Article pubs.acs.org/Langmuir
Multiscale Simulations of Protein G B1 Adsorbed on Charged SelfAssembled Monolayers Jie Liu, Chenyi Liao, and Jian Zhou* School of Chemistry and Chemical Engineering, Guangdong Provincial Key Lab for Green Chemical Product Technology, South China University of Technology, Guangzhou, Guangdong, 510640, People’s Republic of China ABSTRACT: The orientation of an antibody plays an important role in the development of immunosensors. Protein G is an antibody binding protein, which specifically targets the Fc fragment of an antibody. In this work, the orientation of prototypical and mutated protein G B1 adsorbed on positively and negatively charged self-assembled monolayers was studied by parallel tempering Monte Carlo and all-atom molecular dynamics simulations. Both methods present generally similar orientation distributions of protein G B1 for each kind of surface. The root-mean-square deviation, DSSP, gyration radius, eccentricity, dipole moment, and superimposed structures of protein G B1 were analyzed. Moreover, the orientation of binding antibody was also predicted in this work. Simulation results show that with the same orientation trends, the mutant exhibits narrower orientation distributions than does the prototype, which was mainly caused by the stronger dipole of the mutant. Both kinds of proteins adsorbed on charged surfaces were induced by the competition of electrostatic interaction and vdW interaction; the electrostatic interaction energy dominated the adsorption behavior. The protein adsorption was also largely affected by the distribution of charged residues within the proteins. Thus, the prototype could adsorb on a negatively charged surface, although it keeps a net charge of −4 e. The mutant has imperfect opposite orientation when it adsorbed on oppositely charged surfaces. For the mutant on a carboxylfunctionalized self-assembled monolayer (COOH-SAM), the orientation was the same as that inferred by experiments. While for the mutant on amine-functionalized self-assembled monolayer (NH2-SAM), the orientation was induced by the competition between attractive interactions (led by ASP40 and GLU56) and repulsive interactions (led by LYS10); thus, the perfect opposite orientation could not be obtained. On both surfaces, the adsorbed protein could retain its native conformation. The desired orientation of protein G B1, which would increase the efficiency of binding antibodies, could be obtained on a negatively charged surface adsorbed with the prototype. Further, we deduced that with the packing density of 12 076 protein G B1 domain per μm2, the efficiency of the binding IgG would be maximized. The simulation results could be applied to control the orientation of protein G B1 in experiments and to provide a better understanding to maximize the efficiency of antibody binding.
1. INTRODUCTION The ability to control and manipulate antibody orientation on surfaces plays an important role in the development of immunosensors.1 One of the common methods is to immobilize the antibody via protein G. It is indirectly supported that the ordered binding layer of protein G would promote the ordered immobilization of immunoglobulin G (IgG) layer. Various results2−5 suggested that the presence of the ordered binding protein layer might improve the antigen binding efficiency. Researchers had studied the ordered binding of IgG by controlling the orientation or immobilization of protein G.6,7 Immobilization of protein G on surfaces is generally done through either noncovalent adsorption or covalent bonding. Chung and co-workers8 performed a protein G-DNA conjugate study which ensured the controlled immobilization of antibody onto the intended area of the surface. In another work,9 they developed a genetically engineered glutathione S-transferase layer contained domains of protein G to immobilize antibodies. The layers exhibited at least a 2-fold enhancement in the © XXXX American Chemical Society
immunoglobulin density with its antigen capture capability totally conserved, compared with a covalently tethered protein G. Kim et al.10 found that the cysteine-specific immobilization of Cys-protein G through the N-terminal cysteine resulted in a 2.2-fold higher binding efficiency of IgG2a antibody than the random immobilization via lysine residues. Lee et al.11 demonstrated a method using a multimeric protein G for the immobilization of antibodies on magnetic silica-nanoparticles. Johonson and Mutharasan12 suggested that pH could affect the orientation of protein G, while the most properly oriented protein G for antibody binding was achieved at near neutral pH. Both chemical and physical immobilization had been studied by Castner and co-workers13,14 to investigate the orientation of surface-immobilized protein G B1. Particularly, they could induce opposite end-on orientations of protein G B1 fragment onto two oppositely charged self-assembled monoReceived: March 28, 2013 Revised: July 8, 2013
A
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
layer (SAM) surfaces (NH3+ and COO−) by creating a charge distribution. Choi et al.7 developed various techniques such as thiolated protein G methods, mutated by adding two cysteine residues at its C-terminus,15 immobilized protein G on mixture of the 11-mercaptoundecanoic acid, and hexanethiol SAMs16 for the immobilization of IgG. They found that the binding capacity of the IgG layer was not proportional to the surface concentration of IgG because that the dense packing hindered the reaction between the Fab domain of IgG and the antigen. Yet, few studies focus on the orientation of protein G, which is one of the key issues to modulate IgG binding. As the orientation and binding site of protein adsorption on the atomic level are difficult to be obtained from experiments, molecular simulation is an effective and accurate technique to study protein behaviors on surfaces. Various research17−23 has been done to investigate the protein adsorption behaviors on solid surfaces with molecular dynamic (MD) simulation. Latour and co-workers24 thought that the chemical property of the surface could control protein orientation prior to inducing protein conformational change. The developed method was applied to analyze the preferred orientations of lysozyme on various functionalized alkanethiol SAM surfaces. Hsu et al.25 developed a method to find the global minimum of the interaction potential energy of an adsorbed protein. Subsequently, they delineated the preferred orientations of human serum albumin (HSA) adsorbed on a hydrophilic SAM with that method. Shea and co-workers26,27 performed studies to investigate the stability of proteins tethered to surfaces. Wang et al.28 supported that the orientation of adsorbed FnIII7−10 could be modulated by changing the charge property of the surface. Boughton et al.29 combined MD with experiments to study the orientational distribution of magainin 2 near polystyrene surfaces. Nordgren et al.30 investigated the orientation of cytochrome c attached to polar and nonpolar soft surfaces. They found that the polar SAM appears to interact more strongly with the protein than the nonpolar SAM does. Simultaneously, increased hydration of the system tends to reduce the effects of other parameters. Trzaskowski et al.31 suggested that the orientation of cytochrome c on the surface could be controlled by means of designing specific structural motifs on the surface. In our early work,32 a generalized residue model was developed to predict the orientation of IgG1 and IgG2a on charged surfaces. The orientation and conformation of cytochrome c adsorbed on charged SAM surfaces were studied by combining Monte Carlo (MC) with MD simulations.33 Hagiwara et al.34 also combined MC with MD simulations to investigate the orientation and adsorption state of β-Lactoglobulin on the Au(100) surface. However, in the energy-based method to probe possible orientations of adsorbed protein, one awkward problem is the representativeness of configuration space sampling. Conventional MC and MD simulations in a canonical ensemble often suffer from the quasi-ergodic problem, i.e., simulations at low temperatures tend to get trapped in local-minimum-energy states and thus fail to sample the whole configuration space. Therefore, to improve the efficiency of configuration space sampling, parallel tempering algorithm has been combined with MC and MD method in many researches. Knotts et al.35,36 successfully applied a method combining density-of-states based technique and parallel tempering to study the stability of proteins on surfaces. Xie et al.37 developed a parallel tempering Monte Carlo algorithm (PTMC) based on the united-residue model32 with implicit water model; the PTMC
algorithm can estimate the primary orientation of proteins on charged surfaces. The main purpose of PTMC37 is to help the protein jump out from the local-energy-minimum orientation state and get the global-energy-minimum orientation state. As in PTMC, although the rotation move is nonphysical, it could break the time scale limit and more efficiently attain the desired orientation. Presently, simulation studies of protein G were focused on the folding kinetics and structural organization.38,39 Yet, the adsorption mechanism between protein G and metal/organic surfaces and the further binding orientation of antibody is not clear. Previously, Baio et al.14 found that with mutating specific negatively charged amino acids in protein G B1 domain into neutral amino acids, the mutant could adsorb on different charged surfaces in opposite end-on orientations. However, the mechanism of the binding behavior, and the differences of the antibody binding efficiency between the prototype and mutant are still unknown at the atomistic level. In this work, we combine the PTMC37 with all-atom MD method to investigate the behavior of prototypical and mutated protein G adsorbed on oppositely charged SAM surfaces in order to investigate the orientation of protein G B1 and to find suitable surfaces for the rational orientation control for antibody binding. Proteins in bulk are also performed here as references to be compared with adsorbed proteins on surfaces. New insights which complement those obtained from experiments14 are presented here by molecular simulations. Due to the different composition of charged residues between the prototype and the mutant, the orientation and conformation of both prototype and mutant adsorbed on charged SAMs are compared. Finally, we further forecast the probable orientation for antibody binding.
2. MATERIALS AND METHODS Protein G B1 domain (as shown in Figure 1) consists of 56 residues with four β-sheets and one α-helix. The high-resolution crystal
Figure 1. Secondary structure of protein G B1 domain. structure of protein G B1 domain (PDB: 1PGB), refined to 0.192 nm by Gallagher et al.,40 served as the structure model for both PTMC and MD simulations. 2.1. PTMC. For a united-residue model of protein, each amino acid is reduced to a sphere centered at the α-carbon of the residue, in which the basic structure information of a protein is well kept.32 The charged surface is treated as a flat surface. The interactions of the charged surface with charged particles compose a series of parameters featuring both van der Waals (vdW) and electrostatic interactions. The parameters were taken from our previous works.32,37 To get the preferred orientation of protein adsorbed on charged surfaces, we performed PTMC simulations before MD study. During PTMC simulations, the protein was kept rigid. N replicas had been calculated in parallel, each in the canonical ensemble as well at different temperature T. The potential energy of protein−surface interaction (Utot) is composed of vdW interaction energy (UvdW) and electrostatic interaction energy (Uele). The details are the same as B
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
those interpreted in our previous works.32,37 The MC simulation in each replica was carried out in a box of 10 × 10 × 10 nm3. As the protein in PTMC simulations are performed by a united-residue model, the amino acids in proteins are reduced to spheres centered at α-carbon of the residues, in which the basic structure information of a protein is well preserved. The protein was initially placed in the center of the simulation box, then with translational and rotational moves around its center of mass. A total of 40 000 000 MC cycles were carried out with 20 000 000 cycles for equilibrium and another 20 000 000 cycles for production. The displacement of each move was adjusted to ensure an acceptance ratio of 0.5. Six replicas, with temperatures of 300 K, 350 K, 450 K, 550 K, 650 and 750 K, respectively, had been used to ensure sufficient energy overlap between neighboring replicas to allow for the acceptance of configuration swaps in each protein-SAMs system. The swaps are performed every 500 cycles. The final protein orientation with the lowest energy was taken as the initial state for further MD simulations. 2.2. MD. For MD simulations, the initial atomistic protein coordinates were obtained by best fit of α-carbon atoms with the coarse-grained structure determined by PTMC with the preliminary preferred orientation. Hydrogen atoms within the protein were added by GROMACS 4.5.3.41 The protein was simulated at pH 7.0. The lysine (LYS) residues as well as the N-terminus of protein G B1 domain were taken to be protonated; whereas glutamic acid (GLU) and aspartic acid (ASP) residues along with the C-terminus were taken to be deprotonated. For the all-atom structure of protein G B1 domain, the prototypical protein has a net charge of −4 e. The mutated protein G was referenced from the work of Baio et al.14 The mutant (D4′), with four charge-neutralizing mutants at one end of the Protein G B1 domain (E19Q, D22N, D46N, and D47N), was constructed by PyMOL software.42 Only one protein molecule was considered in the MD simulations. The potential parameters for proteins are from the OPLS force field.43 The √3 × √3 structures of HS(CH2)10COOH-SAM and HS(CH2)10NH2-SAM on Au(111) were adopted in the all-atom MD simulations, which consist of 144 thiol chains and 1296 gold atoms. For thiol chains in the studied systems, 10 chains are protonated or deprotonated for NH2-SAM and COOH-SAM respectively, representing surface charge densities of ±0.05 C/m2 to mimic the experiment. The surface has the dimension of 5.994 ×5.191 nm2. S atoms in SAM and Au atoms are kept fixed during the whole simulation. The potential parameters for SAMs are from the OPLS force field.44 With the preliminarily optimized orientation of protein on SAM surfaces from PTMC simulations, water molecules were added to a simulated box of 5.994 × 5.191 × 6.0 nm3. The added water molecules were selected such that no water oxygen atom was closer than 0.28 nm to the protein and surfaces. There are 5880 water molecules in the system, which are described by the SPC/E model.45 Counter ions were added to keep the system neutral. Each simulation system has approximately a total of 25 275 atoms. The simulation box size is 5.994 × 5.191 × 30.000 nm3. There is a vacuum layer (22 nm) beyond the water layer. The 22 nm vacuum layer was added to produce a pseudo2d summation. The SAM with gold is about 2 nm high. Before MD simulations, the whole system was optimized with the steepest descent method to eliminate the steric overlap or inappropriate geometry after converting the coarse-grained structure into the atomistic structure; as the side chain of protein is ignored in PTMC simulations. With the GROMACS 4.5.3. package, the MD simulations were performed in a canonical ensemble with integrating time steps of 2.0 fs and the temperature was controlled by a Nose-Hoover thermostat46,47 with a time constant of 0.2 ps. The temperature of the simulated system was 300 K. Bonds containing hydrogen atoms were constrained by the LINCS algorithm.48 For the nonbonded interactions, the cutoff distance was 1.0 nm. Electrostatic interactions were calculated by the Particle mesh Ewald (PME) method49 for slab system. A 50 ns MD simulation was performed for each system. The prototype and the mutant in bulk with the box size of 5.994 × 5.191 × 6.0 nm3 were performed as references to be compared with adsorbed proteins on surfaces. In the bulk solution, the protein should be stable and preserve
its native structure. If the surface induces significant structural change of the protein compared with that in bulk, that means the surface is not suitable for the immobilization of proteins. Other running parameters of proteins in bulk are the same as those mentioned for surface systems. For structure visualization, the visual molecular dynamics (VMD) program50 was used. 2.3. Data Analysis. 2.3.1. Protein Orientation. The orientation angle is used to quantitatively characterize the protein orientation on surfaces. It is defined as the angle between the unit vector normal (n) to the surface and the unit vector along the protein dipole (m) (as shown in Figure 2). The cosine value of this angle is used to represent
Figure 2. Illustration of orientation angle within protein. The direction of normal to surface is noted as n and the direction of dipole of protein is noted as m. the orientation of the adsorbed protein. In short, the cosine value could quantify the direction of the dipole within the protein refer to the normal to the surface. For the cosine value of −1, the dipole is antiparallel to the normal to the surface; for that of +1, the dipole is parallel to the normal to the surface; for the cosine value of 0, the dipole is parallel to the surface. 2.3.2. Conformation of Protein. The root-mean-square deviation (RMSD) represents the minimum root-mean-square deviation between one simulated structure and its reference structure. It is defined as RMSD = [(1/M)∑i N= 1mi||ri(t1) − ri(t2)||2]1/2, where M = ∑i N= 1mi and ri(t) is the position of atom i at time t. It can be used to analyze the structure changes of proteins on surfaces referring to that in bulk or under other conditions. The gyration radius of a protein, Rg, is defined as Rg = [(∑i||ri||2mi)/ (∑imi)]1/2, where mi is the mass of atom i and ri is the position of atom i with respect to the center of mass of the molecule. It represents a mass-weighted root-mean-square average distance of all atoms in a protein from its center of mass, which could characterize the overall size of a protein. The eccentricity of a protein is another parameter that could be used to characterize the overall shape of a protein. It is defined as 1 − Iave/Imax, in which Imax is the maximal principal moment of inertia, and Iave is the average of three principal moments of inertia.30,33 The dipole of a protein is defined as μ⃗ = ∑iN= 1qi(ri − rCOM), where qi is the partial charge of each atom, and rCOM is the position of the center of mass of the protein. The dipole moment is the module of the dipole; it can illustrate the distribution of charged residues in protein. The dipole moments and other properties of the prototype and the mutant are listed in Table 1.
Table 1. Some Properties of Prototypical Protein G B1 and Mutated Protein G B1
prototype mutant
no. of residues
mol wt (Da)
dipole moment (D)
net charge (e)
56 56
6192 6192
144 318
−4 0
3. RESULTS AND DISCUSSION The orientation and conformation of prototypical and mutated protein G B1 domain on oppositely charged SAMs were investigated by PTMC and MD simulations. For comparison, behaviors of proteins in the bulk solution were also studied. The preliminary orientations of protein G B1 on SAM surfaces C
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
were obtained from PTMC simulations. Furthermore, the subsequent orientation of antibody had been predicted. The optimal orientation, gyration radius, eccentricity, RMSD, superimposed structures and dipole moment of protein were calculated during the MD simulations. Simulation results are summarized in Tables 2, 3, and 4 and Figures 3−10. Table 2. Interaction Energies of Protein G B1 Adsorbed on Charged Surfaces prototype on COOH-SAM prototype on NH2-SAM mutant on COOH-SAM mutant on NH2-SAM
Utot (kJ·mol−1)
UvdW (kJ·mol−1)
Uele (kJ·mol−1)
−421.2
−67.9
−353.3
−228.9
−22.0
−206.9
−406.3
−92.2
−314.1
−256.4
−44.0
−212.5
Table 3. Interaction Energies between Charged Residues of the Prototype with COOH-SAM Surface UvdW(kJ·mol−1) Uele(kJ·mol−1)
LYS10
LYS13
ASP36
GLU56
−21.6 −180.6
−1.6 −158.6
−8.7 −14.1
−0.2 0.1
Figure 3. Orientation distributions of mutated and prototypical proteins adsorbed on oppositely charged surfaces by PTMC (above) and MD (below) simulations. The distribution of the mutant adsorbed on COOH-SAM is presented with a black line, whereas that adsorbed on NH2-SAM is presented with a red line. The distributions of the prototype adsorbed on COOH-SAM and NH2-SAM are presented in blue and mauve, respectively.
3.1. Energy and Orientation. As shown in Table 2, for four systems, the protein adsorption behaviors are driven by the competition of electrostatic and vdW interactions, while the absolute value of electrostatic interaction energy is distinctly higher than that of the vdW interaction energy. The electrostatic interaction is the dominant driving force during the adsorption. Particularly, the electrostatic interaction energy between the prototype and COOH-SAM is even lower than that between the prototype and NH2-SAM, although the prototype has the same charge property with the COOH-SAM. That is mainly because of the binding of positively charged residues (LYS10, LYS13) with the negatively charged surface. We further analyze the energies between the surface and charged residues (LYS10, LYS13, ASP36, and GLU56) of the prototype near the negatively charged surface, as shown in Table 3. Obviously, the adsorption is induced by the positively charged LYS10 and LYS13 residues simultaneously. In Ravichandran’s work,51 they also indicated that the net charge were not the major criterion for the protein adsorption; while the distribution of charged amino acid residues on the protein surface was the dominant factor in determining the electrostatic interaction between the protein and the surface. The similar finding had also been concluded by Xie et al.37 The orientational distributions of protein on different charged surfaces during PTMC and MD simulations are shown in Figure 3. As was mentioned in our previous
works,32,33,37,52 the dipole moment of protein was an important factor in determining the orientation of protein on charged surfaces. From Table 4 and Figure 3, it can be seen that the orientation distribution in both PTMC and MD simulations presented the same trend. Due to the stronger dipole of the mutant, the orientation distributions of the mutant appear narrower than those of the prototype; the cosine of orientation angle of the mutant was closer to ±1 than that of the prototype. It is worth mentioning that although the prototype has a net charge of −4e, it can stably adsorb on negatively charged surfaces; because the positively charged domain (LYS10, LYS13) of the prototype protein contributes the electrostatic interactions to induce the adsorption process (as shown in Table 3). Figures 4 and 5 display the final protein configurations by PTMC and MD simulations on oppositely charged SAMs. For the mutant, it has “end-on” orientation (i.e., N-terminus orient toward the surface) on negatively charged SAM (COOH-SAM, as shown in Figures 4a and 5a) with a cosine value of −0.9, which is the same as that assumed in Baio et al.’s work.14 For
Table 4. Averaged Properties of the Prototype and Mutant of Protein G B1 in Bulk Solutions and Adsorbed on SAM Surfaces by MD Simulations orientation prototype
mutant
bulk NH2-SAM COOH-SAM bulk NH2-SAM COOH-SAM
0.83 ± 0.10 0.05 ± 0.2 0.95 −0.85 ± 0.05
Rgyr(nm)
eccentricity
RMSD (nm)
± ± ± ± ± ±
0.20 0.18 0.16 0.20 0.17 0.20
0.13 0.13 0.11 0.12 0.13 0.12
1.08 1.04 1.03 1.07 1.05 1.04 D
0.03 0.03 0.02 0.02 0.02 0.02
dipole (D) 154 165 153 308 318 311
± ± ± ± ± ±
18 19 27 21 21 23
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
adsorbed on COOH-SAM (Figures 4c and 5c). The “lyingon” orientations became the preferred ones when the prototype adsorbed on positively charged surface (NH2-SAM, as shown in Figures 4d and 5d). Interestingly, the dipole of the prototype was almost parallel to the negatively charged surface. The adsorption was induced by the competition of electrostatic and vdW interactions. Therefore, the orientation of protein is the compromise of these two interactions. Moreover, in the prototype, the positively charged residues LYS10 and LYS13 contribute more interaction energies with the negatively charged residues. Thus, the perfectly perpendicular orientation cannot be obtained. From this result, it can be found that the orientation of protein adsorbed on charged surfaces could be controlled by altering the charge distribution within the protein (mutation). Meanwhile, the larger the dipole of the protein becomes, the narrower the orientation distribution would be. 3.2. Binding Sites. From the preferred orientation of proteins on charged SAMs, we further analyze the residues close to surfaces. For the mutant adsorbed on the negatively charged surface (Figure 5a), the MET1, which is the Nterminus, and LYS50 provided the major contribution to the adsorption, while GLU56 and ASP40 were the dominated binding sites when the mutant was adsorbed on the positively charged surface (Figure 5b). As for the prototype, the LYS10 and LYS13 lead the protein adsorbed on the negatively charged surface (Figure 5c); while on the positively charged surface, the prototype was adsorbed by ASP40, GLU42, and GLU56 (Figure 5d). From these results, we can find that the charged residue (e.g., lysine, glutamate, and aspartate) dominated the binding behavior. It was consistent with what we have found before, i.e., the protein adsorption was mainly induced by strong electrostatic interactions.32,33,37,53 3.3. RMSD, DSSP, and Superimposed Structures. In this work, the crystal structures of mutated and prototypical proteins were used as reference structures in the analysis of RMSD. From Table 4, it can be seen that the RMSD value of the prototype in bulk solution was 0.13 nm and was comparable to that of protein G B1 in water by Sheinerman and Brooks54 (i.e., 0.12 nm). The RMSD values of mutant were 0.13 and 0.11 nm on NH2-SAM and COOH-SAM, respectively. While for the prototype, the RMSD values were 0.13 and 0.12 nm on NH2-SAM and COOH-SAM, respectively. The information about the stability of the secondary structures of adsorbed proteins was illustrated by the DSSP (Definition of Secondary Structure of Proteins) analysis55 and superimposed structures. In Figure 6, we presented the DSSP analysis for each residue of the prototype and the mutant. As can be seen from the DSSP analysis, properties of secondary structures were retained during the whole MD simulation. Moreover, the simulated structures of adsorbed proteins on surfaces were superimposed on their crystal structures as shown in Figure 7. This provided a visual assessment of the overall structure of the protein in solution and on surfaces. The overall structures of the prototype and the mutant adsorbed on charged surfaces were kept, as well as that in solution. The proteins that had not gone through distinct conformational change were mainly caused by the low charge density of surface. This is consistent with our previous work,33 the conformation of cytochrome c could be largely retained when it adsorbed on a slightly charged surface. As a result, the protein keeps its biological activity either in bulk solution or on surfaces, and could be used to promote the succeeding antibody binding.
Figure 4. Preferred configurations of protein orientations on oppositely charged surfaces from PTMC simulations. The surface in red is for negatively charged surface while that in blue is for positively charged surface. The direction of dipole of protein is denoted as m. Orientations of mutants are shown in parts (a) and (b); while those of prototypes are shown in parts (c) and (d).
Figure 5. Preferred configurations of protein orientations on oppositely charged surfaces from MD simulations. The terminal group of self-assembled monolayer in red is for COOH-SAM, while that in blue is for NH2-SAM. The vdW space-filling representation in red is for negatively charged residues (i.e., GLU, ASP), and that in blue is for positively charged residues (i.e., LYS, N-terminal residue). The direction of dipole of protein is noted as m. Orientations of mutants are shown in parts (a) and (b); while those of prototypes are shown in parts (c) and (d).
positively charged surfaces (NH2-SAM), although the mutant oriented with “head-on” orientation (i.e., N-terminus orient toward the solution), the orientation was not perfectly opposite to that on COOH-SAM (as shown in Figures 4b and 5b). As was mentioned in Baio et al.’s work,14 with substituting specific negatively charged amino acids with neutral amino acids, the protein could adsorb on charged surfaces in opposite end-on orientations. Moreover, they illustrated by diagram that the opposite orientation was induced by the one side distribution of negatively charged residues. In this work, we provide a more detailed explanation to illustrate the real adsorption states. The imperfect opposite orientation was mainly caused by the residue of LYS10 (as shown in Figure 5b). When the mutant adsorbed on NH2-SAM, ASP36, and GLU56 performed the electrostatic attraction between the protein and surface, while the adjacent LYS10 contributed the electrostatic repulsion. The competition of these two interactions resulted in the imperfect opposite orientation relative to that on COOH-SAM. While for the prototype, it performed “head-on” orientation when E
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
Figure 8. Dipole moment as a function of time for protein G B1 adsorbed on oppositely charged surfaces.
constructed of 56 residues with four β-sheets and one αhelix; the structure is particularly stable because of the major βsheet constitution. The distributions of charged amino acids within the protein are scattered, it is hard to induce the change of dipole moment. The slightly charged surface was another factor that preserves the conformation of proteins from denaturation, as mentioned in section 3.3. The dipole moment was another parameter to illustrate that the native state conformation can be conserved when proteins adsorbed on charged surfaces. Moreover, the dipole moments of the mutant on both charged surfaces were two times as large as that of the prototype. It had been mentioned that we had created a distinct charge distribution of the mutant. Thus the larger dipole moment makes the mutant have a narrower protein orientation distribution than that of the prototype. 3.6. Orientation of Antibody. Antibodies (IgGs) could be immobilized stably onto surfaces with protein G due to physical interactions. It is significant to confirm that the antibodies adsorbed on surfaces have ordered orientation. Controlling the orientation of antibodies by ordering the orientation of protein G B1 domain has been investigated by many researchers, as mentioned in the Introduction. Researchers56−58 focused on interactions between protein G B1 domain and the Fc fragment of IgG. Sauer-Eriksson et al.59 suggested that the residues from protein G which were involved in binding were situated within the C-terminal part of the α-helix, the N-terminal part of the third β-strand and the loop region (GLU27, LYS28, LYS31, GLN32, TRP43, ASN35, VAL39, GLU42, THR44) connecting these two structural elements. Wiseman and Frank60 suggested that an antibody could adsorb on the surface in four kinds of orientations (i.e., Fab-up (end-on), Fab-down (end-on), side-on, and flat-on (see Figure 9). Obviously, we preferred the Fab-up
Figure 6. Time evolution of the secondary structures of the mutant and the prototype adsorbed on surfaces.
Figure 7. Simulated structures of the mutant (red) adsorbed on COOH-SAM (a) and NH2-SAM (b), the prototype (red) adsorbed on COOH-SAM (c) and NH2-SAM (d) superimposed on their crystal structures of protein G B1 domain (cyan).
3.4. Gyration Radius and Eccentricity. From Table 4, it can be seen that the gyration radius of the mutant was 1.07 nm when solvated in bulk solution; while 1.05 and 1.04 nm when adsorbed on NH2-SAM and COOH-SAM, respectively. The gyration radii of the mutant on positively charged surface and negatively charged surface were 1.87% and 2.80% smaller when compared with that in bulk solution. The gyration radii of the prototype on NH2-SAM (1.04 nm) and COOH-SAM (1.03 nm) were 3.70% and 4.63% smaller when compared with that in bulk solution (1.08 nm). The values of the eccentricity were also close to each other. In other words, either the mutant or the prototype adsorbed on charged surfaces had the same conformation as that in bulk solution. Thus, the slightly charged surfaces do not affect the overall shapes of the prototype and the mutant. 3.5. Dipole. The evolutions of the dipole moment as a function of time for protein in bulk solution and on SAMs were displayed in Figure 8. As is shown in Table 4 and Figure 8, for the mutant, the dipole moments of protein adsorbed on surfaces are almost equal to that in bulk solution. The same phenomenon could be found for the prototype. One reason for the unchanged dipole moments is that the protein is
Figure 9. Terms used for antibody orientation. The red beads indicate antigen-binding sites.
orientation to maximize the antigen binding of antibody. In this work, we initially controlled the orientation of protein G B1 by strong electrostatic interactions. Then we predicted the conceivable orientation of the antibody which bound to the protein G B1 domain obtained from the charged surface. The structure of the complex between Fc and protein G (PDB: 1FCC) was rotated to make the protein G B1 domain within F
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
the complex, best fit with the orientation of protein G B1 that adsorbed on charged surfaces. Thus, the orientation of antibody was obtained. In Figure 10, the red beads are for the hinge
LYS13 and LYS10; while the ASP40, GLU42, and GLU56 displayed as the key binding sites when it was adsorbed on NH2-SAM with a “lying-on” orientation. For the mutant, LYS50 and MET1 (the N-terminus) led the mutant to adsorb on COOH-SAM with “end-on” orientation; while the mutant adsorbed on NH2-SAM, with the orientation of “head-on” by GLU56 and ASP40. The mutant has imperfect opposite orientations when it is adsorbed on oppositely charged surfaces, which is different from that inferred from experiments. For the mutant on COOH-SAM, the orientation was the same as that assumed by experiments. For the mutant on NH2-SAM, the orientation was induced by the competition between attractive interaction (ASP40 and GLU56) and repulsive interaction (led by LYS10); thus, the perfect opposite orientation could not be performed. With the same orientation trends, the prototype presents wider orientation distributions than does the mutant, because of the twice as large dipole moment of the mutant. The prototype with a net charge of −4 e can still be stably adsorbed on COOH-SAM, due to the local distribution of charged residues of the prototype. For four systems, the protein adsorption behaviors are driven by the competition of electrostatic and vdW interactions, while the absolute value of electrostatic interaction energy is distinctly higher than that of the vdW interaction energy. Although protein adsorption is dominated by strong electrostatic interactions, the adsorbed proteins have few conformational changes on the whole and the protein bioactivity is preserved. This had been proven by the analysis of the root-mean-square deviation, DSSP, gyration radius, eccentricity, dipole moment, and superimposed structures of proteins. Moreover, the orientations of binding antibodies have been predicted in our study. As a result, the prototype absorbed on the negatively charged surface would control the orientation of binding antibody with practical applications. Furthermore, we deduce that with the packing density of 12 076 protein G B1 domain per μm2, the efficiency of the binding IgG would be maximized. This work sheds lights on the mechanism of controlling the desired orientation of antibody by assessing the conditions for attaining preferred orientation of protein G B1 domain. The findings could provide useful information for the design and development of antibody biosensors and other biological diagnosis devices.
Figure 10. Preferred orientations of antibody bound to confirmed configurations of (a,b) mutated and (c,d) prototypical protein G B1 domains. Protein in blue is for the Fc fragment of the antibody, whereas that in lime green is for the mutant and the prototype. The red beads are for the hinge region bound to Fab fragments, which are the binding sites of the antigen. Reference surfaces are performed to imitate the negatively charged surface (in red) and the positively charged surface (in blue).
region bound to Fab fragments, which are the binding sites of the antigen. From Figure 10, it is revealed that the orientation of the mutant adsorbed on charged surfaces (NH2-SAM and COOH-SAM) would induce the binding antibody adsorbed with “flat-on” orientation. For the prototype adsorbed on the positive NH2-SAM, the subsequent binding antibody adsorbed with typical “Fab-down” orientation. That is to say, while the prototype can contain an ordered orientation when adsorbed on the positive NH2-SAM, it cannot promote the antibody binding for practical applications. Furthermore, we studied the orientation of antibody controlled by the prototype adsorbed on the negative COOH-SAM. After analyzing the trajectory of this system, the statistical results show that antibody adsorbed on “Fab-up” orientation (Figure 10c) with a frequency of 98%. Interestingly, from the adsorption energy viewpoint, this system is also the most stable. Thus, we suggest an experimental strategy as below to maximize the efficiency of antibodies. First, the protein G B1 domain could be adsorbed on COOH-SAM. Then, chemical cross-linker should be added to immobilize the protein G B1 domain onto the surface. Moreover, we predicted the rational packing scheme of the protein G B1 domain on charged surfaces. The predicted packing density was calculated by the real size of the complex of IgG (PDB: 1IGT) and protein G B1 domain (refer to PDB: 1FCC). The aspect ratio of the complex is 16.3 ×10.16 nm2, and each Fc fragment can bind with two protein G B1 domains, we deduce that with the packing density of 12 076 protein G B1 domain per μm2, the sensitivity of the binding IgG would be maximized.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work is supported by National Key Basic Research Program of China (No. 2013CB733504), Program for New Century Excellent Talents in University (NCET-07-0313), National Natural Science Foundation of China (Nos. 20706019, 20876052, 21376089), and Guangdong Science Foundation (No. S2011010002078). The computational resources for this project are provided by SCUTGrid at South China University of Technology.
4. CONCLUSIONS In this work, the orientation and conformation of prototypical and mutated protein G B1 adsorbed on oppositely charged selfassembled monolayers (i.e., NH2-SAM and COOH-SAM) are investigated by a combined PTMC and MD simulation approach. Both the prototype and the mutant adsorb on charged surfaces with preferred orientations. For the prototype, it was adsorbed on COOH-SAM with “head-on” orientation by
■
REFERENCES
(1) Gray, J. J. The interaction of proteins with solid surfaces. Curr. Opin. Struc. Biol. 2004, 14 (1), 110−115.
G
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
(2) Tanaka, T.; Matsunaga, T. Fully automated chemiluminescence immunoassay of insulin using antibody-protein A-bacterial magnetic particle complexes. Anal. Chem. 2000, 72 (15), 3518−3522. (3) Neubert, H.; Jacoby, E. S.; Bansal, S. S.; Iles, R. K.; Cowan, D. A.; Kicman, A. T. Enhanced affinity capture MALDI-TOF MS: Orientation of an immunoglobulin G using recombinant protein G. Anal. Chem. 2002, 74 (15), 3677−3683. (4) Song, H. Y.; Zhou, X. D.; Hobley, J.; Su, X. D. Comparative study of random and oriented antibody immobilization as measured by dual polarization interferometry and surface plasmon resonance spectroscopy. Langmuir 2012, 28 (1), 997−1004. (5) Wang, H.; Castner, D. G.; Ratner, B. D.; Jiang, S. Y. Probing the orientation of surface-immobilized immunoglobulin G by time-offlight secondary ion mass spectrometry. Langmuir. 2004, 20 (5), 1877−1887. (6) Gao, D.; McBean, N.; Schultz, J. S.; Yan, Y. S.; Mulchandani, A.; Chen, W. F. Fabrication of antibody arrays using thermally responsive elastin fusion proteins. J. Am. Chem. Soc. 2006, 128 (3), 676−677. (7) Bae, Y. M.; Oh, B. K.; Lee, W.; Lee, W. H.; Choi, J. W. Study on orientation of immunoglobulin G on protein G layer. Biosens. Bioelectron. 2005, 21 (1), 103−110. (8) Jung, Y.; Lee, J. M.; Jung, H.; Chung, B. H. Self-directed and selforiented immobilization of antibody by protein G-DNA conjugate. Anal. Chem. 2007, 79 (17), 6534−6541. (9) Ha, T. H.; Jung, S. O.; Lee, J. M.; Lee, K. Y.; Lee, Y.; Park, J. S.; Chung, B. H. Oriented immobilization of antibodies with GST-fused multiple F-c-specific B-domains on a gold surface. Anal. Chem. 2007, 79 (2), 546−556. (10) Kim, E. S.; Shim, C. K.; Lee, J. W.; Park, J. W.; Choi, K. Y. Synergistic effect of orientation and lateral spacing of protein G on an on-chip immunoassay. Analyst. 2012, 137 (10), 2421−2430. (11) Lee, J. H.; Choi, H. K.; Lee, S. Y.; Lim, M. W.; Chang, J. H. Enhancing immunoassay detection of antigens with multimeric protein Gs. Biosens. Bioelectron. 2011, 28 (1), 146−151. (12) Johnson, B. N.; Mutharasan, R. pH effect on protein G orientation on gold surfaces and characterization of adsorption thermodynamics. Langmuir 2012, 28 (17), 6928−6934. (13) Baugh, L.; Weidner, T.; Baio, J. E.; Nguyen, P. C. T.; Gamble, L. J.; Slayton, P. S.; Castner, D. G. Probing the orientation of surfaceimmobilized protein G B1 using ToF-SIMS, Sum Frequency Generation, and NEXAFS Spectroscopy. Langmuir. 2010, 26 (21), 16434−16441. (14) Baio, J. E.; Weidner, T.; Baugh, L.; Gamble, L. J.; Stayton, P. S.; Castner, D. G. Probing the orientation of electrostatically immobilized protein G B1 by time-of-flight secondary ion spectrometry, sum frequency generation, and near-edge X-ray adsorption fine structure spectroscopy. Langmuir 2012, 28 (4), 2107−2112. (15) Kim, H.; Kang, D. Y.; Goh, H. J.; Oh, B. K.; Singh, R. P.; Oh, S. M.; Choi, J. W. Analysis of direct immobilized recombinant protein G on a gold surface. Ultramicroscopy 2008, 108 (10), 1152−1156. (16) Choi, J. W.; Park, K. S.; Lee, W. C.; Oh, B. K.; Chun, B. S.; Paek, S. H. Regulation of anti-LDL immobilization on self-assembled protein G layer using CHAPS and its application to immunosensor. Mater. Sci. Eng., C 2004, 24 (1−2), 241−245. (17) Xie, Y.; Liu, M. F.; Zhou, J. Molecular dynamics simulations of peptide adsorption on self-assembled monolayers. Appl. Surf. Sci. 2012, 258 (20), 8153−8159. (18) Zhang, A. J.; Xie, Y.; Zhou, J. Experimental control and characterization of proteins orientation on surfaces. Prog. Chem. 2009, 21, 1408−1417. (19) Sheng, Y. B.; Wang, W.; Chen, P. Adsorption of an ionic complementary peptide on the hydrophobic graphite surface. J. Phys. Chem. C 2010, 114 (1), 454−459. (20) Yu, X.; Wang, Q. M.; Lin, Y. A.; Zhao, J.; Zhao, C.; Zheng, J. Structure, orientation, and surface interaction of alzheimer amyloidbeta peptides on the graphite. Langmuir 2012, 28 (16), 6595−6605. (21) Wei, T.; Carignano, M. A.; Szleifer, I. Lysozyme adsorption on polyethylene surfaces: Why are long simulations needed? Langmuir 2011, 27 (19), 12074−12081.
(22) Trzaskowski, B.; Leonarski, F.; Les, A.; Adamowicz, L. Modeling tubulin at interfaces. Immobilization of microtubules on self-assembled monolayers. J. Phys. Chem. B 2005, 109 (37), 17734−17742. (23) Xie, Y.; Liao, C. Y.; Zhou, J. Effects of external electric fields on lysozyme adsorption by molecular dynamics simulations. Biophys. Chem. 2013, 179, 26−34. (24) 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 (3), 1103−1117. (25) Hsu, H. J.; Sheu, S. Y.; Tsay, R. Y. Preferred orientation of albumin adsorption on a hydrophilic surface from molecular simulation. Colloids Surf. B 2008, 67 (2), 183−191. (26) Friedel, M.; Baumketner, A.; Shea, J. E. Effects of surface tethering on protein folding mechanisms. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (22), 8396−8401. (27) Friedel, M.; Baumketner, A.; Shea, J. E. Stability of a protein tethered to a surface. J. Chem. Phys. 2007, 126, 095101. (28) Wang, H.; He, Y.; Ratner, B. D.; Jiang, S. Y. Modulating cell adhesion and spreading by control of FnIII(7−10) orientation on charged self-assembled monolayers (SAMs) of alkanethiolates. J. Biomed. Mater. Res. A 2006, 77A (4), 672−678. (29) Boughton, A. P.; Andricioaei, I.; Chen, Z. Surface orientation of magainin 2: Molecular Dynamics Simulation and sum frequency generation vibrational spectroscopic studies. Langmuir 2010, 26 (20), 16031−16036. (30) Nordgren, C. E.; Tobias, D. J.; Klein, M. L.; Blasie, J. K. Molecular dynamics simulations of a hydrated protein vectorially oriented on polar and nonpolar soft surfaces. Biophys. J. 2002, 83 (6), 2906−2917. (31) Trzaskowski, B.; Leonarski, F.; Les, A.; Adamowicz, L. Altering the orientation of proteins on self-assembled monolayers: A computational study. Biomacromolecules 2008, 9 (11), 3239−3245. (32) Zhou, J.; Chen, S. F.; Jiang, S. Y. Orientation of adsorbed antibodies on charged surfaces by computer simulation based on a united-residue model. Langmuir 2003, 19 (8), 3472−3478. (33) Zhou, J.; Zheng, J.; Jiang, S. Y. Molecular simulation studies of the orientation and conformation of cytochrome c adsorbed on selfassembled monolayers. J. Phys. Chem. B 2004, 108 (45), 17418− 17424. (34) Hagiwara, T.; Sakiyama, T.; Watanabe, H. Molecular simulation of bovine beta-lactoglobulin adsorbed onto a positively charged solid surface. Langmuir 2009, 25 (1), 226−234. (35) Knotts, T. A.; Rathore, N.; de Pabloz, J. J. An entropic perspective of protein stability on surfaces. Biophys. J. 2008, 94 (11), 4473−4483. (36) Wei, S.; Knotts, T. A. Predicting stability of alpha-helical, orthogonal-bundle proteins on surfaces. J. Chem. Phys. 2010, 133, 115102. (37) Xie, Y.; Zhou, J.; Jiang, S. Y. Parallel tempering Monte Carlo simulations of lysozyme orientation on charged surfaces. J. Chem. Phys. 2010, 132 (6), 065101. (38) Bauer, M. C.; Xue, W. F.; Linse, S. Protein GB1 folding and assembly from structural elements. Int. J. Mol. Sci. 2009, 10 (4), 1552− 1566. (39) Wu, X. M.; Yang, G.; Zhou, L. J. Identifying the intermediates during the folding/unfolding of protein GB1 with MD simulations. Theor. Chem. Acc. 2012, 131 (5), 1229. (40) Gallagher, T.; Alexander, P.; Bryan, P.; Gilliland, G. L. Two crystal structures of the B1 immunoglobulin-binding domain of streptococcal protein G and comparison with NMR. Biochemistry-U.S.. 1994, 33 (15), 4721−4729. (41) Hess, B.; van der Spoel, D.; Lindahl, E.; Apol, E.; Apostolov, R.; Berendsen, H. J. C.; van Buuren, A.; Bjelkmar, P.; van Drunen, R.; Feenstra , A.; Groenhof, G.; Kasson, P.; Larsson, P.; Meulenhoff, P.; Murtola, T.; Pall, S.; Pronk, S.; Schultz, R.; Shirts, M.; Sijbers, A.; Tieleman, P.Gromacs User Manual v.4.5; RIT: Uppsala, Sweden, 2010 (42) DeLano, W. L. The PyMOL Molecular Graphics System; DeLano Scientific: San Carlos, CA, 2002; http://www.pymol.org. H
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
Langmuir
Article
(43) Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.; Jorgensen, W. L. Evaluation and reparametrization of the OPLS-AA force field for proteins via comparison with accurate quantum chemical calculations on peptides. J. Phys. Chem. B 2001, 105 (28), 6474−6487. (44) Kony, D.; Damm, W.; Stoll, S.; van Gunsteren, W. F. An improved OPLS-AA force field for carbohydrates. J. Comput. Chem. 2002, 23 (15), 1416−1429. (45) Chatterjee, S.; Debenedetti, P. G.; Stillinger, F. H.; Lynden-Bell, R. M. A computational investigation of thermodynamics, structure, dynamics and solvation behavior in modified water models. J. Chem. Phys. 2008, 128 (12), 124511. (46) Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A 1985, 31 (3), 1695−1697. (47) Nose, S. A unified formulation of the constant temperature molecular dynamics methods. J. Chem. Phys. 1984, 81 (1), 511−519. (48) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. LINCS: A linear constraint solver for molecular simulations. J. Comput. Chem. 1997, 18 (12), 1463−1472. (49) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Hsing, L.; Pedersen, L. G. A smooth particle mesh Ewald method. J. Chem. Phys. 1995, 103 (19), 8577−8593. (50) Humphrey, W.; Dalke, A.; Schulten, K. VMD: visual molecular dynamics. J. Mol. Graphics 1996, 14 (1), 33−38. (51) Ravichandran, S.; Madura, J. D.; Talbot, J. A Brownian dynamics study of the initial stages of hen egg-white lysozyme adsorption at a solid interface. J. Phys. Chem. B 2001, 105 (17), 3610−3613. (52) Zhou, J.; Tsao, H. K.; Sheng, Y. J.; Jiang, S. Y. Monte Carlo simulations of antibody adsorption and orientation on charged surfaces. J. Chem. Phys. 2004, 121 (2), 1050−1057. (53) Chen, S. F.; Liu, L. Y.; Zhou, J.; Jiang, S. Y. Controlling antibody orientation on charged self-assembled monolayers. Langmuir 2003, 19 (7), 2859−2864. (54) Sheinerman, F. B.; Brooks, C. L. A molecular dynamics simulation study of segment B1 of protein G. Proteins 1997, 29 (2), 193−202. (55) Kabsch, W.; Sander, C. Dictionary of protein secondary structure: Pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 1983, 22 (12), 2577−637. (56) Guss, B.; Eliasson, M.; Olsson, A.; Uhlen, M.; Frej, A. K.; Jornvall, H.; Flock, J. I.; Lindberg, M. Structure of the IgG-binding regions of streptococcal protein G. EMBO J. 1986, 5 (7), 1567−1575. (57) Kato, K.; Lian, L. Y.; Barsukov, I. L.; Derrick, J. P.; Kim, H.; Tanaka, R.; Yoshino, A.; Shiraishi, M.; Shimada, I.; Arata, Y. Model for the complex between protein G and an antibody Fc fragment in solution. Structure 1995, 3 (1), 79−85. (58) Sloan, D. J.; Hellinga, H. W. Dissection of the protein G B1 domain binding site for human IgG Fc fragment. Protein Sci. 1999, 8 (8), 1643−1648. (59) Sauer-Eriksson, A. E.; Kleywegt, G. J.; Uhlen, M.; Jones, T. A. Crystal structure of the C2 fragment of streptococcal protein G in complex with the Fc domain of human IgG. Structure 1995, 3 (3), 265−278. (60) Wiseman, M. E.; Frank, C. W. Antibody adsorption and orientation on hydrophobic surfaces. Langmuir 2012, 28 (3), 1765− 1774.
I
dx.doi.org/10.1021/la401171v | Langmuir XXXX, XXX, XXX−XXX
