Trifluoromethyl group as a pharmacophore: Effect of replacing a CF 3 group on binding and agonist activity of a glucocorticoid receptor ligand

Article abstract of DOI:10.1016/j.bmcl.2005.07.025

Compound 1, a potent glucocorticoid receptor ligand, contains a quaternary carbon bearing trifluoromethyl and hydroxyl groups. This paper describes the effect of replacing the trifluoromethyl group on binding and agonist activity of the GR ligand 1. The r

Full text of DOI:10.1016/j.bmcl.2005.07.025

Bioorganic & Medicinal Chemistry Letters 15 (2005) 4761–4769
Trifluoromethyl group as a pharmacophore: Effect of replacing
a CF3 group on binding and agonist activity
of a glucocorticoid receptor ligand
Raj Betageri,a,* Yan Zhang,d Renee M. Zindell,a Daniel Kuzmich,a
Thomas M. Kirrane,a Jo¨rg Bentzien,a Mario Cardozo,e Alison J. Capolino,a
Tazmeen N. Fadra,a Richard M. Nelson,a Zofia Paw,b Daw-Tsun Shih,c Cheng-Kon Shih,b
Ljiljana Zuvela-Jelaska,c Gerald Naboznyc and David S. Thomsona
aDepartment of Medicinal Chemistry, Boehringer Ingelheim Pharmaceuticals Inc., 900 Ridgebury Road, P.O. Box 368,
Ridgefield, CT 06877, USA
bDepartment of Cardiovascular Disease, Boehringer Ingelheim Pharmaceuticals Inc., 900 Ridgebury Road, P.O. Box 368,
Ridgefield, CT 06877, USA
cDepartment of Immunology and Inflammation, Boehringer Ingelheim Pharmaceuticals Inc., 900 Ridgebury Road,
P.O. Box 368, Ridgefield, CT 06877, USA
dDrug Discovery Division, Johnson & Johnson Pharmaceutical Research and Development, L.L.C., 1000 Route 202,
P.O. Box 300, Raritan, NJ 08869, USA
eDepartment of Structural Biology, Amgen San Francisco, 1120 Veterans Boulevard, South San Francisco, CA 94080, USA
Received 27 May 2005; revised 20 July 2005; accepted 21 July 2005
Available online 19 August 2005
Abstract—Compound 1, a potent glucocorticoid receptor ligand, contains a quaternary carbon bearing trifluoromethyl and hydrox-
yl groups. This paper describes the effect of replacing the trifluoromethyl group on binding and agonist activity of the GR ligand 1.
The results illustrate that replacing the CF3 group with a cyclohexylmethyl or benzyl group maintains the GR binding potency.
These substitutions alter the functional behavior of the GR ligands from agonists to antagonists. Docking studies suggest that
the benzyl analog 19 binds in a similar fashion as the GR antagonist, RU486. The central benzyl group of 19 and the C-11 dime-
thylaniline moiety of RU486 overlay. Binding of compound 19 is believed to force helix 12 to adopt an open conformation and this
leads to the antagonist properties of the non-CF3 ligands carrying a large group at the center of the molecule.
ꢀ 2005 Elsevier Ltd. All rights reserved.
The glucocorticoid receptor (GR) belongs to the family
of steroid receptors that include the mineralocorticoid
(MR), progesterone (PR), estrogen (ER), and androgen
(AR) receptors. The anti-inflammatory effects of endo-
genous steroids have stimulated the development of glu-
cocorticoid (GC) derivatives, such as dexamethasone
and prednisolone,1–4 which have found wide use in the
treatment of various inflammatory, immune, and aller-
gic disorders including rheumatoid arthritis, COPD,
CrohnÕs disease, systemic lupus erythematosus, and
osteoarthritis.5 However, use of GCs is associated with
a number of side effects that include edema, weight gain,
muscle weakness, diabetes mellitus, and osteoporosis.6,7
Recent investigations have led to an understanding of
the molecular mechanisms that mediate GC effects.
The anti-inflammatory and immune suppressive proper-
ties of GCs have largely been attributed to transrepres-
sion, whereas some of the side effects (such as
diabetes, glaucoma) have been ascribed to transactiva-
tion. In addition, cross-reactivity of GC ligands with
other steroid receptors may also lead to a number of side
effects.8,9 GR agonists showing dissociation between
transactivation and transrepression activities could
provide therapeutic agents with a reduced side-effect
profile.10–23 Furthermore, GCs stimulate the production
of glucose in the liver and this can exacerbate Type 2
Keywords: Glucocorticoid receptor; Glucocorticoid receptor ligand;
Trifluoromethyl group; Pharmacophore; Glucocorticoid receptor
antagonist; RU486; Helix 12.
*
Corresponding author. Tel.: +1 203 798 5428; fax: +1 203 791
6072; e-mail: rbetager@rdg.boehringer-ingelheim.com
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.07.025

4762
R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
diabetes.24 Studies with the prototype GR antagonist
RU486 have shown that GR antagonism could be an
effective way of regulating peripheral glucose metabo-
lism. A liver-selective GR antagonist could decrease
the hepatic glucose production without the risk of
peripherally driven side effects and may find therapeutic
utility in treating Type 2 diabetes.25 The discovery of
dissociated GC agonists, as well as liver-selective GR
antagonists, has been the focus of many drug discovery
programs in the recent years.25–33 Studies directed to-
ward understanding the structural basis for the agonist
and antagonist activities will aid in the design of GR li-
gands with the desired functional profile.
Several classes of ligands for the glucocorticoid receptor
have been described in the scientific and patent litera-
ture.26–33 For example, compound 131 and more recent-
ly, analog ZK 21634832 (Fig. 1) have been reported as
GR agonists demonstrating dissociation between trans-
repression and transactivation activities.
Our attention was drawn to the central quaternary cen-
ter containing trifluoromethyl and hydroxyl groups, in
the GR ligand 1. A similar structural feature, that is, a
quaternary center containing methyl and hydroxyl
groups is present in the androgen antagonists, bicaluta-
mide,34 and hydroxyflutamide35 (Fig. 2). The quaternary
center has shown to be an essential pharmacophore36 for
the biological activity of these ligands. Replacing the
central methyl group with a CF3 affects the functional
activity and the CF3 analogs behave as partial agonists.
We were intrigued by the role of the CF3 group in the
GR ligand 1 and asked the following questions: are
CF3 and hydroxyl groups required for the biological
activity? How does replacing CF3 and modifying other
Figure 2. GR and AR antagonists.
parts of the molecule, such as the backbone amide bond
and/or the anilide moiety, affect the biological activity?
We present here our results on investigating the effect
of these modifications on the binding potency, nuclear
receptor selectivity, and the agonist activity of com-
pound 1.
Structure–activity relationships (SAR) were driven by
the GR-binding affinities of the ligands. The IC50 values
for binding to GR, PR, MR, and ER were determined
using a fluorescence polarization competitive binding as-
say.37 Initially, we focused on exploring lipophilic CF3
substituents, such as alkyl and cycloalkyl groups, and
the SAR are shown in Table 1. Compound 1 has an
IC50 of 6 nM in the GR-binding assay. To quickly assess
the effect of replacing the CF3 group in 1, SAR was ini-
tiated with the preparation of readily accessible analogs
containing an unsubstituted phenyl ring at the right side
of the molecule. Toward this goal, we prepared the
phenyl analog 2 and the data showed that this com-
pound retains a good binding potency (IC50 79 nM).
Replacing CF3 by a hydrogen atom (3) leads to a signif-
icant loss in GR binding, whereas a CF3 to CH3 substi-
tution (4) leads to a 5-fold loss in the GR binding. These
results clearly indicate that in compound 3, the central
secondary-hydroxyl group alone is not sufficient to con-
fer good GR binding. Isopropyl analog 5 was similarly
potent compared to the methyl analog 4. Increasing
the size of the CF3 substituent from isopropyl (5) to
cyclohexylmethyl (6) leads to a dramatic improvement
in the GR binding potency. With an IC50 of 9 nM,
ligand 6 is 9-fold more potent than the corresponding
CF3 analog 2. These results clearly demonstrate that
the CF3 group is not essential for potent GR binding
and this prompted us to investigate further the CF3 sub-
stitution SAR. The optically active enantiomers (7 and
8) of 6, separated by chiral HPLC,38 showed a 35-fold
difference in the GR binding potencies, indicating a
stereochemical preference of the hydroxyl and cyclo-
Figure 1. GR agonists reported in the literature.

R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
4763
Table 1. Receptor binding data; alkyl, cycloalkyl, and arylalkyl CF3
replacements
analog 10 show that replacing the CF3 in 2 with a benzyl
group (10) (Table 1) results in a 6-fold improvement of
the GR binding. However, attaching a phenyl group
directly to the central carbon decreases the binding affin-
ity (9). A slight loss in GR binding is observed when the
central phenyl ring is attached by a two-carbon atom
spacer (11). Analogs 12 and 14 carrying 4-tert-butyl
phenyl and 2-naphthyl groups, respectively, bind to
GR with IC50s of 70–110 nM. These results suggest that
the effect of the bulkier substitutions is not dramatic
(6- to 8-fold loss in the GR binding compared to 10).
In contrast, the biphenyl analog 13 is significantly less
active than 10, pointing to a limitation in the size of
the hydrophobic group tolerated at this position.
R
OH
H
N
O
O
O
Compd
R
IC50 (nM) mean SD
PR
GR
1a
2
—
6
2
8
16
36
2
3
CF3
H
79
>2000
3
>2000
4
CH3
370 10
85 22
5
6
200 47
210 70
Encouraged by the initial results, we extended the SAR
to evaluate analogs of 10 in which various substituents
have been introduced on the central phenyl ring. Either
2-hydroxy-5-fluorophenyl, the preferred right side in
the CF3 series (Table 1), or an unsubstituted phenyl
group was employed for the right side of the molecule.
The SAR on the selected analogs having substituents at
the ortho or meta positions are given in Table 2. We have
shown earlier (Table 1) that analog 10, having an unsub-
stituted phenyl ring at both the center and the right side
of the molecule, is a potent GR ligand. In contrast to the
CF3 analogs 1 and 2, incorporating 2-hydroxy-5-fluor-
ophenyl at the right side of the molecule decreases the
GR binding affinity (19 and10). Small hydrophobic
groups, such as CH3, are tolerated at both ortho and
meta positions (20, 21). However, meta-disubstitution
leads to an 8-fold loss in GR binding (22). Introducing
a hydroxyl group at the ortho or meta positions decreases
the GR binding affinity (15, 16). Analogs bearing a polar
3,5-bis-hydroxybenzyl (17) or 2-pyridylmethyl side chain
(18) are significantly less potent than 10. These results
suggest that the central benzyl ring in 10 binds in a lipo-
philic environment. Data for the methyl substituted ana-
logs reflect the spatial constraints of the binding pocket.
9
3
19 12
7b
8b
9
320 60
210 70
9
2
35
6
460 40
240 40
10
13
1
30
6
11
12
57 18
59 14
72 15
330 70
The CF3 compound 24 containing a benzoxazinone ani-
lide moiety binds to GR with an affinity similar to the
corresponding phthalide 1 (Table 3). To expand the
scope of CF3 replacement SAR, we investigated the ef-
fect of incorporating the phenolic group, as well as the
benzoxazinone anilide moiety, on the affinities of our
most potent cyclohexylmethyl and benzyl analogs. The
CF3 analog 2 with an unsubstituted phenyl ring at the
right side is 15-fold less potent than the 2-hydroxy-5-
fluoro analog 1. Methylation of the phenolic group in
1 leads to a 40-fold loss in the GR binding (23). These
results suggest that the phenolic group may be involved
in hydrogen bonding interactions, which may account
for a higher binding affinity of the CF3 compound 1
compared to 2 and 23. In contrast, incorporating the
right side phenolic group does not improve the GR
binding in the non-CF3 series as illustrated by the data
for the cyclohexylmethyl analogs (6, 25), (26, 27) and
benzyl analogs (10, 19), (28, 29). In fact, results with
the phthalides (6, 25) and (10, 19) indicate a trend to-
wards a slight decrease in GR affinity when a phenolic
group is introduced at the right side of the molecule.
A slight improvement in GR binding is observed when
13
740 310
180 40
14
110 20
48 10
a See Figure 1.
b Optically active enantiomers of 6.
hexylmethyl groups for optimal GR binding. A similar
(50-fold) difference in the GR binding affinities has been
reported32 for the optically active enantiomers of the
CF3 compound ZK 216348.
Next, we focused our attention on replacing the CF3
group in 2 with aryl or arylalkyl groups. Data for the

4764
R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
Table 3. Receptor binding data; R1, R2, and R3 modifications
Table 2. Receptor binding data; substituted benzyl analogs
R1 OH
R2 OH
H
H
N
R2
N
R1
R3
O
O
O
O
R1
Compd.
R1
R2
R3
IC50 (nM)
mean SD
Compd.
R2
IC50 (nM) mean SD
GR
PR
GR
79
PR
10
15
13
1
30
6
6
O
O
2
1
CF3
CF3
8
2
36
16
3
2
OH
61 11
47
O
O
OH
OH
6
F
OMe
16
17
18
19
120 10
>2000
44
0
O
O
23
24
6
CF3
CF3
230 50
85 48
22 15
19 12
OH
HO
F
OH
160
4
9
1
3
8
6
1
1
1
3
2
F
N
760 230
400 30
35 11
O
O
OH
60
1
OH
F
F
O
O
25
26
27
10
19
28
29
20
9
42
8
OH
20
48 33
36 16
CH3
CH3
F
40 22
OH
21
22
52
5
34
9
OH
F
CH3
F
OH
N
O
7
35
30
9
6
CH3
H3C
470 230
51 17
O
F
13
60
7
the left side phthalide is replaced with benzoxazinone
(analogs 27, 25 and 29, 19). This trend is seen only with
the non-CF3 analogs having a phenolic group at the
right side. The SAR shown in Table 3 suggests that
the right side phenolic group in the non-CF3 analogs
is not involved in hydrogen bonding interactions, unlike
CF3 compounds 1 and 24, supported by the docking
studies of the compounds 19, 1,33 and 2433,39 (vide mod-
eling section at the end).
OH
F
35 11
30
5
OH
F
To assess the influence of the central hydroxyl group on
GR binding potency, the hydroxyl group in the benzyl
analog 10 was replaced with a hydrogen atom. This sub-
stitution leads to a poor GR ligand (30) (Table 4). We
have shown earlier that the presence of a central hydrox-
yl group alone will not lead to potent GR binding (3,
Table 4). Taken together, binding data for 3, 10, and
30 clearly indicate that the central quaternary center in
10 (and the other non-CF3 analogs) is essential for
potent GR binding. Both the hydroxyl group and the
8
32 13
CF3 substituent (benzyl or cycloalkyl group) are in-
volved in important interactions with GR and function
as essential pharmacophores. Comparison of the bind-
ing data for the analogs 31 and 32 suggests that the
backbone carbonyl is not involved in critical interac-
tions with the GR.

R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
4765
Table 4. Receptor binding data; backbone modifications and CF3
replacements
Table 5. Nuclear receptor selectivity; GR, PR, MR, and ER binding
data for selected compounds
R1 R2
H
Compd.
IC50 (nM) mean SD
N
X
R3
GR
PR
MR
30
ER
O
1
2
6
79
4
2
8
1
16
36
2
3
6
n.b.a
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
n.b.
O
215 35
41
24
4
22 15
85 22
210 70
19 12
7
Compd R1
R2
R3
X
IC50 (nM)
mean SD
370 10
200 47
60 16
315 50
170 30
160 14
175 55
190 42
140 26
104 14
5
GR
PR
6
9
13
3
1
10
12
19
25
27
29
31
30
6
3
10
30
31
32
H
OH
OH
H
CO
CO
CO
CH2
CO
>2000
13
2000
30
72 15
330 70
35 11
60
20
7
1
8
1
2
6
42
35
8
9
8
81
32 13
80 20
104
195 29
9
1
6
a n.b.: no detectable binding at 2 lM.
>2000
81
2000
respectively. A dissociated GR ligand is expected to
block the dexamethasone-induced transactivation in
the MMTV assay, while maintaining the transrepression
activity. CF3 compounds 1 and 24 do exhibit such a pro-
file in these functional assays. Both compounds 1 and
24 show potent agonist activities with EC50s of 45 nM
(efficacy 59% of dexamethasone) and 10 nM (efficacy
87% of dexamethasone), respectively (Table 6). In the
antagonist assay, while compound 1 has an EC50 of
150 nM and was able to completely antagonize the
MMTV transactivation, 24 gave a maximum inhibition
of 60% at 5 lM. Interestingly, replacing the 2-hy-
droxy-5-fluorophenyl group in the agonist 1 with an
unsubstituted phenyl ring leads to the full antagonist 2
with an EC50 of 150 nM. Thus, compound 2 showed
no IL-6 agonism at 2 lM, while completely inhibiting
the transactivation. Non-CF3 compounds behaved sim-
ilarly, showing no agonist activity at a similar concen-
tration (Table 6). Potent antagonist activities with
EC50s of 120–760 nM and 100% inhibition at 5 lM were
observed for most of the non-CF3 compounds tested
OMe
OH
OH
6
3
80 20
F
OMe
52
35
1
F
The following observations were made on the GR/PR
selectivity. Binding data show that the CF3 analogs 1
and 24 have similar affinities toward GR and PR (Tables
1 and 3). Replacing CF3 with a cyclohexylmethyl or ben-
zyl group (6, 10) leads to a slight improvement in the
GR/PR selectivity (Table 1). Introducing a bulky or po-
lar group on the central phenyl ring decreases the GR
binding affinity. However, the PR binding is not signif-
icantly altered (Tables 1 and 2). Both the right side phe-
nolic group and the left side anilide moieties do not
influence the GR/PR selectivity (Table 3). Overall,
replacing the CF3 group and modifying other parts of
the molecule do not improve the binding selectivity over
PR. Similarly, selectivity over MR and ER is not affect-
ed by the CF3 replacements and other modifications de-
scribed in the present work. Nuclear receptor selectivity
data for selected compounds are shown in Table 5.
Table 6. IL-6 agonism and MMTV antagonsim data for selected
compounds
Compd
IL-6
EC50 (nM) % efficacy EC50 (nM) % inhibition
mean SD at 2 lM mean SD at 5 lM
MMTV
Dex.
RU486
1
1
100
44
—
2
1
100% at 100 nM
Non-CF3 compounds with the GR binding affinities less
than 2 lM were tested for their agonist and antagonist
activities. The agonist activities were determined by
measuring the transcriptional repression of IL-6 produc-
tion in IL-1-stimulated human foreskin fibroblasts.40
Mifepristone (RU486) is shown to compete with
dexamethasone and inhibit the dexamethasone-induced
GR transactivation. This assay33,41,42 was employed to
evaluate antagonist activities of GR ligands by measur-
ing the inhibition of the dexamethasone-induced GR
transactivation of the mouse mammary tumor virus
(MMTV) luciferase gene in HeLa cells.43 Dexametha-
sone and RU486 were used as standards for GR agon-
ism and antagonism, and have EC50s of 1 and 2 nM,
45
—
10
—
—
—
—
—
—
—
—
—
—
9
3
59
150 35
150 90
100
100
60
2
n.d.a
87
24
4
—
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
2370 580
620 200
140 65
70
95
5
6
100
10
12
19
25
29
27
31
205 110 100
760 270 90
420 170 100
130 90
130 90
120 40
100
100
100
610 160 100
a n.d.: no detectable inhibition at 2 lM.

4766
R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
(Table 6). Lack of agonist activity of the methyl analog
4 could be due to the poor GR binding affinity. In con-
trast, the non-CF3 compounds with a large group at the
center of the molecule are potent GR ligands and the
lack of agonist activity of these compounds cannot be
attributed to poor GR binding affinities. For example,
analogs 6, 25, 27, and 29 bind to GR with affinities com-
parable to that of dexamethasone (IC50 6 nM) and yet
have no agonist activities. Similar to RU486, these
compounds 6, 25, 27, and 29 behave as potent GR
antagonists by completely inhibiting the dexametha-
sone-induced MMTV transactivation with EC50s in the
range of 120–140 nM. The benzyl analogs 10 and 19 also
showed 100% inhibition at 5 lM and had EC50s of 205
and 420 nM, respectively. These results clearly demon-
strate that replacing the CF3 group in 1 and 24 with a
large lipophilic moiety affects the functional profile
and imparts antagonist properties to these ligands.
To gain insight into the structural reasons for the ob-
served agonistic and antagonistic behavior, we have
docked the CF3 compound 1 and the benzyl analog 19
into the GR-ligand binding domain (GR-LBD) binding
pocket.39,45–49 Compound 1 can be docked into the
binding pocket observed in the co-complex X-ray struc-
ture of GR-LBD/fluticasone.33 The docking study shows
that compound 1 occupies a similar space as the known
GR agonist, fluticasone propionate. The binding mode
is highlighted by the position of helix 12 (H12) that sits
on the binding pocket (Fig. 3A). This position of helix
12 has shown to be a salient feature of the agonist con-
formation. Benzyl analog 19 cannot bind in the same
fashion, as the large benzyl moiety requires the reloca-
tion of helix 12. Compound 19 was docked into the
binding pocket using the co-complex X-ray structure
of GR-LBD/RU486.46 The docking study shows that
19 and RU486 occupy similar space and the benzyl moi-
ety of 19 overlays with the C-11 dimethylaniline group
of RU486 (Fig. 3B). The benzyl group, like the C-11
dimethylaniline group of RU486, forces helix 12 to
adopt an open conformation. In the case of RU486,
the helix 12 is displaced into a coactivator recognition
surface known as activation function-2 (AF-2) domain.
The displaced helix 12 blocks the recruitment of coacti-
vators and provides a molecular basis for RU486 antag-
onsim of GR.41 Our docking studies indicate that
binding of compound 19 to GR induces similar confor-
mational changes, which may account for the antagonist
activity of the benzyl analog 19 and other non-CF3 ana-
logs that contain a large group at the quaternary center.
Our results on the non-CF3 analogs of compound 1 and
24 indicate that: (i) the CF3 group can be replaced with a
cyclohexylmethyl or benzyl group and potent GR bind-
ing can be retained; (ii) the central hydroxyl group is
essential; (iii) the central quaternary center is required
for binding to GR; (iv) the backbone carbonyl can be re-
placed with a methylene group and good GR affinity can
be maintained; (v) the right side phenolic group does not
improve the GR binding; (vi) the selectivity over PR,
MR, and ER is not affected dramatically by the CF3
substitutions. The CF3 group, by increasing the acidity
of the central hydroxyl group, could enhance its hydro-
gen bond donor ability. Replacing the CF3 with a cyclo-
alkyl or arylalkyl group weakens the ability of the
central hydroxyl moiety to be involved in hydrogen
bonding interactions.35,36,44 Consequent loss in GR
binding could be compensated by a gain in the binding
energy due to the hydrophobic interactions of the CF3
substituents (benzyl or cyclohexylmethyl groups) with
the receptor. In addition, replacing the CF3 group with
a cyclohexylmethyl or benzyl moiety changes the func-
tional behavior of these compounds from agonists to
antagonists.
A more detailed analysis of the binding interactions of
the agonist 1 and the antagonist 19 shows similarities,
as well as differences (Figs. 4A and B). The left side het-
erocyclic group of the CF3 compound 1 overlays with
the A-ring of fluticasone and is likely to interact with
Gln570 and Arg611. The central hydroxyl of 1 overlays
with the C-11 hydroxyl of the steroid and similarly
forms a hydrogen bond with the amide side chain of
Asn564. The ligand amide nitrogen is suggested to form
an additional hydrogen bond with the backbone
Figure 3. (A) Docking of compound 1 into the GR-LBD using the GR-LBD/fluticasone co-complex X-ray structure. (B) Docking of compound 19
into the GR-LBD using the GR-LBD/RU486 co-complex X-ray structure.

R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
4767
carbonyl of Leu563. No hydrogen bond interaction is
observed between the ligand amide carbonyl group
and the GR-LBD. The right side phenol is engaged in
potential hydrogen bonding interactions with the side
chain of Thr739. This is in line with the SAR reported
in Table 3 where substitution of the 2-hydroxy-5-fluor-
ophenyl right side by phenyl (2) or 2-methoxy-5-fluor-
ophenyl (23) leads to a loss in GR binding. We believe
that compounds 1, 2, and 24 bind in a similar fashion
with helix 12 capping the binding pocket. Absence of
the key interaction between Thr739 and compound 2
stabilizes an inactive GR conformation, which accounts
for the lack of the agonist activity of 2. But the non-CF3
compounds discussed in this paper carry a bulky group
at the center of the molecule, which pushes the helix 12
out of the agonist position, leading to a classical RU486-
like mode of GR antagonism.
the paper. The central phenyl ring of the antagonist 19
is surrounded by the side chains of the hydrophobic
amino acids Val571, Trp600, Met604, and Phe737.
The spatial constraints of this pocket could explain the
effect of the methyl substitutions on the central phenyl
ring (Table 2). The model suggests that groups extend-
ing out from the para position of the central phenyl ring
are pointing toward the protein surface. The poor bind-
ing affinity of the biphenyl analog 13 could be due to ste-
ric interactions of the para phenyl group with the nearby
amino acid side chains. Perhaps, introducing a spacer
group between the two phenyl rings will extend the para
phenyl group and may improve the GR binding of 13. In
this context, it is interesting to note that large groups
have been introduced in RU486 using the amine group
on the C-11 phenyl ring as the point of attachment with-
out significantly affecting the GR binding.25 In conclu-
sion, these modeling studies suggest that replacing the
CF3 group in 1 and 24 by large substituents changes
the position of helix 12, leading to the antagonistic activ-
ity of these ligands. The observed binding interactions
between 19 and GR-LBD also offer explanations for
the SAR of other regions in these non-CF3 compounds.
The antagonist binding mode of 19 shows potential
interactions between the left side heterocyclic ring and
the side chains of Gln570 and Arg611, similar to the
agonist 1. The central hydroxyl of the benzyl analog
19 could be involved in forming a hydrogen bond with
˚
the amide side chain of Asn564, although at 4 A this dis-
tance is large for a normal hydrogen bond. The amide
group of the antagonist 19 shows similar interactions
with the GR-LBD as the agonist 1. The SAR shown
in Table 4 is supported by the presence of a potential
hydrogen bond between the amide nitrogen and the
backbone carbonyl of Leu563, and a lack of hydrogen
bonding interaction between the ligand amide carbonyl
and the GR-LBD. Different interactions were observed
between the GR-LBD and the right sides of the agonist
1 and the antagonist 19. Repositioning of helix 12 in the
antagonist binding mode alters the location of Thr739,
which is no longer able to form a hydrogen bond with
the ligand. This could explain why no increase in the
GR binding affinity and no agonist activity were ob-
served for 19 when a phenolic group was introduced
on the right side phenyl ring in 10 (Table 3). This is in
contrast to analogs 1 and 2 where the presence of the
phenolic group at the right side phenyl ring has an effect
on the functional profile (Table 6). These results further
lend support to our suggestion of different binding
modes for the CF3 and non-CF3 analogs presented in
The syntheses of the compounds described in the pres-
ent work have been published elsewhere.50 Analogs with
an unsubstituted phenyl ring at the right side of the
molecule were synthesized using commercially available
2-phenyl-2,2-dimethylethylmagnesium bromide. The
reaction of this Grignard reagent with diethyl oxalate
and hydrolysis of the resulting ester gave the desired
keto-acid. The keto-acid (35) was synthesized, accord-
ing to a modified literature procedure31 (Scheme 1).
Organolithium reagent generated from 2-bromo-4-flu-
oroanisole (33) was quenched with acetone to give the
tert-alcohol (34). Tin chloride-mediated coupling of this
alcohol (34) with the trimethylsilyl enol ether of ethyl
pyruvate and hydrolysis of the resulting ester led to
the keto-acid (35). The keto-acids were converted into
the corresponding acid chlorides and reaction with the
appropriate anilines gave the keto-amides. Treatment
of the keto-amides with a corresponding Grignard (or
organolithium) reagent, followed by demethylation of
the anisole intermediates, led to the desired para-fluoro-
phenols (36).
Figure 4. (A) Binding mode of compound 1. (B) Binding mode of compound 19. Solid lines indicate potential hydrogen bonds.

4768
R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
Scheme 1. Reagents and conditions: (a) n-BuLi, THF, À78 ꢁC, acetone, 85%; (b) trimethylsilyl trifluoromethanesulfonate, di isopropylethylamine,
CH2Cl2, 0 ꢁC, 90%; (c) SnCl4, CH2Cl2, À78 to À50 ꢁC, 35%; (d) KOH, aq. ethanol, reflux, 93%; (e) thionyl chloride, dimethylacetamide, 0 ꢁC,
ArNH2, 0 ꢁC to rt, 60%; (f) Grignard reagent, 0 ꢁC to rt, 60%, (g) BBr3, CH2Cl2, 0 ꢁC to rt, 75%.
10. Adcock, I. M. Pulm. Pharmacol. Ther. 2000, 13, 115.
11. Eggert, M.; Schulz, M.; Neeck, G. J. Steroid Biochem.
Mol. Biol. 2001, 77, 185.
12. Elmore, S. W.; Pratt, J. K.; Coghlan, M. J.; Mao, Y.; Green,
B. E.; Anderson, D. D.; Stashko, M. A.; Lin, C. W.; Falls,
D.; Nakane, M.; Miller, L.; Tyree, C. M.; Miner, J. N.;
Lane, B. Bioorg. Med. Chem. Lett. 2004, 14, 1721.
13. Hermoso, M. A.; Cidlowski, J. A. IUBMB Life 2003, 55,
497.
14. Heck, S.; Kullmann, M.; Gast, A.; Ponta, H.; Rahmsdorf,
H. J.; Herrlich, P.; Cato, A. C. EMBO J. 1994, 13, 4087.
15. Jenkins, B. D.; Pullen, C. B.; Darimont, B. D. Trends
Endocrinol. Metab. 2001, 12, 122.
16. Reichardt, H. M.; Kaestner, K. H.; Tuckermann, J.;
Kretz, O.; Wessely, O.; Bock, R.; Gass, P.; Schmid, W.;
Herrlich, P.; Angel, P.; Schutz, G. Cell 1998, 93, 531.
17. Schaaf, M. J.; Cidlowski, J. A. J. Steroid Biochem. Mol.
Biol. 2002, 83, 37.
In conclusion, we have discussed our efforts in probing
the pharmacophoric elements required for binding to
GR with a focus on the role of CF3 group in the ligands
1 and 24. Although, further work is needed to fully
understand the role of the CF3 group and its effect on
the agonist activity we believe our work sheds light on
the binding modes of agonists and antagonists based
on the compounds 1 and 24. Docking studies suggest a
binding mode similar to that of RU486, wherein the
conformational changes involving the repositioning of
helix 12 explain the observed GR antagonism. We have
shown that by introducing a large group in a non-steroi-
dal GR ligand, functional activity can be altered from
agonism to antagonism. These observations may be
applicable to other potential GR scaffolds wherein
appropriate modifications of the size and shape, while
maintaining the key interactions, could be an effective
way of modulating the functional activity.
18. Schaaf, M. J.; Cidlowski, J. A. Mol. Cell. Biol. 2003, 23,
1922.
19. Smoak, K. A.; Cidlowski, J. A. Mech. Aging Dev. 2004,
125, 697.
Acknowledgment
20. Tronche, F.; Kellendonk, C.; Reichardt, H. M.; Schutz, G.
Curr. Opin. Genet. Dev. 1998, 8, 532.
21. Webster, J. C.; Cidlowski, J. A. Trends Endocrinol. Metab.
1999, 10, 396.
22. Cato, A. C.; Schacke, H.; Sterry, W.; Asadullah, K. Curr.
Drug Targets Inflamm. Allergy 2004, 3, 347.
We thank John Regan for critically reading the
manuscript.
23. Schacke, H.; Hennekes, H.; Schottelius, A.; Jaroch, S.;
Lehmann, M.; Schmees, N.; Rehwinkel, H.; Asadullah, K.
Ernst Schering Res. Found. Workshop 2002, 357.
24. Friedman, J. E.; Sun, Y.; Ishizuka, T.; Farrell, C. J.;
McCormack, S. E.; Herron, L. M.; Hakimi, P.; Lechner,
P.; Yun, J. S. J. Biol. Chem. 1997, 272, 31475.
25. von Geldern, T. W.; Tu, N.; Kym, P. R.; Link, J. T.; Jae,
H. S.; Lai, C.; Apelqvist, T.; Rhonnstad, P.; Hagberg, L.;
Koehler, K.; Grynfarb, M.; Goos-Nilsson, A.; Sandberg,
J.; Osterlund, M.; Barkhem, T.; Hoglund, M.; Wang, J.;
Fung, S.; Wilcox, D.; Nguyen, P.; Jakob, C.; Hutchins, C.;
Farnegardh, M.; Kauppi, B.; Ohman, L.; Jacobson, P. B.
J. Med. Chem. 2004, 47, 4213.
References and notes
1. Ali, S. L. Anal. Profiles Drug subst. Excipients 1992, 21, 415.
2. Cohen, E. M. Anal. Profiles Drug subst. Excipients 1973, 2,
163.
3. Adcock, I. M. Curr. Allergy Asthma Rep. 2003, 3, 249.
4. Parente, L. In Glucocorticoids; Goulding, N. J., Flowers,
R. J., Eds.; Birkhauser: Boston, 2001, pp 35–54.
5. Toogood, J. In Glucocorticoids; Goulding, N. J., Flowers,
R. J., Eds.; Birkhauser: Boston, 2001, pp 161–174.
6. Gupta, R.; Jindal, D. P.; Kumar, G. Bioorg. Med. Chem.
2004, 12, 6331.
7. Schacke, H.; Docke, W. D.; Asadullah, K. Pharmacol.
Ther. 2002, 96, 23.
26. Ali, A.; Thompson, C. F.; Balkovec, J. M.; Graham, D.
W.; Hammond, M. L.; Quraishi, N.; Tata, J. R.; Einstein,
M.; Ge, L.; Harris, G.; Kelly, T. M.; Mazur, P.; Pandit, S.;
Santoro, J.; Sitlani, A.; Wang, C.; Williamson, J.; Miller,
D. K.; Thompson, C. M.; Zaller, D. M.; Forrest, M. J.;
Carballo-Jane, E.; Luell, S. J. Med. Chem. 2004, 47, 2441.
27. Coghlan, M. J.; Kort, M. E. Exp. Opin. Ther. Patents
1999, 9, 1523.
8. Zhi, L.; Tegley, C. M.; Edwards, J. P.; West, S. J.;
Marschke, K. B.; Gottardis, M. M.; Mais, D. E.; Jones, T.
K. Bioorg. Med. Chem. Lett. 1998, 8, 3365.
9. Zhi, L.; Tegley, C. M.; Kallel, E. A.; Marschke, K. B.;
Mais, D. E.; Gottardis, M. M.; Jones, T. K. J. Med. Chem.
1998, 41, 291.

R. Betageri et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4761–4769
4769
28. Coghlan, M. J.; Kym, P. R.; Elmore, S. W.; Wang, A. X.;
Luly, J. R.; Wilcox, D.; Stashko, M.; Lin, C. W.; Miner,
J.; Tyree, C.; Nakane, M.; Jacobson, P.; Lane, B. C.
J. Med. Chem. 2001, 44, 2879.
38. The chiral HPLC conditions used to separate the enan-
tiomers 7 and 8: 250 · 21.1 mm, Chiral Technologies,
CHIRALPACK OD column, isocratic 10% isopropanol
in hexanes, 8 ml/min.
29. Lin, X.; Hueber, V. Curr. Opin. Drug Discov. Devel. 2000,
3, 383.
39. Bledsoe, R. K.; Montana, V. G.; Stewart, E. L.; Xu, H. E.
EP1375517 A1, 2004.
30. Shah, N.; Scanlan, T. S. Bioorg. Med. Chem. Lett. 2004,
14, 5199.
31. Lehmann, M.; Krolikiewicz, K.; Skuballa, W.; Strehlke,
P.; Kalkbrenner, F.; Ekerdt, R.; Giesen, C. WO 0032584,
2000.
32. Schacke, H.; Schottelius, A.; Docke, W. D.; Strehlke, P.;
Jaroch, S.; Schmees, N.; Rehwinkel, H.; Hennekes, H.;
Asadullah, K. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,
227.
33. Barker, M.; Clackers, M.; Demaine, D. A.; Humphreys,
D.; Johnston, M. J.; Jones, H. T.; Pacquet, F.; Pritchard,
J. M.; Salter, M.; Shanahan, S. E.; Skone, P. A.; Vinader,
V. M.; Uings, I.; McLay, I. M.; Macdonald, S. J. J. Med.
Chem. 2005, 48, 4507.
34. Mukherjee, A.; Kirkovsky, L.; Yao, X. T.; Yates, R. C.;
Miller, D. D.; Dalton, J. T. Xenobiotica 1996, 26, 117.
35. Tucker, H.; Crook, J. W.; Chesterson, G. J. J. Med. Chem.
1998, 31, 954.
36. Katchen, B.; Buxbaum, S. J. Clin. Endocrinol. Metab.
1975, 41, 373.
40. Human foreskin fibroblasts are stimulated with 1 ng/mL
of recombinant human IL-1 in the presence of test
compound. After 24 h, the degree of GR agonist activity
(transrepression) is determined by measuring IL-6 in the
tissue culture media and calculating the EC50 values. Top
concentration in the assay was 2 lM.
41. Honer, C.; Nam, K.; Fink, C.; Marshall, P.; Ksander, G.;
Chatelain, R. E.; Cornell, W.; Steele, R.; Schweitzer, R.;
Schumacher, C. Mol. Pharmacol. 2003, 63, 1012.
42. It should be pointed out that RU486 shows weak agonist
activity with a 44% efficacy (at 1 lM) of IL-6 production
in our IL-6 assay.
43. HeLa cells stably transfected with MMTV Luciferase
construct were preincubated with 20 nM dexamethasone
for 15 min. After preincubation, cells were treated with the
compounds or their vehicle (0.2% DMSO) and incubated
for 6 h. After the incubation, the induction of luciferase
was measured. Top concentration in the assay was 5 lM.
44. Jeffrey, J. M.; Hughes, L. R.; Glen, A. T.; Taylor, P.
J. Med. Chem. 1991, 34, 447.
37. Fluorescence polarization competitive binding assays were
performed to quantitate the ability of test compounds to
displace ligands from GR, ER, MR, and PR in solution.
Binding reactions were assembled in 96-well microplates
and consisted of Baculovirus lysate containing either GR,
MR or PR, 5 nM tetramethyl-rhodamine conjugates of
dexamethasone or RU486, respectively, and test com-
pound dilutions in an assay buffer containing 10 mM TES,
50 mM KCl, 20 mM sodium molybdate, 1.5 mM EDTA,
0.04% w/v CHAPS, 10% v/v glycerol, and 1 mM DTT, pH
7.4. The ER fluorescence polarization competitive binding
assays were performed according to the protocol provided
with the Panvera kit P-2614. IC50 values shown are the
mean and standard deviation from at least two separate
experiments (IC50 value for compound 17 is from a single
experiment); each consisting of duplicate 11-point con-
centration–effect curves and top concentration in the assay
was 2 lM.
45. Bledsoe, R. K.; Montana, V. G.; Stanley, T. B.; Delves, C.
J.; Apolito, C. J.; McKee, D. D.; Consler, T. G.; Parks, D.
J.; Stewart, E. L.; Willson, T. M.; Lambert, M. H.; Moore,
J. T.; Pearce, K. H.; Xu, H. E. Cell 2002, 110, 93.
46. Kauppi, B.; Jakob, C.; Farnegardh, M.; Yang, J.; Ahola,
H.; Alarcon, M.; Calles, K.; Engstrom, O.; Harlan, J.;
Muchmore, S.; Ramqvist, A. K.; Thorell, S.; Ohman, L.;
Greer, J.; Gustafsson, J. A.; Carlstedt-Duke, J.; Carlquist,
M. J. Biol. Chem. 2003, 278, 22748.
47. (R)-enantiomers of the compounds 1 and 19 were docked
intotheGR-LBDpocketusingtheGR-LBD/fluticasoneand
the GR-LBD/RU486 co-complex X-ray structures, respec-
tively. The docking studies were done using Glide49 and
Figures 3A, B and 4A, B were prepared using Maestro48.
48. Maestro Version 6.5.007, Schro¨dinger, New York, 2004.
49. Glide 2.5, Schro¨dinger, New York, 2003.
50. Betageri, R.; Thomson, D.; Zhang, Y.; Zindell, R. M.
WO 03059899, 2003.