Cascade and staggered dielectrophoretic cell sorters

Abstract

We have developed two new microfluidic cell sorters based on conventional negative dielectrophoresis (DEP) for continuous flow operations. The first one is a cascade configuration sorter designed to increase purity of isolated target cell. The second one has two staggered side channels in opposite side walls to increase sample throughput without compromising enrichment factor. Particles (carboxylate microspheres) of different sizes were first used to demonstrate the feasibility of the present DEP sorters for cell isolation. Then biological cells, i.e. human prostate cancer cell line LNCaP and human colorectal cancer cell line HCT116 were used to test the performance of the DEP sorters. In the present work, applied voltage was in the range of 0 to 20 Vp-p, and frequency was from 0 to 10 MHz. Comparing to a single side channel DEP cell sorter, the isolation purity was improved from 80% to 96% by single cascade sorter and the sample throughput was increased from 0.2 μL/min to 0.65 μL/min by a single staggered side channel sorter. In this manuscript, we report the cell sorter designs, cell separation and enrichment factors.

1. Introduction

Dielectrophoresis (DEP) is a phenomenon in which a force is exerted on dielectric particles (either biological or non-biological) when they are subjected to a non-uniform electric field. It is not necessary for the particles to be charged. When an electric field is applied to systems consisting of particles suspended in a liquid, dipole moments are induced on the particles, due to the electrical polarizations at the interface between the particles and the suspending liquid [1, 2]. The polarizability of living cells depends strongly not only on their composition, size, shape, morphology and phenotype, but also on the frequency of the applied electrical field [3, 4]. This means that for a given particle type and suspending medium, the particles can experience, at a certain AC frequency applied to the electrodes, a translational force directed towards regions of high electric field strength (this phenomenon is called positive DEP, i.e. pDEP). Alternatively, by simply changing the frequency, they may experience a force that will direct the particles away from the high electric field strength regions (this phenomenon is called negative DEP, i.e. nDEP) [5].

DEP offers advantages of contactless, label free, easy operation and high specificity. In addition, it can be used to manipulate, transport, separate and sort different types of particles [6-8] including biological particles, such as cells [9-14], bacteria [15, 16], viruses [17-20], yeast [21, 22] and even breast cancer cells [23, 24].

The dielectrophoresis can be easily adopted in integrated microfluidic systems to manipulate, separate particles and cells. In recent years, microfluidics has generated great excitement due to its potential of providing rapid analyses with high resolutions and low costs for a wide range of biological and chemical applications [25-32]. The integrated microfluidic system will play a key role in biochemistry, molecular biology and synthesis protocols [33], and is expected to have major impacts on biomedical research, clinical diagnosis [29], point of care, food pathogen screening [34] and environmental monitoring by providing automated and portable solutions to a wide range of fluid based problems [35-37].

In our previous work [38, 39], a continuous flow DEP cell sorter with a single side channel and a main channel was developed. It was applied to isolate HCT116 colorectal cancer cells from the mixture with Human Embryonic Kidney 293 cells (HEK 293) and Escherichia coli (E. coli) bacteria. However, by its nature, this DEP sorter has a limited capability to achieve a high purity of target cells with only one sorter. Furthermore, like all other microfluidics based cell sorters [25, 28, 40, 41], the microfluidic DEP sorter handles normally only small volume of sample because of the small size of the microchannel, and the enrichment factor decreases with the increased throughput (i.e. flow rate Q). (Note some other DEP based systems have reported significant enrichment/concentration of bioparticles under particular conditions [42-44]) Increasing Q will cause large hydrodynamic force, which can compete with DEP force and reduce the DEP effect for cell separation at high Q.

In reality, many applications require not only highly purified target cells, but also a high throughput of samples. Hence, in the present work, we introduce a microfluidic cascade DEP sorter to increase the purity of the target cells and a staggered side channel DEP cell sorter to increase throughput without compromising the enrichment factor of cells.

2. Principle of the method

DEP force mainly depends on a sphere particle radius $a$, its complex dielectric constant ${\epsilon }_p^{\ast }$, dielectric constant of the surrounding medium ${\epsilon }_m^{\ast }$ and electric field $\overrightarrow{E}$. Based on this, when an electric field with constant $\overrightarrow{E}$ and frequency $\omega$ is applied in a medium with a constant ${\epsilon }_m^{\ast }$, we could predict that particles with different $a$ or ${\epsilon }_p^{\ast }$ in this medium will experience different DEP forces. Consequently, selective separation can be achieved by applying an additional force such as the gravity or a hydrodynamic force in fluid flow.

Based on the aforementioned concept, we have designed a cascade and a staggered DEP cell sorter respectively, as shown in Fig. 1. Both DEP sorters were fabricated using lamination technique. This method offers benefits to build complex, three dimensional devices, including the use of a variety of inexpensive, high quality, bio-compatible materials that can easily accommodate the incorporation of other planar structures [45-47]. For both cascade and staggered side channel cell sorters, transparent acrylic plastic substrates of 24 mm x 20 mm x 1.25 mm were used as top (cap) layer. One inlet and three outlet wells are drilled on the top layer. The diameters of all the wells are 1.6 mm. Channels are formed on a middle layer, with the same height of 40 μm. Indium tin oxide (ITO) thin film as electrodes are coated on the glass substrate (Delta Technologies, Limited). It is transparent and colorless in thin layers, by which we can visualize the separation process. In the cascade sorter the width of the first and second main channels are 550 μm and 300 μm, respectively. The widths of the first and second side channels are 150 μm and 110 μm, respectively. In the staggered sorter, the widths of the main and side channels are 600 μm and 260 μm, respectively. The gaps between the electrodes in all electrode pairs are 20 μm.

Figure 1.

Cascade and staggered DEP sorter. (a) Schematic of the cascade DEP cell sorter. (b) Picture of the fabricated cascade DEP cell sorter under a microscope. (c) Schematic of the staggered side channel DEP sorter; (d) Picture of the staggered side channel DEP sorter.

Cascade:

The cascade DEP cell sorter has two DEP sorters and each has a main channel and a side channel. The side channel of the first sorter is the main channel of the second one. In this design, there are two electrodes on the bottom surfaces of the microchannel. These electrodes are in parallel with a gap, and have an angle of approximately 45° to the streamwise direction across the two main channels. AC electric signals are applied to the two electrodes having a phase difference of 180°. The operational principle is given below. As the target particles, driven by the hydrodynamic force, approach the gap between the electrode pair, they will experience a strong nDEP force, which repels them against the hydrodynamic force in the direction with a certain angle to the flow direction in the main channel, if we apply an AC signal with frequency in the particles’ nDEP force regime. The net force of DEP f_DEP and hydrodynamic force f_h1 push the target particles to the side channel, while other particles that do not experience the DEP force will keep their motion to the end of the main channel by the hydrodynamic force f_h2. However, the side channel must have a small portion of flow to carry the isolated target cells to a collection well. Thus, there will always be a small portion of non-target cells that follows this portion of fluid and flows into the side channel. This will limit the purity of the target cells isolated in the side channel [39]. The purity $P$ was defined as the ratio of target particles number $T$ collected in the side channel to the total particles number $T+N$ collected in the side channel, where $N$ is the number of non-target particles.

$$P=T∕(T+N)$$ (1)

To avoid this problem, one can make the entrance of side channel as small as possible, but this method has the risk of leading to the block of the side channel. Moreover, this method could also raise the difficulty of the fabrication of the chip. In the present work, we use the cascade configuration to solve this problem. As shown in Fig. 1(a), the first side channel is directly connected to the second main channel. Therefore, repelled target cells and a small portion of non-target cells will flow into the second main channel. As there is an electrode pair located at the second main channel with the same gap between them, the target cells will be repelled again by the nDEP force before the second electrode pair, and will move to the second side channel, but the non-target cells will pass over the electrode pair along the second main channel by the hydrodynamic force. The sorting process is repeated in the second main channel. Consequently, the cells are further purified. Fig. 1(b) is a picture of the fabricated cascade chip under a microscope. The fabrication method is given in a previous work [39].

Staggered:

In a DEP separation process, sample throughput is limited by the strength of the DEP force, i.e. as the throughput increases, the hydrodynamic force will be increased as well, and the percentage of DEP repelled cells will decrease [48]. A straight forward way to solve this problem is to increase the electric field strength, which will enhance the DEP force on the cells. However, high electric field could result in Joule Heating, which is potentially harmful to the cells. It is expected that the targeted cells near the side wall (on the bottom side in Fig. 1(c)) opposite to the side channel have to move a longer distance, and thus, need longer time to deflect to the first side channel, than those cells near the side channel do. As cells approach the electrode pair, the cells near the side wall opposite to the side channel are more difficult to be repelled to the side channel. Therefore, they are most likely be pushed to the main channel without being repelled to the first side channel.

In order to solve this problem, we introduce a second DEP sorter downstream the original one along the main channel. In the second DEP sorter, another side channel is designed on the side wall that is opposite to the first side channel, and another electrode pair is placed right downstream the second side channel as the first electrode pair is, but in an angle of approximately 135° to the main flow direction, as shown in Fig. 1 (c). In this case, those cells that are not collected in the first side channel will be collected in the second side channel. As a result, the throughput will be increased without reducing the enrichment factor. Picture of the fabricated staggered DEP sorter under a microscope is shown in Fig. 1(d).

To quantify the presented DEP sorter performance for the target cell separation and isolation, we define the cell enrichment factor as the ratio of target particles blocked by the nDEP force to the total target particles or cells that flow into the sorter [44]:

$$\eta =A∕(A+B)$$ (2)

where $A$ is the number of the total target particles or cells that are pushed by DEP force and the hydrodynamic force that flow into the side channels, $B$ is the number of total target particles or cells flowed through the gaps in the main channel. $\eta$ is calculated based on movie obtained from the visualization.

3. Materials and experimental setup

3.1. Sample preparation

To demonstrate the feasibility of the present DEP sorters, two kinds of fluorescent particles with different sizes were used in the experiments. Diameters of the particles were 6 μm (Fluoresbrite^® YO carboxylate microspheres), with wavelength of excitation max of 529 nm and emission max 546 nm and 3 μm (Fluoresbrite^® YG carboxylate microspheres), with wavelength of excitation max of 441 nm and emission max 486 nm, respectively. Both kinds of the particles were mixed in DI water with approximately the same concentration of 2.0 × 10⁷ particles/ml. Conductivity and pH of DI water were about 2 μS/cm and 7, respectively. Under a microscope, 3-μm particles were seen as the small dark spheres, and 6-μm particles as large bright spheres.

Another aim of developing the DEP sorter is to separate different biological cells. Therefore, two types of biological cells were selected for demonstration. A human prostate cancer cell line LNCaP (ATCC^™) was used as the target cells to be separated from a human colorectal cancer cell line HCT116 (ATCC^™). Both LNCaP and HCT 116 cells were in the shape of sphere in the suspending medium and have almost the same diameters of about 20 μm [49]. To distinguish LNCaP from HCT 116 cells under microscope, we labeled HCT 116 cells with dye Hoechst 33342 (InvitrogenTM, whose peaks of absorption and emission are around 350 and 461 nm, respectively [50]). Thus, under a microscope, the LNCaP cells were seen as dark spheres, and the HCT116 cells as bright spheres. Both types of the cells were mixed in a suspending medium, Phosphate-buffered saline (PBS) buffer (TEKnova, Inc.) with approximately the same concentration of 1.5× 10⁶ cells/ml. The conductivity of the medium was 3000 μS/cm.

3.2. Experimental setup

The DEP cell sorters were placed on the test bed of an inverted epi-fluorescent microscope (Olympus-IX70). A syringe pump (Harvard PHD 2000) was used to deliver the sample to the chip. To balance pressure, outlets of each sorter were connected into a reservoir. A function generator (Tektronix, Model AFG3102) was used to supply AC electrical signal to these ITO electrodes. An oscilloscope (LeCroy 1GHz) was used to observe the real signal output of the function generator. The particle motion was captured by a high-resolution CCD camera (SensiCam-QE, Cooke Corp). Green light (532 nm in wave length) was used to excite these particles and UV light (350 nm in wavelength) to excite the biological cells. Brief experimental conditions for particles and cells separations are listed in Table 1.

Table. 1.

Experimental conditions of particle and cell separations

	Buffer conductivity	AC electrical frequency	Flow rate
Particles separation	2 μS/cm	10 MHz	0.3 ~0.6 μL/min
Cells separation	3000 μS/cm	2.5 MHz	0.15 ~ 1.2 μL/min

4. Experimental results and discussion

4.1. Separation of different size particles

Particle size is a key effective parameter, which influences the DEP force on a particle. Separation of two kinds of particles with different sizes is relatively easy to achieve, due to their obvious DEP force difference at a certain applied frequency. Hence, to demonstrate the separation process in the cascade and staggered side channel DEP cell sorters, two kinds of different size fluorescent particles were first separated in both sorters.

4.1.1. Cascade chip to increase purity

In the experiment of the cascade DEP cell sorter, Q was kept at 0.3 μL/min, and a sinusoidal signal was applied to the two electrode pairs with a 180° phase shift. The voltage and frequency of the applied signal were 20 V_p-p and 10 MHz respectively (At this frequency, the particles experienced high nDEP force [51, 52], measured by a straight forward wedge DEP chip [39]). Under these conditions, the two kinds of particles were separated in the cascade chip, as shown in Fig. 2. The fluid flowed from the inlet at the left side to the outlet at the right side in the main channel.

Figure 2.

Visualization of DEP separation of 6-μm and 3-μm particles in the cascade DEP cell sorter. (a) Without AC activation, all particles flowed mostly through the electrode gap. (b) With applied AC signals, almost all 6-μm particles were repelled to the side channel due to the nDEP force. Then in the second main channel, 6-μm particles were separated again from the 3-μm particles before the second electrode pair. (c) Zoom-in view of particle separation near the first electrode pair. (d) Zoom in view of the separation near the second electrode pair.

As shown in Fig. 2, 6-μm particles were isolated from the premixed solution of the two kinds of particles in the cascade DEP cell sorter. Fig. 2(c) and (d) give clearer view of the separation of both kinds of the particles near the electrode pairs in the first and second main channel. The results suggest that the cascade DEP cell sorter not only has the capability of isolating and enriching target particles, but also has the capability of improving $P$ of isolated particles. Note that the streaklines between the two electrodes in Fig. 2(d) were caused by the particles’ light reflection from the side wall surface (facing the particles) of the glass gap. These particles that generated the streaklines were right upstream the electrode pair. Similar streaklines can be observed in other figures below.

4.1.2. Staggered side channels to increase throughput

In the experiment of the DEP cell sorter having two staggered side channels (each of which has an electrode pair), a relatively low Q of 0.3 μL/min and a relatively high Q of 0.6 μL/min were used.

The voltage and the frequency of the applied signal were 20 V_p-p and 10 MHz respectively. Under these conditions, the two kinds of particles were separated in the DEP cell sorter, as shown in Fig. 3.

Figure 3.

Visualization of separation of the 6-μm and 3-μm in the staggered side channel DEP cell sorter. (a) Without AC activation, all particles flowed through the electrode gap. (b) With applied AC signals, most of 6-μm particles were repelled to the first side channel due to the nDEP force, at low Q of 0.3μL/min. (c) While most of 6-μm particles were repelled to the side channel when the high Q of 0.6 μL/min was used, some 6-μm particles passed over the electrode gap and escaped to the main channel. (d) At the second electrode pair, those escaped 6-μm particles were repelled by the second electrode pair to the second side channel, but 3-μm particles remained flowing down to the end of the main channel.

As shown in Fig. 3, 6-μm particles were separated from the 3-μm ones in the DEP cell sorter with the staggered side channels. These results suggest that this staggered side channel DEP sorter, compared with the single side channel sorter, has the capability of keeping high $\eta$, even though the sample throughput is increased.

4.2. Separation of prostate cancer cells (LNCaP) and colorectal cancer cells (HCT116)

One of the major goals of developing these DEP cell sorters was to manipulate and isolate biological cells. In the present work, we selected LNCaP prostate cancer cells as the target cells for isolation from HCT116 colorectal cancer cells for demonstration. These two types of cells were relatively difficult to separate, since they are similar in diameter (around 20 μm) and are both epithelial cancer cells. The DEP force difference between these two types of cells was mainly due to their different dielectric properties. To separate them, selection of the AC frequency was crucial. According to the measurement of DEP spectrum of each cell type by a wedge chip we fabricated earlier [39], it was found that at the frequency of 2.5 MHz, we could efficiently separate these two types of cells, since at this frequency, while HCT116 cells had a negligible DEP force (the cross-over frequency for LNCaP was 20 MHz), LNCaP cells experienced a relatively strong nDEP force (Unless specialized, the applied frequency was kept at 2.5 MHz). Increase of purity in cascade sorter and increase of throughput without compromising the enrichment factor in the staggered sorter were investigated.

The purity represents the ratio of target particles number $T$ collected in the side channel to the total particles number ($T+N$), collected in the side channel, and the enrichment factor is the ratio of target particles blocked by the nDEP force to the total target particles or cells that flow into the sorter.

4.2.1. Cell separation in the cascade DEP cell sorter

In the experiment of the cascade sorter, the sample Q was kept at 0.15 μL/min, the voltage of applied signal was 12 V_p-p. Visualization of the separation between the LNCaP and the HCT116 cells in the cascade sorter is given in Fig. 4, with and without the electrical activation. In Fig. 4(c), it is shown that in the second main channel, LNCaP cells were again repelled to the side channel due to the nDEP force, but the HCT116 cells were not. By this process, $P$ of isolated LNCaP cells was further enhanced.

Figure 4.

Visualization of the separation of LNCaP and HCT116 cells in the cascade sorter. (a) Without AC activation, most cells flowed along the main channel. (b) With AC activation, LNCaP cells were repelled to the side channel, when a small portion of HCT116 cells also flowed in the side channel (c) Most LNCaP cells were repelled to the second side channel, while most HCT116 cells passed over the gap. The dark spheres were the LNCaP cells, and the bright spheres are the HCT116 cells.

To quantify and characterize the purification of the prostate cancer cells from the cascade sorter, we estimated $P$ of the prostate cancer cells (prostate cancer cells number divided by total cells number) in the side channel. Cells entered into the side channel were counted based on movie obtained from the visualization (the same movie of Fig. 4). Tests were repeated 5 times to obtain average value and error bar. As shown in Fig. 5, purities of LNCaP at the entrance of the side channel were compared for three cases: without AC signal, with AC signal in a single electrode pair sorter, and with AC signal in a cascade configuration sorter.

Figure 5.

Quantitative comparison of LNCaP cell purity at the entrance of the side channel.

Without electric field activation, the LNCaP cells and HCT116 cells had the same possibility to enter the side channel with portion of fluid that flowed into each side channel, therefore, the percentage of LNCaP cells was about 50%, which was identified with the experiment result. As the electrodes were activated, $P$ of LNCaP was increased to 80% in the single sorter, suggesting that most cells repelled to the side channel were prostate cancer cells. However, there were still 20% colorectal cancer cells that flowed into the side channel. Nevertheless, in the cascade sorter, the prostate cancer cells were separated again from the colorectal cancer cells near the electrode pair in the second main channel. Here, $P$ of the LNCaP cells was increased to 96% by the cascade sorter. Compared with the single DEP sorter, $P$ of the prostate cancer cells was increased by 20% in the cascade sorter. After all, since the LNCaP cells were repelled into the second side channel with a smaller volume of suspending medium, the concentration of cells was also increased from 1.5× 10⁶ cells/ml to 4.5× 10⁶ cells/ml. These results are very useful for the situation, where high purity and concentration of target cells is required. From Fig. 5, it is anticipated that with additional increase of the cascade number, $P$ of the target cell can be further improved.

4.2.2. Cell separation in the staggered side channel DEP cell sorter

To measure and quantify the throughput improvement in the staggered sorter, the applied voltage was kept at 12 V_p-p, and Q was changed from 0.1 μL/min to 1.6 μL/min. As shown in Fig. 6, the LNCaP cells were isolated from the HCT116 cells in the single side channel sorter after electrical activation at low Q, compared Fig. 6(a) to Fig. 6(b). Unfortunately, comparing Fig. 6(b) to Fig. 6(c), the number of target cells, which escaped from the first electrode pair, increased with the increased throughput. This effect will reduce $\eta$. However, the design of the staggered side channel sorter, aiming at increasing the throughput without compromising $\eta$, can indeed overcome this problem as shown in Fig. 6(d).

Figure 6.

Visualization of separation between LNCaP and HCT116 cells in the staggered side channel sorter. (a) Without AC activation, both kinds of cells flowed through the electrode gap along the main channel. (b) At relatively low Q of 0.2 μL/min, with applied AC signals, almost all LNCaP cells were repelled to the first side channel due to the nDEP force. (c) When a relatively high Q of 0.4 μL/min was delivered, a small portion of LNCaP cells escaped to the main channel. (d) For the staggered DEP sorter, the escaped LNCaP cells were repelled by the second electrode pair, but the HCT116 cells remained flowing down to the end of the main channel.

We first measured $\eta$ by increasing throughput in the single side channel sorter. As shown in Fig. 7, when Q was lower than 0.2 μL/min, $\eta$ remained at 90%. As Q was increased from 0.2 μL/min, $\eta$ started to decrease. Finally, when Q reached to 0.9 μL/min, $\eta$ decreased to 20%, and the sorter then had no more effect for cell isolation.

Figure 7.

Enrichment factor as a function of sample flow rate.

We then measured $\eta$ by increasing throughput in the staggered sorter. As also shown in Fig. 7, $\eta$ remained at 90% as long as Q was lower than 0.6 μL/min. When Q reached to 1.2 μL/min, $\eta$ decreased to 20%, and the sorter then had no more effect for cell isolation with further increasing of Q. For the single side channel sorter, to remain an enrichment factor of e.g. 90%, Q needed to be lower than 0.2 μL/min. Meanwhile, for the staggered sorter, as long as Q was lower than 0.6 μL/min, the enrichment facor could be kept at 90%. Thus, in keeping the same $\eta$ of 90%, Q was increased three times in the staggered sorter compared with that in the single side channel sorter. These results suggest that, without compromisng $\eta$ the sample throughput was successfully increased in the staggered sorter.

5. Conclusion

In the present work, two microfluidic cell sorters, cascade and staggered have been developed based on conventional negative dielectrophoresis for continuous operations. Carboxylate particles have been first separated in both sorters based on their difference in sizes. Then we have successfully isolated the prostate cancer cells LNCaP from the colorectal cancer cells HCT116 by optimizing electric properties in these cell sorters. $P$ of isolated cells was increased by 20% through adding only one more sorter in the cascade configuration. Sample throughput was increased to three times in the staggered side channel DEP sorter compared with that of a single DEP sorter without compromising the enrichment factor. The present work indicates that, not only biological cells can be separated and isolated from each other by DEP, but also $P$ of the isolated target cells and sample throughput could be increased by the designed cascade and staggered DEP sorter. This would provide new opportunities to enhance DEP sorter performance for lab-on-a-chip applications.