Influence of Precipitants on Molecular Arrangements and Space

Dec 27, 2016 - Atom charges of the ammonium ion, sulfate ion, tartrate ion, and PEG 1000 molecule used in the MD simulation were determined by perform...
2 downloads 12 Views 2MB Size
Download PDF
Article pubs.acs.org/crystal

Influence of Precipitants on Molecular Arrangements and Space Groups of Protein Crystals Satoshi Fudo,† Fei Qi,† Michiyoshi Nukaga,‡ and Tyuji Hoshino*,† †

Graduate School of Pharmaceutical Sciences, Chiba University, Inohana 1-8-1, Chuo-ku, Chiba 260-8675, Japan Faculty of Pharmaceutical Sciences, Josai International University, Gumyo 1, Togane-shi Chiba 283-8555, Japan

‡

S Supporting Information *

ABSTRACT: In protein crystallization, precipitants are used to control the final protein concentration in the solution and/ or to decrease the protein solubility for nucleation and growth. In this study, we obtained three crystal structures for the same kind of protein with three different crystallizing agents, in which one of the three different chemicals, ammonium sulfate, potassium sodium tartrate, and polyethylene glycol (PEG), was contained as a main precipitant. The space group of the protein crystal obtained by PEG was different from those obtained by the other two precipitants. Molecular dynamics simulations were carried out for the protein in the presence of each of the three precipitants at a concentration equivalent to the crystallizing condition or without any precipitant. The simulations showed that all of the three precipitants enhanced protein stability by decreasing the conformational fluctuation. The distribution of precipitant molecules was found to be not isotropic around the protein in every case. In the simulations with ammonium sulfate and potassium sodium tartrate, high-concentration areas of precipitants on the protein surface coincided with noncontact sites with other protein molecules in the crystals. In the simulations with PEG, low-concentration areas coincided with noncontact sites with other protein molecules in the crystal. The results suggest that precipitants play multiple roles not only of decreasing the protein solubility but also in restricting contact sites on the protein surface. This restriction is reflected in the molecular arrangement in protein crystals, thereby resulting in crystal growth with a specific space group.

■

molecules by dealing with crystal structures statistically20,21 or by using computer simulation.22−24 However, despite these broad studies, there are still many factors to be clarified for a better understanding of protein crystallization. From a practical point of view, screening and optimization of crystallizing agents that are suitable for obtaining protein crystals with good quality require many protein samples and much time and effort. Therefore, an efficient approach to find the optimal composition of a crystallizing agent is needed. Precipitants, including salts, polymers, and organic solvents, have the greatest influence on protein solubility. The role of salt in crystallization has been discussed from many aspects including Derjaguin, Landau, Verwey, and Overbeek (DLVO) theory,25 Hofmeister series,26−28 electroselectivity,28 and common ion effect.29 Salts are thought to be excluded from the protein surface by being strongly hydrated, leading to salting-out of the protein.30 However, direct interaction with the protein and significant alteration of protein stability have been reported in many cases.31,32 Polyethylene glycol (PEG) is one of the most successful polymers, and its action as a precipitant can be explained by depletion forces.33,34 Although

INTRODUCTION Information on protein structure is fundamental for understanding the function of a protein and often enables rational structure-based design in drug discovery. X-ray crystallography has been widely used for determining the structures of protein molecules. Although protein crystallization is indispensable in X-ray structure analysis, crystallization is one of the hurdles in such studies.1 Therefore, an atomistic-scale understanding of protein crystallization is necessary to further advance crystallization techniques and to accelerate research on structural biology. A phase diagram of a protein solution is essential to explain the behavior of a solute in solution and to control the protein crystallization.2−9 The second virial coefficient, which is closely related to the strength of protein interactions, correlates with the solubility of a protein in an aqueous solution.10 In protein solutions that can promote protein crystal growth, the second virial coefficient is in a narrow range called the crystallization slot.11 Nucleation of the protein crystal occurs in the outside state of the soluble phase and is affected by the supersaturation rate and liquid−liquid phase separation.3,7,8,12,13 While classical nucleation theory can account for the experimental protein nucleation in many cases,8,14,15 recent studies have suggested that so-called nonclassical nucleation events also occur.16−19 Several studies have focused on the contact of protein © XXXX American Chemical Society

Received: September 19, 2016 Revised: December 5, 2016 Published: December 27, 2016 A

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

coordinates for crystal structures 2 and 3 were determined by molecular replacement with the MOLREP program46 using crystal structure 1 (4ZQQ in PDB code)44 as a search model. Structure refinement and model building were carried out using PHENIX47 and COOT.48 Molecular Dynamics Simulation. Atom charges of the ammonium ion, sulfate ion, tartrate ion, and PEG 1000 molecule used in the MD simulation were determined by performing quantum chemical calculation using the Gaussian09 program49 in the following steps. First, geometric optimization was performed at the B3LYP/631G** level. PEG 8000 was replaced by PEG 1000 because PEG 8000 is too long to deal with in a boundary box model. As a starting geometry of PEG 1000, a linear conformation was selected. Next, the electrostatic potentials of those compounds were calculated at the B3LYP/cc-pVTZ level under the solvation condition of ether (ε = 4) by the IEFPCM method.50 Then atom charges were assigned to the respective atoms using the RESP method.51 Crystal structure 1 was used as a template for building the models for MD simulation. The template was placed in a rectangular periodic boundary box filled with TIP3P water,52 and counterions were added to neutralize the model system using the LEaP module in AMBER 14.53 The model at this step was used as an initial structure for MD simulation without any precipitant (simulation 0). To build models that include ammonium sulfate as a precipitant, three water molecules were chosen randomly in the model for simulation 0, and then two of them were replaced with ammonium ions and the other one was replaced with a sulfate ion. This operation was repeated until the concentration of ammonium sulfate was the same as that in the crystallizing droplet in experiments. The procedure for replacing water molecules with ammonium sulfate was done twice to make two independent initial models for MD simulations with ammonium sulfate (simulations 1-a and 1-b). Similarly, the water molecules chosen randomly in the model for simulation 0 were replaced with either potassium, sodium, or tartrate ions, and the procedure was duplicated to make two independent initial models for MD simulations with potassium sodium tartrate (simulations 2-a and 2b). Since the length of PEG 1000 in the linear conformation is almost the same as that of the longest edge of the rectangular box, random generation, as done for the other precipitants, was not applicable. Hence, PEG 1000 molecules were generated to surround the protein in the model for simulation 0, and the water molecules that overlapped with a PEG molecule were deleted. The coordinates of PEG molecules were changed between the two initial models for MD simulations by rotating every other PEG by 180° around the axis perpendicular to the primary molecular axis (simulations 3-a and 3-b). It should be emphasized again that the number of precipitant molecules in each model was set to reflect the initial concentration of the precipitant in the hanging drop in experiments. Therefore, the concentration of the precipitant in each model is half of that in the well solution. The general AMBER force field (GAFF)54 was applied to the ammonium ion, sulfate ion, tartrate ion, and PEG 1000 molecule, and the ff10 force field55 was applied to the other molecules. Energy minimization, heating, pre-equilibration, and production runs were performed with the pmemd module in AMBER 14. First, energy minimization was executed only for water molecules and counterions. Energy minimization was executed again without any restraint for all of the atoms. Energy minimizations were performed by the steepest descent method for the first 3000 cycles and by the conjugated gradient method for the subsequent 7000 cycles. Second, the system was heated to 291 K for 0.1 ns under the NVT-ensemble condition. Third, pre-equilibrated calculation was performed for 0.4 ns under the NPT-ensemble condition. Then the production runs of MD simulation were carried out for 100 ns on models 0, 1-a, 1-b, 2-a, and 2-b and for 200 ns on models 3-a and 3-b with an integration time step of 2.0 fs under the NPT-ensemble condition (1 atm, 291 K). The cutoff distance for the noncovalent interaction terms was set to 12.0 Å. Expansion and shrinkage of all covalent bonds involving hydrogen were constrained by the SHAKE algorithm.56 A periodic boundary condition was applied to avoid an edge effect.

PEG is thought to be preferentially excluded from the protein surface,35 direct interaction of PEG with proteins has also been reported.36 Similarly, while many organic precipitants have been suggested to be generally excluded from the protein surface,37,38 it has been reported that they are possibly bound to hydrophobic regions of the protein surface.39,40 In our previous works, we carried out cluster analyses of crystal structures for the same kinds of proteins, exemplified by human immune-deficiency virus type 1 (HIV-1) protease, hemoglobin, myoglobin, human serum albumin, and hen eggwhite lysozyme.41,42 For each kind of protein, crystal structures were separated into several groups based on their structural similarities. From the analyses, it was found that crystal structures belonging to the same group were crystallized by using the same kind of precipitant. It was also shown that crystal structures belonging to the same group almost always had a common space group. These results suggest that precipitants strongly affect how each protein molecule makes contact with other protein molecules in the crystal and thereby precipitants determine the space group. The aim of the present study was to determine the behavior of precipitants in terms of the interaction with a protein molecule in solution by performing molecular dynamics (MD) simulation and to clarify the influence of precipitants on the manner of molecular arrangement of proteins in a crystal. The protein used for this study was influenza virus polymerase acidic subunit N-terminal domain one-loop deletion recombinant (PANΔloop), which was used in our previous works.43,44 First, the protein was crystallized using three different precipitants: ammonium sulfate, potassium sodium tartrate, and PEG 8000. Next, MD simulations were carried out for the protein solution with each of the precipitants. Then the relationship between molecular packing in the crystal and molecular geometry in the simulation was analyzed.

■

METHODS

Protein Expression and Purification. Truncated PAN protein, PANΔloop, from influenza virus A/Puerto Rico/8/34 (PR8) strain (H1N1) was expressed and purified as previously described.43,44 Briefly, an Escherichia coli strain transformed with the pET50b(+) vector containing the PANΔloop gene was cultured in LB medium. The protein was expressed at 17 °C for 48 h after induction with 0.2 mM IPTG. The protein was purified by a His-affinity column, followed by cleavage of a 6× His-fused Nus-tag by HRV 3C protease. The cleaved protein was again purified by Ni-NTA resin. The protein was further purified by gel filtration with a running buffer of 20 mM Tris-HCl at pH 8.0 and 100 mM NaCl. Finally, the protein was concentrated to 9.9 mg/mL. Protein Crystallization and Structure Determination. Crystal structure 1, which was obtained by using crystallizing agent A (100 mM MES at pH 5.8, 1.1 M ammonium sulfate, 0.1 M potassium chloride and 9% (v/v) trehalose), was reported in our previous work.44 Two other crystals of PANΔloop were grown by the vapor diffusion method with hanging drops consisting of 1.0 μL of 9.9 mg/mL protein solution containing 4.0 mM MnCl2 and 1.0 μL of crystallizing agent B or C at 18 °C. Crystallizing agent B includes 100 mM MES at pH 6.0 and 0.8 M potassium sodium tartrate, and agent C includes 100 mM Tris-HCl at pH 7.5, 28% PEG 8000, 0.2 M calcium acetate, and 10% Jeffamine M-600. The crystals were cryoprotected by brief immersion in a well solution containing 22.5% (v/v) glycerol, followed by flashfreezing in liquid nitrogen. X-ray diffraction data for crystal structures 2 and 3 were obtained at 100 K on the BL5A beamline of Photon Factory (PF, Tsukuba, Japan). The diffraction data were indexed, scaled, and merged with HKL2000.45 Intensities were converted into structure factors, and 5% of the reflections were flagged for Rfree calculations. Atom B

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Analysis of Molecular Dynamics Simulation. The ptraj module of AMBER tools was utilized to obtain root-mean-square deviation (RMSD) and B-factor with handling MD trajectories, in which only main chain atoms (N, Cα, and C) were taken into account. The last 50 ns trajectories were used for B-factor calculation. The coordinates of 200 snapshots for the last 20 ns were used to make distribution maps of precipitant molecules. All of the snapshots were aligned, with the active site of PANΔloop positioned at the center of the box. Grids of 1.0 Å spacing were generated in the box, and distribution maps were constructed by counting the number of precipitant molecules for the respective grids in which each precipitant molecule was assigned to one of the grids at the shortest distance from its center of mass. The counts were summed up for all of the snapshots and visualized using PyMOL.57 Calculation of Electrostatic Potential. Electrostatic potential of crystal structure 1 (PDB code: 4ZQQ)44 was calculated by the Delphi program58 with a protein dielectric constant of 4.0 and a solvent dielectric of 80. Electrostatic potential maps were visualized using the Chimera program.59

sodium tartrate (simulations 2-a and 2-b), and PEG 1000 (simulations 3-a and 3-b). PEG 1000 is a substitute for PEG 8000 in the simulation. The models built for the simulations are shown in Figure 2. The number of molecules, total number of atoms, and box dimensions in each simulation are summarized in Table S2. These models (except for that for simulation 0) were constructed so as to reproduce the concentrations of precipitants in the hanging drops at the moment when they were set up by mixing the protein solution and one of the crystallizing agents. In order to focus on the precipitants, other minor chemicals included in the crystallizing agents were ignored. In simulations 3-a and 3-b, simulation time was set to 200 ns, while simulation time in other simulations was set to 100 ns because the movement of PEG 1000 was relatively slow and equilibration that was sufficient for reliable analysis needed a longer simulation time. In all of the models, RMSD values for main chain atoms of the protein from the initial structure remained below 3.0 Å during the simulation (Figure S1), and B-factor values for most of the residues were below 50 Å2 (Figure 3). It should be noted that in most residues, B-factors in all of the simulations with precipitants were lower than those of simulation 0. This effect caused by precipitants was the most significant at the residues around 20−30, where the B-factors were much higher in the simulation 0. This result means that the precipitants have a suppressive effect on the conformational fluctuation of this protein. Less fluctuation of the protein molecule leads not only to enhancement of molecular stability but also to reduction of protein solubility, both of which are crucial roles required for precipitants. Therefore, this B-factor analysis partly explains the action of these precipitants on the protein molecules in the solution for crystallization. Among the simulations with precipitants, B-factor values in simulations 3-a and 3-b were higher than those in simulations 1-a, 1-b, 2-a, and 2-b. This is compatible with the experimental results in crystallography, in which it was difficult to obtain high-quality crystals using PEG 8000 as a precipitant, and the resolution of crystal structure 3 was 2.75 Å (Table 1). The averaged distributions of precipitant molecules were analyzed for all of the simulations using the trajectories of the last 20 ns. Figures 4a and S2a show the distributions of ammonium ions in dots and sulfate ions in mesh for simulations 1-a and 1-b, respectively. The distributions of these two ions are similar, and, to our surprise, the distributions are not isotropic around the protein. To clarify what these anisotropic distributions mean, the molecular arrangement in the crystal structure was contrasted to the distribution. In Figure 1b, each protein molecule has the same surrounding environment and has direct contact with four surrounding protein molecules. One protein molecule and its four surrounding molecules were extracted as shown in Figures 4b and S2b. Then the distributions of the two kinds of ions (Figures 4a and S2a) and the extracted protein molecules (Figures 4b and S2b) were superimposed (Figures 4c and S2c, backside views in Figures 4d and S2d). Intriguingly, there were almost no distributions of precipitants in the areas where the protein molecule has contact with other molecules in the crystal structure. In other words, there was almost no overlap between the distribution areas of precipitants (results of the simulation) and contact sites between proteins (results of crystal structure analysis). Figures 5a and S3a show the averaged distributions of potassium and sodium ions in dots (as a sum of the two ions)

■

RESULTS AND DISCUSSION Crystal structures 1, 2, and 3 were obtained by using the different precipitants for crystallizing agents (Table 1 and Table Table 1. Crystallographic Propertiesa crystal structure precipitant space group a, b, c (Å) α, β, γ (deg) resolution (Å) no. reflections (Rfree set) Rwork/Rfree a

1

2

3

ammonium sulfate P41212 66.33, 66.33, 127.34 90.00, 90.00, 90.00 29.41−1.80 25617 (1316)

potassium sodium tartrate P41212 66.86, 66.86, 128.00

32.34−1.75 29882 (1519)

C2 64.24, 87.40, 66.50 90.00, 94.31, 90.00 36.49−2.75 9578 (514)

0.224/0.274

0.194/0.238

0.197/0.234

90.00, 90.00, 90.00

PEG 8000

More detailed data are provided in Table S1.

S1). The numbers of PANΔloop molecules in an asymmetric unit were 1, 1, and 2 for crystal structures 1, 2, and 3, respectively. The structures of these protein molecules are almost the same as shown in Figure 1a. For crystal structure 3, the protein molecule of only chain A is displayed because the structural difference between chain A and chain B is small. The space groups of crystal structures 1, 2, and 3 are P41212, P41212, and C2, respectively. According to our previous works, the precipitant used for crystallization strongly affects the space group of the crystal.41,42 Since the space groups for crystal structures 1 and 2 are the same, ammonium sulfate and potassium sodium tartrate, which were used as precipitants for the crystals, are considered to have a similar effect on the protein. Figure 1b−d shows how protein molecules in multiple asymmetric units are packed in the crystals. Obviously, the packing modes in crystal structures 1 and 2 are the same since they have the same space group and almost the same cell constants. There are eight protein molecules per one unit cell in all of the three crystal structures, while the size of the unit cell of crystal structure 3 is much smaller than those of the others, which means the proteins in crystal structure 3 are tightly packed. To clarify the effects of precipitants on protein molecules and crystal structures, we carried out MD simulations for a PANΔloop molecule without any precipitant (simulation 0) and with ammonium sulfate (simulations 1-a and 1-b), potassium C

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 1. Crystal structures 1, 2, and 3. (a) Superimposition of the three crystal structures 1 (green), 2 (cyan), and 3 (yellow, chain A only). (b−d) Crystal packing of structures 1 (b), 2 (c), and 3 (d). Unit cells are depicted by lines, and a, b, and c axes are depicted by arrows. In all of the three crystals, one unit cell contains eight protein molecules, each of which is depicted in a different color.

and tartrate ion in mesh for the last 20 ns of simulations 2-a and 2-b, respectively. The distributions of cations and tartrate ions were similar but not isotropic around the protein. In the same manner as that for ammonium sulfate, a protein molecule and four protein molecules making direct contact with the protein were extracted from crystal structure 2 (Figures 5b and S3b) and superimposed on the distributions of cations and tartrate ions in simulations 2-a and 2-b (Figures 5c and S3c, backside view in Figures 5d and S3d). Again, there was no overlap between the distribution areas of precipitants and contact sites of proteins. Ammonium sulfate and potassium sodium tartrate showed similar distributions in the simulations, which seems to be reasonable because these two precipitants produce crystal structures with the same space group. The next question is why these two precipitants are distributed anisotropically, in other words, why the precipitants gather at specific sites on the protein. This anisotropy can be explained by the electrostatic potential around the protein. Figure 6a,b shows the areas with large positive or negative potential values around the protein. The areas match well with the areas where precipitants are observed at a high rate (Figures 4a, 5a, S2a, and S3a). Therefore, the electrostatic potential around the protein determines the distribution of these precipitants, and the precipitant anisotropic distribution is responsible for the molecular packing in crystals. It is notable that ammonium sulfate and potassium sodium tartrate have similar ion charges;

that is, both of them are composed of monovalent cation(s) and divalent anion. On the basis of the findings in this study, we propose the mechanism of protein crystallization under the conditions of first two precipitants as follows. First, precipitant molecules are attracted to the protein surface due to the electrostatic potential and are distributed heterogeneously around the protein. This causes electrostatic screening of the protein surface charge, leading to suppression of the electrostatic repulsion of protein molecules and stabilization of the conformational fluctuation of protein molecules (Figure 3), both of which make the protein solubility lower. Then the rate of molecular contact between proteins is increased. Precipitant molecules covering some sites of the protein surface prevent other protein molecules from accessing the covered sites. Hence, the contact sites on the proteins are restricted, helping protein molecules to form regularly ordered clusters, that is, to crystallize with a specific space group. Figures 7a and S4a show the averaged distributions of PEG molecules for the last 20 ns of simulations 3-a and 3-b, respectively. PEGs cover most of the protein surface, being considerably different from the results for the other two precipitants. In crystal structure 3, each protein molecule has direct contact with 10 other molecules (Figures 7b and S4b). The protein molecule of chain A in the crystal structure was used as a center molecule in these figures (colored green) since the two molecules in an asymmetric unit are almost identical in structure and contact geometry. In a manner similar to that for D

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 2. Initial structures of the models built for simulations 0 (a), 1-a (b), 2-a (c), and 3-a (d). Protein molecules are depicted in cartoon (cyan), water molecules are in lines (red for oxygen, white for hydrogen), sodium ions are in spheres (purple), chloride ions are in spheres (light green), ammonium ions are in sticks (blue for nitrogen, white for hydrogen), sulfate ions are in sticks (yellow for sulfur, red for oxygen), potassium ions are in spheres (light pink), tartrate ions are in sticks (green for carbon, red for oxygen), and PEG molecules are in sticks (green for carbon, red for oxygen).

Figure 3. B-factors for protein amino acid residues. (a) Comparison among simulations 0 (red line), 1-a (green line), 2-a (blue line), and 3-a (magenta line). (b) Comparison among simulations 0 (red line), 1-b (green line), 2-b (blue line), and 3-b (magenta line). In the PANΔloop recombinant protein, residues 51−72 were replaced by an Ala-Ser dipeptide linker (numbered 51, 52), and five additional residues were connected to the N-terminal side (numbered −4, −3, ..., 0).

On the basis of the finding that protein molecules preferentially make contact with each other at the sites covered by PEGs in simulation, we propose the crystallization process using PEG as a precipitant as follows. In the initial stage of crystallization just after mixing the protein solution and crystallizing agents for droplets, PEGs are bound to some specific sites of protein molecules. Molecular contacts of proteins are formed with the progress of crystallization, accompanying the release of PEG molecules that were

the other two precipitants, the distributions of PEG molecules (Figures 7a and S4a) and the extracted crystal structure (Figures 7b and S4b) were superimposed and compared (Figures 7c,d and S4c,d). Although many parts of the protein surface are covered by PEGs or surrounding protein molecules, the sites that are not covered by PEGs in the simulations have no contact with other protein molecules in the crystal structure (those sites are indicated by red arrows in Figures 7c,d and S4c,d). E

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 4. Distributions of ammonium and sulfate ions in simulation 1-a and comparison with crystal structure 1. (a) Distributions of ammonium and sulfate ions in simulation 1-a. The distribution of ammonium ions is depicted in dots and that of sulfate ions is in mesh. They are colored according to the rate of the presence of those ions in each point over the last 20 ns in the simulation (Color scales for both ions are depicted at the bottom.). The protein molecule is depicted in cartoon (yellow). (b) Direct contact of protein molecules in crystal structure 1. A protein molecule (green) has direct contact with four other protein molecules (red, orange, magenta, and pink). (c, d) Superimposition of (a) and (b), viewed from two different angles.

observed in the crystal structure. While those previous studies focused on specific ion-residue interactions, our study shed light on the distribution of precipitant molecules. It is notable that a certain range of ion concentrations induced patchy attraction of lactoferrin molecules in Monte Carlo simulation with ions implicitly treated by Li et al.61 Analysis of protein crystal structures crystallized with different precipitants will provide important information on how precipitants affect crystal structures. Two crystallographic studies on thaumatin62 and lysozyme63 crystallized with stereoisomeric precipitants showed that the stereoisomers sometimes influenced the protein crystal. For example, different tartrate stereoisomers gave different crystal habits, i.e., space group and unit cell dimension, of thaumatin, which was explained from the difference in interactions of the precipitants with the protein. In contrast, the difference in protein crystals of lysozyme grown in the presence of one of the two enantiomers of 2-methyl-pentanediol (MPD) was marginal, although there were local conformational differences in a few residues because of the presence or absence of agents. These results suggested that, rather than the precipitant, the resulting crystal packing had a greater influence on the crystal structure. Therefore, biological aspects of a protein can be assessed better by comparing crystal structures with different packing conditions, as examined in our previous works.41,42

originally bound to the contact sites. The release of PEG molecules leads to an increase in the entropy of the system, which efficiently drives the crystallization. This can be regarded as a kind of depletion force, in which decrease of the depletion zone by increase of contacts of macromolecules is promoted by the entropic gain of PEG molecules in the solution.33,34 Accordingly, PEGs play a role of guiding protein molecules for making adequate contacts and packing, thereby leading to a specific space group in the crystal. The shape of the protein surface is a fundamental factor determining how proteins are packed in a crystal with formation of specific intermolecular contacts. Therefore, it is important to take into account the protein shape for understanding the protein crystallization. Computer simulation is one of the rational means to assist the analysis. In this study, we showed that precipitant molecules were distributed anisotropically around the protein surface, being directly linked to the crystal packing. Some computational studies have been carried out to analyze how ions affect the property of proteins. MD simulations by Vrbka et al.28 and Lund et al.60 showed that different kinds of ions preferred to be bound to different types of amino acid residues. An MD simulation by Fusco et al.23 showed that a high salt concentration caused electrostatic screening of charged residues on the protein surface, which led to stabilization of the intermolecular interaction that was F

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 5. Distributions of cations and tartrate ions in simulation 2-a and comparison with crystal structure 2. (a) Distributions of cations and tartrate ions in simulation 2-a. The distribution of cations (as a sum of sodium and potassium ions) is depicted in dots and that of tartrate ions is in mesh. They are colored according to the rate of the presence of those ions. The protein molecule is depicted in cartoon (yellow). (b) Direct contact of protein molecules in crystal structure 2. A protein molecule (green) has direct contact with four other protein molecules (red, orange, magenta, and pink). (c, d) Superimposition of (a) and (b), viewed from two different angles.

Figure 6. (a, b) Electrostatic potential map of crystal structure 1 (PDB code: 4ZQQ), viewed from two different angles. Positive and negative potential areas are depicted in blue and red mesh, respectively. The protein molecule is depicted in cartoon (yellow).

■

CONCLUSION

the distribution showed a strong anisotropy. MD simulations with precipitants at a high concentration equivalent to the crystallizing condition enabled us to clarify the molecular interactions of precipitants with the protein. Our calculations suggested that a heterogeneous distribution of precipitants played a role in restricting contact sites of the protein, thereby causing a specific molecular packing in the crystal and leading to growth in the corresponding space group. Accordingly, a precipitant that can sufficiently restrict the contact sites of

Three kinds of crystals were obtained with three different precipitants for the same kind of protein. The space group of one of the crystals was different from that of the others. The relationship between the precipitants used in protein crystallization and molecular packing in the crystals was investigated by combining crystallographic structure analysis and MD simulation. It was found that precipitant molecules were not randomly distributed around the protein, and, instead, G

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 7. Distribution of PEG molecules in simulation 3-a and comparison with crystal structure 3. (a) Distribution of PEGs in simulation. The distribution is shown in mesh. Mesh is colored according to the rate of PEG atom presence at each point over the last 20 ns in the simulation. The protein molecule is depicted in cartoon (yellow). (b) Direct contact of protein molecules in crystal structure 3. A protein molecule (green) has direct contact with 10 other protein molecules (10 different colors). (c, d) Superimposition of (a) and (b), viewed from two different angles.

protein molecules will be one of the critical elements for successful crystallization.

■

Tyuji Hoshino: 0000-0003-4705-4412 Notes

The authors declare no competing financial interest.

ASSOCIATED CONTENT

■

S Supporting Information *

ACKNOWLEDGMENTS This work was performed under the approval of the Photon Factory Program Advisory Committee (Proposal No. 2012G658, 2014G563). Calculations were performed at the Research Center for Computational Science, Okazaki, Japan, and at Information Technology Center of the University of Tokyo. A part of this work was supported by a grant for Scientific Research C from the Japan Society for the Promotion of Science. This work was also supported by a grant from the Japanese Agency for Medical Research and Development.

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.cgd.6b01385. Detailed crystallographic data. Number of molecules, total number of atoms, and box dimension in simulation. RMSD values of main chain atoms for MD simulation. Distributions of ammonium and sulfate ions in simulation 1-b. Distributions of cations and tartrate ions in simulation 2-b. Distribution of PEG molecules in simulation 3-b (PDF)

■

Accession Codes

Atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession codes 5JHT for crystal structure 2 and 5JHV for crystal structure 3.

■

REFERENCES

(1) McPherson, A.; Gavira, J. A. Acta Crystallogr., Sect. F: Struct. Biol. Commun. 2014, 70, 2−20. (2) Lomakin, A.; Asherie, N.; Benedek, G. B. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 9465−9468. (3) Galkin, O.; Vekilov, P. G. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 6277−6281. (4) Haas, C.; Drenth, J. J. Phys. Chem. B 2000, 104, 368−377. (5) Curtis, R. A.; Blanch, H. W.; Prausnitz, J. M. J. Phys. Chem. B 2001, 105, 2445−2452. (6) Vivares, D.; Bonnete, F. J. Phys. Chem. B 2004, 108, 6498−6507.

AUTHOR INFORMATION

Corresponding Author

*Phone: +81-43-226-2936, Fax: +81-43-226-2936 E-mail: [email protected]. ORCID

Satoshi Fudo: 0000-0001-9925-5131 H

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

(7) Dumetz, A. C.; Chockla, A. M.; Kaler, E. W.; Lenhoff, A. M. Cryst. Growth Des. 2009, 9, 682−691. (8) Liu, Y.; Wang, X.; Ching, C. B. Cryst. Growth Des. 2010, 10, 548− 558. (9) Vekilov, P. G. J. Phys.: Condens. Matter 2012, 24, 193101. (10) Haas, C.; Drenth, J.; Wilson, W. W. J. Phys. Chem. B 1999, 103, 2808−2811. (11) George, A.; Chiang, Y.; Guo, B.; Arabshahi, A.; Cai, Z.; Wilson, W. W. Methods Enzymol. 1997, 276, 100−110. (12) ten Wolde, P. R.; Frenkel, D. Science 1997, 277, 1975−1978. (13) Bhamidi, V.; Varanasi, S.; Schall, C. A. Langmuir 2005, 21, 9044−9050. (14) Lutsko, J. F.; Duran-Olivencia, M. A. J. Chem. Phys. 2013, 138, 244908. (15) Sleutel, M.; Lutsko, J.; Van Driessche, A. E. S.; Duran-Olivencia, M. A.; Maes, D. Nat. Commun. 2014, 5, 5598. (16) Savage, J. R.; Dinsmore, A. D. Phys. Rev. Lett. 2009, 102, 198302. (17) Streets, A. M.; Quake, S. R. Phys. Rev. Lett. 2010, 104, 178102. (18) Tan, P.; Xu, N.; Xu, L. Nat. Phys. 2013, 10, 73−79. (19) Sauter, A.; Roosen-Runge, F.; Zhang, F.; Lotze, G.; Jacobs, R. M. J.; Schreiber, F. J. Am. Chem. Soc. 2015, 137, 1485−1491. (20) Price, W. N., II; Chen, Y.; Handelman, S. K.; Neely, H.; Manor, P.; Karlin, R.; Nair, R.; Liu, J.; Baran, M.; Everett, J.; Tong, S. N.; Forouhar, F.; Swaminathan, S. S.; Acton, T.; Xiao, R.; Luft, J. R.; Lauricella, A.; DeTitta, G. T.; Rost, B.; Montelione, G. T.; Hunt, J. F. Nat. Biotechnol. 2009, 27, 51−57. (21) Cieslik, M.; Derewenda, Z. S. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2009, 65, 500−509. (22) Pellicane, G.; Smith, G.; Sarkisov, L. Phys. Rev. Lett. 2008, 101, 248102. (23) Fusco, D.; Headd, J. J.; De Simone, A.; Wang, J.; Charbonneau, P. Soft Matter 2014, 10, 290−302. (24) Taudt, A.; Arnold, A.; Pleiss, J. Phys. Rev. E 2015, 91, 033311. (25) Muschol, M.; Rosenberger, F. J. Chem. Phys. 1995, 103, 10424− 10432. (26) Collins, K. D. Methods 2004, 34, 300−311. (27) Zhang, Y.; Cremer, P. S. Curr. Opin. Chem. Biol. 2006, 10, 658− 663. (28) Vrbka, L.; Jungwirth, P.; Bauduin, P.; Touraud, D.; Kunz, W. J. Phys. Chem. B 2006, 110, 7036−7043. (29) Annunziata, O.; Payne, A.; Wang, Y. J. Am. Chem. Soc. 2008, 130, 13347−13352. (30) Arakawa, T.; Timasheff, S. N. Biochemistry 1984, 23, 5912− 5923. (31) Iyer, G. H.; Dasgupta, S.; Bell, J. A. J. Cryst. Growth 2000, 217, 429−440. (32) Vaney, M. C.; Broutin, I.; Retailleau, P.; Douangamath, A.; Lafont, S.; Hamiaux, C.; Prange, T.; Ducruix, A.; Ries-Kautt, M. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2001, 57, 929−940. (33) Asakura, S.; Oosawa, F. J. Polym. Sci. 1958, 33, 183−192. (34) Vivares, D.; Bonnete, F. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002, 58, 472−479. (35) Arakawa, T.; Timasheff, S. N. Biochemistry 1985, 24, 6756− 6762. (36) Israelachvili, J. Proc. Natl. Acad. Sci. U. S. A. 1997, 94, 8378− 8379. (37) Pittz, E. P.; Timasheff, S. N. Biochemistry 1978, 17, 615−623. (38) Bolen, D. W. Methods 2004, 34, 312−322. (39) Anand, K.; Pal, D.; Hilgenfeld, R. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002, 58, 1722−1728. (40) Deshpande, A.; Nimsadkar, S.; Mande, S. C. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2005, 61, 1005−1008. (41) Qi, F.; Fudo, S.; Neya, S.; Hoshino, T. Chem. Pharm. Bull. 2014, 62, 568−577. (42) Qi, F.; Fudo, S.; Neya, S.; Hoshino, T. J. Chem. Inf. Model. 2015, 55, 1673−1685. (43) Fudo, S.; Yamamoto, N.; Nukaga, M.; Odagiri, T.; Tashiro, M.; Neya, S.; Hoshino, T. Bioorg. Med. Chem. 2015, 23, 5466−5475.

(44) Fudo, S.; Yamamoto, N.; Nukaga, M.; Odagiri, T.; Tashiro, M.; Hoshino, T. Biochemistry 2016, 55, 2646−2660. (45) Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276, 307− 326. (46) Vagin, A.; Teplyakov, A. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66, 22−25. (47) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis, I. W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; GrosseKunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R. J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66, 213−221. (48) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66, 486−501. (49) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J.; Brothers, E. N.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö .; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Revision D.01; Gaussian, Inc.: Wallingford, CT, USA, 2009. (50) Cances, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032−3041. (51) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J. Phys. Chem. 1993, 97, 10269−10280. (52) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926−935. (53) Case, D. A.; Berryman, J. T.; Betz, R. M.; Cerutti, D. S.; Cheatham, T. E., III; Darden, T. A.; Duke, R. E.; Giese, T. J.; Gohlke, H.; Goetz, A. W.; Homeyer, N.; Izadi, S.; Janowski, P.; Kaus, J.; Kovalenko, A.; Lee, T. S.; LeGrand, S.; Li, P.; Luchko, T.; Luo, R.; Madej, B.; Merz, K. M.; Monard, G.; Needham, P.; Nguyen, H.; Nguyen, H. T.; Omelyan, I.; Onufriev, A.; Roe, D. R.; Roitberg, A.; Salomon-Ferrer, R.; Simmerling, C. L.; Smith, W.; Swails, J.; Walker, R. C.; Wang, J.; Wolf, R. M.; Wu, X.; York, D. M.; Kollman, P. A. AMBER 2015; University of California: San Francisco, 2015. (54) Wang, J. M.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004, 25, 1157−1174. (55) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Proteins: Struct., Funct., Genet. 2006, 65, 712−725. (56) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977, 23, 327−341. (57) The PyMOL Molecular Graphics System; Schrödinger, LLC: Cambridge, MA, 2010. (58) Li, L.; Li, C.; Sarkar, S.; Zhang, J.; Witham, S.; Zhang, Z.; Wang, L.; Smith, N.; Petukh, M.; Alexov, E. BMC Biophys. 2012, 5, 9. (59) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J. Comput. Chem. 2004, 25, 1605−1612. (60) Lund, M.; Vrbka, L.; Jungwirth, P. J. Am. Chem. Soc. 2008, 130, 11582−11583. (61) Li, W. M.; Persson, B. A.; Morin, M.; Behrens, M. A.; Lund, M.; Zackrisson Oskolkova, M. J. Phys. Chem. B 2015, 119, 503−508. (62) Asherie, N.; Jakoncic, J.; Ginsberg, C.; Greenbaum, A.; Stojanoff, V.; Hrnjez, B. J.; Blass, S.; Berger, J. Cryst. Growth Des. 2009, 9, 4189−4198. (63) Stauber, M.; Jakoncic, J.; Berger, J.; Karp, J. M.; Axelbaum, A.; Sastow, D.; Buldyrev, S. V.; Hrnjez, B. J.; Asherie, N. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2015, 71, 427−441.

I

DOI: 10.1021/acs.cgd.6b01385 Cryst. Growth Des. XXXX, XXX, XXX−XXX