Microfluidic affinity separation chip for selective capture and release of label-free ovarian cancer exosomes

Colin L. Hisey a, Kalpana Deepa Priya Dorayappan b, David E. Cohn b, Karuppaiyah Selvendiran b and Derek J. Hansford *a
aDepartment of Biomedical Engineering, The Ohio State University, Columbus, OH, USA. E-mail: hansford.4@osu.edu
bDivision of Gynecologic Oncology, Comprehensive Cancer Center, The Ohio State University Wexner Medical Center, Columbus, OH, USA

Received 11th August 2018 , Accepted 29th August 2018

First published on 4th September 2018


Exosomes are nanoscale vesicles found in many bodily fluids which play a significant role in cell-to-cell signaling and contain biomolecules indicative of their cells of origin. Recently, microfluidic devices have provided the ability to efficiently capture exosomes based on specific membrane biomarkers, but releasing the captured exosomes intact and label-free for downstream characterization and experimentation remains a challenge. We present a herringbone-grooved microfluidic device which is covalently functionalized with antibodies against general and cancer exosome membrane biomarkers (CD9 and EpCAM) to isolate exosomes from small volumes of high-grade serous ovarian cancer (HGSOC) serum. Following capture, intact exosomes are released label-free using a low pH buffer and immediately neutralized downstream to ensure their stability. Characterization of captured and released exosomes was performed using fluorescence microscopy, nanoparticle tracking analysis, flow-cytometry, and SEM. Our results demonstrate the successful isolation of intact and label-free exosomes, indicate that the amount of both total and EpCAM+ exosomes increases with HGSOC disease progression, and demonstrate the downstream internalization of isolated exosomes by OVCAR8 cells. This device and approach can be utilized for a nearly limitless range of downstream exosome analytical and experimental techniques, both on and off-chip.


Introduction

Most healthy and diseased cells continually secrete extracellular vesicles (EVs) into the surrounding fluid (saliva, milk, blood, etc.).1,2 Among these EVs, exosomes have received significant attention because they contain biomolecules such as lipids, nucleic acids, and proteins with a similar profile to their cell of origin and can be absorbed by distant cells through several processes including endocytosis, phagocytosis, and membrane fusion as a means of cell-to-cell communication.3–8

In cancer medicine, their ability to noninvasively provide information about their tumour cells of origin has potential applications in diagnostics and prognostics. In addition, their ability to deliver cargo that promotes proliferation, promotes metastasis, or inhibits proper immune response makes them a useful tool to elucidate the underlying mechanisms of cancer progression.9–15 Furthermore, they are overexpressed by cancer cells compared to normal cells of the same tissue type, making them a smaller but much more abundant liquid biopsy target compared to circulating tumour cells (CTCs).16 However, the current standard of isolation of exosomes is ultracentrifugation, which is time intensive and requires expensive equipment, typically separating both healthy and diseased exosomes from other biofluid components nonspecifically.17–21 To overcome the limitations of centrifuge-based isolation methods, many microfluidic approaches have been developed to improve the speed and specificity.22,23

Generally, these approaches can be classified based on the property used to separate the exosomes from each other and the rest of the sample: either physical or biochemical. Physical separation techniques differentiate exosomes based on their physical properties such as density or size, and include acoustic,24,25 membrane filtration,26–29 nanowire trapping,30 lateral displacement,31 and viscoelastic flow systems.32 While physical separation techniques can yield higher numbers of exosomes, often without labelling or modification, they do not target surface biomarkers and cannot directly target exosomes originating from specific types of cells. In many cases, they also require some type of post-separation processing before they can be characterized.

On the other hand, biochemical or affinity-based separation techniques show the most promise for the collection of specific subtypes of exosomes by targeting known surface proteins, often members of the tetraspanin family (e.g., CD9, CD41, CD63, and CD81).33,34 The majority of these systems have performed the characterization of the exosomes directly on the capture surface or lysed the exosomes for downstream analysis, with no release or recovery of intact exosomes.35,36 Recent developments have demonstrated the release of affinity-captured exosomes from channel surfaces, in which the linker molecule that attaches the antibody to the capture surface is cleaved, the exosome–antibody pair are released together using protein A/G chemistry, or magnetically trapped bead–exosome complexes are released together by removal of the magnetic field.37–39 These techniques, however, yield exosomes with some of the surface receptors blocked by the residual antibody and in some cases magnetic beads, which could present issues with downstream internalization by cells, sequential affinity-based filtrations, and reusability of the device. The captured exosomes should retain their in vivo structure, contents, and surface chemistry to ensure that researchers can use collected specific exosomes for as many applications as possible. Affinity chromatography adapted to microfluidic devices to isolate intact and label-free specific exosomes provides this capability.

Antibody–antigen interactions involve hydrogen bonds, van der Waals forces, hydrophobic–hydrophobic interactions, electrostatic interactions, and three-dimensional conformational matching.40–42 To release a captured target from affinity capture surfaces, the interaction must be disrupted by one or more of four general approaches: heat, reduction of ionic strength, organic solvents, and/or pH adjustments.41,43 This affinity chromatography process has been examined in the past primarily in columns and heparin-conjugated agarose beads,44 but has only recently been made commercially available for magnetic bead kits designed for biomarker-specific exosome isolation (System Biosciences). These approaches are typically expensive and require large sample volumes, or do not provide the potential for on-chip experimentation downstream of the capture surface, and are either slower or more expensive. In microfluidics, a number of affinity chromatography systems have been developed for the isolation of various molecules,45 but only one which could be considered a liquid biopsy, where colorectal cancer cell lines were spiked into whole blood, captured via anti-EpCAM, and enzymatically released with trypsin.46 In this paper we demonstrate the use of affinity chromatography to isolate intact and label-free exosomes from patient biofluids in a microfluidic system.

We present a microfluidic device that captures circulating exosomes from high-grade serous ovarian cancer (HGSOC) serum using a herringbone grooved surface that is functionalized with antibodies against CD9 (a common exosome surface marker) or EpCAM (epithelial cell-specific marker), as HGSOC is epithelial in origin. Herringbone grooves were used because they have been shown to increase fluid interaction with the antibody-functionalized inner channel surface in previous liquid biopsy devices.35 Serum samples obtained from healthy, benign, and HGSOC patients (stage I–IV) were flowed through the device, and following capture the exosomes were eluted intact and label-free using a low pH buffer and rapidly neutralized downstream to ensure their stability in solution. We used a combination of nanoparticle tracking analysis (NTA), western blotting, flow cytometry, fluorescence microscopy, and scanning electron microscopy (SEM) to verify the device performance. With our approach, biomarker-specific exosomes could be isolated intact and label-free, as demonstrated by their uptake by OVCAR-8 ovarian cancer cells. The ability to be characterized or manipulated downstream will aid current efforts to use exosomes as early diagnostic and prognostic tools, elucidate their role in cancer progression using internalization studies, or engineer them for semi-synthetic therapeutic delivery vehicles.47–49

Experimental

Microfluidic device design and fabrication

Channels with herringbone patterns were fabricated using a multilayer photolithography process (EVG 620 Contact Aligner). Herringbone patterns were designed with 135 μm short segments and 400 μm long segments with 45° vertices, and spaced by 70 μm in alternating sets of 7. First, 30 μm channel bases were patterned with SU8-2025, followed by a second 10 μm SU-8 2025 layer containing the grooves. These grooves have been shown to improve the contact of the solution with the surface and optimal mixing has been demonstrated using positive grooves in the forward flow direction.50 Following treatment of the silicon surface with hexamethyldisilazane (HMDS), Sylgard 184 polydimethylsiloxane (PDMS) was used to create moulds of the SU-8 patterns using a 10[thin space (1/6-em)]:[thin space (1/6-em)]1 ratio of base to curing agent (Fig. 1A–D). Inlet and outlet ports were punched and the channels were bonded to thin PDMS base layers using air plasma at 1000 mTorr for 3 min (Harrick Plasma Cleaner) and baked at 60 °C for 15 min before the surface chemistry modification. To quantify the capture and release improvements caused by the herringbone grooves, a 37 μm smooth channel was also fabricated to compare channels of equal total volume.
image file: c8lc00834e-f1.tif
Fig. 1 Device fabrication: (A) fabrication schematic showing photolithography of main channel layer (1–3), aligned photolithography of herringbone groove layer (4–6), followed by PDMS molding and packaging (7–9); (B) herringbone design (bar = 200 μm); (C) SEM image of PDMS herringbones (bar = 200 μm); and (D) assembled devices.

Surface chemistry modification

Functionalization of the PDMS was performed to covalently bond antibodies to the inner channel surface. Briefly, 10% [3-(2-aminoethylamino)propyl]trimethoxysilane in ethanol was flowed through channels at 5 μl min−1 for 15 min then left stagnant for 35 min, followed by a 15 min ethanol wash at 16.67 μl min−1. A similar treatment and wash process was performed with 5% glutaraldehyde in dH2O followed by a dH2O wash. 3 μg ml−1 anti-EpCAM (Santa Cruz) or anti-CD9 (Sigma) in PBS was flowed through the channel at 5 μl min−1 for 15 min, then left stagnant overnight at 4 °C. For antibody concentration optimization experiments, devices were also made using 10 and 30 μg ml−1 anti-EpCAM solutions. Another PBS wash was then performed for 15 min before the addition of serum.

For surface chemistry optimization experiments, the above chemistry was performed on smooth channels and compared to channels that had recently undergone plasma treatment and channels that were plasma bonded and left to sit for one week to return to the native, nonreactive state. Surface antibody retention following a PBS flush was qualitatively measured using FITC-IgG in a similar treatment to that described above (for anti-EpCAM/anti-CD9), followed by imaging the resulting fluorescence and quantifying mean grayscale values in ImageJ. Exposure time, binning, and all other settings were held constant.

Immunocapture and elution

Cell-free patient serum was thawed slowly and diluted from 1[thin space (1/6-em)]:[thin space (1/6-em)]4 to 1[thin space (1/6-em)]:[thin space (1/6-em)]10 in 1× PBS, depending on the number of replicates in the experiment, then filtered through a 0.2 μm pore syringe filter. 100 μl of serum was flowed through the device at 20 μl min−1. Devices were then flushed with filtered PBS at 16.67 μl min−1 for 15 min. Effluent tubing was replaced and the endings of the tubing were placed in 2 ml microcentrifuge tubes that were preloaded with 10 μl pH 8.5 Tris-Cl buffer for neutralization. Captured exosomes were eluted with 100 μl of pH 2.2 glycine-HCl buffer at 80 μl min−1 followed by manually pushing air through channels to ensure complete collection. A 10[thin space (1/6-em)]:[thin space (1/6-em)]1 ratio of eluent to neutralization buffer was chosen to achieve a final pH of ∼7.4 based on the titration curve (Fig. 2A and B). The final exosome solutions were stored at −20 °C until characterization for stability.51
image file: c8lc00834e-f2.tif
Fig. 2 Elution verification and analysis: (A) capture and release protocol; (B) buffer neutralization curve; (C) fluorescence verification of elution protocol with Alexa Fluor IgG (red, bar = 400 μm) and CFSE (green, bar = 200 μm): following treatment with the fluorescent secondary antibody against the surface anti-EpCAM, localized antibody fluorescence is observed (1) with no observable exosome CFSE fluorescence (4). Following passage of serum and PBS over the surface, there is a decrease in antibody fluorescence (2), and an increase in exosome fluorescence (5). Following elution, there is a return of antibody fluorescence (3) and a decrease in exosome fluorescence (6); and (D) NTA quantification showing the elution efficiency of ph 2.2 glycine-HCL buffer compared to PBS elution and controls (n = 3).

Fluorescence microscopy

Surface functionalization of the primary EpCAM antibody and capture and release of exosomes were assessed qualitatively using fluorescent microscopy. Antibody location and density were assessed using Alexa Fluor anti-rabbit IgG secondary antibody (Thermo Fisher), and Exo-glow green dye carboxyfluorescein succinimidyl ester (CFSE) (System Biosciences) stain was used to characterize the location and density of exosomes in devices before and after the flow of serum, and after elution and washing with 1× PBS. Devices were imaged under the fluorescence microscope with exposure time and binning held constant for all the images captured. Cropping and scale bars were done using ImageJ and brightness was increased on all images equally.

Scanning electron microscopy (SEM)

Channels were disassembled and treated with Karnovsky's fixative (2.5% EM-grade glutaraldehyde, 2% formaldehyde and 0.1 M sodium phosphate buffer) for 10 min, washed with PBS for 2 × 5 min, followed by a series of dehydration treatments of 35% ethanol for 10 min, 50% ethanol for 2 × 10 min, 70% ethanol for 2 × 10 min, 90% ethanol for 2 × 10 min and 100% ethanol for 4 × 10 min. Samples were then coated with palladium/gold and imaged with an InLens detector at 3 keV and 100% charge compensation (Zeiss Ultra 55 SEM). Size distributions from SEM images were determined based on Feret's diameter and calculated from surface area using threshold and binary tools in ImageJ followed by the particle analysis tool. Particle analysis area limits were set as 10–2000 nm in diameter, the same as NTA.

Nanoparticle tracking analysis (NTA)

A NanoSight NS300 system was used to analyze particles from 10–2000 nm diameter in solution and provided the size distribution and concentration of the purified exosomes. Briefly, exosome samples (10 μL) were diluted in PBS to reach the recommended concentration (1 × 108–1 × 109 particles per mL). The light from a green laser (532 nm) is scattered by the particles and captured by a CCD camera, visualized with a standard 20× light microscope, and the motion of each particle is tracked from frame to frame. Five 30 second videos obtained from the optimized measurement settings were used to measure the statistics of size and distribution with the NTA 3.2 Analytical software (ESI).52

Comparison to ultracentrifugation and ExoQuick

The efficiency of specific exosome capture was characterized using the cell culture medium from 107 OvCAR8 cells that had been cultured for 48 hours. For all samples, the conditioned medium was centrifuged at 400 × g for 10 min to remove whole cells and then centrifuged at 10[thin space (1/6-em)]000 × g for 30 min to remove large vesicles in solution, and finally filtered through a 0.22 μm porous membrane. The ultracentrifuged samples were spun at 100[thin space (1/6-em)]000 × g for 1 h at 4 °C to isolate the exosomes, and the ExoQuick samples isolated the exosomes by mixing the ExoQuick-TC reagent (System Biosciences) 1[thin space (1/6-em)]:[thin space (1/6-em)]5 with culture medium and incubating overnight at 4 °C, followed by centrifugation at 3000 × g for 10 min. The pellets for both ultracentrifugation and ExoQuick processes were re-suspended in 100 μL of cold PBS for analysis. 10 μL of biotynilated EpCAM antibody was mixed with 40 μL streptavidin-coated magnetic beads in solution (System Biosciences) and incubated for 1 h. Following exosome isolation, all samples were mixed with the coated magnetic beads M to capture EpCAM+ exosomes, and excess exosomes were rinsed. EpCAM+ exosomes were eluted using the kit elution buffer, and exosome concentration was measured using NTA.

Flow cytometry

CFSE labeled exosomes eluted from the device were quantified using an ARIA III flow cytometer equipped with a 4-laser setup (50 mW Coherent Sapphire 488 nm blue, 65 mW Coherent Genesis CX 355 nm violet, 100 mW Coherent OBIS 640 nm red, and 100 mW Coherent Sapphire 561 nm yel/grn) and FACS Diva 6.0 software (BD Biosciences). Flow cytometer setup was performed using CS&T instrument setup beads (BD Biosciences). Forward scatter (FSC) and side scatter (SSC) parameters were set to log mode and the lowest threshold allowed by the cytometer (200) was selected for each. FSC/SSC voltages were set to the highest values that minimized most of the background noise. Each sample tube was run at the same amount of time and flow rate to subtract false positive events detected over an equal time frame. The FITC 530/30 band pass filter was used to measure the fluorescence of the CFSE.

ImageStream (IS)

ImageStream flow cytometry was used to quantify the eluted EpCAM+ and CD9+ exosomes. Fluorescent markers are required for ImageStream analysis requires due to the low level of exosome scattering, and their brightfield (BF) images being below the resolution of the BF camera for the majority of EVs. We used CFSE, a fluorescent cell-permeable dye that covalently binds through its succinimidyl group to intracellular proteins at their lysine residues and other amine sources.53,54 The ImageStream was equipped with a 60× objective to detect particles in the EV size range, with a numerical aperture (NA) of 0.9 and a resolution of 0.3 μm2 per pixel. Although many of the vesicles are below the size of the pixel resolution, when labelled with fluorescent molecules these smaller vesicles become detectable due to the intensity of the signal, and the width of the core stream is reduced to 7 μm to increase the frequency of in-focus objects. Also, the speed was set to the lowest setting for maximum resolution. All sheath buffers were filtered with 0.1 μm filters to ensure that there were minimal background particulates. The instrument was equipped with 488 nm lasers to be compatible with the CFSE, and 200 mW was used to increase the number of photons generated per fluorochrome molecule, including the 758 nm laser for scatter measurements.

Traditionally, submicron polystyrene beads (PSB) were used to demonstrate the ability of cytometer to detect particles smaller than 1 μm. However, PSB have a refractive index (RI) close to 1.6 and scatter up to 10 times more than biological EVs, which are primarily composed of lipids and proteins and have a RI < 1.4. All samples were collected using the INSPIRER instrument acquisition software. Samples were run in the following order to make sure there was no carryover of any fluorophores from sample to sample: 1) buffer only, 2) unstained EVs, 3) buffer plus reagents and/or stained EVs with detergent, 4) single stained compensation controls, and 5) samples. Based on the high scatter intensity of the PSB, a collection gate was established that eliminated the beads from the final measurement. Depending on the concentration of EVs, the collection was stopped on the amount of time required to achieve a reasonable calibration bead count (5000 events). All samples were collected for the same amount of time except the compensation controls. A total of 2000 events were sufficient for the compensation controls. Here we used the polystyrene beads as a reference standard gate to help identify the range of scatter for these exosome particles. All data analysis was performed using the ImageStream Data Exploration and Analysis Software (IDEAS R 6.2 EMD) or FCS Express 6.0 (De Novo Software) as described previously.55

Internalization and confocal microscopy

50 μL of 10× Exo-Glow (System Biosciences) was added to 500 μL volume of resuspended intact exosomes in 1× PBS and mixed well by flicking. The exosome suspension was then incubated at 37 °C for 10 minutes and placed on ice for 30 min, after which the samples were centrifuged at 100[thin space (1/6-em)]000g for 1 h in an ultracentrifuge. The supernatant (with excess label) was removed and the labelled exosome pellet was again resuspended in 500 μL of 1× PBS. The labelled exosomes were cultured with OVCAR8 high-grade serous ovarian cancer cells and imaged using an Olympus FV1000 confocal microscope and Imaris software.

Statistical analysis

Unpaired t-test, one-way ANOVA, and Tukey-HSD multicomparison tests were performed for all datasets in MATLAB and Graphpad Prism (α = 0.05). Results are reported as averages ± standard error of the mean and error bars are standard error of the mean.

Results and discussion

Capture and release verification

To qualitatively assess the various stages of our capture and release protocol, we labelled the primary EpCAM antibody with Alexa Fluor anti-rabbit IgG (red) and stained captured exosomes with CFSE (green) (Fig. 2C, surface antibodies in panels 1–3 and CFSE stained exosomes in panels 4–6). Each step of analysis (image columns) was produced by running a separate clinical sample, with location and relative intensities of fluorescent signals remaining relatively consistent. Secondary labelling of surface-functionalized anti-EpCAM produced localized fluorescence on herringbone edges and vertices (panel 1) and was less intense once serum had passed through the channel (panel 2), indicating that the exosomes may be blocking the secondary antibody. Following elution, localized fluorescence appeared to return to a level near that of the original surface (panel 3), indicating that much the original anti-EpCAM antibody remains intact following elution. Almost no fluorescence was observed from CFSE treatment of the functionalized surface (panel 4), but it increased significantly on the herringbone edges and vertices following exosome capture (panel 5). Post-elution, it is clear that a few exosomes remain on the surface, as the fluorescence is still localized and more significant than the pre-capture surface (panel 6). This presence of residual exosomes indicates that the elution protocol can be further optimized to maximize the exosome recovery. Since the columns are from different samples, an exact recovery cannot be determined using this fluorescent approach, but a conservative estimate of ∼60% recovery was determined using relative fluorescent intensity between images. While the reason for the highly localized fluorescence signals is unclear, it is likely due to a flow effect, the high surface energy at the corners of the herringbones, or some combination of the two. We have also verified that the localized fluorescence is not simply an optical effect based on SEM images, which also show this corner localization of exosomes.

Initial NTA tests on the device eluate demonstrated that the low pH buffer releases significantly more exosomes than filtered PBS alone at 80 μl min−1 and that the buffer eluate contained significantly more particles than stock buffer or PBS (Fig. 2D). Mean nanoparticle concentrations increased from 4.16 × 108 ± 4.77 × 107 particles per ml using PBS elution to 8.05 × 108 ± 4.96 × 107 particles per ml using buffer elution (p < 0.01), indicating that the lower pH within the buffer can disrupt the antibody–exosome bonds as hypothesized. More significantly, the concentrations of the buffer elution experiments are significantly higher than the controls (PBS and low pH buffer samples), while the PBS elution samples are statistically identical to the controls. Because NTA quantifies any particles in solution, including non-exosomal particles, flow cytometry of CFSE-stained exosomes was used to confirm the NTA results.

Device optimization and analysis

Fluorescence intensity of the surface FITC-IgG was used to compare antibody adsorption on the channel surfaces from our three surface chemistry conditions. Mean grayscale values quantified using ImageJ demonstrated an increase in fluorescence from 49.67 ± 1.55 a.u. on recently plasma-treated channels to 73.23 ± 1.51 a.u. on channels one week post-treatment (p < 0.05). A maximum and significantly greater average fluorescence was seen from our linker chemistry at 145.42 ± 9.34 a.u. (p < 0.001), demonstrating that it is a much more effective means of bonding and maintaining antibodies on the inner PDMS surface (Fig. 3A).
image file: c8lc00834e-f3.tif
Fig. 3 Device performance analysis. (A) Surface chemistry optimization comparing fluorescence of passively adsorbed FITC-IgG on plasma treated and untreated PDMS to our functionalized surface (n = 8), (B) antibody concentration optimization from NTA counts of exosomes isolated using three different concentrations of anti-EpCAM showing no significant difference between groups (n = 3), (C) ImageStream quantification of CD9 positive and CFSE-labeled exosomes isolated using smooth vs. grooved channels (n = 3), and (D) comparison of EpCAM+ exosome concentrations from ultracentrifugation, ExoQuick kit, and our microfluidic chip (n = 3).

Three anti-EpCAM concentrations including 3, 10, and 30 μg ml−1 were tested to determine if there was a surface saturation point and to minimize the cost per device. Based on the NTA counts of eluted exosomes, no significant difference was found and 3 μg ml−1 was used for the remaining experiments (Fig. 3B, ESI video).

The efficacy of the herringbone grooves at generating flow-folding and increasing the probability of exosome–antibody interactions on the channel walls was tested by running identical exosome samples. Eluted exosomes isolated from stage IV HGSOC serum, using anti-CD9 functionalized grooved and smooth channels of equal total volume, were stained with CFSE and compared using ImageStream. A significant increase from 8.89 × 105 ± 3.04 × 105 (smooth) to 2.03 × 106 ± 2.28 × 105 (grooved) particles per ml was seen between the channels (p < 0.05, Fig. 3C). The higher surface area to volume ratio may also cause this increase in captured and eluted exosomes.

Specific exosome capture for identical samples was compared between our microfluidic device, ultracentrifugation (100[thin space (1/6-em)]000 × g), and ExoQuick to demonstrate the enhanced selectivity (Fig. 3D). The samples prepared using our microfluidic capture and release chip had significantly more EpCAM+ exosomes (4.83 × 1010 ± 3.85 × 109 particles per mL) compared to both the ExoQuick kit samples (1.71 × 1010 ± 6.36 × 109 particles per mL, p = 0.0125) and the ultracentrifugation samples (1.12 × 1010 ± 6.36 × 108 particles per mL, p = 0.0005). This points to a loss of exosomes using the other two techniques, possibly due to aggregation during centrifugation without proper redispersion for selective capture.

Evaluating clinical utility

All patient serum used in this study was obtained from frozen samples and was approved by the Institutional Review Board (IRB Study Number: 2016C0099). To compare the relative abundance of total exosomes in varying disease stages, ExoQuick was used to isolate the total exosome population from healthy, benign, stage I, and stage IV HGSCOC patient serum samples and demonstrated a significant increase between all groups with increasing disease stage (p < 0.05), from 6.71 × 108 ± 5.18 × 107 particles per ml using healthy serum to 7.27 × 1011 ± 1.51 × 1010 using stage IV HGSOC serum (Fig. 4A). Furthermore, to determine the relative abundance of total EpCAM, an important epithelial cell-based biomarker in HGSOC, western blots of the total exosome populations yielded distinctly stronger EpCAM bands with increasing disease stage (Fig. 4A).
image file: c8lc00834e-f4.tif
Fig. 4 Clinical application in HGSOC, using ExoQuick, western blot, microfluidic capture, and flow cytometry: (A) NTA count and western blot of ExoQuick isolated exosomes from healthy, benign and HGSOC serum (n = 3); (B) NTA count of microfluidically isolated EpCAM+ exosomes comparing healthy and stage IV serum (n = 3, ESI videos); and (C) flow cytometer count of microfluidically isolated EpCAM+ and CFSE-stained exosomes from benign, stage I, and stage IV HGSOC.

Based on the significant increases in total exosomes from ExoQuick isolation and increase in EpCAM found in the western blot, an initial test comparing eluted EpCAM positive exosomes from healthy and stage IV HGSOC serum was performed, yielding a significant increase in NTA count from 3.72 × 108 ± 6.26 × 107 to 1.27 × 1010 ± 1.349 particles per ml (p < 0.001). This initial test confirmed that EpCAM+ exosomes warranted further experimentation to assess their value in predicting HGSOC disease stages (Fig. 4B, ESI videos).

To determine the relative abundances of EpCAM+ exosomes in HGSOC serum and whether significant differences could be measured with greater sensitivity, isolated exosomes from serum at different stages of HGSOC were CSFE stained and counted using flow cytometry (ESI). HGSOC serum was found to contain significantly more EpCAM+ exosomes compared to benign serum, with mean flow cytometry counts of 700 ± 61, 862 ± 53, and 1089 ± 17 for benign, stage I, and stage IV serum, respectively (Fig. 4C, p < 0.05). This result strongly supports the potential of EpCAM positive exosomes in the eluate from our devices as diagnostic or prognostic indicators in HGSOC and the need for further experimentation.

It is important to note that the data presented in Fig. 4 is intended to demonstrate the reproducibility inherent in the device and our capture and release protocol. One sample per disease stage was split, diluted, and run through multiple devices in each experiment. Our experimental numbers (n) refer to the number of devices through which a given serum sample was processed (using equal dilutions for all comparisons). Given the heterogeneous nature of serum exosome levels (both total and biomarker-specific), a much larger number of patient samples must be processed and compared to make a more general conclusion about EpCAM+ exosomes in the general population.

To determine the potential of our device to profile multiple biomarker-specific exosome levels in a single sample, a comparison of exosomes which were captured and released based on two commonly used targets (CD9 and EpCAM) was performed. Relative abundances were quantified using ImageStream (ESI) and demonstrated that the general marker CD9+ exosomes had a higher mean concentration of 1.78 × 106 ± 3.57 × 105 particles per ml compared to the more specific marker EpCAM+ exosomes of 6.78 × 105 ± 1.04 × 105 particles per ml (Fig. 5A, p < 0.05). Size distributions of released exosomes of both targets were measured using NTA and calculated from SEM images using ImageJ (ESI). Calculated size distributions based on Feret's diameter and projected areas yielded similar profiles to those from NTA, but with smaller and more physiologically accurate diameters, comparable to those calculated in a previous study35 (Fig. 5B). This difference in size distribution could be due to the fixing and dehydration protocols used to prepare the samples for SEM, or clusters of particles which can form in the solution, resulting in many particles measured via NTA which are greater than the known sizes of individual exosomes (<200 nm). Furthermore, as part of our size calculation using SEM images, the watershed function separates perceived exosome clusters into single particles.


image file: c8lc00834e-f5.tif
Fig. 5 Stage IV HGSOC exosome profile: (A) ImageStream comparison of CD9+ and EpCAM+ CFSE-stained exosomes isolated using our microfluidic device from a single stage IV HGSOC serum sample (n = 3) and (B, left to right) SEM images of grooves and captured exosomes on surface (50 μm bar and 1 μm bar, respectively), size distribution from SEM images calculated using ImageJ to obtain the Feret's and projected area diameters (ESI), and size distribution obtained by NTA showing larger size distribution in NTA but similar shape.

Internalization by cell cultures

Culturing isolated and intact patient-derived exosomes with OVCAR8 HGSOC cells yielded clear visual evidence of intact released exosomes and their cellular internalization (Fig. 6A–C, ESI video). The Exo-Glow exosome stain used in this experiment only becomes fluorescent in the presence of active esterases, and only remains localized if contained within an intact membranous structure, demonstrating that the released exosomes were both intact and biologically active. These images and video qualitatively demonstrate the potential of intact exosome isolation and how downstream internalization, on or off-chip, could potentially be used to study changes in cellular behaviour in vitro, such as proliferation or migration.
image file: c8lc00834e-f6.tif
Fig. 6 OVCAR8 cells cultured with isolated intact serum exosomes allows cell internalization: (A) DAPI staining showing the cell nuclei in blue; (B) DAPI combined with CSFE labeled exosomes (green); and (C) combined differential interface contrast and fluorescent micrographs clearly showing internalization.

Conclusions

We have developed a microfluidic platform which isolates exosomes directly from serum using two commonly targeted membrane biomarkers, CD9 and EpCAM. Utilizing principles of affinity chromatography, we hypothesized that biomarker-specific exosomes could be rapidly captured and released, both intact and label-free, and kept stable by immediately neutralizing the eluate downstream. Furthermore, we postulated that the relative abundance of these released exosomes could indicate a positive correlation with HGSOC severity and that the released exosomes could be internalized by cultured cells in vitro.

Using a combination of NTA, flow cytometry, fluorescence microscopy, and SEM, we confirmed our capture and release protocol and performed some initial surface chemistry, antibody concentration, and channel design optimization. In addition, we demonstrated that our low pH buffer elution yielded significantly greater exosomes than PBS alone (p < 0.01) and significant differences were found in the number of both total and EpCAM+ exosomes between healthy, benign, stage I, and stage IV HGSOC patient serums (p < 0.05). Also, significantly more exosomes were captured and released from stage IV HGSOC serum using CD9 as a target compared to EpCAM (p < 0.05). Lastly, using confocal microscopy we demonstrated that post-elution, these intact and label-free exosomes can be internalized by cultured cells.

Compared to conventional isolation methods, such as ultracentrifugation and ExoQuick-enhanced centrifugation followed by capture on antibody-coated magnetic beads, our technology demonstrates higher yield, higher specificity, is inexpensive, comparatively rapid (<20 minutes for capture and release), and requires minimal sample volume to detect significant differences in HGSOC disease stages (<100 μl), critical for its application to clinical samples. Compared to many microfluidic exosome isolation technologies, our approach allows the use of precise downstream characterization techniques designed for intact vesicles in solution, such as NTA and ImageStream flow cytometry, rather than having to lyse the captured exosomes or perform on-chip fluorescence analysis. Furthermore, because selectively captured exosomes are released intact and label-free, thereby retaining their full bioactivity while containing known biomarkers on their surface, they are readily available for further downstream experimentation, both on and off chip. Examples of potential experiments enabled by this technology include studies on the effects of patient-derived exosomes on cellular proliferation, migration, and phenotype modification. The release of exosomes with no attached labelling molecules enables the possibility of combine capture and release chambers to rapidly isolate highly specific exosomes with both positive and negative enrichment of multiple surface markers. The combination of these possibilities will bring exosome profiling closer to clinical translation, provide new avenues for internalization studies, and enhance current efforts in post-elution exosome modification for therapeutic delivery.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

NCI RO1-CA176078 grant and KOH Ovarian Cancer Foundation Grant. Fig. 1D photo credit: Jonathan Smith. Proofreading credit: Candace Hisey. We would also like to thank Roman A. Zingarelli for his help in sample processing and Alex Cornwell from shared analytical services at Ohio State for ImageStream analysis assistance.

References

  1. Y. Sun and J. Liu, Clin. Ther., 2014, 36, 863–872 CrossRef PubMed.
  2. J. Kowal, M. Tkach and C. Thery, Curr. Opin. Cell Biol., 2014, 29, 116–125 CrossRef PubMed.
  3. G. Raposo and W. Stoorvogel, J. Cell Biol., 2013, 200, 373–383 CrossRef PubMed.
  4. S. Pant, H. Hilton and M. E. Burczynski, Biochem. Pharmacol., 2012, 83, 1484–1494 CrossRef PubMed.
  5. M. Simons and G. Raposo, Curr. Opin. Cell Biol., 2009, 21, 575–581 CrossRef PubMed.
  6. A. V. Vlassov, S. Magdaleno, R. Setterquist and R. Conrad, Biochim. Biophys. Acta, 2012, 1820, 940–948 CrossRef PubMed.
  7. C. Bang and T. Thum, Int. J. Biochem. Cell Biol., 2012, 44, 2060–2064 CrossRef PubMed.
  8. H. Mollaei, R. Safaralizadeh and N. Pouladi, Meta Gene, 2017, 11, 70–74 CrossRef.
  9. P. Kharaziha, S. Ceder, Q. Li and T. Panaretakis, Biochim. Biophys. Acta, 2012, 1826, 103–111 Search PubMed.
  10. D. D. Taylor and C. Gercel-Taylor, Gynecol. Oncol., 2008, 110, 13–21 CrossRef PubMed.
  11. B. Liang, P. Peng, S. Chen, L. Li, M. Zhang, D. Cao, J. Yang, H. Li, T. Gui, X. Li and K. Shen, J. Proteomics, 2013, 80, 171–182 CrossRef PubMed.
  12. D. S. Choi, J. O. Park, S. C. Jang, Y. J. Yoon, J. W. Jung, D. Y. Choi, J. W. Kim, J. S. Kang, J. Park, D. Hwang, K. H. Lee, S. H. Park, Y. K. Kim, D. M. Desiderio, K. P. Kim and Y. S. Gho, Proteomics, 2011, 11, 2745–2751 CrossRef PubMed.
  13. C. Lau, Y. Kim, D. Chia, N. Spielmann, G. Eibl, D. Elashoff, F. Wei, Y. L. Lin, A. Moro, T. Grogan, S. Chiang, E. Feinstein, C. Schafer, J. Farrell and D. T. Wong, J. Biol. Chem., 2013, 288, 26888–26897 CrossRef PubMed.
  14. H. Peinado, M. Aleckovic, S. Lavotshkin, I. Matei, B. Costa-Silva, G. Moreno-Bueno, M. Hergueta-Redondo, C. Williams, G. Garcia-Santos, C. Ghajar, A. Nitadori-Hoshino, C. Hoffman, K. Badal, B. A. Garcia, M. K. Callahan, J. Yuan, V. R. Martins, J. Skog, R. N. Kaplan, M. S. Brady, J. D. Wolchok, P. B. Chapman, Y. Kang, J. Bromberg and D. Lyden, Nat. Med., 2012, 18, 883–891 CrossRef PubMed.
  15. A. Hoshino, B. Costa-Silva, T. Shen, G. Rodrigues, A. Hashimoto, M. M. Tesic, H. Molina, S. Kohsaka, A. Di Giannatale, S. Ceder, S. Singh, C. Williams, N. Soplop, K. Uryu, L. Pharmer, T. King, L. Bojmar, A. Davies, Y. Araso, T. Zhang, H. Zhang, J. Hernandez, J. Weiss, V. Dumont-Cole, K. Kramer, L. Wexler, A. Narendran, G. Schwartz, J. Healey, P. Sandstrom, K. Labori, E. Kure, P. Grandgenett, M. Hollingsworth, M. de Sousa, S. Kaur, M. Jain, K. Mallya, S. Batra, W. Jarniagin, M. Brady, O. Fodstad, V. Muller, K. Pantel, A. Minn, M. Bissell, B. Garcia, Y. Kang, V. Rajasekhar, C. Ghajar, I. Matei, H. Peinado, J. Bromberg and D. Lyden, Nature, 2015, 527, 329–335 CrossRef PubMed.
  16. Y. H. Soung, S. Ford, V. Zhang and J. Chung, Cancers, 2017, 9, 8 CrossRef PubMed.
  17. K. W. Witwer, E. I. Buzas, L. T. Bemis, A. Bora, C. Lasser, J. Lotvall, E. N. Nolte-'t Hoen, M. G. Piper, S. Sivaraman, J. Skog, C. Thery, M. H. Wauben and F. Hochberg, J. Extracell. Vesicles, 2013, 2, 20360 CrossRef PubMed.
  18. A. Cvjetkovic, J. Lotvall and C. Lasser, J. Extracell. Vesicles, 2014, 3, 23111 CrossRef PubMed.
  19. E. Zeringer, T. Barta, M. Li and A. V. Vlassov, Cold Spring Harb. Protoc., 2015, 2015, 319–323 Search PubMed.
  20. K. Rekker, M. Saare, A. M. Roost, A. L. Kubo, N. Zarovni, A. Chiesi, A. Salumets and M. Peters, Clin. Biochem., 2014, 47, 135–138 CrossRef PubMed.
  21. I. Helwa, J. Cai, M. D. Drewry, A. Zimmerman, M. B. Dinkins, M. L. Khaled, M. Seremwe, W. M. Dismuke, E. Bieberich, W. D. Stamer, M. W. Hamrick and Y. Liu, PLoS One, 2017, 12, e0170628 CrossRef PubMed.
  22. J. C. Contreras-Naranjo, H. J. Wu and V. M. Ugaz, Lab Chip, 2017, 17, 3558–3577 RSC.
  23. S. Gholizadeh, M. Shehata Draz, M. Zarghooni, A. Sanati-Nezhad, S. Ghavami, H. Shafiee and M. Akbari, Biosens. Bioelectron., 2017, 91, 588–605 CrossRef PubMed.
  24. M. Wu, Y. Ouyang, Z. Wang, R. Zhang, H. Po-Hsun, C. Chen, H. Li, P. Li, D. Quinn, M. Dao, S. Suresh, Y. Sadovsky and T. Huang, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 10584–10589 CrossRef PubMed.
  25. K. Lee, H. Shao, R. Weissleder and H. Lee, ACS Nano, 2015, 9, 2321–2327 CrossRef PubMed.
  26. H. K. Woo, V. Sunkara, J. Park, T. H. Kim, J. R. Han, C. J. Kim, H. I. Choi, Y. K. Kim and Y. K. Cho, ACS Nano, 2017, 11, 1360–1370 CrossRef PubMed.
  27. R. T. Davies, J. Kim, S. C. Jang, E. J. Choi, Y. S. Gho and J. Park, Lab Chip, 2012, 12, 5202–5210 RSC.
  28. L. G. Liang, M. Q. Kong, S. Zhou, Y. F. Sheng, P. Wang, T. Yu, F. Inci, W. P. Kuo, L. J. Li, U. Demirci and S. Wang, Sci. Rep., 2017, 7, 46224 CrossRef PubMed.
  29. S. Cho, W. Jo, Y. Heo, J. Y. Kang, R. Kwak and J. Park, Sens. Actuators, B, 2016, 233, 289–297 CrossRef.
  30. Z. Wang, H. J. Wu, D. Fine, J. Schmulen, Y. Hu, B. Godin, J. Zhang and X. Liu, Lab Chip, 2014, 13, 2879–2882 RSC.
  31. B. H. Wunsch, J. T. Smith, S. M. Gifford, C. Wang, M. Brink, R. L. Bruce, R. H. Austin, G. Stolovitzky and Y. Astier, Nat. Nanotechnol., 2016, 11, 936–940 CrossRef PubMed.
  32. C. Liu, J. Guo, F. Tian, N. Yang, F. Yan, Y. Ding, J. Wei, G. Hu, G. Nie and J. Sun, ACS Nano, 2017, 11, 6968–6976 CrossRef PubMed.
  33. Y. J. Chiu, W. Cai, Y. R. Shih, I. Lian and Y. H. Lo, Small, 2016, 12, 3658–3666 CrossRef PubMed.
  34. D. D. Taylor and C. Gercel-Taylor, Semin. Immunopathol., 2011, 33, 441–454 CrossRef PubMed.
  35. C. Chen, J. Skog, C. H. Hsu, R. T. Lessard, L. Balaj, T. Wurdinger, B. S. Carter, X. O. Breakefield, M. Toner and D. Irimia, Lab Chip, 2010, 10, 505–511 RSC.
  36. S. S. Kanwar, C. J. Dunlay, D. M. Simeone and S. Nagrath, Lab Chip, 2014, 14, 1891–1900 RSC.
  37. H. Im, H. Shao, Y. I. Park, V. M. Peterson, C. M. Castro, R. Weissleder and H. Lee, Nat. Biotechnol., 2014, 32, 490–495 CrossRef PubMed.
  38. Y. T. Kang, Y. J. Kim, J. Bu, Y. H. Cho, S. W. Han and B. I. Moon, Nanoscale, 2017, 9, 13495–13505 RSC.
  39. Z. Zhao, Y. Yang, Y. Zeng and M. He, Lab Chip, 2016, 16, 489–496 RSC.
  40. R. Reverberi and L. Reverberi, Trasfus. Sangue, 2007, 5, 227–240 Search PubMed.
  41. P. L. Howard, Transfusion, 1981, 21, 447–482 CrossRef.
  42. I. S. Mian, A. R. Bradwell and A. J. Olson, J. Mol. Biol., 1991, 217, 133–151 CrossRef PubMed.
  43. C. J. van Oss, R. J. Good and M. K. Chaudhury, J. Chromatogr., 1986, 176, 111–119 CrossRef.
  44. L. Balaj, N. A. Atai, W. Chen, D. Mu, B. A. Tannous, X. O. Breakefield, J. Skog and C. A. Maguire, Sci. Rep., 2015, 5, 10266 CrossRef PubMed.
  45. J. P. Grinias and R. T. Kennedy, TrAC, Trends Anal. Chem., 2016, 81, 110–117 CrossRef PubMed.
  46. U. Dharmasiri, S. K. Njoroge, M. A. Witek, M. G. Adebiyi, J. W. Kamande, M. L. Hupert, F. Barany and S. A. Soper, Anal. Chem., 2011, 83, 2301–2309 CrossRef PubMed.
  47. S. Viaud, C. Thery, S. Ploix, T. Tursz, V. Lapierre, O. Lantz, L. Zitvogel and N. Chaput, Cancer Res., 2010, 70, 1281–1285 CrossRef PubMed.
  48. I. S. Zeelenberg, M. Ostrowski, S. Krumeich, A. Bobrie, C. Jancic, A. Boissonnas, A. Delcayre, J. B. Le Pecq, B. Combadiere, S. Amigorena and C. Thery, Cancer Res., 2008, 68, 1228–1235 CrossRef PubMed.
  49. J. L. Hood and S. A. Wickline, Wiley Interdiscip. Rev.: Nanomed. Nanobiotechnol., 2012, 4, 458–467 Search PubMed.
  50. T. J. Kwak, Y. G. Nam, M. A. Najera, S. W. Lee, J. R. Strickler and W. J. Chang, PLoS One, 2016, 11, e0166068 CrossRef PubMed.
  51. V. Sokolova, A. K. Ludwig, S. Hornung, O. Rotan, P. A. Horn, M. Epple and B. Giebel, Colloids Surf., B, 2011, 87, 146–150 CrossRef PubMed.
  52. Z. Sun, A. Li, W. Jia, S. Ye, G. Van Eps, J. Yu and W. Yang, Oncotarget, 2017, 8, 92839–93855 Search PubMed.
  53. E. van der Vlist, E. N. Nolte-'t Hoen, W. Stoorvogel, G. Arkesteijn and M. Wauben, Nat. Protoc., 2012, 14, 1311–1326 CrossRef PubMed.
  54. V. Pospichalova, J. Svoboda, Z. Dave, A. Kotrbova, K. Kaiser, D. Klemova, L. Ilkovics, A. Hampl, I. Crha, E. Jandakova, L. Minar, V. Weinberger and V. Bryja, J. Extracell. Vesicles, 2015, 4, 25530 CrossRef PubMed.
  55. J. Lannigan and U. Erdbruegger, Methods, 2017, 112, 55–67 CrossRef PubMed.

Footnotes

Electronic supplementary information (ESI) available: 4 figures and 2 videos. See DOI: 10.1039/c8lc00834e
These authors contributed equally.

This journal is © The Royal Society of Chemistry 2018