Insight into the Dynamic Interaction of Different Carbohydrates with

May 7, 2010 - The unbinding process of three monosaccharides―galactose, glucose, and mannose―from human surfactant protein D (hSP-D) was ...
12 downloads 0 Views 4MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 7383–7390

7383

Insight into the Dynamic Interaction of Different Carbohydrates with Human Surfactant Protein D: Molecular Dynamics Simulations Jilong Zhang, Qingchuan Zheng, and Hongxing Zhang* State Key Laboratory of Theoretical and Computational Chemistry, Institute of Theoretical Chemistry, Jilin UniVersity, Changchun 130023, P. R. China ReceiVed: NoVember 28, 2009; ReVised Manuscript ReceiVed: April 21, 2010

The unbinding process of three monosaccharides;galactose, glucose, and mannose;from human surfactant protein D (hSP-D) was investigated by the molecular docking and molecular dynamics methods to explore the cause of different dynamic interaction between these monosaccharides and the protein. The results show that the low affinity of galactose for hSP-D is attributed to the different binding conformation from the other two monosaccharides. The sugar coordinates to the calcium ion by the hydroxyl groups in the C2 and C3 atoms, so it cannot form the effective interaction with hSP-D. Glucose and mannose have similar binding conformations with hSP-D. Their difference in the affinity is induced by the interaction between the hydroxyl group in the C2 atom and the residue Asp325. The direction of the hydroxyl group in mannose results in the formation of the hydrogen bond with Asp325 and further makes mannose hydrogen-bond to the residues Glu329 and Arg343 by the hydroxyl groups in the C3, C4, and C6 atoms. As glucose only forms three hydrogen bonds with the residues Glu321, Asn323, and Glu329 by the hydroxyl groups in the C3 and C4 atoms, its interaction with hSP-D is weaker than that of mannose. Thus glucose has a lower energy barrier of dissociation. This work could provide the more penetrating understanding of hSP-D physiological functions. 1. Introduction As the body’s gas-exchange organ, the lung is inevitably exposed to air that is contaminated with all kinds of pathogens, allergens, and pollutants. Hence its host-defense functions, which must facilitate clearance of inhaled pathogens and particles, are especially important for a human’s health. Pulmonary surfactant protein D (SP-D) is an important protein in a lung innate immune system. It binds a variety of potential ligands, including glycoconjugates expressed by micro-organisms and complex oligosaccharides associated with viral envelope proteins, fungal cell wall, and particulate organic antigens. In addition, some interactions also occur between SP-D and host cells. The consequences of these interactions include microbial aggregation and enhanced cellular uptake, inhibition of bacterial and fungal growth through effects on membrane permeability, modulation of cytokine production by phagocytic cells in response to microorganisms, enhanced clearance of apoptotic cells, and modulation of acquired immunity.1-8 Consequently, the protein is involved in many diseases, including allergic asthma,7,8 chronic obstructive pulmonary disease,9 influenza,4,10-12 human immunodeficiency virus (HIV),4,13 and so on. SP-D is the member of the collectin family of C-type lectins named for their amino-terminal collagen-like region and carboxyl-terminal lectin or carbohydrate recognition domain (CRD).14,15 It exists predominately as a tetramer of trimeric subunits (dodecamer) assembled into a cruciform structure. Each subunit is composed of three polypeptide chains characterized by an N-terminal region, a triple helical collagen-like region, a coiled-coil neck region, and the Ca2+-dependent CRD (Figure 1).15,16 The binding to many potential ligands frequently takes place in the collectin CRDs. For example, the CRD of human SP-D (hSP-D) binds influenza virus through the interaction with * To whom correspondence should be addressed. Tel: +86-43188498966. Fax: +86-431-88498966. E-mail: [email protected].

Figure 1. Trimeric monomer and dodecamer of SP-D. Four identical trimeric monomers link together by the N-terminal domain to form one dodecamer.

viral hemeagglutinin protein and neuraminidase envelope glucoproteins.10-12 Crouch et al.17 reported that SP-D exhibits calcium-dependent binding to solid-phase phosphatidylinositol. Meschi and co-workers13 pointed out that the binding of hSP-D to the envelope protein (gp 120) of HIV is mediated by its calcium-dependent carbohydrate-binding activity and is dependent on glycosylation of gp 120. In recent research, Nie et al.18 proved that hSP-D binds MD-2 through the CRD. Taken together, these studies indicated that the CRD of SP-D is extremely important for its immune responses and opsonization. Since the first structural details of the neck plus CRDs (NCRDs) of homotrimeric hSP-D complexed with carbohydrate ligands

10.1021/jp9113078  2010 American Chemical Society Published on Web 05/07/2010

7384

J. Phys. Chem. B, Vol. 114, No. 21, 2010

were reported,19 a lot of research on SP-D has focused on the key residues in its CRD. The species differences in the carbohydrate binding preferences show that the residues Asp324 and Asp325 in the primary carbohydrate binding site of hSP-D play vital roles in the recognition of ligands such as N-acetylmannosamine.20 The variation of the residue at position 343 of SP-D affects the binding of phosphatidylinositol, which is identified as the primary surfactant-associated ligand in crude surfactant.17,21,22 Additionally, the mutant R343 V of hSP-D shows increased binding to a subset of the oligosaccharides and markedly enhanced influenza A viral neutralizing activity.12 In a word, the residues in the CRD and adjacent site contribute to the binding to many potential ligands and are responsible for SP-D’s function as the immune mediator. The carbohydrate binding preferences of SP-D affect its functions in the body of animals. The high affinity of the collectin for clustered oligosaccharides is considered to be important for their ability to distinguish nonself from self, as most carbohydrates in animals are terminated by sugars, such as galactose or sialic acid, which are poorly recognized by the collectin. In most experiments, the affinity for SP-D toward different carbohydrates has been tested in a competitive enzymelinked immunosorbent assay (ELISA).16,20,23 The relative estimates of affinities are thermodynamic quantities where the amount of SP-D bound to a certain ligand is determined by the lowest free-energy state. The results of these experiments show that SP-D has the highest affinity for glucose and the lowest affinity for galactose of the three monosaccharides, mannose, glucose, and galactose. But the recent atomic force microscope (AFM) experimental results24 first revealed that the dynamic strength of the interaction between hSP-D and saccharide ligands is different from the thermodynamics stability of the same molecular complex. The unbinding forces indicate that mannose has higher dynamic strength than glucose. Considering the physiological environment of the lung, which always exibits inflation and deflation, a significant liquid flow will take place in the alveolar hypophase with the changes of the thickness of the aqueous layer. Also, in the alveolar, the size of the invading pathogens is comparable with the thickness of the aqueous layer. Thus the bond between glycoconjugates of pathogens and SP-D will be subjected to a high degree of mechanical stress. This makes the study on the dynamic characteristics more significant. Accordingly, based on the results from the molecular docking and conventional molecular dynamics (MD) simulations, steered molecular dynamics (SMD) simulations and the calculations of the dissociation free energies were performed to study the changes of the pulling forces in the unbinding process of monosaccharides from hSP-D and explore the mechanism that results in the different dynamic affinities. In the SMD simulations, mimicking the AFM experiment, time-dependent external forces were applied to the monosaccharide ligands to facilitate their unbinding from hSP-D, which usually could not be achieved by conventional MD simulations. At the same time, the free-energy profiles along the dissociation pathways also give the corresponding energy barriers. According to the profiles, the snapshots of the dynamics trajectories corresponding to the peaks were statistically checked and analyzed to determine the causes of the different dynamic strengths on the atomic level. 2. Materials and Methods The molecular docking of monosaccharides to hSP-D was performed on a Dell Precision WorkStation T5400 with the CDOCKER25 module in the Discovery Studio 2.1 software package.26 The conventional MD and SMD simulations were

Zhang et al. carried out on the Dawning 5000A server with the parallel MD program NAMD 2.6.27 Topology and force field parameters were assigned from the CHARMM27 protein lipid parameter set28 for hSP-D and from the CHARMM22 sugar parameter set29 for three monosaccharides. 2.1. Molecular Docking. Initial coordinates of hSP-D for molecular docking were downloaded from the Protein Data Bank (www.rcsb.org), PDB code: 1PWB, which included three saccharide ligands (two glucoses and one maltose) and the NCRDs of the protein. All the saccharide ligands in the crystal structure were deleted to perform the molecular docking study. Following that, CDOCKER was used to dock galactose, glucose, and mannose (Figure S1, Supporting Information) into the CRDs of the hSP-D in chains A, B, and C, respectively. CDOCKER is a grid-based molecular docking method that employs CHARMm. The receptor hSP-D was held rigid while three monosaccharide ligands were allowed to flex during the refinement. Because of the existence of the original ligands in the crystal structure, prior knowledge of the binding site was acquired. Thus it was possible to specify the ligand placement in the active site using a binding site sphere with a radius of 10 Å. Random monosaccharide ligands’ conformations were generated from the initial ligand structures through high-temperature MD at 1000 K, followed by random rotations. The random conformations were refined by grid-based simulated annealing and a final full force field minimization. The top 10 poses were saved for comparison and analysis. Finally, the pose with the lowest CDOCKER interaction energy was used for the subsequent MD simulations. 2.2. MD Simulations. The conformations obtained by the molecular docking were used to perform the MD simulations. The hydrogen atoms were added with the VMD program.30 TIP3P31 water molecules were used to solvate the whole system in two steps. A 10 Å water shell was added by using Solvate 1.0 (http://www.mpibpc.mpg.de/home/grubmueller/downloads/ solvate/index.html) in the first step. In the second step, VMD’s solvate plugin was used to supplement more water molecules, forming a cubic box with dimensions 11.17 × 11.02 × 11.29 nm3. According to the physiological concentration, 154 mM, 87 Na+, and 116 Cl- ions were incorporated to ensure the overall neutrality of the system. The final system contains 160051 atoms. Periodic boundary conditions were applied to the system to obtain consistent behavior. van der Waals interactions were gradually turned off at a distance between 1.2 and 1.4 nm. The particle mesh Ewald (PME) method32,33 was employed every one step for the computation of electrostatic forces. The nonbonded pair list was updated every 20 steps. An integration time step of 1 fs was assumed. The whole system was first energy minimized with 10 000 steps of conjugate gradients, by keeping the coordinates of protein backbone atoms and ligands restrained with a spring constant of 10 kcal mol-1 Å-2. After minimization, at the same restraint, the temperature of the system was gradually increased with the Langevin dynamics34 method from 100 to 300 K by stepwise reassignment of velocities every 1 ps. Then another 200 ps MD simulation was performed in the canonical (NVT) ensemble. During this period, the constraint was gradually decreased to 0 with a step length of 2 kcal mol-1 Å-2. Following that, the pressure of the system was coupled to a reference pressure of 1 bar with a modified Nose´-Hoover Langevin piston method.35 In the constant pressure/constant temperature (NPT) conditions, a 4 ns MD simulation was performed on the whole system to ensure the accomplishment

Dynamic Interaction of Different Carbohydrates with hSP-D

Figure 2. Sideview of the NCRDs of hSP-D. The arrowheads in the springs indicate the pulling directions of external forces in the SMD simulations. Image rendering was done with VMD.30

of the equilibrium. The final stable state was saved as a restart point for the following SMD simulations. 2.3. SMD Simulations. In our recent work,36 glucose was taken as an example to explore the effect of all kinds of different factors on the outcomes of the pulling force. The results show that, under the velocity of 0.05 Å ps-1 and the force constant of 0.5 kcal mol-1 Å-2, the simulations give a better curve of the pulling force, while the different pulling directions and reference points have no apparent effect on the outcomes of the pulling force. Consequently, in the present simulations, the same velocity and force constant were adopted to pull three monosaccharide ligands along the same direction as the 3-fold axis of the neck region of hSP-D (Figure 2). In this process, keeping the R-carbon (CR) atom of Ser206 in each subunit fixed, the external steering force was in turn applied on the mass center of each monosaccharide. The pulling force experienced by the reference point was calculated by employing eq 1:

F(t) ) k(Vt - x)

(1)

where k is the force constant for the spring, and x is the displacement of the reference point from its original position. During the SMD simulations, the steering force was only applied along the pulling direction. Each monosaccharide was free from constraint in the plane perpendicular to the pulling direction. The trajectories were saved every 0.5 ps, and steering forces were recorded every 20 fs. The SMD simulations for each monosaccharide along the same pathway were repeated three times with the same starting structure and different random seeds. 2.4. Free-Energy Calculations. To investigate the dissociation of monosaccharides from hSP-D, the reaction coordinate, ξ, was chosen as the distance separating the center of mass of monosaccharide ligands from the corresponding coordinated calcium ion. Variation of the free energy, ∆G(ξ), along ξ was determined using the adaptive biasing force (ABF) method,37 which relies upon the integration of the average force acting on ξ. In the NAMD implementation of ABF,38 the force is evaluated within the classical thermodynamic integration formalism.39 The free-energy derivative, dG(ξ)/dξ, is estimated locally throughout the simulations, thus providing a continuous update of the biasing force. Once applied to the system, the bias yields a Hamiltonian in which no net average force acts

J. Phys. Chem. B, Vol. 114, No. 21, 2010 7385

Figure 3. Superimposition of the docking conformation of glucose on that in the crystal structure. The glucose is rendered as licorice. The orientation in the docking result is colored in purple, and that in the crystal structure is colored in yellow. Image rendering was done with VMD.30

along ξ. As a result, all values of the reaction coordinate are sampled with the same probability, which, in turn, greatly improves the accuracy of the calculated free energies. To further increase the efficiency of the calculation, the dissociation pathway of monosaccharides from hSP-D, that is, 4 e ξ e 15 Å, was divided into 11 nonoverlapping windows. For each window, up to 3 ns of MD trajectory was generated. The initial Cartesian coordinates of the system for each window were obtained from the SMD simulations. Instantaneous values of the force were accrued in bins 0.01 Å wide. The 33 ns ABF simulations were performed for every monosaccharide ligand. The expression given by Rodriguez-Gomez et al.40 was used to estimate the standard error of the free-energy difference. 3. Results and Discussion 3.1. Docking of Three Monosaccharides. Glucose was included in the crystal structure downloaded from the Protein Data Bank. However, in order to assess the accuracy of CDOCKER, it was deleted from the CRDs of hSP-D and then docked into the same site in chain B of the protein in the molecular docking study. The docking conformation of glucose was superimposed onto that in the crystal structure. It can be seen in Figure 3 that two conformations fit very well. The rootmean-square deviation (rmsd) between the two conformations is only 0.43 Å, which is comparable with the rmsd value induced by the thermal fluctuation in MD simulations. So the docked result of glucose proves that CDOCKER can provide reasonable conformations for the following molecular docking of the other two monosaccharides. Subsequently, galactose, and mannose were docked into chain A and chain C of hSP-D in the same procedure as glucose, respectively. The docked results (Figure 4) show that mannose adopts the similar binding manner with glucose. The hydroxyl groups of mannose in the C3 and C4 atoms coordinate to the calcium ion in the CRD of chain C. As the diastereomer of glucose, mannose just has the opposite direction of the hydroxyl group in the C2 atom. Thereby it is not unexpected that the monosaccharide shares the same coordination pattern as glucose. In the crystal structures with PDB access code 2RIA, 2ORJ, 3G81, 3G83, 3G84, each of which binds the ligand with mannose moiety, all the mannose moieties coordinate to the

7386

J. Phys. Chem. B, Vol. 114, No. 21, 2010

Zhang et al.

Figure 4. Docking conformations of (a) galactose, (b) glucose, and (c) mannose obtained by CDOCKER. Image rendering was done with VMD.30

corresponding calcium ions by the hydroxyl groups in the C3 and C4 atoms. The coincidence between the docking result of mannose and the X-ray experimental structures further verifies the accuracy of CDOCKER. However, the docked conformation of galactose is different from that of the other two monosaccharides. It coordinates to the corresponding calcium ion by the hydroxyl groups in the C2 and C3 atoms (Figure 4a). In order to evaluate the rationality of the docked conformation, several X-ray crystal structures with bound ligands were inspected and compared in our research. Although the ligands in these structures are different, most of them coordinate to the corresponding calcium ions by the hydroxyl groups in the C3 and C4 atoms of pyranoses. The hydroxyl groups in the two sites form the equatorial bonds with the correlative carbon atoms in the pyran ring of each saccharide with chair form. Additionally, the dihedral angles between two hydroxyl groups are mostly around -60°. In the crystal structure named 2RIB, a kind of alternative pattern is the coordination of two hydroxyl groups in the side chain of its ligand to the calcium ion. Although they do not belong to the pyran ring, the dihedral angle of -60° is still kept. This case looks like that of two hydroxyl groups with the equatorial bonds in the

pyranose ring. However, the situation is distinct for galactose. As another diastereomer of glucose, the only difference between them is the opposite direction of the hydroxyl group in the C4 atom (Figure S1). But the difference results in the change of the hydroxyl group in the site from an equatorial bond to an axial bond. So there are not two adjacent hydroxyl groups with a dihedral angle of -60°. The hydroxyl groups in the C2 and C3 atoms are the only pair of hydroxyl groups with equatorial bonds. On the basis of the comparison and analysis above, the pair of hydroxyl groups is most likely coordinated to the calcium ion. Therefore the docking result of galactose is reasonable. Such a different coordination pattern leads to a different orientation of galactose, which makes the monosaccharide seem a little far from the hSP-D. So it is not difficult to imagine the different nature of galactose from the other two monosaccharides in the nonequilibrium procedure. In the present study, the three monosaccharides are all R-anomers. It should be pointed out that only the anomer of galactose has influence on the dissociation from hSP-D, which will be analyzed in the following discussion. Because of the specific binding pattern of glucose and mannose (Figure 4b,c), the hydroxyl groups in their C1 atoms are far away from the

Dynamic Interaction of Different Carbohydrates with hSP-D

Figure 5. Time dependence of the external forces for three carbohydrates during the SMD simulations. The peak values of the force for galactose, glucose, and mannose correspond to 637, 884, and 913 pN, respectively. After the peak value, the curves show the motion of each monocarbohydrate in water.

binding site of hSP-D. On the other hand, the carbohydratebinding site of hSP-D looks like bit shallow, so no effective interaction occurs between the protein and the two monosaccharides. Under such conditions, their anomers have no significant influence on the dissociation. However, the binding pattern of galactose is unique according to the discussion above. The result is that the hydroxyl group in its C1 atom affects its dissociation, which causes the differences between the anomers. 3.2. MD Simulations. On the basis of the molecular docking results of CDOCKER mentioned above, conventional MD simulations were carried out to obtain a more stable conformation for the sequential SMD simulations. Although the crystal structure used in the present study is only composed of the neck region and the carbohydrate recognition region of hSP-D, many studies have shown that its NCRDs contain all the information necessary to direct normal CRD folding, intrachain disulfide bond formation, coordination of calcium ions, and ligand binding activity.5,26 A 4 ns MD simulation in the NPT ensemble was performed on the whole system. The structural stability of hSP-D was assessed by calculation of the rmsd values. The rmsd’s of all atoms and of CR atoms of the protein from the starting point in the stage are shown as a function of the simulation time in Figure S2 in the Supporting Information. The rmsd profiles indicate that the conformation of the protein appears to be equilibrated in a short time at the beginning of the NPT stage. The rmsd values are generally stabilized near 2.9 and 2.4 Å for all atoms and CR atoms, respectively. After that, SMD simulations were performed to explore the different dynamic strengths. 3.3. SMD Simulations. The time scales for the conventional MD simulations are generally confined to the order of tens of nanoseconds, in which the unbinding event is difficult to reproduce. Thus, in order to drive the complex into the dissociation state of ligands, SMD simulations were further run based on the stable structure from the conventional MD simulation of the carbohydrates-hSP-D complex. The force curves from SMD simulations were used to compare the values of the external forces and determine the nonequilibrium natures. As shown in the force profiles in Figure 5, the pulling forces of the three carbohydrates keep increasing linearly in the initial stage of the pulling process. Afterward, at nearly 0.5 ns, the pulling force of galactose increases to the maximum, which is just about 637 pN. At this moment, galactose is first pulled out from the CRD of hSP-D. Its pulling force subsequently rapidly decreases to zero and shows minor fluctuations near zero. For

J. Phys. Chem. B, Vol. 114, No. 21, 2010 7387 the other two monosaccharides, the pulling forces do not increase until they reach the peak value near 0.7 ns. The maximum of the exerted force for glucose is smaller than that for mannose, and the peak value comes up later. After the peaks, the curves of the pulling forces show that the unbound monosaccharides are pulled in the physiological solution, so the force values show a minor fluctuation near zero. The results of the SMD simulations are shown to be in agreement with the AFM experiment in quality.24 The differences in the numerical values are to a great extent caused by the pulling velocity in the SMD simulations, as described in eq 1. For the three monosaccharides selected in our simulations, galactose has the much lower peak value of pulling force than the other two sugars, which is attributed to its weaker dynamic interaction with hSP-D. The ELISA data20 also shows that galactose has the lowest affinity for hSP-D in the thermodynamic stable state. As discussed in the molecular docking above, the reason is that galactose adopts an obviously different binding pattern from the other two monosaccharides, which weakens the interaction between galactose and hSP-D, not only in the equilibrium state but also probably in the nonequilibrium state. Glucose and mannose have the similar docking conformations, and the peak values of the pulling forces are also proximal. However, the peaks are still distinguishable. In the present research, the SMD simulation for every monosaccharide was repeated three times. In these simulations, all three sugars show fine reproducibility. The pulling force of glucose has an average value of 888 pN with a fluctuation of about 4 pN. Also, the average value of the pulling force for mannose is 915 pN with a fluctuation of less than 5 pN. These computational values confirm that mannose withstands higher force than glucose in the pulling procedure. However, in the steady state of thermodynamics, the relative affinities of the other two monosaccharides for hSP-D are just opposite to that. Their docking conformations only give an explanation for the affinities in the equilibrium state, but not in the nonequilibrium state. In order to further explore the nonequilibrium interaction between the sugars and hSP-D, the ABF method was used to compute the variation of the free energies during the dissociation process. 3.4. Free Energies of Dissociation. The free-energy profiles characterizing the dissociation of three monosaccharides from the CRDs of hSP-D are shown in Figure 6. The minima for three sugars in the profiles are located at the sites where the reaction coordinates are equal to 4.94 Å for galactose, 4.65 Å for glucose, and 4.64 Å for mannose, respectively. These minima correspond to the equilibrium states with the lowest free energy, which give the same affinity order as the ELISA experiments20 with free-energy values of -2.89, -5.75, and -5.52 kcal/mol for galactose, glucose, and mannose, respectively. At the same time, the values of the reaction coordinates of three sugars also prove the above observation that galactose is farther from hSP-D than the other two monosaccharides. After the minima in the free-energy profiles, as three sugars move away from the hSPD, the interaction between them is progressively disrupted. This leads to an abrupt increase of the free energy. As the separation of three monosaccharides and hSP-D further increases, the freeenergy profiles reach peak values, which are 1.83, 2.28, and 3.97 kcal/mol for galactose, glucose, and mannose, respectively. Then the profiles begin to drop down and, before long, gradually increase and reach plateaus. At the stages of the plateau, the free-energy profiles show the dissolvation of three monosaccharides into water. Accordingly, the corresponding free-energy profiles can be used to anchor the ABF curves. Also, some small errors in the computation could possibly result in a small shift

7388

J. Phys. Chem. B, Vol. 114, No. 21, 2010

Figure 6. The free-energy profiles for three carbohydrates in the unbinding process. For each curve, the first valley corresponds to the stable state with the lowest free energy, and the following peak corresponds to the intermediate in the dissociation process. After that, the curve shows the change in energy during the solvation progress of each monocarbohydrate.

of the computed profile of the dissociation free energy, but not affect the shape of the free-energy profile. The corresponding free-energy values of the plateaus are 6.0, 2.9, and 1.1 kcal/ mol for galactose, glucose, and mannose, respectively. The enthalpy values of solution for the three sugars measured in the experiments are 3.85, 2.63, and 1.92 kcal/mol, respectively.41,42 The computed free-energy values are in agreement with the experimental values, which proves that the computed profiles are reliable. As shown in Figure 6, the minima in the profiles, which correspond to the stable states with the lowest free energy, show the order of affinities as follows: glucose > mannose > galactose. However, the following peak values display a different order: mannose > glucose > galactose. As glucose and mannose have proximal free-energy values in the lowest points of the profiles and mannose has a larger value than glucose in the peaks, the free-energy barriers for the two sugars will become different from the affinities in the stable states. The free-energy barriers of mannose and glucose between the two states are 9.49 and 8.03 kcal/mol, respectively. This indicates that, in order to overcome the free-energy barriers, more work needs to be done or larger force needs to be exerted for mannose. For galactose, the value of the free-energy barrier is 4.72 kcal/mol. The computed results of these energy barriers are consistent with the pulling forces in the SMD simulations and comparable with those of the AFM experiment. The measured force values for mannose, glucose, and galactose in the AFM experiment24 are 54, 47, and 38 pN, respectively. These values, together with the above-mentioned free-energy values, suggest that the exerted forces have an intrinsic relationship. The free-energy barriers can be used to explore the cause of the different nonequilibrium nature. For galactose with a different binding pattern, the free-energy profile gives a small barrier, so it is apparent that pulling the carbohydrate will not need a larger force. At the same time, because galactose also has a lower affinity for hSP-D in the equilibrium state, the properties between the equilibrium and nonequilibrium states remain consistent. For mannose and glucose, they have a similar binding pattern and a proximal affinity in the stable state. However, mannose can bear a higher dynamic strength than glucose, which is just opposite to the affinity in the stable state,

Zhang et al. so the affinities of the two sugars in the stable state are not the only crucial factors of the dynamic strength. In the free-energy profiles, it can be clearly seen that the peak values of the freeenergy profiles also overwhelmingly affect the values of the free-energy barriers and accordingly affect the dynamic strength of mannose and glucose. It is the main cause of the deviation of the nonequilibrium nature from the equilibrium nature. 3.5. Atomic Insight of the Dynamic Strength. In order to obtain an explanation at the atomic level, by the statistical method, we checked the MD trajectories corresponding to the peak values of the free energies for the existence of hydrogen bonds. The hydrogen bonds with the biggest occupancy are all rendered in Figure 7. For galactose, the two hydrogen bonds rendered in Figure 7a have an occupancy of 35% in all the checked trajectories, while the occupancy of the other hydrogen bonds has a maximal value of just 3%. Also, in Figure 7b, the occupancies of the hydrogen bonds for the residues Glu321, Asn323, and Glu329 are 39%, 32%, and 37%, respectively. For mannose, the value for the residue is 41% for Asp325, 30% for Glu329, and 27% for Arg343. The corresponding conformations of the three sugars were checked to determine the atomic mechanisms that induce their dynamic differences. In the docking conformation, galactose adopts a pose That is far away from the CRD of hSP-D. During the dissociation procedure, the affinity of galactose for hSP-D is a bit lower than that of the other two monosaccharides. As shown in Figure 7a, at this time of unbinding, galactose only forms two hydrogen bonds with the coordination residue Glu321 in the CRD, while the other coordination residues such as Asn 323, Glu329, and Arg343 have larger distances from galactose and interact poorly with the monosaccharide. The result for galactose is caused by its distinct binding conformation with hSP-D. Because of the coordination of the hydroxyl groups in the C2 and C3 atoms to the calcium ion, the other hydroxyl groups of galactose are distant from the relative residues in the CRD. During the dissociation process, these residues cannot effectively interact with the hydroxyl groups as much as possible to prevent the departing of galactose. Only the residue Glu321 forms two hydrogen bonds with the C1 and C2 hydroxyl groups of galactose and slightly delays the dissociation of galactose. On the other hand, for the R-anomer of galactose, the direction of the hydroxyl group in the C1 atom also makes it possible to form a hydrogen bond between the hydroxyl group and the residue Glu321, while the β-anomer will lose the hydrogen bond with Glu321. For glucose and mannose, they have a similar binding conformation with hSP-D. The only differences between the two sugars are the directions of the hydroxyl groups in their C2 atoms. This little distinction results in glucose’s higher affinity in the thermodynamic stable state. It can be inferred that the different dynamic strength of interaction between the two monosaccharides and hSP-D also results from the difference in the directions of the hydroxyl group. In Figure 7b, glucose forms three hydrogen bonds with the residues Glu321, Asn323, and Glu329 through the hydroxyl groups in the C3 and C4 atoms. However, effective interaction between the other hydroxyl groups of glucose and the residues in the CRD does not take place, which makes the binding of glucose to hSP-D weaker than that of mannose during the dissociation process. Especially, the hydroxyl group of glucose in the C2 atom does not have any hydrogen bond with the relative residues. The case is different for mannose. In Figure 7c, the hydroxyl group of mannose in the C2 atom, which locates in the opposite direction to glucose, strongly interacts with the residue Asp325 by the

Dynamic Interaction of Different Carbohydrates with hSP-D

J. Phys. Chem. B, Vol. 114, No. 21, 2010 7389

Figure 7. Conformations of (a) galactose, (b) glucose, and (c) mannose corresponding to the peak values of the free-energy profiles. Image rendering was done with VMD.30

hydrogen bond. The consequence of the interaction is the approach of mannose to the residues Glu329 and Arg343. Although it results in the disappearance of the possible hydrogen bonds between mannose and the residues Glu321 and Asn323, the hydroxyl groups of mannose in the C3 and C4 atoms form two hydrogen bonds with another residue Glu329. Moreover, an additional hydrogen bond forms between the hydroxyl group in the C6 atom and the residue Arg343 owing to mannose’s approach to Arg343. So a total of four hydrogen bonds have emerged between mannose and hSP-D during its dissociation process from the protein. These hydrogen bonds strengthen the interaction and increase the unbinding energy barrier of mannose. Comparing mannose with glucose, most of the hydroxyl groups in the former (four-fifths) interact with hSP-D by the hydrogen bonds, while glucose just contributes two-fifths of its hydroxyl groups. The difference in the usage rate of the hydroxyl

groups is actually caused by their different conformations at the time of dissociation. The different conformations just result from the different directions of the hydroxyl groups in the C2 atom of the two monosaccharides. In the thermodynamic stable state, two sugars adopt the similar binding pose with hSP-D. The direction of the hydroxyl group of glucose in the C2 atom makes its affinity for hSP-D stronger than mannose. However, during the departing process, because of the interaction between the hydroxyl group and the residue Asp325, mannose is drawn close to Asp325 and thereafter close to the residues Glu329 and Arg343. A stronger interaction is produced between mannose and hSP-D and thereby raises the unbinding energy barrier of mannose. 4. Conclusions In the present study, molecular docking and MD simulations were used to investigate the unbinding process of three

7390

J. Phys. Chem. B, Vol. 114, No. 21, 2010

monosaccharides;galactose, glucose, and mannose;from hSP-D and explore the cause of the different dynamic strengths of the interaction between the sugars and hSP-D. The docking results of CDOCKER show that the binding pattern of galactose to hSP-D is different from that of the other two monosaccharides. This results in the low affinity of galactose for hSP-D not only in the stable state but also in the dynamic process. The SMD forces and ABF energy profiles have both proved that galactose bears a smaller force during the unbinding process. For glucose and mannose with a similar docking conformation to hSP-D, the investigation at the atomic level reveals that glucose forms three hydrogen bonds with the residues Glu321, Asn323, and Glu329, while four hydrogen bonds between mannose and hSP-D are formed between the four hydroxyl groups of mannose and the residues Asp325, Glu329, and Arg343. Because the hydroxyl group of mannose in the C2 atom is in the opposite direction of glucose, it strongly interacts with the residue Asp325 during the dissociation and thereby leads to the formation of hydrogen bonds with the other two residues. These interactions delay the departing process of mannose and increase the unbinding energy barrier. The corresponding hydroxyl group of glucose does not have an analogous interaction with the relative residues, so hSP-D cannot effectively prevent its dissociation. The result is that glucose has a lower energy barrier than mannose. Thus it can be seen that the exact discrimination of hSP-D to different sugars, which is important for its physiological functions, to a large extent relies on the hydroxyl group’s direction of sugars and the relative residues of the protein. Acknowledgment. This work is supported by the Natural Science Foundation of China, Key Projects in the National Science & Technology Pillar Program, Specialized Research Fund for the Doctoral Program of Higher Education, and the Specialized Fund for the Basic Research of Jilin University (Grant Nos. 20903045, 2006BAE03B01, 20070183046, and 200810018). Supporting Information Available: Figure S1 showing the geometries of galactose, glucose and mannose and Figure S2 containing the rmsd’s of the protein in the NPT ensemble simulations. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Wright, J. R. Nat. ReV. Immunol. 2005, 5, 58. (2) Erpenbeck, V. J.; Krug, N.; Hohlfeld, J. M. Naunyn-Schmiedeberg’s Arch. Pharmacol. 2009, 379, 217. (3) Kingma, P. S.; Whitsett, J. A. Curr. Opin. Pharmacol. 2006, 6, 277. (4) Vigerust, D. J.; Shepherd, V. L. Trends Microbiol. 2007, 15, 211. (5) Gupta, G.; Surolia, A. Bioessays 2007, 29, 452. (6) Kishore, U.; Greenhough, T. J.; Waters, P.; Shrive, A. K.; Ghai, R.; Kamran, M. F.; Bernal, A. L.; Reid, K. B. M.; Madan, T.; Chakraborty, T. Mol. Immunol. 2006, 43, 1293. (7) Pastva, A. M.; Wright, J. R.; Williams, K. L. Proc. Am. Thorac. Soc. 2007, 4, 252. (8) Haczku, A. J. Allergy Clin. Immunol. 2008, 122, 861.

Zhang et al. (9) Dahl, M. Am. J. Respir. Crit. Care Med. 2008, 177, 1177. (10) Hartshorn, K. L.; Crouch, E. C.; White, M. R.; Eggleton, P.; Tauber, A. I.; Chang, D.; Sastry, K. J. Clin. InVest. 1994, 94, 311. (11) Hartshorn, K. L.; Webby, R.; White, M. R.; Tecle, T.; Pan, C.; Boucher, S.; Moreland, R. J.; Crouch, E. C.; Scheule, R. K. Respir. Res. 2008, 9, 65. (12) Crouch, E.; Hartshorn, K.; Horlacher, T.; McDonald, B.; Smith, K.; Cafarella, T.; Seaton, B.; Seeberger, P. H.; Head, J. Biochemistry 2009, 48, 3335. (13) Meschi, J.; Crouch, E. C.; Skolnik, P.; Yahya, K.; Holmskov, U.; Leth-Larsen, R.; Tornoe, I.; Tecle, T.; White, M. R.; Hartshorn, K. L. J. Gen. Virol. 2005, 86, 3097. (14) Holmskov, U. L. APMIS Suppl. 2000, 100, 1. (15) Kuroki, Y.; Voelker, D. R. J. Biol. Chem. 1994, 269, 25943. (16) Persson, A.; Chang, D.; Crouch, E. J. Biol. Chem. 1990, 265, 5755. (17) Crouch, E.; McDonald, B.; Smith, K.; Roberts, M.; Mealy, T.; Seaton, B.; Head, J. Biochemistry 2007, 46, 5160. (18) Nie, X. M.; Nishitani, C.; Yamazoe, M.; Ariki, S.; Takahashi, M.; Shimizu, T.; Mitsuzawa, H.; Sawada, K.; Smith, K.; Crouch, E.; Nagae, H.; Takahashi, H.; Kuroki, Y. Biochemistry 2008, 47, 12878. (19) Shrive, A. K.; Tharia, H. A.; Strong, P.; Kishore, U.; Burns, I.; Rizkallah, P. J.; Reid, K. B. M.; Greenhough, T. J. J. Mol. Biol. 2003, 331, 509. (20) Crouch, E. C.; Smith, K.; McDonald, B.; Briner, D.; Linders, B.; McDonald, J.; Holmskov, U.; Head, J.; Hartshorn, K. Am. J. Respir. Cell Mol. Biol. 2006, 35, 84. (21) Persson, A. V.; Gibbons, B. J.; Shoemaker, J. D.; Moxley, M. A.; Longmore, W. J. Biochemistry 1992, 31, 12183. (22) Ogasaware, Y.; Kuroki, Y.; Akino, T. J. Biol. Chem. 1992, 267, 21244. (23) Lu, J.; Willis, A. C.; Reid, K. B. M. Biochem. J. 1992, 284, 795. (24) Thormann, E.; Dreyer, J. K.; Simonsen, A. C.; Hansen, P. L.; Hansen, S.; Holmskov, U.; Mouritsen, O. G. Biochemistry 2007, 46, 12231. (25) Wu, G. S.; Robertson, D. H.; Brooks, C. L.; Vieth, M. J. Comput. Chem. 2003, 24, 1549. (26) DiscoVery Studio 2.1; Accelrys Inc.: San Diego, 2008. (27) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale´, K.; Schulten, K. J. Comput. Chem. 2005, 26, 1781. (28) MacKerell, A. D., Jr.; Bashford, D.; Bellott, M.; Dunbrack, R. L., Jr.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; Joseph-McCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E., III; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wio´rkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102, 3586. (29) Guvench, O.; Greene, S. N.; Kamath, G.; Brady, J. W.; Venable, R. M.; Pastor, R. W.; MacKerell, A. D., Jr. J. Comput. Chem. 2008, 29, 2543. (30) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14, 33. (31) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926. (32) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089. (33) Toukmaji, A. Y.; Board, J. A., Jr. Comput. Phys. Commun. 1996, 95, 73. (34) Hoover, W. G. Phys. ReV. A 1985, 31, 1695. (35) Feller, S. E.; Zhang, Y. H.; Pastor, R. W. J. Chem. Phys. 1995, 103, 4613. (36) Zhang, J. L.; Zheng, Q. C.; Zhang, H. X. Chem. Phys. Lett. 2010, 484, 338. (37) Darve, E.; Pohorille, A. J. Chem. Phys. 2001, 115, 9169. (38) He´nin, J.; Chipot, C. J. Chem. Phys. 2004, 121, 2904. (39) den Otter, W. K. J. Chem. Phys. 2000, 112, 7283. (40) Rodriguez-Gomez, D.; Darve, E.; Pohorille, A. J. Chem. Phys. 2004, 120, 3563. (41) Hendricks, B. C.; Steinbach, W. H., Jr.; LeRoy, R. H.; Moseley, A. G., Jr. J. Am. Chem. Soc. 1934, 56, 99. (42) Schwarz, F. P. J. Solution Chem. 1996, 25, 471.

JP9113078