Virus free EVs?

A question that surfaces repeatedly in extracellular vesicle (EV) method development is whether two desirable properties can be held at the same time: a complete, unbiased EV population isolated without affinity capture, and a preparation that carries no risk of contaminating virus. Each goal on its own is achievable. The difficulty is that the two pull in opposite directions, and the reason they do is biological rather than technical.

This article sets out why the tension exists, where physical separation genuinely resolves the problem, where it cannot, and what a defensible claim of "free of viral contamination" actually looks like for an EV preparation.

Two Questions Hidden Inside One

The first question is about completeness. Affinity capture using antibodies against tetraspanins such as CD9, CD63 and CD81 gives an exceptionally clean product, but it defines the product by the capture reagent. EVs that carry little or no target tetraspanin, larger vesicles that are poorly represented in the classical exosome fraction, and vesicle subsets from non-endosomal routes are all under-recovered. If the biological question concerns the whole secreted vesicle repertoire, positive affinity selection is the wrong tool, because it answers a narrower question than the one being asked.

The second question is about purity, specifically the presence of viral particles. Purity of that kind normally argues for selectivity, and selectivity is exactly what has just been given up. Hence the apparent impasse.

The impasse is real but not absolute, and the reason is worth stating early: selectivity can be applied to the contaminant instead of to the target. That single change of direction is what makes the combination of completeness and viral clearance tractable, and it is developed further below.

Why Enveloped Viruses Are the Hard Case

Enveloped viruses and EVs are not merely similar in size. They are, in many cases, products of the same cellular machinery. Retroviruses bud using the ESCRT machinery that also drives intraluminal vesicle formation in multivesicular bodies, an overlap formalised two decades ago as the Trojan exosome hypothesis. Reviews of the field have since described EVs and viruses as occupying a continuum rather than two discrete categories, with viral proteins incorporated into vesicles and host membrane proteins displayed on virions.

The most striking illustration is the quasi-enveloped virus. Hepatitis A and hepatitis E viruses, both classically described as non-enveloped, are released from infected cells cloaked in host-derived membrane bearing exosomal markers. In that form they are, by any physical measure applied at the bench, an EV that happens to contain a viral capsid. No membrane chemistry, no density gradient and no size-based method distinguishes them from the vesicles around them, because in every property those methods measure, they are the same object.

The practical consequence is that for enveloped and quasi-enveloped agents, a physical separation cannot be relied upon to deliver clearance. It can enrich, it can reduce, and it can sometimes achieve useful partial resolution, but it cannot exclude.

The Geometry of the Problem

Setting the numbers out side by side makes the constraint explicit. The properties that non-affinity methods act on are size, buoyant density, surface charge and hydrophobicity. For enveloped viruses, all four fall inside the range occupied by EVs. For non-enveloped viruses, density is the exception, and it is a decisive one.

Values are indicative and depend strongly on the gradient medium, run duration and whether equilibrium was reached. They are given to show the shape of the separation problem, not as specification limits.

What Each Non-Affinity Method Can and Cannot Do

The methods available for building a complete EV population all discriminate on physical properties, and each carries a distinct completeness cost and a distinct viral clearance capability.

The pattern is consistent. Every method that preserves completeness is blind to the distinction that matters, and every method that sharpens the distinction does so by discarding part of the population.

Subtractive Affinity: Selectivity Aimed at the Contaminant

The way out of the trade-off is to stop asking affinity to define the product. If an affinity step is used to capture and discard the virus, and the EV population is recovered in the flow-through, then the EV population is still defined entirely by physical criteria. Nothing about the vesicle repertoire has been pre-selected.

Chromatographic work on this problem is instructive. In one two-step process developed for HEK293 supernatant, EVs dominated the flow-through of a heparin affinity column while HIV-1 gag virus-like particles were enriched in the salt-eluted peak, giving a scalable separation of two particle classes that conventional methods leave mixed. Related pseudo-affinity chemistries, including sulphated cellulose adsorbers, have long been used to capture a range of viruses, and immunodepletion against a viral envelope glycoprotein achieves the same directional logic with greater specificity where a known agent is the concern.

Two caveats need stating plainly. First, depletion delivers log reduction, not absolute absence, so the outcome is a reduced titre rather than a guarantee. Second, the shared-composition problem cuts both ways: EVs displaying viral envelope protein will be co-depleted, and virions carrying host membrane proteins may escape capture. Subtractive affinity therefore improves the situation substantially without closing the question by itself.

Source Control Is the Dominant Lever

If the producer cell does not make virus, there is no separation problem to solve. This is not a evasion of the question, it is the single highest-impact intervention available, and it is where most of the real risk reduction happens in practice.

The elements are familiar from biologics manufacturing and are set out formally in the ICH Q5A(R2) guideline, which came into effect in November 2023 and now covers a wider range of product types. The essentials are a fully qualified cell bank with a documented virological history, removal of animal-derived raw materials (foetal bovine serum introduces both bovine EVs and bovine viral risk, and EV-depleted serum still carries residual material), control of the production environment, and screening of the unprocessed bulk.

Two categories resist this approach. Endogenous retroviral elements are present in the genome of many production lines, and rodent-derived lines in particular release non-infectious retrovirus-like particles constitutively. Human lines carry endogenous retroviral sequences that can be expressed under some conditions. These particles cannot be designed out of the cell, they are physically indistinguishable from EVs, and they are the specific reason that a claim of complete freedom from viral particles is difficult to sustain for any mammalian EV preparation.

Why the Standard Viral Clearance Toolbox Does Not Transfer

Plasma-derived and recombinant protein products achieve viral safety partly through dedicated inactivation and removal steps. Almost none of these survive contact with an EV product, because the properties they exploit (a lipid envelope, small size, protein robustness at extremes of pH) are properties the EV shares.

This table explains why EV viral safety strategies lean so heavily on prevention and detection. The middle tier of the conventional three-tier strategy, dedicated inactivation, is largely unavailable.

What "Free of Viral Contamination" Can Defensibly Mean

Absence cannot be proved in the abstract. It can be demonstrated against a stated panel of agents, by stated methods, to a stated limit of detection. That formulation is not a hedge, it is how viral safety is expressed throughout biologics, and adopting it turns an unanswerable question into an answerable one.

The regulatory direction of travel supports this approach. ICH Q5A(R2) formally introduced PCR and next-generation sequencing for adventitious agent detection, and a dedicated European Pharmacopoeia chapter on high-throughput sequencing for viral extraneous agents has followed. Broad, sequence-agnostic detection is precisely what is needed when the contaminant cannot be removed by physical means and may not be on anyone's list.

A Practical Framework

Bringing the elements together gives a workable sequence for a laboratory that wants both properties as far as they can be had at once.

• Choose the producer cell for its virological history as well as its EV yield. This decision constrains everything downstream.
• Eliminate animal-derived raw materials, and confirm that serum replacement has not simply substituted one uncharacterised particle source for another.
• Build the separation on unbiased physical methods. TFF with diafiltration retains the widest size range with the least compositional bias, and pairs well with size exclusion for a polishing step.
• Add a density step where non-enveloped virus is the credible risk, accepting the recovery cost, since this is the one place where physical separation truly clears an agent.
• Add subtractive affinity depletion where a specific enveloped agent is the concern, recovering EVs in the flow-through so that the population remains physically defined.
• Define what "free" means before generating data: which agents, by which assays, to which limit of detection.
• Report separation and characterisation in line with MISEV2023, including the co-isolated material that was not removed. Transparency about residual risk is more useful than an unqualified purity claim.

So, What Is the Answer?

A complete EV population can be prepared without affinity capture. A preparation can also be shown to be free of viral contamination to a defined and defensible limit. What cannot be done is to prove absolute freedom from viral particles by physical sorting alone, because enveloped and quasi-enveloped viruses occupy the same physical coordinates as the vesicles they are being separated from, and because endogenous retrovirus-like particles are released constitutively by many mammalian production lines.

The realistic and scientifically honest position is this: completeness is delivered by unbiased physical separation, viral risk is controlled principally at source, residual specific risk is reduced by subtractive depletion aimed at the contaminant rather than the target, and the result is demonstrated by broad and targeted testing to a stated limit. That combination gives a population that is complete by physical criteria and free of virus to a level that can be defended with data. No single sorting method delivers both, and any protocol claiming to do so is worth examining closely.

Implementing the Physical Separation Step

The unbiased separation stage of this workflow is where most preparations are won or lost, because recovery bias introduced here cannot be recovered later. The EVlution TFF system from Cell Guidance Systems is designed for reproducible concentration and buffer exchange of EV preparations at research scale, with hollow fibre cartridge formats that retain the full EV size range while allowing free protein and small molecules through to permeate. The related article on TFF diafiltration covers the operating parameters in detail.

For characterisation of the resulting population, NTA size profiling establishes the particle size distribution and concentration before and after processing, and ExoLISA assays provide quantitative surface marker detection. A full EV and exosome processing service is available for groups that prefer processed material to in-house instrumentation.

Summary

The question of whether a complete EV population can be free of viral contamination has a precise answer once the terms are separated. Completeness requires non-selective physical methods. Viral exclusion by those same methods works well for non-enveloped viruses, where density provides a decisive difference, and fails for enveloped and quasi-enveloped viruses, which are built from the same membranes by the same machinery as the vesicles themselves. The gap is closed not by a better sorting method but by a different strategy: control the source, apply selectivity to the contaminant rather than the product, and express the result as absence to a defined limit of detection against a defined panel. That is achievable today, and it is a stronger position than an unqualified claim of purity.

Further Reading

• Welsh JA, et al. Minimal information for studies of extracellular vesicles (MISEV2023): from basic to advanced approaches. J Extracell Vesicles. 2024;13(2):e12404.
• Gould SJ, Booth AM, Hildreth JEK. The Trojan exosome hypothesis. Proc Natl Acad Sci USA. 2003;100(19):10592-10597.
• Nolte-'t Hoen E, Cremer T, Gallo RC, Margolis LB. Extracellular vesicles and viruses: are they close relatives? Proc Natl Acad Sci USA. 2016;113(33):9155-9161.
• McNamara RP, Dittmer DP. Modern techniques for the isolation of extracellular vesicles and viruses. J Neuroimmune Pharmacol. 2020;15(3):459-472.
• Feng Z, et al. A pathogenic picornavirus acquires an envelope by hijacking cellular membranes. Nature. 2013;496(7445):367-371.
• van der Grein SG, Defourny KAY, Slot EFJ, Nolte-'t Hoen ENM. Intricate relationships between naked viruses and extracellular vesicles in the crosstalk between pathogen and host. Semin Immunopathol. 2018;40(5).
• Reiter K, Aguilar PP, Wetter V, Steppert P, Tover A, Jungbauer A. Separation of virus-like particles and extracellular vesicles by flow-through and heparin affinity chromatography. J Chromatogr A. 2019;1588:77-84.
• ICH Q5A(R2). Viral safety evaluation of biotechnology products derived from cell lines of human or animal origin. November 2023.

Learn more about EVlution TFF and EV purification tools from Cell Guidance Systems: