Protein strangles membrane necks by polymerizing into a spiral collar

Signals travel along neurons as pulses of electrical polarization, but they’re carried between neurons by glutamate, serotonin, and other neurotransmitter molecules. Neurons keep neurotransmitters ready for use inside nanoscale bags called synaptic vesicles, whose skins are made from the same lipids as the neuron’s outer membrane.

When it’s time to pass on a message, the vesicles fuse with the inside surface of the neuron’s membrane and break open. The debouched neurotransmitters diffuse across a gap of a few nanometers to reach the receiving neuron, where, by binding to the surface, they deliver the message. Spent neurotransmitters are broken up by enzymes floating in the gap.

Neurons remake vesicles in a process that’s the reverse of the vesicles’ destruction. A concave pit forms out of the neuron’s membrane. As the pit becomes more spherical, the neck that connects it to the membrane narrows. Vesicle formation ends when the neck is cut to seal the vesicles and trap them inside the neuron. Transporter proteins in the vesicle membrane reload the vesicles with fresh neurotransmitters.

Biologists call the release and reabsorption of vesicles exocytosis and endocytosis. Both processes are rich ground for biophysical study. They involve the topological transformation and continuum mechanics of thin, almost liquid membranes and the participation of several molecular actors, among them a motor protein called dynamin.

Dynamin polymerizes to form a spiral collar around the vesicle’s neck during the final stages of endocytosis. But until now, it wasn’t clear how the protein is recruited at the right moment or how it begins squeezing the vesicle’s membrane. Aurélien Roux and his coworkers in Patricia Bassereau’s group at the Curie Institute in Paris have cleared up both mysteries. ¹ Surprisingly, the resolution lies not in a varied cast of biochemical actors, as is often the case in molecular biology, but in the polymerization process itself.

Endocytosis

Figure 1 depicts some of the steps in endocytosis. In the first, molecules of protein called clathrin bind to the inside surface of the neuron’s membrane. Clathrin molecules also bind to each other, creating a crystalline cage that deforms the membrane into a spherical pit. Dynamin shows up when the clathrin cage is almost complete and a short neck has formed.

Clathrin is recruited by the advent in the membrane of a lipid called PIP₂. Like clathrin, dynamin also recognizes PIP₂, but without an additional means to sense membrane curvature, dynamin would bind promptly—and therefore uselessly—to the pit and not to the more highly curved neck that forms later. Certain proteins that bind to dynamin are sensitive to curvature and could in principle recruit dynamin at the right moment.

Other molecules might help dynamin when it forms a spiral collar. Energy is needed to deform the membrane into the thin tube that dynamin envelops. Conceivably, that energy could come from dynamin’s polymerization, from a molecule that binds to dynamin, or from an unknown molecule that squeezes the membrane in advance of dynamin’s adsorption.

Roux didn’t set out to find how dynamin senses curvature. At first, he was more interested in dynamin’s polymerization. In 1998 Sharon Sweitzer and Jenny Hinshaw of the National Institute of Diabetes and Digestive and Kidney Diseases in Bethesda, Maryland, mixed dynamin and lipids in a test tube and found that the two ingredients form thin membrane tubes wrapped by dynamin spirals. ² The spirals have the same inside radius, about 10 nm, as that of the dynamin monomers. Given that dynamin can be coaxed to form spirals of the same radius even without lipids, Roux figured that dynamin’s polymerization could provide the energy needed to squeeze and elongate a vesicle’s neck. Electron micrographs of dynamin on curved, never flat, membranes were consistent with that proposal.

For verification, Roux sought to measure the force dynamin exerts on membranes when it polymerizes. Cell membranes are thin and delicate. At 300 K, the benchmark temperature of living things, the energy needed to bend or stretch membranes barely exceeds the energy of their thermal fluctuations. They are also transparent. Measuring their mechanical properties is difficult even without added dynamin.

When he joined Bassereau’s group in 2007, Roux and coworkers began adapting a method developed by another member of the group, Pierre Nassoy. Nassoy had previously shown how one could measure a membrane’s elastic moduli by manipulating cell surrogates called giant unilamellar vesicles. Micron-scale GUVs are much larger than synaptic vesicles, and even some cells (hence “giant”); they have one layer of lipids as opposed to the usual two (hence “unilamellar”); and they carry little more than ambient fluid (hence “vesicle”).

Figure 2 shows the basics of Nassoy’s method. A micropipette applies suction pressure to hold a GUV in place while a glass bead is chemically attached to the surface. An optical trap pulls the bead away to create a thin tube. The force required to hold the bead in place depends on the stretching modulus, which increases with suction pressure, and on the bending modulus, which doesn’t. Adjusting the pressure calibrates a relationship, first derived in 1994 by Evan Evans and Anthony Yeung, ³ between the measurable force on the bead and the otherwise unmeasurable radius of the tube.

Figure 2 also shows Roux’s modification: the introduction of dynamin (red dots) via a second micropipette. Roux could adjust two control parameters: the concentration of dynamin and, through suction on the GUV, the radius of the tube. By fluorescently tagging dynamin molecules, he could observe the protein sticking to a tube through a microscope. Although the polymer’s spiral structure could not be resolved, the accumulation grew in a way consistent with polymerization: by extending at both ends.

At first, Roux found it hard to fix the dynamin concentration. Sometimes the micropipette was too close to the GUV and the concentration too high; sometimes the micropipette was too far and the concentration too low. To his surprise, dynamin polymerized on a tube even at modest concentration, provided the tube radius was reduced to around 10 nm, dynamin’s inside radius. After seeing that polymerization depends on both tube radius and dynamin concentration, Roux controlled the concentration more precisely.

A typical experimental run at low dynamin concentration appears in figure 3. Micrographs are at the top, a plot of the tube radius versus time is at the bottom. Fluorescently tagged PIP₂ (top, red) delineates the GUV and the tube. Fluorescently tagged dynamin (top, green) shows up only when the tube is at its lowest radius, 10.4 nm.

By taking measurements over a range of concentration and radii, Roux and his coworkers could map a phase diagram of dynamin’s polymerization. Below a critical concentration c ₁, dynamin never polymerized. At c ₁, dynamin polymerized only on tubes whose radius matched dynamin’s. As the concentration increased, dynamin polymerized on tubes of a widening range of radii. The lower limit of the range decreased weakly with concentration; dynamin couldn’t polymerize on tubes that were too thin. The upper limit increased rapidly with concentration. Evidently, the vigor of dynamin polymerization was enough to deform ever thicker tubes.

A simple mathematical model that balances polymerization energy, which depends on concentration, and the elastic energy, which depends on concentration and tube radius, could reproduce the phase diagram. The molecular picture that emerges from the experiment is of matching geometries. Curved monomers slide around on a curved surface and readily link if the two curvatures match. If they don’t match, the linking can force a match, provided enough monomers are present.

Roux could also measure the polymerization force, which was manifest as a reduction in the force required to hold the bead in place. In general, the polymerization force depends on dynamin concentration and membrane tension. At a concentration of 12 µmol/L, the force is 18.1 ± 2.0 pN.

Interestingly, Roux’s results imply that dynamin cannot exert enough force to overcome the higher membrane tensions measured in real neurons. However, certain membrane proteins appear in a nascent pit to reduce the tension and regulate the onset of endocytosis.

In the final stage of endocytosis, polymerized dynamin obtains energy from molecules of GTP (guanosine triphosphate, a common cellular fuel), twists into a tighter spiral, and garrotes the neck. When Roux was a postdoc in Pietro De Camilli’s lab at Yale University, he, De Camilli, and their coworkers had verified the GTP-fueled twisting. ⁴ Now, Roux plans to measure the twisting force.