5708
W. Shan et al. / Bioorg. Med. Chem. Lett. 26 (2016) 5707–5711
O
N
between arginine 752 (R752) and the aryl nitrile of compound 5
are also evident. Furthermore, there is a lipophilic pocket between
threonine 877 (T877) and the bicyclic ring of the sultam. It is
believed that the orientation of Helix-12 is, in part, responsible
for agonist/antagonist function of nuclear hormone receptors.11
Therefore we investigated substitutions on the bicyclic sultam ring
at C5 to take advantage of the lipophilic pocket and potentially per-
turb Helix-12 to a more favorable antagonist conformation.
A series of sultams was synthesized by the methods shown in
Scheme 1. Dienophile 6 was synthesized according to literature
precedent19 and subsequent Diels–Alder reaction with cyclopenta-
diene in dichloromethane at room temperature resulted in forma-
tion of the desired bicycle 7 as a 95:5 ratio of endo and exo isomers
in 75% yield. Fortuitously, the endo and exo isomers were easily
separated by silica gel chromatography. The endo isomer was then
reduced to the corresponding alcohol 8 by treatment with sodium
borohydride. At this stage, normal phase chiral HPLC (AD column,
Hexane/IPA/DEA 73%/27%/0.1%) was performed to give enan-
tiomers 9A and 9B in >99% ee. Antipode 9A which is shown in
Scheme 1, led to the active series whereas the enantiomer 9B led
to inactive analogs (Data not shown). The absolute stereochemistry
was confirmed by single crystal X-ray diffraction measurement.
With enantiomerically pure alcohol 9A in hand, the sultam ring
was closed by an intramolecular Mitsunobu reaction with triph-
enylphosphine and diisopropyl azodicarboxylate which afforded
10 in 76% yield over 2 steps. Catalytic hydrogenation (Pd/C) of 10
led to formation of the saturated analog 5. Treatment of either 5
or 10 with lithium bis (trimethylsily) amide followed by methyl
iodide yielded C5 methylated analogs 11 and 12.
O
O
OH
O
H
N
S
CF3
CN
N
F
NC
F3C
H
N
S
F
Me
O
1.Bicalutamide
2. Enzalutamide
N
O
H
O
S
O
O
N
CN
H
N
O
H
H
CF3
O
H
H
O
4. BMS-641988
3. Abiraterone acetate
Figure 1. Known modulators of androgen receptor mediated signaling.
Table 1
Androgen receptor binding (Ki) and functional antagonist activity (IC50) of sultam
analogs
Compound no.
MDA-MB-453
MDA-MB-453
IC50 (nM)21
Ki (nM)20
1
4
5
10
11
12
37
2
3
12
3
2
173
16
130
219
73
30
H
The data in Table 1 reveals some exciting results with C5 methyl
substituted compounds 11 and 12, both of which have improved
functional antagonist potency compared to compounds 5 and 10.
This was in line with prediction from our models. Most notable
was compound 12, which was comparable to our clinical com-
pound 4 in terms of in vitro potency.
N
O
CN
CF3
S
O
H
5
Figure 2. Initial [2.2.1]-bicyclic sultam.
Based on this promising profile, compound 12 was evaluated in
the immature rat prostate weight (IRPW) PK/PD model, where the
compound effect on AR dependent growth of the prostate and sem-
inal vesicles was measured.22 In this model, compounds were
dosed orally once daily for 4 days with plasma concentrations of
drug measured 2 h post-dose on day 4. Agents that effectively
block the proliferative effect of the AR in these tissues would result
in a decrease in the total weight of organs relative to a control
group. While the exposure of compound 12 was very low
(0.05 0.014 lM) compared to compound 4, it still had a robust
PD effect (35 5% at 25 mg/kg) compared to the castration control
(32% 4%). Therefore, compound 12 was progressed into an effi-
cacy study in the CWR22-LD1 human prostate cancer xenograft
model which has been shown to be refractory to treatment with
bicalutamide.14 In this study (Fig. 4), 12 and bicalutamide (1) were
administered with daily oral dosing at 150 mg per kg for 20 days.
As shown in Figure 4, 1 had modest activity (39% Tumor Growth
Inhibition). Animals receiving compound 12 exhibited tumor stasis
during the course of treatment (87% TGI on the last day of dosing).
There was no observed toxicity in this study.
Similar to the results from the IRPW study where the robust PD
effects could not be rationalized on the basis of the observed very
low exposures in animals, the observed efficacy in tumor models
was not consistent with the very low exposure of compound 12
(Not detected) in mouse serum. To investigate this further, we per-
formed mouse liver microsome incubations of compound 12. In
this study, multiple oxidative metabolites were observed, suggest-
ing the possible presence of multiple active agents in vivo. Previous
experience in our lab suggested that sustained drug exposure over
24 h was necessary for an AR antagonist to be effective. Based on
the metabolic profile of [2.2.1]-bicyclic imides from our lab,10,11
Figure 3. Compound 5 docked into WT AR Ligand Binding Domain. Helix-4 is
hidden to facilitate visualization of the ligand. Helix-11 is colored in red.
IC50 of 130 nM (Table 1). Notably, this compound was found to be
stable under a wide pH range. Previous efforts from our laborato-
ries led to the first crystal structures of the AR with DHT, as well
as several small molecule agonists from our earlier bicyclic imide
series.16 Using the available crystal structure of an AR modulator
(PDB1XNN), we constructed a model (Fig. 3) of compound 5 bound
to the AR ligand binding domain (LBD) using the software Glide17
followed by Macro Model18 energy minimization. We explored
potential sites to increase interactions between compound 5 and
the AR LBD protein backbone to improve antagonist activity.
In this model, phenylalanine 764 (F764) forms an edge-face
interaction with the aromatic ring of compound 5. The interactions