Endocytosis and Signaling Endosomes

A macrophage takes in about 3 percent of its surface membrane every minute. It swallows the equivalent of its whole surface about every half hour, and it does not shrink. The membrane must come back as fast as it goes in, and every receptor sitting in that membrane makes the trip too. So for any receptor you have met in this course, ask where it goes once it is inside the cell, and what the trip does to its signal.

What’s endocytosis used for?

Endocytosis is the pinching of a patch of plasma membrane inward to make a vesicle, a small closed bag of membrane, that carries in whatever was bound to the patch. It does three jobs.

The first is food. Cholesterol travels in the blood packed into particles of low-density lipoprotein (LDL), which are far too large for any channel or carrier. Michael Brown and Joseph Goldstein found that cells bind LDL to a surface receptor and carry the pair inside, a process that came to be called receptor-mediated endocytosis. Iron enters the same way, bound to the protein transferrin.

The second is membrane upkeep. Secretion adds membrane to the surface and endocytosis takes it back, and worn membrane proteins are pulled in for disposal.

The third is why this lesson sits in a signaling course. For a signal, endocytosis can mean two things. It can end the signal, by taking the receptor off the surface and destroying it. Or it can move the signal, by carrying an active receptor to a new platform inside the cell, where it keeps signaling or signals differently. Which one happens is decided after the receptor is in.

It all starts with endocytosis: a coat, then a cut

The best understood way in is the clathrin-coated pit. Clathrin is a protein shaped like a three-legged pinwheel, and many copies lock together into a curved cage on the inner face of the membrane. Clathrin touches neither the membrane nor the cargo. Adaptor proteins do that. The main adaptor at the plasma membrane, AP2, holds the lipid PIP2 (an address from the lesson on lipid anchors), short sequences in receptor tails, and clathrin. So a receptor enters a pit because its cytoplasmic tail carries a sequence that an adaptor can read.

Brown and Goldstein met the proof in a patient known as J.D. This patient’s LDL receptors could bind LDL but stayed scattered over the surface instead of gathering in pits, and the cells could not take LDL in. The fault was later traced to one tyrosine in the receptor’s tail, replaced by cysteine.

A pit deepens until it hangs from the surface by a thin neck. Dynamin, a large GTPase, assembles into a collar around the neck, and when it hydrolyzes GTP the neck is cut. The coat is then stripped off and the bare vesicle can fuse with its target.

There are other doors. Caveolae are small flask-shaped pits built around a protein called caveolin, macropinocytosis gulps large volumes of fluid, and several more routes that need no clathrin have been described and are less well mapped.

Signaling defines which internalization pathway it takes. The best studied case is the EGF receptor. In HeLa cells, Sara Sigismund, Pier Paolo Di Fiore and their colleagues reported that at low doses of EGF the receptor enters almost entirely through coated pits and is mostly recycled. At high doses the receptor also becomes tagged with ubiquitin, and a share of it enters by a route without clathrin and is mostly destroyed. Treat that as a result in one cell line, since other laboratories find clathrin carrying most of the receptor at high doses too. The modest version is safe: how hard a receptor is signaling can change how it enters and where it ends up.

The early endosome is a sorting station

The early endosome is the first stop for almost all endocytosed cargo. It lies near the plasma membrane, and it sorts cargo into separate patches of its own membrane that set the cargo’s fate: back to the surface, or on to the lysosome to be destroyed.

Two physical features do much of the sorting. One is acid. A proton pump in the endosome membrane holds the inside at about pH 6 to 6.5, and at that pH many ligands let go. LDL falls off its receptor, and iron falls off transferrin. The other is shape. An early endosome has a round body with thin tubes growing out of it. A tube is nearly all membrane and almost no volume, so membrane proteins spread into the tubes while dissolved contents stay in the body. The tubes pinch off and go back to the surface, so a receptor with no other instruction mostly goes back too.

There is a fast way back, taking a few minutes, straight from tubes rich in the protein Rab4. There is a slow way, taking tens of minutes, that passes through a second compartment and uses Rab4 and Rab11. An LDL receptor makes the round trip about every ten minutes, again and again.

Rab proteins define and drive the maturation state. Rabs are small GTPase switches of the kind in the lesson on GTPase switches: on with GTP, off with GDP, flipped by GEFs and GAPs. Human cells have more than sixty, and each marks one compartment or one step of traffic. Rab5 marks the early endosome. In its GTP form it sits on the membrane and recruits effectors, proteins that bind only the on state, which tether incoming vesicles and help them fuse.

How does a Rab find its membrane? Each Rab carries geranylgeranyl groups, 20-carbon lipids, on its tail, and a greasy tail would stick to any membrane it met. Suzanne Pfeffer’s laboratory worked on this with Rab9. A protein called GDI wraps the GDP-bound Rab and hides the tail, so the Rab stays dissolved in the cytosol. Her group found an activity on endosome membranes, a displacement factor, that frees the Rab from GDI, and showed that GDP is exchanged for GTP as the Rab lands. GDI does not pull off the GTP form, so the Rab stays put until a GAP switches it off.

Headed to the lysosome: the receptor is sealed inside

A receptor leaves the default route when it carries a tag, and the tag is one you know: ubiquitin. For an active EGF receptor the ligase Cbl adds it, as the lesson on receptor tyrosine kinases describes. This ubiquitin is not read by the proteasome. A set of proteins on the endosome called ESCRT binds it, gathers tagged receptors into a patch, and pushes the patch inward until it pinches off as a small vesicle inside the endosome. An endosome filled with such vesicles is a multivesicular endosome (MVE), also called a multivesicular body.

What is the purpose of MVEs? Think about which way the receptor faces. On the outer membrane of an endosome, the receptor’s tail still points into the cytosol and can still reach its partners. On an internal vesicle, the tail is sealed inside and cannot reach the cytosol at all. That step ends the signal before anything is destroyed. It also puts the whole receptor where the lysosome’s enzymes can reach it.

Meanwhile the compartment itself changes. Marino Zerial’s laboratory filmed single endosomes in living cells and saw Rab5 leave as Rab7 arrived, which they called Rab conversion. With Rab7 come new effectors, stronger acid (about pH 5.5 in late endosomes) and a growing load of internal vesicles. Their membrane is rich in an unusual lipid, lysobisphosphatidic acid, which has not been detected in early endosomes. The late endosome then fuses with a lysosome, a bag of digestive enzymes at about pH 4.5 to 5, and the receptor is broken down.

Design problem

Two receptors on the same cell bind their ligands at the same moment. One must signal for about ten minutes and then be silent for hours. The other must keep signaling for hours. Using only the sorting decisions made at the early endosome, design the difference. Say what tag or sequence each receptor would need.

A receptor that keeps talking, and one that escapes

Between the coated pit and the internal vesicle, a receptor spends time on the outer membrane of an endosome with its tail in the cytosol. Is it still signaling? In 1996 Sandra Schmid’s laboratory tested this by blocking endocytosis in HeLa cells with a dynamin mutant defective in GTP binding and hydrolysis. If endocytosis only ended signals, every response to EGF should have gone up. The cells did divide more, but some responses fell, including full activation of the MAP kinase ERK. On this evidence part of the EGF signal needs the receptor to be inside, although later experiments that blocked endocytosis in other ways did not all find the same fall in ERK.

Other receptors change what they say when they move. Toll-like receptor 4 signals through the adaptor MyD88 at the plasma membrane. After it is taken in, it signals through a second adaptor, TRIF, which switches on interferon genes. The Toll-like receptors that detect nucleic acids, such as TLR3, TLR7 and TLR9, work only in endosomes, where the remains of digested microbes arrive. And some G-protein-coupled receptors, taken in with arrestin as the lesson on turning signals off describes, go on activating G proteins from endosomes.

Now break the sorting. In roughly 3 percent of non-small-cell lung cancers, a splicing error removes exon 14 from the messenger RNA for MET, a receptor tyrosine kinase. The missing stretch holds tyrosine 1003, the docking site for Cbl. The receptor is poorly tagged with ubiquitin, escapes the lysosome and keeps signaling for longer than it should. The fault is not an overactive kinase. It is a receptor that misses its turn to the lysosome. Two MET kinase inhibitors, capmatinib and tepotinib, are approved for lung cancers with this change.

A neuron poses the hardest version of the keep-talking problem, because its endosomes must carry a signal the length of an axon. That is the next lesson.