replica-exchange molecular dynamics simulation of small peptide in water and ethanol

8/3/2019 Replica-Exchange Molecular Dynamics Simulation of Small Peptide in Water and Ethanol

http://slidepdf.com/reader/full/replica-exchange-molecular-dynamics-simulation-of-small-peptide-in-water-and 1/5

Replica-exchange molecular dynamics simulation of small peptide in water and in ethanol

Koji Yoshida a, Toshio Yamaguchi a,*, Yuko Okamoto b,c,1

a Department of Chemistry, Faculty of Science, Fukuoka University, Nanakuma, Jonan-ku, Fukuoka 814-0180, Japanb Department of Theoretical Studies, Institute for Molecular Science, Okazaki, Aichi 444-8585, Japan

c Department of Functional Molecular Science, The Graduate University for Advanced Studies, Okazaki, Aichi 444-8585, Japan

Received 13 May 2005; in final form 22 June 2005

Available online 25 July 2005

Abstract

Replica-exchange molecular dynamics simulations have been performed on a 10-residue peptide in ethanol as well as in water

that were treated explicitly with 32 replicas in the range of 298–600 K. It has been found that the peptide tends to form compact

structures in ethanol whereas it is extended in water. The numbers of intramolecular hydrogen bonds and turn structures are both

larger in ethanol than in water, and this difference enhances the tendency of forming secondary structures in ethanol.

Ó 2005 Elsevier B.V. All rights reserved.

1. Introduction

Alcohol-induced denaturation of peptides and pro-

teins that enhances the a-helical structure has widely

been used in protein technology [1]. Solvent effects on

denaturation of proteins have often been discussed in

terms of electrostatic interactions by treating medium

as a dielectric continuum in biophysics and biochemistry

fields, but the underlying mechanism has yet to be clar-

ified at the microscopic level [2]. Kinoshita et al. [3] ap-

plied the reference interaction site model (RISM) theory

to calculate the thermodynamic properties of various

conformations of Met-enkephalin and C-peptide frag-

ment of ribonuclease A in methanol, ethanol, and water.Their results showed that alcohols facilitate the peptide

molecules to form the secondary structures with intra-

molecular hydrogen bonds such as the a-helix. They

concluded that the free energy of solvation in alcohols

becomes less dependent on the conformational change

than in water because of less solvation number of pep-tide in alcohols than that in water and that the confor-

mational stability in alcohols is governed mostly by

the conformational energy [3]. However, these results

are based on calculations of solvation energy of a few

fixed solute structures [3]. They selected a few typical

structures of the peptide (native, extended, in vacuum,

and others). Because the solute molecules in water and

in ethanol are always changing their structures, the sol-

ute and solvent molecules should be treated on an equal

basis to discuss the solvent effects more exactly.

Recently, we have investigated liquid structures in

tert-butanol (TBA)–water mixtures without peptideswith a RISM theory [4]. The calculated radial distribu-

tion functions showed that hydrogen-bond formation

between hydrophilic pairs was enhanced as the concen-

tration of large hydrophobic TBA increases (here water

was treated as a hydrophilic molecule). It is thus ex-

pected that hydrophilic interactions within peptides are

more enhanced in ethanol than in water [4].

In simulations of protein folding, a generalized-

ensemble algorithm has been proven effective because

0009-2614/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.cplett.2005.06.114

* Corresponding author. Fax: +81 92 865 6030.

E-mail address: [email protected] (T. Yamaguchi).1 Present address: Department of Physics, Graduate School of

Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602,

Japan.

www.elsevier.com/locate/cplett

Chemical Physics Letters 412 (2005) 280–284

mailto:[email protected]

mailto:[email protected]



it overcomes the common problem in protein simula-

tions that they can easily get trapped in one of a huge

number of local-minimum-energy states (for a review,

see [5]). In the present study, one of the generalized-

ensemble algorithms called the replica-exchange molecu-

lar dynamics (REMD) [6] was performed with a 10-res-

idue model peptide in water and in ethanol that weretreated explicitly as individual molecules in order to

compare solvent effects on the peptide conformations.

We have indeed observed quite significant differences

in the behaviors of the peptide in these solvents.

2. Methods

A model peptide of 10 residues was taken from the a-

helix part of 28-residue designed peptide with b – b – a

motif whose 3D structure have been determined by 2D

NMR [7]. The amino-acid sequence is GLUÀ-LEU-

ARG+-ASPÀ-PHE-ILE-GLUÀ-LYS+-PHE-LYS+. The

N-terminus and the C-terminus were taken to be the

usual zwitter ionic groups (NHþ3 and COOÀ, respec-

tively). REMD [6] was incorporated in the MD package

program DL _ POLY [8]. The utilities of DL _ PROTEIN [9],

which is a version of DL _ POLY specified for larger and

complex biomolecular simulations, were used to con-

struct the input files for the model peptide. The force-

field parameters were taken from CHARMM22 [10,11].

For water and ethanol, the TIP3P model [12] and the

OPLS united atom model [13] were used, respectively.

Periodic boundary conditions were imposed. The unit

cell sizes of water and ethanol systems were 31.08 and41.03 A, respectively, which were determined from the

densities of the corresponding bulk solvents at 298 K

and 0.1 MPa. The volume of the cell was fixed and the

temperature was controlled by the Berendsen method

(NVT ensemble). The potential cut-off distance for the

Lennard-Jones interactions was 10 A. The electrostatic

interaction was calculated by the Ewald method. The

structure of the model peptide was taken from the cor-

responding part of the PDB file (PDB code: 1FSD)

and solvated in water. The solvent molecules within

1.5 A from the atoms of the peptide were removed.

The numbers of water and ethanol molecules finally

adopted were 910 and 702, respectively. The time step

was set to 1.0 fs. A canonical MD simulation of this sys-

tem was performed at 600 K for 100 ps (i.e., 105 MD

steps), from which we randomly extracted an extended

structure of the peptide without any secondary struc-

tures. This structure was used as the initial conforma-

tion for all the replicas in both solvent conditions.

Replicas at 32 temperatures from 298 to 600 K were

used for the REMD simulations. Their values were

298, 305, 311, 316, 323, 330, 337, 345, 353, 361, 369,

378, 387, 396, 405, 414, 424, 433, 443, 454, 464, 475,

486, 498, 510, 523, 535, 547, 560, 573, 587, and 600 K.

After an equilibration canonical MD run at each tem-

perature for 10 ps, a production REMD simulation of

106 time steps (1.0 ns) for each replica was finally per-

formed. Replica exchange was tried every 100 MD steps

(0.1 ps). The data from trajectories were stored every

1000 MD steps (1 ps) for later analysis.

3. Results and discussion

3.1. Results of REMD simulations

We first examine whether the REMD simulations

were performed properly. Each replica of the REMD

simulations should realize a random walk in the temper-

ature space, which in turn induces a random walk in the

potential energy space. We depict in Fig. 1 the time

series of temperature and potential energy for one of

the replicas during the REMD simulation in ethanol.

We observe a random walk between the lowest temper-

ature (298 K) and the highest temperature (600 K) in

Fig. 1a. Likewise, we do observe in Fig. 1b a random

walk in the potential energy space, which covers a range

as large as %2300 kcal molÀ1. Note that there is a strong

Fig. 1. Time series of: (a) temperature and (b) total potential energy in

ethanol for one of the replicas.

K. Yoshida et al. / Chemical Physics Letters 412 (2005) 280–284 281



correlation between the two random walks as there

should be. All the other replicas exhibit essentially thesame behavior. To be more quantitative, it is required

for an optimal performance of REMD simulations that

the acceptance ratios of the replica exchange should be

uniform and sufficiently large (say, >10%). The accep-

tance ratios of the replica exchange from the REMD

simulation in water range from 12% to 21%, while those

from the REMD simulation in ethanol range from 11%

to 23%. Therefore, we conclude that the present REMDsimulations performed properly.

3.2. The radii of gyration of the peptide in water and

ethanol

The radii of gyration of the model peptide at different

temperatures in water and ethanol are shown in Fig. 2.

The radius of gyration, Rg, is defined here by

R2g ¼

1

N

X N

i¼1

ðri À rÞ2; ð1Þ

where N , ri , and r ð¼

1 N P N

i¼1riÞ are the total number of the atoms in the peptide, the coordinate vector of atom

Fig. 2. Radii of gyration of the peptide in water (open circles) and in

ethanol (filled circles) at 32 temperatures of the REMD simulations.

Fig. 3. Snapshots of the peptide: (a) in water and (b,c) in ethanol. Hydrogen atoms in the peptide and solvent molecules are suppressed for clarity.

Red and blue parts in (c) indicate a-helix and turn structures, respectively, depicted by the RASMOL program.

282 K. Yoshida et al. / Chemical Physics Letters 412 (2005) 280–284



i , and the center of geometry of the peptide, respectively.

The Rg values in water are larger than those in ethanol

at all the temperatures. The results indicate that the pep-

tide tends to take more compact structures in ethanol

than in water. Moreover, the Rg values in water are al-

most constant throughout the temperatures, whereas

those in ethanol tend to decrease with increasing temper-ature. This trend suggests that the electrostatic interac-

tions within the peptide are strengthened in ethanol

than in water and that these interactions are further

strengthened in ethanol as the temperature is raised.

Note that 6 out of 10 residues of this peptide are

charged. This is why the above effects are outstanding.

3.3. Secondary structure of the peptide in water and in

ethanol

Snapshots from the simulations in these solvents are

shown in Fig. 3. For the simulation in water (Fig. 3a),

we see that the side chains of all the charged residues

are pointed away from the peptide backbone and hy-

drated with bulk water. Therefore, the backbone struc-

ture is extended. For the simulation in ethanol

(Fig. 3b), on the other hand, we see that all the side

chains of charged residues (GLU-1À, ARG-3+, ASP-

4À, GLU-7À, LYS-8+, and LYS-10+) are electrostati-

cally attracted to the other parts of the peptide. The

backbone conformation is thus compact. These results

support the above claim that intrachain electrostatic

interactions are more strengthened in ethanol than in

water. These intrachain interactions enhance second-

ary-structure formation more in ethanol than in water.While no secondary structures were seen at all in water

during the REMD simulation, we did observe forma-

tions of an a-helix structure in ethanol as shown in

Fig. 3c, though the a-helix appeared less than 1% of

the time in ethanol during the REMD simulation at

all temperatures.

Fig. 4 shows the number of turn structures and the

number of intramolecular hydrogen bonds of the pep-

tide in water and in ethanol. Both numbers are larger

in ethanol than in water. These findings suggest that

the intramolecular interactions, such as the intramolec-

ular hydrogen bonds, in the peptide are dominant in

forming the structures of the peptide in ethanol, in com-

parison with those in water. A previous RISM calcula-

tion of TBA–water mixtures has shown that the

probability of hydrogen-bond formation of water–water

pairs increases as the concentration of TBA increases [4];

there, we considered water as hydrophilic solute mole-

cules. In water, the hydrophilic groups of the peptide

are exposed to bulk water to form hydrogen bonds with

water molecules. In ethanol, on the other hand, the sol-

vent molecule has a large hydrophobic part as well as a

hydrophilic part. Under this condition, the hydrophilic

groups of the peptide tend to combine with each other,

resulting in a compact structure via the intramolecular

hydrogen bonds.

4. Conclusions

We have compared the solvent effects of pure water

and pure ethanol on the conformations of a small pep-

tide. We have found that the peptide tends to form a

compact structure in ethanol, while it is extended in

water. This is ascribed to the strengthening of intramo-

lecular electrostatic interactions in the peptide in etha-

nol. Although definite secondary-structure formations

are rarely observed in both solvent conditions, intra-

chain hydrogen-bond formations are more enhanced in

ethanol than in water. We interpret this finding as a con-

sequence of strong side-chain electrostatic interactions,

because as many as 6 out of 10 residues are charged.

These interactions actually hinder the formation of

backbone secondary structures. We expect clear

enhancement in secondary-structure formations, such

as a-helices, in ethanol than in water when we replace

Fig. 4. Numbers of: (a) intramolecular hydrogen bonds and (b) the

turn structures of the peptide in water (open circles) and in ethanol

(filled circles) at 32 temperatures of the REMD simulations.

K. Yoshida et al. / Chemical Physics Letters 412 (2005) 280–284 283



some of the charged residues by a-helix-forming neutral

ones.

Acknowledgments

The present work was supported by the Joint Study

Program (2002–2003) of the Institute for Molecular Sci-ence and Grants-in-Aid for Young Scientists (No.

13740335, K.Y.), for Scientific Research in Priority

Areas, ÔWater and BiomoleculesÕ, and for the NAREGI

Nanoscience Project from the Ministry of Education,

Culture, Sports, Science and Technology of Japan.

References

[1] M. Buck, Q. Rev. Biophys. 31 (1998) 297.

[2] D.P. Hong, M. Hoshino, R. Kuboi, Y. Goto, J. Am. Chem. Soc.

121 (1999) 8427.

[3] M. Kinoshita, Y. Okamoto, F. Hirata, J. Am. Chem. Soc. 122

(2000) 2773.

[4] K. Yoshida, T. Yamaguchi, A. Kovalenko, F. Hirata, J. Phys.

Chem. B 106 (2002) 5042.

[5] A. Mitsutake, Y. Sugita, Y. Okamoto, Biopolymers (Pept. Sci.)

60 (2001) 96.

[6] Y. Sugita, Y. Okamoto, Chem. Phys. Lett. 314 (1999) 141.

[7] B.I. Dahiyat, S.L. Mayo, Science 278 (1997) 82.

[8] W. Smith, T.R. Forrester, DL _ POLY, Daresbury Laboratory at

Daresbury, Nr. Warrington, UK, 1996.

[9] S. Melchionna, S. Cozzini, DL _ PROTEIN, The Istituto Nazionale di

Fisica della Materia, 1997.

[10] B.R. Brooks, R.E. Bruccoleri, B.D. Olafson, D.J. States,

S. Swaminathan, M. Karplus, J. Comp. Chem. 4 (1983)

187.

[11] A.D. MacKerell Jr., B. Brooks, C.L. Brooks III, L. Nilsson, B.

Roux, Y. Won, M. Karplus, CHARMM, in: P.v.R. Schleyer et al.

(Eds.), The Encyclopedia of Computational Chemistry, vol. 1,

Wiley, Chichester, 1998, p. 271.

[12] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey,

M.L. Klein, J. Chem. Phys. 79 (1983) 926.

[13] W.L. Jorgensen, J. Phys. Chem. 90 (1986) 1276.

284 K. Yoshida et al. / Chemical Physics Letters 412 (2005) 280–284

replica-exchange molecular dynamics simulation of small peptide in water and ethanol

Documents