Binding of a homologous series of anthraquinones to DNA

Article abstract of DOI:10.1016/j.bmcl.2003.10.054

The DNA binding characteristics of a series of homologous 2, 6-disubstituted anthraquinone threading intercalators bearing one to four ethylene glycol units in their side chains have been studied. Binding constants were measured via surface plasmon resonance (SPR). These compounds bind to an AT-rich hairpin with slightly higher affinity than to a GC-rich hairpin. The binding constants decrease as the length of the side chain increases.

Full text of DOI:10.1016/j.bmcl.2003.10.054

Bioorganic & Medicinal Chemistry Letters 14 (2004) 401–404
Binding of a homologous series of anthraquinones to DNA
Ruel E. McKnight, Jianguo Zhang and Dabney W. Dixon*
Department of Chemistry, Georgia State University, Atlanta, GA 30303, USA
Received 18 July 2003; revised 16 October 2003; accepted 30 October 2003
Abstract—The DNA binding characteristics of a series of homologous 2,6-disubstituted anthraquinone threading intercalators
bearing one to four ethylene glycol units in their side chains have been studied. Binding constants were measured via surface plas-
mon resonance (SPR). These compounds bind to an AT-rich hairpin with slightly higher affinity than to a GC-rich hairpin. The
binding constants decrease as the length of the side chain increases.
# 2003Elsevier Ltd. All rights reserved.
The factors that control small molecule intercalation in
DNA continue to be the focus of study.1ꢀ9 Overall, the
hydrophobicity of the molecule and net charge are
usually the most significant. Additional factors specific
to the side chains include their steric bulk, isohelicity of
the side chains with the minor groove, and phasing of
the ligand subunits with the edges of the base pairs.10ꢀ12
Loss of rotational and translational degrees of freedom
upon binding have also been considered, though con-
tributions from these have generally been difficult to
assess.2 One approach to analyzing the contributions of
the factors controlling binding is to compare binding
of a series of homologous compounds.
turn from 2,6-diaminoanthracenedione and bromopro-
pionyl chloride.13ꢀ15 The amines were either available
from commercial sources (I and II) or synthesized from
the glycol methyl ether and acrylonitrile, followed by
reduction with hydrogen and a catalyst (III and IV).16
Surface plasmon resonance (SPR) based Biacore tech-
nology17 was used to measure the binding of com-
pounds I–IV to AT- and GC-rich DNA hairpins (Fig.
2).18 The instrument response (RU) was plotted as a
function of time to generate sensograms. Data from the
sensograms were cast into binding plots of response as a
function of the concentration of free drug (Cfree) (Fig.
2). In each case, the observed response (RUobs) was
In the present study, we report the synthesis and DNA
binding characteristics of a homologous series of 2,6-
disubstituted anthraquinones with increasing side chain
size (Fig. 1). These symmetrical molecules are threading
intercalators, that is with one chain each in the major
and minor groove.1,3 The side chains were designed such
that the cationic charge (protonated secondary amine)
would have a constant position in the series. The side
chains were based on the polyethylene glycol repeating
unit. This allows a significant increase in the length of
the side chain with a relatively small change in the
hydrophobicity of the molecule.
greater than the maximum predicted response
18
(RUmax,pred
)
indicating that there were multiple bind-
ing sites. Scatchard and Hill analyses revealed that each
compound bound non-cooperatively to each DNA
sequence. Therefore, the data were fit by non-linear
least squares fitting analysis to a multiple equivalent site
equilibrium model (eq 1):19
RUobs=RUmax;pred ¼ nKCfree=ð1 þ KCfree
Þ
ð1Þ
where n is the number of binding sites and K is the
binding constant. Fits of the data to other binding
models gave much larger errors.
The anthraquinones were synthesized by condensing an
excess of the appropriate amine with 2,6-bis(3-bromo-
propionamido)anthracene-9,10-dione, synthesized in
Table 1 shows that the binding affinity for the AT-rich
sequence was 2–3times greater than for the GC-rich
sequence for all four compounds in the series. Tanious
et al. had previously studied the binding of several
* Corresponding author. Tel.: +1-404-651-3908; fax: +1-404-651-
1416; e-mail: ddixon@gsu.edu
0960-894X/$ - see front matter # 2003Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.10.054

402
R. E. McKnight et al. / Bioorg. Med. Chem. Lett. 14 (2004) 401–404
anthraquinone threading intercalators to poly(dA-dT)
and poly(dG-dC) oligonucleotide sequences.20 They
found binding constants for the poly(dA-dT) sequence
that were on average about 5 times greater than that for
the poly(dG-dC) sequence, consistent with the current
study.
workers, using optical titration methods and an exten-
ded McGhee–von Hippel model, found that related
anthraquinones occluded 3–4 base pairs when binding
to calf thymus DNA.13 These two methods of measur-
ing binding are consistent with one another, in that the
stem portion of the hairpin (presumably the major
binding site) has 8–9 base pairs.
The observed instrument response at high concentra-
tions of added anthraquinone showed that each hairpin
bound more than one molecule. Data fits indicated that
the AT-rich sequence had approximately one more
binding site (n=3.1ꢁ0.4) than the GC-rich sequence
(n=2.1ꢁ0.5) for all four compounds. Neidle and co-
The binding constant decreases with increasing length
of the side chain for anthraquinones I–IV for both the
AT- and GC-rich sequences. This result was consistent
with DNA thermal melting studies,21 which showed a
decrease in ꢀTm from compound I–IV (Table 1). Figure
3 shows the binding constants in this study plotted as a
function of the molecular volume of the molecule.22 The
data set can be expanded if we also look at the previous
work of Neidle and co-workers on compounds with the
same central core but different side chains.13 The entire
data set shows a consistent pattern: the binding con-
stant decreases as the molecular volume increases.
It should also be noted that the hydrophobicity of the
molecule decreases as the chain length increases for the
intercalators studied herein. Specifically, the calculated
octanol/water partition coefficient, ClogP, changes by
ꢀ0.14 with the addition of each –CH2CH2O– group.23
Figure 3 plots the binding constants as a function of
ClogP for both compounds I–IV of this study and the
five related compounds from the study of Neidle and
co-workers.13 Overall, there is no dependence of the
binding constant on hydrophobicity. Thus, molecular
size is a better predictor of DNA binding than hydro-
phobicity for these anthraquinones, with molecules
bearing larger side chains binding more poorly.
Figure 1. Homologous series of 2,6-disubstituted anthraquinones with
polyethylene glycol side chains.
Our data may be compared with two other studies in
which the DNA binding constant was measured as a
function of the length of the side chain within a homo-
logous series. Wakelin et al. studied five homologous 9-
aminoacridinecarboxamides.24 The series was synthe-
sized by incorporating methylene units in the side chain,
thus increasing the hydrophobicity of the molecule with
increasing length of the side chain (ClogP increases by
+0.53with the addition of each methylene group). For
this series, binding to poly(dA-dT) was independent of
chain length; binding to poly(dG-dC) decreased slightly
(by a factor of 2) over the series as the chain became
longer. Maleev et al. studied the binding of four homo-
logous disubstituted actinocin derivatives to calf thymus
DNA.25 They observed about an order of magnitude
decrease in the binding constant for intercalation as the
number of methylene groups in the side chains were
Table 1. Binding constants for compound I–IV binding to the AT-
and GC-rich DNA sequence
a
Compd
Molecular
volume (A3)
ꢀTm
(ꢂC)
K (AT)
(105 Mꢀ1
K (GC)
(105 Mꢀ1
)
)
I
II
III
IV
652
774
889
1009
29.1
26.7
24.0
22.2
7..61ꢁ0.33
5.5ꢁ0.2
ꢁ0.2
2.2ꢁ0.3
ꢁ0.4
2.5ꢁ0.31.1
Figure 2. AT- and GC-rich DNA hairpins used for study (top). Bind-
ing plot for the binding of compound II to the AT-rich sequence
(bottom).
1.9ꢁ0.1
0.9ꢁ0.1
a ꢀTm (TmcomplexꢀTmDNA) were determined using calf thymus DNA.

R. E. McKnight et al. / Bioorg. Med. Chem. Lett. 14 (2004) 401–404
403
References and notes
1. Johnson, D. S.; Boger, D. L. DNA Binding Agents. In
Comprehensive Supramolecular Chemistry; Atwoods, J. L.,
Davies, J. E. L., MacNicol, D. D., Vogtle, F., Lehn, J.-
M., Eds.; Pergamon: Oxford, 1997; Vol. 4, p 73.
2. Chaires, J. B. Biopolymers 1997, 44, 201.
3. Wilson, W. D. DNA Intercalators. In DNA and Aspects of
Molecular Biology; Kool, E. T., Ed.; Elsevier: New York,
1999; Vol. 7, p 427.
4. Graves, D. E.; Velea, L. M. Curr. Org. Chem. 2000, 4,
915.
5. Qu, X.; Chaires, J. B. Methods Enzymol. 2000, 321, 353.
6. Haq, I.; Jenkins, T. C.; Chowdhry, B. Z.; Ren, J.; Chaires,
J. B. Methods Enzymol. 2000, 323, 373.
7. Haq, I.; Chowdhry, B. Z.; Jenkins, T. C. Methods Enzy-
mol. 2001, 340, 109.
8. Qu, X. G.; Chaires, J. B. J. Am. Chem. Soc. 2001, 123, 1.
9. Haq, I. Arch. Biochem. Biophys. 2002, 403, 1.
10. Bailly, C.; Chaires, J. B. Bioconjug. Chem. 1998, 9, 513.
11. Wemmer, D. E. Annu. Rev. Biophys. Biomol. Struct. 2000,
29, 439.
12. Neidle, S. Nat. Prod. Rep. 2001, 18, 291.
13. Agbandje, M.; Jenkins, T. C.; McKenna, R.; Reszka,
A. P.; Neidle, S. J. Med. Chem. 1992, 35, 1418.
Figure 3. Plots of binding constant (log units) as a function of mole-
cular volume (top) and ClogP (bottom). Data shown are for com-
pound I–IV binding the AT-rich (closed triangles) and GC-rich (closed
squares) sequences, and from work by Neidle and co-workers (open
diamonds).
14. Breslin, D. T.; Schuster, G. B. J. Am. Chem. Soc. 1996,
118, 2311.
15. As an example, for the synthesis of III, a 10-fold excess of
3-[2-(2-methoxyethoxy)ethoxy]propylamine (a mixture
with the nitrile and starting alcohol) was allowed to react
directly with 0.20 g (0.39 mmol) of bis(3-bromopropion-
amido)anthracene-9,10-dione by refluxing in ethanol for
36 h. The product was filtered and the filtrate evaporated
under reduced pressure. Recrystallization was carried out
from CHCl3–EtOH (3:2, v/v) to give a yellow powder
(0.19 g, 70%). 1H NMR (300 MHz, DMSO-d6) d 1.82 (m,
4H, CH2CH2CH2), 2.82 (t, J=6.6 Hz, 4H), 2.93(t, J=6.9
Hz, 4H), 3.23 (s, OCH3), 8.13(dd, J=2.4, 8.4 Hz, 2H,
ArH), 8.19 (d, J=8.4 Hz, 2H, ArH), 8.50 (d, J=2.4 Hz,
2H, ArH). The other peaks are presumably underneath
DMSO-d6 and between 3.2 and 3.6 ppm. FAB-MS (thio-
glycerol) m/z 701.3753 (M+H+ calcd 701.3762 for
C36H53N4O10). For III and IV, the final products for
DNA studies were purified by HPLC on a Beckman
Ultrasphere ODS C-18 column with methanol as the
mobile phase. The other anthraquinones were prepared in
a similar fashion.
increased from two to five. The observation that the
binding constants decrease as a function of the chain
length for series where the hydrophobicity either
decreases (anthraquinone series herein) or increases (9-
aminoacridinecarboxamide and actinocin series) indi-
cates the importance of chain length in determining the
binding constant. Further analysis of structure-binding
correlations among these series is complicated by dif-
ferences in the distance between the intercalating moiety
and the cationic charge in the side chains; the distance is
constant in the anthraquinone series but increases as the
side-chain length increases in the 9-aminoacridine-
carboxamide and actinocin series.
The effect of side-chain size on DNA binding pre-
sumably results from loss of rotational degrees of free-
dom of the side chains on binding of the ligand to the
DNA. The energetic cost resulting from this should
increase from compound I–IV, resulting in the observed
decrease in the binding constant. The observation that
side-chain length is a better predictor of DNA binding
than hydrophobicity for the anthraquinones, and the
observation of decreased DNA binding with increasing
side-chain length for series that both increase and
decrease in hydrophobicity, indicate the importance of
side-chain length as a determinant of the equilibrium
constant for DNA binding.
16. 3-[2-(2-Methoxyethoxy)ethoxy]propionitrile was synthe-
sized by condensing di(ethylene glycol) methyl ether with
an excess of acrylonitrile.26 This was hydrogenated over
Raney nickel at 55 psi26 to give 3-[2-(2-methoxyethoxy)
ethoxy]propylamine as a mixture with the starting materi-
als. 3-2-(2-(2-Methoxyethoxy)ethoxy)ethoxypropylamine
was synthesized similarly, except that 5% rhodium on alu-
mina was used as the hydrogenation catalyst.27
17. Davis, T. M.; Wilson, W. D. Methods Enzymol. 2001, 340,
22.
18. Binding interactions were measured with a BIACORE
3000 optical biosensor instrument (Biacore, Uppsala,
Sweden). All interactions were measured on a streptavidin
coated sensor chip (Biacore) at 25 ꢂC. 50-Biotin-labeled
DNA hairpins were obtained as a gift from Dr. David
Wilson. They were originally obtained as anionic
exchange and HPLC purified products from Midland
Certified Reagents (Midland, TX, USA) and were used
without further purification. A GC base pair was placed
at the end of the AT-rich hairpin in order to minimize the
risk of fraying. DNA hairpins (25 nM) were dissolved in
HBS-EP buffer and immobilized via biotin capture. A
Acknowledgements
We thank BASF for the gift of the amine precursor for
II. We thank Dr. David Wilson and Farial Tanious for
their assistance with the Biacore work and helpful dis-
cussions. This work was supported by the National
Institutes of Health AI127196. The Biacore was pur-
chased with funds from the Georgia Research Alliance.

404
R. E. McKnight et al. / Bioorg. Med. Chem. Lett. 14 (2004) 401–404
flow rate of 2 mL/min was used in order to control the 20. Tanious, F. A.; Jenkins, T. C.; Neidle, S.; Wilson, W. D.
amount of DNA immobilized. In each case, the amount
of DNA immobilized was equivalent to approximately
350 RU. For the binding analysis, several concentrations
Biochemistry 1992, 31, 11632.
21. Thermal melting (Tm) experiments were conducted with a
Cary 3E spectrophotometer (Varian) attached to a tem-
perature controller. Experiments were done in MES buffer
without NaCl. Tm melting profiles were obtained by
measuring the absorbance at 260 nm as a function of
temperature (25–95 ꢂC) and at a ramp rate of 0.5 ꢂC/min.
The concentration of calf thymus DNA was generally
5ꢃ10ꢀ5 M with a drug/DNA ratio of 0.3:1.
(5 nM–5 mM) of compound in MES 10 buffer [1ꢃ10ꢀ2
M
MES (2-(N-morpholino)ethanesulfonic acid) containing
1ꢃ10ꢀ3 M EDTA and 0.1 M NaCl with the pH adjusted
to 6.25 with NaOH] were injected at a flow rate of 20 mL/
min. After each injection the compound was dissociated
and the surface regenerated using running buffer and
regeneration buffer (10 mM Gly, pH 2). The response
from the reference channel (channel 1) was subtracted
from the responses from the channels with DNA (chan-
nels 2 and 3). At steady-state, the instrument response is
proportional to the amount of bound compound.19 Aver-
age fitting of the response at steady-state was performed
with the BIAevaluation 3.1 program and the binding
constants determined using KaleidaGraph (version 3.5,
Synergy Software, Reading, PA, USA). The maximum
22. Molecular volumes were calculated using PCModel (ver-
sion 8.5, Serena Software, Bloomington, IL, USA). Data
shown are for the Monte Carlo calculation. Data using
the Connolly analytical formulation29 were the same
within 1%.
23. ClogP values were calculated using ChemDraw (version
7.0, CambridgeSoft, Cambridge, MA, USA). In view of
the atom limit of the algorithm, the ClogP values are
reported for the anthraquinone and one side chain only.
24. Wakelin, L. P. G.; Atwell, G. J.; Rewcastle, G. W.;
Denny, W. A. J. Med. Chem. 1987, 30, 855.
25. Maleev, V.; Semenov, A.; Kruglova, E.; Bolbukh, T.;
Gasan, A.; Bereznyak, E.; Shestopalova, A. J. Mol.
Struct. 2003, 645, 145.
predicted response per bound compound (RUmax,pred
)
is given by: RUmax,pred=(RUDNAꢃMWcompoundꢃR)/
MWDNA, where RUDNA is the response of the immobi-
lized DNA, MWcompound and MWDNA are the molecular
weights for the compound and DNA, respectively, and R
is the ratio of the refractive index increment of the com-
pound to DNA. For this series of compounds, R was
taken to be 1.2.28
26. Utermohlen, W. P. J. Am. Chem. Soc. 1945, 67, 1505.
27. Cativiela, C.; Diaz-de-Villegas, M. D.; Galvez, J. A. J.
Org. Chem. 1994, 59, 2497.
19. Carrasco, C.; Facompre, M.; Chisholm, J. D.; Van Vran-
ken, D. L.; Wilson, W. D.; Bailly, C. Nucleic Acids Res.
2002, 30, 1774.
28. Davis, T. M.; Wilson, W. D. Anal. Biochem. 2000, 284,
348.
29. Connolly, M. L. J. Mol. Graph. 1993, 11, 139.