Up-regulation, Modification, and Translocation of S100A6 Induced by Exposure to Ionizing Radiation Revealed by Proteomics Profiling*S

The cellular response to genotoxic stress is a complex cascade of events including altered protein expression, interactions, modifications, and relocalization, leading to cell cycle arrest and DNA repair or to apoptosis. p53 protein has a central role in this process, and p53 status is an important factor in the response of a tumor to genotoxic anticancer therapy. We studied p53-related changes postexposure to ionizing radiation using top-down mass spectrometry. Initially two cell lines were compared, HCT116 p53 wild type (wt) and p53−/−, in a time course study postirradiation. In the p53 wt cell line a striking increase of a 10.2-kDa protein was detected, and this protein was identified with MS/MS analysis as S100A6. Further MS profiling led to detection of two post-translationally modified variants of S100A6, namely glutathionylated and cysteinylated forms. In p53 wt cells, a specific shift from glutathionylated to cysteinylated S100A6 occurred postirradiation. The p53 dependence of this specific change in protein level and modification pattern of S100A6 postirradiation was confirmed in a panel of four lung cancer cell lines (H23, U1810, H69, and A549) with different p53 status and using small interfering RNA against p53. Interestingly the closely related S100 family protein S100A4 showed the same changes in modification pattern post-ionizing radiation in the p53 wt lung cancer cell line, and S100A4 also showed p53-dependent expression. Using confocal microscopy, relocalization of S100A6 from nucleus to cytosol and a colocalization with tropomyosin in stress fibers was detected in A549 cells postirradiation. This relocalization coincided with the change in S100A6 modification pattern. Based on these results, we suggest that S100A6 and S100A4 are regulated via redox modifications in vivo and that these proteins are involved in the cellular response to genotoxic stress.

The cellular response to genotoxic stress is complex, and great effort is put into determining the network of events taking place in the cell to maintain genomic stability and prevent tumorigenesis. The tumor suppressor protein p53 is one of the key components in the DNA damage signaling cascade (for a review, see Ref. 1). Following activation by DNA damage surveillance proteins, p53 has the ability to direct cells with damaged DNA toward cell cycle arrest to allow DNA repair or in cases with severe damage toward apoptosis. p53 is also one of the most commonly mutated proteins in human cancers (2). The frequent mutations of p53 in cancer and the central position of p53 in the DNA damage signaling network directed our interest to studying p53-dependent signaling in response to genotoxic stress.
Exposure of cells to ionizing radiation (IR) 1 results in DNA damage, generation of reactive oxygen species, and activation of the p53 signaling cascade. The effective killing of cells by IR is used in the clinic, and today 40 -50% of cancer patients receive radiation therapy. Experimentally IR exposure of cells is used as a model system to study the DNA damage signaling cascade. Omics methods are well suited for studying these complex signaling networks, and efforts have been made to elucidate p53-dependent signaling post-treatment of cells with radiation and drugs but to date have been limited to studies of global mRNA expression (3,4).
We have earlier studied p53-related differences in response to anticancer therapy, focusing on differences in drug and radiosensitivity and DNA repair (5). In the present work, we used a proteomics approach to find new p53-related IRresponsive proteins in colon (HCT116 p53 wt and p53 Ϫ/Ϫ cell lines) and lung cancer cell lines with different p53 status (A549, U1810, H23, and H69). Protein profiling using SELDI-TOF-MS in a time course study revealed a p53-dependent increase of a 10.2-kDa protein postirradiation that was identified as S100A6. Closer analysis indicated three forms of S100A6, namely glutathionylated, cysteinylated, and unmodified S100A6. Additionally the same change in post-translational modification pattern postexposure to IR was detected for the closely related protein S100A4. These findings directed our studies to further characterize this modification pattern in vitro using mass spectrometry profiling and to study the subcellular localization of S100A6 postexposure to IR.

EXPERIMENTAL PROCEDURES
Cell Culturing, Irradiation, and Protein Extraction-Colon cancer cell lines HCT116 wt and HCT116 p53 Ϫ/Ϫ cells (6) were cultured in McCoy's 5A medium with L-glutamine, 10% calf serum, and 1% penicillin/streptomycin. Lung cancer cell lines U1810, H69, A549, and H23 were cultured in Dulbecco's modified Eagle's medium with 10% calf serum and 1% penicillin/streptomycin (Invitrogen). Twenty-four hours after seeding, monolayer cultures of cells were exposed to ␥-radiation (12 Gy for HCT116 or 8 Gy for U1810, H69, A549, and H23) using a 60 Co source at a dose rate of 1.5 Gy/min. Cells were irradiated at room temperature and harvested at the indicated time points postirradiation. Cells were then lysed using repeated freezethaw cycles, and soluble proteins were extracted in a pH 7.5 lysis buffer including 0.1% Triton X-100, 1% CHAPS, and Complete Mini EDTA-free protease inhibitors (Roche Diagnostics). To measure response to oxidative stress, cells were treated with the indicated concentrations of H 2 O 2 (Sigma-Aldrich) 24 h after seeding and harvested 48 h post-treatment. Soluble proteins were extracted as above.
SELDI-TOF-MS-based Protein Profiling-Cell extracts were profiled using a ProteinChip SELDI-TOF-MS system, PBSIIC (Bio-Rad). Each sample was analyzed on four separate ProteinChip arrays with different types of chemical surfaces, including strong anionic exchange (Q10, pH 7.5), cation exchange (CM10, pH 7.0), metal (copper) ion binding (IMAC-Cu), and reversed-phase (H50, 15% acetonitrile). Standard protocols recommended by the manufacturer were used for all experiments. After incubation, washing, and application of matrix solution (sinapinic acid), the ProteinChip arrays were analyzed, and spectra were collected in the 0 -150-kDa range using the SELDI-TOF Protein Biology System IIC. The data analysis was performed using the CE software package (Bio-Rad). Each spot was analyzed twice with separate settings of laser intensity, detector sensitivity, and time lag focusing to allow optimal detection of spectral regions between 3 and 10 kDa and 10 and 35 kDa, respectively. All spectra were base line-corrected (using a peak width value of 8), and peak detection was performed using a signal/noise value of 4 and valley depth value of 2 (first pass). Representative SELDI-TOF-MS data from one of three independent experiments is shown.
Protein Purification-Isolation of S100A6 was performed using an Á KTA purifier chromatography system (GE Healthcare). Cell extract from HCT116 p53 wt cells 72 h postexposure to ionizing radiation was fractionated according to size with a gel filtration column (Superdex 200 10/300 GL, GE Healthcare) using 50 mM ammonium acetate, pH 6.0, as running buffer. The protein of interest was located using screening of collected fractions with SELDI-TOF-MS NP20 Protein-Chip arrays (normal phase). The fraction containing the protein was loaded into a reversed-phase column (SOURCE 15RPC ST 4.6/100, GE Healthcare) using 20 mM ammonium acetate, pH 5.5, and a linear gradient from 5 to 90% acetonitrile as running buffer. Again fractions were scanned for the protein of interest using SELDI-TOF-MS and NP20 ProteinChip arrays. The fraction containing the 10.2-kDa protein was sufficiently pure for protein identification.
Protein Identification-Reversed-phase chromatography fractions containing the protein of interest were digested with trypsin (Promega, Madison, WI) according to standard protocols and analyzed using LC-MS/MS on a CapLC-Q-TOF Ultima atmospheric pressure ionization system (Waters Corp.). Peak list files were generated, and database searching was performed using the manufacturer-supplied Pro-teinLynx Global Server 2 software (release PLGS2.2.5). The Swiss-Prot database was searched using human taxonomy (Swiss-Prot release 50.7, containing 14,636 human protein sequences) for pro-teins of molecular mass between 5 and 200 kDa and theoretical pI values between 0 and 14. Specifically search parameters used were: enzyme specificity, trypsin; precursor ion tolerance, 0.1 Da; fragment ion tolerance, 0.1 Da; fixed modification, carbamidomethylation of Cys; and variable modifications, deamidation of Asn/Gln and oxidation of Met. An initial search using default parameters for processing raw data (peak list generation) returned no significant hits. To improve the sensitivity of analysis, more time-consuming preprocessing parameters were used for peak list generation (specifically each spectrum was processed using the MaxEnt3 algorithm of the software). This search resulted in a single significant protein identification, the S100A6 protein (AC: P06703/GI: 116509). The protein was identified with two peptides, LMEDLDR and LQDAEIAR, giving a sequence coverage of 16.7%. The two peptides were sequenced from doubly charged precursors of m/z 446.25 and 458.30 and had "ladder scores" of 71.8 and 73.3, respectively (the two MS/MS spectra are shown in Supplemental Fig. 1). Due to the somewhat ambiguous nature of the "ladder score" value, the identification was confirmed by performing the same search using Mascot. In short, the MassLynx 4.0 software was used to generate MaxEnt3-processed spectra that were exported as Sequest (.dta) files and submitted to Mascot MS/MS ion search on line (Matrix Science). The Mascot search resulted in ion scores of 35 and 48 for LMEDLDR and LQDAEIAR, respectively (scores Ͼ31 indicate significance, i.e. p Ͻ 0.05). The total (average) sequence mass (minus N-terminal Met) of S100A6 is 10,048.6 Da. Allowing for one N-acetylation, (ϩ42 Da) this mass (10,090.6 Da) is in close agreement with the observed mass of the protein of interest (ϳ10,090 Da) from SELDI experiments (see Fig. 3A). The identification was further validated using Western blot analysis and immunocapture with anti-S100A6 antibody (see below).
Western Blot Analysis-Total cell extracts (50 g) were resolved by a 10% SDS-PAGE (NuPAGE Bis-Tris, Invitrogen) and blotted onto a nitrocellulose membrane. After blocking with nonfat milk, the membranes were probed with the following antibodies against S100A6 (Sigma-Aldrich), p53 and p21 (Santa Cruz Biotechnology Inc.), phospho-p53 (Ser-15) (Cell Signaling Technology Inc.), and S100A4 (Lab Vision) followed by appropriate horseradish peroxidase-conjugated secondary antibodies (GE Healthcare). Anti-tubulin antibody (Lab Vision) was used as loading control and showed equal total protein loading in all Western blot experiments.
Immunocapture Experiments-Anti-S100A6 antibody (Sigma-Aldrich) coupled to an RS100 ProteinChip array (BioRad) was used to capture S100A6 protein from HCT116 cell extract according to the standard protocol recommended by the manufacturer. RS100 Pro-teinChip arrays were analyzed, and spectra were collected in the 0 -150-kDa range using SELDI-TOF Protein Biology System IIC. Seize primary mammalian immunoprecipitation kit (Pierce) was used for extraction of S100A4 from A549 cell extract. Anti-S100A4 antibody (Lab Vision) was primarily subjected to protein G purification to exclude BSA. Anti-p21 antibody (Santa Cruz Biotechnology Inc.) was used as a negative control. The protocol for immunoprecipitation was then followed, and the eluate was analyzed using SELDI-TOF-MS and NP20 ProteinChip arrays (normal phase).
Small Interfering RNA-For suppression of p53 ON-TARGETplus SMARTpool TP53 was purchased (Dharmacon). The protocol for siRNA treatment of A549 was followed. In short, 24 h after seeding cells were treated with siRNA against p53 or with siCONTROL Non Targeting (Dharmacon) and with DharmaFECT 1 as transfection reagent. Twenty-four hours after siRNA treatment cells were irradiated as described above, and 48 h postexposure to IR cells were harvested and lysed.
Fluorescence Cytochemistry-A549 cells were grown on Falcon culture slides (BD Biosciences). Twenty-four hours after seeding cells were irradiated as described above. At the indicated time points postirradiation, cells were fixed in 4% phosphate-buffered formalin solution. Cells were then washed using PBS, permeabilized using 0.5% Triton X-100, and blocked in 3% BSA. After incubation with primary antibody (anti-S100A6, Sigma-Aldrich) overnight at 4°C the cells were washed in PBS and incubated with secondary antibody (Alexa Fluor 488 goat anti-mouse IgG, Invitrogen) at room temperature for 30 min. The cells were then washed in PBS and counterstained in Vectashield hard set mounting medium with 4Ј,6-diamidino-2-phenylindole (Vector Laboratories). Fluorescence images were obtained using an LSM 510 META spectral laser scanning microscope (Zeiss, Gö ttingen, Germany). For colocalization experiments cells were irradiated and treated as described above with primary antibodies against S100A6 (GenWay) and tropomyosin (TM311, Sigma-Aldrich) followed by secondary antibodies (FITC goat anti-IgY (GenWay) and Alexa Fluor 568 goat anti-mouse IgG (Invitrogen)). Fluorescence images were obtained using a Leica DMRXA fluorescence microscope (Leica, Wetzlar, Germany).
In Vitro Modification Study-All in vitro modification experiments were performed using total cell extract (6 mg/ml protein concentration) in 50 mM HEPES, pH 7.5. Addition of Cys, GSH, or DTT (Sigma-Aldrich) was done to reach the indicated final concentrations, and reaction tubes were incubated for 2 h at 4°C. The modification pattern of S100A6 was then analyzed using SELDI-TOF-MS and Q10 ProteinChip arrays at pH 7.5.

Screening of p53-related Changes in Proteomic Pattern
Postirradiation-Four SELDI ProteinChip array surfaces were used to monitor changes in the proteomic profile of HCT116 wt and HCT116 p53 Ϫ/Ϫ cells postirradiation. Using CM10, IMAC-Cu, H50, and Q10 ProteinChip arrays we were able to detect between 400 and 500 protein peaks per sample in the 3-30-kDa mass area (Fig. 1). As a measurement of peak intensity variability we calculated the average cv of peak intensities in peaks with signal/noise Ͼ5 in the 12-20-kDa mass area in mass spectra generated on anion exchange surface (Q10). In HCT116 wt cells (control ϩ six time points postirradiation) the average cv was 20.6%, and in p53 Ϫ/Ϫ cells (control ϩ six time points postirradiation) the average cv was 16.6%, indicating low general variability in the profiling experiment. Few overall changes in protein expression pattern were seen postirradiation, and in the low molecular weight proteome of HCT116 the most striking change was a time-dependent increase of a 10.2-kDa peak appearing in p53 wild type cells postirradiation in Q10 mass spectra ( Fig. 2A). In the same peak cluster we could also detect a decrease of a 10.4-kDa peak in p53 wt cells. The 10.2-kDa peak could be detected in p53 Ϫ/Ϫ cells as well but with considerably lower peak intensity ( Fig. 2A). All experiments were performed in biological triplicates showing the same changes in protein expression pattern.
Isolation and Identification of S100A6 Protein-To identify found proteins, the cell lysate was first fractionated using liquid chromatography. Size exclusion chromatography was used followed by reversed-phase chromatography. Fractions were scanned using SELDI-TOF-MS NP20 ProteinChip arrays to locate the protein of interest. The fraction containing our target protein was freeze-dried and digested with trypsin. The tryptic digest was then analyzed using LC-MS/MS, and peptide fragments were sequenced. Two peptides were sequenced matching the S100A6 protein sequence. Up-regulation of S100A6 in HCT116 could be confirmed using a monoclonal anti-S100A6 antibody in Western blot analysis (Fig. 2B). During protein purification of the 10.2-kDa protein, two other proteins were co-purified (10.1 and 10.4 kDa) suggesting proteins with similar biochemical characteristics. However, only S100A6 peptides could be sequenced from this fraction.
Post-translational Modifications of S100A6 Protein-The 10.2-kDa peak seen in the SELDI-TOF analysis was part of a cluster containing three major peaks. The mass difference between peak 1 (10,090 Da) and peaks 2 and 3 was 119.5 and 306.1 Da, respectively (Fig. 3A). As these three peaks cofractionated after chromatography and only gave rise to S100A6 peptides after trypsinization we postulated that the three peeks were in fact one protein with different posttranslational modifications. A database search (Delta Mass, Association of Biomolecular Resource Facilities) for modifications suggested cysteine (ϩ119 Da) and glutathione (ϩ305 Da) as possible modifications (Fig. 3A). The S100A6 protein contains a single cysteine residue (position 3 from the N terminus) that permits the suggested modifications. To further investigate the S100A6 modifications we performed a series of in vitro experiments using crude HCT116 wt p53 cell lysate (Fig. 3B). By addition of 1 mM cysteine to HCT116 cell lysate the major S100A6 peak was shifted to the mass of the cysteinylated form. Addition of 1 mM GSH shifted the major peak to the mass of glutathionylated S100A6. Addition of equal concentrations of cysteine and GSH resulted in two major peaks corresponding to cysteinylated and glutathionylated S100A6. Increasing the GSH concentration to 10 times the cysteine concentration shifted the peak to the mass of glutathionylated S100A6. Addition of DTT (reducing agent) to the

FIG. 3. S100A6 is modified by glutathionylation and cysteinylation in HCT116 cells.
A, SELDI-TOF-MS spectrum from HCT116 p53 wt cells 24 h postirradiation showing annotated peaks in the 10-kDa peak cluster. Peak mass differences suggest modification of S100A6 by cysteinylation and glutathionylation. B, densitometric presentation of mass spectra showing in vitro characterization of S100A6 modifications. By addition of 1 mM cysteine to HCT116 cell lysate the major S100A6 peak is shifted to the mass of the cysteinylated form. Addition of 1 mM GSH shifts the major peak to the mass of glutathionylated S100A6. Addition of equal concentrations of cysteine and GSH results in two major peaks corresponding to cysteinylated and glutathionylated S100A6. Increasing the GSH concentration to 10 times the cysteine concentration shifts the peak to the mass of glutathionylated S100A6. Addition of DTT (reducing agent) to the lysate removes S100A6 redox modifications. lysate removed S100A6 modifications. All three forms of S100A6 could be captured using an anti-S100A6 antibody coupled to an RS100 ProteinChip array confirming that the three peaks corresponded to S100A6 with different modifications (data not shown). S100A6 Regulation and p53 Relation in a Lung Cancer Cell Line Panel-To validate the S100A6 and p53 relation we chose to study the S100A6 expression pattern in an additional set of cell lines with differences in p53 status. Four lung cancer cell lines were used for analysis of S100A6 expression and modification postexposure to ionizing radiation: A549 (wt p53 (7)), H23 (mut. p53 (8)), U1810 (mut. p53 (9)), and H69 (mut. p53 (10)). Western blot data showed strongest S100A6 expression in A549 cells as well as an increase in S100A6 protein expression postirradiation (Fig. 4A). Functional p53 status postirradiation in A549 and HCT116 wt cell lines was confirmed by p53 phosphorylation and downstream protein p21 up-regulation (A549, Fig. 4A; HCT116, Ref. 5). Components of the S100A6 cluster were detected in all lung cancer cell lines using Q10 protein arrays (anion exchange) with the strongest signal seen in A549 (p53 wt) (Fig. 4B). The modifi-cation pattern of S100A6 in A549 cells postirradiation (Fig. 4B) paralleled the pattern detected in HCT116 wt cells ( Fig. 2A) with an increase in the cysteinylated S100A6 and a decrease in glutathionylated S100A6. S100A4 Expression and Post-translational Modification in a Lung Cancer Cell Line Panel-The distinct modification pattern of S100A6 protein postexposure of cells to ionizing radiation raised the question of whether this pattern could be found in other proteins as well. A similar pattern could be found in A549 cells in the sub-12-kDa mass area (Fig. 4B). A database search on S100 proteins suggested S100A4 as a candidate based on protein molecular weight and N-terminal sequence homology to S100A6. Western blot with antibody against S100A4 confirmed strong S100A4 expression in A549 cells (Fig. 5A), and immunoprecipitation followed by SELDI-TOF-MS confirmed the identity of the sub-12-kDa peaks as S100A4 (Fig. 5B). S100A4 could not be detected in the three p53 mutated lung cancer cell lines (H23, U1810, and H69; Fig. 5A).
Reduced Expression of S100A6 and S100A4 Post-IR in Cells Treated with siRNA against p53-To confirm the p53  4. p53-dependent expression and modification of S100A6 in lung cancer cell lines. A, Western blot showing S100A6 protein level in four lung cancer cell lines untreated (c) and postexposure to ionizing radiation (24 and 48 h). A549, the only cell line in the panel with wt p53, shows the highest expression of S100A6 protein in untreated cells and an increase in S100A6 protein level postirradiation. p53 expression level and phosphorylation and p21 expression in A549 cells postirradiation confirmed the expression of functional p53. B, densitometric presentation of mass spectral data showing S100A6 modification pattern in lung cancer cell lines postexposure to ionizing radiation. In p53 wt cells (A549) the modification of S100A6 is shifted from glutathionylation to cysteinylation postirradiation (left panel). The same pattern was seen for a protein with a mass of 11.6 kDa in A549 cells (right panel). This protein was later shown to be S100A4 (see Fig. 5). relation in S100A6 and S100A4 expression we treated A549 cells with siRNA directed against p53. Western blot data showed lower S100A6 and S100A4 protein expression postirradiation in cells treated with siRNA against p53 compared with cells treated with non-targeting control siRNA (Fig. 6). Using antibodies against p53 and the well established p53regulated protein p21 we could also show efficiency in the siRNA-mediated suppression of p53.
p53-dependent Up-regulation of S100A6 in Response to Oxidative Stress-To study the role of oxidative stress in S100A6 up-regulation, we treated HCT116 wt and p53 Ϫ/Ϫ cells with H 2 O 2 . Interestingly we detected a p53-dependent up-regulation of S100A6 in response to treatment with 1000 M H 2 O 2 for 48 h (Fig. 7). In HCT116 p53 wt cells we detected strong S100A4 expression in control cells, but rather than an increase we detected a slight decrease after treatment of cells with 1000 M H 2 O 2 for 48 h. In p53 Ϫ/Ϫ HCT116 cells we were not able to detect S100A4, supporting the p53 dependence in S100A4 expression.
Subcellular Localization of S100A6 -Fluorescence immunocytochemistry was used to determine subcellular localization of S100A6 in A549 cells postexposure to IR. In untreated cells S100A6 protein showed nuclear localization. Interestingly we could detect a relocalization of S100A6 24 h postexposure to ionizing radiation after which the protein was mainly cytoplasmic. This pattern was even more pronounced 48 h postirradiation (Fig. 8).
Colocalization of S100A6 and Tropomyosin in Stress Fibers-One of the suggested cytoplasmic interaction partners of S100A6 is tropomyosin. Because we detected a translocation of S100A6 to the cytoplasm in response to irradiation we wanted to study possible colocalization with tropomyosin. Using fluorescence cytochemistry and antibodies against S100A6 and tropomyosin we detected colocalization of these two proteins in A549 cells treated with ionizing radiation (Fig.  9). Tropomyosin staining showed typical stress fibers, and S100A6 localized in these fibers. DISCUSSION We used a proteomics approach to further elucidate the complex cellular signaling induced by genotoxic stress. Studying the low molecular weight proteome using SELDI-TOF-MS we performed a time course study to identify p53dependent changes in protein pattern postirradiation.
We report here an increase in S100A6 protein level in colon and lung cancer cells postexposure to ionizing radiation. S100A6 is a member of the S100 protein family of small   5. p53-dependent expression of S100A4 in lung cancer cell lines. A, Western blot confirming higher S100A4 protein expression in p53 wt lung cancer cells (A549). S100A4 protein could not be detected in the three cell lines with mutated p53 (H23, U1810, and H69). B, immunoprecipitation (IP) of S100A4 from A549 cell lysate followed by SELDI-TOF-MS of precipitate confirming the S100A4 identity suggested in Fig. 4B. Anti-p21 antibody (Ab) was used as a negative control. c, control.  6. siRNA against p53 inhibits up-regulation of S100A6 and S100A4 postirradiation. A Western blot showing higher expression of S100A6 and S100A4 postirradiation in A549 cells treated with non-targeting control siRNA compared with A549 cells treated with siRNA against p53 is shown.
Tubulin p53 S100A6 S100A4 FIG. 7. p53-dependent up-regulation of S100A6 by oxidative stress. A Western blot showing a dose-dependent up-regulation of S100A6 in response to treatment with H 2 O 2 in HCT116 p53 wt cells is shown. This up-regulation was not seen in p53-deficient cells. We could not detect up-regulation of S100A4 after treatment with H 2 O 2 ; however, HCT116 also cells show clear p53-dependent S100A4 expression.
Ca 2ϩ -binding proteins associated with a variety of cellular and extracellular processes (for recent reviews, see Refs. [11][12][13]. S100A6 is reported to be up-regulated in several diseases especially in cancer. So far the protein level of S100A6 has been studied using immunohistochemistry and shown to be up-regulated in colorectal adenocarcinoma (14,15), malignant melanoma (16), pancreatic cancer (17,18), breast cancer (19), thyroid carcinoma (20,21), and osteosarcoma (22). However, the clinical significance of this observed up-regulation is controversial because in some cases high expression is reported to correlate with increased cell proliferation and poor clinical outcome (pancreatic cancer and melanoma), whereas in osteosarcoma high S100A6 has been shown to be associated with decreased metastasis (22). The biological functions of S100A6 remains obscure, but some conclusions can be drawn from studying its interacting proteins. The S100A6 interaction with Sgt1, involved in ubiquitin-mediated protein degradation of the known oncogene ␤-catenin, suggests S100A6 involvement in regulation of cell proliferation (23,24). S100A6 also seems to play a key part in other cellular processes such as tropomyosin-related cytoskeleton rearrangements as indicated by interaction and antisense studies (25,26). The closely related S100A4 protein is known to affect cell migration and is correlated to metastasis of cancer (for reviews, see Refs. 27 and 28). S100A4 has also been shown to interact with p53 and disrupt its tetramerization and thereby regulate its function (29).
To our knowledge this is the first time S100A6 or any of the other S100 family proteins have been shown to be up-regu-lated in response to cytotoxic treatment. We also showed a p53 dependence in the protein expression level of S100A6 and the closely related protein S100A4. p53 is not a known S100A6 or S100A4 transcription factor; nevertheless a 25-fold difference in S100A4 mRNA expression between untreated HCT116 wt and HCT116 p53 Ϫ/Ϫ has been reported previously using cDNA microarray and confirmed by Western blot (30). Interestingly a reverse correlation between S100A4 expression and positive p53 staining, indicating mutated p53, was shown by immunohistochemistry in clinical lung adenocarcinoma samples (31) supporting our results.
Using mass spectrometry-based protein profiling we demonstrated for the first time in vivo three different forms of S100A6 and S100A4: non-modified, cysteinylated, and glutathionylated. Using a time course experimental setup, we also showed a distinct change in the pattern of these modifications postexposure of cells to ionizing radiation. To our knowledge the shift from glutathionylation to cysteinylation that we showed in this study is previously unknown. This effect was seen in two different p53 wild type cell lines expressing S100A6 (HCT116 and A549) as well as in two different proteins from the same protein family (S100A6 and S100A4). None of the other 400 -500 proteins and peptides detected in this proteomics experiment showed the same modification pattern, suggesting a specific, well regulated modification.
Protein S-thiolation, i.e. the covalent and reversible addition of glutathione or cysteine to a cysteine residue in a protein, has lately emerged as a redox-sensitive mechanism to regulate protein function (for reviews, see Refs. [32][33][34]. Glutathionylation is a fairly well characterized protein modification, and to date close to 100 proteins have been demonstrated as potential targets of glutathionylation. So far very few proteins have been described as targets of cysteinylation. There are only a handful of publications showing in vivo protein Sthiolation or its role in regulation of protein function, most likely due to difficulties in measuring protein S-thiolation in vivo. In the sample preparation protocol we used for low molecular weight proteome profiling, we did not include reduction and alkylation. This allowed us to detect redox-dependent post-translational modifications such as glutathionylation or cysteinylation. The wide use of reducing and alkylating agents in proteomics protocols can explain why so few redox-dependent modifications have been described so far. In vitro characterization of the modifications showed that it was possible to push the equilibrium toward either the cysteinylated or glutathionylated form of S100A6 simply by adding substrate (free cysteine or glutathione). The reversibility of the covalent modification in these experiments points toward enzymatic regulation of the modification. Glutaredoxin has been suggested as the key enzyme responsible for dethiolation of protein S-thiolated proteins (35). Whether glutaredoxin is involved in dethiolation of S100A6 remains to be proven.
Ionizing irradiation induces generation of reactive oxygen  8. S100A6 is translocated postirradiation in A549 cells. Confocal microscopy images showing localization of S100A6 protein in untreated and irradiated A549 cells using immunofluorescence are shown. 4Ј,6-Diamidino-2-phenylindole (DAPI) staining of DNA shows nuclei, and the right panel shows an overlay of the two images. In untreated cells S100A6 is localized in the nucleus with some staining in the cytoplasm. Postexposure to ionizing radiation S100A6 is translocated to the cytoplasm. species (ROS) and cellular oxidative stress (36), and we suggest that this change in intracellular redox balance could be responsible for the alterations in S100A6 protein level and modifications that we detected. wt p53 has the ability to enhance the production of ROS and thereby potentiate the oxidative stress signal (37). This p53-potentiated ROS formation could explain the p53-dependent differences in S100A6 protein level and modification pattern we report postirradiation. Recently the presence of an antioxidant response element in the promoter region of the S100A6 gene was shown (38). We could confirm up-regulation of S100A6 protein in response to treatment of cells with H 2 O 2 , and we also report a p53 dependence in this up-regulation. Carcinoma cells are frequently under persistent oxidative stress (39), and this could be one of the reasons S100A6 is found to be upregulated in many cancer diseases. However, p53 can be activated via a variety of different stimuli, and further studies are needed to fully elucidate the connections between p53 activation and S100A6 up-regulation.
In addition to up-regulation and modification, we also report S100A6 translocation in response to irradiation. In untreated p53 wt lung cancer cell line A549, S100A6 localized mainly to the nucleus. Postexposure to ionizing radiation S100A6 translocated to the cytoplasm. The subcellular localization of S100A6 and S100A4 has earlier been shown to be Ca 2ϩ -dependent (40,41). Elevation of the intracellular Ca 2ϩ level leads to a translocation of the protein from the nucleus/perinuclear membrane toward the cytoplasm/plasma membrane. One effect of cellular exposure to ionizing radiation is a p53-dependent BIK activation resulting in endoplasmic reticulum Ca 2ϩ release into the cytoplasm (42), and this Ca 2ϩ release could potentially explain the S100A6 translocation we saw postirradiation. The S100 family of proteins are Ca 2ϩ -binding proteins, and the conformation of the proteins is strongly dependent on this binding (43). The translocation of S100A6 postirradiation could be an effect of a Ca 2ϩ -dependent conformational change leading to altered affinity of S100A6 to interacting proteins. The cytoplasmic localization of S100A6 postirradiation directed our interest to tropomyosin, which is a cytoplasmic protein with a reported Ca 2ϩ -dependent interaction with S100A6 (25). It has also been reported that expression of antisense S100A6 resulted in a reduction and disorganization of tropomyosin-associated cytoskeleton filament networks without down-regulation of tropomyosin and a parallel reduction in F-actin stress fibers (26). However, no actual colocalization of S100A6 and tropomyosin was seen. We showed, for the first time, colocalization of S100A6 and tropomyosin. This colocalization was a consequence of IR-induced translocation of S100A6 to the cytoplasm. Tropomyosins bind to and stabilize actin filaments, and these so-called stress fibers constitute an important part of the cytoskeleton. It has been reported that overexpression of tropomyosin 3 leads to formation of stress fibers and inhibition of cell motility and that treatment of cells with siRNA against tropomyosin inhibits stress fiber formation in response to transforming growth factor-␤ (44). Antisense experiments also show that reducing the protein level of tropomyosin 1 results in anchorage-independent growth (45). It seems reasonable to believe that actin-tropomyosin stress fibers are important in the regulation of cellular processes linked to metastasis, and our data reported here suggest that S100A6 interacts with and possibly stabilizes these structures. In osteosarcoma increased expression of S100A6 has been associated with decreased metastasis, and overexpression of S100A6 in osteosarcoma cell lines led to inhibition of cell migration and anchorage-independent growth (22). Our data present a possible mechanism for these findings through stabilization of tropomyosin-actin stress fibers.
Interestingly it has been shown in vitro that the Ca 2ϩ affinity of recombinant S100A1, another S100 family protein, is increased by up to 4 orders of magnitude by glutathionylation or cysteinylation of its single Cys residue (46). Thus it seems possible that the modification of S100A6 Cys-3 that is reported in the present study is important for S100A6 affinity to Ca 2ϩ and that protein S-thiolation could be a way to potentiate the effect of endoplasmic reticulum Ca 2ϩ release on S100A6.
In conclusion, the novel up-regulation, modification, and translocation of S100A6 postexposure to ionizing radiation shown here suggest a role for S100A6 in the cellular stress response. Furthermore the colocalization of S100A6 with tropomyosin stress fibers indicates that S100A6 is involved in regulation of the cytoskeleton in response to stress. Involvement of S100A6, S100A4, and possibly other S100 proteins in stress response pathways could potentially explain the upregulation of S100 proteins in the numerous and diverse pathological conditions reported in the literature. Modification of cysteine residues in S100 proteins is likely to change the affinity for Ca 2ϩ and thereby protein conformation and target protein interaction. We suggest that protein S-thiolation con-S100A6 Tropomyosin Overlay+DAPI A549 8Gy 48h FIG. 9. Colocalization of S100A6 and tropomyosin in A549 cells postirradiation. Fluorescence microscopy images showing colocalization of S100A6 and tropomyosin in stress fibers postexposure to ionizing irradiation in A549 cells are shown. DAPI, 4Ј,6-diamidino-2-phenylindole.
stitutes an important regulatory mechanism for the S100 protein family. We also hypothesize that S100 proteins could act as linkers between Ca 2ϩ and redox signaling pathways and that there exists a cross-talk through S100 Ca 2ϩ binding and protein S-thiolation that ultimately guides S100 proteins to specific protein interactions and subcellular localizations.
Acknowledgments-We are grateful to Dr. Salvador Rodriguez-Nieto and Birgitta Mö rk for experimental assistance with confocal microscopy.
* The work was supported by grants from the Stockholm County Council, Swedish Cancer Foundation, and Cancer Society in Stockholm. 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.