Quantification of capture efficiency, purity, and single-cell isolation in the recovery of circulating melanoma cells from peripheral blood by dielectrophoresis

This paper describes a dielectrophoretic method for selection of circulating melanoma cells (CMCs), which lack reliable identifying surface antigens and are extremely rare in blood. This platform captures CMCs individually by dielectrophoresis (DEP) at an array of wireless bipolar electrodes (BPEs) aligned to overlying nanoliter-scale chambers, which isolate each cell for subsequent on-chip single-cell analysis. To determine the best conditions to employ for CMC isolation in this DEP-BPE platform, the static and dynamic dielectrophoretic response of established melanoma cell lines, melanoma cells from patient-derived xenografts (PDX) and peripheral blood mononuclear cells (PBMCs) were evaluated as a function of frequency using two established DEP platforms. Further, PBMCs derived from patients with advanced melanoma were compared with those from healthy controls. The results of this evaluation reveal that each DEP method requires a distinct frequency to achieve capture of melanoma cells and that the distribution of dielectric properties of PBMCs is more broadly varied in and among patients versus healthy controls. Based on this evaluation, we conclude that 50 kHz provides the highest capture efficiency on our DEP-BPE platform while maintaining a low rate of capture of unwanted PBMCs. We further quantified the efficiency of single-cell capture on the DEP-BPE platform and found that the efficiency diminished beyond around 25% chamber occupancy, thereby informing the minimum array size that is required. Importantly, the capture efficiency of the DEP-BPE platform for melanoma cells when using optimized conditions matched the performance predicted by our analysis. Finally, isolation of melanoma cells from contrived (spike-in) and clinical samples on our platform using optimized conditions was demonstrated. The capture and individual isolation of CMCs, confirmed by post-capture labeling, from patient-derived samples suggests the potential of this platform for clinical application.


Microfluidic device design and fabrication
The dimensions for two device designs are described here. For the continuous-flow DEP microfluidic chip, the width and length of the microfluidic channel were 2 mm and 13 mm, respectively, and the channel had a height of 25 μm, the width of the sample inlet and buffer inlet was 250 μm and 1.75 mm, respectively. The interdigitated electrode array contained 50 electrodes, 50 μm in width with a gap of 50 μm.
For the DEP-BPE, the 4 parallel microchannels with each being 7 mm long 110 μm × wide 25 μm tall were arranged in parallel and separated by 590 μm. Each channel had 20 × chambers extruded at each side (160 chambers). Each chamber was 110 μm diameter and edge-toedge distance of two adjacent chambers was 330 μm. The microchannels were interconnected to a common inlet and outlet by a bifurcation (branching) scheme. A leak channel (7 μm wide) was affixed to each chamber to make a fluidic connection to the main channel.
The features were designed using AutoCAD (Autodesk, San Rafael, CA) and written on a chrome mask. Soft-lithography was used to pattern the structures of the continuous-flow DEP and DEP-BPE device. Negative photoresist (SU-8 2025, MicroChem Corp., MA, USA) was spincoated onto a four-inch silicon wafer and was subsequently subjected to several physicochemical processes, namely soft baking, UV light exposure, post-exposure baking, development, and hard baking. As a result, the SU-8 master template was formed on a silicon wafer with a thickness of 25 μm to the proposed design. Next, for replica molding, an instant barrier was made by wrapping the master silicon wafer in aluminum foil. A 10:1 volumetric mixture of PDMS (Sylgard 184, Dow Corning Corp., MI, USA) and a curing agent were then poured onto the master wafer. After degassing the polymer mixture, the master wafer overspread with clear PDMS was cured in room temperature, then the PDMS replica was removed from the master wafer and perforated at the channel inlet and outlet using a punch. The punch was used 1 mm for inlet and 2 mm for outlet.
The DEP electrodes were fabricated on a four-inch glass wafer using a conventional photolithography process. Photoresist (PR, AZ4620) was spin-coated at 3000 rpm for 35 seconds to achieve approximately a thickness of 7 μm, exposed under UV light and developed to define the electrode patterns. AZ developer, gold etchant and chrome etchant were used successively to define the electrodes, and NMP was used to remove the patterned PR. Lastly, the PDMS replica was assembled with fabricated electrode patterns onto the glass wafer after air plasma treatment using a plasma cleaner and was stored overnight in an oven (65 ℃) to ensure permanent bonding.
Each device was pretreated with 3 μM pluronic F-127 solution and incubated at 4 ℃ overnight to effectively inhibit cell adhesion on PDMS and electrodes. Each sample was run using a new disposable polydimethylsiloxane (PDMS) chip to prevent contamination.

Biological material and blood
Four melanoma cancer cell lines, A375 (V600E, homozygous), SK-MEL-1 (V600E, heterozygous), SK-MEL-2 (WT) and SK-MEL-28 (V600E, homozygous) (American Type Culture Collection, Manassas, VA) and a cell line (PDX-10) derived from a patient-derived melanoma xenograft at the University of Iowa were used to characterize DEP response. They were cultured in DMEM media with 1% non-essential amino acid and 10% fetal bovine serum (FBS) supplementation at 37 ℃ and 5% CO 2 in a humidified environment. In preparation of DEP experiments, A375, SK-MEL-2 and SK-MEL-28 were detached from culture flask using 0.25% Trypsin-EDTA (1x), followed by pelleting by centrifugation (1100 rpm, 5 min), washed with culture media or DPBS and resuspended in working buffer. The dielectric characteristics of the cell lines were measured within 8 passages after initial receipt.
Whole human blood, individually drawn from healthy donors, was provided by University of Iowa. Each 4-mL draw was collected into one Strek tube containing K 3 EDTA anticoagulant, stored at 4 ℃ before shipping, and used right after receiving. PBMCs were processed using density gradient separation, whole blood cells were first floated on Ficoll-Pague (GE Henlthcare Bio-Sciences Corp., Uppsala, Sweden) and then centrifuged for 40 min at 400 xg, followed by careful isolation of the buffy coat. The isolated buffy coat was washed and diluted in DPBS to remove residual Ficoll-Pague. The final step was an additional centrifugation for 10 min at 100 xg to reduce contamination of the platelets. Each melanoma cell line and PBMC viability was evaluated with a trypan blue. The cell sample was diluted in trypan blue (1:1 ratio) and incubated for 2 min. 20 μL of the sample was then transferred into the cell counting chamber slide, then the slide was inserted into Countess™. Viable cells remain unstained and non-viable turn blue.
Patients with Stage IV metastatic cancer were recruited according to the protocol approved by the University of Iowa's Institutional Review Board. PDX-10 samples were obtained from University of Iowa.

Cell spike-in experiment
The number of melanoma cells spiked in will be controlled either by serial dilution (for spike in of 100-1000 cells) or by spiking in from a capillary (for <100 cells). The capillary spikein experiments were followed as Zhao's method.

Characterize the DEP response of melanoma cells and PBMCs using 3DEP system
The working buffer ("DEP buffer") was made right before each experiment, it is comprised of 8.0% sucrose, 0.3% dextrose, and 0.1% BSA in 1.0 mM Tris buffer (pH 8.1) and used within 72 h. DEP spectral measurement was performed with unlabeled cells on a 3DEP dielectrophoretic analyzer (DepTech, Uckfield, U.K.), experimental methods and data analysis followed previously published procedures. (3,4) Briefly, approximately 80 μL of cell suspension was injected into the DEP well chip (3DEP 806), and a cover glass placed on top to avoid the formation of a meniscus, due to surface tension, which could interfere with the measurement of light intensity changes. The chip was mounted on a camera setup where a light beam is directed from the top of the chip, and the door was closed to prevent interference from ambient room lighting. A recording interval was set to 30 s at 10 V pp , with data collected over 20 points between 1 kHz and 45 MHz. This procedure was repeated for 15 distinct samples (for cultured cells, 5 samples from each of 3 culture flasks).

The intensity of light changes depending on the movement of the cells by the DEP force.
Light intensity, in each microwell, was measured for 30 s sweeping a frequency range from 1 kHz to 45 MHz. The light intensity vs frequency spectrum generated is fitted by the 3DEP software to a core-shell model using an iterative least square method to yield values of membrane and cytoplasmic conductivity as well as specific membrane capacitance, . 3DEP light = / intensity bands from 4 to 9 only, from the platform spectrum output, were selected for each experiment to be fitted, resulting in a better DEP spectrum with Pearson correlation coefficient, R > 0.95. Cells were measured immediately (< 20 min) after being suspended into the DEP media to minimize artifacts due to cell stress by the non-physiological ion composition of the medium.
While cells can respond rapidly to their environment, the DEP response of cells is stable over this time period.

Quantify the DEP response of melanoma cells using continuous-flow DEP device
The experimental test protocol was established as follows: (1) prior to all experiments, the channel, syringe and tubes were coated with 3 μM Pluronic solution overnight to reduce cell adhesion to the microchannel and electrode surfaces. The channel was washed using the DEP buffer for 10 minutes. (2) Cell samples were injected into the device from the sample inlet. DEP buffer was injected into the buffer inlet through tubes connected to a syringe pump. Sample flow rate and buffer flow rate were fixed at 200 nL/min and 1 μL/min (5:1 sheath to sample flow ratio), respectively. (3) A waveform generator, connected to the DEP device via driving electrodes, was used to apply AC voltage. The AC field was alternated between on and off. (4) Experimental results were observed using the Nikon AZ100 fitted with an Andor Zyla 5.5 sCMOS camera. Each experiment was performed to process at least a total volume of 120 μL.
To quantify the cell response for the entire frequency range (5-200 kHz) at resolution as 5 kHz, the cell concentration was maintained at 2 10 6 cells/ mL, the flow rate of cell solution was × fixed at 200 nL/min, each frequency was maintained as least 1 min and frequency was sweeping from 5 kHz to 200 kHz then back to 5 kHz, the total number of examined cells was around 2.5 10 4 for each experiment. × To test the high throughput of DEP-FFF, a different cell concentration was made and tested. The cell concentration was increased to 5 10 6 to 1 10 7 cells/ mL, the other experimental × × parameter was set as the same. However, when high concentration was used, the aggregation of cells was observed and the isolation efficiency was decreased. To quantify the incidence of unwanted PBMC capture in the DEP-BPE device. The PBMCs was re-suspended at a concentration of approx. 10 7 cells/100 μL in DEP buffer followed by injection of 20 μL (around 2 10 6 cells) into the DEP-BPE device under the same conditions employed × for melanoma cells. The fraction captured will be quantified by counting the pockets occupied by cells.   . Immuno-labeling effect was evaluated by pre-labelling PDX-10 cell with each staining molecules, nuclei stain has a big effect on DEP response, as shown in (B). Figure S3. DEP response measurement was taken when blood sample was stored at room temperature for 3 days, the cell diameter measured was 6 μm, which could be reasonable considered as red blood contamination.