Bioconjugate Chem. 1999, 10, 419−423
419
Electrostatic Interactions of Redox Cations with Surface-Immobilized and Solution DNA Adam B. Steel,*,† Tonya M. Herne, and Michael J. Tarlov* Process Sensing Group, NIST, Gaithersburg, Maryland 20899. Received September 18, 1998; Revised Manuscript Received January 5, 1999
Association constants for ruthenium(III) hexaamine and cobalt(III) tris(2,2′-bipyridine) with solution and surface-immobilized DNA were determined. The interaction of the cationic redox molecules with calf thymus DNA was monitored via normal pulse voltammetry with analysis of the mass-transfer limited current assuming a discrete binding-site model. Single-stranded DNA was immobilized on gold via self-assembly of a 5′ hexanethiol linker. Double-stranded surface-immobilized DNA was produced by hybridization of a complementary target to surface-immobilized single strands. The interaction between the metal complexes and surface-immobilized DNA was determined using chronocoulometry to construct adsorption isotherms. The measured binding constants for the cationic redox molecules with solution, surface-immobilized single-stranded, and double-stranded DNA are well-correlated, even as a function of ionic strength. The agreement between the determined association constants for the surface-immobilized and solution DNA indicates the potential utility of DNAderivatized electrodes for examination of small molecule interactions with nucleic acids.
INTRODUCTION
Heterogeneous DNA hybridization, the association of a solution-phase single strand with its immobilized complementary strand at a solid surface, has received considerable attention as an alternative to current DNA diagnostic methods because it is more labor, time, and material efficient (1). While biosensors based on heterogeneous hybridization at arrays are now being commercially developed, the physical characteristics of the surface-immobilized DNA are at present poorly understood (2). Of particular interest is the influence of surface immobilization on small molecule binding to DNA. In this report, we compare the electrostatic binding of small redox molecules to DNA in solution and immobilized on alkanethiol-modified gold surfaces. Small molecule binding to DNA has been researched in great detail due to its importance in strand scission mediation, as a structural probe, and as a model for protein-nucleic acid interactions (3). Binding modes include covalent attachment, electrostatic attraction, and intercalation into the hydrophobic regions of the DNA. The interaction between a molecule and DNA is often an interplay of the available binding modes. Studies of the binding of electrochemically active small molecules to solution-phase DNA have provided association constants for a wide range of organic and metal chelate compounds (3, 4). The binding constants can be subsequently decomposed into contributions for each type of interaction to provide a more detailed picture of the bound complex. DNA biosensors have been developed based on an electrochemical signal change for the interaction of small molecules with oligonucleotides immobilized on an electrode surface (1, 5). We recently reported an electrochemical method to determine DNA immobilized on gold based on the electrostatic accumulation of electroactive * To whom correspondence should be addressed. † Current address: Gene Logic, Inc. 708 Quince Orchard Rd., Gaithersburg, MD 20878.
metal complexes with the nucleotide phosphate backbone (6). The quantitative validity of this method rests heavily on using molecules that interact with DNA predominately through electrostatic attraction. As such, small, highly charged redox cations such as ruthenium hexaamine (RuHex) and cobalt trisbipyridyl (CoBPY) were selected as electrochemical markers. In this paper, we report the binding constants for RuHex and CoBPY with solution and surface-immobilized DNA. EXPERIMENTAL SECTION
Materials. All reagents were obtained from Aldrich Chemical, unless otherwise noted. All solutions were made with deionized water (18 MΩ cm resistivity) from a Barnstead NANOpure system (7). Calf thymus DNA was obtained from Sigma Chemical Co. (St. Louis, MO) and was used as received. Concentrated DNA solutions (3-6 mM in nucleotide phosphate) were prepared by dissolving the solid sodium salt of DNA in the measurement electrolyte. Synthetic DNA oligonucleotide probes were purchased from Research Genetics (Huntsville, AL). The probe oligonucleotide contained a 6-mercaptohexyl linker at the 5′ end and 25 bases with the sequence 5′HS(CH2)6-CAC GAC GTT GTA AAA CGA CGG CCA G-3′. The complementary single-stranded target was a 25mer with the sequence 5′-CTG GCC GTC GTT TTA CAA CGT CGT G-3′. 6-Mercapto-1-hexanol (MCH) and 2-mercapto1-ethanol (MCE) were purified using flash chromatography (silica, chloroform). Hexaamineruthenium(III) chloride (99%) (Strem Chemical, Newburyport, MA) was used as received. Tris(2,2′-bipyridyl) cobalt(III) perchlorate was prepared using the procedure of Dollimore and Gillard and recrystallized twice from water prior to use (8). Electrode Preparation. Gold electrodes were prepared by rf sputter deposition of ca. 20 nm of a chromium adhesion layer followed by ca. 200 nm of gold (each 99.99% purity) onto glass microscope slides. Electrodes were cleaned by exposure to warm piranha solution [70% concentrated sulfuric acid, 30% peroxide solution (30%); WARNING: piranha reacts violently with organic sol-
10.1021/bc980115g Not subject to U.S. Copyright. Published 1999 by American Chemical Society Published on Web 03/06/1999
420 Bioconjugate Chem., Vol. 10, No. 3, 1999
vents] for 12 min. The electrodes were rinsed thoroughly with deionized water and placed in monolayer deposition solutions while wet. MCE- and MCH-modified surfaces were prepared from 1 mM aqueous solutions. DNA-modified surfaces were prepared using our previously described procedure (9). Briefly, mixed monolayer surfaces containing thiolated probe DNA (ss-DNA) and MCH were prepared by immersing the clean gold substrate in a 1.0 µM solution of probe oligonucleotide in 1.0 M potassium phosphate buffer, pH 7, for 15 min, rinsing with 10 mM NaCl, 5 mM Tris buffer, pH 7.4 (R-BFR) for 5 s, immersing in a 1.0 mM MCH solution in deionized water for 1 h, and finally rinsing with R-BFR for 5 s and drying under a stream of nitrogen. Surface-immobilized double-stranded DNA (ds-DNA) was obtained by hybridization of the immobilized ss-DNA to the complementary target at 35 °C for 60 min in 1.0 M NaCl, 10 mM Tris buffer, and 1 mM EDTA, pH 7.4. The concentration of complementary target was 1.0 µM. Upon removal from the hybridization reaction solution, the electrodes were rinsed with R-BFR for 5 s and dried under a stream of nitrogen prior to electrochemical characterization. Instrumentation. Cyclic voltammetry (CV), normal pulse voltammetry (NPV), and chronocoulometry (CC) were performed with a CH Instruments (Cordova, TN) model 660 electrochemical analyzer (7). The following parameters were employed: CV, sweep rate ) 100 mV/ s; NPV, pulse duration ) 100 ms, pulse frequency ) 1 s-1, step increment ) 4 mV; CC, pulse period ) 500 ms, pulse width ) 500 mV. Ultraviolet-visible spectra were obtained on a Perkin-Elmer Lambda Bio 20 spectrophotometer (7). Procedures. All electrochemical characterizations were performed in a single-compartment cell with a 10 mL volume. The working electrodes were prepared as described above. A saturated calomel electrode (SCE) and platinum wire gauze served as the reference and counter electrodes, respectively. Supporting electrolyte, 10 mM Tris buffer, pH 7.4 (E-BFR), was deoxygenated via purging with nitrogen gas for 10 min prior to measurements, and the cell was blanketed with nitrogen for the experiment duration. All experiments were carried out at laboratory ambient temperature (21-24 °C). Binding constants for redox cations with calf thymus DNA in solution were calculated following the DNA titration strategies of the Bard (3e) and Thorp (3h) groups. Aliquots of solution containing 50 µM redox cation and 3.0 mM nucleotide phosphate were added to a solution containing 50 µM redox cation in 10 mM Tris buffer. For competitive binding studies, NaCl was added to the Tris buffer. Normal pulse voltammograms were collected for each nucleotide phosphate concentration at a MCE-modified electrode. MCE-modified electrodes were used in place of a bare gold electrode because the selfassembled monolayer blocks nonspecific adsorption of the redox molecule and DNA without affecting the reaction kinetics (10). Binding constants for redox cations with surfaceimmobilized ss- and ds-DNA were calculated from adsorption isotherms. The amount of redox cation accumulated at the electrode surface due to the presence of nucleotide phosphate was determined via chronocoulometry as previously described (6). In brief, double layer charge determinations were made in the electrolyte buffer without redox cation. When a redox cation is added to the low ionic strength electrolyte, the redox cation exchanges with the native charge compensation cation, accumulating at that interface. The amount of ac-
Steel et al.
cumulated redox cation can then be measured using chronocoulometry, a current integration technique, under equilibrium conditions. The number of redox cations at the electrode surface is determined from the difference in chronocoulometric intercepts for the identical potential step experiment in the presence and absence of redox cation. Accumulation of the redox cation at the electrode surface does not significantly alter the double layer charge for the electrode based on experiments using an equivalent potential step magnitude but over a region where the redox cation is not electroactive. Adsorption isotherms were constructed by adding aliquots of a redox cation solution to 10 mL of electrolyte buffer, stirring the solution, and waiting 1 min prior to recording the chronocoulometric trace. Variable waiting periods were used to ensure that the system had reached equilibrium; 30 s of equilibration time was found to be sufficient. RESULTS AND DISCUSSION
Solution-Phase DNA Interactions. Association constants for electrostatic binding of RuHex and CoBPY to DNA were determined following the method of the Bard and Thorp groups (3e,h). High-affinity binding to DNA, as is the case for the redox molecules under consideration here, is best described by a discrete site model. Free small molecule and DNA-binding sites, of a fixed size s in base pairs, are in equilibrium with the bound species. An expression for the bound fraction as a function of the association constant, K, and the binding site size, s, can be derived from the equilibrium constant expression
(
)
2K2[M][NP] s 2K[M]
b - b2 Xb )
b ) 1 + K[M] +
K[NP] 2s
1/2
(1a) (1b)
where Xb is the bound mole fraction, [M] is the total metal ion concentration, and [NP] is the nucleotide phosphate concentration. Equation 1 is valid for noncooperative, nonspecific binding to a unique binding site on DNA. Values for the binding constant, K, and the binding site size, s, are determined by nonlinear regression analysis of bound fraction versus nucleotide phosphate profiles. The bound fraction of redox molecule was determined from the plateau current in normal pulse voltammograms obtained during the titration of redox molecule with calf thymus DNA (3h). The plateau current diminishes upon the addition of DNA because the effective diffusion coefficient of the redox molecule is lowered by binding to DNA. The bound mole fraction, assuming rapid exchange between free and bound states on the voltammetric time scale, is given by a ratio of squared current differences
Xb )
i2 - i2o i2sat - i2o
(2)
where io is the plateau current in the absence of DNA and isat is the current at complete saturation. Because we are interested in promoting electrostatic attraction between the redox cations and nucleotide phosphate backbone, DNA titrations were performed in the absence of excess electrolyte cation. Titration data are given in Figure 1 for both RuHex and CoBPY with the associated nonlinear regression best-fits to eqs 1a and 1b. The regression analysis results are summarized in Table 1. The association constant for the smaller, harder RuHex is an order of magnitude larger than that for
Electrostatic Interactions of Redox Cations
Bioconjugate Chem., Vol. 10, No. 3, 1999 421
Figure 1. Plot of the bound mole fraction of 50 µM metal complex upon titration with calf thymus DNA in 10 mM Tris buffer, pH 7.4, no excess Na+, and associated nonlinear leastsquares fits to eq 1. Binding titrations are given for ruthenium(III) hexaamine (RuHex, closed circles) and cobalt(III) trisbipyridal (CoBPY, open circles).
Figure 2. Plot of the bound mole fraction of 50 µM RuHex upon titration with calf thymus DNA in 10 mM Tris buffer, pH 7.4 with increasing excess Na+ and associated nonlinear leastsquares fits to eq 1. Binding titrations were performed with excess sodium concentrations of 0 (squares), 10 (circles), and 100 (diamonds) mM.
Table 1. Binding Constants of RuHex and CoBPY to Calf Thymus DNA in Solution
Surface-Immobilized DNA. Study of the interactions of small molecules with surface-immobilized oligonucleotides is important for the developing field of DNA diagnostics. Of particular interest here is the interaction of electrostatically bound molecules to DNA immobilized on gold. Attachment of DNA to an electrode surface provides a convenient means to probe the interactions of redox molecules and DNA, as shown by Kelley et al. (1a). Surfaces modified with ss-DNA and ds-DNA were prepared to determine if molecule binding affinity is a function of DNA structure (i.e., single-strand vs duplex DNA). Cationic RuHex interacts strongly with the polyanionic backbone of surface-immobilized DNA. The strong interaction is manifested in cyclic voltammograms as a postwave shoulder attributable to surface-confined species (11). Five consecutive voltammograms for a ss-DNAmodified electrode in 50 µM RuHex are given in Figure 3. The overlap of all of the voltammograms is evidence that equilibrium has been established in the system. Also shown in Figure 3 is a series of voltammograms taken after exposing the electrode to 50 µM RuHex solution, then transferring the electrode to a redox-free buffer for measurement. Removal of the electrode from the redoxcontaining solution allows us to ascribe the peak at -272 mV to surface-confined redox molecules. Further confirmation that the peak at -272 mV is due to a surfaceconfined species is determined from scan rate dependence studies. The peak currents at -168 mV (diffusion-limited) and -272 mV (surface-confined) are plotted as a function of scan rate in Figure 4. The peak current at -272 mV is linear with scan rate, indicating a surface-confined reaction, while the peak current data for -168 mV is indicative of the mass-transfer limited reaction. The voltammetry of CoBPY at DNA-modified electrodes does not exhibit a post-wave, as was observed for RuHex, reflecting the difference in the relative strength
complex
excess [Na+] (mM)
Ka (M-1)
sa (bp)
RuHex
0 10 100 0 10 15b
1.2 × 106 5.0 × 105 6.0 × 104 2.0 × 105 4.5 × 104 5.8 × 104
2.0 2.5 4.0 2.0 3.0 3.0
CoBPY
a The binding constant, K, and binding site size, s, were calculated from linear regression analysis of titration data in electrolytes with increasing excess sodium ion concentration. b Reference 3h.
CoBPY. The binding site sizes obtained for RuHex and CoBPY in the absence of excess electrolyte cation are the same, within the experimental uncertainty. Furthermore, the binding site size of 2.0 base pairs approaches the theoretical saturated binding site size of 1.5 base pairs for complete charge compensation of the nucleotide phosphate by trivalent redox cation. We note that the sodium ion concentration in these experiments is equivalent to the nucleotide phosphate concentration, roughly 1 mM, so that a significant amount of sodium cation is available in solution to compete with the redox cation for the anionic phosphate-binding sites. It is not surprising then that the theoretical saturated binding site size of 1.5 base pairs/trivalent redox cation is not achieved. The influence of a competing cation, e.g., sodium, was studied further by measuring the binding constant, K, and site size, s, in solutions with increasing ionic strength. Binding isotherms for RuHex are given in Figure 2 with the accompanying nonlinear regression analysis fits. The data are summarized in Table 1. As would be expected, the binding constant decreases and the effective binding site size increases with increasing ionic strength. Our results are in good agreement with those reported by Welch et al. (3h).
422 Bioconjugate Chem., Vol. 10, No. 3, 1999
Steel et al.
Figure 3. Cyclic voltammetry of RuHex at ss-DNA modified electrode. The first five cycles, v ) 100 mV/s, for an electrode in 50 µM RuHex, 10 mM Tris, pH 7.4 (solid line), and an electrode that was incubated in same solution for 1 min, washed, and transferred to 10 mM Tris, pH 7.4, for measurement (dotted line).
Figure 4. Plot of the peak currents at -168 (squares) and -272 mV (circles) versus scan rate for a ss-DNA modified electrode in 50 µM RuHex, 10 mM Tris, pH 7.4. The lines are given as a guide to the eye only.
of the redox cation/DNA interaction (11). Loading experiments, such as those included for RuHex in Figure 3, show that CoBPY does accumulate at the electrode surface. Determination of Redox Molecule Affinities for Surface-Immobilized DNA. The binding constants of redox molecules with ss- and ds-DNA were determined from adsorption isotherms. The amount of surfaceconfined redox molecule was measured using chronocou-
Figure 5. Adsorption isotherms of RuHex at ss-DNA (closed circles) and ds-DNA (open circles) modified electrodes. The interfacial charge due to adsorbed RuHex was determined from the increase in chronocoulometric intercept with increasing metal complex concentration.
lometry. Binding isotherms for RuHex at ss- and ds-DNA modified electrodes in E-BFR are given in Figure 5. An association constant, K, was calculated from the isotherm data using the Langmuir adsorption isotherm model. The Langmuir model assumes that the every binding site is equivalent and that the ability of a molecule to bind is independent of occupation of nearby sites. A linearized form of the Langmuir isotherm in terms of the solution
Electrostatic Interactions of Redox Cations
Bioconjugate Chem., Vol. 10, No. 3, 1999 423
Table 2. Binding Constants of RuHex and CoBPY to ssand ds-DNA Immobilized on an Electrode Surface complex RuHex CoBPY
excess [Na+] (mM)
K/ss-DNAa (M-1)
K/ds-DNAa (M-1)
0 10 100 0 10
(2.2 ( 0.9) × 106 (3.7 ( 2.2) × 105 (2.7 ( 1.6) × 104 (2.7 ( 0.5) × 105 (6.9 ( 0.3) × 104
(1.3 ( 0.4) × 106 (3.0 ( 0.6) × 105 (4.1 ( 0.8) × 104 (1.6 ( 0.3) × 105 (5.6 ( 0.4) × 104
a The binding constants, K, were determined from Langmuir analysis of adsorption isotherm data in electrolytes containing the listed excess sodium ion concentrations. The average and standard deviation on at least three measurements per entry are given.
shown that the interaction of RuHex with surfaceimmobilized DNA can be used as a means to quantitate the surface density of DNA when the length of the probe and target are known (6). The DNA surface densities calculated from the saturated RuHex coverages in Figure 6 are 3.2 × 1012 probes/cm2 and hybrids/cm2. Comparison of the binding constants in Table 2 shows that these cations interact equivalently with ss- and dsDNA over the range of ionic strengths studied. The correspondence of these binding constants indicates that electrostatic attraction is the predominant contributor to the overall metal chelate-DNA interaction. Comparison of the association constants in Tables 1 and 2 shows that binding affinity is not affected by end-tethering DNA to a surface. CONCLUSION
We have shown that electrostatic binding of select cationic molecules to DNA is not affected by immobilization of the nucleic acid to a surface. Thus, DNA-derivatized electrodes provide a convenient and potentially powerful means to study the interactions of small molecules with nucleic acids. We intend to explore the utility of the techniques outlined herein for the investigation of other small molecule-DNA interactions. LITERATURE CITED
Figure 6. Linearized Langmuir isotherm. The adsorption isotherm data for RuHex at ss-DNA (closed circles) and ds-DNA (open circles) modified electrodes in accordance with eq 3. The lines are the least-squares best-fit to the data used to calculate the binding constants given in Table 2.
redox cation concentration, [M], the accumulated charge at the electrode surface, Q, and the saturated charge, Qsat, is given by eq 3.
[M] [M] 1 ) + Q Qsat KQsat
(3)
Results of Langmuir analysis of isotherm data are given in Table 2. The binding isotherm data for RuHex at ssand ds-DNA modified electrodes in E-BFR (no added NaCl) are plotted in the linearized Langmuir plot in Figure 6. The saturation coverages of RuHex are 0.6 and 1.2 µC/cm2 for ss- and ds-DNA modified surfaces, respectively. Binding constants calculated from the plots in Figure 6 are 2.1 × 106 and 1.1 × 106 M-1 for ss- and dsDNA, respectively. The doubling of the redox molecule coverage in going from ss- to ds-DNA is particularly satisfying because it indicates that all of the ss-DNA is hybridized when we form our ds-DNA samples. The doubling of saturated redox molecule coverage is a result of our probe and target having the same length so that forming the hybrid doubles the number of binding sites (nucleotide phosphates) at the electrode surface. We have previously
(1) (a) Kelley, S. O., Barton, J. K., Jackson, N. M., and Hill, M. G. (1997) Bioconjugate Chem. 8, 31. (b) Chan, V., Graves, D. J., and McKenzie, S. E. (1995) Biophys. J. 69, 2243. (c) Wang, J., Cai, X., Rivas, G., Shiraishi, H., Faria, P. A. M., and Donta, N. (1996) Anal. Chem. 68, 2629. (d) Hashimoto, K., Ito, K., and Ishimori, Y. (1994) Anal. Chem. 66, 3830. (e) Millan, K. M., and Mikkelsen, S. R. (1993) Anal. Chem. 65, 2317. (2) (a) Chan, V., Graves, D. J., Fortina, P., and McKenzie, S. E. (1997) Langmuir 13, 320. (b) Chan, V., Graves, D. J., and McKenzie, S. E. (1995) Biophys. J. 69, 2243. (3) (a) Barton, J. K., Danishefsky, A. T., and Goldberg, J. M. (1984) J. Am. Chem. Soc. 106, 2172. (b) Kelly, J. M., Tossi, A. B., McConnell, D. J., and OhUigin, C. (1985) Nucleic Acids Res. 13, 6017. (c) Barton, J. K. (1986) Science 233, 727. (d) Dervan, P. B. (1986) Science 233, 464. (e) Carter, M. T., Rodriguez, M., and Bard, A. J. (1989) J. Am. Chem. Soc. 111, 8901. (f) Meyer-Almes, F. J., and Porschke, D. (1993) Biochemistry 32, 4246. (g) Murphy, C. J., Arkin, M. R., Ghatlia, N. D., Bossmann, S., Turro, N. J., and Barton, J. K. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 5315. (h) Welch, T. W., and Thorp, H. H. (1996) J. Phys. Chem. 100, 13829. (i) Wang, J., Rivas, G., Cai, X., Donta, N., Shiraishi, H., Luo, D., and Valera, F. S. (1997) Anal. Chim. Acta 337, 41. (j) Pang, D., and Abruna, H. D. (1998) Anal. Chem. 70, 3162. (4) Mahadevan, S., and Palaniandavar, M. (1997) Inorg. Chim. Acta 254, 291. (5) (a) Zhao, Y., Pang, D., Wang, Z., Cheng, J., and Qi, Y. (1997) J. Electroanl. Chem. 431, 203. (b) Palanti, S., Marrazza, G., and Mascini, M. (1996) Anal. Lett. 29, 2309. (c) Napier, M. E., and Thorp, H. H. (1997) Langmuir 13, 6342. (6) Steel, A. B., Herne, T. M., and Tarlov, M. J. (1998) Anal. Chem. 70, 4670. (7) Certain commercial products and instruments are identified to adequately specify the experimental procedure. In no case does such identification imply endorsement by the authors or NIST. (8) Dollimore, L. S., and Gillard, R. D. (1973) J. Chem. Soc., Dalton Trans. 933. (9) Herne, T. M., and Tarlov, M. J. (1997) J. Am. Chem. Soc. 119, 8916. (10) Terrattaz, S., Cheng, J., Miller, C. J., and Guiles, R. D. (1996) J. Am. Chem. Soc. 118, 7857. (11) Bard, A. J., and Faulkner, L. R. (1980) Electrochemical Methods, p 527, John Wiley & Sons, New York.
BC980115G
