| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Oak Ridge Conference |
1 Center for Nanotechnology, NASA Ames Research Center, Moffett Field, CA 94035.
aAuthor for correspondence. Fax 650-604-5244; e-mail jli{at}mail.arc.nasa.gov.
Abstract
Background: Reducing cost and time is the major concern in clinical diagnostics, particularly in molecular diagnostics. Miniaturization technologies have been recognized as promising solutions to provide low-cost microchips for diagnostics. With the recent advancement in nanotechnologies, it is possible to further improve detection sensitivity and simplify sample preparation by incorporating nanoscale elements in diagnostics devices. A fusion of micro- and nanotechnologies with biology has great potential for the development of low-cost disposable chips for rapid molecular analysis that can be carried out with simple handheld devices.
Approach: Vertically aligned multiwalled carbon nanotubes (MWNTs) are fabricated on predeposited microelectrode pads and encapsulated in SiO2 dielectrics with only the very end exposed at the surface to form an inlaid nanoelectrode array (NEA). The NEA is used to collect the electrochemical signal associated with the target molecules binding to the probe molecules, which are covalently attached to the end of the MWNTs.
Content: A 3 x 3 microelectrode array is presented to demonstrate the miniaturization and multiplexing capability. A randomly distributed MWNT NEA is fabricated on each microelectrode pad. Selective functionalization of the MWNT end with a specific oligonucleotide probe and passivation of the SiO2 surface with ethylene glycol moieties are discussed. Ru(bpy)2+-mediator-amplified guanine oxidation is used to directly measure the electrochemical signal associated with target molecules.
Summary: The discussed MWNT NEAs have ultrahigh sensitivity in direct electrochemical detection of guanine bases in the nucleic acid target. Fewer than 
1000 target nucleic acid molecules can be measured with a single microelectrode pad of 20 x 20 µm2, which approaches the detection limit of laser scanners in fluorescence-based DNA microarray techniques. MWNT NEAs can be easily integrated with microelectronic circuitry and microfluidics for development of a fully automated system for rapid molecular analysis with minimum cost.
Clinical laboratories are undergoing a revolution aimed at reducing the cost and time for molecular diagnostics through organizing, innovating, and implementing new technologies (1). This is increasingly important because of the great success of the Human Genome Project. The lack of a means to quickly and inexpensively detect specific molecular signatures has become the bottleneck limiting the full usage of genetic information that has been accumulated. However, current molecular diagnostics are highly labor-intensive and lack automation and integration, which can be done only in central clinical laboratories. Miniaturization technologies, which integrate engineering capabilities well established in the silicon microfabrication industry and microelectromechanical systems with the expertise in molecular biology, allow molecular tests to be carried out on microchips. Tremendous efforts have been made in the past decade to commercialize miniaturized instruments for molecular diagnostics, including thermocyclers, microfluidics chips, DNA microarrays, and other types of biosensors (1)(2).
An integrated molecular testing system involves innovation at each step, such as sample collection, nucleic acid extraction/separation, gene amplification, and signal detection/analysis. The major emphasis in the development of miniaturization technologies has been on highly sensitive biosensors that can be integrated with new technologies in other steps. Electronic techniques are of particular interest for this purpose because they can be directly integrated with microelectronics and microfluidics systems to gain advantages in miniaturization, multiplexing, and automation. Electrochemical methods are among the most popular electronic techniques that naturally interface the biomolecules in solutions with solid-state electronics (3). Many reports have demonstrated individually addressed microelectrode arrays for molecular analysis (4)(5)(6). However, the sensitivity of electrochemical detection based on microelectrodes is typically substantially lower than conventional laser-based fluorescence techniques. In recent years, nanoscale sensing elements such as carbon nanotubes (CNTs)1 and semiconducting nanowires have been incorporated in electronic biosensors to improve detection sensitivity (7)(8)(9)(10), which can greatly reduce efforts in sample preparation for a clinical test. CNTs have also been directly functionalized with biomolecules to serve as electrode materials with improved properties for use in electrochemical biosensors (11)(12)(13)(14).
In previous reports (14)(15)(16)(17), we demonstrated that nanoelectrodes, particularly a multiwalled carbon nanotube (MWNT)-based nanoelectrode array (NEA), can be integrated into an electrochemical system for ultrasensitive DNA detection. This technique uses a mediator-amplified guanine oxidation mechanism based on the use of the inherent guanine bases in target DNA molecules as the signal moieties. PCR amplicons can thus be measured directly without labeling. The detection sensitivity is comparable to the sensitivity of laser-based fluorescence methods in conventional DNA microarray techniques. Here we report further development of a multiplexing DNA chip and novel solutions for selective passivation/functionalization to increase the reliability of the system. The potential of such micronano-biofusion in developing low-cost biochips for molecular analysis will be discussed.
Materials and Methods
fabrication of mwnt nea
As described previously (14)(15)(16)(17)(18)(19), the fabrication scheme consists of the following steps, using a bottom-up approach:
Metal contact deposition.
A metal film (typically 200 nm thick chromium or platinum) is deposited on a silicon wafer covered with 500 nm SiO2 or Si3N4. The metal film serves as the electrical contact wiring individual microelectrode pads to the measuring circuitry. The size and layout of the microelectrode pads can be fabricated down to a few micrometers by use of ultraviolet lithography to satisfy the requirements for miniaturization and multiplexing for specific applications.
Catalyst deposition.
A nickel catalyst film of 10–30 nm thickness is deposited at the microelectrode pads to serve as the catalyst so that MWNTs are grown only specifically at nickel-covered locations.
CNT growth.
A forest-like vertically aligned MWNT array is grown on nickel catalyst film by plasma-enhanced chemical vapor deposition using a DC-biased hot-filament chemical vapor deposition system (20)(21). The mean diameter of the MWNTs can be varied from 30 to 100 nm, and the mean length is controlled at 5–10 µm.
Dielectric encapsulation.
A SiO2 film is deposited by thermal chemical vapor deposition using tetra-ethylorthosilicate at a vapor pressure of 400 mTorr and at a temperature of 715 °C in a quartz-tube furnace. SiO2 forms a conformal film filling the space between MWNTs as well as covering the substrate surface.
Planarization.
The dielectric film dramatically increases the mechanical strength of the CNT array so that a chemical mechanical polishing process can be applied to planarize the surface. Excess SiO2 and part of the MWNTs are removed so that the very ends of some MWNTs are exposed. We control this process so that only a small number of MWNTs are exposed to form a low-density inlaid NEA.
array-in-array: multiplex mwnt nea
Clinical diagnostics require addressing each individual microspot in an array, which is affixed with a specific oligonucleotide probe. In principle, a single MWNT electrode can be individually wired to achieve the highest sensitivity and multiplexing, but there will be a large uncertainty in the test results because of variations in the properties of individual MWNTs. To avoid this problem, the size of the microelectrode pad is constructed to be >10 µm so that 100 or more MWNT nanoelectrodes are formed on each pad. This array-in-array format can maintain the high-degree multiplexing desired as well as provide improved statistical reliability because of automatic averaging of signals collected from many MWNT nanoelectrodes. Shown in Fig. 1
is a 3 x 3 array with a MWNT NEA on each 200 x 200 µm2 microelectrode pad, which was fabricated with the method described above.
|
Shown in Fig. 1A
are 30 dies with the 3 x 3 array pattern on a 4-inch Si3N4-covered silicon wafer after chromium metallization. Fig. 1B
is an optical picture of a single die after MWNT growth by plasma-enhanced chemical vapor deposition. The black area is covered with a forest-like MWNT array, and the white area is free of CNTs. The 3 x 3 array, at the lower portion Fig. 1B
, consists of nine microelectrode pads for DNA detection, which are individually connected to larger 1 x 1 mm2 contact pads, at the top portion of Fig. 1B
, through a metal line at the center of the white channels. The chip can then easily be connected to a potentiostat through these large contact pads. The contact pads as well as the massive surrounding area are intentionally covered with forest-like MWNT arrays to produce raised surfaces at the same height as the microelectrode surface, which makes the chemical mechanical polishing process much easier. Shown in panels C through E of Fig. 1
are scanning electron microscope images zoomed in step-by-step at one of the microelectrode pads. Clearly, the MWNT growth is highly selective at the area covered with nickel catalyst (i.e., the gray area in panels C and D of Fig. 1
). The metal line is isolated from the surrounding MWNT arrays by a 20-µm gap (as shown in Fig. 1D
). MWNTs have uniform alignment and size distribution over the whole chip. The mean spacing between neighboring MWNTs is 200–300 nm, with 20% variation in the length. Stopping chemical mechanical polishing at the proper stage allows only a small number of MWNTs to be exposed at the surface to form a low-density NEA as shown in Fig. 1F
.
Results
electrochemical properties of the mwnt nea
It is known that there is a hemispherical diffusion layer around each nanoelectrode exposed at the SiO2 surface. In previous studies (14)(15)(16)(17), we found that the distance between neighboring MWNT nanoelectrodes is critical in determining the electrochemical behavior. Only NEAs with a mean nearest spacing >1 µm show sigmoidal shape in the cyclic voltammetry measurements (14)(15)(16)(17), indicating that the diffusion layers of neighboring MWNT nanoelectrodes are not significantly overlapped and that each nanoelectrode behaves nearly as an independent nanoelectrode, originally referred to as ultramicroelectrodes (22). This is consistent with the empirical estimation (23) for ultramicroelectrode arrays, which require a nearest spacing over six times the radius of the electrodes. Such low-density NEAs are essential for achieving ultrahigh sensitivity for biosensor applications. Fig. 2A
shows a typical cyclic voltammetry measurement obtained in 1.0 mmol/L K4[Fe(CN)6] and 1.0 mol/L KCl with one of the electrodes in the 3 x 3 array discussed in Fig. 1
. Clearly, a steady-state feature is observed corresponding to the radial diffusion process at each nanoelectrode. The nanoscale size and low density ensure low background noise and a small cell time constant so that pulsed techniques such as differential pulse voltammetry (DPV) can be reliably applied (as shown in Fig. 2B
). Both cyclic voltammetry and DPV confirm the well-defined electrochemical properties of the low-density MWNT NEA.
|
selective passivation and functionalization
In a previous report (15), we found that there is substantial nonspecific DNA adsorption on the SiO2 surface. A rigorous three-step washing procedure in 2x standard saline citrate (SSC) containing 1 g/L sodium dodecyl sulfate, 1x SSC, and 0.1x SSC, respectively, had to be applied to eliminate the nonspecific adsorption and reduce false-positive results. However, this procedure increases false-negative results. To solve this problem, the SiO2 surface needs to be passivated with protective moieties such as ethylene glycol, which is known to resist the nonspecific adsorption of biomolecules (24). For MWNT NEAs, it is possible to selectively passivate the SiO2 surface with ethylene glycol moieties and have the MWNT surface refreshed and functionalized with desired probe biomolecules, as shown schematically in Fig. 3A
.
|
In the first step, the MWNT NEA is thoroughly cleaned by first soaking the sample in 1.0 mol/L HNO3 for 15 min, which forms –OH groups on the SiO2 surface. The sample is then immersed in an 8 g/L solution of 3-aminopropyltriethyoxysilane (Aldrich) in ethanol for 20 min to produce a primary amine surface functionality (7). After rinsing with ethanol and water, a terminal ethylene glycol functionality is introduced by coupling of the surface amine with the terminal carboxylic acid group of 2-(2-methoxyethoxy)acetic acid. A solution (50 µL) of 0.1 mmol/L 2-(2-methoxyethoxy)acetic acid (Aldrich) is mixed with 5 mg of 1-ethyl-3(3-dimethylaminopropyl) carbodiimide hydrochloride (Fluka) and 2.5 mg of N-hydroxysulfosuccinimide (Aldrich), which act as coupling reagents (25). The solution is applied to the functionalized MWNT NEA and incubated at room temperature for 2 h. The sample is then thoroughly rinsed with water.
In the second step, the ethylene glycol-passivated MWNT NEA is electrochemically etched at 1.5 V (vs saturated calomel electrode) for 120 s in 1.0 mol/L NaOH, which selectively removes the molecules attached to MWNT during the passivation step. Even some carbon atoms at exposed MWNTs are removed to regenerate a fresh MWNT surface that is rich with –COOH and –OH functionalities (26). A 10 µmol/L oligonucleotide probe related to the BRCA1 cancer gene (27) [Cy3-5'-CTIIATTTCICAIITCCT-3' (AmC7-Q); QIAGEN] in 50 µL of phosphate-buffered saline (PBS; Sigma) is mixed with the coupling reagents [0.5 mg of 1-ethyl-3(3-dimethylaminopropyl)carbodiimide hydrochloride and 0.25 mg of N-hydroxysulfosuccinimide]. Electroactive guanine bases are substituted by nonelectroactive inosine bases in the DNA probe. The reaction mixture is applied to the MWNT NEA and incubated at room temperature for 1 h. The sample is then thoroughly rinsed with PBS.
This procedure takes advantage of the localized addressing property of electronic techniques so that the functionalization can be applied specifically at desired MWNT electrodes. For a multiplex array, MWNTs at each microelectrode pad can be sequentially etched and flushed with specific probe molecules mixed with the coupling agents. Localized functionalization can be achieved down to a few micrometers without use of photosynthesis techniques. This could be a great advantage for low-cost automation in fabricating high-degree multiplex DNA chips.
Because SiO2 covers >99% of the surface, the ethylene glycol passivation dramatically reduces the inaccuracy caused by nonspecific adsorption on the SiO2 surface. This is particularly important for clinical tests with a very small amount of target molecules. Panels B through E in Fig. 3
show the dramatic difference in Cy3 fluorescence after probe functionalization between a nonpassivated MWNT NEA and one with ethylene glycol passivation. The high intensity of Cy3 fluorescence (Fig. 3B
) is mainly the result of nonspecific adsorption on the SiO2 surface, which cannot be removed by just rinsing. The rigorous three-step washing removed most of the Cy3, as shown in Fig. 3C
. After passivation of SiO2 with ethylene glycol moieties, a simple rinsing step is enough to remove as many of the nonspecifically adsorbed molecules as are removed by rigorous washing, which greatly simplifies the testing procedure.
electrochemical detection of dna hybridization
Target hybridization.
The target oligonucleotide, with a sequence of Cy5–5'-AGGACCTGCGAAATCCAGGGGGGGGGGG-3' (QIAGEN) and containing additional 10 guanine bases (in bold) as electrochemical signal moieties, was hybridized by incubating the probe-functionalized electrodes in 0.1 µmol/L target solutions in 3x SSC buffer over 2 h at 40 °C. A wild-type sequence of 300 bases within the BRCA1 gene containing 5'-AGGACCTGCGAAATCCAG-3', which is complementary to the oligonucleotide probe, and an unrelated wild-type sequence of 400 bases within the BRCA1 gene were hybridized under the same conditions after denaturation at 95 °C for 5 min and quenching in an ice bath before hybridization. The electrodes were then rinsed thoroughly with PBS.
Electrochemical detection mechanism.
As shown in Fig. 4
, the guanine groups in the target DNA serve as electrochemical signal moieties that provide a small current produced by the oxidation at 1.02 V (vs saturated calomel electrode). This signal is passed through MWNTs to the underlying circuit, but this current is very small because of the very limited number of guanine bases in the target molecules. A metal complex ion, Ru(bpy)32+, needs to be introduced as a mediator that amplifies the guanine oxidation signal by an electrocatalytic mechanism (28), as illustrated in Fig. 4B
. This works for both oligonucleotide and PCR amplicon targets. In fact, the specific PCR amplicon with 300 bases contains 75 inherent guanine bases. These guanine bases are near the MWNT surface but may not be in direct contact with it. However, the mediator likely serves as a shuttle to transfer electrons from any guanine base approximately within the hemispherical diffusion layer, so that all of the signal moieties are still active. Because the inherent guanine bases in the dangling single-stranded part of the target DNA can be used as signal moieties, their presence eliminates the requirement of labeling and greatly reduces the cost and time for molecular analysis. The MWNT produced by plasma-enhanced chemical vapor deposition contains bamboo-like multiwalled nanofibers, unlike the ideal MWNT, and presents a series of closed graphitic shells along the tube axis (16), which in turn have the advantage of sealing electrolytes from access to the hollow channels at the center, as shown schematically in panels A and C of Fig. 4
.
|
Electrochemical detection algorithm.
AC voltammetry (ACV) and DPV are straightforward techniques for deriving a signal associated with guanine oxidation (11)(12)(13)(14). Shown in Fig. 5
are ACV measurements in 5.0 mmol/L Ru(bpy)32+ with 0.20 mol/L sodium acetate (pH 5.2) obtained with an AC sinusoidal wave at 10 Hz and 25 mV amplitude superimposed on the staircase DC potential ramp from 0.50 to 1.20 V. Panels A, C, and E of Fig. 5
show the raw data with three consecutive scans in each measurement after incubation of the probe-functionalized MWNT NEA with oligonucleotide targets, with a specific PCR amplicon, and with an unrelated PCR amplicon, respectively.
|
All measurements show well-defined peaks around 1.02 V (vs saturated calomel electrode) as indicated by the long arrows in panels A, C, and E of Fig. 5
. For complementary oligonucleotides and PCR amplicon targets, the first scan always gives a much higher peak, whereas the peaks in the second and scans are approximately the same height. In contrast, the peak height of the sample incubated in unrelated PCR amplicon increases slightly in each subsequent scan. This phenomenon occurs because the peak current at 1.02 V in the first scan consists of two parts:
 | (1) |
 | (2) |
 | (3) |
We can further correct the variation in MWNT density in each NEA by normalizing 
ip1,2 with ip2. The normalized difference between the data obtained in the first and second scans (red solid lines) and between the second and third scans (blue dotted lines) are shown in panels B, D, and F of Fig. 5
.
Clearly, only the difference between the data obtained in the first and second scans is associated with complementary oligonucleotides and PCR amplicon targets, which give a clear positive peak around 1.02 V, as indicated by the short up arrows. The difference between the data obtained in the second and third scans of all measurements and the difference between the data obtained in the first and second scans of the unrelated PCR amplicons are either almost zero or a very small negative number. As a result, the second and third scans can be used as internal controls, with normalized differential peak heights that should fall within approximately ±0.025, as indicated by the two dashed lines in panels B, D, and F of Fig. 5
. If the normalized differential curves of the first and second scans have a positive peak above the top dashed line, a positive conclusion can be drawn. These results obtained with ethylene glycol-passivated MWNT NEAs are consistent with previous data obtained with nonpassivated electrodes (15).
Discussions
We have estimated that the maximum number of target DNAs that are hybridized to the probes at the ends of the CNTs on a 20 x 20 µm2 microcontact is <1000. The actual number could be many times smaller. The detection limit and quantitative analysis need to be calibrated with other techniques with even higher sensitivities, such as isotope or fluorescence photocounting techniques. However, it is clear that the underestimated detection sensitivity reported here already meets the critical requirements for diagnostics applications, for which "yes" or "no" answers are as important as quantitative measurements. In a previous report (15), we demonstrated that the positive differential signal, as shown in Fig. 5
, appeared only when target DNAs specifically hybridized on the electrodes functionalized with the complementary oligonucleotide probes. Other control experiments with bare CNT nanoelectrodes and those with noncomplementary probes all gave negative differential signals after the incubation in the PCR amplicons and polyG solutions.
In this report, we focused on demonstrating the technology in well-defined conditions. Hence, the incubation time was controlled over 2 h to ensure full hybridization. It can be shortened in the future for practical diagnostics. However, because the concentrations of the specific target DNAs in the sample are reduced, the time required for them to find the right probes will be significantly increased if the movement of target DNAs depends solely on diffusion. The electrochemical platform has great advantages in solving these problems. Electrical fields can be applied directly at each microcontact to speed up the movement of target DNAs during the hybridization process. In addition, negative potentials at the electrode surface have been reported to repulse negatively charged DNA molecules away from the surface and may provide additional stringent control (4)(29). For this purpose, the ACVs in Fig. 5A
were set to start at –0.10 V. However, we found that a redox wave may appear around +0.45 V in subsequent scans if the ACV scan is set to start at –0.10 V and run up to 1.20 V (as shown in Fig. 5
, A and B). This is associated with the formation of some electroactive functional groups on the carbon electrodes. The baseline is very stable only if ACV measurements start at +0.20 V or higher. This limited our current attempt to use negative electropotential for additional stringent control. Preliminary results show that application of a negative potential pulse (5 s at –0.10 V) or a sequence of bipolar pulses before ACV starts at +0.50 V can prevent the occurrence of the redox peak at 0.45 V. As seen in panels B and E of Fig. 5
, rather stable baselines are obtained with this protocol. Further investigation is underway to determine the optimum conditions in electrical fields to detach unmatched target molecules and accelerate the specific hybridization.
In conclusion, the feasibility of integrating vertically aligned MWNTs as individually addressed NEAs in a miniaturized multiplex electronic DNA chip has been demonstrated. MWNTs and the substrate are encapsulated in SiO2 matrix, leaving only a small number exposed at the surface, which forms a NEA with very low density. The SiO2 surface can be passivated with ethylene glycol moieties to reduce nonspecific adsorption, whereas the exposed MWNT end can be selectively functionalized with DNA probes by taking advantage of the localized addressing property of nanoelectrodes. The specific hybridization of both polyG-tagged oligonucleotide target and label-free PCR amplicon can be measured directly by electrochemical methods. This method provides a unique way to fuse microelectronic circuitry with biomolecules through nanostructured materials, which dramatically improves the detection sensitivity. There is a great potential to integrate this detection technology with other miniaturization technologies, such as microfluidics for sample separation and microthermal cycler for amplification, to develop a fully automated, low-cost electronic chip for rapid molecular analysis.
Acknowledgments
A.M.C. and J.H. are with the University Affiliated Research Center at NASA Ames operated by University of California, Santa Cruz. H.C. and Q.Y. are with ELORET Corporation. This work was supported by NASA under contract NAS2-99092. We acknowledge Drs. Hou Tee Ng and Wendy Fan for helpful discussions and Barbara Nguyen-Vu for participation in part of the work.
Footnotes
1 Nonstandard abbreviations: CNT, carbon nanotube; MWNT, multiwalled carbon nanotube; NEA, nanoelectrode array; DPV, differential pulse voltammetry; SSC, standard saline citrate; PBS, phosphate-buffered saline; and ACV, AC voltammetry. 
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
