Figure 1. Schematic representation and description of the main steps involved in the three methodologies used for the isolation of extracellular vesicles. (Gámez-Valero et al., 2016)
1. Define the Starting Material and EV Claim
Describe source material, collection, donor or culture conditions, volume, storage, preprocessing, and intended EV population. Extracellular vesicles are heterogeneous in size, composition, and origin. Avoid assigning a biogenesis-specific label solely from particle diameter or one marker.
Define the application: exploratory analysis, biomarker work, functional assays, cargo studies, or process development. The acceptable tradeoff between recovery, soluble-protein removal, lipoprotein separation, concentration, and throughput depends on the endpoint.
Characterize the main contaminants
Potential contaminants include abundant soluble proteins, protein aggregates, lipoproteins, nucleic acids, cell debris, media components, and particles from supplements. Their abundance and size overlap determine what SEC can and cannot resolve. A clean chromatogram does not establish particle identity.
2. Prepare Feed without Creating or Losing Particles
Remove cells and large debris using a defined centrifugation or filtration sequence appropriate to the sample. Record force, time, temperature, membrane, pore rating, and recovery. Aggressive handling can remove large EVs, create aggregates, or release new particles.
Control freeze-thaw, storage duration, and container surfaces. Compare fresh and stored samples when the workflow will use both. Mix gently and avoid foam. Measure input volume, particle signal, total protein, and selected markers before SEC so recovery has a defensible denominator.
Concentrate only when needed and validate the method. Concentration can co-enrich soluble proteins, cause vesicle loss, or change aggregation. Account for hold-up in membranes and devices. Do not overload SEC to compensate for a poorly controlled concentration step.
3. Use SEC according to Its Nonbinding Size Mechanism
SEC separates by differential access to pores. Particles excluded from most pore volume elute earlier; smaller soluble molecules enter more pores and elute later. The method does not bind EVs and usually uses an isocratic mobile phase compatible with particle stability.
Select matrix and pore range so EV-sized particles are excluded or partially excluded while smaller contaminants are retarded. If all large particles elute together near the void, SEC cannot distinguish EVs from similarly sized lipoproteins, aggregates, or nonvesicular particles.
Use size exclusion chromatography products with a justified bed geometry and sample. Product categories provide starting options; the actual fractionation must be confirmed for the source matrix.
4. Control Sample Volume, Concentration, and Column Geometry
Injection volume relative to bed volume strongly influences resolution. A large sample plug broadens early and late populations, causing overlap between vesicle-rich and protein-rich fractions. Determine the maximum sample volume that preserves the required separation rather than using the largest volume the hardware can accept.
High concentration can increase viscosity, particle interactions, and nonspecific effects. Clarify or filter cautiously before injection when particles that can block the bed are present. Measure pressure and flow under the actual sample and buffer.
Use consistent bed height, packing, and flow. Extra-column tubing and fraction size can be large relative to early peaks. Characterize void and total permeation behavior with suitable standards, recognizing that globular standards do not fully model heterogeneous vesicles.
5. Design Fraction Collection before the Run
Choose fraction volume and collection timing based on bed volume, expected void region, system delay, and analytical sensitivity. Smaller fractions improve mapping but increase assay demand and surface exposure. Correct for postcolumn hold-up so labels reflect the material actually collected.
Analyze individual fractions during development before combining them. Plot particle measurements, total protein, lipid or membrane signal, EV-associated markers, contaminant markers, and volume. A particle-rich fraction with high soluble protein is not necessarily the preferred pool.
Define pooling rules prospectively. Depending on the application, early fractions may provide higher purity but lower recovery, while broader pooling improves recovery at the cost of protein or lipoprotein contamination. Report this tradeoff rather than presenting one pool as universally optimal.
6. Use Orthogonal Evidence for Identity, Recovery, and Purity
No single assay establishes EV identity or purity. Combine particle concentration and size with imaging or other structural evidence, protein and lipid measurements, positive EV-associated components, and markers of expected non-EV contaminants as appropriate. Use method controls and state detection limits.
| Question | Example evidence | Interpretation caution |
|---|---|---|
| Are particles present? | Particle count and size distribution | Non-EV particles can be counted |
| Are vesicle-associated components enriched? | Multiple appropriate markers | Marker presence alone does not show purity |
| Were soluble proteins reduced? | Protein mass and source-specific contaminants | Low protein can also reflect EV loss |
| Are lipoproteins separated? | Relevant apolipoprotein or lipid analysis | Size overlap can limit SEC resolution |
Use ratios such as particles per protein only with clear units and assay conditions. Ratios can rise because protein is removed or because particle measurements are biased. Retain the underlying values.
7. Calculate Recovery with a Complete Volume Balance
Define recovery as a measured EV-related quantity in the selected pool divided by the same quantity in the applied sample. Correct for dilution and include all fractions when examining mass balance. Particle counting methods can be matrix sensitive; use consistent dilution and instrument settings.
Track hold-up and adsorption in the column, tubing, filters, and collection vessels. A low recovery may reflect excluded early material collected before the chosen window or late tailing. Analyze wash or subsequent blank fractions when carryover is suspected.
Report pool concentration and total recovered quantity. A highly pure but dilute pool can be difficult to use and may require a concentration step that changes recovery. Include post-SEC concentration in the workflow evaluation.
8. Address Lipoproteins, Aggregates, and Size Overlap
SEC alone may not separate EVs from lipoproteins or protein aggregates of overlapping hydrodynamic size. Characterize source-specific contaminants across fractions. If overlap remains unacceptable, add an orthogonal step based on density, charge, affinity, or another justified property.
Sequence matters. A preclear or selective depletion can reduce the load of a co-eluting contaminant, while a post-SEC affinity step can enrich a defined subpopulation. Each added step changes recovery and may bias the EV population.
Avoid claiming absolute purity. State which contaminants were measured, which were below detection, and which were not assessed. Use functional controls to determine whether observed biological activity follows EV-rich fractions or a soluble co-eluting component.
9. Scale SEC by Preserving Resolution Drivers
Preserve sample volume relative to bed volume, bed height, appropriate velocity, buffer, temperature, and fraction resolution. Increasing diameter can raise throughput at comparable bed height, but distribution, system delay, and collection capacity must be rechecked.
Monitor pressure and packing; particulate feeds can foul the inlet. Develop cleaning compatible with the matrix and intended reuse, then verify removal of particles and proteins with blank runs. Consider single-use operation when cleaning evidence or sample risk makes reuse unsuitable.
Scale analytics with the process. A larger number of fractions or batches can strain particle and marker assays. Define a development panel and a smaller routine panel that remains sensitive to the critical separation.
10. Choose an EV Pool from Fraction-Resolved Evidence
Consider a plasma-derived sample concentrated before SEC. Early fractions show high particle counts and low total protein, while the next fractions add substantially more particles but also a strong lipoprotein marker. Pooling all particle-rich fractions maximizes count recovery but weakens compositional purity.
The team plots particle number, protein mass, selected EV-associated markers, lipoprotein markers, and fraction volume. Imaging confirms vesicular structures in the earliest region but also reveals nonvesicular particles later. A particle-to-protein ratio alone would not expose this change because both numerator and denominator vary.
Define two pools for two applications
For a molecular cargo study, a narrow early pool is selected to reduce lipoprotein contamination even though particle recovery is lower. For an exploratory functional assay, a broader pool is evaluated, but soluble-protein and lipoprotein controls are included so biological activity is not assigned automatically to EVs.
A blank column run and a postrun wash identify low carryover of particle signal. Recovery calculations include the concentrated input, every SEC fraction, tubing hold-up where measurable, and the post-SEC concentration step. The latter accounts for more loss than SEC itself, changing where optimization effort is directed.
When injection volume is doubled, the early EV-rich and later protein regions overlap. The process returns to the lower sample-to-bed-volume ratio and instead uses parallel runs or a larger-diameter bed at comparable height. This preserves the resolution mechanism rather than accepting a high-throughput overload.
The team then tests a source with different lipoprotein abundance. The same fraction numbers no longer produce the same purity. Pooling is therefore defined by verified column volumes and fraction analytics, with a reference profile, rather than a fixed tube list assumed to work for all sources.
Cleaning is developed with representative material and verified by particle, protein, and marker blanks. Because some contaminants are difficult to remove and the intended analyses are sensitive, the team also evaluates restricted reuse. Column history is linked to every sample.
The final report presents narrow- and broad-pool recovery, protein and lipoprotein reduction, EV evidence, volume, and postconcentration yield. It states which contaminants were assessed and avoids claiming absolute purity. The chosen pool is explicitly tied to the downstream application.
11. Report an SEC EV Workflow Transparently
Report source, preprocessing, storage, input volume, column matrix and dimensions, bed volume, flow, buffer, sample load, fraction volume, pooling, recovery, concentration, identity evidence, and contaminant measurements. Include instrument and assay settings needed to interpret particle data.
Preserve chromatograms and fraction-level results. A summary purity ratio cannot show whether the pool boundary was robust. Document deviations, column history, and any evidence of carryover.
For workflow support, share the source matrix, desired EV population, contaminant profile, sample volume, analytical capabilities, and downstream use through custom chromatography services. The goal is a measured balance of recovery and purity for the stated application.
Assess column-to-column and run-to-run reproducibility with a common reference sample when possible. Record packed bed volume, void position, flow, pressure, and fraction boundaries. A gradual shift in void volume or early-peak width can indicate packing or fouling before routine purity metrics fail.
Use source-matched blanks and process controls. Cell-culture media, platelet-poor plasma, or other matrices can contribute particles and signals that survive preprocessing. Run controls through the same concentration and SEC sequence so contamination introduced by devices or reagents is visible.
If biological activity is the endpoint, test activity across several neighboring fractions and include soluble-protein or lipoprotein controls. Depletion, neutralization, or re-isolation experiments can help determine whether activity follows EVs. Avoid attributing function to an EV marker that merely co-occurs with a soluble effector.
For repeated small samples, evaluate whether miniature beds preserve sufficient resolution and recovery. Extra-column volume and fraction-collector timing can dominate at low bed volume. A device that is convenient for throughput may require a different fraction scheme and separate qualification.
Document the population bias created by the workflow. Pre-filters, concentration membranes, and SEC pores can favor certain sizes or surface properties. The recovered pool represents vesicles compatible with the full process, not necessarily the complete starting EV population.
Where size overlap remains severe, test an orthogonal method on fraction-defined material rather than the crude feed first. This reveals whether the added mechanism removes the actual co-eluting contaminant and how much EV signal it sacrifices. Compare sequences in both orders when practical, because affinity or charge treatment before SEC can change aggregation and the particle population presented to the size-exclusion bed.
Retain enough of the starting sample and individual fractions for repeat analysis. This allows a questionable particle count, marker result, or functional observation to be checked without reconstructing the entire isolation and helps distinguish analytical drift from a true fractionation difference.
Develop an EV fractionation strategy with measurable tradeoffs
Share the source, preprocessing, sample volume, contaminant profile, EV assays, and downstream use to define an informative SEC study.
References
- Benayas B, Morales J, Egea C, Armisén P, Yáñez‐Mó M. Optimization of extracellular vesicle isolation and their separation from lipoproteins by size exclusion chromatography. J of Extracellular Bio. 2023;2(7):e100. doi:10.1002/jex2.100
- Carta G, Jungbauer A. Protein Chromatography: Process Development and Scale-Up. 2nd edition. Wiley-VCH; 2020. doi:10.1002/9783527824045
- Gámez-Valero A, Monguió-Tortajada M, Carreras-Planella L, Franquesa M, Beyer K, Borràs FE. Size-Exclusion Chromatography-based isolation minimally alters Extracellular Vesicles’ characteristics compared to precipitating agents. Sci Rep. 2016;6(1):33641. doi:10.1038/srep33641
