Cortactin stabilization of actin requires actin-binding repeats and linker, is disrupted by specific substitutions, and is independent of nucleotide state

The actin-binding protein cortactin promotes the formation and maintenance of actin-rich structures, including lamellipodial protrusions in fibroblasts and neuronal dendritic spines. Cortactin cellular functions have been attributed to its activation of the Arp2/3 complex, which stimulates actin branch nucleation, and to its recruitment of Rho family GTPase regulators. Cortactin also binds actin filaments and significantly slows filament depolymerization, but the mechanism by which it does so and the relationship between actin binding and stabilization are unclear. Here we elucidated the cortactin regions that are necessary and sufficient for actin filament binding and stabilization. Using actin cosedimentation assays, we found that the cortactin repeat region binds actin but that the adjacent linker region is required for binding with the same affinity as full-length cortactin. Using total internal reflection fluorescence microscopy to measure the rates of single filament actin depolymerization, we observed that cortactin–actin interactions are sufficient to stabilize actin filaments. Moreover, conserved charged residues in repeat 4 were necessary for high-affinity actin binding, and substitution of these residues significantly impaired cortactin-mediated actin stabilization. Cortactin bound actin with higher affinity than did its paralog, hematopoietic cell-specific Lyn substrate 1 (HS1), and the effects on actin stability were specific to cortactin. Finally, cortactin stabilized ADP–actin filaments, indicating that the stabilization mechanism does not depend on the actin nucleotide state. Together, these results indicate that cortactin binding to actin is necessary and sufficient to stabilize filaments in a concentration-dependent manner, specific to conserved residues in the cortactin repeats, and independent of the actin nucleotide state.

Actin filament networks provide mechanical support as well as protrusive and contractile forces essential for directed cell motility and cell and tissue morphogenesis (1,2). Proteins that regulate actin polymerization, stability, and branching are critical for cells to form diverse actin-rich subcellular structures from a common pool of monomers (3,4). The actin-binding protein cortactin promotes the formation and maintenance of actin-rich structures in diverse cell types. Cortactin is essential for dynamic protrusions of lamellipodia (5)(6)(7), maturation of invadopodia in invasive cancer cells (8 -11), integrity of endothelial cell adherens junctions (12,13), and maintenance of dendritic spine structure in neurons (14,15). Cortactin functions in cellular structures have been predominantly attributed to its binding and activation of the Arp2/3 complex, which stimulates actin filament branching (16 -18), as well as its participation in Rho family GTPase signaling (11,19,20). The necessity of cortactin Arp2/3 complex activation in cell protrusion and motility depends on cellular context (21)(22)(23)(24), suggesting that cortactin has underappreciated, Arp2/3 complexindependent functions in actin-rich structures.
In addition to its roles in Arp2/3 complex activation and cell signaling, cortactin has direct effects on actin filament stability. Cortactin employs 6.5 37-amino acid "cortactin repeats" to bind actin filaments (25). These cortactin repeats and its N-terminal acidic (NTA) 2 domain mediate proper cortactin localization in fibroblasts and neurons (5,15). EM has revealed a 23-Åresolution structure of cortactin-bound actin filaments (26), but the cortactin repeat region is natively unfolded in solution (27), limiting any high-resolution structural details of the cortactin repeat fold or the cortactin-actin interaction interface. Previous analyses of cortactin truncation or deletion mutants indicate that cortactin repeat 4 is necessary for actin binding, lamellipodial protrusion, and exosome secretion (5,6,28) and that cortactin splice variants lacking repeats 5 and 6 have reduced abilities to bind actin and support cell migration (29). In vitro, cortactin-bound actin filaments depolymerize significantly slower than actin filaments alone, and this effect depends on the concentration of cortactin in solution (30). However, it is not known whether cortactin-mediated actin stabilization requires the NTA domain, proline-rich domain, or Src homo-logy 3 (SH3) domain, which serve as scaffolding sites for protein-protein interactions. Although previous work was performed on filaments grown from ATP-actin, it is also unknown whether cortactin stabilizes ADP-actin filaments or modifies the actin-bound nucleotide to slow depolymerization.
Here we used purified cortactin proteins to define the minimal requirements for high-affinity cortactin binding to actin. These studies revealed that the cortactin repeat region together with the adjacent 77-amino acid linker region was sufficient to mediate high-affinity binding to actin. Using total internal reflection fluorescence (TIRF) microscopy to measure the depolymerization of individual actin filaments, we demonstrated that this cortactin fragment stabilized actin filaments even at subsaturating concentrations. We identified charged, highly conserved residues in the cortactin repeat region that are necessary for both high-affinity binding to and stabilization of actin filaments. We found that cortactin stabilizes ADP-actin filaments, indicating that stabilization is independent of nucleotide state. Together, these experiments demonstrate that the cortactin repeat region and adjacent linker are necessary and sufficient to bind actin and that cortactin-actin interactions are necessary and sufficient to stabilize actin filaments.

The cortactin repeat region binds actin, and the affinity is increased by a 77-amino acid adjacent linker
We used cosedimentation assays to identify the regions of cortactin that mediate high-affinity binding to actin filaments.
We also tested which repeats within the CR region (Fig. 2, A and B) mediate actin binding. Only trace amounts of CR 1-4 bound to actin in the pellet, even when present at 12 M, suggesting that this region binds actin weakly, if at all (Fig. 2, C, F, and G). CR 3-6 bound actin, albeit with significantly lower affinity than CR16. We could not saturate filaments with CR 3-6 and can only place a limit on the affinity of K d Ն 7.0 M (Fig. 2, D, F, and G). CR 3-6 with the adjacent linker region, CR36L, bound actin filaments with a higher affinity (K d ϭ 2.6 Ϯ 0.7 M) (Fig. 2, E-G), but this affinity was significantly lower than that of CR16L (p Ͻ 0.0001). Our data indicate that six cortactin repeats and the adjacent segment are sufficient to recapitulate high-affinity actin binding similar to full-length cortactin.

Cortactin-actin interactions stabilize actin filaments
Differences in cortactin:actin stoichiometry have been observed in the presence of other actin-and cortactin-interacting proteins, like the Abl2 tyrosine kinase (31). Here, all cortactin fragments lacking the NTA, proline-rich, and SH3 domains bound actin filaments at higher stoichiometry than full-length cortactin. Removal of the flanking NTA and SH3 domains may remove a steric hindrance to CR-actin interactions.

Cortactin inhibits actin filament depolymerization
We next sought to determine the minimal domains in cortactin capable of stabilizing actin filaments (30) by measuring fluorescent actin filament depolymerization rates in vitro using TIRF microscopy. In the initial phase of our work, we noted frequent pausing during filament depolymerization. Although this phenomenon has been ascribed to photoinduced crosslinking under some circumstances (33), we observed pausing under even very low laser illumination (0.015 mW/100 ms), and pausing frequency did not scale with illumination intensity. We reasoned that pauses may result from the tethering of biotinylated actin to streptavidin linkages on the coverslip. To circumvent this limitation, we first grew biotinylated actin filament seeds tethered to the coverslip via streptavidin. We then grew nonbiotinylated untethered barbed ends from these seeds with 0.24 M actin monomers for 15 min, washed out free actin monomers, and monitored filament shortening rate (Fig. 3A). Consistent with other studies of untethered filaments (34 -36), filaments shortened with fewer pauses. The slow growth of untethered barbed ends for 15 min provided several half-lives of phosphate release from ADP-P i -bound subunits in the filament (35,37).
We used this single-filament approach to measure the impact of cortactin on actin stability. Following actin monomer washout, control actin filaments grown slowly from ATP-actin monomers depolymerized at a rate of Ϫ5.8 Ϯ 1.5 subunits/s (Fig. 3, B-D), a rate very close to that reported previously for ADP-actin depolymerization using TIRF microscopy (38) 3E and Movie S1). In all, observation of untethered barbed ends reveals that cortactin stabilizes actin in a concentrationdependent manner and that it can function even at subsaturating concentrations.

Cortactin-actin interactions are necessary and sufficient to stabilize filaments
We next tested which cortactin domains were necessary and sufficient for actin filament stabilization using truncated cortactin proteins (Fig. 4, A and B). The cortactin fragment lacking the Arp2/3 complex-binding NTA domain (cort⌬NTA) stabilized actin filaments when present at 1 M (shortening rate of

Cortactin-actin interactions stabilize actin filaments
Ϫ1.4 Ϯ 0.6 subunits/s). Likewise, a cortactin fragment lacking the proline-rich and SH3 domains (cort⌬ProSH3) also slowed filament depolymerization (Ϫ1.4 Ϯ 0.1 subunits/s) (Fig. 4C). Both mutants slowed filament depolymerization to the same degree as full-length cortactin, demonstrating that the NTA, proline-rich, and SH3 domains of the protein are not necessary to stabilize actin. We also found that 1 M CR16L fragment, which binds filaments with the same affinity as full-length cortactin, was sufficient to stabilize actin filaments (Ϫ1.5 Ϯ 0.4 subunits/s depolymerization rate), whereas CR16, which has a lower affinity for actin, had a greatly reduced ability to stabilize actin (Ϫ3.5 Ϯ 1.2 subunits/s) (Fig. 4C). Together, these results demonstrate that cortactin binding to actin is sufficient to stabilize actin filaments and that cortactin truncations with lower affinity for actin filaments are less able to slow depolymerization.
To test whether the effects on actin stability were specific to cortactin, we tested whether the cortactin paralog HS1 could bind and stabilize actin filaments (Fig. 5A). HS1, which only contains 3.5 actin-binding repeats, bound actin with significantly lower affinity than cortactin (K d Ն 4.4 M, Fig. 5, B and C). When present at 1 M in TIRF microscopy studies of depolymerization, HS1 did not affect actin depolymerization, in nota-ble contrast to the stabilizing effects of all cortactin CR-containing constructs at the same concentration ( Fig. 5D). At this concentration, HS1 binds with density 1 HS1:11 actin monomers, and we demonstrate that 0.35 M cortactin slows actin depolymerization when bound 1 cortactin:16 actin monomers. Thus, although cortactin has concentration-dependent effects on actin stability, HS1 does not stabilize actin filaments when bound at a similar density.

Substitutions in conserved, charged regions in repeat 4 disrupt cortactin-actin interactions and actin filament stabilization
Previous work demonstrated that deletion of CR4 reduces/ disrupts cortactin binding to actin in vitro (5) and its ability to support cell protrusion and exocytosis (6,39). In addition, substitutions that reduce the net positive charge in cortactin repeats 3 and 4 decrease binding (28,40). We tested whether specific substitutions of well-conserved charge clusters or sites of covalent modifications within repeats 3 and 4 impacted actin binding (Fig. 6A). Replacement of residues 181-184 (KTEK) or 188 -190 (QKD) in the context of CR16L with stretches of alanine did not impact binding affinity relative to CR16L   Fig. 6, B, C, E, and F). However, replacement of residues 207-209 (DKS) in CR16L significantly reduced binding affinity (Fig. 6D). Although saturated binding was not achieved, we can place a lower limit on the affinity (DKS/AAA K d Ն 3.8 M) (Fig. 6, E and F). Mass spectrometry analysis has shown that serine 209 may be phosphorylated (41). Specific substitutions that prevented (S209A) or mimicked phosphorylation (S209D) at this site both greatly reduced the affinity for actin filaments Fig. 7), indicating that the serine residue may be important for actin binding. To test whether these specific residues impact actin stability, we measured filament depolymerization with 1 M CR16L bearing the DKS/ AAA substitutions (Fig. 6G). CR16L(DKS/AAA) did not stabilize filaments, and it yielded rates significantly different from CR16L or full-length cortactin (Ϫ4.3 Ϯ 0.4 subunits/s). Together, these results demonstrate that limited or point substitutions in the cortactin repeat region disrupt actin binding and that these residues are critical for actin filament stabilization by cortactin.

Cortactin stabilization of actin is independent of the nucleotide state of filaments
Cortactin and CR16L decreased the rate of filament depolymerization to Ϫ1.1 to Ϫ1.5 subunits/s. This closely corresponds to the rate of ATP-actin depolymerization observed previously in single-filament TIRF measurements (38). Slowing the rate of ATP hydrolysis or the rate of P i release from fila-ments represents a potential mechanism by which cortactin stabilizes actin. To evaluate this mechanism, we tested whether cortactin could stabilize filaments grown from ADP-actin independent of filament aging, ATP hydrolysis, and phosphate release (Fig. 8). ADP-actin filaments depolymerized at a rate of Ϫ7.1 filaments/s, corresponding well with rates extrapolated from EM measurements of actin filaments (42). This measurement was not statistically significantly different from the control filaments observed after aging ATP-actin, indicating that the events observed from aged filaments were predominantly from ADP-bound barbed ends. Cortactin stabilized ADP-actin filaments from depolymerization to a similar extent as aged ATP-actin (depolymerization rate, Ϫ1.8 Ϯ 0.6 subunits/s) and ATP-actin filaments measured previously (30). These data indicate that cortactin effects on filament depolymerization are not due to decreased rates of ATP hydrolysis or phosphate release, which would be expected to stabilize actin.

Discussion
Cortactin promotes the formation and maintenance of actinrich structures, including lamellipodia, invadopodia, and dendritic spines (5-11, 14, 15). Cortactin binds and activates the Arp2/3 complex and signals through other actin-binding proteins (43) and small GTPases (11,19,20). Here we demonstrate that interactions between cortactin and actin filaments are sufficient to promote actin stability. Although the CR region is sufficient to bind actin, a 77-amino acid adjacent linker is necessary to achieve the high-affinity binding of cortactin. Cortac- , which binds actin with the same affinity as full-length cortactin, also stabilizes filaments to the same degree as full-length cortactin when present at 1 M. CR16, which has reduced binding affinity, stabilizes filaments but to a significantly lesser degree than full-length cortactin. Cortactin and control data are replicated from Fig. 3 for clarity. n ϭ 3 to 5 chambers for each condition ϫ 8 -12 filaments analyzed per chamber (16 chambers for control rates). Analysis was performed grouping both by chamber and all filaments, plotting mean Ϯ S.D. One-way ANOVA was used to compare control rates and varying concentrations of cortactin (*, relative to controls; #, relative to FL-cortactin; **, p Ͻ 0.01; ****, p Ͻ 0.0001; #, p Ͻ 0.05; ####, p Ͻ 0.0001; N.S., not significant).

Cortactin-actin interactions stabilize actin filaments
tin stabilizes actin filaments at subsaturating concentrations and decreases the depolymerization rate up to 6-fold at saturation. The CR region and adjacent linker are sufficient to stabilize filaments, and specific substitutions in conserved residues within CR4 impair actin binding and actin stabilization. The cortactin paralog HS1 binds actin with lower affinity but does not stabilize actin filaments. Finally, cortactin stabilizes ADPactin filaments as well as filaments grown in ATP-actin that undergo hydrolysis and phosphate release, indicating that cortactin stabilizes actin irrespective of which nucleotide is bound. Together, these results indicate that, in addition to promoting Arp2/3 complex activation and supporting Rho GTPase activation, cortactin stabilizes actin filaments through direct cortactin-actin interactions. Cortactin binds actin with its 6.5 cortactin repeats, but here we demonstrate that the 77-amino acid adjacent linker is required to recapitulate high-affinity actin binding of fulllength cortactin (Figs. 1 and 2). High-resolution structural information is not yet available for the cortactin repeats or this adjacent segment. Some studies have proposed that cortactin is well-folded and globular in solution (40,44), but more recent biophysical data indicate that the cortactin repeats are natively unfolded and extended (27,32). The adjacent segment has often been referred to as the "helical domain" or "coiled coil," but no structural data support this or any other nomenclature (25). The isolated cortactin repeats, with or without this adjacent region, behave as a monomer in solution (27). Still, the requirement for this adjacent segment for high-affinity actin binding is also observed in the cortactin paralog HS1 (46). It is possible that the adjacent linker directly contacts actin filaments to contribute to high-affinity binding. Alternatively, this linker portion might support folding of the CR and its proper binding to filaments. In this work as well as that of others (26), we demonstrate changes in binding stoichiometry between the isolated CR and actin depending on the regions flanking the cortactin repeat region.
We measure actin stability here by observing depolymerization of untethered barbed ends from actin filaments (Figs. 3 and  4). Previous work has observed slow depolymerization of untethered actin filaments before the disassembly rate increased over time (35). Those studies used a microfluidic chamber to observe depolymerization during the washout step, labeled actin with Alexa 488 on lysine, and observed depolymer-izationimmediatelyafterfilamentgrowthwith2Mactinmonomers. Here, there was a delay in imaging (ϳ90 s) while actin monomers were washed out of the chamber, during which time any residual ADP-P i subunits at the barbed end may have dissociated, resulting in the linear depolymerization events we see. Instead of observing filaments immediately after polymerization in ATP-actin, growing actin filaments for 15 min at 0.24 M provided several half-lives for P i release, which increased the uniformity and rates of depolymerization relative to our previous studies (30). These aged ATP-actin filaments depolymerize at a similar rate as ADP-actin, strongly suggesting that the depolymerization events we measured after 15 min of slow untethered growth were from ADP-bound barbed ends. ADP-actin filaments depolymerize in our study at rates identical to those observed previously in EM measures of untethered filaments (42), confirming a lack of interference from coverslip interactions. To our knowledge, these are the first measurements of depolymerization of untethered actin filaments labeled with Oregon Green 488 on Cys-374. It is possible that, in addition to aging filaments and delayed imaging after washout, differences in the dye label could impact actin depolymerization rates.
Cortactin stabilizes ADP-actin as well as filaments grown from ATP-actin (Fig. 8), ruling out ATP hydrolysis and phosphate release as potential mechanisms of cortactin filament stabilization. Cortactin stabilizes ADP-actin despite a reduced affinity for filaments relative to ATP-actin. Based on the affinity of cortactin for ADP-actin (K d ϭ 3.9 M) (6), 1 M cortactin should only have reached 20% binding saturation on ADPactin, again indicating that cortactin stabilizes actin at subsaturating concentrations.
Substitutions that reduce cortactin affinity for actin impair its ability to stabilize actin filaments. We demonstrate that substitution of the well-conserved residues Asp-207 through Ser-209 reduces actin binding and actin stabilization by cortactin ( Fig. 6 and 7). These residues undergo serine phosphorylation in a cellular phosphoproteomics assay (41) and are cross-linked to actin when incubated with filaments (44). Substitution of serine 209 with alanine or aspartic acid disrupts actin binding, indicating that the serine side chain may have a specific effect on actin binding independent of phosphorylation or charge. CR16 alone binds actin with reduced affinity relative to CR16L and cortactin, and this reduced binding correlates with a diminished effect on actin stability. Although these trunca-

Cortactin-actin interactions stabilize actin filaments
tion mutations indicate that cortactin-actin interactions are sufficient for actin stabilization, specific substitutions of Asp-207 through Ser-209 demonstrate the necessity of cortactin-actin binding in decreasing the rate of filament depolymerization.
In contrast to cortactin, HS1 does not slow filament depolymerization (Fig. 5). We found that purified monomeric HS1 binds actin with a lower affinity than GST-tagged (hence, dimerized) HS1 (47). HS1 binds actin with a lower affinity but higher stoichiometry (1 HS1:2 actin monomers) relative to cortactin, and HS1 does not stabilize actin filaments compared with cortactin at similar concentrations or binding density on filaments. The actin-binding repeats of HS1 share 86% sequence identity with the cortactin repeats, but HS1 only has 3.5 repeats compared with 6.5 repeats in cortactin (48). We found that cortactin constructs containing only 4 repeats do not bind actin with high affinity, even when including the adjacent linker (Fig. 2). In addition, cortactin isoforms lacking repeats 5 and 6 fail to support cell migration (29), indicating that the number of cortactin repeats may be critical to the overall affinity of the protein for actin filaments in vivo.

Cortactin-actin interactions stabilize actin filaments
Cortactin may stabilize filaments through two nonmutually exclusive mechanisms. Cortactin decoration of filaments during depolymerization may change filament conformation in a way that indirectly stabilizes monomers at the barbed end. Cortactin binding changes actin conformation, but this has only been investigated at low resolution. EM suggests that cortactin widens the gap between the long-pitch strands (26), whereas biochemical studies indicate that lateral and longitudinal crosslinks among actin subunits are disrupted upon cortactin binding instead of strengthened (32). Alternatively, cortactin may stabilize filaments by directly binding and bridging adjacent actin protomers. This mode of binding has been observed previously for actin-stabilizing proteins such as coronin, Xin, nebulin, SipA, and drebrin (49 -52). Many actin-stabilizing proteins bundle filaments in vitro (53), and although cortactininduced actin bundles have been observed before (22), all filaments studied here are not part of bundles, excluding bundling as a means by which cortactin stabilizes actin. The mechanism through which cortactin interactions with actin protomers sta-bilize filaments could potentially be resolved via a high resolution cryo-EM structure of cortactin-actin complexes.
Cortactin repeat truncations (CR14, CR36) were generated using PCR-based amplification and cloned into the pMAL-TEV vector (New England Biolabs, Ipswich, MA). CR truncations were purified from E. coli as described above, with the maltose-binding protein (MBP) tag cleaved with TEV protease. Human HS1 complementary DNA was inserted into pFastBac1 with a His 6 tag using restriction digest-based cloning and purified in the same manner as cortactin above.
Actin was purified from an acetone powder of frozen chicken breast muscle (Trader Joe's, Orange, CT) by one cycle of polymerization and depolymerization (55), followed by gel filtration over S300 resin (GE Healthcare) in G buffer (2 mM Tris-HCl (pH 8.0), 0.5 mM ATP, 0.5 mM DTT, 0.1 mM CaCl 2 , and 1 mM NaN 3 ) and storage in G buffer at 4°C.
Actin was labeled at Cys-374 with Oregon Green 488 or biotin as described previously (56). Unlabeled actin monomers cortactin stabilizes ADP-actin to the same degree as filaments grown from ATP-actin undergoing ATP hydrolysis and phosphate release over 15 min. 14 total experiments were performed with ADP-actin controls and nine experiments with cortactin and ADP-actin. Aged ATP-actin filament data are replicated from Fig. 3 for clarity. Analysis was performed grouping both by chamber and all filaments, plotting mean Ϯ S.D. One-way ANOVA; *, relative to ADP-actin; ****, p Ͻ 0.0001. There was no significant difference (N.S.) between cortactin-decorated ADP-actin and slowly grown ATP-actin.

Cortactin-actin interactions stabilize actin filaments
were clarified of actin polymers/aggregates by centrifugation at 135,000 ϫ g for 2 h and polymerized in a 1:1 mixture with 2ϫ labeling buffer (2ϫ ϭ 200 mM KCl, 4 mM MgCl 2 , 50 mM imidazole (pH 7.0), 0.6 mM ATP, and 6 mM NaN 3 ) for 30 min. Actin was diluted to 24 M in 1ϫ labeling buffer. For Oregon Green labeling, a 30 mM solution of Oregon Green 488 iodoacetamide (Thermo Fisher) was prepared in dimethyl fluoride and added dropwise to polymerized actin in a 12.5ϫ molar excess. For biotin labeling, a 20 mM solution of EZ-Link Maleimide-PEG2-Biotin (Thermo Fisher) was prepared in N,N-dimethylformamide and added dropwise to polymerized actin in a 20ϫ molar excess. The labeling reactions were stirred overnight in darkness at 4°C. Labeled actin was pelleted by centrifugation at 135,000 ϫ g for 2 h and resuspended by Dounce homogenization in G buffer. After dialysis for 48 h in G buffer, labeled actin monomers were clarified at 100,000 ϫ g for 2 h, followed by gel filtration over S300 resin. The extent of biotinylation was quantified using a change in HABA/avidin (Millipore Sigma, St. Louis, MO) absorbance at 500 nm.

Actin cosedimentation assays
Cosedimentation assays were performed as described previously (31). Actin monomers were polymerized by addition of 10ϫ ME buffer (1ϫ ϭ 0.05 mM MgCl 2 and 0.2 mM EGTA) and 10ϫ KMEI buffer (1ϫ ϭ 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, and 10 mM imidazole (pH 7.0)) for 45 min. Filaments were stabilized with equimolar phalloidin. Increasing concentrations of cortactin proteins were incubated with actin filaments for 45 min with final buffer concentrations of 20 mM Hepes (pH 7.25), 110 mM KCl, 2 mM MgCl 2 , and 10 mM imidazole (pH 7). Filament-cortactin reactions were pelleted by centrifugation at 100,000 ϫ g for 45 min, with 27% of the supernatant and the entire pellet (resuspended in water) analyzed by SDS-PAGE gel and staining with Coomassie Blue R250. Gels for full-length cortactin and HS1 contained 4 M urea. The binding affinity (K d ) was quantified by plotting densitometry values of cortactin proteins in the actin pellet compared with actin in the pellet as a molar ratio. A specific binding curve (GraphPad Prism) with the equation y ϭ B max ϫ x/(K d ϩ x) was used to fit the curve, where y is specific binding, x is the concentration of the ligand, B max is the maximal binding in the same units as y, and K d is the binding affinity in the same units as x. Stoichiometry was determined at saturating concentrations using densitometry and normalizing the density to both molecular weight and a linear curve of staining standards.

TIRF microscopy experiments
Open-ended flow chambers of ϳ12 l were made using a channel cut from Parafilm placed between a slide and coverslip arranged perpendicularly. Coverslips were plasma-cleaned for 2 min using H 2 /O 2 . Coverslips were coated overnight at 60°C with 2 mg/ml methoxy-PEG5000-silane (Laysan Bio, Arab, AL) and 0.02 mg/ml biotin-PEG3400-silane in 95:4:1 (v/v) ethanol: water:acetic acid, sonicated for 2 min in ethanol, and dried with filtered air before immediately forming flow chambers. Slides were sonicated for 10 min each in 2% Alconox, water, and ethanol and stored in ethanol up to 2 days before making flow chambers. Chambers were sealed by a metal block at 90°C for a total of 10 s, followed by firm pressure with a gloved finger.
Actin monomers (0.1% biotin-labeled, 33% Oregon Greenlabeled) were polymerized at a 1 M in equal parts 2ϫ TIRF buffer and reaction buffer and flowed into the chamber using Whatman paper. After 5 min of incubation, polymerization was imaged for 2 min to identify barbed and pointed ends. An actin mixture (0% biotin-labeled, 33% Oregon Green-labeled) at 0.24 M was flowed into the chamber twice to allow nontethered barbed ends to grow for 15 min, the last 5 min of which also contained cortactin proteins of interest at varying concentrations. Finally, the chamber solution was replaced with cortactin proteins of interest in equal parts 2ϫ TIRF buffer and reaction buffer (with or without cortactin constructs) to observed depolymerization events. Exposure times were 100 ms at ϳ0.015-mW laser power, with frames collected 10 s apart during polymerization and depolymerization and 5 min apart during growth of nontethered ends.
Mg-ADP-actin monomers were prepared as described previously (57,58). Ca-ATP-actin monomers (6 M) were incubated with 10ϫ ME buffer for 5 min (to a final concentration of 1ϫ ME buffer). The resulting Mg-ATP-actin monomers were incubated with 2 mM glucose and 20 units/ml hexokinase (Millipore Sigma, St. Louis, MO) for 35 min, clarified at 100,000 ϫ g for 10 min, and used for depolymerization experiments. TIRF buffers were identical to ATP-actin experiments with two exceptions: ADP was present in the TIRF buffer to replace ATP, and TIRF buffers contained 20 units/ml hexokinase. Tethered ADP-actin seeds (1 M) were incubated in TIRF chambers for 2.5 min with little to no growth of seeds seen. Untethered ADPactin filaments were polymerized from seeds in 2 M ADPactin monomers for 5 min with and without cortactin present. Filaments depolymerized in a 1:1 mixture of reaction buffer:2ϫ TIRF buffer containing 0 M or 1 M cortactin.

Cortactin-actin interactions stabilize actin filaments
Changes in filament length were measured using ImageJ software with the JFilament plugin (59) for 8 -12 filaments per experiment and at least three experiments per condition. Background correction was performed with a rolling ball radius of 7 pixels followed by drift correction with the TemplateMatching plugin (60). Depolymerization events were observed in nontethered filaments with few pauses. If a pause occurred during depolymerization, depolymerization was quantified separately before and after the pause, and the rate for that filament was the weighted average of the two events over time. Events were seen as well in tethered filaments but not quantified because of much more pronounced pauses. For quantification, we assumed 370 actin subunits/m of filament (45).

Statistical analysis
Differences in dissociation constant values were performed with unpaired, two tailed t tests as appropriate. Significance was defined as p Ͻ 0.05. Comparisons of depolymerization rates in TIRF microscopy experiments were performed with one-way ANOVA as appropriate. We quantify and analyze our depolymerization results using each chamber as an experimental replicate. Although each filament is analyzed individually, the variance between each experiment is higher than that between filaments in one experiment. Although this approach requires more data collection, it more accurately reflects small differences in buffers, concentrations, and conditions across multiple experiments compared with considering each filament as a replicate alone.