Mechanism of Actin Filament Turnover by Severing and Nucleation at Different Concentrations of ADF/Cofilin

Transcription

1 Molecular Cell 24, 13 23, October 6, 2006 ª2006 Elsevier Inc. DOI /j.molcel Mechanism of Actin Filament Turnover by Severing and Nucleation at Different Concentrations of ADF/Cofilin Ernesto Andrianantoandro 1,4 and Thomas D. Pollard 1,2,3, * 1 Department of Molecular, Cellular and Developmental Biology 2 Department of Molecular Biophysics and Biochemistry 3 Department of Cell Biology Yale University New Haven, Connecticut Summary ADF/cofilins are key regulators of actin dynamics during cellular motility, yet their precise role and mechanism of action are shrouded in ambiguity. Direct observation of actin filaments by evanescent wave microscopy showed that cofilins from fission yeast and human do not increase the rate that pointed ends of actin filaments shorten beyond the rate for ADP-actin subunits, but both cofilins inhibit elongation and subunit dissociation at barbed ends. Direct observation also showed that cofilins from fission yeast, Acanthamoeba, and human sever actin filaments optimally at low-cofilin binding densities well below their K d s, but not at high binding densities. High concentrations of cofilin nucleate actin assembly. Thus, the action of cofilins in cells will depend on the local concentration of active cofilins: low concentrations favor severing, whereas high concentrations favor nucleation. These results establish a clear paradigm for actin turnover by coflin in cells. Introduction Cell motility is founded on the polymerization of actin monomers into filaments that push the leading edge of the cell forward, allowing cells to extend pseudopodia (Pollard and Borisy, 2003). Subunits in filaments must be recycled to maintain a monomer pool and allow further polymerization and motility. Hydrolysis of ATP bound to polymerized actin and release of the g-phosphate (P i ) promote recycling of actin subunits back to the monomer pool, because ADP-actin dissociates faster from barbed ends than ATP or ADP-P i subunits. Filaments of pure actin turn over much slower than most cells move, so cellular factors must drive the recycling process. Observations on bacterial actin comet tails (Carlier et al., 1997; Rosenblatt et al., 1997) and reconstitution of motility in vitro (Loisel et al., 1999) showed that ADF/cofilins increase the rate of actin-based motility. Cofilins also inhibit the exchange of nucleotide bound to actin monomers (Nishida, 1985), accelerate P i release from ADP-P i subunits in filaments (Blanchoin and Pollard, 1999), and sever filaments (Maciver et al., 1991). The effects of cofilins on polymerization of bulk samples *Correspondence: thomas.pollard@yale.edu 4 Present address: Department of Electrical Engineering, Princeton University, J319, EQuad, Princeton, New Jersey, of actin were interpreted as acceleration of the rates of subunit dissociation from the pointed ends (up to 22-fold) and of elongation at barbed ends (Carlier et al., 1997). Dozens of publications cite accelerated pointed-end depolymerization as the accepted mechanism for cofilins to promote actin filament disassembly, but this activity has never been demonstrated directly. Neither the concentration of ends nor the length distributions of filaments were known in the relevant experiments, making it impossible to measure subunit dissociation from individual filament ends. Assays using bulk samples do not isolate the many reactions occurring during polymerization and cannot distinguish between the effects of cofilin on each reaction. Cofilins can also accelerate spontaneous assembly of actin monomers (Carlier et al., 1997; Du and Frieden, 1998; Yeoh et al., 2002), and activation of microinjected caged cofilin increases actin polymer in carcinoma cells (Ghosh et al., 2004). Possible mechanisms for enhancing assembly were severing, faster elongation, or nucleation, but these activities were not individually explored. Here, we observe the various activities of cofilins in isolation, clearly delineating the separate effects of cofilins on subunit association and dissociation from filament barbed and pointed ends, filament fragmentation, and nucleation. Real-time microscopic assays show that cofilins sever filaments optimally at concentrations far below the dissociation equilibrium constant, inhibit depolymerization at barbed ends, and increase subunit dissociation at pointed ends only to the normal rate for ADP-actin. Concentrations of cofilins that bind actin monomers enhance actin filament nucleation. Results To assure that our new assays of cofilin activities revealed properties shared across species, we compared cofilins from three widely divergent species: Acanthamoeba, S. pombe, and human. A 1.7 Å resolution crystal structure of S. pombe cofilin (Supplemental Results in the Supplemental Data available with this article online) verified its similarity to other cofilins. Fortuitously, the electron density also includes the N-terminal serine of the protein, homologous to serine 1 or 3, that can be phosphorylated in other cofilins to inhibit their activity. Nucleotide exchange experiments with actin monomers (Supplemental Results) established that S. pombe cofilin has a higher affinity for ADP-actin monomers than for ATP-actin monomers under physiological salt conditions, much like other cofilins. Filament Severing Disassembly of actin filaments by severing with cofilin is best investigated and analyzed visually. We used TIR fluorescence microscopy to observe severing of individual 25% Oregon Green-labeled ATP-actin filaments polymerized for 6 min. Immobilized proteins anchored these filaments to the glass surface of a flow cell. In experiments with S. pombe cofilin, GST-Cdc12(FH1FH2)p

2 Molecular Cell 14 Figure 1. Actin Filament Severing by Cofilins Observed by Evanescent Wave Fluorescence Microscopy Conditions were as follows: 50 mm KCl, 10 mm imidazole, 1 mm MgCl 2, 1 mm EGTA, 0.2 mm ATP, 0.25% methyl cellulose, 100 mm DTT, 15 mm glucose, 20 mg/ml catalase, and 100 mg/ml glucose oxidase at 22 C and ph 7.0. (A) Fluorescence micrographs of entire microscopic fields of 25% Oregon Green-labeled actin filaments polymerized for 6 min taken at indicated time points after wash in from time-lapse movies of actin filament severing. First row, actin filaments washed with S. pombe cofilin at the optimal severing concentration of 10 nm. Filaments were attached to the glass surface at their barbed ends by S. pombe GST-Cdc12(FH1FH2)p. Second row, actin filaments washed with actophorin at the optimal severing concentration of 100 nm. Filaments were attached to the glass surface by NEM-inactivated muscle myosin II. (B) Detailed montage of one actin filament being severed by 100 nm actophorin over time. anchored barbed ends to the slide (Kovar and Pollard, 2004). In experiments with Acanthamoeba cofilin (actophorin), NEM-myosin anchored one or two points along each filament (Kuhn and Pollard, 2005). We exchanged the volume in the flow cell with buffer containing cofilin to initiate severing. We measured severing activity as breaks per second per micrometer of filament. Ten nanomolar S. pombe cofilin severed most filaments within 90 s (Figure 1A, top row). Each broken filament separated into two fragments (Figure 1B). One segment remained bound to the immobilized proteins, while the other segment diffused away. Severing activity was maximal with 10 nm S. pombe cofilin and fell dramatically at both lower and higher concentrations (Figure 2A). At this optimal cofilin concentration, severing was 15% more active when filaments were anchored at one or two sites along their lengths with low concentrations of NEM-myosin and half as active when filaments were labeled on lysines with Alexa 488. Severing by S. pombe cofilin was reduced 1000-fold with 20 mm phosphate in the buffer (Table 1). Severing was also maximal with 10 nm human cofilin, but the rate was 20-fold lower than S. pombe cofilin. Severing was optimal with a 10-fold higher concentration of actophorin, but the maximum rate was 4-fold lower than with S. pombe cofilin (Table 1). Severing depended on filament length for lengths <16 mm. Longer filaments severed with the maximum probability (data not shown). The dependence of the severing probability on filament length was nonlinear. Plots of log severing probability versus log length had slopes of 1.4 for both S. pombe cofilin and actophorin (Figure 2B). This slope is the exponent of length (L) where severing probability k sev = k(l n ). To determine the density of cofilins on actin filaments during severing, we analyzed binding with an unlimited nearest-neighbor cooperativity model (Kowalczykowski

3 Nonmass Action Mechanism of ADF/Cofilin Activity 15 Figure 2. Analysis of Actin Filament Severing by Cofilins (A) Dependence of severing activity on the concentration of S. pombe cofilin under the conditions of Figure 1. Degree of saturation of the subunits in the filaments was calculated by using Equation. 2. The units of severing activity are severing events per mm of actin filament per second. Optimal severing activity occurs at 10 nm S. pombe cofilin with cofilin bound to 0.12% of the actin subunits. No severing occurred without cofilin or with 1 mm S. pombe cofilin, which saturates 49% of the actin subunits. (B) Log-log plot of the dependence of the severing probabilities of S. pombe cofilin (filled circles) and actophorin (empty circles) on actin filament length under the conditions of Figure 1. Slopes are the exponents of length in the equation k sev =k(l n ). For both cofilins tested, severing probability is proportional to L 1.4 (gray lines). Dashed lines are fits to exponents of L 1 and L 2. (C and D) Kinetics of S. pombe cofilin binding to and dissociating from ADP-actin filaments (generated by aging ATP-actin filaments for 1 hr). Conditions were as follows: 10 mm imidazole, 2 mm Tris-HCl, 50 mm KCl, 1 mm MgCl 2, 1 mm EGTA, 0.2 mm ATP, 0.5 mm DTT, and 1 mm NaN 3 at 22 C and ph 7.0. (C) Kinetics of 0.2 mm S. pombe cofilin binding to pyrenyl ADP-actin filaments monitored by quenching of fluorescence. The inset shows the time course of the rapid decrease in fluorescence after mixing of 0.2 mm S. pombe cofilin with 6 mm pyrenyl ADP-actin filaments fit to a single exponential. The main plot shows the dependence of k obs for cofilin binding to pyrenyl ADP-actin filaments on the concentration of actin polymer. (D) Time course of dissociation of 2 mm S. pombe cofilin from 2 mm pyrenyl ADP-actin filaments. Labeled filaments were incubated with S. pombe cofilin for 5 min then mixed with 20 mm unlabeled actin filaments. The time courses for 2 mm, 0.4 mm, and 0.2 mm S. pombe cofilin fit to double exponentials, and rate constants were extracted from the large amplitude fast phase with an average k 2 =0.11s 21. (E and F) Cooperative binding of S. pombe cofilin to ADP-actin filaments (generated by aging ATP-actin filaments for 1 hr) at equilibrium. Conditions as in (C) and (D). (E) Titration of 2 mm pyrenyl actin filaments with 50 nm to 10 mm S. pombe cofilin. Binding density (fraction of actin bound determined from the extent of quenching of fluorescence) is plotted versus free cofilin concentration and curve fit by using Equation 2. (F) Calculation of the cluster size distribution. Fraction of bound cofilin in a cluster of a given size is plotted versus cluster size for actin lattice saturations of 0.12% (solid line), 1.4% (dotted line), and 49% (dashed line) corresponding to severing assay conditions with 10 nm, 100 nm, and 1 mm S. pombe cofilin and 0.4 nm actin. The inset shows the full cluster size distribution for 49% lattice saturation. et al., 1986; McGhee and von Hippel, 1974), first applied to actin and cofilin by De La Cruz (2005). The McGheevon Hippel equation for binding sites of size 1 (Equation 1) was reformulated in terms of binding density as a function of free cofilin concentration, intrinsic binding affinity (K a for isolated cofilin binding the actin lattice), and a cooperativity parameter (w) (Equation 2). All kinetic and equilibrium binding experiments were done with pyrenyl ADP-actin filaments (generated by aging ATP-actin filaments for 1 hr). We calculated the association equilibrium constant K a for isolated S. pombe cofilin binding to the actin lattice from the association and dissociation rate constants determined by observing the kinetics of binding and dissociation (Figures 2C and 2D; Supplemental Results). The time course of association of cofilin was followed by mixing 0.2 mm S. pombe cofilin with 2 8 mm actin and observing quenching of pyrene fluorescence (Figure 2C, inset). Fitting these time courses to single exponentials gave the apparent rate constant k obs at each concentration of filaments. The dependence of k obs on the concentration of actin yielded an association rate constant, k + = mm 21 s 21 (Figure 2C). To follow the time course of spontaneous dissociation of S. pombe cofilin from 2 mm

4 Molecular Cell 16 Table 1. Severing Activity in Severing Events/mm Actin/Second and Optimal Severing Concentration (nm of Cofilin) as a Function of Nucleotide State of Actin Filament and ADF/Cofilin Isoform Nucleotide in Buffer ADF/Cofilin Isoform Optimal Severing Concentration Severing Activity ATP Human cofilin 10 nm ATP actophorin 100 nm ATP S. pombe cofilin 10 nm ADP S. pombe cofilin 10 nm ATP+ P i S. pombe cofilin 10 nm To form filaments, ATP-actin was polymerized for 6 minutes (when ATP or ATP+Pi was in the buffer) or polymerized and aged for 1 hr to generate ADP-actin filaments (when ADP was in the buffer). pyrenyl actin, we added 20 mm unlabeled filaments to bind dissociating cofilin and observed the recovery of pyrene fluorescence (Figure 2D). The average dissociation rate constant (k 2 ) for three concentrations of S. pombe cofilin was 0.11 s 21. The ratio of k + to k 2 gave K a = 0.12 mm 21. To measure equilibrium binding of S. pombe cofilin to actin filaments, we titrated 2 mm actin with 50 nm to 10 mm S. pombe cofilin and observed the quenching of pyrene fluorescence. Fitting the titration data to Equation 2 with the constraint of the K a from kinetics (0.12 mm 21 ) gave a cooperativity parameter (w) of 8.5 (Figure 2E). A cooperativity parameter of 1 indicates noncooperative binding. Cooperative binding along a linear lattice favors the formation of clusters of ligands (Kowalczykowski et al., 1986). For a given saturation of lattice, Equation 3 predicts the fraction of bound cofilins in a cluster of a given size (Figure 2F). At low binding densities, single S. pombe cofilins bound in an isolated fashion, but even at 49% saturation, most clusters consisted of ten or fewer cofilins (Figure 2F, inset). This indicates very weak cooperativity compared to some DNA binding proteins with w = 1000 or more and cluster sizes of 100 ligands or more (Kowalczykowski et al., 1986; McGhee and von Hippel, 1974). At 10 nm, S. pombe cofilin bound to only 0.12% of actin subunits (Equation 2) and severing was maximal. At 100 nm, S. pombe cofilin bound to 1.4% of the actin subunits and the severing rate was 40-fold lower. At 1 mm, S. pombe cofilin bound to 49% of actin subunits and no severing was observed. During maximal severing, a single S. pombe cofilin molecule (cluster size = 1) is bound for every 833 subunits and the absolute severing rate (see Experimental Procedures)is severing events per micrometer of filament per second. This corresponds to one severing event per 75,000 subunits per second or one severing event per bound cofilin every w90 s. Dissociation of Subunits from Actin Filament Ends in the Presence of ADF/Cofilins Cofilin may also disassemble actin filaments by increasing subunit dissociation from pointed ends (Carlier et al., 1997). We tested this hypothesis by observing shortening filaments by TIRF microscopy in the presence of vitamin D binding protein to sequester free-actin monomers. To identify filament ends, we observed filaments growing over 6 min from ATP-actin with or without saturating concentrations of cofilin. With actin alone, most subunits likely bind ADP-P i, due to rapid ATP hydrolysis and slow dissociation of P i. Cofilin promotes P i release, so most subunits in saturated filaments bind ADP. Incorporation of biotinylated actin monomers bound to streptavidin on the glass surface anchored the growing filaments, so we could observe both barbed and pointed ends shortening when we washed out free monomers with control buffer or buffer with cofilin. Compared with control barbed ends shortening at 0.41 s 21 (Figure 3A), human cofilin inhibited barbedend depolymerization 6-fold (Figure 3B) and S. pombe cofilin inhibited barbed-end depolymerization 14-fold (Figure 3C). A two-tailed Student s t test showed a statistically significant difference in the distributions of instantaneous depolymerization rates (Figures 3G 3I) for controls and filaments treated with either cofilin (p = <0.0001). Compared with the control pointed-end dissociation rate of 0.06 s 21 (Figure 3D), dissociation was 4.8-fold faster with human cofilin (Figure 3E) and 5.5-fold faster with S. pombe cofilin (Figure 3F). A two-tailed Student s t test showed statistically significant differences in the distributions of instantaneous depolymerization rates (Figures 3J 3L) between controls and filaments treated with human cofilin (p = 0.006) or S. pombe cofilin (p = 0.039). We confirmed this effect of cofilins on pointedend depolymerization with observations of filaments polymerized from ATP-actin for 6 min with their barbed ends anchored to the slide by GST-Cdc12(FH1FH2)p. A 1000-fold range of S. pombe cofilin concentrations increased the rate of pointed-end subunit dissociation from 0.06 s 21 (the rate for ADP-P i -actin subunits) to an average of 0.34 s 21 (Table 2). The values for pointedend depolymerization of control filaments and filaments treated with cofilin closely approximate values for pointed-end depolymerization of aged actin filaments with ATP in the buffer (0.06 s 21 ) and with ADP in the buffer (0.26 s 21 )(Kuhn and Pollard, 2005). ADP-actin dissociates from pointed ends at 0.27 s 21 (Pollard, 1986). Effects on Actin Filament Elongation and Critical Concentration In addition to its effects on subunit disassembly, cofilin reportedly increases the rate of bulk polymerization, perhaps by increasing the rate of filament elongation (Carlier et al., 1997). We tested this hypothesis with TIRF microscopy of growing filaments anchored by incorporation of biotinylated monomers bound to streptavidin on the slide. We used rates of growth prior to adding actin monomers saturated with cofilins to identify the barbed and pointed ends. Neither S. pombe nor human cofilin accelerated elongation at either end of growing filaments. The rate constant for barbed-end elongation was the same for actin alone (Figure 4A) and actin with 32 mm human cofilin (Figure 4B) and lower by a factor of two with 24 mm S. pombe cofilin (Figure 4C). The critical concentrations (Cc = k 2 /k + ) differed at most by a factor of 2 (Figures 4A 4C). The slow elongation of pointed ends was difficult to measure accurately, but rates were similar in the presence and absence of cofilins. Cofilin has little or no effect on elongation, thus its effect on bulk polymerization must occur in some other way.

5 Nonmass Action Mechanism of ADF/Cofilin Activity 17 Figure 3. Depolymerization of Actin Filaments in the Presence of Cofilins Observed by Evanescent Wave Fluorescence Microscopy Conditions as in Figure 1. ATP-actin filaments were grown for 6 min from 1.5 mm 50% Oregon Green-labeled actin alone, or with 32 mm human cofilin, or with 24 mm S. pombe cofilin. These samples were washed with buffer alone or with 10 mm human cofilin or with 10 mm S. pombe cofilin and observed. The data are presented in two ways: (A) (F), time course of length versus time, where dark gray lines are the global fit rates with dark dotted lines at 95% confidence intervals and light gray lines are the average individual filament rates with light dotted lines at minimum and maximum slopes and dashed lines indicating standard deviations, and (G) (L), histograms of the distribution of measured length changes during each 9 s interval. (A C) Barbed ends. (A) Actin filaments alone, (B) actin with 10 mm human cofilin, and (C) actin with 10 mm S. pombe cofilin. (D F) Pointed ends. (D) Actin filaments alone, (E) actin with 10 mm human cofilin, and (F) actin with 10 mm S. pombe cofilin. (G I) Barbed ends. (G) Actin filaments alone, (H) actin with 10 mm human cofilin, and (I) actin with 10 mm S. pombe cofilin. (J L) Pointed ends. (J) Actin filaments alone, (K) actin with 10 mm human cofilin, and (L) actin with 10 mm S. pombe cofilin. Effects on Actin Filament Nucleation We observed many more filaments by microscopy in polymerization experiments with cofilin than actin alone (Figures 4D 4F). No severing occurred in these experiments with high concentrations of cofilin, so these filaments formed de novo by nucleation. To quantify nucleation activity, we measured the generation of new filament ends by diluting samples 1000-fold into 2 mm pyrene-labeled ATP-actin monomers and observing the rate of elongation (Blanchoin and Pollard, 1999; Ressad et al., 1999). With 6 mm Mg-ATP-actin monomers and 10 mm S. pombe cofilin, the concentration of ends plateaued within 10 min (Figure 4G). Under these conditions, 10 mm S. pombe profilin delayed nucleation for w50 s but produced the same number of ends (Figure 4G, inset). With a fixed concentration of S. pombe cofilin, the dependence of the initial rate of end formation on ATP-actin monomer concentration had a sharp inflection point at a stoichiometry of two cofilins to one actin (Figure 4H). The nucleation activity of human cofilin was slightly lower than S. pombe cofilin, and S. pombe cofilin with glutamate substituted for R78 and K80 inhibited rather than promoting actin nucleation (Supplemental Results). These conserved basic residues are required for human and S. cerevisiae cofilins to bind actin filaments.

6 Molecular Cell 18 Table 2. Average Barbed-End and Pointed-End Depolymerization Rates (Change in Filament Length over Time in Subunits per Second) in the Presence of ADF/Cofilins at the Indicated Concentrations ADF/Cofilin Isoform Barbed-End Rate Pointed-End Rate Actin alone sub/sec sub/sec 1 mm human cofilin sub/sec 2 10 mm human cofilin sub/sec sub/sec 10 nm S. pombe cofilin sub/sec 100 nm S. pombe cofilin sub/sec 1 mm S. pombe cofilin sub/sec sub/sec 10 mm S. pombe cofilin sub/sec sub/sec ATP-actin was polymerized for 6 min before depolymerization. ATP was in the buffer for the course of the assay. Although actin fully labeled with rhodamine on cysteine 374 is incapable of polymerization (Otterbein et al., 2001), it formed short filaments in the presence of S. pombe cofilin that we observed by epifluorescence microscopy (data not shown) and quantified by measuring their activity as seeds for elongation of pyrene-actin (Figure 4I). The concentration of ends increased linearly with the concentration of S. pombe cofilin. Discussion We investigated the biochemical activities of cofilin to provide explanations founded on thermodynamics and polymer mechanics for interpreting studies of cofilin in cells. Although further mechanistic insight (e.g., structure function) may be gained from investigating ph dependence and exhausting all possible conditions and reactions, we focus on those most relevant to the cellular environment. Because cofilin has different effects on actin over a wide range of concentrations, it is important to examine each effect in isolation. Our experiments isolate and characterize cofilin binding, severing, effects on subunit association and dissociation, and nucleation. Many previous experiments on cofilin did not isolate these reactions, making it difficult to interpret the results. Our observations allow us to propose a molecular mechanism of actin turnover in cells that forms the basis for sound alternatives to past explanations for cofilin function. Mechanism of Severing and Its Impact on Actin Filament Disassembly Despite early evidence for severing, its causal role in filament disassembly was largely ignored due to the popularity of the assumption that cofilin accelerates subunit dissociation from filament pointed ends. The best evidence for increased subunit dissociation comes from single-turnover experiments measuring the exchange of fluorescent or radiolabeled ADP bound to actin filaments incubated with cofilin for ATP in the medium (Blanchoin et al., 2000; Carlier et al., 1997; Yeoh et al., 2002). The nucleotide exchange rate rises with cofilin concentration, peaks at a ratio of cofilin:actin of 1:2, and drops at higher ratios, even in the presence of barbed-end capping agents. Nucleotide exchange normally occurs only on free-actin monomers, not subunits in filaments, thus the nucleotide bound to actin subunits in filaments exchanges only after the subunits dissociate from the filament. The increase in nucleotide exchange rate with cofilin was interpreted to arise from rapid dissociation of monomers (Carlier et al., 1997), but such experiments did not measure subunit dissociation from individual filaments, or the number or lengths of the filaments. We find that subunit dissociation from individual filaments in the presence of cofilin is too slow to account for the observed rates of nucleotide exchange without any change in the number and length of filaments. Both severing and nucleation by cofilin are capable of producing many filament ends. A combination of these effects at steady state will likely produce filament length distributions responsible for the results of singleturnover assays. This deserves further analysis with mathematical models of the actin cycle that account for all of the individual reactions. If further work shows that increased filament number cannot account for the single-turnover results, then cofilin may increase the rate of nucleotide exchange within actin filaments. Our results suggest that severing by cofilin should be considered as an alternative to accelerated subunit dissociation for the mechanism of actin filament disassembly. Severing is optimal at a concentration of S. pombe cofilin 100-fold lower than the apparent K d for cooperative binding (1/wK a = 1.02 mm). The probability of severing is lower at lattice saturations below the optimum concentration due to a paucity of binding events, whereas saturations greater than the optimum stabilize the filaments. We propose a severing mechanism based on alteration of actin filament twist that is consistent with cooperative binding (De La Cruz, 2005), severing at nanomolar concentrations of several cofilins (Ichetovkin et al., 2002; Orlova et al., 2004), the effects of cofilin on thermal stability of actin filaments (Dedova et al., 2004), and the effects of cofilin on torsional rigidity of actin filaments (Prochniewicz et al., 2005). Actin filaments bend and twist (Gittes et al., 1993; Rebello and Ludescher, 1998), but flexural rigidity is greater than torsional rigidity (Prochniewicz et al., 2005). The characteristic length scale for bending is the persistence length. At length scales below the persistence length, bending is negligible, because it occurs in higher elastic modes at very low amplitude (Gittes et al., 1993). Severing depends on filament length (Figure 2B) below the 16.7 mm (Ott et al., 1993) persistence length, so the amplitude of higher-mode bending is unlikely to disrupt filament structure. Thus, an effect of cofilin on actin filament twist is a more attractive explanation for the observed length dependence of severing. Saturation with cofilin shifts average helical twist of actin filaments by five degrees per subunit (McGough et al., 1997). Filaments partially decorated with cofilin have the same mean twist as fully decorated filaments, so single cofilins impose a change in twist that propagates along the filament (Galkin et al., 2001). These studies describe variability of twist as static disorder, but information about torsional dynamics is necessary to understand the mechanism of severing. Transient phosphorescence anisotropy showed that binding of single cofilins increases the torsional flexibility of hundreds of adjacent subunits (Prochniewicz et al., 2005). Cofilin shifts the equilibrium distribution between

7 Nonmass Action Mechanism of ADF/Cofilin Activity 19 Figure 4. Effect of Cofilins on Actin Filament Elongation and Nucleation (A C) Elongation was observed by evanescent wave fluorescence microscopy with 50% Oregon Green-labeled ATP-actin. Conditions as in Figure 3. Dependence of the elongation rate on the concentration of unlabeled actin: (A) actin alone, (B) actin with 32 mm human cofilin, and (C) actin with 24 mm S. pombe cofilin. Filled circles represent barbed-end rates and filled triangles represent pointed-end rates. Slope = k +, y-int. = k 2, and x-int. = Cc. (D F) Nucleation of actin filaments by cofilins. Evanescent wave fluorescence micrographs of filaments formed by 0.8 mm 50% Oregon Greenlabeled ATP-actin after 400 s in polymerization buffer. Conditions as in Figure 1. (D) Actin alone, (E) actin with 32 mm human cofilin, and (F) actin with 24 mm S. pombe cofilin. (G I) Quantitation of barbed ends formed by nucleation by dilution of samples into 2 mm 25% pyrenyl ATP-actin monomers and measurement of the rate of elongation. Conditions for nucleation and subsequent elongation were as in Figures 2C and 2D. (G) Time course of barbed-end formation by 6 mm ATP-actin with 10 mm S. pombe cofilin fit with a hyperbola. Inset, time course of ends formed by 6 mm ATP-actin with 10 mm S. pombe cofilin (filled circles), or with 10 mm S. pombe cofilin and 10 mm S. pombe profilin (filled triangles). (H) Dependence of end formation on actin monomer concentration. A range of ATP-actin concentrations was polymerized for 30 s with 10 mm S. pombe cofilin. (I) Dependence of the rate of end formation by 10 mm rhodamine-actin on S. pombe cofilin concentration. Samples were polymerized for 60 s before dilution. thermally equivalent but structurally distinct conformers of filaments by binding preferentially to one of the two states (Prochniewicz et al., 2005). Filaments longer than a few hundred subunits have sections in both of these conformations, with their abundance depending on the degree of saturation by cofilin. Each section can be thought of as a different phase, C (cofilin preferred) or N (not preferred), because one conformation preferentially binds cofilin and allows greater degree and speed of torsion than the other. According to the thermodynamics of interfaces, interactions between subunits within each phase will be

8 Molecular Cell 20 more thermally stable than at interfaces between phases (Berry et al., 2000). Thus, severing can relieve strain if dissociation of interfacial bonds requires less energy than dissociation of cofilin from the filament. Severing eliminates strained interfaces, leaving products with lower energy. Increasing cofilin binding density can also eliminate the strained interfaces by conversion of the entire filament into C phase. The entirely C phase filament may be more thermally stable (Dedova et al., 2004) than a bare actin filament. Binding of cofilin thus acts as a thermomechanical band-pass filter for filament disassembly, restricting severing activity to low concentrations of active cofilin, contrary to the expectations of mass action kinetics. The mechanism of severing at low cofilin concentrations explains the involvement of cofilin in actin filament disassembly, as follows. Under nonphysiological depolymerizing conditions, when actin filaments are brought below their critical concentration, severing at a low optimal cofilin concentration liberates barbed and pointed ends for depolymerization, promoting bulk disassembly. Under polymerizing conditions in cells, severing in conjunction with barbed-end capping by capping protein, gelsolin, or Aip1 may increase bulk disassembly by multiplying the number of pointed ends available for depolymerization. These conditions favor bulk pointed-end disassembly, particularly if the monomer pool is bound to profilin and thymosin-b4 (Pantaloni and Carlier, 1993). Mechanism of Nucleation and Its Impact on Actin Assembly Just as there is debate as to how cofilins promote actin filament disassembly, cofilins could promote actin filament assembly in a variety of ways. Cofilins might increase the rate of elongation (Carlier et al., 1997), but we observed that cofilin actually slows barbed-end elongation of individual filaments. Severing by cofilin might create more ends for polymerization (Ichetovkin et al., 2002) as suggested to explain an increase of actin polymer in carcinoma cells after EGF stimulation or local release of microinjected caged cofilin (Ghosh et al., 2004; Zebda et al., 2000). Severing contributes to the formation of new ends at nanomolar concentrations of cofilin, but another mechanism is required for micromolar concentrations of cofilin, which do not sever, and to explain how mixtures of cofilin and actin monomers produce ends rapidly without a lag in time. All our evidence is consistent with promotion of nucleation as the main contribution of cofilin to increasing the rate of spontaneous actin polymerization. We suggest that cofilins stimulate nucleation by stabilizing long-pitch actin dimers, the first intermediate in spontaneous assembly (Sept and McCammon, 2001). Cofilin likely lowers the rate of dissociation of these dimers. Several lines of evidence support this interpretation. (1) Reconstructions of electron micrographs show cofilins bound between subunits along the long pitch helix of the actin filament, contacting subdomain 2 of one subunit and subdomain 1 of the next subunit (McGough et al., 1997). (2) Substitution of two conserved basic residues with alanine, uncharged, or acidic residues reduces affinity for actin filaments, but not monomers, indicating a site on cofilins specific for binding filaments (Lappalainen et al., 1997; Pope et al., 2000). S. pombe cofilin with these two basic residues replaced by glutamic acid inhibits rather than promotes de novo actin nucleation (Supplemental Results). (3) Nucleation activity lessens at ratios of cofilin:actin >2:1, indicating that excess cofilin may bind a second cofilin binding site on actin monomers, preventing dimerization of doubly bound monomers. (4) Cofilin induces nucleation of rhodamine-actin, perhaps by substituting for a missing actin-actin contact compromised by rhodamine. Actin Filament Turnover in Cells Our results are inconsistent with the hypothesis that cofilins accelerate treadmilling of individual filaments. We propose that cofilins contribute to the collective turnover of actin filaments in cells by promoting disassembly by severing and assembly by nucleation. Low concentrations of cofilin favor severing, whereas high concentrations favor nucleation (Figure 5) as well as P i release from ADP-P i filaments (Blanchoin and Pollard, 1999) and dissociation of branches formed by Arp2/3 complex (Blanchoin et al., 2000). The usual assumption that mass action drives function does not hold in this system, so cofilin has different effects on actin depending on the local concentration of active cofilin throughout a cell. For both components of turnover to occur at the same time, active cofilin must be differentially distributed in space, along a concentration gradient. Conversely, for the components of turnover to occur uniformly in space, active cofilin concentration might be differentially distributed, in time, through oscillations. Observations on cells may favor the former possibility (Supplemental Discussion). This actin turnover mechanism offers tools to make hypotheses and interpret findings in future investigations of cofilin function in cells and an opportunity to reconsider interpretations of previous work. For example, nucleation is more likely than severing to explain how local release of a high concentration of cofilin stimulates actin assembly in carcinoma cells (Ghosh et al., 2004). On the other hand, neither severing nor rapid pointedend depolymerization can explain how high concentrations of cofilin reduce the length of actin comet tails of Listeria in cell extracts (Carlier et al., 1997; Rosenblatt et al., 1997). The debranching activity of cofilin (Blanchoin et al., 2000) should be considered as an alternative mechanism. Reducing expression of the adf1 gene in fission yeast increases the filamentous structures in fission yeast (Nakano and Mabuchi, 2006). The authors attributed this change to a lack of depolymerizing activity, but extensive severing near the optimal concentration is a likely alternative. Overexpression of cofilin not only diminished some actin filament structures but also produced large bars containing GFP-cofilin. These bars stained with actin antibody, but not phalloidin. The nucleation activity of high-cofilin concentrations may have produced aberrant actin bars saturated with cofilin, thus excluding phalloidin (Nishida et al., 1987). A high concentration of cofilin may also compromise organized actin filament structures by excluding tropomyosin (an essential component of cables) and myosin II (an essential component of the contractile ring) (Nishida et al., 1984).

9 Nonmass Action Mechanism of ADF/Cofilin Activity 21 Figure 5. Concentration Dependence of Cofilin Activity Very low concentrations of cofilin (blue ovals) do not bind actin filaments (green). At an optimal low concentration of cofilin, single cofilins bind and sever actin filaments. Capping of severed filament barbed ends promotes dissociation of actin monomers (green circles) from pointed ends, leading to filament disassembly. Higher concentrations of cofilin bind cooperatively to actin filaments and promote the release of inorganic phosphate (P i ) but do not sever them. Very high concentrations of cofilin bind actin monomers and stimulate nucleation, leading to actin filament assembly. Experimental Procedures Evanescent Wave Microscopy Flow cells were constructed from coverslips and acidcleaned slides (Kuhn and Pollard 2005). Stretched strips of Parafilm form the side walls of the chamber. Flow-cell volumes were between 5 and 7 ml. All actin experiments were done in polymerization buffer: 50 mm KCl, 10 mm imidazole (ph 7.0), 1 mm MgCl 2, 1 mm EGTA, 0.2 mm ATP, 0.25% methyl cellulose, 100 mm DTT, 15 mm glucose, 20 mg/ml catalase, and 100 mg/ml glucose oxidase. Images of filaments were captured with an Orca ER CCD camera (Hamamatsu Corp., Bridgewater, NJ) on an Olympus IX70 microscope with prism-based total internal reflection illumination. The shutter on the exciting beam was opened every 9 s for 150 milliseconds to acquire images during severing, depolymerization, or elongation. At least two movies were taken for each experiment, done on different days with different preparations of proteins. Filament lengths were measured manually with Image J and plugins developed by Jeffery Kuhn (Kuhn and Pollard, 2005). Severing Assays and Analysis Flow cells were prepared by filling with a succession of solutions. First, the glass was coated with an anchor for actin filaments, either 250 nm GST-Cdc12p(FH1FH2) (Kovar and Pollard, 2004) or 200 nm NEM-myosin, for 5 min. The surface was then blocked with either two volumes of 1% BSA or one volume of 2% BSA, followed by either another two volumes of 1% BSA or one volume of 2% BSA. Then, the chamber was washed with two volumes of polymerization buffer. ATP-actin monomers 25% labeled with Oregon Green were polymerized in the chamber by mixing equal parts 23 polymerization buffer and actin solution, then drawing the mixture into the flow cell by capillary action using filter paper. After filaments grew to lengths of mm, one volume of cofilin in polymerization buffer was added to the flow cell, displacing free monomers and unattached filaments. Methyl cellulose in the polymerization buffer and nonspecific electrostatic interactions kept the filaments near the surface of the slide. In experiments with ADP-P i -actin, ADP-actin subunits were saturated with phosphate by including 20 mm potassium phosphate and 25.9 mm KCl (to balance the ionic strength) in polymerization buffer. Control samples had 20 mm K 2 SO 4. ADPactin filaments were generated by polymerizing ATP filaments in the flow cell then washing out free monomer and aging for 1 hr in the presence of the critical concentration of actin in polymerization buffer. We counted severing events by eye and observed at most one event per filament per frame. Filament lengths were measured manually (see above). The total filament content in all frames was estimated from the filament lengths in the first frame after adding cofilin to the chamber, multiplied by the total number of frames. Severing activity was calculated by summing the total number of severing events in all frames, dividing by the total filament content to yield an average number of severing events per mm of filament per frame, then dividing by time interval between frames to yield severing events per mm of filament per second. We used this to compare

10 Molecular Cell 22 relative severing activities for different types and concentrations of cofilins under various conditions. A more accurate estimate for absolute severing activity of severing events/mm actin/second at the optimal 10 nm S. pombe cofilin used a total filament content value derived from automated individual length measurements for the entire time course (below). Length dependence of severing was determined by automated measurement in Image J of areas and perimeters of every filament in individual frames of the optimal severing concentrations of S. pombe cofilin and actophorin. Severed filaments were binned according to length and divided by total filaments binned the same way to generate histograms of severing probability versus length. Analysis of Cooperative Binding We analyzed cofilin binding to pyrene-actin filaments with the McGhee-von Hippel unlimited nearest neighbor cooperativity model, expressed in Scatchard format for binding sites of size 1 as v=½lš =K a ð1 2 vþ½ð1 2 2v + RÞ=ð2 2 2vÞŠ 2 (1) where R = ðð1 2 2vÞ 2 + 4wvð1 2 vþþ 1=2, v is the fraction of actin lattice bound (binding density), K a is the equilibrium constant for isolated cofilin binding to actin filament, w is the cooperativity parameter, and [L] is the concentration of free cofilin (calculated from total cofilin concentration and concentration of bound actin). Algebraic rearrangement using Mathematica (Wolfram Media) yields v=1=2ð1+ð 2 1+K a ½LŠwÞ=ð1+K a ½LŠð4+wð 2+K a ½LŠwÞÞÞ 1=2 Þ: (2) The titration curve of v versus [L] was fit to this equation by using nonlinear least squares, with K a = mm 21 (determined from the ratio of the dissociation and association rate constants measured kinetically). The cooperativity parameter w was determined from the best fit of the data. Cluster size distribution (fraction of bound cofilin as a function of cluster size) was determined by applying F c =cð1 2 b n b 1 Þ 2 ðb n b 1 Þ c 2 1 (3) where F c is the fraction of bound cofilin in a cluster, c is the cluster size (number of cofilins), and b n b 1 is the conditional probability that a bound ligand will be located next to another bound ligand on the right (Kowalczykowski et al., 1986). For binding sites of size 1, b n b 1 = ð1 2 ð2 2 2wÞv 2 RÞ=ð2vðw 2 1ÞÞ where v is the fraction of actin lattice bound (binding density), w is the cooperativity parameter, and R = ðð1 2 2vÞ 2 + 4wvð1 2 vþþ 1=2. Depolymerization and Elongation Assays and Analysis Flow cells were prepared and treated with 3 mg/ml biotinylated BSA (Supplemental Experimental Procedures) for 5 min, washed with two volumes of 0.33 mg/ml streptavidin (Molecular Probes, Invitrogen), one volume of 100 nm biotinylated actin monomers (Supplemental Experimental Procedures), and four volumes of polymerization buffer. Actin monomers 50% labeled with Oregon Green were then polymerized in the chamber by mixing equal parts 23 polymerization buffer and actin solution, then drawing the mixture into the flow cell by capillary action using filter paper. Filaments were anchored to the slide during polymerization by incorporation of biotinylated actin monomers sparsely attached to the streptavidin on the surface. For elongation assays, a range of concentrations of actin monomers 50% labeled with Oregon Green were observed during polymerization in the absence and presence of cofilins. For depolymerization assays, filaments were observed growing for 6 min to lengths of mm, to mark the fast-growing barbed ends. Depolymerization from both filament ends was initiated by adding one volume of 1 mmor8mm vitamin D binding protein (alone or supplemented with cofilin) in polymerization buffer to the flow cell, displacing the free monomer and unattached filaments. Vitamin D binding protein (Calbiochem) captured dissociating monomers. To monitor pointed-end depolymerization from filaments capped at their barbed ends with GST-Cdc12(FH1FH2)p, we used the slide preparation and filament growth procedures and conditions from the severing assays (above), then proceeded as with the depolymerization assay. Filament lengths were measured manually (see above). Barbedand pointed-end shortening were measured relative to a set point on a stationary filament. For pointed-end depolymerization from filaments capped at their barbed ends by GST-Cdc12(FH1FH2)p, total lengths were measured. Dissociation rate constants were determined from plots of filament length versus time for ten filaments and time intervals for pointed-end depolymerization or filaments and time intervals for barbed-end depolymerization. Filament growth rates of actin-cofilin complexes for filaments and time intervals were plotted versus concentration to yield association rate constants and critical concentration. Nucleation Assays The assays were carried out in two steps. First, unlabeled actin was polymerized with cofilin in F buffer (Supplemental Experimental Procedures) for a given time at 22 C. Actin concentration, cofilin concentration, and incubation time varied depending on tested parameters. An aliquot of the first reaction was diluted 1:100 into F buffer, then 1:10 into F buffer containing 2 mm 25% pyrenyl actin monomers, for which the rate of spontaneous polymerization is much slower than the rate of elongation from the added seeds. Polymerization was followed by increase in pyrenyl actin fluorescence on a PTI Alpha-scan spectrofluorometer with excitation at 365 nm and emission at 406 nm. The number concentration of filaments (N) was calculated from the initial rate of polymerization according to Rate = k + ½AŠ½NŠ 2 k 2 ½NŠ (4) where k + is the association rate constant (10 mm 21 s 21 ), k 2 is the dissociation rate constant (1 s 21 ), and [A] is the actin monomer concentration. Supplemental Data Supplemental Data include Supplemental Results, Supplemental Discussion, Supplemental Experimental Procedures, Supplemental References, five figures, and one table and can be found with this article online at 13/DC1/. Acknowledgments This work was funded by National Institutes of Health research grant GM26338 to T.D.P. We thank Enrique De La Cruz, Mark Mooseker, Paul Forscher, and Joao Cabral for advice, Jeff Kuhn for invaluable training and advice, Eric Dufresne, Rachel Mahaffy, and Dimitris Vavylonis for insightful discussion and critique, Brad Nolen for assistance with Molscript and rmsd calculations, David Kovar for providing GST-Cdc12(FH1FH2)p, and Ikuko Fujiwara for providing biotinylated actin. Received: December 2, 2005 Revised: June 1, 2006 Accepted: August 4, 2006 Published: October 5, 2006 References Berry, S.R., Rice, S.A., and Ross, J. (2000). Physical Chemistry, Second ed. (New York: Oxford University Press). Blanchoin, L., and Pollard, T.D. (1999). Mechanism of interaction of Acanthamoeba actophorin (ADF/cofilin) with actin filaments. J. Biol. Chem. 274, Blanchoin, L., Pollard, T.D., and Mullins, R.D. (2000). Interaction of ADF/cofilin, Arp2/3 complex, capping protein and profilin in remodeling of branched actin filament networks. Curr. Biol. 10, Carlier, M.F., Laurent, V., Santolini, J., Melki, R., Didry, D., Xia, G.X., Hong, Y., Chua, N.H., and Pantaloni, D. (1997). Actin depolymerizing factor (ADF/cofilin) enhances the rate of filament turnover: implication in actin-based motility. J. Cell Biol. 136, De La Cruz, E.M. (2005). Cofilin binding to muscle and non-muscle actin filaments: isoform-dependent cooperative interactions. J. Mol. Biol. 346,

11 Nonmass Action Mechanism of ADF/Cofilin Activity 23 Dedova, I.V., Nikolaeva, O.P., Mikhailova, V.V., dos Remedios, C.G., and Levitsky, D.I. (2004). Two opposite effects of cofilin on the thermal unfolding of F-actin: a differential scanning calorimetric study. Biophys. Chem. 110, Du, J., and Frieden, C. (1998). Kinetics studies on the effect of yeast cofilin on yeast actin polymerization. Biochemistry 37, Galkin, V.E., Orlova, A., Lukoyanova, N., Wriggers, W., and Egelman, E.H. (2001). Actin depolymerizing factor stabilizes an existing state of F-actin and can change the tilt of F-actin subunits. J. Cell Biol. 153, Ghosh, M., Song, X., Mouneimne, G., Sidani, M., Lawrence, D.S., and Condeelis, J.S. (2004). Cofilin promotes actin polymerization and defines the direction of cell motility. Science 304, Gittes, F., Mickey, B., Nettleton, J., and Howard, J. (1993). Flexural rigidity of microtubules and actin filaments measured from thermal fluctuations in shape. J. Cell Biol. 120, Ichetovkin, I., Grant, W., and Condeelis, J. (2002). Cofilin produces newly polymerized actin filaments that are preferred for dendritic nucleation by the Arp2/3 complex. Curr. Biol. 12, Kovar, D.R., and Pollard, T.D. (2004). Insertional assembly of actin filament barbed ends in association with formins produces piconewton forces. Proc. Natl. Acad. Sci. USA 101, Kowalczykowski, S.C., Paul, L.S., Lonberg, N., Newport, J.W., McSwiggen, J.A., and von Hippel, P.H. (1986). Cooperative and noncooperative binding of protein ligands to nucleic acid lattices: experimental approaches to the determination of thermodynamic parameters. Biochemistry 25, Kuhn, J.R., and Pollard, T.D. (2005). Real-time measurements of actin filament polymerization by total internal reflection fluorescence microscopy. Biophys. J. 88, Lappalainen, P., Fedorov, E.V., Fedorov, A.A., Almo, S.C., and Drubin, D.G. (1997). Essential functions and actin-binding surfaces of yeast cofilin revealed by systematic mutagenesis. EMBO J. 16, Loisel, T.P., Boujemaa, R., Pantaloni, D., and Carlier, M.F. (1999). Reconstitution of actin-based motility of Listeria and Shigella using pure proteins. Nature 401, Maciver, S.K., Zot, H.G., and Pollard, T.D. (1991). Characterization of actin filament severing by actophorin from Acanthamoeba castellanii. J. Cell Biol. 115, McGhee, J.D., and von Hippel, P.H. (1974). Theoretical aspects of DNA-protein interactions: co-operative and non-co-operative binding of large ligands to a one-dimensional homogeneous lattice. J. Mol. Biol. 86, McGough, A., Pope, B., Chiu, W., and Weeds, A. (1997). Cofilin changes the twist of F-actin: implications for actin filament dynamics and cellular function. J. Cell Biol. 138, Nakano, K., and Mabuchi, I. (2006). Actin-depolymerizing protein Adf1 is required for formation and maintenance of the contractile ring during cytokinesis in fission yeast. Mol. Biol. Cell 17, Nishida, E. (1985). Opposite effects of cofilin and profilin from porcine brain on rate of exchange of actin-bound adenosine 5 0 -triphosphate. Biochemistry 24, Nishida, E., Maekawa, S., and Sakai, H. (1984). Cofilin, a protein in porcine brain that binds to actin filaments and inhibits their interactions with myosin and tropomyosin. Biochemistry 23, Nishida, E., Iida, K., Yonezawa, N., Koyasu, S., Yahara, I., and Sakai, H. (1987). Cofilin is a component of intranuclear and cytoplasmic actin rods induced in cultured cells. Proc. Natl. Acad. Sci. USA 84, Orlova, A., Shvetsov, A., Galkin, V.E., Kudryashov, D.S., Rubenstein, P.A., Egelman, E.H., and Reisler, E. (2004). Actin-destabilizing factors disrupt filaments by means of a time reversal of polymerization. Proc. Natl. Acad. Sci. USA 101, Ott, A., Magnasco, M., Simon, A., and Libchaber, A. (1993). Measurement of the persistence length of polymerized actin using fluorescence microscopy. Phys. Rev. E Stat. Phys. Plasmas Fluids Relat. Interdiscip. Topics 48, R1642 R1645. Otterbein, L.R., Graceffa, P., and Dominguez, R. (2001). The crystal structure of uncomplexed actin in the ADP state. Science 293, Pantaloni, D., and Carlier, M.F. (1993). How profilin promotes actin filament assembly in the presence of thymosin b4. Cell 75, Pollard, T.D. (1986). Rate constants for the reactions of ATP- and ADP-actin with the ends of actin filaments. J. Cell Biol. 103, Pollard, T.D., and Borisy, G.G. (2003). Cellular motility driven by assembly and disassembly of actin filaments. Cell 112, Pope, B.J., Gonsior, S.M., Yeoh, S., McGough, A., and Weeds, A.G. (2000). Uncoupling actin filament fragmentation by cofilin from increased subunit turnover. J. Mol. Biol. 298, Prochniewicz, E., Janson, N., Thomas, D.D., and De la Cruz, E.M. (2005). Cofilin increases the torsional flexibility and dynamics of actin filaments. J. Mol. Biol. 353, Rebello, C.A., and Ludescher, R.D. (1998). Influence of tightly bound Mg2+ and Ca2+, nucleotides, and phalloidin on the microsecond torsional flexibility of F-actin. Biochemistry 37, Ressad, F., Didry, D., Egile, C., Pantaloni, D., and Carlier, M.F. (1999). Control of actin filament length and turnover by actin depolymerizing factor (ADF/cofilin) in the presence of capping proteins and ARP2/3 complex. J. Biol. Chem. 274, Rosenblatt, J., Agnew, B.J., Abe, H., Bamburg, J.R., and Mitchison, T.J. (1997). Xenopus actin depolymerizing factor/cofilin (XAC) is responsible for the turnover of actin filaments in Listeria monocytogenes tails. J. Cell Biol. 136, Sept, D., and McCammon, J.A. (2001). Thermodynamics and kinetics of actin filament nucleation. Biophys. J. 81, Yeoh, S., Pope, B., Mannherz, H.G., and Weeds, A. (2002). Determining the differences in actin binding by human ADF and cofilin. J. Mol. Biol. 315, Zebda, N., Bernard, O., Bailly, M., Welti, S., Lawrence, D.S., and Condeelis, J.S. (2000). Phosphorylation of ADF/cofilin abolishes EGF-induced actin nucleation at the leading edge and subsequent lamellipod extension. J. Cell Biol. 151, Accession Codes Atomic coordinates were deposited in the Protein Data Bank under accession code: 2I2Q.