Feng
Wang
a,
Hua
Wan
b,
Jian-ping
Hu
c and
Shan
Chang
*b
aSchool of Information Science & Engineering, Changzhou University, Changzhou, China
bCollege of Informatics, South China Agricultural University, Guangzhou, China. E-mail: schang@scau.edu.cn
cFaculty of Biotechnology Industry, Chengdu University, Chengdu, China
First published on 21st October 2014
Botulinum neurotoxins (BoNTs) are known as the most poisonous biological substances, and they are also used to treat a wide range of medical conditions as well as in the cosmetic applications. Recently, the complex structures of the BoNT/A receptor-binding domain (BoNT/A-RBD) and the synaptic vesicle protein 2C luminal domain (SV2C-LD) were determined by X-ray crystallography. In this article, the wild type (WT) and four mutants of the new structure are studied by molecular dynamics (MD) simulations. The differently decreased structural stabilities of the mutants relative to WT are shown to be consistent with the experimental data of binding affinities. The conformational changes of the five systems are explored by using principal component analysis (PCA) and free energy landscape (FEL) methods. Based on the calculation of interactions at the binding interface, we divide the interface between BoNT/A-RBD and SV2C-LD into two crucial binding regions. Through the comparison of WT and four mutants, we further propose the relationship between the conformational changes of BoNT/A-RBD:SV2C-LD and the interfacial interactions. This study would provide some new insights into the understanding of the dynamics and the interaction mechanism of BoNT/A-RBD:SV2C-LD.
The X-ray structures of BoNTs promote our understanding of the properties and the mechanism of these toxins. The structures of BoNTs are composed of a light chain (LC, ∼50 kDa) and a heavy chain (HC, ∼100 kDa), which are linked by a disulfide bond.8 The light chain (residues 1–448) is a Zn2+-dependent metalloprotease. The heavy chain contains an N-terminal translocation domain (residues 449–872) and a C-terminal receptor-binding domain (residues 873–1295). In 1998, the crystal structure of the entire 1285 amino acid BoNT/A was determined at 3.3 Å resolution.9 In 2007, Allen et al. reported the crystal structures of BoNT/A-LC in complexes with two potent small-molecule inhibitors, and explored the active site and conformational flexibility of BoNT/A-LC.10 In 2008, Stevens et al. solved the crystal structure of the BoNT/A binding domain alone and in complex with the ganglioside GT1b at 1.7 Å and 1.6 Å, respectively.11 These important structural data provided some new clues to the functional study of BoNTs.
Theoretical studies, such as the coarse-grained models12–14 and the all-atom molecular dynamics (MD) simulation,15–17 have become important tools in exploring the dynamics and interaction mechanisms of proteins. These simulation methods have also been performed to investigate the interactions and conformational flexibility of BoNTs. Some groups have applied MD methods to explore the conformational flexibility of BoNT/A-LC and the inhibitor binding modes.18,19 The entire BoNT/A structure has also been studied by the long-time MD simulations and the properties of BoNT/A at various temperatures and pH values were analyzed to understand the toxicity and structural variations.20 The all-atom MD simulations have been used to investigate the conformational properties of the oligosaccharide structures of GD1A and GD1B in an aqueous environment.21,22 In addition, both the urea denaturation experiments and theoretical modeling were performed to understand the folding mechanism of BoNT/A.23
In the previous studies,24 the synaptic vesicle protein 2 (SV2) has been proposed to be a protein receptor of BoNT/A, but the structural details and theoretical studies of the interactions between BoNT/A and SV2 are relatively sparse. Recently, Kammerer et al. determined the high-resolution crystal structure of the BoNT/A receptor-binding domain (BoNT/A-RBD) in complex with the SV2C luminal domain (SV2C-LD).25 This study provides a strong structural basis for the interactions of BoNT/A and SV2. However, several questions still remain unclear. What specific interactions are formed at the interface between BoNT/A-RBD and SV2C-LD? How do the different mutants reduce the binding affinities? Do the conformational changes occur when the interface residues of BoNT/A-RBD:SV2C-LD are mutated? What is the relationship between the conformational changes and the interfacial interactions?
In this article, in order to probe the above issues, the new crystal structure of BoNT/A-RBD:SV2C-LD and four relevant mutant systems were analyzed by MD simulations. The principal component analysis (PCA) and free energy landscape (FEL) methods were applied to explore the functional dynamics and conformational changes of the five systems. We investigated the crucial interfacial interactions of BoNT/A-RBD:SV2C-LD and explained the loss of binding affinities of four mutants relative to the wild type. Finally, we further suggested the relationship between the conformational changes of BoNT/A-RBD:SV2C-LD and the interfacial interactions.
(1) |
The free energy landscape (FEL) can promote an understanding of the dynamic process occurring in a biological system.33,35 In the FEL, the free energy minima usually represent the conformational ensemble in the stable states while the free energy barriers denote the transient states.36 The FEL was constructed on the basis of the above PCA data.32 The corresponding expression is:
ΔG(X) = −KBTlnP(X) | (2) |
In order to analyze the relative binding energy between BoNT/A-RBD and SV2C-LD, the program g_mmpbsa41 was applied to the five systems. It implements the molecular mechanics Poisson–Boltzmann surface area (MM-PBSA) approach to estimate interaction free energies. This program can calculate molecular mechanics potential energy (including both electrostatic and van der Waals interactions) and free energy of solvation (including polar and nonpolar solvation energies). Although the entropy contribution was not included in the calculations, the binding energy calculated using g_mmpbsa showed an apparent correlation of 0.80–0.85 with the experimental binding free energy in the previous study.41
The flexibility of each residue is assessed by its root mean square fluctuation (RMSF). Fig. 3C and D show the RMSF values of BoNT/A-RBD and SV2C-LD calculated from 12 to 42 ns trajectories, respectively. For the five systems, the middle part of the C-terminal BoNT/A-RB subdomain (residues K1170–I1250) and the N-terminal of SV2C-LD (residues V473–N480) exhibit large fluctuation values. In contrast, the binding site residues (see Fig. 1) of BoNT/A-RBD:SV2C-LD show relatively low fluctuation values. For the WT system, the RMSF values of BoNT/A-RBD and SV2C-LD are lower than those of the mutant systems. For the R1294A, R1156E and TT1145/6AA systems, the mutations lie in BoNT/A-RBD, so these mutations mainly lead to the high fluctuation of BoNT/A-RBD. Similarly, the F563A mutation increases the fluctuation of SV2C-LD. Notably, the R1156E mutation also causes the RMSF increase of SV2C-LD. This result is consistent with the above RMSD analysis. The mutation of R1156E induces the electrostatic repulsion, which will affect the stability of both BoNT/A-RBD and SV2C-LD.
In the mutant systems, the motion modes are different from those of the WT system. As shown in Fig. S1A (ESI†), the N-terminal BoNT/A-RB subdomain in the R1294A system has larger movement than the WT system, so the RMSF values of some parts in the N-terminal BoNT/A-RB subdomain are increased remarkably as shown in Fig. 3C. In PC1 and PC2 modes, SV2C-LD and the N-terminal BoNT/A-RB subdomain have the opposite motions. The free energy landscape of the R1294A system also has two local basins, but the magnitudes of the PC1 and PC2 motions are larger than those of the WT system.
As shown in Fig. S1B (ESI†), since the mutant residue F563 lies in SV2C-LD, the average structure of SV2C-LD changed remarkably in the F563A system. In the PC1 mode, the C-terminal BoNT/A-RB subdomain has the opposite movement with SV2C-LD. The free energy landscape of the F563A system also has two local basins, but most of the conformations are located in the negative direction of PC1 and the positive direction of PC2.
In the R1156E system, the average structure has large changes at the binding site (see Fig. S1C, ESI†). The left side of SV2C-LD moves away from BoNT/A-RBD. In the PC1 and PC2 modes, SV2C-LD has the opposite movements against the C-terminal and the N-terminal subdomains of BoNT/A-RB, respectively. The free energy landscape of the R1156E system has some basins in the negative direction of the PC1 mode, so SV2C-LD moves anticlockwise and the left side will be far away from BoNT/A-RBD.
Similar to R1156E, the average structure of TT1145/6AA also has large changes at the binding site, but the position is on the right side (see Fig. S1D, ESI†). In the PC1 mode, BoNT/A-RBD rotates anticlockwise and SV2C-LD moves clockwise. The free energy landscape of the TT1145/6AA system has one basin in the positive direction of the PC1 mode, so SV2C-LD moves away from BoNT/A-RBD on the right side.
Similar to the WT system, Fig. 5 shows the representative structures of mutant systems in different energy basins compared with the crystal structure. As shown in Fig. 5A, the two representative conformations of R1294A have large conformational changes in the N-terminal BoNT/A-RB subdomain, but most of the interface residues are relatively stable. Compared to the crystal structure, the mutation site A1294 exhibits some fluctuations in the representative conformations. In the F563A system, besides BoNT/A-RB, SV2C-LD in the representative conformations also changes remarkably (see Fig. 5B). The interface between BoNT/A-RBD and SV2C-LD becomes unstable, especially at the mutation site A563. In the R1156E system, the representative structure has large changes at the binding site (see Fig. 5C). The left side of SV2C-LD moves away from BoNT/A-RBD. Similar to R1156E, the representative structure of TT1145/6AA also has conformational changes at the binding interface, but the position is on the right side (see Fig. 5D).
In order to take a consistent view of the conformational changes of WT and mutants, the mutant trajectories are also projected onto the first and second motion modes of WT. As shown in Fig. S2 (ESI†), each mutant system has only one basin in this free energy landscape. The local basins of the R1294A and F563A systems located in the negative direction of WT's PC2 mode, but the PC1 values are relatively small (see Fig. S2A and B, ESI†). Therefore, although the conformational changes of R1294A and F563A are different from that of WT, the fluctuations of SV2C-LD in these two systems are not remarkable. The basin of R1294A seems to be nearer to the local minimum of WT, so its average structure is more similar to the WT system. The local basin of the R1156E system is located in the negative direction of WT's PC1 mode, so the SV2C-LD moves away from BoNT/A-RBD on the left side (see Fig. S2C, ESI†). In contrast, the local basin of TT1145/6AA located in the positive direction of the PC1 mode (see Fig. S2D, ESI†). Then, SV2C-LD rotates clockwise and the right side is far away from BoNT/A-RBD.
From the above motion analysis, it is found that SV2C-LD has swing motions in all the five systems. In the WT system, this motion is reciprocating swing and the magnitude is small, so the structure remains stable. In R1294A and F563A, the mutation may break some interactions at the interface, and the motion magnitude increases. Fortunately, these systems also have two basins in the free energy landscape, so the average structures are similar to the WT system. In contrast, the R1156E and TT1145/6AA systems have only one basin or some basins with same motion directions in the free energy landscape. The mutations in these two systems may break the important interactions at the interface, so SV2C-LD only swings in the fixed direction. Hence, the average structures and the representative structures are different from those of the WT system, and then the binding affinities are reduced remarkably. It can be seen that these conformational changes in the four mutant systems are correlated with the mutation positions and the interactions at the interface. Then, the detailed interactions at the interface will be analyzed below.
WT | ||
---|---|---|
BONT/A-RBD | SV2C-LD | Occupancy (%) |
M1144-N | C560-O | 93.0 |
S1142-O | F562-N | 90.8 |
M1144-O | C560-N | 79.2 |
T1146-OG1 | F557-N | 78.8 |
S1142-N | F562-O | 71.6 |
T1146-N | F557-O | 54.8 |
T1145-OG1 | N559-N | 47.2 |
T1146-OG1 | E556-OE2 | 29.0 |
T1146-OG1 | E556-OE1 | 23.2 |
K951-NZ | N559-OD1 | 19.1 |
R1294-NH1 | D539-O | 16.4 |
Type | BONT/A-RBD | SV2C-LD | E elec | E vdW | Occupancy (%) |
---|---|---|---|---|---|
a E elec and EvdW are the electrostatic and van der Waals energies, respectively. The unit of energy is kcal mol−1. | |||||
WT | R1156 | F563 | −4.4 ± 1.2 | −3.2 ± 0.5 | 96.3 |
R1294A | R1156 | F563 | −4.7 ± 1.2 | −3.2 ± 0.6 | 96.6 |
F563A | — | — | — | — | — |
R1156E | R1294 | Y478 | −2.0 ± 0.6 | −1.5 ± 0.8 | 0.1 |
R1294 | F563 | −2.2 ± 0.9 | −1.6 ± 0.5 | 3.0 | |
TT1145/6AA | R1156 | F563 | −2.9 ± 0.8 | −2.3 ± 0.4 | 98.8 |
Table S1 (ESI†) shows the hydrogen bonds with an occupancy of over 10% in the four mutant systems. Interestingly, the mutant systems possess more hydrogen bonds compared to WT. The R1294A system only loses some low occupancy hydrogen bonds, and most of the high occupancy hydrogen bonds are maintained at the interface. The F563A and R1156E mutations mainly break the cation–π interaction, and have little influence on the high occupancy hydrogen bonds formed by S1142–T1146. Notably, the residue R1294 forms much more hydrogen bonds in F563A and R1156E systems. It implies that after the cation–π interaction is broken, R1294 contributes much more interactions to stabilize the systems. The TT1145/6AA mutation has a remarkable influence on the high occupancy hydrogen bonds. In this system, the side-chain hydrogen bonds of T1145 and T1146 are completely broken. Moreover, the residue M1144 lose a backbone hydrogen bond with C560, whereas R1156 forms many side-chain hydrogen bonds with N565 and D543.
Another important interaction at the interface is the cation–π interaction (see Table 2). The cation–π interaction between BoNT/A-RBD and SV2C-LD is mainly formed by residues R1156 and F563, and located on the left side of the interface. The F563A and R1156E mutations mainly break the cation–π interaction. In the F563A system, the cation–π interaction is completely lost at the interface. Nevertheless, in the R1156E system, R1294 forms the cation–π interactions with F563 or Y478, but both the energy and occupancy are very low. The WT, R1294A and TT1145/6AA systems still maintain the cation–π interactions. Although the occupancies are more than 96% in the three systems, the binding energies are definitely different. The WT and R1294A systems possess the similar binding energies, where Eelec < −4.0 kcal mol−1 and EvdW < −3.0 kcal mol−1. It implies that the cation–π interactions are strong in these two systems. In the R1294A system, the binding energy is even lower than that of the WT system. The position of R1294 is close to the cation–π interaction, so the cation–π interaction will also help in stabilizing the system after the hydrogen bond of R1294 is lost. Compared with other systems, the cation–π interaction has much higher occupancy in the TT1145/6AA system, but the energy is not favourable. As shown in Fig. 7, the angle in the TT1145/6AA system decreases to around 55°, and the distance also increases to more than 4.5 Å. The motion modes in Fig. S1D (ESI†) can give the explanation for these changes. SV2C-LD moves away from BoNT/A-RBD on the right side, which causes the changes in the angle and distance. In addition, R1156 forms some hydrogen bonds with N565 and D543 in the TT1145/6AA system (see Table S1, ESI†), so these wrong hydrogen bonds may also attract the side-chain of R1156 and then weaken the cation–π interaction.
Fig. 7 Time series of the distance and angle for the cation–π interactions in the WT (red), R1294A (purple) and TT1145/6AA (orange) systems. The smooth values are represented by the black curves. |
The binding energies of the five systems are calculated by g_mmpbsa and are shown in Table 3. The van der Waals energies of WT, R1294A and F563A are lower than those of R1156E and TT1145/6AA. In the R1156E and TT1145/6AA systems, SV2C-LD moves away from BoNT/A-RBD at the interface, so the van der Waals energies between the two proteins are increased. In the F563A and TT1145/6AA systems, the electrostatic energies are much lower than other systems. These two systems formed much more new hydrogen bonds at the interface, so electrostatic energies become more favourable. Notably, the R1156E structure also formed the new hydrogen bonds, but the electrostatic energy is higher compared to other systems. In the R1156E system, the positively charged residue R1156 is mutated to negatively charged glutamic acid. Then, this mutation causes the electrostatic repulsion at the interface, so the electrostatic energy is increased in the R1156E system. Because the polar residues on the surface form much more new hydrogen bonds in the F563A, R1156E and TT1145/6AA systems, the polar interactions between protein and water molecules may be decreased. Thus, the polar solvation energies are increased in these three systems. The values of nonpolar solvation energies seem to be similar, except the TT1145/6AA system. The movement of SV2C-LD in TT1145/6AA breaks the interfacial interactions and decreases the interface area between BoNT/A-RBD and SV2C-LD. Therefore, the van der Waals energy and nonpolar solvation energy are increased in the TT1145/6AA system. Although the total binding energy of WT is very low, it is not the lowest one in the five systems. The entropy contribution was not included in the g_mmpbsa,41 so this binding energy will have some deviations compared with the real binding free energy.
TYPE | ΔEvdWa | ΔEelec | ΔGpolar | ΔGnonpolar | ΔGbinding |
---|---|---|---|---|---|
a ΔEvdW, ΔEelec, ΔGpolar, and ΔGnonpolar are binding energy components of van der Waals, electrostatic, polar and nonpolar solvation energies, respectively. ΔGbinding is the total binding energy. The unit of energy is kcal mol−1. | |||||
WT | −60.4 ± 3.4 | −234.1 ± 6.3 | 100.3 ± 8.9 | −7.1 ± 0.3 | −201.4 ± 8.3 |
R1294A | −59.4 ± 5.6 | −232.7 ± 9.8 | 100.7 ± 9.4 | −6.7 ± 0.4 | −197.6 ± 9.2 |
F563A | −61.8 ± 6.3 | −297.0 ± 9.3 | 156.6 ± 10.0 | −7.4 ± 0.5 | −209.5 ± 9.8 |
R1156E | −47.0 ± 6.1 | −214.2 ± 8.1 | 133.6 ± 11.1 | −6.8 ± 0.5 | −134.7 ± 8.5 |
TT1145/6AA | −34.9 ± 3.4 | −292.9 ± 9.7 | 188.3 ± 9.9 | −5.8 ± 0.3 | −145.2 ± 9.2 |
From the above analyses, we can divide the interface between BoNT/A-RBD and SV2C-LD into two binding regions. One is the hydrogen bonds between S1142–T1146 in BoNT/A-RBD and residues E556–F562 in SV2C-LD, which is located at the middle and on the right side of the interface. The other one consists of the cation–π interaction and the R1294–D539 hydrogen bond, which lies on the left side of the interface. The TT1145/6AA mutation mainly breaks the first binding region. The high occupancy hydrogen bonds are lost at the middle and on the right side, so SV2C-LD moves away from BoNT/A-RBD on the right side. This movement also changes the angle and the distance of cation–π interaction. Thus, this mutation reduces the binding affinity remarkably. The R1156E mutation breaks the second binding region. The correct cation–π interaction is lost, but R1294 makes some wrong cation–π interactions with F563 or Y478. These wrong interactions also affect the correct hydrogen bond between R1294 and D539. However, the hydrogen bonds in the first region are kept, so SV2C-LD moves away from BoNT/A-RBD on the left side. The F563A mutation also breaks the cation–π interaction. However, the hydrogen bonds between R1294 and D539 increase the occupancy significantly, which helps in stabilizing the left side of the interface. In the R1294A system, the mutation only breaks the low occupancy hydrogen bond between R1294 and D539. Meanwhile, the energy of cation–π interaction decreases to even lower than that of the WT system, which also helps in stabilizing the left side of the interface. Therefore, the F563A and R1294A systems have the relatively low binding energies and some binding affinities.
Footnote |
† Electronic supplementary information (ESI) available: Fig. S1 and S2; Table S1. See DOI: 10.1039/c4mb00383g |
This journal is © The Royal Society of Chemistry 2015 |