Exosome therapy and extracellular vesicles

Cells continuously shed membrane. Exosomes proper form inside the cell: inward budding of the endosomal membrane creates intraluminal vesicles within a multivesicular body, largely through the ESCRT machinery, and these are released when that body fuses with the plasma membrane. Microvesicles instead bud straight off the cell surface. Apoptotic bodies come from dying cells. The three overlap in size and in buoyant density, and once they are in a tube nothing about a vesicle records where it came from.

That is not a pedantic distinction; it is the field’s central problem. The International Society for Extracellular Vesicles addressed it in its MISEV guidance by recommending the neutral term extracellular vesicle unless biogenesis has actually been demonstrated — an unusual case of a research community formally conceding that its preparations are not what their names imply.

Why nothing separates cleanly

Every isolation method sorts on a property that EVs share with something else. Differential ultracentrifugation sorts by sedimentation, and co-pellets protein aggregates. Size-exclusion chromatography sorts by hydrodynamic radius, which in plasma puts EVs in the same window as chylomicrons and lipoproteins. Density gradients help, but HDL and LDL sit close enough to overlap. Tangential-flow filtration concentrates efficiently and discriminates poorly. Polymer precipitation is fast and the least selective of all.

The characterisation tools inherit the ambiguity. Nanoparticle tracking analysis measures the Brownian motion of scattering particles; it cannot tell a vesicle from a lipoprotein of the same diameter. So a “particle count” is a count of particles, and dosing by particles per millilitre is dosing by an undefined mixture.

The cargo arithmetic

The proposed mechanism is transfer of functional cargo — miRNA, mRNA, protein — into recipient cells. Quantifying RNA per vesicle has repeatedly given averages well below one copy of any given miRNA per vesicle, so at plausible doses most recipient cells receive no copies of the molecule credited with the effect. Some of the observed activity is therefore more likely surface protein and lipid signalling than cargo delivery. Uptake adds a second discount: internalised vesicles largely follow the endolysosomal route, and endosomal escape is as inefficient here as it is for synthetic lipid nanoparticles. Systemically injected EVs also clear within minutes into liver and spleen macrophages.

Potency, not yield

Manufacturing scale is a solved-enough engineering question — hollow-fibre perfusion raises vesicle concentration substantially over flask culture. What is not solved is what the product is. The vesicle population is a readout of the producer cell’s state, and it shifts with donor, passage number, oxygen tension and medium; serum-derived vesicles contaminate cultures unless the serum is depleted, and depletion is never complete.

A medicine needs a defined active substance, a potency assay tied to its mechanism, and specifications a batch can pass or fail. Here the active substance is a heterogeneous population, the mechanism is contested, and potency assays remain surrogate. That, rather than efficacy signals, is why no exosome therapeutic has been approved as such, and why the FDA has issued public safety warnings about unapproved exosome products sold by clinics.