This article was downloaded by: [University of Groningen]On: 01 November 2012, At: 10:40Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Journal of Biomolecular Structure and DynamicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/tbsd20

A Theoretical Elucidation of Glucose Interaction withHSA's DomainsRasoul Nasiri a , Homayoon Bahrami a , Mansour Zahedi a , Ali Akbar Moosavi-Movahedi b &Naghmeh Sattarahmady ba Department of Chemistry, Faculty of Sciences, Shahid Beheshti University, Evin,19839-63113, Tehran, Iranb Institute of Biochemistry and Biophysics, University of Tehran, Tehran, IranVersion of record first published: 15 May 2012.

To cite this article: Rasoul Nasiri, Homayoon Bahrami, Mansour Zahedi, Ali Akbar Moosavi-Movahedi & Naghmeh Sattarahmady(2010): A Theoretical Elucidation of Glucose Interaction with HSA's Domains, Journal of Biomolecular Structure and Dynamics,28:2, 211-226

To link to this article: http://dx.doi.org/10.1080/07391102.2010.10507354

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form toanyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses shouldbe independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly inconnection with or arising out of the use of this material.

211

Journal of Biomolecular Structure & Dynamics, ISSN 0739-1102 Volume 28, Issue Number 2, (2010) ©Adenine Press (2010)

*Phone: 98-21-22431661Fax: 98-21-22431663E-mail: m-zahedi@sbu.ac.ir

Rasoul Nasiri1 Homayoon Bahrami1

Mansour Zahedi1*

Ali Akbar Moosavi-Movahedi2

Naghmeh Sattarahmady2

1Department of Chemistry, Faculty of

Sciences, Shahid Beheshti University,

Evin, 19839-63113, Tehran, Iran2Institute of Biochemistry and

Biophysics, University of Tehran,

Tehran, Iran

A Theoretical Elucidation of Glucose Interaction with HSA’s Domains

http://www.jbsdonline.com

Abstract

The interaction of different domains belonging to Human Serum Albumin (HSA) with open form of glucose have been investigated using molecular dynamics simulation methods. Applying docking, primary structures involving interaction of some residues with glucose have been obtained. Subsequently, equilibrium geometries at 300 K and minimum geometries have been determined for each of aforementioned structures by employing MD simulation and simulated annealing. The stability of species has been evaluated using a SAWSA v2.0 model. Ultimately, NBO analysis has been carried out to specify possible hydrogen bonding regarding the HSA interaction with glucose. Results obtained show that glucose can interact with Lys195, Lys199, and Glu153. In these interactions, each lysine forms an H-bonding with glucose. The H-bonding is obtained by stretching of N-H bond belonging to NH3

+ group of lysine along an oxygen atom of glucose. In addition, the above mentioned lysines are protonated, and there is an electrostatic interaction between glucose with Lys195 or Lys199. In addition, an H-bonding is formed between O atom of –COO group belonging to Glu153 and H atom of OH group belonging to glucose. Because, the N-H group of Lys195 interacts with the O atom of latter OH group, reaction of Lys195 is more desirable than that of Lys199. In fact, glucose is placed in the vicinity of Lys195 along with electrostatic interaction and H-bonding to Lys195 and Lys199 as well as H-bonding with Glu153, which subsequently reacts with Lys195. Thus, Lys195 is the primary site in reaction of glucose with HSA.

Key words: Human Serum Albumin; Glucose; Molecular Dynamics Simulation; Simulated annealing; NBO Analysis.

Introduction

Sugars have been utilized to maintain physiological and biochemical function and energy providing in biological systems. Dietary sugars comprise from mono- and di- saccharides, including sucrose, lactose, maltose, glucose, and fructose (1). Glu-cose and fructose are the main physiological carbon sources and are present in virtually all processed and unprocessed foods and glucose is the main sugar of serum blood. These two sugars are classified as glycaemic carbohydrates, which are readily absorbed and metabolized by various tissues, especially the liver (2). Consumption of glucose and fructose in processed foods have increased in food

Abbreviations: HSA - Human Serum Albumin; MD - simulation Molecular Dynamics simulation; MM - Molecular Mechanics; LGA - Lamarckian Genetic Algorithm; QM - Quantum Mechanics; B3LYP/6-31G* method - A quantum mechanics calculation level; SAWSA - atom-Weighted Solvent Accessible Surface Areas; Z-matrix - Perform the optimization in internal coordinates; HF/6-31G* - This keyword requests a Hartree-Fock calculation with 6-31G* basis set; CHELPG - Produce charges fit to the elec-trostatic potential at points selected according to the CHELPG scheme; DFT method - Density Function Theory; Annealing simulation - A technique for searching global minimum of a system by heating and cooling; NBO - Natural Bond Orbital.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

212

Nasiri et al.

industry and resulted in the development of obesity and other physiological and cellular consequences (e.g., diabetes mellitus, aging) (3).

A high concentration of reducing sugars produces chemical modifications in macromolecules, such as proteins, lipids, and nucleic acids. One highly studied chemical alteration is “non-enzymatic glycation”, which was first described by the French biochemist Louis Maillard at the beginning of the 20th century (4). Dur-ing this process, reducing sugars react non-enzymatically with free amino groups on target molecules and produce early glycation products that are reversible and named as Schiff base (an unstable aldimine). This process is followed by a com-plex series of reactions to form Amadori product (a stable ketoamine) and finally yields a class of heterogeneous chemical compounds collectively referred to as advanced glycation end products (AGE) (5, 6). Increased amounts of glycated proteins may induce long-term tissue damage, which is generally accompanied by pathogenic consequences of diabetes, including poor circulation to extremities, retinopathy, nephropathy and coronary artery disease (7). In principle, glycated proteins function differently from their non-glycated form. This has been sug-gested for human serum albumin (HSA), hemoglobin, and a series of intracellular enzyme proteins (8).

Glycation of albumin is of special interest because human serum albumin (HSA) has 58 lysines and 18 arginines residues making it a favorable target for glycation (9). It is a single-chain protein with 585 amino acid residues, three homologous domains (I-III), and a helical heart-shaped globule structure. It has one tryptophan residue (Trp214), 17 disulphide bridges, and one free thiol (Cys34) (10, 11). HSA is quantitatively the most important depot and transport protein in blood plasma. It is a major antioxidant with important roles in maintaining normal osmolarity of plasma and interstitial fluids (12). The identification of the sites of glycation, struc-tural and functional changes during glycation of HSA, and AGE formation in the presence of different carbohydrates have been the subject of recent in vitro studies (13-19). In addition, we have previously studied the structural and thermodynamic properties of glycated HSA (20-23).

It has been found that during the AGE-formation process, lysine and Arginine side chains are modified to a high degree and to a minor extent, respectively (24). Previ-ous studied report Lys199, lys281, lys439 and lys525 as the major glycation sites in HSA (25). A recent study (26) has proposed new sites for minimally glycated HSA using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). Based on the latter study (26), lysine residues in location of 12, 51, 195, 199, 205, 439 and 538 were found to be modified through the forma-tion of fructosyllysine, while the modification of Lys159 and Lys286 involved the formation of pyrraline or Nε-carboxymethyl-lysine respectively. Lysine 378 was found to give Nε-carboxyethyl-lysine in some forms of glycated HSA but fruc-tosyl-lysine in other forms. Residues Arg160 and Arg472 produced a modifica-tion based on Nε-(5-hydro-4-imidazolon-2-yl) ornithine. Arg222 was modified to produce argpyrimidine, Nε-[5-(2,3,4-trihydroxybutyl)-5-hydro-4-imidazolon-2-yl] ornithine or tetrahydropyrimidine. The difference among the main sites for glyca-tion of HSA is related to the fact that many of these glycation-related modifications are located at or near known drug binding sites on HSA (26).

The free energy of hydration (∆Gh) is often correlated with the stability of the protein (27) and it can be divided into non-polar and electrostatic free energies contributions. Non-polar free energy can be represented by the accessible surface area (ASA) (28). In addition, electrostatic free energy contributions are treated using the Poisson-Boltzmann (PB) equation, as previously studied, (29, 30). The combination of these two approximations forms the ASA/PB implicit solvent model that yields an approximation of the total free energy, ∆G = ∆G(nonpolar) + ∆G(elec).

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

213

MD Study of Glucose Interaction with HAS

Domains

A different implicit solvent model, also using ASA, is based on the assumption that the solvation free energy of a solute can be expressed in terms of a linear sum of the atomic contribution weighted by partial exposed surface area: ∆G(X) = ∑ Wi × Si(X). Here, Si(X) is the partial ASA of atom i, and Wi is an atomic free energy per unit area associated with atom i. Such model is referred to as full ASA. Being simple and inexpensive, this approach is widely used in computations on biomolecules (31, 32). Full ASA models have been employed to investigate thermal denaturation of proteins (33) and to examine protein-protein associations (34). In this study, a full ASA model called SAWSA v2.0 is employed to evaluate the stability of species in interaction of glucose with HSA. As stated elsewhere (35), the predicted values of aqueous solvation free energies from this model are in good agreement with those from the PB/SA model and are much better than (36-38) models developed for proteins. In SAWSA model, atoms are classified into 20 basic types for proteins which outnumber some ASA models in terms of atom types. Thus, the employed model benefits considerable flexibility in evaluation of hydration free energies. Based on above reported, we have used SAWSA v2.0 model in our calculation.

The HSA is named a multifunctional plasma carrier protein, because of its ability to bind to an unusually broad spectrum of ligands. These include inorganic cations, organic anions, various drugs, amino acids, and important hydrophobic molecules like bilirubin (39, 40).

It is obvious that glycosylation of HSA in diabetic patients must alter the affinity of many ligands. The affinity binding of long chain fatty acids is reduced considerably by one-twentieth (5) and aspirin by half (41). Thus, determination of the glucose binding sites on HSA is very important in diabetic patients.

This work attempts to determine the primary site of HSA upon reaction with open form of glucose using docking, molecular dynamics simulations and simulated annealing methods as well as employing a SAWSA model and NBO analysis. Sev-eral articles dealing with the use of many of these methodologies in a variety of systems are routinely and currently being published in the Journal of Biomolecular Structure and Dynamics (42-54). Initially, equilibrium geometry of HSA at 300 K has been obtained using MD simulation. Latter geometry has been used to specify primary structures for glucose-HSA interaction by employing glucose docking. Results achieved from docking treatment, has led to domain I-III dissection of HSA and provided structures involving interaction of each domain of HSA with glucose, separately. Equilibrium geometries at 300 K and minima geometries have been determined using molecular dynamics simulations and simulated annealing for each latter structure.

Stability of geometries obtained has been determined by evaluation of hydration free energy under a WASA model. Results obtained have shown that Lys195 is the primary reaction site of HSA with glucose. Hydrogen bonding has been ultimately specified in Lys195-HSA interaction.

Computational Scheme

Molecular dynamics simulation (MD) and molecular mechanic minimization (MM) in a condensed medium were performed using a GROMACS 3.3.3 package in a Gromos force field (G43A1) (55-59). In gromacs manual, G43a1 is recom-mended to study proteins. AutoDockTools (ADT), a graphical user interface for AutoDock4.0 (60, 61) was employed to obtain primary geometries, which involve glucose-HSA interaction. Grid maps used by the empirical free-energy scoring function in AutoDock were generated with the AutoGrid utility. A grid map of 60 − 60 − 60 grid points in size with a grid-point spacing of 0.375 Å was created for the protein. The map was centered on the glucose binding-site region, and the volume

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

214

Nasiri et al.

of the box was large enough to encompass the binding region of I-III domains of HSA. All bond rotations and torsions for the ligand and some residues in HSA were set in the AutoTools utility of AutoDock. A Lamarckian genetic algorithm (LGA) in AutoDock 4.0 software was used to generate ligand populations.

Gaussian 98 program revision A-9 (62) was employed for quantum mechanical calculation belonging to NBO analysis as well as charge calculation of glucose. Based on conclusion drawn from previous works (63-64) and due to the desired structures being of medium size, B3LYP/6-31G* method was used for all above calculations (65-67).

The MD simulations were carried out by particle mesh Ewald method (68) for the electrostatic interactions. The van der Waals cutoff was 14 Å. The integration time step was 1 fs, with the neighbor list being updated every fifth step by using the grid option and a cutoff distance of 12 Å. Periodic boundary condition has been used with constant number of particles in the systems, constant pressure, and constant temperature simulation criteria (NPT). In this simulation the systems were coupled to external constant temperature (100, 200, 300 K, τ = 0.1 ps) in three steps and external constant pressure bath (1 atm, τ = 0.5 ps). For all simulations, the water molecules were added as a simple point charges (SPC) model. Force field param-eters for the glucose were extracted from website belonging to the Dundee Prodrg Server (69). Also, geometry of glucose is not distorted during MD simulation and MM minimization. Thus, employed force field parameters of glucose are valid. Protonated state of various residues have been determined from the pKa calculation by PROPKA at PH = 7.4 (70).

Point charges for glucose was obtained from a HF/6-31G* single point calculation in Gaussian 98 using the ChelpG fitting procedure (71). It is worth mentioning that the ligand charges in docking procedure are RESP charges which are calculated automatically by AutoDock software. ESP Charges utilized for glucose in MD simulations and MM minimization are obtained by ChelpG method. Note that the limitation of this approach is that the polarization effect associated with the con-densed phase environment is not explicitly included, although the tendency for the HF/6-31G* QM level of theory to overestimate dipole moments has been suggested to account for this deficiency (72).

The numbering system of atoms in interaction of some residues with glucose, as shown in (Figures 2, 3 and 4) is used to specify results, which arise from NBO anal-ysis. In addition, as mentioned above, the SAWSA v2.0 model (atom-Weighted Solvent Accessible Surface Areas model) is also employed for evaluation of sta-bility of species in interaction of some residues with glucose (35). In fact, Wi (an atomic free energy per unit area associated with atom I) for each atom of HSA and glucose is extracted from SAWSA v2.0 model (35).

Results and Discussion

Previous experimental studies have indicated that the high affinity binding sites of interaction of glucose with HSA have been spread over HSA at its different domains (I-III), namely Lysines 12, 51, 159, 195, 199, 205, 286, 378, 439, 525, and 538 as well as Arginines 160, 222 and 472 (26). Based on such fact, the fol-lowing approach was considered. Initially, docking was used in order to achieve primary structures suitable for HSA-glucose interaction. Subsequently, molecular dynamics simulation was employed to obtain equilibrium geometries involving aforementioned interaction. In addition, Annealing simulation was further ben-efited to obtain near local structures for the desired interaction. Finally, minimum geometries of latter structures were achieved by employing optimization tech-niques, namely steepest descent and conjugate gradients. Stability of these mini-mum geometries was evaluated using a SAWSA v2.0 model. Ultimately, highest

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

215

MD Study of Glucose Interaction with HAS

Domains

affinity binding site of each domain was specified in glucose-HSA interaction. Besides, hydrogen bonding of highest stable selected geometries were evaluated using NBO analysis. Details of all calculations and results as well as discussions have been summarized as follow.

HSA Equilibrium Structure and Docking

Docking has been performed to achieve initial structure(s) for the glucose-HSA interaction. To have an HSA species, which mimics natural characters, a thorough equilibrium approach has been carried out as explained below. Thus, an initial equilibrium geometry of HSA at 300 K has been obtained using MD simulations, and subsequent docking has been performed.

The X-ray structure of HSA (code: 1AO6) was placed in a solvent box with about 42,350 SPC water molecules being equal to an 18 Å distance for ADA to the box edges. Neutralization of the system required addition of twelve Na+ ions. First, the solvent alone and any Na+ ions were subjected to energy minimization with the solute kept fixed in its initial configuration. The solvent and Na+ ions were then allowed to evolve using a molecular dynamics simulation for 20 ps with a step time of 1 fs, again keeping the structure of the solute molecule fixed. Next, the entire system was minimized using the steepest descent of 1000 steps followed by conjugate gradients of 9000 steps. In order to obtain equilibrium geometry at 300 K and 1 atm, the system was heated at a weak temperature (t = 0.1 ps) and pressure (t = 0.5 ps) coupling by taking advantage of the Berendesen algorithms (58). Heating time for molecular dynamics simulations at 100 K, 200 K and 300 K was 100 ps. All above simulations were performed at constant tem-perature and pressure with a non-bonded cutoff of 14 Å. A molecular dynamics simulation was further carried out for 700 ps at 300 K, followed by structural minimization calculation. The latter minimization was performed at the steepest descent of 1000 steps followed by conjugate gradients of 9000 steps. This way, a minimum geometry was obtained for HSA. It is noteworthy that the backbone root mean square deviation (RMSD) of whole HSA relative to its own starting structure was ~1.6Å, at the last 200 ps of simulations. This low RMSD value indicates that the MD run was stable and protein atoms did not significantly devi-ate from the starting temperatures during MD simulations (73, 74). In addition, average temperature of 700 ps of simulation at 300 K was equal to 300 ± 0.8 K. Therefore, the extracted equilibrium structure was obtained under temperature stability conditions.

By achieving the latter HSA structure in the presence of water molecules and ions, glucose docking was carried out using a standard AutoDock. Note that X-ray struc-ture of HSA is not used in docking process. In fact, latter geometry is simulated in a 1ns time of MD simulation (300 ps of heating time followed 700 ps in 300 K). Because, we only want refined X-ray geometry and eliminate probable bad contact between residues in the X-ray geometry. Thus, it is expected that flexibility of HSA will a more correspondence with experience, compared to X-ray geometry of HSA. As mentioned above, HSA is a single-chain protein with 585 amino acid residues. However, size of HSA limit us in employing of a long time simulation. As was mentioned in previous section, experimental studies have shown that sites in reaction of HSA with glucose are lysines 12, 51, 159, 195, 199, 205, 286, 378, 439, 525, and 538 as well as Arginines 160, 222 and 472 in I-III domains. Thus, afore-mentioned residues must be relaxed in docking process. Considering these results, some dockings were performed and in every docking process, some of mentioned residues and glucose were relaxed. Ultimately, fourteen structures were obtained, which involve glucose-HSA interaction. Latter structures were chosen by reference to experimental data and output of docking process, namely binding energies and statistical results. Note that, after elimination of solvent molecule and ions after simulation, glucose is docked to HSA.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

216

Nasiri et al.

Molecular Dynamics Simulation of Structures Involving Glucose-HSA Interaction

Based on above section, fourteen structures were obtained via docking processes. In each one of these structures, glucose is surrounded by some residues. As has been justified by others and in order to decrease the calculation cost, one of the known domains of HSA along with glucose were bisected from the mentioned structures each time (75, 76). Bisected domains include that part of HSA, which involves glucose-HSA interaction, and are as follows:

1- IA (Residues 1-105) and IB (Residues 120-201) which include Lys 12, 51, 159, 195, 199 and Arg 160

2- IIA (Residues 177-291) and IIB (Residues 316-395) which include Lys 205, 286, 378 and Arg 222

3- IIIA (Residues 367-491) and IIIB (Residues 512-585) which include Lys 439, 525, 538 and Arg 472

Ultimately, fourteen structures were chosen based on experimental results and docking outputs. It is noteworthy that each of latter structures includes glucose interacting with one of the above bisected HSA domains. Each of the fourteen starting structures was placed in a solvent box with ~20,000 SPC water mol-ecules again maintaining a distance of 18 Å to the box edges. In order for the water molecules to adopt themselves to the potential due to presence of HSA molecule, an equilibrium approach was carried out as explained in previous sec-tion in molecular dynamics simulation of HSA. Same approach as that explained for the whole HSA simulation was employed to obtain equilibrium geometries at 300 K and 1 atm. Thus all above fourteen systems were heated by a weak temperature coupling (t = 0.1 ps) and pressure (t = 0.5 ps) while heating time for molecular dynamics simulation at 100 K, 200 K and 300 K chosen to be 100 ps. Subsequently, molecular dynamics simulations at 300 K were performed for 700 ps, followed by structural minimization calculations. The minimization algorithms were the same as those given in section 3.1. This way, a minimum geometry was obtained for glucose while interacting with one of HSA domains. It is noteworthy that the backbone root mean square deviation (RMSD) of glucose-HSA domain structures relative to their own starting structures were ~1.5 Å after 200 ps of simulations at 300 K, as illustrated in (Figure 1) as a prototype. These low RMSD values indicate that the MD run were stable and protein atoms did not significantly deviate from the starting temperatures during MD simulations.

Figure 1: The backbone root mean-square deviation (RMSD) of protein structure relative to its own starting structure.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

217

MD Study of Glucose Interaction with HAS

Domains

Same temperature stability conditions were experienced in these simulations as that mentioned in section 3.1.

Conformational Analysis

To have a more chance of avoiding being trapped in a high energy minimum, conformational analysis on above fourteen minimum geometries is necessary. As stated in section 3.2, local geometries involving one of HSA domain’s interaction with glucose were obtained using MM minimization of equilibrium geometries in 300 K. In order to obtain more local minimized geometries for each species, conformational sampling on each equilibrium geometry at 300 K is necessary. To carry out such task, a known technique called simulated annealing method was employed. Starting structures are equilibrium geometries at 300 K, which were obtained in section 3.2. It is noteworthy that latter geometries are structures including each domain of HSA with glucose. Note that each of the mentioned structures was placed in a solvent box with about 20,000 SPC water molecules maintaining a distance of 18 Å of the box edges, and neutralization of the sys-tem required addition of Na+ ions. Heating treatment up to 550 K was initially performed on each system in 50 ps duration keeping all parameters the same as before. Annealing treatment on these systems was then carried out. The strategy being decrease from 550 K to 50 K followed by increase from 50 K to 550 K. Time duration for each temperature increment was 100 ps. This way carrying out a simulated annealing of 1900 ps resulted in 10 near local minima structures for each starting system. These near local minima can be extracted from 100, 300, 500 up to 1900 ps of annealing simulation time. Inspection of the annealed systems showed no indication of any bond breakage or tension in structures as expected. Thus using a temperature as high as 550 K did not have any effect in destroying aforementioned structures. Then, in order to obtain local minima structures, the annealed near local geometries were optimized by tacking advan-tage of steepest descent of 1000 steps followed by conjugate gradients of 9000 steps methods.

SAWSA v2.0 Model

The evaluation of the relative stability of the geometries including glucose-HSA interaction must take an entropic quantity into account. As has been reported, empirical SAWSA v2.0 model can be used in hydration free energy calculations with reasonable precision (35). In addition, these methods have a minor compu-tational cost in contrast to free energy perturbation techniques. In this study, the hydration free energy of the desired species was considered a criterion for their relative stability and was calculated using a SAWSA v2.0 model.

In this section, final MM minimized geometries obtained in section 3.2 and 3.3 have been considered. These geometries have been obtained from equilibrium geom-etries of MD simulation at 300 K (see section 3.2) and near local minima geom-etries of annealing simulation (see section 3.3). Thus, there are many minimized geometries in which glucose can interact with one or more residues. Hydration free energy of the minimized geometries was calculated as summarized in (Table I). As mentioned above, in this study, free energy of hydration is used to determine relative stability between geometries with various interaction of glucose. Note that this comparison is separately done for every domain of HAS, as it is evident from first row in the table I.

Namely, geometry involving II domain of HSA and glucose is not compared with that of III or II domain of HSA with glucose. Each column of table I belongs to the interaction of glucose with one of the I-III domains. Namely, 18 geometry for each the domains have been obtained which involve a glucose molecule, and Data of each column belongs to geometry of the one of domain of HSA with glucose.

Table IThe relative hydration free energy belonging to the some geometries which might arise from interac-tion of glucose with different domain of HSA.a

I∆(∆ Gh)kJmol–1

II∆(∆ Gh)kJmol–1

III∆(∆ Gh)kJmol–1

0 0 0103 46 116128 144 200152 162 229175 191 242189 220 263195 231 276220 265 289241 272 295257 279 307292 289 319317 312 331328 319 358415 323 370440 343 387458 372 453511 464 504527 544 585606 660 778

a The hydration free energies have been evaluated

relative to the most negative hydration free energy.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

218

Nasiri et al.

We have used a WASA model, because, it is reported that an ASA model has minor cost computation and an accuracy of about 2 kj.mol–1. We have not con-sidered effect of glucose interaction on HSA folding, thus WASA have been not used to evaluate folding changes due to the presence of glucose. Latter goal can be investigated, by simulation of HSA in a medium with the considerable number of glucose in a long time MD simulation. Thus, In this study we do never attempt to evaluate folding change of has due to the presence of glucose in medium. We will investigate latter subject in the future. Energy of each of minimized geometry relative to hydration free energy of a structure which has the most negative hydra-tion free energy is given in each column of the table. As mentioned above, data of every column in the table belong to structures involving the one of domain of HSA with a glucose molecule, and must be compared to each other. Nevertheless, data of two columns are for geometry including different domain of HSA, must be not compared. In fact, data of rows involve results of various domains in interaction of glucose with HSA, and are absolutely independent. Obtained results indicate that conformational analysis is necessary to obtain low energy minimum geometries for glucose-HSA interaction. This is because the most negative value of hydration free energy for minimum geometries belongs to the minimum obtained in simu-lated annealing.

Considering data of the table I, and minimum geometries obtained, it is clear that glucose is surrounded by Glu153, Lys195, and Lys199 as shown in (Figure 2), when glucose is docked in domain I of HSA. In addition, a closer look at the table I and final geometries reveals that in the geometry with highest affinity belonging to domain II of HSA- glucose interaction, glucose is near some residues, namely Asp340, Glu382, and Val381. Latter fact is also shown in (Figure 3). Besides, data

Distance After docking After 1 ns simulation

r O3-H35.694 1.84

r O3-N15.911 2.827

r O4-H43.091 1.879

Figure 2: Core interaction belonging to the most stable geometry involving domain I of HSA and glucose

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

219

MD Study of Glucose Interaction with HAS

Domains

of the same table while considering minimum geometries reveal that in the most stable structure involving domain III of HSA-glucose interaction, glucose is sur-rounded by Glu491, Lys538, and Ala539. By a closer look at the latter interaction, it becomes clear that glucose is in fact placed in the vicinity of Glu491, as shown in (Figure 4). Thus, glucose is surrounded by some residue in the most stable geom-etry involving glucose with I-III domains of HSA interaction. NBO analysis has been performed on above geometries in order to investigate the interaction types of glucose with its surrounding residues.

NBO Analysis

Natural bond orbital (NBO) analysis originated as a technique for studying hybridization and covalency effects in polyatomic wave functions. NBOs were conceived as a “chemists’ basis set” that would correspond closely to a picture

Distance After docking After 1 ns simulation

r O1-H1 16.917 1.661

r O2-C2 17.278 3.522

r O1-C1 16.647 3.158

r O3-O4 5.366 2.607

r O4-C1 4.525 3.302

r O6-H5 10.36 2.899

r C5-H5 9.47 3.535

Figure 3: Core interaction belonging to the most stable geometry involving domain II of HSA and glucose.D

ownl

oade

d by

[U

nive

rsity

of

Gro

ning

en]

at 1

0:40

01

Nov

embe

r 20

12

220

Nasiri et al.

Figure 4: Core interaction belonging to the most stable geometry involving domain III of HSA and glucose.

of localized bonds and lone pairs as basic units of molecular structure. NBO analysis was also employed to obtain second order interaction energies and inter-molecular donor-acceptor interactions, arising from the noncovalent terms (77). Latter interaction is used to present H-bonding and other strong forms of van der Waals complexation. Note that there is an electrostatic interaction between two atoms or groups including opposite charges, when donor-acceptor interaction is observed between them. In this section, NBO analysis is employed to evaluate probable H-bonding and other interactions among bonding or anti-bonding orbit-als in glucose-HSA interactions as follow. In a force field, H-bondings between atoms are evaluated based on some atomic distances and bond angles. To deter-mine if an H-bond exits, a geometrical criterion is used in pages of 162-163 in the Gromacs manual):

r ≤ r HB H

α ≤ α HBA = 60º B A

This section of macromolecule is individually considered and H-bondings are determined and the reminder of macromolecule is ignored. In NBO analysis in this text, interaction core of chosen geometries are dissected and then NBO analysis has been performed. In this analysis, No geometry optimization is done, and only it is trying to evaluate probable H-bonding and other interactions among bond-ing or anti-bonding orbital in interaction glucose with some residues belonging a domain of HSA (page 12 of text). Thus, effect of other residues and flexibility of protein resulting from MD simulation on a studied H –bonding is not destroyed, when NBO analysis is performed, because dissected section is not optimized, and only a powerful tool is used to evaluate interaction in these dissected sections. Residues dissected have been chosen as the reminders of residues have not effect or have minor effect on interaction studied in NBO analysis. Thus, ignorance of large section of a macromolecule is occurred, when H-bonding is investigated in even employing of a force field. But, NBO analysis give us a more accurate and quanti-tative determination in interactions in site binding between a ligand and a protein using a valid QM method (B3LYP method).

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

221

MD Study of Glucose Interaction with HAS

Domains

NBO Analysis Upon the Most Stable Geometry, which Includes Interaction of Glucose with Domain I of HSA

As shown in (Tables II A and B), it is obvious that charge transfer from an almost sp hybrid orbital and an almost p orbital of O1 of the Glu153 (Figure 2) have weakened O2-H1 and O3-H2 σ-bonds of glucose. Former fact is evident from ∆Eij

2 values of 19.13 and 16.97 kJmol1 in (Table II) for interaction of lone pairs Lp(1) and Lp(3) of O1 as donors with σ-anti bonding of O2 and H1 atoms (BD*(1) O2-H1) as acceptors, respectively. latter point is clear from ∆Eij

2 values of 7.32 and 31.38 kJmol1 in (Table II) for interaction of lone pairs Lp(1) and Lp(2) of O1 as donors with σ-anti bonding of O3 and H2 atoms (BD*(1) O3-H2) as accep-tors, respectively. In addition, N1-H3 σ-bond of Lys195 is weakened by charge transfer from a p orbital of O3 belonging to glucose. This fact is apparent from ∆Eij

2 value of 16.46 kJmol1 in (Table II) for interaction of lone pair Lp(2) of O3 as donor with σ-anti bonding of N1 and H3 atoms (BD*(1) N1-H3) as acceptors. Similar trends have been observed in weakening of N2-H4 σ-bond of Lys199 by charge transfer from sp4.54 and sp2.47 hybrid orbitals of O4 atom belonging to glucose (Figure 2) in above species. These effects are obvious from ∆Eij

2 values as 1.06 and 15.72 kJmol1 in (Table II) for interaction of O4 lone pairs (Lp(1) O4 and Lp(2) O4) as donors with σ-anti bonding of N2 and H4 atoms (BD*(1) N2-H4) as acceptors, respectively. Thus, Glu153, Lys195, and Lys199 show stronger interac-tions than other residues, when glucose interacts with aforementioned domain of HSA. Based on above discussions, COO group of Glu153 stretches H atom of OH group belonging to glucose. O atom of latter group interacts to an N-H group of Lys 195. Thus, reaction of Lys195 with glucose is more desirable than that of Lys199 due to the presence of Glu153.

NBO Analysis upon the Most Stable Geometry, which Includes Interaction of Glucose with Domain II of HSA

As shown in (Tables III A and B), it is obvious that charge transfer from a p orbital and from an almost sp hybrid orbital of acetate O6 belonging to the Val381

Table IIThe main results of NBO analysis belonging to the most stable geometry which may arise from inter-action of glucose with domain I of HSA.

A- Some donor-acceptor interactions

Donor Acceptor ∆Eij2 (kJmol–1)

LP (1) O1 BD*(1) O2-H1 19.13LP (1) O1 BD*(1) O3-H2 7.32LP (2) O1 BD*(1) O3-H2 31.38LP (3) O1 BD*(1) O2-H1 16.97LP (2) O3 BD*(1) N1-H3 16.46LP (2) O4 BD*(1) N2-H4 15.72LP (1) O4 BD*(1) N2-H4 1.06

B- Hybridization of some lone pairs and σ-anti bonding interactions

HybridizationOccupation

number

LP (1) O1 sp0.97 1.92LP (2) O1 p 1.85LP (3) O1 p 1.64LP (2) O3 p 1.95LP (1) O4 sp4.54 1.97LP (2) O4 sp2.47 1.94BD*(1) N1-H3 sp2.8 N-sH 0.05BD*(1) O2-H1 sp2.91

O-sH 0.08BD*(1) O3-H2 sp2.97O-sH 0.09BD*(1) N2-H4 sp2.86 N-sH 0.05

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

222

Nasiri et al.

(Figure 3) has weakened O5-H5 σ-bond of glucose. This fact is evident from ∆Eij

2 value of 23.09 and 11.68 kJmo1 in (Table III) for interaction of lone pairs Lp(2) and Lp(1) of O6 as donors with σ-anti bonding of O5 and H5 atoms (BD*(1) O5-H5) as acceptors, respectively. Also, aforementioned bond of glucose is weak-ened by charge transfer from C5-O6 σ-bond of Val381. Latter fact is evident from ∆Eij

2 value of 1.27 kJmol1 in (Table III) for interaction of BD(1) C5-O6 as donor with σ-anti bonding of O5 and H5 atoms (BD*(1) O5-H5) as acceptors. Similar trends have been observed in weakening of O3-H3 σ-bond of glucose by charge transfer from a sp0.77 hybrid orbital and two p orbitals of O4 atom of the Glu382 (Figure 3) in above species. These effects are obvious from ∆Eij

2 values as 14.79, 3.11 and 32.06 kJmol1 in (Table III) for interaction of O4 lone pairs (Lp(1) O4, Lp(2) O4 and Lp(3) O4) as donors with σ-anti bonding of O3 and H3 atoms (BD*(1) O3-H3) as acceptors, respectively. In addition, charge transfer from C3-O4 σ-bond of Glu382 weakens O3-H3 bond of glucose. Latter fact is evident from ∆Eij

2 value of 2.18 kJmol1 in (Table III) for interaction of BD(1) C3-O4 as donor with σ-anti bonding of O3 and H3 atoms (BD*(1) O3-H3) as acceptors. Another charge transfer is observed from a p orbital of O5 of glucose to C4-H4 of Glu382. This charge transfer is clear from ∆Eij

2 value of 1.81 kJmol1 in (Table III) for interaction of LP(2) O5 as donor with σ-anti bonding of C4 and H4 atoms (BD*(1) C4-H4) as acceptors. In addition, three charge transfers have occurred from two hybrid orbitals and a p orbital of oxygen atom belonging to Asp340 as donor with

Table IIIThe main results of NBO analysis belonging to the most stable geometry which may arise from interaction of glucose with domain II of HSA.

A- Some donor-acceptor interactions

Donor Acceptor ∆Eij2 (kJmol–1)

LP (1) O1 BD*(1) O2-H1 10.76LP (2) O1 BD*(1) O2-H1 44.38LP (3) O1 BD*(1) O2-H1 6.89LP (3) O1 BD*(1) C1-H2 2.04BD (1) C2-O1 BD*(1) O2-H1 2.09LP (1) O4 BD*(1) O3-H3 14.79LP (2) O4 BD*(1) O3-H3 3.11LP (3) O4 BD*(1) O3-H3 32.06BD (1) C3-O4 BD*(1) O3-H3 2.18LP (2) O5 BD*(1) C4-H4 1.81LP (1) O6 BD*(1) O5-H5 11.68LP (2) O6 BD*(1) O5-H5 23.09BD (1) C5-O6 BD*(1) O5-H5 1.27

B- Hybridization of some lone pairs and σ- anti bonding interactions

Hybridization Occupation number

LP (1) O1 sp1.31 1.94LP (2) O1 sp5.89 1.83LP (3) O1 p 1.64LP (1) O4 sp0.77 1.94LP (2) O4 p 1.89LP (3) O4 p 1.62LP (2) O5 p 1.95LP (1) O6 sp0.94 1.96LP (2) O6 p 1.88BD (1) C2-O1 sp2.61

C-sp1.37O 1.996

BD (1) C3-O4 sp2.13C-sp1.41

O 1.99BD (1) C5-O6 sp2.45

C-sp1.61O 1.99

BD* (1) O2-H1 sp2.54O-sH 0.13

BD* (1) C1-H2 sp2.61C-sH 0.02

BD* (1) C4-H4 sp2.66C-sH 0.02

BD* (1) O3-H3 sp2.73O-sH 0.10

BD* (1) O5-H5 sp3.11O-sH 0.07

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

223

MD Study of Glucose Interaction with HAS

Domains

σ-anti bonding of O2 and H1 atoms (BD*(1) O2-H1) of glucose, as acceptors. This fact is evident from ∆Eij

2 values of 10.76, 44.38 and 6.89 kJmol1 in (Table III) for interaction of lone pairs Lp(1), Lp(2) and Lp(3) of O1 with BD*(1) O2-H1 of glu-cose, respectively. By considering above discussion based on results summarized in (Table III), there are charge transfers between a lone pair (LP(3)) of aspartate’s O atom (O1) and σ-anti bonding of C1 and H2 atoms (BD*(1) C1-H2) of glucose. All above charge transfers are of n→σ* character, and since atoms such as O and H are involved, hydrogen bonding is plausible. Thus, H-bonding is a factor in glucose absorption by domain II of HSA and Asp340 and Glu382 have stronger H-bonding interaction with glucose than other residues in domain II.

NBO Analysis Upon the Most Stable Geometry, which Includes Interaction of Glucose with Domain III of HSA

As shown in (Tables IV A and B), it is obvious that charge transfer from two sp1.71 and sp3.63 hybrid orbitals and an almost p orbital O1 of COO group belonging to the Glu491 (Figure 4) have weakened O3-H1 and O2-H2 σ-bonds of glucose. Former fact is evident from ∆Eij

2 values of 3.56, 21.34, and 15.46 in (Table IV) for interaction of lone pairs Lp(1), Lp(2), and Lp(3) of O1 as donors with σ-anti bonding of O3 and H1 atoms (BD*(1) O3-H1) as acceptors, respectively. Also, latter fact is clear from ∆Eij

2 values of 8.55, 20.34, and 6.69 kJmol1 in (Table IV) for interaction of lone pairs Lp(1), Lp(2), and Lp(3) of O1 as donors with σ-anti bonding of O2 and H2 atoms (BD*(1) O2-H2) as acceptors, respectively. Besides, results obtained show that there is no considerable interaction between glucose and Lys538 as well as glucose and Als539. Thus, glucose can interact with domain III of HSA by formation of H-bonding with COO group of Glu491.

Based on experimental results, the primary site in reaction of glucose with HSA is the one with resiudes such as Lysines 12, 51, 159, 195, 199, 205, 439, 286, 378, 525, and 538 as well as Arginines 160, 472 and 222 in I-III domains. Results obtained show that Lysines 199 and 195 are those Lysine, which can interact with glucose by employing H-bonding, among aforementioned Lysines, when glucose is docked in HSA. Latter two Lysines are protonated, and electrostatic interaction is expected in interaction of glucose with Lys195 and Lys199. As mentioned above, N-H group of Lys195 interacts to an O atom of OH group belonging to glucose, while at the same time COO group of Glu153 stretches H atom of this OH group. Such finding reveals that reaction of Lys195 with glucose is more desirable than

Table IVThe main results of NBO analysis belonging to the most stable geometry, which may arise from inter-action of glucose with domain III of HSA.

A- Some donor-acceptor interactions

Donor Acceptor ∆Eij2 (kJmol–1)

LP (1) O1 BD* (1) O3-H1 3.56LP (1) O1 BD* (1) O2-H2 8.55LP (2) O1 BD* (1) O3-H1 21.34LP (2) O1 BD* (1) O2-H2 20.34LP (3) O1 BD* (1) O3-H1 15.46LP (3) O1 BD* (1) O2-H2 6.69

B- Hybridization of some lone pairs and σ- anti bonding interactions

Hybridization Occupation number

LP (1) O1 sp1.71 1.94LP (2) O1 sp3.63 1.84LP (3) O1 p 1.66BD* (1) O3-H1 sp4.01

C-sH 0.10BD* (1) O2-H2 sp3.55

C-sH 0.08

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

224

Nasiri et al.

that of Lys199. Ultimately, it can be concluded that glucose is placed in the vicin-ity of Lys195 by forming H-bonding with Glu153, Lys199 and Lys195 as well as having electrostatic interaction with Lys195 and Lys199, and reacts to Lys195 subsequently. Thus, Lys195 is the primary site for reaction of glucose with HSA. In previous works (Jacobsen, et al. 1977; Moosavi-Movahedi, et al. 2007), it is reported that LYS 195 can be the primary site in interaction of HSA with bilirubin. Above results show that after glycation of HSA the affinity binding of bilirubin must be reduced in diabetic patients. Thus, diabetic patients can be exposed to disease due to bilirubin increase in blood.

Conclusion

This work has been an attempt to determine the primary site in reaction of glucose with HSA using molecular dynamics simulations, B3LYP method and a SAWSA v.20 model. Equilibrium and minimum geometries have been obtained in interac-tion of glucose with I-III domains of HSA by employing MD simulation as well as simulated annealing. The relative stability of minimum geometries has been deter-mined using a SAWSA v2.0 model. Interaction core of the most stable structure in interaction of glucose with each of domains belonging to HSA, has been bisected, and B3LYP method has been employed to evaluate the H-bonding and other van der Waals forces in the bisected segments. Results show that Lysines 195 and 199 are primary species, which can interact with glucose among all Lys residues of HSA in 1:1 interaction of glucose with HSA. H-bonding forms between N-H group of Lysines and O atom of glucose as well as O atom of –COO group of Glu153 and an OH group of glucose. Former H-bonding is obtained by charge transfer of O atom of glucose to σ-anti bonding of N and H atoms of Lysines, namely by stretch-ing the N-H bond along the O atom. Latter H-bonding is formed via charge transfer of O atom of Glu153 to σ-anti bonding of O and H atoms of glucose, namely by stretching of the O-H bond of glucose along the O atom of Glu153. Because, afore-mentioned lysines are protonated, an electrostatic interaction is expected in interac-tion of glucose with Lys199 and Lys195. Note that N-H group of Lys195 interacts to an O atom of OH group of glucose while simultaneously Glu153 stretches H atom of latter OH group. This fact shows that reaction of Lys195 is further facili-tated compare to that of Lys199. Thus, glucose is placed in the vicinity of Lys195 by formation of H-bonding with Lys195, Lys199 and Glu153 as well as having electrostatic interaction with both lysines and then reacts to Lys195.

Acknowledgements

We are grateful to Professor Seik Weng Ng for making us available his soft-ware (G98W) and hardware (machine time) facilities. The financial support of Research Council of Shahid Beheshti University, the Research council of the University of Tehran as well as Iran National Science Foundation (INSF) is grate-fully acknowledged.

References

G. H. Anderson and D. Woodend. 1. Am J Clin Nutr 78, 843S-849S (2003).N. G. Asp. 2. Am J Clin Nutr 59, 679S-681S (1994).M. Sigman-Grant and J. Morita. 3. Am J Clin Nutr 78, 815S-826S (2003).R. G. Khalifah, J. W. Baynes, and B. G. Hudson. 4. Biochem Biophys Res Commun 257, 251-258 (1999).N. Shaklai, R. L. Garlick, and H. Franklin Bunn5. . J Biol Chem 259, 3812-3817 (1984).P. A. Voziyan, R. G. Khalifah, C. Thibaudeau, A. Yildiz, J. Jacob, A. S. Serianni, and 6. B. G. Hudson. J Biol Chem 278, 46616-46624 (2003).M. Brownlee. 7. Annu Rev Med 42, 159-166 (1991).B. Trueb, C. G. Holenstein, R. W. Fischer, and K. H. Winterhalter. 8. J Biol Chem 255, 6717-6720 (1980).N. Iberg and R. Fluckiger. 9. J Biol Chem 261, 13542-13545 (1986).X. M. He and D. C. Carter. 10. Nature 358, 209- 215 (1992).

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

225

MD Study of Glucose Interaction with HAS

Domains

D. C. Carter, X. M. He, S. H. Munson, P. D. Twigg, K. M. Gernert, M. B. Broom, and 11. T. Y. Miller. Science 244, 1195-1198 (1989).E. Bourdon, N. Loreau and D. Blache. 12. FASEB J 13, 233-243 (1999).P. J. Coussons, J. Jacoby, A. McKay, S. M. Kelly, N. C. Price, and J. V. Hunt. 13. Free Radical Biol Med 22, 1217-1227 (1997).S. D. Sharma, B. N. Pandey, K. P. Mishra, and S. Sivakami. 14. J Biochem Mol BiolBiophys 6, 233-242 (2002).D. M. Mendez, R. A. Jensen, L. A. McElroy, J. M. Pena, and R. M. Esquerra. 15. Arch Biochem Biophys 444, 92-99 (2005).H. Zoellner, Y. H. Hou, T. Hochgrebe, A. Poljak, M. W. Duncan, J. Golding, T. Henderson, 16. and G. Lynch. Biochem Biophys Res Commun 284, 83-89 (2001).P. A. Voziyan, R. G. Khalifah, C. Thibaudeau, A. Yildiz, J. Jacob, A. S. Serianni, and 17. B. G. Hudson. J Biol Chem 278, 46616-46624 (2003).M. E. Westwood and P. J. Thornalley18. . J Protein Chem 4, 359-372 (1995).K. M. Biemel, D. A. Friedl and M. O. Lederer. 19. J Biol Chem 277, 24907-24915 (2002).N. Sattarahmady, A. A. Moosavi-Movahedi, F. Ahmad, G. H. Hakimelahi, M. Habibi-Rezaei, 20. A. A. Saboury and N. Sheibani. Biochim Biophys Acta 1770, 933-942 (2007).A, Barzegar, A. A. Moosavi-Movahedi, N. Sattarahmady, M. A. Hosseinpour-Faizi, 21. M. Aminbakhsh, F. Ahmad, A. A. Saboury, M. R. Ganjali, and P. Norouzi. Protein Pept Lett 14, 13-18 (2007).N. Sattarahmady, A. A. Moosavi-Movahedi, M. Habibi-Rezaei, S. Ahmadian, A. A. 22. Saboury, H. Heli, and N. Sheibani. Carbohyd Res 343, 2229-2234 (2008).N. Sattarahmady, F. Khodagholi, A. A. Moosavi-Movahedi, H. Heli, and G. H. Hakimelahi. 23. Int J Biol Macromol 41, 180-184 (2007).A. Schmitt, J. Schmitt, G. Münch, and J. Gasic-Milencovic24. . Anal Biochem 338, 201-215 (2005).N. Iberg and R. Fluckiger. 25. J Biol Chem 261, 13542-5 (1986).C. Wa, R. L. Cerny, W. A. Clarke, and D. S. Hage. 26. Clin Chim Acta 385, 48-60 (2007).Z. He and Z. Zhang. 27. J Protein Chem 18, 557-564 (1999).B. Lee and F. M. Richards. 28. J Mol Biol 55, 379-400 (1971).J. G. Kirkwood. 29. J Chem Phys 2, 351-61 (1934).P. Debye and E. Hückel. 30. Physik Zeitschr 24, 305-325 (1924).F. Fraternali and W. F. Van Gunsteren. 31. J Mol Biol 256, 939-948 (1996).D. Eisenberg and A. Mcclachlan. 32. Nature 319, 199-203 (1986).T. Lazaridis and M. Karplus. 33. Science 278, 1928-1931 (1997).M. D. Cumming, T. N. Hart, and R. J. Read. 34. Protein Sci 4, 2087-2089 (1995).T. Hou, W. Zhang, Q. Huang, and X. Xu. 35. J Mol Model 11, 26-40 (2005).T. Ooi, M. Oobatake, G. Nemethy, and H. A. Scheraga. 36. Proc Natl Acad Sci (USA) 84, 3086-3090 (1987).L. Wesson and D. Eisenberg. 37. Protein Sci 1, 227-235 (1992).J. Vila, R. L. Williams, M. Vazquez, and H. A. Scheraga. 38. Proteins 10, 199−218 (1991).U. Kragh-Hansen. 39. Pharmacol Rev 33, 17-53 (1981).J. R. Brown and P. Shockley40. . In “Lipid-Protein Interactions” (P. Jost and O. H. Griffith, eds.), 1, 25-68 (1982).J. F. Day, S. R. Thorpe, and J. W. Bayness. 41. J Biol Chem 254, 595-597 (1979).G. H. Braun, D. M. M. Jorge, H. P. Ramos, R. M. Alves, V. B. da Silva, S. Giuliatti, 42. S. V. Sampaio, C. A. Taft, Carlos, and H. T. P. Silva. J Biomol Struct Dyn 25, 347-355 (2008).J. Yoon, J. Park, S. Jang, K. Lee, and S. Shin43. . J Biomol Struct Dyn 25, 505-515 (2008).A. Cordomí, E. Ramon, P. Garriga, and J. J. Perez. 44. J Biomol Struct Dyn 25, 573-587 (2008).C. Y-C. Chen, Y-F. Chen, C-H Wu and H-Y. Tsai. 45. J Biomol Struct Dyn 26, 57-64 (2008).J-H. Zhao, C-T. Yang, J. W. Wu, W-B. Tsai, H-Y Lin, H-W. Fang, Y. Ho, and H-L. Liu. 46. J Biomol Struct Dyn 26, 65-73 (2008).B. N. Macchion, R. Strömberg, and L. Nilsson. 47. J Biomol Struct Dyn 26, 163-173 (2008).U. B. Sonavane, S. K. Ramadugu, and R. R. Joshi. 48. J Biomol Struct Dyn 26, 203-214 (2008).F. Mehrnejad, and N. Chaparzadeh. 49. J Biomol Struct Dyn 26, 255-262 (2008).J. Yoon, S. Jang, K. Lee, and S. Shin. 50. J Biomol Struct Dyn 27, 259-269 (2009).D. Ilda, C. Giovanni, and D. Alessandro. 51. J Biomol Struct Dyn 27, 307-317 (2009).A. Cordomí, and J. J. Perez. 52. J Biomol Struct Dyn 27, 127-147 (2009).H. R. Bairagya, B. P. Mukhopadhyay, and K. Sekar. 53. J Biomol Struct Dyn 27, 149-158 (2009).J. Zhang. 54. J Biomol Struct Dyn 27, 159-162 (2009).D. Van Der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark, and H. J. C. Berendsen. 55. J Comp Chem 26, 1701-1718 (2005).H. Bekker, H. J. C. Berendsen, E. J. Dijkstra, S. Achterop, R. van Drunen, D. van der Spoel, 56. A. Sijbers, H. Keegstra, B. Reitsma, M. K. R. Renardus, R. A. de Groot, and J. Nadrchal. GROMACS: A parallel computer for molecular dynamics simulations, In Physics Comput-ing 92. Singapore World Scientific (1993).E. Lindahl, B. Hess and D. van der Spoel. 57. J Mol Mod 7, 306-317 (2001).H. J. C. Berendsen, D. van der Spoel and R. van Drunen. 58. Comp Phys Comm 91, 43-56 (1995).

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012

226

Nasiri et al.

W. F. van Gunsteren, S. R. Billeter, A. A. Eising, P. A. Hünenberger, P. Krüger, A. E. 59. Mark, W. R. P. Scott, and I. G. Tironi. Biomolecular Simulation, the GROMOS96 Manual and User Guide, Hochschulverlag AG der ETH, Zürich (1996).R. Huey, G. M. Morris, A. J. Olson, and D.S. Goodsell. 60. J Comp Chem 28, 1145-1152 (2007). G. M Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K. Belew, and 61. A. J. Olson. J Comp Chem 19, 1639-1662 (1998).M. J. Frisch, 62. et al. GAUSSIAN 98, Version A.9, Gaussian Inc, Pittsburgh, PA. (2000).M. Zahedi, H. Bahrami, Sh. Shabazian, N. Safari and S.W. Ng. 63. J Mol Struct (THEOCHEM) 633, 21-23 (2003).M. P. Gleeson, N. A. Burto, and I. H. Hillier64. . Chem Comm 2180-2181 (2003).A. D. Becke65. . J Chem Phys 98, 5648-5652 (1993).G. A. Peterson, and M. A. Al-Laham. 66. J Chem Phys 94, 6081-6090 (1991).W. Andreoni, A. Curioni, and T. Mordasini. 67. IBM J Res and Dev 45, 397-406 (2001).D. M. York, A. L. Wlodawer, G. Pedersen, and T. A. Darden. 68. Proc Natl Acad Sci USA 91, 8715- 8718 (1994).A. W. Schuettelkopf, and D. M. F. van Aalten. Acta Crystallogr. 69. Sect D-Biol Crystallogr 60, 1355-1363 (2004).H. Li, A. D. Roberton, and J. H. Jensen. 70. Proteins 61, 704-721 (2005).C. M. Breneman, and K. B. Wiberg. 71. J Comp Chem 11, 361-373 (1990).O. M. Becker. 72. Computational Biochemistry and Biophysic. 1st edition, Marcel Dekker. ISBN: 082470455X. (2001).N. Mogharrab, H. Ghourchian, and M. Amininasab73. . Biophy J 92, 1192-1203 (2007).W. Wriggers, E. Mehler, F. Pitici, H. Weinstein, and K. Schulten74. . Biophy J 74, 1622-1639 (1998).Z. Moosavi-Movahedi, H. Bahrami, M. Zahedi, K. Mahnam, J. Chamani, S. Safarian, 75. A. A. Saboury, and A. A. Moosavi-Movahedi. Biophy Chem 125, 375-387 (2007).N. Diaz, D. Suarez, T. L. Sordo, and K. M. Merz. 76. J Med Chem 44, 250-260 (2001).A. E. Reed, L. A. Curtiss, and F. Wienhold. 77. Chem Rev 88, 899-926 (1998).

Date Received: September 19, 2009

Communicated by the Editor Thomas Cheatham

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

10:

40 0

1 N

ovem

ber

2012