Efficient and specific strand scission of DNA by a dinuclear copper complex: Comparative reactivity of complexes with linked tris(2-pyridylmethyl)amine moieties

Article abstract of DOI:10.1021/ja020039z

The compound [CU^II₂(D¹)(H₂O)₂] (ClO₄)₄ (D¹ = dinucleating ligand with two tris(2-pyridylmethyl)-amine units covalently linked in their 5-pyridyl positions by a -CH₂CH₂- bridge) selectively promotes cleavage of DNA on oligonucleotide strands that extend from the 3′ side of frayed duplex structures at a site two residues displaced from the junction. The minimal requirements for reaction include a guanine in the n (i.e. first unpaired) position of the 3′ overhang adjacent to the cleavage site and an adenine in the n position on the 5′ overhang. Recognition and strand scission are independent of the nucleobase at the cleavage site. The necessary presence of both a reductant and dioxygen indicates that the intermediate responsible for cleavage is produced by the activation of dioxygen by a copper(I) form of the dinuclear complex. The lack of sensitivity to radical quenching agents and the high level of site selectivity in scission suggest a mechanism that does not involve a diffusible radical species. The multiple metal center exhibits a synergy to promote efficient cleavage as compared to the action of a mononuclear analogue [Cu^II(TMPA)-(H₂O)](ClO₄)₂ (TMPA = tris(2-pyridylmethyl)amine) and [Cu(OP)₂]²⁺ (OP = 1,10-phenanthroline) at equivalent copper ion concentrations. The dinuclear complex, [Cu^II₂(D¹)(H₂O)₂] (ClO₄)₄, is even capable of mediating efficient specific strand scission at concentrations where [Cu(OP)₂]²⁺ does not detectably modify DNA. The unique coordination and reactivity properties of [Cu^II₂(D¹)(H₂O)₂] (ClO₄)₄ are critical for its efficiency and site selectivity since an analogue, [Cu^II₂(DO)(Cl₂)] (ClO₄)₂, where DO is a dinucleating ligand very similar to D¹, but with a -CH₂OCH₂- bridge, exhibits only nonselective cleavage of DNA. The differences in the reactivity of these two complexes with DNA and their previously established interaction with dioxygen suggest that specific strand scission is a function of the orientation of a reactive intermediate.

Full text of DOI:10.1021/ja020039z

Published on Web 05/03/2002
Efficient and Specific Strand Scission of DNA by a Dinuclear
Copper Complex: Comparative Reactivity of Complexes with
Linked Tris(2-pyridylmethyl)amine Moieties
Kristi J. Humphreys,† Kenneth D. Karlin,*,† and Steven E. Rokita*,‡
Contribution from the Department of Chemistry, The Johns Hopkins UniVersity,
Baltimore, Maryland 21218, and Department of Chemistry and Biochemistry, UniVersity of
Maryland, College Park, Maryland 20742
Received January 8, 2002
Abstract: The compound [CuII (D1)(H2O)2](ClO4)4 (D1 ) dinucleating ligand with two tris(2-pyridylmethyl)-
2
amine units covalently linked in their 5-pyridyl positions by a -CH2CH2- bridge) selectively promotes
cleavage of DNA on oligonucleotide strands that extend from the 3′ side of frayed duplex structures at a
site two residues displaced from the junction. The minimal requirements for reaction include a guanine in
the n (i.e. first unpaired) position of the 3′ overhang adjacent to the cleavage site and an adenine in the n
position on the 5′ overhang. Recognition and strand scission are independent of the nucleobase at the
cleavage site. The necessary presence of both a reductant and dioxygen indicates that the intermediate
responsible for cleavage is produced by the activation of dioxygen by a copper(I) form of the dinuclear
complex. The lack of sensitivity to radical quenching agents and the high level of site selectivity in scission
suggest a mechanism that does not involve a diffusible radical species. The multiple metal center exhibits
a synergy to promote efficient cleavage as compared to the action of a mononuclear analogue [CuII(TMPA)-
(H2O)](ClO4)2 (TMPA ) tris(2-pyridylmethyl)amine) and [Cu(OP)2]2+ (OP ) 1,10-phenanthroline) at equivalent
copper ion concentrations. The dinuclear complex, [CuII (D1)(H2O)2](ClO4)4, is even capable of mediating
2
efficient specific strand scission at concentrations where [Cu(OP)2]2+ does not detectably modify DNA.
The unique coordination and reactivity properties of [CuII (D1)(H2O)2](ClO4)4 are critical for its efficiency
2
and site selectivity since an analogue, [CuII (DO)(Cl2)](ClO4)2, where DO is a dinucleating ligand very similar
2
to D1, but with a -CH2OCH2- bridge, exhibits only nonselective cleavage of DNA. The differences in the
reactivity of these two complexes with DNA and their previously established interaction with dioxygen suggest
that specific strand scission is a function of the orientation of a reactive intermediate.
Introduction
widely applied as a footprinting agent of both proteins and
DNA,7,8 a probe of the dimensions of the minor groove,9 and
The previous two decades have seen a proliferation of
research on the interactions of transition metal complexes with
nucleic acids.1-4 Numerous parameters including the size, shape,
and chirality of complexes as well as the structure and
composition of target DNA control both binding and modifica-
tion.5,6 Although many coordination compounds have been
characterized extensively, the 1,10-phenanthroline-copper com-
plex, [Cu(OP)2]2+, has garnered a large degree of attention due
to its high nucleolytic efficiency. This complex has also been
an identifier of transcription start sites.10,11 All of these
investigations were conducted with the 2:1 complex of [Cu-
(OP)2]2+, in which two phenanthroline ligands chelate one
copper ion and are assumed to require 3 equiv of complex for
DNA cleavage. This complex is proposed to bind to DNA
through intercalation of one phenanthroline into the DNA or
general hydrophobic interactions between the ligand and the
minor groove of DNA. Reaction has also been directed to
specific nucleotide sequences by covalently linking phenan-
throline to a DNA binding element (i.e. proteins or oligonucle-
otides) through a derivatized amido moiety.12,13 In both cases,
* To whom correspondence should be addressed. E-mail: karlin@jhu.edu.,
rokita@umd.edu.
† The Johns Hopkins University.
‡ University of Maryland.
(1) Perrin, D. M.; Mazumder, A.; Sigman, D. S. In Progress in Nucleic Acid
Chemistry and Molecular Biology; Cohn, W., Moldave, K., Eds.; Academic
Press: New York/Orlando, 1996; Vol. 52, pp 123-151.
(2) Pratviel, G.; Bernadou, J.; Meunier, B. Angew. Chem., Int. Ed. Engl. 1995,
34, 746-769.
(7) Kuwabara, M. D.; Yoon, C.; Goyne, T. E.; Thederahn, T.; Sigman, D. S.
Biochemistry 1986, 25, 7401-7408.
(8) Sigman, D. S. In DNA and RNA CleaVers and Chemotherapy of Cancer
and Viral Diseases; Meunier, B., Ed.; Kluwer Academic Publishers: The
Dordrecht, Netherlands, 1996; pp 119-132.
(3) Pratviel, G.; Bernadou, J.; Meunier, B. AdV. Inorg. Chem. 1998, 45, 251-
(9) Sigman, D. S.; Mazumder, A.; Perrin, D. M. Chem. ReV. 1993, 93, 2295-
312.
2316.
(4) Sigel, A.; Sigel, H., Eds. Metal Ions in Biological Systems; Marcel
Dekker: New York, 1996; Vol. 33.
(10) Spassky, A.; Sigman, D. S. Biochemistry 1985, 24, 8050-8056.
(11) Thederahn, T. B.; Spassky, A.; Kuwabara, M. D.; Sigman, D. S. Biochem.
Biophys. Res. Commun. 1990, 168, 756-762.
(5) Barton, J. K. In Bioinorganic Chemistry; Bertini, I., Gray, H. B., Lippard,
S. J., Valentine, J. S., Eds.; University Press: Mill Valley, CA, 1994; pp
455-503.
(12) Chen, C.-h. B.; Sigman, D. S. Science 1987, 237, 1197-1201.
(13) Chen, C.-h. B.; Sigman, D. S. Proc. Natl. Acad. Sci. 1986, 83, 7147-
7151.
(6) Pyle, A. M.; Barton, J. K. Prog. Inorg. Chem. 1990, 38, 413-475.
9
10.1021/ja020039z CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 6009-6019
6009

A R T I C L E S
Humphreys et al.
Chart 1
the binding and recognition of DNA are dominated by ligand
interactions with the double-helical structure. The copper ion,
although crucial for generating the intermediate responsible for
strand cleavage, does not directly interact with DNA.
Our laboratories have begun to explore the intriguing
chemistry of copper-mediated O2 chemistry with DNA by using
a complementary set of coordination compounds. Attention has
initially focused on multinuclear copper complexes due to their
potential for efficient intramolecular activation of bound O2 and
for binding in a selective manner to particular nucleic acid
conformations. Both features were recently demonstrated by the
Chart 2
ability of a trinuclear complex, [CuII (L)(H2O)3(NO3)2](NO3)4‚
3
5H2O (1) (L
)
2,2′,2”-tris(dipicolylamino)triethyl-
amine),14-16 to recognize a helix/coil junction of DNA and
promote selective strand scission upon reduction of 1 in the
presence of dioxygen. Unlike [Cu(OP)2]2+, the copper ions in
1 appear to perform a dual role in both recognition and
reactivity. The unique reactivity of 1 poses several questions
regarding how the ligand and copper ion centers control binding
and how the multiple metal ions produce the oxygen-derived
reactive intermediate.
The unique recognition and specificity of 114,15 and the high
nucleolytic efficiency of [Cu(OP)2]2+ encouraged us to examine
other multinuclear copper complexes for related properties. This
survey identified a dinuclear complex [CuII (D1)(H2O)2](ClO4)4
2
(2) (see Chart 1 for identity of D1) that mediated strand scission
at junctions between single-stranded and double-helical regions
of DNA. The site selected by 2 is not recognized or cleaved by
1, although both compounds mediate cleavage on the 3′
overhang strand at single target positions. Most importantly,
are also known to activate dioxygen.20,21 The observed differ-
ences in overall reactivity and specific cleavage by the three
complexes may be linked to the formation and orientation of
O2-derived intermediates.
the nucleolytic efficiency of [CuII (D1)(H2O)2](ClO4)4 (2) is
2
much greater than that of [CuII (L)(H2O)3(NO3)2](NO3)4‚5H2O
3
(1) and [Cu(OP)2]2+ at low concentrations and correlates well
Experimental Section
with the established dioxygen reactivity of the reduced form of
2, [CuI2(D1)(MeCN)]2+ 17
Further comparison with known
.
Materials. Oligodeoxynucleotides were purchased from Gibco BRL.
T4 kinase and its buffer were obtained from New England Biolabs
and [γ-32P]ATP (3000 Ci/mmol) was obtained from Amersham. The
binucleating ligand D1 and the mononucleating analogue TMPA and
its coordination complex (3)17,18,22 were synthesized according to
published procedures. Synthesis of the ligand DO and its corresponding
dicopper(II) complex (4) will be published elsewhere. Solutions of [Cu-
analogues of 2 that are both mono- and dinuclear provides
insight into the origins of recognition and reactivity. The
mononuclear compound, [CuII(TMPA)(H2O)](ClO4)2 (3), and
the dinuclear analogue, [CuII (DO)(Cl)2](ClO4)2 (4) (Chart 2),
2
possess similar metal coordination environments18,19 to 2 and
(14) Humphreys, K. J.; Karlin, K. D.; Rokita, S. E. J. Am. Chem. Soc. 2001,
123, 5588-5589.
(18) Jacobson, R. R.; Tyeklar, Z.; Karlin, K. D.; Zubieta, J. Inorg. Chem. 1991,
30, 2035.
(15) Humphreys, K. J.; Karlin, K. D.; Rokita, S. E. Manuscript submitted for
publication.
(19) Nagao, H.; Komeda, N.; Mukaida, M.; Suzuki, M.; Tanaka, K. Inorg. Chem.
1996, 35, 6809-6815.
(16) Frey, S. T.; Sun, H. H. J.; Murthy, N. N.; Karlin, K. D. Inorg. Chim. Acta
1996, 242, 329-338.
(17) Lee, D.-H.; Wei, N.; Murthy, N. N.; Tyekla´r, Z.; Karlin, K. D.; Kaderli,
S.; Jung, B.; Zuberbu¨hler, A. D. J. Am. Chem. Soc. 1995, 117, 12498-
12513.
(20) Tyekla´r, Z.; Jacobson, R. R.; Wei, N.; Murthy, N. N.; Zubieta, J.; Karlin,
K. D. J. Am. Chem. Soc. 1993, 115, 2677-2689.
(21) Karlin, K. D.; Lee, D.-H.; Kaderli, S.; Zuberbu¨hler, A. D. J. Chem. Soc.,
Chem. Commun. 1997, 475-476.
9
6010 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Efficient Strand Scission of DNA by a Dicopper(II) Complex
A R T I C L E S
(OP)2]2+ were prepared according to the literature23 and examined under
conditions identical with those described for reaction of 2. Solutions
of the metal complexes and other reagents for strand scission were
prepared fresh daily in distilled-deionized water (18 MΩ‚cm). Stock
solutions of 3-mercaptopropionic acid (MPA), glutathione (GSH), and
dithiothreitol (DTT) were titrated with Ellman’s reagent to determine
free thiol concentration.24 All other chemicals were used as supplied
by the manufacturer.
D1. To 1.45 g of [Cu(CH3CN)4]PF6 (3.9 mmol) in a 100 mL Schlenk
flask under argon was added dropwise 40 mL of freshly distilled and
degassed CH3CN containing 1.05 g of TMPA-Cl (3.11 mmol). The
reaction was allowed to stir under argon for 18 h, then concentrated
by rotary evaporation to give a green solid. The solid was dissolved in
100 mL of CH2Cl2 and extracted three times with 100 mL of
concentrated NH4OH(aq). The organic layer was washed three times
with 100 mL of saturated sodium carbonate and dried over MgSO4.
Filtration through Celite and concentration yielded a yellow oil. The
oil was purified on alumina by initial elution with ethyl acetate to
remove any remaining TMPA-Cl followed by elution with 5:95
methanol/ethyl acetate to give the reductively coupled product (Rf )
0.15). Rotary evaporation of the product fractions yielded a white solid
Methyl(6-bromomethyl)nicotinate. Recrystallized N-bromosuccin-
imide (13.36 g, 75 mmol), 13.4 g of methyl(6-methyl)nicotinate (89
mmol), and 174 mg of benzoyl peroxide (0.7 mmol) were added to
350 mL of CCl4. The mixture was heated to reflux and irradiated with
a 90 W bulb for 18 h. The orange solution containing a brown solid
was filtered through Celite to remove the solid and washed three times
with 100 mL of saturated sodium carbonate. The organic layer was
dried over MgSO4 for 30 min, filtered through Celite, and concentrated
by rotary evaporation to give a pink solid. The solid was purified by
flash column chromatography on silica (70-230 mesh), eluting with
1:4 ethyl acetate/hexanes. Fractions containing the monobrominated
product (Rf ) 0.22) were concentrated by rotary evaporation to give a
1
in 84% yield. H NMR (CDCl3): δ 2.88 (s, 4 H), 3.85 (s, 4 H), 3.87
(s, 8 H), 7.12-7.14 (m, 4 H), 7.41-7.68 (m, 12 H), 8.34 (s, 2 H), 8.53
(d, 4 H). FAB mass spectrum: m/z 607 (M + 1)+.
[CuII2(D1)(H2O)2](ClO4)4 (2). A methanolic solution (5 mL) of Cu-
(ClO4)2‚6H2O (0.25 g, 0.67 mmol) was added dropwise to a test tube
containing a 5 mL solution of D1 (0.2 g, 0.3 mmol) in MeOH. After 1
h at room temperature, blue crystals developed which were isolated
by filtration, washed with Et2O, and dried in vacuo to yield 0.35 g of
crystalline solid (88%). Anal. Calcd (Found) for C38H42N8Cu2Cl4O18:
C, 39.09 (38.46); H, 3.63 (3.63); N, 9.6 (9.33). UV-vis λmax 870 nm
(ꢀ 420 M-1 cm-1). IR (Nujol, cm-1): 3217 (m, br, OH), 1610 (m,
HOH), 1107 (vs, ClO4-). EPR: g 2.198, g| 2.0155, A 85 G. The
EPR spectrum is characteristic of that observed for trigonal bipyramidal
Cu(II) complexes of TMPA.25,26
1
crystalline white solid in 57% yield. H NMR (CDCl3): δ 3.97 (s, 3
H), 4.59 (s, 2 H), 7.53-7.55 (d, 1 H), 8.29-8.32 (d, 1 H), 9.17 (d, 1
H).
TMPAE. To 6.28 g of methyl(6-bromomethyl)nicotinate (27 mmol)
was added 6.65 g of PY1 (dipicolylamine) (33 mmol), 300 mL of THF,
and 8 mL of diisopropylethylamine (46 mmol). The mixture was stirred
for 24 h at room temperature, filtered through Celite, and concentrated
by rotary evaporation to give a dark brown oil. The oil was dissolved
in diethyl ether, filtered through Celite, and placed at -20 °C to give
a pale yellow solid, which was isolated by filtration. The solid was
dried in vacuo resulting in a 64% yield (Rf ) 0.27, alumina, ethyl
Purification and Labeling of DNA. Oligonucleotides were purified
prior to use by denaturing (7 M urea) polyacrylamide gel electrophoresis
and elution with 50 mM NaOAc and 1 mM EDTA (pH 5.2). The
resulting solution was extracted with phenol/chloroform and the DNA
was precipitated by the addition of ethanol. DNA was dried under
reduced pressure and redissolved in water. Concentrations were
determined spectrophotometrically at 260 nm using calculated extinction
coefficients.27 DNA was radiolabeled by incubation with [γ-32P]ATP
and T4 kinase according to the supplier. The 5′-32P-labeled DNA was
isolated by passage over a MicroBioSpin P-6 column (Bio-Rad). Frayed
duplex structures containing a 5′-32P-labeled oligonucleotide (100 nM,
90 nCi) and a complementary sequence (200 nM) were annealed in
sodium phosphate (10 mM, pH 7.5) by heating to 90 °C followed by
slow cooling to room temperature. Single-stranded constructs were
formed by mixing 5′-32P-labeled oligonucleotides (100 nM, 90 nCi) in
sodium phosphate (10 mM, pH 7.5).
1
acetate). H NMR (CDCl3): δ 3.80 (s, 3 H), 3.87 (s, 4 H), 3.88 (s, 2
H), 7.15-7.20 (m, 2 H), 7.54-7.73 (m, 6 H), 8.18-8.22 (d, 1 H),
8.45-8.47 (d, 2 H), 9.01 (d, 1 H).
TMPAOH. To 0.85 g of lithium aluminum hydride (22 mmol) under
argon and cooled to 4 °C in an ice bath was added dropwise 3.48 g of
TMPAE (10 mmol) dissolved in 225 mL of dioxygen-free diethyl ether.
The reaction mixture was stirred for 18 h at ambient temperature, then
cooled to 4 °C in an ice bath and quenched by the dropwise addition
of 1 mL of H2O, 1 mL of 10% NaOH, and 1 mL of H2O. The quenched
reaction was allowed to stir for 2 h at room temperature, extracted
with 500 mL diethyl ether, and filtered through Celite. The filtrate was
dried over MgSO4, filtered through Celite, and concentrated by rotary
evaporation to give a pale yellow oil in 99% yield. This crude product
was used without further purification (Rf ) 0.22, alumina, 5:95
methanol/ethyl acetate). 1H NMR (CDCl3): δ 3.83 (s, 6 H), 4.65 (s, 2
H), 7.13-7.15 (m, 2 H), 7.52-7.62 (m, 6 H), 8.42-8.47 (m, 3 H).
Copper-Dependent Strand Scission. Various concentrations of the
dinuclear or mononuclear copper complexes were combined with a
labeled DNA sample (100 nM, 90 nCi) in sodium phosphate buffer
(10 mM, pH 6.8) and strand scission was initiated by addition of the
reductant (5 mM). The reaction was quenched after a 15-min incubation
at ambient temperature with 10 mM diethyl dithiocarbamic acid
(5 µL).28 DNA was then isolated by ethanol precipitation and dried
under high vacuum. As indicated, certain samples were further treated
with 20 µL of piperidine (0.2 M) for 30 min at 90 °C. These samples
were dried under reduced pressure and twice redissolved with 20 µL
of water, then subsequently dried to remove trace quantities of
piperidine. The isolated DNA was resuspended in water, normalized
to 45 nCi per sample, and mixed with loading buffer (0.25%
bromophenol blue, 0.25% xylene cyanol, 3% sucrose, and 7 M urea).
The samples with loading buffer were then separated by denaturing (7
M urea) polyacrylamide (20%) gel electrophoresis and visualized by
TMPACl. To a solution of 2.93 g of crude TMPA-OH (9 mmol) in
75 mL of CHCl3 cooled to 4 °C in an ice bath was added dropwise 5
mL of thionyl chloride (69 mmol). The reaction was stirred for 30 min
at 4 °C, then warmed to room temperature and allowed to stir for
another 18 h. The dark green solution was concentrated by rotary
evaporation to give a green solid. The solid was redissolved in 100
mL of CH2Cl2 and washed three times with 100 mL of saturated sodium
carbonate. The organic layer was dried over MgSO4 for 2 h, then filtered
through Celite and absorbed onto alumina. Purification by column
chromatography on alumina, eluting with ethyl acetate, and concentra-
tion by rotary evaporation yielded a pale yellow solid in 79% yield (Rf
1
) 0.33, alumina, 5:95 methanol/ethyl acetate). H NMR (CDCl3): δ
3.90 (s, 6 H), 4.57 (s, 2 H), 7.13-7.17 (m, 2 H), 7.54-7.69 (m, 6 H),
8.53-8.55 (m, 3 H).
(25) Karlin, K. D.; Cruse, R. W.; Kokoszka, B. F.; Orsini, J. J. Inorg. Chim.
Acta 1982, 66, L57-L58.
(26) Karlin, K. D.; Hayes, J., C.; Shi, J.; Hutchinson, J. P.; Zubieta, J. Inorg.
Chem. 1982, 21, 4106-4108.
(27) Fasman, G. D.; Sober, H. A., Eds. Handbook of Biochemistry and Molecular
Biology, 3rd ed.; CRC Press: Boca Raton, FL, 1975; Vol. 2.
(28) Humphreys, K. J.; Johnson, A. E.; Karlin, K. D.; Rokita, S. E. J. Biol.
Inorg. Chem., JBIC, 2002, in press.
(22) Anderegg, v. G.; Hubmann, E.; Podder, N. G.; Wenk, F. HelV. Chim. Acta
1977, 60, 123-140.
(23) Yoon, C.; Kuwabara, M. D.; Law, R.; Wall, R.; Sigman, D. S. J. Biol.
Chem. 1988, 263, 8458-8463.
(24) Ellman, G. L. Arch. Biochem. Biophys. 1959, 82, 70-77.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 6011

A R T I C L E S
Humphreys et al.
Figure 2. Autoradiograms of 20% polyacrylamide denaturing gels (7 M
urea) showing cleavage products of 5′-32P-labeled Ia (100 nM) induced by
multinuclear copper complexes in the presence of MPA for 15 min in sodium
phosphate (10 mM, pH 6.8) at ambient temperature. (A) Watson strand of
Ia and (B) Crick strand of Ia.
mM MPA for 15 min at ambient temperature (Figure 2A, lanes
5 and 6). The site preferentially cleaved by 2 is on the 3′
overhang and displaced from the central duplex by one
intervening nucleotide, G23. Reaction with the Crick strand of
Ia (shown on the right in Figure 1A) did not result in specific
cleavage. However, 2 demonstrated weak recognition of the
junction by a slightly enhanced level of cleavage at relatively
high concentrations of the complex (Figure 2B, lane 6).
The observed activity of 2 directly contrasted with that of
Figure 1. Structures and sequences of the frayed duplex constructs I and
II. The sites of specific strand scission by 2 are indicated by an arrowhead.
autoradiography and PhosphorImager (Molecular Dynamics). Quanti-
tation of the products relied on ImageQuant software.
the previously studied trinuclear complex, [CuII (L)(H2O)3-
Dioxygen Dependence of Strand Scission. To test the dependence
of the strand scission reaction on O2, the standard reaction conditions
were modified to limit the amount of dioxygen in the reaction mixture.
To achieve an O2-limited environment, a solution containing 100 nM
DNA IIa and 1 µM 2 was degassed by bubbling the solution with
prepurified nitrogen using a syringe needle for 15 min prior to initiation
with 5 µL of undegassed MPA (5 mM). During the 15 min incubation,
the reaction was kept under a nitrogen atmosphere. Quenching and
analysis followed the same procedures as described above.
3
(NO3)2](NO3)4‚5H2O (1),14,15 which mediated specific cleavage
at C20 of the Crick strand of Ia under the same conditions
(Figure 2B, lane 3). Reaction of 1 with the Watson strand of Ia
did not produce specific strand scission (Figure 2A, lane 3).
The selectivity of 1 had previously been characterized and found
to promote scission directly adjacent to a junction between
single- and double-stranded DNA on the 3′ single-stranded
overhang and require adjacent purine and G at the n (i.e. first
unpaired) and n + 1 (i.e. second unpaired) positions.14,15
Although both complexes mediate specific cleavage near helix/
coil junctions, they do not promote these reactions at the same
junctions. This may indicate similar, but not identical structural
requirements for recognition and selective cleavage. The absence
of a guanine in the n + 1 position of the 5′ overhang of the
Crick strand explains the lack of specific reactivity of 1 at this
junction. Reaction of 2 also appears to be more efficient than 1
at equivalent copper ion concentrations as evident from the
greater level of selective and background cleavage observed
(Figure 2A, lane 5 vs Figure 2B, lane 2).
Strand Scission in the Presence of Radical Scavengers. To test
for the presence of radical intermediates formed during strand scission,
100 mM ethanol, D-mannitol, and tert-butyl alcohol were added
alternatively to a standard reaction yielding a total volume of 50 µL
and final concentrations of 1 µM 2, 100 nM IIa, and 10 mM radical
scavenger. After initiation with 5 µL of MPA (50 mM) and incubation
for 15 min at ambient temperature, quenching and analysis followed
that described above.
Results and Discussion
Selective Strand Scission Induced by [CuII2(D1)(H2O)2]-
(ClO4)4 (2). The dinuclear complex, [CuII (D1)(H2O)2](ClO4)4
2
(2), was initially investigated for reactivity with a range of DNA
structures to test the properties of multiple metal ions for
selective oxidation. Specific strand scission was observed on
the Watson strand at A24 of the frayed duplex IA (Figure 1a,
Watson strand shown on the left) during reaction with 2 and 5
To elucidate the origins of recognition, specific strand
scission, and reactivity of [CuII (D1)(H2O)2](ClO4)4 (2), a series
2
of new targets was designed to identify positional and possible
sequence dependences for efficient reaction at helix/coil junc-
tions. The new constructs (II, Figure 1B) moved the cleavage
9
6012 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Efficient Strand Scission of DNA by a Dicopper(II) Complex
A R T I C L E S
amount of cleavage with a concomitant decrease in the percent-
age of that directed to A21 due to increasing amounts of
sequence-neutral background cleavage in the presence of higher
concentrations of 2. A similar response was observed for
cleavage of Ia by 1 at a higher concentration range of 0.5 to 5
µM.14
Both the selective strand cleavage at A21 and general
background cleavage of IIa mediated by [CuII (D1)(H2O)2]-
2
(ClO4)4 (2) and MPA occurred spontaneously and did not require
further treatment of the DNA with heat or alkali. The cleavage
products of these reactions are therefore likely derived through
oxidation of C-3′, C-4′, or C-5′, but not the C-1′ carbon of the
deoxyribose moiety of the DNA backbone.29 The degree of
competing oxidation resulting in an abasic site or occurring
directly at the nucleobases may be indicated by cleavage induced
by piperidine (90 °C) after reaction with 2. In the case of IIa,
the degree of base release and base oxidation was dependent
on the concentration of 2. At 0.1 µM of the dinuclear complex,
the amount of total cleavage both before and after piperidine
treatment was small and approximately equivalent.30 However,
as the concentration of 2 increased from 0.25 to 1 µM, the
amount of total cleavage following piperidine treatment in-
creased to 41 ( 9%. At these higher complex concentrations,
alternative modes of oxidation accounted for nearly 50% of the
total reaction, demonstrating that 2 can be equally efficient at
oxidizing various functional groups within DNA, and yet the
general site selectivity of reaction remained the same whether
spontaneous or piperidine-induced strand scission was ob-
served.
Figure 3. Autoradiogram of a 20% polyacrylamide denaturing gel (7 M
urea) showing cleavage products of 5′-32P-labeled IIa (100 nM) incubated
with 2 and MPA for 15 min in sodium phosphate (10 mM, pH 6.8) at
ambient temperature. (A) Lanes 1-4: IIa (Watson strand ) and 5 mM MPA
with 0.1, 0.25, 0.5, and 1 µM 2. Lane 5: A+G sequencing lane. (B) Lanes
1-4: IIa (Crick strand) and 5 mM MPA with 0.1, 0.25, 0.5, and 1 µM 2.
Lane 5: A+G sequencing lane.
Specific strand scission was not observed after treatment with
2 and MPA on the complementary Crick strand at either junction
in contrast to the Watson strand of IIa. The total yield of
cleavage for the Crick strand was only 6% at 1 µM 2 and no
single nucleotide accounted for more than 10% of this value
(Figure 3B, lane 4). At all concentrations of 2, the total cleavage
on the Crick strand of IIa was no more than half that observed
on the Watson strand. However, when the specific cleavage at
site further from the 3′ terminus while retaining the same central
duplex region, single-stranded portions, and sequence near the
helix/coil junction as Ia (Figure 1A). The features of the frayed
duplex that were conserved between Ia and IIa were those that
had a potential to play a role in selective cleavage near the
junction. Extension of the overhangs near the junction was
necessary to enable facile observation and analysis of specific
strand scission with 5′-32P-labeled oligonucleotides. The con-
structs were also varied at the cleavage site and key residues
thought to be crucial in reactivity.
Reaction of IIa with 2 and MPA under the standard
conditions demonstrated specific cleavage at A21 on the 3′
overhang (Figure 3A) identical with that observed at A24 of Ia
(Figure 2B). Notably, 2 once again revealed its greater efficiency
relative to 1 by mediating specific cleavage of the new target
in the presence of only 0.1 µM 2 as compared to the previous
standard of 0.5 µM set by 1.14,15 At the lower concentration of
2, approximately 1% of the oligonucleotide IIa was degraded
with 79% of that cleavage directed at A21 (Figure 3A, lane 1).
This represents a 5-fold decrease in complex concentration
relative to the lowest concentration of 1 required for mediation
of selective cleavage. Recognition and specific cleavage by 2
proceeded not only at a lower concentration than 1, but with
significantly less copper present in the reaction mixture. As the
concentration of 2 is increased 10-fold to 1 µM, the total
cleavage of IIa approached 14% with 65% of that occurring at
A21 (Figure 3A, lane 4). There was a steady increase in the
A21 on the Watson strand was subtracted from its cleavage yield
at low concentrations of 2, the sequence-neutral background
cleavage on the two strands was nearly equal. As the concentra-
tion of 2 was increased, the difference between the levels of
total cleavage on the Watson and Crick strands decreased and
corresponded to a decrease in the percentage of this cleavage
directed at A21 on the Watson strand of IIa. At low concentra-
tions of 2, strand scission is more frequently directed at A21
than at any other site on the Watson and Crick strands of IIa.
In contrast, when this site is not present as in the Crick strand
of IIa or at higher complex concentrations, sequence-neutral
strand scission becomes more prevalent. This partitioning of
reactivity between specific and nonspecific sites indicates a
preferential interaction of 2 with the junction of IIa that results
in specific cleavage at A21 of the Watson strand of this frayed
duplex.
Copper-Induced Cleavage of Single-Stranded DNA. Reac-
tion with either the Watson or Crick strand of IIa in the absence
of its complementary strand produced very little cleavage unless
the concentration of [CuII (D1)(H2O)2](ClO4)4 (2) was raised
2
to levels greater than 5 µM.30 At concentrations less than 1 µM,
(29) Pogozelski, W. K.; Tullius, T. D. Chem. ReV. 1998, 98, 1089-1107.
(30) See Supporting Information.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 6013

A R T I C L E S
Humphreys et al.
Figure 4. Effect of Y21 on specific and total cleavage of IIa, IIb, IIc, and
IId (Watson strands) with 0.5 µM 2 and 5 mM MPA under the standard
reaction conditions (see Supporting Information). Total degradation of each
construct is designated by the dark shading and the percentage of specific
strand scission at Y21 by the light shading.
degradation of the single strands was comparable to the
nonspecific reaction detected for the Crick strand of IIa and
showed little preference for any individual nucleotide either
before or after piperidine treatment.30 Interestingly, at higher
complex concentrations preferential cleavage of unclear origin
was observed at C11 of the Watson strand, but no other
nucleotides. Thus, reaction of A21 in IIa duplex cannot be due
to an intrinsic reactivity of that particular adenosine. Similarities
between the amount of nonspecific cleavage observed with
single-stranded and double-helical structures of DNA argue
against association by 2 in the major or minor groove of the
duplex region of IIa. This result suggests that specific strand
scission observed for the Watson strand of IIa arises in part
from interaction of the dinuclear complex with a unique structure
such as the helix/coil junction.
Figure 5. Autoradiogram of a 20% polyacrylamide denaturing gel (7 M
urea) showing cleavage products of 5′-32P-labeled IIe (Watson strand, 100
nM) and IIf (Watson strand, 100 nM) incubated with 2 and MPA for 15
min in sodium phosphate (10 mM, pH 6.8) at ambient temperature. (A)
Lanes 1-4: IIe and 5 mM MPA with 0.1, 0.25, 0.5, and 1 µM 2. Lane 5:
A+G sequencing lane. (B) Lanes 1-4: IIf and 5 mM MPA with 0.1, 0.25,
0.5, and 1 µM 2. Lane 5: A+G sequencing lane.
Role of the Targeted Nucleotide, Y21. To determine whether
there is also a sequence requirement for reaction at the junction,
systematic variations were made on both the 3′ and 5′ overhangs
near the duplex region of the frayed duplex construct IIa. The
first sequence modifications of IIa were made at the site of
reaction, A21, on the 3′ overhang. The nucleobase at this site
was determined to have no effect on specific strand scission by
replacing the target with thymine, guanine, or cytosine (IIb,
would base-pair with the adjacent adenine on the 5′ overhang
and shift the position of the helix/coil junction.
The total cleavage of the Watson strands of IIe and IIf by 2
(Figure 5) decreased nearly 3-fold to 5% relative to that of the
Watson strand of IIa. The strands lacking G20 produced only
nonspecific cleavage in quantities comparable to that detected
for the Crick strand of IIa under the same concentration of 2.
Neither IIe nor IIf was subject to efficient oxidation of A21.
Instead a 5-fold decrease in the percentage of total cleavage
directed at A21 from 53% to 10% was observed for IIe at 1 µM
2 (Figure 5A, lane 4), and this was accompanied by an increase
in reaction at C20 and T22 flanking A21. A similar cleavage
pattern was detected from reaction of IIf (Figure 5B, lane 4).
In this case, a slight enhancement in cleavage at A20 and T22
was observed, indicating that selectivity for A21 requires a
guanine. The small increase in cleavage near the junction in
IIe and IIf resembled the pattern observed for reaction of 2
with the Crick strand of Ia (Figure 2B). Even in the absence of
the guanine at X20 or other possible sequence requirements, the
dinuclear complex appears to have a weak intrinsic affinity for
junctions of single- to double-stranded DNA. This is further
illustrated by preferential oxidation of A4, A5, and G6, which
are part of the other helix/coil junctions in IIa, IIe, and IIf
(Figure 5). The enhancement at both junctions is distributed
over several contiguous nucleotides in contrast to the specific
reaction, which targets a single nucleotide.
IIc, and IId, Figure 1B). With 0.5 µM [CuII (D1)(H2O)2](ClO4)4
2
(2), the amount of total cleavage of IIa, IIb, IIc, and IId
remained relatively constant at 4 ( 1% and the percentage of
the total cleavage directed at Y21 varied little from 56 ( 3%
(Figure 4). Because specific strand scission was maintained at
the same level for all four sequence variants, the nucleobase at
the cleavage site could not be involved in binding or reaction
of the dinuclear complex.
Role of Guanine at the n (First Unpaired) Position of the
3′ Overhang, X20. The dinuclear complex, [CuII (D1)(H2O)2]-
2
(ClO4)2 (2), could not have similar sequence requirements to
[CuII (L)(H2O)3(NO3)2](NO3)4‚2.5H2O (1) since 2 mediates
3
strand scission at a site that is not recognized by 1. However,
the previous work with 1 suggested that interaction with a
guanine near the cleavage site might be important for recogni-
tion.14,15 In the frayed duplex construct IIa (Figure 1B), two
guanines are in close proximity to the cleavage site. One of
these, G20, precedes the reactive A21 on the 3′ overhang and is
directly adjacent to the central duplex. The other guanine is
located on the 5′ overhang at a site three nucleotides removed
from the duplex. The role of G20 was investigated first by
replacing this guanine with cytosine and adenine (IIe and IIf,
Figure 1B). Thymine could not be used equivalently since it
The ability of the guanine in the n position of the 3′ overhang
(X20, Figure 1B) to direct specific strand scission at Y21 by
9
6014 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Efficient Strand Scission of DNA by a Dicopper(II) Complex
A R T I C L E S
[CuII (D1)(H2O)2](ClO4)4 (2) was considered likely due to
2
coordination of copper to the N7 of the base and its ability to
influence the site selectivity of 1.14,15 Coordination of Cu(II)
salts to G-N7 is well precedented in duplex DNA.31 However,
in metal complexes the steric bulk of the ligand may restrict
binding of the metal center to single-stranded regions as
proposed for several macrocyclic nickel complexes.32-34 Co-
ordination through G-N7 is consistent with a lack of reactivity
when the guanine is replaced with either cytosine (IIe) or
adenine (IIf), which do not interact as strongly with transition
metals.31 Although coordination of copper to the N7 of adenine
has also been observed,31,35 binding to the G-N7 is preferred
because of its greater basicity.36 The redox chemistry of guanine,
which determines many reactions, is likely not involved in
selective strand scission of DNA by 2. Despite a requirement
for its presence, G20 is neither the site of specific strand scission
nor subject to base oxidation30 during reaction (Figures 2A and
3A). Since there are no other guanines on the 3′ overhang close
to the junction, G20 is then likely to be the primary regulator of
recognition at the helix/coil junction. Of course, this would also
rely on the ability of copper to bind without oxidizing G20. This
is possible since oxidation is likely controlled by an O2-derived
radical rather than the copper itself. Orientation of a possible
copper-peroxo intermediate may be key to the contrasting
selectivity of 2 and 1.14,15 Unique recognition of a G‚A mismatch
could provide an alternative source of selectivity. This proposal
would not necessarily suggest G-N7 binding and is currently
subject to further investigation.
Figure 6. Structures of the frayed duplex structures IIh, IIi, IIj, and IIk
with truncated 5′ overhangs. Specific strand scission is indicated by an
arrowhead.
Role of the 5′ Overhang. Single-stranded guanines are also
present in the 5′ overhang near the site of strand scission induced
by [CuII (D1)(H2O)2](ClO4)4 (2) and consequently their role in
2
reactivity was next examined. Recognition by the trinuclear
complex, [CuII (L)(H2O)3(NO3)2](NO3)4‚5H2O (1), was entirely
3
directed through a guanine and another purine on the 5′
overhang, even though cleavage occurred on the opposite 3′
extension.15 The dependence of guanine on the 5′ overhang for
reaction by 2 was first investigated with DNA IIg containing a
thymine in place of the guanine nearest to the junction (Figure
1B). Reaction of this DNA and 2 proceeded equivalently to
that of IIa and exhibited no loss in efficiency or selectivity for
A21.37 The only other guanine on the 5′ overhang is five
nucleotides away from the junction and was not present in the
frayed duplex Ia. This original DNA target also exhibited
selective strand cleavage at the n + 1 site and thus the distal
guanine in IIa-g is not required for selective reaction.
To test the general dependence of reaction on the presence
of a 5′ overhanging sequence, modification of a truncated
structure IIh completely lacking in the overhang was character-
ized (Figure 6). If only the 3′ overhang were required for
recognition, then specific strand scission on the 3′ overhang
would have proceeded without change. However, no specific
Figure 7. Densitometer trace exhibiting specific and nonspecific cleavage
of IIi and IIk by 1 µM 2 under standard conditions.
strand scission at A21 was detected for IIh.30 Moreover, the
absence of the 5′ overhang resulted in a decrease in the total
cleavage of this DNA by 5-fold compared to that of IIa at 1
µM [CuII (D1)(H2O)2](ClO4)4 (2). These results suggest that the
2
5′ overhang is required for recognition and likely helps to
stabilize the frayed duplex structure and its interaction with 2.
Selective cleavage at A21 was recovered by restoring the first
two nonhelical nucleotides of the 5′ overhang in IIi (Figure 7).
Still, the total cleavage of the Watson strand of IIi at all
concentrations of 2 was less than that observed under identical
conditions for the Watson strand of IIa. The decrease in overall
reactivity was also accompanied by a relative decrease in
selectivity for A21, which was targeted only 50% of the time (1
µM 2) relative to the 65% for IIa.30 A similar result had
(31) Sigel, H. Chem. Soc. ReV. 1993, 22, 255-267.
(32) McLachlan, G. A.; Muller, J. G.; Rokita, S. E.; Burrows, C. J. Inorg. Chim.
Acta 1996, 251, 193-199.
previously been observed for [CuII (L)(H2O)3(NO3)2](NO3)4‚
3
5H2O (1) when comparing another helix/coil junction and its
derivative with a partially truncated 5′ overhang.15
Stimulation of selective scission at A21 by the presence of
adenine and thymine on the 5′ overhang suggested that one or
both of these nucleotides could be involved in recognition by
2. The sequence requirements of this truncated 5′ extension were
(33) Shih, H.-C.; Kassahun, H.; Burrows, C. J.; Rokita, S. E. Biochemistry 1999,
38, 15034-15042.
(34) Shih, H.-C.; Tang, N.; Burrows, C. J.; Rokita, S. E. J. Am. Chem. Soc.
1998, 120, 3284-3288.
(35) Sigel, H. In Metal-DNA Chemistry; Tullius, T. D., Ed.; American Chemical
Society: Washington, DC, 1989; pp 159-204.
(36) Burrows, C. J.; Rokita, S. E. Acc. Chem. Res. 1994, 27, 295-301.
(37) Data not presented.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 6015

A R T I C L E S
Humphreys et al.
Figure 8. Effect of varying reductants on specific and total cleavage of
IIa (Watson strand) with 1 µM 2 and 5 mM reductant under the standard
reaction conditions. Total degradation of the construct is designated by the
dark shading and the percentage of specific strand scission at A21 by the
light shading.
Figure 9. Effect of standard radical scavengers on cleavage of IIa (Watson
strand) by 1 µM 2 and 5 mM MPA. Total degradation of the construct is
designated by the dark shading and the percentage of specific strand scission
at A21 by the light shading.
background cleavage of IIa, a standard reaction mixture was
degassed with nitrogen for 15 min prior to the addition of MPA.
The reaction was further kept under a nitrogen atmosphere
during its standard incubation at ambient temperature. This
resulted in a 3.5-fold decrease in reactivity relative to a parallel
sample that was not degassed.30 However, little relative change
was observed in the cleavage directed at A21 (41%). Based on
these results, the reactive intermediate or intermediates respon-
sible for specific cleavage and sequence-neutral background
cleavage appear to be derived from O2. This is also consistent
with the O2 requirement of strand scission by 1.14,15 The
dependence on O2 for both complexes is indicative of a
mechanism akin to those proposed for [Cu(OP)2]2+, which also
requires dioxygen to mediate cleavage of DNA.8
explored next by substituting the nonhelical adenine of the
truncated frayed duplex (IIi) with either a thymine or a guanine
(IIj and IIk, Figure 6). No specific strand scission at A21 on
the 3′ overhang was observed after reaction of either IIj or IIk
and [CuII (D1)(H2O)2](ClO4)4 (2) (Figure 7).30 At most, the
2
nonhelical thymine in IIi, IIj, and IIk could be relegated to a
minor role in recognition of 2 since it was present in both
reactive (IIi) and unreactive targets (IIj and IIk). The loss of
specific strand scission when the nonhelical adenine is replaced
on the 5′ overhang with thymine or guanine implicates adenine
as an absolute requirement for recognition and specific reactivity
at helix/coil junctions. Selective cleavage of the DNA also
requires a guanine in a corresponding position on the 3′
overhang.
Effect of Standard Radical Scavengers. The presence of a
diffusible radical species can be diagnosed by monitoring
quenching of DNA cleavage in the presence of alternative
Role of the Reductant. Generation of the dicopper(I) form
of [CuII (D1)(H2O)2](ClO4)4 (2) appears to be crucial for
2
formation of the reactive intermediate based on the lack of strand
scission when either 3-mercaptopropionic acid (MPA) or 2 is
absent from the reaction mixture. To determine if MPA played
a unique role in strand scission either as a reductant or through
interactions with 2, the extent and pattern of DNA scission were
assessed by testing alternative reductants, glutathione (GSH),
and dithiothreitol (DTT). The efficiency of the three reductants
to promote scission of the Watson strand of IIa varied according
to MPA > GSH > DTT (Figure 8). The total cleavage of IIa
(Watson strands) by 1 µM 2 and 5 mM GSH was approximately
3-fold less than with an equivalent concentration of MPA.
Reaction in the presence of 5 mM DTT resulted in even less
total cleavage with only 1% of the Watson strand degraded.
However, in the presence of either GSH or DTT, the amount
of cleavage directed at A21 was relatively consistent at 62 (
6% and exhibited the same selectivity observed in the presence
of MPA.30 Despite the differences in the overall reactivity, the
similarities in the cleavage patterns suggest that a common
intermediate is formed in each case resulting from reduction of
Cu(II) complex to Cu(I).38,39 The observed differences in total
cleavage are likely related to the reduction potentials of the thiols
and their respective abilities to generate Cu(I) from Cu(II).
Role of Dioxygen. The expected Cu(I) species generated by
the thiols has the potential to activate O2 by forming intermedi-
ates such as superoxide anion, peroxide, and hydroxyl radical.
To directly assess the importance of dioxygen in specific and
•
H-atom donors, which would scavenge radicals (such as OH)
in solution. To this end, standard radical scavengers were added
to the reaction of IIa and [CuII (D1)(H2O)2](ClO4)4 (2) prior to
2
initiation with MPA. The individual addition of ethanol,
D-mannitol, and tert-butyl alcohol (10 mM) was found to have
no effect on sequence-neutral cleavage, which remained constant
at 9 ( 1% (Figure 9).30 Selective cleavage at A21 represented
52 ( 3% of the total yield in the presence or absence of each
scavenger. Thus, oxidation of the deoxyribose within the DNA
backbone leading to both specific and nonspecific strand scission
by 2 is not mediated by a diffusible, or at least trappable, radical
species. Additional cleavage effected by subsequent treatment
with piperidine also remained unaffected by these H-atom
donors. The lack of diffusible intermediates is consistent with
the ability of 2 to direct cleavage with single nucleotide
resolution.
Target selective reaction mediated by distamycin or oligo-
nucleotide conjugates of EDTA‚Fe(II) generates a range of
cleavage products 7-8 bases on either side of the binding site
of their recognition element.40,41 This is most typical of localized
generation of a diffusible hydroxyl radical species. When such
a complex is not linked to a DNA recognition element, then
scission occurs randomly at all four nucleotides and is inhibited
by addition of radical scavengers. In contrast, [Cu(OP)2]2+
,
which is believed to react through a coordinated radical species
generated close to the cleavage site, cannot be quenched by
(38) John, D. C. A.; Douglas, K. T. Biochem. J. 1993, 289, 463-468.
(39) Veal, J. M.; Merchant, K.; Rill, R. L. Nucleic Acids Res. 1991, 19, 3383-
3388.
(40) Taylor, J. S.; Schultz, P. G.; Dervan, P. B. Tetrahedron 1984, 40, 457-
465.
(41) Dreyer, G. B.; Dervan, P. B. Proc. Natl. Acad. Sci. 1985, 82, 968-972.
9
6016 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Efficient Strand Scission of DNA by a Dicopper(II) Complex
A R T I C L E S
mannitol or alcohols.42 Specific and nonspecific cleavage of
DNA by the trinuclear complex, [CuII (L)(H2O)3(NO3)2](NO3)4‚
3
5H2O (1), also was not subject to inhibition by the hydrogen
atom donors ethanol, D-mannitol, and tert-butyl alcohol.15 In
addition, specific strand scission was directed at one nucleotide
suggestive of a nondiffusible intermediate. These results indicate
that the reactive species that effects selective cleavage by [CuII -
2
(D1)(H2O)2](ClO4)4 (2) is likely formed close to the site of
reaction and not released from the complex for diffusion to the
cleavage site. The reactive intermediate in strand scission
mediated by 2 may be related to the known µ-peroxo dicopper-
(II) (Cu-O-O-Cu) intermediate formed when the dicopper-
(I) complex of 2 reacts with dioxygen.17 A similar end-on
µ-O-O dimer was proposed as the reactive intermediate in the
nucleobase specific oxidation of a 514-bp fragment of pBR322
by a mononuclear Co(II)‚Lys-Gly-His complex.43 The formation
of the peroxo species in this example was postulated from the
known dioxygen chemistry of the cobalt-peptide and an
observed dependence in specific oxidation on the concentration
of the cobalt-peroxo dimer.
Figure 10. Densitometer traces demonstrating differential reactivity of 2,
3, and 4 with IIa and 5 mM MPA: (A) 1 µM 2, (B) 1 µM 4, and (C)
2 µM 3.
results of these experiments demonstrate that multiple metal ions
within the same complex are essential for efficient cleavage of
DNA.
The lack of activity exhibited by [CuII(TMPA)(H2O)](ClO4)2
(3) vs [CuII (D1)(H2O)2](ClO4)4 (2) may be related to their
2
different ability in activating dioxygen. In nitrile solvent both
complexes initially form a short-lived superoxo intermediate
that is observable only by stopped-flow kinetics.45 One Cu(II)-
Reactivity of [CuII(TMPA)(H2O)](ClO4)2 (3) or [CuII
-
2
(DO)(Cl)2](ClO4)2 (4) with IIa. The dinuclear metal complex
-
[CuII (D1)(H2O)2](ClO4)4 (2) likely controlled both its recogni-
O2 can then react with an additional Cu(I) ligand complex to
2
form an end-on µ-peroxo (Cu-O22--Cu). Formation of the
peroxo species by the Cu(I) form of 3 requires that two
complexes come together yielding an intermolecular µ-peroxo
dimer.45-47 The tethering of two copper binding units in the
dicopper(I) form of 2 gives rise to an intramolecular peroxo
that is entropically stabilized.17 Dioxygen and reductant de-
pendent cleavage of DNA by 2 clearly indicates that the
dinuclear complex is reduced to its dicopper(I) form, which
subsequently reacts with O2 to generate the reactive species
responsible for DNA oxidation. The contrasting reactivity of 2
and 3 with DNA cannot by ascribed to differences in their
reduction potentials since they are nearly identical.17 The
absence of nonspecific background cleavage of DNA when 3
is reacted with IIa in the presence of MPA and O2 suggests
difficulties in forming the appropriate O2-derived intermediate
and is not primarily related to a lack of specific DNA
recognition. Instead, the rate of forming the reactive species of
3 may by too slow to mediate strand scission during the standard
incubation time. Alternatively, the mononuclear complex 3
might interact with DNA (i.e. by binding to the biopolymer) in
such a manner that it never comes into contact with another
copper complex to form a reactive dicopper intermediate. The
lack of activity by 3 implies that the dinuclearity of 2 as well
as its metal coordination environment and dioxygen reactivity
are crucial for efficient strand scission.
tion and reactivity with DNA by associating with the frayed
duplex and generating the intermediate responsible for site
specific strand scission. This dual role is intimately linked to
the influence of the ligand in defining the coordination environ-
ment about each copper, which in turn determines the orientation
and type of intermediates and their rate of formation. Since at
least one of the coppers in the dinuclear complex 2 may
coordinate to the N7 of G20, the coordination environment
surrounding the metals could also be crucial for recognition of
the helix/coil junction in IIa. Each copper(II) ion in 2 is chelated
through three pyridyl nitrogens and a tertiary amine yielding
the trigonal bipyramidal geometry that is characteristic for this
tripodal tetradentate motif.26,44 The fifth coordination site could
alternatively be occupied by water, hydroxide ion, or substrate.22
The mononuclear compound, [CuII(TMPA)(H2O)](ClO4)4 (3),
possesses a nearly identical coordination sphere to that of 2 and
its Cu(I) and Cu(II) complexes have been characterized by X-ray
crystallography, EPR, electrochemistry, and UV-vis spectros-
copy.17,19 The effects of the coordination environment and
nuclearity (i.e. di- vs mononuclear) on DNA cleavage were
assessed by comparing nonspecific and specific strand scission
of IIa by 2 and 3. When 3 was reacted with IIa under the
standard reaction conditions, no degradation of the Watson
strand was observed (Figure 10).30 This lack of reactivity
precluded any observation of specific strand scission. The Crick
strand of IIa was also not cleaved by 3 under standard reaction
conditions.30 The absence of strand scission was not related to
differences in the amount of metal ion present since equivalent
copper ion concentrations of 2 and 3 were used (1 and 2 µM,
respectively). Reacting Cu(ClO4)2‚6H2O with IIa under the
standard reaction conditions also failed to produce either specific
or nonspecific cleavage of the Watson and Crick strands.30 The
Another dinuclear complex, [CuII (DO)(Cl)2](ClO4)2 (4), was
2
tested to correlate DNA cleavage with O2 reactivity. The copper
ions in 4 share similar coordination environments to those in
[CuII (D1)(H2O)2](ClO4)4 (2) and [CuII(TMPA)(H2O)](ClO4)2
2
(3).48 Additionally, the dicopper(I) form of 4 yields an intramo-
lecular µ-peroxo species in the presence of O2 with a rate of
formation faster than that formed by 2.21 The 2-atom linker of
(45) Karlin, K. D.; Kaderli, S.; Zuberbu¨hler, A. D. Acc. Chem. Res. 1997, 30,
139-147.
(42) Pope, L. M.; Reich, K. A.; Graham, D. R.; Sigman, D. S. J. Biol. Chem.
1982, 257, 12121-12128.
(43) Ananias, D. C.; Long, E. C. J. Am. Chem. Soc. 2000, 122, 10460-10461.
(44) Kokoszka, G.; Karlin, K. D.; Padula, F.; Baranowski, J.; Goldstein, C. Inorg.
Chem. 1984, 23, 4378-4380.
(46) Karlin, K. D.; Wei, N.; Jung, B.; Kaderli, S.; Zuberbu¨hler, A. D. J. Am.
Chem. Soc. 1991, 113, 5868-5870.
(47) Karlin, K. D.; Wei, N.; Jung, B.; Kaderli, S.; Niklaus, P.; Zuberbu¨hler, A.
D. J. Am. Chem. Soc. 1993, 115, 9506-9514.
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 6017

A R T I C L E S
Humphreys et al.
2 produces a considerable amount of strain in the peroxo
complex and contributes to enthalpic destabilization of its
intermediate. This effect is avoided by addition of an oxygen
atom to the linker of 3. If a dinuclear complex possessing similar
properties with O2 is needed for strand scission, then 4 should
promote cleavage of IIa. Indeed, sequence-neutral cleavage of
the DNA was detected when 4 was reacted with the Watson
strand of IIa under the standard conditions (Figure 10).30 The
sequence neutral reaction was equally efficient for the Watson
and Crick strands and similar to that formed by 2. However,
highly selective strand scission did not occur at any site on IIa
and consequently, the amount of total cleavage on the Watson
strand was about 3-fold less than that generated with 2. The
cleavage pattern generated by 4 did show a slight enhancement
in cleavage at A21 and T22 near the helix/coil junction, suggesting
a weak interaction between the DNA and this alternative
dinuclear complex may be retained.30 The reactivity of both 2
and 4 again demonstrates that multinuclearity is essential in
mediating strand scission. However, site selectivity depends on
more subtle features related to the recognition of DNA and the
structure of the copper complexes, topics of continuing inves-
tigation.
Reactivity of [Cu(OP)2]2+ with IIa. The unique recognition
and efficiency of [CuII (D1)(H2O)2](ClO4)4 (2) is also highlighted
2
by its comparison to the activity of [Cu(OP)2]2+. At equivalent
copper ion concentrations both 2 and [Cu(OP)2]2+ cleaved the
Watson strand of IIa under the standard conditions, but their
dependence on concentration differed dramatically. At 0.5 µM
[Cu2+], a total of 2% of the strand was cleaved by 2 whereas
no detectable cleavage was evident for [Cu(OP)2]2+. When
[Cu2+] was doubled to 1.0 µM, total degradation by 2 doubled
to 4% while degradation by [Cu(OP)2]2+ remained below
detectable levels (Figure 11A). However, after the [Cu2+] was
increased to 2 µM, [Cu(OP)2]2+ became much more reactive
than 2, cleaving 50% of the oligonucleotide in contrast to 9%
cleaved by the dinuclear compound (Figure 11B). However,
[Cu(OP)2]2+ was not capable of mediating specific strand
scission at A21 under any of the concentrations tested. The
phenanthroline-copper complex demonstrated a preference for
oxidation of double-helical 5′ GT sequences consistent with prior
reports.49,50 Cleavage of the Crick strand of IIa by 2 µM [Cu-
(OP)2]2+ was similar to that of the Watson strand of IIa and
again a threshold [Cu2+] of 2 µM was necessary for reaction.30
There is no cleavage on either strand by [Cu(OP)2]2+ at
concentrations below 2 µM. Earlier studies had shown that
strand scission by [Cu(OP)2]2+ appeared to require 2 equiv of
the complex, one bound to the minor groove and one free of
DNA.51 Consequently, if all of the 1,10-phenanthroline-copper
complex is interacting with the biopolymer, then none is left to
form the dimeric intermediate and reaction is prevented.
However, as the complex concentration increases, some complex
will remain unbound and free to interact with the DNA-bound
Figure 11. Densitometer traces demonstrating differential reactivity of 2
and [Cu(OP)2]2+ with IIa and 5 mM MPA: (A) 1 µM [Cu2+] and (B) 2
µM [Cu2+].
complex. If a related dicopper complex must form around the
site of strand scission induced by 2, then no intermolecular
assembly of copper is required. The ligand maintains the two
coppers intramolecularly to promote efficient oxidation of DNA
at low copper concentrations.
Proposed Model of Reactivity and Recognition by [CuII
-
2
(D1)(H2O)2](ClO4)4 (2). The key to understanding the differ-
ences between [CuII (D1)(H2O)2](ClO4)4 (2) and [CuII (DO)-
2
2
(Cl)2](ClO4)2 (4) in cleavage of DNA may lie in understanding
the details of their interactions with dioxygen. Formation of a
µ-peroxo intermediate [Cu-O22--Cu] from the dicopper(I)
form of 2 was shown to be destabilized enthalpically relative
to the same intermediate formed between two monomeric copper
complexes, i.e., in the reaction of [CuII(TMPA)(H2O)](ClO4)2
(3) with dioxygen. This destabilization is caused by strain in
the peroxo intermediate manifested in the short two-atom bridge
connecting the copper binding moieties.17,52 In contrast, the
three-atom ether bridge of 4 participates in an intramolecular
µ-peroxo species that is both entropically and enthalpically
stabilized when the dicopper(I) complex is reacted with dioxy-
gen.17,21,52 Once formed, the intermediates of 2 and 4 likely
initiate H-atom abstraction from the deoxyribose moiety of DNA
and lead to direct strand scission. The slight enhancement of
cleavage at the junction of IIa produced by 4 suggests that this
analogue possessed similar recognition properties to 2.
Although the µ-peroxo intermediate of 4 is thermodynami-
cally more stable than that of 2,21 their similar abilities to oxidize
DNA nonspecifically suggest that neither the rate of formation
nor the lifetime of the intermediate are responsible for selective
cleavage. Molecular mechanics modeling demonstrates that one
of the primary differences between the µ-peroxo intermediates
(48) Spectroscopic studies revealed similar EPR spectra for 2, 3, and 4 typical
of reverse tetragonal coordination. Chloride coordination in 4 produced a
distinct UV-vis spectrum that compared favorably with that of a chloride
coordinated form of 3. Since both of these relate to the electronics
surrounding the metal as affected by the ligand environment, the compounds
are then presumed to have closely related coordination spheres. This is
also supported by the similar bond distances and angles from X-ray
structures of each dicopper(II) complex. Data to be published elsewhere.
(49) Yoon, C.; Kuwabara, M. D.; Spassky, A.; Sigman, D. S. Biochemistry 1990,
29, 2116-2121.
(50) Veal, J. M.; Rill, R. L. Biochemistry 1989, 28, 3243-3250.
(51) Goldstein, S.; Czapski, G. J. Am. Chem. Soc. 1986, 108, 2244-2250.
(52) Comba, P.; Hilfenhaus, P.; Karlin, K. D. Inorg. Chem. 1997, 36, 2309-
2313.
9
6018 J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002

Efficient Strand Scission of DNA by a Dicopper(II) Complex
A R T I C L E S
of 2 and 4 is the tilt of the two equatorial planes encompassing
the metal ion and the three pyridyl nitrogens in relation to one
another.52 All other constraints, including the calculated Cu‚‚‚
Cu distances, Cu-O-O-Cu bond angles, and torsion angles
around the peroxide bond, are nearly identical for both 2 and
4. Therefore, the difference in the ability of the two µ-peroxo
complexes to promote selective cleavage may be the orientation
of the O-O moiety. Further investigations will be needed to
ascertain the identity of and the exact effect of copper-dioxygen
intermediates in mediating specific cleavage.
Generation of the reactive intermediate responsible for
specific and sequence-neutral strand scission requires both a
reductant and dioxygen, implying the formation of an intermedi-
ate derived from the activation of O2 by Cu(I). Addition of
radical scavengers to the standard reaction failed to inhibit
cleavage by [CuII (D1)(H2O)2](ClO4)4 (2), demonstrating that
2
the intermediate responsible for selective and nonspecific strand
scission is not a freely diffusible radical. At equivalent copper
ion concentrations, 2 is more efficient than its mononuclear
analogue, [CuII(TMPA)(H2O)](ClO4)2 (3), and [Cu(OP)2]2+. The
dinuclear analogue, [CuII (DO)(Cl)2](ClO4)2 (4), exhibits a
2
Conclusion
similar yield of sequence-neutral DNA cleavage, but only a very
a weak preference for reaction at a helix/coil junction. Site-
specific strand scission, best illustrated by 2, may derive from
the formation and orientation of its µ-peroxo derivative. Studies
have now been initiated to identify the subtle requirements of
target recognition and dioxygen activation that may reveal
fundamental information about copper and related complexes
as well as provide tools for nucleic acid analysis in vitro and in
vivo.
A dicopper(II) complex, [CuII (D1)(H2O)2](ClO4)4 (2), has
2
been shown to exhibit efficient and specific strand scission at
junctions between single- and double-stranded DNA. Although
the general requirements of a helix/coil junction are similar to
those of previously studied trinuclear complex [CuII (L)(H2O)3-
3
(NO3)2](NO3)4‚5 H2O (1)14,15 studied previously, the exact
cleavage sites and nucleotide sequence requirements of these
complexes are quite different. Specific strand scission by 2
occurs at a site that is two nucleotides displaced from the
junction on the 3′ overhang of a frayed duplex structure.
Cleavage is not dependent on the identity of the base at the
cleavage site, but does require a guanine on the 3′ overhang
directly adjacent to the central duplex and an adenine on the 5′
overhang in the same position. This is in contrast to 1, for which
both sequence requirements were found on the 5′ overhang
opposite the cleavage site on the 3′ overhang. Again, these
differing recognition elements may reflect the available geom-
etries of the dicopper(II)-peroxo intermediate and suggest that
orientation of the intermediate may be more important than its
thermodynamic stability.
Acknowledgment. This work was supported by the NIH
(GM28962 to K.D.K. and GM47531 to S.E.R.).
Supporting Information Available: Autoradiograms of se-
quence variants of IIa in reactions with complex 2, piperidine
treated samples, strand scission in the presence of different
reductants and standard radical scavengers, and cleavage medi-
ated by complexes 3, 4, and [Cu(OP)2]2+ (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA020039Z
9
J. AM. CHEM. SOC. VOL. 124, NO. 21, 2002 6019