Structure-guided design, synthesis, and evaluation of salicylic acid-based inhibitors targeting a selectivity pocket in the active site of human 20α-hydroxysteroid dehydrogenase (AKR1C1)

Article abstract of DOI:10.1021/jm9001633

The first design, synthesis, and evaluation of human 20α- hydroxysteroid dehydrogenase (AKR1C1) inhibitors based on the recently published crystal structure of its ternary complex with inhibitor are reported. While the enzyme-inhibitor interactions observ

Full text of DOI:10.1021/jm9001633

J. Med. Chem. 2009, 52, 3259–3264
3259
Structure-Guided Design, Synthesis, and Evaluation of Salicylic Acid-Based Inhibitors
Targeting a Selectivity Pocket in the Active Site of Human 20r-Hydroxysteroid Dehydrogenase
AKR1C1)
(
,
†
†
†
†
‡
‡
Ossama El-Kabbani,* Peter J. Scammells, Joshua Gosling, Urmi Dhagat, Satoshi Endo, Toshiyuki Matsunaga,
‡
‡
Midori Soda, and Akira Hara
Medicinal Chemistry and Drug Action, Monash Institute of Pharmaceutical Sciences, 381 Royal Parade, ParkVille, Victoria 3052, Australia,
and Laboratory of Biochemistry, Gifu Pharmaceutical UniVersity, Mitahora-higashi, Gifu 502-8585, Japan
ReceiVed February 9, 2009
The first design, synthesis, and evaluation of human 20R-hydroxysteroid dehydrogenase (AKR1C1) inhibitors
based on the recently published crystal structure of its ternary complex with inhibitor are reported. While
the enzyme-inhibitor interactions observed in the crystal structure remain conserved with the newly designed
inhibitors, the additional phenyl group of the most potent compound, 3-bromo-5-phenylsalicylic acid, targets
a nonconserved hydrophobic pocket in the active site of AKR1C1 resulting in 21-fold improved potency
i
(K ) 4 nM) over the structurally similar 3R-hydroxysteroid dehydrogenase isoform (AKR1C2). The
compound is hydrogen bonded to Tyr55, His117, and His222, and the phenyl ring forms additional van der
Waals interactions with residues Leu308, Phe311, and the nonconserved Leu54 (Val in AKR1C2).
Additionally, the metabolism of progesterone in AKR1C1-overexpressed cells was potently inhibited by
3
-bromo-5-phenylsalicylic acid, which was effective from 10 nM with an IC50 value equal to 460 nM.
Introduction
also involved in the development of several human and rodent
tumors, such as lung, endometrial, esophageal, ovarian, and
breast cancers, and its overexpression in cancer cells is related
a
Hydroxysteroid dehydrogenases (HSDs) belong to two
protein superfamilies, the aldo-keto reductase (AKR) super-
1
2-16
1
to drug-resistance against several anticancer agents.
family and the short-chain dehydrogenase/reductase superfam-
2
AKR1C1 is inhibited by a variety of compounds including
bile acids, phytoestrogens, synthetic estrogens, benzodiaz-
ily. Members of the AKR superfamily are NAD(P)(H)-
dependent oxidoreductases that possess a triose phosphate
isomerase (TIM) barrel motif consisting of an eight-stranded
epines, nonsteroidal anti-inflammatory agents, and synthetic
3
10,17-21
ꢀ
-sheet at the core surrounded by eight R-helices. While the
compounds.
Among the inhibitors, benzbromarone and
four human HSDs belonging to the AKR1C subfamily, AKR1C1
20R-HSD), AKR1C2 (type 3 3R-HSD), AKR1C3 (type 2 3R-
3′,3′′,5′,5′′-tetrabromophenolphthalein are the most potent in-
1
0
(
hibitors showing IC50 values of 48 and 33 nM, respectively.
Other inhibitors show IC50 or K values in the micromolar range,
HSD), and AKR1C4 (type 1 3R-HSD), share at least 86%
sequence homology with AKR1C1 and AKR1C2 in particular
differing only by seven residues, they display distinct positional
and stereo preferences with respect to their substrates and are
involved in different physiological roles. AKR1C1 has a major
role in progesterone metabolism that is essential for the
maintenance of pregnancy. The conversion of progesterone to
an inactive progestin, 20R-hydroxyprogesterone, by AKR1C1
is associated with premature birth leading to infant morbidity
i
and either lack the selectivity to AKR1C1 or their selectivity
was not examined using the four AKR1C isoforms. Recently,
we have identified dihalogenated salicylic acid derivatives as
4,5
i
potent inhibitors of AKR1C1 showing K values between 6-9
nM and determined the crystal structure of the first ternary
complex with a potent inhibitor bound in the active site at 1.8
6
2
2,23
Å resolution.
The inhibitor 3,5-dichlorosalicylic acid was
surrounded by side chains contributed by the 11 amino acid
residues Tyr24, Leu54, Tyr55, Trp86, His117, His222, Glu224,
Trp227, Leu306, Leu308, and Phe311 and anchored from its
carboxylate group that formed hydrogen bonds with the catalytic
residues His117 and Tyr55. Analysis of the inhibitor binding
site revealed that four nonconserved residues (Leu54, His222,
Leu306, and Leu308) in the four AKR1C isoforms are present
within van der Waals contacts (<4.0 Å) from the inhibitor, and
thus should be considered when designing specific inhibitors
7,8
and mortality. AKR1C1 has been implicated in brain function
where it modulates the occupancy of γ-aminobutyric acid type
A (GABA ) receptors through its reductive 20R-HSD activity,
A
which converts neuroactive steroids (3R,5R-tetrahydroproges-
terone and 5R-tetrahydrodeoxycorticosterone) and their precur-
sors (5R-dihydroprogesterone and progesterone) into inactive
and ineffective 20R-hydroxysteroids, thereby removing them
9
,10
from the synthetic pathway.
steroids by AKR1C1 is implicated in symptoms of premenstrual
syndrome and other neurological disorders.
The elimination of neuroactive
of AKR1C1. 3,5-Dichlorosalicylic acid shows a K value of 6
i
9
,11
The enzyme is
nM for AKR1C1 and greater than 4000-fold difference in
inhibitor potency between AKR1C1 and the two isoforms
AKR1C3 and AKR1C4, which is derived mainly from the
nonconserved interactions between the inhibitor and residues
*
To whom correspondence should be addressed. Phone: 61-3-9903-9691.
Fax: 61-3-9903-9582. E-mail: ossama.el-kabbani@pharm.monash.edu.au.
†
Monash Institute of Pharmaceutical Sciences.
Gifu Pharmaceutical University.
Abbreviations: HSD, hydroxysteroid dehydrogenase; AKR, aldo-keto
2
3
‡
Leu306 and Leu308 from the C-terminal loop. The inhibitor
also highly inhibits AKR1C2 that plays roles in the elimination
of the potent androgen 5R-dihydrotestosterone and the synthesis
a
reductase; AKR1C1, human 20R-hydroxysteroid dehydrogenase; AKR1C2,
human type 3 3R-hydroxysteroid dehydrogenase; S-tetralol, S-(+)-1,2,3,4-
4,5,10
of the GABA
A
receptor-modulated neuroactive steroids.
The
A
tetrahydro-1-naphthol; GABA , γ-aminobutyric acid type A; BAEC, bovine
aortic endothelial cell; LC/MS, liquid chromatography/mass spectrometry.
high selectivity and similar potencies of the inhibitor for
1
0.1021/jm9001633 CCC: $40.75
 2009 American Chemical Society
Published on Web 04/27/2009

3
260 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10
El-Kabbani et al.
Figure 1. Stereoview of the inhibitor compound 4 (INH; pink color) modeled in the superimposed active sites of AKR1C1 (green color) and
AKR1C2 (red color). The surrounding residues are labeled with their single letter code for clarity, and hydrogen bonds between the inhibitor and
AKR1C1 are shown as black dotted lines with corresponding distances given in angstroms. The shortest contacts between the 5-phenyl moiety of
3
4
the inhibitor and nonconserved residue 54 of AKR1C1 and AKR1C2 are shown in blue. The figure was prepared using MOLSCRIPT after energy
minimization.
Table 1. K
i
Values of 3,5-Dibromosalicylic Acid (DBSA) and Compounds 4 and 9 for the AKR1C Isoforms
values (µM)
K
i
inhibitor
DBSA
compound 4
compound 9
AKR1C1
AKR1C2
AKR1C3
AKR1C4
a
0.009 ( 0.0002
0.004 ( 0.0004
0.14 ( 0.017
0.082 ( 0.0023 (9)
0.087 ( 0.012 (21)
1.97 ( 0.013 (14)
23 ( 1.1 (2600)
4.2 ( 0.15 (890)
21 ( 3.4 (150)
45.7 ( 5.9 (5100)
18.2 ( 2.5 (3900)
b
NI
a
b
Ratios of AKR1C isoforms (1C2-1C4) to AKR1C1 are given in parentheses. No inhibition was observed at inhibitor concentrations upto 100 µM.
AKR1C1 and AKR1C2 are due to the homologous structures
of their active sites that differ by only one amino acid residue,
which is Leu54 in AKR1C1 and is Val54 in AKR1C2. Thus,
it is important that newly designed inhibitors capture the
maximum interactions with Leu54 of AKR1C1 in order to
improve the selectivity over AKR1C2.
pocket forming additional van der Waals interactions with
residues Leu308, Phe311, and the nonconserved Leu54 (Val in
AKR1C2). As previously observed in the structure of the ternary
2
3
2
3
complex, the salicylic acid moiety is present with van der
Waals contacts from Leu306. In addition to the synthesis of
compound 4, 3-phenyl-5-bromosalicylic acid (compound 9) that
has the phenyl and bromo groups present at opposite positions
compared to that in compound 4, was synthesized for compari-
son, and together with 3,5-dibromosalicylic acid, the inhibitory
potencies of the three compounds were measured against the
four isoforms (AKR1C1 to AKR1C4). With the exception of
compound 9 and AKR1C4, the three compounds inhibited the
four isoforms competitively with respect to the substrate S-(+)-
In this study, we report the first design, synthesis, and
evaluation of AKR1C1 inhibitors based on the recently pub-
23
lished crystal structure of the AKR1C1 ternary complex. While
the enzyme-inhibitor interactions observed in the crystal
structure remain conserved with the newly designed inhibitors,
the additional phenyl group of the most potent compound,
3
i
-bromo-5-phenylsalicylic acid (K ) 4 nM), targets a hydro-
phobic pocket in the active site of AKR1C1 interacting with
the nonconserved Leu54, resulting in improved specificity over
AKR1C2. In addition, this compound potently inhibited the
metabolism of progesterone by AKR1C1 in the cells, showing
an IC50 (concentration for 50% inhibition of this metabolism)
value of 460 nM. These results provide the framework needed
for the development of new inhibitors that are more specific to
AKR1C1 for potential use in treatments against cancer and
neurological disorders.
1,2,3,4-tetrahydro-1-naphthol (S-tetralol), and their K values are
i
shown in Table 1. Compound 9 showed the least potency for
AKR1C1 and AKR1C2, likely due to short contacts and
disruption of the hydrogen bonding interaction with His222
(Figure 1). However, compound 4 showed improved potency
and selectivity toward AKR1C1 compared to those of 3,5-
dibromosalicylic acid because of the additional favorable
interactions between the 5-phenyl ring and the residues lining
the selectivity pocket (Figure 1), and was a 21-fold more potent
inhibitor of AKR1C1 than AKR1C2. Additionally, compared
2
3
to 3,5-dichlorosalicylic acid, compound 4 displayed a 2.9-
fold enhancement in the calculated binding energies of the
AKR1C1-inhibitor complex (-114 kcal/mol vs -39 kcal/mol)
and was a 2-fold more potent and selective inhibitor of AKR1C1
than AKR1C2.
Results and Discussion
2
3
24
The crystal structure of AKR1C1 together with a GRID
analysis of the inhibitor-binding site suggested that the replace-
ment of the bromine atom at the 5-position of 3,5-dibromosali-
cylic acid with a phenyl (compound 4) is expected to enhance
the inhibitor potency and selectivity for AKR1C1 over AKR1C2.
Benzene rings and other aromatic systems are common moieties
among compounds used as therapeutic agents, playing major
roles ranging from providing steric bulk to forming an integral
Chemical Synthesis of Compounds 4 and 9. Compound 4
was synthesized using the approach outlined in Scheme 1.
Briefly, this involved a Suzuki coupling between methyl
5
-iodosalicylate (1) and phenyl boronic acid to give the biphenyl
. Bromination of the 3-position (Br ) followed by saponification
O) afforded the target compound
in good yield (75% over two steps).
The synthetic approach of compound 9 is depicted in Scheme
2. Methyl 5-bromosalicylate was reacted with NaI and chloram-
2
5
2
2
part of the pharmacophore. Residues present within van der
of the methyl ester (LiOH.H
2
Waals contacts of the modeled compound 4 are shown in Figure
4
1
. The carboxyl and hydroxyl groups of this compound are
hydrogen bonded to the side-chains of Tyr55, His117, and
His222, and its phenyl moiety enters a hydrophobic selectivity

AKR1C1 Inhibitor Design
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10 3261
a
Scheme 1. Synthesis of 3-Bromo-5-phenylsalicylic Acid (4)
a
Reagents and conditions: (i) PhB(OH)
2 2 3
, K CO , Pd(PPh
3 4 2 2 2
) , DMF, 70 °C, 24 h; (ii) Br , MeOH, rt, 20 min; (iii) LiOH.H O, THF:H O (4:1), 70 °C, 24 h.
a
Scheme 2. Synthesis of 3-Phenyl-5-bromosalicylic Acid (9)
Figure 2. LC/MS analysis of the metabolism of progesterone (p) in
control and AKR1C1-overexpressed BAECs. The cells were cultured
for 6 h in the medium containing 30 µM progesterone in the absence
or presence of 10 µM compound 4. Chromatograms: (a) the overex-
pressed cells incubated without inhibitor, (b) the overexpressed cells
incubated with compound 4, and (c) the control cells without inhibitor.
The position of the metabolite was identical to that of the authentic
20R-hydroxyprogesterone (20R). The peak other than that of the
substrate and metabolite is due to unknown substances (*), which are
present in the medium.
a
Reagents and conditions: (i) NaI, Chloramine T, DMF, rt; (ii) MeI,
K
2
CO
3
, DMF, rt; (iii) PhB(OH)
2 3 4 3 4
, K PO , Pd(PPh ) , DMF, 100 °C; (iv)
BBr (1 M in CH
3
2
Cl ), CH Cl , -78 °C.
2
2
2
ine T to give the corresponding 3-iodo analogue 6. Suzuki
coupling of this compound with phenyl boronic acid did not
proceed cleanly (by TLC); therefore, it was decided to first
protect the phenol as the corresponding methyl ether 7. This
straightforward transformation was achieved using methyl iodide
and potassium carbonate. Coupling of methyl 2-methoxy-3-iodo-
5
-bromobenzoate (7) with phenyl boronic acid afforded the
desired biphenyl 8 in 84% yield. The reaction of 8 with excess
boron tribromide effected the cleavage of both the methyl ether
and ester to yield the target compound, 5-bromo-3-phenylsali-
cylic acid (9).
Bovine aortic endothelial cells (BAECs) that are transfected
with the cDNA for AKR1C1 were used to evaluate the inhibitory
potency of 3,5-dibromosalicyclic acid, compound 4, and com-
pound 9 at the cellular level. The expression of AKR1C1 in
the cells was confirmed by assaying the S-tetralol dehydrogenase
activity. The activity in the extract of the transfected cells (0.27
unit/mg) was 10-fold higher than that of the control cells
transfected with the vector alone. This was also evident in the
metabolism of progesterone in the transfected cells (Figure 2),
but not in the control cells. The transfected BAECs metabolized
progesterone only into 20R-hydroxyprogesterone as shown in
the liquid chromatography/mass spectrometry (LC-MS) chro-
matogram (Figure 2a), where the retention time of this
metabolite was identical to that of authentic 20R-hydroxy-
progesterone and distinct from that (16 min) of authentic 20ꢀ-
hydroxyprogesterone. Compound 9, compound 4, and 3,5-
dibromosalicylic acid inhibited the metabolism of progesterone
in the cells, as shown in the representative LC-MS chromato-
gram for the addition of compound 4 (Figure 2b). Comparison
of the dose-response curves for the three inhibitors (Figure 3)
indicated that compound 4 most potently inhibited the metabo-
lism of progesterone. It was effective from 10 nM, and the IC50
value was 460 nM. 3,5-Dibromosalicylic acid was effective from
Figure 3. Dose-response curves for 3,5-dibromosalicylic acid,
compound 4, and compound 9 in the inhibition of progesterone
metabolism by the AKR1C1-overexpressed BAECs. (O), 3,5-dibro-
mosalicylic acid; (2), compound 9; and (b), compound 4. The cell
culture conditions were the same as those described in the legend of
Figure 2.
compound 9 was low, as evident by only 40% inhibition at its
high concentration of 10 µM. It should be noted that the three
compounds did not affect the cell viability up to their concentra-
tions of 10 µM. The ratio of the IC50 value of compound 4 to
that of 3,5-dibromosalicylic acid was 5, which is higher than
the ratio of K
i
values of the two compounds for AKR1C1. This
0
.1 µM, and its IC50 value was 2.3 µM, while the inhibition by
suggests that the addition of the phenyl ring at 5-position of

3
262 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10
El-Kabbani et al.
the salicylic acid resulted in increasing the cell permeability of
the inhibitor.
In summary, the use of the recently determined crystal
structure of AKR1C1 complexed with an inhibitor in conjunc-
tion with a GRID analysis of the inhibitor-binding site has
allowed the design of a new salicylic acid-based inhibitor
respect to the inhibitor. The most favored probe suggested by the
program that may improve the binding and specificity of the 3,5-
dihalogenated salicylic acids to AKR1C1 over AKR1C2 was the
phenyl because of its binding in a hydrophobic pocket and
interaction with the nonconserved Leu54 (Val in AKR1C2).
3
-Bromo-5-phenylsalicylic acid (compound 4) was manually
docked into the active site of AKR1C1, on the basis of the observed
orientation of 3,5-dicholorosalicylic acid in the 1.8 Å resolution
(
compound 4) with improved potency (K
selectivity (21-fold) over that of AKR1C2. Moreover, compound
significantly decreased the metabolism of progesterone in the
i
) 4 nM) and
23
crystal structure, and any water molecules in the active site
4
coinciding with the docked compound were removed. While there
was no apparent steric hindrance in the model of the complex, to
relieve any strain associated with the crystal coordinates the model
was subjected to energy minimization using the Discover 2.7
program (Biosym Technologies, San Diego, California, USA) on
a Linux workstation following published protocols found to be
effective for visualizing a protein-ligand complex in its lowest
cells with an IC50 value of 460 nM, which is comparable or
superior to the IC50 values of the previously known two most
potent inhibitors of AKR1C1, benzbromarone and 3′,3′′,5′,5′′-
1
0
tetrabromophenolphthalein. Compound 4 was designed to
target a selectivity pocket in the active site of AKR1C1 lined
by the three apolar residues Leu54, Leu308, and Phe311. Leu308
is one of two nonconserved C-terminal residues (the other
residue is Leu306) responsible for the greater than 4000-fold
difference in inhibitor potency between AKR1C1 and the two
27,28
energy conformation.
Briefly, hydrogen atoms, partial charges,
atomic potentials, and bond orders were assigned by using the
automatic procedures within InsightII. Arginine, lysine, aspartate,
and glutamate amino acids were charged, while histidines were
uncharged, with hydrogen atoms fixed at the Nε2 atoms. Energy
minimization calculations included a constant-valence force field
incorporating the simple harmonic function for bond stretching and
excluding all nondiagonal terms were carried out (cutoff distance
of 26 Å) using the steepest-descent and conjugate-gradient algo-
rithms (down to a maximum atomic root-mean-square derivative
of 10.0 kcal/Å and 0.01 kcal/Å, respectively). Additionally, the
values for the intermolecular binding energy were calculated for
the enzyme-inhibitor complexes with 3,5-dichlorosalicylic acid and
compound 4 bound in the active site using Discover as previously
2
3
isoforms AKR1C3 and AKR1C4. Since the active sites of
AKR1C1 and AKR1C2 differ only by one amino acid residue,
which is Leu54 in AKR1C1 and is Val54 in AKR1C2, and the
current inhibitors show similar potency for the two enzymes,
newly designed inhibitors that capture the maximum interactions
with Leu54 in AKR1C1 are needed in order to improve their
selectivity over AKR1C2. Thus, future developments of new
derivatives of compound 4 are likely to improve on the
selectivity of the currently known AKR1C1 inhibitors. We have
also illustrated that while large chemical database searches are
useful in discovering new enzyme inhibitors, the use of the high
resolution crystal structure of an enzyme-inhibitor complex is
an effective tool in optimizing the enzyme-inhibitor interaction
by exploiting the small structural differences between the
different enzyme isoforms.
26
described.
Chemistry. NMR spectra were recorded on a Bruker 300 WB
1
spectrometer with an Avance console. Unless otherwise stated, H
1
3
3
and C NMR spectra were obtained in CDCl at 300 and 75 MHz,
respectively. High-resolution electrospray mass spectra (HRMS)
were obtained on a Waters LCT Premier XE (TOF) spectrometer.
TLC analyses were run on precoated silica gel 60 F254 aluminum
plates (Merck), and compounds were analyzed under UV light and
stained using phosphomolybdic acid (48 g in 100 mL of EtOH).
Column chromatography was performed using Merck silica gel 60
(particle size 0.063-0.200 µm, 70-230 mesh). The purity of the
target compounds (4 and 9) was determined to be >98% by LC-
MS using an Agilent 6120 Quadrapole LC-MS with an Eclipse
XDB-C18 column (5 µm, 4.6 × 150 mm).
Experimental Section
Molecular Modeling and Inhibitor Design. In an attempt to
develop potent and more specific inhibitors of the enzyme,
compounds were designed based on the crystal structure of the
23
AKR1C1 ternary complex. To aid the design process, the program
2
4
GRID (version 18) was used to search the active site of the
enzyme for the most suitable or favorable positions of a variety of
probes, as previously described by our laboratory for L-xylulose
Methyl 5-phenylsalicylate (2). Methyl 5-iodosalicylate (1)
(0.150 g, 0.539 mmol) was dissolved in anhydrous, degassed DMF
(5 mL). Potassium phosphate tribasic (0.240 g, 1.131 mmol),
26
reductase. Briefly, a total of 25 probes were tested with AKR1C1,
which included the following groups: methyl, aromatic carbon,
amino, amido, and heterocyclic nitrogens, halogens, sulfur, carbonyl,
ether, and hydroxyl oxygen, carboxy group, water, phosphate, and
other ions. Each probe was analyzed independently using the
InsightII 2.1 package (Biosym Technologies Inc., San Diego, CA).
Calculations were performed on a cube (35 Å per side) centered
on the active site, with a grid spacing of 0.5 Å. The interaction
energy between the probe and every atom within the protein
structure was evaluated at each grid point. A dielectric constant of
3 4
Pd(PPh ) (0.035 g, 0.003 mmol, 5 mol %), and phenyl boronic
acid (0.073 g, 0.599 mmol) were added, and the reaction mixture
was stirred at 70 °C overnight. The reaction mixture was partitioned
between CH
was further extracted with CH
organic phase were dried (Na SO
to give brown crystals (0.128 g, 96%). H NMR δ 4.00 (m, 3H),
7.09 (d, 1H, J ) 8.7, 1H), 7.36 (t, 1H, J ) 7.5 Hz), 7.46 (t, 2H, J
) 7.5 Hz), 7.58 (d, 2H, J ) 8.4 Hz), 7.72 (d, 1H, 2.4, 8.4 Hz),
8.11 (d, 1H, J ) 2.4 Hz), 10.75 (s, 1H).
Methyl 3-Bromo-5-phenylsalicylate (3). Methyl 5-phenylsali-
cylate (2) (0.198 g, 0.86 mmol) was dissolved in MeOH (5 mL),
2
Cl
2
(30 mL) and H
2
O (50 mL), and the aqueous layer
Cl
(2 × 10 mL). The combined
), filtered, and reduced in Vacuo
2
2
2
4
1
8
0 was used to simulate a bulk aqueous phase, while areas as
determined by GRID to be excluded from solvent were assigned a
dielectric constant of 4 (i.e., the interior of the protein). The
accompanying program GRIN was used to automatically assign
atom types and charges for the protein, using the standard parameter
file provided with GRID. The output was converted (using GINS
supplied with GRID) into a form suitable for input to the Biosym
utility contour, and contour maps were built up using steps of 1
kcal/mol. The contour map detailed a number of energy levels.
Negative energy levels delineate regions at which ligand binding
is favored in particular, and positive energy levels define the surface
of the target. The contour map was then superimposed on the active
site of AKR1C1 using InsightII. Visual inspection of the superim-
posed 3,5-dichlorosalicylic acid on the active site of AKR1C1
provided information on the predicted position of the probe with
and a solution of Br (50 µL, 0.97 mmol) in MeOH (2 mL) was
2
added dropwise. The reaction mixture was stirred overnight at room
temperature. A precipitate formed during the course of the reaction
and was collected via vacuum filtration and washed with cold
1
MeOH (0.02 g, 87%). H NMR δ 4.02 (s, 3H), 7.36 (d, 1H, J )
7.2 Hz), 7.45 (br t, 2H, J ) 7.5 Hz), 7.54 (d, 2H, J ) 7.5 Hz), 7.99
(d, 1H, J ) 2.1 Hz), 8.06 (d, 1H, J ) 2.1 Hz), 11.45 (s, 1Η).
3-Bromo-5-phenylsalicylic Acid (4). Methyl 3-bromo-5-phe-
nylsalicylate (3) (0.120 g, 0.387 mmol) was dissolved in THF/H
(4:1) (4 mL), LiOH ·H O (0.081 g, 1.937 mmol) added, and the
reaction refluxed overnight. The reaction was diluted with water
and acidified to pH 1 with 5 M HCl (5 mL). Extraction with CH Cl
2
O
2
2
2

AKR1C1 Inhibitor Design
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10 3263
(
4 × 5 mL) followed by drying (Na
2
SO
4
), filtration, and evaporation
software ED50 and IC50 for graded Response version 1.2. The
inhibition patterns were determined by fitting the initial velocities
1
afforded a white solid (0.098 g, 86%). H NMR δ 7.35 (d, 1H, J
)
7.2 Hz), 7.47 (d, 2H, J ) 7.2 Hz), 7.55 (d, 2H, J ) 7.2 Hz),
using five substrate concentrations (0.2-2 × K
0.5-5 × K for other enzymes) in the presence of the inhibitor
concentrations (0-0.5 × IC50) to Lineweaver-Burk and Dixon
plots. The K values were calculated by using the appropriate
m
for AKR1C3 and
8
.05 (d, 1H, J ) 2.3 Hz), 8.13 (d, 1H, J ) 2.3 Hz), 11.2 (s, 1H).
C NMR δ 112.3, 115.9, 127.6, 128.7, 128.8, 130.3, 133.6, 137.6,
m
1
3
1
39.0, 158.4, 172.8. HRMS (ESI) m/z cacld for C13
H
9
BrO
3
[M -
i
-
H] : 290.9662. Found: 290.9654.
programs of ENZFITTER (Biosoft, Cambrige, UK) and are
expressed as the mean ( standard error of at least three determina-
tions.
Methyl 3-iodo-5-bromosalicylate (6). Methyl 5-bromosalicylate
(
5) (15.00 g, 64.92 mmol) was dissolved in DMF (50 mL) prior to
the addition of NaI (11.68 g, 77.92 mmol) followed by chloramine
T (20.60 g, 77.03 mmol). The reaction mixture was stirred overnight
at room temperature. After filtration and evaporation of the solvent,
Evaluation of Inhibitors in the Cells. BAECs, a generous gift
from Taisho Pharmaceutical Co. (Saitama, Japan), were cultured
in Dulbecco’s modified Eagle’s medium supplemented with 10%
fetal bovine serum, penicillin (100 U/mL), and streptomycin (100
the resultant residue was taken up in Et
with H (100 mL), and brine (100 mL).
O (5 × 100 mL), Na
The Et O layer was separated, dried, and evaporated to give a
2
O (250 mL) and washed
S
2
O
3
2
µg/mL) at 37 °C in a 5% CO incubator. In all experiments, the
2
2
cells were used at passages 4-8, and the endothelial cobblestone
morphology was confirmed microscopically before use. The expres-
sion vector with the cDNA for AKR1C1 was constructed according
2
yellow solid, which was recrystallized from EtOH (12.50 g, 52%).
1
H NMR δ 3.99 (s, 3H), 7.98 (d, 1H, J ) 2.4 Hz), 8.05 (d, 1H, J
3
2
)
2.4 Hz), 11.55 (s, 1H).
to the method previously reported. The cDNA was initially
29
Methyl 2-methoxy-3-iodo-5-bromobenzoate (7). Methyl 3-iodo-
amplified from the bacterial expression vector pGEX/AKR1C1
5
-bromosalicylate (6) (0.98 g, 2.74 mmol) was dissolved in DMF
by PCR using the primer pairs consisting of a forward primer (5′-
GAGTCGACgccaccATGGATTCGAAATATCAGTGT-3′) and a
reverse primer (5′-AGGTCGACTTAATATTCATCAGAAAATGGA-
3′), in which SalI site, a Kozak sequence and a start codon are
expressed in italic letters, small letters, and underlined letters,
respectively. The PCR product was verified by automated DNA
sequencing and subcloned at the SalI site of the eukaryotic
(
5 mL), and potassium carbonate (0.53 g, 3.73 mmol) and
iodomethane (0.24 mL, 3.86 mmol) were added. The reaction
mixture was stirred at ambient temperature overnight prior to being
diluted with Et
The organic layer was dried (Na
2
O (20 mL) and washed with H
2
O (5 × 100 mL).
2
SO ), filtered, and evaporated in
4
1
Vacuo to give a yellow solid (0.55 g, 54%). H NMR δ 3.89 (s,
3
2
3
2
H), 3.94 (s, 3H), 7.92 (d, 1H, J ) 2.4 Hz, 1H), 8.07 (d, 1H, J )
expression vector pGW1. The expression vector with the insert
was then transfected into subconfluent BAECs using Lipofectamine
2000 (Invitrogen). The transfected cells were maintained in the
medium containing 2% fetal bovine serum for 24 h and then used
to evaluate the inhibitory effects of 3,5-dibromosalicylic acid and
compounds 4 and 9 on the metabolism of progesterone in the cells.
The cells were pretreated for 2 h with various concentrations of
inhibitors in serum-free growth medium prior to incubating for 6 h
with 30 µM progesterone. The culture media were collected by
centrifugation, and the lipidic fraction of the media was extracted
twice by ethyl acetate. The metabolite, 20R-hydroxyprogesterone,
was quantified on a LC-MS using a Chiralcel OJ-H 5 µm column
.4 Hz).
Methyl 5-bromo-2-methoxybiphenyl-3-carboxylate (8). Methyl
-bromo-3-iodo-2-methoxybenzoate (7) (0.150 g, 0.40 mmol) was
5
dissolved in degassed DMF (8 mL), and tribasic potassium
phosphate (0.103 g, 0.485 mmol), tetrakis(triphenylphosphine)-
palladium (0.025 g, 0.002 mmol), and phenylboronic acid (0.06 g,
0
.492 mmol) were added. The reaction mixture was stirred overnight
at 100 °C. After cooling to room temperature, the solution was
diluted with CH Cl (10 mL) and washed with water (10 mL). The
aqueous layer was acidified with 5 M HCl (5 mL) and extracted
2
2
with CH
dried (Na
2
Cl
2
(3 × 10 mL). The combined organic extracts were
3
3
SO
4
), filtered, and reduced in Vacuo, and the resultant
as described previously.
2
residue was purified by column chromatography using hexane/
CH Cl (2:1) eluent to afford the target compounds as a brown
solid (0.087 g, 89% yield). H NMR δ 3.47 (s, 3H), 3.95 (s, 3H),
.38-7.56 (m, 5H), 7.62 (d, 1H, J ) 2.7 Hz), 7.87 (d, 1H, J ) 2.7
Hz).
Acknowledgment. U.D. is the recipient of a Monash
Graduate School postgraduate scholarship. This work was partly
supported by a Grant-in-Aid for Young Scientists (to S.E.).
2
2
1
7
3
-Phenyl-5-bromosalicylic Acid (9). Methyl 5-bromo-2-meth-
References
oxybiphenyl-3-carboxylate (8) (0.13 g, 0.39 mmol) was dissolved
(
1) Hyndman, D.; Bauman, D. R.; Heredia, V. V.; Penning, T. M. The
aldo-keto reductase superfamily homepage. Chem. Biol. Interact. 2003,
143-144, 621–631.
in dry CH
C in a dry ice/acetone bath, and boron tribromide (0.67 mL, 0.98
mmol) was added dropwise. After stirring overnight in a dry ice/
acetone bath, the reaction mixture was diluted with CH Cl (10
mL) and washed with 1 M citric acid (2 × 30 mL) and saturated
NaHCO (30 mL). The aqueous layer was cooled on ice and
2 2 2
Cl (3 mL) under N . The solution was cooled to -78
°
(2) Oppermann, U.; Filling, C.; Hult, M.; Shafqat, N.; Wu, X.; Lindh,
M.; Shafqat, J.; Nordling, E.; Kallberg, Y.; Persson, B.; J o¨ rnvall, H.
Short-chain dehydrogenases/reductases (SDR): the 2002 update. Chem.
Biol. Interact. 2003, 143-144, 247–253.
2
2
3
(
3) Jez, J. M.; Bennett, M. J.; Schlegel, B. P.; Lewis, M.; Penning, T. M.
Comparative anatomy of the aldo-keto reductase superfamily. Biochem.
J. 1997, 326, 625–636.
acidified with 5 M HCl (10 mL). A white precipitate formed and
1
was collected by vacuum filtration (0.07 g, 57%). H NMR δ
7
.36-7.47 (m, 3H), 7.56 (d, 2H, J ) 7.8 Hz), 7.66 (d, 1H, J ) 2.4
(4) Penning, T. M.; Burczynski, M. E.; Jez, J. M.; Hung, C. F.; Lin, H. K.;
Ma, H.; Moore, M.; Palackal, N.; Ratnam, K. Human 3R-hydroxy-
steroid dehydrogenase isoforms (AKR1C1-AKR1C4) of the aldo-keto
reductase superfamily: functional plasticity and tissue distribution
reveals roles in the inactivation and formation of male and female
sex hormones. Biochem. J. 2000, 351, 67–77.
(5) Bauman, D. R.; Steckelbroeck, S.; Penning, T. M. The roles of aldo-
keto reductases in steroid hormone action. Drug News Perspect. 2004,
17, 563–578.
1
3
Hz), 8.03 (d, 1H, J ) 2.4 Hz), 11.10 (s, 1H). C NMR δ 111.2,
1
12.8, 128.2, 128.4, 129.3, 132.2, 133.0, 135.5, 140.3, 158.7, 173.5.
-
HRMS (ESI) m/z cacld for C13
90.9662.
Enzyme and Activity Assays. The recombinant AKR1C1,
AKR1C2, AKR1C3, and AKR1C4 were expressed in Escherichia
coli JM109 and purified to homogeneity as previously described.
Protein concentration was determined by a bicinchoninic acid
protein assay reagent kit (Pierce) using bovine serum albumin as
the standard. The NADP -linked S-tetralol dehydrogenase activity
of the enzymes was assayed by measuring the rate of change in
NADPH fluorescence (at 455 nm with an excitation wavelength of
40 nm) or its absorbance (at 340 nm) at 25 °C, as described
previously. In the inhibition assays, the IC50 values for the
9 3
H BrO [M - H] : 290.9662. Found:
2
29-31
(
6) Zhang, Y.; Dufort, I.; Rheault, P.; Luu-The, V. Characterization of a
human 20R-hydroxysteroid dehydrogenase. J. Mol. Endocrinol. 2000,
2
5, 221–228.
+
(7) Lewis, M. J.; Wiebe, J. P.; Heathcote, J. G. Expression of progesterone
metabolizing enzyme genes (AKR1C1, AKR1C2, AKR1C3, SRD5A1,
SRD5A2) is altered in human breast carcinoma. BMC Cancer 2004,
4
, 27–39.
3
(
8) Piekorz, R. P.; Gingras, S.; Hoffmeyer, A.; Ihle, J. N.; Weinstein, Y.
Regulation of progesterone levels during pregnancy and parturition
by signal transducer and activator of transcription 5 and 20R-
hydroxysteroid dehydrogenase. Mol. Endocrinol. 2005, 19, 431–440.
23
inhibitors were initially determined with the S-tetralol concentration
0.1 mM for AKR1C1 and 1 mM for other enzymes) using a
(

3
264 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10
El-Kabbani et al.
(
9) Lambert, J. J.; Belelli, D.; Hill-Venning, C.; Peters, J. A. Neurosteroids
(23) Dhagat, U.; Endo, S.; Sumii, R.; Hara, A.; El-Kabbani, O. Selectivity
determinants of inhibitor binding to human 20R-hydroxysteroid
dehydrogenase: Crystal structure of the enzyme in ternary complex
with coenzyme and the potent inhibitor 3,5-dichlorosalicylic acid.
J. Med. Chem. 2008, 51, 4844–4848.
(24) Goodford, P. J. A computational procedure for determining energeti-
cally favorable binding sites on biologically important macromolecules.
J. Med. Chem. 1985, 28, 849–857.
and GABAA receptor function. Trends Pharmacol. Sci. 1995, 16, 295–
3
03.
(
10) Higaki, Y.; Usami, N.; Shintani, S.; Ishikura, S.; El-Kabbani, O.; Hara,
A. Selective and potent inhibitors of human 20R-hydroxysteroid
dehydrogenase (AKR1C1) that metabolizes neurosteroids derived from
progesterone. Chem. Biol. Interact. 2003, 143-144, 503–513.
11) Eser, D.; Sch u¨ le, C.; Baghai, T. C.; Romeo, E.; Uzunov, D. P.;
Rupprecht, R. Neuroactive steroids and affective disorders. Pharmacol.
Biochem. BehaV. 2006, 84, 656–666.
12) Wiebe, J. P. Role of progesterone metabolites in mammary cancer. J.
Dairy Res. 2005, 72, 51–57.
13) Deng, H. B.; Adikari, M.; Parekh, H. K.; Simpkins, H. Ubiquitous
induction of resistance to platinum drugs in human ovarian, cervical,
germ-cell and lung carcinoma tumor cells overexpressing isoforms 1
and 2 of dihydrodiol dehydrogenase. Cancer Chemother. Pharmacol.
(
(25) Lednicer, D. Strategies for Organic Drug Synthesis and Design; Wiley-
Inter science: New York, 1998; pp 31-32.
(
(
(
26) Carbone, V.; Ishikura, S.; Hara, A.; El-Kabbani, O. Structure-based
discovery of human L-xylulose reductase inhibitors from database
screening and molecular docking. Bioorg. Med. Chem. 2005, 13, 301–
3
12.
(
(
(
27) Darmanin, C.; El-Kabbani, O. Modelling studies on the binding of
substrate and inhibitor to the active site of human sorbitol dehydro-
genase. Bioorg. Med. Chem. Lett. 2000, 10, 1101–1104.
28) Darmanin, C.; El-Kabbani, O. Modelling studies of the active site of
human sorbitol dehydrogenase: an approach to structure-based inhibitor
design of the enzyme. Bioorg. Med. Chem. Lett. 2001, 11, 3133–3136.
2
004, 54, 301–307.
(
14) Wang, H. W.; Lin, C. P.; Chiu, J. H.; Chow, K. C.; Kuo, K. T.; Lin,
C. S.; Wang, L. S. Reversal of inflammation-associated dihydrodiol
dehydrogenases (AKR1C1 and AKR1C2) overexpression and drug
resistance in nonsmall cell lung cancer cells by wogonin and chrysin.
Int. J. Cancer 2007, 120, 2019–2027.
15) Rizner, T. L.; Smuc, T.; Rupreht, R.; Sinkovec, J.; Penning, T. M.
AKR1C1 and AKR1C3 may determine progesterone and estrogen
ratios in endometrial cancer. Mol. Cell. Endocrinol. 2006, 248, 126–
29) Matsuura, K.; Hara, A.; Deyashiki, Y.; Iwasa, H.; Kume, T.; Ishikura,
S.; Shiraishi, H.; Katagiri, Y. Roles of the C-terminal domains of
human dihydrodiol dehydrogenase isoforms in the binding of substrates
and modulators: probing with chimaeric enzymes. Biochem. J. 1998,
(
(
(
3
36, 429–436.
1
35.
16) Selga, E.; No e´ , V.; Ciudad, C. J. Transcriptional regulation of aldo-
keto reductase 1C1 in HT29 human colon cancer cells resistant to
methotrexate: role in the cell cycle and apoptosis. Biochem. Pharmacol.
(30) Matsuura, K.; Shiraishi, H.; Hara, A.; Sato, K.; Deyashiki, Y.;
Ninomiya, M.; Sakai, S. Identification of a principal mRNA species
for human 3R-hydroxysteroid dehydrogenase isoform (AKR1C3) that
2
008, 75, 414–426.
2
exhibits high prostaglandin D 11-ketoreductase activity. J. Biochem.
17) Brozic, P.; Cesar, J.; Kovac, A.; Davies, M.; Johnson, A. P.; Fishwick,
C. W.; Lanisnik Rizner, T.; Gobec, S. Derivatives of pyrimidine,
phthalimide and anthranilic acid as inhibitors of human hydroxysteroid
dehydrogenase AKR1C1. Chem. Biol. Interact. 2009, 178, 158–164.
18) Brozic, P.; Turk, S.; Lanisnik Rizner, T.; Gobec, S. Discovery of new
inhibitors of aldo-keto reductase 1C1 by structure-based virtual
screening. Mol. Cell. Endocrinol. 2009, 301, 245–250.
1998, 124, 940–946.
(31) Shiraishi, H.; Ishikura, S.; Matsuura, K.; Deyashiki, Y.; Ninomiya,
M.; Sakai, S.; Hara, A. Sequence of the cDNA of a human dihydrodiol
dehydrogenase isoform (AKR1C2) and tissue distribution of its mRNA.
Biochem. J. 1998, 334, 399–405.
(
(
(
32) Tanaka, N.; Aoki, K.; Ishikura, S.; Nagano, M.; Imamura, Y.; Hara,
A.; Nakamura, K. T. Molecular basis for peroxisomal localization of
tetrameric carbonyl reductase. Structure 2008, 16, 388–397.
19) Byrns, M. C.; Penning, T. M. Type 5 17ꢀ-hydroxysteroid dehydro-
genase/prostaglandin F synthase (AKR1C3): role in breast cancer and
inhibition by non-steroidal anti-inflammatory drug analogs. Chem. Biol.
Interact. 2009, 178, 221–227.
20) Brozic, P.; Smuc, T.; Gobec, S.; Rizner, T. L. Phytoestrogens as
inhibitors of the human progesterone metabolizing enzyme AKR1C1.
Mol. Cell. Endocrinol. 2006, 259, 30–42.
21) Usami, N.; Yamamoto, T.; Shintani, S.; Ishikura, S.; Higaki, Y.;
Katagiri, Y.; Hara, A. Substrate specificity of human 3(20)R-
hydroxysteroid dehydrogenase for neurosteroids and its inhibition by
benzodiazepines. Biol. Pharm. Bull. 2002, 25, 441–445.
(
33) Ishikura, S.; Nakajima, S.; Kaneko, T.; Shintani, S.; Usami, N.;
Yamamoto, I.; Carbone, V.; El-Kabbani, O.; Hara, A. Comparison of
Stereoselective Reduction of 3- and 20-Oxosteroids among Mouse and
Primate 20R-Hydroxysteroid Dehydrogenases. In Enzymology and
Molecular Biology of Carbonyl Metabolism 12; Weiner, H., Plapp,
B., Lindahl, R., Maser, E., Eds.; Purdue University Press: West
Lafayette, IN, 2005; pp 341-351.
(
(
(
34) Kraulis, P. MOLSCRIPT: a program to produce both detailed and
schematic plots of protein structures. J. Appl. Crystallogr. 1991, 24,
9
46–950.
(
22) Dhagat, U.; Carbone, V.; Chung, R. P.; Matsunaga, T.; Endo, S.; Hara,
A.; El-Kabbani, O. A salicylic acid-based analogue discovered from
virtual screening as a potent inhibitor of human 20R-hydroxysteroid
dehydrogenase. Med. Chem. 2007, 3, 546–550.
JM9001633