Discovery of 3-aryloxy-lactam analogs as potent androgen receptor full antagonists for treating castration resistant prostate cancer

Article abstract of DOI:10.1016/j.bmcl.2011.11.068

High throughput cell-based screening led to the identification of 3-aryloxy lactams as potent androgen receptor (AR) antagonists. Refinement of these leads to improve the ADME profile and remove residual agonism led to the discovery of 12, a potent full antagonist with greater oral bioavailability. Improvements in the ADME profile were realized by designing more ligand-efficient molecules with reduced molecular weights and lower lipophilicities.

Full text of DOI:10.1016/j.bmcl.2011.11.068

Bioorganic & Medicinal Chemistry Letters 22 (2012) 1230–1236
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of 3-aryloxy-lactam analogs as potent androgen receptor full
antagonists for treating castration resistant prostate cancer
⇑
⇑
Chuangxing Guo , Susan Kephart , Martha Ornelas, Javier Gonzalez, Angelica Linton, Mason Pairish,
Asako Nagata, Samantha Greasley, Jeff Elleraas, Natilie Hosea, Jon Engebretsen, Andrea N. Fanjul
Oncology Medicinal Chemistry, Pfizer Worldwide Research & Development, San Diego, CA 92121, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
High throughput cell-based screening led to the identification of 3-aryloxy lactams as potent androgen
receptor (AR) antagonists. Refinement of these leads to improve the ADME profile and remove residual
agonism led to the discovery of 12, a potent full antagonist with greater oral bioavailability. Improve-
ments in the ADME profile were realized by designing more ligand-efficient molecules with reduced
molecular weights and lower lipophilicities.
Received 13 October 2011
Revised 14 November 2011
Accepted 18 November 2011
Available online 9 December 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Androgen receptor
AR antagonist
Prostate cancer
Ligand efficiency
Polar surface area
PSA
Castration resistant prostate cancer
CRPC
LBD
NHR
Several androgen receptor (AR) antagonists have been investi-
gated in clinical trials against prostate cancer.1 One example is
bicalutamide (CasodexÒ, Fig. 1), which is used along with chemical
or surgical castration in the combined androgen blockade (CAB)
protocol for prostate cancer treatment.2 Unfortunately, bicaluta-
mide (alone or in combination therapy) loses efficacy after only a
few months to a few years of treatment. Oncogenic mutations or
changes in gene expression make the cancer cells resistant to hor-
mone blockade, resulting in castration-resistant prostate cancer
(CRPC), also known as hormone refractory prostate cancer (HRPC).
Elevated AR expression in tumor cells correlates with a change in
the functional activity of bicalutamide from antagonist to agonist,3
suggesting that prostate tumors may adapt to survive on any resid-
ual agonism a drug molecule may induce. This antagonist-to-ago-
nist switch has also been observed for other drugs,3 and may be
responsible for drug resistance in CRPC patients. Recently, novel
AR antagonists such as MDV-3100 and BMS-641988 (Fig. 1) have
demonstrated efficacy against CPRC in preclinical models,4,5 and
more importantly, in patients.6 These results encouraged our
own search for novel, non-steroidal, pure antagonists of AR in CRPC
cells.
Pfizer has a collection of AR antagonists synthesized for an ear-
lier program targeting topical treatments for hair-loss and excess
sebum production.7 These compounds were screened for AR antag-
onism in a high-throughput assay based on CRPC cells,8 and 1
(Fig. 2) was identified as a potent antagonist (AR antagonism
IC50 = 77 nM). Initial follow-up chemistry efforts on close-in ana-
logs revealed that, after chiral resolution, the more potent diaste-
reomers (e.g., 2) are partial AR antagonists with significant
residual agonism,9 but removal of the methyl group (e.g., 3) elim-
inated AR agonism. These leads were profiled for their ADME prop-
erties and were found to have good permeability (data not shown),
but suffered from generally high unbound clearance in vitro (HLM
unbound Clint10 19.3–115
l
L/min/mg) and in vivo (unbound Cl:11
g/mL
921–1585 mL/min/kg in rat) as well as low solubility (<10
l
at pH 7.4). Both of these factors are likely to be underlying causes
of low oral absorption (Fa <25%) observed in rat PK studies of these
leads.
Since an orally active agent was our goal, we sought to improve
solubility and reduce unbound clearance. Both problems may be
due to the high lipophilicity (cLogP) of these molecules. Our strat-
egy was to reduce lipophilicity by modifying either the substituent
on the lactam nitrogen (right-hand side) or the aryl ether (left-
hand side), in such a way that would maintain the full antagonist
profile. Analysis of the existing SAR on this series revealed that
⇑
Corresponding authors.
E-mail address: alex.guo@pfizer.com (C. Guo).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.11.068

C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
1231
CF3
CF3
CF3
NC
NC
NC
F
O
S
O
O
S
O
H
O
N
H
N
N
N
O
HO
HN
NH
S
O
O
F
H
O
O
MW:
cLogP:
TPSA:
430
464
3.35
471
2.71
1.79
107.26
76.44
116.57
Bicalutamide (Casodex®)
BMS-641988 (Phase I)
MDV-3100 (Phase II/III)
Figure 1. Androgen receptor antagonists in human use.
N
N
O
N
N
N
N
CF3
CF3
CF3
F
O
O
O
N
N(S)
N
NC
NC
NC
O
O
(R)
(R)
O
O
O
1 [(R,S)/(S,R) racemate]
2
3
MW:
488
470
456
cLogP:
TPSA:
3.73
87.28
3.56
87.28
3.25
87.28
AR cell-based
antagonism IC50:
77 nM
62% inh @ 0.1uM
3.09x@1.0 uM
71 nM
agonism:
1.43x@1.0 uM
0.95x@1.0 uM
HLM unbound Clint:
115 uL/min/mg
3.8 ug/mL
0.00825
43.87 uL/min/mg
3.7 ug/mL
0.0181
19.3 uL/min/mg
5.9 ug/mL
0.0127
sol:
hFu:
LE:
0.28
3.38
0.30
3.89
LipE:
Rat PK:
Fa:
10% (PO)
1585 (IV)
25% (PO)
4294 (IV)
10% (PO)
921 (IV)
unbound Cl:
Figure 2. HTS screening lead and examples of close-in analogs.
Gln711
H12
OH
DHT
Asn705
Arg752
O
MW:
cLogP: 3.55
TPSA:
290
Thr877
37.3
dihydrotestosterone (DHT)
Figure 3. AR-DHT co-crystal structure showing recognition and latch residues.
residual agonism is related to apparently subtle structural changes
(e.g., removal of the methyl group of 2 to give 3). Little is known
about the structure of AR when bound to full antagonists, so pre-
dicting these structural effects is difficult.

1232
C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
Co-activator
Gln711
Asn705
Met895
Ile899
Arg752
H12
DHT
Thr877
Figure 4. AR-DHT co-crystal structure showing hydrophobic residues of helix 12.
O
O
1) Further functionalization
of R if necessary
O
1) LiHMDS or NaH
2) R-X
O
O
O
(R)
F3C
NC
F3C
NC
F3C
NC
R
R
R
N
NH
N
N
2) Resolution by chiral SFC
5
6-11
13-25
and
4
(+/-)
Resolution by chiral SFC
O
O
1) LiHMDS or NaH
O
O
(R)
F3C
NC
F3C
NC
NH
2) R-X
4r
(+) enantiomer, 98%ee
5r
Significant degradation of ee
Scheme 1. Synthesis of right-hand analogs by alkylation/chiral resolution
No co-crystal structure of AR bound to a full antagonist is
available, but crystal structures of AR bound to steroid agonists
such as dihydrotestosterone (DHT)12 show that the ligand binding
domain (LBD) contains a hydrophobic, flexible pocket with two
deep residues (ARG752, GLN711) which form hydrogen bonds
necessary for ligand binding recognition (Fig. 3). The arylcyano
group found in Bicalutamide, MDV-3100, BMS-641988, and com-
pounds 1–3 is presumably the H-bond acceptor for these recogni-
tion residues. Steroid agonists also form hydrogen bonds through
the 17b hydroxyl group to two residues (ASN705, THR877), which
are closer to the surface of the protein. By analogy to the structur-
ally-similar estrogen receptor, it is thought that hydrogen bond-
ing to these ‘latch’ residues causes a conformational change
which allows helix 12 (H12) to close over the binding pocket,
thereby revealing the coactivator binding site. Subsequent coacti-
vator binding initiates agonist effect. In the case of the estrogen
receptor, disruption of H12 closing leads to pure antagonist
activity.13
disrupt H12 folding did not seem consistent with our goal of
improving oral bioavailability for this series. In addition, the length
and rigidity of the phenyl 1,3,4-oxadiazole likely contributes to the
observed poor solubility of these molecules. Therefore, truncation
to reduce molecular weight emerged as a preferred strategy.
When the agonist ligands are bound, hydrogen bonds to the
latch residues are satisfied, and H12 can move to cover the binding
pocket. This makes the AR LBD a more hydrophobic environment
than when the ligand is not bound and the pocket is still exposed
to solvent. If hydrophobic interactions in the steroidal core region
and hydrogen bonding to the latch residues are the driving forces
for AR LBD to adopt the agonism conformation (H12 closed), weak-
ening these interactions may favor the H12-open conformation,
therefore leading to full antagonism. The ability of a molecule to
disrupt such hydrophobic interactions may be a function of its
bulk lipophilicity (as predicted by cLogP) or of its polar surface
area (as predicted by TPSA14). The steroid agonist DHT
(cLogP = 3.55) is very similar in bulk lipophilicity to molecules
1–3 (cLogP = 3.73, 3.56, and 3.25, respectively), but differs greatly
from them in polar surface area (TPSA15: DHT, 37.3 Å2; 1–3,
87.28 Å2). The three clinical antagonists shown in Figure 1 also
have significantly higher TPSA values (76.44–116.57 Å2) than
DHT, though their cLogP values vary by more than 1.5 units. Thus
The phenyl 1,3,4-oxadiazole moiety of compounds 1–3 is long
and rigid and could destabilize the H12-closed conformation by
steric interference, but this moiety does not always render full
antagonism, as indicated by compounds 1 and 2. Preventing resid-
ual agonism by building longer, more rigid molecules to physically

C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
1233
Table 1
SAR of the right-hand group
O
O
(R)
F3C
R
N
NC
Compound
R
cLogP
TPSA (Å2)
Antagonism IC50 (or% at 1.0
633 nM
lM)
Agonism fold induction at 1.0 lM
N
N
6
1.42
87.28
0.97Â
O
CH3
4r
7a
H
2.54
4.27
62.12
53.33
105 nM
(88%)
1.31Â
1.45Â
CH3
O
8
3.29
3.55
70.40
79.63
266 nM
303 nM
(23%)
0.91Â
1.05Â
0.85Â
CH3
O
9b
10b
OCH3
O
2.70
2.80
82.43
77.12
NHCH3
CN
CN
OH
11
12
13
(89%)
2.75Â
0.89Â
1.03Â
2.93
2.64
3.10
77.12
73.56
73.56
131 nM
263 nM
371 nM
337 nM
14c
CH3
1.06Â
15c
1.02Â
OH
CH3
H3C
16
3.35
3.10
73.56
73.56
761 nM
0.91Â
OH
17c
18c
OH
CH3
OH
977 nM
216 nM
0.97Â
0.99Â
19
3.50
4.80
73.56
53.33
(29%)
0.90Â
H3C CH3
20a
21
(85%)
(87%)
2.12Â
1.11Â
CH3
2.79
3.36
73.56
73.56
OH
OH
22c
(96%)
371 nM
1.01Â
23c
0.95Â
CH3
24c
25c
3.04
93.79
(17%)
(27%)
1.01Â
0.92Â
OH
H3C OH
a
b
c
Synthesis discussed in Ref. 16
Racemic mixture tested.
Diastereomeric pair, tested separately: both are (R) at the lactam 3-position.
we hypothesized that analogs of 1 with TPSA values >70 Å2 would
be more likely to be full antagonists.
Racemic 4-(4,4-dimethyl-2-oxopyrrolidin-3-yloxy)-2-(trifluo-
romethyl)benzonitrile (4)16 was alkylated on the lactam nitrogen
This suggested a strategy to simultaneously optimize ADME
properties and reduce residual agonism. Analogs with cLogP val-
ues <3.5 and TPSA >70 Å2 should have better aqueous solubility
and lower unbound clearance, and should also disfavor the hydro-
phobic interactions which induce H12 to close over the binding
pocket. We hypothesized that an analog of 1 with a small, polar
group on the right-hand side would have lower overall lipophilicity
and present a hydrophilic surface toward the hydrophobic region
of H12 (MET895 and ILE899, Fig. 4), disfavoring the H12-closed
conformation. Ideally, the polar group would not form hydrogen
bond interaction(s) to the latch residues (ASN705 and THR877),
therefore avoiding induction of the agonism conformation. As such,
we hoped to obtain full antagonism with a smaller, less lipophilic
molecule, which would be more likely to have favorable pharma-
ceutical properties.
with either an allyl bromide or an a-bromo ester in the presence
of lithium bis(trimethylsilyl) amide. Further functionalization
(ozonolysis or dihydroxylation of an alkene; amidation or reduc-
tion of an ester), followed by resolution by chiral SFC17 gave the de-
sired analogs in >95% ee/de.18 The (+) enantiomers were assigned
(R) stereochemistry at the lactam 3-position by analogy to previous
work.7a,16 Cyanomethyl analog 11 resulted from alkylation of 4 by
bromoacetonitrile, but cyanoethyl analog 12 could not be made by
this method. Instead, compound 12 was synthesized by the meth-
od shown in Scheme 2 and discussed later.
Lactam 4 was resolved by chiral SFC to give the (R)-(+)-enantio-
mer 4r in 98% ee. When 4r was subjected to the same alkylation
conditions as above, significant racemization occurred, giving
products with low as 50% ee, so this route was not pursued further.
In all cases the isolated (S)-(À)-enantiomers were much less potent
(data not shown).
A number of lactam analogs which contained polar groups on
low-MW substituents of various lengths and branching patterns
were synthesized by the general route shown in Scheme 1. Assay
results are shown in Table 1.
Truncation of the terminal phenyl ring of 3 to a methyl group
(compound 6) significantly reduced the lipophilicity (cLogP = 1.42
for 6 vs 3.25 for 3), resulting in a dramatic improvement in solubility

1234
C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
O
O
O
CN
1) NaH
THF
O
CN
H2N
HO
O
Ar
N
H
Ar
O
O
THF
2) Ar-F
OH
28
26 (+/-)
27
O
O
Resolution by
chiral SFC
1) MsCl, Et3N
CH2Cl2
O
O (R)
CN
CN
N
N
Ar
Ar
2) NaH, THF
29
Scheme 2. Synthesis of left-hand analogs.
12, 30-33
(6 217 lg/mL; 3 5.9 lg/mL), with no change in the TPSA or residual
Table 2
agonism. But without the phenyl group, antagonist potency is re-
duced. Complete truncation of the alkyl chain (compound 4r) re-
stored antagonist activity, but at the cost of residual agonism
(1.31-fold induction at 1.0 uM)9, perhaps due to the reduction in po-
lar surface area (TPSA = 62.12 Å2).
SAR of the left-hand group
O
O (R)
N
CN
Ar
Analogs with linear alkyl chains in the right-hand side, such as 7
or 20 (TPSA 53.33 Å2 for each), exhibited significant residual agon-
ism (fold induction at 1.0
l
M >1.4Â), but when the terminal car-
Compound Ar
cLogP TPSA
Antagonism Agonism
(Å2)
IC50 (or% at
1.0 M)
fold
induction at
bon was replaced with a ketone (8), ester (9), or nitrile (12), the
residual agonism is suppressed and the antagonism potency im-
proved. Linear alcohols 13 and 21 show a similar effect. Branching
on the alcohol chains (compounds 14–19 and 22–25), has little ef-
fect on the residual agonism and does not improve antagonist po-
tency, so is not worth the added lipophilicity. Chain length seems
important—cyanomethyl analog 11 has a large residual agonist ef-
fect (2.75-fold induction), but the extra carbon in cyanoethyl ana-
log 12 makes it a full antagonist (fold induction 0.89Â). This shows
a limit of the TPSA hypothesis, as both 11 and 12 are calculated to
have TPSA = 77.12 Å2. From this exercise, hydroxyethyl and cyano-
ethyl were identified as preferred right-hand groups. Compared to
lead molecule 1, nitrile 12 and alcohol 13 are cleaner (albeit
slightly less potent) antagonists, and are less lipophilic (cLogP
3.73 vs 2.93, 2.64).
To further optimize the lipophilicity of these molecules, the
most efficient right-hand group, the cyanoethyl, was held constant
while modifications of the left-hand aryl ether were explored.
While preserving the nitrile for recognition, analogs with less lipo-
philic aryl ethers were synthesized by the general route shown in
Scheme 2. Assay results for these compounds are shown in Table 2.
Racemic pantolactone (26) was deprotonated with sodium hy-
dride and then coupled with the appropriate aryl fluoride to make
lactone ethers 27. Amide alcohols 28 were obtained by stirring the
lactones with 3-aminopropionitrile in a sealed tube for 3–5 days at
room temperature. (Heating this reaction led to undesired imine
side products.) Mesylation followed by sodium hydride-promoted
ring closure afforded racemic lactams 29, which were resolved
by chiral SFC to give the more potent (R)-(+)-enantiomers in
>95% ee.19
l
1.0 lM
F3C
12
2.93
2.59
77.12 131 nM
77.12 68 nM
0.89Â
0.99Â
NC
Cl
30
NC
31
1.96
1.90
2.10
77.12 (10%)
89.48 (52%)
89.48 (47%)
0.95Â
0.89Â
1.16Â
NC
F3C
32
N
NC
F3C
33
N
NC
tein. In contrast, when the trifluoromethyl group on the phenyl
ring is replaced by the only slightly less lipophilic chloride (30,
cLogP 2.59), potent AR antagonism is maintained. From this exer-
cise, trifluoromethyl (as in 12, Fig. 5) and chloro (as in 30, Fig. 5)
were identified as preferred C-2 substituents on the left-hand phe-
nyl ring.
Compared to the lead molecule 1, compounds 12 and 30 are less
lipophilic (cLogP 3.73 vs 2.92, 2.59 respectively) and this resulted
in a marked improvement in aqueous solubility. Compound 12
(57
l
g/mL) is nearly 15Â more soluble, and 30 (159
lg/mL) is
g/mL). Though 12
The results from Table 2 suggested that the left-hand region of
these molecules must contain a minimum amount of lipophilicity
for recognition by AR. When the trifluoromethyl group was re-
moved (31, cLogP 1.96) activity was lost. Pyridyl derivatives 32
and 33 (cLogP 1.90, 2.10) were designed to retain the cyano and
trifluoromethyl groups in their desired orientation, but both mole-
cules were significantly less potent likely due to desolvation pen-
alty (by placing the polar nitrogen in a lipophilic pocket) or
mismatch of dipole–dipole interactions between ligand and pro-
greater than 40x more soluble than 1 (3.8
l
and 30 have AR antagonism potency similar the lead 1, in terms
of ligand efficiency (based on antagonism IC50), 1 is moderately
efficient (LE 0.28) while 12 (LE 0.37) and 30 (LE 0.44) are much im-
proved, mostly due to their lower molecular weight. In terms of
lipophilic efficiency,20 12 (LipE 3.98) and 30 (LipE 4.57) are more
efficient than lead 1 (LipE 3.38) and this should lead to better PK
and lower in vivo drug doses.

C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
1235
O
O
(R)
(R)
O
Cl
O
F3C
NC
N
N
CN
CN
NC
12
30
MW:
351
317
2.59
cLogP:
2.92
AR cell-based
antagonism IC50:
agonism:
131nM
68nM
0.89x@1uM
0.99x@1uM
--21
HLM unbound Clint:
19.1 uL/min/mg
57 ug/mL
0.226
sol:
159 ug/mL
hFu:
LE:
0.37
3.96
0.44
4.57
LipE:
Rat PK:
Fa:
64% (PO)
unbound Cl:
170mL/min/kg (IV)
Figure 5. Optimized 3-aryloxy lactam AR antagonists.
Figure 6. Fraction of molecules with residual agonism vs. TPSA for Pfizer’s collection of AR antagonists.
Compound 1221 was further evaluated for its ADME profile. Its
metabolic stability (HLM unbound Clint 19.1) and permeability
(data not shown) are good. The improved ADME profile was con-
firmed by an in vivo rat PK study which showed oral absorption
of 12 at 64%, a many-fold increase over the lead. The unbound
clearance is also reduced considerably (unbound Cl 170 vs 1585
for 1).
To further investigate our hypothesis that polar surface area
above a minimum threshold is required for full antagonism, resid-
ual agonism of our full collection of AR antagonists (3193 com-
pounds of various structural classes) was retrospectively
analyzed against molecular polar surface area (TPSA, Å2). As shown
in Figure 6, the AR antagonists with higher TPSA are more likely to
be full antagonists whereas the ones with lower TPSA are more
likely to exhibit residual agonism (fold induction >1.03Â). A TPSA
cut-off of >60 Å2 (instead of >70 Å2 as originally hypothesized) ap-
pears to be most predictive for the switch from partial antagonist
to full antagonist. Below this threshold, the majority of compounds
(76%) exhibit residual agonism, while above it, the majority are full
antagonists (56% to 70% depending on TPSA). Despite the general
trend that molecules with higher TPSA are more likely to be full
antagonists, a significant number of the molecules with high TPSA
(>90 Å2) still show residual agonism, suggesting that additional
factors, especially induced conformational change(s), should be
incorporated into the design of AR full antagonists. These design
strategies have been reported elsewhere.22
In summary, a series of novel, non-steroidal AR antagonists
were discovered through HTS; the clearance and solubility of the
initial leads was optimized while maintaining a pure antagonist
profile. Residual agonism was avoided using a hypothesis for full
AR antagonism based on increasing the polar surface area above
a minimum threshold. Deficiencies in the ADME profile were over-
come by designing more ligand-efficient molecules with reduced
molecular weight and lipophilicity. The result is compound 12,
which has good oral bioavailability and will be further evaluated
for its potential as a treatment for castration-resistant prostate
cancer.
Acknowledgment
We wish to thank Dr. Paul Richardson for synthesis of valuable
intermediates.
References and notes
1. Chen, Y.; Sawyers, C. L.; Scher, H. I. Curr. Opin. Pharmacol. 2008, 8, 440.
2. Schellhammer, P. Expert. Opin. Investig. Drugs 1999, 8, 849.
3. Chen, C. D.; Welsbie, D. S.; Tran, C.; Baek, S. H.; Chen, R.; Vessella, R.; Rosenfeld,
M. G.; Sawyers, C. L. Nat. Med. 2004, 10, 33.

1236
C. Guo et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1230–1236
4. Tran, C.; Ouk, S.; Clegg, N. J.; Chen, Y.; Watson, P. A.; Arora, V.; Wongvipat, J.;
Smith-Jones, P. M.; Yoo, D.; Kwon, A.; Wasielewska, T.; Welsbie, D.; Chen, C. D.;
Higano, C. S.; Beer, T. M.; Hung, D. T.; Scher, H. I.; Jung, M. E.; Sawyers, C. L.
Science 2009, 324, 787.
5. Attar, R. M.; Jure-Kunkel, M.; Balog, A.; Cvijic, M. E.; Dell-John, J.; Rizzo, C. A.;
Schweizer, L.; Spires, T. E.; Platero, J. S.; Obermeier, M.; Shan, W.; Salvati, M. E.;
Foster, W. R.; Dinchuk, J.; Chen, S.-J.; Vite, G.; Kramer, R.; Gottardis, M. M.
Cancer Res. 2009, 69, 6522.
6. Scher, H. I.; Beer, T. M.; Higano, C. S.; Anand, A.; Taplin, M.-E.; Efstathiou, E.;
Rathkopf, D.; Shelkey, J.; Yu, E. Y.; Alumkal, J.; Hung, D.; Hirmand, M.; Seely, L.;
Morris, M. J.; Danila, D. C.; Humm, J.; Larson, S.; Fleisher, M.; Sawyers, C. L.
Lancet 2010, 375, 1437.
7. (a) Mitchell, L. H.; Johnson, T. R.; Lu, G. W.; Du, D.; Datta, K.; Grzemski, F.;
Shanmugasuddaram, V.; Spence, J.; Wade, K.; Wang, Z.; Sun, K.; Lin, K.; Hu, L.-
Y.; Sexton, K.; Raheja, N.; Kostlan, C.; Pocalyko, D. J. Med. Chem. 2010, 53, 4422;
(b) Mitchell, L.; Wang, Z.; Hu, L.-Y.; Kostlan, C.; Carroll, M.; Dettling, D.; Du, D.;
Pocalyko, D.; Wade, K. Bioorg. Med. Chem. Lett. 2009, 19, 1310; (c) Mitchell, L.
H.; Hu, L.-Y.; Nguyen, M.; Fakhoury, S.; Smith, Y.; Iula, D.; Kostlan, C.; Carroll,
M.; Dettling, D.; Du, D.; Pocalyko, D.; Wade, K.; Lefker, B. Bioorg. Med. Chem.
Lett. 2009, 19, 2176; (d) Li, J. J.; Iula, D. M.; Nguyen, M.; Hu, L.-Y.; Dettling, D.;
Johnson, T. R.; Du, D. Y.; Shanmugasundaram, V.; Van Camp, J. A.; Wang, Z.;
Harter, W. G.; Yue, W.-S.; Boys, M. L.; Wade, K. J.; Drummond, E. M.; Samas, B.
M.; Lefker, B. A.; Hoge, G. S.; Lovdahl, M. J.; Asbill, J.; Carroll, M.; Meade, M. A.;
Ciotti, S. M.; Krieger-Burke, T. J. Med. Chem. 2008, 51, 7010; (e) Hu, L. Y.; Du, D.;
Hoffman, J.; Smith, Y.; Fedij, V.; Kostlan, C.; Johnson, T. R.; Huang, Y.; Kesten, S.;
Harter, W.; Yue, W. S.; Li, J. J.; Barvian, N.; Mitchell, L. H.; Lei, H. J.; Lefker, B.;
Carroll, M.; Dettling, D.; Krieger-Burke, T.; Samas, B.; Yalamanchili, R.; Lapham,
K.; Pocalyko, D.; Sliskovic, D.; Ciotti, S.; Stoller, B.; Hena, M. A.; Ding, Q.; Maiti,
S. N.; Stier, M.; Welgus, H. Bioorg. Med. Chem. Lett. 2007, 17, 5983; (f) Van Camp,
J. A.; Hu, L.-Y.; Kostlan, C.; Lefker, B.; Li, J. J.; Mitchell, L. H.; Wang, Z.; Yue, W.-S.;
Carroll, M.; Dettling, D.; Du, D.; Pocalyko, D.; Wade, K. Bioorg. Med. Chem.Lett.
2007, 17, 5529; (g) Hu, L.-Y.; Lei, H. J.; Du, D.; Johnson, T. R.; Fedij, V.; Kostlan,
C.; Yue, W.-S.; Lovdahl, M.; Li, J. J.; Carroll, M.; Dettling, D.; Asbill, J.; Fan, C.;
Wade, K.; Pocalyko, D.; Lapham, K.; Yalamanchili, R.; Samas, B.; Vrieze, D.;
Ciotti, S.; Krieger-Burke, T.; Sliskovic, D.; Welgus, H. Bioorg. Med. Chem. Lett.
2007, 17, 5693.
Linton, A.; Kephart, S. E.; Pairish, M.; Guo, C. Tetrahedron Lett. 2011, 52, 4760.
An efficient synthesis of the chiral lactams was developed after this research
work was done.
17. Representative separation procedure: Enantiomers separated on Berger MGII
(Waters) by supercritical fluid chromatography (SFC) using a Chiralpak AD-H
21.2 Â 250 mm column eluted with 20% MeOH in CO2 at 140 bar (flow rate
ꢀ60.0 mL/min). Samples collected by UV at peak times. Solvents removed by
rotary evaporation and lyophilization. A Chiracel OD-H column was used for
compounds 19 and 33.
18. Representative procedure: (R)-(+)-4-(1-(2-hydroxyethyl)-4,4-dimethyl-2-
oxopyrrolidin-3-yloxy)-2-(trifluoromethyl)benzonitrile (13).
A solution of
( )-4-(4,4-dimethyl-2-oxopyrrolidin-3-yloxy)-2-(trifluoromethyl)benzonitrile
(4) Ref. 16 (3.46 g, 11.6 mmol) in THF (23 mL) at 0 °C was treated with a 1.0 M
solution of lithium bis(trimethylsilyl)amide in THF (14.0 mL, 14.0 mmol) and
stirred at 0 °C for 40 min before methyl bromoacetate (1.40 mL, 15.2 mmol)
was added. The resulting solution was stirred at room temperature for 1 hour,
quenched with deionized water (25 mL), and concentrated to remove THF. The
aqueous residue was extracted with dichloromethane (2 Â 50 mL), and the
combined extracts dried over magnesium sulfate, filtered, concentrated, and
purified by silica gel chromatography (eluting with 30–100% ethyl acetate in
heptane) to give racemic ester 9 (3.04 g, 8.22 mmol) as a white solid. 1H NMR
(400 MHz, CHLOROFORM-d) d ppm 1.29 (s, 3H) 1.31 (s, 3H), 3.27 (d, J = 9.35 Hz,
1H), 3.37 (d, J = 9.35 Hz, 1H), 3.77 (s, 3H) 3.97 (d, J = 17.68 Hz, 1H), 4.23 (d,
J = 17.68 Hz, 1H), 4.62 (s, 1H), 7.41 (dd, J = 8.59, 2.53 Hz, 1H), 7.49 (d,
J = 2.27 Hz, 1H), 7.76 (d, J = 8.84 Hz, 1H). A THF (0.95 mL) solution of ester 9
(169 mg, 0.456 mmol) was reduced by stirring with lithium borohydride
(13.2 mg, 0.59 mmol) at room temperature for 26 h. The solution was
quenched with 0.5 M HCl (8 mL), causing gas evolution, then extracted with
ethyl acetate (2 Â 20 mL). The combined organics were dried over magnesium
sulfate, filtered, and concentrated to give crude racemic alcohol
5
(R = CH2CH2OH) as colorless gel (113.6 mg, 0.33 mmol). Two previous
a
batches of crude racemic alcohol were combined with this one for
purification by silica gel chromatography (eluting with 100% ethyl acetate)
and resolution by chiral SFC to give (R)-(+)-enantiomer 13 as a white solid.
Chiral HPLC shows >99% ee. Optical rotation +248.57°. 1H NMR (400 MHz,
CHLOROFORM-d) d ppm 1.26 (s, 3H), 1.28 (s, 3H), 2.26 (t, J = 5.43 Hz, 1H), 3.30
(d, J = 9.85 Hz, 1H), 3.35 (d, J = 9.85 Hz, 1H), 3.44 (dt, J = 14.91, 5.05 Hz, 1H),
3.52 (dt, J = 14.65, 5.31 Hz, 1H), 3.83 (q, J = 5.31 Hz, 2 H), 4.61 (s, 1H), 7.40 (dd,
J = 8.59, 2.53 Hz, 1H), 7.50 (d, J = 2.53 Hz, 1H), 7.76 (d, J = 8.59 Hz, 1H). HRMS:
[M+H]+ calcd 343.126403, found 343.126337, error À0.19 ppm.
8. CRPC cell-based HTS assay: In this assay, prostate cancer refractory cells
(LNAR), stably expressing an AR response element DNA sequence (ARE)-
luciferase reporter gene construct (PSA-LUC), were treated with the testing
compounds in the absence (agonistic mode) or the presence (antagonistic
mode) of a potent agonist (R1881). Upon activation and binding of the AR to
the ARE, the luciferase gene transcribed and translated into active luciferase
enzyme and luminescence was read as signal by the plate reader. Testing
compounds were dissolved in 100%DMSO as 10 mM stock solution. Serial
19. Representative
procedure:
(R)-(+)-4-(1-(2-cyanoethyl)-4,4-dimethyl-2-
oxopyrrolidin-3-yloxy)-2-(trifluoromethyl)benzonitrile (12):
A solution of
( )-4-(4,4-dimethyl-2-oxotetrahydrofuran-3-yloxy)-2-
(trifluoromethyl)benzonitrile Ref. 16 (2.9957 g, 10.01 mmol) and 3-
aminopropionitrile (1.55 mL, 21.2 mmol) in THF (5.0 mL) was stirred in a
sealed tube at room temperature for 5 days. The solution was concentrated and
purified by silica gel chromatography (eluting with 40-100% ethyl acetate in
heptane) to give amide/alcohol 28 (Ar = 2-CF3-3-CN phenyl) as a sticky white
dilutions were prepared from 0.17 nM to 10 lM and final DMSO concentration
never exceeded 0.1%. For agonism, values obtained from the compounds under
study were compared to those of untreated cells, which were assigned an
arbitrary number of 1.0 to indicate no agonism. For antagonism, cells were
treated with 0.1 nM R1881 alone (corresponding to max receptor
activation = 100%) or in combination with the various compounds.
foam (3.58 g, 9.15 mmol).
A solution of this alcohol in dichloromethane
(37 mL) and triethylamine (4.0 mL, 29 mmol) was cooled to 0 °C before adding
methane sulfonyl chloride (1.7 mL, 22 mmol). After stirring at 0 °C for 2 h, the
solution was partitioned between ethyl acetate and aqueous ammonium
chloride. The organic layer was washed with brine, dried over magnesium
sulfate, filtered and concentrated to give crude mesylate (3.79 g, 8.47 mmol) as
a sticky yellow foam. The mesylate was dissolved in THF (45 mL), and cooled to
0 °C before sodium hydride (60 wt % dispersion in mineral oil, 1.95 g, 49 mmol)
was added. Stirring continued, gradually warming to room temperature, for
26 hours. The reaction solution was diluted with ethyl acetate (50 mL) and
quenched with aqueous ammonium chloride (30 mL), causing vigorous gas
evolution. The layers were separated and the aqueous layer back-extracted
with ethyl acetate (30 mL). The combined organic extracts were dried over
magnesium sulfate, filtered, concentrated, and purified by silica gel
chromatography (eluting with 40–100% ethyl acetate in heptane) to give
racemic 29 (Ar = 2-CF3-3-CN phenyl) as a light yellow gum (1.39 g, 3.95 mmol).
Enantiomeric resolution by chiral SFC gave (R)-(+)-enantiomer 12 as an off-
white solid (602.63 mg, 1.71 mmol). Chiral HPLC shows >99% ee. Optical
rotation +174.90°. 1H NMR (400 MHz, CHLOROFORM-d) d ppm 1.28 (s, 3H),
1.31 (s, 3H), 2.68 (t, J = 6.32 Hz, 2H), 3.41 (s, 2H), 3.57 (dt, J = 13.89, 6.32 Hz,
1H), 3.64 (dt, J = 13.89, 6.32 Hz, 1H), 4.59 (s, 1H), 7.39 (dd, J = 8.84, 2.53 Hz, 1H),
7.49 (d, J = 2.27 Hz, 1H), 7.77 (d, J = 8.59 Hz, 1H). HRMS: [M+H]+ calcd
352.126738, found 352.126681, error À0.16 ppm.
9. The cut-off for significant residual AR agonism was set to >1.032 fold induction
in CRPC cell-based assay (agonism mode, see Ref. 8) based on statistic analysis
of multiple test results (n = 176) of RD162 in the same assay. RD-162, a full AR
antagonist reported in literature (Ref. 4), was used as a reference standard. In
the agonism assay, the averaged agonism fold induction of RD162 (n = 176) is
0.876 with a standard deviation of 0.0779. An AR ligand with an agonism fold
induction >1.032 (0.876 + 2 Â 0.0779) is likely (>95% confidence) to have more
residual agonism than RD162.
10. HLM unbound Clint was calculated with a formula of HLM Clint/Fum. Here
HLM Clint corresponds to the intrinsic metabolic clearance (Clint) in
microsomes, in
lL/min/mg of microsomal protein whereas Fum is the
calculated unbound fraction of the drug in human liver microsomes by a
computational model developed at Pfizer. Data interpretation as following:
Clint <15 lL/min/mg (low clearance); Clint 15–40 lL/min/mg (moderate
clearance); Clint >40 (high clearance). HLM data from the same assay has
been reported previously. See references: (a) Dong, L.; Marakovits, J.; Hou, X.;
Guo, C.; Greasley, S.; Dagostino, E.; Ferre, R.; Johnson, M.C.; Kraynov, E.;
Thomson, J.; Pathak, V.; Murray, B.W. Bioorg. Med. Chem. Lett. 2010, 20, 2210–
2214. (b) Linton, A.; Kang, P.; Ornelas, M.; Kephart, S.; Hu, Q.; Pairish, M.; Jiang,
Y.; Guo, C. J. Med. Chem. 2011, 54, 7705.
11. In vivo unbound clearance was calculated with a formula of Cl/Fu. Here Cl is
drug plasma clearance measured in rat PK study whereas Fu is the measured
unbound fraction of the drug in rat plasma.
12. Pereira de Jésus-Tran, K.; Côté, P.-L.; Cantin, L.; Blanchet, J.; Labrie, F.; Breton, R.
Prot. Sci. 2006, 15, 987.
13. Pike, A. C. W.; Brzozowski, A. M.; Hubbard, R. E.; Bonn, T.; Thorsell, A.-G.;
Engström, O.; Ljunggren, J.; Gustafsson, J.-Å.; Carlquist, M. EMBO J. 1999, 18,
4608.
20. (a) Ryckmans, T.; Edwards, M. E.; Horne, V. A.; Correia, A. M.; Owen, D. R.;
Thompson, L. R.; Tran, I.; Tutt, M. F.; Young, T. Bioorg. Med. Chem. Lett. 2009, 19,
4406; (b) Guo, C.; Hou, X.; Dong, L.; Dagostino, E.; Greasley, S.; Ferre, R.;
Marakovits, J.; Johnson, M. C.; Matthews, D.; Mroczkowski, B.; Parge, H.;
VanArsdale, T.; Popoff, I.; Piraino, J.; Margosiak, S.; Thomson, J.; Los, G.; Murray,
B. W. Bioorg. Med. Chem.Lett. 2009, 19, 5613.
21. Compound 30 ionizes poorly under the MS conditions used in the HLM assays,
so reliable data for this compound could not be obtained from this assay.
22. Guo, C.; Linton, M. A.; Kephart, S.; Ornelas, M.; Mason, P.; Gonzalez, J.;
Greasley, S.; Nagata, A.; Burke, B. J.; Edwards, M.; Hosea, N.; Kang, P.; Hu,
W.; Engebretsen, J.; Briere, D.; Shi, M.; Gukasyan, H.; Richardson, P.; Dack,
K.; Underwood, T.; Johnson, P.; Morell, A.; Felstead, R.; Kuruma, H.;
Matsimoto, H.; Zoubeidi, A.; Gleave, M.; Los, G.; Fanjul, A. N. J. Med. Chem.
2011, 54, 7693.
14. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.
15. TPSA values were calculated using ChemBioDraw Ultra version 11.0.1,
CambridgeSoft. This program does not include the contributions of polar
sulfur atoms, so it differs slightly from the model presented in Ref. 14
16. (a) Barrett, S.D.; Fedij, V.; Hu, L-Y; Iula, D.M.; Lefker, B.A.; Raheja, R.K.; Sexton,
K.E.; Van Camp, J.A. Androgen Modulators. Patent US2006/252796 A1, 2006.;
(b) Ornelas, M. A.; Gonzalez, J.; Sach, N. W.; Richardson, P. F.; Bunker, K. D.;