Bioconjugate Chem. 1997, 8, 906−913
906
Probing Biomolecule Recognition with Electron Transfer: Electrochemical Sensors for DNA Hybridization Mary E. Napier,† Carson R. Loomis,‡ Mark F. Sistare,† Jinheung Kim,† Allen E. Eckhardt,‡ and H. Holden Thorp*,† Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290, and Xanthon, Inc., Park Research Center, Research Triangle Park, North Carolina 27709-2296. Received June 17, 1997X
Identifying infectious organisms, quantitating gene expression, and sequencing genomic DNA on chips all rely on the detection of nucleic acid hybridization. Described here is a novel assay for detection of the hybridization of products of the polymerase chain reaction using electron transfer from guanine to a transition-metal complex. The hybridization assay was modeled in solution by monitoring the cyclic voltammetry of Ru(bpy)32+ (bpy ) 2,2′-bipyridine) in the presence of a probe strand containing only A, T, and C prior to and after hybridization to a complement that contained seven guanines, which led to high catalytic current due to the oxidation of guanine by Ru(bpy)33+. To allow recognition of all four bases in the target sequence, it was shown that inosine 5′-monophosphate was 3 orders of magnitude less reactive than guanosine 5′-monophosphate, suggesting that effective hybridization sensors could be realized by immobilization of probe strands in which inosine was substituted for guanosine; hybridization to guanosine-containing target strands would then provide high catalytic currents. A sensor design was tested in a model system for the detection of a synthetic 21-mer oligonucleotide patterned on the sequence of the ras oncogene, which gave an increase in charge collected of 35 ( 5 µC after hybridization and of only 8 ( 5 µC after exposure to noncomplementary DNA. Independent quantitation of probe and target by radiolabeling showed that the hybridized electrode contained 3.0 ( 0.3 ng of target. New sensor electrodes were then prepared for the detection of PCR-amplified genomic DNA from herpes simplex virus type II, genomic DNA from Clostridium perfringens, and genomic RNA from human immunodeficiency virus and gave an additional charge of 35-65 µC for hybridization to complementary amplicon and of only 2-10 µC after exposure to noncomplementary DNA.
INTRODUCTION
Electron-transfer reactions of macromolecules are exceptionally well understood at the theoretical level (1, 2), and predictions from these models of the effects of distance and driving force on electron-transfer kinetics have been successfully tested (3, 4). Proper exploitation of this knowledge base should lead to new means for assessing the quantity or quality of the interactions of macromolecules with other biomolecules or small ligands, much in the same way that energy-transfer reactions have been used to study protein-protein complexes (5) and DNA hybridization (6). Energy-transfer approaches, and related photonic methods such as fluoresence polarization (7) and microscopy (8), generally require the attachment of chromophoric labels to one or both of the interacting partners. We report here on an electrontransfer approach to studying DNA hybridization that avoids the attachment of exogenous labels to the target strand by using engineered probes that do not undergo electron-transfer reactions as readily as natural DNA. Electron-transfer reactions of DNA have been of intense recent interest due to intriguing issues regarding the role of base stacking in the electronic coupling of redox partners (9-13). In particular, one-electron oxidation of the guanine nucleobase in native DNA has been detected by high-resolution gel electrophoresis via the formation of base-labile lesions at guanine following electron transfer (14-16). We showed that this one* Author to whom correspondence should be addressed. † University of North Carolina. ‡ Xanthon, Inc. X Abstract published in Advance ACS Abstracts, October 15, 1997.
S1043-1802(97)00114-6 CCC: $14.00
electron oxidation could also be realized by metal complex mediators that were activated electrochemically at potentials accessible in neutral aqueous solution with the appropriate electrode material (17). The guanine-metal electron transfer was detected as a catalytic current enhancement in the cyclic voltammogram of the mediator complex and was observed for complexes with potentials >1.0-1.1 V (all potentials versus Ag/AgCl) (18); later equilibrium titrations confirmed that the thermodynamic potential of guanine oxidation at pH 7 was E7 ) 1.06 V (19). The absolute rate constants for electron transfer can be determined from the cyclic voltammetry and offer a sensitive probe of the sequence and structure at the oxidized guanine via existing models for the effects of distance and driving force on the kinetics of electrontransfer reactions (20, 21). In particular, the kinetics of eq 2 have been used to detect single-base mismatches at guanine in solution (20). These observations suggest the engineering of solid-phase sensors by which biomolecular interactions of immobilized DNA are probed using guanine as the redox-active reporter. Because the metal complex is used to carry the electron to the electrode, the nucleic acid can be immoblized in a manner such that direct charge transfer from guanine to the electrode is not required and in which care is taken to preserve the native conformation and recognition properties of the biomolecule. The electrochemical detection of DNA at solid surfaces has been realized previously (22) via oxidation of guanine and adenine (23, 24); however, in these cases electron transfer is efficient only when the DNA is adsorbed onto the electrode surface or incorporated into a carbon paste electrode. In these latter cases, © 1997 American Chemical Society
Bioconjugate Chem., Vol. 8, No. 6, 1997 907
Electrochemical Sensors for DNA Hybridization
the integrity of the biomolecule structure and macromolecular interaction is a concern. The detection of nucleic acid hybridization at solid surfaces has been used for the identification of infectious organisms in clinical specimens (25, 26), the quantitation of mRNA for gene expression analysis (27), and the sequencing or resequencing of genomic DNA on highdensity “chip” arrays (28). Presently, these efforts involve the attachment of a fluorescent label to the target nucleic acid, which is then hybridized with a probemodified surface and detected after the unhybridized DNA has been washed away from the solid surface. Since detection of photons is required to signal hybridization, analysis of high-density arrays labeled in this manner requires high-resolution fluorescence microscopes. Alternatively, indirect detection of hybridization can be accomplished using sandwich assays when the surface-bound hybrid is subsequently hybridized to an additional signal probe that carries one or more fluorescent labels or enzymes that convert a nonfluorescent substrate to a fluorescent one (25). By attaching multiple enzymes to the signal probes, large signal amplification can be achieved (29); however, the preparation of these multiple enzyme systems is complex. We report here on a system in which the hybridization of unlabeled, native DNA can be detected electrochemically at membranemodified electrodes. Detection of DNA hybridization was realized in model oligonucleotide hybrids, and the practical utility of the sensors was confirmed by detection of the products of polymerase chain reactions (PCR) of genomic material from infectious organisms.
Figure 1. Cyclic voltammograms of Ru(bpy)32+ (25 µM) at a scan rate of 25 mV/s in 50 mM sodium phosphate buffer (pH 7) with 0.7 M NaCl: (A) no added oligonucleotide; (B) with 75 µM d(5′-TTTTATACTATATTT); (C) with the hybrid of the oligonucleotide from (B) with d(5′-GGGAAATATAGTATAAAAGGG). The secondary structure of the hybrid from (C) is indicated on the figure. Reference electrode: Ag/AgCl. Working electrode: unmodified ITO.
RESULTS
Solution Model. Before the modified electrode sensors were engineered, the hybridization assay was simulated in solution at unmodified indium tin oxide (ITO) electrodes to confirm that the redox chemistry was properly conceived. The mediator chosen was Ru(bpy)32+ (bpy ) 2,2′-bipyridine), which exhibits a reversible redox couple at 1.05 V and oxidizes guanine in DNA at high salt by a two-step mechanism (20, 21):
Ru(bpy)32+ f Ru(bpy)33+ + e3+
Ru(bpy)3
(1)
+ DNA f DNAox + Ru(bpy)3
2+
(2)
where DNAox is a DNA molecule in which guanine has been oxidized by one electron. We have shown previously that the rate constant for eq 2 in 50 mM phosphate buffer (pH 7) with 700 mM added NaCl is 1.0 × 104 M-1 s-1 for double-stranded DNA; at lower ionic strength (50 mM phosphate), the rate constant for eq 2 increases to 1.4 × 105 M-1 s-1 because of increased binding of the metal complex to DNA and a consequent increase in the local concentration (21). The simulations and kinetic analyses for these scenarios have been described in detail elsewhere (21). The quasi-reversible cyclic voltammogram of Ru(bpy)32+ is shown in Figure 1A. Addition of an oligonucleotide that does not contain guanine produces a small enhancement in the oxidation current (Figure 1B). This oligonucleotide simulates the probe strand in our hybridization assay. Upon hybridization of the oligonucleotide to a longer complement that contains a single guanine in the duplex region and an overhang of three guanines on each end, dramatic catalytic enhancement is observed (Figure 1C). These results suggest a very effective hybridization assay by which, if the probe strand containing no guanines were immobilized to the electrode,
Figure 2. Cyclic voltammograms of (A) Ru(bpy)32+ (25 µM) with (B) inosine 5′-monophosphate (0.3 m) and (C) guanosine 5′-monophosphate (0.3 mM). Scan rate: 25 mV/s. Reference electrode: Ag/AgCl. Working electrode: unmodified ITO.
little current enhancement would be observed, but after hybridization to a strand containing multiple guanines, large current enhancements would be produced. One limitation to the scenario suggested by Figure 1 is that the probe strand could not contain guanine, so a sequence comprised only of A, T, and C was chosen. Such a design would drastically limit the choice of sequences chosen for hybridization. To circumvent this problem, we chose a commercially available guanine derivative that would still recognize cytidine but not donate an electron to Ru(bpy)33+. Shown in Figure 2 are the cyclic voltammograms of Ru(bpy)32+ in the presence of the mononucleotides guanosine 5′-monophosphate (GMP) and inosine 5′-monophosphate (IMP), in which the guanine base has been substituted by the deaminated hypoxanthine analogue. As expected, a large current enhancement is observed for the guanine mononucleotide, but not for the hypoxanthine mononucleotide, apparently because the exocyclic amine plays a critical electronic role in the guanine oxidation. Digital simulation of the scan rate dependences of the voltammograms shown in Figure 2 gives second-order rate constants for electron transfer to Ru(bpy)33+ of kGMP ) 6.4 × 105 M-1 s-1 and kIMP ) 97 M-1 s-1. Hypoxanthine still recognizes cytidine, so now the engineering of probe strands that recognize all four bases can be envisioned by replacement of guanosine with inosine, and the lower electron-transfer
908 Bioconjugate Chem., Vol. 8, No. 6, 1997
Napier et al.
Figure 4. Difference in charge collected in cyclic voltammetry experiments following hybridization of electrode modified with the appropriate probe from Table 1 with the corresponding PCR compared to the current obtained before hybridization. Control reactions involved exposure of the electrode modified with the 21-mer ras probe and the PCR. Data were collected and analyzed as in Figure 3A and gave similar error limits. The negative charge indicated for HSV results from a greater current for the electrode modified with the control probe before hybridization compared to after hybridization.
Figure 3. (A) Cyclic voltammograms taken at 25 mV/s of Ru(bpy)32+ (100 µM) at the polymer-modified ITO working electrode showing the current (i) measured at the electrode with only the inosine-substituted 21-mer probe attached (dotted) and after exposure to a 200 µL solution of 1 nmol of the complementary (solid) and control (dashed) oligonucleotides. Inset: Integrated charge (in microcoulombs) obtained after subtraction of the current for the unhybridized electrode (dotted) from that after hybridization to the complementary and control oligonucleotides. The charge is 35 ( 5 µC for the hybridization reaction and 8 ( 5 µC for the control. (B) Chronoamperometric traces taken with a potential step from 0 to 1.3 V with Ru(bpy)32+ (100 µM) showing current collected at the electrode modified with the 21mer probe (dotted) and after exposure to the control (dashed) and complementary (solid) oligonucleotides. The dotted and dashed lines appear nearly superimposable on this scale. Inset: Integrated charge in microcoulombs as in (A). (C) Data from (B) replotted on an expanded scale to show the relationship of the complementary (solid), control (dashed), and probe only (dotted) scans.
reactivity of hypoxanthine by nearly 3 orders of magnitude will lead to relatively low currents for the unbound probe. Since hypoxanthine can only form two of the three hydrogen bonds in a Watson-Crick base pair, it may be desirable in the future to use a guanine derivative that is redox-inert but capable of forming all three hydrogen bonds. Accordingly, the reactivity of the nucleotide of 7-deazaguanine was determined as in Figure 2, and the results were very similar to those with IMP (k ) 850 M-1 s-1). In the study of PCR products discussed below, the
specificity afforded by inosine substitution was sufficient (Figure 4); however, 7-deazaguanine is clearly an available and attractive alternative. Sensor Electrodes. As shown by Marchand-Brynaert et al. (30), oxidation of the alcohol chain ends of tracketched microporous poly(ethylene terephthalate) membranes with KMnO4 results in the formation of carboxylic acid functionalities that can be linked to proteins and amino acids via carbodiimide-catalyzed reaction of amine groups to form amide linkages. We show here that single-stranded DNA probes can also be coupled to the carboxylic acid groups of the polymer membrane via the exocyclic amines of the nucleobases. The DNA attachment was followed by X-ray photoelectron spectroscopy, which showed changes at each step similar to those observed for amino acid functionalization (30) except that peaks due to ionization of phosphorus were apparent following DNA coupling. Similarly, cyclic voltammograms of the membranes exposed to probe with no watersoluble carbodiimide (WSC) resembled the voltammograms of membranes with no exposure to DNA, demonstrating a low degree of nonspecific DNA binding. When attached to ITO electrodes, the covalently attached DNA will be accessible to Ru(bpy)33+, which can shuttle electrons from guanine to the electrode surface according to eqs 1 and 2. If inosine-substituted oligonucleotides are used as immobilized probes, then relatively small catalytic currents will be observed. If the immobilized inosine-substituted probes are then hybridized to strands containing guanine, then large catalytic currents are expected due to the shuttling of electrons from the hybridized DNA to the electrode by the Ru(bpy)32+ mediator. The sensor design was first tested using synthetic 21mer oligonucleotides patterned on the ras oncogene sequence (31) (Table 1). No current above background was observed for the probe-modified electrodes in the absence of Ru(bpy)32+ before or after hybridization, indicating that direct electron transfer from the immobilized DNA to the electrode did not occur. However, measurable current due to DNA was always observed in cyclic voltammograms of Ru(bpy)32+ at polymer membranes modified with the inosine-subsituted 21-mer probe and after hybridization. As shown in Figure 3A, a significant increase in current was observed for the polymer film undergoing hybridization to the complementary 21-mer DNA (solid) when compared to the
Electrochemical Sensors for DNA Hybridization Table 1. Oligomer Sequences Used as Probes and Primers oligomer sequence 21-mer ras probe 21-mer complementary 21-mer control HIV primers HIV probe HSV primersa HSV probea C. perfringens primersb C. perfringens probeb a
5′-ITACTCTTCTTITCCAICTIT 5′-ACAGCTGGACAAGAAGAGTAC 5′-ACATCGAGCTTAAGGTGTCGC 5′-CCGGAATTCTGCAACAACTGCTG 5′-CCGCTCGAGATGCTGGTCCCA 5′-AAACAAATTCCACAAACTTGC 5′-CGACATCAACCACCTTCGCT 5′-ATGTAGCACGAGGCTGTCGT 5′-ICCCICACCATCCAACCACCC 5′-TGCTAATGTTACTGCCGTTGATAG 5′-ATAATCCCAATCATCCCAACTATG 5′-CAAAAIAATATICAAIATITT
Reference 36. b Reference 35.
polymer film containing only the inosine-substituted probe (dotted) or after exposure to the 21-mer control sequence (dashed). This current increase suggests that the hybridization event was successful and that electron transfer from the guanines of the hybridized strand to Ru(bpy)33+ is responsible for the increase in oxidation current. In contrast, very little current increase is observed for the polymer membrane exposed similarly to the 21-mer control DNA. The small increase in current is attributed to a small amount of nonspecific binding of the DNA to the polymer membrane. The inset in Figure 3 shows the charge (in microcoulombs, µC) acquired for the hybridization and control reactions after subtraction of the charge acquired at the probe-modified electrode. Measurement of the charge collected at five entirely different modified electrodes gives an error of (5 µC. Exposure of the probe-modified electrodes to 200 µL of 2 mM (in nucleotide) calf thymus DNA produced no current enhancement, demonstrating a significant degree of sequence discrimination. The 21-mer system was also evaluated by chronoamperometry experiments in which the potential was stepped from a resting potential of 0 V to 1.3 V, where Ru(bpy)32+ is oxidized. The current was collected for 10 s after stepping the potential; the results for the first 5 s are shown in Figure 3B. As in the cyclic voltammetry experiments, some catalytic current is observed with the probe-modified electrode (dotted) with a much greater increase after hybridization (solid) than after exposure to the control oligonucleotide (dashed). The quantity of charge collected in Figure 3B determined by subtraction of the current at the probe-modified electrode and integration over the entire time period (inset) was similar to that collected in the cyclic voltammetry experiments. The majority of guanine in the film was consumed in both amperometry conducted for 10 s and cyclic voltammetry at a scan rate of 25 mV/s; accordingly, the electrodes could not be re-interrogated after the first scan by either method. While the insets in Figure 3 clearly show that the total charge is significantly greater in both experiments for the hybridization compared to the control, the chronoamperometry data also show significant enhancements in the instantaneous current at early times in the catalytic cycle. The chronoamperometry data are shown on an expanded scale in Figure 3C, where at t ) 0.05 s there is a 38 µA enhancement in the hybridized case while there is only a 2 µA enhancement for the control. The quantity of attached and hybridized DNA was independently determined using 32P-labeled oligonucleotides that were then quantitated by scintillation counting. First, the 21-mer probe was radiolabeled and attached to the film, and the quantity of attached probe was determined. This experiment showed that 8.2 ( 0.8 pmol of probe was attached to the electrode. In a second
Bioconjugate Chem., Vol. 8, No. 6, 1997 909
series of experiments, the unlabeled 21-mer probe was attached to the electrode and then exposed to radiolabeled 21-mer complement. This experiment showed that 0.43 ( 0.04 pmol of oligonucleotide was hybridized to the film. Thus, the electrodes used in Figure 3 contained 3.0 ( 0.3 ng of hybridized target DNA, and the ratio of hybridized target to total attached probe was 5.2 ( 0.5%. A number of experiments support the idea that the immobilized DNA resembles its native structure and conformation. The effect of salt concentration in the hybridization buffer is the same as for solution hybridization (32); that is, more efficient and stable hybridization occurred at high salt concentration. Similarly, the effect of salt concentration on the guanine-metal electron transfer was the same as we have observed previously in solution (21): more efficient electron transfer was observed at low salt, at which there is more binding of the metal complex to DNA. The relative reactivities of single-stranded and duplex guanine were similar to that observed in solution (20). We have used the complex Os(bpy)32+, which is identical in structure to Ru(bpy)32+ but not sufficiently oxidizing to abstract an electron from guanine (33, 34), to assess whether the films impede diffusion of the mediator. Cyclic voltammograms of Os(bpy)32+ were identical at the unmodified polymer, probemodified polymer, and hybridized polymer, demonstrating that the macromolecule does not impede diffusion of the mediator. Finally, the thermal denaturation characteristics for poly(dC)‚poly(dG) models were the same in solution and on the polymer film. The utility of the sensors was demonstrated by showing that the sensitivity and specificity were sufficient for detection of PCR products of biologically significant length in unpurified amplification reactions. Genomic DNA from HSV and Clostridium perfringens was amplified using PCR as described elsewhere (35, 36), and the unpurified PCR were hybridized to the polymer films modified with the appropriate probes. Genomic RNA from HIV was amplified by RT-PCR (26) and similarly hybridized. The cyclic voltammograms of the probemodified and hybridized films were collected along with the appropriate controls. The charge passed during each voltammogram (in microcoulombs) was determined as in Figure 3, inset. The bar graph in Figure 4 shows the differential charge obtained following subtraction of the charge measured at the probe-modified films from that measured at the hybridized films. Controls involved exposing films modified with the 21-mer ras probe to the raw PCR of the HIV, HSV, or C. perfringens genomes under the same conditions. An increase in the charge above that for the probe-modified films signals either nonspecific binding in the control case or a successful hybridization event for the complementary case. In all three cases, the complementary hybridization showed a significant increase in charge when compared to the control exposure. The charge increase signifies that the HIV, HSV, and C. perfringens amplicons successfully hybridized to the probe attached to the polymer film. In contrast, little or no residual charge was observed for the controls, indicating a minimal amount of nonspecific binding of amplicon or other components of the unpurified PCR to the polymer membrane. The specificity is emphasized by the ability of the electrodes to discriminate large PCR amplicons (247 bp for C. perfringens, 561 bp for HSV, and 3.2 kb for HIV) and by the fact that constituents of the PCR, such as genomic DNA or RNA, polymerase, mononucleotides, and primers, do not interfere with the reaction. Finally, controls performed using the HSV probe against the C. perfringens PCR and vice
910 Bioconjugate Chem., Vol. 8, No. 6, 1997
versa were identical with those obtained using the 21mer control probe. DISCUSSION
A notable feature of our approach is that the unlabeled target DNA is distinguished from the synthetic probes by a difference in the electron-transfer reactivity. In our assay, instead of distinguishing duplex and singlestranded DNA [as in ethidium bromide fluorescence (37)], we distinguish probe and target. The importance of this difference is that our assay is not dependent on bringing double-stranded DNA to the electrode. Thus, our assay is equally well suited for single- or double-stranded DNA as a target. Even more important, single-stranded RNA is an equally suitable target since it contains guanine, which is of particular interest in identifying unamplified genomic RNA from viruses such as HIV or hepatitis C (29), in detecting ribosomal RNA from bacteria (38), or in quantitating cellular mRNA for gene expression analysis (27). For the studies described here, the probes were attached to the membrane via the endogenous exocyclic amines of the nucleobases, as has also been done for direct attachment to glassy carbon electrodes (22, 39). We have prepared films that behave similarly with synthetic oligonucleotides to which an alkylamine linker was appended. The latter method provides for greater hybridization efficiency and specificity and will be used in the future as required; attachment via the endogenous amines was apparently sufficient for these studies. Further, the data in Figure 3 suggest that the probe strands are present in excess since hypoxanthines are 3 orders of magnitude less reactive than guanines (Figure 2), while the current at the probe-modified electrode prior to hybridization is measurable. Accordingly, the quantitation of the radiolabled probes shows that there is in fact a 20-fold excess of probe to target at the hybridized electrode. The ability to tune the electron-transfer reactivity of the synthetic probe allows us to maintain the probe strand in excess, which greatly increases the hybridization efficiency, while maintaining requisite sensitivity in terms of the total charge collected (Figure 3, insets) and in chronoamperometric currents (Figure 3C). The ability to vary both the electron-transfer reactivity and loading of the probes allows for tuning of sensitivity and hybridization efficiency in individual assays. To our knowledge, Figure 4 shows the first use of electrochemical sensors to detect PCR amplicons, although such strategies have been suggested (40). The hybridization reactions were performed with no purification of the reaction mixture following PCR, which indicates that nonspecific binding of other DNA in the reaction mixture did not occur. Further, the results in Figure 2 and elsewhere clearly show that guanosine mononucleotides are particularly reactive, but the results in Figure 4 show that the membranes did not adsorb dGTP, which was present in millmolar concentration in the PCR and would have contributed to the catalytic current in the control reaction if bound to the membrane. As discussed above, the methodology was effective for amplicons varying in length from 247 bp for C. perfringens up to 3.2 kb for HIV. The strategy reported here is distinct from related approaches in which hybridization can be detected without labeling the target nucleic acid. The other electrochemical sensors that operate via voltammetric activation can be divided into two groups. In the first group, the redox chemistry of the nucleobases is monitored by direct electrochemistry (23, 40). This strategy requires that the
Napier et al.
DNA is immobilized in a manner that allows electron transfer from the nucleobases directly to the electrode, which is accomplished either by adsorption of the nucleic acid to mercury or by inclusion in carbon paste. In this strategy, maintaining the native structural and recognition properties of the nucleic acid is a concern. The other approach is to monitor the electrochemistry of an exogenous redox indicator that binds more tightly to doublestranded DNA than to single strands (22, 39, 41-43). In this case, higher currents are observed when doublestranded DNA is brought to the electrode. This strategy will be less attractive for single-stranded targets, such as those discussed above, because the only duplex brought to the electrode by the hybridization event will be the small region where the probe is hybridized to the single-stranded target. Other biosensor approaches to hybridization detection are strategies based on surface plasmon resonance (SPR) (44) or the quartz crystal microbalance (QCM) (45), both of which require considerably more expensive instrumentation and immobilization surfaces than the electrochemical apparatus described here with the polymer-modified ITO electrode. The advantages of SPR, QCM, and the double-stranded redox indicator approaches are that engineered (i.e., inosinesubstituted) probes are not required and that neither technique consumes the analyzed DNA as occurs with the guanine oxidation described here. In the future, we envision numerous applications of this approach beyond the detection of PCR amplicons. The attraction of electrochemical biosensors for DNA diagnostics has been discussed elsewhere (41), but, briefly, electrochemical techniques are particularly suited to miniaturization, which brings about a consequent increase in sensitivity due to both the ability to detect small currents and the small volumes analyzed by small electrodes (46, 47). The studies here were conducted using cyclic voltammetry and chronoamperometry; however, modern pulsed methods may offer additional advances in sensitivity and precision (48). The ability of eq 2 to distinguish a single-base mismatch in solution (21) should be conveniently translated to the solid-phase format described here. EXPERIMENTAL METHODS
Reagents and DNA. The inorganic reagents used in these experiments were of analytical grade or higher. HCl and H2SO4 were obtained from Fisher Scientific (Pittsburgh, PA). Water-soluble carbodiimide [WSC, 1-[3(dimethylamino)propyl]-3-ethylcarbodiimide hydrochloride] and potassium permanganate were obtained from Aldrich (Milwaukee, WI). Na2HPO4, NaH2PO4, and NaCl were obtained from Maillinckrodt (Phillipsburg, NJ). Water was obtained from a Milli-Q Plus purification system (Millipore, Bedford, MA). Synthetic oligonucleotides (Table 1) were synthesized by the University of North Carolina Department of Pathology and purified using Amicon Micron 3 concentrators with a cutoff of 3000 molecular weight. Genomic DNA from herpes simplex II virus (HSV) and C. perfringens bacterial DNA was obtained from Sigma (St. Louis, MO). The RT-PCR of a 3.2 kb fragment from the gp160 gene from HIV was performed using the primers in Table 1 as described elsewhere (26) and was a gift from Trimeris. Preparation and Activation of the Carboxylated Polymer Membranes. Cyclopore poly(ethylene terephthalate) track-etched membranes with a pore size of 0.4 µm and a diameter of 25 mm were obtained from Whatman International (Hillsboro, OR). Four circular sample disks, ∼8 mm in diameter, were cut from each 25 mm membrane. The carboxylated polymer mem-
Electrochemical Sensors for DNA Hybridization
branes were prepared following an adaptation of a published procedure (30). The polymer disks were treated with a solution of KMnO4 in 1.2 N H2SO4 (2.5 g/50 mL) for ∼18 h at room temperature. The polymer disks were then washed with 6 N HCl (2 × 30 min, 25 °C) to remove the brown manganese oxide followed by water rinses (3 × 30 min, 25 °C). The carboxylation of the polymer film was confirmed by XPS analysis with an increase in the O/C ratio from 0.363 to 0.398 following treatment. These results and those obtained at each step in the functionalization compared favorably to those obtained by others in protein immobilization (30). The activation of the surface carboxylate moieties was performed by application of 30 µL of a freshly prepared 10 mM WSC in 20 mM sodium phosophate buffer (pH 7.0) to each side of the polymer disk. After each application, the polymer was allowed to dry. The polymer disks were then rinsed twice with 20 mM sodium phosphate buffer and once with water. The DNA probes were coupled to the activated polymer membranes by application of 0.75 nmol of probe in 10 µL of water to each side of the membrane to give a surface density of ∼5 nmol/cm2. The polymer was allowed to dry after each application. The polymer disks were then rinsed twice with 20 mM sodium phosphate buffer and once with water. The polymer disks with the probe attached were now ready for hybridization. PCR Amplification. All DNA samples were amplified using the PCR method as described elsewhere (35, 36). The reaction tubes contained 80 µL of a solution of ∼1 pmol (in nucleotide) sample DNA, 10 mM Tris-HCl (pH 8.4), 50 mM KCl, 200 µM dNTP, and either 2.5 mM MgCl2 (HIV and HSV) or 3 mM MgCl2 (C. perfringens). The primer sequences (Table 1) were for a 247 bp fragment of the R-toxin gene of C. perfringens (35, 49), for a 561 bp amplicon from the BglII N fragment from HSV (36), or for a 3.2 kb amplicon from the gp160 gene of HIV. The primers were added to a final concentration of 4 µM, and 1-2 units Taq polymerase (Perkin-Elmer, Norfolk, CT) was added to the mixture using the hotstart method of preheating the reaction tube to 94 °C for 5 min. The mixtures were subjected to 45 cycles at 94 °C for 1 min, 55 °C for 1 min, and 70 °C for 2 min with an automatic thermal cycler (Perkin-Elmer Model 2400). Each set of amplification reactions contained both positive and negative controls, and each mixture was analyzed by agarose gel electrophoresis to verify that fragments of the appropriate size were produced and that the positive and negative controls behaved appropriately. Synthetic 21-mer Hybridization. Polymer films to which an inosine-substituted 21-mer probe (Table 1) was coupled were placed into 200 µL of a hybridization buffer of 50 mM sodium phosphate and 800 mM NaCl with 1 nmol of the complementary or control 21-mer DNA. The hybridization buffer and polymer film were heated for 1 h at 50 °C and slowly cooled to room temperature. The polymer membrane was removed from the liquid and washed twice in 20 mM sodium phosphate buffer and once with water prior to electrochemical analysis. Hybridization of PCR Products. Polymer films modified with the appropriate inosine-substituted probe (Table 1) were placed into the hybridization buffer with a 60 µL fraction of the unpurified HIV, HSV, or C. perfringens PCR and heated for 10 min at 95 °C in a total volume of 200 µL. The polymer films were allowed to hybridize in the buffer overnight. As a control, the noncomplementary inosine-substituted ras 21-mer probe was coupled to a polymer membrane and exposed similarly to the PCR. The polymer membrane was removed from the liquid and washed twice in 20 mM sodium
Bioconjugate Chem., Vol. 8, No. 6, 1997 911
phosphate buffer and once with water prior to electrochemical analysis. Electrochemical Analysis. Cyclic voltammograms were collected using a PAR 273A potentiostat/galvanostat with a single-compartment voltammetric cell equipped with an ITO working electrode (area 0.32 cm2), a Pt-wire counter electrode, and an Ag/AgCl reference electrode. In a typical experiment, the polymer membrane was placed on top of the ITO electrode at the base of the voltammetric cell and held in place with an O-ring. Two hundred microliters of 100 µM Ru(bpy)32+ in 50 mM sodium phosphate buffer (pH 7.0) was placed into the voltammetric cell and allowed to equilibrate with the polymer film for 20 min. The polymer film covered all of the exposed ITO electrode area so that the Ru(bpy)32+ was forced to diffuse through the film to access the working electrode. The porosity of the polymer membrane allowed for adequate diffusion of the Ru(bpy)32+. Cyclic voltammograms from 0.0 to 1.3 V were taken at a scan rate of 25 mV/s. A freshly cleaned ITO electrode was used for each experiment, and a background scan of buffer alone was collected and subtracted from subsequent scans. The quantities of charge (in microcoulombs) shown in the bar graphs in Figures 3 and 4 were obtained by integrating the cyclic voltammograms between 900 and 1300 mV and subtracting the charge for the probemodified films from the charge for the hybridization and control reactions. Quantitation of Probe and Target by Radiolabeling. The 21-mer probe and complementary target oligonucleotides were 5′-32P-labeled using T4 polynucleotide kinase and γ-32P-ATP (6000 Ci/mmol) according to standard procedures (32), and the unreacted ATP was removed by chromatography over a Stratagene NucTrap column. The radiolabeled 21-mer probe (370 cpm/pmol) was attached to the PET membrane by the same strategy described above. The films were washed with buffer and water as in the attachment for electrochemistry, and then the membrane was added to 4 mL of SafetySolve scintillation fluid and counted using a scintillation counter. In a separate experiment, a film to which unlabeled 21-mer probe was attached was hybridized to 44 µM 21-mer complementary target that was 32P-labeled (360 cpm/ pmol). The quantity of hybridized target was determined similarly. ACKNOWLEDGMENT
This research was supported by the David and Lucile Packard Foundation and by Xanthon, Inc. We gratefully acknowledge the assistance of Trimeris in the RT-PCR amplification of HIV RNA. LITERATURE CITED (1) Beratan, D. N., Onuchic, J. N., Winkler, J. R., and Gray, H. B. (1992) Electron-Tunneling Pathways in Proteins. Science 258, 1740-1741. (2) Marcus, R. A., and Sutin, N. (1985) Electron Transfers in Chemistry and Biology. Biochim. Biophys. Acta 811, 265. (3) Wuttke, D. S., and Gray, H. B. (1993) Protein Engineering as a Tool for Understanding Electron Transfer. Curr. Opin. Struct. Biol. 3, 555-563. (4) Wuttke, D. S., Bjerrum, M. J., Winkler, J. R., and Gray, H. B. (1992) Electron-Tunneling Pathways in Cytochrome c. Science 256, 1007-1009. (5) Mitra, R. D., Silva, C. M., and Youvan, D. C. (1996) Fluorescence Resonance Energy Transfer Between BlueEmitting and Red-Shifted Excitation Derivatives of the Green Fluorescent Protein. Gene 173, 13-17. (6) Parkhurst, K. M., and Parkhurst, L. J. (1995) DonorAcceptor Distance Distributions in a Double-Labeled Fluo-
912 Bioconjugate Chem., Vol. 8, No. 6, 1997 rescent Oligonucleotide Both as a Single Strand and in Duplexes. Biochemistry 34, 293-300. (7) Walker, G. T., Nadeau, J. G., Linn, C. P., Devlin, R. F., and Dandliker, W. B. (1996) Strand displacement amplification (SDA) and transient-state fluorescence polarization detection of Mycobacterium tuberculosis DNA. Clin. Chem. 42, 9-13. (8) Fodor, S. P. A., Rava, R. P., Huang, X. C., Pease, A. C., Holmes, C. P., and Adams, C. L. (1993) Multiplexed Biochemical Assays with Biological Chips. Nature 364, 555-556. (9) Arkin, M. R., Stemp, E. D. A., Holmlin, R. E., Barton, J. K., Ho¨rmann, A., Olson, E. J. C., and Barbara, P. F. (1996) Rates of DNA-Mediated Electron-Transfer Between Metallointercalators. Science 273, 475-480. (10) Priyadarshy, S., Risser, S. M., and Beratan, D. N. (1996) DNA Is Not a Molecular Wire: Protein-like Electron Transfer Predicted for an Extended π-Electron System. J. Phys. Chem. 100, 17678-17682. (11) Olson, E. J. C., Hu, D., Ho¨rmann, A., and Barbara, P. F. (1997) Quantitative Modeling of DNA-Mediated Electron Transfer between Metallointercalators. J. Phys. Chem. B 101, 299-303. (12) Melvin, T., Botchway, S., Parker, A. W., and O’Neill, P. (1995) Migration of Photoinduced Oxidative Damage in Models for DNA. Chem. Commun., 653-654. (13) Meade, T. J., and Kayyem, J. F. (1995) Electron Transfer in Synthetic Oligonucleotides Labeled with Ruthenium. Angew. Chem., Int Ed. Engl. 34, 352-354. (14) Hall, D. B., Holmlin, R. E., and Barton, J. K. (1996) Oxidative DNA Damage Through Long-Range Electron Transfer. Nature 384, 731-735. (15) Sugiyama, H., and Saito, I. (1996) Theoretical Studies of GG-Specific Photocleavage of DNA via Electron Transfer: Significant Lowering of Ionization Potential and 5′-Localization of HOMO of Stacked GG Bases in B-Form DNA. J. Am. Chem. Soc. 118, 7063-7068. (16) Armitage, B., Yu, C., Devadoss, C., and Schuster, G. B. (1994) Cationic Anthraquinone Derivatives as Catalytic DNA Photonucleases: Mechanisms for DNA Damage and Quinone Recycling. J. Am. Chem. Soc. 116, 9847-9859. (17) Johnston, D. H., Welch, T. W., and Thorp, H. H. (1996) Electrochemically Activated DNA Oxidation. Metal Ions Biol. Syst. 33, 297-324. (18) Johnston, D. H., Cheng, C.-C., Campbell, K. J., and Thorp, H. H. (1994) Transdioxorhenium(V)-Mediated Electrocatalytic Oxidation of DNA at Indium Tin-Oxide Electrodes: Voltammetric Detection of DNA Cleavage in Solution. Inorg. Chem. 33, 6388-6390. (19) Steenken, S., and Jovanovic, S. V. (1997) How Easily Oxidizable is DNA? One-Electron Reduction Potentials of Adenosine and Guanosine Radicals in Aqueous Solution. J. Am. Chem. Soc. 119, 617-618. (20) Johnston, D. H., Glasgow, K. C., and Thorp, H. H. (1995) Electrochemical Measurement of the Solvent Accessibility of Nucleobases Using Electron-Transfer Between DNA and Metal Complexes. J. Am. Chem. Soc. 117, 8933-8938. (21) Johnston, D. H., and Thorp, H. H. (1996) Cyclic Voltammetry Studies of Polynucleotide Binding and Oxidation by Metal Complexes: Homogeneous Electron-Transfer Kinetics. J. Phys. Chem. 100, 13837-13843. (22) Millan, K. M., Saraullo, A., and Mikkelsen, S. R. (1994) Voltammetric DNA Biosensor for Cystic Fibrosis Based on a Modified Carbon Paste Electrode. Anal. Chem. 66, 29432948. (23) Wang, J., Chicharro, M., Rivas, G., Cai, X., Dontha, N., Farias, P. A. M., and Shiraishi, H. (1996) DNA Biosensor for the Detection of Hydrazines. Anal. Chem. 68, 2251-2254. (24) Palecek, E., and Fojta, M. (1994) Differential Pulse Voltammetric Determination of RNA at the Picomole Level in the Presence of DNA and Nucleic Acid Components. Anal. Chem. 66, 1566-1571. (25) Spargo, C. A., Haaland, P. D., Jurgensen, S. R., Shank, D. D., and Walker, G. T. (1993) Chemiluminescent detection of strand displacement amplified DNA from species comprising the Mycobacterium tuberculosis complex. Mol. Cell. Probes 7, 395-404.
Napier et al. (26) Martin, W. J. (1994) Infectious Diseases. In The Polymerase Chain Reaction (K. B. Mullis, F. Ferre´and, and R. A. Gibbs, Eds.) pp 406-417, Berkhauser, Boston, MA. (27) Schena, M., Shalon, D., Davis, R. W., and Brown, P. O. (1995) Quantitative Monitoring of Gene Expression Patterns with a Complementary DNA Microarray. Science 270, 467470. (28) Chee, M., Yang, R., Hubbell, E., Berno, A., Huang, X. C., Stern, D., Winkler, J., Lockhart, D. J., Morris, M. S., and Fodor, S. P. A. (1996) Accessing Genetic Information with High-Density DNA Arrays. Science 274, 610-613. (29) Holodniy, M., Mole, L., Margolis, D., Moss, J., Dong, H., Boyer, E.. Urdea, M., Kolberg, J., and Eastman, S. (1995) Determination of Human Immunodeficiency Virus RNA in Plasma and Cellular Viral DNA Genotypic Zidovudine Resistance and Viral Load during Zidovudine-Didanosine Combination Therapy. J. Virol. 69, 3510-3516. (30) Marchand-Brynaert, J., Deldime, M., Dupont, I., Dewez, J.-L., and Schneider, Y.-J. (1995) Surface Functionalization of Poly(ethylene terephthalate) Film and Membrane by Controlled Wet Chemistry: Chemical Characterization of Carboxylated Surfaces. J. Colloid Interface Sci. 173, 236244. (31) Neubaeur, A., Dodge, R. K., George, S. L., Davey, F. R., Silver, R. T., Schiffer, C. A., Mayer, R. J., Ball, E. D., WursterHill, D., Bloomfield, C. D., and Liu, E. T. (1994) Prognostic Importance of Mutations in the ras Proto-Oncogenes in De Novo Acute Myeloid Leukemia. Blood 83, 1603-1611. (32) Maniatis, T., Fritsch, E. F., and Sambrook, J. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Press, Cold Spring Harbor, NY. (33) Welch, T. W., Corbett, A. H., and Thorp, H. H. (1995) Electrochemical Determination of Nucleic Acid Diffusion Coefficients through Noncovalent Association of a RedoxActive Probe. J. Phys. Chem. 99, 11757-11763. (34) Welch, T. W., and Thorp, H. H. (1996) Distribution of Metal Complexes Bound to DNA Determined by Normal Pulse Voltammetry. J. Phys. Chem. 100, 13829-13836. (35) Daube, G., China, B., Simon, P., Hvala, K., and Mainil, J. (1994) Typing of Clostridium perfringens by in vitro amplification of toxin genes. J. Appl. Bacteriology 77, 650-655. (36) Lulitanond, V., Chantratita, W., Thammaprasert, K., Nimmanahaeminda, K., Matangkasombut, P., and Yoosook, C. (1994) Detection of herpes simplex virus type 2 Bgl II N fragment in paraffin-embedded cervical tissue sections using nested polymerase chain reaction. Mol. Cell. Probes 8, 441447. (37) Waring, M. J. (1965) Complex Formation Between Ethidium Bromide and Nucleic Acids. J. Mol. Biol. 13, 269. (38) Scholin, C. A., Villac, M. C., Buck, K. R., Krupp, J. M., Powers, D. A., Fryxell, G. A., and Chavez, F. P. (1994) Ribosomal DNA Sequences Discriminate Among Toxic and Non-Toxic Pseudonitzschia Species. Nat. Toxins 2, 152-165. (39) Millan, K. M., and Mikkelsen, S. R. (1993) SequenceSelective Biosensor for DNA Based on Electroactive Hybridization Indicators. Anal. Chem. 65, 2317-2323. (40) (a) Palecek, E. (1996) From Polarography of DNA to Microanalysis with Nucleic Acid-Modified Electrodes. Electroanalysis 8, 7-14. (b) Mikkelsen, S. R. (1996) Electrochemical Biosensors for DNA Sequence Determination. Electroanalysis 8, 15-19. (41) (a) Wang, J., Cai, X., Rivas, G., Shiraishi, H., Farias, P. A. M., and Dontha, N. (1996) DNA Electrochemical Biosensor for the Detection of Short DNA Sequences Related to the Human Immunodeficiency Virus. Anal. Chem. 68, 26292634. (b) Wang, J., Palecek, E., Nielsen, P. E., Rivas, G., Cai, X., Shiraishi, H., Dontha, N., Luo, D., and Farias, P. A. M. (1996) Peptide Nucleic Acid Probes for Sequence-Specific DNA Biosensors. J. Am. Chem. Soc. 118, 7667-7670. (42) Kelley, S. O., Barton, J. K., Jackson, N. M., and Hill, M. G. (1997) Electrochemistry of Methylene Blue Bound to a DNA-Modified Electrode. Bioconjugate Chem. 8, 31-37. (43) Hashimoto, K., Ito, K., and Ishimori, Y. (1994) SequenceSpecific Gene Detection with a Gold Electrode Modified with
Bioconjugate Chem., Vol. 8, No. 6, 1997 913
Electrochemical Sensors for DNA Hybridization DNA Probes and an Electrochemically Active Dye. Anal. Chem. 66, 3830-3833. (44) Peterlinz, K. A., Georgiadis, R. M., Herne, T. M., and Tarlov, M. J. (1997) Observation of Hybridization and Dehybridization of Thiol-Tethered DNA Using Two-Color Surface Plasmon Resonance Spectroscopy. J. Am. Chem. Soc. 119, 3401-3402. (45) Okahata, Y., Matsunobu, Y., Ijiro, K., Mukae, M., Murakami, A., and Makino, K. (1992) Hybridization of Nucleic Acids Immobilized on a Quartz Crystal Microbalance. J. Am. Chem. Soc. 114, 8299-8300. (46) Fan, F.-R. F., Kwak, J., and Bard, A. J. (1996) Single Molecule Electrochemistry. J. Am. Chem. Soc. 118, 96699675.
(47) Licht, S., Cammarata, V., and Wrighton, M. S. (1989) Time and Spatial Dependence of the Concentration of Less Than 105 Microelectrode-Generated Molecules. Science 243, 11761178. (48) Osteryoung, J. (1993) Voltammetry for the Future. Acc. Chem. Res. 26, 77. (49) Titball, R. W., Hunter, S. E. C., Martin, K. L., Morris, B. C., Shuttleworth, A. D., Rubidge, T., Anderson, D. W., and Kelly, D. C. (1989) Molecular Cloning and Nucleotide Sequence of the Alpha-Toxin (Phospholipase C) of Clostridium perfringens. Infect. Immun. 57, 367-376.
BC9701149
