Nucleic acid analysis has played an important role in the detection of pathogens and genetic diseases. In recent years, its usefulness has been seen in many decentralized applications such as point-of-care diagnostics, environmental and food monitoring, and the detection of biological warfare agents. Among the available analytical techniques for DNA analyses, real-time polymerase chain reaction (PCR) developed for the simultaneous amplification and detection of deoxyribonucleic acid (DNA) has been a key technology for high-speed testing and accurate quantification.¹ Various assay formats, utilizing fluorescence-linked reporters such as SYBR green 1, hydrolysis probe, and hybridization probes, have been reported. Despite the wide acceptance, their use is largely limited to clinical and research laboratory apparatuses. The difficulty in advancing this technology for point-of-care testing (POCT) applications lies in the requirement for bulky and complex optical systems for the DNA amplicon detection. Toward the goal of performing a complete DNA analysis in a handheld instrument, optical-based detection schemes will need to be replaced by an alternative detection method that is compatible with the concept of POCT. One of the possible choices could be the electrochemistry-based route.

Over the past years, numerous studies have been carried out on electrochemical DNA sensors,^2-5 with some research groups focusing on the detection of PCR amplicons.^6,7 Efforts have also been made to develop integrated DNA microchips with electrochemical detectors for PCR product identification.^8-10 However, these post-PCR hybridization-based platforms suffer from a long assay time and a narrow dynamic range as compared to the fluorescence-based real-time PCR methods. To achieve an electrochemical real-time PCR (ERT-PCR), the detection scheme must fulfill the basic requirement of thermal stability as well as negligible PCR inhibition. In a previous work, our group demonstrated, for the first time, that the extension of solution-phase and electrode-bound oligonucleotide probe with some of the deoxythymine triphosphate (dTTP) substituted with ferrocene-labeled deoxyuridine triphosphate (Fc-dUTP) can be utilized to monitor the DNA amplification process in real-time.¹¹ Briefly, an oligonucleotide probe specific to the PCR amplicon was immobilized onto an indium tin oxide (ITO) electrode, which was then dipped into the PCR solution in a conventional PCR tube format. At the annealing step, the heat-denatured single-stranded amplicon containing some Fc-dUTP moieties hybridizes to the capture probe on the ITO electrode. Then, the electrode-bound probe was extended by the polymerase (usually termed as solid-phase PCR^12,13), producing a progressive accumulation of the redox marker onto the electrode surface in response to the target amplicon generation. For these proof-of-concept experiments, a series of samples were run in parallel and the electrodes were taken out from the PCR tube at the desired cycle number followed by water rinsing. The incorporated ferrocene signal was then electrochemically measured.

Although the electrochemistry-based real-time PCR technique has shown to be a very sensitive technique,¹¹ with an onset PCR cycle down to ~3, its starting target DNA analyte concentration nevertheless is required to be higher than that of the fluorescence-based counterparts. In this study, two important issues that would greatly affect the performance of this on-chip electrochemistry-based technique were studied. These were electrode surface passivation and electrochemical scanning during the PCR. Our approach to improving the ratio of electrochemical signal-to-background noise is discussed. Performance comparison of the on-chip ERT-PCR versus its fluorescence counterpart, using M13mp18 DNA template as a model analyte, is reported. The ability to conduct real-time PCR in a microchip is a crucial step in realizing a portable device for POCT applications.

Materials and Methods

Reagents and Instrumentation. All general chemicals were obtained from Sigma-Aldrich (St. Louis, MO), whereas PCR reagents were obtained from Invitrogen (Carlsbad, CA), unless otherwise stated. Note that Fc-dUTP was synthesized according to Wlassoff and King's protocol,¹⁴ with a minor modification in the purification step where the Fc-dUTP filtrate in DMSO was evaporated to dryness in a vacufuge concentrator (Eppendorf vacufuge concentrator 5301) without using an HPLC DEAE-cellulose column. The dried filtrate was then dissolved in deionized water, followed by a final purification step with a 0.45 m polypropylene membrane filter (Millipore). The thermal control system for the PCR consisted of a data acquisition card (PCI-MIO-16E-1, National Instruments, Austin, TX) along with a signal-conditioning board (SC-2042-RTD, National Instruments) connected to the temperature sensors. A digital feedback proportional-integral-derivative (PID) control algorithm was implemented in LabVIEW software (National Instruments) to control voltage supply to the heater by a power source (HP6629A, Hewlett-Packard, Rockville, MD). Temperature sensors and heaters were connected to the corresponding instruments via electrical contact pins.⁹ Electrochemical measurements were performed with an Autolab PGSTAT30 potentiostat/galvanostat (Eco Chemie, The Netherlands) controlled by the General Purpose Electrochemical System (GPES) software (Eco Chemie).

Silicon-Glass Microchip Fabrication. The integrated silicon-glass microchip consisted of Si (thickness of 400 m) and ITO-coated glass substrates (Delta Technologies, Stillwater, MN) with metallic patterns and microstructures (top half of Figure 1A). Thin-film Pt heaters and temperature sensors (thickness of 100 nm) were patterned on the front side of the Si to control thermal cycling of the PCR. On the other side of the Si, a micromachined chamber (length and width of 5 mm, depth of 325 m), formed by the inductively coupled plasma deep reactive ion etching (ICP-DRIE) process, was used for the solution-phase PCR. ICP-DRIE etched feed holes (diameter of 500 m, depth of 100 m) were used for the injection and removal of PCR solution. The ITO-coated glass chip had thin-film Pt pseudoreference and counter electrodes (thickness of 100 nm) coated on the center with four ITO-based (thickness of 100 nm) circular working electrodes on the surrounding for the immobilization and solid-phase extension of oligonucleotide capture probe. Ultraviolet (UV) curing optical cement (type UV-69, Summers Optical, Hatfield, PA) was used to seal the silicon and glass chips, and the curing procedure was performed according to the manufacturer's instruction. A side view schematic of the bonded microchip with features of heaters and sensor, reaction microchamber, and working electrode is shown on the bottom half of Figure 1A. A similar DNA microchip was previously reported by our group for the multiplexed detection of Escherichia coli and Bacillus subtilis.⁹

Figure 1 Schematic diagrams of the microchip showing (A) the layout of the silicon and glass substrates (top half) and the side view (bottom half), (B) the initial status of the ERT-PCR microsystem containing target analyte and Fc-dUTP on the solution phase and electrode-bound oligonucleotide capture probe and a blocking layer on the ITO surface, and (C) a snapshot of the ERT-PCR microchamber at the end phase of an ERT-PCR cycle where the incorporation of Fc-dUTP moieties on the hybridized PCR amplicon and the extended capture probe linked on the electrode amplicon can be seen.

Probe Immobilization and Electrode Passivation. Prior to the UV bonding, the patterned glass substrates were sequentially sonicated in an Alconox solution (8 g of Alconox/L of water), propan-2-ol, and twice in water. Each sonication lasted for 15 min. They were then dried with nitrogen gas and treated in an air plasma for 10 min (power setting ~7 W, operating pressure ~2 × 10^-4 bar) using a Harrick plasma cleaner (Harrick Plasma, Ithaca NY). The hydrolyzed glass substrates were then immersed in a 10% (3-glycidoxypropyl)trimethoxysilane (dissolved with 95% ethanol) for 1 h. The silanized substrates were dried at 50 C under vacuum for 3 h and bonded with the silicon substrate. After the UV bonding, a 1 M oligonucleotide probe solution (sequence: 5'-NH₂-TTT TTT TTT TTT TTT TTT TTA AGG AAA CAG CTA TGA C-3') in phosphate-buffered saline (PBS, 100 mM NaCl/10 mM sodium phosphate, pH 7.0) was introduced into the microchamber and incubated overnight. Excess probe was washed off with PBS. Residual epoxide groups were blocked with ethanolamine for 12 h, unless otherwise stated. Finally, the microchamber was flushed thoroughly with autoclaved double-deionized water and dried with nitrogen gas.

Electrochemical Real-Time PCR Protocol. The PCR master mix contained 1× ThermoPol reaction buffer (20 mM Tris-HCl, 10 mM (NH₄)₂SO₄, 10 mM KCl, 2 mM MgSO₄, 0.1% Triton X-100, pH 8.8), 0.2 mM dNTPs (with 0.06 mM dTTP substituted by Fc-dUTP), 0.2 M forward primer (5'-GTA AAA CGA CGG CCA G-3'), 0.2 M reverse primer (5'-AAG GAA ACA GCT ATG AC-3'), 0.04 ng/L M13mp18 template, 0.5 g/L bovine serum albumin, and 0.04 units/L Vent_R (exo-) DNA polymerase (New England BioLabs, Ipswich, MA). The master mix was pipetted into the microchamber, and the injection holes were sealed with Bostik's Blu-Tack. The chip was subjected to the following thermal cycling profile: initial denaturation at 94 C for 2 min; 30 cycles at 94 C for 20 s, at 55 C for 20 s, 72 C for 10 s. It should be noted that, apart from on-chip thermal control by the patterned heater and temperature sensors, the PCR thermal cycling of the microchip could also be performed by the conventional cycler (Eppendorf Mastercycler personal) with very similar results (data not shown). Differential pulse voltammetric measurements were performed with a pulse amplitude of 100 mV and a scan rate of 25 mV/s.

Fluorescent Real-Time PCR Protocol. SYBR green fluorescence-based assay was conducted to provide a benchmark comparison with the on-chip ERT-PCR. The real-time PCR kit was purchased from Invitrogen (SYBR GreenER qPCR Super Mix, catalog no. 11760-100), and the SYBR assay was performed using an Applied Biosystems 7300 real-time PCR system. The reaction condition was set according to the vendor's recommended protocol.

Results and Discussion

A schematic of the ERT-PCR system in a silicon-glass microchip is shown in Figure 1, parts B and C. One of the key elements to the implementation of on-chip ERT-PCR lies in successfully reading the accumulated electrochemical signal resulting from the increased PCR amplicon amount at each PCR thermal cycle. Unlike the previous proof-of-concept ERT-PCR demonstrated in the Eppendorf tube PCR process,¹¹ the on-chip ERT-PCR process involves oligonucleotide extension of solution and solid phases in a closed microenvironment. It requires repetitive electrochemical potential scanning without the removal of solution and impurities adsorbed on the surface. Several crucial factors (e.g., passivation of the sensing electrode, strategy on the electrochemical scanning, and control of enzyme and oligo probe concentrations) that would greatly affect the analytical signal of this new assay platform are presented in the following paragraphs.

Electrode Surface Passivation. One of the biggest challenges in implementing the ERT-PCR in a microchip format is to minimize, if not totally eliminate, the background noise caused by the nonspecific adsorption of solution-phase Fc-dUTP. Figure 2 shows that the background signal is strongly dependent on the duration of the ethanolamine blocking treatment. Characteristic redox peaks from Fc-dUTP (~+0.57 V vs Pt pseudoreference electrode) are noticeable with a blocking time shorter than 3 h, which was likely caused by the nonspecific adsorption of the Fc-dUTP to the unreacted epoxide groups on the ITO surface. For a long enough blocking time (i.e., 12 h), the baseline is nearly flat. This implies that almost all the residual epoxide groups had reacted or were blocked with the ethanolamine and nonspecific adsorption of the Fc-dUTP was minimized. The signal-to-noise ratio can then be increased, and hence, a lower detection limit is expected.

Figure 2 Differential pulse voltammetric scans of the ITO electrodes in the absence of the template. The chips were incubated with ethanolamine for different times: (a) 1, (b) 2, (c) 3, and (d) 12 h. Electrochemical measurements were carried out using a pulse amplitude of 100 mV/s and a scan rate of 25 mV/s. The background signals were obtained at the end of 30 cycle PCRs.

Electrode Scanning Strategy. Different from the assay platform of a fluorescence-based real-time PCR, the ERT-PCR technique requires electrochemical scanning at intermittent intervals during the thermal cycling process in order to electrochemically monitor the amplified PCR products in a real-time setting. There are four circular ITO-based electrodes patterned in our silicon-glass microchip (Figure 1A). Ideally, it would be desirable to have each of the four electrodes scanned for many times (e.g., once for every PCR cycle and 30 times in a 30 cycle PCR) without affecting the PCR performance. However, realistically the adsorption characteristics of charged species (e.g., dNTPs, Mg²⁺) and enzyme polymerase could change when exposed to the repetitive potential scanning environment. This subtle change could affect the process of nucleotide extension on the solid electrode surface.

To investigate the effect of electrochemical scanning, the ERT-PCR was performed in our microchips with all the ITO electrodes immobilized with the same capture oligonucleotide probes. In one set of the microchips, all the four ITO electrodes were electrochemically scanned for multiple times in every five cycles, whereas in the other chips, the four ITO electrodes were selectively and singly scanned at the thermal cycle of 0th, 10th, 20th, and 30th, respectively. It can be seen from Figure 3 that a cycle-by-cycle increase of the electrochemical signal was obtained, indicating a successful incorporation of the Fc-dUTP redox marker on the extended probe. A low background noise, for the amplification case without DNA template, suggests an effective minimization of the unspecific electrochemical interference. The leveling-off gain of the electrochemical signal at high cycle numbers (data points shown as squares) on the multiply scanned electrode is a good indication that the multiple electrochemical scanning has indeed an adverse effect on the process when compared to the signal measured on the single-scanning electrode (data points shown as circles). The working mechanism of the potential scanning is not yet fully understood, and the investigation of this mechanism is ongoing. Since our experiments suggested that the effect of multiple scanning can be reduced by adding more polymerase (data not shown), it is plausible that irreversible adsorption of enzyme and other species on the electrode could play a role in this process. Moreover, it is possible to avoid the issue of multiple scanning by employing more ITO-based working electrodes in our microchip. In this case, each of the ITO electrodes will only be scanned once for the electrochemical measurement at a specific thermal cycle. The linear relationship between the signal versus cycle number in a single-scanning measurement (Figure 3) reflects this strategy. Therefore, experimental data presented in the next session (calibration plot) were all done based on a single-scanning electrode. Actually, similar plots of signal saturation at high cycle numbers were also observed with fluorescence-based real-time PCR measurements. However, the reasons for the signal saturation for these two scenarios are very different in nature. In the fluorescence-based case, the phenomena were attributed to the depletion of limiting reagents, whereas in ERT-PCR, the electrochemical scanning was the main reason. A benchmark comparison of the ERT-PCR versus conventional fluorescence-based counterpart technique will be discussed in the following paragraphs.

Figure 3 ERT-PCR plot of peak current signal in differential pulse voltammetric scans vs PCR cycle number with different electrode scanning strategies: · single scan in the presence of template (3 × 106 copies/L); multiple scan in the presence of template (3 × 106 copies/L); single scan in the absence of template.

Calibration Plot. For any real-time PCR technique, it is important to evaluate its performance in the quantification of target DNA molecules. In ERT-PCR, the standard curve was obtained by setting the threshold value at 0.1 nA for the current-cycle number plots with different initial template concentrations. Unlike a linear calibration plot exhibited by the fluorescence-based real-time PCR (inset of Figure 4B), the electrochemical one can be approximated by two linear regimes (inset of Figure 4A), with a crossover point at ~10⁵ copies/L. As shown in Figure 4, with an initial template concentration of 3 × 10⁶ copies/L, the onset cycle number, at which the analytical signal in ERT-PCR (~3-5 cycles) can be distinguishable from the background signal, is much smaller than that for the fluorescence-based real-time PCR counterparts (around 18-20 cycles). This suggests that fewer PCR cycles are needed for the ERT-PCR technique. It is interesting to note that the electrochemical real-time PCR methodology has a relatively better performance (in terms of the threshold cycle number) than the SYBR green fluorescence-based real-time PCR at high template concentrations (>10⁵ copies/L). For template concentrations lower than 10³ copies/L, the threshold cycle number for the electrochemical scheme exceeds that of the fluorescence scheme. One possible explanation of this phenomenon is the competition between the solution-phase primer and the immobilized probe for the target amplicon. At low target copy number, the solution-phase primer dominates in the annealing step. Hence, it is necessary to build up sufficient amount of PCR products to facilitate solid-phase probe extension, leading to the need for a large threshold cycle number for a low initial DNA content.

Figure 4 (A) ERT-PCR plot of peak current signal vs PCR cycle number for a series of dilutions from 102 to 106 DNA copies/L (* 3 × 102, × 3 × 103, 3 × 104, 3 × 105, 3 × 106). Inset: standard curve with a threshold set at 0.1 nA. (B) SYBR green assay plot of fluorescent signal vs PCR cycle number for a series of dilutions from 102 to 106 DNA copies/L (a, 3 × 106; b, 3 × 105; c, 3 × 104; d, 3 × 103; e, 3 × 102; f, negative control). Inset: standard curve of the corresponding SYBR green assay.

Effects of Enzyme and Probe Concentrations on the Peak Current Signals. As mentioned in the previous paragraph, a high PCR threshold cycle number is required for samples with low template concentrations. In the cases with the fluorescence-based schemes, little can be done to increase the signal-to-background ratio given that background fluorescence cannot be easily eliminated. On the other hand, the enzyme and probe concentrations can be used to enhance the sensitivity of this ERT-PCR technique. Figure 5 illustrates that the analytical signals are greatly increased without increase in background signal (data not shown) with increasing enzyme concentrations. When the enzyme concentration is increased by 8 times, there is an 8-fold increase in the analytical signal. More importantly, at an initial template concentration of 3 × 10³ copies/L, the threshold cycle number reduces from 25 to less than 5. Another way to improve the signal-to-background ratio is achieved by using a higher probe concentration during the immobilization step (Table 1). The probe concentration during the immobilization step for all previous experiments was 1 M. When the concentration was raised to 100 M, the analytical signal was increased by a factor of 8.

Figure 5 ERT-PCR plot of current signal vs PCR cycle number in the presence of target DNA template (3 × 103 copies/L) at different Vent polymerase concentrations: 0.32 units/L, · 0.24 units/L, 0.04 units/L.

Conclusions

We have successfully demonstrated the implementation of the ERT-PCR technique in an integrated silicon-glass microchip. Key technical issues including electrode surface passivation, effect of potential scanning on the fidelity of the electrochemical detection platform, quantification performance, as well as effects of enzyme and probe concentrations on the signal-to-background ratio have been carefully evaluated. This new detection platform compares very well with the state-of-the-art fluorescence-based real-time PCR techniques in terms of speed and portability. With the design of multiple working electrodes on the same microchip, this technique is also very promising for the real-time multiplexing detection. Progress has been made to incorporate sample preparation functionality into the miniaturized device. Once fully developed, it offers tremendous opportunities in point-of-care nucleic acid analysis.

Acknowledgment

The authors express thanks for the funding support from the Research Grants Council of the Hong Kong Special Administrative Region Government (RGC CERG Project No. 601106). Laboratory facilities provided by the Bioengineering Graduate Program and Nanoelectronics Fabrication Facility are also acknowledged.