Synthesis of?phenanthridinium-bis-nucleobase conjugates, interactions with poly U, nucleotides and?in?vitro antitumour activity of?mono- and?bis-nucleobase conjugates

Article abstract of DOI:10.1016/j.ejmech.2006.05.005

Novel bis-nucleobase-phenanthridinium conjugates were synthesised and their aqueous solutions spectroscopically characterised. Bis-adenine conjugate revealed in aqueous solutions significantly more pronounced intramolecular aromatic stacking interactions

Full text of DOI:10.1016/j.ejmech.2006.05.005

European Journal of Medicinal Chemistry 41 (2006) 1153–1166
http://france.elsevier.com/direct/ejmech
Original article
Synthesis of phenanthridinium–bis-nucleobase conjugates,
interactions with poly U, nucleotides and in vitro antitumour
activity of mono- and bis-nucleobase conjugates
L.-M. Tumir a, I. Piantanida a,*, M. Žinić a, I. Juranović Cindrić b, Z. Meić b, M. Kralj c, S. Tomić d
a Laboratory of Supramolecular and Nucleoside Chemistry, Division of Organic Chemistry and Biochemistry,
Ruđer Bošković Institute, Bijenicka 54, HR 10002 Zagreb, P.O.B. 180, Croatia
b Laboratory of Analytical Chemistry, Department of Chemistry, Faculty of Science, Horvatovac 102a, HR 10000 Zagreb, Croatia
c Laboratory of Functional Genomics, Division of Molecular Medicine, Ruđer Bošković Institute, P.O.Box 180, HR-10002 Zagreb, Croatia
d Laboratory for Chemical and Biological Crystallography, Division of Physical Chemistry, Ruđer Bošković Institute, HR 10002 Zagreb, P.O.B. 180, Croatia
Received in revised form 25 April 2006; accepted 4 May 2006
Available online 21 June 2006
Abstract
Novel bis-nucleobase–phenanthridinium conjugates were synthesised and their aqueous solutions spectroscopically characterised. Bis-adenine
conjugate revealed in aqueous solutions significantly more pronounced intramolecular aromatic stacking interactions than bis-uracil analogue. In
contrast with previously reported poly A recognition by bis-uracil conjugate, recognition of complementary nucleotides and poly U was not
observed due to the strong interference of bulk water with hydrogen bonding between nucleobases. The screening of anticancer activity on six
human cell lines revealed that tethering of a nucleobase to phenanthridinium moiety diminished antiproliferative potential of phenanthridinium.
However, among mono-nucleobase conjugates adenine derivative was found to be the most selective one (MiaPaCa–2, Hep–2).
© 2006 Elsevier Masson SAS. All rights reserved.
Keywords: Phenanthridinium–bis-nucleobase conjugates; Nucleotides; Polynucleotides; In vitro antitumour activity
1. Introduction
are also susceptible to the influence of ligand binding with
consequences for recognition by repair enzymes, a number of
small molecules were synthesised that bind specifically at such
lesions with an idea to inhibit the DNA repair system and in
that way pronounce the action of antitumour drugs [8]. Most of
studied molecules incorporated three structural units: (a) an in-
tercalator unit for strong binding to DNA, (b) a nucleic base for
recognition of the complementary non-paired base at the abasic
site, and (c) a linker connecting the intercalator and the base of
a suitable length and flexibility to enable pairing of a tethered
base with that at the abasic site. Linkers of many such interca-
lator–nucleobase conjugates actively contributed in binding to
DNA and under some experimental conditions exhibited also
strand cleavage [8]. In previous research we have prepared a
series of phenanthridinium–nucleobase conjugates (Scheme 1),
characterised by short aliphatic linkers inert toward DNA/
RNA, that have shown intriguingly different specificity of
binding to ss-polynucleotides [9–11]. Inspired by observed se-
lectivity we have studied interactions of novel phenanthridi-
It is well known that non-covalent interactions between
DNA/RNA and peptides or smaller molecules are essential
for all processes in living cells. Both types of polynucleotides
exhibit a wide range of structural topologies, among which dif-
ferent single stranded (ss-) sequences are quite numerous.
While ss-sequences are a ubiquitous part of the RNA folding
landscape, there are fewer observations of stable ss-DNA
cases, e.g. hairpins [1] and abasic sites [2], to name some of
them. Since ds-DNA is protected from reaction with a number
of chemical and biological nucleases (see reference [3] and the
references therein), some studies have been aimed at exploiting
the vulnerable ss-DNA, e.g. hairpin loop regions [4–6].
Furthermore, many of the antitumour drugs act by forming
abasic sites in DNA [7]. Since DNA lesions and strand breaks
* Corresponding author. Tel.: +385 1 45 71 210; fax: +385 1 46 80 195.
E-mail address: pianta@irb.hr (I. Piantanida).
0223-5234/$ - see front matter © 2006 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2006.05.005

1154
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
combined with molecular modelling and interactions with nu-
cleotides and single stranded uracil polynucleotide. Here pre-
sented in vitro screening of the antiproliferative activity of phe-
nanthridinium–mono- and bis-nucleobase conjugates (2, 3, 5–
8) addressed the impact of tethered nucleobases on the bioac-
tivity of typically DNA intercalative phenanthridinium moiety
(1,4).
2. Chemistry
2.1. Synthesis
Scheme 1. Previously studied protonated (4–8) [9,10] and methylated (9–11)
[11] phenanthridinium–nucleobase conjugates.
The general strategy used for the synthesis of the novel phe-
nanthridine–bis-nucleobase conjugates 1–3 comprises the alky-
lation of the amino substituents of phenanthridine by dibromo-
propane, followed by the introduction of nucleobase at the
other end of alkyl linker. Novel compound 1, as well as pre-
viously known Ade-C3 (9-(n-propyl)adenine) and Ura-C3 (1-
(n-propyl)uracil) [17] were prepared for comparison purposes.
nium–bis-nucleobase conjugates presented in Fig. 1 with DNA
and RNA, among which bis-uracil derivative (2) has shown
strong selectivity toward homo polynucleotides containing
consecutive adenines [12]. Here are presented syntheses of
phenanthridinium–bis-nucleobase conjugates, detailed spectro-
scopic properties in aqueous solution, interactions with nucleo-
tides and poly U and comparable in vitro biological activity on
six human cell lines of all till now prepared phenanthridinium
mono- and bis-nucleobase conjugates (Fig. 1 and Scheme 1).
As outlined in Scheme 2, 3,8-bis-tosylamino-6-methylphe-
nantridine was alkylated in the presence of potassium carbo-
nate in dry DMF due to a low solubility in other solvents,
e.g. acetonitril. Only the addition of excess of mono- or dibro-
moalkane afforded bis-alkylated phenanthridines 12 and 13 in
acceptable yields (30% and 55%, respectively). For preparation
of 12, a large excess of dibromoalkane was essential in order to
prevent formation of undesired alkyl-bis-phenanthridine.
Again, the large excess of adenine or uracil in reaction with
12 was necessary to obtain compounds 14 and 15 in yields of
34 and 84%, respectively. Under these conditions (Scheme 2)
the alkylation of uracil selectively occurred at the N1 position
giving 14 as main product but precise temperature control in
reaction (T = 40–50 °C) should be stressed since at higher tem-
peratures 1,3-N alkylated uracil became dominant while at
room temperature reaction was unreasonably slow (few
weeks). Removal of tosyl from 13–15 was achieved by heating
at 100 °C in acidic conditions (Scheme 2) and giving com-
pounds 1–3 in 40–93% yields. They were found to be suffi-
ciently soluble in water at acidic conditions, which enabled
studies in aqueous media.
Newly synthesised compounds 1–3 also introduce two ma-
jor differences compared to previously studied 4–11: a) an ad-
ditional nucleobase offers a possibility of doubled recognition
between conjugate and complementary polynucleotide [13], b)
two linker–nucleobase units tethered to the opposite sites on
the longer axis of phenanthridinium result in significantly re-
stricted possibility of phenanthridinium intercalation into poly-
nucleotide. Considering the case b), steric restriction in non-
covalent complex formation has already in some cases resulted
in strong ss- over ds-polynucleotide preferential binding [14,
15]. Also, intercalative binding mode significantly differenti-
ates between ss- polynucleotides and ds-polynucleotides by
means of “neighbour exclusion principle” valid only for latter
polynucleotides [16]. Therefore, in this work we present synth-
esis of compounds 1–3, detailed spectroscopic analysis of in-
tramolecular interactions of bis-nucleobase conjugates 2, 3
Fig. 1. General structure of novel phenanthridine–bis-nucleobase conjugates.

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1155
Scheme 2. i) Br(CH2)3Br or Br(CH2)2CH3, K2CO3, dry DMF, Ar, r.t., ii) NaH, dry DMF, Ar, 40–50 °C, iii) a) 2:1 CH3COOH: H2SO4, 80–100 °C, b) NaOH/H2O.
2.2. Spectroscopic properties
tion range of 5 × 10–6–8 × 10–5 mol dm–3. By controlling UV–
vis experiments it was observed that aqueous solutions of 1–3
were stable for a few months. In all, hypochromic effect was
observed at higher concentrations pointing the presence of in-
termolecular interactions.
The UV–vis spectrum of bis-adenine conjugate (3), if com-
pared with the referent 1, revealed a pronounced hypochromic
effect and bathochromic shift at the one of absorption maxi-
mum of phenanthridinium moiety (Table 1, λ = 287–290 and
466–494 nm; Fig. 2). An additional maximum in the UV–vis
spectrum of 3 at 267 nm can be attributed to covalently linked
2.2.1. UV–vis and fluorescence spectra
UV–vis spectra of aqueous solutions of 1–3 are strongly pH
dependent, exhibiting a one step change at pKa ≈ 6, attributed
to protonation of phenanthridine nitrogen [9]. Due to the low
solubility of phenanthridine at pH 7, all further experiments in
aqueous media were done at pH = 5, all of compounds being in
a protonated (phenanthridinium) form. The UV–vis spectra of
compounds 1–3 (at pH = 5, sodium citrate–HCl buffer,
I = 0.027 mol dm–3) obey Lambert–Beer law in the concentra-

1156
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
Table 1
a
Electronic absorption data of 1, 2, 3 and reference compounds
λmax / nm (ε / cm2 mmol–1
49500 (287, λmax); 6000 (466, λmax
58501 (289, λmax); 57949 (287 nm); 6900 (485, λmax
25011 (267 nm, λmax); 32810 (290 nm, λmax); 31995 (287 nm); 4500 ≈ 38% hypochromicity (287 nm)
(494 nm, λmax
c
)
Hypochromicity (Phen)
1
2
3
)
)
≈ 8% hyperchromicity (287 nm)
)
b
Ura–C3
Ade–C3
a
9280 (261 nm, λmax) 2104 (287 nm)
12871 (261 nm, λmax) 923 (287 nm)
b
Sodium citrate–HCl buffer, I = 0.027 mol dm–3, pH = 5.
b
c
Sodium cacodilate–HCl buffer, I = 0.05 mol dm–3, pH = 5 [9].
{[ε287 nm (1) + 2 × ε287 nm (Ura–C3 or Ade–C3)–ε287 nm (2 or 3)]/[ε287 nm (1) + 2 × ε287 nm (Ura–C3 or Ade–C3)]} × 100.
Fig. 2. UV–vis spectra of 1, 2, 3 (at pH = 5, sodium citrate–HCl buffer, I = 0.027 mol dm–3).
adenines since it agrees well with the maximum of the referent
adenine analogue Ade-C3 [9] (Table 1, λ = 262 nm). Under
assumption of ideal additivity of chromophores in 3 (phenan-
thridinium, two adenines), the calculated hypochromic effect at
287 nm (38%, Table 1) is quite significant. All effects ob-
served for 3 point toward intramolecular a stacking interactions
between phenanthridinium and covalently linked adenines.
294 nm). A further temperature increase (Fig. 3b) yielded
roughly proportional increase of the complete spectrum. Upon
cooling to 25 °C the starting spectrum was not completely re-
covered. In the same temperature range (25–90 °C) UV–vis
spectra of 1 and 2 do not change.
The increase of absorbance at λ = 267 nm in the spectrum of
3 upon temperature raise is consistent with a breaking (disso-
ciation) of aromatic stacking interactions of adenines and phe-
nanthridinium. However, a peculiar decrease of absorbance at
about 290 nm in 25–60 °C temperature range and bathochro-
mic shift suggested even more efficient aromatic stacking inter-
actions of phenanthridinium. Since, according to the isosbestic
point at λ = 270 nm, only two species are in equilibrium it
seems reasonable to propose unstacking of one adenine (absor-
bance increase at 267 nm) followed by formation of a second
adenine–phenanthridinium complex characterised by its UV–
vis spectrum with even more pronounced hypo- and bathochro-
mic effects than observed for two adenines–phenanthridinium
complex (dominant at 25 °C). A proportional increase in the
complete spectrum of 3 at temperatures higher than 60 °C
agrees well with the dissociation of second adenine–phenan-
thridinium complex. All afore mentioned suggested that at
25 °C both adenines are stacked on the same side of phenan-
thridinium moiety. Such a structure could explain dissociation
UV–vis spectrum of bis-uracil conjugate 2 does not reveal
any obvious hypochromic effect if compared with that of 1 but
clearly shows bathochromic shift of phenanthridinium maxi-
mum (Table 1, λ = from 287 to 289 and λ = from 466 to
485 nm; Fig. 3). Even under presumption of ideal additivity
of chromophores in 2 (phenanthridinium, two uracils), hypo-
chromic effect was not evident (Table 1). A new shoulder at
about 270 nm in the spectrum of 2 corresponds well to the
maximum of the referent uracil analogue Ura-C3 [9] (Table 1,
λ = 268 nm) and therefore can be attributed to covalently
linked uracils.
UV–vis spectrum of aqueous solution of 3 has shown strong
temperature dependence (Fig. 3). In the temperature range be-
tween 25 and 60 °C (Fig. 3a) the absorbance maximum at
λ = 267 nm (attributed to adenines of 3) was increasing and
the maximum attributed to the phenanthridinium of 3 was de-
creasing and shifting to longer wavelengths (λ = from 290 to

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1157
Fig. 3. Temperature dependent changes of UV–vis spectrum of 3 (c (3) = 3.8 × 10–5 mol dm–3 at pH = 5, sodium citrate–HCl buffer, I = 0.027 mol dm–3): a) 25, 40,
50, 60 °C; b) 60, 70, 80, 90 °C.
of one adenine in the range 25–60 °C followed by a more effi-
cient stacking of the remaining adenine thus causing effects
presented on Fig. 3a.
All studied compounds (1–3) in aqueous solutions exhibit
fluorescence emission linearly dependent on the compound
concentration up to c = 7.0 × 10–6 mol dm–3. Excitation spectra
monitored at emission maxima of all compounds agree well
with the UV–vis spectra. Significantly higher fluorescence
emissions of compounds 2 and 3 compared with the reference
compound 1 (Table 2) support intramolecular aromatic stack-
ing interactions between phenanthridinium and linked nucleo-
bases. Quenching of 2 and 3 fluorescence upon temperature
increase, while virtually no change in the emission spectra of
1 being observed, was consistent with the unstacking of intra-
molecular complexes of 2 and 3 [9,11].
Hypochromic shift of 2 and 3 emission maximum compared
to reference 1 (Table 2) agrees well with the effect observed for
most aromatic compounds upon intercalation into DNA [18],
pointing also toward intramolecular stacking interactions.
Table 2
Fluorescence emission and excitation properties of 1–3 in aqueous solutions (at
2.2.2. NMR experiments in aqueous solutions
pH = 5, sodium citrate–HCl buffer, I = 0.027 mol dm–3
)
1H NMR spectra of compounds 1, 2 and 3 were recorded at
600 MHz in water solutions (sodium cacodylate in D2O–DCl
buffer, I = 0.005 mol dm–3, pD = 5, water signal suppressed
using pre-saturation technique). Chemical shifts (δ/ppm) of
aromatic protons of 1 and 2 were found to be concentration
a
λexc (nm)
285
λem (nm)
585
INTcompd/INT5
1
1
2
3
290
290
580
580
1.5
2.7
a
λex = 290 nm.

1158
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
dependent (c = 1.0 × 10–4–1.0 × 10–3 mol dm–3) due to self-
stacking interactions (estimated Ksa of 102 mol–1 dm3) [9].
Proton chemical shifts δ0 of 1 and 2 were obtained by ex-
trapolation of dependence of δ versus concentration to infinite
dilution, where no intermolecular self-association was present
[19]. Chemical shifts δ0 of aromatic protons of 2 were strongly
shifted upfield (0.1–0.3 ppm) relative to 1 signals, due to in-
tramolecular aromatic π…π interactions between phenanthridi-
nium and covalently attached uracil moieties.
Table 3
Results of the semiempirical, AM1 calculations (in vacuum) and of the force
field calculations (AMBER, with the effect of the bulky water simulated by the
distance dependent dielectric constant)
Compound
2
AM1
AMBER
Conformation
E
S
G
ΔE (kcal mol–1
)
ΔE (kcal mol–1
)
10.6
3.6
12.2
3.4
2.5
0.0
1.4
0.0
G0
3
Conformation
ΔE (kcal mol–1
)
ΔE (kcal mol–1
)
Interestingly, using the same experimental conditions, aqu-
eous solution of 3 didn’t show any signals in H NMR spec-
E
S
G0
4.1
0.0
0.4
15.9
0.0
0.6
1
trum. Proton signals could be observed only at probe tempera-
tures higher then 50 °C. Adding DMSO in water solution of 3
caused also appearance of proton signals (volume portion of
DMSO more than 30%; c (3) > 1 × 10–4 mol dm–3). This be-
haviour can be explained by formation of large aggregates of 3
molecules in water solution, even at unexpectedly low concen-
tration.
different sides of the plane) and G-shape (with nucleobases in
stacking interaction with phenanthridinium from the same side
of the plane) conformations using the program InsightII [20]
(Fig. 4).
2.3.1. Semiempirical MO and MM calculations for 2 and 3
Quantum mechanical, AM1 calculations were performed in
order to determine partial atom charges (to be used for empiri-
cal calculations), and relative potential energies of different
conformations of 2 and 3 in vacuum.
2.3. Molecular modelling, molecular mechanics, molecular
dynamics and semiempirical study performed for 2 and 3
Molecules 2 and 3 were built in extended (E) and folded S-
shape (with the nucleobases stacking phenanthridinium from
Fig. 4. Optimised structures of different conformations (E, extended, S, with nucleobases stacked on different sides of phenanthridinium plane, G, with nucleobases
stacked on the same side of phenanthridinium plane) of compounds 2 (Fig. 4a) and 3 (Fig. 4b).

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1159
Molecular modelling of the possible conformations of com-
pounds 2 and 3 in water was accomplished by molecular me-
chanics (MM) calculations and molecular dynamics (MD) si-
mulations performed with the AMBER force field and the
partial charges determined by the semiempirical (AM1) calcu-
lations. Two types of calculations were accomplished: a) with
the effect of the bulky water simulated by the distance depen-
dent dielectric constant (ε = 80r) and, b) using the explicit
water molecules. According to the molecular modelling results,
both semiempirical and MM, for 2 and 3, folded G- and S-
shape conformations, are more favourable than the extended
one (Table 3). The lowest energy conformation G0 of 2 has
three intramolecular H-bonds: two O····H–N between uracils
and the O····H–N bond between the uracil and phenanthridi-
nium.
The screening of electrostatic interactions slightly changed
the relative stability of conformers in such a way that the stack-
ing between nucleobases and phenanthridinium became more
pronounced (Fig. 4a). The energy difference between the
folded, G-shape and S-shape, and the extended E conformation
increased. This is specially pronounced in the case of bis-ade-
nine compound 3 with the S-shape conformation being the
most stable. However, the potential energy difference between
the S- and G-shape conformers of 3 is small and we expect to
have both of them similarly populated in the solution. MD si-
mulations revealed different behaviour of 2 and 3.
close to the G-shape, were the most populated. The extended
conformation changed to the folded one after only a few tenths
of ps. Firstly, folded one of the side chains, and then the other,
both on the same side of the phenanthridinium plane. After
about 200 ps the molecule folded to so called G-shape confor-
mation and remained as it, with some minor fluctuations, until
the end of simulation (Fig. 5), total time of the simulations
varied from 0.5–1.5 ns. The S-shape conformation also chan-
ged to the G-shape one during the MD simulation and re-
mained in it most of the time. During the MD simulation with
the G-shape conformation as an initial one the changes were
less significant although transitions to the semi E- and S-shape
conformations also occurred.
As a measure of strength of the stacking interactions we
considered distance between the centre of the uracil ring and
middle of the closer phenyl of the phenanthridinium unit (d
(X1,XL) and d(X2,XD), see Fig. 5). Percentages of conforma-
tions sampled during the MD simulation (every 100 fs) that
have these distances below 3.5, 4.0, and 4.5 Å are given in
Table 4. In the same table given is the distance between two
Table 4
Percentages of distances between the uracil rings and between each uracil and
the phenanthridinium unit plane (Fig. 5) sampled during four MD simulations
(0.5 ns each). The initial conformation for each simulation is given in the first
column on the left
Initial
% d(X1,X2)
% d(X1,XL)
% d(X2,XD)
conformation > 10 Å, 13 Å, 17 Å < 3.5 Å, 4.0 Å,
4.5 Å
< 3.5 Å, 4.0Å, 4.5 Å
2.3.2. MD simulations of 2
E
E*
S
G
a
85, 28, 3
3, 0, 0
13, 1, 0
21, 3, 0
18, 58, 70
29, 81, 96
11, 53, 80
23, 72, 88
1, 23, 62
MD simulation of 2 at room temperature (298 K, the effect
of the bulky water simulated with the distance dependent di-
electric constant) revealed transitions between different confor-
mations. Independently of a starting conformation, (S, G0 and
E were used as initial conformations) the folded conformations,
a
16, 68, 89
26, 80, 96
23, 72, 89
E* is continuation of E simulation. Both simulations were running for
0.5 ns.
Fig. 5. Conformational changes (distances between aromatic rings) that occurred during the simulation and the final, geometry optimised conformation of 2.
Hydrogens are omitted for simplicity.

1160
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
Fig. 6. Conformational changes (distances between adenine rings) that occurred during MD simulation starting from E conformation (Fig. 6a) and S conformation
(Fig. 6b) of 3.
uracil centres, d(X1,X2). In the folded, G-shape and S-shape
conformations it is slightly below 7 and 9 Å, respectively. A
value of this distance between 10 and 13 Å corresponds to
partly folded conformation, transition from E to either S or
G, with only one nucleobase stacked with phenanthridinium.
From the results presented in Table 4 it is apparent that 2 is
in solution mostly folded (less than 1% of molecules are in
extended form when results of MD simulations started from
different conformations are combined).
No one of MD simulations with explicit water molecules,
starting from different initial conformations, has revealed any
conformational transition during the first 200 ps at room tem-
perature. Increasing temperature to 350 K revealed a conforma-
tional change from the extended to the folded conformation,
after about 400 ps. At the end of 0.5 ns of MD simulation at
the increased temperature the molecule is in ‘half S’ conforma-
tion (Fig. 7).
Since the molecular dynamic simulations in water were
rather short we suspect that influence of the water molecules
on solute was not properly investigated. In order to estimate
changes in hydrophobic to hydrophilic ratio related to the con-
formational changes we determined the ratio between polar and
non-polar solvent accessible surface area for each conformation
of bis-nucleobase conjugates of phenanthridinium (data not gi-
ven). The ratio is smallest for the extended, and the highest for
the folded, S-shape, conformation. This means that ratio be-
tween the stabilizing, polar solute - solvent interactions, and
the negative influence of water due to its contact with non-po-
lar parts of the molecule is, in agreement to the results of mo-
lecular mechanics and dynamics calculations, the highest in the
folded, S conformation, the most favourite conformation of 3.
According to the results of MD we can conclude that influ-
ence of water and its possibility to form the hydrogen bond
with N1 hydrogen of phenanthridinium favours stacking inter-
actions.
2.3.3. MD simulations of 3
The course of MD simulation for 3 much more depends of
the starting conformation than it is observed for 2. MD simula-
tions with the effect of the bulky water simulated by the dis-
tance dependent dielectric constant (ε = 80r) at room tempera-
ture (298 K) revealed conformational transitions only in the
case of E conformation as the starting one (Fig. 6a). In this
case, the conformational transition to the folded, G-shape con-
formation occurred after about 50 ps of simulation at room
temperature (Fig. 6a). In the case of S conformation as the
starting one, no conformational transition was recorded during
the 0.5 ns of simulation at room temperature (Fig. 6b). The
mean distance between the geometrical centres of adenines is
7.2 Å in the first case, and 8.4 Å in the second case (the first
50 ps were not used for averaging).
2.4. Interactions of 1–3 with nucleotides
Addition of nucleotides to aqueous solutions of 1–3 didn’t
induce any significant changes in their UV–vis spectra, quite
similar as noted for previously studied phenanthridinium–nu-
cleobase conjugates [9,11]. Nevertheless, titrations of 1–3 with
nucleotides resulted with changes in their fluorescence spectra,
quantity of emission change being strongly dependent on a nu-
cleobase covalently attached to phenanthridinium and also on a
type of nucleotide added (Table 5). To obtain the binding con-
stants (Ks), fluorescence titration data were processed by
SPECFIT giving in general the best fit for the 1:1 stoichiome-
try of complexes (Table 5).
Calculated binding constants (Ks) were of the same order of
magnitude as those found for ethidium bromide [21,22] and
other phenanthridinium–nucleobase conjugates [9,11]. By
comparing the fluorescence changes (Table 3, ΔINT %) in
most cases the tendency ΔINT % (3) < ΔINT % (2) < ΔINT
Fig. 7. Conformation of 3 after the MD simulation with explicit water
molecules (omitted for clarity) and the mild geometry optimization.

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1161
Table 5
Table 6
a,b
Binding constants (logKs) for various 1–3/nucleotide complexes
Growth inhibitory effects of different phenanthridinium–nucleobase conjugates
2, 3 and 5–8 and referent compounds 1 and 4 on the growth of malignant
tumour cell lines in comparison with their effects on the growth of normal
human diploid fibroblast (WI 38)
AMP
97
1.63
70
1.72
25
1.2
GMP
68
1.23
49
1.02
17
1.27
UMP
13
< 1
CMP
–10
1.2
–4
1.38
–7
c
c
c
1
2
3
ΔINT
log Ks
ΔINT
log Ks
ΔINT
log Ks
%
%
%
d
a
IC50 (μM)
5.5
< 1
d
HeLa
MCF-7
0.4
> 100
MiaPaCa-2 Hep-2
10
SW 620 WI 38
6
< 1
1
2
3
4
5
6
7
8
4
0
4
5
7
2
4.4
> 100
> 100
39 45
38 13
≥ 100
5
14 14
> 100
> 100
≥ 100
> 100
> 100
d
1.2
> 100
> 100
43 18
≥ 100
> 100
> 100
58.5
66 25
50
51 14
> 100
> 100
77 40
47 38
42 12
Fluorimetric titrations were performed at pH = 5 (I = 0.027 mol dm–3, so-
a
> 100
49 19
44 38
> 100
dium citrate–HCl buffer).
AMP2– = adenosine monophosphate; GMP2– = guanosine monophosphate;
b
43
3
9
UMP2– = uridine monophosphate.
> 100
15 10
84 20
c
41
1
1
46
6
61 33
62 24
88
≥ 100
8
ΔINT % = [(INT0–INT)/INT0] × 100; INT0, relative fluorescence intensity
of pure compound, INT, relative fluorescence intensity calculated for 100% of
complex formed.
59 18
57 32
a
IC50; the concentration that causes a 50% reduction of the cell growth.
d
Less than 50% of complexation was reached allowing only estimation of
cleotide or nucleotide. Another interesting point is that affinity
of 1 toward poly U (logKs = 4.2, n = 0.5) is one order of mag-
nitude higher than found for EB under same experimental con-
ditions (logKs ≈ 3) [26]. Opposite spectroscopic changes found
for 2 and 3 upon titration with poly U point toward existence
of at least two different complexes, thus allowing only estima-
tion of stability constants (logKs < 3) and ratios n for condi-
tions of high excess of polynucleotide (fluorescence increase).
Neither 2 nor 3 (complementary to poly U) didn’t show any
selectivity toward poly U, pointing toward minor (if any) im-
pact of interactions between nucleobase attached to phenanthri-
dinium with uracil of poly U.
binding constant.
% (1) is evident. Since intramolecular stacking interactions
have a stronger impact on fluorescence properties of 3 than
on 2 it is reasonable to expect that addition of nucleotide will
have less influence on the emission of the former. Same expla-
nation agrees well with highest ΔINT % values found for 1
where no intramolecular stacking interactions can exist.
Fluorescence increase of 1–3 solutions upon addition of
AMP, GMP and UMP can be explained by mechanism pro-
posed for ethidium bromide and its derivatives having primary
or secondary amino groups at 3,8 positions of the phenanthri-
dinium ring [23]. However, unexpected quenching of fluores-
cence was observed for all studied compounds upon addition
of CMP. Up till now, no ethidium bromide derivative is known
to exhibit such selectivity for any nucleotide. Actually, small
molecules that recognise any nucleotide by opposite spectro-
scopic changes upon formation of non-covalent complex are
extremely rare (e.g. proflavine recognition of GMP [24], bis-
acridine differentiation between purine and pyrimidine nucleo-
tides [25] and none has shown recognition of only CMP com-
pared to other nucleotides. Therefore, it is not possible to pro-
pose explanation of observed quenching of 1–3 solutions upon
addition of CMP without detailed photophysical studies which
is out of the scope of this publication.
3. Pharmacology
3.1. Antiproliferative effect of compounds in vitro
The compounds 1–3 and 4–8 were tested for the potential
antiproliferative effect on the panel of six human cell lines, five
of which are derived from five cancer types: HeLa (cervical
carcinoma), MCF–7 (breast carcinoma), SW 620 (colon carci-
noma), MiaPaCa–2 (pancreatic carcinoma), Hep–2 (laryngeal
carcinoma) and WI 38 (diploid fibroblasts). The compounds
showed different antiproliferative effect on the presented panel
cell lines (Table 6). The compounds 2 and 3 produced little or
no growth inhibition. The compounds 4–8 showed accentuated
inhibitory effect on all cell lines, but mostly at the highest
tested concentration. Among these compounds 5 and 6 were
the most selective ones, since they did not inhibit a growth of
normal fibroblasts. On the other side, the compound 1 was the
most active one. Besides, the compound 1 was cytotoxic to all
cell lines. It should be also taken into consideration that 3, 4, 6
and 8 formed colloid DMSO stock solutions used in experi-
ments.
2.5. Interactions of 1–3 with polynucleotides
Although 1–3 didn’t show any recognition of complemen-
tary nucleotides, 2 showed selective interactions within more
lipophilic microenvironment of polynucleotides characterised
by consecutive adenines [12]. Also, in our previous research
analogue of 3 (mono–adenine conjugate) [10] has shown re-
cognition of complementary poly U. Therefore, we have per-
formed detailed studies of 1–3 interactions with single stranded
(ss-) polynucleotide poly U. Contrary to previously studied ds-
polynucleotides [12], addition of ss-polynucleotide poly U to 1
resulted in fluorescence quenching, while emission of 2 and 3
was first quenched up to the ratio r(2) = 0.05, r(3) = 0.15 and
than strongly enhanced. The observed quenching of fluores-
cence was quite unexpected (especially for referent 1) since
fluorescence of EB was increased by addition of any polynu-
4. Conclusions
Spectroscopic methods (UV–vis, fluorescence, NMR)
clearly point to the intramolecularly stacked conformation of
adenine conjugate 3. Thermal UV–vis experiments of 3 reveal
unstacking of one adenine at up to 60 °C and second adenine at
much higher temperatures (Fig. 8). These observations are ad-

1162
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
Fig. 8. Schematic representation of possible temperature dependent changes of intramolecular conformation of 3.
ditionally confirmed by molecular modelling experiments,
which showed that the S conformation (with adenines stacked
on different sides of the phenanthridinium plane) is the most
stable one. However, the potential energy difference between
the G-shape and the S-conformers is low (about 0.6 kcal mol–1,
at the room temperature) and their population in water solution
is similar, according to MM and MD results.
actually close structural analogue of ethidium bromide (EB).
This structure proximity of 1 and EB was also reflected in poor
selectivity toward cancer cell lines compared to a normal hu-
man fibroblast line. One order of magnitude lower activity of 4
pointed out that the absence of one amino group (compared to
1) had a significant negative impact. Introduction of one nu-
cleobase on to phenanthridinium system (5–8) yielded compar-
able activity toward cell lines as observed for 4 but in some
cases had a significant impact on selectivity toward cancer
cells compared to a normal human fibroblast. This selectivity
was more pronounced for compounds with shorter linker be-
tween phenanthridinium and nucleobase (5 and 6). On the
other hand, compound 7 having more flexible pentamethylene
chain between phenanthridinium and uracil showed 2–3 times
higher inhibitory activity against HeLa cell lines than its ana-
logue 5 with shorter linker. A simple explanation and correla-
tion of biological data with structure due to mode of activity is
not possible owing to complexity of biological system. How-
ever, possible argument of observed selectivity could be steric
hindrance of less flexible compound 5 being predominantly in
folded conformation in water, which could partially hamper
contact with the cell DNA. Bis-nucleobase conjugates 2 and
3 were found to be inactive, very likely due to a poor solubility
in water under experimental conditions used.
None of the studied compounds is an intriguing candidate
for the anticancer therapy. Nevertheless, systematic variation a)
of number of substituents (one or two amino groups), b) varia-
tion of binding-active substituents (uracil, adenine) and c) flex-
ibility of phenanthridinium–nucleobase system (linker length),
has pointed out to measurable structure–biological activity cor-
relation. Comparatively small structural changes in the simplest
possible model of intercalator–nucleobase system have shown
considerable impact not only on interactions with DNA and
RNA but also on antiproliferative activity toward different cell
lines and in some cases actually generated selectivity toward
cancer cells compared with normal ones. This observation
could be the starting point to the future research for more se-
lective antitumour agents. Due to their selectivity toward spe-
cific polynucleotide sequences, as well as some selectivity
against cancer cells compared to normal human fibroblast,
compounds 5–8 are highly interesting candidates for examina-
Differently, the extended conformation has significantly
higher potential energy. The NMR and fluorescence spectra
of 2 also revealed the folded conformation of molecule char-
acterised by aromatic stacking interactions between phenanthri-
dinium and covalently linked uracils although the effects are
less pronounced than for 3. Smaller NMR and fluorescence
effects of 2 and negligible differences in UV–vis spectrum
(compared to referent 1) suggested much weaker intramolecu-
lar interactions between phenanthridinium and uracils than ob-
served for 3. The molecular modelling revealed frequent tran-
sition between different conformations, with the population of
fully extended being less than 1%. The G-shape conformation,
additionally stabilised by hydrogen bonds between uracil rings,
is according to the results of molecular modelling the most
stable one, but the energy difference between this and extended
conformation is lower than in the case of 3. Weaker intramo-
lecular stacking interactions in 2 than in 3, same as observed
for previously examined compounds 5–8, 10–11 [9,11], also
agreed well with the finding that large condensed aromatics
in water bound more strongly purine than pyrimidine nucleo-
sides and nucleotides [21].
The lack of recognition of the complementary nucleotides
by 2 and 3 can be explained by strong competition of bulk
water with possible hydrogen bonds between nucleobases.
More intriguing is the observation that bis-adenine conjugate
3 showed the affinity toward poly U comparable to one ob-
served for bis-uracil conjugate 2, while in earlier studies we
have observed that mono-adenine conjugate 6 formed signifi-
cantly more stable complex with poly U than its uracil analo-
gue 5 [10]. The steric hindrance of two adenines tethered to
phenanthridinium and strong intramolecular interactions of 3
are the most probable causes of the failing selectivity.
Considering the in vitro biological activity assay, the most
active of all studied compounds is the referent derivative 1,

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1163
tion of synergic effects in combination with currently available
alkylating drugs that can produce multiple abasic lesions in the
cell [8c]. Abasic site can be lethal if not repaired [27], thus
compounds that are able to bind to specific abasic site can dis-
semble the lesion to the AP-endonuclease and therefore inhibit
a repair process in damaged cell, so they have great potential as
enhancers for anticancer therapy. Furthermore, future research
in the field of model intercalator–nucleobase systems targeting
specific polynucleotide sequences will be focused on the place-
ment of recognition unit (nucleobase) into hydrophobic pocket
between intercalator units.
nucleotides–polynucleotides and changes in emission at maxi-
ma (Table 1) were monitored. The binding constants and stoi-
chiometries of complexes of 1–3 with the nucleotides were
calculated for the concentration range corresponding to ca.
20–80% complexation by non-linear least-square fitting pro-
gram SPECFIT [29].
5.3. Synthesis
5.3.1. 3,8-Bis-(3-bromopropyltosyl)amino-6-
methylphenantridine (12)
The 1,3-dibromopropane (5.7 ml, 56 mmol) and K2CO3
(5.2 g, 38 mmol), were suspended in dry DMF (40 ml). To this
suspension, solution of 3,8-bis-tosylamino-6-methylphenantri-
dine (1 g, 1.88 mmol) in dry DMF (80 ml) was added drop-
wise during 30 min. and the reaction mixture was stirred dur-
ing 48 hours under argon atmosphere at room temperature.
Then, water and CH2Cl2 were added to this suspension. Water
layer was washed twice with CH2Cl2, organic extracts were
dried over Na2SO4 and evaporating. Water was added to brown
oil to give light brown precipitate. Pure compound 12 was ob-
tained by TLC (SiO2, 2% MeOH in CH2Cl2, Rf = 0.46) as col-
ourless oil (450 mg, 31%), that was crystallised from MeOH,
5. Experimental
5.1. General procedures
1H NMR spectra were recorded on a Varian–Gemini 300
operating at 300 MHz, as well as on Bruker Avance DRX
1
500 operating at 500 MHz for H. Chemical shifts (δ) are ex-
pressed in ppm, and J values in Hz. Signal multiplicities are
denoted as s (singlet), d (doublet), t (triplet), q (quartet) and m
(multiplet). The electronic absorption spectra were measured
on a Varian Cary 100 Bio spectrometer. IR spectra were re-
corded on a Perkin–Elmer 297 instrument using KBr pellets.
Fluorescence spectra were recorded on Perkin–Elmer LS 50
fluorimeter. Mass spectra were obtained using Extrel 2001
DD spectrometer. Preparative thin layer chromatography
(TLC) was carried out using Kieselgel HF254 “Merck”. Melting
points were determined on Kofler apparatus and are uncor-
rected. 3,8-Bis-tosylamino-6-methylphenantridine was pre-
pared starting from 4′-nitro-2-aminobiphenil, as previously de-
1
white crystals: m.p. 157–160 °C; H NMR (CDCl3) δ: 2.02–
2.14 (m, 2 × CH2, 4 H), 2.42 (s, Ts–CH3, 3 H), 2.45 (s,
Ts–CH3, 3 H), 2.89 (s, Phen–CH3, 3 H), 3.46 (m, 2 × CH2Br,
4 H), 3.83 (m, 2 × NCH2, 4 H), 7.22–7.48 (m, 2 × Ts, 8H),
7.53 (d.d., Phen–H9, 1 H, J1 = 8.72 Hz, J2 = 2.05 Hz), 7.59–
7.63 (m, Phen–H2, Phen–H4, 2 H), 7.88 (d, Phen–H7, 1 H,
J = 1.79 Hz), 8.48 (d, Phen–H1, 1 H, J = 8.46 Hz), 8.55 (d,
Phen–H10, 1 H, J = 8.71 Hz) ppm; 13C NMR (CDCl3) 21.34,
23.01, 29.64, 31.32, 31.5, 48.76, 49.05, 122.49, 122.97,
123.77, 126.32, 126.45, 127.03, 127.66, 127.77, 128.25,
129.61, 130.69, 131.38, 134.38, 138.56, 139.92, 143.87,
144.1, 159.43 ppm; IR (KBr) n: 2940, 2860, 1600, 1580,
1540, 1350, 1230, 1185, 1160, 1090, 945, 830, 810, 730, 710,
670 cm–1. Anal. Calcd for C34H35N3O4S2Br2 (Mr = 773.60): C
52.79, H 4.56, N 5.43%; Found: C 52.74, H 4.69, N 5.38%.
1
scribed [28] For all products, purity was checked by H NMR
and for most of them correct elemental analyses were obtained.
Only for 2 hygroscopic character of precipitate yielded elemen-
tal analyses with non-stoichiometric amounts of water; how-
ever, since NMR spectra of this compound were correct and
2 was obtained by simple chemical reactions form well char-
acterised starting compounds, its structure is not questionable.
5.2. UV–vis and fluorescence measurements
5.3.2. 3,8-Bis-(n-propyltosyl)amino-6-methylphenantridine
(13)
Compound 13 was obtained as described for 12; K2CO3
(1.17 g, 8.47 mmol), 1–bromopropane (1 ml, 11.3 mmol) and
3,8-tosylamino-6-methylphenantridine (300 mg, 0.56 mmol) in
dry DMF (3 + 5 ml) gave light brown oil, that was purified by
TLC (SiO2, 2% MeOH in CH2Cl2, Rf = 0.44) to afford colour-
less oil (192 mg, 55%), that was crystallised from
MeOH/CH2Cl2, white powder: m.p. 79–83 °C; 1H NMR
(CDCl3) δ: 0.93 (m, 2 × CH3, 6 H), 1.49 (m, 2 × CH2, 4 H),
2.41 (s, Ts–CH3, 3 H), 2.44 (s, Ts–CH3, 3 H), 2.87 (s,
Phen–CH3, 3 H), 3.64 (m, 2 × NCH2, 4 H), 7.21–7.51 (m,
2 × Ts, Phen–H2, 9H), 7.59 (d, Phen–H4, 1 H, J2–4
= 1.93 Hz), 7.63 (dd, Phen–H9, 1 H, J7–9 = 2.2 Hz, J9–10
= 8.79 Hz), 7.87, (d, Phen–H7, 1 Hz), 8.46 (d, Phen–H10, 1
H), 8.53 (d, Phen–H1, 1 H, J1–2 = 8.82 Hz) ppm; 13C NMR
(CDCl3) 9.41, 9.64, 9.71, 39.9, 40.27, 110.42, 110.82,
Nucleotides and polynucleotides were purchased from Sig-
ma and Aldrich, and used without further purification. Polynu-
cleotide was dissolved in the sodium cacodylate buffer,
0.05 mol dm–3, pH = 7 and its concentration determined spec-
troscopically as the concentration of phosphates. The measure-
ments were performed in aqueous buffer solution (pH = 5,
I = 0.027 mol dm–3, sodium citrate–HCl buffer). Under the ex-
perimental conditions used the absorbance and fluorescence in-
tensities of 1–3 were proportional to their concentrations. Spec-
troscopic titrations were performed at constant ionic strength
(buffer, I = 0.027 mol dm–3) by adding portions of nucleotide
or polynucleotide solution into solution of the tested com-
pound. Obtained data were corrected for dilution. In fluori-
metric titrations excitation wavelengths at λmax > 440 nm were
used in order to avoid absorption of excitation light by added

1164
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
111.69, 114.97, 115.64, 115.79, 117.6, 122.87 ppm; IR (KBr)
n: 2960, 2940, 2880, 1650, 1610, 1600, 1580, 1570, 1480,
1350, 1210, 1170, 1070, 1030, 1010, 960, 890, 840, 810,
740, 730, 710, 670, 665, 640 cm–1; Anal. Calcd for
C34H37N3O4S2 (Mr = 615.79): C 66.31, H 6.06, N 6.82%;
Found: C 66.18, H 5.87, N 7.07%.
Phen–H9, 1 H, J9–10 = 8.40 Hz), 7.71 (s, Phen–H4, 1 H),
7.82 (s, Phen–H7, 1 H), 8.1 (m, 2 × Ade–H2, 2 × Ade–H8, 4
H), 8.71 (d, Phen–H1, 1 H, J1–2 = 9.02 Hz), 8.81 (d, Phen–
H10, 1 H) ppm; 13C NMR (DMSO-d6) δ: 21.08 (Ts–CH3),
22.82 (Phen–CH3), 27.99 (CH2), 47.64, 118.99, 122.95,
123.12, 124.22, 125.73, 126.45, 127.04, 127.81, 129.22,
129.44, 130.02, 131.25, 131.48, 134.33, 137.86, 141.18,
143.65, 144.15, 149.74, 152.6, 156.24, 158.48 ppm; IR (KBr)
n: 3470, 3300, 3100, 2940, 1650, 1600, 1480, 1420, 1375,
1350, 1310, 1240, 1220, 1160, 1110, 1090, 1070, 1010, 950,
820, 800, 780, 730, 710, 670, 650 cm–1; Anal. Calcd for
C44H43N13O4S4 (Mr = 881.30): C 59.91, H 4.92, N 20.66%;
Found: C 59.74, H 5.17, N 20.57%.
5.3.3. 3,8-Bis-[3-(urac-1-il)propyltosyl)]amino-6-
methylphenantridine (14)
Uracil (217 mg, 1.94 mmol) that was previously dried, and
NaH (77 mg, 60% w.w., 1.94 mmol), were suspended in dry
DMF (5 ml) and stirred during 1 h in argon atmosphere at
room temperature. To this suspension, solution of 12
(100 mg, 0.13 mmol) in dry DMF (10 ml) was added drop-
wise and the reaction mixture was stirred during 48 hours un-
der argon atmosphere at 40–50 °C. Then, water and CH2Cl2
were added to this suspension. Water layer was washed twice
with CH2Cl2, organic extracts were dried over Na2SO4 and
evaporating. Water was added to brown oil to give light grey
precipitate 14 (83 mg, 77%) that was filtered washed with
water, and without further purification used in next step. Pure
compound 14 was obtained by TLC (SiO2, 10% MeOH in
CH2Cl2, Rf = 0.57) as white solid, that was recrystallised from
MeOH and small amount of water; m.p. 152–155 °C; 1H NMR
(DMSO-d6) δ: 1.7 (br, 2 × CH2, 4 H), 2.4 (s, Ts–CH3, 3 H),
2.42 (s, Ts–CH3, 3 H), 2.8 (s, Phen–CH3, 3 H), 3.74 (m, 2 ×
NCH2, 2 × CH2Ura, 8 H), 5.52 (m, 2 × Ura–H5, 2 H), 7.37–
7.59 (m, 2 × Ts, Phen–H2, 9 H), 7.62 (m, 2 × Ura–H6, Phen–
H9, 3 H), 7.71 (d, Phen–H4, 1 H, J2–4 = 1.87 Hz), 7.88 (d,
Phen–H7, 1 H, J7–9 = 1.87 Hz), 8.73 (d, Phen–H1, 1 H, J1–2
= 9.03 Hz), 8.82 (d, Phen–H10, 1 H, J9–10 = 9.02 Hz), 11.21 (s,
2 × Ura–NH, 2 H) ppm; 13C–NMR (DMSO-d6) δ: 21.18,
22.94, 27.35, 28.93, 38.77, 42.18, 49.88, 118.84, 122.87,
122.98, 124.0, 125.6, 125.96, 126.96, 127.66, 129.1, 129.31,
129.88, 131.21, 131.28, 134.47, 138.04, 140.93, 143.46,
143.87, 149.63, 152.44, 156.05, 158.3 ppm; IR (KBr) n :
3040, 2920, 1670, 1470, 1450, 1370, 1335, 1225, 1150,
1075, 940, 805, 760, 720, 700, 660, 645 cm–1; Anal. Calcd for
C42H41N7O8S2 × 3 H2O × CH3OH (Mr = 889.98): C 56.68, H
5.32, N 11.01%; Found: C 56.84, H 5.2, N 10.57%; ES-MS
(m/z) calcd: 836.3 (M+ + 1), 418.6 (M2+ + 2); found: 836.0
(M+ + 1), 418.7 (M2+ + 2).
5.3.5. 3,8-Bis(propyl)amino-6-methylphenantridine (1)
Compound 13 (100 mg, 0.16 mmol) was dissolved in 5 ml
conc. H2SO4 and heated under reflux at 80–100 °C for 2 h.
Reaction mixture was cooled, poured on ice and made alkaline
(pH = 8–9) by addition of 2 M NaOH. Red solid 1 (74 mg,
78%) was precipitated, filtered and washed with a lots of
water, then purified by TLC (SiO2, 10% MeOH in CH2Cl2,
Rf = 0.65) and finally recrystallised from MeOH and small
1
amount of water; m.p. 118–121 °C; H NMR (DMSO-d6) δ:
0.98 (m, 2 × CH3, 6 H)), 1.61 (m, 2 × CH2, 4 H), 2.75 (s,
Phen–CH3, 3 H), 3.07 (m, 2 × NCH2, 4 H), 5.85 (br, NH, 1
H), 5.97 (br, NH, 1 H), 6.82 (s, Phen–H4, 1 H), 6.90–6.93
(m, Phen–H7, Phen–H2, 2 H), 7.17 (d, Phen–H9, 1 H,
J = 8.97 Hz), 8.14 (d, Phen–H1, 1 H, J = 8.97 Hz), 8.24 (d,
Phen–H10, 1 H, J = 8.98 Hz) ppm; 13C NMR (DMSO-d6) δ:
11.9, 11.97, 23.15, 44.88, 103.08, 106.11, 115.92, 120.11,
121.79, 122.3, 123.77, 125.31, 143.5, 146.79, 147.95, 156.82
ppm; IR (KBr) n: 3420, 3280, 2960, 2930, 2880, 1620, 1580,
1510, 1475, 1430, 1390, 1365, 1250, 1220, 1180, 1150, 1075,
1010, 950, 830, 810, 720, 670 cm–1; Anal. Calcd for C20H25N3
(Mr = 307.44): C 78.14, H 8.20, N 13.67%; Found: C 78.41, H
7.98, N 13.56%.
5.3.6. 3,8-Bis-[3-(urac-1-il)propyl)]amino-6-
methylphenantridine (2)
Compound 2 (150 mg, 0.18 mmol) was obtained as de-
scribed for 1; 14 in mixture of 1 ml conc. H2SO4 and 2 ml
conc. acetic acid, gave red solid 2 (74 mg, 78%) that was pur-
ified by TLC (SiO2, 20% MeOH in CH2Cl2, Rf = 0.45) and
finally recrystallised from MeOH and H2O; yellow powder:
5.3.4. 3,8-Bis[3-(aden-9-il)propyltosyl)]amino-6-
methylphenantridine (15)
1
m.p. 129–134 °C; H NMR (DMSO-d6) δ: 1.96 (br, 2 × CH2,
Compound 15 was obtained as described for 14; adenine
(270 mg, 1.94 mmol), NaH (77 mg, 60% w.w., 1.94 mmol)
and 12 (100 mg, 0.13 mmol) in dry DMF (3 + 10 ml) gave
white powder 15 (96 mg, 84%), that was without further pur-
ification used in next step. Pure compound 15 was obtained by
TLC (SiO2, 10% MeOH in CH2Cl2, Rf = 0.26) as white solid,
that was recrystallised from MeOH; m.p. 185–188 °C; 1H
NMR (DMSO-d6) δ: 1.94 (br, 2 × CH2, 4 H), 2.39 (s,
Ts–CH3, 3 H), 2.41 (s, Ts–CH3, 3 H), 2.77 (s, Phen–CH3, 3
H), 3.75 (br, 2 × NCH2, 4 H), 4.24 (m, 2 × CH2Ade, 4 H), 7.3
(s, 2 × NH2, 4 H), 7.37 (m, 2 × Ts, Phen–H2, 9 H), 7.64 (d,
4 H), 2.85 (s, Phen–CH3, 3 H), 3.18 (br, 2 × NCH2, 4 H), 3.83
(t, 2 × CH2Ura, 4 H, J = 6.84 Hz), 5.57 (d, 2 × Ura–H5, 2 H,
J = 7.16 Hz), 6.18 (s, 2 × NH, 2 H), 6.91 (s, Phen–H4, 1 H),
6.99–7.02 (m, Phen–H7, Phen–H2, 2 H), 7.27 (d, Phen–H9, 1
H, J = 9.03Hz), 7.71 (d, 2 × Ura–H6, 2 H, J = 7.78 Hz), 8.26
(d, Phen–H1, 1 H, J = 9.03 Hz), 8.34 (d, Phen–H10, 1 H,
J = 9.03 Hz), 11.29 (s, 2 × Ura–NH, 2 H) ppm; IR (KBr) n:
3350, 2920, 1660, 1610, 1570, 1505, 1450, 1340, 1330,
1280, 1230, 1170, 1150, 1050, 800, 750, 705 cm–1; ES-MS
(m/z) 528.2 (M+ + 1), 264.6 (M2+ + 2); found: 528.1
(M+ + 1), 264.7 (M2+ + 2).

L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
1165
5.3.7. 3,8-Bis[3-(aden-9-il)propyl)]amino-6-
methylphenantridine (3)
Compound 3 was obtained as described for 1; 15 (90 mg,
0.1 mmol) in 1 ml conc. H2SO4 and 3 ml conc. acetic acid
gave yellow powder (50 mg, 84%) 3, that was purified by TLC
(SiO2, 20% MeOH in CH2Cl2, Rf = 0.08) and recrystallised
standard 96-well microtiter plates on day 0. The cell concentra-
tions were adjusted according to the cell population doubling
time (PDT): 1 × 104 per ml for HeLa, Hep-2, MiaPaCa-2 and
SW 620 cell lines (PDT = 20-24 hours) 2 × 104 per ml for
MCF-7 cell lines (PDT = 33 hours) and 3 × 104 per ml for
WI 38 (PDT = 47 hours). Test agents were then added in five,
10-fold dilutions (10–8–10–4 mol dm–3) and incubated for
further 72 hours. Working dilutions were freshly prepared on
the day of testing. The solvent was also tested for eventual
inhibitory activity by adjusting its concentration to be the same
as in working concentrations. After 72 hours of incubation, the
cell growth rate was evaluated by performing the MTT assay
[31], which detects dehydrogenase activity in viable cells. The
absorbency (OD, optical density) was measured on a micro-
plate reader at 570 nm. The percentage of growth (PG) of the
cell lines was calculated according to one or the other of the
following two expressions:
1
from MeOH and small amount of water; m.p. >300 °C; H
NMR (DMSO-d6) 2.28 (br, 2 × CH2,
4 H), 2.84 (s,
Phen–CH3, 3 H), 3.27 (m, 2 × NCH2, 4 H), 4.42 (br, 2 ×
CH2Ade, 4 H, J = 6.84 Hz), 6.12 (br, NH, 1 H), 6.26 (br,
NH, 1 H), 6.97 (s, Phen–H4, 1 H), 7.03–7.07 (m, Phen–H7,
Phen–H2, 2 H), 7.31 (m, Phen–H9, NH2, 3 H), 8.27 and 8.30
(s, 2 × Ade–H2, 2 × Ade–H8, 4 H), 8.43 (m, Phen–H1, Phen–
H10, 2 H) ppm; IR (KBr) n: 3300, 3100, 2910, 2840, 1650,
1600, 1570, 1505, 1460, 1405, 1380, 1330, 1300, 1240, 1190,
1155, 1000, 800, 710, 640 cm–1; Anal. Calcd for C30H31N13
× 2 H2O × CH3OH (Mr = 641.75): C 58.01, H 6.12, N
28.37%; Found: C 57.74, H 5.77, N 28.0%; ES-MS (m/z)
calcd: 574.3 (M+ + 1), 287.6 (M2+ + 2), 192.1 (M2+ + 3);
found: 574.0 (M+ + 1), 287.7 (M2+ + 2), 192.2 (M2+ + 3).
If (mean ODtest – mean ODtzero) ≥ 0 then
PG = 100 × (mean ODtest – mean ODtzero)/(mean ODctrl
meanODtzero).
–
If (mean ODtest – mean ODtzero) < 0 then:
PG = 100 × (mean ODtest – mean ODtzero)/ODtzero
Where:
.
5.4. Molecular modelling, molecular mechanics, molecular
dynamics and semiempirical study
Mean ODtzero = the average of optical density measurements
before exposure of cells to the test compound.
Mean ODtest = the average of optical density measurements
after the desired period of time.
Mean ODctrl = the average of optical density measurements
after the desired period of time with no exposure of cells to the
test compound.
Each test point was performed in quadruplicate three indivi-
dual experiments. The results are expressed as IC50, which is
the concentration necessary for 50% of inhibition. The IC50
values for each compound are calculated from dose–response
curves using linear regression analysis by fitting the test con-
centrations that give PG values above and below the reference
value (i.e. 50%). If however, for a given cell line all of the
tested concentrations produce PGs exceeding the respective re-
ference level of effect (e.g. PG value of 50), then the highest
tested concentration is assigned as the default value, which is
preceded by a “>” sign.
Semiempirical calculations were performed with the pro-
gram MOPAC, using AM1 calculations. The geometry optimi-
sation of a molecule was performed in two steps: a) with the all
heavy atoms fixed, b) with all atoms free to move.
Force field calculations were performed using the all-atom
AMBER force field [20], and the partial atomic charges de-
rived from the electrostatic potential fitted AM1 calculations.
Energy minimization was performed by the steepest descent
and conjugate gradients algorithms. Systems were optimised
up to the convergence of about 0.01 kcal mol–1. The molecular
dynamic simulation was accomplished using the time step of
1 fs and the Varlet integration method. MD simulations with
the explicit water molecules were performed using PBC with
the cubic unit cell dimensions 3 1 × 3 1 × 31 Å. NVT assemble
was used and the cut-off distances of 13 and 15 Å. System was
equilibrated for 3000 steps and than the temperature was
slowly increasing 50 K/1000 steps up to 300 K. At room tem-
perature simulation was performed in duration of about 1 ns.
Acknowledgments
5.5. Antitumour activity assays
Support for this study by the Ministry of Science, Education
and Sport of Croatia is gratefully acknowledged (Projects
0098053, 0119641, 0098093, 0098036)
The growth inhibition activity was assessed according to the
slightly modified procedure performed at the National Cancer
Institute, Developmental Therapeutics Program [30]. The HeLa
(cervical carcinoma), MCF–7 (breast carcinoma), SW 620 (co-
lon carcinoma), MiaPaCa–2 (pancreatic carcinoma), Hep–2
(laryngeal carcinoma) and WI 38 (diploid fibroblasts) cells
were cultured as monolayers and maintained in Dulbecco’s
modified Eagle’s medium (DMEM) supplemented with 10%
fetal bovine serum (FBS), 2 mM L-glutamine, 100 U ml–1 pe-
nicillin and 100 μg ml–1 streptomycin in a humidified atmo-
sphere with 5% CO2 at 37 °C. The cells were inoculated onto
References
[1] S. Yoshizawa, G. Kawai, K.A. Watanabe, K.I. Miura, I. Hirao, Biochem-
istry 36 (1997) 4761–4767.
[2] H. Naegeli, Mechanism of DNA Damage Recognition in Mammalian
Cells, R.G. Landes Company, Austin, Texas, USA, Springer-Verlag,Hei-
delberg, Germany, 1997.
[3] C. Bailly, M.J. Waring, in: K.R. Fox (Ed.), Methods in Molecular Biol-
ogy, Humana Press, Totowa, NJ, 1997, pp. 51–79 (vol. 90).

1166
L.-M. Tumir et al. / European Journal of Medicinal Chemistry 41 (2006) 1153–1166
[4] J.-C. François, N.T. Thuong, C. Hélène, Nucleic Acids Res. 22 (1994)
3943–3950.
[5] P.T. Henderson, B. Armitage, G.B. Schuster, Biochemistry 37 (1998)
2991–3000.
[6] G. Juan, W. Pan, M. Darzynkiewicz, Exp. Cell Res. 227 (1996) 197–
202.
[16] a)D. M. Crothers, Biopolymers, 6 (1968) 575–584.; b) J.L. Botour, E.
Delain, D. Coulaud, J.B. LePecq, J. Barbet, B. P, Roques, Biopolymers
17 (1978) 872–886.
[17] N.J. Leonard, R.F. Lambert, J. Org. Chem. 34 (1969) 3240–3248.
[18] DNA and RNA binders, in: M. Demeunynck, C. Bailly, W.D. Wilson
(Eds.), Wiley-VCH, Weinheim, 2002.
[7] K. Hemminki, D.B.J. Ludlum, Natl. Cancer Inst. 73 (1984) 1021–1028.
[8] a) Y. Coppel, J.-F. Constant, C. Coulombeau, M. Demeunynck, J. Gar-
cia, J. Lhomme, Biochemistry 36 (1997) 4831¯4843; b) P. Belmont, M.
Jourdan, M. Demeunynck, J-F. Constant, J. Garcia, J. Lhomme, D. Car-
ez, A. Croisy, J. Med. Chem. 42 (1999) 5153¯5159; c) J. Lhomme, J-F.
Constant, M. Demeunynck, Biopolymers 52 (1999) 65-83; d) J. M. Bar-
ret, C. Etievant, J. Fahy, J. Lhomme, B.T. Hill, Anticancer Drugs 10
(1999) 55-65; e) K. Alarcon, M. Demeunynck, J. Lhomme, D. Carrez,
A. Croisy, Bioorg. Med. Chem. 9 (2001) 1901-1910; f) A. Martelli, J.F.
Constant, M. Demeunynck, J. Lhomme, P, Dumy, Tetrahedron 58 (2002)
4291–4298.
[9] L.-M. Tumir, I. Piantanida, P. Novak, M. Žinić, J. Phys. Org. Chem. 15
(2002) 599–607.
[10] I. Juranović, Z. Meić, I. Piantanida, L.-M. Tumir, M. Žinić, Chem. Com-
mun. (2002) 1432–1433.
[11] L.-M. Tumir, I. Piantanida, I. Juranović Cindrić, T. Hrenar, Z. Meić, M.
Žinić, J. Phys. Org. Chem. 16 (2003) 891–899.
[19] J.-L. Dimicoli, C. Hélène, J. Am. Chem. Soc. 95 (1974) 1036–1044.
[20] W.D. Cornell, P. Cieplak, C.I. Payly, I.R. Gould, K.M. Merz, D.M. Fer-
guson, D.C. Spellmeyer, T. Fox, J.W. Caldwell, P.A. Kollman, J. Am.
Chem. Soc. 117 (1995) 5179–5197.
[21] I. Piantanida, V. Tomišić, M. Žinić, J. Chem. Soc. Perkin Trans. 2 (2000)
375–383.
[22] A. Odani, H. Masuda, O. Yamauchi, S. Ishiguro, Inorg. Chem. 30 (1991)
4486–4488.
[23] N.W. Luedtke, Disertation, University of California, San Diego, 2002.
[24] S. Georghiou, Photochem. Photobiol. 26 (1977) 59–68.
[25] M.-P. Teulade, J.-M. Lehn, J.-P. Vigneron, J. Supramol. Chem. 5 (1995)
139–147.
[26] I. Piantanida, B.S. Palm, M. Žinić, H.-J. Schneider, J. Chem. Soc. Perkin
Trans. 2 (2001) 1808–1816.
[27] A. Gentil, G. Renault, C. Madzak, A. Margot, J.B. Cabral-Neto, J.J. Vas-
seur, B. Rayner, J.L. Imbach, A. Sarasin, Biochem. Biophys. Res. Com-
mun. 173 (1990) 704–710.
[28] a) J. Cymerman, W.F. Short, J. Org. Chem. (1949) 703-704; b) J.W.
Lown, B.C. Gunn, K.C. Majumdar, E. McGoran, Can. J. Chem. 57
(1979) 2305–2313.
[29] a) S P E C F I T G L O B A L A N A L Y S I S, a Program for Fitting,
Equilibrium and Kinetic Systems, using Factor Analysis & Marquardt
Minimization; b) H. Gampp, M. Maeder, C.J. Meyer, A.D. Zuberbuehler,
Talanta 32 (1985) 257-264; c) M. Maeder A.D. Zuberbuehler, Anal.
Chem. 62 (1990) 2220–2224.
[12] L.-M. Tumir, I. Piantanida, I. Juranović, Z. Meić, S. Tomić, M. Žinić,
Chem. Commun. (2005) 2561–2563.
[13] S. Takenaka, M. Manabe, M. Yokoyama, M. Nishi, J. Tanaka, H. Kon-
do, Chem. Commun. (1996) 379–380.
[14] I. Piantanida, B.S. Palm, P. Čudić, M. Žinić, H.-J. Schneider, Tetrahe-
dron Lett. 42 (2001) 6779–6783.
[15] a) N. Berthet, J. Michon, J. Lhomme, M.P. Teulade-Fichou, J.P. Vigner-
on, J-M. Lehn, Chem. Eur. J. 5 (1999) 3625-3630; b) M. Jourdan, J. Gar-
cia, J. Lhomme, M.P. Teulade-Fichou, J.P. Vigneron, J-M, Lehn, Bio-
chemistry 38 (43) (1999) 14205–14213.
[30] M.R. Boyd, K.D. Paull, Drug Dev. Res. 34 (1995) 91–109.
[31] T. Mossman, J. Immunol. Methods 65 (1983) 55–63.