Home

scd.uhp-nancy.fr - Bibliothèques de l`Université de Lorraine

image

Contents

1. SSS Target and r Hit d Lead e Regulatory hit identification refinement 4 refinement development s VHS LSI CCS Early knovdedge 5 improved decision making Reduce attrition rates tapes du d veloppement d un m dicament potentiel avant essais 80 2 Fd 9 L encore on ne pr sume en rien de la forme que prennent les mol cules dans ces banques Il peut s agir d une forme mat rielle plaques adapt e un screening exp rimental ou bien d une forme immat rielle base de donn es informatiques adapt e un screening virtuel Les r gles de Lipinski nomm es rule of 5 cause des param tres et non cause du nombre de r gles sont 1 masse mol culaire lt 500 Da 2 donneurs de liaisons H lt 5 3 accepteurs de liaisons H atomes N et O lt 10 4 logP ccefficient de partition octanol eau lt 5 indique que la mol cule est capable de franchir les membranes biologiques On entend par l une banque de taille raisonnable destin e tre employ e de fa on routini re sur une cible quelconque Une telle banque se doit d avoir une diversit chimique importante et d exclure des compos s non drug like en premier lieu les toxiques 16 Essais La vie d un m dicament s organise en plusieurs tapes suivant un protocole rigoureux mise au point essais pr cliniques essais cliniques phase I III puis suivi post commercialisation
2. Appariement des nucl otides au sein de la structure en double h lice de l ADN Du g nome au prot ome Les prot ines sont galement des polym res constitu s partir des 20 acides amin s naturels La signification du con ic oo codage g n tique donne le lien entre les s quences a 7 nucl iques et peptidiques les s quences g n tiques codent U Phe c les s quences prot iques 57 par s quences de trois p nucl otides de l ARN ou codon voir ci contre 58 59 La 7 a y synth se d une prot ine a plus pr cis ment pour origine la Leu A transcription d une partie sp cifique fonctionnelle de l ADN un g ne sous forme d ARN 60 Le m canisme de A ie transcription de ARN vers des prot ines fait intervenir le Met G ribosome ensemble d une centaine de prot ines bs enzymatiques qui sont les ouvri res de cette chaine de Ms z production 61 Au niveau du prot ome un niveau de complexit suppl mentaire est pr sent par rapport au g nome alors que l ADN et ARN poss dent des structures invariables l activit et la fonction des prot ines d pendent intrins quement de leur g om tries qui d finissent un large espace conformationnel On appelle ce processus transcription de l ADN proche sur de nombreux points du processus de r plication de l ADN intervenant lors de la division cellulaire Certaines prot ines dites r gulatrices de g nes
3. 62 Brugge J S New intracellular targets for therapeutic drug design Science 260 1993 918 919 63 Boutin J A Tyrosine proteine kinase inhibition and cancer International Journal of Biochemistry 26 issue 10 11 1994 1203 1226 64 Gay B Suarez S Caravatti G Furet P Meyer T and Schoepfer J Selective Grb2 SH2 inhibitors as anti Ras therapy International Journal of Cancer 83 1999 235 241 65 Guilloteau J P Fromage N Ries Kautt M Reboul S Bocquet D Dubois H Faucher D Colonna C Ducruix A and Becquart J Purification stabilization and crystallization of a modular protein Grb2 Proteins Structure Function and Genetics 25 issue 1 1996 112 119 66 Thornton K H Mueller W T McConnell P Zhu G Saltiel A R and Thanabal V Nuclear magnetic resonance solution structure of the growth factor receptor bound protein 2 Src homology 2 domain Biochemistry 35 1996 11852 11864 67 Senior M M Frederick A F Black S Murgolo N J Perkins L M Wilson O Snow M E and Wang Y S The three dimensional solution structure of the Src homology domain 2 of the growth factor receptor bound protein 2 Journal of Biomolecular NMR 11 1998 153 164 68 Nioche P Liu W Q Broutin I Charbonnier F Latreille M T Vidal M Roques B Garbay C and Ducruix A Crystal structures of the SH2 domain of Grb2 Highlight on the binding of a new high affinity inhibitor Journal of Molecular Biology
4. 75 Marcou G and Rognan D Optimizing fragment and scaffold docking by use of molecular interaction fingerprints Journal of Chemical Information and Modeling 47 issue 1 2007 195 207 Should structure based virtual screening techniques be used more extensively in modern drug discovery V Leroux and B Maigret Nancy universit Universit H Poincar Nancy I UMR CNRS UHP 7565 eDAM group BP 239 54506 Vandceuvre les Nancy Cedex France Abstract The drug discovery processes used by academic and industrial scientists are nowadays being questioned The approaches of the pharmaceutical industry that were successful 20 years ago are simply not suitable anymore for the increasing complexity of available biological targets and the raising standards for medical safety While the current scientific context resulting from significant developments in genomic proteomic organic synthesis and biochemistry seems particularly favorable the efficiency of drug research does not appear to be following the trend In particular the in silico approaches often considered as potential enhancements for classic drug discovery are an interesting case Techniques such as virtual screening did undergo many significant progresses in the past 5 10 years and have proven their usefulness in hit discovery approaches for who wants to avoid carrying out too many expensive experimental tests while exploring an important molecular diversity However reliabilit
5. Ras signalling BS 2 molecule SH2 m a Ras GAP Ubiquitination __ 4H E Ubi O Uba con Kinases Btk Tec Cytoskeletal PTPc ee y regulator DSPs sm PT Tensin Signal ATE regulator Sh2 SAP Phospholipid sco e a no messenger Positionnement des domaines SH2 pr sents dans diff rentes classes de prot ines 31 Les fonctions correspondantes dans l organisme sont indiqu es gauche Par exemple Src SH2 reconna t pr f rentiellement la s quence pYEEI PI3K p85a pY M V XM X acide amin quelconque Crk pYXXP Grb2 pYXN Cbl D N D XpYXXX P F STAT pYXXQ SAP T S IpYXX I V Syk pYEXL Sh PTP1 L V S XpYXX I L V Sh PTP2 V S XpY I V X I L V X W F Vav pYMEP etc 25 La prot ine Grb2 Description et fonction Grb2 growth factor receptor bound protein 2 est une petite prot ine globulaire d environ 200 r sidus pr sente en milieu intra cellulaire et simplement constitu e d un domaine SH2 compris entre deux domaines SH3 32 35 Les fonctions que peuvent prendre les prot ines sont d termin es par leur r activit chimique qui d pend en grande partie de leur g om trie et est d finie par les s quences particuli res d acides amin s La diversit fonctionnelle des prot ines est particuli rement importante Grb2 est une prot ine adaptateur elle permet de mettre en relation des prot ines g n ralement de taille plus importante Grb2 poss de
6. popraoid sjdios yno sreo ppoqony SuLyoop 0 Jord uns aq JOJ punoo V9 Yoreas oo priqAy LTI woneajosap ued SaINonns 10 da001 souloyos Suryoseas a quyiear tke 5 Surpnypour peorridus eae 2p09 991n0 arqissod sqof asnu 0 0 0 uonisod qe ais aq ose ueo LTI VO wenpreureTt esaAas sjuasaidar STI cl pue Sur109S tg st IND euondo se ays are srojouresred Te pue JA T Baie xs AS mm4 SOILOPEIE jo uonnqinsiq date JO 19JU99 Surunsse SIOJEM eINjONNS VO sulpeouue poye nus Wey pus Jus y sdnois ypoqomy JUJA SOUTQUIOD uy ay Jojoue red g add e st ed WIN uonouny 0 991 ON pusome weisoid ayeredos uorouny OW 2IQIIBAR UI9A9S patJioods 10 pue suryo oun pueuop uonewunysa SIOU9 291 Suroos yordu apis jq x qy euondo 109 9 Apoidea Trt asn 0 Asea 10 OUS y woy porroods 10 puey 390219791 Aue Ji SI9JEM SIdV d 109 perddns lt 3 ease ees ai A ym Suruuwei o1d sanmba sdno13 peuruno Sproe somnjonys purest r EI tz Joya payers sqof punog e JO UOTJEULIOJUOS parepout Aprordxo pape OTI 811 IPIUUOT 7 es 3 rel VO mq ajgrssod uonouny OUIUE IUOS JO UOTLIOI TMA punoq Sunorpaid WAd S94A aqy uonengyuos y woy s1oydoso1 adeys Jo SUIS da109 eusayxa ue JO s suonouny g qissod payrunif A190 W AQA SUIS e yno u do y m Aysu WoyejuIod uorouny pi Sunoos eondu opnio g uonezinoueieg DOU9I9JOI WOT PAJONTISUOT Suu09s yordu a gt
7. 25 Marangere L E M and Pawson T Structure and function of SH2 domains Journal of Cell Science Suppl 18 1994 26 Pawson T and Gish G D SH2 and SH3 domains from structure to function Cell 71 1992 359 362 27 Pawson T Protein modules and signalling networks Nature 373 1995 573 580 28 Moran M F Koch C A Anderson D Ellis C England L Martin G S and Pawson T Src homology region 2 domains direct protein protein interactions in signal transduction Proceedings of the National Academy of Sciences USA 87 1990 8622 8626 29 Pawson T Olivier P Rozakis Adcock M McGlade J and Henkemeyer M Proteins with SH2 and SH3 domains couple receptor tyrosine kinases to intracellular signalling pathways Philosophical Transactions of the Royal Society of London Series B Biological Sciences 340 issue 1293 1993 279 285 30 Smithgall T E SH2 and SH3 domains Potential targets for anti cancer drug design Journal of Pharmacological and Toxicological Methods 34 1995 125 132 31 Waksman G Kumaran S and Lubman O Y SH2 domains role structure and implications for molecular medicine Expert Reviews in Molecular Medicine 6 issue 3 2004 32 Matuoka K Shibata M Yamakawa A and Takenawa T Cloning of ASH a ubiquitous protein composed of one Src homology region SH 2 and two SH3 domains from human and rat cDNA libraries Proceedings of the National Academy of Sciences USA 89 1992 9015 9019 33 L
8. Wilson R Jayatilake H Gusterson B A Cooper C Shipley J Hargrave D Pritchard Jones K Maitland N Chenevix Trench G Riggins G J Bigner D D Palmieri G Cossu A Flanagan A Nicholson A Ho J W C Leung S Y Yuen S T Weber B L Seigler H F Darrow T L Paterson H Marais R Marshall C J Wooster R Stratton M R and Futreal P A Mutations of the BRAF gene in human cancer Nature 417 2002 949 954 12 Garbay C Liu W Q Vidal M and Roques B P Inhibitors of Ras signal transduction as antitumor agents Biochemical Pharmacology 60 2000 1165 1169 13 Liu W Q Vidal M Math C P rigaud C and Garbay C Inhibition of the Ras dependent mitogenic pathway by phosphopeptide prodrugs with antiproliferative properties Bioorganic amp Medicinal Chemistry Letters 10 2000 669 672 14 Barbacid M Ras genes Annual Review of Biochemistry 56 1987 779 827 15 Warne P H Viciana P R and Downward J Direct interaction of Ras and the amino terminal region of Raf 1 in vitro Nature 364 1993 352 355 16 Zhang X F Settleman J Kyriakis J M Takeuchi Suzuki E Elledge S J Marshall M S Bruder J T Rapp U R and Avruch J Normal and oncogenic p21ras proteins bind to the amino terminal regulatory domain of c Raf 1 Nature 364 1993 308 313 17 Basu T Warne P H and Downward J Role of Shc in the activation of Ras in response to epidermal growth factor and nerve growth factor Oncogen
9. pYpY N B turn Les r sultats obtenus ne permettent l heure actuelle aucune validation il est difficile de distinguer la moindre sp cificit parmi les 8000 combinaisons la distribution des scores est troite et en particulier en ce qui concerne Grb2 la sp cificit du r sidu Ny n est pas mise en vidence Situation pr sente Les r sultats de docking peuvent difficilement donner lieu une publication tant donn qu aucune validation n est donn e quant la capacit de GOLD classer correctement les mol cules entrantes suivant leur affinit pour Grb2 SH2 Toutefois nous conservons les r sultats du protocole de validation On peut mettre l hypoth se que l chec actuel de la validation soit d la pr cision insuffisante de la fonction de score de GOLD point faible souvent voqu pour ce programme Les structures obtenues semblent correctes les peptides pYXXX sont correctement orient s et le groupe pTyro est bien positionn dans sa cavit sp cifique Si l avenir un protocole automatis de post traitement est mis en place par exemple par minimisation en utilisant un champ de force il sera utile l utiliser afin de tenter de valider le docking tout en pr cisant ventuellement quels niveaux le protocole actuel pourrait tre am lior De plus pour des mol cules ponctuelles la condition la conduite de simulations de m canique mol culaire ult rieures plus pr cises ne requ
10. 13 Mervis J Productivity counts but the definition is key Science 309 2005 726 727 14 Vogel H G RieB G and Vogel W F Strategies in drug discovery and evaluation in Drug discovery and evaluation pharmacological assays ed H G Vogel 2002 Springer Verlag Berlin Heidelberg New York p 1 21 15 Anderson D Public computing Reconnecting people to science in Conference on shared knowledge and the Web 2003 Madrid Spain 16 Rauwerda H Roos M Hertzberger B O and Breit T M The promise of a virtual lab in drug discovery Drug Discovery Today 11 issue 5 6 2006 228 236 17 Garrett M D and Workman P Discovering novel chemotherapeutic drugs for the third millenium European Journal of Cancer 35 issue 14 1999 2010 2030 18 Searls D B Using bioinformatics in gene and drug discovery Drug Discovery Today 5 issue 4 2000 135 143 19 Shneiderman B Inventing discovery tools combining information visualization with data mining Information Visualization 1 issue 1 2002 5 12 20 Leach A R Molecular modelling Principles and applications 1996 Addison Wesley Longman Essex 585 pages 21 Ooms F Molecular modeling and computer aided drug design Examples of their applications in medicinal chemistry Current Medicinal Chemistry 7 2000 141 158 22 Oprea T I and Matter H Integrating virtual screening in lead discovery Current Opinion in Chemical Biology 8 issue 4 2004 349 358 23
11. Cancer and Metastasis Reviews 13 issue 1 1994 67 89 5 Vojtek A B and Der C J Increasing complexity of the Ras signaling pathway Journal of Biological Chemistry 273 issue 32 1998 19925 19928 6 Reuter C W M Morgan M A and Bergmann L Targeting the Ras signaling pathway a rational mechanism based treatment for hematologic malignancies Blood 96 issue 5 2000 1655 1669 7 Olsson A K Ras MAPK signaling in differentiating SH SYSY human neuroblastoma cells in Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 2000 Academic press Acta Universitatis Upsaliensis Uppsala p 66 8 Alberola Ila J and Hern ndez Hoyos G The Ras MAPK cascade and the control of positive selection Immunological Reviews 191 2003 79 96 9 Brambilla R Gnesutta N Minichiello L White G Roylance A J Herron C E Ramsey M Wolfer D P Cestari V Rossi Arnaud C Grant S G Chapman P F Lipp H P Sturani E and Klein R A role for the Ras signalling pathway in synaptic transmission and long term memory Nature 390 1997 281 286 10 Bos J L Ras oncogenes in human cancer a review Cancer Research 49 1989 4682 4689 11 Davies H Bignell G R Cox C Stephens P Edkins S Clegg S Teague J Woffendin H Garnett M J Bottomley W Davis N Dicks E Ewing R Floyd Y Gray K Hall S Hawes R Hughes J Kosmidou V Menzies A Mould C Parker A Stevens C Watt S Hooper S
12. Carpenter C Duckworth B Graziani A Kapeller R and Soltoff S Oncogenes and signal transduction Cell 65 issue 5 1991 914 10 Foulds L The experimental study of tumor progression I III 1954 Academic press London 11 Nowell P C The clonal evolution of tumor cell populations Science 194 1976 23 28 12 Land H Parada L F and Weinberg R A Cellular oncogens and multistep carcinogenesis Science 222 1983 771 778 13 Hanahan D and Weinberg R A The hallmarks of cancer Cell 100 2000 57 70 14 Weinberg R A Tumor suppressor genes Science 254 1991 1138 1146 15 Greenblatt M S Bennett W P Hollstein M and Harris C C Mutations in the p53 tumor suppressor gene clues to cancer etiology and molecular pathogenesis Cancer Research 54 issue 18 1994 4855 4878 16 Fedi P Tronick S R and Aaronson S A Growth factors in Cancer Medicine eds J F Holland R C Bast D L Morton E Drei D W Kufe and R R Weichselbaum 1997 Williams and Wilkins Baltimore MD p 41 64 17 Hunter T Oncoprotein networks Cell 88 1997 333 346 18 Kerr J F Wyllie A H and Currie A R Apoptosis a basic biological phenomenon with wide ranging implications in tissue kinetics 26 1972 239 257 19 Kerr J F Winterford C M and Harmon B V Apoptosis Its significance in cancer and cancer therapy Cancer 73 issue 8 1994 2013 2026 20 Wright W E Pereira Smith O M and Shay J W Reversible cellular
13. This data was translated in terms of filtering efficiency E f in table 3 The main result can be interpreted as follows if we apply respectively 10 and 50 filtering using the multiple target SHEF filter amongst all hits we will retain respectively 90 8 and 52 8 of what would have been lost using random selection The comparison between the four available filters based on SHEF rankings suggests that the use of the multiple target consensus ranking should be the best choice This is in agreement with the observations made analyzing figures 7 g 7 h and 7 i More interestingly analysis of SHEF efficiency of the different hits sub groups reveals that molecules specific to the 1PQ6 target conformation according to GOLD are performing poorly with SHEF see table 3 1PQ6 specific line It has been shown that the specific 1PQ6 shape is taken into account by SHEF but 1PQ6 also presents a second particularity the accessibility of a charged residue The corresponding 1PQ6 specific ligands most probably share a binding mode dominated by electrostatic effects that SHEF as it only compares geometries is unable to assess Contrarily the molecules that are defined as hits for all of the three LXRB pocket conformations are those for which SHEF filtering is the most efficient for both values of filtering see table 3 1P8D 1PQ6 1PQ9 line These molecules might have a high degree of adaptability allowing SHEF to perform well in identifying the conf
14. form e la jonction de l h lice aA et du feuillet BB On peut souvent lire que pYXNX est un motif sp cifique Grb2 SH2 ce qui n est pas tout fait exact En effet ce motif peut tre actif avec n importe quel domaine SH2 toutefois il l est alors moins que d autres motifs Concernant Grb2 ces autres s quences sont quasiment toutes inactives C est bien cette inactivit qui est sp cifique et non l activit de pYXNX 31 Ligands pseudo peptidiques ou non peptidiques optimis s Substitution de groupes peptidiques Novartis La premi re approche visant a d couvrir des inhibiteurs plus efficaces de Grb2 SH2 consista pour une quipe de Novartis apr s avoir r solu la structure RX d un complexe de r f rence Grb2 SH2 KPFpYVNV 106 code PDB 1ZFP se baser sur la petite s quence peptidique Ac pYIN NH qui conserve une affinit micromolaire Les optimisations de cette s quence ont ensuite consist tester l effet de substituants non peptidiques 89 109 aux diff rents groupes Ac 69 110 pTyr 111 112 Ile 78 88 113 114 Asn 71 113 et NHb 4 78 115 La combinaison de la plupart des substituants les plus efficaces donna naissance au ligand pseudo peptidique CGP78850 64 qui s av ra 200 fois plus actif que Ac pYIN NH 60 Le squelette tri peptidique lui m me fut dans une d marche s par e remplac afin d aboutir un petit ligand phosphat totalement non pept
15. guliers de la puissance informatique accessible une croissance bien plus spectaculaire que celle des avanc es technologiques au niveau des appareillages de la recherche biom dicale classique R cemment de nouveaux concepts de calcul distribu popularis s par le projet SETI Home ont encore accru l int r t de la communaut scientifique envers les approches in silico en g n ral Ensuite l utilisation de mod les num riques peut permettre lorsque ceux ci se basent sur la structure des assemblages mol culaire nous utiliserons ici dans ce domaine les m thodes de la dynamique mol culaire et du docking de visualiser l action d une drogue sur sa cible Loin d tre une abstraction les m thodes num riques permettent ainsi l acc s le plus frontal qui soit aux ph nom nes biologiques l chelle microscopique Enfin dans la continuit des investissements colossaux effectu s dans la g nomique la communaut scientifique se tourne d sormais vers le prot ome et l interactome qui seuls permettront une bonne compr hension des m canismes fondamentaux du vivant En particulier les structures exp rimentales disponibles de biomol cules travers des bases publiques telles que la Protein Data Bank croissent exponentiellement Dans ce contexte les m thodes num riques sont de plus en plus int ressantes face aux techniques purement exp rimentales qui ne sont gu re en mesure d appr hender seules le p
16. ne de vie un r gime alimentaire trop riche 100 en graisses par exemple ou bien un environnement pollu 2000 favorisent l apparition de cancers Des mutations g n tiques pr disposant certaines formes de cancer peuvent galement tre h rit es 25 c est le cas d un fran ais sur 60 par exemple les mutations du g ne BRCAI reli es aux cancers 1900 1920 1940 1960 1980 Year 50 1000 du sein et des ovaires Cet aspect semble cependant avoir moins d incidence que les facteurs environnementaux 26 Enfin de fa on moins fr quente certains virus contenant dans leur g nome des oncog nes ou g nes inactivant les suppresseurs de tumeur ou bien favorisant indirectement des proto oncog nes de leur cellule h te lors de leur r plication peuvent provoquer des mutations pouvant aboutir la formation de tumeurs canc reuses 27 28 Les diff rentes approches de la lutte contre le cancer La lutte contre le cancer est constitu e de multiples approches souvent compl mentaires Il en va de m me en ce qui concerne les aspects aussi bien m dicaux que scientifiques Dans un tel contexte il est vident que la lutte contre les diff rentes formes de cancer ne pourra pas tre le fait d une seule technique scientifique d un seul axe de recherche En num rant bri vement les principales directions qui peuvent tre suivies nous pouvons donner l impression qu il s agit de domaines distincts bien au
17. phase IV Les essais pr cliniques ont pour objectif d valuer avant l tude chez l homme la s curit du produit toxicit mutagen se canc rogen se son action sur les organes cibles ainsi que son cycle de vie dans l organisme absorption propagation limination La phase I correspond la premi re administration l homme sur une cinquantaine de volontaires sains durant 6 18 mois afin de d terminer plus pr cis ment le dosage du produit La phase II effectu e en g n ral en milieu hospitalier sur plusieurs centaines de malades durant 2 3 ans a pour but de d terminer les conditions optimales d administration dose et posologie et de les relier aux concentrations passant effectivement dans l organisme La phase II pouvant tre tendue plusieurs milliers de patients mesure l efficacit du m dicament rapport traitement de la pathologie effets secondaires dans les conditions d utilisation pr conis es en concurrence avec un placebo et d ventuels m dicaments de r f rences Une autorisation de mise sur le march est d livr e en cas de succ s Bilan financier Les experts valuent le co t total du d veloppement d un nouveau m dicament l heure actuelle 2000 800 millions de 81 noter que ce montant correspond aux co ts totaux investis par l industrie pharmaceutique divis s par le nombre de m dicaments franchissant toutes les tapes menant leur comme
18. que la tumeur est localis e et solide Afin d viter sa r surgence apr s l op ration on utilise la radioth rapie 30 et la chimioth rapie 31 qui peuvent d autre part s av rer suffisantes lorsque la tumeur n a pas atteint une taille critique et peuvent constituer au minimum des traitements palliatifs efficaces dans les cas o le d veloppement et la nature de la tumeur ne permettent plus un traitement curatif Toutefois ces deux m thodes sont connues pour provoquer nombre d effets secondaires d sagr ables car elles ne ciblent pas exclusivement les cellules canc reuses D autres m thodes th rapeutiques telles que l hormonoth rapie 32 33 sont accessibles ne ciblant plus les cellules canc reuses mais favorisant ou non des biomol cules prot ines anticorps hormones qui leurs sont associ es cependant elles n ont souvent encore qu un r le compl mentaire du fait de leur efficacit plus modeste ou de leur domaine d action plus limit Ainsi il est courant d employer simultan ment plusieurs techniques parmi celles que nous venons d voquer 34 Parall lement cela de nouvelles th rapies sont en d veloppement Ainsi de nombreux travaux de recherche portent sur la th rapie g nique 35 consistant ins rer dans les cellules des g nes fonctions th rapeutiques Une des grandes difficult s de cette approche r side dans la mise au point de vecteurs efficaces virus bact ries permetta
19. tyrosine residues Journal of Organic Chemistry 53 1988 3621 3624 155 Liu W Q Olszowy C Bischoff L and Garbay C Enantioselective synthesis of 2S 2 4 phosphonophenylmethyl 3 aminopropanoic acid suitably protected for peptide synthesis Tetrahedron Letters 43 2002 1417 1419 156 Smyth M S Ford Jr H and Burke Jr T R A general method for the preparation of benzylic a a difluorophosphonic acids non hydrolyzable mimetics of phosphotyrosine Tetrahedron Letters 33 1992 4137 4140 157 Burke T R Smyth M S Nomizu M Otaka A and Roller P P Preparation of fluoro 4 phosphonomethyl D L phenylalanineand hydroxy 4 phosphonomethyl D L phenylalanine suitably protected for solid phase synthesis of peptides containing hydrolytically stable analogues of O phosphotyrosine Journal of Organic Chemistry 58 1993 1336 1340 46 158 Yao Z J Gao Y Voigt J Ford Jr H and Burke Jr T R Synthesis of Fmoc protected 4 carboxy difluoromethyl L phenylalanine A phosphotyrosyl mimetic of potential use for signal transduction studies Tetrahedron 55 issue 10 1999 2865 2874 159 Gao Y Luo J Yao Z J Guo R Zou H Kelley J Voigt J H Yang D and Burke Jr T R Inhibition of Grb2 SH2 domain binding by non phosphate containing ligands 2 4 2 malonyl phenylalanine as a potent phosphotyrosyl mimetic Journal of Medicinal Chemistry 43 2000 911 920 160 Kole H K Ye B Akamatsu M Yan X Barf
20. une cellule canc reuse sont pr sent bien caract ris es 13 Les mutations favorisant la formation d une cellule canc reuse conf rent celle ci une autonomie par rapport l organisme h te la cellule canc reuse g n re ses propres facteurs de Evading croissance 16 ce qui d r gule les voies de signalisation correspondantes 17 et est insensible aux m canismes de r gulation de la population cellulaire par l organisme en particulier la mort cellulaire programm e 18 19 D autre part la cellule canc reuse acquiert une grande capacit de multiplication 20 21 et de prolif ration dans l organisme ainsi qu une meilleure r sistance aussi bien au syst me immunitaire qu un traitement ext rieur cela se traduit en particulier par la mutation du g ne suppresseur de tumeur p53 Insensitivity to anti growth signals Sustained 22 Une cellule canc reuse est une cellule pour laquelle Rs l ensemble de telles mutations rend irr versible sa mutation Be li infinie dans l organisme moins qu un traitement appropri Limitless replicative ne soit effectu sur la personne affect e Il est encore impossible de d terminer l origine d un cancer donn ant rieurement la s rie de mutations g n tiques qui le provoque on estime m me souvent que ces mutations peuvent survenir spontan ment Toutefois il s agit d v nements anormaux et tant donn que seule une comb
21. 14308 14314 46 Neel B G and Tonks N K Protein tyrosine phosphatases in signal transduction Current Opinion in Cellular Biology 9 issue 2 1997 193 204 47 Lawrence D S and Niu J Protein kinase inhibitors the tyrosine specific protein kinases Pharmacology amp Therapeutics 77 issue 2 1998 81 114 48 Levitzki A Protein tyrosine kinase inhibitors as novel therapeutic agents Pharmacology amp Therapeutics 82 issue 2 3 1999 231 239 49 Baselga J Targeting tyrosine kinases in cancer the second wave Science 312 issue 5777 2006 1175 1178 50 Force T Kuida K Namchuk M Parang K and Kyriakis J M Inhibitors of protein kinase signaling pathways Emerging therapies for cardiovascular disease Circulation 109 2004 1196 1205 51 Botfield M C and Green J SH2 and SH3 domains choreographers of multiple signaling pathways Annual Reports in Medicinal Chemistry 30 1995 227 237 52 Sawyer T K Src homology 2 domains Structure mechanisms and drug discovery Biopolymers Peptide Science 47 1998 243 261 53 Cody W L Lin Z Panek R L Rose D W and Rubin J R Progress in the development of inhibitors of SH2 domains Current Pharmaceutical Design 6 2000 59 98 54 Madema R H and Bos J L The role of p21 ras in receptor tyrosine kinase signaling Critical Reviews in Oncology Hematology 4 1993 615 661 55 Tari A M Arlinghaus R and Lopez Berestein G Inhibition of Grb2 and Crkl protei
22. 615 56 Beddell C R Goodford P J Norrington F E Wilkinson S and Wootton R Compounds designed to fit a site of known structure in human haemoglobin British Journal of Pharmacology 57 issue 2 1976 201 209 57 PDB RCSB Protein Data Bank http www rcsb org pdb home home do 58 Kellenberger E Rodrigo J Muller P and Rognan D Comparative evaluation of eight docking tools for docking and virtual screening accuracy Proteins Structure Function and Bioinformatics 57 2004 225 242 59 Kontoyianni M McClellan L M and Sokol G S Evaluation of docking performance comparative data on docking algorithms Journal of Medicinal Chemistry 47 issue 3 2004 558 565 60 Perola E Walters W P and Charifson P S A detailed comparison of current docking and scoring methods on systems of pharmaceutical relevance Proteins Structure Function and Bioinformatics 56 issue 2 2004 235 249 61 Kubinyi H Drug research myths hype and reality Nature Reviews Drug Discovery 2 2003 665 668 62 Verkhivker G M Bouzida D Gehlhaar D K Rejto P A Arthurs S Colson A B Freer S T Larson V Luty B A Marrone T and Rose P W Deciphering common failures in molecular docking of ligand protein complexes Journal of Computer Aided Molecular Design 14 2000 731 751 63 Clarke C Woods R J Gluska J Cooper A Nutley M A and Boons G J Involvement of water in carbohydrate protein binding Journal of th
23. 737 738 56 Crick F H C The structure of the hereditary material Scientific American 191 issue 4 1954 54 61 57 Yanofsky C Gene structure and protein structure Scientific American 216 issue 5 1967 80 94 58 Crick F H C On protein synthesis Symposia of the Society for Experimental Biology 12 1958 138 163 59 Crick F H C Barnett L Brenner S and Watts Tobin R J General nature of the genetic code for proteins Nature 192 1961 1227 1232 60 Brenner S Jacob F and Meselson M An unstable intermediate carrying information from genes to ribosomes for protein synthesis Nature 190 1961 576 581 61 Lake J A The ribosome Scientific American 245 issue 2 1981 84 97 62 Anfinsen C B Principles that govern the folding of protein chains Science 181 issue 4096 1973 223 230 63 Richards F M The protein folding problem Scientific American 264 issue 1 1991 54 57 64 James L C and Tawfik D S Conformational diversity and protein evolution a 60 year old hypothesis revisited Trends in Biochemical Sciences 28 issue 7 2003 361 368 65 Figeys D Combining different omics technologies to map and validate protein protein interaction in humans Briefings in functional genomics and proteomics 2 issue 4 2004 357 365 66 Uetz P and Finley Jr R L From protein networks to biological systems FEBS Letters 579 2005 1821 1827 67 Berman H M Westbrook J Feng Z Gilliland G Bha
24. Lyne P D Structure based virtual screening an overview Drug Discovery Today 7 issue 20 2002 1047 1055 24 Anderson A C The process of structure based drug design Chemistry amp Biology 10 2003 787 797 25 Veselovsky A V and Ivanov A S Strategy of computer aided drug design Current Drug Targets Infectious Disorders 3 issue 1 2003 33 40 26 Kuntz I D Blaney J M Oatley S J Langridge R and Ferrin T E A geometric approach to macromolecule ligand interactions Journal of Molecular Biology 161 issue 2 1982 269 288 27 Goodford P J A computational procedure for determining energetically favorable binding sites on biologically important macromolecules Journal of Medicinal Chemistry 28 1985 849 857 28 Gschwend D A Good A C and Kuntz ID Molecular docking towards drug discovery Journal of Molecular Recognition 9 1996 175 186 29 Brooijmans N and Kuntz ID Molecular recognition and docking algorithms Annual Review of Biophysics and Biomolecular Structure 32 2003 335 373 30 Kitchen D B Decornez H Furr J R and Bajorath J Docking and scoring in virtual screening for drug discovery methods and applications Nature Reviews Drug Discovery 3 2004 935 949 31 Hilbert M B hm G and Jaenicke R Structural relationships of homologous proteins as a fundamental principle in homology modeling Proteins Structure Function and Genetics 17 issue 2 1993 138 151 32 Kubinyi H QSAR an
25. MAPK Dans un organisme vivant une voie de signalisation correspond une cascade d interactions entre prot ines dont la finalit est la propagation d un signal biologique De tels processus ont lieu l chelle de la milliseconde voire de la seconde Ils sont localis s aussi bien l ext rieur qu l int rieur des cellules ceux qui sont la fois extra et intra cellulaires permettent aux cellules de communiquer avec leur environnement Souvent un stimulus mineur au d but de la voie de signalisation donne lieu une r ponse biologique bien plus importante L ensemble des voies de signalisation constitue un circuit logique d finissant en grande partie le fonctionnement d un organisme L identification la caract risation et la classification fonctionnelle des diff rentes voies de signalisation d un organisme constitue donc un aspect crucial dans l tude de son interactome CO Growth factor Extracellular membrane Rae a SS Le l a ae way Cell growth and _ ka differenciation Tyrosine kinase receptor SH2 D gt SH3 pY PTB domain Tyrosine phosphorylation p Fonctionnement simplifi des m canismes de croissance et diff renciation cellulaire incluant la voie de signalisation Ras MAPK partie gauche 1 23 La voie de signalisation Ras MAPK 2 6 correspond une cascade d interactions intra cellulaires essentielle dans la croissance la r gulation et
26. Molecular Graphics amp Modelling issue 4 2002 313 328 37 Humphrey W Dalke A and Schulten K VMD Visual Molecular Dynamics Journal of Molecular Graphics 14 1996 33 38 38 Verdonk M L Chessari G Cole J C Hartshorn M J Murray C W Nissink J W M Taylor R D and Taylor R Modeling water molecules in protein ligand docking using GOLD Journal of Medicinal Chemistry 48 2005 6504 6515 39 Wong C F Kua J Zhang Y Straatsma T P and McCammon J A Molecular docking of balanol to dynamics snapshots of protein kinase A Proteins Structure Function and Bioinformatics 61 issue 4 2005 850 858 40 Phillips J C Braun R Wang W Gumbart J Tajkhorshid E Villa E Chipot C Skeel R D Kal L and Schulten K Scalable molecular dynamics with NAMD Journal of Computational Chemistry 26 issue 16 2005 1781 1802 41 So S S and Karplus M Evaluation of designed ligands by a multiple screening method Application to glycogen phosphorylase inhibitors constructed with a variety of approaches Journal of Computer Aided Molecular Design 15 2001 613 647 42 Lyne P D Structure based virtual screening an overview Drug Discovery Today 7 issue 20 2002 1047 1055 43 Wang J Kollman P A and Kuntz I D Flexible ligand docking A multistep strategy approach Proteins Structure Function and Genetics 36 issue 1 1999 1 19 44 Miteva M A Lee W H Montes M O and Villoutreix B O Fast str
27. Tout d abord il a t d montr que le domaine SH2 de Grb2 a une activit biologique ind pendante de ses deux domaines SH3 91 en cons quence ces derniers n ont pas a tre inclus dans les mod les Ensuite la liaison Grb2 SH2 ligand fait intervenir une unit Grb2 SH2 et une unit du ligand l inverse la liaison Grb2 SH3 ligand implique en g n ral les deux domaines SH3 ce qui complique la mod lisation au vu d une publication r cente 74 il pourrait toutefois tre n cessaire de reconsid rer ce point Enfin comme allons le voir la forme et le mode de liaison des ligands de Grb2 SH2 est sp cifique et bien d termin e ce qui devrait simplifier les d marches de mises au point de nouveaux inhibiteurs On peut explorer certaines structures issues de la base de donn es PDB pour se faire une id e Le site http www rcsb org pr sente chaque mois une section consacr e aux structures les plus remarquables cliquer sur le lien Molecule of the month de la page d acceuil t Celle ci indique que certains ligands synth tiques pourraient se lier avec deux unit s de Grb2 SH2 Reste d terminer si une telle structure observ e par crystallographie RX caract rise bien le complexe dans les conditions physiologiques Pour ce faire la publication de la structure RMN de ce complexe serait id ale 29 tat actuel des connaissances sur l inhibition de Grb2 SH2 Ligands peptidiques D t
28. acc der aux techniques dites de structure based drug design dont cette th se se fait l cho D autres aspects sont particuli rement importants Quelle est l accessibilit exp rimentale de la cible aussi bien in vitro que in vivo Dans les deux cas il est utile de disposer de protocoles de mesures biologiques de l activit d une mol cule donn e qui soient les plus fiables et accessibles qu il se peut si le co t d un tel test est important cela limitera consid rablement le nombre de mol cules que l on pourra envisager de tester exp rimentalement par la suite Et avant tout toujours dans l optique du drug design est ce que la cible peut tre accessible dans les conditions physiologiques par une mol cule administr e de fa on externe Dans tous les cas toute information initiale sur la cible est susceptible de fournir des moyens d adapter les protocoles de recherche de m dicaments qui s ensuivront L tude de l interactome validation d une cible et tude sa fonction biologique travers l identification du mode d action des prot ines associ es vient compl ter dans ce contexte l tude du g nome identification d une cible partir de l expression des g nes codant pour les prot ines impliqu es Identification des compos s prometteurs hits Une fois la cible pharmaceutique identifi e il faut soit tester un ensemble de mol cules candidates sur cette cible selon un processus q
29. ad nine A thymine T cytosine C et guanine G 0 0 N A FAN M N pes H Adhi ZE SN N JE N SN LT LC lt rt I i 2 NS NT NH 07 Sy H NT a nn 07 Sw TH Adenine A Cytosine C Guanine G Thymine 1 Le squelette pentose phosphate de l ADN et les 4 bases possibles correspondantes Dans I ARN les thymines sont mut es en uraciles et les d oxyriboses du squelette en riboses 8 ADN et ARN ont des fonctions diff rentes de par leur structure La structure de ADN fut d termin e par Watson Crick et Franklin en 1953 54 56 et a pour particularit le positionnement des nucl otides l int rieur d une double h lice form e par deux cha nes pentose phosphate antisym triques Ce positionnement dans un espace si confin est rendu possible par un appariement des nucl otides par paires ad nine et thymine d une part cytosine et guanine d autre part sont reli es d une h lice l autre par des liaisons hydrog ne L information g n tique est ainsi doubl e et stabilis e le r le biologique de conservation tenu par l ADN appara t alors vident L ARN a une structure en simple h lice dans laquelle les nucl otides sont expos s permettant la r plication et l expression directe de la s quence CH l Kini de de N N N ER Oo 4 g 5 e N N ee ol N 4 N H N w d N w d ntt H Adenine A Thymine T Guanine G Cytosine C
30. and precise could lead to the limitations of each method being compensated by features of the others and programs being judged with regard to complementarities rather than to individual performance Such an approach termed consensus scoring 65 66 is already attempted regarding scoring functions alone Its extension to docking techniques is the basis of Glide s hierarchical docking over which unfortunately the user do not have total control Glide being a commercial program Ultimately such an approach should be extended outside the boundaries of SBDD This is the deep seated motive of the VSM G project 67 The drug discovery process can docking and structure based design be integrated more efficiently The drug discovery process which includes CADD can be simplified into several well defined steps One can first start from the chemical space which includes all molecules that a chemist could make The size of this galaxy is crudely estimated to 1060 68 Of course we cannot explore that either experimentally or by using computing What we need is to focus on the biochemical space the ensemble of molecules that are relevant to biology For that purpose we can use pharmacokinetics 69 and various other approaches 70 71 to estimate a drug like probability of molecules Simple empirical guidelines are proposed such as the well known Lipinski s rule of five 72 Molecular databases merging compounds from suppliers and applying b
31. avec la m me facilit que des traitements antiviraux sont certainement profitables sous un angle publicitaire et les acteurs de l industrie pharmaceutique communiquent volontiers sur le sujet On peut envisager que l effort de recherche r ellement consenti soit moindre tant donn que ces entreprises d pensent d sormais plus d argent dans leurs d partements marketing que dans la R amp D 18 R f rences bibliographiques 1 Pierce G B Shikes R and Fink L M Cancer A problem of developmental biology 1978 Prentice Hall Englewood Cliffs N J 2 Jemal A Murray T Ward E Samuels A Tiwari R C Ghafoor A Feuer E J and Thun M J Cancer statistics 2005 CA A Cancer Journal for Clinicians 55 2005 10 30 3 Cancer Pronostics long terme Expertise collective Inserm 2005 Les ditions Inserm 4 Pinell P How do cancer patients express their points of view Sociology of Health amp Illness 9 issue 1 1987 25 44 5 Moulin P Imaginaire social et cancer Revue Francophone de Psycho Oncologie 4 issue 4 2005 261 267 6 Pardee A B Dubrow R Hamlin J L and Kletzen R F Animal cell cycle Annual Review of Biochemistry 47 1978 715 750 7 Yanishevsky R M and Stein G H Regulation of the cell cycle in eucaryotic cells International Review of Cytology A Survey of Cell Biology 69 1981 223 259 8 Sporn M B The war on cancer Lancet 347 1996 1377 1381 9 Cantley L C Auger K R
32. c dent ligand 2 tant plac par superposition avec ligand 1 toutes les mol cules d eau issues de la RX sont supprim es Enfin Grb2 SH2 non li est obtenu en supprimant ligand 2 Les trois types de syst me sont ensuite immerg s dans des bo tes d eau cubiques de 60 ou 80 A neutralis s lectriquement contre ions et proton s pH 7 Trois champs de force sont s lectionn s CFF91 AMBER et CHARMM22 Discover est utilis dans les deux premiers cas NAMD dans le troisi me Les dynamiques P 1 atm T 300 K sont lanc es apr s une phase de minimisation et une phase d quilibration Aucune contrainte n est impos e au syst me mis part la rigidit des liaisons O H des mol cules d eau Nous obtenons sept trajectoires distinctes complexe ligand 1 CFF91 1 ns complexe ligand 1 AMBER 2 ns complexe ligand 1 CHARMM22 2 ns complexe ligand 2 CFF91 1 ns complexe ligand 2 AMBER 1 17 ns complexe ligand 2 CHARMM22 2 ns domaine SH2 non complex CHARMM22 2 ns CO pour validation du mod le du r cepteur 2 t A Lae y j Par exemple des effets de polarisation de grande amplitude Ce type de ph nom nes marque typiquement les limites des m thodes de la m canique mol culaire face celles en g n ral trop co teuses pour des syst mes de cette taille de la m canique quantique Nous avons utilis Discover et NAMD pour effectuer les dynamiques mol cula
33. calculs seront si complexes que l accessibilit du syst me sera la plupart du temps impossible atteindre limitant les simulations des mod les n impliquant qu un petit nombre d atomes lt lt 1000 La simulation de syst mes de grande taille tels que les biomol cules n cessite de rabaisser le r alisme et la repr sentativit des simulations En particulier les effets lectroniques seront particuli rement difficiles caract riser 14 Drug design Etapes de la mise au point d un m dicament Identification de la cible pharmaceutique L identification d une cible pharmaceutique peut se faire par diff rents moyens parmi lesquels viennent en premier lieu les m thodes de la biologie cellulaire et mol culaire Il faut leur ajouter des techniques plus r centes issues de la g nomique et de la bioinformatique Une fois la cible identifi e il convient de rassembler scrupuleusement toutes les informations accessibles 4 son sujet 4 commencer par les informations bibliographiques Une cible peut enfin tre caract ris e de diff rentes fa ons une structure un m canisme biologique une s quence nucl ique ou prot ique etc Chacune de ces caract risations ouvre des possibilit s d tude dans l optique de la d couverte de m dicaments actifs On peut num rer plusieurs exemples d informations utiles diff rents niveaux Ainsi il est essentiel de disposer d une structure fiable de la cible afin d
34. cat gories fonctionnelles de prot ines peuvent tre d finies Le domaine SH3 de Grb2 reconna t sp cifiquement des motifs peptidiques riches en proline tels que ceux pr sents sur Sos 26 Une grande vari t de pathologies peuvent tre meme em ravea ww Tyrosine kinases corr l es au niveau de l interactome a la ABL KI POSER Gleevec ST 571 Approved CML Novas EGFR 201839 ressa Approved lung cancer AstraZeneca d r gulation d une voie de signalisation bien ene ost recap Cnet pr cise En particulier parmi celles qui ont trait a la NT fa fren ee Merck Kok EMD 72000 mAb Phase cancer Merck KgaA r gulation de la croissance cellulaire la Genistein Phase I cancer ner ye i RHS mAb Phase Il cancer York Medical Bioscience Inc surexpression des signaux prot ine kinase est sae heed ose eue aaa particuli rement fr quente Leur inhibition Parts Pres ec ot A x F A VEGFR POGFR FGFR su6668 Phase cancer Pharmacia Corp constitue ainsi une cible th rapeutique POGER F3 rs3s18 Phase cance Milenium Pharmaceuticals gt N VEGFR SU5416 Phase IN cancer Pharmacia Corp extr mement int ressante 47 49 laquelle wemmer _ Prase I mc bit l industrie pharmaceutique accorde une grande wmm pr feet Emo copa g A 3 sent HER 2 neu 17 AAG Phase cancer Kosan importance ci contre s lection d inhibiteurs de Trastuzumab mj Approved cancer Genetech 2C4 mAb Phase cancer Gen
35. code is 1JYQ Typical peptidic or pseudo peptidic ligands for Grb2 SH2 26 34 36 including the natural recognized sequences pYXNX have a similar binding mode to Grb2 SH2 than ligand 1 indeed all of those ligands present a pTyr residue bound to cavity 1 and an Asn residue bound to cavity 2 The binding of Grb2 SH2 ligands is mainly based on these two interactions Ligand 1 specificity relies on the optimization of the mAz and a Me pTyr residues which does not alter the binding of pTyro and Asn Consequently it can be said that proposals of novel drug design strategies for Grb2 SH2 inhibition deriving from the study of ligand 1 would also be of interest when applied to most Grb2 SH2 ligands Additional work was conducted in order to further increase the affinity provided by ligand 1 Molecular modelling studies 37 were performed in order to design novel compounds by optimizing the corresponding interactions starting from the published X ray structure of ligand 1 complex 26 PDB entry 1JYQ A new ligand hereby noted ligand 2 figure 1 b was thus designed and synthesized and its efficiency was measured using ELISA 38 tests However contrary to in vacuo MD predictions ligand 2 was shown to be experimentally unable to inhibit Grb2 As these calculations were made without taking any water molecule into account assuming that solvent interactions with the protein ligand system were not crucial we made the hypothesis that the discr
36. d tre l origine d erreurs parfois difficilement d celables Les recherches th oriques que nous voulons conduire s orientent la fois vers la mise en place d infrastructures combinant les point forts des diff rentes m thodes disponibles plut t que de chercher syst matiquement en cr er de nouvelles et vers l emploi de protocoles de validation plus stricts ne se limitant pas la seule reproduction d un petit nombre de complexes exp rimentaux de r f rence Un objectif long terme consiste garantir qu un screening virtuel puisse identifier tous les hits d une base de mol cules pour une cible et non plus au moins un avec une bonne probabilit comme actuellement 56 Le projet VSM G Cette partie d crit bri vement le projet VSM G Une description plus d taill e de VSM G ainsi que son protocole de validation sont pr sents dans l article 2 Multiple step virtual screening using VSM G overview and validation of fast geometrical matching enrichment GENERATION SELECTION Pr sentation Comme nous venons de le voir les techniques de screening virtuel pr sentent une h t rog n it tr s importante qui s exprime au niveau de leurs conditions d utilisation possibilit s d application et pertinence Cette caract ristique fondamentale peut s av rer tre b n fique d s lors qu elle est explicitement prise en compte afin qu un besoin sp cifique l approche la p
37. de force comme cause de la non repr sentativit des pr c dentes simulations est invalid e du moins pour les trois champs de force employ s Validation du protocole L analyse de la trajectoire de la prot ine Grb2 non complex e avec CHARMM272 permet de v rifier la pertinence du protocole employ La g om trie de la prot ine volue de fa on coh rente la conformation des r sidus du site actif volue afin de maximiser les interactions intramol culaires ce qui compense le retrait des interactions prot ine ligand Aucun changement notable n est observ en dehors du site actif Ci dessous cluster graphs des trajectoires obtenues Deux graphes sont pr sent s pour Grb2 SH2 non complex e avec CHARMM22 celui du haut correspond Grb2 SH2 complet tandis que celui du bas correspond au site actif seul CHARMM22 ligand 1 ligand 2 uncomplexed 50 Nouvelles connaissances concernant l interaction de ligands sur le r cepteur Grb2 SH2 Effets du solvant La mod lisation explicite du solvant constituant le seul ajout m thodologique au protocole des simulations le fait que celles ci soient d sormais en conformit avec les observations exp rimentales accr dite Vhypoth se d un r le actif de l eau au niveau de la liaison de ligands sur le domaine SH2 de Grb2 Dans le cas de la dynamique du complexe Grb2 SH2 ligand 1 les mol cules d eau issues de la structure cristallographique ne s av rent
38. de modules Le d veloppement de VSM G n en est qu ses d buts On trouvera toutefois dans l article un exemple d application suffisamment probant pour justifier les investissements consentis jusqu pr sent tout en soulignant la n cessit de poursuivre cet effort Les objectifs directs se situent au niveau 1 de l utilisation des m thodes de calcul massivement distribu es grilles de calcul r seaux h t rog nes 2 de l int gration de l ensemble du spectre des m thodes de la chemoinformatique et de la bioinformatique m thodes statistiques et empiriques fouille de donn es et 3 vers une conception plus modulaire les modules du funnel sont pour le moment cod s en dur alors que leur impl mentation devrait se faire au niveau de l interface utilisateur Perspectives long terme vers un screening virtuel intelligent Chaque technique de screening virtuel est susceptible de produire un certain taux de faux positifs et de faux n gatifs Si un faux positif constitue une simple perte de temps car il sera priori d tect ult rieurement en particulier dans le contexte du funnel impl ment par VSM G l exclusion tort d une mol cule est bien plus probl matique car et priori ind celable Une volution conceptuelle de VSM G devra donc consister substituer au m canisme de filtrages successifs par les modules du funnel un syst me de prioritisation compl t par des algorithm
39. des s quences d environ 100 acides amin s d abord observ es en commun dans les oncoprot ines Src et Fps 20 puis ensuite dans une grande vari t de prot ines intra cellulaires 21 telles que Fyn PI3K p85a p56 Lek Syk Sap Grb2 soit plus de 100 prot ines dans le g nome humain 22 La conformation des domaines SH2 correspond deux h lices encadrant plusieurs feuillets 8 centraux anti parall les voir ci contre La s quence de la plupart des domaines SH2 suit une g om trie BaBBBBBaB on d compose alors la s quence en zones g om triques not es successivement BA AA aA AB BB BC BC CD BD 23 Ras est un oncogene qui a t historiquement isol sur des rats et est associ au virus Sarcoma d o son sigle MAPK signifie mitogen activated protein kinase La voie de signalisation Ras MAPK est parfois d sign e diff remment Ras kinase Ras Erk Ras Raf Mek Erk MAPK Ras peut d signer aussi bien la prot ine qui intervient dans la voie de signalisation Ras MAPK que le g ne qui code pour la production de cette prot ine Ici les mutations du g ne dont on parle sont bien entendu corr l es 4 des mutations sur la prot ine concernant principalement les r sidus 12 13 et 61 Ces prot ines n ont pas n cessairement les m mes fonctions et le domaine non plus s il est expos La s quence des domaines SH2 de Cbl et STATI a une g om trie aBBBo 24 La particu
40. discrepancy between experimental results and MD simulations is the force field choice for the dynamics All the simulations presented in this paper have been performed using CHARMM22 and the NAMD program The initial in vacuo simulations that prompted for ligand 2 synthesis and testing prior to this work were performed using CFF91 and the Discover program Accelrys In order to have a standardized protocol we recreated those simulations using CHARMM22 CFF91 and CHARMM22 in vacuo dynamics were in agreement predicting that both ligand 1 and ligand 2 binding were stable on the Grb2 SH2 receptor The MD simulations in water for ligand 1 and ligand 2 complexes were also performed using the CFF91 and AMBER force fields with Discover and parameters close to the ones employed with NAMD and CHARMM22 The results were again in agreement between the force fields predicting that when water is explicitly modelled ligand 1 is binds to Grb2 SH2 in a stable manner while ligand 2 does not Structural analysis for both ligand 1 and ligand 2 solvated complexes showed conformations similar to those described previously for the CHARMM22 dynamics These calculations allowed us to verify that for the system under investigation the choice of the force field is not crucial while the modelling of the solvent is Discussion The importance of modelling water molecules Until recently upon modelling of protein ligand complexes by docking strategies solvent molecules were
41. doit donc se limiter l emploi de m thodes statistiques sur des bases de mol cules actives connues ce qui est tr s limit tant la portion de l interactome couverte par ces mol cules est restreinte Contrainte de stabilit Dans l organisme les r actions de phosphorylation et de d phosphorylation sont courantes ce qui pose probl me au niveau des inhibiteurs de Grb2 SH2 disposant d un r sidu pTyr ou de tout autre r sidu phosphat celui ci est susceptible d tre d phosphoryl par les phosphatases le ligand perdrait alors une grande partie de son affinit pour Grb2 SH2 De fa on int ressante l tude par screening r alis e par Hart et al 96 sur les s quences non phosphoryl es de type X 5 YoXN 2 X 10 Si elle confirme que la d phosphorylation aboutit une baisse importante de l affinit du ligand de un trois ordres de grandeur semble indiquer galement que l absence du groupe phosphate en position 0 a pour cons quence de d placer la s lectivit sur les autres r sidus du ligand avec en particulier pour un Glu et de fa on moins notable aux positions 4 5 et 6 Cela pourrait traduire un mode de liaison l g rement diff rent des s quences peptidiques non phosphat es et souligne l importance des groupes distants dans la mise au point de ligands pour Grb2 SH2 dans lesquels le groupe pTyr est substitu 34 Contrainte d accessibilit Il existe des r gles g n ra
42. drop of approximately 100 kcal mol Such a disruption of ligand 2 binding is not compensated by a strengthening of the remaining interactions Thus the interaction of ligand 2 with cavity 1 is finally much weaker than the one of ligand 1 Regarding the interactions of the residues forming cavity 3 the values obtained with ligand 2 give more details about the large energy loss discussed previously and observed in figures 7 b 8 b and 9 The interactions with the S14 Ser BG and Nj43 Asn BG residues are lost while the interaction with Ry Arg BG is conserved as observed with the analysis of the conformations Thus until the Kioo Lys BD interaction in cavity 1 is lost at 1800 ps the weakening of the binding energy between ligand 2 and Grb2 SH2 is linked to the loss of well defined H bonds whereas these are maintained throughout the 2 ns of simulation in the case of ligand 1 Interestingly during the simulation of the ligand 1 complex the Nj43 Asn BG interaction with ligand 1 undergoes significant fluctuations without being lost This could indicate that this interaction is prone to be disrupted upon taking solvent effects into account but is maintained with the help of the interactions with S 4 Ser BG and Ry Arg BG which both remain stable However the ligand 2 case indicates that the main Ry42 Arg BG interaction could not be of enough amplitude so as to alone maintain the ligand in cavity 3 as it is also strongly interacting with water
43. en bleu sont issues de la base cibl e eDAM on remarque ligand 1 situ en seconde position en haut droite celles encadr es en jaune de la base Yao pharmacop e chinoise enfin la mol cule encar e en vert provient de la base ZINC 54 Validation En tant que tels ces r sultats de docking n ont qu une faible signification En l absence de tests exp rimentaux des structures sugg r es le protocole de docking doit tre valid sur la base de r sultats existants avant d envisager le test d un plus grand nombre de mol cules Nous disposons pour ce faire de r sultats d taill s d exp riences de screening r alis es sur diff rents domaines SH2 dont Grb2 Les domaines SH2 reconnaissant tous une ou plusieurs s quences du type pYXXX le screening des 8000 combinaisons possibles a t effectu et les pr f rences pr cises chaque position 1 2 et 3 sont connues Le protocole de validation est bas sur 4 structures Code Prot ine s quences Ligand mode de RX de pr cision suffisante de r cepteurs SH2 PDS a a ert anew complex s Les 8000 peptides pYXXX sont dock s 1S0 Src pYEED pYEEI lin aire sur chacun de ces r cepteurs sous deux formes 1P13 pYANF 8 turn avec et sans conservation des mol cules d eau ILKK P56 Lck pYEEI pYEEI lin aire cristallographiques soit un total de 64000 dockings 1JYQ Grb2 pYVNV pYENW
44. enthalpic contributions to the free energy of binding are of almost the same magnitude for a hydrogen bond at 300 K However when a significant number of water molecules are found in several experimental structures interposed between the ligand and the receptor making multiple H bond bridges one should investigate further in order to determine if such waters would conserve their position in biological conditions and therefore be structurally involved in ligand binding On the opposite if no such water is observed in X ray structures most probably no water molecule is involved in protein ligand binding because the inclusion of a water molecule from the solvent is more entropy costly at 300 K than in X ray conditions With the X ray structures of complexes of SH2 domains 26 34 36 we are in the former situation one water molecule is observed in all cases at the bottom of cavity 1 forming H bond bridges with one of the phosphate oxygens and the receptor The X ray structure of Grb2 SH2 complexed by ligand 1 26 also embodies an additional number of bridging waters MD shows that the presence of such waters is not due to crystallization effects since one of them appears to maintain its position during the simulation In addition upon exchange with waters from the bulk the positions of several other water molecules are conserved This confirms that specific significant stabilizing water mediated interactions occur at the interface between Grb2 SH2 and
45. generally not taken into account and were believed not to exert a significant effect on the binding Inclusion of some water molecules in the molecular complex could have two opposite effects i a favorable effect on the binding enthalpy due to the formation of additional hydrogen bonds and ii an unfavorable entropic effect due to the extraction of the water molecules from the bulk restraining their available number of degrees of freedom These two effects being in most cases considered of comparable impact their sum is assumed to be negligible regarding the free energy of binding and have a similar magnitude in all structures considered Additionally a good inhibitor for a given receptor is usually supposed to remove water molecules initially present on the binding site While generally this assumption is correct 49 it has been demonstrated that in some cases specific water molecules are to be considered as part of the receptor and are better kept conserved upon ligand binding as they provide multiple stabilizing H bonds that compensate for the absence of the entropic gain that could occur upon removal 50 The removal of water molecules from the modelled system was demonstrated here to be misleading in the case of Grb2 SH2 because only explicit inclusion of the water environment of two reference complexes was able to give results in agreement with experimental measurements Moreover the results of this study clearly indicate that the involv
46. ines 63 a un int r t vident mais la t che s av re si complexe qu elle en constitue une sorte de saint Graal de la biochimie Ainsi alors que l tude du g nome correspond fondamentalement l exploration d un espace s quentiel sur quatre bases celui du prot ome en plus de l extension d une partie de cet espace cette fois cod sur 20 bases lui ajoute un espace conformationnel beaucoup plus vaste 64 L exploration de ce prot ome est tr s d licate et ce m me si on ne s int resse qu une prot ine bien d finie moins qu une structure exp rimentale de celle ci soit disponible ce qui fut le cas pour ce travail Les techniques de la mod lisation mol culaire peuvent alors tre mises en uvre Par exemple les r sidus Met et Glu se retrouvent plus fr quemment dans des h lices a Val et Ile dans des feuillets B Gly et Pro dans les tours L Arg est le seul acide amin qui ne semble pas avoir de pr f rence particuli re On passe alors de la structure primaire lin aire des prot ines qui se limite la s quence la structure secondaire globulaire On parle de structure tertiaire lorsque l on consid re en plus les interactions longue distance sur les cha nes peptidiques qui entre autres diff rencient structuralement les r sidus hydrophiles des r sidus hydrophobes Enfin la structure quaternaire se r f re l organisation de cha nes peptidique
47. issue 11 2001 1617 1620 129 Shi Z D Lee K Wei C Q Roberts L R Worthy K M Fisher R J and Burke Jr T R Synthesis of a 5 methylindolyl containing macrocycle that displays ultrapotent Grb2 SH2 domain binding affinity Journal of Medicinal Chemistry 47 2004 788 791 130 Oishi S Shi Z D Worthy K M Bindu L K Fisher R J and Burke T R Ring closing metathesis of C terminal allylglycine residues with an N terminal B vinyl substituted phosphotyrosyl mimetic as an approach to novel Grb2 SH2 domain binding macrocycles Chemistry amp Biochemistry 6 2005 668 674 131 Oishi S Karki R G Shi Z D Worthy K M Bindu L Chertov O Esposito D Frank P Gillette W K Maderia M Hartley J Nicklaus M C Barchi Jr J J Fisher R J and Burke Jr T R Evaluation of macrocyclic Grb2 SH2 domain binding peptide mimetics prepared by ring closing metathesis of C terminal allylglycines with an N terminal B vinyl substituted phosphotyrosyl mimetic Bioorganic amp Medicinal Chemistry Letters 13 2005 2431 2438 132 Shi Z D Wei C Q Lee K Liu H Zhang M Araki T Roberts L R Worthy K M Fisher R J Neel B G Kelley J A Yang D and Burke Jr T R Macrocyclization in the design of non phosphorus containing Grb2 SH2 domain binding ligands Journal of Medicinal Chemistry 47 2004 2166 2169 133 Lee K Zhang M Liu H Yang D and Burke Jr T R Utilization of a B aminophosphotyrosyl mimetic in
48. le cas de Grb2 o seul le second est possible WEEGKIPRAKABEMES KQRHDGABDIRESESAPGDESESVKE GN DVOHERVERDGAGRYEINV VRDENSINELVDYHRSTSVSRNQO LRO WNVGSSNENKAENEER GKR DGHEDVRES skocCYACSVWV vc SVKHCVINKTATCYGFAEP NLYSSEKELVEHYOHTsLVoHNDSLNVEE WEEKNLSRKDAEROLLA PGNTHGSELIRESES TAGSPSLSVRDF DONOGEVWKHYKIRNLEDNGGFYISPRIT FPGLHELVRHYTNASD GLCTRLSRE WYECK1TRRESERULLNAENPRCTELVRESETTKCAYCUSVSDF DNAKGLNVKHYKIRKEDSCGEYETSRIQUENSLOOLVAYYSKHAD GLCHRLTEV S quences des domaines SH2 de de haut en bas Grb2 PI3K p85a P56 Lck et C Sre En jaune r sidus formant une h lice a en bleu r sidus formant un feuillet f En orange r sidus fondamentaux pour la reconnaissance des ligands i L importance de ce motif nous semble toutefois sur estim e car il diverge sur un certains nombre de domaines SH2 par exemple avec Cbl on a YIFRLS avec Syk FLIRAR avec SAP YLLRDS sans pour autant que la s lectivit de ces domaines SH2 pour les phosphotyrosines soit remise en question L analyse des structures RX montre que seul le r sidu arginine de FLIRES conserv dans tous les domaines SH2 est vraiment crucial ainsi dans une moindre mesure la s rine et une autre arginine sur l h lice aA conserv es dans la plupart des cas Il est donc sans doute plus juste de d finir plus g n ralement la sp cificit pTyr des domaines SH2 comme la r sultante de la pr sence de plusieurs r sidus arginine et s rine accessibles dans une cavit
49. ligand 1 Thus the very strong affinity of ligand 1 compared to other known ligands could be assumed to rely not only on the targeting of cavity 3 but also on the involvement of those interfacial water molecules stabilizing the protein ligand interaction by means of several additional H bonds overcompensating for their loss of degrees of freedom upon binding Indeed this specificity as well as cavity 3 targeting could distinguish ligand 1 amongst other known inhibitors of Grb2 SH2 signalling and thus those waters should be conserved upon modelling ligand 1 analogues The present study only accounting for ligand 1 and ligand 2 binding we cannot predict if they should also be conserved upon modelling other Grb2 SH2 ligands whose structure differ significantly However new methods should enable starting from the structure of Grb2 SH2 including all possibly stabilized water molecules upon ligand binding to sort those waters that should be conserved upon docking a given molecule 52 Disruption of ligand 2 binding by bulk water molecules Independently from the stabilizing effect of a limited number of water molecules located at the protein ligand interface MD of both ligand 1 and ligand 2 complexes highlighted the destabilizing effects of bulk solvent It first appears that bulk solvent can force specific ligand groups to keep enough solvent accessibility resulting in a conformational change that could be unfavorable In this study this is observ
50. ligand 1 and ligand 2 complexes in vacuo uncomplexed Grb2 SH2 in 80 A waterbox is 2 ns A conformation was recorded each 5 ps except for the uncomplexed Grb2 SH2 MD 10 ps giving 400 conformations per trajectory 200 for uncomplexed Grb2 SH2 During the dynamics it was verified that Grb2 SH2 did not undergo abnormal structural changes that the size of the unit cell did not fluctuate much and that the counter ions always stayed out of reach gt 10 A from Grb2 SH2 and thus did not directly influence its dynamics Analysis of the MD trajectories Analysis of the MD data was performed either using NAMD or InsightII Accelrys In the latter case we had to make an in house conversion program to obtain arc format trajectory files from the dcd format outputted by NAMD We also had to truncate the resulting arc files that were too big gt 2 Gb for the solvated complex trajectories for InsightII to load them Consequently the snapshots of ligands conformations obtained with the protein represented with its Connolly surface and colored upon its distance to the ligand using an InsightII script were generated from arc files with all water molecules suppressed For the generation of cluster graphs only the ligands were retained For the analysis of the position of water molecules regarding the protein ligand interaction we programmed a filter that works as follows i In the case of the ligand 1 complex all waters retained from the X ray struc
51. ligand groups were next computed We retained i the 4 phosphate oxygens of the two phosphates of ligand 1 the 3 phosphone oxygens on the corresponding phosphones of ligand 2 which are the only atoms of these groups that make significant favorable interactions ii the Asn 2 side chain bound to cavity 2 iii the aromatic group in 1 position that was modelled in order to make stacking interactions and iv the backbones of the ligands peptidic in the case of ligand 1 non peptidic and cyclic in the case of ligand 2 These results are presented as energy bars in figure 8 a for ligand 1 and in figure 8 b for ligand 2 Figure 8 Interaction energy between various ligands atoms groups and Grb2 SH2 a ligand 1 complex left b ligand 2 complex right pY 0 OH n a pY 0 01 pY 0 02 pY 0 03 pY 1 OH n a pY 1 01 pY 1 02 pY 1 03 J j ia pY 2 pY 1 TRE backbone 500 1000 1500 500 1000 1500 ps 2000 0 coe 2000 o erao PE lt 75 121m0 0 The ligand 1 pTyro phosphate oxygens in agreement with the structural analysis are bound more tightly to cavity 1 than their pTyr counterparts in cavity 3 In both cases it appears that the non terminal phosphate oxygens make interactions equivalent to those of the terminal ones which is surprising because only the latter are H bonded to Grb2 SH2 as predicted by the X ray structure In the case of ligand 2 as expected the oxygens of both phosphones are equivalent and their interacti
52. ment un screening exp rimental son usage s av re assez simple et une telle approche est devenue assez r pandue car elle permet souvent d obtenir moindre co t une quantit d informations suffisante lors des premi res phases d une recherche de m dicaments la d couverte de hits Lors des phases plus avanc es les techniques de screening virtuel peuvent apporter un enrichissement des connaissances Par exemple sur la base d un hit il peut tre proc d la g n ration d une base de d riv s par chimie combinatoire virtuelle Cette base peut ensuite tre screen e par exemple en suivant les protocoles classiques de docking d j employ s afin de d terminer les hits Une analyse statistique des r sultats peut enfin apporter des pr cisions permettant d optimiser les structures int ressantes s int grant ainsi parfaitement aux diff rentes m thodes de la phase de lead optimization Ainsi les techniques de screening virtuel sont consid r es comme efficaces lorsqu elles accompagnent d autres approches elles n ont alors pas un r le central dans un protocole de recherche de m dicaments Pour que cela puisse tre le cas il faudrait qu elles puissent combiner efficacit et fiabilit Si des m thodes th oriques relativement fiables existent elles sont d licates mettre en uvre Par exemple les simulations de dynamique mol culaire n cessitent des informations structurales pr cises souvent d li
53. mes biomol culaires en g n ral la fois complexes chimiquement et difficilement accessibles exp rimentalement les moyens informatiques sont dans un tel contexte indispensables Le champ d application des techniques de mod lisation mol culaire est immense l image du spectre des connaissances et des domaines scientifiques qui sont susceptibles d intervenir Nous nous limiterons aux syst mes que nous nous appr tons tudier de type biologique La mod lisation mol culaire peut tre pr cieuse dans l identification de cibles th rapeutiques typiquement des r cepteurs prot iques dont le r le sp cifique dans une pathologie est caract ris Elle peut s av rer ensuite indispensable lors de la mise au point d inhibiteurs chimiques pour cette cible Un bon exemple du r le positif de telles m thodes dites de structure based drug design 74 est constitu par la mise au point d inhibiteurs de l HIV prot ase 75 Des techniques telles que la mod lisation par homologie peuvent permettre de pallier l absence de structures exp rimentales compl tes dans le cas de certaines cibles d int r t majeur telles que les r cepteurs de type GPCR 76 77 De nombreuses approches sont disponibles et peuvent tre combin es afin de produire des r sultats scientifiques de grand int r t en particulier sur le plan m dical 78 La mod lisation peut plus g n ralement profiter directement des progr s dans une gran
54. molecules This leads to the ligand 2 phosphone moving towards the bulk simultaneously with R 4 Arg BG Behavior of specific water molecules along the trajectories The positions of water molecules interposed between ligand 1 and Grb2 SH2 as depicted in the X ray structure were tracked during the MD trajectory It appeared that only one of those waters HOH324 in 1JYQ monomer B did maintain itself interacting with both Grb2 SH2 and ligand 1 during the whole trajectory HOH position in cavity 1 is fluctuating and differs slightly from its conformation in the X ray structure where it is H bonded to the terminal atoms of residues Fos Phe BC Sos Ser BC and Kio Lys BD During the MD simulation HOH324 makes H bonds mostly with the non terminal phosphate oxygen and the Kyjo9 Lys BD residue terminal nitrogen but also with the terminal oxygens of the three cavity 1 serines and the Ao Ala BC residue The other water molecules from X ray quickly moved in the bulk the most stable staying 100 ps and 350 ps Throughout the trajectory several water molecules from the water box retained stable interactions with Grb2 residues but did not interact significantly with ligand 1 with the exception of one water molecule that bridged Sos Ser BC and Fios Phe BD while remaining around 4 A from the phosphates Regarding water molecules exchanged at conserved positions it is noted that in cavity 3 the phosphate is always H bonded to at least two w
55. molsoft com 137 Cavasotto C N Kovacs J A and Abagyan R A Representing receptor flexibility in ligand docking through relevant normal modes Journal of the American Chemical Society 127 issue 26 2005 9632 9640 138 Fernandez Recio J Abagyan R and Totrov M Improving CAPRI predictions optimized desolvation for rigid body docking Proteins Structure Function and Bioinformatics 60 issue 2 2005 308 313 139 Totrov M and Abagyan R Flexible protein ligand docking by docking energy optimization in internal coordinates Proteins Structure Function and Genetics Suppl 1 1997 215 220 140 Freisner R A Banks J L Murphy R B Halgren T A Klicic J J Mainz D T Repasky M P Kmoll E H Shelley M Perry J K Shaw D E Francis P and Shenkin P S Glide A new approach for rapid accurate docking and scoring 1 Method and assessment of docking accuracy Journal of Medicinal Chemistry 47 issue 7 2004 1739 1749 141 Halgren T A Murphy R B Freisner R A Beard H S Frye L L Pollard W T and Banks J L Glide A new approach for rapid accurate docking and scoring 2 Enrichment factors in database screening Journal of Medicinal Chemistry 47 issue 7 2004 1750 1759 142 Glide Grid based LIgand Docking with Energetics http www schrodinger com ProductDescription php m 143 Maestro Unified interface for Schr dinger software http www schrodinger com ProductDescription php m ID 6 amp
56. multiple step procedure Thus even considering the limitations of our validation test results are clear enough to demonstrate that SHEF and by extension its association as the first module in the VSM G screening protocol can actually be useful for in silico drug discovery This paper has highlighted precisely the conditions for obtaining good performance from MSSH SHEF It appears that for flexible receptors prone to induced fit effects upon complexation a filtering based on a consensus ranking of SHEF results for multiple target conformers should be favored More importantly basic information regarding the types of interactions involved in ligand binding is crucial for deciding if MSSH SHEF should be used and if so to what extent Enrichment can only be expected when binding is not largely dominated by chemical interactions such as electrostatic effects or hydrogen bonding Active sites that are known to favor hydrophobic interactions might be targets of choice for a structure based drug design strategy involving MSSH SHEF as part of a multiple step VS procedure set up using the VSM G program Limitations of the spherical harmonics based geometrical matching procedure have been pointed out As with all structure based in silico techniques there are two fundamental aspects of how the protein ligand binding is modeled Firstly the way search space is defined and secondly how this space is explored An improvement of SHEF in the first area would inv
57. on peut rencontrer en utilisant ces m thodes ont t mis en vidence Lors de l tape d optimisation de mol cules de r f rence l usage de m thodes pouss es telles que la dynamique mol culaire est un choix naturel pour le mod lisateur On s attend g n ralement par ce biais d crire de fa on pr cise l interaction entre deux compos s d int r t biologique et la question de savoir si cette interaction est suffisante est souvent r solue ce stade de la recherche Cette tude a pourtant d but par une contradiction ce niveau pr cis Il a t possible de surmonter ce probl me en mettant en vidence les effets de solvant qui pour la cible Grb2 SH2 et un ligand en particulier s av rent cruciaux La prise en compte de ce param tre ouvre la voie des tudes ult rieures plus fiables d un tel syst me d autant plus que le protocole de dynamique mol culaire mis en place pour Grb2 SH2 que l on peut consid rer comme satisfaisant en l tat actuel de nos connaissances est ais ment r utilisable Les tudes de docking ne prennent en compte le solvant que de fa on tr s superficielle Cet aspect n est pas n cessairement redhibitoire tant donn que la technique utilis e est par nature approch e bien d autres niveaux Au final si des pistes sont donn es pour la mise au point de nouveaux inhibiteurs pour Grb2 SH2 le protocole employ n a pu tre valid si bien que contrairement
58. pancr atiques La r union de biologistes pharmacologues chimistes et physicochimistes acad miques et du secteur industriel d sireux de mettre en commun leurs comp tences va permettre dans un premier temps de concevoir des agents peptidiques comp titeurs des interactions cibl es et de les valider par des tests cellulaires et vivo en les vectorisant puis de proc der des tudes structurales des complexes entre prot ine et peptide inhibiteur afin de d finir les points d interaction pour concevoir grace l aide de la mod lisation mol culaire in silico des mol cules non peptidiques selon une approche de chimie combinatoire cibl e L tude th orique du vivant l chelle mol culaire La lutte contre le cancer comme objectif de la recherche scientifique s inscrit dans le cadre plus g n ral des sciences du vivant On entend par l un ensemble de disciplines scientifiques telles que la biochimie et la g n tique issues des progr s r alis s au XX si cle autour de la biologie et dont les applications fondent d sormais une tr s grande part des d couvertes en pharmacologie et en m decine Les sciences du vivant reposent avant tout au niveau th orique sur la connaissance du fonctionnement d un organisme vivant Une des approches possibles pour d crire les m canismes correspondants et sur laquelle ce travail repose est celle de la biochimie pour laquelle les m canismes de base sont des int
59. pas li es fortement Grb2 ou au ligand tandis que le positionnement de certaines d entre elles dans leur voisinage est conserv Cela semble traduire la stabilit du complexe plut t que l expression d un r le structural notable de ces mol cules d eau tant donn que le mode de liaison du ligand s av re identique celui pr dit ant rieurement in vacuo et conforme la structure RX L analyse de la trajectoire de la dynamique du complexe ligand 2 r v le quant elle que Le premier groupe phosphat a un mode de liaison alternatif qui n est pas enti rement localis dans la cavit 1 voir article et figure page 33 En effet si le r sidu Rgs est bien cibl l interaction avec R est consid rablement affaiblie au profit de Kio constituante de la cavit 2 Cette derni re interaction s annule t 1 8 ns sans que le mode de liaison de ligand 1 impliquant R amp soit r obtenu voir figure 10 de l article La liaison du second groupe phosphat dans cavit 3 se fait exclusivement avec le r sidu Ry42 Cela permet t 1 ns un changement conformationnel d favorable bien visible sur la figure 5 au cours duquel R 4 se d plie vers le solvant en entra nant ligand 2 figure 6 Ce mouvement concert optimise cette interaction au d triment de celles impliquant S 4 et Ny43 figure 10 au final environ 100 kcal mol d interactions avec ligand 2 sont transf r s de Grb2 SH2 vers le solvant fi
60. point in proteomics The role of docking in computer aided drug design Computer aided drug design CADD techniques can be broadly classified by defining two categories 22 structure based drug design SBDD 23 24 and ligand based drug design LBDD 25 When studying the interaction of compounds with a given target of biological interest the prerequisite for SBDD is the three dimensional structure of the biomolecular target SBDD is thus often referred to target based or receptor based drug design For screening purposes docking 26 30 is by far the most popular SBDD method When this information is not available one can resort to homology modelling 31 if enough structural data is known from analogues of the target If this is not possible LBDD techniques focussed on the ligands only often constitute an alternative choice 25 The most known method used in that area is QSAR 32 33 which describes molecules through a set of physico chemical descriptors generally related to the ligand structure from which statistical analysis could show relationships with biological activity and ultimately model a reliable empirical formula It can be noted that both docking and QSAR often constitute computational challenges as their search space geometrical degrees of freedom or descriptors is very large For that reasons complex algorithms are often used e g genetic algorithms GA 34 in the case of docking and neural networks 35 36 for
61. precision Glide Docking and scoring incorporating a model of hydrophobic enclosure for protein ligand complexes Journal of Medicinal Chemistry 49 issue 21 2006 6177 6196 97 Swiss Prot Protein knowledgebase http www expasy org sprot 98 Allen F H Bellard S Brice M D Cartwright B A Doubleday A Higgs H Hummelink T Hummelink Peters B G Kennard O Motherwell W D S Rodgers J R and Watson D G The Cambridge Crystallographic Data Centre Computer based search retrieval analysis and display of information Acta Crystallographica B 35 1979 2331 2339 99 Bairoch A Serendipity in bioinformatics the tribulations of a Swiss bioinformatician through exciting times Bioinformatics 16 issue 1 2000 48 64 100 Allen F H The Cambridge Structural Database a quarter of a million crystal structures and rising Acta Crystallographica B 58 2002 380 388 101 Boeckmann B Bairoch A Apweiler R Blatter M C Estreicher A Gasteiger E Martin M J Michoud K O Donovan C Phan I Pilbout S and Schneider M The SWISS PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Research 31 issue 1 2003 365 370 102 Yang G Desvignes M D Smail Tabone M and Maigret B TLdb Target Ligand database Manuscript in preparation 2006 103 CSD Cambridge Structural Database http www ccdc cam ac uk products csd 104 Stoica I Morris R Karger D Kaashoek M F and Balakris
62. q of a molecular database of n structures as the ratio between the number of hit compounds and the total number of structures _ N hits n q The enrichment e of a database by a filtering process and for a given filtering ratio f 0 lt f lt 1 f being the amount of filtered out candidates can be defined as the ratio between the quality of the reduced database and the quality of the initial database aS O Enrichment is commonly used to evaluate the efficiency of molecular database method By definition random selection does not affect quality so its efficiency is 1 for any filtering amount The maximum enrichment that can be obtained for a given filtering level is when all hits are retained which corresponds to em f F The filtering efficiency E is eventually defined as the relative distance of the filtering method from random filtering E 0 to maximum enrichment E 1 e f 1 eS ETE Figure 7 Density plots between rankings The 6 first plots a b c d e f depict the relationships between the different target conformations for SHEF a b c and GOLD d e f Target conformation influence on these two programs can therefore be observed The 4 last plots g h i j show the relationship between SHEF and GOLD results for the three target conformations g h i then using multiple target rankings j The scale is set so that the average 5 block density is 8383 400 21 Further explanations on
63. rences a t jug e trop faible pour convaincre des performances de filtrage de VSM G Cette ancienne version de l article est disponible sur demande Article 3 Leroux V Maigret B Should structure based virtual screening techniques be used more extensively in modern drug discovery Computers and Applied Chemistry 24 2007 1 10 61 62 Role of water molecules for binding inhibitors in the SH2 domain of Grb2 a molecular dynamics study Vincent Leroux Nohad Gresh Wang Qing Liu Christiane Garbay Bernard Maigret 1 Universit Henri Poincar Nancy I UMR CNRS UHP 7565 groupe eDAM BP 239 54506 Vandceuvre les Nancy Cedex France 2 Universit Ren Descartes Paris V UFR Biom dicale Laboratoire de Pharmacochimie Mol culaire et Cellulaire 75006 Paris France Inserm U648 75006 Paris France CNRS FRE 2718 75006 Paris France Corresponding author Bernard Maigret edam uhp nancy fr http www edam uhp nancy fr Keywords Grb2 SH2 Molecular Dynamics solvent effects water bridging ligand binding drug design Abstract Growth factor receptor bound protein 2 Grb2 plays an essential role in the Ras MAPK signalling pathway which is an important target for anti cancer drug design The precise mechanisms by which effective and selective ligands can bind to the Src homology 2 SH2 domain of Grb2 and interrupt the signalling pathway are not fully understood We report in this paper the results of
64. the design and synthesis of macrocyclic Grb2 SH2 domain binding peptides Journal of Medicinal Chemistry 46 2003 2621 2630 134 Shi Z D Lee K Liu H Zhang M Roberts L R Worthy K M Fivash M J Fisher R J Yang D and Burke Jr T R A novel macrocyclic tetrapeptide mimetic that exhibits low picomolar Grb2 SH2 domain binding affinity Biochemical and Biophysical Research Communications 310 2003 378 383 135 Sayos J Wu C Morra M Wang N Zhang X Allen D van Schaik S Notarangelo L Geha R Roncarolo M G Oettgen H de Vries J E Aversa G and Terhorst C The X linked lymphoproliferative disease gene product SAP regulates signals induced through the co receptor SLAM Nature 395 issue 6701 1998 441 444 136 Morra M Simarro Grande M Martin M Chen A S Lanyi A Silander O Calpe S Davis J Pawson T Eck M J Sumegi J Engel P Li S C and Terhorst C Characterization of SH2D1A missense mutations identified in X linked lymphoproliferative disease patients Journal of Biological Chemistry 276 issue 39 2001 36809 36816 137 Poy F Yaffe M B Sayos J Saxena K Morra M Sumegi J Cantley L C Terhorst C and Eck M J Crystal structures of the XLP protein SAP reveal a class of SH2 domains with extended phosphotyrosine independent sequence recognition Molecular Cell 4 1999 555 561 138 Hwang P M Li C Morra M Lillywhite J Muhandiram D R Gertler F Terhorst C Kay
65. transformation by overexpression of Shc proteins Oncogene 1994 9 2827 2836 14 Rozakis Adcock M Fernley R Wade J Pawson T Bowtell D The SH2 and SH3 domains of mammalian Grb2 couple the EGF receptor of the Ras activator mSos1 Nature 1993 363 83 85 15 Rozakis Adcock M McGlade J Mbamalu G Pelicci G Daly R Li W Batzer A Thomas S Brugge J Pelicci M G Schlessinger J Pawson T Association of the She and Grb2 Sem5 SH2 containing proteins is implicated in activation of the Ras pathway by tyrosine kinases Nature 1992 360 689 692 16 Schlessinger J How receptor tyrosine kinases activate Ras Trends Biol Sci 1993 18 273 275 17 Khosravi Far R Der C J The Ras signal transduction pathway Cancer Metastasis Rev 1994 13 67 89 18 Davies H Bignell GR Cox C Stephens P Edkins S Clegg S Teague J Woffendin H Garnett M J Bottomley W Davis N Dicks E Ewing R Floyd Y Gray K Hall S Hawes R Hugues J Kosmidou V Menzies A Mould C Parker A Stevens C Watt S Hooper S Wilson R Jayatilake H Gusterson B A Cooper C Shipley J Hargrave D Pritchard Jones K Maitland N Chenevix Trench G Riggins G J Bigner DD Palmieri G Cossu A Flanagan A Nicholson A Ho J W Leung S Y Yuen S T Weber B L Siegler H F Darrow T L Paterson H Marais R Marshall C J Wooster R Strat
66. une activit significative dans un nombre vari de processus biologiques 36 39 En particulier lorsqu elle est complex e avec Sos par ses domaines SH3 40 Grb2 se lie tr s efficacement par son domaine SH2 avec des r cepteurs tyrosine kinase phosphoryl s tels que le facteur de croissance EGF R 41 She 42 43 ou FAK 44 Le complexe Grb2 Sos active alors la voie de signalisation Ras MAPK 18 dont nous avons d crit pr c demment les caract ristiques principales en se positionnant comme interm diaire entre l expression initiale de facteurs de croissance et activation de Ras Il est important de noter que inhibition de l activit de Grb2 via son domaine SH2 ou ses deux domaines SH3 constituerait un blocage du m canisme de la voie de signalisation Ras MAPK et plus g n ralement de diff rents processus de r gulation cellulaires 45 Int r t pharmaceutique Le domaine SH2 de Grb2 se situe dans un contexte multiple particuli rement int ressant du point de vue pharmacologique les m canismes de croissance cellulaire en g n ral et la voie de signalisation Ras MAPK en particulier les prot ines activit tyrosine kinase ainsi que les domaines SH2 constituent tous diff rents niveaux des syst mes activement tudi s dans l optique de la mise au point de m dicaments innovants Un grand nombre de voies de signalisation relatives la croissance cellulaire sont r gul es par des r actions de p
67. 0 success in reproducing conformations and binding free energies of protein ligand complexes 67 Hence the reference set is approximate and cannot be used to measure SHEF performance precisely However our aim here is simply to demonstrate SHEF usefulness as part of the VSM G screening funnel in a large scale VS context Consequently a reference set large enough statistically and chemically diverse seems appropriate despite GOLD related limitations In order to evaluate filtering the reference molecular database has to be divided in two subsets the first corresponding to the presumably most potent molecules referred to as the hit compounds subset that shall be conserved upon filtering and the second subset being considered as the inactive structures for the target The GOLD score values are being used to rank ligands against the three target conformations and the 10 best ranked ligands are selected from each of the three sets This cutoff value is set arbitrarily Ranks are being used to select ligands instead of the score values because molecular dynamics simulations performed in our laboratory on LXRB indicate that important induced fit effects 68 could occur upon ligand binding This suggests the GOLD scoring function which does not account for the receptor internal energy may only correlate with the global free energy of binding across a single receptor conformer 69 As shown in figure 5 the three ensembles of 838 selected structur
68. 15 2005 1385 1388 87 Songyang Z Shoelson S E McGlade J Olivier P Pawson T Bustelo X R Barbacid M Sabe H Hanafusa H Yi T Ren R Baltimore D Ratnofsky S Feldman R A and Cantley L C Specific motifs recognized by the SH2 domains of Csk 3BP2 fps fes GRB 2 HCP SHC Syk and Vav Molecular and Cellular Biology 14 issue 4 1994 2777 2785 88 Garc a Echeverr a C Furet P Gay B Fretz H Rahuel J Schoepfer J and Caravatti G Potent antagonists of the SH2 domain of Grb2 Optimization of the X 1 position of 3 amino Z Tyr PO3H2 X Asn NH3 Journal of Medicinal Chemistry 41 issue 11 1998 1741 1744 89 Fretz H Furet P Garcia Echeverria C Rahuel J and Schoepfer J Structure based design of compounds inhibiting Grb2 SH2 mediated protein protein interactions in signal transduction pathways Current Pharmaceutical Design 6 2000 1777 1796 90 Shakespeare W C SH2 domain inhibition a problem solved Current Opinion in Chemical Biology 5 2001 409 415 91 Lemmon M A Ladbury J E Mandiyan V Zhou M and Schlessinger J Independent binding of peptide ligands to the SH2 and SH3 domains of Grb2 Journal of Biological Chemistry 269 issue 50 1994 31653 31658 92 Cantley L C Auger K R Carpenter C Duckworth B Graziani A Kapeller R and Soltoff S Oncogenes and signal transduction Cell 65 issue 5 1991 914 93 Ward C W Gough K H Rashke M Wan S S Tribbick G and
69. 2 Shoichet B K Virtual screening of chemical libraries Nature 432 2004 862 865 3 Stahura F L and Bajorath J Virtual screening methods that complement HTS Combinatorial Chemistry and High Throughput Screening 7 issue 4 2004 259 269 4 Perola E Xu K Kollmeyer T M Kaufmann S H Prendergast F G and Pang Y P Successful virtual screening of a chemical database for farnesyltransferase inhibitor leads Journal of Medicinal Chemistry 43 issue 3 2000 401 408 5 Griineberg S Stubbs M T and Klebe G Successful virtual screening for novel inhibitors of human carbonic anhydrase Strategy and experimental confirmation Journal of Medicinal Chemistry 45 2002 3588 3602 6 Vangrevelinghe E Zimmermann K Schoepfer J Portmann R Fabbro D and Furet P Discovery of a potent and selective protein kinase CK2 inhibitor by high throughput docking Journal of Medicinal Chemistry 46 issue 13 2003 2656 2662 7 Kraemer O Hazemann I Podjarny A D and Klebe G Virtual screening for inhibitors of human aldose reductase Proteins Structure Function and Bioinformatics 55 2004 814 823 8 Doman T N McGovern S L Witherbee B J Kasten T P Kurumbail R Stallings W C Conolly D T and Shoichet B K Molecular docking and high throughput screening for novel inhibitors of protein tyrosine phosphatase 1B Journal of Medicinal Chemistry 45 2002 2213 2221 9 Bajorath J Integration of virtual and high throughput
70. 2 Thornton K H Mueller W T McConnell P Zhu G Saltiel A R Thanabal V Nuclear magnetic resonance solution structure of the growth factor receptor bound protein 2 Src homology 2 domain Biochemistry 1996 35 11852 11864 33 Kimber M S Nachman J Cunningham A M Gish G D Pawson T Pai E F Structural basis for specificity switching of the Src SH2 domain Mol Cell 2000 5 1043 1049 34 Rahuel J Garcia Echeverria C Furet P Strauss A Caravatti G Fretz H Schoepfer J Gay B Structural basis for the high affinity of amino aromatic SH2 phosphopeptide ligands J Mol Biol 1998 279 1013 1022 35 Berman H M Westbrook J Feng Z Gilliland G Bhat T N Weissig H Shindyalov I N Bourne P E The Protein Data Bank Nucl Acids Res 2000 28 235 242 36 Ettmayer P France D Gounarides J Jarosinski M Martin M S Rondeau J M Sabio M Topiol S Weidmann B Zurini M Bair K W Structural and conformational requirements for high affinity binding to the SH2 domain of Grb2 J Med Chem 1999 42 971 980 37 Gresh N Liu W Q Garbay N Unpublished work 38 Gilmer T Rodriguez M Jordan S Crosby R Alligood K Green M Kimery M Wagner C Kinder D Charifson P Hassell A M Willard D Luther M Rusnak D Sternbach D D Mehrotra M Peel M Shampine L Davis R Robbins J Patel LR Kassel D Burkhart W Mo
71. 27 131 La combinaison de ces deux approches 132 133 donna lieu une classe de ligands cycliques non peptidiques efficaces 131 134 La structure RX de l un d entre eux 74 PDB 2AOA 2AOB indique une possibilit d action simultan e sur deux domaines SH2 Il peut s agir d artefacts de cristallisation d autant plus que pour les besoins de la RX le domaine SH2 a t isol des deux domaines SH3 qui l encadrent dans Grb2 et qui auraient pu constituer une g ne st rique interdisant toute dim risation SH2 ligand SH2 de ce type Cette dim risation observ e par RX souligne tout de m me un caract re versatile du ligand capable de se lier sous certaines conditions Grb2 SH2 selon deux modes de liaison distincts 33 Recherche de ligands actifs in vivo contraintes Une mol cule active in vitro ob it des contraintes de compl mentarit g om trique et chimique par rapport a la cible biologique qu elle vise Une telle mol cule ne sera pas n cessairement active in vivo car pour cela elle doit ob ir trois contraintes suppl mentaires qui sont en rapport avec le milieu biologique 1 elle doit tre sp cifique sa cible 2 elle doit tre stable chimiquement dans l organisme et 3 elle doit pouvoir se diffuser dans l organisme jusqu la localisation du r cepteur cible Contrainte de sp cificit Au r cepteur cible peuvent correspondre dans l organisme plusieurs homologues dont
72. 315 2002 1167 1177 69 Rahuel J Garc a Echeverr a C Furet P Strauss A Caravatti G Fretz H Schoepfer J and Gay B Structural basis for the high affinity of amino aromatic SH2 phosphopeptide ligands Journal of Molecular Biology 279 1998 1013 1022 70 Ettmayer P France D Gounarides G Jarosinski M Martin M S Rondeau J M Sabio M Topiol S Weidmann B Zurini M and Bair K W Structural and conformational requirements for high affinity binding to the SH2 domain of Grb2 Journal of Medicinal Chemistry 42 1999 971 980 41 71 Furet P Garcia Echeverria C Gay B Schoepfer J Zeller M and Rahuel J Structure based design synthesis and X ray crystallography of a high affinity antagonist of the Grb2 SH2 domain containing an asparagine mimetic Journal of Medicinal Chemistry 42 1999 2358 2363 72 Ogura K Tsuchiya S Terasawa H Yuzawa S Hatanaka H Mandiyan V Schlessinger J and Inagaki F Solution structure of the SH2 domain of Grb2 complexed with the Shc derived phosphotyrosine containing peptide Journal of Molecular Biology 289 1999 439 445 73 Schiering N Casale E Caccia P Giordano P and Battistini C Dimer formation through domain swapping in the crystal structure of the Grb2 SH2 Ac pYVNV complex Biochemistry 39 2000 13376 13382 74 Phan J Shi Z D Burke Jr T R and Waugh DS Crystal structures of a high affinity macrocyclic peptide mimetic in complex with t
73. 4 or via She 15 then activating the Ras MAPK signalling pathway 16 17 This pathway is essential for cell growth and differentiation and its mutations are amongst the most frequently observed ones in human cancers 18 Anarchic cell proliferation characterizing acute myeloid leukaemia 19 breast ovarian and prostate cancers 20 was shown to be related to over expression of HER2 ErbB2 a truncated analog of the EGF receptor and Grb2 in the Ras pathway Thus the search for Grb2 SH2 inhibitors as for other compounds blocking the HER2 signalling is a particularly interesting approach for anti cancer drug design 5 7 21 22 The binding sites of SH2 domains are known to recognize very short sequences of amino acids the most active part being composed of pTyr and three other residues C terminal to it Mutagenesis studies of small sequences of amino acids demonstrated each family of SH2 domains to recognize a well defined sequence namely pY VNV the two valine residues not being crucial in the case of Grb2 23 24 The binding mode of such a pattern is depicted by X ray crystallography measurements 25 26 and thus constituted the basis of several drug design studies for more efficient and specific pseudo peptidic analogues 27 28 One compound mAz pY a Me pY N NHb hereby noted ligand 1 see figure 1 a is a result of this research and one of the most potent Grb2 SH2 inhibitors known to date 29 Ligand 1 targets three distinct bin
74. 6 1232 1240 78 Ooms F Molecular modeling and computer aided drug design Examples of their applications in medicinal chemistry Current Medicinal Chemistry 7 2000 141 158 79 Lipinski C A Lombardo F Dominy B W and Feeney P J Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings Advanced Drug Discovery Reviews 23 issue 1 1997 3 25 80 Bleicher K H B hm H J M ller K and Alanine A I Hit and lead generation beyond high throughput screening Nature Reviews Drug Discovery 2 issue 5 2003 369 378 81 DiMasi J A Hansen R W and Grabowski H G The price of innovation new estimates of drug development costs Journal of Health Economics 22 2003 151 185 82 Rowinsky E K and Donehower R C Drug therapy paclitaxel Taxol New England Journal of Medicine 332 1995 1004 1114 83 Nygren P and Larsson R Overview of the clinical efficacy of investigational anticancer drugs Journal of Internal Medicine 253 2003 46 75 84 Jorgensen W L The many roles of computation in drug discovery Science 303 issue 5665 2004 1813 1818 85 Drews J Drug discovery A historical perspective Science 287 2000 1960 1964 86 Garrett M D and Workman P Discovering novel chemotherapeutic drugs for the third millenium European Journal of Cancer 35 issue 14 1999 2010 2030 21 22 Cible Description La voie de signalisation Ras
75. 6 c corresponds to a perfect correlation A given filtering process will obviously have results between these two Figure 6 b is another ideal case for filtering but only for a precise filtering amount which may or may be not satisfactory Figure 6 Explanation of the density plots representation for ranks correlation Three particular cases are exemplified a random selection b ideal filtering c perfect correlation p 0 1 0 p 0 42 0 28 p l t 1 high c o g 2 2 5 S Fi mn F g e g e a reference data a reference data b reference data c Block population worst ranks ranking 2 best ranks es ranking 1 The Spearman 70 p and Kendall 71 t coefficients are employed as measures of correlation p 1 2 Yar n n D n 1l 4 n a aan a D 6 r gt r i l j i l ri is the SHEF ranking of the i ranked GOLD structure Ar is the difference between these two ranks Ar ri 1 6 is the boolean function 6 true 1 while false 0 The rankings in both cases are in ascending order from the best predicted binding molecule to the worst We also have O lt p lt 1 and 1 lt q lt 1 0 indicating an absence of correlation random selection and 1 perfect correlation same rankings Other metrics are used in order to evaluate filtering performance Given a definition of what is a hit structure and what is not for a specific target we can describe the quality
76. Algorithm Parameters Number of Population Size 100 Selection Pressure 1 1 Operations 100000 n cessaire m me si certains syst mes en particulier HIV prot ase sont connus pour pr senter au moins une mol cule d eau structurellement importante Number of 5 Nace Niche Se 2 Migrate 10 Mutate 85 Crossover 85 Click mouse button 2 over window for help Afin de tester les effets de l inclusion du solvant dans le mod le de docking nous avons retenu les mol cules d eau pr sentes dans la structure RX de r f rence du complexe Grb2 SH2 ligand 1 en modifiant le fichier PDB afin de faire en sorte que GOLD les consid re comme des r sidus li s de fa on covalente Grb2 ci contre la structure correspondante ligand 1 est en jaune et les mol cules d eau en bleu tandis que les zones en rouge indiquent des volumes vacants l interface Test et validation des param tres Afin de tester la fiabilit du protocole de docking il convient de l prouver en v rifiant que les structures de complexes connues exp rimentalement sont reproduites Il s av ra dans le cas de GOLD que seul le mod le du r cepteur de Grb2 SH2 d fini explicitement incluant les mol cules d eau cristallographiques associ aux param tres standard et un nombre d chantillonnage de 50 permettait de reproduire les conformations des complexes exp rimentaux 1BMB 1JYR 1JYW 1ZFP noter que dans sa versi
77. Figure 6 Snapshots of complexes of ligand 1 and ligand 2 during the MD simulations The colors on the Grb2 surface on the main snapshots are related to the distance of the ligand to the surface The colors on the smaller snapshots on the top red corners of the pictures are related to the main Grb2 SH2 interacting residues Cavity 1 Re7 Arg aA Rss Arg BB red Ssg Ser BB Soo Ser BC Sos Ser BC orange Cavity 2 K 09 Lys BD dark blue Li Leu BE Wi21 Trp EF light blue Cavity 3 Ry4 Arg BG purple S141 Ser BG Nia3 Asn BG green ligand 1 AE ligand 1 Mo TE ligand 1 ER r t 1ns 24 y La t 2ns N K 8 AE kcal mol Evolution of the ligands binding energies The decomposition of the interaction energy between ligand 1 residues and Grb2 SH2 indicates that pTyro located in cavity 1 contributes to approximately 55 of the total protein ligand binding energy pTyr targeting cavity 3 to 40 and the Asn residue which is required in order to have the specific Grb2 SH2 B turn conformation to 5 We have plotted the evolution of the binding energies between ligand 1 and ligand 2 on one hand and the Grb2 SH2 domain and the water molecules on the other hand in figures 7 a and 7 b respectively These graphs clearly indicate that ligand 1 binding energy only undergoes acceptable fluctuations as a loss of less than 10 of the interaction energy is observed after 2 ns In an opposed way a significa
78. H H Q nc A HN Ps PN Ne T N k p j pa F3 di o C H2C H2C H2C N H H N H H l a se i se j H3N coo H3N coo H3N CO0 H3N coo H3N CO0 Aspartate Glutamate Lysine Arginine Histidine Asp D Glu E Lys K Arg R His H H2N NH c a fa HA gt HO CH ps Hac pia p RE 4 se H3N CO0 H3N coo H3N coo H3N coo H3N coo Glycine Serine Threonine Asparagine Glutamine Gly G Ser S Thr T Asn N Gin Q H H H H H S EX i BA Hs 7 ue mr ee 2 H3N coo H3N coo H3N coo H3N Co0 H3N CO0 Cysteine Alanine Valine Phenylalanine Tyrosine Cys Ala A Val v Phe F Tyr Y H f H C Hs CH Ss HN we vs HC Hc a x H j ae b H H au p a oe a H2C Y s y CO0 H3N coo H3N coo H3N Co0 H3N coo Proline Leucine Isoleucine Methionine Tryptophan Pro P Leu L Ile 1 Met M y Liste des 20 acides amin s class s naturels class s par propri t s respectivement acides D E basiques K R H structure sp cifique G C P hydrophiles S T N Q et hydrophobes A V F Y L L M W Il a t d montr que la s quence d une prot ine d termine la structure g om trique globulaire adopt e par un processus de repliement sur elle m me qui survient apr s sa synth se Les principes de base de ce m canisme sont connus 62 ainsi que des r sultats statistiques Sa pr diction d sign e comme le probl me de repliement des prot
79. I Action Concert e Incitative dans la cat gorie Mol cules et cibles th rapeutiques nomm e Nouvelles approches th rapeutiques du cancer conception et validation de mol cules inhibant des interactions prot iques et ciblant la voie Ras induite par les prot ines activit tyrosine kinase Fonctionnement du cycle cellulaire et causes du cancer Les cellules d un organisme ont une dur e de vie limit e Ainsi elles ont la capacit de se multiplier afin que l organisme puisse se maintenir en vie et il existe de nombreux m canismes permettant de r guler la population des cellules 6 7 en favorisant leur prolif ration les g nes codant ainsi pour la division cellulaire sont appel s proto oncog nes ou au contraire en la freinant les g nes correspondants sont appel s suppresseurs de tumeurs Une trop faible population de cellules peut provoquer l activation de m canismes favorisant l expression de proto oncog nes et l inverse la d tection de mutations pourra favoriser les suppresseurs de tumeurs ce qui permettra de proc der la r paration ou ventuellement P limination des cellules endommag es avant que celles ci ne puissent mettre en danger l organisme en se multipliant En temps normal l ensemble de ces m canismes permet de maintenir un quilibre entre le nombre de cellules r sultantes de la division cellulaire et le nombre de cellules mortes tout en r gulant ventuelles alt ra
80. I 000 0007 MT IC 008 MALI VIRE HNEME I HE NM ER SE TT Se EMRE HAE A M AL 4 ORBE AMMEN FE MD OBEA CHE BUNA AR RE V RRE MERE ae Lh Sela LE TEE SHC 941 E L HET 900 KOK HEAMFRIC OOT MARE FH CHU ERM tf pi LESART eC EY EE ELEC TOK YAR TG EY OE ph feed HERE to BMW WX AS GA ERA LE WCA hs DI Bs YOU I eh NG ICT GE COTES Ze Fe y RAGS a BE A fe DEN Reh FEA N S RA LAK LENS ERR He HEN HET EE BAF ake RCE GUE Red A roses rad WOT Sh EPE sex Woes WUE CU BY ES al EK hw LE GUN SE MOY Y WIS EAB CGY SEM Y ach Dk Pr KENK WH Ee HS MAT Lesa FAKAM uff TE SMALL UNE ae a al ne YO BMA fe AH ARE VE EPA Er AVA HEC SEAN Cr ET AAT HT G ZAS AUG SRA BE Ly EB SY BPE Bl Bie EI GH LIB EE 2 NFA PE A A IC QUE HOA MTR GH TS ANA she A EE BOF R EC ESA GARE BTE TH EARR A Yh LA A H E ke A MI EHE Pk GS ACTE HG EE ch SE ARE 7 E Bd fk SEA 2 E EHER HVALP RI A ee A GS SE HHK re SE M el ee GRACE IE id St YT 7 YY G1 Ht 091 L00L NSST 4 Pape pe a D FESFIHNYHONS IHETH FEEFTI SHAMHS HE TH rth A ol AYLSIINNAHD GAliddV GNV SYSLNdNOD it Bit dS it Bit LE cs l id Bit BE 44 5 AC St afar ik Bit 224 AE h 93HAAT NSGO0 d1 9 11I NO 091 100 NSSI
81. I T E DeBolt S Ferguson D Seibel G and Kollman P AMBER a package of computer programs for applying molecular mechanics normal mode analysis molecular dynamics and free energy calculations to simulate the structural and energetic properties of molecules Computer Physics Communications 91 1995 1 41 82 Nelson M Humphrey W Gursoy A Dalke A Kal L Skeel R and Schulten K NAMD A parallel object oriented molecular dynamics program International Journal of Supercomputer Applications and High Performance Computing 10 1996 251 268 83 MacKerell Jr A D Brooks B Brooks III C L Nilsson L Roux B Won Y and Karplus M CHARMM The energy function and its parametrization with an overview of the program in The Encyclopedia of computational chemistry ed P V R Schleyer et al 1998 John Wiley amp sons Chichester p 271 277 84 Kal L Skeel R Bhandarkar M Brunner R Gursoy A Krawetz N Phillips J Shinozaki A Varadarajan K and Schulten K NAMD2 Greater scalability for parallel molecular dynamics Journal of Computational Physics 151 1999 283 312 85 Case D A Cheatham III T E Darden T Gohlke H Luo R Merz Jr K M Onufriev A Simmerling C Wang B and Woods R J The Amber biomolecular simulation programs Journal of Computational Chemistry 26 issue 16 2005 1668 1688 86 Johnson M A Galvan LF and Villa Freixa J Framework based design of a new all purpose molecular
82. J Schoepfer J Fretz H Structure based design and synthesis of high affinity tripeptide ligands of the Grb2 SH2 domain J Med Chem 1998 41 3442 3449 46 Ogura K Tsuchiya S Terasawa H Yuzawa S Hatanaka H Mandiyan V Schlessinger J Inagaki F Solution structure of the Sh2 domain of Grb2 complexed with the Shc derived phosphotyrosine containing peptides J Mol Biol 1999 289 439 445 47 Kal L Skeel R Bhandarkar M Brunner R Gursoy A Krawetz N Phillips J Shinozaki A Varadarajan K Schulten K NAMD2 Greater scalability for parallel molecular dynamics J Comp Phys 1999 151 283 312 48 Ray N Cavin X Paul J C Maigret B Intersurf dynamic interface between proteins J Mol Graph Modell 2005 23 347 354 49 De Lucca G V Erickson Viitanen S Lam P Y S Cyclic HIV proteases inhibitors capable of displacing the active site structural water molecule Drug Discovery Today 1997 2 6 18 50 Clarke C Woods R J Gluska J Cooper A Nutley M A Boons G J Involvement of water in carbohydrate protein binding J Am Chem Soc 2001 123 12238 12247 51 Dunitz J D Win some lose some enthalpy entropy compensation in weak intermolecular interactions Chem Biol 1995 2 709 712 52 Verdonk M L Chessari G Cole J C Hartshorn M J Murray C W Nissink J W M Taylor RD Taylor R Modelling water molecules in protein l
83. J From atoms and bonds to three dimensional atomic coordinates Automatic model builders Chemical Reviews 93 1993 2567 2581 31 http www rcsb org pdb static do p file_formats pdb i ndex html 32 Davis I W Leaver Fay A Chen V B Block J N Kapral G J Wang X Murray L W Arendall W B II Snoeyink J Richardson J S and Richardson D C MolProbity all atom contacts and structure validation for proteins and nucleic acids Nucleic Acids Research in the press 2007 33 Gordon J C Myers J B Folta T Shoja V Heath L S and Onufriev A H a server for estimating pKas and adding missing hydrogens to macromolecules Nucleic Acids Research 33 issue Web server issue 2005 W368 W371 34 Neshich G Mancini A L Yamagishi M E Kuser P R Fileto R Pinto I P Palandrani J F Krauchenco J N Baudet C Montagner A J and Higa R H STING Report convenient web based application for graphic and tabular presentations of protein sequence structure and function descriptors from the STING database Nucleic Acids Research 33 issue Database issue 2005 D269 D274 35 Cai W Zhang M and Maigret B New approach for representation of molecular surface Journal of Computational Chemistry 19 issue 16 1998 1805 1815 36 Cai W Shao X and Maigret B Protein ligand recognition using spherical harmonic molecular surfaces towards a fast and efficient filter for large virtual throughput screening Journal of
84. Koshland Jr D The key lock theory and the induced fit theory Angewandte Chemie International Edition in English 33 issue 23 24 1994 2375 2378 69 Redocking experiments of LXRB reference ligands present in the X ray structures back up this hypothesis Using GOLD the 1PQ6 ligand redocked in the 1PQ6 binding pocket conformation yields a significantly higher score than the 1PQ9 ligand redocked in the 1PQ9 conformation However according to experimental data the 1PQ9 ligand is indeed clearly more potent on LXRB than the 1PQ6 one further indicating that the protein ligand interaction could not be the dominant term in the free energy of binding 70 Spearman C The proof and measurement of associtaion between two things American Journal of Psychology 15 issue 1 1904 72 101 71 Kendall M A new measure of rank correlation Biometrika 30 issue 1 2 1938 81 89 72 Mavridis L Hudson B D and Ritchie D W Toward high throughput 3D virtual screening using spherical harmonic molecular surface representations Journal of Chemical Information and Modeling 47 2007 1787 1796 73 Massova I and Kollman P A Combined molecular mechanical and continuum solvent approach MM PBSA GBSA to predict ligand binding Perspectives in Drug Discovery and Design 18 issue 1 2000 113 135 74 Gilson M K and Zhou H X Calculation of protein ligand binding affinities Annual Review of Biophysics and Biomolecular Structure 36 2007 21 42
85. L E Pawson T Forman Kay J D and Li S C A three ponged binding mechanism for the SAP SH2D1A SH2 domain structural basis and relevance to the XLP syndrome The EMBO Journal 21 issue 3 2002 314 323 139 McGuigan C Devine K G O Connor T J Galpin S A Jeffries D J and Kinchington D Synthesis and evaluation of some novel phosphoramidate derivatives of 3 azido 3 deoxythymidine AZT as anti HIV compounds Antiviral Chemistry amp Chemotherapy 1 1990 107 113 140 Burke Jr T R Yao Z J Liu D G Voigt J H and Gao Y Phosphotyrosyl mimetics in the design of peptide based signal transduction inhibitors Biopolymers Peptide Science 60 2001 32 44 141 Burke Jr T R Luo J H Yao Z J Gao Y Zhao H Milne G W A Guo R Voigt J H King C R and Yang D Monocarboxylic based phosphotyrosyl mimetics in the design of Grb2 SH2 domain inhibitors Bioorganic amp Medicinal Chemistry Letters 9 1999 347 352 142 Long Y Q Voigt J H Lung F D T King CR and Roller P P Significant compensatory role of position Y 2 conferring high affinity to non phosphorylated inhibitors of Grb2 SH2 domain Bioorganic amp Medicinal Chemistry Letters 9 1999 2267 2272 143 Lung F D T Long Y Q Roller P P King C R Varady J Wu X W and Wang S Functional preference of the constituent amino acid residues in a phage library based nonphosphorylated inhibitor of the Grb2 SH2 domain Journal of Peptide R
86. LE EMAN CURE A ONU A EE RENE BRUN MON A T 1 UNI EE 500 t ps 2000 0 a E 20 pc E s 150 kcal mol b t ps With most of the residues on the ligand 1 complex the interactions with water slightly decrease during the simulation In the case of the ligand 2 complex at 1800 ps as expected the Kio Lys BD interaction with water increases as its interaction with ligand 2 is lowered Additionally stabilizations of S 4 Ser BG and N14 Asn BG are observed This corresponds to water molecules filling the gap in cavity 3 that results from ligand 2 escaping towards the bulk along with Ri4 Arg BG The interaction of the latter residue slightly decreases while its salt bridge with the phosphone is reinforced This was not obvious as R 4 Arg BG and phosphone move altogether away from cavity 3 and are therefore more exposed to the solvent In uncomplexed Grb2 SH2 it is noted that Ry4 Arg BG is particularly hydrophilic as compared to Re7 Arg aA and Rgs Arg BB which are less exposed to the solvent in cavity 1 Interestingly it is noted that the interactions of the Egy Glu BC residue with water differ with both complexes and with the uncomplexed Grb2 SH2 At the beginning of the simulation the Egy Glu BC makes much larger interactions with water in the case of the ligand 2 complex with a difference of approximately 80 kcal mol with its interaction with water in the case of the ligand 1 complex However such an interac
87. Nancy Universite Universite H enri Poincar AVERTISSEMENT Ce document est le fruit d un long travail approuv par le jury de soutenance et mis disposition de l ensemble de la communaut universitaire largie Il est soumis la propri t intellectuelle de l auteur Ceci implique une obligation de citation et de r f rencement lors de l utilisation de ce document D autre part toute contrefa on plagiat reproduction illicite encourt une poursuite p nale gt Contact SCD Nancy 1 theses sciences scd uhp nancy fr Code de la Propri t Intellectuelle articles L 122 4 Code de la Propri t Intellectuelle articles L 335 2 L 335 10 http www cfcopies com V2 leg leg_droi php http www culture gouv fr culture infos pratiques droits protection htm Universit Henri Poincar Nancy I U F R Sciences et techniques de la mati re et des proc d s Ecole doctorale lorraine de chimie et physique mol culaires Th se pr sent e pour l obtention du titre de Docteur de l Universit Henri Poincar en Chimie informatique et th orique par Vincent LEROUX Mod lisation d inhibiteurs du domaine SH2 de la prot ine Grb2 par dynamique mol culaire docking et criblage virtuel sous la direction de Bernard MAIGRET quipe de dynamique des assemblages membranaires UMR UHP CNRS 7565 Soutenue publiquement le 15 d cembre 2006 Rapporteurs Dr Nohad GRESH Dr Didier ROGNAN Examinateurs P
88. ON 7e Je ang 6 MO pu OH 009 fe fe N fe O 9 a N N H9991y s O0 N ZHN i a N_ O o o 9 NT So N N en o rol 7 x TNT oF NH eta ZtN awy ZUN eng 3119 z NOX 9dd ION 72 Je 481104 OH A ION 7238 ng 23 000 a 8 9 fe fe o 9 PS N NE O ley b o8oy N ans o ss o O O ae ps ie amp i M ET N i J L 6 j e RER e Z si A N 9 G gy W 6 58d99 baton 05882499 H999y SIBAON SINBAON Ho o f0 Z Ho o N N N N N w N N ZHN O O ZHN 0 z Z dnu N ie S dny ae N o d nga oA 2 Od dwg p2399301d ye O dwd fe DAFL 80d SHNO NHASNI 72 Je Aeque S Fzywu Had 2 OdO ZHN O ZHN ZHN t O 2 OdO oad LOL 80d SIHEAON v H999y O O M oN oO N T PEAT eoayoy u s 2 OdO ZHN oad a9ua13491 38 R f rences bibliographiques 1 Vidal M Gigoux V and Garbay C SH2 and SH3 domains as targets for anti proliferative agents Critical Reviews in Oncology Hematology 40 2001 175 186 2 Robbins D J Cheng M Zhen E Vanderbilt C A Feig L A and Cobb M H Evidence for a Ras dependent extracellular signal regulated protein kinase ERK cascade Proceedings of the National Academy of Sciences USA 89 1992 6924 6928 3 Schlessinger J How receptor tyrosine kinases activate Ras Trends in Biochemical Sciences 18 1993 273 275 4 Khosravi Far R and Der C J The Ras signal transduction pathway
89. Pharmacol 2000 60 1165 1169 6 Liu W Q Vidal M Math C P rigaud C Garbay C Inhibition of the Ras dependant mitogenic pathway by phosphopeptide prodrugs with antiproliferative properties Bioorg Med Chem Lett 2000 70 669 672 7 Vidal M Gigoux V Garbay C SH2 and SH3 domains as targets for anti proliferative agents Crit Rev Oncol Hematol 2001 40 175 186 8 Lowenstein E J Daly R J Batzer A G Li W Margolis B Lammers R Ullrich A Skolnik E Y Bar Sagi D Schlessinger J The SH2 and SH3 domain containing protein Grb2 links receptor tyrosine kinases to Ras signalling Cell 1992 70 432 442 9 Maignan S Guilloteau J P Fromage N Arnoux B Becquart J Ducruix A Crystal structure of the mammalian Grb2 adaptor Science 1995 268 291 293 10 Chardin P Cussac D Maignan S Ducruix A The Grb2 adaptor FEBS Lett 1995 369 47 51 11 Buday L Membrane targeting of signalling molecules by SH2 SH3 domain containing adaptor proteins Biochem Biophys Acta 1999 1422 187 204 12 Xie Y Li K Hung M C Tyrosine phosphorylation of Shc proteins and formation of Shc Grb2 complex correlate to the transformation of NIH3T3 cells mediated by the point mutation activated neu Oncogene 1995 10 2409 2413 13 Salcini A E McGlade J Pelicci G Nicoletti I Pawson T Pelicci P G Formation of the Shc Grb2 complexes is necessary to induce neoplastic
90. QSAR It should also be remarked that such a distinction between SBDD and LBDD can be misleading because it refers more to the nature of the main input data is a structure of the biological target of interest already available rather than to what kind of data will be manipulated While we did use the de facto definitions here this would benefit from some clarification Additionally we will see that such a frontier between SBDD and LBDD should be outshined In this short review we will now focus on SBDD and molecular docking The basic aim of docking as a SBDD technique is to solve the question of how a specific molecule would interact with a given biological target whose molecular structure is known One of its current goals is to simulate HTS experiments reliably and quickly enough through virtual experiments VHTS making the virtual screening approach VS 23 37 39 a potent cost saving alternative or complement to screening campaigns 40 Another highlight of docking is that valuable structural information is obtained that can be used readily as a starting point for optimization using a great variety of molecular modeling techniques Issues and concerns about docking It is often said that SBDD methods have proven their value having a more than 10 years track record of successes 41 51 But more generally most methods of modern drug discovery are being questioned even the efficiency of HTS techniques one of the most important too
91. R Development and validation of a genetic algorithm for flexible docking Journal of Molecular Biology 267 1997 727 748 120 GOLD Genetic Optimization for Ligand Docking http www ccdc cam ac uk products life_sciences gold 121 Verdonk M L Chessari G Cole J C Hartshorn M J Murray C W Nissink J W M Taylor R D and Taylor R Modeling water molecules in protein ligand docking using GOLD Journal of Medicinal Chemistry 48 2005 6504 6515 122 Geist A Beguelin A Dongarra J Jiang W Manchek R and Sunderam V S PVM Parallel Virtual Machine A users guide and tutorial for network parallel computing 1994 MIT Press 299 pages 123 Sunderam V S PVM A framework for parallel distributed computing Concurrency Practice and Experience 2 issue 4 1990 315 340 124 Goodsell D S Morris G M and Olson A J Automated docking of flexible ligands applications of AutoDock Journal of Molecular Recognition 9 issue 1 1996 1 5 125 AutoDock http autodock scripps edu 126 Osterberg F Morris G M Sanner M F Olson A J and Goodsell D S Automated docking to multiple target structures Incorporation of protein mobility and structural water heterogeneity in AutoDock Proteins Structure Function and Genetics 46 issue 1 2001 34 40 127 Morris G M Goodsell D S Halliday R S Huey R Hart W E Belew R K and Olson A J Automated docking using a Lamarckian genetic algorithm and an empi
92. Rahuel J Gay B and Furet P Structure based design of a non peptidic antagonist of the SH2 domain of Grb2 Bioorganic amp Medicinal Chemistry Letters 9 1999 1973 1978 117 Tong L Warren T C Lukas S Schembri King J Batageri R Proudfoot J R and Jakes S Carboxymethyl phenylalanine as a replacement for phosphotyrosine in SH2 domain binding Journal of Biological Chemistry 273 issue 32 1998 20238 20242 118 Vu C B Recent advances in the design and synthesis of SH2 inhibitors of Src Grb2 and ZAP 70 Current Medicinal Chemistry 7 2000 1081 1100 119 Sawyer T K Bohacek R S Dalgarno D C Eyermann C J Kawahata N Metcalf II C A Shakespeare W C Sundaramoorthi R Wang Y and Yang M G Src homology 2 inhibitors Peptidomimetic and nonpeptide Mini Reviews in Medicinal Chemistry 2 issue 5 2002 475 488 120 Cussac D Vidal M Leprince C Liu W Q Cornille F Tiraboschi G Roques B P and Garbay C A Sos derived peptidimer blocks the Ras signaling pathway by binding both Grb2 SH3 domains and displays antiproliferative activity FASEB Journal 13 1999 31 38 44 121 Liu W Q Vidal M Olszowy C Million E Lenoir C Dh tel H and Garbay C Structure activity relationships of small phosphopeptides inhibitors of Grb2 SH2 domain and their prodrugs Journal of Medicinal Chemistry 47 issue 5 2004 1223 1233 122 Domchek S M Auger K R Chartterjee S Burke Jr T R and Shoelson S E I
93. Wang J X Systematic mapping of potential binding sites for Shc and Grb2 SH2 domains on Insuline Receptor Substrate 1 and the receptors for Insulin Epidermal Growth Factor Platelet derived Growth Factor and Fibroblast Growth Factor Journal of Biological Chemistry 271 issue 10 1996 5603 5609 94 M ller K Gombert F O Manning U Grossm ller F Graff P Zaegel H Zuber J F Freuler F Tschopp C and Baumann G Rapid identification of phosphopeptide ligands for SH2 domains Journal of Biological Chemistry 271 issue 28 1996 16500 16505 95 Oligino L Lung F D T Sastry L Bigelow J Cao T Curran M Burke Jr T R Wang S Krag D Roller P P and King C R Nonphosphorylated peptide ligands for the Grb2 Src homology 2 domain Journal of Biological Chemistry 272 issue 46 1997 29046 29052 96 Hart C P Martin J E Reed M A Keval A A Pustelnik M J Northrop J P Patel D V and Grove J R Potent inhibitory ligands of the Grb2 SH2 domain from recombinant peptide libraries Cellular Signalling 11 issue 6 1999 453 464 97 McNemar C Snow M E Windsor W T Prongay A Mui P Zhang R Durkin C Le H V and Weber P C Thermodynamic and structural analysis of phosphotyrosine polypeptide binding to Grb2 SH2 Biochemistry 36 1997 10006 10014 98 Suenaga A Hatakeyama M Ichikawa M Yu X Futatsugi N Narumi T Fukui K Terada T Taiji M Shirouzu M Yokoyama S and Konagaya A Molecu
94. aine SH2 L analyse des structures RX de complexes peptidiques de Grb2 SH2 68 70 compar es celles de complexes d autres domaines SH2 23 77 101 105 permet de v rifier que cette sp cificit se retrouve dans le mode de liaison correspondant L analyse des structures RX montre en effet que la surface accessible du r cepteur de Grb2 SH2 diff re totalement de celle des autres r cepteurs SH2 par le r sidu W12 Trp EF dont l importance avait d j t mise en vidence par mutagen se la d termination des structures g om triques 25 W 12 Trp EF de par son encombrement impose une contrainte g om trique sp cifique Grb2 en plus de la n cessit de la pr sence d un r sidu pTyr commune tous les domaines SH2 Seules les cha nes peptidiques pr sentent un tour B en position pY 1 d o une conformation coud e position pY 2 permet de garantir ne sont pas d favoris s 106 108 que seul un r sidu asparagine en Ainsi une s quence ne pr sentant pas le motif pYXNX sera sp cifiquement inactive avec Grb2 SH2 Ainsi la s quence pYINQ a la m me activit avec Grb2 SH2 et Lek SH2 tandis que la s quence pYEEI n est active qu avec Lck SH2 et de fa on plus pouss e que pour pYINQ 107 L interaction d un ligand avec un domaine SH2 peut donc se faire suivant au moins deux modes de liaison l un lin aire et l autre courb le premier tant priori plus favorable sauf dans
95. alorim triques ont d montr que sa substitution par une alanine divise l affinit de la s quence pY VNV 2000 fois 97 Deux tudes 93 98 sugg rent toutefois que des s quences ne disposant pas d un r sidu asparagine en position pY 2 en particulier pYQQD sont favorables un degr moindre ce qui semble indiquer que la pr sence de l asparagine constitue plus une contrainte conformationnelle tr s favorable qu une obligation structurale Les r sultats divergent quant la pr f rence sur les sites pY 1 et pY 3 ceux ci ne doivent donc pas tre cruciaux en regard de interaction sur le r cepteur de Grb2 SH2 bien qu il ait t propos de placer un r sidu charg positivement en pY 1 afin d augmenter la sp cificit du ligand sans pour autant handicaper trop fortement son affinit 99 A la lumi re de ces informations il est l gitime de consid rer la s quence naturelle pY VNV comme une r f rence en vue de mettre au point des inhibiteurs peptidiques ou non plus performants ciblant le domaine SH2 de Grb2 Mode de liaison Comme avec l ensemble des domaines SH2 le r cepteur du domaine SH2 de Grb2 poss de deux cavit s bien distinctes la premi re reconna t sp cifiquement un r sidu pTyr tandis que la seconde plus sp cifique est hydrophobe La zone de reconnaissance pTyr est caract ris e principalement par la pr sence de deux arginines Re Arg aA et Rgs Arg BB elle poss de donc u
96. and binding pocket it is able to accommodate ligands with noticeably different shapes and sizes 54 In this work we study more specifically the LXRB isoform for which we took as a starting point different X ray structures available from the Protein DataBank PDB 55 1P8D 56 1PQ6 54 and 1PQ9 54 For each of these structures the most complete chain was retained chain A for 1P8D and chain B for 1PQ6 and 1PQ9 In all cases the binding area was complete and the Ca trace superimposed well allowing missing fragments to be added using homology modeling Protonation was performed at pH 7 with VSM G The imidazole tautomer of the active site histidine residue is the N H one 57 Figure 4 shows that the three binding site conformations represented by their MSSH generated surfaces imported into VMD 37 are clearly distinct geometrically The 1PQ9 cavity is significantly smaller 810 A than 1PQ6 996 A and 1P8D 1014 1PQ6 has a less spherical more specific shape Therefore it could be expected that 1 1P8D is the least selective upon ligand binding ii 1PQ6 shape specificity could be overcome by ligand flexibility and iii the 1PQ9 conformation should filter out more structures based on their size Figure 4 Shapes of the 1P8D 1PQ6 and 1PQ9 active sites from left to right as approximated by spherical harmonics expansion surfaces using MSSH The X ray ligands filling the active sites are shown The protein ligand bindin
97. and computational approaches to estimate solubility and permeability in drug discovery and development settings Advanced Drug Discovery Reviews 23 issue 1 1997 3 25 73 Irwin J J and Shoichet B K ZINC a free database of comercially available compounds for virtual screening Journal of Chemical Information and Modeling 45 issue 1 2005 177 182 74 A free database for virtual screening ZINC Zinc Is Not Commercial http blaster docking org zinc 75 Bleicher K H B hm H J M ller K and Alanine A I Hit and lead generation beyond high throughput screening Nature Reviews Drug Discovery 2 issue 5 2003 369 378 76 Bajorath J Integration of virtual and high throughput screening Nature Reviews Drug Discovery 1 2002 882 894 77 Alder B J and Wainwright T E Studies in molecular dynamics I General method Journal of Chemical Physics 31 issue 2 1959 459 466 78 van Gunsteren W F and Berendsen J C Computer simulation of molecular dynamics Methodology applications and perspectives in chemistry Angewandte Chemie International edition in english 29 1990 992 1023 79 Karplus M and McCammon J A Molecular dynamics simulations of biomolecules Nature Structural Biology 9 issue 9 2002 646 652 80 Chipot C and Pearlman D A Free energy calculations the long and winding gilded road Molecular Simulation 28 2002 1 12 81 Pearlman D A Case D A Caldwell J W Ross W S Cheatham II
98. ar la d termination de la structure RX du complexe correspondant code PDB 1JYQ 68 B o _ Structure exp rimentale du ligand mAZ pTyr a Me pTyr Asn NH li au r cepteur de Grb2 SH2 Les principaux r sidus actifs sont mentionn s on distingue clairement trois cavit s not es 1 2 et 3 Substituants pTyr et cyclisation Affymax NCI Du moment qu un r sidu Asn est pr sent dans la s quence on peut supposer que la cyclisation du ligand n est pas incompatible avec le maintien de la conformation turn requise par Grb2 SH2 Une quipe du centre de recherche Affymax montra ainsi que la cyclisation d une s quence peptidique C pYXN C par un pont disulfure peut augmenter l affinit pour Grb2 SH2 d un ordre de grandeur dans certains cas 96 Cela souligne le grand int r t que constitue la strat gie de cyclisation afin de mettre au point des ligands plus efficaces de Grb2 SH2 En se basant sur l inhibiteur CGP78850 de Novartis Burke Jr et al NCI qui auparavant avaient d j tudi les effets d analogues de pTyr sur des inhibiteurs de SH2 122 123 orient rent les optimisations ult rieures de l inhibition de Grb2 SH2 dans deux directions distinctes d une part l emploi de divers substituants au groupe principal d riv de pTyr 124 126 et d autre part la cyclisation du squelette peptidique afin de stabiliser la conformation B turn n cessaire la reconnaissance du ligand 82 1
99. asic such drug like filtering are currently available 73 74 The next steps of drug discovery aim to further reduce the numbers of molecules considered from 108 drug like candidates to 103 hits then a limited number of leads which after optimization and careful inspection could be proposed for clinical trials We define hits as molecules confirmed active regarding the biological target in vitro while leads are hits which are also active in vivo An excellent review of the associated protocols including definitions strategies and issues is available 75 At both the drug candidates to hit VS and hit to lead optimization steps experiments can be conducted In the former case HTS can greatly benefit from its association with VHTS 40 76 but the complexity and heterogeneity of docking methods should be taken into account SBDD is not limited to docking techniques Outside the VS approach when hits have to be optimized to the largest extent possible more versatile and much more computationally costly methods such as molecular dynamics 77 79 and free energy calculations 80 are clearly more suited It is worthy to note that contrary to docking programs depicted on Table I the internals of the most popular molecular dynamics packages 81 93 share the same theoretical principles differences between the forcefields and the integrators being more technical than conceptual Furthermore unlike vHTS that can cut down costs drastically in hit d
100. ate if water could affect differently the latter s binding modes or energies Figure 1 Chemical structures of ligand 1 and ligand 2 POS pY 1 pY 2 N 2 H o A o NAVA N PO HN N NH i y H z OC A N CH O NH oO pY 1 DY 1 O py 2 PO pYd OPO pY0 a ligand 1 b ligand 2 Material and Methods Design and Synthesis The design and synthesis of the pseudo peptidic linear ligand 1 were already described in previous works 6 29 In its complex with Grb2 SH2 as observed in the crystal structures ligand 1 adopts a B I turn conformation B II turn mimetics have been largely described in the literature contrary to B I turn mimetics 41 In such a turn structure the oxygen of the 1 carbonyl group forms an H bond with the backbone NH of the 2 residue Eguchi et al have developed a bicyclic structure tetrahydro 2H pyrazino 1 2 a pyrimidines 4 7 diones as a B I turn mimetic The advantage in this structure is that all the side chains of a four residues peptide are reserved which is uncommon with such mimetics 42 43 For this reason this bicycle structure was adopted as the starting point for the development of non peptidic Grb2 SH2 inhibitors first leading to a mimetic of Ac pTyr aMe pTyr Asn NH CH 2 3 1 naphtyl In the proposed molecule ligand 2 both phosphate groups of pTyr and aMe pTyr are replaced by phosphonate groups which are stable to phosphatases The pTyr mimetic building block was prepared from 2 4 bromo ph
101. ater molecules thus defining a water H bond network where the S14 Ser BG and Nj43 Asn BG residues are also involved Conversely in the case of ligand 2 one water molecule remains in the vicinity of the phosphone during almost 1 ns This water makes H bonds with the 3 phosphoney oxygens and with the terminal atoms of residues Rgs Arg BB Sss Ser BB and Soo Ser BC It is thus located between the phosphone and the entrance of cavity 1 All other water molecules interacting with both the ligand and the protein exchange very rapidly The averaged number of water molecules H bonded to ligand 2 has increased from 4 to 9 as the simulation proceeded This is correlated with the higher exposure of the ligand escaping from the active site Influence of bulk solvent on protein residues As the tracking of a limited number of stable water molecules interacting both with the protein and the ligands could not alone account for the solvent effects on ligand binding we measured the interaction energies between all protein residues and all water molecules The results restricted to the Grb2 SH2 receptor residues are presented as energy bars in figure 11 a for the complex with ligand 1 in figure 11 b for the complex with ligand 2 and in figure 11 c for the uncomplexed Grb2 SH2 Figure 11 Interaction energy between Grb2 SH2 residues and water a ligand 1 complex top left b ligand 2 complex top right c uncomplexed Grb2 SH2 bottom TUE
102. aux r sultats Dynamique mol culaire sur deux complexes de r f rence Situation initiale Mise en ceuvre Syst mes mod lis s Techniques d analyse Choix du champ de force Validation du protocole Nouvelles connaissances concernant l interaction de ligands sur le r cepteur Grb2 SH2 Effets du solvant Propositions au niveau du design de nouveaux ligands Docking flexible sur des bases de mol cules Situation initiale Prise en charge du solvant Test et validation des param tres Limitations Screening virtuel sur la cible Grb2 SH2 Mol cules s lectionn es comme candidates R sultats Validation Situation pr sente Screening virtuel mise en place d outils innovants Principales probl matiques li es au screening virtuel Le projet VSM G Pr sentation D veloppements en cours et objectifs Perspectives long terme vers un screening virtuel intelligent Conclusions Publications 30 30 30 30 31 32 32 32 33 34 34 34 35 36 36 36 39 47 47 47 48 48 49 49 50 51 51 51 52 52 52 52 53 53 53 54 55 55 56 56 57 57 58 58 59 61 Pr sentation L application de simulations num riques reposant sur les bases th oriques de la chimie est une approche relativement r cente dans le monde de la recherche pharmaceutique Plusieurs facteurs la rendent particuli rement attrayante Tout d abord son champ d application cro t parall lement aux progr s r
103. avatti G Fretz H Schoepfer J Griitter M G Structural basis for specificity of GRB2 SH2 revealed by a novel ligand binding mode Nat Struct Biol 1996 3 586 589 26 Nioche P Liu W Q Broutin I Charbonnier F Latreille M T Vidal M Roques B Garbay C Ducruix A Crystal structures of the SH2 domain of Grb2 Highlight on the binding of a new high affinity receptor J Mol Biol 2002 315 1167 1177 27 Garcfa Echeverria C Furet P Gay B Fretz H Rahuel J Schoepfer J Caravatti G Potent antagonists of the SH2 domain of Grb2 optimization of the X 1 position of 3 amino Z Tyr PO3H2 X 1 Asn NH J Med Chem 1998 41 1741 1744 28 Gay B Furet P Garcia Echeverria C Rahuel J Ch ne P Fretz H Schoepfer J Caravatti G Dual specificity of Src homology 2 domains for phosphotyrosine peptide ligands Biochemistry 1997 36 5712 5718 29 Liu W Q Vidal M Gresh N Roques B P Garbay C Small peptides containing phosphotyrosine and adjacent aMe phosphotyrosine or its mimetics as highly potent inhibitors of Grb2 SH2 domain J Med Chem 1999 42 3737 3741 30 Eck M J Shoelson S E Harrison S C Recognition of a high affinity phosphotyrosyl peptide by the Src homology 2 domain of p56lck Nature 1993 362 87 91 31 Machida K Mayer B J The SH2 domain versatile signaling module and pharmaceutical target Biochim Biophys Acta 2005 1747 1 25 3
104. base 65 66 Diversity of each of the three subsets the 598 375 database the 8 383 diversity set and the 31 220 CN was measured as fractions of the total diversity 62 Results are depicted in table 1 It appears that the 8 383 subset and the larger CN database are of comparable diversity The former is therefore suitable as input data for a VS validation experiment Interestingly from the large scale database to the diversity subset we only traded 40 of the diversity for a 98 6 size reduction The 8 383 compounds database was pre processed into VSM G ligand preparation modules which made it suitable for the docking programs used afterwards Molecules were first converted into 3D and then their protonation state was set arbitrarily at pH 7 As MSSH SHEF is a rigid shape matching procedure a conformational search was performed retaining at most 400 conformers per compound giving 1 102 299 conformers Table 1 Diversity analysis of the reference database used in this paper here referred to as the diversity subset of 8 383 compounds In the table 100 diversity is that of the union of the large scale and CN databases All values are computed by the ScreeningAssistant software Please refer to Monge et al for details on how drug like and lead like compounds are defined and how molecular database diversity is measured number of drug like lead like drug like lead like global compounds compounds compounds diversity diversity di
105. by polarizable molecular mechanics Validation on model binding sites by quantum chemistry J Comp Chem 2005 26 1131 1147 61 Gresh N Shi G B Conformation dependent intermolecular interaction energies of the triphosphate anion with divalent metal cations Application to the ATP binding site of a binuclear bacterial enzyme A parallel quantum chemical and polarizable molecular mechanics investigation J Comp Chem 2004 25 160 168 62 Langlet J Claverie P Caillet J Pullman A Improvements on the continuum model Application to the calculation of the vaporization thermodynamic quantities of nonassociated liquids J Phys Chem 1988 92 1617 1631 63 Phan J Shi Z D Burke Jr T R Waugh D S Crystal structures of a high affinity macrocyclic peptide mimetic in complex with the Grb2 SH2 domain J Mol Biol 2005 353 104 115 Multiple step virtual screening using VSM G Overview and validation of fast geometrical matching enrichment Alexandre Beautrait Vincent Leroux Matthieu Chavent L o Ghemtio Marie Dominique Devignes Malika Smail Tabbone Wensheng Cai Xuegang Shao Gilles Moreau Peter Bladon Jianhua Yao Bernard Maigret 1 Nancy Universit LORIA Groupe ORPAILLEUR Campus scientifique BP 239 54506 Vandceuvre l s Nancy Cedex France 2 Nankai University Department of Chemistry Tianjin 300071 P R China 3 30 Avenue Jean Jaur s 94220 Charant
106. cates obtenir pour des assemblages biologiques de taille importante et exigent de larges moyens de calculs qui les excluent du champ d application du screening virtuel Ne sont principalement accessibles que des programmes de docking ou des approches empiriques et statistiques m thodes fondamentalement approch es On consid re donc souvent raison que telles qu elles les m thodes de screening virtuel ne sauraient se substituer totalement aux approches classiques de la recherche pharmaceutique Il existe l heure actuelle un tr s large choix de programmes de docking chacune de ces approches correspond une mod lisation sp cifique au niveau de la flexibilit des structures de l algorithme d exploration de l espace conformationnel ainsi que de la fonction de score utilis e pour d finir celui ci De plus la possible prise en compte du solvant aspect qui comme nous l avons vu peut s av rer d terminant dans certains cas est souvent insuffisante au del des limitations intrins ques du docking De fa on remarquable une analyse d taill e du fonctionnement interne des principaux programmes de docking r v le que chacun poss de des points forts et des points faibles caract ristiques Cette h t rog n it est double tranchant si elle peut conf rer une certaine libert quiconque poss de suffisamment d expertise elle induit une complexit qui limite l usage du screening virtuel et est susceptible
107. ch is a surface based method appears more sensitive to the active site shape specificities than GOLD which relies on a classical atom coordinates based representation of molecular structures But in contrast to GOLD SHEF appears unable to assess size constraints correctly This could be related not to SHEF itself but rather to its current implementation within the VSM G screening funnel Indeed only the best conformer score is retained for ranking each compound the diversity of geometrically acceptable conformations referred to as adaptability is not taken into account This could lead to SHEF producing false positives with ligands occupying almost all the active site volume These ligands might require a minimal adaptability in order to provide a good chance to satisfy chemical constraints upon binding in addition to geometrical complementarity Relationship between SHEF and GOLD classifications Figures 7 g 7 h and 7 i depict the relationships between SHEF and GOLD ranks for 1P8D 1PQ6 and 1PQ9 respectively Given the fundamental differences between these two programs it is not surprising to see lower correlation between SHEF and GOLD than between two different target conformations for either SHEF or GOLD We are however far from the random case depicted in figure 6 a thus it is clear that noticeable enrichment using SHEF is already observed at this point If the general profile of the three density plots is similar they differ regarding t
108. cking et enfin du screening virtuel On pourra constater que de nouvelles connaissances sur la nature physico chimique de la liaison de certains ligands sur Grb2 SH2 sont mises en vidence dans le premier cas tandis que les deux autres approches ont surtout donn naissance a des avanc es m thodologiques Les publications auxquelles ce travail a donn lieu sont enfin disponibles Contexte Cancer Avant propos Le cancer est une famille de maladies caract ris e par une d r gulation des m canismes de division cellulaire d un organisme 1 La croissance cellulaire incontr l e qui en r sulte parasite le fonctionnement normal de la machinerie biologique et peut aboutir des d ficiences graves d un ou plusieurs organes vitaux principalement reins foie et poumons Ainsi la plupart des cancers humains peuvent causer la mort et il s av re que les cancers sont devenus la premi re cause de mortalit dans les pays d velopp s 2 Le nombre de d c s en France provoqu s par le cancer cro t constamment 150000 en 2000 contre 125000 en 1980 parall lement l accroissement de l esp rance de vie les cancers touchant principalement et avec une gravit accrue les personnes les plus g es Cette augmentation est toutefois moins importante que celle du nombre de cas diagnostiqu s 278000 contre 170000 3 qui traduit aussi bien les progr s de la m decine anti cancer que les efforts de sant publique principal
109. clustering averaging and minimization of conformations from NMR data or MD sampling The screening funnel a multiple step strategy Presently a wide variety of virtual screening programs are available and it is generally assumed that a well chosen combination of methods will give better results than a single one The interest for such multiple step VS protocols has been stressed in various papers often as a combination of a single structure based docking calculation with ligand based approaches as pre filters 5 41 Post processing refinements starting from docking results have also been reviewed 15 42 Alternatively several methods can be employed at different stages within a given docking program 43 The use of several docking programs in the same protocol 44 is less frequent Moreover most programs require significant expertise in setting up and analyzing the results More generally each technique features a specific balance between the speed of calculations and the reliability of results 45 Open software tools overcoming such limitations are lacking The virtual screening implementation in the VSM G platform is constituted by a series of different structure based methods organized sequentially in a funnel strategy The techniques range from simple methods to more sophisticated ones profiting from the speed of the former and the accuracy of the latter At each step of the process the filter discards inappropriate compounds T
110. contraire dans le cadre le la recherche anti cancer l interdisciplinarit est d une importance cruciale tant le sujet d tude est complexe En premier lieu la pr vention est fondamentale car il est tabli que plus un cancer est d cel pr cocement et les traitements th rapeutiques effectu s en cons quence meilleures sont les chances de survie du patient La pr vention inclut des actions de d pistage et de sensibilisation du public mais aussi des tudes cliniques et pist mologiques Ces derni res permettent cerner avec plus de pr cision les facteurs de risque ce qui constitue un pr alable indispensable toute action de pr vention visant les populations risque mais permet galement de recouper les connaissances issues des recherches en biologie et de fournir au milieu m dical des donn es statistiques pr cieuses pour la prise en charge et le suivi des patients Si des examens r guliers sont indispensables pour les formes de cancer les plus courantes on peut galement isoler les facteurs de risques potentiels chez un individu Une fois le cancer diagnostiqu le choix d une ou plusieurs approches d pend de la nature du cancer de son avancement ainsi que de l tat de sant g n ral du patient concern L intervention chirurgicale 29 extraction de tumeur ou amputation de l organe affect plus ancienne m thode th rapeutique de traitement contre le cancer est toujours largement pratiqu e d s lors
111. contre les croix indiquent les substitutions envisageables pour synth se Le squelette en R3T SN question a finalement t abandonn Parmi les 30000 mol cules dock es par la suite de fa on pr visible les mol cules construites manuellement partir de ligand 1 et d autres ligands actifs connus produisent les meilleurs r sultats Plusieurs modifications sugg r es par l analyse des dynamiques de ligand 1 donnent lieu des scores am lior s De fa on coh rente avec les connaissances pr alables sur l inhibition de Grb2 SH2 la s quence de r f rence pYVNV interagit moins fortement que ligand 1 tandis que pYpYN dont ligand 1 est issu est encore moins actif orientation du r sidu pY non optimis e Diff rentes mol cules disposant d une charge lectrique inf rieure ou gale celle de la r f rence et ne poss dant pas de groupe phosphate instable sont mises en vidence Ces structures pourraient s av rer utiles en tant que bases pour la mise au point d inhibiteurs de Grb2 SH2 plus int ressants en tant que m dicaments potentiels La structure des complexes correspondants pr dite par docking peut constituer la base d optimisations par mod lisation mol culaire Mol cules int ressantes issues des calculs de docking flexible class es suivant leur score GOLD vis vis de Grb2 SH2 d croissant de haut en bas De gauche droite mol cules neutres charg es 1 2 et 4 Les mol cules encadr es
112. cording to the one found in the NMR structure of the Grb2 protein complexed with a Shc derived peptide PDB code 1QG1 46 This histidine protonation state appears optimal for intramolecular interactions with neighbor residues All the water molecules from X ray data that were distant by more than 5 A from the protein or the ligand were suppressed We thus obtained a set of 16 water molecules located in or near the protein ligand binding site and these were labeled specifically in order to follow their individual trajectories during the MD simulations The complex 1551 atoms from 96 residues 87 ligand 1 atoms was next immersed at the center of a large cubic water box of 80 A length which contains approximately 16500 water molecules to simulate the biological environment in a realistic way TIP3 was the model chosen for all waters Figure 2 Preparation of the pTyr mimetic building block COH CO tBu CO tBu COH 2 tBuOH Pd PPh 4 Br Br PO3Et POsEty Figure 3 Synthesis of ligand 2 HO Resin o 4 o S Br Resin gt Resin __ PmocNH Os Resin BrCH CH OEt y p UK Se ie i o laph NHTit ve NHFmoc OT io tog Et03P MAA kean NAA esn o A o o NH Naph NHT Naph POsEty POsEtz t H ZO i OY a i N ya EROP ow NAG EROP 8 nA gm No 2 LE z NHTrt NHT POsEt POsEt P o OXY o H203P Sy nee N nA Et 05P TOR Mann So NALS Y Da NH gt POsEty oO PO3H2 Th
113. ctes let 000001 1 mat 604 0001 _10 mat How would you lie to fer your molecular DA Prevost crestes OF D Oummeune s a Sub atructure fle generated using GIS Draw The VSM G features regarding the ligand database preparation and its target related capabilities are listed on chart 1 and chart 2 respectively Current development is mostly concentrated on the screening funnel part Chart 1 Current VSM G features ligand database preparation Database creation and handling generation of virtual combinatorial libraries from chemical scaffolds and fragments merging of molecular files with detection of duplicate structures support for different file formats the most popular SDF 23 and MOL2 24 as output conversion between formats using in house code or OpenBabel 25 implementation of the MarvinBeans library 26 and VIDA 27 for database browsing if available Substructure search flexible criteria through combinations of simple operators and or not have at least at most support for SMILES 28 SDF and RDF 23 as input internal use of a canonical topology coding that greatly reduces the complexity of the requests quickly searches through millions of compounds on desktop computers once the coding is performed Toxicity prediction implementation of PCT 29 a carcinogenicity prediction program based on SAR exclusion of presumably toxic compounds possible enrichment of the databa
114. d 3D QSAR in drug design part 1 Methodology Drug Discovery Today 2 issue 11 1997 457 467 33 Kubinyi H QSAR and 3D QSAR in drug design part 2 Applications and problems Drug Discovery Today 2 issue 12 1997 538 546 34 Forrest S Genetic algorithms principles of natural selection applied to computation Science 261 1993 872 878 35 Schneider G and Wrede P Artificial neural networks for computer based molecular design Progress in Biophysics and Molecular Biology 70 issue 3 1998 175 222 36 Zapan J and Gasteiger J Neural networks in chemistry and drug design 2nd ed 1999 Wiley VCH 402 pages 37 Walters W P Stahl M T and Murcko M A Virtual screening an overview Drug Discovery Today 3 issue 4 1998 160 178 38 Bissantz C Development and application of new methods for the virtual screening of chemical databases Thesis 2002 Swiss Federal Institute of Technology Zurich 292 pages 39 Shoichet B K Virtual screening of chemical libraries Nature 432 2004 862 865 40 Mestres J Virtual screening a real screening complement to high throughput screening Biochemical Society Transactions 30 issue 4 2002 797 799 41 von Itzstein M Wu W Y Kok G B Pegg MS Dyason J C Jin B van Phan T Smythe M L White H F Oliver S W Colman P M Varghese J N Ryan D M Woods JM Bethell R C Hotham V J Cameron J M and Penn C R Rational design of potent sialidase based inhibitors o
115. d biomolecular simulation Philosophical Transactions of the Royal Society of London Series A Mathematical Physical and Engineering Sciences 363 issue 1833 2005 2017 2035 111 Grid 5000 http www grid5000 fr 112 Paolini G V Shapland R H B van Hoorn W P Mason J S and Hopkins A L Global mapping of pharmacological space Nature Biotechnology 24 issue 7 2006 805 815 113 Moustakas D T Lang P T Pegg S Pettersen E T Kuntz I D Brooijmans N and Rizzo R C Development and validation of a modular extensible docking program DOCK 5 Submitted for publication 2006 114 UCSF DOCK http dock compbio ucsf edu 115 Ewing T J A Makino S Skillman A G and Kuntz I D DOCK 4 0 Search strategies for automated molecular docking of flexible molecule databases Journal of Computer Aided Molecular Design 15 2001 411 428 116 Gropp W Lusk E and Skjellum A Using MPI Portable parallel programming with the message passing interface 1994 MIT Press 328 pages 117 Gropp W Lusk E Doss N and Skjellum A A high performance portable implementation of the MPI message passing interface standard Parallel Computing 22 issue 6 1996 789 828 118 Jones G Willett P and Glen R C Molecular recognition of receptor sites using a genetic algorithm with a description of desolvation Journal of Molecular Biology 245 issue 1 1995 43 43 119 Jones G Willett P Glen R C Leach A R and Taylor
116. d high throughput screening for novel inhibitors of protein tyrosine phosphatase 1B Journal of Medicinal Chemistry 45 2002 2213 2221 48 Gr neberg S Stubbs M T and Klebe G Successful virtual screening for novel inhibitors of human carbonic anhydrase Strategy and experimental confirmation Journal of Medicinal Chemistry 45 2002 3588 3602 49 Shoichet B K McGovern S L Wei B and Irwin J J Lead discovery using molecular docking Current Opinion in Chemical Biology 6 2002 439 446 50 Vangrevelinghe E Zimmermann K Schoepfer J Portmann R Fabbro D and Furet P Discovery of a potent and selective protein kinase CK2 inhibitor by high throughput docking Journal of Medicinal Chemistry 46 issue 13 2003 2656 2662 51 Kraemer O Hazemann I Podjarny A D and Klebe G Virtual screening for inhibitors of human aldose reductase Proteins Structure Function and Bioinformatics 55 2004 814 823 52 Lahana R How many leads from HTS Drug Discovery Today 4 issue 10 1999 447 448 53 Golebiowski A Klopfenstein S R and Portlock D E Lead compounds discovered from libraries Current Opinion in Chemical Biology 5 issue 3 2001 273 284 54 Golebiowski A Klopfenstein S R and Portlock D E Lead compounds discovered from libraries part 2 Current Opinion in Chemical Biology 7 issue 3 2003 308 325 55 Shoichet B K Screening in a spirit haunted world Drug Discovery Today 11 issue 13 14 2006 607
117. de vari t de domaines En premier lieu on peut citer les progr s de l informatique aussi bien sur le plan mat riel puissance de calcul brute architectures parall les et distribu es que logiciel outils sp cialis s techniques de programmation La disponibilit plus importante de structures exp rimentales est galement tr s profitable les progr s technologiques des appareillages ont un impact positif pour les mod lisateurs Il en va de m me pour les avanc es purement th oriques ainsi que les nouvelles d couvertes en biologie et biochimie qui peuvent permettre d am liorer les mod les au niveau conceptuel Nous nous situons donc avec ce travail dans un contexte r solument interdisciplinaire Relier conceptuellement des domaines si diff rents constitue bien videmment un d fi scientifique particuli rement int ressant mais la m daille a son revers En effet nous allons tre confront s rapidement une grande h t rog n it au niveau des moyens mis en uvre des appareillages des concepts des donn es A notre niveau on peut supposer que les difficult s qui vont appara tre seront plus d ordre technique au niveau de la mise en uvre des moyens informatiques que purement conceptuels Dans le domaine de la mod lisation mol culaire les conditions permettant d obtenir un mod le tr s r aliste sont connues il suffit d appliquer les lois de la m canique quantique Mais en contrepartie les
118. ding sites on Grb2 SH2 respectively labeled cavity 1 cavity 2 and cavity 3 in the present work Cavity 1 is the main pTyr binding pocket common to all SH2 domains In Grb2 SH2 it is composed of residues Re7 Arg aA Rss Arg BB Sss Ser BB Soo Ser BC and Sos Ser BC This Grb2 SH2 residue notation corresponds to the residue numbering in Grb2 as reported in the 1JYQ PDB structure alternate between brackets is as proposed by Eck 30 for defining the a helix B sheet succession BaBBBBBa seen in most SH2 domains 31 Cavity 2 is composed of Kio Lys BD Li29 Leu BE and W21 Trp EF While such an hydrophobic binding site is present in all SH2 domains in the case of Grb2 the specific Wj 2 Trp EF residue restrains its size and forces ligands to adopt a B turn conformation 32 Peptidic and pseudo peptidic ligands can adopt such a required conformation with the presence of an Asn residue 33 the 2 in subscript indicates that the Asn is 2 residues C terminal to the main pTyro Cavities 1 and 2 are both targeted by the natural Grb2 SH2 ligand peptide sequence pY VNV and it should be the case as well for potential ligands 34 Cavity 3 constitutes a second pTyr binding pocket specific to Grb2 and to our knowledge only targeted by ligand 1 It is composed of residues S141 Ser BG Rig Arg BG and N 43 Asn BG The X Ray structure of the SH2 domain of Grb2 in complex with ligand 1 is available in the Protein Data Bank 35 PDB
119. du domaine SH2 de SAP pouvait tre li des inhibiteurs non seulement non phosphat s mais aussi non charg s inhibiteurs peptidiques qui plus est En effet SAP SH2 peut se lier avec une affinit Ka 0 6 uM SLAM autour du r sidu Y gt lorsque celui ci n est pas phosphoryl 135 136 L analyse de la structure RX 137 ou RMN 138 du complexe form indique que le mode de liaison diff re de celui des ligands typiques des domaines SH2 en g n ral car il implique la liaison dans trois cavit s sur le r cepteur et non plus deux Des exp riences de screening d termin rent la s quence de r f rence comme tant T S XXoXX V D ce qui indique que la s lectivit n est pas d termin e par un r sidu Yo ou pYo mais principalement aux positions 2 et 3 138 Cela indique au niveau de la s quence de r f rence TIY AQV sur SLAM que la somme des interactions des r sidus T et V de la s quence TIYoAQV est plus importante que celle du r sidu Yo Ces travaux sur SAP SH2 n ont pas notre connaissance donn lieu la mise au point d un inhibiteur SH2 neutre et actif in vivo Ils montrent toutefois que le ciblage de cavit s lointaines par des groupes neutres est susceptible du moins pour certains domaines SH2 de donner lieu des interactions qui pourraient apr s optimisation compenser celles du r sidu pTyr de r f rence En ce qui concerne Grb2 SH2 deux cavit s distinctes de celle du r sidu pTyr e
120. dus et automatis s permet l application un mod le ou une combinaison de mod les des syst mes dont la taille et la complexit ne les rend pas traitables par d autres moyens Les techniques informatiques permettent ainsi de se rapprocher de la r alit physique validant ou invalidant les mod les avec plus de rigueur Si l informatique n est pas l origine de la d couverte de th ories scientifiques bien souvent elle seule permet d tendre le champ d application des mod les qui en d coulent Les simulations num riques une nouvelle m thode de recherche L apparition de moyens de calcul informatique de plus en plus importants la loi de Moore jusqu ici v rifi e postule que la puissance de calcul des microprocesseurs double tous les 18 mois partir des ann es 1950 a alt r le bin me th orie exp rience de la recherche scientifique en ins rant une m thode interm diaire la simulation num rique Fondamentalement aussi bien l aspect th orique que l aspect exp rimental de la recherche sont int gr s dans l outil informatique le premier travers des programmes et des algorithmes et le second travers des simulations et des donn es Ainsi la conduite d une simulation peut tre analogue selon les circonstances la celle d une exp rience ou la mise en place d une th orie rendant l une comme l autre plus accessibles Un autre avantage consid rable de l informatique doit galemen
121. e 9 1994 3483 3491 18 Skolnik E Y Batzer A Li N Lee C H Lowenstein E Maohammadi M Margolis B and Schlessinger J The function of GRB2 in linking the insulin receptor to Ras signaling pathways Science 260 issue 5116 1993 1953 1955 19 Machida K and Mayer B J The SH2 domain versating signaling module and pharmaceutical target Biochimica et Biophysica Acta 1747 2005 1 25 20 Sadowski I Stone J C and Pawson T A noncatalytic domain conserved among cytoplasmic protein tyrosine kinases modifies the kinase function and transforming activity of Fujinami sarcoma virus P130 gag fps Molecular and Cellular Biology 6 issue 12 1986 4396 4408 21 Russel R B Breed J and Barton G J Conservation analysis and structure prediction of the SH2 family of phosphotyrosine binding domains FEBS Letters 304 1992 22 Pawson T and Nash P Assemby of cell regulatory systems through protein interaction domains Science 300 2003 445 452 23 Eck M J Shoelson S E and Harrison S C Recognition of a high affinity phosphotyrosyl peptide by the Src homology 2 domain of p56lck Nature 362 1993 87 91 39 24 Songyang Z Shoelson S E Chaudhuri M Gish G Pawson T Haser W G King F Roberts T Ratnofsky S Lechleider R J Neel B G Birge R B Fajardo J E Chou M M Hanafusa H Schaffhausen B and Cantley L C SH2 domains recognize specific phosphopeptide sequences Cell 72 issue 5 1993 767 778
122. e American Chemical Society 123 2001 12238 12247 64 Ladbury J E Just add water The effect of water on the specificity of protein ligand binding sites and its potential application to drug design Chemistry amp Biology 3 issue 12 1996 973 980 65 Charifson P S Corkery J J Murcko M A and Walters W P Consensus scoring A method for obtaining improved hit rates from docking databases of three dimensional structures into proteins Journal of Medicinal Chemistry 42 issue 25 1999 5100 5109 66 Feher M Consensus scoring for protein ligand interactions Drug Discovery Today 11 issue 9 10 2006 421 428 67 Beautrait A Leroux V Chavent M Maigret B Cai W Shao W Moreau G Bladon P Yao J Liao Q Yu F and Souchet M VSM G the Virtual Screening Manager platform for computational Grids Example of use for the identification of putative liver X receptor ligands Manuscript in preparation 2006 68 Dobson C M Chemical space and biology Nature 432 2004 824 828 69 Newton C G and Lockey P M The importance of early pharmacokinetics Current Drug Discovery April 2003 2003 33 36 70 Sadowski J Optimization of the drug likeness of chemical libraries Perspectives in Drug Discovery and Design 20 2000 17 28 71 Irwin J J How good is your screening library Current Opinion in Chemical Biology 10 issue 4 2006 352 356 72 Lipinski C A Lombardo F Dominy B W and Feeney P J Experimental
123. e YIM poyeroosse ABOU UOTIOVIDIUL JA Ina Ima lest Joy ANTIQIXOI MA sv p9199J9P qjeonewomy CI pourensuod AO MON sqol AIARIVHO snoprezey se ize Surumeds puewop Aquo xun xeuds WWAYHO pouodor juounean zs1 wer3oxd IST uo oJqeIIeAr f H Jo uorsuayxo we1801d 34 0 Jo 2q WO UONBZIWITUIU UOALIP YO NIARIVHO 2endod oy ON Ima ainpacoid SA 2p09 9910 NASA ue se pajuouro dur ne JIM pardnos st NIMAVQ ese ojqusn AJIPEO P NOS SE P 14u tordu euondo IM poy ST NIMAVA s LOST Lv HSSIN a DEAR SA 8p1 LSdv asn HSSIN q passaoord aid spuesy isp ovr svt pmu g UT apnpouu sroyne A pm podnoo Suryoyeur eoLnow0es amd rri i TIIM SUOISIOA P TIe 10 QUES syuarolyooo ON uonoung Auejrurs odeys 5 i ON ON ASIIJ AU SE ROpT au yoRIUOD 19 D INSA ds i soovjins soruowey eotiayds AAHS sey Ayowonxg D WSA 2 ur poyuauaydury SOIUOULIEY jeonayds Jo yas oy JO UONZIUTUTA GA HR ASNA l ur pouuejd se pourgop adeys 10jda00y One UOISIOA Uroods spuest SIOULIOJUOD JA v pue uonouny duos Sursn a qissod bs P y 7 puey punoq 7 gt i p z das uolstoa1d paods SOILI9PB98 0 WIM oqissod ert au YIM Suoye 159q 24 JO JUSWIOUTJOI Sulioos omduw ue woy s1ojdasar jdy mu 7 i UMOUY amp JO UOTJEULIOJUOS Suryoop 1511 trI Or1 oy Sutaosdurt 3J SI OIJSOUIA SI9JSNIO IND onsoryy nr mod PUS J0818 ou OW uonezrurunu uage 9A97 ssa001d Suryoop ym SUTJO0P SSOIT ie Pasties 3D Jo
124. e starting structure of the complex with ligand 2 was obtained by superimposing ligand 2 over ligand 1 using the two phosphorous atoms as superimposition points and then suppressing ligand 1 Two other systems were also preparated by simply removing all water molecules from the previously described models in order to perform again the in vacuo MD simulations for ligand 1 and ligand 2 complexes using the same parameters and force field than the dynamics with explicit water Another model was prepared from the ligand 1 complex in vacuo by removing ligand 1 and recreating the 80 A waterbox The initial conformation of Grb2 SH2 is not initially relevant in such a system because it includes the induced fit effects upon ligand binding while the ligand has been removed Our first goal here is to retrieve the unbound Grb2 SH2 conformation that is known experimentally from the complexed structures which would be an indication of the correctness of these models Besides after this conformational change this MD simulation will provide interesting information regarding the effects of solvent on receptor residues To generate trajectories of the complexes we used the NAMD program 47 coupled with the adapted CHARMM22 force field Before running the calculations appropriate counter ions were added replacing water molecules at the boundaries of the water box Figure 4 represents the starting structure as modelled for the ligand 1 complex MD simulation Periodic boundar
125. e starts from the even larger ensemble of synthesizable small chemical structures The exploration here involves sorting out molecules with no or unwanted biological effects restraining the chemical space 21 to the so called drug space 22 Eventually merging the target space with the drug space leads to a third ensemble of receptor ligand associations that have to be explored successfully in order to solve the drug discovery problem Provided that the ensembles of targets and candidate molecules have been previously reduced efficiently to avoid a combinatorial explosion this is still a long and arduous process VSM G rationalizes these searches It focuses on the exploitation and management of current knowledge of the proteome to target and chemical to drug steps It also relies on a specific protocol relying on structure based virtual screening methods regarding the final ligand to hit process Its workflow has been designed so to match the processes described above and is summarized in figure 1 The basic organization of the platform is therefore divided in three distinct parts two for the preparation of input data ligands and protein targets respectively and a third one which is a multi layer funnel for the in silico screening Figure 1 Basic VSM G workflow for hit discovery Substructure Toxicity Database search prediction 3D tools F x generation Ligand preparation modules N Screening funnel Can Protein Target preparat
126. ed with the ligand 1 mAz group as well as with the ligand 2 naphthalene However in the former case this did not appear to lead to the disruption of the binding as all other groups did remain in their initial positions bound to cavities 1 2 and 3 This observation is in agreement with biological assays which demonstrated that the mutation of mAz did not have a significant impact on ligand 1 affinity 26 More importantly the analysis of MD data revealed that the lack of affinity of ligand 2 originates from the evolution of the binding of both phosphone groups particularly the one targeting cavity 3 This does not seem to be caused by a weakness of the interaction between ligand 2 and cavity 3 residues as it is initially of the same order of magnitude than the phosphate interaction of ligand 1 but rather in the hydrophilic nature of the very flexible arginine residue in cavity 3 This particular residue appears to have a clear tendency to move away from its serine and asparagine neighbors thus changing the shape of cavity 3 Consequently a ligand bound to cavity 3 has to maintain strong interactions with these residues in order to force the arginine to keep its position in cavity 3 as observed in the ligand 1 complex X ray structure If this is not the case there would be a strong probability that the arginine can move towards the solvent while maintaining a large interaction with a negatively charged group of the ligand this group would t
127. eening Nature Reviews Drug Discovery 2 issue 5 2003 369 378 18 Veselovsky A V and Ivanov A S Strategy of computer aided drug design Current Drug Targets Infectious Disorders 3 issue 1 2003 33 40 19 Jain A N Virtual screening in lead discovery and optimization Current Opinion in Drug Discovery amp Development 7 issue 4 2004 396 403 20 Ofran Y Punta M Schneider R and Rost B Beyond annotation transfer by homology novel protein function prediction methods to assist drug discovery Drug Discovery Today 10 issue 21 2005 1475 1482 21 Dobson C M Chemical space and biology Nature 432 2004 824 828 22 Oprea T I and Gottfries J Chemography The art of navigating in chemical space Journal of Combinatorial Chemistry 3 issue 2 2001 157 166 23 http www mdl com solutions white_papers ctfile form ats jsp 24 http www tripos com data support mol2 pdf 25 http www openbabel sourceforge net 26 http www chemaxon com products html 27 http www eyesopen com 28 Weininger D SMILES a chemical language and information system 1 introduction to methodology and encoding rules Journal of Chemical Information and Computer Sciences 28 issue 1 1988 31 36 29 Liao Q Yao J H Li F Yuan S G Doucet J P Panaye A and Fan B T CISOC PCST a predictive system for carcinogenic toxicity SAR and QSAR in Environmental Research 15 issue 3 2004 217 235 30 Sadowski
128. eld of docking is constantly growing with new programs being introduced regularly fuelling a strong competition In most cases the programs are being benchmarked for their ability to reproduce known structures while less often the ranking capability on a reference VS experiment including more or less potent binders as well as decoys is also tested 58 Emphasis is put on reliability which is of course of premium importance for who wants to pick the most suitable program for his needs However in our opinion such information is just not enough relying solely on benchmark results eludes the fact that performing docking requires control over the process We will now survey a representative set of docking programs focusing on these aspects How do docking programs work The choice of docking methods is large see table I for providing a limited set of all docking programs and evaluating them is difficult for at least four reasons 1 each method corresponds to at least one ratio between speed and value of results two methods that clearly do not play in the same league here most probably do not have the same applicability in VS 2 the efficiency of a given method is more or less system dependant 3 each program is specific regarding its requirements concerning the nature and format of input data parameters and results establishing a benchmark in such a context could be very tedious 4 most importantly each method has its very own way to m
129. ement ax s sur la pr vention et le d pistage Le cancer n est pas seulement un axe crucial de sant publique c est galement un probl me social 4 5 Cela peut principalement s expliquer du fait que la vision populaire du cancer se d cline et se d forme bien en dehors de la r alit m dicale Tout d abord au cancer est quasi syst matiquement associ e l image de mourants condamn s pour lesquels la m decine et la science ne peuvent que s acharner inhumainement et en vain le cancer repr sente la mort violente pr c d e de la d ch ance physique celle que personne ne souhaite et qui inspire universellement la peur cela vient s ajouter le fait que le cancer est bien souvent interpr t de fa on exclusive soit comme fatalit soit sous langle causal Dans le premier cas le cancer est symbole d injustice et d incompr hension il peut frapper n importe qui n importe quand Dans le second cas sachant que certaines formes de cancer sont bien connues comme amplifi es par des comportements risques une g n ralisation peut s op rer si on en est malade c est qu on l a bien cherch Enfin le cancer est parfois consid r comme une maladie contagieuse Tous ces l ments combin s font du cancer une maladie honteuse ainsi qu un facteur important d exclusion sociale En 2002 la lutte contre le cancer est d clar e chantier prioritaire par l tat Un plan Cancer est al
130. ement of water molecules in the binding between a protein and a ligand can be separated in two classes i favorable enthalpic interactions by a limited number of waters stabilized between the ligand and the protein surface with some of them to be considered as part of the protein namely those retained in both ligand bound and free protein states and reported as stable in MD simulations and ii interactions with bulk solvent which are unfavorable for ligand binding as they compete with it In the present study it appeared that while the potency of ligand 1 could be enhanced by the contributions of a limited number of water molecules characterized by X ray crystallography the lack of affinity of ligand 2 was ascribed to destabilizing effects on the ligand caused directly by bulk solvent and or to conformational changes of protein residues which occur only when solvent is explicitly modelled Such effects could be correlated to specificities of the Grb2 SH2 system as there are numerous examples of successful in vacuo drug design strategies indicating that most protein ligand systems do not seem particularly sensitive to solvent effects such as those observed here Involvement of stabilized water molecules in the ligand 1 complex Predicting how a given water molecule could affect the binding without resorting to complex calculations like the MD simulations performed in this study is not obvious because as remarked by Dunitz 51 the entropic and
131. enyl acetic acid figure 2 The preparation of the aMe pTyr mimetic a amino acid was made by the method described by Liu et al 44 The C terminal hydrophobic building block 2 1 naphtyl propylamine was prepared by the method described by Furet ef al 45 The bicyclic molecule was synthesized following the method described by Eguchi ef al starting from a NovaSyn TG hydroxyl resin The diethyl 2 bromoacetal was condensed with the hydroxyl resin and then alkylated by 3 1 naphthyl propylamine Fmoc Asn Trt OH suitably protected Fmoc B aminoacid and pTyr mimetic building block were introduced successively by HATU HOAt DIEA coupling and 20 piperidine Fmod deprotection The peptidyl resin was then freed from the resin by formic acid catalyzed hydrolysis of the acetal group to form the aldehyde function which cyclized spontaneously to give the bicyclic molecule The final product was obtained by treating the resin free molecule with TMSI to hydrolyze the phosphonate protections figure 3 Computational simulations We have used the 1JYQ PDB structure 26 as a starting point It corresponds to the SH2 domain of Grb2 complexed with ligand 1 Since it is a dimer we only retained one monomer chain identifier B chain A and its ligand was removed Hydrogen atoms were then added according to the expected charge distribution of amino acids at pH 7 The particular protonation state of Hio His BD present in the binding site was taken ac
132. epancy between simulation and experiment could originate from that point Recent papers about a different SH2 domain raised our interest towards that direction in the case of Src SH2 in its complex with the peptide pYIpYV 39 the pTyr42 residue appears bound more to the water network around its position than to the protein indicating that water could play a significant role in the binding process Furthermore an analysis of available thermodynamical data for the SH2 domain of Src complexed by the pYEEI sequence suggested that solvent effects could contribute by as much as 25 to the total binding free energy 40 In the case of the complexes of the Grb2 protein with ligands high resolution X ray crystallographic data 26 34 36 also reveal that a limited number of water molecules are present at the protein ligand interface some of them buried inside cavity 1 with the pTyr residue The purpose of the present study is to contribute to unravel the possible role of the solvent and of these water molecules in particular in the binding of inhibitors to Grb2 SH2 and to provide insight towards improved inhibitor design Therefore in order to account for solvent effects on binding we have performed MD studies of the complexes of both ligand 1 and ligand 2 docked to Grb2 SH2 with the interface waters explicitly included the complex being immersed in a water box with periodic boundary conditions Applying the same procedure to ligands 1 and 2 should indic
133. eractions entre biomol cules L tude th orique de ces biomol cules est rendue possible par le progr s de certaines techniques exp rimentales et surtout grace a la banalisation des calculs informatiques il s agit d un axe de recherche relativement r cent Dans cette partie nous d crirons tout d abord les deux classes fondamentales de biomol cules les acides nucl iques ADN et ARN et les prot ines Les premi res contiennent le mode d emploi de I organisme tandis que les secondes uvrent son fonctionnement en assurant la communication entre les cellules Nous expliquerons ensuite en quoi le champ de vision des sciences du vivant s est consid rablement largi ces derni res ann es entre autres gr ce au poids croissant des m thodes de calcul informatique dans la recherche scientifique passant d une vision centr e sur le g nome une vision plus large qui inclut l ensemble des prot ines ou prot ome puis ensuite l ensemble des interactions entre biomol cules ou interactome Structure des biomol cules Des acides nucl iques l information g n tique Les acides nucl iques s observent sous deux formes polym riques l acide ribonucl ique ARN et l acide d soxyribonucl ique ADN dans lesquelles des groupes fonctionnels appel s nucl otides sont greff s sur un squelette polym re pentose phosphate Quatre nucl otides diff rents sont rencontr s dans la structure de l ADN
134. ering capability This will include a proof of concept study of the usefulness of post docking optimizations and molecular dynamics calculations as funnel modules following geometrical matching and flexible docking Next we will illustrate the whole screening funnel strategy through an actual large scale hit discovery campaign using computer grid architectures The relevance of using advanced techniques like target sampling and grid computations in such a context will also be highlighted Acknowledgments We thank Yesmine Asses Safia Kellou and Amel Maouche for their feedback Alexandre Beautrait was supported by grants from INRIA Institut National de Recherche en Informatique et en Automatique R gion Lorraine and ARC Association pour la Recherche sur le Cancer Vincent Leroux by a post doctoral fellowship from the INCa Institut National du Cancer Matthieu Chavent by a joined fellowship between CNRS Centre National pour la Recherche Scientifique and R gion Lorraine We thank Openeye for providing free access to OMEGA and VIDA software according to an academic license Chemaxon for supplying MarvinBeans Java library CCDC for the trial version of the GOLD program and the laboratory of chemoinformatics at the Orl ans University for the ScreeningAssistant program References 1 DiMasi J A Hansen R W and Grabowski H G The price of innovation new estimates of drug development costs Journal of Health Economics 22 2003 151 185
135. ermination de la s quence de r f rence Tous les domaines SH2 sont des r cepteurs naturels pour les s quences peptidiques de type pYXXX o pY d signe une tyrosine phosphoryl e et X n importe quel acide amin 92 En ce qui concerne Grb2 SH2 les s quences naturelles connues sont principalement pYINQ ligand EGF R 41 93 et pYVNV Shc 42 La premi re tape dans la mise au point d inhibiteurs pour Grb2 SH2 consista a identifier la s quence peptidique optimale Ce probl me fut r solu par Songyang et al qui effectu rent un screening par chromatographie des s quences pY XXX sur un grand nombre de domaines SH2 24 87 Il s av ra que les trois r sidus pr f rentiels cons cutifs la phosphotyrosine sont sp cifiques chaque domaine SH2 bien que des similitudes permettent d effectuer une classification des domaines SH2 en familles De fa on int ressante il est constat que certains domaines SH2 dont Grb2 sont hautement s lectifs sur la position pY 2 d autres par exemple Src le sont pour la position pY 3 Des travaux ult rieurs montr rent que certains domaines SH2 sont s lectifs bien en dehors du motif pYXXX 19 En ce qui concerne le domaine SH2 de Grb2 les s quences de type pYXNX sont pr f r es Diff rents screenings effectu s plus sp cifiquement sur Grb2 SH2 84 93 96 soulign rent ensuite V importance d un r sidu asparagine en position pY 2 De fa on plus pr cise des mesures c
136. es de grandeur d autant plus importante que la charge n gative port e par le groupe phosphate de pTyr est diminu e 36 Roller ef al NCI r ussirent ainsi d terminer la s quence peptidique cyclique nomm e GITE CELY EN 2VGMYC qui pr sente une affinit int ressante et pour laquelle de fa on remarquable les deux r sidus acides Glu et Glu semblent compenser l absence de groupe phosphate en position 0 95 Des exp riences de mutagen se sur cette base montr rent que chaque r sidu de la s quence de GITE except Gly 4 a un r le dans l affinit pour Grb2 SH2 et confirm rent l importance particuli re des r sidus Glu et Asn 95 142 143 De fa on int ressante la structure RMN du complexe de GITE avec Grb2 SH2 indiqua que GITE ne forme aucune liaison hydrog ne avec Grb2 SH2 les cha nes lat rales des peptides tant toutes tourn es l ext rieur du r cepteur de Grb2 SH2 Ainsi on peut supposer d une part que l affinit de GITE pour Grb2 SH2 provient de la conformation adopt e par le cycle qui serait particuli rement favorable 144 D autre part GITE constitue une base int ressante pour la mise au point d inhibiteurs non phosphoryl s de Grb2 SH2 au m me titre que la s quence Ac pYIN NH utilis e initialement par Novartis pour la mise au point d inhibiteurs phosphoryl s Diff rentes substitutions donn rent en effet lieu des compos s certes actifs 145 151 mais dont la charge to
137. es de d tection de faux n gatifs Une telle d tection pourrait impliquer le passage forc d une petite proportion de mol cules rejet es afin de valider la fiabilit d un filtre par les pr dictions de filtres ult rieurs La mise en place d un tel proc d n cessite une analyse en temps r el des r sultats l impl mentation de m thodes de clustering de bases mol culaires et peut d boucher sur une optimisation la vol e des param tres du funnel De tels m canismes d adaptation et d apprentissage s int grent naturellement l objectif d tendre VSM G une plateforme d acquisition de connaissances Cela n cessite une expertise des diff rents programmes employ s et l int gration de codes de manipulation de donn es plus particuli rement afin d assurer la conversion entre les formats de fichiers utilis s t Id alement il faudrait que VSM G puisse tre utilis ais ment par des scientifiques ayant peu ou pas d exp rience en mod lisation mol culaire et en chemoinformatique le rendant accesible en dehors du cercle des th oriciens Nous n en sommes pas encore 1a On trouvera dans l article plus de d tails concernant ces trois modules ainsi que sur les autres composantes de VSM G On utilise alors le funnel tant qu il reste du temps de calcul disponible et si ce temps est infini toutes les mol cules de d part passeront l ensemble des modules Un tel protocole peut tre e
138. es of both complexes at different times are shown in figure 6 Only limited variations around the initial X ray conformation take place with the ligand 1 complex trajectory conserving the main interactions of the two phosphate groups The most persistent hydrogen bonds are those occurring between the Rss Arg BB guanidinium group and the pTyro phosphate This phosphate is held in cavity 1 by a network of H bonds donated by Rss Arg BB Sss Ser BB Ego Glu BC and S s Ser BC The pTyr phosphate is anchored less deeply in cavity 3 through H bonds mostly involving R14 Arg BG guanidinium and S141 Ser BG hydroxyl groups The linker moiety between the two phosphate groups is maintained on the binding surface by H bonds involving mainly the guanidinium group of Rez Arg aA The stacking of the mAz aromatic ring with R Arg aA that was expected from in vacuo simulations appears unstable flipping between its initial parallel conformation and an anti parallel conformation oriented towards the solvent By contrast the ligand 2 complex trajectory presents several important conformational changes as illustrated by cluster graph analysis figure 5 b During the simulation the two phosphone groups present a clear tendency to escape from cavities 1 and 3 increasing the number of hydrogen bonds with the solvent at the expense of those with the protein Moreover the naphthalene moiety has immediately positioned itself in the bulk instead of reta
139. es overlap giving a classification of hits into different families regarding their selectivity for the three target conformations Out of a total of 1 414 molecules 670 47 bind specifically on one of the three conformations 356 25 bind on all the three conformations the rest binding on two out of three The amount of selective molecules on each conformation is 20 24 and 36 for 1P8D 1PQ6 and 1PQ9 respectively which is in agreement with the structural specificities highlighted previously Figure 5 Populations of hits defined from GOLD results of the 8 383 compounds diverse database For each target conformation 1P8D 1PQ6 and 1PQ9 the top scoring 10 structures are defined as hits The overlapping of these three sets is represented There are a total of 1 414 hit compounds that is defined as the target subset that has to be conserved through the filtering process 1P8D 838 Calm a 838 1PQ9 Total 838 1414 unique compounds Analysis of results An in house program was created for representing relationships between the screening results of two different techniques for the same set of input data Figure 6 explains the principles of the generated graphical representation Both ranks ranges are divided in twenty 5 blocks a sensible trade off between graphics clarity and the amount of represented information Three particular cases are provided as examples Figure 6 a depicts random selection and on the opposite figure
140. esearch 57 2001 447 454 144 Lou Y G Lung F D T Pai M T Tzeng S R Wei S Y Roller P P and Cheng J W Solution structure and dynamics of GTIE a nonphosphorylated cyclic peptide inhibitor for the Grb2 SH2 domain Archives of Biochemistry and Biophysics 372 issue 2 1999 309 314 145 Long Y Q Yao Z J Voigt J H Lung F D T Luo J H Burke Jr T R King C R Yang D and Roller P P Structural requirements for Tyr in the consensus sequence Y E N of a novel nonphosphorylated inhibitor to the Grb2 SH2 domain Biochemical and Biophysical Research Communications 264 issue 3 1999 902 908 146 Li P Zhang M Peach M L Zhang X Liu H Nicklaus M Yang D and Roller P P Structural basis for a non phosphorus containing cyclic peptide binding to Grb2 SH2 domain with high affinity Biochemical and Biophysical Research Communications 307 2003 1038 1044 147 Li P Zhang M Peach M L Liu H Yang D and Roller P P Concise and enantioselective synthesis of Fmoc Pmp Bu 2 OH and design of potent Pmp containing Grb2 SH2 domain antagonists Organic Letters 5 2003 3095 3098 148 Li P Peach M L Zhang M Liu H Yang D Nicklaus M and Roller P P Structure based design of thioether bridged cyclic phosphopeptides binding to Grb2 SH2 domain Bioorganic amp Medicinal Chemistry Letters 13 issue 5 2003 895 899 45 149 Long Y Q Guo R Luo J H Yang D and Roller P P Potentiating effect
141. esidue appears to be an interesting target for the binding of a negatively charged residue on the ligand This study indicates that one single doubly charged residue such as a phosphate group can bind to either Kio Lys BD and Rss Arg BB or Re7 Arg aA and Rss Arg BB Consequently the design of analogues with two distinct charged groups one targeting the two cavity 1 arginines and the other Kio Lys BD could lead to more potent compounds iii It was shown that in cavity 3 while the interaction with the R 2 Arg BG residue constitutes the main contribution of the interaction energy the binding with the neighboring residues S 4 Ser BG and N 43 Asn BG have to be strong enough to constitute a constraint on Ry42 Arg BG Such a constraint is required as R 42 Arg BG has a clear tendency to move towards the solvent disrupting the conformation of ligands targeting cavity 3 This information should be taken into account when modelling such ligands Conclusions As remarked by Ladbury 53 54 leaving water out of a drug design strategy reduces the likelihood that the strategy will be successful This work shows that solvent effects can impact significantly the binding of ligands to the SH2 domain of Grb2 while the force field choice does not Consequently not taking solvent into account with such a system could turn de novo design of active ligands to be unpredictive 55 This is also in line with the conclusions of previous studies beari
142. etech prot ines kinase en phase de test clinique 50 Au Elavil pan ene OSI Parmeet ee P 3 7 Serine threonine kinases niveau mol culaire ces recherches visent en PKC CR POGFR cae Phase 1 cancer retinapatiy Novartis R A P kgs p G PKC ISIS 3521 Phase Iil cancer ISIS Pharmaceuticals premier lieu a identifier les prot ines de la voie de ospet251 Phase 1 cancer parts r i A PES A 7 UCN O1 Phase Wl cancer Kyowa Hakko Kogyo signalisation vis e les plus impliqu es dans ate oe cot hry an ge r Phase III diabetic neuropathy l expression exag r e du signal biologique Un cos Fleopii Ps tam ents F CSi E7070 Phase cancer ESAI nombre important de liens de causalit entre a alae Bristol Myers Squibb diff rentes pathologies et une prot ine contenant uj UA adh ase Pc pa ee cas TE MLK CEP 1347 Phase II neurodegenerat Cephalon un domaine SH2 a ainsi t tabli 51 L inhibition RAF BAYa3 0006 Happee ia Onyx Pharmaceutical Bayer 2 1855132 Phase cancer tsis pharmaceuticals 1794 Phase Merck de la fonction de ce dernier la reconnaissance des m 1 779450 pan tea we s quences pr sentant un r sidu pTyr est alors Poa prose Sins 21 R115777 Phase MI Johnson amp Johnson susceptible de constituer un traitement m dical ns C0779 Prase 1 canen yee p RADOO Phase cancer Novartis Phase IN efficace et sp cifique 1 19 52 La d termination eames ene d inhibiteurs pour les diff
143. etween automated screening and manual modeling 95 96 More importantly as stated before docking limitations are mostly conceptual and protocols making wise use of the complementarities between different techniques could boost the efficiency of VS simulations considerably Perspectives for structure based drug design As purely empirical LBDD methods are correlated from the start to experimental data validation is part of the design On the opposite SBDD methods rely on molecular models whose design depend on existing concepts algorithmically constraints and assumptions at different levels the validation is all but obvious and can ultimately appear to invalidate design Considering this it is striking to notice that in the field of molecular docking the development of new methods is observed at a very fast pace while the validation of existing ones is still prone to improvements 61 Unlike LBDD closely bound to applied biochemistry with SBDD we are often on the verge of theoretical chemistry Taming SBDD is therefore difficult by nature despite some appearances to the contrary SBDD also needs to mature On the other hand it can be postulated that the great variety of SBDD proposals is a chance for who possesses enough expertise Constructing protocols using complementary docking techniques as a solution to the VS accuracy issues is an approach that should be extended to the other CADD techniques as well as to advances in data mining and c
144. ews in Oncology Hematology 48 2003 45 63 46 Sastry L Cao T and King C R Multiple Grb2 protein complexes in human cancer cells International Journal of Cancer 70 1997 208 213 47 Buolamwini J K Novel anticancer drug discovery Current Opinion in Chemical Biology 3 1999 500 509 48 Clark G J and Der C J Aberrant function of the Ras signal transduction pathway in human breast cancer Breast Cancer Research and Treatment 35 issue 1 1995 133 144 49 Garbay C Liu W Q Vidal M and Roques B P Inhibitors of Ras signal transduction as antitumor agents Biochemical Pharmacology 60 2000 1165 1169 50 Sebolt Leopold J S Development of anticancer drugs targeting the MAP kinase pathway Oncogene 19 issue 56 2000 6594 6599 51 Boldt S Weidle U H and Kolch W The role of MAPK pathways in the action of chemotherapeutic drugs Carcinogenesis 23 issue 11 2002 1831 1838 52 Kaelin Jr W G Choosing anticancer drug targets in the postgenomic era Journal of Clinical Investigation 104 issue 11 1999 1503 1506 53 Bange J Zwick E and Ullrich A Molecular targets for breast cancer therapy and prevention Nature Medicine 7 issue 5 2001 548 552 54 Watson J D and Crick F H C Genetic implications of the structure of deoxyribonucleic acid Nature 171 1953 964 967 55 Watson J D and Crick F H C Molecular structure of nucleic acids A structure for deoxyribose nucleic acid Nature 171 1953
145. f Molecular Biology 245 issue 1 1995 43 43 51 Jones G Willett P Glen R C Leach A R and Taylor R Development and validation of a genetic algorithm for flexible docking Journal of Molecular Biology 267 1997 727 748 52 Lala D S The liver X receptors Current Opinion in Investigational Drugs 6 issue 9 2005 934 943 53 Collins J L Therapeutic opportunities for liver X receptor modulators Current Opinion in Drug Discovery amp Development 7 issue 5 2004 692 702 54 F rneg rdh M Bonn T Sun S Ljunggren J Ahola H Wilhelmsson A Gustafsson J A and Carlquist M The three dimensional structure of the liver X receptor B reveals a flexible ligand binding pocket that can accommodate fundamentally different ligands Journal of Biological Chemistry 278 issue 40 2003 38821 38828 55 Berman H M Westbrook J Feng Z Gilliland G Bhat T N Weissig H Shindyalov I N and Bourne P E The Protein Data Bank Nucleic Acids Research 28 issue 1 2000 235 242 56 Williams S Bledsoe R K Collins J L Boggs S Lambert M H Miller A B Moore J McKee D D Moore L Nichols J Parks D Watson M Wisely B and Willson T M X ray crystal structure of the liver X receptor beta ligand binding domain regulation by a histidine tryptophan switch Journal of Biological Chemistry 278 issue 29 2003 27138 27143 57 Steiner T and Koellner G Coexistence of both histidine tautomers in the solid state a
146. f influenza virus replication Nature 363 1993 418 423 42 Schevitz R W Bach N J Carlson D G Chirgadze N Y Clawson D K Dillard R D Draheim S E Hartley L W Jones N D Mihelich E D Olkowski J L Snyder D W Sommers C and Wery J P Structure based design of the first potent and selective inhibitor of human non pancreatic secretory phospholipase A2 Nature Structural Biology 2 1995 458 465 43 Tondi D Slomczynska U Costi M P Watterson D M Ghelli S and Shoichet B K Structure based discovery and in parallel optimization of novel competitive inhibitors of thymidylate synthase Chemistry amp Biology 6 issue 5 1999 319 331 44 Filikov A V Mohan V Vickers T A Griffey R H P D C Abagyan R A and James T L Identification of ligands for RNA targets via structure based virtual screening HIV 1 TAR Journal of Computer Aided Molecular Design 14 issue 6 2000 593 610 45 Hopkins S C Vale R D and Kuntz I D Inhibitors of kinesin activity from structure based computer screening Biochemistry 39 issue 10 2000 2805 2814 46 Perola E Xu K Kollmeyer T M Kaufmann S H Prendergast F G and Pang Y P Successful virtual screening of a chemical database for farnesyltransferase inhibitor leads Journal of Medicinal Chemistry 43 issue 3 2000 401 408 47 Doman T N McGovern S L Witherbee B J Kasten T P Kurumbail R Stallings W C Conolly D T and Shoichet B K Molecular docking an
147. flexibilit des ligands dispose d un algorithme de recherche volutionnel performant et propose un rapport int ressant entre fiabilit et vitesse de traitement La prise en charge du solvant propos e ici se limite au positionnement de mol cules d eau interfaciales En aucun cas un tel mod le ne pourra pr dire des ph nom nes de d stabilisation du bulk tels que ceux observ s lors des dynamiques de ligand 2 avec Grb2 SH2 Ainsi il est pr voir dans le meilleur des cas que de telles situations donnent lieu au m me type de faux positifs que ceux qui seraient probablement obtenus avec une dynamique mol culaire in vacuo De plus le fait que l inclusion des mol cules d eau cristallographiques am liore la pr cision du docking peut para tre surprenant tant donn que celles ci n apparaissaient pas jouer un r le important dans les trajectoire des dynamiques On peut donc mettre l hypoth se que cela corresponde non pas une am lioration de la pertinence du mod le mais plut t une contrainte structurale limitant l espace de recherche En effet en positionnant ces mol cules d eau l tendue du site actif est restreinte ce qui peut grandement faciliter la t che du programme de docking Il faut alors garder l esprit que cela pourrait aussi artificiellement d favoriser le docking de mol cules poss dant un mode de liaison pr f rentiel diff rent de celui de ligand 1 Cette crainte peut toutefois
148. g factor hSos1 binds to Grb2 and links receptor tyrosine kinases to Ras signalling Nature 363 1993 85 88 41 Rozakis Adcock M Fernley R Wade J Pawson T and Bowtell D The SH2 and SH3 domains of mammalian Grb2 couple the EGF receptor to the Ras activator mSos1 Nature 363 1993 83 85 42 Rozakis Adcock M McGlade J Mbamalu G Pelicci G Daly R Li W Batzer A Thomas S Brugge J Pelicci P G Schlessinger J and Pawson T Association of the She and Grb2 Sem5 SH2 containing proteins is implicated in activation of the Ras pathway by tyrosine kinases Nature 360 1992 689 692 43 Skolnik E Y Lee C H Batzer A G Vicentini L M Zhou M Daly R J Myers Jr M J Backer J M Ullrich A White M F and Schlessinger J The SH2 SH3 domain containing protein Grb2 interacts with tyrosine phosphorylated IRS1 and Shc implications for insulin control of Ras signaling EMBO Journal 12 1993 1929 1936 44 Schlaepfer D D Hanks S K Hunter T and van der Geer P Integrin mediated signal transduction linked to Ras pathway by Grb2 binding to focal adhesion kinase Nature 372 1994 786 791 45 Atabey N Gao Y Yao Z J Breckenridge D Soon L Soriano J V Burke Jr T R and Bottaro D P Potent blockade of hepatocyte growth factor stimulated cell motility matrix invasion and branching morphogenesis by antagonists of Grb2 Src homology 2 domain interactions Journal of Biological Chemistry 276 issue 17 2001
149. g modes depicted in the three experimental structures have also been analyzed The shared characteristics are dominated by hydrophobic interactions with F271 F329 and F349 1PQ6 allows for a possible specific charge charge interaction with R319 R319 already makes an internal interaction with E gt in the 1P8D conformation dampening the strength of possible ligand interaction In the case of 1PQ9 neither of those residues is accessible as the pocket size is restricted by a particular F3 orientation Ligand database preparation The starting database is composed by compounds commercially available in March 2006 from three suppliers ChemDiv 58 Enamine 59 and Comgenex 60 Filtering using Lipinski s rule of five 61 was performed allowing a single violation for each structure giving a total of 598 375 unique molecules In order to reduce the database size while retaining as many chemical diversity as possible we used the ScreeningAssistant software 62 This tool characterizes each molecule of the database using SSKey 3D 54 bit fingerprints 63 allowing for similarity estimation between pairs by computing Tanimoto coefficients 64 Database clustering can then govern the generation of diversity maximized subsets In our case we targeted a 10 000 molecule subset and obtained a database of 8 383 compounds A reference diverse database was defined by merging the initial 598 375 molecules database with the Chimioth que Nationale CN data
150. gures 7 et 9 Propositions au niveau du design de nouveaux ligands Les interactions de l atome d oxyg ne non terminal du groupe phosphate des deux phosphotyrosine du ligand 1 s av rent non n gligeables ces deux atomes ne sont pas pr sents dans ligand 2 En cons quence bien que le choix de deux groupes phosphone pour ligand 2 la place des phosphates soit parfaitement justifi sur le plan pharmaceutique on peut mettre l hypoth se que cette diff rence structurale soit directement l origine aussi bien du mode de liaison alternatif dans cavit 1 que du changement conformationnel d favorable au niveau de cavit 3 De plus on a pu observer que les groupes en position 1 ne conf raient Pi 5 aucune activit sur le r cepteur Grb2 SH2 La substitution ou la suppression o de ces groupes est donc propos e Ces deux modifications s av rant priori CO simples chimiquement la synth se et le test de l activit biologique d un N OPO d riv de ligand 2 sans le groupe naph et pr sentant deux r sidus O phosphotyrosine non modifi s voir ci contre s av rerait sans doute plus probant que la conduite d une simulation de dynamique mol culaire suppl mentaire OPO Proposition de d riv pour ligand 2 Un phosphone tant beaucoup plus stable face aux r actions de d phosphorylation qui peuvent survenir dans l organisme qu un phosphate ligand 2 a un caract re drug like nettement plus i
151. he Grb2 SH2 domain Journal of Molecular Biology 353 2005 104 115 75 Berman H M Westbrook J Feng Z Gilliland G Bhat T N Weissig H Shindyalov I N and Bourne P E The Protein Data Bank Nucleic Acids Research 28 issue 1 2000 235 242 76 Smith Schmidt T Banking on structures BiolT World 1 issue 8 2002 77 Gilmer T Rodriguez M Jordan S Crosby R Alligood K Green M Kimery M Wagner C Kinder D Charifson P Hassell A M Willard D Luther M Runsak D Sternbach D D Mehrotra M Peel M Shampine L Davis R Robbins J Patel I R Kassel D Buckhart W Moyer M Bradshaw T and Berman J Peptide inhibitors of src SH3 SH2 phosphoprotein interactions Journal of Biological Chemistry 269 issue 50 1994 31711 31719 78 Furet P Gay B Caravatti G Garcia Echeverria C Rahuel J Schoepfer J and Fretz H Structure based design and synthesis of high affinity tripeptide ligands of the Grb2 SH2 domain Journal of Medicinal Chemistry 41 1998 3442 3449 79 Liu W Q Vidal M Gresh N Roques B P and Garbay C Small peptides containing phosphotyrosine and adjacent aMe phosphotyrosine or its mimetics as highly potent inhibitors of Grb2 SH2 domain Journal of Medicinal Chemistry 42 1999 3737 3741 80 Burke Jr T R Yao Z J Gao Y Wu J X Zhu X Luo J H Guo R and Yang D N Terminal carboxyl and tetrazole containing amides as adjuvants to Grb2 SH2 domain ligand binding Bioorga
152. he distributions of false positives i e populations located at the bottom right corners corresponding to molecules whose binding ranks are overestimated by SHEF according to GOLD results In agreement with previous observations it appears that SHEF generates most false positives when docking on the 1PQ9 conformation while correlation between GOLD and SHEF is best in the 1PQ6 case which presents a more specific shape that should favor SHEF efficiency Interestingly figure 7 j shows that the correlation between the SHEF and GOLD consensus rankings is higher than the average of the GOLD SHEF correlation for the three receptor conformations Additionally such an approach could be more interesting than the 1PQ6 only filtering of figure 7 h which naturally favors ligands more specific to 1PQ6 Even if the corresponding correlation is higher it is probably more important to favor diversity regarding target conformations when no precise information is known concerning their relative stability SHEF as a first step enrichment filter in the screening funnel protocol It should first be noted that the ligands present in the 1P8D 1PQ6 and 1PQ9 experimental structures redocked using GOLD fall into the range of the hits subset as defined previously These reference ligands are also amongst the top 2 structures according to SHEF calculations Therefore unless the filtering ratio is set too high they would be retrieved in a SHEF GOLD screening funnel expe
153. he most simple and quick filters are being used at an early stage in the filtering process allowing the more time consuming processes to be used in later stages The multiple step screening funnel strategy is shown in figure 3 Figure 3 Basic principle of the virtual screening funnel process gt molecular databases T number of molecules decreasing processing time per molecule increasing calculations precision increasing drug candidates Methodology Outline of the proof of concept study Most docking methods are not efficient enough for use in high throughput VS i e the time required to process gt 10 molecules is out of reach with modern hardware Fast filtering prior to docking might be a workaround Ligand based methods can also prove useful here but unless large training sets are available for the target they are of limited value Geometrical matching procedures which are orders of magnitude faster than common docking methods can be employed in this particular context 46 and can lead to discovery of hits 47 but few studies exist to estimate their impact in a general VS experiment The geometrical matching procedure evaluated here is a two step process First the MSSH program 35 36 approximates the geometry of molecular structures using a series of spherical harmonics functions This representation is very compact as all information is contained in the expansion coefficients while the cor
154. he regulatory domains of the Src family tyrosine kinase Lck Nature 368 1994 764 769 105 Tong L Warren T C King J Batageri R Rose J and Jakes S Crystal structures of the human p56lck SH2 domain in complex with two short phosphotyrosyl peptides at 1 0 A and 1 8 A resolution Journal of Molecular Biology 256 1996 601 610 106 Rahuel J Gay B Erdmann D Strauss A Garc a Echeverria C Furet P Caravatti G Fretz H Schoepfer J and Griitter M Structural basis for specificity of GRB2 SH2 revealed by a novel ligand binding mode Nature Structural Biology 3 issue 7 1996 586 589 107 Gay B Furet P Garcia Echeverria C Rahuel J Ch ne P Fretz H Schoepfer J and Caravatti G Dual specificity of Src homology 2 domains for phosphotyrosine peptide ligands Biochemistry 36 1997 5712 5718 108 Kimber M S Nachman J Cunningham A M Gish G D Pawson T and Pai E F Structural basis for specificity switching of the Src SH2 domain Molecular Cell 5 issue 6 2000 1043 1049 109 Garcia Echeverria C Antagonists of the Src Homology 2 SH2 domains of Grb2 Src Lck and ZAP 70 Current Medicinal Chemistry 8 2001 1589 1604 110 Furet P Gay B Garcia Echeverria C Rahuel J Fretz H Schoepfer J and Caravatti G Discovery of 3 aminobenzyloxycarbonyl as an N terminal group conferring high affinity to the minimal phosphopeptide sequence recognized by the Grb2 SH2 domain Journal of Medicinal C
155. hemistry 40 1997 3551 3556 43 111 Furet P Caravatti G Denholm A A Faessler A Fretz H Garcia Echeverria C Gay B Irving E Press N J Rahuel J Schoepfer J and Walker C V Structure based design and synthesis of phosphinate isosteres of phosphotyrosine for incorporation in Grb2 SH2 domain inhibitors Part 1 Bioorganic amp Medicinal Chemistry Letters 10 2000 2337 2341 112 Walker C V Caravatti G Denholm A A Egerton J Faessler A Furet P Garcia Echeverria C Gay B Irving E Jones K Lambert A Press N J and Woods J Structure based design and synthesis of phosphinate isosteres of phosphotyrosine for incorporation in Grb2 SH2 domain inhibitors Part 2 Bioorganic amp Medicinal Chemistry Letters 10 2000 2343 2346 113 Schoepfer J Gay B Caravatti G Garc a Echeverr a C Fretz H Rahuel J and Furet P Structure based design of peptidomimetic ligands of the Grb2 SH2 domain Bioorganic amp Medicinal Chemistry Letters 1998 2865 2870 114 Garc a Echeverr a C Gay B Rahuel J and Furet P Mapping the X 1 binding site of the Grb2 SH2 domain with o a disubstituted cyclic a amino acids Bioorganic amp Medicinal Chemistry Letters 9 1999 2915 2920 115 Schoepfer J Fretz H Gay B Furet P Garcia Echeverria C End N and Caravatti G Highly potent inhibitors of the Grb2 SH2 domain Bioorganic amp Medicinal Chemistry Letters 9 1999 221 226 116 Caravatti G
156. hen follow the arginine and eventually be bound to water as strongly as to Grb2 This would correspond to a significant destabilizing conformational change of the ligand coupled with a large drop of the binding energy Such a transition occurs during the first nanosecond of the ligand 2 complex MD and was not observed in the corresponding in vacuo MD simulation The fact that in ligand 2 the ligand 1 phosphates have been replaced by phosphones could explain why ligand 1 maintains its binding to Grb2 SH2 while ligand 2 does not We demonstrated that in both cases all oxygen atoms bound to the phosphorus atom are making interactions of the same order with Grb2 SH2 and that as a whole the phosphate and phosphone groups are making interactions of similar strength when initially positioned with the same conformations regarding cavity 1 and cavity 3 In this regard the only difference between those two groups relies in the geometrical distribution of the interactions in three directions with the phosphone in four directions with the phosphate While this difference did not appear to lead to major consequences in in vacuo MD simulations it seems to be crucial when solvent is modelled With both cavity 1 and cavity 3 it appears that the four directions conformation of a phosphate group leads to a more stable binding than the three directions conformation obtained with ligand 2 phosphones In this regard the non terminal oxygen of ligand 1 phosphate groups co
157. hnan H Chord a scalable peer to peer lookup service for internet applications in Proceedings of the 2001 conference on Applications technologies architectures and protocols for computer communications 2001 105 Anderson D P Cobb J Korpela E Lebofsky M and Werthimer D SETI home An experiment in public resource computing Communications of the ACM 45 issue 11 2002 56 61 106 Chen S Zhang W Ma F and Shen J A cooperative computing platform for drug discovery and design in Proceedings of the 2004 IEEE international conference on services computing 2004 107 Montgomery S B Fu T Guan J Lin K and Jones S J An application of peer to peer technology to the discovery use and assessment of bioinformatics programs Nature Methods 2 issue 8 2005 563 108 Cappello F Caron E Dayde M Desprez F Jegou Y Primet P Jeannot E Lanteri S Leduc J Melab N Mornet G Namyst R Quetier B and Richard O Grid 5000 a large scale and highly reconfigurable grid experimental testbed in The 6th IEEE ACM International Workshop on Grid Computing 2005 109 Coveney P V Scientific grid computing Philosophical Transactions of the Royal Society of London Series A Mathematical Physical and Engineering Sciences 363 issue 1833 2005 1707 1713 110 Woods C J Ng M H Johnston S Murdock S E Wu B Tai K Fangohr H Jeffreys P Cox S Frey J G Sansom M S P and Essex J W Grid computing an
158. hosphorylation gr ce aux kinases de d phosphorylation via les phosphatases souvent gr ce un quilibre entre ces deux processus antagonistes 46 En effet le changement de l tat de phosphorylation de certaines prot ines au niveau de r sidus tyrosine s rine et thr onine principalement constitue alors un interrupteur pour le signal propag par la voie de signalisation Celle ci poss de dans de nombreux cas une tape qui implique la reconnaissance mol culaire d un r sidu pTyr t che pour laquelle sont sp cialis s les domaines SH2 26 Certaines prot ines extra cellulaires ont surtout un r le m canique anatomique Ainsi la collag ne dont la structure est une triple h lice extr mement stable forme par empilement les tendons fibres musculaires particuli rement r sistantes la tension L lastine l inverse est une constituante essentielle des tissus fournissant l lasticit indispensable au fonctionnement art riel De nombreuses prot ines sont des enzymes leur r le est de catalyser des r actions biochimiques Des prot ines peuvent galement avoir une sp cialisation biologique par exemple en constituant le syst me immunitaire anticorps ou bien en transcrivant l information g n tique ribosome On qualifie de transporteurs les prot ines dont le r le est la diffusion dans l organisme de substrats telle l h moglobine qui transporte l oxyg ne dans le sang De nombreuses autres
159. idique 116 mais il semble que des optimisations sur cette nouvelle base ne furent pas tent es Extension du r cepteur cibl INSERM CNRS L approche d optimisation initi e par Novartis consistant substituer les diff rents groupes d une s quence peptidique de r f rence pYXN dans le cas de Grb2 est reprise par la plupart des quipes cherchant a mettre au point des inhibiteurs de domaines SH2 52 53 83 86 109 117 119 Une de ces quipes a permis gr ce l une de ces substitutions de mettre en vidence qu il tait possible pour un ligand d atteindre des zones du r cepteur Grb2 SH2 non cibl es par les ligands connus jusqu a pr sent tendant ainsi la taille du r cepteur Apr s avoir mis au point un ligand inhibant simultan ment l activit des deux domaines SH3 de Grb2 120 l quipe dirig e par Christiane Garbay s int ressa l inhibition du domaine SH2 de Grb2 autre cible possible pour bloquer la voie de signalisation Ras MAPK Pour ce faire ils prirent comme base un des pseudo peptides d velopp s par Novartis mAz pTyr Acec Asn NH p 88 dont les substituants aux positions 1 et 1 sont optimis s par rapport la r f rence Ac pYIN NHb et tent rent de d terminer un substituant encore plus efficace que Ac c 1 79 121 D autres travaux conduits Novartis avaient indiqu qu une s quence pYopYiiNi2 s av rait 25 fois plus active sur le domaine SH2 de Src que la s q
160. iert qu une structure satisfaisante Dans un contexte d optimisation et non de screening l impr cision possible d un score de docking n est pas handicapante On peut donc envisager m me si la validation d crite pr c demment n est pas encore men e bien de tester par dynamique mol culaire les structures les plus int ressantes pr sent es la page pr c dente Cela pr sente le risque de gaspiller une quantit importante de temps de calcul sur des faux n gatifs mais une telle situation n est pas inhabituelle dans le cadre g n ral du drug design On se r f re ici aux publications suivantes Songyang et al Cell 72 1993 767 778 Songyang et al Mol Cell Biol 14 1994 2777 2785 Gram et al Eur J Biochem 246 1997 633 637 Kessels et al PNAS 99 2002 8524 8529 Vetter et al Curr Prot Pept Sci 3 2002 365 397 Si la conservation de ces mol cules d eau semble tre utile pour le docking sur Grb2 SH2 nous n tions pas en mesure d en d duire si ce serait ou non le cas pour les trois autres r cepteurs SH2 55 Screening virtuel mise en place d outils innovants Principales probl matiques li es au screening virtuel Cette partie se place dans le contexte d taill par l article 3 Should structure based virtual screening techniques be used more extensively in modern drug discovery G n ralement lorsque le screening virtuel a pour simple vis e de servir de compl
161. if SBDD VS techniques should be used more extensively the answer is yes but we have to specify that the focus should be on the association with the other techniques of modern drug discovery Resulting views should be more and more linked with the advances in understanding the mechanisms of life from the genome to the interactome This open approach is promising while on the exclusive side inventing new docking techniques could be necessary for improving the tools but as the choice there is somewhat cluttered and presently limited to hit discovery this should not be overemphasized Figure 1 flowchart of the computer aided drug discovery process Y i Chemical space CE 10 database chemical bank management virtual combinatorial chemistry pharmacokinetics L y sai Candidates database enrichment Ss Ss i knowledge experimental 108 ligand based e oO y knowledge based E 103 optimization selectivity assessment CM experimental validation Leads 10 SS Cm og Boge YA Legend number of compounds costs time and money knowledge variations when chemoinformatics techniques are used wisely SUBIS014d SUTDOp p sn uruw dy JO SanaAIdOAd T AQEL pue31 y pue goejins urajord y TOZTUNTUTUL urajo1d au yoq i Sok i uoe aow 0 p9910 puey
162. igand 1 et ligand 2 Le mode de liaison r solu exp rimentalement pour le complexe de ligand 1 est reproduit correctement Toutefois il tait galement pr vu que ligand 2 partage ce m me mode de liaison avec une nergie d interaction plus importante on aurait donc pu pr voir que ligand 2 poss de une affinit plus importante pour Grb2 SH2 que ligand 1 Nous observons ici une contradiction vidente entre r sultats exp rimentaux et pr dictions th oriques L objectif de cette th se tant l tude th orique du r cepteur Grb2 SH2 il s av ra n cessaire avant toute chose de fournir une explication coh rente cette situation Bien entendu les protocoles exp rimentaux n tant pas sujets caution nous avons commenc par v rifier et reproduire les simulations effectu es aucune erreur dans la mod lisation n a t d tect e On pouvait alors supposer que le syst me tait d crit de fa on insuffisante un param tre important n tant pas pris en compte Nous avons d cid de tester simultan ment deux hypoth ses d une part le choix du champ de force d autre part les effets de solvant Mise en uvre Syst mes mod lis s De la structure exp rimentale du complexe Grb2 SH2 ligand 1 1JYQ on conserve un monom re B le ligand 1 li ainsi que les mol cules d eau situ es moins de 5 de la prot ine ef du ligand Le syst me du complexe avec ligand 2 est mod lis sur la base du pr
163. igand docking using GOLD J Med Chem 2005 45 6504 6515 53 Ladbury J E Just add water The effect of water on the specificity of protein ligand binding sites and its potential application to drug design Chem Biol 1996 3 973 980 54 Morton C J Ladbury J E Water mediated protein DNA interactions The relationship of thermodynamics to structural detail Protein Sci 1996 5 2115 2118 55 Mancera R L De novo ligand design with explicit water molecules an application to bacterial neuraminidase J Comput Aided Mol Des 2002 16 479 499 56 Tame J R H Sleigh S H Wilkinson A J Ladbury J E The role of water in sequence independent ligand binding by an_ oligopeptide transporter protein Nat Struct Biol 1996 3 998 1001 57 Rostom A A Tame J R H Ladbury J E Robinson C V Specificity and interactions of the protein OppA Partitioning solvent binding effect using mass spectrometry J Mol Biol 2000 296 269 279 58 Jones G Willett P Glen R C Molecular recognition of receptor sites using a genetic algorithm with a description of desolvation J Mol Biol 1995 245 43 53 59 Jones G Willett P Glen R C Leach AR Taylor R Development and validation of a genetic algorithm for flexible docking J Mol Biol 1997 267 727 748 60 Antony J Piquemal J P Gresh N Complexes of thiomandelate and captopril mercaptocarboxylate inhibitors to metallo B lactamase
164. ilter In the literature database filtering against a given target is often referred to as enrichment 11 12 The second purpose is the identification of a small number of candidates likely to be potent by ranking input compounds In all VS filters there is a trade off between speed and accuracy filters are optimized for speed The fastest filters can handle up to a few million molecules but are notoriously imprecise in reducing this number to less than a thousand while retaining all potential hits More costly techniques which can be used in lead optimization strategies can tackle this problem 13 14 but not with several million molecules as input and sensible computation times 15 Therefore VS protocols are often based on a single or a few fast filters and used prior to experimental screening However in that case VS usage is limited to that of a pre filter for HTS reducing the number of compounds to be tested experimentally and hence the cost of experiments by at least one order of magnitude 16 17 We have devised a platform for virtual screening called VSM G Virtual Screening Manager for computational Grids Our objective with VSM G is to provide a user friendly tool that would give scientists a large range of in silico strategies for finding hits Two kinds of approaches can be employed here ligand based and structure based 18 19 At present VSM G uses structure based methods to rank input compounds according to their affinit
165. inaison de plusieurs types de mutations provoque le cancer on estime qu un des pr alables la survenue du cancer est une instabilit du g nome sur un intervalle de temps important Il a t d montr que certains facteurs ext rieurs pr cis peuvent en tre la cause On peut citer l exposition r p t e des substances dites canc rig nes 23 24 qui peuvent tre des drogues alcool tabac des produits chimiques solvants pesticides ou bien des produits naturels amiante mat riaux radioactifs De nombreux canc rig nes ont t et sont toujours pour certains d entre eux massivement employ s dans l industrie arsenic amiante nickel goudrons si bien que la part des expositions professionnelles parmi les causes du cancer n est certainement pas n gligeable Certains canc rig nes sont caract ristiques d un type pr cis de cancer par exemple le tabac dont la consommation est 20 Year Lag Time Between Smoking and Lung Cancer H 414 A i Cigarettes Lung parfaitement corr l e d mographiquement au nombre de cas sen Cones de cancer du poumon La combinaison de plusieurs Pe ersen Deaths Per Year Per substances canc rig nes augmente encore les risques en pres Pose particulier l association alcool tabac est connue pour son gna 150 incidence tr s importante sur les cancers de la bouche de la gorge et de la vessie On estime plus g n ralement qu une i mauvaise hygi
166. ing energies They are however very useful for identifying the most relevant ligand conformational changes that occur during the dynamics NAMD has a feature described in the manual as pair interactions that changes the energy output to just a part of it corresponding to the interactions between two user defined atom groups It is possible to extract such values multiple times with different group definitions by reprocessing the trajectory file in a series of forced runs at O K this procedure is described in detail in the NAMD manual Using this method we computed the interactions between various groups Grb2 SH2 the ligands residues of the Grb2 SH2 receptor specific parts of the ligands water molecules The calculated energy values solely represent the expression of the interaction potential between specific non bonded groups of atoms this is just the sum of the van der Waals and Coulomb interactions Entropy is not evaluated at all Such values which we designate as interaction energy or binding energy are not the most representative of the interaction between the protein and the ligand the free energy of binding is However our goal is not to compute binding energy values precisely but rather to gather data that is able to clearly point out the differences between ligand 1 and ligand 2 binding In order to obtain an intuitive representation of that data programs were made in order to build automatically the files describing the at
167. ining a stacking interaction with R Arg aA as expected from in vacuo simulations Compared to the complex of ligand 1 conformational changes of both ligand 2 and Grb2 SH2 occurred so that several side chains of the protein binding site initially interacting with ligand 2 have formed intra molecular H bonds or H bonds with waters instead of the original ones In cavity 3 this is especially notable for the S 4 Ser BG residue which only interacts significantly with the phosphone group during the first 600 ps Conversely the phosphone and the R 4 Arg BG guanidinium moved altogether towards the bulk conserving their salt bridge interaction At t 1 ns the phosphone appears only bound to this residue as shown on figure 6 In cavity 1 Sos Ser BC hydroxyl group no longer interacts with the phosphone after 200 ps and the one of Soo Ser BC moves rapidly between 4 6 A from the phosphorous atom In the same cavity Re7 Arg aA guanidinium fluctuates several times between 4 8 A from the closest phosphoney oxygen By contrast Rgs Arg BB and Ssg Ser BB keeps interacting with the phosphones The weak interaction with R Arg aA compared to ligand 1 binding is compensated by the formation of a stable salt bridge involving K Lys BD outside cavity 1 which is lost abruptly after 1800 ps in favor of bulk hydration as shown at t 2 ns on figure 6 Figure 5 Cluster graphs of the a ligand 1 top and b 2 ne ee RMSD A
168. ion PC T Handling z MD Receptor sampling definition Current status The key features of VSM G are as follows 1 Wide coverage of the VS process from ligand and target preparation to the screening setup the monitoring of the calculation processes and finally the results analysis Unified and user friendly graphical interface see figure 2 Seamless integration of the modules e g intercommunication procedures such as file format conversions are automatic and transparent to the user Easy maintenance of the code with modular design and choice of widely used programming languages Java C C and Fortran Access to grid technology to take advantage of distributed computing involving computer and cluster grids VSM G relies on third party software for performing specific tasks or in order to provide several choices of techniques for a given purpose Due to its modular design VSM G is readily useable even if those external programs are not installed on the host computer One of the main development goals of VSM G is to provide at least one free open source solution for each task which is not currently the case e g at the moment GOLD is the only choice for performing flexible docking Figure 2 Some screenshots of the VSM G graphical interface A YSM G Job submission Wend arectory D OntabasesUOtconts Tasks totat 100 Furring 19 Mew 38 done 45 taled canceled Passe choose a Me format O POG MOL Sne
169. ions and applicability of such an approach are discussed Possible improvements of both the geometrical matching technique and its implementation within VSM G are suggested Keywords multiple step virtual screening VSM G structure based drug design geometrical matching spherical harmonics surfaces SHEF GOLD molecular database enrichment Introduction The search for new drugs is time consuming and expensive 1 any method that speeds up the process is beneficial Recently virtual screening VS techniques 2 have gained much interest in many drug development strategies 3 VS has two obvious advantages the speed with which one can screen a large library of compounds and the small initial capital investment compared to the cost of an in vitro high throughput screening HTS program The first aim of HTS and VS is to reduce a molecular database to few hit compounds for a protein target VS is considered successful when combined or not with HTS it leads to confirmed hits for a cost lower to that of HTS alone Research in this area is particularly active and several success stories have been reported 4 7 Thus it is now widely accepted that VS calculations can complement HTS experiments 8 9 VS methods can have two distinct purposes The first one is the exclusion of a large number of compounds which have little or no activity leading to a limited set of molecules which are more probable hits 10 such a method is referred to as a f
170. ions of the residue in 1 position is approximately constant during the whole simulation Additionally the interaction of the group in position 0 is abruptly lowered in magnitude at 1800 ps as expected from the observation of the loss of the phosphone binding with the Kioo Lys BD residue Figure 9 Interaction energy of the phosphate ligand 1 phosphone ligand 2 group of the ligands in dots ligand 2 pY l water ligand 2 pY l ligand 1 pY 3 ligand 2 i pYo j li 1 po 2000 Evolution of the protein residues binding energies We also followed the variation of the interaction energy of all Grb2 SH2 residues with both ligands through each trajectory This has permitted to establish the list of the Grb2 SH2 residues involved in ligand binding As expected residues R 7 Arg aA Rgs Arg BB Seg Ser BB Soo Ser BC and Sos Ser BC in cavity 1 Kio Lys BD in cavity 2 and S44 Ser BG Rix Arg BG and Nig Asn BG in cavity 3 have favorable ligand binding energies The binding energy of residues Li29 Leu BE and W121 Trp EF in cavity 2 is very low In contradiction to what could be expected from the X ray structure of the ligand 1 complex the H107 His BD residue does not interact significantly with either ligand 1 or ligand 2 Additionally Ks4 Lys AA makes minor interactions with both ligands We will now focus on the data regarding the aforementioned interacting residues The correspondi
171. ire 13 L tude sur le plan mol culaire du fonctionnement d un organisme vivant peut ainsi s effectuer diff rents niveaux conceptuels successifs Le g nome repose sur l espace des s quences de nucl otides le prot ome y ajoute l espace g om trique des prot ines correspondantes et l interactome la liste et la nature des interactions possibles qui en d coulent Le g nome humain est pr sent d crypt et concernant les prot ines le nombre de structures exp rimentales r solues et accessibles 67 68 croit exponentiellement La cartographie de l interactome n en est quant elle qu ses balbutiements t Dans le cadre de ce travail nous tudions une cible prot ique pr cise et en particulier l inhibition de son activit au sein d une voie de signalisation correctement caract ris e Il ne nous appara t pas important dans ces conditions de d tailler l tat actuel des recherches sur l interactome qui peuvent inclure des aspects autres que purement g om triques et chimiques tels les effets cin tiques Au 23 ao t 2005 la base PDB Protein Data Bank recence 29532 structures prot iques La portion de ces structures qui correspondent des complexes et qui sont donc d un grand int r t afin de relier prot ome et interactome au niveau de l espace g om trique et non plus seulement au niveau de l espace des interactions n est toutefois que de 1316 4 4 Si l interactome es
172. irements for designing new potent water friendly ligands The synthesis and activity measurement of analogues of ligand 1 ligand 2 and a new class of Grb2 SH2 ligands 63 if possible will therefore be performed shortly Acknowledgements The MD computations were performed on the computers of the Centre d Informatique de l Enseignement Sup rieur CINES Montpellier France This work was supported by the ACI Mol cules et cibles th rapeutiques Molecules and medicinal targets of the French Ministry of Research The work of Vincent Leroux was funded by an MRT fellowship related to the ACI References 1 Sadowski I Stone J C Pawson T A noncatalytic domain conserved among cytoplasmic protein tyrosine kinases modifies the kinase function and transforming activity of Fujinami sarcoma virus P130gag fps Mol Cell Biol 1986 6 4396 4408 2 Pawson T Gish G D SH2 and SH3 domains from structure to function Cell 1992 71 359 362 3 Moran M F Koch C A Anderson D Ellis C England L Martin G S Pawson T Src homology region 2 domains direct protein protein interactions in signal transduction Proc Natl Acad Sci USA 1990 87 8622 8626 4 Smithgall T E SH2 and SH3 domains potential targets for anti cancer drug design J Pharmacol Toxicol Methods 1995 34 125 132 5 Garbay C Liu W Q Vidal M Roques B P Inhibitors of Ras signal transduction as anti tumor agents Biochem
173. ires Il s av ra que Discover n est pas capable de g rer un fichier en entr e contenant plus de 10000 r sidus Or chaque mol cule d eau est consid r e comme un r sidu il fut impossible de mod liser des bo tes d eau de 80 A dans ce cas 48 Techniques d analyse Outre l analyse visuelle avec InsightII ou VMD et les v rifications usuelles taille de bo te structure de la chaine prot ique volution des distances intermol culaires int ressantes deux types de mesures permettant une analyse graphique ais e ont t effectu es sur les trajectoires Les graphes de RMSD obtenus avec InsightII permettent de rep rer les changements conformationnels les plus significatifs Les barres d nergies obtenues pour les trajectoires calcul es par NAMD permettent une d composition des interactions du syst me ainsi que la surveillance de leur volution au cours du temps Cette technique qui utilise des programmes et scripts cr es sp cifiquement s av re tr s utile pour tous types de trajectoires et a t appliqu e avec succ s au sein du laboratoire afin de d finir ais ment une carte des interactions pour des simulations de type prot ine prot ine voir l exemple ci dessous Smad3 MH2 domain 1280 1290 E 1300 2 1310 E 1320 250 E 1330 40 1340 30 1350 20 aa 10 0 4 6 time ns Choix du champ de force L analyse compar e des trajectoires des deux complexes avec les tr
174. iscovery computer aided lead discovery techniques add more cost to the hit to lead phase As the admission charge for candidates to clinical trials is so high rushing through experiments is absolutely out of the question at this point Moreover no in silico technique can at present handle realistically the degree of specificity of a given interaction even if there are early attempts in that direction 94 Consequently experimental approaches of classic biochemistry will continue to dominate the lead discovery phase This does not mean that SBDD is a waste of time in the hit to lead phase as a great deal of meaningful knowledge sometimes unreachable experimentally e g local structural dynamics cannot be observed without resorting to molecular dynamics can be potentially gained When such data is reinserted it is likely to enrich the whole drug discovery process Regarding these points Figure 1 highlights how CADD could be of use in modern drug discovery protocols To conclude this part we will say that concerns about docking and SBDD being inadequately integrated in drug discovery could appear linked to misunderstandings regarding the abilities of the different methods In SBDD the hit discovery screening and lead discovery predicting techniques are clearly distinguished while in LBDD this is more likely only a matter of size and pertinence of data It should be noted that some recent developments may start to fill the methodological gap b
175. it partir de ligands actifs in vitro soit en tentant de mettre au point des bases structurales nouvelles en int grant d s le d but les contraintes sp cifiques de stabilit et d accessibilit Modification de ligands existants Une modification int ressante de la phosphotyrosine a t propos e par Gay ef al afin de permettre la p n tration cellulaire d un ligand CGP78850 dont nous avons parl pr c demment comprenant d j un analogue de phosphotyrosine plus r sistant aux phosphatases Afin de permettre la p n tration de la membrane cellulaire ce groupe est est rifi ce qui donne le compos CGP85793 Celui ci apr s p n tration cellulaire sera hydrolys par les est rases intra cellulaires 139 relachant CGP78850 qui peut alors inhiber Grb2 SH2 60 64 D une fa on g n rale une d marche couramment entreprise afin d augmenter l accessibilit in vivo des ligands a base de phosphotyrosine consiste 4 employer un substituant au groupe pTyr afin de contourner les probl mes d accessibilit et de stabilit qui lui sont li s 77 124 125 140 voir tableau page suivante Celui ci peut tre un autre groupe charg a base de phosphore en particulier Pmp et ses d riv s un groupe acide souvent a base carboxyl 126 141 ou bien un groupe neutre lectriquement typiquement Tyr A noter que sur aucun ligand connu de Grb2 SH2 tout comme pour Src SH2 77 la substitution du groupe pT
176. la diff rentiation des cellules 7 8 Elle semble galement tre impliqu e dans des processus biologiques vari s tels que le fonctionnement de la m moire long terme 9 Elle suscite un grand int r t scientifique car les mutations du g ne Ras se signalent parmi les mutations les plus fr quemment observ es dans les cas de cancers humains 6 10 11 De plus certaines de ces mutations sont en particulier reli es l activit de Ras MAPK ce qui fait d elle une cible th rapeutique 12 13 Au sein de Ras MAPK on trouve deux formes complex es de Ras Ras GDP et Ras GTP De fa on g n rale la prot ine Ras a besoin d tre modifi e souvent par complexation pour acqu rir une activit biologique Dans Ras MAPK seule la forme Ras GTP est active le passage de Ras GDP Ras GTP provoque un changement conformationnel qui expose un site actif de Ras 14 pour Raf 15 16 ce qui permet de poursuivre la cascade d interaction de Ras MAPK Cette transformation est favoris e entre autres par l intervention de complexes de la prot ine adaptatrice Grb2 17 18 Les domaines SH2 On d signe par domaine une famille de s quences d acides amin s pr sente sur un nombre important de prot ines et dont les membres pr sentent un tr s fort taux d homologie Les domaines correspondent des codes g n tiques g n ralement fortement conserv s au cours de l volution Les domaines SH2 Src homology 2 19 sont
177. la dynamique mol culaire nous ne sommes pas en mesure de soumettre des propositions pr cises afin d am liorer le protocole de screening virtuel qui a t mis en place Toutefois une analyse des diff rentes techniques de docking disponibles donne un aper u des aspects th oriques et techniques dont l am lioration pourrait tre la plus profitable Ce travail peut bien s r tre soumis plusieurs critiques Si on peut estimer qu il repr sente un certain avancement dans la compr hension du m canisme d inhibition du r cepteur de Grb2 SH2 tout en contribuant l am lioration des approches utiles une telle tude il est aussi clair que de nombreuses possibilit s restent ouvertes En ce qui concerne la dynamique mol culaire si le r le des mol cules d eau constituant l environnement biologique du syst me a t mis en vidence pour un ligand pr cis avec Grb2 SH2 peut on proc der une quelconque g n ralisation La r ponse est n gative en l absence de donn es ult rieures issues de simulations suppl mentaires ou bien d exp rimentations Nous nous contentons d indiquer qu en absence de connaissances pr cises pr alables sur les effets de solvant possibles d un syst me prot ine ligand il convient par prudence de mod liser le solvant dans la limite de ce que les m thodes employ es permettent Le protocole de docking flexible n est galement pas valid l heure actuelle Da
178. lar dynamics free energy and SPR analyses of the interactions between the SH2 domain of Grb2 and ErbB phosphotyrosyl peptides Biochemistry 42 2003 5195 5200 99 Kessels H W H G Ward A C and Schumacher T N M Specificity and affinity motifs for Grb2 SH2 ligand interactions Proceedings of the National Academy of Sciences USA 99 issue 13 2002 8524 8529 100 Broadshaw J M Grucza R A Ladbury J E and Waksman G Probing the two pronged plug two holed socket model for the mechanism of binding of the Src SH2 domain to phosphotyrosyl peptides A thermodynamic study Biochemistry 37 1998 9083 9090 101 Booker G W Breeze A L Downing A K Panayotou G Gout I Waterfield M D and Campbell I D Structure of an SH2 domain of the p85a subunit of phosphatidylinositol 3 OH kinase Nature 358 1992 684 687 102 Overduin M Mayer B Rios C B Baltimore D and Cowburn D Secondary structure of Src homology 2 domain of c Abl by heteronuclear NMR spectroscopy in solution Proceedings of the National Academy of Sciences USA 89 issue 24 1992 11673 11677 103 Waksman G Kominos D Robertson S C Pant N Baltimore D Birge R B Cowburn D Hanafusa H Mayer B J Overduin M and al e Crystal structure of the phosphotyrosine recognition domain SH2 of v src complexed with tyrosine phosphorylated peptides Nature 358 1992 625 626 104 Eck M J Atwell S K Shoelson S E and Harrison S C Structure of t
179. larit des domaines SH2 r side dans une tr s haute affinit pour des petites s quences prot iques d butant par un r sidu tyrosine phosphoryl pTyr 24 25 Chaque famille de domaines SH2 reconna t pr f rentiellement des s quences plus sp cifiques selon la nature des r sidus proximit imm diate de la phosphotyrosine 24 Les domaines SH2 en g n ral se lient aux r sidus pTyr pr sents sur les r cepteurs tyrosine kinase 26 et r gulent ainsi leur activit 27 ce qui les relie en particulier diff rentes cascades de r actions dont Ras MAPK contr lant la division cellulaire 28 29 Les domaines SH2 sont ainsi des cibles int ressantes pour la mise au point de r gulateurs de la division cellulaire classe de mol cules potentiellement anticanc reuses De plus la surexpression de prot ines contenant des domaines SH2 impliquant une d r gulation des voies de signalisation correspondantes caract rise souvent le d veloppement de tumeurs canc reuses ce qui fait des domaines SH2 tout comme Ras MAPK des cibles th rapeutiques de choix dans la recherche pharmaceutique contre le cancer 1 12 13 30 Adaptors sH2 SH3 SH3 Crk Crkl SH3 SH2 SH3 Grb2 Grap SAM sh Sip 76 Scaffolds me s She sv Q Dapp1 Ms SH3 SH22 Tyrkinase Src family a i sR suzen MS Fps A Oro a Phosphatases shz s ae SHP 1 SHP 2 laa ona Bo sh STAT
180. les fonctions biologiques sont diff rentes Dans pareil cas l inhibiteur se doit d tre sp cifique afin d tre suffisamment actif sur la cible sans avoir recourir des dosages importants qui pourraient engendrer des effets secondaires ind sirables Dans le cadre de la recherche de ligands actifs in vivo la sp cificit est ainsi une contrainte qui s impose aussi bien la cible qu l inhibiteur La sp cificit d un r cepteur peut tre estim e par des recherches bibliographiques et des techniques de base de la prot omique tels que les logiciels d alignement de s quences A ce niveau les sp cificit s structurales du domaine SH2 de Grb2 contribuent grandement son int r t en tant que cible th rapeutique au sein de la voie de signalisation Ras MAPK Estimer la sp cificit d un inhibiteur quelconque pour une cible donn e est par contre une t che bien plus d licate il s agit d une des raisons expliquant le taux d chec lev des protocoles de recherche de m dicaments Pour garantir qu un ligand soit sp cifique il faudrait en effet le tester sur l ensemble des biomol cules avec lesquelles il est susceptible d interagir au sein d un organisme ce qui ne peut se faire qu exp rimentalement en milieu clinique Au niveau th orique non seulement l ensemble du prot ome n est pas connu mais en plus l espace connu est d j trop vaste pour envisager son exploration syst matique On
181. les simples que doit respecter une mol cule quelconque pour pouvoir tre administr e l homme la mani re d un m dicament Il est vident que de telles r gles doivent tre consid r es durant la recherche d inhibiteurs pour une cible biologique donn e L identification de mol cules non drug like efficaces in vitro peut toutefois s av rer utile En effet il peut tre possible partant de ces mol cules inutiles comme r f rence d effectuer des modifications afin de leur conf rer un caract re drug like sans pour autant restreindre de fa on significative leur efficacit Le respect de la contrainte d accessibilit pour des inhibiteurs de Grb2 SH2 se heurte une difficult de taille Grb2 est une prot ine qui agit en milieu intra cellulaire au sein de la voie de signalisation Ras MAPK Une mol cule destin e inhiber cette activit devra donc passer la barri re biologique que constitue la membrane cellulaire Or tous les inhibiteurs connus de Grb2 SH2 y compris ceux pour lesquels le groupe pTyr a t substitu sont hautement charg s n gativement ce qui leur interdit a priori le franchissement de la membrane La mise au point d inhibiteurs neutres ou faiblement charg s conservant une activit suffisante partir des ligands connus de Grb2 SH2 constitue une t che encore plus difficile que celle de la simple substitution de la phosphotyrosine Pourtant 1l a t montr que le r cepteur
182. ls of pharmaceutical research is being debated 52 55 Regarding SBDD even if VS techniques are now common in the pharma world a patented drug entirely designed in silico has not been reported yet 3 VS is also well known for its unpredictability which is annoying for such an approach that at first sight seems to provide an escape route from the empirical practices in medicinal chemistry 39 This also contradicts the dream of a modelling approach that could totally replace experiments that was foreseen in the 1970s as SBDD was coined 56 Docking programs in particular still boost poor accuracy as redocking validation simulations performed on a limited number of protein ligand structures extracted from the PDB 5 57 show that the current state of the art solutions only provide success rates of 60 70 58 60 This clearly indicates that docking methods regardless the progress that is made constantly in this area still need to be improved Before attempting to improve docking methods one has to explain why current more and more sophisticated approaches still do not perform as well as one can expect Solving this problem is a prerequisite for answering the question raised here should we use VS through SBDD methods more extensively Recently Shoichet postulated that there exist three main problems that need to be assessed 39 1 the possible chemical space is so wide that most probably we are not able to explore all of its p
183. lus appropri e soit s lectionn e La mise en place d un protocole de screening virtuel fiable et suffisamment efficace pour g rer des larges bases de mol cules gt 10 et pouvant tre employ de fa on routini re au del d un simple accompagnement des techniques exp rimentales est un d fi qu il est tentant de relever si l on consid re l avantage que peut conf rer les progr s technologiques et conceptuels de l informatique Or aucune technique de screening virtuel ne r unit seule les crit res de fiabilit et de rapidit de traitement on doit donc envisager l utilisation combin e de plusieurs techniques Nous avons vu qu une m thode de docking est d autant plus approch e que le nombre de dimensions du syst me est faible Dans ces conditions les m thodes les plus rapides ne sont pas aptes tablir un classement pr cis des mol cules test es mais sont priori capables d en exclure une partie suffisante En cons quence DRUG CANDIDATE l utilisation d une s quence de techniques de la plus rapide la plus pr cise chacune constituant non pas un valuateur mais un filtre mol culaire est envisageable On obtient ce qu on peut assimiler un entonnoir de screening voir sch ma de principe ci contre chaque tape de l entonnoir un nombre significatif de mol cules est limin du screening virtuel Les mol cules exclues en premier sont celles dont on peut mesurer le man
184. ly prior to docking Some programs can construct more or less efficiently the active site if they are given a single starting coordinate while others can try to find it when it is not known In some cases the lack of information can simply prevent docking The nature of the user interface and the possibility of launching docking jobs in parallel make programs more or less convenient to use but in the case of vHTS the absence of parallel execution can render calculation execution and management particularly painful Lastly the legal status is in our opinion quite important Not being able to inspect the program code means that the docking program is a sealed black box The users have to delegate part of the control they could exert on their simulations provided they have the technical skills to track bugs or customize an external source code The information presented on the accompanying table raises a serious consideration Amongst the programs that were presented in Table I DOCK GOLD AutoDock FlexX ICM Dock and Glide being the most popular while SHEF DARWIN and MORDOR were added for diversity there is not a single consensus on one of the key features presented This is not very trusty at first sight as docking foundations appear foggy to say the least But as a consequence of such a total heterogeneity highlights and weaknesses are quite distinct and method specific A well chosen combination of techniques e g fast and crude to slow
185. m riques une nouvelle m thode de recherche La simulation des syst mes biologiques par mod lisation mol culaire Drug design tapes de la mise au point d un m dicament Identification de la cible pharmaceutique Identification des compos s prometteurs hits Mise au point et optimisation de compos s actifs sp cifiques leads Essais Bilan financier La recherche de nouveaux m dicaments anti cancer R f rences bibliographiques Cible Description La voie de signalisation Ras MAPK Les domaines SH2 La prot ine Grb2 Description et fonction Int r t pharmaceutique Accessibilit exp rimentale Accessibilit th orique Ge NW 00 11 11 12 12 13 13 13 14 15 15 15 15 16 17 17 18 19 23 23 23 24 26 26 26 28 29 Etat actuel des connaissances sur inhibition de Grb2 SH2 Ligands peptidiques D termination de la s quence de r f rence Mode de liaison Sp cificit Ligands pseudo peptidiques ou non peptidiques optimis s Substitution de groupes peptidiques Novartis Extension du r cepteur cibl INSERM CNRS Substituants 4 pTyr et cyclisation Affymax NCI Recherches de ligands actifs in vivo contraintes Contrainte de sp cificit Contrainte de stabilit Contrainte d accessibilit Recherches de ligands actifs in vivo strat gies pour Grb2 SH2 Modification de ligands existants Nouvelles bases structurales R f rences bibliographiques Princip
186. matching enrichment J Mol Model 14 2008 135 148 DOI 10 1007 s00894 007 0257 9 Alexandre Beautrait en tant que programmeur de VSM G est naturellement le premier auteur de cet article En dehors de la r daction avec Alexandre j ai pour ma part effectu les calculs utilisant GOLD ai fourni des codes de bas niveau en particulier pour la conversion de formats ainsi que le programme de cr ation de barres d nergie d j utilis dans l article 1 J ai galement con u et mis en place l ensemble des techniques d analyse validation des r sultats L ancienne r vision de ce document disponible au moment de la soutenance de th se pr sentait une version ant rieure de cet article alors en cours de soumission L article avait pour titre VSM G the Virtual Screening Manager for computational Grids Example of use for the identification of putative liver X receptor ligands Cette version est identique l article finalement publi en ce qui concerne la description de l outil VSM G mais diff rait au niveau du protocole de screening utilis pour valider celui ci Cette validation moins tendue reposait sur la r cup ration par VSM G de 3 ligands connus de LXR au sein d une base de leurres Elle a t abandonn e car elle ne permettait pas de mettre en vidence l efficacit de MSSH SHEF au sein de VSM G dans un contexte de criblage haut d bit en aveugle En particulier la similarit des leurres avec les r f
187. molecular dynamics simulations of the Grb2 SH2 domain structure derived from X ray in water solution and without constraints The protein was complexed with two reference molecules one is a potent and extensively studied inhibitor while the other was shown to have no affinity for the Grb2 SH2 domain in contradiction with previously performed in vacuo simulations Analysis of the MD trajectories for the two complexes reveals interesting features which may explain the stability of the complex obtained for the first molecule versus the poor binding of the other ligand It is shown that water plays a critical role in the stability of these complexes and it is proposed to consider this solvent behavior in forthcoming structure based drug design strategies In this respect structural details are given concerning water molecules reported to be stabilized between the active ligand and the protein receptor We have also identified which residues from Grb2 SH2 and the two ligands could be involved in destabilizing interactions with bulk solvent Finally we propose guidelines for optimizing both ligands by preventing such unwanted effects Abbreviations EGF R epidermal growth factor receptor Grb2 growth factor receptor bound protein 2 ICs concentration required for 50 inhibition MAPK mitogen activated protein kinase MD molecular dynamics pTyr pY phosphotyrosine phosphorylated tyrosine Ras oncogene isolated from rats and associated with sa
188. mploy sur de petites bases pour tester l efficacit de diff rents modules et ainsi estimer plus pr cis ment les param tres de filtrage optimaux On peut supposer que la probabilit de faux n gatifs d une m thode donn e et corr l e des param tres physico chimiques pr cis Il convient donc de maximiser la diversit des mol cules test es pour la d tection d erreurs tandis que lorqu un faux n gatif est mis en vidence il faudra tester en priorit les mol cules rejet es les plus similaires tt Par exemple on peut augmenter le taux de rejet d un module par petits paliers tant qu aucun faux n gatif n est d tect Dans le cas contraire il faut le diminuer et ventuellement supprimer le module du funnel s il s av re que le rapport entre capacit de filtrage et vitesse de traitement devient insuffisant face un autre module 58 Conclusions Dans le cadre de cette th se j ai t amen conduire aussi bien des simulations tr s pr cises de dynamique mol culaire que des calculs de docking flexible l chelle du screening virtuel J ai enfin particip la mise en place de prototypes de protocoles plus haut d bit Ce travail couvre ainsi les deux tapes de la recherche pharmaceutique auxquelles les simulations num riques de syst mes biomol culaires peuvent contribuer la phase de d couverte de hits suivie de la phase d optimisation de leads Certains des obstacles que l
189. mportant que ligand 1 51 Docking flexible sur des bases de mol cules Situation initiale Pr r r GOLD 2 1 a Ise 67 cha ae du solvant e 0LD Genetic Optimisation for Ligand Docking nf Derun smsent Submitaext ent Configuration re goidcont eef Les r sultats de dynamique mol culaire allant dans le sens d un ee ee GAE RE impact potentiellement important du solvant lors de la liaison d un manes owe MBSE FT Pr p m ligand sur Grb2 SH2 il s av ra n cessaire de tenir compte de cet un cs aie a aspect au niveau de la mise en place d un protocole de docking eee A es Covalent 1 Display Output Edit Parameters Parameter File DEFAULT La plupart des programmes de docking ne pr voient pas une telle Fitness Function Settings possibilit et dans le cas de GOLD version 2 que nous avons caldscore v chenseare v User Deed Score sees oar 3 Fitness Flags Edit Constraints Late m ones E utilis les interactions caract risant la d solvatation du r cepteur a Torsion Angle Distributions sa ce A a ie 7 Use Distributions WY Edit Distributions Distributions File DEFAULT la liaison du ligand sont favoris es posteriori Le fait qu un Annealing Parameters A gt g n r 5 van der Waals 40 Hydrogen Bonding 25 _ ligand doive imp rativement chasser les mol cules d eau du site PET eee actif est traditionnellement consid r comme une condition Genetic
190. n biochimie ne doivent pas tre consid r s comme des syst mes chimiques classiques dans lesquels les mol cules sont en interaction en milieu dilu Le fait que le corps humain soit constitu d environ 70 d eau signifie en particulier du point de vue chimique que la plupart des biomol cules se trouvent en solution des concentrations particuli rement lev es d environ 30 71 formant une masse en mouvement et en interaction 72 L re de la post g nomique Conceptuellement le code g n tique d un organisme vivant correspond son programme de fonctionnement ainsi le d chiffrage de ce code constitua pour la communaut scientifique un objectif fondamental Etant donn que biologiquement toute fonction d coule de ce codage le but ultime de la g n tique fut l interpr tation de toute fonction biologique partir d un code g n tique Cet enthousiasme est pr sent retomb La g n tique s av ra un outil remarquable sur le plan m dical et a permis d identifier la source de nombreuses pathologies tout en fournissant de pr cieuses indications sur des traitements pharmaceutiques potentiels Mais il est pr sent vident que l tude du g nome ne peut pas tre la recette miracle qui r volutionnera la compr hension du vivant En particulier l hypoth se r ductionniste un g ne une fonction biologique s av ra fausse Pour un grand nombre de g nes la fonction biologique c
191. nd stabilisation of the unfavourable N6 H form by intramolecular hydrogen bonding crystalling L His Gly hemihydrate Chemical Communications Cambridge United Kingdom 13 1997 1207 1208 58 http www chemdiv com 59 http www enamine net 60 http www amridirect com 61 Lipinski C A Lombardo F Dominy B W and Feeney P J Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings Advanced Drug Delivery Reviews 23 issue 1 1997 3 25 62 Monge A Arrault A Marot C and Morin Allory L Managing profiling and analyzing a library of 2 6 million compounds gathered from 32 chemical providers Molecular Diversity 10 issue 3 2006 389 403 63 Xue L Godden J and Bajorath J Database searching for compounds with similar biological activity using short binary bit string representations of molecules Journal of Chemical Information and Computer Sciences 39 issue 5 1999 881 886 64 Tanimoto T T Non linear model for a computer assisted medical diagnostic procedure Transactions of the New York Academy of Sciences 2 issue 23 1961 576 580 65 Hibert M and Haiech J Des g nes aux m dicaments nouveaux d fis nouvelles strat gies M S M decine Sciences 16 issue 12 2000 1332 1339 66 http chimiotheque nationale enscm fr 67 http www ccdc cam ac uk products life_sciences valid ate gold_validation 68
192. ne Je ne regrette pas une seule seconde d avoir fait cette th se cette exp rience m a apport beaucoup et chacun d entre vous m y a aid Note importante Ce document constitue une version r vis e du manuscrit de th se Il r percute les suggestions formul es par le jury lors de la soutenance La section publications a t modifi e car au moment de l impression du manuscrit de th se les articles n taient pas publi s deux d entre eux taient toutefois sous presse On trouvera ici les publications dans leur version finale La liste des changements intervenus pour l un d entre eux pourra tre consult e en exergue de la section publications Ce document ainsi que sa r vision initiale est disponible sous forme lectronique sur demande leroux vincent free fr Table des mati res Pr sentation Contexte Cancer Avant propos Fonctionnement du cycle cellulaire et causes du cancer Les diff rentes approches de la lutte contre le cancer L tude th orique du vivant a l chelle mol culaire Structure des biomol cules Des acides nucl iques l information g n tique Du g nome au prot ome Du g nome l interactome description du fonctionnement d un organisme vivant Du prot ome l interactome Constitution mol culaire d un organisme vivant L re de la post g nomique Importance des simulations num riques L informatique une r volution scientifique Les simulations nu
193. ne charge 2 ce qui permet d interpr ter la force de la liaison pTyr SH2 Le reste de la cavit est compos de trois s rines Sgg Ser BB Soo Ser BC et Soo Ser BC La cavit hydrophobe quand elle est d limit e par Kio Lys BD et constitu e de L 20 Leu La num rotation correspond l ordre des r sidus sur Grb2 tandis que la notation entre parenth ses distingue les r sidus suivant la zone g om trique de SH2 laquelle ils appartiennent dans l ordre BA AA aA AB BB BC 30 BE et Wiz Trp EF qui interagissent plus faiblement avec l asparagine de la s quence de r f rence pYXNX Ce mod le est conforme la description classique et simpliste 100 des complexes de domaines SH2 qui comprennent pour la plupart un ligand deux branches li es dans deux cavit s distinctes l une pour pTyr et l autre hydrophobe La s quence du domaine SH2 de Grb2 contient galement le motif FLIRES analogue au motif FLVRES souvent cit comme tant d une importance capitale pour la reconnaissance de la phosphotyrosine et pr sent sur la plupart des domaines SH2 Sp cificit Grb2 SH2 comme nous l avons vu est doublement s lectif sur les s quences peptidiques auquel il se lie si la pr sence d un r sidu pTyr est attendue car syst matique avec les domaines SH2 la s lectivit pour un r sidu asparagine en position pY 2 distingue Grb2 des autres prot ines contenant un dom
194. netics 41 issue 2 2000 173 191 152 Brooks B R Bruccoleri R E Olafson B D States D J Swaminathan S and Karplus M CHARMM A program for macromolecular energy minimization and dynamics calculations Journal of Computational Chemistry 4 issue 2 1983 187 217 153 MORDOR MOlecular Recognition with a Driven dynamics OptimizeR http mondale ucsf edu science mordor html ID 6 amp sID 6 Mo VIB Loote ALTE osz SYA Wares SUCRES SET Mery HAHAE puonpq ayy q 1 p10 tuonraodiog 5 Irl0 K SET BU 14 Je fu eee a 1 euo Aq painqunsiq peoiqy SIH AEC ch SZ Y AA BS Bd cs 10 Aq painqunsiq a fo Fa OS BS UXME umay Suifiog q pa ERA ECE ea 1oy xYs pew Z8YBSSTI 010 9800 PL uo ow adr awoy xysl pew T8P8BSSTI 010 9800 LH 080001 Surfrog 080001 Sih HF asi TST IESE ch als bd eH ne unouen33uou7 SL A Je CAA HY OS 1 14 Buipei yoog peuo MIEHET a a RESTES REET HORS ARABE PA LOOZ T ON 2 TA HIS SW 4002 p861 ur poeg Auon CHRE 861 HE Ans poyddy pue s197ndu07 ENHANE UEH SLEMTEHANVMESNSETLTNER SULEMS SNN HS NERH EXYHT EYVHERE TR TE AY AOC ETR AR ZW CHERE PSLEPI09 010 980 HH AB uo ov adi owoy AXSI feux 7886679 010 980 54 HIFH 00 TI 00 O1 HT BY fy HJE 8 MICHEL HUM Eh g LE TPR Ee EE N OSOOOT GMT Qh TSE HE Coe AYE X EX RE SE Ek FEM fel Hp X 007 YHA L 0008 MF IC 000 ME fra
195. ng Z Gilliland G Bhat T N Weissig H Shindyalov I N and Bourne P E The Protein Data Bank Nucleic Acids Research 28 issue 1 2000 235 242 6 Smith Schmidt T Banking on structures BioIT World 1 issue 8 2002 7 Chanda S K and Caldwell J S Fulfilling the promise drug discovery in the post genomic era Drug Discovery Today 4 2003 168 174 8 Nordhaus W D An economic history of computing 2006 http nordhaus econ yale edu computing June2006 pdf 9 Neshich G Borro L C Higa R H Kuser P R Yamagishi M E B Franco E H Krauchenco J N Fileto R Ribeiro A A Bezerra G B P Velludo T M Jimenez T S Furukawa N Teshima H Kitajima K Bava A Sarai A Togawa R C and Mancini A L The Diamond STING server Nucleic Acids Research 33 issue Web server issue 2005 W29 W35 10 Neshich G Mancini A L Yamagishi M E Kuser P R Fileto R Pinto I P Palandrani J F Krauchenco J N Baudet C Montagner A J and Higa R H STING Report convenient web based application for graphic and tabular presentations of protein sequence structure and function descriptors from the STING database Nucleic Acids Research 33 issue Database issue 2005 D269 D274 11 STING Sequence To and withIN Graphics http www cbi cnptia embrapa br SMS 12 DiMasi J A Hansen R W and Grabowski H G The price of innovation new estimates of drug development costs Journal of Health Economics 22 2003 151 185
196. ng information is represented as energy bars in figures 10 a and 10 b Figure 10 Interaction energy between significant Grb2 SH2 residues and a ligand 1 left b ligand 2 right Background colors for residues names are the same as the ones employed in figure 6 small snapshots Ke4 Lys AA Ese Glu BC F101 Phe BC Q106 Gin BD H107 His BD F108 Phe BD The variations of the binding energies of the involved Grb2 SH2 residues are small in the case of ligand 1 particularly with the residues forming cavity 1 The corresponding binding mode in cavity 1 involving Rez Arg aA and Rss Arg BB but not Kio Lys BD is less favorable energetically than the initial binding of ligand 2 which mostly involve Rgs Arg BB and Kio Lys BD Indeed the interactions with Re Arg aA and the three cavity 1 serines are lower in the case of ligand 2 by 40 and 30 kcal mol respectively This is compensated by the interaction with Kio Lys BD which fluctuates around 140 kcal mol as compared to 40 kcal mol with ligand 1 However after 1800 ps of dynamics the corresponding H bonds are lost which is correlated with the ligand 2 binding disruption observed in figures 7 b and 9 This leads to a conformation where Rgs Arg BB is the only positively charged residue making H bonds with the phosphones and where the interaction of Kio Lys BD with ligand 2 is now of the same order of magnitude as with ligand 1 translating a
197. ng on the SH2 domain of Src 39 40 which indicated that solvent effects could intervene at the pTyr binding site common to all SH2 domains as well as in specificity zones The present MD simulations confirm this behavior of water concerning the SH2 domain of Grb2 but also highlight two critical issues i the involvement of a limited number of water molecules bridging the ligand and the receptor in positions persistently occupied during the simulations and more importantly ii the interaction between bulk water and specific protein and ligand residues which appeared to lead to disruption of ligand binding in one case Therefore this work suggests that the Grb2 SH2 binding site could be partly hydrated upon ligand binding as was observed with other systems 56 57 and gives precise information for the prospective inclusion of discrete water molecules in Grb2 SH2 models as input in library screening methods that use structural data In this regard future studies will use the GOLD docking program 58 59 for which the structural water issue was recently addressed 52 and as a more accurate step in screening the SIBFA molecular mechanics procedure 60 61 where the additional effects of bulk solvation can be accounted for by a Continuum reaction field 62 Moreover the knowledge of the potential solvent effects on the binding of ligands on the SH2 domain of Grb2 as well as the MD protocol that can be used to unravel them fulfills the requ
198. nhibition of SH2 domain phosphoprotein association by a nonhydrolyzable phosphonopeptide Biochemistry 31 1992 9865 9870 123 Burke Jr T R Smyth M S Otaka A Nomizu M Roller P P Wolf G Case R and Shoelson S E Nonhydrolyzable phosphotyrosyl mimetics for the preparation of phosphatase resistant SH2 domain inhibitors Biochemistry 33 1994 6490 6494 124 Yao Z J King C R Cao T Kelley J Milne G W A Voigt J H and Burke Jr T R Potent inhibition of Grb2 SH2 domain binding by non phosphate containing ligands Journal of Medicinal Chemistry 42 1999 25 35 125 Burke Jr T R and Lee K Phosphotyrosyl mimetics in the development of signal transduction inhibitors Accounts of Chemical Research 36 issue 6 2003 426 433 126 Gao Y L W Luo J H Guo R Yang D Zhang Z Y and Burke Jr T R Examination of novel non phosphorus containing phosphotyrosyl mimetics against protein tyrosine phosphatase 1B and demonstration of differential affinities toward Grb2 SH2 domains Bioorganic amp Medicinal Chemistry Letters 10 2000 923 927 127 Gao Y Voigt J Xu J X Yang D and Burke Jr T R Macrocyclization in the design of a conformationally constrained Grb2 SH2 domain inhibitor Bioorganic amp Medicinal Chemistry Letters 11 2001 1889 1892 128 Gao Y Wei C Q and Burke Jr T R Olefin metathesis in the design and synthesis of a globally constrained Grb2 SH2 domain inhibitor Organic Letters 3
199. nic amp Medicinal Chemistry Letters 9 2001 1439 1445 42 81 Wei C Q Li B Guo R Yang D and Burke Jr T R Development of a phosphatase stable phosphotyrosyl mimetic suitably protected for the synthesis of high affinity Grb2 SH2 domain binding ligands Bioorganic amp Medicinal Chemistry Letters 12 2002 2781 2784 82 Wei C Q Gao Y Lee K Guo R Li B Zhang M Yang D and Burke Jr T R Macrocyclization in the design of Grb2 SH2 domain binding ligands exhibiting high potency in whole cell systems Journal of Medicinal Chemistry 46 2003 244 254 83 Lung F D T and Tsai J Y Grb2 SH2 domain binding peptide analogs as potential anticancer agents Peptide Science 71 2003 132 140 84 Gram H Schmitz R Zuber J F and Baumann G Identification of phosphopeptide ligands for the Src homology 2 SH2 domain of Grb2 by phage display European Journal of Biochemistry 246 1997 633 637 85 Lung F D T Tsai J Y Wei S Y Cheng J W Chen C Li P and Roller P P Novel peptide inhibitors for Grb2 SH2 domain and their detection by surface plasmon resonance Journal of Peptide Research 60 2002 143 149 86 Shi Z D Karki R G Oishi S Worthy K M Bindu L K Dharmawardana P G Nicklaus M C Bottaro D P Fisher R J and Burke Jr T R Utilization of a nitrobenzoxadianole NBD fluorophore in the design of a Grb2 SH2 domain binding peptide mimetic Bioorganic amp Medicinal Chemistry Letters
200. nontransforming avian tumor viruses Proceedings of the National Academy of Sciences USA 67 issue 4 1970 1673 1680 19 28 zur Hausen H Papillomaviruses and cancer From basic studies to clinical application Nature Reviews Cancer 2 2002 342 350 29 Pahlman L Beger H and Kroon B The place of surgical oncology in general surgery European Journal of Surgical Oncology 25 1999 619 621 30 Dobbs J Barrett A and Ash D Practical radiotherapy planning 3rd ed 1999 Arnold London 31 Cairns J Boyle D and Frei E Cancer chemotherapy Science 220 issue 4594 1983 252 256 32 Grady D Rubin S M Petitti D B Fox C S Black D Ettinger B Ernster V L and Cummings S R Hormone therapy to prevent disease and prolong life in postmenopausal women Annals of Internal Medicine 117 issue 12 1992 1016 1037 33 Muss H B Endocrine therapy for advancer breast cancer A review Breast Cancer Research and Treatment 21 issue 1 1992 15 26 34 Azria D Lemanski C Zouhair A Gutowski M Belkac mi Y Dubois J B Romieu G and Ozsahin M Hormonoradioth rapie adjuvante concomitante des cancers du sein tat de l art Cancer Radioth rapie 8 2004 188 196 35 Dachs G U Dougherty G J Stratford I J and Chaplin D J Targeting gene therapy to cancer a review Oncology Research 9 1997 313 325 36 Pardoll D M Cancer vaccines Trends in Pharmacological Sciences 14 issue 5 1993 202 208 37 Pa
201. ns ces conditions nous ne pouvons pas qualifier la valeur des r sultats obtenus Mais l encore un protocole a t clairement tabli et les r sultats sont disponibles il s agit de la premi re tape d un travail de longue haleine qui s int gre dans le d veloppement de la plateforme VSM G Sur un plan personnel cette tude s est av r e au final bien plus complexe qu elle ne le paraissait au premier abord Si j tais conscient au d but de mon doctorat que les techniques de la chimie informatique ne sauraient se limiter lancer des calculs et analyser leurs r sultats et implique une remise en question permanente des mod les des techniques et des m thodes d analyse j ai fait preuve d un certain optimisme Je souhaite que le lecteur de ce document s il n en tait pas d j convaincu aura pu percevoir la complexit mais aussi la richesse et le potentiel actuel des m thodes d riv es de la mod lisation mol culaire 59 60 Publications Article 1 Leroux V Gresh N Liu WQ Garbay C Maigret B Role of water molecules for binding inhibitors in the SH2 domain of Grb2 a molecular dynamics study J Mol Struct 806 2007 51 66 DOI 10 1016 j theochem 2006 11 010 Article 2 Beautrait A Leroux V Chavent M Ghemtio L Desvignes MD Smail Tabbone M Cai W Shao X Moreau G Bladon P Yao J Maigret B Multiple step virtual screening using VSM G overview and validation of fast geometrical
202. ns results in growth inhibition of Philadelphia chromosome positive leukemic cells Biochemical and Biophysical Research Communications 235 1997 383 388 56 Daly R J Binder M D and Sutherland R L Overexpression of the Grb2 gene in human breast cancer cell lines Oncogene 9 issue 9 1994 2723 2727 57 Sastry L Cao T and King C R Multiple Grb2 protein complexes in human cancer cells International Journal of Cancer 70 1997 208 213 58 Kairouz R and Daly R J Modulation of tyrosine kinase signalling in human breast cancer through altered expression of signalling intermediates Breast Cancer Research and Treatment 2 2000 197 202 59 Pendergast A M Quilliam L A Cripe L D Bassing C H Dai Z Li N Batzer A Rabun K M Der C J Schlessinger J and others BCR ABL induced oncogenesis is mediated by direct interaction with the SH2 domain of the GRB 2 adaptor protein Cell 75 issue 1 1993 175 185 60 Gay B Suarez S Weber C Rahuel J Fabbro D Furet P Caravatti G and Schoepfer J Effect of potent and selective inhibitors of the Grb2 SH2 domain on cell motility Journal of Biological Chemistry 274 issue 33 1999 23311 23315 61 Soriano J V Liu N Gao Y Yao Z J Ishibashi T Underhill C Burke Jr T R and Bottaro D P Inhibition of angiogenesis by growth factor receptor bound protein 2 Src homology 2 domain binding antagonists Molecular Cancer Therapeutics 3 issue 10 2004 1289 1299
203. nt de d crire un gaz ou un cristal quelconque Et en chimie th orique les m thodes ab initio se limitent des assemblages simplistes en l absence de moyens num riques t Remarquons que de nombreux appareillages modernes incorporent afin d affiner ou d interpr ter les mesures par exemple des donn es spectrom triques des traitements par simulations num riques L outil informatique n est donc pas seulement indispensable au niveau conceptuel mais galement au niveau technique m me si ce dernier point n est pas aussi visible Dans le cas de syst mes mol culaires on peut franchir des barri res de potentiel afin d explorer au mieux un espace conformationnel De plus la mise en place de transformations alch miques est la base des m thodes de calcul d nergie libre 13 Pour de nombreux syst mes la mise en uvre de la simulation correspond la mise en place d un mod le et d un protocole qui constituent un compromis acceptable entre le r alisme des r sultats et l accessibilit en termes de temps de calcul Nous d taillerons quel niveau des approximations devront tre effectu es dans le cadre de la simulation de syst mes biomol culaires La simulation de syst mes biologiques par mod lisation mol culaire Les diff rentes m thodes que nous avons utilis es pour ce travail appartiennent la classe de la mod lisation mol culaire Nous les avons appliqu es l tude de syst
204. nt de plus en plus fr quent et ses ve a Head Disease implications m dicales et sociales sont comme a er nous l avons vu un probl me majeur Il z TE constitue aussi pour l industrie pharmaceutique 3 5000 un march mergeant particuli rement i prometteur Le cancer est d j la premi re 3 e cause de mortalit dans les pays d velopp s ci i ss o contre volution des cause de mortalit aux A aR PEN Etats Unis 2 On estime que d ici 2008 le n cancer deviendra aussi le premier march ki mondial pour l industrie pharmaceutique Celle o o i ie LEE RAR PARA z ci semble effectivement consentir cet axe de 3332423328323333 recherche des moyens particuli rement importants En r sum les enjeux de la recherche d une nouvelle classe de m dicaments anti cancer sont aussi bien scientifiques que m dicaux et commerciaux Chimioth rapie comme radioth rapie s attaquent la r plication de l ADN et la division cellulaire et sont efficaces contre le cancer lorsqu ils sont dirig s vers les zones de l organisme h bergeant les tumeurs Il s agit de traitements anti prolif ration et non sp cifiquement anti cancer qui s accompagnent d effets secondaires toxiques non n gligeables On peut toutefois mettre quelques r serves sur une telle r alit Des annonces concernant les m dicaments du futur agissant sur des pathologies aussi complexes que les cancers
205. nt de transf rer l information g n tique dans les cellules tandis qu ils sont cibl s par le syst me immunitaire de l individu Le d veloppement de vaccins anti cancer ou immunoth rapie 36 40 dont le but est aussi bien de pr venir l infection de l organisme par des agents canc rig nes que de favoriser les r ponses immunitaires au d veloppement d une tumeur ult rieure potentielle fait galement l objet d efforts de recherche pouss s Un autre domaine de recherche prometteur de la lutte anti cancer et dans lequel nous nous situons dans le cadre de cette tude consiste 4 mettre au point des inhibiteurs chimiques plus sp cifiques 41 42 avec potentiellement une efficacit accrue et des effets secondaires 43 moins importants pour le patient 44 De tels m dicaments s av rent d ja utiles en compl ment de traitements anti cancer plus conventionnels 45 Ces inhibiteurs peuvent tre des mol cules en partie ou en totalit synth tiques et ne ciblent ni le systeme immunitaire immunoth rapie ni certaines hormones hormonoth rapie mais des prot ines 46 47 et des voies de signalisation 48 51 dont le r le dans le d veloppement du cancer est bien caract ris Cela n cessite pr alablement une connaissance plus pouss e des m canismes biomol culaires correspondants 52 ce niveau l interpr tation de r sultats de recherches en g n tique peut s av rer pr cieux 53 Quelques liens utiles
206. nt part of the initial ligand 2 interaction is lost in favor of a better interaction with the solvent Thus as already observed with the analysis of the conformations the behaviors and ligand 1 and ligand 2 regarding the interaction energy with Grb2 SH2 differ completely As expected from in vacuo simulations ligand 2 in its initial conformation interacts more strongly with Grb2 SH2 than ligand 1 interaction energy of 552 7 kcal mol instead of 532 1 kcal mol and less with water 192 5 kcal mol instead of 257 7 kcal mol However during the course of the MD it appears that the ligand 2 interaction decreases significantly to reach a level of interaction of the same order than its interaction with the solvent This implies that ligand 2 can not be considered bound to Grb2 SH2 anymore contrary to ligand 1 Figure 7 Variations of the interaction energy between the Grb2 SH2 protein and the ligand P L plain curves and between the ligand and water L W dotted curves a corresponds to the ligand 1 complex left b to the ligand 2 complex right 200 32 1 keal mol LL E Liw E 257 7 keal mol 100 E 484 6 keal mol E m Y 271 8 kcal mol lt a EE 52 7 kcal mol E 192 kcalmol L Eg 3942 keal mol Bym 342 4 kcalmol In order to account more precisely for the binding of ligand 1 and ligand 2 on Grb2 SH2 the interactions of specific
207. ntitative structure activity relationships MM molecular mechanics FF molecular mechanics forcefield MC Monte Carlo methods Trends in modern drug discovery Nowadays the drug discovery process is being more and more complex and costly As an interdisciplinary field by essence 1 it is correlated to the increasing knowledge provided by the large number of the omics issues and has to follow the advent of many new chemical biological and computer techniques The highest impact was certainly provided by the progress in computational capabilities which are now a key factor in modern drug design 2 3 In addition to such advances there is an explosion of available data and knowledge that can be used to designate new targets of interest for drug research In particular the great efforts in genetics provided a much deeper understanding of biological processes and allowed the identification of more and more targets of medical interest However in that case the early enthusiasm was dampened as it appeared that biological function is indeed related mostly to interactions between biomolecules mainly proteins In that regard the study of putative targets can be conducted at different levels as in fine their three dimensional structure must be taken into account to understand or predict their behavior giving a much larger space to be investigated than the genomics sequential space itself 4 Currently the human genome is decrypted and the
208. number of available protein experimental structures is growing exponentially 5 6 Simultaneously as biologists increased their understanding of cell components it became clear that simply obtaining a full list of them will not tell us how a cell works Rather even for a substructure that has been well characterized there are significant difficulties in understanding how components interact to produce the observed behavior Consequently a level of perception higher than genomics has to be assessed encompassing the list and the nature of the interactions between biomolecules Even if it is an incredibly complex task it is now admitted that the effort in the drug discovery process must be shifted towards the proteome 7 and ultimately to the interactome Given all those advances and considering the rate at which progresses are made more generally in all the domains associated to drug design 1 computing technology 8 and biochemical information access and management 9 11 it could be postulated that in consequence drug discovery would benefit from that trend and follow it A quick look in this direction indicates however that this is not the case the average cost 800 millions in 2000 and duration 15 years for the development of a new drug actually increases 12 as well as the failure rate amongst compounds submitted to clinical trials by the pharmaceutical industry 13 Several popular explanations involve parameters external to the
209. odel the system define the search space and evaluate conformations In this regard rather than focusing on precision and speed we defined the main internals of docking programs presented in decreasing importance order Firstly decisions have to be taken regarding molecules and target flexibility Typically if the docking simulations are performed on the 3D structure of a target extracted from an X ray structure of this target associated with a given ligand then if the receptor flexibility treatment is too limited the docking program will artificially favor the molecules that share the binding mode corresponding to the ligand present in the X ray complex If other binding modes can exist because of large induced fit effects the docking is biased The scoring function and search engine are the driving forces of the docking algorithms and are therefore directly related to their efficiency For example the GOLD docking program having a very good search engine and a relatively weak scoring function is good at predicting structures but is known to perform poorly if the score value is used to rank VS results Next solvation handling appears mandatory in some systems as discrete water molecules can bridge specific protein ligand interactions 63 64 Not permitting the treatment of such waters would again introduce a harmful bias and the programs are not equal in that area Target definition must also be taken into account if it is not detailed precise
210. of distant sites in non phosphorylated cyclic peptide antagonists of the Grb2 SH2 domain Biochemical and Biophysical Research Communications 310 2003 334 340 150 Song Y L Roller P P and Long Y Q Development of 1 3 aminotyrosine suitably protected for the synthesis of a novel nonphosphorylated hexapeptide with low nanomolar Grb2 SH2 domain binding affinity Bioorganic amp Medicinal Chemistry Letters 14 2004 3205 3208 151 Song Y L Peach M L Roller P P Qiu S Wang S and Long Y Q Discovery of a novel nonphosphorylated pentapeptide motif displaying high affinity for Grb2 SH2 domain by the utilization of 3 substituted tyrosine derivatives Journal of Medicinal Chemistry 49 issue 5 2006 1585 1596 152 Liu W Q Carreaux F Meudal H Roques B P and Garbay C Synthesis of constrained 4 phosphonomethyl phenylalanine derivatives as hydrolytically stable analogs of O phosphotyrosine Tetrahedron 52 issue 12 1996 4411 4422 153 Stankovic C J Surendran N Lunney E A Plummer M S Para K S Shahripour A Fergus J H Marks J S Herrera R Hubbell S E Humblet C Saltiel A R Stewart B H and Sawyer T K The role of 4 phosphonodifuloromethyl and 4 phosphono phenylalanine in the selectivity and cellular uptake of SH2 domain ligands Bioorganic amp Medicinal Chemistry Letters 7 1997 1909 1914 154 Marseigne I and Roques B P Synthesis of new amino acids mimicking sulfated and phosphorylated
211. ois champs de force indique clairement que sur le plan qualitatif les r sultats d un champ de force l autre sont similaires De fa on satisfaisante les trois simulations du complexe ligand 1 confirment la stabilit de celui ci tandis que les complexes de ligand 2 s av rent instables Les cluster graphs des 6 trajectoires des complexes indiquent toutefois que des diff rences mineures peuvent tre observ es On peut ainsi noter pour les trajectoires du complexe ligand 1 avec les champs de force CFF91 et AMBER une l g re r organisation structurale post rieure la phase d quilibration qui n appara t pas dans la dynamique avec CHARMM22 La trajectoire obtenue avec ce dernier champ de force reste donc la plus stable par rapport a la structure exp rimentale de d part m me si on peut consid rer que la stabilit avec les deux autres champs de force s av re tout fait satisfaisante Merci Matthieu Chavent pour ces images 49 Il n a pas t jug bon de proc der dans l article une analyse structurale d taill e de toutes les trajectoires Nous avons choisi arbitrairement de nous focaliser sur les dynamiques obtenues avec CHARMM22 NAMD d une part pour des raisons de clart et d autre part car l analyse des r sultats de NAMD est facilit e barres d nergie Ainsi 1 le protocole de simulation mis en place reproduit les r sultats exp rimentaux et 2 l hypoth se du choix du champ
212. olve taking into account basic chemical properties to extend the complementarity score that is currently computed Such an approach has already been tried out in the ligand based drug design area 72 Regarding the exploration strategy in its current implementation in VSM G SHEF acts as a rigid docking program that only selects a single conformer out of a list for a given structure this approach has been shown here to produce significant numbers of false positives in some cases An alternative could be to use a diverse set of docked conformers for each ligand the selection between them being made by a second module in the screening funnel protocol Various techniques are being considered in this regard 73 75 In any case it is uncertain that improvements of the SHEF algorithm would necessary be worthwhile At the present time the main advantage of the MSSH SHEF approach is its speed With the safe parameters used in this report SHEF is typically 2 3 orders of magnitude faster for processing gt 10 conformers than GOLD for docking the corresponding 10 structures MSSH is still 1 order of magnitude faster than GOLD and its calculations can be done once and for all for a given molecular database Enhancements of the MSSH and SHEF programs should obviously not be made at the cost of the loss of such a computing speed advantage that allows for performing large scale structure based VS In further work we will focus on selection rather than on filt
213. om groups as well as the corresponding NAMD configuration files and to start the NAMD jobs Another program was made for converting a series of time energy values such as those obtained with the aforementioned calculations into graphic files containing energy bars with time as x axis variable using a black and white or a color scale from E gt 0 white to E lt user defined negative value black This procedure for representing the distribution of the binding interactions will be used systematically in our group to obtain a quick representation of the energetical repartition of the interactions between bio molecules and to monitor their fluctuation during MD simulations We also plan to implement it in VMD and its InterSurf plugin 48 so that protein binding sites can be spotted easily and the corresponding interactions clearly represented graphically during dynamics analysis Results Experimental measurements The measurement of the affinity of ligand 1 for Grb2 SH2 has already been presented 29 According to the ELISA experiments while ligand 1 exhibits a very strong inhibitory activity on Grb2 SH2 signalling with an Co of 11 1 nM ligand 2 presents no such effect C59 gt 10 M Evolution of the conformation of ligand 1 and ligand 2 complexes The behaviors of ligand 1 and ligand 2 are entirely different during the MD simulations The conformational variations of the ligands are depicted as cluster graphs in figure 5 Structur
214. omputing science Interfaces overcoming the obstacles associated with heterogeneous programs building bridges between LBDD SBDD database filtering techniques and more generally knowledge in drug design are strongly desired One of the goals of the post genomic era is indeed to benefit fully from the explosion of biological chemical and structural data for which web based services 9 11 57 74 97 and specific databases 5 57 73 74 97 103 provide unprecedented access Its application to drug discovery certainly involves linking this data to the variety of methods of SBDD that will most probably benefit more and more from the advances in modern computing such as peer to peer networks 15 104 107 and computational grids 108 111 Interestingly the knowledge enrichment is currently taking off inside the boundaries of specific approaches The gathering of SAR data starting to link LBDD with modern biological knowledge has been reported recently 112 The VSM G platform currently being developed 67 at the SBDD VS level within the framework of computational grids has the long term purpose to implement a similar united we stand philosophy The fusion of the great richness of complementary approaches benefiting from the explosion of genomics and proteomics data should eventually lead in the future to a more efficient global CADD driving concept that we could name knowledge based drug design In conclusion to our question regarding
215. on France 4 Interprobe Chemical Services Gallowhill House Larch Avenue Lenzie Kirkintilloch Glasgow G66 4HX Scotland UK 5 Shanghai Institute of Organic Chemistry Laboratory of Computer Chemistry and Chemoinformatics 354 Fenglin rd Shanghai 200032 P R China Corresponding author bernard maigret loria fr 33 354 958 608 telephone 33 383 275 652 fax Abstract Numerous methods are available for use as part of a virtual screening strategy but yet none of those is solely able to guarantee both a level of confidence comparable to experimental screening and a computing efficiency that could drastically cut down the costs of early phase drug discovery campaigns We present here VSM G Virtual Screening Manager for computational Grids a virtual screening platform that combines several structure based drug design tools VSM G aims to be as user friendly as possible while retaining enough flexibility to accommodate other in silico techniques as they are developed In order to illustrate VSM G concepts we present a proof of concept study of a fast geometrical matching method based on spherical harmonics expansions surfaces This technique is implemented in VSM G as the first module of a multiple step sequence tailored for high throughput experiments We show that using this protocol notable enrichment of the input molecular database can be achieved against a specific target here the LXR nuclear receptor The benefits limitat
216. on 3 GOLD propose un algorithme qui sur la base d une liste de positions possibles typiquement connues par cristallographie RX d termine la liste des mol cules d eau sur le site actif qui seraient probablement concerv es pour une conformation donn e d un ligand et en tient compte au niveau de la valeur du score Il s agit d une volution particuli rement bienvenue Cette fonctionnalit n tait malheureusement pas disponible lorsque les simulations pr sent es ici ont t conduites I1 s agit d une n cessit absolue dans la mesure o les tests de validation bas s sur cette d marche ne donnent pour les meilleurs programmes que des taux de r ussite de l ordre de 60 70 De plus l efficacit d un programme de docking donn d pend de la nature du syst me qu on lui applique il fallait donc v rifier que GOLD pouvait bien s appliquer Grb2 SH2 et dans quelles conditions 52 Limitations L utilisation d un programme de docking impose fondamentalement des limitations plus importantes que celles des techniques de la dynamique mol culaire Principalement les effets dynamiques sont par nature inaccessibles une fonction de score empirique et approximative est employ e au lieu d un calcul nerg tique faisant intervenir un champ de force et enfin l espace de recherche conformationnel est restreint en particulier au niveau du r cepteur Le programme GOLD a t choisi car il tient compte de la
217. on en 3D puis protonation avant d tre dock es sur Grb2 SH2 Au moment de d terminer le protocole de docking GOLD semblait tre le programme potentiellement le plus pr cis nous nous sommes content s de v rifier qu il l tait suffisamment pour notre cible en utilisant les structures exp rimentales connues Les techniques de docking sont en constante volution et de nouveaux programmes apparaissent r guli rement M me si GOLD a connu des am liorations notables en particulier au niveau de la prise en charge du solvant son int r t est actuellement nettement moins vident face aux versions r centes de programmes comme ICM ou Glide On peut donc dire par rapport aux calculs de docking que s ils taient refaire il faudrait sans doute avant toute chose prendre plus de temps pour estimer dans la mesure du possible l efficacit des diff rentes solutions disponibles L article 3 d crit une partie des probl matiques associ es une telle situation 53 R sultats Une des premi res simulations de docking apr s validation du protocole consista partir d un squelette mol culaire propos par des exp rimentateurs poss dant quatre zones de substitution pour deux substituants possibles d terminer la position optimale de ces derniers pour Grb2 SH2 Il s av ra clairement que les positions les plus favorables en vert ci contre correspondaient pr cis ment aux synth ses les plus d licates ci
218. on with Grb2 SH2 is similar in magnitude to the one of ligand 1 phosphate oxygens Regarding the residues in 1 position both the ligand 1 mAz and the ligand 2 naphthalene contributions to the binding energy are negligible during the whole 2 ns of trajectories which confirms that these groups do not make stacking interactions with Grb2 SH2 whereas they clearly interact with water Weak favorable interactions observed with mAz are solely due to the presence of the terminal NH group which however is not H bonded to Grb2 SH2 as observed in the ligand 1 complex X ray structure The binding energy curves of ligand 1 OPO and ligand 2 PO groups including the unfavorable contributions of the phosphorous atoms are presented on figure 9 Figure 8 b indicates that the interaction between the second ligand 2 phosphone and Grb2 SH2 decreases during the simulation Figure 9 shows more clearly that the discrepancy of ligand 2 binding observed in figure 7 b corresponds solely to this evolution of the 1 group binding This interaction is initially approximately 100 kcal mol greater than the phosphone interaction with the solvent a feature also observed in ligand 1 Between 600 ps and 900 ps the former lost 50 kcal mol while the latter gained 50 kcal mol ligand 2 is not bound to cavity 3 anymore After 900 ps ligand 2 moves away from cavity 3 as it interacts significantly more with water than with Grb2 Interestingly the sum of both interact
219. onible et permet de mesurer in vitro de fa on fiable et relativement simple le pouvoir d inhibition d un produit donn sur le fonctionnement d un domaine SH2 quelconque dans l organisme Au moment de d buter cette th se des mesures obtenues par ce protocole pour un nombre significatif d inhibiteurs de Grb2 taient d j publi es 69 78 83 L tude th orique de Vinhibition Grb2 SH2 inclut la mise en place d un protocole permettant d estimer l affinit d un ligand donn toutefois la valeur d un tel protocole serait tr s limit e s il n tait pas possible de valider exp rimentalement ses r sultats La pr sence en parall le d un protocole exp rimental robuste fournissant directement des valeurs d affinit correspond ce niveau la meilleure des situations Ajoutons enfin que parmi les quipes de chercheurs impliqu es dans ACI ayant permis la conduite de cette th se sont pr sentes diff rentes quipes d exp rimentateurs aptes proc der aussi bien la synth se chimiques d inhibiteurs potentiels qu la mesure de leur affinit par le protocole ELISA C est un point extr mement positif aussi bien dans ce qu il implique de collaboration entre chercheurs issus d horizons scientifiques diff rents qu au niveau de la vari t des approches qui pourront tre entreprises pour am liorer la compr hension de la cible d tude Diff rentes techniques exp rimentales autres que le pr
220. ord D Roller P P and Burke Jr T R Protein tyrosine phosphatase inhibition by a peptide containing the phosphotyrosyl mimetic O malonyl tyrosine OMT Biochemical and Biophysical Research Communications 209 1995 817 822 161 Ye B Akamatsu M Shoelson S E Wolf G Giorgetti Peraldi S Yan X J Roller P P and Burke T R L O 2 malonyl tyrosine A new phosphotyrosyl mimetic for the preparation of Src homology 2 domain inhibitory peptides Journal of Medicinal Chemistry 38 1995 4270 4275 162 Burke T R Ye B Akamatsu M Ford H Yan X J Kole H K Wolf G Shoelson S E and Roller P P 4 O 2 2 fluoromalonyl L tyrosine A phosphotyrosyl mimic for the preparation of signal transduction inhibitory peptides Journal of Medicinal Chemistry 39 1996 1021 1027 163 Burke Jr T R Yao Z J Zhao H Milne G W A Wu L Zhang Z Y and Voigt J Enantioselective synthesis of nonphosphorus containing phosphotyrosyl mimetics and their use in the preparation of tyrosine phosphatase inhibitory peptides Tetrahedron 54 issue 34 1998 9981 9994 Principaux r sultats Dynamique mol culaire sur deux complexes de r f rence Cette partie r sume les principaux points de l article 1 Role of water molecules for binding inhibitors in the SH2 domain of Grb2 a molecular dynamics study On trouvera le texte de cette publication dans la section suivante Situation initiale De fa on classique une structure c
221. ormations that have the best steric complementarity Figure 8 Enrichment curve of SHEF as measured in the validation experiment depending on the amount of applied filtering Reference results are the GOLD rankings which were used to define a target subset of 1414 structures referred to as hits out of the starting 8383 The multiple target rankings were used in both cases i e the rank of each molecule is the best rank amongst the 1P8D 1PQ6 and 1PQ9 classifications The two dotted curves represent random selection and perfect correlation in which SHEF would reproduce GOLD results perfectly thus the filtering efficiency E for a filtering ratio f can be measured as the relative y position of the SHEF enrichment curve between these two filtering ratio 0 10 20 30 40 50 60 70 80 90 100 1400 100 90 2 1200 80 TD 1 PR 2 me S 60 2 800 T O 2 Cc S 600 40 o E as oO 8 400 30 2 20 200 10 0 0 8000 7000 6000 5000 4000 3000 2000 1000 0 number of molecules Table 2 Evolution of the GOLD hits subsets population when applying 10 and 50 SHEF based filtering The groups defined on figure 6 are studied separately while regarding SHEF filtering the results for each of the 3 target conformations are presented as well as those using the multiple target consensus ranking Note the multiple target all hits results bottom right can be measured directly on the figure 8 curve Table contents pop
222. orrespondante est inconnue et il n est m me pas possible de d terminer s il y en a une Pour ces raisons on estime souvent que l effort investi dans la g n tique doit pr sent tre tendu en direction de l interactome En effet si la g n tique permet d identifier de nombreuses pathologies elle reste souvent impuissante d s lors qu il s agit de les corriger Si la th rapie g nique est un domaine de recherche plus connu la mise au point de m dicaments ciblant l interactome est galement un axe prometteur en particulier sur le plan pharmaceutique 73 Le travail de cette th se se situe dans ce dernier domaine 12 Importance des simulations num riques L informatique une r volution scientifique Dans le pass les mod les physiques th oriques de la mati re ne pouvaient tre v rifi s que dans des circonstances exp rimentales particuli res dans lesquelles des variables physiques isol es peuvent tre mesur es D une fa on g n rale la mise au point de mod les th oriques par le scientifique r sulte classiquement d une approche r ductionniste de la nature Le point faible d une telle d marche est que les mod les valid s obtenus seront par nature r ducteurs la complexit n est pas accessible L informatique a t une r volution scientifique car elle a permis la complexit d tre en partie accessible scientifiquement La possibilit de conduire des calculs ten
223. ors mis en place rapidement sur une dur e de cinq ans et dot d un budget de 1 5 milliard d euros Cependant le plan Cancer est principalement ax sur des mesures de pr vention et de d pistage effectivement prioritaires sur le court terme au niveau de la sant publique les actions de recherche centralis es autour de canc ropoles pilot s au niveau minist riel sont d une importance toute relative et comprennent au final peu de projets th oriques ou orient s vers l aspect pharmaceutique Ainsi la part de financement des recherches de nouveaux m dicaments anti cancer dans le plan Cancer ne s l ve qu 10 millions d euros fin 2004 Ces progr s sont particuli rement flagrants en ce qui concerne le cancer du sein qui reste le plus fr quent pour les femmes et dont le taux de mortalit associ a baiss plus que de moiti de 1980 2000 Maladie honteuse jusque dans la mort N emploie t on pas syst matiquement la formule pudique mort des suites d une longue maladie sur les faire part de d c s Les organismes de recherche publics tels que l Inserm ou le CNRS d j impliqu s dans les projets du plan Cancer participent galement de fa on ind pendante la lutte contre le cancer soit en finan ant directement des projets de recherche soit en initiant et participant activement des projets financ s de fa on externe Cette th se r sulte ainsi de la mise au point en 2002 d une AC
224. ossibilities 2 the structure of biomolecules is very complex particularly regarding flexibility and induced fit effects and 3 the efficient simulation of ligands binding capabilities related to free energy calculations is a difficult task In another interesting opinion Kubinyi indicates that problems could rather originate mostly from oversimplifications and wrong assumptions made by scientists performing in silico research relying too much on computer results and forgetting that proper knowledge and expertise the in cerebro research is always the first requirement 61 While there is a large consensus around the directions indicated by Shoichet and others that motive most of the propositions for enhancing current docking methods 62 experience made us to focus more on the importance of some issues raised by Kubinyi Docking programs as well as SBDD techniques are very likely to give the inexperienced user a false sense of security by controlling and visualizing directly what happens on the screen in structural details it is easy to assume that there is no problem as long as there is no sign of the contrary This can be a fatal mistake as above all molecular modelling requires rigor and prudence More specifically setting up a docking simulation without carefully checking how the internals of the program work and how this could be related to the results is often a straight road to more or less strange failures The fi
225. otentiel d un tel flux de donn es Parmi les m thodes num riques dont il est question ici la plupart ont pour origine la chimie th orique Pour l expert dans ce domaine les recherches en biologie et en pharmacologie constituent des champs d application passionnants de par la nature des progr s scientifiques qu ils conditionnent et du fait qu ils se situent toujours a la limite des progr s technologiques De nombreux d fis sur les plans th orique aussi bien que technique dans ce secteur de plus en plus interdisciplinaire attendent d tre relev s Ce manuscrit constitue une tude th orique de l inhibition de l activit du domaine SH2 de la prot ine Grb2 par la liaison de ligands La premi re partie d crit le contexte de ce travail qui se situe plus g n ralement travers le ciblage de Grb2 SH2 parmi les multiples approches de lutte contre le cancer de m tude du vivant l chelle mol culaire domaine interdisciplinaire par excellence et plus pr cis ment de la recherche pharmaceutique La seconde partie s emploiera d tailler les connaissances disponibles sur la cible au moment de d buter ce travail en incluant les avanc es qui sont apparues en parall le a ce travail On trouvera dans la troisi me partie un r sum des principaux r sultats obtenus au cours de cette th se Ceux ci se classent selon trois approches m thodologiques distinctes celle de la dynamique mol culaire du do
226. otocole ELISA ont galement t propos es afin de caract riser l inhibition de la cible Grb2 SH2 84 86 Bien que leur emploi ne semble pas pr f rable si l on cherche obtenir des valeurs d affinit quantitatives elles peuvent se r v ler utiles dans un contexte purement qualitatif par exemple pour identification d inhibiteurs par screening Un tel cas de figure est videmment bien plus favorable que celui qui requiert la construction par homologie d un mod le de structure dont la qualit conditionnera celle de toutes les simulations ult rieures qui l emploieront 28 Accessibilit th orique Nous venons de voir que Grb2 SH2 s av rait accessible sur le plan exp rimental de par la disponibilit des structures exp rimentales n cessaires a l emploi de m thodes de la mod lisation mol culaire ainsi qu au niveau des exp riences qui pourront tre la fois conduites de fa on compl mentaire aux simulations et afin de valider celles ci Qu en est il de l accessibilit de Grb2 SH2 au niveau des simulations elles m mes x Tout d abord d une fa on g n rale nous avons mod liser des complexes biologiques constitu s d une prot ine li e un inhibiteur Un syst me mol culaire d une telle taille ne peut pas tre tudi par toutes les m thodes de la chimie th orique il peut toutefois tre trait par un certain nombre d entre elles Dans le cadre de cette tude nous a
227. owenstein E J Daly R J Batzer A G Li W Margolis B Lammers R Ullrich A Skolnik E Y Bar Sagi D and Schlessinger J The SH2 and SH3 domain containing protein GRB2 links receptor tyrosine kinases to ras signaling Cell 70 issue 3 1992 431 442 34 Chardin P Cussac D Maignan S and Ducruix A The Grb2 adaptor FEBS Letters 369 1995 47 51 35 Maignan S Guilloteau J P Fromage N Arnoux B Becquart J and Ducruix A Crystal structure of the mammalian Grb2 adaptor Science 268 issue 5208 1995 291 293 40 36 Salcini A E McGlade J Pelicci G Nicoletti I Pawson T and Pelicci P G Formation of Shc Grb2 complexes is necessary to induce neoplastic transformation by overexpression of Shc proteins Oncogene 9 issue 10 1994 2827 2836 37 Xie Y Li K and Hung M C Tyrosine phosphorylation of Shc proteins and formation of Shc Grb2 complex correlate to the transformation of NIH3T3 cells mediated by the point mutation activated neu Oncogene 10 issue 12 1995 2409 2413 38 Buday L Membrane targeting of signalling molecules by SH2 SH3 domain containing adaptor proteins Biochimica et Biophysica Acta 1422 1999 187 204 39 Tari A M and Lopez Berestein G Grb2 a pivotal protein in signal transduction Seminars in Oncology 28 issue 5 suppl 16 2001 142 147 40 Li N Batzer A Daly R Yajnik V Skolnik E Chardin P Bar Sagi D Margolis B and Schlessinger J Guanine nucleotide releasin
228. probabilit faible d exclusion de mol cules valables Les plus connues sont celles de Lipinski rule of five 79 La mesure et la pr diction de la toxicit sont des probl mes plus complexes mais qui doivent tre abord s en particulier afin d am liorer la qualit d une banque de mol cules g n rique Mise au point et optimisation de compos s actifs sp cifiques leads On d finit un lead t te de s rie comme une mol cule ou structure de base d un ensemble de mol cules qui non seulement pr sente une activit significative pour la cible hit mais qui en plus est s lective pour celle ci lors d un test exp rimental Cette d finition est arbitraire et peut varier d un protocole de s lection un autre De fa on g n rale partir d un nombre important de mol cules de d part on effectue au moins trois filtrages successifs le premier sert identifier les hits le second s lectionner les leads enfin le troisi me niveau de filtrage correspond la s lection ventuelle apr s optimisations des leads d un ou plusieurs compos s candidats pour les tests cliniques Cette derni re tape correspond en g n ral l emploi de techniques manuelles contrairement aux pr c dentes pour lesquelles le filtrage peut tre plus ou moins automatis Leads Clinical iid A optimizan candidate 26 ee ere ae Hits gt Hit generation gt 3 i
229. que d int r t pour la cible sur la base des mod les les plus simples Globalement on cherche s assurer que lorsqu une mol cule est exclue une tape donn e on a pour ce faire d pens le minimum de temps de calcul th orique Le temps de calcul total est ainsi optimis d s lors que les diff rentes techniques et param tres sont bien choisis VSM G Virtual Screening Manager for computational Grids constitue fondamentalement une infrastructure logicielle permettant d impl menter ais ment un tel protocole de screening virtuel et ce de fa on totalement automatis e 57 Dans ce contexte l h t rog n it des programmes disponibles induit une grande complexit et peut s av rer handicapante Un des principaux objectifs fix VSM G est donc de faciliter autant que possible la conception par l utilisateur d un protocole multi tapes efficace Pour ce faire le fonctionnement du funnel est automatis et transparent tandis que l interface graphique est con ue dans une optique de simplicit D veloppements en cours et objectifs La conduite d une simulation avec VSM G s organise globalement en trois tapes la pr paration de la base des ligands la pr paration de la cible qui peut tre repr sent e par un nombre quelconque de structures et enfin le funnel l heure actuelle cette derni re partie est limit e trois tapes de docking et les deux autres sont de simples collections
230. rcialisation ce chiffre inclut donc toutes les d penses li es au d veloppement de mol cules qui finalement n ont pas pu tre commercialis es Le laps de temps entre le d marrage du processus de recherche et la commercialisation lorsque la recherche pharmaceutique est couronn e de succ s est d environ 15 ans Dans le cas du Taxol la d couverte de la mol cule remonte 1963 tandis que l autorisation de mise sur le march s est faite en 1990 82 La d couverte de m dicaments suit clairement un processus en entonnoir au cours duquel chaque tape de moins en moins de mol cules satisfont l ensemble des crit res successifs et dont le taux d chec est au final particuli rement important Si la phase de recherche de hits peut comprendre de plusieurs milliers plusieurs millions de mol cules il n est m me pas certain qu une seule de ces mol cules passe la phase des essais pr cliniques Lorsqu une mol cule arrive ce stade la probabilit qu elle doive tre abandonn e suite aux essais cliniques est encore lev e Une tude statistique mentionne 12 succ s c est dire 12 mol cules dont l effet th rapeutique suppos a t prouv par les tests cliniques sur 209 mol cules candidates 83 Il est de plus crucial de noter que les co ts croissent exponentiellement d une tape l autre si bien que les progr s conceptuels des phases pr cliniques durant lesquelles les m thodes de sim
231. rcoma virus RMSD root mean square deviation RTK receptor with endowed tyrosine kinase activity SH2 Src homology 2 SH3 Src homology 3 She Src homology 2 domain containing Sos son of sevenless protein Src sarcoma viral oncogene homolog Introduction Several signalling and adaptor proteins have an SH2 Src homology 2 domain which regulates the activity and specificity of kinases and binds phosphotyrosine pTyr residues of growth factor receptors 1 2 Such a binding triggers a cascade of metabolic events resulting into nuclear transcription 3 SH2 domains could thus constitute an essential target for the design of small molecules that upon competing with the binding of pTyr residues would act as regulators of signal transduction An important pharmaceutical interest arises from the fact that over expression of such proteins corresponding to a deregulation of the pathway and thus of cell cycle progression is taking part in processes of cellular hyper proliferation and cancer development 4 7 One of the most important SH2 domain encompassing proteins is Grb2 which is an adaptor composed of one SH2 domain linking two SH3 domains 8 10 It has a significant activity as a linker with numerous receptors of the cell membrane 11 also involved in fibroblast 12 and neoplastic transformations 13 Complexed with Sos the exchange factor of Ras the SH2 domain of Grb2 binds directly to auto phosphorylated RTKs such as EGF R 1
232. rentes familles de iad Prisha HR Rs us 010 323 Phase RA stroke diabetes Scios inc domaines SH2 constitue donc un sujet de recherche 010 469 Phase I RA Crohn s Sais nc ai SX 6 Pa POKI UCN 01 Phase MI cancer Kyowa Hakko Kogyo d int r t pharmaceutique particuli rement suivi m3 ems se gee 19 53 myelogenous leukemia RA rheumatoid arthritis and ACS acute coronary syndromes Inhibitors are of tao types monoclonal antibodies mAs gt which are directed at the extracefiuder domain of various receptor tyrosine kinases and smal molecuie inhibitors Dans environ 25 des cas de cancers humains des prot ines exprimant le signal Ras sont pr sentes sous des formes alt r es dans lesquelles elles ne sont plus r gul es normalement permettant ainsi aux cellules canc reuses d acqu rir leur autonomie fonctionnelle 54 La recherche d inhibiteurs de Ras MAPK suscite pour cette raison un tr s grand int r t 6 Dans ce contexte nous avons tudi plus particuli rement Vinhibition du domaine SH2 de Grb2 Il a t d montr que la prolif ration anarchique des cellules caract risant certaines formes de leuc mie 55 ainsi que les cancers du sein des ovaires du poumon et de la prostate 56 58 est reli e 4 une surexpression de HER2 ErbB2 un analogue du complexe Grb2 EGF 59 Des inhibiteurs de Grb2 SH2 sont galement susceptibles de r duire non seulement la capacit de multiplication mais aussi la mobili
233. responding surfaces still provide a good level of detail Additionally this process can be done once and for all for each protein and ligand conformer Afterwards evaluating the surface complementarity between a target active site and a ligand is performed through simple and efficient operations 48 specific to spherical harmonics algebra This very fast procedure is performed with the SHEF program 49 which identifies and scores the geometrically optimal orientation of each ligand conformation for the target These techniques are described in depth by Cai et al 35 36 49 In this paper we study a VSM G operated screening funnel using MSSH SHEF followed by flexible docking using GOLD 50 51 Such an approach involves using SHEF results to filter out part of the input ligand database before proceeding to the second funnel step relying on GOLD In this proof of concept study we did no such filtering all molecules of the test set are evaluated with both techniques in order to simulate the screening funnel for all levels of filtering between the two steps Target preparation The liver X receptors LXRs 52 represent attractive targets for the development of new therapeutic agents for treating multiple especially cardiovascular diseases 53 Several structures of the ligand binding domain of LXR co crystallized with various ligands have been determined by X Ray crystallography Reports on structural analysis reveal great plasticity of the lig
234. rical binding free energy function Journal of Computational Chemistry 19 issue 14 1998 1639 1662 128 Rarey M Kramer B Lengauer T and Klebe G A fast flexible docking method using an incremental construction algorithm Journal of Molecular Biology 261 issue 3 1996 470 489 129 FlexX http www biosolveit de FlexX 130 ClauBen H Buning C Rarey M and Lengauer T FlexE efficient molecular docking considering protein structure variations Journal of Molecular Biology 308 issue 2 2001 377 395 131 Stahl M and Rarey M Detailed analysis of scoring functions for virtual screening Journal of Medicinal Chemistry 44 issue 7 2001 1035 1042 132 Hoffmann D Kramer B Washio T Steinmetzer T Rarey M and Lengauer T Two stage method for protein ligand docking Journal of Medicinal Chemistry 42 issue 21 1999 4422 4433 133 Rarey M Kramer B and Lengauer T The particle concept Placing discrete water molecules during protein ligand docking predictions Proteins Structure Function and Bioinformatics 34 1999 17 28 134 Cavasotto C N and Abagyan R A Protein flexibility in ligand docking and virtual screening to protein kinases Journal of Molecular Biology 337 2004 209 225 135 Bursulaya B D Totrov M Abagyan R and Brooks III C L Comparative study of several algorithms for flexible ligand docking Journal of Computer Aided Molecular Design 17 2003 755 763 136 MolSoft http www
235. riment Taking as reference the SHEF consensus ranking we plotted the variation of the population of GOLD hits as a function of the filtering ratio The resulting curve is shown in figure 8 together with the enrichment curves that would result from random selection and from the ideal case where the 1 414 hits are all ranked before the other 6 969 molecules A clear enrichment is observed on all ranges of filtering There is still much room for improvement but present SHEF performance is interesting considering that SHEF and GOLD are not in the same league in terms of speed and precision In the virtual screening context if the number of molecules to screen is too high for GOLD using available computing power SHEF could provide a rational solution for decreasing the number of candidates molecules without limiting too much the chances of finding novel hit compounds for a given target Correlation between SHEF efficiency and the nature of the protein ligand binding mode We will now focus on results for two particular filtering ratios chosen arbitrarily 0 1 low filtering 90 of molecules retained and 0 5 half the molecules filtered out In order to determine whether particular families of molecules could influence SHEF filtering efficiency the variation of all hits populations as defined in figure 5 for the four possible SHEF rankings on the 1P8D 1PQ6 1PQ9 targets and multiple target consensus was collected The results are shown in table 2
236. ristallographique de r solution suffisante d un complexe de type prot ine ligand correspond a un point de d part potentiel pour une tude par les techniques de la mod lisation mol culaire Le mod lisateur d s lors que son travail lui permet d estimer l affinit d un ligand donn pour la cible peut sugg rer de nouvelles structures potentiellement int ressantes aux biochimistes ou au contraire les dissuader de synth tiser une mol cule donn e Dans un tel contexte une approche d optimisation peut tre constitu e par des simulations par dynamique mol culaire Notre point de d part est constitu au niveau purement structural de la structure exp rimentale d un complexe Grb2 SH2 ligand Ce ligand de r f rence est not ligand 1 dans l article on tudie galement un d riv not ligand 2 Les informations structurales sont compl t es par des donn es de tests biologiques ligand 1 a une affinit nanomolaire lors de tests in vitro sur Grb2 SH2 tandis que les m mes tests ont pu montrer que l affinit de ligand 2 s av rait n gligeable en comparaison HN N NH 2 L H 2 O CH NH o Ma OPO NH A o Ligand 1I haut et ligand 2 bas Diagramme des interactions du complexe Grb2 SH2 ligand 1I structure PDB 1JYQ 47 Ant rieurement au travail effectu dans le cadre de cette th se des simulations par dynamique mol culaires avaient t effectu es sur les complexes de Grb2 SH2 avec l
237. rmiani G Castelli C Dalerba P Mortarini R Rivoltini L Marincola F M and Anichini A Cancer immunotherapy with peptide based vaccines What have we achieved Where are we going Journal of the National Cancer Institute 94 issue 11 2002 805 818 38 Adam J K Odhav B and Bhoola K D Immune responses in cancer Pharmacology amp Therapeutics 99 2003 113 132 39 Berzofsky J A Terabe M Oh S Belyakov I M Ahlers J D Janik J E and Morris J C Progress on new vaccine strategies for the immunotherapy and prevention of cancer Journal of Clinical Investigation 113 issue 11 2004 1515 1525 40 Rosenberg S A Yang J C and Restifo N P Cancer immunotherapy moving beyond current vaccines Nature Medicine 10 issue 9 2004 909 915 41 Nam N H and Parang K Current targets for anticancer drug discovery Current Drug Targets 4 issue 2 2003 159 179 42 Dancey J and Sausville E A Issues and progress with protein kinase inhibitors for cancer treatment Nature Reviews Drug Discovery 2 2003 296 313 20 43 Laurence V and Espi M Effets secondaires de la chimioth rapie Le G n raliste 2118 2001 44 Boutin J A Tyrosine proteine kinase inhibition and cancer International Journal of Biochemistry 26 issue 10 11 1994 1203 1226 45 Awada A Cardoso F Atalay G Giuliani R Mano M and Piccart M J The pipeline of new anticancer agents for breast cancer treatment in 2003 Critical Revi
238. rof Daniel CANET Prof Christiane GARBAY Dr Bernard MAIGRET Invit s Dr Peter BLADON Dr Gilles MOREAU niversit Ren Descartes Paris V niversit Louis Pasteur Strasbourg niversit Ren Descartes Paris V U U Universit Henri Poincar Nancy I U U niversit Henri Poincar Nancy I Interprobe Chemical Services Glasgow Ecosse 30 avenue Jean Jaur s 94220 Charanton Mes remerciements les plus chaleureux vont mon directeur de th se Bernard pour son encadrement sa patience face mon fonctionnement irr gulier mais surtout pour avoir fait en sorte que je termine cette th se en ayant une vision de la recherche scientifique qui a d pass le simple stade de l int r t et de la curiosit Un grand merci galement Nohad Christiane Peter et Gilles pour leur soutien et leur gentillesse Je tiens galement saluer l ensemble des membres du laboratoire eDAM ainsi qu tous ceux avec qui j ai travaill sp cialement mes compagnons d armes Alexandre Jean Paul Matthieu J r me Adil Amel Yesmine Muhannad et Werner Bon courage ceux qui n en ont pas encore fini et pareillement ceux qui sont partis vers d autres aventures Enfin je garde beaucoup de reconnaissance pour ma famille et mes amis qui ont su rester pr sents m me lorsque j tais un peu d connect par mon travail et la distance Merci en particulier mes parents Sophie Pascal Estelle et Catheri
239. s analogues en super structures 10 Du g nome a l interactome description du fonctionnement d un organisme vivant Du prot ome l interactome Les interactions entre prot ines peuvent tre identifi es regroup es au sein de voies de signalisation et repr sent es sch matiquement sous la forme de graphes d interactions 65 66 WNT gt gt Disheveled Anti growth factors 4 e g TGFB GSK 3p pa TCF TGFER Celis Cadherin En pcan WE cmmnmiter m AA Vue 4 sa Ecm gt Cintes 1 Cyel D CDKA H p15 Smads Sre cee gt On i L ape Rb HPV E7 SATA it R p27 lt i tY Be 1 j Mos MKKs JNKs gt Jun Cycl E CDK2 p21 lt Growth Factors Grb2 t t t ae 4 NA damage 0 g TGFa gt ATK P sos gt Ras gt Rat gt MEK gt MAPK gt MAPK gt EIK gt Fos gt Changes Coli gt sensi ot Z Ji wi Mad Max in Gene_ Proliferation fs MEKK yo re H Cen cycled 553 Hormones CdC42 gt Rao gt Rho Gea e mam amma lt 2 e g ronson D atte gt PU 1 ARF MOM2 1 e g Estrogen NHR e g ER i n I f Moi NEB NF KB aa a TMM gt Mitochondria l as A Bel 2 een gt DS MT i Akka iB Gentine ar Beth Stat 3 5 4 DNS t MELC 4 Decoy R 4 Death Factors 4 e g FasL Cytokines _ e g IL 36 Sch ma des principales voies de signalisation connues intervenant au niveau cellula
240. sID 15 144 Cai W Xu J Shao X and Maigret B SHEF An efficient approach for virtual screening using coefficients of spherical harmonics surfaces Submitted for publication 2006 145 Ritchie D W and Kemp G J L Fast computation rotation and comparison of low resolution spherical harmonic molecular surfaces Journal of Computational Chemistry 20 issue 4 1999 383 395 146 Cai W Shao X and Maigret B Protein ligand recognition using spherical harmonic molecular surfaces towards a fast and efficient filter for large virtual throughput screening Journal of Molecular Graphics and Modelling 20 issue 4 2002 313 328 147 Yamagishi M E Martins N F Neshich G Cai W Shao X Beautrait A and Maigret B A fast surface matching procedure for protein ligand docking Journal of Molecular Modeling 12 issue 6 2006 965 972 148 Casanova H and Berman F Parameter sweeps on the grid with APST in Grid Computing eds F Berman G Fox and T Hey 2003 John Wiley amp Sons p 773 787 149 van der Raadt K Yang Y and Casanova H Practical divisible load scheduling on grid platforms with APST DV in Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium IPDPS 05 Papers 2005 150 APST A Parameter Sweep Tool http grail sdsc edu projects apst 151 Taylor J S and Burnett RM DARWIN A program for docking flexible molecules Proteins Structure Function and Ge
241. scientific world particularly management concerns the industry preferring to favor marketing over research and the raising standards in medical safety However most probably limitations of research techniques should also be taken into account carefully Drug research methods can be divided into two non exclusive classes of approaches the well established classic experimental drug discovery 14 and the ever growing ensemble of theoretical and computational methods now integral parts of today s drug design following the most significant advances and emerging concepts of modern scientific research 15 16 A crucial paradigm of modern drug discovery is that most of the interesting targets that were reachable by the protocols successful 20 30 years ago have probably been investigated The corresponding marketed drugs encompass a very limited part of the interactome 500 biomolecules 1 Studying other targets that could provide the basis for discovering innovative compounds e g anti cancer specific 17 at the genomics and proteomics level is significantly more demanding and requires the use of computer aided techniques 18 Data mining and visualisation techniques 19 associated with statistical analysis approaches are without any doubt mandatory here Molecular modelling methods 20 are also very important 21 as they directly simulate the possible relationships between molecular structure and biological function which is the central
242. screening Nature Reviews Drug Discovery 1 2002 882 894 10 Abagyan R and Totrov M High throughput docking and lead generation Current Opinion in Chemical Biology 5 2001 375 382 11 Xu H and Agrafiotis D K Retrospect and prospect of virtual screening in drug discovery Current Topics in Medicinal Chemistry 2 2002 1305 1320 12 Krovat E M and Langer T Impact of scoring functions on enrichment in docking based virtual screening An application study on renin inhibitors Journal of Chemical Information and Computer Sciences 44 issue 3 2004 1123 1129 13 Huo S Wang J Cieplak P Kollman P A and Kuntz I D Molecular dynamics and free energy analyses of Cathepsin D inhibitor interactions Insight into structure based ligand design Journal of Medicinal Chemistry 45 issue 7 2002 1412 1419 14 Jenwitheesuk E and Samudrala R Improved prediction of HIV 1 protease inhibitor binding energies by molecular dynamics simulations BMC Structural Biology 3 2003 15 Alonso H Bliznyuk A A and Gready JE Combining docking and molecular dynamic simulations in drug design Medicinal Research Reviews 26 issue 5 2006 531 568 16 Waszkowycz B Perkins T D J Sykes R A and Li J Large scale virtual screening for discovering leads in the postgenomic era IBM Systems Journal 40 issue 2 2001 360 376 17 Bleicher K H B hm H J M ller K and Alanine A I Hit and lead generation beyond high throughput scr
243. se of substructures associated with poor chemical stability or toxicity 3D structure generation fragment based 3D structure generation program the fragment database gt 10 000 structures can be enhanced extended by the user CORINA 30 which shares the same concept can be used alternatively if available post processing options protonation at pH 7 conformational sampling using OMEGA 27 if available Chart 2 Current VSM G features target preparation Handling of protein structures automatic checking and cleaning of input PDB files with respect to PDB standards 31 protein structures can be checked using the MOLPROBITY server 32 correction of protonation states link to the H web server 33 relaxation of the hydrogen positions upon energy minimization link to the STING 34 web based suite of programs for data mining Receptor definition holoproteins receptor assumed to be located at the center of mass of the ligand apoproteins generation of an interactive protein 2D map with MSSH 35 36 and VMD 37 for picking up surface receptors manual definition of receptors can be imported from VMD selections and exported to funnel modules handling of resident water molecules potentially useful with some docking programs 38 Multiple target conformations management handling of multiple X ray structures enrichment through MD sampling 15 39 using VMD and NAMD 40
244. senescence implications for immortalization of normal human diploid fibroblasts Molecular and Cellular Biology 9 1989 3088 3092 21 Shay J W and Bacchettti S A survey of telomerase activity in human cancer European Journal of Cancer 33 issue 5 1997 787 791 22 Levine A J p53 the cellular gatekeeper for growth and division Cell 88 1997 323 331 23 Heidelberger C Freeman A E Pienta R J Sivak A Bertram J S Casto B C Dunkel V C Francis M W Kakunaga T Little J B and Schechtman L M Cell transformation by chemical agents a review and analysis of the literature A report of the U S Environmental Protection Agency Gene Tox Program Mutation Research 114 issue 3 1983 283 385 24 Ashby J and Tennant R W Definitive relationships among chemical structure carcinogenicity and mutagenicity for 301 chemicals tested by the US NTP Mutation Research 257 issue 3 1991 229 306 25 Lynch H T Fusaro R M and Lynch J Heredity cancer in adults Cancer Detection and Prevention 19 issue 3 1995 219 233 26 Lichtenstein P Holm N V Verkasalo P K Iliadou A Kaprio J Koskenvuo M Pukkala E Skytthe A and Hemminki K Environmental and heritable factors in the causation of cancer Analyses of cohorts of twins from Sweden Denmark and Finland New England Journal of Medicine 343 issue 2 2000 78 85 27 Duesberg P H and Vogt P K Differences between the ribonucleic acids of transforming and
245. servent d terminer quelles parties de l ADN doivent tre transcrites et quelles parties sont ignor es d finissant ainsi les segments sur les chromosomes correspondant aux g nes On peut noter qu alors que dans l organisme humain la taille moyenne des prot ines est de 300 r sidus on est tr s loin d observer 20 soit environ 10 s quences peptidiques possibles L ordre de grandeur des s quences significatives est estim 10 De fa on remarquable l adoption par une prot ine d une conformation anormale peut changer sa fonction biologique et aboutir des pathologies particuli rement graves C est le cas de la maladie de la vache folle dans laquelle une prot ine du cerveau un prion subit un changement conformationnel se propageant aux autres prions formant des aggr gats la source d un processus d g n ratif mortel La conformation d une prot ine peut tre caract ris e par I orientation relative des diff rentes unit s du squelette On distingue pour ce faire les h lices a caract ris es par des liaisons H intra cha ne et repr sent es graphiquement sous la formes de cylindres des feuillets 8 correspondent l association parall le ou anti parall le de deux cha nes distinctes par des liaisons H inter cha nes par des fl ches plates On appelle tour B ou Q lorsque l angle est tr s marqu le passage d une forme g om trique l autre NH N N
246. simulation application The Adun simulator Journal of Computational Chemistry 26 issue 15 2005 1647 1659 87 Phillips J C Braun R Wang W Gumbart J Tajkhorshid E Villa E Chipot C Skeel R D Kal L and Schulten K Scalable molecular dynamics with NAMD Journal of Computational Chemistry 26 issue 16 2005 1781 1802 88 van der Spoel D Lindahl E Hess B Groenhof G Mark A E and Berendsen H J C GROMACS fast flexible and free Journal of Computational Chemistry 26 issue 16 2005 1701 1718 89 Adun molecular simulation project http diana imim es Adun 90 Amber Assisted Model Building with Energy Refinement http amber scripps edu 91 CHARMm Chemistry at HARvard Molecular Mechanics http www accelrys com products charmm index html 92 GROMACS GROningen MAchine for Chemical Simulation http www gromacs org 93 NAMD Not Another Molecular Dynamics http www ks uiuc edu Research namd 94 Mason J S Good A C and Martin EJ 3 D pharmacophores in drug discovery Current Pharmaceutical Design 7 issue 7 2001 567 597 95 Cho A E Guallar V Berne B J and Freisner R Importance of accurate charges in molecular docking Quantum mechanical molecular mechanical QM MM approach Journal of Computational Chemistry 26 issue 9 2005 915 931 96 Freisner R A Murphy R B Repasky M P Frye L L Greenwood J R Halgren T A Sanschagrin P C and Mainz D T Extra
247. sur le cancer Ligue contre le cancer http www ligue cancer asso fr F d ration Nationale des Centres de Lutte Contre le Cancer http www fnclec fr Rapport de la commission d orientation sur le cancer 2002 http www sante gouv fr htm dossiers cancer Mission interminist rielle pour la lutte contre le cancer http www plancancer fr Cancer Medicine 5th edition BC Decker http www ncbi nlm nih gov books by fcgi call bv View ShowTOC amp rid cmed TOC amp depth 10 American Cancer Society http www americancancersociety org National Comprehensive Cancer Network http www nccn org R sum de l ACI qui a permis de financer cette th se Le but de ce projet est de concevoir des agents pharmacologiques cibl s capables d inhiber des voies de signalisation d r gul es par la pr sence de prot ines cod es par des oncog nes et conduisant la formation de cancers L approche choisie est nouvelle mais donne d j des r sultats Elle consiste inhiber les interactions entre ces prot ines et leurs effecteurs ou entre des prot ines situ es en aval dans la voie d r gul e Les cibles s lectionn es sont situ es dans la voie de signalisation Ras d pendante impliqu e dans des cancers la fois nombreux et souvent fr quents ERBB2 sein ovaire Grb2 sein ovaire et certaines leuc mies comme la LMC RasGAP dans les cancers exprimant Ras oncog nique 30 des tumeurs 90
248. t et les capacit s de propagation de cellules canc reuses 45 60 61 Leur mise au point en tant qu ventuels m dicaments anti cancer est donc pleinement justifi e 1 12 13 62 64 Actuellement certains produits de la classe des inhibiteurs de farnesyltransferase actifs sur Ras MAPK et pouvant tre prescrits par voie orale sont test s cliniquement en tant que m dicaments anti cancer sp cifiques 27 Accessibilit exp rimentale Grb2 tout d abord est une biomol cule accessible pour l exp rimentateur les protocoles de cristallisation sont bien d finis 65 Des structures pour Grb2 SH2 libre 35 66 68 et complex avec diff rents ligands 68 74 sont disponibles dans la base PDB 75 76 aussi bien par RX 35 68 71 73 74 que par RMN 66 67 70 72 Cela constitue un argument de poids en faveur de l utilisation de Grb2 comme cible th rapeutique travers des tudes de mod lisation mol culaire En effet de telles tudes reposent sur le mod le de d part employ dans le cas de Grb2 un tel mod le sera d riv directement de structures exp rimentales dont on peut assurer la fiabilit Cristaux de Grb2 65 L accessibilit exp rimentale Grb2 SH2 d j garantie sur un plan purement structural est galement assur e au niveau de la grandeur physique qui quantifie l int r t ou non d un inhibiteur donn l affinit En effet un protocole standardis ELISA 77 est disp
249. t tre mentionn Une simulation num rique n est soumise aucune autre contrainte mat rielle que celle du temps de calcul n cessaire Le choix entre mesure exp rimentale et simulation num rique n a ainsi pas lieu d tre lorsque l exp rience est trop contraignante par exemple elle requiert des conditions de temp rature et de pression d licates elle peut s av rer dangereuse voire impossible Les simulations num riques peuvent aller encore plus loin en permettant de repousser artificiellement certaines contraintes physiques afin de mieux appr hender le comportement d un syst me L approche r ductionniste consiste fractionner un syst me complexe en sous syst mes plus simples jusqu ce que tous les sous syst mes puissent tre d finis de fa on satisfaisante par des mod les th oriques Pour valider un sous syst me donn pour lequel des hypoth ses th oriques sont avanc es il faut mettre en place une exp rience dans laquelle interviennent le moins possible de ph nom nes physiques externes au sous syst me d crit Cette contrainte est souvent particuli rement difficile satisfaire Les premiers mod les physiques des gaz parfaits ne pouvaient tre valid s que par des mesures sur des gaz rares et les mod les de cristaux devaient l tre avec des chantillons tr s purs S ils permettent bien d appr hender les lois fondamentales qui leur sont associ es ils ne permettent aucuneme
250. t T N Weissig H Shindyalov I N and Bourne P E The Protein Data Bank Nucleic Acids Research 28 issue 1 2000 235 242 68 Smith Schmidt T Banking on structures BiolT World 1 issue 8 2002 69 Goto S Okuno Y Hattori M Nishioka T and Kanehisa M LIGAND Database of chemical compounds and reactions in biological pathways Nucleic Acids Research 30 2002 402 404 70 Bohacek R S McMartin C and Guida W C The art and practice of structure based drug design a molecular modelling perspective Medicinal Research Reviews 16 1996 3 50 71 Dobson C M Chemical space and biology Nature 432 2004 824 828 72 Ellis R J and Minton A P Join the crowd Nature 425 2003 27 28 73 Chanda S K and Caldwell J S Fulfilling the promise drug discovery in the post genomic era Drug Discovery Today 4 2003 168 174 74 Anderson A C The process of structure based drug design Chemistry amp Biology 10 2003 787 797 75 Wlodawer A and Vondrasek J Inhibitors of HIV 1 protease A major success of structure assisted drug design Annual Review of Biophysics and Biomolecular Structure 27 1998 249 284 76 Lundstrom K Structural genomics of GPCRs Trends in Biotechnology 23 issue 2 2005 103 108 77 H nin J Maigret B Tarek M Escrieut C Fourmy D and Chipot C Probing a model of a GPCR ligand complex in an explicit membrane environment The human cholecystokinin 1 receptor Biophysical Journal 90 200
251. t dans lesquelles des ligands peuvent se lier partiellement sont connues 68 Cette caract ristique est encourageante dans l optique de la mise au point de ligands moins charg s et donc plus susceptibles de passer la barri re membranaire Les r gles de Lipinski pour citer les plus connues permettent d exclure de fa on relativement fiable des mol cules qui ne pourraient pas tre administr es oralement et ensuite se diffuser librement dans l organisme Anglicisme pratique d signant le caract re m dicamenteux d une mol cule sur le plan structural ce qui a trait entre autres au respect ou non des r gles de Lipinski C est dire un million de fois plus efficaces que la s quence peptidique de r f rence pY VNV charg e doublement n gativement et qui contient le r sidu pTyr L affinit augmente apr s phosphorylation mais seulement d un facteur 3 comparer aux quatres ordres de grandeur observ s en ce qui concerne la plupart des domaines SH2 Lorsque le r sidu Yo est phosphoryl on peut supposer que son interaction avec SAP SH2 devient alors comparable celle des r sidus voisins Cela contraste avec le cas de la plupart des ligands de domaines SH2 ciblant les deux cavit s du r cepteur dans lesquels le r sidu pY semble irrempla able 35 Recherche de ligands actifs in vivo strat gies pour Grb2 SH2 Le d veloppement de ligands actifs in vivo pour Grb2 SH2 peut se faire so
252. t un r seau routier on n en conna t pour le moment qu une partie des autoroutes sous la forme de voies de signalisation La plupart des interactions entre biomol cules sont inconnues l heure actuelle En effet plusieurs interactions peuvent tre en concurrence au sein du m me milieu extra ou intra cellulaire Ces consid rations peuvent tre cruciales pour des applications d ordre m dical 11 Constitution mol culaire d un organisme vivant La composition chimique d un organisme vivant ne saurait bien videmment se r duire aux acides nucl iques et aux prot ines Toutefois ces deux classes fondamentales de biomol cules sont la base du fonctionnement des organismes et suffisent caract riser le cadre de cette th se De fa on g n rale si l on exclut les biomol cules de taille importante acides nucl iques prot ines membranes la diversit en terme chimiques d un organisme vivant est limit e les organismes vivants les plus simples sont ainsi construits autour d une centaine de petites mol cules organiques 69 en comparaison du nombre de mol cules organiques th oriquement synth tisables estim 10 70 L espace biologique est de nombreux ordres de grandeur plus petit que l espace chimique 71 Un aspect important de la nature d un organisme vivant observ e sous l angle de sa constitution chimique doit n anmoins tre mentionn Les syst mes e
253. tale reste de 2 pTyr phosphotyrosine R CH D 0 PO 7 Tyr tyrosine R CH OH phosphonotyrosine R CH PO phosphinate isosteres R CH D CH PO X 111 112 Pmp phosphonomethyl phenylalanine R CH D CH PO 64 123 147 152 155 F gt Pmp phosphonodifluoromethyl phenylalanine R CH CF PO 123 153 156 1571 cmF carboxymethyl phenylalanine R CH CH2 COO 117 141 F cmF carboxydifluoromethyl phenylalanine R CH CF COO 141 158 pmF P malonyl phenylalanine R CH CH COO 132 159 Gla acide y carboxyglutamique R CH2 CH COO gt 143 146 OMT O malonyl tyrosine R CH O CH COO 160 161 FOMT fluoro O malonyl tyrosine R CH O CF COO 162 cmT carboxymethyl tyrosine R CH2 O CH2 COO7 163 Substituants pTyr employ s dans les inhibiteurs de Grb2 SH2 De gauche droite abr viation usuelle nom chimique formule r f rences Page suivante S lection d inhibiteurs de Grb2 SH2 R f rences 1 78 113 2 64 3 64 4 71 5 68 79 6 82 134 7 74 8 95 144 9 146 Abr viations Acc acide l aminocyclohexyl carboxylique hNp 3 2 hydroxynaphthalen 1 yl propyle mAZ m aminobenzylo carbonyle S R Achec acide cis 2 amino cyclohex 3 yl carboxylique naph D 2 naphthyl alanine meF methyl phenylalanine tBuAla tertiobutyl alanine AmF amino phenylalanine 37 ION 72 Je 491104 FJen OV VOVZ add z 4 I
254. tation du risque avec l ge et le long cycle de d veloppement de la maladie que l on observe g n ralement Le d veloppement du cancer est souvent li une surexpression de prot ines exprim es par les proto oncog nes mut s en oncog nes donc impliqu es lorsque leur concentration dans l organisme est appropri e dans les processus normaux de croissance cellulaire et aboutissant une multiplication incontr l e des cellules affect es Le cancer peut galement tre favoris par la mutation d un ou plusieurs g nes suppresseurs de tumeurs 14 15 aboutissant la diminution ou la disparition de biomol cules charg es de r parer ou d liminer les cellules endommag es La plupart des cancers combinent i http www inserm fr fr questionsdesante dossiers cancer http www inserm fr fr recherches etats_des_lieux att00002003 CANCER2002_v3 4 num pdf t www cnrs ft SDV cnrscancer html 4 ces deux causes une mutation pouvant en provoquer d autres par cascade ceci tant acc l r 4 chaque fois qu un g ne suppresseur de tumeurs est touch Cela aboutit 4 la formation d au moins une cellule canc reuse pour laquelle les signaux g n tiques qu elle contient et favorisant sa prolif ration pr valent suite un nombre critique de mutations un nombre trop faible de mutations tant r gul naturellement sur ceux codant pour sa r gulation Les diff rentes mutations n cessaires 4 la formation d
255. these representations can be found on figure 6 p 0 73 1 0 53 p 0 85 t 0 66 p 0 78 0 59 SHEF 1PQ6 S a a pr n SHEF 1PQ9 SHEF 1P8D a SHEF 1P8D b SHEF 1PQ6 c p 0 90 t 0 75 p 0 70 t 0 55 p 0 66 t 0 51 r Block population GOLD 1PQ6 GOLD 1PQ9 GOLD 1PQ9 GOLD 1P8D d GOLD 1P8D e GOLD 1PQ6 f p 0 22 t 0 15 p 0 27 t 0 19 p 0 15 t 0 11 p 0 23 t 0 16 pe SHEF 1P8D SHEF 1PQ6 SHEF 1PQ9 SHEF multi target GOLD 1P8D 9 GOLD 1PQ6 h GOLD 1PQ9 i GOLD multi target j Results Influence of target conformation on GOLD and SHEF results The density plots of figure 7 give a picture of how target conformation specificities influence GOLD and SHEF results The SHEF correlation between 1P8D and 1PQ9 figure 7 b is greater than those between 1PQ6 and both 1P8D figure 7 a and 1PQ9 figure 7 c This is in agreement with the observation that the 1PQ6 shape is the most specific In the case of GOLD it first appears that 1P8D and 1PQ6 results are highly correlated figure 7 d The correlations with 1PQ9 figures 7 e and 7 f are lower A significant amount of structures performing well with both 1P8D and 1PQ6 are ranked low with 1PQ9 indicating a group of ligands whose size fits well into the former active site conformations but not in the smaller 1PQ9 Surprisingly such an expected group does not appear in SHEF results Therefore SHEF whi
256. tion with water goes to the same level as with the ligand 1 complex after 800 ps of simulation The differences observed for the uncomplexed Grb2 SH2 correspond to the ligand not interacting with cavity 1 arginines thus stabilizing their interaction with the glutamate In all cases at the end of the trajectories the Glu side chain is displaced as to make a strong ionic interaction with the K 4 Lys AA cationic side chain therefore only the Glu backbone is then located at the bottom of cavity 1 2000 2000 Significant differences in the interactions of the charged Grb2 SH2 residues that are not directly involved in the protein ligand binding are also observed For example Keo Lys aA is more solvent exposed in ligand 1 complex than in ligand 2 complex itself more exposed than in uncomplexed Grb2 SH2 The opposite is observed with K7 Lys AB The E7 Glu aA residue only interacts significantly with water molecules in the ligand 1 complex The Eg Glu BB residue similar to the Rgs Arg BB residue involved in ligand binding is much more solvent exposed when Grb2 SH2 is ligand bound All those variations are correlated to induced fit effects upon ligand binding Differences found in some cases between ligand 1 and ligand 2 could be the consequence of their adoption of different binding modes concerning their group in position 0 bound to cavity 1 Force field dependence One aspect that could be of importance for explaining the initial
257. tions du code g n tique Cela permet aussi galement de r parer les pertes cellulaires accidentelles ponctuelles dues par exemple des blessures assurant ainsi le bon fonctionnement des organes et des tissus Lorsque cet quilibre est rompu cela peut avoir pour effet une multiplication anarchique de certaines cellules lesquelles peuvent se concentrer formant des tumeurs lesquelles caract risent le cancer lorsque leur taille ne permet plus leur retrait par des interventions m dicales simples Dans de nombreux cas ces tumeurs peuvent menacer directement le bon fonctionnement d un ou plusieurs organes vitaux ou du syst me sanguin ou sont susceptibles de le faire en migrant dans l organisme on parle alors de m tastases Le cancer met alors gravement en danger la vie de l individu affect et aboutit souvent sa mort en l absence de traitement un tel stade le cancer met alors gravement en danger la vie de l individu affect et aboutit souvent sa mort en l absence de traitement 90 des cas de mortalit dues au cancer sont cons cutives la formation de m tastases 8 Quelle qu en soit la cause le d veloppement d un cancer a toujours pour origine la mutation d un ou plusieurs proto oncog nes qui sont alors appel s oncog nes 9 le cancer est une maladie g n tique La totalit des cancers r sultent une s rie concert e de mutations g n tiques et non d une seule 10 13 ce qui explique l augmen
258. ton M R Futreal P A Mutations of the BRAF gene in human cancer Nature 2002 417 906 907 19 Pendergast A M Quilliam L A Cripe L D Bassing C H Dai Z Li N Batzer A Rabun K M Der C J Schlessinger J BCR Abl induced oncogenesis is mediated by direct interactions with the SH2 domain of the Grb2 adaptor protein Cell 1993 75 175 185 20 Daly RJ Binder M D Sutherland R L Overexpression of the Grb2 gene in human breast cancer cell lines Oncogene 1994 9 2723 2727 21 Boutin J A Tyrosine protein kinase inhibition and cancer Int J Biochem 1994 26 1203 1226 22 Brugge J S New intracellular targets for therapeutic drug design Science 1993 260 918 919 23 Songyang Z Shoelson S E Chaudhuri M Gish G Pawson T Haser W G King F Roberts T Ratnofsky S Lechleider R J Nell B G Birge R B Fajardo J E Chou M M Hanafusa H Schaffhausen B Cantley L C SH2 domains recognize specific phosphopeptide sequences Cell 1993 72 767 778 24 Songyang Z Shoelson S E McGlade J Olivier P Pawson T Bustelo X R Barbacid M Sabe H Hanafusa H Yi T Ren R Baltimore D Ratnofsky S Feldman R A Cantley L C Specific motifs recognized by the SH2 domains of Csk 3BP2 fps fes Grb2 HCP SHC Syk and Vav Mol Cell Biol 1994 14 2777 2785 25 Rahuel J Gay B Erdmann D Strauss A Garcia Echeverria C Furet P Car
259. tre relativis e D une part l interaction de ligands sur les domaines SH2 est sp cifique et en ce qui concerne Grb2 comme nous l avons vu pr c demment un seul mode de liaison en f turn est connu On peut raisonnablement supposer dans de telles conditions qu aucun autre mode ne soit favorable m me si l analyse des dynamiques du complexe de ligand 2 semble indiquer que ce point m riterait de plus amples investigations D autre part il est vident qu afin de pouvoir pr dire par docking la liaison d un ligand dans un mode radicalement diff rent il faudrait d j disposer de la conformation du r cepteur correspondante sachant que les effets d induced fit d limitent une limitation importante de la quasi totalit des programmes de docking disponibles dont GOLD Screening virtuel sur la cible Grb2 SH2 Mol cules s lectionn es comme candidates Nous avons eu recours diff rentes bases mol culaires existantes provenant de diff rentes sources m ta base ZINC fournisseurs de produits chimiques collaborateurs afin de s lectionner les mol cules tester par docking sur Grb2 SH2 Les ligands de Grb2 SH2 connus des d riv s de ceux ci ainsi qu un certain nombre de mol cules construites manuellement ont t regroup s au sein d une petite base cibl e de r f rence Au total environ 30000 mol cules ont t pr par es si n cessaire correction d erreurs suppression de parasites et conversi
260. ture are conserved ii Only the water molecules from the waterbox that are distant by less than 5 A from both the protein and the ligand during a continuous period of at least 50 ps of simulation are maintained iii Only the protein residues distant by less than 5 A to the ligand at any time are maintained iv Of course all ligand atoms are maintained Please note that such truncated trajectory files are only used for graphical analysis using InsightII all other analyses with NAMD make use of the original dcd files We used cluster graphs to study the stability of both ligand 1 and ligand 2 conformational behavior during the MD simulations Such graphs represent the RMSD between the different conformations obtained during the simulation The x and y axis correspond to the simulation time and each point represents the RMSD between conformation at timestep x and timestep y thus on the diagonal we have RMSD 0 A and every cluster graph is symmetric regarding its diagonal With the color scale used black white blue pink and red correspond to slight variations that indicate normal fluctuations at 300K while green cyan yellow then black again highlight more or less important conformational changes The cluster graphs are obtained by reprocessing InsightII generated graphics They only represent the absolute stability of the ligands conformation over time and are not directly related to its stability regarding the receptor nor to the bind
261. ualifi de screening criblage soit proc der la mise au point rationnelle de mol cules soit opter pour une combinaison de ces deux approches de base Dans l id al des structures exp rimentales r solues par RMN ou RX seront disponibles si tel n est pas le cas il faudra construire un mod le et tre particuli rement vigilant lors de sa validation En effet si le mod le de d part n est pas correct le temps qui sera ensuite pass dans les tapes suivantes de recherche le sera en pure perte On entend par accessibilit in vitro la disponibilit de protocoles de test en laboratoire de pr f rence quantitatifs In vivo il est utile de disposer d un marqueur mol culaire appropri et non toxique afin de visualiser l activit de la mol cule dans l organisme Il ne s agit pas toutefois de disposer d une connaissance compl te de la cible mais d une connaissance suffisante afin de pouvoir mettre en place un protocole de recherche de m dicaments robuste Le screening peut tre exp rimental ou bien virtuel Dans le premier cas des tests biochimiques exp rimentaux mesurent l affinit d une mol cule donn e pour la cible dans le second cette affinit est estim e par des calculs num riques Ces m thodes doivent se baser sur une r f rence quelconque qui peut tre la structure de la cible si celle ci est disponible on parle alors de structure based drug design la structure d
262. ucture based virtual ligand screening combining FRED DOCK and Surflex Journal of Medicinal Chemistry 48 2005 6012 6022 45 Leroux V and Maigret B Should structure based virtual screening techniques be used more extensively in modern drug discovery Computers and Applied Chemistry 24 issue 1 2007 1 10 46 Yamagishi M E B Martins N F Neshich G Cai W Shao X Beautrait A and Maigret B A fast surface matching procedure for protein ligand docking Journal of Molecular Modeling 12 2006 965 972 47 Singh J Chuaqui C E Boriack Sjodin P A Lee W C Pontz T Corbley M J Cheung H K Arduini R M Mead J N Newman MN Papadatos J L Bowes S Josiah S and Ling L E Successful shape based virtual screening the discovery of a potent inhibitor of the type I TGF receptor kinase TRRD Bioorganic amp Medicinal Chemistry Letters 13 issue 24 2003 4355 4359 48 Ritchie D W and Kemp G J L Fast computation rotation and comparison of low resolution spherical harmonic molecular surfaces Journal of Computational Chemistry 20 issue 4 1999 383 395 49 Cai W Xu J Shao X Leroux V Beautrait A and Maigret B SHEF a vHTS geometrical filter using coefficients of spherical harmonics molecular surfaces Journal of Molecular Modeling to be submitted 2007 50 Jones G Willett P and Glen R C Molecular recognition of receptor sites using a genetic algorithm with a description of desolvation Journal o
263. uence pY YN ob issant au motif pYXN exig par Grb2 SH2 107 Le compos mAz pTyr pTyr Asn NH2 fut donc synth tis mais s av ra deux fois moins actif que mAz pTyr Ac c Asn NH indiquant que dans le cas de Grb2 le r sidu pTyr n tait pas dans une conformation optimale 79 Il fut donc a m thyl dans le but de le stabiliser dans la conformation B turn sp cifique aux ligands de Grb2 SH2 Le ligand r sultant mAz pTyr a Me pTyr Asn NH pr senta cette fois ci une activit dix fois plus importante Ka 30 5 nM IC5o 11 1 nM 79 ce qui en fait un des inhibiteurs les plus actifs pour Grb2 SH2 connus ce jour Ce ligand s av ra en effet un peu moins actif que la r f rence Ac pYIN NH ICso 26 uM et 9 uM respectivement ce qui n en faisait pas priori une base d optimisation plus int ressante d autant plus que des modifications sur une telle base non peptidique sont notoirement plus difficiles 4 r aliser pour les exp rimentateurs que des substitutions de groupes peptidiques 32 De fa on int ressante des tudes par mod lisation mol culaire de mAZ pTyr a Me pTyr Asn NH indiqu rent que si les r sidus mAZ pTyr et Asn taient li es dans les deux cavit s connues du r cepteur Grb2 SH2 a Me pTyr interagissait dans une troisi me cavit jusqu alors non cibl e compos e des r sidus Sia Ser BG Riy Arg BG et Nia Asn BG 79 Cette observation fut confirm e exp rimentalement p
264. ulation of the different GOLD hit groups after SHEF filtering initial population 1P8D SHEF based filters 1PQ9 1PQ6 multiple target 10 50 10 50 10 50 10 50 1P8D specific 1PQ6 specific 1PQ9 specific 1P8D 1PQ6 1P8D 1PQ9 1PQ6 1PQ9 1P8D 1PQ6 1PQ9 all hits GOLD based hit groups 1401 1080 Table 3 Values of the SHEF filtering efficiency E f for f 10 and f 50 These values are directly correlated to those of table 2 GOLD based hit groups SHEF based filters Sper qian ce 1P8D 1PQ6 1PQ9 multiple target 10 50 10 50 10 50 10 50 1P8D 92 8 64 2 1PQ6 85 7 53 7 1PQ9 100 64 0 94 0 56 6 gt 1PQ9 specific 100 42 0 1P8D 1PQ6 75 7 37 9 1P8D 1PQ9 100 69 1 1PQ6 1PQ9 100 61 1 1P8D 1PQ6 1PQ9 all hits 88 7 463 95 0 56 7 816 429 Discussion and concluding remarks In this study we wanted to present an overview of VSM G and then more precisely to evaluate the usefulness of the SHEF geometrical matching procedure as part of the VSM G multiple step high throughput VS procedure We have chosen as the reference data score values from the flexible docking program GOLD This allows for a qualitative assessment of MSSH SHEF efficiency as a first fast filter for the VSM G
265. ulations informatiques jouent un r le crucial 84 doivent concentrer une bonne part des efforts des chercheurs et des industriels Enfin les m dicaments actuellement sur le march ne ciblent qu une toute petite partie de l organisme moins de 500 biomol cules 85 Les m thodes de recherche pharmaceutiques classiques ne sont plus efficaces les cibles tant de plus en plus complexes et sp cifiques Dans ce contexte l apport des techniques de la bioinformatique permettant une r duction des co ts et un traitement plus pouss en compl ment des approches exp rimentales prend naturellement un poids de plus en plus important On peut rajouter ici que si l industrie pharmaceutique milite fr quemment pour une simplification et surtout un racourcissement des processus de validation cliniques des m dicaments il s agit plus pour elle d un probl me conomique que d un enjeu scientifique L am lioration des protocoles de s lection de candidats ant rieurement aux essais cliniques est l inverse un probl me m thodologique purement scientifique 17 La recherche de nouveaux m dicaments anti cancer Durant de nombreuses ann es la communaut scientifique a pens que le d cryptage du g nome pourrait tre l origine d une r volution scientifique Si tel n a pas t le cas les efforts sont d sormais tourn s vers l interactome un d fi scientifique de taille est pos comment interpr ter et e
266. uld constitute a conformational constraint which is very useful against bulk solvent destabilizing effects or potential unwanted conformational changes of specific protein residues caused by their interaction with the solvent Concerning cavity 1 with OPO as a four directional binding group only deep binding in cavity 1 as performed by ligand 1 seems favorable With the three directional binding of PO a conformation locating ligand 2 partly outside cavity 1 is possible The loss of binding observed at 1 8 ns at this level could have two causes i the conformational change that is made on both the residue in 1 position targeting cavity 3 and the naphthalene residue and ii the fact that the phosphoney group not bound into cavity 1 like the ligand 1 phosphates but rather at cavity 1 entrance is much more exposed to the solvent and thus to possible destabilizing effects At this point it is only possible to affirm that at least one of these assumptions is correct we cannot determine whether 1 is true or ii is true or both Proposals for modelling new Grb2 SH2 inhibitors Useful additional information could be obtained by performing MD simulations on ligand 1 and ligand 2 analogues Several strategies could be attempted in order to model such analogues i The 1 group does not interact with Grb2 SH2 in the case of ligand 1 as well as with ligand 2 its suppression or replacement should be easily performed ii The Kio Lys BD r
267. un ligand de r f rence ligand based drug design des jeux de donn es statistiques QSAR etc 15 La d finition que l on fait d un compos prometteur ou hit touche est arbitraire Dans le cas d un screening il s agit d un compos qui appara t comme interagissant significativement plus que la moyenne des compos s test s sur la cible vis e Cela ne signifie en rien que ce compos sera comp titif vis vis d une r f rence connue ni m me qu il aura une quelconque activit in vivo Dans le cas de mol cules mises au point sp cifiquement c est la comp tence du chercheur qui d terminera en premier lieu la qualit des hits mis en place Il est bien s r impossible de tester par screening l ensemble des mol cules possibles celles qui le sont font partie de banques de mol cules pr existantes ou bien sont d finies sp cifiquement par rapport la cible screening aveugle d une part test de structures d termin es rationnellement d autre part Divers facteurs permettent de caract riser l utilit dans le cadre du drug design d une base de mol cule utilis e pour le screening La diversit chimique doit tre importante et surtout les mol cules doivent avoir les caract ristiques physico chimiques caract risant un m dicament Des r gles l mentaires permettent de filtrer rapidement bon nombre de mol cules ne r pondant pas ce dernier crit re hors des banques avec une
268. versity Database large scale 598 327 563 777 94 2 195 332 32 6 84 3 823 81 8 diversity subset 8 383 7 875 93 9 3 178 37 9 50 0 43 5 48 3 CN 31 220 27 403 87 8 20 295 65 414 44 8 437 Parameterization of the virtual screening programs 1 102 299 conformers were docked using SHEF in the three target conformations giving a total of 3 306 897 rigid docking calculations Using GOLD 8 383 molecules were docked giving 25 149 flexible docking calculations The programs parameters that were used favoring reliability over speed are listed in chart 3 Chart 3 Parameters for MSSH SHEF and GOLD used for the validation study simulating the use of MSSH SHEF for filtering prior to GOLD calculations MSSH 35 36 SHEF 49 spherical harmonics expansion of order 10 cavity coordinates defined using the ligand center of mass GOLD 51 default genetic algorithm parameters 50 dockings molecule early termination option docking stopped if the top 5 conformations fall within 1 5 A RMSD range cavity definition flood fill works well when the receptor is not open and extended same cavity coordinates as with MSSH SHEF scoring function GoldScore Definition and relevance of reference data The reference data for evaluating SHEF performance is constituted by GOLD results and not by experimental data Like all docking programs GOLD does not provide 10
269. vons principalement effectu des calculs par dynamique mol culaire et docking flexible On peut remarquer que relativement la plupart des complexes mol culaires repr sentatifs d une activit biologique Grb2 SH2 correspond une structure de taille modeste Les ligands peptidiques de Grb2 SH2 sont ainsi des s quences constitu es de 4 acides amin s actifs 87 ce qui est raisonnable et permet d envisager que la diversit chimique des inhibiteurs possibles soit accessible A moyens de calcul gaux et pour un protocole de simulation donn il est donc priori possible d obtenir des r sultats plus pr cis ou en plus grand nombre en comparaison avec la moyenne des cibles biologiques potentielles Ant rieurement a cette th se des travaux de recherche th oriques sur Grb2 SH2 ont r v l de nombreuses informations int ressantes 70 78 79 88 89 et constituent une base de travail appr ciable Pour autant si on dispose d une connaissance initiale satisfaisante du domaine SH2 de Grb2 celle ci n apparait pas pour autant suffisante pour r soudre toutes les interrogations s y rapportant On peut souligner que c est le cas des domaines SH2 en g n ral 90 La cible Grb2 SH2 est toujours en cours d investigation traduisant son int r t de la part de la communaut scientifique Notons enfin que les complexes form s par Grb2 SH2 avec des inhibiteurs pr sentent certains autres avantages facilitant leur mod lisation
270. wo S131 RUS Dao posn S11 os wopuer au 107 pasn siuou as pasn SI 3109S uonezrnururut uoddns p9p y uonouny SULI09S UOTJONTHSUO9 EJUAUOIOUT get YAPIV JI uoneztumdo pozrumdo uoym PL 2p09 anos LIT 911 Ayjenueut pourap 10 jurod pee eae f d pisu puesryq quiojnod en A du re aAnoe Afrepnonted fs Jojourered y m J jo Surpu d p Jo UOTJEZTUTUTU Surmp jqrssod jqrx juo CTI 9d f 5 SOUPE Id S94 Jo U99 W01 poyesoues PHO MPA SU IE N99 OU TUI 2 NESISA AIO A our QUI pUEUUOT soyorordde B90 Suryoyeur A SU uo Apsour peseq aroos are ainjonys 10j da001 Jo pues 48o7ens wood J A E yordu snone A PUS ROLNOWIONDH MAGI sIqrssod eroAag SJUOWI9SUCLIEII porun JsIJ IOYDUY A syusweambat sulpuey AVITIQIxog aureu SSH snJe s esa amp P9ZIPIIE184 998HI9JUI 19S uoriuyep 1014539 uoneajog auIsUa PILIS uomuny SULIOIS AMEIXA 103d999y puesry we18014 References 1 Drews J Drug discovery A historical perspective Science 287 2000 1960 1964 2 Charifson P S Practical application of computer aided drug design 1997 Marcel Dekker 552 pages 3 Jorgensen W L The many roles of computation in drug discovery Science 303 issue 5665 2004 1813 1818 4 James L C and Tawfik D S Conformational diversity and protein evolution a 60 year old hypothesis revisited Trends in Biochemical Sciences 28 issue 7 2003 361 368 5 Berman H M Westbrook J Fe
271. xploiter de fa on utile la masse des donn es obtenues Le cancer tant la maladie g n tique par excellence il constitue un axe de recherche id al pour l exploitation des connaissances de la g nomique Un grand nombre d approches d sign es sous le terme d oncologie mol culaire ont permis l identification d un nombre important de g nes impliqu s dans les processus de prolif ration canc reuse Le cancer tant la r sultante d un ensemble de mutations g n tiques il ne s agit pas tant de caract riser de fa on particuli rement d taill e ces anomalies mais plut t de d tecter celles qui surviennent de fa on r currente De nombreuses tudes bas es l origine sur une telle approche et visant la mise au point de mol cules anti cancer sp cifiques sont actuellement conduites ce qui souligne l int r t de la communaut scientifique 86 Des m thodes de recherche innovantes sont susceptible d en r sulter La faible sp cificit des traitements anti cancer actuels tels que la chimioth rapie et la radioth rapie massivement employ es pose probl me La mise au point d une nouvelle classe de m dicaments sp cifiquement anti cancer est dans ce contexte particuli rement prometteuse car porteuse la fois d une plus grande efficacit et d une plus faible toxicit Enfin au fur et mesure que l esp rance de vie augmente dans les soci t s modernes le cancer m0 pres 8000 NS devie
272. y ainposoid RIOIOUIWIOD uo uonepmnoreo OU ur poyeIsaquy P J FO qu uo poxoop SISJEM z sosod pentur Jo y uo Surpuadap suwa pls v se possaooid FOYE penser PHS J9 U99 W01 payes9ues PUO a g 29eds yoreag dajs a1dn mn panqinsiq Sui puiq wordxg uonsoyes 1 TEPIH Jo suoneurquios SurAre A oid 10 da901 ON Koem es sxopulq soseyord g nts pue paods RAS popraoid ore sprezim 6 1 uonezrumdo Jun099e OUT S1997j9 Ado qua UMOUX JIM Sox9 du109 U99MJ9q JJO 0pUT I 2 a ee js uonsajop AA eorydess uonouny OW pure uopezrunuru pue g uoneaosap Jo uornepnus DIN a 9 I re1 poo3 g suras en PO 4 N O1d WOD SA 2AISU9U9IdWIO9 UMOUX Jou Suuo9s yordu YIM S9JBUIPI009 Sun sjuouodu09 LEI paresoues 10 ndur Ina 04 NOI AMTIQrxayy 10 d2991 x a Aor gy J Aqrenuew pourJop oq uep BUIO UT ur Yoreas PLD qeuonrppe UJIM poseq JA Sv Jos SUOTRULIOJUOS JO juouuissosse OY se a J01d9991 pis e Jo 3SN Feni s TEI uonezrununu eal 3 asu Je puesi souarajor Lesti swm Aq pamoy oy nn Loe 1 mpout o8es gt TEIOBFIOQUI SJATOSTP 3 TET suonouny Xp Sarmonns 39818 uonezrumdo 671 8Z1 ysvj AIDA odAj 9s89T WaAd S94 sidus 10 QD punog e jo UOTEULIOJUO SUBI plSLI Sursn z Jo juaurooe d Sur109s eoridus apni Jo ojquiosus ue juasardor 904 XXI prrwuop oy WOJ Jo ENUCIA US IAN OY UT PUVSIT a onewomy ueo pus Joydao01 oy se ouwes ened ay JO UONONSUOION j 971 fie g 971 pesn eq ose A
273. y conditions were used for the three models including the 80 A waterbox and all models were processed at constant pressure of 1 atm and at a constant temperature of 300 K using Langevin piston No constraints were applied except for the O H water bonds that are kept rigid in the simulations with water Switching cut off was set from 8 to 10 A Figure 4 Modelled starting structure for the solvated ligand 1 complex MD simulation The calculations were performed on an SGI 3800 supercomputer and a Linux cluster A minimization of 10000 steps using the conjugate gradients method was first performed At the end of the minimization phase it was verified that the binding mode of ligand 1 and ligand 2 did not change significantly by checking the hydrogen bonds It was also verified that the binding mode of ligand 1 and ligand 2 was similar as predicted from previous works on ligand 2 we superimposed ligand 1 and ligand 2 complexes after minimization and noted that the positions of the three functional groups of both ligands the two phosphate phosphone groups and the Asn residue relative to the Grb2 SH2 receptor matched 10 ps of equilibration were then performed giving the initial coordinates and velocities for the dynamics The consistency of the structures was verified another time at this point The time steps of the dynamics were set at 1 fs The total simulated time for each of the five trajectories ligand 1 and ligand 2 complexes in 80 A waterbox
274. y for a target Thus it can prioritize them for experimental testing Ligand based modules such as substructure search can be involved as pre processing steps to screen molecular databases and reduce the number of compounds to be considered subsequently This initial operation can precede the central element of the platform the screening funnel a multi step structure based filtering process that hierarchically combines several docking methods After describing the VSM G platform we will present a proof of concept study in the filtering enrichment context using the liver X receptor B LXRf as a target for a screening calculation against a diverse ligand database The VSM G screening funnel was used consisting of a fast geometrical matching filter preceding flexible docking This approach is compared to using flexible docking alone for VS The benefits and limitations of geometrical matching as part of the screening funnel approach in terms of computing efficiency applicability and relevance are discussed Overview of the VSM G platform Aims and scope of VSM G The first step of the pre clinical drug discovery process can be simplified as a work of exploration at the intersection of distinct spaces 20 The first of these is the proteome whose exploration in the drug design context involves its restriction to the sub space of proteins whose interactions could be significant therapeutically as novel targets the target space The second spac
275. y is still deceiving despite constant enhancements and results are unpredictable What are the origins of such issues In this short review we will first summarize the current status of computer aided drug design then we will focus on the structure based class of virtual screening approaches for which docking programs constitute the main part Can such methods give something more than cost savings in the early banks to hit phases of the drug discovery process We will try to answer this question by exploring the highlights and pitfalls of the great variety of docking approaches It will appear that while the structure based drug design field is not yet ready to fulfill all of its early promises it should still be investigated extensively and used with caution Most interestingly structure based methods are best used when combined with other complementary drug design approaches such as the ligand based ones In this regard they will have an increasing role to play in modern drug discovery which is more and more interdisciplinary Keywords modern drug discovery structure based drug design computer aided drug design virtual screening docking hit discovery hit to lead grid computing Abbreviations CADD computer aided drug design VS virtual screening SBDD structure based drug design LBDD ligand based drug design HTS experimental high throughput screening vHTS virtual high throughput screening GA genetic algorithm QSAR qua
276. yer M Bradshaw T Berman J Peptide inhibitors of src SH3 SH2 phosphoprotein interactions J Biol Chem 1994 269 31711 31719 39 Lubman O Y Waksman G Structural and thermodynamic basis for the interaction of the Src SH2 domain with the activated form of the PDGF f receptor J Mol Biol 2003 328 655 668 40 Henriques D A Ladbury J E Inhibitors to the Src SH2 domain A lesson in structure thermodynamic correlation in drug design Arc Biochem Biophys 2001 390 158 168 41 Hanessian S McNaughton Smith G Lombart H G Lubell W D Design and synthesis of conformationally constrained amino acids as versatile scaffolds and peptide mimetics Tetrahedron 1997 53 12789 12854 42 Eguchi M Lee M S Nakanishi H Stasiak M Lovell S Kahn M Solid phase synthesis and structural analysis of bicyclic f turn mimetics incorporating functionality at the I and I 3 positions J Am Chem Soc 1999 121 12204 12205 43 Eguchi M Lee M S Stasiak M Kahn M Solid phase synthesis and solution structure of bicyclic B turn peptidomimetics diversity at the I position Tetrahedron Lett 2001 42 1237 1239 44 Liu W Q Olszowy C Bischoff L Garbay C Enantioselective synthesis of 2S 2 4 phosphonophenylmethyl 3 aminopropanoic acid suitably protected for peptide synthesis Tetrahedron Lett 2002 43 1417 1419 45 Furet P Gay B Caravatti G Garcia Echeverria C Rahuel
277. yro n a donn lieu un gain d affinit significatif 89 119 126 La limite de cette approche r side dans le fait que l impact sur l affinit d un substituant donn en particulier sur pTyro d pend grandement de la base structurale sur laquelle la substitution est effectu e Par exemple Pmp peut s av rer suivant la plateforme sur laquelle il se substitue pTyro soit quivalent soit largement d favorable 19 Nouvelles bases structurales Une autre approche pour la d termination de ligands non phosphoryl s pour Grb2 SH2 consiste reprendre les m thodes qui ont permis par optimisations successives de d terminer des ligands phosphoryl s efficaces sachant que les s quences peptidiques pY5X 1N 2 sont pr f r es il peut s av rer int ressant d effectuer des screenings de s quences Y5X 1N 2 Sur de telles s quences on remarque alors que l absence du groupe phosphate en position 0 entra ne une forte s lection pour un r sidu Glu gt qui compenserait ainsi l absence de charge n gative sur la position 0 96 On observe dans le meilleur des cas pour des ligands peu ou moyennement actifs et avec des substituants doublement charg s tels que Pmp pmF et FOMT une activit l g rement meilleure sans que ce gain soit significatif au regard de la pr cision intrins que des protocoles exp rimentaux de mesure d affinit La plupart du temps on observe une perte d affinit d un ou plusieurs ordr

Download Pdf Manuals

image

Related Search

Related Contents

User Manual  Metal 3D - CYPE Ingenieros    Boîtier à fixation murale pour Flair 2xD Wall  操作ガイド(基本編)  Edbak TR7  Proyección Brillante En gran Pantalla  User Manual, V1.6 - Viola Systems Ltd.  

Copyright © All rights reserved.
Failed to retrieve file