SAR study of 2,3,4,14b-tetrahydro-1H-dibenzo[b,f]pyrido[1,2-d][1,4]oxazepines as progesterone receptor agonists

Article abstract of DOI:10.1016/j.bmcl.2007.12.065

We have developed a new class of progesterone receptor agonists having a tetracyclic dibenzo-oxazepine structure 1. In this paper, the synthesis and structure-activity relationships of this new class are described. This work led to the identification of p

Full text of DOI:10.1016/j.bmcl.2007.12.065

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1461–1467
SAR study of 2,3,4,14b-tetrahydro-1H-dibenzo[b,f]pyrido-
[1,2-d][1,4]oxazepines as progesterone receptor agonists
Paul P. M. A. Dols,a, Brigitte J. B. Folmer,a Hans Hamersma,b Cor W. Kuil,c
Hans Lucas,a Lourdes Ollero,a Jos B. M. Rewinkela and Pedro H. H. Hermkensa,*
aDepartment of Medicinal Chemistry, N.V. Organon, PO Box 20, 5340 BH Oss, The Netherlands
bDepartment of Molecular Design & Informatics, N.V. Organon, PO Box 20, 5340 BH Oss, The Netherlands
cDepartment of Molecular Pharmacology, N.V. Organon, PO Box 20, 5340 BH Oss, The Netherlands
Received 3 October 2007; revised 20 December 2007; accepted 22 December 2007
Available online 28 December 2007
Abstract—We have developed a new class of progesterone receptor agonists having a tetracyclic dibenzo-oxazepine structure 1. In
this paper, the synthesis and structure–activity relationships of this new class are described. This work led to the identification of
potent progesterone agonists up to 1 nM activity. Substitution at positions 6, 7 and 1 has proven to be crucial for activity, indicating
that probably these positions are involved in important interactions with the receptor.
Ó 2008 Elsevier Ltd. All rights reserved.
Over the last 10 years, considerable attention has been gi-
ven to the identification of non-steroidal ligands for ste-
roidal receptors.1 Synthetic steroidal ligands designed to
interact with these receptors have the downside that they
also often bind to other, structurally closely related, ste-
roid receptors as there is a substantial degree of homology
in their ligand binding domains. This cross-reactivity is
often the reason for side effect profiles. Different chemo-
types might have different properties (e.g., PK, physico-
chemical parameters or ADMET profiles) which might
cause different tissue distribution profiles and therefore
differences in functionality with respect to steroid ligands.
It has also been demonstrated2 that different chemotypes
alter binding modes, which in turn influences the binding
of cofactors (functional selectivity).
An important challenge, however, remains the identifi-
cation of ligands which show no cross-reactivity to other
steroid hormone receptors and therefore have, for exam-
ple, no androgenic (side effects: acne, hirsutism, fatigue,
mood swings/depression, rashes, weight gain) or oestro-
genic (side effects: fluid retention, breast tenderness,
weight gain, headaches, breakthrough bleeding, hyper-
tension) activity. Non-steroidal progestins have a high
potential of having reduced cross-reactivity and could
thereby show the desired functional selectivity. There-
fore, one potential therapeutic use of non-steroidal
progestins is to replace the steroidal progestins in the
next generation of OCs.5
In a HTS campaign the tetracyclic compound 1a was
identified as a confirmed hit which showed in vitro ago-
nistic PR activity with an EC50 in the CHO luciferase as-
say of 513 nM. The subsequent SAR exploration of
substituents on various positions on this tetracyclic core
is the topic of the present investigation.
Progestin ligands find their application mainly in the
area of oral contraceptives (OC) and a wide variety of
oncological and gynaecological indications. The OC
field has undergone evolutionary changes over the
years.3 Four generations of steroidal progestin agents
have been developed. The driving force has been to im-
prove desired activity and reduce adverse side effects.4
11
O
14a
H
7
Keywords: Non-steroidal; Progesterone receptor; Agonist.
*
Cl
N
NH2
5
Corresponding author. Tel.: +31 412 662339; fax: +31 412 662519;
1
e-mail: pedro.hermkens@organon.com
Present address: AtosOrigin, Papendorpseweg 93, Utrecht, The
Netherlands.
1a
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.12.065

1462
P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
A synthesis of 1-amino substituted tetracyclic core has
been described6 which yielded the key amine intermedi-
ate, 1a–d in 11 steps (Scheme 1). In the first five steps the
tetracyclic scaffold was prepared and the last six steps
were used to introduce the proper substituents and rela-
tive stereochemistry at positions 1 and 14a.
methods (e.g., anhydrides, amidation). The resulting
amide structures 9 were target of subsequent modifica-
tion. Treatment of the amide 9a with phosphorus penta-
sulfide afforded the thioamide 11. Alkylation of the
amide in 9a with methyl iodide in the presence of so-
dium hydride afforded tertiary amide 12. Reaction of
amine 1a with acetic anhydride gave acetamide 10.
The amidine compound 13 was prepared via amidine
formation of the amine functionality of 1a by treatment
with trifluoroacetonitrile. Urethane derivative 14a was
prepared by standard procedure via coupling corre-
sponding chloroformate with 1a. The thiourea com-
pound 14b was formed from reaction of isobutyl
isothiocyanate with 1a. Reductive alkylation of corre-
sponding aldehydes with 1a provided the amines 15.
Our first goal was the design of a shorter route to the key
amine intermediate. This resulted in a 6-step synthesis
(Scheme 2). Central key intermediates in the new synthesis
route are the tricyclic imines 5,7 which can be prepared via
a number of routes. Treatment of 5 with glutaric anhy-
dride yielded tetracycles 6. A Curtius rearrangement with
DPPA and an alcohol resulted in the formation of ure-
thane structures 7. Selective reduction of the amide func-
tionality was accomplished with borane in THF.
Treatment of the resulting urethane structures 8 with
HBr in acetic acid afforded the desired amines 1.
The direct chlorination of 9a or 9b with NCS in the pres-
ence of a catalytic amount of HCl afforded a mixture of
6- and 8-substituted compounds. The two regioisomers
(9i and 9j or 9k and 9l, respectively) were easily sepa-
rated (Scheme 4).
Scheme 3 shows the acylation of the amine functionality
of structure 1 via several different routine synthetic
X
X
X
R7
1a: O Cl
1b: O
1c: CH2 Cl
F
HX
+
6 steps
5 steps
R7
R7
N
H
N
H
R7
NO2
NH2
2a-d
1a-d
1d: S
Cl
Scheme 1.
X
X
OH
F
a
+
H
R7
R7
R7
NH2
NH
NH2
79-100%
O
R6
O
H
3e,f
4e,f
64-99%
5g-i
b
c
52-91%
5e,f
R6 R7
X
O
N
X
d
e: H Cl NMe
H
f:
H
g: H
Br
F
O
O
O
O
R7
R7
N
34-85%
COOH
R6
6e-i
R6
O
h: Br
H
Cl
5
i:
H
e
80-100%
R= small alkyl
X
N
X
N
X
f
g
H
H
H
H
R7
H
R7
R7
N
NH2
N
25-100%
N
19-92%
R6
OR
OR
R6
1e-i
R6
O
O
O
8e-i
7e-i
Scheme 2. Reagents and conditions: (a) HCOOH, reflux; (b) compound 5e: PPA, POCl3, compound 5f: PPA; (c) i—toluene, TsOH or EtOH; ii—
K2CO3, DMSO, 18-crown-6, 140 °C or Et3N, DMSO, 160 °C, microwave; (d) glutaric anhydride, xylene, 140 °C; (e) i—toluene, Et3N, DPPA, reflux;
ii—ROH; (f) BH3, THF; (g) HBr, AcOH, 100 °C.

P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
1463
O
O
O
Cl
Cl
Cl
c
H
N
N
H
H
N
H
H
H
N
N
N
S
11
CF3
CF3
X
13
10
O
HN
R6 R7
e
b
a: H Cl
O
O
b: H
c: H Cl CH2
d: H Cl
e: H Cl NMe
H
X
X
S
R7
R7
a
H
N
N
H
H
N
f: H
g:H
h: Br
Br
F
H
O
O
O
R6
R6
NH2
CF3
1a-h
9a-h
O
d
g
f
O
O
N
O
Cl
Cl
Cl
H
H
H
N
N
N
H
N
H
N
R16
W
CF3
R15
Y
O
12
Y W R15
15a: R16 = Me
15b: R16 = CHMe2
14a: O O Cl
b: S NH CHMe2
Scheme 3. Reagents and conditions: (a) (TFA)2O, pyridine, CH2Cl2 or CF3COOEt, Et3N, MeOH; (b) Ac2O, pyridine, CH2Cl2; (c) P2S10, dioxane,
reflux; (d) NaH, DMF, MeI; (e) F3CCN, THF; (f) compound 14a: ClCOOCH2Cl, CH2Cl2, satd NaHCO3 (aq); compound 14b: i-BuN@C@S, THF;
(g) NaB(OAc)3H, CH2Cl2, R16CHO.
Cl
Cl
O
O
O
a
Cl
Cl
N
H
H
N
N
N
H
H
H
N
H
N
Cl
CF
3
CF
CF
3
3
O
9a
O
O
9i
9j
Cl
O
O
O
a
N
H
H
N
N
N
H
H
H
N
H
N
Cl
CF
3
CF
CF
3
3
O
9b
O
O
9k
9l
Scheme 4. Reagents and condition: (a) NCS, 1 N hydrochloric acid, acetone.
The bromo derivatives 9f and 9h were subjected to the
Stille transformation, affording compounds with a vinyl
(9m and 9o) or acetyl functionality (9n and 9p)
(Scheme 5).
The (anti)-progestin activity of the compounds de-
scribed in Tables 1–4 was determined in the CHO
luciferase assay (see Supplementary material). Of the
two possible diastereoisomer pairs formed during the

1464
P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
synthesis, only the racemate of the diastereoisomer
with the cis configuration between the amine at posi-
tion 1 and the bridgehead hydrogen atom at position
14b demonstrated in vitro potency and the racemate
of the trans diastereomer turned out to be completely
inactive. The compounds described in Tables 1, 3 and
4 are racemates of the cis-diastereoisomer.
sult suggests that the trifluoro group might also be in-
volved in an important interaction with the
progesterone receptor. Combination of the observa-
tions that Cl as R7 and COCF3 as R1 were both po-
tential interaction points with the receptor resulted in
the very potent agonist 9a. However, this compound
has a mixed profile. It was decided to investigate first
the SAR on the amino functionality (R1 and R17) and
for this purpose R7 was kept constant as a chlorine.
The acetamide 10 was also a compound with a mixed
profile showing weak agonistic and reasonable antago-
nistic activity. The diminished potency of secondary
amide compound 12 indicated the importance of a
free NH. Shifting from the trifluoroamide to the more
After identification of 1a the compounds 1b and 15a,
published in 1996,6 were also tested. Both compounds
proved to be not active, thereby indicating that
R7 = Cl is involved in a relevant interaction with the
receptor. The only compound with R7 = H which
showed substantial activity is compound 9b. This re-
O
O
O
O
Br
b
a
H
N
H
H
N
N
H
N
H
N
H
N
CF3
CF3
CF3
O
9n
O
O
9m
9f
O
O
O
b
a
H
N
H
H
N
N
H
N
H
N
H
N
Br
9h
O
CF3
CF3
CF3
9o
O
9p
O
O
Scheme 5. Reagents and conditions: (a) PdCl2(PPh3)2, vinyltributyltin, toluene, 110 °C; (b) i—PdCl2(PPh3)2, (1-ethoxyvinyl)tributyltin, toluene,
reflux; ii—3 N hydrochloric acid, THF.
Table 1.
O
R17
H
R7
N
N
R1
Compound
R7
R1
R17
PRB luciferase
EC50 (nM)
i.a.
IC50 (nM)
i.a.
1a
Cl
H
H
H
H
H
H
H
H
Me
H
H
H
H
H
513
NA
10.6
222
4450
1.3
0.79
NA
NA
43
1b6
9a
H
Cl
H
COCF3
COCF3
COCH3
CSCF3
COCF3
C(@NH)CF3
COOCH2Cl
0.63
0.67
0.36
0.97
0.56
0.76
0.79
0.27
0.62
9b
10
Cl
Cl
Cl
Cl
Cl
Cl
H
193
133
11
12
450
10.3
598
NA
NA
1340
13
0.14
0.97
14a
14b
15a6
15b
CSNHCH2CHMe2
Et
CH2CHMe2
314
NA
Cl
0.74

P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
1465
Table 2.
O
14b
H
N
Cl
H
1
N
CF3
O
Compound
Config. C1
Config. C14b
PRB luciferase
EC50 (nM)
i.a.
IC50 (nM)
i.a.
9a
9a(SR)
9a(RS)
Racemate
10.6
3.0
54
0.63
0.75
0.65
43
26
0.27
0.19
S
R
S
R
Table 3.
X
H
H
Cl
N
N
CF3
O
Compound
X
PRB luciferase
EC50 (nM)
i.a.
IC50 (nM)
i.a.
9a
9c
9d
9e
O
10.6
2.0
2.8
9.8
0.63
0.74
0.70
0.66
43
26
0.27
0.20
CH2
S
NMe
polarizable trifluorothioamide 11 resulted in a 10-fold
increase of the potency while the trifluoroamidine
compound 13 proved to be equipotent to the amide
9a. The carbamate analogue 14a and the amine deriv-
ative 15b turned out to be micromolar active agonists,
while the thiourea derivative 14b is a full antagonist,
albeit a rather weak one.
It can be concluded that R1 is an important substitu-
ent by which potency and agonism versus antagonism
can be controlled. An electron-withdrawing group
like CXCF3 (X = O, S, NH) appears to be especially
favourable. Taken together, the effects of the substi-
tutions at the 1-amine and at position 7 suggest that
these two functionalities mimic the two carbonyl
groups of progesterone. The distance between the 3-
˚
and 20-keto groups in progesterone is 11.9 A,
whereas that between the 7-Cl substituent and the
˚
fluoroacetyl group in compound 17d is 11.5 A. If this
Figure 1. Two poses: assumed eutomer (1S,14bR) on progesterone
(orange) in the active site of PR. Yellow: orientation 1; blue:
orientation 2.
conjecture were true, additional (especially hydro-
philic) substituents in positions 2–4 and 11–14 should
hardly increase progestin activity, because of the pre-
dominantly lipophilic nature of the progesterone
receptor active site.
From the activities of the separated enantiomers8 we
learned that the eutomer has the (1S,14bR) configura-

1466
P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
Table 4.
R8
R7
O
N
H
H
N
CF3
R6
O
Compound
R6
R7
R8
PRB luciferase
EC50 (nM)
i.a.
IC50 (nM)
i.a.
9a
9g
9i
H
Cl
H
H
Cl
H
Cl
H
H
H
H
H
10.6
198
38
0.63
0.73
0.47
0.96
0.17
0.87
43
730
318
0.27
0.20
0.28
H
F
Cl
H
Cl
9j
Cl
H
1.6
9k
9l
H
Cl
993
19
2220
0.62
H
9m
9n
9o
9p
H
Vinyl
COMe
H
>1000
258
0.81
0.96
H
0.47
1.26
0.93
Vinyl
COMe
H
tion. The other enantiomer (1R,14bS) showed a more
than 15-fold reduction in potency (Table 2).
Investigation of substitution on the aromatic A ring
indicated that small polar and non-polar groups (e.g.,
chlorine (9l), vinyl (9o), acetyl (9p)) are permitted in
the 6 position (R6), resulting in potent agonists. 6,7-
Disubstitution resulted in the potent agonist 9j. For
the 7 position (R7) we clearly see that chlorine (9a) is still
the most active substituent, followed by fluorine (9g).
Substitution at this position by an acetyl (9n) or vinyl
(9m) moiety diminished the potency. Substitution of
the 8 position (R8) with chlorine (9k, 9i) is not favoured,
which might be an indication for restricted space at this
area of the receptor site.
To date, we have not been able to obtain an X-ray dif-
fraction structure of any of the ligands described above
bound to PR. In order to assess the positioning of the
tetracyclic scaffold in the progesterone receptor, a dock-
ing experiment was performed (see Supplementary
material). Compound 9a was selected as the model com-
pound. Figure 1 shows these minimal-energy poses for
the two orientations. When the two eutomers are com-
pared, orientation 2 (COCF3 group coinciding with
the C20 carbonyl) has a slightly lower interaction en-
ergy. This would seem to make orientation 2 more
likely. However, if we compare the resulting space occu-
pied within PR between the two orientations, we observe
that in orientation 1 (COCF3 group coinciding with the
C3 carbonyl), aromatic ring C occupies the space above
C10 and C11 in steroids. Indeed, active progestagens
with rather large 11b substituents are known.9 On the
contrary, in orientation 2, ring C occupies a space below
and beyond C7 and C8, and only slightly touches the
pocket frequently postulated10 below ring D of steroids.
Based on these data, an unequivocal choice between the
two orientations is hardly possible.
In conclusion, we have identified a novel series of po-
tent progesterone receptor agonists, the 2,3,4,14b-tet-
rahydro-1H-dibenzo[b,f]pyrido[1,2-d][1,4]oxazepines. SAR
studies demonstrated the importance of a 6- and/or
7-substituent for activity. Incorporation of a trifluoro-
acetamide or a trifluorothioamide substituent at the
N-1 position significantly improved the potency in this
series, producing the very potent agonists 9a, 9o, 9p
and 11. Probably these two functionalities, 6- or 7-
substituent and trifluoroacetamide or trifluorothioa-
mide, mimic the two carbonyl groups of progesterone.
Further developments of this series will be published
in due course.
Investigation of the bridgehead atom by compounds
9c–e demonstrated that C, S, O and N are all well tol-
erated but that CH2 and S are more potent than O
and NMe. Although X = O was not the most potent
option in the bridgehead position, the difference with
either methylene or sulfur is not substantial. There-
fore, we decided to keep oxygen as the most preferred
bridgehead, both for reasons of synthetic feasibility as
well as because of the fact it is expected to be less
prone to metabolic conversions or instability as is
known by potential oxidation of S and N and radical
formation of CH2.
Acknowledgments
The following co-workers are acknowledged for their
contribution to this project: Ingrid Zanders-van der
Ley, Arnold Broess, Ruud Tas, Robert Lazeroms, Co
Peters, Hemen Ibrahim, Rinie Bouwman, Abasin Mo-
mand, Lian de Bruin-Timmerman, Irene Korbl-Gol-
stein, Dirk de Weys, Jeffry Wisse, Marcel van der Rijst
and Marjo Schepers.

P. P. M. A. Dols et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1461–1467
1467
Fensome, A.; Wrobel, J.; Winneker, R.; Zhang, Z. Expert
Opin. Ther. Patents 2003, 13, 1839.
Supplementary data
6. Caulfield, W. L.; Gibson, S.; Rae, D. R. J. Chem. Soc.,
Perkin Trans. 1 1996, 545.
7. Note of caution: tricyclic imines with the general
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2007.12.065.
structure
5 are reported to be potentially strongly
irritant and sensitising compounds. See the aquatic
toxicity of the sensory irritant and riot control agent
dibenz(b,f)-1,4-oxazepine (CR). Johnson, Dennis W.;
Haley, Mark V.; Landis, Wayne G. ASTM Special
Technical Publication 1990, 1096 (Aquat. Toxicol. Risk
Assess.: 13th Vol.), 176.
References and notes
1. Reviews on non-steroidal ligands (a) Coghlan, M. J.;
Kort, M. E. Expert Opin. Ther. Patents 1999, 1523; (b)
Lin, X.; Huebner, V. Curr. Opin. Drug Discov. Devel.
2000, 3, 383; (c) Buijsman, R. C.; Hermkens, P. H. H.; van
Rijn, R. D.; Stock, T.; Teerhuis, M. M. Curr. Med. Chem.
2005, 12, 1017; (d) Hermkens, P. H. H.; Kamp, S.; Lusher,
S.; Veeneman, G. H. Idrugs 2006, 7, 488.
2. Veeneman, G. H. Curr. Med. Chem. 2005, 12, 1077.
3. Madauss, K. P.; Stewart, E. L.; Williams, S. P. Med. Res.
Rev. 2007, 27, 374.
4. (a) Schindler, A. E.; Campagnoli, C.; Druckman, R.;
Huber, J.; Pasqualini, J. R.; Schweppe, K. W.; Thijssen, J.
H. H. Maturitas 2003, 46S1, S7; (b) Edwards, L. A.
Formulary 2004, 39, 104.
5. (a) Zhi, L.; Marschke, K. B. Expert Opin. Ther. Patents
1999, 9, 695; (b) Allan, G.; Macielag, M. Expert Opin.
Ther. Patents 1999, 9, 955; (c) Woolfrey, J. R.; Avery, M.
A. Chemtracts-Org. Chem. 2000, 13, 20; (d) Zhang, P.;
8. Racemate 1a was separated into its enantiomers by
chiral HPLC (Chiracel OJ column with heptanes/ethanol
85:15 as eluent). Both enantiomers were converted to
the corresponding trifluoroamides 9a. The absolute
stereochemistry has been determined by means of
VCD (see Supplementary material). It was established
with many compounds that the activities of the eutom-
ers correlate well with those of the racemates (data not
shown).
9. Schoonen, W. G. E. J.; Deckers, G.; de Gooijer, M. E.; de
Ries, R.; Mathijssen-Mommers, G.; Hamersma, H.;
Kloosterboer, H. J. J. Steroid Biochem. Mol. Biol. 2000,
74, 109.
10. (a) van der Sijde, D.; Kooreman, H. J.; Jaitly, K. D.;
Marx, A. F. J. Med. Chem. 1972, 15, 909; (b) Solo, A. J.;
Kapoor, J. N. J. Med. Chem. 1973, 16, 270.