A Third Look at the Structure of Leiomodin Bound to Actin

Main Text

Our quest in science is to understand Nature by seeking the truth. One element of this quest is verification, usually by building directly on past work with new experimental data, theory, or interpretation. Progress can also be made by reevaluating published data; I have done this multiple times during my career, so it must be a common experience. Such is the case in an article on a crystal structure of the protein leiomodin bound to actin by Boczkowska et al. in this issue of the Biophysical Journal (1). Fortunately, crystallographers deposit their diffraction data in the publicly available Protein Data Bank, making it possible for anyone to recalculate electron density maps and build new models using original data. Boczkowska et al. have done so with the data from an article by Chen et al. (2) and have come up with a model that differs from that proposed by Chen et al. (Fig. 1, B and C). Together with new biochemical data, Boczkowska et al. use their new model to propose how leiomodin nucleates actin filaments in muscle cells.

Figure 1.

Comparison of models for tropomodulin- and leiomodin-binding actin filaments and monomers. (A) Model of tropomodulin bound to the pointed end of an actin-tropomyosin filament. The figure is modified from Pollard (12). (B) Model by Chen et al. (2) of one Lmod2 molecule bound to two actin molecules in the unit cell of their crystal structure. This image from the Chen et al. article was rotated and relabeled. (C) Model by Boczkowska et al. (1) of one Lmod2 molecule bound to one actin molecule based on recalculating the electron density map from the data of Chen et al. and rebuilding the model. In the revised structure, the unit cell contains two such complexes related by a local twofold symmetry axis, shown below. These images from the Boczkowska et al. article were relabeled.

So, what is leiomodin and why is it important? Leiomodin is a member of a family of proteins that regulate actin polymerization by interacting with the slow-growing “pointed end” of actin filaments (3). Tropomodulin, the founding member of the family, was discovered in the spectrin-actin membrane skeleton of red blood cells (4), where it binds the pointed ends of short actin-tropomyosin filaments (5). Three other isoforms are found in other tissues. An unrelated protein called αβ-adducin binds to the fast-growing “barbed end” of these filaments and anchors them to the plasma membrane. Neither protein fully caps the ends, so subunits exchange slowly at both ends (6). Binding several long, tetrameric spectrin molecules to each short actin filament creates an elastic, hexagonal, two-dimensional network anchored to and supporting the plasma membrane of red blood cells (6). Subsequently, it was discovered that tropomodulin stabilizes the pointed ends of the actin thin filaments in striated muscle cells (7).

Biochemical characterization of the tropomodulin domains that interact with actin and tropomyosin and crystal structures of tropomodulin domains bound to actin have given us a detailed understanding of how tropomodulin binds to the pointed end of the actin-tropomyosin filament and inhibits actin subunit addition and dissociation (8). A single tropomodulin molecule caps the pointed end by draping the extended actin binding site 1 (ABS1) over the terminal subunit of the filament, while the leucine-rich repeat (LRR) domain comprising ABS2 binds to the back of subdomains 1 and 2, positioned so that it can interact with three adjacent actin subunits in a filament (8) (Fig. 1 A). Interactions of sequences flanking ABS1 with tropomyosin increase the affinity for the pointed end of the filament 5- to 10-fold.

Leiomodins were discovered later (9) and were shown to contribute to the assembly of actin filaments in muscle cells (10). Mutations in the genes for each of the three human leiomodins predispose to diseases including myopathies of heart, skeletal, and smooth muscles. These leiomodin isoforms share several domains with tropomodulins, but they are larger owing to C-terminal extensions containing proline-rich and WH2 domains. Despite having a common ancestor, tropomodulins and leiomodins differ in biochemical properties and cellular functions. Leiomodins nucleate actin filaments but do not cap their pointed ends, opposite in both regards to tropomodulins (10). Nevertheless, a crystal structure showed that the leiomodin ABS2 domain contacts three actin subunits like that of tropomodulin, whereas subtle differences in sequence make ABS2 the key element that promotes nucleation by leiomodin (11).

Chen et al. (2) reported a co-crystal structure of leiomodin-2 (Lmod2) with mutant actins that do not polymerize. They interpreted their electron density maps as one leiomodin molecule connecting two actin monomers arranged suitably to nucleate actin polymerization despite being disposed in a conformation distinct from that of actin subunits in the filament (Fig. 1 B). Their model had ABS2 placed in contact with one of the actin molecules, similar to the tropomodulin ABS2 domain (8), whereas the C-terminal region including the WH2 domain was associated with the second actin molecule in the unit cell. The authors noted “an extra LRR domain in the structure (likely resulting from partial degradation of Lmod2 (residues 162–495) during crystallization),” but did not consider it in their nucleation model. They also reported that mutations in the LRR and WH2 domains compromised the ability of leiomodin to stimulate the spontaneous polymerization of actin, most likely by interfering with nucleation.

Boczkowska et al. used the primary data of Chen et al. to recalculate electron density maps and built a new model of the complex of Lmod2 with actin that had better refinement statistics than the model of Chen et al. (1). The electron density maps of Boczkowska et al. lack density for the elements of the Lmod2 molecule that connect the two actin subunits in the Chen et al. model. More importantly, their model includes two Lmod2 ABS2 and WH2 domains, each bound to one of the actin molecules in the unit cell, rather than one Lmod2 molecule connecting two actin molecules (Fig. 1 C). They also note that by using a different but equivalent crystallographic definition of the content of the unit cell, the two independent complexes in the unit cells are related by a local twofold symmetry axis, an arrangement clearly distinct from subunits in the actin filament. The LRR ABS2 domain is associated with actin subdomains 1 and 2 (as shown previously for leiomodin by Chen et al. (1) and Boczkowska et al. (11) and for tropomodulin ABS2 by Rao et al. (8)). The leiomodin WH2 domain binds in the barbed-end groove (also called the target binding site) of each actin molecule, where Chen et al. had modeled a different portion of the Lmod2 sequence that they called helix h1. Boczkowska et al. used biochemical experiments to establish that the leiomodin ABS2 domain has most of the nucleation activity. The WH2 domain augments the nucleation activity of the ABS2 domain, whereas the proposed helix h1 neither binds actin nor appears to contribute to nucleation. This revision of the Chen et al. model reconciles most of the differences between the new crystal structure of Lmod2 and previous work.

Although the homologous ABS2 domains of leiomodin and tropomodulin both interact with three actin subunits, leiomodin stimulates the polymerization of bulk samples of actin monomers, whereas tropomodulin does not (3). In such experiments, the rate-limiting step is the formation of nuclei of actin trimers, which then rapidly elongate (12). Computer simulations of the time course of spontaneous polymerization showed that dimers of actin dissociate at ∼10⁶/s and trimers dissociate a subunit at ∼100/s, so nucleation is very unfavorable (13). The leiomodin ABS2 domain most likely stabilizes actin dimers or trimers better than the tropomodulin ABS2 domain. Slowing the dissociation of dimers or trimers by a small amount can increase bulk polymerization of actin monomers dramatically (14). Calculating the nucleation rate constants in the presence of tropomodulin or leiomodin should explain the differences in their capacities to stimulate the assembly of actin monomers.

Editor: David Thomas.