Bleomycin mimics. Design and synthesis of an acridine derivative which cleaves DNa in a sequence-neutral manner

Article abstract of DOI:10.1039/a606218k

Bleomycin 1 is a clinically used anticancer drug that cleaves DNA at guanine-pyrimidine dinucleotides by a free-radical mechanism. The compound FTP1 2 was designed as a bleomycin mimic to cleave DNA in a non-sequence selective manner. Compound 2 may provide a route to novel, chemically simple, bleomycin analogues that also have potential as new DNA footprinting agents. FTP1 2 was synthesised from chelidamic acid, 1,3-phenylenediamine and 9-chloroacridine. An untargeted agent 8 was also made. Compound 2 cleaves DNA in a relatively non-sequence selective manner although a number of hotspots of similar sequence were found. It has a 20-fold higher cutting activity than the untargeted agent 8 and has been used to footprint two antibiotics, actinomycin D 9 and a minor groove binder, distamycin A 10. It gives clear, well-defined footprints which compare favourably with those reported by MPE-Fe^II, DNase I and iron-EDTA and is a useful addition to the range of footprinting agents. Its cytotoxicity against human colon carcinoma cells in culture is 20 times lower than that of bleomycin although its ability to cleave naked DNA is 10 000-fold less, suggesting that on a mole-for-mole basis FTP1-induced DNA damage is more biologically harmful.

Full text of DOI:10.1039/a606218k

View Article Online / Journal Homepage / Table of Contents for this issue
Bleomycin mimics. Design and synthesis of an acridine derivative
which cleaves DNA in a sequence-neutral manner
Mark Searcey, Siobhán McClean, Bernie Madden and
Laurence P. G. Wakelin*
Cancer Drug Discovery, Chemistry Department, University College Dublin, Belfield, Dublin 4,
Ireland
Bleomycin 1 is a clinically used anticancer drug that cleaves D NA at guanine–pyrimidine dinucleotides by
a free-radical mechanism. The compound F TP1 2 was designed as a bleomycin mimic to cleave D NA in a
non-sequence selective manner. Compound 2 may provide a route to novel, chemically simple, bleomycin
analogues that also have potential as new D NA footprinting agents. F TP1 2 was synthesised from
chelidamic acid, 1,3-phenylenediamine and 9-chloroacridine. An untargeted agent 8 was also made.
Compound 2 cleaves D NA in a relatively non-sequence selective manner although a number of hotspots
of similar sequence were found. It has a 20-fold higher cutting activity than the untargeted agent 8 and
has been used to footprint two antibiotics, actinomycin D 9 and a minor groove binder, distamycin A 10. It
gives clear, well-defined footprints which compare favourably with those reported by M PE–FeII, D N ase I
and iron–ED TA and is a useful addition to the range of footprinting agents. Its cytotoxicity against
human colon carcinoma cells in culture is 20 times lower than that of bleomycin although its ability to
cleave naked D NA is 10 000-fold less, suggesting that on a mole-for-mole basis F TP1-induced D NA
damage is more biologically harmful.
moiety, is the chief constituent of the clinically used mixture
Introduction
blenoxane. Recently, studies2 of bleomycin analogues in which
the iron chelating portion has been modified to contain a
pyridine with an electron-donating substituent in the 4-
position, have shown that these compounds, which are targeted
to DNA through the usual bithiazole moiety but which lack the
disaccharide of the naturally occurring compound, have a
DNA cleaving activity about 10% of that of bleomycin. An
analogue of the minor groove binding compound distamycin,
to which a similar pyridine derived iron (II) chelating moiety
was attached, was shown to cleave DNA 3 at recognised
distamycin binding sites. A derivative of chelidamic acid has
also been appended to an oligonucleotide and shown to cleave
Bleomycins are a family of antitumour antibiotics used in com-
bination chemotherapy against several tumours.1 Bleomycin A2
1, which comprises an iron(II) binding β-aminoalanine–
pyrimidine–β-hydroxyhistidine and a DNA-binding bithiazole
DNA in
a
sequence specific manner.4 Such ‘minimum
analogues’ of bleomycin are of interest for a number of
reasons. The targeting of iron(II) binding pyridine derivatives to
DNA by the use of minor groove binders, oligonucleotide
analogues or intercalators may lead to compounds with
improved antitumour activity compared to the natural com-
pound. It also allows investigation of the sequence preferences
of the targeting moiety through determination of DNA cleav-
age patterns by the compound in the presence of iron(II) and a
reducing agent. Similarly, the resulting compounds may in
themselves be useful tools in the investigation of drug–DNA
interactions by the technique of footprinting.
There is growing interest in compounds that bind to DNA in
a sequence specific manner since they may have therapeutic
application in the treatment of cancer, viral and bacterial infec-
tions.5 Investigations into the sequence specificities of minor
groove binders,6 intercalators7 and oligonucleotide analogues8
have been carried out using various techniques including those
known collectively as footprinting. Footprinting involves the
cutting of a strand of 32P end-labelled DNA of known sequence
both in the absence and in the presence of varying concen-
trations of ligand. From the resulting autoradiograms, infor-
mation regarding the sequence preference of the DNA binding
compound, its binding affinity and even some kinetic data 9 may
be obtained. DNA cleavage reactions may be carried out either
by enzymes or by chemical agents. Enzymic reactions suffer
J. Chem. Soc., Perkin Trans. 2, 1997
523

View Article Online
from some inherent sequence preferences. For instance, DNase
I preferentially cleaves sequences of mixed nucleotide com-
position, and cleaves more efficiently in GC regions than AT.10
Hydroxyl radical cleavage by the FeII–EDTA complex also has
limitations, particularly when attempting to footprint inter-
calating compounds.11 The preferred chemical agent, known as
MPE–FeII [methidiumpropylethylenediaminetetraacetic acid–
iron(II)], is a derivative of methidium and EDTA and its syn-
thesis is not trivial.7 Until recently, it has not been commercially
available.
In order to develop a synthetically more accessible com-
pound which will cleave DNA in a non-sequence specific man-
ner and which has potential as both a new footprinting agent
and also as an experimental antitumour ‘bleomycin mimic’, we
have designed the chelidamic acid derivative 2. The compound
is targeted to DNA via a 9-anilinoacridine group. 9-Amino-
acridines have been shown 12 to bind to DNA with little
sequence preference and would be expected to give compounds
which cleaved in essentially a non-sequence specific manner.
Here, we describe the synthesis of the new compound, its ability
to cleave DNA, initial studies on its application to the footprint-
ing of distamycin A and actinomycin D and measurements of
its cytotoxicity against human colon carcinoma cell line. DNA
cleavage by the new agent, FTP1 2 was compared to an
analogue 8 which is not targeted to DNA through an acridine
moiety, although it carries a similar charge and also to the well
known footprinting agents MPE–FeII, FeII–EDTA and DNase
I. A proposed structure for iron(II) binding is shown by 2a.
Conversion to the acridine derivative 5 was carried out via reac-
tion with freshly prepared 9-chloroacridine and gave, after
purification by flash column chromatography,13 an 84% yield
of the acridine methyl ester. Hydrolysis under acid-catalysed
conditions followed by precipitation from ethanol with ethyl
acetate gave the target compound 2 in 74% yield as the hydro-
chloride salt.
In order to investigate the degree of activity conferred on the
compound by targeting to DNA via intercalation of the acrid-
ine moiety, and also to explore the sequence preference, if any,
of an untargeted compound, the N,N-dimethylacetylamino
analogue 8 was also prepared (Scheme 2). This compound
Results and discussion
Synthesis of FTP1 and a non-targeted analogue
The synthesis of FTP1 is outlined in Scheme 1. Methyl 6-
Scheme 2 (i) Chloroacetic anhydride, CH2Cl2, room temp., 3.5 h. (ii)
Dimethylamine, CH3CN, room temp., 21 h. (iii) NaOH, H2O–MeOH,
room temp., 3 h.
should still have some non-specific interaction with the
negatively charged nucleic acid by virtue of its positive charge
and should also be reasonably soluble. Its synthesis was
straightforward, proceeding directly from the aniline methyl
ester 4 to the chloroacetyl derivative 6 via reaction with chloro-
acetic anhydride under relatively mild conditions. Addition of
the dimethylamino group via alkylation with dimethylamine in
acetonitrile gave the methyl ester derivative 7 of the required
compound, which was not characterised but immediately
hydrolysed to the acid 8.
Scheme 1 (i) Methanesulfonyl chloride, Et3N, CH2Cl2, room temp.,
1 h. (ii) 1,3-phenylenediamine (3 equiv.), room temp., 16 h. (iii) 9-
Chloroacridine, anhydrous methanol, room temp., 2 h. (iv) 1 M HCl, ∆,
3 h, then precipitated from ethanol–ethyl acetate.
hydroxymethyl-4-methoxypyridine-2-carboxylate2 was con-
verted to the anilino-derivative 4 via mesylation and reaction
with 1,3-phenylenediamine. Phenylenediamine was used as the
linker, rather than flexible bridges such as propylenediamine, to
avoid potential intramolecular reactions of 9-aminoacridines.
Reaction via the mesylate, rather than by the previously
described 2,3 Swern oxidation–reductive alkylation, gave the
required product in reasonable yield in a one-pot reaction.
DNA Cleavage by FTP1
The cleavage of both strands of the SV40 early promoter frag-
ment of DNA by 20 µM FTP1 2, equimolar iron(II) in the pres-
ence of dithiothreitol (DTT, excess) was analysed by scanning
laser densitometry (Fig. 1). Investigation of the concentration
dependence of the extent of cleavage (data not shown) shows
this concentration of FTP1 and iron(II) to be optimal for
524
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Fig. 1 DNA cleavage patterns obtained with 20 µ FTP1, 20 µM Fe(NH4)2(SO4)2 and 5 mM dithiothreitol analysed by laser densitometry. (a) 3Ј-
Labelled DNA, sequence 5Ј to 3Ј. (b) 5Ј-Labelled DNA, sequence 3Ј to 5Ј. (c) Summary of hotspot sequences. Dotted vertical lines represent
potential binding sites, bold filled lines represent the extent of the hotspot and smaller lines the site of most intense cutting.
‘single hit kinetics’.14 Bleomycin cleaves DNA to a similar ex-
tent at a concentration of 2 nM.15 At 20 µM, FTP1 2 cuts DNA
at all nucleotides with a general five-fold variation in cutting
intensity. In addition, there are four hotspots which show an
increased, up to 15-fold, variation in cleavage, labelled 1–4 in
Fig. 1(a) and (b) and shown in Fig. 1(c). Three of the four
J. Chem. Soc., Perkin Trans. 2, 1997
525

View Article Online
hotspots comprise alternating purine–pyrimidine sequences
with the two sites of highest cleavage intensity (1 and 2) con-
taining a similar GpyATGC motif. The cutting pattern at these
sites suggests that the intercalator is positioned between the
base pairs of the central ApT step. Site 3 appears to be centred
around a TpA step in the sequence 5Ј-TTAGGG. Site 4 is simi-
lar to 1 and 2 in being a purine–pyrimidine sequence but is of
the form (GT)n. The acridine may be intercalating at any of the
GpT steps, although as cleavage appears to be asymmetric with
cutting only clearly visible on the 5Ј-strand, it is not possible to
identify a precise binding site. Both MPE–FeII and FeII–EDTA
also show the hotspot at position 3 (data not shown), but not at
the other three sites, which suggests that enhanced cleavage by
hydroxyl radical at this position is a consequence of sequence
rather than being attributable to the nature of the cleaving
agent.
secondary sites for actinomycin D. Sites 1, 2 and 4 clearly dis-
play a 3Ј-overhang which suggests that cleavage is taking place
in the minor groove, similar to observations for DNase I, an
enzyme known to cut the sugar phosphate backbone from this
groove.10 Site 3 is extended on the 5Ј-strand, perhaps suggesting
some binding to the secondary CpC and GpG sites to the 5Ј-
and 3Ј-sides respectively of the main site. Site 5, which is not
revealed on the 3Ј-strand, is a poor binding site for actinomycin
D. Its sequence 5Ј-TATGCAAA contains a GC step in the
centre of an AT rich region which suggests a narrow minor
groove, perhaps too narrow to accommodate the cyclic
depsipeptides. It has previously been shown that actinomycin D
dissociates rapidly from the sequence T9GCA9, suggesting that
a GC step preceded by thymine and followed by a run of
adenines is a poor binding site for the drug.17 Sites 1, 2 and 5
extend over 4–6 base pairs, which concurs with the number of
base pairs physically occluded by actinomycin D as revealed by
crystallography.18
DNA cleavage by the untargeted compound 8
Comparison of cleavage by FTP1 2 and the untargeted agent 8
in the presence of equimolar iron(II) showed that the untargeted
compound cleaves to a similar extent to 2 only at a concen-
tration of 200 µM (data not shown), indicating that targeting of
the chelidamic acid moiety to DNA via the acridine chromo-
phore leads to a compound which is at least 20 times more
effective. It should also be noted that iron(II) cleaves DNA in
the absence of added ligand to a similar extent to compound 8
at comparable concentrations.
A number of interesting features are revealed by comparing
FTP1-derived actinomycin D footprints with those generated
by MPE–FeII and DNase I (Fig. 2). No inhibition of cleavage is
seen at site 1 for MPE–FeII whereas site 2 is almost identical to
that reported by FTP1. Site 5 is again not seen on the 3Ј-strand
while site 3 is longer, including the sequence five base pairs to
the 3Ј-side of the site given by FTP1. Site 4, which includes the
two secondary GpG steps is smaller and misses much of the
3Ј-end given by the new reagent. A number of studies with
MPE 11,19 have suggested that the strong affinity of methidium
for DNA may lead to displacement of the ligand under investi-
gation from some weaker, secondary sites. It is possible that this
is also occurring at sites in this study. Comparing the FTP1 and
DNase I footprints reveals that the sequence selectivity of the
enzyme leads to profound variations in cutting frequency,
making site size difficult to define. However, FTP1 and DNase I
clearly display the same five sites, although sites 2, 3 and 5
appear to coalesce into one large footprint on the DNase
I-cleaved fragment. Iron–EDTA shows no footprints for
actinomycin D.
Footprinting of actinomycin D
In order to investigate the suitability of FTP1 2 as a foot-
printing agent, its ability to reveal the sequence selectivities
of two known DNA binding antibiotics was assessed, namely
the GC-selective intercalator actinomycin
D 9 and the
AT-selective minor groove binder distamycin A 10. Result-
ing autoradiographs were analysed using scanning laser
densitometry.
Footprinting of distamycin A
Distamycin A 10 is a member of the family of oligopyrrole
peptide antibiotics. It binds to DNA by displacing the spine of
hydration from the minor groove of AT-rich sequences where
its curved planar structure is isohelical with the duplex. It
makes hydrophobic contacts with the walls of the groove and
electrostatic, hydrophobic and hydrogen bonding interactions
in the base of the groove. Previous studies20 have shown that the
precise ordering of the bases in the site modifies the width of
the groove and determines the binding mode for distamycin.
Monomeric binding predominates for the narrow sequence
AAAAA while the sequence AAATT, with a slightly wider
minor groove, accommodates side-by-side dimers. When
distamycin binds to the SV40 early promoter fragment, three
sites are clearly protected from cleavage by FTP1 2, as revealed
by scanning laser densitometry (see Fig. 3). Site 1 is a classical
distamycin binding site with an AATTA sequence flanked by
GC. Sites 2 and 3 show an acceptance of GC base pairs in the
body of an AT tract. Again, all three sites have a 3Ј-overhang,
suggesting that cleavage by FTP1 is in the minor groove.
MPE reveals similar distamycin footprints to FTP1, there
being a number of qualitative differences. For instance, inhibi-
tion of cleavage by MPE at site 1 is weaker, the site is longer and
displaced to the 3Ј-end. Site 2, while covering essentially the
same sequence, is also weaker in intensity. By contrast, site 3 is
longer, inhibition of cleavage is enhanced and the footprint is
displaced to the 5Ј-end. As reported by DNase I, site 1 becomes
more complex and is difficult to define. It is displaced to the 3Ј-
end and an enhancement of cleavage occurs within the site
defined by FTP1. Site 2 is comparable with that given by FTP1
but, due to the sequence selectivity of DNase I, no cutting
Actinomycin D 9 is an antitumour antibiotic which binds to
DNA by intercalation, positioning its two cyclic depsipeptides
in the minor groove. Previous footprinting studies have revealed
a selectivity for binding to 5Ј-GpC steps with some secondary
affinity for GpG and CpC sequences.16
Footprinting studies of actinomycin D using FTP1 2 as the
cleavage agent show protection from cleavage at four sites on the
the 3Ј-labelled strand, a fifth site being revealed on the 5Ј-labelled
strand (Fig. 2). Four of the five sites contain a GpC step while
the fifth (labelled site 4) contains two GpG sequences, potential
526
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Fig. 2 Comparison of differential cleavage plots for actinomycin D-induced protection from cleavage of 3Ј end-labelled SV40 early promoter DNA
by FTP1 with those obtained by cleavage with (a) MPE, (b) DNase I. FTP1 = filled circles, other agents open circles. Sites 1–4 are labelled for clarity.
Vertical scales are in units of ln fa Ϫ ln fc, where fa is the fractional cleavage at any bond in the presence of the drug (0.1 molecule /bp) and fc is the
fractional cleavage of the same bond in the control. Positive values indicate enhancement, while negative values indicate protection. (c) Sequence
summary showing sites of footprints obtained in the presence of actinomycin D using FTP1 as the cleaving agent.
J. Chem. Soc., Perkin Trans. 2, 1997
527

View Article Online
Fig. 3 Comparison of differential cleavage plots for distamycin-induced protection from cleavage of 3Ј end-labelled SV40 early promoter DNA by
FTP1 with those obtained by cleavage with (a) MPE, (b) FeII–EDTA or (c) DNase I. FTP1 = filled circles, other agents open circles. Vertical scales
are in units of ln fa Ϫ ln fc, where fa is the fractional cleavage at any bond in the presence of the drug (0.1 molecule/bp) and fc is the fractional cleavage
of the same bond in the control. Positive values indicate enhancement, while negative values indicate protection. Sites 1–3 are shown for clarity. (d)
Sequence summary showing sites of footprints obtained in the presence of actinomycin D using FTP1 as the cleaving agent.
occurs at certain bases in either control or drug-treated lanes.
Site 3 is not discernible with DNase I. Iron()–EDTA shows
very weak footprints, with only site 1 clearly discernible.
less. If FTP1 kills cells by a free radical mechanism, this indi-
cates that on a mole-for-mole basis FTP1 induces strand breaks
500-fold more cytotoxic than those caused by bleomycin. This
may be due to differences in intracellular uptake and/or due to
the number and type of strand breaks induced. For instance,
differences may derive from the ratio of double to single strand
breaks generated by the two compounds or from the sequence
selectivity of bleomycin for GpC and GpT sequences compared
to the sequence neutrality of FTP1. Analogues of FTP1 which
cleave DNA in a sequence-selective fashion are currently under
investigation.
Cytotoxicity studies
The toxicity of FTP1 2 was measured in the human colon car-
cinoma cell line HT-29. The concentration required to inhibit
growth by 50% (IC50) was found to be 57 µM. In the same cell
line, bleomycin wasfound to havean IC50 of 3.1µM. Although the
ability of FTP1 to cleave naked DNA is 10 000-fold lower than
that for bleomycin, its cytotoxicity in this cell line is only 20-fold
528
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
However, FTP1 also shares many structural characteristics
with the antitumour agent amsacrine, which is also a 9-anilino-
acridine derivative substituted at the 2Ј- and 4Ј-positions. The
mechanism of action of amsacrine involves inhibition of the
DNA processing enzyme topoisomerase II and its IC50 in the
Ht-29 cell line was found to be 170 nM. This suggests that if the
activity of FTP1 stems from something other than DNA cleav-
age, that it is a poor inhibitor of topoisomerase II.
calated in the DNA has been a matter of some debate. Models
have been suggested for amsacrine, based on computational
calculations, in which the anilino-portion resides in the minor
groove,23 whereas a kinetic study suggested that a major groove
interaction could be feasible.22 Once again, NMR studies fail to
resolve this problem and no crystal structures have been
reported of 9-anilinoacridine–oligonucleotide complexes.
The studies reported here provide some insights into the reso-
lution of these issues. By tethering a chelidamic acid derivative
to 9-anilinoacridine, which, as a bleomycin mimic, has the abil-
ity to cleave DNA in the presence of iron() and a reducing
agent, it is possible to ascertain whether the acridine displays
any sequence selectivity. By using the compound as a footprint-
ing agent, it is possible to analyse the overhang generated by
inhibition of cleavage on the 3Ј- and 5Ј-labelled strands and to
find the orientation of binding of the cleavage agent. We find
that the sequence selectivity of FTP1 2 appears to be slight,
with some preference for an alternating purine–pyrimidine
sequence, in particular the sequence GpyATGC. The overhang
generated by both actinomycin D and distamycin A in the foot-
printing experiments is directed towards the 3Ј-end of the DNA
Sequence selectivity and binding orientation of FTP1
The broad sequence neutrality of intercalation by 9-anilino-
acridines was first established by the results of binding studies
using natural DNAs of varying sequence composition and
synthetic oligonucleotides such as poly(dA᎐dT) and poly
(dG᎐dC).21 In recent times, NMR and footprinting studies have
failed to give any more detailed information as regards the
sequence preference of 9-anilinoacridines for mixed sequence
DNA. This is presumably due to the kinetics of binding of the
9-anilinoacridines where the residence time on the duplex has
been shown to be of the order of milliseconds.22 Similarly, the
orientation of the 9-substituent when the chromophore is inter-
J. Chem. Soc., Perkin Trans. 2, 1997
529

View Article Online
fragment, suggesting that FTP1 binds to DNA with the 9-
substituent located in the minor groove.
was purified by flash column chromatography (gradient
elution : chloroform–5% methanol in chloroform) to give
pure product as an orange solid (Rf 0.20, 10% methanol in
chloroform), 172 mg (84%).
δH (CDCl3) 8.02 (d, 2 H, J 9), 7.98 (d, 2 H, J 11), 7.57 (t, 2 H, J
7.63, 8.16), 7.52 (d, 1 H, PyH, J 2.54), 7.20 (t, 2 H, J 7.03, 8.16),
7.06 (t, 1 H, J 8.15, 7.88), 7.01 (d, 1 H, PyH J 2.25), 6.36 (t, 2 H,
ArH, J 6.75, 7.87), 6.28 (s, 1 H, ArH), 4.40 (s, 2 H, CH2), 3.98
(2, 3 H, OCH3) and 3.86 (s, 3 H, CO2CH3).
6-{N-[3-(acridin-9-yl)aminophenyl]aminomethyl}-4-methoxy-
pyridine-2-carboxylic acid (2) (FTP1). The methyl ester 5 (406
mg, 1.58 mmol) was dissolved in hydrochloric acid (1 M, 20 ml)
and heated under reflux for 3 h. At the end of this period the
solvent was removed under reduced pressure and the residue
azeotroped with toluene (2 × 25 ml). The residue was then dis-
solved in ethanol (10 ml) and precipitated with excess ethyl
acetate. The resulting solid was filtered and immediately dried
over P2O5 in vacuo. The yield was 283 mg (74%).
Microanalysis. C27H22N4O3ؒ3HClؒ4H2O requires C 51.31 H
5.23 N 8.87 Cl 16.86, found C 51.56 H 5.03 N 8.72 Cl 17.31.
δH [(CD3)2SO] 11.64 (s, 1 H, OH), 8.26 (d, 2 H, J 8.72), 8.17
(d, 2 H, J 8.72), 7.93 (t, 2 H, J 7.60, 7.87), 7.46 (d, 1 H, PyH, J
2.25), 7.32 (t, 2 H, J 7.32, 8.15), 7.21 [m (d overlapping t), 1 H,
J 8.15, 7.88], 6.73 (d, 1 H, J 9), 6.66 (s, 1 H), 6.62 (d, 1 H, ArH, J
9), 4.40 (s, 2 H, CH2) and 3.89 (s, 3 H, OCH3). Mass Spectrum
(FAB+). 451 (M + 1), 407, 329, 274, 176, 136, 107, 89, 77, 63, 51
and 39.
6-{N-[3-(N,N-dimethylaminoacetyl)aminophenyl]amino-
methyl}-4-methoxypyridine-2-carboxylic acid (8). The aniline
derivative (4) (90 mg, 0.31 mmol) was stirred in dry dichloro-
methane (1 ml). Chloroacetic anhydride (90%, 60 mg, 0.31
mmol) was added and the solution stirred for 3.5 h. The solvent
was removed under reduced pressure, the residue redissolved in
chloroform (10 ml) and washed with water (2 × 5 ml), sodium
hydrogen carbonate (1 M, 2 × 5 ml) and water (2 × 5 ml). The
organic solution was dried (MgSO4) and evaporated to give the
chloroacetyl derivative 6 as a green oil, 91 mg (81%) which was
unstable over long periods at room temp.
Whether these conclusions can be extended to other acridines
is not altogether clear. It is probable that the 9-substituent in
FTP1 is orientated orthogonally to the chromophore, as previ-
ously described for amsacrine,24 and as such it is likely that both
compounds can bind in a similar groove orientation. The 9-
substituent of FTP1 contains a number of potential hydrogen
bond donors and acceptors which could endow a sequence
recognition different to that for other 9-anilinoacridines. How-
ever, compound 8 does not appear to generate the same cleav-
age hotspots as FTP1, supporting the inference that it is the
acridine and not the 9-substituent which endows the sequence
selectivity on DNA.
Experimental
Synthesis of FTP1
Materials. With the exception of chelidamic acid (Fluka
Chemical Ltd., Dorset, UK), all other chemicals were from
Aldrich (Dorset, UK). Anhydrous solvents were from Aldrich
and were used without further purification. Chromatography
was carried out using Silica gel 60 TLC plates (Merck 5554
and 13792) and silica gel 60 for flash chromatography (Merck
particle size 0.040–0.063 mm). Methyl 6-hydroxymethyl-4-
methoxypyridine-2-carboxylate 3 was made via esterification of
chelidamic acid with acidic methanol,4 alkylation of the 4-
hydroxy group with methyl iodide and reduction with sodium
borohydride.2
Methods. TLC plates were visualised by UV (254 nm) and
were also developed using cerium(IV) sulfate (5% mass/vol in 4
M sulfuric acid) followed by heating. Flash column chromato-
graphy was carried out by the method of Still et al.13 NMR
spectra were recorded using a JEOL 270 MHz spectrometer
and were referenced to SiMe4 or to residual Me2SO. δ values are
given in ppm; J values are given in Hz. Microanalyses were
provided by the Chemical Services Unit of University College
Dublin. Fast Atom Bombardment Mass Spectra using 3-
nitrobenzyl alcohol as a matrix were obtained from the Mass
Spectroscopy Service of University College London, UK.
Methyl 6-[N-(3-aminophenyl)aminomethyl]-4-methoxypyri-
dine-2-carboxylate 4. Methyl 6-hydroxymethyl-4-methoxy-
pyridine-2-carboxylate 3 (475 mg, 2.41 mmol) was stirred in dry
dichloromethane (5 ml). Methanesulfonyl chloride (205 ml,
2.65 mmol), triethylamine (369 ml, 2.65 mmol) and 4-
dimethylaminopyridine (32 mg, 0.27 mmol) were added and the
solution stirred for 30 min. 1,3-Phenylenediamine (783 mg, 7.32
mmol) was dissolved in anhydrous dichloromethane (5 ml) and
added to the above solution in one portion. The resulting solu-
tion was stirred at room temp. for 16 h. Water (10 ml) and
dichloromethane (10 ml) were then added. The layers were sep-
arated and the organic layer washed with water (10 ml), then
extracted with hydrochloric acid (1 M, 2 × 10 ml). The pH was
adjusted to 4–5 (1 M NaHCO3) and the resulting solution
extracted with chloroform (2 × 20 ml). The organic layers were
dried (MgSO4) and volatiles removed under reduced pressure to
give a brown oil, pure by TLC (Rf 0.30, 10% MeOH in CHCl3),
495 mg (72%), which was used without further purification.
δH (CDCl3) 7.54 (d, 1 H, PyH, J 2.25), 7.08 (d, 1 H, PyH, J
2.54), 6.93 (t, 1 H, ArH, J 8.16, 7.87), 6.07 (dd, 2 H, ArH, J
1.97, 7.88, 1.97), 4.80 (s, 2 H, CH2), 4.00 (s, 3 H, OCH3) and
3.84 (s, 3 H, CO2CH3).
δH (CDCl3) 8.16 (s, 1 H, NH), 7.57 (d, 1 H, PyH, J 2.53), 7.06
(m, 3 H, ArH), 6.78 (d, 1 H, ArH, J 7.87), 6.43 (d, 1 H, ArH, J
7.59), 4.51 (s, 2 H, CH2), 4.15 (s, 2 H, CH2), 4.01 (s, 3 H, CH3)
and 3.87 (s, 3 H, CH3).
This compound 6 (43 mg, 0.12 mmol) was dissolved in
acetonitrile (1 ml) and dimethylamine (33% in ethanol, 19 ml,
0.14 mmol) was added. The solution was stirred for 16 h, after
which time the reaction was still incomplete (TLC, 10% meth-
anol in chloroform, Rf starting material 0.50, Rf product 0.20).
1 equiv. of dimethylamine (33% in ethanol, 19 ml, 0.14 mmol)
was added, the mixture stirred for 3 h, 1 equiv. of dimethyl-
amine (33% in ethanol, 19 ml, 0.14 mmol) added and the mix-
ture again stirred for 3 h. At the end of this period, the solvents
were removed in vacuo and water (2 ml) and chloroform (1 ml)
added. The organic fraction was separated, and the water layer
extracted with chloroform (3 × 2 ml). The organic layers were
combined, dried (MgSO4) and evaporated. The resulting oil was
purified by preparative TLC (Silica gel, eluting solvent 10%
methanol in chloroform, Rf 0.20) to give the methyl ester 7 as an
oil, 37 mg (84%). This compound was immediately dissolved in
methanol (1 ml) and sodium hydroxide (4 mg, 0.1 mmol) in
water (0.5 ml) was added. After 3 h at room temp., the solvents
were removed under reduced pressure and the resulting oil
azeotroped with toluene (1 × 5 ml). It was redissolved in dry
methanol (1 ml) which was then saturated with dry hydro-
gen chloride. Addition of ethyl acetate precipitated a product 8
(24 mg, 69%) which, although correct by NMR analysis, con-
tained 1.5 equiv. of sodium chloride by microanalysis. It was
used without further purification for the DNA cleavage
experiments.
Methyl 6-{N-[3-(acridin-9-yl)aminophenyl]aminomethyl}-4-
methoxypyridine-2-carboxylate (5). 9-Chloroacridine (95 mg,
0.44 mmol) was dissolved in dry methanol (2 ml) and the
above product (4) (127 mg, 0.44 mmol) was added. The mix-
ture was stirred at room temp. for 16 h, volatiles removed
under reduced pressure and the residue dissolved in water (1
ml). The solution was basified (2 M NaOH) and extracted
with chloroform (4 × 10 ml). The organic layers were dried
(MgSO4), and evaporated to give an orange oily solid. This
δH [(CD3)2SO] 10.91 (s, 1 H, OH), 7.57 (d, 1 H, PyH, J 2.53),
7.35 (s, 1 H, PyH), 7.10 (m, 2 H, ArH), 7.01 (d, 1 H, ArH, J 8),
530
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Table 1
(10) was incubated with DNA (30 000 cpm SV40 early pro-
moter fragment plus 1 µg µlϪ1 calf thymus DNA, in 20 mM Tris
HCl, pH 7.6, 50 mM NaCl) for 30 min at room temp., before
addition of 1µl FTP1 (200 µM), 1 µl FeII (200 µM) and 1µl DTT
(50 mM) in the order stated. After a further 15 min, the reac-
tions were stopped with thiourea and the DNA processed as
outlined above.
Footprinting with MPE, FeII–EDTA and DNase I. To foot-
print ligand binding sites with MPE, FeII–EDTA or DNase I,
the labelled DNA was incubated as outlined above with either
of the two ligands for 30 min. Then the cleaving agents were
added under conditions which resulted in 10–20% cleavage of
the DNA. For MPE footprinting, a 1:1 mix of MPE and
[Fe(NH4)2(SO4)2] was added at a final concentration of 2 µM
followed by 1 µl DTT (50 mM), reactions proceeded for 10 min.
For generating FeII–EDTA footprints, a mixture of reagents
was prepared by adding the following in the order listed:
[Fe(NH4)2(SO4)2] (200 µM) EDTA (4 µM), ascorbic acid (10 mM)
and 0.3% H2O2 in the ratio of (1:1:2:2). An aliquot of 1 µl
of this mix was added to each reaction and the cleavage allowed
to proceed for 3 min. DNase I footprinting was carried out by
incubating DNA–ligand complexes with DNase I (1/2000 dilu-
tion of 1 µg mlϪ1 stock) for 1 min on ice. Reactions were stopped
by addition of thiourea (100 mM) in the case of MPE and FeII–
EDTA, or addition of DNase I stop solution (200 mM NaCl, 30
mM EDTA, 1% SDS, 100 µg mlϪ1 yeast tRNA) and the DNA
processed as outlined above.
Densitometry. Densitometric analysis of autoradiographs was
carried out using a Molecular Dynamics scanning laser densi-
tometer and the scanned images subsequently analysed using
the PHORETIX 1D software package (Non-linear Dynamics).
The lanes analysed were corrected for the counts loaded and for
the extent of strand breaks in the untreated DNA control.
Growth inhibition assays. Human colon carcinoma HT29/219
cells were obtained from the American Type Cell Culture
Collection and grown at 37 ЊC in the presence of 5% CO2 in
minimum essential medium with Earle’s salts (EMEM) sup-
plemented with 10% heat inactivated foetal calf serum, 2 mM
glutamine, 0.1 mM (MEM) non-essential amino acids, 10 mM
HEPES, pH 7.3, 50 U mlϪ1 penicillin and 50 mg mlϪ1 strepto-
mycin. Cells were routinely seeded at a density of 2 × 104 cells
mlϪ1 in 6 well plates, 48 h prior to drug addition.
All drug stock solutions, with the exception of bleomycin,
were in Me2SO at a concentration of 10 µM and stored at Ϫ20 ЊC
until use. Subsequent dilutions were made at the time of drug
treatment using sterile distilled H2O. Bleomycin was dissolved
in sterile distilled H2O at a concentration of 10 µM and stored at
Ϫ20 ЊC until use.
For each drug treatment the test compound was added to the
cells in concentrations of 0.01, 0.1, 1.0 and 10 µM. Cells were
exposed continuously to drug for 72 h after which time the cells
were trypsinized and counted using a Coulter Counter (ZM
model). IC50 values were determined by calculating the drug
concentration required to reduce cell growth to 50% of the con-
trol cell values.
6.48 (d, 1 H, ArH, J 8.16), 4.49 (s, 2 H, CH2), 4.13 (d, 2 H, CH2,
J 4), 3.93 (s, 3 H, CH3) and 2.84 [d, 6 H, N(CH3)2, J 3.94].
DNA cleavage reactions. The plasmid pSP72 containing the
SV40 early promoter was linearised with Eco R1. The SV40
early promoter sequence is given in Table 1.
To label the DNA at the 5Ј-end, the plasmid was dephos-
phorylated with calf intestinal phosphatase, and subsequently
labelled with [γ32P]ATP and T4 polynucleotide kinase. Alter-
natively the linearised plasmid was 3Ј-end labelled with
[α32P]dATP and Klenow fragment. The labelled DNA was then
cut with Hind III and the 300 bp SV40 early promoter fragment
purified on a 4% agarose gel.
The cleavage reactions were carried out with 30 000 cpm of
labelled SV40 early promoter DNA (approx 3 nM bp) plus 1 µg
µlϪ1 sonicated calf thymus DNA in 20 mM Tris-HCl (pH 7.6). In
a final volume of 10 µl, the DNA was preincubated in the
presence of Tris-HCl (pH 7.6) containing 50 mM NaCl with
increasing concentrations of FTP1, before addition of an equi-
molar concentration of [Fe(NH4)2(SO4)2] followed by addition
of 1 µl DTT (50 mM). The reactions were incubated for 15 min,
stopped with 5 µl thiourea (100 mM), the DNA precipitated in
absolute ethanol, and washed twice in 70% ethanol. The dried
DNA was loaded on 8% polyacrylamide sequencing gels and
cleavage products separated at 55 ЊC at constant power of 60 W.
The gels were fixed in 5% acetic acid, 5% methanol, dried and
exposed to autoradiographic film.
References
1 J. Stubbe and J. W. Kozarich, Chem. Rev., 1987, 87, 1107.
2 A. Kittaka, Y. Sugano, M. Otsuka and M. Ohno, Tetrahedron, 1988,
44, 2821.
3 L. Huang, A. R. Morgan and J. W. Lown, Bioorg. Med. Chem. Lett.,
1993, 3, 1751.
4 I. O. Kady and J. T. Groves, Bioorg. Med. Chem. Lett., 1993, 3, 1376.
5 J. W. Lown, Chemtracts–Organic Chemistry, 1993, 6, 205.
6 P. G. Schultz, J. S. Taylor and P. B. Dervan, J. Am. Chem. Soc., 1982,
104, 6861.
7 R. P. Hertzberg and P. B. Dervan, J. Am. Chem. Soc., 1982, 104, 313.
8 N. Colocci, M. D. Distefano and P. B. Dervan, J. Am. Chem. Soc.,
1993, 115, 4468.
Footprinting experiments
Footprinting with FTP1. Distamycin A (9) or actinomycin D
9 J. C. Dabrowiak, A. A. Stankus and J. Goodisman, in Nucleic Acid
J. Chem. Soc., Perkin Trans. 2, 1997
531

View Article Online
Targeted Drug Design, ed. C. L. Propst and T. J. Perun, Marcel
Dekker, New York, 1992, p. 93.
17 M. C. Fletcher and K. R. Fox, Eur. J. Biochem., 1996, 237, 164.
18 S. Kamitori and F. Takusagawa, J. Am. Chem. Soc., 1994, 116, 4154.
19 M. M. Van Dyke and P. B. Dervan, Science, 1984, 225, 1122.
20 D. Rentzeperis, L. A. Marky, T. J. Dwyer, B. H. Geierstanger,
J. G. Pelton and D. E. Wemmer, Biochemistry, 1995, 34, 2937.
21 W. R. Wilson, B. C. Bagueley, L. P. G. Wakelin and M. J. Waring,
Mol. Pharmacol., 1981, 20, 404.
10 H. R. Drew and A. A. Travers, Cell, 1984, 37, 491.
11 C. Bailly and M. J. Waring, J. Biomol. Struct. Dyn., 1995, 12, 869.
12 L. P. G. Wakelin and W. A. Denny, in Molecular Basis of Specificity
in Nucleic Acid-Drug Interactions, eds. B. Pullman and J. Jortner,
Kluwer, Netherlands, 1990, p. 191.
13 W. C. Still, M. Kahn and A. Mitra, J. Org. Chem., 1978, 43, 2923.
14 J. C. Dabrowiak and J. Goodisman, in Chemistry and Physics of
DNA-Ligand Interactions, ed. N. R. Kallenbach, Adenine Press,
1989, p. 143.
15 K. P. Nightingale and K. R. Fox, Nucl. Acids Res., 1993, 21, 2549.
16 (a) K. R. Fox and M. J. Waring, Nucl. Acids Res., 1984, 12, 9271; (b)
J. Portugal and M. J. Waring, FEBS Lett., 1987, 225, 195; (c)
M. W. Van Dyke, R. P. Hertzberg and P. B. Dervan, Proc. Natl.
Acad. Sci., USA, 1982, 72, 5470.
22 W. A. Denny and L. P. G. Wakelin, Cancer Res., 1986, 46, 1717.
23 K.-X. Chen, N. Gresh and B. Pullman, Nucl. Acids Res., 1988, 16,
3061.
24 S. Neidle, G. O. Webster, B. C. Bagueley and W. A. Denny, Biochem.
Pharmacol., 1986, 35, 3915.
Paper 6/06218K
Received 9th September 1996
Accepted 6th November 1996
532
J. Chem. Soc., Perkin Trans. 2, 1997