The Calcium-binding Protein S100A2 Interacts with p53 and Modulates Its Transcriptional Activity*

Head and neck squamous cell carcinoma express high levels of the EF-hand calcium-binding protein S100A2 in contrast to other tumorigenic tissues and cell lines where the expression of this protein is reduced. Subtractive hybridization of tumorigenic versus normal tumor-derived mammary epithelial cells has previously identified the S100A2 protein as potential tumor suppressor. The biological function of S100A2 in carcinogenesis, however, has not been elucidated to date. Here, we report for the first time that during recovery from hydroxyurea treatment, the S100A2 protein translocated from the cytoplasm to the nucleus and co-localized with the tumor suppressor p53 in two different oral carcinoma cells (FADU and SCC-25). Co-immunoprecipitation experiments and electrophoretic mobility shift assay showed that the interaction between S100A2 and p53 is Ca2+-dependent. Preliminary characterization of this interaction indicated that the region in p53 involved with binding to S100A2 is located at the C terminus of p53. Finally, luciferase-coupled transactivation assays, where a p53-reporter construct was used, indicated that interaction with S100A2 increased p53 transcriptional activity. Our data suggest that in oral cancer cells the Ca2+- and cell cycle-dependent p53-S100A2 interaction might modulate proliferation.

S100A2 is a member of the subfamily of S100 Ca 2ϩ -binding proteins, characterized by two distinct EF-hand structural motifs. It is a homodimeric protein that upon binding of calcium undergoes a conformational change (1,2). The transduction of calcium signals in that form regulates many cellular functions such as the control of cell growth and proliferation (3), transcription (4), and p53-dependent growth arrest and apoptosis (5,6). The S100A2 protein has been first detected in lung and kidney and is mainly expressed in a subset of tissues and cells such as breast epithelia and liver (6 -10). S100A2 and twenty other S100 genes are located on a cluster on human chromosome 1q21, a region frequently rearranged in human cancer (6,11). Interestingly the cDNA coding for the S100A2 protein was identified as a novel tumor suppressor gene by subtractive hybridization between normal and tumorderived human mammary epithelial cells (12). Expression studies showed that the S100A2 gene is markedly down-regulated in several tumor tissues of various origins like melanomas (13) and breast carcinoma (14). Moreover, growth factors were reported to alter the S100A2 gene expression at late G 1 /S-phase, indicating that S100A2 is cell cycle-regulated (15). Site-specific DNA methylation of the S100A2 gene promoter region in normal versus tumorigenic breast cancer cell lines indicated repression of gene expression in tumor cells, thus suggesting a role for S100A2 in suppression of tumor cell growth and possibly inhibition of tumor progression (16).
Recently, much attention has been paid to the expression of the S100A2 gene and gene product in head and neck squamous cell carcinoma (HNSCC). 1 In contrast to breast (14,16) and colon carcinoma (17), the S100A2 protein is overexpressed in a subset of HNSCC (18,19). Furthermore, S100A2 expression in HNSCC has been positively associated with squamous cell differentiation and negatively with tumor grading (20). Immunolocalization studies revealed that the protein, preferably located in the nucleus in normal tissue (21)(22)(23), becomes both nuclear and cytoplasmic in tumorigenic HNSCC tissues (24).
The transcription factor p53 is overexpressed in HNSCC (25,26), and its subcellular localization correlates with tumor stage (27) and tumor progression (28). Induction of p53 transactivation activity by DNA damage results in increased S100A2 transcription (29). Furthermore, S100B, which is present in neuronal tissues and is associated with brain tumors (8, 30 -32), and S100A4, overexpressed in metastatic breast cancer cell lines (33), were recently reported to interact with p53, and this in turn was shown to cause decreased p53 transcriptional activity. However, although interactions between p53 and S100 proteins are of particular interest, the mechanism of S100-p53regulated growth arrest, and in particular the role of S100A2 in carcinogenesis in HNSCC, has not been elucidated to date.
In an attempt to study the biological role of S100A2, we investigated the subcellular localization of S100A2 and p53 proteins in HNSCC cell lines synchronized with the DNAreplication inhibitor hydroxyurea (34). Next, we examined the calcium dependence of S100A2-p53 association in vivo using two distinct HNSCC lines as well as cells derived from breast cancer epithelia. The interaction between the two proteins was confirmed through in vitro pull-down assay using full-length and truncated p53 proteins. Moreover, the effect of this interaction on p53 transactivation was examined in a luciferasecoupled reporter assay. Our data provide the first insights into the regulation of p53 activity by S100A2.

EXPERIMENTAL PROCEDURES
DNA Constructs-Full-length cDNA of human p53 1-393 and the deletion construct p53 73-393 , both containing a Kozak consensus start, were cloned into the mammalian expression vector pcDNA3 (Invitrogen) by PCR employing a 5Ј BamHI site and a 3Ј NotI site. p21-luc (in pGL2), p53 Asp2813 Gly , and p53 1-362 (lacking the S100B binding site) were gifts from Patrick Chène, Novartis, Basel and have been described previously (35,36). The full-length S100A2 cDNA was cloned as a fusion into pGEX-3X vector (Amersham Biosciences) by PCR using a 5Ј BamHI site and 3Ј EcoRI site. All constructs were controlled by sequencing.
Cell Synchronization Assay and Cell Cycle Analysis-Exponentially growing FADU, SCC-25, and HBL100 cells were treated with 2 mM HU (Fluka, Buchs, Switzerland) for 24, 28, and 20 h, respectively, to obtain G 1 /S-phase arrest. Synchronized cells were released from the HU block and subjected to cell cycle analysis. For FACS analysis, samples were collected at the indicated time points, trypsinized, washed, and stained with propidium iodide (25 g/ml) according to the manufacturer's guidelines (Cycle Test TM Plus DNA Reagent Kit, BD Biosciences). Stained samples were analyzed in a fluorescence-activated cell sorter (FACSCalibur, BD Biosciences), and cell cycle distribution was analyzed with WinMDI software.
Immunofluorescence-FADU and SCC-25 cells were grown on glass coverslips and synchronized with HU. Cells were fixed in 2% paraformaldehyde for 15 min at room temperature, washed four times with phosphate-buffered saline, permeabilized using 0.1% Triton for 1 min, and blocked in Dulbecco's modified Eagle's medium-horse serum (1%) for 1 h at room temperature after extensive washes with phosphatebuffered saline. Slides were incubated with the following antibodies: monoclonal anti-p53-DO1 at a dilution of 1:100 and polyclonal anti-S100A2 at a dilution of 1:500 for 1 h at 37°C. Samples were washed with phosphate-buffered saline and incubated with the secondary CY2and CY5-conjugated anti-mouse and the CY3-conjugated anti-rabbit antibody (Dianova, Hamburg, Germany) both at a dilution of 1:200 as described previously (37). Nuclear stainings were performed using 4Ј,6diamidino-2-phenylindole. Control stainings were performed on untreated cells and with the secondary antibody alone. Localization of the proteins was obtained with a Leica confocal microscope (DMIRE, Wetzlar, Germany).
Western Blot-For whole cell extracts, cells were lysed in ice-cold lysis buffer (50 mM Tris buffer (pH 7.5), 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 1% Triton, 5 mM ␤-mercaptoethanol, 100 mM NaCl), and samples were clarified by centrifugation for 15 min at 14,000 rpm in an Eppendorf centrifuge. 100 g of protein extracts were resuspended in loading buffer, heated for 5 min at 95°C and loaded onto a 4 -12%-gradient SDS-PAGE gel (Invitrogen). Upon separation of proteins under denaturing conditions, proteins were transferred to Nitrocellulose membranes (Schleicher and Schuell, Dassel, Germany) and probed with the following antibodies: polyclonal anti-S100A2 at 1:2000 dilution, monoclonal anti-p53-DO1 antibodies at 1:5000 dilution for 1 h at room temperature in TBST/3% milk powder. Membranes were washed in TBST, incubated with the secondary antirabbit-or the anti-mouse-horseradish peroxidase-conjugated antibody (1:10000), respectively, for 1 h at room temperature, and exposed to ECL detection reagent (ECL, Amersham Biosciences). Protein bands were visualized on Kodak films.
Electrophoretic Mobility Shift Assay-An oligonucleotide derived from the p21/WAF promoter (33) was labeled with [␥-32 P]ATP (Amersham Biosciences). Human p53 proteins in nuclear extracts were obtained from H1299 cells after transient transfection using the calcium phosphate method. Human recombinant S100A2 protein was purified as previously described (8). Nuclear extracts were incubated in the presence of 5ϫ gel shift binding buffer (20% glycerol, 5 mM MgCl 2 , 2.5 mM EDTA, 2.5 mM dithiothreitol, 250 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.25 mg/ml poly(dI-dC)) and 0.5 and 1 g of recombinant S100A2 protein. For the S100A2 interaction with p53, the reaction mix was incubated in the presence of 2 mM calcium or 5 mM EDTA for 1-2 h at 4°C. Radiolabeled oligonucleotide p21/WAF was added, and the reaction was allowed to proceed for an additional 20 min at room temperature. The assay was terminated with 1 l of gel loading 10ϫ buffer (250 mM Tris-HCl, pH 7.5, 0.2% bromphenol blue, 40% glycerol). Samples were run on 5% non-denaturing polyacrylamide gel (CleanGel TM System 25S, Amersham Biosciences) according to the manufacturer's guidelines. The gel was dried and exposed to x-ray film at Ϫ80°C overnight.
GST-Pull-down Assay-The GST-S100A2 fusion construct was expressed from pGEX-3X and purified as previously described (38). p53 full-length and deletion mutants were cloned in pcDNA3 plasmid (Invitrogen) and used to produce 35 S-labeled proteins (Amersham Biosciences) with the TNT-coupled transcription-translation system (IVT, Promega, Madison, WI): 2 g of GST-S100A2 coupled to glutathione-Sepharose beads washed with 2 mM CaCl 2 were incubated with 4 l of the IVT reaction mix and NET-80 buffer (80 mM NaCl, 20 mM Tris) for 3 h at 4°C. Radiolabeled proteins were pulled down, washed extensively with NET-80 buffer, and separated on a 4 -12% SDS-Gel Nu-PAGE (Invitrogen). The gel was dried and exposed to x-ray film at Ϫ80°C to visualize radiolabeled proteins.
Transcriptional Activation Assay-A luciferase construct driven under the control of the p21 promoter (p21-luc) was used to study the transcriptional activity of p53. H1299 cells (6 ϫ 10 5 ) were plated in 60-mm dishes in the presence of complete medium and transfected after 24 h using the calcium phosphate method. The following constructs were used: p53 1-393 , p53 73-393 , and p53 1-362 (lacking the S100B-p53 binding site) and S100A2. After 48 h of transfection, luciferase activity was measured according to the manufacturer's guidelines (Promega). Transfection efficiency was normalized using ␤-galactosidase activity. Results are the mean of three independent experiments. Statistical significance was evaluated using the Student t test.

RESULTS
p53 and S100 Proteins Are Differentially Localized in Human HNSCC Tissue Biopsies-S100 and p53 display a characteristic nuclear and cytoplasmic staining in tumor biopsy sections ( Fig. 1) originating from the tongue (upper panel) and the hypopharynx (lower panel). Wild-type p53 predominantly accumulated in the nucleus in both tissue sections but was found in the cytoplasm of sections of the tongue (upper panel, a). S100A2 was expressed in the cytoplasm and diffusely in the nucleus in sections of the tongue (upper panel, c) as well as the hypopharynx (lower panel, c). Compared with S100A2, S100A4 (upper and lower panels, d) and S100A6 (upper and lower panels, e) were both exclusively expressed in the cytoplasm of HNSCC tissue biopsies. The staining for the S100B protein was negative (upper and lower panels, f). To further investigate the evidence on the cytosolic localization of S100A2 and p53, which are normally nuclear proteins (14,22), we examined cell lines derived from HNSCC tissues (FADU and SCC-25 cells).
Cell Cycle-dependent Localization of S100A2 in Human FADU and SCC-25 Cells-To assess whether transition through the cell cycle affects S100A2 subcellular localization, FADU and SCC-25 cells expressing both endogenous S100A2 and p53 proteins were treated with HU. HU blocks ribonucleoside diphosphate reductase and leads to rapid depletion of deoxyribonucleotide pool, thereby arresting the cells at the G 1 /S boundary. Immunofluorescence staining of untreated FADU cells showed that p53 and S100A2 were predominantly localized in the nucleus, whereas cytoplasmic staining for both proteins was evident in SCC-25 cells (Fig. 2, A and B). The merged images indicated partial co-localization of p53 and S100A2 in the nucleus in FADU but mainly cytoplasmic colocalization in SCC-25 cells. Treatment of FADU or SCC-25 with 2 mM HU for 24 or 28 h, respectively, resulted in synchronization at G 1 /S as shown by flow cytometric analysis (Fig. 3, A  and B). Under these conditions, p53 and S100A2 co-localized in the nucleus and the cytoplasm in both cell lines (Fig. 2, t ϭ 0, panels C and D). Upon HU removal (t ϭ 1), cells synchronously moved into S-phase (Fig. 3, A and B), and this was characterized by translocation of p53 to the nucleus in FADU and SCC-25 cells. At this time point, S100A2 translocated to the nucleus in both cell lines (Fig. 2, C and D). At 8 and 10 h, respectively, after release from the HU block (t ϭ 2), cells were mostly in G 2 phase (Fig. 3, A and B), and both p53 and S100A2 proteins were exclusively present in the nucleus. During transition through the next cell cycle, at 12 and 16 h after release from the HU block (t ϭ 3), p53 and S100A2 were redistributed FIG. 1. Immunohistochemical localization of S100A2 and p53 proteins in the cytoplasm and the nucleus in HNSCC tissues. Expression of S100 and p53 proteins in malignant HNSCC tissues, originating from the tongue (upper panel, a-f) and the hypopharynx (lower panel, a-f) using the following antibodies: anti-p53-1801 (a), anti-p53-DO-1 (b), anti-S100A2 (c), anti-S100A4 (d), anti-S100A6 (e), and anti-S100B (f) antibodies. The arrows indicate positive cytoplasmic, and the arrowheads positive nuclear staining of p53 and S100A2. S100A4 and S100A6 proteins are mainly in the cytoplasm. Note the negative staining of S100B (ϫ20 objective, H&E stain; scale bar, 20 m).

FIG. 2. Subcellular localization of S100A2 and p53 in HU synchronized FADU and SCC-25 cells. Localization of p53 (green) and S100A2 proteins (red) in untreated FADU cells (A) and SCC-25 cells (B).
Arrowheads indicate nuclear and cytoplasmic staining of p53 and S100A2 in both cell lines. C, synchronization of FADU cells with 2 mM HU and staining of anti-p53-DO1 (green) and anti-S100A2 (red) antibodies at the indicated time points. D, detection of p53 (blue) and S100A2 proteins (red) in SCC-25 cells, treated with 2 mM HU and stained at the corresponding time points. DNA staining was performed with 4Ј,6-diamidino-2-phenylindole. Scale bar, 40 m. to the cytoplasm where they co-localized. Finally, in FADU cells, 24 h after release from the HU block (t ϭ 4), p53 and S100A2 staining returned to the pattern displayed in untreated cells.
FACS analysis of FADU and SCC-25 cells indicated that the less differentiated SSC-25 cells had a longer cell cycle as compared with the former (Fig. 3, A and B). Taken together these data show cell cycle-dependent shuttling of p53 and S100A2 in FADU and SSC-25 cells with co-localization in the nucleus at late S/G 2 -phase.
Endogenous S100A2 and p53 Interact in FADU and HBL100 Cells in a Calcium-dependent Manner-Co-localization of p53 suggested that the two proteins may physically interact. To substantiate this finding, we performed co-immunoprecipitation experiments using total extracts of FADU cells. Precleared FADU cell lysates were immunoprecipitated with p53-DO-1 antibody in the presence of increasing calcium concentrations or 5 mM EDTA. Proteins were resolved by SDS-PAGE and detected by Western blot analysis with anti-S100A2 (Fig. 4A). The results showed that S100A2 co-immunoprecipitated with p53 in the presence of calcium concentrations higher than 0.1 mM (lanes [3][4][5]. In contrast, no interaction between p53 and S100A2 could be detected in the presence of a calcium chelator (lane 7) or at low calcium concentration (lane 2). Identical results were obtained when immunoprecipitations (IP) were performed with S100A2 antibody and p53 was detected with p53-DO-1 antibody (Fig. 4B). To rule out the possibility that the signal given by the IgG-heavy chain of the antibody used in IP could be erroneously interpreted as p53, we performed control-IP using anti-S100A2 antibody in the absence of cell extract. The results indicated that no protein band at the level of ϳ50 kDa could be detected when using antibody p53-DO-1 for Western blotting (Fig. 4, B and D, lane 1). Western blotting of S100A2 confirmed the efficiency of the antibody used for IP (Fig. 4, C and E).
To support the evidence obtained from FADU cells, we employed HBL100 cells that originate from breast cancer tissue and, like FADU cells, express wild-type p53 and S100A2 (Fig.  4D). As shown in Fig. 3C, HBL100 cells could also be synchronized with HU. Precleared HBL100 cell lysates were incubated with anti-S100A2 antibody under the same conditions used for FADU cells. Also in the case of HBL100 cells, the interaction between p53 and S100A2 was readily observable though at slightly higher calcium concentrations (lane 5) than in FADU cells. As observed in the latter, addition of 5 mM EDTA during IP completely abrogated this interaction (lane 7). Taken together, these findings suggest that the interaction between p53 and S100A2 is Ca 2ϩ -dependent and occurs in cell lines expressing the two proteins, independent of the tumor type from which the cells originate. S100A2 Binding to p53 Affects Its DNA Binding Activity-To characterize the physiological consequences of S100A2 binding to p53, we examined p53 DNA binding activity (Fig. 5). To this end, we performed electrophoretic mobility shift using nuclear extracts from H1299 cells and a labeled oligonucleotide containing the p53-consensus sequence of the p21 promoter. Because H1299 cells lack both p53 and S100A2, they were transiently transfected with the p53 1-393 construct (lane 1). Addition of 0.5 g of recombinant S100A2 protein to H1299 nuclear extracts resulted in supershift of the oligonucleotide-p53 complex in the presence (lane 4) but not in the absence of 2 mM calcium (lane 3). Addition of 5 mM EDTA to the p53-S100A2 complex completely reversed the effect of calcium (lane 5).
Furthermore, we tested DNA binding in the presence of increasing amounts of S100A2. Previous reports have shown that increasing amounts of S100B reduce the intensity of p53 complex binding to its responsive element (30, 31, 39 -41). Similarly, in the presence of 1 g of S100A2 and 2 mM calcium, the intensity of the S100A2-p53 complex was clearly reduced (lane 7). Again, calcium was essential for the induction of the supershift, and the latter could be reversed by 5 mM EDTA (lane 8). Ectopic expression of p53 proteins in transiently transfected H1299 cells was controlled by Western blot analysis (Fig.  5B). These results show that the Ca 2ϩ -dependent interaction between p53 and S100A2 facilitates binding of p53 to its responsive element. FIG. 4. S100A2 physically interacts with p53 in a Ca 2؉ -dependent manner. A, FADU cell extracts subjected to immunoprecipitation (IP) with monoclonal anti-p53-DO-1 to p53 in the presence of calcium (0.1, 0.5, 1, and 2 mM) or a calcium chelator (5 mM EDTA). Western blotting was performed with polyclonal anti-S100A2 antibody as described under "Experimental Procedures." B, IP of FADU cell extract with a polyclonal anti-S100A2 under conditions described in A, followed by detection of p53 with monoclonal anti-p53-DO-1. C, analysis of the extent of S100A2 precipitated from FADU cell extracts. D and E, IP of HBL100 extracts with polyclonal anti-S100A2 antibody, followed by detection of p53 and S100A2 as indicated in B and C. Control-IP with p53-DO-1 to p53 or anti-S100A2 antibody in the absence of cell extracts are shown in lane 1 of each panel. Anti-p53-DO-1 were used at a dilution of 1:5,000, anti-S100A2 at a dilution of 1:2,000, and the corresponding secondary-horseradish peroxidase-coupled antibodies at a dilution of 1:10,000. S100A2 Interacts with Wild-type and the C Terminus of p53-We next examined the domains required for the interaction of S100A2 with p53. We performed in vitro GST-pull down assay (Fig. 6) using in vitro translated and 35 S-labeled p53 1-393 , p53 73-393 , and p53  . Recombinant GST-S100A2 proteins were immobilized on glutathione-Sepharose beads, incubated with the in vitro translated p53 proteins, and the eluted complexes were separated on SDS-PAGE. Under these conditions, full-length p53 protein was pulled down in the presence of 2 mM calcium, whereas the interaction was almost completely abolished in the presence of 5 mM EDTA (Fig. 6,  upper panel lanes 3 and 4).
Deletion of the p53 transactivation domain (p53 73-393 ) did not affect the calcium-dependent interaction between the Cterminal domain of p53 and recombinant S100A2 (Fig. 6, middle panel, lane 3). In contrast, deletion of the C-terminal domain, where the S100B-binding site was mapped (residues 367-388) (1), resulted in complete loss of the binding to GST-S100A2 (Fig. 6, lower panel, lane 3). These data showed that the p53 region necessary for binding to S100A2 is located in the C terminus of p53. S100A2 Modulates Transcriptional Activity of p53-To test the functional consequences of the S100A2-p53 interaction, we performed transient transfections in H1299 cells, which are deficient in p53 and S100A2. Transfection of the p53 wild-type expression vector resulted in a ϳ19.5-fold activation of transcription of the p21 reporter gene construct (Fig. 7A, lane 3), whereas transfection of S100A2 had no effect on the luciferase reporter (lane 2). Co-transfection of the p53 expression vector with S100A2 resulted in a further ϳ7.2-fold increase of p53-dependent transcriptional activation. Titration experiments showed that maximal p53 activation was obtained at 100 ng of S100A2 DNA (lane 4) with a reduction of transcriptional activity to almost basal level at 500 ng of S100A2 (lane 5). Furthermore, we employed a p53 construct containing a mutation in the N-terminal region (p53 Asp2813 Gly ) that caused an overall reduction of activity while displaying little induction of the p21-reporter construct (ϳ1.3-fold, Fig. 7B, lane 3). Co-transfection with S100A2 resulted in the same extent of increase in transactivation observed with wild-type p53. The effect was maximal at 100 ng of S100A2 (ϳ3.5-fold) with a less pronounced induction at 500 ng of S100A2 (ϳ1 .3-fold, lanes 4 and 5). These results indicated that S100A2 was able to activate p53 transcriptional activity in a statistically significant manner.
To further verify that the far C-terminal domain of p53 (residues 362-393) was necessary for the observed induction of p53-dependent transcription by S100A2, we transiently transfected H1299 cells with the p53 1-362 deletion construct (Fig. 7C). p53 1-362 displayed an overall lower transactivation activity as compared with the full-length protein but was nevertheless able to activate the reporter construct (ϳ4.2-fold, lane 3). Interestingly, the transcriptional activation of p53 was slightly reduced at 100 ng of S100A2 (ϳ0.5-fold, lane 4) as well as 500 ng of S100A2 DNA (ϳ0.6-fold, lane 5). These results indicate that S100A2 is able to activate the transcriptional activity of wild-type p53 but lack of the oligomerization domain of p53 abrogates this function. The p values in the corresponding experiments indicate that there is a statistical significance of the transcriptional activation of wild-type and mutant p53 by S100A2.

DISCUSSION
The S100A2 protein belongs to the calcium binding, EF-hand protein family of S100 proteins and has attracted interest due to the fact that it is localized to the nucleus (22), it is downregulated in various tumor tissues (24,42,43), and it is presumably a potential tumor suppressor gene (12,16). Site-specific methylation of the S100A2 gene promoter region and transcriptional activation of the S100A2 promoter by p53 (29) further support the hypothesis on involvement of this protein in carcinogenesis. The tumor suppressor p53 is a key regulator of the cell cycle and triggers apoptosis in response to stress (44). Stability and cellular localization of p53 is regulated by interaction with HDM2 (45), and its transactivation activity is modulated by cell type and tissue-specific cofactors (41).
Based on our previous finding that the S100A2 is localized to the nucleus (22), we investigated possible physical and functional interactions with p53 in HNSCC tissues and cell lines (FADU and SCC-25) that express both p53 (25,46,47) and S100A2 (18,48). We found that the S100A2 protein was localized in the cytoplasm with focal nuclear staining in HNSCC primary tumor tissues, whereas S100A6 and S100A4 were exclusively in the cytoplasm (Fig. 1). On the contrary, S100B was undetectable in tissue staining. p53 was present in nuclei as well as in the cytoplasm. This pattern of p53 and S100A2 staining was reflected in studies performed on cell lines derived from HNSCC tumors. Predominantly nuclear co-localization of p53 and S100A2 was found in FADU, whereas SCC-25 cells displayed both cytosolic and nuclear staining. Several studies have shown that wild-type p53 accumulates in the cytoplasm where it is rapidly degraded in an HDM-2-dependent manner. Cytoplasmic sequestration of p53 has been described in a variety of primary tumor tissues such as neuroblastoma (49), colon cancer (50,51), and presumably melanoma (52). Changes of the subcellular localization of p53 in these tumors have been associated with inhibition of the p53 tumor suppressive function and, as a consequence, with poor prognosis (51). Overexpression of p53 is an early event in the carcinogenesis of HNSCC and correlates with increased recurrence, poor prognosis, and poor survival rate (53,54). Members of the S100 protein family have been described to enhance nuclear accumulation of p53 (32). Moreover, a recent report showed disruption of the p53 tetramerization equilibrium by S100A4 and S100B that lead to accumulation of p53 in the nucleus (55). To examine whether p53 and S100A2 localization and interaction was cell cycle-dependent, we considered cells synchronized with HU. This drug is widely used for cell synchronization, and the reversibility of its effect makes it particularly suitable to use in cell culture. Moreover, HU is used in the treatment of thrombocythemia (56). Similar to HU, fluorouracil applied in chemotherapy of colon and HNSCC inhibits DNA synthesis by thymidine starvation resulting in growth arrest and cytotoxicity (57). We found that during transition through the S-and G 2 -phases that follow synchronization with HU, p53 and S100A2 translocated to the nucleus and co-localized. After transition trough M-phase, the two proteins were redistributed to the nuclear and cytosolic compartments, as observed in non-synchronized cells. The mechanism of the p53-S100A2 complex shuttling between the cytoplasm and the nucleus remains to be determined.
We further investigated the physical interaction of endogenous S100A2 and p53 proteins in FADU and HBL 100 cells (Fig. 4) in the presence or the absence of calcium. Titration experiments indicated that p53 and S100A2 proteins interacted only at calcium concentrations higher than 0.5 mM in FADU and higher than 2 mM in HBL100 cells. The property of S100A2 to bind calcium with lower affinity as compared with other S100 family members may explain the results obtained. The conditions employed in our assay closely reflected the elevated intracellular calcium concentrations caused by deregulated calcium homeostasis that has been reported to occur in responsive luciferase reporter containing binding sites for p53 and the p53 1-393 wild-type construct (A), mutated p53 Asp281-Ͼ Gly (B), and truncated p53 1-362 (C) concomitantly with increasing amounts of full-length S100A2. A, lane 1, 1.5 g of p21-luc only; lane, 1.5 g p21-luc with 100 ng of S100A2 only; lane 3, 1.5 g of p21-luc with 100 ng of p53 1-393 ; lane 4, 1.5 g of p21-luc with100 ng of p53 p53 1-393 and 100 ng of S100A2; lane 5, 1.5 g of p21-luc with 100 ng of p53 1-393 and 500 ng of S100A2. p value, 3-4, 0.029. B, lane 1, 1.5 g of p21-luc only; lane 2, 1.5 g of p21-luc with 100 ng of S100A2 only; lane 3, 1.5 g of p21-luc with100 ng p53 Asp281-Ͼ Gly ; lane 4, 1.5 g of p21-luc with 100 ng p53 Asp2813 Gly and 100 ng of S100A2; lane 5, 1.5 g of p21-luc with 100 ng p53 Asp2813 Gly and 500 ng of S100A2. p value, *3-4, 0.0065. C, lane 1, 1.5 g of p21-luc only; lane 2, 1.5 g of p21-luc, 100 ng of S100A2 only; lane 3, 1.5 g of p21-luc with100 ng p53 1-362 ; lane 4, 1.5 g of p21-luc with 100 ng p53 1-362 and 100 ng of S100A2; lane 5, 1.5 g of p21-luc with 100 ng p53 1-362 and 500 ng of S100A2. p value, 3-4, 0.0073. Results are representative of three independent experiments and normalized as described under "Experimental Procedures. "   FIG. 7. Luciferase reporter assay to measure the transcriptional activity of wild-type, mutated, and truncated p53 in the presence of S100A2. H2199 cells were transfected with the p21-human tumors (58). Moreover, interaction between S100B and p53 has also been studied using high calcium concentrations (1-2 mM) (30,31,39,59,60). Furthermore, our data showed that the Ca 2ϩ -dependent interaction of S100A2 with p53 involved the C-terminal domain of p53 (Fig. 6). These regions encompasses the tetramerization domain and contains the S100B binding site (residue 367-388) (1,31,61). Given the high sequence similarity between the S100 family members, it is likely that S100A2 interacts at the same binding site as S100B binding. Finally, our studies indicate that S100A2 activates the transcriptional activity of full-length p53 (Fig. 7). A p53 mutant at the residue that is a common hot spot for mutation in numerous tumors of the colon and breast (62), and that causes reduction of p53 transactivation activity responded similarly to wild-type p53 when co-transfected with S100A2. Furthermore, our finding that transactivation of a reporter construct by the deletion mutant p53 1-362 was reduced in cotransfection experiments with S100A2 (Fig. 7C) supports our conclusion of a potential S100A2 binding motif in the C terminus of p53. S100A2 protein expression is high in tumors of the oral cavity but not in breast and neuroblastoma. Prognosis of S100A2-positive HNSSC than of negative tumors is generally better. S100A2 has been used as a tumor marker (19). The interaction of S100A2 and p53 that we report in this study is of particular interest, because, contrary to other S100 family members, S100A2 activates p53 transcriptional activity, presumably helping to restore p53 function in growth arrest and apoptosis. The recent report of an interaction between S100A2 and p63 (63), a p53 family member, substantiates the importance of our findings.