Originalarticle
Pharmacophorebasedvirtualscreening,moleculardockingstudiestodesign
potentheatshockprotein90inhibitors
SugunadeviSakkiah,SundarapandianThangapandian
1
,ShaliniJohn
1
,KeunWooLee*
DivisionofAppliedLife Science(BK21Program),PlantMolecularBiologyandBiotechnologyResearchCenter(PMBBRC),DepartmentofChemistryandResearchInstitute,
GyeongsangNationalUniversity(GNU),900Gazwa-dong,Jinju660-701,RepublicofKorea
a r r ti c c l e e i i nf o
Article history:
Received7February2011
Receivedinrevisedform
30March2011
Accepted4April2011
Availableonline15April2011
Keywords:
Heat shockprotein90
Commonfeaturehypothesis
Structure-basedpharmacophore
LigandScout
Virtualscreening
GOLD
a b b s s tr r a a c t
The identificationofimportant chemicalfeatures ofHeat ShockProtein90(HSP90)inhibitorswillbe
helpfultodiscoverthepotentcandidatetoinhibittheHSP90activity.ThebesthypothesisfromHip-Hop,
Hypo1,onehydrogenbonddonor(HBD),twohydrogenbondacceptors(HBA),andtwohydrophobic(H)
andstructure-basedhypothesis,SB_Hypo1,oneHBA,oneHBDandfourHfeatures,weregeneratedusing
DiscoveryStudio andLigandScout,respectively.Testanddecoysetswereusedtocorroboratethebest
hypothesesandthevalidatedhypotheseswereusedtoscreenthechemicaldatabases.Subsequently,the
screenedcompoundswerefilteredbyapplyingtheruleoffive,ADMETandmoleculardocking.Finally,
fourcompoundswereobtainedasnovelleadstoinhibittheHSP90activity.
2011ElsevierMassonSAS.Allrightsreserved.
1. Introduction
Molecularchaperones,mainly responsiblefor protein folding
andits 3D conformation inthecell,alsoplaysacrucialroles in
balancing the degradation and synthesis of f many y proteins [1].
Among manychaperones,Heat Shock Protein(HSP) is themost
importantbecauseitprotectsthecellswhenstressedelevatedby
temperature[2].The90-kDaHeatShockProtein(HSP90),anATP-
dependentmolecularchaperon,isoneofthevitaltargetsinche-
mogenomic approaches [3] as s well l as highly y conserved from
bacteriatohuman[4e6].HumanHSP90familyincludes17genes
whichwereclassifiedinto4classes:HSP90AA,HSP90AB,HSP90B,
andTRAP[7].HSP90playsamajorroleinfoldingandmaturationof
variousclientproteinssuchassteroidreceptors,p53,ErbB2,Src,Abl,
Raf,Aktandcyclin-dependentserinekinases[2].HSP90consistsof
three distinct t domains [8]: N-terminal, middle e and C-terminal
domains.TheN-andC-terminaldomains aremainly focusedin
many pharmaceutical l companies s to design the e new inhibitors.
N-terminal, ATP binding domain interacts s with many y synthetic
inhibitorsaswellasthenaturalproductsandtheC-terminaldomain
playsanimportantroleinhomo-dimerisationprocess[9].Inhibition
ofHSP90leadstoderegulationofmanycrucialpathwayssuchas(a)
self-sufficiencyingrowthsignals(b)tissuesinvasion/metastasis(c)
insensitivityto antigrowth h signals (d) sustainedangiogenesis(e)
evasionofapoptosisand(f)limitlessreplicativepotentialwhichare
responsibleforthecancercell’ssurvival[10]aswellasitinvolvesin
thedestabilizationanddegradationofoncogenicclientproteinsto
stop the cancer cell l growth [5]. Blocking the ATPase activity of
HSP90willbeanimportantpharmacologicalplatforminanticancer
therapy[11].Themajoritiesofinhibitorsdevelopedsofarinhibitthe
HSP90bybindinginATP-bindingpocketwhichdrivesthechap-
eron’scycleanddirectinteractionsinducetheactiveconformation
[12].Geldanamycinanditsderivativesarethenaturalproductsto
inhibittheATPaseactivitybybindingintheN-terminaldomainof
HSP90 [13] and reported as s a potent t cancer drugs. . Hence, the
N-terminal domain n is a substantial l target in n structure biology
approachesthat facilitatethestructure-basedinhibitoroptimiza-
tion[14].InhibitorsofHSP90areclassifiedintoseveralclassesbased
ondistinct modesofinhibitionlike (a) blockadeofATP binding,
(b)disruptionofco-chaperon/HSP90interactions,(c)antagonism
of client/HSP90 0 associations s and d (d) ) interference with post-
translationalmodificationsofHSP90[15].Alltheaboveinhibition
makesHSP90asapotentialtargetformanydiseasesrangingfrom
the disruption of multiple signaling pathways s associated d with
cancer[16,17] andalso intheclearanceof f proteinaggregatesin
neurodegenerativediseases[18].Thus,discoveryofsmallmolecule
inhibitorsofHSP90remainsanactivefieldofcancerresearch[19].
* Correspondingauthor.Tel.:þ82557516276;fax:þ82557527062.
E-mailaddress:kwlee@gnu.ac.kr(K.W.Lee).
1
Contributedequallyassecondauthor.
ContentslistsavailableatScienceDirect
EuropeanJournalofMedicinalChemistry
journal homepage: http://www.elsevier.com/locate/ejmech
0223-5234/$eseefrontmatter2011ElsevierMassonSAS.Allrightsreserved.
doi:10.1016/j.ejmech.2011.04.018
EuropeanJournalofMedicinalChemistry46(2011)2937e2947
Convert pdf to url - Convert PDF to html files in C#.net, ASP.NET MVC, WinForms, WPF application
How to Convert PDF to HTML Webpage with C# PDF Conversion SDK
convert pdf to url online; convert pdf to html file
Convert pdf to url - VB.NET PDF Convert to HTML SDK: Convert PDF to html files in vb.net, ASP.NET MVC, WinForms, WPF application
PDF to HTML Webpage Converter SDK for VB.NET PDF to HTML Conversion
change pdf to html format; embed pdf into html
The computationaltechniques helpindrug designprocessto
dramaticallywidenthechemicalspaceandreducethenumberof
candidates for experimental l validation [20]. Pharmacophore
modelingandvirtualscreeningisaninexpensiveandfastalterna-
tivepowerfultooltoidentifythepotentialleadforvarioustargets
[17,21].Henceinthisstudy,ligandandstructure-basedpharmaco-
phore models were generatedandvalidated using the test t and
decoysets.Inawidevarietyofapplications,virtualscreeningwas
found to o be successful method d especially when combined with
moleculardockingstudies.Hencethepredictedhypotheseshave
beenemployedinvirtualscreeningtoidentifythepotentleadfrom
variousdatabasessuchasMaybridge[22]andChembridge[23].The
retrievedhitmolecules aresortedoutby applying several filters
suchasmaximumfitvalueofthebestpharmacophoremodelsfrom
ligand-based,structure-basedhypothesesandsatisfytheLipinski’s
Rule of five, , ADME properties and subsequently subjected to
moleculardocking process.The molecular docking studieswere
performed using g two different programs, , LigandFit/Discovery
Studiov2.5(DS,Accelrys)[24]andGeneticOptimizationforLigand
Docking (GOLD, Cambridge CrystallographicDataCenter) [25]to
findthesuitableorientationofleadsintheactivesiteofHSP90.
2. Resultsanddiscussions
2.1. LigandbasedpharmacophoremodelingusingDS
Thequalitativetoptenhypothesesweregeneratedbasedonthe
trainingset molecules (Fig.1) using Common n Feature Pharmaco-
phoreGeneration/DStoidentifythecommonfeaturesnecessaryto
inhibittheHSP90function.Directandpartialhitmaskvalueof‘1’
and ‘0’ ’ for hypothesis indicated that t the molecules present in
dataset are well mapped d to all l the e chemical features s in n the
hypothesis andthere is no partial mapping or missing features
specifiedinTable1.TheClusteranalysis(Fig.2)wasusedtoevaluate
and categorized the difference between n the compositions of
hypothesischemicalfeaturesandlocations.Thisprocessgivesthe
numberofclusters,inthispapertwoclustersareselectedbasedon
the features s similarities. . Cluster I contains six hypotheses with
thecombination of three chemical l featureslike hydrogen n bond
acceptor(HBA),hydrogenbonddonor(HBD)andhydrophobic(H)
andClusterIIcontainsfourhypothesesthatcontainstwochemical
features(HBAandH).FocusingonabovethetwoClusterstherewas
onlyonechemicalfeature(HBAorHBD)wasdifferentfromClusterI
andII.Hence,initiallyonepharmacophoremodelwasselectedas
a good hypothesis from each cluster with high h ranking g score:
ClusterI-Hypo1:twoHBA,twoHBD,oneHandClusterII-Hypo2:
twoHBA andthreeH.Amongthese e two hypotheses,Hypo1has
the highest ranking score(54.562) when compared withHypo2
(53.562).HenceHypo1wasselectedasabestqualitativepharma-
cophore models s based on n the e chemical features similarities,
rankingscoreandtheircorrespondinggeometricconstrains.Hypo1
shows the e good alignment with the training set molecules has
showninFig.3.
2.2. Structure-basedpharmacophoremodeling
The40co-crystalstructuresfromProteinDataBank(PDB,www.
rcsb.org) has s been checked for a better r understanding g of the
specificity
andpharmacophorerequirementsofHSP90activesite
(Table 2). The e binding site was s characterized d by several l direct
Fig.1. StructureoffivecompoundsusedastrainingsetinHip-Hop,PDBID,IC
50
valuesandtheircorrespondingco-crystalnamewasgiveninbracket.
Table1
DetailsofthetoptenhypothesesgeneratedusingHip-HopforHSP90.
HypothesisName Featuresa
RankDirecthitPartialhitMax.fit
Hypo1
H,H,HBD,HBA,HBA 54.56 6 11111
00000
5
Hypo2
H,H,HBA,HBA,HBA 53.56 6 11111
00000
5
Hypo3
H,H,HBD,HBA,HBA 53.30 0 11111
00000
5
Hypo4
H,H,HBD,HBA,HBA 52.13 3 11111
00000
5
Hypo5
H,H,HBA,HBA,HBA 52.30 0 11111
00000
5
Hypo6
H,H,HBA,HBA,HBA 52.13 3 11111
00000
5
Hypo7
H,H,HBD,HBA,HBA 52.03 3 11111
00000
5
Hypo8
H,H,HBD,HBA,HBA 51.90 0 11111
00000
5
Hypo9
H,H,HBD,HBA,HBA 51.54 4 11111
00000
5
Hypo10
H,H,HBA,HBA,HBA 51.03 3 11111
00000
5
a
HBA ¼ Hydrogen n Bond Acceptor; HBD ¼ ¼ Hydrogen n Bond Donor; ; H H ¼
Hydrophobic.
b
Highertherankingscore,lessertheprobabilityofchancecorrelation.Thebest
hypothesisshowsthehighestvalue.
c
Direct Hit, , Partial Hit t indicates s whether (1) ) or (0) ) a a training g set molecule
mappedeveryfeatureofthehypothesisandmappedtoallbut onefeatureinthe
hypothesis.Thenumbersfrom(right toleft) correspondtothecompounds(from
toptobottom).
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2938
C# PDF url edit Library: insert, remove PDF links in C#.net, ASP.
C#.NET PDF SDK - Edit PDF Hyperlink (url) in C#.NET. Enable users to copy and paste PDF link. Help to extract and search url in PDF file.
adding pdf to html page; how to change pdf to html format
VB.NET PDF url edit library: insert, remove PDF links in vb.net
VB.NET PDF - Edit PDF Hyperlink (url) in VB.NET. Make PDF link open in a new window, blank page or tab. Edit PDF url in preview without adobe PDF reader control.
online pdf to html converter; embed pdf into web page
interactionssuchashydrogenbondinteractionandliphophilicside
chainscomplement whichreflect the hydrophobic nature ofthe
HSP90activesite.Theactivesiteofthefiveco-crystal structures
were visualized and analyzed for proteineligand interactions
(Fig.4)[26]whichwasthemostimportantstartingpointforthe
generationofstructure-basedpharmacophore.
2.2.1. Generationofstructure-basedpharmacophoremodelsusing
LigandScout
LigandScout present t the interactions between n protein n and
ligandaswellaswithsomeexcludedvolumespherescorrespond-
ingtotheir3Dstructuresofprotein.Inthisstudy,fivedifferent3D
structureofHSP90boundwithitsinhibitorssuchas3BM9[27],
3BMY[28],3FT8[27],3HHU[29],3EKR[30]wereselectedasinput
forstructure-basedpharmacophoregeneration.Thegeneratedfive
pharmacophore models withits geometrical constrain andtheir
active sites s were e represented in Fig. 5. For 3BM9 complex, the
generatedpharmacophorecontainstwoHBApointedtowardsthe
Lys58,Thr184,oneHBD chemical featurewhichpointedtowards
Asp93,twoHgroupsand12excludedvolumespheres.Fourfeatures
hypothesiswasgeneratedfrom3BMYcomplexwhichcomposedof
oneHBA,oneHBD,andtwoH groups with10excludedvolume
spheres. The HBD and HBA A groups pointed d towards s the Asp93
andThr184,respectively.3FT8complexesconsistsofsix features
hypothesisincludesoneHBAwhichpointedtowardsThr184,one
Fig.2. PhylogenicClusteranalysisforligand-basedcommonfeaturehypothesis.
Fig.3. TrainingsetcompoundsshowngoodalignmentwithHypo1.Greencolorindicateshydrogenbondacceptor(HBA);cyanindicateshydrophobic(H)andmagentaindicates
hydrogenbonddonor(HBD)(forinterpretation ofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
Table2
AnalysesofcriticalaminoacidsforHSP90inhibitionfrom40co-crystalstructures
depositedinproteindatabank.
PDBID
Asn51 Asp93 Lys58 Asp93 Gly97 Phe138 Gly97 Thr184
3BM9[27]
O
O
O
O
O
O
3BMY[27] O
O
O
O
O
O
3EKR[30] O
O
O
O
O
O
2BYI[41]
O
O
O
O
O
O
O
O
2BYH[41]
O
O
O
O
O
O
O
2VCJ[42]
O
O
O
O
O
O
2VCI[42]
O
O
O
O
O
O
O
O
3D0B[43]
O
O
O
O
1YC3[44] O
O
O
O
O
O
O
O
1YET[45] O
O
O
O
O
O
2QG0[46] O
O
O
O
O
O
O
2QFO[46] O
O
O
O
O
O
2QG2[46] O
O
O
O
O
2QF6[46]
O
O
O
O
O
1OSF[47] O
O
O
O
O
O
2CDD[48]
O
O
O
O
O
O
O
O
2CCS[49]
O
O
O
O
O
O
O
2CCU[49]
O
O
O
O
O
2CCT[49] O
O
O
O
O
O
O
1UY6[50] O
O
O
O
O
O
1UY7[50] O
O
O
O
O
1UYM[50]
O
O
O
O
1UYI[50]
O
O
O
O
O
O
O
1UYK[50]
O
O
O
O
O
O
1UYH[50] O
O
O
O
O
O
O
1UYG[50]
O
O
O
O
O
O
O
1UYF[50] O
O
O
O
O
O
O
1UYE[50]
O
O
O
O
O
1UY9[50]
O
O
O
O
O
1UYC[50]
O
O
O
O
O
O
1UY8[50]
O
O
O
O
O
1UYD[50]
O
O
O
O
O
2UWD[51]
O
O
O
O
O
2BSM[52] O
O
O
O
O
O
O
O
2BT0[52]
O
O
O
O
O
O
2FWY[53] O
O
O
O
O
O
2H55[53] O
O
O
O
O
O
O
2FWZ[53] O
O
O
O
O
O
2JJC[54]
O
O
O
O
2BZ5[14]
O
O
O
O
O
O
O
O
3EKO[55]
O
O
O
O
O
O
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2939
C#: How to Open a File from a URL (HTTP, FTP) in HTML5 Viewer
PDF to svg, C#.NET convert PDF to text, C#.NET convert PDF to images VB.NET How-to, VB.NET PDF, VB.NET Word, VB.NET Excel C# HTML5 Viewer: Open a File from a URL.
convert pdf fillable form to html; convert pdf to url link
C#: How to Add HTML5 Document Viewer Control to Your Web Page
new RECommand("convert")); _tabFile.addCommand new UserCommand("pdf"); _userCmdDemoPdf. addCSS new customStyle({ background: "url('RasterEdge_Resource_Files/images
best website to convert pdf to word; convert url pdf to word
HBDpointedtowardsAsp93,andfourHgroupswith16excluded
volumespheres.Sixfeatureshypothesiswasgeneratedfrom3HHU
complexthatincludestwoHBA,oneHBD,andthreeHgroupswith
14excludedvolumespheres.ThetwoHBAgroupspointedtowards
theIle110,Gly135andHBDpointedtowardsAsp92.3EKRcomplex
producedsevenfeatureshypothesisconsistsoftwoHBA,oneHBD,
andfourHgroupswith11excludedvolumespheres.ThetwoHBA
andHBDgroups pointedtowardstheThr184,Asn51and Asp93,
respectively.
Comparingtheabovefivepharmacophoremodels,oneHBDand
HBAfromallthemodelswerepointedtowardsAsp93andThr184
whichplaysamajorroleinHSP90activity,respectively.Hence,HBD
andHBAfeaturesareconsideredasimportantchemicalfeaturesto
discoverthenovelHSP90inhibitors.Thedynamicstructure-based
pharmacophorewasgeneratedbysuperimposingthefivestructure-
based hypothesis and the e overlapped d chemical l features were
removed.Finallyeightfeaturesdynamicsstructure-basedhypoth-
esis(Fig.6)wasproducedthatconsistsof3HBA,1HBDand4H
features.Duetothehighnumberofchemicalfeaturesthedynamics
structure-basedmodel(8featureshypothesis)hasbeensplitinto
twodifferenthypothesiswhichcontainsixfeatureseach,SB_Hypo1
andSB_Hypo2containoneHBA,oneHBD,fourHfeaturesandthree
HBA,oneHBD,twoHgroups,respectively(Fig.7).
2.3. Validationofligandandstructure-basedpharmacophoremodels
The best t hypothesis s from m Hip-Hop (Hypo1) and dynamics
structure-basedmodels(SB_Hypo1andSB_Hypo2)werevalidated
usingtwodifferentmethods:(i)testset,tovalidatehowwellour
selectedhypothesis pick theactivefrominactivecompounds (ii)
decoyset,toevaluatethepredictabilityoftheselectedhypothesis
usingstatisticalparameters.Thetestsetcontains30structurally
diversemoleculesthatwasclassifiedintothreedifferentcategories
basedonitsIC
50
values(explainedinMethodsandmaterials),i.e.,
highly active,moderatelyactiveandlowactivecompounds. The
qualitative hypothesis (Hypo1) and dynamics structure-based
hypothesis(SB_Hypo1)arescreenedallthemoleculespresentin
the test set but it produces a a high h fit t value e forthe active and
moderatelyactivecompoundswhencomparedwiththelowactive
compounds. In case of dynamics structure-based d hypothesis,
SB_Hypo2,selectfewactiveandmoderatecompoundsbutfailsto
choosethelowactivecompounds aswell assomeofthe active
compounds(Table3).Comparingthestructure-basedhypothesis,
the hit compounds s from SB_Hypo1 1 comprises s all l compounds
screenedbySB_Hypo2hypothesis.Hence,SB_Hypo1wasselected
as a best t structure-based d pharmacophore model for the HSP90
receptor.
2.4. Enrichmentofdatabase
Indrugdiscoveryprocess,thebesthypothesisshouldidentify
the active compoundsfrominactive.Decoyset,comprises of 25
known good inhibitors and 1975 decoy y molecules s of HSP90
inhibitors,wasusedto validatewhetherthehypotheses(Hypo1,
SB_Hypo1) could d able e to discriminate the active e from m inactive
compounds ornot.Thedatabasescreeningwasperformedusing
Fig.4. ActivesiteoffivecrystalstructuresofHSP90boundwithitsinhibitors.
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2940
VB.NET Image: VB Code to Download and Save Image from Web URL
Apart from image downloading from web URL, RasterEdge .NET Imaging SDK still dedicated to provide powerful & profession imaging controls, PDF document, image
convert pdf to web page; convert pdf form to html form
C# HTML5 PDF Viewer SDK to view PDF document online in C#.NET
NET extract PDF pages, VB.NET comment annotate PDF, VB.NET delete PDF pages, VB.NET convert PDF to SVG. Able to load PDF document from file formats and url.
how to add pdf to website; converting pdf into html
LigandPharmacophore Mappingmodule.Theresult was analyzed
using asetofparameters suchas hitlist(H
t
),numberofactive
percentofyields(%Y),percentratio ofactivesinthehitlist(%A),
enrichmentfactor(E),falsenegatives,falsepositives,andgoodness
of hit t score e (GH) ) (Table 4[31]. Hypo1 1 and d SB_Hypo1 were
successfullyretrieved100%and88%ofactivecompoundsfromthe
decoy set, , respectively. In n addition Hypo1 1 and d SB_Hypo1 1 have
shownanenrichmentfactorof4.1and3.2aswellastheGHscoreof
0.89and0.71,respectively,whichindicatesthatthequalityofthe
pharmacophoremodelsareacceptable.Byoverallvalidations,we
canassurethatboththehypotheseswereabletopredictmostof
thecompoundsinthesameorderofmagnitudeanditcanableto
discriminate the e active e inhibitors s form inactive e or r low active
compounds. Hence, , we e suggested that Hypo1 and SB_Hypo1
hypothesesaregoodtoselectordiscriminatedthesuitableinhib-
itorsofHSP90.
2.5. Databasescreening
Fromtheabovevalidationmethods,itwasprovedthatHypo1
andSB_Hypo1havesuperiorabilitytodistinguishtheactiveand
inactiveinhibitors ofHSP90.Therepresentative pharmacophore,
Fig.5. Structure-basedhypothesesweresuperimposedontheactivesiteof3DstructureofHSP90.Greencolorindicateshydrogenbondacceptor(HBA);cyanindicateshydrophobic
(H)andmagentaindicateshydrogenbonddonor(HBD).Redarrowrepresentsthehydrogenbondacceptor(HBA);Greenarrowrepresentsthehydrogenbonddonor(HBD);Brown
colorindicateshydrophobic(H)(forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2941
VB.NET PDF Library SDK to view, edit, convert, process PDF file
Able to render and convert PDF document to/from supported document package offers robust APIs for editing PDF document hyperlink (url), which provide
convert pdf into html online; pdf to html
C# Image: How to Download Image from URL in C# Project with .NET
convert image to byte, and how to convert an image jpg, or bmp image from a URL to your to provide powerful & profession imaging controls, PDF document, tiff
convert pdf to html for online; convert pdf to url
Hypo1and SB_Hypo1, have used as a 3D queries to o screen n the
chemical databases like Maybridge and d Chembridge consists s of
60,000and50,000compoundsrespectively.Theinitialscreeningof
Hypo1retrieved16,236fromMaybridge,12,923fromChembridge
and the retrieved hits s were e narrow down to 1100 0 and d 543
compoundsbyapplyingthemaximumfitvalueof4,respectively.In
caseofSB_Hypo1,15,187and12,814compoundswereselectedand
sortedoutbyapplyingacutoffvaluemorethan4(maximumfit
value)635fromMaybridgeand216moleculesfromChembridge,
respectively.Moreover,ADMEandLipinski’sruleoffivewereused
toeliminatethenon-druglikecompoundsfromthehitmolecules
fromMaybridgeandChembridgecorrespondedto166,166and36,
55,fromHypo1andSB_Hypo1,respectively.Totally,17compounds
wereselectedfromMaybridgeandChembridgewhichsatisfiedall
the chemical featurespresent in Hypo1and19compounds was
sortedoutfromMaybridgeandChembridgeusingSB_Hypo1.These
17and19compoundswereusedforfurtheranalysislikemolecular
docking studies to avoidthe false positive hits from the virtual
Fig. 6. . Eightfeaturehypothesis s afterremovalofoverlappedchemicalfeaturesfrom
structure-basedmodels.Greencolorindicates hydrogen bondacceptor(HBA); cyan
indicateshydrophobic (H) and magenta indicateshydrogen bond donor(HBD). Red
arrow represents s the hydrogen bond acceptor (HBA); Green arrow represents the
hydrogen bond donor(HBD);Browncolorindicateshydrophobic(H)(forinterpreta-
tionofthereferencestocolourinthisfigurelegend,thereaderisreferredtotheweb
versionofthisarticle.)
Fig.7. Sixfeatureshypothesis:SB_Hypo1andSB_Hypo2anditsgeometricconstrains.Greencolorindicateshydrogenbondacceptor(HBA);cyanindicateshydrophobic(H)and
magentaindicateshydrogenbonddonor(HBD)(forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
Table3
FitvaluesofeachcompoundsinthetestsetusingHypo1,SB_Hypo1andSB_Hypo2.
CompoundNo.
Exp.IC
50
nMa
FitValue
Hypo1
SB_Hypo1
SB_Hypo2
1
6
4.453
4.916
2.759
2
6
4.432
4.285
3.434
3
11
3.893
4.352
2.94
4
13
3.740
4.379
3.642
5
14
4.169
4.076
1.764
6
14
3.446
3.952
2.506
7
26
3.755
3.011
2.169
8
30
3.361
3.839
e
9
36
3.676
3.333
1.888
10
40
3.93
3.002
e
11
64
3.784
3.151
3.698
12
115
3.446
2.853
3.865
13
142
3.238
3.093
3.608
14
146
3.472
3.087
3.643
15
231
3.613
3.157
3.442
16
239
3.672
3.125
3.502
17
280
3.622
3.248
0.768
18
343
3.327
2.646
3.302
19
600
3.512
2.515
e
20
728
3.127
2.269
2.857
21
914
2.944
2.496
3.727
22
1000
2.548
2.935
2.522
23
1290
2.983
2.567
3.577
24
2630
2.708
2.742
3.785
25
4400
0.961
0.988
e
26
4730
1.239
2.619
e
27
6900
0.81
1.835
e
28
12800
0.567
3.046
e
29
75000
1.952
1.852
e
30
200000
0.078
0.358
e
a
Activity scale: IC
50
< 300 0 nM (highly y active), , 300 nM   IC
50
< 3000 nM
(moderatelyactive),IC
50
3000nM(lowactive).
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2942
VB.NET PDF- View PDF Online with VB.NET HTML5 PDF Viewer
PDF to Text. Convert PDF to JPEG. Convert PDF to Png to PDF. Image: Remove Image from PDF Page. Image Link: Edit URL. Bookmark: Edit Bookmark. Metadata: Edit, Delete
convert pdf to web pages; converting pdf to html code
C# PDF Library SDK to view, edit, convert, process PDF file for C#
provides C# users abilities to view, annotate, convert and create NET package offers robust APIs for editing PDF document hyperlink (url) and quick
convert pdf to webpage; best website to convert pdf to word online
screening process.Thefinal leads which satisfiedall thecritical
chemicalfeaturesofHypo1andSB_Hypo1aswellastheADMEand
DruglikenesspropertiesaretabulatedinTable5.
2.6. Moleculardockingstudies
Moleculardockingprovidesavisualizationofpotentialbinding
orientationsthatdisplayclearlythehydrogenbondinginteractions
withcriticalresiduessuchasThr184andAsp93.
2.6.1. LigandFit/DS
Moleculardocking is a computational technique that sample
conformationsofsmallcompoundinproteinbindingsite;scoring
functionsareusedtoassesswhichoftheseconformationswerebest
complements to the protein n binding g site. Molecular r docking
programsconsistsoftwoessentialparts:analgorithmthatsearches
theconformational,rotationalandtranslationalspaceavailableto
candidatemoleculeswithinbindingsiteandanobjectivefunction
tobeminimizedduringtheprocess.Alltheselectedhitcompounds
(36)andtestsetmolecules(30)weredockedintotheactivesiteof
HSP90toconfirmthesuitablebindingorientationoftheligandsand
also to ensure e its s geometric fit within the active e site. . Flexible
dockingfollowedbyconsensusscoringmethodwasperformedto
identifyasuitableorientationofligandsintheactivesiteofHSP90.
LigScore1,LigScore2,PiecewiseLinearPotential1(PLP1),Piecewise
LinearPotential2(PLP2),Potentialofmeanforce(PMF),Jain,Ludi
andDockscorewereusedforthedetectionofHSP90inhibitors.PLP
scoreswerecalculatedbasedonthedescriptionsofhydrogenbonds
forming.PMFscores werecalculatedby the summing pair-wise
interaction terms s over r all inter-atomic pairs s of the recep-
toreligandcomplex.Dock scorewasconsideredasthedegreeof
difficultyaboutligandmovingintothebindingsite.JainandLudi
scoreswereconsultedin hydrophobic interaction anddegreeof
freedom,respectively.Beforeanalyzingthedockingresultofthehit
molecules, docking g procedure e was validated d using g the test set
molecules.IntheinterpretationofdockedtestsetmoleculesJain
scoringfunctionwasabletoretrievereasonableposeoftheactive
molecules.Mostofthesorted molecules,using Jainscore,show
goodinteractionswiththecriticalresidueslikeAsp93andTry184.
Hencethesameanalysisprocedurehasbeenperformedforthehit
moleculesderivedfromavirtualscreeningprocesswhichpicked
out8and7moleculesfromMaybridgeandChembridge,respec-
tively.Tosortoutthesemoleculesvisualinspectionofthedockedhit
compoundswerecarriedouttofindthecriticalinteractionsofthe
smallmoleculeswithreceptor.Themainaimofthisvisualization
processistoeliminatethemoleculeswhicharenotabletoshowthe
hydrogenbondinteractionswiththecriticalresiduespresentinthe
HSP90activesite.Basedontheaboveanalyzes,3and1fromMay-
bridgeandChembridgedatabaseswereselectedasapotentinhib-
itorsoftheHSP90.
2.6.2. GOLDanalysis
GOLDdockingwasperformedtofindtheinteractionsbetween
the ATP binding site of HSP90 and ligands. . The e databases hit
compounds(36)aswellasthetestsetmolecules(30)wereusedas
inputforGOLDdocking,foreachmolecule10posesweresavedand
the corresponding goldfitness score were generated.Larger the
fitnessscoreoftheligandposewasbetterbecauseitwascalculated
basedonthenegativeofthesumofthecomponentenergyterms.
The fitness functionwas optimizedfor the prediction of ligand
bindingposition.Thewellorientatedligandposehaslowestenergy
withaverageGoldfitnessscore[32]wereenumerated.Inthecase
of test t set molecules, most t of the e active e compounds s show w an
averagefitnessscoresmorethan25andtheinactivecompounds
havetheaveragefitnessscorelessthan25.Fromthisanalyzesit
wasconfirmedthattheactivemoleculesofHSP90showsthegood
fitnessscore.So,theabovepre-validatedanalysiswasusedtosort
Table4
StatisticalparameterfromscreeningDecoyset.
No.
Parameter
Hypo1
SB_Hypo1
1
Totalnumberofmoleculesindatabase(D)
1200
1200
2
Totalnumberofactivesindatabase(A)
25
25
3
Totalnumberofhitmoleculesfromthe
database(Ht)
29
33
4
Totalnumberofactivemoleculesinhitlist
(Ha)
25
22
5
%Yieldofactives[(Ha/Ht)100]
86.21
66.66
6
%Ratioofactives[(Ha/A)100]
100
88
7
EnrichmentFactor(EF)
4.1
3.2
8
Falsenegatives[A-Ha]
0
3
9
FalsePositives[HteHa]
4
11
10
Goodnessoffitscore
a
(GF)
0.89
0.71
a
[(Ha/4HtA)(3AþHt)x(1-((HteHa)/(D-A))]
Table5
ADMEvaluesfortheselectedcompoundsfromMaybridgeandChembridgedatabaseswhichsatisfiedallthechemicalfeaturesofHypo1andSB_Hypo1.
Maybridge
Chembridge
Name
BrainBlood
Barrier(BBB)
Solubility
Name
BrainBlood
Barrier(BBB)
Solubility
BTB03651
0.876
3.057
Compound10905
1.185
2.717
BTB07807
1.319
2.67
Compound12503
0.754
3.388
BTB08814
0.753
3.661
Compound13574
1.4
2.288
BTB13767
0.705
3.428
Compound15649
0.746
2.714
BTB14330
0.943
3.38
Compound19817
1.127
2.824
DSH00533
0.774
3.645
Compound20342
0.6
3.225
HTS01124
0.764
3.434
Compound20351
0.76
2.669
HTS02708
0.883
3.292
Compound20354
0.996
2.466
HTS05096
0.759
3.882
Compound20887
1.055
3.358
HTS10008
0.905
3.635
Compound21245
0.538
3.478
HTS12786
1.066
3.439
Compound23403
0.974
3.801
JFD03179
0.693
3.872
Compound25039
0.941
3.474
KM01884
1.001
3.112
Compound26856
0.702
3.27
NRB02089
0.949
2.177
Compound30002
0.641
2.876
NRB02765
1.114
2.387
Compound31095
1.058
2.498
RDR03170
1.247
2.813
Compound32927
0.574
2.812
SCR00085
0.721
3.861
Compound8921
1.041
2.81
e
e
e
Compound8964
0.953
3.496
e
e
e
Compound9112
1.185
2.7
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2943
outtheretrievedhit moleculesfromthedatabases.Totally,8hit
molecules from m Maybridge e and Chembridge e databases s were
selectedaspotentinhibitorsbasedonthefitnessscorecutoffvalue
of25.Theseselectedinhibitorswerefurthervalidatedbyusingthe
visualization method to find the suitable e binding mode e of the
inhibitors based on the critical interactions withthe active site
residues.FourandthreecompoundsfromMaybridgeandChem-
bridgehaveshownbifurcateinteractionswithAsp93andThr184.
Totally,7moleculeswereselectedas goodcompoundsbasedon
GOLDprogram.
Comparing the e results s of f LigandFit t and d GOLD D docking, , 3
compoundsfromMaybridge and1compoundfrom Chembridge
databases were selected d as potent t inhibitors s of HSP90 0 which
showedahydrogenbondinteractionswiththeimportant amino
acidslikeAsp93andTyr184aswellasallthemoleculeshavegood
score (Jain score/LigandFit and gold fitness score/GOLD) values
(Fig.8).TheHBAofcandidatecompoundshaveshownaninterac-
tionwithAsp93andtheHBDofcandidatecompoundstointeract
with COofAsp93.Inaddition,thecompoundboundinthepre-
dicted orientation andsatisfied the expectedhydrophobic c inter-
action as defined by our r query y (Hypothesis). Furthermore,
the candidate compounds s satisfied d the demands of f a a chemical
features-basedandstructure-basedhypothesismodels.Fromthe
aboveresults,ithasbeenprovedthatthescreenedmoleculescan
effectivelydisruptthebindingofATPaseandmatchthestructural
requirementofanewtypeofHSP90inhibitors.Hence,finally,4
molecules from m the databases s were selected as most t potent
inhibitorsforHSP90.
2.7. Similarityanalysis
ThePubChem [33]andScifinderscholar[34]searchtoolsare
usedtoconfirmthenoveltyofthehitcompounds.SID4259781and
SID 859899 9 compounds show a partial similarity with our hit
compounds. The partial similarity y compounds s from PubChem
databases were e tested d with the e SB_Hypo1, , Hypo1 as well as
moleculardockingstudies.InSB_Hypo1,SID4259781showsafit
valueof0.9andSID859899failstoscreen.IncaseofHypo1both
thecompoundsarenotabletofitthegeometricconstrains.Overall
boththecompoundsformthePubChemdatabaseisnotabletofit
withourbesthypotheses.Hencetoconfirmtheorientationofthese
molecules intheATP binding siteofHSP90,LigandFitandGOLD
moleculardockingwasperformed.ThedockscoreandJainscore
forthehitmoleculesarefarbetterwhencomparedwiththePub-
Chemmolecules.ThedockscoreforSID4259781andSID859899
compoundswaslessthan85butthedatabasehitmoleculesshows
avalue greaterthan90.To checkthehydrophobic natureofthe
molecules Jain n scoring g was calculated, , the hit molecules shows
avaluegreaterthan4.5butthePubChemmoleculesshowsavalue
lessthan3.5.Goldfitnessscorevalueforthesetwocompoundsare
lessthan25whichwereinactiveinourcase(Table6).Molecular
dockingstudiesalsoprovethatthesecompoundsarenotsuitableto
form hydrogen bondinteractions s with h the critical residues like
Asp93andThr184.Thenoveltiesofthefourhitcompoundswere
furtherconfirmedbytheScifinderscholarsearchtool.Hencewe
suggestthatthesefourcompoundsarenovelscaffoldtoinhibitthe
HSP90activity.
Fig.8. HitcompoundsfromMaybridge(DSHS00533,JFD03179,BTB14330)andChembridge(Compound31095)intheactivesiteofheatshockprotein90.Dottedlinerepresents
thehydrogenbonds.LigandsareingreencolorandthecriticalaminoacidsAsp93andThr184wasrepresentinballandstick(forinterpretationofthereferencestocolourinthis
figurelegend,thereaderisreferredtothewebversionofthisarticle.)
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2944
3. Conclusions
InhibitionofHSP90hasemergedasanewpromisingtargetin
thefieldofantitumortherapysinceitinfluencesmanysignaling
pathways. Several l structurally diverse compounds possessing
growthinhibitorypotencyagainst HSP90overexpressing cancer
cellswereidentifiedusingpharmacophore,virtualscreening and
moleculardockingstudies.Inthisstudy,softwarepackagesDSand
LigandScout were usedfortheidentification andvisualization of
proteineligandinteractionsites,pharmacophoremodelgeneration
and database screening. . The e qualitative pharmacophore model
(Hip-Hop) was compare with the structure-based hypothesis to
elucidatingthecriticalchemicalfeaturestoinhibittheactivityof
HSP90. The qualitativebest t hypothesis, , Hypo1, contains 2 HBA,
1HBDand2Hfeatureshavebeenselectedbasedonthecluster
analysis, geometrical parameters of the e hypothesis as well as
validatedusing testanddecoysets.Forstructure-basedpharma-
cophore,fivedifferent hypothesesweregeneratedbasedon the
HSP90crystalstructuresboundwithfivedifferent ligands.These
five hypotheses were superimposed and removed d the e overlaid
features.Finally,thedynamicsstructure-basedhypothesisconsist
of8chemicalfeatures whichwassplit intotwopharmacophore
models: SB_Hypo1consistsof1HBA,1HBD and4H groups and
SB_Hypo2, 3HBA,1 HBD and 2H chemical l groups. . These two
hypotheseshavebeenvalidatedusingtestset,SB_Hypo1,selected
as abest hypothesis basedon itsability inretrieving the active
molecules andconfirmedby the decoyset whichshowed good
predictive power. . Hence, the e best hypothesis Hypo1 1 and d the
SB_Hypo1 were e used as a a 3D D structural l query to screen the
chemical databases s like e as s Maybridge and Chembridge for
retrieving new w potent t inhibitorsofHSP90.A A total l of17 7 and19
compounds from m Maybridge e and Chembridge were selected as
commonhit moleculesfromthe virtual l screening g processusing
Hypo1andSB_Hypo1aswellasthesemoleculessatisfiedthedrug
like properties, respectively.Hence,theselectedmoleculeswere
subjected to two molecular docking g methods: : LigandFit/DS S and
GOLD were applied d to o select t the e potent t inhibitors of f HSP90.
LigandFit/DS picked 3 3 and d 1 molecules from m Maybridge e and
Chembridgedatabasesapplyingconsensusscoringfunctions and
also theselectedmoleculeswerevisualizedtofindthehydrogen
bonds with h Asp93 and Thr184. From GOLD docking, 4 and 3
moleculesfrom MaybridgeandChembridgewereselectedbased
onthe goldfitness scoreaswell as thecritical hydrogen bonds
with Asp93andTyr184.Bycomparing thebothdockingresults,
finally, 3 and 1 compounds from m Maybridge e and d Chembridge
databases were e selected as potent t inhibitors s of HSP90 0 which
showedgoodscorevaluesandnecessaryhydrogenbondinterac-
tions with both the critical l amino o acids Asp93 3 and Tyr184. All
these molecules s show w good interactions. In n conclusion, it t has
shown that, modification of typical pharmacophore and combi-
nationofdockingwithpharmacophorebasedvirtualscreeningcan
improvetheactivityofleads.Thesehitsareunderoptimizationfor
furtherdrugdevelopmenttogetmoredruggablelead.
4. Methodsandmaterials
4.1. Pharmacophoremodeling
Theligandbasedandstructure-basedpharmacophoremodeling
studies were carried out using theHip-Hop/DSandLigandScout,
respectively. Hip-Hop p mainly focused d on the critical l common
featurespresentinthesetofinhibitorsandLigandScoutgenerate
the structure-basedpharmacophore model based on the critical
interactionsbetweentheproteineligand.
Ligand and d structure-based pharmacophore modeling g is the
productivetooltodiscoverthecompoundswithimprovedpotency
andpharmacokineticproperties.Ligandbasedpharmacophorewas
classifiedintotwocategories,first,basedonthecommonfeatures
present inthe set of molecules andsecond,is purely basedon
activityvaluesandthestructureofligandswhichwerepresentin
trainingsettogenerateapharmacophoremodels.Structure-based
pharmacophore model l will utilize the e interactions s between
receptoreligandcomplexestogenerateahypothesis.Thestructure-
basedmethodbecomesincreasinglyimportantbecausethedeposit
ofX-raycrystalstructuresinPDBwasgrowingrapidly.Itwassug-
gestedthattheinformationabouttheproteinstructureisagood
sourcetobringforththestructure-basedpharmacophoreandused
asfirst-screeningbeforedockingstudies[35,36].
4.2. LigandbasedapproachusingDS
Thesignificanceofpharmacophoremodelspurelydependson
thequalityofthemoleculesusedinpharmacophoregeneration.So
themainattentionwasgiventothetrainingsetmoleculeselection,
in this work k five different HSP90 co-crystal structures s were
selected from m PDB.These five e inhibitors, , PDB ID: 3BM9,3HHU,
pdb:3BMY,3FT8,and3EKRwereselectedbasedonthe sizeand
activity values(IC
50
) of the ligands.All the five inhibitors were
extractedfrom its boundstructure andtheCHARMm force field
was applied toverify the bondorders.The‘principal’and‘Max-
OmitFeat’ values were set to o 2 2 and 0forall compounds s in the
trainingset.HBA,HBDandHfeatureswereconsideredasimpor-
tantchemicalfeaturesofHSP90inhibitorsbasedonthepublished
resultofHSP90HypoGenmodel[3].Tenhypothesesweregener-
atedandthe best hypothesis was selectedbasedonthe cluster
analysismoduleaswellastherankingscoreofthehypothesis.
4.3. Generationofreceptor/structure-basedpharmacophoremodels
usingLigandScout
FiveX-rayco-crystalstructuresofHSP90(PDBID:3BM9,3FT8,
3HHU, 3EKR,3BMY) wereusedto o generatethe structure-based
pharmacophore models. These five structures were selected
basedontheresolutionanditsdepositeddate.
4.3.1. Structure-basedpharmacophoremodeling
Theligandinteractionswithcriticalaminoacidspresentinthe
active site of protein was s a a sufficient t input to o generate e the
structure-basedpharmacophore.Twosoftwaretoolswereusedto
keyoutthecrucialpharmacophorepatterns:(i)LigandScout,used
tostudytheinteractionsbetweentheinhibitorsandaminoacidsin
the active site of HSP90, , also o used as s a tool l for automatic
construction andvisualization ofpharmacophoremodel l derived
from the 3D coordinates of f the proteins [31,37]. The e software
extracts and d interprets ligandereceptor interactions s such as
hydrogen bond, , charge e transfer, , hydrophobic regions s of their
macromolecularenvironment from m PDB files. Multiple e chemical
features were detected and mapped onto the ligand functional
groupsthatareallowingtheusertoexportknowledgeonHSP90
Table6
Similaritysearchvalidation.
Name
Fitvalue
JainScore DockScore e GoldFitness
Score
Hypo1 SB_Hypo1
Compound31095 2.49
2.61
4.93
93
53
DSHS00533
4.66
3.56
5.81
91
58
JFD03179
3.71
4.23
4.66
105
57
BTB14330
4.06
3.59
4.46
99
56
SID4259781
e
0.9
3.38
84
22
SID8595899
e
e
3.42
83
17
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2945
inhibitors.AlternativeHBAand/orHBDsiteareconsideredsimul-
taneouslyontheproteinwithinthelimitsofgeometricconstraints.
Excludedvolumesphereswerealsoaddedtothestructure-based
model onto o coordinatesdefined d by protein side chain atoms to
characterizetheinaccessibleareasforanypotentialligand(ii)DS
was usedfor theconversion of.hypoedit to .chm m file, whichis
a suitable format t for screening g the multi-conformational three-
dimensionalchemicalstructuredatabases.The3Dcoordinationof
the interaction point was obtained from LigandScout pharmaco-
phoredefinitionsandresultsinspecificinteractionmodelthatare
abletomaptheligandsintheirbioactiveconformation.
4.3.2. PharmacophoremodelgenerationusingLigandScout
TheLigandScout[37]softwarewasusedtoobserveanddetect
thecrucialinteractionsbetweenthecriticalaminoacidspresentin
theactivesiteofHSP90anditsinhibitors.Stepwiseinterpretation
ofthefunctionalgrouppatternshaveperformedforligands:planar
ringdetection,assignmentoffunctionalgrouppatterns,determi-
nation of the hybridization state and finally the e assignment of
Kekule pattern. We built the pharmacophore modelsfor HSP90
complexes with five e different t inhibitors. All l the e generated
hypothesesin.hypoedit formatwhichwere translatedinto .chm
usingDS/Hypoeditscripttouseas3Dqueryforthevirtualscreening
process.
4.4. Pharmacophoreselectivityandevaluation
To validate e the e generated qualitative and structure-based
hypotheses, two kinds s of dataset were constructed, test t and
decoy sets. . Test t set contains s 30 0 diverse compounds s with the
activity value e (IC
50
) between 6 6 nMe200,000 nM, which were
classified into o active (IC
50
< 300nM M ¼ ¼ þþþ),moderate active
(300 nM
 IC
50
< 3000 nM
¼ þþ) and d low
active
(IC
50
3000nM¼þ)HSP90inhibitors[3].MDL-ISISDrawv2.5was
used to o sketch h the e two-dimensional chemical structures of f all
compoundsandconvertedintotheircorresponding3Dformatby
exportedintoDS.BestConformationmodulewasusedtogenerate
the 255conformations ofeach compound to assurethe energy
minimized conformation by applying g CHARMm m force e field d and
Polingalgorithm[38].Theconformationswithenergyvaluehigher
than20kcal/molfromtheglobalminimumwererejected.Testset
is mainly used to evaluate, how well l the e selected hypothesis
distinguishedbetweenactiveandinactiveinhibitors.
Decoyset,waspreparedbycalculatingthe1Dpropertyof1200
moleculesThemainpurposeofthisvalidationistocheckhowwell
the generated d hypothesis differentiates the active from m inactive
molecules.Twenty-fiveactiveHSP90antagonistsalsoincludedin
thedecoysettocalculatethestatisticalparametersfortheselected
best hypothesis such as goodness of hit score (GH),enrichment
factor(EF).GHandEFarethetwomainfactorswhichcanpredict
thecapabilityofthehypothesis.
4.5. Insilicoscreening
Virtualscreeningwasoneofthefastandaccuratetechniqueto
obtainthenewleadswithdesiredactivityprofiles[3].Therepre-
sentativepharmacophore canbeusedas asearchquery forthe
virtual screening of f the multi-conformational databases by
applyingtwodistinctalgorithms,so-calledFASTandBESTFlexible
search,toretrievethecompoundswithnovel scaffoldsandwith
desiredchemical features.Here,FastFlexible search methodwas
usedtoscreenthesmallmoleculedatabaseslikeMaybridgeand
Chembridge. It handles the conformational l flexibility by pre-
generatingarepresentativesetofdiverseandlowenergyconfor-
mations with h the poling algorithm. Maximum m Omitted Features
option was chosen n as s ‘0’ for r the e ligand-based pharmacophore
models (Hypo1) whichwas changed into ‘1’ forstructure-based
pharmacophore(SB_Hypo1)becausemappingallfeaturespresent
in the structure-based hypothesis s will l reduce the hit t rate. The
resultinghitmoleculeswererankedaccordingtotheirgeometricfit
valuethatindicateshowwellthemoleculesweremappedontothe
hypothesisfeatureslocationconstraintsandtheirdistancedevia-
tion from m the e feature centers. The compounds s with highest t fit
valueswereextractedandsubjectedtoADMEandLipinski’sruleof
fivetorefinethehitmolecules.Testsetmoleculesandthehitleads
whichsatisfiedthechemicalfeaturespresentinboththehypoth-
eses were e selected d as s an n input for molecular docking using
LigandFit/DSandGOLDprograms.
4.6. Moleculardockingprotocols
Combining the virtual screeningandmoleculardockingtech-
niques have become one of the e reputable e methods in n drug
discovery andenhancing the efficiency in leadoptimization.To
evaluatetheaccuracyofdocking,twodockingsoftware’swereused
inthisstudyforthepurposeofgettingunbiasedresultsofdocking,
LigandFit/DSandGOLD.
Mostofthedockingalgorithmsassumetheproteinasarigid
objectwhichleadstopoorcorrelationofdockingscoreswiththe
experimentalbindingaffinitiesofligands[39,40].Thereisnosingle
dockingalgorithmorscoringfunctionthatcancorrectlypredictthe
binding affinities of ligands in every proteineligand d complex
[39,40]. Hence e in n this s study two o different docking programs
(LigandFitandGOLD)wereinvestigatedonthetestsetmoleculesas
well as the leadsto identify the binding modeofligandsin n the
active pocket of a protein n and d to o predict the binding affinity
betweentheligandandtheprotein[20].
4.6.1. LigandFit/DSdocking
LigandFitwasexecutedforaccurate orientationoftheligands
intotheproteinactivesite.Fordockingstudy,proteinwasprepared
byremovingallthewatermoleculesandCHARMmforcefieldwas
applied.Theactivesitewasidentifiedbythevolumeoccupiedby
co-crystalandthecriticalresidueswereselectedbyexaminingthe
40HSP90co-crystalstructuresfromPDB.Throughoutthedocking
processtoptenconformationsweregeneratedforeachligandafter
theenergyminimizationusingthesmartminimizemethod,which
beginswithsteepestdescentmethodandfollowedbytheconju-
gategradientmethod.Eachofthesavedconformationwasevalu-
atedandrankedusingthescoring functionsincludingLigScore1,
LigScore2,PLP1,PLP2,JAIN,PMFandLUDI.Inthiswork,Jainscore
wasusedtoselecttheleadswhichwerefurtherevaluatedbyvisual
inspection ofproteineligandinteractions. All l theligands,which
formedthegoodhydrophobicandhydrogenbondinteractionswith
Asn51, Ala55, Lys58, , Gly97, Met98, Phe138 and d Asp93, , Thr184,
respectively,wereselectedasthepotentleadsforHSP90.
4.6.2. GOLDdocking
GOLDv4.1(http://gold.ccdc.cam.ac.uk./index.php)wasusedfor
predict, how well l the flexible molecules s bind d into the proteins
activesite.Thisprogramisusingageneticalgorithmmethodology
for proteineligand docking that t allows full l ligand and partial
proteinflexibility.CrystalstructureofHSP90wasselectedfromPDB
fordocking process,hydrogen atomswereadded, andall water
moleculeswereremovedfromtheprotein.Tosimplifythedocking
calculations,thebindingsite oftheligandwasdefinedas a8Å
radius from m the bound ligand and d used d as s an input for r GOLD
calculations.Dockingcalculationswereperformedusingthedefault
GOLDfitnessfunctionanddefaultevolutionaryparameters:pop-
ulationsize¼100;selectionpressure¼1.1;#operations¼100,000;
S.Sakkiahetal./EuropeanJournalofMedicinalChemistry46(2011)2937e2947
2946
Documents you may be interested
Documents you may be interested