Studies on the mechanism of action of azinomycin B: Definition of regioselectivity and sequence selectivity of DNA cross-link formation and clarification of the role of the naphthoate

Article abstract of DOI:10.1021/ja025563k

Evaluation of the sequence selectivity, noncovalent association, and orientation of the DNA cross-linking agent azinomycin B on its duplex DNA receptor is described. A strong correlation between sequence nucleophilicity and cross-linking yield was observe

Full text of DOI:10.1021/ja025563k

Published on Web 10/09/2002
Studies on the Mechanism of Action of Azinomycin B:
Definition of Regioselectivity and Sequence Selectivity of
DNA Cross-Link Formation and Clarification of the Role of the
Naphthoate
Robert S. Coleman,* Ronelito J. Perez, Christopher H. Burk, and Antonio Navarro
Contribution from the Department of Chemistry, The Ohio State UniVersity,
1
00 West 18th AVenue, Columbus, Ohio 43210-1185
Received January 10, 2002
Abstract: Evaluation of the sequence selectivity, noncovalent association, and orientation of the DNA cross-
linking agent azinomycin B on its duplex DNA receptor is described. A strong correlation between sequence
nucleophilicity and cross-linking yield was observed, and steric effects due to the thymine C5-methyl group
were identified. Detailed studies on the role of the azinomycin naphthoate using viscometry, fluorescence
contact energy transfer, and DNA unwinding assays point to a nonintercalative binding mode for this group.
A kinetic assay for agent regioselectivity was used to determine the orientation of binding and covalent
cross-link formation.
1
. Introduction
dino[1,2-a]pyrrolidine ring system. They exhibit potent in vitro
cytotoxic activity and promising in vivo antitumor activity
against P388 leukemia in mice that is comparable to mitomycin
New prototypes of antitumor agents generate tremendous
scientific interest in the synthetic organic and medicinal
chemistry communities, and the DNA cross-linking agents
azinomycins A (1a) and B (1b) are typical in this regard.
4
C.
1
,2
Evaluation of the azinomycins has been hampered by
chemical instability and poor availability from natural sources.
Information on the biological activity of these agents is limited
4
to the original 1987 report. We recently reported the first total
5
synthesis of a member of this natural product family, and we
now disclose our studies that begin to clarify the biological
mechanism of action of these agents.
The azinomycins are among a small group of molecules that
interact with duplex DNA in the major groove. Azinomycin B
6
forms a covalent interstrand cross-link via the N7 positions of
suitably disposed purine bases in the duplex DNA sequence
5
′-d(PuNPy)-3′, via the electrophilic C10 and C21 carbons, but
the structural origin of the sequence selectivity and binding
affinity have yet to be defined. Early studies by Lown and
7
Majumdar provided the original demonstration that azinomycin
(
3) Nagaoka, K.; Matsumoto, M.; Oono, J.; Yokoi, K.; Ishizeki, S.; Nakashima,
T. J. Antibiot. 1986, 39, 1527. Yokoi, K.; Nagaoka, K.; Nakashima, T.
Chem. Pharm. Bull. 1986, 34, 4554.
These naturally occurring agents were isolated in 1986 from
(
4) Ishizeki, S.; Ohtsuka, M.; Irinoda, K.; Kukita, K.; Nagaoka, K.; Nakashima,
T. J. Antibiot. 1987, 40, 60. In vitro cytotoxicity: IC50 ) 0.07 µg/mL (1a)
and 0.11 µg/mL (1b) against L5178Y cells. In vivo antitumor activity:
3
cultures of Streptomyces griseofuscus S42227 and possess an
intricately functionalized structure containing the unusual aziri-
1
93% ILS at 16 µg/kg 1b (3/7 survivors) against P388 leukemia; 161%
ILS at 32 µg/kg 1b (5/8 survivors) against Erlich carcinoma. In the same
system, mitomycin C exhibited a 204% ILS at 1 mg/kg against P388
leukemia.
*
Address correspondence to this author. E-mail: coleman@chemistry.ohio-
state.edu
(
1) For references to synthetic work on the azinomycins, see: Coleman, R. S.
Synlett 1998, 1031. Coleman, R. S.; Kong, J.-S.; Richardson, T. E. J. Am.
Chem. Soc. 1999, 121, 9088. Hodgkinson, T. J.; Shipman, M. Tetrahedron
(5) Coleman, R. S.; Li, J.; Navarro, A. Angew. Chem., Int. Ed. 2001, 40, 1736.
(6) For a review on the synthesis of DNA cross-linking agents, and a discussion
of the historical development of DNA cross-linking agents, see: Coleman,
R. S. Curr. Opin. Drug DiscoV. DeV. 2001, 4, 435. For a comprehensive
review on DNA cross-linking agents, see: Rajski, S. R.; Williams, R. M.
Chem. ReV. 1998, 98, 2723. For a review of agents that covalently modify
DNA, see: Gates K. S. Covalent Modification of DNA by Natural Products.
In ComprehensiVe Natural Products Chemistry; Barton, D., Nakanishi, K.,
Meth-Cohn, O., Eds.; Pergamon Press/Elsevier Science: Oxford, U.K.,
1999; Vol. 7, p 491.
2
001, 57, 4467.
(
2) (a) Hartley, J. A.; Hazrati, A.; Kelland, L. R.; Khanim, R.; Shipman, M.;
Suzenet, F.; Walker, L. F. Angew. Chem., Int. Ed. 2000, 112, 3609. (b)
Hartley, J. A.; Hazrati, A.; Khanim, R. Hodgkinson, T. J.; Shipman, M.;
Suzenet, F.; Kelland, L. R. Chem. Commun. 2000, 2325. (c) Hodgkinson,
T. J.; Kelland, L. R. Shipman, M.; Suzenet, F. Bioorg. Med. Chem. Lett.
2
000, 10, 239.
1
3008 J. AM. CHEM. SOC. 2002, 124, 13008-13017
9
10.1021/ja025563k CCC: $22.00 © 2002 American Chemical Society

Mechanism of Action of Azinomycin B
A R T I C L E S
B (n e´ e carzinophilin)8 covalently cross-links native DNA
without prior activation. A more recent, preliminary study on
azinomycin/DNA interactions by Armstrong and co-workers
Benzyl (2S,3S)-3,4-Epoxy-2-(3-methoxy-5-methyl-1-naphthoyloxy)-
-methylbutanoate. A stirred solution of benzyl (2S,3S)-3,4-epoxy-
3
15
9
2-hydroxy-3-methylbutanoate (255.0 mg, 1.15 mmol), Et
3
N (0.25 mL,
.72 mmol), and DMAP (4-(N,N-dimethylamino)pyridine) (14.0 mg,
.11 mmol) in dry CH Cl (15 mL) at 0 °C was treated with a solution
1
0
demonstrated cross-link formation between the agent and N7
of dG and N7 of dG or dA within the major groove of DNA.
These results were confirmed recently by Saito and co-
workers,10 who proposed the orientational binding mode of
azinomycin B to its triplet base-pair receptor, which we
2
2
of 3-methoxy-5-methyl-1-naphthoyl chloride (320.0 mg, 1.38 mmol)
in CH Cl (10 mL). The reaction mixture was stirred at this temperature
2
2
for 3 h before being quenched by the addition of water (50 mL). The
organic layer was separated and the aqueous layer was extracted with
dichloromethane (3 × 25 mL). The combined organic extracts were
dried (Na SO ), filtered, and concentrated. Purification of the residue
11-13
confirmed in a series of computational studies,
established experimentally.
and now
2
4
In none of these studies was a clear role for the naphthoic
acid defined. Computational work was consistent with either
by flash chromatography (2 × 10 cm silica, 20% ethyl acetate/hexane)
1
afforded the desired compound (367 mg, 76%) as a colorless oil:
H
12,13
NMR (400 MHz, CDCl
3
) δ 8.60 (dd, J ) 6.2, 3.7 Hz, 1H), 7.90 (d, J
an intercalative or nonintercalative binding model.
Experi-
mental work on partial structures by Zang and Gates14 indicated
that azinomycin fragments bearing the naphthoate were weak
) 2.6 Hz, 1H), 7.51 (d, J ) 2.6 Hz, 1H), 7.38 (m, 2H), 7.35 (m, 5H),
5
3
(
.29 (d, J ) 12.0 Hz, 1H), 5.21 (d, J ) 12.0 Hz, 1H), 5.19 (s, 1H),
.91 (s, 3H), 2.95 (d, J ) 4.6 Hz, 1H), 2.65 (d, J ) 4.6 Hz, 1H), 2.60
3
-1
DNA binding agents (10 M ), and these workers proposed
an intercalative binding mode.
s, 3H), 1.42 (s, 3H); 13C NMR (100 MHz, CDCl
) δ 167.3, 166.0,
3
1
1
34.9, 133.9, 133.5, 131.2, 130.6, 128.4, 128.3, 128.2, 128.1, 127.8,
26.2, 125.5, 125.4, 124.3, 75.2, 67.2, 55.0, 51.7, 17.8; HRMS (ESI),
We now provide our initial results evaluating the covalent
and noncovalent interactions of both azinomycin B with duplex
DNA. Herein, we provide evidence that more clearly defines
the regioselectivity of the agent in the cross-linked lesion,
delineates the sequence selectivity of cross-link formation, and
demonstrates a lack of effective intercalation by the naphthoate
group of the agents.
m/z 443.1471 (calcd for C H O + Na: 443.1471).
25
24
6
2. Materials and Methods
Azinomycin B was obtained from fermentation broths of Strepto-
myces sp. (NRRL 15902) using growth and isolation conditions similar
to those originally published. Samples were stored in Et O or toluene
solution at -80 °C to prevent decomposition; samples degraded
significantly after several months, and best results were obtained with
freshly isolated samples.
2
(
2S,3S)-3,4-Epoxy-2-(3-methoxy-5-methyl-1-naphthoyloxy)-3-me-
thylbutanoic Acid. A solution of the above benzyl ester (368 mg, 0.88
mmol) in dry methanol (50 mL) was treated with 10% palladium on
carbon (54 mg) and the resulting suspension was stirred under hydrogen
Methyl 3-Methoxy-5-methyl-1-naphthoate (4). A solution of
diazomethane in ether was prepared by the addition of 1-methyl-3-
nitro-1-nitrosoguanidine (201.5 mg, 1.157 mmol) to a mixture of 40%
aqueous NaOH (20 mL) and ether (10 mL). A stirred solution of 2 (50
mg, 0.23 mmol) in THF (5 mL) at room temperature was treated with
the diazomethane solution. The reaction mixture was quenched by the
addition of 1 N aqueous HCl (2 mL), and the mixture was washed
with saturated aqueous NaCl (2 × 2 mL). The organic layer was dried
(
1 atm) for 2 h at room temperature. The reaction mixture was filtered
through a pad of Celite, and the filtrate was concentrated in vacuo to
afford the corresponding crude acid as a colorless solid, which was
used without further purification: 1H NMR (400 MHz, CDCl
) δ 8.61
dd, J ) 6.2, 3.7 Hz, 1H), 7.89 (d, J ) 2.5 Hz, 1H), 7.49 (d, J ) 2.5
Hz, 1H), 7.38 (m, 2H), 5.32 (s, 1H), 3.91 (s, 3H), 2.95 (d, J ) 4.6 Hz,
H), 2.65 (d, J ) 4.6 Hz, 1H), 2.67 (s, 3H), 1.42 (s, 3H).
3
(
1
2 4
(Na SO ), the solvent was removed, and the residue was purified by
flash chromatography (2 × 5 cm silica, 15% ethyl acetate/hexane) to
1
afford pure 4 (52.3 mg, 99%) as a crystalline solid: H NMR (500
MHz, CDCl
3
) δ 8.58 (dd, J ) 6.2, 3.7 Hz, 1H), 7.78 (d, J ) 2.6 Hz,
1
3
1
H), 7.42 (d, J ) 2.6 Hz, 1H), 7.32 (m, 2H), 4.00 (s, 3H), 3.94 (s,
1
3
H), 2.64 (s, 3H); C NMR (125 MHz, CDCl
33.5, 129.9, 128.0, 127.2, 125.3, 124.3, 121.8, 108.4, 56.0, 52.7, 20.5;
O, 1:1) λmax 240 (ꢀ 34 000), 298 (ꢀ 6300), 338 nm
ꢀ 7400); HRMS (ESI), m/z 253.0822 (calcd for C14 + Na:
53.0841).
3
) δ 168.3, 156.4, 134.8,
UV (MeOH/H
(
2
2
14 3
H O
(2S,3S)-3,4-Epoxy-2-(3-methoxy-5-methyl-1-naphthoyloxy)-3-me-
thylbutanamide (3). A solution of the above crude acid (43 mg, 0.13
mmol) in tetrahydrofuran (THF) (5 mL) at 0 °C was treated with
triethylamine (0.05 mL, 0.33 mmol) and ethyl chloroformate (31 mg,
0
.26 µmol). After 15 min at 0 °C, aqueous 30% ammonium hydroxide
0.05 mL, 0.35 mmol) was added, and the reaction mixture was stirred
for 30 min at 0 °C before being diluted with EtOAc and filtered through
(
(
10) Fujiwara, T.; Saito, I.; Sugiyama, H. Tetrahedron Lett. 1999, 40, 315.
(
7) Lown, J. W.; Majumdar, K. C. Can. J. Biochem. 1977, 55, 630.
(11) Alcaro, S.; Coleman, R. S. J. Org. Chem. 1998, 63, 4620.
(12) Alcaro, S.; Coleman. R. S. J. Med. Chem. 2000, 43, 2783.
(13) Alcaro, S.; Ortuso, F.; Coleman, R. S. J. Med. Chem. 2002, 45, 861.
(14) Zang, H.; Gates, K. S. Biochemistry 2000, 39, 14968.
(
8) Hata, T.; Koga, F.; Sano, Y.; Kanamori, K.; Matsumae, A.; Sugawara, R.;
Hoshi, T.; Shimi, T.; Ito, S.; Tomizawa, S. J. Antibiot., Ser. A. 1954, 7,
1
07.
(9) Armstrong, R. W.; Salvati, M. E.; Nguyen, M. J. Am. Chem. Soc. 1992,
14, 3144.
(15) Coleman, R. S.; Sarko, C. R.; Gittinger, J. P. Tetrahedron Lett. 1997, 38,
5917. Coleman, R. S.; McKinley, J. D. Tetrahedron Lett. 1998, 39, 3433.
1
J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002 13009

A R T I C L E S
Coleman et al.
a pad of silica gel. Evaporation of the solvent and purification of the
addition, the solution was mixed with a gentle stream of air. Each
mixture was equilibrated for 10 min at room temperature.
residue by flash chromatography (2 × 10 cm silica, 50% ethyl acetate/
1
hexane) afforded 3 (37.9 mg, 93%) as a colorless solid: H NMR (400
Fluorescence spectra were determined using a FluoroMax-2 fluo-
rimeter from ISA, Inc. Concentrations of DNA in BPE buffer were
calculated from A260 values. Control experiments without DNA were
performed by combining 2 µL of test compound in DMSO (dimethyl
sulfoxide) and 180 µL BPE buffer. DNA stock solution (2 µL) was
added, and the mixture was incubated 5 min at room temperature.
Topoisomerase I unwinding assays were performed by incubating a
mixture of 2 µL of Topo I buffer (10 mM Tris-HCl, 5% glycerol, 0.1
mM spermidine, 1 mM EDTA, 0.15 M NaCl, 1% BSA, pH 7.0), 0.5
ng pHOT1 supercoiled DNA, 12 units topoisomerase I, and double
distilled water (18 µL total volume) at 37 °C for 1 h. The relaxed
plasmid DNA topoisomers were treated with 2 µL of a DMSO solution
of the test compound, and the mixture was incubated at 37 °C for 30
min. The mixture was treated with 5 µL of 1% SDS and 4 µL of
proteinase K (50 µg/mL) solutions, and was incubated at 56 °C. After
MHz, CDCl
H), 7.45 (d, J ) 2.5 Hz, 1H), 7.32 (m, 2H), 6.18 (br s, 1H), 5.96 (br
s, 1H), 5.20 (s, 1H), 3.94 (s, 3H), 3.00 (d, J ) 4.6 Hz, 1H), 2.76 (d, J
3
) δ 8.60 (dd, J ) 6.2, 3.6 Hz, 1H), 7.88 (d, J ) 2.5 Hz,
1
13
)
4.6 Hz, 1H), 2.64 (s, 3H), 1.52 (s, 3H); C NMR (100 MHz, CDCl
3
)
δ 169.3, 166.0, 156.2, 134.9, 133.6, 128.5, 128.2, 127.3, 125.6, 124.2,
1
3
22.5, 108.8, 76.3, 62.9, 56.3, 53.7, 20.5, 18.0; HRMS (ESI), m/z
52.1150 (calcd for C18 + Na: 352.1161).
H
19NO
5
1
5-20 min, 2 µL of 10× loading dye (5% sarkosyl, 0.0025%
bromophenol blue, 25% glycerol) was added to the mixture, which
was mixed with an equal volume of 1:1 CHCl /i-AmOH. The aqueous
layer was loaded onto a 1% agarose gel (12 × 12.5 cm) cast in TAE
buffer (40 mM Tris, 1 mM EDTA, and 20 mM glacial acetic acid).
The gel was run at 120 V for 4.5 h, stained with ethidium bromide
(
2S,3R)-3,4-Dihydroxy-2-(3-methoxy-5-methyl-1-naphthoyloxy)-
-methylbutanamide (5). A solution of amide 3 (20 mg, 0.06 mmol)
O (7 mL) was treated with concentrated
3
3
in a 6:1 mixture of THF/H
aqueous perchloric acid (1 drop), and the reaction mixture was warmed
at reflux for 15 min. The mixture was diluted with EtOAc and washed
2
(0.5 µg/mL) for 2 h, and viewed with a Biorad Gel Doc 2000
with saturated aqueous solution of NaHCO
3
(5 mL) and saturated
SO ),
transilluminator. Two important control reactions were performed. The
test compound was added simultaneously with Topo I to verify that
the enzyme is active in the presence of the compound being tested,
and the test compound was added in the absence of the enzyme to
establish whether the compound damages DNA.
aqueous NaCl (3 × 3 mL). The organic layer was dried (Na
2
4
filtered, and concentrated. Purification of the residue by flash chro-
matography (2 × 5 silica, 50% ethyl acetate/hexane) afforded 5 (10
1
mg, 50%) as a colorless solid: H NMR (500 MHz, CDCl
3
) δ 8.66
(
dd, J ) 6.2, 3.6 Hz, 1H), 8.02 (d, J ) 2.5 Hz, 1H), 7.57 (d, J ) 2.5
Hz, 1H), 7.43 (m, 2H), 5.84 (s, 1H), 4.46 (d, J ) 9.9 Hz, 1H), 4.31 (d,
J ) 9.9 Hz, 1H), 4.03 (s, 3H), 2.73 (s, 3H), 1.59 (s, 3H); C NMR
100 MHz, CDCl ) δ 169.3, 166.0, 156.2, 134.9, 133.6, 128.5, 128.2,
27.3, 125.6, 124.2, 122.5, 108.8, 76.3, 62.9, 56.3, 53.7, 20.5, 18.0;
HRMS (ESI), m/z 370.1363 (calcd for C18 + Na: 370.1369).
DNA unwinding assays with T4 ligase were performed by lineariza-
tion of 1 µg of pBR322 supercoiled plasmid DNA with 3 units of
BamH1 at 37 °C for 1 h in a buffer system consisting of 150 mM
NaCl, 10 mM Tris-HCl, 10 mM MgCl , 1 mM dithiothreitol (DTT),
2
and 100 µg/mL bovine serum albumin (BSA), pH 7.9. The total reaction
volume was 50 µL. The DNA was extracted with an equal volume of
13
(
1
3
H21NO
6
3
1:1 CHCl /phenol, and the aqueous layer was separated and mixed with
three volumes of cold ethanol. The mixture was incubated in a dry
ice/acetone bath for 30 min, and the DNA was pelleted by centrifugation
at 10000 rpm for 20 min. The DNA was dried, dissolved in T4 ligase
buffer (50 mM Tris-HCl, 10 mM MgCl
2
, 10 mM DTT, 1 mM ATP,
25 µg/mL BSA, pH 7.5), and incubated with the test compound (added
in DMSO) for 10 min at room temperature. The total reaction volume
was 100 µL (10% DMSO). T4 Ligase (400 units) was added, and the
mixture was incubated at 12 °C for 2 h, and mixed with an equal volume
All polyacrylamide gel images were recorded using a Storm
phosphorimager from Molecular Dynamics. Both the vertical adjustable
slab gel kit (Model ASG-250) for polyacrylamide gel electrophoresis
and the agarose gel electrophoresis kit (Model SGE-014-02) were
purchased from CBS Scientific. The topoisomerase I DNA unwinding
kit was a gift from TopoGen, Inc. of Columbus, OH. The BamH1 and
T4 ligase enzymes for the T4 ligase unwinding experiments were from
New England Biolabs (NEB). Calf thymus DNA, sheared herring sperm
DNA, and supercoiled pBR322 were purchased from Sigma, Promega,
and NEB, respectively. Other oligodeoxynucleotides used were syn-
thesized using an Applied Biosystems 392 DNA/RNA synthesizer.
Relative viscosities were calculated using the equation η ) (t -
3
of 1:1 CHCl /phenol mixture to extract the test compound. The aqueous
layer was collected and mixed with three volumes of cold ethanol to
precipitate the DNA. The mixture was incubated in a dry ice/acetone
bath for 30 min, and the DNA was pelleted by centrifugation at 10000
rpm for 20 min. The dried pellet was dissolved in 3 µL loading dye
(5% sarkosyl, 0.0025% bromophenol blue, 25% glycerol) and loaded
onto a 1% agarose gel cast in TAE buffer. The gel was run at 120 V
for 4.5 h and was stained with ethidium bromide (0.5 µg/mL) for 2 h.
Control experiments similar to those used in Topo I assays were
performed to eliminate concerns analogous to those with the topois-
merase I assay.
Cross-linking reactions were performed with 15-mer oligodeoxy-
32
t
0
)/t
the buffer, using a Cannon-Manning semimicro (no. 50) viscometer
total volume ) 440 µL) immersed in a room temperature water bath.
Flow times ranged from 268 s for buffer to 1123 s for ethidium bromide.
0
, where t ) flow time of the DNA solution and t
0
) flow time of
nucleotides. One strand of the duplex was 5′- P end-labeled and an
4
aliquot with activity of 10 cpm was added to a solution containing
(
0.1 nmol of the unlabeled strand, 0.1 nmol of the complement, 10 µL
of 10 mM Tris-HCl (pH 7.5), and 1 µL of 50 mM NaCl. The resulting
mixture was dried and redissolved in 10 µL double distilled water to
make 10 µM DNA duplex solution in 10 mM Tris-HCl buffer
containing 5 mM NaCl. The DNA solution was annealed and cooled
to 8 °C. In a separate microcentrifuge tube, the desired volume of a 2
mM toluene stock solution of azinomycin B was evaporated. The DNA
solution was added to the azinomycin B, which was incubated at 8 °C
1/3
Values of (η/η
were plotted against the binding ratio r. Approximately 0.8 mM agent
per base-pair DNA solution in BPE buffer (6 mM Na HPO , 2 mM
NaH PO , 1 mM Na EDTA (EDTA ) ethylenediaminetetraacetic acid),
0
) , where η
0
) relative viscosity of blank DNA solution,
2
4
2
4
2
pH 7.0) was used. DNA solutions were titrated with the test compounds
by adding small volumes of concentrated aqueous solutions. After each
13010 J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002

Mechanism of Action of Azinomycin B
A R T I C L E S
Table 1. Sequence Selectivity of Azinomycin Ba
triplet sequence
cross-link (%)
mono (%)
starting DNA (%)
1
2
3
4
5
6
5′-GCC-3′
77
52
18
30
24
24
3
4
0
0
2
5
48
12
23
37
16
19
0
18
0
3
′-CGG-5′
5′-GTC-3′
′-CAG-5′
5′-GCT-3′
′-CGA-5′
5′-GTT-3′
′-CAA-5′
5′-ATC-3′
′-TAG-5′
5′-ACC-3′
′-TGG-5′
70
46
39
28
73
96
98
77
91
0
3
3
3
0
21
7
3
3
5
95
a
Reaction conditions: 10 mM Tris-HCl (pH 7.5), 5 mM NaCl, 10 µL
reaction volume, 20 h, 8 °C, 10 µM duplex DNA, and 400 µM azinomycin
B. Numbers opposite sequences refer to yields when that strand of the duplex
DNA was 5′ end-labeled with 32P.
Figure 1. DNA cross-linking by azinomycin B (1b). Reaction conditions:
1
0 mM Tris-HCl (pH 7.5), 5 mM NaCl, 10 µL volume, 8 °C, 20 h, 1 µM
duplex DNA, 120 µM 1b. Sequence used: 5′-d(TATTATGCCTATTAT)-
′ (A)/3′-d(ATAATACGGATAATA)-5′ (B). Lanes 1-3, 5′ end-labeled
Maxam-Gilbert sequencing to be the italicized dG bases (G-
lane data not shown). When the 3′-CGG-5′ strand was 5′ end-
labeled and the duplex was reacted with azinomycin B (lanes
3
strand is B. Lanes 4-6, 5′ end-labeled strand is A. Lane 1, control; lane 2,
DNA + 1b; lane 3, DNA + 1b and then piperidine; lane 4, control; lane
1-3), two electrophoretically distinct species were formed (lane
5
, DNA + 1b; lane 6, DNA + 1b and then piperidine.
2
) that were of lower mobility than the starting oligonucle-
for 24 h and then was treated with 11 µL of 20 mM NaOH solution at
7 °C for 45 min to form the formamidinopyrimidine (FAPY) base in
otide: (1) a band directly above the starting DNA that
corresponded to monoalkylation by the drug; (2) a band
corresponding to a cross-linked duplex at much lower mobility.
Lane 3 is the fragmented oligomer resulting from piperidine
treatment of the cross-linked duplex and demonstrates that the
DNA is site-specifically alkylatedsat a single guaninesboth
in the monoalkylated intermediate and in the cross-linked
product. When the complementary 5′-GCC-3′ strand was end-
labeled, only a trace of monoalkylated species was formed, and
the cross-linked duplex was by far the major product of the
reaction. This study served to define the strand that reacted in
the initial monoalkylation reaction (the 3′-CGG-5′ strand) and
provided evidence on which strand underwent subsequent cross-
link formation (the 5′-GCC-3′ strand) in addition to providing
clear definition of which bases in the DNA are alkylated.
In a full investigation of the sequence selectivity of covalent
cross-linking by azinomycin B (1b), we used DNA 15-mer
duplexes with the general sequence shown (Py ) pyrimidine;
Pu ) purine). The triplet binding sequence contained 5′-disposed
reactive purines (Pu) flanked by a nonnucleophilic T-residue.
Cross-linking experiments were performed with each DNA
3
order to minimize spontaneous depurination. The sample was desalted
using a Sephadex G-25 column and evaporated. The residue was
dissolved in 3 µL loading dye and loaded onto a 20% polyacrylamide
gel (28 × 16 cm). The gel was run at 20 W for 70 min, and cross-
linking was quantified by phosphorimagery.
Kinetic experiments were performed using a duplex consisting of
3
2
different sized strands, with both strands 5′- P end-labeled. Labeled
4
oligodeoxynucleotides (10 cpm each strand) were combined with 10
pmol each of the unlabeled strands, 10 µL of 10 mM Tris-HCl (pH
7
.5), and 1 µL of 50 mM NaCl. The mixture was dried in a speed
vacuum and dissolved in 10 µL of double distilled water to reconstitute
a 1 µM solution of the duplex DNA in 10 mM Tris-HCl buffer and 5
mM NaCl. The DNA was annealed and reacted with dried azinomycin
B prepared as above. The reaction was allowed to proceed at 8 °C for
the specified time and was quenched by the addition of 5 µL of 10
mM NaSEt. Parallel reactions requiring different reaction times were
run simultaneously. At the termination of all the reactions, they were
each desalted using a Sephadex G-25 column and dried. Each sample
was dissolved in 3 µL loading dye and loaded onto a 20% polyacryl-
amide gel (28 × 16 cm). The gel was run at 20 W for 70 min and
viewed using a phosphorimager.
3
2
strand alternately end-labeled with P phosphate.
3. Results and Discussion
5
′-d(TATTATPuPyPyATTATT)-3′
Sequence Selectivity of Cross-Link Formation. We began
our evaluation of mechanism of action of azinomycin B with a
study of the sequence dependence of cross-link formation, using
the recognition sequence 5′-d(GCC)-3′. Initial interstrand DNA
cross-link assays were performed using the following 15-mer
duplex containing a single binding site embedded within an
unreactive A-T sequence.
3
′-d(ATAATAPyPuPuTAATAA)-5′
Results from these studies are presented in Table 1, and the
yields for the various species formed in the reaction of
azinomycin B with the targeted DNA were determined by
densitometry of denaturing polyacrylamide gels using a phos-
phorimager. Analysis of the sequence selectivity data was
complicated by a competing, facile, and sequence-dependent
depurination of the cross-linked duplex, and the column labeled
5′-d(TATTATGCCTATTAT)-3′
3′-d(ATAATACGGATAATA)-5′
“
mono” contains the combined percentage of monoalkylated
In our preliminary evaluation using this d(GCC‚CGG) binding
triplet, we found that azinomycin B afforded exceptionally high
levels of formation of the cross-linked product, oftentimes in
excess of 90% as measured by densitometry (Figure 1). The
site of alkylation of the end-labeled strand was confirmed by
and depurinated oligomer. In all cases, Maxim-Gilbert se-
quencing confirmed the site of alkylation.
Cross-link formation with azinomycin B was optimal between
two guanine bases when the intervening base pair was G‚C
(Table 1, entry 1). With the d(GGC‚CCG) sequence, yields for
J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002 13011

A R T I C L E S
Coleman et al.
cross-link formation were reproducibly 75-80%, and oftentimes
in excess of 90%. The lower level of cross-link formation
observed and the accompanying increase in monoalklation when
the 3′-CGG-5′ strand was 5′ end-labeled are due to selective
depurination of the guanine in the 5′-GCC-3′ containing strand.
The 5′-GCC-3′ containing strand underwent essentially no
monoalkylation compared to the more nucleophilic 3′-CGG-5′
strand, as expected from a consideration of sequence-dependent
1
6
nucleophilicity.
Cross-linking yields dropped dramatically when the interven-
ing base pair between the reactive guanines was A‚T (Table 1,
entry 2). This result is expected based on the higher degree
nucleophilicity of the 5′-dG in GG sequences compared to AG
and because of steric hindrance of cross-link formation by the
C5-methyl group of the thymine that is 5′ to one of the reactive
guanines.17 In this case, there is still a greater degree of
monoalkylation of the more nucleophilic 3′-CAG-5′ strand than
of the 5′-GTC-3′ strand, but the overall level of cross-link
formation is reduced.
Figure 2. Plot of relative specific viscosity (η/η0)1/3 of calf thymus DNA
versus binding ratio (per base pair) for the indicated compounds. Ap-
proximately 0.8 mM agent per base-pair DNA solution in BPE buffer (6
mM Na2HPO4, 2 mM NaH2PO4, 1 mM Na2EDTA, pH 7.0) was used.
ments designed to define the intercalative ability of the
azinomycins, we have examined the interaction of the natural
Cross-linking between G and A was modest to minimal within
a variety of sequence contexts (Table 1, entries 3-6), contrasting
previous descriptions of some of these sequences being ef-
fectively cross-linked by azinomycin B. With the d(GCT‚CGA)
sequence that has been reported recently by Saito and co-
1
9
agent 1b and the partial structures 2, 3, 4, and 5 with duplex
DNA.
1
0
9
workers and earlier by Armstrong and co-workers to be
efficiently cross-linked by azinomycin B, the cross-linking yield
was 24% with either strand end-labeled (Table 1, entry 3), which
is comparable to that observed by Saito and co-workers (32%).
In this entry, for the CGA-containing strand, yields do not sum
to 100% because a measurably significant amount of radioactiv-
ity was nonspecifically localized in the gel. When the interven-
ing base pair is changed to T‚A (Table 1, entry 4), cross-link
formation is essentially abolished, and only minimal levels of
monoalkylation of the guanine in the 5′-GTT-3′ strand was
1
7
observed.
With sequences containing the more reactive 3′-TGG-5′ and
′-TAG-5′ sequences (Table 1, entries 5 and 6), only monoalky-
3
lation of the reactive guanine bases was observed with little or
no cross-link formation, and yields were proportional to guanine
nucleophilicity.
Viscometry is the classic assay for DNA intercalation. Under
the assumption that duplex DNA is a rigid rodlike structure,
hydrodynamic properties such as viscosity are highly dependent
upon changes in length. In the classic intercalation model
where ligands lengthen the DNA helix, viscometry provides a
sensitive way of determining the intercalation between the
stacked base pairs of DNA. We determined the effect of
incremental levels of azinomycin B (1b) and naphthoic acid
1
6
Making a comparison of our results (Table 1) with those of
9
an earlier study by Armstrong and co-workers is difficult
2
0
because the earlier data are qualitative. Nevertheless, the two
sequences that underwent cross-linking most effectively in our
work (GCC‚CGG and GCT‚CGA) were the same as the best
sequences from the Armstrong study. While we observed
significant differences in the yield of cross-link formation
between these two sequences where Armstrong and co-workers
did not, there are subtle flanking sequence differences between
the oligodeoxynucleotide sequences used in the two studies
(
2
2) on the viscosity of solutions of calf thymus DNA (Figure
).
Adding incremental levels of azinomycin B to a solution of
calf thymus DNA failed to measurably affect the viscosity of
the solution. This is in marked contrast to the pronounced
increase in the viscosity when a solution calf thymus DNA was
(there is an A‚T transversion adjacent to the purine of the
PyPuPu-containing strand).
Role of the Naphthoate. Given the structural similarity
2
1
titrated with the intercalator ethidium bromide (Figure 2).
These results contradict the report by Gates and co-workers14
in which they demonstrated an increasing trend in the relative
viscosity of a solution of sheared herring sperm DNA when
1
8
between the azinomycin and neocarzinostatin naphthoate, we
had antiticipated that the drugs would share a common,
intercalative mode of association with duplex DNA. In experi-
(
16) Sugiyama, H.; Saito, I. J. Am. Chem. Soc. 1996, 118, 7063.
17) For a similar report of steric hindrance of cross-link formation due to the
C5-methyl group of thymine, see: Coleman, R. S.; Pires, R. M. Nucleic
Acids Res. 1997, 25, 4771.
(
(19) Shishido, K.; Omodani, T.; Shibuya, M. J. Chem. Soc., Perkin Trans. 1
1992, 2053.
(20) Cohen, G.; Eisenberg, H. K. Biopolymers 1969, 8, 45.
(21) For essentially identical data comparing ethidium bromide with the minor
groove binder Hoechst 33258, see Figure 5 in ref 23.
(
18) Gao, X.; Stassinopoulos, A.; Gu, J.; Goldberg, I. H. Bioorg. Med. Chem.
1
995, 3, 795.
13012 J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002

Mechanism of Action of Azinomycin B
A R T I C L E S
Figure 3. Plot of relative specific viscosity (η/η0)1/3 of sheared herring
sperm DNA versus binding ratio (per base pair) for the indicated compounds.
Approximately 0.8 mM agent per base-pair DNA solution in BPE buffer
(6 mM Na2HPO4, 2 mM NaH2PO4, 1 mM Na2EDTA, pH 7.0) was used.
Figure 4. Excitation spectra of opened epoxide 5 and ethidium bromide
in the absence and presence of sheared herring sperm DNA in BPE buffer
(
6 mM Na2HPO4, 2 mM NaH2PO4, 1 mM Na2EDTA, pH 7.0). Emission
treated with epoxyamide 3, a compound that interacts with DNA
covalently.
was measured (y-axis, arbitrary units) at 636 nm for ethidium bromide and
435 nm for 5 as a function of excitation wavelength (x-axis).
Because shorter strands of DNA are generally more sensitive
to length changes resulting from intercalation, we employed
sheared herring sperm DNA (average length of 300-1000 bp)
in a second set of viscosity experiments. The model compound
known binding site 5′-d(GGC)-3′.
5′-d(ATAATAGGCATAATA)-3′
5
was investigated (Figure 3). This compound failed to show
3
′-d(TATTATCCGTATTAT)-5′
an increase in the relative viscosity of the DNA sample,
indicating a lack of intercalative binding. These results contradict
the report by Zang and Gates14 in which they demonstrated an
increasing trend in the relative viscosity of a solution of sheared
herring sperm DNA when treated with a compound analogous
to 5. However, there are significant differences in the experi-
mental designs of the two studies, and a comparative interpreta-
tion is made even more difficult because of the extremely low
dynamic range of their viscometry measurements and their lack
of either positive or negative controls.
In the excitation spectra shown in Figure 4 with ethidium
25
bromide (10 µM)-a well-characterized DNA intercalator with
4
5
-1
binding constants between 10 -10 M -in the presence of
sheared herring sperm DNA, there is a 6-fold enhancement of
ethidium fluorescence measured at 636 nm when a solution of
DNA (50 µM base pairs) is added. Conversely, there is a
decrease in naphthalene fluorescence measured at 435 nm when
DNA (50 µM base pairs) is added to a solution of opened
epoxide 5 (10 µM). This reduction in fluorescence quantum
yield is due to screening of the fluorophore by the DNA
absorbance. Essentially identical excitation spectra were obtained
with an excess of agent (10 µM 5 and 2 µM DNA base pairs;
data not shown). Similar results were obtained for naphthoic
acid 2 and methyl naphthoate 4, using sheared herring sperm
and calf thymus DNA. In each case, either there was no change
or a decrease was observed in the measured fluorescence
intensity when DNA was added to a solution of the agent (data
not shown).
Fluorescence contact energy transfer22 from DNA bases to a
bound ligand is the most definitive method to confirm intercala-
tion in duplex DNA.23 Light absorbed by the DNA bases is
transferred to a bound fluorophore under specific geometric and
spectroscopic conditions: (1) there must be spectral overlap
between the donor and acceptor systems (i.e., they both must
absorb in the same region of the spectrum); (2) the donor-
acceptor distance must be appropriate, as energy transfer exhibits
-
6
24
a distance dependence of r (4-7 Å in DNA); (3) the donor
and acceptor dipoles must be suitably aligned for efficient energy
transfer. For a molecule bound to duplex DNA, these three
conditions are met simultaneously only when the acceptor is
bound intercalatively. In the case of drugs bound along the
grooves or on the surface of the DNA duplex, the larger drug-
to-DNA distance and the poor orbital overlap prohibits energy
transfer. Experimentally, an increase in fluorescence quantum
yield of an intercalated molecule is observed when the DNA is
irradiated.
In the presence and absence of the 15-mer duplex shown
above containing the known recognition sequence 5′-GCC-3′,
an identical result was obtained (Figure 5). Using different agent/
base pair ratios (1:3, 1:1, and 5:1) to determine the behavior of
the fluorescence excitation spectra when an excess amount of
DNA was added to the compound and vice versa, the fluores-
cence intensities for both emission (data not shown) and
excitation spectra of azinomycin B were insignificantly altered
upon the addition of DNA. Identical results were obtained using
calf thymus DNA (data not shown). With each DNA system,
hν f DNA f agent f hν
(donor) (acceptor)
(
22) LePeck, J. B.; Paoletti, C. J. Mol. Biol. 1967, 27, 87.
23) Suh, D.; Chaires, J. B. Bioorg. Med. Chem. 1995, 3, 723.
(24) F o¨ rster, T. Ann. Phys. 1948, 2, 55.
25) Wilson, W. D.; Krishnamoorthy, C. R.; Wang, Y.-H.; Smithy, J. C.
Biopolymers 1985, 24, 1941. Bresloff, J. L.; Crothers, D. M. Biochemistry
1981, 20, 3547.
(
In the present studies, three different nucleic acid systems
were examined: calf thymus DNA, sheared herring sperm DNA
(
(300-1000 bp), and the 15-mer oligomer shown containing the
J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002 13013

A R T I C L E S
Coleman et al.
the DNA duplex in g75% yield, thereby localizing the drug
irreversibly onto its DNA receptor. The gel electrophoresis
autoradiogram insert (Figure 6) shows the conversion of native
DNA to alkylated DNA at the end of this incubation period.
In the interpretation of negative results such as those obtained
in the fluorescence contact energy transfer experiments, it is
reasonable to ask whether the experimental design was ap-
propriate for detecting potentially weak binding. There is no
information on the threshold of agent binding for this experiment
to yield positive results, but in a recent paper Viola and co-
workers observed a strong positive enhancement of agent
fluorescence in the presence of DNA using a furocoumarin with
4
a Kapp ) 3.3 × 10 . The DNA/agent ratio in the Viola work
2
8
was 10:1. In the results shown in Figure 5, the DNA/agent
ratio was 0.067, and the agent is covalently bound to the DNA
duplex. In Figure 6, gel electrophoresis was used to confirm
that the agent was covalently bound to the duplex, making
discussions of agent/DNA ratios and binding constants irrelevant
in this case.
Figure 5. Excitation spectra of azinomycin B (1b; 40 µM) in the absence
and presence of 15-mer duplex DNA (40 µM in base pairs) in BPE buffer
(
6 mM Na2HPO4, 2 mM NaH2PO4, 1 mM Na2EDTA, pH 7.0). Emission
was measured (y-axis, arbitrary units) at 435 nm as a function of excitation
wavelength (x-axis). For clarity, the positive ethidium bromide control is
not shown.
In three separate systems (1b, which forms covalent cross-
links; opened epoxide 5, which can only interact with DNA
noncovalently; and epoxyamide 3, which has been demonstrated
to effectively alkylate both single and double-stranded DNA),27
using three different forms of duplex DNA (sheared herring
sperm; calf thymus; and a synthetic 15-mer), we observed no
enhancement of agent fluorescence upon the addition of DNA
under conditions where ethidium bromide exhibited a strong,
positive enhancement of fluorescence.
To confirm the negative results obtained with viscometry and
fluorescence spectroscopy, we examined two enzyme-based
assays for intercalation. Topoisomerase I unwinds supercoiled
DNA. The topoisomerase assay for intercalation is based on
the fact that intercalating agents unwind the DNA double helix,
and with relaxed plasmid DNA as a substrate, the unwinding
caused by a bound intercalating agent introduces positive
superhelical twists in the DNA structure. Hence, the plasmid
behaves as if it were positively supercoiled. Treatment of the
intercalator-DNA complex with topoisomerase I (topo I) repairs
the extra superhelical twists in the DNA structure, resulting in
an apparently relaxed plasmid that in reality is negatively
supercoiled because of the presence of the intercalator. Extrac-
tion of the intercalator from the helix causes the formerly relaxed
DNA to negatively supercoil. Intercalating agents thus promote
the conversion of relaxed plasmid DNA to the negatively
supercoiled form in the presence of topo I. This is manifested
in the electrophoretic analysis as a shifting of the relaxed
topoisomeric ladder to the more highly mobile, negatively
supercoiled form of the plasmid. The known intercalating agent
4′-(acridin-9-ylamino)-3′-methoxymethanesulfonalinide (m-
AMSA) is used as a positive control.
Figure 6. Excitation spectra of epoxyamide 3 (100 µΜ) in the absence
and presence of 15-mer duplex DNA (5 µM) in BPE buffer (6 mM Na2-
HPO4, 2 mM NaH2PO4, 1 mM Na2EDTA, pH 7.0). The drug + DNA
spectrum was obtained after 20 h incubation of 3 and the duplex at 8 °C.
Emission was measured (y-axis, arbitrary units) at 435 nm as a function of
excitation wavelength (x-axis). Denaturing polyacrylamide gel electrophore-
32
32
sis insert: lane 1, 5′ P end-labeled DNA; lane 2:, 5′ P end-labeled DNA
drug after incubation.
+
ethidium bromide was used as a positive control, and a large
enhancement of fluorescence intensity was observed upon the
2
+
addition of duplex DNA (cf., Figure 4). The Barton Ru(phen)3
(
26
phen ) 1,10-phenanthroline) major groove binder was used
as a negative control (data not shown).
Epoxyamide partial structure 3 is highly effective at covalent
modification of duplex DNA, alkylating guanine bases in the
GCC‚CGG receptor with yields as high as 86%.27 The excitation
spectrum of 3 (Figure 6) showed no enhancement upon the
addition of 15-mer duplex DNA; the observed decrease in
fluorescence intensity is due to the DNA absorbance shielding
the naphthoate absorbance. In this experiment, the drug was
incubated with the duplex DNA for 20 h at 8 °C. Polyacrylamide
gel electrophoresis confirmed that epoxyamide 3 had alkylated
(
26) For a recent review, see: Kane-Maguire, N. A. P.; Wheeler J. F. Coord.
Chem. ReV. 2001, 211, 145, and references therein.
(
27) Coleman, R. S.; Burk, C. H.; Navarro, A.; Brueggemeier, R. W.; Diaz-
Cruz, E. S. Org. Lett. 2002, 4, 3545.
13014 J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002

Mechanism of Action of Azinomycin B
A R T I C L E S
Figure 7. DNA unwinding experiment with naphthoic acid 2 and topo I.
Lanes 1-3, 1000, 500, 100 µM 2, respectively; lane 4, 100 µM m-AMSA;
lane 5, relaxed plasmid DNA; lane 6, supercoiled plasmid DNA.
Figure 9. DNA unwinding experiment with 2 and T4 ligase. Lanes 1-4,
1000, 100, 50, and 10 µM 2, respectively; lane 5, linearized DNA + T4
ligase; lane 6, linearized DNA.
Figure 10. DNA unwinding experiment with 4 and T4 ligase. Lanes 1-4,
Figure 8. DNA unwinding experiment with methyl naphthoate 4 and topo
I. Lane 1, control (supercoiled plasmid DNA + 1000 µM 4 without topo
I); lane 2, 1000 µM 4 and topo I added simultaneously; lanes 3-5, 1000,
1
000, 100, 50, and 10 µM 4, respectively; lane 5, linearized DNA + T4
ligase; lane 6, linearized DNA.
5
00, 100 µM 4, respectively, added to plasmid DNA preincubated with
topo I; lane 6, 100 µM m-AMSA added to plasmid DNA preincubated with
topo I; lane 7, relaxed plasmid DNA; lane 8, supercoiled plasmid DNA.
Unwinding assay results obtained with compounds 2 and 4
are shown in Figures 7 and 8, respectively. Lanes 5 and 7
(Figures 7 and 8, respectively) show the behavior of relaxed
plasmid DNA in 1% agarose gel. The ladder of bands represents
the topoisomeric forms (topoisomers) of the plasmid. The
distribution of the topoisomers shifts when an intercalating agent
is added, as was observed with the intercalator m-AMSA (lanes
Figure 11. DNA unwinding experiment with 5 and T4 ligase. Lane 1,
linearized DNA + 1000 µM 5; lanes 2, 1000 µM 5 and T4 ligase added
simultaneously to linearized DNA; lane 3, 100 µM 5 and T4 ligase added
simultaneously to linearized DNA; lanes 4-7, 1000, 100, 50, and 10 µM
5, respectively; lane 8, linearized DNA + T4 ligase; lane 9, linearized DNA.
4
and 6, Figures 7 and 8, respectively). Compounds 2 and 4
were examined at concentrations as high as 1000 µM. In
agreement with the results obtained in viscosity and fluorescence
experiments, the compounds showed little or no ability to induce
electrophoretic shifts in plasmid DNA mobility, indicating no
or an extremely weak ability to bind to DNA intercalatively.
While this is not surprising for a compound such as 2 that bears
a negative charge at neutral pH, the identical lack of shifts was
observed the corresponding methyl ester. In particular, with
methyl naphthoate 4 (Figure 8), there was no detectable shift
in DNA mobility with concentrations of 4 as high as 1 mM.
This experiment is based on a concentration-dependent shift in
topoisomer mobility, and an examination of lanes 1-3 in Figure
hence negatively supercoiled. Such DNA will migrate more
rapidly upon agarose gel electrophoresis compared to linearized
DNA ligated in the absence of an intercalating agent.
As shown in Figure 9 (lanes 1-4), ligation of linearized DNA
in the presence of 2 did not promote the production of negatively
supercoiled DNA even at very high concentrations (1000 µM).
Similar results were obtained with compound 4 (Figure 10, lanes
1
-4). With 5, products with slower mobility started to appear
at a compound concentration of 1000 µM (Figure 11, lane 1).
We interpreted this result as a sign of inhibition of the activity
of the enzyme because the product resembles nicked or
incompletely ligated DNA. Inhibition of the activity of the
enzyme was confirmed on lane 2 (Figure 11), where the enzyme
was added simultaneously with 5 to the linearized DNA
substrate. The inhibition is manifested by the small change in
the amount of the linearized DNA starting material during the
ligation reaction and the formation of a small amount of nicked
or incompletely ligated DNA observed just above the band of
the DNA substrate.
Orientation of the Agent. Definition of the orientation of
binding of azinomycin B to the target sequence d(GGC)‚(CCG)
was determined using a gel electrophoresis-based assay that
differentiated kinetically between the initial monoalkylation and
subsequent cross-link formation. Using a short 15-mer strand 6
7
or lanes 2-5 in Figure 8 leads to the conclusion that such a
shift does not occur.
We also performed a DNA unwinding experiment with T4
ligase. Similar to the topoisomerase I experiments, this assay
makes use of the ability of intercalating agents to unwind double
stranded DNA. Linearized plasmid DNA is treated with an
intercalating agent followed by the enzyme T4 ligase, which
links the two ends of the duplex to re-form the closed circular
plasmid. If the agent under examination binds to duplex DNA
intercalatively, then upon ligation and subsequent removal of
the agent, the closed circular DNA will be underwound and
(28) Viola, G.; Uriarte, E.; Gia, O.; Moro, S. Farmaco 2000, 55, 276.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002 13015

A R T I C L E S
Coleman et al.
sequence selectivity of azinomycin B that is reported in
1
3
Table 1.
Conclusion
From the preliminary set of experiments evaluating the
sequence selectivity of binding, it would appear that sequence
selectivity of cross-link formation is related generally to
nucleobase nucleophilicity and specifically to steric effects in
the major groove of duplex DNA. There are some subtle
sequence effects on agent binding, but at a basic level it appears
that the effectiveness of the agent in the formation of covalent
cross-links is related more to the nucleophilicity of the DNA
sequence rather than to any specific recognition elements.
However, this simplistic assessment will certainly evolve, as
additional details of the specific contacts between the agent and
the DNA receptor are uncovered.
The conclusion that can be drawn from the assays for
intercalative binding is that the azinomycin naphthalene does
not detectably intercalate within duplex DNA. In both the full
and partial structures examined, we could find no evidence for
intercalative binding. Given the structural analogy between the
azinomycin and neocarzinostatin naphthoate systems, the latter
of which has been demonstrated to bind duplex DNA inter-
Figure 12. Kinetic assay for regioselective of cross-link formation. Lanes
1
-5, 2.0, 1.5, 1.0, 0.5, and 0 h reaction, respectively; lane 6, 24 h reaction;
18
calatively, we had anticipated that the two systems would share
lane 7, single strand 7; lane 8, single strand 6.
mechanistic similarity. However, it is not clear whether the
neocarzinostatin naphthoate provides binding affinity on its own
or whether other structural features of the molecule are driving
a weak intercalative binding. Apparently, in the structures
examined, including the parent azinomycin molecules, there is
no appreciable intercalative binding. These results raise the
interesting question of what role this hydrophobic aromatic
group plays in the mechanism of action of these antitumor
agents. Our experimental results with epoxide-bearing partial
structures have shown that the naphthoate is essential for
effective covalent interaction of the agents with DNA,27 so we
can draw the limited conclusion that the naphthoate provides
an important degree of noncovalent association that may be due
simply to hydrophobic interactions with the DNA duplex. It
also raises interesting questions in drug design with respect to
optimizing noncovalent interactions of these agents with DNA
in efforts to increase selectivity and potency.
and long 21-mer strand 7, we 5′ end-labeled both strands with
P and followed the progression of covalent bond formation
3
2
from initial monoalkylation to cross-link formation. We worked
under the assumptions that the aziridine ring of 1b is the most
reactive electrophile, consistent with observations made during
the course of synthetic work, and that the initial covalent bond
is formed between one of the DNA dG-N7’s and C10 of 1b.
Which strand underwent the initial alkylation was not known.
Following the progression of covalent adduct formation by
denaturing polyacrylamide gel electrophoresis (Figure 12), we
observed the rapid monoadduct formation with a concomitant
decrease in the amount of longer strand 7 (lane 5) accompanied
by a small amount of cross-link formation. As the reaction
progresses, the long DNA strand 7 reacts completely (lane 1)
and with kinetics distinct from that for cross-link formation.
Furthermore, the rate of cross-link formation is identical with
that for the disappearance of short strand 6, and these observa-
tions suggest that the aziridine alkylates the dG of the 3′-CGG-
With respect to the discrepancy between our results and those
1
4
of Zang and Gates, their conclusion of intercalative binding
by the azinomycin naphthoate was based on a single viscometry
experiment. (Their remaining experiments are equivocal with
respect to intercalative versus nonintercalative binding.) In their
viscometry results, using a compound similar to our 4 (com-
pound 9 in their work), they observed a marked increase in the
viscosity of solutions of sheared herring sperm DNA (100-
200 bp). However, since these workers did not include controls
(i.e., a known intercalator or a molecule demonstrated to bind
nonintercalatively), and because of the extremely narrow
dynamic range for flow times that were observed (88-94 s),
we do not feel that this experiment can be unequivocally or
reliably interpreted to indicate intercalative binding.
5
′-containing strand as the site of the initial monoalkylation.
We know from our synthetic work that the aziridine of the
azinomycin azabicyclic system is exceptionally reactive toward
even weak nucleophiles such as bromide or water, whereas the
epoxide is not. While it is strictly true that we have no conclusive
evidence that the aziridine is more reactive than the epoxide in
the drug-DNA complex, it would be extremely unlikely that
the large difference in reactivity between the aziridine and
epoxide in the native agent would be completely reversed upon
binding to DNA.
This work is consistent with the model put forth by Saito
10
and co-workers, who proposed that the aziridine reacts in the
initial monoalkylation event, followed by subsequent cross-
linking by the epoxide. Detailed molecular modeling studies
Concerning the T4 DNA ligase assay used by Zang and
Gates,1 wherein unwinding of linearized plasmid DNA in the
presence of an intercalating agent is measured, when these
workers added increasing concentrations of a compound related
to 5 in the presence of T4 DNA ligase, an increase in the linear
4
1
2
not only support the proposed mode of cross-link formation
but also reproduce accurately the experimentally observed
13016 J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002

Mechanism of Action of Azinomycin B
A R T I C L E S
form of the plasmid was observed, rather than the expected
supercoiled form, as was observed with the intercalator dauno-
mycin. This may be due to inhibition of the ligase enzyme by
the drug, but it is does not appear to be the result of intercalation
by the agent.
conformational properties of the drug-DNA monoadducts and
complexes confirmed the most probable mechanism of action
involves an initial aziridine and subsequent epoxide alkylation
in the orientation proposed. The different hydrogen bonding
networks in the monoadducts and in the complexes between
the drug and the three-base-pair receptor is the primary reason
for the sequence-dependent cross-linking reactivity. In addition,
steric hindrance of the major groove exposed methyl group of
central thymine-based triplets plays an important role in the lack
of the reactivity of those sequences.
Our results on the regioselectivity of agent-DNA cross-link
formation provide evidence that the more nucleophilic purine
reacts with the most electrophilic site of the agent. The agent
forms a covalent cross-link by an initial reaction of the aziridine
C10 with the 5′-G of the 3′-PyPuPu-5′-containing strand in the
monoalkylation event with subsequent cross-link formation
occurring between the epoxide C21 and the 5′-disposed purine
two bases removed on the complementary strand. This conclu-
sion is based on observations made in the course of synthetic
studies that the aziridine ring of azabicyclo[3.1.0]hexane systems
is exceptionally electrophilic, and readily reacts with weak
nucleophiles such as bromide under dilute reaction conditions
The results presented are selected from an extensive set of
experiments, and they provide a preliminary picture of our work
on the interaction of the azinomycins with duplex DNA. In
particular, we have begun to define the origin of the sequence
selectivity exhibited by the natural agent, and we have defined
the regioselectivity of the covalent cross-link. There appears to
be no appreciable intercalative binding of the natural agents or
naphthalene-bearing partial structures to duplex DNA. In related
(
CH2Cl2, 25 °C).29 This assumption remains to be proven
27
unequivocally.
work, we have examined the covalent interactions of epoxide-
bearing partial structures with duplex DNA, and we have found
that the naphthalene is essential for achieving appreciable
covalent bond formation. The issue of identifying the role of
the aromatic group in noncovalent interactions of the agents
with DNA currently remains incompletely resolved.
Acknowledgment. This work was supported by the National
Institutes of Health (Grant CA-65875). R.S.C. was the recipient
of an Alfred P. Sloan Foundation Research Fellowship (1995-
8
). A.N. was a fellow of the Spanish Ministerio de Educacion
(
EX 0052672518). We thank Matthew Mendlik for developing
In a series of detailed investigations using Monte Carlo
simulations of DNA-drug adducts, we observed a good
correlation between experimental data and computational esti-
the isolation protocol for azinomycin B, Dan Ordaz and Jon-
David Sears of the OSU Fermentation Facility for performing
fermentations, and Professor Claudia Turro for help with the
fluorescence contact energy transfer experiments.
12,13
mations of sequence selectivity.
The comparison of the
(
29) Coleman, R. S.; Richardson, T. E.; Carpenter, A. J. J. Org. Chem. 1998,
6
3, 5738.
JA025563K
J. AM. CHEM. SOC.
9
VOL. 124, NO. 44, 2002 13017