Anal. Chem. 2005, 77, 5189-5195
Electrical Detection of DNA Hybridization Based on Enzymatic Accumulation Confined in Nanodroplets Gilles Marchand,* Cyril Delattre, Raymond Campagnolo, Patrick Pouteau, and Fre´de´ric Ginot
Equipe Commune LETI/bioMe´ rieux, CEA/LETI/DTBS, 17 avenue des Martyrs, 38054 Grenoble Cedex 09, France
Electrical monitoring of DNA hybridization is one way to reduce the cost and size of the DNA chip reader in comparison with the more classical optical detection. Within electrical methods, electrochemical detection shows very high performances in terms of accuracy and sensitivity, especially when an enzymatic accumulation is used to amplify the signal. However, signal multiplexing for miniaturized systems based on both enzymatic accumulation and electrochemical detection remains challenging due to the Brownian diffusion of the detected product of the enzymatic reaction. We present here a DNA chip with electrical detection based on the following sequence: (i) hybridization of nucleic acids and washing in a liquid layer as usual, (ii) formation of independent nanodroplets on each detection site, (iii) enzymatic accumulation in each droplet avoiding cross-contamination between neighboring sites, and (iv) electrochemical detection of the product accumulated during the enzymatic reaction. The simple and fast transition from the liquid layer (hybridization step) to an array of nanodroplets (enzymatic accumulation and detection steps) was performed through the filling of the hybridization chamber with a solution containing the enzymatic substrates, the drawing of this solution, and the simultaneous creation of droplets thanks to retention areas based on circular rims or hydrophilic rings. Using this approach, hybridization is achieved in a liquid layer as usual, followed by the enzymatic accumulation in nanodroplets to avoid the cross-talk between neighboring sites. Moreover, working in droplets enables a fast increase in the concentration of the product generated by the enzymatic reaction and thus an improvement of the detection limit of the system. Since their first development in the late 1980s, DNA chips have evolved into a powerful tool providing complex and informative data from nucleic acid sequences. Taking advantage of the acceleration of genomics discoveries, this technology has proven invaluable in many fields of research and diagnostics.1 All commercially available methods appear to rely on fluorescence emission or optical reflectance for detection. Although these detection methods are highly sensitive, they need fluorescence scanners that are inherently costly and not transportable. This prevents or limits the use of DNA chips, for point-of-care testing * To whom correspondence should be addressed: (e-mail) gilles.marchand@ cea.fr; (fax) +33 (0) 438 785 787. (1) Epstein, J. R.; Biran, I.; Walt, D. R. Anal. Chim. Acta 2002, 469, 3-36. 10.1021/ac0505066 CCC: $30.25 Published on Web 07/02/2005
© 2005 American Chemical Society
or as a routine diagnostics tool. Other detection methods have been proposed in which hybridization of nucleic acids is directly related to an electric signal. This enables a drastic simplification of the reader. Contact imaging of DNA chips with photodetector arrays has been proposed to be one of such promising systems. Authors2,3 showed the feasibility of such a system using a chargecoupled device chip combined with radioactive labeling of target molecules. In 2005, Mallard et al.4 presented a system using direct functionalization of a low-cost complementary metal oxide semiconductor photodetector and labeling of the hybridized DNA target with an enzyme specific for chemiluminescent reaction. This system shows high performances in terms of accuracy and sensitivity in a multiplex format. However, the density of sites is limited by the diffusion of the excited molecules before photon emission. Electrical systems were developed for detection of both labelfree and labeled targets.5 Generally, the systems using direct detection (label free), based on oxidation of nucleic acids,6 oxidation or reduction of an electrochemical probe,7 or quartz crystal microbalances,8,9 are not sensitive enough for diagnostic applications. Indirect detection (labeled targets) is a way to improve the sensitivity of the electrochemical systems. Systems have been reported in which electrochemical detection relies on redox reactions of molecules generated via enzyme labels grafted on DNA targets.10-12 Enzyme labeling of target molecules gives signal amplification through generation of a large number of electroactive molecules from each enzyme (typically, a molecule of horseradish peroxidase or alkaline phosphatase4 has a turnover frequency ranging from 1000 to 10 000 s-1). Such a system (2) Eggers, M.; Hogan, M.; Reich, R. K.; Lamture, J.; Ehrlich, D.; Hollis, M.; Kosicki, B.; Powdrill, T.; Beattie, K.; Smith, S. Biotechniques 1994, 17, 516525. (3) Lamture, J. B.; Beattie, K. L.; Burke, B. E.; Eggers, M. D.; Ehrlich, D.; Fowler, R.; Hollis, M. A.; Kosicki, B. B.; Reich, R. K.; Smith, S. R. Nucleic Acids Res. 1994, 22, 2121-2125. (4) Mallard, F.; Marchand, G.; Ginot, F.; Campagnolo, R. Biosens, Bioelectron, 2005, 20, 1813-1820. (5) Drummond, T. G.; Hill, M. G.; Barton, J. K. Nat. Biotechnol. October 2003, 21, 10, 1192-1199. (6) Palecek, E. Anal. Biochem. 1988, 170, 421-431. (7) Ihara, T.; Maruo, Y.; Takenaka, S.; Takagi, M. Nucleic Acids Res. 1996, 24, 21, 4273-4280. (8) Wang, J.; Jiang, M.; Palecek, E. Bioelectrochem. Bioenerg. 1999, 4, 477480. (9) Galasso, K.; Livache, T.; Roget, A.; Vieil, E. J. Chim. Phys. 1998, 95, 15141517. (10) Azek, F.; Grossiord, C.; Joannes, M.; Limoges, B.; Brossier, P. Anal. Biochem. 2000, 284, 107-113. (11) Wang, J. Anal.Chim. Acta 2002, 469, 63-71. (12) Hwang, S.; Kim, E.; Kwak, J. Anal. Chem. 2005, 77, 579-584.
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005 5189
achieves high sensitivity.10 However, site density is limited by the diffusion of electroactive species. This diffusion leads to crosscontamination and false reading for DNA hybridization when using a microarray of electrochemical cells. Thus, the use of this enzymatic concept is problematic for arrays with a high site density. In some cases, systems with 3D structures13 were proposed to avoid contamination between neighboring sites. However, with this kind of approach, fluidic problems disturb the filling of the liquid layer used for the hybridization step. Others authors14 proposed to monitor the product generated during the enzymatic reaction (enzymatic product) by amperometric detection, but such an approach does not solve the issue of diffusion of the electroactive molecules (after some times, a crosscontamination will occur). In a different way, Willner15 suggested the biocatalyzed precipitation of an insoluble product on a transducer. However, with this approach, a broadening of sites was observed and the dynamic range was small because the enzyme activity is affected by the precipitation it generates. In this paper, we present an electrochemical system based on (i) hybridization of nucleic acids in a liquid layer as usual, (ii) washing with hybridization buffer, (iii) formation of independent nanodroplets on each detection site (comprising an electrochemical cell and a retention area), (iv) enzymatic accumulation in each droplet avoiding cross-contamination between neighboring sites, and (v) multiplexed electrochemical detection of the accumulated enzymatic product. EXPERIMENTAL SECTION Reagents. The following oligonucleotides, supplied by Eurogentec, were used in this study: specific hybridization probe, pyrrole 5′TTTTTTTTTTGCCTTGACGATACAGCTA3′; nonspecific hybridization probe, pyrrole 5′TTTTTTTTTTTTGGAGCTGCTGGCG3′; horseradish peroxidase (HRP)-labeled target oligonucleotide, 5′TAGCTGTATCGTCAAGGC-HRP-3′. For the specific and nonspecific hybridization probes, the oligonucleotide is grafted on position 1 of the pyrrole (pyrrole-1-oligonucleotide). All reagents used were reagent grade. MilliQ water (18.2 MΩ) was used to prepare aqueous solutions. Lithium perchlorate (Sigma Aldrich 20,528-1), o-phenylenediamine tablets (OPD; Sigma P7288), hydrogen peroxide 30% (w/w) solution (Sigma H1009), Triton X-100 (Sigma T8787), sodium chloride (SDS 1380517), pyrrole (Sigma Aldrich 13,170-9), tris(hydroxymethyl)aminomethane (Prolabo 33621.260), ethylenediaminetetraacetic acid (Acros Oganics 118432500), and 1H,1H,2H,2H-perfluorooctyltrichlorosilane (Fluka 77279) were used as received. Pyrrole-3-ethanol was supplied by Apibio. Instrumentation. Electrodeposition and voltammetric measurements were performed on a chip connected to an EGG-PAR model 283 potentiostat equipped with Echem software or an Autolab PGSTAT 20 potentiostat from Eco Chemie BV equipped with general purpose electrochemical system software. The wetting properties of surfaces were characterized by their static contact angle with MilliQ water as reference material. Contact angles were measured using a Digidrop system from GBX. Fabrication of Devices. Microelectrode Fabrication. Each step was performed in a cleanroom environment with standard pro(13) Infineon, patent WO 02/090573. (14) Dill, K.; Montgomery, D. D.; Ghindilis, A. L.; Schwarzkopf, K. R.; Ragsdale, S. R.; Oleinikov, A. V. Biosens. Bioelectron. 2004, 20, 736-742.
5190 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
Figure 1. Schematic diagram of devices microfabrication (in transverse section). (a) Pt microelectrodes on oxidized silicon wafer. (b) Pt microelectrodes surrounded by circular rims. (c) Hydrophobic functionalization outside circular rims. (d) Geometrical parameters for circular rims (See text for typical values).
cesses for silicon micromachining. The initial substrate was a 100mm n-type silicon wafer with a 500-nm layer of silicon oxide obtained by furnace oxidation at 1050 °C under a flow of water vapor. A platinum layer of 500 nm was deposited by sputtering. A 1.4-µm photoresist layer (S1813/Shipley) was deposited by spincoating on the Pt film and exposed through a first photomask to produce a pattern for microelectrodes, connecting tracks, and electrical connection pads. This pattern was subsequently completely etched using an argon ionic etcher (Argon Ion Beam Etch System, Veeco Microetch 801 machine). After removal of the remaining photoresist, plasma-enhanced chemical vapor deposition SiO2 film (300 °C/SiH4 and N2O mixture/STS Multiplex machine) of 500 nm was deposited and further baked at 500 °C for 3 h under nitrogen flow. A 1.85-µm photoresist layer (S1813/ Shipley) was deposited on top of this film and exposed through a second photomask to open the SiO2 layer at locations of microelectrodes and electrical connection pads. This layer was then completely etched using reactive ion etching (CHF3 and O2 mixture/Nextral 100 machine) process (Figure 1a). Circular Rim Fabrication. Circular rims were fabricated around an electrochemical cell by a single photolithography step with a negative photoresist (Durimide D7510) (Figure 1b). Speed of spincoating for photoresist was chosen to give the desired thickness at the end of the overall process. After UV exposure through a photomask and development, the photoresist was hard baked for 4 h at 350 °C. This final step gives the durimide a chemical resistance to almost all solvents (for example, acetone and ethanol) and aqueous solutions. A hydrophobic patch surrounding the circular rims was added on some chips by performing further lithographic steps (Figure 1c). As an example, a classical liftoff process can be followed using a sacrificial photoresist layer deposited inside the circular rim. In this way, a hydrophobic silane was grafted only outside the circular rims (see Chip Functionalization). The mechanism of droplet formation was investigated using matrixes of circular rims with varying dimensions (diameter (d) measured at rim center ranging from 0.06 to 1 mm, minimal wall to wall distance (L) of 60 µm, rim height (h) from 8 to 15 µm, rim width (w) from 10 to 30 µm) fabricated on a Si wafer with 500-nm SiO2 layer (Figure 1d). Fluidic Packaging. Both type of chips were first electrically connected to a printed circuit board using gold wire bonding. Electrical wires and platinum connection pads were then isolated using a UV-cured glue (Vitralit UV 6181). Silica capillaries of 100µm internal diameter (Polymicro Technologies, LLC TSP100200) were used as inlet and outlet tubing. A wall (1 mm high and 300
µm wide) was created following chip edges using UV-cured glue, and capillaries were included inside this wall. Finally, a polycarbonate cover plate was glued on top of the wall to create a cavity with an inlet and an outlet. This packaging scheme creates a liquid layer in the chamber with a volume close to 4 µL. Experimental Procedures. Chip Functionalization. Hydrophobic Treatment. Chips were incubated 2 h at room temperature in a 1/3/4 mass proportions of NaOH, deionized water, and ethanol or submitted to an oxygen plasma atmosphere (600 W, 45 s, 25 cm3 min-1, 165 mTorr, Plassys MDS 150 machine) in order to generate silanol groups on the chip surface. After drying, chips were incubated at room temperature for 10 min in a mixture of 9 mM 1H,1H,2H,2H-perfluorooctyltrichlorosilane in anhydrous toluene. After extensive washing with ethanol, chips were incubated in an ultrasonic bath in ethanol for 5 min. Chips were finally baked for 1 h at 110 °C. Electrodeposition of Pyrrole-1-oligonucleotide and Pyrrole-3ethanol. The anodic copolymerization of pyrrole and pyrrole-1oligonucleotide16,17 on the counter electrode was carried out in an electrolyte LiClO4 0.1 M in water containing 20 mM pyrrole and 5 µM pyrrole-1-oligonucleotide at 1 V versus a pseudoreference platinum electrode during 2 s. The anodic polymerization of pyrrole-3-ethanol on the auxiliary electrode was carried out in an electrolyte 0.5 M LiClO4 in water containing 100 mM pyrrole-3ethanol at 1 V versus platinum electrode during 5s. After each electropolymerization, modified chips were extensively washed with deionized water and dried. Hybridization and Enzymatic Accumulation. Chips were incubated for 30 min at 37 °C in the presence of different concentrations of HRP-labeled target in a hybridization buffer (10 mM Tris pH 8.0, 1 mM EDTA, 1 M NaCl, and 0.05% Triton X-100) in a wet room. After hybridization, the chips were then washed extensively with hybridization buffer. The remaining washing buffer was discarded, and the enzymatic mix (20 mL of phosphatecitrate buffer 50 mM, 20 mg of OPD, 4 µL of H2O2) was added on to the chips and drawn immediately. Droplets stayed only on the functionalized sites. OPD is a classical substrate used in colorimetry, for example, in ELISA tests. In our application, it was used as an HRP substrate in order to generate a product with electrochemical properties (2,3-diaminophenazine, DAP). The chips were incubated at room temperature for 5 min in darkness, and DAP was then detected by electrochemical detection. Electrochemical Detection. The enzymatic product was measured by cyclic voltammetry at 50 mV/s or by differential pulse voltammetry (DPV) (pulse height of 100 mV (∆Ep), pulse width of 100 ms (tp), scanning velocity 5 mV/s) in the range 0 to -600 mV versus a pseudoreference platinum electrode. RESULTS AND DISCUSSION Principle of the System. The DNA chip described in this paper uses microelectrodes as both a solid support for the DNA probes and an electrochemical measurement device. Enzyme labeling of target molecules gives signal amplification through (15) Willner, I.; Willner, B. Trends Biotechnol. 2001, 19, 6, 222-229. (16) Caillat, P.; David, D.; Belleville, M.; Clerc, F.; Massit, C.; Revol-Cavalier, F.; Peltie´, P.; Livache, T.; Bidan, G.; Roget, A.; Crapez, E. Sens. Actuators, B 1999, 61, 154-162. (17) Bidan, G.; Billon, M.; Galasso, K.; Livache, T.; Mathis, G.; Roget, A.; TorresRodriguez, L. M.; Vieil, E. Appl, Biochem. Biotechnol. 2000, 89, 183-193.
generation of a large number of electroactive molecules for each enzyme molecule. However, this simple system is not efficient on its own for microarrays of electrochemical cells because electroactive molecules diffuse freely from one detection site to another. This diffusion could lead to cross-contamination and false reading for DNA hybridization. A new way of dealing with this issue consists of making a transition from a liquid layer (hybridization step) to multiple nanodroplets (enzymatic accumulation and detection steps). The creation of independent nanodroplets containing the electroactive product on each detection site can then be achieved. Diffusion is not an issue anymore as a physical barrier is in place. The transition from the liquid layer to a matrix of droplets was performed through filling of the hybridization chamber with a solution containing the enzymatic substrates, drawing of this solution, and then droplet formation by retention of liquid thanks to specific retention areas (circular rims or hydrophilic rings). The principle of this system is illustrated in Figure 2. In a more detailed way, DNA probes are grafted by electrocopolymerization onto the electrode used as counter electrode for measurement of the electroactive product. Complementary DNA targets that hybridize to these probes are injected in a liquid layer (hybridization chamber) in order to have access to all available targets in the bulk during the hybridization step. These targets have been initially coupled to HRP enzymes. After washing with the hybridization buffer and then a surfactant-free buffer, a solution containing the enzymatic substrates (OPD and hydrogen peroxide) is introduced into the chamber (Figure 2b) and immediately drawn out (Figure 2c), leaving nanodroplets on detection sites (Figure 2d). Easy and total dewetting of the surface occurs during the drawing step, using a hydrophobic treatment on the chip if necessary. Retention of the aqueous solution was achieved by retention areas formed by circular rims or a hydrophilic ring (made of a conducting polymer). In this way, the enzymatic reaction can take place in a well-defined volume, which on one hand avoids the cross-contamination between neighboring sites and on the other hand leads to a faster increase in species concentration. After the accumulation time during which the enzymatic reaction occurs, the enzymatic product (electroactive) is detected on the central electrode (working electrode) by a standard electrochemical detection technique. Description of the Chips. In our system, the electrochemical cell for detection is constituted of three microelectrodes: working electrode (for the measurement), counter electrode, and reference electrode. Two types of chips were used in this study (Figure 3). Chip 1 was composed of an electrochemical cell with the following dimensions: 300-µm-diameter working electrode, 130-µm-wide circular counter electrode, 50 × 130 µm square reference electrode, and 70-µm gap between each electrode. Around the counter electrode, a 130-µm-wide auxiliary electrode was added with a 70-µm gap. On this last electrode, the hydrophilic polymer will be deposited (see Chip Functionalization). Chip 2 was composed of four electrochemical cells with the following dimensions: 60-µm-diameter working electrode, 30-µm-wide circular counter electrode, 10 × 30 µm square reference electrode, and 10-µm gap between each electrode. Around the counter electrode, a circular rim (8-µm height and 1-µm width) was added (see Circular Rim Technology). Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
5191
Figure 2. Principle of transition from a liquid layer to a nanodroplets array by drawing the aqueous solution. (a) Schematic representation of a chip before introduction of the solution. (b) Filling of the chamber. (c) An intermediate state of drawing. (d) Droplet retention onto each site (comprising an electrochemical cell and a retention area).
Further, the behavior of the other electrodes was verified by measuring the peak current at various scan rates and comparing it with the Randles-Sevcik equation18.
ip ) (2.69 × 105)n3/2AD1/2Cv1/2
Figure 3. (a) Chip 1 composed of an electrochemical cell (working electrode (1), counter electrode (2), reference electrode (3)) and an auxiliary electrode (4). (b) Chip 2 composed of four electrochemical cells individually surrounded by a 8-µm-high circular rim (5). A hydrophobic patch was fabricated around the electrochemical cells (6).
Characterization of the Electrochemical Cell. Initial electrochemical experiments in the fluidic chamber were carried out both to check the electrochemical behavior of the electrodes and to characterize the platinum reference electrode (integrated pseudoreference electrode). Potassium ferro/ferricyanide (FeII/ FeIII) couple was chosen as a redox model to investigate the detection performances. A 0.3 M KCl aqueous solution with both compounds at the same concentration was used. Then, other measurements were carried out after electrocopolymerization of pyrrole-1-oligonucleotide and pyrrole on the counter electrode to check the negligible influence of these polymers on the behavior of the electrodes. First, a thin platinum wire macroelectrode was introduced in the liquid layer to characterize the behavior of the integrated pseudoreference electrode. Cyclic voltammetry showed an identical FeII/III redox behavior on chips 1 either with the platinum wire macroelectrode or with the integrated pseudoreference electrode (E′1/2 ) 20 mV versus platinum and ∆Ep )100 mVsdata no shown). 5192
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
(1)
where ip is the peak current, n is the number of electrons of the redox reaction, A is the area of the electrode, D is the diffusion coefficient, C is the concentration, and v is the scan rate. As shown in eq 1, the peak current of a solution species is proportional to the square root of the scan rate. For an FeII/III concentration of 10-5 M, a signal of 32 nA with chip 1 was obtained comparable with theoretical 35.7 nA. Thus, electrochemical results are well described by eq 1, which means the voltammetric response was a result of the linear diffusion of species. A relative standard deviation (RSD, ratio of standard deviation to the mean value) on the current value of 2% was observed between chips with an FeII/III concentration of 10-5 M. Shao et al. reported that a steady-state diffusion current can appear when very small electrodes are used.19 This was explained by a spherical element introduced in the diffusion flow due to edge effects. In this case, current intensity for a microelectrode with ring structure in a bulk solution can be described by eq 2.
Ilim ) 4nFDCr
(2)
where n is the number of electrons, F is the faradic constant, D is the diffusion coefficient, C is the concentration of redox species in the bulk, and r is the working electrode radius. (18) Bard, A. J.; Faulkner, L. R. Electrochemical Methods, Fundamentals and Applications; Wiley: New York, 1980. (19) Shao, Y.; Mirkin, M. V.; Fish, G.; Kokotov, S.; Palanker, D.; Lewis, A. Anal. Chem. 1997, 69, 1627-1634.
Figure 4. Cyclic voltammetry response of 6.6 × 10-5 M FeII/III in 0.3 M KCl at 50 mV/s versus platinum pseudoreference electrode with chip 2. Measurement in liquid layer format (- - -) and in droplet format (s).
For chips 2, a limiting current (15 nA for 6.6 × 10-5 M FeII/ was generated at the working electrode as shown on Figure 4. This steady-state diffusion current was higher than the value predicted by eq 2 by 1 order of magnitude. Cooper et al.20 observed the same phenomenon with a similar cell and a very small volume (600 pL) and explained it by a redox recycling process occurring in small volume. They claimed that this process is sufficiently fast so that it does not limit the electrochemical process. In our case, the chamber volume was much higher (4 µL), and the same phenomenon was observed. Diffusion law in homogeneous solution leads to an estimation of the diffusion length using the following equation21 FeIII)
l ≈ (4Dt)1/2
(3)
where l is the mean diffusion length of species, D is the diffusion coefficient, and t is the time Cooper and co-workers’ explanation can still be applicable in our case because the diffusion length estimated with eq 3 (120 µm for 5-s measurement) is compatible with a recycling process between working and counter electrodes (interelectrodes distance 10 µm). Moreover, the diffusion of species generated at the counter electrode toward the working electrode can be accelerated by the concentrations gradient created during the measurement. On chips 2, an RSD between chips of 15% was observed with 10-5 M FeII/III (an RSD of
