Conformation of actin subunits at the barbed and pointed ends of F-actin with and without capping proteins

Abstract

Advances in cryo-electron microscopy (cryo-EM) have made possible the determination of structures of the barbed and pointed ends of F-actin, both in the absence and the presence of capping proteins that block subunit exchange. The conformation of the two exposed protomers at the barbed end resembles the “flat” conformation of protomers in the middle of F-actin. The barbed end changes little upon binding of CapZ, which in turn undergoes a major conformational change. At the pointed end, however, protomers have the “twisted” conformation characteristic of G-actin, whereas tropomodulin binding forces a flat conformation upon the second subunit. The structures provide a mechanistic understanding for the asymmetric addition/dissociation of actin subunits at the ends of F-actin and open the way to future studies of other regulators of filament end dynamics.

Actin accounts for over 10% of the total protein in most cell types and about 20% of the protein in muscle cells (Pollard, 2016). Actin exists in regulated equilibrium between monomeric (G-actin) and filamentous (F-actin) forms and plays crucial roles in countless cellular processes including motility, division, organelle transport, muscle contraction, endocytosis, and exocytosis (Pollard, 2017). Actin is also a major factor in disease, including cancer (Ampe, Witjes, & Van Troys, 2021), several muscle myopathies (Marston, 2018), and pathogen motility (Lamason & Welch, 2017). Crystal structures have been determined of G-actin alone (Otterbein, Graceffa, & Dominguez, 2001; Rould, Wan, Joel, Lowey, & Trybus, 2006) and in complexes with numerous actin-binding proteins (ABPs) (Dominguez & Holmes, 2011). A model of F-actin was first obtained more than 30 years ago by fitting the crystal structure of G-actin from its complex with DNase-I (Kabsch, Mannherz, Suck, Pai, & Holmes, 1990) to the x-ray fiber diffraction pattern of oriented F-actin gels (Holmes, Popp, Gebhard, & Kabsch, 1990). Subsequent improvements to this method led to a more refined model of F-actin and the realization that G-actin undergoes a major conformational change upon polymerization (Oda, Iwasa, Aihara, Maeda, & Narita, 2009). G-actin consists of two major domains which, due to their disposition in F-actin, are referred to as the outer (subdomains 1 and 2) and inner (subdomains 3 and 4) domains. In F-actin, the inner domain is positioned closer to the longitudinal axis, whereas the outer domain is more exposed and is frequently involved in interactions with ABPs, including myosin. G-actin is described as having a “twisted” conformation, but upon polymerization a ~20° rotation of the outer domain relative to the inner domain leads to a “flat” conformation of actin protomers in F-actin. As a result of subunit flattening, the catalytic site becomes primed for hydrolysis, explaining why G-actin is a slow ATPase whereas F-actin hydrolyzes ATP fast (Oda et al., 2009).

With the development of novel methods and instrumentation (Callaway, 2020), cryo-electron microscopy (cryo-EM) has now become the method of choice to study F-actin and its interactions with numerous ABPs. Recent cryo-EM structures of F-actin reach near-atomic resolution, producing a detailed account of the conformational changes that take place upon polymerization and how the nucleotide state influences the structure of F-actin (Oosterheert, Klink, Belyy, Pospich, & Raunser, 2022). However, these structures have been typically obtained by helical reconstruction and thus reveal the average conformation of protomers in the middle of the filament, which changes very little as a function of the bound nucleotide (ATP, ADP-Pi, ADP) or cation (Ca²⁺ or Mg²⁺) (Dominguez, 2019).

The ends of F-actin are the sole sites for filament growth/shrinkage and are targeted by regulatory proteins that control filament assembly/disassembly dynamics by either blocking or accelerating subunit exchange. It has been known for about 40 years that F-actin is kinetically asymmetric, i.e., F-actin has polarity (Pollard, 1986). At steady-state, the barbed (or +) end grows whereas the pointed (or -) end shrinks, a phenomenon known as filament treadmilling. But the source of filament polarity and end-specific protein-protein interactions have remained obscure. Specifically, until recently we did not know the conformation of actin protomers at the barbed and pointed ends or why one end grows whereas the other shrinks. We also did not know how proteins can specifically target the ends, including capping proteins that block subunit exchange, such as CapZ (also known as capping protein or CP) at the barbed end (Edwards et al., 2014) and tropomodulin (Tmod) (Fowler & Dominguez, 2017) at the pointed end. In a recent study, we addressed this knowledge gap, reporting structures of the free and capped ends of F-actin (Carman, Barrie, Rebowski, & Dominguez, 2023).

When polymerized alone, actin forms long filaments in cryo-EM micrographs (Fig. 1A). This is ideal for helical reconstruction, since filaments can be segmented, resulting in many views of their middle region but, at most, only two ends are observed per filament. To increase the amount of filament ends per micrograph, we generated short filaments (~45 nm in length) by polymerizing actin in the presence of barbed and pointed end capping proteins (CapZ and Tmod) and tropomyosin (Fig. 1B). To identify filament ends, we manually picked a small subset of filament end particles, which was then used to train the program Topaz, a particle-picking pipeline that uses neural networks trained with a general-purpose positive-unlabeled learning method (Bepler et al., 2019). Barbed and pointed ends were separated during the initial data processing steps, and subsequent classification focused on the terminal subunits to separate capped from free ends and obtain homogeneous reconstructions. Importantly, helical symmetry was not imposed during data processing. This strategy produced high-resolution reconstructions of the free barbed and pointed ends and their complexes with CapZ and Tmod, respectively.

Fig. 1: Conformation of actin subunits at the barbed and pointed ends of F-actin.

(A-B) Representative cryo-EM micrograph of actin polymerized in the absence (A) or the presence (B) of capping proteins. (C) Merged cryo-EM reconstructions of the free pointed (top) and barbed (bottom) ends of F-actin. Terminal protomers at the free pointed end have a G-actin-like, twisted conformation (pink), whereas protomers at the free barbed end have an F-actin-like, flat conformation (gray). The long- and short-pitch helices of F-actin as indicated by yellow and cyan traces, respectively. (D) Two views of the CapZ heterodimer (CapZα, blue, CapZβ, cyan) bound to the barbed end. CapZ undergoes a major conformational change upon binding, but the barbed end changes little as a result of this interaction. (E) Two views of Tmod (lime) bound to the pointed end. Tmod binds at the interface between the first three protomers at the pointed end and forces protomer-2 into an F-actin-like, flat conformation (gray). (F) The structures depicted in parts C-E demonstrate that subunits in F-actin have different conformations depending on whether they are in middle or at the ends of F-actin. These structural differences correlate with kinetic differences in the association/dissociation constants of subunits at the filament ends. Only the main pathway at equilibrium is depicted, with ATP-actin adding preferentially to the barbed end and ADP-actin dissociating from the pointed end (Pollard, 1986). CapZ and Tmod inhibit subunit exchange at the barbed and pointed ends, respectively.

Exposed actin protomers at the free barbed end have a flat conformation, similar to that of protomers in the middle of the filament (Fig. 1C). In contrast, protomers at the free pointed end have a twisted conformation, similar to that of G-actin. Thus, the barbed end appears conformationally primed for subunit addition, whereas the pointed end appears conformationally primed for subunit dissociation. In the middle of F-actin, each protomer interacts with four other protomers, including two along the long-pitch helix, one toward the pointed end and one toward the barbed end, and two along the short-pitch helix, i.e. on the opposite long-pitch helix (or strand) (Fig. 1C). The last protomer at the barbed end, however, only interacts with two other protomers, one at the pointed end along the long-pitch helix and one on the opposite strand. The penultimate protomer interacts with three other protomers, one at the pointed end and two on the opposite strand. At the pointed end, the situation is similar. The first protomer at the pointed end only contacts two other subunits, one toward the barbed end along the long-pitch helix and one on the opposite strand. The second protomer contacts three other protomers, one at the barbed end and two on the opposite strand. Because of the different environments of protomers at the ends of F-actin compared to protomers in the middle or in solution, their conformations cannot be simply described as F-actin (barbed end) or G-actin (pointed end), but rather as pseudo-F-actin and pseudo-G-actin, with some local rearrangements resulting from missing interactions with other protomers.

Reconstructions of the capped ends show that CapZ undergoes a substantial conformational change upon binding to the barbed end, which in turn remains mostly unaffected by this interaction (Fig. 1D). Tmod binding forces a flat, F-actin conformation upon the second protomer at the pointed end. Both interactions sterically block protomer addition and dissociation (Fig. 1E).

In summary, together these structures reveal the source of kinetic asymmetry in F-actin and the mechanism of barbed and pointed end capping (Fig. 1F). The strategy used in this work to obtain structure of the ends of F-actin can now be applied to other ABPs that regulate filament end dynamics.