Synthesis, pharmacological characterization and QSAR studies on 2-substituted indole melatonin receptor ligands

Article abstract of DOI:10.1016/S0968-0896(00)00322-9

A number of (6-methoxy-1-2-propionylaminoethyl)indoles, carrying properly selected substituents at the C-2 indole position, were prepared and tested as melatonin receptor ligands. Affinities and intrinsic activities for the human cloned mt₁ and

Full text of DOI:10.1016/S0968-0896(00)00322-9

Bioorganic & Medicinal Chemistry 9 (2001) 1045±1057
Synthesis, Pharmacological Characterization and QSAR Studies
on 2-Substituted Indole Melatonin Receptor Ligands
Marco Mor,a Gilberto Spadoni,b,* Barbara Di Giacomo,b Giuseppe Diamantini,b
Annalida Bedini,b Giorgio Tarzia,b Pier Vincenzo Plazzi,a Silvia Rivara,a
Romolo Nonno,c Valeria Lucini,c Marilou Pannacci,c Franco Fraschinic
and Bojidar Michaylov Stankovc
a
Á
Dipartimento Farmaceutico, Universita degli Studi di Parma, Parco Area delle Scienze 27A, I-43100 Parma, Italy
b
Á
Istituto di Chimica Farmaceutica e Tossicologica, Universita degli Studi di Urbino, Piazza Rinascimento 6, I-61029 Urbino, Italy
Cattedra di Chemioterapia, Dipartimento di Farmacologia, Universita degli Studi di Milano, Via Vanvitelli 32, I-20129 Milano, Italy
c
Á
Received 1 August 2000; revised 22 November 2000; accepted 28 November 2000
AbstractÐA number of 6-methoxy-1-(2-propionylaminoethyl)indoles, carrying properly selected substituents at the C-2 indole
position, were prepared and tested as melatonin receptor ligands. Anities and intrinsic activities for the human cloned mt1 and
MT2 receptors were examined and compared with those of some 2-substituted melatonin derivatives recently described by us. A
quantitative structure±activity relationship (QSAR) study of the sixteen 2-substituted indole compounds, 5a±k, 1, 8±11, using par-
tial least squares (PLS) and multiple regression analysis (MRA) revealed the existence of an optimal range of lipophilicity for the C-
2 indole substituent. There are also indications that planar, electron-withdrawing substituents contribute to the anity by estab-
lishing additional interactions with the binding pocket. No mt1/MT2 subtype selectivity was observed, with the relevant exception
of the 2-phenethyl derivative 5e, which exhibited the highest selectivity for the h-MT2 receptor among all the compounds tested
(MT2/mt1 ratio of ca. 50). Conformational analysis and superposition of 5e to other reported selective MT2 ligands revealed
structural and conformational similarities that might account for the MT2/mt1 selectivity of 5e. # 2001 Elsevier Science Ltd. All
rights reserved.
Introduction
Most of the functions of MLT in vertebrates are medi-
ated by the action of MLT on speci®c high-anity
membrane receptors,9À12 recently classi®ed as mt1, MT2,
putative MT3 and Mel1c.13 Mel1c is found only in lower
vertebrates and birds while mt1 and MT2 have been
found in higher vertebrates and humans.
Melatonin (N-acetyl-5-methoxytryptamine, MLT, 1) is
the principal hormone secreted by the pineal gland during
the dark period of the light/dark cycle.1 MLT in¯uences
a variety of functions such as seasonal reproduction in
photoperiodic animals, circadian rhythms, sleep pro-
cesses, immune functions, thermoregulation, oxidant/
antioxidant balance, cancer biology, and others.2 Although
its exact role in humans still remains unclear, research on
MLT suggests possible clinical implications, especially in
the treatment of circadian-rhythm-based sleep disorders.3À7
Protective e€ects on the cardiovascular system by reducing
the risk of atherosclerosis and hypertension8 have also
been reported. The potential therapeutic implications of
MLT justify the considerable e€orts devoted to under-
stand how MLT interacts with its receptors.
Previous extensive studies on indole derivatives and
indole bioisosteres of MLT demonstrated that suitably
spaced methoxy and amido moieties are important for
binding to and activation of the melatonin receptor.14,15
To improve the information about the essential phar-
macophoric features of the MLT molecule, and to
establish which regions can tolerate structural modulation,
we have described several indole compounds exhibiting
anity and intrinsic activity for the melatonin receptors
to varying degrees.16,17 Numerous conformationally
constrained MLT analogues have also been synthesized
and tested to elucidate the bioactive conformation of
MLT (for extensive reviews see refs. 14,15,18). Starting
from these studies, we and other authors have recently
*Corresponding author. Tel.: +39-722-2545; fax: +39-722-2737;
e-mail: gilberto@uniurb.it
0968-0896/00/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(00)00322-9

1046
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
proposed molecular models of the putative melatonin
binding site (for a review see ref. 15).
Selectivity toward mt1 and MT2 melatonin receptor
subtypes was also investigated; ®nally molecular modeling
was employed to investigate the steric features of the
most ¯exible substituents which showed some MT2
selectivity.
Although the limited series of the known 2-substituted
compounds suggested that C-2 halogens or phenyl
group19À21 contribute signi®cantly to the receptor binding,
a systematic investigation on the relevance of this C-2
substituent is lacking. The modulatory e€ect on anity
had been previously ascribed to the in¯uence of the C-2
substituent on the ethylamido side chain folding.22
However, the remarkable increase in binding anity
observed in 2-substituted rigid melatonergic ligands
favours the existence of an auxiliary binding site around
the C-2 indole position.23 Increased binding anity and
MT2 selectivity were also observed in some melatonin
antagonists/partial agonists, by introducing a benzyl or
a phenyl group at the C-2 indole or in a topologically
equivalent position.24,25
Results
The goal of this study was to prepare a series of 2-sub-
stituted-1-(2-propionylaminoethyl)indoles (Schemes 1
and 2) and to measure their anity and ecacy at the h-
mt1 and h-MT2 receptors (Table 1). A total of 16 com-
pounds were then examined in a quantitative structure±
activity relationship (QSAR) analysis (Tables 3 and 5)
to investigate quantitatively the e€ect of properly selec-
ted lipophilic and hydrophilic C-2 substituents on mel-
atonin receptor anity, ecacy and selectivity.
In a previous paper,26 we reported that the shift of the
MLT side chain from C-3 to N-1 indole position a€ords
a series of melatonin bioisosteres (II) with anity and
SAR comparable to that of the natural melatonin series,
provided that the methoxy group is in the C-6 indole
position.
Chemistry
The 2-substituted melatonin analogues 8±919,20 and
related bioisosters 5a±k (Table 1) were synthesized by
known chemical processes. The general synthetic route
involved an intermediary 2-substituted-5- or 6-methoxy
indole. Some of these indoles were prepared according
to previously described procedures (2a,32 2b,33 and 2d25).
By reacting 6-methoxy-1H-indole with dimethyldisul®de
and sulfonylchloride, according to a method previously
described for related compounds,34 we obtained a mix-
ture of mono- and bis-methylthio indole isomers that
was converted to 6-methoxy-2-methylthio-1H-indole by
acid-catalyzed isomerization with polyphosphoric acid
(PPA) at 100 ꢀC. The thiomethyl group was then readily
oxidized to 2-methanesulfonyl-1H-indole (2c) with
magnesium monoperoxyphtalate hexahydrate (MMPP)
under biphasic conditions (H2O/CH2Cl2/benzyltriethyl-
ammonium chloride).
In our continuing interest in the characterization of MLT
receptors, we have further synthesized and pharmacologi-
cally characterized new 2-substituted MLT-bioisosteres
(II) in order to get quantitative correlations, explaining
the dependence of the biological activity and mt1/MT2
selectivity on the physicochemical features of the C-2
indole substituent. Substituents were chosen to provide
both a broad distribution of physicochemical properties
(lipophilicity and electronic and steric e€ects) and minimal
cross-correlation. Their selection was based on sub-
stituent properties, synthetic feasibility and on pre-
viously reported 2-substituted MLT analogues.24,27,28
Three disjoint principal properties (DPPs), obtained by
van de Waterbeemd et al.29 from an extended set of
variables for 59 substituents, were employed for a con-
cise description of physicochemical properties necessary
for an optimization study.30 Starting from a putative
`optimal' set of substituents, composed by the eight
combinations of high and low levels for each property
(corresponding to a factorial design31), substituent
replacements and additions were performed (see
Experimental).
A literature procedure26 (Scheme 1) was used for the
synthesis of N-[2-(2-substituted-6-methoxy-1H-indol-1-yl)-
ethyl]propionamido derivatives (5a±c, e). N-cyanomethyl-
ation of indoles 2a±d with sodium hydride and chloro-
acetonitrile in DMF gave the required 1-cyanomethyl-
indoles 3a±d. Intermediates 3a±c were converted to the
®nal compounds 5a±c by hydrogenation over Raney
nickel and concomitant N-acylation with propionic
anhydride. Aldehyde 3d was subjected to Wittig reac-
tion conditions35 using benzyltriphenylphosphonium
chloride in the presence of 1,7-diazabicyclo[4.5.0.]undec-
6-ene (DBU) to give an E/Z mixture of the 2-styryl
derivative 4 which was converted to the 2-phenethyl
derivative 5e by hydrogenation over Raney nickel and
concomitant N-acylation with propionic anhydride. (In
some cases it was necessary to terminate the reduction
of the styryl group by catalytic hydrogenation with 10%
palladium over carbon at room temperature for 16 h.)
The pharmacological results of the corresponding
derivatives, synthesized and tested on mt1 and MT2
melatonin receptor subtypes, are presented in Table 1,
and the correlation matrix of their DPPs is reported in
Table 2.
1-(2-Propionylamino-ethyl)-6-methoxy-1H-indol-1-yl-2-
carboxylic acid 6, obtained by alkaline ester hydrolysis
of 5a, was reacted with thionyl chloride in dry tetra-
hydrofuran (THF) and the crude acid chloride was treated
with a saturated solution of ammonia in dichloromethane

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1047
Table 1. Binding anitya and intrinsic activity of MLT and compounds 5a±k, 8±11 for the human mt1 or MT2 melatonin receptors stably
expressed in NIH3T3 cells
Human mt1
Human MT2
c
Compd.
R1
pKi ÆSEM
Relative intrinsic activityb
pKi ÆSEM
Relative intrinsic activityb
pKi1ÀpKi2
5a
5b
5c
5d
5e
5f
5g
5h
5i
COOMe
CF3
SO2Me
CHO
CH2CH2Ph
CONH2
NHCONH2
CH2OH
H
9.82Æ0.07
9.52Æ0.12
6.83Æ0.07
8.95Æ0.04
7.20Æ0.07
7.57Æ0.07
6.65Æ0.03
7.46Æ0.0
0.98Æ0.02
0.83Æ0.01
0.79Æ0.01
0.98Æ0.05
0.75Æ0.01
0.98Æ0.02
0.89Æ0.04
0.96Æ0.04
0.93Æ0.06
0.92Æ0.03
0.95Æ0.01
10.12Æ0.05
10.08Æ0.12
7.34Æ0.13
9.34Æ0.05
8.88Æ0.12
7.79Æ0.08
7.21Æ0.15
7.87Æ0.05
9.21Æ0.09
10.36Æ0.09
10.21Æ0.06
1.33Æ0.08
1.02Æ0.06
0.78Æ0.07
1.11Æ0.06
0.27Æ0.05
1.02Æ0.07
0.72Æ0.07
0.94Æ0.01
1.03Æ0.07
1.17Æ0.01
1.27Æ0.08
À0.30
À0.56
À0.51
À0.39
À1.68
À0.22
À0.56
À0.41
À0.44
À0.38
+0.07
8.77Æ0.09
9.97Æ0.05
10.28Æ0.03
5j
5k
Ph
Br
MLT (1)
H
Br
Ph
I
CH2Ph
9.63Æ0.03
10.54Æ0.04
10.66Æ0.06
10.64Æ0.03
7.5d
1
9.43Æ0.03
9.94Æ0.06
10.42Æ0.07
10.29Æ0.05
9.6d
1
+0.20
+0.60
+0.24
+0.35
À2.10
2-BrMLT (8)
2-PhMLT (9)
2-IMLT (10)
2-BnMLT (11)d
0.98Æ0.01
1.05Æ0.02
0.98Æ0.01
1.02Æ0.01
0.96Æ0.01
apKi values were calculated from IC50 values obtained from competition curves by the method of Cheng and Pruso€40 and are the means of at least
three independent determinations performed in duplicate.
bThe relative intrinsic activity values were obtained by dividing the maximal analogue-induced G-protein activation by that of MLT.
cThe di€erence (pKi1ÀpKi2) represents selectivity towards the mt1 (positive values) or the MT2 (negative values) subtype.
dpKi values from ref 24.
Table 2. Correlation matrix for disjoint principal properties of 2-
substituents in compounds 5a±k
human mt1 and MT2 receptor subtypes stably expressed
in NIH3T3 rat ®broblast cells and the results are sum-
marized in Table 1. The characterization of NIH3T3-
mt1 and MT2 cells was described in detail elsewhere.28,37
S_1
L_1
E_1
S_1
L_1
E_1
1.00
0.45
0.04
0.45
1.00
0.40
0.04
0.40
1.00
The activation of the G-protein subsequent to the
binding of compounds 5a±k to mt1 and MT2 receptor
subtypes was determined by measuring the speci®c
to yield the 2-carboxylic acid amide derivative 5f
(Scheme 2). The acid 6 was subjected to a modi®ed
Curtius reaction36 to obtain the carbonylazido deriva-
tive 7. Treatment of 7 with ammonium acetate in
re¯uxing toluene provided the 2-ureido derivative 5g
(Scheme 2).
binding
of
[35S]-guanosine-50-(3-thiotriphosphate)
([35S]GTPgS) to membranes prepared from NIH3T3
cells expressing human cloned mt1 or MT2 melatonin
receptors. The detailed description and validation of
this method was reported elsewhere.25,28,37 In both
human mt1 and MT2 receptor-expressing cell lines,
MLT produced a concentration-dependent stimulation
of basal [35S]GTPgS binding with a maximal stimula-
tion of 370% in mt1 and 250% in MT2 above basal
levels, respectively. The amount of bound [35S]GTPgS is
proportional to the level of the analogue-induced G-
protein activation, and it is related to the intrinsic
activity of the compounds. Full agonists increased the
basal [35S]GTPgS binding in a concentration-dependent
manner, likewise the natural ligand MLT, whereas par-
tial agonists increased it to a much lesser extent than
MLT and antagonists were without any e€ect. The
relative intrinsic activity values were obtained by divid-
ing the maximal G-protein activation of a test com-
pound by that of MLT and the results are summarized
in Table 1.
The 2-carboxymethyl derivative 5a was reduced by
LiAlH4 to the hydroxymethyl derivative 5h, that was
oxidized by MnO2 to the 2-carboxaldehyde derivative
5d (Scheme 2).
Pharmacology
Evaluation of the compounds was performed by mea-
suring the melatonin receptor binding anity and in
vitro functional activity.
The binding anity of compounds 5a±k was determined
using 2-[125I]iodomelatonin (100 pM) as a labelled
ligand in competition binding analyses on cloned

1048
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
Scheme 1. Reagents and conditions: (i) NaH, ClCH2CN, DMF, room temperature, 16 h; (ii) (Ph)3P+CH2C6H5ClÀ, DBU, toluene, re¯ux, 4 h;
(iii) Raney nickel, H2, 4 atm, (EtCO)2O, THF, 50 ꢀC, 6 h (24 h for 5e); (iv) Pd-C, H2, THF, room temperature, 16 h.
Scheme 2. Reagents and conditions: (i) 3 N KOH, MeOH, THF, room temperature, 16 h; (ii) (PhO)2P(O)N3, Et3N, benzene, room temperature,
24 h; (iii) CH3COOÀNH4+, toluene, re¯ux, 5 min; (iv) SOCl2, THF, 50 ꢀC, 4 h; (v) NH3, CH2Cl2/THF, room temperature, 2 h; (vi) LiAlH4, THF,
10 ꢀC, 1 h; (vii) MnO2, CH2Cl2, room temperature, 7 h.
QSAR
pKi ꢀMT2† ˆ 0:93ꢀÆ0:04† pKi ꢀmt1† ‡ 0:97ꢀÆ0:37†
n ˆ 10
R2 ˆ 0:98 s ˆ 0:17
F ˆ 476:
ꢀ1†
Relationship between mt1 and MT2 anity. As can be
observed from Figure 1, representing pKi values of
compounds 5a±k for the two receptor subtypes, the
anity for MT2 receptor is strictly correlated with that
for mt1 one. A relevant exception is the phenethyl deri-
vative 5e, which shows much lower anity for mt1
receptor than the parent compound 5i, but almost the
same anity as 5i for MT2 receptor. Another feature of
this compound, partially shared by the methylsulphonyl
derivative 5c, is its low intrinsic activity on cells expres-
sing MT2 receptor subtype. Excluding from the corre-
lation the outlier 5e, the following regression equation
can be obtained:
The positive intercept indicates that pKi values on the
MT2 subtype are slightly, but systematically, higher
than on the mt1 one. This systematic di€erence is not
observed for MLT derivatives 8±10, as can be seen from
Table 1.
Multiple regression analysis (MRA). No single variable,
among those reported in Table 3, provided a statistically
good linear model either for mt1 or for MT2 receptor
binding anity, whereas a parabolic relationship
between the logarithm of the relative anities (pRA1

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1049
Table 3. Regression analysis (eqs (7) and (8)) of structure-relative
binding anity (pRA1, pRA2) of 2-substituted-1-(2-propionylamino-
ethyl)indoles (5a±k), MLT (1) and 2-melatonin derivatives (8±11)
Table 5. Descriptive (R2) and predictive (Q2) power of PLS models
with increasing number of latent variables (LV); weights of physico-
chemical properties (WX) and of relative binding anity (C) in each LV
Compd
Subst.
pRA1
pRA2
LV
1st
2nd
3rd
4th
Obs. Calc.a Res. Obs. Calc.b Res.
R2X
R2Y
Q2
0.306
0.612
0.401
0.607
0.832
0.645
0.789
0.864
0.645
0.929
0.895
0.687
5a
5b
5c
5d
5e
5f
5g
5h
5i
COOMe
CF3
1.05 0.82 0.23 0.91 0.49 0.42
0.75 1.34 À0.59 0.87 1.09 À0.22
SO2Me À1.94 À1.54 À0.40 À1.87 À1.65 À0.22
WX
CHO
0.18 0.02 0.16 0.13 À0.25 0.38
ꢀ
MR
L
0.495
À0.064
À0.053
0.295
À0.323
À0.304
0.269
0.590
0.493
0.456
0.044
0.266
0.013
À0.063
0.005
À0.171
0.101
0.301
0.191
0.103
0.548
0.430
0.579
CH2CH2Ph À1.57 À1.24 À0.33 À0.33 À0.16 À0.17
CONH2 À1.2 À1.71 0.51 À1.42 À1.69 0.27
NHCONH2 À2.12 À1.95 À0.17 À2.00 À1.75 À0.25
CH2OH À1.31 À1.35 0.04 À1.34 À1.26 À0.08
B1
B5
Sb
sm
sp
À0.757
0.031
0.229
H
Ph
Br
0.00 0.05 À0.05 0.00 0.01 À0.01
1.20 0.04 1.16 1.15 0.52 0.63
1.51 1.26 0.25 1.00 1.03 À0.03
0.180
À0.220
5j
5k
À0.226
0.182
À0.216
0.163
1 (MLT)
0.00 0.05 À0.05 0.00 0.01 À0.01
0.91 1.26 À0.35 0.51 1.03 À0.52
1.03 0.04 0.99 0.99 0.52 0.47
1.01 1.15 À0.14 0.86 1.04 À0.18
À1.57 À0.31 À1.26 À0.14 0.32 À0.46
ꢀ2
À0.598
0.017
À0.541
À0.048
8 (2-BrMLT)
9 (2-PhMLT)
10 (2-IMLT)
11 (2-BnMLT)c
C
pRA1
pRA2
0.605
0.513
0.190
0.364
0.141
0.280
0.152
0.186
aFrom eq (7).
bFrom eq (8).
cCalculated from Ki values reported by ref 24 for Mel1a (2-BnMLT:
32.7 nM, MLT: 0.88 nM) and Mel1b (2-BnMLT: 0.25 nM, MLT:
0.18 nM) receptors expressed in COS-7 cellular lines.
Table 4. Structural descriptors employed for QSAR analysis29
ꢀ
MR
Aromatic substituent constants for lipophilicity
Molar refractivity
LSecond generation length STERpaIMramOeLter
B1
B5
Sb
sm
sp
Second generation minimum width STERIMOLparameter
Second generation maximum width STERIMOLparameter
Austel's steric branching parameter
Hammett constant for meta substitution
Hammett constant for para substitution
and pRA2) and lipophilicity can be recognized from
Figure 2, and can be expressed by the following equa-
tions:
pRA1 ˆ 0:90ꢀÆ0:17† ꢀ À 0:58ꢀÆ0:12† ꢀ2 ‡ 0:74ꢀÆ0:30†
n ˆ 11 R2 ˆ 0:80 s ˆ 0:67 F ˆ 16:5 Q2 ˆ 0:38
Figure 1. Relationship between mt1 and MT2 anity for compounds
5a±k. The line represents eq (1) (see text).
SDEP ˆ 1:01 Optimal ꢀ ˆ 0:78
ꢀ2†
pRA2 ˆ 0:93ꢀÆ0:12† ꢀ À 0:43ꢀÆ0:08† ꢀ2 ‡ 0:51ꢀÆ0:20†
Experimental for a de®nition), as the low anity of the
phenethyl derivative 5e suggests rather a cut-o€ value
than a parabolic trend, and the good performance of the
phenyl substituent seems to be underestimated. On the
other hand, eqs (2) and (3) represented by far the best
two-variable regression models both for pRA1 and
pRA2. Therefore, whereas other structural properties
are supposed to play a role, their in¯uence is statistically
shadowed by the parabolic correlation.
n ˆ 11 R2 ˆ 0:89 s ˆ 0:46 F ˆ 31 Q2 ˆ 0:74
SDEP ˆ 0:59 Optimal ꢀ ˆ 1:08
ꢀ3†
These equations can explain 80% and 89% of the total
variation in the dependent variable, respectively, as
expressed by the square of the correlation coecient, R2
values; the standard error of the residuals (s) is rather
large in both cases, if compared to the uncertainty of the
anity values on log scale (a di€erence of 0.3 corre-
sponds to a doubling of Ki), but the F values, being
signi®cantly higher than 1, allow to exclude, with good
con®dence, the hypothesis of chance correlation.
The best three-variable MRA models were those
including, in addition to lipophilicity, the electronic
e€ect of the substituent, without strong di€erences
between sm and sp. Eqs (4) and (5), using sm are repor-
ted, but similar correlation could also be obtained with
sP.
Although a biphasic pro®le of these relationships is evi-
dent, eq (2) lacks a good predictive power (Q2, see

1050
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
(5-methoxyluzindole, 11) behaved as 5e, being much less
potent on mt1 receptor than on MT2 in displacing the
labelled ligand.
In the 16-compound set, comprising MLT derivatives,
parabolic relationships between relative anity and
lipophilicity could still be observed, and electron-with-
drawing e€ects (described by either sm or sp) improved
the ®tting.
pRA1 ˆ 0:87ꢀÆ0:15† ꢀ À 0:48ꢀÆ0:11† ꢀ2 ‡ 2:10ꢀÆ0:89†
ꢁm ‡ 0:05ꢀÆ0:34†
n ˆ 16 R2 ˆ 0:79 s ˆ 0:65 F ˆ 15:1 Q2 ˆ 0:64
SDEP ˆ 0:74
ꢀ7†
Figure 2. Relative anity of compounds 5a±k versus lipophilicity.
Hollow circles: pRA1; ®lled circles: pRA2; dotted line: eq (2); con-
tinuous line: eq (3).
pRA2 ˆ 0:89ꢀÆ0:09† ꢀ À 0:34ꢀÆ0:06† ꢀ2 ‡ 1:31ꢀÆ0:51†
ꢁm ‡ 0:01ꢀÆ0:20†
n ˆ 16 R2 ˆ 0:90 s ˆ 0:38 F ˆ 36:4 Q2 ˆ 0:84
pRA1 ˆ 0:93ꢀÆ0:15† ꢀ À 0:51ꢀÆ0:11† ꢀ2 ‡ 1:81ꢀÆ0:90†
ꢁm ‡ 0:23ꢀÆ0:36†
SDEP ˆ 0:42
ꢀ8†
n ˆ 11 R2 ˆ 0:88 s ˆ 0:57 F ˆ 16:5 Q2 ˆ 0:53
SDEP ˆ 0:88
As in the case of the 5a±k set, the correlation is slightly
better for pRA2 values; the pRA1 value of 2-benzyl-
MLT was overestimated (thus leading to a decrease of
Q2), while the contribution to mt1 anity of the phenyl
group was underestimated for both 5j and 2-PhMLT. It
is worth noting that the coecients of eqs (8) and (5) are
very similar, thus suggesting that electron-withdrawing
substituents with an optimal lipophilicity (ꢀ value
around 1) should lead to better anity, and that this
QSAR is common to the whole series of indole derivatives.
Observing the residuals, reported in Table 3, it can also be
inferred that this simple regression model under-
estimates the anity contribution of planar substituents
having ꢀ-electron clouds (i.e., phenyl group, and sub-
stituents with a carbonyl group directly attached to the
indole ring), giving positive residuals for them. This
feature could not be accounted for by the electronic
descriptors employed, and a similar result occurred with
sp instead of sm.
ꢀ4†
pRA2 ˆ 0:95ꢀÆ0:09† ꢀ À 0:37ꢀÆ0:07† ꢀ2 ‡ 1:41ꢀÆ0:56†
ꢁm ‡ 0:11ꢀÆ0:22†
n ˆ 11 R2 ˆ 0:94 s ˆ 0:35 F ˆ 36:6 Q2 ˆ 0:83
SDEP ˆ 0:47
ꢀ5†
Although the introduction of the electronic variable
improved the predictive power, compound 5e remained
a severe outlier in these models, and it can be suspected
to a€ect the analysis for its excessive leverage. Its
exclusion from the test set caused a high correlation
between the two biological variables (see above) leading
to very similar QSARs for pRA1 and pRA2; for this
reason, only the equations involving pRA2 are further
discussed.
In spite of the exclusion of the most lipophilic sub-
stituent (5e), the parabolic dependence of anity on
lipophilicity was still observed, as illustrated by eq (6),
indicating that this trend did not depend only on the
loss of anity observed for 5e. In eq (6) the Q2 value is
closer to the R2 one, indicating that the ability of the
model to explain pRA2 variation is similar in prediction
and in calculus, and that no compound is in¯uencing
the model with an excessive leverage.
Partial least squares projections to latent variables (PLS
analysis)
PLS has the advantage, over MRA, that the variables in
the Y matrix (pRA) are correlated with latent variables
(LV), calculated from the X matrix, which are mutually
orthogonal; the LVs can therefore be regarded as
describing statistically independent e€ects of the physico-
chemical properties on the biological ones.
pRA2 ˆ 0:98ꢀÆ0:12† ꢀ À 0:32ꢀÆ0:12† ꢀ2 ‡ 0:40ꢀÆ0:21†
In the PLS analysis, pRA1 and pRA2 were used as the
Y matrix, and the eight variables reported in Table 4,
plus a squared expansion of ꢀ (ꢀ2), as the X matrix.
When the 11 compounds 5a±k were considered, no simple
PLS model with appreciable predictive power (at cross-
validation) could be achieved. Including MLT (1) and
the MLT derivatives 8±11, the increased number of
compounds (16) led to an improvement in the statistics
of the PLS model. R2, Q2, and the PLS weights (WX) of
n ˆ 10 R2 ˆ 0:91 s ˆ 0:44 F ˆ 34:2 Q2 ˆ 0:84
SDEP ˆ 0:48 Optimal ꢀ ˆ 1:53
ꢀ6†
A comparison of the e€ects of 2-substituents on the
melatonin sca€old can be attempted, considering the
logarithm of relative anity data (pRA) of MLT deri-
vatives 8±11 (Table 3). The benzyl derivative of MLT

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1051
the PLS model obtained are reported in Table 5 for the
®rst four latent variables (LV). The PLS weights allow
the calculation of the coordinates on the latent variables
(scores, t) from the X matrix: T=XÂWX; the WX values
can therefore be used to explain the chemical meaning
of the latent variables, considering that the calculated Y
values are obtained as Y=TÂC, where C are the coe-
cients relating each Y variable to each LV.
Table 6. Distances (A) between the phenyl ring in position 2 and
other key features for minimum-energy conformations of 5e and 11
which can be superposed
5e
11
Superposition 1
Superposition 2
Superposition 3
Amide Oa
7.39
7.92
6.02
2.81
7.31
7.46
6.26
2.21
Amide Hb
Indole benzenec
Height above the planed
The ®rst two LVs describe the most relevant e€ects of
physicochemical properties on anity, as they improve
the predictive power of the model (Q2) and explain 83%
of pRA variation, with an R2 in the X space of 0.607
(i.e., they use 60.7% of the whole variance of the X
matrix). The ®rst LV is dominated by the parabolic
e€ect of lipophilicity, as argued by the WX values of ꢀ
and ꢀ2; the second one, which has a stronger in¯uence
on pRA2 than on pRA1, is mainly due to a linear e€ect
of lipophilicity, and can be explained by the fact that
more lipophilic substituents are more tolerated by MT2
receptor. The next two LVs should be inspected looking
for additional e€ects, even if they could not signi®cantly
improve the predictive power, and each of them could
explain only an additional 3% of Y variation. The third
LV, which is still more related to pRA2 than to pRA1,
gives the highest weight to the Verloop minimum-width
parameter (B1); the negative sign means that better a-
nity can be obtained for substituents with small B1, that is
the planar ones. The fourth LV contains electronic infor-
mation, meaning that electron-withdrawing substituents
have a positive in¯uence both on pRA1 and pRA2.
Amide Oa
Amide Hb
Indole benzenec
Height above the planed
3.81
4.00
6.45
3.29
3.59
4.60
6.34
2.46
Amide Oa
Amide Hb
Indole benzenec
Height above the planed
9.48
7.88
5.53
À3.04
9.06
7.32
6.30
À2.60
aDistance between the centroid of the phenyl ring in position 2 and the
amide oxygen.
bDistance between the centroid of the phenyl ring in position 2 and the
amide hydrogen.
cDistance between the centroid of the phenyl ring in position 2 and the
centroid of the benzene fragment of the indole nucleus.
dDistance between the centroid of the phenyl ring in position 2 and the
plane of the indole ring; a positive value means that the phenyl ring
points to the same direction as the acylaminoethyl side chain.
Molecular modeling
The conformational analysis performed on the phe-
nethyl group of 5e and the benzyl group of 11 showed
that, superposing the putative pharmacophore elements
of MLT receptor ligands, as de®ned in ref 23, the C-2
ending phenyl ring of the two compounds can occupy the
same position, lying out of the plane of the indole ring.
Table 6 reports the distances among some key features
for the couples of conformers having di€erences in
those distances lower than 1 A, corresponding to di€er-
ent superposition models. Those referred as super-
position 1 and 3 (Fig. 3), correspond to the arrangement
of the phenyl ring of the C-2 substituent above and
below the plane of the indole ring, respectively. These
two models also allow the superposition of the known
MT2 selective antagonist 4-phenyl-2-acetylamino-tetra-
lin (4-P-ADOT, not shown), which has the phenyl ring
attached to a tetrahedral carbon, and thus projected out
of the plane of the tetralin nucleus; the same was not pos-
sible for superposition 2, presenting the phenyl ring of the
2-substituent too close to the acylaminoethyl chain.
Figure 3. Superpositions 1 (left) and 3 (right) of 5e (green carbons)
and 11 (yellow carbons) as described in Table 6. The arrangement of
the phenyl rings out of the plane of the indole nucleus is evidenced.
Non-carbon atoms are color coded.
mt1 and MT2 anity, as its parabolic relationship
explains most of the anity variation, and additional
e€ects can hardly be observed with sucient statistical
signi®cance. However, the relatively high residuals of
the corresponding models (eqs (2) and (3)) indicate that
other structural features are playing a role in receptor
binding. In fact, from MRA and PLS models, indica-
tions arise that planar, electron-withdrawing sub-
stituents can establish additional interactions with the
binding pocket.
Discussion
It is apparent from the data presented in Table 1 that
2-substituted indole melatonergic ligands bind at h-mt1
or h-MT2 receptors with varying anity. Lipophilicity
of the 2-substituent is clearly the major determinant for
The QSARs observed for the substituents at the 2 posi-
tion of 1-acylaminoethylindoles seem to be followed
also by the corresponding MLT derivatives 8±11, at
least for the few data on cloned receptors available so

1052
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
far (see eqs (7) and (8)). Binding data on native tissues,
with heterogeneous populations of receptor subtypes,
have been reported for some 2-alkyl-MLT derivatives:
while in 2-methyl-MLT a small increase in anity had
been achieved, both iso-propyl and cyclohexyl sub-
stituents had led to a drop of anity.20 This behavior
can qualitatively be compared to the QSARs reported
above, suggesting that branched alkyl substituents could
be detrimental for receptor binding, in spite of positive
lipophilic contribution, due to their electronic features
and non-planar shape.
(CONH2, NHCONH2, CH2OH) had lower anity both
for mt1 and MT2 receptors. In general, 2-substitution
(5a±d, 5f±h, 5j±k) had a minimal e€ect on mt1/MT2
subtype selectivity. An interesting contrast between the
e€ects of the C-2 substitution on subtype selectivity is
found for 2-substitution with phenethyl, resulting in the
compound (5e) with the highest MT2/mt1 selectivity
ratio (ca. 50) among all of the ligands tested in this
study; the introduction of a C-2 phenethyl group also
causes a substantial decrease in intrinsic activity for the
h-MT2 receptor. These results suggest a di€erent shape
for the two binding pockets; the occupancy of the
MT2 putative cavity by an appropriately sized group
could induce a di€erent MT2 receptor conformation
responsible for the loss of intrinsic activity observed for
5e.
The descending branches of the parabolic relations,
illustrated in Figure 2, represent the main di€erence
between mt1 and MT2 SARs; as this is mainly due to the
phenethyl group of 5e, it can be attributed to higher
steric tolerance of the MT2 subtype. In fact, the e€ect of
the phenethyl substituent (5e), leading to a drop of a-
nity which is much higher on the mt1 than on the MT2
receptor, points out to a di€erent shape for the two
binding pockets. On the other hand, for 5e, and for the
other non-planar substituent of 5c, a higher depression
of MT2 intrinsic activity is observed on the cellular line
employed; this behavior recalls that of 5-methox-
yluzindole (11) and of 4-phenyl-2-acetyl-amino-tetralin
(4-P-ADOT) derivatives, which have been found to dis-
play good selectivity for the MT2 receptor subtype, with
low intrinsic activity.24
Experimental
General
2-[125I]Iodomelatonin (speci®c activity=2000 CimmolÀ1
)
and [35S]-GTPgS (speci®c activity=1070 mmolÀ1) were
purchased from Amersham (Bucks, UK). 2-Iodomela-
tonin was obtained from RBI (Natick, MA, USA). 2-
Bromomelatonin,19 2-phenylmelatonin,20 and the mela-
tonergic derivatives 5a,26 5i±k26 were synthesized as
described elsewhere. Melting points were determined on
a Buchi SMP-510 capillary melting point apparatus and
Molecular models of 5e and 11 indicated that the ending
phenyl ring of the two compounds can be superposed
out of the plane of the indole nucleus, either above or
below it (see Fig. 3); this is also possible for 4-P-ADOT,
which can be considered the prototype of selective MT2
antagonists. In contrast, a 3D-QSAR model built for
non-selective melatonin receptor ligands16 had revealed
the presence of a favourite steric region near the 2-
position of MLT, in the same plane of the indole, which
can be occupied, for example, by the non-selective
ligands 2-phenyl-MLT and 5j. Therefore, the selectivity
observed for 5e and 11 could be due to the presence, in
the MT2 binding site, of an additional steric tolerance in
a region di€erent from the lipophilic pocket pointing
towards the position 2 of the indole nucleus. The fact
that the mt1 receptor lacks this additional pocket could
explain the anity drop for the phenylalkyl groups on
this subtype. The di€erent receptor conformation that
could be induced by substituent accommodation into
the MT2 binding site could be seen as a possible expla-
nation for the loss of intrinsic activity on this receptor
subtype.
1
are uncorrected. H NMR spectra were recorded on a
Bruker AC 200 spectrometer; chemical shifts (d scale)
are reported in parts per million (ppm) relative to the
central peak of the solvent. Coupling constants (J
values) are given in hertz (Hz). EI±MS spectra (70 eV)
were taken on a Fisons Trio 1000. Only molecular ions
(M+) and base peaks are given. Infrared spectra were
obtained on a Bruker FT-48 spectrometer; absorbances
are reported in n (cmÀ1). Elemental analyses for C, H
and N were performed on a Carlo Erba analyzer, and
were within Æ0.4% of theoretical values.
6-Methoxy-2-methylthio-1H-indole. SO2Cl2 (0.8 mL,
10 mmol) was added to a cooled (À25 ꢀC) solution of
dimethyl disulphide (1.0 mL, 11 mmol) in CH2Cl2
(25 mL); the cooling was removed and the mixture was
stirred and let warm to room temperature. 2.6 mLof this
solution were added to a stirred solution of 6-methoxy-
indole32 (0.294 g, 2 mmol) in CH2Cl2 (100 mL), and the
mixture was stirred at room temperature for 5 min. The
reaction mixture was washed with a 2 N Na2CO3 solution
and, after drying over Na2SO4, the solvent was
removed. The crude residue was ®ltered on silica gel
(cyclohexane/EtOAC, 8:2) to give 0.29 g of a mixture of
four products (starting material, two isomers of the
thiomethyl derivative and a dithiomethyl derivative)
detected by GC±MS (EI). This mixture was added to
polyphosphoric acid (PPA) (20 g), preheated to 100 ꢀC
under mechanical stirring and a nitrogen atmosphere,
and the mixture was stirred at 100 ꢀC for 1 h. After
cooling, water was added, the mixture was left under
stirring for 30 min, and then extracted three times with
ethyl acetate. After drying over Na2SO4 the solvent was
Conclusions
The in¯uence of substitutions at the C-2 indole position
of indole melatonergic ligands on binding at both h-mt1
and h-MT2 receptors has been examined through
QSAR studies (PLS and MRA methods). Substitution
with electron-withdrawing substituents having an opti-
mal lipophilicity (ꢀ value around 1), such as I, Br,
COOMe, CF3, resulted in a general increase in binding
anity. Compounds with hydrophilic substituents

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1053
removed to give the crude 2-methylthio-6-methoxy-
indole that was puri®ed by column ¯ash chromato-
graphy (silica gel; cyclohexane±ethyl acetate, 8:2 as
eluent) followed by crystallization from dichloro-
methane-cyclohexane (0.08 g, 21%). Mp 57 ꢀC; 1H
NMR (CDCl3): d 2.47 (s, 3H), 3.85 (s, 3H), 6.55 (d, 1H,
J=1.4 Hz), 6.80 (m, 2H), 7.43 (d, 1H, J=8.3 Hz), 8.03
J=8.9 Hz), 9.76 (s, 1H); EIMS m/z 264 (M+, 100); IR
(cmÀ1, nujol): 2259, 1621.
(2-Formyl-6-methoxy-1H-indol-1-yl)acetonitrile (3d). White
crystalline solid (0.38 g, 18%); mp 152 ꢀC (EtOAc/hexane);
1H NMR (CDCl3): d 3.94 (s, 3H), 5.64 (s, 2H), 6.80 (d,
1H, J=1.9 Hz), 6.93 (dd, 1H, J=1.9, 8.8 Hz), 7.29 (s,
1H), 7.65 (d, 1H, J=8.8 Hz), 9.76 (s, 1H); EIMS m/z
214 (M+, 100); IR (cmÀ1, nujol): 1658.
(br s, 1H); EIMS m/z 193 (M+), 178 (100); IR (cmÀ1
,
nujol): 3395, 1621.
2-Methanesulfonyl-6-methoxy-1H-indole (2c). A solution
of magnesium monoperoxyphtalate hexahydrate (MMPP
80%) (0.492 g, 0.99 mmol) and of benzyltriethylammonium
chloride (0.015 g, 0.1 equiv) in water (5 mL) was added to a
0ꢀC cooled solution of 2-methylthio-6-methoxyindole
(0.128 g, 0.66 mmol) in dichloromethane (10 mL). After
stirring for 5 min, the layers were separated and the
aqueous layer was extracted with dichloromethane. The
combined organic layers were washed with a saturated
NaHCO3 solution and then with water. The organic
phase was dried (Na2SO4), ®ltered, concentrated in
vacuo and puri®ed by ¯ash chromatography (silica gel;
cyclohexane/ethyl acetate, 8:2) followed by crystal-
lization from ethyl acetate/cyclohexane, to yield 0.12 g
(6-Methoxy-2-styryl-1H-indol-1-yl)acetonitrile (4). Benzyl-
triphenylphosphonium chloride (0.23 g, 0.59 mmol) was
added to a stirred solution of 3d (0.09 g, 0.42 mmol) and
DBU (0.12 mL) in toluene (9 mL) and the mixture was
re¯uxed for 4 h. The mixture was cooled to room tem-
perature, dichloromethane was added, and the organic
phase was washed with water, dried (Na2SO4), and
concentrated under reduced pressure to give the crude
ole®n 4 as a mixture of E and Z isomers. Puri®cation by
¯ash chromatography (silica gel; dichloromethane as
eluent) gave 0.072 g (59.5% yield) of the stereoisomeric
ole®ns which was used in the next step without further
puri®cation. EIMS m/z 288 (M+, 100).
(79%) of the desired product 2c. Mp 137±138 ꢀC; H
General procedure for the synthesis of 2-substituted-N-
[2-(6-methoxy-1H-indol-1-yl)ethyl]propionamido deriva-
tives (5a±c, e). A solution of the suitable indol-1-yl-aceto-
nitrile 3a±c, 4 (1 mmol) in THF (5 mL) and propionic
anhydride (2 mL) was hydrogenated for 6 h at 50ꢀC under
pressure (4 atm) using Raney nickel as catalyst. Following
catalyst removal (by ®ltration on Celite1), the evapora-
tion of the solvent gave a residue which was partitioned
between ethyl acetate and 2 N NaOH. The organic layer
was washed with brine, dried (Na2SO4) and evaporated
under reduced pressure to give crude title compounds (5a±
c, e) which were puri®ed by ¯ash-chromatography (silica
gel; cyclohexane/ethyl acetate, 3:7) and crystallization.
1
NMR (CDCl3): d 3.20 (s, 3H), 3.67 (s, 3H), 6.89 (m,
2H), 7.15 (d, 1H, J=1.5 Hz), 7.58 (d, 1H, J=9.4 Hz),
8.98 (br s, 1H); EIMS m/z 225 (M+, 100); IR (cmÀ1
nujol): 3296, 1621.
,
General procedure for the synthesis of (1H-indol-1-yl)
acetonitrile derivatives (3a±d). A solution of the appro-
priate indole 2a±d (10 mmol) in dry DMF (10 mL) was
added dropwise to a stirred ice-cooled suspension of
sodium hydride (0.42 g of an 80% dispersion in mineral
oil, 14mmol) in dry DMF (30 mL) under a N2 atmo-
sphere. After the addition, the mixture was stirred at 0ꢀC
for 30min, then chloroacetonitrile (1.06 g, 14mmol) was
added dropwise and the resulting mixture was stirred at
room temperature for 16 h, then poured into ice-water
(250 g) and extracted with ethyl acetate. The organic
phase was washed with brine, dried over sodium sulphate
and concentrated under reduced pressure to give a residue
which was puri®ed by ¯ash column chromatography
(silica gel; cyclohexane/ethyl acetate, 7:3 as eluent for
3a±b and dichloromethane/ethyl acetate, 98:2 for 3c±d)
and crystallization.
1-(2-Propionylamino-ethyl)-6-methoxy-1H-indol-1-yl-2-
carboxylic acid methyl ester (5a). This was prepared as
described in literature.26
N-[2-(6-Methoxy-2-tri¯uoromethyl-1H-indol-1-yl)ethyl]-
propionamide (5b). White crystalline solid (0.23 g, 73%);
mp 100±101 ꢀC (dichloromethane/hexane); 1H NMR
(CDCl3): d 1.12 (t, 3H), 2.16 (q, 2H), 3.64 (q, 2H), 3.91
(s, 3H), 4.36 (t, 2H), 5.57 (br t, 1H), 6.85 (dd, 1H,
J=2.0, 8.7 Hz), 6.90 (s, 1H), 7.03 (d, 1H, J=2.0 Hz),
7.52 (d, 1H, J=8.7 Hz); EIMS m/z 314 (M+), 241 (100);
1-Cyanomethyl-1H-indole-2-carboxylic acid methyl ester
(3a). This was prepared as described in the literature.26
IR (cmÀ1
, nujol): 3323, 1647. Anal. calcd for
C15H17F3N2O2 (314.31): C, 57.32; H, 5.45; N, 8.91.
Found: C, 57.70; H, 5.48; N, 8.94.
(6-Methoxy-2-tri¯uoromethyl-1H-indol-1-yl)acetonitrile
(3b). White crystalline solid (3.92 g, 86%); mp 68±69 ꢀC
(Et2O/hexane); H NMR (CDCl3): d 3.93 (s, 3H), 5.06
(s, 2H), 6.83 (d, 1H, J=2.0 Hz), 6.94 (dd, 1H, J=2.0,
8.9 Hz), 6.99 (m, 1H), 7.58 (d, 1H, J=8.9 Hz); EIMS m/z
254 (M+), 214 (100); IR (cmÀ1, nujol): 1627.
1
N-[2-(2-Methanesulfonyl-6-methoxy-1H-indol-1-yl)ethyl]
propionamide (5c). White crystalline solid (0.22 g, 68%);
mp 135 ꢀC (ethyl acetate/cyclohexane); 1H NMR
(CDCl3): d 1.07 (t, 3H), 2.12 (q, 2H), 3.19 (s, 3H), 3.68
(m, 2H), 3.91 (s, 3H), 4.54 (t, 2H), 6.08 (br t, 1H), 6.87
(dd, 1H, J=2.1, 8.8 Hz), 6.99 (d, 1H, J=2.1 Hz), 7.21
(s, 1H), 7.55 (d, 1H, J=8.8 Hz); EIMS m/z 324 (M+),
251 (100); IR (cmÀ1, nujol): 3367, 1639. Anal. calcd for
C15H20N2SO4 (324.39): C, 55.54; H, 6.21; N, 8.64.
Found: C, 55.94; H, 6.29; N, 8.62.
(2-Methanesulfonyl-6-methoxy-1H-indol-1-yl)acetonitrile
(3c). White crystalline solid (1.82 g, 69%); mp 164 ꢀC
(EtOAc/cyclohexane); 1H NMR (CDCl3): d 3.29 (s,
3H), 3.95 (s, 3H), 5.48 (s, 2H), 6.80 (d, 1H, J=2.1 Hz),
6.97 (dd, 1H, J=2.1, 8.9 Hz), 7.35 (s, 1H), 7.63 (d, 1H,

1054
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
N-[2-(6-Methoxy-2-phenethyl-1H-indol-1-yl)ethyl]propion-
amide (5e). Beige crystalline solid (0.29 g, 82%); mp
1H NMR (CDCl3): d 1.07 (t, 3H), 2.12 (q, 2H), 3.67 (m,
2H), 3.91 (s, 3H), 4.66 (t, 2H), 5.84 (br t, 1H), 6.83 (dd,
1H, J=2.4, 8.8 Hz), 6.85 (d, 1H, J=2.4 Hz), 7.35 (s,
1H), 7.53 (d, 1H, J=8.8 Hz); EIMS m/z 315 (M+), 32
(100); IR (cmÀ1, nujol): 3311, 2151, 1663.
ꢀ
1
125 C (EtOAc/cyclohexane); H NMR (CDCl3): d 1.06
(t, 3H), 2.09 (q, 2H), 3.02 (m, 4H), 3.51 (m, 2H), 3.86 (s,
3H), 4.16 (t, 2H), 5.45 (br t, 1H), 6.28 (s, 1H), 6.75 (dd,
1H, J=2.2, 8.4 Hz), 6.83 (d, 1H, J=2.2 Hz), 7.22±7.32
(m, 5H), 7.42 (d, 1H, J=8.4 Hz); EIMS m/z 350 (M+),
259 (100); IR (cmÀ1, nujol): 3309, 1642. Anal. calcd for
C22H26N2O2 (350.46): C, 75.40; H, 7.48; N, 7.99.
Found: C, 75.15; H, 7.56; N, 8.04.
N-[2-(6-Methoxy-2-ureido-1H-indol-1-yl)ethyl]propion-
amide (5g). Ammonium acetate (0.15 g, 1.9 mmol) was
added to a solution of 7 (0.21 g, 0.66 mmol) in toluene
(15 mL) and the mixture was heated at re¯ux for 5 min
under nitrogen. The mixture was cooled to room tem-
perature, the solid ®ltered, washed sequentially with
toluene, water and diethyl ether, and dried in vacuo to
give 0.11 g (54%) of the ureido product 5g as a white
solid. An analytical sample was obtained by crystal-
lization from ethyl acetate/methanol. Mp 191 ꢀC; 1H
NMR (DMSO-d6): d 0.90 (t, 3H), 1.98 (q, 2H), 3.30 (m,
2H), 3.72 (s, 3H), 4.03 (t, 2H), 6.03 (br s, 2H,
exchangeable with D2O), 6.14 (s, 1H), 6.62 (dd, 1H,
J=2.2, 8.4 Hz), 6.97 (d, 1H, J=2.2 Hz), 7.27 (d, 1H,
J=8.4 Hz), 7.93 (br t, 1H, exchangeable with D2O), 8.17
(br s, 1H, exchangeable with D2O); EIMS m/z 304
(M+), 188 (100); IR (cmÀ1, nujol): 3453, 3288, 1668,
1-(2-Propionylamino-ethyl)-6-methoxy-1H-indol-1-yl-2-
carboxylic acid (6). A solution of 5a26 (3.04 g, 10 mmol)
in THF (20 mL), MeOH (24 mL) and 3 N KOH (10 mL)
was stirred at room temperature for 16 h. After cooling
to 0 ꢀC, the solution was acidi®ed with 6 N HCl and the
white solid that precipitated was collected, washed with
water, and dried at 50 ꢀC under vacuum to give 2.64 g
(91%) of the acid 6 which was directly used without
further puri®cation. Mp 204 ꢀC; H NMR (DMSO-d6):
1
d 0.87 (t, 3H), 1.92 (q, 2H), 3.36 (q, 2H), 3.81 (s, 3H),
4.54 (t, 2H), 6.75 (dd, 1H, J=2.1, 8.8 Hz), 7.07 (d, 1H,
J=2.1 Hz), 7.15 (s, 1H), 7.51 (d, 1H, J=8.8 Hz), 7.90
(br t, 1H), 12.70 (br s, 1H); EIMS m/z 290 (M+), 217
(100); IR (cmÀ1, nujol): 3260, 1664.
.
1636. Anal. calcd for C15H20N4O3 0.6H2O (315.16): C,
57.17; H, 6.78; N, 17.78. Found: C, 56.94; H, 6.39; N,
17.46.
1-(2-Propionylamino-ethyl)-6-methoxy-1H-indol-1-yl-2-
carboxylic acid amide (5f). Thionyl chloride (0.08 mL)
was added to a solution of the acid 6 (0.18 g, 0.62 mmol)
in dry THF (10 mL) and the mixture was stirred at
room temperature for 15 min under nitrogen, then
heated at 50 ꢀC for 4 h, and ®nally allowed to stand at
room temperature for 2 h. The solvent and excess thionyl
chloride were removed under reduced pressure and the
residue was dissolved in dry THF (40 mL). A saturated
solution of ammonia in dichloromethane was poured
into the ice-cooled solution of the acid chloride and the
mixture was stirred at room temperature for 2 h. Petroleum
ether was added to the reaction mixture and the yellow
solid that precipitated upon cooling to 0 ꢀ C was collected,
washed with water, and dried (0.15 g; yield 85%). An
analytical sample of 5f was prepared by recrystallization
N-[2-(2-Hydroxymethyl-6-methoxy-1H-indol-1-yl) ethyl]-
propionamide (5h). A solution 5a26 (0.73 g, 2.4 mmol) in
dry THF (8 mL) was added dropwise to a stirred sus-
pension of LiAlH4 (0.11 g) in dry THF (5 mL) at 0 ꢀC
under nitrogen. The mixture was stirred for additional
1 h at 10 ꢀC, and the unreacted LiAlH4 was destroyed by
careful addition of water. The resulting mixture was ®l-
tered through a Celite1 pad and the ®ltrate was dried
over Na2SO4. The dried solution was concentrated
under reduced pressure to give a crude oil. Puri®cation
of the oil by ®ltration on silica gel (ethyl acetate as elu-
ent) gave 0.60 g (91% yield) of 5h as orange viscous oil.
1H NMR (CDCl3): d 1.05 (t, 3H), 2.12 (q, 2H), 3.68 (q,
2H), 3.88 (s, 3H), 4.31 (t, 2H), 4.79 (s, 2H), 6.15 (br t,
1H), 6.42 (d, 1H, J=0.6 Hz), 6.78 (dd, 1H, J=2.2,
8.6 Hz), 6.84 (d, 1H, J=2.2 Hz), 7.46 (dd, 1H, J=8.6,
from methanol. Mp 206±207 ꢀC; H NMR (DMSO-d6):
1
d 0.83 (t, 3H), 1.96 (q, 2H), 3.69 (q, 2H), 3.78 (s, 3H),
4.50 (t, 2H), 6.71 (dd, 1H, J=2.0, 8.7 Hz), 6.99 (s, 1H),
7.07 (s, 1H), 7.25, 7.87 (2 br s, 2H exchangeable), 7.46
(d, 1H, J=8.7 Hz), 7.99 (br t, 1H); EIMS m/z 289 (M+),
174 (100); IR (cmÀ1, nujol): 3290, 3149, 1679. Anal.
calcd for C15H19N3O3 (289.33): C, 62.27; H, 6.62; N,
14.52. Found: C, 62.02; H, 6.60; N, 14.35.
0.6 Hz); EIMS m/z 276 (M+), 160 (100); IR (cmÀ1
,
nujol): 3327, 2925, 1647. Anal. calcd for C15H20N2O3:
(276.34) C, 65.20; H, 7.30; N, 10.14. Found: C, 65.34;
H, 6.93; N, 9.78.
N-[2-(2-Formyl-6-methoxy-1H-indol-1-yl)ethyl]propion-
amide (5d). Activated manganese dioxide (0.6 g) was
added to a solution of 5h (0.414 g, 1.5 mmol) in dry di-
chloromethane (15 mL). The mixture was stirred for
10 h at room temperature and then ®ltered. The ®lter
cake was washed with hot acetone (4Â8 mL), and the
combined ®ltrates were concentrated under reduced
pressure to yield 0.42 g of a crude orange oil. Puri®ca-
tion of the crude product by ¯ash column chromato-
graphy (silica gel; cyclohexane/ethyl acetate, 3:7)
a€orded 0.1 g (24%) of 5d as white solid, and 0.29 g
(70%) of the unreacted starting material 5h. Mp 120±
1-(2-Propionylamino-ethyl)-6-methoxy-1H-indol-1-yl-2-
carbonyl azide (7). Diphenyl phosphorazidate (0.18 mL,
0.83 mmol) was added to a solution of 6 (0.2 g,
0.69 mmol) in benzene (5 mL) and Et3N (0.14 mL) and
the reaction mixture was stirred for 24 h at room tem-
perature under nitrogen. The resulting suspension was
dissolved in ethyl acetate, washed sequentially with
water and a saturated NaHCO3 aqueous solution, and
dried (Na2SO4). The solvent was removed under
reduced pressure to give the crude product 7 (0.21 g,
97%) that was used in the next step without further
puri®cation.
121 ꢀC (dichloromethane/hexane); H NMR (CDCl3): d
1
1.07 (t, 3H), 2.11 (q, 2H), 3.67 (q, 2H), 3.91 (s, 3H), 4.66
(t, 2H), 5.98 (br t, 1H), 6.82±7.61 (m, 4H), 9.72 (s, 1H);

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1055
EIMS m/z 274 (M+), 201 (100); IR (cmÀ1, nujol): 1659.
Anal. calcd for C15H18N2O3: (274.32) C, 65.68; H, 6.61;
N, 10.21. Found: C, 65.34; H, 6.23; N, 10.10.
compounds and was taken as 100%. The maximal G-
protein activation was measured in each experiment by
using MLT (100 nM). Compounds were added to a
concentration equivalent to 100 nM of MLT, and the
percent stimulation above basal was determined. The
equivalent concentration was estimated on the basis of
the ratio of the anity of the test compound over that
of MLT. It was assumed that at the equivalent con-
centration the test compound occupies the same number
of receptors as MLT 100 nM. All measurements were
performed in triplicate. The SEM values were below
15% of the mean.
Pharmacological evaluation
Membrane preparation. The preparation of membranes
was described elsewhere.28,37 In short, NIH3T3 cells
stably expressing the cloned human mt1 or MT2 receptor
subtypes were grown to con¯uence. On the day of assay
the cells were detached from the ¯asks with ethylenedi-
aminetetracetic acid (EDTA) (4 mM)/Tris±HCl (50 mM),
pH 7.4, room temperature, and collected by centrifugation
at 1000Âg for 10 min at 4 ꢀC. The cells were suspended
in EDTA (2 mM)/Tris±HCl (50 mM), homogenized in
10±15 volumes of ice-cold EDTA (2 mM)/Tris±HCl
(50 mM) with an ultra-Turrax apparatus and centri-
fuged at 50,000Âg at 4 ꢀC for 25 min. The ®nal pellet
was then resuspended in ice-cold Tris±HCl (50 mM)
assay bu€er.
Melatonin caused a 370% increase over basal (100%)
[35S]GTPgS binding in NIH3T3 expressing the mt1
receptor and 250% increase in NIH3T3 expressing the
MT2 receptor. The relative intrinsic activity values were
obtained by dividing the maximal G-protein activation
of a test compound by that of MLT. Thus the relative
intrinsic activity of MLT is always 1, for partial agonists
between 0 and 1 and for antagonists 0.
Membrane protein level was determined according to a
previously reported method.38
Experimental design for substituent selection
2-[125I]Iodomelatonin binding assays. Anities of com-
pounds were determined in competition binding assays,
by the displacement of 2-[125I]iodomelatonin from mt1
and MT2 receptors expressed in NIH3T3 cells.
For a set of 59 substituents, a large number of structural
descriptors has been reported,41 and a smaller set of
principal properties, named disjoint principal properties
(DPP), has been derived from separate groups of
descriptors.29 These DPPs are classi®ed into three
groups, containing information on the steric (s), elec-
tronic (e), lipophilic (l) and hydrogen-bonding proper-
ties of the substituents; due to the method of
calculation, they are not mutually orthogonal, but their
chemical meaning is clear, and their ability to condense
information from several descriptors makes them useful
for compound selection in experimental design.
2-[125I]Iodomelatonin (100 pM) and a range of con-
centrations of test compound were incubated with the
receptor preparation for 90 min at 37 ꢀC. The ®nal
membrane concentration was 5±10 mg protein per tube.
The binding conditions were described in detail else-
where.39 IC50 values were determined by non linear ®t-
ting strategies and pKi values were calculated from the
IC50 values using the Cheng±Pruso€ equation.40 In
saturation studies 2-[125I]iodomelatonin was added at
concentrations ranging from 10 to 1000 pM.
The substituents to be introduced in position 2 of the
indole ring were selected on the basis of the ®rst steric
(S_1), electronic (E_1) and lipophilic (L_1) DPPs,
through an iterative procedure employing the program
DESDOP.42 This program calculates an index, called
D-eciency, based on the determinant of the X0X
matrix, and related to both the volume occupied by the
substituents in the property space, and to inter-property
correlation.43 During the selection procedure, some
reaction-based and knowledge-based constraints were
taken in account. Some substituents which could pre-
sent synthetic diculty were in fact excluded from the
experimental domain, while some others, known to
grant good anity when introduced on the MLT struc-
ture (i.e., Ph, Br, COOMe),16 were forced into the
selection. As the e€ect of 2-alkyl substituents on MLT
binding anity had been previously investigated,20
and it was found to be detrimental for receptor a-
nity, the presence of alkyl groups was limited to the
phenethyl one. This group also represented the upper
limit of lipophilicity and steric hindrance for the pres-
ent study.
Determination of the intrinsic activity: [35S]GTPꢀS binding
assays. [35S]GTPgS Binding studies in NIH3T3 cells
expressing human cloned mt1 or MT2 receptors were
performed as previously described.25,28,37 The ®nal pellet,
obtained as described above (membrane preparation),
was resuspended in ice-cold Tris±HCl assay bu€er
(50 mM) to give a ®nal membrane concentration of 20±
30 mg/mL. The membranes (15±25 mg of protein) were
then incubated at 30 ꢀC for 30 min in the presence and
in the absence of melatonin analogues, in an assay bu€er
consisting of [35S]GTPgS (0.3±0.5 nM), GDP (50 mM),
NaCl (100 mM) and MgCl2 (3 mM). The ®nal incuba-
tion volume was 100 mL. The incubation was terminated
by the addition of ice-cold Tris±HCl bu€er, pH 7.4
(1 mL), rapid vacuum ®ltration through Whatman GF/
B glass-®ber ®lters followed by three (3 mL) washes with
ice-cold Tris±HCl bu€er, pH 7.4. Bound radioactivity was
determined by liquid scintillation spectrophotometry
after overnight extraction in 4 mLFilter-Count scintil-
lation ¯uid. Basal binding was assessed in the absence of
ligands, and non-speci®c binding was de®ned using
GTPgS (10 mM). Basal stimulation is the amount of
[35S]GTPgS speci®cally bound in the absence of
The correlation matrix for compounds 5a±k, reported in
Table 2, shows low correlation between couples of the
®rst three DPP.

1056
M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
QSAR
was applied, in order to reduce the number of con-
formations, and each of the conformers thus obtained
was energy-minimized. The six conformers obtained for
11, and the 16 for 5e, were submitted to the following
analysis.
QSARs were analyzed expressing the anity for each
receptor subtype as a contribution of the substituent,
relative to that of the unsubstituted compound
À
Á
KXi =KHi . The Àlog of relative anity was indicated as
pRA.
Therefore,
pRA(R1=X)=pKi(R1=X)Àp-
For each conformer the distances between the centroid
of the phenyl ring in position 2 and three key features of
the molecule were measured, that is (1) the carbonyl
oxygen of the amide group; (2) the hydrogen atom of
the amide group; (3) the centroid of the benzene frag-
ment of the indole ring. The three distances found for
each conformer of 11 were compared to those measured
for the conformers of 5e. The couples of conformers
having di€erences, in all the distances, lower than 1 A
were mutually superposed ®tting the putative pharma-
cophore elements, that is the methoxy oxygen, the ben-
zene fragment of the indole nucleus, the four atoms of
the amide group and the centroid of the phenyl ring of
the 2-substituent.
Ki(R1=H) both for compounds 5a±k and for MLT
derivatives 8±11. Relative anities for the mt1 and MT2
receptors are indicated as pRA1 and pRA2, respectively.
pRA values for compounds 5a±k, and for MLT derivatives
8±11, are reported in Table 3.
Structural descriptors used for regression analyses were
selected from the databank compiled by van de Water-
beemd et al., which can be downloaded from the web
site of the QSAR and Modeling Society (www.pharma.
ethz.ch/qsar/). Detailed description of the variables can
be found in the original paper.29 A limited set of sub-
stituent constants, reported in Table 4, was employed,
to avoid an excessive risk of chance correlation.
Multiple regression analysis (MRA) was performed by
Excel,44 using an in-house macro routine to provide all the
possible combinations of a certain number of variables.
Partial least squares (PLS) models were calculated, after
scaling of the variables to unit variance, with the pro-
gram Simca 6.0.45 The lipophilicity variable (ꢀ) was
squared, in order to include parabolic e€ects. The
descriptive power of the models are estimated by R2Y,
expressing the fraction of the variance of the Y matrix
explained by the latent variables, and by R2X, expressing
the same for the X matrix. Cross validation excluding one
compound at a time (leave-one-out method) was
employed to calculate Q2 values (Q2=1ÀPRESS/SSy,
where PRESS is the sum of squares of the residuals in
prediction, i.e., for the compounds excluded from the
training set, and SSy is the sum of squares of the devia-
tions from the mean value of y), as an estimate of the
predictive power of the models. The standard deviation of
the errors in prediction (SDEP) was calculated as
(PRESS/N)_, where N is the total number of compounds.
Acknowledgements
This work was supported by grants from Ministero
dell'Universita e della Ricerca Scienti®ca e Tecnologica
(MURST) and University of Urbino. We thank Mr.
Francesco Palazzi for technical assistance.
References
1. Reiter, R. J. Endocr. Rev. 1991, 12, 151.
2. Bubenik, G. A.; Blask, D. E.; Brown, G. M.; Maestroni,
G. J.; Pang, S. F.; Reiter, R. J.; Viswanathan, M.; Zisapel, N.
Biol. Signals Recept. 1998, 7, 195.
3. Zhdanova, I. V.; Wurtman, R. J.; Lynch, H. J.; Ives, J. R.;
Dollins, A. B.; Morabito, C.; Matheson, J. K.; Schomer, D. L.
Clin. Pharmacol. Ther. 1995, 57, 552.
4. Arendt, J.; Deacon, S. Chronobiol. Int. 1997, 14, 185.
5. Arendt, J.; Skene, D. J.; Middleton, B.; Lockley, S. W.;
Deacon, S. J. Biol. Rhythms 1997, 12, 604.
6. Van Reeth, O. Horm. Res. 1998, 49, 158.
7. Oldani, A.; Ferini-Strambi, L.; Zucconi, M.; Stankov, B.;
Fraschini, F.; Smirne, S. NeuroReport 1994, 6, 132.
8. Arangino, S.; Cagnacci, A.; Angiolucci, M.; Vacca, A. M.;
Longu, G.; Volpe, A.; Melis, G. B. Am. J. Cardiol. 1999, 83,
1417.
9. Ebisawa, T.; Karne, S.; Lerner, M. R.; Reppert, S. M.
Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 6133.
10. Reppert, S. M.; Weaver, D. R.; Ebisawa, T. Neuron 1994,
13, 1177.
Conformational analysis and conformer superposition
Molecular modeling studies were performed with the
Sybyl 6.6 software46 running on a Silicon Graphics O2
workstation. Three-dimensional models of the mole-
cules were energy-minimized using the standard Tripos
force ®eld,47 ignoring the electrostatic contribution, to
.
an energy gradient of 0.01 Kcal/mol A, with the Powell
method.
11. Reppert, S. M.; Godson, C.; Mahle, C. D.; Weaver, D. R.;
Slaugenhaupt, S. A.; Gusella, J. F. Proc. Natl. Acad. Sci.
U.S.A. 1995, 92, 8734.
12. Mazzucchelli, C.; Pannacci, M.; Nonno, R.; Lucini, V.;
Fraschini, F.; Stankov, B. M. Mol. Brain Res. 1996, 39, 117.
13. Dubocovich, M. L.; Cardinali, D. P.; Hagan, R. M.;
Krause, D. N.; Sudgen, D.; Vanhoutte, P. M.; Yocca, F. D. In
The IUPHAR Compendium of Receptor Characterization and
Classi®cation; Girdlestone D., Ed.; IUPHAR Media: London,
1998, pp 187±193.
Â
14. Mathe-Allainmat, M.; Andriex, J.; Langlois, M. Exp.
Opin. Ther. Patents 1997, 7, 1447.
15. Mor, M.; Plazzi, P. V.; Spadoni, G.; Tarzia, G. Curr.
Med. Chem. 1999, 6, 501.
Compounds 11 and 5e were submitted to a systematic
conformational search (Systematic Search routine of
Sybyl) to generate the conformations having di€erent
orientations of the side chains in position 2. The meth-
oxy group and the acylaminoethyl side chain were kept
in a conformation corresponding to the pharmacophore
model previously described (model B in ref 23), while
the two single bonds in the benzyl group of 11 and the
three single bonds in the phenethyl group of 5e were
rotated by steps of 30ꢀ. An energy ®lter of 100 Kcal/mol

M. Mor et al. / Bioorg. Med. Chem. 9 (2001) 1045±1057
1057
16. Mor, M.; Rivara, S.; Silva, C.; Bordi, F.; Plazzi, P. V.;
Spadoni, G.; Diamantini, G.; Balsamini, C.; Tarzia, G.; Fra-
schini, F.; Lucini, V.; Nonno, R.; Stankov, B. M. J. Med.
Chem. 1998, 41, 3831.
17. Spadoni, G.; Balsamini, C.; Bedini, A.; Carey, A.; Dia-
mantini, G.; Di Giacomo, B.; Tontini, A.; Tarzia, G.; Nonno,
R.; Lucini, V.; Pannacci, M.; Stankov, B. M.; Fraschini, F.
Med. Chem. Res. 1998, 8, 487.
29. van de Waterbeemd, H.; Costantino, G.; Clementi, S.;
Cruciani, G.; Valigi, R. Disjoint Principal Properties of
Organic Substituents. In Chemometric Methods in Molecular
Design; Mannhold, R., Krogsgaard-Larsen, P., Timmerman,
H., Eds.; VCH: Weinheim/New York, 1995; pp 103±111.
30. Mager, P. P. Med. Res. Rev. 1997, 17, 453.
31. Box, G. E. P.; Hunter, W. G., Hunter, J. S. Statistics for
Experimenters; Wiley: New York, 1978; pp 306±373.
32. Allen, M. S.; Hamaker, L. K.; La Loggia, A. J.; Cook,
J. M. Synth. Commun. 1992, 22, 2077.
18. Takaki, K. S.; Mahle, C. D.; Watson, A. J. Curr. Pharm.
Des. 1997, 3, 429.
19. Duranti, E.; Stankov, B.; Spadoni, G.; Duranti, A.;
Lucini, V.; Capsoni, S.; Biella, G.; Fraschini, F. Life Sci. 1992,
51, 479.
20. Spadoni, G.; Stankov, B.; Duranti, A.; Biella, G.;
Lucini, V.; Salvatori, A.; Fraschini, F. J. Med. Chem. 1993, 36,
4069.
21. Garratt, P. J.; Jones, R.; Rowe, S. J.; Sugden, D. Bioorg.
Med. Chem. Lett. 1994, 4, 1555.
22. Garratt, P. J.; Vonho€, S.; Rowe, S. J.; Sugden, D.
Bioorg. Med. Chem. Lett. 1994, 4, 1559.
33. Miyashita, K.; Kondoh, K.; Tsuchiya, K.; Miyabe, H.;
Imanishi, T. J. Chem. Soc. Perkin Trans. 11 1996, 1261.
34. Dillard, R. D.; Bach, N. J.; Draheim, S. E.; Berry, D. R.;
Carlson, D. G.; Chirgadze, N. Y.; Clawson, D. K.; Hartley, L. W.;
Johnson, L. M.; Jones, N. D.; McKinney, E. R.; Mihelic, E. D.;
Olkowski, J. L.; Schevitz, R. W.; Smith, A. C.; Snyder, D. W.;
Sommers, C. D.; Wery, J.-P. J. Med. Chem. 1996, 39, 5119.
35. Wittig, G.; Schollkopf, U. Chem. Ber. 1954, 87, 1318.
36. Noguchi, N.; Kuroda, T.; Hatanaka, M.; Ishimaru, T.
Bull. Chem. Soc. Jpn. 1982, 55, 633.
23. Spadoni, G.; Balsamini, C.; Diamantini, G.; Di Giacomo,
B.; Tarzia, G.; Mor, M.; Plazzi, P. V.; Rivara, S.; Lucini, V.;
Nonno, R.; Pannacci, M.; Fraschini, F.; Stankov, B. M. J.
Med. Chem. 1997, 40, 1990.
37. Nonno, R.; Pannacci, M.; Lucini, V.; Angeloni, D.; Fra-
schini, F.; Stankov, B. M. Br. J. Pharmacol. 1999, 127, 1288.
38. Bradford, M. M. Anal. Biochem. 1976, 72, 248.
39. Stankov, B.; Cozzi, B.; Lucini, V.; Fumagalli, P.; Sca-
glione, F.; Fraschini, F. Neuroendocrinology 1991, 53, 214.
40. Cheng, Y.-C.; Pruso€, W. H. Biochem. Pharmacol. 1973,
22, 3099.
41. Van de Waterbeemd, H.; Testa, B. Adv. Drug Res. 1987,
16, 85.
42. DESDOP, v. 2.0; MIA srl: Perugia, Italy.
43. Baroni, M.; Clementi, S.; Cruciani, G.; Kettaneh-Wold,
K.; Wold, S. D. QSAR 1993, 12, 225.
44. Excel 97; Microsoft Corp., 1997.
45. Simca-S 6.01; Umetri AB & Ericsson AB, 1997.
46. Sybyl, v. 6.6; Tripos Inc.: 1699 South Hanley Rd., St.
Louis, MO 63144, USA, October 1999.
24. Dubocovich, M. L.; Masana, M. I.; Iacob, S.; Sauri, D. M.
Naunyn-Schmiedeberg's Arch. Pharmacol. 1997, 355, 365.
25. Spadoni, G.; Balsamini, C.; Bedini, A.; Diamantini, G.; Di
Giacomo, B.; Tontini, A.; Tarzia, G.; Mor, M.; Plazzi, P. V.;
Rivara, S.; Nonno, R.; Pannacci, M.; Lucini, V.; Fraschini, F.;
Stankov, B. M. J. Med. Chem. 1998, 41, 3624.
26. Tarzia, G.; Diamantini, G.; Di Giacomo, B.; Spadoni, G.;
Esposti, D.; Nonno, R.; Lucini, V.; Pannacci, M.; Fraschini,
F.; Stankov, B. M. J. Med. Chem. 1997, 40, 2003.
27. Sugden, D.; Pickering, H.; Teh, M.-T.; Garrattt, P. J. Biol.
Cell 1997, 89, 531.
28. Nonno, R.; Lucini, V.; Pannacci, M.; Mazzucchelli, C.;
Angeloni, D.; Fraschini, F.; Stankov, B. M. Br. J. Pharmacol.
1998, 124, 485.
47. Clark, M.; Cramer, R. D., III; Van Opdenbosch, N. J.
Comp. Chem. 1989, 10, 982.