaked protein structure conformation were compared by using a user-defined overall
4. CONCLUSIONS In this study, crystalline poly(ethylene glycol)−hexamethylene diisocyanate (PEG-HDI) and amorphous castor oil−hexamethylene diisocyanate (CO-HDI) polyurethanes were modeled in order to gain insight into the adsorption behavior of type 1 module of the blood protein fibronectin on hydrophilic and hydrophobic surfaces. The protein surface is heterogeneous with hydrophobic and hydrophilic regions. Hence, protein orientation has significant effect on protein−surface interactions and on the adsorption behaviors. Therefore, molecular mechanics calculations were employed in the first stage to determine the preferred orientation of fibronectin on the surfaces by calculating the protein−surface interaction energies. These calculations 12626
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
(4) Aggarwal, P.; Hall, J. B.; Mcleland, C. B.; Dobrovolskaia, M. A.; McNeil, S. E. Nanoparticle interaction with plasma proteins as it relates to particle biodistribution, biocompatibility and therapeutic efficacy. Adv. Drug Delivery Rev. 2009, 61, 428−437. (5) Somasundaran, P.; Hubbard, A. T. Encyclopedia of Surface and Colloid Science, 2nd ed.; Taylor & Francis: New York, 2006. (6) Agashe, M.; Raut, V.; Stuart, S. J.; Latour, R. A. Molecular simulation to characterize the adsorption behavior of a fibrinogen gamma-chain fragment. Langmuir 2005, 21, 1103−1117. (7) Braun, R.; Sarikaya, M.; Schulten, K. Genetically engineered goldbinding polypeptides: structure prediction and molecular dynamics. J. Biomater. Sci., Polym. Ed. 2002, 13, 747−757. (8) Dalal, P.; Knickelbein, J.; Haymet, A. D. J.; Sönnichsen, F. D.; Madura, J. D. Hydrogen bond analysis of Type 1 antifreeze protein in water and the ice/water interface. Phys. Chem. Commun. 2001, 4, 32− 36. (9) Raut, V. P.; Agashe, M. A.; Stuart, S. J.; Latour, R. A. Molecular dynamics simulations of peptide-surface interactions. Langmuir 2005, 21, 1629−1639. (10) Cole, D. J.; Payne, M. C.; Ciacchi, L. C. Water structuring and collagen adsorption at hydrophilic and hydrophobic silicon surfaces. Phys. Chem. Chem. Phys. 2009, 11, 11395−11399. (11) Kang, Y.; Li, X.; Tu, Y.; Wang, Q.; Agren, H. On the mechanism of protein adsorption onto hydroxylated and nonhydroxylated TiO2 surfaces. J. Phys. Chem. C 2010, 114, 14496−14502. (12) Shen, J. W.; Wu, T.; Wang, Q.; Pan, H. H. Molecular simulation of protein adsorption and desorption on hydroxyapatite surfaces. Biomaterials 2008, 29, 513−532. (13) Hagiwara, T.; Sakiyama, T.; Watanabe, H. Molecular simulation of bovine beta-lactoglobulin adsorbed onto a positively charged solid surface. Langmuir 2009, 25, 226−234. (14) Kubiak-Ossowska, K.; Mulheran, P. A. Multiprotein interactions during surface adsorption: a molecular dynamics study of lysozyme aggregation at a charged solid surface. J. Phys. Chem. B 2011, 115, 8891−8900. (15) Starzyk, A.; Cieplak, M. Denaturation of proteins near polar surfaces. J. Chem. Phys. 2011, 135, 235103. (16) Cormack, A. N.; Lewis, R. J.; Goldstein, A. H. Computer simulation of protein adsorption to a material surface in aqueous solution: biomaterials modeling of a ternary system. J. Phys. Chem. B 2004, 108, 20408−20418. (17) Madura, J. D.; Wierzbicki, A.; Harrington, J. P.; Maughon, R. H.; Raymond, J. A.; Sikes, C. S. Interactions of the D- and L-forms of winter flounder antifreeze peptide with the {201} planes of ice. J. Am. Chem. Soc. 1994, 116, 417−418. (18) Raffaini, G.; Ganazzoli, F. Molecular dynamics simulation of the adsorption of a fibronectin module on a graphite surface. Langmuir 2004, 20, 3371−3378. (19) Raffaini, G.; Ganazzoli, F. Protein adsorption on the hydrophilic surface of a glassy polymer: a computer simulation study. Phys. Chem. Chem. Phys. 2006, 8, 2765−2772. (20) Zheng, J.; Li, L.; Tsao, H.; Sheng, Y.; Chen, S.; Jiang, S. Strong repulsive forces between protein and oligo (ethylene glycol) selfassembled monolayers: a molecular simulation study. Biophys. J. 2005, 89, 158−166. (21) Wei, T.; Mu, S.; Nakano, A.; Shing, K. A hybrid multi-loop genetic-algorithm/simplex/spatial-grid method for locating the optimum orientation of an adsorbed protein on a solid surface. Comput. Phys. Commun. 2009, 180, 669−674. (22) Mungikar, A.; Forciniti, D. Effect of co-solvents on the adsorption of peptides at the solid-liquid interface. Biomacromolecules 2006, 7, 239−251. (23) Noinville, V.; Vidalmadjar, C.; Sebille, B. Modeling of protein adsorption on polymer surfaces. Computation of adsorption potential. J. Phys. Chem. 1995, 99, 1516−1522. (24) Zhdanov, V. P.; Rechendorff, K.; Hovgaard, M. B.; Besenbacher, F. Deposition at glancing angle, surface roughness, and protein adsorption: Monte Carlo simulations. J. Phys. Chem. B 2008, 112, 7267−7272.
neglected the competitive adsorption of the protein and water molecules on the hydrophilic PEG-HDI surface, since they were carried out using the implicit solvent model. Consequently, the crystalline structure of the PEG-HDI surface promoted adsorption by providing an energetically more efficient contact with the protein, as opposed to the rougher surface of the amorphous CO-HDI. Next, the most preferred orientations were used for a more realistic treatment of the adsorption process via MD simulations using explicit water molecules. In this case, competitive adsorption between the protein and water molecules disfavors protein adsorption on the hydrophilic PEG-based polyurethane due to competitive binding of water molecules, in agreement with expectations. The hydrophobic residues concentrated at the N-terminal end of the fibronectin fragment were mainly responsible on the adsorption characteristics of the protein on surfaces with different hydrophilicity. However, the relatively small difference on the surface−protein interaction energies for both surfaces underlined the interplay between the surface hydrophilicity and the surface roughness on protein adsorption. The simulations provide useful insight into understanding these complex interactions, which cannot be easily identified directly by experiments, bearing in mind that the choice of force field and other approximations are significant considerations in interpreting the results, as in most of the simulation studies. The understanding of these interactions will help to tailor polyurethane surfaces by varying the compositions of PEG and CO components in the copolymer to optimize biocompatibility, swelling resistance, and mechanical performance.
■
ASSOCIATED CONTENT
* Supporting Information S
Figures S1−S4 and Tables S1−S4. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Prof. F. Seniha Güner from Istanbul Technical University is greatly acknowledged for sharing unpublished experimental data and for helpful discussions. This work is supported in part by Istanbul Technical University through the Scientific Research Fund. High-performance computing resources used in this work were provided by the National Center for High Performance Computing of Turkey (UYBHM) under Grant 1001342011.
■
REFERENCES
(1) Puleo, D. A.; Biziosi, R. Biological Interactions on Materials Surfaces: Understanding and Controlling Protein, Cell, and Tissue Responses; Springer: New York, 2009. (2) Dee, K. C.; Puleo, D. A.; Bizios, R. An Introduction to TissueBiomaterial Interactions; Wiley: Hoboken, NJ, 2002. (3) Cedervall, T.; Lynch, I.; Foy, M.; Berggård, T.; Donnelly, S. C.; Cagney, G.; Linse, S.; Dawson, K. A. Detailed identification of plasma proteins adsorbed on copolymer nanoparticles. Angew. Chem., Int. Ed. 2007, 46, 5754−5756. 12627
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628
Langmuir
Article
(25) Corneillie, S.; Lan, P. N.; Schacht, E.; Davies, M.; Shard, A.; Green, R.; Denyer, S.; Wassall, M.; Whitfield, H.; Choong, S. Polyethylene glycol-containing polyurethanes for biomedical applications. Polym. Int. 1998, 46, 251−259. (26) Lamba, N. M. K.; Woodhouse, K. A.; Cooper, S. L. Polyurethanes in Biomedical Applications; CRC Press: Boca Raton, FL, 1998. (27) Shelke, N. B.; Sairam, M.; Halligudi, S. B.; Aminabhavi, T. M. Development of transdermal drug-delivery films with castor-oil-based polyurethanes. J. Appl. Polym. Sci. 2007, 103, 779−788. (28) Sirkecioglu, A.; Mutlu, H.; B., Citak, C.; Koc, A.; Guner, F. S. Physical and surface properties of polyurethane hydrogels in relation with their chemical structure. Int. Polym. Sci. 2012, submitted. (29) Williams, M. J.; Phan, I.; Harvey, T. S.; Rostagno, A.; Gold, L. I.; Campbell, I. D. Solution structure of a pair of fibronectin type 1 modules with fibrin binding activity. J. Mol. Biol. 1994, 235, 1302− 1311. (30) Sun, H. COMPASS: An ab initio force-field optimized for condensed-phase applications - Overview with details on alkane and benzene compounds. J. Phys. Chem. B 1998, 102, 7338−7364. (31) Heuchel, M.; Hofmann, D.; Pullumbi, P. Molecular Modeling of small-molecule permeation in polyimides and its correlation to freevolume distributions. Macromolecules 2004, 37, 201−214. (32) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank. Nucleic Acids Res. 2000, 28, 235−242. (33) Dauber-Osguthorpe, P.; Roberts, V. A.; Osguthorpe, D. J.; Wolff, J.; Genest, M.; Hagler, A. T. Structure and energetics of ligand binding to proteins: E. coli dihydrofolate reductase-trimethoprim, a drug-receptor system. Proteins: Struct., Funct., Genet. 1988, 4, 31−47. (34) Bhowmik, R.; Katti, K. S.; Katti, D. Molecular modeling of polyacrylic acid-hydroxyapatite interface. Polymer 2007, 48, 664−674. (35) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J. Intermolecular Forces; Pullman, B., Ed.; Reidel: Dordrecht, Holland, 1981; pp 331−342. (36) Kabsch, W.; Sander, C. Dictionary of protein secondary structure - pattern-recognition of hydrogen-bonded and geometrical features. Biopolymers 1983, 22, 2577−2637. (37) Raffaini, G.; Ganazzoli, F. Understanding the performance of biomaterials through molecular modeling: crossing the bridge between their intrinsic properties and the surface adsorption of proteins. Macromol. Biosci. 2007, 7, 552−566. (38) Mücksch, C.; Urbassek, H. M. Molecular dynamics simulation of free and forced BSA adsorption on a hydrophobic graphite surface. Langmuir 2011, 27, 12938−12943.
12628
dx.doi.org/10.1021/la301546v | Langmuir 2012, 28, 12619−12628