Integrin-bound talin head inhibits actin filament barbed-end elongation

Focal adhesions (FAs) mechanically couple the extracellular matrix to the dynamic actin cytoskeleton, via transmembrane integrins and actin-binding proteins. The molecular mechanisms by which protein machineries control force transmission along this molecular axis (i.e. modulating integrin activation and controlling actin polymerization) remain largely unknown. Talin is a major actin-binding protein that controls both the inside-out activation of integrins and actin filament anchoring and thus plays a major role in the establishment of the actin-extracellular matrix mechanical coupling. Talin contains three actin-binding domains (ABDs). The N-terminal head domain contains both the F3 integrin-activating domain and ABD1, whereas the C-terminal rod contains the actin-anchoring ABD2 and ABD3. Integrin binding is regulated by an intramolecular interaction between the N-terminal head and a C-terminal five-helix bundle (R9). Whether talin ABDs regulate actin polymerization in a constitutive or regulated manner has not been fully explored. Here, we combine kinetics assays using fluorescence spectroscopy and single actin filament observation in total internal reflection fluorescence microscopy, to examine relevant functions of the three ABDs of talin. We find that the N-terminal ABD1 blocks actin filament barbed-end elongation, whereas ABD2 and ABD3 do not show any activity. By mutating residues in ABD1, we find that this activity is mediated by a positively charged surface that is partially masked by its intramolecular interaction with R9. Our results also demonstrate that, once this intramolecular interaction is released, the integrin-bound talin head retains the ability to inhibit actin assembly.


Focal adhesions (FAs) mechanically couple the extracellular matrix to the dynamic actin cytoskeleton, via transmembrane integrins and actin-binding proteins. The molecular mechanisms by which protein machineries control force transmission along this molecular axis (i.e. modulating integrin activation and controlling actin polymerization) remain largely unknown. Talin is a major actin-binding protein that controls both the inside-out activation of integrins and actin filament anchoring and thus plays a major role in the establishment of the actinextracellular matrix mechanical coupling. Talin contains three actin-binding domains (ABDs). The N-terminal head domain contains both the F3 integrin-activating domain and ABD1, whereas the C-terminal rod contains the actin-anchoring ABD2 and ABD3. Integrin binding is regulated by an intramolecular interaction between the N-terminal head and a C-terminal fivehelix bundle (R9).
Whether talin ABDs regulate actin polymerization in a constitutive or regulated manner has not been fully explored. Here, we combine kinetics assays using fluorescence spectroscopy and single actin filament observation in total internal reflection fluorescence microscopy, to examine relevant functions of the three ABDs of talin. We find that the N-terminal ABD1 blocks actin filament barbed-end elongation, whereas ABD2 and ABD3 do not show any activity. By mutating residues in ABD1, we find that this activity is mediated by a positively charged surface that is partially masked by its intramolecular interaction with R9. Our results also demonstrate that, once this intramolecular interaction is released, the integrin-bound talin head retains the ability to inhibit actin assembly.
Cell adhesion to the extracellular matrix plays a critical role in many physiological functions, such as cell migration, invasion, or epithelial basement membrane attachment. Among the multiple adhesion structures, focal adhesions (FAs) 5 play a major role (1,2). These multiprotein complexes couple various extracellular matrices to the actin cytoskeleton via the transmembrane heterodimeric ␣␤ integrins and actin-binding proteins (ABPs) (3,4). The control of actin polymerization by ABPs is thought to play an important role to initiate the formation of the actomyosin stress fibers and control their tension by modulating their elongation. We showed previously that vinculin blocks actin filament barbed-end elongation (5), whereas others reported that VASP promotes the elongation of actin filament barbed ends in a processive-like manner (6,7). Several formins may also play a role in the formation and elongation of stress fibers (8). However, despite these isolated characterizations, the respective roles of the multiple ABPs and their coordination in this process are poorly understood.
The actin-binding protein talin plays a major role in FAs (9) (Fig. S1). First, it acts very early to activate integrins. In this process, the N-terminal PTB (phosphotyrosine-binding) domain, located in the head domain of talin, also known as the F3 subdomain of the FERM (four-point-one, ezrin, radixin, moesin) domain, binds to the cytoplasmic tail of the ␤ subunit of integrins to induce an allosteric conformational change through the membrane, leading to the binding of the extracellular domains to ECM with high affinity (10,11). Talin head is not constitutively active because F3 is buried by an intramolecular interaction with the R9 five-helix bundle of the rod (12,13). Several mechanisms have been proposed to explain talin head exposure in FAs. The binding of the N-terminal head of talin to a PIP2-enriched membrane or to the RIAM (Rap1-GTPinteracting adaptor molecule) protein could disrupt the F3-R9 interaction to expose F3, allowing integrin binding (13)(14)(15)(16). Talin head is also released by calpain-mediated proteolytic cleavage (17). The isolated head domain of talin, resulting from calpain cleavage, regulates adhesion dynamics (18).
In addition to its role in the inside-out activation of integrins, talin anchors the actin cytoskeleton to FAs (19). Talin contains three actin-binding domains (ABDs), also named actin-binding sites (ABSs) (20). ABD1 spans along the F2 and F3 subdomains of the FERM domain (21). ABD2 is located in the center of the rod (22). ABD3, also known as THATCH (talin/HIP1R/Sla2p actin-tethering C-terminal homology), is located at the C terminus of the protein (23). The nanoscale organization of mature FAs, revealed by superresolution, showed that the N-terminal ABD1 of talin is located proximal to the membrane, whereas ABD3 extends deeper in the cell to anchor actomyosin stress fibers (24). In this conformation, talin behaves as a mechanosensor to transduce the force conveyed by the actin cytoskeleton to FAs (25). These properties allow talin to sense the rigidity of the ECM (26). Single molecule experiments showed that force transduction involves the mechanical stretching of the talin rod domain, which exposes cryptic vinculin-binding sites (27). Our previous work showed that vinculin binding to mechanically stretched talin exposes its ABD, known as vinculin tail, and reinforces actin anchoring (28,29). In cells, actomyosin force acts on talin via its ABD3 to expose vinculin-binding sites and ABD2, locking talin into an actinbinding configuration that stabilizes FAs (30).
Although actin binding to talin plays a major role in the establishment of the actin-ECM mechanical coupling, the contribution of the three ABDs of talin to the regulation of actin polymerization is poorly understood. Here, we report the detailed characterization of the ABDs of talin. We found that the N-terminal ABD1 blocks actin filament barbed-end elongation, whereas ABD2 and ABD3 do not show any activity. This activity is masked by the intramolecular interaction between the F3 subdomain of ABD1 and R9. When ABD1 is exposed, it associates with the cytoplasmic tail of integrin-␤3 and inhibits actin filament barbed-end elongation. Our study reveals a new activity of talin and paves the way to understand the complex regulation of actin assembly in FAs.

Talin ABD1 inhibits actin filament barbed-end elongation
To determine the ability of talin constructs to regulate actin assembly (Fig. S1, A and C), we first monitored barbed-end elongation by measuring the increase in pyrenyl-labeled actin fluorescence upon polymerization in the presence of spectrinactin seeds (5). In cells, multiple parameters can modify the activity of proteins, such as the charge of membrane lipids, the addition of phosphate groups by kinases, local gradients of pH and cations, crowding effects, or the binding of partners. Hence, in many instances, it proved informative to screen a variety of physico-chemical conditions, including a mild reduction in ionic strength, before establishing the activity of actinbinding proteins, such as formins, VASP, or vinculin (5,31,32). Here, varying the ionic strength was sufficient to reveal a new inhibitory activity of ABD1 (Fig. 1A). This activity of ABD1 is higher at low ionic strength, with a half-maximum effect at 50 mM KCl (Fig. S2), which is the KCl concentration used by many groups to establish the activity of actin-binding proteins (33). In conditions where ABD1 inhibited actin polymerization, fulllength talin, ABD2, and ABD3 remained inactive (Fig. 1A).
The inhibition curve displays a sigmoidal shape (Fig. 1B), suggesting a cooperative mode of binding. Alternatively, at low concentration, ABD1 could be sequestered along the side of the filaments, preventing it from altering barbed-end dynamics. ABD1 abolished the elongation of spectrin-actin seeds but did not inhibit, or slightly enhanced, the pointed-end elongation of barbed-end capped gelsolin-actin complexes, indicating that ABD1 blocks the elongation of actin filament barbed ends specifically (Fig. 1B).
For the activity of isolated ABD1 to play a physiological role, a significant amount of ABD1-containing talin head, produced by calpain-mediated proteolysis, must exist in cells. To verify this point, we performed the biochemical fractionation of chicken gizzard tissue, extensively used in studies on FA pro- Talin head inhibits actin assembly teins, followed by Western blotting with an antibody directed against talin head (Fig. S3A). We detected talin head in the cytosolic (S1) and membrane (P1) fractions, whereas full-length talin was found associated with the membrane fraction (P1) only. In mild extraction conditions, full-length talin dissociated from the membrane fraction (S2), whereas a large amount of talin head remained associated with the membrane (P2) (Fig.  S3, B and C). These observations indicate that ABD1 probably plays a role independent of the context of the full-length protein.
The inhibition of barbed-end elongation is usually the signature of actin filament barbed-end capping proteins. However, our kinetic assays did not rule out the possibility that talin inhibitory activity results from the sequestration of actin monomers. Our results, showing that the saturation of the actin filaments by phalloidin completely abolished ABD1 activity, indicate that ABD1 acts on the filaments and not on the monomers (Fig. S4A). The effect of ABD1 on the linear relationship between the elongation rate and the concentration of polymerizing actin can also provide valuable information about the mechanism. In this assay, a capping protein is expected to decrease the slope of the curve, whereas a monomersequestering protein is expected to shift the curve without affecting the slope. We observed that, at nearly saturating concentrations of ABD1, the elongation rate is dramatically reduced (Fig. S4B). The abscissa intercept, which gives the critical concentration for actin assembly, is slightly shifted toward the critical concentration of the free pointed end, which is the usual signature of a barbed-end capping protein, but not further, ruling out an actin-monomer sequestering activity.
In cells, profilin-actin is the main polymerizable actin species, and profilin can participate in several mechanisms as a co-factor in an unpredictable manner. It was therefore important to determine whether ABD1 inhibits the elongation of actin filament barbed ends in the presence of a saturating amount of profilin. We found that profilin did not affect ABD1 activity or slightly enhanced its activity (Fig. S4A).
Finally, we directly observed the effect of ABD1 on the elongation of single actin filaments in total internal reflection fluorescence (TIRF) microscopy. In this assay, 5% biotin-labeled actin filaments are trapped on a streptavidin-biotin-PEGcoated surface to facilitate the quantification of the barbed-end elongation rate. In these conditions, we also observed that ABD1 inhibits actin assembly. At the single filament level, the details of the elongation clearly show that pauses occur in the presence of ABD1, resulting in the presence of blocked small filaments, whereas filaments elongate at the rate of the control between pauses ( Fig. 2 (A-C) and Movies S1 and S2). In the presence of ABD1, the fraction of blocked filaments increases with time to reach 50% after 600 s at 2 M ABD1 (Fig. 2D). Altogether, these observations clearly show that ABD1 blocks actin filament barbed-end elongation in a capping-like manner.

ABD1 activity is masked by an intramolecular interaction with R9
Interestingly, full-length talin, in contrast with isolated ABD1, remained inactive (Fig. 1A), suggesting that ABD1 activ-ity is masked by an intramolecular interaction. To identify the domain that inhibits ABD1 activity, we compared the activity of talin constructs encompassing the N-terminal ABD1 domain and various deletions of the rod domain. We found that a talin construct (talin 196 -1822) ending after the R9 five-helix bundle was almost inactive, whereas a talin construct (talin 196 -1659) ending before R9 was fully active (Fig. 3, A and B). These observations indicate that R9 plays a critical role to keep ABD1 inactive. Talin constructs, containing combinations of exposed ABD1 and ABD2 (talin 196 -1659) or exposed ABD1 and ABD3 (196 -2541-⌬912-2299), inhibit actin assembly, indicating that ABD1 activity does not combine with ABD2 or ABD3 to create a different activity (Fig. 3B).
To determine whether R9 was sufficient to inhibit ABD1, we expressed R9 as an isolated protein and tested its role on the inhibitory activity of isolated ABD1. Our results demonstrated that R9 is sufficient to reverse the activity of ABD1, whereas R9 alone did not affect actin assembly (Fig. 4, A and B). This result, indicating that actin and R9 binding to ABD1 are mutually exclusive, was confirmed by a co-sedimentation experiment in which increasing amounts of R9 efficiently displaced ABD1 from actin filaments (Fig. 4, C and D), without binding to actin (Fig. S5).
To define the ABD1-actin binding interface involved in the barbed-end capping-like activity, we performed a systematic mutagenesis analysis of ABD1. Because the sensitivity of ABD1 activity to ionic strength indicates that the ABD1-actin interaction is mainly electrostatic, we restricted this analysis to acidic and basic amino acids. We verified that our mutants were not unfolded by showing that they all retained the ability to interact with the side of actin filaments in a co-sedimentation assay (Fig. S6). Interestingly, we found several point mutations in basic residues that abolish ABD1 activity (Fig.  5, A and B). We also showed that the ABD1 K324E mutant, which is defective for barbed-end inhibition but not for side binding (Fig. 5 (A and B) and Fig. S6), reverses the inhibitory activity of the wildtype form of ABD1 in a kinetic assay (Fig.  S7). This result indicates that actin filament side binding is necessary but not sufficient for the inhibitory activity of ABD1, which requires an additional set of basic residues to alter barbed-end dynamics.
Because most of these basic residues have been found by previous studies (14) to be involved in PIP2 binding, we tested the effect of PIP2 micelles and found that they reverse the activity of ABD1 (Fig. 5C). When placed on the known crystal structure of ABD1 (talin F2F3), these residues form a clear binding surface that spans along the F2 and F3 subdomains of ABD1 (Fig. 5D). Interestingly, several of these basic residues of F3 were also found to interact with R9 in the crystal structure of the F3-R9 complex (12) (Fig. 5E), explaining why R9 binding to ABD1 interferes with the activity of ABD1 on actin assembly (Fig. 4).

Evidence for an integrin-talin complex that inhibits actin assembly
Previous reports also showed that R9 and the cytoplasmic tail of integrin-␤ compete for ABD1 binding (12). Whether the binding of integrin-␤ to ABD1 can release the R9-ABD1 auto-

Talin head inhibits actin assembly
inhibitory contact or keep it dissociated after talin activation to form an integrin-talin complex that inhibits barbed-end elongation is not known. Studying the binding of the cytoplasmic tail of integrin-␤ to talin head in our assays is difficult because the affinity is very low (K d ϭ 273 M for talin-1-integrin-␤3, (34)). In cells, integrin clustering could favor the binding of integrin to talin through an avidity effect. To determine the activity of an integrin-talin-actin complex, we designed a synthetic peptide of integrin-␤3 carrying mutations (DTAN)/(VE) between the membrane-distal (MD) and membrane-proximal (MP) regions that are expected to increase the integrin-talin affinity by a factor of 1000, according to similar mutations performed in integrin-␤1D (34) (Fig. S1B). We tested the ability of this integrin peptide to reverse the ABD1-R9 interaction, using the inhibition of actin assembly as a readout of ABD1 exposure. First, we showed that integrin restores the ability of ABD1 to inhibit actin assembly in the presence of a saturating concentration of R9, indicating the existence of an active integrin-ABD1-actin ternary complex that is protected from R9 inhibition (Fig. 6A). We also tested the ability of this integrin peptide to activate longer autoinhibited constructs, such as full-length talin and talin 196 -1822. This peptide induced the inhibition of

Talin head inhibits actin assembly
actin assembly by talin 196 -1822 (Fig. 6, B and C). However, this peptide could activate neither full-length talin nor full-length talin carrying the E1770A mutation, previously described to break the ABD1-R9 interaction (12), suggesting that the dimerization of full-length talin, together with additional autoinhibitory contacts, stabilizes the inactive conformation of talin (Fig. S8). We also found that the integrin-␤3 peptide phosphorylated on tyrosine 747 of the NPLY motif was slightly less efficient at activating talin 196 -1822 than the nonphosphorylated one (Fig. 6C), in agreement with previous find-ings showing that this phosphorylation reduces talin-integrin binding (35).

Discussion
Our study revealed that talin ABD1 inhibits actin filament barbed-end elongation in a regulated manner (Fig. 7, A and B). Talin adds to the list of proteins that target actin filaments barbed ends in FAs, including VASP, vinculin, tensin, and formins (5, 7, 8, 36). Like vinculin tail, talin ABD1 binds to the side of actin filaments and inhibits barbed-end elongation. In

Talin head inhibits actin assembly
both mechanisms, barbed-end inhibition requires amino acids that are not necessary for side binding. Determining the result of the combination of these similar activities in the force-dependent talin-vinculin complex will be the subject of future studies. In FAs, the elongation of actin filaments is expected to release the actomyosin-based tension in stress fibers. Conversely, the stable anchoring provided by actin filament barbedend capping in FAs could be a means to increase the traction force applied by stress fibers on FAs and ECM. Another interesting possibility is that all the barbed-end capping proteins mentioned above, including talin, anchor the filaments with their barbed ends facing FAs and their pointed ends pointing outward, to create the appropriate orientation for actomyosin contraction.
Interestingly, we identified four lysines in the F3 part of ABD1 (Lys-318, Lys-320, Lys-322, and Lys-324) that are involved in the inhibitory activity and that were previously described in the F3-R9 autoinhibitory interaction (12,13). This observation explains why an autoinhibited construct of talin (residues 196 -1822 or full-length) does not inhibit actin polymerization (Fig. 7A). Although we used a high-affinity mutant of integrin-␤3 to disrupt this interaction, as a tool to demonstrate the existence of an integrin-talin-actin complex, the affinity of the wildtype integrin-␤3 tail for talin (K d ϭ 273 M (34)) is probably not sufficient to break the ABD1-R9 intramolecular interaction. It is more likely that other regulators, such as RIAM, synergize with integrin to disrupt the ABD1-R9 interaction of the autoinhibited conformation (16), allowing ABD1 to bind actin. Alternatively, the calpain-mediated cleavage of

Talin head inhibits actin assembly
the region linking the head and the rod of talin produces enough talin head that constitutively interacts with both integrin and actin (18). Our findings also shed light on the mechanism by which actin could modulate integrin activation by talin (Fig. 7). Integrin activation occurs through outside-in and inside-out mechanisms. The outside-in mechanism involves the binding of ECM to integrin, inducing integrin extension. The inside-out mechanism involves the binding of talin F3 to the cytoplasmic tail of integrin-␤. Talin first interacts with the integrin MD site, inducing its ordering (37). This first step is followed by the binding of the lysine 324 of F3 (Lys-327 in talin-2) to an acidic residue of the integrin MP site (Asp-723 in integrin-␤3 or Asp-729 in integrin-␤1). This interaction brings the positively charged interface, made of basic residues of F2 and F3, in close proximity to the negative charges of PIP2 heads at the inner face of the plasma membrane (Fig. 7C). The interaction between talin Lys-324 and integrin-␤3 Asp-723 also disrupts the salt bridge between this acidic residue and a basic residue of the ␣ subunit (Arg-995 in ␣IIb), allowing the dissociation of the ␣ and ␤ subunits and their extension in a high-affinity conformation for ECM (38). In this conformation, talin F3 interacts with integrin MD and MP sites, and both F2 and F3 are anchored to the PIP2-containing membrane surface. The actin-binding surface that we identified along F2 and F3 is nearly identical to the one that interacts with PIP2-containing membranes, because all of the mutations that reverse the inhibition of actin assembly by ABD1 have also been described to alter the binding of PIP2 to liposomes in vitro (14). Our results confirm that PIP2 prevents ABD1 to inhibit actin polymerization. Interestingly, our mutagenesis study, showing that the lysine Lys-324 interacts with actin barbed ends, implies that actin prevents lysine Lys- . D, the amino acids whose mutation affects the activity of ABD1 are represented on the known structure (side chains in blue) (43). E, overlay of ABD1 (F2F3) (yellow) and the complex formed by F3 (orange) and R9 (gray) (12,43). The four basic amino acids (Lys-318, -320, -322, and -324) in F3 (side chains in blue) involved in the inhibition of actin polymerization are also involved in the autoinhibitory contact between ABD1 and R9. The representations of the known structures in D and E have been generated by PyMOL (Schroedinger, LLC, New York).

Talin head inhibits actin assembly
324 from interacting with the MP site, which is normally required for integrin activation. Therefore, in the integrin-ABD1-actin complex, integrin would remain inactive because the F3 part of ABD1 only associates with the integrin MD site, whereas the MP site remains dissociated. In this conformation, ABD1 would not associate with the PIP2-containing membrane leaflet, leaving enough space for an actin filament to bind. In support of this hypothesis, the interaction between talin F3 and the isolated integrin MD site has been reported by others (37,39). In addition, mutations in the MP region of integrin, including a K324D charge swapping, similar to the K324E we used, abolish integrin activation in cells and abrogate talin binding to the MP site but not to the MD site (37,38). It is generally believed that, in FAs, the ABD1-containing talin head lies on the membrane, where it activates integrin by the inside-out pathway (11), whereas the C-terminal ABD3 domain senses the actomyosin force to induce the formation of the mechanosensitive talin-vinculin complex, which is the first step toward the maturation of FAs (9). More recently, ABD2 has been involved in force transmission too (30). Our findings suggest an alternative mechanism to couple integrin to the actin cytoskeleton through the N-terminal ABD1. In this conformation, the talin rod does not sense force, and the integrin MP region cannot contribute to the inside-out integrin activation. This mechanism would allow the retrograde actin flow to drag inactive integrins inside FAs. The mechanism that we identified could also prevent inactive integrins from diffusing away from FAs between two activation events, which appears necessary for a very dynamic and adaptive process like integrin activation. Finally, it is also possible that the integrin-ABD1actin complex transmits enough force to induce the mechanically induced transition from the closed to the open high-affinity conformation of integrin (40,41).

Protein purification
All of the recombinant proteins were expressed using a similar protocol. First, the plasmid of interest was transformed in Escherichia coli (BL21 DE3, Invitrogen). Typically, transformed bacteria were grown in 4 -8 liter of LB medium containing the appropriate antibiotics at 37°C until the absorbance measured at 600 nm reached 0.6 -0.8. The culture was then incubated on ice until the temperature reached 16°C precisely. The expression of the recombinant proteins of interest was induced by adding 1 mM isopropyl 1-thio-␤-D-galactopyranoside to the medium during 16 h at 16°C. After centrifugation, the pellet was lysed by sonication in 50 mM Tris, pH 7.8, 500 mM NaCl, 1% Triton X-100, 1 mM ␤-mercaptoethanol, 10 g/ml benzamidine, and 1 mM PMSF.

Chicken gizzard fractionation
To fractionate chicken gizzards, we adapted the beginning of the classic protocol used for talin purification (42). Chicken gizzards (7 g) were mixed in 70 ml of ice-cold buffer (5 mM Tris, pH 7.0, 1 mM PMSF (protease inhibitor), and 5 mM EGTA) and homogenized in a Waring blender with three 10-s bursts at full speed. The suspension was centrifuged at 4°C for 10 min at 10,000 ϫ g. The supernatant (S1) was kept. The pellet (P1) was resuspended in 70 ml of ice-cold buffer (50 mM Tris, pH 9.0, 1 mM PMSF, and 5 mM EGTA) and homogenized in a Waring blender with three 10-s bursts at full speed. The suspension was then incubated at 37°C for 30 min and centrifuged at 4°C for 10 min at 10,000 ϫ g to separate the soluble FA proteins (S2) and the membrane-associated proteins (P2). Comparable samples of S1, P1, S2, and P2 were submitted to Western blot analysis with an antibody directed against talin head (TA205, Bio-Rad). This experiment was reproduced twice with the same results.

F-actin co-sedimentation assay
We performed co-sedimentation assays to measure the competition between actin filaments and R9 for talin ABD1. In this complex, ABD1 inhibits actin filament barbed-end (ϩ) elongation. In this situation, integrin is inactive, and the ␣ and ␤ subunits remain associated. C, ABD1 binding to a PIP2-containing membrane allows its binding with integrin MD and MP sites. In this situation, integrin is active, and the ␣ and ␤ subunits are dissociated. Actin cannot interact with ABD1.

Talin head inhibits actin assembly
software. This experiment was reproduced three times in similar conditions with the same results.

Polymerization assay
Actin polymerization was monitored by the increase in fluorescence of 10% pyrenyl-labeled actin. Actin polymerization was induced by the addition of 25 mM KCl (unless specified), 1 mM MgCl 2 , and 0.2 mM EGTA to a solution of 10% pyrenyl-labeled Ca-ATP-G-actin containing the proteins of interest. Fluorescence measurements were carried out in a Xenius spectrofluorimeter (Safas, Monaco). For kinetic experiments, 100 pM spectrin-actin seeds were added to the reaction for barbed-end elongation measurements, and 25 nM gelsolin-actin (1:2) complexes were added for pointed-end elongation measurements. All of the experiments have been reproduced two or three times with the same conclusions.

Observation and measurement of single actin filament elongation by TIRF microscopy
Our protocol is a modification of the protocol used to study vinculin activity (5). To force the filaments to grow at the surface of the coverslip, we first irradiated coverslips with deep UV radiation for 1 min and incubated them with 10% biotin-labeled PLL-PEG for 1 h at room temperature. The coverslip was then washed extensively with water and dried with a nitrogen stream. Flow cells containing 40 -60 l of liquid were prepared by sticking the PLL-PEG-coated coverslip to a slide with double-faced adhesive spacers. The chamber was first incubated with 0.1 mg/ml streptavidin for 5 min and washed with washing buffer (5 mM Tris, pH 7.8, 200 M ATP, 10 mM DTT, 1 mM MgCl 2 , 25 mM KCl). The chamber was then saturated with 10% BSA for 5 min and washed with washing buffer. The final reaction was then injected into the chamber. A typical reaction was composed of 1.5 M 5% Alexa 488, 5% biotin-labeled Mg-Gactin in 5 mM Tris, pH 7.8, 200 M ATP, 0.8% methylcellulose, 1 mM 1,4-diazabicyclo(2,2,2)-octane, 25 mM KCl, 1 mM MgCl 2 , 200 M EGTA, 10 mM DTT supplemented with 0, 0.1, 0.5, 1, or 2 M talin ABD1. Finally, we sealed the flow chamber with VALAP (a mixture of Vaseline, lanolin, and paraffin) and observed the reaction on an Olympus AX71 inverted microscope equipped with a ϫ60 (numerical aperture 1.45) objective (Olympus) and a Blues 473-nm laser (Cobolt). The time-lapse videos were acquired by Metamorph and subsequently analyzed by the ImageJ software.  Fig. 4. A. F. performed some of the experiments shown in Fig. 3. S. C. performed some of the experiments shown in Fig. 1 and Fig. S2. C. V. performed some of the experiments shown in Fig. 3. B. F. made cDNA constructs and purified proteins (Fig. S1). C. L. C. performed some of the experiments shown in Figs. 3 and 6 and Figs. S3-S8, conceived and coordinated the study, and wrote the paper.