Analytica Chimica Acta 375 (1998) 197±203
Indicator-free electrochemical DNA hybridization biosensor Joseph Wanga,*, Gustavo Rivas1,a, Joao Roberto Fernandes2,a, Jose Luis Lopez Paza,b, Mian Jianga, Russel Waymirea a
Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, NM 88003, USA Departamento de QuõÂmica Analitica, Facultad de Ciencias QuõÂmicas, Universidad de Valencia, Burjasot, Valencia, Spain Received 29 March 1998; received in revised form 18 June 1998; accepted 30 June 1998
Abstract A new electrochemical hybridization biosensor protocol without an external indicator is described. The biosensor format involves the immobilization of inosine-substituted (guanine-free) probe onto the carbon paste transducer, and a direct chronopotentiometric detection of the duplex formation by the appearance of the guanine oxidation peak of the target. Such a use of the intrinsic DNA electrochemical response for monitoring hybridization events offers several advantages (over the common use of external indicators), including the appearance of a new peak, a ¯at background, or simplicity. A 4 min short hybridization period allows a detection limit around 120 ng/ml. Performance characteristics of the sensor are described along with future prospects. # 1998 Elsevier Science B.V. All rights reserved. Keywords: DNA hybridization; Biosensors; Guanine; Inosine; Carbon paste electrodes
1. Introduction DNA biosensors based on nucleic-acid recognition processes are rapidly being developed towards the goal of rapid and inexpensive testing of genetic and infectious diseases. Electrochemical transducers are often being used for detecting the DNA hybridization event, due to their high sensitivity, small dimensions, low cost, and compatibility with microfabrication technology [1,2]. Such devices rely on the immobili*Corresponding author. Tel.: +1-505-646-2505; fax: +1-505 646 2649; e-mail: [email protected]
1 Permanent Address: Departamento FõÂsico QuõÂmica, Universidad Nacional de CoÂrdoba, CoÂrdoba, Argentina. 2 Permanent Address: Departamento de QuõÂmica/FC/UNESP, PO Box 473, Bauru SP 17033-360, Brazil. 0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0003-2670(98)00503-0
zation of a single-stranded (ss-) probe onto the electrode surface to recognize through base pairing the complementary (target) DNA strand in a sample solution. The duplex formation is commonly being detected in connection with the use of an appropriate electroactive hybridization indicator [3±6]. Most indicator-based electrochemical DNA biosensors have used cationic metal complexes (e.g., Co
phen3 3 or [3,6]) or intercalating organic compounds Co
bpy3 3 (e.g., acridine orange ) that interact in a different way with ss- and ds-DNA. The increased electrochemical response due to the indicator association with the surface duplex thus serves as the hybridization signal. Transition metal complexes (e.g., Ru
bpy2 3 , mediating the electrooxidation of the guanine base, have also been used by Thorp's group for the detection of DNA hybridization . Little attention has been
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203
given to the development of electrochemical DNA biosensors that do not require an external redox-active tag or mediator. In this paper we report on the development of electrochemical biosensors for direct monitoring of DNA hybridization without the use of an external redox indicator. The method relies on the intrinsic DNA signals, associated with the electroactivity of nucleic-acid bases . Previously, we exploited the sensitivity of the guanine oxidation signal to the DNA structure for detecting the duplex formation [9,10]. In particular, the decreased guanine response of the immobilized probe upon the formation of the surface duplex was used for monitoring the hybridization event. However, such a measurement of the decreased anodic signal of the immobilized probe is very limited, as it cannot be used for detecting targets containing guanine bases. Such a limitation is overcome in the present work through the use of immobilized inosinesubstituted (guanine-free) probes. While the inosine moiety preferentially forms base pair with the target cytosine residue , its oxidation signal is well separated from the guanine response. Direct and convenient detection of the DNA hybridization can thus be accomplished through the appearance of the target guanine oxidation signal. The attractive performance characteristics of the new indicatorfree electrochemical detection scheme are described in the following sections. 2. Experimental 2.1. Apparatus Constant-current chronopotentiometric measurements were performed in a usual three-electrode electrochemical cell (4 ml, Kimble Glass, Vineland, NJ), using a TraceLab potentiometric stripping unit (PSU 20 Radiometer, Denmark) in connection with an IBM PS/2 55SX computer. According to the TraceLab protocol, the potentials were sampled at a frequency of 30 kHz and the derivative signal (dt/dE) versus potential (E) was recorded following baseline ®tting. The peak area served as the analytical signal. The threeelectrode system consisted of a carbon-paste working electrode (CPE), a Ag/AgCl reference electrode
(Model RE-1, BAS, W. Lafayette, IN) and a platinum wire auxiliary electrode which entered the cell through a Te¯on cover. 2.2. Reagents The 29-mer inosine-substituted DNA probe (50 -AIA CIA TCA IAT ACC ITC ITA ITC TTA AC-30 ), 29-mer guanine-containing DNA-probe (50 -AGA CGA TCA GAT ACC GTC GTA GTC TTA AC-30 ), 29-mer DNA target (50 -GTT AAG ACT ACG ACG GTA TCT GAT CGT CT-30 ) and 27-mer non-complementary oligonucleotide (50 -GTC GTC AGA CCC AAA ACC CCG AGA GGG-30 ) were purchased from Life Technologies (Grand Island, NY, USA) as their ammonium salts. Sodium chloride (Cat. no. S3014), monosodium phosphate (Cat. no. S3139), sodium acetate buffer (3 M, pH 5.20.1 at 258C, Cat. no. S7899), Tris± HCl buffer (1.000.05 M, pH 7.000.05 at 258C, Cat. no. T 2413) and Tris±EDTA (TE) buffer (100 concentrate, 1.0 M Tris±HCl, 0.1 M EDTA, pH 8.0, Cat. no. T9285) were received from Sigma. The above reagents were free of DNase and RNase. Tris (1,10phenanthroline) cobalt(III) perchlorate was synthesized at NMSU using the method described by Dollimore and Gillard . The surfactant Tween 20 (Cat. no. 27434-8) and polyvinylalcohol (PVA) (Cat. no. P 8136) were obtained from Aldrich and Sigma, respectively. Stock solutions of the target, probe and non-complementary oligonucleotides (nominally 1000 mg/l) were prepared with the TE buffer solution (1 concentrate, 10 mM Tris±HCl, 1 mM EDTA, pH 8.0) and stored in the freezer until use. Sterile distilled water was used for preparing all the solutions. Fresh (<2 h) probe solutions were used for the surface immobilization. All glassware, containers, pipette tips and the cells (with the exception of the electrodes) were sterilized by autoclaving for 30 min. Sterilized water was used to rinse the electrodes prior to use. 2.3. Electrode preparation Carbon paste was prepared in the usual way by hand-mixing graphite powder (Grade 38, Cat. no. G67-500, Fisher Scienti®c, Pittsburgh, PA) and mineral oil (Cat. no. M5904, free of DNase, RNase and
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203
protease, Sigma) in a 70:30 mass ratio. The surface was polished on a weighing paper to a smoothed ®nish before use. The body of the working electrode was a Te¯on sleeve (3.5 mm, i.d.) tightly packed with the carbon paste. Electrical contact was provided by a stainless steel screw. 2.4. Procedure Each measurement involved the immobilization/ hybridization/detection cycle at a fresh carbon paste surface. All the experiments were performed at room temperature (22.00.5)8C. 2.4.1. Indicator-free hybridization protocol Electrode pretreatment. The CPE was activated by applying 1.6 V for 1 min in a stirred 0.2 M acetate buffer solution pH 5.0. Probe immobilization. The 29-mer inosine-substituted probe was immobilized on a pretreated CPE by applying a potential of 0.5 V for 2 min in a stirred acetate buffer solution (0.2 M, pH 5.0) containing 25 mg/l guanine-free oligonucleotide. The electrode was then rinsed with a solution containing 0.5 M NaCl and 0.02 M phosphate buffer solution pH 7.0 for ca. 5 s. Hybridization. The probe-coated electrode was immersed into the stirred hybridization solution (0.02 M phosphate solution pH 7.00.5 M NaCl containing the 29-mer target or 27-mer non-complementary sequences) for the desired time while holding the potential at 0.5 V. The electrode was subsequently rinsed with 0.2 M acetate buffer solution pH 5.0 for ca. 5 s. Chronopotentiometric transduction. The guanine oxidation signal was measured after transferring the electrode to a blank acetate buffer solution using constant-current chronopotentiometry with an initial potential of 0.5 V and a current of 6 mA. 2.4.2. Indicator
phen3 3 based protocol Some comparative experiments used the Co
phen3 3 redox hybridization indicator. The surface pretreatment and hybridization conditions were similar to those employed in the indicator-free protocol (with the exception of using a guanine-containing probe). After the hybridization the electrode was rinsed with 0.02 M Tris±HCl solution pH 7.0 for ca.
5 s and then it was immersed in a stirred Tris±HCl buffer solution containing 0.1 mM Co
phen3 for 3 1 min while holding the potential at 0.5 V. After rinsing the electrode with Tris buffer solution, it was transferred to an unstirred Tris solution where the surface-accumulated Co
phen3 3 was measured using a constant-current chronopotentiometry with an initial potential of 0.5 V and a current of ÿ6 mA. The difference between the indicator peak areas in the presence and absence of the target served as the hybridization signal. 3. Results and discussion Computerized chronopotentiometry was employed for assessing and demonstrating the new indicator-free hybridization biosensor due to its effective discrimination against background contributions at carbon transducers and convenient detection of the guanine DNA peak [10,13]. A 29-mer inosine substituted oligonucleotide probe, complementary to a sequence speci®c to the deadly pathogen Cryptosporidium parvum , was immobilized by adsorption onto the carbon paste surface. Fig. 1 compares chronopotentiograms for increasing levels of the 29-mer target (between 1 and 3 mg/l, (b±d)), as well as for the absence of the target (a), obtained with the new indicator-free approach (A) and through the common electroactive indicator (B). use of the Co
phen3 3 (The latter employed a regular guanine-containing probe.) The use of the redox indicator results in a large blank response (at ca. 0.1 V), related to the association of Co
phen3 3 with the single-stranded probe (B,a). This indicator peak increases after successive additions of the Cryptosporidium-DNA target. Quantitative work thus relies on measuring the difference in peak areas (with and without the target). In contrast, the new indicator-free approach, offers a ¯at baseline and no peak in the absence of the target (A,a). Well-de®ned and sharp guanine peaks (Ep1.01 V) are observed in the presence of increasing levels of the target. The area or height of the guanine response can thus serve as the hybridization signal. Such a use of the intrinsic electroactivity of the target not only eliminates the need for external indicators, but offers a more reliable detection of the duplex formation, as it relies on the appearance of a new (guanine) peak ±
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203
Fig. 2. Response of the inosine-substituted probe-modified electrode following 4 min hybridization in a phosphate-buffer/sodiumchloride solution containing 0 (a) and 6 (b) mg/l of the 29-mer target, as well as in the presence of 6 mg/l of the noncomplementary sequence (c). Also shown (d) is the corresponding response of the bare carbon paste electrode to the 6 mg/l noncomplementary oligonucleotide. Other conditions as in Fig. 1(A). Fig. 1. Comparison of the response of the indicator-free (A) and indicator-based (B) electrochemical sensing strategies. Target guanine (A) and [Co(phen)3]3 (B) chronopotentiometric signals at the probe 29-mer inosine-substituted and 29-mer DNA-modified carbon pastes, respectively, in the presence of increasing 29-mer target concentrations: 0 (a), 1 (b), 2 (c) and 3 mg/l (d). Pretreatment: 1 min at 1.6 V in a stirred 0.2 M acetate buffer solution, pH 5.0; probe accumulation: 2 min at 0.5 V in a stirred 0.2 M acetate buffer solution, pH 5.0 containing 25 mg lÿ1 of 29mer inosine-substituted DNA (A) or 29-mer DNA (B) probes. Hybridization: 4 min at 0.5 V in a stirred phosphate buffer solution (0.02 M, pH 7.0) containing 0.5 M NaCl and different target concentrations. Indicator association (B): 1 min at 0.5 V in a stirred Tris±HCl buffer solution, pH 7.0 containing 0.1 mM [Co(phen)3]3. Chronopotentiometric transduction in an unstirred 0.2 M acetate (A) or Tris±HCl (B) buffer solutions with a constant current of 6 (A) or ÿ6 (B) mA and an initial potential of 0.5 V.
without a background signal ± rather than on the increased indicator response. The attractive performance of Fig. 1 can be understood in terms of the electroactivity of the nucleic-acid bases, and particularly the separation of the guanine and inosine chronopotentiometric peaks. Fig. 2 displays the response of the probe-modi®ed electrode, over a wider potential range, following a 4 min incubation in the blank solution (a), target solution (b), and in the presence of the non-complementary oligonucleotide (c). The incubation in the blank solution results in a large peak at 0.77 V, associated with
the oxidation of the probe inosine residue (a). A second peak (Ep1.01 V), related to the oxidation of the target guanine moieties is observed following hybridization in the target solution (b). Such a peak, which serves as the hybridization signal, can be better quantitated without recording the preceding inosine signal (and hence in connection with a more sensitive scale). Signi®cantly (30-fold) smaller guanine response is observed upon replacing the target solution with that of the non-complementary sequence (c). Such a guanine signal is attributed to the oxidation of guanine residues in the non-speci®cally adsorbed non-complementary oligomer. Also shown in Fig. 2 is the corresponding response of the non-complementary sequence at the bare carbon paste electrode (d). As expected from the absence of immobilized probe layer, the non-complementary oligonucleotide can be readily adsorbed onto the carbon paste to yield a large guanine response (note the different scales). The ratio of the non-complementary guanine peak area (bare/coated surfaces) is around 400, indicating an effective minimization of non-speci®c adsorption effects by the probe layer. Careful attention to the probe immobilization conditions is crucial for reducing the guanine contribution of non-speci®cally adsorbed oligonucleotides. In general, the surface coverage of the probe (and hence the minimization of non-speci®c adsorption) is depen-
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203
dent on the probe concentration and its adsorption time. Different probe concentrations ranging 5± 25 mg/l were assessed in connection with 0.5 to 10 min adsorption times. For example, the use of a 5 mg/l probe solution and a 5 min immobilization resulted in a large non-speci®c adsorption contribution. Best results were obtained using a 25 mg/l probe solution in connection with a 2 min adsorption. With this probe solution, the inosine response (of the adsorbed probe) increased rapidly with the adsorption time at ®rst (up to 2 min) and then it started to level off, indicating attainment of full surface coverage. The use of blocking agents (such as Tween 20 or PVA) further minimized non-speci®c adsorption effects but greatly compromised the reproducibility of the hybridization signal. Under the optimal probe immobilization conditions (and in the absence of blocking agents), the guanine hybridization response is ca. 30-fold larger than the guanine contribution of the non-complementary sequence (at an equal concentration). Occasionally (1 out of 20 runs) the non-complementary strand displayed larger than expected guanine signals. It should be pointed out that nonspeci®c adsorption effects can also greatly affect the results of indicator-based hybridization sensors . The guanine hybridization signal is strongly affected by the surface pretreatment. While no guanine response was observed at the untreated carbon paste, a well de®ned one was observed following a short (1 min) pretreatment at 1.6 V (or higher potentials up to 1.8 V). It should be pointed out that the inosine probe response is less affected by this treatment (only a 4-fold increase with a de®ned peak even at the untreated surface). Such a behavior indicates that the pretreatment effect is mainly upon the electrochemistry of purine bases, with a lesser effect on the interfacial accumulation. The immobilized probe strongly adheres to the pretreated surface. No apparent decrease in the probe inosine peak was observed upon immersing the coated electrode in a stirred phosphatebuffer blank solution over periods (up to 15 min) relevant to the hybridization experiments. Such a behavior is similar to that common for guanine-containing probes [10,13], indicating that the inosine substitution has little effect upon the stability of the adsorbed probe. The positive hybridization potential (0.5 V) further enhances the stability of the immobilized probe (through electrostatic attraction of the
Fig. 3. (A) Chronopotentiograms obtained following different hybridization times: 0 (a), 2 (b), 4 (c), 6(d) and 8 min (e) using a target solution of 2 mg/l. (B) Dependence of the hybridization signal upon the hybridization time. Other conditions as in Fig. 1(A).
negatively charged phosphate backbone). Yet, such a potential is lower than that required for the inosine oxidation (ca. 0.77 V), and thus has little effect upon the hybridization properties of this base. Fig. 3(A) displays chronopotentiograms for a solution containing 2 mg/l of the 29-mer Cryptosporidium target following different hybridization times (0± 8 min, (a)±(e)). Well-de®ned guanine peaks are observed, indicating that short hybridization periods are suf®cient for detecting the duplex formation. Also shown (Fig. 3(B)) is the resulting plot of peak area vs. hybridization time. The hybridization signal increases linearly with the hybridization time at ®rst (up to 6 min) and levels off above 8 min. Analogous timedependent experiments using a 10 mg/l solution of the non-complementary sequence (i.e., 5-fold excess in comparison to the target) displayed a negligible (<2 ms) guanine response up to 7 min hybridization. Longer times resulted in a larger guanine peak, indicating some displacement of the adsorbed probe. A judicious selection of the hybridization time is thus
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203
Fig. 4. Calibration plots for increasing levels of the target (A) and non-complementary (B) sequences. Hybridization time 4 min. Other conditions as in Fig. 1(A).
required for minimizing non-speci®c adsorption effects. Fig. 4(A) displays a typical calibration plot obtained with the indicator-free chronopotentiometric strategy using a 4 min hybridization time. The guanine hybridization signal increases linearly with the concentration of the Cryptosporidium target sequence up to 3 mg/l, and levels off at higher concentrations (slope of the linear portion, 18 ms l/mg). Shorter hybridization times can be used to extend the linear range, in connection to some loss in sensitivity. For example, with 2 min hybridization, linearity prevailed up to 10 mg/l with a sensitivity of 6 ms l/mg (not shown). Also shown in Fig. 4 is the corresponding plot for increasing concentrations of the non-complementary oligonucleotide (B). The latter displays a negligible response up to 3 mg/l, with very small signals appearing thereafter. For example, at the 4 and 6 mg/l levels, the response for the non-complementary sequence is 20- and 16-fold smaller than that of the target. Measurements of 250 mg/l of the target oligonucleotide were used for estimating the detection limit (not shown). A relatively short hybridization time of 6 min yielded a de®ned guanine signal of 6.4 ms, and a detection limit of around 120 mg/l (based on S/N3). A series of 10 repetitive measurements of the 2 mg/l target oligomer resulted in an average hybridization signal of 30.6 ms, and a relative standard deviation of
11.4% (4 min hybridization; not shown). Improved reproducibility and convenience may be achieved in connection with mechanically-renewable bulk modi®ed carbon composite electrodes . In conclusion, we have demonstrated the utility and advantages of an indicator-free sequence-speci®c hybridization biosensor based on the target guanine oxidation process at the carbon paste transducer. Such a use of the intrinsic DNA redox signal for monitoring hybridization events offers several advantages over the use of external indicators, including the appearance of a new peak, ¯at background or simplicity. The indicator-free operation is well suited for other carbon electrode transducers (including ``one shot'' disposable strips). Proper attention to the attachment of the guanine-free probe, through the use of different immobilization schemes and/or blocking agents, should further minimize non-speci®c adsorption effects. The use of guanine-free peptide±nucleic acid (PNA) probes should couple the indicator-free detection with the high speci®city inherent to PNA recognition . Work is in progress in these directions. We hope that this and similar developments will accelerate the adaptation of electrochemical devices in DNA diagnostics. Acknowledgements JW acknowledges the ®nancial support from the NM Water Resources Research Institute. GR acknowledges the ®nancial support from Universidad Nacional de Cordoba, Argentina. JRF and JLP acknowledge fellowships from FAPESP, Brazil and NATO, respectively. References  S.R. Mikkelsen, Electroanalysis 8 (1996) 15.  J. Wang, X. Cai, G. Rivas, H. Shiraishi, N. Dontha, Biosens. Bioelectron. 12 (1997) 587.  K.M. Millan, S.R. Mikkelsen, Anal. Chem. 65 (1993) 2317.  K. Hashimoto, K. Ito, Y. Ishimori, Anal. Chem. 66 (1994) 3830.  K. Hashimoto, K. Miwa, M. Goto, Y. Ishimoro, Supramolecular Chem. 2 (1993) 265.  J. Wang, G. Rivas, X. Cai, Electroanalysis 9 (1997) 395.  D.H. Johnston, K.C. Glasgow, H.H. Thorp, J. Am. Chem. Soc. 117 (1995) 8933.
J. Wang et al. / Analytica Chimica Acta 375 (1998) 197±203  E. Pallecek, Electroanalysis 8 (1996) 7.  J. Wang, X. Cai, J. Wang, C. Jonsson, E. Palecek, Anal. Chem. 67 (1995) 4065.  J. Wang, X. Cai, B. Tian, H. Shiraishi, Analyst 121 (1996) 965.  S. Casegreen, E. Southern, Nucleic acids Res. 22 (1994) 131.  L. Dollimore, R. Gillard, J. Chem. Soc., Dalton Trans. (1973) 933.
 J. Wang, X. Cai, G. Rivas, H. Shiraishi, Anal. Chim. Acta 326 (1996) 141.  J. Wang, G. Rivas, C. Parrado, X. Cai, M. Flair, Talanta 44 (1997) 2003.  J. Wang, E. Palecek, P. Nielsen, G. Rivas, X. Cai, H. Shiraishi, N. Dontha, D. Luo, P. Farias, J. Am. Chem. Soc. 118 (1996) 7667.  J. Wang, J.R. Fernandes, L.T. Kubota, Anal. Chem. 70 (1998) 3699.