Investigation of structure-activity relationships in a series of glibenclamide analogues

Article abstract of DOI:10.1016/j.ejmech.2004.06.004

In this study, the synthesis of 15 new glibenclamide analogues is described. The conformational trends of these analogues were investigated using Monte Carlo conformational analysis. The conformational analysis results resolved the discrepancy between previous molecular modelling simulations of glibenclamide and allowed rationalizing the effect of aqueous environment on the overall conformation. The 3D-QSAR study was carried out with respect to the compounds' ability to antagonize the [³H]-glibenclamide binging in rat cerebral cortex. Superimposition of the antagonists was performed using the conformations derived from atom-by-atom fit to the glibenclamide crystal structure and this alignment was used to develop CoMFA models. CoMFA provided a good predictability: number of PLS components = 2, q² = 0.876, R ² = 0.921, SEE = 0.455 and F = 70. Best CoMFA models showed the steric and lipophilic properties as the major interacting forces whilst the electrostatic property contribution was a minor factor.

Full text of DOI:10.1016/j.ejmech.2004.06.004

European Journal of Medicinal Chemistry 39 (2004) 835–847
www.elsevier.com/locate/ejmech
Original article
Investigation of structure–activity relationships in a series
of glibenclamide analogues
ElizabethYuriev a, David C.M. Kong b, Magdy N. Iskander a,*
a Department of Medicinal Chemistry, Victorian College of Pharmacy, Monash University, 381 Royal Parade, Parkville 3052, Victoria, Australia
b Department of Pharmacy Practice, Victorian College of Pharmacy, Monash University, 381 Royal Parade, Parkville 3052, Victoria, Australia
Received 24 July 2003; received in revised form 11 June 2004; accepted 14 June 2004
Available online 23 July 2004
Abstract
In this study, the synthesis of 15 new glibenclamide analogues is described. The conformational trends of these analogues were investigated
using Monte Carlo conformational analysis. The conformational analysis results resolved the discrepancy between previous molecular
modelling simulations of glibenclamide and allowed rationalizing the effect of aqueous environment on the overall conformation. The
3D-QSAR study was carried out with respect to the compounds’ ability to antagonize the [3H]-glibenclamide binging in rat cerebral cortex.
Superimposition of the antagonists was performed using the conformations derived from atom-by-atom fit to the glibenclamide crystal
structure and this alignment was used to develop CoMFA models. CoMFA provided a good predictability: number of PLS components = 2,
q2 = 0.876, R2 = 0.921, SEE = 0.455 and F = 70. Best CoMFA models showed the steric and lipophilic properties as the major interacting forces
whilst the electrostatic property contribution was a minor factor.
© 2004 Elsevier SAS. All rights reserved.
Keywords: CoMFA; Conformational analysis; Glibenclamide; KATP channel
1. Introduction
ceptance of glibenclamide and several other sulfonylurea
drugs, several risks limit their applications [3,12]. Specifi-
cally, sulfonylureas are associated with: hypoglycemia and
weight gain; hyperinsulinemia and thus a deterioration in
diabetes; the increasing the risk of cardiovascular mortality
(in long-term treatment) [13–15].
Developing new K+ antagonists for KATP-channels [1] is a
challenging area of drug design with research in this field
growing exponentially. It is anticipated that these molecules
will have considerable therapeutic value in diabetes and
cardiac dysfunction [2].
Glibenclamide (Fig. 1) is a potent and selective KATP
channel blocker and a potent oral hypoglycaemic agent
[2–4]. It is a member of the sulfonylurea family and has a
high affinity inhibition of the KATP-Kir 6.2 (subfamily) sul-
fonylurea binding site, which appears to be important for
KATP channel regulation [5]. The mechanism of action of the
drug consists of the interaction with a sulfonylurea receptor
(SUR) and the inhibition of the ATP-sensitive K+ channel,
which depolarizes the cells and leads to insulin secretion
[1,6,7]. The same mechanism of drug action is at play at the
extrapancreatic sites in the liver, kidney, brain, skeletal,
heart, and smooth muscle [3,6,8–11]. Despite the wide ac-
Previously, we had tested glibenclamide and analogues
for their ability to antagonize the vasorelaxant actions of the
K+ channel opener levcromakalim in rat thoracic aorta and to
displace [3H]-glibenclamide from rat cerebral cortex mem-
branes [16]. That study demonstrated that glibenclamide and
analogues were active as antagonists of levcromakalim-
mediated responses. Our glibenclamide analogues were also
1000 to 10,000 times more potent at displacing [3H]-
glibenclamide from rat cerebral cortex membranes. The re-
sults also revealed the quantitative differences in the activi-
ties of the sulfonamide/sulfonylurea-based compounds
studied. Specifically, the sulfonamide analogues had a more
dramatic effect as levcromakalim antagonists than as agents
that displace [3H]-glibenclamide. Therefore, the main find-
ing from that study was that the above two propensities of
glibenclamide analogues are based on different structure–
* Corresponding author. Tel.: +61-3-99039545; fax: +61-3-99039582.
E-mail address: magdy.iskander@vcp.monash.edu.au (M.N. Iskander).
© 2004 Elsevier SAS. All rights reserved.
doi:10.1016/j.ejmech.2004.06.004

836
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
critically revisit the conclusions of earlier reported confor-
mational studies of glibenclamide [18–20]. CoMFA models
as well as CoMFA predictions for modified analogues con-
firmed the importance of several structural elements for the
KATP channel blocking activity and allowed arriving at a set
of rules for the design of new compounds with improved
activity.
2. Chemistry and pharmacology
Amidoethylbenzenesulphonamides 1a–h and amidoeth-
ylbenzenesulphonylureas 2b–h were synthesized as outlined
in Scheme 1. Benzenesulphonylureas 4a–d were synthesized
as outlined in Scheme 2. Drugs that were generously donated
(Scheme 3) were: glibenclamide 2a (Hoechst Pharmaceuti-
cal Co., Melbourne, Australia), glipizide 2i (Farmitalia Carlo
Erba, Milan, Italy), iodoglibenclamide 2j (from Dr. Donald
Gehlert, Lilly Research Laboratories, Indianapolis, IN). Ben-
zamide 5 (Scheme 3) was prepared by a procedure similar to
that outlined for amidoethylbenzenesulphonamides 1a–h in
Scheme 1. The measurement of the pKi values of the inhibi-
tion of [3H]-glibenclamide binding in rat cerebral cortex by
these compounds (Table 1) was described elsewhere [16].
3. Results and discussion
3.1. Conformational analysis
In order to further elucidate glibenclamide mechanism of
action and to develop more potent agents free of the limita-
tions mentioned in the introduction, the structural informa-
tion relating to the SUR–glibenclamide interaction is neces-
sary. Unfortunately, no 3D structural data is available with
respect to this interaction. The lack of such information
leaves structural investigation of isolated glibenclamide as
the main line of attack aimed at determination of biologically
relevant conformation(s).
Previously, conformational preferences of glibenclamide
have been studied experimentally by NMR and X-ray crys-
tallography [17]. It has been shown that, in the solid state,
glibenclamide exhibits an extended conformation (Fig. 1,
top). In the same investigation, the conformational behaviour
of glibenclamide was studied in solution and the free rotation
around bonds t2 and t3 (as designated in Fig. 2) was demon-
strated.
Conformational trends of glibenclamide analogues have
also been a subject of molecular mechanics studies [18–20],
which resulted in contradictory conclusions. While Grell et
al. [20] generated the lowest energy structure with the con-
formation identical to that in the solid state, Lins et al. [18,19]
produced lowest energy structures displaying a U-shaped
conformation with hydrophobic cycles at the extremity of
each branch and a peptidic bond at the bottom of the U.
Thus, we undertook to carry out a detailed conformational
analysis of the glibenclamide system in order to achieve a
Fig. 1. 3D structure of glibenclamide. Hydrogens are omitted for clarity.
Colour coding: C, grey; O, red; N, blue; S, yellow; Cl, green. (Top) Crystal
structure [17]. Coordinates are taken from CSD (refcode DUNXAL). (Mid-
dle) Lowest energy conformation resulting from the systematic conforma-
tional search in vacuo. (Bottom) Lowest energy conformation resulting from
the systematic conformational search in explicit aqueous solution.
activity relationships (SARs). It has been also shown that
different structure–activity relationships exist for different
structural groups of analogues (i.e., amidoethylbenzene-
sulphonylureas, sulphonamides, benzenesulphonylureas,
benzamide, benzoylcyclohexylureas, and benzoylphenylth-
ioureas).
Such differentiation may be important for developing bet-
ter compounds with greater specificity in antagonistic activ-
ity. The aim of the current study was therefore to analyse the
three-dimensional (3D) structural properties of a number of
closely related glibenclamide analogues in order to clarify
the structure–activity relationships in quantitative terms and
to assess the relative contributions of different structural
elements to the antagonistic activity. In particular, conforma-
tional analysis and Comparative Molecular Field Analysis
(CoMFA) study of these compounds have been carried out.
Detailed conformational analysis, now carried out both in
vacuo and in explicit aqueous solution, provides data
complementary to previously published NMR and X-ray
investigations of glibenclamide [17]. It also allows us to

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
837
Scheme 1. Synthesis of sulfonamide and amidoethylbenzenesulfonylurea analogues. (a) Dry acetone. (b) Cyclohexylisocyanate, anhydrous potassium
carbonate.
three-fold aim: (i) to correlate the theoretically-observed
conformational trends with those observed in the relevant
experimental structures; (ii) to resolve the discrepancy be-
tween previous conformational analysis simulations
[18–20], and (iii) to investigate the differences of the confor-
mational trends in vacuo and in solvated medium.
3.1.1. Monte Carlo conformational analysis
Monte Carlo conformational analysis was carried out in
vacuo and yielded the clusters of torsion values, accessible
for each rotatable bond, which are best visualized as confor-
mational wheels (Fig. 2). Additional wheels (not shown)
were also generated, where frequency and strain energy of
particular conformations were used to further identify the
preferred conformational ranges.
Scheme 2. Synthesis of benzenesulfonylurea analogues. (a) 25% Ammonia
solution. (b) Cyclohexylisocyanate, anhydrous potassium carbonate.
Scheme 3. Additional glibenclamide analogues used for 3D-QSAR analysis.

838
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
Table 1
pKi values of the inhibition of [3H]-glibenclamide binding in rat cerebral
cortex
Compound number
pKi
6.76
1a
2a
2b
2c
2d
2e
2f
2g
2h
2i
10.19
9.64
8.58
9.44
9.07
8.66
9.52
9.28
9.88
8.86
6.14
6.51
6.50
6.02
2j
4b
4c
4d
5
See reaction schemes for compounds’ structures.
The values of the torsional angle t1 (Fig. 2) reflect rotation
of the terminal phenyl ring. This rotation is of special interest
in those glibenclamide analogues, in which the terminal
phenyl ring is substituted, particularly in the ortho-position.
In our conformational analyses, this substituent was com-
monly found in the anti-orientation with respect to the car-
bonyl. The same arrangement was observed in the crystal
structure of glibenclamide [17] and was prevalent in the
relevant structures obtained from the Cambridge Structural
Database (CSD).
Previously, an NMR study of glibenclamide demonstrated
the free rotation around the Namide–Csp3 and Csp3–Csp3 bonds
[17]. The respective torsion angles t2 and t3 (Fig. 2), together
with the angle t4 discussed below, are determinative of the
relative orientation of two aromatic rings of glibenclamide.
Therefore, they influence the overall shape of the molecule,
affecting either an extended or a folded shape [18–20]. Our
conformational searches indicated that, indeed, the rotation
around the Namide–Csp3 bond is free. With respect to the
rotation around the Csp3–Csp3 bond, the preferred torsional
angle ranges for t3 were around 60° and 180° (Fig. 2), as
expected.
The values of the torsional angle t4 (Fig. 2) reflect rotation
of the middle-of-chain phenyl ring around the Csp3–Caromatic
bond. The conformational searches pointed to the near per-
pendicular orientation of the –CH2–CH2– bond with respect
to the aromatic ring as preferred. Such orientation (e.g.,
Fig. 1, middle) leads to a partial overlap of the aromatic rings
and, as a result, to the compact conformation of the benza-
mide part of the molecule.
The conformational environment around the sulfo group
SO2 determines the orientation of the middle-of-chain aro-
matic ring with respect to the cyclohexyl ring. The conforma-
tional analysis indicated the virtually unhindered rotation of
this aromatic ring around the Caromatic–S bond (torsion angle
t5, Fig. 2). The rotation of the cyclohexylurea fragment
around the S–N bond is represented by the torsion angle t6
Fig. 2. Torsion angles t1–t10, varied during the conformational search, are
indicated. Conformational wheels are shown for torsion angles t2–t6 and t9.
Torsion angle distributions are from: (i) Monte Carlo conformational search
and (ii) CSD.

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
839
(Fig. 2), which clustered around 60° and 180° regions
(Fig. 2), with regions around 60° being significantly
favoured, both statistically and energetically.
mally exposed leading to a folded conformation. However, in
this case, the groups that surround these fragments (i.e.,
chloro, methoxy, carbonyl, sulfo, amide) are all polarizable
and are good hydrogen bonding donors/acceptors. As a re-
sult, these groups make the hydrophobic fragments more
soluble and push them into the bulk solvent and away from
each other, thus favouring the extended conformation. On the
other hand, the folded conformation is preferred in vacuum
because the effect of the polar/hydrogen bonding groups is
not as pronounced as it is in aqueous solution. Furthermore,
in vacuum the van der Waals interactions, especially stack-
ing, are significant. Whereas this rationalisation accom-
plishes our third aim (to investigate the differences of the
conformational trends in vacuo and in solvated medium–see
above), it leaves open to discussion the identity of the bio-
logically active conformation(s) of glibenclamide and the
related structures and, consequently, the conformation(s) to
be used 3D-QSAR analyses.
It is unarguable that the selection of bioactive conforma-
tion is the critical point in a 3D-QSAR study. Generally,
albeit not always, the best choice for a bioactive conforma-
tion is the one extracted from a receptor–ligand structure. In
the case of SUR–sulfonylurea interaction, the complex struc-
ture, even the structure of the isolated receptor that would
lend itself to a docking approach, is not yet available. In cases
such as that it is usually assumed that a ligand will adopt a
bioactive conformation so as to achieve a relatively low strain
energy. Thus, if a crystal structure for a ligand is available it is
usually used for a 3D-QSAR analysis. An example of such
application is a recent CoMSIA study on sulfonylureas [21].
However, the present case offers three options with respect to
low energy conformations. Specifically, the crystal structure
of glibenclamide [17]–extended conformation, the lowest
energy conformation generated in vacuum–fully folded, and
the lowest energy conformation generated in explicit aque-
ous environment and resembling that in the solid state. This
choice is confounded by the great flexibility of the glibencla-
mide and its analogues, as demonstrated by conformational
analyses presented above. It is accepted that conformational
flexibility is one of the greatest challenges to selecting a
ligand for CoMFA [22]. Such flexibility results in a number
of diverse local conformational minima within a narrow
range of energies (including the conformation of the crystal
structure). We agree with Lins et al. [19] that there is prob-
ably a link between the lowest energy conformation of glib-
enclamide and its biological activity. However, we also sup-
port Grell et al. [20] notion that it is reasonable to consider a
range of low energy conformations as favourable potential
binding conformations. Furthermore, it should not be ne-
glected that, in case of flexible molecules, binding to the
receptor may modify the interaction energy resulting in a
ligand conformation not revealed by a conformational search
of an isolated ligand [23]. Having weighed out these argu-
ments and our data, we decided that, out of the accessible low
energy candidates, the crystal structure of glibenclamide [17]
is the best available approximation for a bioactive conforma-
tion and should be used as a basis for a 3D-QSAR analysis.
Finally, the torsion angle t9 (Fig. 2) and the conformation
of the cyclohexyl ring control the orientation of the cyclo-
hexyl ring with respect to the urea moiety. The conforma-
tional search yielded the chair conformation of the cyclo-
hexyl in all found structures. The rotation around the torsion
angle t9 was found to be virtually free (Fig. 2), but the areas
around 120° were significantly preferred, both energeti-
cally and statistically.
The torsion angles t7, t8, and t10 (Fig. 2) were not in-
cluded into the Monte Carlo conformational searches al-
though, in principle, rotation is possible around the respec-
tive bonds. The rotation around the amide bonds of the urea
moiety is controlled by the torsion angles t7 and t8 (Fig. 2). In
the crystal structure of glibenclamide [17] both bonds are in
the trans arrangement and this arrangement was found to be
prevalently represented in CSD. Similarly, the analysis of the
crystal structures showed that the t10 torsion angle values
cluster around 180° (i.e., the methyl group is pointing away
from the carbonyl). This arrangement is intuitively apparent
(due to steric restrictions otherwise) and is observed in the
crystal structure of glibenclamide [17].
Thus, we have accomplished the first aim (to correlate the
theoretically-observed conformational trends with those de-
tected in the relevant experimental structures) and concluded
that the results of our conformational analyses agree well
with the torsion angle distributions found in the relevant
structures from CSD and are in accord with previous experi-
mental structural investigations of glibenclamide [17].
3.1.2. Systematic conformational analysis
Systematic conformational analyses based on the acces-
sible angle ranges, determined as described above, were
carried out in vacuo and in explicit aqueous solution. Such
simulations allow appropriate accounting for solvent effects
on the energy and conformation. The conformations of the
lowest energy structures obtained in vacuo and in solution are
depicted in Fig. 1, middle and bottom, respectively. While
similar torsion ranges were observed in vacuo and in the
aqueous solution (data not shown), the overall conformations
of the lowest energy structures were significantly different.
The systematic conformational search carried out in vacuo
favoured the highly folded conformation (Fig. 1, middle),
resembling the U-shaped structure modelled by Lins et al.
[18,19]. Adding the explicit solvent to the modelled environ-
ment benefited a more extended molecular shape of glib-
enclamide (Fig. 1, bottom), as predicted by Grell et al. [20].
While these results resolve the discrepancy between previous
molecular modelling simulations of glibenclamide (second
aim of the conformational analysis effort–see above), they
also raise issues with respect to the effect of aqueous envi-
ronment on the overall conformation. A priori, in an aqueous
environment, the hydrophobic fragments of the molecule
(i.e., aromatic and aliphatic rings) would prefer to be mini-

840
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
3.2. CoMFA of glibenclamide analogues
CoMFA results are presented in Table 2 and demonstrated
in Fig. 3. The best model (B) utilizes steric fields and LogP
values as variables and only requires two PLS components to
achieve the best statistics observed. This model predicts well
the biological activity studied as demonstrated by the predic-
tion plot (Fig. 4). However, a caveat is warranted with respect
to this prediction plot: while the formal statistical measures
of the quality of this model are good (Table 2), it must be
pointed out that the activity data used is distributed in a
bimodal fashion. The second-best model is model D, which
Fig. 4. Correlation of experimental vs. predicted biological activity for
glibenclamide and glibenclamide analogues (amidoethylbenzenesulfonylu-
reas 2a–j, sulfonamide 1a, benzenesulfonylureas 4b–d, and benzamide 5)
using CoMFA model B. pKi, inhibition of [3H]-glibenclamide binding in rat
cerebral cortex [16]. The diagonal represents the y=x line (optimal predic-
tion).
utilizes steric and electrostatic fields and LogP values as
variables and, again, just two PLS components are sufficient
to produce good statistics. It should be commented that these
models combine two different types of variables: interaction
fields (which are ‘multicolumn’ data) and scalar LogP (a
single data point for each compound). Although, such com-
binations are an extension of a traditional molecular field
analysis, their use is a common tactic in the 3D-QSAR field
(e.g., [24]). In addition to the variables in model B, model D
accounts for electrostatic fields, however the electrostatic
contribution is small (6%). Thus, both models suggest that
electrostatics had very little contribution to the quantitative
structure–activity relationship in this set of compounds.
However, this may be not the intrinsic feature of the relation-
ship, but rather the artefact resulting from limited structural
variation in the studied compounds.
Fig. 3. Steric CoMFA map for model B showing contributions to glibencla-
mide activity. Yellow denotes regions where steric bulk is detrimental to
activity and green denotes regions where steric bulk enhances activity
(standard 80/20 settings). Two views allowing the clear observation of all
features of the glibenclamide molecule are shown.
Table 2
CoMFA models
Models
Variables a
A
B
C
D
Steric electro
Steric LogP
Electro LogP
Steric electro LogP
Statistics
No. of components
2
2
3
2
q2
0.623
0.909
0.994
0.488
60
0.876
0.921
0.569
0.455
70
0.850
0.924
0.655
0.466
44
0.829
0.895
0.669
0.524
51
R2
SEP
SEE
F
Prob
0.000
0.000
0.000
0.000
Contributions
Steric
Electro
LogP
0.722
0.278
0.447
0.319
0.060
0.621
0.525
0.475
0.553
a Variables: no. components, optimum number of components obtained from cross-validated PLS analysis; R2, correlation coefficient; q2, cross-validated
correlation coefficient; SEP, standard error of prediction; SEE, standard error of estimate; F, F-statistic; prob, probability of the null hypothesis (both F and prob
are tests for R2 = 0 hypothesis).

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
841
Fig. 5. Structures of glibenclamide analogues used for predictive CoMFA calculations. X, halogen.
According to the visual representation of model B (Fig. 3),
steric bulk is desirable (green) around the phenyl ring of the
benzamide moiety with preference for the side of the ring
anti to the direction of the carbonyl of the amide bond; it is
also more favourable in the plane of the ring, which indicates
preference for halogen substituents. Steric bulk is unfavour-
able (yellow) ‘above’ the cyclohexyl ring, i.e., in the direc-
tion of the carbonyl of the urea moiety. Since in this series of
analogues the structural and, hence, conformational variation
is limited to the substitution of the phenyl ring and to the
presence/absence of certain structural features in compounds
1a, 4b–d, and 5, the steric field clouds are concentrated
mainly around the phenyl ring. Therefore, the design of new
analogues based on this CoMFA study may be done prima-
rily with respect to phenyl ring substitution.
decline significantly upon hydroxylation of the cyclohexyl
ring, the acyclic analogues showed markedly reduced activ-
ity compared to glibenclamide.
Since the test compounds in this CoMFA study vary
mostly in the nature and the arrangement of substituents in
the terminal phenyl ring of glibenclamide, the following
calculations were focused on that part of the molecule. Simi-
larly to the above, replacement of the methoxy group with
methyl in the terminal phenyl ring of glibenclamide (Fig. 5b)
resulted in the decreased pKi value of 9.66, indicating the
requirement for an electronegative atom at this position. This
particular modification was also found to have a pronounced
effect on the structural features of this derivative. Specifi-
cally, it caused the conformational change, i.e., the 180°
rotation of the terminal phenyl ring. In glibenclamide, the
methoxy group is in the position anti to benzamide carbonyl.
However, the methyl group in this derivative is in the syn
arrangement (as shown in Fig. 5). While it is obvious that
such conformational adjustment is due to steric reasons (less
repulsion between the carbonyl oxygen and the methyl car-
bon), it also indicates that the anti arrangement, observed in
glibenclamide, is likely to be preferred for biological activity.
Thus, whereas OCH3 → CH3 modification leads to the in-
crease in LogP, which could have been expected to result in
the increase in pKi, it is clearly detrimental for steric and
electronic properties of the molecule.
3.3. CoMFA-based predictions of biological activity
Based on the CoMFA results presented above for the
studied series of compounds, pKi values were calculated for
some glibenclamide derivatives (Fig. 5) using model B.
When the cyclohexyl ring in 2f (pKi = 8.66) was replaced
with the methyl group (Fig. 5a), the resulting pKi (7.90)
decreased, suggesting the importance of the cyclohexyl ring
for the observed activity. This is in agreement with other
studies on structural requirements of sulfonylureas and their
analogues for interactions with SUR subtypes. Specifically,
Meyer et al. [25] found that exchanging the cyclohexyl ring
of glibenclamide by a methyl group reduces its selectivity for
SUR1. Also, it has been shown in a recent study on inhibition
of [3H] glibenclamide binding to rat brain synaptomes [11]
that while the activity of glibenclamide analogues does not
Substitution of meta-hydrogen of 2h (pKi = 9.28) with
halogens X (Fig. 5c) was studied because glibenclamide
itself carries chlorine at this position. Whereas the substitu-
tion with chlorine produced virtually no effect on pKi (9.26),
the iodine at this position resulted in the increase of pKi to
9.72, which is closer to the pKi of iodoglibenclamide (9.88),

842
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
Table 3
which carries the iodine substituent in the same position.
Taken in isolation, this modification would point toward the
preference for iodine at this position. However, the compari-
son of activities of glibenclamide and iodoglibenclamide
suggests the preference for chlorine at the same location.
This seeming discrepancy is resolved by the examination of
the conformations of these derivatives, in particular–of the
orientations of the terminal phenyl ring with respect to the
carbonyl group of the amide. Specifically, as outlined above
for OCH3 vs. CH3 substitution, the preferred arrangement of
the ortho-methoxy group in glibenclamide and iodoglib-
enclamide is in the anti orientation with respect to the carbo-
nyl due to steric reasons. This restriction places a halogen
atom into a particular location in space and brings out the
preference for chlorine. On the other hand, the methoxy
group in 2h is in the meta-position with respect to the amide
bond and is not sterically restricted. This results in the rela-
tively free rotation of the phenyl ring and in no observable
preferences of the orientation of the substituents with respect
to the amide bond. Thus, the comparison of glibenclamide
activity to its two actual and two hypothetical derivatives
shows the importance of both the nature of groups/atoms
used for substitution as well as their substitution location
and, in particular, the resulting conformations.
Using CoMFA model B, biological activity values (pKi)
were also calculated for some sulfonamide-related hypogly-
caemic agents: glimepiride, meglitinide, and meglitinide de-
rivative (Fig. 5d–f). The resulting pKi values were 10.1, 6.67,
and 6.76, respectively. It is impossible to absolutely verify
these predictions, because the experimental pKi of these
compounds were not determined alongside the experimental
pKi values used in the current study. However, the result of
pKi = 10.1 for glimepiride indicates the good predicting
ability of this model since, according to literature (for ex-
ample [26,27]), this compound shows activity similar to
glibenclamide. On the other hand, for the purpose of this
comparison it must be born in mind that glibenclamide and
glimepiride may be binding to the different receptor sites
[7,28,29]. These predictions lend themselves to cautious
speculations with respect to the structure–activity relation-
ships. Specifically, the decreased pKi values of the benzoic
acid analogues indicate the importance of the sulfo group (as
compared to carboxy), probably due to the altered geometry
of sulfur (compared to carbon). Again, this theoretical
CoMFA-based prediction is in accord with experimental
findings related to the structural requirements of sulfony-
lureas and their analogues for interactions with SUR sub-
types. Namely, Meyer et al. [25] found that the replacement
of the carboxyl group of meglitinide by a sulfonylurea group
significantly increased the affinities for SUR1 and SUR2. On
the other hand, there is a strong opinion in the field of the
anti-diabetic agents that it is the acidic/anionic function (e.g.,
carboxy [25] or phosphate [30]), rather than the sulfonylurea
group per se of the first and second generation agents, that is
essential for drug activity [3]. The more recently evolved
group of compounds, styled glinides (such as meglitinide and
CoMFA-predicted diiodo analogue inhibition activities
Analogue
Model F
Model H
2,4-Diiodo
10.60
2,5-Diiodo
10.65
3,5-Diiodo
10.69
10.71
10.74
10.76
nateglinide [31]), which are based on the non-sulfonylurea
part of glibenclamide, do block KATP channels, albeit seem to
interact with different regions of SUR [3].
Therefore, CoMFA models as well as CoMFA predictions
pointed to the importance of the following structural ele-
ments for the biological activity: sulfo group, cyclohexyl,
and selective substitution at the terminal phenyl ring (Fig. 2).
The above analysis also produced positive indicators for
halogen substitution. Specifically, such substitution allows
significant LogP contribution to the model and agrees with
the preference for steric bulk localization in the plane of the
phenyl ring. Encouraged by these indicators, we carried out
systematic predictive CoMFA calculations of biological ac-
tivity for iodo-substituted glibenclamide analogues in the
terminal phenyl ring. The predicted pKi values for mono-
substituted analogues (9.81–9.87) demonstrated that, while
the position of iodo-substitution does not affect the activity,
this choice of halogen, compared to chlorine (8.58–9.44)
[16], significantly improves it. These values were very close
to that of iodoglibenclamide, suggesting that the methoxy
group in iodoglibenclamide may be irrelevant for the activity.
Further encouraged by these findings, we calculated the pKi
values for diiodo substituted analogues, which are summa-
rized in Table 3. Interesting observation was made while
optimizing the analogue structures for these calculations.
The 2,6-diiodo analogue was highly strung sterically and
exhibited the strain energy more than 8 kcal/mol above that
for all other compounds in this series. While this could be an
indication of the steric repulsion between the amide carbonyl
and the neighbouring iodo-substituents, it was interesting to
observe that in the 2,4- and 2,5-diiodo analogues the iodine
atoms, neighbouring with the carbonyl group, preferred to
occupy the syn rather than anti-orientation with respect to the
carbonyl oxygen. Out of three sterically and conformation-
ally allowed diiodo substitutions (Table 3), 3,5-diiodo ana-
logue was found to be the most energetically favoured (with
the energy well of approximately 6 kcal/mol). All three
compounds produced the highest predicted activity, super-
seding that of glibenclamide!
4. Conclusions
Molecular modelling of glibenclamide and its derivatives,
presented in this work, was aimed at clarifying the structure–

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
843
activity relationships of glibenclamide analogues in quanti-
tative terms using conformational analysis and CoMFA study
of these compounds. The detailed conformational analysis in
vacuum and in explicit aqueous solution provided data
complementary to previously published NMR and X-ray
investigations of glibenclamide. It also allowed us critically
revisiting the conclusions of earlier reported conformational
study of glibenclamide done purely in vacuum. CoMFA
models as well as CoMFA predictions for modified ana-
logues confirmed the importance of several structural ele-
ments for the KATP channel blocking activity and allowed
arriving at a set of rules for the design of new compounds
with improved activity–rules to be used at the bench in order
to improve new pharmacomodulations. It must be noted that
the studied series of compounds is limited and this study
focused on developing a robust model and not in investigat-
ing its predictive power with an external and large test set.
compound was filtered and recrystallized from methanol and
decolourising charcoal to give white plates.
5.1.1.1.4-(2-(5-Chloro-2-methoxybenzamido)ethyl)benzene-
sulphonamide (1a). Compound 1a was synthesized from
5-chloro-2-methoxybenzoyl chloride (2.05 g, 10 mmol) and
4-(2-aminoethyl)benzenesulphonamide (2.00 g, 10 mmol).
Yield, 1.89 g (59.8%); mp: 213–215 °C (lit. mp: 214–216 °C,
[32]); TLC Rf 0.74, 8:2 chloroform/methanol. m/z (relative
intensity) 368 (M+. +1, 86%), 169 (100%). Accurate mass
369.06758 requires C16 H18 N2 O4 S1 Cl1. mmax 3380,
3292, 1616, 1544, 1340, 1158, 682 cm–1.
5.1.1.2.4-(2-(4-Chloro-2-methoxybenzamido)ethyl)benzene-
sulphonamide (1b). Compound 1b was synthesized from
4-chloro-2-methoxybenzoyl chloride (4.10 g, 20 mmol) and
4-(2-aminoethyl)benzenesulphonamide (4.00 g, 20 mmol).
Yield, 4.36 g (59.2%); mp: 186–187 °C (lit. mp: 187–188 °C,
[32]); TLC Rf 0.73, 8:2 chloroform/methanol. m/z (relative
intensity) 368 (M+. +1, 60%), 169 (80%). mmax 3388, 3334,
1626, 1530, 1332, 1239, 1152, 642 cm–1.
5. Experimental
5.1. Chemistry
5.1.1.3. 4-(2-(2-Chlorobenzamido)ethyl)benzenesulphona-
mide (1c). Compound 1c was synthesized and from
2-chlorobenzoyl chloride (2.70 g, 15.4 mmol) and 4-(2-
aminoethyl)benzenesulphonamide (3.10 g, 15.4 mmol).
Yield, 3.58 g (70.6%); mp: 198–199 °C (lit. mp: 195–196 °C,
[33]); TLC Rf 0.87, 8:2 chloroform/methanol. m/z (relative
intensity) 339 (M+. +1, 63%), 139 (100%). mmax 3370, 3292,
1629, 1548, 1317, 1290, 1152, 660 cm–1.
Melting points were determined on Gallenkamp MFB-
595 apparatus and are uncorrected. Infra-red spectra were
recorded using a Hitachi 270-30 grating spectrophotometer
as potassium bromide discs. Positive ion Fast Atom Bom-
bardment (FAB) mass spectra were obtained on a Joel JMX
DX-300 double focussing instrument. Incident particle used
was Xenon. Microanalysis was conducted by the National
1
Analytical Laboratory Ltd and had an error of 0.4%. H
5.1.1.4. 4-(2-(3-Methoxybenzamido)ethyl)benzenesulphona-
mide (1h). Compound 1h was synthesized from
3-methoxybenzoyl chloride (1.71 g, 10 mmol) and 4-(2-
aminoethyl)benzenesulphonamide (2.00 g, 10 mmol). Yield,
2.02 g (60.5%); mp: 163–164 °C (lit. mp: 162 °C, [33]); TLC
Rf 0.84, 8:2 chloroform/methanol. m/z (relative intensity)
335 (M+. +1, 100%), 135 (100%). Accurate mass 335.10395
requires C16, H19, N2, O4, S1. mmax 3334, 3280, 1629, 1536,
1317, 1236 1155, 684 cm–1.
NMR data were recorded using a Bruker AM-300 MHz
instrument. Chemical shifts were expressed in parts per mil-
lion (ppm) using tetramethylsilane (TMS) as the standard
reference. All spectra were recorded in 5 mm tubes. Analyti-
cal thin-layer chromatography (TLC) was performed on
Merck Kieselgel 60 F254 precoated aluminium backed plates.
Column chromatography was performed either on Merck
Kieselgel 60 (70–230 meshASTM) silica or Merck Kieselgel
60 (230–400 mesh ASTM) silica. All chemicals used were
reagent grade and were distilled, recrystallized or dried with
molecular sieves prior to use. For synthesis of related com-
pounds which follows the same procedure, a general proce-
dure has been outlined for one compound. All compounds
following the same procedure have been indicated and any
variations from the general procedure are mentioned. Unless
otherwise stated, all compounds used were purchased from
Aldrich.
5.1.2. General procedure for the synthesis of 4-(2-(benza-
mido)ethyl)benzenesulphonamides (1d–g)
Appropriate benzoyl chloride was added slowly to a
stirred solution of 4-(2-aminoethyl)benzenesulphonamide in
dry pyridine (60 ml). The resultant mixture was stirred for
16 h. After the removal of pyridine under vacuum, a viscous
yellow liquid was obtained. When this viscous residue was
treated with hydrochloric acid (1 M, 50 ml), a yellowish solid
was formed. This was filtered, washed with water and recrys-
tallized as white needles from absolute ethanol.
5.1.1. General procedure for the synthesis of 4-(2-(benza-
mido)ethyl)benzenesulphonamides (1a–c, 1h)
Appropriate benzoyl chloride was added slowly to a
stirred solution of 4-(2-aminoethyl)benzenesulphonamide in
dry pyridine (60 ml). The resultant mixture was stirred for
16 h. Pyridine was removed under reduced pressure and
absolute ethanol (40 ml) was added to the residue. The crude
5.1.2.1. 4-(2-(3-Chlorobenzamido)ethyl)benzenesulphona-
mide (1d). Compound 1d was synthesized from 3-chloro-
benzoyl chloride (3.48 g, 20 mmol) and 4-(2-aminoethyl)-
benzenesulphonamide (4.00 g, 20 mmol). Yield, 2.14 g

844
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
(31.7%); mp: 161–163 °C (lit. mp: 161–163 °C, [32]); TLC
Rf 0.81, 8:2 chloroform/methanol. m/z (relative intensity)
339 (M+. +1, 52%), 139 (100%). mmax 3328, 1626, 1533,
1332, 1152 cm–1.
+1, 5%), 369 (23%), 169 (100%). Accurate mass 494.15232
requires C23, H29, N3, O5, C11, S1. Micro analysis ex-
pected values C 55.9%, H 5.7%, N 8.5%, and S 6.5%. Found
C 55.5%, H 5.9%, N 8.6%, and S 6.5%. mmax 3364, 2932,
1623, 1593, 1527, 1449, 1248, 1155 cm–1.
5.1.2.2. 4-(2-(4-Chlorobenzamido)ethyl)benzenesulphona-
mide (1e). Compound 1e was synthesized from 4-chloro-
benzoyl chloride (3.48 g, 20 mmol) and 4-(2-amino-
ethyl)benzenesulphonamide (4.00 g, 20 mmol).Yield, 3.46 g
(51.2%); mp: 222–223 °C (lit. mp: 225–226 °C, [32]); TLC
Rf 0.86, 8:2 chloroform/methanol. m/z (relative intensity)
339 (M+. +1, 36%), 139 (71%). Accurate mass 339.05915
requires C15, H16, N2, O3, S1, Cl1. mmax 3400, 3328, 1632,
1542, 1323, 1155, cm–1.
5.1.3.2. 1-(4-(2-(4-Chlorobenzamido)ethyl)benzenesulpho-
nyl)-3-cyclohexylurea (2e). Compound 2e was synthesized
from 1e (676 mg, 2 mmol), anhydrous potassium carbonate
(830 mg, 6 mmol) and cyclohexylisocyanate (400 mg,
3.2 mmol). Yield, 564 mg (60.9%) of white needles; mp:
197–199 °C (lit. mp: 196–197.5 °C, [33]); TLC Rf 0.62,
9.5:0.5 chloroform/methanol. m/z (relative intensity) 464
(M+. +1, 2%), 339 (29%), 139 (100%). mmax 3384, 2932,
1634, 1532, 1442, 1232, 1154 cm–1.
5.1.2.3. 4-(2-(Benzamido)ethyl)benzenesulphonamide (1f).
Compound 1f was synthesized from benzoyl chloride
(1.41 g, 10 mmol) and 4-(2-aminoethyl)benzenesulpho-
namide (2.00 g, 10 mmol). Yield, 1.70 g (56.0%); mp: 249–
250 °C; TLC Rf 0.81, 8:2 chloroform/methanol. m/z (relative
intensity) 305 (M+. +1, 43%), 105 (29%). Accurate mass
305.09385 requires C15, H17, N2, O3, S1. Microanalysis
expected values C 59.2%, H 5.3%, N 9.2% and S 10.5%.
Found C 59.2%, H 5.6%, N 9.3%, and S 10.5%. mmax 3388,
3190, 1617, 1542, 1326, 1149, 684 cm–1.
5.1.3.3.
1-(4-(2-(Benzamido)ethyl)benzenesulphonyl)-3-
cyclohexylurea (2f). Compound 2f was synthesized from 1f
(608 mg, 2 mmol), anhydrous potassium carbonate (830 mg,
6 mmol) and cyclohexylisocyanate (400 mg, 3.2 mmol).
Yield, 72 mg (8.34%) of white needles; mp: 187–189 °C (lit.
mp: 189–191 °C, [33]); TLC Rf 0.38, 9.5:0.5 chloroform/
methanol. m/z (relative intensity) 430 (M+. +1, 7%), 305
(100%). Accurate mass 430.17894 requires C22 H28 N3 O4
S1. mmax 3340, 2932, 1632, 1512, 1440, 1215, 1158 cm–1.
5.1.2.4. 4-(2-(2-Methoxybenzamido)ethyl)benzenesulphona-
mide (1g). Compound 1g was synthesized from 2-methoxy-
benzoyl chloride (1.71 g, 10 mmol) and 4-(2-amino-
ethyl)benzenesulphonamide (2.00 g, 10 mmol). Yield, 1.25g
(37.4%); mp: 177–179 °C (lit. mp: 173–174 °C, [33]); TLC
Rf 0.84, 8:2 chloroform/methanol. m/z (relative intensity)
335 (M+. +1, 38%), 135 (100%). Accurate mass 335.10689
requires C16, H19, N2, O4, S1. mmax 3352, 3184, 1623, 1539,
1323, 1239 1155, 696 cm–1.
5.1.3.4.1-(4-(2-(2-Methoxybenzamido)ethyl)benzenesulpho-
nyl)-3-cyclohexylurea (2g). Compound 2g was synthesized
from 1g (334 mg, 1 mmol), anhydrous potassium carbonate
(415 mg, 3 mmol) and cyclohexylisocyanate (200 mg,
1.6 mmol). Yield, 138 mg (30.1%) of white rods; mp: 180–
182 °C (lit. mp: 184 °C, [33]); TLC Rf 0.57, 9.5:0.5
chloroform/methanol. m/z (relative intensity) 460 (M+. +1,
8%), 335 (100%), 135 (100%). mmax 3346, 2932, 1614, 1527,
1452, 1245, 1155 cm–1.
5.1.3. General procedure for the synthesis of 4-(2-(benza-
mido)ethyl)benzenesulfonyl-3-cyclohexylureas (2b, 2e–h)
Appropriate 4-(2-(benzamido)ethyl)benzenesulphona-
mide was refluxed in a stirred solution of anhydrous potas-
sium carbonate (415 mg, 3 mmol) in acetone (25 ml) for 1.5 h
under dry conditions. At that temperature, cyclohexylisocy-
anate (200 mg, 1.6 mmol) in dry acetone (4 ml) was added
dropwise. The resultant mixture was refluxed for another
16 h. Acetone was removed under vacuum and water (30 ml)
was added to the white residue. The mixture was acidified
with concentrated hydrochloric acid to pH 1. White precipi-
tate was filtered and recrystallized from methanol.
5.1.3.5.1-(4-(2-(3-Methoxybenzamido)ethyl)benzenesulpho-
nyl)-3-cyclohexylurea (2h). Compound 2h was synthesized
from 1h (334 mg, 1 mmol), anhydrous potassium carbonate
(415 mg, 3 mmol) and cyclohexylisocyanate (200 mg,
1.6 mmol). Yield, 122 mg (26.6%) of white rods; mp: 181–
182 °C (lit. mp: 184-185°C, [33]); TLC Rf 0.65, 95:5
chloroform/methanol. m/z (relative intensity) 460 (M+ ·+1,
4%), 335 (60%), 135 (100%). mmax 3322, 2932, 1623, 1584,
1521, 1446, 1245, 1155, 1026 cm–1.
5.1.4. 1-(4-(2-(2-Chlorobenzamido)ethyl)benzene-
sulphonyl)-3-cyclohexylurea (2c)
Compound 2c was prepared by refluxing 1c in a stirred
solution of anhydrous potassium carbonate (415 mg,
3 mmol) in acetone (25 ml) for 1.5 h under dry conditions. At
that temperature, cyclohexylisocyanate (200 mg, 1.6 mmol)
in dry acetone (4 ml) was added dropwise. The resultant
mixture was refluxed for another 16 h. On removal of acetone
under vacuum, 1 M sodium hydroxide (50 ml) was added.
The reaction mixture was filtered and then acidified with
5.1.3.1. 1-(4-(2-(4-Chloro-2-methoxybenzamido)ethyl)ben-
zenesulfonyl)-3-cyclohexylurea (2b). Compound 2b was
synthesized from 1b (368 mg, 1 mmol), anhydrous potas-
sium carbonate (415 mg, 3 mmol) and cyclohexylisocyanate
(200 mg, 1.6 mmol). Yield, 309 mg (62.2%) of white rods;
mp: 197–199 °C (lit. mp: 200–201 °C, [34]); TLC Rf 0.59,
95:5 chloroform/methanol. m/z (relative intensity) 494 (M+.

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
845
concentrated hydrochloric acid to pH 1. The resultant pre-
cipitate was filtered and recrystallized from methanol. Yield,
80 mg (8.6%) of white needles; mp: 202–204 °C (lit. mp:
204–205 °C, [33]); TLC Rf 0.58, 9.5:0.5 chloroform/
methanol. m/z (relative intensity) 464 (M+. +1, 5%), 339
(46%), 139 (100%). mmax 3358, 2932, 1635, 1593, 1527,
1449, 1221, 1155 cm–1.
tallized from water/methanol and decolourising charcoal was
used. Yield, 1.18 g (43.0%) of white rods; mp: 133–134 °C;
TLC Rf 0.18, chloroform. m/z (relative intensity) 214 (M+.
+1, 100%), 197 (53%), 133 (36%). mmax 3384, 3284, 1328,
1164 cm–1.
5.1.7. General procedure for the synthesis
of benzenesulphonyl-3-cyclohexylureas (4a–d)
5.1.5. 1-(4-(2-(3-Chlorobenzamido)ethyl)benzene-
sulphonyl)-3-cyclohexylurea (2d)
Benzenesulphonyl-3-cyclohexylureas (4a–d) were pre-
pared from appropriate benzenesulphonamides (3a–d), an-
hydrous potassium carbonate and cyclohexylisocyanate ac-
cording to the methods described for 2b and 2c.
Compound 2d was synthesized from 1d (676 mg,
2 mmol), anhydrous potassium carbonate (830 mg, 6 mmol)
and cyclohexylisocyanate (400 mg, 3.2 mmol). After reflux-
ing the reaction mixture for 16 h, acetone was removed under
reduced pressure and then 1 M HCl (50 ml) was added. The
white precipitate was filtered and recrystallized twice from
methanol. Yield, 374 mg (40.4%) of white needles; mp:
191–192 °C (lit. mp: 190–191 °C, [34]); TLC Rf 0.60, 9.5:0.5
chloroform/methanol. m/z (relative intensity) 464 (M+. +1,
2%), 339 (29%), 139 (100%). mmax 3384, 2930, 1634, 1528,
1445, 1224, 1154 cm–1.
5.1.7.1. 1-Benzenesulphonyl-3-cyclohexylurea (4a). Com-
pound 4a was prepared from 3a (471 mg, 3 mmol), anhy-
drous potassium carbonate (1.242 g, 9 mmol) and cyclohexy-
lisocyanate (600 mg, 4.8 mmol) according to the methods
described for 2c. Yield, 378 mg (44.7%) of white rods; mp:
187–189 °C (lit. mp: 188–189 °C, [38]); TLC Rf 0.47, chlo-
roform. m/z (relative intensity) 283 (M+. +1, 100%), 200
(43%), 158 (36%), 143 (90%), 141 (51%). mmax 3364, 3080,
2944, 1656, 1544, 1444, 1338, 1162 cm–1.
5.1.6. General procedure for the synthesis
of benzenesulphonamides (3a–d)
5.1.7.2.
1-(4-Methylbenzenesulphonyl)-3-cyclohexylurea
(4b). Compound 4b was prepared from 3b (513 mg,
3 mmol), anhydrous potassium carbonate (1.242 g, 9 mmol)
and cyclohexylisocyanate (600 mg, 4.8 mmol). After the
removal of acetone under vacuum, 2 M sodium hydroxide
solution (50 ml) and methanol (20 ml) were added. Yield,
414 mg (46.6%) of white rods; mp: 170–172 °C (lit. mp:
171–173 °C, [38]); TLC Rf 0.46, 9.5:0.5 chloroform/
methanol. m/z (relative intensity) 297 (M+. +1, 84%), 215
(46%), 172 (49%), 155 (100%). mmax 3328, 3172, 2932,
1641, 1527, 1446, 1338, 1155, 1029 cm–1.
Appropriate benzenesulphonyl chloride was added drop-
wise to a stirred 25% ammonia solution (45 ml). The result-
ant mixture was heated to 100 °C and cooled to room tem-
perature. The product was recrystallized from water.
5.1.6.1. Benzenesulphonamide (3a). Compound 3a was syn-
thesized from benzenesulphonyl chloride (9.2 g,
0.052 mole). Yield, 4.62 g (56.6%) of white plates; mp:
146–147 °C (lit. mp: 147–148 °C, [35]); TLC Rf 0.48, 9:1
chloroform/methanol. m/z (relative intensity) 158 (M+. +1,
100%), 141 (33%). mmax 3370, 3274, 1335, 1155 cm–1.
5.1.7.3. 1-(4-Ethylbenzenesulphonyl)-3-cyclohexylurea (4c).
Compound 4c was prepared from 3c (555 mg, 3 mmol),
anhydrous potassium carbonate (1.242 g, 9 mmol) and cyclo-
hexylisocyanate (600 mg, 4.8 mmol).Yield, 278 mg (29.9%)
of white rods; mp: 159–161 °C; TLC Rf 0.51, chloroform.
m/z (relative intensity) 311 (M+. +1, 47%), 229 (29%), 186
(8%), 169 (41%), 143 (36%), 105 (100%). Micro analysis
expected values C 58.0%, H 7.1%, N 9.0% and S 10.3%.
Found C 57.6%, H 7.3%, N 9.2% and S 10.4%. mmax 3316,
3228, 2920, 1676, 1522, 1444, 1330, 1158 cm–1.
5.1.6.2. 4-Methylbenzenesulphonamide (3b). Compound 3b
was synthesized from 4-methylbenzenesulphonyl chloride
(10.0 g, 0.052 mole). Yield, 7.83 g (88.1%) of colourless
rods; mp: 136–137 °C (lit. mp: 138 °C, [36]); TLC Rf 0.5, 9:1
chloroform/methanol. m/z (relative intensity) 172 (M+. +1,
100%), 154 (57%). mmax 3328, 3244, 3132, 1322, 1146,
1086 cm–1.
5.1.6.3. 4-Ethylbenzenesulphonamide (3c). Compound 3c
was synthesized from 4-ethylbenzenesulphonyl. The crude
product was recrystallized from water/methanol. Yield,
0.93 g (9.5%) of white plates; mp: 103–105 °C (lit. mp:
104–107.5 °C, [37]); TLC Rf 0.51, 9:1 chloroform/methanol.
m/z (relative intensity) 186 (M+. +1, 86%), 169 (64%), 105
(100%). mmax 3344, 3268, 1326, 1290, 1173 cm–1.
5.1.7.4. 1-(4-ter-Butylbenzenesulphonyl)-3-cyclohexylurea
(4d). Compound 4d was synthesized and purified from 3d
(639 mg, 3 mmol), anhydrous potassium carbonate (1.242 g,
9 mmol) and cyclohexylisocyanate (600 mg, 4.8 mmol).
Yield, 44 mg (6.1%) of white rods; mp: 210–211 °C; TLC Rf
0.53, chloroform. m/z (relative intensity) 239 (M+. +1,
100%), 256 (33%), 214 (31%), 197 (54%), 143 (40%), 133
(86%). Accurate mass 339.17461 requires C17 H27 N2 O3
S1. Micro analysis expected values C 60.3%, H 7.7%, N
8.3% and S 9.5%. Found C 60.0%, H 7.9%, N 8.4% and S
5.1.6.4. 4-ter-Butylbenzenesulphonamide (3d). Compound
3d was synthesized from 4-ter-butylbenzenesulphonyl chlo-
ride (3.0 g, 13 mmol) in dry acetone (12 ml) and 25%
ammonia solution (30 ml). The crude compound was recrys-

846
E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
Table 4
9.6%. mmax 3364, 3096, 2944, 1534, 1464, 1342, 1164,
1034 cm–1.
MM+ parameters for the geometry optimisation of glibenclamide and its
analogues
Bond
r0 (Å)
1.750a
1.700a
Kr (mdyn/Å)
4.0 [42]
10.0b
5.1.8. N-(2-Phenylethyl)-5-chloro-2-methoxybenzamide (5)
Compound 5 was synthesized by adding 5-chloro-2-
methoxybenzoyl chloride (4.10 g, 20 mmol) slowly to a
stirred solution of 2-phenylethylamine (2.42 g, 20 mmol) in
chloroform (20 ml) at 0 °C. The reaction mixture was filtered,
washed three times with 1 M HCl solution and the organic
layer dried with anhydrous sodium sulphate. Chloroform
was then removed under reduced pressure to give a viscous
yellow oil. This was triturated with hexane and the precipi-
tate obtained was recrystallized from diethyl ether. Yield,
0.28 g (9.7%) of colourless plates; mp: 61–63 °C (lit. mp:
61–64 °C [39]); TLC Rf 0.91, 8:2 chloroform/methanol. m/z
(relative intensity) 290 (M+. +1, 100%), 169 (100%), 105
(100%). mmax 3310, 2950, 1626, 1482, 1263 cm–1.
CA–S4
C3–S2
Angle
h0 (degree)
120.0 a [43]
104.0 a [44]
107.7 a [43]
120.00 a [41]
Kh (mdyn × Å/rad2)
0.420 [43]
CA–CA–S4
CA–S4–N2
CA–S4–O1
S4–N2–CO
1.440 [44]
0.597 [43]
0.695 b
Torsional angle b
CA–CA–CA–S4
CA–CA–S4–N2
CA–CA–S4–O1
CA–S4–N2–CO
CA–S4–N2–HV
O1–S4–N2–CO
H–CA–CA–S4
S4–N2–CO–N2
S4–N2–CO–O1
HV–N2–CO–CA
CO–CA–CA–O2
CO–N2–CO–CA
N2–CO–C3–NA
CO–C3–C3–NA
CO–C3–NA–C3
CO–C3–N2–HV
NA–C3–CO–O1
V1 (kcal/mol)
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
V2 (kcal/mol)
6.250
0.033
0.033
0.033
0.033
0.033
6.250
1.250
1.250
1.250
6.250
1.250
1.250
11.25
2.500
1.250
1.250
V3 (kcal/mol)
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
5.2. Computational procedures
5.2.1. Molecular mechanics protocol
All modelling studies were carried out using HyperChem
modelling package (version 5.11). Molecular mechanics en-
ergy minimizations were done using the MM+ force field of
HyperChem, an extension of Allinger’s MM2 force field
[40]. The missing force field parameters, associated with
glibenclamide and its analogues, were retrieved from litera-
ture or generated based on the structural information avail-
able from Cambridge Structural Database (for bond lengths
and angles) and following the standard procedures [41] (for
force constants). These are summarized in Table 4. Input
molecular systems were either taken from crystal structures
(where available from CSD) or model-built by modifying the
crystal structures of related molecular systems. Prior to con-
formational searching, molecular systems were initially en-
ergy minimized to remove residual strain and bad contacts
resulting from modified experimental structures and model
building.
a Based on the statistical analysis of the CSD data.
b HyperChem default scheme for generating unknown parameters.
options: torsion angles t1–t10 (Fig. 2); ranges for variation:
1 to 8 simultaneous variations; acyclic variation: 60°
to 180°; torsion flexing: 30° to 120°; search method:
usage directed; acceptance energy criterion: 6 kcal/mol
above best conformation. The newly generated conforma-
tions were excluded on the pre-optimization stage if atoms
were closer than 0.5 Å and if torsions were within 15° of an
already accepted conformation. The geometry optimized
conformations were considered duplicates of an already ac-
cepted conformation if exhibiting the torsions within 10°–
15° of already accepted conformation.
5.2.2. Geometry optimization
The molecular structures were optimized using the conju-
gate gradient algorithm (Polak-Ribiere) with the termination
condition being RMS gradient less than 0.1 kcal/(Å × mol)
(in vacuo) and 0.5 kcal/(Å × mol) (in explicit solvent (H2O)).
No cut-offs were used in vacuo, in explicit solvent the
switched cut-off was used with inner and outer radii of
8.5 and 12 Å, respectively. Electrostatic interactions were
modelled using the bond dipoles. For the explicit solvent
calculations, the cubic periodic box with the side of 25 Å was
used.
5.2.4. Analysis of conformational preferences in
experimental structures
Analysis of conformational preferences in experimental
structures was carried out by searching CSD for structures
containing relevant fragments with torsion angles of interest.
The database output files were examined visually to remove
the outliers and then subjected to GSTAT for a statistical
analysis. GSTAT is a multi-functional geometry program
within the CSD, which can perform geometrical calculations
for complete CSD entries and for substructural fragments
and carry out statistical and numerical analyses of fragment
geometry.
5.2.3. Monte Carlo conformational analysis
Conformational analysis was carried out at the molecular
mechanics level of theory using the Conformational Search
module of ChemPlus (version 1.6) with the following search

E. Yuriev et al. / European Journal of Medicinal Chemistry 39 (2004) 835–847
847
[10] S. Bavirti, J.A. Tayek, Metabolism 52 (2003) 407–412.
5.2.5. Comparative molecular field analysis
[11] R.A. Hill, S. Rudra, B. Peng, D.S. Roane, J.K. Bounds, Y. Zhang,
A. Adloo, T. Lu, Bioorg. Med. Chem. 11 (2003) 2099–2113.
[12] R. Rosskamp, K. Wernicke-Panten, E. Draeger, Diabetes Res. Clin.
Pract. 31 (Suppl) (1996) S33–S42.
[13] UK Prospective Diabetes Study Group, Lancet 352 (1998) 837–853.
[14] M.C. Riddle, J. Clin. Endocrinol. Metab. 88 (2003) 528–530.
[15] P.A. Brady, A. Jovanovic, J. Am. Coll. Cardiol. 42 (2003) 1022–1025.
[16] J.L. Challinor-Rogers, D.C. Kong, M.N. Iskander, G.A. McPherson,
J. Pharmacol. Exp. Ther. 273 (1995) 778–786.
[17] S.R. Byrn, A.T. McKenzie, M.M. Hassan, A.A. Al-Badr, J. Pharm.
Sci. 75 (1986) 596–600.
[18] L. Lins, R. Brasseur, W.J. Malaisse, Pharmacol. Res. 34 (1996) 9–10.
[19] L. Lins, R. Brasseur, W.J. Malaisse, Biochem. Pharmacol. 50 (1995)
1879–1884.
[20] W. Grell, R. Hurnaus, G. Griss, R. Sauter, E. Rupprecht, M. Mark,
P. Luger, H. Nar, H. Wittneben, P. Muller, J. Med. Chem. 41 (1998)
5219–5246.
[21] T.J. Hou, Z.M. Li, Z. Li, J. Liu, X.J. Xu, J. Chem. Inf. Comput. Sci. 40
(2000) 1002–1009.
[22] G. Klebe, U. Abraham, J. Med. Chem. 36 (1993) 70–80.
[23] H. Lanig, W. Utz, P. Gmeiner, J. Med. Chem. 44 (2001) 1151–1157.
[24] A.K. Chakraborti, R. Thilagavathi, Bioorg. Med. Chem. 11 (2003)
3989–3996.
[25] M. Meyer, F. Chudziak, C. Schwanstecher, M. Schwanstecher,
U. Panten, Br. J. Pharmacol. 128 (1999) 27–34.
[26] S.A. Raptis, E. Hatziagelaki, G. Dimitriadis, K.E. Draeger, C. Pfeiffer,
A.E. Raptis, Exp. Clin. Endocrinol. Diabetes 107 (1999) 350–355.
[27] Prescrire Int. 7 (1998) 106–107.
[28] W. Kramer, G. Muller, F. Girbig, U. Gutjahr, S. Kowalewski, D. Hartz,
H.D. Summ, Biochim. Biophys. Acta 1191 (1994) 278–290.
[29] G. Muller, D. Hartz, J. Punter, R. Okonomopulos, W. Kramer, Bio-
chim. Biophys. Acta 1191 (1994) 267–277.
[30] K. Hastedt, U. Panten, Biochem. Pharmacol. 65 (2003) 599–602.
[31] S. Hu, S. Wang, B. Fanelli, P.A. Bell, B.E. Dunning, S. Geisse,
R. Schmitz, B.R. Boettcher, J. Pharmacol. Exp. Ther. 293 (2000)
444–452.
Comparative molecular field analysis was carried out with
the CoMFA unit of the molecular modelling suite of pro-
grams Sybyl (version 6.2). Input structures were based on the
crystal structure of glibenclamide [17] and were modelled as
described above. Gasteiger-Marsili charges were assigned
for CoMFA runs. Two different positioning (database and
rigid field fit) were performed, the former yielding a better
alignment. In this approach, all the compounds were aligned
by atom-by-atom fitting using the heavy atoms of the termi-
nal phenyl ring and the carbon atom of the carbonyl group of
the benzamide substructure. Cross-validated CoMFA runs
were performed by means of PLS (leave one out) regression
analysis in order to obtain the optimal number of principal
components to be used in the subsequent analyses (the opti-
mum number of components obtained from cross-validated
PLS analysis and q2–the cross-validated correlation coeffi-
cient are shown in Table 2). The number of principal compo-
nents thus determined was used in the non-cross-validated
CoMFA runs so as to obtain the highest correlation coeffi-
cient R2 and the lowest standard error (Table 2). The pKi of
the inhibition of [3H]-glibenclamide binding in rat cerebral
cortex [16] was used as the biological activity indicator
(Table 1). CoMFA was carried out with the default settings
except for 2.0 kcal/mol column filtering applied at cross-
validation and no-validation runs. CoMFA variable used: S,
steric fields; E, electrostatic fields; P, octanol-water partition
coefficient (LogP), calculated with ChemPlus.
Acknowledgements
[32] H. Weber, W. Aumuller, K. Muth, R. Weyer, Benzenesulfonyl ureas,
1976 US patent 3932503.
[33] H. Weber, W. Aumuller, R. Weyer, K. Muth, F.H. Schmidt, Benzene-
sulfonyl ureas and process for their manufacture, 1969 US patent
3426067.
This work was supported in part by The Australian Post-
graduate Awards (APA) and Monash Graduate Scholarship
(MGS) to DCMK.
[34] H. Weber, W. Aumuller, K. Muth, R. Weyer, R. Heerdt, E. Fauland,
A. Bander, W. Pfaff, F.H. Schmidt, H. Stork, Arzneim-Forsch. 19
(1969) 1346–1362.
References
[35] J. Buckingham (ed.), Dictionary of Organic Compounds, Chapman
and Hall, New York, 1982.
[36] B.S. Furniss, A.J. Hannaford, P.W.G. Smith, A.R. Tatchell, Vogel’s
Textbook of Practical Organic Chemistry, Longman Scientific &
Technical, London, 1989.
[1] M.J. Coghlan, W.A. Carroll, M. Gopalakrishnan, J. Med. Chem. 44
(2001) 1627–1653.
[2] M.L.J. Ashford, in: N.S. Cook (Ed.), Potassium Channels: Structure,
Classification, Function and Therapeutic Potential, Ellis Horwood,
Chichester, 1990, pp. 300–325.
[37] Hilton-Davis, US patent 4874894, 1989.
[38] I. Fabricius, GB patent 11885013, 1970.
[3] F.M. Gribble, F. Reimann, Diabetologia 46 (2003) 875–891.
[4] V. Kecskemeti, Z. Bagi, P. Pacher, I. Posa, E. Kocsis, M.Z. Koltai,
Curr. Med. Chem. 9 (2002) 53–71.
[5] M.V. Mikhailov, E.A. Mikhailova, S.J. Ashcroft, FEBS Lett. 499
(2001) 154–160.
[6] L. Luzi, G. Pozza, Acta Diabetol. 34 (1997) 239–244.
[7] W. Kramer, G. Muller, F. Girbig, U. Gutjahr, S. Kowalewski, D. Hartz,
H.D. Summ, Diabetes Res. Clin. Pract. 28 (Suppl) (1995) S67–S80.
[8] C.E. Schotborgh, A.A. Wilde, Cardiovasc. Res. 34 (1997) 73–80.
[9] A.A. Konstas, M. Dabrowski, C. Korbmacher, S.J. Tucker, J. Biol.
Chem. 277 (2002) 21346–21351.
[39] G.R. Brown, A.J. Foubister, J. Med. Chem. 27 (1984) 79–81.
[40] U. Burkert, N.L. Allinger, Molecular Mechanics, ACS, Washington,
DC, 1982.
[41] N.L. Allinger, X. Zhou, J. Bergsma, J. Mol. Struct. (Theochem) 312
(1994) 69–83.
[42] J. Kao, C. Eyermann, E. Southwick, D. Leister, J.Am. Chem. Soc. 107
(1985) 5324.
[43] D.L. Mattern, X. Chen, J. Org. Chem. 56 (1991) 5903–5907.
[44] J.B. Nicholas, R. Vance, E. Martin, B.J. Burke, A.J. Hopfinger, J.
Phys. Chem. 95 (1991) 9803–9811.