New structure-activity relationships of A-and D-ring modified steroidal aromatase inhibitors: Design, synthesis, and biochemical evaluation

Article abstract of DOI:10.1021/jm300262w

A- and D-ring androstenedione derivatives were synthesized and tested for their abilities to inhibit aromatase. In one series, C-3 hydroxyl derivatives were studied leading to a very active compound, when the C-3 hydroxyl group assumes 3β stereochemistry (1, IC₅₀ = 0.18 μM). In a second series, the influence of double bonds or epoxide functions in different positions along the A-ring was studied. Among epoxides, the 3,4-epoxide 15 showed the best activity (IC₅₀ = 0.145 μM) revealing the possibility of the 3,4-oxiran oxygen resembling the C-3 carbonyl group of androstenedione. Among olefins, the 4,5-olefin 12 (IC₅₀ = 0.135 μM) revealed the best activity, pointing out the importance of planarity in the A,B-ring junction near C-5. C-4 acetoxy and acetylsalicyloxy derivatives were also studied showing that bulky substituents in C-4 diminish the activity. In addition, IFD simulations helped to explain the recognition of the C-3 hydroxyl derivatives (1 and 2) as well as 15 within the enzyme.

Full text of DOI:10.1021/jm300262w

Article
pubs.acs.org/jmc
New Structure−Activity Relationships of A- and D-Ring Modified
Steroidal Aromatase Inhibitors: Design, Synthesis, and Biochemical
Evaluation
Carla Varela,† Elisiario J. Tavares da Silva,*,† Cristina Amaral,‡,§ Georgina Correia da Silva,‡,§
́
Teresa Baptista,∥ Stefano Alcaro,⊥ Giosue Costa,⊥ Rui A. Carvalho,# Natercia A. A. Teixeira,‡,§
̀
́
and Fernanda M. F. Roleira*,†
†CEF, Center for Pharmaceutical Studies, Pharmaceutical Chemistry Group, Faculty of Pharmacy, University of Coimbra, 3000-548
Coimbra, Portugal
‡Biochemistry Laboratory, Faculty of Pharmacy, University of Porto, Rua Jorge Viterbo Ferreira, 228, 4050-313 Porto, Portugal
§Institute for Molecular and Cellular Biology (IBMC), University of Porto, Rua do Campo Alegre, 823, 4150-180 Porto, Portugal
∥CEF, Center for Pharmaceutical Studies, Bromatology, Pharmacognosy and Analytical Sciences Group, Faculty of Pharmacy,
University of Coimbra, 3000-548 Coimbra, Portugal
⊥CCLab, Dipartimento di Scienze della Salute, Universita
̀
“Magna Græcia” di Catanzaro, Italy
#Department of Life Sciences, Faculty of Science and Technology, University of Coimbra, Portugal
S
* Supporting Information
ABSTRACT: A- and D-ring androstenedione derivatives were
synthesized and tested for their abilities to inhibit aromatase.
In one series, C-3 hydroxyl derivatives were studied leading to
a very active compound, when the C-3 hydroxyl group
assumes 3β stereochemistry (1, IC50 = 0.18 μM). In a second
series, the influence of double bonds or epoxide functions in
different positions along the A-ring was studied. Among
epoxides, the 3,4-epoxide 15 showed the best activity (IC50
=
0.145 μM) revealing the possibility of the 3,4-oxiran oxygen
resembling the C-3 carbonyl group of androstenedione.
Among olefins, the 4,5-olefin 12 (IC50 = 0.135 μM) revealed
the best activity, pointing out the importance of planarity in
the A,B-ring junction near C-5. C-4 acetoxy and acetylsalicy-
loxy derivatives were also studied showing that bulky substituents in C-4 diminish the activity. In addition, IFD simulations
helped to explain the recognition of the C-3 hydroxyl derivatives (1 and 2) as well as 15 within the enzyme.
INTRODUCTION
■
Aromatase is a cytochrome P-450 enzyme that catalyzes the
aromatization of androgens, the final step in the biosynthesis of
estrogens. Aromatase inhibitors (AIs) reduce the synthesis of
estrogens and offer a therapeutic alternative for the treatment of
estrogen-dependent cancers such as breast cancer.1−3 There are
two classes of AIs, steroidal and nonsteroidal compounds,4−6
which cause potent estrogen suppression.3,7 The nonsteroidal
AIs are mostly azole type compounds such as the clinically used
anastrazole and letrozole, which compete with the substrate for
binding to the enzyme active site. Steroidal AIs, as exemestane
and formestane (23) (Figure 1), mimic the natural substrate
androstenedione (8) and are converted by the enzyme in
reactive intermediates, which bind irreversibly to the enzyme
active site, resulting in the inactivation of aromatase. Despite
the success of the third-generation nonsteroidal (anastrazole
and letrozole) and steroidal (exemestane) AIs, they still have
some major side effects, such as increase of bone loss, joint
Figure 1. Steroidal aromatase inhibitors.
pain, and heart problems.8 In addition, after some years of
usage they can develop cellular resistance. For these reasons, it
is important to search for other potent and specific molecules
with lower side effects and which can overcome the resistance
phenomena.
Received: February 21, 2012
Published: April 4, 2012
© 2012 American Chemical Society
3992
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
a
Scheme 1. Synthesis of Aromatase Inhibitors from 3 and 2
a
Reagents and conditions: (i) NaBH4, methanol, rt; (ii) (t-butoxi)3AlLiH, THF, reflux, 8 h 20 min; (iii) (CH3CO)2O, pyridine, rt, 21 h 25 min.
Recently, the active site of aromatase has been elucidated,
and the molecular basis for enzyme−substrate interaction has
been established.9 It was found that the volume of the binding
pocket taken from the enzyme−substrate complex is relatively
short (no more than 400 Å3) allowing to enter into the cleft
only molecules with appropriate dimensions such as derivatives
of 8 with small substituents. These findings can lead to a more
rational design of new AIs and consequently to a more
efficacious intervention at the level of estrogen production.
Keeping this in mind, the design, synthesis, and biochemical
evaluation of new substrate-based inhibitors, containing the A-
and D-ring chemical key-features important for enzyme−drug
interaction, would greatly contribute to establish new
structure−activity relationships (SARs). These SAR are
valuable tools for understanding the enzyme inhibition
mechanism and to find more selective, potent, and lower side
effect AIs.
Among other interactions, aromatase establishes two hydro-
gen bonds with the carbonyl functions at C-3 and C-17 of 8.9
From a previous study, it was concluded that the presence of
the carbonyl group at C-3 is not mandatory to bind steroid
molecules to the enzyme aromatase.10 Other authors described
AIs without the C-17 carbonyl group.11 Recently, our group
confirmed those facts and postulated that, at least, one carbonyl
group (C-3 or C-17) is necessary in order to allow the binding
of steroid molecules to the enzyme aromatase.12 In the present
work, one of our aims is to study steroid molecules as AIs in
which the carbonyl group at C-3 (a hydrogen bond acceptor)
of 8 was replaced by a hydroxyl group (a hydrogen bond
donor) and also in which the two carbonyl groups (C-3 and C-
17) were replaced by two hydroxyl groups. The effect of the
stereochemistry of the hydroxyl group at C-3 was also explored
(Scheme 1).
double bond or the epoxide function along the A-ring, in
aromatase inhibitory activity. For this, we prepared two series of
steroid compounds (Scheme 2) and studied their inhibitory
activity against aromatase.
Compound 23 is the C-4 hydroxy analogue of 8 and was the
first steroidal AI clinically used. Its C-4 acetoxy derivative 24
also revealed strong aromatase inhibitory activity.13 Related
with this, we present a preliminary SAR study based on the
synthesis (Scheme 3) and aromatase inhibitory activity
evaluation of C-4 acetoxy and acetylsalicyloxy derivatives of
23 and of 3,4-olefin 14 (Scheme 2), another potent AI
prepared and evaluated by our team.10
Finally and based on the importance of the C-17 carbonyl
group in steroidal AIs, the chemical modification of the C-17
carbonyl oxygen of compound 14 by its sulfur isoster 28
(Scheme 4) was performed.
Complementary docking studies by Glide/IFD simulations14
have also been done. Actually, cytochromes P450 can adopt
multiple conformations depending on the bound ligand
because of the enzyme’s plasticity.15 For this reason, we
generate poses of our compounds using IFD simulations, in
order to take into account the binding site flexibility and to
optimize the network of protein−ligand interactions as
compared to rigid docking. In our analysis, we have considered
the binding contributions, the induced fit effects onto the
enzyme residues, and the deviation of the steroidal scaffold by
the experimentally determined position of 8 with respect to the
reference crystallographic model.9
RESULTS
■
Chemistry. The synthesis of the 3β-hydroxy derivative of
testosterone 4 was performed through two different methods
(Scheme 1). In the first approach, testosterone (3) reacted with
sodium borohydride, in methanol, at room temperature
yielding a mixture of the 3β- and 3α-isomers, which after
crystallization, afforded 21% of compound 4 (Method A). In
the second method, 3 was refluxed in tetrahydrofuran with
lithium tri-t-butoxyaluminum hydride.16 This reduction also led
to a mixture of the 3β- and 3α-isomers, which after
In our recent studies, we also noticed that some planarity in
the A-ring and in the A,B-ring junction is important for the
inhibitory activity of steroids against aromatase.10,12 This
planarity can be conferred by a double bond or by an epoxide
function both containing similar bond geometries. Presently,
we are interested in studying the influence of the position of the
3993
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
a
Scheme 2. Synthesis of Aromatase Inhibitors from 3, 16, and 18
a
Reagents and conditions: (i) CrO3, H2SO4, acetone, rt; (ii) CH3COOH, Zn dust, reflux, 15 min; (iii) H2O2, HCOOH, CH2Cl2, rt; (iv)
CH3COOOH, CH3COONa·3H2O, CHCl3, rt, 7 h 30 min; (v) HSCH2CH2SH, anhydrous p-toluenesulfonic acid, anhydrous THF, rt, 4 h 5 min;
(vi) Na, NH3, anhydrous THF, −65 °C, 25 min; (vii) (t-butoxi)3AlLiH, anhydrous THF, reflux, 3h 15 min; (viii) SOCl2, benzene, 5−8 °C, 3 h 15
min; (ix) AlLiH4, diethyl ether, reflux, 10 h 30 min; (x) CrO3, H2SO4, acetone, 0 °C.
crystallization gave compound 4 in 22% yield (Method B).
Compound 3 was also converted to its 17β-acetate 6 in acetic
anhydride in 84% yield (Scheme 1).17 Compound 6 was then
reduced by sodium borohydride, leading to a mixture of the 3β-
7 and 3α-epimers (90:10, respectively, NMR and HPLC).
Attempts to isolate 7 by crystallization led, however, to
enrichment of the mixture in the 3α-isomer. Compound 5 was
obtained quantitatively from the commercially available
compound 2, by reducing this last compound with sodium
borohydride in methanol, at room temperature (Scheme 1).
Compound 8 was prepared through the oxidation of 3 with
Jones Reagent in 98% yield (Scheme 2).12 Protection of the C-
3 carbonyl group of 8 was undertaken by treatment with
ethane-1,2-dithiol in anhydrous THF and in the presence of
anhydrous p-toluenesulfonic acid.18 The crude product
obtained was purified by column chromatography affording
the protected compound 9 in 84% yield and also the product
resulting from the simultaneous protection of the C-3 and C-17
carbonyl groups, in 5% yield. Desulfurization of compound 9
with sodium-ammonia in anhydrous THF19 afforded com-
pound 12 in 46% yield and compound 10 in 26% yield. The
crude material resulting from the treatment of compound 12, in
dichloromethane, with performic acid was purified by
crystallization and column chromatography allowing the
isolation of the main product of the reaction (one TLC
spot). NMR analysis of this product revealed it to be a mixture
of the isomers 13a and 13b, in 66:34 proportion. Further
purification by column chromatography allowed the isolation of
3994
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
a
Scheme 3. Synthesis of Aromatase Inhibitors from 23
a
Reagents and conditions: (i) acetyl chloride, dry pyridine, rt, 21 h 50 min; (ii) o-acetylsalicyloyl chloride, dry pyridine, rt, 24 h 40 min; (iii) Zn dust,
CH3COOH, rt.
a
From this reaction, a complex mixture of products was obtained
and used as starting material in the next reaction. Treatment of
this mixture with lithium aluminum hydride led, after workup,
to the main product 20 which after Jones oxidation afforded
compound 21. Treatment of 21 with a solution of performic
acid in dichloromethane gave compound 22 in 15% yield.
Compound 23 was treated with acetyl chloride in dry
pyridine leading, after crystallization, to the pure compound 24,
in 72% yield (Scheme 3). Compound 24 was treated with dust
zinc in acetic acid solution leading to a mixture of both 5α- and
5β-epimers which after crystallization with petroleum ether
allowed the isolation of the pure epimer 25. Treatment of 23
with o-acetylsalicyloyl chloride in pyridine led, after purification
by column chromatography, to a main fraction (one single TLC
spot) which after NMR analysis was revealed to be composed
by a mixture of compounds 24 and 26, in 40:60 proportion.
The formation of compound 24 probably resulted from an
intramolecular transesterification occurring in 26. Further
purification by column chromatography using a different
mixture of solvents allowed the isolation of the pure compound
26. Treating 26 with zinc dust in acetic acid solution led to an
inseparable epimeric mixture of the ester 27a and its 5β-epimer
27b in a 70:30 proportion (NMR analysis) (Scheme 3).
The synthesis of thione 28 was performed with Lawesson’s
reagent22 and required an environment with controlled
humidity. This reaction occurred within 7 h, although it was
not complete. The reaction mixture was worked up with a prior
elimination of Lawesson’s reagent using an aluminum oxide
neutral column. This additional step to the conventional
workup was developed after observing that there was a reaction
of Lawesson’s reagent with silica gel leading to a complete
decomposition of the product when the crude was directly
chromatographed through a silica gel column. After reagent
elimination, the crude product was purified by silica gel column
Scheme 4. Synthesis of 28 from 14
a
Reagents and conditions: (i) Lawesson’s reagent, dry toluene, reflux,
7 h.
the pure compound 13a in 7.6% yield. Compound 10 was
treated in the same way as compound 12, allowing the isolation
of the main product of the reaction, which after NMR analysis
revealed itself to be a mixture of the two 4,5-epoxide isomers
11a and 11b, in a 60:40 proportion (Scheme 2). In agreement
with a previous description of our group,10 a Clemmenson-type
reduction of 8 with zinc dust in acetic acid solution gave a
mixture of 5α- and 5β-epimers from which the 5α-epimer 14
was isolated by n-hexane crystallization in 60% yield. Treatment
of 14 with performic acid in dichloromethane led to the
epoxide derivative 15, in 96% yield (Scheme 2).10 The 2,3-
epoxide 17 was synthesized from the commercially available
2,3-olefin 16 by two different pathways: via performic acid,
Method A,12 and via peracetic acid, Method B.20 As the starting
material 16 was only available in 92% purity (it is supplied in a
92:8 inseparable mixture with 14), the resulting compound 17
was also obtained in 92% purity (NMR and HPLC analysis)
(Scheme 2), by both methods. 1,2-Olefin 21 was prepared
following a described strategy.17,21 Briefly, treatment of enone
18 with lithium tri-t-butoxyaluminum hydride gave the desired
allylic alcohol 19 in 94% yield (Scheme 2). Compound 19 was
then treated with thionyl chloride in benzene, under nitrogen.
3995
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
chromatography and afforded thione 28, in 54% yield (Scheme
4).
substrate (Table 2). These steroids were revealed to be
competitive inhibitors. Representative Lineweaver−Burk and
Dixon plots for compound 13a are presented in Figures 2 and
3, respectively.
Biochemistry. Inhibition of aromatase activity of the
studied A- and D-ring modified steroids was evaluated in
human placental microsomes by a radiometric assay in which
tritiated water, released from [1β-3H] androstenedione into the
incubation medium, was used as an index of estrogen
formation.23 A screening assay was performed, and the results
(Table 1) are shown as a percentage of inhibition (%) for all
Table 1. Aromatase Inhibition of Tested Compounds at 2
a
μM in Human Placental Microsomes
compds
aromatase inhibition (%) SEM
1
93.14 2.36
60.84 2.61
56.82 5.59
6.59 2.00
4.60 0.40
97.84 0.19
84.59 0.51
95.90 0.60
96.40 0.10
72.05 2.60
70.70 4.25
55.99 1.86
40.01 2.05
33.90 1.68
68.73 3.53
41.45 2.05
99.65 0.06
2
4
5
7
12
13a
1410,12
1510,12
Figure 2. − Lineweaver−Burk plot of inhibitor 13a. Different
concentrations of inhibitor (0, 0.1, 0.3, and 0.6 μM), with 8 as
substrate at 10, 20, and 30 nM, were used. Each point represents the
mean of three independent determinations done in triplicate. The
experiments with the other compounds gave similar plots to the results
shown for 13a.
16
17
21
22
25
26
28
formestane (23)
DISCUSSION AND CONCLUSIONS
■
a
Concentrations of 40 nM [1β-3H]androstenedione, 20 μg of protein
As referred to in the introduction, aromatase establishes two
main hydrogen bonds with the carbonyl functions at C-3 and
C-17 of its natural substrate 8.9 Nevertheless, it is known that
the presence of the carbonyl group at C-3 is not mandatory to
bind steroid molecules to the enzyme aromatase and to get
aromatase inhibition.10 The C-17 carbonyl group, however,
seems to have a more important role in steroidal AIs. In any
case, at least one of the referred carbonyl groups must exist in
order to allow the binding of steroid molecules to the enzyme
aromatase.12 In this work, several steroid molecules were
studied as AIs in which the carbonyl group at C-3 (a hydrogen
bond acceptor) of 8 was replaced by a hydroxyl group (a
hydrogen bond donor) and also in which the two carbonyl
groups (C-3 and C-17) were replaced by two hydroxyl groups.
Looking at compounds 1 and 2 (Scheme 1 and Table 1), we
from human placental microsomes, 2 μM of the compounds, and 15
min of incubation were used. The experiments were done in triplicate.
The results represent the mean SEM of three different experiments.
Compound 23 at 0.5 μM was used as reference.
compounds at 2 μM, relative to an assay carried out in the
absence of the inhibitor. AI 23 at 0.5 μM (99.65 0.06%) was
used as reference. For the steroids 1, 12, 13a, 16, and 17 with
aromatase inhibition higher than 70%, the IC50 with a
concentration of [1β-3H] androstenedione of 100 nM (Table
2) was determined. For steroids 1, 13a, and 16, kinetic studies
to characterize the type of binding to the active site of the
enzyme and the apparent inhibition constant (Ki) were also
performed, using different concentrations of inhibitor and
Table 2. IC50 and Kinetic Studies for the Most Potent Inhibitors
a
b
compds
IC50 (μM)
type of inhibition
competitive
kinetic studies Vm (mol/min/μg prot)
Ki (μM)
real affinity (Km/Ki) (nM)
1
0.183
0.135
0.970
0.225
0.145
1.733
1.150
0.042
0.225 0.025
0.100
1.026 0.026
12
13a
competitive
competitive
competitive
competitive
0.015 0.001
0.200 0.010
0.086
0.050
0.038
9.501
0.636 0.058
0.012 0.002
1410,12
1510,12
16
17
formestane (23)
a
Concentrations of 100 nM [1β-3H]androstenedione, 20 μg protein from human placental microsomes, different concentrations of the compounds,
b
and 15 min of incubation were used. Concentrations of 10, 20, 30, and 40 nM [1β-3H]androstenedione, 20 μg of protein from human placental
microsomes, different concentrations of the compounds, and 5 min of incubation were used. Apparent inhibition constants (Ki) were obtained by
Dixon Plot. Inhibition type was based on analysis of the Lineweaver−Burk plot. The experiments were done in triplicate in three independent
experiments.
3996
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
Figure 4. Best pose of inhibitor 1 within the aromatase binding pocket
shown as a transparent cartoon. The ligand is represented as a yellow
carbon polytube model. The heme and the labeled residues interacting
via hydrogen bonds with 1 are, respectively, displayed by cyan and
magenta carbon polytube models. Hydrogen bonds are depicted as
dashed black lines, and their distances are measured in Å. Nonpolar
hydrogen atoms are omitted for the sake of clarity.
Figure 3. Dixon plot of inhibitor 13a. Different concentrations of
inhibitor (0, 0.1, 0.3, and 0.6 μM), with the substrate 8 at 10, 20, and
30 nM, to determine the apparent inhibition constant (Ki), were used.
Each point represents the mean of three independent experiments
done in triplicate. The experiments with the other compounds gave
similar plots to the results shown for 13a.
having antiaromatase activity.10,12 This planarity could be
achieved introducing a double bond or an epoxide function into
the A-ring, both possessing similar bond geometries. In this
work, we are interested in studying the effect of the double
bond or he epoxide function and its different positioning along
the A-ring, in aromatase inhibitory activity. Considering the A-
ring olefins (12, 14, 16, and 21) and the corresponding
epoxides (13a, 15, 17 e 22) (Scheme 2), it was observed that
4,5-olefin 12 and 1,2-olefin 21 are better AIs than the
corresponding epoxides 13a and 22. Interestingly and on the
contrary, the 3,4-epoxide 15 and the 2,3-epoxide 17 are better
AIs than the corresponding olefins 14 and 16 (Tables 1 and 2).
As published previously by the authors, for a similar 3,4-
epoxide,12 the oxiran oxygen in the 3,4 position may resemble
the C-3 carbonyl oxygen of the aromatase substrate 8, allowing
it to establish hydrogen bonds with aromatase active site
residues. IFD results also indicate 15 as being able to accept
two hydrogen bonds with the aromatase. Notably, they are both
established with the Thr 310 residue (Figure 5), in particular
with the NH backbone (2.4 Å) and OH side chain (2.3 Å).
observed that the substitution of the C-3 carbonyl group of 8
by a C-3 hydroxyl group, maintaining the C-17 carbonyl group,
led to a potent AI when the C-3-OH assumes the 3β
stereochemistry. Compound 1 is in fact a very strong AI, with
an IC50 of 0.18 μM, having also a high affinity to the enzyme
(Ki of 0.1 μM) (Table 2). The C-3α-OH analogue 2 was not as
good an AI as 1. Changing the two carbonyl groups of 8, at C-3
and C-17, by hydroxyl groups, a dramatic decrease in activity
was observed (compounds 4 and 5), particularly if the C-3-OH
assumes the 3α stereochemistry (compound 5) (Table 1). This
decrease was also observed by the authors for other compounds
submitted to the same transformation in C-17.12 The lack of
both C-3 and C-17 carbonyl groups in 4 and 5 can explain the
inability of these steroids to bind conveniently to the enzyme
with the consequent loss of activity. As the C-3-OH
stereochemistry seems to play an interesting role in the
aromatase inhibitory capacity of this kind of compound, a
molecular docking study in the aromatase active site was done
for compounds 1 and 2. IFD simulations revealed the ability to
establish one hydrogen bond between the C-17 carbonyl group
and Met 374 (1.8 Å) by both compounds. Accordingly, the
different aromatase recognition of 1 and 2 can be addressed to
the hydrogen bond network of C-3-OH. In the former case, 1
donates and accepts two hydrogen bonds, respectively, with
Thr 310 (2.1 Å) and Ile 305 (2.4 Å) (Figure 4), and in the
latter case, 2 establishes only one hydrogen bond with Asp 309
(2.1 Å) (data not shown). Considering 100% of the volume of
the aromatase binding site in the crystallographic model
3EQM, the IFD simulations with ligand 1 induced an increase
of the cavity equal to 113%. Concerning compound 7 (Scheme
1), as expected the substitution of the C-17 hydroxyl group by
the C-17 acetoxy group dramatically reduces the aromatase
inhibitory activity (Table 1). The acetoxy group is, in fact, a
bulky group and may cause steric hindrance at the enzyme
active site. This is in accordance with previous studies of the
authors12 and is consistent with the short volume of the
aromatase binding pocket, as recently described.9
Figure 5. Best pose of inhibitor 15 within the aromatase binding
pocket shown as a transparent cartoon. The ligand is represented as a
yellow carbon polytube model. The heme and the labeled residues
interacting via hydrogen bonds with 15 are, respectively, displayed by
cyan and magenta carbon polytube models. Hydrogen bonds are
depicted as dashed black lines, and their distances are measured in Å.
Nonpolar hydrogen atoms are omitted for the sake of clarity.
As reported before, some planarity in the A-ring and in the
A,B-ring junction has been revealed to be important for steroids
3997
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
Interestingly, in the lowest energy pose, inhibitor 15 fits the
receptor core adopting an inverted non-canonical positioning
when compared with aromatase substrate 8, keeping the C-19
methyl group in the opposite side of the heme group (Figure
5). As observed for compound 1, the volume of the binding site
was quite affected by the presence of compound 15 with
respect to 8. The IFD simulations with ligand 15 induced a
remarkable increase of the cavity equal to 123% (considering
100% the volume with the substrate 8). The main reason can
be attributed to the conformational change of Phe 221 and Trp
224 side chains, both located in the same aromatase α helix
(Figure 6).
reinforcing the importance of having planarity in the A,B-ring
junction at C-5.
In Scheme 4, we present the synthesis of thione 28 from
ketone 14. Compound 28 was designed based on the isosterism
concept by performing the chemical substitution of the C-17
carbonyl oxygen by the sulfur atom. It was observed that this
transformation resulted in a loss of aromatase inhibitory
activity, which prompted us to conclude that the sulfur atom
was not as able as the oxygen atom to establish a hydrogen
bond with the Met 374 residue, in the aromatase active site.
In summary, new AIs and new SARs have been found.
Briefly, the β configuration of C-3 hydroxyl derivatives of
aromatase substrate 8 is beneficial for aromatase inhibition; A-
ring olefins are better inhibitors than the corresponding
epoxides, except 2,3-epoxide 17 and 3,4-epoxide 15, probably
due to the possible establishment of additional hydrogen bonds
with aromatase residues; planarity near the C-5 A,B-ring
junction seems to be important for inhibitory activity since 4,5-
olefin 12 is a better inhibitor than 3,4-olefin 14, 14 is a better
inhibitor than 2,3-olefin 16, and 16 is a better inhibitor than
1,2-olefin 21; and bulky substituents at C-4 and substitution of
the C-17 carbonyl oxygen by the sulfur atom seem not to be
beneficial for aromatase inhibition.
Although the new AIs found are not as potent as those
already used as drugs, they showed some promising activities
and can be studied for their effects in bone and in resistant cell
lines in order to find molecules that can overcome the side
effects and resistance shown by the drugs in clinical use.
Figure 6. Aromatase binding site volume comparison between 1 and
15 induced fit models reported, respectively, in magenta and yellow
colors. The ligands, the heme, and the most involved residues in the
conformational change (Phe 221 and Trp 224) are displayed in
polytube models. The rest of the protein is rendered as transparent
cartoons. Nonpolar hydrogen atoms are omitted for the sake of clarity.
EXPERIMENTAL SECTION
■
Chemistry. Melting points (mps) were determined on a Reichert
Thermopan hot block apparatus and were not corrected. IR spectra
were recorded on a Jasco 420FT/IR spectrometer. The 1H NMR
spectra were recorded at 600 MHz, on a Varian Unity 600. The 13C
NMR spectra were recorded at 150 MHz on a Varian Unity 600.
Chemical shifts were recorded in δ values in parts per million (ppm)
downfield from tetramethylsilane as an internal standard. All J-values
are given in Hz. HRMS analyses were made on a QTof instrument
from Applied Biosystems using the electrospray technique. Mass
spectra ESI and LC-MS were obtained with a mass spectrometer QIT-
MS Thermo Finningan, model LCQ Advantage MAX coupled to a
Liquid Chromatograph of High Performance Thermo Finningan. 3β-
Hydroxyandrost-4-en-17-one (1), 3α-hydroxyandrost-4-en-17-one (2),
5α-androst-2-en-17-one (16), and 3-oxo-5α-androst-1-en-17β-yl ac-
etate (18) were purchased from Steraloids, Inc. (Newport RI, USA).
Testosterone (3) was purchased from STEROID, (Cologno, Monzese,
MI). Formestane (23) was purchased from Sigma-Aldrich, Inc. (St.
Louis, USA). Reagents and solvents were used as obtained from the
suppliers without further purification. Yields have not been optimized.
Compounds 8, 14, and 15 were prepared as described.12
Another observation is related to the fact that when the
double bond is closer to the A,B-ring junction in C-5, the
highest aromatase inhibition is reached (Table 1), confirming
that planarity in C-5 A,B-ring junction is very important to
aromatase inhibition. Among the studied olefins and epoxides,
compound 12, which has a C-4 double bond, showed effectively
the best aromatase inhibition. Numazawa et al. had already
described this compound as a very strong AI.24 Among the
epoxides, compound 15 showed the best aromatase inhibitory
activity (IC50 = 0.145 μM).
As pointed out in the Introduction, the C-4 acetoxy
derivative 24 of the first steroidal AI clinically used, 23, also
revealed strong aromatase inhibitory activity.13 On the basis of
this, a new AI (compound 26) based on 23 was designed and
synthesized, by esterification of its C-4 hydroxyl group with the
acetylsalicylic acid moiety (Scheme 3). Unfortunately, the C-4
acetylsalicyloxy derivative 26 showed a pronounced decrease in
aromatase inhibition when compared with its precursor 23
(Table 1). The results obtained so far suggest that the presence
of bulky substituents in C-4 diminishes the aromatase
inhibitory activity, which is consistent with the short volume
of the binding pocket of aromatase.9 The C-4 acetoxy derivative
25 of olefin 14 also showed a dramatic reduction in the
aromatase inhibitory activity, when compared with that of 14
(Table 1). In this case, the introduction of a C-4 acetoxy group
did not maintain aromatase inhibitory activity, as observed in
24 relative to 23 (Scheme 3). Further, we assist reduction in
the activity when we go from 24 to 25, which is probably due to
the displacement of the 4,5 double bond to the 3,4 position,
All compounds possess a purity superior to 95%, except compound
7 (90% purity) and compounds 16 and 17 (92% purity). The purity
was checked by HPLC with a C18-reversed phase column and water/
acetonitrile 40:60 as solvent. The purity of individual compounds was
determined from the peak areas in the chromatogram of the sample
solution.
Androst-4-ene-3β,17β-diol (4). Method A. To a solution of 3
(1.0 g, 3.47 mmol) in dry methanol (30.0 mL), sodium
borohydride (350.1 mg, 9.25 mmol) was added, and the
reaction was stirred at room temperature for 1 h 20 min, after
which a supplement of 100 mg of sodium borohydride was
added to completely transform the starting material (2 h 40
min, TLC). After methanol removal under vacuum, water (100
mL) was added and the product extracted with ethyl acetate
(3× 200 mL). The organic layer was then washed with water
(200 mL), dried over anhydrous MgSO4, filtered, and
3998
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
concentrated to dryness giving a white solid residue (1.04 g)
composed by a mixture of 4 and its 3α-epimer 5. Crystallization
from methanol gave pure compound 4 (214.7 mg, 21%) as
white crystalline plates. Mp(methanol) 149−151 °C. IR (NaCl
plates, CHCl3) υmax cm−1: 3361 (OH), 1656 (CC), 1051
(C−O). 1H (600 MHz, DMSO-d6) δ: 0.64 (3H, s, 18-H3), 0.98
(3H, s, 19-H3), 3.41 (1H, ddd, J17α−16α = 8.5, J17α−16β = 8.5,
J17α‑17βOH = 4.5, 17α-H), 3.90 (1H, m, 3α-H), 4.44 (1H, d,
J17βOH‑17α = 4.5, 17β−OH), 4.54 (1H, d, J3βOH‑3α = 5.4, 3β−OH),
5.17 (1H, bs, 4-H). 13C (150 MHz, DMSO-d6) δ: 11.2 (C-18),
18.5 (C-19), 20.2, 23.0, 29.0, 29.7, 31.5, 32.3, 35.3, 35.5, 36.4,
36.8, 42.4, 50.2, 54.2, 65.9 (C-3), 79.9 (C-17), 125.3 (C-4),
144.2 (C-5). ESI: 289.2 ([M − H]+, 84%).
Method B. To a solution of 3 (1.0 g, 3.47 mmol) in dry
tetrahydrofuran (40.0 mL), lithium tri-t-butoxyaluminium hydride
(1.15 g, 4.51 mmol) was added, and the reaction was heated under
reflux for 2 h 30 min. To completely transform the starting material
(TLC), 874 mg of tri-t-butoxyaluminum hydride was added in several
portions and the reaction stirred at room temperature for a further 8 h
20 min. After methanol removal under vacuum, water (100 mL) was
added and the product extracted with ethyl acetate (3 × 200 mL). The
organic layer was then washed with water (200 mL), dried over
anhydrous MgSO4, filtered, and concentrated to dryness giving a white
solid residue (1.04 g) composed by a mixture of 4 and 5.
Crystallization from methanol gave the pure compound 4 (222.1
mg, 22%) as white crystalline plates.
Androst-4-ene-3α,17β-diol (5). To a solution of 2 (10 mg, 0.035
mmol) in methanol (3 mL), sodium borohydride (4.64 mg, 0.123
mmol) was added, and the reaction mixture was stirred at room
temperature until all of the starting material had been consumed (10
min, TLC). After methanol evaporation and dissolution of the residue
obtained with ethyl acetate (50 mL), the organic layer was washed
with water (3 × 40 mL), dried over anhydrous MgSO4, filtered, and
evaporated affording pure compound 5, in quantitative yield as a white
residue. Mp(dichloromethane/methanol) 202−205 °C. IR (KBr) νmax cm−1:
3328 (OH), 1655 (CC), 1054 (C−O). 1H NMR (600 MHz,
DMSO-d6) δ: 0.66 (3H, s, 18-H3), 0.92 (3H, s, 19-H3), 3.41−3.45
(1H, m, 17α-H), 3.85 (1H, bs, 3-H), 4.29 (1H, d, J = 4.8, 3α−OH or
17β−OH), 4.38 (1H,d, J = 4.8, 3α−OH or 17β−OH), 5.30 (1H, s, J =
4.2, 4-H). 13C NMR (150 MHz, DMSO-d6) δ: 11.1 (C-18), 18.0 (C-
19), 20.7, 23.0, 27.7, 29.7, 31.3, 31.6, 32.1, 35.4, 36.5, 36.9, 40.0, 42.4,
50.3, 53.2, 62.3 (C-3), 79.9 (C-17), 122.2 (C-4). ESI: 289.5 ([M −
H]+, 100%).
General Procedure to Obtain 4α,5α- and 4β,5β-Epoxyandro-
stan-17β-ol (11a and 11b), 4α,5α-Epoxyandrostan-17-one
(13a), 2α,3α-Epoxy-5α-androstan-17-one (17), and 1α,2α-
Epoxyandrostan-17-one (22). To a solution of olefin (10, 12, 16,
or 21) in dichloromethane, a solution of performic acid, generated in
situ by addition of 98−100% HCOOH to 35% H2O2, was added. The
reaction mixture was stirred at room temperature until complete
transformation of the starting material. Dichloromethane (100 mL)
was added, and the organic layer was washed successively with 10%
aqueous NaHCO3 (2 × 100 mL) and water (4 × 100 mL), dried over
anhydrous MgSO4, filtered, and concentrated to dryness. The obtained
residue was purified by silica gel 60 column chromatography (n-
hexane/ethyl acetate).
4α,5α- and 4β,5β-Epoxyandrostan-17β-ol (11a and 11b).
Olefin 10 (82.7 mg, 0.30 mmol); dichloromethane (4.0 mL); 98−
100% HCOOH (0.04 mL); and 35% H2O2 (0.11 mL) were used; total
reaction time, 7 h (TLC). Before column chromatography, 83.6 mg of
a white solid residue was obtained. Purification by column
chromatography afforded 39.7 mg of an inseparable epimeric mixture
(60:40, by NMR) of 11a and 11b. 4α,5α-Epoxyandrostan-17β-ol
1
(11a): H NMR (600 MHz, DMSO-d6) δ, 0.65 (3H, s, 18-H3), 1.02
(3H, s, 19-H3), 3.42−3.46 (1H, m, 17α-H), 2.86 (1H, d, J4β‑3α = 4.3,
4β-H), 4.42 (1H, d, J17βOH‑17α = 4.8, 17β−OH). 4β,5β-Epoxyandro-
stan-17β-ol (11b): 1H NMR (600 MHz, DMSO-d6) δ, 0.64 (3H, s, 18-
H3), 0.93 (3H, s, 19-H3), 3.42−3.46 (1H, m, 17α-H), 2.86 (1H, d,
J4α‑3α = 4.6, 4α-H), 4.44 (1H, d, J17βOH‑17α = 4.9, 17β−OH).
4α,5α-Epoxyandrostan-17-one (13a). Olefin 12 (140.7 mg,
0.52 mmol); dichloromethane (4.0 mL); 98−100% HCOOH (0.07
mL); and 35% H2O2 (0.19 mL) were used; total reaction time, 6 h 30
min (TLC). Before column chromatography 141.9 mg of a white solid
residue was obtained. Purification by column chromatography afforded
76.6 mg of the 4,5-epoxyandrostan-17-one in a mixture (66:34 by
NMR) of 4α,5α- and 4β,5β-epimers 13a and 13b. Further purification
by column chromatography with chloroform allowed us to isolate 11.4
mg (7.6%) of the pure 4α,5α-epimer 13a. 4α,5α-Epoxyandrostan-17-
one (13a): mp(chloroform) 142−145 °C. IR (NaCl plates, CHCl3) νmax
1
cm−1: 1734 (CO), 1216 (C−O−C). H NMR (600 MHz, CDCl3)
δ: 0.88 (3H, s, 18-H3), 1.03 (3H, s, 19-H3), 2.46 (1H, ddd, J16β‑16α
=
19.0, J16β‑15β = 9.0, J16β‑15β = 1.0, 16β-H), 2.90 (1H, d, J4β‑3α = 4.6, 4β-
H). 13C NMR (150 MHz, CDCl3) δ: 13.7 (C-18), 15.3 (C-19), 19.2,
20.5, 21.8, 23.6, 29.3, 29.6, 31.2, 31.4, 34.7, 35.8, 36.4, 46.7, 47.7, 51.2,
61.3, 65.3 (C-5), 220.9 (C-17). ESI: 287.1 ([M − H]+, 100%). 4β,5β-
1
Epoxy-androstan-17-one (from the mixture of 13a + 13b): H NMR
(600 MHz, CDCl3) δ, 0.88 (3H, s, 18-H3), 1.08 (3H, s, 19-H3), 2.43
(1H, ddd, J16β‑16α = 19.0, J16β‑15β = 9.0, J16β‑15β = 1.0, 16β-H), 2.94 (1H,
d, J4α‑3β =4.1, 4α-H).
3β-Hydroxyandrost-4-en-17β-yl Acetate (7). To a solution of 6
(676.5 mg, 2.04 mmol) in dry methanol (20 mL), sodium borohydride
(127.5 mg, 3.37 mmol) was added, and the reaction was stirred at
room temperature until complete transformation of the starting
material (45 min, TLC). After removal of methanol under vacuum,
water (200 mL) was added and the product extracted with ethyl
acetate (3 × 200 mL). The organic layer was washed with water (200
mL), dried over anhydrous MgSO4, filtered, and concentrated to
dryness giving a white solid residue (706.4 mg). This residue was
purified by a silica gel 60 column chromatography (petroleum ether
60−80 °C/ethyl acetate) affording 430.0 mg of a 90:10 3β/3α-
epimeric mixture of 3-hydroxyandrost-4-en-17β-yl acetate (NMR and
HPLC control). An attempt to isolate 3β-epimer 7 by crystallization
just enriched the mixture in the 3α-epimer. 3β-Hydroxyandrost-4-en-
17β-yl acetate (7): 1H NMR (600 MHz, DMSO-d6) δ, 0.76 (3H, s, 18-
H3), 0.99 (3H, s, 19-H3), 1.98 (3H, s, CH3COO), 3.90 (1H, m, 3α-H),
4.49 (1H, dd, J17α‑16α = 9.0, J17αH‑16βH = 8.0, 17α-H), 4.54 (1H, d,
J3βOH‑3α = 5.5, 3β−OH), 5.19 (1H, bs, 4-H). 13C NMR (150 MHz,
DMSO-d6) δ: 11.8 (C-18), 18.4 (C-19), 20.0, 20.8, 22.9, 27.0, 28.9,
31.4, 32.2, 35.1, 35.2, 36.2, 36.7, 41.9, 49.7, 53.8, 65.8 (C-3), 81.8 (C-
17), 125.5 (C-4), 143.9 (C5), 170.2 (OCO). ESI: 331.0 ([M − H]+,
2α,3α-Epoxy-5α-androstan-17-one (17). Method A. Olefin
16 (100 mg, 0.37 mmol); dichloromethane (2.0 mL); 98−
100% HCOOH (0.1 mL); and 35% H2O2 (0.3 mL) were used;
total reaction time, 6 h (TLC). Before column chromatography,
103.5 mg of a white solid residue was obtained. Purification by
column chromatography (chloroform) afforded compound 17
(6.7 mg, 6%) in 92% purity (NMR and HPLC analysis). IR υmax
(NaCl plates CHCl3) cm−1: 1738(CO), 1013 (C−O−C). 1H
NMR (600 MHz, CDCl3) δ: 0.78 (3H, s, 19-H3), 0.84 (3H, s,
18-H3), 2.05 (1H, ddd, J16α‑16β = 19.0, J16α‑15β = 9.0, J16α‑15α = 9.0,
16α-H), 2.42 (1H, ddd, J16β‑16α = 19.0, J16β‑15β = 9.0, J16β‑15α = 1.0,
16β-H), 3.11 (1H, m, J2β‑1α = 6.0, J2β‑3β = 3.95, 2β-H) and 3.16
(1H, m, J3β‑4α =6.0, J3β‑2β = 3.95, J3β‑4β = 1.69, 3β-H). 13C NMR
(150 MHz, CDCl3) δ: 15.6 (C-19), 16.3 (C-18), 22.8, 24.4,
30.7, 31.6, 33.1, 34.1, 36.4, 37.8, 38.5, 38.9, 40.9, 50.2, 53.5,
53.9, 55.0 (C-2), 56.4 (C-3) and 223.9 (C-17). ESI: 287.0 ([M
− H]+, 100%).
Method B. To a stirred 9% aqueous peracetic acid solution (1.0
mL) at 10 °C, trihidrated sodium acetate (79.6 mg) and olefin 16
(200.2 mg, 0.73 mmol) in chloroform (2 mL) were added. The
reaction was then stirred at room temperature until complete
transformation of the starting material (7 h 30 min, TLC).
Dichloromethane (150 mL) was added, and the organic layer was
washed with 10% aqueous NaHCO3 (100 mL), water (4 × 100 mL),
1
100%). 3α-Hydroxyandrost-4-en-17β-yl acetate: H NMR (600 MHz,
DMSO-d6) δ, 0.77 (1H, s, 18-H3), 0.92 (1H, s, 19-H3), 1.98 (3H, s,
CH3COO), 3.85 (1H, m, 3β-H), 4.36 (1H, d, J3αOH‑3β = 4.4, 3α−OH),
4.50 (1H, dd, J17α‑16α = 9.0, J17α‑16β = 8.0, 17α-H), 5.31 (1H, d, J4−3β
=
4.5, 4-H).
3999
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
dried over anhydrous MgSO4, and concentrated to dryness giving
162.9 mg of a white residue. Purification of this residue by the usual
procedure gave compound 17, also in 92% purity.
gel 60 column chromatography (petroleum ether 40−60 °C/ethyl
acetate) allowing us to separate 610.6 mg of compound 26 in a
mixture with compound 24 (60:40 respectively, NMR). A portion of
this mixture (127.7 mg) was further purified by further silica gel 60
column chromatography (n-hexane/diethyl ether) allowing us to
isolate pure compound 26 (51.2 mg) as a white crystalline residue.
1α,2α-Epoxyandrostan-17-one (22). Olefin 21 (46.4 mg, 0.17
mmol); dichloromethane (3 mL); 98−100% HCOOH (0.03 mL); and
35% H2O2 (0.07 mL) were used; total reaction time, 9 h 30 min
(TLC). Before column chromatography, 38.6 mg of an oily residue
was obtained. Purification by column chromatography afforded 7.4 mg
(15%) of the pure compound 22. Mp(chloroform) 119−122 °C. IR (NaCl
Mp(diethyl
183−185 °C. IR (KBr disk) υmax cm−1: 3453
ether/n‑hexane)
(CHAr), 1769 (CO ester), 1739 (CO), 1687 (CC), 1606 (C
1
CAr), 1195 (C−O). H NMR (600 MHz, CDCl3) δ: 0.92 (3H, s, 18-
1
plates, CHCl3) υmax cm−1: 1738 (CO), 1049 (C−O). H NMR
H3), 1.32 (3H, s, 19-H3), 2.28 (3H, s, CH3COO), 7.12 (1H, d, J4−3 =
(600 MHz, CDCl3) δ: 0.87 (3H, s, 18-H3), 0,92 (3H, s, 19-H3), 2.07
(1H,ddd, J16α‑16β = 19.0, J16α‑15α = 9.0, J16α‑15β = 9.0, 16α-H), 2.42 (1H,
ddd, J16β‑16α = 19.0, J16β‑15β = 9.0, J16β‑15α = 1.0, 16β-H), 2.99 (1H, d,
J1β‑2β = 4.0, 1β-H), 3.12 (1H, dd, J2β‑1β = 4.0, J2β‑3α = 3.0, 2β-H). 13C
NMR (150 MHz, CDCl3) δ: 11.4 (C-18), 13.8 (C-19), 20.4, 21.7,
22.7, 23.4, 27.6, 30.4, 31.3, 34.9, 35.8, 36.6, 37.3, 47.7, 49.0, 51.3, 52.9
(C-1), 59.1 (C-2), 221.1 (CO). ESI: 286.9 ([M − H]+, 100%).
4-Acetoxyandrost-4-en-3,17-dione (24). To a solution of 23
(750.6 mg, 2.48 mmol) in dry pyridine (12.5 mL), at 0 °C, acetyl
chloride (0.27 mL, 3.80 mmol) was added dropwise. The reaction was
stirred for 15 min at 0 °C, and then the temperature was raised to the
ambient. Three subsequent additions of acetyl choride (3 × 0.1 mL)
were made allowing the reaction to be complete (total reaction time:
21 h 50 min, TLC). The solvent was then evaporated under vacuum,
and the obtained residue was crystallized with ethyl acetate/n-hexane
after activated charcoal decoloration giving the pure compound 24 as
white crystals (616.1 mg, 72%). Mp(chloroform) 169−171 °C. IR (NaCl
plates, CHCl3) νmax cm−1: 3018 (CH), 1739 (CO), 1680 (C
7.8, 4-HAr), 7.32 (1H, dd, J3−4 = 7.8, J3−2 = 7.8, 3-HAr), 7.57 (1H, dd,
J2−3 = 7.8, J2−1 = 9.2, 2-HAr), 8.13 (1H, d, J1−2 = 9.2, 1-HAr). 13C NMR
(150 MHz, CDCl3) δ: 13.7 (C-18), 17.8 (C-19), 20.4, 20.9, 21.7, 24.0,
29.9, 31.5, 33.4, 34.8, 34.9, 35.7, 39.4, 47.4, 50.9, 54.0, 122.8, 123.8
(CAr-4), 125.9 (CAr-3), 132.2 (CAr-2), 134.0 (CAr-1), 139.2, 151.1,
155.3, 162.1, 169.3, 189.8 (C-3), 219.5 (C-17). ESI: 463.7 ([M − H]+,
100%).
4-(o-Acetylsalicyloxy)-5α-androst-3-en-17-one (27). A solu-
tion of a crude containing compound 26 as the main product (272.9
mg) in glacial acetic acid (25 mL) was sonicated with an ultrasound
probe in the presence of excess of dust zinc (<10 μm) (4.73 g, 17.43
mmol) until the transformation of all of the starting material (20 min,
TLC). Zinc was filtered and washed with glacial acetic acid, and then
the filtrate was concentrated under vacuum. To the oily residue, water
(200 mL) was added, and the product was extracted with
dichloromethane (3 × 100 mL). The organic phase was sequentially
washed with 10% aqueous NaHCO3 (2 × 150 mL) and water (3 ×
150 mL), dried over anhydrous Na2SO4, filtered, and concentrated to
dryness giving an oily residue (242.9 mg). This residue was purified by
a silica gel 60 column chromatography (petroleum ether 40−60 °C/
ethyl acetate) allowing us to isolate 101.1 mg of 4-(o-acetylsalicyloxy)-
5α-androst-3-en-17-one 27a in an inseparable mixture with its 5β-
epimer 27b (70:30, respectively, by NMR). 4-(o-Acetylsalicyloxy)-5α-
androst-3-en-17-one (27a): 1H NMR (600 MHz, CDCl3) δ, 0.87 (3H,
s, 18-H3), 0.93 (3H, s, 19-H3), 2.33 (3H, s, CH3COO), 5.36 (1H, dd,
J3−2β = 6.6, J3−2α = 3.3, 3-H), 7.12 (1H, d, J4−3 = 8.7, 4-HAr), 7.33 (1H,
dd, J3−4 = 8.7, J3−2 = 8.7, 3-HAr), 7.58 (1H, dd, J2−3 = 8.7, J2−1 = 9.6, 2-
HAr), 8.07 (1H, d, J1−2 = 9.6, 1-HAr). 4-(o-Acetylsalicyloxy)-5β-
androst-3-en-17-one (27b): 1H NMR (600 MHz, CDCl3) δ, 0.87 (3H,
s, 18-H3), 1.05 (3H, s, 19-H3), 2.34 (3H, s, CH3COO), 5.51 (1H, dd,
J3−2β = 6.9, J3−2α = 3.6, 3-H), 7.12 (1H, d, J4−3 = 8.7, 4-HAr), 7.33 (1H,
dd, J3−4 = 8.7, J3−2 = 8.7, 3-HAr), 7.58 (1H, dd, J2−3 = 8.7, J2−1 = 9.6, 2-
HAr), 8.05 (1H, d, J1−2 = 9.6, 1-HAr).
5α-Androst-3-ene-17-thione (28). To a solution of olefin 14
(420.7 mg, 1.46 mmol) in dry toluene (30 mL), Lawesson’s reagent
(624.9 mg, 1.54 mmol) was added, and the reaction mixture was
heated under reflux for 7 h, in an inert atmosphere. The remaining
Lawesson’s reagent was removed through an aluminum oxide neutral
column leading to an orange residue, which was further purified by
silica gel 60 column chromatography (petroleum ether 40−60 °C)
affording pure compound 28 (238.5 mg, 54%), as a light orange solid.
Mp(petroleum ether 40−60 °C) 95 °C. IR (NaCl plates, CHCl3) υmax cm−1:
3015 (CH), 1650 (CC), absence of CO (peak around 1715 in
compound 14). 1H NMR (600 MHz, CDCl3) δ: 0.79 (3H, s, 19-H3),
0.88 (3H, s, 18-H3), 2.60 (1H, ddd, J16α‑16β = 22.0, J16α‑15β = 9.0, J16α‑15α
1
C), 1059 (C−O). H NMR (600 MHz, CDCl3) δ: 0.91 (3H, s, 18-
H3), 1.26 (3H, s, 19-H3), 2.23 (3H, s, CH3COO). 13C NMR (150
MHz, CDCl3) δ: 13.7 (C-18), 17.6 (C-19), 20.2, 20.3, 21.7, 23.9, 29.7,
31.2, 33.3, 34.6, 34.7, 35.7, 39.1, 47.4, 50.7, 53.8, 139.2, 154.9, 168.6,
190.4 (C-3), 220.2 (C-17).
4-Acetoxy-5α-androst-3-en-17-one (25). To a solution of 24
(90.3 mg, 0.26 mmol) in glacial acetic acid (7.5 mL), zinc dust (500.0
mg, 7.65 mmol) was added. The reaction was sonicated in an
ultrasound bath at room temperature for 25 min, after which an excess
of dust zinc (500.0 mg, 7.65 mmol) was added, and the reaction
proceeded until all of the starting material had been consumed (2 h,
TLC). Zinc was then filtered and washed with diethyl ether (50 mL),
and the filtrate was concentrated under vacuum. To the oily residue
obtained, water (100 mL) was added, and the product was extracted
with dichloromethane (3 × 100 mL). The organic phase was
sequentially washed with 10% aqueous NaHCO3 (2 × 100 mL) and
water (3 × 100 mL), dried over anhydrous MgSO4, filtered, and
concentrated to dryness giving an oily residue (105.9 mg) composed
by a mixture of 4-acetoxy-5α-androst-3-en-17-one 25 and its 5β-
epimer. Further crystallization with petroleum ether afforded the pure
compound 25, as white crystals. Mp(petroleum
116−119 °C. IR
ether)
(NaCl plates, CHCl3) νmax cm−1: 3018 (CH), 1737 (CO), 1681
1
(CC), 1158 (C−O). H NMR (600 MHz, CDCl3) δ: 0.86 (3H, s,
18-H3), 0.88 (3H, s, 19-H3), 2.11 (3H, s, CH3COO), 5.24 (1H, dd,
J3−2β = 6.6, J3−2α = 3.3, 3-H). 13C NMR (150 MHz, CDCl3) δ: 12.4
(C-18), 13.9 (C-19), 20.5, 20.6, 20.7, 21.4, 21.7, 30.1, 31.5, 33.4, 34.7,
35.8, 36.5, 47.0, 47.8, 51.3, 53.0, 112.5 (C-3), 148.8 (C-4), 169.7,
221.0 (C-17). HRMS: m/z [M + Na]+ calcd for C21H30O3, 353.2087;
found, 353.2080.
= 9.0, 16α-H), 2.93 (1H, ddd, J16β‑16α = 22.0, J16β‑15β = 9.0, J16β‑15α
=
1.0, 16β-H), 5.27 (1 H, ddd, J4−3 = 9.5, J4−5α = 4.0, J4−2α = 2.0, 4-H),
5.54 (1 H, ddd, J3−4 = 9.5, J3−2β = 6.0, J3−2α = 3.0, 3-H). 13C NMR
(150 MHz, CDCl3) δ: 14.5 (C-19), 20.5 (C-18), 23.7, 26.1, 27.1, 29.9,
34.3, 36.7, 37.7, 38.3, 38.6, 48.5, 51.8, 55.7 (2 carbons), 62.1, 128.2 (C-
4), 133.7 (C-3), 274.1 (CS). ESI: 287.2 ([M − H]+, 100%).
Biochemistry. Preparation of Placental Microsomes. Placental
microsomes were obtained as described by Yoshida and
Osawa,25 with some modifications as reported previously by
our group.10 Human placentas, obtained after delivery from a
local hospital, were placed in cold 67 mM potassium phosphate
buffer (pH 7.4) containing 1% KCl. The cotyledon tissue was
separated and homogenized in a Polytron homogenizer with 67
mM potassium phosphate buffer (pH 7.4) containing 0.25 M
sucrose and 0.5 mM dithiothreitol (DTT, 1:1, w/v). The
4-(o-Acetylsalicyloxy)androst-4-en-3,17-dione (26). To a
solution of 23 (500.2 mg, 1.65 mmol) in dry pyridine (6.5 mL) at 0
°C, o-acetylsalicyloyl chloride (492.0 mg, 2.48 mmol) was added. The
reaction mixture was stirred at room temperature for 22 h 30 min, and
after that, an excess of o-acetylsalicyloyl chloride (247.4 mg, 1.25
mmol) was added. The reaction proceeded until complete trans-
formation of the starting material (24 h 40 min, TLC). After
evaporation of the solvent under vacuum, the residue was dissolved in
dichloromethane (100 mL), and the organic layer was sequentially
washed with 0.25 N aqueous HCl (4 × 100 mL), 10% aqueous
NaHCO3 (2 × 100 mL), and water (2 × 100 mL), dried over
anhydrous MgSO4, filtered, and concentrated to dryness giving a
yellow oily residue (865.8 mg). This residue was then purified by silica
4000
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
homogenate was centrifuged at 5000g for 30 min, and the
supernatant was centrifuged at 20,000g for 30 min, and
afterward at 54,000g for 45 min to yield the microsomal pellet.
The microsomes were washed and ressuspended in 67 mM
potassium phosphate buffer (pH 7.4) containing 0.25 M
sucrose, 20% glycerol, and 0.5 mM DTT and stored at −80 °C.
All procedures were carried out at 0−5 °C. Protein content was
estimated by the Bio-Rad protein assay (Bio-Rad Laboratories,
Munich, Germany) using BSA as a standard.
ASSOCIATED CONTENT
* Supporting Information
Mass spectrometry (MS) analysis for compounds 1, 2, 14, 15,
and 16; and synthetic procedures as well as spectroscopic and
physical data for compounds 6, 9, 10, 12, 19, 20, and 21. This
material is available free of charge via the Internet at http://
pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*Tel: +351 239 488 400. Fax: +351 239 488 503. E-mail:
etavares@ff.uc.pt (E.J.T.d.S.); froleira@ff.uc.pt (F.M.F.R.).
Aromatase Assay Procedure. Aromatase activity was measured
according to Thompson and Siiteri,23 and Heidrich et al.,26 by
measuring the tritiated H2O released from [1β-3H] androstenedione
(PerkinElmer Life Sciences, Boston, MA, USA), during the
aromatization process. All tested compounds were dissolved in
DMSO and diluted in 67 mM potassium phosphate buffer (pH 7.4).
Briefly, for the screening assay, (1 mL) 20 μg of protein of the
microsomes, 40 nM of [1β-3H] androstenedione (1 μCi), and 2 μM of
each of the AIs were used for the reaction mixture. The aromatase-
catalyzed reaction was initiated by the addition of NADPH (150 μM),
and incubations were performed in a shaking water bath at 37 °C for
15 min. For the IC50 determination, 100 nM (1 μCi) of [1β-3H]
androstenedione and different concentrations of the inhibitors were
used. For the kinetic studies and to minimize the time-dependent loss
of the initial aromatization rate, 5 min of incubation time was used,
and assays were performed with different concentrations of [1β-3H]
androstenedione (10−40 nM). All of the aromatase reactions were
terminated by the addition of 250 μL of 20% trichloroacetic acid. The
mixture was transferred to microcentrifuge tubes containing a
charcoal−dextran pellet, vortexed, and incubated for 1 h. After
centrifugation at 14,000g for 10 min, the supernatants were transferred
to new charcoal−dextran pellets, incubated for 10 min, and
subsequently pelleted by a new centrifugation cycle. The supernatant
containing the tritiated water product was mixed with a liquid
scintillation cocktail (ICN Radiochemicals, Irvine, CA, USA) and
counted in a liquid scintillation counter (LS-6500, Beckman Coulter,
Inc., Fullerton, CA). All experiments were carried out in triplicate.
Docking Experiments. The X-ray coordinates of human
aromatase in complex with its substrate 8 were extracted from the
PDB (code 3EQM).9 Initially, both ligand and enzyme were
pretreated. For ligand preparation, the 3D structures of all the studied
compounds were generated with the Maestro Build Panel. The target
structure was prepared through the Protein Preparation Wizard of the
graphical user interface Maestro 9.127 and the OPLS-2005 force field.28
All water molecules were removed, hydrogen atoms were added, and,
finally, energy minimization was performed until the rmsd of all heavy
atoms was within 0.3 Å of the original PDB model.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank FCT (Fundaca
̧
o para a Cien
̃
̂
cia e Tecnologia) for
financial support and for the Ph.D. grants for C.V. and C.A.
(SFRH/BD/44872/2008 and SFRH/BD/48190/2008, respec-
tively).
ABBREVIATIONS USED
■
AI/AIs, aromatase inhibitors; DRY, GRID hydrophobic probe;
3EQM, PDB code of the aromatase androstenedione crystallo-
graphic complex; FLAP, fingerprints for ligands and proteins
(molecular software by Molecular Discovery Inc.); Glide,
docking software belonging to the Scrodinger suite http://
www.schrodinger.com/products/14/5/; GlideSP, Glide single
precision version; GRID, software created by Professor Peter
Goodford; IFD, induced fit docking; OPLS-AA, optimized
potentials for liquid simulations, version all atoms (force field of
the Glide software); OPLS-2005, optimized potentials for
liquid simulations, version 2005 (force field of the Glide
software); PDB, Protein Data Bank; PYMOL, Python-
enhanced molecular graphics tool; SEM, standard error of the
mean
REFERENCES
■
(1) Winer, E. P.; Hudis, C.; Burstein, H. J.; Wolff, A. C.; Pritchard, K.
I.; Ingle, J. N.; et al. American Society of Clinical Oncology
Technology Assessment on the Use of Aromatase Inhibitors as
Adjuvant Therapy for Postmenopausal Women with Hormone
Receptor-Positive Breast Cancer: Status Report 2004. J. Clin. Oncol.
2005, 23, 619−29.
Docking studies were carried out using IFD.14 An initial Glide SP
docking of each ligand was performed by using a softened potential, a
van der Waals radius scaling factor of 0.50 for receptor/ligand atoms,
and a number of 20 poses per ligand to be energy minimized with the
OPLS-AA force field.29 The poses were saved for each ligand and
submitted to the subsequent Prime side chain orientation prediction of
residues with a distance cutoff of 5 Å around each ligand. After the
Prime minimization of the residues and the ligand for each pose, a
Glide SP redocking of each protein−ligand complex structure within
30 kcal/mol above the global minimum was performed. Finally, each
output pose was estimated by the binding energy (IFD score) and
visually examined. The pretreated 3EQM model was also submitted to
a redocking test, evaluating the computational protocol in terms of
rmsd deviation onto the ligand coordinates. The rmsd value was equal
to 0.18 Å. The 3D Figures 4−6 were obtained with PyMOL graphics
and modeling package, version 0.98.30 The volume analysis of the
binding sites has been carried out by means of the FLAPsite utility,31
as implemented in the FLAP program.32 The GRID probe was set to
DRY, and the other parameters were set as default.
(2) Perez, E. A. Appraising Adjuvant Aromatase Inhibitor Therapy.
Oncologist 2006, 11, 1058−69.
(3) Jordan, V. C.; Brodie, A. M. H. Development and Evolution of
Therapies Targeted to the Estrogen Receptor for the Treatment and
Prevention of Breast Cancer. Steroids 2007, 72, 7−25.
(4) O’Reilly, J. M.; Brueggemeier, R. W. Development of Steroidal
and Nonsteroidal Inhibitors of Aromatase for the Treatment of
Hormone-Dependent Breast Cancer. Curr. Med. Chem. 1996, 3, 11−
22.
(5) Banting, L.; Nicholls, P. J.; Shaw, M. A.; Smith, H. J. Recent
Developments in Aromatase Inhibition as a Potential Treatment for
Estrogen-Dependent Breast Cancer. Prog. Med. Chem. 1989, 26, 253−
298.
(6) Banting, L. Inhibition of Aromatase. Prog. Med. Chem. 1996, 33,
147−184.
(7) Miller, W. R.; Anderson, T. J.; Jack, W. J. L. Relationship between
Tumor Aromatase Activity, Tumor Characteristics and Response to
Therapy. J. Steroid Biochem. Mol. Biol. 1990, 37, 1055−1059.
(8) Dutta, U.; Pant, K. Aromatase Inhibitors: Past, Present and
Future in Breast Cancer Therapy. Med. Oncol. 2008, 25, 113−24.
4001
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002

Journal of Medicinal Chemistry
Article
(9) Ghosh, D.; Griswold, J.; Erman, M.; Pangborn, W. Structural
Basis for Androgen Specificity and Oestrogen Synthesis in Human
Aromatase. Nature 2009, 457, 219−223.
(10) Cepa, M. D. S.; Tavares-da-Silva, E. J.; Correia-da-Silva, G.;
Roleira, F. M. F.; Teixeira, N. A. A. Structure-Activity Relationships of
New A,D-Ring Modified Steroids as Aromatase Inhibitors: Design,
Synthesis, and Biological Activity Evaluation. J. Med. Chem. 2005, 48,
6379−6385.
(11) Sherwin, P. F.; McMullan, P. C.; Covey, D. F. Effects of Steroid
D-Ring Modification on Suicide Inactivation and Competitive
Inhibition of Aromatase by Analogues of Androsta-1,4-diene-3,17-
dione. J. Med. Chem. 1989, 32, 651−658.
(12) Cepa, M. M. D. S.; Tavares da Silva, E. J.; Correia-da-Silva, G.;
Roleira, F. M. F.; Teixeira, N. A. A. Synthesis and Biochemical Studies
of 17-Substituted Androst-3-enes and 3,4-Epoxyandrostanes as
Aromatase Inhibitors. Steroids 2008, 73, 1409−1415.
(28) Banks, J. L.; Beard, H. S.; Cao, Y.; Cho, A. E.; Damm, W.; Farid,
R.; Felts, A. K.; Halgren, T. A.; Mainz, D. T.; Maple, J. R.; Murphy, R.;
Philipp, D. M.; Repasky, M. P.; Zhang, L. Y.; Berne, B. J.; Friesner, R.
A.; Gallicchio, E.; Levy, R. M. Integrated Modeling Program, Applied
Chemical Theory (IMPACT). J. Comput. Chem. 2005, 26, 1752−
1780.
(29) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. Development
and Testing of the OPLS All-Atom Force Field on Conformational
Energetics and Properties of Organic Liquids. J. Am. Chem. Soc. 1996,
118, 11225−11236.
(30) Delano W. L. The PyMOL Molecular Graphics System, 2002.
http://www.pymol.org
(31) Henrich, S.; Salo-Ahen, O. M. H.; Huang, B.; Rippmann, F.;
Cruciani, G.; Wade, R. C. Computational Approaches to Identifying
and Characterizing Protein Binding Sites for Ligand Design. J. Mol.
Recognit. 2010, 23, 209−219.
(32) Baroni, M.; Cruciani, G.; Sciabola, S.; Perruccio, F.; Mason, J. S.
A Common Reference Framework for Analyzing/Comparing Proteins
and Ligands. Fingerprints for Ligands and Proteins (FLAP): Theory
and Application. J. Chem. Inf. Model. 2007, 47, 279−294.
(13) Marsh, D. A.; Brodie, H. J.; Garrett, W.; Tsai-Morris, C.; Brodie,
A. M. H. Aromatase Inhibitors. Synthesis and Biological Activity of
Androstenedione Derivatives. J. Med. Chem. 1985, 28, 788−795.
(14) Sherman, W.; Day, T.; Jacobson, M. P.; Friesner, R. A.; Farid, R.
Novel Procedure for Modeling Ligand/Receptor Induced Fit Effects. J.
Med. Chem. 2006, 49, 534−554.
(15) Zhao, Y.; White, M. A.; Muralidhara, B. K.; Sun, L.; Halpert, J.
R.; Stout, C. D. Structure of Microsomal Cytochrome P450 2B4
Complexed with the Antifungal Drug Bifonazole: Insight into P450
Conformational Plasticity and Membrane Interaction. J. Biol. Chem.
2006, 281, 5973−5981.
(16) Parish, E. J.; Scott, A. D. Selective Oxidation of Steroidal Allylic
Alcohols Using 3,5-Dimethylpyrazole and Pyridinium Chlorochro-
mate. J. Org. Chem. 1983, 48, 4766−4768.
(17) Varela, C. L.; Roleira, F. M. F.; Carvalho, R. A.; Teixeira, N. A.
A.; Tavares-da-Silva, E. J. Insights and Mechanisms towards the
Reactivity of Steroid A-Ring Δ1- and Δ4−3-Ketoenones and
Corresponding Allylic Alcohols, unpublished results.
(18) Oh, S. S.; Robinson, C. H. Synthesis of and Chemical Model
Reaction Studies with 3-Deoxyandrogens: Evidence Supporting a 2,3-
Enolization Hypothesis in Human Placental Aromatase Catalysis. J.
Chem. Soc., Perkin Trans. 1 1994, 2237−2243.
(19) Numazawa, M.; Mutsumi, A.; Hoshi, K.; Koike, R. 19-Hydroxy-
4-androsten-17-one: Potential Competitive Inhibitor of Estrogen
Biosynthesis. Biochem. Biophys. Res. Commun. 1989, 160, 1009−1014.
(20) Campbell, M. M.; Craig, R. C.; Boyd, A. C.; Gilbert, I. M.;
Logan, R. T.; Redpath, J.; Roy, R. G.; Savage, D. S.; Sleigh, T. Amino-
steroids. Part 6. Stereospecific Syntheses of Eight, Isomeric, Steroidal
Vicinal 2,3-Amino-alcohols. J. Chem. Soc., Perkin Trans. 1 1979, 2235−
2247.
(21) Bowers, A.; Cross, A. D.; Edwards, J. A.; Carpio, H.; Calzada, M.
C.; Denot, E. Steroids. CCV. Ring A Modified Hormone Analogs. Part
I. Some Ring A Olefins. J. Med. Chem. 1963, 6, 156−161.
(22) Mohamed, N. R.; Elmegeed, G. A.; Younis, M. Studies on
Organophosphorus Compounds VII: Transformation of Steroidal
Ketones with Lawesson’s Reagent into Thioxo and Heterofused
Steroids. Results of Antimicrobial and Antifungal Activity. Phosphorus,
Sulfur Silicon Relat. Elem. 2003, 178, 2003−2017.
(23) Thompson, E. A.; Siiteri, P. K. Utilization of Oxygen and
Reduced Nicotinamide Adenine Dinucleotide Phosphate by Human
Placental Microsomes during Aromatization of Androstenedione. Biol.
Chem. 1974, 249, 53−64.
(24) Numazawa, M.; Kamiyama, T.; Tachibana, M.; Oshibe, M.
Synthesis and Structure-Activity Relationships of 6-Substituted
Androst-4-ene Analogs as Aromatase Inhibitors. J. Med. Chem. 1996,
39, 2245−2252.
(25) Yoshida, N.; Osawa, Y. Purification of Human Placental
Aromatase Cytocrome P-450 with Monoclonal Antibody and Its
Characterization. Biochemistry 1991, 30, 3003−3010.
(26) Heidrich, D.; Steckelbroeck, S.; Klingmuller, D. Inhibiton of
Human Cytocrome P-450 Aromatase Activity by Butylins. Steroids
2001, 66, 763−769.
(27) Maestro, version 9.1; Schrodinger, LLC: New York, NY, 2010.
̈
4002
dx.doi.org/10.1021/jm300262w | J. Med. Chem. 2012, 55, 3992−4002