CHAPTER 3. Fully automated, label-free isolation of EVs from whole blood on a disc
3.3 A fully automated plasma-driven EV isolation on a disc
3.3.1 Experimental details
Design, fabrication and operation of the disc
The Exodisc structure was designed using a three-dimensional (3D) computer-aided design (CAD) program and fabricated by means of a CNC milling machine (M&I CNC Lab, Osan, South Korea). The top, middle, and bottom layers of polycarbonate (PC; IComponents Co. Ltd., Pyongtaek, South Korea) were milled in accordance with the 3D CAD model. Post milling, all layers were laminated using two pressure sensitive, double-sided adhesives (DFM 200 clear 150 POLY H-9 V-95;
FLEXcon, Spencer, MA, USA) and a customized pressing apparatus (Figure 3.1). The general procedure of fabricating a lab-on-a-disc integrated with a membrane filter has been described in our previously published works.28, 164 Briefly, the procedure can be described as follows: each layer of the Exodisc device is fabricated via CNC milling, and the reverse side of the filtration chamber was carved for insertion of commercially available membranes, including a track-etched PC membrane (13-mm diameter and 0.6-μm pore size; SPI Supplies, West Chester, PA, USA) and an AAO membrane (13-mm diameter and 0.02- or 0.1-μm pore size; Whatman; Thermo Fisher Scientific, Waltham, MA, USA), as filters I and II, respectively. Reversible ID valves were automatically actuated in the same manner as that reported in extant studies (Figure 3.2).21, 165
The overall process of the EV enrichment from the whole-blood samples on a spinning disc was summarized in Figure 3.3 and Table 3.1. First, plasma samples (200 μL) were obtained from whole-blood (600 μL) by spinning the disc at 3600 rpm for 5 min (Figure 3.3A). Following plasma separation, valve #1 was opened, and the disc was made to rotate via spin step #2 to transfer 200 μL of separated plasma into the pre-filtration chamber (Figure 3.3B). Valve #2 was then opened, and the loaded plasma solution was transferred through the 600-nm TEPC and 100-nm AAO membranes into the waste chamber via spin step #3 (Figure 3.3C). During filtration, large particles are trapped by the
69
Figure 3.1. (A) Expanded view of disc depicting top, middle, and bottom layers of polycarbonate (PC) with two layers of double-sided adhesive tapes and filters. (B) Photograph and (C) schematic showing microfluidic layout of Exodisc-P to enrich EVs starting from plasma, which is identical to Exodisc-B, except for the blood separation chamber.
70
Figure 3.2. Schematic illustration depicting operation of individually addressable diaphragm valves (ID valves). Side view of the fluidic channel shows the ID valve in open and closed states; the valve can be reversibly actuated by manipulating electromagnet polarity. For valve actuation, spinning disc was stop-aligned and spun again for liquid transfer.
Figure 3.3. Schematics (left) and CCD images (right) of overall process of EV-enrichment from whole-blood samples on a spinning disc—(A) Plasma separation from whole blood; (B) transfer of plasma to filtration chamber by opening valve #1; (C) EV enrichment on filter II by opening valve
#2; (D) washing of enriched EV by opening valve #3; (E) removal of solution under filter II by opening valve #4; (F) transfer of enriched EV to collection chamber for recovery by opening valve
#5.
71
Table 3.1. Detailed disc operation program for EV isolation from whole blood.
No. Step
Spin speed (rpm)
Duration Operation
1 Plasma
separation 3600 5 min Plasma separation from whole blood (600 μL) 2 Plasma
transfer 2400 20 s
Open valve #1 to transfer plasma sample (200 μL) to pre-filtration chamber integrated with TEPC membrane having pore diameter of 600 nm
3 EV filtration 1200 10 min
Open valve #2 to filtrate EV sample from pre-filtered plasma through AAO membrane having pore diameter of 100 nm
4 Wash 1200 20 min Open valve #3 to wash EV sample on 100-nm filter 5 Bottom-solution
removal 1800 10 s Open valve #4 to remove solution under 100-nm AAO membrane
6 EV retrieval 1800 10 s Open valve #5 to retrieve EV sample Total time < 36 min
EV = extracellular vesicle; TEPC = track-etched polycarbonate; AAO = anodic aluminum oxide.
72
600-nm filter, and residual proteins and lipoproteins are removed by the 100-nm AAO filter. After filtration, valve #3 was opened, and a washing solution (1.6 mL) was transferred through the EV filtration chamber into the waste chamber via spin step #4 (Figure 3.3D). Next, the bottom solution below the AAO filter was removed via spin step #5 to eliminate impurities (Figure 3.3E). Finally, the enriched-plasma EVs were transferred into the elution chamber (Figure 3.3F).
Clinical samples preparation, handling, and storage
Whole-blood was collected from five healthy donors, three prostate cancer patients, and six lung cancer patients. The plasma samples were collected from 30 healthy donors and 43 prostate-cancer patients. All blood and plasma samples were obtained following approval by an institutional review board (IRB 1702-008-051). The biospecimens and corresponding data used in this study were provided by the Biobank of Pusan National University Hospital—a member of the Korea Biobank Network.
Cancer patient blood samples were obtained from the Pusan National University Hospital (IRB H1612- 019-049), while blood/plasma samples of healthy donors were obtained from commercial sources (Innovative Research, MI) and volunteers at the Yeungnam University Medical Center (IRB 2018-04- 011). Blood samples measuring 3 mL each were collected in a vacutainer EDTA collection tube and processed within 2 h after collection.
Preparation of LNEVs for spiking experiments
The EVs used for spiking whole-blood samples were isolated from LNCaP cells obtained from ATCC and cultured at the Roswell Park Memorial Institute medium (Gibco, Thermo Fisher Scientific) supplemented with 10% Exo-Free fetal bovine serum (Systems Biosciences Inc., CA, USA) and 1%
antibiotics/antimycotics. The cells were incubated at 37 °C with 5% CO2. The cell-culture supernatant was collected after 48 h of culture and centrifuged at 300 g for 10 min to remove dead cells, followed by spinning at 2,000 g for 15 min at 4 °C to completely remove dead cells and cellular debris. The supernatant was filtered through a 600-nm filter to remove vesicles measuring more than 600 nm. The EVs were enriched using a disc comprising a 20-nm AAO membrane filter. A standard curve was drawn for CD9-CD81 ELISA using isolated EVs (Figure 3.4), and a known amount of enriched EVs was used to spike whole-blood samples for process optimization.
Ultracentrifugation (UC)
The EVs were isolated from whole-blood samples with or without spiked LNEV using the standard ultracentrifugation process. For plasma preparation, all steps were performed at RT. The whole-blood samples were initially centrifuged at 500 g for 10 min at RT to separate cellular
73
Figure 3.4. CD9-CD81 sandwich ELISA results from LNCaP-cell-derived EVs spiked in the plasma sample. At each experimental condition, the plasma volume and the total sample volume was fixed at 100 μL and 200 μL, respectively. The remaining volume was adjusted by adding PBS. All measurements were performed thrice. The mean values were plotted with the standard deviations represented by error bars.
74
components, and the supernatant, which contained platelet-rich plasma, was centrifuged twice at 2500 g for 15 min each at RT to remove platelets. The clear plasma was filtered through a 600-nm syringe filter and centrifuged in a Beckman Coulter Ultracentrifuge at120,000 g for 90 min at 4 °C using a TLA 120.2 Ti fixed- angle rotor (Beckman Coulter) and 1.2 mL polycarbonate ultracentrifuge tubes to pellet EVs. To eliminate protein contaminants, the supernatant was carefully removed, and the EV pellets were resuspended in PBS and centrifuged again at 120,000 g for 90 min at 4 °C. The supernatant was removed, and the resulting EV pellets were resuspended in the desired volume of PBS to facilitate further analysis.
Nanoparticle tracking analysis (NTA)
The concentration and the size distribution of EVs were measured using an NTA system Nanosight NS500; Malvern Instruments, Malvern, UK). The isolated EV samples were vortexed and diluted with 200-nm pre-filtered PBS to obtain the recommended 25–100 particles/frame of the NTA system. All measurements were performed under identical settings to ensure consistent results. Each sample was analyzed thrice, and mean values were plotted.
CD9-CD81 sandwich ELISA
An EVs solution was prepared to maintain identical input volumes for both isolation methods and to compare the efficiencies of the two EVs isolation methods (Exodisc and UC). A 96-well plate (Corning Inc., NY, USA, cat#3590) was coated with 50 μL of coating antibodies (10 μg/mL anti-CD9 in PBS buffer; MEM 61; Abcam, Cambridge, UK) and incubated overnight at 4 °C. The following morning, the plate was blocked with 1% bovine serum albumin (BSA)-PBS buffer at 37 °C for 1 h.
After washing with 0.1% BSA-PBS buffer (washing buffer), the plate was further incubated with an EV solution in PBS buffer (50 μ L) a t RT f or 2 h. Following removal of the solution, the plate was washed twice with washing buffer, followed by addition of biotin-conjugated secondary antibodies (anti-CD81; LifeSpan Biosciences, Inc., Seattle, WA, USA) in PBS buffer (50 μL; 500 ng/mL), and incubated at RT for 1 h. After washing thrice with the washing buffer, the plate was incubated with a solution of HRP-conjugated streptavidin in PBS buffer (50 μL; 1:500 for LNEV spiked samples and 1:200 for clinical samples) at RT for 30 min followed by three washing procedures using the washing buffer. TMB solution (50 μL) was subsequently added, and the plate was incubated at RT for 15 min, following which, 50 μL of stop solution was added to each well. Solution absorbance was measured using a plate reader spectrophotometer (TECAN, Morrisville, NC, USA) at 450 nm.
SEM imaging
75
The EVs enriched on D100 were fixed using 4% paraformaldehyde in PBS buffer at RT for 20 min. Subsequently, the EVs were washed once with PBS and successively subjected to serial dehydration with 50%, 70%, 90%, and 100% ethanol for 15 min each. Finally, 100% ethanol treatment was repeated, followed by drying of samples. The SEM images of the EVs isolated on D100 were acquired using the SU-8220 cold FE-SEM instrument (Hitachi high technologies, Japan).
Direct ELISA for EV & cancer specific markers
The EVs were lysed for 30 min using RIPA buffer containing 1% proteinase inhibitors on ice with gentle vortices generated at 10-min intervals. A 50-μL volume of 1:50 diluted EV lysates in PBS was used to coat each well of a 96-well plate (Corning Inc., NY, USA, cat#3590) and incubated overnight at 4 °C. The following morning, the plate was blocked with 1% bovine serum albumin (BSA)- PBS buffer at RT for 1 h. After washing with 0.1% BSA-PBS buffer washing buffer), the plate was loaded with primary antibodies (Table 3.2) in PBS buffer (50 μL; 500 ng/mL) and incubated at RT for 1 h. After washing thrice with washing buffer, the plate was incubated along with a solution of HRP- conjugated detection antibodies in PBS buffer (50 μL) at RT for 20 min, followed by three washing procedures using the washing buffer. TMB solution (50 μL) was then added, and the plate was, again, incubated at RT for 15 min. Following which, 50 μL of stop solution was added to each well. Solution absorbance was measured using a plate reader spectrophotometer at 450 nm.
Mouse xenograft model to study tumor progression
Mice were cared for in compliance with the protocol approved by the Institutional Animal Use and Care committee of UNIST (IACUC-2013-013). To perform EV isolation and monitor protein markers within plasma, 5 × 106 PC3 cells in 200 μL of PBS per mouse were subcutaneously injected into the left flank of three 8-week-old male nude mice. Likewise, in the control group, 200 μL PBS was injected into each of the three mice in the group. The blood samples (approximately 100 μL) were collected every week via retro-orbital blood collection for 13 weeks. Subsequently, the animals were sacrificed, and tumor masses were removed and photographed. Calipers were used to determine the tumor length and width, Tumor volume was estimated using the volume formula—D/2 × d2—where D denotes the largest diameter, while d refers to the shortest diameter.
Western blotting and SDS-PAGE gel
Exosome pellets were lysed in RIPA buffer with a protease inhibitor mixed with the sample buffer (Cell biolabs, CA, USA) and boiled for 5 min. Subsequently, the lysates were separated on a 10%
SDS-PAGE gel (Pierce, Rockford, IL, USA) using a Mini-Protean TGX electrophoresis apparatus Bio-
76 Table 3.2. List of antibodies used in this study.
Type Antibody Company Catalogue
Number Species
Primary antibodies
Anti-CD9 antibody [MEM-61] Abcam ab2215 Mouse
monoclonal Anti-PSMA antibody [YPSMA-
1] Abcam ab19071 Mouse
monoclonal Anti-HSP90 antibody [16F1] Abcam ab13494 Rat
monoclonal Anti-EpCAM antibody [MOC-
31] Abcam ab187270 Mouse
monoclonal
Anti-PSA antibody SD Mouse
monoclonal
Human CD63 Purified H5C6 BD BD556019 Mouse
monoclonal
Human CD81 Purified JS-81 BD BD555675 Mouse
monoclonal
Human CD9 Purified M-L13 BD BD555370 Mouse
monoclonal
Purified Mouse Anti-HSP90 BD BD610418 Mouse
monoclonal
Secondary /
Detection antibodies
Anti-CD81 antibody LS LS-C134650
Mouse Monoclonal [clone 1.3.3.22]
Anti-EGFR1 antibody [EFGR-1] Abcam ab24293 Mouse monoclonal Human
PSMA/FOLH1/NAALADase I R&D BAF4234 Polyclonal Sheep IgG Goat Anti-Rabbit IgG H&L
(HRP) Abcam ab6721
rabbit polyclonal IgG
Streptavidin-HRP R&D DY998
ELISA kit/Duo set
Human Serum Albumin DuoSet
ELISA R&D DY1455
Human Kallikrein 3/PSA
immunoassay R&D DKK300
SD company (SD), BD Biosciences (BD), LS Bioscience (LS), R&D Systems (R&D)
77
rad, CA, USA). To facilitate immunoblotting, gels were equilibrated with a transfer buffer [250 mM Tris, 20% methanol (v/v), 200 mM glycine, pH 8.0] 10 min prior to transfer onto nitrocellulose membranes. The said transfer was performed via a cassette of tank transfer with a Mini-protean unit for 1 h at 100 V in accordance with the manufacturer instructions. Post transfer, the membranes were probed into using antibodies specific to albumin (R&D systems, USA). Immuno-detection was then performed using HRP-labelled secondary antibodies and visualized using a LAS 4000 detection system in accordance with the manufacturer protocol (Amersham, UK).
RNA extraction and RT-qPCR
To analyze gene expressions, total RNA was extracted from EVs isolated from 200 μL of plasma spiked with LNEVs using the miRNeasy kit (Qiagen). RNA integrity and quantity were analyzed using an RNA pico-sensitivity kit on a bioanalyzer (Perkin- Elmer). cDNA was prepared using a SuperScript VILO cDNA synthesis kit (Thermo Fisher Scientific). Real-time PCR was performed using the gene expression master mix kit (Thermo Fisher Scientific) and Taqman probe with a QuantStudio 6 real-time PCR instrument (Thermo Fisher Scientific) using the following protocol: 50 °C for 2 min, 95 °C for 10 min followed by 40 cycles of 60 °C for 15 s, using the forward, probe, and reverse primers listed in Table 3.3. All samples were analyzed in triplicate. Data presented as mean ± SE.
Statistical analysis
Optical density (OD) measurements of ELISA assays were performed on 73 samples (i.e., 43 and 30 samples each from prostate-cancer patients and healthy donors, respectively). The measured OD values were normalized along each biomarker for all samples and sorted in descending order with respect to the HSP90-OD level. The normalized OD values were visualized on a heat map (Figure 5A), and eight markers were validated in patients diagnosed with prostate cancer and healthy donors by means of a box plot. A student’s t-test was performed, and p-values below 0.05 were considered statistically to be significant—here, p < 0.05, p < 0.01, p < 0.001, and p < 0.0001 have been represented as *, **, ***, and ****, respectively (Figure 5B). Receiver operation characteristic (ROC) curves and metrics for classifying prostate-cancer patients and normal donors were obtained for each biomarker (Figures 5C–5D and Figure S11). The ROC metrics, (i.e., AUC, SE, accuracy, sensitivity, specificity, and negative likelihood ratio (NLR) cutoff) were calculated using standard approaches. OD measurements corresponding to HSP90 ELISA were depicted in the waterfall plot (Figure 5E). The cutoff value was given by NLR. All computations and visualizations were performed using MATLAB (MathWorks, USA) and OriginPro (OriginLab, USA) packages.
78
Table 3.3. List of primer sequences used in RT-PCR analysis.
GAPDH Forward 5’-ATGGGTGTGAACCATGAGAA-3’
Probe 5’-CCTCAAGATCATCAGCAATGCCTCC-3’
Reverse 5’-GTGCTAAGCAGTTGGTGGTG-3’
CD9 Forward 5’-GGCTTCCTCTTGGTGATATTCG-3’
Probe 5’-TCCTGGACTTCCTTAATCACCTCATCCT-3’
Reverse 5’-GGCTCATCCTTGGTTTTCAG-3’
CD63 Forward 5ʹ-AACGAGAAGGCGATCCATAAG-3ʹ Probe 5ʹ-CCTCGACAAAAGCAATTCCAAGGGC-3ʹ Reverse 5ʹ-GCAGGCAAAGACAATTCCC-3ʹ
CD81 Forward 5’-AGATCGCCAAGGATGTGAAG-3’
Probe 5’-AGCAGTCAAGCGTCTCGTGGAAG-3’
Reverse 5’-AGGTGGTCAAAGCAGTCAG-3’
79 SVM classification
Binary support vector machine (SVM) classification models were created using a set of ELISA data (Figure S12). The SVM models are defined with the third-order polynomial kernel function and are validated by five-fold cross-validation. All the computation was performed using MATLAB classification learner application (MathWorks).