Novel N-(Arylalkyl)indol-3-ylglyoxylylamides T argeted as L igands of the
Benzodiazepine R eceptor: Synthesis, Biological E valuation, and Molecular
Modeling Analysis of the Structure-Activity R elationships†
Giampaolo Primofiore,*,§Federico Da Settimo,§Sabrina Taliani,§Anna Maria Marini,§Ettore Novellino,£
Giovanni Greco,£Antonio Lavecchia,£Franc ¸ois Besnard,‡Letizia Trincavelli,#Barbara Costa,#and
Dipartimento di Scienze Farmaceutiche and Dipartimento di Psichiatria, Neurobiologia, Farmacologia e Biotecnologie,
Universita ` di Pisa, Via Bonanno 6, 56126 Pisa, Italy, Dipartimento di Chimica Farmaceutica e Tossicologica, Universita ` di
Napoli “Federico II”, Via D. Montesano, 49, 80131 Napoli, Italy, and Department of Molecular and Functional Genomics,
Synthe ´labo, 10 rue des Carrie `res, 92500 Rueil-Malmaison, France
Received J anuary 23, 2001
A series of N-(arylalkyl)indol-3-ylglyoxylylamides (4-8) was synthesized as ligands of the
benzodiazepinereceptor (BzR) and tested for their ability todisplace[3H]flumazenil from bovine
brain membranes. The new compounds, bearing a branched (4) or a geometrically constrained
benzyl/phenylethyl amide side chain (5-8), represent the continuation of our research on
N-benzylindol-3-ylglyoxylylamides 1 (Da Settimo et al., 1996), N′-phenylindol-3-ylglyoxylo-
hydrazides 2 (Da Settimo et al., 1998), and N-(indol-3-ylglyoxylyl)alanine derivatives 3
(Primofiore et al., 1989). A few indoles belonging to the previously investigated benzylamides
1 and phenylhydrazides 2 were synthesized and tested toenrich the SARs in these twoseries.
The affinities and the GABA ratios of selected compounds for clonal mammalian R1?2γ2, R3?2γ2,
and R5?3γ2BzR subtypes were alsodetermined. It was hypothesized that the reduced flexibility
of indoles 4-8 would both facilitate the mapping of the BzR binding cleft and increase the
chances of conferring selectivity for the considered receptor subtypes. In the series of indoles
4, the introduction of a methyl group on the benzylic carbon with the R configuration improved
affinity of the 5-substituted (5-Cl and 5-NO2) derivatives, whereas it was detrimental for their
5-unsubtituted (5-H) counterparts. All S enantiomers were less potent than the R ones.
Replacement of themethyl with hydrophilic substituents on thebenzylic carbon lowered affinity.
The isoindolinylamide side chain was tolerated if the 5-position was unsubstituted (Kiof 5a )
123 nM), otherwise affinity was abolished (5b, c). All the 2-indanylamides 6 and (S)-1-
indanylamides 8 were devoid of any appreciable affinity. The 5-Cl and 5-NO2 (R)-1-
indanylamides 7b (Ki80 nM) and 7c (Ki28 nM) were the most potent among the indoles 5-8
geometrically constrained about the side chain. The 5-H (R)-1-indanylamide 7a displayed a
lower affinity (Ki675 nM). The SARs developed from the new compounds, together with those
collected from our previous studies, confirmed the hypothesis of different binding modes for
5-substituted and 5-unsubstituted indoles, suggesting that the shape of the lipophilic pocket
L1(notation in accordance with Cook’s BzR topological model) is asymmetric and highlighted
the stereoelectronic and conformational properties of the amide side chain required for high
potency. Several of the new indoles showed selectivity for the R1?2γ2subtype compared with
the R3?2γ2and R5?3γ2subtypes (e.g.: 4t and 7c bind tothese three BzR isoforms with Kivalues
of 14 nM, 283 nM, 239 nM, and 9 nM, 1960 nM, 95 nM, respectively). The GABA ratios close
to unity exhibited by all the tested compounds on each BzR subtype were predictive of an
efficacy profile typical of antagonists.
The γ-aminobutyric acid type A (GABAA) receptor is
the major inhibitory ligand-gated ion channel in the
mammalian brain.1,2This membrane-bound hetero-
pentameric receptor is made up of five subunits out
of the 18 which have so far been cloned and sequenced
(6R, 4?, 4γ, 1δ, 1?, and 2F). Three subunits (R, ?, and γ)
are required toform a fully functional GABAAreceptor.
The so-called benzodiazepine receptor (BzR) is located
between the R and γ subunits, and its occupation by a
ligand can allosterically modulate the affinity of the
GABA neurotransmitter for its specific binding site. BzR
agonists and inverse agonists potentiate or decrease,
respectively, theGABA-induced chlorideinflux, whereas
antagonists have minimal or no effects on the chloride
flux. These substances exhibit a wide variety of phar-
macological actions spanning in a continuum from full
agonists (anxiolytic, anticonvulsant, sedative-hypnotic,
and myorelaxant agents) through antagonists toinverse
* Towhom correspondence should be addressed. Tel: 39 50 500209.
Fax: 39 50 40517. E-mail: firstname.lastname@example.org.
†Presented at the Italian-Hungarian-Polish J oint Meeting on
Medicinal Chemistry, Giardini Naxos-Taormina, September 28-
October 1, 1999 and at the IX Meeting on Heterocyclic Structures in
Medicinal Chemistry Research, Palermo, May 14-17, 2000.
§Dipartimento di Scienze Farmaceutiche, Universita `di Pisa.
£Universita `“Federico II” di Napoli.
‡Department of Molecular and Functional Genomics, Synthe ´labo.
#Dipartimento di Psichiatria, Neurobiologia, Farmacologia e Bio-
tecnologie, Universita `di Pisa.
2286J . Med. Chem. 2001, 44, 2286-2297
10.1021/jm010827j CCC: $20.00© 2001 American Chemical Society
Published on Web 06/12/2001
agonists (anxiogenic and proconvulsant agents). Partial
agonists exist within this efficacy spectrum and are of
particular interest, as they may display antianxiety
properties devoid of the undesirable side effects typical
of full agonist-type ligands.3Partial inverse agonists
have been described which can enhance general memory/
learning and block or reverse the effects of barbiturate
toxicity but are devoid of proconvulsant activity.4,5
Moreover, the availability of cloned benzodiazepine
receptor subtypes will probably lead soon tothe discov-
ery of subtype-selective ligands, which will open the
exciting possibility toseparate the many pharmacologi-
cal actions of BzR ligands permitting the selective
treatment of anxiety, sleep disorders, convulsions, and
memory deficits with fewer side effects.1,6
Structure-activity relationships (SARs) of structur-
ally diverse classes of ligands7-14were rationalized by
Cook and co-workers15through a comprehensive phar-
macophore/receptor model consisting of several BzR
interaction (sub)sites: (i) a hydrogen bond acceptor (A2),
(ii) a hydrogen bond donor (H1), (iii) a bifunctional
hydrogen bond donor/acceptor (H2/A3), and (iv) four
lipophilic pockets (L1, L2, L3, and Ldi). The boundaries
of the receptor were defined in terms of sterically
forbidden sites (S1, S2, and S3). Finally, it was assumed
that agonists, antagonists, and inverse agonists share
the same binding cleft. Figure 1 describes the inter-
actions of our compounds 1 and 2 at the BzR in the
framework of Cook’s pharmacophore model.
We have recently reported on a new class of BzR
ligands designed as open chain analogues of ?-carbo-
lines, the N-(benzyl)indol-3-ylglyoxylylamides 1.16In-
terestingly, in this series the effects of the R5 and
X substituents on potency are not constant but inter-
dependent. Particularly, affinity is favored by electron-
donating or electron-attracting X substituents depend-
ing on whether the 5-position of the indole nucleus is
substituted (R5) Cl/NO2) or not (R5) H). Thus, while
the optimum of affinity in the 5-Cl/NO2 series was
reached with X ) 3′,4′-(OMe)2(Ki 11 nM), in the 5-H
series potency was optimized with X ) 4′-Cl (Ki67 nM).
A few selected benzylamide derivatives were alsoevalu-
ated by in vivo tests, but none of them showed any
activity. Their lack of efficacy was explained in terms
of poor absorption and bioavailability, partly depending
on low water solubility.
Thenanomolar binding constants exhibited by several
benzylamides 1 prompted the design of closely related
but more water-soluble and bioavailable analogues.
Therefore, we prepared a series of N′-phenylindol-3-
ylglyoxylohydrazides 2 formally derived from the previ-
ously described benzylamides by replacing the CH2
spacer with the isosteric NH group.17Surprisingly,
affinity was restricted to 5-H phenylhydrazides, the
5-Cl/NO2 counterparts being invariably inactive for
either electron-donating or electron-attracting X sub-
stituents on the side phenyl ring. This discrepancy in
the SARs of the isosteric series of 5-Cl/NO2indoles was
suspected todepend on differences in theconformational
properties of the NHNHAr and NHCH2Ar side chains:
the former being forced in a gauche disposition about
the N-N bond and the latter, more flexible, capable of
assuming a staggered conformation about the N-C
bond. As observed in the 5-H benzylamides, the affinity
of 5-H phenylhydrazides is enhanced by electron-
withdrawing X substituents such as 4′-NO2(Ki11 nM)
and is lowered considerably by methylation of theindole
nitrogen. Selected phenylhydrazides tested in vivo
revealed efficacy profiles typical of partial agonists.
Taken together, the binding data of compounds 1
and 2 suggested that 5-Cl/NO2 indoles interact with
the receptor differently from their 5-H counterparts.
Particularly, two alternative binding modes of the
ligands were hypothesized, called A and B, exemplified
in Figure 1 using the most potent benzylamide and
phenylhydrazide derivatives, respectively.
Binding mode A requires (i) a transoid conformation
of the side chain (not feasible for phenylhydrazides) and
(ii) non-electron-withdrawing X substituents on theside
phenyl ring. According toour hypothesis, benzylamides
would engage interactions with the A2site (through the
indole NH), the H1and H2sites (through the CdO2 and
CdO1), and the L1, L2, and LDilipophilic regions (filled
by the CH2, the phenyl and the fused benzene ring,
F igure 1. Binding modes A and B exemplified through the most potent ligands among benzylamides 1 and phenylhydrazides 2,
respectively. Labeling of BzR subsites are in accordance with Cook’s pharmacophore model.15
N-(Arylalkyl)indol-3-ylglyoxylylamidesJ ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14 2287
respectively). An electron-withdrawing group in the
5-position of the indole, such as Cl or NO2, strengthens
the NH‚‚‚A2 hydrogen bond. The favorable electronic
effects exerted by electron-donating X substituents in
the series of 5-Cl/NO2 benzylamides suggest that the
side phenyl ring might be involved in a charge-transfer
interaction with an electropositive function within the
Binding mode B takes place through the following
interactions: (i) CdO2 and CdO1 are hydrogen-bound
to the H1 and H2 sites; (ii) the lipophilic L1 and L2
regions are occupied by the pyrrole and benzene moi-
eties of the indole nucleus; and (iii) a hydrogen bond
is donated by the indole NH to a heteroatom belonging
to the S1 site. Electron-withdrawing substituents on
the side phenyl ring, such as 4′-NO2 in the most
potent phenylhydrazide, make the terminal 2′-CH hy-
drogen moreelectropositivethus favoring its interaction
with the electron-rich A2 site. Binding mode B is
accessible only to 5-H indoles because the sterically
forbidden S2 site closely faces the 5-position and is
unable to host substituents larger than a hydrogen.
Alternatively (or additionally), mode B might not
be feasible for 5-Cl and 5-NO2 derivatives owing to
unfavorable electron-attracting effects of these substit-
uents on the indole π-system. Collinearity between size
and electron-attracting power within the limited data
set (Cl and NO2) of 5-substituents did not allow us to
single out which property actually disables mode B.
Benzylamides 1 are closely related analogues of
N-(indol-3-ylglyoxylyl)amino acid derivatives with the
general formula 318,19displaying nanomolar potency
when incorporating a (D)-alanineresidue(R′ ) Me). The
much lower affinities of the corresponding glycine (R′
) H) and (L)-alanine derivatives led us to hypothesize
that the L1pocket surrounding the R-carbon is asym-
metric, sothat it can be filled by the (D)-alanine methyl
on one side but has no room for the (L)-alanine methyl
on the opposite side.
In light of the SARs in series 1 and 3, we felt that a
fruitful continuation of our research would betoprepare
rivatives 4 which retain the benzylamide scaffold 1 and
bear an R-methyl group (R′ ) Me) in the same spatial
position (R configuration) as in the previously described
(D)-alanine derivatives 3. Compounds 4 were synthe-
sized as pure enantiomers whenever the starting prod-
ucts werecommercially available, otherwisetheracemic
mixture was prepared and tested.
The following indole derivatives, all featuring a geo-
metrically constrained N-phenylalkyl side chain, were
included in the same project: 2-(indol-3-ylglyoxylyl)-
isoindolines 5, N-(indan-2-yl)indol-3-ylglyoxylylamides
6, (R) and (S) enantiomers of N-(indan-1-yl)indol-3-
ylglyoxylylamides 7 and 8. We reasoned that the
reduced flexibility of these structures would more mark-
edly discriminate between the two putative binding
modes A and B, facilitate mapping of the BzR binding
cleft, and increase the chances of improving affinity.
Among the newly investigated compounds there are
alsosome benzylamides and phenylhydrazides of series
1 and 2 bearing a methoxy group in position 5 of the
indole nucleus. By adding these 5-MeO derivatives to
the data set of 5-Cl and 5-NO2 indoles, we broke the
collinearity between size and the electron-withdrawing
character of the 5-substituent (R5), so as to unambigu-
ously identify the property of this substituent leading
to binding mode A or B. Finally, benzylamides of type
1 were prepared featuring a nitro group in position 4′
of the side phenyl ring (X ) 4′-NO2) to compare their
binding affinities with those of the 4′-nitro derivatives
of type 4 (R′ ) Me, X ) 4′-NO2).
Since it is currently recognized that subtype-selective
BzR ligands might represent potential selective drugs
for the treatment of anxiety, sleep disorders, convul-
sions, and memory deficit with fewer side effects,1,6a
few selected indole derivatives were evaluated by the
radioligand techniqueon recombinant rat R1?2γ2, R3?2γ2,
and R5?3γ2GABAA/BzR subtypes. The in vitro efficacy
profile of the selected compounds for all three GABAA/
BzR subtypes was assessed by means of theGABA ratio.
This paper describes the synthesis, the biological
evaluation, the SARs, and the molecular modeling
analysis of the novel indole derivatives 1, 2, 4-8
targeted as ligands of the BzR.
2288 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14 Primofiore et al.
The general synthetic procedure used in the prepara-
tion of compounds 1, 2, and 4-8 involved the acylation
of the appropriate indole with oxalyl chloride in ac-
cordance with a published procedure.20The indolylgly-
oxylyl chlorides obtained were allowed to react in mild
conditions with the appropriate amine in the presence
of triethylamine in toluene solution (in THF solution
for 4i and 4bb) (Scheme 1). All products were purified
by recrystallization from the appropriate solvent, and
their structures were confirmed by IR,1H NMR, MS,
and elemental analysis (Table 1). Spectral data of all
the newly synthesized compounds 1, 2, and 4-8 are
reported in the Supporting Information.
R esults and Discussion
The binding affinity of each newly synthesized indole
derivatives at the BzR in bovine brain membranes was
determined by competition experiments against the
radiolabeled antagonist [3H]flumazenil21and expressed
as the Ki value only for those compounds inhibiting
radioligand binding by more than 80% at a fixed
concentration of 10 µM. The in vitro efficacy of active
compounds was measured by the GABA ratio which
predicts thepharmacological profileof a BzR ligand.22-24
Table 2 summarizes the biological data. The affinities
of some previously reported16,17,19indoles (compounds
1a′-l′, 2a′-g′, and 3a′-c′) discussed in the present
paper are listed in Table 3. Molecular modeling studies,
performed toaid the interpretation of SARs, were based
on semiempirical quantum-mechanics and molecular
mechanics calculations using theAM1 method25and the
Tripos force field26available within the SYBYL suite of
programs27(computational details are given in the
Property of the R5Substituent Disabling Bind-
ing Mode B. As stated, binding mode B might not be
feasible for 5-Cl/NO2 indoles, due to a steric clash
between a 5-substituent (R5) larger than a hydrogen and
the S2site (see Figure 1). Alternatively or additionally,
mode B might be forbidden for 5-substituted indoles
owing to the unfavorable electron-withdrawing effect
exerted by a 5-Cl or a 5-NO2 on a putative charge-
transfer interaction between the indole moiety and an
electron-poor function within the L2 site. It was not
possible to establish whether binding mode B is dis-
favored by the size and/or the electron-withdrawing
ability of R5, as these twoproperties remained collinear
in a set of the two substituents 5-Cl and 5-NO2.17The
5-OMe derivatives 1d,e and 2a,b were purposely pre-
pared and tested to obtain a slightly larger data set in
which the steric and electronic properties of R5are not
correlated. Only compound 1d (bearing a 4′-OMe on the
side phenyl ring) possessed an appreciable potency,
implying that a 5-OMe group affects the affinity of
indole derivatives, like a 5-Cl or a 5-NO2substituent.
Specifically, 5-Cl/NO2benzylamides 1 elicit nanomolar-
submicromolar Kivalues if X is electron-donating or a
hydrogen (1e′-l′), whereas 5-Cl/NO2phenylhydrazides
2 are inactive for any other type of substituent X (2d′-
g′). What makes 5-OMe, 5-Cl, and 5-NO2 similar is
clearly a steric rather than an electronic property.
Taken together, the above-summarized SARs suggest
that a 5-substituted indole cannot attain binding mode
B because it would be sterically repelled by the S2site.
Consequently, we speculate that compound 1d binds to
the receptor in accordance with mode A. The relatively
low affinity of 1d (Ki 494 nM) compared with its 5-H,
5-Cl, and 5-NO2counterparts 1b′,f ′,j′ (163 nM, 107 nM,
and 53 nM, respectively) is probably related to the
electron-donating effect of the 5-OMe, weakening the
hydrogen bond between the indole NH and the A2site.
E ffects of the R ′ and X Substituents on Affinity
in Compounds 4. Compounds with thegeneral formula
4 are moderately to highly potent when bearing a
methyl group on the benzyl R-carbon with the R con-
figuration; 4t is the most potent among the newly
synthesized indoles (Ki 17 nM). The remaining com-
pounds of type 4, where R′ is a hydrophilic (CH2OH,
CN or COOEt) or a Me group with the S configuration,
displayed no affinity at the BzR, with the exception of
the cyano derivative 4c′ (Ki241 nM). A comparison of
the binding constants of 4a, 4l, and 4t with those of the
corresponding R-desmethyl analogues 1a′, 1e′, and 1i′
(all the six ligands being unsubstituted on the side
phenyl ring) reveals that (R)-R-Me improves the affinity
of the 5-Cl/NO2derivatives by 4.5- and 7-fold, respec-
tively, whereas it lowers the affinity of the 5-H deriva-
tive by 11-fold. These divergent effects of (R)-R-Me are
probably related to the different binding modes of the
5-Cl/NO2and 5-H indoles, which direct the same (R)-
R-Me todifferent regions of the BzR. The SARs outlined
sofar areconsistent with our hypothesis19of a lipophilic
L1pocket available to the (R)-R-Me of indoles 3 and 4
binding in accordance with mode A. The shape of this
pocket is asymmetric, so that it hosts the (R)-R-Me on
one side, while it has no room available for the (S)-R-
Me on the opposite side. The affinity of 5-H indoles,
binding in accordance with mode B, is significantly
disfavored by R-methylation, probably because the
receptor cleft surrounding the benzylic R-carbon is
Within the set of 5-Cl/NO2 indoles 4, none of the
substituents X on the side phenyl ring increases affinity
(compare 4l vs 4n, p-r and 4t vs 4v, x-z). In contrast,
4′-OMe and 3′,4′-(OMe)2improve potency of 5-Cl/NO2
benzylamides 1 by 16- and 10-fold (compare 1e′ vs 1g′
and 1i′ vs 1k′). In the series of 5-H indoles 4, affinity is
favored by an electron-donating X group (compare 4a
against 4c,e,f) or abolished if X is an electron-with-
N-(Arylalkyl)indol-3-ylglyoxylylamidesJ ournal of Medicinal Chemistry, 2001, Vol. 44, No. 142289
T able 1. Physical Properties of Indolylglyoxylylamide Derivatives 1, 2, and 4-8
2290 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14Primofiore et al.
drawing 4′-NO2group (compare 4a vs 4g). These data
contrast with the effects of X in the set of 5-H indoles
1, wherein affinity is favored by an electron-withdraw-
ing X substituent such as a 4′-Cl (compare 1a′ vs 1d′)
or a 4′-NO2 (featured by the newly synthesized 1a,
which is the most potent 5-H benzylamide derivative).
Thelack of parallelism between theeffects of X in series
1 and 4 might be due to different orientations of their
side phenyl rings within the BzR depending on the
absence/presence of the (R)-R-Me group.
Indoles 5-8 Geometrically Constrained about
the N-Phenylalkyl Side Chain. Also in the series of
ligands 5-8, 5-H derivatives exhibit divergent SARs
from their 5-Cl/NO2counterparts. Unsubstituted indoles
tolerate the isoindolinylamide side chain (5a being
practically equipotent to1a′) but not the (R)-1-indanyl-
amide moiety (7a is 5.6-fold less potent than 1a′). In
contrast, 5Cl/NO2indoles show no potency at the BzR
when bearing the isoindolinylamide side chain, while
the (R)-1-indanylamide residue produces an enhance-
ment of the affinity, seeing that 7b and 7c are 6-fold
and 4-fold more potent than the corresponding benzyl-
amides 1e′ and 1i′. None of the compounds 6a-c and
8a-c possesses any significant potency. These data
further support our hypothesis of different interaction
modes available for 5-H and 5-Cl/NO2 indoles at the
BzR. The binding data of the 1-indanylamides 7a,b,c
and 8a,b,c parallel those of their open chain analogues
4a,l,t and 4b,m,u, thus suggesting that the methylene
in position 2 of the 1-indane ring (2-CH2) fits into the
same lipophilic pocket of the L1site hosting the (R)-R-
Isoindolinylamides 5 are the most rigid structures
among thosediscussed in thepresent paper: thetorsion
angles (OdC)-N-C-C1′ and N-C-C1′-C2′ are both
frozen in a staggered conformation (their values are
180°). Based on our model of binding mode A,17the
transoid disposition of the former torsion angle should
favor affinity of the 5-Cl and 5-NO2derivatives 5b and
5c. Actually, the inactivity of these two compounds
T able 1. (Continued)
aElemental analyses for C, H, N, were within (0.4% of the calculated values.
N-(Arylalkyl)indol-3-ylglyoxylylamidesJ ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14 2291
T able 2. Inhibition of [3H]flumazenil Specific Binding to Bovine Brain Membranes and GABA Ratios of Indolylglyoxylylamide
Derivatives 1, 2, and 4-8
2292 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14Primofiore et al.
depends on poor shape complementarity with the recep-
tor rather than on lack of conformational requirements.
In fact, Figure 2 shows that when 5c is oriented in
accordancewith modeA, theisoindolinebenzenemoiety
projects into the sterically hindered S2site mapped by
ring F of theinactivebenzopyridodiindole 9.28Theactive
(R)-R-Me benzylamide 4t, also aligned in accordance
with mode A, avoids the steric clash with the S2site by
virtue of its side phenyl ring twisted out of the main
plane of the molecule. More specifically, the torsion
angle N-C-C1′-C2′ is 60° in 4t, whereas, as already
mentioned, it is fixed to 180° in the isoindolinylamides
5. On the other hand, it is worth noting that when the
active 5-H isoindoline 5a adopts binding mode B, the
indole moiety fills the L2site without contacting the S2
The lack of affinity exhibited by the 2-indanylamides
6a-c and the (S)-1-indanylamides 8a-c implies that
neither mode A nor mode B of interaction are allowed
for thesecompounds. The5-Cl/NO2(R)-1-indanylamides
7b and 7c are the only ones, among the newly investi-
T able 2. (Continued)
aKivalues are means ( SEM of three determinations.bGABA ratio ) (Kiwithout GABA)/(Kiwith GABA).cNot determined for the
compounds (10 µM) showing percentages of inhibition of specific [3H]flumazenil binding e80%.
F igure 2. The isoindolinylamides 5c (inactive, in red) and
5a (active, in yellow) oriented according to mode A and,
respectively, mode B are aligned on the (R)-R-Me-benzylamide
4t (active, in gray) and the benzopyridodiindole 9 (inactive,
in cyan). The fused benzene ring F of 9 maps the sterically
N-(Arylalkyl)indol-3-ylglyoxylylamides J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14 2293
gated closed-chain analogues, capable of ensuring a
tight binding (probably in accordance with mode A).
Figure 3 shows an overlay of the 5-NO2derivatives
7c (active), 6c (inactive), and 8c (inactive) illustrating
how we interpret the binding data of indoles 6-8 at the
molecular level. The side chains of the three ligands
attain a transoid conformation (required for binding
mode A) which ensures the match of the corresponding
terminal fused-benzene rings expected to fill the L2
pocket. An arrow highlights a region of the receptor
whose occupancy by a ligand, such as 6c or 8c, com-
promises binding for steric reasons. The indane 2-CH2
fragment of the (R)-1-indanylamide 7c, supposed to
enhance affinity through a hydrophobic interaction, is
surrounded by a cartoon of the lipophilic pocket L1. We
believe that this BzR site hosts similarly also the (R)-
R-Me group of indole derivatives 4.
The 5-H derivatives 6a, 7a, and 8a are poorly active
or not active at all, probably because branching at the
carbon bound tothe amidic nitrogen, a structural motif
common to these compounds as well as to the 5-H
R-methylbenzylamides 4, interferes sterically with bind-
T able 3. Inhibition of [3H]flumazenil Specific Binding to Bovine Brain Membranes and GABA Ratios of Indolylglyoxylylamide
a-cSee Table 2 footnotes.
F igure 3. Superposition of the (R)-1-indanylamide 7c (active,
in yellow) on the (S)-1-indanylamide 8c (inactive, in red) and
the 2-indanylamide 6c (inactive, in cyan), all oriented in
accordance with mode A. Sterically unfavorable fragments of
the inactive compounds 8c and 6c are marked by an arrow. A
cartoon of the L1 lipophilic pocket surrounds the favorable
2-CH2fragment of the potent ligand 7c.
2294J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14Primofiore et al.
ing mode B. This hypothesis is consistent with the
significantly higher potency elicited by the isoindolin-
ylamide 5a which is characterized by twononbranched
Using an exhaustively washed membrane prepara-
tion, the GABA ratio values of the most active com-
pounds of each series 1, 4, 5, and 7 were evaluated. All
the products tested showed values close to unity,
predicting antagonist properties. A correspondence be-
tween the GABA ratio and the pharmacological profile
has already been reported for thestructurally analogous
Binding of Selected Compounds at r1?2γ2, r3?2γ2,
and r5?3γ2GABAA/Bz R eceptor Subtypes. A number
of compounds (1a, 4a, 4t, 5a, 7a, 7c, 1j′, and 2c′) were
tested for their ability to displace [3H]flumazenil from
recombinant rat R1?2γ2, R3?2γ2, and R5?3γ2GABAA/Bz
receptor subtypes (Table 4). A good correlation exists
between the affinities at wild-type and R1?2γ2subtype
receptors. Most of the ligands showed enhanced affini-
ties for the R1?2γ2isoform, with compounds 4t and 7a
exhibiting the highest selectivity over both R3?2γ2and
The GABA ratios close to unity exhibited by these
compounds at the R1?2γ2 receptor subtype were pre-
dictive of antagonist properties. Interestingly, it has
recently been proposed that R1-selective ligands, such
as the antagonist tert-butyl ?-carboline-3-carboxylate
(BCCT), may be useful for the treatment of alcohol
The SARs developed from the new series of N-
(arylalkyl)indol-3-ylglyoxylylamide derivatives 1 and
4-8 further support our hypothesis of two different
binding modes selected by these ligands depending on
the size of the substituent in the 5-position of the indole
nucleus. Specifically, 5-substituted and 5-unsubstituted
indoles bind preferentially in accordance with mode A
and B, respectively. Using molecular modeling methods,
we inferred the conformational and stereoelectronic
properties of the amide side chains leading to a high
affinity. The binding data of optically active indole
derivatives suggested that the shape of the lipophilic
pocket L1 is asymmetric. A subset of the compounds
tested on recombinant GABAA/BzR R1?2γ2, R3?2γ2, and
R5?3γ2subtypes showed an enhanced affinity at the R1
containing GABAA isoform.
E xperimental Section
Reichert Ko ¨fler hot-stage apparatus and are uncorrected. The
[R]D values were measured with a Perkin-Elmer Model 241
polarimeter in freshly distilled DMF solution. Infrared spectra
were obtained on a PYE/UNICAM mod. PU 9561 spectropho-
tometer in Nujol mulls. Nuclear magnetic resonance spectra
were recorded in DMSO-d6on a Varian CFT-20 spectrometer
operating at 80 MHz using tetramethylsilane (TMS) as the
internal standard. Mass spectra were obtained on a Hewlett-
Packard 5988 A spectrometer using a direct injection probe
and an electron beam energy of 70 eV. Magnesium sulfate was
always used as the drying agent. Evaporations were made in
vacuo (rotary evaporator). Analytical TLC was carried out on
Merck 0.2 mm precoated silica gel aluminum sheets (60 F-254).
Elemental analyses were performed by our Analytical Labora-
tory and agreed with theoretical values to within (0.4%.
Besides the commercially available starting materials, the
following products were prepared in accordance with reported
General Procedure for the Synthesis of N-[(5-Substi-
tuted indol-3-yl)glyoxylyl]amide Derivatives 1, 2, and
4-8. Triethylamine(3.0 mmol) was added dropwisetoa stirred
suspension, cooled at 0 °C, of indolylglyoxylyl chloride (2.5
mmol) and the appropriate amine (2.75 mmol) in 50 mL of
dry toluene (THF for compounds 4i and 4bb). The reaction
mixture was left towarm toroom temperature, stirred for 24-
36 h (TLC analysis), and then filtered. Theprecipitatecollected
was triturated with a saturated NaHCO3 aqueous solution,
washed with water, and collected again to give a first portion
of crudeproduct. Thetoluene(or THF) solution was evaporated
to dryness, and the residue was treated with saturated
NaHCO3aqueous solution, washed with water, and collected
toyield an additional amount of crude product. The quantities
of amide derivatives obtained from the initial insoluble
precipitateor from thetoluene(or THF) solution werevariable,
depending upon the solubility of the various compounds. All
products 1, 2, and 4-8 were purified by recrystallization from
the appropriate solvent. Yields, recrystallization solvents, and
melting points are listed in Table 1. IR,1H NMR, and MS
spectral data are reported in the Supporting Information.
Binding Studies. [3H]Flumazenil (specific activity 70.8 Ci/
mmol) was obtained from NEN Life Sciences Products. All
other chemicals were of reagent grade and were obtained from
Bovine cerebral cortex membranes were prepared in ac-
cordance with ref 32. The membrane preparations were
subjected to a freeze-thaw cycle, washed by suspension and
centrifugation in 50 mM tris-citrate buffer pH 7.4 (T1), and
then used in the binding assay. Protein concentration was
assayed by the method of Lowry et al.33
[3H]Flumazenil binding studies were performed as previ-
Clonal mammalian cell lines expressing relatively high
levels of rat GABAA receptor subtypes (R1?2γ2, R3?2γ2, R5?3γ2)
were maintained, as previously described34in Minimum Es-
sential Medium Eagle with EBSS, supplemented with 10%
fetal calf serum, L-glutamine (2 mM), penicillin (100 units/
mL), and streptomycin (100 µg/mL) in a humidified atmo-
sphere of 5% CO2/95% air at 37 °C. Cells were harvested and
then centrifuged at 500 × g. The crude membranes were
prepared after homogenization in 10 mM potassium phos-
phate, pH 7.4, and differential centrifugation at 48 000 × g
for 30 min at 4 °C. The pellets were washed twice in this
manner before final resuspension in 10 mM potassium phos-
phate, pH 7.4, containing 100 mM potassium chloride.34
Melting points were determined using a
T able 4. Inhibition of [3H]Flumazenil Specific Binding and
GABA Ratios of Selected Compounds at Rat R1?2γ2, R3?2γ2, and
R5?3γ2GabaA/Bz Receptor Subtypesa
Ki(nM)bor % inhibition (10 µM)c
16 ( 2
1150 ( 86
14 ( 2
224 ( 20
225 ( 13
9 ( 0.6
42 ( 3
25 ( 3
50 ( 3
1600 ( 110
5% ( 0.4
283 ( 19
3200 ( 150
28% ( 3
1960 ( 150
137 ( 11
40 ( 4
765 ( 63
58 ( 6
5500 ( 360
239 ( 21
43% ( 4
2160 ( 160
95 ( 8
126 ( 11
43 ( 5
35% ( 3
aThe ability of the compounds to displace [3H]flumazenil was
measured in membranes from HEK293 cells expressing theR1?2γ2,
R3?2γ2, and R5?3γ2 subtypes, as described in the Experimental
Section.bKi values are means ( SEM of three determinations.
cPercentage inhibition values of specific [3H]flumazenil binding
at 10 µM concentration aremeans ( SEM of threedeterminations.
dGABA ratio ) (Kiwithout GABA)/(Kiwith GABA).
N-(Arylalkyl)indol-3-ylglyoxylylamides J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 142295
[3H]Flumazenil binding assays to transfected cells mem-
branes were carried out as previously described.34In brief, the
cell lines membranes were incubated in a volume of 500 µL
which contained [3H]flumazenil at a concentration of 1-2 nM
and the test compound in the range 10-9-10-5M. Nonspecific
binding was defined by 10-5M diazepam. Assays were
incubated to equilibrium for 1 h at 4 °C.
The potencies of the new synthesized compounds toinhibit
[3H]flumazenil binding in the presence and absence of GABA
were compared. The differences obtained were expressed as
the GABA ratio, namely the ratios of the Ki values obtained
in the absence of GABA over the Ki values obtained in the
presence of GABA.
Computational Chemistry. All molecular modeling was
performed using the software package SYBYL27running on a
Silicon Graphics R10000 workstation. Most of themodels were
built starting from benzylamide structures of type1 (available
from our previous works)16,17in accordance with SYBYL
standard bond lengths and valence angles. Atom centered
charges were calculated by the Gasteiger-Hu ¨ckel method.35,36
Preliminary geometry optimizations werecarried out using the
SYBYL/MAXIMIN2 minimizer based on the molecular me-
chanics Tripos force field26and the BFGS (Broyden, Fletcher,
Goldfarb, and Shanno) algorithm.37A root-mean-square gradi-
ent of the forces acting on each atom of 0.05 kcal/mol Å was
set as the convergence criterion.
Global minimum and pharmacophore-consistent conforma-
tions wereidentified using theSYBYL/SEARCH routine. With
theexception of compounds 5a and 5c, featuring a totally rigid
isoindolinylamide moiety, rotatable bonds of the arylalkyl side
chain were generally scanned through 10° increments within
the 0-350° interval. A 0.75 van der Waals scaling factor was
applied to “soften” steric contacts in the rigid rotamers. All
the conformations subjected to further modeling had a strain
energy (difference with respect to the global minimum con-
formation) not greater than 3 kcal/mol.38For each of the
compounds 4t, 6c, 7c, and 8c we selected a pharmacophore-
consistent conformation as the one featuring the largest
distance between the amidic nitrogen and the phenyl ring
centroid. This criterion allowed us to identify geometries
characterized by a side chain aryl moiety positioned within
the plane of the indole-COCONH system and therefore com-
patible with binding mode A.
The selected global minimum and pharmacophore-consis-
tent conformations were subjected to full geometry optimiza-
tions performed with the semiempirical quantum-mechanics
method AM125available in the MOPAC program.39MOPAC
was run using the keywords “XYZ” and “MMOK”. The result-
ing pharmacophore-consistent conformers, defined by the
following torsion angles, were all coincident with global
minima: 4t: (O))C-N-C-C1′Ar ) 159°, N-C-C1′Ar-C2′Ar
) 60°; 5a and 5c: (O))C-N-C-CAr) 180°; 6c: (O))C-N-
C2-C1 )108°; 7c: (O))C-N-C-CAr) 155°; 8c: (O))C-N-
C-CAr ) 162°.
The model of the benzopyridodiindole 9 shown in Figure 2
was availablefrom a previous work.17Molecular superpositions
wereaccomplished following theprocedures described by Cook
Acknowledgment. This work was supported by
grants from the Ministry of University and Scientific
and Technological Research (MURST).
Supporting Information Available: Table containing
theIR,1H NMR, and MS spectral data of compounds 1, 2, 4-7.
This material is available free of charge via the Internet at
(1) Huang, Q.; He, X.; Ma, C.; Liu, R.; Yu, S.; Dayer, C. A.; Wenger,
G. R.; McKernan, R.; Cook, J . M. Pharmacophore/Receptor
Models for GABAA/BzR Subtypes (R1?3γ2, R5?3γ2, and R6?3γ2) via
a Comprehensive Ligand-Mapping Approach. J . Med. Chem.
2000, 43, 71-95.
(2) Sieghart, W. Structure and Pharmacology of γ-Aminobutyric
AcidAReceptor Subtypes. Pharmacological Rev. 1995, 47, 181-
(3) Haefely, W.; Martin, J . R.; Schoch, P. Novel Anxiolytics That
Act as Partial Agonists at Benzodiazepine Receptors. Trends
Pharmacol. Sci. 1990, 11, 452-456.
(4) Gardner, C. R. Interpretation of the Behavioral Effects of
Benzodiazepine Receptor Ligands. Drugs Future1989, 14, 51-
(5) Haefely, W. E.; Kyburz, E.; Gerecke, M.; Mohler, H. P. Recent
Advances in the Molecular Pharmacology of Benzodiazepine
Receptors and in the Structure-Activity Relationship of Their
Agonists and Antagonists. Adv. Drug Res. 1985, 14, 166-322.
(6) Korpi, E. R.; Mattila, M. J .; Wisden, W.; Luddens, H. GABAA-
receptor Subtypes: Clinical Efficacy and Selectivity of Benzo-
diazepine Site Ligands. Annals Med. 1997, 29, 275-282.
(7) Gu, Z. Q.; Wong, G.; Dominguez, C.; de-Costa, B. R.; Rice, K. C.;
Skolnick, P. Synthesis and Evaluation of Imidazo[1,5-a][1,4]-
benzodiazepine Esters with High Affinities and Selectivities at
Diazepam Insensitive (DI) Benzodiazepine Receptors. J . Med.
Chem. 1993, 36, 1001-1006.
(8) Wong, G.; Koehler, K. F.; Skolnick, P.; Gu, Z. Q.; Ananthan, S.;
Schonholzer, P.; Hunkeler, W.; Zhang, W.; Cook, J . M. Synthetic
and Computer Assisted Analysis of theStructural Requirements
for Selective, High Affinity Ligand Binding to “Diazepam-
Insensitive” Benzodiazepine Receptors. J . Med. Chem. 1993, 36,
(9) Liu, R.; Hu, R. J .; Zhang, P.; Skolnick, P.; Cook, J . M. Synthesis
and Pharmacological Properties of Novel 8-Substituted Imida-
zobenzodiazepines: High-Affinity, Selective Probes for R5-
Containing GABAAReceptors. J . Med. Chem. 1996, 39, 1928-
(10) Arbilla, S.; Depoortere, H.; George, P.; Langer, S. Z. Pharma-
cological Profile of the Imidazopyridine Zolpidem at Benzodiaz-
epine Receptors and Electrocorticogram in Rats. Naunyn-
Schmiedebergs Arch. Pharmacol. 1985, 330, 248-251.
(11) Yokoyama, N.; Ritter, B.; Neubert, A. D. 2-Arylpyrazolo[4,3-c]-
quinolin-3-ones: Novel Agonist, Partial Agonist, and Antagonist
of benzodiazepines. J . Med. Chem. 1982, 25, 337-339.
(12) Shindo, H.; Takada, S.; Murata, S.; Eigyo, M.; Matsushita, A.
Thienylpyrazoloquinolines with High Affinity toBenzodiazepine
Receptors: Continuous Shift from Inverse Agonist to Agonist
Properties Depending on the Size of the Alkyl Substituent. J .
Med. Chem. 1989, 32, 1213-1217.
(13) Allen, M. S.; Hagen, T. J .; Trudell, M. L.; Codding, P. W.;
Skolnick, P.; Cook, J . M. Syntheses of Novel 3-substituted
?-Carbolines as Benzodiazepine Receptor Ligands: Probing the
Benzodiazepine Receptor Pharmacophore. J . Med. Chem. 1988,
(14) Trudell, M. L.; Lifer, S. L.; Tan, Y. C.; Martin, M. J .; Deng, L.;
Skolnick, P.; Cook, J . M. Synthesis of Substituted 7,12-Dihy-
dropyrido[3,2-b:5,4-b′]diindoles: Rigid Planar Benzodiazepine
Ligands with Inverse Agonist/Antagonist properties. J . Med.
Chem. 1990, 33, 2412-2420.
(15) Zhang, W.; Koehler, K. F.; Zhang, P.; Cook, J . M. Development
of a Comprehensive Pharmacophore Model for the Benzodiaz-
epine Receptor. Drug Design Discovery 1995, 12, 193-248.
(16) Da Settimo, A.; Primofiore, G.; Da Settimo, F.; Marini, A. M.;
Novellino, E.; Greco, G.; Martini, C.; Giannaccini, G.; Lucacchini,
A. Synthesis, Structure-Activity Relationships, and Molecular
Modeling Studies of N-(Indol-3-ylglyoxylyl)benzylamine Deriva-
tives Acting at the Benzodiazepine Receptor. J . Med. Chem.
1996, 39, 5083-5091.
(17) Da Settimo, A.; Primofiore, G.; Da Settimo, F.; Marini, A. M.;
Novellino, E.; Greco, G.; Gesi, M.; Martini, C.; Giannaccini, G.;
Lucacchini, A. N′-Phenylindol-3-ylglyoxylohydrazide Deriva-
tives: Synthesis, Structure-Activity Relationships, Molecular
Modeling Studies, and Pharmacological Action on Brain Benzo-
diazepine Receptors. J . Med. Chem. 1998, 41, 3821-3830.
(18) Martini, C.; Gervasio, T.; Lucacchini, A.; Da Settimo, A.;
Primofiore, G.; Marini, A. M. Specific Inhibition of Benzodiaz-
epine Receptor Binding by Some N-(Indol-3-ylglyoxylyl)amino
Acid Derivatives. J . Med. Chem. 1985, 28, 506-509.
(19) Primofiore, G.; Marini, A. M.; Da Settimo, F.; Martini, C.;
Bardellini, A.; Giannaccini, G.; Lucacchini, A. Specific Inhibition
of Benzodiazepine Receptor Binding by Some N-(Indol-3-
ylglyoxylyl)aminoAcid Derivatives: StereoselectiveInteractions.
J . Med. Chem. 1989, 32, 2514-2518.
(20) Da Settimo, A.; Primofiore, G.; Marini, A. M.; Ferrarini, P. L.;
Franzone, J . S.; Cirillo, L.; Reboani, M. C. N-(indol-3-ylglyoxyl-
Secretory Activity. Eur. J . Med. Chem. 1988, 23, 21-24.
(21) Giannaccini, G.; Giacomelli, M.; Martini, C.; Lucacchini, A.;
Piccolino, M. Binding of the Benzodiazepine Ligand [3H]Ro15-
1788 toMembranePreparations of theRabbit and TurtleRetina.
Comp. Biochem. Physiol. 1992, 101C, 337-342.
Preparation and Gastric Anti-
2296J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 14Primofiore et al.
(22) Braestrup, C.; Nielsen, M.; Honore ´, T.; J ensen, L. H.; Petersen,
E. M. Benzodiazepine Receptor Ligands with Positive and
Negative Efficacy. Neuropharmacology 1983, 22, 1451-1457.
(23) Braestrup, C.; Nielsen, M. Benzodiazepine Receptors. In Hand-
book of Psychopharmacology; Iversen, L. L., Iversen, S. D.,
Snyder, S. H., Eds.; Plenum Press: New York, 1983; Vol. 17, pp
(24) Fryer, R. I.; Rios, R.; Zhang, P.; Gu, Z. Q.; Wong, G.; Basile, A.
S. Skolnick, P. Structure Activity Relationships of 2-Phenylpyra-
zolo[4,3-c]quinolin-3-ones and their N- and O-Methyl Analogues
at Benzodiazepine Receptors. Med. Chem. Res. 1993, 3, 122-
(25) Dewar, M. J .; Zoebisch, E. G.; Healy, E. F.; Stewart, J . J . P.
AM1: A New General Purpose Quantum Mechanical Molecular
Model. J . Am. Chem. Soc. 1985, 107, 3902-3909.
(26) Clark, M.; Cramer, R. D., III.; van Opdenbosch, N. Validation
of theGeneral PurposeTripos 5.2 ForceField. J . Comput. Chem.
1989, 10, 982-1012.
(27) SYBYL Molecular Modeling System, version 6.5; Tripos Inc.: St.
(28) Martin, M. J .; Trudell, M. L.; Diaz-Arauzo, H.; Allen, M. S.; Deng,
L.; Schultz, C. A.; Tan, Y.-C.; Bi, Y.; Narayanan, K.; Dorn, L. J .;
Koheler, K. F.; Skolnick, P.; Cook, J . M. Molecular Yardsticks:
Rigid Probes to Define the Spatial Dimensions of the Benzodi-
azepine Receptor Binding Site. J . Med. Chem. 1992, 35, 4105-
(29) Bianucci, A. M.; Da Settimo, A.; Da Settimo, F.; Primofiore, G.;
Martini, C.; Giannaccini, G.; Lucacchini, A. Benzodiazepine
Receptor Affinity and Interaction of Some N-(Indol-3-ylglyoxyl-
yl)amine Derivatives. J . Med. Chem. 1992, 35, 2214-2220.
(30) Kametani, T.; Fukumoto, K. Cularine and Related Compounds.
Part VI. A Total Synthesis of (()-Cularine. J . Chem. Soc. 1963,
(31) Bornstein, J .; Lashua, S. C.; Boisselle, A. P. Facile Synthesis of
Dihydroisoindole. J . Org. Chem. 1957, 22, 1255-1256.
(32) Martini, C.; Lucacchini, A.; Ronca, G.; Hrelia, S.; Rossi, C. A.
Isolation of Putative Benzodiazepine Receptors From Rat Brain
Membranes by Affinity Chromatography. J . Neurochem. 1982,
(33) Lowry, O. H.; Rosebrough, N. J .; Farr, A. L.; Randall, R. J .
Protein Measurement With the Folin Phenol Reagent. J . Biol.
Chem. 1951, 193, 265-275.
(34) Besnard, F.; Even, Y.; Itier, V.; Granger, P.; Partiseti, M.; Avenet,
P.; Depoortere, H.; Graham, D. Development of StableCell Lines
Expressing Different Subtypes of GABAAReceptors. J . Rec. Sign.
Transd. Res. 1997, 17, 99-113.
(35) Gasteiger, J .; Marsili, M. IterativePartial Equalization of Orbital
Electronegativity. Tetrahedron 1980, 36, 3219-3228.
(36) Purcel, V. P.; Singer, J . A. A Brief Review of Table and
Semiempirical Parameters Used in theHu ¨ckel Molecular Orbital
Method. J . Chem. Eng. Data 1967, 12, 235-246.
(37) Head, J .; Zerner, M. C. A Broyden-Fletcher-Goldfarb-Shanno
Optimization Procedure for Molecular Geometries. Chem. Phys.
Lett. 1985, 122, 264-274.
(38) Bostro ¨m, J .; Norrby, P.-O.; Liljefors, T. Conformational Energy
Penalties of Protein-Bound Ligands. J . Comput.-Aided Mol.
Design 1998, 12, 383-396.
(39) MOPAC, version 6.0; Quantum Chemistry Program Exchange.
N-(Arylalkyl)indol-3-ylglyoxylylamidesJ ournal of Medicinal Chemistry, 2001, Vol. 44, No. 142297