Insight into the Stability of Cross-b Amyloid Fibril from MolecularDynamics Simulation

Yue Chen,1 Yong-Jie He,1 Maoying Wu,1 Guanwen Yan,2 Yixue Li,3 Jian Zhang,4 Hai-Feng Chen1,31 Department of Bioinformatics and Biostatistics, College of Life Sciences and Biotechnology, Shanghai Jiaotong University,

800 Dongchuan Road, Shanghai, 200240, China

2 Shanghai High School, 400 Shangzhong Road, Shanghai, 200231, China

3 Shanghai Center for Bioinformation Technology, 100 Qinzhou Road, Shanghai, 200235, China

4 Institute of Medical Science, School of Medicine, Shanghai Jiaotong University, 280 Chongqing Road,

Shanghai, 200025, China

Received 30 November 2009; revised 20 January 2010; accepted 3 February 2010

Published online 9 February 2010 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/bip.21405

This article was originally published online as an accepted

preprint. The ‘‘Published Online’’ date corresponds to the preprint

version. You can request a copy of the preprint by emailing the

Biopolymers editorial office at biopolymers@wiley. com

INTRODUCTION

Amyloid-like fibrils are found in many fatal diseases,

including Alzheimer’s disease, type II diabetes mel-

litus, and the transmissible spongiform encephalo-

pathies, and prion disease.1 These diseases are

linked to protein misfolding. Then, aggregation of

misfolding protein into amyloid-like fibrils is a key factor in

these diseases.2 It was found that these fibrils contain a cross-

b spine and b-strands perpendicular to the fibril axis.3

Because the character of amyloid fibril is noncrystalline and

insoluble, it is difficult to crystalize atomic-level structures of

cross-b spine with traditional experimental methods. Up to

2005, Eisenberg et al. determined X-ray crystal structures of

Insight into the Stability of Cross-b Amyloid Fibril from MolecularDynamics Simulation

Additional Supporting Information may be found in the online version of this

article.Correspondence to: Jian Zhang; jian.zhang@sjtu.edu.cn or Hai-Feng Chen, e-mail:

haifengchen@sjtu.edu.cn

ABSTRACT:

Amyloid fibrils are considered to play causal roles in the

pathogenesis of amyloid-related degenerative diseases

such as Alzheimer’s disease, type II diabetes mellitus, the

transmissible spongiform encephalopathies, and prion

disease. The mechanism of fibril formation is still hotly

debated and remains an important open question. In this

study, we utilized molecular dynamics (MD) simulation

to analyze the stability of hexamer for eight class peptides.

The MD results suggest that VEALYL and MVGGVV-1

are the most stable ones, then SNQNNY, followed by

LYQLEN, MVGGVV-2, VQIVYK, SSTSAA, and

GGVVIA. The statistics result indicates that hydrophobic

residues play a key role in stabilizing the zipper interface.

Single point and two linkage mutants of MVGGVV-1

confirmed that both Met1 and Val2 are key hydrophobic

residues. This is consistent with the statistics analysis. The

stability results of oligomer for MVGGVV-1 suggest that

the intermediate state should be trimer (3-0) and

tetramer (2-2). These methods can be used in

stabilization study of other amyloid fibril. # 2010 Wiley

Periodicals, Inc. Biopolymers 93: 578–586, 2010.

Keywords: amyloid-like fibril; aggregation mechanism;

mutation; stability; oligomer

Contract grant sponsor: Instrumental Analysis Center of Shanghai Jiaotong Uni-

versity

Contract grant sponsor: National Natural Science Foundation of China

Contract grant numbers: 30770502, 20773085

Contract grant sponsor: Natural Science Foundation of Shanghai China

Contract grant number: 10ZR1414500

Contract grant sponsor: Ministry of Science and Technology China

Contract grant number: 2010CB833601

Contract grant sponsor: National 863 High-Tech Program

Contract grant number: 2007DFA31040

VVC 2010 Wiley Periodicals, Inc.

578 Biopolymers Volume 93 / Number 6

amyloid-like fibril from a yeast prion-derived peptide by X-

ray microcrystallography.4 The research brings a break in this

field. Then a set of crystal structures from different protein

precursors were released with the same method.5 These

atomic-resolution structures make it possible to investigate

the common characters of amyloid formation by atomic mo-

lecular dynamics methods, which can directly compare with

experimental results.6,7

The aggregation mechanism of amyloid fibril is still hotly

debated and remains an important unresolved open ques-

tion. To explain the conversion of peptides from soluble to fi-

brous forms, several types of atomic-level models have been

proposed, such as refolding, natively disordered and gain of

interaction.3 A set of computational studies provide an

insight into the characteristics of the amyloid aggregate.8–16

Toschi et al. suggests that electric fields favor the switch of

Ab-peptides from helical to b-sheet conformation.17 Mas-

man et al. explored the contributions of the different struc-

tural elements of trimeric and pentameric full-length Ab(1-42) aggregates in solution to their stability and conforma-

tional dynamics.18 Kent et al. reports that a solvent-exposed

hydrophobic patch is believed to be important for aggrega-

tion of Ab(10-35).14 Nussinov et al. studies Ab40 elongation

and lateral association and the aggregation pathway of b2-microglobulin amyloid with molecular dynamics simula-

tions.13,19 Garcia et al. research the flexibility of C terminus

of Ab42 and find this responsible for the higher propensity of

this peptide to form amyloids.20 DeMarco and Daggett have

studied the aggregation process of prion fibril using atomic

molecular dynamics.8 Wu et al. has reported the time scale of

amyloidogenic hexapeptide NFGAIL aggregation.9 Further-

more, Gnanakaran et al. has investigated the aggregation of

simple amyloid b-dimer with replica-exchange molecular dy-

namics.10 Besides the research of aggregation mechanism,

there are also some investigations about inhibitors. N-meth-

ylated inhibitors can disassemble the early steps of Ab16-22protofibril.2 These researches tell us the aggregation process

of amyloid fibril. However, we still do not know if there is an

intermediate state and transition state during the process of

aggregation. The purpose of this study is to answer these

questions.

Common structure character for these amyloid fibrils

implies common mechanism of pathogenesis.21 This indi-

cates that the study of short peptide aggregation could reveal

some common fundamental mechanism that governs fibril

formation in large protein systems. In this study, we intend

to research the stability of eight hexamer peptides to under-

stand their aggregation mechanisms using room-temperature

molecular dynamics simulation in explicit water. A represen-

tative MVGGVV-1 hexamer model was shown in Figure 1.

This simulation study can shed light on possible mechanisms

of aggregation.

Computational MethodsMolecular Dynamics Simulation. In this work, the atomic

coordinates of the cross-b amyloid hexamers were con-

structed with WinCoot22 from the X-ray structure using the

symmetry operations of structure space group (P21).4,5 Resi-

due mutant was performed using SCWRL3.23 Hydrogen

atoms were added using the LEAP module of AMBER8.24

Particle Mesh Ewald (PME)25 was employed to treat long-

range electrostatic interactions with the default setting in

AMBER8.24 A revised parm99 force field was used for intra-

molecular interactions.26,27 1000-step steepest descent mini-

mization was performed to relieve any structural clash in the

solvated system. The SHAKE algorithm28 was used to con-

strain bonds involving hydrogen atoms so that a 2fs time

step was used. The minimized system was heated up and

brief equilibrated for 20 ps in the NVT ensemble at 298K

with PMEMD of AMBER8. Langevin dynamics was used in

the heating and equilibration runs with a friction constant of

1 ps21. Ten independent trajectories of 10.0 ns each in the

NPTensemble at 298 Kwere then simulated with PMEMD of

AMBER8. The protocol is also shown in literature.29–32 A

total of 2.1 ls trajectories were collected, respectively, takingabout 50,770 CPU hours on the in-house Xeon (1.86GHz)

cluster.

The Definition of Oligomer

The native contacts are the main interactions among these

interfaces and class into two categories. One is inter-strands,

such as between Strands 1 and 2, Strands 2 and 3, Strands 4

and 5, Strands 5 and 6. The other is inter-sheets, such as

between Strands 1 and 4, Strands 1 and 5, Strands 2 and 5,

Strands 2 and 6, Strands 3 and 6 (shown in Figure 1).6

FIGURE 1 Hexamer of MVGGVV-1 cross-b prion.

The Stability of Cross-b Amyloid Fibril 579

Biopolymers

According to the arrangement of peptide, there is one possi-

bility for hexamer and pentamer. For tetramer, there are two

possibilities, one three strands in one sheet and the fourth on

the other sheet, or each two in the same sheet (3-1 versus 2-

2). For trimer, there are also two possibilities, all three

strands belong to one single sheet, or two strands in one

sheet and the third on the other sheet (3-0 versus 2-1). For

dimer, two conformations are defined. Two strands are on

one sheet or belong to different sheets (2-0 versus 1-1).

According to the arrangement of peptide, the atomic coordi-

nates of these oligomers were constructed and extracted from

the structure of hexamer.

RESULTS AND DISCUSSION

The Stability of Hexamer for Eight Class Peptides

The stability of hexamer for 10 trajectories of 10.0 ns each

was simulated at 298 K. The simulation condition is listed in

Table I. The Ca atom RMSDs are shown in Figure 2. The av-

erage root mean square deviation (RMSD) is about 18 A for

SSTSAA, 10 A for LYQLEN, 9 A for MVGGVV-2, VQIVYK,

GGVVIA, 6 A for SNQNNY, 2.5 A for VEALYL, and 2 A for

MVGGVV-1, respectively, at the end of 10-ns simulation.

This suggests that the hexamers of SSTSAA, LYQLEN,

MVGGVV-2, VQIVYK, and GGVVIA are unstable. VEALYL

and MVGGVV-1 are very stable. The stability of SNQNNY is

situated between the two ones. Therefore, the hexamer

of peptide MVGGVV-1 was chosen to study the stability

mechanism.

To study the stability in detail, Ca variations of eight pep-

tides are illustrated in Figure 3. This indicates that all chains

have common characteristics of small variation for the five

central residues whereas large variations for the two end resi-

dues, suggesting that the center residues are more rigid than

the residues in the termini regions. This is in agreement with

the reported of Zheng et al.33 The Ca RMSFs of MVGGVV-1

and VEALYL are much smaller than those of SNQNNF,

SSTSAA, LYQLEN, VQIVYK, MVGGVV-2, and GGVVIA.

Table I Summary of Simulation Conditions

Class Sequence Strand/Sheet organization Counter Ion Water Trajectories Time(ns)

Class 1 SSTSAA Hexamer_3_3 / 3810 10 100

VQIVYK 6 Cl- 3812 10 100

Class 2 SNQNNF / 3812 10 100

Class 4 GGVVIA / 3815 10 100

Class 7 VEALYL 6 Na1 3811 10 100

LYQLEN 6 Na1 3814 10 100

Class 8 MVGGVV-2 / 3812 10 100

MVGGVV-1 / 3813 10 100

Oligomer MVGGVV-1 Dimer_1_1 / 1273 10 100

MVGGVV-1 Dimer_2_0 / 1269 10 100

MVGGVV-1 Trimer_2_1 / 1907 10 100

MVGGVV-1 Trimer_3_0 / 1907 10 100

MVGGVV-1 Tetramer_2_2 / 2543 10 100

MVGGVV-1 Tetramer_3_1 / 2542 10 100

MVGGVV-1 Pentamer_3_2 / 3173 10 100

Mutant GVGGVV-1 Hexamer_3_3 / 3813 10 100

MGGGVV-1 / 3814 10 100

MVGGGV-1 / 3811 10 100

MVGGVG-1 / 3813 10 100

GGGGVV-1 / 3813 10 100

MVGGGG-1 / 3814 10 100

FIGURE 2 Ca RMSD of eight hexamer peptides.

580 Chen et al.

Biopolymers

This is consistent with the results of Ca RMSD. However, the

fluctuation of residue 6-7 is larger than those of 1-2 for

Strands 1-3, and the fluctuation for Strands 4-6 is reverse. A

little twist of b-strand for these hexamer peptides during

room temperature was found. This is in agreement with

other simulation.18,34 The previous molecular dynamics sim-

ulation also suggests that the 10-stranded b-sheets of

SSTSAA and VQIVYK have high fluctuations and significant

distortions.35

To further monitor the interaction responsible for the

steric zipper motif stability, the average number of native

contact and hydrogen bond for each residue of eight hexamer

peptides was calculated. A hydrogen bond is assigned if the

distance between donor and acceptor atom is less than 3.5 A.

There are two types of native contact. One is the contact of

interstrand, and the other is intersheet. An intersheet chain

contact is defined if the distance between the center mass of

two side chains is less than 6.0 A (shown in Figure 4). In gen-

eral, the average number of hydrogen bond is between 0.5

and 2.5. The average number of hydrogen bond for SNQNNF

is the largest among eight hexamer peptides. This can explain

the higher zipper stability of SNQNNF. However, the average

hydrogen bond value of residue F is very small. This might

decrease the zipper stability of whole system. For MVGGVV-

1 and VEALYL, the average number of hydrogen bond is

about 1.5 for each residue. These hydrogen bonds play a key

role in stabilizing the zipper motif. This is consistent with the

secondary structure evolution of MVGGV-1 and VEALYL

(shown in Figure 5). Their secondary structures almost keep

b-sheet during 10 ns MD simulation. For SSTSAA, GGVVIA,

and VQIVYK, the variation of hydrogen bond is relative large

and the number is almost smaller than those of MVGGVV-1

and VEALYL. This might influence their zipper stabilities.

Besides hydrogen bond, we also count the average native

contact of eight hexamer peptides. The average native contact

of each residue for MVGGVV-1 is similar to VEALYL. These

interactions also can keep their secondary structures. The

variation of SSTSAA and GGVVIA is the largest; this indi-

cates that SSTSAA and GGVVIA are not stable. Not surpris-

ingly, different arrange forms will induce different stability.

There are three hydrophobic contacts between intersheet for

MVGGVV-2 (V11/M190V11/V260and V14/V35), and six

hydrophobic contacts for MVGGVV-1 (V6/V200V12/V200V12/

M250V12/V290V17/M250and M13/V32). Therefore, the stabil-

ity of MVGGVV-1 is higher than that of MVGGVV-2. In

summary, these native contacts of interstrand and intersheet

should be major driving forces for the aggregation of peptides.

The distances of interstrand and intersheet for eight hex-

amers are listed in Table II. The difference of distance

between simulation and crystal for SSTSAA and GGVVIA is

the largest among eight hexamer peptides. This suggests that

SSTSAA and GGVVIA are the less unstable. These distorts

will disaggregate their hexamer structures. The simulation

distance of interstrand and intersheet for VEALYL and

MVGGVV-1 is almost similar to the crystal. These results are

consistent with those of RMSF and RMSD. This is also in

agreement with the report that LVEALYL is the main contrib-

utor to the spine formation of fibrils for full-length insulin.36

To study the influence of the property of residue to the

stability of hexamer peptides, we count the property of resi-

due having intersheet native contact for eight hexamer pep-

tides. The results are listed in Figure 6. The fraction of hydro-

phobic residues is larger than 70% with the population of

native contact higher than 30, 40 and 50%, respectively. This

suggests that hydrophobic residues likely provide a strong

contribution to stabilizing these hexamer peptides.14 How-

FIGURE 3 Ca variation of residues for eight hexamer peptides.

FIGURE 4 Average number of hydrogen bond and native contact

for eight hexamer peptides.

The Stability of Cross-b Amyloid Fibril 581

Biopolymers

ever, Esposito et al. reports that polar and aromatic residues

play a key role in the steric zipper motif from explicit solvent

molecular dynamics simulation.35

The alignment between average structure and crystal

structure of MVGGVV-1 is shown in supplement file (Sup-

porting Information Figure 1S). The Ca rms between them is

about 1.65 A. This suggests that MVGGVV-1 is rather stabil-

ity. Because the fluctuation of RMSF for MVGGVV-1 is the

smallest among eight hexamer peptides, it was chosen to

study aggregation mechanism in detail.

Mutant Research

To further monitor the interaction responsible for the aggre-

gation stability, native contacts of interstrand and intersheet

for MVGGVV-1 were studied. There are two types of native

contact. One is the contact of interstrand, and the other is

intersheet. An intersheet chain contact is defined if the dis-

tance between the center of mass of two side chains is less

than 6.0 A. The populations of native contacts for a couple of

peptide of interstrand and intersheet in simulation are shown

in Figure 7. Six stable interstrand and two stable intersheet

native contacts can be found with populations higher than

40%. The native contacts of intersheet focus on Met1/Val2.

This suggests that these native contacts of interstrand and

intersheet should be major driving forces for the aggregation.

The native contact of MVGGVV-1 suggests that Met1 and

Val2 are key residues to stabilize the cross-b zipper interface.

To confirm these key residues, we mutate each residue with

tiny Gly. According to the distribution of residue in

sequence, mutation research can be classed into two catego-

ries: single-point residue mutation and two linkage muta-

tion. The fraction of native contact (Qf) for wild type and

mutations is shown in Figure 8. Qf of wild type is larger than

90% and keeps constant. This suggests that wild type of

MVGGVV-1 is very stable. This is consistent with the results

of RMSF. For single mutation, the mutant of M1G and V2G

induces a significant decrease of Qf, which value is about

Table II The Distance of Interstrand and Intersheet for

Simulation and Crystal

Sequence

Simulation (A) Crystal (A)

dintersheet dinterstrand dintersheet dinterstrand

SSTSAA 13.796 1.91 12.126 2.65 10.79 4.83

VQIVYK 16.096 0.36 5.49 6 0.31 14.74 4.86

SNQNNF 16.796 0.56 6.26 6 0.46 13.91 4.88

GGVVIA 11.736 1.36 7.64 6 1.21 8.96 4.79

LYQLEN 16.266 1.40 5.14 6 0.22 17.35 4.89

MVGGVV-2 13.426 1.48 5.30 6 0.42 15.15 4.86

VEALYL 13.956 0.69 5.03 6 0.098 13.71 4.83

MVGGVV-1 11.456 0.26 4.92 6 0.10 12.75 4.90FIGURE 6 Statistical result of hydrophobic residues with inter-

sheet native contact.

FIGURE 5 Secondary structure evolution of MVGGVV-1 and VEALYL. A: MVGGVV-1; B: VEALYL.

582 Chen et al.

Biopolymers

40% of wild type. V5G and V6G just bring a slight decrease

for native contact. This suggests that Met1 and Val2 are key

residues for the zipper stability. For two linkage mutations,

the Qf of M1GV2G is also significant smaller than that of

V5GV6G. Therefore, the two-linkage residues of Met1Val2

are more important than those of Val5Val6. This is in agree-

ment with the result of single point mutation.

The distances of interstrand and intersheet for wild type

and mutants are listed in Figures 9 and 10. The distance of

interstrand for WT, V5GV6G and V6G is about 4.5 A and

their variations are very small. The distance of interstrand for

V5G is about 6.5 A. The distance of interstrand for M1G,

V2G, and M1GV2G is about 10 A, 9 A, and 10 A, respec-

tively. This suggests that the strands of M1G, V2G, and

M1GV2G have the propensity of expend. The distance of

intersheet for WT, V5G, V6G is between 10 A and 11 A.

Their hexamers almost keep stable. Surprisingly, the distance

of intersheet for V5GV6G, M1G, and V2G decreases. The b-sheets have the propensity of compression. On the contrary,

the distance of intersheet for M1GV2G increases. The hex-

amer of M1GV2G is almost disaggregation.

To reveal the structural adjustment for mutants, the inter-

actions between peptides are shown in Figures 11 and 12. In

general, the number of hydrogen bond and native contact for

M1G, V2G, and M1GV2G is significant smaller than that of

wild type, V5G, V6G, and V5GV6G, respectively. The values

for V5G, V6G, and V5GV6G are similar to that of wild type.

This suggests that the mutation of Met1 and Val2 signifi-

cantly decreases the hydrogen bonds and native contacts of

residues. This shows that Met1 and Val2 are key residues for

MVGGVV-1 aggregation.

To study the stability of mutation, Ca variations of wild

type and mutations are illustrated in Figure 13. The RMSF of

M1GV2G is the largest, then followed by V5G, M1G, V2G,

FIGURE 9 The interstrand distance for wild type and mutants.

FIGURE 8 The fraction of native contact for wild type and

mutants.

FIGURE 7 Native contacts of interstand and intersheet for

MVGGVV-1.

FIGURE 10 The intersheet distance for wild type and mutants.

The Stability of Cross-b Amyloid Fibril 583

Biopolymers

V6G, WT, and V5GV6G. Exception the mutant of V5G, the

variation order of RMSF is consistent with the result of inter-

actions between sheets.

The Stability of Oligomer for MVGGVV-1

In order to confirm the aggregation kinetics, the stabilities at

room temperature of dimer, trimer, tetramer, and pentamer

were studied and simulation conditions were also gathered in

Table I. As shown in shown in supplement file (Supporting

Information Figure 2S), the RMSDs quickly increased to 17

A for dimer (1-1) and �10 A for dimer (2-0) after 6 ns. This

suggests that dimer is not stable and discards their original

organization of structure. This is consistent with the report

of Zheng et al. that the dimer of GNNQQNY is not thermo-

dynamically stable state.33 Their average structures absolutely

depart from initial coordination of dimer. The b-sheet struc-

ture of dimer (2-0) also loses. Then, how about the stability

of trimer for the addition of a strand based on the dimer?

The trimer (2-1) is neither stable and its RMSD was about 15

A after 10 ns simulation. However, the RMSD of trimer (3-0)

was about 5 A, indicating significant stability of the struc-

tures. Using implicit solvent model, Gosponer et al. have

reported that three-stranded parallel in-register aggregates as

nucleus from three peptides simulation in an implicit sol-

vent.37 For the model system of tetramer with the addition

another strand, tetramer (2-2) is much more stable than tet-

ramer (3-1). Furthermore, the RMSD of tetramer (2-2) was

about 7 A after 8-ns simulation. The result suggests hat tet-

ramer (2-2) might be another stable state. The RMSD of pen-

tamer (3-2) is similar to that of trimer (3-0), indicating the

pentamer (3-2) is a stable state. The residue fluctuation of

these oligomers was shown in Figure 14. The fluctuations of

trimer (3-0), tetramer (2-2) and pentamer (3-2) were the

FIGURE 12 The number of native contact for wild type and

mutants.

FIGURE 13 Ca RMSF for wild type and mutants.FIGURE 11 The number of hydrogen bond for wild type and

mutations.

FIGURE 14 Ca variation of residues for oligomer.

584 Chen et al.

Biopolymers

smallest among these oligomers. The average native contacts

for these oligomers are shown in Figure 15. The average

numbers of native contact for trimer (3-0), tetramer (2-2),

and pentamer (3-2) are the largest among these oligomers,

and are consistent with the result of RMSF. There are two

types of conformer for trimer and tetramer. According to

their stabilities, the intermediate state should be trimer (3-0)

and tetramer (2-2). Collins et al report that fibers grow by

monomer addition.38 The possibility mechanism is that pen-

tamer is aggregated by monomer added after the intermedi-

ate state is formed.

CONCLUSIONSIn summary, our all-atom explicit solvent molecular dynam-

ics study reveals the stability of eight class peptides. The

results suggest that MVGGVV-1 and VEALYL are more stable

than other peptides. Statistical results indicate that hydro-

phobic interactions play key role in the stability of amyloid

fibril-like peptides. Then the most stability peptide was cho-

sen to furthermore study. Mutant research confirmed that

Met1 and Val2 are key residues to stabilize the cross-b zipper

interface. Finally, the stability results of oligomer for

MVGGVV-1 suggest that the flexibility of trimer (3-0), tet-

ramer (2-2), and pentamer (3-2) were the smallest among

these oligomers. According to their stabilities, the intermedi-

ate state should be trimer (3-0) and tetramer (2-2). These

results are helpful to understand the early aggregation of

amyloid fibril-like peptides.

REFERENCES1. Dobson, C. M. Trends Biochem Sci 1999, 24, 329–332.

2. Chebaro, Y.; Derreumaux, P. Proteins 2009, 75, 442–452.

3. Nelson, R.; Eisenberg, D. Curr Opin Struct Biol 2006, 16, 260–265.

4. Nelson, R.; Sawaya, M. R.; Balbirnie, M.; Madsen, A. O.; Riekel,

C.; Grothe, R.; Eisenberg, D. Nature 2005, 435, 773–778.

5. Sawaya, M. R.; Sambashivan, S.; Nelson, R.; Ivanova, M. I.;

Sievers, S. A.; Apostol, M. I.; Thompson, M. J.; Balbirnie, M.;

Wiltzius, J. J.; McFarlane, H. T.; Madsen, A. O.; Riekel, C.;

Eisenberg, D. Nature 2007, 447, 453–457.

6. Chen, H. F. Comput Biol Chem 2009, 33, 41–45.

7. Wang, J.; Tan, C.; Chen, H. F.; Luo, R. Biophys J 2008, 95, 5037–

5047.

8. DeMarco, M. L.; Daggett, V. Proc Natl Acad Sci USA 2004, 101,

2293–2298.

9. Wu, C.; Lei, H.; Duan, Y. J Am Chem Soc 2005, 127, 13530–

13537.

10. Gnanakaran, S.; Nussinov, R.; Garcia, A. E. J Am Chem Soc

2006, 128, 2158–2159.

11. Nguyen, H. D.; Hall, C. K. Proc Natl Acad Sci USA 2004, 101,

16180–16185.

12. Lipfert, J.; Franklin, J.; Wu, F.; Doniach, S. J Mol Biol 2005, 349,

648–658.

13. Zheng, J.; Jang, H.; Nussinov, R. Biochemistry 2008, 47, 2497–

2509.

14. Kent, A.; Jha, A. K.; Fitzgerald, J. E.; Freed, K. F. J Phys Chem B

2008, 112, 6175–6186.

15. Boucher, G.; Mousseau, N.; Derreumaux, P. Proteins 2006, 65,

877–888.

16. Derreumaux, P.; Mousseau, N. J Chem Phys 2007, 126, 025101.

17. Toschi, F.; Lugli, F.; Biscarini, F.; Zerbetto, F. J Phys Chem B

2009, 113, 369–376.

18. Masman, M. F.; Eisel, U. L.; Csizmadia, I. G.; Penke, B.; Enriz,

R. D.; Marrink, S. J.; Luiten, P. G. J Phys Chem B 2009, 113,

11710–11719.

19. Zheng, J.; Ma, B.; Chang, Y.; Nussinov, R. Front Biosci 2008, 13,

3919–3930.

20. Sgourakis, N. G.; Yan, Y.; McCallum, S. A.; Wang, C.; Garcia, A.

E. J Mol Biol 2007, 368, 1448–1457.

21. Kayed, R.; Head, E.; Thompson, J. L.; McIntire, T. M.; Milton,

S. C.; Cotman, C. W.; Glabe, C. G. Science 2003, 300, 486–489.

22. Emsley, P.; Cowtan, K. Acta Crystallogr D Biol Crystallogr 2004,

60, 2126–2132.

23. Canutescu, A. A.; Shelenkov, A. A.; Dunbrack, R. L.; Jr. Protein

Sci 2003, 12, 2001–2014.

24. Case, D. A.; Darden, T. A.; Cheatham, T. E.; Simmerling, C. L.,

III; Wang, J.; Duke, R. E.; Luo, R.; Merz, K. M.; Wang, B.; Pearl-

man, D. A.; Crowley, M.; Brozell, S.; Tsui, V.; Gohlke, H.; Mon-

gan, J.; Hornak, V.; Cui, G.; Beroza, P.; Schafmeister, C.; Cald-

well, J. W.; Ross, W. S.; Kollman, P. A. 2004.

25. Darden, T.; York, D.; and Pedersen, L. J Chem Phys 1993, 98,

10089–10092.

26. Lwin, T. Z.; Lu, Q.; Luo, R. Protein Sci 2006, 15, 2642–2655.

27. Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D.

A. J Comput Chem 2004, 25, 1157–1174.

28. Rychaert, J. P.; Ciccotti, G.; and Berendsen, H. J. C. Comput

Phys 1977, 23, 327–341.

29. Chen, H. F.; Luo, R. J Am Chem Soc 2007, 129, 2930–2937.

30. Chen, H. F. PLoS One 2009, 4, e6516.

31. Qin, F.; Chen, Y.; Li, Y. X.; Chen, H. F. J Chem Phys 2009, 131,

115103.

32. Chen, H. F. J Chem Theory Comput 2008, 4, 1360–1368.

FIGURE 15 Average number of native contact for oligomer.

The Stability of Cross-b Amyloid Fibril 585

Biopolymers

33. Zheng, J.; Ma, B.; Tsai, C. J.; Nussinov, R. Biophys J 2006, 91,

824–833.

34. Esposito, L.; Pedone, C.; Vitagliano, L. Proc Natl Acad Sci USA

2006, 103, 11533–11538.

35. Vitagliano, L.; Stanzione, F.; De Simone, A.; Esposito, L. Biopol-

ymers 2009, 91, 1161–1171.

36. Ivanova, M. I.; Sievers, S. A.; Sawaya, M. R.; Wall, J. S.; Eisen-

berg, D. Proc Natl Acad Sci USA 2009, 106, 18990–18995.

37. Gsponer, J.; Haberthur, U.; Caflisch, A. Proc Natl Acad Sci USA

2003, 100, 5154–5159.

38. Collins, S. R.; Douglass, A.; Vale, R. D.; Weissman, J. S. PL oS

Biol 2004, 2, e321.

Reviewing Editor: J. McCammon

586 Chen et al.

Biopolymers