Discovery and optimization of highly ligand-efficient oxytocin receptor antagonists using structure-based drug design

Article abstract of DOI:10.1016/j.bmcl.2008.11.064

A novel oxytocin antagonist was identified by 'scaffold-hopping' using Cresset FieldScreen molecular field similarity searching. A single cycle of optimization driven by an understanding of the key pharmacophoric elements required for activity led to the

Full text of DOI:10.1016/j.bmcl.2008.11.064

Bioorganic & Medicinal Chemistry Letters 19 (2009) 990–994
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and optimization of highly ligand-efficient oxytocin receptor
antagonists using structure-based drug design
a
a
a
b
Benjamin R. Bellenie a, , Nicholas P. Barton , Amanda J. Emmons , Jag P. Heer , Cristian Salvagno
*
a GlaxoSmithKline Pharmaceuticals, New Frontiers Science Park (North), Coldharbour Road, Harlow, Essex CM19 5AD, England, United Kingdom
b GlaxoSmithKline S.p.A., Medicines Research Centre, Via Fleming 2, Verona, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
A novel oxytocin antagonist was identified by ‘scaffold-hopping’ using Cresset FieldScreen molecular field
similarity searching. A single cycle of optimization driven by an understanding of the key pharmaco-
phoric elements required for activity led to the discovery of a potent, selective and highly ligand-efficient
oxytocin receptor antagonist. Selectivity over vasopressin receptors was rationalized based on differences
in the structure of the natural ligands.
Received 13 October 2008
Revised 17 November 2008
Accepted 18 November 2008
Available online 24 November 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Oxytocin antagonist
Cresset FieldScreen
Oxytocin
Vasopressin
Molecular field
Hit discovery
Structure-based drug design
The cyclic nonapeptide hormone oxytocin 1 acts at the seven-
transmembrane oxytocin (OT) receptor and is associated with
numerous physiological roles including control of uterine contrac-
tions, regulation of sexual response, behaviour and emotions.1,2
The endogenous oxytocin peptide is used clinically to induce la-
bour in pregnant women,3 and the oxytocin antagonist atosiban
is an established acute treatment to delay the onset of pre-term la-
bour.4 Although there are no confirmed OT receptor subtypes, OT
receptors show close structural similarity to vasopressin receptors
(particularly V1a), and the oxytocin 1 and vasopressin 2 peptides
only differ by two amino acids (Fig. 1).1 This similarity is demon-
strated by the binding affinity of oxytocin 1 at the two receptors
(OT Ki = 6.8 nM; V1a Ki = 34.9 nM).5
Atosiban 3 (Fig. 2) is non-selective (OT Ki = 397 nM; V1a
Ki = 4.7 nM),5 and requires dosing by continuous infusion for up
to 48 h,4 hence there has been significant interest in developing
an orally active, selective OT antagonist.2,6
A number of chemotypes have been investigated (Fig. 2), includ-
ing triazole 4 discovered by researchers at Pfizer,7 and diketopiper-
azine (DKP) GSK221149 (5) which shows greater than 1400-fold
selectivity over vasopressin receptors, and a greater than 10-fold
increase in potency at the oxytocin receptor compared to atosiban,
despite a greater than 50% reduction in molecular weight.8
OH
NH
O
H
N
H2N
N
H
O
O
O
O
S
S
NH
O
N
N
H
O
H
NH2 NH2
O
N
N
O
H2N
H
O
O
oxytocin,1
OH
O
H
N
H2N
N
H
O
O
NH
H2N
O
O
NH
S
S
HN
NH
O
N
N
H
O
H
N
NH2 NH2
O
N
O
H2N
H
O
O
vasopressin,2
* Corresponding author. Tel.: +44 1279 622000.
E-mail address: ben.r.bellenie@gsk.com (B.R. Bellenie).
Figure 1. Comparison of structures of oxytocin and vasopressin.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.11.064

B. R. Bellenie et al. / Bioorg. Med. Chem. Lett. 19 (2009) 990–994
991
O
O
H
N
N
H
O
NH
S
S
O
O
H2N
O
O
N
NH OH
NH2
H
N
N
H2N
H
O
N
O
H
O
O
atosiban, 3
O
N
N
N
N
OMe
O
O
O
N
N
N
H
HN
O
N
Figure 3. Cresset FieldScreen representation of GSK221149. Blue field points
(spheres) highlight energy minima for a positively charged probe, red for a negative
probe. Yellow spheres represent an attractive van der Waals minima for a neutral
probe and orange spheres represent hydrophobic centroids. Oxygen atoms are
shown in red, nitrogen in blue. The size of the points is related to the strength of the
interaction.
OMe
4
GSK221149, 5
Figure 2. Oxytocin antagonists.
A virtual screening approach was adopted to search for a novel
chemical series of small molecule antagonists. The extended elec-
tron density representation offered by the Cresset XED forcefield
provides a way to characterize the calculated field around a mole-
cule. In this approach, probes are used to identify interaction min-
ima and depicted as molecular field points showing the location
and significance of the minima.9
O
N
O
6
An appropriate conformation of GSK221149 was required in
order to generate molecular field points with which to perform a
virtual screen. Docking studies were performed to generate a puta-
tive binding mode for the compound in a rhodopsin based homol-
ogy model of the oxytocin receptor. The initial ligand conformation
was based on the small molecule crystal structure of GSK221149
and fitted to the receptor model. Receptor–ligand complexes were
then optimized with molecular mechanics and the preferred
hypothesis selected. The resulting conformation of the ligand was
used to perform a virtual screen of two million compounds in
the FieldScreen system provided by Cresset (Fig. 3). The resulting
hit list of optimized overlays was ranked using the similarity score
of the field points and after being filtered to remove compounds
with undesirable functionalities and properties, the ranked list
was inspected visually in 3D. Two thousand five hundred overlays
were assessed in the context of the SAR established for the DKPs
which showed that a variety of groups were tolerated in the mor-
pholine position, while the SAR around the indane group is rela-
tively tight.8 219 compounds were selected for screening.
The biaryl amide, 6, was one of fourteen active oxytocin antag-
onists identified from this search, and demonstrated a fpKi of 7.3 in
a functional FLIPR assay using recombinant human oxytocin recep-
tor CHO cells. The low molecular weight (335) and very low polar
surface area (29.5)10, made this an extremely attractive template
with good ligand efficiency (LE = 0.40, BEI = 22, SEI = 25).11 The
synthetic tractability of the compound also made further optimiza-
tion readily accessible.
Figure 4. Cresset FieldScreen overlay of biaryl amide 6 (grey with spherical field
points, structure shown at top) and GSK221149 (5) (green with octahedral field
points).
contributing atoms which overlay, but the fields they present for
interaction with a receptor.
Multiple conformations are possible for both the biaryl amide
and the DKP, and before designing modifications to the lead com-
pound 6, alternative overlays were considered in order to be able
to generate and test multiple binding hypotheses. Biaryl amide 6
was mapped to a consensus pharmacophore derived from a set
of published oxytocin antagonists including triazole 4 and diketo-
piperazine 5. The overlay of 6 with 5 shown in Figure 5 suggests an
alternative possible alignment of these two molecules.
The field print image (Fig. 4) used to identify 6 shows one pos-
sible overlay of this molecule with GSK221149. The amide func-
tionality in 6 generates a molecular field point co-located with
that from one of the carbonyls of the diketopiperazine (DKP). The
THF moiety also has a projected nucleophilic/acceptor feature
which is mimicked by the other DKP acceptor. This demonstrates
an advantage of a field based approach in which it is not the

992
B. R. Bellenie et al. / Bioorg. Med. Chem. Lett. 19 (2009) 990–994
Table 1
Effect of biaryl substitution on potency
R2
R4
N
O
R5
12
O
R1
R3
R3
R1
R2
R4
R5
fpKi
6
–H
–H
–H
–H
–H
–H
–Me
–H
–H
–H
–H
–H
–H
–H
–Me
–H
–H
–H
–H
–Cl
–OMe
–SO2Me
–CN
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–Cl
–H
–H
–H
–H
–H
–H
–H
–H
–H
–Cl
7.3
8.5
7.2
6.2
5.7
9.1
6.9
8.1
7.1
12a
12b
12c
12d
12e
12f
12g
12h
Table 2
Effect of nitrogen incorporation on potency
Figure 5. Mapping of the DKP compound 5 (orange) and biaryl amide 6 to a
consensus pharmacophore of published oxytocin antagonists. The matched features
shown are three H-bond donor features (green), aromatic pi-system (brown) and
hydrophobe (cyan).
R2
R4
N
R5
12
O
O
R1
R3
A3
In both cases, the distal aromatic of the biaryl system overlays
with the indane of the DKP. The biaryl adopts a ring-twist, suggest-
ing that locking this conformation by incorporating a group ortho
to the ring junction may be beneficial. The relative position of
the cyclopropylmethyl and tetrahydrofuranyl (THF) groups is dif-
ferent in the two overlays; in the pharmacophore-derived instance
the THF oxygen is superposed with the hydrogen bond acceptor of
the morpholine amide of 5, and the cyclopropylmethyl group of 6
is co-located with the isobutyl group in the DKP. It has been
proposed that the isobutyl group of the DKP mimics the isoleucine
group in oxytocin.12 This is replaced by a phenylalanine in vaso-
pressin; hence this position may be important for selectivity.
Based on these hypotheses, small arrays were designed and
synthesized as shown in Scheme 1. Key SAR findings are sum-
marized in Tables 1 and 2. The introduction of a chlorine group
in the distal ring (12a) or a methyl group in the proximal ring
X
A1
A2
fpKi
D
fpKi vs 6
6
O
O
O
O
O
N(Et)
–CH–
–CH–
–N–
–CH–
–CH–
–CH–
–CH–
–C(Me)–
–CH–
–N–
–CH–
–CH–
–CH–
–CH–
–CH–
–N–
7.3
9.1
6.1
5.5
<5.5
<5.5
À
12e
13a
13b
13c
13d
+1.8
À1.2
À1.8
>À1.8
>À1.8
–C(Me)–
–CH–
potency. Small lipophilic groups are tolerated in other positions
but have a smaller effect (12f–h). Polarity in general appears to
be poorly tolerated within the biaryl portion of the molecule (
12c,d, 13a–c). Replacement of the THF–oxygen of compound
12e with N-ethyl (13d) results in a complete loss of activity
(fpKi < 5.5).
Compound 12e was selected for further profiling and shows
excellent selectivity versus vasopressin V1a in the functional assay
(Table 3). The replacement of cyclopropylmethyl with benzyl con-
ferred a reversal of selectivity, with 14 showing higher potency at
V1a than OT. This effect is consistent across all analogues prepared
as shown in Chart 1 and Table 3. The data supports the proposal
above that this group can be superposed with the DKP isobutyl
group, the isoleucine group of oxytocin or the benzyl group of
vasopressin.
The binding affinity of 12e at oxytocin, and selectivity versus
multiple vasopressin receptors was investigated using a filtration
binding assay.13 10-fold selectivity was observed versus V1a, and
the compound showed greater selectivity versus V1b and V2
(Table 4).
In summary a novel, low molecular weight oxytocin antagonist
6 has been identified using molecular field print similarity search-
ing. Multiple binding hypotheses were used to design an array
which led to the discovery of 12e, an OT receptor antagonist with
nM potency and selectivity over vasopressin receptors in only one
cycle of optimization. The molecule has low molecular weight and
polar surface area making it a highly efficient ligand (LE = 0.48,
BEI = 26, SEI = 31).
(12e) in positions which can cause
a
ring-twist increase
O
a)
O
O
H
N
H
NH2
7
8
9
b)
b)
O
O
c)
N
N
R1
R1
O
O
R2
Br
11
10
Scheme 1. Reagents and conditions: (a) i—Na2SO4, DCM, 100 °C, 10 min; ii—NaBH4,
DCM/MeOH, rt, 1 h; (b) ArCO2H, PS-EDC, HOAt, DCM/NMP/THF, rt, 2 days then add
PS-NCO, MP-carbonate; (c) R2–B(OH)2, Pd(PPh3)2Cl2, Na2CO3, DME/EtOH/H2O,
140 °C, 15 min.

B. R. Bellenie et al. / Bioorg. Med. Chem. Lett. 19 (2009) 990–994
993
Table 3
Effect on vasopressin V1a selectivity of replacement of methylcyclopropyl group with benzyl
R1
N
R2
O
O
R2
R1 = benzyl
V1a fpKi
R1 = methylcyclopropyl
OT fpKi
7.4
OT – V1a
OT fpKi
V1a fpKi
OT – V1a
2
15a
15b
15c
15d
15e
15f
15g
15h
15i
8.5
8.6
7.9
7.3
7.1
8.4
6.8
6.3
5.8
À1.1
12e
16b
12b
12f
16e
16f
12d
13b
16i
9.1
8.5
7.2
6.9
6.7
6.5
5.7
5.5
5.2
7.1
7.8
7.2
6.2
6
À0.8
À0.7
À1.1
À1.1
À1.3
À0.8
À0.8
À0.6
7.6
6.1
5.7
5.1
5.4
5.4
5.4
5
0.9
1.1
1.2
1.6
1.1
0.3
0.1
0.2
F
O
CF3
N
CF3
CN
N
7.1
6
N
5.5
5.2
O
R1
Table 4
N
Selectivity profiling of compound 12e
R2
O
O
OT
V1a
V1b
V2
Filtration binding pKi
Functional FLIPR fpKi
8.9
9.1
7.9
7.1
<4.9
5.9
6.6
—
9
8.5
8
7.5
7
Acknowledgement
6.5
6
The authors acknowledge Ami Simmons and thank her for her
contribution to the preparation of example compounds.
5.5
5
References and notes
5
5.5
6
6.5
7
7.5
8
8.5
V1a fpKi
1. Gimpl, G.; Fahrenholz, F. Physiol. Rev. 2001, 81, 629.
2. Borthwick, A. D. Annu. Rep. Med. Chem. 2006, 41, 409.
3. Stubbs, T. M. Clin. Obstet. Gynecol. 2000, 43, 489.
4. Tsatsaris, V.; Carbonne, B.; Cabrol, D. Drugs 2004, 64, 375.
Chart 1. Plot showing OT and V1a potency where R1 = methylcyclopropyl (blue)
and R1 = benzyl (red). Lines link molecules with the same R2 groups. Structures
shown in Table 3.

994
B. R. Bellenie et al. / Bioorg. Med. Chem. Lett. 19 (2009) 990–994
5. Åkerlund, M.; Bossmar, T.; Brouard, R.; Kostrzewska, A.; Laudanski, T.;
9. Cheeseright, T.; Mackey, M.; Rose, S.; Vinter, J. G. J. Chem. Inf. Model. 2006, 46,
665; Cheeseright, T. J.; Mackey, M. D.; Melville, J. L.; Vinter, J. G. J. Chem. Inf.
Model 2008, 48, 2108.
Lemancewicz, A.; Secradeil-Le Gal, C.; Steinwall, M. Brit. J. Obstet. Gynaecol.
1999, 106, 1047.
6. Allen, M. J.; Livermore, D. G. H.; Mordaunt, J. E. Prog. Med. Chem. 2006, 44, 331.
7. Brown, A. D.; Ellis, D.; Smith, C. R. U.S. Patent 7,291,640, 2007.
8. Liddle, J.; Allen, M. J.; Borthwick, A. D.; Brooks, D. P.; Davies, D. E.; Edwards, R.
M.; Exall, A. M.; Hamlett, C.; Irving, W. R.; Mason, A. M.; McCafferty, G. P.;
Nerozzi, F.; Peace, S.; Philp, J.; Pollard, D.; Pullen, M. A.; Shabbir, S. S.; Sollis, S.
L.; Westfall, T. D.; Woollard, P. M.; Wu, C.; Hickey, D. M. B. Bioorg. Med. Chem.
Lett. 2008, 18, 90.
10. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.
11. Abad-Zapatero, C. Expert Opin. Drug Discov. 2007, 2, 469.
12. Borthwick, A. D.; Davies, D. E.; Exall, A. M.; Livermore, D. G.; Sollis, S. L.;
Nerozzi, F.; Allen, M. J.; Perren, M.; Shabbir, S. S.; Woollard, P. M.; Wyatt, P. G. J.
Med. Chem. 2005, 48, 6956.
13. Filtration binding was determined by displacement of 3H-oxytocin from
recombinant human oxytocin receptors stably transfected in CHO cells.