Structure-based optimization and biological evaluation of human 20α-hydroxysteroid dehydrogenase (AKR1C1) salicylic acid-based inhibitors

Article abstract of DOI:10.1016/j.ejmech.2010.08.052

The tertiary structure of the Leu308Val mutant of human 20α-hydroxysteroid dehydrogenase (AKR1C1) in complex with the inhibitor 3,5-dichlorosalicylic acid (DCL) has been determined. Structures and kinetic properties of the wild-type and mutant enzymes indicate that Leu308 is a selectivity determinant for inhibitor binding. The Leu308Val mutation resulted in 13-fold and 3-fold reductions in the inhibitory potencies of DCL and 3-bromo-5-phenylsalicylic acid (BPSA), respectively. The replacement of Leu308 with an alanine resulted in 473-fold and 27-fold reductions in the potencies for DCL and BPSA, respectively. In our attempts to optimize inhibitor potency and selectivity we synthesized 5-substituted 3-chlorosalicylic acid derivatives, of which the most potent compound, 3-chloro-5-phenylsalicylic acid (K_i = 0.86 nM), was 24-fold more selective for AKR1C1 relative to the structurally similar 3α-hydroxysteroid dehydrogenase (AKR1C2). Furthermore, the compound inhibited the metabolism of progesterone in AKR1C1-overexpressed cells with an IC₅₀ value equal to 100 nM.

Full text of DOI:10.1016/j.ejmech.2010.08.052

European Journal of Medicinal Chemistry 45 (2010) 5309e5317
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Structure-based optimization and biological evaluation of human
20 -hydroxysteroid dehydrogenase (AKR1C1) salicylic acid-based inhibitors
a
,
a
a
a
b
Ossama El-Kabbani a , Peter J. Scammells , Tom Day , Urmi Dhagat , Satoshi Endo ,
*
Toshiyuki Matsunaga b, Midori Soda b, Akira Hara b
a Medicinal Chemistry and Drug Action, Monash Institute of Pharmaceutical Sciences, Monash University, 381 Royal Parade, Parkville, Victoria 3052, Australia
b Laboratory of Biochemistry, Gifu Pharmaceutical University, Daigaku-Nishi, Gifu 501-1196, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
The tertiary structure of the Leu308Val mutant of human 20a-hydroxysteroid dehydrogenase (AKR1C1)
Received 22 June 2010
Received in revised form
8 August 2010
Accepted 24 August 2010
Available online 17 September 2010
in complex with the inhibitor 3,5-dichlorosalicylic acid (DCL) has been determined. Structures and
kinetic properties of the wild-type and mutant enzymes indicate that Leu308 is a selectivity determinant
for inhibitor binding. The Leu308Val mutation resulted in 13-fold and 3-fold reductions in the inhibitory
potencies of DCL and 3-bromo-5-phenylsalicylic acid (BPSA), respectively. The replacement of Leu308
with an alanine resulted in 473-fold and 27-fold reductions in the potencies for DCL and BPSA,
respectively. In our attempts to optimize inhibitor potency and selectivity we synthesized 5-substituted
3-chlorosalicylic acid derivatives, of which the most potent compound, 3-chloro-5-phenylsalicylic acid
Keywords:
Aldo-keto reductase
3-Bromo-5-phenylsalicylic acid
Cancer
(Ki ¼ 0.86 nM), was 24-fold more selective for AKR1C1 relative to the structurally similar 3
a-hydrox-
ysteroid dehydrogenase (AKR1C2). Furthermore, the compound inhibited the metabolism of proges-
terone in AKR1C1-overexpressed cells with an IC50 value equal to 100 nM.
Ó 2010 Elsevier Masson SAS. All rights reserved.
Drug design
Human 20
Human 3
a
-hydroxysteroid dehydrogenase
a
-hydroxysteroid dehydrogenase
1. Introduction
enzyme scaffold adaptable to cofactor and/or substrate induced
conformational changes [5,7,8]. The cofactor-binding and active
Hydroxysteroid dehydrogenases (HSDs) catalyze the synthesis
and metabolism of all classes of steroid hormones and play a pivotal
role in regulating nuclear receptor signaling by controlling the
concentration of active steroid hormones at the pre-receptor level
[1,2]. HSDs are broadly classified into two protein superfamilies, the
short-chain dehydrogenase/reductase and the aldo-keto reduc-
taseo (AKR) [3,4]. Members of the AKR superfamily are NAD(P)
(H)-dependent oxidoreductases that adopt a TIM-barrel motif
sites are conserved across all AKRs where the active site is
comprised of a catalytic tetrad which forms an oxyanion pocket
along with the nicotinamide ring of the cofactor via a hydrogen
bonding network. The reactions catalyzed by AKRs follow an
ordered biebi kinetic mechanism where the cofactor binds first and
leaves last [5,9,10]. Structural studies have shown that significant
conformational changes occur in the AKRs upon cofactor-binding
resulting in a tight binary complex and a mature substrate binding
site [8,11,12].
consisting of a cylindrical core of eight stranded
rounded by eight -helices running antiparallel to the
b
-sheet sur-
a
b
-sheet [5,6].
There are four human NADP(H)-dependent HSDs belonging to
While the TIM-barrel motif is well conserved within the family
members, it is interrupted by a number of loops and helices of
varying lengths. These loops provide the structural variations
required for binding chemically diverse substrates and make the
the AKR1C subfamily, namely AKR1C1 (20
-HSD), AKR1C3 (type 2 3 -HSD and type 517
(type 1 3 -HSD), which share over 86% amino acid sequence
a
-HSD), AKR1C2 (type 3
3a
a
b
-HSD) and AKR1C4
a
homology [1,2,8,13]. The AKR1C isoforms have different expression
patterns in human tissues, where they regulate ligand occupancy
and trans-activation of nuclear receptors by controlling the
concentrations of active steroid hormones such as androgens,
estrogens and progestins in target tissues [2,13e15]. Although the
AKR1C isoforms catalyze the oxidation and reduction reactions
with comparable efficiencies, they thermodynamically favor the
ketosteroid reduction in vitro [16]. The direction of catalysis in vivo
is also governed by the cellular ratio of NADPH/NADPþ. In
Abbreviations: HSD, hydroxysteroid dehydrogenase; AKR, aldo-keto reductase;
AKR1C1, human 20
hydroxysteroid dehydrogenase; S-tetralol, S-(þ)-1,2,3,4-tetrahydro-1-naphthol;
GABAA, -aminobutyric acid type A; BPSA, 3-bromo-5-phenylsalicylic acid; DCL,
a-hydroxysteroid dehydrogenase; AKR1C2, human type 3 3a-
g
3,5-dichlorosalicylic acid; WT, wild-type; MT, mutant.
* Corresponding author. Tel.: þ61 3 9903 9691; fax: þ61 3 9903 9143.
E-mail address: ossama.el-kabbani@monash.edu (O. El-Kabbani).
0223-5234/$ e see front matter Ó 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2010.08.052

5310
O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
metabolically active cells, reduction is favored over oxidation since
NADPH is the predominant form of the cofactor and the dehydro-
genase activities of the four isoforms are potently inhibited by
NADPH [17,18]. Although the four AKR1C isoforms have overlapping
catalytic properties [15], they display distinct positional and stereo
preferences for steroid substrates. Differences in affinity for the 3-,
17- or 20-ketosteroid substrates and mode of steroid binding in the
active site among the four AKR1C isoforms are dictated by residues
lining their substrate binding pockets [2,8,13].
the inhibitor binding site. In addition, the effect of shortening of the
Leu308 side-chain on inhibitor binding was investigated by
molecular modeling of DCL and BPSA in wild-type (WT) and MT
AKR1C1s. Optimization of AKR1C1 inhibitors was carried out where
the most potent compound, 3-chloro-5-phenylsalicylic acid
(Ki ¼ 0.86 nM), was 24-fold more selective for AKR1C1 over
AKR1C2. Furthermore, the compound potently inhibited the
metabolism of progesterone by AKR1C1 in the cells with an IC50
value of 100 nM.
Although AKR1C1 exhibits both 3
preferentially acts as 20-ketosteroid reductase converting
progesterone and its metabolites (5 -dihydroprogesterone and
,5 -tetrahydroprogesterone) into the corresponding 20
hydroxysteroids [17,19e22]. Since the metabolite 20 -hydrox-
yprogesterone has a low affinity for the progesterone receptors,
AKR1C1 plays an important role in controlling the cellular
concentration of progesterone, which is an essential hormone
required for endometrial development and maintenance of preg-
nancy [23]. Progesterone is a natural hormone, synthesized in both
males and females, and also affects neuronal function by modu-
lating gene expression and cellular receptors abundant in the
central nervous system [24]. Progesterone and/or its above
metabolites exert neuroprotective effects in traumatic and ischemic
brain injury [24], nerve crush injury [25], peripheral neuropathy
induced by an antineoplastic agent docetaxel [26], and diabetic
a/b- and 20a-HSD activities, it
a
2. Results and discussion
a
3
a
a
a-
2.1. Crystal structure of AKR1C1 Leu308Val MT
a
Crystals of the Leu308Val MT in ternary complex with NADPþ
and DCL were obtained by optimizing the salt and precipitant
concentrations of the crystallization buffer that was used to crys-
tallize the WT AKR1C1 ternary complex [39]. The final concentra-
tions of zinc sulfate and polyethylene glycol monomethyl ether 550
(PGMME) were reduced from 20 mM to 10 mM and from 25% (v/v)
to 15% (v/v) respectively, while keeping the pH (6.5) and the final
concentration of MES buffer (0.1 M) unchanged. The asymmetric
unit contained one molecule of the Leu308Val MT, one molecule of
the cofactor NADPþ, one molecule of the inhibitor DCL, 396 solvent
molecules, and a zinc ion. The zinc ion from the crystallization
buffer was present in coordination with a water molecule, the side-
chain of Glu133, and the side-chains of Glu292 and His248 from the
neuropathy [27]. In addition, 3a,5a-tetrahydroprogesterone is
a potent positive modulator of the GABAA receptors in the brain
[28]. Since AKR1C1 efficiently reduces progesterone and its
metabolites, and is ubiquitously expressed in human tissues, its
inhibition would lead to elevating the concentrations of neuroac-
tive steroids. On the other hand, the expression of AKR1C1 was
reported to be up-regulated in ovarian endometriosis [29], endo-
metrial and breast cancer [30,31], and cancer cells derived from
lung and colon [32,33]. Recent work has shown that the enzyme
plays a crucial role in induction of neoplastic transformation of
NIH3T3 cells [34]. Thus, AKR1C1 is an interesting target for the
development of not only new drugs for the treatments of the
neurological disorders and endometriosis, but also novel anticancer
drugs with pain relief effect through the activation of the GABAA
receptors.
Several classes of AKR1C1 inhibitors have been identified,
including, benzodiazepines, benzofurans, phenolphthalein deriva-
tives, benzoyl benzoic acids, steroid carboxylates and N-phenyl-
anthranilic derivatives [21,22,35e37]. More recently, we have
designed 3-bromo-5-phenylsalicylic acid (BPSA) as a selective and
potent inhibitor of AKR1C1 with a Ki value of 4 nM based on the
crystal structure of the enzyme in ternary complex with a lead
compound [38,39]. Since members of the AKR1C subfamily share
high sequence homology and show overlapping substrate speci-
ficity, it is important to identify selectivity determinants to aid the
development of inhibitors that are specific for AKR1C1. Structural
analysis of the inhibitor binding site in the AKR1C1 ternary complex
with 3,5-dichlorosalicylic acid (DCL) bound facilitated the identi-
fication of nonconserved residues (Leu54, His222, Leu306, and
Leu308) involved in hydrogen bonding and van der Waals inter-
actions with the inhibitor. While the amino acid residue Leu308 in
AKR1C1 makes van der Waals contacts with both DCL and BPSA, in
the other AKR1C isoforms with the exception of AKR1C2 this
interaction is not conserved [38,39].
Table 1
Data collection and refinement statistics.
Space group
P21
Cell dimensions
a ¼ 39.41 Å
b ¼ 83.89 Å
c ¼ 48.91 Å
b
¼ 91.0ꢀ
Radiation source
Wavelength (Å)
Rotating anode
1.54178
Diffraction dataa
Resolution (Å)
Number of observed reflections
Number of unique reflections
Redundancy
Completeness (%)
I/s(I)
Rmerge (%)
30e1.9
24030 (2377)
22733 (2237)
4.0 (4.8)
94.6 (94.1)
11.7 (2.7)
4.9 (36.0)
Refinement statistics
Resolution (Å)
Protein residues
Solvent molecules
Zinc ion
Cofactor
Inhibitor
Rfree (%)
Rcryst (%)
30e1.9
318
396
1
1
1
26.0
19.4
RMSDs
Bonds (Å)
Angles (ꢀ)
0.022
2.0
Ramachandran plot
Residues in most favored regions (%)
Residues in allowed regions (%)
93.3
6.3
Estimated coordinated error
Luzzati mean coordinate error (Å)
0.244
In an effort to investigate the role of Leu308 in inhibitor binding
we have examined the effects of the Leu308Val and Leu308Ala
mutations in AKR1C1 on the potency of inhibition. A crystal
structure of the Leu308Val mutant (MT) enzyme in ternary
complex with NADPþ and DCL was determined to identify any
structural changes that may have been caused by the mutation in
Mean B factors (Å2)
Protein
26.0
19.3
27.5
36.4
NADPþ
Inhibitor
Water molecules
a
Statistics for the highest resolution shell (1.97e1.9 Å) are shown in parentheses.

O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
5311
Table 2
neighboring symmetry-related molecule. There were 319 amino
Effects of the mutations of Leu308 in AKR1C1 on Ki values for DCL and BPSA.
acids in the final structure that fitted into the electron density map,
with the exception of five residues at the N-terminal end. As the
electron density was poorly defined for these residues due to
disorder in the N-terminus they were omitted from the structural
model. The structure was refined to a crystallographic R-factor of
0.194 (Rfree ¼ 0.260) and during the refinement process, double
conformations were assigned to residues Ile49 and Cys242. The
root mean square deviations (RMSD) of bond lengths and angles of
the refined model were 0.022 Å and 2.0ꢀ, respectively. A Ram-
achandran plot of the backbone torsions had 93.3% of the residues
in the most favored region and 6.3% of the residues in the addi-
tionally allowed regions [40]. Only Ser221 was in the disallowed
region due to the hydrogen bond between its main-chain and the
cofactor molecule. A summary of the data collection and refine-
ment statistics is presented in Table 1.
Clear electron density allowed unambiguous fitting of Val308
and the inhibitor molecule into the respective difference Fourier
maps (FoꢁFc and 2FoꢁFc). The RMSD of the MT structure from the
WT AKR1C1 was 0.2 Å. As shown in Fig. 1, the Leu308Val mutation
did not affect the surrounding residues of the active site except for
the side-chain of Phe311 where a change in the torsion angle of the
phenyl ring was noticed resulting from the additional interaction
with the side-chain of Val308. The key hydrogen bonding interac-
tions between the inhibitor and WT enzyme were conserved in the
Leu308Val MT structure. The inhibitor molecule was anchored from
its carboxylate group that formed hydrogen bonds with the cata-
lytic residues His117 (2.8 Å) and Tyr55 (2.5 Å), while the hydroxyl
group was hydrogen bonded to His222 (3.0 Å). Van der Waals
contacts were present between the inhibitor and residues Leu54,
Leu306, and Phe311. Unlike the WT enzyme, the side-chain of the
mutated Val308 made long contacts with the inhibitor DCL, the
closest point of contact being 3.9 Å.
Enzyme
Ki (nM)a
DCL
BPSA
WT
Leu308Val
Leu308Ala
5.9 ꢂ 0.8
4.1 ꢂ 0.4
78 ꢂ 9 (13)
14 ꢂ 2 (3)
2790 ꢂ 300 (473)
113 ꢂ 12 (27)
a
The inhibition patterns were competitive with respect to the substrate
S-tetralol. Ki ratio of MT/WT is given in parenthesis.
selectivity [38,39]. In the case of BPSA, an inhibitor designed based
on the structure of AKR1C1 in ternary complex with NADPþ and
DCL, its phenyl group targets a nonconserved hydrophobic pocket
in the active site of the enzyme lined by residues Leu54, Leu308 and
Phe311, resulting in a 21-fold improved potency (Ki ¼ 4 nM) over
the structurally similar AKR1C2 [38]. In this study, the contribution
of Leu308 in inhibitor binding to AKR1C1 was investigated by
mutating Leu308 to valine and alanine residues, and determining
the effects of the mutations on the binding constants (Ki values) for
DCL and BPSA. The Leu308Val mutation resulted in a 13-fold
increase in the Ki value for DCL and a 3-fold increase in the Ki value
for BPSA compared to the WT, while in the case of the Leu308Ala
MT the Ki values for DCL and BPSA increased by 473-fold and 27-
fold, respectively (Table 2).
The above results illustrate the role of the bulky side-chain of
Leu308 in inhibitor binding through the formation of important
van der Waals contacts with the DCL and BPSA molecules (Fig. 2).
The effect resulting from shortening the side-chain of Leu308 on
inhibitor binding was reflected in the Ki values where the inhibition
potency decreased for the mutants as the side-chain shortened
from leucine to valine to alanine. In the Leu308Ala MT the alanine
side-chain was not present within van der Waals contacts from the
inhibitor, resulting in higher Ki values than those of the Leu308Val
MT. Furthermore, the Ki values for DCL in comparison with those for
BPSA, showed that the mutations of Leu308 had a lesser impact on
the binding of BPSA than for DCL. This effect is likely due to the
additional van der Waals contacts made by the phenyl ring of BPSA
2.2. The role of Leu308 in inhibitor binding
Structural and kinetic studies on the binding of the inhibitors
DCL and BPSA to the AKR1C isoforms suggested that the interac-
tions between the inhibitor and the four nonconserved amino acid
residues at positions 54, 306, 308, and 311 confer inhibitor
Fig. 1. Superimposition of the crystal structures of the WT AKR1C1 (yellow) and the
Leu308Val MT (purple) with DCL bound in the active site. Shortest distance between
DCL and residue Val308 (orange dotted lines) and hydrogen bonds with the catalytic
residues (black dotted lines) are given in angstroms. (For interpretation of the refer-
ences to colour in this figure legend, the reader is referred to the web version of this
article).
Fig. 2. Docked inhibitors DCL (green) and BPSA (grey) in the active site of AKR1C1
Leu308Ala MT (cyan) with Leu308 WT (grey) and Leu308Val MT (purple) super-
imposed. Both inhibitors formed conserved hydrogen bonds with the catalytic residues
(black dotted lines). The 5-phenyl ring of BPSA extended deeper into the binding site
interacting with Phe311 (green dotted lines). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article).

5312
O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
Scheme 1. Synthesis scheme for prepration of compounds 5aec and 5eei. Reagents and conditions: (i) MeI, K2CO3, acetone, 96%; (ii) SO2Cl2, CH2Cl2, 96%; (iii) ReB(OH)2,
PdCl2(PPh3)2, K3PO4, DMF (for 4aed and 4fei); (iv) BBr3, CH2Cl2.
which extended deeper into the selectivity pocket interacting with
the phenyl ring of Phe311 (Fig. 2).
methoxybiphenyl (6) was O-demethylated using boron tribromide
to give the phenol 7. A Duff reaction, which introduces a formyl
group using hexamethylenetetramine (HMT), afforded the salicy-
laldehyde 8 which was subsequently oxidized to 3-fluoro-5-phe-
nylsalicylic acid (9) (Scheme 2).
2.3. Synthesis of 5-substituted 3-chlorosalicylic acid derivatives
In our attempts to optimize inhibitor potency and selectivity, 5-
substituted 3-chlorosalicylic acid derivatives were prepared using
the synthetic approach depicted in Scheme 1. 5-Iodosalicylic acid
(1) was protected and chlorinated in high yield to give the key
intermediate 3. A range of phenyl substituents were then intro-
duced into the 5-position via SuzukieMiyaura coupling with the
appropriate phenyl boronic acid (compounds 4aed and 4fei).
The final step in the synthesis was the simultaneous deprotection
of the methyl ester and methyl ether moieties using boron tri-
bromide, as reported in our earlier study [38]. In general the cor-
responding salicylic acid derivative 5 was formed in good yield.
However, the BBr3 mediated deprotection of the 5-(4-ethylphenyl)
substituted derivative 4d produced an unexpected product. In this
case, the major product was 5-chloro-4-hydroxy-40-(1-methox-
yethyl)biphenyl-3-carboxylic acid (5e), which presumably resulted
from benzylic bromination and methoxy substitution of the 5-(4-
ethylphenyl) group (Scheme 1).
2.4. Structureeactivity relationships
We have previously reported that 3,5-dihalosalicylic acids are
potent AKR1C1 inhibitors, with the 3,5-dichloro analog (i.e. DCL)
being the most potent in this series [38,39]. A 5-phenyl substituent
was introduced to target a selectivity pocket in the active site of
AKR1C1 lined by the three non-polar residues Leu54, Leu308, and
Phe311 [38]. BPSA proved to be a potent inhibitor with a Ki value of
4.1 nM and 21-fold selectivity for the AKR1C1 relative to AKR1C2
(Table 3), where the selectivity pocket differs by one amino acid
residue (Leu54 in AKR1C1 and Val54 in AKR1C2).
In the present study, the combination of a 3-chloro and
a 5-phenyl group (5a) afforded a substantial increase in potency
(Ki ¼ 0.86 nM) with similar level of selectivity compared to BPSA.
The effect of further substitution of the 5-phenyl moiety of 5a was
subsequently explored through the evaluation of the salicylic acid
derivatives 5b, 5c and 5eej. Analogs with 3-methylphenyl (5b),
4-methylphenyl (5c), 4-butylphenyl (5f) and 4-isobutylphenyl
(5g) groups in the 5-position all proved to be potent AKR1C1
3-Fluoro-5-phenylsalicylic acid (9) was prepared using the
synthetic approach shown in Scheme 2. This synthesis was based
on the approach developed by Micklatcher and Cushman for the
preparation of 3-fluorosalicylic acid [41]. Briefly, 3-fluoro-4-
Table 3
Ki values of DCL, BPSA and the 5-substituted 3-chlorosalicylic acids towards AKR1C1
and AKR1C2.
Compound
R
X
Ki (nM)a
AKR1C2
Ratiob
AKR1C1
DCL
BPSA
5a
5b
5c
5e
5f
5g
5h
5i
Cl
Ph
Ph
Cl
Br
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
F
5.9 ꢂ 0.8
4.1 ꢂ 0.4
0.86 ꢂ 0.01
1.3 ꢂ 0.1
2.6 ꢂ 0.1
340 ꢂ 23
2.1 ꢂ 0.2
2.0 ꢂ 0.1
96 ꢂ 5
70 ꢂ 2.0
11
21
24
13
24
9
7
15
5
87 ꢂ 12
21 ꢂ 3
3-MePh
4-MePh
4-CH3CH(OMe)Ph
4-BuPh
4-IsoBuPh
4-tert-BuPh
4-CF3OPh
PhC^C
Ph
17 ꢂ 1
63 ꢂ 4
3000 ꢂ 120
14 ꢂ 1
29 ꢂ 2
470 ꢂ 17
26 ꢂ 2
29 ꢂ 2
0.9
3
1.2
5j
9
64 ꢂ 10
168 ꢂ 40
1.5 ꢂ 0.2
1.3 ꢂ 0.1
a
Scheme 2. Synthesis scheme for preparation of compound 9. Reagents and conditions:
(i) BBr3, CH2Cl2, 88%; (ii) hexamethylenetetramine, TFA, H2SO4, 41%; (iii) NaClO2, sul-
famic acid, NaH2PO4$H2O, dioxane/H2O, 61%.
The inhibition patterns were competitive with respect to the substrate S-
tetralol. Ratio represents AKR1C2/AKR1C1.
b
Ratio represents AKR1C2/AKR1C1 Ki values.

O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
5313
inhibitors which exhibited Ki values less than 3 nM. Interestingly,
the 5-(3-methylphenyl)salicyclic acid 5b proved to be twice as
potent as the corresponding 4-methylphenyl analog 5c, but was
only around half as selective versus AKR1C2. Substitution of the 5-
phenyl moiety with a bulky 4-tert-butyl group (5h), 4-ethyl-
methoxy group (5e) or a more polar 4-trifluoromethoxy group (5i)
significantly reduced potency. Finally, replacement of the Cl atom
of 3-chloro-5-phenylsalicylic acid (5a) with a fluorine atom failed
to improve potency (compound 9; Ki ¼ 1.3 nM) and essentially
abolished selectivity.
and examined the effects of the Leu308Val and Leu308Ala muta-
tions on the potency of inhibitor. The effect of shortening of the
Leu308 side-chain on inhibitor binding was also investigated by
modeling the inhibitors DCL and BPSA in the WT and MT AKR1C1
structures. The modeling results were in agreement with the Ki
values where the inhibitor potency for the enzyme decreased as the
side-chain shortened from leucine to valine to alanine. Attempts to
optimize the inhibitors were then carried out by preparing
5-substituted 3-chlorosalicylic acid derivatives. The most potent
compound in the series, 3-chloro-5-phenylsalicylic acid
(Ki ¼ 0.86 nM), was 24-fold more selective for AKR1C1 over the
structurally similar AKR1C2, and potently inhibited the metabolism
of progesterone by AKR1C1 in the cells with an IC50 value equal to
100 nM. On the other hand, the replacement of the Cl atom of
3-chloro-5-phenylsalicylic acid with a F atom failed to improve
potency and essentially abolished selectivity.
2.5. Evaluation of inhibitory potency at cellular level
The inhibitory potency of compounds 5a and 9 was evaluated
using bovine aortic endothelial cells that were transfected with the
cDNA for AKR1C1. The cells exhibited 10-fold higher S-tetralol
dehydrogenase activity than that of the control cells transfected
with the vector alone, and only the AKR1C1-overexpressed cells
4. Experimental section
efficiently metabolized progesterone into 20a-hydroxyprogester-
4.1. Site-directed mutagenesis and purification of recombinant
enzymes
one in the metabolic analysis using liquid chromatography/mass
spectrometry, as described previously for BPSA [38]. Compounds
5a and 9 inhibited the metabolism of progesterone in the cells, and
their dose-response curves are shown in Fig. 3. Compound 5a
showed 92% inhibition at 2 mM and 26% inhibition at 0.01 mM, and
its IC50 value was 100 nM, while the IC50 value for compound 9 was
Mutagenesis was performed using a QuickChange site-directed
mutagenesis kit (Stratagene) and pGEX-2T expression plasmid
harbouring cDNA for AKR1C1 as the template [42], according to the
protocol described by the manufacturer. The primer pair used for
the mutagenesis was composed of sense and antisense oligonu-
cleotides to alter a codon of the AKR1C1 cDNA. The 30-mer primers
were synthesized to give the Leu308Val and Leu308Ala mutations.
The coding regions of the cDNAs in the expression plasmids were
sequenced by using a Beckman CEQ2000XL DNA sequencer in order
to confirm the presence of the desired mutation and ensure that no
other mutation had occurred. The recombinant WT and MT
AKR1C1s and AKR1C2 were expressed in Escherichia coli JM109 and
purified to homogeneity as previously described [42,43].
300 nM. It should be noted that the two compounds did not affect
the cell viability up to their concentrations of 10 mM. The IC50 values
determined at cellular level for the two compounds are lower than
that (460 nM) of the previously synthesized inhibitor, BPSA [38],
correlating with the Ki values determined in vitro using AKR1C1
(Table 3). This indicates that the difference in the type of halogen
atom at the 3-position of the 5-phenylsalicylic acid affects the
affinity for the enzyme rather than cell permeability. Thus, the
presence of the Cl atom at this 3-position is an important structural
requisite for the 5-phenylsalicylic acid as
inhibitor.
a potent AKR1C1
4.2. Assay of enzyme activity
3. Conclusions
The dehydrogenase activities of WT and MT enzymes were
determined at 25 ꢀC by measuring the rate of change in NADPH
fluorescence (at 455 nm with an excitation wavelength of 340 nm)
[21]. The inhibitor was dissolved in methanol and added to the
reaction mixture, in which the final concentration of methanol was
less than 2.5%. When the fluorescence due to the high concentra-
tion of inhibitor interfered with the fluorometric assay, the enzyme
activity was determined by measuring the rate of change in NADPH
absorbance at 340 nm. The inhibition pattern was determined by
fitting the initial velocities using five substrate concentrations
(0.5e5 ꢃ Km) in the presence of the three inhibitor concentrations
to LineweavereBurk and Dixon plots. The Ki value was calculated by
using the appropriate programs of the ENZFITTER (Biosoft, Cam-
bridge, UK) and is expressed as the mean ꢂ standard error of at
least three determinations.
In summary, in our efforts towards the development of potent
AKR1C1 inhibitors we have determined the crystal structure of the
Leu308Val MT enzyme in ternary complex with NADPþ and DCL,
4.3. Crystallization
Prior to crystallization, AKR1C1 was concentrated to 21 mg/mL
in 20 mM HEPES buffer (pH 7.5) with 5 mM 2-mercaptoethanol. A
small sample of the enzyme was mixed with NADPþ and the
inhibitor dissolved in DMSO (molar ratio of AKR1C1/NADPþ/
inhibitor was 1:3:3). The final concentration of DMSO did not
exceed 2% of the total sample volume. The AKR1C1/NADPþ/inhib-
itor solution was mixed with an equal volume of the crystallization
buffer (15% (v/v) polyethylene glycol monomethyl ether 550,
0.01 M zinc sulfate in 0.1 M MES buffer (pH 6.5)). The ternary
Fig. 3. Dose-response curves for compounds 5a and 9 in the inhibition of progesterone
metabolism by the AKR1C1-overexpressed bovine aortic endothelial cells. The cells
were cultured for 6 h in the medium containing 30 mM progesterone in the absence
(control) or presence of compounds 5a (C) and 9 (B), and the product 20a-hydrox-
yprogesterone in the media was determined.

5314
O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
complex was crystallized using the hanging drop vapor-diffusion
method where 3 L droplets of the enzyme-buffer solution were
placed above a well containing 1 mL of the crystallization solution.
Rectangular crystals appeared at 295 K within one week and grew
to maximum dimensions of 0.3 ꢃ 0.1 ꢃ 0.05 mm. The inhibitor DCL
(97% purity) was obtained from the SigmaeAldrich Chemical Co.
plotting was managed using XWINPLOT software running on the
Silicon Graphics workstation. High resolution mass spectra (HRMS)
were obtained on Waters LCT Premier XE time-of-flight spectrom-
eter fitted with an electrospray (ESI) ion source and controlled with
MassLynx software (version 4.1). Analytical purity was assessed
using a Waters 2960 Separation Module coupled with a Waters 998
Photodiode Array Detector with a Phenomax Luna 5u C8 (2) 100A
(150 ꢃ 4.60 nM ID) column. Samples were run over a gradient of
20e100% Buffer C (20% H2O, 80% MeOH) in Buffer A (0.1% TFA, 99.9%
H2O) for 10 min followed by isocratic 100% Buffer A for 2 min. In all
cases purity was >95%. Thin-layer chromatography was performed
on Merck Silica Gel 60 F254 plates. 3-Fluoro-4-methoxybiphenyl (6)
was prepared using the methods described by Mewshaw and co-
workers [49].
m
4.4. X-ray data collection and structure determination
Crystals of the AKR1C1 Leu308Val MT ternary complex were
briefly soaked in the cryo-protecting buffer comprising 0.1 M MES
(pH 6.5), 0.01 M zinc sulfate, 20% (v/v) polyglycol monomethyl
ether 550 and 30% (v/v) glycerol. The diffraction data were collected
at 100 K on a MAR-345 image plate system mounted on a Rigaku
RU300 rotating anode generator operating at 50 kV and 90 mA. The
X-ray diffraction data were collected to a resolution of 1.9 Å and
processed using HKL2000 and SCALEPACK [44]. The structure was
solved by the molecular replacement method using the program
MOLREP in the CCP4 suite of crystallographic software [45], and the
atomic coordinates of AKR1C1 (PDB code 3C3U) without the
inhibitor, cofactor and water molecules were used as the search
model [39]. The initial model was subjected to iterative cycles of
manual fitting into 2FoꢁFc and FoꢁFc electron density maps using
Coot [46], followed by structural refinement using REFMAC [47]. The
inhibitor molecule, cofactor, waters and zinc ion were added
towards the end of the refinement. The atomic coordinates were
deposited in the Protein Data Bank (ID code 3GUG).
4.6.1. Methyl 5-iodo-2-methoxybenzoate (2)
5-Iodosalicylic acid (5.00 g, 18.93 mmol) was dissolved in
acetone (100 mL) and then K2CO3 (10.47 g, 75.72 mmol) and MeI
(10.75 g, 75.72 mmol) were added. The mixture was refluxed for
48 h, cooled and filtered through Celite. The solvent was removed
in vacuo and the residue was redissolved in EtOAc (25 mL), washed
with 10% NaHCO3 (15 mL), H2O (2 ꢃ 15 mL) and brine (10 mL), and
then dried over MgSO4. The solvent was then evaporated to yield 2
as white crystals (5.32 g, 96%); mp 55e57 ꢀC; 1H NMR
J ¼ 2.4 Hz), 7.71 (d, 1H, J¼2.4, 8.7 Hz), 6.73 (d, 1H, J ¼ 8.7 Hz), 3.68 (s,
6H); 13C NMR
165.2,159.1, 142.1,140.1, 122.3, 114.5, 81.9, 56.3, 52.4.
d 8.05 (d, 1H,
d
4.6.2. Methyl 3-chloro-5-iodo-2-methoxybenzoate (3)
4.5. Molecular docking
Compound 2 (2.00 g, 6.84 mmol) was dissolved in CH2Cl2
(10 mL) and then SO2Cl2 (2.20 mL, 27.39 mmol) was added over
a 5 min period. The solution was then heated to 50 ꢀC and stirred at
this temperature for 15 h. The resulting mixture was then cooled,
diluted with H2O (200 mL) and extracted with CH2Cl2 (3 ꢃ 20 mL).
The combined organic layers were washed with brine (10 mL),
dried over MgSO4 and concentrated in vacuo. The resulting residue
was purified by flash chromatography (0e10% EtOAc/petroleum
spirit) to yield beige crystals (2.15 g, 96%); mp 53e54 ꢀC; 1H NMR
The coordinates for AKR1C1 were obtained from the RCSB
Protein Data Bank (ID code 3C3U code). The structure was prepared
using the Maestro (Shrödinger, LLC, Portland, OR) software package
Version 8.5 where hydrogen atoms in the structure were generated.
The Protein Preparation module in Maestro was used to perform
a brief relaxation on the starting structure with the ‘‘Refinement
Only” option, which optimizes the hydroxyl and thiol torsions as
well as performing an all-atom constrained minimization to relieve
any clashes. The formal charges and appropriate bond orders of the
ligands were manually adjusted in Maestro. In order to eliminate
any potential bond length and bond angle biases in the structure,
the ligands were subjected to a full minimization prior to the
docking. The docking calculations were performed using Glide 5.0
on a Linux workstation [48]. A grid box was generated, where the
centre of the grid box was defined by the centre of the bound
ligand. The calculations were run in the ‘‘Extra Precision” (XP)
mode of Glide where the ligands were docked flexibly, and to soften
the potential for the non-polar parts of the ligands, the van der
Waals radius was scaled by 0.8. In order to ensure that the poses
generated were conformationally distinct, as well as increasing the
diversity of the retained poses, poses with an RMSD less than 0.5 Å
and a maximum atomic displacement less than 1.3 Å were dis-
carded as duplicates. All figures for the structural models were
prepared by PyMOL (DeLano Scientific, San Carlos, CA, USA).
d
7.93 (d, 1H, J ¼ 3.3 Hz), 7.79 (d, 1H, J ¼ 3.3 Hz), 3.96 (s, 6H); 13C
NMR
d 164.2, 155.8, 143.0, 138.4, 130.7, 128.3, 86.2, 62.0, 52.6.
4.6.3. General procedure for SuzukieMiyaura coupling of 3 to
phenyl derivatives 4aei
To a stirred solution of 3 (0.200 g, 618 mmol) in anhydrous DMF
(5 mL) were added K3PO4 (0.212 g, 1.22 mmol) and the desired
boronic acid (1.85 mmol). This solution was then purged with N2 for
1 h, after which PdCl2(PPh3)2 (0.043 g, 0.06 mmol) was added and
the reaction was stirred at 70 ꢀC under N2 overnight. Once the
starting material had been consumed (monitored by TLC), the
reaction mixture was filtered through SiO2 by the aid of diethyl
ether (25 mL). The organic layer was washed with H2O (5 ꢃ 50 mL),
brine (20 mL) and dried over MgSO4. The solvent was removed in
vacuo and the residue was purified via flash chromatography
(0e10% EtOAc in petroleum ether) to yield the desired product.
4.6.3.1. Methyl 5-chloro-4-methoxybiphenyl-3-carboxylate (4a). Co
mpound 4a was prepared using phenyl boronic acid (0.224 g) and 3
4.6. Chemistry
(0.200 g) as an amorphous solid (144 mg, 85%). 1H NMR
d
7.92 (d,
1H, J ¼ 2.4 Hz), 7.77 (d, 1H, J ¼ 2.4 Hz), 7.35e7.56 (m, 2H) 7.41 (m,
3H) 3.98 (s, 3H, CH3) 3.96 (s, 3H, CH3); 13C NMR
166.0, 155.1, 138.6,
138.0, 132.6, 130.1, 129.2, 128.5, 128.2, 127.1, 62.2, 52.7.
Melting points were determined using a Barnstead electro-
thermal 9100 variable melting point apparatus and are uncorrected.
1H NMR spectra were recorded on a Bruker Avance 300 WB spec-
trometer. Unless otherwise stated,1H NMR spectra were obtained in
CDCl3 at 300 MHz 13C NMR were recorded on a Varian 600 MHz
Inova NMR spectrometer. Unless otherwise stated, 13C NMR spectra
were obtained in CDCl3 at 151.4 Hz. Data acquisition and processing
was managed using XWINNMR software package version 3.5 and
d
4.6.3.2. Methyl
5-chloro-4-methoxy-30-methylbiphenyl-3-carbox-
ylate (4b). Compound 4b was prepared from 3-methylphenyl
boronic acid (0.250 g) and 3 (0.200 g) as a brown oil (120 mg, 69%).
1H NMR
d
7.91 (d, 1H, J ¼ 2.4 Hz), 7.76 (d, 1H, J ¼ 2.4 Hz), 7.16e7.36

O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
5315
(m, 4H), 3.98 (s, 3H), 3.96 (s, 3H), 2.42 (s, 3H); 13C NMR
155.0, 138.8, 138.6, 138.1, 132.6, 130.0, 129.1, 128.9, 128.4, 127.8,
127.02, 124.1, 62.2, 52.7, 21.7.
d
166.0,
The aqueous layer was then collected and acidified to pH 2. The
aqueouslayerwasthenextractedwithEtOAc (3ꢃ15mL). Theorganic
layers were combined and dried over Na2SO4. The solvent was
removed in vacuo to yield the deprotected product 5.
4.6.3.3. Methyl
5-chloro-4-methoxy-40-methylbiphenyl-3-carbox-
ylate (4c). Compound 4c was prepared from 4-methylphenyl
boronic acid (0.250 g) and 3 (0.200 g) as a brown oil (125 mg, 70%).
4.6.4.1. 5-Chloro-4-hydroxybiphenyl-3-carboxylic acid (5a). Compo
und 5a was obtained as white solid (88 mg, 69%); mp 200e205 ꢀC
1H NMR
d
7.89 (d, 1H, J ¼ 2.4 Hz), 7.75 (d, 1H, J ¼ 2.4 Hz), 7.45 (d, 2H,
(dec.); 1H NMR (CD3OD)
d
7.96 (d, 1H, J ¼ 2.3 Hz), 7.75 (d, 1H,
J ¼ 7.9 Hz) 3.97 (s, 3H), 3.95 (s, 3H), 1.58 (s, 3H). 13C NMR
d
165.9,
J ¼ 2.3 Hz), 7.22e7.49 (m, 5H); 13C NMR (CD3OD)
d 173.4, 158.4,
154.8, 138.0, 137.9, 135.6, 132.2, 129.9, 129.8, 128.1, 127.0, 126.8, 62.1,
52.6, 21.2.
140.3, 135.0, 134.2, 130.3, 128.8, 128.3, 127.7, 123.6, 115.8; HRMS
(ESI) m/z calcd for C13H9ClO3 [M ꢁ H]ꢁ 247.0167, found 247.0173.
4.6.3.4. Methyl 5-chloro-40-ethyl-4-methoxybiphenyl-3-carboxylate
(4d). Compound 4d was prepared from 4-ethylphenyl boronic acid
(0.276 g) and 3 (0.200 g) as a brown oil (163 mg, 87%). 1H NMR
4.6.4.2. 5-Chloro-4-hydroxy-30-methylbiphenyl-3-carboxylic acid (5b).
Compound 5b was obtained from 4d as a beige solid (78 mg, 90%);
mp 205e210 ꢀC (dec.); 1H NMR (CD3OD)
d
7.83 (d, 1H, J ¼ 2.4 Hz),
7.62 (d, 1H, J ¼ 2.4 Hz), 7.10e7.17 (m, 3H), 6.97e6.99 (m, 1H), 2.22 (s,
3H); 13C NMR (CD3OD)
171.7, 156.6, 138.6, 138.4, 133.4, 132.6,
d
7.90 (d, 1H, J ¼ 1.6 Hz), 7.75 (d, 1H, J ¼ 1.6 Hz), 7.47 (d, 2H,
J ¼ 8.1 Hz), 7.27 (d, 2H, J ¼ 8.1 Hz), 3.95 (s, 3H), 3.97 (s, 3H), 2.69 (q,
d
2H, J ¼ 7.6 Hz), 1.28 (t, 3H, J ¼ 7.6 Hz); 13C NMR
d
165.8, 154.7, 144.3,
128.5, 127.8, 126.7, 126.6, 123.2, 121.8, 114.1, 20.1; HRMS (ESI) m/z
137.8, 135.8, 132.2, 129.8, 128.5, 128.1, 126.8, 126.8, 62.0, 52.5, 28.5,
15.5.
calcd for C14H11ClO3 [M ꢁ H]ꢁ 261.0324, found 261.0321.
4.6.4.3. 5-Chloro-4-hydroxy-40-methylbiphenyl-3-carboxylic acid (5c).
Compound 5c was obtained as a beige solid (90 mg, 99%); mp
4.6.3.5. Methyl 40-butyl-5-chloro-4-methoxybiphenyl-3-carboxylate
(4f). Compound 4f was prepared from 4-butylphenyl boronic acid
(0.322 g) and 3 (0.200 g) as a brown oil (183 mg, 89%). 1H NMR
210e215 ꢀC (dec.); 1H NMR (CD3OD)
d
8.05 (d, 1H, J ¼ 2.4 Hz), 7.85
(d, 1H, J ¼ 2.4 Hz), 7.44 (d, 2H, J ¼ 7.8 Hz), 7.26 (d, 2H, J ¼ 7.8 Hz),
2.40 (s, 3H); 13C NMR (CD3OD)
173.1, 157.9, 138.5, 137.2, 134.7,
d
7.89 (s, 1H), 7.75 (s, 1H), 7.45 (d, 2H, J ¼ 7.8 Hz), 7.25 (d, 2H,
d
J ¼ 7.8 Hz), 3.96 (s, 3H), 3.94 (s, 3H), 2.64 (t, 2H, J ¼ 7.7 Hz),
133.9, 130.7, 127.9, 127.4, 123.3, 115.5, 21.1; HRMS (ESI) m/z calcd for
C14H11ClO3 [M ꢁ H]ꢁ 261.0324, found 261.0327.
1.59e1.67 (m, 2H), 1.31e1.44 (m, 2H), 0.94 (t, 3H, J ¼ 7.4 Hz); 13C
NMR
d 166.1, 154.9, 143.2, 138.0, 136.0, 132.4, 130.0, 129.2, 128.3,
127.0, 126.9, 62.2, 52.7, 35.5, 33.8, 22.6, 14.1.
4.6.4.4. 5-Chloro-4-hydroxy-40-(1-methoxyethyl)biphenyl-3-carbox-
ylic acid (5e). Compound 5e was obtained as a beige solid (75 mg,
4.6.3.6. Methyl 5-chloro-40-isobutyl-4-methoxybiphenyl-3-carbox-
ylate (4g). Compound 4g was prepared from 4-isobutylphenyl
boronic acid (0.327 g) and 3 (0.200) as a brown oil (155 mg, 76%). 1H
66%); mp 185e190 ꢀC (dec.); 1H NMR (CD3OD)
d
8.03 (d, 1H,
J ¼ 0.9 Hz), 7.82
d
(d, 1H, J ¼ 0.9 Hz), 7.53 (d, 2H, J ¼ 7.6 Hz), 7.37 (d,
2H, J ¼ 7.6 Hz), 4.36 (q, 1H, J ¼ 6.3 Hz), 3.22 (s, 3H), 1.42 (d, 3H,
NMR
d
7.91 (d, 1H, J ¼ 2.4 Hz), 7.76 (d, 1H, J ¼ 2.4 Hz), 7.46 (d, 2H,
J ¼ 6.3 Hz); 13C NMR (CD3OD)
d 171.5,156.1,142.9,136.9,133.2,131.4,
J ¼ 8.0 Hz), 7.22 (d, 2H, J ¼ 8.0 Hz), 3.97 (s, 3H), 3.95 (s, 3H), 2.51 (d,
126.7, 126.5, 126.4, 121.5, 114.9, 78.2, 55.8, 23.5; HRMS (ESI) m/z
2H, J ¼ 7.2 Hz), 1.85e1.94 (m, 1H), 0.93 (d, 6H, J ¼ 6.6 Hz); 13C NMR
calcd for C16H15ClO4 [M ꢁ H]ꢁ 305.0586, found 305.0597.
d
166.1, 154.9, 141.9, 138.0, 136.0, 132.4, 130.0, 129.9, 128.3, 127.0,
126.8, 62.2, 52.7, 45.2, 30.4, 22.5.
4.6.4.5. 40-Butyl-5-chloro-4-hydroxybiphenyl-3-carboxylic acid (5f).
Compound 5f was obtained as a beige solid (80 mg, 78%); mp
4.6.3.7. Methyl 40-tert-butyl-5-chloro-4-methoxybiphenyl-3-carbox-
ylate (4h). Compound 4h was prepared from 4-tert-butylphenyl
boronic acid (0.328 g) and 3 (0.200 g) as a brown oil (149 mg, 90%).
190e195 ꢀC (dec.); 1H NMR (DMSO-d6)
d
8.00 (d, 1H, J ¼ 2.4 Hz), 7.79
(t, 1H, J ¼ 2.4 Hz), 7.42 (d, 2H, J ¼ 8.4 Hz), 7.22 (d, 2H, J ¼ 8.4 Hz), 2.62
(t, 2H, J ¼ 7.7 Hz), 1.56e1.66 (m, 2H), 1.31e1.43 (m, 2H), 0.94 (t, 3H,
1H NMR
d
7.91 (d, 1H, J ¼ 2.4 Hz), 7.76 (d, 1H, J ¼ 2.4 Hz), 7.45e7.48
J ¼ 7.4 Hz); 13C NMR (DMSO-d6)
d 171.5, 155.9, 141.8, 135.2, 133.1,
131.7,129.0,126.4,126.2,121.4,114.8, 34.3, 33.0, 21.7,13.7; HRMS (ESI)
(m, 4H), 3.97 (s, 3H), 3.95 (s, 3H),1.35 (s, 9H); 13C NMR
d
166.1,154.9,
151.4, 137.9, 135.8, 132.5, 130.0, 128.3, 127.1, 126.8, 126.2, 62.3, 52.7,
34.8, 31.5.
m/z calcd for C17H17ClO3 [M ꢁ H]ꢁ 303.0793, found 303.0789.
4.6.4.6. 5-Chloro-4-hydroxy-40-isobutylbiphenyl-3-carboxylic acid (5g).
Compound 5g was obtained as a beige solid (114 mg, 98%); mp
4.6.3.8. Methyl 5-chloro-4-methoxy-40-(trifluoromethoxy)biphenyl-
3-carboxylate (4i). Compound 4i was prepared from 4-tri-
fluoromethoxyphenyl boronic acid (0.378 g) and 3 (0.200 g) as
175e180 ꢀC (dec.); 1H NMR (CD3OD)
d
8.05 (d, 1H, J ¼ 2.4 Hz), 7.85
(d, 1H, J ¼ 2.4 Hz), 7.49 (d, 2H, J ¼ 7.9 Hz), 7.24 (d, 2H, J ¼ 7.9 Hz),
a brown oil (135 mg, 61%). 1H NMR
d
7.87 (d, 1H, J ¼ 2.4 Hz), 7.72 (d,
2.52 (d, 2H, J ¼ 6.9 Hz), 1.92 (m, 1H) 0.93 (t, 6H, J ¼ 6.2 Hz); 13C NMR
1H, J ¼ 2.4 Hz), 7.56 (m, 2H), 7.28 (m, 2H), 3.96 (s, 6H); 13C NMR
(CD3OD) d 173.2, 157.9, 142.3, 137.5, 134.6, 133.9, 130.8, 127.9, 127.2,
d
165.6, 155.3, 149.1, 137.2, 136.4, 132.3, 130.1, 128.3, 128.2, 127.1,
123.3, 115.6, 46.0, 31.5, 22.7; HRMS (ESI) m/z calcd for C17H17ClO3
121.4, 120.4 (J ¼ 256 Hz), 62.1, 52.6.
[M ꢁ H]ꢁ 303.0793, found 303.0804.
4.6.4. General procedure for demethylation of compounds 4aei
To a stirred solution of the appropriate carboxylate 4 in CH2Cl2
(3 mL) at ꢁ78 ꢀC under N2 BBr3 (10 equiv.) in CH2Cl2 was added over
a 5 min period. The reaction mixture was stirred overnight at rt and
once thestarting materialhadbeen consumed (asmonitoredby TLC),
the reaction was quenched via the addition of MeOH (10 mL). The
solvent was removed in vacuo to give a crude solid, which was then
treated by repeated addition and evaporation of MeOH (5 ꢃ 5 mL) to
remove volatile impurities. H2O (15 mL) was added and the pH was
adjustedto14.TheaqueouslayerwaswashedwithEtOAc(3ꢃ10mL).
4.6.4.7. 40-tert-Butyl-5-chloro-4-hydroxybiphenyl-3-carboxylic acid
(5h). Compound 5h was obtained as a beige solid (130 mg, 90%);
mp 215e220 ꢀC (dec.); 1H NMR (CD3OD)
d
8.03 (s, 1H), 7.81 (s, 1H),
7.47 (br s, 4H), 1.35 (s, 9H); 13C NMR (CD3OD)
d
171.6, 156.5, 150.2,
135.6, 133.2, 132.3, 126.4, 125.7, 125.5, 121.9, 33.9, 30.3; HRMS (ESI)
m/z calcd for C17H17ClO3 [M ꢁ H]ꢁ 303.0793, found: 303.0787.
4.6.4.8. 5-Chloro-4-hydroxy-40-(trifluoromethoxy)biphenyl-3-carbox-
ylic acid (5i). Compound 5i was obtained as a beige solid (91 mg,
98%); mp 220e225 ꢀC (dec.); 1H NMR (CD3OD)
d 8.00 (d, 1H,

5316
O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
J ¼ 2.4 Hz), 7.65 (d, 1H, J ¼ 2.4 Hz), 7.59 (d, 2H, J ¼ 8.6 Hz), 7.24 (d,
and transfected into sub-confluent bovine aortic endothelial cells
using the Lipofectamine 2000 (Invitrogen) as described previously
[38]. The transfected cells were maintained in the medium without
fetal bovine serum for 2 h, and then used to evaluate the inhibitory
effects of compounds 5a and 9 on the metabolism of progesterone
in the cells. The cells were pretreated for 2 h with various
2H, J ¼ 8.6 Hz); 13C NMR (CD3OD)
d 173.1, 157.4, 148.2, 138.7, 131.0,
129.5, 127.5, 127.0, 121.4, 121.0, 120.6 (q, J ¼ 259 Hz), 119.3; HRMS
(ESI) m/z calcd for C14H8ClF3O4 [M ꢁ H]ꢁ 330.9990, found:
330.9994.
4.6.5. 3-Fluorobiphenyl-4-ol (7)
concentrations of inhibitors prior to incubating for 6 h with 30
progesterone. The metabolite, 20 -hydroxyprogesterone, in the
culture media was extracted twice by ethyl acetate, and quantified
mM
To a stirring solution of 6 (0.42 g, 2.1 mmol) in CH2Cl2 (5 mL)
at ꢁ78 ꢀC was added dropwise BBr3 (2.81 mL, 16.5 mmol) over
5 min. The reaction mixture was stirred overnight at rt. Once all
starting material had been consumed (as monitored by TLC), the
reaction mixture was quenched via the dropwise addition of MeOH
(5 mL). The solvent was removed in vacuo to yield a brown solid.
This crude mixture was then purified via flash chromatography
(0e8% Et2O in petroleum ether) to yield solid white needles
a
on an LC/MS using a Chiralcel OJ-H 5
previously [19].
m column as described
Acknowledgements
UD is the recipient of a Monash Graduate School postgraduate
scholarship. This work was partly supported by a Grant-in-Aid for
Young Scientists (to SE).
(0.338 g, 88%); mp 80e85 ꢀC (dec.); 1H NMR
d 7.52 (d, 2H,
J ¼ 7.5 Hz), 7.42 (t, 2H, J ¼ 7.3 Hz), 7.31 (m, 3H), 7.10 (t, 1H, J ¼ 8.6 Hz,
1H).
References
4.6.6. 5-Fluoro-4-hydroxybiphenyl-3-carbaldehyde (8)
[1] T.M. Penning, Endocr. Rev. 18 (1997) 281e305.
To a solution of 7 (0.320 g, 1.7 mmol) in CF3CO2H (5 mL) was
added portion wise, hexamethylenetetramine (0.477 g, 3.4 mmol)
over a 10 min period. The solution was refluxed for 14 h, cooled and
poured into H2O (30 mL). A solution of 50% H2SO4 (15 mL) was then
added and the solution was stirred for 2 h at rt. The solution was
then extracted with EtOAc (3 ꢃ 20 mL) and the combined organic
extracts were washed with 1 M HCl (5 ꢃ10 mL), brine (20 ml), dried
over MgSO4 and concentrated in vacuo. The resulting yellow oil was
purified using flash chromatography (0e50% CH2Cl2 in petroleum
ether) to yield a yellow solid (0.152 g, 41%); mp 55e59 ꢀC (dec.); 1H
[2] T.M. Penning, Hum. Reprod. Update 9 (2003) 193e205.
[3] B. Persson, Y. Kallberg, J.E. Bray, E. Bruford, S.L. Dellaporta, A.D. Favia,
R.G. Duarte, H. Jörnval, K.L. Kavanagh, N. Kedishvili, M. Kisiela, E. Maser,
R. Mindnich, S. Orchard, T.M. Penning, J.M. Thornton, J. Adamski,
U. Oppermann, Chem. Biol. Interact. 178 (2009) 94e98.
[4] J.M. Jez, T.M. Penning, Chem. Biol. Interact. 130e132 (2001) 499e525.
[5] O.A. Barski, S.M. Tipparaju, A. Bhatnagar, Drug Metab. Rev. 40 (2008)
553e624.
[6] J.M. Jez, M.J. Bennett, B.P. Schlegel, M. Lewis, T.M. Penning, Biochem. J. 326
(1997) 625e636.
[7] G. Sanli, J.I. Dudley, M. Blaber, Cell Biochem. Biophys. 38 (2003) 79e101.
[8] T.M. Penning, Y. Jin, V.V. Heredia, M. Lewis, J. Steroid Biochem. Mol. Biol. 85
(2003) 247e255.
NMR (CDCl3)
NMR (CDCl3)
d
10.96 (s, 1H), 10.01 (s, 1H), 7.36e7.63 (m, 6H); 13C
196.4, 151.1 (d, J ¼ 250 Hz), 148.9 (d, J ¼ 12.7 Hz),
[9] L.J. Askonas, J.W. Ricigliano, T.M. Penning, Biochem. J. 278 (1991) 835e841.
[10] B.P. Schlegel, J.M. Jez, T.M. Penning, Biochemistry 37 (1998) 3538e3548.
[11] C.E. Grimshaw, K.M. Bohren, C.J. Lai, K.H. Gabbay, Biochemistry 34 (1995)
14356e14365.
d
138.3, 133.4, 133.5, 129.1, 127.9, 126.6, 122.4, 121.5 (d, J ¼ 18.5 Hz).
[12] B. Nidetzky, M. Klimacek, P. Mayr, Biochemistry 40 (2001) 10371e10381.
[13] T.M. Penning, M.E. Burczynski, J.M. Jez, C.F. Hung, H.K. Lin, H. Ma, M. Moore,
N. Palackal, K. Ratnam, Biochem. J. 351 (2000) 67e77.
[14] D.R. Bauman, S. Steckelbroeck, T.M. Penning, Drug News Perspect. 17 (2004)
563e578.
[15] T.M. Penning, Y. Jin, T.L. Rizner, D.R. Bauman, Mol. Cell. Endocrinol. 281 (2008)
1e8.
[16] W.C. Cooper, Y. Jin, T.M. Penning, J. Biol. Chem. 282 (2007) 33484e33493.
[17] S. Steckelbroeck, Y. Jin, S. Gopishetty, B. Oyesanmi, T.M. Penning, J. Biol. Chem.
279 (2004) 10784e10795.
[18] Y. Jin, T.M. Penning, Biochemistry 45 (2006) 13054e13063.
[19] S. Ishikura, S. Nakajima, T. Kaneko, S. Shintani, N. Usami, I. Yamamoto,
V. Carbone, O. El-Kabbani, A. Hara, in: H. Weiner, B. Plapp, R. Lindahl, E. Maser
(Eds.), Enzymology and molecular biology of carbonyl metabolism, vol. 12,
Purdue University Press West Lafayette, 2006, pp. 341e351.
[20] Y. Zhang, I. Dufort, P. Rheault, V. Luu-The, J. Mol. Endocrinol. 25 (2000)
221e228.
[21] Y. Higaki, N. Usami, S. Shintani, S. Ishikura, O. El-Kabbani, A. Hara, Chem. Biol.
Interact. 143e144 (2003) 503e513.
[22] N. Usami, T. Yamamoto, S. Shintani, S. Ishikura, Y. Higaki, Y. Katagiri, A. Hara,
Biol. Pharm. Bull. 25 (2002) 441e445.
4.6.7. 5-Fluoro-4-hydroxybiphenyl-3-carboxylic acid (9)
To a stirred solution of 8 (0.140 g, 0.65 mmol) in 1,4-dioxane
(3 mL) and H2O (1 mL) was added sulfamic acid (0.094 g,
0.97 mmol) and NaH2PO4$H2O (0.390 g, 2.58 mmol). The mixture
was cooled in an ice-bath for 15 min and a solution of NaClO2
(0.076 g, 0.84 mmol) in H2O (1 mL) was added dropwise to the
mixture, and the solution was stirred for an additional 15 min.
Na2SO3 (0.098 g, 0.78 mmol) was then added and stirring was
continued for an additional 15 min. The solution was then acidified
to pH 1 by dropwise addition of conc. HCl and the solution was
extracted with EtOAc (3 ꢃ 10 mL). The organic layers were
combined and washed with H2O (20 mL) and dried over MgSO4.
The solvent was removed in vacuo and the crude residue was
purified by flash chromatography (0e10% MeOH in CH2Cl2) to yield
a white solid (0.092 g, 61%); mp 213e215 ꢀC (dec.); 1H NMR
(CD3OD)
d 7.80 (s, 1H), 7.53e7.63 (m, 3H), 7.41 (br t, 2H), 7.31 (br t,
[23] R.P. Piekorz, S. Gingras, A. Hoffmeyer, J.N. Ihle, Y. Weinstein, Mol. Endocrinol.
19 (2005) 431e440.
1H); 13C NMR (CD3OD)
d
173.1152.9 (d, J ¼ 246 Hz), 150.9 (d,
J ¼ 13.8 Hz), 140.2, 133.2, 130.1, 128.6, 127.6, 124.5, 120.7 (d,
J ¼ 18.5 Hz), 116.4; HRMS (ESI) m/z calcd for C13H9FO3 [M ꢁ H]ꢁ
231.0463, found: 231.0459.
[24] I. Sayeed, D.G. Stein, Prog. Brain Res. 175 (2009) 219e237.
[25] I. Roglio, R. Bianchi, S. Gotti, S. Scurati, S. Giatti, M. Pesaresi, D. Caruso,
G.C. Panzica, R.C. Melcangi, Neuroscience 155 (2008) 673e685.
[26] I. Roglio, R. Bianchi, F. Camozzi, V. Carozzi, I. Cervellini, D. Crippa, G. Lauria,
G. Cavaletti, R.C. Melcangi, J. Peripher. Nerv. Syst. 14 (2009) 36e44.
[27] E. Leonelli, R. Bianchi, G. Cavaletti, D. Caruso, D. Crippa, L.M. Garcia-Segura,
G. Lauria, V. Magnaghi, I. Roglio, R.C. Melcangi, Neuroscience 144 (2007)
1293e1304.
4.7. Evaluation of inhibitors in the cells
Bovine aortic endothelial cells, a generous gift from Taisho
Pharmaceutical Co. (Saitama, Japan), were cultured in Dulbecco’s
modified Eagle medium supplemented with 10% fetal bovine
[28] J.J. Lambert, D. Belelli, C. Hill-Venning, J.A. Peters, Trends Pharmacol. Sci. 16
(1995) 295e303.
[29] T. Smuc, N. Hevir, M. Ribic-Pucelj, B. Husen, H. Thole, T.L. Rizner, Mol. Cell.
Endocrinol. 301 (2009) 59e64.
[30] T. Smuc, T.L. Rizner, Mol. Cell. Endocrinol. 301 (2009) 74e82.
[31] M.J. Lewis, J.P. Wiebe, J.G. Heathcote, BMC Cancer 4 (2004) 27.
[32] D.C. Seo, J.M. Sung, H.J. Cho, H. Yi, K.H. Seo, I.S. Choi, D.K. Kim, J.S. Kim, A.A. El-
Aty, H.C. Shin, Mol. Cancer 6 (2007) 75.
[33] E. Selga, V. Noé, C.J. Ciudad, Biochem. Pharmacol. 75 (2008) 414e426.
[34] C.W. Chien, I.C. Ho, T.C. Lee, Carcinogenesis 30 (2009) 1813e1820.
serum, penicillin (100 U/mL) and streptomycin (100 m
g/mL) at 37 ꢀC
in a 5% CO2 incubator. In all experiments, the cells were used at
passages 4e8, and the endothelial cobblestone morphology was
confirmed microscopically before use. The eukaryotic expression
vector pGW1 harbouring the cDNA for AKR1C1 was constructed

O. El-Kabbani et al. / European Journal of Medicinal Chemistry 45 (2010) 5309e5317
5317
[35] S. Gobec, P. Brozic, T.L. Rizner, Bioorg. Med. Chem. Lett. 15 (2005) 5170e5175.
[36] M.C. Byrns, S. Steckelbroeck, T.M. Penning, Biochem. Pharmacol. 75 (2008)
484e493.
[37] D.R. Bauman, S.I. Rudnick, L.M. Szewczuk, Y. Jin, S. Gopishetty, T.M. Penning,
Mol. Pharmacol. 67 (2005) 60e68.
[43] H. Shiraishi, S. Ishikura, K. Matsuura, Y. Deyashiki, M. Ninomiya, S. Sakai,
A. Hara, Biochem. J. 334 (1998) 399e405.
[44] Z. Otwinowski, W. Minor, Meth. Enzymol. 276 (1997) 307e326.
[45] National Collaborative Computational Project, Acta Crystallogr. D Biol. Crys-
tallogr. 50 (1994) 760e763.
[38] O. El-Kabbani, P.J. Scammells, J. Gosling, U. Dhagat, S. Endo, T. Matsunaga,
M. Soda, A. Hara, J. Med. Chem. 52 (2009) 3259e3264.
[39] U. Dhagat, S. Endo, R. Sumii, A. Hara, O. El-Kabbani, J. Med. Chem 51 (2008)
4844e4848.
[40] G.N. Ramachandran, V. Sasisekharan, Adv. Protein Chem. 23 (1968) 283e438.
[41] M.L. Micklatcher, M. Cushman, Synthesis (1999) 1878e1880.
[42] K. Matsuura, A. Hara, Y. Deyashiki, H. Iwasa, T. Kume, S. Ishikura, H. Shiraishi,
Y. Katagiri, Biochem. J. 336 (1998) 429e436.
[46] P. Emsley, K. Cowtan, Acta Crystallogr. D Biol. Crystallogr. 60 (2004) 2126e2132.
[47] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Acta Crystallogr. D Biol. Crystallogr.
53 (1997) 240e255.
[48] R.A. Friesner, J.L. Banks, R.B. Murphy, T.A. Halgren, J.J. Klicic, D.T. Mainz,
M.P. Repasky, E.H. Knoll, M. Shelley, J.K. Perry, D.E. Shaw, P. Francis,
P.S. Shenkin, J. Med. Chem. 47 (2004) 1739e1749.
[49] R.J. Edsall Jr., H.A. Harris, E.S. Manas, R.E. Mewshaw, Bioorg. Med. Chem. 11
(2003) 3457e3474.