Synthesis, Biological Evaluation, and Molecular Modeling of 1 -Benzyl-1H-imidazoles as Selective Inhibitors of Aldosterone Synthase (CYP11B2)

Article abstract of DOI:10.1021/jm901356d

Reducing aldosterone action is beneficial in various major diseases such as heart failure. Currently, this is achieved, with mineralocorticoid, receptor antagonists, however, aldosterone synthase (CYP 11B2) inhibitors may offer a promising alternative. In this study, we used three-dimensional modeling of CYP11B2 to model, the binding modes of the natural substrate 18-hydroxycorticosterone and the recently published CYP11B2 inhibitor R-fadrozole as a rational guide to design 44 structurally simple and achiral 1-benzyl-1H-imidazoles. Their syntheses, in vitro inhibitor potencies, and in silico docking are described. Some promising CYP11B2 inhibitors were identified, with our novel lead, MOERAS 115 (4-((5-phenyl-1H-imidazol-1-yl)methyl) benzonitrile) displaying an IC₅₀ for CYP11B2 of 1.7 nM, and a CYP11B2 (versus CYP11B1) selectivity of 16.5, comparable to R-fadrozole (IC ₅₀ for CYP11B26.0 nM, selectivity 19.8). Molecular docking of the inhibitors in the models enabled us to generate posthoc hypotheses on their binding modes, providing a valuable basis for future studies and further design of CYP11B2 inhibitors.

Full text of DOI:10.1021/jm901356d

1712 J. Med. Chem. 2010, 53, 1712–1725
DOI: 10.1021/jm901356d
Synthesis, Biological Evaluation, and Molecular Modeling of 1-Benzyl-1H-imidazoles as Selective
Inhibitors of Aldosterone Synthase (CYP11B2)
Luc Roumen,† Joris W. Peeters,‡ Judith M. A. Emmen,§ Ilona P. E. Beugels,§ Erica M. G. Custers, Marcel de Gooyer,^
Ralf Plate, Koen Pieterse,† Peter A. J. Hilbers,† Jos F. M. Smits,§ Jef A. J. Vekemans,‡ Dirk Leysen,§ Harry C. J. Ottenheijm,§
Henk M. Janssen,‡ and J. J. Rob Hermans*,§
†BioModeling and bioInformatics, Eindhoven University of Technology, PO Box 513, Eindhoven 5600 MB, The Netherlands, ‡SyMO-Chem BV,
Den Dolech 2, Eindhoven 5612 AZ, The Netherlands, §Department of Pharmacology and Toxicology, Cardiovascular Research Institute
Maastricht, Maastricht University, P.O. Box 616, Maastricht 6200 MB, The Netherlands, Department Medicinal Chemistry,
Schering-Plough Research Institute, P.O. Box 20, Oss 5340 BH, The Netherlands, and ^Department of Pharmacology,
Schering-Plough Research Institute, P.O. Box 20, Oss 5340 BH, The Netherlands
Received September 14, 2009
Reducing aldosterone action is beneficial in various major diseases such as heart failure. Currently, this
is achieved with mineralocorticoid receptor antagonists, however, aldosterone synthase (CYP11B2)
inhibitors may offer a promising alternative. In this study, we used three-dimensional modeling of
CYP11B2 to model the binding modes of the natural substrate 18-hydroxycorticosterone and the
recently published CYP11B2 inhibitor R-fadrozole as a rational guide to design 44 structurally simple
and achiral 1-benzyl-1H-imidazoles. Their syntheses, in vitro inhibitor potencies, and in silico docking
are described. Some promising CYP11B2 inhibitors were identified, with our novel lead MOERAS115
(4-((5-phenyl-1H-imidazol-1-yl)methyl)benzonitrile) displaying an IC50 for CYP11B2 of 1.7 nM, and a
CYP11B2 (versus CYP11B1) selectivity of 16.5, comparable to R-fadrozole (IC50 for CYP11B2 6.0 nM,
selectivity 19.8). Molecular docking of the inhibitors in the models enabled us to generate posthoc
hypotheses on their binding modes, providing a valuable basis for future studies and further design of
CYP11B2 inhibitors.
Introduction
As we and others9 have put forward, an alternative manner
to reduce aldosterone action is to inhibit its biosynthesis. This
approach may solve the problems associated with MR anta-
gonism such as side-effects,10 interindividual variations in
patient response regarding the pharmacodynamics and phar-
macokinetics of MR antagonists,11 and the induced compen-
satory aldosterone synthesis of which long-term effects are
unknown.10,12 The latter might be of relevance in view of
the fact that not all aldosterone actions seem to involve the
mineralocorticoid receptor.2
The mineralocorticoid hormone aldosterone is an impor-
tant regulator of the sodium and potassium balance, extra-
cellular fluid volume, and blood pressure. It exerts its
biological effects via (classical) genomic as well as nonge-
nomic (rapid) actions that are mainly due to interaction with
the mineralocorticoid receptor (MRa), although interaction
with alternative but yet to be identified receptors may also
contribute to the effects of aldosterone.1
Apart from its role in physiology, aldosterone seems to play
a prominent role in the pathogenesis of various diseases as, e.g.,
heart failure,2 arrhythmias,2 hypertension,3 nephropathy,4,5
and edema.6 In particular, in the case of chronic heart failure
or postmyocardial infarction, clinical studies have convincingly
shown benefit of blocking the action of aldosterone with MR
antagonists and a reduction of patient mortality.7,8
Aldosterone is mainly produced in the adrenal cortex,
where its biosynthesis from the precursor cholesterol involves
a number of catalytic steps and enzymes (Figure 1).13,14 The
early steps of aldosterone biosynthesis share pathways and
precursors with other steroidal hormones. The final steps of
aldosterone biosynthesis are conducted by the cytochrome
P450 enzymes 11B1 (cortisol synthase) and 11B2 (aldosterone
synthase), which will further be denoted as CYP11B1 and
CYP11B2, respectively. These enzymes catalyze the 11-hydro-
xylation of 11-deoxycorticosterone to corticosterone, which is
then further hydroxylated to 18-hydroxycorticosterone
(18OH-B). Finally, CYP11B2 (but not CYP11B1) oxidizes
the 18-hydroxyl group of 18OH-B to the corresponding
aldehyde, thus resulting in the formation of aldosterone.
CYP11B1 shows a high structural overlap with CYP11B2
and shares some ligand conversion features with the latter
(Figure 1). However, CYP11B1 is mainly responsible for
glucocorticoid formation.13,14 Thus, to obtain specific inhibition
*To whom correspondence should be addressed. Phone: þ31 433 88
1331. Fax: þ31 433 88 4149. E-mail: r.hermans@farmaco.unimaas.nl.
a Abbreviations: CYP11B1 and -11B2, cytochrome P450 11B1 and
-2, respectively; MR, mineralocorticoid receptor; 18OH-B, 18-hydro-
xycorticosterone; B1model_Sfad, CYP11B1 model optimized with
S-fadrozole; B1model_18OHB, CYP11B1 model optimized with 18-
hydroxycorticosterone; B2model_Rfad, CYP11B2 model optimized
with R-fadrozole; B2model_18OHB, CYP11B2 model optimized with
18-hydroxycorticosterone; MD, molecular dynamics; LC-MS, liquid chro-
matography-mass spectrometry; GC-MS, gas chromatography-mass
spectrometry; HPLC, high performance liquid chromatography; LOO,
leave-one-out.
r
pubs.acs.org/jmc
Published on Web 02/01/2010
2010 American Chemical Society

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1713
Table 1. In Vitro Binding Potencies in CYP11B1 and CYP11B2 for the
Synthesized Inhibitors
Figure 1. Final steps of the biosynthesis of aldosterone by the
CYP11B family.
of the biosynthesis of aldosterone, it is essential to focus on
CYP11B2 and to avoid extensive CYP11B1 inhibition.
Recently, R-fadrozole (also referred to as FAD286) was
identified as a CYP11B2 inhibitor15 (Table 1). Animal studies
have shown that R-fadrozole does indeed lower aldosterone
levels in vivo and that it reduces end organ damage and
pathologies in which aldosterone is known to play a role, i.e.,
cardiac fibrosis,16,17 kidney fibrosis,4 primary aldosteronism,18
cardiac hypertrophy,15 diabetes associated cardiac arrythmias
and nephropathy,19,20 and salt-induced hypertension and
sympathetic hyperactivity.21 This indicates that CYP11B2 is
indeed a promising drug target.
Apart from R-fadrozole or its analogues, a number of
specific CYP11B2 inhibitors have been described. For in-
stance, Hartmann et al. described a variety of compounds
featuring an aromatic heterocycle such as a pyridine,22-24 an
imidazole,23,24 or an isoquinoline moiety,25 which are thought
to strongly interact with the cytochrome heme iron atom
(Supporting Information Figure 1). These hypothesized inter-
actions between the ligand and the heme iron are based on
interactions of nitrogen containing aromatic heterocycles as
shown in numerous crystal structures of cytochrome P450
enzymes combined with their ligands (see review de Graaf
et al.26). Furthermore, the aromatic moieties designed by
Hartmann et al. are linked either directly or via a methylene
group to two fused rings that are provided with various
substituents. Recently, structures featuring a pyridine moiety
have been optimized through 3,6- and 3,7-substitution
(Supporting Information Figure 1, compound D)27 and those
featuring an isoquinoline moiety through introduction of a
dihydroquinolinone.28
In the present study, we aim at the design and synthesis of a
different class of CYP11B2 inhibitors: structurally relatively
simple nonchiral 1-benzyl-1H-imidazoles that, in contrast to
the fadrozoles or the above-mentioned compounds of the
group of Hartmann, do not contain fused ring systems and
chiral carbon atoms. Using aninsilicomodeling approach, we
have designed a number of compounds that were subse-
quently synthesized and tested for their ability to inhibit
human CYP11B2 and CYP11B1 in an in vitro assay. As a
rational basis for CYP11B2 inhibitor design, we have used our
published three-dimensional CYP11B1 and CYP11B2 models.29
These models have previously been compared to those of
Ulmschneider et al.30 and of Belkina et al.31 and have proven
to accurately describe protein-substrate interactions and
protein-inhibitor interactions such as with R- and S-fadro-
zole and R-etomidate, another well-known CYP11B inhibi-
tor.32 In designing the 1-benzyl-1H-imidazoles as CYP11B2
inhibitors, we have combined the modeled R-fadrozole inter-
action with the structural investigation of the protein-inhibitor
interactions by performing molecular docking in the CYP11B1
and CYP11B2 models. The ligand poses resulting from the
molecular docking analyses have subsequently been used in
a retrospective analysis to rationalize the structure-activity

1714 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
Figure 2. Three-dimensional overlays of 18-hydroxycorticosterone (cyan) in the active site of CYP11B2 with (A) R-fadrozole (green), (B) R-
etomidate (orange), (C) the 5-methyl propanoate substituted 4-cyanobenzyl-1H-imidazole (3b, orange) and the 5-propanol substituted 4-
cyanobenzyl-1H-imidazole (4b, purple), and (D) the 5-phenyl substituted 4-cyanobenzyl-1H-imidazole (5b, green). 2D structural formulas of
the inhibitors can be found in Table 1.
relationship of the synthesized compounds. Because the protein
models feature a rigid structure, various protein conforma-
tions29 were included in our docking analysis to account for
possible active site perturbations. If the docking protocol
showed a sufficient trend with the pIC50, these protein struc-
tures were further analyzed to determine which active site
features they contained that might be of importance for
protein-ligand interactions.
iron but in contrast to R-fadrozole possesses the interaction
with Thr318.
Next, we designed three main modifications for the R-
fadrozole structure to mimic the interactions of the steroid.
These substituents are placed at the 5-position of the benzyl-
imidazole group and are the introduction of a methyl
n-propanoate moiety (Figure 2C purple, Table 1 series 3),
an n-propanol moiety (Figure 2C orange, Table 1 series 4),
and a phenyl group (Figure 2D green, Table 1 series 5),
respectively. The n-propanol moiety is overlaid with the C21-
hydroxyl group of 18OH-B, the methyl n-propanoate moiety
with the C20-carbonyl group of 18OH-B, and the phenyl
group partially fills the space of the D-ring of 18OH-B as well
as the C20,21-hydroxyacetyl group of 18OH-B. The resulting
5-imidazole substituted analogues were further modified by
replacing the nitrile group present in R-fadrozole. We also
opted for designing compounds that do or do not mimic the
interaction with Thr318 that is characteristic for R-etomi-
date. Thereby, the importance of this type of interaction with
Thr318 can be elucidated. All of the compounds that were
synthesized for this investigation possess the 4-nitrile benzyl
group that is present in R-fadrozole. An overview of all the
compounds that have been synthesized and tested for
CYP11B2 and -11B1 inhibition are listed in Table 1.
Synthesis. The N-benzyl imidazole compounds 1b-e and
2b-e have been prepared by alkylation of imidazole with the
appropriate substituted benzyl bromides, providing the pro-
ducts in isolated yields of 32-68%. Compounds 1g has been
Results
Rationale for the Inhibitor Design. As a starting point for
the inhibitor design, we have compared the interaction of the
natural ligand 18-hydroxycorticosterone (18OH-B) in the
active site of the CYP11B2 model with the modeled interac-
tions of the two CYP11B2 inhibitors R-fadrozole (Figure 2A,
green) and R-etomidate (Figure 2B, orange).29 In these
models, the binding mode of 18OH-B is stabilized by the
formation of the following three hydrogen bond interactions
with the protein active site: (1) between the C3 carbonyl and
Arg123, (2) between the C21-hydroxyl group and the Phe381
backbone, and (3) between the C18-hydroxyl group and
Thr318. In addition, the natural ligand possesses an internal
hydrogen bond between its C18-hydroxyl and its C20-carbonyl
groups. Modeling of R-fadrozole shows that this inhibitor
strongly binds to the heme iron atom with its aromatic
nitrogen atom and that it mimics the interaction of 18OH-B
with Arg123 via its nitrile group. When modeling R-etomidate,
we observed that this ligand also strongly binds to the heme

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1715
Scheme 1. Synthesis of the Imidazole Series 3 and 4 from Urocanic Acid 1133 a
a (a) TMS-Cl, Et3N, CHCl3, RT, 1 h; (b) (i) corresponding 4-substituted benzyl bromides 10, acetonitrile, 55-80 ꢀC, 18 h, (ii) acetic acid, water, reflux,
10 min; (c) LiAlH4, THF, 0 ꢀC; (d) (i) corresponding 4-substituted benzyl bromide, acetonitrile, 55-80 ꢀC, 18 h, (ii) MeOH, reflux, 30 min.
Scheme 2. Synthesis of the Imidazole Series 5 and 6 from Substituted Imidazolesa
a (a) ClCPh3, Et3N, CHCl3, RT, 1 h; (b) (i) acetonitrile, CHCl3, 80 ꢀC, 18 h, (ii) acetic acid, H2O, reflux, 10 min.
Scheme 3. Synthesis of Imidazole 6g and the Series 7 by Suzuki Coupling Reactionsa
a (a) DME, EtOH, water, Na2CO3, Pd(PPh3)4, 75 ꢀC, 18 h.
derived from its nitro precursor by a catalytic hydrogenation
reaction. The syntheses of the other inhibitors prepared are
outlined in Schemes 1-Scheme 3. The tritylated imidazole
building blocks 14 and 16 have been made in high-yielding
steps from urocanic acid 11 by using procedures as described
by Stark et al. (see Scheme 1).33 The methyl ester 14 has been
N-alkylated with a series of benzyl bromides, followed by
removal of the trityl group, to acquire compounds 3a-e and
3h. Similarly, the imidazole derivative 16 has been benzy-
lated and detritylated to prepare inhibitors 4b-4e. In an
alternative synthetic approach, the methyl esters 3a and 3h
have been reduced to give the alcohols 4a and 4h in good
yields.
Commercially available phenyl imidazole can conveni-
ently be tritylated to provide building block 9g, enabling
subsequent N-alkylation with various benzyl bromides to
furnish the series of compounds 5, while other commercially
available substituted imidazoles have also been tritylated
and then converted with 4-cyano benzyl bromide to prepare
compounds 6a-6e (see Scheme 2). The isolated yields for the
applied alkylations-deprotections of the tritylated imida-
zoles to acquire the series of compounds 3, 4, 5, and 6 depend
on the specific reactants and can vary between 11% and
100%, but are typically around 50%. Compound 6f has been
made from 6a by hydrolysis of the ester group. Finally,
compound 6g and the series of compounds 7a-f have been
prepared in isolated yields between 35% and 89% by a
Suzuki coupling reaction of the bromide 6d and the appro-
priate aryl boronic acids (Scheme 3).
Even though these prepared 1-benzyl-1H-imidazoles are
simple in structure and bear resemblance to the well-known
fadrozole structure, quite a number of the presented mole-
cules are novel, in particular those with aryl groups in the R5-
position of the imidazole ring, such as most molecules in the 5
and 7 series, and molecule 6g.
In Vitro Screening. The compounds were tested as inhibi-
tors of human CYP11B1 and CYP11B2 in an in vitro
screening method. This method consisted of an incubation
of test compounds and the substrate 11-deoxycorticosterone
with cultured V79 cells that stably overexpress CYP11B1 or
CYP11B2,34-36 followed by HPLC analysis of cell medium
extracts to determine substrate conversion rates. The ob-
tained IC50 values are shown in Table 1.
Overall, the IC50 values of the inhibitors range from as low
as 0.5 nM to more than 1000 nM (i.e., the highest tested
inhibitor concentrations). The CYP11B2 versus CYP11B1
selectivity defined as the quotient of the IC50 values for
CYP11B1 and CYP11B2 varies from 0.2 to 16.5. The most

1716 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
potent CYP11B2 inhibitors are 3b, 3e, 4e, 5b, 5d, 5e, 6b, and
7a-d (IC50 < 10 nM), whereas the most selective CYP11B2
inhibitors are 3b, 3e, 4c, 5b, 5d, 6a, 6b, 6c, 6e, 6f, 7a, and 7c
(selectivity > 4). The most potent CYP11B1 inhibitors are
5a, 5e, 7b, and 7d, whereas the most selective CYP11B1
inhibitors are 2c, 2f, and 4h.
Inspection of the IC50 values of the para- and meta-
substituted series 1a-g and 2b-g shows that these series
are not very potent for CYP11B2 and generally better
inhibitors for CYP11B1. Meta substitution seems to yield
an inhibitor that is selective for CYP11B1. For the para
substituted compounds (1a-g), this does not seem to be a
general rule because marginal selectivity for CYP11B2 is
observed when the para-substituent is a fluoride group
(selectivity of 1.5), while no selectivity is seen (selectivity =
1) when it is a nitrile group. Importantly, if a substituent is
introduced at the 5-position of the imidazole group, the
CYP11B2 potency generally is markedly increased as is the
selectivity.
Investigation of series 3a-e, 3h, 4a-e, 4h, and 5a-e, 5h,
reveals that within these series, the nitrile, fluoride, and
bromide groups are profitable for both CYP11B2 potency
(2-50 nM) as well as CYP11B2 versus CYP11B1 selectivity
(factor 1.4-16.5). For compounds 3e and 5b, the CYP11B2
versus CYP11B1 selectivity even approaches that of R-
fadrozole (factor 19.8). It is striking that of the compounds
3c-e and 4c-e, the chloride bearing analogues 3d and 4d
have the lowest CYP11B2 potency. Comparing the
CYP11B2 potencies, it is clear that both the bromide and
nitrile substituents are most favored in the active site of
CYP11B2. Overall, the bromide substituent also infers the
best CYP11B1 potency.
The introduction of other 5-imidazole substituents (series
6a-g) also increases the potency and selectivity for
CYP11B2. Comparing the potency of R-etomidate to 6b
shows a slightly decreased potency for CYP11B1 as well as
for CYP11B2. The drop in potency is larger for CYP11B1,
which can be attributed to the introduction of the nitrile
group that seems to be suboptimal for binding to CYP11B1.
Introduction of a sterically small substituent (e.g., 6c) leads
to increased selectivity for CYP11B2; however, when a
hydrogen bond donor (6f) or acceptor (6e) is added, neither
an increase of potency nor an increase of selectivity is
observed. The introduction of a methyl substituent confers
more selectivity than the introduction of a propanoate (3b)
and propanol (4b) but less selectivity than the phenyl group
(5b) (compare the selectivity of 11.8 to 8.9, 3.7, and 16.5,
respectively).
specifically chosen S-fadrozole in combination with
CYP11B1 because it is a strong and selective inhibitor for
this isoform (Table 1).29
The docking results of the ligands in all of the four initial
optimized structures (homology models) can be seen in
Figure 3. In our previous modeling study, we have observed
a trend between the GoldScore and the IC50. Because the
GoldScore is an artificial fitness function, we have investi-
gated the occurrence of chance correlation between the
docking score and the potency (Supporting Information
Figures 2 and 3). To validate the docking procedure in the
CYP11B homology models, regression analysis has been
carried out for the GoldScore and the binding affinity of
the compounds (Table 2). This analysis suggests that a
relation between the GoldScore of our docked compounds
and the pIC50 exists. Hence, we have decided that the trend
between the GoldScore and pIC50 can be used to select
protein conformations from the MD study, which may
possess active site features that related to the SAR of the
synthesized compounds.
The regression analysis showed that compound 6g is a
clear outlier in the docking. On the basis of the consideration
that the GoldScore does not discriminate any electrostatic
properties of the pyridine and phenyl moieties, compound 6g
is consistently scored equal to the more potent phenyl
analogue 5b. Because the GoldScore cannot capture the
behavior of the ring and the electrostatic interactions of the
ring are likely to play a role in protein-ligand interactions,
we have decided to leave compound 6g out of the analysis.
Indeed, by doing so, the obtained correlation was signifi-
cantly improved (P < 0.05 for all models in F-test), such that
we find the omission of compound 6g justified. The char-
acteristics of the regression analysis seem sufficient to con-
clude that our CYP11B models may be used to rationalize
structure-activity relationships in
a posthoc fashion
(Table 2, R2 > 0.75, q2LOO > 0.75, and q2k-fold > 0.7).
Molecular Dynamics and the Effect of the Choice of Ligand
Used for Model Optimization on Docking Performance. In-
vestigation of the various protein conformations of the
CYP11B1 and CYP11B2 models reveals that around half
of the protein conformations sampled by the molecular
dynamics (MD) simulations possess a similar trend between
the GoldScore and the pIC50 as the original models. Hence,
they seem feasible for the docking, ranking, and scoring of
the inhibitors synthesized and for generating hypotheses on
active site features that may determine protein-ligand inter-
actions (Supporting Information Figure 4). During the MD,
the protein conformation remained steady, resulting in
several protein conformations in which the docking poses
are seemingly in line with the observed IC50s. Only the
B1model_Sfad had in the MD several successive protein
conformations in which the relationship between the Gold-
Score and the pIC50 is lost (conformations 15-22). However,
generally speaking, all of the homology models possess an
adequate performance for docking.
Molecular Docking. Validation of the Docking Procedure.
In addition to the design, synthesis, and biological testing of
the inhibitors presented in this study, we have investigated
the binding mode of the inhibitors in the three-dimensional
models of CYP11B1 and CYP11B2, respectively. For the
evaluation of the binding mode of the inhibitors, flexible
docking has been carried out on several different protein
conformations extracted from molecular dynamics simula-
tions presented in an earlier study.29 The molecular dynamics
of the CYP11B1 and CYP11B2 models have been performed
with either 18-hydroxycorticosterone (models will be de-
noted as B1model_18OHB and B2model_18OHB, re-
spectively) or fadrozole. The CYP11B1 model contains the
S-enantiomer in the active site (denoted as B1model_Sfad),
whereas the CYP11B2 model contains the R-enantiomer
in the active site (denoted as B2model_Rfad). We have
If for CYP11B1 18OH-B was used as a ligand in our
homology models (B1model_18OHB), the performance of
the homology models in their basal conformation
(conformation 1) was poorer than if S-fadrozole was used
(B1model_Sfad). Also, if the various MD conformations
were compared for the basal B1model_18OHB, MD con-
formation 23 seemed to perform better in the inhibitor
dockings than the basal B1model_18OHB. A possible reason
for this is that while 18OH-B is not a substrate for CYP11B1,

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1717
Figure 3. Relation between the GoldScore and the pIC50 of the inhibitors in the four optimized protein structures of (A) B1model_Sfad, (B)
B1model_18OHB, (C) B2model_Rfad, and (D) B2model_18OHB. Red diamonds, series 1; green diamonds, series 2; blue diamonds, series 3;
red squares, series 4; green squares, series 5; blue squares, series 6; red circles, series 7. The red line indicates the line corresponding to the original
regression model (see Table 1, Supporting Information), whereas the blue line corresponds to the regression model without outlier 6g.
Table 2. Correlation Data for the Formulae of the GoldScore in Relation to the pIC50 Values for the Different Models as pIC50 = A ꢀ GoldScore þ Ba
model
A
B
R2
q2 LOO
SPRESS LOO
q2 k-fold
SPRESS k-fold
F
B1model_Sfad
0.154 ( 0.002
0.111 ( 0.002
0.195 ( 0.001
0.172 ( 0.001
-0.877 ( 0.096
0.772 ( 0.126
-3.404 ( 0.070
-2.853 ( 0.083
0.865
0.849
0.846
0.919
0.848
0.833
0.828
0.910
0.250
0.262
0.377
0.273
0.846
0.832
0.826
0.716
0.252
0.263
0.379
0.443
261.639
230.477
225.126
462.659
B1model_18OHB
B2model_Rfad
B2model_18OHB
a k-Fold is taken as 1000 times a k-fold cross-validation with k = 20% of the data set. Outlier 6g is not included in these formulas. Regression
constants are the correlation coefficient (R2), the cross-validated correlation coefficient (q2), the standard error (S), test statistic (F), and the prediction
error sum of squares (PRESS).
it fills a larger portion of the active site pocket than
S-fadrozole, especially around Phe381 (compare Figure 2A
and Figure 4). The MD conformation of 18OH-B pushes
Phe558 to the top of the active site, whereas in the MD
conformation of S-fadrozole, Phe558 forms a neat packing
to narrow the pocket. The docking of ligands in the narrow
pocket is more favorable than in the larger cavity in which
the ligands can fit more flexibly. In addition, a narrow
pocket probably will also result in a better ranking because
the GoldScore is optimized for steric interactions. To
summarize, molecular dynamics of the B1model_18OHB
has evolved the active site to conformation 23, of which the
active site conformation is comparable to that of the
For CYP11B2, B2model_18OHB performs better in the
docking of the inhibitors synthesized than the B2model_R-
fad. Both models show a similar active site packing (Figure 5).
The active site influences causing this difference in ranking
effectiveness are more subtle than for the protein conforma-
tions of CYP11B1 and cannot be easily rationalized.
Observed Inhibitor-Protein Interactions in the Docking
Studies. Docked inside the B2model_18OHB, the com-
pounds 4a-e and 4h interact with the protein backbone
carbonyl of Phe381 via its n-propanolyl moiety. In addition,
the docking pose of compound 4b possesses a stabilizing
interaction with Arg123 (Figure 5). These interactions are
similar to the ligand-based design discussed in the Rationale
section. The compounds 3a-e and 3h all fit with the propano-
ate group in its extended conformation as is demonstrated
˚
B1model_Sfad (rmsd of Phe558 side chain is 0.5 A, data
not shown).

1718 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
Figure 4. Active site difference between the CYP11B1 model opti-
mized with S-fadrozole (cyan) and 18OH-B (white). The molecules
shown in the active site are S-fadrozole and 4b. The induced fit of
S-fadrozole features a movement of Phe558, narrowing the active
site pocket.
Figure 6. Active site differences between CYP11B1 optimized with
S-fadrozole (cyan) and CYP11B2 optimized with 18OH-B (white).
Indicated are the docking poses of 5b (green), 5e (yellow), and 6b
(orange) in the CYP11B2 model. The arrow indicates the limited
volume around the para-position.
Because of the interaction with Phe558, the protrusion
of the para-benzyl substituents toward Arg123 is limited.
Figure 6 illustrates that the bromide substituent fits more
comfortably into the CYP11B1 pocket than the nitrile,
whereas the corresponding space of the pocket in CYP11B2
is larger (compare 5e and 7f). The binding orientation of the
imidazole above the heme for series 6 is varying in accor-
dance with the substituent size. The larger compounds 6a
and 6g bind in a similar fashion as observed for the series 3, 4,
5, and 7. The other inhibitors bind with their small substi-
tuent oriented into the small pocket around Thr318, forming
a hydrogen bond with the threonine alcohol if possible (see
compound 6b, Figure 6). The space around this substituent is
similar for CYP11B1 and CYP11B2, although the pocket is
slightly wider in CYP11B2. On the basis of these docking
observations, we hypothesize that the compounds of series 6
are selective for CYP11B2 (versus CYP11B1) because of its
more favorable pocket size for the para-nitrile substituent of
these compounds near Arg123. No conclusions could be
drawn from the docking results to explain the higher potency
of compound 6c over compounds 6e and 6f.
Figure 5. Active site differences for CYP11B2 optimized with
R-fadrozole (cyan) and 18OH-B (white). Only very small active site
differences are observed. Inside the cavity, the docking results for
compounds 3b (orange), 4b (purple), and 5b (green).
for compound 3b in Figure 5. As such, these compounds are
not docked in the active site pocket as anticipated in the design
of mimics for the C20-carbonyl group of 18OH-B. In retro-
spect, this is understandable because the C20-carbonyl group of
the steroid is stabilized by the internal hydrogen bond with its
C18-hydroxyl group, but this particular interaction is absent in
case of the inhibitors.
Instead of the designed curled conformation, the n-pro-
panoate moiety adopts a flat orientation in the active site
pocket similar to the phenyl ring of compounds 5a-e and 5h.
This orientation is visualized in Figure 5 for compound 5b.
The compounds from series 5 and 7 all form stable ring
stacking interactions with their phenyl moiety and Phe558,
forming an out-of-plane rotation with respect to the imida-
zole ring. With regard to the phenyl ring of Phe558, there is
room for the introduction of ortho and meta substituents,
whereas the para position allows only limited substitution
for the CYP11B2 models. This is indicated with an arrow for
compound 5b in Figure 6.
Discussion and Conclusion
The in silico design of novel CYP11B2 inhibitors based on
its overlap with the steroid 18-hydroxycorticosterone has led
to the synthesis of 1-benzyl-1H-imidazoles. The structures of
these inhibitors are relatively simple, they are nonchiral, and
they seem to achieve a good fit within the volume of the active
site of CYP11B2. Of these 44 substances, 20 compounds
possess a potency in the low nanomolar range (IC50 1-
30 nM) for CYP11B2 and 12 compounds for CYP11B1. In
addition, nine compounds possess a considerable selectivity
(versus CYP11B1) for CYP11B2 (5-20) that is close to that
observedforR-fadrozole, a compoundwhichhasproventobe
effective in vivo in reducing aldosterone levels and in reducing
aldosterone related pathological events.4,15-21 On the other
hand, the CYP11B2 versus CYP11B1 selectivities of our
compounds are quite poor when compared to those described
by the group of Hartmann et al.22-25 Their compounds
possess scaffolds different from ours and display a selectivity

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1719
up to a 1000-fold. This shows that higher CYP11B2 versus
CYP11B1 selectivity can be obtained but whether further
optimization of the type of compounds that is described in
the present study would indeed produce more selective com-
pounds still awaits further investigation. Additionally, it has
to be mentioned that in the present study, we have only
screened for CYP11B1 and CYP11B2 inhibition. Given the
quite simple structures of the compounds and their structural
relationship with S-fadrozole (a known CYP19 inhibitor), it
may well be that (some of) these compounds display inhibi-
tory activity to the steroid converting enzymes CYP17 and
CYP19. Future studies on optimizing the here described type
of compounds for CYP11B2 inhibition should take this into
account.
A molecular docking study of the inhibitors synthesized in
various protein conformations has shown that our homology
models are suitable to rationalize structure-activity relation-
ships. In particular, the homology model of CYP11B1 opti-
mized with S-fadrozole and that of CYP11B2 optimized with
18OH-B are the most appropriate for the structural investiga-
tion of fadrozole derivatives. Active site differences between
these models near the amino acids Phe558, Thr318, and
Arg123 have been shown to be important for ligand binding
and ligand specificity. In particular, we propose that the use of
a para-nitrile substitutent on the benzyl ring leads to the
highest CYP11B2 selectivity due to an interaction with
Arg123 that is slightly different with the corresponding inter-
action in CYP11B1.
We have observed that a simultaneous substitution on the
5-imidazole and the 1-H-benzyl moieties does not lead to clear
structure-activity relationships. Our models and docking
investigations indicate that this may be due to the flexibility
of the substituents placed on the 5-imidazole as in compound
series 3 and 4. The phenyl group of 5d seems to fit more rigidly
into the active site pocket of CYP11B2, thereby limiting the
protrusion of the benzylimidazole moiety, locking the com-
pounds of series 5 in one particular conformation. This lock-
ing isnot observedwiththe compound series 3 and 4 wherethe
imidazole is able to rotate freely.
Inaddition, wepresume that the dualeffect of polarizability
as well as a spatial arrangement may play a role in the potency
of the para-benzyl halogen containing inhibitors. The polar
effects of the nitrile and fluoride groups allow these substitu-
ents to interact strongly with the Arg123 in the active site of
our CYP11B2 models. On the other hand, based on our
models, it seems that the large bromide substituent can form
a weakly polar interaction with Arg123, and the para-bromide
compoundspossessa better steric fit in the active sitepocket of
CYP11B2. These various effects may play a role in the
preferred docking of bromide compounds over fluoride and
chloride compounds.
Not all of the measured selectivities of the 5-substituted
1-benzyl-1H-imidazoles can be rationalized with the in silico
models. For instance, the docking results are not able to
rationalize the higher potency of compound 6c over com-
pounds 6e and 6f. Steric differences near Thr318 seem to have
an effect, however, that does not explain why the esther group
of compounds 6b and R-etomidate does not result in a lower
potency. Future optimizations of the homology models are
required to shed light on these observations.
Finally, we propose that compound 5b, compound code
MOERAS115, is a good lead structure for future optimiza-
tions because in our models, it limits spatial movements of the
flexible benzyl-imidazoles in the pocket of CYP11B2. It also
possesses the important nitrile substituent on the para posi-
tion of the benzyl moiety. Indeed, of the here-studied com-
pounds, it possesses the highest CYP11B2 versus CYP11B1
selectivity (16.5) and a low IC50 value (1.7 nM) for CYP11B2.
Early explorations of modifications of its 5-imidazole phenyl
group and the molecular docking of these inhibitors (series 7)
have already given insight to the possibilities for structural
optimization.
Experimental Section
Chemistry. Molecular characterization of the prepared com-
pounds was performed by NMR spectroscopy using a Varian
Mercury Vx 400 MHz or a Varian Gemini 300 MHz spectro-
meter, where spectra were recorded at a temperature of 298 K.
Found coupling constants (J) are given in Hz. The purity of all
products was determined using either LC-MS of GC-MS, where
the chromatograms showed only a single peak corresponding to
the desired product (purity g95%). LC-MS measurements were
executed using a Finnigan LCQ Decca XP Max ESI-MS
spectrometer and applying an Alltima HP C18 50 mm, 3 μm
column with an acetonitrile/water eluent gradient, and 0.1%
formic acid in the eluent for ionization. In addition to the MS-
detection, photo diode array (PDA) detection was also per-
formed. Some products have alternatively been characterized
by GC-MS, using a Shimadzu GCMS-QP5000 spectrometer
applying a Zebron ZB-5 15 m, 0.1 μm, or a Zebron ZB-35 30 m,
0.25 μm capillary column in a Shimadzu GC-17A chromato-
graph.
Compounds 1a, 1f, 2f, 2g, the substituted imidazoles 8, the
benzyl bromides 10, urocanic acid 11, and the aryl boronic acids
17 have been acquired from commercial sources (mainly Acros
and Sigma-Aldrich). All further reagents, catalysts, chemicals,
materials and solvents were also obtained commercially and
were used without further purification. When appropriate,
drying of solvents was done by applying molsieves that were
prepared by drying in an oven at 500 ꢀC.
The series of imidazoles 1 and 2 have been prepared using a
similar procedure as that for compound 1c. Details and changes
with respect to the procedure for 1c are given below.
1-(4-Fluorobenzyl)-1H-imidazole (1c). 4-Fluorobenzyl bro-
mide 10c (1.0 g, 5.3 mmol), imidazole (0.36 g, 5.3 mmol), and
potassium carbonate (1.46 g, 10.6 mmol) were dissolved in
10 mL of acetonitrile and stirred for 4 h at 60 ꢀC. After cooling
down, the mixture was filtered and the filtrate was evaporated in
vacuo, giving crude product that was purified by column
chromatography over silica eluting with chloroform/methanol
(95:5), yielding 0.41 g (44%) of a colorless oil. 1H NMR (CDCl3)
δ 7.52 (s, 1H), 7.12 (m, 2H), 7.08 (s, 1H), 7.03 (m, 2H), 6.87 (s,
1H), 5.08 (s, 2H). GC-MS: m/z = 176 (molecular ion). LC-MS:
m/z = 177 (M þ 1).
Overall, the bromide substituent also infers the correspond-
ing ligands the higher CYP11B1 potency, often even higher
than the corresponding nitrile bearing ligands (1b,d, 4b,e, and
5b,d, though not 3b,d). Intuitively, the presence of a para-
nitrile substituent ought to make a better interaction with
Arg123, but for the tested compounds, this substituent pro-
trudes further into the active site than a para-bromide sub-
stituent. This may indicate that in CYP11B1 the three-
dimensional space close to Arg123 is more limited than in
case of CYP11B2, as is in line with our earlier observations.29
As such, the presence of a para-nitrile substituentseems to be a
determining factor for the CYP11B2 selectivity of the benzyl-
imidazoles.
1-(4-Cyanobenzyl)-1H-imidazole (1b). Using 4-cyanobenzyl
bromide 10b, sodium carbonate as base and chloroform as the
solvent. The crude product was dissolved in 1 M aqueous
hydrochloride and washed twice with ether. After bringing the

1720 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
pH to 10, the aqueous solution was extracted with chloroform,
after which the organic layer was dried over Na2SO4 and
evaporated to dryness, yielding 3.96 g (42%) of a white solid.
1H NMR (DMSO-d6) δ 7.85 (d, 2H, J = 8.4), 7.79 (s, 1H), 7.38
(d, 2H, J = 8.4), 7.22 (s, 1H), 6.95 (s, 1H), 5.32 (s, 2H). LC-MS:
m/z = 184 (M þ 1).
The series of imidazoles 3 were prepared using a similar
procedure as that for compound 3b. Details and changes with
respect to the procedure for 3b are given below.
Methyl 3-[1-(4-Cyanobenzyl)-1H-imidazol-5-yl]-propanoate
(3b). Methyl 3-(1-trityl-1H-imidazol-4-yl)-propanoate 14 (1.0
g, 2.5 mmol) and 4-cyanobenzyl bromide 10b (0.5 g, 2.5 mmol)
were dissolved in 10 mL of acetonitrile and stirred overnight at
80 ꢀC. The mixture was concentrated and triturated using 10 mL
of ethyl acetate. The precipitate was collected by filtration and
washed with ethyl acetate. The solids were refluxed for 30 min in
20 mL of methanol. The resulting solution was evaporated in
vacuo and triturated with 10 mL of ethyl acetate. After washing
with ethyl acetate, the solid was taken up in a saturated
NaHCO3 solution (15 mL), and this solution was extracted
twice with 15 mL of chloroform. The combined organic layers
were washed with a saturated NaHCO3 solution, dried over
Na2SO4, and evaporated to dryness, giving a yellowish oil that
was purified by column chromatography on silica gel eluting
with CHCl3/MeOH (95:5-90:10) resulting in 0.51 g (75%) of a
slightly yellow solid. 1H NMR (CDCl3) δ 7.65 (d, 2H, J = 8.4),
7.51 (s, 1H), 7.13 (d, 2H, J = 8.4), 6.89 (s, 1H), 5.19 (s, 2H),
3.66 (s, 3H), 2.69 (t, 2H, J = 7.7), 2.60 (t, 2H, J = 7.7). LC-MS:
m/z = 270 (M þ 1).
1-(4-Chlorobenzyl)-1H-imidazole (1d). Using 4-chloroben-
zylchloride 10d. Purification was performed by column chro-
matography on silica eluting with chloroform/methanol (95:5)
1
to give 0.81 g (68%) of product. H NMR (CDCl3) δ 7.52 (s,
1H), 7.32 (d, 2H, J = 8.4), 7.09 (s, 1H), 7.06 (d, 2H, J = 8.4),
6.87 (s, 1H), 5.08 (s, 2H). LC-MS: m/z = 193 (M þ 1). GC-MS:
m/z = 192 (molecular ion).
1-(4-Bromobenzyl)-1H-imidazole (1e). Using 4-bromobenzyl-
bromide 10e. Purification was performed by column chroma-
tography on silica eluting with chloroform/methanol (95:5-
1
90:10) to give 0.42 g (44%) of a slightly yellow solid. H NMR
(CDCl3) δ 7.54 (s, 1H), 7.43 (d, 2H, J = 8.4), 7.10 (s, 1H), 7.02 (d,
2H, J=8.4),6.88(s, 1H),5.08(s, 2H).LC-MS:m/z=237(Mþ 1).
1-(4-Aminobenzyl)-1H-imidazole (1g). 4-Nitrobenzyl bro-
mide 10i and imidazole were reacted in refluxing acetone,
applying potassium carbonate as base. Purification was per-
formed by column chromatography on silica eluting with
chloroform/methanol (97.5:2.5-95:5) to give 1.97 g (42%) of
a white solid, the nitro precursor. 1H NMR (CDCl3) δ 8.19 (d,
2H, J = 8.4), 7.58 (s, 1H), 7.24 (d, 2H, J = 8.4), 7.13 (s, 1H),
6.88 (s, 1H), 5.33 (s, 2H). LC-MS: m/z = 204 (M þ 1). 1-(4-
Nitrobenzyl)-1H-imidazole (0.5 g, 2.5 mmol) was weighed in a
high pressure vessel and dissolved in 2.5 mL of ethanol, 2.5 mL
of ethyl acetate, and 1 mL of a 1 M aqueous hydrochloride
solution. The mixture was purged with argon, and Pd/C (10%,
0.05 g) was then added. The mixture was shaken in a Parr
apparatus at 40 psi H2 pressure. After 30 min, the mixture was
filtered over celite and the filtrate was concentrated and
divided between chloroform (10 mL) and a saturated sodium
bicarbonate solution (10 mL). The aqueous layer was ex-
tracted with chloroform (10 mL), and the combined organic
layers were washed with water, dried over Na2SO4, and
evaporated in vacuo to dryness yielding 0.32 g (75%) of a
white solid. 1H NMR (CDCl3) δ 7.49 (s, 1H), 7.04 (s, 1H), 6.97
(d, 2H, J = 8.4), 6.87 (s, 1H), 6.64 (d, 2H, J = 8.4), 4.96 (s, 2H),
3.75 (bs, 2H). LC-MS: m/z = 174 (M þ 1).
Methyl 3-(1-Benzyl-1H-imidazol-5-yl)-propanoate (3a). Using
14 and benzyl bromide 10a and applying a reaction temperature
of 65 ꢀC. Purification by column chromatography on silica gel
eluting with chloroform/methanol (95:5) resulting in 1.18 g
1
(64%) of a white solid. H NMR (CDCl3) δ 7.48 (s, 1H), 7.34
(m, 3H), 7.05 (d, 2H, J = 7.3), 6.85 (s, 1H), 5.09 (s, 2H), 3.65
(s, 3H), 2.75 (t, 2H, J = 7.7), 2.56 (t, 2H, J = 7.7). LC-MS:
m/z = 245 (M þ 1).
Methyl 3-[1-(4-Fluorobenzyl)-1H-imidazol-5-yl]-propanoate
(3c). Using 14 and 4-fluorobenzyl bromide 10c. Purification
was performed by column chromatography on silica gel eluting
with chloroform/methanol (95:5) resulting in 0.51 g (77%) of a
white solid. 1H NMR (CDCl3) δ 7.47 (d, 1H, J = 1.1), 7.04 (s,
2H), 7.03 (s, 2H), 6.85 (s, 1H), 5.08 (s, 2H), 3.67 (s, 3H), 2.74 (t,
2H, J = 7.5), 2.57 (t, 2H, J = 7.5). GC-MS: m/z = 262
(molecular ion).
Methyl 3-[1-(4-Chlorobenzyl)-1H-imidazol-5-yl]-propanoate
(3d). Using 14 and 4-chorobenzyl chloride 10d. Purification by
column chromatography on silica gel eluting with chloroform/
methanol (99:1-98:2) resulting in 0.09 g (13%) of a colorless oil.
1H NMR (CDCl3) δ 7.48 (s, 1H), 7.32 (d, 2H, J = 8.4), 6.99 (d,
2H, J = 8.4), 6.85 (s, 1H), 5.08 (s, 2H), 3.67 (s, 3H), 2.72 (t, 2H,
J=7.5), 2.59 (t, 2H, J=7.5). GC-MS: m/z = 278, 280 (molecular
ions).
1-(3-Cyanobenzyl)-1H-imidazole (2b). Using 3-cyanobenzyl
bromide 10j. Purification was performed by column chroma-
tography on silica eluting with chloroform/methanol (95:5) to
give 0.43 g (46%) of product. 1H NMR (CDCl3) δ 7.63 (d, 1H,
J = 7.7), 7.57 (s, 1H), 7.49 (t, 1H, J = 7.7), 7.43 (s, 1H), 7.36
(d, 1H, J = 7.7), 7.14 (s, 1H), 6.91 (s, 1H), 5.18 (s, 2H). LC-MS:
m/z = 184 (M þ 1). GC-MS: m/z = 183 (molecular ion).
1-(3-Fluorobenzyl)-1H-imidazole (2c). Using 3-fluorobenzyl
bromide 10k. Purification was performed by column chroma-
tography on silica eluting with chloroform/methanol (95:5) to
give 0.46 g (49%) of product. 1H NMR (CDCl3) δ 7.55 (s, 1H),
7.32 (m, 1H), 7.11 (s, 1H), 7.01 (m, 1H), 6.93 (d, 1H, J = 7.3),
6.90 (s, 1H), 6.83 (d, 1H, J = 9.1), 5.12 (s, 2H). LC-MS: m/z =
177 (M þ 1). GC-MS: m/z = 176 (molecular ion).
Methyl 3-[1-(4-Bromobenzyl)-1H-imidazol-5-yl]-propanoate
(3e). Using 14 and 4-bromobenzyl bromide 10e. Purification
was performed by column chromatography on silica gel eluting
with chloroform/methanol (95:5) resulting in 0.49 g (60%) of a
white solid. 1H NMR (CDCl3) δ 7.48 (s, 1H), 7.47 (d, 2H, J =
8.6), 6.92 (d, 2H, J = 8.6), 6.85 (s, 1H) 5.06 (s, 2H), 3.67 (s, 3H),
2.72 (t, 2H, J = 8.0), 2.58 (t, 2H, J = 8.0). LC-MS: m/z = 325
(M þ 1).
Methyl 3-[1-(4-Methoxybenzyl)-1H-imidazol-5-yl]-propano-
ate (3h). Using 14 and 4-methoxybenzyl bromide 10h and
applying a reaction temperature of 55 ꢀC. Purified by column
chromatography on silica gel eluting with chloroform/methanol
(100:0-95:5) resulting in 0.49 g (71%) of a slightly yellow solid.
1H NMR (CDCl3) δ 7.45 (s, 1H), 7.00 (d, 2H, J = 8.4), 6.86 (d,
2H, J = 8.4), 6.82 (s, 1H), 5.01 (s, 2H), 3.79 (s, 3H), 3.66 (s, 3H),
2.76 (t, 2H, J = 7.5), 2.56 (t, 2H, J = 7.5). LC-MS: m/z = 275
(M þ 1).
1-(3-Chlorobenzyl)-1H-imidazole (2d). Using 3-chlorobenzyl
bromide 10l. Purification was performed by column chroma-
tography on silica eluting with chloroform/methanol (95:5) to
give 0.30 g (32%) of product. 1H NMR (CDCl3) δ 7.54 (s,
1H), 7.29 (m, 2H), 7.12 (m, 2H), 7.01 (m, 1H), 6.89 (t, 1H, J =
1.8), 5.09 (s, 2H). LC-MS: m/z = 193 (M þ 1). GC-MS: m/z =
192 (molecular ion).
1-(3-Bromobenzyl)-1H-imidazole (2e). Using 3-bromobenzyl-
bromide 10m. Purification was performed by column chroma-
tography on silica eluting with chloroform/methanol (95:5) to
give 0.42 g (44%) of product. 1H NMR (CDCl3) δ 7.55 (s, 1H),
7.46 (d, 1H, J = 8.1), 7.30 (s, 1H), 7.23 (t, 1H, J = 7.7), 7.11 (s,
1H), 7.07 (m, 1H), 6.90 (t, 1H, J = 1.2), 5.09 (s, 2H). LC-MS:
m/z = 239 (M þ 1). GC-MS: m/z = 236 and 238 (molecular ions).
The series of imidazoles 4b-4e were prepared using a similar
procedure as that for compound 4b, while 4a was prepared in a
similar way as 4h.
3-(1-Benzyl-1H-imidazol-5-yl)-1-propanol (4a). Using a simi-
lar procedure as that for 4h starting with methyl 3-(1-benzyl-
1H-imidazol-5-yl)-propanoate 3a yielding 0.44 g (57%) of a

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1721
colorless oil. 1H NMR (CDCl3) δ 7.40 (s, 1H), 7.29 (m, 3H), 7.03
(d, 2H, J = 7.7), 6.79 (s, 1H), 5.04 (s, 2H), 3.95 (bs, 1H), 3.61 (t,
2H, J = 6.3), 2.52 (t, 2H, J = 7.5), 1.79 (m, 2H). LC-MS: m/z =
217 (M þ 1).
(7.5 mL) was added, and after cooling down to room tempera-
ture, the mixture was extracted twice with 5 mL portions of
diethyl ether. The organic phase was extracted with 7.5 mL of a 1
M solution of HCl. The combined aqueous phases were adjusted
to pH 8 with a 6 M NaOH solution and were then extracted twice
with 10 mL portions of methylene chloride. The methylene
chloride extracts were dried over Na2SO4 and evaporated in
vacuo to give a slightly yellow oil. This crude product was
purified by column chromatography using silica gel and eluting
with a chloroform/methanol gradient yielding 60 mg (37%) of a
colorless oil. 1H NMR (CDCl3) δ 7.56 (s, 1H), 7.36 (m, 3H), 7.27
(m, 2H), 7.13 (s, 1H), 6.97 (m, 4H), 5.12 (s, 2H). LC-MS: m/z =
253 (M þ 1).
3-[1-(4-Cyanobenzyl)-1H-imidazol-5-yl]-1-propanol (4b). A
solution of 4-(3-((trimethylsilyl)oxy)propyl)-1-trityl-1H-imida-
zole 16 (19.8 g, 45 mmol) and 4-cyanobenzyl bromide 10b (8.8 g,
45 mmol) in 250 mL of acetonitrile was refluxed overnight. The
solution was concentrated and dissolved in a mixture of 80 mL
of acetic acid and 20 mL of water, and this mixture was refluxed
until a white solid precipitated. Water (200 mL) was added, and
after cooling down to room temperature the mixture was
extracted twice with diethyl ether. The aqueous phase was
adjusted to pH 9 with a 4 M NaOH solution and was then
extracted with methylene chloride. The organic extracts were
dried over Na2SO4 and evaporated to give a slightly yellow solid.
Trituration with cold acetone yielded 6.5 g (59%) of a white
solid. 1H NMR (CDCl3) δ 7.62 (d, 2H, J = 8.6), 7.48 (s, 1H),
7.12 (d, 2H, J = 8.6), 6.87 (s, 1H), 5.17 (s, 2H), 3.65 (t, 2H, J =
6.1), 2.69 (bs, 1H), 2.52 (t, 2H, J = 7.7), 1.84 (m, 2H). 13C NMR
(CDCl3) δ 141.8, 137.4, 132.8, 131.6, 127.0, 126.9, 118.3, 112.0,
61.2, 47.8, 31.0, 20.1. LC-MS: m/z = 242 (M þ 1).
1-Benzyl-5-phenyl-1H-imidazole (5a). Using 9g and benzyl
1
bromide 10a yielding 93 mg (61%) of a white solid. H NMR
(CDCl3) δ 7.55 (d, 1H, J = 0.8), 7.29 (m, 8H), 7.14 (s, 1H), 7.01
(d, 2H, J = 8.0), 5.14 (s, 2H). LC-MS: m/z = 235 (M þ 1). GC-
MS: m/z = 234 (molecular ion).
1-(4-Cyanobenzyl)-5-phenyl-1H-imidazole (5b). Using 9g and
4-cyanobenzyl bromide 10b yielding 5.5 g (68%) of white solid.
1H NMR (CDCl3) δ 7.61 (s, 1H), 7.58 (d, 2H, J = 8.4), 7.36 (m,
3H), 7.22 (m, 2H), 7.17 (s, 1H), 7.06 (d, 2H, J = 8.4), 5.23 (s,
2H). 13C NMR δ 142.1, 138.6, 133.4, 132.7, 129.2, 128.8, 128.4,
127.1, 118.3, 112.0, 48.3. LC-MS: m/z = 260 (M þ 1). GC-MS:
m/z = 259 (molecular ion).
1-(4-Chlorobenzyl)-5-phenyl-1H-imidazole (5d). Using 9g and
4-chlorobenzyl chloride 10d. For speeding up the reaction, ca. 1
mol equiv of NaI was added to the reaction mixture to in situ
convert the benzyl chloride to the benzyl iodide. Purification
yielded 21 mg (12%) of a yellowish oil. 1H NMR (CDCl3) δ 7.57
(s, 1H), 7.36 (m, 3H), 7.26 (m, 4H), 7.14 (s, 1H), 6.92 (d, 2H, J =
8.8), 5.12 (s, 2H). LC-MS: m/z = 269 and 271 (M þ 1 isotopic
ions).
3-[1-(4-Fluorobenzyl)-1H-imidazol-5-yl]-1-propanol (4c). Using
16 and 4-fluorobenzyl bromide 10c. Purification was performed
by column chromatography on silica gel eluting with chloroform/
1
methanol (95:5-90:10) to give 0.3 g (55%) of a white solid. H
NMR (CDCl3) δ 7.46 (s, 1H), 7.04 (s, 2H), 7.02 (s, 2H), 6.86 (s,
1H), 5.05 (s, 2H), 3.66 (t, 2H, J = 6.1), 2.53 (t, 2H, J = 7.3), 1.82
(m, 2H), 1.65 (bs, 1H). LC-MS: m/z = 235 (M þ 1).
3-[1-(4-Chlorobenzyl)-1H-imidazol-5-yl]-1-propanol (4d). Using
16 and 4-chlorobenzyl chloride 10d. Purification was performed
by column chromatography on silica gel eluting with chloroform/
1
methanol (95:5-90:10) to give 0.2 g (25%) of a white solid. H
NMR (CDCl3) δ 7.44 (s, 1H), 7.29 (d, 2H, J = 8.4), 6,96 (d, 2H,
J = 8.4), 6.83 (s, 1H), 5.04 (s, 2H), 3.65 (t, 2H), 2.51 (t, 2H, J =
7.5), 2.25 (bs, 1H), 1.81 (m, 2H). LC-MS: m/z = 251 (M þ 1).
3-[1-(4-Bromobenzyl)-1H-imidazol-5-yl]-1-propanol (4e). Using
16 and 4-bromobenzyl bromide 10e. Purification was performed
by column chromatography on alumina eluting with chloroform/
1-(4-Bromobenzyl)-5-phenyl-1H-imidazole (5e). Using 9g and
4-bromobenzyl bromide 10e yielding 79 mg (39%) of a slightly
yellow solid. 1H NMR (CDCl3) δ 7.57 (d, 1H, J = 0.9), 7.42 (d,
2H, J = 8.4), 7.36 (m, 3H), 7.26 (m, 2H), 7.15 (d, 1H, J = 0.9),
6.87 (d, 2H, J = 8.4), 5.11 (s, 2H). LC-MS: m/z = 315 (M þ 1).
1-(4-Methoxybenzyl)-5-phenyl-1H-imidazole (5h). Using 9g
and 4-methoxybenzyl bromide 10h yielding 96 mg (56%) of an
oil. 1H NMR (CDCl3) δ 7.52 (s, 1H), 7.32 (m, 5H), 7.12 (s, 1H),
6.94 (d, 2H, J = 8.6), 6.82 (d, 2H, J = 8.6), 5.06 (s, 2H), 3.76 (s,
3H). LC-MS: m/z = 265 (M þ 1).
1
methanol (99:1-95:5) to give 0.7 g (52%) of a white solid. H
NMR (CDCl3) δ 7.47 (s, 1H), 7.46 (d, 2H, J = 8.2), 6.90 (d, 2H,
J = 8.2), 6.85 (s, 1H), 5.03 (s, 2H), 3.66 (t, 2H, J = 6.2), 2.50 (t,
2H, J=7.7),2.1(bs,1H),1.82(m,2H).LC-MS:m/z= 297 (M þ 1).
3-[1-(4-Methoxybenzyl)-1H-imidazol-5-yl]-1-propanol (4h). A
solution of methyl 3-[1-(4-methoxybenzyl)-1H-imidazol-5-yl]-
propanoate 3h (0.48 g, 1.76 mmol) in 3.5 mL of freshly distilled
THF was added dropwise (carefully) to an ice-cooled suspen-
sion of lithium aluminiumhydride (0.1 g, 2.64 mmol) in 1.5 mL
of freshly distilled THF. After stirring for 2 h, the reaction
mixture was quenched by carefully adding 1.5 mL of a 0.1 M
aqueous solution of NaOH. The resulting suspension was
filtered and washed with THF, after which the filtrate was
concentrated and dissolved in 10 mL of dichloromethane. The
dichloromethane solution was washed twice with 10 mL of a 0.1
M aqueous solution of NaOH, dried over Na2SO4, and evapo-
rated to dryness, giving 0.36 g (83%) of a colorless oil. 1H NMR
(CDCl3) δ 7.43 (s, 2H), 6.99 (d, 2H, J = 8.6), 6.85 (d, 2H, J =
8.6), 6.82 (s, 1H), 4.99 (s, 2H), 3.79 (s, 3H), 3.65 (t, 2H, J = 6.2),
2.55 (t, 2H, J = 7.5), 2.18 (bs, 1H), 1.81 (m, 2H). LC-MS: m/z =
247 (M þ 1).
1-(4-Cyanobenzyl)-5-(methylene-acetate)-1H-imidazole (6a).
This compound was made starting from commercially available
4-(hydroxymethyl)-imidazole 8f following the literature proce-
dure of Millet et al.,37 involving subsequent tritylation, acetyl-
ation, alkylation with 4-cyanobenzyl bromide 10b, and depro-
tection of the trityl group. Yield: 0.87 g (65%) of a white solid.
1H NMR (CDCl3) δ 7.65 (d, 2H, J = 8.2), 7.58 (s, 1H), 7.20 (s,
1H), 7.14 (d, 2H, J = 8.2), 5.26 (s, 2H), 4.97 (s, 2H), 1.85 (s, 3H).
LC-MS: m/z = 256 (M þ 1).
1-(4-Cyanobenzyl)-5-(methyl carboxylate)-1H-imidazole (6b).
Using methyl 1-trityl-1H-imidazole-4-carboxylate 9b and 4-
cyanobenzyl bromide 10b yielding 27 mg (11%) of a white solid.
1H NMR (CDCl3) δ 7.80 (s, 1H), 7.71 (s, 1H), 7.63 (d, 2H, J =
8.3), 7.22 (d, 2H, J = 8.3), 5.59 (s, 2H), 3.80 (s, 3H). LC-MS:
m/z = 242 (M þ 1).
1-(4-Cyanobenzyl)-5-methyl-1H-imidazole (6c). Using 4-
methyl-1-trityl-1H-imidazole 9c and 4- cyanobenzyl bromide
10b yielding 0.71 g (78%) of a slightly yellow solid. 1H NMR
(CDCl3) δ 7.64 (d, 2H, J = 8.4), 7.50 (s, 1H), 7.12 (d, 2H, J =
8.4), 6.87 (s, 1H), 5.14 (s, 2H), 2.07 (s, 3H). LC-MS: m/z = 198
(M þ 1). GC-MS: m/z = 197 (molecular ion).
The series of imidazoles 5 and 6 have been prepared using a
similar procedure as that for compound 5c. Details and changes
with respect to the procedure for 5c are given below.
1-(4-Fluorobenzyl)-5-phenyl-1H-imidazole (5c). A solution of
4-phenyl-1-trityl-1H-imidazole 9g (0.25 g, 0.65 mmol) and 4-
fluorobenzyl bromide 10c (0.12 g, 0.65 mmol) in 2 mL of
acetonitrile and 0.5 mL of chloroform was stirred overnight at
80 ꢀC. The solution was concentrated and dissolved in a mixture
of 1.25 mL of acetic acid and 1 mL of water, and this mixture was
refluxed until a white solid precipitated. A 1 M HCl solution
1-(4-Cyanobenzyl)-5-bromo-1H-imidazole (6d). Using 4-bro-
mo-1-trityl-1H-imidazole 9d and 4- cyanobenzyl bromide 10b
yielding 84 mg (31%) of a white solid. 1H NMR (CDCl3) δ 7.64
(d, 2H, J = 8.4), 7.63 (s, 1H), 7.19 (d, 2H, J = 8.4), 7.10 (s, 1H),
5.20 (s, 2H). LC-MS: m/z = 262 (M þ 1). GC-MS: m/z = 261
and 263 (ionic isotopes).

1722 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
1-(4-Cyanobenzyl)-5-formyl-1H-imidazole (6e). Using 4-for-
myl-1-trityl-1H-imidazole 9e and 4-cyanobenzyl bromide10b yield-
ing 57 mg (37%) of a slightly yellow solid. 1H NMR (CDCl3) δ9.74
(s, 1H), 7.88 (s, 1H), 7.79 (s, 1H), 7.63 (d, 2H, J = 8.5), 7.26 (d, 2H,
J = 8.5), 5.58 (s, 2H). LC-MS: m/z = 212 (M þ 1).
chromatography over silica eluting with chloroform/methanol
(98:2) and a second column eluting with ethyl acetate/hexane
(1:1-4:1) yielded 29 mg (35%) of a white solid. 1H NMR
(CDCl3) δ 7.60 (s, 1H), 7.59 (d, 2H, J = 8.4), 7.25-6.95 (m,
7H), 5.22 (s, 2H), 2.32 (s, 3H). LC-MS: m/z = 274 (M þ 1).
1-(4-Cyanobenzyl)-5-(4-fluorophenyl)-1H-imidazole (7e). Using
4-fluorophenyl boronic acid 17e and purifying by column chro-
matography over silica eluting with chloroform/methanol
(99:1-98:2) yielded 75 mg (89%) of a white solid. 1H NMR
(CDCl3) δ 7.64 (s, 1H), 7.58 (d, 2H, J = 8.0), 7.18 (m, 2H), 7.14 (s,
1H), 7.05 (m, 4H), 5.20 (s, 2H). LC-MS: m/z = 278 (M þ 1).
1-(4-Cyanobenzyl)-5-(4-methylphenyl)-1H-imidazole (7f). Using
4-methylphenyl boronic acid 17f and purifying by column chro-
matography over silica eluting with chloroform/methanol
(99:1-98:2) yielded 37 mg (71%) of a white solid. 1H NMR
(CDCl3) δ 7.60 (s, 1H), 7.58 (d, 2H, J = 8.1), 7.2-7.0 (m, 7H),
5.21 (s, 2H), 2.36 (s, 3H). LC-MS: m/z = 274 (M þ 1). GC-MS:
m/z = 273 (molecular ion).
1-(4-Cyanobenzyl)-5-hydroxymethyl-1H-imidazole (6f). This
compound was prepared by hydrolysis of 6a, as described by
1
Millet et al.37 Yield: 0.3 g (72%) of a white solid. H NMR
(DMSO-d6) δ 7.83 (d, 2H, J = 8.2), 7.72 (s, 1H), 7.30 (d, 2H, J =
8.2), 6.86 (s, 1H), 5.35 (s, 2H), 5.12 (t, 1H, J = 5.2), 4.30 (d, 2H,
J = 5.2). LC-MS: m/z = 214 (M þ 1).
1-(4-Cyanobenzyl)-5-(4-pyridyl)-1H-imidazole (6g). 1-(4-Cy-
anobenzyl)-5-bromo-1H-imidazole 6d (100 mg, 0.38 mmol),
4-pyridinyl boronic acid 17h (52 mg, 0.42 mmol), 0.38 mL of a
2 M aqueous solution of K3PO4 (0.76 mmol), and 1 mL of DMF
were weighed in a Schlenck tube and capped with a rubber
septum. Three cycles of freeze-pump-thaw were performed to
remove the oxygen from the reaction mixture, and tetrakis-
(triphenylphosphine)palladium(0) (18 mg, 15 μmol) was added.
The reaction mixture was then stirred overnight at 85 ꢀC. To the
reaction mixture were added 3 mL of ethyl acetate and 6 mL of
water. After stirring a few moments, the layers were separated
and the water layer was extracted with ethyl acetate. The
combined organic layers were dried over MgSO4, and the
solvent was evaporated in vacuo, giving crude product that
was purified by column chromatography over silica eluting with
ethyl acetate/hexane/methanol (2:1:0-96:0:4) and a second
column eluting with chloroform/methanol (95:5) yielding 58
mg (58%) of product. 1H NMR (CDCl3) δ 8.57 (d, 2H, J = 6.0),
7.66 (s, 1H), 7.60 (d, 2H, J = 8.1), 7.33 (s, 1H), 7.13 (d, 2H,
J = 6.0), 7.09 (d, 2H, J = 8.1), 5.31 (s, 2H). LC-MS: m/z = 261
(M þ 1). GC-MS: m/z = 260 (molecular ion).
The tritylated imidazole building blocks 9 were prepared as
described below for 9c, unless indicated otherwise.
4-Methyl-1-trityl-1H-imidazole (9c). A solution of 4-methyl-
1H-imidazole 8c (0.99 g, 12 mmol) and triethyl amine (3.4 g, 34
mmol) in 15 mL of chloroform was added to an ice-cooled
solution of trityl chloride (3.7 g, 13 mmol) in 15 mL of chloro-
form. The reaction mixture was stirred for 1 h at ambient
temperature and was then washed twice with 15 mL portions
of a 0.5 M HCl solution and once with 15 mL of a saturated
NaHCO3 solution. The organic layer was dried over Na2SO4
and evaporated in vacuo. The resulting solid was triturated with
15 mL of diethyl ether, filtered, and washed with diethyl ether
yielding 3.7 g (93%) of a white solid. 1H NMR (CDCl3) δ 7.32
(m, 10H), 7.15 (m, 6H), 6.51 (s, 1H), 2.10 (s, 3H). LC-MS: m/z =
325 (M þ 1) and 243 (trityl radical cation). GC-MS: m/z = 243
(trityl radical cation), the molecular ion is not observed.
4-(Methylene acetate)-1-trityl-1H-imidazole (9a). This com-
pound was made by acylation of 9f following the procedure of
Millet et al.37
Methyl 1-Trityl-1H-imidazole-4-carboxylate (9b). A few ml of
N,N-dimethylformamide (DMF) was used as cosolvent to dis-
solve the starting compound methyl 1H-imidazole-4-carboxy-
late 8b. The DMF was removed by trituration with water,
followed by trituration with diethyl ether to yield 0.4 g (56%)
of a white solid. 1H NMR (CDCl3) δ 7.59 (s, 1H), 7.62 (s, 1H),
7.35 (m, 9H), 7.12 (m, 6H), 3.85 (s, 3H). LC-MS: m/z = 737
(2Mþ1) and 243 (trityl radical cation). GC-MS: m/z = 243
(trityl radical cation).
4-Bromo-1-trityl-1H-imidazole (9d). Using 4-bromo-1H-imi-
dazole 8d yielding 0.77 g (73%) of a white solid. 1H NMR
(CDCl3) δ 7.35 (m, 9H), 7.30 (s, 1H), 7.10 (m, 6H), 6.80 (s, 1H).
LC-MS: m/z = 391 (M þ 1) and 243 (trityl radical cation).
GC-MS: m/z = 243 (trityl radical cation).
4-Formyl-1-trityl-1H-imidazole (9e). A few mL of N,N-di-
methylformamide (DMF) were used as cosolvent to dissolve
the starting compound 1H-imidazole-4-carbaldehyde 8e. The
DMF was removed by trituration with water, followed by
trituration with diethyl ether to yield 0.6 g (69%) of a
white solid. 1H NMR (CDCl3) δ 9.85 (s, 1H), 7.61 (s, 1H),
7.52 (s, 1H), 7.35 (m, 9H), 7.10 (m, 6H). LC-MS: m/z = 677
(2M þ 1) and 243 (trityl radical cation). GC-MS: m/z = 243
(trityl radical cation).
The imidazole series 7 were prepared using a similar proce-
dure as that for compound 7b.
1-(4-Cyanobenzyl)-5-(2-methylphenyl)-1H-imidazole (7b). 1-(4-
Cyanobenzyl)-5-bromo-1H-imidazole 6d (80 mg, 0.31 mmol),
2-methylphenyl boronic acid 17b (46 mg, 0.34 mmol), and
Na2CO3 (0.76 mmol) were weighed in a Schlenck tube, capped
with a rubber septum. The reaction tube was filled with argon and
subsequently with 1.5 mL of a 1:1:1 mixture of water, dimethoxy
ethane, and ethanol. The mixture was heated to 75 ꢀC upon which
a clear solution developed. Three cycles of freeze-pump-thaw
were performed to remove the oxygen from the reaction mixture,
and tetrakis(triphenylphosphine)palladium(0) (0.7 mg, 0.6 μmol)
was added. The reaction mixture was then stirred overnight at
75 ꢀC. The mixture was concentrated, 2.5 mL of chloroform was
added, and the mixture was washed twice with 2 mL portions of a
saturated NaHCO3 solution. The organic layer was dried over
Na2SO4, and the solvent was evaporated in vacuo, giving crude
product that was purified by column chromatography over silica
eluting with chloroform/methanol (98:2) and a second column
eluting with ethyl acetate/hexane (1:1-4:1), yielding 46 mg (55%)
of a colorless oil. 1H NMR (CDCl3) δ 7.67 (s, 1H), 7.51 (d, 2H,
J = 8.3), 7.3-7.1(m, 3H), 7.04 (m, 2H), 6.95 (d, 2H, J =8.3), 4.98
(s, 2H), 2.02 (s, 3H). LC-MS: m/z = 274 (M þ 1).
1-(4-Cyanobenzyl)-5-(2-fluorophenyl)-1H-imidazole (7a). Using
2-fluorophenyl boronic acid 17a and purifying by column chro-
matography over silica eluting with ethyl acetate/hexane (1:1-4:1)
yielded32mg(38%) ofaslightlyyellowsolid.1HNMR(CDCl3) δ
7.63 (s, 1H), 7.53 (d, 2H, J = 8.4), 7.36 (m, 1H), 7.14 (m, 4H), 7.04
(d, 2H, J = 8.4), 5.15 (s, 2H). LC-MS: m/z = 278 (M þ 1).
1-(4-Cyanobenzyl)-5-(3-fluorophenyl)-1H-imidazole (7c). Using
3-fluorophenyl boronic acid 17c and purifying by column chro-
matography over silica eluting with ethyl acetate/hexane (2:1-4:1)
yielded 52 mg (61%) of a white solid. 1H NMR (CDCl3) δ 7.63 (s,
1H), 7.60 (d, 2H, J = 8.4), 7.32 (m, 1H), 7.19 (s, 1H), 7.08 (d, 2H,
J = 8.4), 7.02 (m, 2H), 6.94 (m, 1H), 5.26 (s, 2H). LC-MS: m/z =
278 (M þ 1).
4-(Hydroxymethyl)-1-trityl-1H-imidazole (9f). This com-
pound was made by tritylation of commercially available
4-(hydroxymethyl)-imidazole 8f following the procedure of
Millet et al.37
4-Phenyl-1-trityl-1H-imidazole (9g). Using 4-phenyl-1H-imi-
dazole 8g yielding 2.4 g (90%) of a white solid. 1H NMR
(CDCl3) δ 7.7 (d, 2H, J = 8.4), 7.48 (s, 1H), 7.32 (m 11H),
7.20 (m, 7H), 7.10 (s, 1H). LC-MS: m/z = 387 (M þ 1) and 243
(trityl radical cation). GC-MS: m/z = 386 (molecular ion) and
243 (trityl radical cation).
1-(4-Cyanobenzyl)-5-(3-methylphenyl)-1H-imidazole (7d). Using
3-methylphenylboronic acid 17d and purifying by column

Article
Methyl 3-(1-Trityl-1H-imidazol-4-yl)-propanoate (14). This
precursor was synthesized starting with urocanic acid 11 by
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1723
conducted at least in duplicate. IC50 values are presented in
Table 1.
following the literature procedure of Stark et al.33
Molecular Modeling. The CYP11B1 and CYP11B2 protein
conformations were extracted from the homology modeling and
molecular dynamics study as described in Roumen et al.29 The
protein states sampled during the molecular dynamics were
extracted for every 25 ps of the simulation. The inhibitor struc-
tures of Table 1 were built with MOE38 using the MMFF94x
force field.
4-(3-((Trimethylsilyl)oxy)propyl)-1-trityl-1H-imidazole (16).
A solution of trimethylsilyl chloride (36.0 g, 331 mmol) in
200 mL of chloroform was added dropwise to an ice-cooled
solution of 3-(1-(trityl)-1H-imidazol-4-yl)propanol 1533 (101.8 g,
276 mmol) and triethylamine (39.0 g, 386 mmol) in 600 mL of
chloroform. After stirring for 1 h, the solution was concentrated
and diluted with 600 mL of diethyl ether, giving a white precipi-
tate. The precipitate was collected by filtration and washed with
600 mL of diethyl ether. Concentration of the filtrate and coeva-
poration with diethyl ether yielded a quantitative yield of a slightly
All inhibitors were docked into our CYP11B1 and CYP11B2
models using the program GOLD v3.239 using the GoldScore
scoring function. The flexible docking of compounds was
performed using the parameters from the default I settings
supplied by GOLD. The GOLD cytochrome P450 parameters
were used to obtain ligand poses of the aromatic nitrogen
interaction with the protein iron atom.40 Each ligand was
docked in each of the protein states for 50 poses, using the heme
iron atom as the definition of the ligand binding site. The
docking score of each ligand was averaged over the top 10
different poses using a clustering root-mean-square distance
1
yellow solid. H NMR (CDCl3) δ 7.35 (m, 10H), 7.15 (m, 6H),
6.55 (s, 1H), 3.61 (t, 2H, J = 6.2), 2.60 (t, 2H, J = 7.5), 1.85 (m,
2H), 0.10 (s, 9H). LC-MS: m/z = 369 and 243 for the TMS-
deprotected Hþ-species and the trityl radical cation, the (M þ 1)þ
species is not observed. GC-MS: m/z = 243 (trityl radical cation),
the molecular ion is not observed.
Biological Assays (Inhibition of CYP11B1 and CYP11B2).
Inhibitor potencies for human CYP11B1 and CYP11B2 were
determined in an in vitro assay using V79 cells in which the
human CYP11B1 or CYP11B2 are stably (over)expressed. The
human CYP11B1 (over)expressing V79 cells (stably transfected
with a pcDNA3.1 vector, carrying a hygromycin resistance box)
were constructed at Organon (Newhouse, UK) as described in
Roumen et al.29 The V79 cells stably (over)expressing human
CYP11B2 were developed in the laboratory of Prof. Rita
˚
cutoff of 2.0 A. A binding mode was deemed proper when the
aromatic nitrogen atom of the imidazole moiety made an
interaction with the heme iron atom at a distance of 2.2 ( 0.2
A and possessed a 90ꢀ ( 20ꢀ angle between the Fe-N bond and
the heme plane.
˚
Regression Analysis. All regression models were constructed
on the basis of a linear correlation using the formula Y = AX þ
B, where A and B are constants, Y is described by compound
affinity, and X is described by the docking GoldScore. Com-
pound affinity as used in these equations is expressed as pIC50
values. To determine the validity of the regression analysis,
leave-one-out (LOO) cross-validation has been performed. Be-
cause the data set is of moderate size (n = 44), LOO may
overestimate the predictivity of the correlation because it only
leaves out one compound for internal validation. For this
reason, we have chosen to also perform k-fold cross-valida-
tion, which uses more compounds for internal validation. If n
is the amount of compounds in the data set, the LOO proce-
dure iteratively builds an internal regression model based on
n - 1 compounds for the prediction of the remaining com-
pound, until each compound has been left out exactly once.
The k-fold cross-validation is a variant of the LOO method.
For k-fold cross-validation, the data set is randomly parti-
tioned into k sets (folds) of approximately the same size. Then
k - 1 of these sets are used for the internal prediction of the
remaining set, until all k sets have been left out once. Because
of the high number of possibilities for the randomized choos-
ing of the k sets, k-fold cross-validation needs to be carried out
a sufficient amount of times. Here we have chosen an iterative
k-fold analysis of 1000 evaluation runs. In turn, k-fold cross-
validation can underestimate the performance of the regres-
sion models if too many chemicals are held out for medium
size data sets (for instance if 50% is left out). Therefore, we
have chosen to partition our data set into folds of nine
compounds, which equals just over 20% of the total amount
of compounds.
For each of the CYP11B homology models, the following
statistical entities have been determined: the squared correlation
coefficient, R2, the Fisher F value, and the cross-validated
squared correlation coefficient, q2, and standard deviation of
the predictions, SPRESS for both LOO and k-fold cross-valida-
tion, respectively (i.e., LOO: q2LOO, SPRESS,LOO and k-fold:
q2k-fold, SPRESS,k-fold). These are indicated in Table 2.
In addition to cross-validation, all data series (x- and
y-vectors) have been checked for chance correlation by a process
referred to as scrambling, and the regressions of the scrambled
y-vectors with the original x-vectors are compared to that of the
original y-vector. Because less than 1% of the permutations
exceeded correlation with the original vectors and none of
the 1000 permutations provided a better correlation than the
Bernhardt, Institute of Biochemistry, Saarland University,
€
34-36
Saarbrucken, Germany.
Cells were cultured under standard conditions in DMEM/
FK12 medium (Gibco, Gaithersburg, MD) supplemented with
10% fetal calf serum (Hyclone, Logan, UT), penicillin/strepto-
mycin (100 U/mL and 100 μg/mL, respectively, both from
Gibco). Cell culturing and incubation experiments with the cells
were performed at 37 ꢀC in a humid environment and 5% CO2
atmosphere.
For assessing inhibitor potencies, cells were transferred to
12-well plates and grown until they were confluent. Next, cells
were incubated for 1 h in serum free medium with cumulative
inhibitor concentrations. Subsequently, 11-deoxycorticosterone
(Sigma-Aldrich, St Louis, MO) was added as substrate to obtain
a final concentration of 500 nM (0.5 ꢀ Km), following incuba-
tion of the plates for another hour for V79 CYP11B1 or another
3 h for V79 CYP11B2. After these incubation times, medium
was collected for steroid analysis by HPLC. To extract the
(produced) steroids from the medium, 1 mL aliquots of medium
were mixed with 1 mL of 1 M sodium-glycine buffer (pH 10.5)
containing 500 nM methylprednisolone (Sigma-Aldrich) as
internal standard and 5 mL of diethylether (Biosolve, Valkens-
waard, The Netherlands). After shaking for 3 min, the samples
were briefly centrifuged and the organic phase (containing the
steroids) transferred into another tube and blown to dryness
under a stream of nitrogen. The dried extracts were dissolved in
100 μL of mobile phase to be used for automated HPLC
analysis. The HPLC system comprised a MR column (4.6 mm
ꢀ 50 mm, particle size 2.5 μm; Shimadzu, Tokyo, Japan) as a
stationary phase and a mixture of 680/320/1 (v/v/v) Milli-Q
water, acetonitrile and trifluoroacetic acid as mobile phase at a
flow rate of 1.0 mL/minute. Steroids were detected by UV
absorption at 243 nm (Shimadzu, Tokyo, Japan). To quantitate
the steroids, peak area ratios versus the internal standard
(methylprednisolone) were determined and compared with
calibration curves.
Determination of IC50 values (potencies) was based on at
least five inhibitor concentrations and was derived from the
generic equation for enzyme inhibition at fixed substrate con-
centration vi = v0/(1 þ I/IC50) with vi and v0 being the reaction
velocities in the presence or absence of inhibitor respectively,
and I being the inhibitor concentration.41 All analyses were

1724 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4
Roumen et al.
original regression model, the likelihood of chance correlations
was found to be low (Supporting Information Figures 2 and 3).
Enantioselectivity Versus Reduction of Plasma Aldosterone.
Endocrinology 2008, 149 (1), 28–31.
(17) Huang, B. S.; White, R. A.; Ahmad, M.; Tan, J.; Jeng, A. Y.;
Leenen, F. H. Central infusion of aldosterone synthase inhibitor
attenuates left ventricular dysfunction and remodeling in rats after
myocardial infarction. Cardiovasc. Res. 2008, 81 (3), 574–581.
(18) Menard, J.; Pascoe, L. Can the dextroenantiomer of the aromatase
inhibitor fadrozole be useful for clinical investigation of aldoster-
one-synthase inhibition? J. Hypertens. 2006, 24 (6), 993–997.
(19) Shimoni, Y.; Chen, K.; Emmett, T.; Kargacin, G. Aldosterone and
the autocrine modulation of potassium currents and oxidative
stress in the diabetic rat heart. Br. J. Pharmacol. 2008, 154 (3),
675–687.
(20) Shiragy, H.; Xue, C. Local renal aldosterone production induces
inflammation and matrix formation in kidneys of diabetic rats.
Exp. Physiol. 2008, 93 (7), 817–824.
(21) Huang, B. S.; White, R. A.; Jeng, A. Y.; Leenen, F. H. Role of
central nervous system aldosterone synthase and mineralocorticoid
receptors in salt-induced hypertension in Dahl salt-sensitive rats.
Am. J. Physiol.: Regul. Integr. Comp. Physiol. 2009, 296 (4), R994–
R1000.
Acknowledgment. This research is supported by the Dutch
Technology Foundation STW, Applied Science Division of
NWO, and the Technology Program of the Ministry of
Economic Affairs, grant no. MFA 6504.
Supporting Information Available: Chemical structures of
known CYP11B2 inhibitors, results of the Y-scrambling analy-
sis, performance of the GoldScore in various protein conforma-
tions, Gold docking values for the different protein models. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Chai, W.; Danser, A. H. Why are mineralocorticoid receptor
antagonists cardioprotective? Naunyn Schmiedeberg’s Arch. Phar-
macol. 2006, 374 (3), 153–162.
€
(22) Ulmschneider, S.; Muller-Vieira, U.; Klein, C. D.; Antes, I.;
Lengauer, T.; Hartmann, R. W. Synthesis and evaluation of
(pyridylmethylene)tetrahydronaphthalenes/-indanes and structu-
rally modified derivatives: potent and selective inhibitors of aldo-
sterone synthase. J. Med. Chem. 2005, 48 (5), 1563–1575.
(2) Struthers, A. D. Aldosterone: cardiovascular assault. Am. Heart. J.
2002, 144 (5), S2–S7.
(3) Weber, M. A. Clinical implications of aldosterone blockade. Am.
Heart. J. 2002, 144 (5), S12–S18.
€
(23) Voets, M.; Antes, I.; Scherer, C.; Muller-Vieira, U.; Biemel, K.;
Barassin, C.; Marchais-Oberwinkler, S.; Hartmann, R. W. Hetero-
aryl-substituted naphthalenes and structurally modified deriva-
tives: selective inhibitors of CYP11B2 for the treatment of
congestive heart failure and myocardial fibrosis. J. Med. Chem.
2005, 48 (21), 6632–6642.
(4) Lea, W. B.; Kwak, E. S.; Luther, J. M.; Fowler, S. M.; Wang, Z.;
Ma, J.; Fogo, A. B.; Brown, N. J. Aldostereone antagonism or
synthase inhibition reduces end-organ damage induced by treat-
ment with angiotensin and high salt. Kidney Int. 2009, 75 (9), 936–
944.
(5) Sato, A.; Hayashi, K.; Naruse, M.; Saruta, T. Effectiveness
of aldosterone blockade in patients with diabetic nephropathy.
Hypertension 2003, 41 (1), 64–68.
(6) Schrier, R. W. Water and sodium retention in edematous disorders:
role of vasopressin and aldosterone. Am. J. Med. 2006, 119 (7),
S47–S53.
(7) Pitt, B.; Williams, G.; Remme, W. J.; Martinez, F.; Lopez-Sendon,
J.; Zannad, F.; Neaton, J.; Roniker, B.; Hurley, S.; Burns, D.;
Bittman, R.; Kleiman, J. The EPHESUS trial: eplerenone in
patients with heart failure due to systolic dysfunction complicating
acute myocardial infarction. Cardiovasc. Drugs Ther. 2001, 15 (1),
79–87.
€
(24) Voets, M.; Antes, I.; Scherer, C.; Muller-Vieira, U.; Biemel, K.;
Marchais-Oberwinkler, S.; Hartmann, R. W. Synthesis and eva-
luation of heteroaryl-substituted dihydronaphthalenes and in-
denes: potent and selective inhibitors of aldosterone synthase
(CYP11B2) for the treatment of congestive heart failure and
myocardial fibrosis. J. Med. Chem. 2006, 49 (7), 2222–2231.
(25) Heim, R.; Lucas, S.; Grombein, C. M.; Ries, C.; Schewe, K. E.;
Negri, M.; Muller-Vieira, U.; Birk, B.; Hartmann, R. W. Over-
coming Undesirable CYP1A2 Inhibition of Pyridylnaphthalene-
Type Aldosterone Synthase Inhibitors: Influence of Heteroaryl
Derivatization on Potency and Selectivity. J. Med. Chem. 2008, 51
(16), 5064–5074.
(26) de Graaf, C.; Vermeulen, N. P. E.; Feenstra, K. A. Cytochrome
P450 in silico: an integrative modeling approach. J. Med. Chem.
2005, 48 (8), 2725–2755.
(27) Lucas, S.; Heim, R.; Negri, M.; Antes, I.; Ries, C.; Schewe, K. E.;
Bisi, A.; Gobbi, S.; Hartmann, R. W. Novel aldosterone synthase
inhibitors with extended carbocyclic skeleton by a combined
ligand-based and structure-based drug design approach. J. Med.
Chem. 2008, 51 (19), 6138–6149.
(28) Lucas, S.; Heim, R.; Ries, C.; Schewe, K. E.; Birk, B.; Hartmann,
R. W. In vivo active aldosterone synthase inhibitors with improved
selectivity: lead optimization providing a series of pyridine sub-
stituted 3,4-dihydro-1H-quinolin-2-one derivatives. J. Med. Chem.
2008, 51 (24), 8077–8087.
(8) Pitt, B.; Remme, W.; Zannad, F.; Neaton, J.; Martinez, F.;
Roniker, B.; Bittman, R.; Hurley, S.; Kleiman, J.; Gatlin, M.
Eplerenone, a selective aldosterone blocker, in patients with left
ventricular dysfunction after myocardial infarction. N. Engl. J.
Med. 2003, 348 (14), 1309–1321.
(9) Hartmann, R. W.; Muller, U.; Ehmer, P. B. Discovery of selective
CYP11B2 (aldosterone synthase) inhibitors for the therapy of
congestive heart failure and myocardial fibrosis. Eur. J. Med.
Chem. 2003, 38 (4), 363–366.
(10) Krum, H.; Nolly, H.; Workman, D; He, W.; Roniker, B.; Krause,
S.; Fakouhi, K. Efficacy of eplerenone added to renin-angiotensin
blockade in hypertensive patients. Hypertension 2002, 40 (2), 117–
123.
(29) Roumen, L.; Sanders, M. P. A.; Pieterse, K.; Hilbers, P. A. J.; Plate,
R.; Custers, E.; de Gooyer, M.; Smits, J. F. M.; Beugels, I.; Emmen,
J.; Ottenheijm, H. C. J.; Leysen, D.; Hermans, J. J. R. Construction
of 3D models of the CYP11B family as a tool to predict ligand
binding characteristics. J. Comput.-Aided. Mol. Des. 2007, 21 (8),
455–471.
(11) Sica, D. A. Pharmacokinetics and pharmacodynamics of miner-
alocorticoid blocking agents and their effects on potassium home-
ostasis. Heart. Failure Rev. 2005, 10 (1), 23–29.
ꢀ
(12) Rousseau, M. F.; Gurne, O.; Duprez, D.; van Mieghem, W.;
Robert, A.; Ahn, S.; Galanti, L.; Ketelslegers, J.-M. Beneficial
neurohormonal profile of spironolactone in severe congestive heart
failure. J. Am. Coll. Cardiol. 2002, 40 (9), 1596–1601.
€
(30) Ulmschneider, S.; Muller-Vieira, U.; Mitrenga, M.; Hartmann,
R. W.; Oberwinkler-Marchais, S.; Klein, C. D.; Bureik, M.;
Bernhardt, R.; Antes, I.; Lengauer, T. Synthesis and evaluation
of imidazolylmethylenetetrahydronaphthalenes and imidazolyl-
methyleneindanes: potent inhibitors of aldosterone synthase.
J. Med. Chem. 2005, 48 (6), 1796–1805.
(13) Fisher, A.; Friel, E. C.; Bernhardt, R.; Gomez-Sanchez, C.;
Connell, C.; Fraser, J. M. C.; Davies, E. Effects of 18-hydroxylated
steroids on corticoid production by human aldosterone synthase
and 11beta-hydroxylase. J. Clin. Encrinol. Metab. 2001, 86 (9),
4326–4329.
(31) Belkina, N. V.; Lisurek, M.; Ivanov, A. S.; Bernhardt, R. Model-
ling of three-dimensional structures of cytochromes P450 11B1 and
11B2. J. Inorg. Biochem. 2001, 87 (4), 197–207.
(32) Dorr, H. G.; Kuhnle, U.; Holthausen, H.; Bidlingmaier, F.; Knorr,
D. Etomidate: a selective adrenocortical 11 beta-hydroxylase
inhibitor. Klin. Woschenschr. 1984, 62 (21), 1011–1013.
(14) Okamoto, M.; Nonaka, Y. Molecular biology of rat steroid 11
beta-hydroxylase [P450(11 beta)] and aldosterone synthase
[P450(11beta, aldo)]. Eur. J. Biochem. 1991, 202 (3), 897–902.
(15) Fiebeler, A.; Nussberger, J.; Shagdarsuren, E.; Rong, S.; Hilfenhaus,
G.; Al-Saadi, N.; Dechend, R.; Wellner, M.; Meiners, S.;
Maser-Gluth, C.; Jeng, A. Y.; Webb, R. L.; Luft, F. C.; Muller,
D. N. Aldosterone synthase inhibitor ameliorates angiotensin II-
induced organ damage. Circulation 2005, 111 (223), 3087–3094.
(16) Minnaard-Huiban, M.; Emmen, J. M. A.; Roumen, L.; Beugels,
I. P. E.; Cohuet, G. M. S.; van Essen, H.; Ruijters, E.; Pieterse, K.;
Hilbers, P. A. J.; Ottenheijm, H. C. J.; Plate, R.; de Gooyer, M. E.;
Smits, J. F. M.; Hermans, J. J. R. Fadrozole Reverses Cardiac
Fibrosis in Spontaneously Hypertensive Heart Failure Rats: Discordant
€
(33) Stark, H.; Purand, K.; Huls, A.; Ligneau, X.; Garbarg, M.;
Schwartz, J.-C.; Schunack, W. [125I]Iodoproxyfan and related
compounds: a reversible radioligand and novel classes of antago-
nists with high affinity and selectivity for the histamine H3 receptor.
J. Med. Chem. 1996, 39 (6), 1220–1226.
(34) Erdmann, B.; Denner, K.; Gerst, H.; Lenz, D.; Bernhardt, R.
Human adrenal CYP11B1: localisation by in situ hybridization

Article
and functional expression in cell cultures. Endocr. Res. 1995, 21
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 4 1725
(38) MOE (The Molecular Operating Environment) Version 2007.08;
Chemical Computing Group Inc.: Sherbrooke Street West, Suite 910,
Montreal Canada H3A 2R7. http://chemcomp.com
(39) Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.;
Taylor, R. D. Improved protein-ligand docking using GOLD.
Proteins 2003, 52 (4), 609–623.
(40) Kirton, B.; Murray, C. W.; Verdonk, M. L.; Taylor, R. D.
Prediction of binding modes for ligands in the cytochromes P450
and other heme-containing proteins. Proteins: Struct., Funct.,
Bioinf. 2005, 58 (4), 836–844.
(1-2), 425–435.
(35) Denner, K.; Doehmer, J.; Bernhardt, R. Cloning of CYP11B1 and
CYP11B2 from normal human adrenal and their functional ex-
pression in COS-7 and V79 Chinese hamster cells. Endocr. Res.
1995, 21 (1-2), 443–448.
(36) Bottner, B.; Denner, K.; Bernhardt, R. Conferring aldosterone
synthesis to human CYP11B1 by replacing key amino acid residues
with CYP11B2 ones. Eur. J. Biochem. 1998, 252 (3), 458–466.
(37) Millet, R.; Domarkas, J.; Houssin, R.; Gilleron, P.; Goossens,
ꢀ
J.-F.; Chavatte, P.; Loge, C.; Pommery, N.; Pommery, J.;
Henichart, J.-P. Potent and selective farnesyl transferase inhibi-
(41) Cornish-Bowden, A. Analysis of Enzyme Kinetic Data; Oxford
ꢀ
University Press: Oxford, UK, 1995.
tors. J. Med. Chem. 2004, 47 (27), 6812–6820.