The mouse formin, FRLalpha, slows actin filament barbed end elongation, competes with capping protein, accelerates polymerization from monomers, and severs filaments.

Formins are a conserved class of proteins expressed in all eukaryotes, with known roles in generating cellular actin-based structures. The mammalian formin, FRLalpha, is enriched in hematopoietic cells and tissues, but its biochemical properties have not been characterized. We show that a construct composed of the C-terminal half of FRLalpha (FRLalpha-C) is a dimer and has multiple effects on muscle actin, including tight binding to actin filament sides, partial inhibition of barbed end elongation, inhibition of barbed end binding by capping protein, acceleration of polymerization from monomers, and actin filament severing. These multiple activities can be explained by a model in which FRLalpha-C binds filament sides but prefers the topology of sides at the barbed end (end-sides) to those within the filament. This preference allows FRLalpha-C to nucleate new filaments by side stabilization of dimers, processively advance with the elongating barbed end, block interaction between C-terminal tentacles of capping protein and filament end-sides, and sever filaments by preventing subunit re-association as filaments bend. Another formin, mDia1, does not reduce the barbed end elongation rate but does block capping protein, further supporting an end-side binding model for formins. Profilin partially relieves barbed end elongation inhibition by FRLalpha-C. When non-muscle actin is used, FRLalpha-C's effects are largely similar. FRLalpha-C's ability to sever filaments is the first such activity reported for any formin. Because we find that mDia1-C does not sever efficiently, severing may not be a property of all formins.

Non-muscle cells contain a variety of actin filament-based structures, including lamellipodia, ruffles, filopodia, microvilli, and sarcomeric contractile structures (including cytokinetic actin rings and stress fibers). Assembly mechanisms for these structures are being vigorously investigated. Spontaneous nucleation of actin monomers occurs very slowly (1), and specific actin-associated proteins that promote rapid actin assembly are required for creating each actin-based structure. Arp2/3 1 complex is a well characterized nucleation factor, forming networks of branched actin filaments that are present in lamellipodia and ruffles (2). In contrast, the proteins controlling assembly of many other actin-based structures have not been identified.
Formins are a conserved class of actin-associated proteins that have been found in all eukaryotes examined and accelerate filament assembly independently of Arp2/3 complex (3). Two unifying structural features of formins are the Formin Homology 1 and 2 (FH1, FH2) domains, generally found in the C-terminal half of the protein. The FH1 domain contains proline-rich sequences capable of binding profilin and SH3 (Src homology 3) domain-containing proteins. The FH2 domain forms a multimeric structure (4,5).
Budding yeast formins Bni1p and Bnr1p are required for the assembly of actin cables and cytokinetic actin rings in vivo (6,7). Bni1p has barbed end nucleation activity in vitro, for which the FH2 domain is sufficient (7,8). Bni1p also slows barbed end elongation, while blocking complete barbed end elongation inhibition by capping protein (4,5,9). Thus, although Bni1p slows elongation, it allows filaments to elongate in the presence of capping protein. Because capping protein usually caps newly assembled filaments within seconds, formins may allow prolonged filament elongation in cells.
In fission yeast, Cdc15 is required for actin ring assembly in vivo. Cdc12 is a barbed end nucleator when bound to the actin monomer-binding protein profilin. In the absence of profilin, Cdc12 tightly caps barbed ends, allowing only pointed end elongation (10).

and this study).
Here we characterize the biochemical properties of the mammalian formin, FRL (formin-related gene in leukocytes), first identified from mouse spleen as the 1094-amino acid FRL␣ splice variant (15), although a number of other variants exist in the data base. We restrict our current study to the C-terminal region of FRL␣ (FRL␣-C, amino acids 449 -1094), which contains the complete FH1 and FH2 domains, as well as the full C terminus. In several assays, a similar construct of the FRL␤ splice variant, differing in its C-terminal 30 amino acids (15), behaves similarly.
FRL␣-C is a dimer and has multiple effects on muscle actin. FRL␣-C binds filaments tightly, with an apparent K d Ͻ 0.1 M. In addition, FRL␣-C slows barbed end elongation with an IC 50 of about 2 nM, demonstrating that it binds preferentially to filament barbed ends. This inhibition of elongation is only partial, and FRL␣-C protects the barbed ends from complete elongation inhibition by capping protein. In pyrene-actin polymerization assays, FRL␣-C accelerates actin polymerization in a concentration-dependent manner, with a persistent lag being observed even at micromolar FRL␣-C concentrations. FRL␣-C's polymerization activity is much weaker than that observed for mDia1 (14). FRL␣-C also severs actin filaments, creating new barbed ends capable of elongation. Additional experiments with platelet actin demonstrate that FRL␣-C has similar effects on non-muscle actin. We believe that polymerization acceleration by FRL␣-C is due both to weak nucleation and filament severing. In addition, we postulate that FRL␣-C's multiple effects on actin dynamics are due to its ability to interact with filament sides, with a preference for the side of the barbed end.

EXPERIMENTAL PROCEDURES
DNA Constructs-Our construct of FRL␣ (GenBank TM /EBI accession number AF215666) and FRL␤ (GenBank TM /EBI accession number AF006466) was generated by reverse transcriptase-PCR from 300.19 murine pre-B lymphoma cell RNA. Total RNA was isolated from exponentially growing cell cultures with TRIzol reagent (Invitrogen), and cDNA was produced with oligo(dT) primer and SuperScript II reverse transcriptase (Invitrogen). The coding region fragments 449 -1094 for FRL␣ (FRL␣-C) and 423-1064 for FRL␤ (FRL␤-C) were amplified with Pfu DNA polymerase (Stratagene). The PCR product was cloned into pGEX-KText vector (a gift from Jack Dixon).
Protein Preparation and Purification-For FRL␣-C and FRL␤-C, Rosetta DE3 Escherichia coli (Novagen) were transformed with expression construct and grown to A 600 0.6 -0.8 in TB (12 g/liter Tryptone, 24 g/liter yeast extract, 4.5 ml/liter glycerol, 14 g/liter dibasic potassium phosphate, and 2.6 g/liter monobasic potassium phosphate) with 100 g/ml ampicillin and 34 g/ml chloramphenicol at 37°C. After reduction to 16°C for 30 min, 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside was added, and the cultures were grown overnight. All subsequent purification steps were performed at 4°C or on ice. Pelleted bacteria were resuspended in EB (50 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5 mM EDTA, 1 mM DTT, and 1 pill/50 ml Complete protease inhibitors (Roche Applied Science)) and extracted by sonication. After ultracentrifugation, supernatant was loaded onto glutathione-Sepharose 4B (Amersham Biosciences) and washed with EB (without protease inhibitors but with 0.05% Polidocanol (Sigma P-9641)). Thrombin (Sigma T-4265) was added to a 50% slurry of the beads to 10 units/ml, and the suspension was mixed for 4 h. Cleaved protein was washed from the column, and thrombin was inactivated with 5 mM diisopropyl fluorophosphate/1 mM phenylmethylsulfonyl fluoride for 15 min, after which DTT was added to a concentration of 10 mM. FRL␣-C was further enriched by SourceS15 chromatography (Amersham Biosciences). Final protein pools were concentrated with Centricon P-20 (Amicon) and dialyzed in Na50MEPD (50 mM NaCl, 0.1 mM MgCl 2 , 0.1 mM EGTA, 2 mM NaPO 4 , pH 7.0, 0.5 mM DTT). Aliquots were stored at 4°C or at Ϫ70°C, with no resulting change in protein activity levels from either storage method. MDia1 748 -1255 was expressed as lovingly described previously (14). In contrast to FRL␣-C, mDia1 748 -1255 lost 90% of its nucleation activity upon freezing, so was stored at 4°C. For human profilin I, BL21(pLysS)DE3 E. coli (Novagen) were transformed with expression construct (gift from Thomas Pollard) and grown to A 600 0.6 -0.8 in LB (10 g/liter Tryptone, 5 g/liter yeast extract, 10 g/liter NaCl), with 100 g/ml ampicillin and 34 g/ml chloramphenicol at 37°C. 1 mM isopropyl-1-thio-␤-D-galactopyranoside was added, and cultures were incubated an additional 4 h at 37°C. All subsequent purification steps were performed at 4°C or on ice. Pelleted bacteria were resuspended in EB and extracted by sonication. After ultracentrifugation supernatant was filtered through a 0.45-m syringe filter and loaded onto poly-L-proline (Sigma P-3886) linked to CNBr-activated Sepharose (Amersham Biosciences). The column was washed with buffer 1 (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1 mM DTT), buffer 2 (buffer 1 with 3 M urea), and buffer 3 (buffer 1 with 8 M urea) sequentially. Buffer 3 eluate was dialyzed in DB (2 mM Tris-HCl, pH 8.0, 0.2 mM EGTA, 1 mM DTT, 0.01% NaN 3 ) overnight and then for an additional 2 h in fresh DB. Profilin was stored at 4°C. Capping protein was a kind gift from David Kovar and Thomas Pollard. ␣-Actinin (AT01-A) and full-length muscle myosin (MY02-A) were purchased from Cytoskeleton. Rabbit skeletal muscle actin was purified from acetone powder (16) and labeled with pyrenyliodoacetamide (17). Both unlabeled and labeled actin were gel filtered on S200 (18), which was crucial to obtain reproducible polymerization kinetics. Platelet actin was purchased from Cytoskeleton (APHL95). Before use, platelet actin was resuspended with water then centrifuged at 100,000 rpm for 30 min at 4°C in a TLA-120 rotor (Beckman). Supernatant was loaded onto a SourceQ 5/5 column (Amersham Biosciences) equilibrated with G0 buffer (2 mM Tris-HCl, pH 8.0, 0.5 mM DTT, 0.1 mM CaCl 2 ). Actin was eluted with a linear gradient from G0 to G300 (G0 plus 300 mM NaCl) and eluted as a single peak at 250 mM NaCl. ATP was added immediately to a final concentration of 0.2 mM. Peak fractions were pooled, and actin was polymerized by addition of EGTA, MgCl 2 , and imidazole (pH 7.0) to 1, 1, and 10 mM, respectively, and incubated at room temperature for 4 h, then overnight at 4°C. Polymerized actin was centrifuged at 100,000 rpm for 30 min at 4°C in a TLA-120 rotor. Pellets were resuspended in G buffer (2 mM Tris-HCl, pH 8.0, 0.5 mM DTT, 0.2 mM ATP, 0.1 mM CaCl 2 , and 0.01% NaN 3 ), pushed through a 30-gauge needle, and dialyzed in G buffer for 48 h. After dialysis actin was centrifuged at 100,000 rpm for 3 h at 4°C in a TLA-120 rotor. The upper two-thirds of supernatant was removed and used for subsequent assays. An alternative purification process, in which actin was polymerized, depolymerized, then gel-filtered over a Superdex200 10/30 column (Amersham Biosciences) in G buffer, was also performed. Both purification procedures produced actin monomers with similar polymerization properties and removed gelsolin and Arp2/3 complex effectively.
Protein Size Analysis Techniques-Gel filtration chromatography was conducted using a Superdex200 10/30 column (Amersham Biosciences) calibrated with both high and low molecular weight standards (Amersham Biosciences). A Stokes radius was calculated following the manufacturer's instructions. Analytical ultracentrifugation was conducted using a Beckman Proteomelab XL-A and an AN-60 rotor. For sedimentation velocity analytical ultracentrifugation, 0.7 M FRL␣-C in 100 mM NaCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM NaPO 4 (pH 7.0), 0.5 mM DTT was centrifuged at 30,000 rpm and 20°C, and A 220 was monitored every 2 min by continuous scan at 0.003-cm steps. Protein partial specific volume, buffer density, and buffer viscosity were determined using the Sednterp program (David Hayes and Tom Laue). Scans 1-200 were analyzed using Sedfit87 (www.analyticalultracentrifugation.com), resulting in one major species (Ͼ90%) centered at 4.2 S with a frictional ratio of 2.02. For sedimentation equilibrium analytical ultracentrifugation, various concentrations of FRL␣-C (0.25, 0.333, 0.5, 0.667, 0.75, 1, and 1.25 M) in the same buffer as for velocity centrifugation was centrifuged at 7,000, 10,000, and 14,000 rpm for 15, 10, and 10 h, respectively, at 20°C. A 220 at 0.001-cm steps were recorded every hour. Winmatch software (David Yphantis) was used to confirm equilibrium, and Winreedit software (D. Yphantis) was used to trim the data. Winnonln software (D. Yphantis) was used to fit the data. First, individual concentrations at individual speeds were analyzed, and minor speed-dependent and concentration-dependent changes in apparent molecular mass were detected. Next, data at multiple concentrations and speeds were fit to a single species, resulting in an apparent molecular mass of 110 kDa (root mean square deviation ϭ 0.00481). The systematic residual error pattern of this fit suggested non-ideality (19). Finally, the same multiple concentrations and speeds were fit to a monomer-dimer equilibrium, with a calculated monomer molecular mass of 71,335 Da. The resulting fit (root mean square deviation ϭ 0.00441) had little systematic error and suggested an apparent K d for dimerization of 0.1 M.
Actin Filament Binding Assays-Actin (10 M) was polymerized for 2 h at 23°C in polymerization buffer (G-Mg buffer (2 mM Tris-HCl, pH 8.0, 0.5 mM DTT, 0.2 mM ATP, 0.1 mM MgCl 2 , and 0.01% NaN 3 ) plus 1ϫ KMEI (10 mM imidazole, pH 7.0, 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA)), followed by addition of 10 M phalloidin (Sigma P-2141). This actin stock was diluted to desired concentration in polymerization buffer in the absence or presence of putative binding proteins. Mixing was conducted in polycarbonate 7-ϫ 20-mm centrifuge tubes (Beckman 343775) to a final volume of 200 l. Filaments were pipetted using cut Pipetteman tips to minimize shearing. After 5 min at 23°C, samples were centrifuged at 80,000 rpm for 20 min at 20°C in a TLA-100.1 rotor (Beckman). 150 l of supernatant was removed, lyophilized, and resuspended in 15 l of SDS-PAGE sample buffer. After removal of the remaining supernatant, pellets were washed briefly with 200 l of 1ϫ KMEI, then resuspended in 20 l of SDS-PAGE sample buffer. Supernatants and pellets were analyzed by Coomassie-stained SDS-PAGE. Similar assays omitting phalloidin or adding 5 mM NaPO 4 , pH 7.0, were also conducted.
Actin Polymerization by Fluorescence Spectroscopy-A detailed procedure is described in a previous study (20). Unlabeled and pyrenelabeled actin were mixed in G buffer to produce an actin stock of the desired pyrene-labeled actin percentage (5% unless otherwise stated). This stock was converted to Mg 2ϩ salt by 2-min incubation at 23°C in 1 mM EGTA/0.1 mM MgCl 2 immediately prior to polymerization. Polymerization was induced by addition of 10ϫ KMEI (500 mM KCl, 10 mM MgCl 2 , 10 mM EGTA, and 100 mM imidazole, pH 7.0) to a concentration of 1ϫ, with the remaining volume made up by G-Mg. Added proteins were mixed together for 1 min prior to their rapid addition to actin to start the assay. Pyrene fluorescence (excitation 365 nm, emission 407 nm) was monitored in a PC1 spectrofluorometer (ISS, Champaign, IL). The time between mixing of final components and start of fluorometer data collection was measured for each assay and ranged between 10 and 15 s.
Calculating Filament Concentration-Slopes of pyrene fluorescence from polymerization time courses were determined at the 50% point of polymerization with Kaleidagraph (Synergy Software, Reading, PA). Slopes were converted initially to filament concentration under the assumption of unrestricted ATP-actin monomer addition to barbed ends with a rate constant (K ϩ ) of 11.6 M Ϫ1 s Ϫ1 (1) according to the equation, where F is filament concentration in micromolar, SЈ is slope converted to micromolar/s, and M 0.5 is micromolar monomer concentration at 50% polymerization. SЈ is calculated by the equation SЈ ϭ (S ϫ M t )/(f max Ϫ f min ), where S is raw slope in arbitrary units/s, M t is micromolar concentration of total polymerizable monomer in micromolar, and f max and f min are fluorescence of fully polymerized and unpolymerized actin respectively, in arbitrary units. Because FRL␣-C slows elongation rate by 80%, filament concentrations from assays conducted in the presence of FRL␣-C were multiplied by 5.
Barbed End Elongation Assays-Unlabeled actin (10 M) was polymerized for at least 1 h at 23°C, then diluted to 5 M in the presence of 10 M phalloidin and centrifuged at 100,000 rpm for 20 min in a TLA-120 rotor. The pellet was resuspended in 3ϫ polymerization buffer to 5 M, then sheared by two passes through a 27-gauge needle. 37.5 l of this mixture was aliquoted into Eppendorf tubes and allowed to re-anneal overnight at 23°C. Polymerization buffer containing FRL␣-C, capping protein, or mixtures of both (37.5-l total) were added to filaments, mixed by gentle flicking, and incubated at 23°C for 3 min. In competition assays between FRL␣-C and capping protein, these two proteins were pre-mixed then added simultaneously to filaments. After 3 min at 23°C, 75 l of 2 M monomers (5% pyrene, Mg 2ϩ -converted) and 8 M profilin in G-Mg buffer were added to the filaments with a cut p200 tip, mixed by pipetting up and down three times, and placed into the fluorometer cuvette. Fluorescence (365/407 nm) was recorded for 180 s. Initial elongation velocity was obtained by linear fitting the initial 100 s of elongation. Final concentrations in the assay were 1.25 M phalloidin-stabilized polymerized actin (0.1-0.15 nM barbed ends) and 1 M monomer. In competition experiments between capping protein and formins, apparent K d of formin binding to barbed end was determined by fitting elongation data to a competition binding model as described previously (20), assuming a K d of 0.1 nM for capping protein binding to barbed ends.
Re-annealing Assay-Actin (10 M) was polymerized for 2 h at 23°C in polymerization buffer, then diluted to 1 M in polymerization buffer with 1.5 M rhodamine-phalloidin (Sigma P-1951). This sample was pulled/pushed four times through a 1-ml syringe with a 27-gauge needle attached, then 10 l was immediately aliquoted into tubes containing 10 l of polymerization buffer with or without 200 nM FRL␣-C or 200 nM capping protein. To stop the reactions, 1 ml of fluorescence buffer (25 mM imidazole, pH 7.0, 25 mM KCl, 4 mM MgCl 2 , 1 mM EGTA, 100 mM DTT, 0.5% methylcellulose, 3 mg/ml glucose, 18 g/ml catalase, 100 g/ml glucose oxidase) was added at various time points. Samples (2 l) were adsorbed to 12-mm round glass coverslips previously coated with 0.01% poly-L-lysine. Filaments were pipetted with cut Pipetteman tips to reduce shearing. Samples were visualized on a Zeiss Axioplan 2 microscope (Carl Zeiss, Thornwood, NY) using 100ϫ 1.4 numerical aperture objective, and images were acquired with a Hamamatsu Orca II cooled charge-coupled device camera using Openlab software (Improvisions Inc., Boston, MA). Filament length was measured using Openlab software. At least two images were analyzed for every condition and time point, and all filaments with both ends discernable were measured on each image, resulting in 200 -1400 individual filaments being meas-ured for each condition depending on filament length.
Severing Assay Using Fluorescence Microscopy-Actin (4 M) was polymerized for at least 1 h at 23°C in polymerization buffer. 10-l aliquots from this stock were carefully pipetted into Eppendorf tubes and incubated 10 min at 23°C. 10 l of protein or buffer was added, mixed by gentle flicking, and incubated for the desired time at 23°C. Polymerization buffer (20 l) containing rhodamine-phalloidin (1 M final) was added, and samples were immediately diluted 200-fold with fluorescence buffer. Samples (2 l) were adsorbed to 12-mm round glass coverslips previously coated with 0.01% poly-L-lysine. Great care was taken during sample preparation to minimize manual severing. Filaments were only pipetted twice during the procedure: once after initial polymerization but before addition of FRL␣-C or buffer and once to apply to coverslips. Also, we used cut Pipetteman tips to reduce shearing. Samples were visualized, and images were acquired as described above. Filament length was measured using Openlab software. At least five images were analyzed for every condition, and all filaments with both ends discernable were measured for each image, resulting in 200 -1500 individual filaments being measured depending on filament length.
Dual Color Filament Assay Using Fluorescence Microscopy-Actin (4 M) was polymerized for at least 1 h at 23°C in polymerization buffer. 10 l of polymerized actin was incubated with 10 l of 200 nM FRL␣-C or buffer for 10 min at 23°C then 20 l of rhodamine-phalloidin was added to 1 M. Samples were diluted 10-fold with 0.5 M actin monomers and 0.5 M Alexa 488-phalloidin (Molecular Probes A-12379), incubated for 6 min at 23°C, and diluted with 2 ml of fluorescence buffer. Samples (2 l) were adsorbed to 12-mm round glass coverslips previously coated with 0.01% poly-L-lysine. Filaments were handled as described for the severing assay. Images were taken through both rhodamine and fluorescein filters. Filament length was measured using Openlab software. For each individual filament with both ends discernable, rhodamine-and Alexa 488-labeled segments were measured separately. At least two images were analyzed for each condition, totaling ϳ150 filaments.
Kinetic Severing Assay Using Time-lapse Microscopy-This method was adapted from previous studies (10,21). Actin (4 M) was polymerized for 2 h at 23°C in polymerization buffer, then rhodamine-phalloidin was added to label 50% of polymerized actin. Filaments were diluted to 0.067 M in modified fluorescence buffer (25 mM imidazole, pH 7.0, 25 mM KCl, 4 mM MgCl 2 , 1 mM EGTA, 100 mM DTT, 0.5% methylcellulose, 6 mg/ml glucose, 36 g/ml catalase, 200 g/ml glucose oxidase) plus 5 mg/ml BSA. Coverslips (30 ϫ 22 mm) were mounted onto glass slides with two pieces of double stick tape, forming a perfusion chamber. NEM-treated myosin was diluted to 50 nM in high salt buffer (50 mM Tris-HCl, pH 7.5, 600 mM NaCl), and 70 l was perfused into chambers for 1 min. Chambers were washed once with 70 l of high salt buffer (50 mM Tris-HCl, pH 7.5, 600 mM NaCl, 1% BSA) and once with 70 l of low salt buffer (same except with 150 mM NaCl). 70 l of labeled actin filaments was perfused into chambers using a cut Pipetteman tip and incubated 5 min. Chambers were washed once with fluorescence buffer plus 5 mg/ml BSA to remove any unbound filaments. Time-lapse recording was initiated, and, after the first image was acquired, 70 l of FRL␣-C (diluted to 200 nM in modified fluorescence buffer with 5 mg/ml BSA) was perfused into chamber. Only filaments clearly breaking in the middle, resulting in two discernable pieces that could be tracked, were counted as severing events. Thirteen sequences for both control and FRL␣-C were analyzed, and the sequences containing the highest and lowest number of severing events for each condition were excluded.
Western Blotting of Actin Preparations-Gel-filtered actin from rabbit muscle, or platelet actin after initial resuspension or additional purification, was Western blotted at 2 g on polyvinylidene difluoride (Millipore) against antibodies to human gelsolin (monoclonal GS-2C4 from Sigma), the ARPC1b subunit of Arp2/3 complex, human profilin I, human cofilin (gifts from Thomas Pollard), human WASp (22), and capping protein ␣2 subunit (monoclonal antibody 5B12.3 was developed by John Cooper and obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD, National Institutes of Health and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA). Blots were developed by chemiluminescence and stained for protein by Amido Black.

FRL␣-C Is a Dimer-
We expressed a C-terminal construct of FRL␣ (FRL␣-C, amino acids 449 -1094) as a glutathione Stransferase (GST) fusion protein in bacteria (Fig. 1, A and B). After cleavage from GST, FRL␣-C elutes with an apparent Stokes radius of 71 Å by gel filtration, consistent with a spherical particle of 530 kDa (Fig. 1C). Sedimentation velocity analytical ultracentrifugation reveals a single species of 4.2 S and a frictional ratio of 2.02 (Fig. 1D). Equilibrium analytical ultracentrifugation of eight FRL␣-C concentrations at three speeds results in curves that best fit a monomer-dimer equilibrium with an apparent K d of 0.1 M (Fig. 1E). These data suggest that FRL␣ -C might dimerize in a reversible fashion. The high frictional ratio suggests that this dimer is elongated, not spherical.

FRL␣-C Binds Muscle Actin Filament Sides and Barbed
Ends-We first characterized FRL␣-C's effects on muscle actin then performed additional targeted experiments on non-muscle actin. FRL␣-C binds tightly to pre-formed actin filaments, because 0.4 M phalloidin-stabilized polymerized actin quantitatively pellets 0.2 M FRL␣-C ( Fig. 2A). A comparable construct of mDia1 binds much more weakly, with an apparent K d of 3 M (Fig. 2B) (14). As with mDia1 (14), filament binding by FRL␣-C is unaffected by the presence of phalloidin or of filament-bound phosphate (not shown). FRL␤-C binds filaments with similar affinity to FRL␣-C (Fig. 2B).
We tested FRL␣-C's effect on barbed end elongation from phalloidin-stabilized actin filaments. FRL␣-C slows elongation 5-fold, as indicated by lower slopes of pyrene-actin fluorescence compared with actin alone (Fig. 3, inset). The half-maximal concentration for this effect is 2 nM (Fig. 3).
We hypothesized that, similar to yeast formins (4, 5, 10), FRL␣-C may partially block filament barbed ends resulting in the reduced elongation rate. To examine this possibility further we tested filament annealing in the presence of FRL␣-C, because annealing requires free barbed ends (23). Phalloidinstabilized filaments were sheared through a 27-gauge needle, then allowed to re-anneal in the absence or presence of FRL␣-C or capping protein and observed by fluorescence microscopy. FRL␣-C slows filament re-annealing (Fig. 4, B, E, and H), which becomes evident at early time points (Fig. 4J). This inhibition, however, is not complete. For comparison, capping protein allows no significant re-annealing over a 24-h period (Fig. 4, C, F, and I).
FRL␣-C Accelerates Polymerization from Muscle Actin Monomers-We tested FRL␣-C's effect on actin polymerization kinetics by pyrene-actin polymerization assays. FRL␣-C accelerates polymerization of 4 M actin monomers (Fig. 5A), but with a considerable lag before new filament production even at M FRL␣-C concentration. We calculated filament concentration using slopes at 50% polymerization and assuming an elongation rate of 3.62 m Ϫ1 s Ϫ1 , based on rates of 11.16 and 1.3 m Ϫ1 s Ϫ1 for barbed and pointed ends, respectively (1), and from the 80% inhibition of barbed end elongation by FRL␣-C. FRL␣-C enables formation of 10 nM filaments maximally, with a ratio of about 1:5 of filaments produced to FRL␣-C dimer in the linear range (Fig. 5B). FRL␣-C decreases the polymerization lag 5-fold, where lag is defined as the time to reach 10% polymerization (Fig. 5C). FRL␤-C produces a similar effect (data not shown).
FRL␣-C's ability to accelerate polymerization is dependent on the concentration of actin monomers (Fig. 5D). Decreasing monomer concentration over a range from 4 M to 1.5 M causes an exponential decrease in the concentration of filaments assembled (Fig. 5E), whereas the time required to reach 10% polymerization increases linearly (Fig. 5F).
Profilin Modulates the Activities of FRL␣-C on Muscle Actin-Profilin is a highly abundant actin monomer-binding protein that inhibits actin nucleation and prevents monomer addition to filament pointed ends (24). Profilin binds polyproline sequences in the FH1 domain of formins, and mDia1 containing the FH1 domain can use profilin-bound actin monomers for nucleation (14). Consequently we tested the effect of profilin on elongation from actin filaments and polymerization from monomers in the presence of FRL␣-C.
A profilin concentration that binds Ͼ95% of the monomer reduces FRL␣-C's inhibition of elongation from 5-to 2-fold (Fig.  6A). A similar effect occurs with FRL␤-C (Fig. 6B). Profilin has no measurable effect on capping protein-induced elongation inhibition under these conditions (Fig. 6B).

FRL␣-C Competes with Capping Protein for the Barbed End of Muscle Actin
Filaments-Capping protein tightly binds filament barbed ends preventing monomer addition, whereas FRL␣-C inhibits elongation 50% in the presence of profilin. We asked whether FRL␣-C could protect barbed ends against capping protein, similar to Bni1 and mDia1 (4,5). When FRL␣-C and capping protein are added simultaneously to phalloidinstabilized filaments, FRL␣-C increases elongation rate in a concentration-dependent manner and with an apparent K d for barbed ends of 3.2 nM (Fig. 7A). Thus, FRL␣-C and capping protein compete for barbed end binding.
MDia1 Does Not Slow Elongation but Competes with Capping Protein-In contrast to other formins, mDia1 has no effect on barbed end elongation (Fig. 7B). This experiment was conducted with an mDia1 construct lacking the FH1 domain and in the presence of high profilin concentration to block the potent nucleation activity of mDia1 (14). Surprisingly, mDial blocks barbed end capping by capping protein, with similar potency to FRL␣-C (2.6 nM apparent K d for barbed ends, Fig. 7B).
FRL␣-C Severs Muscle Actin Filaments-Using a fluorescence microscopy assay, we found that FRL␣-C severs actin filaments. When FRL␣-C was incubated with actin filaments in suspension followed by stabilization with rhodamine-phalloidin, a marked decrease in filament length was observed (Fig. 8,  A, B, and G). When rhodamine-phalloidin was added to filaments prior to FRL␣-C incubation, severing was inhibited (Fig.  8, D and E). We were concerned that this effect was artifactual, due to the barbed end binding ability of FRL␣-C, and because barbed end capping proteins slow re-annealing of sheared filaments (Fig. 4, C, F, and I) (22). Therefore we took great care during sample preparation to minimize filament shearing during manipulation (see "Experimental Procedures"). As a confirmation of our methods, filament length was not significantly affected by the presence of 100 nM capping protein (Fig. 8C) or ␣-actinin (Fig. 8F). Measurements of filament length were highly reproducible. Because incubation of filaments with FRL␣-C occurs in solution, the possibility that surface-bound filaments might be artifactually destabilized by an effect of FRL␣-C side binding on filament helical pitch is not an issue.
Severing depends on FRL␣-C concentration and incubation time (Fig. 8, H and I)  incubated with filaments for 10 min (Fig. 8H), which resulted in a median filament length of 2.2 m, with 46.8% of filaments shorter than 2 m. Incubation of buffer with filaments for 10 min resulted in a median filament length of 6.7 m, and only 4.61% of filaments under 2 m. The largest filaments were severely depleted by FRL␣-C, with filaments Ͼ10 m going from 48% to 14%.
To monitor barbed end elongation from severed filaments, we performed a dual filament assay similar to those previously described (25). Polymerized actin was incubated with buffer ( Fig. 9A) or 200 nM FRL␣-C (Fig. 9B), then equimolar rhodam-ine-phalloidin was added. Subsequently, additional monomers and Alexa 488-phalloidin were added, and the new monomers were allowed to elongate. Thus, the original filaments were labeled with rhodamine-phalloidin, whereas the newly elongated segments of these filaments were labeled with Alexa 488-phalloidin. A concentration of 0.5 M actin monomers was added to minimize pointed end growth. Rhodamine-and Alexa 488-labeled segments of each filament were measured separately. Filaments incubated with buffer had median lengths of 8.16 and 7.24 m for rhodamine-labeled and Alexa 488-labeled ends, respectively, whereas the numbers for filaments incu- bated with FRL␣-C were 3.87 and 3.92 m. Thus, FRL␣-C severs pre-formed filaments then slows barbed end elongation.
To observe severing activity of FRL␣-C more directly, we employed time-lapse fluorescence microscopy. Rhodamine-labeled actin filaments (1:2 molar ratio of phalloidin to actin) were tethered to coverslips by NEM-treated myosin. After perfusion of FRL␣-C into chambers, severing could be observed as the appearance of gaps or large breaks in the filaments over several frames (Fig. 10, arrows). Severing was scored only when a clear gap or break in the middle of a filament occurred during observation, and when both resulting pieces could be observed for several subsequent frames. Although overall shortening of filaments was observed with FRL␣-C addition, these events were not scored, because they may have been due to photobleaching or movement of filaments out of the plane of focus. A total of 50 severing events was observed with FRL␣-C, whereas 13 events were observed for buffer alone.
Severing by FRL␣-C Increases the Concentration of Muscle Actin Filament Ends-If FRL␣-C severs filaments, then the increase in filament concentration should be reflected in an increased elongation rate upon addition of monomers. When FRL␣-C was incubated with pre-formed filaments, followed by dilution with 0.5 M pyrene-actin monomers, we observed a slight decrease in elongation rate compared with filaments alone (Fig. 11). However, this decrease actually represented a 4-fold increase in filament concentration, because FRL␣-C decreased barbed end elongation 5-fold.
Effects of FRL␣-C on Platelet Actin-To examine FRL␣-C's effects on non-muscle actin, we purchased platelet actin (ϳ85% ␤-actin and 15% ␥-actin non-muscle isoforms) from Cytoskeleton (Denver, CO) then performed additional purification procedures (see "Experimental Procedures"). This additionally purified platelet actin was used for all subsequent experiments.
FRL␣-C slows barbed end elongation from platelet actin filaments 2-fold (Fig. 12A, closed circles) compared with the 5-fold reduction we observe with muscle actin (Fig. 3). The apparent K d for this effect is similar to the apparent K d with muscle actin. Profilin did not reduce FRL␣-C inhibition of elongation from platelet actin (data not shown), in contrast to its effect on muscle actin. FRL␣-C relieved elongation inhibition by capping protein on platelet actin (Fig. 12A, closed circles), with a similar apparent K d to that observed on muscle actin.
We also tested FRL␣-C's ability to accelerate polymerization from platelet actin monomers. Whereas 4 M platelet actin monomers polymerized more rapidly than 4 M muscle actin monomers (compare Fig. 5A to Fig. 12B), FRL␣-C's effects were similar (Fig. 12B), and the same trends in concentration of filaments produced and reduction in lag time were observed (data not shown).
Finally, we tested FRL␣-C's ability to sever platelet actin filaments. Incubation of 500 nM FRL␣-C with platelet actin filaments resulted in a reduction of median filament length from 9.77 to 2.34 m (Fig. 12, C and D). Thus, FRL␣-C produces qualitatively similar results on muscle and non-muscle actin, with the largest difference being on elongation rate. DISCUSSION In this study, we found FRL␣-C to be a dimer that has multiple effects on muscle actin. FRL␣-C bound tightly to actin filament sides. In addition, FRL␣-C bound and partially occluded actin filament barbed ends as evidenced by its 5-fold inhibition of barbed end elongation, its inhibition of filament re-annealing, and its inhibition of complete barbed end capping by heterodimeric capping protein. Although not slowing barbed end elongation, mDia1 strongly inhibited capping protein, evidence that further supports barbed end binding by formins. Profilin partially relieved FRL␣-C's inhibition of elongation from actin filaments. FRL␣-C also accelerated polymerization from actin monomers, an effect that might be partially due to its ability to sever filaments. In addition, we find FRL␣-C had similar effects on non-muscle actin: it inhibited barbed end elongation 2-fold, inhibited complete barbed end capping by capping protein, accelerated polymerization from monomers, and severed. This study is the first to demonstrate severing by any formin.
FRL␣-C is an elongated dimer, in agreement with biochemical data for a Bni1p FH2 construct (amino acids 1348 -1750) (5). A slightly longer Bni1p FH2 construct appears tetrameric (4). Possibly, the core FH2 region forms a dimer, whereas more C-terminal sequences induce higher order oligomerization for some formins. Because our construct contains the entire C terminus, this region of FRL␣ does not appear to affect higher order oligomerization with high affinity. Our equilibrium analytical ultracentrifugation results suggest that FRL␣-C dimerization may be reversible. Because reversible dissociation of a dimer is difficult to detect definitively, the details of dimerization equilibrium await further experimentation. If FRL␣-C is in monomer-dimer equilibrium, binding to the filament side or barbed end may stabilize the dimeric state.
The ability of formins to inhibit barbed end dynamics has previously been demonstrated, and a gradient of inhibition has emerged. Cdc12 completely inhibits barbed end elongation (10), FRL␣-C inhibits elongation from muscle actin filaments 80% (this study), Bni1p inhibits 25-50% (4), and mDia1 causes no inhibition (this study). To allow elongation at a reduced rate, formins may either change the twist of the filament helix by side binding or bind near the barbed end in a manner that inhibits access to monomers. We prefer the latter explanation, because low concentrations of formin are sufficient for elongation inhibition. In addition, three formins, mDia1 (14), Bni1 (4), and FRL␣ (data not shown) inhibit barbed end depolymerization To slow barbed end elongation, formins must move with the advancing barbed end. Models for processive capping by Bni1p have been proposed by Zigmond et al. (4) and Mosely et al. (5), which predict that Bni1p moves along with the barbed end as it elongates. This effect is predicted to be dependent on multimerization, because at least one formin subunit must remain bound while the other moves to a new barbed end actin subunit. Processive capping seems even more likely for FRL␣-C, because it binds so tightly to actin filament sides. Our observation that low FRL␣-C concentrations inhibit elongation (IC 50 of 2 nM) suggests that FRL␣-C must bind preferentially to the barbed end. If FRL␣-C bound with equal affinity to the side and barbed end, then higher concentrations than those observed would be required to inhibit elongation.
We propose a "side ratchet model" for FRL␣-C binding to elongating barbed ends (Fig. 13). FRL␣-C binds to both actin filament sides and barbed ends, but barbed end binding is preferred. FRL␣-C does not sit on the end but binds the side of the two barbed end subunits ("end-side" binding). End-side binding partially occludes access by monomers, slowing elongation (Step 1a). Different formins occlude the barbed end to different degrees, resulting in the gradient of elongation inhi- bition observed (Cdc12 Ͼ FRL␣ Ͼ Bni1 Ͼ mDia1). When a monomer does add, the interaction between one FRL␣-C subunit and the actin subunit (now the penultimate subunit) weakens, allowing the FRL␣-C subunit to release and re-bind the new barbed end subunit. The other FRL␣-C subunit remains bound to its actin subunit, thus maintaining association with the actin filament (Step 2a). The other formin subunit then advances analogously (Step 3). Our model extends those of Mosley et al. (5) and Zigmond et al. (4) in that we propose formin binding to the sides of barbed end subunits, instead of directly on the barbed end itself.
The reason for formin binding preference at the "end-side" over the side is unclear. We agree with Zigmond et al. (4) that binding is probably not affected by actin subunit nucleotide state, because both FRL␣-C (data not shown) and mDia1 (14) slow barbed end depolymerization from ADP-actin filaments. We postulate that the end-side preference arises from topology differences between end-sides and sides.
Profilin-bound monomers alter FRL␣-C's effect on elongation from muscle actin filaments, reducing elongation inhibition by Ͼ50%. This effect suggests that profilin binds to FRL␣-C, because addition of profilin-actin to barbed ends would be more inhibited by FRL␣-C then would free actin if profilin did not associate. FRL has been shown to associate with profilin in vitro and in vivo, and this interaction occurs specifically through its FH1 domain (15). Profilin probably increases the elongation rate by providing a closely apposed monomer that better overcomes partial barbed end occlusion by FRL␣-C.
FRL␣-C protects filament barbed ends from complete elongation inhibition by capping protein. This inhibition of capping protein has a similar concentration dependence as elongation inhibition, supporting our model that FRL␣-C remains associated at the barbed end as the filament is elongating, and is not binding and releasing with each new monomer addition. One would expect FRL␣-C to be less effective at blocking capping protein than at inhibiting elongation if FRL␣-C were continuously dissociating and re-associating. The x-ray crystal structure of capping protein (26) has lead to development of a model in which two C-terminal "tentacles" each contact two or three subunits at the barbed end in an "end-side" manner (27). This information supports our side ratchet model, because end-side binding by formin would inhibit end-side tentacle binding. The fact that mDia1 does not inhibit elongation but inhibits capping protein at concentrations similar to those of FRL␣-C (Ͻ5 nM) is remarkable and supports the end-side binding model.
As proposed by Zigmond et al. (4) and by Mosley et al. (5), the ability to protect against capping protein provides a mechanism for prolonging the elongation phase of filaments. Although in vitro FRL␣-C slows down elongation, in vivo filaments protected against capping protein would elongate efficiently, while unprotected filaments would get capped immediately (within 1 s) by capping protein. This possibly provides a mechanism for making longer filaments found in filopodia and microvilli as opposed to the short branched filaments in lamellipodia/ruffles (28).
FRL␣-C severs filaments, a property not previously demonstrated for formins. This property may be unique to FRL␣-C, because preliminary data suggest mDia1 does not sever. Our hypothesis is that the severing mechanism is related to the side binding and barbed end binding properties. In the absence of free barbed ends, FRL␣-C binds to the filament side (Step 1b). The filament then flexes due to its normal thermal motion, creating space for FRL␣-C to bind as it does at the filament barbed end (Step 2b). When the filament flexes back, FRL␣-C occludes part of the barbed end, preventing re-association and causing severing. After FRL␣-C severs, it remains associated with the new barbed end as it elongates (Step 2a, bottom). In contrast to FRL␣-C, mDia1-C does not sever efficiently, which may be due to two properties that are dissimilar to FRL␣-C: 1) weak side binding and 2) lack of barbed end occlusion.
Severing is probably a major contributor to FRL␣-C's acceleration of polymerization from monomers we observe in vitro for the following reasons. The lag before polymerization, even with high concentrations of FRL␣-C, suggests that FRL␣-C must wait for filaments to nucleate spontaneously before amplifying filament production. FRL␣-C's effect on polymerization is highly dependent on free actin concentration. Filaments produced by FRL␣-C decrease exponentially with either decreasing monomer concentration or increasing concentration of profilin. Severing may not be the only contributor to polymerization acceleration in vitro, because FRL␣-C might also enhance nucleation similar to other formins. The partial inhibition of re-annealing of FRL␣-C may also accelerate polymerization in vitro by keeping the filament number high.
We postulate that nucleation is also due to end-side binding, which causes stable dimerization (Fig. 13). Other nucleators are thought to operate by stabilizing actin dimers, but do so by binding the dimer end and not the side. Arp2/3 complex may form a dimer mimic, allowing elongation toward the barbed end (2). Capping protein binds dimers in a barbed end manner, allowing elongation toward the pointed end (29). Formins might be a third variety of nucleators, stabilizing dimers from the side and potentially allowing elongation in both directions.
Although the overall effects of FRL␣-C on non-muscle actin are similar to those for muscle actin, there are some important differences. FRL␣-C inhibits elongation from platelet actin filaments less than from muscle actin filaments. We observed a 2-fold inhibition, compared with the 5-fold inhibition seen on muscle filaments. Also in contrast to muscle filaments, profilin does not cause any relief to this inhibition. One possible reason for these differences could be due to slight differences in the structure of the barbed ends of non-muscle filaments, which may decrease occlusion of the barbed end by FRL␣-C. A second possible reason for these observed differences could be the presence of contaminating proteins in one of the actin preparations. As reported by the manufacturer, we detected gelsolin in the platelet actin after preparation according to their instructions. In addition, some Arp2/3 complex appeared present in this preparation. In contrast we did not detect profilin, capping protein, WASp, or cofilin. Additional purification steps appeared to remove both gelsolin and Arp2/3 complex from the platelet actin preparation (data not shown). How-ever, additional contaminants may persist and affect the actions of FRL␣-C.
It is unclear which biochemical properties of FRL␣-C are relevant in cells. The severing and polymerization acceleration activities are weak, raising doubts that they alone could generate filaments rapidly enough to support formation of large actin-based cellular structures. However, other binding proteins may enhance severing and/or nucleation by FRL␣-C. Capping protein inhibition is a potent effect and certainly is a possibility for creating long filaments in cells.

Filaments with Capping Protein, Accelerates Polymerization from Monomers, and Severs
, Slows Actin Filament Barbed End Elongation, Competes α The Mouse Formin, FRL