Article Cite This: Anal. Chem. 2019, 91, 8244−8251
pubs.acs.org/ac
Divergent Pair of Ultrasensitive Mechanoelectronic Nanoswitches Made out of DNA Fen Ma,†,‡ Lin Qi,‡ Owen Einarson,§ Dipankar Sen,*,‡,§ and Hua-Zhong Yu*,‡,§ †
Downloaded via GUILFORD COLG on July 24, 2019 at 07:40:20 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Key Laboratory of Synthetic and Natural Functional Molecular Chemistry (Ministry of Education), College of Chemistry and Materials Science, Northwest University, Xi′an, Shaanxi 710127, China ‡ Department of Chemistry and §Department of Molecular Biology & Biochemistry, Simon Fraser University, Burnaby, BC V5A 1S6, Canada S Supporting Information *
ABSTRACT: Mechanoelectronic DNA nanoswitches refer to designed oligonucleotide constructs that are composed of conduction-interrupted duplex stems functionally coupled to ligand recognition motifs; they have been shown to undergo remarkable conduction switching upon binding molecular ligands/analytes. Herein we report a divergent pair of such mechanoelectronic DNA switches, the “signal-on” 3′AA-1 switch and the “signal-off” NB-1 switch, both activated by and responded to mercury ions (Hg2+) at nM levels. We first investigated their charge transport efficiency at a biochemical level, by studying how distinct base sequence at the switches’ central three-way junction and at the recognition motif (capable of forming T-Hg2+-T metallo-base pairs) influences their overall conductivity. Gel electrophoresis assays revealed that the presence of two unpaired adenines (AA) at the junction led to “signal-on” behavior with increasing Hg2+ concentration; divergently, absence of these adenines led to a “signal-off” behavior. Upon immobilization on gold electrodes, both DNA switches, with enhanced and inhibited conductivity, respectively, showed excellent sensitivity as well as selectivity toward Hg2+ and can be regenerated for multicycle applications. The high performance of these devices, as both nanoswitches and biosensors with robust and reproducible properties, highlights their potential as an outstanding new class of DNA mechanoelectronic components with built-in biosensing capabilities.
B
principle relies on analyte binding-induced conformational change, or association/dissociation of DNA complexes, which in turn produce varied electrochemical responses.10−17 One of the limitations of this approach has been both the need for identifying functional DNA structures (e.g., aptamers) that undergo global scale conformational changes, and the necessity of multiple washing steps to complete the strand association or dissociation processes. In the past two decades, we have been exploring a novel sensing mechanism based on the analyte-induced conductivity change of specially designed DNA switches.18−22 Our design of so-called mechanoelectronic DNA switches consists of a disrupted conduction path (two adjacent but discontinuous Watson−Crick paired duplex stems) structurally coupled to a ligand recognition motif (an unstructured DNA domain with specific sequences for ligand/analyte binding). The binding of analyte/ligand to the recognition motif induces subtle conformational changes at the three-way junction between the conduction path and the coupled recognition arm, to generate remarkably enhanced conductivity through otherwise
esides the genetic storage and propagation functions, DNA has gained extensive interest as a versatile building block for fabricating nanostructures and devices, due to its unique base-pairing, structural, and conformational properties.1,2 Functional DNA switches are specially designed oligonucleotide constructs that can undergo either configuration or conductivity changes in the presence of certain molecular stimuli. Such a molecule/ligand-gated switching behavior promises the application of DNA switches for designing and constructing biosensors, where stimuli-induced conformation and conductivity changes are transduced to different types of signal readouts. Many different types of functional DNA (e.g., aptamers, DNAzymes, G-quadruplexes, and i-motif) have been explored as building blocks for constructing DNA switch-based biosensors for analytes beyond nucleic acids (DNA or RNA), i.e., proteins, cells, and small molecules or ions.3−8 Electrochemistry is one of the most popular readout methods for biosensors owing to its advantages of rapidity, instrumental simplicity, and high sensitivity. In addition, the facile immobilization of DNA on metal electrodes through strong gold−sulfur interaction,9 makes it convenient to develop DNA-based electrochemical biosensors. In such DNA switch-based biosensors, the most common sensing © 2019 American Chemical Society
Received: February 18, 2019 Accepted: May 28, 2019 Published: May 28, 2019 8244
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry
Scheme 1. Sequences and Modifications of 3′AA-1 and NB-1 Mechanoelectronic DNA Switches for Gel Electrophoresis Assays of Hg2+-Responsive Charge Transport Conductivity Testsa
a The sequences in black and blue are the duplex conducting stems and the recognition arm of Hg2+, respectively. The three nucleotides in red are the reporter guanines for examining the DNA charge flow induced oxidative damage.
disrupted duplex stems.21 Beyond adapting aptamers as the recognition motif for creating ultrasensitive electrochemical biosensors for protein biomarkers,18,20 we have shown the general versatility of such functional DNA switches for building robust nanoelectronics via incorporating K+-responsive Gquadruplexes.22 It has been recently demonstrated biochemically that mechanoelectronic DNA switches can be built with relatively simple DNA recognition sequences, for example, mercury cation (Hg2+) binding thymine−thymine (T-T) mismatches.19 In this case, concomitant mechanical (structural) and electrical (conductivity) switching are activated with the formation of THg2+-T base pairs in the recognition arm. Herein, we investigate the properties of a pair of such mechanoelectronic DNA nanoswitches to explore if authentic “turn-on” and “turnoff” responses can be achieved in response to the same analyte stimulus. We first studied the charge transport properties of the DNA switch pair (3′AA-1 and NB-1, Scheme 1) using standard biochemical gel assays for monitoring charge-flow induced guanine oxidative damage in response to various amounts of Hg2+. The confirmed “signal-on” and “signal-off” DNA switches, with redox tags (ferrocene) appended, were then immobilized onto gold electrodes through a thiol linker, and their on-chip performance as a function of Hg2+ concentration was evaluated using both cyclic voltammetry (CV) and squarewave voltammetry (SWV).
Sigma-Aldrich (St. Louis, MO). All chemicals were of ACS reagent-grade and were used as received. Deionized water (>18.2 MΩ·cm) was produced from a Barnstead EasyPure UV/UF compact water system (Dubuque, IA) and used for preparing all solutions and buffers. The oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA), for which the sequences are listed in Table S1. Gold-coated glass slides (100 nm Au and 5 nm Cr) were purchased from Evaporated Metal Films, Inc. (Ithaca, NY). Biochemical Assays of Hg2+-Responsive Charge Transport of DNA Switches. All oligonucleotides were pretreated with 10% aqueous piperidine at 90 °C for 30 min to cleave any pre-existing DNA damage sites. The unmodified oligonucleotides (NB-1 and 3′AA-1) were also purified by denaturing polyacrylamide gel electrophoresis (PAGE) containing 7 M Urea in Tris-borate-EDTA (TBE) buffer. DNA was eluted from gel pieces by crush and soak into 10 mM TrisEDTA (TE) buffer, precipitated by using ethanol, and redissolved in water. 5′-Anthraquinone (AQ) conjugated C1 oligonucleotide was prepared by reacting 5′-C6-NH2-modified C1 with the N-hydroxysuccinimidyl (NHS)-ester of anthraquinone-2-carboxylic acid, as described previously.21 The AQmodified C1 oligonucleotide (AQ-C1) was then purified by HPLC through a 4.6 × 100 mm Agilent Eclipse XDB 3.5 μmC18 column (Phenomenex, Torrance, CA). The mobile phase consisted of solvent A: 0.1 M triethylammonium acetate (pH 7.0)/CH3CN (92:8) and solvent B: CH3CN; a linear gradient from 0 to 40% solvent B was used for each run at a flow rate of 1.0 mL/min for 30 min. Hg2+-binding oligonucleotides (NB-1 and 3′AA-1) were 5′-32P-labeled by OptiKinase (Affymetrix, Inc., Santa Clara, CA) and γ-32P-ATP (PerkinElmer, Inc., Waltham, MA) following the manufacturer instruction, precipitated using ethanol, and redissolved in H2O. The DNA constructs were annealed by heating AQ-C1 and 5′-32P-labeled NB-1 or 3′AA-1 to 95 °C for 2 min in 10 mM Tris-HOAc buffer (pH = 8.0), followed by cooling down to
■
EXPERIMENTAL SECTION Reagents and Materials. Tris(hydroxymethyl)aminomethane (Tris) was purchased from Caledon Laboratories Ltd. (Georgetown, ON). 6-Mercapto-1-hexanol (MCH, 97%), [Ru(NH3)6]Cl3, tris(2-carboxyethyl) phosphine hydrochloride (TCEP), ethylenediaminetetraacetic acid (EDTA), mercury(II) acetate (Hg(OAc)2), and other salts (MgCl2, CaCl2, Co(NO3)2, Ni(NO3)2, Cd(NO3)2, Cu(NO3)2, Cr(NO3)2, Zn(NO3)2, and AgNO3) were obtained from 8245
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry room temperature over 30 min. Subsequently, the Hg2+ in binding buffer (5 mM MgCl2, 50 mM Tris-HOAc, pH = 8.0) with desired concentrations was added, and the final concentration of AQ-C1 and 5′-32P-labeled NB-1 or 3′AA-1 were kept at 125 and 80 nM, respectively. The samples were then transferred into a 96-well ELISA plate and equilibrated in an ice−water bath (0 °C). After that, the samples were positioned 4 cm under a UVP Black-Ray UVL56 lamp (Analytik Jena AG, Jena, Germany) and irradiated at 365 nm in a cold room (4 °C) for 30 min (only the 2 rows in the plate center were used to ensure uniform irradiation). Following UV irradiation, the samples were transferred to microcentrifuge tubes for ethanol precipitation. The precipitated DNA samples were resuspended in 100 μL of 10% aqueous piperidine (v/v), heated to 90 °C, and followed by several cycles of lyophilization and resuspension. The products were analyzed by 12% denaturing PAGE. Autoradiography imaging and quantification of the gel results were performed on a Typhoon 9410 phosphoimager with ImageQuant TL7.0 software (GE Healthcare, Little Chalfont, U.K.). Modification and Immobilization of DNA Switches on Gold. Ferrocenecarboxylic acid (Fc-COOH) was first covalently attached to the 5′-termini of C1′ oligonucleotide following the established protocol.26 The resulting Fc-COOH coupled C1′ (Fc-C1′) was further purified with HPLC as described above; the procedure was also reported in the literature.20 Small pieces of gold glass slide (1.0 × 2.5 cm2 were cleaned by immersion into a 90 °C Piranha solution (30% H2O2/96% H2SO4 = 1:3 v/v) for 5−7 min. (Warning: the so-called Piranha solution is highly reactive when contact with organic materials; extreme precautions must be taken during use.) The gold slides were then rinsed thoroughly with water, followed by drying with N2. To avoid potential contaminations, the cleaned gold slides were immediately used for DNA immobilization. To reduce the disulfide bond, the 3′AA-1′ or NB-1′ oligonucleotides were treated with 10 mM TCEP in 40 μL of 100 mM Tris buffer (pH = 7.4) for 5 h. After reduction, the DNA solution was desalted by using an Illustra MicroSpin G25 column (Amersham Bioscience, Bath, U.K.) pretreated with 100 mM Tris buffer in order to remove the cleaved short thiols (HO-(CH2)6-SH). The concentration of the purified, thiolate 3′AA-1′ and NB-1′ oligos were determined by using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA). Subsequently, 1.0 μM of freshly prepared thiolate 3′AA-1′ or NB-1′ were hybridized with 2.0 μM Fc-C1′ in deoxygenated immobilization buffer (10 mM Tris-HCl, 0.1 M NaCl and 5 mM MgCl2, pH = 7.4) at 80 °C for 5 min, followed by slow cooling down to room temperature. Then 10 μL of the above hybridized DNA sample (diluted to desired concentrations with immobilization buffer) was dropped onto a freshly cleaned gold slide and incubated in a humidity chamber under dark for 12−18 h. After immobilization, the gold slides were thoroughly washed with immobilization buffer and incubated with 10 μL of 1.0 mM MCH aqueous solution for 60 min in a humidity chamber in order to passivate the gold surface and remove physically adsorbed DNA strands.27 Electrochemical Measurements. All electrochemical experiments were carried out in a three-electrode, singlechamber cell (made of Plexiglas V-series acrylic resin) with a CHI 660D Electrochemical Analyzer (Austin, TX). The working electrode, a DNA switch-modified gold slide, was attached to the opening side of the cell. The area of working
electrode exposed to the electrolyte (0.14 cm2) was restricted by an O-ring seal. A platinum wire and an Ag|AgCl|3 M NaCl electrode were used as the counter and reference electrode, respectively. For the detection of Hg2+, a Hg(OAc)2 stock solution was diluted using the incubation buffer to desired concentrations. A total of 2 mL of the freshly diluted Hg2+ solution was added into the cell and incubated for at least 30 min, followed by CV and SWV measurements. All electrochemistry experiments were carried out under ambient conditions, and the electrolyte solution was deoxygenated by bubbling Ar for at least 15 min.
■
RESULTS AND DISCUSSION Design of Divergent Mechanoelectronic DNA Nanoswitch Pair. The design of mechanoelectronic DNA switches is based on the hypothesis that the formation of coaxial base stacking between the two adjacent duplex stems allows relatively efficient electron−hole (h+) transport through the three-way junction, and that such coaxial base stacking can be triggered via the local conformational change of the recognition arm in response to analyte/ligand binding.21 Previous studies have revealed that DNA three-way junctions with two unpaired bases located right at the center typically form stable, approximately T-shaped structures with proper and strong coaxial stacking between two of the three duplex arms.23 In contrast, fully complementary DNA three-way junctions (i.e., without unpaired bases at the junction) normally adopt trigonal (or Y-shaped) conformation where the three duplex arms are roughly at 120° to each other; such a conformation leads to poor base-stacking at the junction.24 In our earlier work, we biochemically surveyed a number of junction structures for switching properties via T-Hg2+-T base pair formation in the recognition arm.19 It was found that the highest enhancement of charge transport conductivity was achieved with two “bulged” adenosines at the junction (the 3′AA-1 DNA switch); this was observed biochemically either (i) in the presence or absence of a single Hg2+ concentration (10 μM) against a backdrop of 0.5−5.0 mM Mg2+ also present in the solution; or (ii) in the presence of increasing concentrations of Hg2+ (0−10 μM). A single biochemical data point (10 μM Hg2+) obtained with a different DNA switch (NB-1), however, appeared to show a contrasting behavior, with a decrease of signal upon adding Hg2+. With neither of these DNA switches, however, were any direct electrochemical experiments carried out. While the above biochemical study focused on proving the concept of constructing mechanoelectronic DNA switches with combined PAGE and FRET characterization, herein we explore the feasibility of adapting these functional DNA constructs for authentic, highly sensitive mechanoelectronic switching on electrode for electrochemical sensing. To begin with, we systematically investigated the charge transport property for the above-mentioned two DNA switches in the presence of a full set of concentrations of Hg2+ (Scheme 1). As described previously, the biochemical assay for evaluating the analyte-driven DNA charge transport efficiency relies on examining the charge-flow generated guanine oxidative damage patterns and intensities.21,25 Briefly, the photoexcitation of anthraquinone (AQ) introduces an “electron-hole” into its adjacent DNA stem (to which the AQ is covalently linked). With well aligned coaxial base stacking, the injected electron− hole is able to transport into the other conductive DNA stem through the junction and cause oxidative damage to the three 8246
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry
Figure 1. Electrophoresis gel assays of the charge transport flow generated guanine damage patterns of 3′AA-1 and NB-1 DNA switches in the absence (−) and presence (+) of UV irradiation, AQ labeling, and different concentrations of Hg2+.
consecutive “reporter guanines” at specific positions (G8-G10, highlighted in red in Scheme 1). Such oxidized guanines are notably susceptible to aqueous piperidine-catalyzed DNA hydrolysis and concomitant breakage of the DNA strand at such sites.25 Thus, the degree of guanine oxidative damage, which reflects the hole transport efficiency, can be evaluated by reading the electrophoretic pattern and quantifying the respective band intensities of the DNA fragments. As shown in Figure 1, for both the 3′AA-1 and NB-1 constructs, the charge-flow generated guanine oxidative damage does not occur for the constructs without AQ modification (lanes 1−3) and those without UV irradiation (lanes 4−6), since only very weak bands can be observed at the reporter guanine sites (the activity of piperidine-catalyzed DNA cleavage is much lower at pristine guanine sites25). In contrast, for AQ-modified DNA switches and upon UVirradiation, clear DNA cleavage bands were evident in both cases and are apparently dependent on the concentrations of Hg2+ added (quantitation of oxidative damage at G8 is achieved by dividing its band intensity in a given lane as a percentage of the total intensity of all bands present in that lane). Crucially, as shown in Figure 1 (lanes 1′−9′) for the 3′AA-1 switch, higher concentrations of Hg2+ show stronger DNA cleavage bands in the gel; while the NB-1 switch shows a reverse trend, that is, the DNA cleavage bands became weaker with increased concentrations of Hg2+. The diametrically opposed charge transfer behaviors of these two DNA switches are quantitatively depicted in Figure 2, based on the relative change in the fraction of G8 cleavage (ΔF/F0, where ΔF is the change in the fraction of G8 cleavage band at a given concentration of Hg2+, and F0 is the fraction of G8 cleavage band without Hg2+). These observations confirmed that with specifically designed sequence of the recognition arm and the
Figure 2. Relative change in the fraction of DNA cleavage (ΔF/F0) at G8 damage site as a function of the Hg2+ concentration for 3′AA-1 and (sold circles) NB-1 (open circles) DNA switches. The black lines are the fitting results of their binding isotherm with Hg2+ based on eq 1 (see main text for details).
junction structure, the charge transport property of the DNA switches can be effectively turned to either generating increased signal (“signal-on”) or decreased responses (“signal-off”). The fact that the charge transfer efficiency can be quantitatively modulated through the DNA switches for obtaining either enhanced or inhibited responses, forms the foundations of designing dual-signal electronic sensors for the same analyte (vide infra). 8247
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry
gold electrode for evaluating their practical enhanced and inhibited conductivity switching, as well as the quantitation abilities. As illustrated in Figure 3A, for the purpose of
For a quantitative comparison, we have determined the dissociation constants (Kd) of the two Hg2+ -DNA switch complexes by fitting the binding isotherms shown in Figure 2 with the Hill model (nonequivalent multiple ligand binding sites), as described by eq 1: [Hg 2 +]α ΔF /F0 = [ΔF /F0]sat Kd + [Hg 2 +]α
(1)
The fitting yields the apparent Kd of 4.2 ± 0.6 μM for 3′AA-1 switch and 11 ± 2 μM for NB-1 switch, with their Hill coefficients (α) of 1.5 ± 0.3 and 1.3 ± 0.2, respectively. The slight difference between these two Kd values may attribute to the different number and distribution of T−T mismatches in their recognition arms and the variation of the junction structure.19 Nevertheless, these low-μM dissociation constants determined from the gel experiments are close to those (hundreds of nM to μM) of other DNA−Hg2+ binding constructs determined from either fluorescence spectrometry28 or isothermal titration calorimetry.29 Besides varied cleavage intensities at G8−G10 sites (the reporter guanines for evaluating charge transport efficiency between the two conductive stems of each of the DNA switches), the gel results displayed in Figure 1 showed the impact of Hg2+ binding on photoinduced cleavage at other G sites, a significant source (in addition to relatively large signal deviations inherent in gel assays) of uncertainties in the G8 quantitative data (shown in Figure 2). Specifically, for the 3′AA-1 DNA switch in the absence of Hg2+ binding, degrees of cleavage are also noted at G18, G22, and G24−G26 sites, which are located in the “floppy” recognition arm, and such floppiness may enable collision and possible direct charge transport between these G sites and the AQ moiety on the conductive stem.19,30 This type putative intramolecular contact is impacted by Hg2+-binding, as revealed by significantly decreased cleavage intensities at the extraneous G sites at higher Hg2+ concentrations. Such an inhibition of potential direct charge transport between the AQ moiety and these G sites (G18, G22, and G24−G26) is likely a result of subtle conformational changes within the 3′AA-1 DNA switch upon T-Hg2+-T pair formation in its recognition arm, which separates the AQ moiety away from those nonreporter G sites. For the NB-1 DNA switch, the cleavage intensity at the G14 site (located precisely at the junction) strengthens significantly at higher Hg2+ concentrations. The enhanced activity of cleavage at this guanine site could be attributed to abrupt structural changes that likely occur precisely at the junction of NB-1 DNA switch (e.g., one double helix transitions into another), likely resulting in a higher degree exposure of its junction bases to the ambient solution environment.31 Correspondingly, such an abruptly changed junction structure corresponds well to the poorer stacking and reduced conductivity between the two conductive stems of NB-1 switch upon Hg2+ binding. All these striking observations are highly provocative and will likely contribute to deeper understandings of the exact mechanism(s) of T-Hg2+-T pairing-induced conformational switches in DNA three-way junctions. Such understandings and further investigations, though, lie beyond the scope of the present work. Built-in Biosensing Capabilities of the Divergent Mechanoelectronic DNA Nanoswitch Pair. Following the successful biochemical characterization, we then “transferred” the divergent mechanoelectronic DNA nanoswitch pair onto
Figure 3. (A) Design of dual-signal electrochemical biosensors for Hg2+ with immobilized 3′AA-1′ and NB-1′ switches on gold. (B) Representative square-wave voltammograms (SWVs) of immobilized 3′AA-1′ and NB-1′ DNA switches after incubation with different concentrations of Hg2+; the SWV measurements (negative scan) were carried out as follows: preconditioning at higher potential for a period of time; the step potential, step amplitude, and frequency are 0.004 V, 0.025 V, and 30 Hz, respectively.
electrochemical detection, the easily oxidized guanine triplet was removed from the conductive stem of the DNA switches; the 32P-radiolabeling and AQ modification on the conductive stems were also replaced with a hexanethiol linker for immobilization on gold electrode via strong Au−S interactions9 and an electroactive ferrocene group for direct electrochemical measurements, respectively. The square-wave voltammetry (SWV) was used for measuring the Hg2+-binding responsive charge transport flow generated from the Fc+/Fc redox couple labeled on the immobilized DNA switches. Compared to conventional CV measurements, the SWV combines the linear potential scan with staircase potential pulses, and records differential redox currents generated at the end of forward and reverse potential pulses, which has higher sensitivity and is less affected by the background capacitive current.32 Figure 3B displays the SWV responses of immobilized 3′AA-1′ and NB-1′ DNA switches upon incubation with different concentrations of Hg2+ (the reduction current was recorded in SWV with the negative sign in order to differentiate it from the oxidation current). It is clear that the 3′AA-1′ DNA switch shows enhanced SWV responses upon incubation with higher concentrations of Hg2+, while the peak potentials remained constant. In contrast, for NB-1′ DNA switch the SWV peak becomes smaller when increasing the concentration of Hg2+; similarly, no potential 8248
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry
concentration of Hg2+. For 3′AA-1′ DNA switch, a “signalon” sensing response is clearly observed, that is, the SWV peak current (Ip) increases with higher concentrations of Hg2+; in contrast, the NB-1′ DNA switch shows “signal-off” sensing response, for which the peak current decreases with increased concentrations of Hg2+; moreover, for both 3′AA-1′ and NB-1′ DNA switches their responses become saturated around 20 nM (Figure 4A). By fitting the electrochemically determined binding isotherms shown in Figure 4A with eq 1, it was found that the affinities of these two DNA switches toward Hg2+ binding are significantly improved upon immobilization on the gold surface (Kd = 6 ± 1 nM for 3′AA-1′ switch, and Kd = 8 ± 2 nM for NB-1′ switch). Presumably there are significant structural fluxionalities of these two DNA switches concomitant with their immobilization onto solid substrates, such that those conformations that are more conducive to T-Hg2+-T binding were serendipitously enriched. Interestingly, such high binding affinities have also been reported recently in other immobilized DNA switches of Hg2+, whose binding isotherms presented close to nM-level Kd values.33−35 We acknowledge that the origin of this notable improvement in binding affinities is intriguing and is not fully understood at this point. This is such an observation that clearly deserves future investigation but lies beyond the purpose and scope of this paper. Figure 4B shows the linear sensing responses observed at lower concentrations of Hg2+ (0.5−10 nM for 3′AA-1′ and 1− 10 nM for NB-1′). Owing to the much improved Hg2+ binding affinity, as well as the high sensitivity and small deviations of electrochemical sensing readout, the Hg2+ detection limits of 3′AA-1′ and NB-1′ DNA sensors determined from their linear sensing responses are as low as 0.5 nM and 0.9 nM, respectively. These values are much lower than those required for drinking water quality by the Health Canada (5 nM),36 the U.S. Environmental Protection Agency (10 nM),37 and the National Health Commission of the People’s Republic of China (5 nM).38 The above results confirmed unequivocally that the immobilized DNA switches as electrochemical sensors with “signal-on” and “signal-off” dual responses have exceptional quantitation capabilities for Hg2+. More importantly, when combining the dual-signal responses from these two versions of DNA switches, that is, by fitting the differential peak currents, a more sensitive linear response was obtained (red line in Figure 4B), which provides an even better detection limit (0.3 nM). We have essentially shown here the possibility of developing ultrasensitive biosensors based on the application of dual-signal DNA switches at the same time; such a novel design would be a significant improvement from the traditional ratiometric sensors in terms of simplicity and selectivity (vide infra).39,40 The sensing selectivity of the pair was subsequently confirmed by comparing the responses of 10 nM Hg2+ with other common divalent metal cations at a much higher concentration (e.g., 200 nM; Figure S5 in the SI). More impressively, the regeneration of these two DNA switch-based Hg2+ biosensors were demonstrated by monitoring the SWV responses upon the alternating immersion in 10 nM Hg2+ and 5 μM EDTA. EDTA has extremely high binding affinity toward Hg2+ (Kf > 1020 M−1)41 and is able to remove Hg2+ from THg2+-T pairs.40 For both 3′AA-1′ and NB-1′ DNA switches, the sensing responses of Hg2+ revert to the blank level upon immersing the two Hg2+-bounded DNA switches into EDTA solution (Figure 5); moreover, the sensing and background
shifts were observed in this case either. The SWV responses shown in Figure 3B are consistent with the enhanced and inhibited charge transport efficiency observed in biochemical assay described above, which proves that the immobilization of DNA switches are ideal candidates as electronic sensors for chip-based quantitation of Hg2+ with dual-signal responses, that is, biochemically surveyed mechanoelectronic DNA switches can be readily translated into electrochemical biosensors with dual-signal responses with minimal modification of the sequences and 3′-/5′-end derivatization. The experimental conditions for the best electrochemical sensing results have been optimized, including the surface density of the DNA switches (1.6 × 1012 molecules/cm2) and the incubation time of Hg2+ solution (30 min; Figures S2−S4 in Supporting Information (SI)). Based on the changes of the SWV peak currents with respect to the blank (absence of Hg2+), we show in Figure 4A quantitatively how the sensor responds depend on the
Figure 4. (A) Changes on SWV peak current (ΔIp) after incubating immobilized 3′AA-1′ (solid circles) and NB-1′ (open circles) DNA switches with different concentrations of Hg2+; the black lines are the fitting results of their binding isotherm with Hg2+ based on eq 1 (see main text for details). (B) The linear fittings of the Hg2+ sensing responses at lower concentration range for immobilized 3′AA-1′and NB-1′ DNA switches (y = 0.1072x + 0.1622 and y = −0.0593x − 0.1351, respectively); the red line is the linear fitting of the differential sensing responses (y = 0.1657x + 0.3058). 8249
DOI: 10.1021/acs.analchem.9b00896 Anal. Chem. 2019, 91, 8244−8251
Article
Analytical Chemistry
enhanced (e.g., a wide range of sensing response could be achieved with more T-T mismatches added into the recognition motif; the charge transport conductivity of the initial constructs of DNA switches could be modulated for minimizing the background signal for either “signal-on” or “signal-off” sensors). Second, by simply replacing the recognition motif with other functional DNA sequences (e.g., aptamers), such mechanoelectronic DNA switch-based dualsignal sensors can be extended to many different types of analytes, such as protein biomarkers as explored in our earlier studies.18,20 Third, as shown in Figure 4B, the dual-signal responses originating from switching between enhanced and inhibited DNA conductivities, allow for a net differential sensing readout with further improved sensitivity. In principle, by making a 1:1 mixed monolayer of these two DNA switches labeled with different redox species, a “real” ratiometric sensing readout can be achieved with simultaneously enhanced and reduced redox responses.39,40 Last but not least, with independent dual-signal responses rather than the conventional synchronized ratiometric signals, a better selectivity can be expected for the mechanoelectronic DNA switch-based sensors, since it would be more challenging for nonspecific binding to simultaneously induce dual-signal responses. We are currently working on expanding the application of such divergent pairs of mechanoelectronic DNA switches for dualsignal sensing with other electrochemical approaches and analytes of interest.
■
CONCLUSION In summary, with Hg2+ binding as a trial system, it was confirmed the T-Hg2+-T formation induced enhanced and inhibited charge transport conductivities of a divergent pair of mechanoelectronic DNA switches with subtle differences at the junction and recognition motif via standard biochemical electrophoresis gel assays. More significantly, based on the biochemical survey results of the divergent mechanoelectronic DNA switch pair, we successfully constructed the “signal-on” and “signal-off” electrochemical biosensors with excellent sensing performances toward Hg2+, which shed light on the great potential of this new type of functional DNA-based dualresponse nanodevices and biosensors with satisfactory sensitivity, selectivity, and reproducibility.
Figure 5. SWV peak current of immobilized (A) 3′AA-1′ and (B) NB-1′ DNA switches during three cycles of alternating addition of Hg2+ (10 nM) and EDTA (5 μM).
signals of both remain stable during three cycles of alternating immersion in Hg2+ and EDTA solutions (variations
