A.J. Nielsen, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
reaction of compounds 11a/12a with Raney nickel, followed by de-
protection with TBAF. Compounds 14 and 15 and were readily pre-
pared from the reaction of 1,2,4-triazole with the allyl bromide derived
from 7a.
The collection of compounds was next screened for AI activity
against recombinant human aromatase via kinetic monitoring of the
conversion of dibenzylfluorescein (DBF) substrate to fluorescein.
Fluorometric measurement of emission was made at 535 nm after ex-
citation at 485 nm utilizing ketoconazole as a positive control following
The results of this assay demonstrated that newly synthesized tria-
zoles 5, 6, 14 and 15 all inhibited human recombinant aromatase with
a Ki in the low nanomolar range in vitro. Several structure-activity-re-
lations (SARs) were immediately discernible upon examination of the
data. First, halogenated analogs were consistently more potent than
their non-halogenated derivatives, consistent with the ketone-bioisos-
tere hypothesis. Secondly, there is a clear trend when comparing the
geometric isomers 5 and 6, as the (Z)-compounds 6 proved consistently
more potent than their (E)-counterparts 5, possibly reflecting a better
overlap with the steroidal core of the native substrate 3. Third, com-
parison of derivatives 6a, 6b and 6c directly (Table 3, entries 6–8)
showed a clear synergistic effect with double halogenation resulting in
the most potent activity. This result is fully consistent with the ketone
bioisostere hypothesis (i.e. in relation to androstenedione). Fourth, aryl
chloride analogs were consistently more potent than the corresponding
aryl bromides, for example, comparing 6b/6c with 6d/6e, indicating
the carbon-chlorine bond to be a better ketone mimic (sterically and
electronically) than the carbon-bromine bond. Finally, conformational
rigidity improves activity, as the hydrogenated product 13 was less
active than either corresponding isomer, 5a and 6a. The 1,2,4-triazole
derivatives 14 and 15 were also 2–3 times more potent than their re-
lated 1,2,3-triazole derivatives 5a and 6a, presenting an interesting
avenue for further investigation. Finally, when comparing these new
compounds with previously developed inhibitors such as 4, it is now
clear that hydrogen bonding acceptor groups in the core of the in-
hibitors (such as methoxymethyl) are not required for potent anti-ar-
omatase activity.
Scheme 3. Synthesis of stilbene derived AIs. Reagents and conditions: (a) NBS,
BPO, benzene, 70 °C, 4 h; (b) NaN3, ACN, rt, 12 h; (c) TMS-acetylene, CuI, THF,
50 °C, 12 h; (d) TBAF, THF, 50 °C, 12 h; (e) Separation over silica gel; (f) Raney
Ni, MeOH, rt, 12 h. (g) NaH, 1,2,4-triazole, THF, 0 °C – rt, 36 h.
the desired α-methylstilbene in 97% yield (3:1 E/Z). The reaction
proceeds more slowly at lower temperatures, with an 80% yield ob-
tained after heating overnight at 70 °C. Similar yields and E/Z ratios
(∼2.3:1) were obtained using all salts prepared.
Overall, this structure-activity study has identified compound 6c as
a potent aromatase inhibitor with a Ki of 8 nM. This compound is a
structurally rigid stilbene featuring synergistic para-chloro substitution
on each phenyl ring. Compound 6c is the most potent analog discovered
thus far in the series. All SAR data agree with previous studies from our
group, suggesting that aryl chlorides act as bioisosteres for the keto-
groups of the native aromatase substrate, androstenedione 3.5
In conclusion, a novel, high yielding aqueous Wittig methodology
for the synthesis of α-methylstilbenes has been developed. We have
To compare our aqueous Wittig method to more classical condi-
tions, a reaction was performed with the unsubstituted α-methylbenzyl
tripropylphosphonium salt and benzaldehyde using n-BuLi in THF at
−78 °C with warming to room temperature. The yield proved to be
substantially lower (32%) and the E/Z (3.2:1) ratio was not sig-
nificantly better. Moreover, the product required purification using
column chromatography. A major benefit of the aqueous methodology
is that the product precipitates from solution and is readily isolated via
filtration and washing with water, as all salts and the phosphine oxide
side-product are water soluble.
Table 3
Inhibitory activity of select α-methylstilbene derivatives on recombinant
human aromatase. Values are the means of three separate experiments.
Having developed a high yielding synthesis of these α-methyl-
stilbenes, we explored a short synthesis of 1,2,3-triazole stilbenes
5(a–e) and 6(a–e), as well as the hydrogenated derivative 13 and 1,2,4-
triazole derivatives 14 and 15 (Scheme 3). Briefly, the α-methyl-
using n-bromosuccinimide and benzoyl peroxide as a radical initiator,
followed by substitution with sodium azide; the two step yields were
typically between 70 and 80%. The azides were then used to synthesize
the TMS-protected triazoles 11 and 12 via a copper catalyzed Huisgen
dipolar cycloaddition with TMS-acetylene in 60–80% yield.9 The E and
Z isomers 11 and 12 were separable using silica gel chromatography,
but thermally isomerized during the subsequent deprotection of the
triazole group using tetra-n-butylammonium fluoride (TBAF), which
gave deprotected isomers 5 and 6 in combined yields of 75–90%. The
final products were also separable using silica gel chromatography. The
hydrogenated derivative 14 was prepared in 80% two-step yield via the
Entry
Compound
X=
Y=
Ki (µM)
1
5a
5b
5c
5d
5e
6a
6b
6c
6d
6e
13
14
H
Cl
Cl
Br
Br
H
Cl
Cl
Br
Br
H
–
H
H
Cl
H
Br
H
H
Cl
H
Br
H
–
0.180
0.081
0.097
0.125
0.45
2
3
4
5
6
0.100
0.057
0.008
0.074
0.024
0.290
0.079
0.032
0.020
7
8
9
10
11
12
13
14
–
–
–
–
* Inhibitory activity on human aromatase previously reported.5a
3