Clinical translation of exosome-derived biomarkers for diseased and malignant cells is hindered by the lack of rapid and accurate quantification methods. This report describes the use of low-magnification dark-field microscope images to quantify specific exosome subtypes in small volume serum or plasma samples.
Infected or malignant cells frequently secrete more exosomes, leading to elevated levels of disease-associated exosomes in the circulation. These exosomes have the potential to serve as biomarkers for disease diagnosis and to monitor disease progression and treatment response. However, most exosome analysis procedures require exosome isolation and purification steps, which are usually time-consuming and labor-intensive, and thus of limited utility in clinical settings. This report describes a rapid procedure to analyze specific biomarkers on the outer membrane of exosomes without requiring separate isolation and purification steps. In this method, exosomes are captured on the surface of a slide by exosome-specific antibodies and then hybridized with nanoparticle-conjugated antibody probes specific to a disease. After hybridization, the abundance of the target exosome population is determined by analyzing low-magnification dark-field microscope (LMDFM) images of the bound nanoparticles. This approach can be easily adopted for research and clinical use to analyze membrane-associated exosome biomarkers linked to disease.
Exosomes are released from most cell types and play a key role in cell-to-cell communications, including pathophysiological processes associated with various diseases, since they can be home to specific tissues or cell types, and contain a variety of nucleic acids, proteins, and lipids that reflect their cell of origin and can exert regulatory effects on their recipient cells1,2,3,4. Exosomes are often secreted at elevated levels in disease states, can interact with both adjacent and distant cells, and are found at relatively high concentration in the circulation as well as most other body fluids, including saliva, urine, pancreatic and bile juice, and bronchoalveolar lavage fluid5,6,7,8,9,10,11. This abundance and stability of exosomes in human body fluids, coupled with their information-rich nature, makes them ideal biomarkers for disease diagnosis and treatment monitoring.
This includes tumor-derived exosomes (TDEs), which contain tumor-specific or selective factors, which can serve as disease biomarkers, including tumor-associated mutant alleles. TDEs can participate in the remodeling of the tumor microenvironment to facilitate tumor development and metastasis, and regulate anti-tumor responses12. Increased TDE secretion is a common phenotype of most cancers and several features of the tumor microenvironment, including hypoxia, acidic pH, and inflammation, are known to promote exosome secretion. Surprisingly, given the number of cells that secrete exosomes, an increase in total exosome level can, itself, function as a cancer biomarker. For example, a recent study found that the total EV concentration in bile juice discriminates malignant and nonmalignant in common bile duct stenosis patients with 100% accuracy7. Similar results have been found with studies using other body fluids, including plasma. However, due to the potential for subject to subject variation, and other confounding factors, most studies investigating exosomes as disease biomarkers have focused on detection of biomarker that are selectively associated with TDEs instead of total exosome numbers.
Translating exosome biomarkers into clinical practice, however, remains challenging since most reported exosome assay approaches require time-consuming and labor-intensive isolation procedures 13. Currently popular exosome isolation methods include ultracentrifugation, density gradients, size-exclusion, co-precipitation, affinity capture, and microfluidic isolation approaches. Ultracentrifugation is the “gold standard” method, and is most commonly used for exosome isolations, but this procedure is time-consuming and results in exosome damage and exosome membrane clustering, and produces exosome samples that are contaminated with proteins, lipoproteins and other factors that can influence subsequent analyses14. Most exosome isolation methods, including ultracentrifugation, cannot separate exosomes (30–150 nm) from micro-vesicles (100–1000 nm) and apoptotic bodies (100–5000 nm), which arise through different mechanisms and have different functions, due to the size overlap among these groups, and the diversity of exosome populations15. New approaches are needed to improve the sensitivity and reproducibility of exosome assays by improving exosome recovery while reducing exosome damage and contamination, although any assays based on such methods will also need to be optimized to render them suitable for translation to applications in clinical settings.
Several recent studies have proposed to employ integrated platforms to capture and analyze exosomes directly from body fluids. These methods employ microfluidic, electrokinetic, affinity capture, and various other methods for exosome isolation, and electrochemistry, surface plasmon resonance, and other methods to detect captured exosomes. It is not clear how feasible many of these approaches will be in clinical settings, due to their complexity, expense, low-throughput or other issues.
We have developed a rapid and inexpensive assay that can be used for sensitive and specific quantification of total exosomes and specific exosome subtypes, including disease-associated exosomes, such as TDEs, which requires only a small amount of sample and which employs a streamlined workflow suitable for clinical environments. In this assay, a slide is coated with antibodies that bind either an exosome-specific or disease-specific marker expressed on the exosome surface in order to directly capture target exosomes present in small volume plasma or serum samples applied to wells on the slide. Captured exosomes are then hybridized with an antibody-conjugated nanoparticle that recognizes the biomarker of interest on these exosomes, which can be either a general exosome marker or a factor specific for an exosome subtype of interest. Images of these sample wells are then captured using a dark field microscope (DFM) and analyzed to measure the light scattered from nanoparticles bound to exosomes of interest captured in each sample well6,16,17. Notably, imaging an entire sample well by low-magnification DFM (LMDFM) avoids a selection bias encountered with high magnification DFM analyses when users must directly choose which fields to capture for subsequent image analysis. LMDFM image analysis is subject to light scattering artifacts from surface irregularities, including scratches and sample debris, but this background can be reduced using a simple noise-reduction algorithm we developed to run on the NIH image analysis program, ImageJ (https://imagej.nih.gov/ij/). This algorithm first applies an input contour threshold that is used to detect the boundaries of the sample well to define the region of the image for subsequent analysis. The region defined by this contour region is then split to separate signal present in the red, blue and green channels of the image and the blue channel is subtracted from the red channel to remove signal arising from surface artifacts and uneven lighting from nanorod signal.
This article describes how to use this assay to rapidly quantify either total or specific exosome levels in plasma or serum samples.
1. Preparation of nanoparticle probes
NOTE: This assay utilizes Functionalized Gold Nanorods (AuNRs; 25 nm diameter x 71 nm length) that are covalently conjugated with neutravidin polymers (AV) and have a surface plasmon resonance peak that produces a red (641 nm peak) scattering signal upon DFM illumination.
2. Preparation of EV capture slides
3. Standard curve preparation
4. Processing human plasma or serum samples
5. Exosome capture and detection
6. DFM Image Capture
7. DFM image analysis
Multi-well protein A/G coated slides (Figure 1A) were functionalized with anti-EphA2 antibody and then used to specifically capture EphA2-positive exosomes from serum samples of patients with and without pancreatic cancer (1 µL/well) and incubated with gold nanorods conjugated with an anti-CD9 antibody (Figure 1B). DFM images of light scattered from these nanoparticles were analyzed using the DSM plugin in ImageJ software to quantify the bound EphA-2 positive exosomes in each well. The DSM algorithm automatically defines the boundary of a sample well, filters the noise from artifacts, calculates the scattered signal from each well, and outputs this information (Figure 2). The DSM algorithm strongly attenuates light scattering artifacts from scratches or debris present in the sample and improves the sensitivity and reproducibility of nanoparticle detection and can automatically process a batch of slide images for high-throughput use. This algorithm uses ImageJ commands and parameters input by the user to subtract image background, calculate the scattering signal from each well, and output a data and image file (Figure 3). Regions of interest are defined by the high intensity well boundaries in the capture image using the contour threshold function of the ImageJ macro program. Image analyses use a predefined contour threshold and image parameters to calculate the intensity of nanoparticle scattering for each well.
As reported in our previous work (supplementary information of Liang et al.6), exosomes isolated from PANC-1 cell cultures characterized by transmission electron microscopy and Western blot exhibited the size range, morphology, and protein marker expression consistent with a high purity exosome sample. The PANC-1 exosomes, prepared with the same procedure as our previous work, were used here to validate our nPES assay for exosome quantification. This assay used an anti-EphA2 antibody to capture a large population of exosomes from the total exosome population and an antibody against the general exosome protein CD9 to detect captured exosomes. Results obtained using serially diluted PANC-1 exosome samples, with protein concentrations ranging from 0.24 to 1.2 µg/µL, showed good reproducibility in replicate wells (Figure 4A) and a strong linear correlation between the scatter response and exosome protein concentration (Figure 4B).
To demonstrate the potential application of this method, serum samples from patients with and without pancreatic cancer were analyzed to detect the abundance of serum exosomes expressing the cancer-associated biomarker EphA2, using an anti-EphA2 antibody to directly capture target exosomes from serum and nanoparticles conjugated with an anti-CD9 antibody to detect bound exosomes. This analysis revealed that serum samples from the pancreatic cancer patients had significantly higher levels of EphA2+ exosomes (Figure 5) than their controls.
Figure 1: Exosome quantification scheme. (A) Schematic of multi-well protein A/G slides (192 wells) used in this assay. (B) Target exosomes are directly captured from samples, including serum and plasma, by surface immobilization on the capture antibody (e.g. anti-EphA2 antibody) bound to the slide, and then incubated with gold nanorods conjugated with a detection antibody (e.g. anti-CD9 antibody) before analysis by DFM image analysis. Please click here to view a larger version of this figure.
Figure 2: Exosome quantification by applying the DSM algorithm to LMDFM images. Low-magnification images are processed with the DSM algorithm to eliminate background signal and signal artifacts that can arise from scratches, mixing voids, debris, and uneven sample illumination to allow robust detection of gold nanorod signal, which correlates with exosome concentration. This figure has been adapted with permission from Sun, D. et al. Noise Reduction Method for Quantifying Nanoparticle Light Scattering in Low Magnification Dark-Field Microscope Far-Field Images. Analytical Chemistry. 88 (24), 12001-12005 (2016). Copyright (2016) American Chemical Society. Please click here to view a larger version of this figure.
Figure 3: Schematic of the DSM algorithm commands and outputs. Indicated steps use native commands from ImageJ and all input parameters are selected according to the requirements of experiments via graphical user interface. This figure has been adapted with permission from Sun, D. et al. Noise Reduction Method for Quantifying Nanoparticle Light Scattering in Low Magnification Dark-Field Microscope Far-Field Images. Analytical Chemistry. 88 (24), 12001-12005 (2016). Copyright (2016) American Chemical Society. Please click here to view a larger version of this figure.
Figure 4: Representative nPES LMDFM images and DSM output data. (A) LMDFM image of an nPES assay analyzing a concentration gradient of PANC-1 exosomes and (B) the linear correlation of the optical signal and exosome concentration from this slide (left to right: 0.24, 0.356, 0.53, 0.80, 1.20 µg/µL respectively for each column). Data are presented as Mean±SE, n=6, with a Pearson correlation coefficient R2 = 0.99 and coefficient of variation for the replicates of each concentration < 0.2. Please click here to view a larger version of this figure.
Figure 5: LMDFM signal of EphA2+ exosomes differs in serum from patients with and without pancreatic cancer. Serum samples analyzed by nPES using an anti-EphA2 capture antibody (cancer-associated) and an anti-CD9 detection antibody (general exosome marker) exhibited a significant difference in the concentration of EphA2+ exosomes in serum samples of patients with and without pancreatic cancer (N = 7/group). Results are presented as Mean±SE. **p = 0.002 by Mann-Whitney U-test (two-sided). This figure has been modified from [Sun, D. et al. Noise Reduction Method for Quantifying Nanoparticle Light Scattering in Low Magnification Dark-Field Microscope Far-Field Images. Analytical Chemistry. 88 (24), 12001-12005 (2016)]. Please click here to view a larger version of this figure.
Exosomes arise from regulated invaginations of the outer endosome membrane that produce multivesicular bodies, a specialized subset of endosomes that contain a large number of intraluminal vesicles that undergo fusion with the plasma membrane to release mature exosomes into the extracellular space. Due to this biogenesis pathway, exosomes can carry membrane-bound factors associated with membrane fractions that comprise or fuse with the endosome membrane, as well as multiple different types of cytosolic components, and thus contain cargoes of proteins, DNA and various RNA subtypes (mRNAs, microRNAs, long non-coding RNAs) that can reflect the phenotype of their cell of origin20. Since exosomes are secreted by most if not all cell types, can exhibit increased secretion from diseased or malignant cells, and accumulate in most body fluids, exosomes are the subject of comprehensive and systematic investigation as a promising minimally invasive means to detect specific disease conditions and monitor their responses to treatment21.
Exosome isolation, which is required for most current exosome analyses, is a lengthy and labor-intensive procedure, restricting the clinical translation of exosome-associated biomarkers with potential medical relevance. Many common isolation methods (ultracentrifugation, size-exclusion, precipitation, etc.) often do not sufficiently distinguish exosomes (30–100 nm) from micro-vesicles (100–1000 nm) and apoptotic bodies (100–5000 nm) due to overlaps in their size ranges or physical properties or may damage exosome integrity15. New approaches are under development that may permit more rapid exosome analyses, but it is not clear how feasible it may be to implement many of these platforms in clinical settings due.
In this report, we present a novel approach that allows nanoparticle-based high-throughput exosome quantification using low magnification dark field microscope images. This method does not require exosome purification, expensive dedicated equipment, or novel technical expertise and should thus be amenable to rapid translation in most research and clinical settings. Our assay can be applied to precisely quantify the concentration of a target exosome population bearing a specific biomarker when assay results are compared to a standard curve, since our results exhibit there is a strong linear correlation (R2 = 0.99) between optical response and exosome concentration. To demonstrate the real world potential of this approach, we have provided data where we employed this method to quantify the concentration of an exosome biomarker associated with pancreatic cancer in serum samples obtained from patients with and without pancreatic cancer.
In LMDFM, the entire sample well is imaged to avoid the selection bias found in high-magnification DFM analyses, where users must directly choose the sample fields to capture for subsequent image analysis, but is subject to light scattering artifacts from surface irregularities, including scratches and sample debris. This background can be reduced to detect target exosome-derived signal using our DSM noise-reduction algorithm that runs on the NIH image analysis program, ImageJ, but care must still be taken to avoid introducing such artifacts which may reduce the dynamic range of the assay.
Materials used in this assay:
Multi-well SuperProtein A/G slides holding 1 µL/well were purchased from Arrayit Corporation (AGMSM192BC). Nanoparticles were obtained from Nanopartz (C12-25-650-TN-DIH-50-1, 6.4 x 1012/mL). DFM images were captured with a Nikon DiR2 digital camera attached to a Nikon Ti-Eclipse microscope, with consistent lighting and a 1/220 s exposure time. The PANC-1 cell line used in this study was purchased from the American Type Culture Collection.
The authors have nothing to disclose.
The work was primarily supported by research funding provided by NIH (U01CA214254, R01HD090927, R01AI122932, R01AI113725 and R21Al126361-01), Arizona Biomedical Research Commission (ABRC) young investigator award.
Eppendorf Repeater stream | Fisher Scientific | 05-401-040 | |
Eppendorf Research plus | Eppendorf | 3120000011 | 0.1 – 2.5 µL, dark gray |
Functionalized Gold Nanorods | Nanopartz | C12-25-650-TN-DIH-50-1 | In vitro neutravidin polymer functionalization |
HulaMixer Sample Mixer | Thermo Fisher Scientific | 15920D | |
Incu-shaker 10L | Benchmark Scientific | H1010 | |
Inverted Research Microscope | Nikon | Ti-DH | With Dark field condenser, DS-Ri2 camera, and Ti-SH-U universal holder, and motorized stage |
NIS-Elements | Nikon | Microscope imaging software | |
Phosphate Buffered Saline (1X) | GE Healthcare Life Sciences | SH30256.02 | HyClone |
Protein A/G Treated Glass Substrate Slides | Arrayit Corp. | AGMSM192BC | Premium microarray substrate |
Q500 Sonicator | Qsonica, LLC | Q500-110 | With standard probe (#4220) |
Superblock blocking buffer | Thermo Scientific | ||
TWEEN 20 | Sigma Life Sciences | 9005-64-5 |