Molecular Simulations for Anti-amyloidogenic Effect of Flavonoid Myricetin Exerted against Alzheimer’s 必 "Amyloid Fibrils Formation
Youngjin Choi： ThomasDonghyunKim,* Seung R. Paik,§ KarpjooJeong,#,* and Seunho Jung**
'Biochip Research Center,Hoseo University, Asan336-795, Korea
^Department ofBiochemistry, University of Chicago, Chicago, Illinois60637, U.S.A.
^School ofChemical and BiologicalEngineering, College of Engineering, SeoulNational University, Seoul151-744, Korea 姓College of Informationand Communication & Department ofAdvanced TechnologyFusion, Konkuk University,
Seoul143-701, Korea. E-mail:jeongk@konkuk.ac.kr
"Bio/MolecularInformaticsCenter & DepartmentofBioscience and Biotechnology, Konkuk University, Seoul 143-701, Korea E-mail: shjung@konkuk.ac.kr
Received June 27,2008
Comparativemolecular simulations wereperformed to establish molecularinteractionand inhibitory effect of flavonoid myricetin on formation of amyloid fibris. For computational comparison, the conformational stability of myricetin with amyloid ^-peptide (AQ) and ^-amyloid fibrils (fAQ) were tracedwith multiple moleculardynamics simulations (MD) using the CHARMMprogram from Monte Carlo docked structures.
Simulations showed that theinhibitionbymyricetin involves bindingof theflavonoid tofAQrather than AQ. Even in MDsimulationsover 5nsat 300K, myricetin/fAQcomplex remained stable in compactconformation for multiple trajectories. In contrast, myricetin/AQ complexmostly turned into the dissociatedconformation duringthe MDsimulationsat 300 K. ThesemultipleMD simulationsprovideatheoretical basis for thehigher inhibitory effect of myricetinonfibrillogenesis of fAQrelativeto AQ. Significant binding betweenmyricetin and fAQobserved from the computational simulations clearly reflectsthe previous experimental results in whichonly fAQhad boundtothemyricetin molecules.
KeyWords : Amyloid fibril,Conformationalanalysis, Flavonoid, Moleculardynamics simulations, Myricetin
Introduction
Amyloidosis is a clinical condition marked by the de positionoffibrousamyloids invarious organs and tissues of the body.1 The protein fibril formationmay be causativefor various fatal human disordersincludingAlzheimer'sdisease (AD).2 Common structural feature of amyloid fibrils has been characterized by critical cross Q-sheet core structure that perpetuatesalong thefibrillongaxis.3 Fibrillogenesis of the monomeric amyloid Q-peptide (AQ) found in AD generallyoccurs via nucleation-dependent polymerization.4 If supersaturated AQsolution forms a nucleus,the growth of fibril is accelerated to evolve into the Q-amyloid fibrils (fAQ).5 The structure-toxicity relationship study performed by Riek and co-workers indicates that fAQ contain an intrinsic toxicity that correlates with their morphologies as well as quantities.6 They showed that the neurotoxicitywas generally proportional to the size of fAQ. Therefore, it would be crucial to regulate the fAQ growth by chemical additivestodevelop potent anti-Alzheimer's diseasedrugs.
Recent studies for thefAQfibrilization revealedthat some flavonoid compounds are able to inhibit the fibril growth and destabilize the preformed fAQ.7 Ono et al. foundthat myricetin exhibited its anti-amyloidogenic effect via pre
ferentialbindingto the fAQ ratherthanthe AQmonomers.8 In order to develop anti-amyloidogenic drugs, it is man datory for structuralbiologists to acquire a detailed under standing of molecularinteractions between flavonoids and
fAQ.
In this report, comparative molecular simulations were performed for both myricetin/AQ and myricetin/fAQ in order to explain the experimentally observed inhibitory effect ofmyricetin for thefibrilgrowthof AQ.To reproduce real interactions between myricetin and amyloid peptide, Monte Carlo (MC)docking and molecular dynamics (MD) simulations were carried out for the molecularmodels of myricetin/AQ and myricetin/fAQ. Significant binding bet weenmyricetin and fAQ observedfrom the computational simulationsclearlyreflectsthepreviousexperimentalresults8 inwhich only fAQ bind to the myricetin molecules. From the simulation offibrillar fAQ with myricetin, it has been foundthat the myricetin interactions with theoutermost part offibrils are responsible for the experimentally observed inhibitoryeffecton thefibrillogenesisof fAQ.
Results andDiscussion
Differences in dockedconformationsandenergetics of the myricetin with A0 and fA0. The interaction energy profilesof each myricetin complex with AQand fAQduring MC docking simulations are shown inFigure 1.The inter- molecular energy between myricetin and AQ was -18.83 kcal/mol and those ofmyricetin and fAQwas -21.41 kcal/
mol during last 6x 105 MC trials. This means that the myricetin moleculeforms a more stable complex with fAQ thanthe monomeric AQ. This result seems to be consistent
o( 트
e 오
4얀 읃
山UOQ。
뜬 £ 드
Figure 1. Interaction energy profiles of myricetin with helical (red line) or fibrous (blue line) form of amyloid peptide during MC docking simulations.
with the experimental observationthat the flavonoid myri
cetinexerted its binding abilityonly tofA。 rather thanA。. Interaction energy profiles of the myricetin/A。 and the myricetin/fA。are shown in Figure 2. Myricetinwasdocked onto a region between o-helix and C-termini ofthe A。. Therewasno special binding site on the A。and the myri
cetin just clung to the surface of the peptide (Fig. 2).
Contrary to thecaseofmyricetin/A。, the myricetinformeda stable docked complex with the fA。. The U-shaped con
formation offA。 provided suitable binding site for a flat shaped myricetin molecule.5 The myricetin wasfully stuck intothe hydrophobic bindingpatch ofthe fA。. These two conformationswere further investigated byMDsimulations.
Figure 2. The 2-D structure of myricetin (a) and its docked con
formation with helical (b) or fibrous (c) form of amyloid peptide.
(E u) 엉 U S S ssewjoQ
ua。」외
(E u) 응 U 42으
asse>jo
ua。」외
(E u) 응 U 42S sse>joQ
」외
ua。
Time (ps) Time(ps)
1000 2000 3000 4000 5000 1000 2000 3000 4000 5000
Figure 3. Center of mass distances between myricetin and helical form of amyloid during multiple 5ns-MD simulations.
Complex stability ofmyricetin/A0 and myricetin/fA0 during multiple MD simulations. Multiple MDsimulations wereperformedfromthe lowest energyconformation of MC simulations.TheseMDsimulations for each myricetincom plex evidenced more stable conformation of myricetin/fA^ comparedto myricetin/A0. Figure 3 shows centerofmass distance between myricetin and A& during 5-ns of MD simulations for six trajectories. Four trajectories showed gradual dissociation ofthe myricetin from A。with some variations (Figure3a-d). Stable dockedconformations were only maintained forjust two trajectories with showinghigh- fluctuations (Figure 3e, f). Therefore, we can conclude that the interaction betweenmyricetin and A。maybe weak and unsuitable for keeping the complex stability during MD simulations. Compared with the myricetin/A。complex, the myricetin showed a more stable structure with the fA。 during multiple MD simulations.Finaldissociationof myri cetin from fA。 were observed for only two trajectories (Figure4c, d). Othersweretightlyassociated with each other during MD simulations, inwhich averaged center of mass distances were 0.83, 0.70, 0.53, and 0.52 nm, respectively (Figure4a, b, e, f).
Thedifferencesintheconformationalstabilityof the myri- cetin/A。 and the myricetin/fA。 were also supported by minimum distances between the myricetin andA。 or fA。 (Figure 5). After 5-ns of MD simulations, trajectory-aver aged minimum distance between myricetin and A。 was about 3 nm, while the distance between myricetinand fA。 was below 2 nm. This means that the atomic interactions between myricetinand fA。 were more stablethanthose of myricetin andA。. Figure 6is a solvent-accessible surface area change upon formation ofcomplexes between myri-
E드 eo ua Qs
ss 을
M-o」와 U 8 튿(
) oo us ass으
을
M-o J U£
<UQ E 드 OO C5 W5
ss 을
oJ
」와 QU O
Figure 4. Center of mass distances between myricetin and fibrous form of amyloid during multiple 5ns-MD simulations.
Time (ps)
Figure 5. Averaged minimum distance between myricetin and helical (red line) or fibrous (blue line) form of amyloid peptide during multiple MD simulations.
cetin and A0 or fA0. At early MD phase (0-1 ns), the surface area change frombindingof myricetin to fA0was -5.85 nm2. This indicates that the surface area was de creased by approximately 5.85 nm2 due to the binding of myricetin and fA0. The value was increasedto -3.70 nm2 duringlateMDsimulation phase(4-5 ns). For themyricetin/
A。,thesurface area changewas-4.40 nm2 during earlyMD phase and the valuewasincreased to -1.33 nm2 during late MD phase. The surface area change ofmyricetin/fA。was much lower than that of myricetin/A。at any MD time phase. These results suggesttighter binding of myricetin to the fA。 in comparison to A。. The ratio of surface area changes during late to early MD phase was 0.63 for the myricetin/fA。and 0.30 for the myricetin/A。. This means
E( u) SO6 듀
x:o ra안
v
<D。 요
』r・s
Figure 6. Averaged surface area change upon complexation of myricetin with helical (red line) or fibrous (blue line) form of amyloid peptide during multiple MD simulations.
that thesurface area decreaseofmyricetin/fA。 was two-fold higher than that of myricetin/A。 during the time scale of MD simulations. Therefore,bindingstability ofmyricetinto the fA。 is more tolerant to dynamic fluctuation of the molecular complex during the MD simulation time. The strongbinding ability ofmyricetintofA。in comparisonto A。 could provide a molecular understanding for experi mentallyobservedinhibitory effect ofmyricetinonamyloid fibrilliogenesis.
Figure 7. Destabilization of fA。fibrils with myricetin (a) and stable fA。fibrils (b) during 10ns MD simulations.
Destabilization of fA0 by myricetin during the MD simulations. Myricetin can destabilize existing fibrils as well as have inhibitory effect on formation of emerging fibrils. Figure 7 is thetimecourseof thefibrildestabilization by myricetin during 10-ns of MD simulations. The fibrils were modeled as five sets offA0monomers. After 280 ps, the myricetin made an interactionwith the outermostfA0 sheaf. After 550 ps,themyricetin had bound toC-termini of thefA& sheaf viahydrophobic interaction of aromatic rings.
Thebinding interaction was intensifiedintothe core cavity oftheoutermostfA0sheafafter 1.85 ns ofMD simulation.
In that time, about half of the domain of fA& sheafwas segregated from remaining fA° fibrils. The level ofinter action between fibrils andfA0partwas minimized by the bound myricetin. After 2.69 ns, the myricetin moved out of Mtermini of fA0 sheaf and showed relatively decreased interaction with thefA&sheaf. Finally, themyricetinmoved tothe next part of fA& sheaf todestabilizeremainingpart of fA0fibrilsafter 8.11 ns. From the MD simulationsnapshots in the Figure 7, it is evidenced that the myricetin exert an anti-amyloidogenic effect bydestabilizingthefA0fibrils as well as inhibitingthe formation of fibrils frommonomeric peptides.
For thequantitativecomparisonbetween with andwithout myricetinconditions, Ca atomic distance was analyzed for each fA0 sheafin the fibrils. Figure 8 is time course of interatomicdistance betweenfA0sheaf1-2, 2-3,3-4, and 4
5 for centric Ca atomof eachGly13residue.During theMD simulations, it was foundthe interatomic distancesof outer fA0sheaf 1-2 or 4-5 were dramatically increasedby myri cetin(Figure 8a, d).However thedistances for theinner fAg sheaf were not changedduringthe MDsimulations for both with and without myricetin conditions (Figure 8b, c). It seems that myricetin molecule exert anti-amyloidogenic effect for outerpartoffAgfibrils by perturbing these inter shelf vdw interactions. Figure 9 is intermolecular electro
static andvdw energybetweenoutermostfAg sheaf 1 and 2.
0 2000 4000 6000 8000 10000 2000 4000 6000 8000 10C
Time(ps) Time (ps)
Figure 8. Interatomic Ca distances between fAg sheaf 1 and 2 (a), 2 and 3 (b), 3 and 4 (c), 4 and 5. (d). Red line indicates fAg fibrils with myricetin and blue line indicates fAg fibrils alone.
0 2000 4000 6000 8000 10000
Time (ps)
Figure 9. Intermolecular electrostatic (a) and vdw energy (b) between outermost fAg sheaf. Red line indicates fAg fibrils with myricetin and blue line indicates fAg fibrils alone.
During MD simulations, the intermolecular electrostatic energies were controversial to each fAg fibril with and withoutmyricetin (Figure9a). However,intermolecularvdw energieswere discriminatedbetweenthefAgfibrils (Figure 9b). The vdw energy of fAgfibrils was lowered than the case of fAgfibrils with myricetin system during the simu lation times. Intermolecular vdw energywas-91.2kcal/mol for thefAgfibrils without myricetin and -71.3 kcal/molfor the fAg fibrils with myricetin, respectively. That means myricetinmoleculedestabilizetheintermolecular vdw inter
action between outer part offAgsheaf ratherthaninterior part of fAg fibrils.
Methods of Computation
Construction of the molecular models andprotocolof MC docking simulations. The starting configuration of each Agand fAgfor MCsimulationswas obtained from the Protein Data Bank (PDB). The PDB ID’s for Ag and fAg were 1zoqand 2beg, respectively. To obtainrepresentative conformationsof the3Dcoordinateensemble,theDiscovery Studio/Builder module(version 1.7, AccelrysSoftware Inc.) was usedas a molecular editor. Theatomic coordinatesand atomic partial charges of the myricetin were obtained by utilizing ab initio calculations with 6-311G basis set at Hartree-Fork (HF) level. All simulations were performed using a general molecular modeling program, CHARMM (version30b1), with a parm22 all-atomforcefield.9 TheMC dockingsimulationswereperformedusing the “MC” module
of CHARMM. The short-range non-bonding interactions were truncated witha 13-A cutoff. Animplicitsolvent water model was usedwith adistance-dependent dielectric con
stant. The dockingprocess was assumedto be a 1:1 inter action between each amyloid peptide and myricetin during theMC runs. Theinitial configuration of each amyloidand myricetin molecule was positioned arbitrarily within a neighboring distance. Trials to a new configuration were accomplished by changingeach move set ofa guestmole cule. The MCmove set for flexible docking wascomposed ofrigid translations, rigid rotations, and rotations offreely rotatabledihedralanglesof the bothmyricetinand amyloids.
A single step consists of picking a random conformer, makinga random move, minimizing the energy of a new conformer, and then checkingthe energy witha Metropolis criterion.10 The MC-accepted structures were saved every 4000steps for 8,000,000trials. Multiple 10-MCtrajectories were generated with different randomnumber seedto con
vergence of the simulation. These MC processes produced convergent docked structures for each monomeric amyloid with myricetin.
Multiple molecular dynamics simulations of the each myricetin/A0 andmyricetin/fA0 complexes. The starting configurations of each myricetin/A^ and myricetin/fA0 complex for theMD simulationsin explicit water weretaken from the MC-docked conformations11 with the lowest- energyvalue (Figure2). The geometriesof these molecular models were fully optimized before MD runs. A TIP3P three-site rigid water model was used to solvate the com- plexes.12 Watermoleculeswereremovediftheywerecloser than 2.8 A to any heavy atoms ofthe complexes. In sum mary,eachsystemwasconstructedusingperiodicboundary conditions witha cubic box ofdimensions 50 A 乂 50 A 乂 50 A, consistingofmyricetin/A^ complex and 4,320 water, or myricetin/fA0complex and 4,293 watermolecules. The systemwasminimized by 2,000stepsofconjugate gradient, followedby AdoptedBasis Newton-Raphson until the root
mean-squaregradient was less than0.001 kcal/mol. TheMD simulations were performed using the CHARMM 30b1 program in theisothermal-isobaricensembleat 300 K (P= 1 bar, T = 300 K). The particle mesh Ewald summation method13 was used to treat the long-range electrostatic interactions.14 The bond lengths of all moleculeswere con
strained with the SHAKEalgorithm.15 Thetimestep was 2.0 fs, and thenon-bondedpairlistwasupdated every 25 steps.
The short-range non-bonded interactions were truncated with a 13-A cutoff. The temperature and pressure of the systemwere regulated using the Langevinpiston method in conjunction with Hoover’s thermostat.16 The system was gradually heated to each targetedtemperature for 50 ps and equilibrated at this temperature. The production MD tra jectorywith one snapshot per2 ps was collected for 5,000 ps. Total six-independent MD simulations were performed with different random number seed to cover wide confor mational sampling. Forthe simulation ofwholefibrils with
myricetin, molecular model was constructed with the cubic box of dimensions70 A 乂 70 A 乂 70 A, consisting of four myricetins, five-fA^fibril, and 10,216water molecules. The MD simulationswerecarried out over thecourseof10-nsto observedestabilizationof fAg fibril bymyricetin.
Conclusions
Comparison ofthe dynamics ofmyricetin/Agandmyri- cetin/fAg complex during molecular simulations demon
strate thatmyriceyin disturb upon Alzheimer’s g-amyloid fibrillogenesis by preferential bindinginto the hydrophobic patch of fAg.MultipleMDsimulations of myricetin/Agand myricetin/fAg provided the important theoretical insight intothe experimentally determinedanti-amyloidogenic effect of flavonoid compound, suggesting that myricetin could bind tofibrousformof amyloid peptides,thereby decreasing favorable interaction between monomeric g-sheetsof amyloid fibrils. Based on the present study, we conclude that com putational methodology canmake great contribution toward anti-anyloidogenicdrugdesign usingnatural flavonoidcom pounds.
Acknowledgments. This study was supported by a grant of thee-ScienceProject of KISTI (Korea Instituteof Science and TechnologyInformation)inMOST(Ministryof Science and Technology) and partly supported by a grant of the KoreaResearchFoundation(KRF-2006-005-J03402).
References
1. Pepys, M. B. Annu. Rev. Med. 2006, 57,223.
2. Selkoe, D. J. Nature 2003, 426, 900.
3. Sunde, M.; Serpell, L. C.; Bartlam, M. J. Mol. Biol 1997, 273,729.
4. Harper, J. D.; Lansbury, P. T. Jr. Annu. Rev. Biochem. 1997, 66, 385.
5. Rochet, J. C.; Lansbury, P. T. Jr. Curr. Opin. Struct. Biol. 2000, 10, 60.
6. Luhrs, T.; Ritter, C.; Adrian, M.; Riek-Loher, D.; Bohrmann, B.;
Dobeli, H.; Schubert, D.; Riek, R. Proc. Nat. Acad. Sci. USA 2005, 29, 17342.
7. Ono, K.; Yoshiike, Y.; Takashima, A.; Hasegawa, K.; Naiki, H.;
Yamada, M. J. Neurochem. 2003, 87, 172.
8. Hirohata, M.; Hasegawa, K.; Yasuhara, S. T.; Ohhashi, Y.;
Ookoshi, T.; Ono, K.; Yamada, M.; Naiki, H. Biochemistry 2007, 46,1888.
9. Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J.;
Swaminathan, S.; Karplus, M. J. Comput. Chem. 1983, 4, 187.
10. Metropolis, N.; Rosenbluth, A. W.; Rosenbluth, M. N.; Teller, A.
H.; Teller, E. J. Chem. Phys. 1953, 21,1087.
11. Choi, Y.; Park, S.; Jeong, K.; Jung, S. Bull. Korean Chem. Soc.
2007, 28, 1811.
12. Jorgensen, W. L. J. Chem. Phys. 1983, 79, 926.
13. Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089.
14. Lee, S. H. Bull. Korean Chem. Soc. 2006, 27, 1154.
15. Ryckaert, J. P.; Ciccotti, G; Berendsen, H. J. C. J. Comput. Phys.
1977, 23, 327.
16. Feller, S. E.; Zhang, Y.; Pastor, R. W.; Brooks, B. R. J. Chem.
Phys. 1995, 103, 4613.
