Article pubs.acs.org/ac
Fuel Cell Virus Sensor Using Virus Capture within Antibody-Coated Nanochannels Yanyan Wei, Lai Peng Wong, and Chee-Seng Toh* Division of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological University, 21 Nanyang Link, Singapore 637371 S Supporting Information *
ABSTRACT: We report a unique fuel cell sensor system for the first time direct detection of unlabeled virus particles based on the formation of antibody−virus complexes within the sensor’s membrane nanochannels. This strategy exploits the change in the membrane resistance of the powered system, comprising a Prussian blue nanotubes (PB-nt) membrane cathode and a platinum mesh anode. The method reports an impressive shortest response time of ∼5 min toward the specific virus target, at low concentration values of 3−45 plaque-forming units per milliliter (pfu mL−1) with detection limit of 0.04 pfu mL−1, comparable to state-of-the-art polymerase chain reaction (PCR)-based methods. The sensor can clearly differentiate dengue virus serotype 2 from serotype 3. When filled with Nafion perfluorinated resin, the PB-nt membrane demonstrates powerful utilization as a stand-alone fuel cell based virus sensor, and thus offers the outstanding promise of a sustainable, low-cost, and rapid low-power virus detection tool.
A
electrolyte solution within the membrane nanochannels helps complete the circuit. When closed circuited, the output current signal can be made available to activate/run other devices. Lowpower devices that can switch on upon detection of virus can readily activate alarms to initiate responses to mitigate spread of the disease, identify diseased persons, and apply appropriate therapeutic care. In the fabrication of the fuel cell virus sensor, membrane nanochannels with dimensions (200 nm) larger than the virus−antibody complex (∼70 nm in size) are utilized to enable virus capture by the nanochannel-adsorbed antibody molecules.17 Recently, nanopores are highly promising in protein recognition and sensing works, because of their similar dimensions to protein sizes18,19 and especially when coupled to electrochemical methods.20 Formation of the virus−antibody complexes within the nanochannels reduces the channel ion conductivity during the operation of the fuel cell virus sensor and thus derives observable lowering of its amperometric signal response, useful for high-sensitivity quantitative measurements.21 To derive high selectivity response of the virus sensor, monoclonal antibody specific toward dengue virus, a vectorborne virus of significant local and global relevance to clinical and environmental health studies, is chosen. The sensor’s specific responses toward two structurally similar dengue virus serotypes are successfully demonstrated with shortest analysis time of ∼5 min with high selectivity against each other. Parts A
fuel cell as a sensing system that can respond specifically to viral diseases is of scientific and technological significance, because of global needs for comprehensive and powerful monitoring tools to prevent and control disease epidemics.1−3 These fuel cells can be deployed at strategic sites such as in aircraft, train, and cruise ship cabins, and other publicly accessed spaces including offices, schools, and entertainment centers, besides point-of-care clinics and hospitals. Among the different types of energy sources, H2O2 is frequently used in enzyme-based sensors since many analytes such as glucose can readily generate H2O2 by the catalytic actions of oxidoreductase enzymes.4 H2O2 is also a carbon-free energy source which itself can be generated abundantly from other renewable energy sources.5,6 A variety of sensors relying on chemicals as energy sources based on a biofuel or fuel cell design have been reported.7−12 However, until now there is no report on applying alternative energy sources for virus sensing because the strategy relying on enzyme, cell, or redox reactions at the biocatalytic anodes or cathodes cannot apply to viruses. This lack of virus-specific enzyme and redox reactions remains a significant obstacle to the development of fuel cells as a sensing system for viruses. To provide sufficient current at the microampere level, we chose iron hexacyanoferrate polymer nanotubes also known as Prussian blue (PB-nt), as the cathode material13,14 with high surface areas and known highest electrocatalytic activity toward hydrogen peroxide reduction.15,16 The polymer nanotubes are deposited within the nanochannels of a porous platinum-coated membrane template which separates the cathodic and anodic cell compartments. The anode comprises an auxiliary platinum electrode immersed in 0.5 M KCl in pH 7 Tris buffer, while the © 2013 American Chemical Society
Received: June 18, 2012 Accepted: January 3, 2013 Published: January 11, 2013 1350
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
Scheme 1. (A) Fabrication of the PB-nt Membrane by Sputtering an ∼50 nm Thick Platinum Layer on One Side of a 60 μm Thick Nanoporous Alumina Membrane, Followed by Electrodeposition of PB onto the Porous Pt Membrane Electrode; (B) Schematic of the Fuel Cell Virus Sensor Using a Two-Compartment Cell Separated by a Prussian Blue Nanotubes Membrane, Showing the Binding of Virus to Antibody Molecules within the Nanochannels; (C) Photograph and Schematic of the StandAlone Nafion-Filled Membrane Probe Using the Nanometer Thick Metal Layer of the Membrane at the Air−Electrolyte Interface as the Reference and Counter Electrode
field-emission scanning electron microscope (JEOL JSM6340F). Fabrication of the Prussian Blue Nanotubes Membrane and Immobilization of Antibody. Prussian blue nanotubes membrane was prepared as follows: First, an ∼50 nm thick platinum layer was sputtered on one side of the membrane (either 20 or 200 nm pore size). Then the conductive membrane was subjected to potential cycling from −0.5 to +0.6 V at 50 mV s−1 for 30 cycles in a solution containing 5.0 mM K3FeIII(CN)6, 5.0 mM FeCl3, 0.1 M KCl, and 0.01 M HCl, to give a blue-colored membrane that indicates formation of the PB-nt. This PB-nt membrane as cathode13 is fairly stable in its reaction with hydrogen peroxide where its anodic and cathodic peak currents decrease by only 25% after 100 potential cycles between 0 and 0.5 V (Supporting Information Figure S-1). Steady-state amperometric current obtained under closed-circuit condition shows high stability where the amperometric current changes within 5% error over 50 min (Supporting Information Figure S-2). All PB-nt membranes were immobilized with antibody by physical adsorption in the membrane nanochannels before use as a virus sensor as follows: Mouse antidengue type II or III virus monoclonal antibody stock was diluted to 0.2 μg mL−1 in 1 M Tris buffer, pH 7 and was stored in 4 °C before use. An amount of 50 μL of the 0.2 μg mL−1 antibody was applied onto the PB-nt side of the membrane and incubated at 4 °C for 45 min and subsequently at room temperature for further 15 min before use. Procedure for Open-Circuit Voltage and ClosedCircuit Current Measurements. The open-circuit voltage (OCV) and closed-circuit current were measured using the two-compartment cell setup according to Scheme 1B. Sensing solution refers to the solution in direct contact with 0.12 cm2
and B of Scheme 1 show the design of the two-compartment cell comprising the PB-nt membrane as the sensing electrode (also the cathode for H2O2 reduction) which faces the sensing solution where the hydrogen peroxide and virus sample are added. The reference solution contains the auxiliary (also the anode for H2O oxidation) and reference electrodes. The complete coverage of the platinum metal layer by PB-nt prevents the interfering reaction of H2O2 and O2 with the underlying platinum (see Scheme 1A), and in addition, PB is fairly inert toward oxygen but gives high electrocatalytic rates with H2O2.22 To further simplify the two-compartment cell and three-electrode system which can be difficult to operate under field conditions, an integrated two-electrode setup using a stand-alone dry membrane probe is developed (Scheme 1C).
■
MATERIALS AND METHODS Materials and Instruments. Nanoporous alumina membranes (Anodisc, 13 mm diameter, 20 and 200 nm pore sizes) from Whatman (Maidstone, Kent, U.K.), 37% HCl from P. P. Chemicals, 35% H2O2 from Alfa Aesar, KCl from Sinopharm Chemical Reagent Co. Ltd. (SCRC), potassium hexacyanoferrate(III) from Sigma-Aldrich, anhydrous ferric chloride from Merck, and tris(hydroxymethyl)-aminomethane from Bio-Rad Laboratories were used. All solutions were prepared in ultrapure water (Sartorius ultrapure water system). Nafion perfluorinated resin solution was from Sigma-Aldrich. Mouse antidengue type II and type III virus monoclonal antibodies were purchased from Millipore. A JEOL JFC-1600 auto fine coater was used to prepare the platinum-coated porous membrane electrode. Autolab PGSTAT128N (Ecochemie, The Netherlands) was used for impedance studies. PB-nt membrane was characterized using a 1351
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
Figure 1. Scanning electron micrographs of (A) the cross section of a 200 nm pore size nanoporous membrane template physically sputtered with an ∼50 nm thick conductive porous platinum layer, followed by electrodeposition of Prussian blue. (B) The enlarged micrographs reveal ∼40 nm thick coating of Prussian blue along the nanochannel walls.
membrane cell, so the H2O2 concentration is 1 mM. The background H2O2 current is allowed to reach steady-state condition after ∼10 min. After that, aliquots (in microliters) of 600 pfu mL−1 virus stock solution were successively added into the H2O2 solution. For the dry membrane probe, the electrolyte used was Nafion perfluorinated resin, incorporated within the 60 μm thick membrane structure, and the auxiliary electrode consists of the ∼50 nm thick platinum coating on the other side of the PB-nt membrane side exposed to surrounding air, in a twoelectrode cell arrangement. During virus sensing, 80 μL of 0.5 mM H2O2 was first added onto the PB-nt-coated side of the dry membrane probe under closed-circuit condition and allowed to generate steady-state current values after ∼7 min. After that, aliquots (2 μL) of 1000 pfu mL−1 virus stock solution were successively added into the H2O2 solution.
geometric area of the porous PB-nt membrane working (WE) electrode while reference solution refers to the solution where both the Ag/AgCl reference (RE) and platinum mesh counter (CE) electrodes are placed, using a three-electrode arrangement. Initial sensing and reference solutions in the twocompartment cell were 1.4 mL of 1 M Tris buffer and 0.5 M KCl (in 1 M Tris buffer), respectively. Closed-circuit amperometric current between the PB-nt membrane working and platinum mesh counter electrodes was measured over time using a galvanostat (eDAQ EA161) connected to a fourchannel data acquisition unit (eDAQ e-corder 401), according to Scheme 1B. All amperometric signals were collected using a 1 Hz low-pass filter to remove significant background noise. During the closed-circuit amperometric current measurements of the PB-nt membrane, stock solutions of H2O2 were added successively into sensing solution. The 0.1 M stock hydrogen peroxide solution was prepared by diluting 35% H2O2 in ultrapure water. For OCV measurement of the fuel cell virus sensor, 1 mM H2O2 sensing solution is used in the sensing solution and the voltage is monitored with time during the addition of dengue virus serotype-2 (DENV-2). The working electrode is the PB-nt membrane immersed in the sensing solution, and the counter electrode is a platinum mesh immersed in the reference solution of a two-compartment cell. The voltage difference between the PB-nt membrane working electrode and the platinum mesh counter electrode was measured over time using the high-impedance voltage measurement function of the CHI1220B potentiostat. Virus Culturing. First, Aedes mosquito cells (C6/36) were infected with dengue serotype 2 viruses at a multiplicity of infection (MOI) of 0.01 for 1 h at 28 °C. After that, the medium was removed and fresh growth medium (RPMI, 5% heat-inactivated fetal bovine serum (FBS) and 5 mM 2mercaptoethanol) was added. The cell culture was incubated at 28 °C for 5 days. After 5 days, the medium was collected and a plaque assay was carried out in order to determine the concentration of the dengue virus in terms of plaque-forming units (pfu mL−1). The virus sample was subsequently heatinactivated at 56 °C for 30 min. The virus stock was diluted to 600 pfu mL−1 in 1 M Tris buffer, pH = 7 and was stored in −20 °C before use. Procedure for Virus Sensing. Before virus sensing, 14 μL of 0.1 M H2O2 was added to the 1.4 mL of 1 M Tris buffer sensing solution of the closed-circuited two-compartment
■
RESULTS AND DISCUSSION Morphology of the Prussian Blue Nanotubes Membrane. Figure 1 shows the scanning electron micrographs of the cross-sectional view of a platinum-coated porous membrane filled with Prussian blue polymer after electrodeposition. The magnified view of Figure 1A reveals ∼40 nm thick coating of Prussian blue along the nanochannel wall which reduces the nominal pore diameter of the nanochannels to decrease somewhat to ca. 100 nm from the originally unmodified 200 nm nanochannels (Figure 1B). Thus, the membrane nanochannels remain porous even after the coating of Prussian blue followed by the immobilization of antibody. We have previously shown that the growth of Prussian blue along the nanochannel walls up to 5 μm can be deposited using the one-step potential cycling method.23 The electrode geometric area is 0.12 cm2 calculated from the radius of the electrode area in contact with the solution in the two-compartment cell. However, the effective surface area of Prussian blue in contact with the solution includes the membrane surface as well as the internal nanochannel walls. For nanotubes with ∼5 μm in length, the effective surface area of Prussian blue in contact with electrolyte solution is ∼3.0 cm2, i.e., ∼25 times larger than the membrane geometric area. The hydrogen peroxide current density of the porous membrane supported PB is 208 μA mM−1 cm−2, which is about 3-fold larger than those reported (∼60 mA M−1 cm−2)24 using PB-coated solid planar electrodes. 1352
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
Figure 2. (A) Open-circuit voltage and (B) closed-circuit steady-state current responses of a two-compartment PB-nt membrane cell toward H2O2. (C) Polarization power curves of the PB-nt membrane sensor obtained in the absence and presence of 5 mM H2O2 and 20 pfu mL−1 virus, using steady-state linear sweep voltammetry at a slow scan rate of 1 mV s−1. Solutions are degassed with nitrogen to remove oxygen interference. (D) Closed-circuit current responses of a single-compartment PB-nt membrane cell toward DENV 2.
s−1 scan rate) in the presence of 5 mM H2O2. Initially in the absence of H2O2, the PB-nt based virus sensor gives insignificant power output. Upon the addition of 5 mM H2O2, its power density increases ∼50 times to a maximum power density value of 12.7 μW cm−2. In contrast, when virus is added to the membrane cell, the maximum power density of this H2O2-powered fuel cell virus sensor decreases. Under steady-state closed-circuit current discharge condition in a solution containing 60 mM H2O2, a significantly larger power density can be achieved. The typical closed-circuit current of a PB-nt membrane using a 60 μm thick, 200 nm pore size membrane template can reach 300 μA at a closed-circuit potential of ∼0.3 V (measured using the high-impedance voltage measurement function of the CHI1220B potentiostat). Thus, the practically useful power output of the PB-nt membrane under this optimal condition can achieve an impressive ∼100 mW cm−3 (or 0.625 mW cm−2) and shows great potential in low-power applications. Besides H2O2 concentration, the cell’s power output can be further controlled by adjusting the other current-dependent parameters such as the amount of deposited PB-nt, quality of the sputtered metal layer which influences its conductivity, and membrane pore size. When immobilized with antivirus antibody, the PB-nt membrane can function as a virus sensor. In the situation of virus detection using a single-compartment arrangement, there is, however, no change in the background H2O2 current during the addition of DENV 2 to the sensing solution (Figure 2D). This is in contrast to the amperometric current responses when the virus sensor uses the two-compartment cell (Figure 3A). Thus, all virus detections were carried out using the twocompartment cell arrangement as described in Scheme 1B.
Characterization of the Two-Compartment PB-nt Membrane Cell. Voltammetry of a PB-nt membrane using the membrane cell setup in Scheme 1B gives a typical electrocatalytic response where the normalized cathodic peak current is enhanced compared to the normalized anodic peak current in the presence of H2O2 (Supporting Information Figure S-3A). Since dissolution of some PB-nt occurs during potential cycling, all voltammetric current signals are normalized to the anodic peak current of each voltammogram in order to compare the relative changes in the voltammetric currents. In the presence of virus, the normalized peak current decreases only slightly. Conversely, the normalized limiting currents (plateau shape) at high and low potentials are clearly lowered in the presence of virus, thus suggesting a limiting process that is sensitive toward virus concentration (Supporting Information Figure S-3B). Figure 2A gives the open-circuit voltage response of the PB-nt membrane cell toward the addition of an excess amount of H2O2. The open-circuit voltage gives a typical sigmoidal titration curve which reaches a plateau value and remains unchanged upon further addition of 20 mM H2O2, because most of the PB-nt have been oxidized to its Prussian blue form. In contrast, Figure 2B shows the closedcircuit current response of a PB-nt membrane continues to increase at high 60 mM H2O2 concentration. This is because under closed-circuit condition, the cathodic current which reduces the PB-nt electrode increases as more PB-nt becomes oxidized by H 2 O2 in the presence of increasing H 2 O2 concentration. It is clear from Figure 2B that the magnitude of the closed-circuit current response of the PB-nt membrane is strongly influenced by the H2O2 concentration. Figure 2C shows the power density curve of a PB-nt sensor obtained from steady-state linear sweep voltammetry (at 1 mV 1353
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
range of 0.25 V and 25 μA, so the virus sensor is operated with power of ca. 8.2 to 1.7 mW cm−3 over the concentration range from 3 to 45 pfu mL−1. Figure 3B gives the calibration plot of the fuel cell sensor closed-circuit current in the presence of virus, Ivirus, normalized to its initial steady-state closed-circuit current after the addition of 1 mM H2O2 but before the addition of virus, Ivirus=0. Normalization of the closed-circuit current signals give reasonably low 2−6% standard error for responses obtained from three sensor probes. In the absence of normalization, uncertainties up to ∼20% are commonly observed for three different membrane probes, since it is difficult to achieve identical thicknesses of the Prussian blue nanotubes and loading amounts of the nanochannel or surface-adsorbed antibody molecules. A low detection limit of 0.04 pfu mL−1 can be achieved using this unique virus capture approach to impede the fuel cell sensor current. The detection limit of the fuel cell virus sensor is determined from the minimum DENV-2 concentration that gives a decrease in the sensor current equivalent to 3 times the standard deviation of background current in the absence of DENV-2. This is calculated from the fitting equation y = 1 − (x/A)1/n (where A = 66.97, 1/n = 0.573) derived from the membrane resistance model in eq 2 below. At a limit of detection (LOD) of 0.04 pfu mL−1, the method is comparable to state-of-the-art real-time reverse transcriptase polymerase chain reaction (PCR) (LODs: 0.6,25 0.4,26 0.01 pfu mL−127). One critical measure of the virus sensor response is its selectivity to distinguish closely similar virus targets with highly conserved amino acid residue sequences. Figure 3C clearly shows that the fuel cell virus sensor with antidengue type II monoclonal antibody incorporated in the nanochannels can differentiate between DENV-2 and dengue serotype 3 (DENV-3) added sequentially to the sensing solution of the two-compartment sensor cell. At 6 pfu mL−1 virus concentration, the change in current by DENV-2 is significant. In contrast, Figure 3C shows that DENV-3 gives no change in current after the initial perturbed transient current relaxes back to the background current value after ∼4 min. The same transient current due to solution mixing is observed for a PB-nt membrane without immobilized antibody in its response toward DENV-2 and PB-nt membrane sensors with immobilized antibody toward nonanalytes (see Supporting Information Figure S-3). This outperforms present methods for the evaluation of dengue virus serotypes where the cross-reactivities of dengue serotypes due to common epitopes remain a challenge.28,29 Thus, the fuel cell virus sensor incorporated with antidengue 2 3H5 antibody using the PB-nt membrane is a high-sensitivity, rapid response, and high-specificity sensor for unlabeled dengue serotype 2 virus. Unlike typical electrochemical and optical antibody-based virus sensors, the unique fuel cell virus sensor does not require input of voltage power but relies on a very small amount of 1 mM H2O2 for each series of measurements. H2O2 is the common signaling reagent consumed or generated in situ during enzyme-linked immunosorbent assay (ELISA) for the direct detection of virus particles, virus protein antigens, and antibodies associated with viral infections.30,31 Therefore, H2O2 is an ideal reagent for powering the electroactive Prussian blue nanotubes in the presence of immune reagents. Impressively, the fuel cell virus sensor responds to the virus in shortest analysis time of ∼5 min, which outperforms all existing immune-based and PCR-based methods. Its linear range of less than 100 pfu mL−1 is highly relevant for early
Figure 3. (A) Typical closed-circuit steady-state current response of a fuel cell sensor toward DENV-2 virus and (B) its calibration plot of normalized closed-circuit steady-state current vs virus concentration; the line is the nonlinear best-fitted data; error bars are standard deviation of experiment data (points). (C) Selective and nonselective responses of the fuel cell virus sensor toward DENV-2 and DENV-3, added sequentially to the sensing solution.
Analytical Performances of the Fuel Cell Virus Sensor. It is highly desirable to develop low-cost fuel cell virus sensors that are highly specific and can quickly distinguish between viruses with similar disease symptoms for rapid therapeutic response at points of care. Fuel cell sensors that rely on redox reactions cannot detect unlabeled viruses. Therefore, we incorporate specific recognition antibody molecules into the PB-nt membrane to bind specifically to the target virus particles. Figure 3A shows that, using an antidengue 2 antibody-incorporated PB-nt membrane, the closed-circuit current of the virus sensor responds and decreases within an extremely short time of ∼5 min to each addition of dengue serotype 2 virus (DENV-2). We intentionally chose a low concentration of 1 mM H2O2 for the virus sensor, so the initial closed-circuit current can reach steady-state value at a short time of 7−12 min. Under this optimized condition, the opencircuit voltage and closed-circuit current outputs are in the 1354
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
Figure 4. (A) Nyquist plot showing the electrochemical impedance response of a 200 nm pore size PB-nt-embedded membrane in the presence of antibody and its specific virus target. Inset: magnified view of the Nyquist plot to reveal ohmic cell resistance. (B) Unchanged potential of a 200 nm pore size membrane sensor (immobilized with antidengue 2 antibody) in the presence of increasing virus concentration. Potentials were measured against the counter electrode under open-circuit condition. Inset: magnified view of panel A.
Model of Sensor Response. Electrolyte resistance is a key parameter that controls performance of fuel cells. In the fuel cell virus sensor, the membrane resistance Rmem is correlated to the total cross-sectional area of membrane nanochannels, A, resistivity of electrolyte solution in the nanochannels, ρ, and membrane thickness, l, as follows:35
diagnosis of the dengue disease in human patients and is superior to the “gold standard” plaque assay which requires incubation time of at least 1 week.32 In addition, the detection limit of 0.04 pfu mL−1 for dengue virus is comparable to existing state-of-the-art high-sensitivity electrically powered multiplex reverse transcriptase PCR.25−27 Membrane Resistances of the Virus Sensor. We evaluate the membrane resistances of the fuel cell virus sensor using electrochemical impedance spectroscopy (EIS) in the presence and absence of virus. Figure 4A reveals that the immobilization of ∼15 nm size antibody molecules increased the membrane resistances of 200 nm pore size membrane as expected. It shows a small increase of only ∼1 Ω in the ohmic cell resistance after incubation in the antibody solution. Subsequent addition of 8 pfu mL−1 virus particles further increased the ohmic cell resistance by 4 Ω, ∼3% of ohmic cell resistance (Figure 4A) due to the capture of virus particles (∼50 nm) within its 200 nm nanochannels. The corresponding larger decrease in the steady-state sensor current of ∼20% according to the calibration plot at Figure 3B indicates most of the resistance change occurs at the membrane instead of the solution or the porous Pt electrode. To confirm if the driving force of the fuel cell virus sensor is also affected by the virus, we measure its open-circuit voltage in increasing virus concentration against its Pt mesh counter electrode (Figure 4B). It is clear that the virus does not influence the sensor’s OCV, unlike the sigmoidal-shaped OCV−time plot of the PB-nt membrane cell when titrated with increasing H2O2 concentrations.33 Thus, the steady-state current response of the fuel cell virus sensor toward the virus does not arise from thermodynamic change. This result is significant in that the driving force of many energy conversion devices can deteriorate over time due to cathode or anode “poisoning” by solution species, causing major loss in power outputs and efficiencies over time.34 Control experiments were carried out using nonspecific analytes, silica particles (100 nm) and BSA protein (∼15 nm), which can enter the large 200 nm pores (Supporting Information Figure S-3). The current response of the virus sensor remained unchanged toward these nonvirus analytes, clearly confirming that the antibody is needed to capture the virus particles within the membrane nanochannels in order to influence its internal membrane resistance. At same time, the membrane pore size of the virus sensor needs to be larger than the virus dimension in order to generate responses toward the virus analyte.
R mem =
ρl A
(1)
Changes in the fuel cell virus sensor current response in the presence of virus concentration, [virus], are attributed to the “plugging” of the nanochannels due to virus−antibody binding. Thus, the cross-sectional pore area is reduced which increases the membrane resistance. We can assume the binding between virus and antibody in the nanochannels follows the simple Langmuir relation, so the increase in the fractional coverage θ of the bound virus within the nanochannels reduces the total cross-sectional area of the unblocked membrane nanochannels, A0, to (1 − θ)A0. The correlation between the virus sensor’s closed-circuit current and virus concentration is found to be best explained using the Langmuir−Freundlich model in which θ = ([virus]1/n)/([virus]1/n + K) where K(M) is the dissociation constant of the antibody−virus complex. This is attributed to the initial binding of virus within a single antibody-coated nanochannel which creates the largest change in channel resistance compared to subsequent virus capture within the same nanochannel. Under optimized sensing conditions where [virus] is typically in few plaque-forming units per milliliter and 1 pfu is estimated to contain ca. 100 000 virus particles,36 the virus concentration is thus ∼10−14 M. Since the value of K (in the range of 10−9−10−7 M) is much larger than [virus], the fractional coverage can be simplified to θ = ([virus]1/n)/K. In addition, the sensor’s closed-circuit current in the presence of virus, Ivirus, can be related to the membrane resistance Rmem according to Ohm’s law (V = RI). When eq 1 is combined with Ohm’s law and the simplified θ expression, a direct relation between Ivirus and virus concentration [virus] is derived as follows: I virus = 0 =
A 0V A V [virus]1/ n − 0 ρl Kρl
(2)
Nonlinear curve fitted results using eq 2 normalized to the initial closed-circuit current in the absence of virus shows excellent agreement with experiment (Figure 3B). The calculated dissociation constant K derived from nonlinear 1355
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
Figure 5. Closed-circuit current response of a dry membrane probe immobilized with antidengue type II monoclonal antibody toward (A) DENV-2 and (B) DENV-3. Background slope correction is carried out on the data presented at panel B to compensate for background drift and clearly shows the negligible effect of DENV-3 on the fuel cell virus sensor response.
■
curve fitting is 1.1 × 10−8 M and, using the pfu-to-virus particle ratio of 1:100 000, is within the reported range (10−9−10−7 M) of observed DENV-2−IgG interactions.36 Nafion-Incorporated Membrane Probe. To simplify the two-compartment cell design and three-electrode system which can be difficult to operate under field conditions, we used a Nafion perfluorinated resin incorporated dry membrane probe design. Conductive platinum layers were sputtered on both sides of the nanoporous alumina membrane with nominal pore size of 200 nm, followed by electrodeposition of PB-nt onto one side of the membrane. An amount of 15 μL of Nafion perfluorinated resin solution was applied on the platinum side of the porous PB-nt, metal-coated membrane and allowed to penetrate into the membrane nanochannels before antibody immobilization. Nafion provides the ionic conduction between the two electrodes.37 During sensor operation, the platinumcoated side of the membrane functions as the reference/ counter electrode and the PB-nt-coated side of the membrane is the working electrode (Scheme 1C). Figure 5 shows the rapid specific response of the dry membrane probe toward DENV-2 and the nonresponse toward DENV-3 virus by simple addition of 80 μL of 0.5 mM H2O2 solution, followed by 2 μL of virus solution to the PB-nt membrane side of the sensor. This demonstrates the unique utility of the nanochannel-based antibody-virus binding fuel cell virus sensor under ambient dry condition using single-step analysis without electrical power input.
■
ASSOCIATED CONTENT
S Supporting Information *
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors thank the Singapore Immunology NetworkAgency for Science, Technology and Research (SIgNA*STAR) for Research Grant SIgN09-023, Dr. Katja Fink (SIgN) for her support in dengue virus samples, and Y.W. acknowledges NTU for a Ph.D. scholarship.
■
REFERENCES
(1) Vaughn, D. W.; Green, S.; Kalayanarooj, S.; Innis, B. L.; Nimmannitya, S.; Suntayakorn, S.; Endy, T. P.; Raengsakulrach, B.; Rothman, A. L.; Ennis, F. A.; Nisalak, A. J. Infect. Dis. 2000, 181, 2−9. (2) Rothman, A. L. J. Clin. Invest. 2004, 113, 946−951. (3) Yager, P.; Edwards, T.; Fu, E.; Helton, K.; Nelson, K.; Tam, M. R.; Weigl, B. H. Nature 2006, 442, 412−418. (4) Boyer, P. D.; Lardy, H.; Myrback, K. The Enzymes; Academic Press: New York, 1963. (5) Gray, H. B. Nat. Chem. 2009, 1, 7. (6) Yamada, Y.; Fukunishi, Y.; Yamazaki, S.; Fukuzumi, S. Chem. Commun. 2010, 46, 7334−7336. (7) Kakehi, N.; Yamazaki, T.; Tsugawa, W.; Sode, K. Biosens. Bioelectron. 2007, 22, 2250−2255. (8) Katz, E.; Buckmann, A. F.; Willner, I. J. Am. Chem. Soc. 2001, 123, 10752−10753. (9) Deng, L.; Chen, C. G.; Zhou, M.; Guo, S. J.; Wang, E. K.; Dong, S. J. Anal. Chem. 2010, 82, 4283−4287. (10) Wen, D.; Deng, L.; Guo, S. J.; Dong, S. J. Anal. Chem. 2011, 83, 3968−3972. (11) Zhou, M.; Du, Y.; Chen, C. G.; Li, B. L.; Wen, D.; Dong, S. J.; Wang, E. K. J. Am. Chem. Soc. 2010, 132, 2172−2174. (12) Zebda, A.; Gondran, C.; Le Goff, A.; Holzinger, M.; Cinquin, P.; Cosnier, S. Nat. Commun. 2011, 2, 370. (13) Selvarani, G.; Prashant, S. K.; Sahu, A. K.; Sridhar, P.; Pitchumani, S.; Shukla, A. K. J. Power Sources 2008, 178, 86−91. (14) Karyakin, A. A. Electroanalysis 2001, 13, 813−819.
CONCLUSIONS
The fuel cell virus sensor comprising membrane-supported hexacyanoferrate nanotubes with antibody-loaded nanochannels offers a significantly low-cost advantage over current stateof-the-art energy storage and fuel cell sensors. In addition, the global scale of such operations during epidemic events often requires tremendous amounts of effort, cost, and time to monitor and control spread of disease and is not readily available in developing countries which have been the sources of some highly infectious diseases. Because the fuel cell virus sensor relies on redox reactions of ultrathin dry membrane structures, consequently, the reagents and miniaturized membranes can be contained within small portable units which do not require sophisticated instrumentations and are highly versatile in terms of design and suitable for deployment. 1356
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357
Analytical Chemistry
Article
(15) Karyakin, A. A.; Puganova, E. A.; Budashov, I. A.; Kurochkin, I. N.; Karyakina, E. E.; Levchenko, V. A.; Matveyenko, V. N.; Varfolomeyev, S. D. Anal. Chem. 2004, 76, 474−478. (16) Shaegh, S. A. M.; Nguyen, N. T.; Ehteshami, S. M. M.; Chan, S. H. Energy Environ. Sci. 2011, 5, 8225−8228. (17) Kim, B. Y.; Swearingen, C. B.; Ho, J. A.; Romanova, E. V.; Bohn, P. W.; Sweedler, J. V. J. Am. Chem. Soc. 2007, 129, 7620−7626. (18) Fu, J. P.; Schoch, R. B.; Stevens, A. L.; Tannenbaum, S. R.; Han, J. Y. Nat. Nanotechnol. 2007, 2, 121−128. (19) Howorka, S.; Siwy, Z. Chem. Soc. Rev. 2009, 38, 2360−2384. (20) Murray, R. W. Chem. Rev. 2008, 108, 2688−2720. (21) Amatore, C.; Da Mota, N.; Sella, C. Anal. Chem. 2008, 80, 4976−4985. (22) Koncki, R. Crit. Rev. Anal. Chem. 2002, 32, 79−96. (23) Johansson, A.; Widenkivist, E.; Lu, J.; Boman, M.; Jansson, U. Nano Lett. 2005, 5, 1603−1606. (24) Tseng, K. S.; Chen, L. C.; Ho, K. C. Sens. Actuators, B 2005, 108, 738−745. (25) Das, S.; Pingle, M. R.; Munoz-Jordan, J.; Rundell, M. S.; Rondini, S.; Granger, K.; Chang, G. J. J.; Kelly, E.; Spier, E. G.; Larone, D.; Spitzer, E.; Barany, F.; Golightly, L. M. J. Clin. Microbiol. 2008, 46, 3276−3284. (26) Johnson, B. W.; Russell, B. J.; Lanciotti, R. S. J. Clin. Microbiol. 2005, 43, 4977−4983. (27) Gurukumar, K. R.; Priyadarshini, D.; Patil, J. A.; Bhagat, A.; Singh, A.; Shah, P. S.; Celcilia, D. Virol. J. 2009, 6, 10. (28) Rajamanonmani, R.; Nkenfou, C.; Clancy, P.; Yau, Y. H.; Shochat, S. G.; Sukupolvi-Petty, S.; Schul, W.; Diamond, M. S.; Vasudevan, S. G.; Lescar, J. J. Gen. Virol. 2009, 90, 799−809. (29) Sukupolvi-Petty, S.; Austin, S. K.; Purtha, W. E.; Oliphant, T.; Nybakken, G. E.; Schlesinger, J. J.; Roehrig, J. T.; Gromowski, G. D.; Barrett, A. D.; Fremont, D. H.; Diamond, M. S. J. Virol. 2007, 81, 12816−12826. (30) Engvall, E.; Perlmann, P. J. Immunol. 1972, 109, 129−135. (31) Campbell, C. N.; de Lumley-Woodyear, T.; Heller, A. Fresenius’ J. Anal. Chem. 1999, 364, 165−169. (32) Dulbecco, R. Proc. Natl. Acad. Sci. U.S.A. 1952, 38, 747−752. (33) Wong, L. P.; Wei, Y. Y.; Toh, C. S. J. Electroanal. Chem. 2012, 671, 80−84. (34) Rodriguez, J. A.; Hrbek, J. Acc. Chem. Res. 1999, 32, 719−728. (35) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; John Wiley: New York, 2001. (36) Cheng, M. S.; Ho, J. S.; Tan, C. H.; Wong, J. P. S.; Ng, L. C.; Toh, C. S. Anal. Chim. Acta 2012, 725, 74−80. (37) Zhuo, L.; Huang, Y.; Cheng, M. S.; Lee, H. K.; Toh, C. S. Anal. Chem. 2010, 82, 4329−4332.
1357
dx.doi.org/10.1021/ac302942y | Anal. Chem. 2013, 85, 1350−1357