Protein Disulfide Isomerase Directly Interacts with β-Actin Cys374 and Regulates Cytoskeleton Reorganization*

Background: PDI regulates cytoskeleton reorganization by the thiol-disulfide exchange in β-actin. Results: PDI directly binds to Cys374 of β-actin during cell adhesion and spreading. Conclusion: Interaction of PDI with β-actin is induced by integrin-mediated cell adhesion and promotes cytoskeleton reorganization. Significance: PDI is a new regulator of the intramolecular disulfide-thiol rearrangement of β-actin in response to αIIbβ3 integrin engagement. Recent studies support the role of cysteine oxidation in actin cytoskeleton reorganization during cell adhesion. The aim of this study was to explain whether protein disulfide isomerase (PDI) is responsible for the thiol-disulfide rearrangement in the β-actin molecule of adhering cells. First, we showed that PDI forms a disulfide-bonded complex with β-actin with a molecular mass of 110 kDa. Specific interaction of both proteins was demonstrated by a solid phase binding assay, surface plasmon resonance analysis, and immunoprecipitation experiments. Second, using confocal microscopy, we found that both proteins colocalized when spreading MEG-01 cells on fibronectin. Colocalization of PDI and β-actin could be abolished by the membrane-permeable sulfhydryl blocker, N-ethylmaleimide, by the RGD peptide, and by anti-αIIbβ3 antibodies. Consequently, down-regulation of PDI expression by antisense oligonucleotides impaired the spreading of cells and initiated reorganization of the cytoskeleton. Third, because of transfection experiments followed by immunoprecipitation and confocal analysis, we provided evidence that PDI binds to the β-actin Cys374 thiol. Formation of the β-actin-PDI complex was mediated by integrin-dependent signaling in response to the adhesion of cells to the extracellular matrix. Our data suggest that PDI is released from subcellular compartments to the cytosol and translocated toward the periphery of the cell, where it forms a disulfide bond with β-actin when MEG-01 cells adhere via the αIIbβ3 integrin to fibronectin. Thus, PDI appears to regulate cytoskeletal reorganization by the thiol-disulfide exchange in β-actin via a redox-dependent mechanism.

Recent studies support the role of cysteine oxidation in actin cytoskeleton reorganization during cell adhesion. The aim of this study was to explain whether protein disulfide isomerase (PDI) is responsible for the thiol-disulfide rearrangement in the ␤-actin molecule of adhering cells. First, we showed that PDI forms a disulfide-bonded complex with ␤-actin with a molecular mass of 110 kDa. Specific interaction of both proteins was demonstrated by a solid phase binding assay, surface plasmon resonance analysis, and immunoprecipitation experiments. Second, using confocal microscopy, we found that both proteins colocalized when spreading MEG-01 cells on fibronectin. Colocalization of PDI and ␤-actin could be abolished by the membrane-permeable sulfhydryl blocker, N-ethylmaleimide, by the RGD peptide, and by anti-␣IIb␤3 antibodies. Consequently, down-regulation of PDI expression by antisense oligonucleotides impaired the spreading of cells and initiated reorganization of the cytoskeleton. Third, because of transfection experiments followed by immunoprecipitation and confocal analysis, we provided evidence that PDI binds to the ␤-actin Cys 374 thiol. Formation of the ␤-actin-PDI complex was mediated by integrin-dependent signaling in response to the adhesion of cells to the extracellular matrix. Our data suggest that PDI is released from subcellular compartments to the cytosol and translocated toward the periphery of the cell, where it forms a disulfide bond with ␤-actin when MEG-01 cells adhere via the ␣IIb␤3 integrin to fibronectin. Thus, PDI appears to regulate cytoskeletal reorganization by the thiol-disulfide exchange in ␤-actin via a redox-dependent mechanism.
Actin, a major component of the cytoskeleton, is involved in multiple cellular functions such as proliferation, adhesion, motility, growth, and cytokinesis. It belongs to the most important targets of integrin-mediated signaling (1). Recent studies showed that its intracellular activity is directly regulated by reactive oxygen species (2). The ␤-actin molecule contains six conserved cysteine residues, but only one of them, located at position 374, is exposed on the surface of the molecule (3,4). This cysteine residue takes part in actin polymerization as well as in filament formation (5). Both processes are preceded by the intracellular oxidation of ␤-actin, which results in the formation of the disulfide bond between the sulfhydryl group Cys 374 and glutathione (4). The oxidative modifications of ␤-actin play a key role in the GSH disulfide formation upon integrin engagement and cell spreading. The glutathionylation of ␤-actin that occurs during cell adhesion leads to the disassembly of actomyosin filaments (2). On the other hand, oxidation of ␤-actin creates intermolecular disulfide bonds between Cys 374 residues of neighboring ␤-actin molecules and thus prevents ␤-actin polymerization (6). In consequence, it leads to a cytoskeleton rearrangement (7,8).
The activity of redox-sensitive proteins is usually regulated by enzymes belonging to the glutaredoxin and thioredoxin systems, which convert disulfide bond cross-linked proteins back to their forms containing free sulfhydryl groups (9,10). In this report, we attempted to evaluate the role of PDI 3 in the regulation of cellular ␤-actin activity. PDI, a member of the thioldisulfide oxidoreductase family, displays thiol isomerase, oxidase, and reductase activity. The canonical function of PDI is a disulfide bond formation in the nascent protein and protein folding in the endoplasmic reticulum (11). PDI has multiple biological functions and interacts with numerous proteins (12). Extracellular PDI is known to mediate platelet aggregation and thrombus formation (13,14). After activation of the platelets and endothelial cells, it can be found both inside and outside of the cells, in association with F-actin microfilaments or with cellular membranes, respectively (14 -18).
Confocal Microscopy-For a microscopic examination, the MEG-01 cells (0.5 ϫ 10 6 cells/ml) were, before and after transfection with pEGFP-␤-actin or its mutant pEGFP-␤-actin C374A , placed on sterile glass microscope slides that had been precoated either with fibronectin or poly-L-lysine. After incubation for 0.5, 1.0, and 2.0 h, any nonadherent cells were removed by washing with PBS, and adherent cells were fixed with 4% formaldehyde in PHEM buffer (60 mM Pipes, pH 6.9, containing 25 mM Hepes, 10 mM EGTA, 4 mM MgCl 2 , and protease inhibitors such as 1 mM PMSF, 0.1 mM EDTA, 1 mM leupeptin, and 1 mg/ml aprotinin) for 20 min at room temperature, then washed three times with PHEM buffer, and permeabilized with 0.1% Triton X-100 (v/v). The slides were then blocked with 2% BSA in PHEM buffer for 60 min at room temperature. Endogenous PDI and ␤or ␥-actin were detected in the untransfected cells with a polyclonal antibody specific to PDI (5 g) conjugated with TRITC or a mouse monoclonal to ␤-actin-CYA-FITC, or ␥-actin-CYA-CY5.5, at 37°C. To minimize cross-talk between the fluorochromes, the fluorescent signals were acquired sequentially at excitation wavelengths of 488, 546, and 633 nm. The confocal laser microscope Leica TCS SP5 system was used at the Laboratory of Confocal Microscopy in the Nencki Institute of Experimental Biology for intracellular probe visualization. Series of single 0.2-m optical sections were collected. The images were scanned at high resolution (ϫ63 oil objective, 1.4 NA). Colocalization of proteins was quantified by calculating Mander's overlap coefficient (R), ranging from 0 (minimum colocalization degree) to 1 (maximum colocalization degree) in digital images using Leica Application Suite software.
Forster Resonance Energy Transfer-To study the direct interactions of proteins that appeared to colocalize in immunocytochemical assays in vivo, a confocal microscopy-based FRET acceptor photobleaching (FRET-AP) method was used. Energy transfer was calculated as the percentage of the increase in fluorescence intensity of the donor (TRITC or Cy5.5, respectively) after photobleaching of the acceptor (FITC or TRITC) in a chosen region of the cell with a high intensity laser light. Estimation of donor fluorescence intensity before (IDpre) and after (IDpost) complete acceptor photobleaching allows the quantification of E eff as follows: For the FRET-AP assay, a Leica TCS SP5 confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany) operating with a white laser (between 610 and 660 nm) of 30 milliwatts, argon laser of 40 milliwatts, and a HeNe laser of 1.2 milliwatts was applied. The FRET-AP couple used in this study was an FITC fluorophore or TRITC fluorophore, respectively, as the donor and TRITC or Cy5.5 fluorophore, respectively, as the acceptor with the Forster radius in the range of 45-50 Å. Cells were examined with a 63ϫ, 1.4 NA Leica oil immersion objective. Photobleaching of the acceptor was performed with five sequential illuminations (five frames, 2048 ϫ 2048 resolution, line average 1, laser power 100%) of the region of interest (ROI). To excite molecules before and after bleaching, the laser line of 488 or 546 nm was used (20%). During bleaching the laser line of 546 or 633 nm (100%) was used to scan a smaller region in the cell compartments of interest. To acquire donor fluorescence, the donor channel from 500 to 540 nm or from 450 to 490 nm (donor emission intensity peak of ϳ520 or 480 nm) was opened, and fluorescence from 590 to 630 nm or from 500 to 540 nm (acceptor emission intensity peak ϳ600 nm) was collected for the acceptor channel. For spectral analysis, excitation at 488 nm was used, and the fluorescence ranging from 500 to 700 nm (10 nm steps) before and after bleaching was collected.
Construction of ␤-Actin and Its Mutant Fused with EGFP-To express ␤-actin, total RNA was isolated from the MEG-01 cells with the use of TriPure reagent, and it was reverse transcribed by using the Moloney murine leukemia virus reverse transcriptase according to the manufacturer's instructions. The resulting cDNA was a template for amplification of the fulllength gene of mature ␤-actin. A human ␤-actin sequence was amplified by PCR with Pfu polymerase using primers containing EcoRI and BamHI site, respectively, and cloned into the pEGFP-N1 vector (a forward primer, ACGGAATTCTGATG-GATGATGATATCGC, and a reverse primer, TATGGATCC-CGGAAGCATTTGCGGTG). The pEGFP-␤-actin C374A mutant was produced using the GeneTailor Site-Directed Mutagenesis System, and TCCATCGTCCACCGCAAAGCATTCTAG-GAATTC and TTTGCGGTGGACGATGGAGGGGCCGGA as primers, propagated in E. coli, were purified with Wizard Midiprep (Promega) and sequenced to confirm the open reading frame.
Transfection of Cells-MEG-01 cells were transfected with pEGFP-␤-actin or its mutant pEGFP-␤-actin C374A using Lipofectamine 2000. A total amount of 1.5 g of plasmid DNA and 6 l of Lipofectamine solution was incubated for 15 min in 200 l of Opt-MEM (Invitrogen) and then diluted with 800 l of Opti-MEM. This solution was added to the growing MEG-01 cells in 6-well plates (Falcon, Becton Dickinson). After 6 h of incubation at 37°C without serum, 2 ml of 10% FCS medium was added to the cells, and the medium was changed 12 h later to the RPMI 1640 medium. The yield of transfection was evaluated by analysis of fusion construct expression by fluorescence microscopy using a FITC filter set (Carl Zeiss, Jena, Germany).
Antisense phosphorothioateoligodeoxynucleotide (GGCA-GCGAGACTCCGAACACGGTA) and the control with a scrambled sequence (GATGGCACAAGCCTCAGAGCGA-CGG) were used to silence PDI gene expression (20). Synthetic oligonucleotides (50, 100, or 200 nM) were transfected to the MEG-01 cells with Oligofectamine (Invitrogen) for 24 h at 37°C. After transfection, the cells were gently washed and resuspended in the RPMI 1640 medium until ready for use.
Recombinant ␤-Actin Protein and Its Mutant-DNA coding of the ␤-actin sequence was amplified by PCR using the primers ACGGGATCCATGGATGATGATATCGC and TATGAAT-TCCTAGAAGCATTTGCGGTGGAC. Then the sequence was subcloned into BamHI and EcoRI sites in pRSETa. The pRSETa-␤-actin construct was propagated in E. coli (BL21Star, DE3), purified with Wizard Midiprep, and sequenced to verify the integrity of the fusion protein. Cys 374 was replaced with Ala by site-directed mutagenesis using the Gene Tailor site-directed mutagenesis system. For this purpose, the mutagenic primers TCCATCGTCCACCGCAAAGCATTCTAGGAATTC and TTTGCGGTGGACGATGGAGGGGCCGGA were used. Recombinant ␤-actin and ␤-actin C374A were expressed in E. coli transformed with pRSETa-␤-actin or pRSETA-␤-actin C374A , respectively, and then incubated for 16 h at 30°C. The harvested cells were centrifuged, resuspended in 50 mM Tris-HCl, pH 7.9, containing 2 mM EDTA, 1% Triton X-100, 1 mM PMSF, and 25 M leupeptin, and homogenized by French press at 4°C on ice followed by centrifugation (100,000 ϫ g, 60 min, 4°C) in binding buffer (5 mM imidazole, 150 mM NaCl, 20 mM Tris-HCl, pH 7.9) and sonicated on ice. The bacterial lysate was clarified by ultracentrifugation for 1 h at 100,000 ϫ g, at 4°C, filtered through a 0.45-m syringe filter, and loaded onto a His-Trap column (5 ml; Amersham Biosciences) connected to an FPLC system (Amersham Biosciences). The proteins were eluted with a 5-500 mM imidazole gradient. The purity of recombinant ␤-actin and its mutant ␤-actin C374A was confirmed by reversed phase ultra performance liquid chromatography and by SDS-PAGE.
PDI Activity-PDI activity was evaluated on the basis of its ability to renature the denatured and reduced RNase. For this purpose RNase (30 mol/l) was incubated with PDI (5 mol/ liter) for 24 h at room temperature. Then 10 l of the incubation mixture was added to 200 l of 0.44 mmol/liter cCMP dissolved in 0.1 mol/liter MOPS, pH 7.0. Changes in absorbance at 284 nm were monitored in a Unicam UV-visible spectrophotometer.
Actin Assays Using SDS-PAGE-Polymerization of recombinant ␤-actin protein and its mutant was determined by semiquantitative SDS-PAGE methods that were optimized in parallel with skeletal muscle actin and platelet actin. To ensure that the experiments started with only the actin monomer, all initial actins in G-buffer (5 mM Tris-HCl, pH 8.0, 0.2 mM CaCl 2 , and 0.2 mM ATP) were allowed to depolymerize on ice for 60 min and then were centrifuged for 30 min at 13,000 ϫ g at 4°C to sediment any remaining oligomers. Supernatants were distributed into microcentrifuge tubes, and actin polymerization buffer (500 mM KCl, 20 mM MgCl 2 , and 10 mM ATP) was added to stimulate polymerization. Polymerization was complete after 5, 15, and 60 min, and the samples were centrifuged at 16,000 ϫ g for 60 min at 25°C to separate polymerized F-actin (pellet) from monomeric G-actin (supernatant). Samples without actin polymerization buffer were made as a control (data not shown). Both the pellet and supernatant were separated by SDS-PAGE in 12% gels. Equal volume samples (15 l) were prepared with a sample buffer (1:1 sample to buffer ratio), loaded onto the gel, run at 150 V for 60 min, and visualized with Coomassie Blue. The developed gels were scanned, and the protein bands were quantitated by the Gel Doc 2000 gel documentation system (Bio-Rad).
Actin Cosedimentation Assay-Rabbit muscle actin was kept on ice overnight, diluted in G-buffer (to 20 M) containing 5 mM Tris-HCl, pH 8.0, 0.2 mM CaCl 2 , and finally incubated at room temperature for 1 h. Next, the obtained F-actin (16 M) was mixed with PDI (2 M), ␣-actinin (2 M), and BSA (2 M) in the following buffer: 5 mM Tris, pH 7.0, 2 mM MgCl 2 , 0.2 mM CaCl 2 , 50 mM KCl, 1 mM ATP, and 5 mM guanidine carbonate. As a control, a probe with only PDI was included. The reaction was incubated for 30 min at room temperature and then centrifuged at 100,000 ϫ g for 30 min at room temperature. The supernatant and pellet fractions were analyzed by SDS-PAGE, and the proteins were visualized by Coomassie Blue staining.
Binding Assays-The wells of 96-well microtiter plates were coated overnight at 4°C with PDI (5 g/ml in PBS). Unbound proteins were washed from the wells, and nonspecific binding sites were blocked by incubation with 5% skimmed milk in PBS for 2 h at room temperature. Direct binding assay was performed by adding increasing concentrations of ␤-actin to immobilized PDI or to wells coated with skimmed milk. Unbound ␤-actin was aspirated, and the wells were washed three times with PBS containing 5% skimmed milk. ␤-Actin bound to PDI was detected with polyclonal rabbit antibodies specific to ␤-actin conjugated with horseradish peroxidase. The reaction was developed using o-phenylenediamine, the reaction was stopped with 4 M H 2 SO 4 , and a change in color was determined at 490 nm. To detect nonspecific binding, all assays were done simultaneously on plates coated with skimmed milk alone and processed as described above. The background binding to the skimmed milk was subtracted from all of the samples before data analysis.
Subcellular Fractionation of Cell Proteins-Cells were transferred to fibronectin precoated plates. Then the cells were harvested at 0.5, 1.0, and 2.0 h of incubation, rinsed with cold PBS twice, and finally centrifuged for 5 min at 400 ϫ g at 4°C. Cytoplasmic, endoplasmic reticulum, and cytoskeleton fractions were obtained using the subcellular protein fractionation kit according to the manufacturer's protocol. Finally, the fractions were separated by the SDS-PAGE method in 10% gels (21). For immunoprecipitation assay, the cytoplasm fraction was isolated by NE-PER nuclear and cytoplasmic extraction reagents to obtain nondenatured, active proteins.
Coimmunoprecipitation and Immunoblotting-For the coimmunoprecipitation experiments, lysate and cytoplasmic fraction nontreated MEG-01 cells treated with N-ethylmaleimide and lysate and transfected with vectors expressing ␤-actin or its mutant ␤-actin C374A were transferred to fibronectinprecoated plates. Then the cells were harvested at 0.5, 1.0, and 2.0 h of incubation, lysed in 25 mM Hepes, pH 7.5, containing 1% CHAPS, 150 mM NaCl, 5 mM MgCl 2 , and protease inhibitors (1 mM PMSF, 0.1 mM EDTA, 1 mM leupeptin, 1 mg/ml aprotinin), and finally centrifuged for 20 min at 16,000 ϫ g at 4°C. Then the lysates were precleared with 30 l of protein A/Gagarose bead slurry for 2 h at 4°C to avoid nonspecific binding. Next, the protein concentration was measured by BCA. A total of 500 g of protein from each lysate was incubated with 2 g of anti-PDI antibodies on a rotator overnight at 4°C. Subsequently, 100 l of protein A/G-agarose bead slurry was added to each cell extract, and the incubation was continued for another 3 h. The beads were washed three times with PBS, suspended in 2ϫ concentrated SDS-PAGE loading buffer, and boiled for 5 min. Proteins released from the resin were separated by SDS-PAGE in 10% gels under reducing and nonreducing conditions (22), transferred to a nitrocellulose membrane (Bio-Rad) and immunodetected by anti-␤-actin-CYA, anti-␥actin-CYA, or anti-PDI antibodies, followed by the appropriate secondary antibodies conjugated with HRP (Santa Cruz Biotechnology). Immunodetection was accomplished by using the enhanced chemiluminescence kit and Kodak BioMax light film (Eastman Kodak). The developed films were scanned, and the protein bands were quantitated by the Gel Doc 2000 gel documentation system (Bio-Rad).
Surface Plasmon Resonance Analyses-The kinetics parameters (association and dissociation rate constants, k on and k off , respectively) and the affinity constant (K D ) between recombinant PDI and ␤-actin or its mutant form were measured by surface plasmon resonance using the Biacore X system (GE Healthcare Life Sciences). The carboxymethyl dextran matrix of CM5 Sensor Chips (GE Healthcare Life Sciences) was activated by the injection (35 l) of an 100 mM N-hydroxysuccinimide and 400 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide mixture at a 1:1 volume ratio to give reactive succinimide esters. Subsequently, PDI was covalently attached to the reactive esters of the chip via their amino groups to achieve 2000 resonance units (using a concentration of 5 g/ml in 10 mM acetate buffer, pH 4.5, and a flow rate at 5 l/min). The remaining active esters that were present on the chip surface were blocked by an injection of 1 mM ethanolamine HCl (35 l). Experiments were performed at 25°C using Hepes (150 mM NaCl, 2 mM CaCl 2 , 0.05% (v/v), 2 mM 2 mM ␤-mercaptoethanol, Surfactant P20, 10 mm Hepes, pH 7.4; flow rate was 5 l/min) as a running buffer. Different concentrations of actin or its mutant were injected (15 l), and the amount of bound PDI to immobilized ␤-actin or its mutant was monitored by measuring the variation of the surface plasmon resonance angle as a function of time. The results were expressed in resonance units, which is an arbitrary unit that is specific for the Biacore X instrument (1000 resonance units corresponds to ϳ1 ng of bound protein/mm 2 ). After each injection, the sensor chip was regenerated with a short pulse (5 l) of 10 mM glycine HCl, pH 2.5. The association rate constants, k on and k off , were determined separately from the individual association and dissociation phases, respectively. The overall affinity constant, K D , was derived from k off /k on . All calculations were prepared with BIAevaluation 3.1 software.
Statistical Analyses-The statistical significance of the differences between the experimental conditions was determined by Student's t test for unpaired groups. p values less than 0.05 were considered significant.

PDI Interacts with ␤-Actin during Cell Adhesion and
Spreading-The emerging evidence suggests that actin organizes cell adhesion and that this process can be regulated by redox mechanisms involving reactive oxygen species (2). To explain the role of PDI in the thiol-disulfide exchange occurring in the ␤-actin cytoskeleton during cell adhesion and spreading, in the preliminary experiments, we investigated the interaction between both proteins and the ␤-actin-CYA-PDI complex formation in the adhering cells. First, we assessed the location of PDI in the MEG-01 cellular fractions. For this purpose cell lysates were fractioned according to the manufacturer's protocol, and then aliquots of the cell lysates, the cytosol, membrane (containing endoplasmic reticulum proteins) and the cytoskeletal fractions were analyzed by Western immunoblotting (Fig.  1A). Subcellular fractionation of MEG-01 demonstrated that PDI was present mainly in the endoplasmic reticulum but that it was also observed in the cytosol and weakly in the cytoskele-ton fraction. The purity of cell fractions was monitored by Western blot analysis using an antibody that recognizes the protein markers of relevant cell fractions such as Ero, ERK1/2, and ␤-tubulin, respectively. Second, we searched for interaction of PDI with ␤-actin as analyzed by a solid phase binding assay. In this experiment, PDI was immobilized on microtiter wells, which was followed by incubation with various concentrations of ␤-actin. Bound ␤-actin was quantified using rabbit polyclonal anti-␤-actin antibodies and goat anti-rabbit IgG conjugated with horseradish peroxidase as the first and second antibodies, respectively. Because is visible in Fig. 1B, ␤-actin bound to immobilized PDI in a dose-dependent manner. It neatly fit the Regression Wizard hyperbola, thus indicating the tendency to saturate. Third, we attempted to detect whether the PDI-␤-actin complex can be formed in adhering cells. The MEG-01 cells were placed on fibronectin-coated dishes, collected after different time points (0, 0.5, 1.0, and 2.0 h), and subjected to immunoprecipitation (Fig. 1C). To study the direct interaction between PDI and ␤-actin-CYA, the cell extracts were incubated with mouse monoclonal antibodies specific to ␤-actin-CYA or rabbit polyclonal anti-PDI antibody. The obtained immunoprecipitates were separated by SDS-PAGE under reducing and nonreducing conditions, and after electrophoretic transferring onto a nitrocellulose membrane, they were blotted with monoclonal anti-␤-actin-CYA or polyclonal anti-PDI antibodies, respectively. To control the concentration of cell extracts, before immunoprecipitation they were analyzed by Western immunoblotting using anti-␤-actin and anti-GAPDH antibodies. Data shown in Fig. 1C indicate the presence of PDI in a high molecular complex with ␤-actin evidenced by: (a) an immunoprecipitate acquired through pull-ing down with antibodies against ␤-actin-CYA, blotted with anti-PDI antibodies, and (b) an immunoprecipitate obtained using anti-PDI antibodies blotted with antibodies against ␤-actin-CYA. There was a single band with a molecular mass of 110 kDa that was specifically recognized by the anti-PDI antibodies under nonreducing conditions. After reduction it showed the presence of two components, namely 65 and 45 kDa, corresponding to PDI and ␤-actin-CYA, respectively (Fig. 1C, middle gels). The control MEG-01 cells placed on poly-L-lysine-coated dishes did not contain such a ␤-actin-PDI complex. Interestingly, the formation of the ␤-actin-PDI complex was dependent on the time of adhesion, as evaluated by the staining intensity of the 110-kDa band. Interaction between these two proteins was most efficient after 1 h of cell adhesion to fibronectin, reaching 173% of that observed at the zero time point. It quickly decreased and approached the control values after 2 h of adhesion.
To confirm specific interaction only between PDI and ␤-actin-CYA, not ␥-actin-CYA, the cell extracts were incubated with polyclonal anti-PDI antibody, and the immunoprecipitates were separated by SDS-PAGE as described above. Western blot assay was carried out with monoclonal anti-␤-actin-CYA or anti-␥-actin-CYA antibodies, respectively. To control the protein concentration in each cell extract, before immunoprecipitation, they were analyzed by Western immunoblotting using anti-␤-actin-CYA, anti-␥-actin-CYA, and anti-GAPDH antibodies. Immunoprecipitation assay followed by blotting with monoclonal anti-␤-actin-CYA antibody demonstrated a single 110-kDa band under nonreducing conditions, and a 45-kDa corresponding to ␤-actin after reduction (Fig. 1D, left  panel). For comparison, an analogous analysis with a monoclo-FIGURE 1. Interaction of PDI with ␤-actin in MEG-01 cells adhering to fibronectin. A, subcellular fractionation of MEG-01 cells showed that PDI is localized in both the endoplasmic reticulum and the cytosol fraction. Equal amounts of cell lysate and cell fractions (cytosol, membrane containing endoplasmic reticulum proteins and cytoskeleton) were analyzed by Western immunoblotting using rabbit anti-PDI antibodies. In parallel, gels were blotted with antibodies specific to cellular compartment markers: rabbit anti-ERK1/2, anti-Ero, and mouse anti-␤-tubulin, respectively. B, the interaction of ␤-actin with PDI as analyzed by solid phase binding assay. Interaction of both proteins was monitored using rabbit polyclonal anti-␤-actin-HRP antibodies, visualized with o-phenylenediamine, and detected at 490 nm. Nonspecific binding was subtracted from all readings before data analysis. Each point represents the mean Ϯ S.D. obtained in three separate experiments performed in duplicate. C, MEG-01 cells attached to fibronectin for 0.0, 0.5, 1.0, and 2.0 h were lysed, and the protein extracts were subjected to immunoprecipitation (IP) with antibodies specific to PDI or ␤-CYA and blotted (WB) with anti-␤-actin-CYA or anti-PDI antibodies, respectively. In parallel, MEG-01 cells were placed on poly-L-lysine and analyzed under the same conditions (bottom panel). The immunoprecipitates were analyzed under nonreducing (top gels) and reducing (middle gels) conditions. The quantity of the immunoprecipitated proteins is shown below the relevant gels. For this purpose, after immunodetection using the enhanced chemiluminescence kit, the films were scanned, and the protein bands were quantitated using the Gel Doc 2000 gel documentation system (Bio-Rad). To quantify the densitometric scans, the background was subtracted, and the area for each protein peak was determined. In parallel, samples of the cell extract were blotted with control mouse IgG (not shown). Levels of ␤-actin-CYA and GAPDH (bottom gels) in each cell extract were tested before immunoprecipitation by Western immunoblotting by using specific mouse monoclonal anti-␤-actin-CYA and mouse monoclonal anti-GAPDH antibodies, respectively. In addition, MEG-01 cells placed on poly-L-lysine were analyzed under the same conditions and shown as negative controls. The data are representative of three independent experiments. Additionally, to confirm specify of PDI-␤-actin-CYA interaction, MEG-01 cells attached to fibronectin for 0.0, 0.5, 1.0, and 2.0 h were immunoprecipitated (IP) by polyclonal anti-PDI antibodies. D, then immunochemical assay (WB) with monoclonal anti-␤-actin-CYA and ␥-actin-CYA antibodies blotted under nonreducing (top gels) and reducing (middle gels) conditions was performed. Quantification proceeded as described previously. Levels of ␤-actin, ␥-actin, and GAPDH (bottom gels) in each cell extract were tested before immunoprecipitation by Western immunoblotting by using specific mouse monoclonal anti-␤-actin, anti-␥-actin, and anti-GAPDH antibodies, respectively. In addition, MEG-01 cells placed on poly-L-lysine were analyzed under the same conditions and shown as negative controls (data not shown). The data are representative of three independent experiments. Furthermore, analysis of the cytoplasm protein was done as described above. The obtained cytoplasm fraction from the MEG-01 cells attached to fibronectin for 0.0, 0.5, 1.0, and 2.0 h was treated as described in the legend to D. Quantification proceeded as described previously. E, levels of ␤-actin, PDI, and GAPDH (bottom gels) in each cell extract were tested before immunoprecipitation by Western immunoblotting by using specific mouse monoclonal anti-␤-actin-CYA, anti-GAPDH antibodies, and rabbit polyclonal anti-PDI, respectively. In addition, MEG-01 cells placed on poly-L-lysine were analyzed under the same conditions and shown as negative controls (data not shown). The data are representative of three independent experiments. F, colocalization of ␤-actin or ␥-actin and PDI in MEG-01 cells adhering to fibronectin was further evidenced by confocal microscopy. Representative thin section confocal fluorescence micrographs of MEG-01 cells adhering to fibronectin that were stained with PDI specific antibody conjugated with TRITC. Cells were counterstained with antibodies conjugated with FITC specific to ␤-actin-CYA and conjugated with Cy5.5 specific to ␥-actin-CYA. Average fluorescence derived from PDI and ␤-actin when merged revealed extensive colocalization of these proteins, which is particularly visible in the focal adhesion sites and which increased during cell spreading in contrast to PDI to ␥-actin with was very low (enlarged sections, white points). Cells, primary labeled with normal rabbit IgG, were stained with goat anti-rabbit IgG conjugated with FITC or normal mouse conjugated with TRITC or Cy5.5 antibodies and used as controls (data not shown). The Mander coefficient (R) was calculated from at least 20 confocal sections by Leica confocal software. FRET-AP efficiency analysis indicates an interaction between ␤-actin and PDI that slightly increased during cell adhesion. The increase of fluorescence intensity is converted to pseudocolor (right panels) that displays variations of pixel gray scales with color. FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 5763 nal antibody to ␥-actin-CYA indicated that PDI is not able to interact with ␥-actin (Fig. 1D, right panel). The results (Fig. 1D) suggest that PDI is able to bind only to ␤-actin and that this interaction is dependent on the time of adhesion. The observed interaction was rapidly increasing during the first hour of adhesion (ϳ173% in comparison to zero time point) and then decreased slowly during the second one. Next, interaction of both proteins was tested in the cytoplasmic fraction (Fig. 1E). For this purpose the cytoplasmic fraction was isolated under mild conditions using NE-PER nuclear and cytoplasmic extraction reagents to obtain nondenatured, active proteins. Then PDI was immunoprecipitated by polyclonal anti-PDI antibodies and separated by SDS-PAGE in 10% gels under nonreducing and reducing conditions. Blotting with a monoclonal antibody to ␤-actin-CYA showed that PDI is associated with ␤-actin in the cytoplasm as well. Their interaction, as in the whole cell lysate, reached the highest level after an hour of cell adhesion and then decreased during the second hour. To control the protein concentration in each cell extract, before immunoprecipitation they were analyzed by Western immunoblotting using anti-␤-actin-CYA, anti-PDI, and anti-GAPDH antibodies. Finally, to determine the subcellular distribution of PDI and its colocalization with ␤-actin, MEG-01 cells adhering to fibronectin were stained with antibodies that specifically recognize these proteins and were then examined by immunofluorescence (Fig. 1F). To detect PDI, the cells were treated with a polyclonalanti-PDI-TRITC antibody. Then the cells were stained with monoclonal anti-␤-actin-CYA conjugated with FITC and anti-␥-actin-CYA conjugated with CY5.5 antibodies. PDI was found in the whole cytoplasm by showing strong punctuate staining together with a diffuse cytoplasmic localization. The average fluorescence derived from PDI and ␤-actin when merged revealed perimembrane regions with colocalization of both proteins, especially in the area of cell adhesion and spreading (Fig. 1F). The observed changes in the intensity of interaction between ␤-actin and PDI, and particularly strong colocalization of the examined proteins in the lamellipodial protrusions, suggested that there was participation of PDI in the regulation of cell adhesion, whereas colocalization between ␥-actin and PDI was very poor. The colocalization was confirmed by calculating the Mander coefficient, which amounted to 0.63 Ϯ 0.06, 0.74 Ϯ 0.03, and 0.67 Ϯ 0.06 for ␤-actin and PDI and 0.21 Ϯ 0.02, 0.24 Ϯ 0.03, and 0.23 Ϯ 0.03 for ␥-actin and PDI after 0.5, 1.0, and 2.0 h of cell adhesion, respectively. Moreover, Forester resonance energy was calculated. These examinations showed that marked levels of energy transfer occurred exclusively between ␤-actin and PDI. This process is clearly illustrated in Fig. 1F, where the estimated energy transfer efficiency value between both tested proteins was appreciably high (E eff ϭ ϳ40%). FRET confirmed that direct protein interaction between ␥-actin and PDI did not take place (Fig. 1F). The E eff was very low (1.4 -1.7%).

PDI Regulates Cytoskeleton Reorganization
We applied an antisense strategy to further evidence the role of PDI in actin cytoskeleton reorganization during cell adhesion. Incubation of MEG-01 cells with 50, 100, and 200 nM of antisense oligonucleotides specific to PDI mRNA efficiently and specifically down-regulated expression of this protein, as is shown by SDS-PAGE and blotting with monoclonal anti-PDI antibodies ( Fig. 2A). After scanning these gels, it appeared that antisense oligonucleotide reduced the expression of PDI in a concentration-dependent manner which, at a concentration of 200 nM, caused ϳ70% inhibition when compared with the control oligonucleotide with a scrambled sequence. Consistently, when such cells were analyzed by confocal microscopy, they clearly showed a lack of colocalization of PDI with ␤-actin or ␥-actin (Fig. 2B). The calculated Mander coefficient was very low and ranged from 0.003 Ϯ 0.001 to 0.007 Ϯ 0.001. Downregulation of PDI in MEG-01 radically delayed their spreading onto the fibronectin. 2 h after seeding, they started changing their shape from circular to flattened with some pseudopodia. Treatment with control oligonucleotide did not affect PDI expression in the MEG-01 cells and had no effect on adhesion and spreading of cells (Fig. 2C). These cells showed numerous lamellipodial protrusions. Colocalization of PDI with ␤-actin and ␥-actin was detectable, and the Mander coefficient value was not different from those observed in the control cells.
MEG-01 cells with reduced PDI expression showed extensively diminished adhesion to fibronectin (Fig. 2D). In this experiment, the adhesion of cells transfected with control or antisense oligonucleotides to PDI was studied by using dishes coated with fibronectin or poly-L-lysine.
␤-Actin Cys 374 Is Directly Involved in Interaction with PDI-In the preliminary experiments, to evaluate the role of free thiol groups in the interaction of PDI with ␤-actin, MEG-01 cells adhering to fibronectin were treated with N-ethylmaleimide (NEM), then lysed, and used in immunoprecipitation experiments. PDI was precipitated with specific antibodies, separated by SDS-PAGE, and tested by Western immunoblotting with the anti-␤-actin antibody. Fig. 3A shows that treatment of MEG-01 cells with NEM dramatically reduced the amount of ␤-actin coimmunoprecipitated with PDI, thus indicating that these proteins poorly interact in the cellular fraction of NEM-treated cells adhering to fibronectin. The lowest panel shows concentrations of ␤-actin, PDI, and GAPDH before the experiment in cell lysates tested by Western immunoblotting. This observation was further confirmed by confocal microscopy analysis   (Fig. 3B). MEG-01 cells in which free thiols were blocked by NEM when placed on fibronectin essentially displayed the morphology of control cells adhering to poly-L-lysine. They were characterized by a disorganized actin cytoskeleton and a rounded shape, thus indicating that blocking free thiols inhibited the formation of lamellipodial extensions and effectively PDI Regulates Cytoskeleton Reorganization FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9 abolished colocalization of PDI and ␤-actin and ␥-actin (Fig.  3A). Additionally, the Mander coefficient value was very low (between 0.01 and 0.09), which confirmed the lack of colocalization between these proteins. Moreover FRET analysis showed lack of interaction between PDI and ␤-actin-CYA as well as between PDI and ␥-actin-CYA (E eff of ϳ3.0 and 2.0%, respectively). At the beginning of adhesion, interaction between the ␤-actin-PDI complex and the proteins was very low (ϳ6% in comparison to the control, nontreated cells) and remained practically unaltered for the whole time of the experiment (Fig. 3B).
Because this assay had low specificity in the next series of experiments, we focused on the role of the free thiols that were present in the ␤-actin molecule during interaction with PDI. Recent studies indicated that ␤-actin Cys 374 is glutathionylated and that this modification is essential for cell spreading and cytoskeleton reorganization (2). Therefore, to determine whether ␤-actin Cys 374 is involved in complexing with PDI, recombinant PDI, ␤-actin, and its mutant ␤-actin C374A were expressed in E. coli and purified to homogeneity. Next, their ability to polymerize was compared with that of control rabbit ␤-actin (Fig. 4A). Representative gel images showing F-actinand G-actin band intensities for commercial rabbit ␤-actin, recombinant ␤-actin and its mutant (␤-actin C374A ) formed after 5, 15, and 60 min, respectively, are illustrated in Fig. 4A. As is clearly shown, recombinant ␤-actin and its mutant demonstrated the same efficiency to form F-actin as that of control rabbit ␤-actin in the actin-binding protein buffer. No significant effects were observed in the absence of the polymerization buffer (not shown). Furthermore, they showed the same kinetics of polymerization. Fig. 4B shows that the recombinant PDI used in this experiment was enzymatically active, as evidenced by its ability to renaturate ribonuclease. Real time biomolecular interactions during complex formation between purified PDI and recombinant ␤-actin or its mutant were then monitored by surface plasmon resonance analysis. For this purpose, limiting amounts of PDI were immobilized on sensor chips through amine coupling. The obtained sensograms describing the binding of ␤-actin to immobilized PDI showed a typical association stage but a disturbed dissociation stage, thus indicating that the initial reversible interaction of both proteins became an irreversible one (Fig. 4C). The values of the association constants calculated for the reversible interaction of ␤-actin with PDI showed a value of K D equal to 5.95 ϫ 10 Ϫ10 . There was no binding of ␤-actin to the uncoated surface or to the immobilized albumin used as a control protein. When ␤-actin C374A was used in place of ␤-actin, there was essentially no interaction detectable with PDI, thus indicating that this residue plays a critical role in interaction with the PDI molecule (Table 1).
To test whether PDI can interact with F-actin, initially cosedimentation assay with commercial available rabbit actin was made (Fig. 4D). It revealed that PDI (lanes 2-5) was located in the soluble fraction after centrifugation both in the absence and in the presence of actin. ␣-Actinin, a known actin-binding protein, was used as a positive control and pelleted with actin ( lanes 8 and 9), whereas BSA, known as non-actin-binding protein, was used as a negative control and remained only in the soluble fraction (lanes 6 and 7). The obtained results suggested that PDI preferentially interacted with globular actin (G-actin).
To further examine whether the mutation of ␤-actin Cys 374 affects complex formation with PDI, the MEG-01 cells were transfected with pEGFPN1 expressing ␤-actin or its mutated form, in which cysteine was replaced by alanine (pEGFP-N1actin C374A ). First, we attempted to evaluate the expression of EGFP proteins after transfection of MEG-01 cells (Fig. 5A). Western immunoblotting of pEGFP-N1-␤-actin and pEGFP-N1-␤-actin C374A after separation of the cell lysates by SDS-PAGE showed an additional band with a molecular mass of ϳ75 kDa corresponding to ␤-actin linked to EGFP. Densitometric scanning of the blots and quantification of the expressed proteins showed that EGFP-␤-actin and EGFP-␤-actin C374A corresponded to ϳ50% of total cell ␤-actin. Then after plating the cells on fibronectin and incubation for 0 -2.0 h, the cells were lysed and used in immunoprecipitation experiments with anti-PDI antibodies (Fig. 5B). Western immunoblotting with anti-␤-actin antibodies showed a dramatic decrease in the intensity of the protein band with a molecular mass of 110 kDa that was immunoprecipitated from cells transfected with pEGFP-␤-actin C374A , when compared with that containing control pEGFP-␤-actin. Consistently, the immunoprecipitate produced by the anti-PDI antibodies contained significantly less FIGURE 4. PDI binds to G-actin dependent on the Cys 374 residue. A, the ability of recombinant wild type ␤-actin and its mutant ␤-actin C374A , platelet actin to polymerize skeletal muscle actin was compared with that of muscle ␤-actin (Rb). G-actins were polymerized during 5, 15, or 60 min and separated to the supernatant (S) and pellet (P) fractions, respectively. Equal volume samples (15 l) were prepared with sample buffer and separated by SDS-PAGE in 12% polyacrylamide gels. The gels were scanned, and the protein bands were quantitated using the Gel Doc 2000 gel documentation system (Bio-Rad). The results of these analyses are shown by plotting the amount of F-actin on the duration of time polymerization. As a control, experiments without the polymerization buffer were carried out (data not shown). B, the activity of recombinant PDI was assayed by measuring the renaturation of RNase, which was monitored by measuring absorbance at 284 nm using a Unicam UV-visible spectrophotometer. Interaction of purified recombinant ␤-actin and its mutant ␤-actin C374A with active recombinant PDI was tested by surface plasmon resonance. For this purpose, recombinant PDI was immobilized on a sensor chip CM5. C, 5-l aliquots of ␤-actin or ␤-actin C374A solutions with a concentration ranging from 10 to 25 nM were injected, and the complex formation was monitored at a flow rate of 5 l/min. In parallel, a BSA solution was injected at the same flow rate as a control protein (not shown). D, a cosedimentation assay was performed to evaluate whether PDI can interact with F-actin. PDI, ␣-actinin, and BSA were incubated with or without F-actin and centrifuged at 16,000 ϫ g for 60 min. Subsequently, pellets (P) and supernatants (S) were analyzed by SDS-PAGE in 12% gels. PDI bound to G-actin was indicated by its presence in the actin supernatant after centrifugation. Control BSA did not cosediment, whereas ␣-actinin was interacting with F-actin. These data are representative of three independent experiments. RU, resonance units.  (⌬RU ϭ 2000). The analyte was injected at 25°C, and binding was followed over time by the change in RU. The k on and k off values were determined from the association and dissociation phase, respectively, with three different concentrations of actin or its Cys 374 mutant. The apparent K D corresponds to the k off /k on ratio. ␤-actin when we used MEG-01 cells transfected with pEGFP-␤-actin C374A . Then we switched to confocal microscopy again to explore the impact of ␤-actin Cys 374 on cell adhesion and spreading that was observed at the level of a single cell. In typical experiments, the MEG-01 cells were transfected with pEGFP-␤-actin or pEGFP-␤-actin C374A by using Lipofectamine, respectively. They were placed on fibronectin and incubated for up to 2 h, then fixed with paraformaldehyde, and counterstained with anti-PDI-TRITC and anti-␥-CYA antibodies labeled by Cy5.5. Cells transfected with GFP-tagged ␤-actin (Fig. 5C) or ␤-actin C374A (Fig. 5D) showed much higher expression of the fluorescently labeled intact ␤-actin. Its distribution matched that of the endogenous ␤-actin (Fig. 5C), it showed much higher accumulation in the cytoplasm and was close to the membranes. There were heavy accumulations of GFP-tagged ␤-actin at the lamellar protrusions, and overlapping staining of PDI seen as white points was particularly evident at these sites. In contrast, MEG-01 cells transfected with pEGFP-␤ actin C374A exhibited a round morphology and an impaired actin cytoskeleton organization (Fig. 5D), similar to that observed in cells treated with N-ethylmaleimide (Fig. 3A).
Colocalization of PDI and site-mutated actin C374A was significantly lower than that observed in the control cells or in the pEGFP-␤-actin-transfected cells. Furthermore, cells transfected with pEGFP-␤-actin C374A adhered much less efficiently and showed impaired spreading on the fibronectin-coated slides. Interaction between PDI and ␥-actin-CYA was very low like in the control cells. FRET method confirmed strong interaction between pEGFP-␤-actin and PDI lack of interaction between pEGFP-␤-actin C374A and PDI (E eff between 50 -54% and ϳ6%, respectively). Fig. 5E shows adhesion of the transfected MEG-01 cells to fibronectin and poly-L-lysine. Transfection of the MEG-01 cells with pEGFP-␤-actin C374A significantly reduced adhesion to fibronectin when compared with the control cells or cells transfected with pEGFP-␤-actin. The adhering cells were quantitated by means of CyQUANT (Molecular Probes) using the Wallac VICTOR 1420 multilabel counter (PerkinElmer Life Sciences). ␣⌱⌱b␤3 Is Essential for Interaction of PDI with ␤-Actin-To examine whether ␣IIb␤3, a major integrin receptor of MEG-01, is implicated in the interaction of PDI with ␤-actin, in subse-quent experiments the cells were preincubated with RGD peptide or anti-␣IIb␤3 monoclonal antibodies, placed on fibronectin, and analyzed in terms of colocalization of both proteins. Control MEG-01 cells treated with RGE peptide or normal mouse IgG showed a typical morphology of adhering cells with numerous lamellipodial protrusions and efficient colocalization of PDI with ␤-actin (Fig. 6A). Treatment of cells with the RGD peptide or anti-␣IIb␤3 abolished the ␤-actin-PDI complex formation (Fig. 6, B and C). Such cells displayed a rounded shape with a disorganized peripheric actin cytoskeleton. However, after 2 h they started to spread and showed the appearance of pseudopodia. All those experiments were confirmed by FRET analysis.

DISCUSSION
Cell adhesion and motility play key roles in a variety of biological processes, including angiogenesis, inflammation, and wound healing. Both adhesion and motility involve the expansion of the plasma membrane, which takes place through the formation of either thin actin-rich protrusions called lamellipodia or finger-like projections known as filopodia. Variations in the rate of expansion may have striking effects on the course of numerous phenomena. The formation of both lamellipodia and filopodia engages the polymerization of actin filaments.
The data presented here demonstrate that PDI interacts exclusively with ␤-actin-CYA during integrin-mediated cell adhesion to fibronectin and that it promotes cytoskeleton reorganization. Specifically, we provided evidence that: (a) both proteins can directly interact in a purified system, as is documented by solid phase binding assay and surface plasmon resonance analysis; (b) PDI preferentially binds to G-actin; (c) intracellularly PDI and ␤-actin form a complex with a molecular mass of 110 kDa that can be immunoprecipitated with specific antibodies either to PDI or ␤-actin; the complex dissociates upon reduction; ␤-actin Cys 374 is involved in the creation of a disulfide bond with PDI and is responsible for integrity of the ␤-actin-PDI complex; mutation of the Cys 374 residue abolishes the ability of ␤-actin to interact with PDI; (d) PDI colocalizes with ␤-actin in adhering MEG-01 cells and predominantly in areas close to the membrane in the lamellipodial protrusions; (e) down-regulation of PDI expression with antisense oligonucleotide abolishes adhesion of the MEG-01 cells to fibronectin FIGURE 5. The ␤-actin Cys 374 residue is directly involved in interaction with PDI during cell adhesion to fibronectin. To evaluate the expression of EGFP-␤-actin WT or its mutant EGFP-␤-actin C374A after the transfection of MEG-01 cells, Western immunoblotting (WB) with an antibody specific to ␤-actin was performed. A, the quantity of endo-and exogenous ␤-actin is shown below the corresponding gels. For this purpose, after immunodetection using the enhanced chemiluminescence kit, the films were scanned, and the protein bands were quantitated using the Gel Doc 2000 gel documentation system (Bio-Rad). The background was subtracted, and the area for each protein peak was determined. In parallel, samples of cell extract were blotted with control mouse IgG (data not shown). B, next, MEG-01 cells transfected with pEGFP-␤-actin or its mutant pEGFP-␤-actin C374A were subjected to immunoprecipitation (IP) with rabbit polyclonal anti-PDI antibody, separated by SDS-PAGE under nonreducing and reducing conditions, and blotted with monoclonal anti-␤-actin-CYA antibody (upper gels). In parallel, samples of cell extracts used in the immunoprecipitation experiments were analyzed by Western immunoblotting by using monoclonal anti-␤-actin or anti-GAPDH antibodies (lower gels). The blots are representative of three independent experiments. C, the adhesion of control MEG-01 cells, transfected with pEGFP-␤-actin or pEGFP-␤-actin C374A , to fibronectin. Cells were added to fibronectin-coated 96-well plates, and the number of adherent cells after 0.5, 1.0, and 2.0 h of incubation was determined with the CyQuant proliferation assay Kit (Invitrogen). Nontreated cells adherent to poly-L-lysine were used as a control. The data are presented as means Ϯ S.D. obtained during three independent experiments. D and E, MEG-01 cells transfected with pEGFP-␤-actin or pEGFP-␤-actin C374A and analyzed by confocal microscopy. MEG-01 cells after transfection were placed on fibronectin for 0.5, 1.0, or 2.0 h, and then they were fixed, permeabilized, and treated with rabbit antibodies specific to PDI-conjugated TRITC and anti-␥-actin-CYA-Cy5.5 antibody. The enlarged sections show colocalization of both proteins in MEG-01 cells transfected with pEGFP-␤-actin and the lack of colocalization when the cells were transfected with pEGFP-␤-actin C374A . The pictures are representative of more than 50 cells observed during three independent experiments. The Mander coefficient (R) was calculated from at least 20 confocal sections by Leica confocal software. FRET analysis showed strong interaction between ␤-actin-CYA and PDI, whereas ␥-actin practically did not bind to PDI. The increase of fluorescence intensity is converted to pseudocolor (right panel) that displays variations of pixel gray scales with color. , and anti-␣IIb␤3 antibodies (C) and placed on fibronectin for 0.5, 1.0, and 2.0 h. Then the cells were fixed, permeabilized, and labeled with specific antibodies to PDI and ␤-actin, as described in the legend to Fig. 1. Localization and colocalization of the tested protein was analyzed under a Leica TCS SP5 confocal microscope. Cell images are representative of more than 50 cells observed during three independent experiments. The Mander coefficient (R) was calculated from at least 20 confocal sections by Leica confocal software. FRET analysis was made as previously described. and appreciably delays their spreading; and (f) the binding of PDI to ␤-actin in the adhering cells requires the functionally active ␣IIb␤3 integrin. Treatment of the cells with the RGD peptide or anti-␣⌱⌱b␤3 antibodies prevents colocalization of PDI with ␤-actin.
By using specific monoclonal antibody, we showed that PDI directly interact only with ␤-actin isoform but not with ␥-actin during adhesion. These two isoforms of actin display distinct distribution and different functional roles and participate differently in organization of cell morphology, polarity, and motility. ␤-Actin is preferentially localized in stress fiber, in circular bundles, and at cell-cell contacts, suggesting its role in cell attachment. The crucial role for ␥-actin is control of cell motility (23). Our previous work showed that PDI is able to interact with integrin (16). It is possible that PDI is recruited to the adhesion area by active form of integrin ␣IIb␤3, therefore interacting with ␤-actin in these compartments during cell adhesion and spreading. However, this supposition requires further investigations.
Recent observations indicate that reactive oxygen species that are produced during processes occurring in the ECM, for example, by adhering cells, modify the structure of cytoskeletal proteins, cause cytoskeleton reorganization, and weaken cell spreading (2). Oxidation of cysteine residues resulting from integrin-dependent transmembrane signaling seems to be a key mechanism in the regulation of the ␤-actin function. During cell adhesion, ␤-actin Cys 374 was described to be directly modified by the formation of a disulfide bond with reduced glutathione (2). S-Glutathionylation of this residue is induced by oxidative stress (24). Interestingly, ␤-actin undergoes reversible S-glutathionylation in response to stimulation by the epidermal growth factor. Deglutathionylation of G-actin leads to an ϳ6-fold increase in the rate of polymerization and the F-actin content. F-actin was found in the cell periphery, which suggests that actin polymerization is regulated by a mechanism of reversible glutathionylation (25). S-Glutathionylation is reversible in either an enzyme-dependent or -independent manner. The principle enzyme that removes glutathione from the mixed sulfides is glutaredoxin, although thioredoxin or protein disulfide isomerase can also catalyze this reaction (26,27). In other systems, PDI showed lower deglutathionylation activity when compared with glutaredoxin; however, in contrast to glutaredoxin, only PDI could create the mixed disulfide intermediate (28). Furthermore, PDI is able to form native disulfide bonds in fully glutathionylated protein substrates (29,30).
Our data showed that ␤-actin Cys 374 constitutes the binding site for PDI in cells adhering to ECM after its translocation from endoplasmic reticulum compartments toward the cytosol in response to integrin engagement. Such translocation of PDI to the cytosol was observed after activation of human neutrophils by phorbol ester (31). The disulfide-bonded complex between PDI and ␤-actin may provide a mechanism that protects cells against irreversible cytoskeleton reorganization. It was reported that oxidation of ␤-actin Cys 374 destabilizes intermonomer interactions of ␤-actin (6). On the other hand, reactive oxygen species are known to interfere with the organization of actin microfilaments and motility of many eukaryotic cells (32). Therefore, oxidation of Cys 374 could be critical for the disassembly of filaments, when the environment is more oxidative because of a shifted GSH/ GSSH equilibrium (6).
The formation of disulfide bonds between ␤-actin monomers may be fundamental during cellular redox responses to growth factors or integrin stimulation. Then the resulting ␤-actin dimmers can be incorporated into F-actin during polymerization and can affect the cross-linking of F-actin and its elasticity (33). Our study suggests that PDI can be responsible for the disulfide-thiol exchange by providing the mechanism for the regulation of ␤-actin cytoskeleton rearrangement during cell adhesion.
To sum up, our data show that PDI can be an element of the mechanism leading toward cytoskeleton reorganization. It mediates the intramolecular disulfide-thiol rearrangement of ␤-actin in response to integrin engagement and forms a complex with ␤-actin Cys 374 , which is localized in the focal adhesion plate mediating adhesion and cell spreading.