Anomalous Trends in Nucleic Acid-Based Electrochemical Biosensors with Nanoporous Gold Electrodes

Abstract

Molecular diagnostics have significantly advanced the early detection of diseases, where electrochemical sensing of biomarkers has shown considerable promise. For a nucleic acid-based electrochemical sensor with signal-off behavior, the performance is evaluated by percent signal suppression (% ss), which indicates the change in current after hybridization. The % ss is generally due to more redox molecules (e.g., methylene blue) associating with the probe DNA bases in the single-strand form than the double-strand form upon hybridization with the target nucleic acid. Nanostructured electrodes generally enhance electrochemical sensor performance via several mechanisms, including increased number of capture probes per electrode volume and unique nanoscale transport phenomena. Here, we employ nanoporous gold (np-Au) as a model electrode material to study the influence of probe immobilization solution concentration on sensor performance and the underlying mechanisms. Unlike planar gold (pl-Au) electrodes, where % ss reaches a steady-state with increasing concentration of the grafting solution, the % ss displays peak performance at certain grafting solution concentrations followed by rapid deterioration and reversal of the % ss polarity, suggesting an unexpected case of increased charge transfer upon hybridization. Fluorometric assessments of electrochemically-desorbed nucleic acids for different electrode morphologies reveal significant amount of DNA molecules (unhybridized and hybridized) remain within the nanopores post-hybridization. Analysis of electrochemical signals (e.g., square wave voltammogram shape) suggests that the large unbound nucleic acid concentration may be altering the modes of methylene blue interaction with the nucleic acids and charge transfer to the electrode surfaces.

Graphical Abstract

The rapid detection and identification of specific microbial pathogens including bacteria, viruses and fungi is a grand challenge facing plant and human health. Early disease diagnosis is essential for an effective national healthcare infrastructure, where it is estimated that billions of molecular tests to detect disease-causing pathogens will be required per year globally¹. Examples of such human pathogens include Escherichia coli, Shigella, and Salmonella which cause foodborne diseases, and plant pathogens such as tobacco mosaic virus and cauliflower mosaic virus². Nucleic acid-based diagnostics is a powerful tool, not just for pathogen detection, but for cancer diagnosis and surveillance as well^3–6. Consequently, approaches to use nucleic acids as biomarkers are growing rapidly to replace or complement cell culture-based, biochemical and immunological assays^7,8. Various nucleic acid-based bio-sensors modalities have been developed for point-of-care detection, enabling sensitive and rapid analysis (optical, surface plasmon resonance, piezoelectric, etc.)^9,10. For example, quantitative polymerase chain reaction, achieves low limits of detection and high selectivity, yet requires sample preparation to extract target nucleic acids. Next-generation sequencing technology, while providing impressive sequence information without a priori knowledge of organisms, the necessary sample preparation and target application followed by bioinformatics interpretation, make it less desirable for applications requiring a fast diagnostic turn-around. Alternatively, electro-chemical nucleic acid-based sensors have attracted significant attention as powerful bioanalytical tools, as they can be multiplexed and easily interfaced with data acquisition electronics for portability^11–14.

The material and geometry that constitute the working electrode of an electrochemical nucleic acid-based sensor play a significant role in its performance^15–18. Electrode coatings can be broadly classified into smooth and nanostructured sensors based on the geometry of the working electrode. Smooth electrodes do not have any crevices or undulations. Common materials used for smooth electrodes include gold, platinum, carbon paste, glassy carbon, indium tin oxide and carbon micro-fibers¹⁴. One main limitation of smooth electrodes is that the probe molecules may not be displayed in favorable orientations for target hybridization¹⁹. Recently, nanostructuring of the electrode surface greatly mitigated these issues and enabled significant enhancement in sensor performance¹⁶. These enhancements include increased surface area-to-volume ratio and surface coverage of capture probes¹⁷, favorable orientation of the immobilized probes¹⁵, tunable dynamic range^15,20, and higher electro-catalytic activity at the surface¹⁹.

In order to develop electrochemical biosensors with rationally-designed nanostructures, it is imperative to have a fundamental understanding of the structure-property relationships that dictate sensor performance. It has been shown that nanoporous gold (np-Au) thin films can successfully be used as a model electrode system to investigate molecule-nanostructure interactions in electrochemical biosensors^17,21–24. Specifically, by modifying np-Au pore morphology (i.e., pore size, effective surface area) and employing electrochemical characterization, the effect of morphology on important nucleic acid-based sensor parameters, such as sensitivity and biofouling resilience, could be investigated. Np-Au, which can be easily produced, is emerging as a popular material for electrochemical sensors and recent demonstrations include detection of biomolecules and pharmaceuticals using this novel material^21–25. Np-Au is produced by a nanoscale self-organization process, where silver atoms are dissolved from a silver-rich gold alloy in nitric acid and gold atoms diffuse at the metal-electrolyte interface to create a bicontinuous open-pore structure²⁶. The np-Au electrodes facilitated three orders of magnitude improvement in detection limit compared to traditional planar counter-parts due to increased target penetration into the porous network and hence enhanced hybridization efficiency¹⁷. Additionally, by utilizing the inherent bio-sieving features of np-Au films, sensitive detection of short nucleic acids in biologically-relevant complex samples, such as serum, was achieved²⁰, circumventing the need for laborious sample preparation where most other nanostructured devices often require purified nucleic acids for analysis since the constituents of complex biological fluids (e.g., blood, plant lysate) adversely affect sensor performance through biofouling²⁷.

Unlike smooth electrode-based sensors, which can work at their optimal performance at a wide range of probe grafting densities^28–31, in order to benefit from the enhanced performance of nanostructured electrodes, the optimal working conditions need to be identified. To that end, we report a comprehensive study to expand fundamental understanding of how major stages of an electrochemical sensing process, such as molecule transport limitations and DNA probe grafting density on np-Au electrodes with varying morphologies, lead to unexpected trends in electrochemical nucleic acid detection. We envision the findings to be generalizable to other electrochemical sensing platforms employing nanostructured electrodes, thereby assist in rationale design of electrochemical sensors for point-of-care diagnostics and medical research.

EXPERIMENTAL SECTION

Chemical Reagents.

Electron Microscopy Sciences glass coverslips were used as substrates for film deposition (22 × 22 × 0.15 mm). Gold (Au), silver (Ag), and chrome (Cr) targets (99.95% pure) were obtained from Kurt J. Lesker. Nitric acid (70%), sulfuric acid (96%) and hydrogen peroxide (30%) were purchased from Sigma-Aldrich, USA. Piranha solution consisted of 1:4 ratio (by volume) of hydrogen peroxide and sulfuric acid. CAUTION: Piranha solution and nitric acid are highly corrosive and reactive with organic materials and must be handled with extreme care. Tris(2-chloroethyl)phosphate (TCEP, 0.5 M), magnesium chloride (MgCl₂, powder) and phosphate buffer (PB, 1 M) were obtained from Fisher Scientific. Phosphate buffered saline (PBS, 10×) was purchased from Life Technologies. Zeba™ Spin Desalting Columns were obtained from Thermo Fisher Scientific. Methylene blue (MB, powder) was obtained from Biosearch Technologies. 6-mercapto-1-hexanol, (MCH, 99%) was obtained from Sigma-Aldrich. The oligonucleotides used in this project consisted of 26 bases and were purchased from Integrated DNA Technologies, USA. The 5’ end of ssDNA probe was modified with a C6 linker and thiol group: 5ThioMC6-D/CGT GTT ATA AAA TGT AAT TTG GAA TT. Target DNA sequence matching to the probe DNA was: AAT TCC AAA TTA CAT TTT ATA ACA CG. Quant-iT™ OliGreen™ ssDNA and Quant-iT™ PicoGreen™ dsDNA Assay Kit were obtained from Thermo Fisher Scientific.

Fabrication of Nanoporous Gold (np-Au) Electrodes.

The np-Au gold films were prepared by sputter-deposition. Briefly, a stack of metal layers (160 nm-thick chrome adhesion layer, 80 nm-thick gold seed layer, 600 nm-thick Au_0.36Ag_0.64 (atomic %) alloy on piranha-cleaned glass coverslips followed by dealloying in nitric acid, as described previously³². To obtain np-Au electrodes with coarser pores (referred to as annealed np-Au), a group of dealloyed np-Au samples (referred to as unannealed np-Au) was thermally treated at 250 °C for 1.5 min on a hotplate³³. Planar gold electrodes (control morphology) were fabricated similarly. Fabrication details can be found in Supporting Information (SI).

Microscopic Characterization of Electrode Morphology.

To investigate micro- and nanoscale morphological features of the fabricated samples, high magnification images (100k× magnification) were taken using scanning electron microscopy (SEM; FEI Nova NanoSEM430). The top and cross-sectional images of varying electrode morphology samples were analyzed using ImageJ (National Institutes of Health shareware, http://rsb.info.nih.gov/ij/index.html) in order to determine the median pore size¹⁷.

Electrochemical Characterization of Electrode Morphology with H₂SO₄ and Methylene Blue.

All electrochemical experiments were performed with a Gamry Reference 600 potentiostat. Effective surface area of the metal electrodes was determined using electrochemical cyclic voltammetry (CV) measurements using previously established protocols that report on the charge required to strip the gold oxide layer via CV^21,34, when using 0.5 M sulfuric acid (Figure S2). For all the measurements, enhancement factor (EF) is defined as the ratio of value measured for any test morphology to that for pl-Au under the same testing conditions. To determine presence of possible transport limitations of MB redox reporters within the porous electrodes, a set of bare np-Au and pl-Au electrodes were exposed to 20 μM MB prior to assessment via CV. Similarly, MCH-coated electrodes (treatment with 1 mM MCH for 2 hr), were incubated in 20 μM MB and the total charge transfer was determined via CV and its subsequent analysis with the Gamry software package. See SI for details on electro-chemical characterization parameters and calculations.

Immobilization and Electrochemical Quantification of DNA Probes.

Electrodes were immobilized with incubation at varying probe solution concentrations using previously described protocols¹⁷. Briefly, piranha cleaned electrodes were incubated overnight (~15 hr) with desired probe solution concentration (0. 0.5, 1, 2, 4, 6 and 8 μM), containing reduced DNA probe, 25 mM PB and 50 mM μMgCl₂. To obtain a well-ordered DNA probe monolayer, electrodes were back-filled with MCH as described above. Finally, DNA probe-immobilized electrode was assembled in the electrochemical cell for electrochemical characterization (see SI and Figure S1a for details). To indirectly quantify the amount of DNA probes immobilized on the electrodes, 150 μL of 20 μM MB prepared in 1× PBS was added to the assembled electrochemical cell and incubated for 10 min. In high ionic strength incubation solution, negatively-charged phosphate backbone of ssDNA is screened, allowing MB to bind to guanine bases (G-bases) only^35,36. The electrodes were washed with PB after MB accumulation to remove unbound MB molecules. Finally, electrochemical MB footprint was acquired in 1× PBS using square wave voltammetry (SWV) measurements, and the total charge transfer, Q, under SWV curve was determined directly from Gamry Software and used as an indirect measure of amount of DNA probes immobilized on the electrode surface (baseline signal). Note that total charge transfer, Q, was used to electrochemically quantify nucleic acids presence instead of SWV peak current, I, because it captures electrochemical behavior more precisely for varying SWV curve shapes due to increasing probe solution concentrations (Figure S3). SWV was performed using a pulse size of 40 mV and frequency previously optimized for different electrode morphologies: 18 Hz for unannealed np-Au, 30 Hz for annealed np-Au, and 60 Hz for pl-Au¹⁷. After probe DNA baseline signal determination (Figure S1b), SWV was run until probe signal was depleted to at least 30% of the baseline signal, indicating MB dissociation from the DNA strand. One set of such electrodes was electrochemically cycled following MB depletion to desorb and elute the immobilized DNA for fluorometrically determining the amount of DNA probes, using a previously-established protocol ensuring efficient extraction of immobilized DNA from the pores^37,38. Remaining electrodes were used for target DNA detection and hybridization assays.

Electrochemical Extraction and Fluorometric Analysis of DNA Probe-Immobilized Electrodes.

To decouple the influence of electrochemical effects, which would obscure determination of actual immobilized probe amount, an independent measurement with fluorescent dyes was used on the electro-chemically-extracted DNA probes (details described in SI and Figure S4)^39,40. To fluorometrically quantify the amount of DNA probes, extracted probe DNA solution was analyzed using Quant-iT™ OliGreen™ ssDNA Kit. Briefly, OliGreen dye was added to the extracted samples, and after 5 minutes of incubation at room temperature, a NanoDrop™ Spectrophotometer was used for quantifying the amount of DNA present in the extracted samples.

Electrochemical Target DNA Detection.

The set of electrodes immobilized with various probe solution concentrations that were electrochemically quantified, were next challenged with 500 nM target DNA concentration to determine detection capabilities of the DNA sensor¹⁷. Each electrode was incubated with in PB containing the target DNA fragments and 50 mM MgCl₂ for 35 min at 37°C. The electrodes were then rinsed to remove nonspecifically bound target molecules and incubated with MB as done before for probe DNA characterization. Amount of target DNA captured by the probe recognition layer was determined by the decrease in total charge transfer under SVW curve (due to less MB binding to the hybrids as G-bases are not accessible) compared to that for the DNA probe baseline^35,36. The difference in total charge transfer between the probe baseline and target signal was used to quantify the hybridization efficiency in each case, expressed as percent signal suppression (% ss), and is classified as a ‘signal-off electrochemical detection mode (Figure S1b)^17,41.

Fluorometric Analysis of Hybridized Samples.

Similar to the sample preparation for fluorometric analysis of probe DNA, the electrodes were exposed to a previously established CV treatment following the hybridization step to extract any nucleic acids (hybrids or ssDNA) from the electrodes (Figure S4). The aforementioned fluorometric analysis was employed with Quant-iT™ OliGreen™ ssDNA Kit Quant-iT™ PicoGreen™ dsDNA Assay Kit to determine amount of ssand dsDNA in the extracted samples.

RESULTS AND DISCUSSION

Here, we discuss electrochemical detection mechanisms studied via (i) electrochemical response as a function of electrode morphology and critical sensing components, such as reporter MB and backfill MCH; (ii) electrochemically- and fluorometrically-determined amounts of DNA probes immobilized on the electrode surfaces; (iii) effect of various probe grafting densities on the sensor detection performance; (iv) analysis of MB electrochemical data to provide insight into MB-nucleic acid interactions within the nanostructured coatings. Unless indicated otherwise, all experiments involved at least three different samples, where the data points are the averages and the error bars are the standard deviations of the measured values respectively.

Morphological Characterization.

Np-Au electrode pore morphology is one of the main characteristics that affect both, electrochemical^17,20 and transport events^42,43 for nanostructured sensors. To understand the effect of pore morphology on electrochemical sensing phenomena at nano-level, a group of np-Au samples (unannealed np-Au) was thermally annealed, as routinely performed⁴⁴. The median pore radii of unannealed and annealed np-Au electrodes were 24 nm and 56 nm respectively, determined by analysis of SEM images (Figure 1a and Figure 1b). Interconnected non-columnar pores and resulting ‘funnel-like’ constrictions are evident in the cross-section images of the two morphologies (Figure 1c and Figure 1d).

Figure 1.

Scanning electron micrographs of top-view (a) unannealed np-Au film; (b) annealed np-Au film (obtained via thermal-annealing process at 250 °C); and cross-sectional views of (c) unannealed np-Au film; (d) annealed np-Au film.

Nanostructure Transport Limitations.

As the electrochemical sensor performance is a function of nucleic acids, MB redox reporters, and MCH backfill, we systematically studied the influence of each on the electrochemical signal. The electrochemical surface area characterization with dilute sulfuric acid, which is not prone to hindered diffusion within the pores) confirm previous findings¹⁷ and reveal that the unannealed np-Au has higher EF (surface area) as compared to annealed np-Au (EF = 11.9, 7.1, 1.0; unannealed, annealed, planar Au respectively), as shown in Table 1a and Figure S2a. Next, electrodes were exposed to 20 μM MB solution to test whether MB is subject to transport limitations within the nanostructures. The EF values (total charge transfer) for MB reduction in the porous electrodes (Table 1b and Figure S2b) were at least as high as those obtained with the electrochemical H₂SO₄ measurements for electrochemically-accessible surface area, suggesting that MB does not experience transport-limitations within np-Au pores. Interestingly, the EF values for MB were 14.3 and 9.9 for unannealed and annealed np-Au respectively, compared to 11.9 and 7.1 when characterized with H₂SO₄. This discrepancy is attributed to the hindered transport of MB molecules within the pores ensuring a higher percentage of redox molecules to be reduced before they escape into the bulk solution (as it would be in the case of planar gold) - an electrochemical phenomena recently observed in np-Au for different redox molecule species⁴⁵.

Table 1.

Comparison of EF for varying electrode morphologies (unannealed np-Au, annealed np-Au, pl-Au), obtained using the charge under the CV curve generated by reducing a) H₂SO₄ on bare electrode (converted to electrode surface area); b) MB on bare electrode: c) MB on MCH coated electrode.

While MB alone does not experience significant transport limitations, the introduction of the MCH passivation layer (necessary for favorable a DNA probe monolayer^46–48) may introduce additional transport limitations. Indeed, the EFs obtained by determining total charge transfer from MB for the different electrode morphologies were less than the values obtained for MB without MCH back-fill (Table 1c and Figure S2c). This suggests that MCH may be reducing effective pore sizes (especially at the funnel-like constrictions) and limiting MB transport to deeper surfaces in the porous electrodes; thereby, reducing the surface area that contributes to electro-chemical sensing. It is obvious to expect that nucleic acids will also be prone to such hindered transport during both DNA probe immobilization and target detections steps, which is discussed next.

Trends in Probe DNA Immobilization Density.

One of the main determinants of an electrochemical sensor performance is formation of DNA probe monolayer^15,49,50 with favorable conformation to allow hybridization with a target molecule. For this reason, before testing the target DNA detection performance of the sensor, the effect of probe immobilization solution concentration (0 – 8 μM) on the extent of DNA probe immobilization within the nanostructured electrodes was investigated by charge transfer determination via SWV (Figure 2a), as described under Experimental Section. Unlike the pl-Au (control), which reaches its maximum probe grafting density at 1 μM probe solution concentration, both np-Au morphologies required higher probe immobilization solution concentrations to achieve their maximum grafting density. This suggests that the probe DNA grafting is mass-transport limited, where a larger concentration gradient ensures DNA in-flux despite the crowding within the pores particularly at the funnel-like constrictions, as was suggested by the MCH and MB results. The probe grafting trends for the two np-Au morphologies indicate that there is a necessary probe concentration to achieve immobilization in nanoporous structures, as exhibited in similar probe currents in low probe immobilization concentration range 0.5 – 2 μM for the two nanostructured morphologies. In this low probe concentration regime, annealed np-Au pore surfaces are readily accessible by probes due to larger pore size, while DNA probes within the unannealed np-Au electrode experience diffusion limitations due to the channel constrictions within the porous electrodes. For this reason, two morphologies appear to have similar amounts of probes immobilized in low probe solution concentration conditions, that is, plausibly only the top portion of unannealed np-Au electrode is accessed. Annealed np-Au reaches its maximum grafting density around 2 μM probe solution concentration, while unannealed np-Au needs much higher probe solution concentration to reach its maximum grafting density, seen by the total charge transfer constantly increasing.

Figure 2.

a) Total charge transfer under SWV, Q, curve obtained for different electrode morphologies (unannealed np-Au, annealed np-Au, pl-Au) at varying probe solution concentrations (0, 0.5, 1, 2, 4, 6, 8 μM); b) OliGreen fluorescence measurements to quantify amount of immobilized DNA probes at varying probe solution concentrations (0, 0.5, 1, 2, 4, 6, 8 μM) and extracted from the electrode via CV desorption/elution method; c) Combined electrochemical results from Figure 2a and fluorescence measurements from Figure 2b to provide insight into immobilization trends for different morphologies at varying probe solution concentrations.

Since the total charge transfer is only an indirect measure of how much DNA probe is present in porous electrodes, fluorescence-based nucleic acid measurements were performed on the immobilized DNA probes to more accurately determine the amount of DNA upon immobilization on the electrodes. Once the DNA within the electrodes was electrochemically desorbed and eluted, the eluate containing the DNA probes was fluorometrically analyzed with OliGreen (Figure 2b). In agreement with the electrochemical results, the OliGreen quantification for unannealed np-Au showed a constant increase for the amount of DNA probes in the porous electrodes with increasing probe solution concentration with a similar inflexion point at 2 μM probe solution concentration. This consensus between the two independent methods (i.e., electro-chemical and fluorometric) rules out the possibility that the anomalous trend observed for the electrochemical assessment is a transport-driven effect. Instead, it appears to be an electro-chemical signal enhancement artefact due to phenomena such as redox amplification in nanoconfinement^45,51,52.

To better illustrate the DNA transport phenomenon during probe immobilization, it is instructive to plot the total charge transfer (Figure 2a) with respect to fluorometric (Figure 2b) assessments of ssDNA amounts in the electrodes. The composite plot, Figure 2c, again highlights that nanoporous electrodes need a higher probe solution concentration to immobilize a larger portion of the available surfaces (as determined by dilute sulfuric acid quantification in Table 1). The composite plot (individual morphologies displayed separately in Figure S5) also reveals that while there is a linear increase in the amount of immobilized probe with increasing probe solution for pl-Au and annealed np-Au, a secondary trend emerges for unannealed np-Au. The correlation coefficients (corresponding to the two observed regimes) between the two measurement modalities for unannealed np-Au are 0.90 (0 to 2 μM region) and 0.94 (2 to 8 μM region). On the other hand, the correlation factor for annealed np-Au and pl-Au are 0.95 and 0.90 respectively. The non-linear trend observed for unannealed np-Au again points to the increasing dependence on a larger concentration gradient between the incubation solution and the pores to successfully immobilize the entire surface of the porous electrode with many channel constrictions (Figure 1d) due to small median pore size.

Effect of Probe Grafting Density on Hybridization.

Once probe DNA had been both electrochemically and fluorometrically characterized, the electrodes were exposed to 500 nM target solution (concentration that is within the dynamic range of detection for all three electrode morphologies^17,42). Successful target detection is identified by the SWV signal drop, measured as percent signal suppression (% ss) defined as (Q_probe − Q_target)/Q_probe) × 100, where the total charge transfer values are shown in Figure S6. A positive % ss generally indicates successful hybridization, while zero % ss indicates no DNA target capture, both of which are normal sensor behaviors. A negative % ss, on the other hand, would suggest an anomalous sensor behavior. To that end, unlike the smooth pl-Au electrode, where once the maximum probe grafting density is reached, the sensors display constant performance (% ss reaches plateau); in the case of nanostructured electrodes, sensor reaches peak performance at distinct probe concentrations, after which the performance starts deteriorating, marked by the reversed % ss (Figure 3a). Specifically, annealed np-Au exhibits reversed % ss behavior in the range of 2 – 8 μM probe immobilization solution concentration, but it reaches its plateau as soon as it enters the reversed % ss regime at 2 μM probe solution concentration. On the other hand, unannealed np-Au, begins to exhibit reversed % ss at 4 μM probe solution concentration, but it plateaus only at 6 μM. The reversed % ss suggests a higher total charge transfer, which is an unexpected result, as the electrochemical signal should either remain the same as probe baseline signal (since no additional probes have been immobilized) or decrease (since hybridization reduces the available binding sites for MB) binding, both of which are expected trends with pl-Au electrodes. Fluorometric investigation of DNA molecules is performed next to illuminate possible mechanisms of this anomalous sensor response. After electrochemical post-hybridization characterization, nucleic acids were electrochemically extracted and analyzed using fluorescent dyes (OliGreen and PicoGreen). It is important to note that fluorescent dye calibration studies have shown that Oli-Green does not only detect ssDNA, but it is sensitive to all DNA in the sample, while PicoGreen primarily labels dsDNA. To prevent ambiguous results, we conducted calibration experiments using combinations of varying ssDNA and dsDNA concentrations that are within the range of DNA amounts quantified in this study. Fluorometric measurements with Oli-Green (total DNA amount) and PicoGreen fluorescent dye (double-stranded DNA) reveal significant amount of unbound DNA molecules within the nanoporous electrodes following the hybridization step (Figure 3b). The results from fluorometric assessment agree with the electrochemical trends (Figure 3a), where with a large amount of residual nucleic acids within the electrodes, the sensor performance deteriorates to the extent of producing reversed % ss at high probe solution concentrations. In addition, the trend for average cleaving current magnitudes obtained from CV desorption curves remains constant across each morphology independent of the probe solution concentration (Figure S7). Since the cleaving current magnitudes are attributed to thiol-bond reduction⁵³, this indicates that accumulation of DNA observed using fluorometric and electrochemical readings may be attributed to the unbound DNA found in the pores.

Figure 3.

a) Percent signal suppression (% ss) for grafting at different probe solution concentrations (0, 0.5, 1, 2, 4, 6, 8 μM) and for three electrode morphologies (unannealed np-Au, annealed np-Au, pl-Au), defined as percent decrease in total charge transfer after hybridization. A positive and zero % ss generally indicate successful hybridization or no target capture respectively, while a negative % ss suggests anomalous sensor behavior. b) Concentration of unbound ssDNA in the porous electrodes after hybridization, determined by subtracting PicoGreen dsDNA fluorescence measurement from OliGreen total DNA measurements.

Effect of Probe Solution Concentration on Interactions between MB and DNA.

The SWV signal after hybridization can be higher than the baseline probe signal only if there are more MB molecules present at the electrodes (hence higher total charge transfer) compared to the pre-hybridization baseline condition. Figure 3b shows that an increase in the amount of residual DNA within porous electrodes (with increasing probe solution concentration) follows the inverse trend observed for % ss in Figure 3a for both morphologies of np-Au, albeit more pronounced for unannealed np-Au. Therefore, it is plausible that reversed % ss phenomenon in np-Au electrodes is caused by high concentration of residual nucleic acids in the pores. As described previously, MB-based nucleic acid detection mechanism is based on the principle that positively-charged MB binds to G-bases only at high ionic strength solution (e.g., 20 mM NaCl^35,54), where the negatively-charged phosphate backbone is screened with the sodium counter-ions (Scheme 1a). However, it has also been shown that depending on the nucleic acid concentration and ionic strength within the electrode pores, MB can undergo multiple modes of binding to the DNA hybrid⁵⁵, so that it does not solely show affinity to the G-bases (Scheme 1b). For example, it can also symmetrically/asymmetrically intercalate^56,57, semi-intercalate^58,59 and bind to the minor/major groove^60–63. In addition, it has been reported that MB can change its electrochemical properties, such as reduction voltage^54,64 and extent of reduction^65–67, depending on the ionic strength of the solution, electrode surface conditions and its distance from the electrode. Observed widening and tilting of the SWV baseline signal for the unannealed np-Au electrode (Figure 4a), suggests that MB might be undergoing redox reactions at various conditions at high probe solution concentrations (2 – 8 μM). Kelley et al. also suggested that broadening and widening of electrochemical peaks is due to varying reduction potentials due to spatial distribution of MB along the DNA strand⁶⁸ and within the pore volume. Recall that subtle electrochemical signal amplification (plausibly due to nanoconfinement effects) has been observed earlier (Table 1b) when the electrodes were exposed to MB solution only. The unannealed np-Au pores are in the 10s of nanometer-radius range, in which there could potentially be nanoconfinement effects causing change in the local pore environment^69–73. These include change in ionic strength, especially when a pore is crowded with unbound nucleic acids, which would decrease the effective pore size for MB inclusion. In addition, asymmetry in and shifting of the SWV peaks in the case of unannealed np-Au indicate that some MB might be generating an electrochemical signal through physical proximity to the electrode surface by binding to the unbound DNA in the pores as well (Scheme 1b). The same MB concentration (20 μM) used for all experiments, reinforced by the agreement between fluorometric quantification of unbound DNA (Figure 2b and 3b) and electrochemical response (Figure 2a and 3a) indicating a higher unbound DNA amount in unannealed np-Au with increasing grafting solution concentration, suggests that the anomalous SWV shapes are mainly due to the unbound DNA (and its influence on MB-DNA interactions and MB electrochemical properties) rather than unbound residual MB in the pores.

Scheme 1.

a) At low probe solution concentrations, DNA capture probes are sparsely immobilized allowing for target DNA to more efficiently access and hybridize with the probes through the more common G-base-dependent detection mode. b) At high probe solution concentrations, the crowded pore environment with residual DNA molecules alter the modes of DNA-MB interactions, resulting in MB association with both ssDNA and dsDNA, thereby causing an unexpected increase in the electrochemical signal upon hybridization. Figure S8 provides an extended qualitative description of various modes of probe DNA immobilization and hybridization, as well as related electrochemical events for different electrode morphologies.

Figure 4.

SWV probe current for varying probe solution concentration (0, 0.5, 1, 2, 4, 6, 8 μM) for a) unannealed np-Au, b) annealed np-Au and c) pl-Au. SWV probe current plots illustrate the peak shift for the unannealed np-Au, where due to smaller pore size (triggeringnanoconfinement effects), the MB molecules likely undergo additional electrochemical reactions at high nucleic acid concentrations in contrast to annealed np-Au and pl-Au, for which these anomalous behavior is not observed.

Collectively, the complex pore local environment (crowded with residual DNA molecules, especially for unannealed np-Au) and the associated alterations in MB-DNA binding modes may be contributing to the reversed % ss, observed during high probe solution concentrations (Figure S8). It is important to mention that we have previously demonstrated that MCH incubation durations greatly affect the formation of DNA probe monolayer³³. That study revealed that higher probe grafting densities require longer MCH incubation durations for proper monolayer formation and target detection, making this yet another factor that should be taken into consideration when optimizing electrochemical sensors with nanostructured electrodes. This is particularly important for electrodes with high aspect-ratio cavities and channel constrictions (e.g., funnel-like structures) that are prone to transport limitations, where longer MCH incubation durations may ensure proper mono-layer formation at the deeper pore surfaces. Finally, extending the rinse durations for nanostructured electrodes with high residual nucleic acid content may reduce molecular crowding inside the pores and improve sensor performance. However, the extended rinse durations may not always be feasible, particularly in the context of point-of-care sensors.

CONCLUSIONS

The goal of this study was to provide insight into transport and electrochemical detection events observed in nanoporous gold (np-Au) electrodes, which were used as a model nanostructured electrode system. The total charge transfer as a measure of immobilized probe DNA (obtained from baseline SWV curve) with respect to the OliGreen quantification of detached/unbound probe DNA showed that while there is a linear increase in probe immobilization with increasing probe solution concentration for pl-Au and annealed np-Au, a secondary trend emerges for unannealed np-Au. It is plausible that DNA transport into the pores is hindered by “funnel-like” constrictions; hence, a larger concentration gradient is necessary to efficiently immobilize the deeper pores of unannealed np-Au and more unbound probe accumulated within the pores for unannealed np-Au. Unlike pl-Au electrode, where once the maximum probe grafting density was reached, the sensor displayed constant performance (steady % ss); in the case of nanostructured electrodes, the sensor reached optimal performance at specific probe concentrations, after which the performance (% ss) rapidly deteriorated. The surprising reversed % ss suggests that there is more total charge transfer compared to the baseline probe signal. This may be an effect of altered MB interactions with DNA molecules within the pores crowded with residual DNA. Fluorometric measurements revealed that, indeed, a significant amount of ssDNA was present in the porous electrodes following hybridization, supporting the possibility of mechanisms discussed for the reversal of % ss polarity. We demonstrated that nanostructured electrochemical sensors exhibit anomalous sensing phenomena compared to smooth electrode-based sensors, particularly for samples with high DNA concentration. By highlighting the importance of determining optimal working conditions, such as probe solution concentration, this study is expected to assist in the design and implementation of other nanostructure-based electrochemical sensors, particularly ones that rely on a signal-off detection mode.