Structural insights into the tropomodulin assembly at the pointed ends of actin filaments

Abstract

Tropomodulins are a family of important regulators of actin dynamics at the pointed ends of actin filaments. Four isoforms of tropomodulin, Tmod1‐Tmod4, are expressed in vertebrates. Binding of tropomodulin to the pointed end is dependent on tropomyosin, an actin binding protein that itself is represented in mammals by up to 40 isoforms. The understanding of the regulatory role of the tropomodulin/tropomyosin molecular diversity has been limited due to the lack of a three‐dimensional structure of the tropomodulin/tropomyosin complex. In this study, we mapped tropomyosin residues interacting with two tropomyosin‐binding sites of tropomodulin and generated a three‐dimensional model of the tropomodulin/tropomyosin complex for each of these sites. The models were refined by molecular dynamics simulations and validated via building a self‐consistent three‐dimensional model of tropomodulin assembly at the pointed end. The model of the pointed‐end Tmod assembly offers new insights in how Tmod binding ensures tight control over the pointed end dynamics.

Abbreviations and symbols

3D

three‐dimensional

ABS

actin‐binding site

ABS1

actin‐binding site 1

ABS2

actin‐binding site 2

CD

circular dichroism

HSQC

heteronuclear single quantum coherence

Lmod

leiomodin

Lmod2

leiomodin 2

Lmod2‐TpmBS1

tropomyosin‐binding site 1 of leiomodin 2

MDS

molecular dynamics simulations

NMR

nuclear magnetic resonance

T‐αL

transient α‐helical loop

T_m

melting temperature

Tmod

tropomodulin

Tmod2

tropomodulin 2

Tmod2‐TpmBS1

tropomyosin‐binding site 1 of tropomodulin 2

Tmod2‐TpmBS2

tropomyosin‐binding site 2 of tropomodulin 2

Tpm

tropomyosin

TpmBS

tropomyosin‐binding site

TpmBS1

tropomyosin‐binding site 1

TpmBS2

tropomyosin‐binding site 2

Tpm1‐1b(1‐X)Zip

tropomyosin fragment 1‐1b containing tropomyosin residues 1‐X fused at the C‐terminus with residues of the GCN4 leucine‐zipper domain

1. INTRODUCTION

The processes of actin polymerization and depolymerization have fundamental roles in the organization of the cytoskeleton, in cell morphology, and in motility. Polymerized filamentous actin (F‐actin) is polar, and dissimilar ends of the actin filament are called “pointed” and “barbed.” Intrinsic kinetic rate constants of actin polymerization and depolymerization are greater at the barbed (also known as “fast growing”) end of the filament, than at the pointed (“slow growing”) end. ¹ In vitro, a spontaneously growing actin filament eventually reaches a steady‐state phase, characterized by an unchanged filament length. In the steady state, the monomeric actin molecules (G‐actin) attaching to the barbed end are supplied by depolymerization of the pointed end in a process known as actin treadmilling. During treadmilling, the pointed‐end release of actin monomers is a rate‐limiting step for actin polymerization at the barbed end. ² , ³

In cells, the growth and maintenance of actin filaments are tightly regulated by a variety of actin‐binding proteins. ⁴ Tropomodulin (Tmod) is an actin‐binding protein that has a central regulatory role in actin filament dynamics at the pointed end. Multiple functions of actin filaments in different cell types led to evolution of four Tmod isoforms in vertebrates (reviewed in Reference 5), that is, tropomodulins 1–4. Tmod isoforms perform their regulatory functions by capping the pointed end and blocking pointed‐end elongation or depolymerization. ⁶ , ⁷ , ⁸ The capping of actin filaments by Tmod is strongly dependent on Tmod interaction with tropomyosin (Tpm). ⁶ Tpm is a coiled‐coil dimeric protein, which binds to opposite sides of the actin filament. ⁹ Tmod interacts with Tpm via two Tpm‐binding sites (TpmBS), TpmBS1 and TpmBS2, ¹⁰ in a clamp‐like fashion, where each TpmBS of one Tmod molecule binds to one of two Tpm N‐terminal regions on the opposite sides of the actin filament. ¹¹

With four Tmod isoforms and up to ~40 Tpm isoforms, ¹² the numerous combinations of complexes between them makes regulation of actin dynamics by Tmods very complex. For instance, three Tmod isoforms, that is, Tmod1‐Tmod3, and at least 16 isoforms of Tpm are expressed in the mammalian central nervous system. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ They can potentially produce 48 distinct filament‐capping Tmod/Tpm complexes. Among different Tmod/Tpm isoform pairs, the binding affinity of complexes formed between either TpmBS and Tpm ranges from very high (with the dissociation constant K _d being in nM range) to virtually undetectable. ¹⁷ This variability is a consequence of a relatively small number of amino acid substitutions in otherwise homologous binding sites in Tmods and Tpms. It is hypothesized to be the basis for the tissue‐specific regulation of pointed‐end capping ¹⁸ and exemplified by Tmod isoform‐specific modulation of dendrite development in neurons. ¹⁹

So far, the potential for comprehensive analysis, understanding, and predictions of the regulatory functions exerted by different Tmod and Tpm isoforms has been severely limited by the lack of three‐dimensional (3D) structural information on the binding interfaces between either TpmBS and Tpm. In this study, we gained structural insight into Tmod/Tpm binding by studying interactions between Tmod2 (the brain‐specific Tmod isoform ¹⁴ ) and nonmuscle Tpm isoforms, Tpm1.8/1.9/1.11/1.12, in which the N‐terminal region is encoded by the 1b exon from the Tpm1 gene. ¹² , ²⁰ We chose these isoforms because they are expressed in neurons, ²¹ , ²² , ²³ , ²⁴ and therefore represent a subset of Tpms involved in little‐researched Tmod/Tpm isoform‐dependent regulation of neural morphology. ²⁵ We identified Tpm N‐terminal residues that interact with TpmBS1 and TpmBS2 from Tmod2. Mapping the binding residues allowed us to model 3D structures of molecular complexes of TpmBS1 and TpmBS2 with Tpm. These models were tested for in silico stability and refined by molecular dynamics simulations (MDS). Additionally, we utilized the refined structures as a basis to propose a new 3D model for the Tmod assembly at the pointed end.

2. RESULTS

2.1. Mapping of the Tmod/Tpm binding interface

2.1.1. Design of Tmod2 and model Tpm peptides representing the Tmod/Tpm binding interface

Previous studies have shown that residues 1–14 in long Tpm isoforms and 1–19 in short Tpm isoforms bind to Tmod2‐TpmBS1 (which represents TpmBS1 of Tmod2, Figure 1a) and Tmod2‐TpmBS2 (which represents TpmBS2 of Tmod2, Figure 1a). ¹⁷ , ²⁶ However, it was not clear how many Tpm residues in total interact with each TpmBS of Tmod2. To map these residues, we created a set of Tpm model peptides with a general name of Tpm1‐1b(1‐X)Zip. The set included four peptides containing X number of N‐terminal residues of short α‐tropomyosins Tpm1.8/1.9/1.11/1.12, which share at the N‐terminus the exon 1b of the Tpm1 gene. ¹² , ²⁰ The X N‐terminal residues were fused at the C‐terminus with a “GCN4 sequence,” which is a part of the 33‐residue leucine‐zipper domain of the yeast transcription factor GCN4. ²⁷ X adopted values of 19, 26, 33, and 43 (Figure 1b). Tpm1‐1b(1‐26)Zip and Tpm1‐1b(1‐33)Zip were longer than Tpm1‐1b(1‐19)Zip by one and two consensus heptad (seven‐residue) repeats, respectively. The repeats are known to be the basic unit of stable coiled coils. ²⁷ Tpm1‐1b(1‐43)Zip included 43 residues encoded by the entire exon 1b.

FIGURE 1.

Amino acid sequences of peptides used in the study. (a) The sequences of Tmod2 Tpm‐binding sites, Tmod2‐TpmBS1 and Tmod2‐TpmBS2, shown within the context of schematic Tmod domain organization. TpmBS1, tropomyosin‐binding site 1; TpmBS2, tropomyosin‐binding site 2; ABS1, actin‐binding site 1; ABS2, actin‐binding site 2. (b) The sequences of Tpm1‐1b(1‐X)Zip model peptides. Each peptide contains X N‐terminal residues of the low molecular weight (short) Tpm isoforms Tpm1.8/1.9/1.11/1.12, followed by a partial GCN4 leucine zipper sequence. Residues belonging to the GCN4 sequence are underlined; X adopts values 19, 26, 33 and 43. Letters a and d indicate corresponding residue positions in the consensus coiled‐coil abcdegf heptad repeat. Glycine is added at the N‐terminus of the Tpm1‐1b(1‐X)Zip sequences to mimic native acetylation. Extra four GCN4 residues LLSK in the Tpm1‐1b(1‐43)Zip peptide are introduced to maintain an uninterrupted coiled‐coil heptad repeat register. (c) 3D crystal structure of the Tpm1‐1b(1‐19)Zip peptide (PDB ID: 3AZD). ⁵⁰ Residues in a and d positions are shown in cyan and magenta, respectively. (d) A helical wheel schematic of a coiled coil showing hydrophobic interactions between a (cyan) and d (magenta) residues

The GCN4 sequence served to facilitate formation of a coiled coil typical of a full‐length Tpm and it contained 18 residues for X = 19, 26, 33, and 22 residues for X = 43. The variation of the GCN4 sequence length was necessary to ensure that all Tpm model peptides consisted of uninterrupted coiled‐coil heptad repeats. Additionally, the Tpm peptides included an N‐terminal Gly residue mimicking native N‐acetylation of Tpm. Inclusion of the N‐terminal Gly and the C‐terminal GCN4 sequence was shown in several studies to mimic well the structure and binding properties of the N‐terminal region of Tpm. ²⁶ , ²⁸ , ²⁹ , ³⁰

2.1.2. Structural and stability characterization of purified Tpm peptides and their complexes with Tmod2 peptides

Specific binding of Tmods requires that two chains of each designed Tpm peptide dimerize and fold into a coiled coil. To confirm formation of the coiled coil and to estimate the degree of folding of the Tpm peptides at different temperatures, we performed thermal unfolding experiments using circular dichroism (CD) spectroscopy. Thermal unfolding experiments subject a protein to a gradual increase in temperature, thus inducing the protein to transition from a folded (native) to an unfolded state. For a two‐state unfolding process, the temperature at the midpoint of the transition, where the ratio of folded to unfolded states is equal to 1, is called the melting temperature (T_m) of the protein. Under certain conditions, and particularly for homologous proteins, the T_m values can be used to compare stabilities of proteins. ³¹ , ³² Of four designed Tpm peptides, Tpm1‐1b(1‐43)Zip forms the most stable coiled coil with a T_m value of 53°C (Figure 2), whereas Tpm1‐1b(1‐19)Zip was second in stability, with a T_m value of 45°C. T_m values for two other Tpm peptides, Tpm1‐1b(1‐26)Zip, and Tpm1‐1b(1‐33)Zip, were 28 and 30°C, respectively, demonstrating that they are the least stable in the series (Table 1). In comparison, T_m of the full‐length 33‐residue GCN4 leucine‐zipper domain (also known as GCN4‐p1) was reported to be 64°C. ³³ For convenience of description, we will henceforth calculate molar ratios between Tmod and Tpm peptides with respect to the molar amount of the Tpm dimers.

FIGURE 2.

Circular dichroism (CD) temperature unfolding curves of the Tpm1‐1b(1‐X)Zip peptides. For each Tpm peptide, temperature dependence of mean residue ellipticity at 222 nm was measured in 100 mM NaCl, 10 mM Na‐phosphate, pH 7.0. Arrows indicate the midpoints of transition for Tpm1‐1b(1‐43)Zip (53°C, filled circles), Tpm1‐1b(1‐33)Zip (30°C, open circles), Tpm1‐1b(1‐26)Zip (28°C, filled triangles), and Tpm1‐1b(1‐19)Zip (45°C, open triangles)

TABLE 1.

Binding of Tpm and Tmod2 peptides. Tmod2‐TpmBS1 and Tmod2‐TpmBS2 are Tmod2 fragments containing first and second Tpm‐binding sites correspondingly. Melting temperature (T_m) is a midpoint of an unfolding transition. ⁵⁹

Peptide or complex	Tm (°C)
Tpm1‐1b(1‐19)Zip	44.7 ± 0.8
Tpm1‐1b(1‐26)Zip	28.0 ± 0.8
Tpm1‐1b(1‐33)Zip	30.1 ± 1.2
Tpm1‐1b(1‐43)Zip	52.6 ± 0.2
Tmod2‐TpmBS1/Tpm1‐1b(1‐19)Zip	54.7 ± 0.7 a
Tmod2‐TpmBS2/Tpm1‐1b(1‐19)Zip	54.0 ± 1.7 a
Tmod2‐TpmBS1/Tpm1‐1b(1‐26)Zip	~39.9
Tmod2‐TpmBS2/Tpm1‐1b(1‐26)Zip	39.5 ± 0.7
Tmod2‐TpmBS1/Tpm1‐1b(1‐33)Zip	38.92 ± 0.01
Tmod2‐TpmBS2/Tpm1‐1b(1‐33)Zip	37.7 ± 1.1
Tmod2‐TpmBS1/Tpm1‐1b(1‐43)Zip	57.5 ± 0.4
Tmod2‐TpmBS2/Tpm1‐1b(1‐43)Zip	58.7 ± 0.2

^a See Reference 17.

The finding that stabilities of the Tpm1‐1b(1‐X)Zip peptide do not directly correlate with its length could be anticipated. Canonical coiled coils are formed by heptad repeats stabilized by hydrophobic residues in the first (a) and the fourth (d) positions, which form a hydrophobic core of the coiled coil. The T_m of the designed Tpm peptides is dependent on the hydrophobicity of residues in a and d positions. These positions in the first 19 residues of Tpm are occupied by large hydrophobic Val, Leu, and Ile residues (Figure 1), known to promote Tpm stability. ³⁴ , ³⁵ On the other hand, Ala residues in a and d positions destabilize Tpm coiled coils, ³⁴ , ³⁵ , ³⁶ providing Tpm with patches of local flexibility required for its function. ³⁷ , ³⁸ Elongation of the Tpm sequence in the model peptides to 26 and 33 residues puts one and two Ala residues, respectively, in the a and d positions (Figure 1b), which is expected to destabilize Tpm1‐1b(1‐26)Zip and Tpm1‐1b(1‐33)Zip.

Upon adding either Tmod2‐TpmBS1 or Tmod2‐TpmBS2 to Tpm peptides, we observed further stabilization of the complexes. We performed thermal unfolding experiments for the samples containing Tpm and Tmod2 peptides at stoichiometric ratios (1:1) (Table 1, Figure 3). The increase in T_m for the Tmod2/Tpm is a result of complex formation, as indicated by comparison of melting curves for the mixed peptides and arithmetic sums of melting curves for individual peptides. Increases in T_m upon addition of Tmod2‐TpmBS1 or Tmod2‐TpmBS2 are comparable, suggesting that they have similar affinities for the N‐terminal Tpm region shared by Tpm1.8/1.9/1.11/1.12 (Table 1).

FIGURE 3.

Binding of Tmod2‐TpmBS1 (a) and Tmod2‐TpmBS2 (b) to Tpm1‐1b(1‐43)Zip. Representative temperature unfolding curves of the individual peptides (Tpm1‐1b(1‐43)Zip (filled circles), Tmod2‐TpmBS1 and Tmod2‐TpmBS2 (open circles)) and their respective complexes (Tmod2‐TpmBS1/Tpm1‐1b(1‐43)Zip and Tmod2‐TpmBS2/Tpm1‐1b(1‐43)Zip (filled triangles)) were recorded in 100 mM NaCl, 10 mM Na‐Phosphate, pH 7.0 at 222 nm. Open triangles show the arithmetic sums of individual unfolding curves (Tmod2‐TpmBS1 + Tpm1‐1b(1‐43)Zip and Tmod2‐TpmBS2 + Tpm1‐1b(1‐43)Zip)

2.1.3. Tpm region interacting with Tmod2‐TpmBS1 or Tmod2‐TpmBS2 is localized within the first 26 N‐terminal residues

We recorded and examined 2D ¹⁵N‐HSQC nuclear magnetic resonance (NMR) spectra of free ¹⁵N‐labeled Tmod2‐TpmBS1 and Tmod2‐TpmBS2 peptides and of their complexes with the Tpm peptides (Figure 4). For the free peptides, we observed a comparatively low chemical shift dispersion (~1 ppm) of backbone amide ¹H resonances (Figure 4, black spectra). This is typical of unfolded or only partly folded proteins, ³⁹ and consistent with the disordered nature of Tmod TpmBS peptides. ¹⁷ When a stoichiometric excess of the unlabeled Tpm1‐1b(1‐19)Zip was added to the Tmod peptides ([Tpm1‐1b(1‐19)Zip]/[Tmod] > 1), the peak dispersion increased to >2 ppm indicating formation of well‐folded molecular complexes (Figure 4a,b, blue spectra).

FIGURE 4.

¹⁵N‐HSQC spectra of complexes between ¹⁵N‐labeled Tmod2 TpmBSs and unlabeled Tpm1‐1b(1‐X)Zip. The panels display spectra of free ¹⁵N‐labeled Tmod2‐TpmBS1 (black spectra in Panels a,c,e,g) or free ¹⁵N‐labeled Tmod2‐TpmBS2 (black spectra in Panels b,d,f,h) superimposed with spectra of their complexes with Tpm1‐1b(1‐X)Zip. The spectra of labeled‐peptide complexes with unlabeled Tpm1‐1b(1‐19)Zip, Tpm1‐1b(1‐26)Zip, Tpm1‐1b(1‐33)Zip, and Tpm1‐1b(1‐43)Zip are color coded in blue, red, cyan, and green, respectively. More specifically, Panel (a): free Tmod2‐TpmBS1 (black) and Tmod2‐TpmBS1/Tpm1‐1b(1‐19)Zip (blue); Panel (b): free Tmod2‐TpmBS2 (black) and Tmod2‐TpmBS2/Tpm1‐1b(1‐19)Zip (blue); Panel (c): free Tmod2‐TpmBS1 (black) and Tmod2‐TpmBS1/Tpm1‐1b(1‐26)Zip (red); Panel (d): free Tmod2‐TpmBS2 (black) and Tmod2‐TpmBS2/Tpm1‐1b(1‐26)Zip (red); Panel (e): free Tmod2‐TpmBS1 (black) and Tmod2‐TpmBS1/Tpm1‐1b(1‐33)Zip (cyan); Panel (f): free Tmod2‐TpmBS2 (black) and Tmod2‐TpmBS2/Tpm1‐1b(1‐33)Zip (cyan); Panel (g): free Tmod2‐TpmBS1 (black) and Tmod2‐TpmBS1/Tpm1‐1b(1‐43)Zip (green); Panel (h): free Tmod2‐TpmBS2 (black) and Tmod2‐TpmBS2/Tpm1‐1b(1‐43)Zip (green)

When a seven‐residue longer Tpm1‐1b(1‐26)Zip peptide was used as a binding partner instead of Tpm1‐1b(1‐19)Zip, the dispersion of amide ¹H resonances in the Tmod spectra was also >2 ppm (Figure 4c,d, red spectra) and similar to that with Tpm1‐1b(1‐19)Zip. Nevertheless, for either Tmod2‐TpmBS1 or Tmod2‐TpmBS2, marked differences were observed between the spectra of their complexes with Tpm1‐1b(1‐19)Zip and Tpm1‐1b(1‐26)Zip (Figure 5a,b). This shows that, on one hand, in each of the four complexes, whether with shorter Tpm1‐1b(1‐19)Zip or with longer Tpm1‐1b(1‐26)Zip, the Tmod2‐TpmBS1 and Tmod2‐TpmBS2 peptides are well‐folded. On the other hand, additional 7 residues from Tpm affected the Tmod2‐TpmBS1 and Tmod2‐TpmBS2 spectra, either by proximity to or by direct contacts with the Tmod2 peptides.

FIGURE 5.

Changes in ¹⁵N‐HSQC spectra of Tmod2‐TpmBS1/Tpm1‐1b(1‐X)Zip and Tmod2‐TpmBS2/Tpm1‐1b(1‐X)Zip complexes observed with the increase in the number (X) of tropomyosin residues. The panels display superimposed spectra of complexes between ¹⁵N‐labeled Tmod2‐TpmBS1 (Panels a,c) or ¹⁵N‐labeled Tmod2‐TpmBS2 (Panels b,d) and unlabeled Tpm1‐1b(1‐X)Zip. The spectra of labeled‐peptide complexes with Tpm1‐1b(1‐19)Zip, Tpm1‐1b(1‐26)Zip, Tpm1‐1b(1‐33)Zip, and Tpm1‐1b(1‐43)Zip are color coded in blue, red, cyan, and green, respectively. Panel (a): Tmod2‐TpmBS1/Tpm1‐1b(1‐19)Zip (blue) and Tmod2‐TpmBS1/Tpm1‐1b(1‐26)Zip (red); Panel (b): Tmod2‐TpmBS2/Tpm1‐1b(1‐19)Zip (blue) and Tmod2‐TpmBS2/Tpm1‐1b(1‐26)Zip (red); Panel (c): Tmod2‐TpmBS1/Tpm1‐1b(1‐26)Zip (red), Tmod2‐TpmBS1/Tpm1‐1b(1‐33)Zip (cyan), and Tmod2‐TpmBS1/Tpm1‐1b(1‐43)Zip (green); Panel (d): Tmod2‐TpmBS2/Tpm1‐1b(1‐26)Zip (red), Tmod2‐TpmBS2/Tpm1‐1b(1‐33)Zip (cyan), and Tmod2‐TpmBS2/Tpm1‐1b(1‐43)Zip (green)

Both ¹⁵N‐labeled Tmod2‐TpmBS1 and Tmod2‐TpmBS2 display almost indistinguishable ¹⁵N‐HSQC spectra when they interact with Tpm1‐1b(1‐26)Zip, Tpm1‐1b(1‐33)Zip or Tpm1‐1b(1‐43)Zip (Figure 5c,d). We can conclude that the Tpm‐binding site for both Tmod peptides lies within the first 26 residues of Tpm1.8/1.9/1.11/1.12. The first 19 residues of the nonmuscle Tpms provide enough contacts to cause binding‐induced folding of both Tmod2‐TpmBS1 and Tmod2‐TpmBS2.

The analysis of the CD melting curves and recalculation of equilibrium concentrations to the conditions of NMR experiments estimated free Tmod peptide populations in all NMR samples as <1–4% (see Section 4.5). Therefore, we considered contributions of potential chemical shift averaging due to chemical exchange between free and bound Tmod peptides negligible, and the observed spectral differences between the complexes as being primarily caused by the intrinsic variation of the binding interface.

2.2. A model for Tmod assembly at the pointed end of the actin filament

2.2.1. Secondary structure predictions in both Tmod2‐TpmBS1 and Tmod2‐TpmBS2 suggest formation of two α‐helices separated by a two‐residue turn

We used three popular secondary structure predictors, Jpred4, PsiPred, and PredictProtein, to predict folded regions of Tmod2‐TpmBS1 and Tmod2‐TpmBS2 (Figure 6). To validate these results, we also predicted the secondary structure of Lmod2‐TpmBS1, a homologous TpmBS of another member of the Tmod family of proteins, leiomodin 2 (Lmod2). We compared the prediction with the secondary structure of Lmod2‐TpmBS1 in complex with the N‐terminus of Tpm1.1, which we have recently determined by NMR ⁴⁰ (PDB ID 6UT2).

FIGURE 6.

Secondary structure predictions for Lmod2‐TpmBS1, Tmod2‐TpmBS1, and Tmod2‐TpmBS2. Amino acid residues shown to be in a helix by NMR ⁴⁰ are highlighted in cyan. The helix denoted as T‐αL is a transient helix, while two others, Helix 1 and Helix 2, are stable α‐helices. Amino acid residues predicted to be in a helix by three sequence‐based secondary structure predictors are highlighted in green

The prediction of α‐helices 1 and 2 in Lmod2‐TpmBS1 was consistent with our structural results, ⁴⁰ where we reported that the two helices form an α‐helical hairpin arranged with the Tpm coiled coil in a four α‐helix bundle. Across the three prediction algorithms, there was little variability in residues assigned to the two helices (Figure 6). Notably, no secondary structure prediction was made for residues Leu24 and Ser25 of Lmod2‐TpmBS1. In the determined structure of Lmod2‐TpmBS1 these residues represent a ββ‐turn connecting helices 1 and 2, and the N‐terminal residues Ser2‐Ile16 were shown to be mostly disordered. However, residues Lys12‐Glu14 appeared to form a transient α‐helical loop (T‐αL) that may also interact with Tpm. In agreement with the transient nature of the loop, the algorithms displayed a high degree of variability in identifying this loop, ranging from predicting that only one residue (Lys12) is in a α‐helical conformation (Jpred4) to that as many as eight residues (Tyr6‐Tyr13) form an α‐helix (PredictProtein).

Predictions for Tmod2‐TpmBS1 followed the same pattern, which was not surprising considering the high level of homology between Tmod2‐TpmBS1 and Lmod2‐TpmBS1. The 19 N‐terminal residues of Tpm1.8/1.9/1.11/1.12 include a low complexity 5‐residue N‐terminal region AGSSS (Figure 1) that is absent in Tpm1.1. The remaining 14 residues are homologous to the 14 N‐terminal residues of Tpm1.1. Therefore, the residue mapping results for Lmod2‐TpmBS1 ⁴¹ and Tmod2‐TpmBS1 (this study) are also in agreement with each other; and we can assume that both complexes (Lmod2‐TpmBS1/Tpm1.1 and Tmod2‐TpmBS1/Tpm1.8) are structurally homologous.

Although the NMR results obtained for Tmod2‐TpmBS1 could be anticipated, the results obtained for Tmod2‐TpmBS2 were surprising. Until now, it has been assumed that TpmBS2 of Tmods form a single continuous α‐helix in the complex with Tpms. ¹¹ Most of the residues in Tmod2‐TpmBS2 starting with Pro117 and ending with Leu144 were predicted to be in a α‐helical conformation (Figure 6), and consequently we expected the number of Tpm1.8/1.9/1.11/1.12 residues forming the entire interface to be at least 33, that is, at least seven residues greater than what we observed in NMR experiments.

We resolved the conflict between the expected and observed size of the Tpm1.8/1.9/1.11/1.12 region required to bind Tmod2‐TpmBS2 by aligning sequences and predicted secondary structure elements of Lmod2‐TpmBS1, Tmod2‐TpmBS1 and Tmod2‐TpmBS2. Secondary structure predictions revealed the presence of up to three residues with unassigned conformation in the middle of Tmod2‐TpmBS2 (Figure 6). Of these three residues, Ser127 was consistently excluded from the α‐helical stretch Pro117‐Leu144 by all three algorithms. Ala126 was excluded by Jpred4 and PsiPred, whereas Ser125 was excluded by PsiPred only. This suggests that, like in Lmod2‐TpmBS1, Ala126‐Ser127 form a two‐residue turn between two helices arranged in a α‐helical hairpin over the N‐terminus of Tpm. Notably, tetrapeptides with sequences SLSA (Lmod2‐TpmBS1, residues 24–27), KLSE (Tmod2‐TpmBS1, residues 23–26), and SASD (Tmod2‐TpmBS2, residues 125–128) have high propensities to adopt the conformation of a linking motif α_Rββα_R in α‐helical hairpins. ⁴² In the linking motif, the first and the last positions are occupied by the last residue of the first α‐helix and the first residue of the second α‐helix, respectively. The second and the third positions in the motif are occupied by turn‐forming residues in β‐conformation. The binding‐induced folding of Tmod2‐TpmBS2 into the α‐helical hairpin rather than into a continuous α‐helix would explain why only at most 26 N‐terminal residues of Tpm1.8/1.9/1.11/1.12 provide the entire binding surface for the interaction with Tmod2‐TpmBS2.

2.2.2. MDS produce 3D models for the molecular complexes of Tmod2‐TpmBS1 and Tmod2‐TpmBS2 with the N‐terminus of Tpm1.8/1.9/1.11/1.12

Using the information from the secondary structure predictions and the hypothesis that both Tmod2‐TpmBS1 and Tmod2‐TpmBS2 fold into an α‐helical hairpin upon binding to the N‐terminus of Tpm1.8/1.9/1.11/1.12, we built initial complexes in UCSF Chimera by placing the hairpins over the N‐terminus of Tpm1‐1b(1‐19)Zip. Upon completion of a series of simulated annealing runs, settled structures were subjected to production runs. The equilibrated structures are displayed in Figure 7. In the modeled structures, no part of either Tmod2‐TpmBS1 or Tmod2‐TpmBS2 peptide reaches beyond the first 19 residues of Tpm1‐1b(1‐19)Zip.

FIGURE 7.

Ribbon view of the modeled complexes (a) Tmod2‐TpmBS1/Tpm1‐1b(1‐19)Zip, (b) Tmod2‐TpmBS2/Tpm1‐1b(1‐19)Zip, and (c) the sequence alignment of Tmod2‐TpmBS1 and Tmod2‐TpmBS2. The N‐ and C‐termini of Tmod2‐TpmBS1, Tmod2‐TpmBS2 and Tpm1‐1b(1‐19)Zip are marked. Tmod2‐TpmBS1 and Tmod2‐TpmBS2 are colored green and cyan, respectively. In Tpm1‐1b(1‐19)Zip, the N‐terminal Gly and the amino acid residues from the Tpm1.8/1.9/1.11/1.12 sequence are colored orange and the residues from GCN4 are colored gray. Shown and labeled are side chains of those homologous amino acids in Tmod2‐TpmBS1 and Tmod2‐TpmBS2 that make direct contacts with Tpm1‐1b(1‐19)Zip (fingerprint interfacial residues). Same fingerprint residues are shown in bold in the sequences

The simulated complexes of Tmod2‐TpmBS1 and Tmod2‐TpmBS2 with Tpm share a strikingly similar 3D organization. In Tmod2‐TpmBS1, Glu17‐Lys23, and Glu26‐Leu36, separated by a ββ two‐residue linker, form two α‐helices with a ~31° angle between them. In Tmod2‐TpmBS2, Leu119‐Ser125 and Asp128‐Leu138 form two α‐helices with a ~30° angle between them. The structures are stabilized by an ionic bond and a hydrophobic core formed by residues—called hereafter fingerprint interfacial residues—that are conservative between the two sites of Tmod2 (Figure 7), among Tmod2 from other species, and among different Tmod isoforms (Figure S1). The fingerprint interfacial residues include Val137 and Leu138 in Tmod2‐TpmBS2, which are highly conserved and which have been shown to be essential for Tpm binding in other Tmod isoforms. ¹⁰ , ⁴³ Consistently, the majority of the amino acid residue variability in Tmod homologs is observed for residues that are not fingerprint interfacial residues (Figure S1).

Although the secondary structure predictors included residues Gly139‐Leu144 in the second helix of Tmod2‐TpmBS2, in simulated replicas of the Tmod2‐TpmBS2/Tpm1‐1b(1‐19)Zip complex Gly139 consistently adopted a conformation close to α_L and formed a kink separating the α‐helical turn formed by Val140‐Leu144 from the second helix. The kink turns residues Val140‐Leu144 away from the Tpm coiled coil, indicating that Val140‐Leu144 might not be a part of the Tmod/Tpm binding interface.

2.2.3. 3D model of the Tmod2 pointed‐end assembly

To validate the 3D models of Tmod2‐TpmBS1 and Tmod2‐TpmBS2 complexes with Tpm, we used them to dock a Tmod2 molecule onto the pointed end of the actin filament (Figure 8). The pointed end was modeled from an atomic model for the Tpm cable with F‐actin. ⁴⁴ Interactions of Tmod2 actin‐binding sites (ABS) 1 and 2 with the pointed end were modeled from crystal structures of the homologous Tmod1 ABS/actin complexes. ⁴⁵ Modeled binding sites of Tmod2 were connected with the remaining segments of the Tmod2 amino acid sequence. The absence of steric clashes provides evidence for the feasibility of the model.

FIGURE 8.

Two opposite side views of the Tmod2 pointed‐end assembly. (a) A “frontal” view of the actin filament with Tmod2‐TpmBS1 shown in green on the foreground and bound to a Tpm protomer (orange), and (b) a “back” view of the actin filament showing Tmod2‐TpmBS2 (cyan) bound to a Tpm protomer on the opposite side from Tmod2‐TpmBS1. The pointed end of the actin filament was modeled as before ⁴⁰ using an atomic model for the Tpm cable with F‐actin. ⁴⁴ Actin‐binding sites of Tmod2 actin‐binding site 1 (ABS1) and actin‐binding site 2 (ABS2) were modeled from the crystal structures of the homologous Tmod1 ABS/actin complexes ⁴⁵ (red and magenta, respectively). Linker segments connecting Tmod2 binding sites are shown in blue

3. DISCUSSION

Tmods are key regulators of actin dynamics with important roles in myogenesis, neuronal morphogenesis, and cell motility. ⁸ , ²⁵ , ⁴⁶ They bind and cap actin filaments at the pointed end thus blocking the pointed end from polymerization/depolymerization. ⁶ The binding is Tpm dependent and is improved by orders of magnitude when Tpm is present. ⁶ , ⁷ , ⁴⁷ , ⁴⁸ The effect of Tpm on Tmod binding to the filament is due to formation of a stable complex between F‐actin and Tpm, whereby Tpm supplies additional binding sites for Tmod.

Tpm is an elongated coiled‐coil protein, which, depending on the isoform, spans in vertebrates six or seven successive actin monomers along the filament. ⁴⁹ Apart from associating with actin, each Tpm molecule interacts with adjacent Tpm molecules via head‐to‐tail (N‐terminus to C‐terminus) overlap, thus forming two continuous Tpm cables on the opposite sides of the actin filament. At the pointed end, each Tpm cable has an exposed N‐terminus, which can bind to Tmod.

Tmod is a protein with an intrinsically disordered N‐terminal region and a globular C‐terminal part (reviewed in Reference 18). It forms a complex with the pointed end by associating via a total of four binding sites, two of which (ABS1 and ABS2) interact with actin, and the remaining two (TpmBS1 and TpmBS2) bind to the two N‐termini of Tpm located at the opposite sides of the filament. ¹⁰ , ¹¹ The sequential order of binding sites in Tmod is TpmBS1, ABS1, TpmBS2, and ABS2, with the first three sites located in the intrinsically disordered N‐terminal half of the Tmod protein. The intrinsic flexibility allows Tmod, a relatively small, ~350 amino acid residue long protein, to wrap around the pointed end of the actin filament and interact simultaneously with four protomers of the actin filament assembly (two Tpm molecules and two actin molecules).

Although a schematic model of Tmod assembly was proposed a number of years ago ¹⁰ , ¹¹ and the atomic structure of ABS complexes with actin has been determined, ⁴⁵ the 3D structure of complexes between TpmBS and Tpm has remained elusive and produced contradicting hypotheses on modes of association between TpmBS and Tpm. ¹¹ , ⁴⁵ Possible reasons for difficulties with the determination of the structures are the intrinsic flexibility of TpmBS peptides and relatively fast kinetics of the association/dissociation processes governing TpmBS/Tpm interactions. Indeed, attempts to co‐crystallize TpmBS1 of Tmod or its homolog, leiomodin (Lmod), with Tpm model peptides resulted in crystals composed of Tpm peptides only. ⁵⁰ , ⁵¹

Very recently, by using NMR and NMR‐guided MDS, we gained a new insight into the mode of interaction of Tpm with TpmBS1 of Lmod (PDB ID 6UT2 ⁴⁰ ). Here, we used this structure as a starting point to create 3D models of complexes of Tpm with Tmod2‐TpmBS1 and with Tmod2‐TpmBS2. A striking feature of the two modeled structures is their structural homology, in contradiction to hypothetical modes of binding suggested earlier. ¹¹ , ⁴⁵ Both models represent a four‐helix bundle, with a Tpm‐binding site of Tmod forming a α‐helical hairpin positioned over the N‐terminus of the Tpm coiled coil. Similarly to the complex of Tpm with TpmBS1 of Lmod, ⁴⁰ the complexes of Tpm with Tmod2‐TpmBS1 and with Tmod2‐TpmBS2 mimic the structure of the Tpm head‐to‐tail overlap complex, which was also shown to represent a crisscross arrangement of four α‐helices. ²⁹ , ⁵² , ⁵³ Residues forming the Tmod/Tpm binding interface are conservative between TpmBS1 and TpmBS2 from different Tmod homologs (Figure S1), with some rare exceptions that can potentially be used for isoform‐specific tuning of binding proposed as a basis for tissue‐specific regulation of pointed end capping. ¹⁸

In our simulated models (Figure 7), the 19 N‐terminal residues of the short Tpms Tpm1.8/1.9/1.11/1.12 (which correspond to 14 residues in long Tpms) supply the major part of the interaction interface for the binding of Tmod TpmBSs. However, when 19 N‐terminal residues in Tpm1‐1b(1‐19)Zip were replaced with 26 N‐terminal residues in Tpm1‐1b(1‐26)Zip, we detected changes in Tmod NMR spectra. A similar effect was observed when we studied complexes formed between Lmod TpmBS1 and Tpm1.1 model peptides. ⁴⁰ , ⁴¹ Although we demonstrated that the major Lmod/Tpm interaction occurs within the first 14 residues of Tpm, additional 7 residues included in the model Tpm peptide led to an NMR spectrum change and a modest, two‐fold increase in binding affinity. ⁴¹ We explain this effect by medium‐range allostery and transient interactions between flexible parts of TpmBSs with Tpm. In addition, the α‐helical loop Val140‐Leu144 of Tmod2‐TpmBS2 can also interact with Tpm. The loop is not as conserved between different animal classes as the region containing the fingerprint interface residues (Figure S1), and therefore it might form an isoform‐specific transient rather than a stable interaction with Tpm.

The simulated structures of the two Tmod/Tpm complexes were found to be sterically consistent with the recreated 3D model of the full‐length Tmod assembly at the pointed end (Figure 8). The 3D model of the pointed‐end Tmod assembly offers new insights in how Tmod binding ensures tight control over the pointed end dynamics.

Depending on the cell type and specific process, the actin filament remodeling at the pointed end can manifest as depolymerization (e.g., to supply G‐actin monomers for growing barbed ends of actin filaments in nonmuscle cells) or pointed‐end elongation (e.g., to elongate thin filaments in muscle cells). A key aspect of the Tmod binding is the crisscross packing of the TpmBS1 or TpmBS2 helices over the top of the Tpm N‐terminus. With the crisscross arrangement, the coiled coil of the Tpm molecule can be kept essentially structurally intact, which is critical if the stability of the filament is not to be impaired and the pointed end is to be protected from depolymerization.

On the other hand, pointed‐end elongation is blocked by two ABS and two TpmBS at every surface where incoming actin and Tpm subunits would attach in the course of pointed‐end growth. By forming complexes that mimic the topology of the Tpm overlap complex, TpmBS1 and TpmBS2 occupy the very location where the head‐to‐tail interaction with an adjacent Tpm molecule takes place. This makes TpmBS1 and TpmBS2 effective inhibitors of the Tpm cable elongation, which contributes to the potency of the Tmod capping function.

4. MATERIALS AND METHODS

4.1. Synthetic peptides

Mouse Tmod2 Tpm binding peptides Tmod2‐TpmBS1 (residues 3–40, corresponds to TpmBS1) and Tmod2‐TpmBS2 (residues 112–149, corresponds to TpmBS2) were synthesized and purified at Tufts University Core Facility (Boston, MA).

4.2. Cloning of recombinant DNA constructs

Sequences of mouse Tmod2 (NP_001033799.1) and Tpm1.8 (NP_001157724.1) were from the National Center for Biotechnology Information (Bethesda, MD). DNA primers for cloning were purchased from Integrated DNA Technologies (Coralville, IA). Restriction enzymes, OneTaq DNA polymerase, and T4 ligase were supplied together with relevant solutions and buffers by New England Biolabs (Ipswich, MA). Two DNA fragments, one encoding Tmod2‐TpmBS1 fused at its N‐terminus with a MFH expression tag ⁵⁴ (MFH‐Tmod2‐TpmBS1), and another, corresponding to exon 1b of Tpm1 and encoding the first 43 residues of Tpm1.8/1.9/1.11/1.12, were synthesized at GenScript (Piscataway, NJ). The DNA sequences were optimized for Escherichia coli expression. ⁵⁵ The DNA fragment encoding MFH‐Tmod2‐TpmBS1 was cloned into the pET‐21b(+) vector (EMD‐Millipore, Burlington, MA) between NdeI and XhoI recognition sites. A Tmod2‐TpmBS2‐encoding DNA fragment was amplified from a vector containing the mouse Tmod2 gene ⁵⁶ and cloned as an MFH‐Tmod2‐TpmBS2 fusion protein by replacing Tmod2‐TpmBS1 in the MFH‐Tmod2‐TpmBS1 construct between EcoRI and XhoI recognition sites. Model Tpm peptides were cloned as MFH‐fusion proteins by inserting DNA encoding Tpm fragments into an in‐house pET‐21b(+)‐derived expression vector MFH‐Zip. The MFH‐Zip vector allows cloning of any chosen Tpm fragment between MFH and a part of the leucine‐zipper domain of the yeast transcription factor GCN4 ⁵⁷ using EcoRI and XhoI recognition sites. The resulting recombinant plasmid encodes for a MFH expression tag followed by a Tpm sequence fused in heptad register at its C‐terminus with the GCN4 sequence. To enable the peptide expression in E. coli, all model Tpm peptides included an N‐terminal Gly residue mimicking N‐terminal acetylation. ²⁶ , ²⁸ , ²⁹ All the constructs encoded for a methionine immediately before the peptides of interest to allow separation of the peptides from the MFH expression tags by treatment with cyanogen bromide (CNBr).

4.3. Expression and purification of recombinant peptides

DNA sequences of the recombinant MFH‐Tmod and MFH‐Tpm constructs were confirmed by Sanger sequencing (performed at GeneWiz, South Plainfield, NJ). Expression of the MFH‐Tpm fusion proteins was performed in lysogeny broth as described previously. ^{41 15}N‐labeling of MFH‐Tmod2‐TpmBS1 and MFH‐Tmod2‐TpmBS2 was performed in minimal medium as described previously. ⁴⁰ Fusion proteins were purified on Ni‐NTA resin (Qiagen, Hilden, Germany) as described previously. ⁴¹ Eluted from Ni‐NTA fusion proteins were dialyzed against 10 mM HCl with two dialysis solution changes, brought to 0.2 M in HCl and 8 M urea, and cleaved overnight with a 400‐fold molar excess of CNBr. Cleaved peptides were purified to homogeneity as described before. ⁴¹ The identity of purified recombinant peptides was confirmed by mass spectrometry at Tufts University Core Facility (Boston, MA). Peptide concentrations were determined from their ultraviolet (UV) spectra using a JASCO (Easton, MD) UV/Vis spectrophotometer. Extinction coefficients at 280 nm for the Tpm and Tmod2 peptides were predicted using the ExPASy/ProtParam tool ⁵⁸ as 2,980 and 1,490 M⁻¹ cm⁻¹, respectively.

4.4. CD measurements

CD measurements were performed using an Aviv model 400 spectropolarimeter (Lakewood, NJ) in 1 mm quartz cuvettes. The peptide concentrations were ~0.1 mg ml⁻¹. Melting curves were measured as a temperature dependence of CD signal at 222 nm. Mean residue ellipticities for the designed Tpm fragments were calculated and plotted against temperature using SigmaPlot 12 (Systat Software, San Jose, CA). Interaction between Tpm peptides and Tmod2‐TpmBS1 or Tmod2‐TpmBS2 was measured upon mixing equimolar concentrations of the peptides in 10 mM sodium phosphate, 100 mM NaCl, pH 7.0. ⁵⁹ Thermodynamic parameters were evaluated by fitting thermal unfolding curves to the experimental data. ²⁶ , ⁵⁹ , ⁶⁰ Statistical analysis was done using SigmaPlot 12.

4.5. NMR spectral analysis

NMR samples were prepared in 50 mM sodium phosphate buffer, pH 6.5, 10% D₂O, 0.2 mM EDTA, 0.1% sodium azide, and 1X Pierce EDTA‐free protease inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA). Concentrations of ¹⁵N‐labeled Tmod peptides were 0.2–0.25 mM. For NMR samples of Tmod/Tpm complexes, the Tpm peptide concentration typically exceeded Tmod concentration by ~50–100 μM. Fractions of folded (dimeric) Tpm peptides in NMR samples were evaluated as described by Greenfield et al. ⁶⁰ using CD data for thermal unfolding of Tpms. The evaluation showed that the majority of the excess Tpm peptide was folded (typically >90%). Fractions of unbound Tmod peptides were evaluated ²⁶ using the thermal unfolding CD curves of Tmod/Tpm complexes, and were estimated as <1–4%. This put an upper limit of ~0.04/0.2 ppm on ¹H/¹⁵N chemical shift variability potentially caused by chemical exchange between bound and unbound Tmod species. Therefore, spectral changes due to the chemical exchange were considered negligible for the purposes of this study.

Two‐dimensional ¹⁵N‐HSQC NMR spectra were recorded at 25°C on a Varian Inova 500 MHz spectrometer (Agilent Technologies, Santa Clara, CA) equipped with a 5 mm triple resonance probe. The gNhsqc pulse sequence ⁶¹ found in the BioPack pulse sequence library (Agilent Technologies) was used to record the 2D ¹⁵N‐HSQC spectra. The NMR spectra were processed using NMRPipe. ⁶² NMRViewJ (One Moon Scientific) was used for NMR spectra visualization.

4.6. Secondary structure predictions

Secondary structure predictions for the human Lmod2 sequence representing TpmBS1, Lmod2‐TpmBS1 (res2‐41), and the mouse Tmod2 sequences representing TpmBS1 and TpmBS2, Tmod2‐TpmBS1 (res3‐res40), and Tmod2‐TpmBS2 (res112‐res149), respectively, were made using Jpred4, ⁶³ PsiPred, ⁶⁴ and PredictProtein. ⁶⁵

4.7. Molecular dynamics simulations

Protein structure building, editing, and visualization were performed with UCSF Chimera. ⁶⁶ Structures were neutralized with Na+ or Cl− ions and solvated in TIP3P water ⁶⁷ with at least 10 Å between the structure and the edge of the solvation box. Energy minimizations were performed by the sander protocol in Amber18 ⁶⁸ with 2,500 cycles of the steepest descent method followed by 2,500 cycles of the conjugate gradient method. Simulated annealing and production runs were performed by the GPU‐accelerated pmemd implementation of the sander protocol with a 1 fs time‐step, using the ff14SB force field, ⁶⁹ and periodic boundary conditions. The temperature was controlled by a Langevin thermostat with a collision frequency of 3 ps⁻¹. Covalent bonds to hydrogen were constrained by SHAKE.

To build a starting structure of Tmod2‐TpmBS1 for docking onto Tpm1‐1b(1‐19)Zip, we used the atomic coordinates of the Lmod2‐TpmBS1/αTM1a_1‐14Zip complex (PDB ID: 6UT2) as a starting point. First, the Lmod2‐TpmBS1 amino acid sequence was replaced with the sequence of Tmod2‐TpmBS1. ⁴⁰ This complex was minimized before being subjected to a simulated annealing protocol whereby the complex was equilibrated for 10 ns at several temperatures starting at 300 K, increasing up to 320 K in 5 K increments, and decreasing back to 300 K in 5 K increments. Chains of αTM1a_1‐14Zip were then replaced with Tpm1‐1b(1‐19)Zip before being minimized and subjected to another simulated annealing run and a 400 ns production run at 298 K.

To build a starting structure of Tmod2‐TpmBS2 for docking onto Tpm1‐1b(1‐19)Zip, ⁵² we used the secondary structure prediction. As a result, for Pro117‐Ser125 and Asp128‐Leu144, dihedral angles (φ,ψ) were set to (−57, −47). The dihedral angles for the remaining residues were arbitrarily set to (−139, 135), with the exception of those for Ala126. The dihedral angles for Ala126 were set to (60, 60), which positioned the helices Pro117‐Ser125 and Asp128‐Leu144 at an acute angle with respect to each other and thus generated a conformation of the helix hairpin motif. ⁴² We oriented the Tmod2‐TpmBS2 helix hairpin in a crisscross topology over the N‐terminus of Tpm1‐1b(1‐19)Zip, much like in the Lmod2‐TpmBS1/αTM1a_1‐14Zip structure. ⁴⁰ This complex was minimized before being subjected to the same simulated annealing protocol as the Tmod2‐TpmBS1 complexes. The annealed complex was minimized before a 400 ns production run at 298 K.

For further analyses, a minimum energy structure was chosen from each Tmod2‐TpmBS1 and Tmod2‐TpmBS2 production run trajectory and minimized. The pointed end of the actin filament and its binding with Tmod2‐TpmBS1/Tmod2‐TpmBS2 were modeled as before ⁴⁰ using an atomic model for the Tpm cable with F‐actin ⁴⁴ and simulated structures obtained in this work. The second Tpm cable was added to the opposite side of F‐actin by ~180° rotation of the Tpm/Factin complex combined with an appropriate longitudinal translation. Tmod ABS 1 and 2 were docked to the pointed end as before ⁴⁰ using crystal structures of the homologous Tmod1 ABS/actin complexes ⁴⁵ (PDB IDs 4PKG and 4PKI). The docking was completed by connecting the Tmod2 binding sites with the remaining segments of the Tmod2 sequence followed by energy minimization of the resulting assembly in Amber18.

AUTHOR CONTRIBUTIONS

Dmitri Tolkatchev: Conceptualization; data curation; formal analysis; investigation; resources; supervision; validation; visualization; writing‐original draft; writing‐review and editing. Balaganesh Kuruba: Conceptualization; data curation; formal analysis; investigation; resources; validation; visualization; writing‐original draft; writing‐review and editing. Garry E. Smith: Conceptualization; data curation; formal analysis; validation; visualization; writing‐original draft; writing‐review and editing. Kyle D. Swain: Investigation; resources; writing‐review and editing. Kaitlin Smith: Resources; writing‐review and editing. Natalia Moroz: Resources; writing‐review and editing. Trenton J. Williams: Resources; writing‐review and editing. Alla S. Kostyukova: Conceptualization; formal analysis; funding acquisition; investigation; project administration; resources; supervision; validation; visualization; writing‐original draft; writing‐review and editing.

Supporting information

FIGURE S1 Alignment of homologous Tpm‐binding sites from Tmod1‐Tmod4. Amino acid sequence alignments of TpmBS1 and TpmBS2 from Mus musculus Tmod2 (NP_001033799.1) with those from M. musculus Tmod1 (NP_068683.1), Tmod3 (NP_058659.1), and Tmod4 (NP_057921.1) are shown in (A) and (B), respectively. Amino acid sequence alignment of TpmBS1 and TpmBS2 across several species is shown in (C) and (D), respectively. Residues comprising the α‐helical hairpins of M. musculus Tmod2 TpmBS1 and TpmBS2 in the modeled complexes are shown with bars. In (A)‐(D), fingerprint interfacial residues are highlighted with gray. In (A) and (B) yellow color highlights residues from α‐helical hairpins of Tmod1, Tmod3 and Tmod4 that are dissimilar to those residues that are identical or similar in Tmod2‐TpmBS1 and Tmod2‐TpmBS2, while residue positions highlighted with cyan are occupied with residues dissimilar in Tmod2‐TpmBS1 and Tmod2‐TpmBS2. In (C) and (D) residues dissimilar to corresponding residues in α‐helical hairpins of M. musculus Tmod2‐TpmBS1 and Tmod2‐TpmBS2 are highlighted in red.

Tolkatchev D, Kuruba B, Smith GE Jr, et al. Structural insights into the tropomodulin assembly at the pointed ends of actin filaments. Protein Science. 2021;30:423–437. 10.1002/pro.4000