Exosomes and Extracellular Vesicles

A red blood cell that is nearly mature gets rid of its transferrin receptors by throwing them out of the cell, still sitting in membrane, on vesicles about 50 nanometres across. For years this was read as rubbish disposal. The question here is whether vesicles like these can also carry a signal from one cell to another, and how you would tell a real case from an artefact of the centrifuge.

Intro to exosomes: a way to throw a receptor out

A reticulocyte is a red blood cell in its last stage of maturing. Its nucleus is gone, but it is still making haemoglobin, so it still needs iron. Iron arrives on the blood protein transferrin and enters through the transferrin receptor. A mature red cell makes no haemoglobin and carries essentially no transferrin receptors, so somewhere in between the receptors have to go.

In 1983 two laboratories reported where. In St. Louis, Clifford Harding, John Heuser and Philip Stahl coupled transferrin to particles of gold, which show as black dots in the electron microscope, and followed it into rat reticulocytes. The gold reached endosomes and then appeared in a compartment packed with small vesicles (about 50 nanometres across, by the authors’ later account), with the gold on them. Some of these compartments appeared to have fused with the plasma membrane, and their small vesicles were outside the cell. In Montreal, Bin-Tao Pan and Rose Johnstone cultured sheep reticulocytes and found that the receptor lost from the cells turned up in the medium, on vesicles they could collect in a centrifuge. Their electron micrographs of 1985 showed the same route in the sheep cells.

In 1987 Johnstone’s group used the name exosome for these vesicles. The working description is short: one membrane, usually quoted as about 30 to 100 nanometres across (authors differ on the limits), released from a cell. That is too small to resolve in a light microscope.

Purpose of MVEs: one compartment, two exits

You met the multivesicular body, also called the multivesicular endosome or MVE, earlier in this unit as the road to the lysosome. The membrane of an endosome buds inward, away from the cytosol, and pinches off small vesicles into its own interior, carrying ubiquitin-tagged receptors. If the compartment fuses with a lysosome, vesicles and cargo are digested.

The purpose of MVEs has a second answer. The same kind of compartment can move to the cell surface and fuse with the plasma membrane. Its internal vesicles are then outside the cell, intact, and those are exosomes. Picture a sorting room with two doors, one to the incinerator and one to the street. The word exosome therefore names an origin: a vesicle that formed inside a multivesicular body.

What chooses the door is only partly known. The inward budding is done by protein complexes called ESCRT, with partners such as ALIX. In a mouse oligodendrocyte cell line, loading one myelin protein into the internal vesicles needed the lipid ceramide and not ESCRT. In the human cell line HeLa, the small GTPases Rab27a and Rab27b were needed for multivesicular bodies to dock at the surface. Each result holds for the cells it was measured in.

The direction of budding fixes the geometry. Because the bud forms away from the cytosol, the inside of an exosome is a sample of cytosol, and its outer face is the face that was the outside of the cell. A receptor on an exosome points its ligand-binding domain outward, as it did on the plasma membrane.

Cells also release vesicles that bud straight outward from the plasma membrane, called microvesicles or ectosomes. They have the same orientation and overlap in size, so once both kinds are in the medium no agreed test reliably tells them apart. The usual markers of small vesicles are tetraspanins such as CD9, CD63 and CD81, membrane proteins with four membrane-spanning helices, and they are found on both kinds. For that reason the field’s own society, ISEV, urges authors to use “extracellular vesicle”, or a descriptive term such as “small extracellular vesicle”, unless they can show where a vesicle was made. Synaptic vesicles are a different thing: they stay inside the neuron, release a soluble transmitter when they fuse with the surface, and are recycled.

What is inside, and three ways it could act

The cargo list has four main entries: messenger RNA, microRNA, cytosolic proteins, and membrane proteins, receptors among them. The proteins follow from the geometry. Messenger RNA in vesicles from embryonic stem cells was reported in 2006, and in 2007 Jan Lötvall’s laboratory in Gothenburg found messenger RNA and microRNA (a short RNA that lowers the output of matching messenger RNAs) in exosome preparations from mast cells, a kind of immune cell. When vesicles from mouse mast cells were added to human mast cells, mouse proteins were detected in the human cells. Later counting set a limit on that picture: a 2014 study found far less than one copy of any given microRNA per vesicle on average.

A vesicle could act on another cell in three ways, and each needs different evidence.

1. By a ligand on its surface. A protein on the vesicle binds a receptor on the target cell. Nothing has to get in: this is ordinary receptor signalling with the ligand held on a membrane.
2. By fusing. The vesicle membrane merges with a membrane of the target cell, at the surface or from inside an endosome, and the contents enter the cytosol. Only this route lets an RNA or a cytosolic enzyme act there.
3. By being taken up. The target cell swallows the vesicle by endocytosis. On its own this shows little, because the usual fate of what a cell swallows is the lysosome. A vesicle that has been eaten has signalled only if something it carries is sensed on the way or escapes.

How would you know? The pellet is defined by the centrifuge

Vesicles are isolated by spinning. Slow spins remove cells and debris, a faster spin removes large vesicles, and about 100,000 times gravity for an hour or more brings down the small ones. What comes down is everything of the right size and density, whatever it is: protein aggregates, lipoprotein particles, complexes of RNA and protein, viruses, vesicles from the calf serum in the culture medium, and soluble growth factors stuck to any of these.

The contaminants have changed conclusions. In 2019 Robert Coffey’s laboratory separated such pellets further, with a density gradient and then capture on antibodies, and found that Argonaute proteins, which carry microRNA, were not detected in the exosomes, and that small vesicles were not vehicles of active DNA release.

So a claim needs controls. Purify further, most often by floating the pellet on a sucrose density gradient, where membrane vesicles rise to about 1.13 to 1.19 grams per millilitre and free protein stays below, and ask whether the activity moves with the vesicles. Test the supernatant left after the spin, and a pellet from medium that never saw cells. Ask whether the cargo is inside: a protein or RNA within a vesicle survives protease or RNase until detergent opens the membrane. And count.

copies per vesicle = copies of the cargo measured ÷ particles counted

If that number is far below one, or if the effect needs many more vesicles than the donor cells release, the case for signalling is weak however clean the pellet.

Design problem

You spin the medium from cell type A, add the pellet to cell type B, and B changes its behaviour. Design the controls you need before you may say that exosomes from A carried a signal to B. Name at least three other explanations, and the experiment that removes each one.

Where the evidence is good, and where it is argued

Disposal. The reticulocyte case stands, and no message is needed to explain it.

Antigen presentation. In 1996 Graça Raposo and colleagues showed that B lymphocytes release vesicles from the multivesicular compartments where MHC class II molecules, the proteins that display peptides to T cells, are loaded. The vesicles carried peptide-loaded MHC class II and stimulated T cells in culture, but only T cells whose receptor recognised that peptide on that MHC molecule. The specificity is the control, since a stray growth factor would not care which T cell it met. This is the first route, a ligand on a surface. How much such vesicles add to an immune response in the body is less clear.

Axon repair. Schwann cells are the glia that wrap peripheral axons. The laboratory of Felipe Court proposed in 2012, and reported with data in 2013, that vesicles from cultured Schwann cells are taken up by sensory axons, speed their regrowth in culture and after injury to the sciatic nerve, and lower the activity of the GTPase RhoA, which is involved in growth cone collapse. The evidence is direct, but it rests on adding prepared vesicles, so count it as supported and not yet closed.

“Cancer uses exosomes to evade therapy.” Weigh this as a claim. Tumour vesicles have been reported to carry the target of an antibody drug, such as HER2 or CD20, and to bind the antibody before it reaches the cell. They can carry PD-L1, the ligand for the checkpoint receptor PD-1, and in one study of people with melanoma, the size of the early rise in PD-L1 on circulating vesicles during anti-PD-1 treatment separated responders from non-responders. In one mouse model of prostate cancer, tumour cells with Rab27a deleted, which release far fewer exosomes, failed to grow in mice with a working immune system but grew in mice without one. The size of the effect in patients is uncertain. The decoy experiments were done mostly in culture, where the experimenter chooses the dose. A correlation in blood cannot separate cause from the amount of tumour. And Rab27a, one of the genes deleted to block release, has other jobs in secretion.

As tools. Vesicles in blood and urine carry proteins and RNA from the tissues that made them, so they are being tested as markers of disease that a blood sample can reach. Groups are also loading vesicles with drugs or RNA, and a 2011 report from Matthew Wood’s laboratory described engineered vesicles delivering a gene-silencing RNA to the mouse brain, although a 2013 study with Wood among its authors showed that the method used to load such RNA can form aggregates that make loading look more efficient than it is. Both uses depend on knowing exactly what is in the tube.