Structural basis for the potential antitumour activity of DNA-interacting benzo[kl]xanthene lignans

Article abstract of DOI:10.1039/c0ob00480d

The biological properties and possible pharmacological applications of benzo[kl]xanthene lignans, rare among natural products and synthetic compounds, are almost unexplored. In the present contribution, the possible interaction of six synthetic benzo[kl]x

Full text of DOI:10.1039/c0ob00480d

View Online / Journal Homepage / Table of Contents for this issue
Organic &
Biomolecular
Chemistry
Dynamic Article Links
Cite this: Org. Biomol. Chem., 2011, 9, 701
www.rsc.org/obc
PAPER
Structural basis for the potential antitumour activity of DNA-interacting
benzo[kl]xanthene lignans†
Simone Di Micco,a Fre´de´ric Mazue´,b Carmelo Daquino,c Carmela Spatafora,c Dominique Delmas,b
Norbert Latruffe,b Corrado Tringali,c Raffaele Riccioa and Giuseppe Bifulco*a
Received 23rd July 2010, Accepted 29th September 2010
DOI: 10.1039/c0ob00480d
The biological properties and possible pharmacological applications of benzo[kl]xanthene lignans, rare
among natural products and synthetic compounds, are almost unexplored. In the present contribution,
the possible interaction of six synthetic benzo[kl]xanthene lignans and the natural metabolite
rufescidride with DNA has been investigated through a combined STD-NMR and molecular docking
approach, paralleled by in vitro biological assays on their antiproliferative activity towards two different
cancer cell lines: SW 480 and HepG2. Our data suggest that the benzo[kl]xanthene lignans are suitable
lead compounds for the design of DNA selective ligands with potential antitumour properties.
but these are actually metabolic products deriving from a dietary
1. Introduction
consumption of plant lignans. The biological role of lignans in
In current literature, the term ‘lignans’ is commonly employed to
plant tissues is reasonably related to plant defense, as suggested by
indicate a large family ofwidespread naturalproducts, biosyntheti-
their antimicrobial, antifungal, antifeedant, and insecticidal prop-
cally originated from the shikimate pathway by oxidative coupling
erties. However, in many cases the bioactivities displayed by these
of two phenylpropanoid (C6C3) units, displaying an impressive
plant secondary metabolites are of pharmacological significance,
structural diversity and a comparable variety of biological activ-
and include antitumor, antioxidant, anti-inflammatory, antileish-
manial, antiangiogenic, cardiovascular and antiviral activity.4–11
A frequently cited example of bioactive lignan, whose chemical
modification has led to new useful drugs, is podophyllotoxin (1).
This natural product is known to be a constituent of Podophyllum
peltatum since 1880, and recognized as an antitumor compound
since around 1950;12 from the variety of synthetic podophyllotoxin
analogues, the anticancer drugs etoposide (2), etopophos (3) and
teniposide (4) have emerged.13
The biological activities and the structural variety of lignans
and related compounds make them an attractive target for
chemical synthesis or modification. Although a variety of synthetic
methodologies have been employed to this purpose, dimeriza-
tion reactions carried out through a radical phenolic oxidative
coupling of natural precursors, may afford ‘unnatural’ products
by a mechanism mimicking the ‘natural’ biosynthetic process.14
These biomimetic syntheses can allow us, in principle, to obtain
compounds unprecedented in literature although maintaining a
basic ‘natural’ skeleton and possibly offering a bioactivity profile
similar to, or better than, that of a natural analogue. A limit of this
synthetic approach is that the lack of stereocontrol both in metal or
enzyme mediated phenolic radical coupling15 frequently results in
complex mixtures and racemic compounds. Nevertheless, a num-
ber of interesting products have been obtained in such a way, and
in some cases the bioactive racemate has been resolved to obtain
the most active enantiomer. Thus, some of us have recently carried
out the biomimetic synthesis of natural and ‘unnatural’ lignans
by oxidative coupling of caffeic esters, namely CAPE (caffeic acid
ities. Strictly speaking, lignans are dimers generated by b-b¢ (8-
8¢) oxidative coupling of two cinnamic acid residues; this term
was originally employed by Haworth1 and refers to the wood
from which many of the first specimens were obtained. How-
ever, many other related compounds (neolignans, oxyneolignans,
hybrid lignans and others) with a carbon linkage between two
C6C3 units different from 8-8¢, or even with different constitutive
monomers, are normally included in this family, as described in
recent reviews.2,3
Lignans have been found in various plant parts, such as roots,
stems, bark, leaves, seeds, and fruits. Some kind of lignans
(enterolactones and enterodiols) have been found in mammals,
aDipartimento di Scienze Farmaceutiche, Universita` degli Studi di Salerno,
Via Ponte Don Melillo, 84084, Fisciano (SA), Italy. E-mail: bifulco@
unisa.it; Fax: +39 089 969602; Tel: +39 089 969741
bINSERM U 866, University of Burgundy, Laboratory of Biochemistry of
Metabolism and Nutrition, 6, Bd Gabriel, F-21000, Dijon, France
cDipartimento di Scienze Chimiche, Universita` degli Studi di Catania, Viale
A. Doria 6, I-95125, Catania, Italy
† Electronic supplementary information (ESI) available: DF-STD NMR
spectra for compounds 9, 10 and 16, acquired at 300 K (pH 7.1); calculated
inhibition constant of the complex formed by 7–12 and 16 with models
A and B; 3D interactions of 7–9, 11, 12 and 16 with models A and B;
graphical representation of SW480 and HepG2 cells percentages related
to the untreated control cells after a 48 h treatment with 8, 9, 11 and 12;
phase contrast microscopy observation of SW480 colon cancer cells treated
for 48 h with 10, 11 and 12; three-dimensional coordinates of models A
1
and B; H, 13C and ESI-MS spectra of new compounds 8, 9 and 11. See
DOI: 10.1039/c0ob00480d
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 701–710 | 701
©

View Online
Scheme 1 Molecular structures of podophyllotoxin (1), etoposide (2), etopophos (3), teniposide (4), caffeic acid phenethyl ester (CAPE, 5), caffeic acid
methyl ester (6) and natural and ‘unnatural’ lignans 7–18.
phenethyl ester, 5, Scheme 1) and methyl caffeate (6, Scheme 1).16
CAPE, a well-known component of propolis reported as an anti-
inflammatory, antioxidant and antitumor agent,17a–d was initially
selected because it had never been employed in phenolic oxidative
coupling reactions, and unreported dimerization products were
expected. The reaction, when carried out on 5 or 6 with Mn(AcO)3
as oxidative reagent, afforded with good yields (72% for CAPE
dimerization) the unusual, achiral benzo[kl]xanthene lignans 7
and 10 as the main products, accompanied by minor amounts
of the aryldihydronaphtalene racemic lignans ( )-13 and ( )-14,
respectively.
Benzo[kl]xanthene lignans are very rare both among natural
products and synthetic analogues. Recently reported represen-
tatives are yunnaneic acid H (15), rufescidride (16) and mon-
golicumin A (17) isolated respectively from Salvia yunnanensis,18
Cordia rufescens,19 and Taraxacum mongolicum.20 Interestingly, the
aryldihydronaphtalene lignan rabdosiin (18), was also isolated
from S. yunnanensis,18 and both are dimers of rosmarinic acid
and strictly related respectively to compounds 7 and ( )-13, thus
indicating the biomimetic nature of our synthetic approach. We
also carried out a mechanistic study of this oxidative coupling
reaction, and we applied this methodology to synthesize the
natural benzoxanthene lignans 16 and 17.
The lignans 7 and 10 are strongly fluorescent and are also
attractive in view of their partly planar structure with extensive
conjugation. It is also worth noting that, due to their rarity in
nature and the low yield of previous synthetic reactions to obtain
analogues,21a–b benzoxanthene lignans are almost unexplored with
regard to their biological properties and possible pharmacological
applications. Even for the previously known rufescidride (16,
Scheme 1), the only available reference about its biological
properties is, to the best of our knowledge, a patent referring to
its antimicrobial activity.22 Thus, as the first step in a study of the
putative biomedical properties of these lignans, we have carried
out an evaluation of their possible interaction with DNA through
an NMR based approach and molecular docking, paralleled by
in vitro biological assays on their antiproliferative activity towards
two different cancer cell lines, namely SW 480 (human colon
carcinoma) and HepG2 (human hepatoblastoma). The binding
mode of 7–12 and 16 to DNA has been analysed by means
of the differential frequency-saturation transfer difference (DF-
STD)23 protocol, based on STD-NMR experiments.24 Along with
the NMR binding assays, molecular docking calculations have
been performed to get the three-dimensional complex of putative
ligands and DNA, to get insights on the structural elements
responsible for the affinity to the biological target.
702 | Org. Biomol. Chem., 2011, 9, 701–710
This journal is
The Royal Society of Chemistry 2011
©

View Online
Table 1 Binding mode index values for 7, 9, 10 and 16 calculated by the
2. Results and discussion
eqn (1)23
The aim of this work was firstly to evaluate both the possible
interaction with DNA and the antiproliferative activity of the
benzoxanthene lignans 7–12 and 16. However, we planned to try
to establish at least a minimum of structure–activity relationships
for these lignans, and primarily to acquire data about the possible
role of the phenolic groups. To this purpose, we prepared by simple
chemical conversions carried out on 7 and 10, respectively the
permethyl derivatives 8 and 11 and the peracetyl derivatives 9 and
12. This small panel of seven compounds was submitted to the
DNA interaction and antiproliferative activity studies, as detailed
below.
Compound
BMI
BMI¢
7
9
1.2
1.0
0.7
0.9a
0.6b
0.2c
0.8
—
10
16
1.4
0.9
a BMI¢ calculated for the sole acetoxy in the position 6. b BMI calculated
for the acetoxy in the positions 9 and 10. c BMI calculated for the methoxy
groups in the positions 1 and 2.
The DF-STD analysis of 7 (Fig. 1) revealed that the polycyclic
aromatic portion intercalates between base pairs of the nucleic
acid (BMI, Table 1), whereas the methoxy groups are positioned
in the minor groove of the biological target (BMI¢, Table 1).
2.1. DF-STD analysis
The experimental STD-NMR outcomes revealed for 7, 9, 10 and
16 a binding event with the DNA. Indeed, the appearance of ligand
signals in the difference spectra suggested an interaction with the
biopolymer. On the other hand, compounds 8 and 11 did not show
any STD effects in the recorded spectra, suggesting a very weak
interaction with the DNA or a completely absent recognition by
the nucleic acid. NMR data were not collected for compound
12, due to problems of solubility at experimental conditions, and
possibly leading to erroneous interpretations of the outcomes.
In order to detect the possible binding mode and to understand
the binding contacts of 7–11 and 16 on the DNA surface, the
DF-STD23 protocol was applied, using a poly(dG-dC)·poly(dG-
dC) copolymer as biological target. This method consists in
the acquisition of a set of STD-NMR24 experiments at two
different saturation frequencies, providing useful information on
the binding mode of DNA interacting molecules, discriminating
among base-pair intercalators, minor groove binders, and external
backbone binders. The comparison of the STD effects at two
different saturation frequencies relative to diagnostic protons was
performed through the calculation of the binding mode index
(BMI),23 using the following equation:23
Fig. 1 DF-STD spectra of the 7-DNA complex. A, B) STD spectra
recorded upon saturation in the aromatic (9.0 ppm) and deoxyri-
bose/backbone (2.0 ppm) spectral regions, respectively. C) Reference STD
spectrum with an off-resonance irradiation (-16 ppm).
⎛
⎜
⎜
⎜
⎞
SN aromatic /SN rif
SN aliphatic /SN rif
⎟
⎟
⎟
⎟
⎟
⎠
∑
In particular, the BMI value (Table 1) showed a more effi-
cient saturation diffusion irradiating on the purine/pyrimidine
resonances, compared to the observed STD effects saturating the
deoxyribose/backbone protons (BMI = 1.2), suggesting for the
planar structural portion of 7 an intercalation between the bases
of DNA (Fig. 1). The BMI value of 1.2 is in fact consistent with
data reported for well known DNA intercalators such as thiazole
orange (BMI = 1.4)23 and with the polycyclic aromatic moiety of
doxorubicin (BMI = 1.33).23 Moreover, this value agrees with the
BMI of 1.3 obtained for ethidium bromide stacking between base
pairs of the parallel quadruplex structure [d(TGGGGT)]4.26
The BMI¢ of 0.7 revealed a minor groove binding mode of the
aliphatic portion of 7; such a value is in accordance with similar
values obtained for well known minor groove binders, such as
distamycin A (BMI = 1.1) and netropsin (BMI = 0.9).23 The BMI¢
is also similar to the experimental data describing the interaction
of distamycin A with the minor groove of a DNA quadruplex
(BMI = 0.8).26
(1)
⎜
⎝
i
BMI =
ni
The BMI is a numerical parameter that, accounting the relative
intensities of STD effects at different saturation frequencies, gives
insights on binding contacts of the ligands on the DNA surface.23
In our original contribution, based on the analysis of ligands with
a well known binding mode to DNA, we defined three BMI ranges:
0 < BMI < 0.50 for external (nonspecific) electrostatic backbone
binding; 0.90 < BMI < 1.10 for minor groove binding; and 1.20
(0.90) < BMI < 1.50 for base-pair intercalation.23
For 7, 9, 10 two distinct binding mode indexes (Table 1) were
calculated: the BMI relative to the polycyclic aromatic portion,
and the BMI¢ belonging the chemical appendages in the positions
1, 2, 6, 9 and 10 of the small molecules. Structural consideration
on the small molecules prompted us to calculate two BMI, as
described for the doxorubicin23 in our original contribution on
DF-STD and recently reported by Gomez-Monterrey et al. for
the analysis of DNA-interacting spiro derivatives:25 the presence
of a planar conjugated moiety typical of DNA intercalators with
flexible substituents.
The structural difference between 7 and 10 relates to the
presence of two phenylethyl groups in positions 1 and 2 (Scheme 1).
However, similar interactions with the biological target were
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 701–710 | 703
©

View Online
observed. Indeed, the obtained binding mode indexes were compa-
rable to the values found for 7. The BMI (1.4, similar to the value of
7) confirmed the base pairs intercalation, because the intensities of
the STD effects are higher irradiating aromatic proton resonances
of the nucleic acid (Figure S1†). As observed for 7, the BMI¢ (0.8,
Table 1) is the result of a similar saturation diffusion efficiency
by differential irradiation of the purine/pyrimidine bases and the
deoxyribose/backbone spectral regions (Figure S1).
the DNA surface (intercalation, minor groove accommodation
and deoxyribose/backbone binding) by the potential ligands.
Moreover, the study of intercalation between base pairs was
performed taking into account two binding cavities (Fig. 2), in
order to evaluate if this type of interaction to the DNA, and the
consequent hydrogen bond formations, were base-dependent.
Compound 9 is the three-acetylated derivative of 7 at the
positions 6, 9 and 10. While the expected binding mode of 9
is similar to that of compound 7, as confirmed by a DF-STD
approach, the analysis of the experimental data was complicated
bythepresenceofthebulkyacetylgroups. Indeed, thestericclashes
exerted by acetyl functionalities reduce the maximum overlap
with the flanking DNA base pairs giving raise to a BMI value
of 1 (Table 1). Concerning the experimental outcomes for the
aliphatic moieties of 9, different BMI¢ values were obtained for
the contributions of methoxyl and acetyl groups (see Table 1).
The experimental data investigation revealed that the acetyl group
in position 6 (2.36 ppm, Figure S2†) showed comparable STD
effects at two saturation frequencies (BMI¢ = 0.9), suggesting
an accommodation in the minor groove of the nucleic acid, as
found for the ester groups of 7 and 10. The BMI¢ is consistent
with the values for well known minor groove binders distamycin
A (BMI = 1.1) and netropsin (BMI = 0.9) interacting with
the poly(dG-dC)·poly(dG-dC) copolymer23 and with a BMI of
0.8 for the distamycin A interacting with the minor groove of
the [d(TGGGGT)]4 quadruplex.26 Consequently, the expected
location of the other acetyls of compound 9 on the DNA surface is
along the major groove, where these groups experienced a similar
saturation transfer from aromatic and deoxyribose/backbone
proton irradiation (Table 1). The two methoxy groups are accom-
modated in the major groove of the deoxyribonucleic acid pointing
towards the deoxyribose/backbone, and causing a BMI¢ value
of 0.2. These findings are in agreement with molecular docking
calculations (see the next section).
Fig. 2 Schematic representation of DNA models (A and B) used in
docking calculations. Sequence and numbering of two 10mers (A and B)
DNA duplexes are reported. The arrows indicate the intercalation points.
The analysis of docking calculations, based on visual inspection
and scoring function values, revealed that compounds 7–12 and
16 intercalate the DNA with their chromophores and collocate
the flexible substituents on different points of the DNA surface,
in accordance with the findings obtained by the experimental
DF-STD data. As reported in Fig. 3, compound 7 stacks its
planar portion between two adjacent base pairs establishing p–
p interactions (Fig. 3).
We have also investigated by a DF-STD protocol the possible
DNA interactions for compound 16 (Figure S3†). As predicted
by the simple inspection of structural planar feature of 16,
its binding mode consists of an intercalation between the base
pair aromatic rings of DNA. The experimental data reveal p-
stacking interactions with purinic and pyrimidinic bases of the
biological target, accompanied by interactions with the external
deoxyribose/backbone by the hydroxy groups of 16. In fact, the
obtained BMI of 0.9 is in perfect agreement with the binding mode
index calculated for investigated particular intercalator, ethidium
bromide, in our original paper.23 In detail, the ethidium bromide is
a base-pairs intercalator, able to interact with external negatively
charged phosphate by its amino groups.
Fig. 3 a) and c) 7 and 8 superimposition in the binding site of models A
and B, respectively. b) and d) 7 and 9 superimposition in the binding site
of models A and B, respectively. The ligands and DNA are represented by
tube, and their atoms coloured by atom type: C, gray; polar H, sky blue; N,
blue; O, red. For 7–9, bonds are depicted respectively in yellow, white and
green. The figure highlights the overlap with base pairs and interactions
with minor groove.
In our model, the oxygen of ester functionality (position 1,
Scheme 1) forms a hydrogen bond with the NH2 of G14 (Figure
S4a†). The hydroxy group in 10 establishes a hydrogen bond with
the anomeric oxygen of deoxyribose, whereas the OH in 6 and
9 gives the same type of interaction with N7 of G3 and G14,
respectively. The methyl groups point along the minor groove,
contributing to the complex stability by means of hydrophobic in-
teractions with the macromolecular counterparts. The comparison
of docking calculations performed on models A and B (Figure S4)
2.2. Molecular docking studies
Molecular docking calculations, using AutoDock 3.0.5,27 were
performed to predict a 3D model of the complex between
compounds 7, 9, 10, 12, 16, and the DNA, to get insights
on the structural elements responsible of the binding to the
biological target, and for the eventual design of more selective
and potent analogues. In the calculations different binding sites
were considered, to discriminate the preferred binding mode on
704 | Org. Biomol. Chem., 2011, 9, 701–710
This journal is
The Royal Society of Chemistry 2011
©

View Online
showed similar results. The predicted bioactive conformations,
indeed, are superimposable, giving the same number of hydrogen
bonds with similar distances. In particular, for model B (Figure S4c
and d) the docked pose presents a hydrogen bond with carbonylic
oxygen of G5 and OH 9, instead of OH 6 as found with model A
(Figure S4a and b).
The same considerations apply for compound 10 (Fig. 4).
Fig. 5 3D interactions of 10-model A (a and b) and 10-model B (c and d)
complexes. In a) and c) the DNA is represented by the molecular surface,
and sticks and balls (coloured by atom type: O, red; C, grey; polar H, sky
blue; N, blue), whereas in b) and d) only by sticks and balls. The ligand
is depicted in sticks (orange) and balls (coloured as for DNA). In a) and
c), the figure highlights the intercalation between two adjacent base pairs
and the hydrogen bonds (yellow line) between ester functionality and NH2
of guanine. In b) and d), besides the intercalation, the hydrogen bonds
(yellow line) formed by OH group are shown.
Fig. 4 a) and c) 10 and 11 superimposition in the binding site of models
A and B, respectively. b) and d) 10 and 12 superimposition in the binding
site of models A and B, respectively. The ligands and DNA are represented
by tube, and their atoms coloured by atom type: C, gray; polar H, sky
blue; N, blue; O, red. For 10–12, bonds are depicted respectively in orange,
light green and blue. The figure highlights the overlap with base pairs and
interactions with minor groove.
Within both models A and B, the common polycyclic aromatic
moiety intercalates between base pairs forming p–p interactions
with the macromolecular counterparts and the two phenylethyl
groups are accommodated along the minor groove establishing
van der Waals contacts (Fig. 5). Moreover, for model A, the oxygen
of the ester functionality (in position 1, Scheme 1) is involved in
a hydrogen bond with the NH2 (in 2) of G5. As found for 7, the
OH at C-10 (Scheme 1) interacts with the anomeric oxygen of
deoxyribose, whereas the hydroxy group at C-9 has a contact with
the N7 of G14 (Fig. 5b). For model B, there is a hydrogen bond
between OH at C-6 and carbonylic oxygen (at C-6) of G5 (Fig. 5d).
It is noteworthy that, besides the p-stacking, the two common
planar portions of compounds 7 and 10 show the same orientation
between the two adjacent base-pairs of models A and B. These
theoretical results are in agreement with the similar experimental
BMI values obtained for 7 and 10 (Table 1).
In contrast to the two methyl substituents of 7, the phenylethyl
groups contribute more efficiently to the complex stability, giving
wider van der Waals contacts along the minor groove of the DNA
(Table S1†). The difference in the predicted binding affinity for
7 and 10 is due to the presence of these two bulky groups in 10
(Table S1) and these theoretical results are in good qualitative
agreement with the biological data (see the next section, Table 2).
The analysis of docking results obtained for 9, the three-
acetylated derivative of 7, revealed an intercalative binding mode
with a different arrangement of the ester groups along the grooves
of the DNA (Fig. 3). The acetyl functionality in position 6 is
accommodated in the minor groove giving van der Waals contacts
by the methyl and creating a hydrogen bond with the NH2 of G14
with its carbonyl group. In our theoretical ligand-DNA complex,
the remaining ester functionalities lay along the major groove of
the nucleic acid (Figure S5†). In detail, the methoxy group in 3
follows the bend of deoxyribose backbone, giving close contact
with the external backbone of the DNA, as also outlined by the
BMI¢ value (Figure S5). The methoxy group in 2 is hydrogen
bonded to the NH2 of C4 and its methyl points toward the solvent.
The acetyl group in 10 establishes a hydrogen bond with the NH2
of C13 and by the oxygen of carbonyl and the methyl group is in
contact with the external phosphate backbone and the base pair
Table 2 IC50 values (mM) for 7–12 and 16, obtained on the colon (SW480)
and hepatic (HepG2) cancer cells. IC50 values calculated after 48 h of
continuous exposure relative to untreated controls. Values are the mean
SD of three experiments
Compound
IC50, (mM, SW480)
IC50, (mM, HepG2)
7
32.89 6.60
> 100
25.5 9.35
2.57 0.58
> 100
3.21 0.33
ꢀ 100
34.83 7.15
> 100
8
9
28.61 5.50
4.76 0.56
ꢀ 100
10
11
12
16
26.66 3.85
> 100
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 701–710 | 705
©

View Online
Fig. 6 Phase contrast microscopy observation of SW480 colon cancer cells treated or not for 48 h with 30 mM of 10, 7 and 16.
aromatic rings of the DNA (Figure S5). The acetyl in 11 faces the
base pairs along the groove floor of the biological target. Similar
findings were obtained for model B (Figure S5), where compound
9 shows a similar docked pose compared to the predicted bioactive
conformation interacting with model A. In our theoretical model,
the acetyl group in 6 is hydrogen bonded through its carbonyl
oxygen with the NH2 of G5, and the methoxy group in 2 establishes
a hydrogen contact with the NH2 of C4.
groups give hydrogen bonds with nucleotides G14 and G5, and the
chromophore intercalates between C4·G14 and G5·C15 base pairs
(Figure S9).
2.3. Antiproliferative activities
The antiproliferative activity of lignans 7–12 was evaluated
on SW480 (colon) and HepG2 (hepatic) cancer cells. Results,
expressed as IC50 values, are reported in Table 2. The most active
compound was the benzoxanthene lignan 10, potently active both
on colon (IC50 = 2.57 mM) and hepatic cancer cells (IC50 = 4.76 mM).
Also the three-acetylated derivative 12 strongly inhibits the cellular
proliferation after a 48 h treatment, selectively on colon cells
In the predicted model of the complex formed by the DNA
and compound 12, we observed an overlapping docked pose
for 10 and 12 for both models A and B (Fig. 4). In particular,
the chromophore is intercalated by two base pairs of the DNA,
and the phenylethyl groups establish van der Waals contacts and
hydrogen bonds with the minor groove of the biological target,
giving a great contribution to the affinity for the macromolecule.
In detail, the ester group in 2 appears to be hydrogen bonded
with the NH2 in 7 of G3 and, along with the substituent in 1,
it shares a hydrogen bond with the NH2 of G14 (Figure S6†).
The bulkier acetyl groups are accommodated in the wider major
groove. The acetyl groups in 6 and 9 are involved in hydrogen
bonds with NH2 of C4 and C13 respectively, with the oxygen of
the carbonyl. For model B, two oxygen atoms bounded to the
phenylethyl groups, are hydrogen bonded with the NH2 of G5,
whereas the CO in position 9 makes the same interaction with
the NH2 of C15 (Figure S6). Interestingly, the comparison of 9
and 12 with their progenitor 7 and 10, revealed that the chemical
conversion of hydroxy groups in acetoxy functionalities causes the
conversion of hydrogen bond donors in acceptors without loss of
interaction with macromolecular counterparts (Figure S5 and S6).
Indeed, the acetoxy groups, in our predicted models, are hydrogen
bonded with nucleotides, as found for the hydroxy groups of 7 and
10 (Figure S5 and S6). Moreover, the bulky acetoxy groups do
not favour the maximum overlap with the adjacent aromatic rings
of the DNA, as outlined by the BMI value of 9 (Table 1, Fig. 3
and 4). These findings are responsible of a lower predicted binding
affinity of 9 and 12, with respect to 7 and 10 (Table S1). Conversely,
the three-methylated derivatives 8 and 11 are not able to establish
hydrogen bonds with the nucleotides due to the presence of methyls
groups (Figure S7 and S8), causing a lower affinity for the DNA,
as also revealed by experimental NMR data and biological assays
(Table 2).
(IC50 = 3.21 mM), and to a lesser extent, on hepatic cells (IC50
=
26.66 mM). It is worth noting that compounds 10 and 12, bearing
the phenylethyl pendants at positions 1 and 2, are significantly
more active than compounds 7 and 9, which have methyl ester
groups in the same positions. The chemical conversion of the
hydroxy into methoxy groups in compounds 8 and 11 dramatically
decreases the cytostatic effect of the original molecules (7 and 10),
indicating a fundamental role of hydroxy or acetoxy groups in
the antiproliferative activity. Also rufescidride (16), lacking any
pendant group, exhibited poor activity. This is evident in Fig. 6
were the optical microscopy images of SW480 colon cancer cells
are reported after a 48 h treatment with 30 mM of 7, 10 and 16 in
comparison with the control culture. Also graphs in Fig. 7 clearly
indicate the different activity of these three structurally related
benzoxanthenes, as shown by the viability cell curves of both
SW480 and HepG2 cells treated with increasing concentrations
of the three compounds.
Fig. 7 Graphical representation of SW480 (left) and HepG2 (right) cell
percentages related to the untreated control cells after a 48 h treatment
with 7, 10 and 16. Counting is realized with a haemocytometer after dead
cells exclusion by Trypan Blue staining.
Finally, the docking outcomes for 16 were in line with the
DF-STD analysis. In fact, we find that 16 establishes extended
van der Waals contacts with base-pair aromatic rings, and the
hydroxy functionalities in 6 and 9 are hydrogen bonded with
anomeric oxygen of deoxyribose ring of nucleotides C13 and C4,
respectively (Figure S9†). Similarly, for model B these two hydroxyl
These finding corroborate the structural considerations re-
ported above, obtained from molecular docking studies and
NMR experiments, highlighting the crucial role of phenylethyl
groups on the affinity for the DNA as well as the importance of
hydroxyl groups in the DNA binding. Moreover, the activity of
706 | Org. Biomol. Chem., 2011, 9, 701–710
This journal is
The Royal Society of Chemistry 2011
©

View Online
three-acetylated derivatives, comparable to that of the parent
molecules, could be explained by a better cell membrane pene-
tration and the hydrolytic action of cellular esterases, releasing the
hydroxylated compounds 7 and 10. The importance of the pendant
groups in 7 and 10 is confirmed by the scarce antiproliferative
activity of rufescidride (16). The detected binding to DNA by
NMR experiments could be in accordance with the antimicrobial
activity of 16 recently reported in a patent.22
3.1.3. Dimethyl 6,9,10-trimethoxybenzo[kl]xanthene-1,2-
dicarboxylate (8)
Compound 7 (60.6 mg, 0.158 mmol) was treated with dimethyl
sulfate using the same procedure described above for 10 to afford
a residue which was purified by LC (silica gel, 70% CHCl3 in
n-hexane) so as to obtain 62.5 mg (93.3% yield) of 8.
Yellow amorphous powder: Rf (TLC) = 0.42 (100% CHCl3); 1H
NMR [500 MHz, CDCl3, 298 K]: d 8.19 (s, 1H, 3-H), 7.45 (d, J =
8.5 Hz, 1H, 4-H), 7.35 (d, J = 8.5 Hz, 1H, 5-H), 7.29 (s, 1H, 11-H),
6.81 (s, 1H, 8-H), 4.05 (s, 6H, OCH3), 3.93 (s, 6H, COOCH3),
3.91 (s, 3H, OCH3). 13C NMR [125 MHz, CDCl3, 298 K]: d 171.6,
166.4, 151.3, 147.8, 145.5, 143.9, 139.1, 129.4, 127.4, 125.1, 124.9,
123.5, 121.1, 120.4, 115.3, 110.4, 107.5, 101.1, 56.7, 56.2, 56.0,
52.5, 52.1. ESI MS: m/z = 447.0 [M+ +Na], 462.9 [M+ +K].
3. Experimental
3.1. General
All chemicals were of reagent grade and used as purchased from
Sigma-Aldrich. LiChroprep Si-60 and LiChroprep DIOL 25-40
(Merck) were used as stationary phases for column chromatog-
raphy. TLC was carried out using pre-coated silica gel F254 plates
(Merck). Cerium sulfate and phosphomolybdic acid were used
as spray reagents. ESI-MS spectra were carried out on mass
spectrometer Agilent MS G1956A. NMR spectra were run on
a Varian Unity Inova spectrometer operating at 499.86 (1H) and
125.70 MHz (13C) and equipped with gradient-enhanced, reverse-
detection probe. Chemical shifts (d) were indirectly referred
to TMS using solvent signals. STD-NMR experiments were
performed on a Bruker Avance DRX 600-MHz spectrometer,
equipped with cryoprobe, at 300 K (see below for details).
3.1.4. Dimethyl 6,9,10-tris(acetyloxy)benzo[kl]xanthene-
1,2-dicarboxylate (9)
Compound 7 (26 mg, 0.068 mmol) was dissolved in pyridine (1 mL)
and acetic anhydride (1.2 mL). The solution was stirred at room
temperature for 4 h and, after standard work-up, the peracetate
9 was obtained with 96% yield. Yellow amorphous powder: Rf
(TLC) = 0.35 (100% CHCl3); 1H NMR [500 MHz, (CD3)2CO, 298
K]: d 8.4 (s, 1H, 3-H), 7.73 (d, J = 8.5 Hz, 1H, 4-H), 7.52 (d, J =
8.5 Hz, 1H, 5-H), 7.58 (s, 1H, 8-H), 7.23 (s, 1H, 11-H), 3.96 (s, 3H,
COOCH3), 3.93 (s, 3H, COOCH3), 2.41 (s, 3H, COCH3), 2.33 (s,
6H, COCH3). 13C NMR [125 MHz, (CD3)2CO, 298 K]: d 169.0,
167.9, 167.3, 165.0, 149.7, 144.3, 141.2, 138.8, 134.8, 131.0, 131.1,
130.2, 127.4, 124.9, 123.4, 123.2, 122.9, 121.8, 120.2, 116.3, 112.8,
52.3, 52.1, 19.6, 19.59, 19.57. ESI MS: m/z = 530.9 [M+ +Na].
3.1.1. Synthesis of compounds 7, 10, 12 and 16
Dimethyl 6,9,10-trihydroxybenzo[kl]xanthene-1,2-dicarboxylate
(7), bis(2-phenylethyl)-6,9,10-trihydroxybenzo[kl]xanthene-1,2-
dicarboxylate (10), bis(2-phenylethyl) 6,9,10-tris(acetyloxy)-
benzo[kl]xanthene-1,2-dicarboxylate (12) and rufescidride (16)
were synthesized as previously described.16
3.2. DF-STD NMR
Before running the NMR experiments, the solubility of com-
pounds 7–12 and 16 was evaluated. The mixture of acetone/water
containing phosphate buffered saline (10 mM) at pH 7.1 was
used to verify the solubility of the molecules. For 7 and 8
the final concentration was 1 mM and for the DNA 50 mM
(expressed as molarity of phosphate groups) in 200 ml of solvent:
acetone/deuterated water 10/190 for 7 and a ratio of 30/170 for
8. The concentrations were halved for 9–11 and the DNA with
respect to 7 and 8, using the solvent acetone/deuterated water in
the ratios 60/140, 20/180 and 80/120 for 9–11, respectively. For
16 the final concentration was 1.8 mM and for the DNA 90 mM
(expressed as molarity of phosphate groups) in 200 ml of solvent
(acetone/deuterated water 20/180).
The DNA used as biological target was a poly(dG-dC)·poly(dG-
dC) copolymer (Sigma Aldrich), presenting an average of 750 base
pairs. The copolymer was dissolved in the above described buffer
aqueous solution and underwent at annealing for 5¢ at 80 ◦C. The
copolymer was dried and afterwards dissolved in the deuterated
buffer with the percentage of acetone as describe above.
3.1.2. Bis(2-phenylethyl)-6,9,10-trimethoxybenzo[kl]xanthene-1,2-
dicarboxylate (11)
Compound 10 (32 mg, 0.056 mmol) was placed into a boiling
flask and dispersed in 10 mL of acetone and 25 mg of anhydrous
potassium carbonate. To this suspension 20 ml of dimethyl sulfate
was added. The resulting mixture was heated for 18 h under a reflux
condenser. Acetone was then removed by a rotary evaporator and
the residue was purified by LC (silica gel, CHCl3 in n-hexane from
70% to 75%) so as to obtain 32.5 mg (94.6% yield) of 11: yellow
amorphous powder; Rf (TLC) = 0.33 (80% CHCl3–n-hexane); 1H
NMR [500 MHz, (CD3)2CO, 298 K]: d 8.12 (s, 1H, 3-H), 7.58 (d,
J = 8.5 Hz, 1H, 4-H), 7.54 (d, J = 8.5 Hz, 1H, 5-H), 7.38–7.15 (m,
10H, phenylethyl ring), 7.24 (overlapped with other signals, 1H,
11-H), 6.81 (s, 1H, 8-H), 4.62 (t, J = 9.0 Hz, 2H, 1¢¢-H), 4.47 (t,
J = 9.0 Hz, 2H, 1¢¢-H), 4.03 (s, 3H, OCH3), 3.93 (s, 3H, OCH3),
3.70 (s, 3H, OCH3), 3.09 and 3.08 (two t, partly overlapped, J =
7.0 Hz, 4H, 2¢¢-H and 2¢¢-H). 13C NMR [125 MHz, (CD3)2CO, 298
K]: d 169.9, 165.6, 152.2, 147.8, 145.8, 144.5, 138.8, 138.2, 138.2,
129.0, 129.2, 128.8, 128.4, 128.3, 127.2, 126.4, 126.3, 125.7, 124.6,
121.3, 123.1, 121.2 120.6, 116.1, 109.9, 108.4, 101.1, 66.3, 65.7,
56.2, 55.7, 55.4, 34.7, 34.3. ESI MS: m/z = 627.1 [M+ +Na].
Two STD NMR spectra for each ligand were recorded irradi-
ating on the aromatic and sugar proton resonances of DNA by a
Gaussian train pulses reaching a total saturation time of 4 s.
For 7 the saturation frequencies were 9.0 ppm (aromatic spectral
region) and 2.0 ppm (sugar spectral region). For 8 and 9 the
aromatic DNA protons were irradiated at respectively 9.4 and
9.8 ppm, and the aliphatic resonances at 5.5 ppm. For 10 the
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 701–710 | 707
©

View Online
saturation frequencies of 9.2 ppm and 1.3 ppm, and for 11 9.8
and 1.2 ppm were respectively applied to irradiate aromatic and
deoxyribose/backbone spectral regions of the nucleic acid. For
16, the selective irradiation was applied at 9.4 ppm and 1.2 ppm.
The STD effects of the individual protons were calculated for each
compound relative to a reference spectrum with off-resonance
saturation at d = -16 ppm.
Typically, 32 scans were recorded for the reference STD
spectrum, whereas 64 scans were recorded for each DF-STD
spectrum for 7, 8 and 16. The numbers of scans were doubled
in the set of experiments for 9–11.
The relative STD effects were calculated for each signal as the
difference between the intensity (expressed as an S/N ratio) of
one signal in the on-resonance STD spectrum and that of the
same signal in the off-resonance NMR spectrum divided by the
intensity of the same signal in the off-resonance spectrum. BMI
values were obtained by using eqn (1).23
account the ligand-macromolecule interactions by precalculating
atomic affinity potentials (grid maps) for each atom type in
the substrate molecule by the grid method.33 These maps are
calculated by AutoGrid, where the protein is embedded in a three-
dimensional grid and a probe atom is placed at each grid point.
The energy of interaction of this single atom with the protein is
assigned to the grid point. Thus, to study the binding mode to
DNA, several docking calculations were performed collocating
the grid box on different regions of the DNA surface: minor and
major groove, along the external backbone and the vertical axes
comprising base pairs. For all the docking calculations considering
the minor groove as the binding site, a grid box size of 62 ¥ 62
˚
¥ 62 with spacing of 0.375 A between the grid points was used
and centred on the following x, y and z coordinates: -0.256; 1.561;
15.262. Where the major groove was the binding cavity, a grid box
˚
size of 48 ¥ 56 ¥ 50 with a spacing of 0.375 A between the grid
points used and centred on the following x, y and z coordinates:
4.56; 1.561; 15.262. For all the docking calculations with models
˚
A and B, a grid box size of 54 ¥ 54 ¥ 64 with spacing of 0.375 A
3.3. Computational methods
between the grid points was used and centred on the following x,
y and z coordinates: -1.189; 2.505; 21.5. The above described grid
boxes also included the external deoxyribose/backbone.
In order to achieve a representative conformational space during
the docking calculations on the ligands under investigation, ten
calculations consisting of 256 runs were performed, obtaining
2560 ligand conformations (256 ¥ 10). The Lamarkian genetic
algorithm was used for dockings. An initial population of 650
randomly placed individuals, a maximum number of 10 ¥ 106 en-
ergy evaluations, and a maximum number of 11 ¥ 106 generations
were taken into account. A mutation rate of 0.02, a crossover rate
of 0.8 and a local search frequency of 0.26 were used.
The models of the biological target were built by using the
graphical interface Maestro version 6.0, Schro¨dinger, LLC, New
York, NY, 2003. A B-DNA decamer (dG-dC)·(dG-dC) was built,
in order to have a complete helix turn, using the available
base-pair models of Maestro (see supporting information for
the three-dimensional coordinates of the models†). To study
the intercalation between base pairs, the binding site was built
taking into account the distances from experimental structures
of the complex between the nucleic acids and intercalators and
the induced local unwinding. In particular, we observed that the
˚
experimental distances between the base-pairs ranged from 6.3 A
28
to 7.6 A. Considering this range of experimental distances, and
All the 3D models were depicted using the Phyton software:34
molecular surfaces were rendered using Maximal Speed Molecular
Surface (MSMS).35
˚
˚
applying increments of 0.1 A to the base-pair distance of our
˚
B-DNA duplexes we built different models, finding 6.9 A as the
optimal spacing for the ligand binding. Thus, two intercalations
points were built (models A and B), using the found distance of
3.4. Proliferation assays
˚
6.9 A between base-pairs, and considered in the calculations. In
details, the binding site of model A was formed by the G3·C13
and C4·G14 base pairs, whereas for the model B the intercalation
point was delimited by C4·G14 and G5·C15 base pairs. Both
models were optimized using Macromodel 8.5 software,29 by
All compounds under test (7–12 and 16) were dissolved in
dimethylsulfoxide (DMSO) solution. Stock solutions were pre-
pared at 100 mM or 10 mM. Solutions were stored at 4 ◦C avoiding
light exposure. The human colon carcinoma cell line SW480
was cultured in RPMI-medium with 10% fetal bovine serum
and 1% antibiotics (penicillin and streptomycin from Sigma-
Aldrich) and for HepG2 human hepatoblastoma cell line, DMEM
+10% fetal bovine serum (FSB) + 1% antibiotics. Cells at 90%
confluence were separated with trypsin/EDTA and reseeded at
1/10 (surface/surface) in 12-well plates (final volume of 1 mL
RPMI medium for each well). After 24 h, cells were treated
in separate experiments with lignans 7–12 and 16 at a final
concentration of 0.1% DMSO.
Polak-Ribiere Conjugate Gradient method using the Force Field
-1
Amber30 (threshold 0.005 kJ mol-1 A ). The GB/SA (Generalized
˚
Born/surface area)31 solvent treatment was used, mimicking the
presence of H2O, in the calculations for reducing the artifacts
derived from the absence of the solvent. During the optimization of
the geometries and energies of models A and B, a penalty of 100 kJ
-2
˚
A
was applied for the distance violations along the hydrogen
bonds between nucleotides aromatic rings of each complementary
strand.
All ligands structures were built and their geometries opti-
mized through MacroModel 8.5 software29 package and using
the MMFFs force field.32 The MonteCarlo Multiple Minimum
(MCMM) method (10,000 steps) of the MacroModel package29
was used in order to allow a full exploration of the conformational
space. The so obtained geometries were optimized using the
Polak–Ribier Conjugate Gradient algorithm (PRCG, maximum
derivative less than 0.001 kcal mol-1). Autodock 3.0.527 was used
for all docking calculations. The software rapidly takes into
Proliferation inhibition assays were performed in 24-well plates
in triplicate, and each experiment was conducted three times. In
all, 30,000 cells were seeded per well and after 24 h were treated
with media containing either each compound in dimethylsulfoxide
(0.1% final concentration) or with 0.1% dimethylsulfoxide as
control (0.1% final concentration). After 48 h, adherent cells were
collected by trypsinization and washed with 1X PBS.
Counting of living cells was made on Mallassez cells. Dead
cells were excluded by coloration with Trypan Blue. Subsequently,
708 | Org. Biomol. Chem., 2011, 9, 701–710
This journal is
The Royal Society of Chemistry 2011
©

View Online
48 h IC50 values were determined by performing 1 nM to
100 mM treatments and the IC50 values were obtained after
parametric regressions on the percentages of viable cells versus the
control.
improving the affinity for the biological target and contributing to
sequence selectivity of the DNA-interacting compounds.
Acknowledgements
This research was supported by a grant of the Universita` degli
Studi di Catania (Progetti di Ricerca di Ateneo, Catania, Italy), by
MIUR, Ministero dell’Universita` e della Ricerca (PRIN 2006 and
2007, Rome, Italy), and by the French Cancer League, Regional
Council of Burgundy.
3
Conclusions
A significant portion of all anticancer drugs are DNA-interacting
molecules, and many clinical anticancer compounds are natural
products or their synthetic or semisynthetic derivatives. In the
present study the benzoxanthene lignans 7–12 and 16 have been
analysed as potential DNA interacting molecules, through DF-
STD NMR spectroscopy, and their antiproliferative activity has
been tested towards two different cancer cell lines, SW 480
(human colon carcinoma) and HepG2 (human hepatoblastoma).
Molecular modeling studies have also been performed to deepen
the structure–activity relationships in order to design more potent
and selective analogs. In detail, by DF-STD experimental protocol
we have monitored the binding events that occur at specific
regions of the duplex DNA. The experimental outcomes reveal
that the planar chromophoric moiety of the studied compounds
is intercalated between two base pairs and the flexible chemical
appendages are collocated along the grooves of the nucleic acid
and make contacts with the external deoxyribose/backbone. The
docking results are in agreement with the experimental NMR
data. Indeed, the predicted bioactive conformations of tested
compounds present the polyaromatic system establishing p–p
interactions with two flanking DNA base pairs and the ester
groups give a further contribution to the DNA binding by their
van der Waals contacts and hydrogen bonds with macromolecular
counterparts. In particular, the bulky phenylethyl groups of 10
and 12, giving wider van der Waals contacts with the minor
groove, improve the binding affinity, as highlighted by the higher
calculated inhibition constants and the IC50 values of biological
data. Our investigation has also highlighted a crucial role of the
phenolic groups in the recognition process by the biological target.
Indeed, the chemical conversion of the hydroxy into methoxy
functionalities prevents the hydrogen bond formation between
ligands and macromolecule, lowering the affinity for the DNA,
as outlined by our combined strategy of investigation for 8 and
11, which result by far less active than their parent compounds
against cancer cells. Concerning the acetylated derivatives of 7
and 10, our data revel that their affinity for the biological target
is not negatively affected by the acetylation in the positions 6, 9
and 10. The STD NMR analysis detects the interaction with the
macromolecule and the theoretical results show that the carbonyl
of the inserted acetyls interact through hydrogen bonds with the
nucleotides, unlike the methoxy groups of 9 and 11. The biological
data show a comparable antiproliferative activity of 12 with its
progenitor 10, whereas compound 9 shows a slightly better IC50
than 7. This is probably due to a better cell membrane penetration
and to the intracellular hydrolysis of the acetylated appendages by
esterases, which release that active form of the compounds. The
DF-STD analysis of 16 show an intercalative binding mode to
the DNA and hydrogen bonds with the external backbone by its
hydroxy funcionalities, even though the antiproliferative activity
is lower than the other compounds.
Notes and references
1 R. D. Haworth, J. Chem. Soc., 1942, 448–456.
2 J.-Y. Pan, S.-L. Chen, M.-H. Yang, J. Wu, J. Sinkkonen and K. Zou,
Nat. Prod. Rep., 2009, 26, 1251–1292.
3 K.-H. Lee and Z. Xhiao, Phytochemistry Reviews, 2003, 2, 341–362.
4 J. L. Charlton, J. Nat. Prod., 1998, 61, 1447–1451.
5 D. C. Ayers, and J. D. Loike, in Lignans. Chemical, biological and
clinical properties, Cambridge University Press, Cambridge, 1990, pp
402.
6 P. M. Dewick, in Medicinal Natural Products. A Biosynthetic Approach,
John Wiley & Sons, Chichester, UK, 2nd edn, 2001, pp. 133.
7 S. Apers, A. Vlietinck and L. Pieters, Phytochemistry Reviews, 2003, 2,
201–217.
8 S. Van Miert, S. Van Dyck, T. J. Schmidt, R. Brun, A. Vlietinck, G.
Lemiere and L. Pieters, Bioorg. Med. Chem., 2005, 13, 661–669.
9 A. K. Prasad, V. Kumar, P. Arya, S. Kumar, R. Dabur, N. Singh, A.
K. Chhillar, G. L. Sharma, B. Ghosh, J. Wengel, C. E. Olsen and V. S.
Parmar, Pure Appl. Chem., 2005, 77, 25–40.
10 S. Apers, D. Paper, J. Buergermeister, S. Baronikova, S. Van Dyck,
G. Lemiere, A. Vlietinck and L. Pieters, J. Nat. Prod., 2002, 65, 718–
720.
11 E. L. Ghisalberti, Phytomedicine, 1997, 4, 151–166.
12 J. L. Hartwell and W. E. Detty, J. Am. Chem. Soc., 1950, 72, 246–253.
13 M. Gordaliza, M. A. Castro, J. M. Miguel del Corral and A. San
Feliciano, Curr. Pharm. Des., 2000, 6, 1811–1839.
14 C. Spatafora and C. Tringali, Targets in Heterocyclic Systems. Chemistry
and Properties, 2007, 11, 284–312.
15 L. B. Davin, H.-B. Wang, A. L. Crowell, D. L. Bedgar, D. M. Martin,
S. Sarkanen and N. G. Lewis, Science, 1997, 275, 362–366.
16 C. Daquino, A. Rescifina, C. Spatafora and C. Tringali, Eur. J. Org.
Chem., 2009, 6289–6300.
17 (a) F. M. Da Cunha, D. Duma, J. Assreuy, F. C. Buzzi, R. Niero,
M. M. Campos and J. B. Calixto, Free Radical Res., 2004, 38, 1241–
1253; (b) Y.-T. Lee, M.-J. Don, P.-S. Hung, Y.-C. Shen, Y.-S. Lo, K.-W.
Chang, C.-F. Chen and L.-K. Ho, Cancer Lett., 2005, 223, 19–25; (c) X.
Debing, W. Dong, H. Yujun, X. Jiayin, Z. Zhaoyang, L. Zengpeng and
X. Jiang, Anti-Cancer Drugs, 2006, 17, 753–762; (d) H. H. Jin, P. H.
Joo, C. Hwa-Jin, M. Hye-Young, P. Eun-Jung, H. Ji-Young and L. S.
Kook, J. Nutr. Biochem., 2006, 17, 356–362.
18 T. Tanaka, A. Nishimura, Y. Kouno, G. Nonaka and C-R. Yang, Chem.
Pharm. Bull., 1997, 45, 1596–1600.
19 S. A. Souza da Silva, A. L. Souto, M. de Fatima Agra, E. V. Leitao
da-Cunha, J. M. Barbosa-Filho, M. Sobral da Silva and R. Braz-Filho,
Arkivoc, 2004, 6, 54–58.
20 S. Shi, Y. Zhang, K. Huang, S. Liu and Y. Zhao, Food Chem., 2008,
108, 402–406.
21 (a) S. Maeda, H. Masuda and T. Tokoroyama, Chem. Pharm. Bull.,
1994, 42, 2506–2513; (b) S. Maeda, H. Masuda and T. Tokoroyama,
Chem. Pharm. Bull., 1995, 43, 935–940.
22 X. Li, S. Shi, Y. Xu, Q. Tao, J. Stockhit, and Y. Zhao, CN Pat.,
101024640, 2007.
23 S. Di Micco, C. Bassarello, G. Bifulco, R. Riccio and L. Gomez-Paloma,
Angew. Chem., Int. Ed., 2006, 45, 224–228.
24 M. Mayer and B. Meyer, Angew. Chem., Int. Ed., 1999, 38, 1784–1788;
M. Mayer and B. Meyer, J. Am. Chem. Soc., 2001, 123, 6108–6117.
25 I. Gomez-Monterrey, P. Campiglia, A. Carotenuto, D. Califano, C.
Pisano, L. Vesci, T. Lama, A. Bertamino, M. Sala, A. Mazzella di
Bosco, P. Grieco and E. Novellino, J. Med. Chem., 2007, 50, 1787–
1798.
The presented results can inspire the synthesis of new DNA
ligands designed for establishing additional minor groove contacts,
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 701–710 | 709
©

View Online
26 L. Martino, A. Virno, B. Pagano, A. Virgilio, S. Di Micco, A. Galeone,
C. Giancola, G. Bifulco, L. Mayol and A. Randazzo, J. Am. Chem.
Soc., 2007, 129, 16048–16056.
27 G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart,
R. K. Belew and A. J. Olson, J. Comput. Chem., 1998, 19, 1639–
1662.
28 (a) H. P. Spielmann, D. E. Wemmer and J. P. Jacobsen, Biochemistry,
1995, 34, 8542–8553; (b) J. Gallego and B. R. Reid, Biochemistry, 1999,
38, 15104–15115; (c) A. K. Todd, A. Adams, J. H. Thorpe, W. A.
Denny, L. P. G. Wakelin and C. J. Cardin, J. Med. Chem., 1999, 42,
536–540; (d) A. Canals, M. Purciolas, J. Aymami and M. Coll, Acta
Crystallogr., Sect. D: Biol. Crystallogr., 2005, 61, 1009–1012; (e) A.
Adams, J. M. Guss, W. A. Denny and L. P. Wakelin, Nucleic Acids Res.,
2002, 30, 719–725; (f) J. N. Lisgarten, M. Coll, J. Portugal, C. W. Wright
and J. Aymami, Nat. Struct. Biol., 2001, 9, 57–60; (g) X. Shui, M. E.
Peek, L. A. Lipscomb, A. P. Wilkinson, L. D. Williams, M. Gao, C.
Ogata, B. P. Roques, C. Garbay-Jaureguiberry, A. P. Wilkinson and L.
D. Williams, Curr. Med. Chem., 2000, 7, 59–71; (h) H. Robinson, W.
Priebe, J. B. Chaires and A. H. Wang, Biochemistry, 1997, 36, 8663–
8670.
29 MacroModel, version 8.5, Schro¨dinger LLC, New York, NY, 2003.
30 W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz Jr,
D. M. Ferguson, D. C. Spellmeyer, T. Fox, J. W. Caldwell and P. A.
Kollman, J. Am. Chem. Soc., 1995, 117, 5179–5197; P. K. Weiner and
P. A. Kollman, J. Comput. Chem., 1981, 2, 287–303; S. J. Weiner, P. A.
Kollman, D. A. Case, U. C. Singh, C. Ghio, G. Alagona, S. Profeta Jr
and P. K. Weiner, J. Am. Chem. Soc., 1984, 106, 765–784; S. J. Weiner,
P. A. Kollman, D. T. Nguyen and D. A. Case, J. Comput. Chem., 1986,
7, 230–252.
31 W. C. Still, A. Tempczyk, R. C. Hawley and T. Hendrickson, J. Am.
Chem. Soc., 1990, 112, 6127–6129.
32 T. A. J. Halgren, J. Comput. Chem., 1996, 17, 616–641.
33 P. J. Goodford, J. Med. Chem., 1985, 28, 849–857.
34 M. F. Sanner, J. Mol. Graphics Mod., 1999, 17, 57–61.
35 M. F. Sanner, A. J. Olson and J. C. Spehner, Biopolymers, 1996, 38,
305–320.
710 | Org. Biomol. Chem., 2011, 9, 701–710
This journal is
The Royal Society of Chemistry 2011
©