Selective intercalation of charge neutral intercalators into GG and CG steps: Implication of HOMO-LUMO interaction for sequence-selective drug intercalation into DNA

Article abstract of DOI:10.1021/ja003956i

We have synthesized naphthopyranone epoxide 4 from D-isoascorbic acid together with its three diastereoisomers. DNA alkylation of ODNs containing 5′XGT3′ and 5′TGY3′ by 4 (11R, 13R), where X and Y are any nucleotide bases, occurred at all G residues excep

Full text of DOI:10.1021/ja003956i

J. Am. Chem. Soc. 2001, 123, 5695-5702
5695
Selective Intercalation of Charge Neutral Intercalators into GG and
CG Steps: Implication of HOMO-LUMO Interaction for
Sequence-Selective Drug Intercalation into DNA
Kazuhiko Nakatani,* Takahiro Matsuno, Kaoru Adachi, Shinya Hagihara, and Isao Saito*
Contribution from the Department of Synthetic Chemistry and Biological Chemistry,
Faculty of Engineering, Kyoto UniVersity, CREST, Japan Science and Technology Corporation (JST),
Kyoto 606-8501, Japan
ReceiVed NoVember 14, 2000
Abstract: We have synthesized naphthopyranone epoxide 4 from D-isoascorbic acid together with its three
diastereoisomers. DNA alkylation of ODNs containing 5′XGT3′ and 5′TGY3′ by 4 (11R, 13R), where X and
Y are any nucleotide bases, occurred at all G residues except at G of the 5′TGC3′ sequence. In contrast, the
1
three other diastereoisomers of 4 showed only weak G alkylation activity. Differential H NMR NOE of the
4-G adduct confirmed the G-N7 alkylation at the epoxide carbon of 4 with concomitant SN2 ring opening of
the epoxide. Quantitative HPLC analysis of G alkylation efficiency for 4 showed the order of G alkylation
susceptibility as TGGT ≈ CGT . TGA > AGT > TGT . TGC. The order was fully consistent with those
reported for aflatoxin B1 oxide and kapurimycin A3, suggesting that the sequence selectivity observed for
these DNA alkylating agents is not structure dependent but most likely due to the intrinsic property of DNA
sequences. We found that the order of G alkylation susceptibility obtained for 4 completely matched the
calculated HOMO energy level of G-containing sequences. These results underscore that 4 is a unique molecular
probe for ranking the HOMO level of G-containing sequences by well-known G alkylation chemistry and
suggests that the intercalation of charge neutral intercalators is a HOMO-controlled process.
Introduction
Certain types of intercalators are known to eventually produce
covalent DNA complexes. Aflatoxin B1 oxide (1),10 one of the
most potent environmental mutagens, and kapurimycin A3
(2),11,12 a member of pluramycin antibiotics,13,14 are the
representatives of such DNA-alkylating intercalators (Figure 1).
Both 115-20 and 221-23 alkylate N7 of guanine (G) by an
Intercalation into stacked DNA base pairs is one of the most
fundamental modes of interaction between planar aromatic
molecules and duplex DNAs.1-4 Numerous structures of
DNA-intercalator complexes have been determined by X-ray
and NMR analyses.5,6 Intercalation of these molecules resulted
in a significant change of the DNA conformation being
accompanied by unwinding and elongation of the duplex at the
intercalated base steps.2,3,7 Intercalated molecules are stacked
by flanking 3′ and 5′ base pairs. In DNA-intercalator com-
plexes, a long axis of fused aromatic rings of intercalators may
be parallel to the hydrogen bonds of neighboring base pairs,
whereas an orthogonal orientation of intercalators is also known
as a threading type intercalation.8,9
(10) Busby, W. F., Jr.; Wogan, G. N. In Chemical Carcinogens, 2nd
ed.; Searle, C. E., Ed.; American Chemical Society: Washington, DC, 1984;
pp 945-1136.
(11) Hara, M.; Mokudai, T.; Kobayashi, E.; Gomi, K.; Nakano, H. J.
Antibiot. 1990, 43, 1513-1518.
(12) Yoshida, M.; Hara, M.; Saitoh, Y.; Sano, H. J. Antibiot. 1990, 43,
1519-1523.
(13) Se´quin, U. Fortschr. Chem. Org. Naturst. 1986, 50, 57-122.
(14) Hansen, M. R.; Hurley, L. H. Acc. Chem. Res. 1996, 29, 249-258.
(15) Essigmann, J. M.; Green, C. L.; Croy, R. G.; Fowler, K. W.; Bu¨chi,
G. H.; Wogan, G. N. Cold Spring Harbor Symp. Quant. Biol. 1983, 47,
327-337.
(1) Lerman, S. J. Mol. Biol. 1961, 3, 13-80.
(2) Saenger, W. Principles of Nucleic Acids Structure; Springer-Verlag:
New York, 1983; Chapter 16.
(3) Wilsonm, W. D.; Jones, R. L. In Intercalation Chemistry; Whitting-
ham, M. S., Jacobson, A. J., Eds.; Academic Press: New York, 1982;
Chapter 14.
(16) Essigmann, J. M.; Croy, R. G.; Nadzan, A. M.; Busby, W. F., Jr.;
Reinhold, V. N.; Bu¨chi, G.; Wogan, G. N. Proc. Natl. Acad. Sci. U.S.A.
1977, 74, 1870-1874.
(17) Lin, J. K.; Miller, J. A.; Miller, E. C. Cancer Res. 1977, 37, 4430-
4438.
(4) Wilson, W. D. In Nucleic Acids in Chemistry and Biology; Blackburn,
M., Gait, M., Eds.; IRL Press Ltd.: Oxford, 1990; pp 295-336.
(5) Wilson, W. D.; Li, Y.; Veal, J. M. In AdVances in DNA Sequence
Specific Agents; Hurley, L. H., Ed.; JAI Press Ltd.: Greenwich, 1992; Vol.
1, pp 89-166.
(6) Wang, A. H.-J. In AdVances in DNA Sequence Specific Agents;
Hurley, L. H., Ed.; JAI Press Ltd.: Greenwich, 1996; Vol. 2, pp 59-101.
(7) Waring, M. J. Annu. ReV. Biochem. 1981, 50, 159-192.
(8) Daunomycin: Wan, A. H.-J.; Ughetto, G.; Quigley, G. J.; Rich, A.
Biochemistry 1987, 26, 1152-1163.
(9) Nogalamycin: (a) Liaw, Y. C.; Gao, Y. G.; Robinson, H.; van der
Marel, G. A.; van Boom, J. H.; Wan, A. H.-J. Biochemistry 1989, 28, 9913-
9918. (b) Gao, Y. G.; Liaw, Y. C.; Robinson, H.; Wan, A. H.-J. Biochemistry
1990, 29, 10307-10316.
(18) (a) Muench, K. F.; Misra, R. P.; Humayun, M. Z. Proc. Natl. Acad.
Sci. U.S.A. 1983, 80, 6-10. (b) Misra, R. P.; Muench, K. F.; Humayun, M.
Z. Biochemistry 1983, 22, 3351-3359.
(19) (a) Benasutti, M.; Ejadi, S.; Whitlow, M. D.; Loechler, E. L.
Biochemistry 1988, 27, 472-481. (b) Loechler, E. L.; Teeter, M. M.;
Whitlow, M. D. J. Biomol. Struct. Dyn. 1988, 5, 1237-1257.
(20) Gopalakrishnan, S.; Harris, T. M.; Stone, M. P. Biochemistry 1990,
29, 10438-10448.
(21) Chan, K. L.; Sugiyama, H.; Saito, I. Tetrahedron Lett. 1991, 52,
7719-7722.
(22) Hara, M.; Yoshida, M.; Nakano, H. Biochemistry 1990, 29, 10449-
10455.
(23) Nakatani, K.; Okamoto, A.; Saito, I. Angew. Chem., Int. Ed. Engl.
1997, 36, 2794-2797.
10.1021/ja003956i CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/25/2001

5696 J. Am. Chem. Soc., Vol. 123, No. 24, 2001
Nakatani et al.
Figure 1. Structures of aflatoxin B1 oxide (1), kapurimycin A3 (2),
and an aglycon model 3 of pluramycin antibiotics. Sequence selectivity
for G alkylation of each molecule is shown below the structure. G’s
shown with an arrow are the alkylated G’s.
Figure 2. Naphthopyranone epoxides designed for alkylation of both
3′ and 5′ G’s at the intercalation site.
observed for 4 is fully consistent with those previously observed
for 1, 2, and 3, suggesting that the G alkylation selectivity of
charge neutral DNA intercalators is not structure dependent.
Furthermore, the order of G-alkylation susceptibility is fully
consistent with the calculated HOMO level of G-containing
sequences. These observations indicate that interactions between
the HOMO of DNAs and the LUMO of the intercalators may
significantly contribute for the sequence selectivity of the drug
intercalation.
electrophilic attack of the epoxide upon intercalation into DNA,
forming a covalent bond between an epoxide carbon and G-N7
with concomitant SN2 ring opening of the epoxide. Despite the
lack of obvious DNA binding functionality, both 1 and 2 showed
distinct sequence selectivity for G alkylation. Aflatoxin B1 oxide
(1) alkylates 3′ side G at the intercalation site with sequence
preference of GG > CG > AG ≈ TG,19 whereas 5′ side G was
alkylated by 2 with the order of GG > GA > GT > GC (G’s
shown in italic are the site of alkylation).23 We have recently
reported that a synthetic aglycon model 3 of pluramycin
antibiotics also selectively alkylates 5′ side G of the GG
sequence.24,25 It has been pointed out that the molecular basis
for the G alkylation selectivity is ascribable to the different
susceptibility of these base steps toward drug intercalation.18,24
Accordingly, these three molecules preferentially intercalate into
the GG step. On the basis of high level ab initio calculations,
we have rationalized that the high preference for the GG step
in the intercalation of 3 is due to the strong stabilization of the
intercalated complex at the GG step by stacking of the
intercalator with both 3′ and 5′ side G’s.24 However, the
molecular basis for the difference between the high susceptibility
of the CG sequence for 1 and the low reactivity of the GC
sequence for 2 and 3 remain to be clarified.
Results
Molecular Design and Synthesis of a New DNA-Alkylating
Intercalator. To alkylate both side G’s at the intercalation site,
we have designed naphthopyranone derivative 4 that has an
electrophilic epoxide one carbon away from the pyranone ring
as compared to 3 (Figure 2). Three diastereoisomers 5, 6, and
7 were also synthesized to determine the effect of the absolute
configuration at the two chiral centers on the G alkylation
efficiency. Synthesis of 4 starts from acetylenic alcohol 826
possessing the correct absolute configuration of two chiral
centers (see Scheme 1 and Supporting Information for the
experimental procedure and spectral data). Alcohol 8 is readily
obtained from D-isoascorbic acid. The enantiomeric purity of 8
1
was confirmed to be more than 98% by comparing H NMR
When 1 intercalated into the GC step, 5′ side G was not
alkylated due to its intrinsic 3′ side G selectivity for the
alkylation. Likewise, 2 cannot alkylate the 3′ side G of the 2-CG
complex. Therefore, it is not feasible to determine the efficiency
of the intercalation of 1 into the GC step and 2 into the CG
step by G alkylation experiments. This discrepancy of intrinsic
alkylation selectivity between two drugs makes it difficult to
directly compare their sequence selectivity for the intercalation.
In the present work we have synthesized novel DNA alkylating
intercalator 4 that alkylates both 3′ and 5′ side G’s at the
intercalation site to compare the susceptibility of GC and CG
steps toward drug intercalation. Quantitative HPLC analysis of
G alkylation at 5′XGY3′ by 4, where X and Y are any nucleotide
base, established the order of the G alkylation susceptibility of
G-containing sequences. We here describe that the G selectivity
spectra of (R)-MTPA ester 9 and (R)-MTPA esters obtained
from three diastereoisomers of 8. The primary hydroxyl group
of 8 was selectively removed by the following three-step
sequence. Tosylation of 8 and a subsequent base treatment
produced epoxide 11 which was reductively opened by L-
Selectride to give 12. Desilylation of 12 and a subsequent
hydrolysis of acetonide produced triol 13 that was transformed
into tosylate 14 without isolation. The remaining hydroxyl
groups of 14 were protected as tert-butyldimethylsilyl ether to
give 15. Lithiation of 15 followed by addition to 1-tert-
buthyldimethylsilyloxy-2-naphthoaldehyde gave adduct 16.
Oxidation of 16 with manganese dioxide followed by treatment
of the resulting ketone with potassium fluoride in the presence
of 18-crown-6 in dry DMF furnished both pyranone and epoxide
formation to yield 4.27,28 Diastereoisomer 6 was synthesized
(24) Nakatani, K.; Okamoto, A.; Matsuno, T.; Saito, I. J. Am. Chem.
Soc. 1998, 120, 11219-11225.
(25) Nakatani, K.; Okamoto, A.; Saito, I. Angew. Chem., Int. Ed. Engl.
1999, 38, 3378-3381.
(26) Nakatani, K.; Arai, K.; Hirayama, N.; Matsuda, F.; Terashima, S.
Tetrahedron 1992, 48, 633-650.
(27) Nakatani, K.; Okamoto, A.; Yamanuki, M.; Saito, I. J. Org. Chem.
1994, 59, 4360-4361.

Sequence-SelectiVe Drug Intercalation into DNA
J. Am. Chem. Soc., Vol. 123, No. 24, 2001 5697
Scheme 1a
a Reagents and conditions: (a) (R)-MTPA acid, DCC, DMAP, 63%;
(b) TsCl, Py, DMAP, 98%; (c) LiHMDS, 79%; (d) LiB[CH(CH3)-
CH2CH3]3H, 96%; (e) TBAF; (f) 1 N HCl; (g) TsCl, Py, 59% for 3
steps; (h) TBDMSOTf, 2,6-lutidine, 74%; (i) LiHMDS, CeCl3, 1-tert-
butyldimethylsilyloxy-2-naphthoaldehyde, 63%; (j) MnO2; (k) KF, 18-
crown-6, DMF, 40% for 2 steps.
Figure 3. Sequence-selective G alkylation by naphthopyranone
epoxides 4-7. G alkylation reactions by 4-7 were examined with 29-
mer duplexes of (a) d(TTA TTG TTC GTT AGT TGG TAT ATT TAA
TA) (ODN1) and (b) d(TTA TGT TTG CTT GAT TGG TAT ATT
TAA TA) (ODN2). 32P-5′-end labeled ODN1 and ODN2 were
hybridized with their complementary strands (1 µM) in sodium
cacodylate buffer (10 mM, pH 7.0). The duplexes were incubated with
drugs (100 µM) at 37 °C for 30 min. ODNs were recovered by ethanol
precipitation and heated with piperidine (10% v/v) at 90 °C for 20
min. Recovered samples by ethanol precipitation were analyzed by
electrophoresis on 15% polyacrylamide gel containing 7 M urea. Lane
1, Maxam-Gilbert G+A reaction; lane 2, control without drug; lane
3, 7; lane 4, 6; lane 5, 5; lane 4, 4. Partial sequences are shown in the
left of the gel.
from an epimer of 8 at the carbon attached to an acetylenic
group. The two other diastereoisomers 5 and 7 possessing 13S
configuration were synthesized from L-ascorbic acid (Supporting
Information).
G Alkylation of the 5′XGY3′ Sequence by 4, 5, 6, and 7.
G alkylations of DNA by 4-7 were examined with duplexes
of 29-mer oligodeoxynucleotides d(TTATTGTTCGTTAGT-
TGGTATATTTAATA) (ODN1) and d(TTATGTTTGCTTGAT-
TGGTATATTTAATA) (ODN2) with their complementary
strands ODN1c and ODN2c. ODN1 and ODN2 have 5′XGT3′
and 5′TGY3′ sequences (underlined), respectively. Alkylated
G sites were detected by PAGE analysis after strand cleavage
induced by heating with piperidine. Efficient G alkylation was
observed only for 4 (Figure 3). Three other epoxides 5-7
showed very weak G alkylation reactivity under the identical
incubation conditions. The G alkylation efficiency of ODN1
and ODN2 by 4 was highly dependent on the flanking bases to
the G being alkylated. The G’s in TGGT, CGT, AGT, and TGA
sequences were efficiently alkylated, whereas only weak alkyl-
ation was observed for the TGC sequence. Alkylation of the
GG sequence by 4 was observed at both G sites.
Identification of G-Alkylation Products. To determine the
chemistry of G alkylation, the reaction of 4 with 8-mer self-
complementary duplex d(ATA CGT AT)2 (ACGT/ACGT)
possessing the 5′ACGT3′ sequence was monitored by HPLC.
All oligomers used for HPLC analysis of G alkylation were
specified by the sequence of the interior four bases for
identification (Table 1). Incubation of ACGT with 4 at 0 °C
produced two products 17 and 18 (Figure 4a). MALDI-TOF
MS of 17 shows the molecular weight of 2690.31 that is in
good agreement with the sum of the molecular weight of ACGT
and 4. A mixture of 17 and 18 was isolated by HPLC and heated
at 90 °C for 30 min (Figure 4b). While 18 was stable under
these conditions, 17 decomposed completely to yield two
products 19 and 20. Molecular weights of 18 and 19 separated
by HPLC were determined by MALDI-TOF MS as 2689.48
and 2274.04, respectively. High-resolution FAB-MS of 20
showed the molecular formula of C22H19N5O5 (MW ) 434.1458).
The presence of the 1,2-diol functionality in 20 was suggested
Table 1. ODNs Used for HPLC Analysis of G Alkylation with 4
ACGTa
TAGT
ACTA
ATGT
ACAT
TGCAa
TGAT
ATCA
TGGT
ACCA
d(ATA CGT AT)
d(ATT AGT AT)
d(ATA CTA AT)
d(ATA TGT AT)
d(ATA CAT AT)
d(TAT GCA TA)
d(ATA TGA TT)
d(AAT CAT AT)
d(TAT GGT ATA T)
d(ATA TAC CAT A)
a ACGT and TGCA are self-complementary oligomers.
by rapid disappearance of 20 on HPLC analysis when treated
with sodium periodate. On the basis of these results, 17 was
assigned as the G-N7 adduct, which undergoes thermally
induced depurination to produce adduct 20 and oligomer d(ATA
CφT AT) 19 containing an abasic site φ (Figure 5).21 Since
thermally stable DNA adduct 18 has the same molecular weight
as 17, it is presumably the adduct of the G 2-amino group. No
further structure determination was attempted due to the limited
amount of 18 available.
Large-scale preparation of 20 by incubation of 4 with herring
sperm DNA enables us to assign the structure by 1H NMR with
differential NOE spectra (Figure S1 in Supporting Information).
1
In the H NMR spectra of 20 in DMSO-d6, a distinct singlet
assignable as the C8-H of guanine was observed at 7.74 ppm.
Differential NOE spectra showed a strong NOE between the
singlet of C8-H and a signal at 4.10 ppm that was assigned as
one of two methylene hydrogens.
(28) Nakatani, K.; Okamoto, A.; Saito, I. Tetrahedron 1996, 52, 9427-
9436.

5698 J. Am. Chem. Soc., Vol. 123, No. 24, 2001
Nakatani et al.
Figure 6. Relative rates for the initial G alkylation of G-containing
oligomers in the presence of 4. A linear approximation was obtained
for all oligomers with good correlation coefficient (R2 > 0.97 for all
plots). Slopes for ACGT, TGAT, TAGT, and TGCA are 0.038, 0.016,
0.013, and 0.012, respectively.
Figure 4. HPLC profiles for the reaction of self-complementary duplex
ACGT/ACGT with 4. ACGT (125 µM) was incubated with 4 (125
µM) in a 10% acetonitrile of sodium cacodylate buffer (50 mM, pH
7.0) at 0 °C. dT (500 µM) was added as an internal standard. (a) A
HPLC profile obtained after incubation for 48 h. (b) A HPLC profile
obtained by heating the mixture of 17 and 18 at 90 °C for 30 min.
Figure 7. Time-course for the formation of 17.
Because 4 is able to alkylate both side G’s at the site of
intercalation, two precovalent complexes of ACGT/ACGT with
4 with intercalation into the 5′CG3′ or 5′GT3′ step are
conceivable as the precursor of 4-ACGT adduct 17. According
to the nearest neighbor exclusion principle, intercalation of 4
into the CG step in the palindromic 5′ACGT3′/5′ACGT3′
sequence would result in an alkylation of G in only one strand,
whereas G’s in both strands would be susceptible to alkylation
when 4 intercalates into the GT step.20 Time-course experiments
for the reaction of ACGT with 4 showed that the relative
amount of 17 reaches a plateau with the ratio of ca. 0.4 (Figure
7), indicating that the alkylation is stopped at 40% conversion
of ACGT.
HPLC analysis of the reaction of duplex TGGT/ACCA with
4 showed an efficient G alkylation of TGGT. However, the
rate of G alkylation at the TGGT sequence could not be
determined by the same procedure as for ACGT due to the
formation of two 4-TGGT N7 adducts at both 3′ and 5′ side
G’s. Accordingly, the rate of the G alkylation at the GG site
relative to that at the CG sequence was estimated by comparing
the rate of disappearance of starting oligomers TGGT and
ACGT (Figure 8). As clearly seen from the figure, TGGT was
consumed more rapidly than ACGT. While TGGT is consumed
by the alkylation with 4 at G-N7 and presumably at G-2-NH2,
it is likely that the rate for G-N7 alkylation at the TGGT
sequence is at least comparable to that at the CGT sequence.
Figure 5. Structures of products obtained by the reaction of ACGT
with 4.
Quantitative Analysis for G-Alkylation Efficiency. We next
examined the sequence dependency of the G alkylation using
duplexes of short oligomers. Oligonucleotide 8-mers of ACGT,
TAGT, ATGT, TGCA, and TGAT contain only one G residue
in 5′XGT3′ or 5′TGY3′ sequence, whereas 10-mer TGGT
contains 5′TGGT3′ sequence (Table 1). These duplexes were
incubated with 4 at 0 °C and an aliquot of the mixture was
analyzed by reverse-phase HPLC. Amounts of a starting
oligomer and 4 were determined using dT as an internal
standard. Adducts 4-ODN were isolated by HPLC and digested
to a nucleoside mixture with snake venom phosphodiesterase,
P1 nuclease, and calf intestine alkaline phosphatase. The quantity
of 4-ODN adduct was determined from the amount of dT that
is produced from 4-ODN adduct by enzymatic digestion.
Initial rates of G alkylation of oligomer duplexes ACGT/
ACGT, TAGT/ACTA, ATGT/ACAT, TGCA/TGCA, and
TGAT/ATCA were determined from plots of the amount of
4-ODN adduct versus the incubation period (Figure 6). A linear
approximation for the plot with good correlation coefficient (R2
) 0.97) was obtained at the beginning of G alkylation of ACGT,
AGT, TGT, and TGA sequences. However, the 4-TGCA adduct
was not detected by HPLC analysis under the incubation
conditions. This is quite consistent with the formation of an
only weak cleavage band at the TGC site in PAGE analysis
(cf. Figure 3b). Relative rates of G alkylation by 4 were 1.0 for
CGT (standard), 0.42 for TGA, 0.34 for AGT, and 0.32 for
TGT.
Discussions
Effects of Chiral Centers on G-Alkylation Efficiency. The
marked difference of the G alkylation reactivity among four
diastereoisomers 4-7 indicated that the efficient G alkylation
by 4 proceeds through a distinct precovalent complex. Molecular
modeling studies of the intercalated complex of 4 with the GG
step showed that (i) intercalation of the naphthalene ring parallel

Sequence-SelectiVe Drug Intercalation into DNA
J. Am. Chem. Soc., Vol. 123, No. 24, 2001 5699
contrast, only one conformation that has the dihedral angles of
90° and -160° for O1-C2-C11-C13 and C2-C11-C13-
C14, respectively, was found for 6 within ∆E of 1.6 kcal/mol.
It has been shown that juxtaposition of N7 of G and the epoxide
for in-line SN2 attack is very important for efficient G
alkylation.24,25,29-31 Conformational analysis of 4 indicated that
conformations suitable for 3′ and 5′ side G alkylation are only
0.3 and 0.5 kcal/mol above in energy the conformation at a
global energy minimum. In contrast, conformations of 6 suitable
for 3′ and 5′ side G alkylation are energetically less favorable
by 2.2 and 1.6 kcal/mol, respectively (Figure 10). These energy
differences required for conformational change of 4 and 6 reflect
the efficiency for the G alkylation of these two alkylating drugs.
Thus, the activation energy for G alkylation by 6 should be
larger than that for 4 because an additional energy is required
for the conformational change of 6.
Figure 8. Time-course for the disappearance of ACGT and TGGT
in the reaction with 4.
Molecular Basis of Sequence Selectivity for the G Alkyl-
ation. The order of G alkylation susceptibility obtained for 4
based on HPLC analysis shown in Figures 6 and 8 was
compared to those reported for 1,19 2, and 323 (Table 2). Despite
the difference of drug structure and epoxide reactivity to G
alkylation, the sequence selectivity obtained from 5′XGY3′ and
4 has a very similar tendency for the selectivity previously
reported for the 5′XG3′ sequence with aflatoxin B1 oxide 119
and the 5′GY3′ sequence with kapurimycin A3 2 and its aglycon
model 3.23 The GG sequence is the most preferred site for all
three drugs. The CG sequence is also preferable for 1 and 4,
whereas the GC sequence is the least preferable site for 2, 3,
and 4.
In principle, both precovalent (intercalation) and covalent (G
alkylation) binding steps are responsible for the G alkylation
selectivity for DNA-alkylating intercalators. While it is very
difficult to determine the rates of each step separately, we
measured the relative overall rate of G alkylation by 4 at the
given sequence as shown in Figures 6 and 8. To rationalize G
alkylation selectivity for 1, it has previously been proposed by
two groups that the precovalent binding step is far more
important for the overall G alkylation rate than the covalent
binding step.18,19 On the basis of high-level ab initio calculations
and molecular dynamic simulations, we have also reached a
similar conclusion, i.e., the efficiency of sequence-selective G
alkylation by 3 is primarily controlled by the sequence-selective
binding.24 These conclusions are also applicable to the sequence-
selective G alkylation by 4. Thus, the results shown in Figure
6 indicate that 4 intercalates much more favorably into the CG
step than the GT step. Weak preference for the TGA and AGT
sequences over the TGT sequence suggests that the susceptibility
of GA and AG steps for drug intercalation is slightly higher
than that for TG and GT steps. It is not clear why the TGC
sequence is inert toward G alkylation by 1 and 4, because these
drugs are able to intercalate into the TG step and alkylate 3′
side G.
Figure 9. Proposed precovalent complexes of 4 at the GG step suitable
for alkylation of (a) 3′ side and (b) 5′ side G’s obtained by molecular
modeling of the 8-mer duplex d(GAT GGT AC)/d(GTA CCA TC).
Only the GG step and 4 were shown in the figure for clarity. (c) Partial
illustration of two complexes with the Newman projection regarding a
C2-C11 bond. Selected atoms were labeled for clarity. Dihedral angles
of O1-C2-C11-C13 of 4 in the precovalent complex of parts a and
b are -150 and 90°, respectively. Dihedral angles of C2-C11-C13-
C14 of 4 for both complexes are about 90°.
Preferential intercalation into the GG and CG steps is not
dependent on the individual drug structures. This suggests an
important role of the intrinsic property of DNA sequences in
the sequence selectivity. We previously reported that a preco-
valent complex formed between the GG step and 3 is most
favorable among the complexes with other G-containing steps
due to the favorable enthalpy change accompanied by drug
intercalation.24 We have calculated the enthalpy changes for the
to the neighboring GC base pairs seems to be essential for G
alkylation, (ii) the optimal dihedral angle of C2-C11-C13-
C14 for the efficient SN2 ring opening of the epoxide would be
90° in the precovalent complex, and (iii) optimal dihedral angles
of O1-C2-C11-C13 for 3′ and 5′ side G alkylation are -150
and 90°, respectively (Figure 9). The results of the conforma-
tional analysis of 4 were shown in Figure 10. The dihedral angle
of C2-C11-C13-C14 for 4 is about 90° for all low-energy
conformations (∆E < 0.8 kcal/mol), whereas various values are
feasible for the dihedral angle of O1-C2-C11-C13. In
(29) Sun, D.; Hansen, M.; Hurley, L. H. J. Am. Chem. Soc. 1995, 117,
2430-2440.
(30) Hansen, M.; Yun, S.; Hurley, L. H. Chem. Biol. 1995, 2, 229-240.
(31) Hansen, M.; Hurley, L. H. J. Am. Chem. Soc. 1995, 117, 2421-
2429.

5700 J. Am. Chem. Soc., Vol. 123, No. 24, 2001
Nakatani et al.
Figure 10. Conformational analysis of (a) 4 and (b) 6 regarding C2-C11 and C11-C14 bonds. Dihedral angles of O1-C2-C11-C13 and
C2-C11-C13-C14 for each conformation are shown. The conformation at the lowest energy for 4 has -90 and 90° for dihedral angles of
O1-C2-C11-C13 and C2-C11-C13-C14, respectively.
ionization potential of stacked G.37-40 The base ionization
Table 2. Preferential Sequences for G Alkylation
potential is regarded as a marker for the easiness of the elec-
tronic reorganization that is necessary to accommodate sig-
nificant charge transfer in the transition state.41 Thus, the
reactivity of the base for electrophiles increases as the ioniza-
tion potential decreases. However, these arguments are not
applicable to the G alkylation by 1, 2, 3, and 4, because these
drugs intercalate into stacked base pair to form precovalent
complexes prior to the covalent bond formation. Therefore, in
the transition state of GG alkylation by 4, two G’s are no longer
stacked. Intercalation into the GG step would collapse the
stacking of the GG doublet to result in an increase in the
oxidation potential.
drug
preferential sequencea
1
2, 3
4
GGT > CGT . AGT ≈ TGT
TGG > TGA > TGT > TGC
TGGT ≈ CGT . TGA > AGT > TGT . TGC
a Alkylated G’s are underlined.
formation of a precovalent complex of 4 at the GC and CG
steps at the B3LYP/6-31G(d) level and found that the energy
difference for the formation of two complexes is only 0.1 kcal/
mol.32 These calculation results do not account for the enormous
difference in the overall G-alkylation rate between these two
G-containing sequences, GC and CG.
The binding of metal ions to DNA is also known to be
sequence selective.42-45 The order of the preference for Co(II)
binding determined by line broadening of G-H8 by 1H NMR is
GG > GA > GT . GC. Mn(II) binding to the G of the GC
sequence has not been detected.42,43 The sequence selectivity
has been partially rationalized by negative MEP.44 Recently,
we reported selective oxidation of DNA with Co(II) ion and
benzoyl peroxide at various G-containing sequences including
runs of G’s.37 The oxidation selectivity is well correlated with
the distribution of HOMO of DNA,38,39 but not with MEP,
implying that the oxidation is initiated by HOMO-controlled
binding of Co(II) ion to G-N7. On the basis of ab initio
calculation and Co(II)-mediated G oxidation, we demonstrated
the experimental HOMO mapping of duplex DNA.37 The
HOMO mapping method visualizes the susceptibility of G-
containing sequences to HOMO-LUMO interaction with DNA-
binding molecules.
The preference of the runs of guanines for G alkylation by
positively charged alkylating agents33 such as nitrogen mustards
and chloroethyltriazines has been rationalized in terms of the
molecular electrostatic potential (MEP).34 Negative MEP in-
creases at the runs of guanines with an increase in the number
of stacked G’s.34 The order of negative MEP is GG > GA >
AG ≈ TG > GT > CG > GC. Therefore, positively charged
drugs tend to bind more strongly to the more electronically
negative G in the runs of G’s.33 However, the G-alkylation
selectivity observed for 1, 2, 3, and 4 could not be rationalized
by MEP. Because these drugs are not positively charged, the
order of each base step ranked by the magnitude of negative
MEP does not fit with the order of susceptibility experimentally
observed for the G alkylation, especially with regard to the
marked difference in the reactivity of CG and GC steps. It has
been proposed that selective alkylation at runs of G with
methylating and ethylating agents via SN2 would be rationalized
by low barrier height for the transition state,35,36 due to the low
(36) Kim, H. S.; LeBreton, P. R. J. Am. Chem. Soc. 1996, 118, 3694-
3707.
(37) Saito, I.; Nakamura, T.; Nakatani, K. J. Am. Chem. Soc. 2000, 122,
3001-3006.
(32) Calculations were carried out according to the method we previously
described with the Gaussian 94 program:29 Frisch, M. J.; Trucks, G. W.;
Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman,
J. R.; Keith, T.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.;
Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.;
Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng,
C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E.
S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.;
Baker, J.; Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A.
Gaussian 94, Revision D.4.; Gaussian, Inc.: Pittsburgh, PA, 1995.
(33) Hartley, J. A. In Molecular Basis of Specificity in Nucleic Acid-
Drug Interactions; Pullman, B., Jortner, J., Eds.; Kluwer Academic
Publishers: Boston, 1990; pp 513-530.
(38) Sugiyama, H.; Saito, I. J. Am. Chem. Soc. 1996, 118, 7063-7068.
(39) Satio, I.; Nakamura T.; Nakatani, K.; Yoshioka, Y.; Yamaguchi,
K.; Sugiyama, H. J. Am. Chem. Soc. 1998, 120, 12686-12687.
(40) Prat, F.; Houk, K. N.; Foote, C. S. J. Am. Chem. Soc. 1998, 120,
845-846.
(41) Ford, G. P.; Scribner, J. D. Chem. Res. Toxicol. 1990, 3, 219-230.
(42) Froystein, N. A.; Davis, J. T.; Reid, B. R.; Sletten, E. Acta Chem.
Scand. 1993, 47, 649-657.
(43) Steinkipf, S.; Garoufis, A.; Nerdal, W.; Sletten, E. Acta Chem. Scand.
1995, 49, 495-502.
(34) Pullman, A.; Pullman, B. Q. ReV. Biophys. 1981, 14, 289-380.
(35) (a) Kim, N. S.; Zhu, Q.; LeBreton, P. R. J. Am. Chem. Soc. 1999,
121, 11516-11530. (b) Zhu, Q.; LeBreton, P. R. J. Am. Chem. Soc. 2000,
122, 12824-12834.
(44) Moldrheim, E.; Andersen, B.; Froystein, N. A.; Sletten, E. Inorg.
Chim. Acta 1998, 273, 41-46
(45) Montrel, M.; Chuprina, V. P.; Poltev, V. I.; Nerdal, W.; Sletten, E.
J. Biomol. Struct. Dyn. 1998, 16, 631-637.

Sequence-SelectiVe Drug Intercalation into DNA
J. Am. Chem. Soc., Vol. 123, No. 24, 2001 5701
measured with JEOL JNM R-400 (100 MHz) and Varian Mercury 400
(100 MHz) spectrometers. IR spectra were recorded on a JASCO FT-
IR-5M spectrophotometer. Mass spectra were recorded on a JEOL JMS
SX-102A and a JEOL JMS HX-110 for EI and FAB measurements,
and a Perseptive Elite spectrometer for MALDI-TOF measurement.
Trihydroxyacetophenone and ammonium citrate were used for the
matrix. Wakogel C-200 was used for silica gel chromatography.
Precoated TLC plates of Merck silica gel 60 F254 was used for
monitoring the reactions and for preparative TLC. All reagents and
solvents were used as received. HPLC analysis was performed on a
CHEMCOBOND 5-ODS-H column (4.6 × 150 mm) with a Gilson
Chromatography Model 305 using an UV detector Model 118 at 254
nm. Triethylammonium acetate (TEAA, 0.1 M) and acetonitrile were
used for eluent for HPLC analysis. All enzymes and DNA oligomers
used in these studies were from commercial sources. [γ-32P]ATP (6000
Ci/mmol) was obtained from Amersham.
Figure 11. Calculated HOMO energy levels of two consecutive base
pairs containing G in B-form structure at the B3LYP/6-31G(d) level.
The HOMO energy level decreased in the order GG, CG, GA, AG,
TG, GT, and GC.
The synthetic procedure and spectral data of 4-7 ae available in
the Supporting Information.
We have then calculated the HOMO energy level of all sets
of two base pairs containing G in the B-form structure at the
B3LYP/6-31G(d) level (Figure 11).32 The order of the HOMO
level of G-containing sequences is GG > CG > GA > AG >
TG > GT > GC, which is completely consistent with the order
of the susceptibility experimentally observed for DNA alkyl-
ation. These results showed that 4 could be used as a unique
molecular probe for raking the HOMO energy level of the
G-containing DNA sequence by the well-known chemistry of
DNA alkylation. Furthermore, the consistency of the HOMO
energy level with the G alkylation susceptibility underscores
the possibility that the intercalation of charge neutral inter-
calators is a HOMO-controlled process. These results further
confirmed the importance of HOMO-LUMO interaction in
DNA-drug interactions.
Alkylation of 32P-5′-End-Labeled ODN1 and ODN2 with 4, 5, 6,
and 7. Single-stranded 29-mer DNA oligomers 5′-d(TTA TTG TTC
GTT AGT TGG TAT ATT TAA TA)-3′ (ODN1) and 5′-d(TTA TGT
TTG CTT GAT TGG TAT ATT TAA TA)-3′ (ODN2) (500 pmol)
were separately 5′-end labeled by phosphorylation with [γ-32P]ATP (4
µL, 370 MBq/µL) and T4 polynucleotide kinase (4 µL, 10 units/µL)
using standard procedures. The 5′-end-labeled ODNs were recovered
by ethanol precipitation and further purified by electrophoresis on 15%
denatured polyacrylamide gel and isolated by the crush and soak
method. DNA alkylation experiments were carried out with a solution
of 100 µL total volume. Isolated ODN1 and ODN2 were incubated
with complementary strand ODN1c d(TAT TAA ATA TAC CAA CTA
ACG AAC AAT AA) and ODN2c d(TAT TAA ATA TAC CAA TCA
AGC AAA CAT AA) (1 µM), respectively, in 20 mM Tris-HCl buffer
(pH 7.6) at 90 °C for 5 min, and cooled slowly to ambient temperature
for duplex formation. A solution of 4, 5, 6, and 7 (10 µL, 1 mM in
CH3CN, a final concentration of drug of 100 µM) was added to the
DNA solution and the mixture was incubated at 37 °C for 30 min. The
reaction was quenched by adding calf thymus DNA, and ODNs were
recovered by ethanol precipitation, dissolved in 50 µL of 10% (v/v)
piperidine, and heated at 90 °C for 20 min. The resulting mixture was
concentrated in vacuo and resuspended in 10 µL of 80% formamide
loading buffer (80% formamide, 1 mM EDTA, 0.1% xylene cyanole,
and 0.1% bromophenol blue). The samples (1 µL) were loaded onto
15% polyacrylamide and 7 M urea sequence gel and electrophoresed
at 1900 V for ca. 2 h. The gel was exposed to X-ray film with
intensifying sheet at -70 °C and analyzed by densitometer.
Conclusion
The molecular basis for the sequence selectivity observed for
G alkylation by aflatoxin B1 oxide (1) and kapurimycin A3 (2)
has been a subject of intense study, because these molecules
lack the apparent DNA binding functionality to rationalize
G-alkylation selectivity. By using synthetically designed G-
alkylating drug 4, we have shown that (i) intercalative DNA
alkylating drugs have very similar sequence selectivity for G
alkylation and (ii) this selectivity is derived from the intrinsic
property of the G-containing DNA sequence. On the basis of
these observations and the calculated HOMO energy levels of
G-containing sequences, we propose that one of the most
important driving forces for the intercalation of charge neutral
intercalators is the interaction between the HOMO of DNA and
the LUMO of intercalators. DNA-binding molecules that possess
high sequence selectivity and covalently binding capability are
important targets as therapeutic agents.46-50 Our studies de-
scribed here shed light on the mechanism of the sequence-
selective intercalation of naturally occurring and synthetic drugs.
Determination of the Intercalation Site of 4 at the 5′-ACGT-3′
Sequence. A solution (100 µL) of 125 µM of d(ATA CGT AT) ACGT
and 500 µM of dT (internal standard) in 10% (v/v) CH3CN/50 mM Na
cacodylate buffer (pH 7.0) was incubated at 90 °C for 5 min and cooled
slowly to 0 °C for duplex formation. 4 (a final concentration of 125
µM) was added and the mixture was incubated at 0 °C. An aliquot (10
µL) of the reaction mixture was analyzed by HPLC after incubation
for 0, 3, 6, 12, 24, 48, and 72 h. Elution was performed with 0.1 M
TEAA buffer, 10-40% acetonitrile linear gradient, for 0-40 min at a
flow rate of 1.0 mL/min. Detection was carried out at 254 nm.
Formation of 4-ACGT adduct at G-N7 17 was confirmed by MALDI-
TOF MS (calcd 2691.92, found 2690.31). The mixture of 4-ACGT at
both N7 and amino groups was isolated by HPLC, lyophilized, and
resolved in water. The mixture was heated to induce depurination and
analyzed by HPLC. Formations of 4-ACGT adduct 18 at an amino
group and ODN containing abasic site 19 were confirmed by MALDI-
TOF MS. 18: calcd 2690.91, found 2689.48. 19: calcd 2275.51, found
2274.04.
Synthesis and Purification of 4-G Adduct 20. To a solution of
Herring Sperm DNA (20 mg) in 10% (v/v) CH3CN/50 mM Na
cacodylate buffer (20 mL, pH 7.0) was added 4 (2.8 mg, 0.01 mmol)
and the mixture was incubated at 37 °C for 48 h. The resulting mixture
was heated at 90 °C for 30 min. DNAs were precipitated with ethanol,
and the supernatant solution was separated and concentrated in vacuo.
Crude products were purified by HPLC to give 4-G adduct 20 (2.0
mg, 46%) as a white solid: 1H NMR (400 MHz, d6-DMSO) δ 8.46 (d,
Experimental Section
1H NMR spectra were measured with JEOL JNM R-400 (400 MHz)
or Varian Mercury 400 (400 MHz) spectrometers. The chemical shifts
are expressed in ppm downfield from tetramethylsilane, using residual
chloroform (δ ) 7.24) as an internal standard. 13C NMR spectra were
(46) Dervan, P. B.; Bulri, R. W. Curr. Opin. Chem. Biol. 1999, 3, 688-
693.
(47) Trauger, J. W.; Baird, E. E.; Dervan, P. B. Nature 1996, 382, 559-
561.
(48) Chang, A. Y.; Dervan, P. B. J. Am. Chem. Soc. 2000, 122, 4856-
4864.
(49) Zhi-Fu, T.; Fujiwara, T.; Saito, I.; Sugiyama, H. J. Am. Chem. Soc.
1999, 121, 4961-4967.
(50) Okamoto, A.; Nakamura, T.; Yoshida, K.; Nakatani, K.; Saito, I.
Org. Lett. 2000, 21, 3249-3251.

5702 J. Am. Chem. Soc., Vol. 123, No. 24, 2001
Nakatani et al.
1 H, J ) 7.9 Hz), 8.08 (d, 1 H, J ) 7.5 Hz), 7.95 (d, 1 H, J ) 8.6 Hz),
7.89 (d, 1 H, J ) 8.8 Hz), 7.81-7.73 (2 H), 7.74 (s, 1 H), 6.59 (s, 1
H), 5.97 (s, 1 H), 5.89 (br, 1 H), 5.44 (d, 1 H, J ) 7.3 Hz), 4.45 (dd,
1 H, J ) 2.4, 13.7 Hz), 4.28 (m, 1 H), 4.10 (dd, 1 H, J ) 9.7, 13.7
Hz), 1.65 (s, 3 H); MS (FAB) (NBA) m/z 434 [(M + H)+]; HRMS
calcd for C22H20N5O5 [(M + H)+] 434.1463, found 434.1458.
of water (distant-dependent dielectric electrostatic, cut off for van der
Waals, electrostatic, and hydrogen bonding were 7, 12, and 4 Å,
respectively). The epoxide side chain was then set to point the 5′ side
G by rotating the C2-C11 bond to obtain the energy minimized
structure suitable for alkylation of the 5′ side G. Two energy minimized
structures were virtually indistinguishable from each other beside the
conformation of the epoxide side chain. Only the GG step and 4 were
shown in Figure 4 for clarity.
HPLC Analysis of Sequence Selectivity of G Alkylation by 4.
Duplexes (62.5 µM) and an internal standard of dT (500 µM) in Na
cacodylate buffer (50 mM, pH 7.0) containing acetonitrile (10% v/v)
were incubated at 90 °C for 5 min and cooled slowly to 0 °C for duplex
formation. 4 (125 µM) was added to the DNA solution and the mixture
was incubated at 0 °C. An aliquot of the reaction mixture was analyzed
by HPLC after incubation for 0, 3, 6, 12, 24, 48, and 72 h. Elution was
performed with 0.1 M TEAA Buffer, 10-40% acetonitrile linear
gradient, for 0-40 min at a flow rate of 1.0 mL/min. Detection was
carried out at 254 nm. 4-ODN adducts were isolated by HPLC and
confirmed by MALDI-TOF MS. 4-TAGT: calcd 2706.93, found
2705.45. 4-ATGT: calcd 2706.93, found 2705.09. 4-TGAT: calcd
2706.93, found 2706.02. 4-TGGT: calcd 3340.33, found 3339.72.
Separately, 4-ODN adducts were isolated by HPLC, lyophilized,
resolved in water, and digested to nucleosides by sneak venom
phosphodiesterase, P1 nuclease, and alkaline phosphatase at 37 °C for
2 h. The quantity of 4-ODN adducts was determined by comparing
the amount of the resulting dT with that of authentic sample.
Procedure for Molecular Modeling Shown in Figure 9. Proposed
precovalent complexes of 4 at the GG step suitable for 3′ and 5′ side
G’s were obtained by molecular modeling of 8-mer duplex d(GAT GGT
AC)/d(GTA CCA TC) with 4. Manual docking of 4 to the duplex at
the GG step from the major groove side produced an initial structure
of 4-DNA complex for molecular mechanics calculations. The epoxide
side chain was set to point toward the 3′ side G of the GG step by
rotating the C2-C11 bond. Energy minimization of the complex was
carried out by using Amber* force field with GB/SA solvation treatment
HOMO Calculations. HOMO energies of stacked N-methylated
dinucleotide base pairs of GG, CG, GA, AG, TG, GT, and GC were
calculated as follows: Geometries of stacked N-methylated dinucleotide
bases at N1 (pyrimidine base) and N9 (purine base) were constructed
by using the biopolymer module in Insight II (98.0) program with helical
parameters of 3.38 Å for pitch, 36° for twist, and 1° for tilt. All the
sugar backbones were removed except for the deoxyribose C1′ carbon
and C1′ H and two H atoms were then attached to the C1′ methine to
complete the stacked N-methylated dinucleotide base pairs. With these
geometry, HOMO energies of stacked base pairs were calculated at
the B3LYP/6-31G(d) level by using the Gaussian 94 program.
Conformational Analysis for 4 and 6 Regarding C2-C11 and
C11-C13 Bonds. Forty-nine conformations generated by rotating both
C2-C11 and C11-C13 bonds with an increment of 72° were
minimized by energy with the PM3 method of semiempirical molecular
orbital calculations. Conformers within the energy difference (∆E) of
2.5 kcal/mol from the conformation at the global energy minimum were
plotted in Figure 10.
Supporting Information Available: Synthetic procedure of
4, 5, 6, and 7 and 1H NMR spectra of 20 (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA003956I