Colocalization of Cancer-Associated Biomarkers on Single Extracellular Vesicles for Early Detection of Cancer

Figures

• 

  Figure 1. Overview of the assay method. Schematic representation of the assay workflow. A: Extracellular vesicles are enriched from human cell line conditioned media, plasma, or serum by size-exclusion chromatography (SEC). Enriched EVs contain cargo and surface biomarkers from their cell of origin. B: Antibodies targeting cancer-associated biomarker antibodies (Abs) are conjugated to magnetic beads (capture antibody) or double-stranded DNA (dsDNA) oligonucleotides (detection antibodies). C: SEC-enriched EVs are captured in solution with magnetic bead–antibody conjugates targeting a specific surface biomarker. D: Immunoaffinity captured EVs are incubated with detection antibodies conjugated to complementary double-stranded DNA probes. The dsDNA oligonucleotides contain single-stranded overhangs that ligate only when in proximity to a complementary probe to generate a template for PCR. The abundance of the detection biomarkers captured on the EVs is read out using quantitative PCR. Positive signal or low cycle threshold (CT) is the result of all three biomarker antibody epitopes being present simultaneously on the surface of the same EV, the absence of one or more biomarkers results in…

• 

  Figure 2. Characterization of human cancer cell line extracellular vesicles. Size-exclusion chromatography (SEC)–enriched EV fractions from human ovarian cancer cell line (COV413A) conditioned media were characterized according to minimal information for studies of extracellular vesicles guidelines. A: Nanoparticle tracking analysis (particle count; greenline) and protein quantification (protein; pinkline) of each fraction collected from the SEC column demonstrating the bulk of the nanoparticles elute in EV1 to EV3, distinct from the bulk of soluble protein (EV6 to EV8). The first fraction (Pre-EV) is the sum of the void and wash volumes of the column, which generally contains larger particles, such as lipoproteins, which are not observed in cell line conditioned media. The purified collection volume (PCV), fractions EV1 to EV3, containing the EVs used in downstream assay steps are highlighted. B: Transmission electron microscopy of EVs derived from the total PCV (pooled fractions EV1 to EV3) from COV413A conditioned media at two different magnifications demonstrates intact EVs ranging from approximately 40 to 300 nm. C: Particle size distribution of pooled SEC fractions EV1 to EV3…

• 

  Figure 3. Bone marrow stromal antigen-2 (BST2), folate receptor-α (FOLR1), and mucin-1 (MUC1) are colocalized with CD63 and CD81 in cell line EVs. EVs derived from cell lines predicted to express high (red), medium (white), or low (blue) levels of the indicated biomarkers based on mRNA analysis were captured with antibody-bead conjugates using anti-CD63 or anti-CD81 antibodies. Detection of bound EVs was determined using oligo-conjugated antibodies against BST2 (A), FOLR1 (B), or MUC1 (C). COV413A EVs were used as a positive control because this line expresses high levels of all three biomarkers studied. For BST2, FOLR1, and MUC1, a second positive cell line, NCI-H1781, NCI-H1819, and HCC-4006, respectively, was also included as a control. Heat maps were included to demonstrate relative expression levels [log2 transcripts per million (TPM)] for each marker, and expression data are derived from the Cancer Cell Line Encyclopedia database at the Broad Institute (https://sites.broadinstitute.org/ccle/datasets, last accessed March 3, 2024).

• 

  Figure 4. Colocalization of biomarkers is required for assay performance and improves signal/noise ratio. All three biomarkers must be present on the same EV for productive readout. If only two biomarkers are present on the EVs (one capture + one detect), no ligation occurs, and CT is high. A–C: Doublet [ie, bone marrow stromal antigen-2 (BST2) capture, folate receptor-α (FOLR1) + FOLR1 detection or BST2 capture, mucin-1 (MUC1) + MUC1 detection; A and B] or triplet combinations (BST2 capture, FOLR1 + MUC1 detection; C) of antibodies tested on cell line EVs from COV413A (BST2high/FOLR1high/MUC1high), OVKATE (BST2high/FOLR1high/MUC1low), SW900 (BST2high/FOLR1low/MUC1high), and G-401 (BST2high/FOLR1low/MUC1low) showing that doublets of BST2 capture/FOLR1 + FOLR1 detection (A), BST2 capture/MUC1 + MUC1 detection (B), or a triplet of BST2 capture/FOLR1 + MUC1 detection (C) only yield positive qPCR signal in cell lines with matching high biomarker expression. Biomarker mRNA abundance (log2 TPM) is represented for each biomarker in each cell line. D: Signal from cell line EV-spiked (dark blue) or unspiked (light blue) healthy human plasma from a representative set of one-biomarker (MUC1 capture,…

• 

  Figure 5. Colocalization of bone marrow stromal antigen-2 (BST2), folate receptor-α (FOLR1), and mucin-1 (MUC1) on single EVs using superresolution microscopy. EVs derived from SKMEL1 (BST2low/FOLR1low/MUC1low; A–D) and COV413A (BST2high/FOLR1high/MUC1high; E–H) cells were captured on an EV Profiler microfluidic slide (ONI) coated with a monoclonal antibody (Ab) to MUC1 (MUC1 capture Ab) before staining with a pan-EV membrane dye (PanEV) and Alexa Fluor antibody conjugates to FOLR1 (yellow; B and D), BST2 (magenta; C and G), or both (D and H). EVs were identified using AutoEV software (ONI) and verified manually. Single, double, and triple positive EVs were quantified by cluster analysis and presented for both SK-MEL-1 (I) and COV413A (J). Experimental details, imaging parameters, and software setting are described in Materials and Methods. Representative images are shown. Scale bar = 100 nm (A–H).

• 

  Figure 6. Discrimination of high-grade serous ovarian cancer (HGSOC) from healthy controls and benign ovarian masses. A: Performance of the mucin-1 (MUC1) capture/bone marrow stromal antigen-2 (BST2) + folate receptor-α (FOLR1) detection biomarker combination in a cohort of healthy controls, patients with benign ovarian masses, and patients with stage I, II, III, and IV HGSOC in a proof-of-concept assay. B: Corresponding receiver operating characteristic curve for discrimination of all HGSOC cases against healthy controls and benign adnexal masses (area under the curve = 0.920). n = 24 healthy controls (A); n = 26 patients with benign ovarian masses (A); n = 9 patients with stage I and II HGSOC (A); n = 15 patients with stage III HGSOC (A); n = 4 patients with stage IV HGSOC (A). FPR, false positive rate; TPR, true positive rate.

• 

  Supplemental Figure S1. Characterization of DNA-antibody (Ab) conjugates. Size-exclusion chromatography traces (absorbance at 260 nm) for unconjugated DNA (green line; top), unconjugated antibody (blue line; middle), and a DNA-antibody conjugation mixture (purple line; bottom). DNA-Ab conjugates elute ahead of the unconjugated components in multiple peaks corresponding to different degrees of labeling. DNA is added in excess relative to unconjugated antibody, resulting in unconjugated DNA in the reaction mixture.

• 

  Supplemental Figure S2. Expression of EV and ovarian cancer biomarkers in EV fractions. A–G: Full-panel blots of data shown in Figure 2 and additional Western blot analysis of whole cell extract (WCE) from COV413A cells and size-exclusion chromatography (SEC) purified fractions from COV413A conditioned media (EV1 to EV6) using antibodies against EV surface biomarkers CD9 (A), CD81 (B), and CD63 (C); EV cargo biomarkers flotillin (D) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH; E) demonstrate the EVs were intact during enrichment; and non-EV internal membrane proteins mitochondrial import receptor subunit TOM20 homolog (TOM20) (F) and ribosomal protein S3 (RPS3) (G) demonstrate the EV fractions are the result of EV biogenesis and do not include intracellular vesicles. H: Total protein staining (Coomassie) was used to validate SEC separation of the EV fractions from the soluble protein component and was confirmed by measuring the protein concentration in each fraction (Figure 2). I–K: Full-panel blots of data shown in Figure 2 demonstrating abundance of bone marrow stromal antigen-2 (BST2; I), folate receptor-α (FOLR1; J), and mucin-1 (MUC1)-C (K) in COV413A EVs. L: Representative molecular weight…

• 

  Supplemental Figure S3. Characterization of human plasma EVs. EVs were enriched from a pool of healthy human plasma samples by size-exclusion chromatography (SEC). Individual fractions were collected and subject to nanoparticle tracking analysis (NTA). A and B: Unconcentrated SEC fractions (EV1, EV2, and EV3) were characterized to determine particle counts (A) and overall size distribution (B) of the enriched nanoparticles. A: Plot showing nanoparticle counts (gray bars) and protein concentration (black line) for the EV-containing SEC fractions from human plasma. The pooled collection volume (PCV), consisting of fractions EV1 to EV3, used in the downstream assay, are highlighted. The Pre-EV sample is a sum of the void volume and the wash fraction containing lipoproteins and larger, non-EV particles (data not shown). B: NTA size-distribution plot of individual fractions EV1, EV2, and EV3 demonstrating particles within the size range of 40 to 400 nm. Buffer alone [phosphate-buffered saline (PBS)] was used to determine the background. C: In parallel, concentrated SEC fractions were analyzed by Western blot analysis against normal human plasma (input) and COV413A whole cell extract (WCE) as a positive…

• 

  Supplemental Figure S4. Screening antibodies (Abs) for functionality and specificity for capture and detection of EVs. A–F: Distinct monoclonal antibody clones against folate receptor-α (FOLR1; A and B), bone marrow stromal antigen-2 (BST2; C and D), and mucin-1 (MUC1; E and F) were screened for their ability to capture (A, C, and E) and detect (B, D, and F) size exclusion chromatography–purified EVs from human cell lines known to be positive [COV413A (+)] or negative [SKMEL1 (-)] for FOLR1, BST2, and MUC1. A, C, and E: Each antibody was independently conjugated to magnetic beads at a fixed density and used to capture cell line EVs in solution. Captured EVs were detected using complementary anti-CD81 double-stranded DNA-conjugated detection antibodies. Antibody clones demonstrating both functionality (greatest signal to noise between the positive and negative cell line) and specificity (low CT in positive cell line and high CT in negative cell line) were further tested and optimized (data not shown). No EV controls were used to eliminate cross-reactive or sticky antibodies. On the basis of performance in immunocapture and other criteria, selected antibodies were subsequently tested for their ability to…

• 

  Supplemental Figure S5. Impact of assay configuration on discrimination of spiked and unspiked plasma. Assay signal from COV413A cell line EV-spiked and unspiked human plasma for the biomarker combination bone marrow stromal antigen-2 (BST2)/folate receptor-α (FOLR1)/mucin-1 (MUC1) using different configurations of capture and detection biomarkers. MUC1 capture exhibited the greatest separation and strongest signal among the three configurations tested.

• 

  Supplemental Figure S6. EV capture efficiency in COV413A and SK-MEL-1 cell line EVs. A total of 4 × 1010 EVs from COV413A [mucin-1 (MUC1)high/folate receptor-α (FOLR1)high/bone marrow stromal antigen-2 (BST2)high] or SK-MEL-1 (MUC1low/FOLR1low/BST2low) were captured on EV Profiler 2 microfluidic chips (ONI) using a cocktail of antibodies (Abs) to CD9, CD63 and CD81 (TetTrio; A) or MUC1 (B). Captured EVs were detected using a Pan-EV membrane dye using an ONI Nanoimager (software version 1.19.15.20240417195007) and ONI AutoEV (beta software version) to detect Pan-EV signal clusters, as described in Materials and Methods. No EV controls were used to establish non-specific fluorescence background signal.