The interaction of gelsolin with tropomyosin modulates actin dynamics

Authors


Abstract

We have investigated the interactions between the actin-binding proteins gelsolin and tropomyosin, with special respect to any effects on the functional properties of gelsolin. Limited proteolysis indicated that the loop connecting the gelsolin domains G3 and G4 is involved in tropomyosin binding. Under nonpolymerizing conditions, binding of tropomyosin neither prevented the formation of a 2 : 1 actin–gelsolin complex, nor did it affect the nucleating activity of gelsolin in actin polymerization, likely as a result of competitive displacement of tropomyosin from gelsolin. To evaluate the effect of tropomyosin on the actin filament severing activity of gelsolin, we measured both filamentous actin (F-actin) viscosity and the relative number concentrations of filaments after fragmentation, either by gelsolin alone or by gelsolin–tropomyosin complexes. The interaction of gelsolin with tropomyosin caused a reduction in F-actin severing activity of up to 80% compared to gelsolin alone. Thus, being bound to gelsolin, tropomyosin prevented gelsolin from severing actin filaments. By contrast, the severing activity of gelsolin for F-actin/tropomyosin was similar to that for F-actin alone even at a tropomyosin : actin saturation ratio of 1 : 7. Thus, when bound to actin filaments, tropomyosin did not significantly inhibit the severing of filaments by gelsolin. The interaction between gelsolin and tropomyosin was largely independent of the muscle actin and tropomyosin isoforms investigated. The results obtained in the present study suggest that tropomyosin is involved in the modulation of actin dynamics not via the protection of filaments against severing, but rather by binding gelsolin in solution to prevent it from severing and to promote the formation of new actin filaments.

Abbreviations
F-actin

filamentous actin

GS

gelsolin

TM

tropomyosin

Introduction

Gelsolin is an actin-modulating protein that severs F-actin, caps the barbed ends of actin filaments preventing monomer exchange, and promotes the nucleation step of actin polymerization [1-3]. Tropomyosin was primarily regarded as a muscle protein that regulates the interaction of actin-containing thin filaments with myosin filaments to allow contraction [4]. Multiple tropomyosin isoforms are also expressed in non-muscle cells where tropomyosins participate in a number of cytoskeleton-mediated cellular processes [5, 6]. Thus, gelsolin and tropomyosin may cooperate in the same cytoskeletal system. Gelsolin is also present in smooth muscles [7], as well as in the myofibrils of striated muscles where both exogenous and endogenous gelsolin were shown to locate along the thin filaments [8, 9]. A widely accepted function of tropomyosins in non-muscle cells is to stabilize actin filaments mechanically, and possibly to protect against the action of filament-destabilizing proteins and disassembling drugs [10]. The results obtained in biochemical and cell biological experiments consistently show that tropomyosin inhibits the fragmentation of actin filaments by cofilin [11-13] and increases the resistance of actin filaments against cytochalasin D- and lantrunculin-caused disassembly in an isoform-specific manner [14].

Previous studies have indicated that tropomyosin may inhibit actin filaments being severed by gelsolin [15-19]. In such studies, both skeletal and smooth muscle tropomyosin partially protected F-actin from gelsolin-induced fragmentation [18, 19], whereas different non-muscle tropomyosins provided either full [16, 17] or partial [15, 18] protection, or did not protect F-actin at all [18]. This variability was explained by a higher affinity of muscle tropomyosin to F-actin, as well as by differences in the binding of non-muscle isotropomyosins to F-actin. By contrast, the presence of tropomyosin either along actin filaments or at their barbed ends did not prevent the binding of gelsolin to actin [8, 17, 20, 21]. Moreover, an interaction between gelsolin and tropomyosin was demonstrated previously by gelsolin/tropomyosin cross-linking, affinity chromatography of gelsolin on immobilized tropomyosin, the titration of fluorescein isothiocyanate-labelled gelsolin with tropomyosin [22], and by the effect of gelsolin on tropomyosin aggregation [23]. It was also shown that the complex between gelsolin and tropomyosin is calcium-sensitive, has a dissociation constant of 0.6 μm, and involves the interaction of tropomyosin with the gelsolin domain G2 [22, 23]. However, the functional effects of the complex formation have not yet been investigated.

The present study aimed to further characterize the gelsolin–tropomyosin interaction, and to investigate the effects of tropomyosin on the functional properties of gelsolin. The results obtained reveal the involvement of the loop connecting gelsolin domains G3 and G4 in complex formation, and show that, by forming a complex with gelsolin in solution, tropomyosin prevents gelsolin from severing actin filaments. By contrast, no association of either skeletal or smooth muscle tropomyosin with actin filaments provided significant protection from being severed by gelsolin.

Results

The gelsolin–tropomyosin interaction is revealed by limited proteolysis

Proteolysis of gelsolin with thermolysin and trypsin generates a limited number of well defined fragments [24, 25]. The susceptibility of the cleavage sites to the enzymes depends on the conformation of gelsolin, and is also affected by interaction with gelsolin-binding proteins [25-27], thus providing a tool for detecting the contacts formed by these interactions. Therefore, we employed limited proteolysis to reveal the gelsolin–tropomyosin-binding sites. Consistent with previous data, incubation of calcium-activated ‘open’ gelsolin with thermolysin generated two fragments of 45 and 40 kDa resulting from cleavage within the linker region between domains G3 and G4 [24, 27], whereas thermolysin did not cleave tropomyosin at the times and concentrations used (Fig. 1). The presence of tropomyosin clearly inhibited gelsolin cleavage (Fig. 1A). Although an allosteric mechanism of the inhibition cannot be excluded, it is plausible to assume that the linker G3/G4 is involved in tropomyosin binding. The inhibition was more efficient with smooth muscle tropomyosin than with skeletal muscle skeletal muscle tropomyosin (Fig. 1B).

Figure 1.

Complex formation of gelsolin and tropomyosin as revealed by limited proteolysis with thermolysin. SDS/PAGE of (A) gelsolin (0.5 mg·mL−1), skeletal muscle tropomyosin (0.5 mg·mL−1) and their 1 : 1 mixture, complemented with 0.2 mm CaCl2 and digested with thermolysin at a 1 : 200 mass ratio for the times indicated and (B) comparison of the effects of skeletal and smooth muscle tropomyosin on the susceptibility of gelsolin to proteolysis with thermolysin; conditions as in (A).

Trypsin cleaves both calcium-activated (open) and Ca-free (closed) gelsolin in domain G2, close to the loop between G1 and G2, at the surface of the molecule [25]. In the presence of tropomyosin, this cleavage is accelerated rather than inhibited (Fig. 2A). Because, under these conditions, trypsin also cleaves tropomyosin (Fig. 2), we used the tropomyosin mutant G126R with a strongly increased resistance to trypsinolysis as a result of the point mutation [28]. Incubation of gelsolin with this mutant also rendered the linker G1/G2 more susceptible to tryptic cleavage (Fig. 2B), consistent with allosteric transitions caused by tropomyosin binding to the gelsolin domain G2 [23]. Taken together with the results of the thermolysin cleavage, these data confirm the formation of a gelsolin–tropomyosin complex.

Figure 2.

Interaction of gelsolin with tropomyosin as monitored by limited proteolysis with trypsin. SDS/PAGE of (A) gelsolin (0.5 mg·mL−1), smooth muscle tropomyosin (0.5 mg·mL−1) and their 1 : 1 mixture and (B) gelsolin (0.5 mg·mL−1), recombinant human α-striated tropomyosin G126R mutant [28] and their 1 : 1 mixture. Other conditions match those shown in Fig. 1. Cleavages were performed at a trypsin : gelsolin mass ratio of 1 : 50 in the presence of 0.2 mm CaCl2.

Probing the effects of tropomyosin on the gelsolin–G-actin interaction and on gelsolin-induced nucleation of actin polymerization

Under nonpolymerizing conditions, calcium-activated gelsolin forms a complex with two actin monomers generating a two- to three-fold increase in the fluorescence intensity of pyrenyl-labelled actin [26, 29]. This effect was not observed if the second monomer was not labelled indicating that the enhancement of actin fluorescence in ternary actin–gelsolin complexes is caused by actin–actin interaction [26]. Therefore, addition of unlabelled G-actin to the complex resulted in a rapid decrease in fluorescence intensity caused by replacement of one pyrenyl-labelled actin monomer by unlabelled actin [26]. Figure 3A shows that pre-incubation of gelsolin with tropomyosin does not prevent formation of the 2 : 1 actin–gelsolin complexes. Moreover, tropomyosin does not appear to diminish the affinity of G-actin to gelsolin because the decrease in fluorescence intensity occurred at almost the same rate in both complexes (Fig. 3A).

Figure 3.

G-actin binding and nucleation of actin polymerization by gelsolin in gelsolin–tropomyosin complexes. (A) Interaction of gelsolin with G-actin under nonpolymerizing conditions. 1.2 μm pyrenyl-actin in buffer G was mixed with either 0.024 μm gelsolin or with 0.024 μm gelsolin pre-incubated with smooth muscle tropomyosin (1 : 1 m/m) in the presence of 0.2 mm CaCl2 (arrows). After fluorescence reached the plateau, unlabelled G-actin (6 μm) was added (arrowheads). (B) Time course of actin polymerization nucleated with gelsolin in the presence or absence of tropomyosin. 9.6 μm pyrenyl-actin in buffer G complemented with 0.2 mm CaCl2 was mixed with 0.096 μm gelsolin or gelsolin pre-incubated with smooth muscle tropomyosin (1 : 1 m/m) and the fluorescence intensity was recorded.

Under polymerizing conditions, gelsolin nucleated actin polymerization as monitored by an increase of pyrenyl-fluorescence. Again, tropomyosin affected the nucleating activity of gelsolin in actin polymerization very little (Fig. 3B), as observed by a comparison of gelsolin alone and preformed gelsolin–tropomyosin 1 : 1 complexes.

Effect of tropomyosin on gelsolin-induced fragmentation of F-actin

Severing of F-actin by gelsolin results in the fast generation of numerous short filaments with gelsolin-capped barbed ends [1, 2], a process accompanied by a dramatic drop in viscosity. Incubation of F-actin with gelsolin at a 1 : 100 gelsolin to actin molar ratio resulted in an approximate 70% reduction of specific viscosity (Fig. 4). However, when gelsolin was pre-incubated with tropomyosin at a molar ratio of 1 : 1, and then added to F-actin at a molar ratio of 1 : 100, the viscosity dropped by only 40% (Fig. 4), corresponding to an approximate 65% inhibition of gelsolin activity (Table 1). This holds for both skeletal and smooth muscle tropomyosin. By contrast, gelsolin activity was inhibited by only 9% when the same amount of tropomyosin was pre-incubated with F-actin for several minutes before gelsolin was added to the mixture (Fig. 4 and Table 1). Thus, interaction with tropomyosin apparently interferes with the severing activity of gelsolin.

Table 1. Percentage inhibition of the actin filament severing activity of gelsolin by tropomyosin as determined by viscometry. Gelsolin was added to F-actin either in a mixture with tropomyosin (A + TM/GS) or after mixing tropomyosin with actin (A/TM + GS) in the ratios indicated, and the steady-state viscosities were determined after 5 min. Controls and other conditions were as described in Fig. 8A. The reduction of viscosity induced by gelsolin was calculated from the specific viscosities measured and, from these values, the relative inhibition of gelsolin activity was determined as described in Materials and methods. smTM, smooth muscle tropomyosin; skTM, skeletal muscle tropomyosin. The results show a representative set of experiments
 Actin : Tropomyosin 100 : 1Actin : Tropomyosin 50 : 1Actin : Tropomyosin 7 : 1
Actin : Gelsolin 100 : 1Actin : Gelsolin 50 : 1Actin : Gelsolin 100 : 1Actin : Gelsolin 50 : 1
smTMskTMsmTMskTMsmTMskTMsmTMskTM
A/TM + GS8.88.811.07.018.013.025.023.2
A + TM/GS67.064.785.083.071.575.374.075.0
Figure 4.

Fragmentation of F-actin by gelsolin and gelsolin–tropomyosin complexes, measured by viscometry. The results of a representative experiment are shown. Either gelsolin, tropomyosin or gelsolin–tropomyosin 1 : 1 complexes were added to 24 μm skeletal muscle F-actin (1 : 50 m/m to actin) and the viscosity was measured before and after addition of these substances. A, F-actin (control); A + TM, F-actin incubated with tropomyosin (control); A + GS, F-actin + gelsolin; A/TM + GS, F-actin pre-incubated with tropomyosin for 10 min and gelsolin added subsequently; A + TM/GS, tropomyosin and gelsolin premixed at 1 : 1 m/m, incubated for 10 min and then added to F-actin (1 : 50 complex to actin).

For a more quantitative determination of this effect, we compared the relative number concentrations (see 'Materials and methods') of the filaments in F-actin solutions fragmented by either gelsolin alone or by gelsolin–tropomyosin complexes. In this assay, F-actin is diluted below the critical concentration of actin to induce depolymerization. The initial rate of F-actin depolymerization is linear and proportional to the number concentration of the filaments [30]. To obtain a population of actin filaments with almost homogenous lengths, pyrenyl-labelled actin was polymerized in the presence of gelsolin at a molar ratio to actin of 1 : 600. After attaining steady-state, either gelsolin or gelsolin–tropomyosin complexes (preformed at molar ratios of 1 : 1 or 1 : 4, respectively) were added to F-actin at molar ratios of gelsolin to actin from 1 : 200 to 1 : 30. After 60 min, aliquots of these solutions were diluted below the critical concentration (see 'Materials and methods'). The initial depolymerization rates of F-actin incubated either with gelsolin or with gelsolin–tropomyosin complexes were linear (Fig. 5A). However, depolymerization of F-actin incubated with gelsolin–tropomyosin 1 : 1 complexes was approximately 60% lower than that of F-actin fragmented by gelsolin alone. A similar reduction of severing was observed with a sample containing the same amount of gelsolin but pre-incubated with tropomyosin at a molar ratio of 1 : 4 (not shown).

Figure 5.

The effects of tropomyosin on the relative number concentration of actin filaments, generated by gelsolin severing. (A) Representative time courses for depolymerization of skeletal muscle F-actin fragmented with either gelsolin or gelsolin–tropomyosin complexes. 24 μm pyrenyl-labelled F-actin pre-capped with gelsolin at a 1 : 600 molar ratio was incubated with gelsolin or gelsolin–tropomyosin complexes at a gelsolin : actin molar ratio of 1 : 50 for 1 h. Aliquots (50 μL) of the F-actin solutions were diluted into 2 mL of buffer G complemented with 0.1 m KCl. The solutions were gently mixed in fluorescence cuvettes by inversion, and the decrease in the intensity of pyrenyl fluorescence associated with F-actin depolymerization was recorded. (1) F-actin; (2) F-actin + gelsolin; (3) F-actin + gelsolin–tropomyosin; (4) F-actin + tropomyosin and gelsolin added subsequently. Inset: the initial linear parts of depolymerization curves that were used to calculate the relative filament number concentrations. (B) The relative filament number concentrations of skeletal and smooth muscle F-actin upon severing with gelsolin or gelsolin–tropomyosin complexes. Bars: white background, smooth muscle actin; grey background, skeletal muscle actin. (C) The relative filament number concentrations of skeletal muscle F-actin upon severing with gelsolin or complexes of gelsolin with either skeletal or smooth muscle tropomyosin. (D) The relative filament number concentrations of skeletal muscle F-actin upon severing with gelsolin or complexes of gelsolin with either α- or β-tropomyosin homodimers.

By contrast, when gelsolin was added to F-actin that had been pre-incubated with the same amount of tropomyosin, the severing activity was reduced by only approximately 20%. This indicates that the inhibition was mainly caused by the gelsolin–tropomyosin interaction, rather than being a result of the binding of tropomyosin to actin. The inhibitory effects were independent of the actin and tropomyosin isoforms used, being similar for both skeletal and smooth muscle actins (Fig. 5B), skeletal and smooth muscle tropomyosins (Fig. 5C) and for α- and β-tropomyosin homodimers (Fig. 5D). In general, the activities of both gelsolin and gelsolin–tropomyosin complexes were proportional to the gelsolin/actin ratio, as seen from the linear dependence in Fig. 6. At the same time, the inhibitory activity of tropomyosin increased in a nonlinear manner with respect to the increase of the gelsolin/actin ratio (Fig. 6, inset).

Figure 6.

The effect of tropomyosin on the severing activity of gelsolin varies with the gelsolin/actin ratio. Gelsolin–tropomyosin complexes were formed at a 1 : 1 molar ratio of tropomyosin to gelsolin. The gelsolin index is the increase in the relative number concentration of actin filaments generated by gelsolin or gelsolin–tropomyosin complexes calculated as I = R[A + GS]/R[A] (■) or I = R[A + TM/GS]/R[A] (●), respectively, where R is the initial rate of depolymerization, A is F-actin, GS is gelsolin and GS/TM is the gelsolin–tropomyosin complex. Inset: Inhibitory effect of tropomyosin on the severing activity of gelsolin given as a percentage and calculated as 1 – R(A + GS/TM)/R(A + GS), where R is the initial rate of depolymerization, A is F-actin, and GS/TM and GS are gelsolin–tropomyosin complex and gelsolin, respectively. Abscissa, gelsolin/actin ratio; ordinate, inhibitory effect of gelsolin-bound tropomyosin (%).

A much smaller, if any, effect of tropomyosin on F-actin severing by gelsolin was observed when the relative filament number concentration was estimated by the nucleation assay [30, 31]. In this mode, unlabelled F-actin, pre-fragmented by either gelsolin or by gelsolin–tropomyosin complexes, served as seeds for the polymerization of additional pyrenyl-labelled G-actin. The initial rates of actin polymerization onto these fragments formed either in the presence or absence of tropomyosin exceeded the rate of polymerization onto nonfragmented F-actin by a factor of approximately 4, showing apparently no effect of tropomyosin on the severing activity of gelsolin (Fig. 7). However, viscosity data from identical parallel samples (data not shown) demonstrated a reduction of severing similar to that shown in Fig. 4.

Figure 7.

Nucleation of actin polymerization by seeds obtained from F-actin severed with either gelsolin or gelsolin–tropomyosin complexes. Unlabelled F-actin (24 μm) pre-capped by gelsolin (1 : 400 molar ratio) was fragmented with gelsolin or gelsolin–tropomyosin at a gelsolin : actin ratio of 1 : 100 and a tropomyosin : gelsolin ratio of 1.5 : 1. Nucleation of pyrenyl-labelled G-actin (4.8 μm in buffer G complemented with 0.2 mm CaCl2) was initiated by the addition of 0.1 m KCl and 50 μL of the fragmented F-actin. The figure shows the initial linear parts of representative polymerization curves for (1) actin, (2) actin + gelsolin and (3) actin + gelsolin–tropomyosin.

This discrepancy may be explained by assuming that the seeds contained not only the F-actin fragments, but also free gelsolin–tropomyosin complexes not bound to F-actin, which then served as additional nuclei for actin polymerization.

Severing of tropomyosin-saturated actin filaments by gelsolin

In the previous experiments, the ratio of tropomyosin to F-actin was well below the saturating ratio. Therefore, we next investigated the gelsolin-induced actin filament severing in the presence of tropomyosin under conditions (tropomyosin to actin molar ratio of 1 : 7) where the filaments should be maximally covered with tropomyosin. As before, the gelsolin to actin ratio was 1 : 50 or 1 : 100. The viscosity of the solutions and the relative filament number concentration after fragmentation were determined as described above. Saturation of F-actin with a stoichiometric amount of skeletal or smooth muscle tropomyosin resulted in an approximate 15% increase in specific viscosity (Fig. 8A) as a result of either the stabilization of longer filaments or the reduction of filament flexibility by tropomyosin [32]. Taking this increased viscosity of actin/tropomyosin into account, the addition of gelsolin to F-actin/tropomyosin caused only a minor difference in the drop of specific viscosity compared to that generated with F-actin (Fig. 8A). By contrast, the viscosity drop was much smaller after the addition of a gelsolin–tropomyosin complex preformed with the same amount of tropomyosin (Fig. 8A), indicating significant inhibition of gelsolin severing activity. To evaluate this effect more quantitatively, the tropomyosin-induced inhibition of gelsolin activity was calculated from the viscosity data (Table 1), as described in Materials and methods. These results show that tropomyosin (when bound to F-actin) inhibits gelsolin-induced severing by only 20%. On the other hand, the same amount of tropomyosin pre-incubated with gelsolin inhibited the severing of F-actin by > 70% (Table 1).

Figure 8.

Fragmentation of tropomyosin-saturated F-actin by gelsolin and gelsolin–tropomyosin complexes. The results of representative experiments are shown. (A) Viscometry of F-actin solutions: 24 μm skeletal muscle F-actin was mixed with smooth muscle tropomyosin (7 : 1 molar ratio), and, after 15 min of incubation, gelsolin (1 : 50 to actin) was added. Alternatively, the same amount of tropomyosin as before was mixed with gelsolin and, after 10 min of incubation, added to F-actin. The viscosities of these samples and the respective controls were measured at intervals. (A) F-actin (control); A + TM, F-actin + tropomyosin (control); A + GS, F-actin + gelsolin; A/TM + GS, F-actin/tropomyosin + gelsolin; A + TM/GS, F-actin + gelsolin–tropomyosin complex. (B) Fluorometry (initial parts of F-actin depolymerization curves): 24 μm pyrenyl-F-actin pre-capped with gelsolin at a 1 : 600 molar ratio was saturated by tropomyosin as in (A) and incubated for 1 h with gelsolin at a gelsolin : actin molar ratio of 1 : 50. Alternatively, the same amount of tropomyosin as before was mixed with gelsolin and, after 15 min of incubation, added to F-actin. The initial linear parts of the depolymerization curves were recorded as described in Fig. 5 and used to calculate the relative filament number concentrations (labels as in A).

Corresponding results were obtained by a comparison of the relative number concentration of gelsolin-fragmented F-actin and F-actin/tropomyosin filaments. Depolymerization of tropomyosin-saturated actin filaments was slower than that of F-actin alone (Fig. 8B) as a result of the inhibition of F-actin depolymerization by tropomyosin [33]. Depolymerization of tropomyosin-saturated filaments upon their fragmentation by gelsolin was also slower than that of F-actin fragmented by gelsolin (Fig. 8B), giving the impression that tropomyosin protects actin from being severed by gelsolin. However, when the corresponding controls were taken into account, the effects of gelsolin on the fragmentation of F-actin and F-actin/tropomyosin were essentially identical (Table 2). By contrast, the same amount of tropomyosin pre-incubated with gelsolin reduced F-actin fragmentation by approximately 50% (Fig. 8B).

Table 2. Fragmentation of free and tropomyosin-saturated F-actin by gelsolin. The data were obtained in an experiment similar to that shown in Fig. 8B and are the average of 5–7 measurements. Rate is the slope of the initial part of the F-actin depolymerization curve that is proportional to the relative filament number concentration. The ‘gelsolin index’ is defined as the increase in the relative number concentration of actin filaments generated by gelsolin or gelsolin–tropomyosin complexes; see also Figs 7 and 8
 RateGelsolin index
Actin0.123 
Actin + GS0.4463.6
Actin + TM0.100 
Actin/TM + GS0.3613.6
Actin + GS/TM0.2552.0

Discussion

The actin cytoskeleton responds to numerous cellular stimuli, and involves the interaction of actin with a multitude of actin-binding proteins that promote the assembly, capping, disassembly, bundling or network formation of filaments. It is becoming more and more evident that actin-binding proteins can also interact with each other [5, 6, 34]. Our results demonstrate for the first time that the interaction between tropomyosin and gelsolin [22, 23] considerably affects actin dynamics. Studying the effects of gelsolin–tropomyosin complex formation on the functional properties of gelsolin, we have shown that, once bound to gelsolin, tropomyosin does not interfere with the capability of gelsolin to nucleate actin polymerization but prevents the severing of F-actin by gelsolin through complex formation. On the other hand, once bound to actin filaments, tropomyosin does not significantly prevent the severing of filaments by gelsolin.

In an attempt to reveal the sites involved in gelsolin–tropomyosin interaction, we have used limited proteolysis with enzymes whose cleavage sites on gelsolin have been determined previously [24, 26, 27]. Gelsolin consists of six similar domains (G1–G6) connected by unstructured linker regions that are susceptible to proteolytic cleavage, especially the long linker G3/G4. At a molar ratio of 1 : 1, tropomyosin efficiently protected calcium-activated gelsolin against thermolysin-induced proteolysis within the linker G3/G4, and also increased accessibility of the G1/G2 linker region to proteolysis with trypsin. This is consistent with the reported binding of gelsolin to tropomyosin via domain G2 [23]. The observed difference between smooth and skeletal muscle tropomyosin on the proteolytic cleavage within the linker G3/G4 may be a result of variation in their sequences [35]. The simplest interpretation of these results would be a model of the gelsolin–tropomyosin complex similar to that of gelsolin/F-actin interaction [36]. In this model, tropomyosin is situated between gelsolin domains G1–3 and G4–6, perpendicular to the plane of the gelsolin molecule. The gelsolin linker G1/G2 that makes intimate contact with actin in the G1–3-actin complex [36] could then interact with tropomyosin in the same manner. The sequence differences between skeletal and smooth muscle tropomyosin are located within pseudo-repeat 2 (approximately residues 40–80) and at the C-terminus [35]. The difference in the interaction of these tropomyosins with gelsolin suggests that the variable regions in tropomyosin may be involved in gelsolin binding.

The binding of gelsolin to G-actin and the formation of a gelsolin/actin complex was not affected by the presence of tropomyosin, nor was the ability of gelsolin to promote nucleation of actin polymerization. This can either be explained by the formation of a ternary complex including all three proteins or, more likely, as a competitive displacement of tropomyosin from gelsolin as a result of the high affinity of G-actin to gelsolin [37], whereas tropomyosin does not bind G-actin or actin oligomers formed under these conditions [4, 38]. By contrast, there was a strong effect of the gelsolin–tropomyosin interaction on the severing of F-actin by gelsolin. A reduction of 60–80% gelsolin activity was observed when a tropomyosin/gelsolin complex was added to actin compared to gelsolin alone. However, when tropomyosin was added to F-actin first, and then gelsolin was added subsequently, the interfering effect was low, even at high concentrations of tropomyosin where the actin filaments were saturated by tropomyosin.

Complete inhibition of gelsolin activity was not observed under any of the different study conditions, including a several-fold excess of tropomyosin over gelsolin. Some fragmentation was always measured. This may be the result of a partial dissociation of the gelsolin–tropomyosin complex after dilution into F-actin. Free tropomyosin may bind to actin leaving free gelsolin, which then severs actin filaments. We have shown (Fig. 7) that the tropomyosin/gelsolin complexes cannot interact with F-actin but do bind G-actin. Hence, these inactive complexes will initiate polymerization rather than bind to F-actin.

A comparison of the results with different ratios of tropomyosin/gelsolin (Figs 4 and 8 and Table 1) shows that the inhibition of gelsolin severing activity does not increase significantly at higher ratios, as would be expected for an equilibrium situation. However, the relationships here are more complicated. At gelsolin–tropomyosin 1 : 1, both proteins are mainly present in binary complexes, whereas only small amounts of gelsolin and tropomyosin will bind to actin or are free in solution. With increasing ratios of tropomyosin/gelsolin, the majority of tropomyosin is initially free, whereas only a small amount of it exists in complex with gelsolin. Against expectation, the excess tropomyosin does not drive the residual free gelsolin into complete complex formation, which would lead to a stronger inhibition of severing. Rather, the free tropomyosin binds to F-actin and approaches conditions where, as a result of cooperative effects, the affinity of tropomyosin to actin increases [34-36]. Therefore, the excess tropomyosin disappears out of equilibrium, leading to almost the same inhibition of fragmentation as with low tropomyosin.

Taken together, these findings suggest that tropomyosin may shuttle between actin filaments and gelsolin. The affinity of tropomyosin for F-actin is rather weak. Under conditions similar to those of the present study, the constant Kd for binding of tropomyosin to F-actin was determined to be 0.5 μм [39], which is comparable with a Kd of 0.6 μм for the gelsolin–tropomyosin complex [22]. Moreover, the binding of tropomyosin to F-actin is cooperative [40-42], and the affinity of individual tropomyosin molecules to actin is much lower (Kd = 0.3–0.5 mm) than at the saturating ratio of 1 : 7 [40, 41]. It is also important that the affinity of tropomyosin for actin filaments depends on the magnesium concentration: a small change in [Mg2+] may cause a major change in structural organization of the complex [40, 42].

Therefore, the equilibrium between tropomyosin bound to F-actin or to gelsolin depends on the experimental conditions, including protein and salt concentrations, the time of actin polymerization, pre-incubation with tropomyosin, and whether tropomyosin was added to F-actin or actin was polymerized in its presence. A thorough analysis of these parameters is required to explain the inhibition of gelsolin activity by tropomyosin reported in previous studies [15-19].

Our results show that even the saturation of actin filaments with tropomyosin does not significantly reduce the severing activity of gelsolin, and hence does not protect the filaments from being fragmented. Tropomyosin will only confer a significant resistance of actin filaments to gelsolin when in combination with other actin-binding proteins, such as caldesmon [18, 19] or nebulin [8]. By contrast, the major contribution of the observed reduction of gelsolin activity is a result of the complex formation of gelsolin and tropomyosin. Therefore, it is more likely that, in stable microfilament systems such as muscles, gelsolin is bound to tropomyosin rather than to actin but is transferred onto actin when the microfilament system rearranges.

The ability of tropomyosin to reduce the severing activity of gelsolin, as shown in the present study, may be of functional importance in the processes of cytoskeletal rearrangements. This is consistent with the finding that, in non-muscle cells, tropomyosin, albeit mostly connected with stress fibres and bundles of actin filaments, is also present in isoform-specific particles (‘dots’) in lamellipodia protruding from spreading cells [43, 44]. Stimulation of cells with growth factors induces a rapid, actin polymerization-dependent outgrowth of lamellipodia and filopodia, which is accompanied by a decrease in the number of tropomyosin-containing particles, suggesting their possible involvement in actin filament formation [44]. Several studies have reported that gelsolin may also be diffusely distributed in the cytoplasm [45-47]. We therefore consider that the interaction of tropomyosin with gelsolin may be a factor contributing to the regulation of cytoskeleton dynamics.

Materials and methods

Protein preparations

Rabbit skeletal muscle actin was purified using the procedure described by Spudich and Watt [48] with an additional, final gel filtration step on Sephadex G150 (GE Healthcare, Milwaukee, WI, USA) to remove traces of actin-binding proteins. G-actin in buffer G (0.2 mm ATP, 0.1 mm CaCl2, 0.4 mm β-mercaptoethanol, 5 mm Tris–HCl, pH 8.2 and 1 mm NaN3) was stored on ice and used within 1 week.

Smooth muscle γ-actin [49] was isolated from myosin-extracted, acetone-dried pig stomach muscle powder, using the same procedure as above [48], except that 1 mm ATP was present at all steps of the preparation. After the gel filtration step, only the rear half of the eluted actin peak containing pure monomeric actin was collected and used for experiments.

Actin labelled with N-(1-pyrenyl) iodoacetamide at Cys374 was prepared as described previously [50], lyophilized in the presence of 4 mm sucrose and stored at −80 °C. Before use, the lyophilized pyrenyl-actin was dissolved in buffer G, dialyzed against the same buffer overnight, and clarified by centrifugation at 100 000 g for 3 h to remove traces of polymerized actin. In all experiments, the term ‘pyrenyl-labelled actin’ refers to a mixture of 90% unlabelled actin and 10% pyrenyl-labelled actin.

Gelsolin was purified from pig stomach smooth muscle [7] and stored as ammonium sulfate precipitate in liquid nitrogen. Before use, the precipitate was dissolved and dialyzed against PSAM-buffer (10 mm imidazole, 0.2 mm EGTA, 0.2 mm dithiothreitol and 2 mm NaN3, pH 7.0). The severing activity of the gelsolin samples toward F-actin was tested by viscometry.

Purification of tropomyosins and separation of α- and β-isotropomyosins was carried out in accordance with procedures described by Smillie [51]. Tropomyosin was kept in TM-buffer (40 mm KCl, 2 mm MgCl2, 0.2 mm dithiothreitol, 10 mm imidazole, pH 7.0) on ice.

Recombinant human TPM1 isoform 1 (α-striated tropomyosin) G126R mutant [28] was a generous gift from Dr Ilya Nevzorov (University of Helsinki, Helsinki, Finland).

The concentration of G-actin was determined spectrophotometrically using an absorption coefficient of 0.63 mL·mg−1·cm−1 at 290 nm. The concentrations of gelsolin and tropomyosin were determined by the micro biuret method.

Assays involving F-actin were performed in 100 mm KCl, 2 mm MgCl2, 0.2 mm CaCl2, 1 mm ATP, 10 mm imidazole and 0.2 mm dithiothreitol (pH 7.4).

Proteolytic digestion

For proteolytic digestion, gelsolin and tropomyosin were diluted to 0.5 mg·mL−1 with TM-buffer and PSAM-buffer, respectively, and the concentration of CaCl2 in the solutions was adjusted to 0.2 mm in excess over EGTA. The gelsolin–tropomyosin 1 : 1 mixtures (0.5 mg·mL−1 for each protein) also contained 0.2 mm CaCl2 in excess over EGTA. Digestion of gelsolin, tropomyosin and their mixtures with thermolysin was performed at an enzyme: protein mass ratio of 1 : 200 at 22 °C. The reaction was stopped with 5 mm EGTA. Cleavage with trypsin was performed at an enzyme: gelsolin mass ratio of 1 : 50 at 22 °C. The reaction was stopped with soybean trypsin inhibitor added at a threefold excess over trypsin. The cleavage products were analyzed by SDS/PAGE [52] in 12% regular or 10–20% gradient gels.

Viscometry

Because the viscosity of an F-actin solution strongly depends on the filament length, and fragmentation of filaments by gelsolin leads to a dramatic viscosity drop, viscometry was used to determine the severing activity of gelsolin. Specific viscosity (ηspec) was measured by Ostwald-type capillary viscometers (Cannon Instrument Company, State College, PA, USA) with a sample volume of 1 mL and an outflow time for water of approximately 27 s at 25 °C. The concentration of actin was 24 μm in all cases. Special care was taken to prevent fragmentation of actin filaments by the measurement itself (i.e. during mixing or handling the sample in the viscometer). To control possible measurement-induced fragmentation, the actin reference samples were treated in exactly the same way as the gelsolin-containing samples, and repeated measurements of the actin samples did not cause any decrease in viscosity. The relative actin filament severing activity of gelsolin was assessed by determining the reduction of ηspec of each sample compared to that of F-actin alone (or F-actin + tropomyosin when it was present).

Because the reduction of specific viscosity depends in a nonlinear way on the activity of gelsolin, a more quantitative evaluation of the inhibition of gelsolin activity was established: A calibration curve from viscometry of 10 actin samples (24 μm) containing different amounts of gelsolin (between 24 nm and 1.2 μm) was constructed by plotting the reduction of ηspec for each sample against the molar ratio of actin to gelsolin, and fitting the data to a curve using sigmaplot (Systat Software Inc., Chicago, IL, USA). The relative inhibition of severing activity for any sample was then determined using this curve.

Determination of the relative number concentration of actin filaments

The relative number concentration of the filaments in F-actin solutions was determined by comparing rates of F-actin depolymerization below the critical concentration for polymerization or the abilities of actin filaments to nucleate polymerization of G-actin above its critical concentration. Accordingly, we measured either the initial rates of F-actin depolymerization upon dilution of the solutions below the critical concentration [30] or the initial rates of G-actin polymerization nucleated by F-actin [30, 31]. In the depolymerization assay, 24 μm pyrenyl-labelled actin was initially polymerized in the presence of gelsolin at a molar ratio of gelsolin : actin of 1 : 600 to cap the barbed filament ends. Then F-actin was incubated with gelsolin or gelsolin–tropomyosin for 1 h. Aliquots of pyrenyl-labelled F-actin solutions (50 μL) pre-incubated with gelsolin or gelsolin–tropomyosin were diluted into 2 mL of buffer G complemented with 0.1 m KCl to a final actin concentration of 0.6 μm (which is below the critical concentration for the pointed ends of actin filaments under these conditions). The solutions were gently mixed in fluorescence cuvettes by inversion five times (this resulted in a dead time of approximately 10 s), and the decrease in the intensity of pyrenyl fluorescence associated with F-actin depolymerization was recorded at 25 °C.

In the nucleation assay, aliquots of 12 μm unlabelled F-actin solutions pre-incubated with gelsolin or gelsolin–tropomyosin were diluted 20-fold into 4.8 μm pyrenyl-labelled G-actin supplemented with 0.1 m KCl immediately after the addition of F-actin seeds. The solutions were mixed as described above, and the increase in fluorescence intensity during polymerization of pyrenyl-labelled G-actin onto unlabelled F-actin seeds was recorded at 25 °C. In both the depolymerization and the nucleation assay, the relative number concentration of filaments in F-actin solutions was derived from the slopes of the initial linear parts of the fluorescence curves.

Fluorescence measurements

Fluorescence intensity of pyrenyl-actin was monitored in a PC 5000 recording spectrofluorometer (Shimadzu Corp., Kyoto, Japan) at 407 nm after excitation at 365 nm [49].

Acknowledgements

We are grateful to Dr Ilya Nevzorov for providing us with tropomyosin G126R mutant. This work was partially supported by the Program ‘Molecular and Cell Biology’ of the Russian Academy of Sciences (to S.K.).

Ancillary