Adsorption of Bovine Serum Albumin on Poly(vinylidene fluoride

Jan 24, 2018 - (11) Although pretreatment of water, including coagulation, activated carbon adsorption, and ozone oxidation, is used to prevent membra...
0 downloads 11 Views 6MB Size
Article Cite This: J. Phys. Chem. B 2018, 122, 1919−1928

pubs.acs.org/JPCB

Adsorption of Bovine Serum Albumin on Poly(vinylidene fluoride) Surfaces in the Presence of Ions: A Molecular Dynamics Simulation Abdul Rajjak Shaikh,*,†,§ Hamed Karkhanechi,†,∥ Tomohisa Yoshioka,‡ Hideto Matsuyama,*,† Hiromitsu Takaba,⊥ and Da-Ming Wang# †

Center for Membrane and Film Technology, Department of Chemical Science and Engineering, and ‡Center for Membrane and Film Technology, Graduate School of Science, Technology and Innovation, Kobe University, 1-1 Rokkodai, Nada-ku, Kobe 657-8501, Japan § Department of Chemistry, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia ∥ Chemical Engineering Department, Faculty of Engineering, Ferdowsi University of Mashhad, Mashhad 9177948974, Iran ⊥ Department of Environmental and Energy Chemistry, Faculty of Engineering, Kogakuin University, Hachioji, Tokyo 192-0015, Japan # Department of Chemical Engineering, National Taiwan University, Taipei 10617, Taiwan S Supporting Information *

ABSTRACT: Adsorption of bovine serum albumin (BSA) on poly(vinylidene fluoride) (PVDF) surfaces in an aqueous environment was investigated in the presence and absence of excess ions using molecular dynamics simulations. The adsorption process involved diffusion of protein to the surface and dehydration of surface−protein interactions, followed by adsorption and denaturation. Although adsorption of BSA on PVDF surface was observed in the absence of excess ions, denaturation of BSA was not observed during the simulation (1 μs). Basic and acidic amino acids of BSA were found to be directly interacting with PVDF surface. Simulation in a 0.1 M NaCl solution showed delayed adsorption of BSA on PVDF surfaces in the presence of excess ions, with BSA not observed in close proximity to PVDF surface within 700 ns. Adsorption of Cl− on PVDF surface increased its negative charge, which repelled negatively charged BSA, thereby delaying the adsorption process. These results will be helpful for understanding membrane fouling phenomena in polymeric membranes, and fundamental advancements in these areas will lead to a new generation of membrane materials with improved antifouling properties and reduced energy demands.

1. INTRODUCTION Water scarcity has been a major global problem, and obtaining water that is drinkable or suitable for irrigation and agricultural use remains a major challenge.1,2 Recently, water treatment research has grown considerably, offering new hope for producing clean water. Of the various techniques used for water treatment, membrane filtration techniques are gaining significant attention owing to their ability to filter suspended solid and organic compounds, as well as inorganic contaminants, such as heavy metals, ease of operation, and small footprint. Depending on the size of particles to be retained, various types of membrane filtration processes, including microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and osmosis (reverse osmosis (RO) and forward osmosis (FO)), can be employed for wastewater treatment. However, membrane fouling is a severe problem encountered in all membrane filtration techniques, and it is a major factor in determining the practical application of these techniques in water and wastewater treatment and desalination in terms of technology and cost.3−6 © 2018 American Chemical Society

Membrane fouling occurs by accumulation and deposition of foulants (rejected compounds) on membrane surfaces or inside membrane pores.7 Effective and efficient methods are required to control and minimize fouling. To date, numerous efforts have been dedicated to control membrane fouling to enable more widespread use of membrane technology.8−10 Fouling may be prevented or minimized by methods such as pretreatment of feed streams, chemical modification to improve membrane antifouling properties, and optimization of operating conditions. Despite several experimental studies on membrane fouling and antifouling properties, an understanding of the mechanism of membrane fouling remains elusive.11 Although pretreatment of water, including coagulation, activated carbon adsorption, and ozone oxidation, is used to prevent membrane fouling,12 most membranes are hydrophobic and modification of membrane surfaces is a useful strategy to control fouling. Received: October 15, 2017 Revised: January 23, 2018 Published: January 24, 2018 1919

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B

phase is the most common in membranes for water treatment. However, PVDF is moderately hydrophobic (water contact angle of ∼89°), which is a drawback for specific applications. The intrinsic hydrophobicity of PVDF membranes causes easy adsorption of biomacromolecules and organic matter on the membrane surface, resulting in membrane fouling.31 Bovine serum albumin (BSA),32 humic acids, and sodium alginate are commonly used as model foulants in experimental studies to understand membrane fouling. Many experimental studies have been dedicated to understanding the adsorption of BSA on PVDF surfaces,33 mainly examining the amount of protein adsorbed and the secondary structural evolution of adsorbed proteins. It has been widely hypothesized that the hydrophobic part of a PVDF membrane interacts with the hydrophobic part of the protein.34 However, the detailed molecular mechanism of membrane fouling on PVDF is still not clearly understood. Adsorption of BSA at different substrates has been studied using atomistic MD simulations.35−37 Recently, KubiakOssowska et al.38,39 have studied the adsorption of BSA on negatively charged silica surfaces. However, although there have been a number of simulations of BSA adsorption at different substrates, an atomistic understanding of this phenomenon is lacking on polymeric membranes, and it is the purpose of this paper to provide this insight. Ions are present in high and low concentrations in surface water, municipal wastewater, industrial wastewater, and other wastewater or water to be treated.40 In fact, the ionic strength of seawater can be higher than 400 mM.41 Ionic strength has been identified as a crucial factor that can seriously affect the fouling rate in membrane operation.42 Despite several experimental studies, the effect of ionic strength on membrane fouling is confusing. Some reports indicate that fouling increases with increase in ionic concentration,43−45 whereas other reports indicate that fouling decreases with increase in ionic strength.46−48 Recently Miao et al.49 have studied the fouling behavior of PVDF surfaces by varying ionic concentrations and reported that increase in ionic strength decreases adsorption of BSA on membrane and thereby reduces fouling of membrane. The effect of ions on adsorption of protein molecules on membrane surface is also not clearly understood. So far, no computational studies have been reported to understand fouling of PVDF membrane in the presence of BSA. Therefore, obtaining an in-depth insight into the effect of ionic strength on adsorption of protein behavior could lead to the development of strategies for fouling control and mitigation, which is crucial for the successful and widespread application of membrane technology in the field of wastewater treatment. As protein-like substances have been identified as a major cause of membrane fouling, understanding the key factors that influence protein fouling phenomena is crucial for controlling and mitigating membrane fouling. MD simulations are an insightful approach for studying the dynamics of protein adsorption on a membrane surface and dynamic interplay between membrane surfaces and foulant molecules as they undergo adsorption. Because of the unique ability of MD simulations to translate atomic-scale interactions and processes into a dynamic visual experience, it is anticipated that computational chemistry will play an ever-greater role in the coming years in helping to answer these and related questions about membrane structure and function. So far, interactions between BSA and PVDF surface have not yet been studied using MD simulations owing to the large size of BSA

Several membrane modification methods, such as plasma modification, radiation grafting modification, surface coating modification, and material blending modification, are also widely used.8 Despite these experimental studies, membrane fouling remains a major challenge in membrane technology. Several computational molecular modeling studies have also been reported to understand the solid surface−biomolecular interactions and or membrane−foulant interactions. These techniques have been successfully applied to develop a better molecular-level understanding of various properties and processes in modern chemical, biological, and material sciences. Molecular dynamics (MD) studies have been widely conducted for understanding membrane fouling phenomena. Ahn et al. studied the adsorptive fouling of a polyethersulfone membrane by natural organic matter (NOM) in the presence of Ca2+ and Mg2+ ions using experimental and MD simulations.13 They observed that divalent ions may cause membrane fouling by promoting aggregation of NOM molecules in solution. Sutton et al.14 studied hydration of NOM and interactions with Na+ and Ca2+ and observed a different affinity of ions with NOM molecules. Moreover, several other studies have been carried out to study interactions between NOM and other ions.15−17 Wei et al. studied the adsorption of lysozyme on polyethylene (PE) surfaces using MD simulations and observed the dehydration of hydrophobic surfaces before adsorption of lysozyme on PE surfaces.18,19 Further, MD simulations of the interactions between protein and solid surfaces were reviewed by Szott and Horbett.20 Hughes and Gale21 carried out MD simulations for polyamide membranes with three different foulants (glucose, phenol, and oxygen) and observed that each of the three foulants interact with the membrane in a different manner. It is found that accumulation of glucose and phenol occurs at the membrane−saline solution due to the favorable nature of the interaction in this region and that the presence of these foulants reduces the rate of flow of water molecules over the membrane−solution interface. Several other computational and theoretical studies aided with experimental studies were carried out to understand membrane fouling22−26 and the effect of metals/ions on it. Recently, Ridgway et al.27 have reviewed MD simulation methods of reverse osmosis membrane structure and elucidated the use of atomistic MD simulation for membrane−foulant interactions. With increasing computational technology and resources, now it can be feasible to study interactions of large protein molecules with membrane surfaces to understand and visualize membrane−protein interactions. Typical polymers used in the water industry include polypropylene, polysulfone, polyacrylonitrile, and poly(vinylidene fluoride) (PVDF).28 PVDF has received considerable attention in numerous applications owing to its outstanding properties, such as high mechanical strength, thermal stability, and chemical resistance. PVDF is an attractive material in membrane technologies and is widely used in UF/ MF, membrane distillation (MD), pervaporation, and other processes.29 PVDF forms at least five crystalline phases, including α, β, γ, δ, and ε phases.30 The α phase has a TGTG′ (T: trans, G: gauche+, and G′: gauche−) dihedral conformation, and the β phase has an all-trans conformation. These phases are of most interest because the α phase is thermodynamically stable, whereas the β phase is kinetically stable at room temperature and pressure. β-PVDF exhibits good piezoelectric and pyroelectric properties. In the α phase, the polymeric chains show nonpolar characteristics, and this 1920

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B

Figure 1. (a) Secondary structure of bovine serum albumin. Cysteine salt bridges (yellow) are shown in licorice representation. (b) Initial structure of the molecular system for the adsorption simulation of BSA on a PVDF surface. BSA, shown in cartoon style, is colored mauve. Sodium ions are shown in blue.

(molecular weight, 69 324 Da)32 and PVDF surfaces. In this study, we carried out all-atom MD simulations to examine molecular interactions between PVDF and BSA. The effect of ions on the adsorption of BSA on PVDF surfaces was also studied. To the best of our knowledge, there are no previous reports on the adsorption of BSA on PVDF membrane surfaces using computational methods. The results will be helpful for understanding membrane fouling phenomena and explaining experimental results.

out for 5 ns. Similarly, the temperature was decreased to 300 K by decrements of 10 K. Production simulations were performed using the NPT ensemble for 30 ns. The temperature was maintained at 300 K using velocity rescaling with a coupling time of 0.1 ps. The pressure was maintained at 1 atm for NPT simulations using a Parrinello−Rahman barostat with a coupling time of 2 ps. Equations of motion were integrated using the leapfrog algorithm with a time step of 2.0 fs. The total electrostatic interactions were evaluated using the particle mesh Ewald summation. Coulomb and van der Waals cutoffs of 1.0 nm were employed. Periodic boundary conditions in all directions were employed to mimic the bulk behavior. Bond lengths with hydrogen were constrained with the LINCS algorithm. The bulk density obtained by simulation is 1.72 g/cm3, which is close to the experimental density of 1.68 g/cm3.57 2.2.2. PVDF Surface in Solvent. To simulate the system in the presence of water, we greatly increased the cell axis perpendicular to the surface of interest (z axis), leaving an empty space above and below the PVDF surface. The PVDF surface was initially soaked in water and minimization was carried out. For water, the TIP3P water model was used. After minimization, short NVT dynamics was carried out to relax the water molecules. Further, simulation was carried out to equilibrate the structure. The z axis was increased so that the protein could also be accommodated. 2.2.3. PVDF−BSA System. To prepare the PVDF−BSA system, we used the BSA crystal structure from the protein data bank (PDB id 3V03).32 Chain A of 3V03 with all disulfide bridges (17 disulfide bridges, Table S2) was used as the starting structure in our simulation. The BSA structure contains a total of 583 amino acid residues (note the first two residues, namely, Asp1 and Thr2, both uncharged and hydrophilic, are missing, but these do not affect the protein’s overall structure or properties). Several missing side-chain atoms were rebuild using the Swiss PDB Viewer58 program. The protonation state of each residue was checked using MolProbity program.59 Hydrogen atoms were added to the protein at neutral pH. The secondary structure of BSA is shown in Figure 1a. The overall

2. COMPUTATION DETAILS 2.1. Force Field. The chemical formula for PVDF is −[CH2−CF2]n−, where n is the number of repeat units. The PVDF monomer unit was built using the visual molecular dynamics program. Geometry optimization was carried out using the Gaussian 03 program.50 Density functional theory calculations were performed using B3LYP and 6-31G(d) basis sets. The electrostatic potential surface was generated by the Merz−Kollman method at the HF/6-31G(d) level of theory, followed by multiconfigurational two-stage restrained electrostatic potential fitting using the RED IV program.51 The generalized AMBER force field (GAFF)52 was used to determine the intra- and intermolecular force constants for the PVDF unit. The PVDF surface was constructed by replicating the crystalline vinylidene fluoride (α-form) unit cell in the three Cartesian directions. The slab consists of a total of 320 polymer chains in the simulation box so that each polymer chain has 20 (−CH2CF2)20 monomeric units. The atoms near the edges of the periodic box in the y direction were covalently bonded with the atoms of their periodic images, mimicking infinitely long chains.18,19,53 2.2. Simulation Procedure. 2.2.1. Preparation of PVDF Surface. All of the simulations were performed with the GROMACS 4.6 simulation program.54 The configurations were first optimized using the energy-minimization routine. Simulation details are similar to our previously reported work.55,56 The temperature was gradually increased by increments of 10 K up to 500 K, and each NVT and NPT simulation was carried 1921

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B charge of the protein at pH 7 is −17e; however, there are numerous negatively charged Asp and Glu residues, as well as positively charged residues, such as Arg and Lys, in the protein so that its surface-charge distribution is inhomogeneous, as shown in Table S3.32 The N and C termini were capped uncharged. The AMBER99sb force field,60 which is known as a good force field for biomolecules, was used to model the parameters of BSA. In the following step, the system was assembled with one BSA molecule, leaving a gap of at least 15 Å between the PVDF surface and the protein. Any water molecules within 3 Å of BSA were removed. To prevent adsorption on the periodic image of the surface, a sufficiently large system was chosen with a 20 Å thick layer of water on top of the protein and a 27 Å thick layer below the PVDF surface. In addition, box size was kept such that a minimum distance of at least 20 Å between protein and y axes remains so that protein cannot interact with its mirror image. The system was neutralized by adding 16 sodium ions. Figure 1b shows the initial setup of the PVDF−BSA system. Water molecules were relaxed with a short NVT simulation at 300 K. In the next step, BSA and water molecules were relaxed under NPT conditions at a pressure of 1 bar and a temperature of 300 K for 50 ns. Equilibration of the PVDF−BSA system was carried out using the NVT ensemble for 300 ns. Finally, production runs were carried out using an NVT simulation for 700 ns at 300 K with a 2 fs time step. It is advisable to do several simulations with different initial velocities to get statistical average; we chose to use single trajectory for longer time to study the adsorption process, which mostly depends on time. As we are using allatom force field and have not used any constrain, it will be too expensive to do several simulations under different initial conditions for longer time. We also carried out MD simulations in the presence of 0.1 M of NaCl. All of the simulation details were similar to those reported above. Similarly to the previous system, equilibration was carried out for 300 ns and production run was carried out for 700 ns. The number of atoms used in simulations is shown in Table S1.

Figure 2. Adsorption of BSA on a PVDF surface for PB and PBI systems, as monitored by the protein−surface distance.

Figure 3. Interaction energy between PVDF and BSA in PB and PBI systems.

adsorption of BSA occurring at ∼315 ns. Further, BSA adapts to the PVDF surface in such a way that the interaction energy decreases. The lowest interaction energy is observed at 500 ns, and the interaction energy varies on the basis of direct contact with the PVDF surface. For the PBI system, the interaction energy is only observed within the first 100 ns, after which no interaction is observed between PVDF and BSA. During equilibration, the distance between BSA and the PVDF surface varies from 10 to 15 Å. In the production simulation, the distance between PVDF and the protein is up to 15 Å during the first 100 ns, indicating a low interaction energy. After 100 ns, the protein starts to move away from the PVDF surface, with the distance varying from 18 to 36 Å, and no interaction is observed after 125 ns. These results indicate that BSA was repelled from the PVDF surface after 100 ns, and it was not located close to surface within 700 ns. The distance and interaction energy results (Figures 2 and 3) show that adsorption occurs in the PB system, but adsorption does not occur in the PBI system during the specified time, that is, adsorption is delayed in the PBI system. The diffusion coefficients of the protein, ions, and water molecules in solution, as calculated using mean square deviation from the trajectory over the first 100 ns (after equilibration), are given in Table 1. In the PB system, the diffusion coefficient for the protein is lower than that in the PBI system. This result is expected as the protein is adsorbed on the PVDF surface in the PB system. Diffusion of water is also

3. RESULTS AND DISCUSSION MD simulations were carried out to examine the adsorption of BSA on PVDF surface and to analyze PVDF−BSA interactions in detail. Simulations were carried out in the presence of water, and ions were added to neutralize the charges. Hereafter, we refer to the system without extra ions as the “PB” (PVDF− BSA) system and the system in the presence of 0.1 M NaCl solution as the “PBI” (PVDF−BSA−ions) system. First, we analyzed the closest distance between PVDF (both upper and lower surface) and BSA in these systems. The distance of BSA from the surface is shown in Figure 2. For the PB system, the distance between PVDF and BSA decreased after 300 ns, indicating adsorption of BSA on the surface. After adsorption, the protein was found to be within 2.5 Å of the PVDF surface. The distance between the PVDF surface and BSA increased continuously after 100 ns in the PBI system. Hence, adsorption does not occur within 700 ns. The movement of the protein molecule away from the surface indicates that the protein prefers to be in bulk solution rather than adsorbed on the surface. The interaction energy between PVDF and BSA during the 700 ns production simulation was calculated in both the PB and PBI systems, as shown in Figure 3. In the PB system, BSA is observed to have a favorable interaction with PVDF, with 1922

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B Table 1. Diffusion Coefficients of the Protein, Water, and Ions diffusion coefficients/10−7 cm2/s system

protein

water

Na+

Cl−

PB PBI

0.043 ± 0.14 3.87 ± 5.90

421.71 ± 45.87 495.38 ± 21.30

84.23 ± 16.97 202.32 ± 12.15

248.57 ± 29.38

In the PB system, in the beginning of the simulation, BSA diffuses freely above the PVDF surface. Around 100 ns, Lys316 and His366 come closer to the PVDF surface. It further desorbs from the surface and after that several atoms adsorb on the surface using the Lys316 residue. Further, BSA reorients and approaches the surface in which Asp56, Glu57, and Ser58 come near to the surface at 316 ns. Further, other residues, such as Lys362, Asp363, and Glu320, come closer to the surface from other end. With correct orientation, the extended Lys side chains (Lys232) act as middle pillar. It creates strong anchors to the surface. The protein remains in a stable adsorbed state. Attachment of BSA to the PVDF surface through these two points then remains constant. There were at least six residues within the hydrogen-bonding distance with the surface. Previous studies carried out by Kurrat et al.,61 who studied BSA adsorption kinetics on silica−titania surfaces, also concluded that around four to five H-bonds created between BSA and the surface (or surface water) are enough to immobilize the BSA on the surface. In case of PBI, His366 comes closer to the surface at 100 ns. However, after that, it desorbs from the surface probably because the orientation with respect to the surface is not optimum. Figure 5b shows that His366 does not remain near to the PVDF surface at 300 ns. It was repelled from the surface, as shown in the further snapshot. Subsequently, the orientation of BSA changed several times, but it did not return to the surface. BSA has three primary domains (I, II, and III) that are arranged in a heart shape with 17 disulfide-bond linkages that stabilize the domains. Each domain can be divided into two subdomains (A and B).32,62 Table S3 shows the amino acids found in each domain of BSA. We calculated the distances between the centers of mass of the various domains during the simulations (Figure S2). As shown in Figure S2A, in the PB system, a minimum is observed in the distance between domains II and III at 300 ns. These domains come closer together during adsorption of the protein on the surface. However, after adsorption, these protein domains return to their initial positions. At ∼100 ns, domains I and II are in close proximity as the distance between these domains decreases from 7.9 to 2 nm. Further, the distance between domains I and II changed from 7 to 6 nm after 400 ns. As shown in Figure S3B, the distances between the different domains in the PBI system showed large fluctuations throughout the simulation. These analyses indicate that ions also have a significant effect on internal domain interactions, and this effect varies with time. Changes in the root-mean-square deviation (RMSD) for different domains of the protein are shown in Figure 6. In the PB system, the RMSD varies between 0.10 and 0.40 nm (Figure 6a). The highest RMSD is observed for domain I-A, and it increases considerably after 550 ns. Further analysis revealed that this domain is in direct contact with the PVDF surface. Domain II-B, which is also adsorbed on the PVDF surface, shows a similar increase in RMSD after adsorption. The average RMSD with respect to time is shown in Figure S3. The average RMSD for BSA between 300 and 500 ns is higher when adsorption of BSA on the PVDF surface occurs. At 500 ns, after

different in the PB and PBI systems, with a higher diffusion coefficient observed for water in the PBI system. This difference results from stronger interactions with ions in the PBI system. Sodium ions have a much lower diffusion coefficient in the PB system. As the protein is negatively charged, the Na+ ions can form electrostatic interactions with the protein. However, in the PBI system, sodium and chlorine ions have similar diffusion coefficients. Overall, the presence of excess ions had a significant influence on the diffusion coefficients of the protein, ions, and water. The orientation of the protein relative to the PVDF surface was analyzed. The angle is defined as the angle between the unit vector normal to the PVDF surface and the vector between the Cα atoms of Ser58 and His366 (Figure 4). These residues

Figure 4. Orientation of BSA with respect to the surface of PVDF for PB and PBI systems.

were consistently close to the surface. Beginning with the protein above the surface, we analyzed all trajectories until the protein either fully reached the surface with a stable orientation or failed to adsorb, as in the PBI system. Adsorption occurred after 300 ns, indicating the longer time requirement for removal of water molecules trapped between BSA and the PVDF surface, which is often accompanied by rotation and, in some cases, deformation of the protein. Overall, with increase in ionic strength, the PVDF−BSA interaction forces are weakened due to increase in hydration repulsion force, which is accompanied by increased diffusion coefficient of BSA.49 As shown in Figure 4, drastic changes in the orientation of BSA in the PBI system were observed throughout the simulation. In the PB system, the orientation of BSA varies between 140 and 50° up to 300 ns and subsequently remains constant within the range of 40−90°. Thus, after adsorption at 300 ns, BSA orientation is restricted. A visual inspection of these systems was conducted by determining the orientation of BSA using the angle calculated between the vectors of the Cα atoms of Ser58−His366 and Ser190−Ala569. The representative snapshots are shown in Figure 5a for the PB system and Figure 5b for the PBI system. Other representative orientations are also shown in Figure S1. 1923

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B

Figure 5. Orientation of a BSA molecule on the PVDF surface at different simulation times for (a) PB and (b) PBI systems. Ser58 (yellow), Ser190 (green), His366 (red), and Ala569 (blue) residues are shown in the van der Waals model. The angles between the vectors of Ser58−His366 and Ser190−Ala569 are shown within parentheses.

the adsorption process, the average RMSD again decreases. Overall, the BSA structural elements are maintained after adsorption, and no significant α-helix unfolding or domain reorganization is observed. Figure S3 indicates that the structure is stable during the entire adsorption trajectory. Simulation for adsorption of BSA on negatively charged silicate also confirmed that BSA retains its overall structure even after adsorption. The PBI system shows only a slight variation, with RMSD values between 0.12 and 0.25 nm (Figure 6b). The average RMSD shows a slight increase up to 450 ns, which then briefly decreases before increasing again (Figure S3). Overall, both the PB and PBI systems do not show abrupt changes of the protein configuration and the structure is stable. The abundance of ions around the protein and the PVDF surface in the PBI system was analyzed using distance criteria. Ions within 4 Å of the protein or PVDF surface were analyzed, as shown in Figure 7a,b for Na+ and Cl−, respectively. Figure 7a reveals that there are more Na+ ions around the protein than Cl− ions. This result is expected because the protein is negatively charged (native protein charge: −16e). The average numbers of Na+ and Cl− ions around the protein are 6.57 ± 1.01 and 4.73 ± 0.49, respectively. Figure 7b reveals that Cl− ions are mostly found around the PVDF surface. The average numbers of Na+ and Cl− ions around the PVDF surface are 3.71 ± 1.43 and 15.31 ± 2.61, respectively. These results indicate that the PVDF surface adsorbs more Cl− ions than Na+ ions. Adsorption of these ions makes the PVDF surface more hydrophilic and hence BSA adsorption becomes difficult. Owing to its inherent negative charge, BSA is repelled from the surface. These results confirm the important effect that ions have on the adsorption of BSA. Recent studies by KubiakOssowska et al.38,39 showed that the negatively charged domains of the BSA are kept away from the negatively charged

Figure 6. Positional RMSD of the Cα atoms for different domains of BSA with respect to the initial structure for (a) PB and (b) PBI systems.

1924

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B

Figure 7. Number of Na+ (black) and Cl− ions (red) around (a) the protein and (b) the PVDF surface in the PBI system.

silicate surface and screened from its repulsive influence by the presence of the counterions. Similar observation was made in our simulation. When negatively charged atoms come close together and orientation of protein is not correct, it will be repelled from the surface. This will give a chance to protein to reorient and again come to the surface. However, when ionic concentration is high, charges are screened and protein does not have the tendency to return back to the surface. Further, we analyzed the interaction between the PVDF surface and the protein. During the production simulation in the PBI system, the protein comes within 2.5 Å of the PVDF surface at 135 ns. However, the protein is subsequently repelled to a distance of 7.5 Å. Further analysis of the PB system revealed that acidic amino acids, such as Asp363 and Asp364, came into close contact with F in PVDF, resulting in BSA being repelled from the PVDF surface. Subsequently, the protein reoriented itself and again approached the PVDF surface at 300 ns, with basic residues, such as His3, Lys232, Lys316, and Lys362, coming into direct contact with the surface. These residues were mostly located close to F in PVDF, as shown in Figure 8, with the side chains of the amino acids becoming

The number of intramolecular hydrogen bonds within the protein is shown in Figure S4. The average numbers of hydrogen bonds during the last 100 ns trajectory are 505 (±11) and 498 (±11) for the PB and PBI systems, respectively. The number of hydrogen bonds shows no significant difference between the PB and PBI systems. Water molecules are weakly bound with the PVDF surface. The oxygen atom of water molecules was placed closer to the surface compared to the hydrogen atoms of water molecules.53 The hydrogen-bonding criteria were fulfilled between hydrogen of the PVDF surface and O of water molecules. In the PBI system, water molecules are easily replaced by ions. The protein end-to-end distance shown in Figure 9 reveals that the protein structure remains constant during the

Figure 9. Protein end-to-end distance over time in PB and PBI systems.

simulation in the PB system. After 300 ns, the average endto-end distance remains constant, indicating that the protein does not change conformation in an abrupt manner. Previous studies with lysozyme suggested that the protein conformation starts to change after adsorption,18 but this conformational change for protein is considerably slower. In the present simulation, drastic conformational changes were not observed, verifying that this process is relatively slow and more time would be required to observe changes in BSA conformation. However, a considerable amount of fluctuation was observed in the simulation for the PBI system, but the overall change in protein conformation was minimal, as can be seen in Figure S3. Overall, MD results suggested that excess ions make PVDF surface more negative, which in turn repels negatively charged BSA molecules. On the other hand, the PB system shows

Figure 8. Final snapshot of BSA adsorbed on a PVDF surface. Residues that are in direct contact with the surface are shown in CPK rendering.

aligned to create maximum contact. In addition, acidic amino acids, such as Asp363, Glu57, Asp56, and Glu57, and polar amino acids, such as Ser58, also came into direct contact with −CH2 groups on the PVDF surface, forming C−H···O-type interactions. The final three-point contact between BSA and the PVDF surface is shown in Figure 8. Experimentally, due to higher rate of fouling in PVDF membrane, surface of these membranes was modified to make it more hydrophilic.63 1925

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B

4. CONCLUSIONS The adsorption of proteins on the polymeric surface is of fundamental importance in membrane processes. Membrane fouling is a crucial problem facing water treatment in the membrane industry. Understanding how fouling occurs and what kind of interactions lead to fouling is crucial for preventing membrane fouling. Here, MD simulations of the adsorption of BSA on a PVDF surface were carried out with sufficient ions to neutralize the protein charge and with an excess of ions. In the absence of excess ions, adsorption of BSA on the PVDF surface was observed after 300 ns. However, for the PVDF−BSA system in 0.1 M NaCl solution, no adsorption was observed within 700 ns. The delay in the adsorption of BSA on the PVDF surface at increased NaCl concentrations is consistent with recent experimental results. The RMSD for Cα atoms does not show much deviation in either case, whereas the number of internal hydrogen bonds within the protein is lower in the system with 0.1 M NaCl solution. Our results suggested that Cl− ions adsorbed on the PVDF surface, increasing its electronegativity, which leads to repulsion of negatively charged BSA from the PVDF surface. Our results are helpful for understanding the fouling mechanism of BSA on PVDF surfaces and the role of ions in delaying such adsorption processes.

adsorption of BSA on the PVDF surface. Our results are consistent with recent experimental results. Miao et al.49 experimentally investigated the effect of ionic strength (0−100 mM) on fouling of PVDF UF membranes using BSA as a foulant. Their results indicated that increasing the ionic strength from 0 to 1 mM caused a decrease in the PVDF− BSA and BSA−BSA electrostatic repulsion forces, resulting in a higher deposition rate of BSA on the membrane surface. This process leads to the formation of a denser BSA layer and consequently increases membrane fouling. However, at ionic strengths of 10 and 100 mM, membrane fouling and BSA removal rate decreased significantly, mainly owing to increased hydration repulsion forces, which caused a decrease in the PVDF−BSA and BSA−BSA interaction forces, accompanied by a decreased hydrodynamic radius and an increased diffusion coefficient of BSA. Consequently, less BSA accumulated on the membrane surface. Recently, Miao et al.64 have also showed that hydration forces were a universal phenomenon during the membrane filtration process, when the levels of pH, ion species, and membrane performances were appropriate. Our results, which are consistent with those experimental data, provide an explanation for delayed fouling in the presence of ions. In addition, several experimental results are reported for cleaning of membrane fouled with BSA with saline solution.65,66 Recently, our group has also reported cleaning of PVDF membrane with saline solution to clean membrane fouled with BSA.67 Thus, such MD studies can be helpful for understanding the general fouling mechanism for different polymeric membranes. Membrane fouling is a very complex phenomenon, and several factors can affect it. In this article, we tried to focus on the effect of ions on BSA adsorption using all-atom force field. In the consecutive study, we will focus on the effect of orientation of BSA toward membrane surface and the effect of different types of ions using coarse-grained molecular dynamics and calculating the free energy of adsorption. A simplified model will help to access different orientations of BSA toward membrane surfaces and analyze several other factors that will contribute to membrane fouling. In summary, during adsorption of BSA on PVDF surface for the PB system, we observed the following: 1. Above the PVDF surface, BSA diffuses freely and can rotate to present its domain II toward the surface. 2. Protein with positively charged surface residue (Lys316) and Pro365 of domain II approaches to the PVDF surface around 100 ns. 3. Desorption of protein occurs and after that several atoms adsorb on the surface using Lys316 residue. 4. Reorientation of BSA occurs and again domain I of protein approaches to the surface at 316 ns. Anchoring residues at this point were Asp56, Glu57, and Ser58. 5. Protein reorients itself by keeping domain I interaction intact and its domain II comes closer to the surface. Residues such as Lys362, Asp363, and Glu320 come closer to the surface. 6. With correct orientation, the extended Lys side chains (Lys232) act as middle pillar and domain I and domain II residues bind to the surface from other side. It creates strong anchors to the surface. The protein remains in a stable adsorbed state.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.7b10221. MD simulation setup, salt bridges in BSA, different domains of BSA protein, orientation of BSA with time, distance analysis between BSA domains, RMSD analysis, and intramolecular hydrogen bond analysis (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (A.R.S.). *E-mail: [email protected]. Tel/Fax: +81-78-803-6180 (H.M.). ORCID

Abdul Rajjak Shaikh: 0000-0003-4444-0684 Hideto Matsuyama: 0000-0003-2468-4905 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by research grant from the Organization of Membrane and Film Technology, Japan. REFERENCES

(1) Elimelech, M.; Phillip, W. A. The future of seawater desalination: energy, technology, and the environment. Science 2011, 333, 712−717. (2) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Marinas, B. J.; Mayes, A. M. Science and technology for water purification in the coming decades. Nature 2008, 452, 301−310. (3) Rezaei, H.; Ashtiani, F. Z.; Fouladitajar, A. Effects of operating parameters on fouling mechanism and membrane flux in cross-flow microfiltration of whey. Desalination 2011, 274, 262−271. (4) Xiao, K.; Wang, X. M.; Huang, X.; Waite, T. D.; Wen, X. H. Combined effect of membrane and foulant hydrophobicity and surface charge on adsorptive fouling during microfiltration. J. Membr. Sci. 2011, 373, 140−151.

1926

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B (5) Bai, L. M.; Qu, F. S.; Liang, H.; Ma, J.; Chang, H. Q.; Wang, M. L.; Li, G. B. Membrane fouling during ultrafiltration (UF) of surface water: Effects of sludge discharge interval (SDI). Desalination 2013, 319, 18−24. (6) Le-Clech, P.; Chen, V.; Fane, T. A. G. Fouling in membrane bioreactors used in wastewater treatment. J. Membr. Sci. 2006, 284, 17−53. (7) Radjenović, J.; Matošić, M.; Mijatović, I.; Petrović, M.; Barceló, D. Membrane Bioreactor (MBR) as an Advanced Wastewater Treatment Technology. In Emerging Contaminants from Industrial and Municipal Waste; Barceló, D.; Petrovic, M., Eds.; Springer: Berlin Heidelberg, 2008; Vol. 5S/2, pp 37−101. (8) Rana, D.; Matsuura, T. Surface modifications for antifouling membranes. Chem. Rev. 2010, 110, 2448−2471. (9) Kochkodan, V.; Johnson, D. J.; Hilal, N. Polymeric membranes: surface modification for minimizing (bio)colloidal fouling. Adv. Colloid and Interface Sci. 2014, 206, 116−140. (10) Misdan, N.; Ismail, A. F.; Hilal, N. Recent advances in the development of (bio)fouling resistant thin film composite membranes for desalination. Desalination 2016, 380, 105−111. (11) Guo, W.; Ngo, H.-H.; Li, J. A mini-review on membrane fouling. Bioresour. Technol. 2012, 122, 27−34. (12) Sun, W.; Liu, J.; Chu, H.; Dong, B. Pretreatment and membrane hydrophilic modification to reduce membrane fouling. Membranes 2013, 3, 226−41. (13) Ahn, W. Y.; Kalinichev, A. G.; Clark, M. M. Effects of background cations on the fouling of polyethersulfone membranes by natural organic matter: Experimental and molecular modeling study. J. Membr. Sci. 2008, 309, 128−140. (14) Sutton, R.; Sposito, G.; Diallo, M. S.; Schulten, H. R. Molecular simulation of a model of dissolved organic matter. Environ. Toxicol. Chem. 2005, 24, 1902−1911. (15) Diallo, M. S.; Simpson, A.; Gassman, P.; Faulon, J. L.; Johnson, J. H.; Goddard, W. A.; Hatcher, P. G. 3-D structural modeling of humic acids through experimental characterization, computer assisted structure elucidation and atomistic simulations. 1. Chelsea soil humic acid. Environ. Sci. Technol. 2003, 37, 1783−1793. (16) Kalinichev, A. G.; Kirkpatrick, R. J. Molecular dynamics simulation of cationic complexation with natural organic matter. Eur. J. Soil Sci. 2007, 58, 909−917. (17) Xu, X.; Kalinichev, A. G.; Kirkpatrick, R. J. 133Cs and 35Cl NMR spectroscopy and molecular dynamics modeling of Cs+ and Cl− complexation with natural organic matter. Geochim. Cosmochim. Acta 2006, 70, 4319−4331. (18) Wei, T.; Carignano, M. A.; Szleifer, I. Lysozyme adsorption on polyethylene surfaces: why are long simulations needed? Langmuir 2011, 27, 12074−12081. (19) Wei, T.; Carignano, M. A.; Szleifer, I. Molecular dynamics simulation of lysozyme adsorption/desorption on hydrophobic surfaces. J. Phys. Chem. B 2012, 116, 10189−10194. (20) Szott, L. M.; Horbett, T. A. Protein interactions with surfaces: computational approaches and repellency. Curr. Opin. Chem. Biol. 2011, 15, 683−689. (21) Hughes, Z. E.; Gale, J. D. Molecular dynamics simulations of the interactions of potential foulant molecules and a reverse osmosis membrane. J. Mater. Chem. 2012, 22, 175−184. (22) Ebro, H.; Kim, Y. M.; Kim, J. H. Molecular dynamics simulations in membrane-based water treatment processes: a systematic overview. J. Membr. Sci. 2013, 438, 112−125. (23) Xiang, Y.; Liu, Y. L.; Mi, B. X.; Leng, Y. S. Hydrated polyamide membrane and its interaction with alginate: a molecular dynamics study. Langmuir 2013, 29, 11600−11608. (24) Xiang, Y.; Liu, Y. L.; Mi, B. X.; Leng, Y. S. Molecular dynamics simulations of polyamide membrane, calcium alginate gel, and their interactions in aqueous solution. Langmuir 2014, 30, 9098−9106. (25) Xiang, Y.; Xu, R.-G.; Leng, Y. Molecular dynamics simulations of poly(ethylene glycol)-grafted polyamide membrane and its interaction with calcium alginate gel. Langmuir 2016, 32, 4424−4433.

(26) Stewart, M. B.; Myat, D. T.; Kuiper, M.; Manning, R. J.; Gray, S. R.; Orbell, J. D. A structural basis for the amphiphilic character of alginates − implications for membrane fouling. Carbohydr. Polym. 2017, 164, 162−169. (27) Ridgway, H. F.; Orbell, J.; Gray, S. Molecular simulations of polyamide membrane materials used in desalination and water reuse applications: Recent developments and future prospects. J. Membr. Sci. 2017, 524, 436−448. (28) Cui, Z.; Drioli, E.; Lee, Y. M. Recent progress in fluoropolymers for membranes. Prog. Polym. Sci. 2014, 39, 164−198. (29) Shah, D.; Maiti, P.; Gunn, E.; Schmidt, D. F.; Jiang, D. D.; Batt, C. A.; Giannelis, E. R. Dramatic enhancements in toughness of polyvinylidene fluoride nanocomposites via nanoclay-directed crystal structure and morphology. Adv. Mater. 2004, 16, 1173−1177. (30) Dillon, D. R.; Tenneti, K. K.; Li, C. Y.; Ko, F. K.; Sics, I.; Hsiao, B. S. On the structure and morphology of polyvinylidene fluoridenanoclay nanocomposites. Polymer 2006, 47, 1678−1688. (31) Edwie, F.; Teoh, M. M.; Chung, T. S. Effects of additives on dual-layer hydrophobic-hydrophilic PVDF hollow fiber membranes for membrane distillation and continuous performance. Chem. Eng. Sci. 2012, 68, 567−578. (32) Majorek, K. A.; Porebski, P. J.; Dayal, A.; Zimmerman, M. D.; Jablonska, K.; Stewart, A. J.; Chruszcz, M.; Minor, W. Structural and immunologic characterization of bovine, horse, and rabbit serum albumins. Mol. Immunol. 2012, 52, 174−182. (33) Hashino, M.; Hirami, K.; Ishigami, T.; Ohmukai, Y.; Maruyama, T.; Kubota, N.; Matsuyama, H. Effect of kinds of membrane materials on membrane fouling with BSA. J. Membr. Sci. 2011, 384, 157−165. (34) Zhao, X. Z.; Xuan, H. X.; He, C. J. Enhanced separation and antifouling properties of PVDF ultrafiltration membranes with surface covalent self-assembly of polyethylene glycol. RSC Adv. 2015, 5, 81115−81122. (35) Mücksch, C.; Urbassek, H. M. Molecular dynamics simulation of free and forced BSA adsorption on a hydrophobic graphite surface. Langmuir 2011, 27, 12938−12943. (36) Delgado-Magnero, K. H.; Valiente, P. A.; Ruiz-Pena, M.; PerezGramatges, A.; Pons, T. Unraveling the binding mechanism of polyoxyethylene sorbitan esters with bovine serum albumin: A novel theoretical model based on molecular dynamic simulations. Colloids Surf., B 2014, 116, 720−726. (37) Gu, Z.; Yang, Z.; Chong, Y.; Ge, C.; Weber, J. K.; Bell, D. R.; Zhou, R. Surface curvature relation to protein adsorption for carbonbased nanomaterials. Sci. Rep. 2015, 5, No. 10886. (38) Kubiak-Ossowska, K.; Jachismsca, B.; Mulheran, P. A. How negatively charged proteins adsorb to negatively charged surfaces: A molecular dynamics study of BSA adsorption on silica. J. Phys. Chem. B 2016, 120, 10463−10468. (39) Kubiak-Ossowska, K.; Tokarczyk, K.; Jachismsca, B.; Mulheran, P. A. Bovine serum albumin adsorption at a silica surface explored by simulation and experiment. J. Phys. Chem. B 2017, 121, 3975−3986. (40) Wang, F.; Zhang, M.; Peng, W.; He, Y.; Lin, H.; Chen, J.; Hong, H.; Wang, A.; Yu, H. Effects of ionic strength on membrane fouling in a membrane bioreactor. Bioresour. Technol. 2014, 156, 35−41. (41) Silva, V.; Geraldes, V.; Brites Alves, A. M.; Palacio, L.; Prádanos, P.; Hernández, A. Multi-ionic nanofiltration of highly concentrated salt mixtures in the seawater range. Desalination 2011, 277, 29−39. (42) Yu, Y.; Lee, S.; Hong, S. Effect of solution chemistry on organic fouling of reverse osmosis membranes in seawater desalination. J. Membr. Sci. 2010, 351, 205−213. (43) Ang, W. S.; Elimelech, M. Protein (BSA) fouling of reverse osmosis membranes: Implications for wastewater reclamation. J. Membr. Sci. 2007, 296, 83−92. (44) Wang, Y.-N.; Tang, C. Y. Protein fouling of nanofiltration, reverse osmosis, and ultrafiltration membrane-The role of hydrodynamic conditions, solution chemistry, and membrane properties. J. Membr. Sci. 2011, 376, 275−282. (45) Mo, H.; Tay, K. G.; Ng, H. Y. Fouling of reverse osmosis membrane by protein (BSA): Effects of pH, calcium, magnesium, ionic strength, and temperature. J. Membr. Sci. 2008, 315, 28−35. 1927

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928

Article

The Journal of Physical Chemistry B (46) Chan, R.; Chen, V. The effects of electrolyte concentration and pH on protein aggregation and deposition: Critical flux and constant flux membrane filtration. J. Membr. Sci. 2001, 185, 177−192. (47) Salgın, S.; Takac, S.; Ozdamar, T. H. Adsorption of bovine serum albumin on polyether sulfone ultrafiltration membranes: Determination of interfacial interaction energy and effective diffusion coefficient. J. Membr. Sci. 2006, 278, 251−260. (48) Salgın, S. Effects of ionic environments on bovine serum albumin fouling in a cross-flow ultrafiltration system. Chem. Eng. Technol. 2007, 30, 255−260. (49) Miao, R.; Wang, L.; Mi, N.; Gao, Z.; Liu, T.; Lv, Y.; Wang, X.; Meng, X.; Yang, Y. Enhancement and mitigation mechanisms of protein fouling of ultrafiltration membranes under different ionic strengths. Environ. Sci. Technol. 2015, 49, 6574−6580. (50) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A. et al. Gaussian 09, revision B.01; Gaussian, Inc.: Wallingford, CT, 2009. (51) Dupradeau, F. Y.; Pigache, A.; Zaffran, T.; Savineau, C.; Lelong, R.; Grivel, N.; Lelong, D.; Rosanski, W.; Cieplak, P. The R.ED. tools: advances in RESP and ESP charge derivation and force field library building. Phys. Chem. Chem. Phys. 2010, 12, 7821−7839. (52) Wang, J. M.; Cieplak, P.; Kollman, P. A. How well does a restrained electrostatic potential (RESP) model perform in calculating conformational energies of organic and biological molecules? J. Comput. Chem. 2000, 21, 1049−1074. (53) Darvishi, M.; Forouton, M. Molecular investigation of oil−water separation using PVDF polymer by molecular dynamic simulation. RSC Adv. 2016, 6, 74124−74134. (54) Pronk, S.; Pall, S.; Schulz, R.; Larsson, P.; Bjelkmar, P.; Apostolov, R.; Shirts, M. R.; Smith, J. C.; Kasson, P. M.; van der Spoel, D.; Hess, B.; Lindahl, E. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 2013, 29, 845−854. (55) Zhou, Z.; Rajabzadeh, S.; Shaikh, A. R.; Kakihana, Y.; Ishigami, T.; Sano, R.; Matsuyama, H. Preparation and characterization of antifouling poly(vinyl chloride-co-poly(ethylene glycol)methyl ether methacrylate) membranes. J. Membr. Sci. 2016, 498, 414−422. (56) Shaikh, A. R.; Rajabzadeh, S.; Matsuo, R.; Takaba, H.; Matsuyama, H. Hydration effects and antifouling properties of poly(vinyl chloride-co-PEGMA) membranes studied using molecular dynamics simulations. Appl. Surf. Sci. 2016, 369, 241−250. (57) Ameduri, B. From vinylidene fluoride (VDF) to the applications of VDF-containing polymers and copolymers: Recent developments and future trends. Chem. Rev. 2009, 109, 6632−6686. (58) Guex, N.; Peitsch, M. C. SWISS-MODEL and the SwissPdbViewer: An environment for comparative protein modeling. Electrophoresis 1997, 18, 2714−2723. (59) Chen, V. B.; Arendall, W. B.; Headd, J. J.; Keedy, D. A.; Immormino, R. M.; Kapral, G. J.; Murray, L. W.; Richardson, J. S.; Richardson, D. C. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr., Sect. D: Struct. Biol. 2010, 66, 12−21. (60) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Comparison of multiple amber force fields and development of improved protein backbone parameters. Proteins: Struct., Funct., Bioinf. 2006, 65, 712−725. (61) Kurrat, R.; Prenosil, J. E.; Ramsden, J. J. Kinetics of human and bovine serum albumin adsorption at Silica-Titania surfaces. J. Colloid Interface Sci. 1997, 185, 1−8. (62) Carter, D. C.; He, X. M.; Munson, S. H.; Twigg, P. D.; Gernert, K. M.; Broom, M. B.; Miller, T. Y. 3-dimensional structure of humanserum albumin. Science 1989, 244, 1195−1198. (63) Liu, F.; Hashim, N. A.; Liu, Y. T.; Abed, M. R. M.; Li, K. Progress in the production and modification of PVDF membranes. J. Membr. Sci. 2011, 375, 1−27. (64) Miao, R.; Wang, L.; Zhu, M.; Deng, D.; Li, S.; Wang, J.; Liu, T.; Lv, Y. Effect of hydration forces on protein fouling of ultrafiltration

membranes: the role of protein charge, hydrated ion species, and membrane hydrophilicity. Environ. Sci. Technol. 2017, 51, 167−174. (65) Corbatón-Báguena, M.-J.; Alvarez-Blanco, S.; Vincent-Vela, M. C. Cleaning of ultrafiltration membranes fouled with BSA by means of saline solutions. Sep. Purif. Technol. 2014, 125, 1−10. (66) Corbatón-Báguena, M.-J.; Alvarez-Blanco, S.; Vincent-Vela, M. C. Fouling mechanisms of ultrafiltration membranes fouled with whey model solutions. Desalination 2015, 360, 87−96. (67) Ujihara, R.; Mino, Y.; Takahashi, T.; Shimizu, Y.; Matsuyama, H. Effects of the ionic strength of sodium hypochlorite solution on membrane cleaning. J. Membr. Sci. 2016, 514, 566−573.

1928

DOI: 10.1021/acs.jpcb.7b10221 J. Phys. Chem. B 2018, 122, 1919−1928