Article pubs.acs.org/Langmuir
Biomimetic Design of Platelet Adhesion Inhibitors to Block Integrin α2β1-Collagen Interactions: I. Construction of an Affinity Binding Model Lin Zhang and Yan Sun* Department of Biochemical Engineering and Key Laboratory of Systems Bioengineering of the Ministry of Education, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, People’s Republic of China Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin 300072, People’s Republic of China S Supporting Information *
ABSTRACT: Platelet adhesion on a collagen surface through integrin α2β1 has been proven to be significant for the formation of arterial thrombus. However, the molecular determinants mediating the integrin−collagen complex remain unclear. In the present study, the dynamics of integrin−collagen binding and molecular interactions were investigated using molecular dynamics (MD) simulations and molecular mechanics−Poisson−Boltzmann surface area (MM-PBSA) analysis. Hydrophobic interaction is identified as the major driving force for the formation of the integrin−collagen complex. On the basis of the MD simulation and MM-PBSA results, an affinity binding model (ABM) of integrin for collagen is constructed; it is composed of five residues, including Y157, N154, S155, R288, and L220. The ABM has been proven to capture the major binding motif contributing 84.8% of the total binding free energy. On the basis of the ABM, we expect to establish a biomimetic design strategy of platelet adhesion inhibitors, which would be beneficial for the development of potent peptide-based drugs for thrombotic diseases.
1. INTRODUCTION
hydrophobic contacts with Y157 and L286, the conservation of arginine by E256, and direct H-bonding interactions from N154, Y157, and H258. The arginine and glutamate in collagen were also identified as important. Siebert et al.16 investigated the binding of α2A on a collagen surface using atom force microscopy (AFM), surface plasmon resonance (SPR), and molecular dynamics (MD) simulations. The great importance of the arginine residue in the collagen was identified. Zaman17 performed MD simulations to examine the contribution of individual residues of integrin for its binding on a collagen surface, focusing on the complementary region in the integrin (including residues 151−159, 215−222, 255−261, and 280− 290). The results showed that denatured collagen induced helical conformations in integrin and significantly reduced the polyproline II content, contributing to the stabilization of
Platelet adhesion on exposed collagen is critical for thrombus formation,1−4 which leads to many arterial diseases with high mortality,2,5 such as stroke and acute myocardial infarction. Integrin α2β1 is an important collagen receptor on platelets. Its binding to collagen surfaces has been proven to be important in platelet adhesion and subsequent thrombus formation.6−11 Many efforts have been focused on the integrin−collagen interface and involved molecular interactions, using both experimental approaches and molecular simulations.11−13 Kamata et al.14 proposed a docking model of the α2 I-domain of integrin (α2A) in complex with collagen based on the crystal structures, which included Y157, N154, D219, H258, and Y285. Emsley et al.15 determined the crystal structure of α2A on a collagen surface and performed comprehensive investigations of the complementary region. The driving force for binding and involved important residues in integrin were identified on the basis of the structure analysis, which included salt bridges formed by D219, surface dimples formed by Q215 and N154, © 2014 American Chemical Society
Received: November 29, 2013 Revised: March 22, 2014 Published: April 3, 2014 4725
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
collagen) were first placed in the center of a cubic box (10 × 7 × 8 nm3 for Com and 10 × 7 × 9.5 nm3 for Sep). Then, solvent molecules were added randomly, followed by the neutralization of the system by adding Na+ as the counterion. The numbers of ions and water molecules for each system are summarized in Table 1.
integrin−collagen interactions. Recently, Attwood et al.18 measured the interactions between recombinant α2A and a triple-helical collagen using AFM. The studies mentioned above have provided extensive analyses of the binding of integrin on collagen surfaces. However, the understanding of molecular determinants mediating the binding is still far from adequate. Clarification of the molecular details involved in the binding of integrin on collagen surfaces is urgent, especially for the design of inhibitors for suppressing the binding and consequent platelet adhesion. MD simulation19,20 has been widely used to examine the protein behaviors at interface, especially the molecular interactions and orientation revealed in the binding process.21−29 Moreover, this coupled with molecular mechanics− Poisson−Boltzmann surface area (MM-PBSA) analysis has been proven to be a powerful tool with many successful applications,30−32 which can provide a clear description of the binding dynamics and can also calculate the binding free energy in a thermodynamic sense. Moreover, it can evaluate the favorable/unfavorable contribution of individual residues and provide a quantitative evaluation of their relative importance for the binding. These advantages make MD simulation coupled with MM-PBSA analysis a suitable tool for the examination of molecular determinants, as shown in our previous work on the VWF A1 domain-GPIbα complex33 and a capsomere of viruslike particles.34 In the present study, MD simulationcoupled MM-PBSA analysis was performed to investigate the binding of integrin α2β1 on a collagen surface. The microscopy process and involved molecular interactions were examined to provide fundamental insights into platelet adhesion induced by integrin α2β1. Our purpose in this work is to establish an affinity binding model (ABM) of integrin for its binding on a collagen surface. On the basis of the ABM, an affinity peptide library could be constructed for the screening of high-affinity inhibitors. We call this biomimetic design of platelet adhesion inhibitors. Part I of this work is concerned with the construction of ABM, and part II of this work will report the library design and screening of inhibitors.
Table 1. Atom Numbers in Each Simulation System system
number of Na+ ions
number of Cl− ions
number of water
Com Sep
53 63
52 62
18 729 22 241
2.2. Molecular Dynamics Simulations. MD simulations in the NVT ensemble were performed using GROMACS 4.5.336 (http://www.gromacs.org/). Temperature was controlled at 298.15 K16,17 by the velocity-rescale (v-rescale) method37 with a time constant of 0.5 ps. The linear constraint solver (LINCS) algorithm38 was applied to constrain all bonds. A periodic boundary was used in the x, y, and z directions. A particle-mesh Ewald (PME) algorithm39 was used to deal with the electrostatic interaction. The cutoffs of the neighboring atom list, Lennard-Jones (LJ) potential, and Coulomb potential energies were all set to 0.9 nm. The initial velocities of particles were generated according to a Maxwell distribution. A Verlet algorithm was used for integration with a time step of 2 fs. Prior to simulation, each system was subjected to 10 000 steps of steepest descent energy minimization. Then a 100 ns MD simulation was performed as described above, storing five million coordinates every 0.02 ps for the following analysis. Four independent simulations were performed for each set of condition. 2.3. Molecular Interaction Analysis. Potential energies between integrin and collagen were calculated using the g_energy program in GROMACS to describe the molecular interactions. The minimum distance between integrin and collagen (denoted as dmin) was calculated using the g_mindist program in GROMACS to describe the binding/dissociation. 2.4. Conformational Analysis. The root mean square deviation (RMSD) and radius of gyration (Rg) were calculated to evaluate the protein conformation quantitatively. RMSD representing the structural changes in a protein at time t with respect to a reference structure (herein the initial structure is used) was calculated using the g_rms program in GROMACS. Rg representing the compactness of the protein conformation was calculated using the g_gyrate program in GROMACS. In order to describe the secondary structure of protein molecules, define secondary structure of proteins (DSSP) analysis was performed using the do_dssp program in GROMACS. The lipophilic potential and electrostatic potential along the molecular surfaces of proteins were visualized using the MOLCAD program (Tripos, http://www.tripos.com/) from the SYBYL package. 2.5. MM-PBSA Analysis. The binding free energy (ΔGbind) between integrin and collagen was calculated using the MMPBSA method33,34 and CHARMM40 (http://www.charmm-gui. org/), following the procedure reported previously33,34 with minor modification. Eleven snapshots were extracted from the last 10 ns trajectory of each simulation at an interval of 1 ns for MM-PBSA analysis. ΔGbind was calculated with eq 1.
2. MODELS AND METHODS 2.1. Model Construction. All-atom (AA) models of integrin α2β1 and collagen were constructed on the basis of the crystal structure in the Protein Data Bank (PDB ID 1DZI, http://www.rcsb.org/pdb/)15 and the CHARMM27 force field. The parameters for hydroxyproline were taken from the literature.35 Lysine and arginine were positively charged with a charge number of +1 each, while glutamic acid and aspartic acid were negatively charged with a charge number of −1 each. Then the total charge number in integrin is −1 while the collagen is neutral (pH 7.0). The proteins were dissolved in physiological saline, where the water molecule was treated using the TIP3P model and Na+ and Cl− were considered to be charged beads. Two simulation systems were constructed using these models, including the complex (combined integrin and collagen, denoted as Com) and separated integrin and collagen (with an initial distance of 3.7 nm between the centers of mass of two proteins, denoted as Sep). Herein, the orientation of two proteins in the crystal structure was used in the initial structure in Sep, while the initial distance was adjusted to make two proteins completely separated but not far from each other to reduce the computational cost according to our previous work.33 The proteins (complex or separated integrin and
ΔG bind = ΔGgas + ΔGsol − T ΔS 4726
(1)
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
Figure 1. Dynamic behavior of the complex and the binding of integrin on collagen. The time courses of dmin, ELJ, and EC are shown in (a), and the corresponding snapshots at the time points marked in (a) are shown in (b).
ΔGgas = ΔGelec + ΔGvdW
(2)
ΔGsol = ΔG PB + ΔGnp
(3)
Gnp = γ × SASA + b
(4)
interaction and ΔGnonpolar is the contribution from the hydrophobic interaction.33
where brackets ⟨...⟩ indicate an average of an energy term along the MD simulation trajectory. −TΔS is the entropic contribution arising from changes in the degrees of freedom (translational, rotational, and vibrational) of proteins, which requires large computational resources for calculation. Herein, the analysis is primarily concerned with per-residue electrostatic and hydrophobic contributions rather than the absolute binding free energy. Then the relative free energy was calculated without a consideration of −TΔS such as that in the recent literature41 and our previous work.32−34 ΔGgas is the sum of ΔGelec and ΔGvdW (eq 2). The solvation energy contains the electrostatic solvation energy (ΔGPB) and the nonpolar solvation energy (ΔGnp) (eq 3). ΔGPB was calculated by solving the linear Poisson−Boltzmann (PB) equation using the PBEQ module of the CHARMM program. The solute and solvent dielectric constants were set to 1 and 80, respectively. ΔGnp was calculated according to eq 4, where constants γ and b were set to 0.00542 kcal/(mol Å2) and 0.92 kcal/mol, respectively.42 The solvent-accessible surface area (SASA) was calculated using a water probe radius of 1.4 Å. The free-energy contribution of each residue can be divided into polar (ΔGpolar) and nonpolar interactions (ΔGnonpolar) according to eq 5, and each part is the sum of two energy terms, as given in eqs 6 and 7. In the following analysis, ΔGpolar is considered to be the contribution from the electrostatic
ΔG bind = ΔGpolar + ΔGnonpolar
(5)
ΔGpolar = ΔGelec + ΔG PB
(6)
ΔGnonpolar = ΔGvdW + ΔGnp
(7)
2.6. Construction of Affinity Binding Model. Residues making significant contributions to the binding free energy (with a criterion of ±2.5 kcal/mol) were considered to be the hot spots.43 To account for the binding affinity of integrin on collagen, an ABM of integrin was constructed on the basis of the most important residues favorable for binding in hot spots.
3. RESULTS AND DISCUSSION 3.1. Molecular Behavior of the Complex. The stability of the integrin−collagen complex and involved molecular interactions were examined through 100 ns MD simulation, as shown in Figure 1, where the data from four independent simulation trajectories are presented. Figure 1a shows the time courses of dmin, ELJ, and EC between integrin and collagen. Corresponding snapshots marked at different time points in Figure 1a are shown in Figure 1b. A stable integrin−collagen complex is observed as dmin remains at 0.17 nm during the simulation (see the black curve in Figure 1a). However, ELJ shows a slight increase at the beginning of simulation and then reaches a plateau with small fluctuations around −150 kJ/mol, while EC shows significant fluctuations until the end of simulation. That is, although the combined state of integrin and collagen is retained as indicated by dmin, there is minor structural arrangement attributed to the 4727
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
the electrostatic potential (Figure 2b), patches of opposite charges are observed in the complementary region of two proteins, where small positively charged patches are observed on integrin and negatively charged patches are observed on collagen, leading to favorable electrostatic attraction between the two proteins. Under these favorable interactions, the integrin binds firmly to the triple-helical structure of the collagen (Figure 1b), which is consistent with the simulation results reported elsewhere.15,16 Herein, it should be noted that the metal ion in the crystal structure was deleted in our simulations. The metal ion is proven necessary for complex formation by extensive experimental results.15 However, the stable complex without a metal ion is also maintained during the long-time MD simulations in the present study. The possible reason is that the major function of the metal ion is to induce the formation of specific structure for complex formation. Once the specific structure for binding is formed, the metal ion can be removed. In our simulations, the crystal structure of the complex is used as the initial structure, where the specific structure for binding is already formed in the presence of the metal ion. Then, if the structure of integrin is well maintained during the simulation, then the metal ion is not necessary. To verify this hypothesis, the conformational transition was further examined using RMSD and Rg (Figure 3). No significant change in RMSD is observed for integrin, although a slight increase of 0.2 nm is shown at the beginning of simulation. The increase is very small in consideration of its large molecular size. Meanwhile, a steady Rg of integrin is observed. These results confirm the rare conformational transition of integrin during the simulation. Moreover, the secondary structure evolution of integrin during the dynamics process was determined by DSSP, as shown in Figure 4a. No significant change in secondary structure is observed, which is consistent with the snapshots in Figure 1b. Moreover, little deviation of the final structure of integrin is observed after 100 ns of MD simulation, as compared to the initial structure (Figure 4a). Therefore, a well-maintained conformation of integrin can be confirmed. Similarly, a relative stable structure of collagen is revealed from RMSD and Rg, which can be attributed to its triple-helical structure (Figure 1b). This is helpful for complex formation. Therefore, when the crystal structure is used as the initial structure, a stable complex is maintained without the metal ion.
solution exchange and the thermal motion. In this process, both negative ELJ and EC are observed, indicating that both hydrophobic and electrostatic interactions between the two proteins are favorable to complex formation. This can be further verified using the lipophilic/electrostatic potential along the molecular surface (Figure 2).
Figure 2. The molecular surface of integrin−collagen complex colored according to the lipophilic potential (a) and the electrostatic potential (b). For the lipophilic potential surface, brown is hydrophobic and blue is hydrophilic. For the electrostatic potential surface, red is positive and purple is negative. The complementary region is shown on the right. The crystal structure from the PDB is used. The images were prepared using the MOLCAD program (http://www.tripos. com/).
Dominant hydrophobic patches are observed on both the surface and the complementary region of the complex (Figure 2a), leading to hydrophobic interaction favorable to complex formation, which accounts for the negative ELJ in Figure 1a. For
Figure 3. RMSD from the crystal structure and Rg of integrin and collagen as a function of simulation time for the complex (a) and binding process (b). 4728
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
Figure 4. Secondary structure evolution of integrin during the simulation on the complex (a) and the binding process of integrin on collagen (b). The final structure of integrin after MD simulation is compared to its initial structure and is shown on the right.
Table 2. Binding Free Energies (kcal/mol) of Integrin α2β1 on Collagen
a
system
ΔGvdW
ΔGnp
ΔGelec
ΔGPB
ΔGnonpolara
ΔGpolarb
ΔGbindc
Com Sep
−40 ± 2 −32 ± 2
−6 ± 0 −5 ± 0
−104 ± 11 −124 ± 8
129 ± 15 140 ± 9
−47 ± 3 −37 ± 2
25 ± 12 16 ± 6
−21 ± 11 −21 ± 5
ΔGnonpolar=ΔGvdW + ΔGnp, hydrophobic interaction energy; bΔGpolar=ΔGelec + ΔGPB, electrostatic interaction energy; cΔGbind =ΔGnonpolar +ΔGpolar.
However, the potential energies are still higher than those of the complex, indicating that the binding is not perfect, leading to a minor adjustment in binding and slow decreases in both ELJ and EC in the following simulation. At 100 ns, ELJ is still a little higher than that of the complex but a similar EC is achieved. A binding structure similar to the complex is observed (Figure 1b), confirming successful binding. The conformational transition of protein molecules in the binding process was also examined. A rare conformational transition of integrin can be confirmed by steady RMSD and Rg (Figure 3) and a well-maintained structure (Figure 4b). Then the favorable interaction between integrin and collagen leads to successful binding. An additional simulation trajectory of the binding process containing the metal ion was provided in Figure S1, where successful binding is also observed. Therefore, on the basis of the MD simulation trajectories, the molecular interactions between integrin and collagen can be clarified in detail using MM-PBSA analysis, which is the basis for the design of inhibitors to suppress the binding of integrin on collagen surfaces. 3.3. Binding Free Energy Analysis. The MM-PBSA method was used to calculate the binding free energy between integrin and collagen, and the results are shown in Table 2. Herein, 11 conformations were extracted from the last 10 ns of each simulation trajectory at an interval of 1 ns. The average of
An additional simulation trajectory of the complex with a metal ion is provided in Figure S1 in the Supporting Information, where a similar stable complex is observed. The simulation of integrin−collagen interactions without a metal ion has also been reported elsewhere.16,17 Moreover, the binding process of separated integrin on a collagen surface was examined without a metal ion for further verification. 3.2. Binding Process of Integrin on Collagen. MD simulation (100 ns) was performed on separated integrin and collagen to examine the binding dynamics of integrin on a collagen surface. The results are shown in Figure 1 using the data from four independent simulation trajectories. A decrease of dmin is observed at the beginning of the simulation (see the red curve in Figure 1a), indicating that the integrin moves back to the collagen. This movement can be attributed to the attraction by collagen, indicated by slight decreases in both ELJ and EC. However, the attraction is not strong enough for the formation of a stable complex. The integrin moves away from the collagen surface with additional structure arrangement in the neighborhood of collagen. For example, the integrin moves toward the collagen at 10 ns, but without a strong interaction as compared to the complex (the black curve), leading to further adjustment of the orientation. The integrin tries several times and finally binds firmly after 20 ns, as indicated by a small dmin equal to that of the complex. 4729
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
Figure 5. Binding free energy contribution of each residue in integrin and collagen, including the residues making significant contributions to the complex (a) and the binding process (b). Residues with |ΔGbind| ≥ 0.5 kcal/mol are shown. The error bars indicate the standard deviations. O is hydroxyproline.
residues. The analysis of the complementary region can identify that these residues are important to complex formation. MMPBSA analysis can further point out that not all of them are favorable for the complex, which is important for the regulation of integrin−collagen interactions and is crucial for the construction of ABM. For example, D254 is an important residue located at the interface between integrin and collagen, but it is unfavorable for binding as indicated by the MM-PBSA analysis. It has been reported that residues D151, S153, T221, and D254, locating at metal ion-dependent adhesion site motif, are required for collagen binding to the intact integrin.15 However, from the MM-PBSA analysis, it can be seen that some of these residues are not necessary or are even unfavorable to maintaining the complex once the complex is formed. Second, MM-PBSA analysis can provide a quantitative evaluation of the relative importance of each residue. For example, both Y157 and G217 are favorable for the binding but Y157 makes a larger contribution as indicated by a lower ΔGbind. Therefore, the MM-PBSA method should be used to evaluate the function and relative contribution of each residue, which is crucial to the construction of ABM and the improvement of the binding affinity. Figure 5a shows that in integrin the most important residue favorable to complex formation is hydrophobic Y157. In its counterpart, collagen, there are several hydrophobic residues, for instance, F355. These residues can provide favorable hydrophobic interaction, which is consistent with the analysis of the lipophilic potential surface in Figure 2 and the binding free energy analysis in Table 2. For the electrostatic interaction, the most important residues in collagen are mainly positively
data from four independent simulation trajectories was used for discussion. In the integrin−collagen complex, both negative ΔGvdW and ΔGnp are observed, leading to a negative ΔGnonpolar of −47 kcal/mol, indicating a favorable hydrophobic interaction between the two proteins. Meanwhile, negative ΔGelec is observed, indicating electrostatic attraction between the two proteins, which is consistent with the negative EC in Figure 1. Moreover, the MM-PBSA analysis provides an actual evaluation of the electrostatic interaction by taking into account the presence of solution. It can be seen that the negative ΔGelec is diminished by a positive ΔGPB, leading to a positive ΔGpolar. That is, in consideration of the solution, the electrostatic interaction turns out to be unfavorable to complex formation. For further in-depth discussion, the residues making large free-energy contributions to complex formation were identified, as shown in Figure 5a. The criterion of ±0.5 kcal/mol was employed. On the basis of the quantitative evaluation of each residue, the important residues in integrin are identified, including Y157, N154, L286, Q215, D219, and E256. These important residues are consistent with the complementary region analysis based on the crystal structure, where salt bridges formed by D219, surface dimples formed by Q215 and N154, hydrophobic contacts with Y157 and L286, the conservation of arginine by E256, and direct H-bonding interactions from N154 and Y157 are considered to be important for binding.15 Moreover, the MM-PBSA analysis in the present study can provide more details. First, it can differentiate the favorable residues and the unfavorable ones among these important 4730
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
Figure 6. (a) Construction of the affinity binding model of integrin. Collagen is shown in the surface model while integrin is shown as a ribbon structure. The residue is colored by its contribution to the binding free energy, which ranges from red (most negative) to blue (most positive). The residues in the affinity binding model are shown as a licorice structure in (b). The α-carbon atom of each residue is marked by a yellow bead. The distances between residues (in angstrom) are marked in blue in (c). The figures are prepared using VMD software (http://www.ks.uiuc.edu/ Research/vmd/).
The residues making large free-energy contributions to binding (Sep) were also identified, as shown in Figure 5b. The free-energy values are generally smaller than those in the complex (Com), which is consistent with the analysis in Table 2. Moreover, unlike that in Com, a wider distribution of important residues is observed in collagen. However, the residues with large free-energy contributions are all located at the motif GFOGER. The other residues make only minor contributions, according to the value of ΔG (Figure 5b). The wider distribution is attributed to the dynamics of the binding process. That is, once the integrin approaches the collagen, possible binding can occur at several sites due to the repeated sequence in the collagen along its triple-helical structure. However, the most favorable binding site is the motif GFOGER, including R380, R338, and O356. As in Com, important hydrophobic residues are observed in both integrin and collagen, for example, Y157 in integrin and F376 in collagen, contributing to the favorable hydrophobic interaction for binding. Meanwhile, the most unfavorable residues are mainly negatively charged residues, such as E256, D151, and D254 in integrin and E379 in collagen. For example, E379, near E256 in collagen, has a free-energy contribution of up to 3.5 kcal/mol due to unfavorable electrostatic repulsion. Therefore, the electrostatic repulsion between these residues leads to the final unfavorable contribution of electrostatic interaction. Moreover, these residues with unfavorable contributions should be deleted in the construction of ABM to improve the binding affinity. 3.4. Construction of the Affinity Binding Model. An ABM of integrin was constructed based on the free-energy decomposition. Herein, the binding motif of the integrin rather than the collagen was used. Because exposing collagen is the first step and the trigger in thrombus formation, covering the exposed collagen is a good choice for inhibiting platelet adhesion and consequent thrombus formation. As shown in Table 2 and Figure 5, similar results are observed in Com and Sep, but stronger interaction is observed in the former. In addition, MM-PBSA analysis has been performed on the
charged, including R359 and R338, which is consistent with previous simulation results15,16 and experimental results.15,44,45 In its counterpart, there are several negatively charged residues, for example, E256, D254, D151, and D219, leading to a favorable contribution to the electrostatic interaction. In collagen, the most important residues favorable to complex formation (R359, R338, F355, O335, and F334) are all located at the motif GFOGER, which is consistent with experimental results.15,44,45 Herein, MM-PBSA analysis can provide a clear evaluation of each residue. For example, the importance of arginine and glutamate to the binding has been experimentally determined and reported.15,16 However, from Table 2, the electrostatic interaction is unfavorable as indicated by a positive ΔGpolar. Why? The analysis should be moved to the unfavorable residues (with a large positive free-energy contribution) in these two proteins. The most unfavorable residues are E256, D254, and D151 in integrin and E358 and E379 in collagen, which are all negatively charged residues. Then the electrostatic repulsion between these residues causes an unfavorable contribution. For example, E358, near E256 and D254 in collagen, has a free-energy contribution of up to 20.2 kcal/mol due to the electrostatic repulsion. Similar results are observed in the binding process, which starts from separated integrin and collagen (Sep). From Table 2, both negative ΔGvdW and ΔGnp are observed, leading to a negative ΔGnonpolar of −37 kcal/mol, indicating a favorable hydrophobic interaction. Meanwhile, negative ΔGelec is diminished by positive ΔGPB, leading to a positive ΔGpolar of 16 kcal/mol. Therefore, the hydrophobic interaction is the major driving force in binding. It should be noted that although similar results are observed in Sep and Com, the value of each term in Sep is smaller than that in Com, as a result of weaker binding than in the complex (see the larger ELJ of Sep compared to that of Com in Figure 1a). That is, both favorable hydrophobic interactions and unfavorable electrostatic interactions are diminished in Sep, leading to the same ΔGbind of −21 kcal/mol as in Com. 4731
dx.doi.org/10.1021/la404599s | Langmuir 2014, 30, 4725−4733
Langmuir
Article
collagen and to explore the molecular basis in the binding. A stable integrin−collagen complex as well as the successful binding of integrin on collagen surface is verified using MD simulations. Thereafter, MM-PBSA analysis provides more details involved in the integrin−collagen complex. Hydrophobic interaction is identified as the major driving force. On the basis of these results, an ABM of integrin is constructed, consisting of Y157, N154, S155, R288, and L220. The ABM contributes 84.8% of the total binding free energy of the integrin−collagen complex and thus captures the major binding motif of integrin. On the basis of the ABM, a biomimetic design strategy of platelet adhesion inhibitors will be established and described in part II, which would be helpful for the regulation of integrin−collagen interactions and the development of novel peptide-based drugs for thrombotic diseases.
complex and binding process in the presence of a metal ion, as shown in Figure S2. The inclusion of a metal ion indeed causes some minor differences in the binding free energy contributions of each residue, especially for Sep. Then, the results in Com are mainly used to construct the model, with a secondary consideration of the important residues in Sep. It is also shown that some important residues with the same charge are in close proximity at the interface between integrin and collagen, leading to an unfavorable free-energy contribution as a result of electrostatic repulsion. Reversing the charge may be a good way to make the electrostatic interaction favorable and improve the binding affinity but may induce new unfavorable repulsion with other charged residues. For example, reversing the charge of E256 to R256 is helpful in diminishing the repulsion from E358 but may cause a new repulsion from R338. Therefore, these residues with unfavorable contributions are not adjusted but are deleted in the construction of the ABM. Only the residues with the most favorable contributions (
