Sequence-Specific, Electronic Detection of Oligonucleotides in Blood

Jul 13, 2006 - Sequence-Specific, Electronic Detection of. Oligonucleotides in Blood, Soil, and Foodstuffs with the Reagentless, Reusable E-DNA Sensor...
0 downloads 0 Views 250KB Size
Anal. Chem. 2006, 78, 5671-5677

Sequence-Specific, Electronic Detection of Oligonucleotides in Blood, Soil, and Foodstuffs with the Reagentless, Reusable E-DNA Sensor Arica A. Lubin,† Rebecca Y. Lai,‡ Brian R. Baker,† Alan J. Heeger,‡ and Kevin W. Plaxco†*

Department of Chemistry and Biochemistry and Department of Physics and Institute for Polymers and Organic Solids, University of California, Santa Barbara, California 93106

The ability to detect specific oligonucleotides in complex, contaminant-ridden samples, without the use of exogenous reagents and using a reusable, fully electronic platform could revolutionize the detection of pathogens in the clinic and in the field. Here, we characterize a labelfree, electronic sensor, termed E-DNA, for its ability to simultaneously meet these challenging demands. We find that because signal generation is coupled to a hybridization-linked conformational change, rather than to only adsorption to the sensor surface, E-DNA is selective enough to detect oligonucleotides in complex, multicomponent samples, such as blood serum and soil. Moreover, E-DNA signaling is monotonically related to target complementarity, allowing the sensor to discriminate between mismatched targets: we readily detect the complementary 17-base target against a 50 000-fold excess of genomic DNA, can distinguish a three-base mismatch from perfect target directly in blood serum, and under ideal conditions, observe statistically significant differences between singlebase mismatches. Finally, because the sensing components are linked to the electrode surface, E-DNA is reusable: a 30-s room temperature wash recovers >99% of the sensor signal. This work further supports the utility of E-DNA as a rapid, specific, and convenient method for the detection of DNA and RNA sequences. There are growing demands for point-of-care diagnostics, the rapid detection of biothreat agents, and for field-deployable methods for environmental monitoring. Thus motivated, numerous reagentless, reusable biosensor platforms for the detection of DNA have been developed to date. Many of these sensor platforms are based on the optical detection of DNA hybridization, including approaches such as surface-immobilized molecular beacons1-5 and * To whom correspondence should be addressed. Phone: (805) 893-5558. Fax: (805) 893-4120. E-mail: [email protected]. † Department of Chemistry and Biochemistry. ‡ Department of Physics and Institute for Polymers and Organic Solids. (1) Du, H.; Disney, M. D.; Miller, B. L.; Krauss, T. D. J. Am. Chem. Soc. 2003, 125, 4012-4013. (2) Du, H.; Strohsahl, C. M.; Camera, J.; Miller, B. L.; Krauss, T. D. J. Am. Chem. Soc. 2005, 127, 7932-7940. (3) Palecek, E. Trends Biotechnol. 2004, 22, 55-58. (4) Ramachandran, A.; Flinchbauch, J.; Ayoubi, P.; Olah, G. A.; Malayer, J. R. Biosens. Bioelectron. 2004, 19, 727-736. 10.1021/ac0601819 CCC: $33.50 Published on Web 07/13/2006

© 2006 American Chemical Society

surface plasmon resonance-based sensors,6 or on the detection of changes in surface-adsorbed mass,7 steric bulk,8 or charge.9,10 However, although many of these approaches exhibit superb sensitivity and achieve single-base mismatch discrimination,4,5,9 few have been shown to operate directly in realistically complex, contaminant-ridden samples.11-14 Here, we demonstrate that the reagentless, operationally convenient E-DNA sensor, which is the electrochemical analogue of optical molecular beacons, is specific, readily reusable, and functional, even in samples such as blood serum and soil. The E-DNA sensor consists of a redox-labeled DNA stem-loop probe chemisorbed onto an interrogating electrode via alkanethiol self-assembled monolayer chemistry.15-17 In the absence of target, the stem-loop conformation holds the redox label in close proximity to the electrode, facilitating electron transfer. Upon binding to a complementary DNA or RNA target, hybridization forces the redox tag away from the electrode, impeding electron transfer and producing a large, readily detectable reduction in redox current. The impressive specificity of optical molecular beacons18-21 and the generally low sample background observed (5) Wang, H.; Li, J.; Liu, H.; Liu, Q.; Mei, Q.; Wang, Y.; Zhu, J.; He, N.; Lu, Z. Nucleic Acids Res. 2002, 30, e61. (6) Nakatani, K.; Sando, S.; Saito, I. Nat. Biotechnol. 2001, 19, 51-55. (7) Hang, T. C.; Guiseppi-Elie, A. Biosens. Bioelectron. 2004, 19, 1537-1548. (8) McKendry, R.; Zhang, J.; Arntz, Y.; Strunz, T.; Hegner, M.; Lang, H. P.; Baller, M. K.; Certa, U.; Meyer, E.; Guntherodt, H.-J.; Gerber, C. Proc. Natl. Acad. Sci. 2002, 99, 9783-9788. (9) Sakata, T.; Miyahara, Y. ChemBioChem 2005, 6, 703-710. (10) Pouthas, F.; Gentil, C.; Cote, D.; Bockelmann, U. Appl. Physics Lett. 2004, 84, 1594-1596. (11) Lee, T. M. H.; Hsing, I.-M. Anal. Chem. 2002, 74, 5057-5062. (12) Umek, R. M.; Lin, S. W.; Vielmetter, J.; Terbrueggen, R. H.; Irvine, B.; Yu, C. J.; Kayyem, J. F.; Yowanto, H.; Blackburn, G. F.; Farkas, D. H.; Chen, Y.-P. J. Mol.ecular Diagn. 2001, 3, 74-84. (13) Liu, R. H.; Yang, J.; Lenigk, R.; Bonanno, J.; Grodzinski, P. Anal. Chem 2004, 76, 1824-1831. (14) Marrazza, G.; Chiti, G.; Mascini, M.; Anichini, M. Clin. Chem. 2000, 46, 31-37. (15) Fan, C.; Plaxco, K. W.; Heeger, A. J. Proc. Natl. Acad. Sci 2003, 100, 91349137. (16) Immoos, C. E.; Lee, S. J.; Grinstaff, M. W. ChemBioChem 2004, 5, 11001103. (17) Mao, Y.; Luo, C.; Ouyang, Q. Nucleic Acids Res. 2003, 31, e108. (18) Hwang, G. T.; Seo, Y. J.; Kim, B. H. J. Am. Chem. Soc. 2004, 126, 65286529. (19) Bonnet, G.; Tyagi, S.; Libchaber, A.; Kramer, F. R. Proc. Natl. Acad. Sci 1999, 96, 6171-6176. (20) Tyagi, S.; Kramer, F. R. Nat. Biotechnol. 1996, 14, 303-308. (21) Giesendorf, B. A. J.; Vet, J. A. M.; Tyagi, S.; Mensink, E. J. M. G.; Trijbels, F. J. M.; Blom, H. J. Clin. Chem. 1998, 44, 482-486.

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006 5671

Table 1. Detection of Target DNA by E-DNA in Complex, Contaminant-Ridden Samples sample testeda

sample sample + target

buffer

blood serum

urine

saliva

soil

Guinness

0b 36.5 ( 0.6

3.4 ( 1.6 36.0 ( 14.2

1.1 ( 0.7 34.2 ( 3.8

0.6 ( 0.5 32.4 ( 1.2

1.6 ( 0.8 35.0 ( 8.1

1.2 ( 1.4 36.3 ( 5.6

a Each value and standard deviation is reported as a percentage and is based on an average of measurements from three electrodes. The values reported are sensor signal changes observed in each sample after 30 min. b By definition.

in electrochemistry experiments (even in contaminant-ridden samples) suggest that the E-DNA platform should be both sequence-specific and largely insensitive to proteins and other likely contaminants. Likewise, the reagentless nature of E-DNA and the strong self-assembled monolayer chemistry employed in its fabrication suggest that the sensor should be easily regenerated. The sequence specificity of E-DNA has, however, seen only rather limited investigation,15-17 and to our knowledge, no studies to date have been reported regarding the sensor’s ability to detect RNA or its ability to selectively identify its target among nonnucleic acid contaminants. Similarly, previous E-DNA studies report at best only 80% signal recovery upon washing.15 Here, we explore in detail the selectivity, specificity, and reusability of the E-DNA sensor. EXPERIMENTAL SECTION Materials. The E-DNA probe sequence, 5′-HS-(CH2)6GCGAGGTAAAACGACGGCCAGTCTCGC-(CH2)7-MB, was obtained from BioSource, Int. (Foster City, CA). Methylene blue (MB) was conjugated to the 3′ end of the probe via succinimide ester coupling (MB-NHS obtained from EMP Biotech, Germany).22 The modified oligo was purified via C18 reversed-phase HPLC and polyacrylamide gel electrophoresis (PAGE) and confirmed by mass spectrometry. The target sequences tested are listed in Table 2. All DNA oligos were purchased from Synthegen, LLC (Houston, TX), and Sigma-Genosys (The Woodlands, TX). Perfect match RNA target was obtained from Dharmacon, Inc. (Chicago, IL) and deprotected immediately prior to use with 2′-O-Deprotection Buffer (Dharmacon, Inc.). Target DNA and RNA concentrations of 200 nM (confirmed via UV absorbance measurements at 260 nm) were employed unless otherwise stated. The 6-mercaptohexanol (Sigma-Aldrich, St. Louis, MO), Tris (2carboxyethyl) phosphine hydrochloride (Molecular Probes, Carlsbad, CA), fetal calf serum and calf thymus single-stranded DNA (Sigma-Aldrich), Guinness Draught beer (St. James’s Gate Brewery, Dublin, Ireland), and guanidine hydrochloride (Pierce, Rockford, IL) were used as received. Preparation and Characterization of E-DNA Sensors. The polycrystalline gold disk electrodes (1.6 mm diameter, BAS, West Lafayette, IN) used in this study were prepared by polishing with diamond and alumina (BAS), sonicating in water, and electrochemically cleaning (a series of oxidation and reduction cycles in 0.5 M H2SO4, 0.01 M KCl/0.1 M H2SO4, and 0.05 M H2SO4) before modification with E-DNA. The probe DNA was immobilized onto the gold surface by incubating the clean electrode in 0.1 µM DNA/1 µM TCEP (Tris(2-carboxyethyl) phosphine hydrochloride) (22) Hermanson, G. T. Bioconjugate Techniques; Academic Press: San Diego, 1996.

5672 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

in 100 mM NaCl/5 mM MgCl2/10 mM potassium phosphate pH 7 buffer for 14-16 h. The surface was then rinsed with water and subsequently passivated with 1 mM 6-mercaptohexanol in 1 M NaCl/10 mM potassium phosphate buffer, pH 7, for 4 h. The electrodes were then rinsed with deionized water before electrochemical analysis using alternating current voltammetry (ACV) at 10 Hz frequency, 25 mV amplitude, over a potential range of -0.1 to -0.4V, using a CHI 603 potentiostat (CH Instruments, Austin, TX) in a standard cell with a platinum counter electrode and Ag/AgCl (3 M NaCl) reference electrode. All experiments were conducted in 1 M NaCl/10 mM potassium phosphate buffer, pH 7 (buffered saline), or in various clinical materials diluted with this buffer. Selectivity Measurements. E-DNA selectivity was assessed by incubating each electrode for 30 min at room temperature in 600 µL of fetal calf serum (50% v/v in buffered saline), urine or saliva (both collected, diluted 50% with buffered saline and used within 30 min), Guinness Draught beer (diluted 50% with buffered saline), or 400 mg of soil (diluted to a final volume of 600 µL with buffered saline). Electrodes were interrogated in buffered saline, rinsed with room temperature water, and then reincubated for 30 min in each sample solution spiked with 280 nM of the complementary DNA target. The serum samples were tested in situ. All other samples were tested ex situ to protect our reference electrode from contamination. Reported values represent the mean and standard deviations of measurements conducted using three independent electrodes. Target Sequence-Specificity Measurements. Specificity measurements were performed by incubating the electrodes in 200 nM target DNA or RNA in buffered saline in situ and interrogating with ACV. The electrodes were then rinsed with deionized water and incubated in buffered saline for 5-15 min before being challenged with the next target. For the set done in fetal calf serum (50% v/v in buffered saline), the electrodes were regenerated with a 30-s soaking step in 8 M guanidine hydrochloride (GuHCl) and then incubated in fetal calf serum for 5-15 min before being challenged with the next target. For the measurement done in 50 000-fold excess calf thymus genomic DNA, the electrode was first incubated for 30 min in 1 mg of genomic DNA dissolved in 1 mL of buffered saline and then reincubated for 30 min in the genomic DNA solution spiked with 20 ng (200 nM) of target DNA. A single electrode was used for each set of experiments to avoid signal deviations arising from variations in sensor fabrication, and each set of experiments was repeated using three independent electrodes. DNA Melting Measurements. Tm values were measured for each probe and target combination at concentrations of 600 nM unmodified E-DNA probe (Sigma-Genosys) and target DNA in

Figure 1. E-DNA works even in complex, contaminant-ridden samples. (left) For example, the sensor readily detects 200 nM target DNA in a sample of fetal calf serum (in situ) and is readily regenerated to 98% of its original signal using a simple, room temperature wash with 8 M GuHCl. (right) Similar selectivity is apparent when the sensor is challenged with urine, saliva, mud (soil collected from outside our laboratory suspended at 67% w/v in buffered saline). Whereas only minor signal changes are observed when the sensor is incubated in target-free samples of these materials, normal signal suppressions are observed when the sensor is incubated in these samples doped with 280 nM target DNA. Data shown are average measurements (performed ex situ to achieve the proper electrolyte balance for electrochemistry) from three different electrodes.

buffered saline using a Beckman Coulter DU 800 UV/vis spectrophotometer. Sensor Regeneration. To test E-DNA reusability, electrodes were prepared and modified as described above and then allowed to hybridize, in situ, with perfect complement target in buffered saline for 5-10 min (hybridization is effectively complete by this time). Following interrogation by ACV, the electrodes were rinsed with 18 MΩ Milli-Q filtered water for ∼30 s, incubated in buffered saline for 5 min to allow for the reformation of stem-loops, and reinterrogated, and the process was repeated iteratively over 10 cycles for each of three electrodes. This same method was used for the regeneration of sensors employed in soil, urine, saliva, and our representative foodstuff. The electrodes employed in fetal calf serum were regenerated via a 30-s wash in 8 M GuHCl, which led to near complete signal recovery (see below); regeneration of these electrodes using water recovered only 89% of the signal (data not shown). Preparation of Thin Film Gold Electrodes and Immobilization of DNA Probes. The thin film gold electrodes used in this study were fabricated on a glass plate using standard e-beam lithography/lift-off techniques. First, a negative photoresist (AZ 5214; Clariant, Somerville, NJ) was spin-coated onto a glass wafer (Corning 7740; Dow Corning, Midland, MI) and patterned using photolithography. A 130-nm platinum film was deposited over a thin titanium adhesion layer (20 nm) by an electron-beam evaporator (SEC-600-RAP, CHA Indus., Fremont, CA). Photoresist patterns and unnecessary metal patterns were removed by a liftoff process using an acetone solution. To define the sensing area, a silicon nitride (SiN, 200 nm) film was grown in a plasmaenhanced vapor deposition reactor (VII 790, Plasma Therm Inc., Petersburg, FL). The SiN film was then patterned by a plasma dry etcher (RIE51, MRC, Prangehurg, NY). AZ 5214 was spincoated and patterned prior to the deposition of the titanium adhesion layer (20 nm) and the gold film (250 nm). The patterned photoresist was subsequently removed by a lift-off process. Prior to surface modification, the patterned electrodes were cleaned by immersing in piranha solution (3:1 H2SO4/H2O2) for 5 min and then thoroughly rinsed with deionized water. The surface was

modified by incubating the clean electrodes in 0.5 µM DNA/1 µM TCEP (in 100 mM NaCl/5 mM MgCl2/10 mM potassium phosphate pH 7 buffer) for 45 min, followed by 3 h in 1 mM 6-mercaptohexanol and ∼1 h in buffered saline using the same buffers and procedure as described above for the disk electrodes. RESULTS Selectivity. Because signal generation in the sensor is based on a binding-induced conformational change, rather than on simple adsorption to the sensor surface, E-DNA should be largely impervious to false positives arising from nonoligonucleotide contaminants. Consistent with this claim, we find that the E-DNA response we obtain from target contained within complex, contaminant-ridden samples is almost indistinguishable from the response observed from target suspended in pure buffer (Figure 1, Table 1). For example, target-free samples of blood serum, urine, saliva, soil, and a representative foodstuff (Guinness Draught) produce no significant change in peak current relative to those observed in pure, DNA-free buffer. When these same samples are spiked with target DNA, the corresponding peak decreases are within error of those observed for the target in simple buffered saline (Table 1). Specificity. The E-DNA sensor is capable of discriminating between perfectly complementary DNA and multiple-mismatch targets (Table 2). For example, whereas the current from a typical sensor is suppressed by 32.3% in the presence of perfectly complementary 17-base target, the signal is suppressed by only 2, 9.6, and 24.4% for 11-, 5-, or 3-base mismatches, respectively, at the same concentration (Figure 2, left). Because even the latter represents a large, statistically robust deviation from the signal observed for perfectly complementary target, these results suggest that the E-DNA sensor very reliably distinguishes multiple mismatch targets from a perfect match target; sufficiently robustly, in fact, that the sensor can perform this discrimination even in grossly contaminated samples, such as blood serum (see below). Of note, the selectivity of the E-DNA is great enough that the sensor is insensitive to a 50 000-fold excess (over the amount of perfectly complementary target employed in these studies) of Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

5673

Table 2. DNA and RNA Target Sequences name

sequence (5′-3′)

perfect match 8G-A 1bp mismatch 8G-C 1bp mismatch 8G-T 1bp mismatch 5G-T 1bp mismatch 12T-C 1bp mismatch 3-bp mismatch 5-bp mismatch 11-bp mismatch RNA perfect match

ACTGGCCGTCGTTTTAC ACTGGCCATCGTTTTAC ACTGGCCCTCGTTTTAC ACTGGCCTTCGTTTTAC ACTGTCCGTCGTTTTAC ACTGGCCGTCGCTTTAC ACTGGAATTCGTTTTAC ACTTTAATTCGTTTTAC CGTATCATTGGACTGGC ACUGGCCGUCGUUUUAC

sheared single-stranded genomic DNA, changing the signal by only 0.9 ( 0.2% when incubated for 30 min in the sheared genomic DNA and by 36.0 ( 2.3% when target DNA is added to the genomic DNA solution (Figure 3, left). These results further demonstrate that the sensor is highly specific and that its

specificity is not reduced by the presence of a large amount of contaminating DNA. In contrast, although the current E-DNA sensor can also detect point mutations, it does so with significantly reduced statistical significance (Figure 2, right) and then only in the absence of contaminants and under ideal laboratory conditions. For example, across five unique single-base mismatched target sequences, we observe signal suppression ranging from 32.6 to 36.7% (Figure 2, right). Among these, the best discrimination is observed for a C-C mismatch, which correlates well with the observation that mismatches containing C are relatively destabilizing.23 Although we illustrate here a single set of data, all measurements were repeated using three independent electrodes; however, although each of the three electrodes produced the same ranking in terms of the absolute signal change generated by the mismatched target, the magnitude of the signal change produced by each point mutation is smaller than our typical electrode-to-electrode variability (data not shown), thus limiting the utility of the current

Figure 2. The sequence specificity of the E-DNA sensor is such that it readily distinguishes between a perfectly complementary 17-base target and 11-, 5-, and 3-bp mismatched targets (left). Additionally, although E-DNA’s ability to identify and discriminate between point mutations is more limited (right), it is sufficient to perform this task under ideal laboratory conditions. In a test of several different single-base mismatched targets, the sensor suppression remains greatest for the perfect complement target, but only by a small margin. Here, the signal is suppressed by 38.0% for the perfect complement; 30% for the 3-bp mismatch; and 32.6% (8G-C), 34.7% (8G-T), 36.1% (12T-C), 36.4% (8G-A), and 36.7% (5G-T) for the single-base mismatched targets. Of note, although the origins of the effect are unclear, the slower association kinetics observed for the three-base mismatch target and several of the single-base mismatch targets are reproducible (data not shown). All targets are at 200 nM.

Figure 3. E-DNA demonstrates sequence specificity in complex, contaminant-ridden samples and works in the presence of excess genomic DNA. (left) For example, the sensor can detect its complementary target against a 50 000-fold excess of single-stranded calf thymus genomic DNA. Shown are ACV curves for an E-DNA sensor in the absence of any added DNA, in the presence of a 50 000-fold excess (m/m) of single-stranded calf thymus genomic DNA, and 200 nM target DNA plus genomic DNA (right) The sensor also readily distinguishes 200 nM 11-, 5-, and 3-bp mismatches from a perfectly complementary target in a sample of fetal calf serum (in situ). Under these conditions, the sensor’s response to a single-base mismatch is within error of that observed for its perfect complement target, and therefore, the data is not included for the single-base mismatch target. 5674

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

Figure 4. The E-DNA signal is reasonably well correlated with the melting temperatures (Tm) of the probe-target duplex over the narrow range of targets for which Tm is even measurable. This suggests that, under the conditions employed, sensor signal is limited by the thermodynamics of hybridization.

generation E-DNA sensor for applications in which point-mutation detection is critical. The E-DNA sensor also displays reasonable selectivity, even in complex, clinically relevant samples, such as blood serum: under these conditions, we achieve ready discrimination between perfectly complementary and mismatched targets, with the exception of the single-base mismatched target, which is difficult to distinguish, since the observed signal suppression is within the error of that seen for the perfect complement (Figure 3, right). For example, we observed an average signal suppression of 37.8 ( 2.6% for the serum sample doped with complementary target but only 19.6 ( 4.6% for a 3-base mismatch and 0.4 ( 0.4% for a 5-base mismatch (the 11-base mismatch was tested, as well, and showed no signal suppression). Although this result confirms our claim that E-DNA’s ability to distinguish single-base mismatches is limited, it also demonstrates the sensor’s ability to perform when faced with complex, contaminant-ridden samples. These results suggest the selectivity of the E-DNA sensor is, thus, sufficient to detect and identify target DNA sequences directly in clinically relevant materials and suggests the sensor is suitable for application in a wide range of real-world scenarios. Previous studies have shown that the E-DNA sensor response is logarithmic in target concentration,15 suggesting that hybridization thermodynamics determine the observed signal change. As an additional test of this hypothesis, we have measured the melting temperatures (Tm) of seven of our target-probe duplexes in solution. Over the admittedly rather narrow range of Tm that we have measured, we find that this measure of hybridization free energy is reasonably well correlated (r ) 0.90) with E-DNA signal suppression (Figure 4), providing further support for the argument that signaling is thermodynamically rather than kinetically limited. Finally, the E-DNA sensor appears equally well suited for the detection of RNA. When challenged with a 17-base synthetic RNA target, the E-DNA sensor response is similar to its final, equilibrium response to the equivalent DNA target (Figure 5). The sensor saturates at 30 min with 45.6% final signal suppression. Reusability and Reproducibility. Reusability is often a desirable attribute for real-world sensor applications. Attempts to regenerate the first generation E-DNA sensor, however, achieved only 80% signal recovery.15 In an effort to improve this, we first modified the E-DNA sensor so as to increase its stability. In particular, we have replaced the passivating self-assembled monolayer, originally β-mercaptoethanol, with 6-mercaptohexanol and

Figure 5. The E-DNA sensor response to RNA is similar to that observed for DNA targets. In a time course of E-DNA’s response to 200 nM RNA target, the signal is suppressed by 38.1% after 5 min and appears to saturate at 30 min with 45.6% signal suppression.

replaced the potentially labile ferrocene label24 with the more stable methylene blue (MB). Unfortunately, however, these modifications alone do not significantly improve sensor reusability: upon washing this modified sensor with 90 °C water, we again recover only slightly more than 80% of the original signal intensity (data not shown). In contrast, by employing low ionic strength rather than high temperature to disrupt hybridization, we observe significantly improved E-DNA reusability. We find that a simple 30-s wash with room temperature water is sufficient to produce effectively complete recovery of conventional gold disk electrodes modified with E-DNA (Figure 6, top). After challenging the sensor with complementary target and observing significant signal suppression, we achieve a mean recovery of 99.4% of the original signal per use over 10 uses (Figure 6, top right). Moreover, much of the loss in recovery occurs during the first usage: from this regeneration, we recover only 96.1% of the prior signal. Over the next nine regenerations, in contrast, we achieve an average of 99.8% signal recovery per regeneration cycle. These results imply that the E-DNA sensor can be used for hundreds of cycles before significant signal degradation is observed (0.998100 ) 82%). Of relevance to real-world sensing scenarios, the E-DNA sensor is also reusable even when challenged with contaminantridden samples. For example, we find that a simple 30-s room (23) Frutos, A. G.; Pal, S.; Quesada, M.; Lahiri, J. J. Am. Chem. Soc. 2002, 124, 2396-2397. (24) Han, S. W.; Seo, H.; Chung, Y. K.; Kim, K. Langmuir 2000, 16, 9493.

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

5675

Figure 6. The E-DNA sensor regeneration can be achieved via a simple 30-s rinse in room temperature deionized water. (top left) ACV scans of a conventional gold-disk electrode E-DNA sensor over 10 cycles of hybridization and regeneration demonstrate (top right) a mean signal regeneration of 99.4 ( 1.0% per use. Much of the total loss of signal, however, arises during the first usage; after the first use/wash cycle, the sensor exhibits a mean recovery of 99.8 ( 0.5% per iteration. The sensor signal is also highly reproducible, producing a mean signal reduction of 36.5 ( 0.6% when challenged with a fully complementary target. (bottom) E-DNA sensors built using microfabricated gold electrodes are similarly reusable and reproducible; we achieve 100.9 ( 1.4% signal recovery per wash and 21.4 ( 0.7% signal drop per use over 13 iterations.

temperature wash with 8 M GuHCl is sufficient to regenerate at least 98.0% of our original signal, even for a sensor that has been employed directly in blood serum (Figure 1, left). We achieve similar levels of recovery from soil, urine, saliva, and foodstuffs using nothing more complex than room temperature washing with water (data not shown). Our ability to reuse a single sensor multiple times allows us to quantify the reproducibility of E-DNA. With a typical electrode, we observe mean signal changes of 36.5 ( 0.6% over 10 challenges, corresponding to a coefficient of variation of 1.6% at this target concentration. We note, however, that our ability to reproducibly fabricate E-DNA electrodes remains rather more limited; the signal variation between electrodes is several percent (data not shown). Thus, variations in fabrication currently limit the electrode-toelectrode reproducibility (and, thus, sensitivity and specificity) of the E-DNA sensor. One advantage that electronic biosensors hold over their optical counterparts is the presumed ease with which electronic sensors can be integrated with modern microelectronics. Thus motivated, we have also characterized the reusability and reliability of E-DNA sensors built using microfabricated gold electrodes in place of the conventional gold-disk electrodes employed above. We find that the reusability of these e-beam evaporated gold electrodes is quite similar to that observed for gold-disk electrodes; 5676

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

we observe a mean signal recovery of 100.9 ( 1.4% over 13 rounds of regeneration (Figure 6, bottom). Of note, the signal recovery for these fabricated electrodes is in excess of 100%. We believe this arises because the initial signal intensity observed from the fabricated electrodes is rather sensitive to variations in the force with which the washing step is conducted (data not shown). Upon interrogation with the complementary target, we observe a signal decrease of 21.4 ( 0.7%, corresponding to a coefficient of variation of 3.3% for the detection of nanomolar target DNA. DISCUSSION The E-DNA sensor is sensitive, sequence-specific, and sufficiently selective to work in blood serum, urine, saliva, soil, and foodstuffs. It is also readily reusable (even when employed in complex samples), reagentless, and fully electronic. Whereas other sensors may individually achieve one or even several of these advantages, we are not aware of any other single biosensing platform that simultaneously exhibits all of these potentially important attributes for the detection of oligonucleotides.25 The sequence specificity of the E-DNA sensor is comparable to that of other label-free DNA sensing technologies. For example, although solution-phase molecular beacons have, like E-DNA, (25) Kissinger, P. T. Biosens. Bioelectron. 2005, 20, 2512-2516.

been reported to exhibit single-base discrimination under ideal conditions,18-21 their sequence specificity in more realistic settings has seen only very limited investigation.26 Similarly, widely differing levels of discrimination have been reported for surfaceattached molecular beacons.2,4,5,16,17,27 One recent report, for example, describes excellent single-base specificity when tested at ∼3 µM target concentrations, but significantly poorer discrimination is observed at nanomolar concentrations.2 Similarly, Immoos et al. report an ability to distinguish single-base mismatches from complementary sequences using E-DNA but report this at elevated temperatures, which favors selectivity.16 We anticipate that under similarly stringent conditions, the specificity of our implementation of E-DNA may also be improved. Our observed results for the mismatched targets may also be explained by comparing the Tm values obtained for the probetarget duplexes in solution versus the sensor performance once the probe is immobilized on a surface. The sequence specificity observed in solution as measured by Tm for the various targetprobe duplexes (with apparent differences in Tm values for the complementary target versus the mismatched targets ranging from 10.0 ( 0.4 to 15.3 ( 0.3 °C, Figure 4), though correlated to, does not appear to translate to a comparable selectivity in the E-DNA signal once the probe is immobilized on a surface. This suggests that the specificity of the probe, under the current conditions, is compromised when immobilized on the surface. Previously, it was shown that surface-immobilized DNA duplexes have depressed Tm values and are less stable than duplexes in solution.28,29 Given this, performing the E-DNA experiments at room temperature may diminish sequence specificity more than performing the same experiment in solution. Nevertheless, even if further improvements in specificity are not forthcoming, the ability to distinguish at the multiple base mismatch level is sufficient for a wide range of applications including, for example, pathogen detection.30 In contrast to its sensitivity, the selectivity and reusability of the E-DNA sensor appear to surpass significantly those of (26) Peng, X.-H.; Cao, Z.-H.; Xia, J.-T.; Carlson, G. W.; Lewis, M. M.; Wood, W. C.; Yang, L. Cancer Res. 2005, 65, 1909-1917. (27) Wei, F.; Sun, B.; Liao, W.; Ouyang, J.; Zhao, X. S. Biosens. Bioelectron. 2003, 18, 1149-1155. (28) Peterson, A. W.; Heaton, R. J.; Georgiadis, R. J. Am. Chem. Soc. 2000, 122, 7837-7838. (29) Peterlinz, A. K.; Georgiadis, R.; Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 3401-3402. (30) Kakinuma, K.; Fukkushima, M.; Kawaguchi, R. Biotechnol. Bioeng. 2003, 83, 721-728. (31) Piunno, P. A. E.; Krull, U. J. Anal. Bioanal. Chem. 2005, 381, 1004-1011. (32) Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 8916-8920.

previously reported electronic sensing approaches.31 For example, although we have shown here that E-DNA is virtually unaffected by the abundant and complex contaminants present in bodily fluids, soil, and foodstuffs, few, if any, other sensing technologies have been reported to operate in such challenging samples without significant target amplification or purification.11-14 Last, even when challenged with complex, dirty samples, the E-DNA sensor can be regenerated in many cases using a simple room temperature rinse with distilled water or guanidine hydrochloride solutions. This contrasts sharply with the reported regeneration of other, similar immobilized sensors,1,4,11,32 which require strong bases4 or heated water and buffers2,32 for regeneration, or often report far more limited regeneration than that reported here.1,2,11 CONCLUSIONS An ideal biosensor discriminates its target from similar, related compounds and from realistic contaminants; is reusable; and does not require the addition of exogenous reagents or cumbersome processing steps.25 The E-DNA sensor appears promising with regard to each of these demanding requirements. E-DNA readily distinguishes multiple-mismatch targets from the perfect complement and, under ideal laboratory conditions, can even discriminate perfect target from single-base mismatches. Moreover, because E-DNA signal generation is coupled to a hybridization-linked conformational change rather than to adsorption to the sensor surface, and because the E-DNA sensor is electrochemical rather than fluorescence- or absorbance-based, the approach is not thwarted by realistically complex, contaminant-ridden samples. Last, E-DNA is readily reusable and can be reset via simple, room temperature, aqueous washes. The exceptional selectivity, specificity, and reusability of the E-DNA platform suggest it may be a promising strategy for a wide variety of sensing applications. ACKNOWLEDGMENT This work was supported by the National Institutes of Health through Grant NIH EB002046 and by the Institute for Collaborative Biotechnologies through Grant DAAD19-03-D-0004 from the U.S. Army Research Office. We thank Dr. Sang-ho Lee for fabricating the thin film gold electrodes used in this study. We also thank Morgan Schafer for her assistance in the RNA experiments.

Received for review January 26, 2006. Accepted June 1, 2006. AC0601819

Analytical Chemistry, Vol. 78, No. 16, August 15, 2006

5677