Symmetric bis-benzimidazoles: New sequence-selective DNA-binding molecules

Article abstract of DOI:10.1039/a901074b

A series of bis-benzimidazole compounds with a head-to-head orientation have been designed as sequence-specific DNA binders; crystallographic analysis of oligonucleotide complexes has been combined with DNase I footprinting to confirm that the predicted optimal site for the core bisbenzimidazole motif is the four-base-pair sequence 5'-AATT; this sequence specificity results in inhibition of transcription at A/T sites and may be responsible for the cytotoxic and antitumour effects shown by these head-to-head bis-benzimidazoles.

Full text of DOI:10.1039/a901074b

Symmetric bis-benzimidazoles: new sequence-selective DNA-binding molecules
Stephen Neidle,*a John Mann,*b Emma L. Rayner,a Anne Baron,b Yaw Opoku-Boahen,b Ian J. Simpson,a
Nicola J. Smith,a Keith R. Fox,c John A. Hartleyd and Lloyd R. Kellande
a The CRC Biomolecular Structure Unit, The Institute of Cancer Research, Sutton, Surrey, UK SM2 5NG
b Department of Chemistry, University of Reading, Whiteknights Park, Reading UK RG6 6AD
c Division of Biochemistry and Molecular Biology, University of Southampton, Southampton, UK SO16 7PX
d CRC Drug-DNA Interactions Research Group, UCL Medical School, London, UK W1P 8BT
e CRC Centre for Cancer Therapeutics, The Institute of Cancer Research, Sutton, Surrey, UK SM2 5NG
Received (in Liverpool, UK) 8th February 1999, Accepted 30th March 1999
A series of bis-benzimidazole compounds with a head-to-
head orientation have been designed as sequence-specific
DNA binders; crystallographic analysis of oligonucleotide
complexes has been combined with DNase I footprinting to
confirm that the predicted optimal site for the core bis-
benzimidazole motif is the four-base-pair sequence 5A-
AATT; this sequence specificity results in inhibition of
transcription at A/T sites and may be responsible for the
cytotoxic and antitumour effects shown by these head-to-
head bis-benzimidazoles.
and analogues) to four consecutive A:T bps, with distinctive
cross-strand H-bonding involving each base pair. This arrange-
ment thus extends the effective recognition of each benzimida-
zole sub-unit to two A:T bps. The modelling used the structures
of
the
self-complementary
duplex
sequences
d(CGCGAATTCGCG) and d(CGCAAATTTGCG), as found
in several relevant drug complexes.9,10 It suggested that the
central 5’-AATT sequence would be an optimal site for the
head-to-head motif and for maintenance of helical register with
all four consecutive bps without necessitating significant DNA
conformational change. An initial range of head-to-head bis-
benzimidazoles have now been synthesised, and is reported
here.
The minor groove of duplex DNA is the site of non-covalent
interaction of a large number of anticancer drugs, antibiotics
and antiviral agents,1 which are believed to exert their action by
competing with transcription factors2 or architectural proteins,3
such as E2F, TATA-box binding proteins or DNA topoisomer-
ase I/II. The molecular basis of DNA recognition for a number
of drugs in this super-family (notably distamycin, netropsin,
The highly convergent synthesis of these bis-benzimidazoles
(Scheme 1) involves the condensation between the requisite
4-substitituted aryl aldehyde and 3,3A,4,4A-tetraaminobiphenyl
7. The latter is commercially available, but is of variable quality,
and we preferred to prepare it by the route shown. Thus, 1 and
2 were easily obtained from 4-methoxybenzaldehyde and
4-hydroxybenzaldehyde, respectively, while 3 was obtained
from 4-(3-dimethylaminopropoxy)benzaldehyde prepared from
4-hydroxybenzaldehyde and 3-dimethylaminopropanol using a
Mitsunobu reaction. In each instance the aryl aldehyde and
3,3A,4,4A-tetraaminobiphenyl were heated in nitrobenzene at 150
ºC for 8–10 h to effect the condensation, and the yields of bis-
benzimidazoles were typically 25–40% on the 10 mmol scale.
Crystallographic analyses have been undertaken on (i) the
bis(hydroxyphenyl) compound 2 co-crystallised with the oligo-
nucleotide sequence d(CGCGAATTCGCG), and (ii) its Me2N
derivative 3 with the sequence d(CGCAAATTTGCG). Both
crystal structures11 show a B-form DNA double helix with
ligand bound in the A/T region of the minor groove [Fig. 1(a)].
They show H-bonding between all four central A:T bps and the
benzimidazole subunits [Fig. 1(b)], in full accord with the
modelling predictions. Each inner-facing ring nitrogen atom of
OH
Me
N
N
N
N
NH
NH
Hoechst 33258
berenil, pentamidine and Hoechst 33258) has been extensively
studied, by crystallographic,4 NMR5 and footprinting methods.6
These studies have shown that their AT preferences are a
consequence of, in particular, (i) the sequence-dependent
narrow width of the minor groove of B-form DNA, resulting in
stabilisation by van der Waals interactions with the walls of the
groove,7 and (ii) their ability to form specific H-bonds with
donor and acceptor atoms on the minor groove edge of A:T
base pairs. These factors have been utilised in the design of
molecules with altered and extended recognition properties,
some of which are capable of sequence-specific gene regula-
tion.8
iii, iv
NHR1
R2
NHR1
NHR1
We have previously reported crystallographic analyses of a
number of oligonucleotide complexes with the head-to-tail bis-
benzimidazole compound Hoechst 33258, and several of its
derivatives,9 including a tris-benzimidazole with an extended
recognition site.10 These have shown that each benzimidazole
subunit interacts with two A:T base pairs by means of a pair of
cross-strand H-bonds. The head-to-tail arrangement forces the
site for each successive subunit to overlap the previous one by
one base-pair (bp), so that each benzimidazole group in Hoechst
33258 and other head-to-tail analogues effectively recognises
1.5 A:T bps.
Molecular modelling, based on these structural studies, has
suggested to us that the novel symmetric head-to-head benzimi-
dazole arrangement could also effectively bind in the minor
groove. This arrangement would extend the size of the bis-
benzimidazole recognition site from three (in Hoechst 33258
R2
R2
4 R1 = R2 = H
6 R1 = Ac, R2 = NO2
7 R1 = H, R2 = NH2
i,ii
v,vi
5 R1 = Ac, R2 = HgOAc
vii
N
N
N
H
N
H
1 X = OMe
2 X = OH
3 X = O(CH2)3NMe2
X
X
Scheme 1 Reagents and conditions: i, Ac2O, AcOH; ii, Hg(OAc)2, AcOH,
HClO4, ca. 20% (2 steps); iii, Cu, PdCl2, Py; iv, HNO3, AcOH, ca. 50% (2
steps); v, H2SO4; vi, Raney Ni, H2, ca. 70% for 2 steps; vii, 4-XC6H4CHO
[X+MeO (1), OH (2) or Me2N(CH2)3O (3)], PhNO2, 150 °C.
Chem. Commun., 1999, 929–930
929

View Article Online
Hoechst 33258. The Me2N derivative 3 is over 10-fold more
active than the others in three out of the four lines. 3 also shows
significant activity in several ovarian lines in the NCI 60-cell
line panel (to be published). The compound with least in vitro
activity, the hydrophobic dimethoxy derivative 1, has been
examined in an in vivo tumour model with the ADJ/PC6
plasmacytoma, where it showed 44% tumour inhibition with ip
administration. Compound 3 has been examined in an in vitro
transcription assay. It effectively inhibits transcription at a
number of A/T sites, consistent with the above data. In addition,
compound 3 inhibited transcription at least 10-fold more
efficiently than compound 2, consistent with its increased
cytoxicity. We speculate that the activity of 3 in vitro and in vivo
may be related to its potent ability to specifically inhibit
transcription at a small sub-set of A/T sites. The relatively low
affinity of compound 3 for 5A-TATA sites suggests that it is less
likely to block TATA-box binding proteins at the start of
transcription.
This work has been supported by grants from the Cancer
Research Campaign, the Royal Society of Chemistry (Adrien
Albert Memorial Fund) and the Ghanaian Government.
Notes and references
1 C. Zimmer and U. Wähnert, Prog. Biophys. Mol. Biol., 1986, 47, 31;
P. B. Dervan, Science, 1986, 232, 464.
2 See for example: J. J. Welch, F. J. Rauscher III and T. A. Beerman,
J. Biol. Chem., 1994, 269, 31051; S.-Y. Chiang, J. C. Azizkhan and T.
A. Beerman, Biochemistry, 1998, 37, 3109.
3 For example: Z. Xu, T.-K. Li, J. S. Kim, E. J. LaVoie, K. J. Breslauer,
L. F. Liu and D. S. Pilch, Biochemistry, 1998, 37, 3558; J. S. Kim, Q.
Sun, C. Yu, A. Liu, L. F. Liu and E. J. LaVoie, Bioorg. Med. Chem.,
1998, 6, 163.
Fig. 1 Computer plots of the crystal structure of the complex of 2 with
d(CGCGAATTCGCG): (a) shows the ligand, in space-filling mode,
centrally bound in the minor groove of the 12-mer double helix, while (b)
shows a detailed view of 2 and the H-bond contacts to A:T bp edges.
4 S. Neidle, Biopolymers, 1997, 44, 105.
Table 1 Cytotoxicity of bis-benzimidazole compounds, for 96 h exposure in
a panel of four ovarian cell lines (as IC50/mM)
5 A. Fede, A. Labhardt, W. Bannwarth and W. Leupin, Biochemistry,
1991, 30, 11377; J. A. Parkinson, S. E. Ebrahimi, J. H. McKie and K.
T. Douglas, Biochemistry, 1994, 33, 8442; C. E. Bostock-Smith, C. A.
Laughton and M. S. Searle, Nucleic Acids Res., 1998, 26, 1660.
6 K. D. Harshman and P. B. Dervan, Nucleic Acids Res., 1985, 13, 4825;
A. Abu-Daya, P. M. Brown and K. R. Fox, Nucleic Acids Res., 1995, 23,
3385.
7 S. Neidle, FEBS Lett., 1992, 298, 97; A. Czarny, D. W. Boykin, A. A.
Wood, C. M. Nunn, S. Neidle, M. Zhao and W. D. Wilson, J. Am. Chem.
Soc., 1995, 117, 4716.
Compound
A2780
A2780-PtR
CH1
SKOV-3
1
2
3
13.5
4.3
0.235
12.0
5.1
2.7
0.115
9.5
3.8
1.05
0.24
26.5
16.5
16
0.375
> 100
Hoechst 33258
8 J. M. Gottesfield, L. Neely, J. W. Trauger, E. E. Baird and P. B. Dervan,
Nature, 1997, 387, 205; S. White, J. W. Szewczyk, J. M. Turner, E. E.
Baird and P. B. Dervan, Nature, 1998, 391, 468; L. A. Dickinson, R. J.
Gulizia, J. W. Trauger, E. E. Baird, D. M. Mosier, J. M. Gottesfield and
P. B. Dervan, Proc. Natl. Acad. Sci. U.S.A., 1998, 95, 12890.
9 N. Spink, D. G. Brown, J. V. Skelly and S. Neidle, Nucleic Acids Res.,
1994, 22, 1607; G. R. Clark, D. W. Boykin, A. Czarny and S. Neidle,
Nucleic Acids Res., 1997, 25, 1510.
a benzimidazole ring H-bonds to a thymine O2 atom of one bp
and an adenine N3 atom on the opposite strand of an adjacent
one. It is notable that the absence of formal cationic charge in 2,
does not affect its sequence selectivity, suggesting that the role
of charge is to modulate affinity rather than to determine
sequence selectivity.
These preferences have also been observed in DNase I
footprinting studies with a natural DNA fragment, which have
shown that 2 and 3 both interact preferentially with some A/T
sites. Compound 3 binds to AT sites between 7–10-fold better
than 2, and has an especially high affinity for the 5A-AATT site
(30 nM). Quantitative footprinting data indicated that the order
of preference is 5A-AATT > 5A-ATAT > 5A-TAAT > 5A-TATA
~ 5A-TTAA, similar to that previous determined for Hoechst
33258. The increased affinity of 3 relative to 2 does not affect
its sequence selectivity. This enhanced binding strength may be
a consequence of its cationic charge or its increased DNA site
size, since the crystal structure of the complex with
d(CGCAAATTTGCG)2 shows it to cover 6 bps. The ability of
the head-to-head bis-benzimidazoles to discriminate between
the 136 possible tetranucleotide sequences may be further
enhanced by flanking sequence preferences. These remain to be
systematically explored. Whilst this paper was in preparation, a
report has appeared which describes compound 2 and its
interactions with tRNA.12
10 G. R. Clark, E. J. Gray, S. Neidle, Y.-H. Li and W. Leupin,
Biochemistry, 1996, 35, 13745.
11 Compound 2 was co-crystallised with d(CGCGAATTCGCG) (from
Oswell DNA Service, Southampton) using the hanging-drop method at
286 K. Elongated prismatic crystals were obtained from a 1.5 mM
solution of 1 in 50% w/w MPD, in a droplet containing 10 mM MgCl2,
60 mM KCl, 0.75 mM DNA and 6 mM spermine, equilibrated against
30% w/w MPD. The crystals are orthorhombic, space group P212121,
with cell dimensions a = 25.46(2), b = 39.96(4) and c = 65.66(7) Å.
Intensity data were collected from a flash-frozen crystal maintained at
100 K with a RAXIS-4 image plate detector, using Cu-Ka radiation
from a rotating-anode generator and mirror-focussing optics. A total of
6380 unique reflections were obtained after merging Friedel equivalents
(Rmerge = 0.044) to a resolution of 1.8 Å. The structure was solved using
molecular replacement, and refined with the SHELX-97 package. The
final model included 174 water molecules, and had an R of 0.229 for
6157 reflections with F > 4s(F), and 23.2% for all data to 1.8 Å. Rfree
was 25.1%. Coordinates and reflection data have been deposited at the
Nucleic Acid Database as entry no. dd0013. See http://www.rsc.org/
suppdata/cc/1999/929/ for a 3D image in .pdb format.
12 E. V. Bichenkova, S. E. Sadat-Ebrahimi, A. N. Wilton, N. O’Toole,
D. S. Marks and K. T. Douglas, Nucleosides Nucleotides, 1998, 17,
1651.
These new bis-benzimidazoles have been examined for
potential cytotoxic effects in a group of ovarian carcinoma cell
lines. Table 1 shows that all compounds are cytotoxic at the mM
level, with activity significantly greater than that shown by
Communication 9/01074B
930
Chem. Commun., 1999, 929–930