Structure-activity relationships of novel substituted naphthalene diimides as anticancer agents

Article abstract of DOI:10.1016/j.ejmech.2012.06.045

Novel 1,4,5,8-naphthalenetetracarboxylic diimide (NDI) derivatives were synthesized and evaluated for their antiproliferative activity on a wide number of different tumor cell lines. The prototypes of the present series were derivatives 1 and 2 characterized by interesting biological profiles as anticancer agents. The present investigation expands on the study of structure-activity relationships of prototypes 1 and 2, namely, the influence of the different substituents of the phenyl rings on the biological activity. Derivatives 3-22, characterized by a different substituent on the aromatic rings and/or a different chain length varying from two to three carbon units, were synthesized and evaluated for their cytostatic and cytotoxic activities. The most interesting compound was 20, characterized by a linker of three methylene units and a 2,3,4-trimethoxy substituent on the two aromatic rings. It displayed antiproliferative activity in the submicromolar range, especially against some different cell lines, the ability to inhibit Taq polymerase and telomerase, to trigger caspase activation by a possible oxidative mechanism, to downregulate ERK 2 protein and to inhibit ERKs phosphorylation, without acting directly on microtubules and tubuline. Its theoretical recognition against duplex and quadruplex DNA structures have been compared to experimental thermodynamic measurements and by molecular modeling investigation leading to putative binding modes. Taken together these findings contribute to define this compound as potential Multitarget-Directed Ligands interacting simultaneously with different biological targets.

Full text of DOI:10.1016/j.ejmech.2012.06.045

European Journal of Medicinal Chemistry 57 (2012) 417e428
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Structureeactivity relationships of novel substituted naphthalene diimides as
anticancer agents
Andrea Milelli a, Vincenzo Tumiatti a, Marialuisa Micco a, Michela Rosini a, Guendalina Zuccari a,
Lizzia Raffaghello b, Giovanna Bianchi b, Vito Pistoia b, J. Fernando Díaz c, Benet Pera c, Chiara Trigili c,
Isabel Barasoain c, Caterina Musetti d, Marianna Toniolo d, Claudia Sissi d, Stefano Alcaro e,
,
Federica Moraca e, Maddalena Zini f, Claudio Stefanelli f, Anna Minarini a
*
a Dipartimento di Scienze Farmaceutiche, Alma Mater Studiorum, University di Bologna, Via Belmeloro 6, 40126 Bologna, Italy
b Laboratory of Oncology, G. Gaslini Institute, Via L. Gaslini 5, 16147 Genova, Italy
c Centro de Investigaciones Biológicas, Consejo Superior de Investigaciones Científicas, Ramiro de Maeztu 9, 28040 Madrid, Spain
d Department of Pharmaceutical Sciences, University of Padova, Via Marzolo 5, 35131 Padova, Italy
e Dipartimento di Scienze della Salute, Università degli Studi “Magna Græcia” di Catanzaro, Campus Universitario “S. Venuta”, 88100 Catanzaro, Italy
f Department of Biochemistry “G. Moruzzi”, Alma Mater Studiorum, University of Bologna, Via Irnerio 48, 40126 Bologna, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Novel 1,4,5,8-naphthalenetetracarboxylic diimide (NDI) derivatives were synthesized and evaluated for
their antiproliferative activity on a wide number of different tumor cell lines. The prototypes of the
present series were derivatives 1 and 2 characterized by interesting biological profiles as anticancer
agents. The present investigation expands on the study of structureeactivity relationships of prototypes
1 and 2, namely, the influence of the different substituents of the phenyl rings on the biological activity.
Derivatives 3e22, characterized by a different substituent on the aromatic rings and/or a different chain
length varying from two to three carbon units, were synthesized and evaluated for their cytostatic and
cytotoxic activities. The most interesting compound was 20, characterized by a linker of three methylene
units and a 2,3,4-trimethoxy substituent on the two aromatic rings. It displayed antiproliferative activity
in the submicromolar range, especially against some different cell lines, the ability to inhibit Taq poly-
merase and telomerase, to trigger caspase activation by a possible oxidative mechanism, to downregulate
ERK 2 protein and to inhibit ERKs phosphorylation, without acting directly on microtubules and tubuline.
Its theoretical recognition against duplex and quadruplex DNA structures have been compared to
experimental thermodynamic measurements and by molecular modeling investigation leading to
putative binding modes. Taken together these findings contribute to define this compound as potential
Multitarget-Directed Ligands interacting simultaneously with different biological targets.
Received 28 May 2012
Received in revised form
20 June 2012
Accepted 21 June 2012
Available online 4 July 2012
Keywords:
Anticancer
Naphthalene diimides
Multitarget-directed ligands
Cytotoxicity
Ó 2012 Elsevier Masson SAS. All rights reserved.
1. Introduction
ability [6], to enhance the stabilization of DNA triplexes [7] and to
stabilize [8] or alkylate [9,10] G-quadruplex DNA structure.
Naphthalenediimide (NDI) scaffold is present in a wide number
of anticancer agents which interact with DNA preferentially as
intercalators [1e5]. Additionally, some derivatives characterized by
this core have been optimized to exhibit bis-threading intercalating
One of the main prototype of NDI derivatives is represented by
N-BDMPrNDI (Fig. 1) which showed the ability to intercalate into
steps containing at least one G:C base pair [11]. Recently, we pub-
lished two papers in which we reported a new series of NDI
derivatives as antiproliferative agents [12,13]. These compounds
were characterized by NDI scaffold properly functionalized with
isothiocyanate functions and two basic side chains, respectively.
The two basic functions should enhance, in principle, both solu-
bility in aqueous media and interactions with the DNA phosphate
groups [14]. Accordingly, a 2-methoxy substituent was introduced
on the two aromatic rings, thanks to its ability to improve the
basicity of the two terminal nitrogen atoms of polymethylene
Abbreviations: BrdU, bromodeoxyuridine; CD, circular dicroism; dsDNA, double
strand DNA; DTP, Developmental Therapeutics Program; EtBr, Ethidium bromide;
ERK, extracellular signal regulated kinase; NAC, N-acetylcysteine; NCI, National
Cancer Institute; NDI, 1,4,5,8-naphthalenetetracarboxylic diimide; PDB, Protein
Data Bank; TRAP, telomeric repeat amplification protocol.
* Corresponding author. Tel.: þ39 051 2099709; fax: þ39 051247600.
E-mail address: anna.minarini@unibo.it (A. Minarini).
0223-5234/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2012.06.045

418
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
8, 10, 12, 14, 16, and 18. To note that fluorine derivatives 11 and 12
were investigated for the peculiar effects of this substituent in
many different derivatives [18].
Finally, to improve the positive electronic effects provided by
the methoxy group, the trimethoxybenzyl derivatives 19e22 were
also comprised in this investigation. To note that this residue is
a common pharmacophore characterizing different well-known
anticancer compounds such as Combretastatin A4, Colchicine and
Podophyllotoxin acting on tubulin skeleton, and for this reason
19e22 were also investigated on this additional target.
All derivatives were screened by cytotoxic assays, and the more
potent were also evaluated by focused biological studies in order to
elucidate their molecular mechanism(s) of action. In particular,
DNA interaction, apoptosis activation, specific enzymes inhibition
investigations and molecular modeling studies were performed.
Fig. 1. Chemical structures of N-BDMPrNDI, 1, 2 and general structure of the newly
designed compounds 3e22.
amines [15]. The aim of such investigation was to define the
influence of the length of the two basic side chains on anti-
proliferative activity by exploring linkers comprising from two to
ten methylene units. We identified two lead compounds (1 and 2,
Fig. 1) which showed an interesting antiproliferative activity in the
micromolar range on SKBR-3, CEM and HL60 cell lines. Further-
more, they efficiently interacted with DNA, they triggered caspase
activation, caused p53 protein accumulation, downregulated AKT
survival, and finally, caused a decrease of ERK1/2 and inhibited
ERKs phosphorylation. There are not enough evidences if all these
events are linked each other or if they are independent and related
to the ability of 1 and 2 to interact with these different targets. In
this case these two compounds might be considered as Multitarget-
Directed Ligands (MTDLs) [16]. Since cancer is a complex disease
comprising numerous altered biological networks, this new class of
compounds might represent a more comprehensive treatment of
this pathology [17].
2. Results and discussion
2.1. Chemistry
Compounds 3e22 were synthesized following the procedure
reported in Scheme 1 and were characterized by IR, 1H NMR, 13C
NMR, mass spectra, and elemental analysis. The appropriate
benzaldehydes were treated with 1,2-diaminoethane or 1,3-
diaminopropane to afford the corresponding Schiff bases which
were reduced in situ with sodium borohydride to give compounds
23e42. Condensation of such derivatives with naphthalenete-
tracarboxylic dianhydride led to the corresponding diamine-
diimides 3e22. Bis(p-toluenesulfonate) salts of the final
compounds were prepared to obtain derivatives easier to handle.
In the present work we extended the study of structureeactivity
relationships of prototypes 1 and 2, including the effect of different
substituents on the phenyl rings (3e22, Fig. 1). In particular, the
position of the two 2-methoxy groups on the two aromatic rings as
shifted into 3 and 4, as well as other chemical groups, characterized
by different aromatic effects, were introduced on the side chains
aromatic rings in order to verify a possible influence on the bio-
logical activities. Since the two prototypes differ in the length of the
two basic side chains, varying from two to three carbon units, we
synthesized the two corresponding series of compounds, charac-
terized by different substituents on aromatic rings such as chlorine,
fluorine, nitro, trifluoromethyl, and methyl. In particular the 1-
related series is constituted by the odd final derivatives 3, 5, 7, 9, 11,
13,15, and 17; the 2-related series by the even final compounds 4, 6,
2.2. Biological evaluations
2.2.1. Growth-inhibiting activity
As preliminary screening all new compounds were submitted to
the Developmental Therapeutics Program (DTP) at National Cancer
Institute (NCI) for evaluation of their anticancer activity against
different human cell lines. Among them, compounds 6, 9, 11, 13, 15,
17 were evaluated only to a preliminary test at 10ꢀ5 M concentra-
tion for 48 h incubation (data not shown). Indeed, only compounds
that satisfied predetermined threshold inhibition criteria in
a minimum number of cell lines, based on the analysis of historical
DTP screening data, progressed to the full five concentration assay.
Following these criteria, derivatives 1e3, 5, 7, 8, 10, 12, 19e22 were
Scheme 1. (i) (a) Toluene, reflux, 5 h; (b) NaBH4, EtOH, room temp., 4 h, 24e36%; (ii) 1,4,5,8-naphthalenetetracarboxylic dianhydride, i-PrOH, reflux, 24 h, 20e85%.

A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
419
selected for evaluation in the full panel of about 60 human cancer
cell lines derived from nine human cancer cell types, that have been
grouped in disease sub-panels including leukemia, non small-cell
lung, colon, central nervous system, melanoma, ovarian, renal,
prostate, and breast tumor cell lines. The compounds were dis-
solved in dimethyl sulfoxide and evaluated at five concentrations at
10-fold dilution the highest being 10ꢀ4 M [19e21] in comparison to
our lead compounds 1 and 2. The mean of the cytotoxic activities of
such compounds against different groups of selected cancer cell
lines are reported in Table 1.
Such results are expressed as pGI50 values which are the nega-
tive log of the molar concentration causing 50% growth inhibition.
For several compounds the five concentrations assay was repeated
twice and no significant differences were found.
Finally, to complete the cytotoxic investigation of the 2-series,
the effects of compounds 14, 16, 18, which were not selected by NCI,
in comparison with the reference compound 2 and the most potent
compound 20, on growth inhibition of two representative common
cell lines such as OVCAR-3 and SW620, were additionally investi-
gated (Table 2).
compound displayed the highest values of cytotoxic activity against
different cell lines such as colon and ovarian cells, which were
similar to those reported in literature for some anticancer agents.
These results clearly indicated 20 as the most interesting derivative
of the two series. Thus, it was used as model compound for further
investigations to elucidate its mechanism of action at cellular level.
2.2.2. Effects of compound 20 on tumor cell proliferation
To give some insights into the biological effects of the novel
compounds, the most potent derivative 20, was further investi-
gated in HeLa cells. It showed to largely reduce cell survival (Fig. 2A)
with an IC50 value of 0.85
mM. First, we determined whether it
induced apoptosis by monitoring the activity of caspase proteases
on a proper substrate (Asp-Glu-Val-Asp (EVD), i.e. mainly effector
caspases ꢀ3 and ꢀ7), which represents a marker of apoptotic cell
death [22]. Treatment of HeLa cells with 20 for 24 h, increased
caspase activity in a dose-dependent manner (Fig. 2B), in associa-
tion to the increased number of cells committed to death. Then, we
evaluated the cells nuclear morphology, which is another reliable
indicator of apoptotic cell death. By DAPI staining in cells treated for
Regarding compounds related to 1, we observed that the
removal of the methoxy group (7) or its shift to position 3,
providing 3, caused a decrease of cytotoxic activity of more than
10-folds. At the same time, its introduction in position 4 (5),
restored completely the activity, which is comparable with that of
the lead compound 1. The introduction of two additional methoxy
groups in position 3 and 4 (19) or in 4 and 5 (21), caused a decrease
in the cytotoxic activity in comparison with 1. In the series of 2-
related compounds, we observed that the substitution of the
methoxy group with a hydrogen (8) or a fluorine (12) atom, did not
cause any decrease in activity. On the contrary, the presence of
different substituents such as chlorine (10), nitro (14), tri-
fluoromethyl (16) and methyl (18) groups in 2-position on the two
aromatic rings determined a marked decrease of inhibitory activity
on different cell lines in comparison with 2. In particular, it is to
24 h with 2 mM of 20 (Fig. 2C) it was possible to detect several cells
with characteristic hallmarks of apoptosis, i.e. chromatin conden-
sation, nuclear fragmentation and/or condensation [23].
2.2.3. Signal transduction pathway
The effect of 20 on signal transduction pathway involved in cell
growth was also investigated. In these studies, HeLa cells were
treated with 20 for 20 h, then ERK1/2 mitogen-activated protein
kinases, that are generally associated with cell growth, were
examined. Fig. 3 shows that 20 downregulated the p44 ERK2
protein, and significantly inhibited the phosphorylation of both p42
and p44 ERKs, which are known to influence the survival of cancer
cells [24].
In order to detect whether oxidative stress would play a role in
the cytotoxic mechanism of the novel derivative, we studied the
effect of the antioxidant N-acetylcysteine (NAC) on the survival of
cells treated with 20. Fig. 4 shows that addition of 5 mM NAC to the
culture medium completely inhibited HeLa cell death induced by
20, suggesting the involvement of oxidative mechanisms in the
cytotoxic action of 20.
note that 16 did not show any significant activity up to 10
mM
(Table 2). In this series the pattern of the multi-substitution with
three methoxy groups was different from that observed for 1-
series. In fact compounds 20 and 22, characterized by 2,3,4 and
3,4,5 trimethoxy substitution, respectively, showed a different
biological profile. In particular, 22 had a decreased inhibitory
activity in comparison with 2 while 20 showed the highest levels of
pGI50 values, emerging as the most potent among all the NDI
derivatives toward the different tested cell lines. Notably, this
2.2.4. Annexin V flow cytometric analysis of apoptosis
We evaluated the apoptosis in OVCAR-3 cells treated with 20
(pGI50 value of 6.98, provided by NCI) by staining cells with
Table 1
Growth inhibition, cytostatic and cytotoxic activity of 1e3, 5, 7, 8, 10, 12, and 19e22 in the 60 cell panel in comparison to Vincristine.
Compda
Leukemia
NSCLC
Colon
CNS
Melanoma
Ovarian
Renal
Prostate
Breast
MG-MIDb
1
2
3
5
7
8
10
12
19
20
21
22
Vinc
6.60
6.75
5.57
6.54
5.90
6.85
6.58
6.77
6.42
6.89
6.65
6.48
7.00
6.44
6.69
5.24
6.40
5.64
6.68
6.48
6.75
6.32
6.90
6.28
6.33
6.60
6.70
6.94
5.66
6.75
5.90
7.03
6.86
7.03
6.77
7.06
6.60
6.60
7.00
6.29
6.67
5.50
6.46
5.63
6.81
6.35
6.79
6.20
6.96
6.41
6.34
6.90
6.34
6.72
5.72
6.61
5.79
6.76
6.56
6.77
6.52
6.98
6.28
6.38
6.80
6.53
6.68
5.39
6.57
5.62
6.79
6.65
6.79
6.48
6.85
6.39
6.32
6.50
6.34
6.67
5.46
6.44
5.60
6.75
6.63
6.80
6.50
6.74
6.36
6.18
6.50
6.52
6.73
5.41
6.39
5.57
6.74
6.58
6.74
6.19
6.99
6.66
6.38
6.90
6.34
6.64
5.52
6.49
5.72
6.65
6.49
6.66
6.45
6.82
6.58
6.40
6.50
6.46
6.72
5.50
6.52
5.71
6.78
6.57
6.79
6.43
6.91
6.47
6.38
6.74
a
1e3, 5, 7, 8, 10, 12, and 19e22, bis(p-toluenesulfonate) salt; highest concentration ¼ 10ꢀ4 M unless otherwise reported; only modes showing a value greater than 4.00 are
reported.
b
c
Mean graph midpoint, i.e., the mean concentration for all cell lines.
Vincristine sulfate, highest concentration ¼ 10ꢀ3 M. Data are expressed as the negative log of the molar concentration at three assay end points: the 50% growth inhibitory
power (pGI50).

420
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
Table 2
extent of apoptosis was expressed as the sum of the percentages in
LR and UR quadrants. Compound 20 was effective in inducing
apoptosis in a dose and time dependent manner between the
Cytotoxic activity of 2, 14, 16, 18and 20 against OVCAR-3 and SW620 cells lines.
Compda
n
R
pGI50
b
concentration of 1.5
with 20 at 2.5 M, apoptosis gradually increased to about 3-fold
after 72 h treatment, while at 2.0 M and 1.5 M the increase
was about 5-fold and 6-fold after 72 h, respectively. The mean
values and standard deviations calculated for untreated and treated
cells at different concentrations in combination with the statistical
analysis determined by Mann Whitney test demonstrated that the
differences between control and treated cells are statistically
significant at any time of treatment (P ¼ 0.0286 vs control).
mM and 2.5 mM. Compared to 24 h treatment
OVCAR-3
SW620
MG-MIDd
m
2
3
3
3
3
3
2-OCH3
2-NO2
2-CF3
2-CH3
2,3,4-(OCH3)3
6.62
6.15
nac
5.71
6.98
6.76
6.22
nac
6.04
7.04
6.55
6.18
e
m
m
14
16
18
20
5.88
7.01
a
2, 14, 16, 18, 20, bis(p-toluenesulfonate) salt; see Fig. 1 for general structure.
pGI50 values are the negative log of the molar concentration causing 50% growth
inhibition; results, derived from three different experiments in quadruplicate wells
as compared to that of control cells, are expressed as mean of the three assays.
b
c
Not active at 10
m
M.
d
2.2.5. Tubuline assays
Mean graph midpoint, i.e., the mean concentration for all cell lines.
As reported above, derivatives 19e22 were designed by intro-
ducing the trimethoxyphenyl pharmacophore, peculiar of some
anticancer agents such as colchicine, podophyllotoxin and com-
bretastatin A4 which affect the microtubules structure, with the
aim to hit this specific and important biological target.
These compounds showed notable pGI50 values against A549
cell line and the values are: 6.34 for 19, 6.99 for 20, 6.16 for 21, and
6.37 for 22 (data provided by NCI). To verify their potential inter-
action on this important anticancer target, 19e22 and the reference
compound 2 (pGI50 ¼ 6.70) were tested both in cells and in in vitro
assays. Concerning the first series of assays, A549 cells were incu-
bated for 20 h with 100 times the GI50 of 19e22 and 2. In these
experiments no depolymerization of the cytoplasmic microtubules
as observed (Fig. 6). In particular, when the cells were treated with
21 and 2 (Fig. 6D, F), they became rounded and the microtubule
cytoskeleton looked disorganized but no microtubule
Annexin V-FITC and PI, FACS was used to distinguish and quanti-
tatively determine the percentage of dead, viable, apoptotic and
necrotic cells. As shown in Fig. 5 panel A, OVCAR-3 control and
treated cells were gated into LR (Lower Right), UP (Upper Right), LL
(Lower Left) and UL (Upper Left) quadrants. Cells in LR and UR were
considered as early apoptotic (annexinþ/PIꢀ) and late apoptotic
(annexinþ/PIþ) respectively. Cells in LL and UL were considered
live (annexinꢀ/PIꢀ) and necrotic (annexinþ/PIþ), respectively. In
particular, in the dot plot of flow cytometric analysis the percentage
of apoptosis, detected after 72 h of treatment, increased gradually
according to the 20 concentration: 41% at 1.5
mM, 47% at 2.0 mM and
57% at 2.5 M. These data, taken together with the above experi-
m
ments, confirm the apoptotic mode of cell death. The experiment
was repeated three times with similar results. In Fig. 6 panel B the
Fig. 2. A. HeLa cells were incubated for 48 h in the presence of 20, then the number of alive viable cells was measured by dye exclusion. B. The cells were treated for 24 h with the
indicated dose of 20, afterward cells were collected for caspase activity determination. Data are mean ꢁ S.E.M. of three replicates. C. Morphological evaluation of nuclei stained with
DAPI from control HeLa cells and cells treated with 2 mM of 20 for 24 h.

A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
421
account that NDI derivatives can selectively interact with differ-
ently folded DNA structures accordingly to their substitution
pattern and the nature of the side chains [27,28].
Thus, the G-quadruplex and duplex DNA recognition of the
selected NDI were compared by fluorescence quenching melting
assay using a G-quadruplex folded sequence based on the human
telomeric sequence (G4) and a 18-bp random double stranded DNA
(dsDNA). All derivatives were able to significantly increase the
tested DNAs melting temperatures in a concentration dependent
manner (Fig. 1SI). It emerged that the increment in the melting
temperature was generally more intense for the G-quadruplex
folded sequence then for the dsDNA template reaching saturation
in the low micromolar range and suggesting a strong interaction
with both DNA arrangements.
Fig. 3. HeLa cells were incubated for 20 h in the presence of the indicated concen-
tration of 20. The content and phosphorylation status of ERK1/2 in cell extracts was
determined by Western blotting (50 mg of protein/lane).
In Table 3 we report the shift in the DNA melting temperature
induced by 2.5 mM ligands concentration that confirmed modest
depolymerization was observed. As control 2.5 mM podophillotoxin
was employed and the cellular microtubules completely depoly-
merized (results not shown).
variation in the efficiency of DNA thermal stabilization among the
tested compounds, in particular toward the G-quadruplex struc-
ture. However, 20 appears to be the most efficient G-quadruplex
binder whereas 8, in agreement with EtBr displacement data,
induced the most relevant stabilization on double stranded DNA.
The difference between the recognition of the two DNA foldings
slightly increments accordingly with the number of methoxy
groups whereas, their substitution with a fluorine atom (12) does
not significantly alter the DNA stabilization properties which,
actually, parallels the cytotoxic activity.
To get further insight into the NDI derivatives-DNA interaction,
we performed a quantitative analysis of the binding process toward
the two substrates using 20 as model compound. These data
were obtained by UVeVIS titrations since addition of dsDNA or
G-quadruplex DNA induced a significant reduction of the drug
absorbance (Fig. 7). The results are summarized in Table 4 and
confirmed a relevant affinity of 20 toward both DNA substrates. In
particular, the difference in Ka is likely not sufficient to preclude the
recognition of one target in physiological conditions.
From these data we can additionally confirm that stacking
interactions are relevant for the DNA binding process. Indeed,
dsDNA binding data analysis indicated a complex stoichiometry (n)
of two base pairs for each bound drug molecule, which is in line
with an intercalation binding mode. In the presence of G-quad-
ruplex, we found two NDI molecules bound to one DNA structure.
This likely corresponds to the stacking of the ligands on each
terminal tetrad.
Merging these results, we can infer that by incrementing the
number of the methoxy groups on aromatic rings of the side chains,
the insertion of the aromatic portion (NDI core) between two base
pairs of the double helix can be hindered. Indeed, steric effects can
occur to impair threading of one side chain between two base pairs.
On the opposite, this structural feature should not sensibly affect
the staking mode of the NDI on the external G-quadruplex tetrads.
Since 20 showed to be able to recognize also telomeric
G-quadruplex sequences, we additionally evaluated its ability to
In in vitro assay we tested the assembly of 25
presence of 30 M of 2, 19e21. In the absence of compounds the
critical concentration of tubulin required for assembly was
3.40 ꢁ 0.83 M. In the presence of a stabilizer (docetaxel) the
concentration was of course lower 1.24 ꢁ 0.31 M. In the presence
of a destabilizer (podophyllotoxin) the concentration required was
lower 7.40 ꢁ 0.79 M. In the presence of the compounds 2, 19e21
the required concentration was equal to those determined in the
absence of drug (3.67 ꢁ 0.90 M, 3.65 ꢁ 0.97 M, 3.23 ꢁ 0.86 M,
3.50 ꢁ 1.09 M, respectively) and no effect on tubulin assembly was
mM tubulin in the
m
m
m
m
m
m
m
m
observed. These results demonstrated that 2, 19e21 did not affect
directly microtubules and tubulin. In particular, in A549 cells they
were cytotoxic but they did not impair microtubules functions.
2.2.6. DNA-binding properties
Several NDI derivatives are efficient DNA binders and this
property can result in promotion of cell apoptosis. To preliminary
assess the ability of the most interesting compounds of the 2-series
(8, 12, 20 and 22) to interact with double stranded DNA, a fluoro-
metric intercalator displacement method was applied [25]. DNA-
binding activity was expressed as the drug concentration able to
reduce by 50% the fluorescence of DNA-bound ethidium bromide
(EtBr). This EC50 value allows to estimate the affinity ranking order
of the ligands for calf thymus DNA [26,27]. All tested derivatives
turned out to be strong DNA-interacting molecules (Table 3) with
modest modulation of the EC50 values. Additionally, we take into
inhibit telomerase activity in HeLa cells (IC50 ¼ 0.85
mM) (Table 3).
The results confirmed that 20 as able to reduce the activity of TAQ
demonstrating an interesting biological profile together with the
other cytotoxic effects displayed above (Fig. 2SI).
2.2.7. Molecular modeling studies
The most interesting compound 20 of the novel NDI series has
been submitted to molecular modeling calculations with the aim to
characterize its conformational profile and to propose binding
modes for the duplex and G-quadruplex DNA. Firstly, the ionization
state of 20 has been analyzed with respect to the thermodynamic
experimental conditions (pH ¼ 7.5). The bis cationic form, with
both secondary amines protonated, was indicated as the most
Fig. 4. Cell viability was measured in cells treated for 24
h with the indicated
concentration of 20 in the absence or presence of 5 mM NAC. Results are mean ꢁ S.E.M.
of triplicate determinations.

422
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
Fig. 5. (A) Representative flow cytometry analysis of apoptosis induced by 20 at different concentrations in OVCAR-3 ovarian carcinoma cell line after 72 h of drug exposure, using
Annexin V-FITC/PI double staining method. Quadrant analysis of fluorescence intensity of gated cells in FL-1 (Annexin V-FITC) and FL-3 (PI) channels was from 10,000 events. The
lower left quadrant shows viable cells; the lower right quadrant shows viable cells in early stages of apoptosis; the upper right quadrant shows cells in the later stage of apoptosis;
and the upper lower left quadrant shows dead cells. (B) Quantitative detection of 20-induced apoptosis by Annexin V-FITC/PI staining on OVCAR-3 cell line at different times and
concentrations. White columns: control; light gray columns: 20 at 1.5 mM; gray columns: 20 at 2.0 mM; black columns: 20 at 2.5 mM. Bars represent the mean number of triplicates
wells from three independent experiments; statistical significance of differences between untreated and treated groups at different times was determined by ManneWhitney test,
two tailed, with 95% confidence intervals. *P ¼ 0.0286 vs controls.
prevalent among all ionizable states. Secondly, in this ionization
form, the conformational study has been carried out by the Monte
Carlo search. The number of conformers found within 5 kcal/mol
above the global minimum was equal to 81. Thirdly, the docking
experiments were carried out by selecting the DNA target models
from the Protein Data Bank (PDB) [29]. The selection of the DNA
duplex model was performed by the analysis of intercalative
ligands with resolution factor equal or lower than 1.5 Å. The PDB
entry 1Z3F contains two guanineecytosine intercalative binding
sites and the ligand (ellipticine) characterized by an extended
aromatic core [30]. This structure was selected as template model
for the DNA duplex docking. As concerns the PDB G-quadruplex
models, the availability of an NDI derivative co-crystal with the
human telomeric sequence, led us to select 3CDM entry as refer-
ence template model [31]. Both duplex and G-quadruplex PDB
entries were prepared for the docking by eliminating non nucleic
acid components, such as ligands, counterions and water mole-
cules, included into the original models. As concerns the 3CDM
entry, based on two G-quadruplex subunits, both A and B chains
were considered for the docking simulations. The docking with the
duplex DNA model was performed following the computational
approach successfully tested with intercalative moieties in other
Fig. 6. Effect of 19e22 and reference compound 2 on the cytoplasmic microtubule network of A549 lung carcinoma cells. A549 cells were incubated for 20 h with either A, DMSO, B,
10 M 20, C, 50 M 19, D, 50 M 21, E, 20 M 22, or F, 20 M 2. Microtubules were stained with anti-tubulin monoclonal antibodies (DM1A). The bar is 10 m.
m
m
m
m
m
m

A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
423
Table 3
Table 4
DNA interaction of 1, 2, 8,12, 20 and 22 evaluated by EtBr displacement, fluorescence
quenching melting on dsDNA and G4, and TAQ polymerase inhibition.
Thermodynamic parameters for the DNA binding by 20 determined in 10 mM Tris,
20 mM KCl, pH 7.5, 37 ꢂC. n refers to the number of G-quadruplex or base pairs
involved in the binding of one NDI molecule.
Compda EtBr displacement Fluorescence melting
EC50 (nM)b
D
Tm (ꢂC)c TAQ inhibition
IC50
(m
M)d
Ka$10ꢀ6 (Mꢀ1
)
n
G4
dsDNA
ctDNA
G-quadruplex
dsDNA
6.94 ꢁ 2.05
0.85 ꢁ 0.07
0.49
2.62 ꢁ 0.03
1
2
8
12
20
22
93 ꢁ 4e
122 ꢁ 6
118 ꢁ 8
130 ꢁ 8
166 ꢁ 8
159 ꢁ 8
16.1
15.5
18.5
16.4
20.5
16.1
6.2
7.7
11.3
8.1
8.4
5.2
>40
10 ꢁ 2
3.5 ꢁ 8
>40
8 ꢁ 1
the duplex model. Compound 20 docked with the DNA duplex
model gave free energy of complexation equal to ꢀ92.2 kcal/mol,
and with the B chain of the G-quadruplex equal to ꢀ123.0 kcal/mol,
resulting from the synergy of the top and bottom concurrent poses.
The reason of the high stability found with the DNA duplex can be
understood analyzing the lowest energy pose reported in Fig. 8.
There is a synergy of attractive interactions due to the NDI core
stacking (two contributions) within the guanineecytosine inter-
calation site, one hydrophobic contact and hydrogen bond within
the minor groove and two hydrogen bonds within the major
groove. Conversely, the molecular recognition obtained in the
docking simulations of the ligand 20 with the DNA G-quadruplex
model (B chain) in a 2:1 stoichiometry resulted in a sort of “sand-
wich-type interaction” as shown in Fig. 9. In this ternary complex
the first bound ligand assumed a semi-folded conformation, fitting
the NDI core onto the DNA G tetrads in top position by efficient end
stacking interaction. Both the ionized nitrogen atoms were able to
donate hydrogen bonds, respectively to the phosphate groups of
dG14 and dG20 (Fig. 9b). The second ligand recognized the G tetrad
in the bottom position by stacking of the NDI core, not properly
detected as a pharmacophore feature in Fig. 9c. The reason of the
omitted detection can be addressed to the misalignment among the
aromatic cores of guanines and NDI (data not shown). The side
chains of this second ligand assumed a more extended conforma-
tion each one realizing different DNA interactions. One appeared to
be more involved than the other in stabilizing the interaction with
the G-quadruplex bottom side.
Actually, the ionized nitrogen atom and the trimethoxy aromatic
ring of one side chain established one hydrogen bond with dG22 N3
and hydrophobic contacts with the DNA, respectively. Conversely,
the second side chain engaged interaction only with the dG10
residue. Globally, the “sandwich-type” model seemed to be able to
protect the unwinding process thus explaining the consistent
stabilizing effect of 20 on the G-quadruplex fold. In conclusion, the
docking experiments proposed different binding modes of inter-
action of 20 against duplex and G-quadruplex DNA models. The NDI
core was able to establish stacking interactions via intercalative
recognition with the duplex DNA only.
5 ꢁ 1
a
2, 14, 16, 18, 20, bis(p-toluenesulfonate) salt; see Fig. 1 for general structure.
EC50 values are defined as the drug concentrations which reduce the fluores-
b
cence of the DNA-bound ethidium by 50%.
c
D
m
Tm corresponds to the increment in the DNA melting temperature induced by
2.5
M drug concentration. Errors were ꢁ0.4 ꢂC.
d
IC50 values are defined as the drug concentrations which reduce by 50% the
amplification of a fragment of pBR322 mediated by TAQ polymerase.
e
Data reported in Ref. [12].
antitumor compounds [32]. The target DNA was isolated in
a conformation able to give intercalative recognition with ligands
by stacking and groove interactions. Accordingly to our last re-
ported experiments we have considered two computational
protocols as reported in the experimental section. In both cases,
accordingly to the Moline general protocol [33] and to reported
case studies [34,35], the best poses were submitted to full energy
optimization. The ensembles of optimized poses were then
submitted to the estimation of free energy of complexation at room
temperature. The docking models obtained with Autodock were
able to lead to energy minima lower than those with Moline. So we
have analyzed them in details in order to characterize the recog-
nition of 20 against the duplex and the quadruplex DNA confor-
mations. Following our geometrical descriptor recently described
[34] the Autodock docking exploration of compound 20 around
the G-quadruplex target structure was analyzed in terms of pose
distribution among three main clusters: top, bottom and lateral.
Respectively out of 256 docking solutions we have registered
respectively 49 (19.14%), 132 (51.56%) and 75 (29.30%) poses, con-
firming a reasonably good level of exploration in the docking
procedure. Taking into account the experimentally observed stoi-
chiometry binding data, a 2:1 complex model was built with the
G-quadruplex target. It has been constructed considering the two
lowest Autodock score poses of 20 respectively found in top and
bottom end stacking configurations, as classified in our previous
communication [36]. The ternary model was then submitted to full
optimization and free energy estimation in the same conditions of
Fig. 7. Titrations of 20 (10e20
mM) with DNA in 10 mM Tris, 50 mM KCl, pH 7.5. PANEL A: human telomeric sequence folded in G-quadruplex; PANEL B: ctDNA. Arrows indicate
absorption spectra change upon DNA addition.

424
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
Fig. 8. a) 3D representation of the compound 20 best intercalative pose within the DNA duplex. The ligand and the DNA are respectively displayed as polytube and wireframe
models. The gray surface represents target binding pocket. The hydrophobic features of the ligand and exclusion volume coats onto the target are shown as gray spheres. Inter-
molecular hydrogen bonds are displayed as gray arrows. Positive ionizable target nitrogens are shown as cone features. The ligand aromatic rings detected to interact via
pep
interactions are highlighted with light gray circles with top/bottom triangles. b) 2D ligand representation with the main interaction features with the DNA duplex model. Three
intermolecular hydrogen bonds are represented by dotted vectors. Electrostatic interacting nitrogens are shown as dark gray features. Hydrophobic interacting rings are highlighted
in light gray. NDI core
pep stacked rings are depicted with circles and black arrows. The top ruler indicates the corresponding DNA duplex interaction area.
Nevertheless the ligand 20 was able to establish a good end
stacking interaction of the NDI core with the terminal G-quartet,
especially at the top position. In this docking analysis we have only
considered two conventional, but not exhaustive, DNA target
models: one intercalative sequence for the duplex and one fold for
the G-quadruplex. Further studies, currently in progress, will extend
both target structures and conformations as well as larger number
of ligands.
the two aromatic rings confirming their possible influence on the
biological effects. Derivative 20, characterized by a chain length of
three methylene units and by 2,3,4-trimethoxy groups on the two
aromatic rings, was the most potent of the two series and showed
an interesting biological profile. In fact, it displayed pGI50 values
around 7, demonstrating an improvement of the cytotoxic activity
toward multiple cancer lines, in comparison with lead compounds
1 and 2.
Indeed, 20 showed the ability to tightly bind DNA irrespectively
to its structural arrangement, to inhibit Taq polymerase and telo-
merase, to induce apoptosis by triggering caspase activated
through a possible oxidative mechanism, to downregulate ERK 2
protein and to inhibit ERKs phosphorylation, without acting
directly on microtubules and tubuline. Taken together, these find-
ings contribute to define these new compounds as potential MTDLs,
if interacting simultaneously with the different biological targets.
3. Conclusions
The present article expands on the study of the
structureeactivity relationships of the prototypes 1 and 2, by
investigating the role of the substituents on the two aromatic rings
included in the side chains. The cytostatic and cytotoxic activities of
1 and 2 were affected by the insertion of different substituents on

A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
425
The results of the present investigation allowed us to delineate
the chemical requirements to improve anticancer activity of this
series of difunctionalized NDI derivatives.
4. Experimental section
4.1. Chemistry
4.1.1. General procedure for the synthesis of 3e22
A mixture of the appropriate amine 23e42 and 1,4,5,8-naph-
thalene-tetracarbocylic dianhydride in a 1:5 molar ratio in iso-
propanol were refluxed for 2 h. After cooling down, removal of
the solvent gave residue that was purified by flash chromatography
using as eluent a mixture of dichloromethane/methanol/33%
aqueous ammonia (9:1:0.03) providing the desired products 3e22
that were converted into the corresponding bis-p-toluenesulfo-
nates salt.
4.1.2. 2,7-Bis-[3-(2,3,4-trimethoxybenzylamino)propyl]-benzo[lmn]
[3,8]phenanthroline-1,3,6,8-tetraone (20)
Yellow oil; 46% yield; mp (bis-p-toluenesulfonate salt): 237 ꢂC
dec; 1H NMR (CDCl3, 400 MHz, free base):
d
1.96e2.0 (m, 4H þ 2H
exchangeable with D2O), 2.74 (t, J ¼ 8.0 Hz, 4H), 3.73 (s, 4H), 3.82
(s, 6H), 3.84 (s, 6H), 3.90 (s, 6H), 4.29 (t, J ¼ 8.0 Hz, 4H), 6.56e6.58
(d, J ¼ 8.0 Hz, 2H), 6.92e6.94 (d, J ¼ 8.0 Hz, 2H), 8.73 ppm (s, 4H);
13C NMR (CDCl3, 100 MHz):
d 28.1, 38.9, 46.4, 48.4, 55.8, 60.6, 60.9,
106.8, 123.9, 125.7, 126.3, 126.4, 130.7, 141.9, 151.9, 152.8, 162.6 ppm;
ESI-MS (m/z): 741 [M þ H]þ; Anal. Calcd for C54H60N4O16S2$2H2O: C
57.85, H 5.53, N 5.13; found: C 58.06, H 5.65, N 5.17.
See Supporting Information for further experimental details,
Elemental Analysis, and spectral data (ESI-MS and 1H and 13C NMR).
4.1.3. General procedure for the synthesis of 23e42
A mixture of the appropriate diamine and aldehyde (in a 5:1
molar ratio) in toluene was refluxed in a DeaneStark apparatus for
5 h. Following solvent removal, the residue was taken up in EtOH,
NaBH4 (1:2.5 molar ratio) was added at 0 ꢂC, and the stirring was
continued at room temperature for 4 h. The solvent was then
removed and the residue was dissolved in dichloromethane and
washed with brine. Removal of the dried solvent gave a residue that
was purified by flash chromatography using as eluent a mixture of
dichloromethane/methanol/33% aqueous ammonia (9:1:0.1),
providing the desired products 23e42.
4.1.4. N1-(2,3,4-Trimethoxybenzyl)-propane-1,3-diamine (40)
Yellow oil; 51% yield; 1H NMR (CDCl3, 200 MHz):
d 1.61e1.71 (m,
2H), 2.09 (brs, 3H exchangeable with D2O), 2.64 (t, J ¼ 7.0 Hz, 2H),
2.75 (t, J ¼ 6.6 Hz, 2H), 3.71 (s, 2H), 3.83e3.91 (m, 9H), 6.59e6.63
(d, J ¼ 8.4 Hz, 1H), 6.9e6.94 (d, J ¼ 8.8, 1H); 13C NMR (CDCl3,
100 MHz):
130.2, 145.4, 148.5.
d 22.7, 41.0, 43.3, 51.1, 54.0, 57.2, 57.5, 107.1, 116.3, 121.1,
Fig. 9. a) 3D representation of compound 20 top/bottom best poses obtained with
the DNA G-quadruplex model (PDB entry 3CDM chain B) in 2:1 stoichiometry. The
ligands and the DNA are respectively displayed as polytube and wireframe models.
The gray surface represents target binding pocket. The exclusion volume coats onto
the target are shown as gray spheres. The intermolecular donating hydrogen bonds
are displayed as arrows. Positive ionizable target nitrogens are shown as cone
See Supporting Information for further experimental details,
Elemental Analysis, and spectral data (ESI-MS and 1H and 13C NMR).
4.2. Biology
features. The ligand aromatic ring detected to interact via
pep interactions is high-
lighted with a gray circle with triangles. b) 2D ligand representation with the main
interaction features with the DNA G-quadruplex model in top side. Two intermo-
lecular hydrogen bonds are represented by dotted vectors. Electrostatic interacting
4.2.1. Growth-inhibiting activity
The NCI screening is a two-stage process [20] beginning with
the evaluation of all compounds against the 60 cell lines at a single
concentration of 10ꢀ5 M. Compounds exhibiting significant
growth inhibition were evaluated against the 60 cell panel at five
concentration levels by the NCI according to standard procedures
(http://dtp.nci.nih.gov/branches/btb/ivclsp.html). In both cases the
exposure time was 48 h.
nitrogen atoms are shown as gray features. The
pep stacked ring is depicted with
circles and a black arrow. c) 2D ligand representation with the main interaction
features with the DNA G-quadruplex model in bottom side. One intermolecular
hydrogen bond is represented by a dotted vector. Electrostatic interacting nitrogen
atoms are shown as gray features. Hydrophobic interacting rings are highlighted in
light gray.

426
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
4.2.2. Indirect immunofluorescence of cellular microtubules and
tubulin assembly inhibition assay
bromophenol blue, 4%
6.8), and then denatured by boiling for 4 min. Aliquots corre-
sponding to 80 g protein were analyzed by SDS-PAGE. Proteins
b-mercaptoethanol in 0.25 M TriseHCl, pH
Human A549 non small lung carcinoma cells were cultured in
RPMI 1640 supplemented with 10% FCS, glutamine, and antibiotics
as previously described [36]. Indirect immunofluorescence was
performed in A549 cells that had been cultured overnight in 12 mm
round coverslips and incubated a further 24 h in the absence (drug
vehicle DMSO) or in the presence of different ligand concentrations.
Attached cells were permeabilized with Triton X-100 and fixed with
3.7% formaldehyde. Microtubules were specifically stained with
m
were transferred onto a nitrocellulose membrane and probed with
the specific primary antibody. After further washing, the
membrane was then incubated for 1 h with peroxidase-conjugated
goat anti-rabbit IgG (Santa Cruz). Immunoreactive bands were
visualized by chemiluminescence with the ECL reagent (Amer-
sham). Antibodies against total and phosphorylated ERK1/2 were
from Cell Signaling. Anti actin was obtained from Santa Cruz. Actin
was used as internal control.
DM1A
a-tubulin monoclonal antibodies and DNA with Hoechst
33342 as previously described [37]. The preparations were exam-
ined using a Zeiss axioplan epifluorescence microscope and the
images were recorded in a Hamamatsu 4742-95 cooled CCD camera.
The effect of the compounds in the assembly of purified tubulin
4.2.6. DNA binding assays
The binding to DNA was determined by a fluorometric ethidium
displacement test at pH 8 [25,26] The assay mixture contained 2
mg
was determined by incubating 25
m
M purified tubulin at 37 ꢂC for
of ethidium bromide and 0.5 g of calf thymus DNA in a final
m
30 min in GAB (glycerol assembling buffer, 3.4 M glycerol, 10 mM
sodium phospate, 1 mM EGTA, 1 mM GTP at pH 6.5) in the presence
of 30 mM test compound, docetaxel, or podofilotoxin or 2 mL DMSO
(vehicle). The samples were processed and the critical concentra-
tion for tubulin assembly [38] in the presence of the ligands
calculated as described [39].
volume of 2 ml. Fluorescence (excitation at 525 nm; emission at
600 nm) was determined in triplicate following addition of aliquots
of a 100ꢃ solution of polyamine compounds. The EC50 values are
defined as the drug concentration which reduces the fluorescence
of the DNA-bound ethidium by 50%.
4.2.7. Fluorescence melting assay
4.2.3. Cell death assays: cell culture and treatment
Fluorescence melting curves were determined in a Roche Light
Cycler (lecc 470 nm, lemm 520 nm) in LiP buffer (10 mM LiOH,
50 mM KCl, pH 7.4 with H3PO4). Increasing ligands concentrations
HeLa cells were cultured in HAM’S F-12/MEM containing 10%
fetal bovine serum, 5% glutamine, 1% non-essential amino acids and
antibiotic solution. All tested compounds were dissolved in dime-
thylsulphoxide, diluted to at 1000ꢃ solution and added to cell
cultures. Control cells received the corresponding volume of the
vehicle. Cell viability was determined by trypan blue exclusion by
counting living cells and stained dead cells with a Burker hemo-
cytometer. Samples were done in triplicate, and at least 10 fields
were counted for each sample. At the end of the incubation, once
the number of living cells in samples were measured, cell survival
was calculated as the percentage of living viable cells in treated
samples in respect to the number of viable cells in control samples.
The IC50 value is the concentration of toxic compound required to
reduce cell survival to 50%.
were added to 0.25
mM final concentration of a labeled human
telomeric sequence (50-Dabcyl-AGGGTTAGGGTTAGGGTTAGGGT-
FAM 30, Eurogentec, Belgium). Before scan acquisition the reaction
mixture was first denatured and annealed. Then, samples were
maintained at 30ꢂ for 5 min before being slowly heated to 95 ꢂC
(1 ꢂC/min) and annealed at a rate of 1 ꢂC/min. When double
stranded DNA was used, dsUP (50-FAM-ACTATTCCCGGGTAATGA)
and dsDOWN (TCATTACCCGGGAATAGT-Dabcyl 30) were mixed at
equimolar concentrations, heated to 95 ꢂC for 5 min, and then
cooled to room temperature overnight before use. Recordings were
taken during both the annealing and melting steps. Tm values were
determined from the first derivatives of the melting profiles using
the Roche Light Cycler software. Each curve was repeated at least
three times and errors were ꢁ0.4 ꢂC.
4.2.4. Apoptosis assays
The activity of Caspase protease enzymes was measured by the
cleavage of the fluorogenic peptide substrate AcDEVD-AMC. At the
indicated time points, cells were washed in phosphate-buffered
saline, harvested in 0.4 ml of lysis buffer, and subjected to two
cycles of freeze-thawing. The lysates were centrifuged for 10 min at
28,000ꢃ g at 4 ꢂC and the supernatant was then used to assay in
4.2.8. UVeVIS spectroscopic titrations
Spectrophotometric titrations were performed in 10 mM Tris-,
20 mM KCl, pH 7.5 with a PerkineElmer Lambda 12 apparatus.
Binding was followed by addition of increasing amounts of DNA to
a freshly prepared drug solution. For each drug/DNA ratio the frac-
duplicate enzyme activity. Extract (10
m
L) was incubated for 15 min
tion of bound ligand was calculated (
n
¼ (ε ꢀ ε0)/(εN ꢀ ε0), where
at 37 ꢂC in a final volume of 30
mL to determine caspase activity.
ε0 and εN are the extinction coefficient of the free and DNA-bound
ligand, respectively). To avoid large systematic inaccuracies due to
experimental errors, the range of bound drug fractions utilized for
calculations was 0.15e0.85. For titrations with ctDNA, data were
evaluated according to the equation of McGhee and Von Hippel for
non-coopative ligandelattice interactions [40]. G-quadruplex
isotherms were analyzed according to a single set of identical sites.
The effect of treatments on nuclear morphology was determined
by the visualization of 40,6-diamidino-2-phenylindole (DAPI)-
stained nuclei. At the indicated time point, cells grown in triplicate
on glass coverslips were washed twice with cold PBS, fixed in ice-
cold methanol/acetic acid (1:1, v/v) for 15 min, rinsed with PBS,
and then stained with 0.1 mg/ml of DAPI. After staining, cells were
mounted on standard glass slides and observed with an IX50
Olympus inverted microscope (Olympus, Tokyo, Japan). Apoptotic
cells were identified by nuclear features characteristic of apoptosis
(nuclear shrinkage and/or fragmentation, chromatin condensation).
4.2.9. Taq polymerase assay
Taq polymerase reaction in the presence/absence of ligands and
their metal complexes were performed using pBR322 (2.5 ng) as
a DNA template and appropriate primer sequences Tup and Tdown
4.2.5. Western blotting
(0.5 mM) to amplify the 906e1064 sequence of plasmid by PCR. The
Cells were collected in 5 mM dithiothreitol, 2 mM EDTA, 0.1%
CHAPS, 0.1% Triton X-100, and protease inhibitors in 20 mM HEPES
pH 7.5 and subjected to two cycles of freeze-thawing. The
homogenate was then centrifuged at 15,000ꢃ g for 15 min and the
supernatant, diluted in loading buffer (2% SDS, 5% glycerol, 0.002%
reaction was carried out in an Eppendorf thermocycler performing
25 cycles of: 30 s at 94 ꢂC, 30 s at 65 ꢂC and 30 s at 72 ꢂC. The
reaction products were resolved on a 2% agarose gel in 1X TBE
(89 mM Tris base, 89 mM boric acid, 2 mM Na2EDTA) and stained by
ethidium bromide.

A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
427
4.2.10. Telomerase activity assay
confidence intervals using GraphPad Prism 3.0 software (GraphPad
Software, Inc.).
8 ꢃ 104 HeLa cells in exponential phase of growth were seeded
in a 6-wells plate. After 24 h they were treated with increasing
concentration of drug and incubated for further 24 h. Then, cells
4.2.14. Molecular modeling
were pelleted and lysed for 30 min on ice using 100
m
L of 0.5%
The molecular modeling studies were performed using different
computational methods. The ionization state was analyzed by the
LigPrep program [41]. The conformational analysis of compound 20
was performed by the Monte Carlo search as implemented in
MacroModel (version 9.0, 2009) [42], generating 10000 confor-
mations and energy minimizing them by the AMBER* force field
[43] with the united atom notation and the GB/SA water solvation
model [44]. Docking experiments were carried out with two
computational approaches respectively based on Autodock [45]
and Moline [46] programs. Full energy optimizations of DNA-20
complexes were carried out in the same conditions used for the
conformational search of the isolated compound 20. The G-quad-
ruplex pose distribution with the Autodock ensemble was carried
out setting as dummy atoms DU1, DU2 and DU3 respectively
defined by the four N1 atoms of guanines in the bottom triad, the
four N1 atoms of guanines in the central triad, and the aromatic
carbon atoms of the NDI core. The angle descriptor was used to
address the classification for each configuration: cluster bottom
(angle within 0e45ꢂ); cluster lateral (angle within 45e135ꢂ),
cluster top (angle within 135e180ꢂ) [35]. The visual analysis and
the construction of Figs. 8 and 9 were performed by the Ligand
Scout software [47].
CHAPS, 1 mM EGTA, 25% 2-mercaptoethanol, 1.74% PMSF and 10%
w/v glycerol. The lysate was centrifuged at 13000 rpm for 30 min at
4
ꢂC and the supernatant used as telomerase source. Total protein
concentration of the lysates was determined by Bio-Rad protein
assay kit using BSA as standard. Telomerase activity was assayed
using TRAPeze telomerase detection kit (Chemicon International).
The reaction products were loaded onto a 10% polyacrylamide gel
(19:1) in TBE 0.5X. Gels were stained with SbrGreen I.
4.2.11. In vitro screening in human tumor cell lines
Drug stock solutions (1 ꢃ 10ꢀ2 M) were prepared in cell culture
DMSO. These were initially diluted in phosphate buffer saline
(1 ꢃ10ꢀ3 M) and finally in complete growth medium (1 ꢃ 10ꢀ4 M).
Several mL were taken and put into cell culture plates to obtain
different drug concentrations (highest final DMSO concentration
was 0.05%). All controls contained the highest DMSO concentration
used in the treated samples. Cells were plated in 96-well tissue
culture plates, allowed to attach 24 h, and then left untreated or
treated with growth medium containing 0.25, 0.50, 0.75, 1.0, 1.25,
1.5, 1.75, 2.0 mM 2, 14, 16, or 18. Briefly, fifteen mL of 5 mg/mL 3-(4,5-
diphenyltetrazolium) bromide (MTT) in PBS were then added to
each well and cells were incubated for 4 h at 37 ꢂC. Formazan
crystals were made soluble with DMSO (all reagents were from
SigmaeAldrich) (St. Louis, MO, USA). Optical densities were deter-
mined at 570 nm using a Dynatech MR5000 plate reader. Viability
was expressed as a percentage of control by dividing the absor-
bance of each treated well by the average of the untreated controls.
Cytotoxicity was determined by plotting cell viability versus drug
concentrations. Results, derived from three different experiments,
are expressed as mean percentage from quadruplicate wells as
compared to that of control cells. The GI50 of the tested compounds
were estimated graphically from the doseeresponse curves after
48 h of drug exposure.
Acknowledgment
This research was supported by a grant from MIUR, Rome
(PRIN), University of Bologna (RFO) and Polo Scientifico-Didattico
di Rimini (to V.T.). We thank the National Cancer Institute
(Bethesda, MD) for the anticancer assays. MICINN (Spanish
Government) is acknowledged for grant BIO2010-16351 (to J.F.D.).
Lizzia Raffaghello is a recipient of MFAG Grant. Giovanna Bianchi is
a recipient of a FIRC fellowship.
Appendix A. Supplementary material
4.2.12. Annexin V-FITC/PI stained fluorescence-activated cell sorter
(FACS)
Supplementary material associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.ejmech.
2012.06.045.
To assess the effect of 20 on apoptosis, OVCAR 3.0 cells were
stained with Annexin V-FITC/PI according to the manufacturer’s
instructions (Immunostep, S.L.; Salamanca, Spain). Briefly, adherent
cells were plated at a concentration of 90 ꢃ 105 cells/well, 6-well
References
plate and after 24 h treated with 20 at 1.5, 2.0 or 2.5 mM. After
[1] S. Takenaka, M. Takagi, Threading intercalators as a new DNA structural probe,
Bull. Chem. Soc. Jpn. 72 (1999) 327e337.
48 h or 72 h incubation cells were detached with scraper and
washed twice with cold PBS and then suspended in 1ꢃ Binding
Buffer at a concentration of 1 ꢃ106 cells/mL. Hundred microliters of
[2] M.F. Brana, M. Cacho, A. Gradillas, B. de Pascual-Teresa, A. Ramos, Intercalators
as anticancer drugs, Curr. Pharm. Des. 7 (2001) 1745e1780.
[3] R. Martinez, L. Chacon-Garcia, The search of DNA-intercalators as antitumoral
drugs: what it worked and what did not work, Curr. Med. Chem. 12 (2005)
127e151.
[4] L.M. Tumir, I. Piantanida, Recognition of single stranded and double stranded
DNA/RNA sequences in aqueous medium by small bis-aromatic derivatives,
Mini Rev. Med. Chem. 10 (2010) 299e308.
[5] L. Strekowski, B. Wilson, Noncovalent interactions with DNA: an overview,
Mutat. Res. 623 (2007) 3e13.
[6] T. Nojima, K. Ohtsuka, T. Nagamatsu, S. Takenaka, Bis-naphthalene diimide
exhibiting an effective bis-threading intercalating ability, Nucleic Acids Res.
Suppl. (2003) 123e124.
the suspension were transferred to a 5 mL tube and 5 mL of Annexin
V-FITC and 5
m
L of PI were added. The samples were gently vortexed
and incubated for 15 min at 25 ꢂC in the dark. Finally, 400
mL of 1ꢃ
Binding Buffer were added to each tube and the samples were
analyzed by flow cytometry within 1 h at two wavelengths 515 and
639 nm. The used flow cytometer was a FACScan (Becton Dick-
inson) and the analysis was performed with Cell Quest software.
Vehicle (DMSO) treated unstained and stained cells were used as
controls.
[7] D.A. Gianolio, L.W. McLaughlin, Tethered naphthalene diimide intercalators
enhance DNA triplex stability, Bioorg. Med. Chem. 9 (2001) 2329e2334.
[8] S. Balasubramanian, S. Neidle, G-quadruplex nucleic acids as therapeutic
targets, Curr. Opin. Chem. Biol. 13 (2009) 345e353.
4.2.13. Statistics
For the experiments involving in vitro drug treatment mean
values and standard deviation for untreated and treated groups at
different times were calculated. In addition, statistical significance
of differences between untreated and treated groups at different
times was determined by ManneWhitney test, two tailed, with 95%
[9] M. Di Antonio, F. Doria, S.N. Richter, C. Bertipaglia, M. Mella, C. Sissi,
M. Palumbo, M. Freccero, Quinone methides tethered to naphthalene diimides
as selective G-quadruplex alkylating agents, J. Am. Chem. Soc. 131 (2009)
13132e13141.
[10] M. Gunaratnam, S. Swank, S.M. Haider, K. Galesa, A.P. Reszka, M. Beltran,
F. Cuenca, J.A. Fletcher, S. Neidle, Targeting human gastrointestinal stromal

428
A. Milelli et al. / European Journal of Medicinal Chemistry 57 (2012) 417e428
tumor cells with a quadruplex-binding small molecule, J. Med. Chem. 52
(2009) 3774e3783.
[11] Z.R. Liu, K.H. Hecker, R.L. Rill, Selective DNA binding of (N-alkylamine)-
substituted naphthalene imides and diimides to GþC-rich DNA, J. Biomol.
Struct. Dynam. 14 (1996) 331e339.
[12] A. Minarini, A. Milelli, V. Tumiatti, L. Ferruzzi, M.R. Marton, E. Turrini, P. Hrelia,
C. Fimognari, Design, synthesis and biological evaluation of new naphtalene
diimides bearing isothiocyanate functionality, Eur. J. Med. Chem. 48 (2012)
124e131.
[13] V. Tumiatti, A. Milelli, A. Minarini, M. Micco, A. Gasperi Campani, L. Roncuzzi,
D. Baiocchi, J. Marinello, G. Capranico, M. Zini, C. Stefanelli, C. Melchiorre,
Design, synthesis, and biological evaluation of substituted naphthalene
imides and diimides as anticancer agent, J. Med. Chem. 52 (2009)
7873e7877.
[29] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig,
I.N. Shindyalov, P.E. Bourne, The Protein Data Bank, Nucleic Acids Res. 28
(2000) 235e242.
[30] A. Canals, M. Purciolas, J. Aymami, M. Coll, The anticancer agent ellipticine
unwinds DNA by intercalative binding in an orientation parallel to base pairs,
Acta Crystallogr. D Biol. Crystallogr. 61 (2005) 1009e1012.
[31] G.N. Parkinson, F. Cuenca, S. Neidle, Topology conservation and loop flexibility
in quadruplex-drug recognition: crystal structures of inter- and intra-
molecular telomeric DNA quadruplex-drug complexes, J. Mol. Biol. 381 (2008)
1145e1156.
[32] S. Alcaro, F. Ortuso, R.S. Coleman, DNA cross-linking by azinomycin B: Monte
Carlo simulations in the evaluation of sequence selectivity, J. Med. Chem. 45
(2002) 861e870.
[33] S. Alcaro, F. Gasparrini, O. Incani, S. Mecucci, D. Misiti, M. Pierini, C. Villani,
A “quasi-flexible” automatic docking processing for studying stereoselective
recognition mechanisms. Part I. Protocol validation, J. Comput. Chem. 21
(2000) 515e530.
[14] H.J. Schneider, Ligand binding to nucleic acids and proteins: does selectivity
increase with strength? Eur. J. Med. Chem. 43 (2008) 2307e2315.
[15] V. Tumiatti, M. Rosini, M. Bartolini, A. Cavalli, G. Marucci, V. Andrisano,
P. Angeli, R. Banzi, A. Minarini, M. Recanatini, C. Melchiorre, Structure-
activity relationships of acetylcholinesterase noncovalent inhibitors
[34] S. Alcaro, A. Artese, J.N. Iley, S. Missailidis, F. Ortuso, L. Parrotta, R. Pasceri,
F. Paduano, C. Sissi, F. Trapasso, M.G. Vigorita, Rational design, synthesis,
biophysical and antiproliferative evaluation of fluorenone derivatives with
DNA G-quadruplex binding properties, ChemMedChem 5 (2010) 575e583.
[35] S. Alcaro, G. Costa, S. Distinto, F. Moraca, F. Ortuso, L. Parrotta, A. Artese, The
polymorphisms of DNA G-quadruplex investigated by docking experiments
with telomestatin enantiomers, Curr. Pharm. Des. 18 (2011) 1873e1879.
[36] R.M. Buey, E. Calvo, I. Barasoain, O. Pineda, M.C. Edler, R. Matesanz, G. Cerezo,
C.D. Vanderwal, B.W. Day, E.J. Sorensen, J.A. Lopez, J.M. Andreu, E. Hamel,
J.F. Diaz, Cyclostreptin binds covalently to microtubule pores and lumenal
taxoid binding sites, Nat. Chem. Biol. 3 (2007) 117e125.
based on
a polyamine backbone. 2. Role of the substituents on the
phenyl ring and nitrogen atoms of caproctamine, J. Med. Chem. 46 (2003)
954e966.
[16] A. Cavalli, M.L. Bolognesi, A. Minarini, M. Rosini, V. Tumiatti, M. Recanatini,
C. Melchiorre, Multi-target-directed ligands to combat neurodegenerative
diseases, J. Med. Chem. 51 (2008) 347e372.
[17] A. Petrelli, G. Valabrega, Multitarget drugs: the present and the future of
cancer therapy, Expert Opin. Pharmacother. 10 (2009) 589e600.
[18] W.K. Hagmann, The many roles for fluorine in medicinal chemistry, J. Med.
Chem. 51 (2008) 4359e4369.
[19] M.R. Grever, S.A. Schepartz, B.A. Chabner, The National Cancer Institute:
cancer drug discovery and development program, Semin. Oncol. 19 (1992)
622e638.
[37] C. de Ines, D. Leynadier, I. Barasoain, V. Peyrot, P. Garcia, C. Briand, G.A. Rener,
C. Temple Jr., Inhibition of microtubules and cell cycle arrest by a new
1-deaza-7,8-dihydropteridine antitumor drug, CI 980, and by its chiral isomer,
NSC 613863, Cancer Res. 54 (1994) 75e84.
[20] A. Monks, D. Scudiero, P. Skehan, R. Shoemaker, K. Paull, D. Vistica, C. Hose,
J. Langley, P. Cronise, A. Vaigro-Wolff, et al., Feasibility of a high-flux anti-
cancer drug screen using a diverse panel of cultured human tumor cell lines,
J. Natl. Cancer Inst. 83 (1991) 757e766.
[21] J.N. Weinstein, T.G. Myers, P.M. O’Connor, S.H. Friend, A.J. Fornace Jr.,
K.W. Kohn, T. Fojo, S.E. Bates, L.V. Rubinstein, N.L. Anderson, J.K. Buolamwini,
W.W. van Osdol, A.P. Monks, D.A. Scudiero, E.A. Sausville, D.W. Zaharevitz,
B. Bunow, V.N. Viswanadhan, G.S. Johnson, R.E. Wittes, K.D. Paull, An
information-intensive approach to the molecular pharmacology of cancer,
Science 275 (1997) 343e349.
[22] I.N. Lavrik, A. Golks, P.H. Krammer, Caspases: pharmacological manipulation
of cell death, J. Clin. Invest. 115 (2005) 2665e2672.
[23] J.D. Robertson, S. Orrenius, Molecular mechanisms of apoptosis induced by
cytotoxic chemicals, Crit. Rev. Toxicol. 30 (2000) 609e627.
[24] K. Balmanno, S.J. Cook, Tumour cell survival signalling by the ERK1/2 pathway,
Cell Death Differ. 16 (2009) 368e377.
[38] F. Oosawa, S. Asakura, Thermodynamics of the Polymerization of Proteins.
London (1975).
[39] R.M. Buey, I. Barasoain, E. Jackson, A. Meyer, P. Giannakakou, I. Paterson,
S. Mooberry, J.M. Andreu, J.F. Diaz, Microtubule interactions with chemically
diverse stabilizing agents: thermodynamics of binding to the paclitaxel site
predicts cytotoxicity, Chem. Biol. 12 (2005) 1269e1279.
[40] J. McGhee, P. von Hippel, Theoretical aspects of DNA-protein interactions: co-
operative and non co-operative binding of large ligands to a one-dimensional
homogeneous lattice, J. Mol. Biol. 86 (1974) 469e489.
[41] LigPrep, Version 2.3, Schroedinger LLC, New York, 2009.
[42] F. Mohamadi, N.G.J. Richards, W.C. Guida, R. Liskamp, M. Lipton, C. Caufield,
G. Chang, T. Hendrickson, W.C. Still, Macromodel - an integrated software
system for modeling organic and bioorganic molecules using molecular
mechanics, J. Comput. Chem. 11 (1990) 440e467.
[43] D.Q. Mcdonald, W.C. Still, Amber-asterisk torsional parameters for the peptide
backbone, Tetrahedron Lett. 33 (1992) 7743e7746.
[25] A.R. Morgan, J.S. Lee, D.E. Pulleyblank, N.L. Murray, D.H. Evans, Review:
ethidium fluorescence assays. Part 1. Physicochemical studies, Nucleic Acids
Res. 7 (1979) 547e569.
[44] D. Hasel, T.F. Hendrickson, W.C. Still, A rapid approximation to the solvent
accessible surface areas of atoms, Tetrahedron Comput. Methodol. 1 (1988)
103e116.
[26] A.W. McConnaughie, T.C. Jenkins, Novel acridine-triazenes as prototype
combilexins: synthesis, DNA binding, and biological activity, J. Med. Chem. 38
(1995) 3488e3501.
[27] C. Sissi, L. Lucatello, A. Paul Krapcho, D.J. Maloney, M.B. Boxer, M.V. Camarasa,
G. Pezzoni, E. Menta, M. Palumbo, Tri-, tetra- and heptacyclic perylene
analogues as new potential antineoplastic agents based on DNA telomerase
inhibition, Bioorg. Med. Chem. 15 (2007) 555e562.
[28] F. Cuenca, O. Greciano, M. Gunaratnam, S. Haider, D. Munnur, R. Nanjunda,
W.D. Wilson, S. Neidle, Tri- and tetra-substituted naphthalene diimides as
potent G-quadruplex ligands, Bioorg. Med. Chem. Lett. 18 (2008)
1668e1673.
[45] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew,
A.J. Olson, Automated docking using a Lamarckian genetic algorithm and an
empirical binding free energy function, J. Comput. Chem. 19 (1998)
1639e1662.
[46] S. Alcaro, F. Gasparrini, O. Incani, L. Caglioti, M. Pierini, C. Villani, “Quasi
flexible” automatic docking processing for studying stereoselective recogni-
tion mechanisms, part 2: prediction of Delta Delta G of complexation and H-1-
NMR NOE correlation, J. Comput. Chem. 28 (2007) 1119e1128.
[47] G. Wolber, T. Langer, LigandScout: 3-d pharmacophores derived from protein-
bound Ligands and their use as virtual screening filters, J. Chem. Inf. Model. 45
(2005) 160e169.