Mapping and fitting the peripheral benzodiazepine receptor binding site by carboxamide derivatives. Comparison of different approaches to quantitative ligand-receptor interaction modeling

1134

J . Med. Chem. 2001, 44, 1134-1150

Ma p p in g a n d F ittin g th e P er ip h er a l Ben zod ia zep in e Recep tor Bin d in g Site by

Ca r boxa m id e Der iva tives. Com p a r ison of Differ en t Ap p r oa ch es to Qu a n tita tive

Liga n d -Recep tor In ter a ction Mod elin g

Maurizio Anzini,*^,§Andrea Cappelli,*^,‡Salvatore Vomero,^‡Michele Seeber,^$Maria Cristina Menziani,^$

Thierry Langer,^†Bertram Hagen,^†Cristina Manzoni,^#and J ean-J acques Bourguignon^∧

Dipartimento di Scienze Farmacobiologiche, Universita` degli Studi “Magna Graecia” di Catanzaro, Complesso

“Ninı` Barbieri”, 88021 Roccelletta di Borgia, Catanzaro, Italy, Dipartimento Farmaco Chimico Tecnologico, Universita` di

Siena, Via A. Moro, 53100 Siena, Italy, Dipartimento di Chimica, Universita` degli Studi di Modena, Via Campi 183,

41100 Modena, Italy, Institute of Pharmaceutical Chemistry, University of Innsbruck, Innrain 52A, A-6020,

Innsbruck, Austria, Istituto di Ricerche Farmacologiche “Mario Negri”, Via Eritrea, 62, 20157 Milano, Italy, and

Laboratoire de Pharmacochimie de la Communication Cellulaire (UMR 7081), Universite´ Louis Pasteur, 74 Route du Rhin,

67401 Illkirch Cedex, France

Received April 28, 2000

The synthetic-computational approach to the study of the binding site of peripheral benzodi-

azepine receptor (PBR) ligands related to 1-(2-chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-

isoquinolinecarboxamide (PK11195, 1) within their receptor (Cappelli et al. J . Med. Chem.

1997, 40, 2910-2921) has been extended. A series of carboxamide derivatives endowed with

differently substituted planar aromatic or heteroaromatic systems was designed with the aim

of getting further information on the topological requisites of the carbonyl and aromatic moieties

for interaction with the PBR binding site. The synthesis of most of these compounds involves

Weinreb amidation of the appropriate lactone as the key step. The most potent compound,

among the newly synthesized ones, shows a nanomolar PBR affinity similar to that shown by

1 and the presence of a basic N-ethyl-N-benzylaminomethyl group in 3-position of the quinoline

nucleus. Thus, it may be considered the first example of a new class of water soluble derivatives

of 1. Several computational methods were used to furnish descriptors of the isolated ligands

(indirect approaches) able to rationalize the variation in the binding affinity of the enlarged

series of compounds. Sound QSAR models are obtained by size and shape descriptors (volume

approach) which codify for the short-range contributions to ligand-receptor interactions.

Molecular descriptors which explicitly account for the electrostatic contribution to the interaction

(CoMFA, CoMSIA, and surface approaches) perform well, but they do not improve the

quantitative models. Moreover, useful hints for the identification of the antagonist binding

site in the three-dimensional modeling of the receptor (direct approach) were provided by the

receptor hypothesis derived by the pharmacophoric approach. The ligand-receptor complexes

obtained provided a detailed description of the modalities of the interaction and interesting

suggestions for further experiments.

In tr od u ction

steroidogenesis, the regulation of which represents a

potential clinical application of PBR ligands.³

After the discovery of benzodiazepine binding sites in

the periphery (peripheral benzodiazepine receptor, PBR),

intensive research devoted to the characterization of this

receptor has been undertaken.¹PBR has been primarily

found on the mitochondrial outer membrane even if a

nonmitochondrial localization in some cells has been

suggested. Although the physiological role of PBR is still

unclear, a wide range of pharmacological activities, such

as anticonvulsant, anxiolytic, immunomodulating, and

cardiovascular, has been related to its activation.²In

particular, this receptor appears to be involved in

Human, bovine, rat, and murine PBR have been

isolated, cloned, and sequenced.⁴However, the crystal-

lographic structure is not yet available, since its close

association to the membrane makes this protein difficult

to isolate, purify, and crystallize in its native form.

Models of the secondary⁵and tertiary⁶structure of

the PBR available in the literature are essentially

concerned with the transmembrane region (TM), which

consists of five R-helices composed of 21 hydrophobic

residues. The N-terminus of the sequence is located in

the mitochondrial domain, while the C-terminus is

exposed to the cytoplasm. The transmembrane regions

are connected by loops rich in hydrophilic residues.

Furthermore, recently reported site-directed mutagen-

esis studies demonstrated that the portion of the recep-

tor involved in the interaction with the ligands mainly

consists of the first cytoplasmic loop and that agonists

and antagonists bind the receptor in different, albeit

partly overlapping, sites.⁷

* To whom correspondence should be addressed. Anzini, M. Tel: +39

961 391157. Fax: +39 961 391490. E-mail: anzini@unicz.it. Cappelli,

A. Tel: +39 577 234320. Fax: +39 577 234333. E-mail: cappelli@unisi.it.

^§Dipartimento di Scienze Farmacobiologiche, Universita` “Magna

Graecia” di Catanzaro.

^‡Dipartimento Farmaco Chimico Tecnologico, Universita` di Siena.

$

Dipartimento di Chimica, Universita` degli Studi di Modena.

^†Institute of Pharmaceutical Chemistry, University of Innsbruck.

#

Istituto di Ricerche Farmacologiche “Mario Negri”.

^∧Laboratoire de Pharmacochimie de la Communication Celluleire,

Universite´ Louis Pasteur.

Published on Web 03/20/2001

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1135

Ch a r t 1

Ch a r t 2

PK11195 (1, Chart 1) is the first nonbenzodiazepine

ligand which was found to bind the peripheral benzo-

diazepine receptors (PBR) with nanomolar affinity, and

it is nowadays the most commonly used radioligand for

this receptor along with Ro5-4864 (2). The class of high

affinity PBR ligands also comprises alpidem (3), an

imidazopyridine, capable of stimulating pregnenolone

formation from mitochondria of C6-2B glioma cells.³The

structure-affinity relationship data of the analogues of

1 (e.g., compounds 4a ,b) were reported in the patent

by Dubroeucq and co-workers and were later discussed

by Bourguignon.⁸Georges and co-workers studied a

limited set of 1-like compounds by means of molecular

orbital calculations in the attempt to define and map

the PBR binding site.⁹

Our earlier efforts in the search of new PBR ligands

culminated in the discovery of some new lead com-

pounds 5 and 6, providing submicromolar affinity and

possessing some structural similarity with 1.¹⁰With the

aim of both understanding the interaction of leads 5 and

6 with the PBR binding site at molecular level and

optimizing their PBR affinity, we planned a large

program focused on the structural modification of these

compounds.

In a previous paper,¹¹we presented the results of a

synthetic-computational approach which, by means of

the conformationally constrained derivatives 7-11 (Chart

2), afforded important information on the orientation

of the carbonyl group of compound 1 in the interaction

with the PBR binding site.

Furthermore, compounds 8f-h (see Table 1) have

been [¹¹C]-labeled as PET PBR ligands potentially

useful in the imaging of multiple sclerosis, human

glyoma and glyoblastoma, cerebral infarction, and cal-

cium channel anomalies in heart diseases.¹²

variation in their binding affinities. First, the indirect

approach has been carried out on the isolated ligands.

The molecular determinants for the interaction with the

receptor have been inferred from the comparison of the

van der Waals volume (volume approach), the steric and

electrostatic fields (CoMFA and CoMSIA approaches),

and the molecular surface electrostatic potentials (sur-

face approach) of each ligand with respect to one or more

reference compounds. Moreover, the adequacy of each

ligand to a specific pharmacophoric hypothesis (phar-

macophoric approach) has also been exploited to obtain

molecular descriptors for quantitative structure-activ-

ity relationship (QSAR) analysis. The (QSAR) models

developed have been tested for their predictive power

by means of compound 16, which can be considered a

structural hybrid between the first and the second set

of the PBR ligands studied.

Finally, the clues obtained by the indirect approach

have been exploited, together with the experimental

information available on the receptor, to obtain three-

dimensional models of the ligand-receptor complexes

(direct approach) which provide a detailed description

of the interaction modalities and interesting suggestions

for further experimental investigation.

The present paper deals with the complementary

information obtained by structural modification of the

carboxylic ester moiety, the N-ethyl-benzylamino group,

and the heteroaromatic system of compounds 6. Several

computational methodologies have been applied to the

enlarged series of published (7-11) and newly designed

(12-16, Chart 3) compounds in order to rationalize the

Ch em istr y

Carboxamides 13a -g, 15a -d , and 16 were obtained

by the multistep sequence shown in Scheme 1. The

1136 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

Ta ble 1. PBR Binding Affinities of Previously Described

Compounds 7-11

Ch a r t 3

compd

bridge (B)

X

R₁

R₂IC₅₀(nM) ( SEM^a

7a

7b

7c

8a

8b

8c

8d

8e

8f

8g

8h

8i

8j

8k

8l

9a

9b

9c

9d

9e

10a

10b

10c

11a

11b

1

n-Bu-CH

CH₂-CH₂-CH₂

Bn

s-Bu

Bn

s-Bu

Bn

4-Cl-Bn

s-Bu

Bn

4-Cl-Bn

4-Cl-Ph

4-MeO-Ph Me

s-Bu

Bn

s-Bu

Bn

s-Bu

Bn

810 ( 110

1200 ( 200

170 ( 14

H

F

H

F

H

F

H

F

H

230 ( 70

13 ( 4.9

1200 ( 180

1700 ( 230

270 ( 66

2.1 ( 0.9

2.9 ( 0.5

2.1 ( 0.6

9.8 ( 2.0

3.4 ( 1.0

6.4 ( 1.1

8.8 ( 1.5

620 ( 100

480 ( 29

780 ( 20

490 ( 47

45 ( 15

H

Me

Sch em e 1^a

H

CH₂

CH₂-CH₂-CH₂

CHdCH-CH₂

b

H

Me

550 ( 93

11 ( 2.6

3.1 ( 0.8

8.9 ( 2.5

10 ( 0.7

2.2 ( 0.3

CH₂-CH₂

O-CH₂-CH₂

b

a

b

Each value is the mean ( SEM of three determinations. See

ref 11.

appropriate lactone (9-phenylfuro[3,4-b]quinoline-1(3H)-

one (17),¹³9-phenylfuro[3,4-b]quinoline-3(1H)-one (18),¹³

and commercially available phthalide 19) underwent

aminolysis with the suitable aluminum amide reagent

(AAR) as the key step¹⁴to afford the corresponding

hydroxymethylcarboxamides 20a -e, 21, and 22a -d

(method A). On the other hand, 2-hydroxymethyl-3-

quinolinecarboxamides 20f and 20g were obtained by

ring opening of lactone 17 with the highly reactive

propylamine and benzylamine, respectively (method B).

By treatment of the hydroxymethylcarboxamides

20a -g, 21, and 22a -d with thionyl chloride in meth-

ylene chloride, chloromethyl derivatives 23a -g, 24, and

25a -d were obtained and easily converted into corre-

sponding final products 13a -g, 15a -d , and 16 by

means of N-ethylbenzylamine.

a

Reagents: (i) Method A: Me₃Al, R₃(R₄)NH‚HCl, CH₂Cl₂,

Method B: R₄NH₂, EtOH; (ii) SOCl₂, CH₂Cl₂; (iii) Bn(Et)NH, NaI,

Na₂CO₃, MeCN.

Compound 12b was prepared by alkylation of 1,2,3,4-

tetrahydroisoquinoline with the previously reported

3-carbethoxy-2-chloromethyl-4-phenylquinoline¹³ac-

cording to the standard methods, while carboxamides

14a -c were synthesized as reported in Scheme 2. Ester

26¹⁵was hydrolyzed in basic medium to yield corre-

sponding acid 27¹⁶which was first reacted with phos-

phorus pentachloride in methylene chloride and then

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1137

Sch em e 2^a

respect to those of the target ligands. As for the

molecules of the second set which did not have the

reference fragments, we proceeded to the topological

alignment with the closest compound that allowed the

definition of all dummy atoms.

Volu m e Ap p r oa ch . Compounds 8f, 8h , and 10c

were taken as templates when building the reference

supermolecule. These are the most structurally different

ligands which show high affinity for the PBR. The van

der Waals volume of the supermolecule was then

computed with the molecular modelingprogram Quanta.¹⁸

All the other compounds satisfactorily matched their

analogous ligands constituting the supermolecule by

using their global minimum conformer or conformers,

the energy of which is only 1 kcal/mol above the global

minimum. The best QSAR equations are obtained by

the topological alignment. The following ad-hoc defined

size and shape descriptors were calculated:

a

Reagents: (i) 3 M NaOH, CH₃OCH₂CH₂OH; (ii) (a) PCl₅,

CH₂Cl₂, (b) R₁(R₂)NH, CH₂Cl₂, Na₂CO₃(10% sol).

with the suitable secondary amine in the presence of

Na₂CO₃to afford the expected carboxamide.

Bromination of 2-methyl-4-phenylquinoline^16awith

NBS in the presence of dibenzoyl peroxide gave 2-bro-

momethyl-4-phenylquinoline (28) which was converted

into the tertiary amine 13h by reaction with N-ethyl-

benzylamine in ethanol at reflux in the presence of K₂-

CO₃.

V_in) van der Waals volume of the ligands included

in the supermolecule van der Waals volume.

V_out) excluded van der Waals volume with respect

to the supermolecule volume.

V_dif) (V_in- V_out)/V_sup, the difference between the

above two values normalized to the total supermolecule

volume.

Su r fa ce Ap p r oa ch . A surface enclosing the ag-

gregate aligned ligands, 8f, 8h , and 10c was generated

by means of the program Receptor.¹⁸

Com p u ta tion a l P r oced u r es

To this purpose, a Wyvill soft object function was

used.¹⁹Regions of the receptor surface model were

removed at zones which proved to accept substitution.

Molecular alignment was achieved both by topologically

superimposing the ligands and by using the previously

defined dummy atoms. In the Results and Discussion

section we reported the results obtained with the second

method, since the receptor surface obtained in this way

better rationalizes the observed binding affinity varia-

tion in this molecular series of compounds.

The receptor surface model supports energetic calcu-

lations for the interactions of molecules with the model.

The evaluation was performed on the basis of the

electrostatic potential. The neutral form was considered

for the compounds containing basic atoms. The solvation

correction, which adds a penalty function whenever

polar atoms are placed in hydrophobic regions of the

receptor surface model, did not appear to improve the

results.

The structures of the studied compounds were opti-

mized using the semiempirical AM1 method¹⁷within

the MOPAC 6.0 (QCPE 455) program. The molecular

mechanics correction to amide bond was applied. The

three-dimensional structure of compound 9d , solved by

X-ray crystallography, was used as an input for the

geometry optimization.¹¹The molecules containing at-

oms inclined to be protonated were considered in both

their charged and neutral forms. Enantiomer R was

arbitrarily chosen for compounds showing an asym-

metric carbon in their structures since both R and S

enantiomers of PK11195 (1) were found to be nearly

equipotent toward PBR [(R), IC₅₀) 9 nM; (S), IC₅₀

)

19 nM; (R,S), IC₅₀) 9 nM)].⁸

Molecu la r Align m en t Used in Volu m e, Su r fa ce,

a n d P h a r m a cop h or e Ap p r oa ch es. Since two non-

congeneric molecular sets were considered, we evaluated

two different methods for ligand alignment in the

indirect approaches.

Each ligand was subjected to a few minimization steps

within the model to avoid excessive penalization of those

molecules whose minimum conformation is not compat-

ible with the receptor cavity obtained. It is noteworthy

that the conformation taken up by the ligands inside

the receptor may actually be quite different from the

minimum conformation of the isolated molecule; there-

fore, conformers characterized by an energy of 2 kcal/

mol above the isolated molecule energy minimum were

accepted.

The total interaction energy (E_inter) of the molecule

with the receptor surface model turned out to be the

most interesting descriptor among those calculated by

the program.

P h a r m a cop h or e Ap p r oa ch . Compounds 8f, 8h ,

and 11a were chosen to generate a pharmacophoric

hypothesis with the Catalyst program.¹⁸Compound 11a

In the first we proceeded to a topological alignment

on the basis of the ligand common quinoline or iso-

quinoline nucleus. Whenever that fragment was absent,

the benzene ring directly bound to the carbonyl group

was used.

In the second, we used dummy atoms appropriately

located so that the hypothetical interactions with resi-

dues on the receptor counterpart could be considered:

a dummy atom on the vector defined by the carbonyl

group, 1 Å away from the oxygen; two dummy atoms 2

Å above and below the center of the common bond of

the bicyclic aromatic group, perpendicularly oriented

with respect to the plane defined by the aromatic

moiety; two dummy atoms 2 Å above and below the

center of the pendant phenyl group. The alignment was

carried out by rms minimization of the distance between

the dummy atoms of the different molecules with

1138 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

was chosen because of the peculiar position of its

pendant phenyl group. Four pharmacophoric groups

were used in the model generation process. They are

as follows: (a) one hydrogen bond donor (carbonyl

group); (b) one aromatic ring corresponding to the

pendant phenyl group; (c) one aromatic ring correspond-

ing to the phenyl ring of the quinoline nucleus farthest

from the above-mentioned pendant phenyl ring; (d) one

hydrophobic group corresponding to the aromatic or

aliphatic chains bound to the amide group.

The mutual positions of the pharmacophoric features

were mapped in the 3D space, and a weight was

assigned to each atom on the basis of structure-affinity

relationship (SAFIR) considerations. Thus, we lowered

the tolerance of the hydrogen bond acceptor, since the

carbonyl position is very important for the recognition

of the ligands,¹¹and we allowed more freedom to the

aromatic ring defining the pendant phenyl. Of the

various models (hypotheses) obtained, the ones showing

the highest space occupancy of the pharmacophoric

groups were chosen. The AM1 minimized structures of

the ligands were then evaluated on the basis of the

hypothesis selected.

F igu r e 1. Structural alignment of compounds 8-11 for the

calculation of CoMFA and CoMSIA.

In the evaluation process, the program considers how

each molecule fits the number and space disposition of

the pharmacophores present in the receptor hypothesis

and gives a score called Fit of the molecule.

means of the Protein Health utility (Ramachandran

plots for main chain and side chain conformations,

chirality, solvent accessibility of polar, hydrophilic and

hydrophobic atoms, close contacts, and holes) and the

22

3D Profile Analysis program

framework.

within the Quanta¹⁸

P BR Recep tor Mod elin g. Human, bovine, murine,

and rat PBR sequences were retrieved from the protein

sequences EMBL database (http://www.embl-heidel-

berg.de/srs5). The human PBR hydropatic profile analy-

sis was performed by means of the TopPred method

(http://www.biokemi.su.se/∼server/toppred2/toppred-

server.cgi). The search for three-dimensionally resolved

analogous proteins was performed by means of the

SCOP program (http://scop.rmc-lmb.cam.ac.uk/scop).

CoMF A. A CoMFA²³model was developed using

previously described compounds 8-11.²⁴The molecular

structures of the ligands under investigation were

refined by using the Tripos force field parameter set

available within version 6.3 of the SYBYL molecular

modeling software package.²⁵Conformational analysis

was performed using molecular dynamics together with

simulated annealing, and the geometry of all structures

retrieved at low temperature was fully optimized using

the conjugate gradient algorithm with a convergence

criterion of 0.0001 kcal/mol Å. The most rigid com-

pounds in this series proved to be 9d ,e. Both structures

possess a double bond that reduces flexibility in the

fused azepine ring of the isoquinoline system. Com-

pound 9e is the slightly more active compound, with an

IC₅₀value of 45 nM. The minimum energy conformer

was selected as a reference structure. In this conformer

a torsion angle of -149° was found between the car-

boxyl-oxygen atom of the azepine ring and the planar

aromatic isoquinoline ring system. This torsion angle

((10°) could also be found at all the other compounds

of the set. The alignment for CoMFA was done by

manually fitting the atoms of the aromatic heterocycle

of minimum energy structures with their amide function

exhibiting an appropriate torsion angle to the planar

part of the molecule (Figure 1). For the interaction

energy matrix calculation, the default CoMFA atom

probe parameters were used (i.e., the Lennard-J ones

potential of a sp³type carbon atom [C.3] and charge +1)

and the values were scaled accordingly by means of the

standard CoMFA scaling. The regression analysis was

performed by means of partial least-squares (PLS)

method.²⁶To establish the quality of the CoMFA

derived, cross-validation was done by means of the

The modeling of the receptor and of the complexes

was performed with the QUANTA program.¹⁸The

complete PBR model was obtained by considering the

structural restraints drawn by the Apolipophorin III

(pdb code: 1aep) for the transmembrane region and by

the V9-K27 portion of the Myohemerythrin (pdb code:

2mhr) for the first loop, while the third loop of the

C-terminal part of the receptor was modeled “ab initio”

by considering the conformational preferences of the

short stretch of the amino acids. The Modeller program²⁰

deduces distance and angle constraints from the target

sequences and combines them with energy terms for an

adequate stereochemistry in an objective function, later

optimized in the Cartesian space by conjugate gradients

and molecular dynamics (simulated annealing) methods.

Different models of the PBR receptor were generated

through randomization of the Cartesian coordinates of

all the atoms, with a deviation of (4 Å. Minimization

of the structures was achieved by means of CHARMm

program.²¹The minimization procedure consisted of 200

steps of steepest descent followed by a conjugate gradi-

ent minimization up to a potential energy value of

<0.0001 kcal/mol Å. We used the united atom force field

parameters and a dielectric constant value of 80, since

we were interested in the loop portions of the receptor.

The interaction energies at the binding site were

calculated with a dielectric constant ꢀ ) 4r. A technical

check of the quality of the model was performed by

leave-one-out method. An r²of 0.592 was found

cv

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1139

Ta ble 2. PBR Binding Affinities of Compounds 4-6 and

12-16 [IC₅₀(nM) + SEM]^a

(s_PRESS: 0.720 (three components), n ) 23), indicating

an acceptable predicting capability of the model. The

final model without validation was found to give a

conventional r²of 0.949 (s ) 0.254, n ) 23) and was

used for affinity prediction of newly synthesized com-

pounds 13b and 16. The affinity values of compounds

13b and 16 (containing an amino function) were pre-

dicted with the molecules both in the neutral and in the

protonated form.

CoMSIA. Another 3D QSAR model was developed

based on previously described compounds 8-11 by

means of the CoMSIA approach.²⁷The same structural

alignment as for the CoMFA was taken, and the default

attenuation factor (0.3) was selected for the similarity

index calculation in steric and electrostatic fields. The

regression analysis was performed by means of PLS. To

determine the quality of the CoMSIA derived, cross-

validation was done by means of the leave-one-out

method. An r²_cvof 0.538 was found (s_PRESS: 0.784 (three

components), n ) 23), indicating a predicting capability

comparable to the CoMFA method. The final model

without validation was found to give a conventional r²

of 0.908 (s ) 0.334, n ) 23) and was used for affinity

prediction of the newly synthesized compounds 13b and

16. The affinity values of compounds 13b and 16

(containing an amino function) were predicted with the

molecules both in the neutral and in the protonated

form.

compd

R

R₁

R₂

IC₅₀

4a ^b

4b^b

5a ^b

5b^b

6a ^b

6b^b

12a ^c

12b

13a

13b

13c

13d

13e

13f

13g

13h

14a

14b

14c

15a

15b

15c

15d

16

5

27

410

210

700

540

CH₂N(Et)Bn

N(Et)Bn

Cl

3700 ( 250

1185 ( 134

229 ( 44

373 ( 58

456 ( 132

1317 ( 367

2479 ( 360

872 ( 293

2895 ( 532

3400 ( 280

CH₂THIQ^d

CONMe₂

CONEt₂

CON(n-Pr)₂

CON(Me)Ph

CON(Me)4-Cl-Ph

CON(H)n-Pr

CON(H)Bn

H

Resu lts a n d Discu ssion

St r u ct u r e-Affin it y R ela t ion sh ip s. The impor-

tance of the carbonyl group in the interaction of PBR

ligands with their receptors has been stressed by several

authors.^3,8,9This is one of the most important groups

in the structure of 1 which is, in principle, capable of

giving long-range (strong and directional) electrostatic

interactions with the receptor.

In a previous paper we described the results obtained

in a series of conformationally constrained carboxamide

derivatives 7-11 (see Table 1).¹¹These compounds are

closely related to 1, while they differ from compounds

6 because the amide moiety is attached to C-2 of the

bicyclic heteroaromatic system instead of C-3 as in

compounds 6. The work performed on conformationally

constrained derivatives 7-11 confirmed the essential

role of the carbonyl function as the primary pharma-

cophoric element, and the importance of the role of both

the amide substituents and the pendant phenyl ring for

which a dispersive nature of the interactions with PBR

binding site was suggested.

To obtain information on the binding modalities of

compounds 6, their main structural features (the car-

boxylic ester moiety, the N-ethylbenzylamino group, and

the bicyclic heteroaromatic system with the pendant

phenyl ring) were subjected to structural modification.

The newly synthesized compounds (12-16) were

tested for their potential ability to displace [³H]-1 from

PBR in rat brain cortex (in the same test system used

for compounds 7-11) in comparison with 1, and the

results of these studies are summarized in Table 2.

The results obtained clearly show that the replace-

ment of the ester function of 6a with a secondary (13f)

or tertiary amides (13a -c) affords compounds with

similar micromolar range affinities, when compared

Et

n-Pr

Ph

Me

Et

n-Pr

Ph

>10000

n-Pr

Me

Et

n-Pr

Me

>10000

8209 ( 1016

8867 ( 1164

>10000

4.1 ( 1.03

2.2 ( 0.3

1

a

b

Each value is the mean ( SEM of three determinations. The

IC₅₀value of these compounds was taken from the literature

(4a ,b: see ref 8; 5a ,b and 6a ,b: see ref 10). ^cSee ref 10. THIQ

d

) (1,2,3,4-tetrahydroisoquinolin)-2-yl.

with ester 6a . In addition, the environment of the

carbonyl amide seems to be relatively sensitive to steric

hindrance since the increase in size of the amide

substituents results in a progressive decrease in affinity

(compare compounds 13b-e with 13a ). These findings

appear to be in disagreement with the results described

in the literature.^3,8Moreover, although the variation in

the PBR affinity consequent to the removal of the

lipophilic amide group in compounds 13a ,b (compare

13a ,b with 13h ) is significant and suggests a specific

role of this group in the ligand-receptor interaction, our

work (see also ref 11) clearly shows that a suitably

oriented lipophilic amide group plays a more substantial

role and contributes to the binding strength much more

than the one in the 3-position of the quinoline nucleus

of 13a -g. Taken together, these results suggest that

the lipophilic amide groups in 3-position of the quinoline

nucleus of compounds 13a -g occupy a receptor area

1140 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

different from the one occupied by the lipophilic amide

groups of the high affinity PBR ligands.

Comparison of compounds 6a ,b with 12a and 13h

suggested that the N-ethylbenzylamino group of 6a ,b

plays a significant role in the interaction of these

compounds with PBR. To characterize better the role

of the benzylamino side chain in these compounds, a

short series of 2-methyl derivatives was prepared (14a -

c). Their inactivity highlights the importance of the

presence of a large lipophilic substituent in the 2-posi-

tion of the bicyclic heteroaromatic system (compare

14a -c with 13b-d ) for the binding to PBR. It is also

noteworthy that inclusion of the N-ethylbenzylamino

group into a rigid tetrahydroisoquinoline ring did not

significantly modify the affinity (compare compound 6a

with 12b).

F igu r e 2. Schematic representation of the interactions of

compound 16 with the PBR binding site.

The bicyclic heteroaromatic system with the pendant

phenyl ring of phenylquinolinecarboxamides 13a -g also

plays a major role in the molecular recognition process

as its simplification into a benzo ring led to the

significantly less potent ligands 15 (compare 13b-d

with 15a -c).

The integration of the information obtained from the

SAFIR analysis discussed above with that obtained from

conformationally constrained derivatives 7-11 led to

the design of the positional isomer of 13b in which the

substituents in positions 2 and 3 of the quinoline

nucleus are inverted (compound 16). It is interesting to

note that carboxamide derivative 16, bearing a N-

ethylbenzylamino group in the 3-position of the quino-

line nucleus, shows a nanomolar PBR affinity compa-

rable with that shown by 1.

On the whole, these findings lead to some consider-

ations: (a) the presence of a carbonyl dipole suitably

located (e.g., in 2-position of the bicyclic aromatic

system) and oriented (e.g., in perpendicular position

with respect to the bicyclic system) is needed for high

PBR affinity; (b) lipophilic substituents attached (di-

rectly or rather by interposition of -C(dO)-N- fragment)

to the 2-position are favorable for the interaction with

PBR; (c) the presence of a bicyclic aromatic or het-

eroaromatic system bearing a suitably located pendant

phenyl ring is of fundamental importance for the

interaction with PBR; (d) a zone of tolerance to large

substituents seems to be located in the PBR site

interacting with the 3-position of quinoline nucleus; and

(e) the PBR binding site seems to tolerate well the

positive charge presumably borne by the tertiary amine

nitrogen of 16 at physiological pH. These considerations

on the PBR interaction of carboxamide derivatives

related to 1 are easily translated into the interaction

model of compound 16 shown in Figure 2.

Qu a n tita tive Str u ctu r e-Affin ity Rela tion sh ip s

(QSAR). (a ) Recep tor Ma p p in g. To obtain a quanti-

tative rationalization of the structure-affinity relation-

ships, it is of primary importance to define molecular

descriptors with high information content. A good

quantitative rationalization of the affinity variation of

26 molecules in the series of compounds 7-11 has

recently been obtained¹¹through comparison of the van

der Waals volumes of the different ligands with that of

an ad-hoc defined reference supermolecule.²⁸

showing the highest affinities reflects the overall shape

and conformational freedom of the PBR binding site.

Thus, compounds 8f, 8h , and 10c were chosen to

constitute the supermolecule and were topologically

superimposed as described in the methods section.

Molecular descriptors such as V_in, V_out, and V_difwere

then calculated. V_inand V_outare, respectively, the inner

and outer van der Waals molecular volumes of the

ligands considered with respect to the resultant refer-

ence volume of the supermolecule, and, from an inter-

pretative point of view, they mimic the short-range

attractive (V_in) and repulsive interactions (V_out) with the

receptor. V_difis defined as the difference between V_in

and V_outand is normalized with respect to the volume

of the supermolecule.

The V_difdescriptor gave the most significant results,

as shown by the previously reported¹¹QSAR model

developed for compounds 7-11

pIC₅₀) 10.25((0.97)V_dif+ 1.43((0.56)

(1)

n ) 26, r²) 0.83, s²) 0.21, F ) 111.09

where n is the number of compounds considered, r the

correlation coefficient, s the standard deviation, and F

the value of the Fisher ratio; the numbers in parenthe-

ses are 95% confidence intervals of the regression

coefficient and of the intercept.

Recalling the mathematical formulation of this mo-

lecular descriptor, the correlation suggests that, in the

molecular series of the compounds considered, the

binding affinities are modulated by the molecular shape

of the amide substituents through the optimization of

dispersive and steric interactions.

The previously chosen supermolecule is surprisingly

able to rationalize the affinity of the additional ligands

considered in this work. The following correlation is

obtained by considering the ligands of Table 3:

pIC₅₀) 5.86((0.58)V_dif+ 4.09((0.28)

(2)

n ) 44, r²) 0.71, s²) 0.36, F ) 103.28;

16 omitted

However, the model fails to predict the behavior of

compound 16 (Figure 3). This ligand is the only one in

the quinoline set with a basic nitrogen having a high

affinity for the peripheral benzodiazepine receptor. The

In this approach we assume that the volume obtained

by superimposing the most structurally different ligands

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1141

Ta ble 3. PBR Affinity Indexes (pIC₅₀) and Molecular

Descriptors for Compounds 1, 4-13, 15, and 16

compd pIC₅₀V_difV_out(Å³) V_in(Å³) E_inter(kcal/mol) Fit

Ta ble 4. Comparison of Experimental versus Predicted

Affinities of Compounds 1, 8-11 (Learning Set), 13b, and 16

(Test Set) from CoMFA and CoMSIA Models

CoMFA

CoMFA CoMSIA CoMSIA

1

4a

4b

5a

5b

6a

6b

7a

7b

7c

8a

8b

8c

8d

8e

8f

8g

8h

8i

8j

8k

8l

9a

9b

8.50 0.62

8.30 0.51

7.57 0.48

6.39 0.37 120.75

6.68 0.37 120.38

6.15 0.35 111.50

6.27 0.29 137.62

13.75

60.37

63.75

285.00

282.88

274.25

281.13

282.00

265.88

264.63

274.00

270.38

277.63

250.88

282.50

253.25

252.75

269.38

304.00

304.75

322.63

308.88

314.88

287.75

288.00

234.88

246.50

270.88

268.25

280.50

255.63

294.50

321.50

291.88

304.50

224.38

310.63

289.50

295.13

295.38

303.00

299.88

279.75

288.88

255.75

198.63

217.63

296.25

-3.21

-3.18

-3.30

-1.98

-1.70

5.36

3.68

3.83

3.81

2.57

2.51

2.41

2.33

2.67

2.80

2.58

2.24

3.53

2.12

2.15

4.00

3.82

3.86

3.76

3.78

2.20

2.58

2.62

2.92

3.27

2.55

3.08

3.28

3.11

3.62

2.38

2.86

2.38

2.77

2.66

2.89

2.83

2.34

2.67

2.25

1.98

1.93

3.74

pred IC₅₀pred IC₅₀pred. IC₅₀pred. IC₅₀

(nM)

neutral

IC₅₀(nM)

( SEM

(nM)

NH⁺

(nM)

neutral

(nM)

NH⁺

compd

1

8a

8b

8c

8d

8e

2.2 ( 0.3

230 ( 70

13 ( 4.9

2.0

94

25

6.7

118

24

770

1492

93

8.4

2.4

5.3

13

-1.84

-3.38

-3.08

-3.26

-2.80

-2.83

-3.11

-1.76

-1.93

-3.09

-3.23

-3.45

-3.31

-3.49

-3.48

-1.34

-2.05

-2.26

-1.81

-2.44

-2.73

-3.41

-3.96

-3.53

-3.70

-2.34

-2.12

-1.64

8.14

6.09 0.43

5.92 0.54

6.77 0.52

6.64 0.49

7.89 0.62

5.92 0.46

5.77 0.43

6.57 0.50

8.68 0.70

8.54 0.69

8.68 0.74

8.01 0.66

8.47 0.67

8.19 0.59

8.06 0.57

6.21 0.44

6.32 0.45

6.10 0.53

6.31 0.53

7.34 0.55

6.26 0.51

7.95 0.65

8.51 0.74

8.05 0.60

8.00 0.63

5.43 0.38

5.92 0.48

6.64 0.35 135.25

6.43 0.31 160.25

6.34 0.24 191.00

5.89 0.27 183.63

5.60 0.22 201.75

6.06 0.28 157.63

5.54 0.21 197.50

5.47 0.34 108.63

5.09 0.14 138.62

5.05 0.15 150.25

8.39 0.21 205.50

86.38

37.50

50.75

38.12

11.63

53.00

64.25

52.00

0.00

1200 ( 180 1123

1700 ( 230 2950

270 ( 66

2.1 ( 0.9

2.9 ( 0.5

2.1 ( 0.6

9.8 ( 2.0

3.4 ( 1.0

6.4 ( 1.1

8.8 ( 1.5

620 ( 100

480 ( 29

780 ( 20

490 ( 47

45 ( 15

550 ( 93

11 ( 2.6

3.1 ( 0.8

8.9 ( 2.5

10 ( 0.7

373 ( 58

4.1 ( 1.0

93

4.1

8f

8g

8h

8i

8j

8k

8l

9a

9b

9c

9d

9e

10a

10b

10c

11a

11b

13b

16

2.5

2.4

5.3

4.6

5.8

3.8

2.4

9.8

7.3

4.88

0.00

709

378

2938

261

381

43

539

8.2

1.3

23.75

29.50

40.63

42.87

50.13

38.62

35.25

42.75

31.50

10.13

0.00

887

516

265

65

441

17

4.0

10

24

9c

9d

9e

13.8

12.7

177

13

10a

10b

10c

11a

11b

12a

12b

13a

13b

13c

13d

13e

13f

13g

13h

15b

15c

16

151

44

132

13

2397

7

31.62

28.88

55.87

98.12

On the other hand, both CoMFA and CoMSIA models

developed on compounds 8-11 exhibit a satisfactory

performance in the affinity prediction for compound 16

(Table 4). Interestingly, the affinity of compound 13b

(the positional isomer of 16) appears to be even better

predicted by the CoMFA model and by the CoMSIA

model in the case of the neutral molecule. Thus, the 3D-

QSAR models (CoMFA and CoMSIA) obtained appear

to predict quite correctly the affinity of molecules

showing significant structural differences with respect

to those considered in the training set. In particular, it

should be stressed that the training set is composed of

molecules having lipophilic amide side chains in the

2-position of the quinoline nucleus (or equivalent) and

small substituents in the 3-position, while a bulkier and

basic N-ethylbenzylamino group is present in the 3-po-

sition of the quinoline nucleus of 16 and a lipophilic

amide side chain occupies the 3-position of the quinoline

nucleus of 13b. Taken together, these observations

suggest that either these 3D-QSAR models tolerate

quite well the additional steric bulk in a region for which

the model is relatively trained (Figure 4a) or the

information on the lipophilic amide side chain is domi-

nant (Figure 4b).

-3.06

5.84

10.21

2.96

13.58

-1.32

-3.57

-1.11

-3.51

Therefore, to overcome the steric drawback in the

volume approach previously described and complicate

the binding site model by explicitly taking into account

the electrostatic contribution to the ligand-receptor

complementarity as CoMFA does, a different approach

has been used in this work. It consists of constructing

the surface enclosing a volume common to all the

aligned molecules used in its building. The shape

obtained is assumed to be complementary to the shape

of the receptor site itself. The electrostatic potentials of

the ligands are then mapped on such a surface.²⁹Since

the receptor model is constructed from an aligned

bundle of molecules, each surface point is characterized

F igu r e 3. Correlation between the PBR binding affinity

(pIC₅₀) and the V_difdescriptor. Compound 16, not included in

the regression, is represented as b.

failure is probably due to the nature of the supermol-

ecule used, which tolerates only a limited steric bulk in

position 3 of the quinoline nucleus (or, more generally,

additional steric bulk).

1142 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

F igu r e 4. (a) Relaxed stereoview of compound 16 in CoMSIA steric sdev.*coeff. contour plot. The yellow contoured areas represent

regions in which steric bulk is detrimental for activity; no substucture of this active molecule penetrates this region.

Correspondingly, the green contoured area indicates a zone were steric interactions are favored. (b) Relaxed stereoview of compound

16 in CoMSIA electrostatic sdev.*coeff. contour plot. The important role of the amide dipole becomes obvious since there is a large

red contoured area indicating high electron density in this area favorable to high affinity. On the other hand, the electron-

deficient heteroaromatic ring system overlaps partially with the blue region where positive charge is favorable to high affinity.

by a value that is equal but opposite to the average

value of the electrostatic potential of the closest atoms

in each molecule. Moreover, receptor flexibility can be

taken into account by opening the surface model in those

parts suggested to be particularly tolerant to substitu-

tion by SAFIR analysis. The receptor model obtained

is shown in Figure 5a.

interaction energy (repulsive interactions), while 7a , 7b,

8c, and 15b are overestimated by the model. The

statistical indices of this equation are comparable to

those of the previous model (eq 2) obtained for all 44

ligands considered. However, this model is able to

predict the affinity of compound 16, as shown in Figure

6 (pIC_{50_exp}) 8.39, pIC_{50_calc}) 8.10).

The model was built with the same molecules em-

ployed in the previous approachs8f, 8h , and 10csand

the best results were obtained when the alignment was

performed by means of dummy atoms (see methods

section). The compatibility of each ligand to the model

is evaluated, and the energy interaction (E_inter) gives a

quantitative description of the ligand-receptor site

Finally, the fourth approach used is substantially

different from the previous ones. It consists of defining

a receptor hypothesis on the basis of pharmacophoric

groups shared by a subset of ligands which were chosen

as representatives of the set and appropriately located

in the 3D space.³⁰

complementarity. The more negative the value of E_inter

,

In this case, shape is not taken into account, the only

consideration concerns whether the ligand possesses the

desired characteristics at the defined positions.

the more stable the interaction between ligands and

receptor.

The QSAR model obtained by considering 34 com-

pounds is

The molecules chosen in the building of the hypoth-

eses are 8f, 8h , and 11a (Chart 2). The last one shows

high affinity and a peculiarity in the position of the

pendant phenyl group. The hypothesis chosen for the

evaluation of the ligands studied is shown in Figure 5b

and assumes four pharmacophoric groups: (a) one

hydrogen bond acceptor (carbonyl group); (b) one aro-

matic ring corresponding to the pendant phenyl ring;

(c) one aromatic ring corresponding to the homocyclic

pIC₅₀) -1.14((0.13)E_inter+ 4.08((0.37) (3)

n ) 34, r²) 0.70, s²) 0.38, F ) 73.33

where ligands 7a , 7b, 8c, 6a , 13b, 15b, 13d , 13e, 13f,

and 13g have been omitted.

All these ligands show low binding affinity to the

receptor: 6a , 13b, 13d , 13e, 13f, and 13g have positive

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1143

F igu r e 5. (a) Receptor model of the binding site obtained by overlapping ligands 8f, 8h , and 10c. (b) Pharmacophoric hypothesis

obtained by considering ligands 8f, 8h , and 11a . The fitting of compound 4b to the hypothesis is shown.

pIC₅₀) 1.54((0.13)Fit + 2.36((0.38)

(4)

n ) 44, r²) 0.78, s²) 0.28, F ) 144.72

It can be noted (Figure 7) that ligands showing low

and high affinity are clearly clustered in the graph, the

threshold for the Fit descriptor being equal to 3.

In addition, the activity of compound 16 is adequately

predicted (pIC_{50_exp}) 8.39, pIC_{50_calc}) 8.12).

Any attempt to improve the linear QSAR models

presented by multiregression analysis of the above-

mentioned ad-hoc descriptors with a large variety of

electronic and reactivity indexes was unsuccessful. This

result seems to support the hypothesis that specific

highly directional interactions with a suitable hydrogen-

bonding donor amino acid residue of the receptor can

easily be achieved from the point of view of the elec-

tronic requirements (as the electronic character of the

carbonyl group is quite similar in all the ligands

considered) once the conformational requirements are

satisfied.

(b) Recep tor F ittin g. (P er ip h er a l Ben zod ia z-

ep in e Recep tor Mod elin g.) The modeling of the 3D

structure of the receptor is based on the knowledge of

its primary structure (sequence), the results of molec-

ular biophysics and biochemistry studies available in

the literature, and the hints from the models obtained

with the indirect approaches used.

F igu r e 6. Correlation between the PBR binding affinity

(pIC₅₀) and the E_interdescriptor. Compound 16 is represented

as b.

ring of the quinoline nucleus; (d) one hydrophobic group

for the aliphatic and aromatic chains bound to the amide

group.

A score (Fit) is then assigned to the molecules under

study on the basis of the number of matched functional

groups and of the possible penalization for incomplete

superimposition or orientation.

On the whole, the descriptor is able to account for 78%

of the variance in the receptor affinity for the 44

compounds considered:

The alignment of the available PBR sequences, re-

ported in Figure 8, has emphasized the high degree of

conservation of some parts of the receptor sequences in

1144 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

was performed on the sequences of the same length of

the loop studied whose distances between N-terminus

and C-terminus fit the desired range. Among the

fragments retrieved, those that showed the residues

presumably responsible for the interaction with the

ligands located in a favorable mutual position (i.e., near

enough to be considered part of the same site) were

chosen.

The structural restraints derived by these templates

for the first loop and the Apolipophorin III for the

transmembrane regions of the receptor were used to

obtain several 3D models of PBR. The short third loop

and the C-terminal part were modeled ab initio.²⁰

Bin d in g Site Sp ottin g. Once a family of 3D PBR

models was obtained by randomisation of the Cartesian

coordinates,²⁰a tentative binding site for the ligands

was identified.

F igu r e 7. Correlation between the PBR binding affinity

(pIC₅₀) and the Fit descriptor. Compound 16 is represented

as b.

Site-directed mutagenesis studies⁷have suggested

that residues R24, E29, K39 and L31 may be part of

the binding site for 2 (agonist). These residues are

located in the first cytoplasmic loop and are conserved

through human, bovine, murine, and rat species. The

same amino acids play, on the other hand, a marginal

role in the interaction with 1; therefore, this antagonist

seems to occupy a similar but not identical binding site.

Moreover, studies designed to elucidate the functional

and structural properties of TspO of Rhodobacter

sphaeroides,^31bwhich is assumed to be a model for the

mammalian PBR, have pointed out the importance of

residues L34, K36, W39, and W44 (corresponding to

L37, K39, W42, and W47 of the PBR sequence) in the

first cytoplasmic loop.

the different species. The degree of conservation is

important in the determination of membrane topology.

Highly conserved residues are usually associated with

structural or functional roles, while mutated residues

(nonconservative mutations) may be important for

specificity, or may simply be located in structurally or

functionally unimportant portions of the receptor.

Experimental studies for PBR topography determi-

nation⁵and the analysis of the hydropathy profile have

highlighted the presence of five transmembrane R-he-

lices composed of 21 amino acids, as shown in Figure 8:

TM1 (3-23), TM2 (45-65), TM3 (82-102), TM4 (109-

129), TM5 (135-155).

The search for analogous proteins with a resolved 3D

structure was unsuccessful. In fact, PBR does not show

any analogy with other known proteins, apart from a

significant relationship with CrtK, a member of the

protein family which is responsible for the carotenoid

biosynthesis.³¹Unfortunately, the CrtK 3D structure is

yet to come. Therefore, the assembly of the five trans-

membrane helices has thus been carried out with

Apolipophorin III (1AEP) as a template. This is a

functionally nonrelated lipoprotein of the African mi-

gratory locust whose transmembrane topology has been

resolved at atomic level.³²

Only few mutagenesis experiments (chimeric recep-

tors and site-directed mutagenesis) have been published

up to now.^5,7,31b,33They have pointed out that the ligand

binding site is located in the loop region that develops

outside the mitochondria, in the cell cytoplasm. Thus,

we concentrated our efforts in the modeling of this

region. In particular, the first cytoplasmic loop that

connects the C-terminus of the first helix to the N-

terminus of the second is composed of 19 amino acids

and seems mostly involved in the interaction with the

ligands. The third loop connects the third with the

fourth helix and is composed of six residues only. In

addition, since deletion studies³³have shown that the

last 13 amino acids of the PBR C-terminus may be

deleted without significant effects on the binding of 1

and 2, only eight terminal amino acids were considered

in our 3D model.

We examined the different models with molecular

graphic methods to select the ones which show the

maximum complementarity between the position and

characteristics of the mutated residues and the phar-

macophoric elements of compound 2. Further selection

was made on the basis of energy considerations. We

manually docked the agonist into the hypothetical

binding site of the different PBR 3D structures, and the

complexes we obtained were subjected to minimization.

Different orientations of the agonist gave ligand-

receptor interaction energies which were taken into

account to classify the models and to select the one

shown in Figure 9a (binding site of 2).

Such a model suggests that the binding site is formed

by the mutated residues R24, E29, K39, and L31 and

by P40, S41, W42, W107, and W161.

The mutation of R32 to glycine lowers the agonist

binding constant by a factor of 10. This residue seems

to have a structural role in the model. The electrostatic

interaction between R32 and the C-terminal part of PBR

determines the conformation of the I loop. K39 estab-

lishes an electrostatic interaction with the carbonyl

oxygen of 2. It is noteworthy that the Cl-substituted

condensed phenyl group is settled in a hydrophobic

pocket defined by W42, W107, and W161, while the

chlorine atom on the pendant phenyl ring gives rise to

dispersive interactions with a more polar environment

defined by the ionic couple R24-D157.

On the basis of these results, the receptor hypothesis

produced by the pharmacophore approach applied to the

ligands studied shows, as far as interactions and the

To model the first cytoplasmic loop, formed by 19

residues, we searched on a database of protein frag-

ments of known three-dimensional structure. The search

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1145

F igu r e 8. Alignment of the PBR sequences (human, bovine, murine, and rat). The five transmembrane domain have been

highlighted. Amino acid residues not conserved among species are boxed.

distances between key elements are concerned, the

correspondences listed in Table 5.

Con clu sion s

Among the different chemical classes of PBR ligands,

all the known ligands are highly lipophilic and some-

times their insolubility restricts their use in different

pharmacological assays. Because most of our compounds

bear a tertiary amino group, which is likely to be

protonated at physiological pH, they constitute an

original series of water-soluble ligands of PBR sites. In

particular, compound 16 represents, to the best of our

knowledge, the first example of PBR ligand with affinity

in the low nanomolar range bearing a basic nitrogen

atom, and it appears to be a new interesting pharma-

cological tool for the studies of the PBR function and/or

an interesting candidate for [¹¹C]-labeling to be used in

PET studies.

The combined use of different computational ap-

proaches allowed us to obtain a quantitative rational-

ization of the structural characteristics of peripheral

benzodiazepine receptor antagonists and to correlate the

interaction effects with the receptor to the functional

role of the main structural components. Size and shape

descriptors, defined with respect to a supermolecule,

turned out to be efficient in describing affinity changes.

They quantify the short-range intermolecular interac-

tion contributions to the ligand-receptor interaction,

which seem to be, in this case, the determinant ones

for the modulation of the affinity. In fact, similar

descriptors which also take into account electrostatic

interactions have not generated better QSAR models

either from the interpretative viewpoint or from the

predictive one.

The three ligands 8f, 8h , and 10c, used to build the

indirect approach supermolecule, were also accom-

modated in this site and the complexes obtained un-

derwent minimization. Figure 10 shows in detail the

interaction modes between two of these ligands (8f and

10c) and the amino acids of the antagonist binding site.

The residue S41 has a key role for the directional

interaction with the carbonyl moiety, and L31 estab-

lishes dispersion interactions with the condensed ben-

zene ring. Mutation of these residues causes a weak

decrease of the binding constant of compound 1 (re-

ported in Figure 10 as fitted into its putative binding

site). The coherence between the results obtained with

the direct approach and the receptor hypothesis built

by the indirect approaches, especially concerning the

high flexibility regions, can be estimated by comparing

Figures 5 and 10d. On the contrary, the indirect

approaches seem to allow much more conformational

freedom to the position of the carbonyl moiety with

respect to the 3D model of the receptor. This might

provide a possible explanation for the overestimation

of compounds 7a ,b and 8c by eq 3.

Finally, this receptor site is perfectly able to host a

PBR ligand structurally unrelated to the previous ones

such as alpidem (3), the lead compound for the imida-

zopyridine ligands³⁴(Figure 9b).

Moreover, by assuming the same binding modes of 1

for compound 16, its higher affinity with respect to its

analogue 4b can be explained by the additional interac-

tions of the basic nitrogen in position 3 of the quinoline

nucleus with D157 (Figure 9c). Therefore, on the basis

of the results obtained by the different approaches, we

can suggest site-specific mutations on hPBR which

involve amino acids W42, W107, W161, S41, and D157.

The results obtained suggest that all the ligands

considered have the electronic requirements for specific

and directional interactions of the carbonyl oxygen with

a hydrogen bond donor amino acid of the receptor, but

that not all of them can satisfy the conformational

1146 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

Ta ble 5. Correspondences between the Results Obtained by

Receptor Mapping and Receptor Fitting Approaches

receptor mapping

receptor fitting

(pharmacophoric hypothesis)

(structural hypothesis)

hydrogen bond acceptor

two aromatic rings:

(1) pendant phenyl

(2) condensed ring

hydrophobic group

Ser41

Trp42

Leu31

Trp107, Trp161

restraints imposed by the receptor. The receptor hy-

pothesis obtained by the pharmacophore approach has

provided, beside a good QSAR model, useful hints for

the identification of the antagonist binding site in the

three-dimensional structure modeling process of the

complexes. The computational simulation of the ligand-

receptor interactions gives a detailed picture of the

complementarity obtained from the interacting mol-

ecules and allows us to speculate on the mechanistic

role of the structural components. However, the ligand-

receptor models proposed may be challenged when

further experimental data are available. All experimen-

tal information must be exploited in each step of the

modeling procedure, and only an interactive updating

process of the developed model/experimental check/

experiments project can confirm or invalidate its ad-

equacy or improve its predictivity.

Exp er im en ta l Section

Melting points were determined in open capillaries on a

Gallenkamp apparatus and are uncorrected. Microanalyses

were carried out using a Perkin-Elmer 240C elemental ana-

lyzer. Merck silica gel 60 (70-230 mesh or 230-400 mesh)

was used for column chromatography, and Riedel-de Haen DC-

1

Mikrokarten SI F 37341 were used for TLC. H NMR spectra

were recorded with a Bruker AC 200 spectrometer in the

indicated solvents (TMS as internal standard); the values of

the chemical shifts are expressed in ppm and coupling

constants (J ) in hertz (Hz). Mass spectra (EI, 70 eV) were

recorded on a VG 70-250S spectrometer. NMR spectra and

elemental analyses were performed by the Dipartimento

Farmaco Chimico Tecnologico, Universita` di Siena. Mass

spectra were performed by Centro di Analisi e Determinazioni

Strutturali, Universita` di Siena.

Eth yl 2-[(1,2,3,4-Tetr a h yd r oisoqu in olin e)-2-m eth yl]-4-

p h en ylqu in olin e-3-ca r boxyla te (12b). To a solution of ethyl

2-chloromethyl-4-phenylquinoline-3-carboxylate¹³(0.33 g, 1.0

mmol) in acetonitrile (30 mL) was added sodium iodide, and

the resulting mixture was stirred at room temperature for 0.5

h. Commercially available 1,2,3,4-tetrahydroisoquinoline (0.16

g, 1.2 mmol) and sodium carbonate (0.42 g, 4.0 mmol) were

then added, and the reaction mixture was refluxed for 5 h.

After cooling, the mixture was filtered on Celite, and the

filtrate was evaporated in vacuo. The residue was dissolved

in methylene chloride, and the organic phase was thoroughly

washed with H₂O, dried, and evaporated under reduced

pressure to give a nearly pure light yellow oil which crystal-

lized on standing. A recrystallization from methanol gave an

analytical sample as colorless prisms (yield 70%) melting at

1

119-121 °C. H NMR (CDCl₃): 0.59 (t, J ) 7.3, 3H), 2.8 (bs,

4H), 3.58 (q, J ) 6.9, 2H), 3.77 (bs, 2H), 4.19 (s, 2H), 6.89-

7.10 (m, 4H), 7.25-7.49 (m, 5H), 7.55-7.85 (m, 2H), 8.15 (d,

J ) 8.3, 1H). MS: m/z 422 (M⁺, 1). Anal. (C₂₈H₂₆N₂O₂) C, H,

N.

Gen er al P r ocedu r es for th e P r epar ation of Hydr oxym -

eth yl Ca r boxa m id es 20a -e, 21, a n d 22a -d . Meth od A. A

2.5 M solution of trimethylaluminum in hexane (0.8 mL, 2.0

mmol) was slowly added to a solution of 2.0 mmol of the

suitable amine hydrochloride in 5 mL of dry methylene

chloride under nitrogen at room temperature. The mixture was

stirred at room temperature for 15 min, and 2.0 mmol of the

F igu r e 9. (a) Binding site of agonist 2. Only residues

contributing to the total interaction energy (E ) -42.79 kca/

mol), computed as E ) E_complex- E_ligand, with more than 2 kcal/

mol are shown. (b) Binding site of compound 3, structurally

unrelated to the previous ligands (E ) -46.72 kcal/mol). (c)

Binding site of compound 16 (E ) -73.93 kcal/mol).

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1147

F igu r e 10. (a) Binding site of 1 (E ) -62.09 kcal/mol). (b) Binding site of compound 8f (E ) -66.69 kcal/mol). (c) Binding site

of compound 10c (E ) -62.16 kcal/mol). (d) Superimposition of the binding sites of 1, 8f, and 10c, showing the receptor flexibility.

appropriate lactone (17-19) was added. The mixture was

warmed at 25-41 °C under nitrogen until TLC indicated that

the reaction was completed. The reaction was carefully

quenched with dilute HCl and extracted with methylene

chloride. The organic extract was dried (MgSO₄) and concen-

trated to afford the expected hydroxymethyl carboxamide

which was purified by flash chromatography with ethyl

N-(4-Ch lor op h en yl)-2-h yd r oxym eth yl-N-m eth yl-4-p h e-

n ylqu in olin e-3-ca r boxa m id e (20e). The title compound was

obtained in 92% yield from lactone 17 and 4-chloro-N-methy-

laniline hydrochloride. ¹H NMR (CDCl₃): 2.88 (s), 3.20 (s),

4.68-5.20 (m), 6.22-6.26 (m), 6.58 (d, J ) 7.5), 6.77-6.95 (m),

7.28-7.80 (m), 8.09-8.22 (m).

N,N-Diet h yl-3-h yd r oxym et h yl-4-p h en ylq u in olin e-2-

ca r boxa m id e (21). The title compound was obtained in 98%

yield from lactone 18 and diethylamine hydrochloride. ¹H NMR

(CDCl₃): 1.23-1.40 (m, 6H), 3.42 (q, J ) 7.1, 2H), 3.69 (q, J )

7.2, 2H), 4.06 (t, J ) 5.3, 1H), 4.40 (d, J ) 5.4, 2H), 7.35-7.75

(m, 8H), 8.12 (d, J ) 8.3, 1H).

1

acetate-methanol (9:1) when necessary. The H NMR spectra

of those amides, the nitrogen of which bears two different

substituents, show the presence of two different rotamers in

equilibrium. For the sake of simplification the integral intensi-

ties have not been given.

N,N-Dim eth yl-2-h yd r oxym eth yl-4-p h en ylqu in olin e-3-

ca r boxa m id e (20a ). The title compound was obtained in 80%

yield from lactone 17 and dimethylamine hydrochloride. ¹H

NMR (CDCl₃): 2.52 (s, 3H), 2.78 (s, 3H), 4.79 (d, J ) 15.6,

1H), 5.05 (m, 2H), 7.33-7.64 (m, 6H), 7.67-7.71 (m, 2H), 8.09

(d, J ) 8.1, 1H).

N,N-Diet h yl-2-h yd r oxym et h yl-4-p h en ylq u in olin e-3-

ca r boxa m id e (20b). The title compound was obtained in 64%

yield from lactone 17 and diethylamine hydrochloride. ¹H NMR

(CDCl₃): 0.72-0.94 (m, 6H), 2.63-2.77 (m, 1H), 2.91-3.13 (m,

2H), 3.57-3.75 (m, 1H), 4.75-5.15 (m, 3H), 7.36-7.58 (m, 6H),

7.67-7.74 (m, 2H), 8.13 (d, J ) 8.4, 1H).

N,N-Dip r op yl-2-h yd r oxym eth yl-4-p h en ylqu in olin e-3-

ca r boxa m id e (20c). The title compound was obtained in 70%

yield from lactone 17 and dipropylamine hydrochloride.¹H

NMR (CDCl₃): 0.59-0.73 (m, 6H), 1.05-1.36 (m, 4H), 2.47-

2.59 (m, 1H), 2.80-2.99 (m, 2H), 3.54-3.60 (m, 1H), 4.68-

5.07 (m, 3H), 7.34-7.57 (m, 6H), 7.68-7.79 (m, 2H), 8.15 (d,

J ) 8.5, 1H).

2-Hyd r oxym eth yl-N-m eth yl-N-p h en yl-4-p h en ylqu in o-

lin e-3-ca r boxa m id e (20d ). The title compound was obtained

in 84% yield from lactone 17 and N-methylaniline hydrochlo-

ride. ¹H NMR (CDCl₃): 2.90 (s), 3.27 (s), 4.82-5.40 (m), 6.38-

6.51 (m), 6.87-6.95 (m), 7.22-7.84 (m), 8.11-8.20 (m).

N,N-Dim eth yl-2-h yd r oxym eth ylben za m id e (22a ). The

title compound was obtained in 55% yield from lactone 19 and

dimethylamine hydrochloride. ¹H NMR (CDCl₃): 2.88 (s, 3H),

3.10 (s, 3H), 4.10 (br s, 1H), 4.49 (s, 2H), 7.15-7.55 (m, 4H).

N,N-Dieth yl-2-h ydr oxym eth ylben zam ide (22b). The title

compound was obtained in 62% yield from lactone 19 and

diethylamine hydrochloride. ¹H NMR (CDCl₃): 1.10 (t, J ) 7.1,

3H), 1.28 (t, J ) 7.1, 3H), 3.25 (q, J ) 7.1, 2H), 3.58 (q, J )

7.1, 2H), 4.52 (s, 2H), 7.23-7.47 (m, 4H).

N,N-Dip r op yl-2-h yd r oxym eth ylben za m id e (22c). The

title compound was obtained in 55% yield from lactone 19 and

1

dipropylamine hydrochloride. H NMR (CDCl₃): 0.74 (t, J )

7.4, 3H), 1.0 (t, J ) 7.4, 3H), 1.50 (q, J ) 7.6, 2H), 1.76 (q, J

) 7.4, 2H), 3.16 (t, J ) 7.5, 2H), 3.50 (t, J ) 7.4, 2H), 3.67 (br

s, 1H), 4.50 (br s, 2H), 7.20-7.47 (m, 4H).

2-Hydr oxym eth yl-N-m eth yl-N-ph en ylben zam ide (22d).

The title compound was obtained in 89% yield from lactone

19 and N-methylaniline hydrochloride. ¹H NMR (CDCl₃): 2.96

(s), 3.51 (s), 4.66 (s), 5.32 (s), 6.94-7.90 (m).

Meth od B. The 2-hydroxymethyl carboxamides 20f and 20g

were prepared by addition of a large excess of the suitable

amine (2 mL) to a suspension of lactone 17 (0.26 g, 1.0 mmol)

in ethanol (20 mL). The reaction mixture was refluxed for 3-18

1148 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

h while stirring. After removing the solvent to dryness, the

residue was purified by column chromatography by eluting it

first with methylene chloride-ethyl acetate (8:2) and then

ethyl acetate to obtain pure 20f,g.

2-Hyd r oxym et h yl-4-p h en yl-N-p r op ylq u in olin e-3-ca r -

boxa m id e (20f). The title compound was obtained in 79%

yield from lactone 17 and propylamine. ¹H NMR (CDCl₃): 0.65

(t, J ) 7.4, 3H), 1.00-1.25 (m, 2H), 3.06 (q, J ) 7.2, 2H), 4.89

(m, 3H), 5.41 (br t, 1H), 7.39-7.79 (m, 8H), 8.10 (d, J ) 8.0,

1H).

2-Hydr oxym eth yl-4-ph en yl-N-(ph en ylm eth yl)qu in olin e-

3-ca r boxa m id e (20g). The title compound was obtained in

84% yield from lactone 17 and benzylamine. ¹H NMR

(CDCl₃): 3.80 (br s, 1H), 4.29 (d, J ) 5.5, 2H), 4.96 (s, 2H),

5.68 (br t, 1H), 6.77-6.81 (m, 2H), 7.19-7.26 (m, 3H), 7.37-

7.77 (m, 8H), 8.10 (d, J ) 8.3, 1H).

Gen er a l P r oced u r e for th e P r ep a r a tion of Ch lor om -

eth yl Ca r boxa m id es 23a -g, 24, a n d 25a -d . To a solution

of the suitable hydroxymethyl derivative (20a -g, 21, and

22a -d ) (1.25 mmol) in methylene chloride (20 mL) was added

thionyl chloride (2 mL, 27.6 mmol), and the resulting mixture

was stirred at room temperature for 2-5 h. The solvent was

evaporated in vacuo, and the thionyl chloride excess was

removed by azeotropic distillation with toluene to give the

chloromethyl derivatives in almost quantitative yields as an

oily product which was used without any further purification

(as judged sufficiently pure by TLC analysis) in the next

reaction.

2-Ch lor om eth yl-N-m eth yl-N-p h en yl-ben za m id e (25d ).

¹H NMR (CDCl₃): 3.01 (s), 3.52 (s), 4.85 (s), 5.33 (s), 6.92-

7.95 (m).

Gen er a l P r oced u r e for th e P r ep a r a tion of Ta r get

Ca r boxa m id es 13a -g, 15a -d , a n d 16. To a solution of the

appropriate chloromethyl derivative (23a -g, 24, and 25a -d )

(1.5 mmol) in acetonitrile (25 mL) was added sodium iodide

(1.5 mmol), and the resulting mixture was stirred at room

temperature for 0.5 h. N-Ethylbenzylamine (0.30 mL, 1.5

mmol) and sodium carbonate (0.64 g, 6.0 mmol) were then

added, and the reaction mixture was refluxed for 3 h. After

cooling, the mixture was filtered on Celite and the filtrate was

evaporated in vacuo. The residue was dissolved in methylene

chloride, and the organic phase was thoroughly washed with

H₂O until neutral, dried, and evaporated under reduced

pressure to give an oily residue which gave the analytical

sample after purification by column chromatography with

dichloromethane-ethyl acetate (8:2) as the eluent.

N,N-Dim eth yl-2-[(N-eth yl)ph en ylm eth ylam in o]m eth yl-

4-p h en ylqu in olin e-3-ca r boxa m id e (13a ). The title com-

1

pound was obtained in 88% yield from 23a . H NMR (CDCl₃):

1.02 (t, J ) 7.4, 3H), 2.36-2.46 (m, 1H), 2.50 (s, 3H), 2.58-

2.68 (m, 1H), 2.71 (s, 3H), 3.60 (d, J ) 14.2, 1H), 3.69 (d, J )

12.8, 1H), 3.95 (d, J ) 13.9, 1H), 4.41 (d, J ) 12.8, 1H), 7.14-

7.52 (m, 6H), 7.64-7.74 (m, 2H), 8.15 (d, J ) 8.4, 1H). MS:

m/z 423 (M⁺, 1). Anal. (C₂₈H₂₉N₃O) C, H, N.

N,N-Dieth yl-2-[(N-eth yl)p h en ylm eth yla m in o]m eth yl-

4-p h en ylqu in olin e-3-ca r boxa m id e (13b). The title com-

pound was obtained in 86% yield from 23b. ¹H NMR (CDCl₃):

0.65-0.87 (m, 6H), 1.06 (t, J ) 7.4, 3H), 2.51-3.13 (m, 5H),

3.60-4.05 (m, 4H), 4.20 (d, J ) 13.7, 1H), 7.10-7.78 (m, 13H),

8.18 (d, J ) 8.4, 1H). MS: m/z 451 (M⁺, 1.5). Anal. (C₃₀H₃₃N₃O)

C, H, N.

2-Ch lor om et h yl-N,N-d im et h yl-4-p h en ylq u in olin e-3-

1

ca r boxa m id e (23a ). H NMR (CDCl₃): 2.54 (s, 3H), 2.80 (s,

3H), 4.97 (ABq, J ) 11.9, 2H), 7.38-7.55 (m, 6H), 7.69-7.81

(m, 2H), 8.14 (d, J ) 8.3, 1H).

2-Ch lor om eth yl-N,N-d ieth yl-4-p h en ylqu in olin e-3-ca r -

boxa m id e (23b). ¹H NMR (CDCl₃): 0.76-0.93 (m, 6H), 2.66-

2.84 (m, 1H), 3.02-3.20 (m, 2H), 3.52-3.69 (m, 1H), 4.93 (ABq,

J ) 12.3, 2H), 7.32-7.58 (m, 6H), 7.62-7.87 (m, 2H), 8.16 (d,

J ) 8.4, 1H).

2-Ch lor om eth yl-N,N-dipr opyl-4-ph en ylqu in olin e-3-car -

boxa m id e (23c). ¹H NMR (CDCl₃): 0.65 (t, J ) 7.4, 3H), 0.79

(t, J ) 7.4, 3H), 1.07-1.46 (m, 4H), 2.54-2.69 (m, 1H), 2.88-

3.09 (m, 2H), 3.38-3.52 (m, 1H), 4.90 (ABq, J ) 11.2, 2H),

7.35-7.68 (m, 6H), 7.72-7.80 (m, 2H), 8.18 (d, J ) 8.3, 1H).

2-Ch lor om eth yl-N-m eth yl-N-ph en yl-4-ph en ylqu in olin e-

3-ca r boxa m id e (23d ). ¹H NMR (CDCl₃): 2.90 (s), 3.25 (s),

4.81-4.99 (m), 5.30-5.40 (m), 6.38-6.50 (m), 6.86-6.97 (m),

7.20-7.83 (m), 8.10-8.21 (m).

2-Ch lor om et h yl-N,N-d im et h yl-4-p h en ylq u in olin e-3-

ca r boxa m id e (23e). ¹H NMR (CDCl₃): 2.88 (s), 3.24 (s), 4.80-

4.98 (m), 5.28-5.48 (m), 6.39-6.50 (m), 6.78-6.92 (m), 7.27-

7.34 (m), 7.39-7.62 (m), 7.69-7.85 (m), 8.11-8.21 (m).

2-Ch lor om eth yl-4-p h en yl-N-p r op ylqu in olin e-3-ca r box-

a m id e (23f). ¹H NMR (CDCl₃): 0.68 (t, J ) 7.4, 3H), 1.08-

1.26 (m, 2H), 3.10 (q, J ) 7.2, 2H), 5.03 (s, 2H), 5.46 (br t,

1H), 7.39-7.80 (m, 8H), 8.16 (d, J ) 8.5, 1H).

2-Ch lor om eth yl-4-p h en yl-N-(p h en ylm eth yl)qu in olin e-

3-ca r boxa m id e (23g). ¹H NMR (CDCl₃): 4.35 (d, J ) 5.6, 2H),

5.05 (s, 2H), 5.64 (br t, 1H), 6.83 (m, 1H), 7.21 (m, 2H), 7.35-

7.80 (m, 10H), 8.14 (d, J ) 8.4, 1H). MS: m/z 386 (M⁺, 68).

N,N-Dipr op yl-2-[(N-eth yl)ph en ylm eth yla m in o]m eth yl-

4-p h en ylqu in olin e-3-ca r boxa m id e (13c). The title com-

1

pound was obtained in 74% yield from 23c. H NMR (CDCl₃):

0.55-0.78 (m, 6H), 0.88-1.33 (m, 7H), 2.54-2.95 (m, 5H),

3.50-4.21 (m, 5H), 7.19-7.74 (m, 13H), 8.19 (d, J ) 8.4, 1H).

MS: m/z 479 (M⁺, 1.2). Anal. (C₃₂H₃₇N₃O) C, H, N.

2-[(N-Eth yl)p h en ylm eth yla m in o]m eth yl-N-m eth yl-N-

p h en yl-4-p h en ylqu in olin e-3-ca r boxa m id e (13d ). The title

compound was obtained in 85% yield from 23d . ¹H NMR

(CDCl₃): 1.15 (m), 2.56-2.87 (m), 2.91 (s), 3.17 (s), 3.71-3.84

(m), 3.99-4.26 (m), 4.50-4.69 (m), 6.35 (d, J ) 7.4), 6.68-

6.72 (m), 6.84-6.93 (m), 7.18-7.79 (m), 8.11-8.22 (m). MS:

m/z 485 (M⁺, 1.3). Anal. (C₃₃H₃₁N₃O) C, H, N.

2-[(N-Eth yl)p h en ylm eth yla m in o]m eth yl-N-m eth yl-N-

(4-ch lor oph en yl)-4-ph en ylqu in olin e-3-car boxam ide (13e).

The title compound was obtained in 69% yield from 23e. ¹H

NMR (CDCl₃): 1.15 (m), 2.53-2.90 (m), 2.92 (s), 3.11 (s), 3.67-

3.82 (m), 3.96-4.24 (m), 4.40-4.60 (m), 6.38 (d, J ) 7.4), 6.62-

6.94 (m), 7.26-7.94 (m), 8.11-8.22 (m). MS: m/z 519 (M⁺, 1).

Anal. (C₃₃H₃₀ClN₃O) C, H, N.

2-[(N-E t h yl)p h en ylm et h yla m in o]m et h yl-N-p r op yl-4-

p h en ylqu in olin e-3-ca r boxa m id e (13f). The title compound

1

was obtained in 51% yield from 23f. H NMR (CDCl₃): 0.7 (t,

J ) 7.4, 3H), 1.03-1.24 (m, 5H), 2.67 (q, J ) 6.9, 2H), 3.05-

3.15 (m, 2H), 3.71 (s, 2H), 4.15 (s, 2H), 6.39 (br t, 1H), 7.10-

7.76 (m, 13H), 8.14 (d, J ) 8.4, 1H). MS: m/z 437 (M⁺, 2.5).

Anal. (C₂₉H₃₁N₃O) C, H, N.

2-[(N-E t h yl)p h en ylm et h yla m in o]m et h yl-N-p h en ylm -

eth yl-4-p h en ylqu in olin e-3-ca r boxa m id e (13g). The title

compound was obtained in 34% yield from 23g. ¹H NMR

(CDCl₃): 1.09 (t, J ) 7.2, 3H), 2.64 (q, J ) 7.0, 2H), 3.69 (s,

2H), 4.15 (s, 2H), 4.31 (d, J ) 4.6, 2H), 6.87 (br t, 1H), 6.89-

6.91 (m, 1H), 7.04 (s, 5H), 7.20-7.38 (m, 6H), 7.42-7.55 (m,

5H), 7.66-7.33 (m, 1H), 8.12 (d, J ) 8.5, 1H). MS: m/z 485

(M⁺, 4). Anal. (C₃₃H₃₁N₃O) C, H, N.

N,N-Dim eth yl-2-[(N-eth yl)p h en ylm eth yla m in o]m eth -

ylben za m id e (15a ). The title compound was obtained in 52%

yield from 25a . ¹H NMR (CDCl₃) 1.02 (t, J ) 6.9, 3H), 2.45 (q,

J ) 7.0, 2H), 2.78 (s, 3H), 3.11 (s, 3H), 3.56 (s, 4H), 7.11-7.36

(m, 8H), 7.60 (d, J ) 7.7, 1H); MS m/z 296 (M⁺, 1). Anal.

(C₁₉H₂₄N₂O). C, H, N.

3-Ch lor om eth yl-N,N-d ieth yl-4-p h en ylqu in olin e-2-ca r -

boxa m id e (24). ¹H NMR (CDCl₃): 1.27-1.38 (m, 6H), 3.28

(q, J ) 7.2, 2H), 3.68 (q, J ) 7.1, 2H), 4.70 (s, 2H), 7.35-7.76

(m, 8H), 8.13 (d, J ) 8.5, 1H). Anal. (C₂₁H₂₁ClN₂O) C, H, N.

2-Ch lor om eth yl-N,N-dim eth ylben zam ide (25a). ¹H NMR

(CDCl₃): 2.89 (s, 3H), 3.12 (s, 3H), 4.63 (br s, 2H), 7.17-7.58

(m, 4H).

2-Ch lor om eth yl-N,N-d ieth ylben za m id e (25b). ¹H NMR

(CDCl₃): 1.10 (t, J ) 7.1, 3H), 1.28 (t, J ) 7.1, 3H), 3.18 (q, J

) 7.1, 2H), 3.26-3.57 (br s, 2H), 4.65 (br s, 2H), 7.20-7.48

(m, 4H).

2-Ch lor om eth yl-N,N-dipr opylben zam ide (25c). ¹H NMR

(CDCl₃): 0.76 (t, J ) 7.4, 3H), 0.98 (t, J ) 7.3, 3H), 1.41-1.82

(m, 4H), 3.05 (t, J ) 7.6, 2H), 3.39-3.57 (br s, 2H), 4.64 (br s,

2H), 7.19-7.49 (m, 4H).

Mapping and Fitting the PBR Binding Site

J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8 1149

N,N-Dieth yl-2-[(N-eth yl)p h en ylm eth yla m in o]m eth yl-

ben za m id e (15b). The title compound was obtained in 48%

yield from 25b. H NMR (CDCl₃): 0.90-1.11 (m, 6H), 1.25 (t,

J ) 7.1, 3H), 2.47 (q, J ) 7.0, 2H), 3.07 (q, J ) 7.0, 2H), 3.24-

3.90 (br s, 6H), 7.11-7.36 (m, 8H), 7.68 (d, J ) 7.7, 1H). MS:

m/z 324 (M⁺, 1). Anal. (C₂₁H₂₈N₂O) C, H, N.

for another 15 min. The methylene chloride layer was sepa-

rated, dried, and filtered. Another portion of amine (10 mmol)

was added to the solution, and this mixture was heated to

reflux for 20 min, diluted with methylene chloride, and washed

with water. The organic phase was dried and evaporated to

afford a residue which after purification by column chroma-

tography with methylene chloride-ethyl acetate (1:1) gave an

oily product which crystallized on standing.

N,N-Dieth yl-2-m eth yl-4-p h en ylqu in olin e-3-ca r boxa m -

id e (14a ). Yield 85%; mp 97-99 °C. ¹H NMR (CDCl₃): 0.73

(t, J ) 7.1, 3H), 0.87 (t, J ) 7.2, 3H), 2.63-2.81 (m, 4H), 2.89-

3.17 (m, 2H), 3.61-3.79 (m, 1H), 7.30-7.74 (m, 8H), 8.13 (d,

J ) 8.2, 1H). MS: m/z 318 (M⁺, 22). Anal. (C₂₁H₂₂N₂O) C, H,

N.

N,N-Dip r op yl-2-m eth yl-4-p h en ylqu in olin e-3-ca r boxa -

m id e (14b). Yield 84%, mp 110-111 °C. ¹H NMR (CDCl₃):

0.66 (m, 6H), 1.08-1.37 (m, 4H), 2.50-2.64 (m, 1H), 2.73-

3.02 (m, 5H), 3.51-3.65 (m, 1H), 7.29-7.73 (m, 8H), 8.06 (d,

J ) 8.3, 1H). MS: m/z 346 (M⁺, 20). Anal. (C₂₃H₂₆N₂O) C, H,

N.

N-Meth yl-N-p h en yl-2-m eth yl-4-p h en ylqu in olin e-3-ca r -

boxa m id e (14c). Yield 83%; mp 129-131 °C. ¹H NMR

(CDCl₃): 2.88 (s), 2.95 (s), 3.23 (s), 6.38 (m), 6.52 (m), 6.85

(m), 6.96 (m), 7.23-7.75 (m), 7.98-8.18 (m). MS m/z 352 (M⁺,

32). Anal. (C₂₄H₂₀N₂O) C, H, N.

In Vitr o Bin d in g Assa ys. Male CRL:CD(SD)BR (Charles

River Italia, Calco, CO, Italy) rats were killed by decapitation,

and their brains were rapidly dissected into the various areas

and stored at -80 °C until the day of assay. The binding assays

were performed as described by Cantoni et al.³⁵The frozen

cortices were homogenized in 50 vols of ice-cold phosphate-

buffered saline (PBS) (50 mM, pH 7.4) containing 100 mM

NaCl, using an Ultra Turrax TP-1810 (2 × 20 s). The

homogenate was centrifuged for 10 min at 1000g at 4 °C

(Beckman model J -21B refrigerated centrifuge). The pellet was

washed three more times by resuspension in the same volume

of fresh buffer and centrifuged again at 50000g for 10 min.

The pellet obtained was finally resuspended in the incubation

buffer (PBS) just before the binding assay.

[³H]PK11195 (s.a. 86.4 Ci/mmol, NEN) binding was assayed

in a final incubation volume of 1.0 mL, consisting of 0.5 mL of

membrane suspension, 0.5 mL of [³H]-ligand, and 20 µL of

displacing agent or solvent. Tissue concentration and [³H]-

PK11195 final concentration were 2.3 mg tissue/sample and

1 nM, respectively.

1

N,N-Dip r op yl-2-[(N-et h yl)p h en ylm et h yla m in o]m et h -

ylben za m id e (15c). The title compound was obtained in 40%

1

yield from 25c. H NMR (CDCl₃): 0.66 (t, J ) 7.1, 3H), 0.94-

1.07 (m, 6H), 1.39-1.61 (m, 2H), 1.68-1.82 (m, 2H), 2.48 (q,

J ) 7.0, 2H), 2.96 (t, J ) 7.4, 2H), 3.36-3.56 (m, 6H), 7.10-

7.36 (m, 8H), 7.72 (d, J ) 7.7, 1H). MS: m/z 352 (M⁺, 2). Anal.

(C₂₃H₃₂N₂O) C, H, N.

2-[(N-Eth yl)p h en ylm eth yla m in o]m eth yl-N-m eth yl-N-

p h en ylben za m id e (15d ). The title compound was obtained

in 58% yield from 25d . ¹H NMR (CDCl₃): 1.06 (t, J ) 7.0, 3H),

2.48 (q, J ) 6.9, 2H), 3.46-3.75 (m, 7H), 7.07-7.58 (m, 14H).

Anal. (C₂₄H₂₆N₂O) C, H, N.

N,N-Dieth yl-3-[(N-eth yl)p h en ylm eth yla m in o]m eth yl-

4-p h en ylqu in olin e-2-ca r boxa m id e (16). The title compound

1

was obtained in 78% yield from 24. H NMR (CDCl₃): 0.75 (t,

J ) 7.3, 3H), 1.31-1.40 (m, 6H), 2.25 (q, J ) 6.2, 2H), 3.27 (q,

J ) 6.9, 2H), 3.41 (s, 2H), 3.68 (q, J ) 6.9, 2H), 3.80 (s, 2H),

7.05-7.50 (m, 12H), 7.60-7.68 (m, 1H), 8.09 (d, J ) 8.3, 1H).

MS (FAB): m/z 452 (M + 1, 100). Anal. (C₃₀H₃₃N₃O‚0.25H₂O)

C, H, N.

2-Br om om eth yl-4-p h en ylqu in olin e (28). A mixture of

2-methyl-4-phenylquinoline^16a(1.0 g, 4.57 mmol) in carbon

tetrachloride (40 mL) with N-bromosuccinimide (0.81 g, 4.57

mmol) and dibenzoyl peroxide (0.2 g, 0.8 mmol) was refluxed

for 72 h. The solvent was then evaporated in vacuo, and the

residue was diluted with small portions of the same solvent

and filtered. The filtrate was concentrated under reduced

pressure, and the residue was purified by column chromatog-

raphy eluting with n-hexane-ethyl acetate (9:1) to give

compound 28 (yield 51%) which was used as such in the next

1

step. H NMR (CDCl₃) 4.72 (s, 2H), 7.42-7.62 (m, 7H), 7.65-

7.80 (m, 1H), 7.90 (d, J ) 8.1, 1H), 8.12 (d, J ) 8.1, 1H).

2-[(N-Eth yl)ph en ylm eth ylam in o]m eth yl-4-ph en ylqu in -

olin e (13h ). A mixture of 2-bromomethyl derivative 28 (0.3

g, 1.0 mmol) and N-ethylbenzylamine (0.89 mL, 6.0 mmol) in

ethanol (20 mL) was heated at reflux until the starting

material disappeared (2.5 h). The solvent was removed in

vacuo and the residue dissolved in diethyl ether. The organic

phase was washed to neutrality, dried, and evaporated to

dryness. Purification of the residue by flash chromatography

with n-hexane-ethyl acetate (8:2) as the eluent gave an

analytical sample of 13h as a light yellow oil (0.34 g, yield

96%). ¹H NMR (CDCl₃): 1.10 (t, J ) 7.0, 3H), 2.62 (q, J ) 7.0,

2H), 3.69 (s, 2H), 3.95 (s, 2H), 7.19-7.51 (m, 11H), 7.64-7.71

(m, 2H), 7.86 (d, J ) 8.0, 1H), 8.10 (d, J ) 8.3, 1H). Anal.

(C₂₅H₂₄N₂) C, H, N.

Incubations (120 min at 4 °C) were stopped by rapid

filtration under vacuum through GF/B fiber filters which were

then washed with 12 mL (3 × 4 mL) of ice-cold PBS by means

of a Brandel M-48RP cell harvester.

Nonspecific binding was assayed in the presence of PK11195

(1 µM).

Dried filters were immersed in vials containing 4 mL of

Filter Count (Packard) liquid scintillation cocktail for the

measurement of trapped radioactivity with a Packard LKB

1214 RACKBETA liquid scintillation spectrometer at a count-

ing efficiency of about 60%. The concentration of the test

P r ep a r a tion of 2-Meth yl-4-p h en ylqu in olin e-3-ca r box-

ylic Acid (27). A mixture of ester 26¹⁵(5.9 g, 2.0 mmol),

potassium hydroxyde (3.0 g, 53 mmol), ethanol (40 mL), and

water (4 mL) was refluxed for 24 h. After cooling and

acidification with glacial acetic acid, the reaction mixture was

concentrated under reduced pressure and the product was

precipitated by addition of water. It was collected, washed with

water until neutral, dried, and recrystallized from a mixture

of methanol/2-propanol to yield acid 27 (3.5 g, yield 73%) as

colorless crystals melting at 268 °C (lit.¹⁶265-267 °C). ¹H

NMR (DMSO-d₆): 2.67 (s, 3H), 7.32-7.51 (m, 7H), 7.61-7.68

(m, 1H), 8.00 (d, J ) 8.4, 1H), 13.29 (br s, 1H).

Gen er a l P r oced u r e for th e P r ep a r a tion of 2-Meth yl-

4-P h en ylqu in olin e-3-Ca r boxa m id es (14a -c). Phosphorus

pentachloride (0.45 mL, 3,4 mmol) was added to a suspension

of the acid 27 (0.91 g, 3.4 mmol) in methylene chloride (20

mL) cooled to -20 °C. After being stirred for 30 min at this

temperature, this mixture was added to a stirred solution of

the suitable amine (10 mmol) in methylene chloride (25 mL)

layered with 10% aqueous sodium carbonate (10 mL) and ice

(5 g). The two-phase mixture was stirred for 30 min and

followed by the addition of sodium carbonate solution (10 mL)

compounds that inhibited [³H]ligand binding by 50% (IC₅₀) was

36

determined by means of the Allfit

program with six

concentrations of the displacers, each performed in triplicate.

Ack n ow led gm en t. The authors are grateful to Prof.

Stefania D’Agata D’Ottavi for the careful reading of the

manuscript. Dr Gianluca Giorgi (C.I.A.D.S., Siena) is

also acknowledged for the recording of mass spectra.

A.M. is grateful to Prof. C. G. Wermuth for the scientifi-

cally fruitful months spent at CNRS (LPM) in Stras-

bourg. M.C.M. and M.S. are especially grateful to

CICAIA (Centro Interdipartimentale Calcolo Automat-

ico e Informatica Applicata) for the Cerius²program.

This work was financially supported by Italian MURST

(ex 60% and ex 40% funds) and was partially funded

by a contract with NFCR.

1150 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 8

Anzini et al.

(18) Molecular Simulations Inc., 16 New England Executive Park,

Burlington, MA, 01803-5297.

(19) Wyvill, G.; McPheeters, C.; Wyvill, B. Data Structures for Soft

Objects. Visual Comput. 1986, 2, 227-234.

(20) Sali, A., Blundell, T. L. Comparative Protein Modeling by

Satisfaction of Spatial Restraints. J . Mol. Biol. 1993, 234, 779-

815.

(21) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;

Swaminathan, S.; Karplus, M. CHARMM: A Program for

Macromolecular Energy, Minimization and Dynamics Calcula-

tions. J . Comput. Chem. 1983, 4, 187-217.

(22) Bowie, J . U.; Luthy, R.; Eisenberg, D. A Method to Identify

Protein Sequences that Fold into a Known Three-dimensional

Structure. Science 1991, 253, 164-170.

(23) Cramer R. D.; Pattersn, D. E.; Bunce, J . D. Comparative

Molecular Field Analysis (CoMFA). 1. Effect of Shape on Binding

of Steroids to Carrier Proteins. J . Am. Chem. Soc. 1988, 110,

5959-5967.

(24) Hagen, B. 3D-QSAR Studies of Ligands of Peripheral Benzodi-

azepine Receptors. Diploma Thesis, University of Innsbruck,

1999.

Refer en ces

(1) (a) Verma, A.; Snyder, S. H. Peripheral Type Benzodiazepine

Receptors. Annu. Rev. Pharmacol. Toxicol. 1989, 29, 307-322.

(b) Saano, V.; Ra¨go, L.; Ra¨ty, M. Peripheral Benzodiazepine

Binding Sites. Pharmacol. Ther. 1989, 41, 503-514. (c) Gavish,

M.; Katz, Y.; Bar-Ami, S.; Weizman, R. Biochemical, Physiologi-

cal, and Pathological Aspects of the Peripheral Benzodiazepine

Receptor. J . Neurochem. 1992, 58, 1589-1601.

(2) Peripheral Benzodiazepine Receptors; Giesen-Crouse, E., Ed.;

Academic Press: London, 1993.

(3) Kozikowski, A. P.; Ma, D.; Brewer, J .; Sun, S.; Costa, E.; Romeo,

E.; Guidotti A. Chemistry, Binding Affinities, and Behavioral

Properties of a New Class of “Antineophobic” Mitochondrial DBI

Receptor Complex (mDRC) Ligands. J . Med. Chem. 1993, 36,

2908-2920 and references therein.

(4) Sprengel, R.; Werner, P.; Seeburg, P. H.; Mukhin, A. G.; Santi,

M. R.; Grayson, D. R.; Guidotti, A.; Krueger, K. E. Molecular

Cloning and Expression of cDNA Encoding a Peripheral-Type

Benzodiazepine Receptor. J . Biol. Chem. 1989, 264, 20415-

20421.

(5) J oseph-Liauzun, E.; Delmas, P.; Shire, D.; Ferrara, P. Topological

Analysis of the Peripheral Benzodiazepine Receptor in Yeast

Mitochondrial Membrane Supports a Five-Transmembrane Struc-

ture. J . Biol. Chem. 1998, 273, 2146-2152.

(6) Bernassau, J . M.; Reversat, J . L.; Ferrara, P.; Caput, D.; Le Fur,

G. A 3D Model of the Peripheral Benzodiazepine Receptor and

its Implication in Intramitochondrial Cholesterol Transport. J .

Mol. Graphics 1993, 11, 236-244.

(25) Tripos Associates Inc., St. Louis, MO.

(26) Wold, S.; Ruhe, A.; Wold, H.; Dunn, W. J . The Covariance

Problem in Linear Regression. The Partial Least Squares (PLS)

Approach to Generalized Inverses. SIAM J . Sci. Stat. Comp.

1984, 5, 735-743.

(27) Klebe, G. Comparative Molecular Similarity Indices: CoMSI.

In 3D QSAR in Drug Design; Kubinyi, H., Folkers, G., Martin,

Y., Eds.; Kluver Academic Publisher: Great Britain, 1998, Vol.

3, pp 87-104.

(28) (a) Menziani, M. C.; De Benedetti, P. G.; Karelson, M. Theoretical

Descriptors in Quantitative Structure-Affinity and Selectivity

Relationship Study of Potent N4-substituted Arylpiperazine

5-HT_1AReceptor Antagonists. Bioorg. Med. Chem. 1998, 6, 535-

550. (b) Menziani, M. C.; Montorsi, M.; De Benedetti, P. G.;

Karelson, M. Relevance of Theoretical Molecular Descriptors in

Quantitative Structure-Activity Relationship Analysis of R₁-

Adrenergic Receptor Antagonists. Bioorg. Med. Chem. 1999, 7,

2437-2451.

(29) Hahn, M. Receptor Surface Models. 1. Definition and Construc-

tion. J . Med. Chem. 1995, 38, 2080-2090.

(30) Pan, C. P.; Hirashima, A.; Kuwano, E.; Eto, M. Three-

dimensional Pharmacophore Hypoteses for the Locust Neuronal

Octopamine Receptor (OAR). 1. Antagonists. J . Mol. Model.

1997, 3, 455-463.

(31) (a) Yeliseev, A.; Krueger, K. E.; Kaplan, S. A Mammalian

Mitochondrial Drug Receptor Functions as a Bacterial “Oxygen”

Sensor. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 5101-5106. (b)

Yeliseev, A. A.; Kaplan, S. TspO of Rhodobacter sphaeroides. J .

Biol. Chem. 2000, 275, 5657-5667.

(32) Breiter, D. R.; Kanost, M. R.; Benning, M. M.; Wesenberg, G.;

Law, J . H.; Wells, M. A.; Rayment, I.; Holden, H. M. Molecular

Structure of an Apolipoprotein Determined at 2.5 Å Resolution.

Biochemistry 1991, 30, 603-608.

(33) Farges, R.; J oseph-Liauzun, E.; Shire, D.; Caput, D.; Le Fur,

G.; Loison, G.; Ferrara, P. Molecular basis for the Different

Binding Properties of Benzodiazepines to Human and Bovine

Peripheral-type Benzodiazepine Receptors. FEBS Lett. 1993,

335, 305-308.

(34) (a) Trapani, G.; Franco, M.; Ricciardi, L.; Latrofa, A.; Genchi,

G.; Sanna, E.; Tuveri, F.; Cagetti, E.; Biggio, G.; Liso, G.

Synthesis and Binding Affinity of 2-Phenylimidazo[1,2-a]pyri-

dine Derivatives for both Central and Peripheral Benzodiazepine

Receptors. A New Series of High-Affinity and Selective Ligands

for the Peripheral Type. J . Med. Chem. 1997, 40, 3109-3118.

(b) Trapani, G.; Franco, M.; Latrofa, A.; Ricciardi, L.; Carotti,

A.; Serra, M.; Sanna, E.; Biggio, G.; Liso, G. Novel 2-Phenylimi-

dazo[1,2-a]pyridine Derivatives as Potent and Selective Ligands

for Peripheral Benzodiazepine Receptors: Synthesis, Binding

Affinity, and in Vivo Studies. J . Med. Chem. 1999, 42, 3934-

3941.

(35) Cantoni, L.; Rizzardini, M.; Skorupska, M.; Cagnotto, A.; Cat-

egoni, A.; Pecora, N.; Frigo, L.; Ferrarese, C.; Mennini, T. Hepatic

Protoporphyria is Associated with a Decrease in Ligand Binding

for Mitochondrial Benzodiazepine Receptors in the Liver. Bio-

chem. Pharmacol. 1992, 44, 1159-1164.

(36) Munson, P. G.; Roadbard, D. Computers in Endocrinology; Raven

Press: New York, 1984; pp 117-145.

(7) Farges, R.; J oseph-Liauzun, E.; Shire, D.; Caput, D.; Le Fur,

G.; Ferrara, P. Site-Directed Mutagenesis of the Peripheral

Benzodiazepine Receptor: Identification of Amino Acids Impli-

cated in the Binding Site of Ro5-4864. Mol. Pharmacol. 1994,

46, 1160-1167.

(8) Bourguignon, J . J . Endogenous and Synthetic Ligands of Mito-

chondrial Benzodiazepine Receptors: Structure-Affinity Rela-

tionships. In Peripheral Benzodiazepine Receptors; Giesen-

Crouse, E., Ed.; Academic Press: London, 1993; pp 59-85.

(9) Georges, G.; Vercauteren, D. P.; Vanderveken, D. J .; Horion, R.;

Evrard, G.; Fripiat, J . G.; Andre, J . M.; Durant, F. Structural

and Electronic Analysis of Peripheral Benzodiazepine Ligands:

Description of the Pharmacophoric Elements for Their Receptor.

Int. J . Quantum Chem., Quantum Biol. Symp. 1990, 17, 1-25.

(10) Anzini, M.; Cappelli, A.; Vomero, S.; Cagnotto, A.; Skorupska,

M. Synthesis and Benzodiazepine Receptor Affinity of 2,3-

Dihydro-9-phenyl-1H-pyrrolo[3,4-b]quinolin-1-one and 3-Carbe-

thoxy-4-phenylquinoline Derivatives. Farmaco 1992, 47, 191-

202.

(11) Cappelli, A.; Anzini, M.; Vomero, S.; De Benedetti, P. G.;

Menziani, M. C.; Giorgi, G.; Manzoni, C. Mapping the Peripheral

Benzodiazepine Receptor Binding Site by Conformationally

Restrained Derivatives of 1-(2-Chlorophenyl)-N-methyl-N-(1-

methylpropyl)-3-isoquinolinecarboxamide (PK11195). J . Med.

Chem. 1997, 40, 2910-2921.

(12) Matarrese, M.; Soloviev, D.; Cappelli, A.; Todde, S.; Moresco, R.

M.; Anzini, M.; Vomero, S.; Sudati, F.; Carpinelli, A.; Perugini,

F.; Galli Kienle, M.; Fazio, F. Synthesis of the Novel [¹¹C]-

Labelled Quinoline Carboxamides: Analogues of PK-11195 as

Putative Radioligands for PET Studies of Peripheral Type

Benzodiazepine Receptors. J . Labelled Compd. Radiopharm.

1999, 42, S397-S399.

(13) Anzini, M.; Cappelli, A.; Vomero, S. Synthesis of 2-Substituted-

2,3-dihydro-9-phenyl-1H-pyrrolo[3,4-b]quinolin-3-ones as Poten-

tial Peripheral Benzodiazepine-receptor Ligands. Heterocycles

1994, 38, 103-111 and references therein.

(14) (a) Basha, A.; Lipton, M.; Weinreb, S. M. A Mild, General Method

for Conversion of Ester to Amides. Tetrahedron Lett. 1977, 48,

4171-4174. (b) Levin, J . I.; Turos, E.; Weinreb, S. M. An

Alternative Procedure for the Aluminum-Mediated Conversion

of Esters to Amides. Synth. Commun. 1982, 12, 989-993.

(15) Fehnel, E. A. Friedlander Syntheses with o-Aminoaryl Ketones.

III. Acid-Catalyzed Condensations of o-Aminobenzophenone with

Polyfunctional Carbonyl Compounds (1,2). J . Heterocycl. Chem.

1967, 4, 565-570.

(16) (a) Fehnel, E. A. Friedlander Syntheses with o-Aminoaryl

Ketones. I. Acid-Catalyzed Condensations of o-Aminobenzophe-

none with Ketones. J . Org. Chem. 1966, 31, 2899-2902. (b)

Fehnel, E. A.; Cohn, D. E. Friedlander Syntheses with o-

Aminoaryl Ketones. II. Structure of the Product Formed in the

Condensation of o-Aminobenzophenone with Acetylacetone. J .

Org. Chem. 1966, 31, 3852-3854.

(17) Dewar, M. J . S.; Zoebisch, E. G.; Healey, E. F.; Stewart, J . J . P.

AM1: A New General Purpose Quantum Mechanical Molecular

Model. J . Am. Chem. Soc. 1985, 107, 3902-3909.

J M0009742

Article Doi

Mapping and fitting the peripheral benzodiazepine receptor binding site by carboxamide derivatives. Comparison of different approaches to quantitative ligand-receptor interaction modeling

DOI: 10.1021/jm0009742

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jm0009742

Full text of DOI:10.1021/jm0009742

Products guided by the article

R&D Labs maybe for 342011-73-8

Relevant to this article

Hot Product