Enantioselective nonsteroidal aromatase inhibitors identified through a multidisciplinary medicinal chemistry approach

Article abstract of DOI:10.1021/jm058042r

To identify enantioselective nonsteroidal aromatase inhibitors, a multidisciplinary medicinal chemistry approach was pursued. First, our earlier CoMFA model [Bioorg. Med. Chem. 1998, 6, 377-388] was extended taking purposely into account previously discov

Full text of DOI:10.1021/jm058042r

7282
J. Med. Chem. 2005, 48, 7282-7289
Enantioselective Nonsteroidal Aromatase Inhibitors Identified through a
Multidisciplinary Medicinal Chemistry Approach
Andrea Cavalli,† Alessandra Bisi,† Carlo Bertucci,† Carlo Rosini,‡ Anja Paluszcak,§ Silvia Gobbi,† Egidio Giorgio,‡
Angela Rampa,† Federica Belluti,† Lorna Piazzi,† Piero Valenti,† Rolf W. Hartmann,§ and Maurizio Recanatini*,†
Department of Pharmaceutical Sciences, University of Bologna, Via Belmeloro, 6, I-40126 Bologna, Italy, Department of
Chemistry, University of Basilicata, Via N. Sauro, 85, I-85100 Potenza, Italy, and Pharmaceutical and Medicinal Chemistry,
Saarland University, P. O. Box 151150, D-66041 Saarbru¨cken, Germany
Received July 25, 2005
To identify enantioselective nonsteroidal aromatase inhibitors, a multidisciplinary medicinal
chemistry approach was pursued. First, our earlier CoMFA model [Bioorg. Med. Chem. 1998,
6, 377-388] was extended taking purposely into account previously discovered enantioselective
aromatase inhibitors. The 3D QSAR model was then exploited to design chiral ligands, whose
configurational assignment was obtained, after HPLC separation, by means of a combination
of circular dichroism measurements and time dependent density functional calculations. Finally,
the new enantiomeric inhibitors were separately tested to ascertain both their potency against
the cytochrome P450 aromatase (CYP19; EC 1.14.14.1), and their selectivity relative to another
enzyme of the P450 family. A satisfactory agreement between experimental and predicted data
allowed us to assert that a properly built “enantioselective CoMFA model” might constitute a
useful tool for addressing enantioselective ligands design.
Chart 1
Introduction
One of the approaches currently exploited for the
treatment of postmenopausal estrogen dependent breast
cancer is to reduce the estrogen’s concentration in blood
by means of aromatase inhibitors.1 The aromatase
enzymatic complex is formed by two proteins, namely
the cytochrome P450 XIX (CYP19) and NADPH-cyto-
chrome P450 reductase. The reaction catalyzed by the
complex is the aromatization of the A ring of androgen
substrates that leads to estrogen hormones. Nonsteroi-
dal aromatase inhibitors competitively interact with the
enzyme, thus displacing the natural substrate, and
eventually give rise to the reduction of the estrogen’s
blood concentration.2 Among different classes of aro-
matase inhibitors, the azole family is going to hold an
increasingly prominent position, as demonstrated by
azole derivatives marketed drugs.2
Very recently, new series of nonsteroidal aromatase
inhibitors have been synthesized by different groups
working in the field,3,4 and many efforts are also being
devoted to the search of dual aromatase-steroid sulfa-
tase inhibitors,5 which might represent a new genera-
tion of drugs targeting estrogen biosynthesis.6
In an earlier paper, we presented a new series of
nonsteroidal aromatase inhibitors,7 which were de-
signed taking advantage of previously derived CoMFA
models.8,9 In that paper, we reported that CoMFA-
driven suitable modifications of the (di)benzopyran-4-
one nucleus allowed us to obtain potent and selective
aromatase inhibitors. The most potent inhibitors of that
series belonged to the azole class, and turned out to be
not only almost as potent as the well-known drug
fadrozole (IC50 ) 40 and 17 nM, respectively) but also
selective with respect to P450 17, another cytochrome
of the P450 family involved in androgen biosynthesis.7
In the present paper, we report a further extension
of our previous CoMFA models,8,9 now comprising
seventy structurally different nonsteroidal aromatase
inhibitors. Seven of the twenty-one newly added mol-
ecules bear at least one stereogenic center and show
enantioselectivity toward the aromatase enzyme. Re-
markably, fadrozole itself is highly enantioselective
toward P450 19,10 as its most active enantiomer (S,
pIC50 ) 7.77) is two log units more potent than the least
active one (R, pIC50 ) 5.64) (Chart 1). Notably, since
our earliest 3D QSAR studies, both enantiomers of
fadrozole were taken into account in the CoMFA analy-
ses.8 The newly developed 3D QSAR model turned out
to be statistically robust and, more important, to be able
to predict different inhibitory activities for enantiomers.
The obtained CoMFA model was then exploited to
address the synthesis of new compounds purposely
designed carrying a stereogenic center (compounds 1-3,
Chart 2). The compounds were synthesized, and the
racemic mixtures were solved in the optically active
forms by means of preparative HPLC employing chiral
stationary phases. The configurational assignment of
the pure enantiomers was accomplished by means of a
combination of circular dichroism (CD) measurements
and time dependent density functional theory (TD-DFT)
calculations. Finally, the new inhibitors were biologi-
cally tested against both CYP19 and CYP17 enzymes.
All of them were active at nanomolar level, selective
* Corresponding author. E-mail: maurizio.recanatini@unibo.it. Phone/
fax: +39 051 2099720/34.
† University of Bologna.
‡ University of Basilicata.
§ Saarland University.
10.1021/jm058042r CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/15/2005

Enantioselective Nonsteroidal Aromatase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23 7283
Table 1. Statistics of the PLS Analyses
Chart 2
component
q2
scross
R2
s
1
2
3
4
5
6
0.600
0.692
0.726
0.765
0.791
0.793
0.704
0.623
0.591
0.551
0.524
0.526
0.665
0.833
0.884
0.931
0.958
0.975
0.643
0.459
0.385
0.299
0.234
0.181
Scheme 1a
ing the chiral stationary phases Chiralcel OD (1 and 3) and
Chiralcel OJ (2). Both stationary phases used derivatized
cellulose as chiral selector.
Configurational Assignment. The absolute configuration
of the separated enantiomers was determined by means of a
combined approach based on experimental measurements and
ab initio calculations of the chiroptical properties, i.e., optical
rotation (OR) and CD.
Ab initio calculations of the OR were carried out at the TD-
DFT/B3LYP/6-31G(d) level of theory and using GIAO orbitals
to guarantee gauge-independency, with GAUSSIAN03 soft-
ware.13 CD calculations were carried out by using the same
TD-DFT/B3LYP/6-31G(d) approach in the dipole-velocity for-
malism that guarantees the origin-independency. Theoretical
CD spectra were obtained by overlapping Gaussian functions
for each transition according to the expression
N
1
1
2
i
∆ꢀ(E) )
∆EiRi e-[(E-∆E /2σ)]
∑
2.297 × 10-39
i
a
x
2πσ
Reagents: (i) piperidine, 150 °C/3 h; (ii) N-bromosuccinimide,
CCl4, reflux 5 h; (iii) imidazole, CH3CN, reflux 6 h.
In the equation, σ is the width at half-height and ∆Ei and Ri
are the excitation energies (in eV) and rotary strengths (in
erg esu cm/gauss) for transition i, respectively. Here, a σ value
of 0.2 eV was used.14 The conformational search was performed
by means of SPARTAN02 at both the molecular mechanics
(MMFF94 force field) and semiempirical (PM3) levels. The
obtained geometries were then fully optimized (B3LYP/6-31G-
(d) level) by means of GAUSSIAN03.
Biological Assays. For the determination of biological
activity, the human enzymes were used. In the case of CYP19
the tritium water method15 and placental microsomes were
employed, whereas for CYP17 the enzyme was prepared16 from
Escherichia coli overexpressing human CYP17.17
with respect to CYP17, and, remarkably, enantioselec-
tive toward the biological target. The fairly good agree-
ment between the experimental and calculated pIC50
values for the single enantiomers showed that our work
strategy was able to correctly address the design and
characterization of new chiral derivatives acting enan-
tioselectively as aromatase inhibitors. In this respect,
a multidisciplinary approach to the design of enantio-
selective inhibitors is here presented.
Methods
Results and Discussion
Computational Chemistry. The molecular models were
built by properly modifying crystallographic skeletons re-
trieved from the Cambridge Structural Database.11 The models
were geometrically optimized and then submitted to Monte
Carlo conformational searches. The conformers were further
classified by means of cluster analyses, and finally optimized
at the AM1 semiempirical level. In addition, representative
conformers of the new compounds were geometrically opti-
mized within the density functional theory (DFT) framework
at the B3LYP/6-31G(d) level of theory. The partial atomic
charges of all 70 compounds were calculated according to the
RESP procedure.12
The 3D QSAR analyses were carried out using the CoMFA
method implemented in the SYBYL 6.9 package. The align-
ment of the molecules within the Cartesian space was ac-
complished according to the atom-by-atom procedure using
both enantiomers of fadrozole as templates. For a detailed
description of the alignment strategy, the reader can refer to
the previously reported CoMFA models.8,9
CoMFA Model. The present CoMFA was built by
adding 21 molecules to the model developed in 1998,9
and represents a further extension of previous ones.8,9
In particular, the test set of the 1998 model9 (8 mol-
ecules) along with the series of new aromatase inhibitors
reported in 20017 (13 molecules) was added to the
original training set of 49 molecules.9 Interestingly, 7
of the newly added molecules carried at least one
stereogenic center and showed enantioselectivity toward
the biological target.
The statistics of the PLS analyses obtained with the
70 compounds training set are shown in Table 1. In
addition, in the Supporting Information (SI), the struc-
tures of the complete set of molecules of the present
CoMFA model (Figure 1SI) and the experimental and
calculated pIC50 values of the whole set of compounds
(Table 1SI) are reported. Also, a plot of the calculated
vs experimental pIC50 is provided in Figure 2SI. Despite
some poorly recalculated activity values, the extended
CoMFA model is of acceptably good quality, and the
overall characteristics of the previous model9 are main-
tained (the interested reader is referred to the original
paper for details on the interpretation of the CoMFA
model).
Chemical Synthesis. The newly designed chiral molecules
were prepared as outlined in Scheme 1. The chromanone was
treated with the selected 4-substituted benzaldehyde in the
presence of piperidine to afford the 3-(4-substituted)benzyl
derivatives 4-6, which were then brominated with N-bromo-
succinimide with a catalytic amount of benzoyl peroxide to give
intermediates 7-9. Reaction with imidazole afforded the final
compounds 1-3 reported in Chart 2.
The newly synthesized compounds 1-3 were obtained in
optically active form by preparative HPLC resolution employ-

7284 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23
Cavalli et al.
Figure 1. CoMFA contour maps plotted in the 3D space around S- (orange carbon atoms) and R-fadrozole (green carbon atoms)
(A and B), and around R- (orange carbon atoms) and S-3 (green carbon atoms) (C and D). (A and C) Steric contours: The regions
where increasing the volume increases the activity are green (0.020 level), and the regions where increasing the volume decreases
the activity are yellow (-0.015 level). (B and D) Electrostatic contours: The red regions indicate increase of activity with increasing
positive charge (0.028 level), and the blue contours indicate increase of activity with increasing negative charge (-0.028 level).
The contour maps of the CoMFA model derived with
five components (bold in Table 1) are depicted in Figure
1A,B. In agreement with our previous models,8,9 the
sterically allowed regions (green in Figure 1A) were
mainly located around the para-CN phenyl ring of
S-fadrozole, whereas the para-CN phenyl ring of R-
fadrozole partially protruded into the sterically disal-
lowed region (yellow in Figure 1A). In Figure 1B, the
positive (red) and negative (blue) electrostatic regions
are reported. While the negative CoMFA electrostatic
contours were of difficult interpretation, as they split
in several regions around the template skeleton, the
positive (red) CoMFA contour partially surrounded the
negatively charged atoms of the para-CN phenyl ring
of R-fadrozole (green carbon atoms, oriented toward the
reader). This result seems contradictory, but it actually
means that an increase of positive charges would be
required to increase the inhibitory potency in that region
of the Cartesian space where the poorly active R-
fadrozole locates its negatively charged phenyl ring.
The alignment strategy followed to obtain CoMFA
predictions of the activity of the enantiomers of 1-3 was
simply that of overlapping the (p-substituted)benzylimi-
dazole moiety of the new compounds’ enantiomers onto
that of S- and R-fadrozole, respectively. This choice was
made to be consistent with the alignment of all enan-
tiomer pairs present in the training set. In Figure 1C,D,
both enantiomers of 3 are shown along with the steric
(Figure 1C) and electrostatic (Figure 1D) CoMFA con-
tour maps. The para-CN phenyl ring of the R enanti-
omer (orange carbon atoms in Figure 1C,D) of 3 was
completely embedded in the positive steric CoMFA
contour (Figure 1C), whereas its S enantiomer (green
carbon atoms in Figure 1C,D) placed the same moiety
into a positive electrostatic region (Figure 1D).
The CoMFA equation was able to predict different
inhibitory potencies for each enantiomer of all of the
newly designed compounds, and the biological activities
of the new compounds were predicted as reported in
Table 2. In particular, the R enantiomers were always
calculated more potent than the S ones.
Based on the present CoMFA model, a new series of
chiral compounds was designed (1-3, Chart 2), by
modifying our previously reported chromone deriva-
tives7 in such a way as to keep the pharmacophoric (p-
substituted)benzylimidazole moiety and to introduce an
asymmetric center on the methylene carbon atom.
After organic synthesis, enantiomer separation, con-
figurational assignment, and biological evaluation (see
below), the predicted and observed data as reported in
Table 2 were compared. It turned out that the CoMFA
model was able to fairly well predict the inhibitory

Enantioselective Nonsteroidal Aromatase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23 7285
Table 3. Stereochemical Characterization of the Enantiomeric
Fractions
OR
CD
(2-propanol)
(2-propanol)
at 589 nm (high energy couplet)
compd
fraction
ee (%)
[R]D
∆ꢀ (λ, nm)
less retained (R) 100
most retained (S) 98.2
less retained (R)
most retained (S) 96.2
less retained (R) 98.2
most retained (S) 96.4
+46
-61
+54
-51
+94
-71
3.1(264); -2.7(238)
-2.7(264); 2.2(238)
18.2(233); -10.9(217)
-15.3(233); 12.1(217)
10.3(242); -3.5(222.5)
-9.7(242); 3.3(222.5)
1
2
3
97.2
Figure 2. The enantioselective resolution of rac-3 (A), upon
chiralcel OD; enantiomeric excess (ee) assay of the enriched
(S)-3 sample (B) [rt(first eluted enantiomer) 36.4 min; rt(second
eluted enantiomer) 40.6 min].
Table 2. Experimental and Calculated Inhibitory Activity of
Molecules 1-3
∆pIC50exp ∆pIC50calc
(R - S) (R - S)
a
compd enantiomer pIC50calc pIC50exp
∆
R
S
R
S
R
S
7.74
6.15
7.60
6.44
7.43
6.27
8.64 -0.90
7.02 -0.87
1
2
3
1.62
0.77
0.76
1.59
1.16
1.16
Figure 3. The UV (A) and CD (B) spectra of the first eluted
enantiomers of 1 (green), 2 (red), and 3 (black) in 2-propanol
solution.
7.31
0.29
6.54 -0.10
6.96
6.20
0.47
0.07
a Calculated from the five-component model (see Table 1).
Configurational Assignment. With regard to the
experimental determination of chiroptical properties,
Figure 2 reports the enantioselective resolution of 3
(Figure 2A), and an example of the enantiomeric excess
(ee) assay of one enantiomeric fraction (Figure 2B). The
ee values determined for the optically active forms of
compounds 1-3 are reported in Table 3 along with
optical rotary power (at 589 nm) and the CD data, while
the absorption and CD spectra are collected in Figure
3. The availability of these chiroptical data could offer
the way to assign the absolute configuration (AC) of the
enantiomers of 1-3. In fact, much recent progress has
been made in the correct interpretation of chiroptical
data by means of both coupled oscillator methods20-27
and ab initio calculations allowing then a reliable
assignment of the molecular AC: some different pro-
grams13,28,29 for calculating ab initio the OR30-33 at
any wavelength and even the CD spectra are now
available.14,34-36 This type of approach to the AC as-
signment has become so powerful and reliable that it
has been possible to show37 that AC determinations
carried out by X-ray diffraction analysis, considered the
safest method to perform this structural determination,
are incorrect and must therefore be revised. Therefore
we decided to arrive at the AC of the present compounds
just analyzing their chiroptical properties.
potencies of 1-3. Actually, the activity of both the
enantiomers of 1 was quite underpredicted (∆ ) -0.90
and -0.87 for R and S, respectively). In particular, con-
cerning R-1, this was likely due to its biological activity
(pIC50 ) 8.64), which was somewhat higher than the
average activities of the 3D QSAR training set. As a
matter of fact, the general limit of the CoMFA approach
in predicting potencies and/or affinities of the most and
least active compounds is well-known.18,19 All of the
other new derivatives were reasonably well predicted.
In particular, the S enantiomers of 2 and 3 were pre-
dicted with errors as small as -0.10 and 0.07, respec-
tively. More interesting, the CoMFA model was able to
well predict the difference in inhibitory potency between
the enantiomers of each pair. In Table 2, the ∆pIC50exp
-
(R-S) and ∆pIC50calc(R-S) (i.e., the experimental and
calculated differences between the biological activity of
the enantiomers of 1-3) are also reported. All of the
predictions (1.62 vs 1.59, 0.77 vs 1.16, and 0.76 vs 1.16
for 1, 2, and 3, respectively) may be considered good.
We conclude that the present CoMFA model for a
series of nonsteroidal aromatase inhibitors can reason-
ably be considered as an example of enantioselective 3D
QSAR model.

7286 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23
Cavalli et al.
Figure 4. The weighted average CD spectrum of S-3 (green) compared to the experimental CD spectra recorded for the two
enantiomeric fractions in the rac-3 HPLC resolution [(+)-enantiomer, less retained fraction blue; (-)-enantiomer, most retained
fraction magenta].
Taking into account that intense bands are present
in the absorption spectra (i.e., electronically allowed
transitions occur), the analysis of the CD spectra by
means of coupled oscillator techniques20-27 after a
suitable conformational analysis, and a correct assign-
ment of the absorption bands in the electronic spectrum
could be a possible strategy to provide the absolute
configuration (AC). However, for the present series of
compounds, some difficulties derived by the fact that,
while the para substituted phenyl group represented a
well-known chromophore, not much information is
reported in the literature regarding the chromenone
chromophore.38 In addition, a further problem was
represented by the location of the point-dipoles, which,
in coupled oscillator treatments, represent the allowed
transitions. The chromenone chromophore is not sym-
metric, so a decision could not be made simply on
symmetry grounds. As a consequence, in order to find
the correct location of point-dipoles39-42 further calcula-
tions would have been needed.
To avoid all of these problems we decided to tackle
the problem of the configurational assignment of (+)-
1-3 by means of the quantum mechanical methods of
calculating chiroptical data: in particular we decided
to start with the ab initio calculation of the optical
rotation (OR) at 589 nm. To this end, OR calculations
for 3 were performed, because it presents the simple
symmetric CN substituent, made only by light atoms,
and shows low energy Cotton effects, which determine
the sign and order of magnitude of experimental OR.
Thus, small basis set calculations could be confidently
carried out,43,44 significantly reducing the computational
effort.
Computations at the TD-DFT level were carried out,
aimed at calculating OR values to be compared to the
measured ones. First, detailed conformational analyses
were performed to determine the conformer populations
of the compound under study. Arbitrarily, the S AC for
3 was assumed. A first conformational search carried
out at the molecular mechanics level (MMFF9445 of
Spartan0246) provided 14 different conformers lying in
a range of 3 kcal/mol. All of the conformations were then
fully optimized at the DFT level, giving only four
conformers lying in a range of 2 kcal/mol. For these
conformations that were real energy minima (no imagi-
nary frequencies), the free energy values were calcu-
lated. Actually, a more accurate evaluation of the con-
former populations exploiting free energy values instead
of electronic energies was pursued taking into account
that, since four structures are involved in the OR
calculation, the final result could be heavily affected by
the conformer populations. In Figure 3SI of the Sup-
porting Information, the 3D models of the four conform-
ers of S-3, with their free energy and the corresponding
population coming from Boltzmann statistics, are re-
ported. In Table 2SI, the OR of each conformation along
with the conformer population is reported.
The OR calculations were carried out for each con-
former at the TD-DFT level and using GIAO orbitals to
guarantee gauge-independency. The final result was a
predicted value for S-3 of -5. This number was not
satisfactory, because it represented only 5% of the
optical rotary power (Table 3), while only a prediction
of at least 60-70% of the experimental value could
reasonably constitute a reliable theoretical estimate.
Therefore, to provide a safer configurational assignment,
further ab initio calculations of the CD spectrum in the
range 400-200 nm were carried out. In detail, the 35
lowest-energy transitions were calculated by the same
TD-DFT approach. Then, the theoretical CD spectra
were obtained as described in the Configurational
Assignment section of the Methods.
The UV and CD spectra of the first eluted enantiomer,
as obtained in 2-propanol solution, are reported in
Figure 3A and 3B, respectively. Looking at the values
of the OR (Table 3), it could be immediately noticed that
the first eluted fraction was always dextrorotatory,
suggesting that the less retained enantiomers should
bear the same AC. In addition, the CD spectra of the
first eluted fractions (Figure 3B) were very similar,
showing a sequence of positive/negative Cotton effects
between 280 and 220 nm (the positive one at about 230
nm being particularly intense). In the range 350-280
nm, only weak CD bands were present, which therefore
could be considered less useful for configurational
correlations. On these grounds, we could safely assume
that all of the compounds were eluted with the same
order, irrespective of both the para substitution of the
phenyl ring and the used chiral stationary phase.
In Figure 4, the weighted average CD spectrum of S-3
is compared to the experimental CD spectra recorded
for the two enantiomeric fractions. The calculated
spectrum of S-3 and that of the less retained enantiomer
are in a mirror image relationship. Thus, we could
conclude that the AC of the less retained (+) fraction of
3 was R.

Enantioselective Nonsteroidal Aromatase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23 7287
Table 4. Biological Activities of Compounds 1-3 and
lectivity toward the biological counterpart. 3D QSAR
studies, organic synthesis, enantiomer separation, con-
figurational assignment, and biological assays were
performed in strict collaboration to face such an impor-
tant medicinal chemistry issue. In details, a new
CoMFA model was obtained, adding 21 new molecules
to a previously derived model. Since some of the added
molecules bore a chiral atom and were enantioselective
toward CYP19, the new CoMFA model was purposely
exploited to design new chiral inhibitors. The 3D QSAR
model predicted different activities for either enanti-
omer. The new molecules were synthesized and the
enantiomers separated by means of preparative HPLC
using a chiral stationary phase. The absolute configu-
rational assignment was obtained by means of a com-
bined approach based on CD measurements and TD-
DFT calculations. Finally, both enantiomers of the three
new derivatives were tested against human P450 19
enzyme and the experimental data were compared to
the 3D QSAR equation predictions. The fairly good
agreement between observed and calculated inhibition
data allowed us to conclude that a properly built
“enantioselective CoMFA model” might constitute a
viable tool to address the design of enantioselective
ligands.
Fadrozole against CYP19 (Aromatase) and CYP17
CYP19a
IC50 (nM)
(( se)
CYP17b
% inhibition
compd
enantiomer
at 2.5 µM
R
S
R
S
R
S
S
2.3 ((1.1)
96 ((15)
20
12
17
13
6
1
2
3
49 ((11)
290 ((32)
110 ((28)
630 ((55)
17 ((7)
6
fadrozole
7
a Substrate 1â[3H]androstenedione 500 nM. b Substrate proges-
terone 25 µM.
The calculated CD spectrum did not reproduce the
sign of the band at lower energy. This behavior could
be due to the influence of the nature of the solvent on
the related electronic transition. Indeed, the CD band
of (R)-3 changed sign passing from the polar 2-propanol
(Figure 3) to the less polar solvent chloroform (Figure
4SI). The same behavior, i.e., inversion of the sign of
the lowest energy CD band by changing the solvent
polarity, was observed also in the case of compound
(R)-1 (Figure 3, 2-propanol; Figure 5SI, chloroform).
Biological Activity. CYP19 was prepared from
human placenta.47 The microsomes were incubated with
1â[3H]androstenedione, and product formation was
monitored by measuring the tritiated water formed
during the aromatization of the substrate.47 CYP17 was
prepared from E. coli overexpressing the human en-
zyme.16,17 For the inhibition test an enzyme preparation
from the homogenate was incubated with progesterone.
The separation of the product from the substrate was
performed by HPLC.16 The experimentally determined
inhibitory potencies of 1-3 against aromatase are
reported in Table 4 expressed as IC50 values. The new
compounds are active at the nanomolar level, R-1 being
one of the most potent aromatase inhibitors so far
reported (IC50 ) 2.3 ( 1.1 nM). Moreover, none of the
newly synthesized compounds was significantly active
toward P450 17 (Table 4), thus indicating that the set
showed a good selectivity for P450 19 with respect to
the androgen forming counterpart. The relative potency
of R-1 with respect to the marketed drug S-fadrozole is
7.4. This molecule is also remarkably more potent than
nonsteroidal inhibitors of the azole series reported by
us in the previous paper.7 Likely, enantioselective
inhibitors might fit at best the aromatase active site
requirements. In this respect, it should be noted that
R-1, the most potent compound of the present series,
bears a NO2 group in the para position of the phenyl
ring, which might likely interact with a key serine
residue (Ser478) of the aromatase active site, as previ-
ously suggested to be the case for the CN group of
S-fadrozole.48,49 However, such an argument does not
help to explain the surprisingly low potency of R-3,
which similarly to fadrozole bears a CN group in the
para position of the phenyl ring. In this regard, further
docking experiments using a new homology-built model
will be performed in order to unravel this issue.
Experimental Section
Computational Methods. The conformational searches
were carried out using the Monte Carlo approach50 and the
MMFF94 force field45 as implemented in the MacroModel
software package.51 In the present study, the number of Monte
Carlo steps was set equal to 8000 and the trial conformation
was accepted if the energy was lower than that of the previous
conformation or if its energy was within an energy window of
100 kJ/mol. Then, the conformations were classified by means
of a cluster analysis52 using geometrical parameters as filtering
screens. Minimum energy conformations were further opti-
mized at the AM1 semiempirical level53 as implemented in
MOPAC. The atomic partial charges of each compound were
computed by means of the RESP procedure,12 i.e., fitting them
on an electrostatic potential calculated at the DFT level. S-3
was submitted to a further protocol of conformational analysis
carried out by means of the SPARTAN02 suite of software at
both the molecular mechanics (MMFF94 force field)45 and
semiempirical (PM3)54 levels.
All of the ab initio quantum chemical calculations were
carried out within the DFT framework using the hybrid
exchange and correlation functional B3LYP55 and the Gauss-
ian basis set 6-31G(d). The OR calculations on the inhibitor
S-3 were carried out following the time dependent DFT (TD-
DFT) approach and using GIAO orbitals to guarantee gauge-
independency. All DFT-based calculations were performed by
means of GAUSSIAN03.13
The 3D QSAR equations were obtained by means of the
comparative molecular field analysis (CoMFA) technique.56
The CoMFA fields were generated by using an sp3 carbon atom
with a formal charge of +1 as a probe, while the CoMFA region
was generated automatically around the molecules by fixing
a grid spacing of 2 Å. The statistical analyses were carried
out by applying the PLS procedure to the appropriate variables
and using the standard scaling method (COMFA_STD). Cross-
validated PLS runs were carried out to establish the optimal
number of components (the latent variables) to be used in the
final fitting models. The number of cross-validated groups was
always equal to the number of compounds (leave-one-out
procedure), and the optimal number of components (latent
variables) was chosen by considering the lowest standard error
of prediction (scross). Scrambling analyses were performed to
rule out the possibility of chance correlations. The CoMFA
studies were carried out by means of the SYBYL software
(Tripos Inc. St. Louis, MO).
Conclusions
A multidisciplinary study in the field of nonsteroidal
aromatase inhibitors was carried out aimed at discover-
ing a new way to design molecules bearing enantiose-

7288 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23
Cavalli et al.
Chemistry. General Methods. All melting points were
determined in open glass capillaries using a Bu¨chi apparatus
and are uncorrected. 1H NMR spectra were recorded on a
Varian Gemini 300 spectrometer in CDCl3 solutions, with Me4-
Si as the internal standard. Mass spectra were recorded on a
Waters ZQ 4000 apparatus operating in electrospray (ES)
mode. Elemental analyses were within 0.4% of theoretical
value unless otherwise indicated. Compounds were named
following IUPAC rules as applied by AUTONOM, a PC
software for systematic names in organic chemistry, Beilstein-
Institut and Springer.
Chiralcel OJ (2), from Daicel (Japan). Repeated injections
(0.2-0.3 mg) on the 0.46 cm i.d. x 25 cm columns allowed
collecting 3-6 mg of each fraction. The same columns were
employed for the enantiomeric excess (ee) determination of the
optically active samples. The mobile phase was a hexane/2-
propanol mixture, 1 mL min-1. The enantioselectivity was
calculated as R ) k2/k1, where k2 and k1 are the capacity factors
[k is defined as (tr - t0)/t0; tr ) retention time of the analyte,
t0 ) retention time of a nonretained solute] of the second and
first eluted enantiomers, respectively.
UV, CD, and Optical Rotation Measurements. Circular
dichroism and absorption spectra were carried out using a
Jasco J-810 spectropolarimeter (Jasco, Tokio, Japan) and a
Jasco V-530 spectrophotometer (Jasco, Tokio, Japan). All
measurements were carried out in 2-propanol at room tem-
perature, using 1-0.1 cm cells. Optical rotation measurements
were carried out at 589 nm (RD) in 2-propanol at room
temperature, using an AA1000 polarimeter (Optical Activity,
England), using a 10 cm cell.
3-(4-Nitro-benzyl)-chromen-4-one (4). A mixture of chro-
man-4-one (4.5 g, 0.03 mol), 4-nitrobenzaldehyde (4.6 g, 0.03
mol), and 0.4 mL of piperidine was heated at 150 °C for 3 h.
After cooling, the residue was crystallized from ethyl acetate,
to give 7.5 g (89.9%) of the desired compound, mp 171-172
1
°C. H NMR: δ 3.85 (s, 2H), 7.35-8.25 (m, 8H, arom + H-2).
3-(4-Bromo-benzyl)-chromen-4-one (5). Using the previ-
ous procedure and starting from 4.5 g (0.03 mol) of chroman-
4-one and 5.5 g (0.03 mol) of 4-bromobenzaldehyde (0.03 mol),
Inhibitor Assays. Human placental CYP19 was prepared
as described,47 and the assay was performed according to the
same reference.47 For the CYP17 inhibition tests E. coli cells
overexpressing human CYP17 and P450 reductase were
used.17 These cells were homogenized, and the 50.000 g
sediment was employed as enzyme source.16 The test was run
as described.16
1
compound 5 (7.84 g, 83%) was obtained, mp 164-167 °C. H
NMR: δ 5.30 (s, 2H), 6.90-8.15 (m, 8H, arom + H-2).
4-(4-Oxo-4H-chromen-3-ylmethyl)-benzonitrile (6). Us-
ing the previous procedure and starting from 4.5 g (0.03 mol)
of chroman-4-one and 3.93 g (0.03 mol) of 4-cyanobenzaldehyde
(0.03 mol), compound 6 (6.9 g, 88%) was obtained, mp 154-
1
155 °C. H NMR: δ 3.80 (s, 2H), 7.30-8.25 (m, 8H, arom +
H-2).
Acknowledgment. This work was supported by
grants from MIUR-COFIN2004 (Rome, Italy). The
authors thank Prof. Paolo Biscarini, (Facolta` di Chimica
Industriale, Universita` di Bologna, Italy) for the avail-
ability of the Jasco J-810 spectropolarimeter. Financial
support from Universita` della Basilicata to E.G. and
C.R. is gratefully acknowledged.
3-[Bromo-(4-nitro-phenyl)-methyl]-chromen-4-one (7).
A mixture of 3-(4-Nitro-benzyl)-chromen-4-one (6), (2.0 g, 0.007
mol) and N-bromosuccinimide (1.27 g, 0.007 mol) in the
presence of a catalytic amount of benzoyl peroxide in 70 mL
of CCl4 was refluxed for 6 h and then hot filtered. The solvent
was evaporated to dryness, and the residue was crystallized
from ligroin, to yield 1.8 g of 7, (64%), mp 83-85 °C. 1H NMR:
δ 6.45 (s, 2H), 7.40-8.35 (m, 8H, arom + H-2).
3-[Bromo-(4-bromo-phenyl)-methyl]-chromen-4-one (8).
Using the previous procedure and starting from 2.25 g (0.007
mol) of 7, compound 8 (1.92 g, 70%) was obtained, mp 136-
139 °C (ligroin, lit.57 mp 139-141 °C). 1H NMR: δ 6.40 (s, 2H),
7.45-8.20 (m, 8H, arom + H-2).
4-[Bromo-(4-oxo-4H-chromen-3-yl)-methyl)-benzoni-
trile (9). Using the previous procedure and starting from 1.83
g (0.007 mol) of 8, compound 9 (1.73 g, 73%) was obtained,
mp 181-184 °C (toluene). 1H NMR: δ 6.45 (s, 2H), 7.40-8.25
(m, 8H, arom + H-2).
Supporting Information Available: Elemental analysis
table; structures and (experimental and calculated) aromatase
inhibitory activity of the training set molecules; plot of
calculated vs experimental pIC50 values from the CoMFA
model; 3D structure and calculated OR of the low energy
conformers of S-3; experimental CD spectra of R-3 and R-1 in
chloroform. This material is available free of charge via the
Internet at http://pubs.acs.org.
References
General Method for the Preparation of the Imidazol-
1-yl Derivatives (1-3). A mixture of the selected bromo-
methyl derivative (0.005 mol) and imidazole (0.015 mol) in 50
mL of acetonitrile was refluxed for 7 h under nitrogen. The
solvent was removed under reduced pressure, and the residue
was purified by flash chromatography (toluene/acetone 4:1).
3-[Imidazol-1-yl-(4-nitro-phenyl)-methyl]-chromen-4-
one (1). Yield: 47%. Mp: 135-139 °C (ligroin). 1H NMR
(CDCl3): δ 6.85 (s, 1H), 6.95-8.35 (m, 12H, arom + H-2 +
CH imidazole). MS (ES): m/z 348 (M + H+). Anal. (C19H13N3O4)
C, H, N.
3-[(4-Bromo-phenyl)-imidazol-1-yl-methyl]-chromen-
4-one (2). Yield: 52%. Mp: 133-135 °C (ligroin). 1H NMR
(CDCl3): δ 6.80 (s, 1H), 6.95-8.20 (m, 12H, arom + H-2 +
CH imidazole). MS (ES): m/z 381 (M + H+). Anal. (C19H13-
BrN2O2) C, H, N.
(1) Owen, C. P.; Nicholls, P. J.; Smith, H. J.; Whomsley, R. Inhibition
of aromatase (P450Arom) by some 1-(benzofuran-2-ylmethyl)-
imidazoles. J. Pharm. Pharmacol. 1999, 51, 427-433.
(2) Recanatini, M.; Cavalli, A.; Valenti, P. Nonsteroidal aromatase
inhibitors: recent advances. Med. Res. Rev. 2002, 22, 282-304.
(3) Kim, Y. W.; Hackett, J. C.; Brueggemeier, R. W. Synthesis and
aromatase inhibitory activity of novel pyridine-containing iso-
flavones. J. Med. Chem. 2004, 47, 4032-4040.
(4) Leonetti, F.; Favia, A.; Rao, A.; Aliano, R.; Paluszcak, A.; et al.
Design, synthesis, and 3D QSAR of novel potent and selective
aromatase inhibitors. J. Med. Chem. 2004, 47, 6792-6803.
(5) Woo, L. W.; Sutcliffe, O. B.; Bubert, C.; Grasso, A.; Chander, S.
K.; et al. First dual aromatase-steroid sulfatase inhibitors. J.
Med. Chem. 2003, 46, 3193-3196.
(6) Suzuki, T.; Moriya, T.; Ishida, T.; Kimura, M.; Ohuchi, N.; et
al. In situ production of estrogens in human breast carcinoma.
Breast Cancer 2002, 9, 296-302.
(7) Recanatini, M.; Bisi, A.; Cavalli, A.; Belluti, F.; Gobbi, S.; et al.
A new class of nonsteroidal aromatase inhibitors: design and
synthesis of chromone and xanthone derivatives and inhibition
of the P450 enzymes aromatase and 17 alpha-hydroxylase/C17,-
20-lyase. J. Med. Chem. 2001, 44, 672-680.
(8) Recanatini, M. Comparative molecular field analysis of non-
steroidal aromatase inhibitors related to fadrozole. J. Comput.-
Aided Mol. Des. 1996, 10, 74-82.
4-[Imidazol-1-yl-(4-oxo-4H-chromen-3-yl)-methyl]-ben-
zonitrile (3). Yield: 49%. Mp: 156-158 °C (ligroin). 1H NMR
(CDCl3): δ 6.80 (s, 1H), 6.90-8.25 (m, 12H, arom + H-2 +
CH imidazole). MS (ES): m/z 328 (M + H+). Anal. (C20H13N3O2)
C, H, N.
Enantioselective HPLC. The chromatographic system
consisted of a Jasco PU-980 solvent delivery system, and a
Jasco MD-910 multiwavelength detector connected to a com-
puter station. A Rheodyne model 7125 injector with a 20 µL
loop was used. The enantiomeric fractions of compounds 1-3
have been obtained by preparative HPLC resolution employing
the chiral stationary phases Chiralcel OD (1 and 3), and
(9) Recanatini, M.; Cavalli, A. Comparative molecular field analysis
of non-steroidal aromatase inhibitors: an extended model for
two different structural classes. Bioorg. Med. Chem. 1998, 6,
377-388.
(10) Furet, P.; Batzl, C.; Bhatnagar, A.; Francotte, E.; Rihs, G.; et
al. Aromatase inhibitors: synthesis, biological activity, and
binding mode of azole-type compounds. J. Med. Chem. 1993, 36,
1393-1400.

Enantioselective Nonsteroidal Aromatase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 23 7289
(11) Allen, F. H. The Cambridge Structural Database: a quarter of
a million crystal structures and rising. Acta Crystallogr. 2002,
B58, 380-388.
(12) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. A well-
behaved electrostatic potential based method using charge
restraints for determining atom-centered charges: the RESP
model. J. Phys. Chem. 1993, 97, 10269-10280.
(13) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; et al. Gaussian 03; Gaussian, Inc.: Pittsburgh, PA.
(14) Diedrich, C.; Grimme, S. Systematic investigation of modern
quantum chemical methods to predict electronic circular dichro-
ism spectra. J. Phys. Chem. A 2003, 107, 2524.
(15) Thompson, E. A., Jr.; Siiteri, P. K. Utilization of oxygen and
reduced nicotinamide adenine dinucleotide phosphate by human
placental microsomes during aromatization of androstenedione.
J. Biol. Chem. 1974, 249, 5364-5372.
(16) Hutschenreuter, T. U.; Ehmer, P. B.; Hartmann, R. W. Synthesis
of hydroxy derivatives of highly potent non-steroidal CYP 17
inhibitors as potential metabolites and evaluation of their
activity by a non cellular assay using recombinant human
enzyme. J. Enzyme Inhib. Med. Chem. 2004, 19, 17-32.
(17) Ehmer, P. B.; Bureik, M.; Bernhardt, R.; Muller, U.; Hartmann,
R. W. Development of a test system for inhibitors of human
aldosterone synthase (CYP11B2): screening in fission yeast and
evaluation of selectivity in V79 cells. J. Steroid Biochem. Mol.
Biol. 2002, 81, 173-179.
(18) Oprea, T. I.; Waller, C. L. Theoretical and Practical Aspects of
Three-Dimensional Quantitative Structure-Activity Relation-
ships. Reviews in computational chemistry; Indiana University-
Purdue University at Indianapolis (IUPUI): Indianapolis, 1997;
pp 127-182.
(36) Braun, M.; Hohmann, A.; Rahematpura, J.; Buehne, C.; Grimme,
S. Synthesis and determination of the absolute configuration of
fugomycin and desoxyfugomycin: CD spectroscopy and fungi-
cidal activity of butenolides. Chem.sEur. J. 2004, 10, 4584.
(37) Devlin, F. J.; Stephens, P. J.; Besse, P. Are the absolute
configurations of 2-(1-hydroxyethyl)-chromen-4-one and its 6-bro-
mo derivative determined by X-ray christallography correct? A
vibrational circular dichroism study of their acetate derivatives.
Tetrahedron: Asymmetry 2005, 16, 1557-1566.
(38) Harada, N.; Ono, H.; Uda, H.; Parveen, M.; Khan, N.; et al.
Atropisomerism in natural products. Absolute stereochemistry
of biflavone, (-)-4′,4′′′,7,7′′-tetra-O-methylcupressuflavone, as
determined by the theoretical calculation of CD spectra. J. Am.
Chem. Soc. 1992, 114, 7687.
(39) Mason, S. F. Non-anomalous absolute configuration of 1,5-
disubstituted-9,10dihydro-9,10-bridged anthracenes. J. Chem.
Soc., Chem. Commun. 1973, 239.
(40) Gottarelli, G.; Proni, G.; Spada, G. P.; Fabbri, D.; Gladiali, S.;
et al. Conformational and configurational analysis of 4,4′-
biphenanthryl derivatives and related helicenes by circular
dichroism spectroscopy and cholesteric induction in nematic
mesophases. J. Org. Chem. 1996, 21, 2013.
(41) Tanaka, K.; Pescitelli, G.; Di Bari, L.; Xiao, T. L.; Nakanishi,
K.; et al. Absolute stereochemistry of dihydrofuroangelicins
bearing C-8 substituted double bonds: a combined chemical/
exciton chirality protocol. Org. Biomol. Chem. 2004, 2, 48.
(42) Giorgio, E.; Maddau, L.; Spanu, E.; Evidente, A.; Rosini, C.
Assignment of the Absolute Configuration of (+)-Diplopyrone,
the Main Phytotoxin Produced by Diplodia mutila, the Pathogen
of the Cork Oak Decline, by a Nonempirical Analysis of its
chiroptical properties. J. Org. Chem. 2005, 70, 7.
(19) Greco, G.; Novellino, E.; Connolly, Y. M. Approaches to Three-
Dimensional Quantitative Structure-Activity Relationships.
Reviews in computational chemistry; Indiana University-Purdue
University at Indianapolis (IUPUI): Indianapolis, 1997; pp 183-
240.
(20) Mason, S. F. Optical rotatory power. Q. Rev. 1962, 17, 20.
(21) Mason, S. F. Optical rotatory dispersion and circular dichroism
in organic chemistry; Heyden and Son: London, 1967; Chapter
4, Theory II; p 71.
(22) Gottarelli, G.; Mason, S. F.; Torre, G. The circular dichroism
and absolute configuration of (+)-trans-stilbene oxide. J. Chem.
Soc. B 1971, 1349.
(23) Harada, N.; Nakanishi, K. The exciton chirality method and its
application to configurational and conformational studies of
natural products. Acc. Chem. Res. 1972, 5, 257.
(24) Mason, S. F. Molecular optical activity and the chiral discrimi-
nations; Cambridge University Press: Cambridge, 1982.
(25) Harada, N.; Nakanishi, K. Circular dichroic spectroscopy: exciton
coupling in organic stereochemistry; University Science Books:
Mill Valley, CA, 1983.
(26) Nakanishi, K.; Berova, N. Circular dichroism: principles and
applications; VCH Publishers Inc: New York, 1994; pp 361-
398.
(27) Berova, N.; Nakanishi, K. Chapter 12. Circular dichroism:
principles and applications, 2nd ed.; Wiley-VCH: New York,
2000.
(28) Helgaker, T.; Jensen, H. J. A.; Joergensen, P.; Olsen, J.; Ruud,
K.; et al. DALTON, a molecular electronic structure program,
1.2 ed; University of Oslo: Oslo, Norway.
(29) Ahlrichs, R.; Bar, M.; Baron, H.-P.; Bauernschmitt, R.; Bocker,
S.; et al. TURBOMOLE; 5.6 ed.; Universitat Karlsruhe: Karlsru-
he, Germany.
(30) Polavarapu, P. L. Ab initio molecular optical rotations and
absolute configurations. Mol. Phys. 1997, 91, 551.
(31) Kondru, R. K.; Wipf, P.; Beratan, D. N. Theory-assisted deter-
mination of absolute stereochemistry for complex natural prod-
ucts via computation of molar rotation angles. J. Am. Chem. Soc.
1998, 120, 2204.
(32) Stephens, P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch, M. J.
Calculation of optical rotation using Density Functional Theory.
J. Phys. Chem. A 2001, 105, 5356.
(33) Polavarapu, P. L. Optical rotation: recent advances in determin-
ing the absolute configuration. Chirality 2002, 14, 768.
(34) Pedersen, T. B.; Koch, H. Theoretical electronic absorption and
natural circular dichroism spectra of (-)-trans-cycloctene. J.
Chem. Phys. 2000, 112, 2139.
(43) Giorgio, E.; Minichino, C.; Viglione, R. G.; Zanasi, R.; Rosini, C.
Assignment of the molecular absolute configuration through the
ab initio Hartree-Fock calculation of the optical rotation: can
the circular dichroism data help in reducing basis set require-
ments? J. Org. Chem. 2003, 68, 5186.
(44) Giorgio, E.; Viglione, R. G.; Rosini, C. Assignment of the absolute
configuration of large molecules by ab initio calculation of the
rotatory power within a small basis set scheme: the case of some
biologically active natural products. Tetrahedron: Asymmetry
2004, 15, 1979.
(45) Halgren, T. A. Merck molecular force field. I. Basis, form, scope,
parameterization, and performance of MMFF94. J. Comput.
Chem. 1996, 17, 490-519.
(46) SPARTAN ′02; Wavefunction Inc.: Irvine, CA.
(47) Hartmann, R. W.; Batzl, C. Aromatase inhibitors. Synthesis and
evaluation of mammary tumor inhibiting activity of 3-alkylated
3-(4-aminophenyl)piperidine-2,6-diones. J. Med. Chem. 1986, 29,
1362-1369.
(48) Cavalli, A.; Greco, G.; Novellino, E.; Recanatini, M. Linking
CoMFA and protein homology models of enzyme-inhibitor
interactions: an application to non-steroidal aromatase inhibi-
tors. Bioorg. Med. Chem. 2000, 8, 2771-2780.
(49) Cavalli, A.; Recanatini, M. Looking for selectivity among cyto-
chrome P450s inhibitors. J. Med. Chem. 2002, 45, 251-254.
(50) Chang, G.; Guida, W. C.; Still, W. C. An internal coordinate
Monte Carlo method for searching conformational space. J. Am.
Chem. Soc. 1989, 111, 4379-4386.
(51) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R. M.
J.; Lipton, M. A.; et al. MacroModelsan integrated software
system for modeling organic and bioorganic molecules using
molecular mechanics. J. Comput. Chem. 1990, 1, 440-467.
(52) Shenkin, P. S.; McDonald, D. Q. Cluster analysis of molecular
conformations. J. Comput. Chem. 1994, 15, 899-916.
(53) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. P. P.
AM1: a new general purpose quantum mechanical molecular
model. J. Am. Chem. Soc. 1985, 107, 3902-3909.
(54) Stewart, J. P. P. Optimization of parameters for semiempirical
methods I. Method. J. Comput. Chem. 1989, 10, 209-220.
(55) Becke, A. D. A new mixing of Hartree-Fock and local density-
functional theories. J. Chem. Phys. 1993, 98, 1372-1377.
(56) Cramer, R. D. r.; Patterson, D. E.; Bunce, J. D. Comparative
molecular field analysis (CoMFA). 1. Effect of shape on binding
of steroids to carrier proteins. J. Am. Chem. Soc. 1988, 110,
5959-5967.
(57) Patonay, T.; Dinya, Z.; Le`vai, A.; Molnar, D. Reactivity of
a-arilidene benzoheteracyclanone dibromides toward azide ion:
an effective approach to 3-(substituted-benzyl chromones and
1-thiochromones. Tetrahedron 2001, 57, 2895-2907.
(35) Furche, F.; Ahlrichs, R.; Wachsmann, C.; Weber, E.; Sobanski,
A.; et al. Circular dichroism of helicenes investigated by time-
dependent Density Functional Theory. J. Am. Chem. Soc. 2000,
122, 1717.
JM058042R