Receptor- And ligand-based study on novel 2,2′-bithienyl derivatives as non-peptidic AANAT inhibitors

Article abstract of DOI:10.1021/ci9004805

Arylalkylamine N-acetyl transferase (serotonin N-acetyl transferase, AANAT) is a critical enzyme in the light-mediated regulation of melatonin production and circadian rythm. With the objective of discovering new chemical entities with inhibitory potencies against AANAT, a medium-throughput screening campaign was performed on a chemolibrary. We found a class of molecules based on a 2,2′-bithienyl scaffold, and compound 1 emerged as a first hit. Herein, we describe our progress from hit discovery and to optimization of this new class of compounds. To complete the study, computational approaches were carried out: a docking study which provided insights into the plausible binding modes of these new AANAT inhibitors and a three-dimensional quantitative structure-activity relationship study that applied comparative molecular field analysis (CoMFA) methodology. Several CoMFA models were developed (variable alignments and options), and the best predictive one yields good statistical results (q² = 0.744, r² = 0.891, and s = 0.273). The resulting CoMFA contour maps were used to illustrate the pharmacomodulations relevant to the biological activities in this series of analogs and to design new active inhibitors. This novel series of 2,2′-bithienyl derivatives gives new insights into the design of AANAT inhibitors.

Full text of DOI:10.1021/ci9004805

446
J. Chem. Inf. Model. 2010, 50, 446–460
Receptor- and Ligand-Based Study on Novel 2,2′-Bithienyl Derivatives as Non-Peptidic
AANAT Inhibitors
Alban Lepailleur,*,† Ste´phane Lemaˆıtre,† Xiao Feng,† Jana Sopkova-de Oliveira Santos,†
Philippe Delagrange,‡ Jean Boutin,‡ Pierre Renard,‡ Ronan Bureau,† and Sylvain Rault†
Centre d’Etudes et de Recherche sur le Me´dicament de Normandie, UPRES EA 4258, INC3M FR CNRS
3038, Universite´ de Caen-Basse Normandie, UFR des Sciences Pharmaceutiques, Boulevard Becquerel,
14032 Caen Cedex, France, and Laboratoires Servier, 125, Chemin de Ronde,
78290 Croissy-sur-Seine, France
Received December 13, 2009
Arylalkylamine N-acetyl transferase (serotonin N-acetyl transferase, AANAT) is a critical enzyme in the
light-mediated regulation of melatonin production and circadian rythm. With the objective of discovering
new chemical entities with inhibitory potencies against AANAT, a medium-throughput screening campaign
was performed on a chemolibrary. We found a class of molecules based on a 2,2′-bithienyl scaffold, and
compound 1 emerged as a first hit. Herein, we describe our progress from hit discovery and to optimization
of this new class of compounds. To complete the study, computational approaches were carried out: a docking
study which provided insights into the plausible binding modes of these new AANAT inhibitors and a
three-dimensional quantitative structure-activity relationship study that applied comparative molecular field
analysis (CoMFA) methodology. Several CoMFA models were developed (variable alignments and options),
and the best predictive one yields good statistical results (q2 ) 0.744, r2 ) 0.891, and s ) 0.273). The
resulting CoMFA contour maps were used to illustrate the pharmacomodulations relevant to the biological
activities in this series of analogs and to design new active inhibitors. This novel series of 2,2′-bithienyl
derivatives gives new insights into the design of AANAT inhibitors.
INTRODUCTION
control of the levels of circulating melatonin represents also
a major target for the search of inhibitors. Thus, AANAT
has received wide attention during past years.
Melatonin is a hormone produced predominantly by the
pineal gland.1,2 Extra pineal sites of melatonin production
include the retina,3 Harderian gland,4 gut,5 bone marrow,6
and skin.7 The enzyme pathway from tryptophan to mela-
tonin was identified many years ago and involves 5-hydroxy-
lation by tryptophan hydroxylase, decarboxylation by aromatic-
L-amino acid decarboxylase, acetylation by arylalkylamine
N-acetyl transferase (AANAT), and finally 5-methylation by
hydroxyindole-O-methyltransferase (HIOMT).8 Studies in
both the pineal gland and the retina indicate that there is a
highly conserved relationship between melatonin synthesis
and AANAT activity. Actually, AANAT is considered the
most important enzyme for the rhythm of melatonin syn-
thesis, since its activity is greatly enhanced at night in both
tissues.9
Inhibitors of AANAT would be useful for delineating the
role of melatonin in circadian rhythms and in its other
proposed but more controversial effects on aging, cancer,
cardiovascular function, and the immune system.1,15,16 No
AANAT inhibitor with potent in vivo activity has been
reported. Specific inhibitors of AANAT could potentially
serve as therapeutic agents for alleviating sleep and mood
disorders, both by raising serotonin and reducing melatonin
production.17,18
Up to now, five classes of AANAT inhibitors have been
described in the literature (see Zheng and Cole for a detailed
review).19 In brief, these five classes correspond to:
Melatonin Derivatives. Since it was reported that mela-
tonin is a competitive AANAT inhibitor, this hormone seems
to exert an autoregulatory control on its own biosynthesis.
Thus, loose structural analogues of the indolamine hormone
were evaluated on AANAT, and moderate inhibitors were
discovered.20
Peptidic Inhibitors. Ferry et al. reported the discovery
of tetrapeptides coming from the deconvolution of combi-
natorial libraries.21
Bisubstrate Analogs. It is suggested that AANAT cata-
lyzes the transfer of an acetyl group from acetylcoenzyme
A (acetyl-CoA) to serotonin, with the involvement of an
intermediate ternary complex, to produce N-acetylserotonin.
Based on this mechanism, it might be expected that a
bisubstrate analog inhibitor, derived from the tethering of
Melatonin plays important roles in many biological
processes, especially in circadian rhythms and in seasonal
reproduction.10-12 Melatonin pharmacology is, therefore, an
important field of investigation, with the aim of trying, on
one hand, to understand the mechanisms by which the
hormone exerts its multiple roles, and on the other hand, to
discover and assess new molecules that will permit modulat-
ing melatonin functions. Besides the classical search of
agonists and antagonists of the melatonin receptors,13,14 the
* Corresponding author. Telephone: (33)2-31-56-68-22. Fax: (33)2-31-
56-68-03. E-mail: alban.lepailleur@unicaen.fr.
† Centre d’Etudes et de Recherche sur le Me´dicament de Normandie.
‡ Laboratoires Servier.
10.1021/ci9004805  2010 American Chemical Society
Published on Web 03/02/2010

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 447
the trans-Tripos 24-00037 did not give the expected results
(IC50 ) 2.1 µM), the synthesis of the cis isomer was carried
out. Only cis/trans mixtures were obtained (60/40 and 70/
30 for the relative proportions), and their pharmacological
evaluation showed comparable affinities (IC50 ) 2.4 and 4.6
µM, respectively). Due to the interesting rigidified scaffold
of the byproduct 7, in comparison with 1, we also evaluated
its AANAT activity. With compound 7, not only the
ambiguity on the cis/trans isomer was put aside but also the
AANAT inhibitory potency was enhanced (IC50 ) 0.6 µM).
Starting from the new lead 7, small groups such as methyl
were appended to the piperidine ring and were generally
tolerated (10-13, Table 1). Moreover, replacing the piperi-
dine moiety with a decahydroquinoline (14), a pyrrolidine
(15), or an azepane (16) retained affinity. The replacement
of the piperidine with a thiomorpholine (17) or a morpholine
(18) also caused no change in affinity but seems to improve
the potency on the cellular model. To continue exploring
the structure-activity relationships of the piperidine moiety,
a number of polar groups were incorporated (19-26). In
general, these compounds retained affinity compared to the
piperidine derivatives 7, but the most important result was
their cellular potency improvement.
Among the 2,2′-bithienyl compounds synthesized during
our AANAT inhibitor program, the structure-activity rela-
tionships derived from the piperazine class are relatively
evident (Table 2). Indeed, the ortho substitutions on the
phenyl piperazine moiety (29 and 35) seem to be favored,
since the meta and para substitutions resulted in a degrada-
tion of affinity (30-34, 36, and 37). However, even if
compounds 27, 28, and 38 retained affinity compared to the
piperidine class, they showed a loss of cell-based potency.
To assess the influence of the piperidine or the piperazine
moiety, new 2,2′-bithienyl derivatives carrying a secondary
or tertiary amine (not included in a ring) were also prepared
(Tables 3 and 4). First, varying the substituents in this
position suggested that tertiary amines slightly improved
affinity compared with secondary amines. Second, incorpora-
tion of larger hydrophobic substituents on such an amine
seems to result in a drop in affinity. To continue exploring
the structure-activity relationships of these analogues,
hydroxyl groups were introduced and gave no significant
variation in the IC50 values, excepting for compound 64 in
comparison with compound 56.
X-ray Data. The solved crystal structure, corresponding
to compound 7, showed that the thiophene rings are situated
in a trans position and that the sulfur atoms lie on opposite
side of the molecule (Figure 2). The two thiophene rings
are slightly deviated from a coplanar position, since an angle
of about 17.9(1)° is observed between the planes of the rings.
The atoms of the triple bond are approximately coplanar with
the neighboring thiophene ring, an angle of about 6.1(1)° is
observed between the plane of the triple bond (the plane was
fitted through C9-C12 atoms) and the plane of the neigh-
boring thiophene ring. The piperidine ring is in a chair
conformation and is situated up with respect to the remaining
part of the molecule. In addition, the CdO bond is oriented
down. The deviation of the N14 atom is about 0.73 Å with
respect to the triple-bond plane, and the O13 deviation is
about 0.66 Å. The deviation of the CdO from the plane of
the thiophenes can be explained by a hydrogen (H)-bond-
type interaction, which occurs between the O13 and the
Figure 1. The first 2,2′-bithienyl hit.
indole and CoASH parts, could potentially mimic the ternary
complex and exert strong inhibition of AANAT.22 As
anticipated, the first bisubstrate analog inhibitor, designed
by Khalil and Cole,23 is a very potent AANAT inhibitor (IC50
) 150 nM). Subsequently, several bisubstrate analog inhibi-
tors have been designed.24,25
N-Haloacetylated Derivatives. Besides catalyzing the
acetyl transfer reaction from acetyl-CoA to serotonin to form
N-acetylserotonin, AANAT was also demonstrated to exert
a secondary alkyltransferase activity.24 N-Haloacetyltrypt-
amines were developed based on the pharmacological
exploitation of AANAT’s alkyltransferase activity.24,26
AANAT catalyzes the reaction between these compounds
and CoASH to form the inhibitory species, i.e., the bisub-
strate analogs. With the same idea, other N-haloacetylated
derivatives were developed as precursors of potent bisubstrate
inhibitors. Some of them exhibited high inhibitory activity.27,28
Rhodanine-based Compounds. More recently, a new
AANAT inhibitor class containing a rhodanine scaffold was
described.29 These derivatives, resulting from a high-
throughput docking, exhibited low-micromolar competitive
inhibition against acetyl-CoA and proved to be effective in
blocking melatonin production in pineal cells.
While embarking on an effort to develop nonpeptidic
AANAT inhibitors, we applied a medium-throughput screen-
ing (MTS) campaign on a chemolibrary commercialized by
Tripos (2500 compounds). The 2,2′-bithienyl derivative 1
(Tripos 24-00037, Figure 1) emerged as a first hit, and
structure-activity relationships were derived. A docking
approach was conducted and gave insights into the plausible
interactions between the new series of inhibitors and the
AANAT binding site. The docking process used the Glide
software30 and exploited a reported X-ray structure of
AANAT in complex with a bisubstrate analog.31 To complete
the study, a three-dimensional quantitative structure-activity
relationship (3D-QSAR) study which applied comparative
molecular field analysis (CoMFA) methodology,32 including
the steric and electrostatic fields, was also carried out in order
to obtain a predictive model from a selection of 43 2,2′-
bithienyl derivatives. The model was used to illustrate the
pharmacomodulations relevant to the biological activities and
to design new AANAT inhibitors.
RESULTS AND DISCUSSION
Identification of Lead Scaffold and Structure-
Activity Relationships. Compound 1 (Tripos 24-00037,
Figure 1), a commercially available compound, was identified
during a MTS campaign (2500 compounds) as an inhibitor
of AANAT. The micromolar activity (IC50 ) 0.64 µM) and
the chemical original nature of 1 led us to consider this
compound as a good initial hit.
At first, we realized the synthesis of 1-[(2E)-3-(2,2′-
bithien-5-yl)prop-2-enoyl]piperidine 1 in order to confirm
the observed activity. Since the pharmacological results of

448 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
Table 1. Structure-Activity Relationships around the Piperidine and Related Derivatives
C8-H8 of a neighboring molecule (1 + x, y, z). The distance
between O13 and H8 is about 2.29 Å.
Glide methodology. Among the two H-bond constraints
defined during the receptor grid generation step, at least one
of them had to be achieved (with the main chain amide of
Leu124 or the water molecule 1061). These constraints were
chosen because they are situated at the junction between the
AcCoA and the serotonin binding sites, where the carbonyl
group of compound 7 is supposed to be. The docking results
led us to the consideration of three different positions
(Figures 4-6). First, and as initially envisaged, the bithienyl
moiety could occupy the AcCoA binding site, leading to the
positioning of the piperidine group in the serotonin binding
pocket. In this case, two orientations can be observed for
the carbonyl group of the ligand, depending on the H-bond
constraint which is filled (Figures 4 and 5). In order to
determine the accuracy of this prediction, the piperidine
group of 7 was replaced by a tryptamine moiety, and the
inhibitory potency of the new compound 50 was evaluated
(Table 3). The pharmacological results showed that the
introduction of the tryptamine moiety causes a loss of
AANAT inhibitory activity (IC50 > 100 µM). As a conse-
quence, this single synthetic change to the derivative 7
suggested that the bithienyl moiety should occupy the
serotonin binding pocket rather than the AcCoA binding site,
as highlighted by the third docking possibility (Figure 6).
Molecular docking studies also demonstrated that the car-
bonyl oxygen of 7 is not able to form a H-bonding contact
with the main chain amide of Leu124 when the bithienyl
moiety occupies the serotonin binding pocket. The resulting
position involves two water-mediated H-bonds with both the
In the crystal packing, beside the already described
interaction O· · · H-C, stacking interaction occurs between
the thiophene rings of two neighboring molecules along the
b axis. The distance between the thiophene ring planes of
neighboring molecules is about 3.69 Å.
Molecular Docking Studies. Molecular modeling studies
were initiated in order to probe the possible interactions
between compound 7 and AANAT. The X-ray coordinate
system used for these studies corresponds to 1CJW, a high-
resolution structure of AANAT in which a bisubstrate analog
is bounded at the enzyme active site.31 The bisubstrate analog
mimics the structure of the presumed intermediate formed
during catalysis and was first described by Khalil and Cole.23
The analog is held in a conformation that is approximately
“S”-shaped (Figure 3). The pantetheine group is aligned
parallel to strand ꢀ5 of the central ꢀ sheet, held in place by
a number of H-bonding contacts involving main chain atoms
(Leu124 and Val126). The R- and ꢀ-phosphates of the
adenosine diphosphate (ADP) group also form several
H-bonds with AANAT residues (Gln132, Gly134, Gly136,
and Ser137). On the opposite side, the indole ring of the
tryptamine group is held in place through extensive hydro-
phobic contacts with six residues (Phe56, Pro64, Met159,
Val183, Leu186, and Phe188), which converge to form a
pocket, referred to as the serotonin binding site.
As explained in the Experimental Section, compound 7
was docked into the binding site of AANAT using the SP

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 449
Table 2. Structure-Activity Relationships around the Piperazine and Related Derivatives
Table 3. Structure-Activity Relationships around the Secondary and Tertiary Amine Derivatives (Carrying a Ring)

450 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
Table 4. Structure-Activity Relationships around the Secondary and Tertiary Amine Derivatives (Not Carrying a Ring)
main and side chains of Glu161. Besides, the bithienyl moiety
fills the hydrophobic funnel that led directly toward His120
and His122, which are involved in the catalytic mechanism.31
To go even further, we can postulate that the beneficial effect
observed with the introduction of polar groups could result
from H-bond contacts with Leu124, Val126, and/or Arg131.
This hypothesis is strengthened by the widely different
activities observed for compounds 20 vs 12 and 19 vs 11.
With the same idea, we can assume that the better activity
observed for compound 64, in comparison with compound
56, is due to an effect of a H-bond. In fact, these compounds
differentiate only by the addition of a hydroxyl group. In
addition, the lower activity of compound 63, when compared
with compound 20, seems to indicate that a precise confor-
mation is mandatory. This difference could result from the
formation of an intramolecular H-bond, preventing a H-bond
contact with Val126 and/or Arg131. The same result can be
deduced from compound 62. Moreover, compounds 75-77
also support this hypothesis, since their higher affinity is
coupled with the presence of a tertiary amine, which avoid
the undesired folding behavior of the hydroxylated chain.
By considering the phenyl piperazine derivatives, the H-bond
constraint with the main chain amide of Leu124 could not
be achieved (data not shown). Consequently, the solvent
exposure of the carbonyl group of these bithienyl derivatives
is highly plausible, and a water network could mediate
H-bond interactions with polar residues such as Glu161.
CoMFA Studies. Even though the docking studies gave
insights into the binding mode possibilities for the
bithienyl derivatives, uncertainties remain. Thus, we
realized a 3D-QSAR study which applied the CoMFA
methodology in order to better understand the relations
between the pharmacomodulations and the biological
activities in this series of analogs. The compounds,
referenced in Table 9, were selected as a training set, and
their capacity to inhibit AANAT was expressed in terms
Figure 2. X-ray crystal structure of compound 7.

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 451
Figure 3. AANAT-bisubstrate analog complex (PDB code 1CJW).
Figure 4. Comparison between the first docking position (in gray) and the bisubstrate binding mode (in orange).
of pIC50 or -log(IC50), where IC50 represents the drug
concentration that inhibits 50% of pure AANAT activity.
The pIC50, which covers four logarithmic units, was used
as a dependent variable in the CoMFA calculations. For
these studies, the selection of the training set compounds
was influenced, on one hand, by the availability of the
IC50 values and, on the other hand, by the presence/absence
of a stereocenter. Indeed, the compounds which exhibited
a stereocenter were put aside, since their consideration
increased the complexity of the crucial alignment step.
All compounds were initially aligned on the X-ray
crystallographic structure of 7 (Figure 2). The fitting
processes were achieved according to selected pairs of
atoms, using a SPL program. The fixed atoms correspond-
ing to the template are represented by the entire molecule
7, excepting atoms C15-C19 (Figure 2). Concerning the
geometries of the remaining moieties for the training set
molecules, the structural information we collected on the
Cambridge Structural Database (CSD)33 led us to uncer-
tainties. Consequently, we had to propose four different

452 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
Figure 5. Comparison between the second docking position (in gray) and the bisubstrate binding mode (in orange).
Figure 6. Comparison between the third docking position (in gray) and the bisubstrate binding mode (in orange).
alignments. Table 5 presents examples that illustrate these
various possibilities.
between the statistical results. These results indicated the
good predictive capacity of the models, which yielded a
cross-validated correlation coefficient q2 from 0.549 to 0.728,
with reasonable respective standard errors of prediction SEcv
(from 0.550 to 0.432). As just noted, a q2 value of 0.3
corresponds to a 95% confidence limit, which minimizes the
risk of chance correlation.34 These analyses were then
repeated with a grid of 1 Å because, according to Cramer,32
First of all, we searched the optimal alignment by
considering the CoMFA default parameters (“smooth” op-
tion). The statistical results obtained with all alignments are
recapitulated in Table 6. The CoMFA analyses were initially
performed with probe atoms placed on a regularly spaced
grid of 2 Å. No significant differences were observed

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 453
Table 5. Representation of the Four Alignment Possibilities
37) substitutions on the phenyl ring. On the contrary, ortho
substituents seem to be tolerated (29 and 35).
With regard to the contour maps of the favorable elec-
trostatic fields, a moderate increase of activity could be
achieved by incorporing polar groups. Particularly, these
favorable substitutions are exemplified by compounds 20 and
25. This graphic representation can also be illustrated with
compound 38 (Figure 12). Among the inhibitors presenting
a piperazine moiety, 38 is the most potent. According to the
CoMFA results, its C1-N2-C3-N4 torsion angle (140°)
leads to a correct orientation of the pyridinyl ring, with regard
to the favorable electrostatic area (field e, Figure 12). It
results in a slight increase of activity in comparison with
compound 28.
Using the comparative molecular similarity indices analy-
ses (CoMSIA)36 program, an approach related to the CoMFA
analysis, we added hydrophobic parameters to our 3D-QSAR
equation. Since this additional information did not lead to a
significant improvement of the model, the CoMSIA approach
was put aside.
CoMFA-Based Pharmacomodulations. To validate and
complete our model, we proposed the synthesis of new
compounds, and we attempted to predict their activities. These
propositions started with 48 which exhibits a first affinity and
an interesting aromatic cycle, in view of the envisaged phar-
macomodulations. Indeed, due to the position of the phenyl ring,
appropriate substitutions could lead to a good occupancy of the
previously discussed favorable steric regions.
The predicted and experimental pIC50 values for the six
new compounds (78-83) are specified in the Table 10.
Clearly, their activities are overestimated but are generally
comparable to those of the lead compounds. Besides, we
have highlighted the harmful effect of a meta substitution
on the phenyl (81 and 82). Finally, we proposed a compound
which exhibits a tertiary amine substituted not only by a
phenyl ring but also by an allyl chain. The pharmacomodu-
lation was suggested by the promising compounds 61 and
74, which combine good activities on pure enzyme and
cellular models (Table 4). The resulting derivative 83 retained
affinity in comparison with the best ones.
this operation achieves a quality improvement of sampling.
As anticipated, the diminution of the grid spacing improves
the results.
Since all alignments exhibited comparable performances,
we decided to carry out our studies with the third one
(Figure 7), which presented slightly better q2 and SEcv values.
As mentioned above, these models have been generated
with the default “smooth” option. However, other methods
of field generation are available to users of CoMFA, and in
some cases, better models could be obtained.35 Thus, the
influences of these three methods were studied for the third
alignment. It should be noted that for these analyses, the
“box” option yielded worse results than both the “smooth”
and “abrupt” options (Table 7). Finally, we retained the
model generated from the “abrupt” method.
The complete statistical results are recapitulated in Table
8. Using the optimal number of components (3), the
nonvalidated model was produced and led to satisfying the
correlation coefficient r2 (0.891) and the standard error of
estimate SEE (or s, 0.273). These data suggested a good
correlation between the steric and electrostatic molecular
fields and the activities measured for the training set. The
contributions were 67% for the steric fields and 33% for the
electrostatic fields (final equation). The experimental vs
predicted activity values for the final model (without cross-
validation) are presented in Figure 8 and in Table 9. The
plot of residuals for the data set is shown in Figure 9.
Graphic representation of the CoMFA model is displayed
in Figure 10. It shows regions where variations of steric or
electrostatic nature in the structural features of the training
set compounds lead to variations of activity. Green contours
indicate regions where an increase of steric bulk will enhance
activity, whereas yellow contours indicate regions where an
increase of steric bulk will reduce activity. Regions where
an electrostatic interaction is favorable for activity are
indicated in blue, while regions where an electrostatic
interaction is detrimental for activity are indicated in red.
Within the training set, few inhibitors are able to occupy
simultaneously the two regions corresponding to favorable
steric bulks (fields a and b, Figure 11). Compound 53 (Table
3) is one of them, and this characteristic enables it to belong
to the best-predicted compounds (Table 9).
CONCLUSION
This study initially describes the discovery of a new class
of nonpeptidic AANAT inhibitors based on a 2,2′-bithienyl
scaffold. Compound 1 emerged from a medium-throughput
screening campaign performed on a chemolibrary com-
mercialized by Tripos. Structure-Activity relationships were
derived and led to a first lead, compound 7, whose interesting
rigidified scaffold attracted our attention. Afterward, molec-
ular modeling studies were initiated in order to probe the
possible interactions between compound 7 and AANAT.
First, docking studies pointed to three plausible binding
modes for this new AANAT inhibitor. The most promising
one suggested that the bithienyl moiety should occupy the
serotonin binding pocket and that the carbonyl oxygen should
be involved in water-mediated H-bonds with the main and
side chains of Glu161. Besides, in this case, the bithienyl
moiety should fill the hydrophobic funnel that led directly
toward the His120 and His122, which are involved in the
catalytic mechanism. Second, and in order to better under-
stand the relations between the pharmacomodulations and
The region where an increase of steric bulks will reduce
activity primarily interests the piperazine derivatives related
to inhibitor 28 (Table 2). This region highlights the detri-
mental effect of meta (30, 34, and 36) and para (31, 33, and

454 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
Table 6. Statistical Results for the Four Possible Alignmentsa
steric
contribution, %
electrostatic
contribution, %
alignment
grid, Å
ONC
q2
SEcv
r2
SEE(s)
1
2
1
2
1
2
1
2
1
2
3
2
3
2
3
2
3
0.588
0.712
0.580
0.708
0.607
0.728
0.549
0.688
0.525
0.445
0.531
0.448
0.513
0.432
0.550
0.463
0.780
0.894
0.778
0.896
0.767
0.891
0.796
0.902
0.384
0.270
0.386
0.267
0.395
0.274
0.369
0.260
76
66
75
70
77
67
73
70
24
34
25
30
23
33
27
30
2
3
4
a ONC is the optimal number of components, q2 is the cross-validated correlation coefficient, SEcv is the standard error of prediction, r2 is the
correlation coefficient, and SEE is the standard error of estimate.
Figure 7. Selected alignment (number three).
Table 7. Additional Statistical Results for Alignment Three
steric
contribution, %
electrostatic
contribution, %
method
smooth
grid, Å
ONC
q2
SEcv
r2
SEE(s)
2
1
2
1
2
1
2
3
2
3
3
2
0.607
0.728
0.595
0.744
0.603
0.676
0.513
0.432
0.521
0.420
0.522
0.466
0.767
0.891
0.767
0.891
0.874
0.797
0.395
0.274
0.395
0.273
0.294
0.368
77
67
77
67
68
75
23
33
23
33
32
25
abrupt
box
Table 8. Final Results of the CoMFA Analysis for Alignment
by its chemical structure and its potency on the cellular
model. Presently, supplementary pharmacological evaluations
are ongoing, and synthetic experimental details will be
described elsewhere.
Threea
cross-validation
number of components
final equation
q2
SEcv
r2
SEE(s)
1
2
3
4
5
6
0.565
0.720
0.744
0.740
0.744
0.733
0.533
0.433
0.420
0.428
0.431
0.446
EXPERIMENTAL SECTIONS
0.891
0.273
Chemistry. The synthesis of 1 started from 2-bro-
mothiophene (Scheme 1). The formation of 5-bromo-2,2′-
bithiophene 2 or 5-iodo-2,2′-bithiophene 3 derivatives37,38
were obtained in good yields. The synthesis of methyl (2E)-
3-(2,2′-bithien-5-yl)acrylate 4 was prepared by a palladium-
catalyzed coupling with methyl acrylate under phase-transfer
a With 1 Å spacing and with the “abrupt” method.
the biological activities in this series of analogues, we
realized a 3D-QSAR study that applied CoMFA methodol-
ogy. The best predictive model yields good statistical results
(q2 ) 0.744, r2 ) 0.891, and s ) 0.273), and the resulting
contour maps were used to design new inhibitors with
micromolar activity.
To complete the study, 18 representative compounds were
selected among all our 2,2′-bithienyl derivatives, and their
affinities for MT1 and MT2 melatoninergic receptors were
evaluated according to Audinot et al.45 Satisfying results were
observed since all of them showed a Ki greater than 10 µM
(unpublished data).
To conclude, this work gives new insights into the design
of nonpeptidic AANAT inhibitors. Besides, the selection of
the morpholine derivative, 18 as a new lead, was encouraged
Figure 8. Plot of the experimental vs predicted pIC50 values
(without cross-validation).

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 455
Figure 9. Plot of the residual values (without cross-validation).
Table 9. Experimental vs Predicted Values for the Final Modela
no.
experimental pIC50
predicted pIC50
difference
7
6.22
5.08
5.66
6.22
5.29
5.64
6.52
5.66
4.95
6.30
5.43
5.05
4.33
3.99
3.58
3.60
3.84
5.03
4.60
3.78
5.54
4.40
4.80
6.10
5.82
5.03
5.20
6.30
6.22
6.22
5.38
5.14
6.05
5.29
5.30
5.74
5.80
5.22
6.22
5.96
6.30
4.92
5.82
5.85
5.82
5.62
5.99
5.64
5.53
6.17
5.85
5.39
6.07
5.39
4.71
4.68
3.82
3.60
3.77
3.68
4.87
4.34
3.50
5.86
4.74
4.94
5.82
5.74
5.16
5.30
6.18
6.14
6.18
5.37
5.45
5.67
5.28
5.53
5.81
5.91
5.23
5.93
5.96
5.98
5.42
5.68
0.37
–0.74
0.04
0.23
–0.35
0.11
12
15
16
17
18
20
23
24
25
27
28
29
30
31
33
34
35
36
37
38
39
40
42
43
44
48
51
52
53
55
60
61
64
66
67
68
69
70
71
73
74
75
0.35
–0.19
–0.44
0.23
0.04
0.34
–0.35
0.17
–0.02
–0.17
0.16
0.16
0.26
0.28
–0.32
–0.34
–0.14
0.28
0.08
–0.13
–0.1
0.12
0.08
0.04
0.01
–0.31
0.38
0.01
–0.23
–0.07
–0.11
–0.01
0.29
Figure 10. Standard deviation *PLS coefficients contour plots of
CoMFA steric and electrostatic fields (green: favorable steric bulks
a and b; yellow: detrimental steric bulk c; blue: electrostatic favored
areas d and e; and red: electrostatic disfavored areas f and g).
Compound 7 inside the map.
Figure 11. Standard deviation *PLS coefficients contour plots
of CoMFA steric fields (green: favorable steric bulks a and b;
and yellow: detrimental steric bulk c). Compound 7 inside the
map.
0.00081
0.32
–0.5
Figure 12. Standard deviation *PLS coefficients contour plots
of CoMFA favorable electrostatic areas. Compound 38 inside
the map.
0.14
a Without cross-validation.
conditions.39 The formation of acid (2E)-3-(2,2′-bithien-5-
yl)acrylic was obtained by saponification of the ester, and
finally compound 1 was achieved by substitution of the
mixed anhydride with piperidine.
Another way of synthesis was, thus, envisaged to obtain
the cis product. This method consists, first of all, in realizing

456 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
Table 10. Experimental vs Predicted Values for the New Proposed Inhibitors
Scheme 1. Synthesis of 1-[(2E)-3-(2,2′-bithien-5-yl)prop-2-enoyl]piperidine 1 (Tripos 24-00037)a
a Reagents and conditions: (a) (1) Mg, Et2O, reflux 30 min; (2) Ni(dpp)Cl2, reflux 4 h, 97%; (b) NBS or NIS, CHCl3/AcOH, room temperature,
12 h under argon, 60-70%; (c) Pd(OAc)2, NBu4I, K2CO3, methyl acrylate, DMF, 85 °C, 24 h under argon, 78%; (d) NaOH, MeOH/H2O, HCl,
85%; (e) ClCO2CH3, Et3N, acetone, piperidine, 12 h, room temperature, 68%.
Scheme 2. Synthesis of Methyl 3-(2,2′-Bithien-5-yl)prop-2-ynoate 5a
a Reagents and conditions: (a) Et3N, PdCl2(PPh3)2, CuI, THF, reflux 18 h under argon, 20%.
a coupling of Sonogashira40 from the methyl propiolate with
3. Several assays were realized with various palladium
catalysts (Pd(PPh3)4 and PdCl2(PPh3)2) and with various
solvents, such as DMF, acetonitrile, or THF. The best results
are obtained with the palladium catalyst PdCl2(PPh3)2 in THF
(20% yield, see Scheme 2).
Due to the low yields, we were interested in the synthesis
of propiolamides 6 derivatives from propiolic acid and
amines. Multiple synthetic routes were explored before
optimal conditions were developed. The method involved
the cross-coupling reaction of propiolamides with 3. The
product (compound 7) was carried out at room temperature

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 457
Scheme 3. Synthesis of 1-[3-(2,2′-Bithien-5-yl)prop-2-ynoyl]piperidine 7a
a Reagents and conditions: (a) CHCl3, DCC, 0 °C, 1 h, 58%; (b) 5-iodo-2,2′-bithiophene, Et3N, Pd(PPh3)4, CuI, DMF, room temperature, 18 h,
84%.
Scheme 4. Synthesis of 1,1-[(2Z)-3-(2,2′-bithien-5-yl)prop-2-enoyl]piperidine 8 and 1-[3-(2,2′-Bithien-5-yl)propanoyl]piperidine 9 derivativesa
a Reagents and conditions: (a) H2, Pd/BaCO3, EtOH, room temperature, 1.5 h, 82%.
in the presence of Pd(PPh3)4 (5 mol %), CuI (5 mol %), and
triethylamine in DMF (Scheme 3). Finally, the reaction of
catalytic hydrogenation41 was realized in the presence of
palladium on barium carbonate and quinoleine in anhydrous
ethanol. Only mixtures of cis/trans were obtained whose
reports were of the order 60/40 and 30/70 (separable by
chromatography) and total reduction of the compound
(Scheme 4).
washed (40 mL buffer A) and then eluted with 150 mL of
buffer B (50 mM Tris/HCl, pH 8.0, containing 100 nM
sodium citrate, 10 mM dithiothreitol, and 10% (v/v) glyc-
erol). The glutathione-S-transferase-AANAT was eluted
sequentially with 40 mL of 10 mM glutathione in buffer B.
Active fractions were pooled and stored frozen at 80 °C until
used. The protein concentration was determined by the
Bradford assay44 (Protassay, Bio-Rad, Ivry-sur-Seine, France)
with bovine serum albumin as the standard.
Afterward, we were interested in the synthesis of a
byproduct possessing a triple connection and carrying
different amines (secondary, tertiary, substituted piperidine
or piperazine, and aniline).
Assays for Enzymatic and Cellular AANAT. An enzy-
matic assay for AANAT was performed in a 100 µL final
volume comprising 1 µg of enzyme, in a phosphate buffer
(50 mM sodium phosphate, pH 6.8, 500 mM NaCl, and 2
mM EDTA) with 1 mM acetyl-CoA and 4 mM phenylethy-
lamine. Compounds were dispensed in black 96-well plates
(Costar, Dutscher, Issy-les-Moulineaux, France) under a
volume of 10 µL, at a starting concentration of 100 µM in
pure dimethylsulfoxide. Cellular assays for AANAT were
performed on subconfluent cells seeded in 96-well plates and
incubated with 10 µL of the tritiated substrate [3H]phenyl-
ethylamine ([3H]PEA, Amersham Pharmacia Biotech, 2.78
TBq mmol-1). In brief, for cell studies, 70 µL of incubation
buffer (phosphate buffer detailed above) containing 0.5%
Tween 85 were added to each well of the plates. Plates were
shaken at room temperature for 30 min using an orbital
shaker (Bioblock Scientific, Strasbourg, France). Each well
was then sonicated once (setting 40/100, probe 0.2 cm
diameter) for five seconds. For [3H]bromoacetyltryptamine
experiments ([3H]BAT, Amersham-Pharmacia Biotech, 2.25
TBq mmol-1), 10 µL of the compound was added, along
with cold PEA standard incubation, to the enzyme or to the
cells in 96 well plates. After incubation (30 min, 37 °C),
the reaction was stopped by the addition of 50 µL of 10%
(v/v) trichloroacetic acid in water. Thirty microliters of the
enzymatic reaction, conditioned culture media, or lysed cells
were analyzed by reverse phase HPLC using a C4 ASTEC,
4.6 × 150 mm column (CIL Cluzeau, Ste-Foy-la-Grande,
France). The column was developed with a linear gradient
of 5-35 or 0-100% (v/v) acetonitrile in H2O, 0.1% (v/v)
trifluoroacetic acid at a flow rate of 1 mL min-1 for 15 or
30 min. The area under the peak of the reaction product was
integrated by the software of the HPLC system (HP-
CHEMSTATION, v06.01; Agilent Technologies, Wald-
bronn, Germany), by comparison of surfaces of the peak for
Culture of AANAT-Expressing Cell Line. The selection
of a positive clone, CHO-hAANAT-8, has been described
elsewhere.42 Cells were seeded (1 × 106 cells per well in a
96-well plate) and cultured for 72 h with 5% CO2:95% air
(v/v) in a humidified incubator in 96-well plates containing
Ham 4MF12 medium (GIBCO BRL, Invitrogen SARL,
Cergy-Pontoise, France) supplemented with 10% (v/v) fetal
bovine serum, 50 µg mL-1 penicillin/streptomycin, 2 mM
L-glutamine, 5 mg mL-1 G418, all from Life Technologies.
Expression and Production of hAANAT. Expression,
production, and purification of hAANAT have been de-
scribed by Ferry et al.43 In brief, the human arylalkylamine
N-acetyltransferase cDNA coding five region (kindly pro-
vided by D. C. Klein and S. L. Coon, National Institutes of
Health, Bethesda, MD) was inserted into the bacterial
expression vector pGEX-4T (Pharmacia, Les Ulis, France),
leading to the expression of a protein fused to glutathione-
S-transferase. An Escherichia coli strain [BL21(DE3)pLysS]
was transformed with the resulting plasmid. Expression of
the construct was induced by adding isopropyl thio-ꢀ-D-
galactoside (0.2 mM) at 24 °C for 6 h. The cells were
harvested by centrifugation (5000 g, 4 °C for 10 min). All
purification procedures were performed at 0-4 °C. Ap-
proximately 10 g of a frozen bacteria pellet was thawed in
40 mL of 2 × NaCl/Pi containing 10 mM dithiothreitol, a
cocktail of protease inhibitors (Complete, Roche; 1 tablet
per 50 mL) and Tween 85 (5% v/v). This suspension was
sonicated eight times for 1 min in ice, centrifuged (20 000
g, 20 min), and the supernatant passed through an affinity
column 40 mL h-1 (glutathione-sepharose, Pharmacia), that
was previously equilibrated with buffer A (2× NaCl/Pi, pH
6.9, containing 10 mM dithiothreitol). The column was

458 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
MOLECULAR MODELING
N-acetylPEA in the presence and absence of compounds. At
least two determinations were made for each experiment.
Each inhibition factor was calculated using the mean values
of those duplicates.
Molecular Docking Studies. The molecular docking
studies were performed using Maestro, version 8.5, and SP
Glide, version 5.0.30 The ligand 7 was docked flexibly to
the binding site of 1CJW.31 The coordinates for 1CJW were
downloaded from the RCSB Protein Data Bank.48 In this
structure, AANAT is complexed with a bisubstrate analog.
The protein was prepared for subsequent grid generation and
for docking using the protein preparation wizard tool supplied
with Glide. Using this tool, all H-atoms were added, the
protonation states for histidine residues were optimized,
crystallographic waters not deemed to be important for ligand
binding were deleted, and the entire protein was minimized.
The structural water near Glu161 was preserved; it was
reoriented and adjusted using the commands within the
protein preparation wizard tool available within Glide, so
that H-bonding between the ligand and the protein could be
established. Next, a grid was prepared using the receptor grid
generation tool in Glide. With the bisubstrate analog in place,
the centroid of the workspace ligand was chosen to define
the grid box. The option to dock ligands similar in size
to the workspace ligand was selected for determining the
grid sizing. For the single-ligand docking, the standard
precision mode was selected. The default settings for scaling
the van der Waals radii were selected: a scaling factor of
0.8 and a partial charge cutoff of 0.25. H-bond constraints
were defined for the docking runs. Indeed, the H-bond
between the ligand and the water molecule 1061 or the main
chain amide of Leu124 was required. In pose prediction and
enrichment studies, a maximum of 20 poses per ligand were
saved.
CoMFA Studies. The 3D structures of all compounds in
the training set were obtained using the Sybyl program
package.49 Their geometries were optimized using the Tripos
force field50 with the distance-dependent dielectric function
and the Powell conjugate gradient algorithm51 (convergence
criterion of 0.05 kcal/mol Å). The alignment of the com-
pounds is a crucial step in CoMFA approach. For this study,
each molecule was initially aligned on the X-ray crystal-
lographic structure of 7 (Figure 2). The fitting process was
performed using an “in-house” developed script written in
SPL (Sybyl programming language). This SPL program
achieved the fit of all compounds on the template according
to selected pairs of atoms. The fixed atoms correspond to
the entire molecule 7 excepting atoms C15-C19 (Figure 2).
For the CoMFA studies, a 3D cubic lattice box with a grid
spacing of 2 and 1 Å was automatically defined around the
aligned molecules. The molecular fieldsssteric and electro-
staticswere sampled at each point of the grid using a C sp3
Binding Experiments at Human MT1 and MT2 Recep-
tors. Binding experiments were carried out as described
previously in Audinot et al.45 The CHO-K1 cell lines stably
expressing either hMT1 or hMT2 receptors were grown at
confluence, harvested in phosphate buffer containing 2
mMEDTA, and centrifuged at 1000 g for 5 min at 4 °C.
The resulting pellet was suspended in 5 mM Tris/HCl, pH
7.4, containing 2 mM EDTA, and homogenized using a
Kinematica (Lucerne, Switzerland) polytron. The homoge-
nate was then centrifuged (20 000 g, 30 min, 4 °C), and the
final pellet was suspended in 75 mM Tris/HCl, pH 7.4,
containing 2 mM EDTA and 12.5 mM MgCl2. The deter-
mination of protein content was performed, according to
Bradford,44 using the Bio-Rad kit (Bio-Rad SA, Ivry-sur-
Seine, France). Aliquots of membrane preparations were
stored at -80 °C until further use. 2-[125I]iodomelatonin
binding assay experiments were conducted as follows:
membranes were incubated for 2 h at 37 °C in binding buffer
(Tris/HCl 50 mM, pH 7.4, 5 mM MgCl2) in a final volume
of 250 µL containing 2-[125I]iodomelatonin (25 or 200 pM
for competition experiments in MT1-CHO and MT2-CHO
cells, respectively, according to the slightly better affinity
of this compound to MT1 as compared to MT2). Nonspecific
binding was defined with 10 µM melatonin. The reaction
was stopped by rapid filtration through GF/B unifilters
(Whatmann, Maidstone, Kent, UK), followed by three
successive washes with ice-cold binding buffer. Data were
analyzed with PRISM software (GraphPad Software Inc., San
Diego, CA). For saturation assays, the density of binding
sites (Bmax) and the dissociation constant of the radioligand
(KD) values were calculated using Scatchard plot analysis.
For displacement experiments, inhibition constants (Ki) were
calculated according to the Cheng-Prussof equation: Ki )
IC50/[1 + (L/KD)], where IC50 is the inhibitory concentration
50% and L is the concentration of 2-[125I]iodomelatonin. For
the correlation analysis of pKi values, the Pearson product-
moment correlation coefficient was used.
X-ray Data. Suitable crystals of the title compounds were
obtained by slow evaporation from 7 at room temperature,
and they were mounted on a glass fiber. Diffraction data were
collected on an Enraf-Nonius-CAD4 diffractometer with Mo
KR radiation (λ ) 0.71073 Å) at room temperature. Data
were measured using θ/2θ scan. The data treatment, polar-
ization, decay corrections, and Gaussian numerical absorption
correction were carried out with the JANA98 program.46 The
crystal structure was solved by direct methods using SHELX-
97 package.47 All nonH-atoms were refined anisotropically.
The positions of H-atoms were determined via difference
Fourier maps and refined with isotropic atomic displacement
parameters, with exception of H-atoms of C2, C3, and C4
(Figure 2) that were calculated and fixed on the C-atoms in
the ideal geometry. The temperature factors of fixed H-atoms
were refined isotropically. One title compound is present in
the asymmetric unit. CCDC reference number 151632.
atom with a +1 charge as probe. The default cutoff (Ecut
)
30 kcal/mol) was applied to both fields by considering the
“smooth” method. For all field values greater than 1.2 Ecut,
a value of Ecut is applied, and for field values between 0.8
and 1.2 Ecut, the following interpolation is applied
E2 - 2EupE + E2lo
Esmooth
)
2(Elo - Eup)
where E is the original field value and Elo and Eup ) 0.8 and
1.2 Ecut, respectively.
For the cutoff, another method called “abrupt” was also
tested, instead of the “smooth” approach. This method simply

NON-PEPTIDIC AANAT INHIBITORS
J. Chem. Inf. Model., Vol. 50, No. 3, 2010 459
Lydic, R., Baghdoyan, H. A., Ed.; CRC Press: Boca Raton, FL, 1999;
pp45-59.
sets all probe values that experience interaction energy greater
than Ecut to Ecut. In this study, the Lennard-Jones parameters
and the van der Waals radii were taken from the Tripos force
field, and the partial charges were calculated with the semiem-
pirical MOPAC package using the Mulliken method (PM3
Hamiltonian). The method of partial least squares (PLS)
implemented in Sybyl was used to construct and validate the
models. To determine the optimal number of components, the
SAMPLS algorithm was used.52 The optimum number of
components (ONC) was retained for final PLS analyses (first
minimum in the standard error).53 Cross-validation was per-
formed using the leave-one-out procedure, wherein one com-
pound is removed from the data set, and its activity is predicted
using the model derived from the rest of the data set. A column
filtering of 2.0 kcal/mol was used to reduce noise. Finally,
nonvalidated models were produced, and conventional correla-
tion coefficient r2 and standard error of estimate SEE (or s) were
computed. An alternative to the standard method was also
considered to calculate the CoMFA fields. The “box” option
was used, which consists of replacing the probe atom by eight
probe atoms (making the corners of a cube) centered on the
coordinates of the original one. The interactions between each
probe atom and target molecule are calculated, and the average
value is assigned to the original probe coordinates. With this
“box” option, the cutoff is applied by considering the “abrupt”
method. From fully validated CoMFA models, contour maps
are presented displaying the most relevant regions of the space,
where the variations in statistical steric and electrostatic fields
are the largest. These contour maps highlight the areas where
changes in the molecular field values are strongly associated
with changes in binding affinities.
(10) Axelrod, J. The Pineal Gland: a Neurochemical Transducer. Science
1974, 184, 1341–1348.
(11) Reiter, R. J. Pineal Melatonin: Cell Biology of its Synthesis and of
its Physiological Interactions. Endocr. ReV. 1991, 12, 151–180.
(12) Klein, D. C.; Coon, S. L.; Roseboom, P. H.; Weller, J. L.; Bernard,
M.; Gastel, J. A.; Zatz, M.; Iuvone, P. M.; Rodriguez, I. R.; Begay,
V.; Falcon, J.; Cahill, G. M.; Cassone, V. M.; Baler, R. The Melatonin
Rhythm-Generating Enzyme: Molecular Regulation of Serotonin
N-Acetyltransferase in the Pineal Gland. Recent Prog. Horm. Res.
1997, 52, 307–357.
(13) Mody, S. M.; Hu, Y.; Ho, M. K. C.; Wong, Y. H. In Search of Novel
and Therapeutically Significant Melatoninergic Ligands. Recent Pat.
CNS Drug DiscoV. 2007, 2, 241–245.
(14) Zlotos, D. P. Recent Advances in Melatonin Receptor Ligands. Arch.
Pharm. (Weinheim, Ger.) 2005, 338, 229–247.
(15) Pierpaoli, W.; Regelson, W.; Colman, C. The Melatonin Miracle.
Nature’s Age-ReVersing, Disease-Fighting, Sex-Enhancing Hormone,
Pocket Books: New York, 1995.
(16) Reppert, S. M.; Weaver, D. R. Melatonin Madness. Cell 1995, 83,
1059–1062.
(17) Lewy, A. J.; Sack, R. L.; Miller, L. S.; Hoban, T. M. Antidepressant
and Circadian Phase-Shifting Effects of Light. Science 1987, 235, 352–
354.
(18) Bonhomme, N.; Esposito, E. Involvement of Serotonin and Dopamine
in the Mechanism of Action of Novel Antidepressant Drugs: a Review.
J. Clin. Psychopharmacol. 1998, 18, 447–454.
(19) Zheng, W.; Cole, P. A. Serotonin N-Acetyltransferase: Mechanism
and Inhibition. Curr. Med. Chem. 2002, 9, 1187–1199.
(20) Shen, S.; Bremont, B.; Serraz, I.; Andrieux, J.; Poncet, A.; Mathe-
Allainmat, M.; Chanut, E.; Trouvin, J. H.; Langlois, M. Structure-
Activity Relationships for Substrates and Inhibitors of Pineal 5-Hy-
droxytryptamine-N-Acetyltransferase: Preliminary Studies. Eur.
J. Pharmacol. 1996, 307, 133–140.
(21) Ferry, G.; Loynel, A.; Kucharczyk, N.; Bertin, S.; Rodriguez, M.;
Delagrange, P.; Galizzi, J. P.; Jacoby, E.; Volland, J. P.; Lesieur, D.;
Renard, P.; Canet, E.; Fauchere, J. L.; Boutin, J. A. Substrate
Specificity and Inhibition Studies of Human Serotonin N-Acetyltrans-
ferase. J. Biol. Chem. 2000, 275, 8794–8805.
(22) Page, A. I. Enzyme Inhibition. In ComprehensiVe Medicinal Chemistry,
volume 2; Sammes, P. G., Ed.; Pergamon Press: New York, 1990;
pp. 61-87.
(23) Khalil, E. M.; Cole, P. A. A Potent Inhibitor of the Melatonin Rhythm
Enzyme. J. Am. Chem. Soc. 1998, 120, 6195–6196.
ACKNOWLEDGMENT
(24) Khalil, E. M.; De Angelis, J.; Ishii, M.; Cole, P. A. Mechanism-Based
Inhibition of the Melatonin Rhythm Enzyme: Pharmacologic Exploita-
tion of Active Site Functional Plasticity. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 12418–12423.
(25) Kim, C. M.; Cole, P. A. Bisubstrate Ketone Analogues as Serotonin
N-Acetyltransferase Inhibitors. J. Med. Chem. 2001, 44, 2479–2485.
(26) Zheng, W.; Scheibner, K. A.; Ho, A. K.; Cole, P. A. Mechanistic
Studies on the Alkyltransferase Activity of Serotonin N-Acetyltrans-
ferase. Chem. Biol. 2001, 8, 379–389.
This work received financial support from the Laboratoires
Servier. We thank the Centre de Ressources Informatiques
de Haute Normandie (CRIHAN) and the European Com-
munity (FEDER) for the molecular modeling software.
REFERENCES AND NOTES
(27) Ferry, G.; Ubeaud, C.; Mozo, J.; Pean, C.; Hennig, P.; Rodriguez,
M.; Scoul, C.; Bonnaud, A.; Nosjean, O.; Galizzi, J. P.; Delagrange,
P.; Renard, P.; Volland, J. P.; Yous, S.; Lesieur, D.; Boutin, J. A.
New Substrate Analogues of Human Serotonin N-Acetyltransferase
Produce in Situ Specific and Potent Inhibitors. Eur. J. Biochem. 2004,
271, 418–428.
(28) Beaurain, N.; Mesangeau, C.; Chavatte, P.; Ferry, G.; Audinot, V.;
Boutin, J. A.; Delagrange, P.; Bennejean, C.; Yous, S. Design,
Synthesis and in Vitro Evaluation of Novel Derivatives as Serotonin
N-Acetyltransferase Inhibitors. J. Enzyme Inhib. Med. Chem. 2002,
17, 409–414.
(29) Szewczuk, L. M.; Saldanha, S. A.; Ganguly, S.; Bowers, E. M.;
Javoroncov, M.; Karanam, B.; Culhane, J. C.; Holbert, M. A.; Klein,
D. C.; Abagyan, R.; Cole, P. A. De Novo Discovery of Serotonin
N-Acetyltransferase Inhibitors. J. Med. Chem. 2007, 50, 5330–5338.
(30) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic,
J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shaw, D. E.; Shelley,
M.; Perry, J. K.; Francis, P.; Shenkin, P. S. Glide: A New Approach
for Rapid, Accurate Docking and Scoring. 1. Method and Assessment
of Docking Accuracy. J. Med. Chem. 2004, 47, 1739–1749.
(31) Hickman, A. B.; Namboodiri, M. A.; Klein, D. C.; Dyda, F. The
Structural Basis of Ordered Substrate Binding by Serotonin N-
Acetyltransferase: Enzyme Complex at 1.8 Å Resolution with a
Bisubstrate Analog. Cell 1999, 97, 361–369.
(1) Arendt, J. Melatonin and the Mammalian Pineal Gland, Chapman
and Hall: London, 1995.
(2) Korf, H. W.; Schomerus, C.; Stehle, J. H. The Pineal Organ, its
Hormone Melatonin, and the Photoneuroendocrine System. AdV. Anat.,
Embryol. Cell. Biol. 1998, 146, 1–100.
(3) Iuvone, P. M.; Tosini, G.; Pozdeyev, N.; Haque, R.; Klein, D. C.;
Chaurasia, S. S. Circadian Clocks, Clock Networks, Arylalkylamine
N-Acetyltransferase, and Melatonin in the Retina. Prog. Ret. Eye Res.
2005, 24, 433–456.
(4) Djeridane, Y.; Touitou, Y. Melatonin Synthesis in the Rat Harderian
Gland: Age- and Time-Related Effects. Exp. Eye Res. 2001, 72, 487–
492.
(5) Bubenik, G. A. Gastrointestinal Melatonin: Localization, Function,
and Clinical Relevance. Dig. Dis. Sci. 2002, 47, 2336–2348.
(6) Conti, A.; Conconi, S.; Hertens, E.; Skwarło-Sonta, K.; Markowska,
M.; Maestroni, M. Evidence for Melatonin Synthesis in Mouse and
Human Bone Marrow Cells. J. Pineal Res. 2000, 28, 193–202.
(7) Slominski, A.; Tobin, D. J.; Zmijewski, M. A.; Wortsman, J.; Paus,
R. Melatonin in the Skin: Synthesis, Metabolism and Functions. Trends
Endocrinol. Metab. 2007, 19, 17–24.
(8) Klein, D. C.; Auerbach, D. A.; Namboodiri, M. A. A.; Wheler, G. H. T.
Indole Metabolism in the Mammalian Pineal Gland. In The pineal
Gland V 1, Reiter, R. J., Ed.; CRC Press: Boca Raton, FL, 1981;
pp199-227.
(9) Klein, D. C.; Baler, R.; Roseboom, P. H.; Weller, J. L.; Bernard, M.;
Gastel, J. A.; Zatz, M.; Iuvone, P. M.; Cahill, G. M.; Falco´n, J.; Cahill,
G.; Cassone, V. M.; Coon, S. L. The Molecular Basis of the Pineal
Melatonin Rhythm: Regulation of Serotonin N-acetylation. In Hand-
book of BehaVioral State control: Cellular and Molecular Mechanisms,
(32) (a) Cramer, R. D., III; Patterson, D. E.; Bunce, J. D. Comparative
Molecular Field Analysis (CoMFA). 1. Effect of Shape on Binding
of Steroids to Carrier Proteins. J. Am. Chem. Soc. 1988, 110, 5959–
5967. (b) Cramer , R. D.,III; DePriest, S. A.; Patterson, D. E.; Hecht,
P. The Developing Practice of Comparative Molecular Field Analysis.

460 J. Chem. Inf. Model., Vol. 50, No. 3, 2010
LEPAILLEUR ET AL.
In3D QSAR in Drug Design: Recent AdVances, Kubinyi, H., Ed.;
ESCOM: Leiden, The Netherlands 1993; pp443-485. (c) Cramer,
R. D., III; Patterson, D. E.; Bunce, J. D. Recent Advances in
Comparative Molecular Field Analysis (CoMFA). Prog. Clin. Biol.
Res. 1989, 291, 161–165.
(43) Ferry, G.; Loynel, A.; Kucharczyk, N.; Bertin, S.; Rodriguez, M.;
Delagrange, P.; Galizzi, J. P.; Jacoby, E.; Volland, J. P.; Lesieur, D.;
Renard, P.; Canet, E.; Fauchere, J. L.; Boutin, J. A. Substrate
Specificity and Inhibition Studies of Human Serotonin N-Acetyltrans-
ferase. J. Biol. Chem. 2000, 275, 8794–8805.
(44) Bradford, M. M. A Rapid and Sensitive Method for the Quantitation
of Microgram Quantities of Protein Utilizing the Principle of Protein-
dye Binding. Anal. Biochem. 1976, 72, 248–254.
(45) Audinot, V.; Mailliet, F.; Lahaye-Brasseur, C.; Bonnaud, A.; Le Gall,
A.; Amosse´, C.; Dromaint, S.; Rodriguez, M.; Nagel, N.; Galizzi, J. P.;
Malpaux, B.; Guillaumet, G.; Lesieur, D.; Lefoulon, F.; Renard, P.;
Delagrange, P.; Boutin, J. A. New Selective Ligands of Human Cloned
Melatonin MT1 and MT2 Receptors. Naunyn-Schmiedeberg’s Arch.
Pharmacol. 2003, 367, 553–561.
(33) Allen, F. H. The Cambridge Structural Database: a Quarter of a Million
Crystal Structures and Rising. Acta Crystallogr. 2002, B58, 380–388.
(34) Clark, M.; Cramer III, R. D. The Probability of Chance Correlation
Using Partial Least Square (PLS). Quant. Struct.-Act. Relat. 1993, 12,
137–145.
(35) Melville, J. L.; Hirst, J. D. On the Stability of CoMFA Models.
J. Chem. Inf. Comput. Sci. 2004, 44, 1294–1300.
(36) Klebe, G.; Abraham, U.; Mietzner, T. Molecular Similarity Indices
in a Comparative Analysis (CoMSIA) of Drug Molecules to Correlate
and Predict their Biological Activity. J. Med. Chem. 1994, 37, 4130–
4146.
(37) Albers, W. M.; Canters, G. W.; Reedijk, J. Preparation of Extended
Di(4-pyridyl)thiophene Oligomers. Tetrahedron 1995, 51, 3895–3904.
(38) Wu, R.; Schumm, J. S.; Pearson, D. L.; Tour, J. M. Convergent
Synthetic Routes to Orthogonally Fused Conjugated Oligomers
Directed toward Molecular Scale Electronic Device Applications. J.
Org. Chem. 1996, 61, 6906–6921.
(39) Vallgarda, J.; Appelberg, U.; Arvidsson, L.-E.; Hjorth, S.; Svensson,
B. E.; Hacksell, U. Trans-2-Aryl-N,N-dipropylcyclopropylamines:
Synthesis and Interactions with 5-HT1A Receptors. J. Med. Chem. 1996,
39, 1485–1493.
(40) Van den Hoven, B. G.; Alper, H. The First Regioselective Hydro-
formylation of Acetylenic Thiophenes Catalyzed by a Zwitterionic
Rhodium Complex and Triphenyl Phosphite. J. Org. Chem. 1999, 64,
9640–9645.
(41) Zheng, A.; Wang, W.; Zhang, H.; Wang, B. Two New Improved
Approaches to the Synthesis of Coumarin-Based Prodrugs. Tetrahe-
dron 1999, 14, 4237–4254.
(42) Ferry, G.; Mozo, J.; Ubeaud, C.; Berger, S.; Bertrand, M.; Try, A.;
Beauverger, P.; Mesangeau, C.; Delagrange, P.; Boutin, J. A.
Characterization and Regulation of a CHO Cell Line Stably Expressing
Human Serotonin N-Acetyltransferase (EC 2.3.1.87). Cell. Mol. Life
Sci. 2002, 59, 1395–1405.
(46) Petricek, V.; Dusek, M. JANA98; The Crystallographic Computing
System; Institute of Physics: Praha, Czech Republic, 1998.
(47) (a) Sheldrick, G. M. Phase Annealing in SHELX-90: Direct Methods
for Larger Structures. Acta Crystallogr. 1990, A46, 467–473. (b)
Sheldrick, G. M. SHELX-97; Program for the Refinement of Crystal
Structures; University of Go¨ttingen: Go¨ttingen, Germany, 1997.
(48) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.;
Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank.
Nucleic Acids Res. 2000, 28, 235–242.
(49) Sybyl 6.9; Tripos Inc.:St Louis, MO, 2003.
(50) Clark, M.; Cramer, R. D., III; van Opdenbosch, N. Validation of the
General-Purpose Tripos 5.2 Force-Field. J. Comput. Chem. 1989, 10,
982–1012.
(51) Powell, M. J. D. Restart Procedures for the Conjugate Gradient
Method. Math. Program. 1977, 12, 241–254.
(52) Bush, B. L.; Nachbar, R. B., Jr. Sample-Distance Partial Least Squares:
PLS Optimized for Many Variables, with Application to CoMFA.
J. Comput.-Aided Mol. Des 1993, 7, 587–619.
(53) Kubinyi, H.; Abraham, U. Practical Problems in PLS Analyses. In
3D-QSAR in Drug Design-Theory, Methods and Applications, Kubinyi,
H.,Ed.;ESCOMSciencePub:Leiden,TheNetherlands1993;pp717-728.
CI9004805