A structure-derived mechanism reveals how capping protein promotes nucleation in branched actin networks

Abstract

Heterodimeric capping protein (CP/CapZ) is an essential factor for the assembly of branched actin networks, which push against cellular membranes to drive a large variety of cellular processes. Aside from terminating filament growth, CP stimulates the nucleation of actin filaments by the Arp2/3 complex in branched actin networks through an unclear mechanism. Here, we report the structure of capped actin filament barbed ends, which reveals how CP not only prevents filament elongation, but also controls access to both terminal filament subunits. In addition to its primary binding site that blocks the penultimate subunit, we find that the CP sterically occludes the central interaction site of the terminal actin protomer through one of its C-terminal tentacle extensions. Deletion of this β tentacle only modestly impairs capping. However in the context of a growing branched actin network, its removal potently inhibits nucleation promoting factors (NPFs) by tethering them to capped filament ends. End tethering of NPFs prevents their loading with actin monomers required for activation of the Arp2/3 complex and thus strongly inhibits branched network assembly both in cells and reconstituted motility assays. Our results mechanistically explain how CP couples two opposed processes -capping and nucleation- in branched actin network assembly.

The molecular mechanism of load adaptation by branched actin networks

“…Elevated capping protein levels resulted in sparser networks of shorter filaments that grew faster at comparable loads (Supplementary Figure 2). In line with our previous work (Akin and Mullins, 2008), we observed that the increase in capping was entirely compensated for by elevated nucleation rates, confirming that capping protein stimulates nucleation in branched networks (Supplementary Figure 2, (Akin and Mullins, 2008;Funk et al, 2021)). Remarkably however, filament length remained force insensitive at all capping protein concentrations tested (Figure 4A).…”

“…We show that the increased number of free barbed ends generated under load inhibits Arp2/3 complex activity by tying up WH2 domains and decreasing the ability of the NPFs to activate the Arp2/3 complex. Unlike the previously described 'monomer gating' mechanism (Akin and Mullins, 2008), in which free barbed ends compete with NPFs for monomeric actin, 'barbed end interference' is based on a direct interaction between NPFs and barbed ends of actin filaments (Funk et al, 2021;Mullins et al, 2018). This sort of negative feedback, in which the products of the nucleation reaction directly inhibit branching, also explains how Arp2/3dependent nucleation achieves a constant, homeostatic rate despite its autocatalytic nature (Mullins et al, 2018).…”

“…To test this prediction we constructed a 'bulky' capping protein by fusing dimeric glutathione Stransferase (GST) to the C-termini of both α and β subunits of capping protein (Figure 4C, bottom). From existing structures, we predict that the gap required for this bulky variant to bind to the filament end should be significantly larger than for wildtype capping protein (Funk et al, 2021;Kim et al, 2010;Narita et al, 2006). Under load, these larger gaps open much less frequently, and so the rate of capping by the bulky variant should decrease much more strongly with applied force.…”

“…Why does compressive loading reduce the rate of Arp2/3-dependent nucleation? In addition to promoting nucleation by delivering actin monomers to the Arp2/3 complex, WASP-family proteins associate with actin filament barbed ends via their WH2 domain and thus tether dendritic networks to the boundary they push against (Co et al, 2007;Funk et al, 2021). This means that free barbed ends and actin monomers are in direct competition to occupy available NPF WH2 domains (Figure 3A).…”

“…We adapted a recently developed Förster resonance energy transfer (FRET) assay (Bieling et al, 2018) to directly measure partitioning of WH2 domains between actin monomers and filament ends (Figure 3A). Briefly, we labeled a WH2-adjacent site of WAVE1ΔN with a fluorescent donor (Alexa488) and conjugated a non-fluorescent acceptor (Atto540Q) to monomeric actin using a labeling protocol that does not perturb binding to profilin and WH2 domains (Bieling et al, 2018;Funk et al, 2021).…”