DNA Sensing on a DNA Probe-Modified Electrode Using

Quelle place pour les puces à ADN dans les maladies inflammatoires ? V. Devauchelle , G. Chiocchia. La Revue de Médecine Interne 2004 25 (10), 732-739...
0 downloads 0 Views 157KB Size
Download PDF
Anal. Chem. 2000, 72, 1334-1341

DNA Sensing on a DNA Probe-Modified Electrode Using Ferrocenylnaphthalene Diimide as the Electrochemically Active Ligand Shigeori Takenaka,*,† Kenichi Yamashita,† Makoto Takagi,† Yoshihiro Uto,‡ and Hiroki Kondo‡

Department of Chemical Systems and Engineering, Kyushu University, Fukuoka 812-8581, Japan, and Department of Biochemical Engineering and Science, Kyushu Institute of Technology, Iizuka 820-8502, Japan

Naphthalene diimide derivative 1 carrying ferrocenyl moieties at the termini of imide substituents binds intact calf thymus DNA 4 times more strongly than the denatured DNA, and its complex with the intact DNA dissociates 80 times more slowly than that with the denatured DNA. On the basis of these observations, ligand 1 was applied to a probe of electrochemical DNA sensing. A thiol-linked single-stranded DNA probe was immobilized through the S-Au bonding to 20-30 pmol/mm2 on a gold electrode. Following hybridization with the complementary DNA, the electrode was soaked in a solution containing 1 (intercalation step) and then washed with buffer for 5 s. The cyclic voltammogram and differential pulse voltammogram for this electrode gave an electrochemical signal due to the redox reaction of 1 that was bound to the double-stranded DNA on the electrode. Thus, dA20 and the yeast choline transport gene were quantitated at the subpicomole level. The sensitivity of DNA detection was improved to 10 zmol by reducing the amount of immobilized DNA probe and protecting the uncovered surface of the electrode with 2-mercaptoethanol. Intercalators are a class of DNA ligands that insert or intercalate between adjacent base pairs of double-stranded DNA.1 Some of them carry bulky substituents on the periphery of intercalating moiety, and their substituents become placed in the major and minor grooves simultaneously when intercalated to the DNA duplex.1,2 These types of intercalators are called threading intercalators. Threading intercalators are expected to dissociate only very slowly from the double-stranded DNA because of this peculiar binding mode. On the other hand, they will exert little, if any, stabilizing effect on the complex with single-stranded DNA. Therefore, threading intercalators should be able to discriminate between double- and single-stranded DNA with a large margin. Naphthalene diimide derivatives are typical synthetic threading intercalators. We have been developing new functional DNA ligands based on this unique property of naphthalene diimides by furnishing them with various functional modifications in the †

Kyushu University. Kyushu Institute of Technology. (1) Wilson, W. D. In Nucleic Acids in Chemistry and Biology; Blackburn, G. M., Gait, M. J. Eds.; Oxford University Press: New York, 1996; p 329. (2) Takenaka S.; Takagi, M. Bull. Chem. Soc. Jpn. 1999, 72, 327. ‡

1334 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

imide positions.2-6 Recently, we succeeded in the synthesis of naphthalene diimide derivative 1 carrying ferrocenyl moieties.7 As expected, this ligand formed a stable complex with doublestranded DNA, and owing to the redox activities of the ferrocene moieties, its interaction with DNA could be monitored electrochemically.8-10 This observation prompted us to study the electrochemical quantitation of complementary DNA by using a singlestranded DNA probe which was immobilized on a gold electrode. Similar approaches were taken previously with such electrochemical DNA ligands as classical intercalators,11 groove binders,12 and metal chelates.13,14 There was, however, no solid rationale behind those approaches, since the ligands adopted failed to discriminate the double-stranded from the single-stranded DNAs. Herein, we report for the first time a successful construction of a practical DNA sensor, based on DNA ligand 1. EXPERIMENTAL SECTION Chemicals. 5′-Mercaptohexyloligonucleotides, 5′-HS(CH2)6p-dT20-3′, 5′-HS(CH2)6-p-dA20-3′, and 5′-HS(CH2)6-p-d(CCGCTTATCTTCAGTTTTCA)-3′ and oligonucleotides, dT20 and dA20 were obtained from Takara Co. (Kyoto, Japan). The concentrations of these oligonucleotides were estimated from the molar absorptivities at 260 nm: 162 600 cm-1 M-1 for 5′-HS(CH2)6-p-dT20-3′ and dT20, 243 400 cm-1 M-1 for 5′-HS(CH2)6-p-dA20-3′ and dA20, and 197 200 cm-1 M-1 for 5′-HS(CH2)6-p-d(CCGCTTATCTTCAGTTTTCA)-3′.15 Calf thymus DNA was purchased from Sigma and (3) Takenaka, S.; Nishira, S.; Tahara, K.; Kondo, H.; Takagi, M. Supramol. Chem. 1993, 2, 41. (4) Takenaka, S.; Manabe, M.; Yokoyama, M.; Nishi, M.; Tanaka, J.; Kondo, H. J. Chem. Soc., Chem. Commun. 1996, 379. (5) Takenaka, S.; Yokoyama, M.; Kondo, H. J. Chem. Soc., Chem. Commun. 1997, 115. (6) Takagi, M.; Yokoyama, H.; Takenaka, S.; Yokoyama, M.; Kondo, H. J. Inclusion Phenom. Mol. Recognit. Chem. 1998, 32, 375. (7) Takenaka, S.; Uto, Y.; Saita, H.; Yokoyama, M.; Kondo, H.; Wilson, W. D. J. Chem. Soc., Chem. Commun. 1998, 1111. (8) Takenaka, S.; Uto, Y.; Kondo, H.; Ihara, T.; Takagi, M. Anal. Biochem. 1994, 218, 436. (9) Ihara, T.; Maruo, Y.; Takenaka, S.; Takagi, M. Nucleic Acids Res. 1996, 24, 4273. (10) Uto, Y.; Kondo, H.; Abe, M.; Suzuki, T.; Takenaka, S. Anal. Biochem. 1997, 250, 122. (11) Rolley, S. O.; Barton, J. K.; Jackson, N. M.; Hill, M. G. Bioconjugate Chem. 1997, 8, 31. (12) Hashimoto, K.; Ito, K.; Ishimori, Y. Anal. Chem. 1994, 66, 3830. (13) Carter, M. T.; Rodriguez, M.; Bard, A. J. J. Am. Chem. Soc. 1989, 111, 8901. (14) Millan, K. M.; Mikkelsen, S. R. Anal. Chem. 1993, 65, 2317. 10.1021/ac991031j CCC: $19.00

© 2000 American Chemical Society Published on Web 02/18/2000

was purified as described previously.16 Denaturation of calf thymus DNA was carried out in aqueous solution by heating at 100 °C for 10 min and then immediately cooled on ice. [Poly(dA-dT)]2 and [poly(dG-dC)]2 were also purchased from Sigma and used without further purification. Recombinant plasmid haboring the yeast choline transport gene was a gift of Professor J. Nikawa.17 This DNA has a unique d(CCGCTTATCTTCAGTTTTCA) sequence on 3693 base pairs. Part of this gene (51 bp) was ligated with pBluescript II SK+. The resulting recombinate DNA was propagated in Escherichia coli strain JM 105. The linear DNA fragment was prepared by PstI digestion of this recombinant. General Physicochemical Methods. Melting points were uncorrected. 1H NMR spectra were recorded on a Brucker AC250P or JEOL GSX-400 spectrometer operating at 250 and 400 MHz, respectively, with tetramethylsilane (TMS) as an internal standard. The absorption spectra were recorded with Hitachi U-3210 and 3300 spectrophotometers equipped with an SPR 10 temperature controller. The HPLC system was composed of the following parts: Hitachi L-7300 column oven, L-7450H diode array detector, two L-7100 pumps, D-7000 Interface chromatograph. Circular dichroism (CD) spectra were recorded over the 220550 nm range on a Jasco J7200 spectropolarimeter. Viscosity titrations were carried out with a PC-controlled automatic system (Lauda, BVS1) equipped with a capillary Ubbelohde-type viscometer, an automatic pump/stopwatch unit, and a thermostated water bath at 30 ( 0.1 °C. Aliquots (2 µL) of 1.0 mM solution of 1 were added to a DNA sample solution (0.2 mM) by means of a microsyringe without removing the solution from the viscometer. The relative viscosity ratios of DNA alone and its complex with ligand 1 were calculated using the expression η/η0 ) (t - to)/(tDNA - to), where to is the flow time of the buffer and t and tDNA are the flow times of the DNA sample in the presence and absence of the ligand, respectively. Synthesis. Ferrocenylnaphthalene diimide 1 was synthesized according to the route given in Scheme 1. N,N′-Bis[[4-(3-aminopropyl)piperazinyl]propyl]naphthalene-1,4,5,8-tetracarboxylic Acid Diimide (2). Two grams of naphthalene-1,4,5,8-tetracarboxylic dianhydride (Aldrich) and 40 mL of N,N′-bis(3-aminopropyl)piperazine were refluxed in 30 mL of tetrahydrofuran for 8 h. The solution was allowed to cool and then was poured into ether (1 L). The precipitate formed was dissolved in a small amount of chloroform-methanol (1:1) and the resultant mixture poured into ether (1 L). The precipitate was removed by filtration, the filtrate was concentrated, and the residue was recrystallized from ether to yield 0.8 g (18%) of 2 as a brown solid: mp >300 °C; 1H MNR (400 MHz, CDCl3) δ 1.58 (4H, m), 1.95 (4H, m), 2.27-2.52 (28H, m), 2.71 (4H, t), 4.28 (4H, t), 8.75 (4H, s) ppm. Anal. Calcd for C34H4804N8‚2.5H2O: C, 60.27; H, 7.83; N, 16.57. Found: C, 60.27; H, 7.51; N, 16.58. N,N′-Bis[[4-(3-ferrocenecarboxamidopropyl)piperazinyl]propyl]naphthalene-1,4,5,8-tetracarboxylic Acid Diimide (1). A solution of 2 (300 mg, 0.47 mmol) and the N-hydroxysuccinimide ester of ferrocenecarboxylic acid (3)8 (600 mg, 1.8 mmol) in chloroform (30 mL) was stirred at room temperature for 50 h. The solvent was removed, and the residue was chromatographed (15) Cantor, C. R.; Warshaw, M. M. Biopolymers 1970, 9, 1059. (16) Davidson, M. W.; Griggs, B. G.; Boykin, D. W.; Wilson, W. D. J. Med. Chem. 1997, 20, 1117. (17) Nikawa, J.; Tsukagoshi, Y.; Yamashita, S. J. Biol. Chem. 1991, 266, 11184.

Scheme 1

on a column of silica gel (Merck 60, methanol). The fraction showing an Rf of 0.2 on TLC (methanol) was collected, and the solvent was removed under reduced pressure. The residue was dissolved in a small amount of chloroform, and the solution was poured into ether. The solid obtained by filtration was dissolved in a small amount of methanol and and the resultant mixture poured into water. The solid obtained was further purified by recrystallization from acetone to yield 33 mg (7%) of 1 as a yellow solid. Homogeneity of the product was confirmed by HPLC; Inertsil ODS-3 (5 mm), 4.6 × 250 mm, GL Science Inc.; flow rate 1.0 mL/min; solvent A, 0.1% (trifluoroacetic acid) (TFA) in acetonitrile; solvent B, acetonitrile; 0-70% B in 40 min. Elution was monitored by absorption at 250-400 nm. The retention time of 1 was 30.03 min: mp 237-240 °C; 1H NMR (CDCl3) δ 1.72 (4H, m), 1.96 (4H, m), 2.43 (12H, m), 2.57 (12H, m), 3.44 (4H, m), 4.30 (10H, m), 4.32 (8H, br s), 4.69 (4H, br s), 7.02 (2H, m), 8.78 (4H, s) ppm; IR (KBr) 1650 and 1630 cm-1. Anal. Calcd for C56H64N8O6Fe2‚1.5H2O: C, 62.00; H, 5.95; N,10.34. Found: C, 62.07; H, 6.05; N, 10.19. Equilibria and Kinetics of Binding. The binding ability of ligand 1 for double-stranded DNA was determined by Scatchard analysis using the condition probability method of McGhee and von Hippel.18 The binding abilities of 1 for denatured DNA, [poly(dA-dT)]2, and [poly(dG-dC)]2 were determined according to the Benesi-Hildebrand equation.19 Kinetic experiments were performed with an RSM 1000 rapidscan double-beam spectrophotometer (On-Line Instrument Systems Inc.). Single-wavelength kinetic records of absorbance versus time were collected. The measurements were made at 383 nm, where naphthalene diimide derivatives showed their maximum absorption. The first-order dissociation rate constant (kd) of 1 from intact or heat-denatured calf thymus DNA was determined by the stopped-flow method. Two kinds of reaction solutions (1% sodium dodecyl sulfate (SDS) and DNA-1 complex) were mixed instantaneously using a piston, and the change in the absorption (18) McGhee, J. D.; von Hippel, P. H. J. Mol. Biol. 1974, 86, 469. (19) Shimer, G. H., Jr.; Wolfe, A. R.; Meehan, T. Biochemstry 1988, 27, 7960.

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1335

spectrum was measured soon after mixing. The absorbance change arose from dissociation of the DNA-1 complex and formation of an SDS-1 complex.20 Immobilization of DNA on a Gold Electrode. Many papers have reported the immobilization of thiol-linked DNA on the gold electrode by chemisorption.11,12,21-23 Although the method is straightforward, it involves tricky aspects, and we found the following procedure gives reproducible results in DNA modification. A gold electrode with 2 mm2 in area was purchased from Bioanalytical Systems (BAS). Prior to DNA immobilization, it was soaked in boiling 2 M KOH for 1 h and washed with deionized water. The electrode was then soaked in concentrated nitric acid for 30 min and washed with deionized water. For chemisorption of DNA, a 1 µL solution of 5′-mercaptohexyloligonucleotide (∼200 pmol) was placed on a gold electrode held upside-down, and the end of the electrode was fitted with a rubber cap to protect the solution from evaporation. The assembly was kept standing for 2 h at room temperature. The electrode was then carefully washed with 500 µL of deionized water, and the wash water was subjected to analysis by HPLC for the oligonucleotide that remained unadsorbed. The total oligonucleotide after subtracting the remainder in the wash water was taken to be the amount immobilized. The density of immobilization of 5′-mercaptohexyloligonucleotide was found to be 20-30 pmol/mm2 on the electrode in many independent runs. The electrode thus prepared could be stored in sterile water in a refrigerator for at least 1 month. Electrodes of reduced DNA modification density were prepared analogously by coadsorption of 2-mercaptoethanol (2-ME). Thus, 1 fmol of thiol-linked oligonucleotide in 0.5 µL of water was placed on an electrode and allowed to stand for 2 h at room temperature while the electrode was protected from drying by fitting it with a rubber cap. The electrode was washed with water, and the water remaining on the electrode surface was removed by absorption with tissue paper. The density of modification could not be determined since the amount of probe DNA used was too small. One µL of 1 mM 2-ME in water was then placed on the electrode, which was protected from drying with a rubber cap, and the assembly kept at room temperature for 2 h. The thinly DNAimmobilized electrode thus obtained was washed with water and stored in sterile water until use. Hybridization on Oligonucleotide Probe-Immobilized Electrode. The gold electrode immobilized with a DNA probe was held upside-down, and a proper amount of DNA sample in a 1 µL aqueous solution was placed on the electrode surface as uniformly as possible. The end of the electrode was fitted with a rubber cap to protect the solution from evaporation. The assembly was kept at 25 °C for a desired period of time for hybridization. The electrode was then washed with deionized water and kept in sterile water in a refrigerator until use. Where the DNA sample was double stranded, it was denatured prior to application to the electrode. (20) Tanious, F. A.; Yen, S.-F.; Wilson, W. D. Biochemistry 1991, 30, 1813. (21) Okahata, Y.; Matsunobu, Y.; Ijiro, K.; Mukae, M.; Murakami, A.; Makino, K. J. Am. Chem. Soc. 1992, 114, 8299. (22) Peterlinz, K. A.; Georgiadis, R. M.; Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 3401. (23) Steel, A. B.; Herne, T. M.; Tarlov, M. J. Anal. Chem. 1998, 70, 4670.

1336

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

Electrochemical Measurements. All the measurements were performed with a BAS model CV-50W electrochemical analyzer. Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were carried out at 25 °C with a normal three-electrode configuration consisting of Ag/AgCl reference electrode, Pt counter electrode, and indicator electrode. The composition and pH of the electrolyte solution was varied in pursuit of optimum sensing conditions, and they generally fell in the ranger 0.04-0.1 M acetic acid-potassium acetate buffer (AcOH-AcOK, pH 5.2-5.6), 0-0.1 M KCl, and 0-1.0 mM 1. In some of the preliminary experiments, an electrolyte solution containing 30% dimethyl sulfoxide (DMSO) was also tested. CV and DPV were often considerably influenced by pH and the supporting electrolyte concentration. The solutions were deoxygenated by purging with argon for 20 min prior to measurements. Sensing Procedure A: Intercalation and Then Wash. The DNA-modified electrode was soaked in the aqueous solution of 1 mM 1 for 5 min and then washed with deionized water with vortexing for 5 s. The electrode was then subjected to electrochemical study in 50 mM AcOH-AcOK buffer (pH 5.2) without addition of 1. This procedure was based on the supposition that 1 was held tightly on the double-stranded DNA (but not on the single-stranded probe DNA) on the electrode, even after water wash, enough to afford an electrochemical signal. This proved right experimentally. Sensing Procedure B: Omission of Wash. The intercalation of 1 to the double-stranded DNA could be made in the same electrolyte solution where the electrochemical measurement was carried out. The DNA-modified electrode was dipped in a solution containing 0.1 M AcOH-AcOK buffer (pH 5.6), 0.1 M KCl, and 50 µM 1, and then the electrochemical study followed. The use of a very low concentration of ligand 1 was essential in order to reduce the current due to “free” 1 in solution, i.e., the ligand molecules that were not bound to DNA. Sensing with a Thinly DNA-Immobilized Electrode. The procedure was essentially the same as the foregoing one except that a thinly DNA-immobilized electrode with coadsorped 2-ME was used. After hybridization with sample DNA, the intercalation of 1 and the subsequent electrochemical measurement were carried out in a single step in a solution containing 0.1 M AcOHAcOK buffer (pH 5.6), 0.1 M KCl, and 50 µM 1. RESULTS Intramolecular Stacking between Ferrocene and Naphthalene Diimide in 1. Ligand 1 is soluble in both water and chloroform, and its spectral behavior in the UV and visible regions was studied in the two solvents. The two major absorption peaks of 1, at 360 and 380 nm, both originating from the naphthalene diimide nucleus, showed hypochromicity as much as 17 and 29%, respectively, in water, suggesting strongly that the naphthalene diimide and ferrocene planes in 1 are stacked in water. Hypochromicity was even more pronounced in buffered aqueous solution (data not shown). Intercalation of 1 into DNA. Ferrocenylnaphthalene diimide 1 underwent hypochromic and bathochromic shifts at the absorption band for the naphthalene diimide chromophore at 383 nm upon binding to sonicated calf thymus DNA. This was indicative of DNA intercalation. The view was reinforced by observing the CD spectra of sonicated calf thymus DNA in the absence and

Table 1. Binding Constants and Dissociation Rate Constants of 1 DNA

kd/s-1 a

K/M-1

calf thymus DNA heat-denatured calf thymus DNAd [poly(dA-dT)]2 [poly(dG-dC)]2

0.056 4.2

1.3 × 105 b (3 × 104)c (3 × 104)c (3 × 104)c

a All data were obtained from stopped-flow kinetic traces for the SDSdriven dissociation of 1 from DNA and kd values are the average defined as k1A1 + k2A2, where k and A values are the fractional amplitudes and the first-order rate constants, respectively, for a double exponential fit to the trace. b Scatchard analysis. c Benesi-Hildebrand plot. d Prepared by heating at 100 °C for 10 min and then immediately cooling on ice.

presence of 1. The CD spectrum of DNA alone was composed of negative and positive Cotton effects of similar intensity at around 245 and 270 nm, respectively, which was characteristic of B-form DNA. The CD spectra of its complex with 1 were nearly superimposable in this region, suggesting that the DNA structure remained intact upon binding to 1. In addition to the CD band of DNA, a new band was induced at ∼383 nm, which was ascribable to the naphthalene diimide chromophore of 1. These behaviors were similar to those of naphthalene diimide-type threading intercalators described previously.20 The intercalation of 1 into double-stranded DNA was confirmed further by viscometric titration of a supercoiled plasmid DNA with 1. Thus, the viscosity of a DNA solution first increased and then decreased as 1 was added. This phenomenon was explained in terms of relaxation and the subsequent reversal of the supercoils of DNA.24 All of these observations proved that 1 bound to double helical DNA by intercalation. Binding Affinity. Spectrophotometric titration was carried out for 1 with calf thymus DNA in 10 mM morpholinoethanesulfonate (MES) buffer and 1 mM ethylenediaminetetraacetate (EDTA) at pH 6.24 and 25 °C. The data obtained were analyzed by Scatchard plots and fitted with the theoretical curves generated by the binding equation with a binding constant of 1.3 × 105 M-1, site size of 2, and cooperativity parameter of 0.4 (Table 1). A similar spectral change was observed in the spectrophotometric titration of 1 with heat-denatured calf thymus DNA. Scatchard analysis was not successful in this case, however, due probably to the heterogeneity of the denatured DNA. We roughly estimated the binding constant to be 3 × 104 M-1 by Benesi-Hildebrand plots. The binding constant of 1 with double-stranded DNA was thus ∼4 times greater than that with single-stranded DNA. Similar spectral titrations were carried out for 1 with [poly(dA-dT)]2 and [poly(dG-dC)]2, and the data analyzed by the Benesi-Hildebrand equation yielded the binding constant of 3 × 104 M-1 for both of them (Table 1). Dissociation of the 1-Double-Stranded DNA Complex. As described above, 1 underwent a large hypochromic shift upon binding to double-stranded DNA. When the complex of 1 with DNA broke in the presence of SDS, the absorbance returned to that of free 1. The SDS-driven dissociation kinetics were studied at 20 °C for complexes of 1 with intact or denatured DNA in MESbuffered aqueous solution (data not shown). The first-order (24) Revet, B. M. J.; Schmir, M.; Vinograd, J. Nature 1971, 229, 10.

dissociation rate constants were determined from the time dependence of the absorbance change. Two exponential terms were needed to fit the trace within 0.1% error. The combined dissociation rate constant kd was obtained from the following equation, kd ) A1k1 + A2k2, where A and k refer to the fractional amplitudes and rate constants and the subscripts 1 and 2 refer to the two fit components. Table 1 summarizes the results. Cyclic Voltammetry of the DNA-Modified Electrode in the Presence of 1: General Study. First, CVs of a dT20-immobilized electrode (density of modification, 27 pmol/mm2) were measured at 100 mV/s before and after hybridization with dA20 (25 °C, 5 min). The electrolyte solution 41 mM AcOH-AcOK buffer (pH 5.2) containing 30% DMSO and 0.5 mM 1 was used. Before hybridization, a one-step redox reaction of the ferrocene moiety was observed on the positive side (E1/2 ) 502 mV, ∆Ep ) 79 mV). After hybridization with 54 pmol of dA20, E1/2 was shifted slightly toward the positive side (E1/2 ) 509 mV, ∆Ep ) 71 mV), and the peak current (Ipc) was decreased by 0.15 µA. These results demonstrated that the ferrocene moieties of 1 gave rise to a current with the DNA-immobilized electrode, but its magnitude was greater for single-stranded DNA than for double-stranded DNA, due presumably to the prevailing redox reaction by 1 present in large excess in the bulk solution but not bound to the electrode surface. To reduce such unfavorable concentration effects, the electrode pretreated with 1-containing solution was washed with deionized water with vortexing for 5 s. Then the electrochemical measurement was made in the electrolyte solution that did not contain 1. As expected, the current for the double-stranded DNA electrode was greater than that for single-stranded DNA by 0.13 µA after this treatment. This proved that nonspecifically bound 1 was washed out from the single-stranded DNA to a large extent. Thus, the procedure using an“intercalation of 1 and subsequent washing” technique (sensing method A) was established. Incidentally, DMSO was originally used to facilitate the dissolution of 1, which was slow in dissolving in water. However, since DMSO sometimes obscured the observed CVs, it was omitted in subsequent studies. It should be emphasized that this “intercalation and wash” technique is applicable only to 1, in our hands. Other electrochemically active ligands such as methylene blue, a classical intercalator,11 or Hoechst 33258, a groove binder,12 gave virtually no discernible signal in CV, when they were tested analogously (data not shown). This may be due to the fact that these classical intercalators, groove binders, and metal complexes do not show such high specificity in binding for double-stranded DNA, although they were reported to be capable of quantitating doublestranded DNA electrochemically. Detection and Quantitation of DNA Using Ligand 1 by Sensing Method A. We first tested this method in the quantitation of synthetic oligonucleotide dA20. One microliter of a sample solution of dA20 (60 pmol) was placed on a dT20 probe-immobilized gold electrode (density of modification, 32 pmol/mm2) to allow hybridization at 25 °C for 20 min. The electrode was then soaked in 1 mL of 1 mM aqueous solution of 1 for 5 min and washed with deionized water for 5 s with vortexing. The CVs and DPVs were then determined (Figure 1). In CV, a one-step redox reaction of 1 (E1/2 ) 485 mV, ∆Ep ) 50 mV) was observed after hybridization with dA20, whereas noncomplementary dT20 gave Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1337

Figure 1. Cyclic voltammograms (A) and differential pulse voltammograms (B) of 1 on a dT20-immobilized electrode (density of modification, 32 pmol/mm2) before (a) and after incubation with 60 pmol of dA20 (b) or dT20 (c) at 25 °C for 20 min. After soaking in an aqueous 1 mM solution of 1, the electrode was washed with deionized water for 5 s under vortexing. CVs and DPVs were measured in 50 mM AcOH-AcOK buffer (pH 5.2) containing 40 mM KCl with Ag/ AgCl and Pt wire as reference and counter electrodes, respectively. Pulse period, 200 ms; scan rate, 100 mV s-1; pulse amplitude, 50 mV; pulse width, 50 ms.

virtually no signal (compare (b) and (c) in Figure 1A). A current response at 509 mV was observed in DPV also after hybridization with dA20, while again virtually no current response was obtained for noncomplementary dT20 ((b) and (c) in Figure 1B). Figure 2A shows a correlation of the current response at 509 mV in CV and at 510 mV in DPV with the amount of dA20 (target) or dT20 (negative control) applied from 0 to 70 pmol. In both CV and DPV, the current response increased linearly with the amount of dA20 applied in the range 10-60 pmol, affording a good linear correlation coefficient over 0.990. When dA20 application exceeded 60 pmol, the response leveled off. This obviously reflected the amount of the immobilized dT20 probe on the electrode (64 pmol). We extended this sensing method to the detection of a natural DNA sample. The sample tested was a plasmid DNA carrying a part of the yeast choline transport gene. A unique 20-mer sequence of this gene, [d(CCGCTTATCTTCAGTTTTCA)], was immobilized on the electrode to 19 pmol/mm2 through thiol linking. Linearized plasmid DNA carrying the target sequence complementary to that of the probe was heated at 80 °C for 30 min and then allowed to hybridize with the DNA probe on the electrode (25 °C, 20 min). The current response at 510 mV increased with an increase in the amount of the target gene used, whereas no response was 1338 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

Figure 2. Correlation of the current response on a dT20-immobilized electrode (density of modification, 32 pmol/mm2) at 459 mV in CV and at 520 mV in DPV with the amount of dA20 (a) and dT20 (b) applied. The experiments were conducted under the conditions similar to those in Figure 1 and panels A and B depict picomolar and subpicomolar ranges of DNA applied, respectively.

observed for the plasmid DNA not carrying the target gene (Figure 3). The detection limit of 1 fmol was the same as that for dA20/dT20, proving that the present system applies equally well to the natural gene with a “complex” base sequence. Detection and Quantitation of DNA Using Ligand 1 by Sensing Method B. Sensing method A just described involves two important steps: intercalation of 1 and wash. To simplify the overall sensing procedure, conditions were searched for that could consolidate the two steps. It was found that this could be achieved by carrying out both intercalation and electrochemical measurements in a single electrolyte solution. The use of 1 in some critically low concentration (50 µM) proved essential. In combination with this, other conditions such as higher potassium chloride concentration (0.1 M) and higher pH (pH 5.6) were also important. Figure 4 shows the CVs recorded on a dA20-immobilized electrode (density of modification, 21 pmol/mm2) before and after hybridization with 42 pmol of dT20 (25 °C, 1 h). An anodic current at 476 mV was observed before hybridization, which was due to the binding of 1 to single-stranded probe DNA. After hybridization, this current increased dramatically as shown in Figure 4b, as was

Figure 4. CVs of a dA20-immobilized electrode (density of immobilization, 21 pmol/mm2) before (a) and after hybridization with 30 pmol of dT20 (b). CV of unmodified electrode (bare gold electrode) was also recorded (c). CVs were measured in 0.1 M AcOH-AcOK buffer (pH 5.6), 0.1 M KCl, and 50 µM 1 at a scan rate of 100 mV s-1.

Figure 3. (A) DPVs of d(CCGCTTATCTTCAGTTTTCG)-immobilized electrode (density of immobilization, 19 pmol/mm2) before (a) and after incubation with 1.0 fmol of plasmid DNA which carried (b) or did not carry (c) a part of the yeast choline transport gene. (B) Calibration curve for plasmid DNA which carried (a) or did not carry (b) the target gene. Supporting electrolytes, 50 mM AcOH-AcOK buffer (pH 5.2); scan rate,100 mV s-1; pulse amplitude, 50 mV.

expected. All the three CV curves (a-c in Figure 4) including the one for an unmodified gold electrode indicated a common feature giving greater anodic peak currents. The reason for this is not yet clear, though we have studied the electrochemistry of 1 in some detail.26 Figure 5A shows the DPV on the dA20-immobilized electrode after hybridization with a varying amount of target dT20. The magnitude of the current response increased in proportion to the amount of the hybridized sample, but the observed absolute current itself at a given amount of the sample differed by ∼20% among the electrodes. This happened because the density of probe DNA modification on individual electrodes varied from preparation to preparation. However, this effect could be corrected for by plotting i/io - 1 against the amount of DNA sample applied on (25) Takenaka, S.; Uto, Y.; Takagi, M.; Kondo, H. Chem. Lett. 1998, 989. (26) Takenka, S.; Yamashita, K.; Uto, Y.; Takagi, M.; Kondo, H. Denki Kagaku 1998, 66, 1329.

the electrode, where io and i represented the current before and after hybridization for a given amount of dT20, respectively. Figure 5B depicts the calibration curves obtained in this manner for target dT20 and nontarget dA20. The normalized current response (i/io - 1) at 460 mV showed good correlation with the amount of dT20 applied over the 10-15-10-11 mol range, whereas nontarget dA20 gave virtually no signal. Thus, the effect of variation in the preparation of electrode could be safely eliminated, and the complementary DNA could be quantitated down to femtomole levels. Sensing on the Electrode with a Low Density of DNA Immobilization. To improve the detection sensitivity further, better conditions of DNA immobilization were searched for. It was found that this could be achieved by reducing the density of probe DNA on the electrode and by masking the unmodified region of the electrode with 2-ME. Thus, an electrode modified with 5′-HS(CH2)6-p-dA20-3′ (1 fmol loading) was prepared as detailed in the Experimental Section. The electrode was allowed to hybridize with various amounts of dT20 (25 °C, 30 min) and then studied according to sensing method B (Figure 6A). First, it should be noted that a current response by 1 in the measuring solution was observed even on the electrode modified only with 2-ME; this response, as large as ∼5.2 µA, was taken as a background. The response by a probe-modified electrode in the absence of the right target was larger than this by ∼3.2 µA. This corresponded to the “unspecific” binding of 1 to the single-stranded DNA probe on the electrode, which seemed reasonable since the surface of the dA20-immobilized electrode should be negatively charged. Then, the addition of 0.01 amol target DNA induced a further 3.0 µA increase, but a further addition of 10 amol target DNA resulted in only 0.5 µA increase. The current responses shown in Figure 6A were normalized according to the method described above (Figure 6B). The analytical signal varied roughly linearly with the log [amount (mol)] value of the target DNA (dT20) in the range Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1339

Figure 5. (A) DPVs of a dA20-immobilized electrode (density of immobilization, 30 pmol/mm2) after incubation with a various amount of dT20: 0, 1, 10, 102, 103, and 104 fmol from bottom to top. DPVs were measured in 0.1 M AcOH-AcOK buffer (pH 5.6) containing 0.1 M KCl and 50 µM 1. Pulse period, 200 ms; scan rate, 100 mV s-1; pulse amplitude, 50 mV; pulse width, 50 ms. (B) Calibration curves for target dT20 (a) and negative control dA20 (b).

from 10-20 to 10-18 mol. It should be noted that the treatment with 2-ME was essential in order to obtain a reproducible response for all these measurements on thinly modified dA20 electrodes. The thiol strongly adsorbs on the gold surface, and it was presumed that 2-ME eliminated an undesired secondary interaction of the probe strand with the bare portion of the metal surface by making it chemically inert. DISCUSSION With the progress of molecular biology, the whole genomic structure has started to emerge for organisms with a smaller genome size and will become available soon for higher organisms with a more complex genome. In line with this progress, demand for gene analysis is increasing more than ever. Electrochemical gene analysis is one of the promising methods as far as the analytical sensitivity is concerned. A number of electrochemical techniques has been devised but none is in practical use for some 1340 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

Figure 6. (A) DPVs of 1 bound to the dA20-immobilized electrode (1 fmol of dA20 loaded) after incubation with a various amount of dT20. DPVs were measured in 0.1 M AcOH-AcOK buffer (pH 5.6) containing 0.1 M KCl and 50 µM 1. Pulse period, 200 ms; scan rate, 100 mV s-1; pulse amplitude, 50 mV; pulse width, 50 ms. The solid line in (A) refers to the electrode treated with 2-mercaptoethanol alone. Other lines refer to responses of the electrodes to DNA application of 0, 0.01, 0.1, 1.0, and 10 amol from bottom to top. (B) Calibration curves for target dT20 (a) and negative control dA20 (b).

reasons. For one thing, performance of DNA-immobilized electrodes is variable, which are prepared commonly by making use of an S-Au linkage between a single-stranded DNA probe and a gold electrode. Although this technique is straightforward, it is subject to a large variation in the performance of the electrodes thus prepared. Through trials and errors, we finally established a reproducible and reliable way of immobilization as described above. In essence, a commercial gold electrode, as received, was boiled in 2 M KOH for 1 h and then in concentrated nitric acid for 30 min. By this pretreatment, mercaptohexyloligonucleotide could be immobilized on the electrode consistently to the density of 20-30 pmol/mm2. The electrode thus prepared was successfully used for hybridization with target DNA. Another breakthrough in this work is the development of a ligand that discriminates double-stranded DNA from single stranded by a fairly large margin. This property is essential in developing a practial DNA sensor, and this was achieved by ferrocenylnaphthalene diimide 1, whose complex with double-stranded DNA is more stable thermodynamically and kinetically than that with single-stranded DNA. Following intercalation of 1 into double-

stranded DNA, preformed on the electrode by hybridization, the excess ligand is washed away by vortexing. Subsequent electrochemical analysis enabled us to quantitate synthetic oligonucleotides (dA20 and dT20) and natural gene fragment (choline transport gene) at the 1 fmol-50 pmol level. It later turned out that a similar level of sensitivity can be achieved by omitting the wash step but by running the intercalation at some critically low concentaration of 1 (method B). This simplified the overall manipulation considerably without impairing precision. To meet the requirements for biological samples, the sensitivity of detection needed to be enhanced. This was achieved by lowering the density of DNA modification on the electrode. As a result of this alteration, the electrochemical signal became very small and could be detected only by DPV, but as small as 10 zmol of target DNA could be detected. This sensitivity of detection compares well with the conventional nonradioactive methods such as those using chemiluminescence. Unlike such methods, however, the present method does not require modification of probe or sample DNA with some functional moieties. This is obviously one of the advantages of the present system, because the introduction of foreign groups to DNA could impair the formation of double strands in an unpredictable way. In summary, the ultrasensitive, simple, and rapid detection and quantitation of target DNA fragments were achieved by using a

DNA probe-immobilized gold electrode and electrochemically active, double-stranded DNA-specific ligand 1. The samples tested were complementary DNAs with no mismatch. In order for this system to become useful practically, it should discriminate DNAs with one or more mismatches. Experiments to test this feasibility are now under way in this laboratory. Furthermore, construction of a DNA microarray is being undertaken by integrating multiple electrodes on a small plate. The results of these undertakings will be published in due course. ACKNOWLEDGMENT We thank Prof. W. David Wilson of Georgia State University for advice and expert assistance with the stopped-flow measurement. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan. The authors are also grateful to the financial support from the Japan Society for the Promotion of Science.

Received for review September 8, 1999. Accepted January 3, 2000. AC991031J

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1341