Microfluidic Chip with Integrated Electrical Cell-Impedance Sensing for

Oct 13, 2013 - 3D-Printed Miniaturized Fluidic Tools in Chemistry and Biology ... Microfluidic Impedance Biosensors for Monitoring a Single and Multip...
2 downloads 0 Views 5MB Size
Article pubs.acs.org/ac

Microfluidic Chip with Integrated Electrical Cell-Impedance Sensing for Monitoring Single Cancer Cell Migration in Three-Dimensional Matrixes Tien Anh Nguyen,*,† Tsung-I Yin,† Diego Reyes, and Gerald A. Urban Department of Microsystems Engineering, IMTEK, University of Freiburg, Georges-Koehler Allee 103, 79110 Freiburg, Germany ABSTRACT: Cell migration has been recognized as one hallmark of malignant tumor progression. By integrating the method of electrical cell−substrate impedance sensing (ECIS) with the Boyden chamber design, the state-of-the-art techniques provide kinetic information about cell migration and invasion processes in three-dimensional (3D) extracellular matrixes. However, the information related to the initial stage of cell migration with single-cell resolution, which plays a unique role in the metastasis−invasion cascade of cancer, is not yet available. In this paper, we present a microfluidic device integrated with ECIS for investigating single cancer cell migration in 3D matrixes. Using microfluidics techniques without the requirement of physical connections to off-chip pneumatics, the proposed sensor chip can efficiently capture single cells on microelectrode arrays for sequential on-chip 2D or 3D cell culture and impedance measurement. An on-chip single-cell migration assay was successfully demonstrated within several minutes. Migration of single metastatic MDA-MB-231 cells in their initial stage can be monitored in real time; it shows a rapid change in impedance magnitude of approximately 10 Ω/s, whereas no prominent impedance change is observed for lessmetastasis MCF-7 cells. The proposed sensor chip, allowing for a rapid and selective detection of the migratory properties of cancer cells at the single-cell level, could be applied as a new tool for cancer research.

E

time cell analyzer (RTCA) instrument could provide kinetic information about the cell migration and invasion process in a 3D microenvironment.12 However, the xCELLigence detects the electrical impedance change only when the seeded cells invade and migrate through the 3D matrixes to contact the microporous membrane coated with a microelectrode; the information related to cell migration in the initial stage, which may shed light on the invasion−metastasis cascade of cancer,8 is therefore unavailable. In addition, because of the limited capability to control the cell adhesion process, a large cell population is usually used in those ECIS-based devices for seeding cells randomly on top of the electrodes. Single-cell analysis, which can provide unique insight into the heterogeneous behaviors of cells,13 is otherwise concealed within the averaged response of a cell population. Although various papers focus on cell-impedance measurement at the single-cell level providing a new strategy for single-cell analysis,14−19 none of them have been applied to the 3D cases. Owing to the unique role of cell migration at the single-cell level in the metastasis−invasion cascade,20 an ECIS-based device that allows for a rapid and selective detection of the

lectrical cell−substrate impedance sensing (ECIS) is a valuable method for investigating various cellular events such as attachment, adhesion, growth, and motility through the monitoring of electrical alternations at the interfaces between the cell and electrode in a real-time, label-free, and nondestructive manner.1−3 Recently, the impedance-based sensing technology has gained a great deal of attention for studying cancer cells and monitoring drug-induced cellular events for drug discovery.4−7 One of the research interests is to investigate the migration tendency of cancer cells, which has been recognized as one hallmark of malignant tumor progression.8 The current ECIS-based devices, however, are limited to the application for two-dimensional (2D) culture system, where cell cultures are often performed on synthetic substrates such as glass or plastic that forces cells to adjust themselves to flat and rigid surfaces. The discrepancies between the behaviors of cells in cultures and in vivo have led researchers to switch to threedimensional (3D) cell culture models, which better represent the microenvironment of living tissues.9,10 Conventional methodologies for evaluating cell migration in 3D extracellular matrixes (ECM), such as the Boyden chamber (also called trans-well migration assay), are end-point and labelbased assays.11 The semiquantitative data collection and endpoint determination make visualization of the migration process impossible. By integrating the impedance sensing method with the design of a Boyden chamber, the Roche xCELLigence real© 2013 American Chemical Society

Received: August 30, 2013 Accepted: October 13, 2013 Published: October 13, 2013 11068

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 1. Design of the sensor chip. (a) 3D image of the sensor chip, which consists of three main parts: microelectrode arrays (MEAs), cell capture arrays (CCAs), and a microfluidic channel with an inlet and an outlet. (b) Cross-sectional view of the MFC with the mechanism of passive pumping. (c) A V-shaped structure for single cell trapping.

Figure 2. Schematic diagram of the microfabrication processes. (a) Deposition and pattern of metal layers (Cr/Au/Ti: 10 nm/100 nm/100 nm) on a Pyrex wafer. (b) Deposition and pattern of a passivation layer (SiOx) on top of the metal layers. (c) Pattern of CCAs and a part of the channel wall using SU-8 negative photoresist. (d) Adhering of a PDMS layer to the SU-8 structure for producing the MFC.



migratory properties of single cancer cells in 3D matrixes would serve as a new tool for cancer research. Here we report a microfluidic device integrated with ECIS for investigating single cancer cell migration in 3D matrixes. This device possesses a simple but effective cell-trapping mechanism to position single cells on top of a microelectrodes array. A microfluidic assay combining a 2D and 3D cell culture system can be performed on the chip and monitored in real time using impedance measurement. Electrical equivalent circuits are proposed to model the cell−substrate heterostructure in 2D and 3D microenvironments. Finally, a real-time monitoring of cell migration in 3D matrixes is demonstrated.

EXPERIMENTAL SECTION

Design of the Sensor Chip. Figure 1a depicts the design of the sensor chip, which consists out of three main parts: microelectrode arrays (MEAs), cell capture arrays (CCAs), and a microfluidic channel. By injecting cell suspension into the channel, single cells can be hydrodynamically trapped by the CCAs.15,17 A 2D or 3D microenvironment can be built up inside the microfluidic channel for monitoring various cellular activities such as the cell adhesion and cell migration (upward) in real time by using the MEAs. To provide a pipet-friendly design for the use in traditional cell biology, a passive pumping method was integrated with the proposed sensor chip.21 Briefly, by placing two droplets with 11069

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 3. Experimental protocols. (a−c) The protocol for the 2D cell experiment: (a) the microelectrode surface is modified using an ethanol solution with 95% 11-mercaptoundecanoic acid (11-MUA) to form a self-assembly monolayer (SAM) on the microelectrodes surface; (b) fibronectin (FN) solution is injected into the channel using a micropipet for FN adsorption onto the SAM-modified substrate for at least 1 h; (c) cell suspension is pumped into the channel for single cell trapping using passive pumping method. (d−f) The protocol for the 3D cell experiment: (d) cold Matrigel solution (2.6 mg/mL, at 4 °C) is passive-pumped into the channel to replace the cell culture medium; (e) the PDMS cover is peeledoff, and the chip is incubated for another 15 min for the polymerization of the Matrigel; (f) a porous membrane is adhered on a tape (thickness ∼100 μm) next to the SU-8 structure to produce a thin layer of Matrigel. A drop of cell culture medium with 10% FBS is placed on top of the membrane for generating a chemoattractant gradient.

manually aligned with the SU-8 channel wall forming the microfluidic channel (Figure 2d). Cell Culture and Cell Seeding. Metastatic MDA-MB-231 and less-metastatic MCF-7 were purchased from ATCC (Manassas, VA, U.S.A.). They were cultivated in Gibco Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Life Technologies GmbH, Darmstadt, Germany) under standard conditions (37 °C, 5% CO2). Subcultivation was done in a 1:10 ratio. Cells were detached from culture flasks by treatment with trypsin−EDTA (Life Technologies GmbH, Darmstadt, Germany) for 2 min. After detachment, they were resuspended in the DMEM to stop any remaining trypsin activity. After centrifugation for 10 min, they were suspended in the Gibco CO2 independent medium supplemented with 4 mM L -glutamine (Life Technologies GmbH, Darmstadt, Germany) to 106 cells/mL. Experimental Protocol. Figure 3 illustrates the experimental protocol. Each sensor chip was cleaned in acetone with ultrasound for 5 min, followed by isopropyl alcohol and a DI water rinse. The sensor chips were glued on PC boards for wire bonding. To ensure the cleanness of the Au electrode’s surface, the wire-bonded chips were further cleaned using oxygen plasma for 10 min. A self-assembly monolayer (SAM) was then formed on the Au electrodes by immersing the chips in an ethanol solution with 95% 11-mercaptoundecanoic acid (11MUA) (Sigma-Aldrich Co. LLC., Germany) at room temperature for 15 h. After removing the chips from the ethanol solution, each chip was cleaned with ethanol 70% for 30 s and then dried carefully using nitrogen flow. The PDMS layers were diced and cleaned with isopropyl alcohol and manually aligned to the SU-8 structure of the sensor chips to form the microchannel with a height of 40 μm (Figure 3a). To provide better surface properties for cell adhesion, fibronectin (FN) (Sigma-Aldrich Co. LLC., Germany) in phosphate-buffered saline (PBS) (pH = 7.4, 1 mg/mL) was injected into the microchannels using a micropipet for FN adsorption onto the SAM-modified Au substrate (Figure 3b).23 The sensor chips were then incubated at room temperature for at least 1 h. Before the step of cell trapping, the microchannel was cleaned by PBS using the passive pumping method. The droplets of PBS with FN on the inlet and outlet of the microchannel were aspirated completely, and a droplet of fresh PBS was placed on the inlet to induce a dynamic flow inside the

different sizes and curvatures on the inlet and the outlet of a solution-filled channel, the different Young−Laplace pressure inside the two liquid droplets (ΔP = 2γ/R, where γ is the surface tension of the fluid and R is the radius of the droplet) induces a dynamic flow inside the microfluidic channel.21 The cells suspended in the droplet with a large curvature can therefore be conveyed into the channel and be trapped by the CCAs (Figure 1b). MEAs are composed of four columns of small electrodes (25 × 25 or 30 × 30 μm2) distributed symmetrically on the opposite side of a large counter electrode (350 × 500 μm2) in the middle. Depending on the flow direction in the cell experiment, the small microelectrodes on either side of the counter electrode are served as the working electrodes (with trapped single cells) and references (without cell), respectively. Each cell capture structure in CCAs is composed of two blocks with the same dimensions to form a V-shaped trap (Figure 1c). There is a 5 μm slot between the two blocks to allow the hydrodynamical flow to pass through it while preventing the escape of the trapped cells. Each V-shaped trap is aligned with a small electrode. Fabrication Process of the Sensor Chip. Figure 2 describes the schematic diagram of the microfabrication process for the proposed sensor chip. Metal films, Cr/Au/Ti (10 nm/ 100 nm/100 nm), were deposited on a Pyrex wafer using an electron-beam evaporator. Cr was used as an adhesion layer, and Ti was used as a barrier layer for the following dry etching process. The MEAs, bonding pads, and connecting lines were then patterned by a lift-off process using a reverse photoresist (AZ 5214, 1.4 μm) (Figure 2a). For producing a passivation layer, SiOx (500 nm) was deposited using plasma-enhanced chemical vapor deposition (PECVD) at 300 °C. A photolithography and dry etching process using reactive ion etching (RIE) were subsequently performed to expose the MEAs and the bonding pad regions. The Ti layer on top of the Au layer was removed by dipping the wafers in 1% HF (Figure 2b). Then, a negative photoresist SU-8 (3025, 20 μm thick) was spin-coated and patterned to form the CCAs and a part of the microfluidic channel walls (Figure 2c). A poly(dimethylsiloxane) (PDMS) layer with 20 μm channel height was fabricated using standard micromolding process.22 Two holes with approximately 1 mm in diameter were punched through the PDMS layer for producing the inlet and the outlet of the microfluidic channel. Finally, the PDMS layer was 11070

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 4. (a) Photo image of a packaged sensor chip with a magnified image of the microchannel, where single cells are trapped on top of the working electrodes. (b) Simulation result of flow field in the microchannel. (c) Distribution of flow rate along the dash line AB. The magnitude of the flow rate at the inlet of the channel is assigned 2 mm/s.

used to analyze the acquired data. For impedance spectrum measurements, the Solartron 1260 (machine, equipment) delivered an alternating voltage with a 10 mV amplitude for the frequency range between 100 and 106 Hz. For the real-time measurement, the Solartron 1260 was set to deliver an alternating voltage of 10 mV at 4 kHz.

channel. The same procedure was repeated at least twice. Then, a droplet of CO2 independent medium (0.5−1 μL) with suspended MDA-MB-231 or MCF-7 (106 cells/mL) was placed on the channel inlet. The suspended cells can then be conveyed into the microchannel for single cell trapping at the CCAs (Figure 3c). After trapping the cells, a big drop of cell culture medium was added on top of the whole PDMS layer to stop the flow inside the microchannel. The sensor chips were then placed inside an incubator for on-chip cell culture. Figure 3, parts c and d, describes the protocol for the cell experiment in Matrigel (BD Biosciences, Germany). By using the same pumping method in Figure 3, parts b and c, cold Matrigel solution (2.6 mg/mL, at 4 °C) was pumped into the microchannel to replace the cell medium while covering the attached cells on top of the microelectrodes (Figure 3d). Then the PDMS layer was carefully removed and a droplet of Matrigel was added to fully cover the whole microchannel (Figure 3e). The sensor chip was then incubated inside an incubator at 37 °C for another 15 min. After the polymerization of the Matrigel, a porous membrane with 0.2 μm pore size (Nunc GmbH & Co. KG, Germany) was carefully attached to a tape with a thickness of 100 μm next to the SU-8 structure. A thin layer of Matrigel was therefore formed between the membrane and the substrate. Finally, a big drop of CO2 independent medium with 10% FBS was placed on top of the membrane to generate a chemoattractant gradient along the thin layer of Matrigel (Figure 3f). Impedance Measurement. Electrical impedance spectroscopy measurements were performed using a Solartron 1260 impedance analyzer (Solartron Analytical, U.K.). The instrument was connected to a computer through a GPIB card and controlled with the Zplot software. The Zview software was



RESULTS AND DISCUSSION

Cell Trapping by CCAs. Figure 4a shows a photo image of a packaged sensor chip (0.65 × 1.1 cm2) with a magnified image of the microfluidic channel, where the single cells were trapped on top of the MEAs for working electrode. The arrangement of CCAs on the opposite side of the counter electrode can prevent the cells from occupying on the MEAs for reference (the columns on the right side of the counter electrode). To estimate the distribution of flow velocity in the CCAs regions, a 3D model which imitates the real structures is simulated with COMSOL 4.3 as shown in Figure 4b. The magnitude of the flow rate around the cell trapping region (along the red dash line AB) is plotted in Figure 4c. A laminar flow with constant velocity (2 mm/s) at the inlet is assumed by the model. The value of that flow rate is assigned based on a previous work,24 which has similar parameters, such as the diameter of the inlet (1−1.2 mm) and the volume of the suspended cell droplet (0.5−1 μL). The geometry of the Vshape produces a slow and uniform flow field at the cell capture regions (Figure 4b). The flow rate in the center of the V-shaped region is much smaller compared with that at the outside region, which is 0.5 and 4 mm/s, respectively, as shown in Figure 4c. Depending on the flow rate and the density of cells in suspension, the trapping efficiency of the CCAs design of the 11071

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 5. (a) Simplified circuit model and the setup for ECIS measurement. (b) The Bode impedance spectra measured on two working electrode (w/t or w/o surface modification) with single cells and a reference electrode without cells. An equivalent electrical circuit model is built up to fit the measurement data (measured or acquired spectra). In the circuit model, Rsol refers to the solution resistance and Rseal refers to the cleft between cell membrane and electrode surface. The electrode−electrolyte interface is modeled by a parallel circuit consisting of a constant phase element ZCPE and a charge-transfer resistance Rct. The single cell is modeled by a parallel circuit Rm and Cm.

Table 1. Simulated Values of the Electrical Elements in the Equivalent Circuit Model for Characterization of the Cell−Electrode Interaction in a 2D Microenvironment bare electrode electrode with surface modification

Rsol [Ω]

Rseal [Ω]

Rm [Ω]

Cm [F]

Q

n

Rct [Ω]

1422 1215

3.57 × 107 3.82 × 107

4.24 × 107 4.45 × 107

6.80 × 10−10 5.84 × 10−10

3.40 × 10−7 4.67 × 10−7

0.48 0.60

5.03 × 106 7.84 × 106

constant (0 ≤ n ≤ 1). The ideal capacitor is usually superseded by the constant phase element due to the inhomogeneities on the surface of the electrode.26 In the proposed sensor chip design, the surface interfacial impedance of the counter electrode is negligible because of the large ratio of surface area compared with that of the working electrodes.27 The cell impedance Zcell can be simply modeled by a membrane capacitance Cm in parallel with a membrane resistance Rm.4,28 In addition, a cleft between the cell membrane and the electrode surface is formed due to the finite binding force and the presence of the cell membrane proteins (e.g., integrins), keeping the cell membrane at a certain distance from the surface.29 This electrolyte-filled cleft constitutes a shunt path with a “seal resistance” Rseal bypassing the cell on the electrode.28,30 The seal resistance can be expressed as

proposed sensor chip can be up to 100% (eight positions of MEAs for a single experiment). Characterization of the Cell−Electrode Interaction for a 2D Cell Culture System. By culturing cancer cells on the MEAs surface, cell-induced impedance signals’ change can reflect the characteristics of the cells when they adhere and spread on the electrodes.1 Figure 5a shows the setup of the impedance measurement and an electrical equivalent circuit to model the impedance of cell−electrode heterostructure. Figure 5b shows the measured impedance spectra on a reference electrode without cells and on working electrodes (w/t or w/o surface modification, squares and stars in legends, respectively) induced by single MDA-MB-231 cells after cell culture for 2 h. It is observed that an increase of 52% (at the frequency of 10 kHz) in the impedance magnitude is obtained from the working microelectrode with the surface modification compared with the bare Au electrode, which corresponds to a previous work using a similar surface modification method for improving the sensitivity of the impedance measurement.25 To determine the cellular impedance values from the measured data, a circuit model of the cell−electrode heterostructure is constructed. As shown in Figure 5b, the electrode−electrolyte interface is modeled by a typical circuit which has been widely applied to biomedical applications.26 It comprises the resistance of the cell culture medium Rsol in series with a parallel circuit, which consists of a constant phase element ZCPE and chargetransfer resistance Rct. The value of Rsol depends on the ion concentration inside the medium. The impedance of constant phase is defined as ZCPE = 1/Q(jω)n, where Q is a measure of the magnitude of ZCPE, ω is the angular frequency, and n is a

R seal =

ρs d

δ

(1)

where, ρs is the resistivity of the electrolyte, d is the average cellto-microelectrode distance, and δ is the surface overlapping coefficient that takes into account the contact area between the cell and the microelectrode.30 The total impedance of the single-cell−electrode system can then be expressed as R sealR m R m + R seal + jωR mR sealCm R ctZCPE + R ct + ZCPE

Zcell−electrode = R sol +

11072

(2)

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 6. Real-time monitoring of a single-cell activities on MEAs in the 2D cell culture system. (a) Change in impedance magnitude induced by a single cell trapped on a working electrode. (b) Change in impedance magnitude induced by cell adhesion and spreading on top of the working electrode after an on-chip cell culture for 2 h. (c) A microscopic image of single MDA-MB-231 cells spreading on MEAs. Single cells were trapped on all positions of the MEAs. The red circle indicates the single cell that was not yet spread.

By fitting the measured spectra into the circuit model, the values of each electrical element in the circuit are obtained. Table 1 shows the values of the circuit model extracted with the Zview software. It can be seen that the extracted values of electric elements correspond to the physiology of a single cell on microelectrodes with different surface properties. Cell spreading on top of the microelectrode with surface modification leads to a decrease of the membrane’s capacitance Cm of about 14% and an increase of the sealing resistance Rseal of about 7%. This results agrees with the previous work by seeding a population of cells on interdigitated microelectrodes.31 The results suggest that the capacitance, the resistance, or the total impedance could be used as a potential parameter to identify the characteristic of cell adhesion on the electrode surface. These data could also be used as a potential diagnostic parameters to distinguish the behavior of cancer cells from the normal cells.31 Real-Time Monitoring of Cell Capture, Adhesion, and Spreading on MEAs. Figure 6, parts a and b, shows a realtime monitoring of single MDA-MB-231 cells capture, adhesion, and spreading on top of MEAs. As shown in Figure 6a, depending on the size of a single cell, a sharp increase of the impedance magnitude of about 50 kΩ is induced when a single MDA-MB-231 cell is trapped by the CCAs. After the capture of

the cell, the impedance value slightly increases because of the cell−substrate interaction. Figure 6b shows a real-time monitoring of the attached cell after cell incubation for 2 h. It can be seen that the magnitude of the impedance increased continuously because of the cell spreading on top of the microelectrode.31 After measuring for 1 h, the signal gradually decreases. The decreased signal could be explained by the stop of cell spreading and the effect of cell medium, which induces a continuously decreasing signal on the reference electrode. To improve the stability of the background signal, a bridge circuit which measures a working and a reference electrode simultaneously could be integrated with the sensor chip.27 Figure 6c shows the microscopic images of single MDA-MB231 cells on MEAs after the on-chip cell culture for 2 h. It can be seen that the cell started to spread over the electrode surface. For the MEAs with smaller electrode dimension (25 × 25 μm2), the spread single cells almost cover the whole electrode surface. Exceptionally, the round shape of the cell indicated by the circle is not yet spread. Characterization of the Cell−Electrode Interaction in a 3D Microenvironment. The dimensionality of the microenvironment is known to influence cellular behaviors, such as the migration speed and the persistence in three dimensions.32,33 In conventional cell invasion assay using the 11073

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

circuit model for the 2D case as shown in Figure 5. Table 2 shows the simulated values of all electric elements in the circuit model fitted to the measured impedance spectra. It can be seen that the tendency of the impedance change in the electrical elements, such as Rseal and Cm, is contrary to the impedance change in those elements in the 2D case, which provides a proof of cell detachment and migration in the 3D matrix. Especially the crucial element Rseal, the simulated value decreases at least 2 orders in magnitude. According to eq 1, the average cell-to-microelectrode distance d could reach a few micrometers after cell migration upward for 2 h (in the initial stage of cell attachment, d is in the range of a few nanometers).29 In addition, the magnitude of Rgel decreases about 50%, which could be explained by the filling up of medium inside the Matrigel layer. Real-Time Monitoring of Single-Cell Migration in Matrigel. A real-time measurement of cell migration in Matrigel is further performed using the sensor chips. Following the experimental protocol described in Figure 3d−f, the sensor chips were put inside a temperature control box (37 °C) for 5 min (until reaching temperature equilibrium) before monitoring the impedance change in real time. Figure 8a shows three representative normalized signals recorded by two working electrodes with a single MDA-MB-231 cell and single MCF-7 cell, respectively, and a reference electrode without cells. It can be seen that the impedance value of the reference electrode and the working electrode with MCF-7 remains stable while the impedance value of the working electrode with MDA-MB-231 decreases gradually with the lapse of time. The curves show that the MDA-MB-231 cell responds to the chemoattractant immediately after the diffusion of FBS. Figure 8b shows the change in impedance value every 200 s caused by the single-cell migration. It can be seen that the migration of metastatic MDAMB-231 cells causes a rapid variation of impedance magnitude with a rate of approximately 10 Ω/s, whereas no prominent impedance change is observed for less-metastatic MCF-7 cells. The free diffusion of FBS into the thin Matrigel layer can be simply estimated by a one-dimensional diffusion model based on Fick’s law34

Boyden chamber, the cells are seeded on top of the 3D matrixes, invading through the matrixes, thus beginning in two dimensions and moving into three dimensions. In the proposed sensor chip, an upward cell invasion assay as shown in the inset of Figure 7 is performed to investigate the cell migration in three dimensions.

Figure 7. Bode impedance spectra measured on a working electrode (with a single cell) before and after cell migration. An electrical equivalent circuit model is built up to fit the measurement data. The inset shows the mechanism of an upward cell migration assay in 3D matrixes using the proposed sensor chip.

To characterize the cell−electrode interaction in a 3D microenvironment, impedance spectra are first measured and then fitted to an equivalent circuit model. As the experimental protocol described in Figure 3d−f, a 3D matrix using Matrigel is constructed inside the microchannel. A gradient of chemoattractant (10% FBS) is generated along the thickness of the thin Matrigel layer to induce the cell migration inside the 3D matrix. Figure 7 shows the measured impedance spectra and their fitting curves. The baseline spectrum (red circles) is measured by the reference electrode without cells. The impedance spectrum (blue squares) is measured by the working electrode right after generating a chemoattractant gradient on the sensor chip. The impedance spectrum (pink stars) is measured by the working electrode after cell incubation for 2 h. Compared to the impedance spectra shown in Figure 5b, the impedance magnitude becomes larger because of the coating of Matrigel on top of the microelectrodes. On the basis of the assumption of cell migration inside the 3D matrix, the impedance magnitude is expected to gradually decrease with time. An electrical equivalent circuit is constructed to model the cell−electrode interaction in the 3D microenvironment. To model the influence of Matrigel on the system, R gel (representing the resistance of Matrigel) and Cgel (representing the capacitance of Matrigel) in parallel are added to the original

p(x , t ) =

⎡ −x 2 ⎤ 1 exp⎢ ⎥ 4πDt ⎣ 4Dt ⎦

t>0 (3)

where p(x, t) is the probability of finding a molecule at position x at time t and D is the diffusion coefficient. The probability distribution corresponds to a Gaussian distribution whose variance is 2Dt. Therefore, the root-mean-square displacement, xrms, can be expressed as xrms = (2Dt)1/2. On the basis of the diffusion coefficient of FBS in Matrigel, D = 6.41 × 10−11 m2/ s,35 and the thickness of Matrigel (approximately 100 μm) in the proposed sensor chip, a chemoattractant gradient can be formed along the thickness of Matrigel just within 1 min, which explains the rapid response of the MDA-MB-231 cell to the chemoattractant. Compared with the xCELLigence device, where the response time required for the cell to migrate through the Matrigel is typically longer than 1 h,12 the

Table 2. Simulated Values of the Electrical Elements in the Equivalent Circuit Model for Characterization of the Cell−Electrode Interaction in a 3D Microenvironment

initial state after incubation for 2 h

Rsol [Ω]

Rgel [Ω]

Cgel [F]

Rseal [Ω]

Rm [Ω]

Cm [F]

Q

n

Rct [Ω]

1214 967

3028 1430

8.13 × 10−11 1.03 × 10−10

3.39 × 108 5.60 × 105

6.39 × 108 1.62 × 106

3.89 × 10−10 9.64 × 10−10

3.38 × 10−9 1.70 × 10−9

0.78 0.82

2.42 × 106 8.70 × 107

11074

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

Figure 8. (a) Representative normalized signals recorded by two working electrodes with a single MDA-MB-231 cell and single MCF-7 cell, respectively, and a reference electrode in the upward cell migration assay. (b) The change in impedance value every 200 s caused by the cell migration. The data represents either the mean of three repeated results or mean ± standard deviation (SD), n = 3. Migration of single MDA-MB231 cells induced a rapid variation of impedance magnitude with a rate of approximately 10 Ω/s, whereas no prominent impedance change was observed for less-metastatic MCF-7 cells.



proposed sensor chip is capable of real-time monitoring cell migration within a very short period of time. In addition, the resolution of the proposed sensor chip can be at the single-cell level. Although various microfluidic devices have been proposed for real-time monitoring the process of cell migration in 3D matrixes,36−38 the resolution and data acquisition of those devices are prominently limited by the optical microscopybased detection method. By utilizing the method of cell impedance measurement in the proposed sensor chip, cell migration kinetics can be easily detected and recorded.

AUTHOR INFORMATION

Corresponding Author

*Phone: +49-761-203-7276. Fax: +49-761-203-7262. E-mail: [email protected]. Author Contributions †

T.A.N. and T.-I.Y. contributed equally to this work.

Notes

The authors declare no competing financial interest.



CONCLUSIONS We present a microfluidic device with integrated ECIS for investigating single cancer cell migration in 3D matrixes. By integrating the passive pumping method with the proposed sensor chip, single cells can be efficiently trapped on the microelectrode array for sequential 2D or 3D cell culture and impedance measurement without the requirement of physical connections to off-chip syringe pumps or off-chip pneumatics. The impedance spectra along with the equivalent circuit model indicate that the cellular activities such as cell adhesion, cell spreading, and cell migration at the single-cell level can be identified by the electrical parameters in the circuit. Furthermore, a cell migration assay in 3D matrixes can be performed on the sensor chip to detect cell migration in their initial stage within a short period of time. Although the presented measurements were recorded using a single working or reference electrode for each experiment, the proposed sensor chip is able to measure multiple signals from multiple trapping sites by integrating a multiplexer with the impedance analyzer (Solartron SI 1260). More breast cancer cell lines with distinctive migration abilities will be further tested using the device to establish the statistical variability of the sensor response to cell migration. Such an ECIS-based device that allows for a rapid and selective detection of the migratory properties of cancer cells at the single-cell level could be applied as a new tool for cancer research.



ACKNOWLEDGMENTS



REFERENCES

We are grateful for the funding support from Else KrönerFresenius-Stiftung (2011-A114) and Vietnam International Education DevelopmentVietnam Ministry of Education and Training (VIED). We thank Mr. Dani Zeniieh for valuable discussions and Professor Brabletz’s Lab (University of Freiburg Medical Center) for providing us the materials and related technical support for the cell experiments.

(1) Giaever, I.; Keese, C. R. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 3761−3764. (2) McGuinness, R. Curr. Opin. Pharmacol. 2007, 7, 535−540. (3) Gu, W.; Zhao, Y. Expert Rev. Med. Devices 2010, 7, 767−779. (4) Solly, K.; Wang, X.; Xu, X.; Strulovici, B.; Zheng, W. Assay Drug Dev. Technol. 2004, 2, 363−372. (5) Chen, Y.; Zhang, J.; Wang, Y.; Zhang, L.; Julien, R.; Tang, K.; Balasubramanian, N. Biosens. Bioelectron. 2008, 23, 1390−1396. (6) Liu, Q.; Yu, J.; Xiao, L.; Tang, J. C. O.; Zhang, Y.; Wang, P.; Yang, M. Biosens. Bioelectron. 2009, 24, 1305−1310. (7) Srinivasaraghavan, V.; Strobl, J.; Agah, M. Lab Chip 2012, 12, 5168−5179. (8) Friedl, P.; Wolf, K. Nat. Rev. Cancer 2003, 3, 362−374. (9) Even-Ram, S.; Yamada, K. M. Curr. Opin. Cell Biol. 2005, 17, 524−532. (10) Sodunke, T. R.; Turner, K. K.; Caldwell, S. A.; McBride, K. W.; Reginato, M. J.; Noh, H. M. Biomaterials 2007, 28, 4006−4016.

11075

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076

Analytical Chemistry

Article

(11) Chen, H.-C. Boyden chamber assay. In Cell Migration; Guan, J.L., Ed.; Humana Press Incorporated: Totowa, NJ, 2005; pp 15−22. (12) Bird, C.; Kirstein, S. Nat. Methods 2009, 6. DOI:10.1038/ nmeth.f.263. (13) Levsky, J. M.; Singer, R. H. Trends Cell Biol. 2003, 13, 4−6. (14) Cho, S.; Thielecke, H. Biosens. Bioelectron. 2007, 22, 1764−1768. (15) Jang, L.-S.; Wang, M.-H. Biomed. Microdevices 2007, 9, 737−743. (16) Han, A.; Yang, L.; Frazier, A. B. Clin. Cancer Res. 2007, 13, 139− 143. (17) Malleo, D.; Nevill, J. T.; Lee, L. P.; Morgan, H. Microfluid. Nanofluid. 2010, 9, 191−198. (18) Hong, J.-L.; Lan, K.-C.; Jang, L.-S. Sens. Actuators, B 2012, 173, 927−934. (19) Arya, S. K.; Lee, K. C.; Rahman, A. R. A. Lab Chip 2012, 12, 2362−2368. (20) Giampieri, S.; Manning, C.; Hooper, S.; Jones, L.; Hill, C. S.; Sahai, E. Nat. Cell Biol. 2009, 11, 1287−1296. (21) Walker, G. M.; Beebe, D. J. Lab Chip 2002, 2, 131−134. (22) Jo, B.-H.; Van Lerberghe, L. M.; Motsegood, K. M.; Beebe, D. J. J. Microelectromech. Syst. 2000, 9, 76−81. (23) Mrksich, M.; Chen, C. S.; Xia, Y.; Dike, L. E.; Ingber, D. E.; Whitesides, G. M. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 10775−10778. (24) McPherson, A. A Passive Pumping Microfluidic Coulter Counter. M.S. Thesis, North Carolina State University, Raleigh, NC, 2009. (25) Asphahani, F.; Thein, M.; Veiseh, O.; Edmondson, D.; Kosai, R.; Veiseh, M.; Xu, J.; Zhang, M. Biosens. Bioelectron. 2008, 23, 1307− 1313. (26) Franks, W.; Schenker, I.; Schmutz, P.; Hierlemann, A. IEEE Trans. Biomed. Eng. 2005, 52, 1295−1302. (27) Lind, R.; Connolly, P.; Wilkinson, C.; Breckenridge, L.; Dow, J. Biosens. Bioelectron. 1991, 6, 359−367. (28) Borkholder, D. Cell Based Biosensors Using Microelectrodes. Ph.D. Thesis, Stanford University, Stanford, CA, 1998. (29) Giaever, I.; Keese, C. R. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 7896−7900. (30) Martinoia, S.; Massobrio, P.; Bove, M.; Massobrio, G. IEEE Trans. Biomed. Eng. 2004, 51, 859−863. (31) Mamouni, J.; Yang, L. Biomed. Microdevices 2011, 13, 1075− 1088. (32) Fraley, S. I.; Feng, Y.; Krishnamurthy, R.; Kim, D.-H.; Celedon, A.; Longmore, G. D.; Wirtz, D. Nat. Cell Biol. 2010, 12, 598−604. (33) Menon, S.; Beningo, K. A. PLoS One 2011, 6, e17277. DOI: 10.1371/journal.pone.0017277. (34) Howard, J. Mechanics of Motor Proteins and Cytoskeleton; Sinauer Associates, Inc.: Sunderland, MA, 2001. (35) Shin, Y.; Kim, H.; Han, S.; Won, J.; Jeong, H. E.; Lee, E.-S.; Kamm, R. D.; Kim, J.-H.; Chung, S. Adv. Healthcare Mater. 2012, 2, 790−794. (36) Chaw, K.; Manimaran, M.; Tay, F.; Swaminathan, S. Biomed. Microdevices 2007, 9, 597−602. (37) Liu, T.; Li, C.; Li, H.; Zeng, S.; Qin, J.; Lin, B. Electrophoresis 2009, 30, 4285−4291. (38) Shin, Y.; Han, S.; Jeon, J. S.; Yamamoto, K.; Zervantonakis, I. K.; Sudo, R.; Kamm, R. D.; Chung, S. Nat. Protoc. 2012, 7, 1247−1259.

11076

dx.doi.org/10.1021/ac402761s | Anal. Chem. 2013, 85, 11068−11076