Label-Free Impedance Detection of Cancer Cells - Analytical

---

Article pubs.acs.org/ac

Label-Free Impedance Detection of Cancer Cells Anita Venkatanarayanan, Tia E. Keyes, and Robert J. Forster* Biomedical Diagnostics Institute, National Center for Sensor Research, School of Chemical Sciences, Dublin City University, Dublin 9, Ireland S Supporting Information *

ABSTRACT: Ovarian cancer cells, SKOV3, have been immobilized onto platinum microelectrodes using anti-EPCAM capture antibodies and detected with high sensitivity using electrochemical impedance. The change in impedance following cell capture is strongly dependent on the supporting electrolyte concentration. By controlling the concentration of Dulbecco’s phosphate buffered saline (DPBS) electrolyte, the double layer thickness can be manipulated so that the interfacial electric field interacts with the bound cells, rather than simply decaying across the antibody capture layer. Significantly, the impedance changes markedly upon cell capture over the frequency range from 3 Hz to 90 kHz. For example, using an alternating-current (ac) amplitude of 25 mV, a frequency of 81.3 kHz, and an open circuit potential (OCP) as the direct-current (dc) voltage, a detection limit of 4 captured cells was achieved. Assuming an average cell radius of 5 μm, the linear dynamic range is from 4 captured cells to 650 ± 2 captured cells, which is approximately equivalent to fractional coverages from 0.1% to 29%. An equivalent circuit that models the impedance response of the cell capture is discussed.

C

to be selectively detected.12 A key feature of the impedance of cells captured on an electrode is that different properties can be probed by changing the applied frequency. For example, the low-frequency (100 Hz) response is often sensitive to the surface coverage of the cells, the cell membrane capacitance, and cell migration effects.13 In contrast, the high-frequency (>100 kHz) response is dependent on the intracellular structures14 as well as intercellular events, such as remodeling of actin and myosin filaments during cell motion.15 Previous studies involving EIS measurements have largely focused on antibody−antigen reactions, and binding events are commonly detected using a redox probe in solution whose heterogeneous electron transfer rate constant is dependent on antibody−antigen binding.16−21 Although insightful, these approaches can alter the biological system. For example, if the insulating probe layer on the surface of the electrode is not defect-free and highly specific to the target analyte, the electrolyte solution can ingress, leading to a decreased sensitivity. In this contribution, we report on the effect of the thickness of the diffuse double layer, as controlled by the concentration of the buffer electrolyte, on the detection sensitivity of ovarian cancer cells. As illustrated in Figure 1a, the approach uses antiEPCAM antibodies immobilized on the surface of platinum electrodes to capture ovarian cancer cells (i.e., SKOV3). By

irculating tumor cells (CTCs) are rare, typically malignant, cells found in the peripheral blood circulation that originate from a primary tumor. They play a central role in cancer metastasis, which is the major cause of cancer-related deaths.1 Early detection results in 5-year survival rates of ∼90% for many cancers. In sharp contrast, if the disease has advanced to Stage III or IV, the survival rate decreases to 33%.2,3 Detecting CTCs at low concentrations represents a powerful approach to the diagnosis of the disease and can provide significant insights into the treatment efficacy.1 However, there are significant analytical challenges. For example, even when a significant primary tumor is present, there may be only 5 to 10 cancer cells per milliliter of blood (i.e., the concentration is typically of the order of 10−21 M). Moreover, the concentration of other cells is at least a factor of 105 higher, meaning that a highly selective assay is required. While optical approaches such as fluorescence or surface plasmon resonance offer high levels of sensitivity, there is a great interest in developing biosensors that allow direct labelf ree electrical detection.4−9 Biological electrochemical impedance spectroscopy (bio-EIS) has attracted significant attention, because it is direct and involves simple instrumentation, in addition to being nondestructive, label-free, and highly sensitive.10 Moreover, analytes can be sometimes detected with little or no sample preparation.11 Circulating tumor cells show significant differences in their shape, size, size distribution, and composition, as well as in the thickness and composition of cell membranes, compared to healthy cells. Therefore, their electrical properties may be sufficiently distinct to allow them © 2013 American Chemical Society

Received: October 10, 2012 Accepted: January 20, 2013 Published: January 20, 2013 2216

dx.doi.org/10.1021/ac302943q | Anal. Chem. 2013, 85, 2216−2222

Analytical Chemistry

Article

Figure 1. Schematic of (a) the electrochemical cell used for the impedance measurements with the equivalent circuit model on the right, and (b) confocal fluorescence image of the captured SKOV3 cells on the platinum electrode. The SKOV3 cells were stained with 5 μM DiOC6 and excited at 488 nm; the scale bar is 50 μm, and the cells have been imaged with a 10× objective.

potentiostat (CH Instruments, Model 760D). A Novascan PSD-UV surface decontamination unit was used to clean the surface of the platinum working electrodes. An excitation signal of 25 mV (peak amplitude) was applied to the electrodes at an open circuit potential (OCP) and the frequency of the signal was varied between 0.1 Hz and 100 kHz. Electrochemical measurements were performed using a platinum microelectrode as the working electrode with an aqueous Ag/AgCl/saturated KCl (3 M) reference electrode and a platinum wire counter. All potentials are quoted versus Ag/AgCl reference electrode, and all electrochemical measurements were made at room temperature (22 ± 2 °C). The dyed cells were imaged using a laser module confocal microscope (Zeiss, Model LSM 510 Meta) using an excitation wavelength of 488 nm and a 10× magnification objective. Microelectrode Fabrication. Platinum microelectrodes were fabricated using an established protocol.22 The in-house fabricated microelectrodes were polished manually using alumina microparticles of decreasing sizes from 1 μm to 0.05 μm and were sonicated in deionized water for 2 min prior to moving to the next size of alumina. Finally, the polished electrodes were rinsed copiously with deionized water and electrochemically cycled in 0.5 M sulfuric acid. The number of voltammetric cycles was kept to a minimum, to minimize surface roughening. Antibody Immobilization and CTC Assay. The electrodes were first treated with Novascan PSD-UV plasma for 10 min, followed by activation using sulfo NHS-EDC (5 mM)23 for 15 min and then washed thoroughly with 0.01 M DPBS. The electrodes were then incubated in 100 μg/mL antiEPCAM antibody solution, at room temperature for 1 h. The antibody-coated electrodes were washed thoroughly with DPBS and incubated in 2.5% casein solution for 30 min to block any

optimizing the detection conditions and characterizing the frequency response, a highly sensitive non-Faradaic impedance immunosensor has been developed. Moreover, the ability of an equivalent electrical circuit model to provide new insights into the mechanism of signal transduction and those factors that control analytical sensitivity are discussed.

■

EXPERIMENTAL SECTION Chemicals and Solutions. Potassium chloride (>99%), sodium fluoride (99.99%), bovine serum albumin (98%), magnesium chloride, sulfuric acid, sulfo-N-hydroxy succinimide (NHS), 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC), 2.5% casein, glutaraldehyde, and fetal bovine serum (FBS) were purchased from Sigma−Aldrich and used as received. The SKOV3 ovarian cancer cell line was purchased from ATCC and stored in liquid nitrogen. Details of the cell culture protocol are given in section S1 in the Supporting Information. Purified anti-EPCAM (CD326) antibodies were purchased from Biolegend. McCoy’s 5A (containing L-glutamine), penicillin streptomycin (pen/strep), Dulbecco’s phosphate buffered saline (DPBS) without calcium and magnesium, and Trypan Blue were all purchased from Fisher Scientific. Accutase cell dissociation buffer was purchased from Invitrogen. Experiments involving cells were carried out in a certified Bio safety Level II laboratory under sterile, laminar air flow conditions. Cells were fixed with 2.5% (w/v) glutaraldehyde in 0.1 M sodium cacodylate buffer. In addition, 0.01 M DPBS solution and all aqueous solutions were prepared using Milli-Q water (18 MΩ cm−1).

■

METHODS Impedance spectroscopy was performed using a conventional three-electrode cell placed inside a Faraday cage and a 2217

dx.doi.org/10.1021/ac302943q | Anal. Chem. 2013, 85, 2216−2222

Analytical Chemistry

Article

Figure 2. Confocal fluorescence image of the platinum working electrode following exposure to different concentrations of SKOV3 cells in suspension. From the upper left-hand side, going across: 653, 637, 438, and 386 SKOV3 cells have been captured. From the lower left-hand side, going across: 355, 28, and 4 SKOV3 cells have been captured (the image in the lower right-hand corner shows a negative control where the antiEPCAM antibody capture layer is not present but the unmodified electrode has been exposed to a highly concentrated suspension of the SKOV3 cells). The cells have been fixed with glutaraldehyde and stained with 5 μM DiOC6, and the excitation wavelength was 488 nm; the scale bar is 200 μm, and the cells were imaged with a 10× objective.

its concentration in the deposition solution, and the associated impedance changes were then measured as a function of the electrolyte concentration. Nyquist plots (see Figure S2 in the Supporting Information) all show a linear region arising from the double layer capacitance in series with the solution resistance. Significantly, when the electrolyte concentration is high (0.1 M), the shape of the Nyquist response is not strongly dependent on the antibody concentration. However, for frequencies below 10 Hz, antibody immobilization produces a distinct change in the total impedance and the phase angle (see Figures S3(a) and S3(b) in the Supporting Information). In dilute electrolyte (0.001 M), both the cell resistance (intersection of the right-hand side of the Nyquist semicircle with the x-axis) and the capacitance are strongly dependent on the antibody concentration (surface coverage) (see Figure S2(b) in the Supporting Information. The greatest difference in impedance and phase angle between different antibody concentrations is seen at ∼200 Hz. The measurements in Figure S2 in the Supporting Information were conducted with a three-electrode system, where the reference electrode contributes significantly to the uncompensated resistance, causing a large semicircle to be observed at high frequency. This resistance decreases the range available for binding induced changes and should be minimized. Adopting a two-electrode system significantly reduces the overall cell resistance, and the dynamic range and limit of detection were determined using the low electrolyte concentration and a two-electrode system. Dynamic Range and Limit of Detection. The electrodes were first modified with anti-EPCAM, using a 100 μg/mL deposition solution for 1 h, followed by blocking with 2.5% casein for 30 min. They were then exposed to SKOV3 cells for 3 h at 37 °C at concentrations of 1 × 106−50 × 106 cells per mL. Following capture, the electrodes were exposed to 5 μM DiOC6 for 10 min and then thoroughly washed with DPBS buffer. Figure 2 shows confocal fluorescence images of the antiEPCAM-modified electrodes after exposure to the SKOV3 cells. These images clearly show the captured DiOC6 labeled cells and reveal that the number of cells captured is dependent on their concentration in suspension. Significantly, for this concentration range, the number of captured cells ranges from

bare electrode surface. The blocked electrodes were then washed thoroughly with DPBS before being exposed to different concentrations of SKOV3 cells suspended in DPBS in an incubator for 3 h at 37 °C. After 3 h, the electrodes were thoroughly washed with DPBS to remove any physisorbed cells and the change in impedance measured. In order to visualize the captured cells, they were stained by exposing them to 5 μM solution of 3,3′-dihexyloxacarbocyanine iodide (DiOC6, λmax = 484 nm), an endoplasmic reticulum staining cellular dye for 10 min, in darkness, at room temperature. The electrodes were then washed copiously with DPBS buffer and observed using confocal microscope in order to determine the total number of cells captured.

■

RESULTS AND DISCUSSION Electrochemical Characterization. The adsorption of cells, or biomolecules such as antibodies, typically causes the electrode capacitance and overall cell resistance to decrease and increase, respectively. However, the maximum sensitivity will be triggered by binding events that occur within the diffuse double layer whose thickness can be controlled by changing the ionic strength of the buffer electrolyte.24 Thus, if the electrolyte concentration is high, the double layer is compressed and the capacitance is dominated by changes occurring close to the electrode surface, e.g., within an antibody-based cell capture layer. However, if the electrolyte concentration is reduced, the sensitivity for the detection of larger entities, such as cells, should be increased. For example, in pure aqueous electrolyte systems, where the electrode is not modified with a relatively low dielectric organic material, the double layer thicknesses where the electrolyte concentrations are 0.1 and 0.001 M are ∼0.1 nm and ∼10 nm, respectively. Here, there is a large difference in size between the antibody capture layer (3 ± 2 nm thick) and the SKOV3 cells (diameter of ∼10 μm). Therefore, the detection sensitivity for cells should be higher at low electrolyte concentration. When the deposition time is fixed, the surface coverage of the antibody is expected to be dependent on its bulk concentration. The surface coverage can change both the overall cell resistance and the capacitance, since adsorption will involve at least some displacement of water and ions from the interface. The antiEPCAM surface coverage was systematically varied by changing 2218

dx.doi.org/10.1021/ac302943q | Anal. Chem. 2013, 85, 2216−2222

Analytical Chemistry

Article

4 to ∼650, giving a wide range over which to probe the impedance. Cyclic voltammetry of an electroactive probe in solution, e.g., ferrocenemethanol (FeMeOH), can provide significant insights into the number of cells captured as well as the barrier properties of the absorbed cells. Figure 3 shows cyclic

ferrocene in the presence of captured cells. These observations are consistent with the immobilized SKOV3 cells, reducing the rate of heterogeneous electron transfer. For example, as shown in Figure 3, the anodic branch of the voltammogram can be satisfactorily fit if the rate of heterogeneous electron transfer is reduced by a factor of 20 (0.001 cm s−1) when there is a significant population of captured cells, compared to the situation where the electrode is modified with anti-EpCAM but has no captured cells. However, the voltammetry is electrochemically irreversible and a well-defined reduction process is not observed. If slow heterogeneous electron transfer alone is responsible for not observing a reduction process, then the rate for ferrocene reduction would have to be at least a factor of 100 slower than its oxidation. If the adsorbed cells simply created a physical barrier to electron transfer, then this asymmetric response would not be expected and it is possible that cell capture creates significant electrostatic repulsion of the ferrocenium from the electrode surface. Figure 3 also shows that the interfacial capacitance decreases by ∼50% following cell capture. These changes in both the interfacial capacitance and charge transfer resistance indicate that it should be possible to detect cell capture by measuring the impedance response. CTC Impedance Assay. The impedance response was measured as the concentration of cells in suspension was systematically varied. Confocal microscopy was used to count the absolute number of cells captured on the electrode surface after exposure to a given concentration in suspension. This “absolute counting” approach allows the correlation between the number of cells captured and the relative change in impedance to be determined. Moreover, it avoids any difficulties caused by the nonlinear dependence of the number of captured cells on the suspension concentration. Figure 4 shows a log−log plot of the dependence of the impedance on the frequency as well as the dependence of the change in phase angle on frequency as the coverage of captured cells is systematically varied. These magnitude and phase data give distinct insights into the capacitive and resistive effects of cell capture. For all cell coverages, an increase in frequency causes the overall magnitude of the impedance to decrease. Equivalent Circuit. Figure 4b shows that the change in phase angle, Δϕ (eq 2), also shows two distinct peaks for all cell coverages investigated. For 4 captured cells, these peaks are observed at 1.3 × 105 Hz and ∼56 Hz. With an increase in the number of captured cells, the magnitude of Δϕ of the peak at ∼56 Hz decreases and shifts significantly toward lower frequencies. Significantly, at 1.3 × 105 Hz, no shift in the peak frequency is observed with the change in cell numbers; however, the amplitude of phase change decreases as the captured cell numbers decrease. Because the cell membrane is highly insulating, the current flowing from the region of the electrode that is covered with cells must flow laterally through the resistive junction region between the cell and the electrode, i.e., through the antibody capture layer. Therefore, when cells are captured, there are two branches for the current flow: the exposed electrode portion and the cell-covered portion. At low frequencies and for low surface coverages of captured cells, the impedance associated with the cell covered region is large, compared to the effective cell−electrode junction resistance. Thus, the impedance is essentially unaffected by the cell capture and it decreases as the frequency increases. As the frequency increases, the impedance associated with the cell-modified region becomes less than cell−electrode junction resistance and the measured impedance is approximately equal to the junction

Figure 3. Cyclic voltammogram of 1 mM FeMeOH in 0.01 M DPBS buffer at an unmodified microelectrode (thin black line, no symbols), following antibody immobilization using a 100 μg/mL anti-EPCAM solution (black line, solid diamonds) and following the capture of 355 SKOV3 cells (black line, open circles). The scan rate is 100 mV s−1. The symbols represent the least-squares fit of the experimental response (solid line) using the Butler−Volmer formulation of electrode kinetics under semi-infinite linear diffusion control. For the anti-EPCAM coated electrode, the standard heterogeneous electron transfer rate constant (k°) is equal to 0.02 cm s−1, whereas the k° value following cell capture is 0.001 cm s−1. The working electrode was 250 μm radius platinum, with Ag/AgCl reference and a platinum counter electrode. The inset shows the confocal fluorescence image of the corresponding electrode, where the SKOV3 cells have been stained with 5 μM DiOC6, excited at 488 nm, and imaged with a 10× objective. (Scale bar = 200 μm.)

voltammograms for a 1 mM solution of FeMeOH at an unmodified platinum electrode (thin black line, no symbols) and following modification with an anti-EPCAM capture layer (black line, solid diamonds). Far from the formal potential, the current is dominated by double layer charging and the difference in current between the forward and backward wave is dependent on the interfacial capacitance. Significantly, formation of the anti-EPCAM layer decreases the capacitance by only ∼15%. Moreover, at this scan rate, the peak-to-peak separation for the oxidation of the ferrocene methanol does not change significantly, following antibody immobilization. As shown in Figure 3, the best fit voltammogram using the Butler−Volmer formulation of electrode kinetics, assuming semi-infinite linear diffusion control, yields a standard heterogeneous electron transfer rate constant (k°) of 0.02 cm s−1. The insensitivity of both the interfacial capacitance and k° to antibody immobilization indicates that the capture layer is not compact and does not present a significant barrier to heterogeneous electron transfer to the redox probe in solution. The inset of Figure 3 shows the confocal fluorescence image of the electrode following SKOV3 capture and indicates that, for this concentration of cells in suspension, the electrode is not fully covered: there is a significant population of captured cells. Significantly, SKOV3 capture changes the voltammetric response markedly (black line, open circles). For example, the anodic peak potential (Epa) shifts in a positive potential direction by ∼90 mV, indicating that it is harder to oxidize 2219

dx.doi.org/10.1021/ac302943q | Anal. Chem. 2013, 85, 2216−2222

Analytical Chemistry

Article

Figure 5. Bode plots showing experimental values for log Z as a function of frequency for 100 μg/mL anti-EPCAM antibody (○) and 608 SKOV3 cells (◇) and simulated values for 100 μg/mL antiEPCAM antibody (dashed line, - - -) and 608 SKOV3 cells (solid line, ) immobilized on the platinum working electrode in 10−3 M DPBS buffer electrolyte. The frequency range was 0.1 Hz to 100 MHz, the excitation signal with a peak amplitude of 25 mV was applied at OCP, and a platinum wire counter electrode was included in the assembly.

⎛1⎞ ZQ = ⎜ ⎟(jω)−n ⎝ Y0 ⎠

where ω = n2πf (0 < n < 1), and it behaves like a pure capacitor or resistor when n approaches 1 or 0, respectively.25 The value obtained during data fitting (n = 0.98) suggests that the CPE behaves very similar to a pure capacitor. Upon CTC capture, another constant phase element (CPECell, with n = 0.88) is needed to model the experimental response. The best-fit values for the resistances and capacitances are given in Table 1. It is clear from data fitting that the equivalent circuits closely model the electrical response of the platinum electrode/solution interface with SKOV3 cells captured by the anti-EPCAM antibody. At high frequencies, the impedance is dominated by the spreading resistance. Since a portion of the electrode area is blocked by the cells at high frequencies, the spreading resistance is increased slightly. As a result, plots of the normalized impedance change will show a peak with an amplitude that is dependent on cell coverage and cell− electrode gap, with the peak frequency depending on the size of contiguous cell aggregates (see Figure S4 in the Supporting Information). Significantly, at the higher frequency range (130 kHz), while no shift in the peak frequency is observed, the magnitude of phase change linearly decreases as the number of cells decrease. The relative impedance change at 81.3 × 103 Hz and difference in phase angle at the two frequencies (i.e., 2.09 × 104 and 1.3 × 105 Hz), as a function of the number of cells, are given in Table 2. The relative impedance (ZREL) and difference in phase angle (Δϕ) are given by eqs 1 and 2, respectively:4,11,27,28

Figure 4. (a) Log−log plot showing the dependence of the impedance on the applied frequency of the eight platinum working electrodes shown in Figure 2. At a frequency of 81.3 kHz, the traces, from top to bottom, represent 4, 355, 386, 438, 547, 637, and 653 SKOV3 cells, as well as a control where the anti-EPCAM antibody capture layer is omitted and the unmodified electrode has been exposed to a highly concentrated suspension of the cells, respectively. (b) Phase change as a function of logarithm of applied frequency for electrodes coated with anti-EPCAM antibody (black dashed line, - - -), 4 (black solid line, ), 355 (gray squares, ■), 386 (gray circles, ●), and 653 (gray triangles, ▲) captured cells.

resistance. However, at a critical frequency, the impedance of the uncovered electrode becomes smaller than junction resistance and the measured impedance is dominated by the uncovered region of the electrode. Therefore, the impedance measured at moderate frequencies increases substantially as the coverage of cells increases. Figure 5 illustrates the equivalent electrical circuit model that seeks to describe the anti-EPCAM-modified electrode following cell capture. The constant magnitude observed in the frequency range from 50 Hz to 100 kHz represents a resistive component and it arises from the captured SKOV3 cells. Beyond 100 kHz, a second resistive component (i.e., the resistance of electrolyte solution) is observed. The captured cells are represented by a constant phase element (CPECell) in series. The impedance of the electrode/solution interface comprises a solution resistance (RSol) in series with diffuse layer capacitance (CE).25 Ideally, an antibody-coated electrode is expected have another capacitance in series with the diffuse layer capacitance.26 We observe that a constant phase element (CPEAB) with n = 0.98 in parallel with a resistor provides an excellent fit. The constant phase element is defined as

Z REL = ABS

ZAbCoated − ZCell ZAbCoated − ZSaturated

ΔΦ = ABS(ΦBare electrode − ΦCoated electrode)

(1) (2)

where ZAbCoated and ZCell are the impedances observed for the anti-EPCAM-coated platinum microelectrode and following exposure of these anti-EPCAM-modified electrodes to SKOV3 cells of different concentrations, respectively. ZSaturated is the impedance observed for the highest surface coverage of captured cells achieved. ϕBare electrode and ϕCoated electrode are the 2220

dx.doi.org/10.1021/ac302943q | Anal. Chem. 2013, 85, 2216−2222

Analytical Chemistry

Article

Table 1. Equivalent Circuit Model Parameters DPBS Anti-EPCAM cells

concentration

RSOL [Ω]

1 mM 100 μg/mL 608 cells

761.1 750 180

CPEAB [F] 1.05 × 10−10 4.8 × 10−7

Table 2. Relative Impedance Change at 81.3 × 105 Hz and Δϕ as a Function of Cell Number at 1.30 × 105 Hz and 2.09 × 104 Hz for 8 Different Platinum Working Electrodes with 0, 4, 355, 386, 438, 547, 637, and 653 Captured SKOV3 Cellsa number of cells captured

ZREL @ 81.3 × 10 Hz

0 4 355 386 438 547 637 653

0 7.33 ± 0.06 4.06 ± 0.06 3.86 ± 0.03 3.25 ± 0.01 1.25 ± 0.05 0.20 ± 0.01 0.001 ± 0.03

@ 1.30 × 105 Hz 63.03 50.69 51.00 51.96 52.20 57.29 53.72 56.02

± ± ± ± ± ± ± ±

0.03 0.2 0.2 0.3 0.1 0.5 0.1 0.1

@ 2.09 × 104 Hz 17.59 15.48 15.28 14.64 15.16 21.25 14.36 17.86

± ± ± ± ± ± ± ±

CE [F]

RAB [Ω]

3.05 × 10−10

8.8 × 10−8 8.8 × 10−8 9.8 × 10−6

1.36 × 105 8.9 × 103

capture of bacterial cells by the E. coli-specific polyclonal antibody could be detected using EIS at low frequency (1 kHz).29 However, the limit of detection was ∼10 bacterial cells. Significantly, our results suggest that, by decreasing the ionic strength of the electrolyte so that a greater fraction of captured cell sits within the electrochemical double layer, and by optimizing the ratio of the electrode size to the target cells, as few as 4 captured cells can be detected. Significantly, control experiments in which a freshly polished platinum microelectrode is exposed to a suspension of SKOV3 cells containing 1 × 106 cells/mL, no captured cells are observed in confocal microscopy and the impedance is indistinguishable from the bare platinum electrode (see Figure S5 in the Supporting Information). These observations confirm that the impedance changes observed in Figure 4 arise from specific binding of SKOV3 cells to anti-EPCAM antibodies; they are not due to the physisorption of other components that may be present in the cell suspension.

ΔPhase 3

CPECELL [F]

0.03 0.2 0.2 0.3 0.1 0.5 0.1 0.1

a

Electrochemical impedance spectroscopy was performed using a 250μm platinum working electrode and a platinum counter electrode in 10−3 M DPBS buffer electrolyte. The frequency range was 0.1 Hz to 100 MHz. An excitation signal with a peak amplitude of 25 mV was applied at the open circuit potential (OCP).

■

CONCLUSIONS A highly sensitive electrochemical impedance sensor has been developed for the detection of the ovarian cancer cell, SKOV3, captured onto the electrode surface through their immunological interaction with surface immobilized anti-EPCAM antibody. Significantly, the detection sensitivity is markedly dependent on the ionic strength of the electrolyte and surface coverage of the captured cells. Moreover, the frequency at which the change in electrical impedance is determined influences the detection sensitivity. Electrochemical impedance spectroscopy (EIS) conducted at open circuit potential (OCP) shows great promise as a sensitive, label-free detection strategy for rare circulating tumor cells (CTCs). Moreover, because the detection is label-free and nondestructive, the cells could be released for expression profiling, thus providing deep insights into the phenotype of CTCs. Significantly, four captured cells can be detected, representing a surface coverage of