Anal. Chem. 2003, 75, 1851-1860
Digital Simulation of Catalytic Cyclic Voltammograms for Oxidation of DNA by a Heterobimetallic Dimer: Effects of DNA Binding and Mass Transport Rebecca C. Holmberg and H. Holden Thorp*
Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290
The electrocatalytic oxidation of DNA by a heterodimer, [(bpy)2Ru(tpphz)Os(bpy)2]4+ (tpphz: tetrapyrido[3,2-a: 2′,3′-c:3′′,2′′-h:2′′′,3′′′-j]phenazine) (1), was studied using cyclic voltammetry with digital simulation. This dimer was chosen because the Ru(III/II) couple (E1/2 ) 1.09 V vs Ag/AgCl) is capable of catalyzing guanine oxidation while the Os(III/II) couple (E1/2 ) 0.63 V) provides a convenient reporter on the binding and mass transport of the complex, which can then be determined in the same voltammetric sweep as the electrocatalysis. Proper description of the electrochemical response required careful measurement of the binding constant of 1 to herring testes (HT) DNA, which was (2.0 ( 0.1) × 104 M-1 by both absorption titration and normal pulse voltammetry. Thermal denaturation experiments were consistent with a nonintercalative binding mode and gave a ∆Tm of only (2.4 ( 0.5) °C. The minor groove binder distamycin did not displace 1 from HT DNA, suggesting that the complex binds in the major groove. As expected, acquisition of the cyclic voltammogram of 1 in the presence of DNA produced catalytic current for the Ru(III/II) couple, while a suppression of current was observed for the Os(III/II) couple. Although the catalytic current for the Ru(III/II) couple initially appeared as a current enhancement, higher concentrations suppressed the catalytic wave as a result of the slower mass transport of the DNA-bound complex. The binding studies were used to create a model for digital simulation that reproduced the behavior of 1 with DNA and gave rate constants that were independent of DNA concentration. The apparent second-order rate constant at 25 mV/s for oxidation of guanine in HT DNA (av 1000 bp, 25% guanine) by 1 was 3 ( 1 × 104 M-1 s-1; similar values were obtained for a 200-bp fragment (7 ( 3 × 104 M-1 s-1) and a 435-bp fragment (8 ( 2 × 104 M-1 s-1). As observed in previous studies of these reactions, biphasic kinetics in the catalytic reaction led to a dependence of the rate constant determined by simulation on the sweep rate. Increasing the sweep rate led to a systematic increase in the simulated rate constant, 10.1021/ac0204653 CCC: $25.00 Published on Web 03/21/2003
© 2003 American Chemical Society
consistent with a fast phase of the homogeneous catalytic reaction. The chemical reactivity of nucleobases is under intense study because these reactions play a role in disease and offer an entry to methods for sensing DNA quantity or structure.1-6 Electrocatalytic reactions in which the metal complex Ru(bpy)32+ (bpy ) 2,2′bipyridine) oxidizes guanine upon electrochemical oxidation to Ru(III) provide a sensitive means for studying the chemical reactivity and mass transport of polynucleotides using cyclic voltammetry.7 When the cyclic voltammogram of Ru(bpy)32+ is acquired in the presence of DNA, large catalytic waves are observed as a result of reduction of electrogenerated Ru(III) by guanine in the DNA.8-11 The rate constants for guanine-Ru(III) electron transfer can be obtained by digital simulation of the resulting electrocatalytic waves and confirmed independently by stopped-flow spectrophotometry using chemically generated Ru(III). 7 The redox potentials of guanine and Ru(bpy)33+/2+ are similar (∼1.05 V, all potentials vs Ag/AgCl), resulting in a nearly thermoneutral electron transfer. Determining the rate constants for oxidation of DNA by Ru(bpy)32+ requires an understanding of the binding of the metal complex to the DNA polyanion, which occurs via displacement of buffer cations by the metal complex.12 For the simple Ru(bpy)32+ system, the rate of oxidation is therefore dependent on the solution ionic strength. Under high ionic strength conditions (50 mM phosphate with 800 mM NaCl), the binding of the metal complex (1) Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1235-1248. (2) Burrows, C. J.; Muller, J. G. Chem. Rev. 1998, 98, 1109-1151. (3) Beckman, K. B.; Ames, B. N. J. Biol. Chem. 1997, 272, 19633-19636. (4) Steenken, S.; Telo, J. P.; Novais, H. M.; Candeias, L. P. J. Am. Chem. Soc. 1992, 114, 4701-4709. (5) Zhou, L. P.; Rusling, J. F. Anal. Chem. 2001, 73, 4780-4786. (6) Ropp, P. A.; Thorp, H. H. Chem. Biol. 1999, 6, 599-605. (7) Johnston, D. H.; Glasgow, K. C.; Thorp, H. H. J. Am. Chem. Soc. 1995, 117, 8933-8938. (8) Sistare, M. F.; Holmberg, R. C.; Thorp, H. H. J. Phys. Chem. B 1999, 103, 10718-10728. (9) Johnston, D. H.; Thorp, H. H. J. Phys. Chem. 1996, 100, 13837-13843. (10) Szalai, V. A.; Thorp, H. H. J. Phys. Chem. B 2000, 104, 6851-6859. (11) Szalai, V. A.; Jayawickamarajah, J.; Thorp, H. H. J. Phys. Chem. B 2002, 106, 709-716. (12) Welch, T. W.; Corbett, A. H.; Thorp, H. H. J. Phys. Chem. 1995, 99, 1175711763.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003 1851
Figure 1. Mechanism for oxidation of DNA by mononuclear metal complexes under conditions where binding of the metal complex to DNA must be considered.
to the polyanion can be neglected,13-15 so the reaction is secondorder and proceeds via an EC-type mechanism that has been described previously.9 At low ionic strength (50 mM phosphate), binding of the metal complex to the DNA backbone must be considered, and the reaction mechanism is described by the square scheme model shown in Figure 1.9 This kinetic model was tested using Os(bpy)32+, which binds to DNA but is not a powerful enough oxidant to undergo electron transfer with guanine upon oxidation to Os(III).12 For Os(bpy)33+/2+, addition of DNA causes a decrease in the measured current because of slower diffusion of the bound form, as compared to the free form. Analysis of the current due to Os(bpy)33+/2+ in the presence of DNA, therefore, allows determination of the binding constant of the metal complex to the nucleic acid.16 Because Ru(bpy)32+ and Os(bpy)32+ are nearly isostructural, these parameters can then be used to describe the binding reactions in the kinetic scheme for Ru(bpy)32+mediated electrocatalysis, shown in Figure 1. The scheme in Figure 1 can be used to simulate voltammograms and obtain rate constants for the homogeneous oxidation of guanine by Ru(bpy)33+.8 These simulations provide rate constants that are independent of DNA concentration but do give a systematic increase in the simulated rate constant with increasing sweep rate; this systematic increase results from biphasic kinetics in the guanine-Ru electron transfer, which is observed in chronoamperometry8,9 and is typical of metal-DNA systems.17,18 Additional support for this interpretation is provided in simulations described here. Although we have analyzed the binding and DNA redox chemistry of Os(bpy)32+ and Ru(bpy)32+ in detail as separate complexes, we have not explored the case in which the two complexes are linked together in a heterobinuclear complex. In such a system, the Os(III/II) couple could act as a reporter on binding of the dimer to DNA in the same complex in which the Ru(III/II) couple catalyzes guanine oxidation. Considerable effort has been directed at studying binuclear metal complexes and their interactions with polynucleotide phosphates.19-26 For example, our group has studied Pt2(P2O5H22-)44- as a chemical nuclease for (13) Manning, G. S. Q. Rev. Biophys. 1978, 11, 179-246. (14) Manning, G. S. Acc. Chem. Res. 1979, 12, 443-449. (15) Record, M. T.; Anderson, C. F.; Lohman, T. M. Q. Rev. Biophys. 1978, 11, 103-178. (16) Welch, T. W.; Thorp, H. H. J. Phys. Chem. 1996, 100, 13829-13836. (17) Kumar, C. V.; Barton, J. K.; Turro, N. J. J. Am. Chem. Soc. 1985, 107, 55185523. (18) Pyle, A. M.; Rehmann, J. P.; Meshoyrer, R.; Kumar, C. V.; Turro, N. J.; Barton, J. K. J. Am. Chem. Soc. 1989, 111, 3051-3058. (19) O’Reilly, F. M.; Kelly, J. M. J. Phys. Chem. B 2000, 104, 7206-7213. (20) Takenaka, S.; Uto, Y.; Saita, H.; Yokoyama, M.; Kondo, H.; Wilson, W. D. Chem. Commun. 1998, 1111-1112.
1852
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
cleaving DNA strands through hydrogen abstraction.27-29 Kelly et al. have examined the photophysical properties of two ruthenium centers covalently linked by a polyalkyl chain bound to DNA.19 Takenaka has synthesized a naphthalene diimide threading intercalator carrying two ferrocenyl moieties that can electrochemically distinguish between single- and double-stranded DNA.20 Others have expanded on this threading concept to include ruthenium complexes for intramolecular photoinduced electron transfer in the presence of DNA and sequence specificity of stereoisomeric binding, respectively.21,22 Similarly, Hannon recently studied the binding effects of a di-iron complex that causes condensation of the DNA upon binding to the major groove.23,24 Because our goal was to study a dinuclear system in which one metal center was a reporter on binding of the complex to DNA and the other metal center was a catalyst for guanine oxidation, we synthesized the complex [(bpy)2Ru(tpphz)Os(bpy)2]4+ (tpphz: tetrapyrido[3,2-a:2′,3′-c:3′′,2′′-h:2′′′,3′′′-j]phenazine) (1).30 According to the measured redox potentials for 1 in
acetonitrile,30 the Ru(III) center should be capable of oxidizing guanine, but the Os(III) center should not. Combining Os and Ru into this single metal complex, therefore, allows simultaneous study of guanine oxidation via the Ru(III/II) couple and DNA binding via the Os(III/II) couple. In this report, we describe the interactions of 1 with nucleic acids and the associated electrocatalysis of guanine oxidation. The binuclear complex produces distinctive catalytic voltammograms that we quantitatively and qualitatively describe by digital simulation with appropriate treatment of binding and mass transport. EXPERIMENTAL SECTION Materials. K2[OsCl6], Ru(bpy)2Cl2‚2H2O, 20% oleum, hydroxylamine hydrochloride, 55% hydrazine hydrate, 1,10-phenanthroline (21) Dixon, D. W.; Thornton, N. B.; Steullet, V.; Netzel, T. Inorg. Chem. 1999, 38, 5526-5534. (22) Onfelt, B.; Lincoln, P.; Norde´n, B. J. Am. Chem. Soc. 1999, 121, 1084610847. (23) Hannon, M. M., V; Prieto, M. J.; Moldrheim, E.; Sletten, E.; Meistermann, I.; Isaac, C. J.; Sanders, K. J.; Rodger, A.; Parkinson, A.; Vidler, D. S.; Haworth, I. S. Angew. Chem., Int. Ed. 2001, 40, 879-884. (24) Rodger, A.; Sanders, K. J.; Hannon, M. J.; Meistermann, I.; Parkinson, A.; Vidler, D. S.; Haworth, I. S. Chirality 2000, 12, 221-236. (25) Heater, S. J.; Carrano, M. W.; Rains, D.; Walter, R. B.; Ji, D.; Yan, Q.; Czernuszewicz, R. S.; Carrano, C. J. Inorg. Chem. 2000, 39, 3881-3889. (26) Milkevitch, M.; Storrie, H.; Brauns, E.; Brewer, K. J.; Shirley, B. W. Inorg. Chem. 1997, 36, 4534-4538. (27) Kalsbeck, W. A.; Gingell, D. M.; Malinsky, J. E.; Thorp, H. H. Inorg. Chem. 1994, 33, 3313-3316. (28) Breiner, K. M.; Daugherty, M. A.; Oas, T. G.; Thorp, H. H. J. Am. Chem. Soc. 1995, 117, 11673-11679. (29) Carter, P. J.; Breiner, K. M.; Thorp, H. H. Biochemistry 1998, 37, 1373613743. (30) Bolger, J.; Gourdon, A.; Ishow, E.; Launay, J. P. Inorg. Chem. 1996, 35, 2937-2944.
monohydrate, ammonium hexafluorophosphate, sodium dithionite, and tetrabutylammonium chloride were all obtained from Aldrich. 5-Nitro-1,10-phenanthroline was acquired from Alfa, and 2,2′bipyridine was purchased from Lancaster Synthesis, Inc. Herring testes DNA was obtained from Sigma and used without further purification. Stock solutions were sheared by sonication (30 min) to achieve homogeneous strand length. Water was purified with a MilliQ purification system (Millipore). Buffer salts were purchased from Mallinckrodt. Gel electrophoresis reagents were purchased from BioRad. Synthesis of 1. [(bpy)2Ru(tpphz)Os(bpy)2](PF6)4 (1) was synthesized and purified as previously described by Bolger et al.30 The purity was determined by elemental analysis. Anal. Calcd for (RuC64H44N14Os)(PF6)4‚H2O: C, 40.49; H, 2.44; N, 10.32%. Found: C, 40.49; H, 2.61; N, 10.42%. The final product was converted to the chloride salt by dissolving the complex in acetone and adding tetrabutylammonium chloride. The isolated chloride complex was then recrystallized by dissolving it in acetonitrile and pipetting a double excess of ether on top of the acetonitrile solution. After a few days, the crystals were filtered and washed with ether. Polymerase Chain Reaction. A 200-base-pair fragment of the dacA gene (PBP5) from E. coli (GenBank accession number D90703) was amplified by polymerase chain reaction (PCR) in a 24-well Hybaid Thermocycler using forward primer (5′-CATGAATACCATTTTTTCCGCTCG-3′) and reverse primer (5′-CACGAGCGGCTTGTCTTGCGTC-3′). Reaction tubes of 100 µL each contained 0.5 µL of template (2.4 ng/µL), 2 µL of each primer (20 µM), 2 µL of dNTPs (10 µM), 4 µL of MgCl2 (50 mM), 10 µL of 10× PCR buffer, and 1 µL of Taq polymerase (Life Technologies, 5 U/µL). The thermocycler was programmed for 1 cycle at 94 °C for 3 min, 40 cycles at 94 °C for 1 min, 63 °C for 1 min and 72 °C for 30 s, and 1 cycle at 72 °C for 5 min. Following amplification, a representative sample from each batch was run at 100 V on a 2% agarose gel with ethidium bromide indicator to visualize products against a 1-kb DNA ladder (Gibco BRL). Purification of PCR product was carried out using the Qiaquick (Qiagen) purification kit, and the resulting eluted solution was ethanol-precipitated and resuspended in 50 mM sodium phosphate buffer. The concentration was determined spectrophotometrically. For the 200-bp fragment containing inosine (200-I), the same primers with inosine replacing all guanosines were used, and dITP was used in place of dGTP. The cycle temperature for inosinecontaining primer annealing was 37 °C, and the annealing time was increased to 1.5 min. Inosine replacement resulted in a much lower amplification yield. A 435-bp PCR product was prepared using the same protocol using forward primer (5′-GGC TGT GCC CGC TGC AAG GGG CCA- 3′) and reverse primer (5′-GCA GCC AGC AAA CTC CTG GAT ATT- 3′). Binding Measurements. Solution concentrations of the metal complex or DNA were determined by spectrophotometry using a Hewlett-Packard HP 8452 diode array spectrophotometer. Extinction coefficients used with absorbance measurements in water were 370 ) 46000 M-1cm-1 for 1 and 260 ) 6600 M-1cm-1 for herring-testes (HT) DNA and PCR products.12 All absorption titrations were performed according to a previously described protocol.31 The decrease in absorbance at 370 nm was monitored upon the addition of increasing amounts of herring testes DNA
Table 1. DNA Strands and Rate Constants at 25 mV/s DNA fragment
length (bp)
% guanine
diffusion coeff (cm2/s)
kf (M-1 s-1)a
HT 435-G 200-G 200-I 39 35
1000 435 200 200 39 35
25.0 30.4 25.0 0 5.1 5.7
2 × 10-7 3 × 10-7 4 × 10-7 4 × 10-7 7.5 × 10-7 8 × 10-7
3 (( 1) × 104 8 (( 2) × 104 7 (( 3) × 104 3 (( 3) × 103
a All rate constants are an average from three different DNA concentrations.
in 0 mM NaCl and 400 mM and 800 mM NaCl sodium phosphate solutions. Emission spectroscopy experiments were performed on a Jobin Yvon-SPEX FluoroMax spectrophotometer (Instruments S. A., Inc.). Glass melting point tubes contained 50-µL solutions of 50 µM metal complex, 25 µM or 250 µM [G] DNA, 800 mM NaCl, and 50 mM sodium phosphate. DNA fragments studied, listed in Table 1, included HT (av 1000 bp), PCR products of 435- and 200bp length (435 and 200-G), and two chemically synthesized oligonucleotides of 39- and 35-bp lengths. The samples were excited at 440 nm, and the emission intensity was scanned from 550 to 750 nm (λem at 618 nm). Competitive binding experiments were performed with 50 µM 1, 500 µM HT DNA, and increasing amounts of distamycin A (Sigma) (0-500 µM). Control experiments with 1 and distamycin A and also distamycin A and HT were performed at the same concentrations. Thermal Denaturation. Thermal denaturation experiments were performed on a Cary 300 Bio UV-vis spectrophotometer. A 24-bp double-stranded oligonucleotide (10 µM) was prehybridized by heating to 95 °C for 5 min and slowly cooled to room temperature over 2 h. Varying amounts of complex 1 (0, 5, and 10 µM) in 800 mM NaCl and 50 mM sodium phosphate, pH 5.2, were added to the hybridized DNA solution and placed in a quartz cuvette. Using the thermal melting program, the temperature of the cell containing the cuvette was ramped from 25 °C to 95 °C and from 95 °C to 25 °C at 0.5 °C/min, and the absorbance at 260 nm was measured every 0.5 °C. The data were smoothed at 0.5 °C intervals with a filter of 5, and the derivative of the resulting curve was taken at a data interval of 0.5 °C. Molecular Modeling. The DNA strand in Figure 5A was taken from an X-ray structure of a crystal structure of the indicated DNA sequence with two intercalated nogalamycin molecules.32 This structure was imported into Sybyl, and the intercalated molecules were removed. For Figure 5B, Sybyl was used to generate a 10bp duplex of B-form DNA. Complex 1 was minimized (MM2 energy minimization) in CS ChemBats3D Pro and imported into Sybyl for comparison with the double-helix molecule. Molecules were manipulated spatially to illustrate the possible binding positions of complex 1 to DNA. Electrochemistry. ITO (tin-doped indium oxide) electrodes were obtained from Delta Technologies, LTD (Stilwater, MN). All ITO electrodes were cleaned in a four-step sonication process (31) Sitlani, A.; Long, E. C.; Pyle, A. M.; Barton, J. K. J. Am. Chem. Soc. 1992, 114, 2303-2312. (32) Smith, C. K.; Davies, G. J.; Dodson, E. J.; Moore, M. H. Biochemistry 1995, 34, 415-425.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1853
involving an aqueous solution of Alconox, 2-propanol, and two washes with Milli-Q water (6 min each). Electrodes were then dried in an oven (90 °C) for 5 min. The freshly cleaned electrodes were used within 24 h. Glassy carbon disk electrodes (Bioanalytical Systems) for nonaqueous electrochemistry were cleaned first using diamond paste (1 µm) and methanol and then polished with Al2O3 in H2O on a felt polishing pad according to standard procedures given by the manufacturer. All electrochemistry experiments were performed on an EG&G PAR 273A potentiostat. The electrochemical cell assembly using an ITO working electrode (area ) 0.32 or 0.15 cm2, Delta Technologies), Ag/AgCl pseudoreference, and Pt counter electrode has been described previously.8 A solution of 100 µM 1 and 6 mM HT DNA was added in increments to a solution of 100 µM 1 in 800 mM NaCl and 50 mM sodium phosphate, pH 5.2. The potential was scanned from 0 to 0.95 V at a rate of 5 mV/s and pulse width of 0.1 s. The slopes of the voltammograms were analyzed by the COOL Algorithm software package.16 For cyclic voltammetry (CV) experiments, the electrode potential was scanned between 0 and 1.3 V. Each electrode was electrochemically conditioned prior to data collection by scanning the buffer solution at the intended scan rate (usually seven cycles). The background CV of sodium phosphate buffer was subtracted from the subsequent metal and DNA CV data files before analysis. Experiments were performed in both low ionic strength (50 mM sodium phosphate) and high ionic strength (50 mM sodium phosphate and 800 mM NaCl) buffer solutions. Cyclic voltammograms of 50 µM metal complex in the presence and absence of DNA were recorded for each experiment. All potentials were converted to the Ag/AgCl reference using Fe(CN)64-/3- as an internal standard. Digital Simulations. Second-order rate constants were determined from the CV data by fitting with the DigiSim software package (Bioanalytical systems, West Lafayette, IN). Voltammograms of the metal complex alone were first fit to determine the effective electrode area (A, planar), heterogeneous electrontransfer rate constant between the metal complex and the electrode surface (ks), and the homogeneous rate constant for the spontaneous conversion of Ru(III) back to Ru(II). The following parameters were required for fitting: T ) 298.15 K, RU ) 0, CDL ) 0 and R ) 0.5. The binding constant used for 1 was 20 000 M-1 (as determined by absorption titration and emission experiments). The rate of the binding step is assumed to be diffusion controlled (i.e., kf ) 1 × 109 M-1 s-1). Diffusion coefficient used was 2 × 10-6 cm2/s for 1 (determined by normal pulse voltammetry via the Cottrell equation 33). Diffusion coefficients for DNA fragments are listed in Table 1.12 RESULTS Binding of Complex 1 to DNA. The Ru-tpphz-Os complex (1) was synthesized and characterized by elemental analysis and nonaqueous electrochemistry. The cyclic voltammogram (CV) in acetonitrile showed a reversible wave for each metal center, and the observed redox potentials matched those of the reported values (E1/2(Os(III/II)) ) 0.79 V, E1/2(Ru(III/II)) ) 1.24 V; all potentials versus Ag/AgCl).30 Despite the conjugated bridging (33) Bard, A. J.; Faulkner, L. R. Electrochemical Methods; John Wiley & Sons: New York, 1980.
1854 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
Figure 2. Cyclic voltammogram of 50 µM 1 in the presence of 0 (bold solid line), 250 (‚‚‚), 500 (- - -), and 750 (s) µM HT DNA in 50 mM NaPi, 800 mM NaCl, pH 5.2, at 25 mV/s.
ligand, the redox sites for Ru(III/II) and Os(III/II) are localized, because although the HOMO exhibits major contributions from the metal centers, bpy ligands, and the phenanthroline portion of the tpphz bridging ligand, the contribution from the central phenazine ring of the tpphz ligand that joins the two metal centers is negligible.30 The absence of communication between the two centers was essential for the further studies involving binding and guanine oxidation. The aqueous electrochemistry of 1 was not reported previously. The CV at pH 5.2 with an ITO electrode gave redox potentials of E1/2(Os(III/II)) ) 0.63 V and E1/2(Ru(III/II)) ) 1.09 V (Figure 2, solid line). The cathodic wave for Ru(III/II) at pH 5.2 was more pronounced as compared to that at pH 7 (data not shown); however, no further improvement in the reversibility of the Ru(III/II) wave was observed below pH 5.2. Thus, all electrochemistry with 1 was therefore carried out at pH 5.2. The pH dependence could arise from homogeneous chemistry, such as hydroxide attack on the bridging ligand, which would be inhibited at low pH. Alternatively, protonation of the ITO surface may improve the electrochemistry. The change from pH 7 to pH 5.2 does not affect the redox potentials or acid-base chemistry of the nucleobases.34,35 Under low ionic strength conditions (50 mM sodium phosphate), adsorption of the complex to the ITO surface was observed, so all further experimentation was carried out at high ionic strength (50 mM phosphate, 800 mM NaCl). Since 1 had not previously been studied with nucleic acids, its binding constant to HT DNA was determined by absorbance titration (Figure 3A) using a site-binding model described elsewhere.9 A similar binding constant (2.0 ( 0.1) × 104 M-1 was observed at both low and high ionic strength (50 mM sodium phosphate and 50 mM sodium phosphate + 800 mM NaCl respectively). We expected that at higher salt concentrations, less of the metal complex would be bound because of the competition between the sodium ions and the complex for the negatively charged phosphate backbone, as observed for Ru(bpy)32+;9 however, an intermediate salt concentration of 400 mM NaCl was also examined and yielded the same binding constant as that observed at the other salt concentrations. Hence, binding of 1 to
DNA is independent of ionic strength and must therefore always be included in the analysis of the reaction with DNA. The absorption titration experiment was supported by normal pulse voltammetry experiments. In a titration experiment similar to the absorption titration, the normal pulse voltammogram was measured in the region beyond the Os(III/II) wave in the presence of increasing amounts of DNA. We have previously developed the theory for using electrochemical currents to determine binding constants of redox-active metal complexes.12,16 The mass-transfer-limited current of the Os(III/II) couple decreases as the amount of DNA is increased because of the decrease in the effective diffusion coefficient (Figure 3B) as described by36,37
Deff ) DbXb + DfXf
(1)
where D and X are the diffusion coefficient and mole fraction of the free and bound forms, respectively. Since Xf ) (1 - Xb) and the current, i, is proportional to D1/2, it follows that the mole fraction of bound complex can be determined as12,16
Xb ) (i2 - i02)/(isat2 - i02)
Figure 3. (A) Absorption titration of 1 (50 µM) with HT DNA monitored at 370 nm in low ionic strength buffer (50 mM NaPi, pH 7). The data were fit with a tight binding equation9 to give an association constant of 20 000 M-1 (site size optimized to 2.2). (B) Normal pulse voltammetry of 100 µM complex 1 with increasing amounts of HT DNA (0-3.5 mM) in 50 mM NaPi buffer, 800 mM NaCl, pH 5.2. (C) Slopes of the curves in B used to determine a binding constant of 2.1 × 104 M-1, similar to that obtained by absorption titration.
(2)
where i0 is the current in the absence of nucleic acid, and isat is the current in an excess of DNA. This analysis requires all species to be in equilibrium, a condition we have repeatedly found to obtain in similar experiments.16 These data were fit to the same site-binding model used for absorbance titration (Figure 3C).9 Data from the normal pulse titrations yielded the same value of the binding constant as the absorbance titration. The signal-to-noise ratio is much better for normal pulse voltammetry than for absorbance titration,16 which explains the deviation from the curve at low concentration points in Figure 3A. Since complex 1 has such a high charge, we expected the DNA binding to depend on the ionic strength of the solution and on the oxidation state according to polyelectrolyte theory.13-15,36,37 To determine whether the oxidized form of the heterodimer (5+ charged complex) had a binding constant different from the 4+ complex, the CV was monitored for shift in redox potential upon the addition of nucleic acid. At 250 mV/s, a positive shift (∼35 mV) in the potential of the Os(III/II) couple was observed upon the addition of 750 µM DNA (Figure 4). This behavior indicates that the 5+ species is bound less tightly than the 4+ species (K5+ ) 5 × 103 M-1). Although this observation goes counter to the expectation for electrostatic binding, a similar effect was also observed for Co(phen)33+/2+ and Os(phen)33+/2+ and has been ascribed to a greater hydrophobicity of the lower-charged species. 16 A value of K6+ of 1.5-2 × 103 M-1 was used to obtain good simulations of the catalytic voltammograms described below. The extent of binding of complex 1 with all of the DNA fragments in this study (listed in Table 1) was also determined (34) Blackburn, M. G.; Gait, M. J. Nucleic Acids in Chemistry and Biology; Oxford University Press: New York, 1996. (35) Baik, M. H.; Silverman, J. S.; Yang, I. V.; Ropp, P. A.; Szalai, V. A.; Yang, W. T.; Thorp, H. H. J. Phys. Chem. B 2001, 105, 6437-6444. (36) Carter, M. T.; Rodriguez, M.; Bard, A. J. J. Am. Chem. Soc. 1989, 111, 8901-8911. (37) Carter, M. T.; Bard, A. J. J. Am. Chem. Soc. 1987, 109, 7528-7530.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1855
Figure 4. Cyclic voltammogram of 50 µM 1 (Os(III/II) couple) in the presence of 0 (s) and 750 µM HT DNA (‚‚‚) in 50 mM NaPi, 800 mM NaCl, pH 5.2, at 250 mV/s.
by emission spectroscopy. As observed for related dppz complexes (dppz ) dipyridophenazine),38,39 a large increase in the emission of 1 was observed upon addition of polynucleotide strands. The difference in emission intensity without DNA and with saturating DNA concentrations could be used to determine the extent of binding of the complex at a particular DNA concentration. Over a wide range of nucleotide concentrations (250 and 500 µM for longer fragments and 50 µM for short fragments), the metal complex in the 4+ form bound to the same extent to each fragment when the emission intensities were normalized to equal numbers of nucleotide phosphates. Because we were surprised at the lack of ionic-strength dependence in the affinity of 1 for DNA, a number of experiments were performed to assess the mode of binding. The lack of dependence of complex binding on ionic strength is potentially indicative of an intercalated species,19 which was addressed with both modeling and thermal denaturation studies. Intercalation via threading of molecules through a base pair stack has been observed with two metal complexes joined by a tether containing a large multicylic aromatic compound that inserts between the base pair stack,20,21 so we considered the possibility that complex 1 could intercalate by inserting the planar bridging ligand between the DNA base pairs. This orientation is not likely for 1 because of the close proximity of the metal centers. The model of 1 intercalated into DNA (Figure 5A) shows that the bridging ligand is not long enough to span the base pair stack. Unacceptable steric interactions thereby occur in the minor groove with the ancillary bpy ligands. This conclusion is supported by thermal denaturation experiments that give a ∆Tm of only (2.4 ( 0.5) °C for complex 1 at a DNA-to-metal complex ratio of 12.5:1. Classical intercalators generally give ∆Tm values of 5-16 °C.25,40-44 Thus, the thermal (38) Holmlin, R. E.; Barton, J. K. Inorg. Chem. 1995, 34, 7-8. (39) Friedman, A. E.; Chambron, J. C.; Sauvage, J. P.; Turro, N. J.; Barton, J. K. J. Am. Chem. Soc. 1990, 112, 4960. (40) Neyhart, G. A.; Grover, N.; Smith, S. R.; Kalsbeck, W. A.; Fairley, T. A.; Cory, M.; Thorp, H. H. J. Am. Chem. Soc. 1993, 115, 4423-4428. (41) Kumar, C. V.; Punzalan, E. H. A.; Tan, W. B. Tetrahedron 2000, 56, 70277040.
1856 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
Figure 5. Sybyl model of a short strand of DNA (light) with complex 1 (dark). Ball-and-stick representation on the left, and space-filling on the right. (A) Illustration of the inability of complex 1 to span the base pair stack of DNA. (B) Fit of complex 1 in the major groove of normal B-form DNA.
denaturation and modeling studies do not support intercalative binding of 1 to DNA. Although binding could occur in the minor groove along the backbone, as observed for Ru(bpy)32+,45,46 we also considered binding of 1 to the DNA major groove. Accordingly, the increase in emission intensity of 1 in the presence of HT DNA was not disrupted by the addition of the known minor-groove binder distamycin A47 to the solution. These findings are similar to those of Barton et al. in which distamycin did not disrupt Ru(phen)2(dppz)2+ binding to DNA, which was concluded to occur in the major groove.48 Furthermore, Figure 5B shows a high complementarity of the structure of 1 for the major groove of DNA. This groove binding model is similar to that proposed for a di-iron supramolecular complex.23 A lower effect of ionic strength effect on the binding constant would be expected for such a binding mode. Some highly charged molecules, such as the di-iron supramolecular complex,23 exhibit strong binding to nucleic acids. However, another dimetallic complex similar to complex 1, [Ru(bpy)2]2(dpb)4+, in which dpb is 2,3-bis(2-pyridyl)benzo[g]quinoxaline,49 showed no spectral changes in absorbance in the presence of DNA (42) Dupureur, C. M.; Barton, J. K. Inorg. Chem. 1997, 36, 33-43. (43) Cusumano, M.; Di Pietro, M. L.; Giannetto, A. Inorg. Chem. 1999, 38, 17541758. (44) Cusumano, M.; Giannetto, A. J. Inorg. Biochem. 1997, 65, 137-144. (45) Stradowski, C.; Gorner, H.; Currell, L. J.; Schultefrohlinde, D. Biopolymers 1987, 26, 189-201. (46) Kelly, J. M.; Tossi, A. B.; McConnell, D. J.; Ohuigin, C. Nucleic Acids Res. 1985, 13, 6017-6034. (47) Geierstanger, B. H.; Wemmer, D. E. Annu. Rev. Biophys. Biomol. Struct. 1995, 24, 463-493.
Figure 6. Mechanism for oxidation of DNA by 1.
and was deemed a weak binder. Thus, bimetallic metal species with a 4+ cationic charge can exhibit widely varying binding affinities that are highly dependent on the ancillary ligands. Although the binding mode with regard to groove of 1 for DNA is not totally resolved, the absorption titration, normal pulse, thermal denaturation, and emission studies all suggest nonintercalative, salt-independent binding of modest (K4+ ) 2.0 × 104 M-1) affinity. The confidence with which we have determined this affinity is the primary requirement for the electrochemistry studies involving simultaneous detection of binding and oxidation at the two metal centers. Electrocatalytic DNA Oxidation. Initial experiments involving oxidation of polymeric nucleic acids were conducted using HT DNA. This nucleic acid form extends an average of 1000 base pairs and is 25% guanine. Thus, increased current was expected for the Ru(III/II) couple in the presence of DNA as a result of Ru(III) oxidation of guanines. Because Os(III/II) is not a strong enough oxidant to react with native DNA,12 decreased current for the Os(III/II) couple in the presence of HT DNA was expected because of binding of the metal complex to the polyanion. Initial addition of DNA to the solution of 1 produced the expected response; at 250 µM DNA, the current due to Os(III/ II) was suppressed, and the Ru(III/II) current was increased to produce a catalytic wave (Figure 2). Further increasing the DNA concentration from 250 to 750 µM continued to suppress the Os(III/II) current, as expected, but the effect on the Ru(III/II) wave was surprising. Specifically, the Ru(III/II) assumed an increasingly sigmoidal shape, as expected for increased catalysis; however, the overall current was concomitantly suppressed, presumably as a result of binding of 1 to DNA and a decrease in mass transport. Although the cathodic wave for Ru is not observed in these voltammograms, significant cathodic waves are observed for Ru at lower DNA concentrations and faster sweep rates (i.e., 50 and 100 µM DNA at 250 mV/s). The cyclic voltammetry therefore suggests that catalysis continues to occur at higher DNA concentrations but that as a result of increased binding of 1 at higher DNA concentrations, the mass transport is reduced, and lower overall currents are observed. The dual nature of the binding and catalysis functions of 1 allows us to assess this suggestion in detail. The features of Figure 2 could be reproduced by digital simulation. The model used was one in which the bound Ru(III) form was capable of oxidizing guanine, but also one in which the reversible Os(III/II) couple can bind to DNA. Thus, an additional square was added to Figure 1 to generate the model shown in (48) Holmlin, R. E.; Stemp, E. D. A.; Barton, J. K. Inorg. Chem. 1998, 37, 2934. (49) Carlson, D. L.; Huchital, D. H.; Mantilla, E. J.; Sheardy, R. D.; Murphy, W. R. J. Am. Chem. Soc. 1993, 115, 6424-6425.
Figure 7. DigiSim simulation of CV’s collected for solutions of 50 µM 1 in the presence of 0 (bold solid line), 250 (‚‚‚), 500 (- - -), and 750 (s) µM DNA using the following required parameters: T ) 298.15 K, RU ) 0, CDL ) 0, and R ) 0.5; K4+ ) 2 × 104, K5+ ) 5 × 103 M-1; rate constant for complex formation, kf ) 1 × 109 M-1 s-1; D1 ) 2 × 10-6 cm2/s and DDNA ) 2 × 10-7 cm2/s. The mechanism was used from Figure 6.
Figure 6. This model was used to generate simulations that reproduced the continuous decrease in the current for the Os(III/II) wave and the initial increase, subsequent decrease, and increasing sigmoidal character of the Ru(III/II) wave (Figure 7). Inclusion of either the cross reaction (Ru3+Os3+ + Ru2+Os3+/DNA ) Ru2+Os3+ + Ru3+Os3+/DNA) or the second-order reaction culminating from the unbound metal species (Ru3+Os3+ + DNA ) Ru2+Os3+ + DNAox) did not improve the fit of the simulations to the data and also did not change the simulated rate constants, as observed previously.9,12 The fit was improved by adding a step in which oxidized 1 was converted to a redox-inactive species and in which this reaction was accelerated by the addition of DNA. A similar effect of DNA on the stability of oxidized species has been observed previously.50 The mechanism in Figure 6 was used to determine the rate constant for guanine-Ru(III) electron transfer. Each equilibrium reaction, including DNA binding, heterogeneous electron transfer, and electron transfer between the metal-bound species and DNA, was input as a separate step in the mechanism. The affinities determined independently were used to describe all of the binding steps. The rate of the binding step was assumed to be diffusion controlled (i.e., kf ) 1 × 109 M-1 s-1). In some cases involving (50) Welch, T. W.; Ciftan, S. A.; White, P. S.; Thorp, H. H. Inorg. Chem. 1997, 36, 4812-4821.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1857
Figure 8. DigiSim Fit (- - -) compared to experimental data (s) for 50 µM 1 + 250 µM 200-G, 50 mM NaPi, 800 mM NaCl, pH 5.2 at 25 mV/s.
Ru(bpy)32+, additional oxidation steps in which the product of guanine oxidation undergoes subsequent oxidation reactions with Ru(III) were needed to fit the data.8,9 For 1, no additional oxidation steps were needed. A representative fit is shown in Figure 8. The fitting program results in a first-order rate constant that can be converted into an apparent second-order rate constant via the steady-state equation,
kf(2) ) kf(1)
Kb6+ Kb6+[DNAG] + 1
(3)
where kf(1) is the first-order rate constant extracted from DigiSim, and [DNAG] is the guanine concentration.8,9 The kinetic model for determining the rates of electron transfer from guanine to metal complexes must further account for the fact that not every base is a guanine and that the guanine-metal electron transfer can occur over a distance of 2.5-5 base pairs. We have described this analysis in detail previously.8 In short, the analysis involves multiplying the guanine concentration by a factor that accounts for the fact that the metal complex might bind to a site that is not guanine but is still close enough to a guanine to attract an electron. This factor is then also used to correct the binding constant in the simulation. In the case of 1, a factor of 4 was needed to fit the data. The factor of 4 is consistent with a scenario in which bound 1 is always capable of abstracting an electron from guanine upon generation of Ru(III). This scenario is also consistent with our previous studies of this effect.8 Using the mechanism in Figure 6 and the correction factor of 4, it was possible to fit both the small amplitude wave for the Os(III/II) couple and the catalytic Ru(III/II) wave to extract an average second-order rate constant of (3 ( 1) × 104 M-1 s-1 for guanine oxidation in HT DNA at 25 mV/s in 800 mM NaCl, 50 mM sodium phosphate buffer, pH 5.2 (Table 1). As observed in previous simulations,8 the rate constant did not change upon varying the DNA concentration from 250 µM to 1 mM. The 1858 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
Figure 9. Cyclic voltammogram of 1 (50 µM) with 0 (bold solid line), 250 (‚‚‚), 500 (- - -), and 1000 (s) µM 200-bp PCR product in 50 mM NaPi, 800 mM NaCl, pH 5.2, at 25 mV/s.
observed rate (at 25 mV/s) is similar to that obtained previously for the oxidation of HT DNA by Ru(bpy)33+ at high salt, pH 7 (k2 ) (3.64 ( 0.04) × 104 M-1 s1).51 We further wished to test whether the model properly accounted for oxidation of DNA strands of different length. A shorter strand consisting of 200 base pairs (200-G) that also contained 25% guanine was prepared by PCR. This product was used for electrocatalytic oxidation at the same concentrations used with HT DNA. As with HT DNA, the initial addition of DNA caused the Ru(III/II) current to increase and the Os(III/II) current to decrease followed by a decrease in both currents upon further addition of DNA (Figure 9). Fitting of the data yielded an average rate constant of (7 ( 3) × 104 M-1 s-1 (Table 1), which is similar to the value obtained for oxidation of HT DNA. If the model is correct, an increase in the percentage of guanine should result in a greater amount of catalytic current with the same rate constant. To test this hypothesis, we prepared a 435base-pair PCR product (435-G) containing 30.4% guanine for oxidation by complex 1. As expected, greater catalytic current was observed compared to HT DNA or 200-G at equal nucleotide phosphate concentrations. The resulting rate constant extracted from DigiSim was (8 ( 2) × 104 M-1 s-1 (Table 1), again similar to the others determined. Our model therefore adequately accounts for different overall lengths and different guanine percentages. Multiphasic Kinetics. Voltammograms were collected over a range of scan rates from 25 to 1000 mV/s, and the simulated rate constants showed a systematic increase with scan rate that leveled off at higher scan rates (Table 2). This trend is similar to that observed with mononuclear Ru(bpy)32+, which was attributed to biphasic kinetics in the guanine-Ru electron transfer.8 As in our earlier studies,8 the biphasic kinetics with 1 were apparent in chronoamperometry experiments where plots of the current in the presence of DNA (i) divided by the current in the absence of DNA (id) versus t1/2 showed multiple phases. The observation (51) Weatherly, S. C.; Yang, I. V.; Thorp, H. H. J. Am. Chem. Soc. 2001, 123, 1236-1237.
Table 2. Scan Rate Dependence of Rate Constants for Guanine Oxidation in HT DNA scan rate (mV/s)
kf (M-1 s-1)
25 100 250 500 1000
3.8 (( 0.9) × 104 4.0 (0.8) × 105 1.2 (0.9) × 106 2.3 (1) × 106 2.2 (0.9) × 106
Figure 11. Cyclic voltammograms of 50 µM 1 in the presence of 0 (s), 250 µM 200-G (- - -) or 200-I (‚‚‚) in 50 mM NaPi, 800 mM NaCl, pH 5.2, at 25 mV/s.
Figure 10. Simulation of the effect of biphasic kinetics over a range of scan rates in a simplified model involving electrode oxidation of Ru(II) to Ru(III) and further, second-order reaction of Ru(III) with DNA. The following required parameters were used: T ) 298.15 K, RU ) 0, CDL ) 0, and R ) 0.5; heterogeneous electron transfer: ks ) 0.01 s-1, DRu ) 6 × 10-6 cm2/s, DDNA ) 2 × 10-7 cm2/s. The solid lines show simulations of catalytic oxidation of two noninterconvertible populations of DNA with second-order rate constants of kf,1 ) 1 × 106 M-1 s-1 and kf,2 ) 1 × 104 M-1 s-1). The dotted lines show a fit to each of the simulated voltammograms with a single rate constant (25 mV/s, kf ) 5 × 104 M-1 s-1; 100 mV/s, kf ) 1.5 × 105 M-1 s-1; 250 mV/s, kf ) 2.5 × 105 M-1 s-1; 500 mV/s, kf ) 2.9 × 105 M-1 s-1; 1000 mV/s, kf ) 3 × 105 M-1 s-1).
of multiphase kinetics is a common result for reactions of small molecules in the presence of DNA.17,18 We have observed multiphasic kinetics for oxidation of DNA by Ru(bpy)33+, both by electrochemistry and by stopped-flow absorption spectrophotometry.8 To further test the notion that biphasic reaction kinetics were the source of the scan-rate dependence, additional simulations on a simplified model were performed. In this model, voltammograms were generated using a mechanism by which Ru(III) could react with two distinct, noninterconvertible DNA species with rate constants of 106 M-1 s-1 and 104 M-1 s-1. The generated voltammograms are shown as solid lines in Figure 10. These simulated voltammograms were then fit with a mechanism through which Ru(III) could only react with a DNA species to return rate constants given in the caption for Figure 10 and the dashed line fits. These rate constants showed a systematic increase with scan rate that leveled off at the higher values. This simulation is consistent with our earlier suggestion8 that the scan rate dependence in the DNA system arises from fitting a multiexponential decay with a single rate constant. Because
multiple square schemes are required to the DNA data, we have chosen to use a single reaction rather than to double the number of parameters required for the fit by including two reaction components. Selective Oxidation of Guanine. The redox potential of the Ru(III/II) couple of 1 is slightly higher than that of the simple Ru(bpy)32+ complex, so we wanted to confirm that complex 1 still exhibits the same preferential oxidation of guanine over the other bases, as observed for Ru(bpy)33+. For instance, adenine is the next closest oxidizable base with a redox potential of 1.19 V.52 The same 200-base-pair DNA fragment discussed above (200-G) was therefore prepared with hypoxanthine in place of guanine (200-I). Hypoxanthine, an analogue of guanine with a higher redox potential, can be readily substituted for guanine in PCR products.53 We have shown previously that hypoxanthine is not oxidized efficiently by Ru(bpy)33+.54
Addition of the 200-I PCR product to a solution of 1 gave the expected result: currents for both the Ru(III/II) and Os(III/II) couples were decreased (Figure 11). Addition of the 200-I DNA therefore decreases diffusion of bound 1 to the electrode surface and does not produce catalytic current. DISCUSSION Digital simulation of electrochemical signals potentially offers a convenient means for determining rates and mechanisms of (52) Steenken, S.; Jovanovic, S. V. J. Am. Chem. Soc. 1997, 119, 617-618. (53) Bailly, C.; Payet, D.; Travers, A. A.; Waring, M. J. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13623-13628. (54) Napier, M. E.; Loomis, C. R.; Sistare, M. F.; Kim, J.; Eckhardt, A. E.; Thorp, H. H. Bioconjugate Chem. 1997, 8, 906-913.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1859
complex chemical reactions on small quantities of sample.55,56 This approach has enabled us to determine rate constants for oxidation by Ru(bpy)33+ of many different DNA sequences and structures. These diverse DNA sequences would be difficult to produce in sufficient quantity for similar measurements by absorption spectroscopy. The application of digital simulation to the question of DNA oxidation by Ru(bpy)33+ requires an understanding of a number of factors that influence the effect of chemical reactions on cyclic voltammetry. In particular, we have shown that consistent rate constants are obtained only when there is a thorough treatment of the change in mass transport due to binding of the metal complex to the polyanion.8,9 Further, the fact that nonguanine nucleotides contribute to DNA binding but are not oxidized must also be considered.8 The model we have developed for studying DNA oxidation by small molecules has been tested with a number of scenarios. In particular, we have used Os(bpy)32+ as a substitute for Ru(bpy)32+, since the Os complex exhibits identical binding to DNA but does not oxidize guanine.12 On the DNA side of the reaction, we have varied the strand length, concentration, and fraction of oxidizable guanines present in the sequence.8 Experimenting with reaction time scale has revealed biphasic reaction kinetics similar to our previously reported modeled system of Ru(bpy)32+ with DNA.8 The dimer 1 allows for an even more stringent test of the model in which the Os center, which only reports on binding to DNA, is combined in the same complex with the catalytic Ru center. On the basis of our previous work involving the monomeric complexes, we should be able to extract the appropriate rate constants for Ru(III) oxidation using the modified square scheme in Figure 6.8,9 Here, we demonstrate that such a model does return a rate constant that is consistent with all of our other measure(55) Lerke, S. A.; Evans, D. H.; Feldberg, S. W. J. Electroanal. Chem. 1990, 296, 299-315. (56) Osteryoung, J. Acc. Chem. Res. 1993, 26, 77-83.
1860
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
ments, including stopped-flow spectrophotometry.7,51 The voltammetric response of the dimerswhere the Ru wave shows an initial increase as a result of catalysis followed by suppression of the catalytic wavesoffers a stringent test of our understanding of these reactions and our ability to simulate the voltammetry. Again, careful measurement of the binding properties of 1 is key to simulating the electrochemistry. CONCLUSIONS In summary, we have shown that the dimer 1 offers a novel means to probe DNA oxidation and binding with a single complex in a single voltammetric sweep. This situation is obtained because the Os(III/II) couple is not sufficiently oxidizing to abstract an electron from guanine, whereas the more positive Ru(III/II) couple is an efficient electrocatalyst. The complex exhibits an unusual binding mode, which is salt-independent, despite the 4+ charge on the complex. This unusual behavior is easily discerned using absorption and electrochemical titration, and these results are further supported by thermal denaturation and emission titration. With this understanding of the binding mode, a kinetic model can then be developed that allows for digital simulation of the cyclic voltammetry that accounts for both the catalytic and binding functions of the complex. Such a method allows for mechanistic investigations of DNA oxidation on relatively small quantities of DNA. ACKNOWLEDGMENT We thank S. Feldberg for many helpful conversations. This research was supported by Xanthon, Inc. R.C.H. thanks the Department of Education for a GAANN Fellowship. Received for review July 19, 2002. Accepted January 30, 2003. AC0204653
