Reducing WBC background in cancer cell separation products by negative acoustic contrast particle immuno-acoustophoresis

Abstract

Cancer cells display acoustic properties enabling acoustophoretic separation from white blood cells (WBCs) with 2-3 log suppression of the WBC background. However, a subset of WBCs has overlapping acoustic properties with cancer cells, which is why label-free acoustophoretic cancer cell isolation needs additional purification prior to analysis. This paper reports for the first time a proof of concept for continuous flow acoustophoretic negative selection of WBCs from cells cancer cells using negative acoustic contrast elastomeric particles (EPs) activated with CD45-antibodies that specifically bind to WBCs. The EP/WBC complexes align at the acoustic pressure anti-nodes along the channel walls while unbound cancer cells focus to the pressure node in the channel center, enabling continuous flow based depletion of WBC background in a cancer cell product. The method does not provide a single process solution for the CTC separation challenge, but provides an elegant part to a multi-step process by further reducing the WBC background in cancer cell separation products derived from an initial step of label-free acoustophoresis. We report the recorded performance of the negative selection immuno-acoustophoretic WBC depletion and cancer cell recovery. To eliminate the negative impact of the separation due to the known problems of aggregation of negative acoustic contrast particles along the sidewalls of the acoustophoresis channel and to enable continuous separation of EP/WBC complexes from cancer cells, a new acoustic actuation method has been implemented where the ultrasound frequency is scanned (1.991 MHz ±100 kHz, scan rate 200 kHz msec⁻¹). Using this frequency scanning strategy EP/WBC complexes were acoustophoretically separated from mixtures of WBCs spiked with breast and prostate cancer cells (DU145 and MCF-7). An 86-fold (MCF-7) and 52-fold (DU145) reduction of WBCs in the cancer cell fractions were recorded with separations efficiencies of 98,6% (MCF-7) and 99.7% (DU145) and cancer cell recoveries of 89.8% (MCF-7) and 85.0% (DU145).

1. Introduction

Detectable shedding of malignant cells from a primary tumor into the blood circulation is associated with metastatic disease, which is the major cause of cancer-related deaths [1, 2]. Not all disseminated cells have tumor-initiating capacities, but the number of circulating tumor cells (CTCs) found in patient blood during the course of treatment is prognostic of overall survival [3-6]. Besides already serving as a prognostic biomarker for several metastatic carcinomas, molecular analysis of isolated CTCs holds promise as predictive and efficacy-response biomarkers [7, 8]. Clinically relevant information indicative of treatment benefit, from a minimally invasive sample source, would provide clinicians with an effective tool to guide patient management [9]. Efficacy-response biomarkers can assist in the development of new and effective therapeutic treatments for cancer patients by reducing clinical trial timelines and costs [10]. Hence, CTC detection and isolation is a highly desirable complement in diagnosis and patient stratification. However, CTCs are present in very low numbers in patients with advanced metastatic disease and there is currently an unmet need for sensitive label-free technologies to isolate and enrich these rare cells for molecular profiling.

Recent technical advances have led to several new assays and devices, including microfluidic approaches, to capture, enumerate and characterize CTCs [11-34], but still only one method has been cleared by the US Food and Drug Administration (FDA) to aid in the management of cancer patients, CellSearch (Veridex, NJ, USA) [35, 36]. Microfluidic methods, however, commonly lack in performance in one or several of the required parameters: throughput (≥5 mL/hr), recovery (≥90%), and purity (≈1 CTC/100 WBC). The CellSearch system enriches CTCs by immuno-capturing with magnetic beads activated with antibodies for Epithelial Cell Adhesion Molecule (EpCAM), followed by positive staining for cytokeratin (CK), nuclei (DAPI), and negative staining for CD45. This positive selection strategy is the most common approach for CTC enrichment as epithelial-derived cancer cells commonly express EpCAM, which is absent in leukocytes and erythrocytes, making it a suitable target for positive enrichment of CTCs. However, subpopulations of CTCs with reduced or absent expression of EpCAM would be missed in assays using CellSearch [37]. Cells with this phenotype have been suggested to cause metastasis and progression of disease based on stem cell properties or increased migratory capacity due to epithelial to mesenchymal cell transformation (EMT) [14, 38]. Furthermore, the CellSearch system suffers from background contamination of WBCs (1000-5000 WBCs) when 7.5 mL of whole blood is processed; which will interfere with downstream molecular profiling tumor cells [39]. As a result, alternative methods are needed to more accurately reflect a diverse spectrum of CTC-characteristics using unbiased, label-free, and gentle isolation strategies with reduced WBC background levels. Immuno-magnetic negative selection has been extensively investigated by Chalmers et al. [40, 41] as a means to deplete WBCs from RBC lysed whole blood samples and enable label free isolation of cancer cells. WBC depletion levels of log 2.5-3 is typically reported, which still leaves a need for further reduction of WBC background. A microfluidic negative selection method (CTC-iChip system) was developed for CTC enrichment by Ozkumur et al [19]. Using this method, CTCs were enriched by depleting WBCs with magnetic beads functionalized with anti-CD45 and CD15 IgG antibodies. Though, the CTC-iChip obtained high recovery of spiked cancer cells (> 90%) with 2.5 log depletion of WBCs, the resulting cancer cell fractions still contained significant background contamination of WBCs (17,264-39,172 WBCs mL⁻¹). Other microfluidic methods for label-free CTC separation with high recoveries display similar challenges in yielding a significant WBC background in their separation products [33, 42, 43].

In this perspective, acoustophoresis offers a label-free, high-throughput, non-contact and gentle microfluidic technique that uses acoustic radiation forces to separate microparticles or cell populations [18, 44-49]. Exploiting acoustic property differences between cell populations, cancer cells have been successfully separated from WBCs in RBC lysed blood using acoustophoresis [18, 33, 50]. However, due to overlapping acoustophysical properties between cancer cells and WBC subpopulations a significant number, ≈10³-10⁴ WBC mL⁻¹, of contaminating WBCs commonly remain in the tumor cell fraction. To further improve the tumor cell to WBC ratio following an initial label-free acoustophoresis separation, we have herein developed an immuno-affinity negative acoustic contrast microbead based method to deplete WBCs from a cancer cell separation product.

The conversion of positive contrast properties of cells to a negative acoustic contrast complex using negative contrast particles was first demonstrated by Shields et al [51]. In addition, negative contrast particles have been modified with ferrofluids to generate both negative contrast and magnetic responses under acoustic and magnetic fields [52]. Negative acoustic contrast elastomeric particles (EPs) have been synthesized with Sylgard 184 and used for biomarker (prostate specific antigen: PSA) and particle trapping assays with acoustophoresis [53, 54]. However, using negative acoustic contrast particles to trap cells at pressure antinodes during acoustophoresis does not enable continuous flow based separations. This is due to the inherent effects of aggregation of negative acoustic contrast particles in acoustic hot spots along the microchannel side walls. The aggregation of negative contrast particles at the side walls causes a distortion of laminar streamlines and separation, earlier reported in efforts to separate lipid particles (with negative acoustic contrast) in milk samples, Grenvall et al. [55]. To alleviate the inherent problems of sidewall aggregation Grenvall suggested to operate the acoustics at higher harmonics, which allowed focusing of the negative contrast particles to high flow rate streamlines well distanced from the sidewalls [55, 56]. This was later also investigated by Faridi et al. in a system using antibody activated negative acoustic contrast microbubbles to move microbubble/cell-complexes to the pressure antinode [57]. The use of higher harmonics, however, increases requirements on precision in flow control as the lateral distance between pressure nodes and antinodes in the standing wave field becomes significantly smaller, leading to an increased risk for carry-over between the streamlines at the outlet flow splitter.

As an alternative solution to solve the problems with side wall aggregation of negative acoustic contrast particles we demonstrate for the first time continuous flow based acoustophoretic negative selection of WBCs from cancer cells using anti-CD45 activated negative acoustic contrast elastomeric particles (EPs) in a λ/2 acoustophoresis configuration, where a frequency modulation of ±100 kHz, scan rate 200 kHz msec⁻¹, around a 1.991 MHz centre frequency suppressed sidewall aggregation. This report does not claim to describe a system that can isolate tumor cells from whole blood but rather a method that can complement a primary tumor cell separation step that still yields a significant WBC background. The described acoustophoretic immuno-affinity negative selection enabled label free tumor cell (and MCF-7 DU145) isolation from a WBC background with tumor cell enrichment factors between 52-86 times at separation efficiencies of 99% and tumor cell recoveries ranging between 85-90%.

2. Materials and Methods

2.1. Manufacturing of Acoustophoresis Chip & Instrument Setup

The acoustophoresis chip was manufactured using methods previously described [18]. Briefly, the microchannel where the sheath buffer enters has a length of 10 mm; a width of 300 μm; and a depth of 150 μm. The main separation channel where the cell mixture with activated EPs enters has a length of 20 mm; a width of 375 μm and a depth of 150 μm. The piezo ceramic (PZT) was actuated using a function generator (33120A, Agilent Technologies Inc., Santa Clara, CA, USA) connected to power amplifier circuitries (LT1012, Linear Technology Corp., Milpitas, CA, USA) where the voltage applied onto the PZT was measured with an oscilloscope (TDS 1002, Tektronix UK Ltd., Bracknell, UK). The temperature of the acoustophoresis chip was monitored by a PT100 resistance temperature detector and kept at 25 °C using a Peltier element.

2.2. Synthesis of Biotinylated EPs

Polydisperse EPs were synthesized using a bulk oil-in-water emulsion process that was adapted from our previously reported methods [53, 54]. Sylgard 184 (1 gram at a 10:2 ratio of Sylgard 184 prepolymer to curing agent) was emulsified in 10 mL of MilliQ water containing biotinylated-Pluronic F-108 surfactant (50 mg) using a homogenizer at 24K RPMs for 60 seconds. The elastomeric droplets (uncrosslinked) with the biotinylated-Pluronic F-108 surfactant were thermally-cured at 45°C for 1.5 hours with continuous stirring, followed by incubation at room temperature (RT) with continuous stirring for a minimum of 12 hours to form fully crosslinked biotinylated-EPs. Biotinylated EPs were then fractionated using centrifugation at 100 x g for 10 seconds to pellet excessively large particles. The removed supernatant was then centrifuged at 100 x g for 4 minutes (repeated four times). The supernatant (containing submicron particles) was removed and the pellet was resuspended in washing buffer (PBS at pH 7.4 containing 0.1% weight BSA, 0.1% weight Pluronic F-108 surfactant, and 2mM EDTA) to obtain micron-sized polydisperse EPs that were now appropriately-sized for acoustophoretic handling.

2.3. Activation of EPs with CD45 Antibodies

Biotinylated-EPs (4 x 10⁶) were incubated with 1.5 μL of streptavidin conjugated CD45 monoclonal antibody (mouse anti-human; BD Bioscience) in 1 mL of washing buffer for 1 hour at RT. The antibody had previously been covalently coupled to streptavidin by using a commercially available streptavidin labelling kit (AbCam® Sweden, Cat. #: ab102921). Elastomeric particles were then washed two times (100 x g for 4 minutes) using 1 mL of washing buffer per wash. The pelleted EPs were resuspended in 1 mL of washing buffer. To probe for the presence of streptavidin-conjugated CD45 antibody, an aliquot of the resuspended EPs were surface stained with 2 μL of a fluorochrome-conjugated detection antibody (goat anti-mouse IgG antibody: BD Bioscience) in the dark for 30 minutes at RT. The particles were then washed twice (100 x g for 4 minutes) using 1 mL of washing buffer per wash. The pelleted EPs were then resuspended in 1 mL of washing buffer and subjected to evaluation of functional characteristics using flow cytometry.

2.4. Healthy Blood Donor and Cell Preparations

Whole blood samples anticoagulated with ETDA were acquired from healthy donors at the blood donor center at Lund University Hospital (Lund, Sweden). The collected blood was processed to preserve viable WBCs. Red blood cell lysis was performed by adding 2 mL of 1x lysis buffer (BD Pharm Lyse™) to 200 μL of whole blood, followed by gentle vortexing and incubation for 15 minutes at RT. Cells were then washed with 2 mL of washing buffer and centrifuged at 200 x g for 5 min. The pelleted WBCs were resuspended in washing buffer and cell concentration determined by flow cytometry.

2.5. Cell Culture, Immunostaining and Spiking

The human cancer cell lines MCF-7 (breast cancer) and DU-145 (prostate cancer) were acquired from the American Type Culture Collection (ATCC) and cultured according to ATTC recommendations in media supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich), 55 IU mL⁻¹ penicillin and 55 μg mL⁻¹ streptomycin (Sigma-Aldrich) at 37°C in air containing 5% CO₂. Harvesting of the cells was done with trypsin/EDTA treatment for 5 min, followed by washing with the washing buffer previously described. The cancer cells were resuspended in 80 μL of washing buffer and surface stained with 20 μL of fluorochrome-conjugated EpCAM antibody (BD Bioscience) in the dark for 30 min at RT. The cancer cells were thereafter washed and resuspended in washing buffer before cell concentration was finally determined by flow cytometry. We spiked 1.00 x 10³ non-fixed breast cancer cells (MCF7) or prostate cancer cells (DU145) into 1.0 mL of washing buffer containing 1.00 x 10⁵ WBCs for acoustophoresis enrichment experiments. At lower cancer cell spiking levels the counts in the FACS data becomes noisy, which is why a relatively high spiking level was chosen to ensure robust data that specifically described the performance of the negative selection immuno-acoustophoresis separation. It should however be noted that clinical samples with high CTC counts can reach up to 10³ mL⁻¹.

2.6. Activated EPs Capture WBCs in an RBC-lysed Sample

A 30-fold excess of EPs were incubated with 1.00 x 10⁵ WBCs (stained with fluorescent CellTrace™ Oregon Green™ 488) in 1 mL washing buffer for 1 hour. The sample was centrifugally-washed 3x and then resuspended in 1 mL of the washing buffer. Images were taken with an Olympus BX63 microscope controlled by CellSens Dimension software.

2.7. Enrichment of Cancer Cells Using Activated EPs with Acoustophoresis

Activated EPs were added to cell samples in a 1:30 ratio (WBC:EP). The samples were incubated at RT with continuous rotation for 1 hour in the dark. After incubation, samples were processed in the acoustophoresis chip according to Fig. 1, (sample flow input: 100 μL min⁻¹; sheath flow input 400 μL min⁻¹; center outlet flow: 100 μL min⁻¹, side outlet flow: 400 μL min⁻¹). The acoustophoresis chip was operated at 1.991 MHz and 20 V_peak-to-peak and a frequency modulation of ±100 kHz, scan rate 200 kHz/1 msec, to avoid trapping of the negative acoustic contrast complexes along the side walls. All samples collected from the side and central outlets were enumerated for cellular content using flow cytometry (BD FACS Canto™ II). White blood cells were counted using FSC-A vs. SSC-A scatter plots and cancer cells were identified using fluorescence. Gating controls for WBC and cancer cell enumeration are available in the Supplemental Figure S1 (Supplemental Figure S1).

Figure 1:

Schematic representation of an acoustophoresis chip that is operated using frequency modulation for cancer cell enrichment. Sheath liquid buffer enters from the side inlet while sample containing cancer cells (red), EPs (blue) and EP/WBC complexes (blue/violet) enter through the center inlet. Upon entering the main separation microchannel, EP/WBC complexes are focused to the pressure antinodes and cancer cells to the middle pressure node by the acoustic radiation force. Enriched cancer cells can thereby be collected from the center outlet, while EPs and EP/WBC complexes are collected from the side outlet.

3. Results and Discussion

3.1. Instrument Setup and Acoustophoretic Separation Approach

In contrast to the setup used by Augustsson et al. [18], samples entered through the center inlet and sheath buffer through the side inlet (Figure 1). Upon activation of the ultrasonic acoustic standing wave field EP/WBC complexes were continuously focused to pressure antinodes along the side-walls and cancer cells to the pressure node in the center of the microchannel. The frequency modulation was used to eliminate localized acoustic hot spots [55] that can cause excessive aggregation of EP/WBC complexes at pressure antinodes along the side walls of the microchannel. Removal of localized hot spots allowed for WBC depletion and cancer cell populations to be continuously recovered and enriched in the center outlet for further analysis, Figure 2. The insert images show the operation of the acoustophoresis system with frequency modulation inactive (left insert) and active (right insert), clearly illustrating the aggregation problem when not using frequency modulation.

Figure 2.

Still image of aggregation of negative contrast particles at hot spots along the acoustophoresis channel sidewall (left insert) when operating at a fixed actuation frequency and the corresponding situation when employing frequency modulation (right insert).

3.2. Synthesis of Biotinylated EPs

Polydisperse biotinylated EPs were synthesized with a 10:2 mixing ratio of Sylgard 184 prepolymer to curing agent using a facile bulk emulsion method in the presence of a biotinylated-Pluronic F-108 surfactant. This ratio was chosen based on previous work showing EPs (synthesized with a 10:2 ratio) being focused to pressure antinodes and blood cells being focused to the middle pressure node with high separation efficiency using acoustophoresis [54]. Elastomeric particles after centrifugation based fractionation had a size histogram (measured by Coulter Counter) showing a mean diameter of 7.35 μm with a standard deviation (SD) of ±2.86 μm, and a 39.0% coefficient of variation (CV) (Figure 3a). This centrifugation-based fractionation process was crucial for removing excessively large EPs that may clog the microchannel and submicron EPs that would negligibly contribute to the negative selection process.

Figure 3:

Synthesis and activation of biotinylated EPs for WBC binding. (a) Size histogram of EPs after centrifugation based fractionation. (b) Schematic illustration showing a biotinylated EP activated with a streptavidin conjugated monoclonal anti-CD45 IgG that has bound a WBC. (c) Fluorescent images of activated anti-CD45 IgG EPs bound to non-fixed WBCs (stained with fluorescent CellTrace™ Oregon Green™ 488) derived from RBC lysed blood.

3.3. Activation of EPs with anti-CD45 Antibody

The adsorption of biotinylated surfactant on the surface of the EPs during emulsification and curing, allows for active biotinylated groups to exist on the surface of the EPs. The presence of active biotin groups, allows for straightforward activation of EPs with streptavidin conjugated anti-CD45 monoclonal antibodies for capturing non-fixed WBCs (Figure 3b). To verify the attachment of anti-CD45 IgG to the surface of biotinylated EPs, a fluorescence-labelled detection antibody was employed in conjunction with flow cytometry analysis. Using this detection method, activated EPs had a mean fluorescence intensity (MFI) of 53910, where non-activated EPs had a MFI of 59 and untreated biotinylated EPs had a MFI of 12. The flow cytometry detection assay demonstrated successful activation of the biotinylated EPs with streptavidin-conjugated anti-CD45 IgG. Once EPs were activated with anti-CD45 IgG, their ability to bind non fixed WBCs was demonstrated by incubation with WBCs (derived from RBC-lysed blood) and imaged using fluorescence microscopy (Figure 3c).

3.4. Enrichment of Cancer Cells Using Activated EPs with Acoustophoresis

To investigate the number of activated EPs (anti-CD45 IgG) that is needed for efficient WBC depletion, a titration of the EP concentration was performed in samples containing WBCs (derived from RBC-lysed blood), Figure 4. The WBC control count was derived from 200 uL input sample volumes, reduced with the residual sample volume of the microfluidic system which is approximately 100 uL, why a maximum of 10⁴ WBCs in the control could be expected, not accounting for residual volumes in the sample tube and approximations of the cell concentration.

Figure 4:

Titration of anti-CD45 activated EP vs WBC ratio. A titration series of anti-CD45 activated EPs with WBCs (2.00 x 10⁴ in 200 μL of washing buffer) was performed to obtain the best concentration of activated EPs for efficient depletion of non-fixed WBCs. A WBC control (without activated EPs added) was used as a baseline for relative comparison with samples containing increasing numbers of activated EPs. The notation (1:5, 1:10, 1:20, 1:40, and 1:80) indicates the ratio of WBCs:EPs that were used in the samples. All samples were incubated with agitation for 1 hour prior to acoustophoresis and enumeration with flow cytometry. These results were used to extrapolate that a 30-fold excess of activated EPs to WBCs (or a 1:30 ratio of WBCs to EPs) was sufficient to obtain efficient depletion of WBCs in the RBC lysed samples.

The titration of activated EPs (anti-CD45 IgG) with WBCs demonstrated that a 1:30 ratio of WBCs to EPs (30-fold EP excess) was required for efficient WBC depletion from the center fraction. Performing the corresponding assay to enrich CTCs in undiluted RBC lysed whole blood would not be an option since a WBC to EP ratio of 1:30 would require 120-330 x 10⁶ Eps mL⁻¹ with a WBC content in whole blood of about 4-11 x 10⁶ WBCs mL⁻¹. At such high particle concentrations a volume fraction greater than 1% would prevail and at such high particle volume fractions, hydrodynamic particle-interaction effects become significant, leading to decreased separation performance and particle mobility and increased suspension viscosity [58]. These are all effects that sets the upper particle concentration limit in any external force field based particle separation technology. However, the proposed cancer cell purification method using CD45-activated EPs, in this paper, is designed as the 2^nd step in a tumor cell enrichment sequence where a primary (1^st step), label-free, acoustophoretic step, will reduce the WBC background to 10³-10⁴ WBCs mL⁻¹ [50, 59] and hence the critical particle concentrations are not superseded.

To demonstrate a proof-of-concept of the negative selection immuno-acoustophoretic WBC depletion method proposed in this paper, the performance in the enrichment of cancer cells using anti-CD45 activated EPs with frequency modulated acoustophoresis was monitored (Figures 5 and 6). In all experiments, anti-CD45 activated EPs (+CD45) and non-activated EPs (-CD45) were incubated with non-fixed cell mixture samples containing approximately 1.00 x 10³ cancer cells and 1.00 x 10⁵ WBCs prior to acoustophoresis. This relatively high level of WBC was chosen to cover, with an order of magnitude, the highest expected WBC background level after an initial label-free acoustophoretic cancer cell separation step. The high cancer cell spiking was chosen to minimize the noise in the subsequent FACS analysis and specifically reveal information of the performance of the negative selection immuno-acoustophoretic WBC depletion method. Successful acoustophoresis separation of cancer cells at clinically relevant spiking levels (50 cancer cells in 5 mL RBC lysed blood) have recently been reported at recoveries up to 87% [46].

Figure 5:

Label-free enrichment of MCF-7 breast cancer cells using activated EPs (anti-CD45 IgG) with frequency modulated acoustophoresis. (a) Measured numbers of WBCs identified in the center fraction for cell mixtures (1.00 x 10⁵ WBCs and 1.00 x 10³ MCF-7 cells in 1 mL) that were untreated (Mixture) along with mixtures that were treated with non-activated EPs (− anti−CD45) and activated EPs (+ anti−CD45). (b) Separation efficiency (%) of the collected MCF-7 cells in the center and side fractions. (c) MCF-7 cell recovery (%) in the center fraction. (d) The number of WBC and MCF-7 cells in the center fraction for untreated (Mixture) samples versus cell mixtures treated with activated EPs. Note: The values given are the mean values and the error bars (standard deviations) calculated with N=3 (experiments were performed in triplicate).

Figure 6:

Label-free enrichment of DU145 prostate cancer cells using activated EPs and frequency modulated acoustophoresis. (a) Measured numbers of WBCs identified in the center fraction for cell mixtures (1.00 x 10⁵ WBCs and 1.00 x 10³ DU145 cells) that were untreated (Mixture) along with mixtures that were treated with non-activated EPs (−CD45) and activated EPs (+CD45). (b) Separation efficiency (%) of the collected DU145 cells in the center and side fractions. (c) DU145 cell recovery (%) in the center fraction. (d) The number of WBC and DU145 cells in the center fraction for untreated (Mixture) samples versus cell mixtures treated with activated EPs. Note: The values given are the mean values and the error bars (standard deviations) calculated with N=3 (experiments were performed in triplicate).

System performance was evaluated by FACS measurement of the number of WBCs and cancer cells recovered in the center fraction after passing through the acoustophoresis chip for: a) the cell mixture without EPs (control), denoted Mixture in the bar graphs, b) the cell mixture with non-CD45 activated EPs denoted −CD45 (control sample to monitor influence of nonspecific binding) to and c) the cell mixture with CD45 activated EPs denoted +CD45 (to monitor efficiency in depleting WBCs)

Using negative acoustophoresis selection of WBCs from MCF-7 spiked samples with activated EPs (+CD45) showed an average of 86 ± 30-fold WBC depletion compared to the corresponding cell mixture samples without EPs (Figure 5a). By contrast, MCF-7 spiked samples treated with non-activated EPs (−CD45) as controls showed no depletion of WBCs. These results clearly demonstrated that activated EPs (+CD45) specifically depleted WBCs from the collected center fraction and that the nonspecific binding (−CD45) was negligible.

The separation efficiency [((# in center outlet) / (# in center outlet + # in side outlet)) * 100] of spiked breast cancer cells was found to be on average 98.6 ± 2.9% considering all samples (Figure 5b). This high separation efficiency demonstrated a low nonspecific depletion/loss of cancer cells from the center fraction due to the acoustophoresis. Along with high separation efficiency, the cancer cell recovery (input to output) displayed an average MCF-7 recovery of 89.8 ± 5.3% in the center fraction (Figure 5c). Most of the recovery loss of cancer cells can be explained by the residual-volume (~100 μL) of the microfluidic system. Since the sample volumes used were 1 mL, a 10% recovery loss of cells (WBCs as well as cancer cells) prevails, i.e. a maximum theoretical recovery of 90% could be anticipated taking the residual volume loss into account. Additional recovery losses may be due to remaining aliquots in the sample tube as well as cell loss at the sample tube wall. Using samples with larger volumes (e.g. 5-10 mL, which are the sample volumes expected from the primary acoustophoresis step) would further reduce the residual volume recovery loss proportionally, i.e. a 10 mL sample would have an intrinsic recovery loss due to system residual-volume of only 1%

The ability to specifically and selectively deplete WBCs offers a straightforward method for high-throughput cancer cell enrichment following an upstream acoustofluidic cancer cell pre-separation step. Here, the average 86-fold depletion of WBCs with high separation efficiency and recovery of cancer cells in the center fraction, led to an average 84 ± 30-fold enrichment of the breast cancer cells (Figure 5d). The cell mixture without EPs, collected in the center fraction, contained an average of 86634 ± 4055 WBCs and 895 ± 59 MCF-7 cells after being processed through acoustophoresis chip as a control. In comparison, cell mixtures treated with activated EPs (+CD45) displayed an average recovery in the center fraction of 1108 ± 469 WBCs and 868 ± 54 MCF-7 cells after the corresponding acoustophoresis step. This reduced ratio of WBCs to cancer cells yielded an average 84-fold enrichment of cancer cells in the center fraction.

To address the anticipated diversity in CTCs that are derived from different metastatic cancers (e.g. breast and prostate), the corresponding selective WBC-depletion was also investigated using the prostate cancer cell line DU145. Similar to samples spiked with MCF-7 breast cancer cells, cell mixture samples contained 1.00 x 10³ prostate cancer cells along with 1.00 x 10⁵ WBC mL⁻¹ and were treated with non-activated and activated EPs. The results showed an average of 52 ± 23-fold WBC specific depletion from the center fraction when cell mixtures were treated with activated EPs (Figure 6a). In addition to obtaining efficient WBC depletion, nearly all the collected DU145 cells were found in the center fraction, displaying a 99.7 ± 0.9% average separation efficiency for prostate cancer cells (Figure 6b). Also, the cancer cell recovery was high, an average of 85.0 ± 6.5% of the spiked DU145 cells were found in the center fraction (Figure 6c), not accounting for the 10-15% estimated recovery loss in the residual microfluidic set-up. Efficient WBC depletion along with high recovery of DU145 cells in the center fraction resulted in an average of 53 ± 27-fold enrichment of the prostate cell line (Figure 6d). Cell mixtures without EPs, collected in the center fraction, contained an average of 75975 ± 2450 WBCs with 848 ± 105 DU145 cells in the center fraction after going through the acoustophoresis chip. In comparison, cell mixtures treated with activated EPs (+CD45), displayed an average recovery in the center fraction of 1661 ± 736 WBCs and 840 ± 66 DU145 cells after the corresponding acoustophoresis step. The WBC depletion step yielded a 53-fold enrichment of the prostate cancer cells in the center fraction. The results demonstrate the possibility of using activated EPs together with acoustophoresis for the enrichment of diverse populations of CTCs from upstream cancer cell separations.

The differences in cancer cell line enrichment factor (52 vs. 84) was not dependent on the cell line as such but rather the fact that the negative selection process displayed a variation. As seen in Fig 5 and 6 the recoveries of DU145 and MCF7 were not significantly different, while the remaining WBC in the center fraction differed between the experiments which then also was reflected in the estimation of the cancer cell enrichment factor. It is worth to emphasize that the cancer cell recovery loss in the current study is to a major part an effect of the residual volume in the fluidic system, being about 10-15% of the 1 mL samples. When scaling up to full clinical sample volumes this loss factor will reduce in proportion to the sample volume.

It should however be commented that the entire sample handling process from RBC lysis to final readout in a system that comprises both a primary label-free acoustophoresis step and a subsequent negative acoustophoresis WBC depletion step will comprise additional loss factors. The lysis procedure induces typically a loss in nucleated cells up to 20-30% [41, 50]. Recent data on acoustophoresis based separation reports a cancer cell recovery up to 87% (i.e. 13% loss) on clinical scale sample volumes [46], only accounting for losses from spiking 50 cancer cells in 1 mL, 1:2 diluted RBC lysed blood.

Since microfluidic platforms for CTC separation have previously been designed to combine several separation technologies [20], it should also be feasible to hybridize our acoustic separation platform with other well-known CTC microfluidic separation systems (e.g., post arrays, inertia, dean flow, magnetophoresis) [19, 26, 60, 61]. However, we believe that by using a single separation technology (i.e., acoustophoresis), a simpler system can be accomplished through a 2-step acoustophoresis platform for enrichment of CTCs.

Future studies will investigate negative acoustophoretic selection in non-fixed as well as fixed cell patient samples. The ability to perform cancer cell enrichment using fixed blood samples has some advantages, including a longer processing window. This can be beneficial for the logistics of transporting samples between the clinic and the analysis laboratory. Increasing the processing window can also enable the batch analysis of enriched samples with CTCs, which can save time and potentially reduce costs.

4. Conclusions

In this report, we have demonstrated label-free enrichment of non-fixed cancer cells by negative selection of WBCs (derived from RBC-lysed blood) using anti-CD45-IgG activated EPs and acoustophoresis. The novelty provided in this report, in terms of the technical mode of operation (frequency modulation) of the acoustophoretic system, solves an important unmet need in using negative acoustophoretic selection for the enrichment of cancer cells. We are the first to demonstrate enrichment of cancer cells by WBC-depletion using negative acoustic contrast particles with acoustophoresis.

We envision that the proposed acoustic-based technology in combination with an up-stream primary, label-free acoustophoretic separation of cancer cells from blood as previously reported [18, 50] may lay the foundation for future analytical and clinical studies aimed to validate novel biomarkers for effective treatment management of cancer patients.

Highlights.

1. Negative acoustophoretic selection performed with acoustic frequency modulation.

2. Depletion of WBCs (52-86 fold) with elastomeric particles (EPs) and acoustophoresis.

3. High % recovery of cancer cells (85.0-89.8%) in the center fraction.

4. Negative selection of WBCs allowed for enrichment of cancer cells (53-84 fold).

5. Acoustic frequency modulation reduces EP aggregation at pressure antinodes.