Leiomodin2 is a polymerase at the pointed ends of actin filaments Authors: Thomas D. Pollard Published: 2026 ABSTRACT Biophysical and structural studies have found that leiomodin 2, a protein discovered in muscle cells, is the first example of a molecule that promotes the elongation of actin filaments at their slowly growing “pointed ends.” This breakthrough will stimulate work to answer many puzzling questions about the mechanism of this process. What started as curiosity about an oft neglected behavior of actin filaments in striated muscle has blossomed into a striking discovery of a mechanism that controls the dynamics of the slow-growing (pointed) end of actin filaments. Littlefield, Almenar-Queralt and Fowler1 discovered the neglected behavior, namely that the pointed ends of actin filaments in living striated muscles can add or lose actin subunits over time. This was a puzzle, because the pointed ends were believed to be capped by the protein tropomodulin discovered by Fowler2. If they were capped, how could these filaments add or dissociate actin subunits? The answer is found in two new papers in Nature Communications, biophysical analysis by Biswas et al.3, and cryo-EM structures by Brotzman et al.4. I will summarize the experiments and discuss the fascinating new questions that arise from the work. Actin filaments can grow and shrink at both ends depending on the concentration of free actin monomers in the solution. The two ends are defined as “barbed” and “pointed” based on the arrowhead pattern created when the heads of myosin bind to a filament. Barbed ends grow faster than pointed ends, and the rate constants for these reactions have been known for 40 years5. In striated muscles, barbed ends are capped by a capping protein and anchored to a structure called the Z-disk6. Pointed ends are capped by tropomodulin, which interacts with both the terminal actin subunits and tropomyosin, an alpha-helical coiled coil that binds to the sides of the actin filament7. Leiomodin2 has domains similar to tropomodulin and is essential for normal heart muscle function7. A knockout mutation in mice results in cardiomyopathy with short actin filaments in the sarcomeres8. Other mutations cause a human cardiomyopathy9. Leiomodin 2 nucleates actin filaments in vitro and in muscle cells10 but other functions were not explored. The laboratory of Shashank Shekhar initiated the new work on leiomodin3. They came prepared with prior experience using real-time observations by total internal reflection microscopy of single actin filaments in microfluidics flow cells to characterize how other proteins influence elongation and disassembly at both ends11,12. Barbed ends are easier to study because they grow and shrink faster than pointed ends, so the quality of their new data on the pointed end is impressive. Their method also allowed imaging of single proteins labeled with contrasting fluorophores associated with either end of a filament. Biophysical breakthrough The key new experiment was to ask how leiomodin 2 influences polymerization at pointed ends. Being related to tropomodulin, leiomodin 2 was expected to cap pointed ends. Surprise, leiomodin increased the rate of elongation at pointed ends3! The key figure plots elongation rates at the pointed end as a function of the actin monomer concentration without and with leiomodin 2 associated with the pointed end (Fig. 1). First, consider the behavior of free pointed ends. The X-intercept at 0.7 µM actin monomers is the “critical concentration” where association and dissociation rates are equal. The slope of the line gives the association rate constant of 0.8 subunits µM−1s−1 as measured previously5. The Y-intercept gives the rate constant for the dissociation of actin subunits of 0.5 s−1.Fig. 1Dependence of elongation rates of pointed ends (PEs) on the concentration actin monomers (G-actin) with and without leiomodin 2.Reproduced from Fig. 1G (Biswas et al. 2026)3. Elongation rates (mean ± SD) of free (black) and Lmod2-bound (red) PEs over a range of actin monomer concentrations. Dependence of elongation rates of pointed ends (PEs) on the concentration actin monomers (G-actin) with and without leiomodin 2. Reproduced from Fig. 1G (Biswas et al. 2026)3. Elongation rates (mean ± SD) of free (black) and Lmod2-bound (red) PEs over a range of actin monomer concentrations. With leiomodin 2 on the pointed end, all aspects of elongation are different. First, leiomodin 2 inhibits subunit dissociation from pointed ends by reducing the dissociation rate constant and the critical concentration to zero. Second, pointed ends elongate at rates proportional to low concentrations of actin monomers in the buffer, but the elongation rate plateaus at 1.6 subunits s−1 at actin monomer concentrations above 2 µM. Direct observation confirmed that leiomodin 2 remains bound processively to growing pointed ends. This two-phase dependence on substrate concentration is typical of many enzyme reactions, where the rate of the reaction is limited by substrate binding at low substrate concentrations and by a second, coupled reaction (typically a reversible conformational change) at high substrate concentrations. In the simplest case, the rate on the plateau is equal to the sum of the first-order rate constants for the forward and reverse reactions. One question discussed at the end of this comment is the nature of these reactions, which determine the rate of actin subunit association and dissociation at the pointed end. The pointed end of a filament remains attached to leiomodin 2 monomers (confirmed by one-step photobleaching) bound to the coverslip as they elongate processively at 1.3 subunits s−1 with 2 µM actin monomers in the buffer. These pointed ends dissociate stochastically from immobilized leiomodin 2 with a half-time of ~ 8 min after the addition of hundreds of subunits. Most actin monomers in the cytoplasm are bound to profilin, which blocks association with the pointed ends of actin filaments, so another surprise is that profilin-actin elongates pointed ends associated with leiomodin 2. Without actin monomers, profilin promotes dissociation of subunits from pointed ends with bound leiomodin 2. Structural breakthrough The new paper from the laboratory of Roberto Dominguez has 4 cryo-EM structures of actin filament pointed ends, including ones with bound leiomodin 2 and/or incoming actin subunits4. They offer many insights about the mechanism of processive elongation. Their samples consisted of short actin filaments assembled from a mixture of actin monomers, capping protein, full-length leiomodin 2 and tropomyosin. This produced mixtures of short filaments capped on their barbed ends with capping protein and a mixture of pointed ends associated with none, one or two leiomodin 2 s and none or one incoming actin subunits (Fig. 2).Fig. 2Space-filling models of cryo-EM reconstructions of the pointed ends of actin filaments associated with leiomodin 2 LRR domains and a newly added actin subunit.Adapted from figure 3 (Brotzman et al. 2026)4. The letters (A–D) at the top identify four structures proposed to be intermediates in the elongation pathway. The arrows indicate four reactions proposed to link these structures. The fourth reaction releases the first LRR domain, giving a structure similar to B but with one more actin subunit. The red circle in C highlights the DBS of the LRR binding in the barbed end groove of newly added actin subunit P + 1 in place of the D-loop of actin subunit P-1. Space-filling models of cryo-EM reconstructions of the pointed ends of actin filaments associated with leiomodin 2 LRR domains and a newly added actin subunit. Adapted from figure 3 (Brotzman et al. 2026)4. The letters (A–D) at the top identify four structures proposed to be intermediates in the elongation pathway. The arrows indicate four reactions proposed to link these structures. The fourth reaction releases the first LRR domain, giving a structure similar to B but with one more actin subunit. The red circle in C highlights the DBS of the LRR binding in the barbed end groove of newly added actin subunit P + 1 in place of the D-loop of actin subunit P-1. One of the glories of cryo-EM is that computational methods called classification can separate images of different structures in the same micrograph, so each structure can be reconstructed separately from thousands of images. In addition to a reconstruction of a bare pointed end Fig. 2A, (similar to previous cryo-EM structures), three reconstructions have density for part of leiomodin 2, the leucine-rich repeat (LRR) domain (residues 186–370), including the D-loop binding site (DBS). None of the reconstructions has density for tropomyosin or either the disordered N- and C-terminal parts of the full-length leiomodin 2. Figure 2 shows structures of pointed ends with (B) a single bound leiomodin 2, (C) with a single leiomodin 2 plus an added actin subunit, and (D) with two leiomodin 2 LRRs and the added actin subunit. A reconstruction of the middle of these short ADP-actin filaments is the highest resolution (2.0 Å) achieved to date. The paper describes, and Supplementary Movie 1 from Brotzman et al.4 visualizes functionally important features in each of these reconstructions. The authors propose that the three structures with leiomodin 2 on the pointed end are intermediates along the reaction pathway that incorporates actin subunits to elongate the filament. The proposed mechanism consists of three reactions. In reaction 1 one leiomodin 2 LRR domain binds to a bare pointed end (A), interacting with actin subunits P, P-1 and P-2 to form intermediate B. Conformational changes in subdomain 2 of subunits P and P-1 are required to open a binding site for the LRR domain. Brotzman et al.4 suggest that leiomodin 2 displaces SD2, an induced fit mechanism. An alternative hypothesis is that rapid thermal fluctuations of the terminal subunits13 spontaneously open the sterically hindered binding site frequently enough to allow LRR to bind. Once bound to the filament, the LRR domain stabilizes the high-energy conformation captured in the frozen sample (structure B). In the core of filaments, thermal motions are more restricted13, explaining why the leiomodin 2 LRR does not bind to the side of filaments. A new actin subunit binds to a bare pointed end by capturing in its barbed end groove the D-loop of actin subunit P-1, but this is impossible in reaction 2 when the leiomodin 2 LRR is bound to the pointed end (B); subdomain 2 of actin subunit P-1 is rotated laterally, and the DBS displaces the D-loop. Consequently, the incoming subunit P + 1 binds to the DBS rather than the D-loop of actin subunit P-1 (see detail in Fig. 2C). Despite completely different interactions, the association rate constants are similar with and without leiomodin 2 (Fig. 1). This is remarkable, because association of subunits at the pointed end is limited by conformational changes rather than being diffusion-limited like the barbed end. The third reaction is the binding of a second leiomodin 2 LRR domain to intermediate structure C, where it interacts with four actin subunits, P-2, P-1, P and P + 1 on the opposite side of the filament from the first leiomodin 2. The conformation of bound subunit P + 1 resembles subunit P in intermediate C. In reaction 4, the first (trailing) leiomodin 2 LRR domain dissociates from the pointed end. This reaction has not been characterized, but Brotzman et al. 4 note that addition of subunits to the end of the filament drives progressive flattening of internal subunits as in free pointed ends. This flattening reverses the conformational changes required for leiomodin 2 to bind in reaction 1 and displaces the LRR domain. Dissociation of the LRR domain presumably allows the D-loop of subunit P-1 to bind in the barbed end groove of P + 1, although the details are not known. These changes return the pointed end to the structure of the B intermediate, ready for a new cycle of elongation. The four elegant structures in Fig. 2 are intermediates along the assembly pathway, but additional transition states likely exist to account for the conformational changes associated with each of these binding reactions. Freezing actin filaments for cryo-EM captures low entropy conformations13 at the expense of higher entropy conformations that may include transient intermediates. Computational methods may reveal these intermediates and explain the processes, such as the conformational changes required for the LRR domain to bind to terminal actin subunits. Questions arising from the new papers The two new papers in Nature Communications establish the leiomodin 2 not only nucleates actin polymerization10 but also promotes elongation of pointed ends, even when the actin monomers are saturated with profilin. The papers also provide clues about the mechanism of pointed end elongation by leiomodin 2, but raise many questions that will drive future research. One unexplained observation is that low concentrations of actin monomers bind to pointed ends associated with leiomodin 2 and free pointed ends with similar rate constants (slopes in Fig. 1). This is surprising, because the binding site for subunit P + 1 is so different when the LRR is bound and since this association reaction on free pointed ends is not limited by diffusion14, but rather by rearrangements of the binding site. The ability of leiomodin 2 to reduce the rate of subunit dissociation from pointed ends to zero is easier to understand. The extra interactions of the terminal subunit with the leiomodin 2 LRR domain are likely to strengthen its association with its neighbors. Further work is required to compare the strength of the longitudinal bonds formed by the DBS and the P-loop. The rate of elongation of free pointed ends increases without bound with the concentration of actin monomers, while the rate with Lmod plateaus at 1.6 subunits s-1 (Fig. 1). A conventional explanation for this type of kinetics is that a binding reaction limits the overall rate at low concentrations of monomers, while slower linked reactions limit the rate on the plateau. Typically, these plateau reactions are pairs of reversible first-order reactions, such as conformational changes, and the plateau rate is the sum of the rate constants for the forward and back reactions. Since the plateau rate is insensitive to the actin monomer concentration, reaction 2 is an unlikely candidate. On the other hand, both reactions 1 and 3 involve not only the binding of a leiomodin 2 LRR domain but also reversible, first-order conformational changes, which can be characterized to understand the pointed end elongation mechanism. The ability of actin-profilin to elongate pointed ends associated with leiomodin 2 is perplexing. Profilin binds to the barbed end of the actin monomer, blocking the barbed end groove from binding to the P-1 subunit at the end of the filament. Bound profilin would also preclude interactions of an incoming actin monomer with an LRR domain bound to the pointed end (Fig. 2C). This steric interference is not an issue when actin-profilin binds to the barbed end of a filament on its own or when assisted by a formin. Another conundrum is that pointed ends of actin filaments remain attached to leiomodin 2 monomers immobilized on glass through hundreds of cycles of actin subunit addition before dissociating, while the polymerization cycle in Fig. 2 includes coordinated interactions of two leiomodin 2 molecules and obligatory dissociation of the trailing LRR domain during reaction 4. An alternative to full dissociation and rebinding to terminal actin subunits is that the C-terminal extension of leiomodin 2 binds to the side of the filament near the tip15. However, Biswas et al. 3 did not observe full-length leiomodin 2 binding along the middle of the filaments. Insights might be provided by measurements of the rate constants for leiomodin 2 binding to and dissociating from filament ends and the concentration of leiomodin 2 in the cytoplasm. Comparisons with other actin polymerases What is known about pointed end elongation with leiomodin 2 appears to be strikingly different from the processive barbed end polymerases, formins and vasodilator-stimulated protein (VASP). Leiomodin 2 is a monomer, formins are dimers. Leiomodin 2 binds to only the flexible conformations of actin subunits at the pointed end, while formins favor barbed end subunits, and some formins also bind to internal subunits. Formins are processive because they step reliably onto each new subunit added to a barbed end, while incorporation of each actin at a pointed end is proposed to require the association and disassociation of one leiomodin 2. All characterized formins slow elongation of barbed ends (ranging from little to profoundly), while Lmod does not inhibit elongation of pointed ends. Both formins and leiomodin 2 can add actin-profilin to filament ends, and both have polyproline sequences that bind profilin and actin-profilin. The polyproline sites in the C-terminal domain of leiomodin 2 are not required for elongation3, whereas the polyproline sequences in the flexible FH1 domains of formins are essential to concentrate actin-profilin for rapid delivery to the barbed end at rates faster than the diffusion-limited rate of free actin monomers. The barbed end polymerase, VASP, shares features with both leiomodin 2 and formins. Like leiomodin 2, VASP tetramers bind to actin subunits near the filament end (barbed end in the case of VASP), and like formins, VASP delivers actin-profilin from polyproline sites to the filament end. In closing The papers by Biswas et al. and Brotzman et al. begin a story that will unfold over the years as experiments address the mechanistic questions raised above and other questions. Addressing equally interesting questions at the cellular level will be empowered by the mechanistic insights. I am prepared for surprises.