Reversible Glutathionylation Regulates Actin Polymerization in A431 Cells*

In response to growth factor stimulation, many mammalian cells transiently generate reactive oxygen species (ROS) that lead to the elevation of tyrosine-phosphorylated and glutathionylated proteins. While investigating EGF-induced glutathionylation in A431 cells, paradoxically we found deglutathionylation of a major 42-kDa protein identified as actin. Mass spectrometric analysis revealed that the glutathionylation site is Cys-374. Deglutathionylation of the G-actin leads to about a 6-fold increase in the rate of polymerization.In vivo studies revealed a 12% increase in F-actin content 15 min after EGF treatment, and F-actin was found in the cell periphery suggesting that in response to growth factor, actin polymerizationin vivo is regulated by a reversible glutathionylation mechanism. Deglutathionylation is most likely catalyzed by glutaredoxin (thioltranferase), because Cd(II), an inhibitor of glutaredoxin, inhibits intracellular actin deglutathionylation at 2 μm, comparable with its IC50 in vitro. Moreover, mass spectral analysis showed efficient transfer of GSH from immobilized S-glutathionylated actin to glutaredoxin. Overall, this study revealed a novel physiological relevance of actin polymerization regulated by reversible glutathionylation of the penultimate cysteine mediated by growth factor stimulation.

The mechanisms involving extracellular stimuli-induced ROS 1 generation in cells and subsequent signal propagating events are under active investigation (1)(2)(3)(4)(5). In response to growth factor, ROS, such as O 2 Ϫ or H 2 O 2 , effectively inhibit protein-tyrosine phosphatases (PTPs) by converting the active site cysteines to sulfenic acids. The sulfenic derivatives are readily S-glutathionylated by GSH, which is abundant in mammalian cells (6), thereby avoiding further oxidation to irreversible sulfinic or sulfonic derivatives (5). PTP inhibition leads to elevation of tyrosine-phosphorylated proteins, thereby promoting signal transduction cascades. Reactivation of glutathionylated PTP is most likely catalyzed by glutaredoxin, characterized as the primary enzyme for protein-SSG deglutathionylation in cells (7,8). As the intracellular redox equilibrium shifts toward oxidation and protein phosphorylation in response to extracellular stimuli, cytoskeletal changes also occur. Actin, an abundant and ubiquitous cellular protein, plays a major role in mediating the infrastructure and dynamics of the cytoplasmic matrix (9,10). Actin polymerization is a dynamic process implicated in growth factor-mediated cytoskeletal changes (11). Here we report both in vitro and in vivo data that indicate EGF-induced deglutathionylation of the penultimate cysteine residue of actin facilitates its polymerization and cytoskeletal changes.

EXPERIMENTAL PROCEDURES
Materials-The human epidermal A431 cell line was from ATCC; bovine muscle actin, EGF, DTNB, NADPH, GSSG reductase, carboxymethyl-BSA, CM-Sepharose, Q-Sepharose, phenyl-Sepharose, GSSG, and GSH were from Sigma; rabbit muscle actin was a generous gift from Dr. Xiong Liu; monoclonal anti-GSH antibody was obtained from Virogen; pYFP-actin vector was from CLONTECH; DCF-Ac and Phalloidin-Oregon Green 488 were from Molecular Probes; precast SDS gels and polyvinylidene difluoride membranes were from Invitrogen; Fu-gene6 was from Roche Molecular Biochemicals; H 2 O 2 was from Fisher; CdCl 2 and 2-vinyl pyridine were from Aldrich. [ 14 C]Iodoacetamide was from American Radiolabeled Chemicals.
Recombinant human glutaredoxin was expressed in Escherichia coli (12) and purified as described (13). Typical specific activity of pure enzyme is 100 units/mg in the standard assay with cysteinyl-glutathione disulfide and GSH as substrates (12).
Effect of EGF or H 2 O 2 on Protein Glutathionylation-A431 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum to 90% confluency. After washing and incubating with PBS (30 -60 min at 37°C), the cells were incubated with H 2 O 2 or EGF (at the indicated concentrations) at room temperature for 30 or 10 min, respectively. Cells were lysed with 50 l of 2ϫ SDS sample buffer without reducing agents and subjected to SDS-PAGE (12%) and Western blot analysis using anti-GSH antibody.
Generation of ROS in A431 Cells upon EGF Treatment-A431 cells were grown as described to 70% confluency. Then they were starved in Hanks' balanced salt solution (containing 2 mM Mg 2ϩ and phenol red) at 37°C for 30 min followed by addition of DCF-Ac (3 M). After 45 min, excess DCF-Ac was washed out, and the cells were incubated in Hanks' solution with or without EGF (0.17 M). After 15 min at room temperature, fluorescent images were obtained as described (2).
Analysis of Cysteine pK a -To determine cysteine pK a values, pH-dependent radiolabel incorporation into rabbit muscle actin from [ 14 C] iodoacetamide was carried out analogous to previous studies (15). Buffers (10 mM) (MES (pH 5-7), HEPPSO (pH 7-9), CAPS (pH 9 -10)) were adjusted to 0.5 M ionic strength and contained carboxymethyl-BSA (as * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Increase of F-actin in Cells Treated with EGF-A431 cells were grown to 80% confluency. After starvation for 8 h in serum-free Dulbecco's modified Eagle's medium, the cells were equilibrated in PBS at 37°C for 1 h and then incubated for 20 min more in the absence (control) or presence of EGF (2 M). Medium was removed and the cells were fixed with 3.7% formaldehyde, permeabilized with 0.1% Triton X-100, and stained with phalloidin-Oregon Green 488. After extensive washing with PBS, 1 ml of 2% SDS was added to each well to dissolve cell contents (37°C, 10 h), and absorbance at 280 nm and fluorescence (496 nm excitation and 520 nm emission) were measured. F-actin content was expressed as the ratio of fluorescence intensity to A 280 nm .
Polymerization of Actin Observed at the Single Cell Level upon EGF Treatment-A431 cells grown on a glass plate (70% confluency) were transfected with a yellow fluorescent protein fused to a ␤-actin (pYFPactin) vector using Fugene6. Upon expression of the fused protein (3ϳ4 days later), cells were subjected to starvation in PBS for 12 h prior to EGF treatment (2 M). Fluorescent images of a single cell were recorded before and after 3, 5, and 10 min incubations with EGF (see the legend for Fig. 4).
Inhibition of Actin Deglutathionylation by Cd 2ϩ -A431 cells were grown to 90% confluency. After cells were incubated with PBS alone or with CdCl 2 (2 M) for 1 h at 37°C, 1 M EGF (or no EGF for controls) was added and incubated for 20 min more. Medium was removed, and the extent of actin glutathionylation was monitored using anti-GSH as described above.
MS Analysis of Glutathione Residue Transfer-Rabbit muscle actin (ϳ100 g) was immobilized onto ϳ200-l dry volume of DNaseI-coupled resin and glutathionylated with 1 mM GSSG in PBS for 1 h at room temperature. After thorough washing with PBS, the resin was suspended in 100 l of PBS containing 0.2 nmol (2.3 g) of glutaredoxin and incubated at 37°C for 1 h before centrifugation. The supernatant was subjected to LC-MS analysis with a Finnigan LCQ (Thermoquest, San Jose, CA) mass spectrometer.

RESULTS AND DISCUSSION
Monoclonal anti-GSH antibody was used to investigate growth factor-mediated protein glutathionylation in vivo. Fig.  1A shows that H 2 O 2 treatment induced glutathionylation of a 42-kDa protein in a concentration-dependent manner in human epidermal carcinoma A431 cells. With 10 mM H 2 O 2 (lane 9), the extent of glutathionylation of the 42-kDa protein was enhanced about 10-fold relative to that observed without H 2 O 2 . Equal amounts of loaded protein were verified by Coomassie Blue staining of an equivalent SDS gel (data not shown). In contrast, treatment of the same cell line with increasing concentrations of EGF surprisingly resulted in a progressive decrease in the extent of glutathionylation of the same protein band (Fig. 1B). At 0.5 M EGF (lane 5), the glutathionylated 42-kDa protein was diminished to approximately one-third of the basal level. When EGF was raised to 2 M or higher, glutathionylation was no longer observable (lanes 7 and 8).
The unexpected decrease in glutathionylation of the 42-kDa protein prompted us to examine ROS generation in A431 cells in response to EGF. Using the membrane-permeable fluorescent ROS indicator DCF-Ac, we found that 0.17 M EGF induced a rapid ROS production as indicated by enhancement of fluorescence intensity (data not shown). This is in agreement with Bae et al. (2). Because the [GSSG]/[Total GSH] redox ratio affects the state of glutathionylated proteins, we also examined the effect of EGF-induced ROS production on the redox state of GSH (Fig. 2). For control cells, the redox ratio was 3.4 Ϯ 0.1%. When cells were treated with 1 M EGF or 10 mM H 2 O 2 for 30 min, the [GSSG]/[Total GSH] ratio increased to 6.2 Ϯ 1% or 17 Ϯ 2%, respectively. These results indicate that EGF did shift the cellular redox equilibrium toward more oxidative, so the observed deglutathionylation of the 42-kDa protein most likely is due to an active compensatory process that is required for cellular response. This unexpected dichotomy was also observed with HeLa cells.
To identify the 42-kDa protein, we used HeLa cells to obtain a larger quantity of the protein. The protein was purified to homogeneity using consecutive FPLC chromatographic separations, which included CM-Sepharose, Q-Sepharose, and phenyl-Sepharose, followed by two-dimensional electrophoresis. The purified protein was subjected to trypsin digestion and peptide sequence analysis (Michigan State University Core Facility) and was thereby identified as the cytoplasmic actin. Glutathionylation of actin, among other proteins, has been observed in other cells under varying conditions of oxidative stress, e.g. gastric mucosal cells treated with H 2 O 2 or diamide (16) and PMA-stimulated neutrophils (17). In these cases, increased S-glutathionylation of actin was observed. However, the site(s) of adduction of the GS moiety was not identified, and its physiological implications were not investigated.
Purified actin contains six cysteine residues. To investigate the relationship between actin structure/conformation and the physiological consequences of its S-glutathionylation, the stoichiometry and site(s) of glutathionylation were determined in vitro using rabbit muscle actin as the protein source because of its high purity and sequence homology. When the actin was incubated with 1 mM GSSG for 30 min, the modified actin showed a molecular mass increase of 305 Da relative to untreated actin, based on LC-MS analysis, indicating incorporation of one GSH per actin. The modified actin was then subjected to CNBr cleavage followed by mass spectrometric analysis of the peptides. The 305-Da GSH moiety was found to be associated with a peptide (WITKQEYDEAGPSIVHRKCF) encompassing the C-terminal penultimate cysteine. This result identified the penultimate cysteine (Cys-374) as the most likely glutathionylation site in vivo, consistent with findings that this cysteine appears to be the most reactive; it was preferentially labeled at physiological pH by thiol-modifying reagents (10,18,19). With 7-dimethylamino-4-methyl-(N-maleimidyl)-coumarin as the thiol reagent, Cys-257 of G-actin was found to be modified (20). This suggests that the relative reactivity of the cysteines in actin may not be determined by unusually low pK a values, but by their accessibility. Accordingly, we investigated the pH dependence of the reaction of actin with [ 14 C]IAM. The data (not shown) revealed a single inflection point corresponding to a normal pK a of 8.4, and only one carboxamido moiety was incorporated per actin molecule when iodoacetamide was maintained at a 3-fold molar excess relative to actin. However, when actin was treated with a 12-fold molar excess of IAM at pH 11, nearly six carboxamido moieties were incorporated, corresponding to the six cysteine residues per actin monomer.
Several reports have addressed the relationship between the states of thiol groups on actin and its state of polymerization (16,17,21,22); however, no clear understanding of the effect of oxidative modification can be derived from those studies. Therefore, we studied the relative rates of polymerization of the glutathionylated and non-glutathionylated forms of bovine muscle G-actin induced by 2 mM Mg 2ϩ and 100 mM KCl. Monitored by light scattering, the steady-state rate for non-glutathionylated actin polymerization was at least 5.6-fold faster than that of the glutathionylated actin (Fig. 3). These data are consistent with the finding that a small cleft exists in the actin molecule in the vicinity of Cys-374, and the size of this cleft is reduced upon polymerization (23). Accordingly, the GSH moiety at the penultimate cysteine is expected to hinder the rate of actin polymerization. Furthermore, in vitro modifications of the highly conserved C terminus of actin including cysteine S-glutathionylation and C terminus truncation of the last two residues led to destabilization of filamentous actin (24,25). The physiological effect of EGF-mediated actin deglutathionylation on F-actin formation was investigated both in batch and at the single cell level. When A431 cells, with or without 1 M EGF, were incubated for 20 min and then fixed, permeabilized, and stained with the filamentous actin indicator, phalloidin Oregon Green 488, the results revealed that the F-actin increased by 12 Ϯ 7% with EGF treatment. Our findings could provide a rationale for the report of Rijken et al. (11) that likewise showed EGF treatment of cells induces actin polymerization. For single cell experiments A431 cells were transfected with actin fused with yellow fluorescent protein, and actin polymerization was monitored by confocal fluoromicroscopy. The data show that actin was evenly distributed in the cytosol prior to EGF treatment (Fig. 4A); however, after a 10-min treatment with 1 M EGF, the light intensity increased with the majority of the light emanating from the cell periphery (Fig. 4B). This local fluorescent enhancement was observed 3 min after EGF treatment. Membrane ruffling was transiently observed and started to disappear at around 5 min. Together, these results indicate that EGF causes an increase in F-actin and its translocation toward the periphery of the cell.
The fact that EGF treatment causes deglutathionylation of actin in both A431 and HeLa cells suggests that this deglutathionylation reaction is well regulated and correlated with cell signaling processes. Glutaredoxin is most likely the enzyme for this reaction, based on its specificity and relative efficiency for protein-SSG deglutathionylation (7,8). Supporting data for this notion are derived from both in vitro and in vivo studies.
Rabbit muscle actin was immobilized on DNase I-coupled resin and treated with 1 mM GSSG in PBS to form glutathionylated actin at the penultimate cysteine. The resin was washed and resuspended in PBS buffer. This immobilized glutathionylated actin was then incubated with pure recombinant human glutaredoxin. The supernatants from both control and glutathionylated actin-treated samples were subjected to LC-ESI-MS analysis. Fig. 5A shows that the control glutaredoxin displays the expected molecular mass of 11,648 Da, whereas the glutaredoxin exposed to the immobilized glutathionylated actin displayed an increase in mass of 305 Da (Fig.  5B), consistent with transfer of a glutathione moiety from the immobilized actin to the glutaredoxin.
To address the in vivo role of glutaredoxin, we investigated the effects of Cd 2ϩ , a known glutaredoxin inhibitor that has been shown to inhibit intracellular deglutathionylation (8).
Here, A431 cells were preincubated with 2 M CdCl 2 , the IC 50 range observed in vitro, for 1 h prior to treatment with EGF. Western blot analysis (Fig. 6) revealed that Cd 2ϩ itself had little effect on actin glutathionylation prior to EGF treatment (lanes 1 and 3). However, Cd 2ϩ inhibited EGF-induced actin deglutathionylation (lanes 2 and 4). Together, these results indicate that glutaredoxin is responsible for the deglutathionylation of actin in response to growth factor stimulation in A431 cells.
Because conventional actins serve to provide a framework for cellular structure and mobility, constant movement and morphological changes of cells are achieved by a rapid rearrangement of actin filaments (11, 26). To achieve these multiple cellular functions, actin polymerization, translocation, polarity, and rate of assembly are rigorously regulated by a large number of actin-binding proteins as well as by covalent modifications that may also target the actin-binding proteins (10). Actin phosphorylation (27) and ADP-ribosylation (28) have been reported, and these events are linked to the state of polymerization, although details of the covalent modifications in the signal transduction pathways have not been fully investigated. Previously, it was reported that the EGF receptor is associated specifically with actin filaments in A431 cells (29). Our findings directly link this growth factor-mediated signaling pathway to deglutathionylation of actin at its penultimate cysteine residue, catalyzed by glutaredoxin, which leads to enhancement of the rate of actin polymerization and translocalization of the polymerized actin to the cell periphery. These results also highlight the importance of reversible glutathionylation of protein as a physiologically relevant regulatory mechanism. It is remarkable that the same EGF stimulus results in opposite changes in the intracellular state of glutathionylation of PTP1B (5) and actin (this report). In light of the likely role of cytoskeletal arrangements in transducing extracellular signals (30), regulation of the state of glutathionylation of actin and corresponding changes in its state of polymerization may modulate the localization and dynamic assembly of specific signal transduction scaffolds.  2 and 4). The medium was removed, and the cells were lysed in 2ϫ SDS sample buffer and analyzed for actin glutathionylation by Western blot with the monoclonal anti-glutathione antibody.