Structure-dependent binding of arylimidamides to the DNA minor groove

Article abstract of DOI:10.1002/cbic.201300622

Heterocyclic diamidines are strong DNA minor-groove binders and have excellent antiparasitic activity. To extend the biological activity of these compounds, a series of arylimidamides (AIAs) analogues, which have better uptake properties in Leishmania and Trypanosoma cruizi than diamidines, was prepared. The binding of the AIAs to DNA was investigated by T_m, fluorescence displacement titration, circular dichroism, DNase I footprinting, biosensor surface plasmon resonance, X-ray crystallography and molecular modeling. These compounds form 1:1 complexes with AT sequences in the DNA minor groove, and the binding strength varies with substituent size, charge and polarity. These substituent-dependent structure and properties provide a SAR that can be used to estimate K values for binding to DNA in this series. The structural results and molecular modeling studies provide an explanation for the differences in binding affinities for AIAs. Copyright

Full text of DOI:10.1002/cbic.201300622

CHEMBIOCHEM
FULL PAPERS
DOI: 10.1002/cbic.201300622
Structure-Dependent Binding of Arylimidamides to the
DNA Minor Groove
Yun Chai,[a] Manoj Munde,[a] Arvind Kumar,[a] Leah Mickelson,[a] Sen Lin,[b]
Nancy H. Campbell,[b] Moloy Banerjee,[a] Senol Akay,[a] Zongying Liu,[a]
Abdelbasset A. Farahat,[a] Raja Nhili,[c] Sabine Depauw,[c] Marie-Hꢀlꢁne David-Cordonnier,[c]
Stephen Neidle,[b] W. David Wilson,*[a] and David W. Boykin*[a]
Heterocyclic diamidines are strong DNA minor-groove binders
and have excellent antiparasitic activity. To extend the biologi-
cal activity of these compounds, a series of arylimidamides
(AIAs) analogues, which have better uptake properties in Leish-
mania and Trypanosoma cruizi than diamidines, was prepared.
The binding of the AIAs to DNA was investigated by Tm, fluo-
rescence displacement titration, circular dichroism, DNase I
footprinting, biosensor surface plasmon resonance, X-ray crys-
tallography and molecular modeling. These compounds form
1:1 complexes with AT sequences in the DNA minor groove,
and the binding strength varies with substituent size, charge
and polarity. These substituent-dependent structure and prop-
erties provide a SAR that can be used to estimate K values for
binding to DNA in this series. The structural results and molec-
ular modeling studies provide an explanation for the differen-
ces in binding affinities for AIAs.
Introduction
Protozoal parasitic diseases have caused significant human
health problems for centuries and they continue to do so in
developing countries. Chagas’ disease (American trypanoso-
miasis) and Leishmaniasis are particularly widespread and do
not currently have satisfactory treatments for the affected pop-
ulations.[1,2] Chagas’ disease, caused by Trypanosoma cruzi, is
prevalent from the southern United States to Argentina, with
90 million people estimated to be at risk and 16–18 million
currently infected.[3] Over 300 million people around the world
are at risk of contracting the Leishmania parasite, and drugs
used to treat that disease are also unsatisfactory: most exhibit
considerable toxicity and resistance is developing.[4,5] Aromatic
diamidines, such as pentamidine (Scheme 1) are active against
the Trypanosoma brucei parasite, which causes the sleeping
sickness human African trypanosomiasis (HAT), as well as
against other parasites.[6] Highly fluorescent heterocyclic diami-
[a] Dr. Y. Chai, Dr. M. Munde, Dr. A. Kumar, L. Mickelson, Dr. M. Banerjee,
Dr. S. Akay, Dr. Z. Liu, Dr. A. A. Farahat, Prof. W. D. Wilson,
Prof. D. W. Boykin
Department of Chemistry, Georgia State University
50 Decatur St. SE., Atlanta, GA 30303 (USA)
E-mail: wdw@gsu.edu
dboykin@gsu.edu
[b] Dr. S. Lin, Dr. N. H. Campbell, Prof. S. Neidle
The School of Pharmacy, University College London
29–39 Brunswick Square, London WC1N 1 AX (UK)
[c] Dr. R. Nhili, S. Depauw, Prof. M.-H. David-Cordonnier
INSERM U837-JPARC (Jean-Pierre Aubert Research Center)
Team Molecular and Cellular Targeting for Cancer Treatment
Universitꢀ Lille Nord de France, IMPRT-IFR-114
Scheme 1. Structures of the compounds and DNA sequences used in this
study. For SPR experiments, 5’-biotin labeled DNA sequences were used.
Institut pour la Recherche sur le Cancer de Lille
Place de Verdun, 59045 Lille Cedex (France)
dine analogues of pentamidine have been developed, and
fluorescence microscopy, as well as detailed molecular biology
analysis, has demonstrated that the mitochondrial kinetoplast
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201300622.
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 68

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
is the primary cellular target of the compounds.[7–14] A prodrug
of furamidine (DB75) has been used in clinical trials against
first-stage HAT, and a related compound is active against the
second stage of the disease.[6,15–18] Clearly compounds of this
type are promising for the treatment of HAT and related dis-
eases.[19–21]
Table 1. Comparison of Tm and fluorescence for AIAs.
[a]
[b]
Compound
R
DTm [8C]
F70 [mm]
A
As the heterocyclic diamidines bind to the minor groove of
DNA at AT sites of four or more base pairs, and the DNA of the
kinetoplast has a very large number of appropriate binding
sites, targeting this unique DNA structure of the parasite offers
a number of therapeutic advantages. Replication of the circu-
lar, interlocked kinetoplast DNA appears to be particularly sen-
sitive to the diamidines, perhaps due to compound-induced
conformational changes that disrupt the kinetoplast struc-
ture.[22–24] There are no known DNA structures in humans that
are equivalent to the kinetoplast, so targeting the kinetoplast
offers an excellent route to selective drug action.[11] The diami-
dines are taken up into T. brucei cells by the P2 membrane
transporter, and, because the parasite lives in blood in all initial
stages of infection, uptake is quite effective.[25] These com-
pounds are, however, taken up much less effectively by both
the T. cruzi and Leishmania parasites that live in human macro-
phage cells. They still appear to target the kinetoplast of these
organisms but they are much less active against them, presum-
ably due to poor uptake.[26]
DB667
DB709
DB745
DB766
DB1890
DB1950
H
OCH3
OCH2CH3
OCH(CH3)2
OCH2CH(CH3)2
OCH2CH2CH(CH3)2
19.6
13.1
8.1
6.0
2.0
1.7
1.2
2.0
5.5
>10
>10
1.0
DB1852
DB1880
DB1876
0.2
27.1
25.0
>10
0.55
0.6
B
DB766
6.0
7.0
1.1
0.5
5.5
1.2
A potentially promising route to developing new heterocy-
clic cations that could selectively target the kinetoplast of
T. cruzi and Leishmania but which also have the potential for
better cell uptake is to modify the amidine group while main-
taining the basic structure that is the key component of the
DNA interactions. A promising modification involves conver-
sion of the amidines to arylimidamides (AIAs, previously re-
ferred to as “reverse” amidines) in which a nitrogen of the ami-
dine is linked to the heteroaromatic core (Table 1) rather than
the carbon in classical amidines (Scheme 1). The AIAs have
lower pKa values and typically have much better biological
activity against both T. cruzi and Leishmania than the ami-
dines.[27–31] The AIAs offer a new approach for the development
of drugs against these diseases, but we understand much less
about their DNA interactions than those of amidine derivatives.
As a key step in understanding their biological targets as well
as developing these compounds to effectively treat parasites,
we have evaluated the interactions of an array of modified
AIAs with AT DNA binding sites that are typical of those found
in kinetoplast DNA. This is the first detailed study of DNA com-
plexes of AIAs, and the results show a surprisingly large varia-
tion in DNA interactions with relatively small changes in the
structure of the compounds.
DB1831
DB1855
DB1937
9.0
>10
[a] Poly(dA)·poly(dT), values are an average of three independent trials
with a reproducibility of ꢀ0.58C. [b] The compound concentration re-
quired to reduce the fluorescence of DAPI in complex with A5 to 70% of
the initial value. The values are reproducible within 10%.
determined (Table 1). The DTm values of the DNA complexes
show quite large, structure-dependent variations. Triple helix
formation (TnAnTn) was excluded because it leads to biphasic
melting curves for the duplex and triplex species. As biphasic
curves were not seen with the AIAs (Figure S1 in the Support-
ing Information), and as duplex-specific minor groove binders
are usually very poor triplex inducers, we are able to rule out
any triplex formation.[33]
O-Alkyl substitution: Scaffold A: The R group in scaffold A of
Table 1 has O-alkyl substituents of increasing size or piperidine
groups with two extra charges. The DTm values are quite sensi-
tive to compound structure. With the unsubstituted DB667
(R=H), for example, DTm =19.68C, but as the alkyl group in-
creases in size, there is a large decrease in the DTm. With an
OMe R group, DB709, DTm is down to 13.18C, and the com-
pound with an OCH2CH(CH3)2 group, DB1890, has a DTm of
only 2.08C. DB1880 with a charged O-piperidine group and
DB1876 with a charged O-isopropylpiperidine group, on the
Results and Discussion
Thermal melting: Ranking the compounds
Thermal melting enables rapid qualitative evaluation of the rel-
ative binding affinities of compounds for DNA.[32] As part of
a screen to find new compounds that target kinetoplast AT
sequences, DTm values for AIAs with poly(dA)·poly(dT) were
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 69

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
other hand, have very high DTm
values despite the substantial
size of the substituent.
Variations in the terminal six-
membered rings: Scaffold B:
Compounds in scaffold B have
extra nitrogen atoms added to
the terminal pyridine of DB766,
and their DTm values are sensi-
tive to the N position (Table 1).
The DTm values are relatively
low for all compounds in the
scaffold B group.
Fluorescence displacement
titrations: Comparing the
relative affinities
Unlike classical amidines, AIAs
do not fluoresce by themselves,
therefore, fluorescence-displace-
ment titrations were used as an
additional screen to rank them
according to their binding affini-
ty at 258C rather than the high
temperatures in Tm experiments.
The compounds were tested
with hairpin DNAs containing an
A5 binding sequence, which is
an analogue of the poly(dA)·po-
ly(dT) used in Tm experiments,
and an ATATA sequence to eval-
uate the AT-sequence-depen-
dent interactions with DNA.
DNAs with two fluorophores,
DAPI and DB829, (Scheme 1),
which have very different prop-
Figure 1. Fluorescence displacement titrations. Titration of DB1876 into A) DAPI–A5 and B) DB829–A5 hairpin DNA
complexes. DAPI fluoresces when bound, so as it is displaced, fluorescence intensity decreases. DB829 fluoresces
more strongly in solution, so as it is displaced, the fluorescence intensity increases. Normalized fluorescence dis-
placement titration. Representative normalized plots of intensity vs. compound concentration for fluorescence dis-
!
&
^
*
placement titration of DB1876 ( ), DB1852 ( ), DB766 ( ) and DB709 ( ) titrated into C) DAPI–A5 and D) DB829–
A5 complexes. DB1876 has the highest binding affinity; DB709 has a moderately high binding affinity; DB766 and
DB1852 have low binding affinity.
erties, were used to test the binding affinities. DAPI alone has
low fluorescence, but when bound to DNA it fluoresces strong-
ly, whereas DB829 has the reverse fluorescence behavior. Fig-
ure 1A, B shows typical fluorescence-displacement titrations
with DB1876 for DAPI–A5 and DB829–A5 complexes. Titration
of the DAPI–A5 complex with DB1876 displaces DAPI and re-
sults in a decrease in the intensity, as expected. The opposite
change in fluorescence is observed upon titration of the
DB829 complex. Figure 1C, D compares ratio plots for titration
of A5–DAPI and A5–DB829 with several compounds from
Table 1. The titration displacement curve for DB1876 is sharp
and the compound displaces DAPI and DB829 at low com-
pound concentration. F70, the compound concentration re-
quired to reduce the fluorescence to 70% of the initial value,
for DB1876 is 0.6 (all concentration values in mm) with DAPI–
A5, whereas the values for the other compounds in the plot
are higher: F70 =1.2 (DB709), F70 =5.5 (DB766), and F70 >10
(DB1852). The displacement results with DB829 are in general
agreement with the DAPI and Tm results.
The agreement between the DTm and F70 results is illustrated
with the plots in Figure 2 for the compounds in Table 1. The
fluorescence displacement titration results are in excellent
agreement with the inverse of the DTm values. Both sets of re-
&
&
Figure 2. Comparative histogram plots of DTm ( ) and fluorescence ( ) for
some compounds that vary significantly in DNA affinity.
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 70

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Figure 3. Circular dichroism spectra of A) DB1876–A5, B) DB1876–ATATA, and C) DB1852–A5 titrations for added ratios (compound to DNA hairpin) ranging
&
*
from bottom to top from 0.32 to 3.0. D) A plot of induced CD (ICD) versus DB1876/DNA ratio; : A5, : ATATA.
sults show that the AIAs bind to AT DNA sequences in a struc-
ture-dependent manner, and some compounds bind quite
strongly.
pared directly for stoichiometry differences. The sensorgrams
of DB667 (Figure 4A) show fast kinetics of association and dis-
sociation and they can be fitted to a steady-state analysis.[34–36]
The steady-state values (in response units, RU) were plotted
against Cfree (free compound concentration) and fitted to
a single site model (Figure 4D) that is predicted by the RU
value at saturation of the DNA binding site.[37] The values of
the equilibrium binding constant, K, are collected in Table 2.
DB667 binds strongly to A5 (9.0ꢃ106 mꢁ1) and somewhat more
weakly to ATATA (2.8ꢃ106 mꢁ1). Interestingly, DB709, with
ꢁOMe groups in place of the ꢁH in DB667, has significantly re-
duced affinity (Figure 4C, F). DB766, which was difficult to
study due to poor binding and water solubility, shows very
low SPR signals (data not shown) that confirm its weaker bind-
ing affinity.
CD: Determining the binding mode and stoichiometry
CD spectroscopy with A5 and ATATA hairpin DNA sequences
was used to obtain information on the binding mode of AIA
compounds that were selected based on Tm and F70 results
(Figure 3). Positive induced signals in CD spectroscopy are gen-
erally obtained for compounds that bind in the DNA minor
groove, and this pattern provides a method for evaluating
binding modes. DB1876 gives strong induced CD with satura-
tion at a compound to DNA ratio of about 1:1. A plot (Fig-
ure 3D) of induced CD (ICD) versus compound/DNA ratio indi-
cates a break at approximately 1:1 for A5 and ATATA
and suggests that DB1876 binds as a monomer in
the minor groove of these AT sequences. With A5 it is
also possible to fit the data differently, with a break
at a 1.4:1 ratio. Because these simple oligomers
should bind the compound at either a 1:1 or 1:2
ratio, we think the fit at near the 1:1 ratio is more
reasonable. DB1852, a close analogue of DB1876
with a neutral O-pentacyclic ring, shows poor CD,
thus indicating that it binds weakly, in agreement
with Tm and F70 results. The CD results thus confirm
a minor groove binding mode for the compounds of
Figure 1A, B and generally support the Tm and F70
conclusions.
SPR biosensor: Quantitative and stoichiometric
analysis of complexes
In order to evaluate the interactions of representa-
tive AIA analogues with DNA more quantitatively,
biosensor-SPR experiments were conducted with
DNA hairpin duplexes containing A5 and ATATA se-
quences. The flow cells for A5 and ATATA had essen-
tially the same amount of DNA immobilized so that
their sensorgram saturation levels could be com-
Figure 4. Surface plasmon resonance. Representative SPR sensorgrams for: A) DB667,
B) DB1876, and C) DB709 binding to immobilized A5 and ATATA hairpin DNAs. The com-
pound concentrations were 0.01 to 1.0 mm from bottom to top. SPR binding affinity
plots. RU values from the steady-state region of SPR sensorgrams are plotted against the
unbound compound concentration, Cfree (flow solution) for D) DB667, E) DB1876, and
&
*
F) DB709 with A5 ( ) and ATATA ( ) DNA.
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 71

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
B-form DNA duplex helix, as observed in a large number of
previous studies.[40–43] The helically twisted DB1880 molecule is
bound as a monomer in the minor groove and covers almost
all of the 6-bp sequence (A5/T20) (A6/T19) (T7/A18) (T8/A17)
(C9/G16) (G10/C15) in this narrow central region of the helix
(Figure 6A, B). The two phenyl rings and the furan ring are ori-
ented parallel to the walls of the minor groove and are deeply
embedded, whereas the two terminal pyridine rings are twist-
ed to the mean plane of these three central rings and are
close to the mouth of the minor groove.
Table 2. Selected SPR binding affinities.[a]
K [106 mꢁ1
]
K [106 mꢁ1
]
Compound
A5
ATATA
Compound
A5
ATATA
DB75[23]
DB613
DB667
18
15
9.0
24
3.6
2.8
DB709
DB1876
0.5
0.4
57
85
[a] The listed binding affinities are an average of two independent experi-
ments carried out with two different sensor chips, and the values are re-
producible within 10%.
There are a number of direct hydrogen-bond and water
bridging contacts between the ligand and the DNA, as shown
in Figure 6B. One inward-facing amidine nitrogen atom is ori-
ented out of the plane of the central diphenylfuran moiety by
438 and hydrogen bonds to the O2 atoms of T20 and T19 (2.7
and 3.3 ꢄ, respectively). This nitrogen atom is also involved in
hydrogen bonding to the O4’ atom T20 (3.0 ꢄ) (Figure 6C). At
the other end of the DB1880 molecule (Figure 6D), the amidi-
nium group is twisted by 728 relative to this plane, and one
amidinium nitrogen atom hydrogen bonds to the O2 (2.7 ꢄ)
and O4’ atoms (3.0 ꢄ) of C9. The other amidinium nitrogen
atom is involved in hydrogen bonding to the O4’ atom of A18
(3.2 ꢄ) on the complementary strand.
Sensorgrams for DB1876, with the O-isopropylpiperidine
ring, show that the compound has much slower binding kinet-
ics with DNA than the other AIAs (Figure 4B). At low com-
pound concentrations, the slow kinetics prevent a steady state
from being reached within the experimental time range, which
is limited by injection volume. Even for these low-concentra-
tion samples, it is possible to fit the curves to a 1:1 kinetic
binding model[33–36] to determine predicted steady-state RU
values for DB1876. At higher concentrations of DB1876, a
steady state was reached, and these values can be used direct-
ly in RU versus Cfree plots along with predicted values (Fig-
ure 4E). DB1876 has very strong binding affinity for A5 and
somewhat weaker binding to ATATA (Table 2). These results are
in excellent agreement with the more qualitative Tm and fluo-
rescence results.
Neither piperidinyl group directly hydrogen bonds to the
DNA. One interacts through its nitrogen atom, via a bridging
water molecule, to a phosphate group, as does the adjacent
pyridine ring, to the next phosphate group on the same
strand. The latter water molecule also hydrogen bonds to the
adjacent amidinium nitrogen atom as well as to a short chain
of water molecules that links to a third phosphate group (Fig-
ure 6E). Both piperidine groups extend outwards from the
groove so that their cationic ring nitrogen atoms are relatively
close to the anionic phosphate groups. Although few water
molecules have been located in the vicinity of these nitrogen
atoms, we cannot discount the possibility that the water bridg-
ing referred to above occurs more extensively; indeed, there is
a strong likelihood that further water molecules are present in
these regions.
DNase I footprinting: Sequence specificity
To compare the binding of compounds with high-molecular-
weight DNA to the results obtained with the hairpin DNAs,
DNase I footprinting studies were conducted. DNase I cuts
DNA at all sites but with variations in rates that depend on the
local minor-groove geometry. Small molecules that bind in the
minor groove block access of the enzyme and result in re-
duced cleavage.[38,39]
Footprinting results for selected AIAs from Table 1 are
shown in Figure 5 with an experimental gel. DB75, a well-char-
acterized diamidine, was used as a control in these studies and
displays good footprints at both the alternating and nonalter-
nating AT sequences. Gel results for DB766 show no detectable
footprint up to >1 mm. The more highly charged DB1876 has
strong footprints at 0.5 mm and above, whereas DB1880 (O-pi-
peridine) has strong footprints at 0.25 mm and above. These re-
sults are in agreement with the Tm and SPR results and suggest
that the hairpin DNAs’ affinities reflect those of the higher-mo-
lecular-weight DNAs.
A well-pronounced network of water molecules is apparent,
surrounding the outer edge of the bound ligand. This horse-
shoe-like arrangement extends along the phosphate groups of
both strands and terminates at one end of the ligand in a net-
work of two connected rings of water molecules, closely simi-
lar to previous observations of water networks in DNA minor-
groove–ligand complexes.[44]
Free compound modeling and docking with DNA
In order to better understand the effects of substituents on
AIA–DNA interactions molecular docking studies were per-
formed.
X-ray crystallography: The structure of the DB1880–DNA
complex
O-Alkyl substitution: Scaffold A: With the unsubstituted
DB667 (R=H, Table 1), DFT calculations for optimizing geome-
try at the 631G* approximation level shows that the phenyl–
furan–phenyl system adopts a coplanar conformation (Fig-
ure 7A). As the alkyl group increases in size with the addition
of O-alkyl substituents, steric hindrance overcomes the long-
In order to evaluate the structural basis of AIA–DNA minor-
groove complexes as well as strong binding of the tetracat-
ionic derivatives, crystals of the DB1880–DNA complex were
grown, and its structure was solved by X-ray crystallography.
The d(CGCGAATTCGCG)2 sequence forms a self-complementary
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 72

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Figure 5. DNase I footprinting titration experiments. The 81 bp DNA fragment containing A5 and ATATA sites was incubated with increasing concentrations
[mm] of DB75, DB766, DB1876, and DB1880 prior to being subjected to mild digestion with DNase I. A) The digested products were separated on a 8% poly-
acrylamide gel containing 8m urea. Sequences at the footprinting sites are indicated to the right of the gel. B) Corresponding densitometric analysis reveals
~
&
&
*
^
*
the localization of the footprints (black boxes) along the DNA sequence. Top: 0.1 ( ), 0.25 ( ), 0.5 ( ), 0.75 ( ), 1 mm (+) DB1876; bottom: 0.1 ( ), 0.25 ( ),
^
0.5 mm ( ) DB1880.
range conjugation and causes some twist between the phenyls
and furan. DB1890, which has R=OCH2CH(CH3)2, for example,
has dihedral angles between the phenyl and furan ring of ap-
proximately 148.
DB613 (Scheme 1) has the same central phenyl–furan–
phenyl rings but with terminal substituted phenyl groups in-
stead of the pyridine in DB667. The dihedral angles between
the AIA and the terminal phenyl or pyridinyl planes are 378
and 48 in DB613 and DB667, respectively. The modeling results
suggest that the close planarity of the terminal pyridinyl
groups and amidine group in DB667 could be due to their
ability to form amidine group···pyridine hydrogen bonds.
Meanwhile in DB613, rotation of terminal phenyl rings into the
AIA plane is more hindered by van der Waals repulsion be-
tween the hydrogen atoms, and a H-bond is not possible.
Molecular electrostatic potential (MEP) maps illustrate the
charge distributions of molecules, help to understand the rela-
tive polarity of a molecule, and are useful for evaluating struc-
ture–activity relationships. The maps clearly show a significantly
different distribution of MEPs for the compounds (Figure 7A).
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 73

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Figure 6. A) Crystallographic structure of the DB1880–DNA complex. The DNA sequence is shown in gray, DB1880 molecule in magenta, nitrogen atoms in
blue, and oxygen atoms in red. B) The full sequence with bases numbered and binding schematic. Detailed views of the hydrogen-bonding interactions be-
tween DB1880 and DNA. The distances [ꢄ] were measured between heavy atoms. C) The three hydrogen bonds between one amidinium nitrogen atom and
the oxygen atoms of T19 and T20. D) The hydrogen bonds between two amidinium nitrogen atoms and the oxygen atoms of C9 and A18. E) The network of
hydrogen bonds bridging water molecules between DB1880 and DNA phosphate oxygen atoms.
Figure 7. A) Electrostatic potential maps. Equilibrium geometry of the AIAs calculated by the DFT B3LYP approximation at the 631G* level. Space-filling
models are shown at the top, with electrostatic potential molecular surfaces at the bottom; blue=positive and red=negative potential, the color was set
at the same scale. B) Comparison of dipole moments. Ab initio-calculated electrostatic potential maps for the pyridine, pyrimidine and pyridazine units of
DB766, DB1831 and DB1937, respectively. The dipole moments of these units are shown on the left; the magnitudes of the dipoles are on the right.
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 74

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
As expected, the negative potential is stronger on oxygens
atoms, whereas the positive potential is stronger on the ami-
dine ꢁNH groups. The inner faces of the molecules, towards
the floor of the minor groove, have a positive character. The
outer face, towards the solvent, is relatively more negatively
charged. This favors the formation of hydrogen bonds and
electrostatic interactions between bases at the floor of the
DNA minor groove and AT sequences. With large O-alkyl
groups on the inner phenyl rings, such as with DB1890, the
negative potential is predicted to be partially shifted from the
furan ring to the alkyl substituents.
ence is perhaps due to the ability to form two amidine···pyrimi-
dinyl hydrogen bonds and to the elimination of repulsive CH
van der Waals interactions with the amidine group in DB1831.
The electrostatic potential molecular surfaces of the central ar-
omatic system on DB766, DB1831, DB1855 (R=pyrazine), and
DB1937 (R=pyridazine) are similar; however, the MEP map
clearly shows a significantly different distribution on the termi-
nal rings. In DB1855 and DB1937, the significantly negative po-
tential centralizes on the terminal pyrazinyl or pyridazinyl nitro-
gen atoms, whereas the electrostatic potential is more evenly
spread across the pyridinyl and pyrimidinyl systems in DB766
and DB1831. This quite polar region in DB1855 and DB1937 at-
tracts the bonding electrons more strongly and provides a
greater attractive force for the aqueous proton. In an aqueous
environment, all four compounds have bound water molecules
around them, and some of these water molecules need to be
released in order to bind to DNA. As the pyridazinyl group in
DB1937 would attract bound water molecules more tightly
than DB766 and DB1831, more energy is required to release
the same number of water molecules and this could be
a factor in explaining the different binding affinities. A contri-
bution from a pK effect is also possible and could reflect a par-
tial lack of protonation of the amidine groups under the test
conditions; this would greatly affect DNA binding affinity.
In order to better understand the energy contributions, it is
important to compare the ab initio calculated MEP for the ter-
minal heterocyclic units (Figure 7B). The pyridazine ring in
DB1937 possesses a high dipole moment (4.47 D), which is at-
tributed to the fact that the two nitrogen atoms are located
on the same side of the ring. Therefore, there is a greater pull
of electrons to that side that results in a high dipole moment.
The dipole moment of pyrazine is 0 since it is symmetrical
about the line passing two nitrogen atoms. The higher magni-
tude of dipole moment for pyridazine, compared to pyridine
(2.36 D) and pyrimidine (2.46 D), suggests a possible contribu-
tion toward the lower observed binding affinity of DB1937.
Intermolecular attractions in pyridazine are stronger than in
pyridine and pyrimidine; this is attributed to electrostatic
forces arising from the high permanent dipole.
Molecular-docking studies can provide ideas for variations in
binding affinity across a set of derivatives such as the AIAs and
are particularly powerful if guided by X-ray structural results,
such as those with DB1880. The results obtained from molecu-
lar docking indicate that DB667 binds as a monomer in the
center of the minor groove of the d(CGCGAATTCGCG)2 duplex
(Figure 8A) and covers almost six base pairs. This orientation
Figure 8. Molecular docking. Models of complexes of A) DB667 and
B) DB1890 with the DNA duplex dodecamer sequence d(CGCGAA TTCGCG)2
given the X-ray crystal structure 3OIE as a guide. C) Overlay of the structures
of DB667, DB709 and DB1890 docked into the same DNA sequence. DB667
is displayed by atom type color, DB1890 is green, and DB709 is brown.
fits snugly in the DNA minor groove, and the inward-facing
amidinium nitrogen atoms are involved in hydrogen bonding
to the cytosine and thymine O2 groups. Docking of DB709
into the same DNA sequence gave similar results. As the alkyl
substituent increases in size, the substituted compound no
longer fits properly into the minor groove for optimum bind-
ing. Steric hindrance and unfavorable electrostatic contacts of
the substituents with the minor groove limit binding. An ex-
ample docking result for DB1890 is shown in Figure 8B. The
phenylfuran core is pushed away from the floor of the groove,
and all interactions with the DNA are weakened. An overlay of
DB667 and DB1890, which were extracted from an overlay of
their DNA complexes after deletion of DNA, is shown in Fig-
ure 8C. The suboptimal binding position of DB1890—pushed
out and down the groove—can be seen in this comparison.
Variations in the terminal six-membered rings: Scaffold B: Ad-
dition of a nitrogen atom to the terminal pyridine of DB766
gives scaffold B compounds, which show electrostatic potential
maps that are sensitive to the N position (Figure 7A). For
DB1831 (R=pyrimidine, Table 1), the dihedral angle between
the amidine groups and the terminal pyrimidinyl plane is 08,
compared to 4–68 in the other three compounds. This differ-
Conclusions
It is useful to put the results on DNA binding of the AIA com-
pounds reported here in context relative to other similar
minor-groove-binding dications. The binding of the diphenyl-
furan diamidine, DB75, which has the same central aromatic
system as the AIAs in Table 1, to both the A5 and ATATA se-
quences has been extensively characterized.[23] This compound
has an equilibrium constant of approximately 2ꢃ107 mꢁ1 with
both DNAs. DB613, which has the same central diphenylfuran
but with a phenyl group instead of the pyridine in DB667, has
been evaluated with the same DNA sequences and has a similar
K to DB75 for A5 but one that is nearly seven times lower with
ATATA (Table 2). DB667 has slightly lower K values than DB613
with both DNAs. Interestingly, most amidine and AIA deriva-
tives bind more poorly to the ATATA sequence than to A5, and
it is actually DB75 that is unusual for its similar binding to the
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 75

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
two DNAs. The weaker binding to ATATA in general can be ex-
plained by the wider minor groove with that sequence relative
to A5.[7,23] Molecular modeling with an AATT sequence that has
been crystalized with both amidines and AIAs shows that
DB613 and DB667 can be inserted into the minor groove in
much the same manner as DB75 in the AATT site (Figure 8).
Replacement of the central phenyl hydrogen atom in DB667
by larger groups decreases the DTm values and increases the
amount of compound required to displace DAPI from the
minor groove (Table 1). Adding ꢁOCH3, in an otherwise equiva-
lent compound (DB709), gives a reduction in DTm of 6–78C,
and substitution with O-isopropyl (DB766) causes another 6–
78C decrease. Larger OR substituents have DTm values that
approach 08C (Table 1). Molecular modeling and docking of
these compounds in the minor groove provides a clear ration-
ale for the decrease in Tm with larger substituents (Figure 8).
The large substituents are significantly out of the conjugated
phenyl–furan ring plane and, as shown in Figure 8, they pre-
vent the attached compounds from penetrating as deeply into
the groove as compounds that have smaller substituents. This
weakens H-bond interactions with the base edges at the floor
of the groove and van der Waals contacts to the walls of the
groove in the complex.
In summary, the AIAs form 1:1 complexes in AT sequences of
four to six base pairs and bind with affinities that strongly
depend on substituent size, charge and polarity.
Experimental Section
Compounds, DNAs, and buffers
Syntheses of the compounds of Table 1: Syntheses of DB667, DB709,
DB745 and DB766 have been published;[29,45] syntheses of com-
pounds DB1831, DB1855 and DB1937 will be published elsewhere;
and all other AIAs are described below. All synthetic compounds
1
were characterized by H and 13C NMR and elemental analysis (C,
H, N within ꢀ0.4%). Poly(dA)·poly(dT) obtained from Pharmacia
Co. was used for Tm experiments. In circular dichroism (CD) and
fluorescence experiments, the hairpin DNA oligomers A5 (5’-GCCAA
AAAGC TCTCG CTTTT TGGC-3’) and ATATA (5’-GCCAT ATAGC TCTCG
CTATA TGGC-3’) with the hairpin loop sequences underlined were
used (DNA sequences shown in Scheme 1). In SPR experiments,
the same hairpin DNA oligomers 5’-labeled with biotin were used.
The cacodylic acid buffer (CAC) used in Tm, CD, and fluorescence
experiments contained cacodylic acid (0.01m), NaCl (0.1m), and
EDTA (0.001m) and was adjusted to pH 6.25. The SPR experiments
were conducted in filtered, degassed CAC buffer with 0.005% P20
surfactant. All DNA oligomers were purchased from Integrated
DNA Technologies, Inc. (Coraville, IA), purified by reversed-phased
HPLC, and characterized by mass spectrometry.
Several compounds were prepared with an additional N in
different positions of the terminal pyridyl of DB766. The pyrimi-
dine, DB1831, has essentially the same DTm as DB766 but the
other two compounds with two N atoms have substantially re-
duced DTm values. Modeling of the compounds
again provides suggestions as to why the selective
decrease in Tm occurs. As can be seen in Figure 7B,
the electron density of the pyrimidine in DB1831 is
more evenly spread than in pyridazine, and pyrimi-
dine has a lower dipole moment. If the nitrogens of
the pyridazine in DB1937 face into the AT minor
groove, this quite polar region will be dehydrated
and will have no possible H-bond donors. If the
group faces outwards, the regions of negative poten-
tial will be near the anionic backbone of the DNA, an-
other unfavorable binding orientation.
Preparation of AIAs: The synthetic route to the new compounds
in this paper is shown in Scheme 2. Experimental details and char-
acterization data for all the compounds and intermediates can be
DB1876 and DB1880 are unique in the AIA com-
pound set. The piperidine substituent is relatively
large, at least comparable to the cyclopentyl on
DB1852, but DB1880 and DB1876 have the highest
binding constants of the compounds in Table 1. The
two additional charges on DB1880 and DB1876 cer-
tainly help binding, but their effect is mitigated to
a certain extent by the 0.1m added NaCl in the ex-
periments. In the crystal structure of the DB1880–
DNA complex, both piperidine groups extend out-
wards from the groove, so that the charged piperi-
dine ꢁNH atoms are relatively close to DNA phos-
phate groups, and this interaction is certainly favora-
ble for binding. However, neither piperidyl group di-
rectly hydrogen bonds to the DNA. The very hydro-
Scheme 2. Synthesis of the AIAs. a) RI, tBuOK, THF, DMF; b) i: 4-hydroxypiperidine,
(Boc)2O, Et3N, CH2Cl2; ii: DEAD/PPh3, THF; c) i: 20 mol% CF3COOH, CH2Cl2; ii: 2-iodopro-
pane, K2CO3, DMF; d) 2,5-bis(trimethylstannyl)furan, Pd(PPh3)4, dioxane; e) 10% Pd/C, H2,
345 kPa, EtOAc/EtOH (9:1); f) i: MeCN, EtOH, RT, 24 h; ii: ethanolic HCl.
phobic O-cyclopentyl in DB1852 clearly does not
favorably interact with the minor groove in a similar
way.
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 76

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
found in the Supporting Information. Starting from 2-bromo-5-ni-
trophenol, 1-bromo-2-alkoxy-4-nitrobenzene (1a–d) was prepared
by treatment with either alkyl iodide or tert-butyl 4-hydroxypiperi-
dine-1-carboxylate. In the case of compound 1e, the bis-Boc-pro-
tecting groups of 1d were first removed by using trifluoroacetic
acid in CH2Cl2 and then treated with 2-iodopropane. Stille coupling
of 1a–e and 2,5-bis(trimethylstannyl)furan in the presence of
Pd(PPh3)4 in dioxane gave the corresponding 2,5-bis(2-alkoxy-4-ni-
trophenyl)furans (2a–e). Furans 2a–e were then reduced by cata-
lytic hydrogenation to give the desired diamino compounds 3a–e.
The target AIA salts DB1890, DB1950, DB1852, DB1880 and DB1876
(4a–e) were prepared in a two-step process. First, the free base
was obtained by treating 3a–e with a hydrobromide salt of naph-
thalen-2-ylmethyl pyridine-2-carbimidothioate in ethanol/acetoni-
trile. The free bases were subsequently treated with anhydrous
ethanolic HCl to give the AIA salts in good overall yield. In the case
of 4d (DB1880), Boc-deprotection was accomplished in the process
of the AIA hydrochloride salt formation.
ported spectra are an average of at least five scans. To obtain the
stoichiometry of each complex, a DNA solution was titrated with
a compound solution, and the induced CD (ICD) of the bound
compound was followed at the maximum wavelength.
Biosensor surface plasmon resonance (SPR): SPR measurements
were performed in a four-channel Biacore 2000 optical biosensor
system (GE Healthcare). The 5’-biotin-labeled DNA sequences (A5
and ATATA hairpins, in Scheme 1) were immobilized onto streptavi-
din-coated sensor chips (Biacore), as previously described.[34,35]
Three flow cells were used to immobilize the DNA oligomer sam-
ples, while a fourth cell was left blank as a control. The SPR experi-
ments were performed at 258C in filtered, degassed CAC buffer.
Steady-state binding analysis was performed with multiple injec-
tions of different compound concentrations over the immobilized
DNA surface at a flow rate of 25 mLminꢁ1 and 258C. Solutions of
known AIA concentration were injected through the flow cells
until a constant steady-state response was obtained. Solution flow
was then replaced by buffer flow resulting in dissociation of the
complex. The reference response from the blank cell was subtract-
ed from the response in each cell containing DNA to give a signal
(RU) that is directly proportional to the amount of bound com-
pound. The predicted maximum response per bound compound in
the steady-state region (RUmax) was determined from the DNA mo-
lecular weight, the amount of DNA in the flow cell, the compound
molecular weight, and the refractive index gradient ratio of the
compound and DNA, as previously described.[36,37] The number of
binding sites and the equilibrium constant were obtained from fit-
ting plots of RU versus Cfree. Binding results from the SPR experi-
ments were fit with either a single-site (K2 =0) or with a two-site
model [Eq. (2)]:
Thermal melting (Tm): Tm experiments were conducted on a Cary
300 Bio UV/Vis spectrophotometer (Varian) with the software sup-
plied with the instrument. A thermistor fixed into a reference cuv-
ette was used to monitor the temperature with a computer-con-
trolled heating rate of 0.58Cminꢁ1. The oligomers were added to
CAC buffer (1 mL) in 1 cm path-length, reduced volume quartz
cells; DNA without compound was used as a control. The concen-
trations of the DNA solutions were determined by measuring the
absorbance at 260 nm. Experiments were generally conducted at
a concentration of 2ꢃ10ꢁ5 m base pair for poly(dA)·poly(dT). For
experiments with complexes, a ratio of 0.3 compounds per base
pair was generally used.
Fluorescence: All experiments were conducted on a Cary Eclipse
Fluorimeter (Varian). Before conducting the fluorescence displace-
ment titration for DNA–AIA complexes, it was important to find
the concentration of the fluorophore (DAPI/DB829, in Scheme 1) in
each experiment so that addition of the test compound would dis-
place it. DNA (50 mm stock solution) was titrated into the fluoro-
phore-containing cell at 0.05 mm increments, and scans were re-
corded. A steady change in fluorescent intensity was observed
until saturation at 0.8 mm A5 with 0.5 mm DAPI, or 2.2 mm A5 with
0.5 mm DB829. These fluorophore concentrations were then used
in each fluorescence displacement experiment.
2
K1Cfree þ 2 K1K2Cfree
ð2Þ
r ¼
2
1 þ K1Cfree þ K1K2Cfree
here r represents the moles of bound compound per mole of DNA
hairpin duplex, K1 and K2 are macroscopic binding constants, and
Cfree is the concentration of free compound in equilibrium with the
complex.
Purification and radiolabeling of DNA restriction fragments and
DNase I footprinting: DNase I footprinting experiments were per-
formed essentially as described previously.[34,35] Complementary 5’-
end-phosphorylated oligonucleotides containing A5 and ATATA
sites (underlined) 5’-CGGTAC CAGATC TTCTAG GAAAAA CGGCTC
GATATA GCAGGC TGGATC CCG and 5’-GATCCG GGATCC AGCCTG
CTATAT CGAGCC GTTTTT CCTAGA AGATCT GGTACC GACT were syn-
thesized by Eurogentec (Seraing, Belgium) and hybridized by heat-
ing the mixture at 958C for 5 min followed by a slow temperature
decrease to room temperature. The double-stranded DNA was
then subcloned in pUC19 previously opened at SacI and BamHI
sites. The 81 bp DNA fragment encompassing this subcloned se-
quence was obtained from EcoRI and PstI double digestion of this
new pUC19–ATATA vector and 3’-end labeled by using a-[32P]dATP
(3000 Cimmolꢁ1 each, PerkinElmer) and ten units of Klenow
enzyme (BioLabs, ꢅvry, France) for 30 min at 378C, separated and
isolated from the plasmid remnant by using a 6% native polyacryl-
amide gel, as previously described.[38] Increasing concentrations (as
indicated in the figure legends) of the various tested compounds
were incubated for 15 min at 378C with the radiolabeled DNA frag-
ments prior to digestion with DNase I (0.001 unitmLꢁ1, Sigma) for
3 min in digestion buffer (20 mm NaCl, 2 mm MgCl2, 2 mm MnCl2,
pH 7.3). Reaction was stopped by freeze-drying and lyophilization.
The cleaved DNA fragments were dissolved in formamide-contain-
In the fluorescence displacement assay, the DNA–fluorophore com-
plexes were titrated with each text compound at increments of
0.5 mL (0.5 mm compound stock solutions). lex was set to 342 nm
for DAPI and to 363 nm for DB829. For DAPI, a 2.5 nm (excitation
and emission) slit width was chosen, and for DB829 it was 5.0 nm.
The fluorescence intensity at maximum peak was recorded for
each scan. In order to compare the two assays, they were plotted
as ratio values [Eqs. (1a) and (1b)]:
RDAPI ¼ F=Fmax
RDB829 ¼ Fmin=F
ð1aÞ
ð1bÞ
here F is the observed fluorescence at each point, Fmin and Fmax are
the minimum and maximum intensities in each titration, and each
titration starts at a ratio of 1.0.
Circular dichroism spectroscopy: CD spectra were obtained on
a computer-controlled Jasco J-710 spectrometer in 1 cm quartz
cells. Typically, a buffered solution of DNA hairpin at a strand con-
centration of 3 mm was prepared, and the CD spectrum was col-
lected from 480–230 nm at a rate of 50 mmminꢁ1 at 258C. The re-
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 77

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
ing denaturing loading buffer (4 mL), heat-denatured for 4 min at
908C, and rapidly chilled on ice prior to electrophoresis on a 8%
denaturing polyacrylamide gel. The gels were then soaked in 10%
acetic acid, transferred to Whatman 3 MM paper, and dried under
vacuum at 808C. Dried gels were exposed overnight on storage
screens and scanned by using a Molecular Dynamics STORM 860.
Quantification of the cleaved bands was performed by using
Image Quant 4.1 software.
pounds in Table 1 with a molecular mechanics MMFF approxima-
tion level with the Spartan’10 software package (Wavefunction
Inc.). The Spartan’10 software package was employed to optimize
the final geometry by using ab initio calculations with B3LYP DFT
at the 631G* approximation level. The molecular energy was calcu-
lated by employing the Hartree–Fock approximation also at the
631G* level. To evaluate electrostatic and structural properties,
MEP color-coded maps were generated in the range from 250
(deep red) to 700 kJmolꢁ1 (deep blue) and superimposed onto the
molecular surface; red represents regions with the most negative
electrostatic potential, blue represents the most positive, and
green represents regions of zero potential. Negative electrostatic
potential corresponds to attraction of a proton by the aggregate
electron density in the molecule; positive electrostatic potential
corresponds to repulsion of a proton by the atomic nuclei.
X-ray crystallography: The oligonucleotide sequence, d(CGCGAA
TTCGCG)2, was purchased from DNA Technology A/S (Risskov, Den-
mark), and DB1880 was used as the hydrochloride salt without any
further purification. The final dsDNA stock solution was 3.0 mm by
annealing (6.0 mm in single-stranded DNA in 20 mm sodium caco-
dylate, pH 6.5) through heating the mixture to 858C for 15 min
and then cooling it to room temperature overnight. DNA complex
crystals were grown by the hanging-drop, vapor-diffusion method.
A successful crystallization experiment typically comprised an-
nealed dsDNA (1 mL, 1.5 mm) with ligand DB1880 (2.25 mm), mixed
with reagent solution (1 mL; 7% 2-methylpentane-2,4-diol (MPD),
140 mm MgCl2, 20 mm sodium cacodylate, pH 6.5). The hanging
drop was equilibrated against a well containing 50% MPD. Crystals
grew in one week at 108C. A dataset was collected at 105 K from
a single flash-frozen crystal by using an Oxford Diffraction Xcalibur
NovaT X-ray diffractometer. The data were processed and scaled
by using CrysalisPro (Oxford Diffraction) and Scala (CCP4 suite).
Molecular docking: Molecular docking studies were performed
with the SYBYL-X1.2 software package on a Windows 8 processor
Workstation.[47] The Surflex-Dock module of the SYBYL software
uses a surface-shape-based method that aligns each test ligand to
a “protomol”.[48–51] The protomol consists of a series of molecular
fragments that characterize the surface properties of the target
active site, including steric effects, hydrogen bond acceptors and
hydrogen bond donors.[48] Docking of the selected compounds
(Table 1) into the DNA minor groove consisted of three steps:
1) preparation of the 3D structure of a DNA sequence and con-
struction of the protomol, 2) preparation of each compound, and
3) docking of each compound into the protomol.
The structure was solved by molecular replacement using the
REFMAC 5.5.0109 program (CCP4), using the DB819–d(CGCGAA
TTCGCG)2 complex structure (PDB ID: 2B3E)[46] as a model, and re-
fined by using REFMAC 5.5.0109. Data collection and refinement
statistics are shown in Table 3. The DB1880 ligand and Mg2+ ion
could be clearly seen in the initial sA-weighted 2FoꢁFc electron-
density maps. The final model (including solvent molecules) was
refined by using data between 22.66 and 1.90 ꢄ, with final R and
Rfree values of 0.163 and 0.240, respectively.
The X-ray crystal structure of the DB1880–DNA complex (PDB ID:
3OIE) was used to generate the protomol. The bound compound
was extracted from the DNA crystal structure and was used as the
ligand for protomol generation in the following Surflex-docking
steps. After the crystallographic water molecules and metal ions
had been removed and hydrogen atoms had been added, the
modified DNA sequences were minimized for a maximum of
100 iterations with a termination gradient of 0.01 kcalmolꢁ1 ꢄꢁ1
.
Structural comparisons of free compounds: Molecular modeling
The Surflex-dock module of the SYBYL software suite was then im-
plemented, and the protomol was generated by using a ligand-
based approach with the extracted AIA reference compound from
the crystal structure. The two important factors that can affect the
size and extent of the protomol generated are “proto_thresh” and
“proto_bloat”. “Proto_thresh” determines how far the protomol ex-
tends into the target site; “proto_bloat” affects how far the proto-
mol extends outside the concavity.[47,50] For the purpose of these
experiments, “proto_thresh” was set to 0.2 and “proto_bloat” was
left at the default values of 0.
studies were initiated by conformation analysis of the tested com-
Table 3. Data collection and refinement statistics for the DB1880–
d(CGCGAA TTCGCG)2 complex crystal structure.
Data collection
sequence
d(CGCGAA TTCGCG)2
space group
P212121
unit cell dimensions
a, b, c [ꢄ]
24.11, 38.42, 66.21
resolution [ꢄ]
Rint [%] overall
I/s
completeness [%]
redundancy
66.19–1.71
3.04
37.88
74.10
1.9
SYBYL-X1.2 software was then employed to construct the test com-
pounds in three-dimensional space. They underwent a short MD
simulation of 1 ns at a constant temperature and volume (NTV).[52]
Briefly, 1) the system temperature was set to 300 K with a coupling
constant of 100 fs, 2) a Maxwell–Boltzmann distribution was em-
ployed for initial atom velocities, 3) the nonbonded pair list was
updated every 25 fs, and 4) the duration of the MD simulations in
vacuo was 1 ns with a time step of 100 fs and a snapshot every
1000 fs. Snapshots from the MD simulation displayed several low-
energy structures. These were minimized to convergence by using
the Tripos force field with conjugate gradient algorithm, and Gas-
teiger–Hꢆckel charges.[46] The termination gradient was 0.01 kcal
molꢁ1 ꢄꢁ1, and the maximum iterations were 104. The Surflex-dock
GeomX module of the SYBYL software suite was then implemented
to dock each compound into the protomol. Each docking starts
from six multiple initial poses to ensure good search coverage.
Refinement
resolution limits [ꢄ]
no. of reflections
completeness [%]
Rwork/Rfree [%]
no. of atoms
no. of ions
no. of waters
overall B-factor [ꢄ2]
RMS deviations
bond lengths [ꢄ]
bond angles [8]
PDB ID
22.7–1.90
4962
95.1
16.3/24.0
672
1
136
21.8
0.01
1.2
3OIE
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 78

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
[24] M. Rettig, M. W. Germann, S. Wang, W. D. Wilson, ChemBioChem 2013,
14, 323–331.
Acknowledgements
[25] N. S. Carter, B. J. Berger, A. H. Fairlamb, J. Biol. Chem. 1995, 270, 28153–
28157.
[26] M. N. C. Soeiro, K. Werbovetz, D. W. Boykin, W. D. Wilson, M. Z. Wang, A.
Hemphill, Parasitology 2013, 140, 929–951.
This work was supported by NIH grant AI064200 and by The Bill
and Melinda Gates Foundation through a subcontract with the
Consortium of Parasitic Drug Development (W.D.W. and D.W.B.),
Cancer Research UK grant C129/A4489 (to S.N.), the Fonds Eu-
ropꢀen de Dꢀveloppement Rꢀgional (FEDER, European Communi-
ty) and the Rꢀgion Nord–Pas de Calais, the Ligue Nationale
Contre le Cancer (Comitꢀ du Nord, Septentrion), the Association
Laurette Fugain and the Institut pour la Recherche sur le Cancer
de Lille (IRCL) (to M.H.D.C.). R.N. thanks the IRCL for a post-doc-
toral fellowship.
[27] C. E. Stephens, R. Brun, M. M. Salem, K. A. Werbovetz, F. Tanious, W. D.
Wilson, D. W. Boykin, Bioorg. Med. Chem. Lett. 2003, 13, 2065–2069.
[28] C. F. Silva, M. M. Batista, R. A. Mota, E. M. de Souza, C. E. Stephens, P.
Som, D. W. Boykin, N. Soeiro, Biochem. Pharmacol. 2007, 73, 1939–
1946.
[29] M. Z. Wang, X. H. Zhu, A. Srivastava, Q. Liu, J. M. Sweat, T. Pandharkar,
C. E. Stephens, E. Riccio, T. Parman, M. Munde, S. Mandal, R. Madhubala,
R. R. Tidwell, W. D. Wilson, D. W. Boykin, J. E. Hall, D. E. Kile, K. A. Werbo-
vetz, Antimicrob. Agents Chemother. 2010, 54, 2507–2516.
[30] C. F. Silva, M. B. Meuser, E. M. De Souza, M. N. Meirelles, C. E. L. Stephens,
P. Som, D. W. Boykin, N. Soeiro Mde, Antimicrob. Agents Chemother.
2007, 51, 3803–3809.
[31] X. H. Zhu, Q. Liu, S. H. Yang, T. F. Parman, C. E. Green, J. C. Mirsalis, M. de
Nazarꢀ Correia Soeiro, E. Mello de Souza, C. F. da Silva, D. da Gama Jaen
Batista, C. E. Stephens, M. Banerjee, A. A. Farahat, M. Munde, W. D.
Wilson, D. W. Boykin, M. Z. Wang, K. A. Werbovetz, Antimicrob. Agents
Chemother. 2012, 56, 3690–3699.
Keywords: arylimidamides · DNA · minor-groove binders ·
molecular modeling · X-ray crystallography
[1] J. R. Coura, P. A. ViÇas, Nature 2010, 465, S6–S7.
[2] W. Apt, Drug Des. Dev. Ther. 2010, 243–253.
[32] W. D. Wilson, F. A. Tanious, M. Fernandez-Saiz, C. T. Rigl in Methods in
Molecular Biology, Vol. 90: Evaluation of Drug–Nucleic Acid Interactions
by Thermal Melting Curves (Ed.: K. R. Fox), Humana, Totowa, 1998,
pp. 219–240.
[33] W. D. Wilson, S. Mizan, F. A. Tanious, S. Yao, G. Zon, J. Mol. Recognit.
1994, 7, 89–98.
[34] B. Nguyen, F. A. Tanious, W. D. Wilson, Methods 2007, 42, 150–161.
[35] Y. Liu, W. D. Wilson in Methods in Molecular Biology, Vol. 613: Quantita-
tive Analysis of Small Molecule–Nucleic Acid Interactions with a Biosensor
Surface and Surface Plasmon Resonance Detection (Ed.: K. R. Fox),
Humana, Totowa, 2010, pp. 1–23.
[36] R. Nanjunda, M. Munde, Y. Liu, W. D. Wilson in Methods for Studying
DNA/Drug Interactions, (Eds.: M. Wanunu, Y. Tor), CRC Press, Boca Raton,
2011, Chapter 4.
[37] T. M. Davis, W. D. Wilson, Methods Enzymol. 2001, 340, 22–51.
[38] P. Peixoto, Y. Liu, S. Depauw, M. P. Hildebrand, D. W. Boykin, C. Bailly,
W. D. Wilson, M. H. David-Cordonnier, Nucleic Acids Res. 2008, 36, 3341–
3353.
[39] Y. Liu, A. Kumar, S. Depauw, R. Nhili, M.-H. David-Cordonnier, M. P. Lee,
M. A. Ismail, A. A. Farahat, M. Say, S. Chackal-Catoen, A. Batista-Parra, S.
Neidle, D. W. Boykin, W. D. Wilson, J. Am. Chem. Soc. 2011, 133, 10171–
10183.
[3] W. de Souza, Microbes Infect. 2007, 9, 544–545.
[4] C. J. Schofield, J. Jannin, R. Salvatella, Trends Parasitol. 2006, 22, 583–
588.
[5] M. N. Soeiro, K. Werbovetz, D. W. Boykin, W. D. Wilson, M. Z. Wang, A.
Hemphill, Parasitology 2013, 140, 929–951.
[6] M. P. Barrett, D. W. Boykin, R. Brun, R. R. Tidwell, Br. J. Pharmacol. 2007,
152, 1155–1171.
[7] W. D. Wilson, B. Nguyen, F. A. Tanious, A. Mathis, J. E. Hall, C. E. Ste-
phens, D. W. Boykin, Curr. Med. Chem. Anticancer Agents 2005, 5, 389–
408.
[8] W. D. Wilson, F. A. Tanious, A. Mathis, D. Tevis, J. E. Hall, D. W. Boykin,
Biochimie 2008, 90, 999–1014.
[9] M. C. Machado Motta, Curr. Pharm. Des. 2008, 14, 847–854.
[10] A. M. Mathis, A. S. Bridges, M. A. Ismail, A. Kumar, I. Francesconi, M. An-
bazhagan, Q. Y. Hu, F. A. Tanious, T. Wenzler, J. Saulter, W. D. Wilson, R.
Brun, D. W. Boykin, R. R. Tidewell, J. E. Hall, Antimicrob. Agents Chemo-
ther. 2007, 51, 2801–2810.
[11] D. S. Tevis, A. Kumar, C. E. Stephens, D. W. Boykin, W. D. Wilson, Nucleic
Acids Res. 2009, 37, 5550–5558.
[12] T. A. Shapiro, P. T. Englund, Proc. Natl. Acad. Sci. USA 1990, 87, 950–954.
[13] O. Vꢇzquez, M. I. Sꢇnchez, J. Martꢈnez-Costas, M. E. Vꢇzquez, J. L. Mascar-
eÇas, Org. Lett. 2010, 12, 216–219.
[40] D. G. Brown, M. R. Sanderson, J. V. Skelly, T. C. Jenkins, T. Brown, E.
Garman, D. Stuart, S. Neidle, EMBO J. 1990, 9, 1329–1334.
[41] C. A. Laughton, F. Tanious, C. M. Nunn, D. W. Boykin, W. D. Wilson, S.
Neidle, Biochemistry 1996, 35, 5655–5661.
[42] B. Nguyen, S. Neidle, W. D. Wilson, Acc. Chem. Res. 2009, 42, 11–21.
[43] S. Neidle, Nat. Prod. Rep. 2001, 18, 291–309.
[44] D. Wei, W. D. Wilson, S. Neidle, J. Am. Chem. Soc. 2013, 135, 1369–1377.
[45] C. E. Stephens, F. Tanious, S. Kim, W. D. Wilson, W. A. Schell, J. R. Perfect,
S. G. Franzblau, D. W. Boykin, J. Med. Chem. 2001, 44, 1741–1748.
[46] N. H. Campbell, D. A. Evans, M. A. Lee, G. N. Parkinson, S. Neidle, Bioorg.
Med. Chem. Lett. 2006, 16, 15–19.
[47] SYBYL Molecular Modeling Software, Version X1.2, Tripos Inc., St. Louis,
MO, 2010.
[48] A. N. Jain, J. Med. Chem. 2003, 46, 499–511.
[49] A. N. Jain, J. Comput.-Aided Mol. Des. 2007, 21, 281–306.
[50] P. A. Holt, J. B. Chaires, J. O. Trent, J. Chem. Inf. Model. 2008, 48, 1602–
1615.
[51] M. M. Dailey, C. Hait, P. A. Holt, J. M. Maguire, J. B. Meier, M. C. Miller, L.
Petraccone, J. O. Trent, Exp. Mol. Pathol. 2009, 86, 141–150.
[52] C. J. Collar, X. Zhu, K. Werbovetz, D. W. Boykin, W. D. Wilson, Bioorg.
Med. Chem. 2011, 19, 4552–4561.
[14] M. I. Sꢇnchez, O. Vꢇzquez, J. Martꢈnez-Costas, M. E. Vꢇzquez, J. L. Mascar-
eÇas, Chem. Sci. 2012, 3, 2383–2387.
[15] T. Wenzler, D. W. Boykin, M. A. Ismail, J. E. Hall, R. R. Tidwell, R. Brun, An-
timicrob. Agents Chemother. 2009, 53, 4185–4192.
[16] J. K. Thuita, M. Z. Wang, J. M. Kagira, C. L. Denton, M. F. Paine, R. E.
Mdachi, G. A. Murilla, S. Ching, D. W. Boykin, R. R. Tidwell, J. E. Hall, B.
Brun, PLoS Neglected Trop. Dis. 2012, 6, e1734.
[17] D. W. Boykin, A. Kumar, J. E. Hall, B. C. Bender, R. R. Tidwell, Bioorg. Med.
Chem. Lett. 1996, 6, 3017–3020.
[18] J. H. Ansede, M. Anbazhagan, R. Brun, J. D. Easterbrook, J. E. Hall, D. W.
Boykin, J. Med. Chem. 2004, 47, 4335–4338.
[19] R. R. Tidwell, D. W. Boykin in DNA and RNA Binders: From Small Mole-
cules to Drugs, Vol. 2 (Eds.: M. Demeunynck, C. Bailly, W. D. Wilson),
Wiley-VCH, Weinheim, 2003, pp. 414–460.
[20] M. F. Paine, M. Z. Wang, C. N. Generaux, D. W. Boykin, W. D. Wilson, H. P.
De Koning, C. A. Olson, G. Pohlig, C. Burri, R. Brun, G. A. Murilla, J. K.
Thuita, M. P. Barrett, R. R. Tidwell, Curr. Opin. Invest. Drugs 2010, 11,
876–883.
[21] A. Mayence, J. J. Eynde, F. M. Krogstad, D. J. Krogstad, M. T. Cushion, T. L.
Huang, J. Med. Chem. 2004, 47, 2700–2705.
[22] H. Denise, M. P. Barrett, Biochem. Pharmacol. 2001, 61, 1–5.
[23] R. A. Hunt, M. Munde, A. Kumar, M. A. Ismail, A. A. Farahat, R. K. Arafa,
M. Say, A. Batista-Parra, D. S. Tevis, D. W. Boykin, W. D. Wilson, Nucleic
Acids Res. 2011, 39, 4265–4274.
Received: September 25, 2013
Published online on December 9, 2013
ꢂ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2014, 15, 68 – 79 79