Essential Role of Cofilin-1 in Regulating Thrombin-induced RelA/p65 Nuclear Translocation and Intercellular Adhesion Molecule 1 (ICAM-1) Expression in Endothelial Cells*

Activation of RhoA/Rho-associated kinase (ROCK) pathway and the associated changes in actin cytoskeleton induced by thrombin are crucial for activation of NF-κB and expression of its target gene ICAM-1 in endothelial cells. However, the events acting downstream of RhoA/ROCK to mediate these responses remain unclear. Here, we show a central role of cofilin-1, an actin-binding protein that promotes actin depolymerization, in linking RhoA/ROCK pathway to dynamic alterations in actin cytoskeleton that are necessary for activation of NF-κB and thereby expression of ICAM-1 in these cells. Stimulation of human umbilical vein endothelial cells with thrombin resulted in Ser3 phosphorylation/inactivation of cofilin and formation of actin stress fibers in a ROCK-dependent manner. RNA interference knockdown of cofilin-1 stabilized the actin filaments and inhibited thrombin- and RhoA-induced NF-κB activity. Similarly, constitutively inactive mutant of cofilin-1 (Cof1-S3D), known to stabilize the actin cytoskeleton, inhibited NF-κB activity by thrombin. Overexpression of wild type cofilin-1 or constitutively active cofilin-1 mutant (Cof1-S3A), known to destabilize the actin cytoskeleton, also impaired thrombin-induced NF-κB activity. Additionally, depletion of cofilin-1 was associated with a marked reduction in ICAM-1 expression induced by thrombin. The effect of cofilin-1 depletion on NF-κB activity and ICAM-1 expression occurred downstream of IκBα degradation and was a result of impaired RelA/p65 nuclear translocation and consequently, RelA/p65 binding to DNA. Together, these data show that cofilin-1 occupies a central position in RhoA-actin pathway mediating nuclear translocation of RelA/p65 and expression of ICAM-1 in endothelial cells.

The nuclear factor B (NF-B) 2 represents a ubiquitously expressed family of transcription factor participating in various biological effects ranging from immune, inflammatory, and stress-induced responses to cell fate decisions such as proliferation, differentiation, apoptosis, and tumorigenesis (1,2). The mammalian NF-B family is comprised of five members: RelA (p65), RelB, c-Rel, NF-B1 (p50 and its precursor p105), and NF-B2 (p52 and its precursor p100). A characteristic feature of these proteins is the presence of a conserved N-terminal 300-amino acid Rel homology domain that contains nuclear localization signal and is involved in dimerization, sequencespecific DNA binding, and interaction with inhibitory IB proteins. A distinguishing feature of RelA, RelB, and c-Rel from p50 and p52 is the possession of a transactivation domain within the C-terminal region. Typically, NF-B exists as a heterodimer of p50 and RelA/p65 subunits associated with IB␣, the prototype of a family of inhibitory proteins IBs that keeps NF-B in the cytoplasm by virtue of masking the nuclear localization signal of RelA/p65 (3,4). Activation of NF-B requires phosphorylation of IB␣ on two specific serine residues (Ser 32 and Ser 36 ) by a macromolecular cytoplasmic IB kinase (IKK) complex composed of the catalytic subunits IKK␣ and IKK␤ and the regulatory subunit NEMO/IKK␥ (5,6). Phosphorylation triggers the ubiquitination of IB␣ by the E3-SCF␤-TrCP ubiquitin ligase, which in turn marks it for degradation by the 26 S proteasome (7,8). The unleashed NF-B migrates to the nucleus to activate transcription of target genes including intercellular adhesion molecule-1 (ICAM-1) (9 -14), an inducible endothelial adhesive protein that serves as a ligand for ␤ 2 -integrins (CD11/ CD18) present on the surface of leukocytes (15)(16)(17). Interaction of ICAM-1 with ␤ 2 -integrins enables polymorphonuclear leukocytes to adhere firmly and stably to the vascular endothelium and to migrate across the endothelial barrier (18 -20). We have shown that RelA/p65 is an essential regulator of endothelial ICAM-1 following stimulation of protease-activated receptor-1 by thrombin, a serine protease released during intravascular coagulation initiated by tissue injury or sepsis (21,22). A key signal mediating RelA/p65 activation by thrombin involves stimulation of the small GTPase RhoA and its effector Rhoassociated kinase (23,24). Activated RhoA/ROCK leads to activation of IKK␤, which in turn mediates the release of RelA/p65 for its nuclear uptake and binding to the ICAM-1 promoter, secondary to phosphorylation and degradation of IB␣ (24).
We also showed that translocation of the released RelA/p65 to the nucleus requires dynamic alterations in the actin cytoskeleton and interfering with these alterations, whether by stabilizing or destabilizing the actin cytoskeleton using the drugs jasplakinolide or latrunculin B, respectively, inhibits nuclear accumulation of RelA/p65 and expression of ICAM-1 (25). Considered together, these data implicate RhoA/ROCK pathway in regulating NF-B activation and ICAM-1 expression by a dual mechanism involving IKK-dependent release and actin cytoskeleton-dependent translocation of RelA/p65 to the nucleus.
Among the RhoA/ROCK effectors mediating reorganization of the actin cytoskeleton include the actin-depolymerizing factor/cofilin, a family of small (15)(16)(17)(18)(19)(20) proteins that bind monomeric and filamentous actin (26,27). Cofilin regulates actin dynamics by depolymerizing actin filaments at their pointed ends or by creating new filament barbed ends for F-actin assembly through their severing activity (28,29). The status of actin polymerization/depolymerization depends on the Ser 3 phosphorylation level of cofilin (30). The phosphorylation of cofilin on this residue renders it inactive and prevents it from binding to actin, thus facilitating actin polymerization (30). This phosphorylation event is catalyzed by LIM kinases (LIMK), which in turn are phosphorylated and activated by ROCK (31)(32)(33)(34). The requirement of RhoA/ROCK in the regulation of actin dynamics and activation of NF-B by thrombin (24,25) led us to investigate the possibility that cofilin serves to link the RhoA/ROCK signaling to changes in actin dynamics and thus contributes in the mechanism of RelA/p65 nuclear translocation and ICAM-1 expression. Our data show that cofilin-1 occupies a central position in RhoA-actin pathway controlling nuclear translocation of RelA/p65 and expression of ICAM-1 in endothelial cells.

EXPERIMENTAL PROCEDURES
Reagents-Human thrombin was purchased from Enzyme Research Laboratories (South Bend, IN). Polyclonal antibodies to RelA/p65, IB␣, and ␤-actin, and a monoclonal antibody to ICAM-1 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). A rabbit polyclonal antibody to cofilin-1 was obtained from Cytoskeleton (Denver, CO), and a rabbit polyclonal antibody that detects cofilin-1 when phosphorylated at Ser 3 was from Santa Cruz Biotechnology (Santa Cruz, CA). Y27632 was purchased from Calbiochem-Novabiochem Corp. (La Jolla, CA). In addition, polyvinylidene difluoride membrane was from Millipore Corp. (Bradford, MA); plasmid maxi-kit was from Qiagen Inc.; and the protein assay kit and nitrocellulose membrane were from Bio-Rad. Alexa Fluor 488-phalloidin and Texas Red goat anti-rabbit IgG were purchased from Molecular Probes (Eugene, OR). All other materials were from VWR Scientific Products Corp. (Gaithersburg, MD).
Cell Culture-Human umbilical vein endothelial cell (HUVEC) cultures were established as described previously (35,36) by using umbilical cords collected within 48 h of delivery. The cells were cultured in gelatin-coated flasks using endothelial basal medium 2 with Bullet TM kit additives (BioWhittaker, Walkersville, MD) as described (35). For treatment, the cells were washed twice with serum-free MCDB-131 medium and incubated in same serum-free medium for 0.5-1 h prior to thrombin challenge. The cells used in the experiments were between three and six passages.
Cell Lysis and Immunoblotting-The cells were lysed in radioimmune precipitation buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.25 mM EDTA, pH 8.0, 1% deoxycholic acid, 1% Triton X-100, 5 mM NaF, 1 mM sodium orthovanadate supplemented with complete protease inhibitors (Sigma). Cell lysates were resolved by SDS-PAGE and transferred onto nitrocellulose (Bio-Rad) or polyvinylidene difluoride membranes, and the residual binding sites on the filters were blocked by incubating with 5% (w/v) nonfat dry milk in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, 0.05% Tween 20) for 1 h at room temperature or overnight at 4°C. The membranes were subsequently incubated with the indicated antibodies and developed using an ECL method as described (37).
Immunofluorescence-Cells grown on coverslips were fixed in 3.7% paraformaldehyde/PBS and permeabilized with 0.1% Triton X-100 for 5 min at room temperature as described (25,38). Permeabilized cells were rinsed three times with PBS and incubated in blocking solution (1% bovine serum albumin/PBS) for 30 min at room temperature to remove nonspecific binding of the antibody. All of the subsequent steps were carried out at room temperature, and the cells were rinsed three times in 1% bovine serum albumin/PBS between each of the steps. To localize F-actin filaments, the cells were incubated with Alexa Fluor 488-phalloidin for 20 min at room temperature in a humid chamber. RelA/p65 was detected using a rabbit polyclonal anti-RelA/p65 antibody (C-20; Santa Cruz Biotechnology) and a secondary antibody conjugated to Texas Red (Molecular Probes, Eugene, OR). DNA was stained using Hoechst Dye in PBS to visualize nuclei. The coverslips were rinsed in PBS and mounted on the slide using Vectashield mounting media (Vector Laboratories, Lincolnshire, IL). The images were obtained using fluorescence or confocal microscope (Zeiss Axioplasm).
RNAi Knockdown of Cofilin-1-SMARTpool siRNA duplexes specific for cofilin-1 and a nonspecific siRNA control were obtained from Dharmacon (Lafayette, CO). The cells were transfected with siRNA using DharmaFect1 siRNA transfection reagent (Dharmacon, Lafayette, CO) according to the manufacturer's recommendations. Briefly, 50 -100 nM siRNA was mixed with DharmaFect1 and added to cells that are 50 -60% confluent. After 24 -36 h, the cells were lysed to determine the level of cofilin-1 by immunoblotting.
cDNA Constructs and Reporter Gene Assay-The constructs pRK5-cof, pRK5-cofS3A, and pRK5-cofS3D encoding wild type (Cof-WT), phosphorylation-defective (CofS3A) and phosphomimic (Cof-S3D) forms of cofilin-1 respectively, were from Theo Rein (Max Planck Institute of Psychiatry, Munich, Germany) (39). The CofS3A was generated by replacing serine 3 with alanine, and this mutation rendered cofilin-1 constitutively active. The Cof-S3D was generated by replacing serine 3 with aspartate, and this mutation rendered cofilin-1 constitutively inactive. Expression vector encoding constitutively active form of RhoA (RhoA-CAT) is described elsewhere (24). The construct pNF-B-LUC containing five copies of consensus NF-B sequences linked to a minimal E1B promoter-luciferase gene was purchased from Stratagene (La Jolla, CA). Transfections were performed using the DEAE-dextran method essen-tially as described (40). Briefly, 5 g of DNA was mixed with 50 g/ml DEAE-dextran in serum-free endothelial basal medium 2, and the mixture was added onto cells that were 60 -80% confluent. We used 0.125 g of pTKRLUC plasmid (Promega, Madison, WI) containing Renilla luciferase gene driven by the constitutively active thymidine kinase promoter to normalize transfection efficiencies. After 1 h, the cells were incubated for 4 min with 10% dimethyl sulfoxide in serum-free endothelial basal medium 2. The cells were then washed two times with endothelial basal medium 2, 10% fetal bovine serum and grown to confluence. We achieved transfection efficiency of 16 Ϯ 3 (mean Ϯ S.D.; n ϭ 3) in these cells. Cell extracts were prepared and assayed for firefly luciferase activity using the Promega Biotech dual luciferase reporter assay system. The data were expressed as a ratio of firefly luciferase activity. For experiments examining the effect of cofilin-1 knockdown on NF-B activity, the cells were first transfected with siRNA using DharmaFect1. After 12-16 h, the cells were again transfected with pNF-B-LUC using DEAE-dextran method, and luciferase activity was determined as described above.
Cytoplasmic and Nuclear Extract Preparation-After treatments, the cells were washed two times with ice-cold Tris-buffered saline and resuspended in 400 l of buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). After 15 min, Nonidet P-40 was added to a final concentration of 0.6%. The samples were centrifuged to collect the supernatants containing cytosolic proteins for determining IB␣ degradation by immunoblotting. The pelleted nuclei were resuspended in 50 l of buffer B (20 mM HEPES, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride). After 30 min at 4°C, the lysates were centrifuged, and supernatants containing the nuclear proteins were transferred to new vials. Protein concentration of the extract was measured using a Bio-Rad protein determination kit (Bio-Rad).

RESULTS
Thrombin Induces Cofilin-1 Phosphorylation/Inactivation via a ROCK-dependent Pathway-To understand the role of cofilin-1 in thrombin-induced stress fiber formation, we first evaluated the ability of thrombin to inactivate cofilin by monitoring the phosphorylation of cofilin-1 at Ser 3 . Western blot analysis showed that thrombin induced cofilin-1 phosphorylation in a time-dependent manner. The increase in cofilin-1 phosphorylation was evident within 1 min (data not shown) and peaked between 5 and 15 min after thrombin challenge (Fig.  1A). Phosphorylation of cofilin-1 began to decline at 30 min and returned to base line by 2-4 h after thrombin stimulation (Fig.  1A). We next determined whether phosphorylation of cofilin-1 by thrombin requires the activation of RhoA effector ROCK. Pretreatment of cells with Y27632, a relatively specific inhibitor of ROCK inhibited basal as well as thrombin-induced cofilin-1 phosphorylation (Fig. 1B). We also determined the involvement of ROCK effector LIMK1 in this response. We found that RNAi knockdown of LIMK1 was effective in preventing both the basal and thrombin-induced cofilin-1 phosphorylation (supplemental Fig. S1).

Inhibition of ROCK Prevents Whereas RNAi Knockdown of Cofilin-1 Promotes Stress Fiber Formation-The involvement
of ROCK in cofilin-1 phosphorylation/inactivation prompted us to assess its role in stress fiber formation induced by thrombin. The effect of ROCK inhibition on stress fiber formation was analyzed by fluorescence microscopy. HUVECs stained with Texas Red labeled-phalloidin showed increased formation of actin stress fibers in cells stimulated with thrombin (Fig. 2,  panel c versus panel a). Inhibiting ROCK by Y27632 impaired

Regulation of RelA/p65 Nuclear Translocation by Cofilin-1
the basal as well as thrombin-induced stress fiber formation Fig.  2 (panels b and d). To determine whether ROCK promotes stress fiber formation by its ability to phosphorylate and thereby inactivate cofilin-1, we assessed the effect of depleting cofilin-1 on stress fiber formation. To this end, the cells were transfected with siRNA targeting cofilin-1 (cofilin-siRNA) or control siRNA and analyzed for cofilin-1 expression by immunoblotting. The cells transfected with cofilin-siRNA showed a marked depletion of cofilin-1 compared with cells transfected with control siRNA (Fig. 3A). Analysis of fluorescence microscopy of HUVECs stained with Alexa-488 labeled-phalloidin showed that depletion of cofilin-1 augmented basal as well as thrombin-induced stress fiber formation (Fig. 3B, panels c and d versus panels a and b), consistent with the actin depolymerizing function of cofilin. We also assessed the effect of overexpressing cofilin-1 on stress fiber formation by thrombin. The results showed that overexpression of cofilin impaired thrombin-induced stress fiber formation (supplemental Fig. S2; green cells in panel d versus b), as expected.
RNAi Knockdown of Cofilin-1 Inhibits Thrombin-induced NF-B Activity and ICAM-1 Expression-We next examined whether stabilization of the actin filaments induced by cofilin-1 knockdown had an effect on NF-B activity induced by thrombin. HUVECs were transfected with pNF-B-LUC in combination with cofilin-siRNA or control siRNA. The results showed that thrombin challenge of cells transfected with control siRNA resulted in increased NF-B-dependent reporter activity and that this response was inhibited in the cells transfected with cofilin-siRNA (Fig. 4A). In view of the essential role of NF-B in ICAM-1 transcription (10, 12), we determined whether depletion of cofilin-1 produces a similar effect on ICAM-1 expression. We found that depleting cofilin-1 was also effective in inhibiting thrombin-induced ICAM-1 expression (Fig. 4B), consistent with its effect on NF-B activity (Fig. 4A).
In reciprocal experiments, we addressed the effect of overexpression of cofilin-1, likely to destabilize the actin filaments, on NF-B activity. Transfection of cells with a construct encoding wild type cofilin-1 (Cof1-WT) resulted in inhibition of thrombin-induced NF-B activity (Fig. 5). Similarly, expression of phosphorylation-defective cofilin-1 mutant (Cof1-S3A), a constitutively active form of cofilin-1 implicated in destabilizing the actin filaments (44), also inhibited thrombin-induced NF-B activity (Fig. 5). Expression of a constitutively inactive mutant of cofilin-1 (Cof1-S3D), known to stabilize the actin filaments (26,39), also inhibited NF-B activity by thrombin (Fig. 5). Collectively, these results indicate that the dynamic reorganization of actin cytoskeleton induced by thrombin is  crucial for NF-B activity and are consistent with our previous report showing that drugs that either stabilize or destabilize the cytoskeleton inhibit NF-B activity (25).
RNAi Knockdown of Cofilin-1 Inhibits RhoA-induced NF-B Activity-We previously showed that activation of RhoA/ ROCK and alterations in actin cytoskeleton are required for thrombin-induced NF-B activation (24,25). These findings together with observations that activation of ROCK is required for cofilin-1 phosphorylation and actin stress fiber formation (Fig. 2) led us to investigate the possibility that cofilin-1 couples RhoA/ROCK to actin cytoskeleton in controlling thrombininduced NF-B activation. For this purpose, we determined the effects of depleting cofilin-1 on NF-B activity induced by RhoA. Expression of constitutively active form of RhoA (RhoA-CAT) was capable of inducing NF-B activity in the absence of thrombin challenge, and depletion of cofilin-1 inhibited this response (Fig. 6). Together, these findings suggest that engage-ment of the RhoA/ROCK/cofilin-1/ actin pathway is a critical determinant of NF-B activation following thrombin challenge of endothelial cells.
RNAi Knockdown of Cofilin-1 Inhibits Thrombin-induced Recruitment of RelA/p65 to Endogenous ICAM-1 Promoter but Fails to Prevent IB␣ Degradation-We next determined whether the reduced NF-B activity following depletion of cofilin-1 was due to inhibition of NF-B DNA binding activity. It should be noted that thrombin-induced NF-B complexes are predominantly composed of RelA/p65 homodimer (10). Electrophoretic mobility supershift assay showed that cofilin-1 knockdown substantially reduced the DNA binding of RelA/ p65 in response to thrombin challenge (Fig. 7A). Additionally, we used ChIP assay to investigate the RelA/p65 binding to the endogenous ICAM-1 promoter following thrombin challenge of endothelial cells. We found that thrombin stimulation resulted in recruitment of RelA/p65 to the endogenous ICAM-1 promoter, which was inhibited upon cofilin-1 knockdown (Fig. 7B). In control ChIP experiments using IgG, we failed to detect the occupancy of RelA/p65 to the ICAM-1 promoter (Fig. 7B), indicating the specificity of the response. To ascertain whether depletion of cofilin-1 inhibits RelA/p65 occupancy of the ICAM-1 promoter by influencing IB␣ degradation, we evaluated the effect of cofilin-1 knockdown on this response. Intriguingly, IB␣ degradation was insensitive to depletion of cofilin-1 (Fig. 7C).
RNAi Knockdown of Cofilin-1 Inhibits Thrombin-induced Nuclear Translocation of RelA/p65-We tested the possibility that reduced DNA binding of RelA/p65 following cofilin-1 depletion is a result of impaired translocation of RelA/p65 to the nucleus. Analysis of nuclear extracts by immunoblotting showed that depletion of cofilin-1 inhibited the nuclear uptake of RelA/p65 by thrombin (Fig. 8A). To further verify the suppressive effect of cofilin-1 knockdown on RelA/p65 nuclear accumulation, we employed fluorescence microscopy to analyze the cells. We noted the presence of RelA/p65 primarily in the cytoplasm of unstimulated cells, irrespective of whether they were transfected with control siRNA or cofilin-siRNA, as expected (Fig. 8B, panels a and b, red staining). Stimulation

. Effects of cofilin-1 knockdown on thrombin-induced NF-B activity and ICAM-1 expression.
A, HUVECs were transfected with cofilin-siRNA or control siRNA using DharmaFect1. Twenty-four hours later, the cells were again transfected with NF-BLUC construct using DEAE-Dextran as described under "Experimental Procedures." The cells were challenged with thrombin (5 units/ml) for 6 h. The cell extracts were prepared and assayed for luciferase activity. The data are the means Ϯ S.E. (n ϭ 4 -6 for each condition). *, different from controls (p Ͻ 0.05); #, different from thrombin-stimulated controls (p Ͻ 0.05). B, HUVECs were transfected with cofilin-siRNA or control siRNA using DharmaFect1. After 24 -36 h, the cells were challenged with thrombin (5 units/ml) for 6 h. Total cell lysates were immunoblotted with an anti-ICAM-1 or anti-cofilin-1 antibody. The bar graph represents the effect of cofilin-1 depletion on ICAM-1 protein expression. ICAM-1 protein expression normalized to actin level is expressed as an ICAM-1/actin ratio. The data are the means Ϯ S.E. (n ϭ 3 for each condition). *, different from controls (p Ͻ 0.05); #, different from thrombin-stimulated controls (p Ͻ 0.05).
with thrombin caused nuclear localization of RelA/p65 in cells transfected with control siRNA as indicated by the pink versus blue nuclei (Fig. 8B, panel c versus panel a). In contrast, thrombin failed to induce RelA/p65 nuclear translocation in cells transfected with cofilin-siRNA as indicated by blue versus pink nuclei (Fig. 8B, panel d versus panel c).

DISCUSSION
We have recently shown that activation of NF-B and expression of ICAM-1 induced by thrombin in endothelial cells is mediated by Rho GTPase and requires alterations in actin dynamics. In the present study, we define a central role of cofilin-1 in Rho-actin pathway mediating nuclear translocation of RelA/p65 and ICAM-1 expression. Our experiments show that thrombin induces phosphorylation of cofilin-1, rendering it inactive to promote stress fiber formation via Rho effector ROCK. Consistent with this, depletion of cofilin-1 caused stabilization of actin filaments. Stabilizing the actin filaments by this approach impaired thrombin-and RhoA-induced NF-B activity. In addition, depleting cofilin-1 was effective in inhibiting ICAM-1 expression by thrombin. Importantly, expression of Cof1-WT or Cof1-S3A, known to destabilize the actin filaments (39,44), also interfered with thrombin-induced NF-B activity. The effect of cofilin-1 knockdown on NF-B activity occurred downstream of IB␣ degradation and was a result of impaired RelA/p65 nuclear translocation and, consequently, RelA/p65 binding to ICAM-1 promoter. Together, these data are consistent with the notion that dynamic changes in actin cytoskeleton induced by thrombin via cofilin-1 phosphorylation/inactivation by RhoA/ROCK pathway are necessary for nuclear translocation of RelA/p65 and expression of ICAM-1 in endothelial cells.
The requirement of cofilin-1 phosphorylation in actin stress fiber formation provides further insight into the mechanisms by which thrombin regulates actin dynamics in endothelial cells. Studies have established thrombin as a strong inducer of stress fiber formation in endothelial cells (45)(46)(47); however, the role of cofilin in this response has remained to be clarified. Therefore, we examined the effect of thrombin on cofilin-1 phosphorylation and demonstrate that thrombin promotes cofilin-1 phosphorylation via activation of ROCK. Accordingly, inhibition of ROCK impaired the ability of thrombin to induce stress fiber formation. These results suggest that phosphorylation of cofilin via ROCK plays an important role in actin cytoskeleton reorganization in response to thrombin. To corroborate these results, we used another strategy whereby cofilin-1 was depleted by RNAi approach to ascertain whether knockdown of cofilin-1 augments stress fiber formation. Indeed, we observed an augmentation both in basal and thrombin-induced stress fiber formation in cells depleted of cofilin-1. Because activation of Rho is implicated in thrombin-induced stress fiber formation in endothelial cells (46,47), our findings are consistent with involvement of RhoA/ROCK in cofilin-1 phosphorylation. Recently, Gorovoy et al. (48) showed an important role of LIMK1 in the mechanism of thrombin-induced actin polymerization in endothelial cells. Given that LIMK lies downstream of RhoA/ROCK in phosphorylating cofilin (32,33,49), we also examined the effect of depleting LIMK1 on cofilin phosphorylation. Depletion of LIMK1 inhibited cofilin phosphorylation, indicating that thrombin engages RhoA/ROCK/LIMK in mediating cofilin phosphorylation to alter the actin dynamics in endothelial cell. Considering that  HUVECs were transfected with cofilin-siRNA or control siRNA using Dharma-Fect1. Twenty-four hours later, the cells were again transfected with NF-BLUC in combination with a construct encoding constitutively active form of RhoA (RhoA CAT ) using DEAE-Dextran. pcDNA3 was used as vector alone control. Sixteen hours after transfection, cell extracts were prepared and assayed for luciferase activity. The data are the means Ϯ S.E. (n ϭ 4 -6 for each condition). *, p Ͻ 0.05 compared with control siRNA control; #, p Ͻ 0.05 compared with RhoA CAT -transfected cofilin-siRNA control. thrombin controls several actin-driven responses in endothelial as well as other cell types (46, 47, 50 -52), our data are of broader significance and suggest that cofilin-1 phosphorylation may be an important regulator of essential cellular functions known to be under control of actin dynamics.
We recently identified a novel function of actin cytoskeleton in controlling thrombin-induced NF-B activity and ICAM-1 expression in endothelial cells (25). We showed that thrombin causes dynamic reorganization of actin cytoskeleton in endothelial cells and that this event is essential for NF-B activity because interfering with actin reorganization, either by stabilizing or destabilizing the actin cytoskeleton, impairs thrombin-induced NF-B activity and ICAM-1 expression (25). Therefore, we sought to determine whether cofilin-1 contributes in thrombininduced NF-B activity expression in endothelial cells. We explored this possibility by assessing the ability of phosphorylation-defective cofilin-1 (Cof1-S3A) mutant to modulate NF-B activity. It should be noted that replacing serine 3 with alanine (S3A) renders the cofilin-1 mutant constitutively active and has been shown to destabilize the actin cytoskeleton by promoting actin depolymerization (39). Destabilizing the actin cytoskeleton by expressing Cof1-S3A inhibited NF-B-dependent reporter activity in response to thrombin. Similarly, overexpression of cofilin-1 (Cof1-WT) impaired actin filament formation, consistent with its function to destabilize the actin cytoskeleton (26,53,54), and mimicked the effect of Cof1-S3A in inhibiting thrombin-induced NF-B-dependent reporter activity. These results are in agreement with our earlier data obtained with cytochalasin D and latrunculin B, two classic actin-depolymerizing drugs with distinct modes of action (25).
In reciprocal experiments, we determined the effect of phosphomimic cofilin-1 (Cof1-S3D) mutant in which negatively charged phosphate is mimicked by aspartate. Replacing serine 3 with aspartate generates a constitutively inactive mutant of cofilin-1 that stabilizes actin cytoskeleton by impairing actin depolymerization (26,53,54). Stabilizing actin cytoskeleton by this approach also caused inhibition of NF-B-dependent reporter activity induced by thrombin. To further validate these data, we evaluated whether knockdown of cofilin-1 produces a similar effect on the thrombin response. We found that siRNA-mediated silencing of cofilin-1 resulted in stabilization of the actin cytoskeleton and inhibition of NF-B activity. We also assessed whether the impaired NF-B activity following cofilin-1 knockdown leads to FIGURE 7. A, effect of cofilin-1 knockdown on thrombin-induced DNA binding of RelA/p65. HUVECs were transfected with cofilin-siRNA or control-siRNA using DharmaFect1. After 24 -36 h, the cells were challenged for 1 h with thrombin (5 units/ml). Nuclear extracts were prepared and assayed for DNA binding of RelA/p65 by electrophoretic mobility shift assay as described under "Experimental Procedures." B, effect of cofilin-1 knockdown on thrombin-induced RelA/p65 binding to the endogenous ICAM-1 promoter. HUVECs were transfected with cofilin-siRNA or control-siRNA using DharmaFect1. After 24 -36 h, the cells were challenged for 1 h with thrombin (5 units/ml), and the cross-linked chromatin was immunoprecipitated (IP) with an antibody to RelA/ p65 or IgG. The ICAM-1 promoter was detected by PCR in these samples using promoter-specific primers as described under "Experimental Procedures." Input shows the starting chromatin extracts. C, effect of cofilin-1 knockdown on thrombin-induced IB␣ degradation. HUVECs were transfected with cofilin-siRNA or control siRNA using DharmaFect1. After 24 -36 h, the cells were challenged for 1 h with thrombin (5 units/ml). The cytoplasmic extracts were prepared and immunoblotted with anti-IB␣ antibody. Actin levels were used to monitor loading. The bar graph represents the effect of cofilin-1 depletion on thrombin-induced IB␣ degradation. The IB␣ level normalized to actin level is expressed relative to the untreated control set at 1. The data are the means Ϯ S.E. (n ϭ 3-4 for each condition). *, different from controls (p Ͻ 0.05).
decreased expression of ICAM-1. A marked reduction in ICAM-1 expression was observed in cofilin-1-depeleted cells. These results are consistent with our previous finding that stabilization of actin cytoskeleton following exposure of cells to jasplakinolide, a well established actin stabilizing agent (55), prevented thrombin-induced NF-B activity and ICAM-1 expression in endothelial cells.
We addressed the mechanism by which cofilin-1 controls NF-B activity and thereby ICAM-1 expression. Activation of NF-B involves its release secondary to degradation of IB␣ (4,56). The released NF-B then migrates to the nucleus, where it binds to the promoter and activates the transcription of its target genes including ICAM-1 (10,12). We began the analysis of NF-B signaling pathway by evaluating the effect of cofilin-1 knockdown on thrombin-induced IB␣ degradation. We found that IB␣ degradation was refractory to cofilin-1 depletion. The insensitivity of IB␣ degradation excludes the possibility that depletion of cofilin-1 exerts its inhibitory effect on NF-B activity and ICAM-1 expression by interfering with thrombin activation of its receptor, protease-activated receptor-1. Intriguingly, cofilin-1 depletion caused a substantial reduction in thrombin-induced RelA/p65 binding to the ICAM-1 promoter in the nucleus. The decreased RelA/p65 binding to the ICAM-1 promoter in the face of IB␣ degradation raised the possibility of impairment in the translocation of RelA/p65 to the nucleus. Indeed, we found that cofilin-1 knockdown was associated with impaired nuclear translocation of the released RelA/p65 after thrombin stimulation. It should be stressed that these results are in accord with our previous report that changes in actin dynamics are crucial for the nuclear transport of RelA/p65 (25). Thus, our findings support the notion that cofilin-1-dependent changes in actin dynamics control NF-B activity by facilitating the nuclear translocation of RelA/p65.
The requirement of functional and dynamic actin cytoskeleton in RelA/p65 nuclear translocation is consistent with previous studies documenting an important role of actin dynamics in agonist-mediated endocytosis of ␤-adrenergic and the ␤ isoform of thromboxane A2 receptors (23,57). These studies showed that internalization of these receptors was prevented by both the agents that destabilize (latrunculin B) or stabilize (jasplakinolide) the actin cytoskeleton (23,57). Consistently, expression of Cof1-WT or Cof1-S3A each inhibited ␤-adrenergic receptor internalization induced by isoproterenol (23). Lin et al. (58) showed that shear stress engages Rho-ROCK-LIMK-Cofilin pathway to activate sterol regulatory element-binding proteins in endothelial cells and that this response is inhibited in the presence of Cof1-WT or Cof1-S3A. Extensive studies by Treisman and co-workers (59, 60) have established a crucial role of actin dynamics in regulating the nuclear uptake of MAL, the myocardin-related coactivator of serum response factor. However, unlike RelA/p65 nuclear translocation or endocytosis of ␤-adrenergic and thromboxane A2 receptors (23,25,57), nuclear accumulation of MAL, which is retained in the cytoplasm through its association with G-actin, is facilitated by signals and agents that deplete G-actin by inducing F-actin formation. Accordingly, inhibition of actin polymerization by latrunculin B prevents MAL translocation to the nucleus and consequently, serum response factor activity (59 -61). Importantly, microtubule dynamics is also implicated in the nuclear accumulation of the tumor suppressor protein p53 (62). Studies by Giannakakou et al. (62) showed that p53 associates with microtubules and that this association is important for p53 nuclear localization after DNA damage. Upon disrupting the normal microtubule dynamics, whether by stabilizing or destabilizing the microtubule cytoskeleton, this association is lost, FIGURE 8. Effect of cofilin-1 knockdown on thrombin-induced nuclear translocation of RelA/p65. A, HUVECs were transfected with cofilin-siRNA or control siRNA using DharmaFect1. After 24 -36 h, the cells were challenged for 1 h with thrombin (5 units/ml). The nuclear extracts were separated by SDS-PAGE and immunoblotted with anti-RelA/p65 antibody. Actin levels were used to monitor loading. B, HUVECs grown on coverslips were transfected with cofilin-siRNA or control siRNA using DharmaFect1. After 24 -36 h, the cells were left untreated (panels a and b) or challenged for 1 h with 5 units/ml thrombin (panels c and d). The cells were then fixed, permeabilized, and stained with anti-RelA/p65 antibody and a secondary antibody conjugated with Texas Red as described under "Experimental Procedures." DNA was stained using Hoechst Dye to visualize the nucleus. The coverslips were mounted on the slide and analyzed by fluorescence microscopy. The results are representative of two separate experiments. Bar, 10 m.
leading to impaired translocation of p53 from cytosol to the nucleus (62).
We previously showed that thrombin activation of RhoA/ ROCK pathway stimulates IKK␤, which in turn promotes the release of RelA/p65 for its nuclear uptake by catalyzing the phosphorylation and thereby degradation of IB␣ (24). Considered together, these and our present findings are consistent with a model wherein thrombin engages RhoA/ROCK to serve dual function in regulating NF-B activity and ICAM-1 expression by coordinating the IKK-dependent release and cofilin-dependent changes in actin dynamics required for the transport of RelA/p65 to the nucleus (Fig. 9). In support of this possibility, we found that expression of constitutively active mutant of RhoA (RhoA-CAT) was sufficient to induce NF-B-dependent reporter activity and that this response was inhibited in cells depleted of cofilin-1. It is, however, unclear how cofilin-1-dependent changes facilitate RelA/p65 nuclear accumulation. Given that thrombin induces myosin light chain phosphorylation and thus promotes actin-myosin interaction (46,47), it is reasonable to suspect that the putative transporting system may involve motor proteins such as myosin to mediate RelA/ p65 transport toward nucleus. Such a possibility is supported by the ability of blebbstatin, an inhibitor of myosin, to prevent ICAM-1 expression by thrombin (supplemental Fig. S3). However, additional studies are required to unequivocally address this possibility.
In conclusion, our results identify cofilin-1 as a critical determinant of thrombin-induced RelA/p65 nuclear translocation and thereby ICAM-1 expression by virtue of controlling the actin dynamics downstream of RhoA/ROCK pathway. Thus, the specific targeting of cofilin-1 may be a useful strategy for dampening the thrombin-activated inflammatory responses associated with intravascular coagulation.