The ability to assemble distinct and dynamic actin structures in the same cell is critical for the diverse role of actin in cell motility and morphogenesis [1]. In yeast cells that undergo polarized growth, actin cables and patches are dynamically assembled near growth sites and are required for two different cellular functions: polarized transport and endocytosis, respectively [2]. Actin filaments (F-actin) in cables are nucleated by formins and decorated by tropomyosins, whereas those in patches are nucleated by the Arp2/3 complex and decorated with cofilin and other patch specific markers [3]. Known cable components, Tpm1, Tpm2, Sac6, and Abp140, stabilize and/or bundle actin filaments in vitro and in vivo [6]–[11]. The rapid turnover of actin patches has been shown to depend on cofilin [4], and a recent study found that although cofilin was not detected along actin cables under normal conditions, rapid turnover of actin cables also requires cofilin [5].

Cofilin is a member of the ADF/cofilin family of low molecular weight actin-binding proteins. These proteins can bind to actin monomers (G-actin) and filaments and stimulate depolymerization and fragmentation of F-actin [12] . Every eukaryotic cell type examined to-date contains at least one member of the ADF/cofilin family, suggesting that cofilin performs a universally important function. Null mutations of cofilin in Saccharomyces cerevisiae, Drosophila melanogaster and Caenorhabditis elegans are all lethal. Cofilin-mediated F-actin severing and turnover are thought to be critical for rapid remodeling of actin organization in response to external stimuli, and for recycling of G-actin to sustain actin polymerization during cellular or intra cellular motility processes [13].

Tropomyosin has long been known to protect actin filaments from the depolymerizing and severing actions of DNaseI [14], ADF/cofilin [15], [16] and gelsolin [17]. Tropomyosin is a long protein with two α-helical polypeptides that form a parallel coiled coil, which further associate through head-to-tail interactions to form a long polymer along the side of an actin filament [18]. Whereas yeast has two tropomyosins, Tpm1 and Tpm2, vertebrates have 17 tropomyosin isoforms. In general, tropomyosin inhibits the dissociation of actin subunits from the pointed end of an actin filament [19], [20], which makes actin filament stronger and less likely to bend or break [21], [22]. Tropomyosin may also be a critical factor governing the spatial segregation of Arp2/3- and formin-nucleated actin structures in the cell. Whereas tropomyosin prevents branching nucleation by Arp2/3 complex, possibly due to their overlapping binding sites on F-actin [23], a recent study also suggests that tropomyosin actively facilitates assembly of actin filaments by formin protein [24].

In order to uncover the mechanism governing the disassembly of tropomyosin-bound actin filaments, we used a proteomics approach to identify tropomyosin-binding proteins and found cofilin to be a highly ranked tropomyosin binder. This led us to test if yeast cofilin can directly sever tropomyosin-decorated F-actin. Surprisingly, we found that the major yeast tropomyosin, Tpm1, has little protective effect against cofilin in a variety of assays, however, a cofilin mutant unable to bind Tpm1 showed diminished ability to sever Tpm1-bound F-actin. These results suggest that yeast cofilin is intrinsically capable of promoting the turnover of tropomyosin-bound F-actin in yeast.

Because previous studies suggest that tropomyosin inhibits actin depolymerization by cofilin [15], we hypothesized that additional proteins may be required to facilitate cable turnover, possibly through regulation of tropomyosin binding to actin, and set out to identify such proteins through Tpm1 affinity chromatography. For this and subsequent experiments, we used recombinant Tpm1 purified from E. coli, which we found had similar F-actin-binding properties as Tpm1 purified from yeast cells (data not shown). To minimize proteins that would associate with the Tpm1 affinity column via F-actin, yeast high-speed supernatant was prepared in G-actin buffer and indeed actin was not found among the bound proteins using Multidimensional Protein Identification Technology (MudPIT). Surprisingly, Cof1 was reproducibly detected in two Tmp1 affinity experiments, but not the BSA control columns, and was ranked among the top ten proteins, based on frequency of detection and quantitative NSAF values (Table 1).

To test if the association between Cof1 and Tpm1 was direct, we expressed and purified yeast Cof1 from E. coli and found that it binds to Tpm1-coupled beads but not the control BSA beads (Fig. 1A). Using a quantitative Cof1 pull-down assay [25] the apparent K_d for Cof1 binding to Tpm1 was determined to be 3.34±0.12 µM (Fig. 1B). Despite a weak affinity, this interaction is likely to be specific because it was abolished by the cof1-22 mutation, which was shown previously to inhibit turnover of actin patches and cables in vivo [4], [5], but not the cof1-5 mutation (Fig. 1A). Neither mutation affected the folding stability of Cof1 [26].

The observation of a direct interaction between Cof1 and Tpm1 led us to test whether Cof1 impacts the stability of yeast F-actin decorated with Tpm1 using a co-sedimentation assay. Yeast actin, purified as described [27], was polymerized in the presence of Tpm1 and then incubated with or without an equimolar or excess cofilin for 20 min. In the absence of Cof1, Tpm1 co-sedimented with F-actin in the pellet fraction (Fig. 2A). Upon incubation with Cof1, most Tpm1 no longer sedimented in the pellet, but instead moved with F-actin to the supernatant fraction. Cof1 had a similar effect when incubated with yeast F-actin bound to either the mouse high-molecular-weight tropomyosin, TM1, or low-molecular-weight tropomyosin, TM4. The extents to which actin was solublized by cofilin were identical with or without tropomyosin bound. This is best observed in an experiment where increasing concentrations of Cof1 was used (Fig. 2B). The two yeast cofilin mutants, Cof1-5 and Cof1-22, showed a similar effect in this assay to that of wild-type Cof1 (Fig. 3). In contrast, mouse non-muscle cofilin 1, which is more similar than mouse cofilin 2 to yeast cofilin, showed diminished ability to depolymerize muscle actin bound to TM1 (Fig. 2C,E), TM4 or Tpm1, as expected (Table 2). This result raised the question of whether tropomyosin could indeed protect F-actin against yeast cofilin.

Because the above assay only assessed the steady-state distribution of F- versus G-actin, we next used two assays in which pyrene-labeled F-actin was diluted (see Materials and Methods) to determine the effect of Cof1 on the kinetics of actin depolymerization with or without bound tropomyosin. Light scattering showed that Cof1 drastically increased the rate of F-actin depolymerization after dilution, and Tpm1 binding to F-actin had no effect on the Cof1-induced actin depolymerization (Fig. 4A). Interestingly, measurements of pyrene actin fluorescence in these experiments suggest that binding of Cof1 to F-actin, as indicated by fluorescence quenching [28], was not affected by the presence of Tpm1 (Fig. 4B), On the other hand, mouse tropomyosin 1 reduced the binding of mouse cofilin 1 to F-actin (Fig. 4C), which is consistent with competitive binding of tropomyosin and cofilin to F-actin, as demonstrated by previous studies [16]. Cof1-22 exhibited reduced ability to promote depolymerization of naked actin compared to wild-type Cof1, confirming a previous report [4], yet again, Tpm1 did not alter the depolymerization kinetics in the presence of Cof1-22 (Fig. 4A).

Cofilin possesses both actin depolymerization and severing activities, and the latter requires much lower concentrations of cofilin than the former [29]. We used a microscopy-based assay to test the effect of Tpm1 on actin severing by substoichiometric amounts of Cof1 (see Materials and Methods). Pre-polymerized yeast actin was incubated with Cof1 for various lengths of time. The reactions were stopped by dilution into a buffer containing Alexa488-phalloidin, and the lengths of F-actin were quantified using fluorescence microscopy. Under our experimental conditions, Tpm1 increased the average F-actin length from 9.81±4.56 µm to 15.6±3.45 µm (>100 filaments measured). After addition of 50 nM cofilin to 5 µM F-actin, the average F-actin length decreased rapidly in the presence or absence of Tpm1 (Fig. 5A). Even though Tpm1-bound filaments started longer, the rate of filament severing by Cof1 was similar with or without Tpm1 bound to F-actin. Cof1-5 exhibited an only slightly reduced rate of actin severing, which was not affected by Tpm1 binding (Fig. 5B,C). Interestingly, Cof1-22 only exhibited a moderate defect severing naked F-actin, however, the severing defect was much more enhanced toward Tpm1-bound F-actin, and long filaments were observable even after 40 min of incubation with Cof1-22 (Fig. 5B,C,D).

Tropomyosin-containing actin structures carry out many important functions such as cellular contractility, motility and polarized transport [18]. Current thinking on the turnover of tropomyosin-decorated actin structures is influenced by the notion that tropomyosin protects F-actin from the depolymerizing activity of ADF/cofilin. The data presented above show that tropomyosin does not present a barrier for the action of yeast cofilin, suggesting that Cof1 may be intrinsically capable of mediating the turnover of actin cables. This is consistent with the recent observation of a dramatically reduced rate of cable turnover in cof1 mutant yeast cells, as well as in cells lacking Aip1, an in vivo co-factor that enhances ADF/cofilin-mediated actin fragmentation by capping the barbed ends of severed filaments [5].

The protective effect of tropomyosin against vertebrate ADF/cofilin is thought to be due to competitive binding of tropomyosin and ADF/cofilin to F-actin. We found surprisingly that Cof1 was associated as a high-scoring binder to Tpm1 affinity column, and that purified Cof1 binds directly to tropomyosin in solution. Although the affinity of this interaction is low in solution, it may occur more extensively on the surface of actin cables where there is a high local concentration of tropomyosin. The interaction between Cof1 and Tpm1 is abolished by cof1-22, which encompasses three amino acid substitutions: E134A, R135A, R138A [26]. These mutations were previously shown to reduce the binding of Cof1 to F-actin, and recent structural analysis placed these residues at the F-actin binding interface [30]. Based on these findings, we speculate that whereas the presence of Tpm1 may hinder Cof1's contacts with F-actin, some of these contacts may be replaced by the interaction between Tpm1 and Cof1. The latter could either alter the association of Tpm1 with F-actin or could help stabilize the overall interaction of Cof1 with F-actin in order to cause the structural changes in the filament that lead to fragmentation [31].

A previous study found that cof1-22 also reduces the rate of actin patch turnover, and the extent of this reduction parallels the effect of the mutations on Cof1-induced actin depolymerization in vitro [4]. This is consistent with the high local concentration of Cof1 in actin patches and a specific requirement for Cof1 in the turnover of ADP-actin filaments during the endocytic process [32]. The effect of cof1-22 on the rate of cable turnover, reported recently, appears much more dramatic than that on actin patches [5], and this correlates with the strong defect of the mutant cofilin in severing tropomyosin-decorated actin filaments observed in this study. These observations, together with a lack of localization of Cof1 along actin cables, may suggest that Cof1 promotes cable turnover primarily through its severing activity. Multicellular organisms express many isoforms of ADF/cofilin and tropomyosin due to gene duplication and alternative splicing [13], [18]. Even though some tropomyosin isoforms have been shown to antagonize the activity of ADF/cofilin, certain isoforms appear to actively recruit ADF/cofilin to highly dynamic actin structures [18]. Thus our results further support the view that cofilin and tropomyosin can collaborate in the regulation of rapid actin turnover.