Introduction

Circulating tumor cells (CTCs) found in the peripheral blood originate from solid tumors and are involved in hematogenous metastatic spread [1]. Research on using CTCs as biomarkers has become a hot topic in the cancer community, and the use of peripheral blood as a ‘real-time liquid biopsy’ for the detection, isolation, and characterization of CTC’s continues to be developed as an alternative to standard biopsies [2–6]. Liquid biopsies can be performed repeatedly with low risk of side effects, making them an attractive approach for monitoring cancer progression and response to therapy over time [7–10]. A disadvantage of existing methods to isolate and analyze CTCs is that they often require various transfer, wash, and staining steps, which are time consuming, expensive, and led to the loss of these rare cells [11–14]. To overcome the limitations of existing CTC isolation strategies, we have developed an inexpensive ‘lab on a chip’ device for the enrichment, staining, and analysis of rare cell populations. Our device combines three different principles for the isolation of CTCs, namely immunomagnetic positive selection of antibody-bound cells, isolation of cells through an immiscible interface, and filtration. This combination makes it possible to integrate rare cell isolation with downstream molecular detection methods on a single platform.

Methods and materials

The surface tension and magnetophoretic device is depicted in Fig. 1. The overall device is supported with a base that sits on the laboratory bench. The device consists of a base plate, a sieve, a sieve gasket, and a cap plate that is guided by dowel pins and attaches to the plate. The cap mounts and seals the sieve on the plate.

Fig. 1
figure 1

Surface tension and magnetophoretic device for rare cell isolation

Full size image

Details of the device design are shown in Fig. 2. The base plate includes an input well with a tapered funnel-like surface into a surface tension channel. The surface tension channel links the input and the output well. The well height and widths of both wells are constrained such that the input well has a 500 μL capacity and the output well has a 400 μL capacity. The output well is filled with a washing solution such as phosphate buffered saline (PBS). Furthermore, the output well has a 8.0 μm polycarbonate microporous membrane sieve (PCT8025100 Polycarbonate, Sterlitech Corporation, USA), which is chosen for a pore size that is capable of trapping antibody-bound cells. This membrane sieve is sealed with a gasket under the output well. Furthermore, a syringe attaches to the end of the device, and a drain well located under the channel communicates to the port of this syringe to extract the washing solution by passing it through the filter.

Fig. 2
figure 2

Details of the device: a view, b exploded view, c details, d cross section

Full size image

During operation, the surface tension channel is filled with 10 μL of olive oil, a high surface tension solution. Previous studies have shown that immiscible phase filtration (IPF) holds great promise for the isolation of, e.g., proteins [15], cells [16], and nucleic acids [17]. As the surface tension is more dominant then gravity, the oil acts as a virtual wall between the immiscible phase (channel with olive oil) and both aqueous phases (input and output well with PBS) [18]. As both fluids have a different density, they will not mix and it is thereby not possible for cells or contaminants to move on their own from the input to the output well. By labeling the cells of interest (cancer cells) with magnetic beads and simply passing a magnet underneath the device (from input to output well), only the cells labeled with magnetic beads will be able to overcome this dominant surface tension and will be pulled through the surface channel into the output well. Thereby, it is possible to filter ‘unwanted cells’ in one single step and maintaining the viability of the cells of interest, without the need for centrifugation and multiple washing steps [19].

Cell preparation

Immortalized LNCaP and DU145 prostate cancer cells were utilized to test the ability of the device to isolate cancer cells. Cell lines are maintained at 37 °C and 5% CO2 in RPMI culture media (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (Gibco, USA). Before running experiments, cells were incubated for 15 min at 37 °C with enzyme free, PBS-based Cell Dissociation Buffer (Gibco, USA) and subsequently re-suspended in 2 mL of RPMI. Next, this sample is stained with (1 μL/mL sample) CellTracker Green CMFDA (Life Technologies, Oregon, USA) and incubated for 30 min at 37 °C. This contributed to the visualization of the cytosol of the cancer cell.

Paramagnetic particle preparation

The paramagnetic particles (PMPs) used for the enrichment of the prostatic epithelial tumor cells were 4.5 μm ‘CELLection Epithelial Enrich DynaBeads’ that were coated with anti-EpCAM antibody (ThermoFisher Scientific, Waltham, MA). Furthermore, in separate experiments, 4.5 μm Dynabeads coated with anti-CD45 antibody (ThermoFisher Scientific, Waltham, MA) were used as a negative control to account for nonspecific cell capture. Before using the beads, the stock buffer solution had to be removed from the antibody-labeled PMPs by washing and re-suspending them twice in PBS. In order to attract these PMPs and implicitly the antibody-bound cells, a magnet was manually and freely moved under the device. This motion induced by a magnetic field on a particle of magnetic material in a fluid is called magnetophoresis.

Experimental setup

A known number of stained cancer cells (1000 cells, measured with the Cellometer System (Nexcelom, Bioscience, USA) was spiked into vials filled with 500 μL RPMI. Subsequently, different quantities (0.5, 5.0, 25 or 50 μL) of anti-EpCAM or anti-CD45 PMPs were added to these samples. Next, these cell suspensions were admixed on a shaker at 4 °C for 30 min, while gently twisting and shaking to allow binding.

After 30 min, the prepared cell solution was placed in the input well, and the magnet was placed under this well. With slow advancement of the magnet under the channel toward the output well, bound cells of interest traversed the immiscible oil channel and entered the output well. The syringe was then used to extract the washing solution by passing it through the filter. Once these processes were complete, the filter was removed and handled over on a glass slide (VWR Superfrost Plus, USA), and Diamond anti-fade mounting media containing DAPI (Life Technologies, Grand Island, NY) was used to coverslip the slides.

Slide scanning and analysis

After mounting the membrane on a glass slide, the slides were imaged with Metasystems (Newton, MA) Metafer5—MetaCyte scanning software. Images of the cells were generated by the software based on fluorescence [20]. As we conducted spiking experiments of solely cancer cells (with a fluorescent green cytosol) in RPMI, there was no need to stain for epithelial markers [9, 21]. Therefore, the cell selection algorithm was based on the following criteria: having a nucleus (DAPI+), a green cytosol (CellTracker Green), and a cellular shape (roundness). Extracted image data were then transferred to the MATLAB R2016a program (Mathworks, Natick, MA), which counted the number of captured cells.

Results

Cell-spiking experiments were conducted to determine the recovery rate of the device. Before starting the experiments, EpCAM expression in both cell lines was determined to be high by flow cytometry. These experiments showed an EpCAM expression of nearly 100% in both cell lines.

The initial preclinical spiking experiments showed that both cell types could be captured (Fig. 3) with an increase in recovery up to approximately 95% using 50 μL of anti-EpCAM Dynabeads (n = 10 per concentration for both cell lines) (Fig. 4). Anti-CD45 Dynabeads were also tested as a negative control to account for nonspecific cell capture (n = 5 per concentration for both cell lines). This demonstrated a low rate of recovery across tested concentrations [22].

Fig. 3
figure 3

Example of a prostate cancer cell (DU145) captured and visualized on the membrane. Red, magnetic beads; green, cytosol; and blue, nucleus

Full size image
Fig. 4
figure 4

Percent recovery (%) of: a DU145 prostate cancer cells and b LNCaP prostate cancer cells, using different volumes of anti-EpCAM Dynabeads (n = 10 per concentration), as well as a negative control using anti-CD45 Dynabeads (n = 5 per concentration)

Full size image

Discussion and future directions

Results demonstrated a recovery percentage above 100% in two experiments (Fig. 4). This is likely due to the fact that the Cellometer (which estimates the number of cancer cells in 1 mL of sample) is not 100% accurate, with the result that the number of spiked cancer cells unfortunately could never have been exactly 1000 cells. To increase the reliability of future spiking experiments, a single-cell manipulator can be used by which the preferred number of cells from the plate can be aspirated, whereupon these can be added to the sample. In addition, the software for counting the number of captured cells did not always recognize clusters, adding to potential inaccuracies in counting.

Another potential problem in using anti-EpCAM labeled magnetic beads to select CTCs is the fact that during disease progression the expression of EpCAM and other epithelial cell surface markers may change or decrease [22–25]. Furthermore, several studies have shown that CTCs are not the only cells in peripheral blood on which EpCAM is expressed, as it could also be expressed on cells of hematopoietic lineage [23–25]. Nevertheless, EpCAM was utilized here as a ‘proof of principle’ phenotypic cell marker as this is currently the most common cell surface marker utilized in CTC detection [7].

Furthermore, as patient samples are utilized in future studies, PMPs capable of binding to all the rare cells of interest (e.g., via a linking antibody/streptavidin-labeled Dynabeads) will be utilized. These beads can bind to ‘disease-specific’ markers of interest, which may be expressed during the progression of a patients’ disease [21]. Lastly, we used CellTracker Green as a proof of principle for the sensitivity of the device. When switching to patient samples, this dye will no longer be used for the visualization of the cells, because all cells, including white blood cells, would be labeled. However, following EpCAM-mediated isolation of tumor cells, future strategies for visualization of the cells could include immunofluorescence methods or fluorescence in situ hybridization directly on the membrane of the device.

In this paper, a micro-fluidic device is introduced with which it is possible to isolate tumor cells with a high sensitivity. Future experiments will be conducted to ascertain specificity. The first step will be switching to a cell suspension derived from a biologic fluid. Tumor cells will be spiked into blood (circulating tumor cells) or bone marrow (disseminated tumor cells) and processed as described previously [26–28].

Conclusion

There continues to be a need for simple, noninvasive, and inexpensive tests for the isolation and characterization of circulating tumor cells. This device combines three different methods, namely isolation of cells through an immiscible interface, using the surface tension channel; immunomagnetic positive selection, using the magnetophoretic property of the magnetic-bead-bound cancer cells; and filtration, using the microporous membrane sieve. This combination makes it possible to integrate rare cell isolation with downstream molecular detection methods in a single platform. Initial tests suggest that the device may be well suited for the isolation of CTCs. Processing on one device saves time and reagents and reduces the risk of sample loss. Further development of the presented technology could potentially allow the monitoring of cancer progression and the response to therapy over time and thereby facilitate personalized treatment strategies and improved outcomes.