Rational design of orally-active, pyrrolidine-based progesterone receptor partial agonists

Article abstract of DOI:10.1016/j.bmcl.2009.06.055

Using the X-ray crystal structure of an amide-based progesterone receptor (PR) partial agonist bound to the PR ligand binding domain, a novel PR partial agonist class containing a pyrrolidine ring was designed. Members of this class of N-alkylpyrrolidines

Full text of DOI:10.1016/j.bmcl.2009.06.055

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4777–4780
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Rational design of orally-active, pyrrolidine-based progesterone receptor
partial agonists
a,
Scott K. Thompson a, , David G. Washburn , James S. Frazee a, Kevin P. Madauss c, Tram H. Hoang a,
*
Leahann Lapinski a, Eugene T. Grygielko b, Lindsay E. Glace b, Walter Trizna b, Shawn P. Williams c,
Chaya Duraiswami d, Jeffrey D. Bray b, Nicholas J. Laping b
a Department of Chemistry, Metabolic Pathways Centre for Excellence in Drug Discovery, GlaxoSmithKline Pharmaceuticals, 709 Swedeland Road, King of Prussia, PA 19406, USA
b Department of Biology, Metabolic Pathways Centre for Excellence in Drug Discovery, GlaxoSmithKline Pharmaceuticals, 709 Swedeland Road, King of Prussia, PA 19406, USA
c Department of Computational and Structural Chemistry, Molecular Discovery Research, GlaxoSmithKline Pharmaceuticals, PO Box 13398, Research Triangle Park, NC 27709, USA
d Department of Computational and Structural Chemistry, Molecular Discovery Research, GlaxoSmithKline Pharmaceuticals, 1250 South Collegeville Road, PO Box 5089, Collegeville, PA
19426, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Using the X-ray crystal structure of an amide-based progesterone receptor (PR) partial agonist bound to
the PR ligand binding domain, a novel PR partial agonist class containing a pyrrolidine ring was designed.
Members of this class of N-alkylpyrrolidines demonstrate potent and highly selective partial agonism of
the progesterone receptor, and one of these analogs was shown to be efficacious upon oral dosing in the
OVX rat model of estrogen opposition.
Received 11 February 2009
Revised 10 June 2009
Accepted 15 June 2009
Available online 17 June 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Progesterone receptor
Agonist
Partial agonist
Endometriosis
Nuclear receptor
Hormone receptors
The progesterone receptor (PR) is a ligand-activated transcrip-
tion factor that, along with the estrogen receptor, plays an impor-
tant role in modulating endometriosis. In the pathogenesis of this
disorder, the ectopic endometrial tissue proliferates in response
to estrogen,1 and current therapies that functionally antagonize
estrogen production or action are efficacious in alleviating symp-
toms. One such therapy that acts by antagonizing estrogen action
is treatment with full progesterone receptor agonists (progestins).
However, this method of therapy has associated with it undesir-
able side effects such as break-through bleeding and breast tender-
ness,2 as well as additional adverse effects attributable to the poor
steroid receptor selectivity of the marketed progestins.3 Since the
level of intrinsic PR agonism required to fully oppose estrogen ac-
tion is unknown, a selective PR partial agonist has the potential to
match the efficacy of progestin therapy, but have reduced side ef-
fects related to PR agonism and steroid receptor selectivity. The
work described herein is the result of efforts directed at this
strategy.
The goal of this effort was to identify small molecule PR partial
agonists with high potency at PR and high selectivity over other
steroid receptors: the androgen (AR), glucocorticoid (GR), mineral-
ocorticoid (MR), and estrogen (ER) receptors. The chemistry effort
initiated with the racemic dimethyl amide 1 (Fig. 1). This com-
pound demonstrated a PR binding IC50 of 125 nM in the FP compe-
tition binding assay, partial agonist activity in a CV-1 reporter
assay, and 10–100-fold selectivity over other steroid receptors.4
However, it was found that the dimethyl amide functionality was
F3C
O
N
F3C
NC
N
1
* Corresponding author. Tel.: +1 610 270 4941.
E-mail address: dave.g.washburn@gsk.com (D.G. Washburn).
All enquires and communications should be forwarded to David G. Washburn.
Figure 1. Activity and selectivity data for racemic PR partial agonist ligand 1.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.06.055

4778
S. K. Thompson et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4777–4780
F3C
R1
N
N
N
F3C
NC
X
Figure 3. Bound conformation of compound 22 in the PR LBD.
O
N
NC
no-1-tert-butoxycarbonylpyrrolidine provided pyrrolidine 3. Re-
moval of the Boc protecting group from 3 followed by reductive
methylation on the pyrrolidine nitrogen and subsequent alkylation
on the aniline nitrogen provided compounds 4–11. Target com-
pounds 12–22 were prepared by alkylation of 3 on the aniline
nitrogen with an alkyl or benzyl halide, followed by removal of
the Boc protecting group and alkylation of the resulting free
pyrrolidine.
Modification of the substituents on the pyrrolidine and aniline
nitrogens revealed some interesting SAR trends, much of which
can be explained in light of the X-ray crystal structure of com-
pound 22 bound in the PR LBD (Fig. 3).5 The preference for benzyl
groups on the aniline nitrogen, particularly those with ortho substi-
tutions (Table 1), can be understood by examining this structure.
Smaller groups such as those found in compounds 4–6 clearly
would not adequately fill the binding pocket for the substituent
on the aniline nitrogen, but the shape of the pocket does appear
well suited to accommodate a benzyl group. Within this pocket,
there is a well-defined binding site for a relatively small ortho sub-
stituent such as chlorine, methyl or trifluoromethyl. Interestingly,
the trifluoromethyl substitution led to a significant decrease in
partial agonist efficacy. However, para substitution is not well
accommodated in this pocket, and meta substitution is blocked
by the back of the pocket and by the surface created by Leu-718
and Phe-794 (not shown). The sharp decrease in binding affinity
observed when the phenyl portion of the benzyl group is replaced
with a 2-pyridyl group (compound 8 vs compound 20) is likely a
S
( )-1
Figure 2. Bound conformation of (S)-1 in the ligand binding domain of PR and
proposed structure of a pyrrolidine-based compound.
metabolically unstable, making compounds such as 1 unsuitable
for use in animal pharmacology models. Any attempts to modify
the groups on the amide nitrogen resulted in substantial decreases
in PR binding potency or selectivity. Therefore, efforts were fo-
cused on identification of an isosteric replacement for the amide
functionality.
Compound 1 was co-crystallized with the PR ligand binding do-
main (PR LBD), and the S-enantiomer was found to be bound in the
binding site.5 Additionally, there were two key observations
regarding the conformation of the bound ligand that led to the de-
sign of an isosteric replacement for the amide group (Fig. 2). First,
the carbonyl oxygen of the amide in 1 did not appear to participate
in any interactions with residues within the binding pocket, sug-
gesting that it could be replaced with another atom or functional-
ity. The second key observation was that the amide carbonyl bond
and the C
a-methyl bond existed in an eclipsed conformation rela-
tive to one another, suggesting that the four atoms spanning the
methyl carbon and the carbonyl oxygen could be tied together into
a five-membered ring. This hypothesis led to the design of a series
of N-alkylpyrrolidine-based compounds.
a
-
The N-alkylpyrrolidines were synthesized as shown in Scheme
1.12 Reaction of 2-chloro-4-fluorobenzonitrile (2) and (S)-3-ami-
Table 1
H
In vitro data for substituted aniline analogs
Cl
N
Cl
F
a
Boc
N
R
NC
NC
N
Cl
3
N
2
NC
b, c, d
d, b, then
c, e, f or g
Compds
R
PR binding IC50, nM
PR CV-1 EC50, nM (%P4)
R1
4
5
6
7
20
8
12
9
10
11
21
22
Et
Pr
Bu
1300
400
250
79
>10,000
>10,000
>10,000
>10,000
1300 (58)
>10,000
1.6 (67)
>10,000
>10,000
>10,000
200 (73)
3 (44)
R1
R3
R2
Cl
N
Cl
N
N
N
Cyclohexylmethyl
Benzyl
CH2-2-Pyridyl
o-Cl-Benzyl
m-Cl-Benzyl
p-Cl-Benzyl
o,o-Di-Cl-benzyl
o-Me-Benzyl
o-CF3-Benzyl
40
NC
NC
3000
20a
160
400
40
12-22
4-11
Scheme 1. Reagents and conditions: (a) (S)-3-amino-1-tert-butoxycarbonylpyrr-
olidine, NaHCO3, H2O/DMSO, 90 °C, 22 h; (b) trifluoroacetic acid, CH2Cl2, 25 °C,
1.5 h; (c) H2CO, NaBH4, MeOH/H2O, 25 °C, 30 min; (d) R1CH2-halide, NaH, DMF,
25 °C, 1 h; (e) R1CHO, NaBH(OAc)3, AcOH, CH2Cl2, 25 °C, 2 h (R3 = H); (f) R2CH2Br,
K2CO3, CH3CN, 75 °C, 15 min (R3 = H); (g) (i) R2R3C(CN)OH, 4 Å sieves, EtOH, 60 °C,
2 h; (ii) NaBH4, EtOH, 25 °C, 16 h.
50
16a
a
At or below the tight binding limit of the PR binding assay.

S. K. Thompson et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4777–4780
4779
tions of progesterone8 and compound 22 (Fig. 4). The pyrrolidine
ring in compound 22 forces a significant movement in the Met-
909 side chain from its position in the progesterone structure,
which is expected to produce the partial agonist response that is
observed for compound 22.
Table 2
In vitro data for substituted pyrrolidine analogs
Cl
Compound 22 was further evaluated in additional binding and
functional assays to determine its selectivity for PR vs. the other
steroid receptors (Table 3). Compound 22 binds to PR with
P500-fold higher affinity than for estrogen receptor b and >15-fold
over AR. Since compound 22 is at or near the tight binding limit in
the PR binding assay, a better measure of selectivity over AR and
other NRs may be by the CV-1 assay. Accordingly, compound 22
is >400-fold selective for PR in AR, MR, and GR functional assays,
with >2500-fold selectivity being observed versus GR in the antag-
onist assay. Compound 22 was inactive in the GR agonist assay
(data not shown).
The PR partial agonist activity of compound 22 was also evalu-
ated in T47D breast cancer cells expressing native PR.9 In this as-
say, compound 22 is a potent partial agonist with an EC50 value
of 9 nM and a maximal agonist response equivalent to 65% of the
full progesterone response. In addition to measuring its partial
agonist activity at PR, the ability of compound 22 to inhibit the
estrogen-induced expression of SDF-1 (stromal cell-derived fac-
tor-1) in ELT-3 rat myometrial cells was evaluated.10 Compound
22 was a potent inhibitor in this assay (max inh. = 91%), demon-
strating an IC50 of 0.12 nM .
N
Cl
N R
NC
Compds
R
PR binding IC50, nM
PR CV-1 EC50, nM (%P4)
12
13
14
15
16
17
18
19
Me
i-Pr
20a
20a
13a
25
1.6 (67)
200 (47)
200 (52)
50 (40)
0.5 (64)
10 (45)
200 (47)
>10,000
(CH2)2CF3
(CH2)2OMe
CH2CN
CH2CO2Et
CH2CONH2
CH2-2-pyridyl
8a
8a
63
20a
a
At or below the tight binding limit of the PR binding assay.
result of placement of a polar group in a hydrophobic binding
environment.
The compounds for which the substituent on the pyrrolidine
ring was varied were similar in binding affinity as measured in
the PR FP binding assay, although for all but two of the compounds,
the measured IC50 was at or below the tight binding limit of this
assay. Therefore, for this compound set, the CV-1 functional assay
may be a better measure of the true potency of the compounds.
While a decrease in potency was generally observed with increas-
ing size or polarity of the substituent on the aniline nitrogen, per-
haps the more interesting trend was the decrease in partial agonist
efficacy when the size was increased beyond cyanomethyl (Table
2). This may also be explained by examining the structure.
The binding pocket shown in Figure 3 that is occupied by the
pyrrolidine N-methyl group in compound 22 does not appear to
be large enough to accommodate a group larger than cyanometh-
ylene. Although the ligand binding site of PR is known to have
some degree of flexibility,5 for larger substituents to be accommo-
dated the ligand would need to undergo global movement within
the binding site. Such shifts in the position of ligands in the PR
LBD have been observed crystallographically.5 A movement of this
type might be expected to cause a perturbation of the position of
Met-909 and induce a movement of helix 12 in the AF-2 region,6
leading to a partial agonist response as has been observed with
the partial agonist genistein bound to ERb.7 In fact, displacement
of Met-909 from its position in the agonist conformation is ob-
served in the structure of compound 22 bound in the PR LBD and
is illustrated in the superposition of the PR LBD-bound conforma-
Having demonstrated potent partial agonist activity, high selec-
tivity and potent inhibition of estrogen-stimulated SDF-1 expres-
sion in ELT-3 cells, compound 22 appeared to be an ideal
candidate for evaluation in an animal model. The OVX rat model,
a well-validated model for the measurement of estrogen opposi-
tion,11 was chosen for this study, using uterine wet weight as the
Table 3
Steroid receptor selectivity data for compound 22
FP binding assays
AR ERb
Functional assays
PR
PR CV-1
agonist
AR CV-1
MR CV-1
agonist
GR A549
antagonist
agonist
0.016a,c 0.250a >10a 0.003
(44%)b
1.3 (77%)b
5.0 (100%)b 7.9 (66%)
a
IC50 (binding assays) and EC50 (functional assays) values are given in
For the functional assays, percent activation or inhibition relative to the control
l
M.
b
agonist is given in parentheses.
c
At or below the tight binding limit of the assay.
0.7
0.6
0.5
0.4
0.3
0.2
**
***
***
***
***
0.1
0.0
E2
--
+
--
3
+
1
+
+
+
+
Compound22
(mg/kg)
-- --
3
10 30
P4
**, P<0.01 vs E only; *** P<0.001 vs E only
Figure 4. Superposition of the bound conformations of progesterone (carbon atoms
colored magenta) and compound 22 in the PR LBD.
Figure 5. Inhibition of estrogen-stimulated uterine wet weight gain in ovariecto-
mized rats by compound 22.

4780
S. K. Thompson et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4777–4780
11. Lundeen, S. G.; Zhang, Z.; Zhu, Y.; Carver, J. M.; Winneker, R. C. J. Steroid
Biochem. Mol. Biol. 2001, 78, 137.
phenotypic readout. As shown in Figure 5, oral administration
(u.i.d.) of compound 22 resulted in a dose-dependent reduction
in estrogen-stimulated uterine growth, with statistically-signifi-
cant reductions at 10 and 30 mg/kg/day and an effect at 30 mg/
kg/day comparable to that of progesterone (P4) administered at
10 mg/kg/day.
In summary, structure-based design has been successfully
implemented in the design of a new class of highly potent and
selective pyrrolidine-based progesterone receptor partial agonists,
one of which has demonstrated dose-dependent efficacy in an ani-
mal model of estrogen opposition. This work has provided a solid
foundation for further optimization of this compound class, the re-
sults of which will be reported elsewhere.
12. Example preparation: Compound 22. Step 1:
A mixture of 2-chloro-4-
fluorobenzonitrile (4.26 g, 27.5 mmol), 1,1-dimethylethyl (3S)-3-amino-1-
pyrrolidinecarboxylate (5.11 g, 27.5 mmol) and NaHCO3 (4.62 g, 55 mmol) in
45 mL of DMSO and 5 mL of H2O was heated with stirring at 96 °C for 6 h and
86 °C for 16 h. The reaction was diluted with 200 mL of H2O and extracted with
Et2O (3Â). The extracts were washed with H2O (2Â), dried over Na2SO4, filtered,
and concentrated. The residue was crystallized from Et2O/hexane to yield 1,1-
dimethylethyl (3S)-3-[(3-chloro-4-cyanophenyl)amino]-1-pyrrolidinecarboxy-
late (6.88 g, 78%). LC–MS (ES) m/e 322 [M+H]+. Step 2: NaH (60% dispersion in
mineral oil, 1.95 g, 48.7 mmol) was washed free of mineral oil with hexane,
suspended in 100 mL of DMF stirred, and cooled in an ice bath. A solution of 1,1-
dimethylethyl (3S)-3-[(3-chloro-4-cyanophenyl)amino]-1-pyrrolidinecarboxy-
late (10.43 g, 32.5 mmol) in 40 mL of DMF was added dropwise over 20 min. The
reaction was stirred an additional 40 min, and a solution of 1-(bromomethyl)-2-
(trifluoromethyl)benzene (11.65 g, 48.7 mmol) in 25 mL of DMF was rapidly
added. The reaction mixture was stirred for 30 min at 0 °C and 30 min at rt. The
mixture was poured into 200 mL cold aqueous NH4Cl, and extracted with Et2O
(3Â). The extracts were washed with H2O (2Â), dried over NaSO4, filtered,
and concentrated. The residue was purified by column chromatography (eluted
with 25% EtOAc/hexane) to yield 1,1-dimethylethyl (3S)-3-((3-chloro-4-
cyanophenyl){[2-(trifluoromethyl)phenyl]methyl}amino)-1-pyrrolidinecarb-
oxylate (14.28 g, 92%). LC–MS (ES) m/e 480 [M+H]+. Step 3: A solution of 1,1-
dimethylethyl (3S)-3-((3-chloro-4-cyanophenyl){[2-(trifluoromethyl)phenyl]*
methyl}amino)-1-pyrrolidinecarboxylate (14.0 g, 29 mmol) in 16 mL of CH2Cl2
was treated with TFA (15 mL) and stirred for 1.5 h. The reaction mixture was
concentrated, and the residue was triturated with Et2O and MeOH. After the
solids were collected by filtration, washed with Et2O, and dried, they were
dissolved in 20 mL of MeOH and added to a stirred mixture of aqueous K2CO3 and
Et2O. The Et2O was separated and the aqueous phase was extracted with Et2O
(2Â). The combined Et2O extracts were washed with H2O and saturated NaCl,
dried over Na2SO4, filtered, and concentrated to yield 2-chloro-4-((3S)-3-
pyrrolidinyl{[2-(trifluoromethyl)phenyl] methyl}amino)benzonitrile (10.74 g,
98%). LC–MS (ES) m/e 380 [M+H]+. Step 4: A solution of 2-chloro-4-((3S)-3-
pyrrolidinyl{[2-(trifluoromethyl)phenyl] methyl}amino)benzonitrile (9.0 g,
24 mmol) in 125 mL of MeOH was treated with formaldehyde (10.9 mL of a
37% aqueous solution, 142 mmol). After 30 min, the solution was cooled in an ice
bath. NaBH4 (2.0 g, 52 mmol) was slowly added and the reaction mixture was
stirred for 30 min. Cold H2O (200 mL) and aqueous NH4Cl (200 mL) were added
to decompose excess NaBH4. The mixture was extracted with Et2O and the
combined extracts were washed with H2O and concentrated. The residue
was purified on a short Al2O3 (neutral, Brockman 2.8) column eluting with
Et2O to yield 2-chloro-4-((1-methyl-3-pyrrolidinyl){[2-(trifluoromethyl)-
phenyl]methyl}amino)benzonitrile (8.31 g, 88%). LC–MS (ES) m/e 394 [M+H]+.
1H NMR (CDCl3)d: 7.75 (d, J = 3.6 Hz, 1H), 7.55–7.35 (m, 3H), 7.2 (d, J = 3.6 Hz,
1H), 6.75 (d, J = 2.8 Hz, 1H), 6.52 (dd, J = 8.8 Hz, 2.4 Hz, 1H), 5.07 (d, J = 18.8 Hz,
1H), 4.72 (d, J = 18.8 Hz, 1H), 4.64 (m, 1H), 2.90 (m, 1H), 2.7 (m, 2H), 2.6–2.4 (m,
5H), 1.87 (m, 1H).
Acknowledgments
The authors would like to thank our colleagues in the Screening
and Compound Profiling group for performing the steroid receptor
binding and CV-1 functional assays.
References and notes
1. Clarke, C. L.; Sutherland, R. L. Endocrinol. Rev. 1990, 11, 266.
2. Mihalyi, A.; Simsa, P.; Mutinda, K. C.; Meuleman, C.; Mwenda, J. M.; D’Hooghe,
T. M. Exp. Opin. Emerg. Drugs 2006, 11, 503.
3. Carr, B. R.; Macdonald, P. C. Int. J. Fertil. Women’s Med. 1997, 42, 133.
4. (a) Pathirana, C.; Stein, R. B.; Berger, T. S.; Fenical, W.; Ianiro, T.; Mais, D. E.;
Torres, A.; Goldman, M. E. Mol. Pharmacol. 1995, 47, 630; (b) Madauss, K. P.;
Grygielko, E. T.; Deng, S.-J.; Sulpizio, A. C.; Stanley, T. B.; Wu, C.; Short, S. A.;
Thompson, S. K.; Stewart, E. L.; Laping, N. J.; Williams, S. P.; Bray, J. D. Mol.
Endocrinol. 2007, 21, 1066.
5. (a) Madauss, K. P.; Deng, S.-J.; Austin, R. J. H.; Lambert, M. H.; McLay, I.;
Pritchard, J.; Short, S. A.; Stewart, E. L.; Uings, I. J.; Williams, S. P. J. Med. Chem.
2004, 47, 3381; (b) Compound 1’s PDB code is 3G8N.; (c) Compound 22’s PDB
code is 3HQ5.
6. Feng, W.; Riveiro, R. C. J.; Wagner, R. L.; Nguyen, H.; Apriletti, J. W.; Fletterick, R.
J.; Baxter, J. D.; Kushner, P. J.; West, B. L. Science 1998, 280, 1747.
7. Pike, A. C. W.; Brzozowski, A. M.; Hubbard, R. E.; Bonn, T.; Thorsell, A.-G.;
Engstrom, O.; Ljunggren, J.; Gustafsson, J.-A.; Carlquist, M. EMBO J. 1999, 18,
4608.
8. Williams, S. P.; Sigler, P. B. Nature 1998, 393, 392.
9. Di Lorenzo, D.; Albertini, A.; Zava, D. Cancer Res. 1991, 51, 4470–4475.
10. Hall, J. M.; Korach, K. S. Mol. Endocrinol. 2003, 17, 792.