DOI: 10.1021/la9006058 8111Langmuir 2009, 25(14), 8111–8116 Published on Web 04/27/2009

pubs.acs.org/Langmuir

© 2009 American Chemical Society

Aggregation of Amyloidogenic Peptides near Hydrophobic and HydrophilicSurfaces

Ivan Brovchenko,* Gurpreet Singh, and Roland Winter

Physical Chemistry, Dortmund University of Technology, Otto-Hahn-Strasse 6, 44221 Dortmund, Germany

Received February 18, 2009. Revised Manuscript Received March 31, 2009

The general effect of surface hydrophobicity/hydrophilicity on the aggregation of peptides is studied by simulationsof oversaturated aqueous solutions of hydrophobic and hydrophilic amyloidogenic peptides. Peptide aggregation wasstudied in bulk solution, in solutions confined between hydrophobic boundaries (smooth planar paraffin-like surfacesand liquid-vapor interfaces) and in solutions confined between hydrophilic surfaces (smooth planar silica-likesurfaces). Aggregation of hydrophobic peptides strongly enhances due to the confinement between hydrophobicsurfaces with all peptides adsorbed at the boundaries and aligned predominantly parallel to them. In the other threecases considered, the peptides are repelled from the walls and do not reveal orientational ordering with respect to thesurface. The degree of peptide aggregation in these cases is only slightly affected by the confinement (it is enhanced forhydrophobic peptides and decreased for hydrophilic peptides). Our results show that even a single environmental factorsuch as water-mediated peptide-surface interaction has a drastic effect on the degree and character of peptideaggregation. A wide diversity of possible scenarios can be expected when specific peptide-surface interactions areadditionally taken into account.

Introduction

In recent years, it has become evident that misfolded proteinsand peptides are of enormous biophysical, medical and pharma-ceutical importance. On the one hand, they can aggregate intofully ordered amyloid fibrils consisting of cross β-sheet structures.These fibrils and their precursor oligomeric structures have beenstudied extensively in the last years due to their central role indiseases such as Alzheimer’s, Parkinson’s, and type II diabetesmellitus.1 Remarkably, nonpathogenic amyloid has been found ina diverse range of organisms: frombacteria tomammals aswell. Itis therefore particularly important to understand the underlyingphysical and chemical principles which trigger, control or preventamyloid formation.

The main regularities of the physical aggregation of interactingobjects are described by the laws of statistical physics. The phasetransition is an important guiding line in the description ofphysical aggregation. With approaching the phase transition(by varying the concentration, temperature, etc.), the aggregationof theminor component is facilitated and, at the phase transition,an infinite (macroscopic) aggregate is formed. In the case ofaqueous solutions, the aggregation (clustering) of both water andsolute molecules depends on the proximity to the demixingtransition, where separation of the aqueous solution into awater-rich and an organic-rich (aggregate) phase takes place.For example, the demixing transition of the aqueous solutionof amyloidogenic peptides results in its separation into amyloidfibrils (organic-rich phase) and an aqueous solution exhibiting anextremely low critical peptide concentration (water-rich phase).2,3

The aggregation phenomena of biomolecules in aqueousenvironment can be complicated by several factors. Dueto hysteresis phenomena, the formation of an equilibrated

aggregated (organic-rich) phase may take extremely long times(years in the case of the highly insoluble amyloidogenic peptides,whose solubility limit is below the picomolar range2,4). Also, thefinite size of a system prevents formation of a condensed phasein oversaturated conditions.5 Therefore, aggregation of highlyinsoluble peptides can be suppressed in small volumes(for example, in biological cells and in their compartments).3

Further complications may arise from the irreversible chemicalreactions in the process of aggregation. The presence of exten-ded surfaces is another important factor that affects peptideaggregation. Clearly, the interplay of all theses factors as well asvarying thermodynamic conditions complicates the possibilityto characterize and predict the process of peptide aggregation.To understand its main regularities, it is thus reasonable toreveal separately the key factors affecting aggregation. In thepresent paper, we consider the effect of various surfaces on theaggregation of peptides in liquid water using molecular dynamicsimulations.

Due to the presence of a surface, the peptide concentration aswell as any other system property becomes local, that is, depen-dent on the distance from the surface. Clearly, the adsorption ofpeptides on a surface affects their aggregation behavior. If thepeptide-surface attraction is strong enough, a surface phasetransition (condensation of one or several condensed peptidelayers at the surface) becomes possible. This transition occurs atsome peptide concentration below the solubility limit of peptidesin bulk liquid water. At higher peptide concentrations, theformation of an organic-rich phase (macroscopic amorphous orordered peptide aggregates) becomes possible also in the bulk.However, within some range of peptide concentrations, thesystem is metastable with respect to the bulk demixing transition,but is unstable with respect to the surface phase transitions.Hence, the condensation of peptide layers on surfaces exhibiting

*Corresponding author. E-mail: ivan.brovchenko@tu-dortmund.de.(1) Chiti, F.; Dobson, C. M. Annu. Rev. Biochem. 2006, 75, 333–366.(2) Jarrett, J. T.; Lansbury, P. T. Cell. 1993, 73, 1055–1058.(3) Singh, G.; Brovchenko, I.; Oleinikova, A.; Winter, R. Biophys. J. 2008, 95,

3208–3221.

(4) Hortschansky, P.; Schroeckh, V.; Christopeit, T.; Zandomeneghi, G.;F::andrich, M. Protein Sci. 2005, 14, 1753–1759.(5) Binder, K. Eur. Phys. J. B 2008, 64, 307–314.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058

8112 DOI: 10.1021/la9006058 Langmuir 2009, 25(14), 8111–8116

Article Brovchenko et al.

strongattractive interactionwith peptides (regardless of the originof this interaction) may be the key mechanism of peptideaggregation in a wide concentration range.

The adsorption of peptides on a surface affects not only thedegree of peptide aggregation, but may also change the structureof the peptide aggregate. The two-dimensionality of the surfaceshould provide orientational ordering of the peptides as well asa preferential formation of two-dimensional aggregate. Theseeffects are expected to be dominant when the surface phasetransitions of the peptides occur on strongly attractive surfaces.Another important factor of the presence of a surface is its effecton the conformation of the single peptides, which, in turn, mayaffect both the degree of aggregation and the structure of theaggregate.6-8 Due to such multifaceted effect of a surface onpeptide aggregation, it is useful to analyze various aspects of thesesurface effects separately. One key parameter is the effectivestrength of the peptide-surface interaction. In aqueous solution,this effective strength is determined by the relative strengths ofthe peptide-surface and water-surface interactions. The directinteraction between peptides and biological surfaces includescontributions from Coulombic interactions, dispersion interac-tions, and hydrogen bonding. The presence of a solvent (water)causes appearance of an additional, solvent-induced interactionbetween the peptide and the surface, which crucially depends onthe strengths of the water-surface and water-water interactions.This solvent-induced interaction appears as an attraction betweenthe hydrophobic parts of peptides and a hydrophobic surface, andas a repulsion between the hydrophilic parts of peptides and ahydrophilic surface. This interaction originates from the tendencyof hydrophobic (hydrophilic) particles or surfaces to stay dehy-drated (hydrated).9

Experimental studies have shown that the degree of peptideand protein aggregation as well as the structure of their aggre-gates change near surfaces and strongly depend on the surfaceproperties.6-8,10-18 Hen egg white lysozyme is completely solublein liquid water but forms macroscopic aggregates enriched inintermolecular β-sheets at a surface, whose hydrophobicity,governing the strength of the water-mediated protein-surfaceinteraction, exceeds some value.13 The existence of a thresholdvalue of the surface hydrophobicity necessary for the formationofthe aggregated phase as well as the thickness of the adsorbedprotein film (about 2-3 protein monolayers) indicates existenceof a first-order surface phase transition of the lysozymemoleculesat hydrophobic surfaces.

Generally, amyloidogenic peptides are highly insoluble in water.Due to the long lag times of the bulk aggregation, peptideaggregates can appear at surfaces before their formation starts in

the bulk. The effect of surfaces on the aggregation reaction has beenextensively studied for amyloid-β (Aβ) peptides.6,7,10-12,14-17,19 Aβpeptides exhibit strong adsorption at the intrinsically hydrophobicliquid-vapor interface.12,17,19 Adsorption of Aβ peptides was alsoobserved on both hydrophilic6,11,14-16 and hydrophobic7,11,14-16

surfaces, as well as on both positively16 and negatively10 chargedsolid surfaces. On the other hand, desorption of Aβ peptides fromsome hydrophilic and negatively charged surfaces was also re-ported.16 On hydrophobic surfaces, Aβ peptides form a uniformedtightly packed layer12,16,19 or uniform, elongated sheets,11 whereasless ordered particulate aggregates appear at hydrophilic11 andpositively charged16 surfaces.This points toa stronger interactionofAβ peptides with hydrophobic surfaces. Variation of the propertiesof a membrane by changing the intramembrane ganglioside con-centration evidence that some threshold concentration isrequired for the adsorption of Aβ peptides at such lipid surfaces.6

Fibrillation of prion peptides on solid surfaces strongly decreasesupon heating.18 Both these features are consistent with the presenceof surface phase transitions of amyloidogenic peptides on certainsurfaces.

Elucidation of the dominating forces in peptide adsorption inan experiment is complicated by the fact that it is difficult to varysome surface property (for example, its hydrophilicity), keepingall other properties of the surface (e.g., its structure, heterogene-ity, charge) and peptide system constant. However, this can beachieved by computer simulations. Adsorption atmembranes20,21

and at the liquid-vapor water interface22 has been studied byatomistic simulations for the case of a single amyloidogenicpeptide, whereas their aggregation at surfaces has been studiedon the level of a simplifiedmodel,23 only. In this paper, we presentfully atomistic simulations of the aggregation of peptides inexplicit water near model surfaces. We consider exclusively theeffect of water-mediated peptide-surface interactions on peptideaggregation. In order to probe the range of possible scenarios,we use two kinds of highly insoluble amyloidogenic peptides(hydrophobic and hydrophilic) and two kinds of model smoothsurfaces (hydrophobic and hydrophilic). The effects of thewater-surface interaction on the degree of peptide aggregation and onthe structure of the peptide aggregates formed are analyzed.

Systems and Methods

We performed a series of computer simulation studies of over-saturated solutions of amyloidogenic peptides in liquid water.Two kinds of amyloidogenic peptides were used: the hydrophobicpeptide NFGAIL, (residues 22-27 of the human islet amyloidpolypeptide, molecular weight 633.75 Da), and the polar hydro-philic peptide GNNQQNY (residues 7-13 of the yeast prionSup35, molecular weight 836.82 Da). The solutions were simu-lated in the bulk (with periodic boundary conditions appliedin three dimensions) and in slit-like pores of 6 nm width (withperiodic boundary conditions applied in two dimensions).Besides, the solution of hydrophobic peptides was simulated inan infinite liquid slab of about 6 nmwidth with two liquid-vaporinterfaces. The use of such awide poremakes the water propertiesin the pore center very close to the bulk ones. The aqueoussolutions were represented by six peptides and about 6200-6700watermolecules. This corresponds to aweight concentrationof peptides of about 3.1-3.3% for the solutions of the hydro-phobic peptides and of about 4.0-4.3% for the solutions of thehydrophilic peptides.

(6) Matsuzaki, K.; Horikiri, C. Biochemistry 1999, 38, 4137–4142.(7) Giacomelli, C. E.; Norde, W. Biomacromolecules 2003, 4, 1719–1726.(8) Lopes, D.; Meister, A.; Gohlke, A.; Hauser, A.; Blume, A.; Winter, R.

Biophys. J. 2007, 93, 3132–3141.(9) van Oss, C. J. J. Mol. Recogn. 2003, 16, 177–190.(10) Terzi, E.; H

::olzemann, G.; Seelig, J. J. Mol. Biol. 1995, 252, 633–642.

(11) Kowalewski, T.; Holtzman, D. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,3688–3693.(12) Schladitz, C.; Vieira, E. P.; Hermel, H.; M

::ohwald, H. Biophys. J. 1999, 77,

3305–3310.(13) Sethuraman, A.; Belfort, G. Biophys. J. 2005, 88, 1322–1333.(14) Giacomelli, C. E.; Norde, W. Macromol. Biosci. 2005, 5, 401–407.(15) McMasters, M. J.; Hammer, R. P.; McCarley, R. L. Langmuir 2005, 21,

4464–4470.(16) Rocha, S.; Krastev, R.; Th

::unemann, A. F.; Pereira, M. C.; M

::ohwald, H.;

Brezesinski, G. ChemPhysChem 2005, 6, 2527–2534.(17) Maltseva, E.; Kerth, A.; Blume, A.; M

::ohwald, H.; Brezesinski, G. Chem-

BioChem 2005, 6, 1817–1824.(18) Ku, S. H.; Park, C. B. Langmuir 2008, 24, 13822–13827.(19) Lep�ere, M.; Muenter, A.; Chevallard, C.; Guenoun, P.; Brezesinski, G.

Colloids Surf. A: Physicochem. Eng. Asp. 2007, 303, 73–78.

(20) Xu, Y.; Shen, J.; Luo, X.; Zhu, W.; Chen, K.; Ma, J.; Jiang, H. Proc. Natl.Acad. Sci. U.S.A. 2005, 102, 5403–5407.

(21) Davis, C.; Berkowitz, M. Biophys. J. 2009, 96, 785–797.(22) Knecht, V.; Mohwald, H.; Lipowsky, R. J. Phys. Chem. B 2007, 111, 4161–

4170.(23) Friedman, R.; Pellarin, R.; Caflisch, A. J. Mol. Biol. 2009, 387, 407–415.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058

DOI: 10.1021/la9006058 8113Langmuir 2009, 25(14), 8111–8116

Brovchenko et al. Article

All atomic molecular dynamic simulations were carried out atT = 330 K with Gromacs software using the OPLS force field24

for the peptides and the SPCE model25 for the water molecules.The cutoff of 1.2 nmwas used for intermolecular interactions andthe PME method was used to treat long-range Coulombic inter-actions. The interaction of water molecules with the smooth porewalls was represented by a (9-3) Lennard-Jones potential. Twokinds of pore walls were considered: with a well-depth U0 ofthe water-wall potential equal to-0.30 kcal/mol, and withU0 =-3.90 kcal/mol. The strength of the water-surface interaction ofthese walls approximately corresponds to a strongly hydrophobicparaffin-like surface and to hydrophilic silica-like surfaces,respectively.26 The peptides did not interact with the walls.

Simulation of the bulk aqueous solutions of the peptides wereperformed atP=1bar. In the case of the liquid slab simulations,the conditions of the liquid-vapor coexistence are provided bythe presenceof the vapor phase in the simulationbox. Simulationsof the aqueous solutions in pores were performed in the constant-volume ensemble. First, the pore was filled with liquid water andwater molecules were added or deleted to adjust the water densityin the pore center to the bulk density of SPCEwater at the liquid-vapor coexistence (about 0.985 g/cm3 at 330 K27). Afterward, sixpeptideswere randomly inserted into the simulationbox such thatthe peptides are at least 0.7 nm away from each other and 1.5 nmaway from the surfaces, and water molecules overlapping withpeptides were deleted. Such procedure ensures that the thermo-dynamic conditions of the solutions in the pores are close to thosein the bulk. For each type of surface and peptide combination,five simulationswith different initial velocitieswere carried out forthe duration of 70 ns.

Results

The density profiles of liquid water in the two types of poresconsidered and the density profile in the slab of liquid water areshown in Figure 1. These profiles were obtained from the localdensities of the centers of water oxygens. Note, that the pore wallslocated at-3 and+3nmcorrespond to the locationof the centersof atoms forming the surface layer of the model solid. The datareveal a pronounced density depletion of liquid water near the

hydrophobic surface caused by the domination of the effect ofmissing neighbors over the weakly attractive water-surfacepotential.28 The observed degree of the density depletion nearthe hydrophobic surface agrees with the available experimentaldata.29,30 The thickness of the liquid-vapor interface is largerthan the thickness of the interface between liquid water andhydrophobic surface (see right panel inFigure 1). It is close to thatobtained in other simulation studies31 and is slightly below thevalues obtained in the experiments (see ref 32) for data collection)due to the unavoidable suppression of the capillary waves insimulations. The two pronounced water layers near the hydro-philic surface are similar to those observed in numerous othersimulation studies.28,33 The high value of the local density at themaximum of the first peak does not mean an enhancement of thewater density in the first layer, but a strong localization of watermolecules in a plane parallel to the surface. Strong orientationalordering of water molecules in the first surface layer becomesessentially weaker in the second layer and practically disappearsstarting from the third one.33

The time dependences of the probability distribution of thecenters ofmass of the peptides in the pores and in the slab of liquidwater are depicted in Figures 2, 3. After random insertion ofhydrophobic peptides in liquid water into the hydrophobic pore

Figure 1. Density profiles of liquid water in the hydrophobic andhydrophilic pores and in the slab of liquid water. Pore walls arelocated at-3 and +3 nm. The profiles near the pore walls and atthe liquid-vapor interface are shown in an enlarged scale in theright panel.

Figure 2. Time dependence of the distribution of the centers ofmass of hydrophobic peptides in the slab of aqueous solution(upper panel) and in the aqueous solution inside a hydrophobicpore (two independent simulation runsare shown in themiddle andlower panels). The midpoints of the liquid-vapor interface areshown by dashed lines. The pore walls are indicated by horizontalblack lines. The color scale on the right-hand side indicates changesof the peptide density.

(24) Jorgensen,W. L.; Tirado-Rives, J. J. Am. Chem. Soc. 1988, 110, 1657–1666.(25) Berendsen, H.; Grigera, J.; Straatsma, T. J. Phys. Chem. 1987, 91, 6269–

6271.(26) Werder, T.; Walther, J.; Jaffe, R.; Halicioglu, T.; Koumoutsakos, P.

J. Phys. Chem. B 2003, 107, 1345–1352.(27) Guissani, Y.; Guillot, B. J. Chem. Phys. 1993, 98, 8221–8235.(28) Brovchenko, I.; Oleinikova, A. Interfacial and Confined Water; Elsevier:

Amsterdam, 2008.

(29) Jensen, T. R.; Jensen,M. O.; Reitzel, N.; Balashev, K.; Peters, G. H.; Kjaer,K.; Bjornholm, T. Phys. Rev. Lett. 2003, 90, 086101.

(30) Steitz, R.; Gutberlet, T.; Hauss, T.; Klosgen, B.; Krastev, R.; Schemmel, S.;Simonsen, A. C.; Findenegg, G. H. Langmuir 2003, 19, 2409–2418.

(31) Taylor, R. S.; Dang, L. X.; Garrett, B. C. J. Phys. Chem. 1996, 100, 11720–11725.

(32) Caupin, F. Phys. Rev. E 2005, 71, 051605.(33) Brovchenko, I.; Oleinikova, A.Molecular organization of gases and liquids

at solid surfaces. In Handbook of Theoretical and Computational Nanotechnology;Rieth, M., Schommers, W., Eds.; American Scientific Publishers: StevensonRanch, CA, 2006; Volume 9, Chapter 3, pp 109-206.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058

8114 DOI: 10.1021/la9006058 Langmuir 2009, 25(14), 8111–8116

Article Brovchenko et al.

or in the slab of liquid water, the equilibration of the systemrequires up to 30 ns. After equilibration, all peptides wereadsorbed at the interfaces. Usually, the peptides are adsorbedon the two opposing interfaces, but sometimes all peptides areadsorbed at one of them, only. The situation is quite different inhydrophilic pores, where both hydrophilic and hydrophobicpeptides (upper and lower panels inFigure 3, respectively) quickly(in a few ns) become localized in the center of the pore, and suchscenario is also observed in the case of the hydrophilic peptides inthe hydrophobic pore (middle panel in Figure 3).

The density profiles of the peptides in the pores, calculatedafter the equilibration period of 30 ns and by taking into accountall peptide atoms, are shown in Figures 4 and 5. The degreeof peptide localization in the pore center is quite similar forboth hydrophilic and hydrophobic peptides in the hydrophilicpore. The hydrophilic peptides, however, show a weaker localiza-tion in the center of the hydrophobic pore. An opposite behavioris observed for hydrophobic peptides in the hydrophobic pore orin a slab of liquid water, where in all simulation runs, the peptidesare strongly adsorbed at the interfaces (Figure 5). The degree ofpeptide localization near interfaces essentially exceeds that in thepore center as can be seen from the comparison of the width ofthe density distributions shown in Figures 4 and 5. This can beadditionally illustrated by the distribution of the centers of massof the peptides.When the peptides are localized in the pore center,this distribution (not shown) practically coincides with thatshown in Figure 4, whereas it is much narrower in the case ofpeptide localization near the interfaces (compare blue and redlines in Figure 5).

The probability distribution of the angle R between the poresurface and the vector connecting twomost distant peptide heavyatoms are shown in Figure 6.When the peptides are repelled fromthe pore walls and localized in the pore center, the orientations of

their longest axes are highly isotropic (left panel in Figure 6) andonly a slight preferential orientation of these axes parallel to thewall can be noticed. The situation is quite different in the case ofstrong adsorption of the peptides at interfaces (right panel inFigure 6). The localization near the interfaces essentially enhancesthe orientational ordering of the peptides andmakes their longestaxes align parallel to the interfaces.

The degree of peptide aggregation was characterized by theparameter R, which denotes the probability to find more thantwo-thirds of all peptides in the largest peptide cluster.3 Thedistance between the centers ofmass of two peptideswas used as aconnectivity criterium: two peptides are considered to belong toone cluster if this distance does not exceed some critical value rc.The dependences of the aggregation parameter R on the distancerc are shown in Figure 7. The degree of aggregation of thehydrophobic and hydrophilic peptides in the bulk solutions isquite similar (see upper panel in Figure 7). It can be comparedwith the similar dependences obtained for the aqueous solutionsof FLVHS peptides.3 The degree of aggregation is noticeablyweaker in the latter case in a wide range of rc. This reflects aweaker propensity of these peptides to aggregation. On the otherhand, the presence of the methyl caps in FLVHS peptides (whichwere absent in the peptides studied in the present paper) couldshift the equilibrium interpeptide distances in the aggregate tohigher values.

The aggregation of the hydrophilic peptides becomes weaker inboth hydrophilic and hydrophobic pores (see middle panel inFigure 7), as inferred from the shift of the dependence R(rc) to ahigher rc values. The situation with the hydrophobic peptides isopposite (lower panel in Figure 7). The aggregation of thehydrophobic peptides is slightly fostered due to the confinementin the hydrophilic pores and it enhances drastically upon con-finement in the hydrophobic pore. The shift of the dependenceR(rc) to lower rc values may also indicate formation of closelypacked aggregates characterized by β-sheets with interpeptidedistances of about 0.5 nm. This is supported by the increasingaverage number of interpeptide H-bonds per one peptide (from11.8 to 17.1) and by the decreasing number of intrapeptideH-bonds (from 7.6 to 4.6) upon peptide adsorption on thehydrophobic surface.

Figure 3. Time dependence of the distribution of the centers ofmass of hydrophilic peptides in aqueous solution in the hydrophilic(upper panel) and hydrophobic (middle panel) pores and ofhydrophobic peptides in aqueous solution in the hydrophilic pore(lower panel). The pore walls are indicated by horizontal blacklines. The color-scale on the right-hand side indicates changes ofthe peptide density.

Figure 4. Density profiles of peptides in aqueous solution in thepores: hydrophilic peptides in hydrophobic (red line) and hydro-philic (black line) pores (left vertical axis); hydrophobic peptides inthe hydrophilic pore (blue line, right vertical axis). The scales of theleft and right are proportional to the molecular weights of thehydrophilic and hydrophobic peptides.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058

DOI: 10.1021/la9006058 8115Langmuir 2009, 25(14), 8111–8116

Brovchenko et al. Article

The change of the degree of peptide aggregation by theconfinement also yields information about the character of thepeptide aggregation. InFigure 8, the number nHofwater-peptideH-bonds is shown as a function of the aggregation parameterR atrc = 0.9 nm (the case of the hydrophobic peptides inside thehydrophobic pore is excluded as nH is strongly affected by peptideadsorption). As can be seen from Figure 8, nH decreases uponaggregation of hydrophilic peptides. This indicates that aggrega-tion of these peptides occurs predominantly via direct contactsbetween hydrophilic groups of the peptides (this may be called asthe “hydrophilic aggregation”), which become less accessible forwater molecules upon formation of the peptide aggregate, thusrendering its surface effectively more hydrophobic. Conversely,nH increases upon aggregation of the hydrophobic peptides. Inthis case, formation of the peptide aggregate occurs pre-dominantly via formation of close contacts between hydrophobicpatches of the peptides (“hydrophobic aggregation”). Accord-ingly, the surface of the peptides exposed to liquid water becomeseffectively more hydrophilic upon peptide aggregation. Similarcharacter of aggregation was also observed for the hydrophobicFLVHS peptides.3

Discussion

The results presented show that even a single extrinsic factorsuch as the water-mediated interaction between peptides and

surfaces has a drastic influence on peptide aggregation phenom-ena. Importantly, this influence depends on the character ofpeptide aggregation and on the strength of the water-surfaceinteraction. The effect of water-mediated peptide-surface inter-actions is strongly determined by the balance of direct peptide-surface and water-surface interactions. If hydrophilic residues

Figure 5. Density profiles of hydrophobic peptides in the slab of aqueous solution (left panel, red line) and in aqueous solution in thehydrophobic pore (right panel, red line). The density profiles of the centers ofmass of the peptides are shown by the blue lines. Themidpointsof the liquid-vapor interface and the pore walls are shown by the black dashed and solid lines, respectively.

Figure 6. Probability distribution of the angle R between the porewall or the liquid-vapor interface and the vector connecting twomost distant peptide heavy atoms.

Figure 7. Dependence of the aggregation parameter R on thedistance rc used as a criterion for interpeptide connectivity. Upperpanel: bulk aqueous solutions of the peptides. Middle panel:aqueous solution of hydrophilic peptides in pores. Lower panel:aqueous solution of hydrophobic peptides in pores.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058

8116 DOI: 10.1021/la9006058 Langmuir 2009, 25(14), 8111–8116

Article Brovchenko et al.

dominate the peptide‘s primary structure, it is energetically morefavorable that such peptides stay hydrated, i.e., remain sur-rounded by at least one water layer. This trend is enhanced by ahydrophilic character of the surface and is opposed by a hydro-phobic surface. However, even in the latter case the tendency ofthe hydrophilic peptides to stay hydrated remains dominating andthey are repelled from the surface (see Figure 4). Hence, hydro-philic peptides should usually be repelled from a surface as directpeptide-surface contacts are energetically unfavorable even inthe case of strongly hydrophobic paraffin-like surfaces. In con-trast, hydrophobic peptides prefer to be dehydrated and the directcontacts with a surface provide energetically more favorable

states. However, this trend can be overcome if the surface ishydrophilic enough. Near hydrophilic silica-like surfaces, thetendency of the surface to be hydrated overcomes the tendencyof the hydrophobic peptides to be dehydrated; consequently, thepeptides are repelled from the surface.

The data clearly show that confinement in pores can causeboth enhancement and weakening of peptide aggregation.In particular, we may expect weaker aggregation of hydrophilicpeptides and stronger aggregation of hydrophobic peptidesin pores (see middle and low panels in Figure 7). The possibilityto weaken peptide aggregation by confinement of the aqueoussolution of hydrophilic peptides in pores may be usefulin pharmaceutical applications. The most drastic surface effecton aggregation is observed when hydrophobic peptides adsorbon hydrophobic surfaces or at the liquid-vapor interface. Inboth the cases, the degree of peptide aggregation is stronglyenhanced and the ordering effect of the surface causes alignmentof the peptides parallel to the surface. Both factors shouldpromote formation of intermolecular β-sheets, which is a keystructural element of amyloid fibrils. Thus, for hydro-phobic peptides, the formation of amyloid fibrils may be expectedfirst of all near hydrophobic surfaces. This observationagrees with experimental studies on Aβ peptides, which exhibitformation of tightly packed peptide layers on hydrophobicsurfaces.11,12,16

Acknowledgment. Financial support from the InternationalMax-Planck Research School in Chemical Biology and from theZentrum f

::ur Angewandte Chemische Genomik is gratefully

acknowledged.

Figure 8. Dependence of the number nH of water-peptideH-bonds on the aggregation parameter R for hydrophilic(left panel) and hydrophobic (right panel) peptides in liquid water.

Dow

nloa

ded

by D

OR

TM

UN

D L

IBR

AR

IES

on O

ctob

er 3

0, 2

009

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Apr

il 27

, 200

9 | d

oi: 1

0.10

21/la

9006

058