Framework for rapid comparison of extracellular vesicle isolation methods

  1. Dmitry Ter-Ovanesyan
  2. Maia Norman
  3. Roey Lazarovits
  4. Wendy Trieu
  5. Ju-Hyun Lee
  6. George M Church
  7. David R Walt  Is a corresponding author
  1. Wyss Institute for Biologically Inspired Engineering, United States
  2. Tufts University School of Medicine, United States
  3. Department of Pathology, Brigham and Women’s Hospital, United States
  4. Harvard Medical School, United States

Abstract

Extracellular vesicles (EVs) are released by all cells into biofluids and hold great promise as reservoirs of disease biomarkers. One of the main challenges in studying EVs is a lack of methods to quantify EVs that are sensitive enough and can differentiate EVs from similarly sized lipoproteins and protein aggregates. We demonstrate the use of ultrasensitive, single-molecule array (Simoa) assays for the quantification of EVs using three widely expressed transmembrane proteins: the tetraspanins CD9, CD63, and CD81. Using Simoa to measure these three EV markers, as well as albumin to measure protein contamination, we were able to compare the relative efficiency and purity of several commonly used EV isolation methods in plasma and cerebrospinal fluid (CSF): ultracentrifugation, precipitation, and size exclusion chromatography (SEC). We further used these assays, all on one platform, to improve SEC isolation from plasma and CSF. Our results highlight the utility of quantifying EV proteins using Simoa and provide a rapid framework for comparing and improving EV isolation methods from biofluids.

Introduction

Extracellular vesicles (EVs) are released by all cell types and are found in biofluids such as plasma and cerebrospinal fluid (CSF). EVs contain contents from their donor cells, providing broad non-invasive access to molecular information about cell types in the human body that are otherwise inaccessible to biopsy (Hirshman et al., 2016). Despite the diagnostic potential of EVs, there are several challenges that have hampered their utility as biomarkers. EVs are heterogeneous, present at low levels in clinically relevant samples, and difficult to quantify (Tkach et al., 2018; Hartjes et al., 2019; Shao et al., 2018). Due to these challenges, there is a lack of consensus about the best way to isolate EVs from biofluids (Coumans et al., 2017a; Konoshenko et al., 2018; Théry et al., 2018).

Several techniques have been used in attempts to quantify EVs. These methods, such as nanoparticle tracking analysis (NTA), dynamic light scattering, and tunable resistive pulse sensing, aim to measure both particle size and concentration (Hartjes et al., 2019). A major limitation of these methods is that they cannot discriminate lipoproteins or particles of aggregated proteins from EVs (Coumans et al., 2017a; Sódar et al., 2016; Welton et al., 2015; Lee et al., 2019; Webber and Clayton, 2013; Johnsen et al., 2019). In addition, they are all physical methods that provide no information about the biological nature of the particles being measured. Since biofluids, and plasma in particular, contain an abundance of lipoproteins and protein aggregates at levels higher than those of EVs (Sódar et al., 2016; Simonsen, 2017), these methods are ill-suited for quantifying EVs (Tkach et al., 2018). Lipid dyes have also been used to label and measure EVs (Osteikoetxea et al., 2015; Visnovitz et al., 2019), but these dyes also bind to lipoproteins and lack sensitivity (Tkach et al., 2018). There are also numerous efforts to apply flow cytometry to the analysis of EVs, but due to the small size of EVs, obtaining quantitative measurements using this approach remains challenging (Lucchetti et al., 2020; Kuiper et al., 2021; Welsh et al., 2020).

A feature of EVs that distinguishes them from both lipoproteins and free protein aggregates is the presence of transmembrane proteins that span the phospholipid bilayer (Simonsen, 2017). The tetraspanins CD9, CD63, and CD81 are transmembrane proteins that are widely expressed and readily found on EVs, often referred to as ‘EV markers’ (Tkach et al., 2018). Although none of these proteins is present on every EV, measuring three tetraspanins should be a reliable proxy for EV abundance in many contexts. We reasoned that by using immunoassays to compare the levels of tetraspanins from a given biofluid, as well as albumin as a representative free protein, we could quantitatively compare the purity and yield of different EV isolation methods.

The most commonly used method for measuring proteins in biofluids is enzyme-linked immunosorbent assay (ELISA), but this technique lacks the sensitivity to detect low-abundance proteins (Coumans et al., 2017b). Single-molecule array (Simoa) technology, previously developed in our lab but now commercially available, converts ELISA into a digital readout (Rissin et al., 2010). Simoa assays can be orders of magnitude more sensitive than traditional ELISAs (Cohen and Walt, 2019), which is particularly useful for EV analysis as the levels of EV proteins are often low in clinical biofluid samples (Coumans et al., 2017b). We have previously applied Simoa to the investigation of L1CAM, a protein thought to be a marker of neuron-derived EV, showing it is not associated with EVs in plasma and CSF (Norman et al., 2021).

In this study, we demonstrate the application of Simoa for relative EV quantification by comparing different EV isolation methods from human biofluids. In particular, we applied Simoa to compare EV isolation methods from human plasma and CSF using three of the most commonly used isolation techniques: ultracentrifugation, precipitation, and size exclusion chromatography (SEC). By also measuring levels of albumin using Simoa, we were able to determine both relative purity and yield for each technique in the same experiment. We then applied these Simoa assays to screen several parameters of SEC and develop improved EV isolation methods from plasma and CSF, demonstrating the utility of this approach for EV analysis.

Results

Framework for quantifying relative EV yield and purity

We set out to quantify the relative difference in yield and purity for different EV isolation methods. Starting with aliquots of the same biofluid, we reasoned that by measuring the tetraspanins CD9, CD63, and CD81 using different isolation methods, we could directly compare EV yield. By also measuring albumin, the most abundant free protein in plasma and CSF, we could compare the purity of these methods. Using Simoa technology, an ultrasensitive digital ELISA, to measure all four of these proteins, we could compare EV yield and purity on one platform with high sensitivity (Figure 1a).

Overview of experimental framework for EV detection using Simoa and size exclusion chromatography (SEC).

(a) Different methods of EV isolation can be directly compared to assess yield and purity by measuring the three tetraspanins (CD9, CD63, and CD81) and albumin. (b) Single immuno-complexes are formed by binding the target tetraspanin protein on EVs to a magnetic bead conjugated to a capture antibody and a biotin-labeled detection antibody. Detection antibodies are labeled with a streptavidin-conjugated enzyme. The beads are then loaded into individual wells of a microwell array where each well matches the size of the magnetic bead limiting a maximum of one bead per well. Wells with the full immuno-complex (on wells) produce a fluorescent signal upon conversion of substrate, unlike wells with beads lacking the immuno-complex (off wells”). (c) EV and free proteins such as albumin in a biofluid sample are separated by SEC. Free proteins elute from the column in later fractions than EVs because free proteins are smaller than the pore size of the beads while EVs are larger and are excluded from entering the beads. EV, extracellular vesicle.

Although Simoa is generally used to quantify free proteins, it can also be used to analyze EV transmembrane proteins. In Simoa, unlike in traditional ELISA, individual immuno-complexes are isolated into femtoliter wells that fit only one bead per well. In a given sample, there are many more antibody-bound beads than target proteins, and therefore Poisson statistics dictate that only a single immuno-complex is present per well. This allows counting ‘on wells’ as individual protein molecules (Figure 1b). The percentage of ‘on wells’ can then be converted to protein concentration by comparing to a calibration curve of recombinant protein standard. We previously developed and validated Simoa assays for the proteins CD9, CD63, and CD81, showing that they are 1–3 orders of magnitude more sensitive than the corresponding standard ELISA assays with the same pairs of antibodies (Norman et al., 2021).

Comparison of existing EV isolation methods

We started by using Simoa to directly compare EV isolation methods commonly used in biomarker studies. For each method, we used identical 0.5 ml samples of human plasma or CSF that were pooled and aliquoted, allowing us to directly compare the different methods. To separate EVs from cells, cell debris, and large vesicles, all samples were first centrifuged and then filtered through a 0.45-μm filter. We compared three methods and chose two variations for each method: ultracentrifugation (with or without a wash step), two commercial precipitation kits (ExoQuick and ExoQuick ULTRA), and two commercially available SEC columns (Izon qEVoriginal 35 nm and 70 nm).

SEC separates EVs from free proteins based on size; proteins enter porous beads and elute from the column later than the EVs, which are much larger and less likely to enter the beads (Figure 1c). Whereas the ultracentrifugation and precipitation conditions each yielded a single sample (performed on 2 separate days and averaged; Figure 2—figure supplement 1), we collected several fractions for SEC and analyzed each fraction to assess the distribution of EVs relative to albumin.

We quantified EVs by measuring the levels of CD9, CD63, and CD81 across the different EV isolation methods in both plasma and CSF (Figure 2a). Since we are interested in all EVs, as opposed to subsets with a specific marker, we quantified EV yield by averaging the levels of the three tetraspanins. We first used the Simoa measurement (in picomoles, determined relative to a corresponding recombinant protein standard) to calculate EV recovery for each individual marker by normalizing the level of tetraspanin in each condition to the amount of that tetraspanin in fractions 7–10 of the Izon qEV 35 nm SEC column (the condition with the highest EV levels in plasma). Next, we averaged the relative tetraspanin recovery values across the three tetraspanins to calculate relative EV recovery.

Figure 2 with 1 supplement see all
Comparison of existing methods for EV isolation in plasma and CSF.

(a) Schematic of experimental outline. (b–d) Individual tetraspanin yields using different isolation methods from plasma. (e) Relative EV recoveries from plasma were calculated by first normalizing individual tetraspanin values (in pM) in each technique to those of Izon qEVoriginal 35 nm EV fractions 7-10 and then averaging the three tetraspanin ratios. (f) Albumin levels using different EV isolation methods from plasma. (g–i) Individual tetraspanin yields using different isolation methods from CSF. (j) Relative EV recoveries in CSF were calculated by first normalizing individual tetraspanin values (in pM) in each technique to those of Izon qEVoriginal 35 nm fractions 7-10 and then averaging the three tetraspanin ratios. (k) Albumin levels using different EV isolation methods from CSF. CSF, cerebrospinal fluid; EV, extracellular vesicle.

Figure 2—source data 1

Comparison of existing methods for EV isolation in plasma and CSF.

https://cdn.elifesciences.org/articles/70725/elife-70725-fig2-data1-v2.xlsx

After determining combined relative EV recovery and albumin concentration for each EV isolation method, we could directly compare EV recovery and purity in both plasma and CSF. In plasma, we found that the Izon qEVoriginal 35 nm SEC column (collecting fractions 7–10) yielded both the highest recovery of EVs and the highest purity (ratio of EVs to albumin) of EVs (Figure 2b–f). In contrast, in CSF, ExoQuick yielded the highest recovery of EVs while Izon qEVoriginal 70 nm yielded the highest purity (Figure 2g–k).

Application of Simoa for custom SEC column optimization

Based on the promising results of commercial SEC columns relative to other methods, we sought to use our assays to further investigate SEC using custom columns. First, we designed an SEC stand that allows for reproducible collection of fractions and multiple columns to be run in parallel (Figure 3—figure supplement 1). We next took advantage of Simoa’s high throughput screening capability to help identify the EV-containing fractions in SEC. This enabled us to optimize EV isolation from 0.5 ml samples of plasma and CSF using SEC. We prepared our own columns to systematically test several parameters: column height (10 or 20 ml) and resin (Sepharose CL-2B, CL-4B, or CL-6B).

This comprehensive comparison led us to several conclusions. First, we found that resins with smaller pore sizes led to higher yields of EVs. To confirm this result with another technique, we also observed the same result with Western blotting for the tetraspanins (Figure 3—figure supplement 2). Sepharose CL-6B, which has the smallest pore size, gave the highest yield, although it was accompanied by higher albumin contamination. For all SEC columns, higher purity could also be achieved by taking a smaller number of fractions (e.g., 7–9 instead of 7–10), albeit at the expense of lower EV yield. Additionally, we found that doubling the height of any given column from 10 to 20 ml resulted in better separation between EVs and free proteins, leading to higher purity but lower EV recovery (Figures 3 and 4). When we compared loading different sample volumes in a 10-ml Sepharose CL-6B column, we found that as expected, larger loading volumes led to lower purity in both plasma (Figure 3—figure supplement 3) and CSF (Figure 4—figure supplement 1).

Figure 3 with 5 supplements see all
Comparison of SEC methods for EV isolation in plasma.

(a) Levels of tetraspanins and albumin in plasma after fractionation with 10 ml custom columns filled with Sepharose CL-6B (top), Sepharose CL-4B (middle), and Sepharose CL-2B (bottom). (b) Levels of tetraspanins and albumin in plasma after fractionation with Izon qEVoriginal 35 nm column (top) and Izon qEVoriginal 70 nm column (bottom). (c) Levels of tetraspanins and albumin in plasma after fractionation with 20 ml custom columns; Sepharose CL-6B (top), Sepharose CL-4B (middle), and Sepharose CL-2B (bottom). EV, extracellular vesicle; SEC, size exclusion chromatography.

Figure 4 with 3 supplements see all
Comparison of SEC methods for EV isolation in CSF.

(a) Levels of tetraspanins and albumin in CSF after fractionation with 10 ml custom columns filled with Sepharose CL-6B (top), Sepharose CL-4B (middle), and Sepharose CL-2B (bottom). (b) Levels of tetraspanins and albumin in CSF after fractionation with Izon qEVoriginal 35 nm column (top) and Izon qEVoriginal 70 nm column (bottom). (c) Levels of tetraspanins and albumin in CSF after fractionation with 20 ml custom columns; Sepharose CL-6B (top), Sepharose CL-4B (middle), and Sepharose CL-2B (bottom). CSF, cerebrospinal fluid; EV, extracellular vesicle; SEC, size exclusion chromatography.

Direct comparison of custom SEC and previous methods

Since we used the same pools of biofluids for these experiments, combining all of the data we generated, we were able to perform a direct, quantitative comparison of the relative yields and purities of EVs across all methods tested. We analyzed these results for both plasma (Figure 3—figure supplement 4) and CSF (Figure 4—figure supplement 2). Since all Simoa measurements were performed with two technical replicates, we also confirmed that Simoa had high reproducibility between technical replicates (Figure 3—figure supplement 5 and Figure 4—figure supplement 1). Our analysis shows that 10 ml Sepharose CL-6B column demonstrated the highest recovery in both plasma and CSF. The 20 ml Sepharose CL-4B column gave the highest purity (ratio of EVs to albumin) for plasma, while for CSF, the 10 ml Sepharose CL-4B column had higher purity than the 20 ml Sepharose CL-4B column. Although the 10 ml column had more albumin contamination in the EV fractions than the 20 ml column, the relative ratio of EVs to albumin was higher. Based on these results, we could select the best custom SEC column for either high yield or high purity isolation (Table 1).

Table 1
Recommendations for SEC columns for EV isolation from plasma and CSF.
High yieldHigh purity
PlasmaSepharose CL-6B10 ml column fractions 7–10Sepharose CL-4B20 ml column fractions 14–17
CSFSepharose CL-6B10 ml column fractions 7–10Sepharose CL-4B10 ml column fractions 7–10

Comparison of top custom SEC methods for plasma and CSF

Based on our results surveying the different SEC resins and column heights, we performed additional experiments to more accurately quantify the best high yield and high purity SEC methods for plasma and CSF using another batch of biofluids with more replicates (four columns per condition). For both plasma and CSF, we compared the Sepharose CL-2B 10 ml column, used in the original SEC EV isolation publication (Böing et al., 2014) and in most subsequent SEC publications (Monguió-Tortajada et al., 2019), to the ‘high yield’ Sepharose CL-6B 10 ml column. We also included a Sepharose CL-4B column as the ‘high purity’ column but, as plasma has much higher protein concentration than CSF, used 20 ml of resin for plasma and 10 ml for CSF.

Our results allow us to directly quantify the difference in EVs and albumin across these methods (Figure 5). We found that, in plasma, the Sepharose CL-6B 10 ml column provided over twofold more EVs relative to the Sepharose CL-2B 10 ml column, but also sixfold more albumin. The Sepharose CL-4B 20 ml column, on the other hand, had similar EV levels to that of Sepharose CL-2B 10 ml column in plasma but had sixfold less albumin (Figure 5a–e), demonstrating a large increase in relative purity (EV to albumin ratio) (Figure 5f). In CSF, the Sepharose CL-6B 10 ml column led to a large increase in EV yield relative to the Sepharose CL-2B 10 ml column (Figure 5g–k), but the Sepharose CL-4B 10 ml column did not lead to improved purity (Figure 5l).

Comparison of top custom SEC methods in plasma and CSF.

Error bars represent the standard deviations from four replicates of each column. (a–c). Individual tetraspanin yields using different isolation methods from plasma. (d) Relative EV recoveries from plasma were calculated by first normalizing individual tetraspanin values (in pM) in each technique to those of the Sepharose CL-2B 10 ml column (fractions 7–10) and then averaging the three tetraspanin ratios. (e) Albumin levels using different EV isolation methods from plasma. (f) EV purity for each method in plasma is calculated as the ratio of the sum of tetraspanin concentrations divided by albumin concentration. (g–i) Individual tetraspanin yield using different isolation methods from CSF. (j) Relative EV recoveries in CSF were calculated by first normalizing individual tetraspanin values (in pM) in each technique to those of Sepharose CL-2B 10 ml (fractions 7–10) and then averaging the three tetraspanin ratios. (k) Albumin levels using different EV isolation methods from CSF. (l) EV purity for each method in CSF is calculated as the ratio of the sum of tetraspanin concentrations divided by albumin concentration. CSF, cerebrospinal fluid; EV, extracellular vesicle; SEC, size exclusion chromatography.

Figure 5—source data 1

Top SEC methods in new batches of plasma and CSF.

https://cdn.elifesciences.org/articles/70725/elife-70725-fig5-data1-v2.xlsx

Discussion

In this study, we describe a framework for rapidly quantifying relative EV yield and purity across isolation methods, overcoming the limitations of other commonly used methods used for EV analysis. Several techniques, such as NTA and other methods developed for analysis of synthetic particles, have been applied to EV detection (Hartjes et al., 2019). The utility of these techniques is hindered, however, by an inability to differentiate heterogeneous EVs from other particles with overlapping size, such as lipoproteins or aggregated protein particles. Thus, although previous reports comparing EV isolation methods Lobb et al., 2015; Helwa et al., 2017; Baranyai et al., 2015; Soares Martins et al., 2018; An et al., 2018; Stranska et al., 2018; Diaz et al., 2018; Kalra et al., 2013; Serrano-Pertierra et al., 2019; Gámez-Valero et al., 2016; Takov et al., 2019; Brennan et al., 2020 have yielded some useful insights, the lack of reliable EV quantification has made these studies difficult to interpret (Tkach et al., 2018; Coumans et al., 2017a; Ludwig et al., 2019).

The measurement of EV transmembrane proteins overcomes the limitations of EV quantification with particle detection methods. Since the transmembrane proteins CD9, CD63, and CD81 are present on EVs but are not present in lipoproteins or free protein aggregates, these tetraspanins can be used for relative quantification of EVs. Although not every EV necessarily contains a tetraspanin protein, by detecting three different tetraspanins per sample with Simoa, we minimize the chance that we are measuring a rare subset of EVs. In the experiments reported here, we observed a strong correlation of the relative levels of the three tetraspanins in different SEC fractions. Since we compared isolation methods from the same starting sample, we were able to provide a direct quantitative comparison of tetraspanin levels between the different isolation methods.

We used Simoa in this study, which is particularly well suited for EV analysis due to the technology’s high dynamic range, throughput, and sensitivity. This sensitivity is achieved by converting ELISA to a digital readout via immuno-capture and counting of individual protein molecules in a microwell array. We used the commercially available Quanterix HD-X instrument, but our lab has also developed other digital ELISA methods using commonly available instrumentation (Cohen et al., 2020; Maley et al., 2020; Wu et al., 2020), which could be similarly applied to EVs. One could also follow a similar approach to the one we present here with traditional ELISA or other protein detection methods, but we find that high sensitivity is often necessary for the low levels of EVs in human biofluids. We have previously shown in a direct comparison (using the same antibodies) that Simoa can detect EV markers in cases where traditional ELISA cannot, such as SEC fractions of CSF (Norman et al., 2021).

We used Simoa to directly compare the yield and purity of commonly used EV isolation methods. To obtain the purest EVs possible (and separate EVs from lipoproteins), it has been demonstrated that it is necessary to combine several techniques sequentially, such as density gradient centrifugation (DGC) and SEC (Karimi et al., 2018; Zhang et al., 2020). However, techniques such as DGC are not scalable to many samples and therefore not amenable to biomarker studies. Thus, we focused on EV isolation methods that are amenable to biomarker studies. After finding that commercial SEC columns compare favorably to ultracentrifugation and ExoQuick precipitation, we compared several resins and column volumes to further improve EV isolation by custom SEC columns. SEC requires no instrumentation and allows one to run many columns in parallel. The throughput can be further increased by using SEC stands (such as the one we designed) and preparing columns ahead of time. Although we used freshly prepared columns in this study, we found comparable performance to columns stored at 4°C for 1 week.

Our investigation of SEC parameters led us to improved methods for EV isolation; in particular, we found that Sepharose CL-6B, which is seldom used for EV isolation, yields considerably higher levels of EVs than either Sepharose CL-2B, the most commonly used resin (Monguió-Tortajada et al., 2019), or Sepharose CL-4B. We attribute this result to Sepharose CL-6B beads having a smaller average pore size (reported [Hagel et al., 1996] to be 24 nm vs. 42 nm for CL-4B and 75 nm for CL-2B), leading to a lower probability that EVs will enter the beads. As there is a tradeoff between EV yield and albumin contamination, we envision different SEC columns will be suited for different applications. Using a 10-ml Sepharose CL-6B column for EV isolation from plasma or CSF is the best choice for downstream applications where maximum EV yield is needed and where some free protein contamination is not detrimental—for example, analyzing rare EV cargo or when further purification of EVs will be performed (such as immuno-isolation). On the other hand, if isolating EVs from plasma where minimal free protein contamination is desired (e.g., in EV protein analysis by Western blot), a larger 20 ml column with Sepharose CL-4B would yield better results. For CSF, which has much less protein than plasma, 10 ml columns are preferable to 20 ml ones.

By developing a Simoa assay to measure albumin (the most abundant protein in plasma and main contaminant when isolating EVs), we were able to assess the purity of EV preparations with respect to unwanted co-purification of free proteins. Our methods could be expanded to assess other contaminants that are less abundant than albumin but may, nonetheless, be problematic for some applications, such as lipoproteins. Adding a Simoa assay for ApoB100 (or other protein components of lipoproteins) would allow for the assessment of both lipoprotein and free protein contamination in EV isolation methods. Although lipoproteins are difficult to separate from EVs due to their overlapping size profile (Simonsen, 2017), a recent study demonstrated that a chromatography column combining a cation-exchange resin layer with an SEC resin layer allows for efficient lipoprotein depletion using ‘dual mode chromatography’ (Van Deun et al., 2020). Simoa could be used to evaluate and help improve such techniques in the future.

The general experimental framework presented here could be easily applied to evaluate new EV isolation methods in plasma, CSF, or other biological fluids, such as urine or saliva. While we limited our study to human biofluids, similar methods could also be applied to compare EV isolation methods from cell culture media. As sensitivity of EV detection and specificity in differentiating EVs from contaminants are obstacles in all EV studies, we envision that ultrasensitive protein detection with Simoa will be broadly applicable to assessing EV isolation methods for both the study of EV biology and development of EV diagnostics.

Methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Biological sample (human)PlasmaBioIVTCat #HUMANPLK2PNNPooled gender, K2EDTA
Biological sample (human)Cerebrospinal fluidBioIVTCat# HMNCSFR-NODXRPooled gender, no diagnosis remnant
AntibodyAnti-CD9 (Mouse monoclonal)MilliporeSigmaCat# CBL162RRID:AB_2075914WB (1:1000)
AntibodyAnti-CD9 (Rabbit monoclonal)AbcamCat# ab195422RRID:AB_2893477Simoa capture
AntibodyAnti-CD9 (Mouse monoclonal)AbcamCat# ab58989RRID:AB_940926Simoa detector
AntibodyAnti-CD63 (Mouse monoclonal)BDCat# 556019RRID:AB_396297Simoa detector;WB (1:1000)
AntibodyAnti-CD63 (Mouse monoclonal)R&D SystemsCat# MAB5048RRID:AB_2275726Simoa capture
AntibodyAnti-CD81 (Mouse monoclonal)Thermo Fisher ScientificCat# 10630DRRID:AB_2532984WB (1:666)
AntibodyAnti-CD81 (Mouse monoclonal)AbcamCat# ab79559RRID:AB_1603682Simoa capture
AntibodyAnti-CD81 (Mouse monoclonal)BioLegendCat# 349502RRID:AB_10643417Simoa detector
Commercial assay or kitHuman Serum Albumin DuoSet ELISAR&D SystemsCat# DY1455Simoa capture and detector
Peptide, recombinant proteinCD9AbcamCat# ab152262
Peptide, recombinant proteinCD63OrigeneCat# TP301733
Peptide, recombinant proteinCD81OrigeneCat# TP317508
Peptide, recombinant proteinAlbuminAbcamCat# ab201876
Commercial assay or kitExoQuick exosome precipitation solutionSBICat# EXOQ5A-1
Commercial assay or kitExoQuick ULTRA EV isolation kit for plasma and serumSBICat# EQULTRA-20A-1
Commercial assay or kitqEVoriginal 70 nmIzonCat# SP1
Commercial assay or kitqEVoriginal 35 nmIzonCat# SP5
OtherSepharose CL-2BCytivaCat# 17014001
OtherSepharose CL-4BCytivaCat# 17015001
OtherSepharose CL-6BCytivaCat# 17016001

Human sample handling

Pre-aliquoted pooled human plasma (collected in K2 EDTA tubes) and CSF samples were ordered from BioIVT. The same pools were used for all main figures throughout the paper in order to ensure comparable analysis of methods. For all EV isolation technique comparisons, one 0.5-ml sample was used for each isolation method. Plasma or CSF was thawed at room temperature. After sample thawing, 100× Protease/Phosphatase Inhibitor Cocktail (Cell Signaling Technology) was added to 1×. The sample was then centrifuged at 2000×g for 10 min. The supernatant was subsequently centrifuged through a 0.45 μm Corning Costar SPIN-X centrifuge tube filter (Sigma-Aldrich) at 2000×g for 10 min to get rid of any remaining cells or cell debris.

Simoa assays

Simoa assays were developed and performed as previously described (Norman et al., 2021). A detailed protocol is available: https://www.protocols.io/view/simoa-extracellular-vesicle-assays-bm89k9z6. Capture antibodies were coupled to Carboxylated Paramagnetic Beads from the Simoa Homebrew Assay Development Kit (Quanterix) using EDC chemistry (Thermo Fisher Scientific). Detection antibodies were conjugated to biotin using EZ-Link NHS-PEG4 Biotin (Thermo Fisher Scientific). For CD9, ab195422 (Abcam) was used as capture antibody and ab58989 (Abcam) was used as detector antibody. For CD63, MAB5048 (R&D Systems) was used as capture antibody and 556019 (BD) was used as detector antibody. For CD81, ab79559 (Abcam) was used as capture antibody and 349502 (BioLegend) was used as detector antibody. For albumin, DY1455 (R&D Systems) was used as both capture and detector antibody. The following recombinant proteins were used for CD9, CD63, CD81, and albumin: ab152262 (Abcam), TP301733 (Origene), TP317508 (Origene), and ab201876 (Abcam). On-board dilution was performed with 4× dilution for each of the tetraspanins, while manual 20× dilution was used for albumin. All samples were raised to 160 μl per replicate in sample diluent. For tetraspanin assays, samples were incubated with immunocapture beads (25 μl) and biotinylated detection antibody (20 μl) for 35 min. Next, six washes were performed, and the beads were resuspended in 100 μl of Streptavidin labeled β-galactosidase (Quanterix) and incubated for 5 min. All bead washes were performed with Wash Buffer 1 (Quanterix). After incubation, an additional six washes were performed, and the beads were resuspended in 25 μl Resorufin β-D-galactopyranoside (Quanterix) before being loaded into the microwell array on the Quanterix HD-X instrument. For the albumin assay, samples were incubated first with immunocapture beads (25 μl) for 15 min and then washed six times. Subsequently, 100 μl detection antibody was incubated with the beads for 5 min. Next, six washes were performed, and the beads were resuspended in 100 μl of Streptavidin labeled β-galactosidase (Quanterix) for a final 5-min incubation. After an additional six washes, the beads were resuspended in 25 μl Resorufin β-D-galactopyranoside (Quanterix) and then loaded into the microwell array on the Quanterix HD-X instrument.

Construction of SEC stand

The custom SEC rack was constructed from a total of 22 pieces using CNC milling tools. The rack is made of an aluminum frame (silver, Multipurpose 6061 Aluminum, McMaster-Carr) consisting of eight pieces, 4 sliding plates made from acetal (black, Wear-Resistant Easy-to-Machine Delrin Acetal Resin, McMaster-Carr), and 10 sliding plate grips made from UHMW Polyethylene (white, Slippery UHMW Polyethylene, McMaster-Carr). The rack frame is held together using 20 ¾″ screws (McMaster-Carr, 92210 A113), 20 ½″ screws (McMaster-Carr, 92210 A110), 10 0.375″ Dowel pins (McMaster-Carr, 90145 A470), and 10 0.5625″ Dowel pins (McMaster-Carr, 90145 A483), and includes 20 spring plungers (McMaster-Carr, 84895 A710) that allow the sliding plates to ‘click’ once aligned with the chromatography columns. Details for constructing the rack and SolidWorks files are included in the Supplementary materials.

Preparation of custom SEC columns

The resins Sepharose CL-2B, Sepharose CL-4B, and Sepharose CL-6B (all from GE Healthcare/Cytiva) were washed in phosphate-buffered saline (PBS). The volume of resin was washed with an equal volume of PBS in a glass container and then placed at 4°C in order to let the resin settle completely (several hours or overnight). The PBS was then poured off, and an equal volume of PBS was again added two more times for a total of three washes. Columns were prepared fresh on the day of use. Washed resin was poured into an Econo-Pac Chromatography column (Bio-Rad) to bring the bed volume (the resin without liquid) to 10 or 20 ml. When the desired amount of resin filled the column and the liquid dripped through, the top frit was immediately placed at the top of the resin without compressing the resin. PBS was then added again before sample addition.

Collection of size exclusion chromatography fractions

Once prepared, all columns were washed with at least 20 ml of PBS in the column. Immediately before sample addition, the column was allowed to fully drip out and, after last drop of PBS, sample (filtered plasma or CSF) was added to the column. As soon as sample was added, 0.5 ml fractions were collected in individual tubes. As soon as the plasma or CSF completely entered the column (below the frit), PBS was added to the top of column 1 ml at a time. Fraction numbers correspond to 0.5 ml increments collected as soon as sample is added. For Izon and 10 ml columns, fractions 6–21 were collected (since first few fractions correspond to void volume). For 20 ml columns, fractions 12–27 were collected (since void volume is larger for 20 ml columns than 10 ml columns). For Figure 5, only fractions 7–10 were collected.

Ultracentrifugation

Samples of filtered 0.5 m plasma or CSF were added to 3.5 ml Open-Top Thickwall Polycarbonate ultracentrifuge tubes (Beckman Coulter), and PBS was added to fill tubes to the top. Samples were ultracentrifuged at 120,000×g for 90 min at 4°C in an Optima XPN-80 ultracentrifuge (Beckman Coulter) using an SW55 Ti swinging-bucket rotor (Beckman Coulter). Afterward, all supernatant was aspirated. Pellets were resuspended in PBS for the ‘Ultracentrifugation’ condition. For the ‘Ultracentrifugation with wash’ condition, the ultracentrifuge tubes were filled to the top with PBS, and samples were ultracentrifuged again at 120,000×g for 90 min. Supernatant was then aspirated, and pellets were resuspended in 500 µl PBS. For all ultracentrifugation samples, isolation was performed on 2 separate days and then resulting Simoa values were averaged.

ExoQuick and ExoQuick ULTRA

Samples of plasma or CSF were mixed with ExoQuick Exosome Precipitation Solution (System Biosciences) or ExoQuick ULTRA EV Isolation Kit for Serum and Plasma (System Biosciences), and protocols were performed according to the manufacturer’s instructions. For ExoQuick, 0.5 ml of plasma or CSF was mixed with 126 µl of ExoQuick and incubated at 4°C for 30 min, followed by centrifugation at 1500×g for 30 min. Supernatant was removed, and samples were centrifuged at 1500×g for an additional 5 min. Residual supernatant was removed, and pellets were resuspended in 500 µl PBS. For ExoQuick ULTRA, 250 µl of plasma or CSF was used in accordance with instructions, and Simoa values were corrected by multiplying by 2 to match the 0.5 ml volume used for other samples. For each sample, 500 µl of EVs was eluted per column. For all precipitations, isolation was performed on 2 separate days and then resulting Simoa values were averaged.

Western blotting

Western blotting for tetraspanins was performed as previously described (Kowal et al., 2017), with minor modifications. 4× LDS was added to samples and samples were heated at 70°C for 10 min. Samples were run at 150 V for 1 hr on 4–12% Bolt Bis-Tris protein gels (Thermo Fisher Scientific) and transferred using iBlot two nitrocellulose mini transfer stack (Thermo Fisher Scientific) at 20 V for 3 min. Blocking buffer was made by dissolving milk powder (to 5% w/v) in PBS-T (PBS with 0.1% Tween). Nitrocellulose membranes were blocked on a shaker for 30 min at 4°C, and then incubated with primary antibody overnight. The following antibodies and dilutions were used: 1:1000 BD (H5C6) for CD63, 1:1000 Millipore Clone MM2/57 (CBL162) for CD9 and 1:666 Thermo Fisher Scientific (M38) for CD81. Membranes were washed three times with PBS-T and incubated with 1:2000 human cross-adsorbed, anti-mouse HRP conjugated secondary antibody (Rockland) in blocking buffer for 2 hr. Membranes were washed three times with PBS-T and WesternBright ECL-spray HRP substrate (Advansta) was added. Images were acquired with a Sapphire Biomolecular Analyzer (Azure Biosystems).

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.

References

    1. Théry C
    2. Witwer KW
    3. Aikawa E
    4. Alcaraz MJ
    5. Anderson JD
    6. Andriantsitohaina R
    7. Antoniou A
    8. Arab T
    9. Archer F
    10. Atkin-Smith GK
    11. Ayre DC
    12. Bach JM
    13. Bachurski D
    14. Baharvand H
    15. Balaj L
    16. Baldacchino S
    17. Bauer NN
    18. Baxter AA
    19. Bebawy M
    20. Beckham C
    21. Bedina Zavec A
    22. Benmoussa A
    23. Berardi AC
    24. Bergese P
    25. Bielska E
    26. Blenkiron C
    27. Bobis-Wozowicz S
    28. Boilard E
    29. Boireau W
    30. Bongiovanni A
    31. Borràs FE
    32. Bosch S
    33. Boulanger CM
    34. Breakefield X
    35. Breglio AM
    36. Brennan MÁ
    37. Brigstock DR
    38. Brisson A
    39. Broekman ML
    40. Bromberg JF
    41. Bryl-Górecka P
    42. Buch S
    43. Buck AH
    44. Burger D
    45. Busatto S
    46. Buschmann D
    47. Bussolati B
    48. Buzás EI
    49. Byrd JB
    50. Camussi G
    51. Carter DR
    52. Caruso S
    53. Chamley LW
    54. Chang YT
    55. Chen C
    56. Chen S
    57. Cheng L
    58. Chin AR
    59. Clayton A
    60. Clerici SP
    61. Cocks A
    62. Cocucci E
    63. Coffey RJ
    64. Cordeiro-da-Silva A
    65. Couch Y
    66. Coumans FA
    67. Coyle B
    68. Crescitelli R
    69. Criado MF
    70. D’Souza-Schorey C
    71. Das S
    72. Datta Chaudhuri A
    73. de Candia P
    74. De Santana EF
    75. De Wever O
    76. Del Portillo HA
    77. Demaret T
    78. Deville S
    79. Devitt A
    80. Dhondt B
    81. Di Vizio D
    82. Dieterich LC
    83. Dolo V
    84. Dominguez Rubio AP
    85. Dominici M
    86. Dourado MR
    87. Driedonks TA
    88. Duarte FV
    89. Duncan HM
    90. Eichenberger RM
    91. Ekström K
    92. El Andaloussi S
    93. Elie-Caille C
    94. Erdbrügger U
    95. Falcón-Pérez JM
    96. Fatima F
    97. Fish JE
    98. Flores-Bellver M
    99. Försönits A
    100. Frelet-Barrand A
    101. Fricke F
    102. Fuhrmann G
    103. Gabrielsson S
    104. Gámez-Valero A
    105. Gardiner C
    106. Gärtner K
    107. Gaudin R
    108. Gho YS
    109. Giebel B
    110. Gilbert C
    111. Gimona M
    112. Giusti I
    113. Goberdhan DC
    114. Görgens A
    115. Gorski SM
    116. Greening DW
    117. Gross JC
    118. Gualerzi A
    119. Gupta GN
    120. Gustafson D
    121. Handberg A
    122. Haraszti RA
    123. Harrison P
    124. Hegyesi H
    125. Hendrix A
    126. Hill AF
    127. Hochberg FH
    128. Hoffmann KF
    129. Holder B
    130. Holthofer H
    131. Hosseinkhani B
    132. Hu G
    133. Huang Y
    134. Huber V
    135. Hunt S
    136. Ibrahim AGE
    137. Ikezu T
    138. Inal JM
    139. Isin M
    140. Ivanova A
    141. Jackson HK
    142. Jacobsen S
    143. Jay SM
    144. Jayachandran M
    145. Jenster G
    146. Jiang L
    147. Johnson SM
    148. Jones JC
    149. Jong A
    150. Jovanovic-Talisman T
    151. Jung S
    152. Kalluri R
    153. Kano SI
    154. Kaur S
    155. Kawamura Y
    156. Keller ET
    157. Khamari D
    158. Khomyakova E
    159. Khvorova A
    160. Kierulf P
    161. Kim KP
    162. Kislinger T
    163. Klingeborn M
    164. Klinke DJ
    165. Kornek M
    166. Kosanović MM
    167. Kovács ÁF
    168. Krämer-Albers EM
    169. Krasemann S
    170. Krause M
    171. Kurochkin IV
    172. Kusuma GD
    173. Kuypers S
    174. Laitinen S
    175. Langevin SM
    176. Languino LR
    177. Lannigan J
    178. Lässer C
    179. Laurent LC
    180. Lavieu G
    181. Lázaro-Ibáñez E
    182. Le Lay S
    183. Lee MS
    184. Lee YXF
    185. Lemos DS
    186. Lenassi M
    187. Leszczynska A
    188. Li IT
    189. Liao K
    190. Libregts SF
    191. Ligeti E
    192. Lim R
    193. Lim SK
    194. Linē A
    195. Linnemannstöns K
    196. Llorente A
    197. Lombard CA
    198. Lorenowicz MJ
    199. Lörincz ÁM
    200. Lötvall J
    201. Lovett J
    202. Lowry MC
    203. Loyer X
    204. Lu Q
    205. Lukomska B
    206. Lunavat TR
    207. Maas SL
    208. Malhi H
    209. Marcilla A
    210. Mariani J
    211. Mariscal J
    212. Martens-Uzunova ES
    213. Martin-Jaular L
    214. Martinez MC
    215. Martins VR
    216. Mathieu M
    217. Mathivanan S
    218. Maugeri M
    219. McGinnis LK
    220. McVey MJ
    221. Meckes DG
    222. Meehan KL
    223. Mertens I
    224. Minciacchi VR
    225. Möller A
    226. Møller Jørgensen M
    227. Morales-Kastresana A
    228. Morhayim J
    229. Mullier F
    230. Muraca M
    231. Musante L
    232. Mussack V
    233. Muth DC
    234. Myburgh KH
    235. Najrana T
    236. Nawaz M
    237. Nazarenko I
    238. Nejsum P
    239. Neri C
    240. Neri T
    241. Nieuwland R
    242. Nimrichter L
    243. Nolan JP
    244. Nolte-’t Hoen EN
    245. Noren Hooten N
    246. O’ L
    247. O’ T
    248. Ochiya T
    249. Olivier M
    250. Ortiz A
    251. Ortiz LA
    252. Osteikoetxea X
    253. Østergaard O
    254. Ostrowski M
    255. Park J
    256. Pegtel DM
    257. Peinado H
    258. Perut F
    259. Pfaffl MW
    260. Phinney DG
    261. Pieters BC
    262. Pink RC
    263. Pisetsky DS
    264. Pogge von Strandmann E
    265. Polakovicova I
    266. Poon IK
    267. Powell BH
    268. Prada I
    269. Pulliam L
    270. Quesenberry P
    271. Radeghieri A
    272. Raffai RL
    273. Raimondo S
    274. Rak J
    275. Ramirez MI
    276. Raposo G
    277. Rayyan MS
    278. Regev-Rudzki N
    279. Ricklefs FL
    280. Robbins PD
    281. Roberts DD
    282. Rodrigues SC
    283. Rohde E
    284. Rome S
    285. Rouschop KM
    286. Rughetti A
    287. Russell AE
    288. Saá P
    289. Sahoo S
    290. Salas-Huenuleo E
    291. Sánchez C
    292. Saugstad JA
    293. Saul MJ
    294. Schiffelers RM
    295. Schneider R
    296. Schøyen TH
    297. Scott A
    298. Shahaj E
    299. Sharma S
    300. Shatnyeva O
    301. Shekari F
    302. Shelke GV
    303. Shetty AK
    304. Shiba K
    305. Siljander PRM
    306. Silva AM
    307. Skowronek A
    308. Snyder OL
    309. Soares RP
    310. Sódar BW
    311. Soekmadji C
    312. Sotillo J
    313. Stahl PD
    314. Stoorvogel W
    315. Stott SL
    316. Strasser EF
    317. Swift S
    318. Tahara H
    319. Tewari M
    320. Timms K
    321. Tiwari S
    322. Tixeira R
    323. Tkach M
    324. Toh WS
    325. Tomasini R
    326. Torrecilhas AC
    327. Tosar JP
    328. Toxavidis V
    329. Urbanelli L
    330. Vader P
    331. van Balkom BW
    332. van der Grein SG
    333. Van Deun J
    334. van Herwijnen MJ
    335. Van Keuren-Jensen K
    336. van Niel G
    337. van Royen ME
    338. van Wijnen AJ
    339. Vasconcelos MH
    340. Vechetti IJ
    341. Veit TD
    342. Vella LJ
    343. Velot É
    344. Verweij FJ
    345. Vestad B
    346. Viñas JL
    347. Visnovitz T
    348. Vukman KV
    349. Wahlgren J
    350. Watson DC
    351. Wauben MH
    352. Weaver A
    353. Webber JP
    354. Weber V
    355. Wehman AM
    356. Weiss DJ
    357. Welsh JA
    358. Wendt S
    359. Wheelock AM
    360. Wiener Z
    361. Witte L
    362. Wolfram J
    363. Xagorari A
    364. Xander P
    365. Xu J
    366. Yan X
    367. Yáñez-Mó M
    368. Yin H
    369. Yuana Y
    370. Zappulli V
    371. Zarubova J
    372. Žėkas V
    373. Zhang JY
    374. Zhao Z
    375. Zheng L
    376. Zheutlin AR
    377. Zickler AM
    378. Zimmermann P
    379. Zivkovic AM
    380. Zocco D
    381. Zuba-Surma EK
    (2018) Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for Extracellular Vesicles and update of the MISEV2014 guidelines
    Journal of Extracellular Vesicles 7:1535750.
    https://doi.org/10.1080/20013078.2018.1535750
    1. Tkach M
    2. Kowal J
    3. Théry C
    (2018) Why the need and how to approach the functional diversity of extracellular vesicles
    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 373:20160479.
    https://doi.org/10.1098/rstb.2016.0479

Decision letter

  1. YM Dennis Lo
    Senior and Reviewing Editor; The Chinese University of Hong Kong, Hong Kong
  2. Qing Zhou
    Reviewer; Chinese University of Hong Kong, Hong Kong
  3. Julie A Saugstad
    Reviewer

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Framework for Rapid Comparison of Extracellular Vesicle Isolation Methods" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Y M Dennis Lo as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Qing Zhou (Reviewer #1); Julie A Saugstad (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The most novel aspect of this work is the strategy for quantifying EV yield and purity after EV isolation. To allow the authors to claim the potential superiority of their method, parallel analysis with a number of other currently available techniques (e.g. NTA) are necessary. For example, it would be useful to test if NTA can detect EV yield change between different isolation methods, or whether western blotting could show the same protein level changes reported by Simoa. There is also insufficient data to conclude that 10 mL Sepharose CL-6B column is the best choice for EV isolation from plasma and CSF.

2) A number of sub-conclusions are not solid enough. For example, whether different centrifugation speeds and time may affect the claim that SEC outperforms ultracentrifugation with regard to EV yield. It is well known that longer and higher speed centrifugation will increase the yield of EVs. Moreover, whether 10 mL Sepharose CL-6B column would still be the best choice for EV isolation from plasma or CSF when extending contamination markers to lipoprotein markers. As the size of lipoproteins is much larger, the purity of 10 mL Sepharose CL-6B column may be more affected than 10 mL Sepharose CL-2B columns.

3) Assessment of contamination could have been bolstered by examination of one or more proteins, e.g. lipoproteins.

4) The rationale for choosing some of the methods for comparison are missing – why test with and without the wash step for UC? Why use exoquick? The error bars are missing from most figures, the reproducibility of a method should be an important factor for consideration as a good isolation method.

5) Figure 3 should be reformatted as a supplement.

6) Frozen-thawed samples, which are commonly avoided in studies on EV, were used in this work. Why were such samples used, and would such samples produce bias to the authors' data?

7) The pre-analytical collection of the plasma and CSF should be described more fully in the Materials and methods and a couple of sentences added to the results. The pre-analytical steps are important for understanding the downstream results. While we note that the authors had purchased the samples from a company, we would like to know more details on the sample collection process. Information that would be useful includes: i) were EDTA collections tubes used? If not, state what type of tubes (Streck, etc) ii) was the plasma spun once to separate from buffy coat, or was it spun twice to also reduce platelets? iii) were hemoglobin/hemolysis measurements taken for the pools?

8) Figure 1C should have a graphic of each method tested. The graphic for SEC is clear and easy to understand. There should also be a short graphic for the ultracentrifugation with the speed of spins and the time laid out in order. There should also be a graphic for the exoquick – though that one is a little less exciting as a picture, but should describe incubation times etc.

9) More details should be included in the results describing the assays chosen for isolation – qEV 35 vs 70 and ultraC with or without wash and the two exoquicks. Explanations should be included why each of these was examined and what potential differences they represented – for example between 35 and 70, etc.

10) In Figure 2, the emphasis should be on recovery, not contamination. Consider reordering the results at the bottom based on relative EV recovery rather than albumin concentration. So in this case, d and e and i and j would switch places and the order would go from left to right on EV recovery. It would be useful to add error bars for this figure. An important component of comparing these assays concerns their reproducibility.

11) In the results – the authors claim that SEC is superior, but exoquick performs well in CSF. The authors should provide a set of characteristics which would objectively demonstrate that SEC is indeed superior to exoquick. Why would Izon 35 or 70 work differently in CSF compared with plasma? What might be some of the contributing factors influencing the effects of pore sizes that might aid in choosing one of these – or for understanding the albumin contamination in CSF.

12) The construction of the SEC collection apparatus is more appropriately placed in the supplementary materials. Izon sells and automated fraction collector, a similar stand, though not as sophisticated. Indeed, almost every lab that has done SEC often fashions its own collection apparatus. Thus, while the apparatus is interesting and well-constructed, its details can be regarded as supplemental.

13) A more detailed description of the pore sizes for the Sepharose chosen is needed. The testing of the columns for different size Sepharose beads and different heights is informative. Figures 4 and 5, one would like to more directly compare the 10 and 20 mL columns and would prefer they were A and B while Izon was C. And if the authors could make a measure of relative yield for the EV recovery, as they did in Figure 2 – that would be a nice additional column.

14) It would be useful to readers for the authors to summarize the outcomes of the plasma and CSF studies presented herein as "Recommendations" or as a summary table indicating the best method for each.

15) In Methods, Simoa Assays, first paragraph: It would be helpful to the reader to include the name of the Tetraspanins and detector antibodies before the catalog # and company name.

16) In Methods, Simoa Assays, second paragraph: In the sentence "Next, six washes were performed" it would be helpful to readers to say what buffer was used for the washes, and if these were consistently used for all the following washes.

17) In Methods, Preparation of Custom SEC Columns: "Columns were prepared fresh on the day of use". Can the authors comment in the Discussion whether the columns need to be prepared fresh on the day of use, or if they can be stored in the refrigerator, and if so, for how many days before use?

18) In Images, Figures 4 and 5, and Supplementary Figure 1: Please include higher resolution images. In Images, Supplementary Figure 2 is difficult to interpret as presented. Suggest grouping Relative albumin concentration by column and fraction number (6b, fx 7-9 and 7-10; Izon 35, fx 7-9 and 7-10), or some way other way to visualize effects of each method vs. the albumin concentration.

19) The authors should upload the Raw data from the Simoa runs as a supplementary file. Also, please include the Group Allocation information in the manuscript.

Reviewer #1 (Recommendations for the authors):

1) Add comparisons between this method with other commonly used techniques such as NTA to test the consistency and different performance.

2) Other contamination markers need to be included to strengthen the outperformance of this method, such as lipoprotein markers.

3) Re-evaluate the purity of different isolation methods with different contamination markers. May add one sample with a quicker and longer ultracentrifuge protocol to test the potential influence.

4) As noticed, frozen-thaw samples are used in this manuscript, usually avoided in EV studies. Can authors explain why not using fresh materials?

Reviewer #2 (Recommendations for the authors):

Introduction

The use of CD9, CD63, CD81 is appropriate for characterization of general EVs. Some data should be generated or discussed regarding abundance on different cell types or cell lines. It may be that these tetraspanins are expressed with different amounts in different contexts or cell types – or even on EVs released through different biogenesis pathways – in addition to not being present on every EV. The use of sensitive quantitative assays is very useful for understanding the purity and contamination with each method, though there are many papers out there already that do this with different assays and assess more EV components and contaminants.

Results

1) The pre-analytical collection of the plasma and CSF should be described more fully in the Materials and methods and a couple of sentences added to the results. The pre-analytical steps are important for understanding the downstream results. While you purchased the samples from a company – you can still report on the methods of collection. Information that should be included: 1) were EDTA collections tubes used? If not state what type of tubes (Streck, etc) 2) was the plasma spun once to separate from buffy coat, or was it spun twice to also reduce platelets? 3) were hemoglobin/hemolysis measurements taken for the pools? If aliquots remain in the freezer – you can add this information to the methods section on the samples.

2) Figure 1C should have a graphic of each method tested. The graphic for SEC is clear and easy to understand. There should also be a short graphic for the ultracentrifugation with the speed of spins and the time laid out in order. There should also be a graphic for the exoquick – though that one is a little less exciting as a picture, but should describe incubation times etc.

3) Spend a bit more time in the results describing the assays chosen for isolation – qEV 35 vs 70 and ultraC with or without wash and the two exoquicks. Provide some details on why each of these was examined and what potential differences they represent – for example between 35 and 70, etc.

4) In Figure 2, the emphasis should be on recovery, not contamination. Consider reordering the results at the bottom based on relative EV recovery rather than albumin concentration. So in this case, d and e and I and J would switch places and the order would go from left to right on EV recovery

5) Are there error bars for this? Can they be added? An important component of comparing these assays is also their reproducibility.

6) In the results – you claim SEC is superior, but exoquick performs well in CSF. You should provide a set of characteristics declaring it to be superior and why exoquick gets excluded. Why would Izon 35 or 70 work differently in CSF compared with plasma – what do you think are the contributing factors about the pore sizes that might aide in choosing one of these – or for understanding the albumin contamination in CSF. You do begin to answer this in the next section, but a more significant discussion is warranted.

7) I admire the construction of the SEC collection apparatus, but this should be supplementary material and should not be a main figure or part of the results. Izon sells and automated fraction collector, a similar stand, though not as sophisticated. And almost every lab that has done SEC very frequently has fashioned their own collection apparatus. So while this is interesting and well-constructed, it is supplemental.

8) More fully describe the pore sizes for the Sepharose chosen. The testing of the columns for different size Sepharose beads and different heights is informative. Figure 4 and 5, I would like to more directly compare the 10 and 20 mL columns and would prefer they were A and B while Izon was C. And if you could make a measure of relative yield for the EV recovery, as you did in Figure 2 – that would be a nice additional column.

9) With Figure 3 moved to supplemental, you could have the two supplemental figures become main figures.

10) Measuring a lipoprotein in addition to albumin would have been very useful to see.

Discussion

1) Why did you choose to look at plasma and CSF?

2) A comparison of Simoa with any other conventional assay would have been useful – like using exoView and each of the tetraspanins just to show similar counts for plasma one time.

Reviewer #3 (Recommendations for the authors):

The authors discuss comparing ultracentrifugation (with or without wash step) and two commercially available kits (ExoQuick and ExoQuick ULTRA), but do not include any data for these experiments. Although they state that this is because SEC was superior, it would be useful to include the outcomes from these studies at least in Supplementary data.

It would be useful to readers to summarize the outcomes of the plasma and CSF studies presented herein as "Recommendations" or as a summary table indicating the best method for each.

In Methods, Simoa Assays, first paragraph: It would be helpful to the reader to include the name of the Tetraspanins and detector antibodies before the catalog # and company name.

In Methods, Simoa Assays, second paragraph: In the sentence "Next, six washes were performed" it would be helpful to readers to say what buffer was used for the washes, and if these were consistently used for all the following washes.

In Methods, Preparation of Custom SEC Columns: "Columns were prepared fresh on the day of use" Can you comment in the discussion whether the columns need to be prepared fresh on the day of use, or if they can be stored in the refrigerator, and if so, for how many days before use?

In Images, Figures 4 and 5, and Supplementary Figure 1: Please include higher resolution images

In Images, Supplementary Figure 2 is difficult to interpret as presented. Suggest grouping Relative albumin concentration by column and fraction number (6b, fx 7-9 and 7-10; Izon 35, fx 7-9 and 7-10), or some way other way to visualize effects of each method vs. the albumin concentration.

I encourage the authors to upload the Raw data from the Simoa runs as a supplementary file. Also, please include the Group Allocation information in the manuscript.

https://doi.org/10.7554/eLife.70725.sa1

Author response

Essential revisions:

1) The most novel aspect of this work is the strategy for quantifying EV yield and purity after EV isolation. To allow the authors to claim the potential superiority of their method, parallel analysis with a number of other currently available techniques (e.g. NTA) are necessary. For example, it would be useful to test if NTA can detect EV yield change between different isolation methods, or whether western blotting could show the same protein level changes reported by Simoa. There is also insufficient data to conclude that 10 mL Sepharose CL-6B column is the best choice for EV isolation from plasma and CSF.

Thank you to the reviewers for these comments. We certainly agree that different methods for comparing EV isolation methods would yield different results. Since there is no perfect technique for measuring EVs, there is also no true positive control to compare to. In particular, NTA has been widely recognized to be problematic for quantifying EVs due to its inability to discriminate between EVs and the similarly sized but more abundant non-EV particles such as lipoproteins or protein aggregates (1-3). For example, Welton et al. demonstrated that there is poor overlap between particles measured by NTA and tetraspanin levels after fractionating plasma by SEC, leading the authors to conclude that >70% of particles measured by NTA in plasma are not EVs (4). This was our motivation for using Simoa to measure tetraspanin levels, which overcomes this limitation of NTA and other particle counting methods. Of course, measuring specific proteins, which are not present on every single EV is a limitation of our approach, but since we are comparing EV isolation methods from the same starting sample of biofluid, Simoa enables an “apples to apples” comparison that is quantitative in a way that other techniques (such as NTA) are not. We have previously compared Simoa to ELISA and western blot (in Supplemental Figure 2) from the same sample (5). To show concordance between Simoa and western blot, we repeated the comparison of 10mL SEC columns with Sepharose CL-2B, CL-4B, and CL-6B and evaluated the EV fractions (F7-10) by western blotting for CD9, CD63, and CD81. Our western blot results confirm that using CL-6B leads to the highest levels of tetraspanins and that SEC resins with smaller pore sizes lead to more EVs. We have added a supplemental figure with these results. Whether or not Sepharose CL-6B is the best choice, as we state in the paper, depends on the goal. Using this resin yields more EVs, but also more albumin contamination.

2) A number of sub-conclusions are not solid enough. For example, whether different centrifugation speeds and time may affect the claim that SEC outperforms ultracentrifugation with regard to EV yield. It is well known that longer and higher speed centrifugation will increase the yield of EVs. Moreover, whether 10 mL Sepharose CL-6B column would still be the best choice for EV isolation from plasma or CSF when extending contamination markers to lipoprotein markers. As the size of lipoproteins is much larger, the purity of 10 mL Sepharose CL-6B column may be more affected than 10 mL Sepharose CL-2B columns.

Thank you to the reviewers for these comments. The purpose of our study was not to comprehensively compare all EV isolation methods, as there are many isolation methods (and many variations of each, such as different ultracentrifugation times). Our main goal was to propose a framework for rapidly comparing EV isolation methods using quantitative measurements. It has, indeed, been shown that increasing centrifugation time increases EV yield, but also free protein contamination (6). As we found SEC to be superior to ultracentrifugation in our initial survey of different isolation methods, and SEC is much higher throughput than ultracentrifugation, we decided to demonstrate the utility of our technique by applying it to the optimization of SEC. We certainly agree that albumin is not the only contaminant in EV preparations and there are other contaminants, such as lipoproteins. Additionally, separating EVs from lipoproteins is acknowledged to be challenging due to their overlapping size ranges (4, 7, 8). We agree that it is likely different SEC resins would have different effects on separation of lipoproteins and EVs. We chose albumin as it is the most abundant protein in plasma, and, thus, the main contaminant in EV preparations. As we mention in the Discussion section of our manuscript, for future studies, we plan to add other Simoa assays for ApoB100 or other protein components of lipoproteins and optimize the separation of EVs from both lipoproteins and free proteins.

3) Assessment of contamination could have been bolstered by examination of one or more proteins, e.g. lipoproteins.

As mentioned in #2, we agree with the reviewer’s comments that there is more than one type of contamination when separating EVs from plasma. Although we plan to develop Simoa assays for protein components of lipoproteins in future studies, this was beyond the scope of this study, which was to demonstrate the utility of Simoa to quantitatively compare EV isolation methods in terms of yield and free protein contamination. We chose albumin as it is the most abundant protein in plasma, and also a “model” contaminating, free protein. In subsequent studies, we hope to use Simoa assays to specifically optimize separation of EVs from lipoproteins. It would be interesting to build on the results of our study, for example, to see if we can incorporate Sepharose CL-6B into improved “dual-mode chromatography” columns, which have been previously shown to deplete lipoproteins when combined with a cation exchange resin (9). We plan to explore these questions in the future.

4) The rationale for choosing some of the methods for comparison are missing – why test with and without the wash step for UC? Why use exoquick? The error bars are missing from most figures, the reproducibility of a method should be an important factor for consideration as a good isolation method.

Thank you to the reviewers for these comments. We initially began by comparing the EV isolation methods that are most widely used in the context of biomarker studies for human biofluids. These commonly used methods are ultracentrifugation, precipitation (ExoQuick), and SEC (10). For each of these methods, we chose a common variation: ultracentrifugation is sometimes done with or without a wash step, ExoQuick has a new kit with an extra spin column step that the manufacturers claim gives higher purity (ExoQuick Ultra), and Izon offers two SEC columns (but it is unclear how the yield of EVs compares between the two). After performing a broad survey of these techniques to get a general sense of how they compare, we saw that the commercial SEC columns performed favorably, and we set out to further explore how the various parameters (resin type and volume) affect EV yield and purity using homemade SEC columns. All of these experiments were done using the same pooled and aliquoted batches of plasma and CSF. Using these samples allowed us to directly compare the Simoa measurements to each other. For the ultracentrifugation and precipitation-based methods (where one final sample is collected), we performed isolations on two different days and then performed two technical replicates on each isolation. Because there was a limit to how much pooled CSF we could obtain, and how many Simoa jobs we could reasonably run, we included SEC results from one column per condition, collecting all fractions with two technical replicates per measurement. Although we have performed many column comparisons with other batches of biofluids, we did not include those, as they could not be directly compared to the ones we included the manuscript. Instead, using a new batch of pooled CSF and plasma, we took the best conditions from the broad survey isolation methods, and redid the comparison of high purity and high yield EV isolation methods (shown in Figure 6). For that comparison, we did four replicates for each column, and two technical replicates of each condition, giving us enough data to include error bars. For figure 2, we did not feel that the low number of replicates warranted the addition of error bars. But to address variability, we have added additional supplementary figures: Figure 2—figure supplement 1 to show the variability of precipitation and ultracentrifugation performed on two different days and Figure 3—figure supplement 5 and Figure 4—figure supplement 3 to show the variability of technical replicates across all conditions tested in Figures 2-5.

5) Figure 3 should be reformatted as a supplement.

Thank you for the suggestion. We have moved Figure 3 to the supplement in the revised manuscript.

6) Frozen-thawed samples, which are commonly avoided in studies on EV, were used in this work. Why were such samples used, and would such samples produce bias to the authors' data?

We agree that pre-analytical variables such as freeze-thaw samples are very important to consider when thinking about biomarker studies. In this case, we wanted to evaluate samples that were as close as possible to samples used in clinical biomarker studies. Since such samples always undergo at least one freeze-thaw cycle, we decided to use pooled CSF and plasma that was shipped to us frozen to mimic clinical samples.

7) The pre-analytical collection of the plasma and CSF should be described more fully in the Materials and methods and a couple of sentences added to the results. The pre-analytical steps are important for understanding the downstream results. While we note that the authors had purchased the samples from a company, we would like to know more details on the sample collection process. Information that would be useful includes: i) were EDTA collections tubes used? If not, state what type of tubes (Streck, etc) ii) was the plasma spun once to separate from buffy coat, or was it spun twice to also reduce platelets? iii) were hemoglobin/hemolysis measurements taken for the pools?

Thank you for the suggestion. We have expanded the Materials and methods section to include information about the type of plasma samples that we ordered from BioIVT (EDTA collection tubes). The plasma was spun once to separate from buffy coat prior to shipment to us. However, upon thawing, as mentioned in the Materials and methods, we performed an additional spin at 2000g for 10 minutes and filtered plasma (or CSF) through a 0.45um filter. Hemoglobin/hemolysis measurements were not taken for the pools.

8) Figure 1C should have a graphic of each method tested. The graphic for SEC is clear and easy to understand. There should also be a short graphic for the ultracentrifugation with the speed of spins and the time laid out in order. There should also be a graphic for the exoquick – though that one is a little less exciting as a picture, but should describe incubation times etc.

Thank you for the suggestion. We have added a graphic to illustrate the details of each method tested as Figure 2a.

9) More details should be included in the results describing the assays chosen for isolation – qEV 35 vs 70 and ultraC with or without wash and the two exoquicks. Explanations should be included why each of these was examined and what potential differences they represented – for example between 35 and 70, etc.

Thank you for the suggestion. As mentioned in #4, we initially chose methods that were the most commonly used in context of EV biomarker studies. For each of these methods, we picked one common variation for each. For example, Izon offers two commercial columns and their website states that the 70nm columns is better for isolating large EVs relative to the 35nm column, but no additional information or data are provided. We omitted techniques such as density gradient centrifugation, for example, since this technique is not amenable to biomarker studies where large numbers of samples are used. We have added a few sentences in the Results section of our manuscript providing additional context for this.

10) In Figure 2, the emphasis should be on recovery, not contamination. Consider reordering the results at the bottom based on relative EV recovery rather than albumin concentration. So in this case, d and e and i and j would switch places and the order would go from left to right on EV recovery. It would be useful to add error bars for this figure. An important component of comparing these assays concerns their reproducibility.

Thank you to the reviewer for the suggestions. We reordered the conditions in Figure 2 based on the recovery. As we mentioned in #4, We agree that reproducibility is important and have added two supplemental figures to address this. We added Figure 2—figure supplement 1 to show experimental variability of the non-SEC methods performed on two independent days with the same batch of plasma or CSF. We also added supplementary figures (Figure 3—figure supplement 5 and Figure 4—figure supplement 3) to demonstrate the variability (CVs) of technical replicates (same sample measured twice by Simoa) for all measurements performed in Figures 2-5.

11) In the results – the authors claim that SEC is superior, but exoquick performs well in CSF. The authors should provide a set of characteristics which would objectively demonstrate that SEC is indeed superior to exoquick. Why would Izon 35 or 70 work differently in CSF compared with plasma? What might be some of the contributing factors influencing the effects of pore sizes that might aid in choosing one of these – or for understanding the albumin contamination in CSF.

We also found it interesting that some isolation methods showed a difference in relative performance in plasma compared CSF. Since plasma has a protein concentration that is approximately two orders of magnitude higher than that of CSF, we think this is the most likely reason for the difference. Our data indicating that Exoquick works relatively better in CSF compared to plasma could be due to CSF having much less albumin than plasma. When comparing all of the techniques we tested in the initial broad survey (including the custom SEC columns), as shown in Figure 4—figure supplement 2, we found that Sepharose CL-6B had higher relative EV yield than ExoQuick in CSF.

12) The construction of the SEC collection apparatus is more appropriately placed in the supplementary materials. Izon sells and automated fraction collector, a similar stand, though not as sophisticated. Indeed, almost every lab that has done SEC often fashions its own collection apparatus. Thus, while the apparatus is interesting and well-constructed, its details can be regarded as supplemental.

We have moved Figure 3 to the supplement in the revised manuscript as Figure 3—figure supplement 1.

13) A more detailed description of the pore sizes for the Sepharose chosen is needed. The testing of the columns for different size Sepharose beads and different heights is informative. Figures 4 and 5, one would like to more directly compare the 10 and 20 mL columns and would prefer they were A and B while Izon was C. And if the authors could make a measure of relative yield for the EV recovery, as they did in Figure 2 – that would be a nice additional column.

Thank you to the reviewer for these suggestions. We have included the sizes of the pores for the different resins in our Discussion section. We have also moved Izon and the 10mL column, as suggested, to more easily compare the 10mL and 20mL columns. For relative EV recovery for the SEC columns, that comparison is included in Figure 3—figure supplement 4 and Figure 4—figure supplement 2 comparing all of the different SEC methods to each other and the non-SEC methods.

14) It would be useful to readers for the authors to summarize the outcomes of the plasma and CSF studies presented herein as "Recommendations" or as a summary table indicating the best method for each.

We have included a table indicating our recommendations for “high purity” and “high yield” EV isolation from both plasma and CSF.

15) In Methods, Simoa Assays, first paragraph: It would be helpful to the reader to include the name of the Tetraspanins and detector antibodies before the catalog # and company name.

We have changed the manuscript to incorporate these changes.

16) In Methods, Simoa Assays, second paragraph: In the sentence "Next, six washes were performed" it would be helpful to readers to say what buffer was used for the washes, and if these were consistently used for all the following washes.

We clarified the wash buffer we used and state that the same buffer was used for all bead washes. We have also provided a link to a more detailed description of the methods: https://www.protocols.io/view/simoa-extracellular-vesicle-assays-bm89k9z6

17) In Methods, Preparation of Custom SEC Columns: "Columns were prepared fresh on the day of use". Can the authors comment in the Discussion whether the columns need to be prepared fresh on the day of use, or if they can be stored in the refrigerator, and if so, for how many days before use?

For this paper, we prepared all SEC columns fresh on the date of use, but we compared fresh columns to those stored for one week in the refrigerator and found that they gave similar results by Simoa. We add this comment to the Discussion.

18) In Images, Figures 4 and 5, and Supplementary Figure 1: Please include higher resolution images. In Images, Supplementary Figure 2 is difficult to interpret as presented. Suggest grouping Relative albumin concentration by column and fraction number (6b, fx 7-9 and 7-10; Izon 35, fx 7-9 and 7-10), or some way other way to visualize effects of each method vs. the albumin concentration.

In the revised manuscripts, we have replaced with high resolution images. We also added additional figures to Supplemental Figures S5 and S6 with all of the methods in the initial comparison ordered by relative albumin concentration in addition to by relative EV recovery.

19) The authors should upload the Raw data from the Simoa runs as a supplementary file. Also, please include the Group Allocation information in the manuscript.

We uploaded all raw data as a supplementary file. We now describe the Group Allocation information in the Materials and methods section and also in the Transparent Reporting Form.

References:

1. Coumans FAW, Brisson AR, Buzas EI, Dignat-George F, Drees EEE, El-Andaloussi S, et al. Methodological Guidelines to Study Extracellular Vesicles. Circulation research. 2017;120(10):1632-48.2. Hartjes TA, Mytnyk S, Jenster GW, van Steijn V, van Royen ME. Extracellular Vesicle Quantification and Characterization: Common Methods and Emerging Approaches. Bioengineering (Basel, Switzerland). 2019;6(1).3. Johnsen KB, Gudbergsson JM, Andresen TL, Simonsen JB. What is the blood concentration of extracellular vesicles? Implications for the use of extracellular vesicles as blood-borne biomarkers of cancer. Biochim Biophys Acta Rev Cancer. 2019;1871(1):109-16.4. Welton JL, Webber JP, Botos LA, Jones M, Clayton A. Ready-made chromatography columns for extracellular vesicle isolation from plasma. Journal of extracellular vesicles. 2015;4:27269.5. Norman M, Ter-Ovanesyan D, Trieu W, Lazarovits R, Kowal EJK, Lee JH, et al. L1CAM is not associated with extracellular vesicles in human cerebrospinal fluid or plasma. Nature methods. 2021;18(6):631-4.6. Cvjetkovic A, Lötvall J, Lässer C. The influence of rotor type and centrifugation time on the yield and purity of extracellular vesicles. Journal of extracellular vesicles. 2014;3.7. Simonsen JB. What Are We Looking At? Extracellular Vesicles, Lipoproteins, or Both? Circulation research. 2017;121(8):920-2.8. Sodar BW, Kittel A, Paloczi K, Vukman KV, Osteikoetxea X, Szabo-Taylor K, et al. Low-density lipoprotein mimics blood plasma-derived exosomes and microvesicles during isolation and detection. Scientific reports. 2016;6:24316.9. Van Deun J, Jo A, Li H, Lin HY, Weissleder R, Im H, et al. Integrated Dual-Mode Chromatography to Enrich Extracellular Vesicles from Plasma. Advanced biosystems. 2020:e1900310.10. Monguio-Tortajada M, Galvez-Monton C, Bayes-Genis A, Roura S, Borras FE. Extracellular vesicle isolation methods: rising impact of size-exclusion chromatography. Cellular and molecular life sciences : CMLS. 2019;76(12):2369-82.

https://doi.org/10.7554/eLife.70725.sa2

Article and author information

Author details

  1. Dmitry Ter-Ovanesyan

    Wyss Institute for Biologically Inspired Engineering, Boston, United States
    Contribution
    Conceptualization, Investigation, Methodology, Validation, Writing – original draft, Writing – review and editing
    Contributed equally with
    Maia Norman
    Competing interests
    The authors have filed intellectual property related to methods for isolating extracellular vesicles
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1134-0073
  2. Maia Norman

    1. Wyss Institute for Biologically Inspired Engineering, Boston, United States
    2. Tufts University School of Medicine, Boston, United States
    3. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
    Contribution
    Conceptualization, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Dmitry Ter-Ovanesyan
    Competing interests
    The authors have filed intellectual property related to methods for isolating extracellular vesicles
  3. Roey Lazarovits

    Wyss Institute for Biologically Inspired Engineering, Boston, United States
    Contribution
    Investigation, Methodology, Visualization, Writing – review and editing
    Competing interests
    No competing interests declared
  4. Wendy Trieu

    Wyss Institute for Biologically Inspired Engineering, Boston, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Ju-Hyun Lee

    Wyss Institute for Biologically Inspired Engineering, Boston, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6728-2071
  6. George M Church

    1. Wyss Institute for Biologically Inspired Engineering, Boston, United States
    2. Harvard Medical School, Boston, United States
    Contribution
    Funding acquisition, Resources, Supervision
    Competing interests
    GMC commercial interests: http://arep.med.harvard.edu/gmc/tech.html
  7. David R Walt

    1. Wyss Institute for Biologically Inspired Engineering, Boston, United States
    2. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
    3. Harvard Medical School, Boston, United States
    Contribution
    Funding acquisition, Resources, Supervision, Writing – review and editing
    For correspondence
    dwalt@bwh.harvard.edu
    Competing interests
    DRW has a financial interest in Quanterix Corporation, a company that develops an ultra-sensitive digital immunoassay platform. He is an inventor of the Simoa technology, a founder of the company and also serves on its Board of Directors. Dr. Walt's interests were reviewed and are managed by BWH. The authors have filed a provisional patent (WO2021163416A1) on methods for EV isolationmeasuring and purifying EVs
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5524-7348

Funding

Chan Zuckerberg Initiative (NDCN Collaborative Science Award)

  • Dmitry Ter-Ovanesyan
  • Maia Norman
  • Roey Lazarovits
  • Wendy Trieu
  • Ju-Hyun Lee
  • George Church
  • David R Walt

Open Philanthropy Project

  • Dmitry Ter-Ovanesyan
  • Maia Norman
  • Roey Lazarovits
  • Wendy Trieu
  • Ju-Hyun Lee
  • David R Walt

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors acknowledge David Kalish for help designing and making the SEC stand and Emma Kowal for help with illustrations and comments on the manuscript. Funding for this study was provided by the Chan Zuckerberg Initiative (CZI) Neurodegeneration Challenge Network (NDCN) and Good Ventures Foundation. Schematics were created with BioRender.com

Senior and Reviewing Editor

  1. YM Dennis Lo, The Chinese University of Hong Kong, Hong Kong

Reviewers

  1. Qing Zhou, Chinese University of Hong Kong, Hong Kong
  2. Julie A Saugstad

Publication history

  1. Preprint posted: October 14, 2020 (view preprint)
  2. Received: May 27, 2021
  3. Accepted: November 14, 2021
  4. Accepted Manuscript published: November 16, 2021 (version 1)
  5. Version of Record published: December 7, 2021 (version 2)

Copyright

© 2021, Ter-Ovanesyan et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,708
    Page views
  • 567
    Downloads
  • 11
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Dmitry Ter-Ovanesyan
  2. Maia Norman
  3. Roey Lazarovits
  4. Wendy Trieu
  5. Ju-Hyun Lee
  6. George M Church
  7. David R Walt
(2021)
Framework for rapid comparison of extracellular vesicle isolation methods
eLife 10:e70725.
https://doi.org/10.7554/eLife.70725

Further reading

    1. Biochemistry and Chemical Biology
    Meiling Wu, Anda Zhao ... Dongyun Shi
    Research Article

    Antioxidant intervention is considered to inhibit reactive oxygen species (ROS) and alleviates hyperglycemia. Paradoxically, moderate exercise can produce ROS to improve diabetes. The exact redox mechanism of these two different approaches remains largely unclear. Here, by comparing exercise and antioxidants intervention on type 2 diabetic rats, we found moderate exercise upregulated compensatory antioxidant capability and reached a higher level of redox balance in the liver. In contrast, antioxidant intervention achieved a low-level redox balance by inhibiting oxidative stress. Both of these two interventions could promote glucose catabolism and inhibit gluconeogenesis through activation of hepatic AMPK signaling, therefore ameliorating diabetes. During exercise, different levels of ROS generated by exercise have differential regulations on the activity and expression of hepatic AMPK. Moderate exercise-derived ROS promoted hepatic AMPK glutathionylation activation. However, excessive exercise increased oxidative damage and inhibited the activity and expression of AMPK. Overall, our results illustrate that both exercise and antioxidant intervention improve blood glucose in diabetes by promoting redox balance, despite different levels of redox balance. These results indicate that the AMPK signaling activation, combined with oxidative damage markers, could act as a sensitive biomarker, reflecting the threshold of redox balance defining effective treatment in diabetes. These findings provide theoretical evidence for the precise treatment of diabetes by antioxidants and exercise.

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Edmundo G Vides, Ayan Adhikari ... Suzanne R Pfeffer
    Research Advance

    Activating mutations in the Leucine Rich Repeat Kinase 2 (LRRK2) cause Parkinson's disease and previously we showed that activated LRRK2 phosphorylates a subset of Rab GTPases (Steger et al., 2017). Moreover, Golgi-associated Rab29 can recruit LRRK2 to the surface of the Golgi and activate it there for both auto- and Rab substrate phosphorylation. Here we define the precise Rab29 binding region of the LRRK2 Armadillo domain between residues 360-450 and show that this domain, termed 'Site #1', can also bind additional LRRK2 substrates, Rab8A and Rab10. Moreover, we identify a distinct, N-terminal, higher affinity interaction interface between LRRK2 phosphorylated Rab8 and Rab10 termed 'Site #2', that can retain LRRK2 on membranes in cells to catalyze multiple, subsequent phosphorylation events. Kinase inhibitor washout experiments demonstrate that rapid recovery of kinase activity in cells depends on the ability of LRRK2 to associate with phosphorylated Rab proteins, and phosphorylated Rab8A stimulates LRRK2 phosphorylation of Rab10 in vitro. Reconstitution of purified LRRK2 recruitment onto planar lipid bilayers decorated with Rab10 protein demonstrates cooperative association of only active LRRK2 with phospho-Rab10-containing membrane surfaces. These experiments reveal a feed-forward pathway that provides spatial control and membrane activation of LRRK2 kinase activity.