Effects of Probe Length, Probe Geometry, and Redox-Tag Placement

Effects of Probe Length, Probe Geometry, and Redox-Tag Placement on the Performance of the Electrochemical DNA Sensor ... E-mail: [email protected]. ...
2 downloads 4 Views 1MB Size
Download PDF
Anal. Chem. 2009, 81, 2150–2158

Effects of Probe Length, Probe Geometry, and Redox-Tag Placement on the Performance of the Electrochemical E-DNA Sensor Arica A. Lubin,† Brook Vander Stoep Hunt,† Ryan J. White,† and Kevin W. Plaxco*,†,‡ Department of Chemistry and Biochemistry and Biomolecular Science and Engineering Program, University of California, Santa Barbara, California 93106 Previous work has described several reagentless, electrochemical DNA (E-DNA) sensing architectures comprised of an electrode-immobilized, redox-tagged probe oligonucleotide. Recent studies suggest that E-DNA signaling is predicated on hybridization-linked changes in probe flexibility, which will alter the efficiency with which the terminal redox tag strikes the electrode. This, in turn, suggests that probe length, probe geometry, and redoxtag placement will affect E-DNA signaling. To test this we have characterized E-DNA sensors comprised of linear or stem-loop probes of various lengths and with redox tags placed either distal to the electrode or internally within the probe sequence (proximal). We find that linear probes produce larger signal changes upon target binding than equivalent stem-loop probes. Likewise, long probes exhibit greater signal changes than short probes provided that the redox tag is placed proximal to the electrode surface. In contrast to their improved signaling, the specificity of long probes is poorer than that of short probes, suggesting that sensor optimization represents a trade off between sensitivity and specificity. Finally, we find that sensor response time and selectivity are only minimally affected by probe geometry or length. The results of this comparative study will help guide future designs and applications of these sensors. Recent years have seen reports of a new class of reagentless, electrochemical sensors for the sequence-specific detection of DNA and RNA that are comprised of a redox-labeled DNA probe chemisorbed onto an interrogating electrode.1-5 The first of these devices,1 which we have termed E-DNA sensors, utilized a DNA probe that adopts a stem-loop configuration that holds the attached redox tag in proximity to the electrode. Hybridization forces the * To whom correspondence should be addressed. E-mail: [email protected]. Phone: (805) 893-5558. Fax: (805) 893-4120. † Department of Chemistry and Biochemistry. ‡ Biomolecular Science and Engineering Program. (1) Fan, C.; Plaxco, K. W.; Heeger, A. J. Proc. Natl. Acad. Sci.U.S.A. 2003, 100, 9134–9137. (2) Immoos, C. E.; Lee, S. J.; Grinstaff, M. W. ChemBioChem 2004, 5, 1100– 1103. (3) Mao, Y.; Luo, C.; Ouyang, Q. Nucleic Acids Res. 2003, 31, e108. (4) Jenkins, D. M.; Chami, B.; Kreuzer, M.; Presting, G.; Alvarez, A. M.; Liaw, B. Y. Anal. Chem. 2006, 78, 2314–2318. (5) Zhang, J.; Qi, H.; Li, Y.; Yang, J.; Gao, Q.; Zhang, C. Anal. Chem. 2008, 80, 2888–2894.

2150

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

redox tag away from the electrode upon binding to a complementary oligonucleotide target, impeding electron transfer and producing a large, readily detectable decrease in redox current.1,2 Since this first E-DNA sensor was reported in 2003, the design of this platform has undergone multiple iterations. For example, the original “signal-off” sensor (target binding leads to reduced signaling current from the stem-loop probe) has been expanded into other signal-off sensors based on linear probes6 and several, more complicated “signal-on” architectures based on hairpin formation, strand-displacement, or pseudoknot probes.7-9 All of these sensors, however, share a common theme: each signals the presence of target via hybridization-linked changes in electron transfer to and from the redox tag.1-9 Many fabrication and operational parameters affect the performance of the E-DNA platform. We have found, for example, that the signaling and equilibration time of both linear and stemloop E-DNA sensors depend strongly on the density of the probe oligonucleotides on the electrode surface.6,10 Likewise, sensor signaling (in terms of absolute faradaic current) and stability are strongly affected by the length of the hydrocarbon chain used to link the probe to the electrode.11 Finally, the signaling of the E-DNA sensor is dependent on both the frequency of the ac voltage used to interrogate the sensor and the geometry of the probe oligonucleotide, with a variety of probe geometries, including hairpin, linear, and pseudoknot probes, having been explored to date.1,6,8,9 In an effort to further optimize the E-DNA approach, we expand here on two E-DNA architectures and explore the potentially key variables of probe length, probe geometry, and redox-tag placement. To do so, we have designed six distinct E-DNA sensors (Figure 1), three based on stem-loop probes and three based on linear probes. Each set of probe geometries includes one short (17 base) and two longer (34 base) recognition elements. These recognition elements are modified with a redox reporter either at the terminus distal from the electrode or, for two of the 34 base constructs, at an internal site 17 bases or 27 (6) Ricci, F.; Lai, R. Y.; Plaxco, K. W. Chem. Commun. 2007, 36, 3768–3770. (7) Immoos, C. E.; Lee, S. J.; Grinstaff, M. W. J. Am. Chem. Soc. 2004, 126, 10814–10815. (8) Xiao, Y.; Lubin, A. A.; Baker, B. R.; Plaxco, K. W.; Heeger, A. J. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 16677–16680. (9) Xiao, Y.; Qu, X.; Plaxco, K. W.; Heeger, A. J. J. Am. Chem. Soc. 2007, 129, 11896–11897. (10) Ricci, F.; Lai, R. Y.; Heeger, A. J.; Plaxco, K. W.; Sumner, J. J. Langmuir 2007, 23, 6827–6834. (11) Lai, R. Y.; Seferos, D. S.; Heeger, A. J.; Bazan, G. C.; Plaxco, K. W. Langmuir 2006, 22, 10796–10800. 10.1021/ac802317k CCC: $40.75  2009 American Chemical Society Published on Web 02/12/2009

Figure 1. We have employed three stem-loop and three linear E-DNA probes in this study, varying in length and redox-tag placement. The stem-loop probes were of lengths 27, 39, or 44 bases and contained either 17 (short probe) or 34 base (long probes) recognition elements. The linear probes, which are either 17 or 34 bases, contain the same recognition element as the equivalent stem-loop probe. Table 1. Probe DNA Sequencesa

a

probe name

sequence (5′-3′)

short stem-loop (27 bases) long stem-loop (44 bases) internal stem-loop (39 bases) short linear (17 bases) long linear (34 bases) internal linear (34 bases)

GAC ACT GGA TCG GCG TTT TAT TGT GTC GACAATGG ATC GGC GTT TTA TTC TTG TTC AGA TAT TCA ATTGTC ACA AGT GGA TCG GCG TTT TAT TCT TGT*TCA GAT ATT CAA TGGATCGGCGTTTTATT TGGATCGGCGTTTTATTCTTGTTCAGATATTCAA TGGATCGGCGTTTTATT*CTTGTTCAGATATTCAA

Where T* ) thymine modified with MB. Bolded sequence indicates the portion of the probe that forms the stem in the stem-loop constructs.

bases from the electrode-bound terminus. Here we study the performance of each of these distinct E-DNA constructs in terms of signal gain, equilibration time, specificity, and selectivity. EXPERIMENTAL SECTION Materials. The probe and target sequences we have employed are based on the Salmonella typhimurium (gyrB gene) sequence.12-14 The sequences of the redox-tagged probe oligonucleotides and target oligonucleotides are as follows (and listed in Tables 1 and 2, respectively): short stem-loop, 5′-HS-(CH2)6-GACACTGGATC GGCGTTTTATTGTGTC-(CH2)7-NH-MB-3′; long stem-loop, (12) Lai, R. Y.; Lagally, E. T.; Lee, S.; Soh, H. T.; Plaxco, K. W.; Heeger, A. J. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 4017–4021. (13) Kakinuma, K.; Fukushima, M.; Kawaguchi, R. Biotechnol. Bioeng. 2003, 83, 721–728. (14) McClelland, M.; et al. Nature 2001, 413, 852–856.

5′-HS-(CH2)6-GACAATGGATCGGCGTTTTATTCTTGTTCAG ATATTCAATTGTC-(CH2)7-NH-MB-3′; internal stem-loop, 5′HS-(CH2)6-ACAAGTGGATCGGCGTTTTATTCTTGT*(MB) TCAGATA TTCAA-3′; short linear, 5′-HS-(CH2)6-TGGATC GGCGTTTTATT-(CH2)7-NH-MB-3′; long linear, 5′-HS(CH2)6-TGGATCGGCGTTTTATTCTTGTTCAGATATTCAA(CH2)7-NH-MB-3′; internal linear, 5′-HS-(CH2)6-TGGATC GGCGTTTTATT*(MB)CTTGTTCAGATATTCAA-3′;(T1)5′-AAT AAAACGCCGATCCA-3′; (T2) 5′-TTGAATATCTGAACAAGA ATAAAACGCCGATCCA-3′; (T3) 5′-AATAAAATATCGATCCA3′; (T4) 5′-TTGAATATCTGAACAAGAATAAAACGCCGATCC ACCCGAATATCTTCTAT-3′; (T5) 5′-TTGAATATAATAACAAG AATAAAATATCGATCCACCCGAATATCTTCTAT-3′; (T6) 5′GGCATCAAGGCGTTTGTTGAATATCTGAACAAGAATAAA ACGCCGATCCA-3′; (T7) 5′-GGCATCAAGGCGTTTGTTGAA TATCTGAACAAGAATAAAATATCGATCCA-3′; (T8) 5′-GGC Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

2151

Table 2. Target DNA Sequences target name (T1) (T2) (T3) (T4) (T5) (T6) (T7) (T8)

17-base 34-base 3-base mm-17b 50-base 3′-overhang 6-base mm-3′-overhang 50-base 5′-overhang 3-base mm-5′-overhang 10-base mm-5′-overhang

sequence (5′-3′) AATAAAACGCCGATCCA TTGAATATCTGAACAAGAATAAAACGCCGATCCA AATAAAATATCGATCCA TTGAATATCTGAACAAGAATAAAACGCCGATCCACCCGAATATCTTCTAT TTGAATATAATAACAAGAATAAAATATCGATCCACCCGAATATCTTCTAT GGCATCAAGGCGTTTGTTGAATATCTGAACAAGAATAAAACGCCGATCCA GGCATCAAGGCGTTTGTTGAATATCTGAACAAGAATAAAATATCGATCCA GGCATCAAGGCGTTTGTTTAATATAATAAAAATAATAAAATATCGATACA

ATCAAGGCGTTTGTTTAATATAATAAAAATAATAAAATAT CGATACA-3′, where -(CH2)7-NH-MB-3′ represents a methylene blue (MB) added to the terminal phosphate via a C7-amino linker and T*(MB) represents a thymine nucleotide modified by the addition of MB to a six-carbon, amino-terminated linker attached at the 5 position of the nucleobase. The bolded, italicized lettering in these sequences indicates bases forming the stem in the stem-loop constructs, and the nonitalicized bolded lettering indicates bases in the stem that are in the GyrB gene sequence. All probes were synthesized, labeled, and purified by the BioSearch Technologies (Novato, CA) and used as received. Methylene blue (MB) was conjugated to either the 3′ end of the probe or the internal linker-modified thymine via succinimide ester coupling (MB-NHS, obtained from EMP Biotech, Berlin, Germany). The modified oligonucleotides were purified via C18 reversed-phase high-performance liquid chromatography (HPLC) and polyacrylamide gel electrophoresis (PAGE) and confirmed by mass spectrometry. The 17-, 34-, and 50-base target sequences were adopted from the Salmonella GyrB gene to complement 17-34 bases on the probe with the remaining overhang on the 50-base targets being either on the 3′ or the 5′ end of the target and are denoted (T1-T8, Figure 1, Table 2). Underlined bases in the target sequences represent point mutations. The target DNA oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). Target DNA concentrations were confirmed via UV absorbance measurements at 260 nm and used as received. 6-Mercaptohexanol (Sigma-Aldrich, St. Louis, MO), tris(2carboxyethyl) phosphine hydrochloride (Molecular Probes, Carlsbad, CA), hexaamineruthenium(III) chloride (RuHex; Strem Chemicals Inc., Newburyport, MA), fetal calf serum (SigmaAldrich), and guanidine hydrochloride (Pierce, Rockford, IL) were used as received. Fabrication and Characterization of E-DNA Sensors. Polycrystalline gold disk electrodes (2 mm diameter, CH Instruments, Austin, TX) were prepared by polishing with 1 µm diamond and 0.5 µm alumina (Buehler, Lake Bluff, IL), 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 probe DNA by incubating the clean electrode in 0.1 µM DNA/2 µM TCEP (tris(2-carboxyethyl) phosphine hydrochloride) in 1 M NaCl/10 mM potassium phosphate pH 7 buffer for 30 min. The surface was then rinsed with water and subsequently passivated with 5 mM 6-mercaptohexanol in 1 M NaCl/10 mM potassium phosphate buffer, pH 7 for 3 h. Prior to use, electrodes were rinsed with 2152

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

deionized water. For a more thorough review of the sensor fabrication please see ref 15. Electrochemical analysis was performed using alternating current voltammetry (ACV) at 10 Hz frequency, 25 mV amplitude, over a potential range of -0.05 to -0.45V, using a CHI 630B potentiostat (CH Instruments, Austin, TX) in a standard cell with a platinum counter electrode and Ag/AgCl (3 M NaCl) reference electrode. Here we report total current; similar trends, however, are observed for both the in-phase and out-of-phase components of the ac signal (data not shown). All experiments were conducted in 1 M NaCl/10 mM potassium phosphate buffer, pH 7 (“buffered saline”) or in fetal calf serum (“serum”) diluted 1:1 with high-salt buffered saline (2 M NaCl/10 mM potassium phosphate buffer, pH 7). All reported values represent the mean and standard error of the mean (SEM) of three measurements conducted using three independently fabricated electrodes. Surface Coverage of E-DNA Sensors. In order to compare all of the constructs, surface coverages were determined using a previously described method.16 Briefly, we monitored the association of hexaamineruthenium(III) chloride (RuHex ) [Ru(NH3)6]3+) with the E-DNA probes via chronocoulometry over a range of 0 V to -0.35 V to 0 V at a pulse period of 250 ms in a standard cell with a platinum counter electrode and Ag/AgCl (3 M NaCl) reference electrode using a CHI 630B potentiostat. In order to calculate the probe density, chronocoulometry measurements were taken in 10 mM Tris-EDTA pH 7.4 and then compared with those taken in 500 µM RuHex/ 10 mM Tris-EDTA pH 7.4 (after the sensors incubated in the solution for 30 min). These measurements can be used to quantify surface immobilized DNA because one RuHex cation associates with three DNA-backbone phosphates, allowing the total number of DNA molecules to be determined.16,17 Reported coverages represent the mean and standard deviations of three measurements conducted using three independently fabricated electrodes. Sensor Response to a Fully Complementary Target. The signaling of each construct was compared by testing each against one short (T1 or T2) and two 50-base (T4 and T6), fully complementary targets at 200 nM (previously determined ab intra to constitute saturating conditions) in situ in buffered saline. Following interrogation, the probes were regenerated by rinsing with 18 MΩ Milli-Q filtered water for ∼30 s, followed by incubation in buffered saline for ∼1-2 min to allow for the reformation of the stem-loop structure (when necessary) and reinterrogated. (15) Xiao, Y.; Lai, R. Y.; Plaxco, K. W. Nat. Protoc. 2007, 2, 2875–2880. (16) Lao, R.; Song, S.; Wu, H.; Wang, L.; Zhang, Z.; He, L.; Fan, C. Anal. Chem. 2005, 77, 6475–6480. (17) Steel, A. B.; Herne, T. M.; Tarlov, M. J. Anal. Chem. 1998, 70, 4670–4677.

Sequence Specificity. Specificity measurements were performed by incubating the electrodes in 200 nM of either a 50base complementary target (T4) or target DNA containing a 3-base mismatch (T5) in buffered saline in situ and interrogating with ACV. Electrodes were incubated in target and interrogated every 10 min to monitor target binding over the course of 90 min. Probe regeneration procedures are the same as described above. Sensor Response in Blood Serum. In order to test the selectivity of each sensor, the behavior of all six sensor geometries were also assessed in fetal calf serum diluted 1:1 with 2 M NaCl/10 mM potassium phosphate buffer, pH 7 in order to control pH and ionic strength. Electrodes were first interrogated in buffered saline (for comparison), then incubated in target-free, 50% serum for 15 min and interrogated in situ at room temperature. The serum sample was then doped with the complementary target (T6) at 200 nM, and each electrode was interrogated every 10 min for 100 min The sensors were then regenerated via a 30 s soak in 8 M guanidine hydrochloride followed by a ∼5 min incubation in serum before being interrogated again to determine signal recovery after the initial use and regeneration. RESULTS AND DISCUSSION Effects of Probe Geometry, Probe Length, and Redox-Tag Placement on E-DNA Signaling. We have characterized the effects of probe geometry, probe length, and redox-tag placement on the behavior of the E-DNA sensor using a test bed comprised of three linear probe sensors and three stem-loop probe sensors (Figure 1). Each set of three contains one probe that hybridizes to a 17-base recognition sequence (T1) and two probes that hybridize to a 34-base recognition sequence (T2) (Figure 2). The redox tags on the two shorter probes are attached at the termini distal to the electrode attachment site. Among the two longer probes in each set is one probe in which the redox tag is at the distal terminus and in the other the redox tag is in the middle of the 34-base hybridization region. The length and placement of the MB tag in each probe is held as similar as possible to those of equal-length probes of the other geometry so as to facilitate comparisons between the two. Probe geometry has a significant effect on E-DNA signaling such that we observe that the target-induced signal change of linear probe E-DNA sensors is generally greater than that of the equivalent stem-loop probe. This is seen, for example, in a comparison between the short linear and stem-loop probes: the 17-base linear probe responds with greater signal suppression (81.4%) when challenged with a 17-base target (T1) than does the 27-base stem-loop probe (27.5% signal suppression), even though both probes are hybridized to the same 17-base sequence (Figure 3). We presume this effect occurs for two reasons. First, hybridization with the stem-loop probe requires rupture of the stem, which is unfavorable; a step that is eliminated in hybridization with the linear construct. Second, the five stem-forming unhybridized bases flanking the recognition region of the stemloop probe may increase the flexibility of the “bound” probe, allowing the MB redox tag to transfer electrons more readily in the hybridized state (Figure 3). Optimal probe length appears to represent a compromise between signaling and affinity. That is, while longer regions of complementarity improve hybridization thermodynamics (and

Figure 2. We employed four distinct DNA targets in this study, all corresponding to the Salmonella typhimurium LT2 DNA gyrase subunit B gene. The first pair of targets (top), of identical length to the recognition elements, are 17 or 34 bases (for short and long probes, respectively). The second pair of targets, both 50 bases in length, are significantly longer than the 17- or 34-base recognition elements leading to overhanging, single-stranded tails of 33 or 16 bases at either the 3′ (middle) or 5′ (bottom) termini.

thus increase the population of target-bound probes at a given target concentration), longer probes also reduce the current observed in the absence of target (Figure 3), presumably because they alter the efficiency with which the attached redox tag approaches the electrode. Thus, despite the presumably greater affinity of the latter, the signal suppression of the short linear probe is greater than that of the long linear probe (Figure 3). Consistent with this, in the absence of target we observe significantly more current from the short probe than from the long probe (Table 3).6,18 This finding may explain the difference in signaling if the greater initial signal per probe translates into a larger change in relative current upon target hybridization. In contrast to the linear probes, however, the relationship between probe length and signaling in the stem-loop probes is much less pronounced (Table 3), presumably because formation of the stem holds the redox tag in similar proximity to the electrode irrespective of the length of the loop. The judicious placement of the reporting redox tag can ameliorate the reduced signaling observed for longer probe lengths. For example, although the signal suppression of long linear probes is less than that of short linear probes (Figure 3), relocating the MB tag from the terminus to the center of the long probe produces a signal change surpassing that of (18) Peterson, A. W.; Heaton, R. J.; Georgiadis, R. M. Nucleic Acids Res. 2001, 29, 5163–5168.

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

2153

Figure 3. All six of the constructs we have investigated readily detect their corresponding target, albeit with sometimes significantly differing signal suppression. Here are ac voltammograms of all six constructs before and after the addition of the relevant 17- or 34-base target (T1 or T2) (at 200nM). In general, the linear probes show greater signal suppression then the analogous stem-loop probes (except for the long terminally labeled probes which respond similarly). Within each geometry (stem-loop or linear), the longer internally labeled probe produces the greatest suppression (∼95%). All six probes are easily regenerated with a 30 s rinse in distilled water. Table 3. Surface Coverage vs Peak Current Data for All Probesa

surface coverage (1012 probes/cm2) peak current (Ip/nA) a

short stem-loop

internal stem-loop

long stem-loop

short linear

internal linear

long linear

1.84 ± 0.3 126 ± 30

1.94 ± 0.5 149 ± 40

2.19 ± 0.8 233 ± 80

3.04 ± 0.6 197 ± 50

2.91 ± 0.4 50 ± 20

2.61 ± 0.1 38 ± 4

Averages represent mean measurements and standard deviations from three independent electrodes.

the short probe (Figure 3), presumably because the internally labeled probe produces a similar signal change upon binding, but binds with higher affinity. This trend also holds for stemloop probes: an internal MB on a longer stem-loop probe leads to improved signaling relative to that of the short, terminally modified stem-loop probe (Figure 3). In this case the effect likely occurs because in the target-bound, internally labeled 2154

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

probe, the redox tag is confined within the hybridized portion of the complex (as opposed to the flexible, single-stranded terminus of the distally labeled construct) and is thus more rigidly fixed away from the electrode. The density with which the probe DNAs are packed on the sensor surface affects the sensor signaling and equilibration times19,20 and thus, in order to meaningfully compare unrelated

Figure 4. While all six probes equilibrate within similar times when challenged with short targets, the three linear probes respond more rapidly than their stem-loop counterparts when challenged with longer targets. Illustrated here are the target response times for all six probes with the short targets (T1 and T2), 50-base, 3′-overhanging target (T4), and 50-base, 5′-overhanging target (T6). All three linear probes show similar response times; however, among the stem-loop probes, the long internally labeled probe responds much more rapidly than the other two terminally labeled probes. See Table 4 for data.

probe geometries, it is important to maintain similar probe densities.10 We have employed chronocoulometry with hexaamineruthenium(III) (RuHex) to quantify the packing densities of our probes in order to ensure that this condition is met.16,17 Despite similar packing densities, however, the measured faradaic currents observed for the six architectures vary significantly (Table 3). For example, while the stem-loop and short linear constructs produce effectively indistinguishable currents in the absence of target (126 ± 30 to 233 ± 80 nA, Table 3), both internally and terminally labeled, longer linear probes produce much smaller currents (50

± 20 nA and 38 ± 4 nA, respectively, Table 3). The smaller current observed (per MB at a fixed probe density as measured via RuHex) likely reflects differences in electron transfer between the stem-loops, all of which fix their MB tags close to the electrode, the short linear probe, which likewise supports efficient electron transfer, and the long linear probes, which exhibit relatively limited electron transfer even in the absence of target. (19) Jayaraman, A.; Hall, C. K.; Genzer, J. J. Chem. Phys. 2007, 127, 144912– 144922.

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

2155

Table 4. Sensor Response Timesa target

short stem-loop

internal stem-loop

long stem-loop

short linear

internal linear

long linear

short 50b3′ 50b5′

0.9 ± 0.5 9.7 ± 0.2 24.2 ± 9.0

0.7 ± 0.2 3.2 ± 1.5 4.9 ± 0.5

1.2 ± 0.4 5.0 ± 0.4 30.3 ± 9.0

0.8 ± 0.2 1.9 ± 1.3 6.9 ± 2.0

1.0 ± 0.1 3.7 ± 0.3 2.5 ± 1.7

0.8 ± 0.3 8.8 ± 3.6 3.2 ± 1.6

a Time at which 50% of the signal suppression has occurred. Measurements done at a target concentration of 200 nM. Response times (in minutes) represent the mean and standard deviation from three independent electrodes.

Figure 5. Although long, internally labeled probes produce the greatest suppression, the thermodynamics of hybridization are such that shorter probes exhibit improved sequence specificity. Internally labeled (top) and short (bottom) probes were challenged with either the perfectly matched (T4) or a 6-base mismatch 50-base 3′ overhanging target (T5) (both at 200 nM). When challenged with a 6 out of 34-base mismatched target (top), the internally labeled stem-loop probe responds to the mismatched target with 87.7 ( 1.9% of the complementary target signal suppression while the linear probe responds with 92.8 ( 1.8% of the complementary target signal suppression. Whereas, when the short probes are instead challenged with the same 6-base pair mismatched target (of which 3 mismatched bases are within the 17-base recognition sequence) (bottom), the short linear probe responds with only 62.0 ( 1.2% of the possible response for a complementary target while the stem-loop probe discriminates even more effectively with only 28.5 ( 0.7% of a possible response to the mismatched target. Data represents the mean and standard error of the mean (SEM) for three independent sensors.

For the short and long probes, respectively, the tests described above employed targets of 17 or 34 bases. Testing both the short and long probes with the same target, however, may provide a meaningful comparison as this eliminates any differences in target behavior (for example, propensity for internal secondary structure or differing diffusion rates). We thus also tested all of our sensors against each of a pair of (20) Markham, N. R.; Zuker, M. Nucleic Acids Res. 2005, 33, W577–W581.

2156

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

50-base targets, one of which results in a 3′-overhang (T4) adjacent to the electrode, and the second of which results in a 5′-overhang (T6) that is distal to the electrode (Figure 2). We find that, as was true for the shorter targets, the internally labeled probes produce the greatest signal suppression with both of the longer targets (Figure 4). (21) Relogio, A.; Schwager, C.; Richter, A.; Ansorge, W.; Valcarcel, J. Nucleic Acids Res. 2002, 30, e51.

Table 5. Self-Hybridizing ∆G and Tm Values Calculated for Each 50-Base Target target (T4) (T5) (T6) (T7) (T8)

50-base-3′-overhang 6-base mm-3′-overhang 50-base-5′-overhang 3-base mm-5′-overhang 10-base mm-5′-overhang

∆Ga (kcal/mol)

Tma (°C)

-2.47 -2.23 -8.54 -4.25 -2.23

43.2 38.2 59.4 48.2 34.7

a ∆G and Tm values were calculated using mfold Hybridization Server20 (http://frontend.bioinfo.rpi.edu/applications/hybrid/ twostate.php) and represent the greatest ∆G and Tm values for solution-phase oligos at 25 °C, 1 M NaCl.

Figure 6. All six probes perform well in complex samples, such as blood serum. This histogram depicts signal suppression observed at 200 nM target concentration when challenged in buffer versus challenged in 50% serum (50-base, 5′-overhang target, T6). The only appreciable difference (differences are within error at 95% confidence for the stem-loop constructs and 99% for the linear constructs) is seen with the short stem-loop and linear probes in which there is a notable decrease in signal suppression when the sensor is run in 50% serum. Data represents the mean and standard error of the mean (SEM) for three independent sensors.

Effects of Probe Geometry, Probe Length, and Redox-Tag Placement on E-DNA Response Time. There is little difference in equilibration times between the stem-loop probes and the linear probes when they are challenged with short targets. For example, the linear probes achieve 50% of their final signal change in 0.8 ± 0.2 to 1.0 ± 0.1 min, and the equivalent equilibration times for stem-loop probes are 0.7 ± 0.2 to 1.2 ± 0.4 min (Figure 4, Table 4). Neither the length nor the location of the redox reporter seems to greatly affect the equilibration time in either set of probes (Figure 4). In contrast, however, all six probes hybridize more slowly to long targets than to short targets, with the stem-loop probes responding the slowest. Likewise, while linear probes respond to both of our two long targets with similar kinetics, we observe rather heterogeneous response times with the stem loop probes. For example, the stem-loop probes respond more rapidly to the 3′-overhang target (T4) than to the 5′-overhang target (T6) (Figure 4 and Table 4). We presume this occurs because the 3′overhang is proximal to the electrode surface, leading to kinetically important steric effects. Effects of Probe Geometry, Probe Length, and Redox-Tag Placement on E-DNA Specificity. The various E-DNA sensor probes show varying capabilities of discriminating between

perfectly complementary DNA and multiple-mismatch targets and illustrate an often fundamental tradeoff between specificity and affinity. For example, although we observed the greatest signal suppression with the long, internally labeled probes, our shorter probes exhibit improved specificity. To illustrate this we note that, while the short linear and short stem-loop probes are suppressed by 86.8% and 28.0% in the presence of the perfectly complementary short target (T1), their signals are suppressed only 64.0% and 19.6%, respectively, by the same concentration of a 3-base pair mismatch target (T3) (data not shown). The poorer specificity of longer probes presumably occurs because, as the stability of the probe-target duplex increases, the change in the population of bound probes produced by a mismatch becomes smaller, rendering the differential current similarly small. The specificity observed for longer targets is more complex. For example, challenging the long, internally labeled probes with 50-base 3′-overhang targets (T4 and T5), we observe only a small discrimination for the perfectly matched target over one containing a 6-base pair mismatch (out of the 34 bases in the recognition element) (Figure 5). In contrast, however, the short linear and stem-loop probes exhibit 92.9 ± 0.2% and 45.7 ± 0.5% suppression for the perfectly matched target, respectively, and only 57.6 ± 1.9% and 13.0 ± 0.8% for the mismatched target (Figure 5). Thus, the short probes exhibit improved specificity even though the ratio of correct base pairing to mismatched base pairing is the same for both the long and short probes (6 bases out of 34 versus 3 bases out of 17) and thus the change in calculated free energy for each complex is likewise proportional.20 This discrepancy in target discrimination may occur because the greater stability of the duplexes formed between longer probes and their targets reduces their ability to discriminate mismatches.21,22 In addition to the effects noted above, the probe response to longer targets is also affected by the secondary structure content and geometry of the targets. When challenged with 5′-overhang targets (T6 and T7), for example, the short probes have difficulty in distinguishing between perfect match (T6) and 3-base mismatch (T7) targets. We presume this occurs because the complementary target (T6) exhibits significantly more secondary structure than the mismatched target (T7), thus reducing the difference in binding free energy between the two (Table 5, Figures S-3, S-4, and S-6 in the Supporting Information). The internally labeled probes likewise fail to exhibit significant discrimination until the number of mismatches is increased to 10 (T8) (Figures S-5 and S-6 in the Supporting Information). In addition to these thermodynamic consequences, target secondary structure also affects sensor response times,23,24 which will affect specificity if complete equilibration has not been achieved. This is illustrated in the observation that these sensors respond more rapidly to the 5′overhang target that contains a 3-base mismatch (T7) than the perfectly matched counterpart (T6); even after 100 min with the perfectly matched target (T6), the probes are not fully equilibrated (Table 5 and Figure S-6 in the Supporting Information). Both perfect match and mismatched 3′-overhang targets (T4 and T5), in contrast, equilibrate rapidly and do not exhibit obvious kinetic (22) Koltai, H.; Weingarten-Baror, C. Nucleic Acids Res. 2008, 36, 2395–2405. (23) Kushon, S. A.; Jordon, J. P.; Seifert, J. L.; Nielsen, H.; Nielsen, P. E.; Armitage, B. A. J. Am. Chem. Soc. 2001, 123, 10805–10813. (24) Du, H.; Strohsahl, C. M.; Camera, J.; Miller, B. L.; Kraus, T. D. J. Am. Chem. Soc. 2005, 127, 7932–7940.

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

2157

secondary structure effects (Figure 5, Figures S-1 and S-2 in the Supporting Information, and Table 5). In addition to these secondary structural effects, the placement of the overhanging region of the longer targets (the portion not bound to the probe) may also affect specificity; steric hindrance resulting from the proximal tail of the 3′-overhang target may force greater target discrimination by destabilizing the probe-target duplex relative to the 5′-overhang targets. Effects of Probe Geometry, Probe Length, and Redox-Tag Placement on E-DNA Selectivity. Because signal generation in the sensor is based on a hybridization-linked change in the flexibility of the probe and not on simple adsorption to the sensor surface, E-DNA is largely impervious to false positives arising from the adsorption of nonoligonucleotide contaminants to the sensor surface.11,25,26 Consistent with this, all six of our E-DNA constructs are selective enough to deploy directly in complex, clinically relevant samples, with each producing similar behavior in 50% serum as in pure buffer (Figure 6). The most notable difference in behavior between buffer samples and serum samples is seen with the short probes where there is a drop in signal suppression upon target detection in 50% serum samples (Figure 6). CONCLUSION Here we present a comparative study of a half-dozen distinct E-DNA architectures in an effort to elucidate the effects of probe geometry, probe length, and redox-tag placement on the performance of this electrochemical sensing platform. We find that linear probes exhibit improved signal changes relative to the equivalent stem-loop probes and that, consistent with their improved ther(25) Lubin, A. A.; Lai, R. Y.; Baker, B. R.; Heeger, A. J.; Plaxco, K. W. Anal. Chem. 2006, 78, 5671–5677. (26) Lubin, A. A.; Fan, C.; Schafer, M.; Clelland, C. T.; Bancroft, C.; Heeger, A. J.; Plaxco, K. W. Forensic Sci. Commun. 2008, 10, 1.

2158

Analytical Chemistry, Vol. 81, No. 6, March 15, 2009

modynamics, longer probes exhibit improved signaling relative to shorter probes provided the redox tag is located proximal to the electrode. In addition to improved signaling, linear probes also equilibrate more rapidly than the equivalent stem-loop probe, but we do not observe any consistent relationship between equilibration times and probe length or redox-tag placement. Finally, we find that the specificity and selectivity of linear and stem-loop probes are effectively indistinguishable, the level of specificity is in part driven by the length of the target, its internal secondary structure, and the orientation of the unhybridized tail on the target with the sensor surface. This highlights the importance that choosing an optimal E-DNA construct for a specific application will depend on whether the most crucial attribute is signaling or sequence specificity. ACKNOWLEDGMENT The authors gratefully acknowledge support from the Institute for Collaborative Biotechnologies through Grant DAAD19-03-D0004 from the U.S. Army Research Office and from the Center for Nanoscience Innovation for Defense through Grant H9400307-2-0704 from Defense Microelectronics Activity (DMEA) and a fellowship from the Santa Barbara Foundation Tri-Counties Blood Bank (to R.J.W.). SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review November 3, 2008. Accepted January 20, 2009. AC802317K