Cofilin Promotes ActinPolymerization and Defines theDirection of Cell MotilityMousumi Ghosh,1Xiaoyan Song,1Ghassan Mouneimne,1Mazen Sidani,1David S. Lawrence,2John S. Condeelis1*A general caging method for proteins that are regulated by phosphorylation wasused to study the in vivo biochemical action of cofilin and the subsequent cellularresponse. By acute and local activation of a chemically engineered, light-sensitivephosphocofilin mimic, we demonstrate that cofilin polymerizes actin, generatesprotrusions, and determines the direction of cell migration. We propose a rolefor cofilin that is distinct from its role as an actin-depolymerizing factor.Standard genetic, biochemical, and chemi-cal tools have been unable to define theprecise temporal and spatial contributionsof the individual protein components ofsignaling pathways. For example, the pre-cise intracellular role of cofilin during cel-lular migration has been difficult to deci-pher, because motility depends on localizedtransients of spatially well-defined signal-ing activity. Results obtained from cofilinoverexpression are complicated by issues ofcompensation by phosphorylation (1, 2),modulation of expression of other motility-related proteins (3), inappropriate localiza-tion of overexpressed protein (4–7 ), andlethality of cofilin suppression (8–10). Local-ized photoactivation of protein activity al-lows one to circumvent these problems (11).In order to establish the in vivo role of cofilinand its mechanistic contributions to cell mo-tility, we prepared a mimic of inactive phos-phocofilin that can be rapidly “switched on”by a brief burst of light. This modified proteinis resistant to down-regulation by endogenousbiochemical mechanisms. We examined theeffects of both global and local release ofcofilin activity on actin polymerization, de-polymerization, protrusion, and motility. Theinstantaneous cell-wide photoactivation ofcofilin activity increased free barbed ends,F-actin content, and cellular locomotion. Fur-thermore, highly localized intracellular cofi-lin activation generated lamellipodia and de-termined the direction of cell motility. Thus,cofilin, by defining the site of actin polymer-ization to form a protrusion, acts as a com-ponent of the “steering wheel” of the cell.Activation of cofilin is required for cellmotility (12–14 ). However, it is not clear howthe activities of cofilin, which include barbedend formation and actin polymerization (14 )aswell as depolymerization (12), are coordinated.Nor has it been determined which of theseactivities predominate during protrusion andcell motility. Another unresolved question is therole of cofilin in chemotaxis. Recent analysis ofthe distribution of cofilin and phosphocofilin inmigrating fibroblasts suggests a role for activecofilin at the leading edge (15 ). However, itremains unclear if cofilin actively sets the di-rection of cell motility, alters cell polarity, orserves a more indirect role, such as actin fila-ment turnover. For example, it is possible thatcofilin-induced polymerization could generatean initial asymmetric compartment that definesthe high-affinity receptors for chemoattractants(16) as a site for subsequent recruitment andactivation of phosphatidylinositol 3-kinase andRho family G proteins. This would set thedirection of cell movement and, therefore, che-motaxis. Further, it has been demonstrated invitro that severing of existing actin filaments bycofilin initiates the side binding activity ofArp2/3 (17 ). Inside cells, this could lead toactin polymerization at the leading edge. Thelocalized light-driven activation of cofilin activ-ity is a means to address these issues.We designed a caged (inactive) form of aconstitutively active mutant of cofilin (S3C),which can be microinjected into cells andlocally activated. The serine at position 3 wasmutated to cysteine, and this S3C mutantcofilin was covalently modified with ␣-bromo-(2-nitrophenyl) acetic acid (BNPA) as1Department of Anatomy and Structural Biology,2De-partment of Biochemistry, Albert Einstein College ofMedicine, 1300 Morris Park Avenue, Bronx, NY 10461,USA.*To whom correspondence should be addressed. E-mail: [email protected]. 1. In vitro characterization of caged and photoactivatedcofilin. (A) Spectrophotometric analysis of the polymerization ofpyrenyl G-actin under different conditions as indicated, demon-strating that caged cofilin is not capable of generating new barbedends in the nucleation assay, whereas caged cofilin that has beenactivated by uncaging generates barbed ends comparable to unmodified S3C cofilin that has neverbeen caged. All plots except that for G-actin alone have 300-nM F-actin seeds. (B) Coomassieblue–stained SDS–polyacrylamide gel electrophoresis (SDS-PAGE) of samples from a sedimentationassay of cofilin binding to F-actin. Mixtures (M) of 5 ␮M F-actin, equimolar S3C cofilin, S3E, orcaged cofilin were incubated for 2 hours at room temperature and centrifuged for 20 min.SDS-PAGE analysis was performed on the pellet (P) and the supernatant (S) of each sample.Unmodified cofilin bound and pelleted with F-actin, whereas caged cofilin and S3E cofilin did notand remained in the supernatant. (C) Mass spectrometry data indicate the mass of controlunmodified S3C (top), caged S3C (middle), and uncaged S3C (bottom). The presence of a singlepeak 181 mass units higher than the control indicates a complete conversion of S3C to caged S3Cand then back to uncaged S3C on irradiation with 340-nm light. a.u., arbitrary units.R EPORTSwww.sciencemag.org SCIENCE VOL 304 30 APRIL 2004 743described elsewhere (18). Using a nucleationassay to monitor severing and a sedimenta-tion assay to monitor binding, we found thatcaged cofilin, like the constitutively inactivemutant S3E, did not bind to or sever F-actin(Fig. 1, A and B). Binding and severing werefully recovered after photoirradiation; thus,the caged cofilin mechanistically mimics in-active phosphocofilin. Furthermore, becausethe S3C form of cofilin cannot be phospho-rylated by LIM-kinase (18), photouncaginggenerates a cofilin that is impervious todown-regulation by endogenous biochemicalmechanisms. Mass spectrometric data (Fig.1C) indicate that the protein was caged at asingle site, and the failure to covalently mod-ify the S3A mutant implicates Cys3as thechemically altered residue.Single-site control of cofilin activity of-fers several advantages. First, it reduces thepossibility of unintended