MIFlowCyt-EV: a framework for standardized reporting of extracellular vesicle flow cytometry experiments

Figures & data

Table 1. MIFlowCyt-EV framework.

Framework components	Complete for each component
1.1 Preanalytical variables conforming to MISEV guidelines*
1.2 Experimental design according to MIFlowCyt guidelines*
2.1 Sample staining details*
2.2 Sample washing details*
2.3 Sample dilution details*
3.1 Buffer-only controls*
3.2 Buffer with reagent controls*
3.3 Unstained controls*
3.4 Isotype controls**
3.5 Single-stained controls*
3.6 Procedural controls**
3.7 Serial dilutions*
3.8 Detergent-treated controls
4.1 Trigger channel(s) and threshold(s)*
4.2 Flow rate/volumetric quantification*
4.3 Fluorescence calibration*
4.4 Scatter calibration
5.1 EV diameter/surface area/volume approximation
5.2 EV refractive index approximation
5.3 EV epitope number approximation
6.1 Completion of MIFlowCyt checklist*
6.2 Calibrated channel detection range
6.3 EV number/concentration
6.4 EV brightness
7.1 Sharing of data to a public repository

*Highlights the components that are broadly applicable to the majority of single-EV analysis experiments regardless of design or instrumentation. **Highlights the components that are only applicable in cases where certain reagents or protocols are used.

Table 2. Example of a completed MIFlowCyt-EV framework.

Component	Brief example
1.1	Blood was collected from 100 individuals in 5 mL 0.109 M citrated plastic tubes (BD Vacutainer, Becton Dickinson) via antecubital vein puncture using a 21-gauge needle. The first 1 mL was discarded, prior to collection of 3.5 mL of blood. Tubes were transported vertically at room temperature. Within 1 hour of blood withdrawal, platelet-depleted plasma was prepared by centrifugation (Eppendorf 5810-R centrifuge, S-4-101 Rotor, Eppendorf) twice at 2500 ×g for 15 minutes at 20 °C. The lowest deceleration setting was used, setting ‘1ʹ. The first centrifugation step was done with 3.5 mL whole blood in 5 mL tubes (BD Vacutainer, Becton Dickinson). Supernatant was collected 10 mm above the buffy coat. The second centrifugation step was done with 2.5 mL platelet-depleted plasma in 15 mL conical tubes (Falcon Conical, Corning). The absence of haemolysis is confirmed by the lack of a spectrophotometric absorbance peak of free haemoglobin at 414 nm using a BioDrop DUO spectrophotometer. 1 mL x2 aliquots of platelet-depleted plasma were transferred to 1.5 mL low-protein binding Eppendorf tubes (Thermo Fisher Scientific) and snap frozen in liquid nitrogen before being stored at −80 °C. Age, sex, fasting status and smoking status were recorded for all individuals.
1.2	1.1 Aim: To compare the concentration of CD41a+ platelet-derived EVs in platelet-depleted plasma between individuals with type-2 diabetes mellitus (T2DM) and healthy controls. We hypothesize that the concentration of platelet-derived EVs will be increased in individuals with T2DM as T2DM has been associated with increased platelet activation. 1.2 Keywords: EV; extracellular vesicles, T2DM; type-2 diabetes mellitus. 1.3 Experimental variables: Platelet-depleted plasma samples were measured from 50 individuals with T2DM and 50 healthy controls. There was no significant difference in age, sex, fasting-status or smoking-status between individuals with and without T2DM. Scatter-based triggering was used for the detection of particles.
2.1	The presence of CD41a was determined using CD41a antibody staining. Please see Table 2 for an overview of the reagents used. 5 × 10^9 EVs (estimated by resistive pulse sensing (nCS1, Spectradyne) after size-exclusion chromatography (SEC) treatment) suspended in 495 µL of Dulbecco’s phosphate-buffered saline (DPBS) were incubated with 1x vFRed membrane stain (Cellarcus Biosciences), 5 mM CaCl^2 (Sigma Aldrich), 20 µM PPACK (Sigma Aldrich), and 5 µL of 25 µg mL^−1 anti-human CD41a mouse-IgG1κ-PE (BioLegend) for one hour at 20°C and protected from light. A matched isotype control, mouse-IgG1κ-PE (BioLegend), was incubated in the same conditions and concentration (0.25 µg mL^−1) as the anti-CD41a stained sample. The final concentration used for EV measurement using serial dilution 5 was (0.016 µg mL^−1) for anti-CD41a and its matched isotype.
2.2	Unbound antibody from CD41a-stained EV samples was removed using SEC. 500 µL of sample was added to a size exclusion column and performed according to manufacturer recommendations. Briefly, the column (qEV Original, Izon Science) was next eluted with 0.2 µm-filtered DPBS and 500 µL fractions were collected. Fractions 7–9 were subsequently pooled together for analysis. The column flow rate was ~0.8–1.2 µL min^−1, with 500 µL DPBS buffer manually maintained on top of the column.
2.3	30 µL of platelet-depleted plasma was added to 5 µL of reagents and 265 µL of DPBS, resulting in a 10-fold dilution. This 10-fold dilution was then serially diluted six times, with 150 µL of sample added to 150 µL of DBPS in a 96-well polypropylene plate (Corning). All wells were measured with the fifth serial dilution (320-fold dilution) in the series used for calculating EV concentration in the starting material.
3.1	A buffer-only control of 0.1 µm-filtered DPBS was recorded at the same flow cytometer acquisition settings as all other samples, including triggering threshold, voltages, and flow rate. The buffer-only control had a count of ~100 events s^−1.
3.2	A buffer with reagent control (0.25 µg mL-1 anti-human CD41a mouse-IgG1κ-PE (Clone: HIP8, Manufacturer: BioLegend, Cat No. 303,706, Lot No. B250952) was recorded at the same flow cytometer acquisition settings as all other samples, including triggering threshold, voltages, and flow rate. This control was serially diluted 6 times, with 150 µL of buffer with reagent added to 150 µL of DBPS to allow comparisons between serially diluted stained samples. Buffer with reagent controls had an event rate of ~100 events s^−1 and were therefore not changed from the buffer-only control.
3.3	Unstained controls were measured at the same dilution as matched stained and isotype controls. Flow cytometer acquisition settings were maintained for all samples, including triggering threshold, voltages, and flow rate. The event rate of unstained controls differed by <5% from isotype controls. No substantial changes in scatter or fluorescence signals were observed between unstained and matched isotype controls.
3.4	Isotype controls were used at the same concentration as matched stained controls and were recorded at the same dilution as matched stained and unstained controls and stained samples. Please see Table 3 for further reagent information. Flow cytometer acquisition settings were maintained for all samples, including triggering threshold, voltages and flow rate. No substantial changes in scatter or fluorescence signals were observed between unstained and matched isotype controls.
3.5	vFRed (Cellarcus Biosciences) and anti-human CD41a mouse-IgG1κ-PE (BioLegend) single-stained controls for a reference set of samples were analysed to aid compensation of anti-human CD41a mouse-IgG1κ-PE into the vFRed channel when excited by the 488 nm laser.
3.6	Excess antibody was reduced by further purification of the stained EVs using SEC columns (qEV Original, Izon Science), with fractions 7–9 collected for analysis. To assess whether this step caused artefacts, 500 µL of each control sample (buffer alone, buffer with reagent, unstained, and isotype controls) was run through a SEC column with 500 µL fractions (7–9) collected for analysis. These SEC processed samples were compared to a sample of unprocessed buffer alone, buffer with reagent, unstained sample, isotype sample, and stained sample to demonstrate the removal of antibody with no artefacts being introduced from the procedure.
3.7	Samples were serially diluted six times, with 150 µL of sample added to 150 µL of DBPS and measured using a 96-well polypropylene plate (Corning). The 4th, 5th and 6th dilutions showed a linear decrease between dilution factor and measured particle count over one minute. The median fluorescence and scatter intensity of all events for dilutions 4, 5 and 6 were maintained at 500 ± 18 PE molecules of equivalent soluble fluorophore (MESF; See Component 4.3 for MESF calibration), and 3000 ± 31 side scatter arbitrary units, respectively. Dilutions 1–3 did not show a linear correlation between dilution factor and measured event count, nor were their fluorescence and scatter intensity maintained. The fifth serial dilution (320-fold dilution) in the series was therefore used to calculate EV concentration and report immunostaining data.
3.8	Stained samples, diluted 320-fold, were treated with 0.1% Triton X-100 for 5 min at 21°C to test the lability of vFRed and CD41a-PE stained events. These measurements were used to compare vFRed and CD41a-PE stained samples not treated with detergent. vFRed and CD41a-PE positive events decreased 87% ± 9% upon treatment with 0.1% Triton X-100 for 5 min.
4.1	Based on the buffer alone control (Component 3.1), detection was triggered on the 488 nm laser excited PE-Texas Red channel (630/30 bandpass filter) at a threshold of 500 arbitrary units, equivalent to ~50 PE-Texas Red MESF, determined using Spherotech Rainbow beads and the manufacturer’s calibration values (see Component 4.3 flor fluorescence calibration). The buffer alone control (Component 3.1) had an event rate of ~100 events s^−1.
4.2	Samples were enumerated using the integral instrument flow rate sensors, resulting in a flow rate of 10 µL min^−1. This was calibrated using weighed volumes of deionized water prior to analysis.
4.3	Arbitrary PE fluorescence scale units (channel number), excited by the 561 nm laser and collected using a 586/15 bandpass filter, were converted to MESF units using PE Quantibrite beads (Becton Dickinson, Cat. 100,001, Lot. L1000001). Least-squares regression was performed between log10-transformed values of PE-A intensities versus bead PE MESF units (provided by manufacturer) using the 3 dimmest bead populations. The resulting regression showed a high correlation with an R^2 value of 0.99, with a slope of 0.931 and intercept of 1.345.
4.4	Side scatter calibration was performed using Mie modelling software, taking into account the wavelength (405 nm) and polarization state (perpendicular to detection) of the laser, the light collection geometry (side scatter, numerical aperture 1.2), and the particle diameter and refractive index. Least square fitting was used to relate the median signal of each National Institute of Standards and Technology (NIST)-traceable polystyrene bead population to the theory, resulting in a linear scaling factor of 11.3 and an R^2 of 0.99. The limiting side scatter collection angle range of the instrument was determined to be 38-142°, based on the flow cell dimensions. Scatter calibration was shown by plotting modelled vs acquired polystyrene bead data.
5.1	EV diameter was approximated using the fluorescence intensity of a membrane intercalating dye; vFRed. The vFRed cytometer collection channel was calibrated using vFRed-stained liposomes of known population diameter (median 100 nm, range ~50–150 nm) and surface area distributions, determined using nanoparticle tracking analysis (NTA) and resistive pulse sensing (RPS). To calibrate flow cytometer fluorescence intensity in terms of equivalent surface area, a least-squares linear regression was performed between the liposome population surface area and vFRed fluorescence intensity distributions.
5.2	Particle refractive index was derived from the ratio of side and forward scatter signal (i.e. Flow-SR). NIST traceable polystyrene beads with known size and refractive index (Exometry, Netherlands) were used to create a mathematical model of the optical configuration of the flow cytometer using FCMPASS software. Using this model, a Flow-SR versus diameter lookup table was calculated, which allows determination of the particle diameter from the measured Flow-SR. The determined diameter was subsequently used to derive the refractive index from a lookup table of side scatter versus diameter. Lookup tables were calculated for diameters ranging from 10 to 1000 nm, with step sizes of 1 nm, and refractive indices from 1.35 to 1.80 with step sizes of 0.001.
5.3	Anti-mouse antibody capture beads (ABC) (Quantum simply cellular, Bangs Laboratories, Cat No. 100,001, Lot No. L1000001) were incubated with 5 µL of 25 µg mL^−1 anti-CD41a mouse-IgG1κ-PE (Clone: HIP8, Manufacturer: BioLegend, Cat No. 303,706, Lot No. B250952) for 15 min at 20°C and protected from light. The PE channel units were converted to ABC units by performing regression between PE-A and ABC number using the 3 dimmest bead populations. Regression showed a high correlation with an R^2 value of 0.99.
6.1	The MIFlowCyt checklist v1.0.0 has been completed and attached in the Supplementary Information.
6.2	The median intensity of the unstained EV population was 200 PE MESF units. The maximum channel number (2^18), when scaled to MESF, was 592,435 PE MESF units. Positive events on PE were assumed to be two standard deviations above the unstained EV population, 240 PE MESF units. The detection range of the PE channel used for PE-positive events was therefore 320 to 592,435 PE MESF units.
6.3	The detected concentration of CD41a-PE positive EVs between calibrated detection range (240 and 592,435 PE MESF units) was 5.62 × 108 particle mL-1. Flow cytometer acquisition settings were maintained for all samples, including triggering threshold, voltages and flow rate.
6.4	EV brightness was compared using the median (25^th, 75^th percentile) PE MESF intensity, due to the non-parametric distribution of EV staining. Unstained platelet-derived EVs from healthy patients and T2DM patients had a median fluorescence intensity of 240 (290, 120) PE MESF. CD41a-positive platelet-derived EVs from healthy patients had a median fluorescence intensity of 535 (800, 205) PE MESF units. CD41a-positive platelet-derived EVs from T2DM patients had a median fluorescence intensity of 560 (890, 250) PE MESF units.
7.1	FC files and the analysis workspace have been uploaded FlowRepository and can also be obtained by contacting the corresponding author.

This example is of a hypothetical experiment. The WG does not currently believe there exists a “gold-standard” methodology or endorse any particular purification methods, reagents, assays or equipment. At the time of publication, no literature exists on the described hypothetical experiment. This hypothetical experiment was designed purely to require that every component of the framework be utilized and need a moderate level of detail. In many assays, every component of the framework may not need to be completed in detail if it is not relevant to the assay. In the case of a component not be relevant to an assay, a brief explanation as to why any specific component was not required should be reported. This completed example is purely a reference to the type of details that are relevant for each component and to the extent they should be discussed.

Table 3. Summary of reagents used for FC experiments.

Characteristic(s) being measured	Analyte	Analyte detector	Reporter	Isotype	Clone	Final concentration	Manufacturer	Cat. number	Lot number
Lipid Membrane	Lipid membrane	vFRed	vFRed	NA	NA	Staining = 1x Serial Dilution 5 = 0.064x	Cellarcus Biosciences	CBS4	190,415
Cell surface protein	Human CD41a	Anti-human CD41 antibody	PE	Mouse IgG1κ	HIP8	Staining = 0.25 µg mL-1, Serial Dilution 5 = 0.016 µg mL^−1	BioLegend	303,706	B250952
Non-specific binding of antibody	NA	NA	PE	Mouse IgG1κ	MOPC-21	Staining = 0.25 µg mL-1, Serial Dilution 5 = 0.016 µg mL^−1	BioLegend	400,112	B227349

Figure 1. Overview of the MIFlowCyt EV Reporting Framework. The left column shows each category of the reporting framework and the middle column shows the components within each category, the right-hand column shows the broad objective of each row. *Highlights the component that are broadly applicable to the majority of single-EV analysis experiments regardless of design or instrumentation. **Highlights the components that are only applicable in cases where certain reagents or protocols are used.

Figure 2. Summary of a poll about what extracellular vesicles (EV) flow cytometry (FCM) working group (WG) members expect to be reported in scientific manuscripts on EV-FC. The top chart shows the number of working group members who have experience reviewing manuscripts from the ISEV, ISAC, ISTH journals (red) and those that have not (white). The bottom bar graph summarizes the personal expectations of all co-authors regarding components to be reported in EV-FC manuscript published in ISEV, ISAC and ISTH journals. The expectations fall into categories of all (blue), most (green) or select/specialized (yellow) manuscripts.

Figure 3. (a) Example plot of reporting serial dilutions data, with event count per second on the left y-axis, median PE MESF intensity of the recorded data on the right y-axis and the dilution factor on the x-axis. (b) Example plot of reporting fluorescence calibration using regression. The fluorescent intensity in arbitrary units (channel number) is on the x-axis with the related fluorescence population reference units on the y-axis. (c) Example plot of light scatter calibration at 405 nm using FCM_PASS software default values. (d) Example plot of refractive index and diameter determination using Flow-SR. methodology.