Thiophene-based diamidine forms a "super" AT binding minor groove agent

Article abstract of DOI:10.1021/ja048175m

The DNA minor groove is the interaction site for many enzymes and transcription control proteins and as a result, development of compounds that target the minor groove is an active research area. In an effort to develop biologically active minor groove ag

Full text of DOI:10.1021/ja048175m

Published on Web 09/29/2004
Thiophene-Based Diamidine Forms a “Super” AT Binding
Minor Groove Agent
Sirish Mallena,† Michael P. H. Lee,§ Christian Bailly,*,‡ Stephen Neidle,*,§
Arvind Kumar,† David W. Boykin,*,† and W. David Wilson*,†
Contribution from the Department of Chemistry, Georgia State UniVersity, P.O. Box 4098,
Atlanta, Georgia 30302-4098, INSERM U-524 et Laboratoire de Pharmacologie Antitumorale du
Centre Oscar Lambret, IRCL, Place de Verdun, 59045 Lille, France, and
Cancer Research UK Biomolecular Structure Group, The School of Pharmacy,
UniVersity of London, London WC1N 1AX, United Kingdom
Received March 30, 2004; E-mail: wdw@gsu.edu
Abstract: The DNA minor groove is the interaction site for many enzymes and transcription control proteins
and as a result, development of compounds that target the minor groove is an active research area. In an
effort to develop biologically active minor groove agents, we are preparing and exploring the DNA interactions
of a systematic set of diamidine derivatives with a powerful array of methods including DNase I footprinting,
biosensor-SPR methods, and X-ray crystallography. Surprisingly, conversion of the parent phenyl-furan-
phenyl diamidine to a phenyl-thiophene-benzimidazole derivative yields a compound with over 10-fold-
increased affinity for the minor groove at AT sequences. Single conversion of the furan to a thiophene or
a phenyl to benzimidazole does not cause a similar increase in affinity. X-ray results indicate a small bond
angle difference between the C-S-C angle of thiophene and the C-O-C angle of furan that, when
amplified out to the terminal amidines of the benzimidazole compounds, yields a very significant difference
in the positions of the amidines and their DNA interaction strength.
Introduction
The DNA minor groove is an important target site for
enzymes and transcription control proteins, and it has been a
particularly attractive target site for development of synthetic
agents for therapeutic purposes and for sequence selective
recognition of DNA.1-7 The minor groove binding diamidine,
pentamidine, for example, has been successfully used in the
clinic against a range of human parasitic diseases for many
years.1,8-10 Furamidine (DB75 in Figure 1) is a synthetic
diamidine that has significantly better activity against a number
of microbial parasites with lower toxicity than pentamidine.1,10,11
† Department of Chemistry, Georgia State University.
‡ IRCL, Place de Verdun, 59045 Lille, France.
§ Biomolecular Structure Unit, University of London.
(1) Tidwell, R. R.; Boykin, D. W. In DNA and RNA Binders: from Small
Molecules to Drugs; Demeunynck, M., Bailly, C., Wilson, W. D., Eds.;
Wiley-VCH: 2003; Vol. 2, Chapter 16, pp 414-460.
Figure 1. Structures of the compounds and ∆Tm values with a DNA
oligomer are shown.
(2) (a) Dervan, P. B.; Edelson, B. S. Curr. Opin. Struct. Bio. 2003, 13, 284-
299. (b) Dervan, P. B. Bioorg. Med. Chem. 2001, 9, 2215-2235.
(3) Neidle, S. Nat. Prod. Rep. 2001, 18, 291-309.
A limitation on the use of all diamidines is their lack of oral
bioavailability but in a very exciting development an orally
active prodrug of DB75 has been developed and is currently in
phase II clinical trials as an antitrypanosomal agent.1,10 The
prodrug is readily converted into DB75 after passing through
the intestinal membrane barrier.
In addition to the therapeutic applications described above,
the minor groove has been the recent target of choice for
development of synthetic agents for sequence selective binding
of DNA. Very successful development of sequence-selective
polyamide analogues of distamycin4 has been described by the
Dervan,2 Lown,7 and Lee12 groups. Bruice and co-workers have
(4) Wemmer, D. E. Annu. ReV. Biophys. Biomol. Struct. 2000, 29, 439-461.
(5) Bailly, C.; Chaires, J. B. Bioconjugate Chem. 1998, 9, 513-538.
(6) Chaires, J. B. Curr. Opin. Struct. Biol. 1998, 8, 314-320.
(7) Sondhi, S. M.; Reddy, B.; Lown, J. W. Curr. Med. Chem. 1997, 4, 313-
358.
(8) Cory, M.; Tidwell, R. R.; Fairley, T. A. J. Med. Chem. 1992, 35, 431-
438.
(9) Fairley, T. A.; Tidwell, R. R.; Donkor, I.; Naiman, N. A.; Ohemeng, K.
A.; Lombardy, R. J.; Bentley, J. A.; Cory, M. J. Med. Chem. 1993, 36,
1746-1753.
(10) (a) Bouteille, B.; Oukem, O.; Bisser, S.; Dumas, M. Fund. Clin. Pharmacol.
2003, 17, 171-181. (b) Fairlamb, A. H. Trends Parasitol. 2003, 19, 488-
494. (c) Ismail, M. A.; Brun, R.; Easterbrook, J. D.; Tanious, F. A.; Wilson,
W. D.; Boykin, W. D. J. J. Med. Chem. 2003, 46, 4761.
(11) Das, B. P.; Boykin, D. W. J. Med. Chem. 1977, 20, 531-536.
9
10.1021/ja048175m CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 13659-13669
13659

A R T I C L E S
Mallena et al.
shown that some polyamide-benzimidazole hybrid compounds
can recognize long base pair sequences of DNA in a 1:1
complex.13 Armitage and co-workers have made the exciting
discovery that selectively designed cyanine dyes can stack as
dimers in the minor groove in AT sequences to form a tandem
helical array that has the potential to generate new nanomate-
rials.14 For the heterocyclic diamidines described above, the
DNA complexes of DB75 and the related furan-benzimidazole
derivative DB293 (Figure 1) have been thoroughly studied and
they display many interesting, useful and quite different
properties.1,3,15,16 Both compounds bind strongly as monomer
complexes in the DNA minor groove in AT sequences of at
least four base pairs in length. Both have a significant
hydrophobic energetic component in formation of their AT
complexes but also depend on other energetic terms such as
van der Waals, electrostatic and H-bonding to drive complex
formation.17,18 DB75 binds much more strongly to AT rich
sequences than pentamidine and, because of the fact that many
parasitic microorganisms have AT rich DNA, the AT binding
appears to be correlated with their biological activity.1 DB75
is an important compound because of its therapeutic properties
but it is essential to develop additional compounds in case of
drug resistance as well as to develop compounds with activities
against other organisms. Compounds that exploit the nature of
the AT rich DNAs in both the nucleus and particularly in the
kinetoplast of organisms such as trypanosomes are of particular
interest.
DB293 is also one of the most exciting new minor groove
binding compounds discovered recently since it can form a
unique dimer structure that stacks in an antiparallel arrangement
in the minor groove to recognize sites that contain GC as well
as AT base pairs.1,15,16,18 In addition to its classical binding
monomer binding mode in AT sequences, this compound offers
an entirely new dimer interaction mode for sequence specific
recognition of mixed DNA sequences. Since it is readily taken
up into cells,19 it obviously offers unique opportunities to
develop new types of sequence-selective DNA targeted thera-
peutic agents. For these reasons, we are engaged in a series of
synthetic, biological, and biophysical studies to understand the
DNA complexes of these compounds in detail and to use this
information to design new compounds with additional and
improved therapeutic and DNA recognition abilities. As part
of this effort, thiophene derivatives with amidine cationic groups
(Figure 1), which are directly related to DB75 and DB293, have
been prepared and their DNA complexes characterized with a
broad array of different methods including X-ray determination
of a complex structure with DB818 bound at an AATT minor
groove site. We show in this paper that although the thiophene
derivatives differ by only a single atom from the parent furans,
they have significant shape changes that can make a large
difference in DNA complex formation depending on the other
groups in the molecular system and their interactions with DNA.
Materials and Methods
Buffers and Oligonucleotides. The double stranded polymers
poly(dAT)2 and poly(dGC)2 were purchased from Pharmacia
(U. S.), and were used in UV-Vis titration experiments. In
Biacore and CD experiments three 5′-biotin labeled hairpin
duplexes namely d(Biotin-CGAATTCGTCTCCGAATTCG)
(AATT hairpin), d(Biotin-CATATATA CCCCTATATATG)
(AT hairpin), d(Biotin-CGCGCGC TTTTGCGCGCG) (Figure
1) were purchased with HPLC purification (Midland Certified
Reagent Co., Midland, TX). The MES buffer used in these
experiments contained 0.010 M [2-(N-morpholino) ethane-
sulfonic acid] (MES), 0.001 M EDTA, 0.1 M NaCl, pH 6.25.
Compound Synthesis: 2-{2-[5(6)-Amidino]benzimidazolyl}-
5-[(4-amidino)phenyl] thiophene trihydrochloride (DB818,
Figure 1). A solution of potassium carbonate (4.4 g, 0.32 mole)
in 10 mL water was added to a stirred solution of 5-bromo-2-
(dimethoxymethyl)thiophene19 (3.3 g, 0.014 mole), 4-cy-
anophenylboronic acid (2.35 g, 0.016 mole) in 35 mL dioxane,
under nitrogen, and stirring was continued for 15 min. After
addition of Pd (PPh3)4 (0.32 g, 2mole%) the mixture was heated
at reflux for 8 h (TLC monitored), cooled, diluted with water
and extracted with 2 × 35 mL chloroform. The chloroform layer
was washed with water and stirred with 20% HCl (40 mL) for
35 min. The organic layer was washed with water, 5% sodium
bicarbonate, water, brine, and dried over sodium sulfate, passed
through a bed of silica, dried and the solvent reduced to yield
5-(4-cyanophenyl)-2-thiophene carboxaldehyde as a white solid
(12) Lacy, E. R.; Madsen, E. M.; Lee, M.; Wilson, W. D. In Small Molecule
DNA and RNA Binders; Demeunynck, M., Bailly, C., Wilson, W. D., Eds.;
Wiley-VCH: 2003; Vol. 2, pp 384-413.
(13) (a) Reddy, P. M.; Jindra, P. T.; Satz, A. L.; Bruice, T. C. J. Am. Chem.
Soc. 2003, 125, 7843-8. (b) Satz, A. L.; Bruice, T. C. Acc. Chem. Res.
2002, 35, 86.
(14) (a) Garoff, R. A.; Litzinger, E. A.; Connor, R. E.; Irene, F.; Armitage, B.
A. Langmuir 2002, 18, 6330-6337. (b) Wang, M.; Silva, G. L.; Armitage,
B. A. J. Am. Chem. Soc. 2000, 122, 9977-9986.
(15) (a) Tanious, F. A.; Hemmelberg, D.; Bailly, C.; Czarny, A.; Boykin, D.
W.; Wilson, W. D. J. Am. Chem. Soc. 2004, 126, 143-153. (b) Nguyen,
B.; Hammelberg, D.; Bailly, C.; Colson, P.; Stanek, J.; Brun, R.; Neidle,
S.; Wilson, W. D. Biophys. J. 2004, 86, 1028-1041. (c) Trent, J. O.; Clark,
G. R.; Kumar, A.; Wilson, W. D.; Boykin, D. W.; Hall, J. E.; Tidwell, R.
R.; Blagburn, B. L.; Neidle, S. J. Med. Chem. 1996, 39, 4554-4562.
(16) (a) Wang, L.; Bailly, C.; Kumar, A.; Ding, D.; Bajic, M.; Boykin, D. W.;
Wilson, W. D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 12-16. (b) Wang,
L.; Carrasco, C.; Kumar, A.; Stephens, C. E.; Bailly, C.; Boykin, D. W.;
Wilson, W. D. Biochemistry 2001, 40, 2511-2521. (c) Nguyen, B.; Tardy,
C.; Bailly, C.; Colson, P.; Houssier, C.; Kumar, A.; Boykin, D. W.; Wilson,
W. D. Biopolymers 2002, 63, 281-297. (d) Tanious, F. A.; Wilson, W.
D.; Wang, Lei.; Kumar, A.; Boykin, D. W.; Marty, C.; Baldeyrou, B.; Bailly,
C. Biochemistry 2003, 46, 13576-13586. (e) Bailly, C.; Tardy, C.; Wang,
L.; Armitage, B.; Hopkins, K.; Kumar, A.; Schuster, G. B.; Boykin, D.
W.; Wilson, W. D. Biochemistry 2001, 40, 9770-9779. (f) Bailly, C.;
Dassonneville, L.; Carrosco, C.; Lucas, D.; Kumar, A.; Boykin, D. W.;
Wilson, W. D. Anti-Cancer Drug Des. 1999, 14, 47-60.
1
2.35 g (77%), mp 208-210 °C dec, H NMR (DMSO-d6): δ
9.95 (s, 1H), 8.06 (d, 1H, J ) 4 Hz), 7.98 (d, 2H, J ) 8.8 Hz),
7.91 (d, 1H, J ) 8.8 Hz), 7.88 (d, 1H, J ) 4 Hz); 13C NMR
(DMSO-d6): δ 184.0, 149.6, 143.5, 138.4, 136.5, 133.0, 127.2,
126.7, 118.0, 111.4; MS: m/e 213 (M+). Anal. Calcd. for
C12H7NOS: C, 67.58; H, 3.30; N, 6.56. Found: C, 67.31; H,
3.41; N, 6.44.
A mixture of the thiophene aldehyde (2.13 g, 0.01 mol), 3,4-
diaminobenzonitrile (1.33 g, 0.01 mol) and 1,4-benzoquinone
(1.08 g, 0.01 mol) in 50 mL dry ethanol was heated under reflux
(N2 atmos.) for 8 h, cooled, diluted with ether, and filtered. The
solid was collected and stirred with 1:3 mixture of EtOH and
ether for 20 min., filtered, washed with ether, and vacuum-dried
at 70 °C for 12 h to yield 2.10 g (60%) of the yellow
(17) Mazur, S.; Tanious, F. A.; Ding, D.; Kumar, A.; Boykin, D. W.; Simpson,
J. J.; Neidle, S.; Wilson, W. D. J. Mol. Biol. 2000, 300, 321-337.
(18) (a) Wang, L.; Kumar, A.; Boykin, D. W.; Bailly, C.; Wilson, W. D. J.
Mol. Biol. 2002, 317, 361-374. (b) Lansiaux, A. L.; Tanious, F.; Mishal,
Z.; Dassonneville, L.; Kumar, A.; Stephens, C. E.; Hu, Q.; Wilson, W. D.;
Boykin, D. W.; Bailly, C. Cancer Res. 2002, 62, 7219-7229.
(19) (a) Lansiaux, A. L.; Dassonneville, L.; Faxompre, M.; Kumar, A.; Stephans,
C. E.; Bajic, M.; Tanious, F. A.; Wilson, W. D.; Boykin, D. W.; Bailly, C.
J. Med. Chem. 2002, 45, 1994-2002. (b) Lansiaux, A.; Tanious, F. A.;
Mishal, Z.; Dassonneville, L.; Kumar, A.; Stephens, C. E.; Hu, O.; Wilson,
W. D.; Boykin, D. W.; Bailly, C. Cancer Res. 2000, 62, 7219-7229.
9
13660 J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004

Heterocycle−DNA Interactions
A R T I C L E S
brown 2-{2-[5(6)-cyano(benzimidazolyl)]}-5-(4-cyanophenyl)
thiophene, mp 315-17 °C dec, H NMR (DMSO-d6/D2O): δ
concentration. After 30 min incubation at 37 °C, the digestion
was initiated by the addition of 2 µL of a DNase I solution
adjusted to yield a final enzyme concentration of about 0.01
unit/ml in the reaction mixture. After 3 min, the reaction was
stopped by freeze-drying. Samples were lyophilized and resus-
pended in 5 µL of an 80% formamide solution containing
tracking dyes. The samples were then heated at 90 °C for 4
min and chilled in ice for 4 min prior to electrophoresis. DNA
cleavage products were resolved by polyacrylamide electro-
phoresis under denaturing conditions (0.3 mm thick, 8%
acrylamide containing 8 M urea). After electrophoresis (about
2.5 h at 60 W, 1600 V in Tris-Borate-EDTA buffered solution),
gels were soaked in 10% acetic acid for 10 min, transferred to
Whatman 3 MM paper, and dried under vacuum at 80 °C. A
Molecular Dynamics 425E PhosphorImager was used to collect
data from the storage screens exposed to dried gels overnight
at room temperature. Baseline-corrected scans were analyzed
by integrating all the densities between two selected boundaries
using ImageQuant 3.3 software. Each resolved band was
assigned to a particular bond within the DNA fragments by
comparison of its position relative to sequencing standards
generated by treatment of the DNA with dimethylsulfate
followed by piperidine-induced cleavage at the modified guanine
bases.
Thermal Melting. Thermal melting experiments were con-
ducted with Cary 3 or Cary 4 spectrophotometers interfaced to
microcomputers as previously described.15,16 A thermistor fixed
into a reference cuvette was used to monitor the temperature.
The DNA is added to 1 mL of buffer (MES with 0.1 M NaCl
added) in 1 cm path length quartz cells and the concentration
determined by measuring the absorbance at 260 nm. Experi-
ments were generally conducted at a concentration of 5 × 10-5
M base pairs for poly(dAT), poly(dAT)2 and 5 × 10-5 M duplex
for d(CGCGAATTCGCG)2. For experiments with complexes
a ratio of 0.3 compound per base pair for poly(dAT) and poly-
(dAT)2 and a ratio of one compound per oligomer duplex for
d(CGCGAATTCGCG)2 was used.
Spectrophotometric Titrations. CD spectra were obtained
with a computer controlled Jasco J-710 Spectrophotometer in
a 1 cm quartz cell in MES buffer. The desired ratios of
compound to DNA were obtained by adding compound to the
cell containing a constant amount of DNA. Absorption spectra
were collected on a Cary 4 spectrophotometer under the same
conditions. During the spectrophotometric titrations, the spectra
of the complex at different DNA (bp) to compound ratios were
collected in the region above 300 nm to avoid DNA absorption
interference. The DNA was added to a cell containing a constant
amount of compound until all compound was bound.
Biosensor-Surface Plasmon Resonance. Experiments were
performed on a BIAcore 3000 instrument and BIAcore sensor-
chips with streptavidin linked to dextran were used for DNA
immobilization by using established methods.17,18,20 DNA im-
mobilization was performed at a flow rate of 5 µL/min with
DNA concentrations of 500-1000 pM. The amounts of DNA
hairpin immobilized on the flow cells were in the range of 400
RUs and one flow cell was always left as a blank for reference.
The sensorgrams (SPR angle in resonance units, RU, vs time)
were collected at flow rate of 10, 20, or 50 µL/min with the
1
8.03 (d, 1H, J ) 1.5 Hz), 7.91 (d, 2H, J ) 8.4 Hz), 7.90 (d,
1H, J ) 3.9 Hz), 7.85 (d, 2H, J ) 8.4 Hz), 7.77 (d, 1H, J ) 3.9
Hz), 7.70 (d, 1H, J ) 8.4 Hz), 7.55 (d, 1H, J ) 1.5 Hz, J ) 8.4
Hz); 13C NMR (DMSO-d6): δ 149.1, 144.1, 137.0, 133.4, 132.8,
129.1, 126.9, 125.9, 125.6, 119.5, 118.2, 110.3, 104.1; MS: m/e
326 (M+). Anal. Calcd. for C19H10N4S‚1.25H2O: C, 65.40; H,
3.61; N, 16.06. Found: C, 65.32; H, 3.67; N, 16.47.
The bis-nitrile (1.8 g, 0.0055 mole) in 75 mL ethanol was
saturated with HCl gas at 0 °C and stirred at room temperature
for 8 days (TLC monitored). Ether was added and the resulting
yellow imidate ester was filtered, washed with ether, and
vacuum-dried at 30 °C for 5 h. The imidate ester hydrochloride
[2.43 g (84%). 0.65 g (0.0012 mole)] was suspended in 65 mL
ethanol, saturated with ammonia at 0 °C, and stirred at room
temperature for 24 h. The solvent was removed, ether: ethanol
(6:1) added and the solid filtered. The yellow amidine was
suspended in 30 mL ethanol and saturated with HCl gas. The
yellow crystalline amidine hydrochloride salt was filtered,
washed with ether and dried in a vacuum at 70 °C for 12 h, to
1
yield 0.39 g (68.5%), mp >350 °C dec, H NMR (DMSO-d6/
D2O): δ 8.05 (d, 1H, J ) 1.5 Hz), 7.93 (d, 1H, J ) 3.6 Hz),
7.92 (d, 2H, J ) 8.4 Hz), 7.84 (d, 2H, J ) 8.4 Hz), 7.45 (d,
1H, J ) 3.6 Hz), 7.73 (d, 1H, J ) 8.4 Hz, 7.63 (dd, 1H, J )
1.5 Hz, J ) 8.4 Hz); 13C NMR (DMSO-d6): δ 166.6, 165.8,
149.3, 146.7, 141.2, 138.4, 138.3, 131.9, 131.3, 129.7, 127.9,
127.7, 126.9, 123.9, 123.0, 116.3, 115.8. FABMS: m/e 361-
(M++1). Anal. Calcd. for C19H16N6S‚3HCl‚0.25H2O: C, 48.11;
H, 4.14; N, 17.71. Found: C, 48.18; H, 4.56; N, 17.77.
Purification Of DNA Restriction Fragments And Radio-
labeling for Footprinting. Plasmids were isolated from E. coli
by a standard sodium dodecyl sulfate-sodium hydroxide lysis
procedure and purified by banding in CsCl-ethidium bromide
gradients. The 117-bp and 265-bp DNA fragments were
prepared by 3′-[32P]-end labeling of the EcoRI-PVuII double
digest of the pBS plasmid (Stratagene) using R-[32P]-dATP
(Amersham, 3000 Ci/mmol) and AMV reverse transcriptase
(Roche). The 160-bp tyrT fragment was obtained from plasmid
pKM∆98 after digestion with the restriction enzymes EcoRI
and AVaI. In each case, the labeled digestion products were
separated on a 6% polyacrylamide gel under nondenaturing
conditions in TBE buffer (89 mM Tris-borate pH 8.3, 1 mM
EDTA). After autoradiography, the requisite band of DNA was
excised, crushed, and soaked in water overnight at 37 °C. This
suspension was filtered through a Millipore 0.22 µm filter and
the DNA was precipitated with ethanol. Following washing with
70% ethanol and vacuum-drying of the precipitate, the labeled
DNA was resuspended in 10 mM Tris adjusted to pH 7.0
containing 10 mM NaCl.
DNase I footprinting, Electrophoresis, and Quantitation.
Bovine pancreatic deoxyribonuclease I (DNase I, Sigma Chemi-
cal Co.) was stored as a 7200 units/mL solution in 20 mM NaCl,
2 mM MgCL2, 2 mM MnCl2, pH 8.0. Stock solutions of DNase
I were kept at -20 °C and freshly diluted to the desired
concentration immediately prior to use. Footprinting experiments
were performed essentially as recently described.15,16 Briefly,
reactions were conducted in a total volume of 10 µL. Samples
(3 µL) of the labeled DNA fragments were incubated with 5
µL of the buffered solution containing the ligand at appropriate
(20) (a) Davis, T. M.; Wilson, W. D. Anal. Biochem. 2000, 284, 348-353. (b)
Davis, T. M.; Wilson, W. D. Methods Enzymol. 2001, 340, 22-51.
9
J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004 13661

A R T I C L E S
Mallena et al.
samples dissolved in degassed MES10 buffer with 5 × 10-3
%
Table 1. Crystallographic Data
v/v Surfactant P20. 10 mM glycine, pH 2.0 solution was also
used in surface regeneration. The RU values were averaged over
a one-minute time span in the steady state response region and
were converted to r (moles of compound bound per mole of
DNA duplex) for binding analysis as previously described.15-18
The binding constants of the compounds were obtained from
fitting a plot of r vs free compound concentration (from the
flow solution with one- and two-site interaction models (for
the one-site model, K2 ) 0)
DB818−DNA complex
DNA sequence
space group
unit cell dimensions (Å)
resolution (Å)
d(CGCGAATTCGCG)2
P212121
a)25.677, b)39.921, c)65.185
17-1.77
1.54178
2.80
wavelength (Å)
V
m (Å3/Da)
measured reflections > 1σ
unique reflections
completeness (%)
in the outer shell (%)
I/σ (I) for the data set
I/σ (I) for the outer shell
data redundancy
43881
6667
95.2
92.5 (1.83-1.77 Å)
70.18
13.16
6.58
4.3
r ) (K1‚Cfree + 2‚K1‚K2‚C2free)/(1 + K1‚Cfree + K1‚K2‚C2
)
free
(1)
R
R
merge (%)
merge for the outer shell (%)
13.5
where K1 and K2 are the equilibrium binding constants; Cfree is
the concentration of the compound in the flow solution. This
concentration is fixed by the flow solution concentration because
the solution is continuously refreshed by flow over the chip
surface; r ) RUeq/RUmax and represents moles of compound
bound per mole of DNA hairpin, where RUeq is the averaged
response at the steady-state level, and RUmax is the maximum
response for binding one molecule of compound and DNA per
binding site and is predicted with the following equation
refinement
resolution range
reflections
reflections with Fo > 4σ(Fo)
parameters refined
restraints
8.0-1.77
5894
5495
2307
2538
R-factor {Fo > 4σ(Fo)} (%)
21.50
29.06
R
free {Fo > 4σ(Fo)} (%)
rms deviations
bond lengths (Å)
bond angle distances (Å)
no. of atoms
0.008
0.028
DNA
DB818 ligand
magnesium-water complex
full occupancy waters
mean temperature factor of atoms (Å2)
DNA
DB818 ligand
magnesium-water complex
waters
486
26
7
RUmax ) (RUDNA‚MWcompound‚RIIr)/MWDNA
(2)
where RUDNA is the amount in response units of DNA
immobilized, MW is the molecular weight of compound and
DNA, respectively, and RIIr is the refractive index increment
ratio of compound to refractive index increment of DNA.20 The
RIIr values for these compounds are between 1.25 and 1.40.
Isothermal Titration Calorimetry (ITC). The ITC experi-
ments were carried out with a MicroCal VP-ITC (MicroCal
Inc., USA) isothermal titration calorimeter interfaced to a PC
for instrument control and data collection with Origin 5.0
software. In a typical titration 10 µL of a 0.15 mM compound
solution was added every 300 s to a total of 29 injections to
DNA in the sample cell at 10 µM duplex. The heat associated
with each injection is observed as a peak that corresponds to
the power required to keep the sample and reference cells at
identical temperature.21 The areas of the peaks produced over
the course of a titration can be converted to heat output per
injection by integration and correction for cell volume, sample
concentration, and subtraction of the heats of dilution. Titrations
of compound into buffer were carried out to determine the
contribution to the binding enthalpy from the heat of compound
dilution.17,18,21 Each titration was repeated 3 to 5 times. Finally,
binding enthalpies were obtained by fitting the corrected data
to an appropriate binding model.17,18,21
Crystallization and Data Collection. The oligonucleotide
d(CGCGAATTCGCG) (HPLC-purified, Eurogentec Ltd.) was
annealed at 85 °C for 6 min in 30 mM sodium cacodylate buffer
at pH 7.0 before use. Crystals were grown by vapor diffusion
at 285 K from hanging drops. The crystal used for data
collection was obtained from a 6 µL drop containing 10 mM
MgCl2, 0.5 mM double-stranded DNA, 0.75 mM DB818
compound, 5% v/v (()-2-methyl-2,4-pentanediol (MPD), 1.0
mM spermine tetrahydrochloride, in 30 mM sodium cacodylate
buffer at pH 7.0, equilibrated against a reservoir of 700 µL of
54
24.8
29.6
21.5
33.0
50% v/v MPD solution. Crystals grew as large rods within three
weeks. Typical crystals were approximately 1.0 × 0.3 × 0.3
mm in dimension.
A crystal was mounted in a loop and flash-frozen in a nitrogen
gas stream at 103 K and maintained at that temperature using
an Oxford Cryostream (Oxford Cryosystems). X-ray diffraction
data were collected on a Rigaku RAXIS IV image plate detector
with CuKR radiation from a Rigaku RU200 rotating anode
generator (100 mA, 50 kV) and an Osmic focusing mirror
system. The crystal-to-detector distance was set at 110 mm. Each
frame was exposed for 35 min at 2.5° oscillation. Seventy-three
frames were collected for the structure, from 0° to 180°, which
resulted in a maximum resolution of 1.77 Å. Indexing and data
processing were carried out using the program package DENZO
and SCALEPACK.22
Structure Solution and Refinement. The structure was
solved by molecular replacement. A suitable isomorphous DNA
model was chosen for structure solution and refinement, this
was the berenil-d(CGCGAATTCGCG)2 complex (PDB entry
2DBE or NDB entry GDL009).23 Only the DNA part of this
model was used for structure solution and refinement. Structure
solution involved rigid-body refinement using the CNS program
package24 starting at 4.0 Å resolution, and this was gradually
(22) Otwinowski, Z. M.; Minor, W. Methods Enzymol. 1997, 276, 307-326.
(23) Brown, D. G.; Sanderson, M. R.; Skelly, J. V.; Jenkins, T. C.; Brown, T.;
Garman, E.; Stuart, D. I.; Neidle, S. EMBO J. 1990, 9, 1329-1334.
(24) Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros, P.;
Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.; Pannu,
N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L. Acta
Crystallogr. D 1998, 54 (Pt 5), 905-21.
(21) Wiseman, T.; Willston, S.; Brandts, J. F.; Lin, L.-N. Anal. Biochem. 1989,
179, 131-137.
9
13662 J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004

Heterocycle−DNA Interactions
A R T I C L E S
increased to the highest resolution at 1.77 Å. The R-factor at
this point was 40.5% and Rfree was 41.5%. The Fourier map
was generated using CNS and the region around the minor
groove of DNA was observed using CHAIN25 to check whether
there is a continuous electron density region. This density was
observed in the minor groove. At this stage, repeated simulated
annealing and positional refinement with gradual increase in
resolution was performed again using CNS. The R-factor was
34.0% and the Rfree was 30.0% before the ligand was included.
The model of ligand was added to the structure at 2.4 Å
resolution. The hexa-coordinated magnesium-water complex was
included into the structure at 2.0 Å resolution; the R-factor and
Rfree at this point were 31.9% and 30.8%, respectively.26 The
refinement was continued to 1.77 Å resolution; the R-factor and
Rfree were 33.4% and 32.2%, respectively, at this stage of the
refinement. The remaining part of the refinement was transferred
to SHELX-97 for atomic refinement.27 The refinement began
at 4.0 Å until it reached 1.77 Å. The Fourier map was checked
during refinement using Xfit portion of the XtalView program.28
Water refinement was carried out after this stage. The automated
solvent searching program SHELXWAT was then used to locate
molecules, as well as manual peak-picking to locate water
molecules. The water molecules were accepted according to
B-factor value (<50 Å2), strong electron density, standard
distance and geometry. The SHELX-97 refinement was con-
tinued to give the final R-factor of 21.5% and Rfree of 29.1%.
The atomic coordinates and structure factors have been deposited
to the RCSB Protein and Nucleic Acid Data Banks with
accession number 1VZK. Crystallographic data for the structure
are given in Table 1.
Figure 2. Thermal melting profiles of duplex d(CGCGAATTCGCG)2 in
the absence (circles) and in the presence of DB75 (squares), DB293 (up
triangles), DB351 (diamonds) and DB818 (down triangles). Experiments
were conducted in MES10 buffer.
in Figure 1. Clearly the amidine-benzimidazole-thiophene
system provides a very favorable unit for strong molecular
recognition of AT sequences in DNA.
DNase I Footprinting. Relative affinity and Sequence
Specificity. Footprinting is the method of choice for defining
sequence specificity in a broad sequence context for compounds
that bind to DNA. Footprinting results have been published for
both furan derivatives in Figure 1, DB7516c,f and DB293,16d,e
but not for the two thiophene analogues. The DNA sequence
recognition specificities of these four compounds are directly
compared with three DNA fragments of 117, 160 and 265-bp
in length that provide a diversity of potential binding sites
(Figure 3). As expected for diamidines with a curvature that
closely matches that of the DNA minor groove, both of the
furans have strong footprints and the densitometric analysis of
the gels (Supporting Information, Figure S1) indicates that the
binding sites essentially correspond to AT sequence regions.
Results
Thermal Melting. Relative Affinity Ranking. We have found
for a wide array of different diamidines that determination of
melting temperature increases (Tm of complex - Tm of free
DNA, where Tm is the maximum in a derivative plot of dA260
/
dT) provides a rapid and accurate method for ranking com-
pounds according to their binding affinity.1 Increases in Tm
values for compound-DNA oligomer complexes are shown in
Figure 1 and example Tm curves are in Figure 2. Large increases
in melting temperature are observed for all of the compounds
with the oligomer containing an -AATT- binding site. Under
the same conditions, no significant increase in the Tm of a DNA
duplex oligomer with an alternating CG sequence was observed
with these compounds.
The footprinting profiles obtained with DB818 are distinct
from those obtained with DB293 but are comparable to those
seen with DB75 and DB351. Densitometric results for the gels
for DB818 and DB293 are shown in Figure S1 (those for DB75
and DB351 were omitted for clarity) and all results are
summarized in Table 2. All results show that DB818 binds to
AT sites, as is the case with the other three compounds, but
there is a key difference, the footprints with DB818 are much
stronger than those at the same concentrations for the other
compounds (compare the 1 µM lanes in Figure 3). The footprints
with DB818 are easily seen at 0.5 µM whereas a concentration
of about 5 µM is required to detect similar effects with DB75
and DB351. In contrast, DB293 can recognize additional types
of sequences, mainly -ATGA- sites, and this has been linked
to its ability to form a stacked antiparallel dimer in the minor
groove at ATGA (Wang et al., 2000, 2001, 2002). DB293 can
bind to AT sequences in a monomer complex but it binds more
strongly to -ATGA- sequences as a dimer.
With the AT DNA sequences the furan to thiophene conver-
sion for DB75 to DB351 makes little change in the ∆Tm values
but the same conversion for DB293 to DB818 results in a
surprisingly large Tm increase for a single atom change. In the
same manner conversion of a phenyl to a benzimidazole for
DB75 to DB293 results in a modest Tm increase but the same
change in the thiophene series (DB351-DB818) results in a
significantly larger ∆Tm. In additional Tm determinations with
AT containing DNA polymers, DB818 again had abnormally
large ∆Tm values (not shown) compared to the other compounds
Spectroscopy. Relative Affinities and Binding Modes. Ab-
sorption spectral titrations of the compounds in Figure 1 with
AT DNAs generally result in large decreases in absorbance with
slight red shifts when the compound to DNA ratio is high but
larger red shifts and an absorbance increase or no significant
(25) Sack, J. S. J. Mol. Graphics 1988, 6, 224-225.
(26) Tsunoda, M.; Karino, N.; Ueno, Y.; Matsuda, A.; Takenaka, A. Acta
Crystallogr. D 2001, 57, 345-348.
(27) Sheldrick, G. M.; Schneider, T. R. Methods Enzymol. 1997, 276, 319-
343.
(28) McRee, D. E. J. Struct. Biol. 1999, 125, 156-65.
9
J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004 13663

A R T I C L E S
Mallena et al.
Figure 3. DNase I footprinting with the (a) 117-bp, (b) 160-bp and (c) 265-bp DNA fragments at graded concentrations of the four DB compounds. The
concentration (µM) of the drug tested is shown at the top of the appropriate gel lanes. Tracks labeled “cont” contained no drug. Tracks labeled “G” represent
dimethyl sulfate-piperidine markers specific for guanine. Numbers at the right side of the gel refer to the sequence and nucleotide position of the different
fragments.
Table 2. Sequences of the DNA Fragments Showing DNase I Protection in the Presence of DB818 and DB293, Inferred from Differential
Cleavage Plots.
DB818
DB293
117-mer
160-mer
5′-ATTGTAATA (20-28)a
5′-TTTT (63-66)
+++b
+
5′-TGTAATA (20-26)
+
5′-TTT (29-31)
5′-TAAAGTGTTACGTT (56-69)
5′-TCATATCA (85-92)
+
++
++
5′-ATCCGT (21-26)
5′-GTGT (62-65)
5′-ATCATATCA (85-93)
++
+
++
265-mer
5′-TAAAGG (61-66)
5′-ATT (76-78)
5′-ATTACG (90-95)
5′-ATGA (101-104)
+
+
++
+
5′-ATTA (90-93)
5′-ATGACCATGA (95-104)
+
++
a The position of the sequence is indicated in parentheses. b +++, ++, and + indicate strong, medium and weak protection.
change as more DNA is added and only the strongest binding
sites are populated. Example spectra for DB818 and DB351
with the AATT oligomer (Figure 1) are shown in Figure 4. This
type of behavior does not prove a binding mode but it is
consistent with results for other similar compounds that bind
in the DNA minor groove in AT sequences. The initial decrease
is characteristic of stacking at high compound-DNA ratios
followed by partitioning of the compound to the strong binding
sites in the minor groove as the ratio of compound to DNA is
decreased. At high ratios of DNA to compound in the titrations
only the strongest binding sites are populated.
Titrations of DB351 and DB818 with the AT polymer, poly-
(dA-T)2, give qualitatively similar results to those with the
AATT oligomer (Supplementary Figure S2). The results with
the GC polymer, poly(dG-C)2 (Supplementary Figure S3) also
illustrate that the compounds can bind to GC sequences but only
at higher concentrations. The absorbance decreases on addition
of the GC DNA, as expected, but there was no significant
increase in absorbance as more DNA was added. Similar
behavior, which has been observed for other minor groove
binding compounds, is characteristic of stacking interactions but
no changes characteristic of a minor groove complex is observed
even at high ratios of GC DNA to compound.
Circular dichroism spectral changes provide very useful
information for defining the binding mode of unfused aromatic
cations, such as those in Figure 1, with DNA.29,30 For minor
groove complexes, strong positive induced CD signals in the
compound absorption region are expected on complex formation
with DNA. The CD spectral titrations of DB351 and DB818
(29) Lyng, R.; Rodger, A.; Norden, B. Biopolymers 1992, 32, 1201-1214.
(30) Rodger, A.; Norden, B. In Circular Dichroism and Linear Dichroism;
Oxford University Press: New York, 1997.
9
13664 J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004

Heterocycle−DNA Interactions
A R T I C L E S
Figure 4. Spectrophotometric titration of DB351 (A) and DB818 (B) with
AATT hairpin DNA. The titrations were conducted in MES10 buffer at 25
°C. The concentration of the compound in each case is 2.5 × 10-5 M. In
DB351, the ratios of the compound to AATT hairpin DNA from the top to
the bottom are free compound, 4.24, 2.12, 1.42, 1.06, 0.85, 0.71, 0.61, 0.53,
and 0.47 (A). In DB818, the ratios of the compound to AATT hairpin DNA
from the top to the bottom are free compound, 42.44, 21.26, 14.20, 8.50,
6.08, 4.74, 3.88, 3.29, 2.85, 2.52, 2.25, 2.04, 1.79, 1.72, 1.59, 1.48, and
1.39 (B).
Figure 5. CD spectra of DB351 (A) and DB818 (B) with the AATT hairpin
DNA (10 µM bp). The experiments were conducted in MES10 buffer at
25 °C. In DB351 the ratios of compound to DNA from the top to the bottom
at 375 nm are 0, 0.1, 0.2, 0.3, 0.4, 0.6, 0.8, 1.0, 1.2, 1.5, and 2.0. In DB818
the ratios of compound to DNA from the top to the bottom at 375 nm are
0, 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, 1.6, 1.8, and 2.0.
with AATT hairpin DNA have large positive induced CD signals
above 300 nm where DNA does not absorb (Figure 5). This
large induced CD is characteristic of a minor groove binding
mode. CD spectra for the complexes of DB818 and DB351 with
the alternating AT polymer also display large induced CD
spectral changes as expected for minor groove complexes
(Supporting Information, Figure S4) and similar changes have
been observed for DB75 and DB293.15,16 With both DNAs the
CD changes are characteristic for minor groove complex
formation. It should be noted that with both AT DNAs the CD
changes are significantly larger for DB818 than for DB351 in
agreement with the Tm and footprinting results that indicate
much stronger AT DNA interactions for DB818. Over the same
range of ratios of compound to DNA, much smaller (DB818)
or essentially no (DB351) CD spectral changes are seen in the
compound absorption region on addition of GC polymer DNA
(Figure S4), as expected for weak nonspecific binding interac-
tions.
Quantitative Comparison of DNA Interactions. Biosensor-
SPR. To quantitatively compare the interactions of the com-
pounds of Figure 1 with specific DNA sequences, biosensor
SPR experiments were carried out. 5′-biotin labeled hairpin
DNA duplexes with alternating AT and GC sequences as well
as an AATT sequence (see Materials) were immobilized on three
flow cells of a BIAcore sensor chip and the fourth flow cell
was left blank for reference subtraction. Sensorgrams for the
interaction of DB818 and DB351 with the AATT sequence are
compared in Figure 6 and, for two such closely related
compounds, the sensorgrams are strikingly different. Both the
rates of association and dissociation of DB351 are too fast to
Figure 6. SPR sensorgrams of DB351 (A) and DB818 (B) with the AATT
hairpin DNA. The compound concentrations for DB351 from bottom to
top are 0 to 0.8 µM (A). The compound concentrations for DB818 from
bottom to top are 0 to 10 nM (B).
be determined at the limit of resolution in SPR experiments. A
steady-state SPR response is reached for DB351 in a few
seconds after initiation of compound or buffer flow. With
9
J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004 13665

A R T I C L E S
Mallena et al.
Table 3. Binding Constants and Experimental Thermodynamic Values at 298 K for Binding of Furan and Thiophene Compounds to DNAa
AATT
ATAT
CGCG
∆
G
∆
H
T
‚∆
S
K
(M-
∆
G
K
(M-
∆G
(kcal/mol)
1
1
1
compd
K (M-
)
(kcal/mol)
(kcal/mol)
(kcal/mol)
)
(kcal/mol)
)
DB75
1.4 × 107
7.5 × 106
1.1 × 107
2.8 × 108
-9.7
-9.4
-2.2
-3.0
-3.6
-6.7
7.5
6.4
6.0
4.8
6.0 × 107
-10.6
-8.8
- -
0.8 × 106
-8.0
-4.7
- -
DB351
DB293b
DB818
2.8 × 106
3.0 × 103
-9.6
- -
- -
- -
-11.5
2.7 × 108
-11.5
- -
a Experiments were carried out in MES10 buffer at 25 °C; The K and ∆G values are from biosensor-SPR experiments. Experimental error limits for
DB75, DB351, and DB293 are approximately (10%. The slow kinetics and very strong binding of DB818 increases the error to approximately 20%. The
∆H values are from ITC experiments and have an experimental error of approximately (10%.
Figure 7. RU values from the steady-state region of SPR sensorgrams
were converted to r by r ) RU/RUmax and are plotted against the unbound
compound concentration (flow solution) for DB75, DB293, DB351, and
DB818 binding to the AATT hairpin DNA. The lines in the Figure were
obtained by nonlinear least-squares fits of data by using eq 1. Experiments
were conducted in MES buffer with 0.1 M NaCl added at 25 °C. The inset
shows the same binding curves at higher concentration.
Figure 8. ITC curves for the binding of DB351 (150 µM) to the duplex
d(GCGAATTCGC)2 (10 µM base pairs) in MES10 at 25 °C (A). Results
were converted to molar heats and plotted against the compound/DNA molar
ratio (B). The line shows the fit to the results and gives best fit ∆H (Table
3) values for binding.
DB818 a steady state is only reached after over 100 seconds at
concentrations near the saturation level for the compound. The
responses for DB75 and DB293 are more similar to those for
DB351. The differences in interaction strength for all four
compounds can be more easily visualized in the plot of r (moles
of compound bound/mole of hairpin DNA) versus Cfree, the free
compound concentration (Figure 7). As can be seen, the
concentration for DB818 to reach saturation is at least 10 times
lower than for the other compounds.
whereas that for DB818 is at least 10 times higher and may be
underestimated because of its slow association rate at low
concentration. Similar behavior is seen for the AT hairpin
oligomer for all the compounds. All four compounds bind to
the GC hairpin 100 or more times more weakly than with the
AT sequences in agreement with, spectral changes, their lack
of footprints and low Tm with GC sequences. The GC
interaction may account for part of the secondary binding
interactions described above for the AATT oligomer sequence.
Detailed Thermodynamics of the Interactions. ITC Experi-
ments. To understand the energetic basis of the interaction
differences between DB818 and the other compounds in Figure
1 in more detail, ITC experiments were conducted to determine
the observed binding enthalpy, ∆H (Figure 8) and to allow
calculation of ∆S (from ∆G ) ∆H - T‚∆S). Integration with
respect to time of the heats produced on serial injection of ligand
into the -AATT- duplex yields the corresponding binding
isotherm. When the K value is not too large (with respect to
1/[compound]) these curves can yield a K value. Figure 8B
shows the binding isotherms obtained by titrating DB351 into
DNA, after subtracting the heat of dilution for addition of the
compound into buffer. The observed ∆H value for binding was
The compounds all have a strong binding constant (K > 107
M-1) as well as weaker binding that can be seen at high
concentrations as a gradual increase in the r versus Cfree plots.
We thus found better fits to all SPR results with a nonidentical
two-site binding model. At concentrations above 1 microM with
the compounds in Figure 1 we begin to have baseline subtraction
problems, and it is difficult to quantitate the weak secondary
binding. It is clear that the K2 values are near 1 × 105 M-1
with DB75 and DB351, 100 times less than K1, and are even
smaller with the benzimidazole derivatives. The strong binding
constants (K1) determined by fitting the curves in Figure 7 with
eq 1 (two-site model-see Materials) are collected in Table 3.
The binding constants for DB75, DB293, and DB351 with the
AATT sequence are between approximately (1-3) × 107 M-1
,
9
13666 J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004

Heterocycle−DNA Interactions
A R T I C L E S
Figure 9. Representative ITC results for addition of DB818 to the AATT
hairpin DNA at 35 °C. Experiments were conducted in MES10 buffer.
Essentially all DB818 added in this titration range is bound to the AATT
site and the average value of the integrated peaks was used to calculate
∆H in Table 3.
Figure 11. (A) Comparison of the optimized structures of DB818 and
DB293: calculations were carried out using the MM2 force field parameter
set. The thiophene, DB818 (shown as green) has a significantly less concave
inner surface, as shown by the larger N...N amidinium distance of 14.4 Å
compared to 12.9 Å in the oxygen analogue, DB293 (shown as gray). (B)
Overlay of the optimized structure of DB818 and DB293 into the AATT
binding site of d(CGCGAATTCGCG)2.
strong site are collected in Table 3. The K1 value can be obtained
from the DB351 ITC titration curve and the value is in good
agreement with that from SPR experiments. Even after subtrac-
tion of the heat of dilution for DB351, the ITC curve does not
go to zero at ratios above one due to the weak secondary
interactions that were also observed in SPR experiments at high
ratios. A less extensive titration curve (Figure 9) was obtained
for DB818 due to precipitation of the complex as saturation of
the AATT site was approached at ITC concentrations. Only the
∆H value for binding can be obtained from these results because
of the limited range and large K value. With the K from SPR
experiments the ∆G and ∆S values can be also be calculated
(Table 3). As can be seen from Table 3, replacing O by S in
DB75 yielded only small changes in ∆G and ∆H values for
binding to DNA, whereas replacing O by S in DB293 yielded
much more favorable (more negative) ∆G and ∆H values. Thus
the enhanced binding of DB818 to AATT is largely due to a
more favorable enthalpy contribution and the entropy of binding
for DB818 is actually less favorable than with the other
compounds.
Crystal Structure of the DNA-DB818 Complex. The crystal
structure of the DB818-d(CGCGAATTCGCG)2 complex shows
that the ligand is bound mostly within the AATT minor groove
region of the B-type structure of the duplex DNA. The ligand
is twisted along its length, with the four torsion angles defining
its conformation having values of 42° (amidine-phenyl), 16°
(phenyl-thiophene), 11° (thiophene-benzimidazole) and 24°
(benzimidazole-amidine), on going from the phenyl end to the
benzimidazole end of the molecule (Figure 10A). The net effect
Figure 10. (A) Crystal structure of DB818 bound to the duplex sequence
d(CGCGAATTCGCG)2. The view is into the minor groove with DB818 C
in green, N in blue and S in yellow; DNA C in gray, N in blue, O in red
and P in yellow. The benzimidazole is at the bottom of this view and the
two amidines are clearly seen at the ends of the bound compound. (B) Close-
up view of the bifurcated pair of hydrogen bonds (2.7 and 2.9 Å), between
the benzimidazole inner-facing ring nitrogen atom and O2 atoms of the
central two thymines in the crystal structure of the complex between DB818
and d(CGCGAATTCGCG)2.
obtained from fitting the curve with a two binding site model
as with the SPR results described above (see Materials and
Methods). Only the ∆H values for binding of the compounds
to the strong binding site and ∆H values determined for the
9
J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004 13667

A R T I C L E S
Mallena et al.
of this is that the terminal amidinium groups are quite twisted
with respect to each other.
for the large equilibrium constants obtained with these systems.
The compounds, particularly DB818, do not have strong enough
spectroscopic signals to allow accurate determination of binding
constants of the magnitude in Table 3. The only method that
allowed determination of equilibrium constants for all com-
pounds was the biosensor-SPR system. The interaction differ-
ences between DB818 and the other compounds of Figure 1
can be visualized in the very different footprints in Figure 3
and in the sensorgrams in Figure 7. At 0.1 M NaCl DB75,
DB351, and DB293 all have equilibrium constants near 1 ×
107 M-1, a value typical of dications in this size range under
these conditions, while DB818 has a K of over 1 × 108 M-1
(Table 3). Conversion of the furan of DB293 to a thiophene or
conversion of a phenyl of DB351 to a benzimidazole yields
the large increase in binding constant. These combined results
indicate that both parts of the benzimidazole-thiophene module
of DB818 are required for the large increase in DNA binding
constant.
There are a number of hydrogen-bond contacts between inner-
facing atoms of the ligand and base edges that partially account
for the affinity and specificity of binding. The inner nitrogen
atom of the benzimidazole group is hydrogen-bonded in a
bifurcated manner to atoms O2 of thymines 7 and 19 (2.90 and
2.70 Å respectively, Figure 10B). Such an arrangement has been
observed in a number of benzimidazole-containing minor-groove
complexes, including those of Hoechst 33258 and analogues.
The amidinium group attached to the benzimidazole is hydrogen-
bonded to O2 of T 20 (2.90 Å); and it is also the starting-point
for an incompletely observed network of water-water contacts
that extend along the groove toward the wider G/C region. The
amidinium group at the other end of the ligand is also weakly
hydrogen-bonded to a base O2 atom (3.20 Å), of cytosine 9.
The ligand-DNA hydrogen bonding observed for DB818 is
in contrast to that predicted from computer modeling for the
furan analogue, DB293 (Figure 11 A, there is at present no
crystal structure for the DB293 complex). This ligand has a
significantly less concave inner radius than DB818 (Figure 11
B), due to the smaller van der Waals radius of oxygen compared
to sulfur. Thus DB293 cannot simultaneously make base-
benzimidazole hydrogen bonds and at the same time be in
position for base hydrogen bonding to both terminal amidinium
groups. DB818, by virtue of its larger concave radius, can do
so. This also enables DB818 to cover a five base-pair site
compared to the four of DB293 and to have stronger compound-
base interactions.
ITC experiments provide a direct method to obtain the binding
enthalpy and with SPR K values allow a full thermodynamic
characterization of DNA complexes. Such a full thermodynamic
picture is essential to complement the structural perspective from
crystallography presented here. ITC comparison of the interac-
tions of the four diamidines from Figure 1 with an AATT DNA
sequence indicates that the better binding of DB818 to the DNA
minor groove is enthalpic in origin (Table 3). The T‚∆S term
for DB818 binding to the minor groove of the AATT sequence
is actually less favorable than that for the other three diamidines
while the ∆H for binding is more favorable by 3-4 kcal/mol.
It should be noted that the net binding entropy of all of the
diamidines is favorable at 25 °C.33 Such entropy contributions
to binding are common for minor groove interactions and it
has been shown that the binding of the benzimidazole derivative,
Hoechst 33 258, is almost entirely driven by a favorable binding
∆S term.34 It is interesting that although Hoechst 33 258 and
DB818 have similar binding constants for AT sites in DNA,
the molecular basis for their large binding constants is quite
different, entropy dominated for Hoechst 33 25834 and enthalpy
dominated for DB818.
Discussion
Footprinting, Tm and spectroscopic results (Figures 3-6) show
that all four diamidines in Figure 1 are AT specific DNA minor
groove binding agents as predicted by their structure and from
studies on the furan derivatives, DB75 and DB293.1,16 These
results also qualitatively illustrate that DB818 binds to AT
sequences much more strongly than any of the other three
diamidines. Conversion of the furan of DB75 to a thiophene
does not enhance DNA interactions and conversion of one
phenyl in DB75 to a benzimidazole does not cause a major
change in the AT minor groove binding affinity. It was thus
initially surprising that conversion of the diphenyl-thiophene,
DB351 or the furan-benzimidazole, DB293, to DB818 (Figure
1) yields a compound that binds significantly better to AT
sequences in DNA than the other diamidines. Both DB818 and
DB293 form monomer complexes in -AATT- sequences and
since DB818 differs by only a single atom from DB293, the
significantly stronger binding was unexpected. The increased
DNA interaction of DB818 cannot be due, for example, to
simple differences in solution properties of S and O since DB351
actually binds slightly more weakly than DB75.
The results presented here suggest that strong, favorable
interactions of the molecular groups of DB818 with the DNA
minor groove provide a very favorable energetic component and
account for the larger binding constant of the compound. Such
strong interactions explain the larger negative enthalpy on
complex formation and they are a welcome example of a very
favorable design strategy to increase affinity and binding
specificity for monomer minor groove binding agents. The
strong interactions, however, also apparently make the DB818
complex less dynamic than with the other three compounds and
this more rigid structure could account for the lower entropy
of binding for DB818. Such enthalpy-entropy compensations
are well-known in biochemical interactions and have been
documented for several DNA-protein complexes.35
Biosensor-SPR and ITC experiments were used to quantita-
tively characterize the binding differences among these com-
pounds.31,32 The SPR method is particularly well-suited for
determination of the strong interactions of compounds, such as
the diamidines in Figure 1, with nucleic acids where very low
concentrations must be used in order to obtain reliable values
It is quite unlikely that any direct interactions with DNA of
the sulfur atom in the thiophene of DB818 could account for
(33) Bailly, C.; Chessari, G.; Carrasco, C.; Joubert, A.; Mann, J.; Wilson, W.
D.; Neidle, S. Nucleic Acids Res. 2003, 31, 1514.
(34) Haq, I.; Ladbury, J. E.; Chowdhry, B. Z.; Jenkins, T. C.; Chaires, J. B. J.
Mol. Biol. 1997, 271, 244.
(31) Wilson, W. D. Science 2002, 295, 2103-2105.
(32) Wilson, W. D.; Wang, L.; Tanious, F.; Kumar, A.; Boykin, D. W.; Carrascol,
C. Biacore, J. 2001, 1, 15-20.
(35) Dignam, J. D.; Nadda, S.; Chaires, J. B. Biochemistry 2003, 42, 5333-
5340.
9
13668 J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004

Heterocycle−DNA Interactions
A R T I C L E S
the increase in binding, for example, relative to the oxygen atom
in the furan of DB293.16b DB75 and DB351, for example, have
the same O to S change but DB351 binds slightly more weakly
than DB75. The explanation for the more favorable interactions
and ∆G and ∆H for DB818 binding thus seems likely to be
due to structural differences in the complex of this compound
with AT minor groove sequences relative to complexes of the
other three diamidines of Figure 1. Fortunately, crystallographic
results for DB818 bound to the self-complementary DNA
duplex, d(CGCGAATTCGCG)2, which has been used in
structural analysis of several minor groove-binding compounds
including DB75, provide a molecular rationale for the thermo-
dynamic observations. Both DB818 and DB75 are largely bound
in the AATT sequence of the duplex with both amidine groups
facing the floor of the groove for interactions with DNA base
pair edges (no satisfactory crystals of the DB351 and DB293
AATT complexes have yet been obtained). The curvature of
DB818, however, appears optimized for interactions of the
aromatic system with the walls and floor of the minor groove
(Figure 10A). These results suggests that the optimized van der
Waals interactions and H-bonding capability of the benzimi-
dazole and amidines of DB818 account for its stronger binding
to AATT relative to DB75 and DB351 (Figure 10B). The less
favorable binding entropy of DB818 is probably due to its very
tight interaction with the bases at the floor of the groove and to
the twist in the bound compound that effectively locks it in
place in the groove (Figure 10).
To better understand the binding differences between the
benzimidazole derivatives of similar size, DB818 and DB293,
a molecular model of DB293 was calculated for structural
comparison (Methods Section). There is a small bond angle
difference between the C-S-C angle of the thiophene of
DB818 and the C-O-C angle of the furan of DB293 that is
largely caused by the size difference between S and O. This
angle change results in a small difference in the positions of
the substituents on the thiophene and furan systems in these
two compounds that can best be seen in an overlay diagram
(Figure 11A). When amplified out to the terminal amidines of
the two compounds, the small angle difference at the five-
membered ring system yields very significant differences in the
positions of the amidines on the two compounds. The DNA
binding consequences of this molecular structural difference are
apparent in an overlay diagram of the DB818 and DB293
complexes with the d(CGCGAATTCGCG)2 sequence (Figure
11B). The DB818 structure is experimental while the DB293
structure is an optimized docked model. The smaller radius of
curvature of DB293 is not well suited to match the DNA minor
groove topology as can be seen in the overlay view. Although
both amidines of DB293 can form H-bonding interactions with
base edges at the floor of the minor groove, the curvature of
the molecular system pushes the central region of the compound
away from the floor of the groove and prevents the benzimi-
dazole from forming strong H-bonding interactions of the type
seen in the DB818 complex. The interactions of DB293 with
the AATT sequence are thus similar to the interactions for the
DB75 and DB351 complexes but are weaker than those for the
closely related benzimidazole derivative, DB818.
In summary, subtle compound structural differences provide
the molecular basis for the significant thermodynamic differ-
ences observed for formation of the AATT complexes by the
four diamidines in Figure 1. Even with their significantly
different substituents, the two diphenyl derivatives and DB293
have similar DNA interactions that yield similar ∆G, ∆H, and
∆S values for formation of their AATT complexes. Hydrophobic
effects provide a major driving force for formation of the DNA
complexes of these compounds, while interactions of the two
amidines with AT base pairs provide additional specificity and
affinity that probably arises primarily through an enthalpy
contribution. With the DB818 complex, however, the compound
and DNA minor groove topology are better matched and the
DB818 interactions with the AT base pairs and groove walls at
AATT are better optimized. Both the amidine and benzimidazole
groups can form H-bonds with the base pairs in the DB818
complex and more base pairs are contacted than with either of
the diphenyl compounds or DB293. As a result, the binding
enthalpy is more favorable for complex formation with DB818.
The optimized interactions result in a tighter and somewhat less
dynamic complex with enthalpy-entropy compensation and a
less favorable ∆S for complex formation with DB818. The net
result is that an improved ∆G for binding of DB818 is obtained
because the ∆H improvement is larger than the ∆S decrease.
Addition of other substituents, which could interact with DNA,
to DB818 could yield favorable contacts and molecular interac-
tions at a lower entropy cost, and could result in a more dramatic
improvement in ∆G for binding of derivatives of DB818. On a
larger scale, by choosing an appropriate central ring or an
adjacent benzo-ring (benzothiophene, benzimidazole, benzofu-
ran) it is possible to “tune” the shape of the molecule to match
DNA structurally and chemically for enhanced affinity and
sequence recognition.
Acknowledgment. Supported by NIH Grant No. GM61587,
by the Bill and Melinda Gates Foundation, and by equipment
funds from the Georgia Research Alliance (W.D.W. and
D.W.B.), the Ligue Nationale Contre le Cancer (C.B.), and the
Cancer Research UK (S.N.).
Note Added after ASAP Posting. After this paper was
posted ASAP on September 29, 2004, three authors’ affiliations
and typographical errors in some of the units in the “Crystal-
lization and Data Collection” section were corrected. The
corrected version was posted October 4, 2004.
Supporting Information Available: Supplementary Figures
S1-S4 and legends. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA048175M
9
J. AM. CHEM. SOC. VOL. 126, NO. 42, 2004 13669