DNA Electrochemical Biosensor for the Detection of Short DNA

Mary E. Napier, Carson R. Loomis, Mark F. Sistare, Jinheung Kim, Allen E. ..... Nyström , Martin Malmsten , Lovisa Ringstad , Andrew Nelson , Paul Ge...
0 downloads 0 Views 176KB Size
Download PDF
Anal. Chem. 1996, 68, 2629-2634

DNA Electrochemical Biosensor for the Detection of Short DNA Sequences Related to the Human Immunodeficiency Virus Joseph Wang,* Xiaohua Cai, Gustavo Rivas,† Haruki Shiraishi,‡ Percio A. M. Farias,§ and Narasaiah Dontha

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, New Mexico 88003

An electrochemical biosensor for the detection of short DNA sequences related to the human immunodeficiency virus type 1 (HIV-1) is described. The sensor relies on the immobilization and hybridization of the 21- or 42-mer single-stranded oligonucleotide from the HIV-1 U5 long terminal repeat (LTR) sequence at carbon paste or strip electrodes. The extent of hybridization between the complementary sequences is determined by the enhancement of the chronopotentiometric peak of the Co(phen)33+ indicator. Numerous factors affecting the probe immobilization, target hybridization, and indicator binding reactions are optimized to maximize the sensitivity and speed the assay time. A detection limit of 4 × 10-9 M HIV-1 U5 LTR segment is reported following a 30 min hybridization. The hybridization biosensor format obviates the use of radioisotopes common in radioactive methods for the detection of HIV-1 DNA. We also report on the direct adsorptive chronopotentiometric stripping measurements of trace levels of various HIV-1 DNAs. There has been considerable interest in developing reliable methods for detecting and quantifying the human immunodeficiency virus type 1 (HIV-1).1,2 The standard diagnostic test for HIV infection is an ELISA for the HIV antibody in which viral antigens are adsorbed onto a solid phase.2,3 A Western blot assay, coupling electrophoretic separation with radioisotopic detection, is also common for this task.2 Alternately, nucleic acid hybridization schemes have been proposed for detecting HIV-1 DNA sequences.4-7 These include radioisotopic assays, with the HIV-1 probe labeled with the 32P isotope4 or nonisotopic hybridization procedures employing colorimetric measurements.5,6 Both of these hybridization strategies rely on prolonged (2-3 h) hybrid† Permanent address: Department of Physical Chemistry, Universidad Nacional de Cordoba, Cordoba, Argentina. ‡ Permanent address: Department of Chemistry, Ritsumeikan University, Kusatsu, Japan. § Permanent address: Department Chemistry, Pontificia Universidad Catolica do RJ, Rio de Janeiro, Brazil. (1) Gallo, R.; Montagnier, L. Sci. Am. 1988, 259 (10), 41. (2) Kuby, J. Immunology; W. Freeman Inc.: New York, 1991; Chapter 23. (3) Nishanian, P.; Huskins, K.; Stehn, S.; Detels, R.; Fahey, J. J. Infect. Dis. 1990, 162, 21. (4) Davis,G.; Blumeyer, K.; DiMichele, L.; Whitfield, K.; Chappelle, H.; Riggs, N.; Ghosh, S.; Kao, P.; Fhay, E.; Kwoh, D.; Guatelli, J.; Spector, S.; Richman, D.; Gingeras, T. J. Infect. Dis. 1990, 162, 13. (5) Mulder, J.; Mckinney, N.; Christopherson, C.; Sninsky, J.; Greenfeld, L.; Kwok, S. J. Clin. Microbiol. 1994, 32, 292. (6) Livache, T.; Fouque, B.; Teoule, R. Anal. Biochem. 1994, 217, 248. (7) Rapier, J.; Villamazo, Y.; Schochetman, G.; Ou, C.; Brakel, C.; Jonegan, J.; Maltzman, W.; Lee, S.; Kirtiker, D.; Galita, D. Clin. Chem. 1993, 39, 244.

S0003-2700(96)00243-0 CCC: $12.00

© 1996 American Chemical Society

ization times, with the former complicated by the short half-life and hazardous nature of the radiolabeled probe. While such solution-phase or bead-based sandwich hybridization assays are suitable for diagnostic laboratories, faster, safer, cheaper, and easier-to-use hybridization sensors, based on the integration of DNA recognition layers and physical transducers, are highly desired for decentralized screening (e.g., self-testing) of HIV-1 DNA. Here we describe a novel chronopotentiometric sequencespecific biosensor for the detection of short DNA sequences related to the HIV-1. One recent focus in our laboratory has been the development of new electrochemical probes for the direct monitoring of DNA hybridization/recognition events. Lately, electrochemical transducers have been used successfully for sequence DNA detection in connection with electroactive hybridization indicators (ref 8 and references therein). Such new sensing devices, which couple the inherent specificity of DNA recognition reactions with the remarkable sensitivity and portability of electrochemical transduction, hold great promise for HIV DNA screening. The new hybridization sensor is based on the adsorptive attachment of the 21- or 42-base single-stranded (ss) oligonucleotide probes from the HIV-1 U5 long terminal repeat (LTR) region9 onto the carbon paste transducer, their hybridization with the complementary sequence targets, binding of the tris(1,10-phenanthroline)cobalt [Co(phen)33+] marker to the hybrid, and chronopotentiometric monitoring of the hybridization process (via the increased marker peak). The sensitivity and automation advantages of microprocessor-controlled chronopotentiometry for the detection of hybridization processes have been illustrated recently in our laboratory.10 In the following sections, we report on the optimization and acceleration of the HIV-1 DNA assay format and on the resulting analytical performance. The new biosensor allows direct quantification of nanomolar concentrations of the target HIV-1 U5 LTR sequence following short (10-30 min) hybridization times. Yet, it would require common isolation from the blood matrix, restriction endonuclease digestion, and PCR amplification (in accordance with ref 7), and additional developmental work prior to its clinical implementation. In addition to the sequencespecific hybridization biosensor, we report on the direct trace (8) Mikkelsen, S. R. Electroanalysis 1996, 8, 12. (9) Muesing, M.; Smith, D.; Cabradillo, C.; Benton, C.; Lasky, L.; Capon, D. Nature 1985, 313, 450. (10) Wang, J.; Cai, X.; Rivas, G.; Shiraishi, H., Anal. Chim. Acta, in press. (11) Wang, J.; Cai, X.; Wang, J.; Jonsson, C.; Palecek, E. Anal. Chem. 1995, 67, 4065.

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2629

measurements of the HIV-1 U5 LTR DNA using our newly developed adsorptive stripping potentiometric scheme.11 EXPERIMENTAL SECTION Apparatus. A TraceLab potentiometric stripping unit (PSU 20, Radiometer, Denmark) and an IBM PS/2 55SX computer were used to obtain the chronopotentiograms. 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. The analytical signal was evaluated using the peak areas following baseline fitting. The three-electrode system, consisting of the carbon paste electrode (CPE), reference electrode (Ag/AgCl, Model RE-1, BAS), and platinum wire auxiliary electrode, joined the cell through holes in the Teflon cover. The body of the working electrode was a Teflon sleeve (3.5 mm i.d.) tightly packed with the carbon paste. The electrical contact was made with a stainless steel screw. The carbon paste was prepared in the usual way by handmixing graphite powder (Acheson 38, Fisher Scientific) and mineral oil (Sigma, Catalog No. M5904, free of DNase, RNase, and protease). The ratio of graphite powder to mineral oil was 70:30. The surface was polished to a smoothed finish before use. The 1.6 mm diameter gold disk electrode was received from BAS Inc (Model MF-2014). A semiautomatic screen printer (Model TF-100, MPM Inc., Franklin, MA) was used for fabricating the thick-film electrode strips. The carbon ink was received from Gwent Electric Materials Ltd., U.K. (Product No. C10903D14). The printed strips were dried at 180 °C for 1 h. All measurements were carried out in a BAS VC-2 cell, containing 1.0 or 2.0 mL volumes of solutions. All glassware, containers, pipet tips, and the cell (except for the electrodes) were sterilized by autoclaving for 30 min. The electrodes were rinsed with sterilized water prior to use. Chemicals. The 21- and 42-mer synthetic oligonucleotides, corresponding to portions of the HIV-1 U5 LTR DNA segment, were purchased (as their ammonium salts) from Life Technologies (Grand Island, NY); their base sequences are as below:

target (21-mer sequence A): 5′-ACT-GCT-AGA-GAT-TTT-CCA-CAT-3′ immobilized probe (21-mer sequence B): 5′-ATG-TGG-AAA-ATC-TCT-AGC-AGT-3′ three-base mismatch (21-mer sequence A′): 5′-ACT-GAT-AGA-CAT-TTT-CTA-CAT-3′ target (42-mer sequence A): 5′-ACT-GCT-AGA-GAT-TTT-CCA-CAC-TGA-CTA-AAAGGG-TCT-GAG-GGA-3′ immobilized probe (42-mer sequence B): 5′-TCC-CTC-AGA-CCC-TTT-TAG-TCA-GTG-TGG-AAAATC-TCT-AGC-AGT-3′ The 21-mer A and 42-mer A are complementary to 21-mer B and 42-mer B, respectively; 21-mer sequence A′ is a mutant of the 212630

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

mer A with three bases changed, as indicated by the underlines. Other oligonucleotides were also received from Life Technologies. Single-stranded calf thymus DNA (ssDNA, lyophilized powder, Catalog No. D8899), double-stranded calf thymus DNA (dsDNA, activated and lyophilized, Catalog No. 4522), transfer RNA (tRNA, from bakers’ yeast, lyophilized powder, Catalog No. R8759), and diethyl pyrocarbonate were purchased from Sigma. Total RNA from human lung tissue was prepared using RNAsol (precipitated and suspended in the water treated with diethyl pyrocarbonate). Plasmid supercoiled DNA (scDNA) was isolated in accordance with ref 12. Guanidine hydrochloride and urea were purchased from Aldrich and Dextran T500 from Pharmacia Biotech. Three poly(ethylene glycol)s (PEGs, MW ) 3000, 10 000, and 35 000) were products of Sigma and used as received. Tris(1,10-phenanthroline)cobalt(III) perchlorate was synthesized in this department using the method described by Dollimore and Gillard.13 Solutions were prepared with sterile double-distilled water. Tris-HCl buffer solutions were prepared in accordance with ref 12. Procedure. The HIV-1 DNA sequence detection basically consisted of four steps: probe immobilization, hybridization, indicator binding, and chronopotentiometric transduction. During the electrode transfer to the next solution, the surface was rinsed with a specific buffer solution (see below). All steps were performed at room temperature (23.0 ( 0.5 °C). Probe Immobilization. A freshly smoothed carbon paste electrode was first pretreated by applying a potential of +1.7 V for 1 min in a stirred acetate buffer solution (0.2 M, pH 5.0) containing the DNA probe (typically at 5 × 10-7 M). The probe was subsequently immobilized onto the activated electrode surface by adsorptive accumulation for 2 min at a potential of +0.50 V (vs Ag/AgCl). The electrode was then rinsed with 0.75 M NaCl, 0.02 M phosphate buffer (pH 7.0) for a short time (10 s). Hybridization. The hybridization was performed at room temperature by dipping the electrode into the stirred target solution (0.75 M NaCl, 0.02 M phosphate buffer, pH 7.0) for a desired time (1-30 min, depending on the target concentration) while holding the potential at +0.5 V. The electrode was then rinsed with 0.02 M Tris-HCl buffer (pH 7.0) for 10 s. Indicator Binding to the Hybrid. The Co(phen)33+ was accumulated onto the surface hybrid by immersing the electrode into the stirred Tris-HCl buffer solution containing 5.0 × 10-5 M Co(phen)33+ for 1 min while holding the potential at +0.5 V. The electrode was then rinsed with the Tris-HCl buffer for 10 s. Chronopotentiometric Transduction. The accumulated Co(phen)33+ was measured by using an initial potential of +0.5 V and a constant current of -8 µA in the Tris-HCl buffer solution. Repetitive measurements were carried out by renewing the surface and repeating the above assay format. Except as stated otherwise, the reported PSA signals represent the differences in indicator peak areas at the hybrid- and probe-coated electrodes. Chronopotentiometric Stripping Analysis (PSA). Direct adsorptive stripping measurements of trace levels of various HIV-1 U5 LTR DNAs were performed using the PSA protocol developed recently for trace measurements of RNA.11 (12) Maniatis, T.; Fritzsch, E.; Sambrook, J. Molecular Cloning: A Laboratory Manual; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1982. (13) Dollimore, L. S.; Gillard; R. D. J. Chem. Soc., Dalton Trans. 1973, 933.

Figure 1. Chronopotentiograms for the blank (dashed peaks) and the target (0.5 µM 21-mer HIV DNA (sequence A)) using 21-mer HIV DNA (sequence B) as the probe following increasing hybridization times: 1 (a), 2 (b), 5 (c), 10 (d) 20 (e), and 30 (f) min. Also shown is the resulting plot of signal versus time. CPE pretreatment and probe immobilization, 1 min at +1.7 V and 2 min at +0.5 V in 0.2 M acetate buffer (pH 5.0) containing 5 × 10-7 M 21-mer HIV DNA (sequence B); hybridization, 1-30 min at +0.5 V in 0.02 M phosphate buffer (pH 7.0), 0.75 M NaCl, 0.5 µM 21-mer HIV DNA (sequence A); indicator binding, 1 min at +0.5 V in 0.02 M Tris-HCl, 50 µM Co(phen)33+; measurements in 0.02 M Tris-HCl (pH 7.4) with a current of -8 µA.

RESULTS AND DISCUSSION Sensor Optimization. The new biosensor relies on the electrochemical (chronopotentiometric) transduction of the hybridization (recognition) reaction between the short complementary HIV-1 strand sections. Figure 1 shows chronopotentiograms for the Co(phen)33+ indicator at the 21-mer HIV-1 DNA (sequence B) probe-coated electrode after different hybridization periods with a 5 × 10-7 M concentration of the 21-mer HIV-1 DNA (sequence A) target. A well-defined chronopotentiometric peak is observed for the indicator over a relatively flat baseline (peak potential, 140 mV; peak width, 300 mV). The peak increases rapidly with the hybridization time at first (up to 10 min) and then more slowly. Such an enhancement is attributed to the accumulation of the indicator at the electrode surface, through association with the newly formed hybrid. The magnitude of the indicator peak thus reflects the extent of the hybrid formation (see also calibration data below). In contrast, no peak enhancement is observed upon repeating this time-dependent experiment in the absence of the HIV-1 DNA target, i.e., at the surface coated with the singlestranded probe (dashed peaks). The ratio of the responses in the presence and absence of the target thus increases rapidly with the hybridization time. For example, with hybridization for 10 and 30 min, there are 2- and 3.8-fold enhancements of the peak, respectively. In view of the “blank” Co(phen)33+ response, at the same potential, all quantitative work has relied on measurements of the difference in peak areas (with and without the target). The resulting plot of peak area difference versus hybridization time is also shown in Figure 1 (inset). Overall, these data indicate that submicromolar concentrations of the short HIV-1 DNA sequence target can be readily quantified following short hybridization periods. Factors affecting the probe immobilization, hybridization, and indicator-binding reactions were optimized for maximizing the

sensitivity and speeding the assay. The influence of the probe immobilization time was examined over the 0.5-5 min range. The hybridization response rises gradually with immobilization period up to 2 min, and then it leveled off (not shown). Such profile indicates surface saturation with the probe, as desired for minimizing nonspecific adsorption. The probe concentration in the immobilization solution was varied over the 2.5 × 10-7-1 × 10-5 M range; the highest response was observed using 5 × 10-75 × 10-6 M probe solutions (not shown). Such adsorptive attachment of the probe results in a stable immobilization. For example, transfer of the probe-coated electrode to a stirred blank 0.02 M phosphate buffer, 0.75 M NaCl solution, while holding it at +0.5 V (i.e., under the hybridization conditions) resulted in only 2% and 5% losses of the probe signal after 10 and 20 min, respectively (as was indicated from the change of the guanine residue peak11). Figure 2A assesses the effect of the salt (NaCl) concentration in the hybridization solution on the response. As expected for hybridization reactions, the peak area difference increases rapidly upon raising the sodium chloride concentration between 0 and 0.5 M and then more slowly. The applied potential during the hybridization step has a negligible influence on the response, with the largest signal observed at +0.5 V (Figure 2B). The concentration of the Co(phen)33+ indicator has a pronounced effect on the hybridization response (Figure 2C). The response increases sharply with the indicator concentration up to 5 × 10-5 M, above which it starts to level off. The dependence of the response upon the indicator binding time is displayed in Figure 2D. As the binding period increases, the hybridization response rises rapidly at first (up to 120 s) and then decreases slightly. Such a decrease in the hybridization response is attributed to the increased background peak signal at long indicator binding times (and not to an actual decline of the absolute response). Based on the data of Figures 1 and 2, and for obtaining the best compromise between sensitivity and speed, most quantitative work employed a 0.75 M sodium chloride hybridization solution, a 5 min hybridization time, an indicator concentration of 5 × 10-5 M and a 1 min indicator binding time. A probe concentration of 5 × 10-7 M and an immobilization time of 2 min were used in all subsequent biosensing work. Sensor Performance. The application of the HIV-1 DNA sensor for quantification was investigated by varying the concentration of the 21-mer target over the 1 × 10-7-1.2 × 10-6 M range (5 min hybridization; other conditions as in Figure 1). The welldefined indicator peak increased linearly with the target concentration at first up to 6 × 10-7 M, with a curvature at higher levels (not shown). The initial linear portion has a slope of 270.1 ms/ µM (correlation coefficient, 0.996). Even shorter hybridization times may be useful to extend the linear dynamic range. A similar calibration experiment using the 42-mer oligonucleotide probe (sequence B)-coated electrode and the target 42-mer HIV-1 nucleic acid (sequence A) yielded a slightly lower sensitivity (241.2 ms/ µM) and linearity up to 3 × 10-7 M target (correlation coefficient, 0.990, not shown). Such reduced sensitivity for the longer probe and target may be attributed to steric hindrance influencing the hybridization efficiency,14 as well as to the slower rate of mass transport of the longer target toward the probe-coated surface. Figure 3 demonstrates the potential of the sensor for quantifying nanomolar concentrations of the HIV-1 target sequence. Measurements of 2 × 10-8 M concentrations of the 42- and 21(14) Watts, H.; Yeung, D.; Parkes, H. Anal. Chem. 1995, 67, 4283.

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

2631

Figure 2. Effects of NaCl concentration in the hybridization solution (A), applied potential for hybridization (B), indicator (Co(phen)33+) concentration (C), and indicator binding time (D) on the chronopotentiometric responses of 0.5 µM 21-mer HIV DNA (sequence A) using 21-mer HIV DNA (sequence B) as the probe. Hybridization time: 5 min; other conditions as in Figure 1.

Figure 3. Chronopotentiograms for the targets: 2 × 10-8 M 42mer HIV DNA (sequence A) (A), 2 × 10-8 M (B), and 1 × 10-8 M (C) 21-mer HIV DNA (A sequence) in the absence (A and B) and presence (C) of 1% PEG (MW ) 10 000), using corresponding complementary strand (21- or 42-mer sequence B) as the probes. Dashed peaks denote the chronopotentiometric responses for the corresponding blank solutions; hybridization time, 30 min; other conditions as in Figure 1.

mer targets (traces A and B) yield defined increases in the indicator peak, following a 30 min hybridization, compared to the response for the blank solution (dashed line). The increased hybridization rate for shorter sequences15 results in a lower detection limit for the 21-mer segment. Several condensing agents, known for their ability to accelerate hybridization reactions,15 were examined for further lowering the detection limit and shortening the assay time. While carboxymethyldextran and guanidine did not influence the hybridization rate, poly(ethylene glycol) offered the expected acceleration and improvements. Best results were observed with a 1% w/w PEG (MW ) 10 000) solution. The improved detectability accrued from the use of PEG (15) Singh, P., Sharma, B., Tyle, P., Eds. Diagnostics in the Year 2000; Van Nostrand Reinhold: New York, 1993; Chapter 21.

2632 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

is illustrated in Figure 3C for sensing of the 1 × 10-8 M concentration of the 21-mer HIV-1 DNA target, following a 30 min hybridization period. A detection limit of 4 × 10-9 M (4 pmol, 27 ng) can be expected from the signal-to-noise characteristics (S/N ) 3) of these data. A total quantity of 1.35 ng (200 fmol) would thus be detected upon reducing the sample volume to 50 µL. Even lower detection limits are expected for longer hybridization times. Yet, practical HIV testing, particularly those involving detection of the infection before seroconversion, would require further lowering of the detection limit, and hence coupling of the sensor with a DNA amplification method, such as PCR. The ability to perform PCR on microfabricated microstructures, as desired for point-of-care settings, was illustrated recently.16 Table 1 assesses the selectivity of the sensor. The sensor was challenged with various noncomplementary oligomers, ssDNA, dsDNA, scDNA , tRNA, and total RNA. The largest response (corresponding to 19.2% of that of the target) was observed in the presence of an oligonucleotide containing a three-base mismatch (column A). Similarly, the presence of these interferents in a mixture with target did not disturb significantly the target signal (column B). An even better specificity may be attained by adjusting the hybridization temperature,15 using new probes based on the DNA mimic peptide nucleic acid,17 or working in connection with the PCR amplification. A series of eight repetitive measurements of 5 × 10-7 M concentration of the 21-mer HIV-1 DNA target resulted in reproducible results and no decrease in the response (mean hybridization response of 124 ms with a relative standard deviation (16) Wilding, P.; Shoffiner, M.; Kricka, L. Clin. Chem. 1994, 40, 1815. (17) Wang, J.; Palecek, E.; Nielsen, P.; Rivas, G.; Cai, X.; Shiraishi, H.; Dontha, N.; Luo, D.; Farias, P., unpublished results.

Table 1. Effect of Various Nucleic Acids on the Hybridization Responsea signal change (%) nucleic acid 21-mer DNA (sequence B, target) ssDNA dsDNA tRNA total RNA plasmid scDNA 36-mer DNAb 21-mer DNAc

concn 0.5 µM 1 µg/mL 1 µg/mL 1 µg/mL 1 µg/mL 1 µg/mL 3 µg/mL 0.5 µM

A (nucleic acid alone)d

B (mixture)e

+100 +10.5 +13.5 +18.0 -11.6 +4.2 +7.9 +19.2

-8.5 +6.4 +10.1 +5.0 -14.7 -1.9 +16.1

a The electrode with the immobilized 21-mer HIV DNA (sequence B) was immersed (A) into the solution of the complementary 21-mer HIV DNA (sequence A, 0.5 µM) or noncomplementary nucleic acid (interferent) alone or (B) into the mixture of the complementary 21mer HIV DNA (sequence A, 0.5 µM) with the interferent given in the table. After 5 min incubation, the electrode was rinsed and dipped in 0.05 mM Co(phen)33+ indicator for 1 min, followed by the chronopotentiometric measurement. The enhancement of the chronopotentiometric signal due to hybridization of the probe (21-mer HIV DNA, sequence B) with the complementary target (21-mer HIV DNA, sequence A) was taken as 100%. All measurements were performed at room temperature; other conditions as in Figure 1. b The sequence of 36-mer DNA is 5’-CCA-CAT-GGC-CTG-TAC-TTT-AAA-AGC-TTCCGG-ATG-ACC-3′. c Sequence A′, three-base mismatched, as given in the Experimental Section. d Signal change of PSA response toward the interferent alone (compared to that of target 21-mer DNA (sequence A). e Signal change of PSA response toward a 0.5 µM concentration of the target 21-mer HIV DNA (sequence A) in the presence of the interferents.

of 5%; 5 min hybridization). Such measurements were carried out using fresh carbon paste surfaces in each measurement cycle. Attempts to regenerate and reuse the single-stranded probe on the surface, via dipping the electrode in hot (65 °C) deionized water for 10 min, or incubating it for 10 min in 8 M urea or 6 M guanidine solutions, were not successful (mainly due to an increased background response). The present carbon paste protocol permits easy surface renewal and fast (2 min) probe immobilization by adsorptive accumulation of the single-stranded probe, as desired for obtaining reproducible results. We also employed a gold disk electrode, in connection with the surface-activation/DNA-immobilization protocol of Oyama.18 Such a gold-based biosensor allowed up to three hybridization/ binding/detection cycles, but its overall signal-to-noise characteristics were inferior to those of the carbon paste sensor (not shown). In addition, such probe immobilization is time consuming (∼2 days). Satisfactory results, approaching those of the carbon paste sensor, were observed at microfabricated thick-film sensors. These are indicated from the calibration and hybridization time dependence data obtained with the 21-mer HIV DNA at such screen-printed carbon strips electrodes (Figure 4, A and B, respectively). Such mass-producible single-use devices should further facilitate the task of HIV DNA screening. Chronopotentiometric Stripping Analysis of HIV-1 DNAs. In addition to hybridization biosensing of HIV-1 DNA, we report on the direct trace detection of HIV-1 DNAs using the adsorptive chronopotentiometric stripping protocol developed recently for trace analysis of RNA.11 Figure 5 shows chronopotentiograms for 64 µg/L (∼1 × 10-8 M) of the single-stranded 21-mer HIV-1 DNA sequences A and B after different preconcentration times. (18) Yamaguchi, S.; Shimomura, T.; Tatsuma, T.; Oyama, N. Anal. Chem. 1993, 65, 1927.

Figure 4. Hybridization response of thick-film sensors. (A) Chronopotentiograms for 21-mer HIV DNA (sequence A) with increasing concentrations: 0.2 (a), 0.4 (b), 0.6 (c), and 0.8 µM (d) using 21-mer HIV DNA (B-sequence) as the probe and Co(phen)33+ as electroactive indicator, along with resulting calibration plot (0-1.2 µM). Dotted lines denote the response of the blank. (B) Effect of hybridization time on the PSA response of 0.5 µM 21-mer HIV DNA (sequence A). Screenprinted electrode pretreatment and probe immobilization: 1 min at +1.8 V and 2 min at +0.5 V in 0.2 M acetate buffer (pH 5.0) containing 0.5 µM 21-mer HIV DNA (sequence B); hybridization for 5 min (A) or 0-7 min (B) at +0.5 V in 0.02 M phosphate buffer (pH 7.0), 0.75 M NaCl solution containing 0-1.2 µM (A) or 0.5 µM (B) 21-mer HIV DNA (sequence A); indicator binding and PSA transduction as in Figure 1.

The PSA response, associated with the oxidation of the guanine residue in the accumulated HIV-1 DNA strands, increases rapidly with increasing preconcentration times, indicating enhancement of the DNA concentration on the carbon paste surface. For example, 5 and 15 min accumulation periods yield 7- and 14-fold enhancements of the peak area of the 21-mer sequence B, respectively, relative to that attained with 1 min accumulation. The higher sensitivity toward sequence B reflects its larger number of guanine residues in its 21-mer strand (5 versus 3 in sequence A; see Experimental Section). Such coupling of effective preconcentration and computerized PSA offers convenient quantitation of microgram per liter levels of the DNA sequences related to HIV-1. Linear calibration plots were obtained for five successive 32 µg/L increments in the concentration of sequences A and B (slopes of 0.146 and 0.368 ms‚L/µg, respectively; correlation coefficients of 0.991 and 0.995; 2 min accumulation; not shown). The detection limits, estimated from S/N ) 3, correspond to 12 and 32 µg/L (i.e., 12 and 32 ng) sequences B and A, respectively. For 12 successive measureAnalytical Chemistry, Vol. 68, No. 15, August 1, 1996

2633

1. The new biosensor strategy may obviate the need for traditional solution assays of HIV-1 DNA, eliminates the use of radioisotopes, and requires significantly shorter hybridization times. While this method holds great promise for screening of peripheral blood HIV DNA, the readers are cautioned that the concept of HIV DNA electrochemical biosensing is still in the research stage. Extensive developmental worksrelated to both the sample preparation and amplification stages,7 as well as to further improvement of the sensor performancesis desired for addressing the clinical needs for higher sensitivity and selectivity and related matrix effects. Such point-of-care HIV DNA testing would benefit from the introduction of compact, user-friendly, hand-held PSA instruments and microfabricated thick-film carbon transducers19 and micromachined PCR units.16 The coupling of the new biosensor with high-speed flow analyzers may benefit also the large-scale HIV diagnostics in central laboratories. While the concept has been demonstrated in connection with one oligonucleotide related to the HIV-1 DNA LTR sequence, further improvements may be achieved by using several probes, from different regions of the HIV-1 DNA, in connection with multielectrode array and multiple hybridization events. Even though the work presented here is within the context of electrochemical transducers, other transduction mechanisms (e.g., optical or piezoelectric ones) may also be useful for the biosensing of DNA from HIV-1 and related retroviruses. Similar biosensing strategies are expected for the detection of other infectious agents.

Figure 5. (A) Chronopotentiograms for 64 µg/L (1.06 × 10-8 M) 21-mer HIV DNA (A, B sequences) at the pretreated CPEs following increasing accumulation time: 1 (a), 3 (b), 5 (c), 10 (d), and 15 (e) min. (B) The resulting PSA signal versus accumulation time plot. Electrolyte, 0.2 M acetate buffer (pH 5.0); electrode pretreatment, 1 min at +1.7 V; accumulation potential, +0.5 V; stripping current, +8 µA.

ACKNOWLEDGMENT G.R., H.S., and P.A.M.F. acknowledge fellowships from CONICET (Argentina), Ritsumeikan (Japan), and CNPq (Brazil). J.W. acknowledges the financial support from DuPont Diagnostics R&D. Useful discussions with C. Jonsson and M. Balakrishnan are highly appreciated.

ments of 64 µg/L of the 21-mer HIV-1 DNA sequences A and B, the relative standard deviations were 11% and 4%, respectively.

Received for review March 12, 1996. Accepted May 17, 1996.X AC9602433

CONCLUSIONS This study has demonstrated the utility of an electrochemical biosensor for the detection of short DNA segments related to HIV-

2634 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

(19) Wang, J. Analyst 1994, 119, 763. X Abstract published in Advance ACS Abstracts, July 1, 1996.