RationalDesignof Ligandsfor AsymmetrieCatalysis - ETH E

Diss. ETH No. 12977
Applicationsof Molecular Modeling and SupramolecularChemistry:
De Novo Design of MHC Class II Inhibitors
and
Rational Design of Ligands for AsymmetrieCatalysis
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZÜRICH
for the degree of
Doctorof Natural Sciences
Presented by
LaurentDUCRY
Dipf. Chem. Universityof Lausanne
born 14.02.1971 inMartigny, Switzerland
accepted on the recommendation of
Prof. Dr. Fran§ois Diederich, examiner
Prof. Dr. Hans-Jürg Borschberg, co-examiner
Dr. Gary L. Olson, co-examiner
Zürich 1998
Vll
Abstract
In the past few years, supramolecular chemistry has received much attention,
resulting in a greater understanding of the weak, non-covalent forces governing
intermolecularinteractions (Chapter 1). The emergence of Computer calculations has,
moreover, allowed quantification of these interactions, opening the door to the
new
design of
molecules with pre-definedsupramolecular properties.
chemistry, molecular modeling has changed the way new drugs are
discovered and developed (Chapter 2.1). De novo drug design can be considered as an
alternativeto the traditional Screening methods used for the discovery of lead structures.
The HLA-DR molecules are a family of major histocompatibility complex (MHC) class
II proteins which play a central role in the immune System. They have been shown to
be linked to rheumatoid arthritis. The availabilityof an X-raycrystal structure of HLADR1 complexed with the HA 307-319 peptide makes them suitable for a de novo design
approach to potential inhibitors (Chapter 2.2). First generationinhibitors were designed
and synthesized with the aim of preserving the interactions present between the protein
and a natural peptide in the three-dimensionalstructure (Chapter 2.3). The inhibitors
consist of the cyclic sulfonylurea scaffold I, hydrogen-bondingto the MHC protein.
This scaffold has two side-chains; the first (R1) is aromatic and was designed to
complex a large hydrophobic pocket, while the second (R2) was intended to reach
another region of the binding groove where important functionalities for hydrogen
bonding are present. These Compounds however exhibited only low potency (single
digit millimolarinhibition for the HLA-DR4protein).
In medicinal
v/
?1
R'
H
v/2
M
R
II
H
generation inhibitors were designed by modifying the R2 side-chain (Chapter
2.4). A simplified Compound, prepared as synthetic model, proved inactive. The
synthesis of the second generation nonpeptide inhibitors was however not finished as
priority was moved onto peptidomimetic inhibitors (Chapter 2.5). These Compounds
still incorporate the sulfonylureascaffold, but in addition to the R1 aromatic side-chain,
Second
vm
peptide residue was attached as second functionality (R2). Urea analogs (scaffold II)
were also prepared to investigate the effect of structural modification of the cyclic
template. None of these Compounds showed improved binding affinity for HLA-DR
molecules. A hypothesis for the decreased potency observed when using the
peptidomimetic scaffolds I or II compared to fully peptidic Compounds is presented
(Chapter 2.6), and structuralmodificationsfor next generationinhibitors are proposed.
a
Computer calculations were next applied to asymmetriccatalysis, a field where the
rational design approach has not yet been exploited (Chapter 3.1). Ligands
incorporating a l,l'-binaphthyl unit substituted in the major groove by a Cinchona
alkaloid moiety were designed for enantioselective phase transfer catalysis (Chapter
3.2). The unsymmetricai l,P-binaphthylfragment was prepared via the oxidative crosscoupling of two differently substituted naphthalen-2-ols, and resolved using
camphorsulfonyl chiral auxiliaries. Coupling with the alkaloid unit was achieved by
nucleophilic addition of a l,l'-binaphthyllithium to a Weinreb amide-type quinuclidine.
Subsequent stereoselective reduction of the resulting carbonyl afforded the four
precursors III with different configurations at C(8) and C(9). These Compounds were
quaternized and used as phase transfer agents for the PTC allylation of 6,7-dichloro-5methoxy-2-phenylindanone, giving 6 to 32% ee's. Monte-Carlo Computer calculations
in agreementwith the experimental enantioselectivities and in each case correctly
predicted the preferentially formed enantiomers. An attempt to immobilize phase
were
transfer agents into organically modified
aerogels was made to allow complete and easy
recovery of such catalysts. Covalent incorporationof dihydrocinchonine quats resulted
in
a
an
enantioselective Silicate sol-gel catalyst.
large
decrease in
enantioselectivity (from
7%) in the allylation of 6,7dichloro-5-methoxy-2-phenylindanone,but represents, nevertheless, the first example of
73% to
MeO'
MeO
MeO
sO
in
;?-chlorobenzoate esters of diastereoisomers III were used in the Sharpless
asymmetricdihydroxylation(Chapter 3.3), and revealed in some cases useful catalytic
activity for the enantioselective dihydroxylation of trans-stilbene(2 to 67% ee).
Studies towards the synthesis of f,l'-binaphthyl Compounds substituted in the minor
groove with a Cinchona alkaloid have also been made (Chapter 3.4).
The
IX
Resume
supramoieculaire a rec,u une attention toute
particuliere. II en a resulte une bien meilleure comprehension des interactions faibles
gouvernant la formation d'edifices de ce type (Chapitre 1). De plus, l'apparition de la
Simulation assistee par ordinateur a ouvert la porte au design de nouvelles molecules
ayant des proprietes supramoleculairespredefinies.
Ces dernieres annees, la chimie
En chimie
•
medicinale, la modelisation moleculaire a modifie la maniere dont les
medicaments sont decouverts et developpes (Chapitre 2.1). La conception de novo de
medicaments peut etre consideree comme une alternative aux methodes
d'echantillonage traditionnellement utilisees pour la decouverte de composes tetes de
serie. Les molecules HLA-DR sont une famille de proteines du complexe majeur
d'histocompatibilite (CMH) de classe II, jouant un röle central dans le Systeme
immunitaire et qui sont liees ä l'arthrite rhumatoi'de. L'existence d'une structure
cristallographique de HLA-DR1 complexe par le peptide HA 307-319 a permis la
conception de novo d'inhibiteurs potentiels (Chapitre2.2). Une premiere generation a
ete modellisee et synthetisee avec le souci de preserver les interactions presentes entre
la proteine et le peptide naturel dans la structure en trois dimensions (Chapitre2.3). Ces
inhibiteurs ont la structure I, une sulfonylureeformant des liaisons hydrogene avec le
CMH. Ils possedent en outre deux chalnes laterales; la premiere (R1), aromatique, a ete
modelisee pour complexerune grande cavite hydrophobe, alors que la seconde (R2) doit
atteindre une autre partie du site actif oü d'importantsgroupes fonctionnelspouvant
former des liaisons hydrogene sont presents. Ces composes n'ont cependant qu'une
faible activite (inhibition millimolaire de HLA-DR4).
R^
/
II
R1
R^
H
H
Les inhibiteurs de la seconde generation ont ete modelises en modifiant R2
(Chapitre 2.4). Un compose simplifie, prepare comme modele, s'est revele inactif. La
synthese des inhibiteurs de la seconde generation n'a cependant pas ete terminee, la
priorite ayant ete donnee aux inhibiteurs peptidomimetiques(Chapitre 2.5). Ces
composes,
en
plus
de la
sulfonyluree cyclique
et
du substituant
aromatique R1,
contiennentun peptide comme second groupe fonctionnel (R2). Des urees (structureII)
ont egalement ete preparees comme analogues. Aucune de ces molecules n'a cependant
les proteines HLA-DR. Une hypothese expliquant
la faible affinite des structures de type I ou II est presentee (Chapitre 2.6), et des
montre une meilleureaffinite
envers
modificationssont proposees pour une future generationd'inhibiteurs.
Le calcul assiste par ordinateur a ensuite ete applique ä la catalyse asymetrique, un
domaine oü des methodes de conception rationnelles n'ont pas encore ete utilisees
(Chapitre 3.1). Des ligands incorporant un groupe l,l'-binaphthyl substitue dans le
grand
sillon par un Cinchona alcaloide ont ete concus comme ligands pour la catalyse
enantioselective par transfertde phase (Chapitre 3.2). Le fragment l,l'-binaphthyl non
symetrique
prepare par le couplage croise oxidatif de deux naphthalen-2-ols
differemment substitues et dedoubles ä l'aide d'auxiliairescamphresulfones chiraux. Le
a
ete
couplage avec la partie alcaloide a ete realise par addition nucleophile d'un 1,1'binaphthyllithium sur une amide de Weinreb derivee d'une quinuclidine. Le groupe
carbonyl en resultant a ensuite ete reduit stereoselectivementpour fournir les quatre
precurseurs III ayant differentes configurations en C(8) et C(9). Ces composes ont ete
quaternises puis utilises comme catalyseurs par transfert de phase pour l'allylation du
6,7-dichloro-5-methoxy-2-phenylindanoneavec 6 ä 32% d'ee. Les calculs Monte-Carlo
etaient en accord avec les enantioselectivitesobtenues experimentalement. D'autre part,
ils ont correctement predit quel serait l'enantiomereen exces, ceci dans chaque cas. Un
essai d'immobilisationdes catalyseurs sur des aerogels modifies a ete effectue afin de
permettre un recyclage facile et complet de ceux-ci. Une forte diminution de
l'enantioselectivitepour l'allylation du 6,7-dichloro-5-methoxy-2-phenylindanone (de
73 ä 7%) a resulte de l'incorporation d'une dihydrocinchonine quaternisee. II s'agit
neanmoins du premier exemple de catalyseur sol-gel enantioselectif.
MeO'
MeO
MeO
^7
III
/?-chlorobenzoate des diastereoisomeres III ont ete utilises dans la
dihydroxylation asymetrique de Sharpless (Chapitre 3.3) et ont, dans certains cas,
montre une enantioselectivite lors de la dihydroxylation du frans-stilbene (de 2 ä 67%
ee). Des etudes en vue de la Synthese de derives l,l'-binaphthyl substitues dans le petit
sillon par un Cinchona alcaloide ont egalement ete effectuees (Chapitre 3.4).
Les esters