Microfluidic Impedance Cytometer with Inertial ... - ACS Publications

Jan 27, 2017 - School of Mechanical Engineering, Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast ...
3 downloads 0 Views 1MB Size
Subscriber access provided by University of Florida | Smathers Libraries

Article

A Microfluidic Impedance Cytometer with Inertial Focusing and Liquid Electrodes for High-throughput Cell Counting and Discrimination Wenlai Tang, Dezhi Tang, Zhonghua Ni, Nan Xiang, and Hong Yi Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b04959 • Publication Date (Web): 27 Jan 2017 Downloaded from http://pubs.acs.org on February 7, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

A Microfluidic Impedance Cytometer with Inertial Focusing and Liquid Electrodes for High-throughput Cell Counting and Discrimination Wenlai Tang, Dezhi Tang, Zhonghua Ni, Nan Xiang* and Hong Yi* School of Mechanical Engineering, Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast University, Nanjing, 211189, China *E-mails: [email protected] (Nan Xiang), [email protected] (Hong Yi) ABSTRACT: In this paper, we present a novel impedance microcytometer integrated with inertial focusing and liquid electrode techniques for high-throughput cell counting and discrimination. The inertial pre-focusing unit orders cells into a determinate train to reduce the possibility of cell adhesions and ensure that only one cell passes through detection region at a time, which improves the accuracy of downstream detection. The liquid electrodes are constructed by inserting Ag/AgCl wires into the electrode chambers filled with flowing highly conductive electrolyte solutions, which owns a high detection sensitivity while requiring a simple fabrication process. The effects of main sample flow rate, feed flow rate in electrode chambers and feed solution type on measured impedance signals are experimentally explored. On the basis of the optimized system, we establish a linear relationship between the amplitude of impedance peaks and the volume of size-calibrated particles, and achieve a high detection throughput of ~5000 cells/s. Finally, using the calibrated microcytometer, we further investigate the size distributions of human breast tumor cells (MCF-7 cells) and leukocytes (WBCs) and set a threshold amplitude to successfully distinguish the MCF-7 cells spiked in WBCs. Our impedance microcytometer may provide a potential tool for label-free cell enumeration and identification.

for accurate detection or collect the output light in a highly reproducible manner, various on-chip optical systems have been developed, such as inserted optical fibers,13 integrated waveguides,14 built-in lens15 and so on. However, the employed optical collection and detection systems are quite expensive and complex for fabrication, and an additional biochemical labeling is generally required before detection, which significantly limits the future application of optofluidic microflow cytometers in low-cost POCT. As an alternative to standard optical detection, the impedance flow cytometry can extract cellular biophysical properties through electrical impedance measurement, which can replace the complex optical components by simple electrodes.16,17 A significant advance in the microfluidic impedance flow cytometry has been achieved by fabricating two pairs of coplanar microelectrodes on the bottom of a microchannel.18 Therefore, a differential impedance sensing scheme could be established to directly measure the impedance signals from single cells while passing through the detection region. The major drawback of the cytometers using coplanar microelectrodes is the non-homogeneous distribution of electric fields in detection region, which may lead to low detection stability and repeatability. To address this issue, parallel facing microelectrodes are developed to homogenize the electric field distribution and improve the detection sensitivity.19,20 Although multiple dielectric properties can be obtained for individual cells, the aforementioned AC impedance cytometers suffer from limited throughput and thus are not suitable for measuring large amounts of cells (e.g., rare circulating tumor cells (CTCs) in blood). In addition, using the commercial impedance analyzers or lock-in amplifiers to acquire and analyze impedance signals makes the system hard to achieve portability. Instead, the DC impedance microflow cytometers have great advantages

Since first proposed in 1934, flow cytometry has become an essential technique for quantifying cell populations and offers multi-parameter, high-throughput and high-accuracy detection for widespread research fields, such as cell biology,1 immunology,2 clinical medicine3 and disease diagnostics.4 However, the commercial bench-top flow cytometers suffer from several drawbacks including bulky size, high cost, and complicated optics, which makes it difficult for the conventional flow cytometry to meet the burgeoning requirements of point-of-care testing (POCT) and in-situ pathogen detection. Microfluidic devices,5 which consist of microchannel networks with dimensions comparable to cell diameters, can achieve accurate cell manipulation on the basis of the interactions between cell, fluid and channel geometry. In addition, the small footprint allows for the easy integration of microfluidic devices with other microsensors or microactuators for more diverse functions. These novel features of microfluidics perfectly support the development of low-cost and portable microflow cytometers. Pioneer works on microflow cytometers directly transfer the conventional fluidic and optical components of commercial flow cytometers to microfluidic devices, which are known as optofluidic microflow cytometers.6 In addition to the traditional hydrodynamic focusing using sheath fluids, some sheathless microfluidic focusing schemes have also been employed to ensure the passing of single cells through the interrogation beam, including active acoustophoresis,7 dielectrophoresis,8 magnetophoresis,9 and passive particle focusing.10 Among these schemes, the inertial focusing is especially attractive due to the capability of focusing cells into a single ordered train without any external force fields, providing a potential pretreatment for high-throughput cell analysis.11,12 In order to define a small illumination volume

ACS Paragon Plus Environment 1

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

in the miniaturization of system and the improvement of detection throughput due to the fast transient response and simple sensing circuit.21,22 For instance, a DC impedance microcytometer with two polyelectrolyte gel electrodes (PGEs) and a sheath-flow focusing unit was developed to count various cells at a throughput of 1000 cells/s.23,24 However, the complex photopolymerization of PGEs and the additional control of sheath flows for 2D focusing may hinder its mass production and low-cost applications. In this paper, we develop a high-throughput and label-free liquid electrode-based impedance microcytometer for the efficient enumeration and identification of particles and cells. An inertial pre-focusing unit is employed to reduce the possibility of cell adhesions and ensure that only a single cell passes through detection region at any given time. The liquid electrode, constructed by inserting an Ag/AgCl wire into the electrode chamber filled with flowing highly conductive electrolyte solution, can easily shorten the length of detection region through increasing the flow rate of feed solution, resulting in an improved detection sensitivity. We then optimize the operating conditions of our liquid electrode microcytometer and calibrate the exact linear relationship between impedance peak amplitude and particle volume. Finally, we successfully employ our liquid electrode microcytometer for the discrimination of tumor cells and blood cells. The proposed microcytometer has a small footprint while holding a high throughput of ~5000 cells/s and may become a promising tool for the future low-cost point-of-care testing.

maximum fluid velocity (approximated as ~1.5U). As the focused particle passes through detection region, it displaces conductive fluid and alters the resistance of detection region. The resistance change caused by the displacement of conductive fluid can be approximated as ∆R = 4ap3/(πσmD4),28 while the original resistance of detection region can be simplified as R = 4l/(πσmD2), where σm is the conductivity of electrolyte, D and l are the equivalent circular diameter and length of detection region, respectively. The ratio of resistance change ∆R to original resistance R is defined as the relative change in resistance ∆R/R = ap3/(D2l), which depends only on the channel structure and inherent particle diameter. In order to reduce the original resistance and magnify the relative resistance change caused by particle translocating, a novel liquid electrode was constructed by inserting a non-polarizable Ag/AgCl wire into the electrode chamber filled with 1M KCl solution (using T-shaped fluid connector), as shown in Figure 1. Since the conductivity of KCl solution (~11.2 s/m) in electrode chamber is much higher than that of PBS solution (~1.6 s/m) in main channel, the whole electrode chamber and side channel can be considered as a liquid electrode. Therefore, the length l of detection region is shortened to be smaller than the width of main microchannel, which increases the relative resistance change for improving the detection performance. It is worth noting that the actual length of detection region can be flexibly adjusted by changing the ratio of sample flow rate to feed flow rate.

CONCEPT AND WORKING PRINCIPLE Our liquid electrode microcytometer consists of three key modules: an asymmetrically curved microchannel, two sensing liquid electrodes and a home-made impedance detection system, as illustrated in Figure 1. The asymmetrically curved microchannel was designed as a pretreatment unit for focusing particles into a determinate train. Therefore, the particles suspended in phosphate buffered saline (PBS, PH = 7.4) solution can pass through the downstream detection region one by one, reducing the possibility of cell adhesion to channel walls and the detection errors caused by the overlap of flowing particles. The detailed working principle of the asymmetrically curved microchannel is the newly emerging inertial focusing,25 which can order particles into a single lateral equilibrium position due to the balance of transverse inertial lift force FL and Dean drag force FD (see Figure S-1a). The inertial lift force FL is actually a net force of the shear-gradient induced lift force and the wall induced lift force, and can be estimated as FL = fLρU2ap4/Dh2.26 In this equation, fL is the lift coefficient, ρ is the fluid density, U is the average fluid velocity, ap is the particle diameter and Dh is the hydraulic diameter calculated as 2wh/(w+h) (where w and h are respectively the channel width and height) for rectangular channels. The opposite Dean drag force FD is generated by the secondary flow in the channel cross-section and can be scaled as FD ~ ρU2apDh2/r,26 where r is the radius of channel curvature. The ratio of these two forces directly determines the focusing modes and the lateral positions of particles. To successfully focus particles under specific flow rates, the particle diameter, the radius of channel curvature and the smallest dimension of channel cross-section (i.e., channel height h in this work) need to satisfy the design rule for inertial focusing in curved channels (i.e., ap2r/h3 > 0.04).27 The channel length required for effective focusing in curved channels can be estimated by multiplying the channel length required for straight channels by a factor f as L = fπµw2/(ρUmap2fL),27 where µ is the fluid viscosity, Um is the

Figure 1. System configuration and working principle of the liquid electrode impedance microcytometer.

The resistance change originated from the translocating of particle was then measured using a home-made impedance detection system consisting of a current amplifier, a USB data acquisition (DAQ) card and a computer. The impedance detection system outputs a voltage change of ∆V = KπσmViap3/(4l2), where K is the amplification factor of current amplifier and Vi is the applied DC bias voltage. The base voltage measured by detection system can be expressed as Vb = KπσmViD2/4l. To eliminate the influences of solution conductivity and applied voltage on measured voltage change, we defined the ratio of voltage change ∆V to base voltage Vb as the relative impedance change: Zr = ∆V/Vb = ap3/(D2l). (1) It is clearly shown that, the relative impedance change Zr directly reflects the relative resistance change ∆R/R caused by the translocating of particle. Unless otherwise noted, all the discussions on impedance signals in this paper refer to the relative impedance signals. Each impedance peak corresponds to the translocation of an individual particle and the peak amplitude is proportional to cubic diameter (i.e., particle volume). The customized current amplifier, which

ACS Paragon Plus Environment 2

Page 2 of 9

Page 3 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry includes an I-V converter, a second-stage voltage amplifier and a low-pass filter, was designed to amplify ionic current signals and remove background noises. The generation of DC bias voltages and the acquisition of response voltage signals were performed simultaneously using a portable USB data acquisition (DAQ) card at a sampling rate of 2 MHz. The acquired electrical signals were further processed via MATLAB-based algorithms for noise reduction, baseline compensation and peak feature extraction.

camera (Phantom V611, Vision Research Inc.)

RESULTS AND DISCUSSION Inertial Particle Pre-focusing for Ensuring Accurate Detection. To accurately measure the number and size information of particles, it is essential to ensure that a single particle passes through detection region at any given time. As confirmed in Figure 2a, the simultaneous existence of double particles in detection region was measured to be a single impedance peak whose amplitude is nearly twice as high as that from single particles. This may lead to the false identification of particle size and the loss of count number, which can be overcame through focusing particles into a determinate train. In this work, the inertial focusing with asymmetrically curved microchannel was adopted. We systematically investigated the focusing behaviors of three differently-sized particles (10 µm, 15 µm and 20 µm) at a particular range of flow rates. Considering the construction of downstream liquid electrodes may be destroyed at high sample flow rate due to the diffusion of PBS solution into electrode chamber, the sample flow rates were set to be 50~250 µl/min. To eliminate the random factor, a set of image frames under identical operating condition were stacked to obtain the statistical distribution of particles over a period of time. From the focusing maps illustrated in Figure S-1b, we found that some particles are unable to reach equilibrium positions at the flow rate of 50 µl/min after passing through the entire curved channel, owing to weak inertial lift force and Dean drag force. Therefore, the sample flow rates for following detection experiments were selected to be 100~250 µl/min where all the tested particles can be well focused. We also quantified the lateral focusing positions of particles across channel width at different flow rates, as illustrated in Figure S-1c. The stable focusing positions of all particles are found to be very close to each other. Therefore, the flowing particles of different sizes could not occupy same downstream positions due to the particle-particle interactions (see the inset of Figure S-1c), which reduces the possibility of particle overlaps in detection region and thus reduces the false detection.

EXPERIMENTAL SECTION Device Fabrication. The detailed dimensions of microfluidic device are described in Text S-1 (Supporting Information). The device was cost-effectively fabricated using soft lithography technique. Specifically, the channel pattern was transferred onto a transparency with a high-resolution laser photoplotter (RP525-SST, EIE). The photomask was aligned on the surface of a silicon wafer coated with a layer of ~25 µm thick SU-8 photoresist (2050, MicroChem Corp.). The photoresist was then exposed via UV light using an exposure machine (MJB4, SUSS). Following a process of post-exposure baking, developing and hard baking, an SU-8 master mold with channel pattern was finally obtained. Then, a degassed mixture of polydimethylsiloxane (PDMS) prepolymer and curing agent (Sylgard 184, Dow Corning Corp.) at a weight ratio of 10:1 was poured onto the prepared master mold. After curing, the cured PDMS block was carefully peeled from master mold and cut into acceptable pieces. Finally, the PDMS piece punched with fluidic ports was permanently bonded to a clean glass slide using the plasma cleaner (PDC-002, Harrick Plasma). In addition to channel device, the T-shaped fluid connector was fabricated by punching a cross hole in a PDMS block and sealing one of the four ports and then connected to the inlet of electrode chamber as feed solution inlet. The Ag/AgCl wire was finally inserted into the vertical port of T-shaped fluid connector for connecting the impedance detection system. Sample Preparation. Polystyrene particles with diameters of 10 µm, 15 µm and 20 µm (Bangs Laboratories and Thermo Fisher Scientific Inc.) were suspended in PBS solution to evaluate the performances of our impedance cytometer. The human breast tumor MCF-7 cells were cultured in the high-glucose Dulbecco’s Modified Eagle’s Medium (DMEM, Life technologies) containing 10% fetal bovine serum (FBS, Life technologies) and 1% penicillin-streptomycin (Life technologies) at 37°C in 5% CO2. The cells were detached with 0.05% Trypsin-EDTA solution (Life technologies) and then collected after centrifugation at 1000 rpm for 5 min. Human whole blood samples donated by healthy volunteers were collected by the venipuncture into EDTA vacuum tubes. Red blood cells were lysed by incubation with ACK lysis buffer (Life technologies) for 5 min and the leukocytes were collected by centrifugation at 300 g for 5 min. Finally, the MCF-7 cells and leukocytes were resuspended in PBS solution. Experimental Setup. The inlets of main channel and T-shaped fluid connectors were connected to syringe pumps (KDS 270, KD Scientific Inc.) via PTFE tubings for the injection of particle/cell suspensions and KCl solution, respectively. During the impedance measurement, a low DC voltage of ~1V generated by DAQ card (USB-6366, National Instruments) was used as excitation signal. The amplification factor K of current amplifier was set to be 105 V/A. In order to observe the particle focusing process and verify the impedance peaks caused by particle translocations, the particle migrations in microchannel were recorded using an inverted microscope (IX 71, Olympus) with a high-speed

ACS Paragon Plus Environment 3

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

investigate the influence of sample flow rate on measured impedance signal, particles suspended in PBS solution were measured at four different flow rates ranging from 100 µl/min to 250 µl/min with an interval of 50 µl/min, while the pure KCl solution was injected into electrode chambers at a lower flow rate of 20 µl/min. The representative impedance signals of three different particles at the flow rate of 100 µl/min are illustrated in Figure 3a. For each particle type, the measured impedance peaks show similar amplitudes owing to the uniform size distribution of particles and the number of particles can be easily obtained by counting the peaks. Through comparing the impedance peaks of different particles, we found that the peak amplitudes increase significantly with increasing particle diameter, which agrees well with eq 1. The peak width reflects the duration of the electric field disturbance caused by particle translocation. Due to the diffusion of electric field lines in main channel (see Figure S-1d), the peak width of particle is found to be much longer than the real time to pass through detection region (see Figure S-2). It is also interesting to find that there exist a rising peak and a shoulder peak along with main impedance peak. When appearing in main channel, the particle itself may act as a moving obstacle to the flow of PBS solution. Therefore, before entering into detection region, the particle hinders PBS solution to flow past so that the KCl solution with a high conductivity occupies more space in detection region, leading to a resistance reduction. This process is measured as a rising peak before the main peak. Inversely, when the particle leaves detection region, more PBS solution with a low conductivity is left in detection region due to the existence of a particle in the downstream, resulting in a resistance increase. This process is presented as a shoulder peak after main peak. Since the large particles have a more significant influence on the disturbance of PBS flows, both the width and amplitude of two small peaks increase with increasing particle diameter.

Figure 2. (a) Representative impedance signals of 10 µm particles without using inertial pre-focusing unit. The sample flow rate was fixed at 100 µl/min while the feed flow rate was set to be 20 µl/min. The inset is the microscope image illustrating the simultaneous existence of two particles in detection region which results in a high impedance peak. (b) Distribution of peak amplitudes of randomly-dispersed 10 µm particles. The sample flow rate was 50 µl/min while the feed flow rate was 20 µl/min. The impedance signals caused by the simultaneous existence of multiple particles in the detection region were carefully rejected by synchronizing the signals with particle images from the high speed camera.

It is noticed that the lateral focusing positions deviate from channel centerline for all three tested particles and change with flow rates. To investigate the influence of lateral position on impedance signal, we measured randomly-dispersed 10 µm particles and plotted the distribution of impedance peak amplitudes in Figure 2b. The peak amplitudes are in accordance with a Gaussian distribution and have a low coefficient of variation (CV) of 7.3%, means that the impedance signals are independent of particle focusing positions. To verify this conclusion, the electric field distribution in detection region was simulated (see Figure S-1d). The results show that the electric field distributions near the centerline of side channels in both horizontal plane and vertical direction, which directly determine the peak amplitude of particle translocation, maintain uniform. These uniform distributions of electrical field ensure that the lateral and vertical focusing positions of particles have little influence on the measured electrical signals. Counting and Sizing Particles. We first measured three size-calibrated particles to characterize the performances of our liquid electrode microcytometer. In attempts to

Figure 3. (a) Representative impedance signals for three differently-sized particles at the flow rate of 100 µl/min wiht the feed flow rate fixed at 20 µl/min. (b) Averages and CVs of the peak amplitudes for three differently-sized particles at sample flow rates ranging from 100 µl/min to 250 µl/min with an interval of 50 µl/min. (c) Relationship between impedance peak amplitude and particle volume

ACS Paragon Plus Environment 4

Page 4 of 9

Page 5 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry at a sample flow rate of 100 µl/min. Error bars represent the standard deviations calculated from the data of numerous particles (n ≈ 100). The results from the microcytometer using PBS as feed solution are also included for comparison.

The average amplitudes of three particles at different sample flow rates were normalized with the maximum value of each particle type, and plotted as a function of sample flow rate in Figure 3b. The CVs calculated via statistical analysis are also included in this figure. The results show that the peak amplitudes of all particle types descend while the corresponding CVs ascend with increasing sample flow rate, which indicates that high sample flow rate may lead to the loss of detection sensitivity and stability. When the feed flow rate is fixed, the actual length of detection region lengthens with increasing sample flow rate, resulting in a low peak amplitude. In addition, the high sample flow rate may cause the loss of electrical signals since particles move too quickly to be recorded. At the flow rate of 250 µl/min, the much severer decrease in the impedance amplitude of 20 µm particle may result from the displacement of PBS

solution into the electrode chamber due to the existence of large particle. Therefore, in order to maintain good detection performances, the sample flow rate was set to be 100 µl/min for the following measurements. Through linear fitting to the impedance amplitudes versus the volumes of all particles at the flow rate of 100 µl/min, we obtained an exact linear relationship between peak amplitude and particle volume, as shown in Figure 3c. As a comparison, we also characterized the impedance signals from the microcytometer using PBS as feed solution, the operational parameters are identical to the above experiments using KCl solution. The results show that the sensitivity of impedance peak amplitude to particle volume change increases significantly (from 0.00146 to 0.00409) when adopting the liquid electrode technique. In addition, the small standard deviations of peak amplitudes imply that our microcytometer has a good stability.

Figure 4. (a) Representative impedance peaks of three differently-sized particles at the selected feed flow rates of 10 µl/min, 30 µl/min and 50 µl/min, respectively. (b) Calibration curve illustrating the relationship between impedance peak amplitude and particle volume under optimized experiment condition. (c) Impedance signals of 10 µm particles at a high particle concentration (~3×106 particles/ml).

After the optimization of sample flow rate, we further investigated the effects of feed flow rate on measured impedance signal. During the experiments, KCl solutions were continuously injected into electrode chambers at

different feed flow rates ranging from 10 µl/min to 50 µl/min while the sample flow rate was fixed at 100 µl/min. The average peak amplitudes for three different-sized particles under all feed flow rates are listed in Table 1, the

ACS Paragon Plus Environment 5

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

representative impedance peaks of particles at the selected feed flow rates of 10 µl/min, 30 µl/min and 50 µl/min were plotted in Figure 4a. Intuitively, the amplitudes of impedance peaks may increase continuously with increasing feed flow rate due to the shortening of detection region length. However, it is found that after an initial growth at the feed flow rates of 10~30 µl/min, the measured peak amplitudes reach stable at high feed flow rates (>30 µl/min). Under a fixed sample flow rate, the increase of feed flow rate causes particle to pass through detection region more quickly. Therefore, less data points can be recorded in an impedance peak at a fixed sampling rate, which may result in the weakening of peak amplitudes. The compromise between the decrease of detection length and the increase of particle velocity leads to relatively stable peak amplitudes at feed flow rates above 30 µl/min. It is worth noting that the biological cells in isotonic PBS solution may enter into the harmful KCl solution with high ion concentration when the ratio of feed flow rate to sample flow rate increases considerably. Therefore, the feed flow rate was set to be 30 µl/min for the following cell detection application.

microcytometer for the identification of cell populations, we characterized the size information of human breast tumor MCF-7 cells and white blood cells (WBCs) under optimized experiment condition. To reduce the influence of background cells, the most abundant RBCs were removed by using a simple process of chemical lysis. Before discriminating the tumor cells in lysed blood, pure MCF-7 cells and pure WBCs were first measured respectively to obtain the statistics of impedance peaks for each cell type. The high-speed videos of particle migration through detection region were also simultaneously recorded to ascertain that all the measured impedance peaks are derived from the translocations of cells. The distributions of the peak amplitudes of MCF-7 cells and WBCs are shown in Figure 5a. The peak amplitudes of MCF-7 cells spread over a wide range, but are obviously larger than those of most WBCs. Statistical analysis performed using the one-way analysis of variance (ANOVA) indicates that each type of cells could be clearly differentiated (P-value < 0.001). The detailed statistical impedance data of cells are also compared with those of polystyrene particles, as listed in the inserted table of Figure 5a. It is found that the biological cells show similar translocation signals to those of similar-sized particles (see Figures S-4a and S-4b). However, the standard deviations of the peak amplitudes of biological cells are found to be much larger than those of particles due to the polydisperse size distribution of real cells. Based on the linear calibration equation for peak amplitude versus particle volume (see Figure 4a), we calculated cell diameters on the assumption that cells are spherical. The diameters of WBCs measured using our microcytometer were calculated to vary from 6.07 µm to 12.81 µm with an average of 9.01 µm, which agree well with previous reports.24 The MCF-7 cells were calculated to be in the sizes ranging from 11.15 µm to 22.87 µm with an average of 14.28 µm. The calculated diameters of MCF-7 cells are slightly lower than expected, 29 one possible reason for this deviation is that the confined space in microchannel deforms large tumor cells into nonspherical shapes. In order to better distinguish these two types of cells, the full width at half maximum (FWHM) of impedance peak was also measured and defined as peak width. Figure 5b shows the scatterplot of the peak widths and amplitudes for MCF-7 cells and WBCs. It is found that the peak amplitudes of these two cells have some overlaps between 2.81 and 4.34. However, the peak amplitudes of almost 99% of WBCs were found to be less than 3.5, while over 96% of MCF-7 cells hold the amplitudes greater than this value. Therefore, the threshold amplitude for discriminating MCF-7 cells and WBCs was set to be 3.5. In addition, the peak widths of MCF-7 cells in the amplitude overlapping region range from 0.05 ms to 0.09 ms while those of large WBCs distribute around 0.07 ms, which may help discriminate these two cells. On the basis of above experiments, we further measured the impedance signals of WBCs spiked with MCF-7 cells. The final concentration of spiked MCF-7s was nearly one third of the concentration of pure MCF-7 cell population employed in the aforementioned experiment. Figure 5c displays the representative impedance peaks of mixed cells in a time period of 1.5 s. There is a total of 21 impedance peaks over the threshold (dash line), which are caused by the possible translocations of MCF-7 cells. The discrimination results are consistent with the total count (75 cells) obtained from the experiment using pure MCF-7 cells under same experiment condition (see Figure S-5c). In all, the experimental results clearly demonstrate that our novel liquid electrode microcytometer is capable of identifying and enumerating biological cells with different sizes without

Table 1. Averages of the impedance peak amplitudes for three differently-sized particles at all tested feed flow rates (n ≈ 100). Average peak amplitude (A.U.) Feed flow rate 10 µm

15 µm

20 µm

10 µl/min

1.56±0.07

5.69±0.17

14.63±0.62

20 µl/min

1.84±0.08

6.32±0.25

17.04±0.71

30 µl/min

1.93±0.08

6.92±0.17

16.98±0.56

40 µl/min

1.88±0.10

7.06±0.22

17.40±0.54

50 µl/min

2.02±0.08

7.14±0.27

17.38±0.75

By applying a linear fitting to the impedance amplitudes versus the volumes of all three differently-sized particles under optimized experiment condition, we obtained a good linearity between peak amplitude and particle volume (see Figure 4b). This calibration curve is consistent with the previous theoretical estimation (eq 1), which indicates that particles with different sizes can be easily distinguished based on the measured impedance signals. Figure 4c displays the typical impedance signals of 10 µm particles suspended in PBS solution at a high concentration, eight impedance peaks were observed in a time period of ~1.6ms, means that a throughput of up to 5000 cells/s can be achieved in our microcytometer. The detection throughput of the microcytometer can be further improved by proportionally increasing the flow rates of the PBS and KCl solutions. However, a DAQ card with a higher sampling rate is required to acquire all the necessary data points in the impedance signals. The performance of our microcytometer is also affected by the particle concentration. When the particle concentration is high enough, particles are squeezed to get very close to each other while passing through the detection region. The multiple adjacent particles may be detected as a single ambiguous impedance peak, which influences the detection accuracy of the microcytometer system (see Figure S-3). In all, it is suggested that our liquid electrode impedance microcytometer is a potential tool for sensing even a small change in particle diameter and can be further used to discriminate bioparticles of different sizes at a throughput much higher than those in the existing impedance-based flow cytometers (see Table S-2). Discrimination of Tumor Cells and Blood Cells. In order to validate the feasibility of the proposed

ACS Paragon Plus Environment 6

Page 6 of 9

Page 7 of 9

Analytical Chemistry any complex surface functionalization and fluorescent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

tagging.

Figure 5. (a) Distribution profile of the peak amplitudes of MCF-7 cells and WBCs. The inserted table lists the averages and standard deviations of the peak amplitudes for MCF-7 cells, WBCs and polystyrene particles with diameters of 10 µm and 15 µm. (b) Scatterplot of the impedance peaks (width vs amplitude) for MCF-7 cells and WBCs. The inset shows the enlarged view of amplitude overlapping region. (c) Representative impedance signals from WBCs spiked with MCF-7 cells. The green dash line represents the threshold for discriminating these two type cells.

CONCLUSIONS

ACKNOWLEDGMENT

In this work, we reported a novel impedance microcytometer integrated with inertial focusing and liquid electrodes for label-free counting and discrimination of particles and cells. The flow rates of sample and feed solutions were optimized through measuring particles of different diameters and an exact linear relationship between impedance peak amplitude and particle volume was calibrated. The liquid electrode technology effectively improves the detection sensitivity without any additional fabrication process. The inertial focusing of particles in the asymmetrically curved microchannel creates a single-cell continuous flow, offering a detection throughput up to ~5000 cells/s. On the basis of the optimized system, the average diameters of MCF-7 cells and WBCs were measured to be 9.01 µm and 14.28 µm, respectively. According to the selected amplitude threshold, the MCF-7 cells spiked in WBCs can be successfully distinguished and counted. The simplicity and flexibility of our novel liquid electrode microcytometer may open up the possibility of label-free cell population identification and enumeration.

This work was supported by the National Natural Science Foundation of China (51505082, 51375089 and 81572906), the Natural Science Foundation of Jiangsu Province (BK20150606), the “333” Project of Jiangsu Province (BRA2015291), the Jiangsu Graduate Innovative Research Program (KYLX_0098), the Scientific Research Foundation of Graduate School of Southeast University (YBJJ1428), and the Open Foundation of the State Key Laboratory of Fluid Power and Mechatronic Systems (GZKF-201501).

REFERENCES (1) Rose, J. A.; Erzurum, S.; Asosingh, K. Cytometry, Part A 2015, 87A, 5-19. (2) Loken, M. R.; Stall, A. M. J. Immunol. Methods 1982, 50, R85-R112. (3) Litwin, V.; O'Gorman, M. R. G. J. Immunol. Methods 2011, 363, 101-102. (4) Ferreira-Facio, C. S.; Milito, C.; Botafogo, V.; Fontana, M.; Thiago, L. S.; Oliveira, E.; da Rocha-Filho, A. S.; Werneck, F.; Forny, D. N.; Dekermacher, S.; de Azambuja, A. P.; Ferman, S. E.; de Faria, P. A. S.; Land, M. G. P.; Orfao, A.; Costa, E. S. PLoS One 2013, 8, e55534. (5) Manz, A.; Graber, N.; Widmer, H. M. Sens. Actuators, B 1990, 1, 244-248. (6) Piyasena, M. E.; Graves, S. W. Lab Chip 2014, 14, 1044-1059. (7) Chen, Y.; Nawaz, A. A.; Zhao, Y.; Huang, P.; McCoy, J. P.; Levine, S. J.; Wang, L.; Huang, T. J. Lab Chip 2014, 14, 916-923. (8) Holmes, D.; Morgan, H.; Green, N. G. Biosens. Bioelectron. 2006, 21, 1621-1630. (9) Liang, L.; Xuan, X. Microfluid. Nanofluid. 2012, 13, 637-643. (10) Liu, C.; Xue, C.; Chen, X.; Shan, L.; Tian, Y.; Hu, G. Anal. Chem. 2015, 87, 6041-6048.

ASSOCIATED CONTENT Supporting Information. Additional information as noted in text.

AUTHOR INFORMATION Corresponding Author

*E-mails: [email protected], [email protected]. Notes The authors declare no competing financial interest.

ACS Paragon Plus Environment 7

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(11) Oakey, J.; Applegate, R. W.; Arellano, E.; Di Carlo, D.; Graves, S. W.; Toner, M. Anal. Chem. 2010, 82, 3862-3867. (12) Bhagat, A. A. S.; Kuntaegowdanahalli, S. S.; Kaval, N.; Seliskar, C. J.; Papautsky, I. Biomed. Microdevices 2010, 12, 187-195. (13) Tung, Y.; Zhang, M.; Lin, C.; Kurabayashi, K.; Skerlos, S. J. Sens. Actuators, B 2004, 98, 356-367. (14) Yang, Y.; Liu, A. Q.; Lei, L.; Chin, L. K.; Ohl, C. D.; Wang, Q. J.; Yoon, H. S. Lab Chip 2011, 11, 3182-3187. (15) Tang, S. K. Y.; Stan, C. A.; Whitesides, G. M. Lab Chip 2008, 8, 395-401. (16) Xu, Y.; Xie, X.; Duan, Y.; Wang, L.; Cheng, Z.; Cheng, J. Biosens. Bioelectron. 2016, 77, 824-836. (17) Sun, T.; Morgan, H. Microfluid. Nanofluid. 2010, 8, 423-443. (18) Gawad, S.; Schild, L.; Renaud, P. Lab Chip 2001, 1, 76-82. (19) Cheung, K.; Gawad, S.; Renaud, P. Cytometry, Part A 2005, 65A, 124-132. (20) Holmes, D.; Morgan, H. Anal. Chem. 2010, 82, 1455-1461. (21) Jagtiani, A. V.; Sawant, R.; Zhe, J. J. Micromech. Microeng. 2006, 16, 1530-1539. (22) Sun, J.; Gao, Y.; Isaacs, R. J.; Boelte, K. C.; Lin, P. C.; Boczko, E. M.; Li, D. Anal. Chem. 2012, 84, 2017-2024. (23) Chun, H. G.; Chung, T. D.; Kim, H. C. Anal. Chem. 2005, 77, 2490-2495. (24) Choi, H.; Kim, K. B.; Jeon, C. S.; Hwang, I.; Lee, S.; Kim, H. K.; Kim, H. C.; Chung, T. D. Lab Chip 2013, 13, 970-977. (25) Zhang, J.; Yan, S.; Yuan, D.; Alici, G.; Nguyen, N.; Ebrahimi Warkiani, M.; Li, W. Lab Chip 2016, 16, 10-34. (26) Di Carlo, D.; Irimia, D.; Tompkins, R. G.; Toner, M. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 18892-18897. (27) Amini, H.; Lee W.; Di Carlo, D. Lab Chip, 2014, 14, 2739-2761. (28) DeBlois, R. W.; Bean, C. P. Rev. Sci. Instrum. 1970, 41, 909-916. (29) Zhang, H.; Wang, Y.; Li, Q.; Zhang, F.; Tang, B. Chem. Commun. 2014, 50, 7024-7027.

ACS Paragon Plus Environment 8

Page 8 of 9

Page 9 of 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

For TOC only

ACS Paragon Plus Environment 9