Tris-benzimidazole derivatives: Design, synthesis and DNA sequence recognition

Article abstract of DOI:10.1016/S0968-0896(01)00170-5

Two tris-benzimidazole derivatives have been designed and synthesized based on the known structures of the bis-benzimidazole stain Hoechst 33258 complexed to short oligonucleotide duplexes derived from single crystal X-ray studies and from NMR. In both derivatives the phenol group has been replaced by a methoxy-phenyl substituent. Whereas one tris-benzimidazole carries a N-methyl-piperazine at the 6-position, the other one has this group replaced by a 2-amino-pyrrolidine ring. This latter substituent results in stronger DNA binding. The optimized synthesis of the drugs is described. The two tris-benzimidazoles exhibit high AT-base pair (bp) selectivity evident in footprinting experiments which show that five to six base pairs are protected by the tris-benzimidazoles as compared to four to five protected by the bis-benzimidazoles. The tris-benzimidazoles bind well to sequences like 5′-TAAAC, 5′-TTTAC and 5′-TTTAT, but it is also evident that they can bind weakly to sequences such as 5′-TATGTT-3′ where the continuity of an AT stretch is interrupted by a single G·C base pair.

Full text of DOI:10.1016/S0968-0896(01)00170-5

Bioorganic & Medicinal Chemistry 9 (2001) 2905–2919
Tris-benzimidazole Derivatives: Design, Synthesis and DNA
Sequence Recognition
Yu-Hua Ji,a,y Daniel Bur,a Walter Hasler,a Valerie Runtz Schmitt,a Arnulf Dorn,a
Christian Bailly,b,* Michael J. Waring,c Remo Hochstrassera,d and Werner Leupina,e,*
aF. Hoffmann-La Roche Ltd, Pharma Research Preclinical Gene Technologies and Infectious Diseases, CH-4070
Basel, Switzerland
b
´
INSERM Unite 524 et Laboratoire de Pharmacologie Antitumorale du Centre Oscar Lambret, Place de Verdun,
59045 Lille, France
cDepartment of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QJ, UK
dBiocentre of the University of Basel, Department of Biophysical Chemistry, Klingelbergstrasse 70, CH-4056 Basel, Switzerland
eGymnasium Liestal, Abteilung Chemie, Friedensstrasse 20, CH-4410 Liestal, Switzerland
Received 4 October 2000; accepted 10 May 2001
Abstract—Two tris-benzimidazole derivatives have been designed and synthesized based on the known structures of the bis-benzi-
midazole stain Hoechst 33258 complexed to short oligonucleotide duplexes derived from single crystal X-ray studies and from
NMR. In both derivatives the phenol group has been replaced by a methoxy-phenyl substituent. Whereas one tris-benzimidazole
carries a N-methyl-piperazine at the 6-position, the other one has this group replaced by a 2-amino-pyrrolidine ring. This latter
substituent results in stronger DNA binding. The optimized synthesis of the drugs is described. The two tris-benzimidazoles exhibit
high AT-base pair (bp) selectivity evident in footprinting experiments which show that five to six base pairs are protected by the
tris-benzimidazoles as compared to four to five protected by the bis-benzimidazoles. The tris-benzimidazoles bind well to sequences
like 50-TAAAC, 50-TTTAC and 50-TTTAT, but it is also evident that they can bind weakly to sequences such as 50-TATGTT-30
where the continuity of an AT stretch is interrupted by a single Gꢀ C base pair. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
are believed to contribute little to overall binding affi-
nity.21 The hydrophobic transfer of ligand from solution
on to its DNA binding sites is more likely to represent
the main driving force for the complex formation.21
There is no doubt, however, that the width of the minor
groove of the double helix is a major determinant of
Hoechst 33258 binding to particular AT stretches.
Interactions with the walls of the minor groove furnish
the most important contributions to binding of
benzimidazole derivatives.22
The dyes Hoechst 33342 and Hoechst 33258 (Fig. 1) are
frequently used in cytometry to stain chromosomes in
situ.1,2 These two bis-benzimidazole compounds become
brightly fluorescent when they bind to DNA. For a long
time, it has been known that Hoechst 33258 binds spe-
cifically to AT-rich sequences in DNA.3À10 A variety of
NMR and X-ray crystal structures of Hoechst 33258
bound to different oligonucleotide duplexes have been
published.11À19 Collectively, these structural studies
reveal that the drug fits snugly into the minor groove of
the double helix, covering a run of four contiguous Aꢀ T
base pairs (bp). The Hoechst 33258–DNA interaction
appear to be stabilized by several H-bonding and van
der Waals contacts20 but in fact these molecular forces
The high affinity of Hoechst 33258 for AT-rich sequen-
ces in DNA, together with its interesting anti-microbial
properties, have encouraged the design of analogues
better able to target specific sites in DNA. A number of
bis-benzimidazoles equipped with various side chains
have been reported. For example, the replacement of
the terminal piperazine ring with an amidinium, an
imidazoline or a tetrahydropyridinium group reinforces
significantly the affinity of the drug for AT
stretches.23À25 Substituting a small group for the bulky
*Corresponding author. Tel.: +33-320-169218; e-mail: ,
yPresent address: Advanced Medicine Inc., 901 Gateway Blvd, South
San Francisco, CA 94080, USA.
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00170-5

2906
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
was recently employed using a tripyrrole peptide–
Hoechst conjugate.33
The structures determined for Hoechst 33258 bound to
d(CGCAAATTTGCG)2 were used to model the tris-
benzimidazole derivative B which is one unit longer
than the parent molecule (Fig. 1). According to compu-
tational studies with our in-house modeling package
MOLOC34 (check also the MOLOC homepage under
http://www.moloc.ch), compound B would cover six
base pairs and fit snugly into the narrow minor groove
of the A3T3 stretch. All three benzimidazole NHs
would engage in H-bonds to either O2 of thymines or
N3 of adenines. While benzimidazoles bpi and bmi pre-
fer to contact O(2) of T(8) and O(2) of T(7) in the 50
strand, respectively, benzimidazole bph contacts O(2) of
T one base pair upstream in the 30-strand. Other H-
bond patterns for B were examined but they all showed
higher overall energy. Dihedral angles of 31 and 39ꢁ
were measured between benzimidazole units. The 4-N-
methyl-piperazine ring is located between the base pairs
involving T(9) and G(10) while the methoxy-phenyl
moiety covers base pairs four and five. A number of
possible replacements for the basic 4-N-methyl-piper-
azine ring were modelled among which a 3-NH2-pyrro-
lidine might gave the best results, corresponding to
molecule C (Fig. 1). The computer analysis suggested
that the protonated NH2 group could form two H-
bonds to DNA compared to only one with the charged
4-N-methyl-piperazine ring. The smaller five-membered
ring fits deeper into the narrow minor groove than the
more voluminous six-membered ring, thereby decreas-
ing the water accessible hydrophobic surface of the
buried ligand.
Figure 1. Structures of the DNA ligands used in this study.
piperazine results in a narrower groove at the 30 end of
the drug binding site and therefore a smaller perturba-
tion of groove width.23 At the other end of the molecule,
the relocation of the phenolic hydroxyl group from the
para to the meta position (meta-Hoechst) has relatively
little impact on the sequence-selectivity of the interac-
tion with DNA.26,27 The central part of the parent
molecule has also been modified with the aim of
designing compounds capable of recognizing GC-con-
taining sequences. In some cases, the replacement of one
or both of the benzimidazole units with another hetero-
cycle (e.g., benzoxazole, pyridoimidazole, etc.) has pro-
duced GC-tolerant compounds28À31 but not really
produced highly GC-selective analogues.
The tris-benzimidazole B possessing a piperazine term-
inal ring can be directly compared with Hoechst 33342
and Hoechst 33258. By contrast, the DNA-binding
properties of the tris-benzimidazole C have to be exam-
ined relative to a related bis-benzimidazole derivative A
bearing the same aminopyrrolidine terminal group
(Fig. 1). Here, we report the synthesis of the bis- and
tris-benzimidazoles A, B and C. Their capacity to
recognize specific sequences in DNA has been investi-
gated using deoxyribunuclease I (DNase I) footprinting
methodology, with several different DNA restriction
fragments as targets or substrates.
Inspection of NMR and X-ray structures of Hoechst
33258 bound to d(CGCGAATTCGCG)2 revealed that
the drug is positioned with the phenol ring situated
between base pairs involving G(4) and T(5) and the
methyl-piperazine ring between base pairs involving
T(7) and T(8). A different molecular architecture has
been reported when the drug is bound to the dodeca-
nucleotide d(CGCAAATTTGCG)2. In this case,
Hoechst 33258 was found to bind to the 50-ATTTG
sequence in a unique orientation.19,32 In other words,
Hoechst 33258 was displaced by one base pair com-
pared to the structures determined previously for this
drug with the sequence d(CGCGAATTGCGC)2. Con-
sideration of this fact led us to propose elongating
Hoechst 33258 by one benzimidazole unit to yield tris-
benzimidazole derivatives capable of covering more
base pairs, and perhaps to improve affinity for a longer
A/T-rich tract of DNA or to change the sequence spe-
cificity. Another approach to improve selectivity and
binding strength of ligands derived from Hoechst 33258
Results
Chemistry
The synthesis of bis-benzimidazole derivatives was car-
ried out basically according to the method reported by
Lown et al.35 The benzimidazole 3 resulted from the
condensation of 4-substituted o-arylenediamine and
anisaldehyde. The ester function of 3 was converted to
aldehyde 5 by reduction and oxidation (Scheme 1,
method A). The later was coupled with 4-aminosub-
stituted o-arylenediamino 11 or 12 to afford the bis-
benzimidazole derivatives. However, we introduced two
important modifications. The first one concerns the

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2907
synthesis of aldehyde 5. Although its synthesis by
method A in good yield was reported, we always
obtained variable yields of both alcohol 4 and aldehyde
5 especially on the large scale. This was probably due to
the poor solubility of these compounds in the reaction
solvents such as THF and CH2Cl2. Furthermore, during
the reduction of the ester to alcohol 4, the further reduced
by-product 5-methylbenzimidazole was obtained with
prolonged reaction time under reflux. The reduction of
N-methoxy methylamide to aldehyde has been repor-
ted.36 Therefore, the synthetic route for aldehyde 5 was
improved as follows (Scheme 1, method B): 3,4-dini-
trobenzoic acid was transformed to amide 6, which was
hydrogenated to the corresponding diamine and subse-
quently condensed with aldehyde to yield benzimidazole
7. Treatment of compound 7 with LiAlH4 in THF
afforded aldehyde 5 in high yield.
always proceed sufficiently smoothly; it required rela-
tively high temperatures together with long reaction
times, and the desired benzimidazole derivative was
obtained in variable yields, and in some cases the by-
products such as Schiff base and dihydrobenzimidazole
derivative were also isolated. The daily use of nitro-
benzene as a solvent was also not convenient in the
laboratory. The dehydrogenating capacity of bisulfite
has been demonstrated in the condensation of 1,8-dia-
minonaphthalene and aldehyde to 2-substituted 1H-
pyrimidines.39,40 However, there is no previous report
on the synthesis of bis-benzimidazole derivatives
utilizing bisulfite as an oxidizing reagent. We therefore
explored the possibility of constructing benzimidazole
unit from o-arylenediamine and aldehyde in the pre-
sence of bisulfite. The condensation was carried out in
ethanol. After hydrogenation of o-dinitroarylene to the
corresponding diamine an aqueous solution of sodium
pyrosulfite, which generated bisulfite in situ, was intro-
duced directly in to the reaction mixture. Following
addition of the appropriate aldehyde, the mixture was
heated under reflux for 4 h to give the desired mono, bis-
or tris-benzimidazole derivatives, respectively, in high
yield. This procedure is easy to handle and offers the
possibility of accomplishing the hydrogenation of dini-
tro to diamine and the subsequent condensation
with aldehyde in a one-pot reaction. Therefore, bis-
benzimidazole aldehyde derivative 10 was easily
prepared by using these modified procedures
(Scheme 2). 3,4-Diaminobenzoic acid was coupled with
The second modification concerns the benzimidazole
ring formation. Heating of an equimolar o-arylenediamine
and an aldehyde in nitrobenzene has usually been
adopted as a convenient and widely applicable route for
the construction of the benzimidazole unit. The reaction
consists of two steps: the Schiff base formation and the
oxidative cyclization. Nitrobenzene is used as solvent as
well as oxidant. This method has many advantages over
the condensation of o-arylenediamine and arylcar-
boxylic acid or their derivatives such as imidate ester,
chloride and anhydrides37,38 in strongly acidic condi-
tions. However, using this method the reaction did not
Scheme 1. (A) (a) HCl/EtOH; (b) 4-methoxybenzaldehyde, PhNO2, 140 ꢁC; (c) LiAlH4, THF/dioxane, reflux; (d) PCC, pyridine/CH2Cl2 or MnO2,
DMF/CH2Cl2; (B) (a) SOCl2, 80 ꢁC; (b) CH3ONHCH3ꢀ HCl, Py, CH2Cl2, rt; (c) H2, 5% Pd/C, EtOH, rt; (d) 4-methoxybenzaldehyde, PhNO2,
140 ꢁC; (e) 4-methoxybenzaldehyde, Na2S2O5, EtOH/H2O, reflux; (f) LiAlH4, THF, 0 ꢁC!rt.

2908
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
monobenzimidazole aldehyde 5 in the presence of
bisulfite in ethanol and water to give bisbenzimidazole
carboxylic acid derivative 8. Then the acid function was
converted to amide 9, which was subsequently reduced
to bisbenzimidazole aldehyde 10.
tris-benzimidazoles compared to the bis-benzimidazoles.
As represented in Figure 2, at a drug–DNA (P) ratio of
0.25, a ÁTm value (Tm complexÀTm DNA) of 12 ꢁC was
measured with Hoechst 33258 whereas the ÁTm reaches
24 ꢁC with compound B. A much larger ÁTm value was
also measured with the other tris-benzimidazole deriva-
tive C compared to the corresponding bis-benzimida-
zole A (ÁTm=14 and 25 ꢁC for A and C, respectively).
We can therefore conclude that the additional benzimi-
dazole unit reinforces the interaction of the drugs with
DNA. These Tm data are in agreement with the foot-
printing data reported below.
In conclusion, the convergent synthesis of titled bis- and
tris-benzimidazoles is as follows: 5-chloro-2-nitroaniline
was treated with the appropriate cyclic secondary amine
in DMA to yield the 5-aminosubstituted nitroaniline.
Catalytic hydrogenation of the latter to the corre-
sponding 4-substituted o-phenylenediamine was fol-
lowed by condensation with aldehyde 5 or 10, in the
presence of aqueous sodium pyrosulfite, to furnish bis-
and tris-benzimidazoles, compounds 1b, 2a, 2b,
respectively (Scheme 2).
DNA binding mode
To characterize the mode of binding to DNA, linear
dichroism (LD) spectra of the bis-benzimidazole A
(known as a minor groove binder) and the two tris-
benzimidazoles B and C were measured in the presence
of calf thymus DNA, using a simple, easy to handle and
inexpensive Couette cell for flow oriented linear dichro-
ism.41 All three compounds exhibit a spectral region (at
l>330 nm) where DNA does not absorb the light thus
allowing the study of the mode of binding by this
method. The absorption and LD spectra of compounds
B and C complexed with calf thymus DNA are presented
DNA binding strength
We measured the ability of the bis- and tris-benzimida-
zole derivatives to alter the thermal denaturation profile
of calf thymus DNA. Under the experimental condi-
tions used (16 mM Na+), the helix-to-coil transition
occurs at 60 ꢁC in the absence of drugs. The Tm is shif-
ted to higher temperature with all five drugs but
the extent of stabilization differs significantly for the
ꢁ
ꢁ
.
Scheme 2. (A) (a) Na2S2O5, EtOH/H2O, reflux; (b) SOCl2, 80 C; (c) CH3ONHCH3 HCl, Py, CH2Cl2, rt; (d) LiAlH4, THF, 0 C ! rt; (B) (a) 3-
BocNH-pyrolidine or 4-N-methylpiperazine, K2CO3, DMF, 90 ꢁC; (b) H2, 5% Pd/C, EtOH, rt; (c) Na2S2O5, EtOH/H2O, reflux; (d) only for com-
pounds 1b and 2b: TFA/CH2Cl2.

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2909
in Figure 3. In both cases, the LD is negative in the
DNA absorption band centered at 260 nm, indicative an
orientation of the base pairs perpendicular to the long
axis of the biopolymer. In contrast, positive LD signals
in the absorption bands of the ligands at wavelengths
higher than 330 nm (where DNA does not absorb) are
observed with compounds B and C (Fig. 3), as well as
with compound A (spectrum not shown). These spectra
indicate that in all three cases the orientation of the bis-
or tris-benzimidazole chromophores is colinear to the
orientation of the DNA helix. In other words, com-
pounds A, B and C do not intercalate between the DNA
base pairs but bind to one of the grooves of the double
stranded DNA helix, most likely the minor groove as is
known to be the case with Hoechst 33258 and related
bis-benzimidazoles including compound A. The experi-
mental finding of groove binding of the two tris-benzimi-
dazoles B and C has in the meantime been verified by X-
ray studies of C complexed to the oligonucleotide
27
duplexes d(CGCAAATTTGCG)2 and to the non-self-
complementary dodecamer DNA duplex d(CG[5BrC]A-
TAT-TTGCG)ꢀ d(CGCAAATATGCG).42 Our observa-
tions differ from those reported43 with the tris-
benzimidazole derivative 5PTB which exhibits both
intercalative and minor groove binding properties. In
our experiments, the two tris-benzimidazole derivatives
behave as pure groove binders.
DNA sequence recognition
DNase I footprinting experiments were carried out
using three DNA restriction fragments chosen for their
diverse arrangements of base pairs. With each fragment,
the products of digestion by DNase I in the absence and
presence of the test drugs were resolved by poly-
acrylamide gel electrophoresis. Typical autoradiograms
are illustrated in Figures 4 and 5. Strong differences
between the control lanes and those for the tested drugs
can easily be seen. Areas of decreased intensity (i.e.,
footprints) in ligand-containing lanes relative to ligand-
free lanes reflect cleavage inhibition due to ligand bound
to specific nucleotide sequences, and densitometric ana-
lysis of the different patterns permits estimation of the
Figure 2. Variation of the melting temperatures (ÁTm) for calf thymus
DNA in the presence of the bis- and tris-benzimidazoles. Tm mea-
surements were performed in BPE buffer pH 7.1 (6 mM Na2HPO4,
2 mM Na2H2PO4, 1 mM EDTA) using 20 mM DNA and 5 mM drug, in
1 mL quartz cuvettes at 260 nm with a heating rate of 1 ꢁC/min.
Figure 3. Absorption (upper panel) and linear dichroism (lower panel) spectra of compounds B and C complexed to calf thymus DNA (average
length approx. 2500 base pairs) in 50 mM phosphate buffer of pH 7.0, 100 mM NaCl and 0.1% NaN3.

2910
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
location and relative strength of binding at particular
DNA sites. Figure 6 shows a selection of differential
cleavage plots determined with the three DNA frag-
ments in the presence of the bis-benzimidazole A and
the tris-benzimidazole derivatives B and C. Experiments
were also performed with Hoechst 33258 and Hoechst
33342 for comparative purposes. The dips in these plots
(negative values) indicate sites of protection from
DNase I cleavage, whereas peaks (positive values) indi-
cate regions of drug-induced enhancement of nuclease
cleavage. We will consider the results obtained with
each fragment in turn.
tical to those previously reported with this piece of
DNA. The most significant observation is that, con-
sistent with the finding on the 117-mer DNA fragment,
the sequences protected from DNase I cleavage by the
tris-benzimidazoles B and C are 1–2 bp longer than
those protected by Hoechst 33258. A conspicuous
example occurs in the A-tract at positions 48–52. The 50-
most adenine residue of the A run (i.e., position A52) is
protected from nuclease attack by compounds B and C
whereas this adenine remains fully accessible to the
enzyme in the presence of Hoechst 33258 and the rela-
ted bis-benzimidazole derivatives. The same observation
can be made for the adenine at position 26 (i.e., protec-
tion by B and C but unimpeded cleavage with Hoechst
33258). Therefore, this second set of footprinting
experiments reinforces the previous conclusions that (i)
the test compounds are AT-specific groove binders and
(ii) the DNase I protected sites are slightly longer with
the tris-benzimidazoles compared to the bis-benzimida-
zoles. All the experiments agree that the three benzimi-
dazole units of compounds B and C bind firmly to DNA
and engage in specific contact with Aꢀ T base pairs.
The 117 bp DNA fragment (Fig. 4, panel Band Fig. 6,
panel a). Four major sites of drug protection can be
discerned around nucleotide positions 25, 43, 64 and 85.
These protected zones each extend for about eight
nucleotides. All four are situated in AT-rich sequences
of the DNA. Adjoining these protected sites are regions
where the rate of DNase I cutting is significantly
enhanced relative to the control. These regions of rela-
tively enhanced cleavage always occur in GC-rich
sequences as for example at positions 34–38 and 71–74
in this DNA. The densitometric analysis was necessarily
limited to the regions of the gel where the bands are well
separated: at first sight, the patterns of enhancement
and protection between nucleotide positions 26 and 78
appear virtually identical for the three test compounds
A, B and C and for the two control drugs Hoechst
33258 and Hoechst 33342. However, closer inspection
of the gels reveals an interesting difference between the
bis and the tris-benzimidazole compounds. The sites
protected from DNase I cleavage by the tris-benzimi-
dazoles B and C are often 1–2 bp longer than the cor-
responding sites detected with the bis-benzimidazoles.
This is clearly visible between positions 40 and 46 of this
DNA fragment, as shown in Figure 7. At this site, the
sequence 30-CAAAA is protected from cleavage by
DNaseI in the presence of Hoechst 33258, Hoechst
33342 and compound A. With the tris-benzimidazoles B
and C, it is obvious that the binding site is one base pair
longer and encompasses the sequence 30-CAAAAT. The
inference is clear that the longer footprints obtained
with tris-benzimidazoles compared to the bis-benzimi-
dazoles reflect the binding specificity of the third benzi-
midazole group present in compounds B and C. It is
also worth noting that the tris-benzimidazole C seems to
produce stronger protection from DNase I cutting than
does Hoechst 33258 or the other test drugs. This can be
seen in the differential cleavage plots in Figure 6a as
well as on the gel in Figure 4a around position 85 where
the footprint produced by compound C is substantially
more pronounced than those produced by the other
drugs.
The 178bp DNA fragment (Fig. 5 and Fig. 6, panel c).
With this fragment the footprinting experiments were
carried out using a 5-fold lower drug concentration than
the experiments with the 117-mer and the tyrT DNA.
The differential cleavage plots obtained for Hoechst
33258 and the bis-benzimidazole derivative A super-
impose quite well, showing that compound A selects
AT-rich sites very similar to those bound by Hoechst
33258. The replacement of the piperidine terminal group
of Hoechst with a pyrrolidine group evidently has little
effect on DNA sequence recognition. Interestingly,
under the conditions used here we can observe sig-
nificant differences between the bis- and the tris-benzi-
midazole compounds. Both the gel and the differential
cleavage plots concur that several sequences are pro-
tected from nuclease attack by the tris-benzimidazoles B
and C but not by the bis-benzimidazole derivatives. This
is particularly noticeable at the AT-rich sequence
between nucleotide positions 80 and 91 where, allowing
for the usual 30-shift by 2 bp due to the bias introduced
by the enzyme cutting in the minor groove, we can
locate binding sites for compounds B and C at the
sequences 30-CATTT 87–91 and 30-TATTT 80–84. In
addition, the existence of two more binding sites can be
discerned from the gaps in the gels near positions 105
and 120 (Fig. 5). They lie beyond the range accessible to
densitometry but are clearly visible at the top of the
autoradiograph. At higher concentrations (ꢂ10 mM)
the bis-benzimidazole compounds also produce foot-
prints at the same sequences (data not shown). These
observations again suggest that the tris-benzimidazoles
are endowed with a higher affinity for DNA compared
to the bis-benzimidazole compounds. Interesting also is
the observation that a weak site exists around the
sequence 50-TATGTT 62–67, detectable only with the
tris-benzimidazoles, indicating that drugs B and C can
accept a sequence containing a Gꢀ C base pair as a sec-
ondary binding site, presumably with lower affinity. As
observed with the 117-mer and the tyrT fragment,
regions of the 178 bp DNA can be seen where nuclease
The tyrT DNA fragment (Fig. 4, panel A and Fig. 6,
panel b). As with the 117-mer fragment, the sites pro-
tected by the test drugs A, B and C coincide with those
protected by the control drugs Hoechst 33342 and
Hoechst 33258. All produce strong protection around
positions 29, 44 and 82 of the DNA. The tyrT fragment
has been used previously to map the binding sites of
Hoechst 33258,3 and the sites observed here are iden-

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2911
Figure 4. DNase I footprinting of Hoechst 33258, Hoechst 33342 and compounds A, B, and C (10 mM each) on the 160 bp EcoRI–AvaI tyrT frag-
ment cut out from plasmid pKMÁ-98 (panel A) and the EcoRI–PvuII 117 bp fragment cut out from plasmid pBS (panel B). In both cases, the duplex
was 30-end labeled at the EcoRI site with [a-32P]dATP in the presence of AMV reverse transcriptase. The cleavage products of the DNase I digestion
were resolved on an 8% polyacrylamide gel containing 8 M urea. Tracks labeled Control contained no drug. Lanes labeled G+A represent formic
acid-piperidine markers track specific for purines. The sequences on the left side indicate the sites protected from DNase I cleavage. Numbers at the
right side of the gels refer to the numbering scheme used in Figure 6 (panels a and b).

2912
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
cutting is enhanced in the presence of the test drugs
compared with the control. Good examples (Fig. 6c)
occur at positions 44–47 (30-GGCG) and 71–75 (30-
CGGCC), that is in regions containing contiguous Gꢀ C
base pairs. There is no doubt that binding of these drugs
to pure GC sequences is strongly disfavored.
Discussion
Sequence specific recognition of the minor groove of
DNA by low-molecular weight synthetic ligands is a
fast-growing field of research.44,45 The ligands of inter-
est are derived from different classes of chemical com-
pounds, and generally represent elongated molecules
containing repeating chemical entities capable of form-
ing hydrogen bonds with building blocks of the DNA,
acting both as hydrogen bond donors as well as hydro-
gen bond acceptors. Based on the high propensity of
Hoechst 33258 to bind to AT-rich sequences, particu-
larly those containing four contiguous Aꢀ T base pairs,46
we decided to add one more benzimidazole unit result-
ing in tris-benzimidazole derivatives that should cover
longer sites on DNA. According to modelling studies,
the incorporation of a third benzimidazole unit should
in principle allow the recognition of a specific DNA
sequence that is about 6 bp long. The footprinting
experiments presented here provide clear data that
vindicate the hypothesis fully. The results in Figure 5
show clearly that the AT binding sites occupied by
compounds B and C are 1–2 bp longer than those
occupied by their bis-benzimidazole counterparts. The
Tm and footprinting experiments concur that the tris-
benzimidazoles B and C interact more tightly with
DNA than do the equivalent bis-benzimidazoles. All
this is consistent with the predictions of our modelling
studies.
There exists a general consensus that DNase I detects
sequence–specific interactions between drugs and DNA
with high sensitivity but tends to overestimate the size
of the binding sites. This view has its origins in foot-
printing investigations where the size of drug binding
sites detected with DNase I was compared to those
detected with other probes, mostly metal chelates.47,48
The elucidation of the crystal structure of a DNase I–
DNA complex49,50 has substantiated this belief: in
binding to DNA, the enzyme engages in contacts with
two phosphates on either side of the cleaved bond and
two phosphates on the other strand across the minor
groove, opposite the phosphates contacted on the 50 side
of the cleaved bond. Nevertheless, the results reported
here show that DNase I can sensitively detect the drug–
DNA interaction and also accurately define the region
that is contacted by the drug. From a technical point of
view, the footprinting data reported here confirm our
previous observations51 that the supposed inability of
DNase I to depict the exact position of ligand binding is
not always justified. If it is optimally used (under single
hit kinetic conditions), there is no doubt that DNase I
can provide a very sensitive and accurate estimation of
drug binding site sizes, at least for minor groove-binders
as studied here.
Figure 5. DNase I footprinting of Hoechst 33258, Hoechst 33342 and
compounds A, B, and C (2 mM each) on the 178 bp EcoRI–PvuII
fragment cut out from plasmid pUC12. Numbers at the side of the gels
refer to the numbering scheme used in Figure 6 (panel c). Other details
as for Figure 4.

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2913
The tris-benzimidazole B bearing a piperazine terminal
group has not been described before, but a similar
compound containing an aminopyrrolidine terminal
group (tris-benzimidazole C) has been the subject of a
crystallographic study: Neidle and co-workers solved
the structure of a complex formed between the self-
complementary dodecamer d(CGCAAATTTGCG)2
and compound C, named TRIBIZ.27 In the crystal
Figure 6. Differential cleavage plots comparing the susceptibility of the 117-mer (a), 160-mer tyrT (b), and 178-mer (c) fragments to DNase I cutting
in the presence of the test drugs A, B and C and Hoechst 33258. The plots were measured by densitometry using phosphorimages of the gels shown
in Figures 4 and 5. Negative differential cleavage corresponds to a ligand-protected site and positive differential cleavage occurs where the nucleo-
lytic cutting is enhanced. Vertical scales are in units of ln(fa)Àln(fc), where fa is the fractional cleavage at any bond in the presence of the drug and fc
is the fractional cleavage of the same bond in the control, given a similar extent of digestion in both cases. Each line drawn represents a three-bond
running average of individual data points, calculated by averaging the value of ln(fa)Àln(fc) at any bond with those of its two nearest neighbors.

2914
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
Figure 7. Portions of autoradiographs showing DNase I footprinting of the drugs on the 30-end labeled 117 bp EcoRI-PvuII restriction fragment.
Each pair of lanes corresponds to cleavage by the enzyme in the presence (left) and absence (right) of the test drug. The third lane in the panel
corresponding to Hoechst 33342 shows the G+A sequencing markers. The scale on the right side of this panel refers to the numbering of the DNA
fragment as represented in Figure 6. Other details as for Figure 4.
structure, the drug C occupies the central AAATTT
region of the duplex. The recognition site for the tris-
benzimidazole drug C is best described by the sequence
50-CAAATTTG whereas the bis-benzimidazole Hoechst
33258 interacts with the sequence 50-ATTTG in its
complex with the same dodecamer.52 It is gratifying that
our footprinting results agree with these data, for
example showing that the binding sites for the tris-ben-
zimidazole B are longer than those for the bis-benzimi-
dazole. But the DNase I experiments reveal additionally
that the ligand can accommodate binding sites that do
not merely contain six contiguous Aꢀ T base pairs. For
example, the sequence 50-TTTACACTTTAT around
position 85 on the 178-mer fragment (Fig. 6c) provides a
privileged binding site for the tris-benzimidazole. The
differential cleavage plots in Figure 6 show that the
binding sites for the drugs A, B and C can contain a
minimum of 4 Aꢀ T base pairs.
coated with silica gel 60 F254 (Merck) were used and
detected by UV (254 nm), by treatment with con-
centrated H2SO4 and then 5% vaniline solution in eth-
1
anol followed by heating. H NMR spectra were taken
on Bruker AC 250 (250 MHz) or Bruker AM 400
(400 MHz), with TMS as internal reference and are
expressed in d units (parts per million). Mass pectra
(MS) were recorded using a MS901 AEI spectrometer.
2-(4-Methoxy-phenyl)-benzimidazole-5-carboxylic acid
ethyl ester (3). A mixture of 1,2-diaminobenzoic acid
ethyl ester (1 g, 5.5 mmol) and anisaldehyde (0.67 mL,
5.5 mmol) in 25 mL of nitrobenzene was heated at
140 ꢁC. Thirty-six hours later, an additional portion of
anisaldehyde (0.2 mL, 1.6 mmol) was added and the
heating was continued for another 5 h. After the
removal of nitrobenzene under reduced pressure, the
crude compound was purified by flash chromatography
over silica gel (AcOEt/hexane: 4/6–5/5). The desired
compound 3 (1.06 g) was obtained in 65% yield. 1H
NMR (CDCl3): d 1.40 (t, 3H, CH3, J=7.1 Hz), 3.84 (s,
3H, OCH3), 4.40 (q, 2H, CH2, J=7.1 Hz), 6.95 (d, 2H,
ArH, J=8.9 Hz), 7.61 (d, 1H, ArH, J=8.4 Hz), 7.98
(dd, 1H, ArH, J1=8.4 Hz, J2=1.6 Hz), 8.05 (d, 2H,
ArH, J=8.8 Hz, 8.83 (d, 1H, ArH, J=1.6 Hz). MS:
296 (100, M+). Anal. calcd for C17H26N2O3 (296.326):
C 68.91, H 5.44, N 9.45; found: C 69.13, H 5.59, N
8.87.
Experimental
Chemistry
All the chemicals were purchased from Fluka or
Aldrich. Reagent grade solvents, purchased from Fluka,
were used without further purification. Evaporation
means removal of solvent by using a Buchi rotary eva-
porator at 35–50 ꢁC in vacuo. Normal phase silica gel
used for flash chromatography was Kieselgel-60 (230–
400 mesh) or Florisil (60–100 mesh). Reverse-phase silica
gel used was Opti-up RP 18. For HPLC, Lichrospher1
60 RP-select B (5 mm) column was used. TLC plates
[2-(4-Methoxy-phenyl)-100H00-benzimidazole-5-yl]-methanol
(4). Under argon, to a solution of compound 3
(297 mg, 1 mmol) in dry THF (20 mL), LiAlH4 (140 mg,
3.7 mmol) was added and the mixture was stirred for

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2915
8 h. Another portion of LiAlH4 (116 mg, 3 mmol) was
added and the reaction mixture was heated to reflux for
overnight. After cooling to room temperature, the reac-
tion mixture was poured carefully to a NH4Cl saturated
solution (5 mL) and further acidified to pH 4 with 4 N
HCl. After introducing 5 mL of MeOH and Speedex,
the mixture was stirred for another 30 min and filtered.
The filtrate was concentrated under vaccum and the
obtained residue was purified by chromatography over
silica gel (CH2Cl2/CH3OH: 95/5), to afford the title
compound 4 (171 mg, 67%). 1H NMR (MeOH-d4): 3.87
(s, 3H, OCH3), 4.72 (S, 2H, HOCH2), 7.08 (d, 2H, ArH,
J=8.9 Hz), 7.25 (dd, 1H, ArH, J1=8.4 Hz, J2=1.6 Hz),
7.55 (d, 1H, ArH, J=8.4 Hz), 7.60 (d, 1H, ArH,
J=1.6 Hz), 8.02 (d, 1H, J=8.9 Hz). MS: 254 (100, M+),
237 (40, M+-OH), 225 (46, M+ÀCH2OH).
[2-(4-Methoxy-phenyl)-100H00-benzimidazole-5-carb-alde-
hyde (5). Method A. A mixture of alcohol 4 (800 mg,
3.2 mmol) and active manganese dioxide (1600 mg,
18.4 mmol) in CH2Cl2 (50 mL) and DMF (10 mL) was
stirred at 60 ꢁC for 3 days. After cooling to room tem-
perature, the mixture was filtered through a bed of
Speedex. The filtrate was concentrated under reduced
pressure and the resulting residue was purified by chro-
matography over silica gel (CH2Cl2/CH3OH: 100/0–90/
10). The aldehyde (720 mg, 93%) was obtained as a
beige solid. The alcohol 4 could be also transformed to
aldehyde 5 by oxidation with three equivalents of pyr-
idium chlorochromate in CH2Cl2 at room temperature,
in about 50% yield.
(dd, 1H, J1=8.5 Hz, J2=1.9 Hz), 8.71 (d, 1H, ArH,
J=1.9 Hz). Anal. calcd for C7H3ClN2O5 (230.56): C
36.47, H 1.31, N 12.15; found: C 36.53, H 1.49, N 12.11.
The above acid chloride (20 g, 86 mmol) was dissolved
in CH2Cl2 (150 mL) and the solution was cooled to 0 ꢁC.
After addition of pyridine (21 mL, 260 mmol) and N,O-
dimethylhydroxylamine
hydrochloride
(12.6 g,
129 mmol), the ice-bath was removed and the mixture
was stirred overnight at room temperature. The reaction
mixture was diluted with 200 mL of CH2Cl2 and the
organic layer was washed twice with brine, dried over
Na2SO4 and concentrated. The crude compound was
purified by flash chromatography over silical gel
(AcOEt/hexane: 1/1–8/2), to fournish the title com-
1
pound 6 (20.3 g, 92%) as an orange solid. H NMR
(CDCl3): 3.43 (s, 3H, CH3), 3.59 (S, 3H, CH3), 7.96 (d,
1H, ArH, J=8.4 Hz), 8.11 (dd, 1H, ArH, J1=8.4 Hz,
J2=1.9 Hz), 8.31 (d, 1H, ArH, J=1.9 Hz). MS: 255 (8,
M+), 195 (100, M+ÀMeNOMe). Anal. calcd for
C9H9N3O6 (255.186): C 42.36, H 3.56, N 16.47; found:
C 42.33, H 3.63, N 16.42.
[2-(4-methoxy-phenyl)-100H00-benzimidazole-5-carboxylic
acid methoxy-methylamide (7)
Method I: Coupling in nitrobenzene. A solution of
methoxy methyl 3,4-dinitrobenzoic amide
6 (2 g,
7.8 mmol) was hydrogenated over 5% Pd/C (800 mg) in
EtOH (200 mL) at room temperature for 3 h, to give N-
methoxy N-methyl 3,4-diaminobenzoic amide. After
hydrogenation, 50 mL of nitrobenzene was added and
EtOH was removed under reduced pressure. After
addition of another 50 mL of nitrobenzene and ani-
saldehyde (0.95 mL, 7.8 mmol), the mixture was stirred
at 140 ꢁC for 24 h, then cooled to room temperature and
filtered to remove the catalyst. The filtrate was con-
centrated under reduced pressure and resulting residue
was loaded on a column of silica gel, which was eluted
with AcOEt/MeOH (100/0–98/2) and give the desired
benzimidazole 7 (1.7 g) in 69% yield.
Method B. To a solution of compound 7 (5.9 g,
19 mmol) in 200 mL of THF/Et2O (3/1), LiAlH4 (2.1 g,
57 mmol) was added in small portion at À70 ꢁC. The
stirring was continued for 6 h and the reaction tem-
perature was kept below À20 ꢁC. The reaction was
quenched by carefully introducing of AcOEt (100 mL)
to destroy the excess of LiAlH4 and the temperature was
allowed to raise to room temperature. The mixture was
then acidified to pHꢃ5 with 12.5 N HCl/MeOH solu-
tion, followed by addition of NH4Cl saturated solution
(30 mL). After 15 min stirring, Speedex powder was
added to the mixture and the stirring was continued for
another 30 min. Finally, the mixture was filtered and the
filtrate was concentrated under reduced pressure. The
residue was purified by flash chromatography over silica
gel to give the desired aldehyde 5 (4.25 g, 88%). 1H
NMR (DMSO-d6): 3.86 (s, 3H, OCH3), 7.15 (d, 2H,
ArH, J=8.9 Hz), 7.75 (dd, d, 2H, ArH), 8.14 (d, 1H,
ArH, J=8.4 Hz), 8.17 (d, 2H, ArH, J=8.9 Hz). MS: 252
(100, M+). Anal. calcd for C15H12N2O2 (252.273): C
71.42, H 4.79, N 11.10; found: C 71.22, H 4.85, N 10.90.
Method II: Coupling with Na2S2O5. After hydrogena-
tion, a solution of Na2S2O5 (770 mg, 4 mmol) in H2O
(1.6 mL) was introduced directly to the mixture, fol-
lowed by addition of anisaldehyde (0.98 mL, 8 mmol).
The resulting mixture was stirred at reflux for 4 h, then
cooled to room temperature and filtered through a bed
of Speedex. The filtrate was concentrate under reduced
pressure and the obtained residue was purified by chro-
matography to furnish 1.95 g of the title compound 7 in
1
80% yield. H NMR (CDCl3): 3.40 (s, 3H, CH3), 3.57
(S, 3H, CH3), 3.82 (s, 3H, OCH3), 6.91 (d, 2H, ArH,
J=8.9 Hz), 7.56 (m, 2H, ArH), 7.93 (s, broad, 1H,
ArH), 8.06 (d, 2H, ArH, J=8.9 Hz). MS: 311 (5, M+),
251 (100, M+ÀN(OCH3)CH3). Anal. calcd for
C17H17N3O3 (311.34): C 65.58, H 5.50, N 13.50; found:
C 65.40, H 5.72, N 13.71.
N-Methoxy-N-methyl-3,4-dinitrobenzamide (6). A mix-
ture of 3,4-dinitrobenzoic acid (20 g, 94 mmol) and
SOCl2 (55 mL) was heated at 80 ꢁC for 4 h and then
concentrated under reduced pressure. Dry toluene was
added to the resulting residue and evaporated to dry-
ness, such procedure was repeated three times. Finally,
the acid chloride was obtained as a yellow solid (21.7 g).
1H NMR (CDCl3): 8.06 (d, 1H, ArH, J=8.4 Hz), 8.52
20-(4-Methoxy-phenyl)-[2,50] bisbenzoimidazole-5-carb-
oxylic acid (8). To a mixture of 3,4-diaminobenzoic
acid (2.3 g, 15 mmol) and aldehyde 5 (3.8 g, 15 mmol) in
EtOH (250 mL), 50 mL of solution of sodium pyro-

2916
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
sulfite (1.5 g, 8 mmol) was added and the mixture was
heated to reflux overnight. After cooling to room tem-
perature, 100 mL of H2O was added to the reaction
mixture and resulting precipitate was filtered. The
obtained solid was recrystalized in MeOH to give the
title compound 8 (4.1 g, 71%) as brown solid. 1H NMR
(DMSO-d6): 3.89 (s, 3H, CH3), 7.24 (d, 2H, ArH,
J=8.9 Hz), 7.78 (d, 1H, ArH, J=8.5 Hz), 7.90 (d, 1H,
ArH, J=9.0 Hz), 7.95 (dd, 1H, ArH, J1=8.5 Hz,
J2<2 Hz), 8.22 (d, 2H, ArH, J=8.9 Hz), 8.24 (dd and d,
2H, ArH), 8.49 (d, 1H, ArH, J<2 Hz). MS (ISP): 385
(100, MH+).
was filtered off to give the title compound 11 (38 g,
1
82%) as a yellow solid. H NMR (CDCl3): 2.21 (s, 3H,
CH3), 2.38–2.42 (m, 4H, 2CH2), 3.29–3.34 (m, 4H,
2CH2), 6.21 (d, 1H, ArH, J=2.7 Hz), 6.39 (dd, 1H,
ArH, J1=2.7 Hz, J2=9.8 Hz), 6.27 (broad, 2H, NH2),
7.80 (d, 1H, ArH, J=9.8 Hz). MS: 236 (46, M+), 221
(10, MÀCH3), 165 (16, MÀC4H9N), 70 (100, C4H8N).
Anal. calcd for C11H16N4O2 (236.275): C 55.87, H 6.84,
N 23.71; found: C 55.32, H 6.88, N 23.51.
[1-(3-amino-4-nitro-phenyl)-pyrrolidin-3-yl]-carbamic acid
‘tert’-butyl ester (12). A mixture of 5-chloro-2-nitroani-
line (7.8 g, 46 mmol), 3-tert-butoxycarbonylamino-pyr-
rolidine (8.5 g, 45.6 mol) and potassium carbonate
(6.4 g, 46.3 mol) in DMF (200 mL) was heated at reflux
for 3 h and then cooled to room temperature. After
addition of 200 mL of H2O, the mixture was extracted
with AcOEt (3ꢄ150 mL) and the organic layer was
washed with brine, dried over anhydrous Na2SO4 and
concentrated. The title compound 12 (13.3 g, 91%) was
provided after purification by flash chromatography
over silica gel (CH2Cl2/MeOH), as yellow solid. 1H
NMR (CDCl3): 1.35 (s, 9H, (CH3)3), 2.01, 2.31 (2m, 2H,
CH2), 3.24 (dd, 1H, CH2N, J2=4 Hz, J2=11.6 Hz), 3.47
(m, 1H, CH2N), 3.66 (dd, 1H, J1=5.2 Hz, J2=11.6 Hz),
4.36 (m, 1H, CH), 4.71 (broad, 1H, NH), 5.60 (d, 1H,
ArH, J=2.4 Hz), 6.02 (dd, 1H, ArH, J1=2.4 Hz,
J2=9.5 Hz), 6.18 (broad, 2H, NH2), 8.02 (d, 1H, ArH,
J=9.5 Hz). MS: 322 (10, M+), 249 (10, M+ÀtBuO),
205 (100, M+ÀBocNH2). Anal. calcd for C15H22N4O4
(322.365): C 55.89, H 6.88, N 17.38; found: C 55.88, H
6.94, N 16.95.
20-(4-Methoxy-phenyl)-[2,50] bis-benzoimidazole-5-carb-
oxylic acid methoxy-methyl amide (9). A mixture of 8
(1.1 g, 2.2 mmol) and SOCl2 (1.1 mL, 15 mmol) was
refluxed for 4 h, then concentrated under reduced pres-
sure and further coevaporated twice with toluene
(2ꢄ100 mL). After drying under vaccuum, the obtained
residue was dissolved in CH2Cl2 (100 mL) and N,O-
dimethylhydroxyamine hydrochloride (0.44 g, 4.5 mmol)
and pyridine (0.8 mL, 10 mmol) were added. The mix-
ture was stirred at room temperature for 24 h and the
usual work up was followed. The crude product was
purified by chromatography over silica gel with CH2Cl2/
MeOH as eluent, to give title compound 9 (560 mg,
1
47%) as beige solid (47%). H NMR (DMSO-d6): 3.31
(s, 3H, NCH3), 3.59 (s, 3H, NOCH3), 3.86 (s, 3H,
OCH3), 7.15 (d, 2H, ArH, J=8.8 Hz), 7.49 (dd, 1H,
ArH, J1=8.3 Hz, J2<2 Hz), 7.63 (d, 1H, ArH,
J=8.3 Hz), 7.72 (d, 1H, ArH, J=8.5 Hz), 7.88 (d, 1H,
ArH, J<2 Hz), 8.08 (dd, 1H, ArH, J1=8.5 Hz,
J2<2 Hz), 8.17 (d, 2H, ArH, J=8.8 Hz), 8.38 (d, 1H,
ArH, J<2 Hz), 13.08 (broad, 2H, NH). MS (ISP): 428
(100, M+H+).
1-{20-(4-methoxy-phenyl)-[2,50]bibenzimidazol-5-yl}-pyr-
rolidin-3-ylamine (1b). A solution of compound 12
(1162 g, 3.6 mmol) in EtOH (200 mL) was hydrogenated
over 5% Pd/C (600 mg) at room temperature for 2 h.
After hydrogenation, aldehyde 5 (759 g, 3.0 mmol) and
3 mL of aqueous solution of sodium pyrosulfite (286 mg,
1.5 mmol) in H2O were added. The mixture was stirred
at reflux for 5 h and then cooled to room temperature,
filtered to remove catalyst. The filtrate was concentrated
under reduced pressure and purified by chromatography
over Florisil (AcOEt/MeOH: 100/0–90/10), to provide
the bis-benzimidazole with NH2-protected by Boc
group (1128 mg), in 81% yield. 1H NMR (400 Hz,
DMSO-d6): 1.41 (s, 9H, (CH3)3C), 1.87–1.97 (m, 1H,
CH2), 2.16–2.25 (m, 1H, CH2), 3,10 (m, 1H, CH2N),
3.24–3.56 (3m, 3H, CH2N), 3.86 (s, 3H, OCH3), 4.17
(m, 1H, CHN), 6.55 (m, 2H, ArH), 7.14 (d, 2H, ArH,
J=8.7 Hz), 7.21 (d, 1H, ArH), 7.42 (d, 1H, ArH,
J=8.5 Hz), 7.66 (broad, 1H, ArH or NHCO), 7.99 (d,
1H, ArH), 8.16 (d, 2H, ArH, J=8.7 Hz), 8.26 (broad,
1H, ArH or NHCO), 12.95 (broad, 2H, NH). MS: 424
(86, MH+ÀBoc). Anal. calcd for C30H32N6O3 (524.62)
containing 2.03% of water: C 67.92, H 6.15, N 16.02;
found: C 68.38, H 6.31, N 15.76.
20-(4-Methoxy-phenyl)-[2,50] bis-benzimidazole-5-carb-
aldehyde (10). To a suspension of 9 (427 mg, 1 mmol) in
THF/Et2O (30/20 mL), LiAlH4 (150 mg, 3 mmol) was
added at À70 ꢁC. The mixture was stirred for 5 h and
the temperature was allowed to raise to room tempera-
ture. The reaction was quenched by slow addition of
AcOEt (50 mL) at 0 ꢁC and then NH4Cl saturated solu-
tion (15 mL). After 30 min further stirring at room tem-
perature, the mixture was filtered through a bed of
Speedex and the filtrate was concentrated under reduced
pressure. The resulting residue was dissolved in MeOH
and precipitated with Et2O, to provide compound 10
(220 mg, 60%) as beige solid. 1H NMR (DMSO-d6):
3.89 (s, 3H, OCH3), 7.22 (d, 2H, ArH, J=8.9 Hz), 7.82–
7.89 (m, 3H, ArH), 8.21–8.27 (m, 2H, ArH), 8.27 (d,
2H, ArH, J=8.9 Hz), 8.54(d, 1H, ArH, J<2 Hz), 10.08
(s, 1H, CHO). MS (ISP): 369 (100, M+H+).
5-(4-Methyl-1-piperazinyl)-2-nitroaniline (11). A mixture
of 5-chloro-2-nitroaniline (34.2 g, 0.2 mol), N-methyl-
piperazine (22 g, 0.2 mol) and potasium carbonate (28 g,
0.2 mol) in DMF (100 mL) was heated at reflux for 3 h
and then cooled to room temperature. After addition of
200 mL of H2O, the resulting precipitate was filtered
and the obtained solid was suspended in 1L of 2 N ace-
tic acid and filtered. The filtrate was slightly basified
with ammoniac solution and the resulting precipitate
The Boc protecting group was removed as follows:
503 mg of above compound was dissolved in CH2Cl2
(20 mL) at 0 ꢁC and trifluoroacetic acid (10 mL) was
added. The mixture was stirred for 2 h and the tem-
perature was allowed to raise to room temperature.

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2917
Afterwards, the mixture was evaporated to dryness and
was purified by reverse phase chromatography over a
Opti-up column with CH3CN/H2O (containing 0.4%
TFA) as eluent. The collected fractions were con-
centrated under reduced pressure to remove CH3CN
and then lyophilized to afford the bis-benzimidazole 1b
as orange powder in 81% yield. 1H NMR (400 Hz,
DMSO-d6): 2.19–2.20 (m, 1H, CH2), 2.36–2.47 (m, 1H,
CH2), 3.37–3.47 (m, 2H, CH2N), 3.54–3.69 (2m, 2H,
CH2N), 3.87 (s, 3H, OCH3), 3.99–4.19 (m, 1H, CHN),
6.73 (d, 1H, ArH, J<2 Hz), 6.90 (dd, 1H, ArH,
J1=8.5 Hz, J2<2 Hz), 7.18 (d, 2H, ArH, J=8.8 Hz),
7.66 (d, 1H, ArH, J=8.9 Hz), 7.87 (d, 1H, ArH,
J=8.5 Hz), 8.00 (dd, 1H, ArH, J1=8.9 Hz, J2<2 Hz),
8.16 (m, 3H, NH+3 ), 8.20 (d, 2H, ArH, J=8.8 Hz), 8.42
(d, 1H, ArH, J<2 Hz), 14.75 (broad, 2H, NH). MS
(ISP): 425 (100, MH+). Anal. calcd for
5% Pd/C (100 mg) in EtOH (50 mL). After hydrogena-
tion, 5 mL solution of Na2S2O5 (100 mg, 0.53 mmol) and
aldehyde 10 (500 mg, 1.5 mmol) were introduced. The
mixture was stirred at reflux overnight and then filtered.
After addition of 100 mL of H2O into the filtrate, the
resulting precipitates were collected and recrystallized in
MeOH to give the title tris-benzimidazole 2a (120 mg,
21.7%), which was transformed to the hydrochloride
salt by HCl/MeOH solution. The low yield was due to
the purification step. 1H NMR (DMSO-d6): 2.87 (s, 3H,
NCH3), 3.20–3.51 (m, 8H, CH2N), 3.87 (s, 3H, OCH3),
7.01 (dd, 1H, ArH, J1=8.8 Hz, J2<2 Hz), 7.14 (d, 1H,
ArH, J<2 Hz), 7.16 (d, 2H, ArH, J=8.8 Hz), 7.52 (d,
1H, ArH, J=8.7 Hz), 7.67–7.80 (m, 2H, ArH), 8.06 (d,
1H, ArH, J=8.5–9 Hz), 8.12 (d, 1H, ArH, J=8–9 Hz),
8.18 (d, 2H, J=8.8 Hz), 8.35–8.45 (m, 2H, ArH), 13.08
(broad, 2–3H, NH). MS (ISP): 555 (100, M+H+).
ꢀ
ꢀ
C25H24N601 2.95 mol CF3COOHꢀ 3.4 mol H2O (822.14):
Anal. calcd for C33H30N8O1 3.8 mol HCl (693.210)
containing MeOH (2.78%): C 56.63, H 5.13, N 15.72,
Cl 18.90; found: C 56.61, H 5.18, N 15.41, Cl 19.00.
C 45.14, H 4.14, N 10.22, F 20.45, H2O 7.45; found: C
45.50, H 3.93, N 10.32, F 20.75, H2O 7.93.
1-{200-(4-Methoxy-phenyl)-[2,50:20,500]terbenzimidazole-5-
yl}-pyrrolidin-3-ylamine (2b). Tris-benzimidazole 2b was
synthesized in a similar manner as bisbenzoimidazole
1b. A solution of compound 12 (322 mg, 1 mmol) in
EtOH (100 mL) was hydrogenated over 5% Pd/C
(100 mg) at room temperature for 3 h. After hydro-
genation, aldehyde 10 (352 mg, 1 mmol) and 1 mL of
solution of sodium pyrosulfite (100 mg, 0.52 mmol) were
added. The mixture was stirred at reflux overnight, then
cooled to room temperature, filtered to remove catalyst.
The filtrate was concentrated under reduced pressure to
obtain a brown solid, which was purified by reverse
phase chromatography over an Opti-up column with
CH3CN/H2O (containing 0.1% TFA) as eluent. The
collected fractions were concentrated under reduced
pressure to remove CH3CN and then lyophilized, to
afford tris-benzimidazole with NH2-protected by Boc
group (99 mg, 59%), as orange solid.
Chemicals and biochemicals
Hoechst 33258 and Hoechst 33342 were purchased from
Sigma Chemical Co. (UK) Stock solutions were prepared
in water before subsequent dilution with a 10 mM Tris–
HCl buffer pH 7.0, containing 10 mM NaCl. Ammonium
persulphate, tris base, acrylamide, bis-acrylamide, ultra-
pure urea, boric acid, tetramethylethylenediamine and
dimethyl sulphate were from BDH. Formic acid,
piperidine, hydrazine and formamide were from
Aldrich. Photographic requisites were from Kodak.
Bromophenol blue and xylene cyanol were from Serva.
a-[32P]-dATP was obtained from NEN Dupont. Unla-
beled dATP was ultra-pure grade from Pharmacia.
Restriction endonucleases AvaI, EcoRI and PvuII
(Boehringer Mannheim, Germany) were used according
to the supplier’s recommended protocol in the activity
buffer provided. Avian myeloblastosis virus (AMV)
reverse transcriptase was from Pharmacia. Plasmids
pKMÁ-98, pUC12 and pBS (Stratagene, La Jolla, CA,
USA) were isolated from Escherichia coli by a standard
sodium dodecyl sulphate–sodium hydroxide lysis pro-
cedure and purified by banding in CsCl–ethidium bro-
mide gradients. Ethidium was removed by several
isopropanol extractions followed by exhaustive dialysis
against tris–EDTA buffer. The purified plasmid was
then precipitated and resuspended in appropriate buffer
prior to digestion by the restriction enzymes. All other
chemicals were analytical grade reagents, and all solu-
tions were prepared using doubly deionised, Millipore
filtered water.
The Boc protecting group was removed by treatment
with trifluoroacetic acid (4 mL) in CH2Cl2 (8 mL), from
0 ꢁC to room temperature. After deprotection, the mix-
ture was evaporated to dryness and the crude product
was purified by HPLC reverse phase chromatography
over RP-18 Select-B column with CH3CN/H2O (con-
taining 0.4% TFA) as eluent. The collected fractions
were concentrated under reduced pressure and lyophi-
lized to afford the title tris-benzoimidazole 2b as orange
1
solid, in 59% yield. H NMR (400 Hz, DMSO-d6): 2.14
(m, 1H, CH2), 2.41 (m, 1H, CH2), 3.37–3.47 (m, 2H,
CH2N), 3.54–3.69 (2m, 2H, CH2N), 3.88 (s, 3H, OCH3),
4.18 (m, 1H, NCH), 6.73 (d, 1H, ArH, J<2 Hz), 6.90
(dd, 1H, ArH, J1=8–9 Hz, J2<2 Hz), 7.29 (d, 2H, ArH,
J=8.8 Hz), 7.67 (d, 1H, ArH, J=8.5 Hz), 7.80 (d, 1H,
ArH, J=8.5 Hz), 7.91 (d, 1H, ArH, J=8.5–9 Hz), 8.03
(dd, 1H, ArH, J1=8–9 Hz, J2<2 Hz), 8.13 (m, NH+3 or
NH+), 8.18 (dd, 1H, ArH), 8.20 (d, 2H, ArH, J=8.8 Hz),
8.48 (broad, 2H, ArH). MS (ISP): 541 (100, MH+).
Melting temperature studies
Absorption spectra were recorded on a Uvikon 943
spectrophotometer. The 12-cell holder was thermostated
with a NeslabRTE 111 cryostat. Drug-RNA complexes
were prepared by adding aliquots of a concentrated
drug solution to a DNA solution at constant con-
centration (usually 20 mM) in BPE buffer pH 7.1 (6 mM
Na2HPO4, 2 mM Na2H2PO4, 1 mM EDTA). A heating
rate of 1 ꢁC/min was used and data points were collected
200 -(4-methoxy-phenyl)-6-(4-methylpiperazin-1-yl)-[2,50;
20,500]tertbenzimidazole (2a). 5-(N-Methylpiperazinyl)-2-
nitroaniline 11 (236 mg, 1 mmol) was hydrogenated over

2918
Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
every 30 s. The temperature inside the cuvette was
monitored by using a thermocouple. The absorbance at
260 nm was measured over the range 25–90 ꢁC in 1-cm
path length reduced volume quartz cells. The ‘melting’
temperature Tm was taken as the mid-point of the
hyperchromic transition determined from first deriva-
tives plots. The reproducibility of the Tm measurements
was ꢅ1 ꢁC.
Electrophoresis and autoradiography
DNA cleavage products were resolved by poly-
acrylamide gel electrophoresis under denaturing condi-
tions (0.3 mm thick, 8% acrylamide gels containing 8 M
urea) capable of resolving DNA fragments differing in
length by one nucleotide. Electrophoresis was continued
until the bromophenol blue marker had run out of the
gel (about 2.5 h at 60 Watts, 1600 V in TBE buffer; BRL
sequencer Model S2). Gels were soaked in 10% acetic
acid for 15 min, transferred to Whatman 3MM paper,
dried under vacuum at 80 ꢁC, and subjected to auto-
radiography at À70 ꢁC with an intensifying screen.
Exposure times of the X-ray films (Fuji R-X) were
adjusted according to the number of counts per lane
loaded on each individual gel (usually 24 h).
Linear dichroism measurements
Flow-oriented linear dichroism (LD) spectra were
recorded in a Couette cell using the experimental setup
previously described.41
DNA restriction fragments
Three DNAs of different base composition: (i) a 117 bp
DNA fragment from plasmid pBS, (ii) the 160 bp tyr T
DNA fragment and (iii) a 178 bp DNA fragment from
plasmid pUC12 were used in the DNAaseI footprinting
studies. The tyr T DNA was obtained by digestion of
the plasmid pKMÁ-98 with EcoRI and AvaI.53 The 117
mer and 178 mer were obtained after digestion of the
corresponding plasmid with the restriction enzymes
PvuII and EcoRI. In each case, the fragment was incu-
bated with AMV reverse transcriptase in the presence of
a-[32P]-dATP to label specifically the 30-end at the
EcoRI site. The singly end-labeled DNA fragment was
then purified by preparative non-denaturing poly-
acrylamide gel electrophoresis (6.5% acrylamide,
1.5 mm thick, 200 V, 2 h, in TBE buffer: 89 mM Tris
base, 89 mM boric acid, 2.5 mM Na2 EDTA, pH 8.3).
After rapid autoradiography to locate the DNA, the
band was excised from the gel, minced with a blade and
extracted overnight in 500 mM ammonium acetate,
10 mM magnesium acetate. The purified DNA was then
precipitated twice with 70% ethanol prior to resuspension
in 10 mM Tris, 10 mM NaCl buffer pH 7.0.
Quantitation by storage phosphor imaging
A Molecular Dynamics 425E PhosphorImager was used
to collect data from storage screens exposed to dried
gels overnight at room temperature. Base line-corrected
scans were analyzed by integrating all the densities
between two selected boundaries using ImageQuant
version 3.3 software. Each resolved band was assigned
to a particular bond within the DNA fragment by
comparison of its position relative to sequencing stan-
dards generated by treatment of the DNA with formic
acid (G+A) followed by piperidine-induced cleavage at
the modified bases in DNA.
Acknowledgements
This work was done under the support of research
grants (to C.B.) from the Ligue Nationale Franc¸ aise
contre le Cancer, Comite du Nord and (to M.J.W.)
from the European Union, the Wellcome Trust, CRC
and AICR. The authors thank the Sir Halley Stewart
Trust for a grant to assist cooperation.
DNase I footprinting
These experiments were performed essentially according
to a published protocol.54 Reactions were conducted in
a total volume of 8 mL. Samples (2 mL) of the labeled
DNA fragment were incubated with 4 mL of the buffer
solution containing the ligand at appropriate con-
centration. After 30–60 min incubation at 37 ꢁC to
ensure equilibration of the binding reaction, the diges-
tion was initiated by the addition of 2 mL of a DNAase I
solution whose concentration was adjusted to yield a
final enzyme concentration of about 0.01 unit/mL in the
reaction mixture. The extent of digestion was limited to
less than 30% of the starting material so as to minimize
the incidence of multiple cuts in any strand (‘single-hit’
kinetic conditions). Optimal enzyme dilutions were
established in preliminary calibration experiments.
After 3 min, the digestion was stopped by freeze drying,
samples were lyophilized, washed once with 50 mL of
water, lyophilized again and then resuspended in 4 mL of
an 80% formamide solution containing tracking dyes.
Samples were heated at 90 ꢁC for 4 min and chilled in ice
for 4 min prior to electrophoresis.
References and Notes
1. Stokke, T.; Steen, H. B. J. Histochem. Cytochem. 1985, 33,
333.
2. Hauglang, R. P. In Handbook of Fluorescent Probes and
Research Chemicals, 6th ed.; Spence, M. T. Z., Ed.; Molecular
Probes: Eugene, OR, 1996.
3. Portugal, J.; Waring, M. J. Biochim. Biophys. Acta 1987,
949, 158.
4. Pjura, P. E.; Grzeskoniak, K.; Dickerson, R. E. J. Mol.
Biol. 1987, 197, 257.
5. Kopka, M. L.; Pjura, P.; Goodsell, D.; Dickerson, R. E. In
Nucleic Acids and Molecular Biology; Eckstein, F.; Lilley, D.
M. J., Eds.; Springer: Berlin, 1987; Vol. 1, pp 1–24.
6. Jorgenson, K. F.; Varshney, U.; van de Sande, J. H. J.
Biomol. Struct. Dyn. 1988, 5, 1005.
7. Loontiens, F. G.; Regenfuss, P.; Zechel, A.; Dumortier, L.;
Clegg, R. M. Biochemistry 1990, 29, 9029.
8. Loontiens, F. G.; McLaughlin, L. W.; Diekmann, S.;
Clegg, R. M. Biochemistry 1991, 30, 182.
9. Quintana, J. R.; Lipanov, A. A.; Dickerson, R. E. Bio-
chemistry 1991, 30, 10294.

Y.-H. Ji et al. / Bioorg. Med. Chem. 9 (2001) 2905–2919
2919
10. Sriram, M.; van der Marel, G. A.; Roelen, H. L. P. F.; van
Boom, J. H.; Wang, A. H.-J. EMBO J. 1992, 11, 225.
11. Carrondo, M. A. A.deC.T.; Coll, M.; Aymami, J.; Wang,
A. H. J.; Van der Marel, G. A.; Van Boom, J. H.; Rich, A.
Biochemistry 1989, 28, 7849.
12. Searle, M. S.; Embrey, K. J. Nucleic Acids Res. 1990, 18,
3753.
13. Embrey, K. J.; Searle, M. S.; Craik, D. J. J. Chem. Soc.,
Chem. Commun. 1991, 1770.
30. Kumar, S.; Joseph, T.; Singh, M. P.; Bathini, Y.; Lown,
J. W. J. Biomol. Struct. Dyn. 1992, 8, 331.
31. Singh, M. P.; Joseph, T.; Kumar, S.; Bathini, Y.; Lown,
J. W. Chem. Res. Toxicol. 1992, 5, 597.
32. Spink, N.; Brown, D. G.; Skelly, J. V.; Neidle, S. Nucleic
Acids Res. 1994, 22, 1607.
33. Satz, A. L.; Bruice, T. C. J. Am. Chem. Soc. 2001, 123, 2469.
34. Gerber, P. R.; Muller, K. J. Comput. Aided Mol. Des.
1995, 9, 251.
14. Embrey, K. J.; Searle, M. S.; Craik, D. J. Eur. J. Biochem.
1993, 211, 437.
15. Fede, A.; Labhardt, A.; Bannwarth, W.; Leupin, W. Bio-
chemistry 1991, 30, 11377.
16. Parkinson, J. A.; Barber, J.; Douglas, K. T.; Rosamond,
J.; Sharples, D. Biochemistry 1990, 29, 10181.
17. Parkinson, J. A.; Barber, J.; Douglas, K. T.; Rosamond,
J.; Sharples, D. Biochemistry 1994, 33, 8442.
18. Vega, M. C.; Garcia Saez, I.; Aymami, J.; Eritja, R.; Van
der Marel, G. A.; van Boom, J. H.; Rich, A.; Coll, M. Eur. J.
Biochem. 1994, 222, 721.
35. Bathini, Y.; Lown, J. W. Synthetic Commun. 1990, 20, 955.
36. Nahm, S.; Weinreb, S. M. Tetrahedron Lett. 1981, 22,
3815.
37. Grimmet, M. R. In Compressive Heterocyclic Chemistry;
Katritzky A. R.; Rees C. W., Eds.; 1984; Vol. 5, p 457.
38. Von Loewe, H.; Urbanietz, J. Arzneim. Forsch. 1974, 24,
1927.
39. Starshikov, N. M.; Pozharskii, A. F. Khim. Chem. Het-
erocycl. Comp. 1980, 81.
40. Maquestiau, A.; Berte, L.; Mayence, A.; Vanden Eynde,
J.-J. Synthetic Commun. 1991, 21, 2181.
19. Gavathiotis, E.; Sharman, G. J.; Searle, M. S. Nucleic
Acids Res. 2000, 28, 728.
41. Vogt, H.; Hanisch, G.; Hochstrasser, R. A. Rev. Sci.
Instrum. 1995, 66, 4385.
20. Fede, A.; Billeter, M.; Leupin, W.; Wuthrich, K. Structure
1993, 1, 177.
42. Aymami, J.; Nunn, C. M.; Neidle, S. Nucleic Acids Res.
1999, 27, 2691.
21. Haq, I.; Ladbury, J. E.; Chowdhry, B. Z.; Jenkins, T. C.;
Chaires, J. B. J. Mol. Biol. 1997, 271, 244.
22. Czarny, A.; Boykin, D. W.; Wood, A. A.; Nunn, C. M.;
Neidle, S.; Zhao, M.; Wilson, W. D. J. Am. Chem. Soc. 1995,
117, 4716.
23. Wood, A. A.; Nunn, C. M.; Czarny, A.; Boykin, D. W.;
Neidle, S. Nucleic Acids Res. 1995, 23, 3678.
24. Clark, G. R.; Boykin, D. W.; Czarny, A.; Neidle, S.
Nucleic Acids Res. 1997, 25, 1510.
25. Bostock-Smith, C. E.; Embrey, K. J.; Searle, M. S. Nucleic
Acids Res. 1998, 26, 1660.
26. Ebrahimi, S. E. S.; Parkinson, J. A.; Fox, K. R.; McKie,
J. H.; Barber, J.; Douglas, K. T. J. Chem. Soc., Chem. Com-
mun. 1992, 1398.
27. Clark, G. R.; Gray, E. J.; Neidle, S.; Ji, Y.-H.; Leupin, W.
Biochemistry 1996, 35, 13745.
28. Bathini, Y.; Rao, K. E.; Shea, R. G.; Lown, J. W. Chem.
Res. Toxicol. 1990, 3, 268.
29. Rao, K. E.; Lown, J. W. Chem. Res. Toxicol. 1991, 4,
661.
43. Pilch, D. S.; Xu, Z.; Sun, Q.; LaVoie, E.; Liu, L. F.; Gea-
cintov, N. E.; Breslauer, K. J. Drug Des. Discov. 1996, 13, 115.
44. Thurston, D. E.; Br, J. Cancer 1999, 80, 65.
45. Bailly, C.; Chaires, J. B. Bioconjugate Chem. 1998, 9, 513.
46. Drobyshev, A. L.; Zasedatelev, A. S.; Yershov, G. M.;
Mirzabekov, A. D. Nucleic Acids Res. 1999, 27, 4100.
47. Sigman, D. S.; Chen, C.-H. B. Annu. Rev. Biochem. 1990,
59, 207.
48. Van Dyke, M. W.; Dervan, P. B. Biochemistry 1983, 22,
2373.
49. Suck, D.; Lahm, A.; Oefner, C. Nature 1988, 332, 465.
50. Weston, S. A.; Lahm, A.; Suck, D. J. Mol. Biol. 1992, 226,
1237.
51. Bailly, C.; Waring, M. J. J. Biomol. Struct. Dyn. 1995, 12,
869.
52. Clark, G. R.; Squire, C. J.; Gray, E. J.; Leupin, W.; Nei-
dle, S. Nucleic Acids Res. 1996, 24, 4882.
53. Drew, H. R.; Travers, A. A. Cell 1984, 37, 491.
54. Low, C. M. L.; Drew, H. R.; Waring, M. J. Nucleic Acids
Res. 1984, 12, 4865.