Molecular Environment Modulates Conformational Differences

Article pubs.acs.org/JPCB

Molecular Environment Modulates Conformational Differences between Crystal and Solution States of Human β‑Defensin 2 Jianguo Li,†,‡,∇ Zhongqiao Hu,‡ Roger Beuerman,*,†,⊥,#,∇ and Chandra Verma*,†,‡,§,∥ †

Singapore Eye Research Institute, 11 Third Hospital Avenue, #06-00, Singapore 168751 Bioinformatics Institute (A*-STAR), 30 Biopolis Street, #07-01 Matrix, Singapore 138671 § School of Biological Sciences, Nanyang Technological University, Singapore 637551 ∥ Department of Biological Sciences, National University of Singapore, Singapore 117543 ⊥ Department of Ophthalmology, National University of Singapore, Singapore 119074 # School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459 ‡

S Supporting Information *

ABSTRACT: Human β-defensin 2 is a cysteine-rich antimicrobial peptide. In the crystal state, the N-terminal segment (residues 1−11) exhibits a helical conformation. However, a truncated form, with four amino acids removed from the N-terminus, adopts nonhelical conformations in solution, as shown by NMR. To explore the molecular origins of these different conformations, we performed Hamiltonian replica exchange molecular dynamics simulations of the peptide in solution and in the crystal state. It is found that backbone hydration and specific protein−protein interactions are key parameters that determine the peptide conformation. The helical conformation in the crystal state mainly arises from reduced hydration as well as a salt bridge between the peptide and a symmetryrelated neighboring monomer in the crystal. When the extent of hydration is reduced and the salt bridge is reintroduced artificially, the peptide is successfully folded back to the helical conformation in solution. The findings not only shed light on the development of accurate force field parameters for protein molecules but also provide practical guidance in the design of functional proteins and peptides.

1. INTRODUCTION Proteins are complex biomolecules, and their three-dimensional structures are determined by several factors. The primary factor in determining the conformation is the intrinsic propensity of the amino acid sequence of the protein.1 However, external factors such as the microenvironment in which a protein molecule exists (e.g., solution, crystal, membrane, etc.) and physical parameters (e.g., temperature, ionic strength) also play important roles in determining the tertiary structure of a protein. Thus, a protein can exist in several different conformational states depending on the external conditions. For example, a number of proteins exhibit differences in the overall or local conformations between the crystal state (X-ray structure) and the solution state (NMR structure).2−5 For some proteins, the differences between the X-ray and the NMR conformations can be larger than the difference between the various X-ray-derived conformations or between the various NMR-derived conformations.6 Understanding the origins of such differences is thus of fundamental importance toward understanding the mechanisms that underlie protein folding and protein/peptide design. Human β-defensin 2 (HBD-2) is a naturally occurring antimicrobial peptide (AMP) with 41 amino acids.7,8 It consists of six cysteine residues that are bridged by three intramolecular disulfide bonds, which stabilize its structure. Although the © XXXX American Chemical Society

overall structure of HBD is quite rigid due to the existence of three disulfide bonds, the N-terminal region is quite flexible and adopts different conformations in crystal and in solution.9,10 The X-ray structure resolved at 1.35 Å, which is based on the full-length HBD-2 (henceforth to be referred to as FL-HBD) displays a helical conformation in the N-terminal segment (pdb: 1FD3), while the NMR structure of a truncated HBD-2 (T-HBD, with 4 residues missing from the N-terminus) reveals an unstructured conformation in the N-terminal segment (pdb: 1E4Q) (Figure 1). It is not clear whether this structural difference between X-ray and NMR experiments arises from the differing microenvironments around the peptide or from the different chain lengths of the peptide (FL-HBD for X-ray vs THBD for NMR). The conformation of FL-HBD in solution and the conformation of T-HBD in a crystalline environment have not been determined experimentally. Although numerous secondary structure prediction tools or Web servers are available, most of them are knowledge-based models, which, in general, cannot distinguish the peptide structure in different environments.11,12 We therefore model these two conformations using atomistic models and perform systematic molecular Received: January 4, 2017 Revised: March 15, 2017 Published: March 15, 2017 A

DOI: 10.1021/acs.jpcb.7b00083 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

under the conditions of reduced hydration and in the presence of a salt bridge.

2. METHODS To exhaustively explore the conformational space of the Nterminal segment of HBD-2, we employed HREMD simulations to study the conformational dynamics of the N-terminal segment of HBD-2 (residues 1−11 for FL-HBD and 5−11 for T-HBD) in both solution and crystal environments. Although there have been some conventional MD simulations carried out for protein molecules in crystals,19−26 to our knowledge, there have been no reports of HREMD simulations carried out on proteins in their crystalline environments due to the large size of the unit cell occupied by proteins. In contrast to the TRMED method in which the energy barrier is overcome by increasing the kinetic energy, the HREMD method works by decreasing the energy barriers between local minima on the free energy surface by modifying the Hamiltonian of the system. Hence, in each HREMD simulation, a number of replicas were run in parallel, and the interactions involving the solute molecule (e.g., solute−solute and solute−solvent interactions) are rescaled from 0% for the first replica (no rescaling) to 50% for the last replica while keeping the solvent−solvent interactions as normal in all replicas. At certain time intervals, a Monte Carlo move is carried out to exchange the conformations of adjacent replicas, and the acceptance is governed by the Metropolis rule.13 As the HREMD method does not require rescaling of temperature, it is particularly useful for simulating those temperature-sensitive systems such as crystals or membranes. Studies have shown that both HREMD and TREMD are equally efficient at sampling the conformational spaces of proteins.27 Hence, in conclusion, the simulations of the HBD-2 crystal are carried out using HREMD as it has several advantages over TREMD, including the following: (i) HREMD uses a fewer number of replicas and thus is computationally less expensive than TREMD; (ii) the HBD-2 crystal will develop instabilities in the high-temperature replicas of TREMD due to increased solubility and thermal fluctuations, while HREMD is carried out at room temperature and is thus suitable for simulation of protein crystals; (iii) HREMD enables tempering along only relevant degrees of freedom, thus leaving the irrelevant degrees of freedom unchanged, resulting in a significant acceleration in sampling; in contrast in TREMD, all degrees of freedom are affected. This enables us to specifically manipulate only the interactions involving the N-terminal segment of HBD-2 and treat the rest of the system as normal. Specifically, in the Hamiltonian of each replica, only the interactions involving the N-terminal segment (e.g., internal interactions of the N-segments, interactions between the N-terminal segments, and interactions between the N-terminal segments with other parts of the protein) are rescaled, and the rest of the interactions (e.g., solvent−solvent interactions) are kept unchanged. We carried out three sets of HREMD simulations, and details are summarized in Table S1. In the first set, four HREMD simulations were performed: FL-HBD in water, FL-HBD in crystal, T-HBD in water, and T-HBD in crystal. There are four monomers (e.g., two dimers) in the asymmetric unit of the HBD-2 crystal used here (Figure S1). The first dimer (monomers 1 and 2) experience different symmetry neighbors compared to the second dimer (monomers 3 and 4). In the first dimer, the ASP4 of each monomer forms a salt bridge with the LYS10 of a symmetry-related monomer in the unit cell, while in

Figure 1. Alignment of the crystal structure of FL-HBD (pdb: 1FD3, residue 1−41, blue) and the NMR structure of T-HBD (pdb: 1E4Q, residue 5−41, red).

dynamics (MD) simulations to obtain a detailed understanding of the molecular origins of the conformational difference of HBD-2 in both crystal and solution environments. However, free energy surfaces of proteins or peptides are quite complex, and conventional MD based folding simulations may only reveal pictures of systems trapped in local minima, resulting in inefficient sampling of the phase space. To efficiently sample the phase space, enhanced sampling methods such as temperature-based replica exchange MD (TREMD) simulations have been developed and are widely used.13,14 However, TREMD relies on elevated temperatures (e.g., enhanced kinetic energy) to overcome the potential minima and thus cannot be applied to protein crystals due to several practical reasons: (a) in TREMD, the system is brought out of the free energy minima by enhanced kinetic energy, and the highest temperature used in TREMD usually reaches 500 K, which will result in possible melts of the protein crystal together with phase transitions, and these are not desirable in simulations. Near the phase transition temperature, there is an abrupt change in the potential energy, resulting in nonoverlapping potential energy distributions between adjacent replica, which may also lead to a low acceptance ratio. (b) In TREMD, the number of replicas depends on the total number of degrees of freedom of the system. For a unit cell of defensin with 16 monomers, the TREMD simulation will require a very large number of replicas, thus making it computationally prohibitive. (c) As the number of protein molecules increases, proper sampling in TREMD will require exploration of a very large conformational space of the system, and this will require computationally prohibitive long simulations for convergence. To overcome these limitations, for the first time, Hamiltonian replica exchange MD (HREMD) simulations15−18 of a protein in its crystalline state were performed, which enabled examination of the molecular origins of the conformational difference between solution and crystalline states. We first validate the force field used here by simulating FL-HBD and T-HBD in their original environments, namely, crystal and solution, respectively, and comparing the simulated and available experimental structures. We next predict the conformations of FL-HBD and T-HBD in solution and crystal environments, respectively. We identify two factors that clearly contribute to the helical propensity of the Nterminal segment of HBD-2 in the crystal: reduced hydration and the presence of a salt bridge. To further validate our findings, we perform MD simulations to refold the disordered N-terminal segment of HBD-2 back to helical states in solution B


Article


Figure 2. (a) Snapshot of the unit cell of the FL-HBD crystal taken from the HREMD simulations; protein molecules and water molecules are represented using the cartoon and red dots, respectively. (b) Secondary structure evolution of FL-HBD and T-HBD in the crystal environment over the HREMD simulations. The x-axis is the simulation time, while the y-axis is the residue number. The different colors represent the different secondary structural elements, which are shown in the legend just below the figures. A-helix, 5-Helix, and 3-Helix represents the α-helix, π-helix, and 3−10 helix, respectively. The N-terminal segment is from residue 1 to 11 for FL-HBD and 1 to 7 for T-HBD.

monomers while allowing the N-terminal segments of the two monomers to be flexible. The distance between ASP4 of one monomer and LYS10 of another monomer was constrained to mimic a salt bridge. We first performed two HREMD simulations of a FL-HBD dimer in solution with the salt bridge under normal and reduced hydrations. To further understand the role of the salt bridge, we also performed similar HREMD simulations for monomers 3 and 4 with the salt bridge under normal and reduced hydration. In order to further understand the importance of the salt bridge, we then performed another HREMD simulation of FL-HBD in crystal in which the salt bridge was disabled by setting the partial charges of the side chains of ASP4 and LYS10 to 0. The AMBER99sb force field28 is used to model the protein molecule, and the TIP3P model29 is used for the water molecules. In each HREMD simulation, two end states A and B were chosen. State A corresponds to the normal Hamiltonian, while state B is a virtual Hamiltonian in which interactions involving the peptide were rescaled to 50% with respect to state A, this corresponding to a system with a lowered free energy barrier between metastable states. Then, a series of replicas with intermediate Hamiltonians switching gradually from state A to state B were added to connect the two end states. At every 1 ps, a Monte Carlo move to exchange the configurations of the adjacent replicas was applied. For the HREMD simulations of HBD-2 in the crystal, each unit cell contained 16 HBD monomers. To sample the conformational space efficiently, only the degrees of freedom involving the N-terminal segment of the first monomer were rescaled, while the interactions of the other parts of the system were kept unchanged. In total, 12 replicas were used in each HREMD simulation, which resulted in about 20% acceptance of exchanges. In each HREMD simulation, the initial conformation of the N-terminal segment was kept unstructured. To obtain an unstructured conformation, we performed conventional MD simulations of the HBD-2 at state B, in which the interactions

the second dimer, the ASP4 of each monomer forms a salt bridge with the N-terminal amine of the other monomer in the dimer. In this study, we have chosen monomers 1 and 2 only and focus on the effect of the crystal environment on the conformations of the N-terminal segment of the monomers 1 and 2. In the following, unless specifically mentioned, both FLHBD and T-HBD refer to the full length and truncated monomers 1 and 2, respectively. In the crystal simulation of THBD, as no X-ray structure is available for T-HBD in the crystal, we construct the model of T-HBD in the crystal by removing the first four N-terminal residues of FL-HBD while maintaining the space group as P 21 21 2. The simulations for FL-HBD in the crystal and T-HBD in solution were then compared to the experimental results to evaluate the accuracy of the chosen force field and the efficiency of the sampling method. The other two simulations were performed to predict the conformations of FL-HBD in water and T-HBD in crystal. In order to study the role of reduced hydration as would be the case in a crystal, a second set of HREMD simulations was performed: FL-HBD and T-HBD in solution under reduced hydration. In the simulations of reduced hydration, the repulsion term of the Lennard-Jones (LJ) potential of the nonbonded interactions between oxygen atoms of water molecules and the backbone atoms of the protein molecule was artificially rescaled by increasing the van der Waals radius of oxygen atoms of water molecules by 30% while keeping the water−water and protein−protein interactions unchanged. This results in enhanced repulsions between water molecules and the protein atoms and thus less favorable hydration of the protein atoms. In the third set, three HREMD simulations were carried out to study the role of a salt bridge (between ASP4 and LYS10, as we will see later) on the helical propensity of FLHBD. For this, we took a dimer from the unit cell of the FLHBD crystal and placed it in solution. To mimic and maintain the relative orientations of the monomers in the crystal state, we applied position restraints on the core structures of the two C


Article


Figure 3. Secondary structure evolution of FL-HBD and T-HBD in solution. The N-terminal segment is from residue 1 to 11 for FL-HBD and residue 1 to 7 for T-HBD. Similar to Figure 2; the different colors represent the different secondary structural elements, which are shown in the legend just below the figures.

experimental and MD simulations have shown that α-helical and 3−10 helical conformations are thermodynamically interchangeable and can even coexist in a peptide, with a free energy difference of only 2 kJ/mol/peptide bond.33−36 In addition, MD simulations of short polyalanine peptides have revealed that the 3−10 helix is an intermediate conformation in the transitions between α-helical and nonhelical conformations.37 The helical conformations attained for FL-HBD and THBD in the crystal state suggest that the crystal environment favors helical conformations. Although the α-helix and the 3− 10 helix are thermodynamically similar, a peptide can prefer one over the other depending on the length of the sequence. Considering that T-HBD has four missing residues at the Nterminus, the preference of the 3−10 helix over the α-helix for the N-terminal segment of T-HBD may arise from the reduced chain length. Helix−coil theories,38,39 simulations,40 and electron spin resonance (ESR) spectroscopy41 have suggested that shorter peptides prefer to adopt the 3−10 helix over the αhelix. 3.2. Solution Structures of HBD-2. Next, we performed 150 ns of HREMD simulations for FL-HBD and T-HBD in solution. Figure 3 shows the secondary structure evolution of FL-HBD and T-HBD during the simulations. Because of the existence of the three disulfide bonds, the core structures of both FL-HBD and T-HBD are stable, with little more fluctuation than that in the crystal simulations; this is not unexpected as the environment now is bulk solvent, without the constraints of a crystal. For the N-terminal segment, FL-HBD and T-HBD display quite different conformations. The Nterminal segment of T-HBD (residues 5 to 11) is largely unstructured during the simulations, which is consistent with the NMR observations.10 For the FL-HBD, the N-terminal segment displays a mixture of diverse conformations, with significant population of helical conformations. This suggests that the first four residues at the N-terminus contribute to the helical propensity of this region. When comparing the results of secondary structures in the crystalline state (Figure 2), it is clear

involving the N-terminal segment were rescaled by 50% while the core structure was treated as normal. This resulted in unfolding of the N-terminal segment of HBD-2, and this unfolded conformation was used as the starting conformation in the HREMD simulations. Each replica of the HREMD simulations ran for 150 ns. The secondary structure evolution during the simulations was calculated using DSSP.30 In all simulations, PME was used to calculate the long-range electrostatic interactions,31 while short-range electrostatic interactions and LJ interactions were calculated with a cutoff distance of 1.0 nm.28 Simulations in solution were performed in the NPT ensemble at 300 K and 1 bar, while the NVT ensemble was used for crystal simulations in order to maintain the size and the shape of the unit cell. All simulations were performed using the GROMACS 4.5 package.32

3. RESULTS AND DISCUSSIONS 3.1. Crystal Structures of HBD-2. We first carried out HREMD simulations in the crystal environment for FL-HBD and T-HBD. Figure 2 shows the secondary structure evolution of one HBD-2 monomer during the simulations. In the crystal state, the core structures (residues 12−41) exhibit highly stable conformations with few variations because the conformational space of the molecules is limited by (i) the three disulfide bonds and (ii) the close packing of protein monomers in the crystal environment. For the N-terminal segment of FL-HBD, helicity in the N-terminus begins to appear at ∼30 ns and becomes predominant after ∼50 ns. This is in agreement with the X-ray experimental structures, suggesting that the chosen force field is appropriate and HREMD is a suitable sampling method for simulations of HBD-2 crystals. On the other hand, the N-terminal segment of T-HBD displays a predominant 3− 10 helix conformation, with some population of α-helical conformations in between. Similar to the α-helix, the 3−10 helix that forms hydrogen bonds between residues i and i + 3 appears frequently in protein secondary structures. Previous D


Article


Figure 4. Secondary structure evolution of FL-HBD and T-HBD in solution with reduced hydration. The N-terminal segment is from residues 1 to 11 for FL-HBD and from residues 1 to 7 for T-HBD. Similar to Figure 2; the different colors represent the different secondary structural elements.

backbone atoms and water molecules in the solution state are much larger than that in the crystal (Figure S3). Although a correlation between reduced backbone hydration and helical propensity is apparent, what is not clear is whether reduced backbone hydration favors the helical conformations or is a consequence of the structural constraints imposed by the helical conformations. To explore this, we performed another set of HREMD simulations for both FL-HBD and T-HBD in solution in which the repulsive terms of the LJ interactions of backbone atoms with water molecules are artificially rescaled, corresponding to reduced backbone hydration. If the reduced backbone hydration is the only factor that leads to the helical conformation of the N-terminal segment, the rescaled interactions should result in significantly reduced hydration around the backbone atoms, thus facilitating intramolecular hydrogen bonding between the backbone atoms and hence stabilization of helical folds. The rescaling results in reduced hydration during the HERMD simulations, as is evident from the lower number of water molecules in the hydration shell of the backbone atoms (Figure S4). Figure 4 shows the secondary structure evolution of both FL-HBD and T-HBD under these rescaled conditions. T-HBD begins to form a 3−10 helix from ∼10 ns, and this conformation is dominant during the rest of the trajectory, with some population of α-helices in between. The frequent switch between the 3−10 helix and the α-helix suggests a lower free energy barrier between these two conformations, consistent with previous literature reports.33−37 However, the secondary structure evolution of FL-HBD under rescaled interactions exhibits no enhanced helicity. This indicates that for FL-HBD other factors must modulate the helical propensity. 3.4. Role of the Salt Bridge. A natural consequence of the close packing of molecules in a crystal, together with reduced hydration levels, is the existence of intermolecular contacts, which significantly modulate the protein conformations. We examined the monomers of FL-HBD belonging to different symmetry-related units in the crystal environment and noticed a salt bridge between ASP4 of one monomer and LYS10 of a

that for both T-HBD and FL-HBD, the N-terminal segment displays less helicity when in solution compared to the crystalline state, particularly for T-HBD whose helicity was totally lost in solution, suggesting that the crystal environment favors the helical conformation. 3.3. Hydration and Helical Propensities. From the above HREMD simulations, it is clear that the crystal environment favors the helical conformation for both FLHBD and T-HBD. So, what are the environmental factors that drive such differences in conformations? One of the differences between the microenvironment of a protein in solution and that in a crystal is that the latter is much less hydrated. To understand the effect of hydration, we calculated the radial distribution function (RDF) between the backbone atoms of the N-terminal segment and the water molecules as well as the cumulative number of water molecules around the N-terminal segment (Figure S2). Although the RDF profiles display higher peaks in the case of the crystal than in solution for both FLHBD and T-HBD, they are normalized parameters that cannot reflect the absolute hydration. Instead, the cumulative number of water molecules around the N-terminal segment is much higher in the solution state, implying reduced backbone hydration in the crystalline environment. In addition, the number of water molecules in the hydration shell of the Nterminal segment was found to be 74 in solution and 42 in the crystal, further confirming the reduced hydration in the crystalline environment. As the secondary structure is mainly driven by the intramolecular hydrogen bonds between backbone atoms, the reduced backbone hydration in the crystal leads to less perturbation of backbone hydrogen bonds from water molecules, resulting in stable secondary structures. In contrast, in the solution state, water molecules can competitively form hydrogen bonds with the backbone atoms, thus resulting in disruption of the backbone−backbone hydrogen bonds. As a consequence, the peptide loses its helical conformation and becomes unstructured. Indeed, for both FL-HBD and T-HBD, the number of hydrogen bonds between E


Article


Figure 5. (a) FL-HBD in a crystal and the salt bridges (red line) between ASP4 in one monomer and the LYS10 of an adjacent monomer; (b) FLHBD in solution and the hydrogen bond between ASP4 and the backbone of VAL6; (c) distance distributions of possible electrostatic pairs.

Figure 6. Secondary structure evolutions of FL-HBD dimer (monomers 1 and 2) in solution with an artificial salt bridge between ASP4 of one monomer and LYS10 of another monomer (a) under normal hydration and (b) under reduced hydration. The N-terminal segment is from residues 1 to 11 for FL-HBD and from residues 1 to 7 for T-HBD. Similar to Figure 2; the different colors represent the different secondary structural elements, which are shown in the legend just below the figures.

atoms, and as a consequence, the helix is destabilized. In contrast, in T-HBD, the ASP4 is missing, which allows the VAL6 backbone amide to engage in intramolecular hydrogen bonds with other backbone atoms, resulting in a stable helical state. This suggests that for FL-HBD the salt bridge constraint also plays an important role in modulating its helical propensity. However, in order for FL-HBD to adopt a robust helix at its N-terminus, it is still not clear if the salt bridge alone is sufficient or if reduced hydration is additionally needed. In order to address this issue, we performed another set of HREMD simulations for the FL-HBD dimer in solution with the salt bridge constraint reintroduced during the simulation. In

symmetry-related neighboring monomer. We also found them to be stable throughout the HREMD simulations. However, when FL-HBD is in solution, the simulations are only carried out on a monomer, and hence this constraint does not exist, suggesting that the salt bridge is critical in stabilizing the helical conformation of the N-terminal segment. Although an intramolecular salt bridge between ASP4 and LYS10 of the same protein is possible, it is not formed, as revealed by the distances between the COO group of ASP4 and NH3 group of LYS10 (Figure 5). Instead, ASP4 forms a hydrogen bond with the backbone amide group NH of VAL6 during most of the simulation time. The backbone amide group of VAL6 is no longer free to engage in hydrogen bonds with other backbone F


Article


is partly supported by the fact that even for a peptide displaying helical conformations the helical part is mainly located at the central segment and the N-terminal and the C-terminal regions are usually unstructured because the central segment is less hydrated than the two termini of a peptide.42 Furthermore, reduced backbone hydration is the driving force for the formation of helices in the presence of cosolvents such as trifluoroethanol (TFE). It is likely that TFE forms a layer around the peptide with the hydrophobic CF3 group of TFE facing the peptide atoms while the COO− group faces bulk water, protecting the backbone atoms against water molecules and thus stabilizing the backbone−backbone hydrogen bonds.43 Besides protein−water interactions, protein structures are also stabilized by specific protein−protein interactions. In crystals, protein molecules are densely packed, resulting in a large number of atomic contacts between monomers and their symmetry-related neighbors and, as a consequence, reduced protein flexibility. In contrast, in solution, protein molecules are largely monomeric and are solvated by water molecules. The confinement and specific protein−protein interactions in the crystalline state can also modulate the protein conformation. In the case of FL-HBD, a salt bridge constraint helps stabilize a helical conformation in the monomer. The dependence of the protein conformation on its microenvironment has broad biological implications. The crystal and solution environments are quite different from the intracellular environment. Inside of the cell, protein molecules interact with hundreds of different macromolecules in a confined space, a phenomenon referred to as crowding.44 Various studies have reported that the conformation of a protein in a crowded environment is quite different from that in solution. For example, it has been shown that the crowding effect not only changes the conformation of a protein45 but also affects protein folding kinetics via modulating the stability of intermediate conformations.46,47 Our results of how the microenvironment (e.g., hydration, specific protein−protein interactions, etc.) modulates protein conformations emphasize the significance of the environmental factors such as protein hydration and specific protein−protein interactions in modulating protein conformations and suggest that these environmental factors should be taken into consideration when designing functional proteins. For instance, a large number of AMPs such as melittin display an unstructured conformation in solution but form helical conformations upon adsorption onto membranes or in a crystal environment.48,49 Hence, understanding the dependence of the molecular environment on the conformations of AMPs is critical for AMP design.50,51 Moreover, physical insights into the role of environmental factors in modulating the protein conformation also shed light on the development of force field parameters for the accurate description of protein interaction potentials.

order to study the combined effects of hydration and the salt bridge, two HREMD simulations were carried out: one with both the salt bridge and rescaled interactions and the other with only the salt bridge. It is clear (Figure 6) that the helical conformation rapidly appears in both simulations (in less than 5 ns). Both HREMD simulations resulted in stable helical states compared to the helicity seen for FL-HBD in solution. In particular, the simulation with both the salt bridge and reduced hydration displayed a little higher content of α-helical propensity. This suggests that the α-helix propensity of the N-terminal segment of FL-HBD primarily arises as a result of the salt bridge between ASP4 in one monomer and the LYS10 in the adjacent monomer and is complemented by reduced hydration, albeit to a lesser degree. Similarly, the HREMD simulation for the dimer with monomers 3 and 4 shows helical conformations as well, further suggesting the importance of the salt bridge constraint (Figure S5). To further understand the role of the salt bridge in modulating the helical propensity of FL-HBD in the crystal environment, we performed another HREMD simulation of the FL-HBD dimer in crystal as the negative control in which we artificially disrupt the salt bridge by rescaling the electrostatic interactions between Lys10 and ASP4 of the neighboring monomer to 0. The secondary structure evolution of FL-HBD (Figure S6) revealed a much reduced propensity for helicity at the N-terminus than that in the presence of the salt bridge. This further confirms the importance of the salt bridge in stabilizing the helical conformation of FL-HBD in the crystal environment.

4. DISCUSSIONS AND IMPLICATIONS X-ray or NMR experiments are the most widely used techniques for protein three-dimensional structure determination. Generally X-ray experiments yield images of static conformations within highly ordered crystals, while NMR generates an ensemble of dynamic structures in solution. Due to the differences in experimental conditions and in the microenvironment of a protein, the three-dimensional structures of a protein available from the two techniques can be significantly different. Using HREMD simulations and a model protein HBD-2, we have examined the molecular origins of the differing conformations adopted by the N-terminal segment of FLHBD and T-HBD in both crystals and solution. The differences in the helical propensities of FL-HBD and T-HBD can be interpreted by examining the underlying intermolecular interactions. There are three basic types of these interactions: protein−protein interactions, protein−water interactions, and water−water interactions. In crystals, the protein molecules are closely packed, resulting in a higher number of protein−protein contacts and fewer protein−water contacts than in solution. Moreover, due to the confinement of the intermolecular spaces or pores/channels in a crystal, water molecules can only form small and very dynamic clusters, resulting in lower water density and slower diffusion in crystals than in solution.21 In addition, the reduced density and lower mobility of waters in the crystals results in reduced backbone hydration compared to that in the solution state, which prevents the backbone atoms from being attacked by water molecules. As a consequence, the protein backbone atoms tend to form more backbone− backbone hydrogen bonds in the crystal than in solution. This results in the crystal environment favoring helical conformations for both FL-HBD and T-HBD. This observation

5. CONCLUSIONS We have performed systematic HREMD simulations to explore the external determinants that lead to the conformational differences of HBD-2 in a crystal and in solution. Reduced hydration and specific intermolecular interactions such as a salt bridge between symmetry-related neighboring monomers in the crystal environment were found to significantly modulate peptide conformations such as the helical propensity. Clearly, the difference between the X-ray and NMR structures of the Nterminal region of HBD2 suggests that the intrinsic propensity G


Article


(7) Schröder, J.-M.; Harder, J. Human beta-defensin-2. Int. J. Biochem. Cell Biol. 1999, 31, 645−651. (8) Harder, J.; Bartels, J.; Christophers, E.; Schroder, J. M. A Peptide Antibiotic from Human Skin. Nature 1997, 387, 861−861. (9) Hoover, D. M.; Rajashankar, K. R.; Blumenthal, R.; Puri, A.; Oppenheim, J. J.; Chertov, O.; Lubkowski, J. The Structure of Human β-Defensin-2 Shows Evidence of Higher Order Oligomerization. J. Biol. Chem. 2000, 275, 32911−32918. (10) Bauer, F.; Schweimer, K.; Klüver, E.; Conejo-Garcia, J.-R.; Forssmann, W.-G.; Rösch, P.; Adermann, K.; Sticht, H. Structure Determination of Human and Murine β-defensins Reveals Structural Conservation in the Absence of Significant Sequence Similarity. Protein Sci. 2001, 10, 2470−2479. (11) Petrey, D.; Honig, B. Protein Structure Prediction: Inroads to Biology. Mol. Cell 2005, 20, 811−819. (12) Zhang, Y. Progress and Challenges in Protein Structure Prediction. Curr. Opin. Struct. Biol. 2008, 18, 342−348. (13) Sugita, Y.; Okamoto, Y. Replica-exchange Molecular Dynamics Method for Protein Folding. Chem. Phys. Lett. 1999, 314, 141−151. (14) Miyashita, N.; Straub, J. E.; Thirumalai, D.; Sugita, Y. Transmembrane Structures of Amyloid Precursor Protein Dimer Predicted by Replica-Exchange Molecular Dynamics Simulations. J. Am. Chem. Soc. 2009, 131, 3438−3439. (15) Bussi, G. Hamiltonian Replica Exchange in GROMACS: a Flexible Implementation. Mol. Phys. 2014, 112, 379−384. (16) Liu, P.; Kim, B.; Friesner, R. A.; Berne, B. J. Replica Exchange with Solute Tempering: A Method for Sampling Biological Systems in Explicit Water. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 13749−13754. (17) Terakawa, T.; Kameda, T.; Takada, S. On Easy Implementation of a Variant of the Replica Exchange with Solute Tempering in GROMACS. J. Comput. Chem. 2011, 32, 1228−1234. (18) Mu, Y. Dissociation Aided and Side Chain Sampling Enhanced Hamiltonian Replica Exchange. J. Chem. Phys. 2009, 130, 164107. (19) Hu, Z.; Jiang, J. Assessment of Biomolecular Force Fields for Molecular Dynamics Simulations in a Protein Crystal. J. Comput. Chem. 2009, 31, 371−380. (20) Hu, Z.; Jiang, J.; Sandler, S. I. Water in Hydrated Orthorhombic Lysozyme Crystal: Insight from Atomistic Simulations. J. Chem. Phys. 2008, 129, 075105. (21) Hu, Z.; Jiang, J. Molecular Dynamics Simulations for Water and Ions in Protein Crystals. Langmuir 2008, 24, 4215−4223. (22) Cerutti, D. S.; Freddolino, P. L.; Duke, R. E.; Case, D. A. Simulations of a Protein Crystal with a High Resolution X-ray Structure: Evaluation of Force Fields and Water Models. J. Phys. Chem. B 2010, 114, 12811−12824. (23) York, D. M.; Wlodawer, A.; Pedersen, L. G.; Darden, T. A. Atomic-level Accuracy in Simulations of Large Protein Crystals. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 8715−8718. (24) Heiner, A. P.; Berendsen, H. J. C.; van Gunsteren, W. F. MD Simulation of Subtilisin BPN′ in a Crystal Environment. Proteins: Struct., Funct., Genet. 1992, 14, 451−464. (25) Stocker, U.; Spiegel, K.; van Gunsteren, W. F. On the Similarity of Properties in Solution or in the Crystalline State: A Molecular Dynamics Study of Hen Lysozyme. J. Biomol. NMR 2000, 18, 1−12. (26) Ahlstrom, L. S.; Vorontsov, I. I.; Shi, J.; Miyashita, O. Effect of the Crystal Environment on Side-Chain Conformational Dynamics in Cyanovirin-N Investigated through Crystal and Solution Molecular Dynamics Simulations. PLoS One 2017, 12, e0170337. (27) Smith, A. K.; Lockhart, C.; Klimov, D. K. Does Replica Exchange with Solute Tempering Efficiently Sample Aβ Peptide Conformational Ensembles? J. Chem. Theory Comput. 2016, 12, 5201− 5214. (28) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Comparison of Multiple Amber Force Fields and Development of Improved Protein Backbone Parameters. Proteins: Struct., Funct., Genet. 2006, 65, 712−725. (29) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935.

of an amino acid sequence to adopt a particular secondary structure is also modulated by the environment.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.7b00083. Details of HREMD simulations at different conditions; figures of RDF and the cumulative number of water atoms and the number of hydrogen bonds under conditions of normal and reduced hydration; secondary structure evolution in a crystal with the salt bridge removed (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (C.V.). *E-mail: [email protected] (R.B.). ORCID

Jianguo Li: 0000-0002-5544-6451 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest. ∇ J.L. and R.B. are also part of the Duke−NUS Medical School, Ophthalmology Academic Clinical Program, Singapore.

■

ACKNOWLEDGMENTS This work is supported by a grant from NMRC/TCR/002SERI/2008/R618, NMRC/BNIG/2016/2014, and NMRC/ TCR/R1018, Singapore. The authors also thank BII, NSCC, and CSC in Finland for providing the computational facilities.

■

ABBREVIATIONS HBD-2, human β-defensin 2; FL-HBD, full length human βdefensin 2; T-HBD, truncated human β-defensin 2; HREMD, Hamiltonian replica exchange molecular dynamics; TREMD, temperature-based replica exchange molecular dynamics; RDF, radial distribution function

■

REFERENCES

(1) Anfinsen, C. B. Principles that Govern the Folding of Protein Chains. Science 1973, 181, 223−230. (2) Akke, M.; Chazin, W. J. An Open and Shut Case. Nat. Struct. Biol. 2001, 8, 910−912. (3) Massiah, M. A.; Worthylake, D.; Christensen, A. M.; Sundquist, W. I.; Hill, C. P.; Summers, M. F. Comparison of the NMR and X-ray Structures of the HIV-1 Matrix Protein: Evidence for Conformational Changes During Viral Assembly. Protein Sci. 1996, 5, 2391−2398. (4) Wallace, B. A.; Janes, R. W.; Bassolino, D. A.; Krystek, S. R. A Comparison of X-ray and NMR Structures for Human Endothelin-1. Protein Sci. 1995, 4, 75−83. (5) Kovacs, F.; Quine, J.; Cross, T. A. Validation of the Singlestranded Channel Conformation of Gramicidin A by Solid-state NMR. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 7910−7915. (6) Garbuzynskiy, S. O.; Melnik, B. S.; Lobanov, M. Y.; Finkelstein, A. V.; Galzitskaya, O. V. Comparison of X-ray and NMR Structures: Is There a Systematic Difference in Residue Contacts between X-rayand NMR-resolved Protein Structures? Proteins: Struct., Funct., Genet. 2005, 60, 139−147. H


Article

The Journal of Physical Chemistry B (30) Kabsch, W.; Sander, C. Dictionary of Protein Secondary Structure: Pattern Recognition of Hydrogen-bonded and Geometrical Features. Biopolymers 1983, 22, 2577−2637. (31) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. A Smooth Particle Mesh Ewald Method. J. Chem. Phys. 1995, 103, 8577−8593. (32) Bjelkmar, P.; Larsson, P.; Cuendet, M. A.; Hess, B.; Lindahl, E. Implementation of the CHARMM Force Field in GROMACS: Analysis of Protein Stability Effects from Correction Maps, Virtual Interaction Sites, and Water Models. J. Chem. Theory Comput. 2010, 6, 459−466. (33) Mikhonin, A. V.; Asher, S. A. Direct UV Raman Monitoring of 310-Helix and π-Bulge Premelting during α-Helix Unfolding. J. Am. Chem. Soc. 2006, 128, 13789−13795. (34) Chipot, C.; Pohorille, A. Folding and Translocation of the Undecamer of Poly-l-leucine across the Water−Hexane Interface. A Molecular Dynamics Study. J. Am. Chem. Soc. 1998, 120, 11912− 11924. (35) Dike, A.; Cowsik, S. M. Solution Structure of Amphibian Tachykinin Uperolein Bound to DPC Micelles. J. Struct. Biol. 2006, 156, 442−452. (36) Hu, Z.; Jiang, J. A Helical Peptide Confined in Metal-organic Frameworks: Microscopic Insight from Molecular Simulation. Microporous Mesoporous Mater. 2016, 232, 138−142. (37) Daggett, V.; Levitt, M. Molecular Dynamics Simulations of Helix Denaturation. J. Mol. Biol. 1992, 223, 1121−1138. (38) Sheinerman, F. B.; Brooks, C. L. 310 Helixes in Peptides and Proteins As Studied by Modified Zimm-Bragg Theory. J. Am. Chem. Soc. 1995, 117, 10098−10103. (39) Rohl, C. A.; Doig, A. J. Models for the 310-helix/coil, π-helix/ coil, and α-helix/310-helix/coil Transitions in Isolated Peptides. Protein Sci. 1996, 5, 1687−1696. (40) Armen, R.; Alonso, D. O. V.; Daggett, V. The Role of α-, 310-, and π-helix in Helix→Coil Transitions. Protein Sci. 2003, 12, 1145− 1157. (41) Fiori, W. R.; Miick, S. M.; Millhauser, G. L. Increasing Sequence Length Favors Alpha-helix over 310-helix in Alanine-based Peptides: Evidence for a Length-dependent Structural Transition. Biochemistry 1993, 32, 11957−11962. (42) Kennedy, R. J.; Walker, S. M.; Kemp, D. S. Energetic Characterization of Short Helical Polyalanine Peptides in Water: Analysis of 13CO Chemical Shift Data. J. Am. Chem. Soc. 2005, 127, 16961−16968. (43) Roccatano, D.; Colombo, G.; Fioroni, M.; Mark, A. E. Mechanism by Which 2,2,2-Trifluoroethanol/Water Mixtures Stabilize Secondary-structure Formation in Peptides: A Molecular Dynamics Study. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 12179−12184. (44) Ellis, R. J. Macromolecular Crowding: Obvious but Underappreciated. Trends Biochem. Sci. 2001, 26, 597−604. (45) Qin, S.; Zhou, H.-X. Effects of Macromolecular Crowding on the Conformational Ensembles of Disordered Proteins. J. Phys. Chem. Lett. 2013, 4, 3429−3434. (46) Dong, H.; Qin, S.; Zhou, H.-X. Effects of Macromolecular Crowding on Protein Conformational Changes. PLoS Comput. Biol. 2010, 6, e1000833. (47) Kurniawan, N. A.; Enemark, S.; Rajagopalan, R. Crowding Alters the Folding Kinetics of a β-Hairpin by Modulating the Stability of Intermediates. J. Am. Chem. Soc. 2012, 134, 10200−10208. (48) Hartings, M. R.; Gray, H. B.; Winkler, J. R. Probing Melittin Helix−Coil Equilibria in Solutions and Vesicles. J. Phys. Chem. B 2008, 112, 3202−3207. (49) Blondelle, S. E.; Lohner, K.; Aguilar, M.-I. Lipid-induced Conformation and Lipid-binding Properties of Cytolytic and Antimicrobial Peptides: Determination and Biological Specificity. Biochim. Biophys. Acta, Biomembr. 1999, 1462, 89−108. (50) Li, J.; Koh, J.-J.; Liu, S.; Lakshminarayanan, R.; Verma, C. S.; Beuerman, R. W. Membrane Active Antimicrobial Peptides: Translating Mechanistic Insights to Design. Front. Neurosci. 2017, 11, 73.

(51) Brogden, K. A. Antimicrobial Peptides: Pore Formers or Metabolic Inhibitors in Bacteria? Nat. Rev. Microbiol. 2005, 3, 238− 250.

I


Molecular Environment Modulates Conformational Differences

Recommend Documents