Actin assembly is a physics problem

Actin isn’t the only protein in the cytoskeleton, but it’s a highly prevalent and versatile one. About 5% of all the protein in a typical animal cell is actin; at any given time, about half of it is floating around as free globular proteins and the other half is bound up in cytoskeletal structures.

Those structures take a variety of forms and perform many biological functions. The most basic actin structure is a thin, flexible polymer, also called a filament. Filaments can be cross-linked into two- and three-dimensional meshes, or they can be collected into strong, stiff bundles. Actin bundles—the focus of Graham and colleagues’ new research—are the basis for, among other things, the fingerlike projections cells use to probe their surrounding environment and the contractile rings that pinch off pairs of dividing cells.

Many proteins and other biomolecules are involved in creating actin structures. Graham and colleagues focused on one called VASP (vasodilator-stimulated phosphoprotein), a multifunctional enzyme that forms actin monomers into filaments and filaments into bundles. In cells, VASP molecules tend to cluster together in liquid-like droplets. But it wasn’t known what, if any, purpose the clustering might serve.

Graham didn’t set out to answer that question. Instead, she wanted to build on related work by Feng Yuan, her colleague in Stachowiak’s group, that explored how protein phase separation can influence the bending of biological membranes. Using a synthetic model membrane—chosen to eliminate the complexity of a living cell—Yuan and colleagues found that adding phase-separating proteins transformed the membrane from a perfect spherical bubble to one with deep ruts and pockets.² 

Graham wanted to see how actin filaments would affect the rutting and pocketing. Actin structures are integral to the shape of a cell, she reasoned. “So I thought that maybe adding actin polymerization would bend the membrane even more,” she says.

But when she introduced actin and VASP to the model membrane system, the most striking effect wasn’t on the membrane but on the VASP drops themselves. “They looked like Cheerios,” says Graham, “but there were others that looked like rods. What was going on? What was driving this system?” She decided to investigate further.

It’s often challenging, if not impossible, to discern the mechanism of a biochemical process by watching it in real time, because everything happens so fast: The system can be completely transformed in mere seconds, if not less. To slow the actin–VASP dynamics down, Graham reduced the amount of actin. A typical cell might have 10 or more times as much actin as VASP. Graham reversed those proportions and looked at actin-to-VASP ratios on the order of 1:10.

Some ensemble images at different actin concentrations are shown in figure 2a, and some illustrative single drops are shown in figure 2b. In all of the images, VASP is fluorescently labeled in red and actin in green—so when both molecules are present together, they appear yellow. The actin concentration serves as a rough proxy for time: Systems with less actin resemble earlier stages in the bundling process, and those with more resemble the later stages.

Together, the images start to tell the story. As actin accumulates inside the VASP drops, it’s first polymerized into single, unbundled filaments. Sometimes the filaments are individually resolvable, as in figure 2b, and sometimes they appear as a diffuse haze. Eventually, the filaments coalesce into a ring around the droplet’s edge. After that, the rings elongate into rods.

That sketch of the process still leaves gaps to be filled in: Why rings, and how do the rings turn into rods? To flesh out the details, Stachowiak reached out to her longtime collaborator Rangamani, whose group specializes in mathematical and mechanical models of biological systems, and through her put Graham in touch with Chandrasekaran.

“This was a good system to model,” says Chandrasekaran, “because the experiments are not so far removed from the modeling. When we model living cells, we have to make a lot of assumptions, and we end up talking about spherical cows.” But in Graham’s experiments, she’d stripped away most of the complexity of the cell. So the actin–VASP droplets’ energetics were defined by only a few variables: The bending energy of actin, the surface energy of the VASP drop, and the wetting interaction between them.

Wetting physics explains why the actin filaments remain confined to the VASP drops instead of poking out into the surrounding solution: Actin energetically prefers to be bathed in VASP rather than anything else. So when the filaments grow longer than the drop’s diameter, they bend. They end up hunched up against the inner droplet surface, as shown in the leftmost panel of figure 3.

As more and more filaments form, they all accumulate at the droplet surface. So when VASP starts forging the filaments into bundles, the bundles also are curled around the inner droplet surface. But actin bundles aren’t as flexible as single filaments are. As a bundle grows thicker, it becomes more resistant to bending. Eventually, as shown in the middle panel of figure 3, the bundle’s rigidity stretches the VASP drop from a sphere into a flattened disk.

As the bundling continues, the ring-shaped bundle grows, and the disk-shaped drop becomes wider and flatter. But the growth cannot continue indefinitely. When the bundle bending energy and droplet surface tension reach their collective breaking point, the ring buckles, as shown in the rightmost panel of figure 3, and becomes a rod.

A key insight from the analysis is that the rings Graham saw in her images weren’t all the same thing. Some were spherical VASP droplets uniformly lined with unbundled actin filaments—a 2D projection of a 3D spherical shell looks brighter at the edges. Others were spherical droplets containing ring-shaped actin bundles, and still others were droplets that were deformed into disks. Similarly, the rods and elongated rings that appear in figure 2a also represent a variety of morphologies. Some really are rods and elongated rings, while others are circular disks seen from the side.

It remains to be seen to what extent the droplet-driven processes play out in living cells. Observing actin bundling directly in cells would be difficult for all the same reasons that researchers opt for simplified model systems in the first place: Cells’ chemistry is orders of magnitude more complicated than simple actin–VASP drops, and their high actin concentrations drive the reactions quickly. Furthermore, VASP drops in cells are 1 µm or less in diameter, so they’re harder to see than the ones in the experiments, which are several times as large.

When a system is too challenging to observe directly, one way to figure out how it works is to break part of it and see how the rest does or doesn’t function. A version of that approach is on the researchers’ to-do list. There’s a mutated form of VASP that can still polymerize and bundle actin but can’t form phase-separated droplets. If a cell were genetically engineered to produce the mutant VASP, would its actin-bundling ability be hindered? If so, that would be strong evidence that VASP droplets—and the forces inside them—are essential to actin bundling.