Combining NMR and molecular modelling in a drug delivery context: Investigation of the multi-mode inclusion of a new NPY-5 antagonist bromobenzenesulfonamide into β-cyclodextrin

Article abstract of DOI:10.1016/j.bmc.2003.10.033

NMR spectroscopic and molecular modelling methods have been employed to describe the complexation of trans-N-4-[N′-(4-chlorobenzoyl) hydrazinocarbonyl]cyclohexylmethyl-4-bromobenzenesulfonamide, a new chemotype of NPY-5 antagonist, and β-cyclodextrin, rev

Full text of DOI:10.1016/j.bmc.2003.10.033

Bioorganic & Medicinal Chemistry 12 (2004) 447–458
Combining NMR and molecular modelling in a drug delivery
context: investigation of the multi-mode inclusion of a new NPY-5
antagonist bromobenzenesulfonamide into ꢀ-cyclodextrin
Gloria Uccello-Barretta,a,* Federica Balzano,a Giuseppe Sicoli,a Carmen Frıglola,b
Ignacio Aldana,b Antonio Monge,b Donatella Paolinoc and Salvatore Guccionec,*
a
`
Dipartimento di Chimica e Chimica Industriale, Universita degli Studi di Pisa, via Risorgimento 35, 56126 Pisa, Italy
bUnidad de I+D de Medicamentos. CIFA. Universidad de Navarra. 31008 Pamplona, Spain
` `
DipartimentodiScienzeFarmaceutiche,UniversitadegliStudidiCatania,V.leA.Doria6,Ed.2CittaUniversitaria,I-95125Catania,Italy
c
Received 25 July 2003; accepted 17 October 2003
Abstract—NMR spectroscopic and molecular modelling methods have been employed to describe the complexation of trans-N-4-
[N0-(4-chlorobenzoyl)hydrazinocarbonyl]cyclohexylmethyl-4-bromobenzenesulfonamide, a new chemotype of NPY-5 antagonist,
and b-cyclodextrin, revealing the coexistence of two different kinds of 1:1 complexes where conformational changes of the guest
compound with respect to the free state are also detected
# 2003 Elsevier Ltd. All rights reserved.
1. Introduction
ability. Poor pharmacokinetic properties, such as low
oral absorption, have been associated with 40% of drug
failures.6 Molecular host-guest based systems are prime
candidates for regulating a drug’s rate release in the
body. A very popular hosts family is constituted by
cyclodextrins (CDs), used to improve solubility and
bioavailability of poorly water-soluble compounds and
a matter of interest to pharmaceutical applications since
these compounds can spontaneously form inclusion
complexes with a variety of guest molecules in aqueous
media.7 This property originates from their unique ‘tor-
oidal’ structure, that is, a basked-type shaped central
cavity of 0.6–1 nm in diameter, which can be described
as a truncated cone, endowed with a polar external sur-
face, responsible for their solubility in aqueous means,
and with an internal apolar cavity, where the displace-
ment of water molecules by hydrophobic guests and,
hence, their inclusion is a process thermodynamically
Neuropeptide Y (NPY), a highly conserved 36 amino
acid polypeptide that is widely distributed throughout
the central and peripheral nervous system, is involved in
the regulation of a variety of neuroendocrine functions.
These effects are mediated through activation of a
family of G-protein coupled receptors (known as Y1–
Y6). There is evidence that NPY-5 receptor may play an
important orexigenic role in feeding response in rodents.
Selective Y5 antagonists could become promising anti-
obesity agents without the cardiovascular side effects1 or
at least useful tools to elucidate the role of NPY-5 sub-
type receptor.1À5
Modern drug design not only focuses on the pharma-
cological activity of a drug like compound but also
considers its ability to be absorbed and to reach its site
of action. Since the peroral route for drug administra-
tion is the most convenient and preferred by patients,
the main objective in the process of drug development is
to obtain a drug product with a good oral bioavail-
favoured.8 The above features are clearly displayed by
9À11
the GRID(version 21)
and HINT 2.35S (shortcut
for Hydropatic Interactions)12,13 maps (Figs 1 and 2),
the latter as implemented in Sybyl 6.9.14 The unexpect-
edly less pronounced hydrophobic contours down the
middle of the CDstructure suggest that the polar influ-
ence of the hydroxyl oxygens extends right towards the
* Corresponding authors. Tel.: +39-050-221-9232; fax: +39-050-221-
9260 (G. U.-B.); tel.: +39-095-738-4020; fax: 39-095-443-604 (S. G.);
e-mail addresses: gub@dcci.unipi.it; guccione@unict.it.
middle of the molecule. As a countercheck some of the
9À11
GRID(version 21)
water contours, also tend to
0968-0896/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2003.10.033

448
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
occur down the middle of the CDstructure, for instance
at an energy of À5 kcal/mol there is still a lot of water
contour.
In view of the relevance of these applications, a detailed
knowledge of the dynamics and host-guest orientation
of inclusion complexes formed by cyclodextrins con-
stitutes the basis for their use as drug carriers, prior to a
possible further development of drugs. In this context
we have exploited NMR spectroscopy and molecular
modelling methods in order to give a defined picture of
the complexation phenomena involving cyclomalto-
heptaose (b-cyclodextrin, b-CD) and the sodium salt of
trans-N-{4-[N0 -(4-chlorobenzoyl)hydrazinocarbonyl]-
cyclohexylmethyl}-4-bromobenzenesulfonamide (G-Na),
being the parent compound G (Fig. 3) a representative
of potent and selective N0 substituted carbohydrazides
acting as hNPY-5 antagonists (in vitro: NPY hY5 Ki
nM 4.30, no affinity for the hNPY Y1 receptor, i.e., Ki
values were higher than 10À5 M), possible drugability of
which is negatively affected because of a poor water
solubility at a pH of 5–7 (<0.1 mg/mL). Preliminary ‘in
vivo’ evaluation (data not shown) showed promising
results.
Cyclodextrin complexes in which aromatic amino acids
are involved are of interest as chiral selectors15,16 and
models for enzyme-substrate specific binding.8,17
There are three basic structures that differ only in the
number of units. The a-cyclodextrin (a-CD), of six units,
the b-cyclodextrin (b-CD) of seven units and the g-
cyclodextrin (g-CD) of eight units connected through a-
1,4 chemical bonds. Complexation processes in solution
depend on the size, shape and hydrophobicity of the
guest molecule and are accompanied by an increase in
8
the strain in the CDcavity. The main forces that have
to do with the complexation processes are van der
Waals interactions between guest and host, responsible
for the stability of complexes in water and in vacuo, and
hydrophobic interactions for the case of aqueous solu-
tions. Other important interactions are hydrogen bond-
ing between the guest and the hydroxyl groups of the
CD, release of strain energy in the molecule ring and
dipole–dipole and/or coulombic interactions.8,17
The quantitative analysis of the dependence of proton
chemical shifts on the drug/cyclodextrin molar ratio has
been used as the basis for defining the complexation
stoichiometry,18 as well as the detection of dipolar
interactions by NOE19 or proton selective relaxation
methods20 has been employed for imposing the con-
formational restraints needed to define the stereo-
chemistry and dynamics of the cyclodextrin complexes
formed in solution.
Conformational in vacuo and aqueous analysis using the
software MacroModel (version. 8.0)21,22 as implemented
in the MAESTRO suite (version. 5.0)22 and high level
quantum chemical calculations (HF-MIDI23,24 and
Jaguar version 4.0 release 2322) were carried out to
complement the NMR results and have a definite view
of the spatial disposition as adopted by compound (G),
showing high binding affinity and considered to char-
acterize the ‘active conformation(s)’ by which the mole-
cule might interact with the cyclodextrin structure. High
level quantum chemical approaches were also used to
better define the conformation of the carbohydrazide
moiety (Scheme 1) checking the soundness of the mole-
cular mechanics performance.22À24
9À11
Figure 1. GRID21
maps of the b-cyclodextrin structure as
retrieved from the Cambridge Structural Database (CSD)32À34 version
5.24, November 2002. Data updates April 2003. CSD code:
BCDEXD03 or BISTAY. The co-crystallized ligand was extracted
before the analysis (see Experimental). (a) blue contours (À4.6 kcal/
mol): polar (water) map (side view); (b) red contours (À0.035 kcal/
mol): hydrophobic probe map (front view). A central tunnel down the
centre of the big toroidal structure is clearly displayed.
A further aspect of the study was motivated by the
analysis of the acceptor character of the
sulfonamide25À31 group then extending to the non stan-
dard CH–O hydrogen bonds. The latter type of hydro-
gen bonds and directional C–H. . .p interactions were
detected in a number of CDinclusion complexes, as also
supported by structural evidences.17 The typical energies
of C–H. . .O and C–H. . .p hydrogen bonds depend on
Figure 2. (a) HINT 2.35S12À14 hydropathicity map of the b-cyclodex-
trin structure as retrieved from the Cambridge Structural Database
(CSD)32À34 (version 5.24, November 2002. Data updates April 2003.
CSDcode: GETPAW). The co-crystallized ligand was extracted
before the analysis (see Experimental). The external hydrophilic (red)
and internal (green) apolar areas are displayed. (b) HINT 2.35S12À14
hydropathicity map of a disordered (not toroidal) b-cyclodextrin
structure obtained by manually building from the fragments as in the
SYBYL 6.8 and 6.9 BIOPOLYMER module)14 and roughly minimiz-
ing to emphasize the double face character of the molecule. The bind-
ing cavity shows very visual hydropathic properties. A guest could
bind in the center, even if it is hydrophobic (green contours).
Figure 3. b-Cyclodextrin (b-CD) and guest (G) structure.

G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
449
the nature of the C–H donor. For the very weakly
polarized methyl donors, C–H. . .O bond energies are
around 0.5 kcal/mol, not much above the energies of
van der Waals interactions. For strongly polarized C–H
groups like in chloroform or in terminal alkynes, C–
H. . .O energies may be larger than 2 kcal/mol. For
the other types of C-H donors, hydrogen bond energies
are in between these extremes. The C-H groups forming
the surface of CDs cavities, C-3-H and C-5-H, are of a
significantly activated type.
ried out at 0 ꢀC, in the presence of 1-hydroxy-
benzotriazole (HOBT) to entrap the intermediate (O-
acylisourea) so hampering the intramolecular re-order-
ing which could lead to the N-acylurea.
1
The H NMR spectrum (Fig. 4) of G-Na (5 mM, D2O)
was assigned by comparing the interproton scalar and
dipolar correlations drawn from the COSY and
ROESY maps, respectively.
The characterization data are reported in Table 1.
Searches of the Cambridge Structural Database
(CSD)32À34 version 5.24, November 2002. (Data updates
April 2003) and the Brookhaven Protein Databank
(PDB)35,36 were merged with Density Functional
Theory (DFT)37À42 results (data not shown).
As a starting point for the assignment of the dis-
ubstituted cyclohexane ring, we used the methylene
protons d of the substituent in 1-position, the reso-
nances of which are well recognizable at 2.40 ppm.
These are J coupled to the proton originating the signal
at 1.11 ppm, which must be e, the scalar correlations of
which lead to the pairs of vicinal protons named as f, at
0.66 and 1.60 ppm. The other proton at 2.03 ppm, inte-
grating for 1H, must be assigned to h, which is J cou-
pled to the two pairs of vicinal protons at 1.20 ppm and
1.65 ppm, indicated as g. Among the f and g couples,
those at 0.66 and 1.20 ppm were attributed to the axial
ones (Fig. 5) as each of them generates dipolar interac-
tion (Fig. 6) on only one pair of protons g and f
respectively adjacent to them. Therefore, the other two
couples at 1.60 and 1.65 ppm are f-equatorial (feq) and
g-equatorial (geq) respectively.
2. Results and discussion
Trans-N-{4-[N0 -(4-chlorobenzoyl)hydrazinocarbonyl]-
cyclohexylmethyl}-4-bromobenzenesulfonamide (G) was
prepared (Scheme 1) by reaction of trans-4-[(4-bromo-
benzenesulfonyl)aminomethyl]cyclohexanecarboxylic
acid (3) with 4-chlorobenzoic hydrazide (4). The inter-
mediate 3 was prepared from 4-bromobenzenesulfonyl
chloride (1) and trans-4-(aminomethyl)cyclohexanecar-
boxylic acid (2). The formation of the sulfonamide 3
was carried out according to the Schotten–Baumann
method in 4 M NaOH. Afterwards, the desired car-
boxylic acid is obtained by a dropwise adding of 35%
HCl. The formation of the corresponding carbohy-
drazide G was carried out by activating the corre-
The relative stereochemistry of the two trans-1,4-sub-
stituents on the saturate ring was unequivocally defined
on the basis of the following NOE interactions (Fig. 6):
the methylene protons d originate NOE effects on both
proton pairs f, axial and equatorial, whereas the proton
e on the same site gives dipolar interactions only with
the protons feq and g-axial (gax), therefore the sub-
stituent, which the d methylene protons belong to, and
the e proton are respectively equatorial and axial as
shown in Figure 5. Therefore, the other substituent
must be equatorial itself; as a matter of fact the proton h
generates dipolar interaction with geq and f-axial (fax),
but no NOE is observed at the frequency of gax (Fig. 6),
leading to the conclusion that h must be axial and,
hence, the substituent on the same carbon atom is
equatorial (Fig. 5).
sponding
dimethylaminopropyl)carbodiimide (EDC). In order to
carboxylic
acid
with
1-ethyl-3-(30-
avoid the formation of N-acylurea, the reaction is car-
A diequatorially (Fig. 7) substituted lowest energy con-
former was also found by HF-MIDI (Hartree–
Fock)23,24 calculations. The amide shows a good delo-
calization of each N lone pair into an anti C–N s*.
There are some soft torsions for the phenyl rings to
rotate, but the phenyl conformation is pointless and it
Scheme 1. Synthetic route to G. The computationally investigated
uncommon bound is squared.
Figure 4. 1H NMR spectrum (600 MHz, D2O, 5 mM, 25 ꢀC) of G-Na
(scheme numbering in Fig. 3).

450
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
can adopt whatever position is required by the host. The
delocalization is indeed into a s* orbital by a kind of
anomeric effect or to be a bit careful ‘generalized anomeric
effect’ since this isn’t oxygen lone pair delocalization into a
C–O s*.43 Each nitrogen is p delocalized within its own
amide, but it is s delocalized to the other. Basing on the
antiperiplanar relationship between the nitrogen lone
pairs in the pyramidal nitrogens and the corresponding C–
N bonds [–C(1)–N(2)–N(3)–C(4)– where C(1) and C(4)
are the carbonyl carbons], the anomeric delocalization is
n[N(3)]–>s*[C(1)–N(2)] and n[N(2)]–>s*[C(4)–N(3)].
Such anomeric delocalization of the nitrogen lone pairs is
often invoked in phosphonamides and heterocycles
including endocyclic amide nitrogens.43,44
aromatic ring was achieved by analysing the 2D
ROESY map of G in its acidic form in DMSO-d6, where
the exchangeable amide protons, which are not detect-
able in D2O, produce three well-resolved resonances at
7.71, 9.81 and 10.35 ppm (Table 1). Of these, the reso-
nance at 7.71 ppm has been assigned to the sulfonamide
NH proton on the basis of its inter-NOE with the
methylene protons d. The two high frequency shifted
protons give dipolar interaction with the highest fre-
quency aromatic proton at 7.86 ppm, which has been
assigned to the aromatic protons named as k (Table 1).
Analogously, the highest frequency aromatic protons at
7.62 ppm in D2O have been assigned to the protons k
and, finally, the protons l are those producing the signal
at 7.25 ppm in D2O. Since no dipolar interactions are
observed between the aromatic protons and the alipha-
tic protons of the cyclohexane ring, the compound G
must assume in solution an ‘extended’ conformation
like that shown in Figure 8a.
Within the aromatic protons, which produces four well
resolved resonances between 7.00 and 8.00 ppm, those
belonging to the p-bromophenyl group were easily
assigned as the methylene protons d originate a weak
but observable NOE at 7.45 ppm (Fig. 6), which can be
attributed to the proton pair b adjacent to the sulfona-
mide function. These last are J coupled to the protons a,
at 7.51 ppm, which are in ortho position to the bromine.
The assignment of the aromatic protons of the other
In the presence of b-cyclodextrin an about three fold
increase of the solubility of G-Na is achieved allowing
us to analyse a 12 mM equimolar mixture ꢀ-CD/G-Na
1
and in the corresponding H NMR spectrum chemical
shift variations (Table 1) are measured, which are
markedly superior for the methylene protons d and for
the cyclohexane ring nuclei, feq and geq in particular.
In order to ascertain whether these variations are
simply due to complexation effects or to conformational
transitions as a result of complex formation, we ana-
lysed the 2DROESY maps of the mixture at mixing
times variable from 0.1 s to 0.8 s. The intramolecular
NOEs detected in the mixture are similar but not equal
with respect to those found for the free G-Na, as the
aromatic protons b of the p-bromophenyl moiety show
dipolar interactions with the cyclohexane ring protons e
and f. Since the same aromatic protons do not originate
any NOE on the protons g, we must hypothesize that a
folding of the chain occur as a consequence of the
Figure 5. Cyclohexane conformational arrangement.
Table 1. 1H NMR chemical shift data (d, ppm referenced to TMS as
external standard) of G in DMSO-d6 and G-Na in D2O in the pure
form (5 mM) and in the equimolar mixture (12 mM, D2O) with b-
cyclodextrin (ꢀ-CD). Complexation shifts (Ád=dmixtureÀdfree, ppm)
for the G-Na protons in the presence of equimolar amount of ꢀ-CD
Proton
Ga
d (ppm)
G-Nab
d (ppm)
G-Na/b-CDb
ÁdÂ102
m
m
d (ppm)
m
a
b
c
d
7.81
7.71
7.71
2.59
1.30
1.73
0.86
1.76
1.31
2.15
10.35
9.81
7.86
7.56
d
d
s
7.51
7.45
d
d
7.52
7.44
d
d
6
À6
Figure 6. 2DROESY map (600 MHz, D 2O, 5 mM, mix 0.4 s, 25 ꢀC)
of G-Na: aliphatic region.
br. s
br. s
br. d
br. q
br. d
br. q
br. t
br. s
br. s
d
2.40
1.11
1.60
0.66
1.65
1.20
2.03
d
2.31
1.14
1.67
0.64
1.75
1.25
2.03
m
br. s
m
À54
e
br. s
br. d
br. q
br. d
br. q
br. t
18
42
feq
fax
geq
gax
h
i
j
k
l
m
À12
br. d
br. q
br. t
60
30
0
7.62
7.25
d
d
7.64
7.25
d
d
12
0
d
Figure 7. Most stable (orthographic view) diequatorially substituted
conformer as calculated by the MIDI Hartree–Fock (HF). See text for
explanation.
a 300 MHz.
b 600 MHz.

G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
451
interaction with the cyclodextrin, bringing the p-bro-
mophenyl ring approximately perpendicular to the
cyclohexane plane. Furthermore, the inter-NOE e–gax is
more intense than it is for the pure compound (com-
pared to the other intramolecular effects produced by
the proton e) to indicate a ring distortion consequent
the chain folding (see Fig. 8b for a picture of G-Na in
the complexed state). Dipolar interactions (Fig. 9) are
observed among all G-Na protons and the internal pro-
Although some slight differences in the shape can arise
from the two combined approaches the folded more
stable complexed conformer (Fig. 8b) as found by
NMR is conceptually in good agreement with the com-
putational result ‘in vacuo’ mimicking the apolar CD’s
cavity. Contrary to the solution NMR (Fig. 8a) where
no possibility of overlapping by p-clouds interactions of
the two aromatic seems to be, a substantially folded, a
slightly solvation energy stretched structure, was also
found in the aqueous conformational (Fig. 10b) analysis
by the software MACROMODEL (version 8.0).22 Pro-
vided that false minima might come out from the cal-
culation itself, the two computationally detected folded
conformations (Fig. 10a and b), the substantial differ-
ence of which lies in the spatial location of the freely
rotating p-chlorophenyl group, may represent two
active forms fitting the hydrophobic cavity of the car-
bohydrate ‘receptor’, that is, allowing maximum contact
between the hydrophobic part of the guest molecule and
the internal surface of the CDcavity, and finally leading
to one of the two complexes detected by the NMR
investigation (see below).8,17 Interestingly a higher
energy (À34.92 KJ/mol vs À36.74 KJ/mol) conformer
aligning very well with that linear detected by NMR was
also found by the software MACROMODEL (version
8.0).21,22 The partial fitting of the linear form (Figs 8a
and 10c) in the cyclodextrin cavity further supports the
above computational result and the proposed dynamic
view of the interaction by a reverse switch on mechanism
(from an inactive unfolded to an active folded) leading
to the folded conformer as that active.45,46 Both the
folded conformations might be stabilized by p clouds
interactions between the two aromatic rings.
0
0
tons H3 and H5 of the cyclodextrin, respectively located
on its wider and narrower internal surface, according to
the formation of inclusion complexes. In our case the
0
relative intensities of the inter-NOEs produced on H3
0
and H5 , which well reflect the orientation of G-Na into
the macrocycle, cannot be explained by means of one
kind of inclusion complex only, but different complexes
or complexes having stoichiometry different from the
one to one must be present.
Figure 8. Conformational arrangement of G-Na in D2O solution in
the pure (a) and complexed (b) form.
According to the folded structure having the p-bromo-
phenyl group faced to the saturate ring, the methylene
protons d, the ring proton e and the aromatic protons b,
which are in close proximity each other in the folded
Figure 10. (a) In vacuo lowest energy conformer of compound G as
obtained by Macromodel 8.0;21,22 (b) superposition of the in vacuo
(wire) and the aqueous (capped sticks) lowest energy conformers; (c)
higher energy (À34.92 KJ/mol) aqueous conformer as obtained by
Macromodel 8.0.21,22 See the text for further details.
Figure 9. 2DROESY (600 MHz, D O, 12 mM, mix 0.4 s, 25 ꢀC) ana-
lysis of G-Na/ꢀ-CD (1:1). Spectral region including the cyclodextrin
protons of the traces corresponding to G protons.
2

452
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
structure (Fig. 8b), produce NOEs only on the protons
imum at the molar fraction of 0.5 corresponding to a 1:1
stoichiometry (Fig. 12). Therefore, there is simultaneous
presence of two different 1:1 complexed forms of G-Na.
The slight asymmetries of the curves are probably due
to the fact that the chemical shifts variations are origi-
nated by the formation of the two different complexes
of Figure 11, each having 1:1 stoichiometry, but with
different stabilities.
0
0
H5 and H6 (Fig. 9) respectively located on the narrower
internal and external surface, whereas the protons a,
which are on the same aromatic moiety, and the protons
fax and feq also determine a dipolar interaction with the
0
internal protons H3 at the wider part of the truncated
cone (Fig. 9). Accordingly, for the cyclohexane protons
0
gax or geq and the proton h the NOE effect at the H3
frequency is more intense with respect to that one
0
observed at the H5 frequency (Fig. 9). Therefore, the
As the formation of 1:1 against 1:2 G-Na/ꢀ-CD com-
plexes should affect the dynamic features of the cyclo-
dextrin severely, we determined the interproton cross-
relaxation rates sij of ꢀ-CD to compare its reorienta-
tional correlation times in the presence of G-Na or in
the free state. In the initial rate approximation,48 this
parameter, which describes the magnetization transfer
between the spins i and j, depends on the internuclear
distance rij and on the reorientational correlation time tc
of the vector ij:
G-Na substrate, in its folded structure, is included into
the cyclodextrin as shown in Figure 11a, once again in
agreement with the conformational analysis which
shows a folded conformation (Fig. 10a and b) underlying
a backwards mechanism to that commonly assumed for
the active forms of proteins45 and GPCR (G-Protein
Coupled Receptor Ligands) ligands,46 that is, a reverse
switching on mechanism.46
It might be further speculated on the CDs as receptor
mimic to profile the possible ligand changes or the
mutual ligand–receptor induced fit before binding.45À47
ꢀ
ꢁ
6
2
4
ꢀij ¼ 0:1ꢁ h rÀij 6ꢂc
À 1
ð1Þ
À
1 þ 4!2ꢂc2
On the other hand, the aromatic protons k and l of the
p-chlorophenyl moiety produce comparable NOEs on
0
0
the protons H3 and H5 (Fig. 9) and hence are them-
selves completely included into another unit of cyclo-
dextrin in the same complex or in a different complexed
form. Taking into account that the proton h, which can
be included into both cyclodextrin units, shows a largely
where g is the gyromagnetic ratio, o is the proton Lar-
mor frequency, and h is the reduced Planck’s constant.
À
For a proton pair having a near fixed distance rij, in a
molecule moving in the extreme narrowing limits
(o2t2c«1), the cross-relaxation rate becomes a simple
increasing function of the reorientational correlation
time:
0
prevalent interNOE with the protons H3 (Fig. 9), the
part of the molecule bearing the p-chlorophenyl ring is
included into the cyclodextrin from its larger rim (see
Fig. 11b).
The presence of a 1 to 2 complex G-Na/ꢀ-CD can be
ruled out on the basis of the determination of the com-
plexation stoichiometry and of the proton selective
relaxation measurements.
2
4
ꢀij ¼ 0:5ꢁ h rÀij 6ꢂc
ð2Þ
À
By using the continuous variation method (Job’s
method)18 applied to different protons of G-Na in mix-
tures with the cyclodextrin having the same total con-
centration but different molar ratios, the corresponding
Job plots are nearly symmetrical curves with a max-
Whereas, when the motion slows down to the o2tc2»1
region, sij shows still a linear dependence on tc but it is
negative:
Figure 12. Job plot for G-Na/ꢀ-CD mixture. Protons geq of G-Na was
used as the diagnostic protons, and the total concentration was 5 mM.
Figure 11. Conformational arrangement of the two 1:1 G-Na/ꢀ-CD
inclusion complexes as derived by NOE analysis.

G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
453
2
4
ꢀij ¼ À0:1ꢁ h rÀij 6ꢂc
ð3Þ
The nitrogen lone pairs which may be delocalized into
the amide carbonyl and do not have to be planar to
conjugate are also somewhat delocalized into adjacent
C–N antibonding s* orbitals being available to serve as
weak H-bond acceptors in an active site. Those nitrogen
atoms would be expected to be rather flexible about
their chemical behavior.
À
Hence, sij can be employed as a probe of the dynamic
changes, which occur in a molecule as a consequence of
the aggregation to another molecule. Among the avail-
able experimental methods to determine sij, we selected
the one based on the determination of proton mono-
(Ri) and biselective (Riij) relaxation rates:
The formation of the other complex brings about a
change of the conformation of G-Na from the linear
structure of Figure 8a to the folded one having the p-
bromophenyl plane bent at the cyclohexane ring and
perpendicular to it (Fig. 8b).45,46 Probably this kind of
folding allows the simultaneous deep penetration of the
apolar saturated and aromatic rings into the cyclodex-
trin, indicating the relevance of the van der Waals and
hydrophobic interactions that are mainly dependent on
such structural features. These interactions are the main
cause for the deep penetration of benzene derivatives
into the cavity of the CDthat is mainly apolar and
hydrophobic. The sulfonamide group protruding from
the narrower part of the cavity, works as an auxiliary
hydrogen bonds acceptor to stabilize the complex by
attractive interactions with the primary hydroxyls
(Fig. 11a).17,25À31
ꢀij ¼ Riij À Ri
ð4Þ
Monoselective relaxation rates are measured by follow-
ing the recovery of the signal of the spin i selectively
inverted, by leaving unperturbed all the other spins in
the molecule; the biselective relaxation rates are deter-
mined under conditions of simultaneous inversion of the
proton pair ij. Therefore, we measured the mono-
0
selective relaxation rate of the proton H2 of the cyclo-
dextrin as 1.04 sÀ1 in the pure compound, slightly
increasing to 1.23 sÀ1 in the mixture with G-Na (Table
2). The biselective relaxation rate of the same proton
200
0
under simultaneous inversion of the proton H1 (R2 10
)
underwent a similar slight increase from 0.89 sÀ1 (pure
cyclodextrin) to 1.07 sÀ1 (mixture). From these values
we calculated the cross-relaxation rates for the proton
pair 2010 of À0.15 sÀ1 and À0.16 sÀ1 in the pure and
complexed cyclodextrin, respectively. On the hypothesis
In bioorganic chemistry, the effectiveness of sulfonyl
and sulfonamide oxygens as good hydrogen-bond
acceptors likewise carbonyl or hydroxyl oxygens is fre-
quently questioned. The relevant properties can be
directly determined or inferred from experimental
structural data.32À34 Further structural comparisons,
and direct energetic comparisons, are possible with high
level quantum chemical approaches.25À31,37À42 Results
from an extensive Cambridge Structural Databases
(CSD)32À34 (Fig. 13) and Brookhaven Protein Databank
(PDB) entries35,36,49,50 support the role of the sulfone
group as an acceptor in biological complexes. From a
0
0
that the intramolecular distance r2 1 is not affected by
the complexation significantly, the ratio between the
two above cross-relaxation terms should reflect the ratio
between the reorientational correlation times of the
proton pair 2010 in the mixture and in the pure com-
pound. In our case this ratio is very close to 1 and hence
only 1 to 1 complexes can be present in solution. As a
matter of fact, as previously demonstrated,49 on forma-
tion of an inclusion complex containing two units of
cyclodextrin, the molecular motion of the macrocycle
should slow down remarkably.
total of 2316 hits (1261 inter- and 1135 intramolecular) as
32À34
Therefore, we can conclude that the complexation pro-
cess of G-Na by the cyclodextrin, responsible for the
observed solubility increase, involves the formation of
two different 1 to 1 inclusion complexes, both having
the drug included from the wide rim of the macrocycle
(Fig. 11). In one complex, the p-chlorophenyl group
penetrates into the cyclodextrin, probably due to the
attractive driving ‘hydrophobic’12,13 interaction between
the apolar cavity (Figs 1b and 2a and b) and the aro-
matic moiety assisted by the formation of hydrogen
bonds among the NH–CO groups of the substrate and
the secondary hydroxyls of the cyclodextrin (Fig. 11b).
Generally speaking, polar atoms on the surface of a
molecule contribute to its selectivity, and hydrophobic
atoms give it its affinity.
found in the CSDdatabase,
1346 (58%) are repre-
sented from unique (not duplicates) hits with a percen-
tage of 39.96 and 60.04 for intra- and intermolecular
0
200
Table 2. Mono- (R2 , sÀ1), biselective (R , sÀ1), cross-relaxation
2 10
(s2 1 , sÀ1) rates and reorientational time (tc, s) for the 2010 protons of
b-cyclodextrin in D2O in the pure and complexed state
0
0
0
200
R2
R
s2 1
tc (10À10
)
0
0
2 10
Figure 13. Search criteria in the Cambridge Structural Database using
the ‘Non Bonded Contact Searches’ option (version 5.24, November
2002 release; data uptodates April 2003).
b-CD1.04
G/b-CD1.23
0.89
1.07
À0.15
À0.16
4.85
5.17

454
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
hydrogen bonds, respectively. A percentage of approxi-
mately 20% hydrogen bonds of all hits was guessed as
directional.
FP82 hot stage apparatus equipped with a FP800/FP80
processor and an Olympus 8091 microscope provided
with a video system and were uncorrected. Micro-
analysis of vacuum-dried samples were obtained on a
Carlo Erba 1106 elemental analyzer (over P2O5 at 1–2
mmHg, 24 h at 60–80 ꢀC). Results are within 0.4% of
theoretical values unless otherwise indicated.
Density Functional Theory (DFT) calculations using
the B3LYP methodology37À42 provide useful compar-
isons with the carboxyl group to further support this
role (data not shown). On the other hand this complex
can be also stabilized by the hydrogen bond interactions
between the NH–CO groups protruding from the wide
rim of the cyclodextrin and the secondary hydroxyls of
the macrocycle as above discussed. Hydrogen bonds can
also be formed with the primary O-6 hydroxyl groups
which are placed at the narrower of the two cavity
openings. The interglycosidic O-4 atoms are sterically
poorly accessible, and serve only occasionally as accep-
tors of O–H. . .O hydrogen bonds. O/N–H. . .O hydro-
gen bonds with either hydroxyl groups or O-4 atoms are
often regarded as the only possible host-guest hydrogen
bonds in CDinclusion complexes. If polar guest mole-
cules are included in the CDs cavities, they have limited
opportunity to satisfy their hydrogen bond potentials.
Typically, guest molecules carrying hydroxyl groups are
oriented in such a way that hydrogen bonds can be
formed through the cavity openings to neighbouring
1
The H and 13C chemical shift data directly concerned
with the subject study of G in DMSO-d6 were further
obtained using a Varian VXR300 spectrometer operat-
1
ing at 300 MHz and 75 MHz for H and 13C, respec-
1
tively. H and 13C NMR chemical shifts are referenced
to TMS as external standard. All NMR spectra in D2O
were recorded on a Varian INOVA600 spectrometer
1
operating at 600 MHz for H using a 5-mm broadband
inverse probe with z-axis gradient. The sample tem-
perature was maintained at 25 ꢀC. The 2DNMR spec-
tra were obtained by using standard sequences. Proton
gCOSY 2Dspectra were recorded in the absolute mode
acquiring eight scans with a 3-s relaxation delay
between acquisitions for each of 512 FIDs. The ROESY
(Rotating-frame Overhauser Enhancement Spectroscopy)
spectra were recorded in the phase-sensitive mode, by
employing a mixing time ranging from 0.1 to 0.8 s. The
spectral width used was the minimum required in both
dimensions. The pulse delay was maintained at 8 s; 512
hypercomplex increments of eight scans and 2K data
points each were collected. The data matrix was zero-
filled to 2KÂ1K and a Gaussian function was applied
for processing in both dimensions. The selective relaxa-
tion rates were measured in the initial rate approxima-
tion48 by employing a selective p pulse at the selected
frequency. After the delay t, a non-selective p/2 pulse
was employed to detect the longitudinal magnetization.
For the biselective measurements, the two protons were
inverted consecutively. Each selective relaxation rate
experiment was repeated at least four times.
17
CDor crystal water molecules.
A survey of the Brookhaven Protein Databank (PDB)
for donor, C–H. . .X interactions revealed more than
500 high-resolution structures with a 30% of kinase
aromatic ligands complexes.36,50,51 Arrangements invol-
ving weak hydrogen bonds are more prone to disorder
than those stabilized by conventional hydrogen bonds:
the entropy gain due to disorder can easily exceed the
enthalpy loss due to breaking the weak hydrogen
bonds.17
Overall the depicted dynamic process of binding
between the b-cyclodextrin and the guest molecule
results from a subtle interplay between steric conditions,
the hydration of the free species and the complexes, van
der Waals interactions, dipole–dipole interactions,
hydrogen bonding, and mutual conformational changes
of the interacting molecules. The complexation of the b-
cyclodextrins with benzene type derivatives, which
penetrate deeply into the apolar and hydrophobic cavity
of the host, is controlled by topological and topographic
parameters indicating the relevance of the van der
Waals and hydrophobic interactions.17
All the conformational and modelling studies were car-
ried out on a SGI-R5000 and-OCTANE workstations
operating under IRIX 6.5.+ using the software Macro-
model (version 8.0)21,22 as implemented in the version
5.0 of the MAESTRO suite.21,22 The G structure was
built by the Macromodel fragment library. Default
parameters were used except of a lower convergence
criteria (0.001). Density Functional Theory calculations
(data not shown)37,42 and CSD(Cambridge Structural
Database) searches by the software CONQUEST (ver-
sion 1.5)32À34 were run on a PC working under WIN-
DOWS XP. The same double boot machine was also
used to run the GRID21 software9À11 under the LINUX
operating system.
3. Experimental
3.1. General methods
The cyclodextrin structure was retrieved from the Cam-
bridge Structural Database (CSD)32À34 (version 5.24,
November 2002. Data updates April 2003. Codes:
BCDEXD03 or BISTAY and GETPAW for the
GRID219À11and HINT mappings, respectively12À14). All
the structures were at a resolution lower than 3.2 A. The
CSDfile was exported to .pdb and read in SYBYL 6.8
or 6.9 to carefully check the coordinates and extract the
co-crystallized ligand as well as to delete the water
All the new compounds were characterized by elemental
analysis and 1H NMR. The 1H NMR spectrum of 3 was
obtained on a Bruker Model AC-200E (200 MHz) with
Me4Si as the internal standard. Thin layer chromato-
graphy was performed on aluminum sheets precoated
with silica gel (HF 254, Merck). The developed chro-
matograms were viewed under UV light or iodine reve-
lation. Melting points were determined on a Mettler

G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
455
molecules before the GRID21 (UNIX or LINUX
version)9À11 (Fig. 1) and HINT 2.35S (Sybyl 6.9
version)12À14 (Fig. 2) analyses. The atom potential types
bond orders were carefully checked to evaluate their
correctness with respect to the intended structure. The
lack of bond order records in the PDB format necessi-
tates that this step be diligently performed before add-
ing possible hydrogen atoms to the ligand structure.
Because hydrogen atoms are only (very) rarely located
in biomacromolecular crystallographic studies, there is
little experimental guidance as to their actual positions.
Also worth noting is that the automated procedures to
add hydrogens to biopolymers and small molecules, as
in Sybyl or other software programs, often (and ran-
domly) orient the hydrogen-bonding hydrogen atoms
away from their intermolecular acceptor atoms.13
MMFFs, where the ‘s’ is for ‘static’.58 With the obvious
exception of inversion barriers, the two force fields give
similar and, when delocalized trigonal nitrogens are not
involved, identical relative conformational energies.
MMFF gives slightly pyramidal amide geometries and
rather strongly pyramidal aniline, nucleic-acid base,
and other enamine nitrogens.58 Experimental and theo-
retical studies indicate that this reflects the true ground-
state geometry.
In contrast, MMFFs58 uses modified torsion and out-
of-plane bending parameters for amide or ‘amide like’
nitrogens, giving rise to planar or nearly planar geome-
tries. Such nitrogens appear planar in crystal structures,
due to time averaging of the rapidly inverting pyramidal
geometries.
9À11
The default parameters of the GRID21
and HINT
Regular MMFF is probably most useful for dynamics
studies (where actual time averaging yields planar aver-
age structures), whereas the MMFFs is probably most
useful for conformational search and for minimizations
which are to be compared with crystallographic data.52À58
2.35S (Sybyl 6.9 implementation)12À14 force fields were
used.
Conformational analyses were performed using the
Monte Carlo Multiple Minimum (MCMM) Search
protocol as implemented in the MacroModel soft-
ware.21,22 In the Monte Carlo approach, the dynamic of
a molecule is simulated by randomly changing dihedral
angle rotations or atom positions. Then, the trial con-
formation is accepted if its energy has decreased from
the previous one. If the energy is higher, there are var-
ious criteria to select or not the calculated conformer. In
our simulations, all of the dihedral angles of single lin-
ear bonds were allowed to move freely, and the con-
formation was accepted if the energy was lower than
that of the previous conformation or within a fixed
energy window (5 kJ/mol) as selector. Prior to submit-
ting the ligands to the search protocol, a minimization
was carried out using the MMFF (Merck Molecular
Force Field) force-field as implemented in Macromodel
8.021,22 with the GB/SA continuum water.21,22 Generally
all the minimizations were performed as above reported.
Default options were used with the Polak-Ribiere
Conjugate Gradient (PRCG) scheme, until a gradient of
0.001 kcal AÀ1 was reached.
The MMFF52À58 gives a reasonable account of the
degree of pyramidalization in compound G and should
be adequate for the intended application according to
preliminary calculations (DFT B3LYP methodology
with the 6-31G** basis set) by JAGUAR (version 4.0
release 23)22 on a simplified model (Table 3). Optimiza-
tion of all degrees of freedom was performed in the
DFT calculations, continuing until the standard gra-
dient and energy change criteria were achieved.
Amide nitrogen geometries are notoriously hard to get
right. Most are pyramidal (have some sp3 character), in
spite of organic prejudice (a bit of a myth) to the con-
trary that amides are planar.43 Although, there is some
precedent for treating them as planar in molecular
mechanics studies very few turn out to really be planar.
That causes the N–H bond dipoles to oppose (trans
conformation) one another while at the same time
aligning with the other carbonyl. The pyramidal N–Hs
tend to have very low barriers to inversion, so amides
are a bit like proteins (it is not very meaningful to think
of them as having a single ‘static’ structure), but instead
they can adopt a range of structures all of which are
quite low in energy. Thus, while the calculation says the
bottom of the energy well corresponds to pyramidal
nitrogen, it would probably not cost much to become
planar. The hydrazine dicarbonyl, prefer to be pyr-
amidal having a six-atom p system with eight electrons
To search the conformational space, 5000 MC steps
were performed on each starting conformation. Least
squares superposition of all non-hydrogen atoms was
used to eliminate duplicate conformations. Considering
of the flexibility of the ligands under investigation, an
energy cut-off of 5.0 kJ molÀ1, high enough to map the
conformational space including the bioactive con-
formation, was applied to the search results. The 21
lowest energy conformers in vacuo and the eight in
water within 2 kJ/mol were further optimized using the
multiple minimization protocol as implemented in soft-
ware MACROMODEL (version 8.0).21,22
Table 3. MMFF versus DFT-B3LYP comparison
The MMFF as in the software MACROMODEL (ver-
sion 8.0)21,22 was used for the conformational in vacuo
(assumed as apolar medium) and aqueous analysis.
Atom sequence
MMFF
B3LYP/6-31G**
C1–N2–N3–H5
C4–N3–N2–H6
N2–N3 distance (A)
147.5
148.8
1.410
161.6
155.5
1.390
There are actually two MMFF implemented in Macro-
model,21,22 namely MMFF (the default)52À58 and

456
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
in it (two from each C¼O plus two nitrogen lone pairs)
which is to some extent like to the not so stable hexa-
triene dianion. Caution must be exerted in modeling
G(5)-type compounds. The sulfonamide group is part of
the problem, as is the C(=O)–N–N–C(=O) group.
Neither of these is well parameterized in any force field.
A further ‘problem’ is that the pyramidal nitrogen
atoms are stereogenic being they either R or S. Unfor-
tunately, because the inversion barrier is low, they can
be either, so a large number of conformations can exist.
their steric, electrostatic and hydrogen-bonding proper-
ties and by their hybridization.9À11
The hydrophobic probe is attracted by dispersion and
induction forces, but avoids polar atoms like hydroxyl
or amine or carboxyl. An important allowance is also
made for the entropic component of hydrophobicity. It
must be noted that polar probes give quite large ener-
gies over quite small regions, whether the hydrophobic
probe gives weaker energies over big regions of the
molecular surface.9À11 Generally speaking, polar atoms
on the surface of a molecule contribute to its selectivity,
and hydrophobic atoms give it its affinity (see below).
Actually the MMFF was designed to work best with
drug-like molecules and it has good performance with
the GB/SA model.52À58
The basis of the HINT model is that quantitatively sig-
nificant data of biomolecular association are encoded in
the experimental determination of hydrophobicity, par-
ticularly from the water–octanol system (LogP). The
constant, LogPo/w, is a thermodynamic quantity repre-
senting the free energy of solvent transfer for partition-
ing between the two solvents. As such, it includes the
effects of entropy, solvation, and enthalpic terms such
as hydrogen bonding, Coulombic attractions, and
hydrophobic attractions. It is an intuitive model because
the components of the HINT score are directly related
with the type of interactions present between different
molecules, and the magnitude of the score is indicative
of the strength of the potential interaction. Because
LogP can be directly correlated with free energy, the
HINT score reveals information not only about
enthalpy but also about entropy. This is because a large
proportion of entropy in the biological environment
resulting from biomolecular associations arises from the
transfer of the solute (ligand) from the solvent (water)
to bound position. Concomitant with that process are
the transfer proteins that bind ligands with different
structures and polarity.12À14
CONQUEST version 1.532À34 was used to search the
Cambridge Structural database (CSD) (version 5.24,
November 2002. Data updates April 2003)32À34using the
‘Non bonded contact searches’ option, criteria of which
are shown in Fig. 13. Density Functional Theory calcu-
lations (DFT) were carried out using the B3LYP meth-
odology as implemented in GAUSSIAN9837À42 (data
not shown).
Complementary information were obtained by the soft-
9À11
wares GRID(version 21),
HINT 2.35S (Sybyl 6.9
implementation),12À14 with the additional aim to test the
last code implementation of the latter software to ana-
lyze carbohydrate molecules The capability of the
HINT 2.35S12,13 software as implemented in SYBYL
6.914 must be noted in correctly mapping the cyclodex-
trin structure in accordance with the GRID219À11
software.
9,11
The GRIDmethodology approach
gives a large
volume of good attractive interactions, the so-called
GRID-map, between the molecule and the probe. In
the program Grid, a three-dimensional grid surrounds
the target molecule. Version 21 of GRIDworks like
previous versions by computing the interaction energy
of a probe at every Grid point on an orthogonal matrix
of points around the molecule of interest. The interac-
tion energy between a probe and each atom of the target
was then calculated for each Grid point, and this calcu-
lation generated one Grid map. By contouring the Grid
map at various energy levels, one can display favorable
interaction areas between the molecule and the probe as
in Figure 1.
A highly exhaustive study on both the HINT paradigm,
its user guidance rules and potential was recently
reported by Cozzini et al.13
3.2. Materials
All starting materials were either commercially available
or reported previously in the literature unless noted. b-
cyclodextrin was purchased from Fluka.
Synthesis of trans-4-[(4-bromobenzenesulfonyl)amino-
methyl]cyclohexanecarboxylic acid (3). A solution of 4-
bromobenzenesulfonyl chloride 1 (6.5 g, 25.44 mmol)
was added, alternately and portion wise to a solution of
trans-4-(aminomethyl)cyclohexanecarboxylic acid 2 (4.0
g, 25.44 mmol) in NaOH 4 M (15 mL), with stirring for
30 min. After the addition, the mixture was stirred for
24 h, at room temperature. The aqueous layer was
washed with CH2Cl2 (3Â20 mL), and acidified with HCl
(c) to pH 1–2. The residue obtained was filtered and
washed with H2O (5Â20 mL) and n-hexane (5Â20 mL)
in order to obtain 3 as a white solid (% Yield: 21; mp,
There is a GRID‘flexibility’ option in version 21 which
allows the possible rotable chains in the test molecule to
move in response to the probe. They move toward the
probe when there is attraction and away from it when
there is repulsion, thus simulating the response of the
target to a change in its environment. This flexibility
option was turned off in this study being no big differ-
ences in the map when the option was turned on as a
consequence of the basket type shape structure of the
cyclodextrins. GRIDprobes are chemical groups such
as methyl, aliphatic hydroxyl, NH+3 amine, water and
divalent cations and the interaction energy at each Grid
point is calculated between the chosen probe and every
atom of the molecule. The probes are characterized by
ꢀC: 183–184). H NMR (DMSO-d6, 200 MHz) d 1.15
1
(m, 4H), 1.75 (m, 5H), 2.05 (t, 1H, J=1.4 Hz), 2.55 (s,
2H), 7.68 (m, 3H), 7.78 (d, 2H, J=8.2 Hz) ppm.

G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
457
C14H18BrNO4S: calcd C, 44.69; H, 4.82; N, 3.72.
Found: C, 44.51; H, 4.83; N, 3.58.
(Department of Medicinal Chemistry & Institute for
Structural Biology and Drug Discovery, Virginia Com-
monwealth University, Richmond, Va Usa) have pro-
vided valuable discussion on the softwares GRID21 and
HINT2.35S (Sybyl 6.9 version), respectively. Schro-
dinger, Inc.22 is gratefully acknowledged for the soft-
ware Jaguar v.4.0 release 23. Toughtful advices by Prof.
A. Corsaro (Department of Chemistry, University of
Catania, Italy) and the support service of the Cam-
bridge Structural Database (CSD) were appreciated.
This work was supported by the Ministero della Ricerca
Scientifica e Tecnologica (MURST) and CNR, Italy.
3.2.1. Synthesis of trans-N-{4-[N0-(4-chlorobenzoyl)hy-
drazinocarbonyl]cyclohexylmethyl} - 4 - bromobenzenesul-
fonamide (G). A mixture of 3 (0.8 g, 2.00 mmol) and
HOBT (0.3 g, 29 mmol) in dry CH2Cl2 (25 mL) at 0 ꢀC,
under N2, was treated with EDC (0.4 g, 2.29 mmol).
After 1 h at 0 ꢀC, 4-chlorobenzoic hydrazide 4 (0.4 g,
2.27 mmol) was added. The reaction was stirred at room
temperature for 24 h. The solvent was evaporated and
the residue was taken up with ethyl ether (20 mL) and
water (20 mL). The obtained precipitate was filtered and
washed with ethyl ether (5Â20 mL) and n-hexane (5Â20
mL) in order to obtain G as a white solid (yield: 68%;
References and notes
ꢀ
1
mp 272 C). H NMR (DMSO-d6, 300 MHz) chemical
shift data are reported in Table 1. 13C NMR (300 MHz,
DMSO-d6, 25 ꢀC) d 28.6, 29.3, 36.8, 42.1, 48.7, 126.1,
128.6, 128.7, 129.4, 131.4, 132.3, 136.7, 140.1, 164.7,
174.7. C21H23BrClN3O4S: calcd C, 47.69; H, 4.38; N,
7.95. Found: C, 47.78; H, 4.45; N, 8.05.
1. Inui, A. Trends Pharmacol. Sci. 1999, 20, 43.
2. Stamford, A. W.; Parker, E. M. Annu. Rep. Med. Chem.
1999, 34, 31.
3. Criscione, L.; Rigollier, P.; Batzl-Hartmann, C.; Rueger,
H.; Stricker-Krongrad, A.; Wyss, P.; Brunner, L.; White-
bread, S.; Yamaguchi, Y.; Gerald, C.; Heurich, R. O.;
Walker, M. W.; Chiesi, M.; Schilling, W.; Hofbauer,
K. G.; Levens, N. J. Clin. Invest. 1998, 102, 2136.
4. Youngman, M. A.; McNally, J. J.; Lovenberg, T. W.;
Reitz, A. B.; Willard, N. M.; Nepomuceno, D. H.; Wil-
son, S. J.; Crooke, J. J.; Rosenthal, D.; Vaidya, A. H.;
Dax, S. L. J. Med. Chem. 2000, 43, 346.
5. Nagaaki, S.; Toshiyuki, T.; Takunobu, S.; Yuji, H.; Aya,
S.; Masaaki, H.; Miki, S.; Katsumasa, N.; Yuko, K.;
Hidefumi, K.; Naoko, F.; Yasuyuki, I.; Akane, I.; Akio,
K.; Takehiro, F. J. Med. Chem. 2003, 46, 666.
6. Wenlock, M. C.; Austin, R. P.; Barton, P.; Davis, A. M.;
Leeson, P. D. J. Med. Chem. 2003, 46, 1250.
3.3. Binding Assays
Binding assays for both receptors NPY1 and NPY5
were done as described by Duhault et al.59,60 In brief,
for the human Y1 receptor binding assay, using iodi-
nated Peptide YY (NEN), incubations were performed
at 30 ꢀC for 90 min with various competitors con-
centrations in Buffer A (Hepes/NaOH 20 mM, pH 7.4,
NaCl 10 mM, KH2PO4 220 mM, CaCl2 1.26 mM,
MgSO4 0.81 mM and bovine serum albumin 0.1%) with
SKN-MC cell membranes (50 mg of protein/mL of
assay) in a total volume of 500 mL. Non-specific bind-
ing was determined in the presence of 1 mM NPY.
7. Ventura, C. A.; Fresta, M.; Paolino, D.; Pedotti, S.; Cor-
saro, A.; Puglisi, G. J. Drug Target 2001, 9, 379 (and
references cited therein).
8. Grigera, J. R.; Caffarena, E. R.; de Rosa, S. Carbohydr.
Res. 1998, 310, 253.
9. Goodford, P. J. J. Med. Chem. 1985, 28, 849.
10. Goodford, P. J. J. Chemometr. 1996, 10, 107.
11. Molecular Discovery Ltd.: West Way House, Elms Par-
ade, Oxford (www.moldiscovery.com).
12. Kellogg, G. E.; Abraham, D. J. Eur. J. Med. Chem. 2000,
35, 651 (www.edusoft-lc.com).
13. Cozzini, P.; Fornabaio, M.; Marabotti, A.; Abraham,
D. J.; Kellogg, G. E.; Mozzarelli, A. J. Med. Chem. 2002,
45, 2469.
14. SYBYL Molecular Modelling Software, version 6.8 and
6.9; Tripos Inc.: 1699 S-Hanley Rd, Suite 303, St. Louis,
MO 63144-2913, USA (www.tripos.com).
15. Cucinotta, V.; D’Alessandro, F.; Impellizzeri, G.; Vec-
chio, G. J. Chem. Soc., Chem. Commun. 1992, 23, 1743.
16. Cucinotta, V.; Giuffrida, A.; Grasso, G.; Maccarrone, G.;
Vecchio, G. J. Chromatogr. A 2001, 916, 61.
The reaction was then stopped by filtration, the filters
(GF/B, Whatman, precoated in 0.3% PEI) were exten-
sively washed with buffer A, and counted in a gamma
counter (Packard). For human Y5 receptor binding
assay, the binding was carried out with iodinated pep-
tide YY (NEN) as follows: COS cells transfected with
the human Y5 NPY receptor were lysed and the mem-
branes prepared by differential centrifugation. These
membranes contained about 2 pmol per mg of protein
of this receptor. Incubations were performed in 500 mL
comprising, 20 pM final of [125I]PYY in 50 mL, 400 mL
of membrane suspension (0.15 mg/mL) and competitor
dilutions in 50 mL, at 30 ꢀC for 2 h.
The reaction was stopped by filtration through GF/C
filters (Whatman).
17. Castro, E. A.; Barbiric, D. J. J. Arg. Chem. Soc. 2002, 90,
1.
18. Homer, J.; Perry, M. C. J. Chem. Soc., FaradayTrans. 1
1986, 82, 533.
Acknowledgements
19. Neuhaus, D.; Willianson, M. The Nuclear Overhauser
Effect; WCH: New York, 1989.
20. Valensin, G.; Kushnir, T.; Navon, G. J. Magn. Reson.
1982, 46, 23.
21. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Lis-
kamp, R.; Lipton, M.; Cauleld, C.; Chang, G.; Hen-
drickson, T.; Still, W. G. J. Comput. Chem. 1990, 11, 440.
22. Schrodinger, Inc.: 1500 S. W. First Avenue, Suite 1180
We are grateful to the Minnesota Supercomputing
Institute for providing computational resources in sup-
port of this work and Prof. C. J. Cramer (Department
of Chemistry and Supercomputer Institute, University
of Minnesota) for the helpful discussion on MIDI. Prof.
Peter J. Goodford (University of Oxford, Laboratory
of Molecular Biophysics) and Prof. E. G. Kellogg

458
G. Uccello-Barretta et al. / Bioorg. Med. Chem. 12 (2004) 447–458
Portland OR 97201, USA/One Exchange Place, Suite 604,
Jersey City, NJ 07302, USA (www.schrodinger.com).
23. Easton, R. E.; Giesen, D. J.; Welch, A.; Cramer, C. J.;
Truhlar, D. G. Theor. Chim. Acta 1996, 93, 281.
24. Li, J.; Cramer, C. J.; Truhlar, D. G. Theor. Chem. Acc.
1998, 99, 192.
Ross, R. B., Ziegler, T., Eds.; ACS Symposium Series,
629; American Chemical Society: Washington, DC, 1996.
42. Adamo, C.; Barone, V. J. Chem. Phys. 1998, 108, 664.
43. Cramer, C. J. J. Mol. Struct. (Theochem) 1996, 370, 135.
44. Eliel, E. L.; Wilen, S. H. Stereochemistryof Organic
Compounds; Wiley: New York, 1994.
25. Hohenberg, P.; Kohn, W. Phys. Rev. B 1964, 136, 864.
26. Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
27. Del Bene, J. E. J. Phys. Chem. 1988, 92, 2874.
28. Del Bene, J. E. J. Int. Quantum Chem.: Quantum Chem.
Symp. 1992, 26, 527.
45. Diler, D. J.; Merz, K. M. J. Comput-Aided Mol. Des.
2002, 16, 105.
46. Modica, M.; Santagati, M.; Guccione, S.; Santagati, A.;
Russo, F.; Cagnotto, A.; Goegan, M.; Mennini, T. Eur. J.
Med. Chem. 2001, 36, 287.
29. Pudzianowski, A. T. J. Phys. Chem. 1996, 100, 4781.
30. Del Bene, A. T.; Person, W. B.; Szczepaniak, K. J. Phys.
Chem. 1995, 99, 10705.
31. Dewar, M. J. S.; Jie, C.; Yu, G. Tetrahedron 1993, 23,
5003.
47. Testa, B.; Kier, L. B.; Carrupt, P. A. Med. Res. Rev. 1997,
17, 303.
48. Freeman, R.; Wittekoek, S. J. Magn. Reson. 1969, 1, 238.
49. Salvadori, P.; Uccello-Barretta, G.; Balzano, F.; Bertucci,
C.; Chiavacci, C. Chirality 1996, 8, 423.
32. Allen, F. H.; Kennard, O. Chem. Des. Autom. 1993, 8, 1
and 31.
33. Allen, F. H. Acta Cryst. 2002, B58, 380.
34. Bruno, I. J.; Cole, J. C.; Edgington, P. R.; Kessler, M.;
Macrae, C. F.; McCabe, P.; Pearson, J.; Taylor, R. Acta
Cryst. 2002, B58, 389.
35. Berman, H. M.; Westbrook, J.; Feng, Z.; Gililand, G.;
Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E.
Nucl. Acids Res. 2000, 28, 235.
36. Lovejoy, B.; Welch, A. R.; Carr, S.; Luong, C.; Broka, C.;
Hendricks, T.; Campbell, J. A.; Walker, K. A. M.; Mar-
tin, R.; Van Wart, H.; Browner, M. F. Nature Struct.
Biol. 1999, 6, 217.
37. Gaussian98, PC version; Gaussian: Carnegie Office Park,
Bldg. 6, Pittsburhh, PA 15106, USA (www.gaussian.com).
38. Perdew, J. P.; Yue, W. Phys. Rev. B 1986, 33, 8800.
39. Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
40. Adamo, C.; Barone, V. Chem. Phys. Lett. 1997, 274, 242.
41. Perdew, J. P., Burke, K., Ernzerhof, M. In Chemical
Applications of DensityFunctional Theory ; Laird, B. B.,
50. Wang, S.; Tabernero, L.; Zhang, M.; Harms, E.; Van
Etten, R. L.; Stauffacher, C. V. Biochemistry 2000, 39,
1903.
51. Su, X. D.; Tadei, N.; Stefani, M.; Ramponi, G.; Par
¨
Nordiund, P. Nature 1994, 370, 575.
52. Halgren, T. A. J. Comput. Chem. 1996, 17, 490.
53. Halgren, T. A. J. Comput. Chem. 1996, 17, 520.
54. Halgren, T. A. J. Comput. Chem. 1996, 17, 553.
55. Halgren, T. A. J. Comput. Chem. 1996, 17, 587.
56. Halgren, T. A.; Nachbar, R. J. Comput. Chem. 1996, 17,
616.
57. Halgren, T. A. J. Comput. Chem. 1999, 20, 720.
58. Halgren, T. A. J. Comput. Chem. 1999, 20, 730 (and
references cited therein).
59. Duhault, J.; Boulanger, M.; Chamorro, S.; Boutin, J. A.;
Della Zuana, O.; Douillet, E.; Fauchere, J. L.; Feletou,
M.; Germain, M.; Husson, B.; Renard, P.; Tisserand, F.
Can. J. Biochem. Physiol. 2000, 78, 173.
60. Guery, S.; Rival, Y.; Wermuth, C.-G.; Renard, P.;
Boutin, J.-A. Chem. Pharm. Bull. 2002, 50, 636.