The Zinc- and Calcium-binding S100B Interacts and Co-localizes with IQGAP1 during Dynamic Rearrangement of Cell Membranes*

The Zn2+- and Ca2+-binding S100B protein is implicated in multiple intracellular and extracellular regulatory events. In glial cells, a relationship exists between cytoplasmic S100B accumulation and cell morphological changes. We have identified the IQGAP1 protein as the major cytoplasmic S100B target protein in different rat and human glial cell lines in the presence of Zn2+ and Ca2+. Zn2+ binding to S100B is sufficient to promote interaction with IQGAP1. IQ motifs on IQGAP1 represent the minimal interaction sites for S100B. We also provide evidence that, in human astrocytoma cell lines, S100B co-localizes with IQGAP1 at the polarized leading edge and areas of membrane ruffling and that both proteins relocate in a Ca2+-dependent manner within newly formed vesicle-like structures. Our data identify IQGAP1 as a potential target protein of S100B during processes of dynamic rearrangement of cell membrane morphology. They also reveal an additional cellular function for IQGAP1 associated with Zn2+/Ca2+-dependent relocation of S100B.

S100B is a member of the S100 family of proteins containing two EF-hand-type calcium-binding domains (1). This protein interacts not only with Ca 2ϩ but also with Zn 2ϩ ions, binding Zn 2ϩ ions with an affinity in the nanomolar range (2). The capacity of S100B to bind and release Zn 2ϩ suggests that Zn 2ϩ may not only play a structural role but might also be involved, together with Ca 2ϩ , in concerted regulation of S100B function. The S100B protein is naturally highly expressed in the vertebrate nervous system, where it is present in astrocytes and Schwann cells (3). In the adult central nervous system, the S100B protein is present in the nuclei and cytoplasm of astrocytes and accumulates in the astrocytic dendrites in the perivascular processes (4). Studies in different laboratories suggest a variety of intracellular regulations by S100B, including negative cell growth regulation (5), cell structure (6), and calcium homeostasis (7). The S100B protein is also secreted from astrocytes and has extracellular functions (8). Extracellular S100B acts as a modulator of neuronal synaptic plasticity (9). Although nanomolar quantities have beneficial neurotrophic effects on nerve cells, high levels of this protein have been implicated in glia activation and could contribute to the devel-opment of brain pathology as observed in Down's syndrome and Alzheimer's disease (10). The recent observation that S100B triggers activation of the pro-inflammatory cell surface receptor receptor for advanced glycation end products has shed more light on its extracellular function (11). In cultured human astrocytoma U87 cells, S100B secretion is dependent on relocation of S100B toward vesicle-like structures at the periphery of the cells and is regulated by Ca 2ϩ and Zn 2ϩ (12). S100B can also be secreted into the bloodstream and cerebrospinal fluid and is a biochemical marker of brain damage or dysfunction in acute and chronic diseases (13,14). A relationship between S100B accumulation in the astrocytic end-feet and morphological changes of astrocytes in the perivascular regions has been reported previously (15). These changes may be related to the release of S100B into the blood stream (15). Consistent with dynamic regulation of astrocyte cell shape by S100B, antisense inhibition of S100B production in cultured rat glial C6 cells is correlated with alterations in cellular morphology (6). The mechanisms of regulation of astrocyte cell morphology by S100B and its secretion pathway remain unclear. By analogy with other EF-hand Ca 2ϩ -binding proteins, such as calmodulin, one might suppose that the biological activity of S100B is related to Ca 2ϩ /Zn 2ϩ -dependent interaction with target proteins. In this study, we identify IQGAP1 protein as the first S100B target protein identified to date whose interaction with S100B is regulated by Zn 2ϩ and Ca 2ϩ . IQGAP1 is also the major specific cytoplasmic S100B target protein present in both rat glial C6 and human U373 or U87 astrocytoma cell lines. We also provide evidence that cytoplasmic S100B specifically binds to a sub-population of IQGAP1 molecules that localize at the polarized leading edge and areas of membrane ruffling and that both S100B and IQGAP1 proteins are relocated in a Ca 2ϩdependent manner within vesicle-like structures. The interaction of S100B with IQGAP1 may have important implications for understanding the roles played by S100B in processes of dynamic rearrangement of cell membranes and in the mechanisms of Zn 2ϩ /Ca 2ϩ -dependent relocation and secretion of S100B.

MATERIALS AND METHODS
Cell Cultures and 35 SMet/Cys Labeling-Human astrocytoma U-373MG, U-87MG cells, rat glioma C6 cells, mammary carcinoma MCF7 cells, and NIH 3T3 fibroblasts were cultured in Dulbecco's modified Eagle's medium with Glutamax (Invitrogen) supplemented with 10% fetal calf serum (Invitrogen). Cells were labeled in methionine-free minimal essential medium, 5% fetal calf serum supplemented with 35 SMet/Cys mix (50 Ci/ml) for 6 h. Transfection Experiments-U-373MG, U-87MG, NIH 3T3, and MCF7 cells were transfected with the pcDNA-Neo containing the wildtype or C-terminal deleted S100B cDNA (17) using FuGENE TM 6 reagent transfection according to manufacturer's protocol. For stably transfected S100B-MCF7 cell lines, cells were incubated, 48 h after * This work was supported by grants from the Association pour la Recherche sur le Cancer and la Ligue Nationale Contre le Cancer. 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. transfection, in complete medium supplemented with 500 g ml Ϫ1 neomycin (G418), and neomycin-resistant S100B-MCF7 clones were selected. S100-and CaM-Sepharose Beads-S100B and CaM 1 were purified from bovine brain to homogeneity (16). S100B-and CaM-Sepharose were prepared by reaction of bovine brain S100B and CaM with CnBr-Sepharose in 20 mM HEPES, pH 7.8, 0.5 mM CaCl 2 . Both S100B-and CaM-Sepharose contained 2 mg of protein per milliliter of beads. The purification protocols for human S100A1, S100A6, S100B, and S100A11 recombinant proteins have been described previously (17). Recombinant human S100B, S100A1, S100A6, and S100A11 were coupled to CnBr-Sepharose (1 mg of protein per milliliter of beads) as described above.
Western Blot Analysis-S100B protein was resolved with SDS-Tris-Tricine-11%-PAGE (17). ␤-tubulin, ␤-catenin, and IQGAP1 were analyzed using Laemmli SDS-PAGE. The proteins were transferred to nitrocellulose membrane and incubated with the primary antibodies. Proteins were visualized using an ECL kit (PerkinElmer Life Sciences).
Mass Spectrometric Analysis and Protein Identification-The 170-kDa protein that binds to S100B-Sepharose in the absence of calcium was excised from Coomassie Blue-stained gels and washed with 50% acetonitrile. Gel pieces were dried in a vacuum centrifuge and rehydrated in 20 l of 25 mM NH 4 HCO 3 containing 0.5 g of trypsin (Promega, sequencing grade). After 4-h incubation at 37°C, a 0.5-l aliquot was removed for MALDI-TOF analysis and spotted onto the MALDI sample probe on top of a dried 0.5-l mixture of 4:3 saturated ␣-cyano-4-hydroxy-trans-cinnamic acid in acetone/10 mg/ml nitrocellulose in acetone/isopropanol 1:1. Samples were rinsed by placing a 5-l volume of 0.1% trifluoroacetic acid on the matrix surface after the analyte solution had dried completely. After 2 min, the liquid was blown off by pressurized air. MALDI mass spectra of peptide mixtures were obtained using a Bruker Biflex mass spectrometer (Bruker-Franzen Analityk, Bremen, Germany). Internal calibration was applied to each spectrum using trypsin autodigestion peptides (MH ϩ 842.50, MH ϩ 1045.55, MH ϩ 2211.11). Protein identification was confirmed by tandem mass spectrometry experiments. After in-gel tryptic digestion, the gel pieces were extracted with 5% formic acid solution and then with acetonitrile. The extracts were combined with the original digest, and the sample was evaporated to dryness in a vacuum centrifuge. The residues were dissolved in 0.1% formic acid and desalted using a Zip Tip (Millipore). Elution of the peptides was performed with 5-10 l of 50% acetonitrile/0.1% formic acid solution. The peptide solution was introduced into a glass capillary (Protana) for nanoelectrospray ionization. Tandem mass spectrometry experiments were carried out on a quadruple time-of-flight hybrid mass spectrometer (Micromass, Altrincham, UK) to obtain sequence information. Collision-induced dissociation of selected precursor ions was performed using argon as the collision gas and collision energies of 40 -60 eV. Protein identification was achieved using both MALDI peptide mass fingerprints and MS/MS sequence information. Mass spectrometric data were compared with known sequences using the programs MS-Fit and MS-Edman located at the University of California San Francisco (available at prospector.ucsf.edu/). Tandem mass spectrometry sequencing of three different peptides (LGLAPQIQDLYGK, LEGVLAEVAQHYQDTLIR, and FPD-AGEDELLK) confirmed the strict identity of the 170-kDa protein with IQGAP1. These peptides are not found in other human protein sequences, including IQGAP2 and the putative IQGAP3 sequence.
Cell Extracts-For binding assays and co-immunoprecipitation experiments, cells were lysed at 4°C in TTBS buffer (40 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.3% Triton X-100) plus protease inhibitors (leupeptin, aprotinin, pepstatin, and 4-(2-aminoethyl)benzenesulfonylfluoride hydrochloride, 10 g/ml each) and centrifuged for 10 min. Cell lysates were precleared by incubation for 10 min with 50 l of protein A-Sepharose. S100B and CaM Binding Assays-500-l aliquots of precleared supernatant were supplemented with either 5 mM EDTA/5 mM EGTA, or with 20 M ZnSO 4 , or with 0.3 mM CaCl 2 /10 M ZnSO 4 and mixed with 20 l of affinity beads equilibrated in the same buffers. After mixing at 4°C for 15 min, the beads were spun down and the supernatant was removed. The beads were washed three times with 1 ml of binding buffers. At the last wash, beads were transferred to new Eppendorf tubes and boiled in SDS-sample buffer. For binding assays using recombinant IQGAP1 or IQGAP1 domains, 20 l of reaction lysates were diluted in 500 l of TTBS and processed as described for cell extracts.
Co-immunoprecipitation Analysis-500-l aliquots of pre-cleared supernatant were mixed with 40 M GTP␥S (Sigma) plus 5 mM MgCl 2 , if needed, and with either 5 mM EDTA/5 mM EGTA or 0.3 mM CaCl 2 /10 M ZnSO 4 and mixed with the appropriate antibodies (5 g) plus 20 l of protein A-or protein G-Sepharose equilibrated in the same buffers. After mixing at 4°C for 50 min, the beads were centrifuged briefly (5 s, 13,000 rpm), and the supernatant was removed. The beads were washed three times with 1 ml of appropriate incubation buffers. At the last wash, beads were transferred to new microcentrifuge tubes and boiled in SDS-sample buffer containing 5 mM EGTA and 5 mM EDTA.
Immunofluorescence Analysis-cells grown on permanox slides from Nunc, Inc. were fixed for 30 min with 4% paraformaldehyde in HBS (10 mM HEPES, pH 7.4, 130 mM NaCl, 15 mM KCl, 5 mM MgCl 2 ) and permeabilized for 5 min with TBS (30 mM Tris-HCl, pH 7.4, 150 mM NaCl) containing 0.2% Triton X-100 plus 1 mM CaCl 2 . After washing with TBS/1 mM CaCl 2 , cells were incubated for 30 min in TBS/1 mM CaCl 2 containing 10% normal goat serum and then incubated with primary antibodies for 90 min in the same buffer containing 5% normal goat serum. The cells were then washed with TBS/1 mM CaCl 2 and incubated for 1 h with the appropriate secondary antibodies conjugated with cyanin3 (Jackson ImmunoResearch Laboratories) or with Al-exa488 (Molecular Probes, Inc., Eugene, OR) in the same buffer as described for the primary antibodies. After washing with TBS/1 mM CaCl 2 , cells were incubated in a solution of Hoechst 33258 (2 g/ml) for 5 min and placed under coverslips. Preparations were analyzed with a Zeiss fluorescence microscope (Axiovert 200M) or a Leica confocal microscope (TCS-SP2).

S100B Overexpression Correlates with Changes in U373 Cell
Morphology-In the transformed human astrocytoma U373 cell line, the expression of S100B is up-regulated in post-confluent cells (Fig. 1). Up-regulation of S100B is specific, because it is not observed with ␤-tubulin, calmodulin (CaM), and IQGAP1.
Although all cells are uniformly immunostained with S100A6 antibodies (Fig. 2B), considerable variation in S100B immunostaining characterized post-confluent U373 cells. In post-confluent culture, the strongest S100B-positive cells had grown on top of the cell layer. These cells are characterized by intense cytoplasmic S100B immunoreactivity and have adopted a less flattened morphology with long processes. Confocal microscopy analysis of S100B immunostaining in confluent and post-confluent U373 cells confirmed a relationship between S100B overexpression and change in cell shape (Fig. 2, C and D). In these experiments, cells were double-labeled with S100B polyclonal antibodies (red) and IQGAP1 monoclonal antibody (green). U373 cells that enter confluence have a flattened morphology. In these cells, the weak S100B immunoreactivity is mostly nuclear and IQGAP1 accumulates at the cell periphery ( Fig. 2C). In post-confluent culture, cells characterized by intense cytoplasmic S100B immunoreactivity have adopted a less flattened morphology with long processes (Fig. 2D). Overlapping of the S100B and IQGAP stainings (white pixels) reveals that some of the S100B colocalizes with IQGAP1 at the cytoplasmic membrane and within processes. The correlation between cytoplasmic S100B overexpression with changes in cell shape is consistent with previous studies that showed that selective inhibition of S100B production by antisense strategies in rat glioma C6 cells resulted in a more flattened cellular morphology (6). In rat C6 glioma cells, S100B is also up-regulated in post-confluent cells, and its up-regulation correlates with drastic cell morphological changes (data not shown).
IQGAP1 Is the Major Specific S100B Binding Protein in the Glial C6 and U373 Cells-In an attempt to identify specific S100B target proteins that could mediate the effect of S100B on cell morphology, we compared proteins in astrocytoma U373 and glial C6 cell extracts that bind to S100B-Sepharose beads. A major S100B-binding protein that migrated with an apparent molecular mass of 170 kDa was identified in both cell lines (Fig. 3A). The 170-kDa protein binds to S100B-Sepharose beads in EGTA/EDTA-and Ca 2ϩ /Zn 2ϩ -containing buffer. The 170-kDa human protein from U373 MG cells was further characterized by mass spectrometry. Protein identification was achieved using both MALDI peptide mass fingerprints and MS/MS sequence information (see "Materials and Methods"). Results revealed that it corresponds to human IQGAP1. IQ-GAP1 is a widely expressed protein that acts as a scaffold in recruiting and maintaining the organization of cytoskeletal proteins at the plasma membrane (19 -27). The other high molecular weight Ca 2ϩ /Zn 2ϩ -dependent S100B-binding protein present in glial C6 cells, but not in U373 cell extract, has been previously identified as AHNAK (28). The binding of IQGAP1 to S100B is specific, because it is not observed with S100A6 ( Fig. 3B, lanes 5 and 6), and S100A11 (Fig. 3C, lanes 3 and 7), two other S100 species expressed in U373 cells (17). S100A1, the closest S100B homologue that is not expressed in U373 cells (17), also binds IQGAP1 (Fig. 3B, lanes 3 and 4). IQGAP1 in U373 cell extract also binds to calmodulin-Sepharose beads (Fig. 3C, lanes 4 and 8).
IQGAP1 Co-immunoprecipitates with S100B from U373 Cell Extract-A physical interaction between S100B with IQGAP1 was confirmed by co-immunoprecipitation of a S100B/IQGAP1 complex from confluent U373 cell extract (Fig. 4). In a first set of experiments, S100B was immunoprecipitated with S16 monoclonal S100B antibody that recognizes an epitope located within the N terminus of S100B (18). The presence of IQGAP1 in the S100B immunoprecipitate was revealed with anti-IQ-GAP1 polyclonal antibodies. A small but detectable amount of IQGAP1 is found in the S100B immunoprecipitates in EDTA/ EGTA buffer (Fig. 4A, lane 4). The amount of IQGAP1 immunoprecipitated with S100B monoclonal antibody increased substantially in buffer containing Ca 2ϩ /Zn 2ϩ (Fig. 4A, lane 5). The co-immunoprecipitation of IQGAP1 with S100B is specific, because it is not observed with control anti-MyoD antibodies (Fig.  4A, lanes 2 and 3). The Ca 2ϩ /Zn 2ϩ requirement for the interaction between soluble S100B and IQGAP1 contrasts with the apparent divalent ion-independent interaction observed with S100B cross-linked to Sepharose beads (see below).
In a second set of experiments, IQGAP1 was immunoprecipitated with anti-IQGAP1 AF4 monoclonal antibody (Fig. 4B). S100B is found associated with IQGAP1 immunoprecipitates when using Ca 2ϩ /Zn 2ϩ -containing buffer. Cdc42 and ␤-catenin, two other IQGAP1 target proteins (22)(23)(24) are also found associated with IQGAP1 immunoprecipitates in both EGTA/EDTA and Ca 2ϩ /Zn 2ϩ buffer. Several laboratories have also shown intracellular interactions between CaM and IQGAP1 (19 -20, 26, 29). In Fig. 4C, we compared the association of calmodulin (CaM) and S100B with immunoprecipitate IQGAP1 out of exponentially growing and post-confluent U373 cells in EGTA/ EDTA-and Ca 2ϩ /Zn 2ϩ -containing buffer. CaM and S100B that co-immunoprecipitated with IQGAP1 were sequentially revealed using the same nitrocellulose transfer membrane. CaM co-immunoprecipitates with IQGAP1 in all conditions tested, whereas S100B only co-immunoprecipitates with IQGAP1 from post-confluent cell extracts in buffer containing Ca 2ϩ / Zn 2ϩ (compare lanes 2 and 3). It is noteworthy that, although other laboratories reported that Ca 2ϩ enhances the interaction between CaM and IQGAP1 (26, 29 -30), we found more CaM immunoreactivity associated with IQGAP1 in U373 cell extracts containing EGTA and EDTA (compare lanes 2 and 3 or lanes 5 and 6). This unexpected observation cannot solely be explained by a competition with S100B, because it is also observed with sub-confluent culture characterized by low S100B expression.
Zn 2ϩ -dependent Interaction between S100B and IQ-GAP1-In a pull-down assay using S100B cross-linked onto Sepharose beads, the S100B/IQGAP1 interaction can be detected independently of the presence of EGTA/EDTA or Ca 2ϩ / Zn 2ϩ in binding buffer (Fig. 3). In contrast, co-immunoprecipitation experiments with endogenous cellular proteins revealed that IQGAP1/S100B interaction is markedly strengthened when Ca 2ϩ and Zn 2ϩ are included in binding buffer (Fig. 4). One factor that might explain this apparent discrepancy is the high S100B protein concentration used in the pull-down assay (1.5-3 M) compared with the soluble S100B in cell extracts. To evaluate the effect of S100B concentration on complex formation with IQGAP1, we performed co-immunoprecipitation analysis using MCF7 cells extracts (which do not express the S100B protein) supplemented with increasing concentrations of recombinant human S100B (Fig. 5A). Results confirm that the S100B/IQGAP1 interaction is regulated by divalent ions and that high concentrations of S100B are not sufficient to promote ion-independent interactions.
Previous studies from our laboratory have shown that chemical modifications within the S100B molecule may have profound effects on the protein quaternary and tertiary structures (2,31). We thus investigated the possibility that cross-linking of S100B onto Sepharose beads modifies S100B conformation as to favor interaction with IQGAP1. To test this, we compared the interaction of recombinant IQGAP1 produced in rabbit reticulocytes with the equal amount of soluble S100B, by means of co-immunoprecipitation, and S100B cross-linked onto Sepharose beads by pull-down assays. As shown in Fig. 5B, S100B cross-linked onto Sepharose beads, but not soluble S100B, interacts with in vitro translated IQGAP1 in EGTA/ EDTA-containing buffer (compare lanes 3 and 7). We next evaluated the contribution of individual divalent ions, Zn 2ϩ and Ca 2ϩ , to the S100B/IQGAP1 interaction. In the co-immunoprecipitation and pull-down assays, addition of Zn 2ϩ (10 M) to binding buffer stimulates interaction between S100B and IQGAP1 (lanes 4 and 8). Further addition of Ca 2ϩ repeatedly enhanced that interaction (lanes 5 and 9). Quantitative evaluation of the radioactivity associated with S100B-Sepharose beads in different buffer conditions is shown in Fig. 5C. Together, these data suggest that, when cross-linked onto Sepharose beads, S100B adopts a conformation that favors its interaction with IQGAP1. This conformational state is probably very similar to that induced upon Zn 2ϩ binding. They also suggest that Ca 2ϩ might also strengthen the S100B/IQGAP1 interaction.
Zn 2ϩ -dependent interaction of S100B with IQGAP1 was also observed with cellular proteins (Fig. 5D). NIH-3T3 cells were  1-2), S100A1-Sepharose (lanes 3-4), S100A6-Sepharose (lanes 5-6), and S100B-Sepharose (lanes 7-8) in the absence (lanes 1, 3, 5, and 7) or in the presence of calcium (lanes 2, 4, 6, and 8). C, comparison of [ 35 S]methionine-labeled proteins in U373 MG cell extracts that interact with control Sepharose beads (lanes 1 and 5), S100B-Sepharose (lanes 2 and 6), S100A11-Sepharose (lanes 3 and 7), and CaM-Sepharose (lanes 4 and 8) in the absence (lanes 1-4) or in the presence of calcium (lanes 5-8). In A-C, the bound proteins were resolved on 5% SDS-PAGE. In C, the bound proteins were resolved on 5% SDS-PAGE and autoradiographed. In panel C, proteins with low molecular weight were left to run out of the gel. Values at left indicate molecular size in kDa of protein standards. Arrowheads indicate the position of IQGAP1 and of the giant protein AHNAK. transfected with S100B expression plasmid and S100B/IQ-GAP1 complex formation analyzed by co-immunoprecipitation with S16 monoclonal antibody in different buffer conditions. When transfected cells are lysed in binding buffer containing 20 M EGTA, the S100B/IQGAP1 interaction is almost undetectable (lane 1). If cell extract containing 20 M EGTA is supplemented with Zn 2ϩ (40 M), binding of IQGAP1 to S100B is rescued (lane 2). Addition of Ca 2ϩ (300 M) or Zn 2ϩ plus Ca 2ϩ to EGTA containing cell extract also rescues S100B/IQGAP1 interaction (lanes 3 and 4). As observed with in vitro translated 35 S-labeled IQGAP1 (Fig. 5, B and C), a slight but significant increase in IQGAP1 immunoreactivity is found associated with S100B immunoprecipitates in buffer containing Ca 2ϩ . That stimulation was more clearly seen with lower exposure of the Western blot membrane to ECL film. All together these data suggest that, in solution, the S100B/IQGAP1 interaction requires either Zn 2ϩ or Ca 2ϩ ions and that Zn 2ϩ is sufficient to promote that interaction. IQGAP1 is thus the first S100B target protein identified whose interaction with S100B is mediated by Zn 2ϩ -dependent conformational change on S100B.
Mechanism of Zn 2ϩ -dependent Interaction of S100B with IQGAP1-To further confirm the essential role of Zn 2ϩ -dependent conformational change on S100B for interaction with IQGAP1, we next compared the mechanism of interaction of S100B with IQGAP1 and with a strict calcium-dependent target protein, AHNAK (28). We first studied the interactions of the wild-type S100B and of a C-terminal deleted mutant S100B (S100B⌬Ct) with IQGAP1 and AHNAK. We used S100B⌬Ct because the C terminus domain of S100B is required for interactions between S100B and strict Ca 2ϩ -dependent target protein (32,33). NIH 3T3 cells were transfected with expression vectors encoding S100B or S100B⌬Ct, and complex formation was assayed by co-immunoprecipitation using the N-terminal S100B monoclonal antibody S16 (Fig. 6A). Although we repeatedly observed a much lower expression of S100B⌬Ct compared with wild-type S100B, both wild-type S100B and mutant S100B⌬Ct co-immunoprecipitate with IQGAP1 from cell ly-sates in Zn 2ϩ /Ca 2ϩ -containing buffer (lanes 6 and 9). Deletion of the C terminus of S100B specifically abrogated Zn 2ϩ /Ca 2ϩdependent interaction of S100B with AHNAK (compare lanes 6 and 9). These results suggest that the C terminus of S100B is not implicated in Zn 2ϩ -dependent interaction of S100B with IQGAP1. We next compared the contribution of individual divalent ions, Zn 2ϩ and Ca 2ϩ , to the S100B⌬Ct/IQGAP1 interaction (Fig. 6B). When transfected NIH 3T3 cells are lysed in binding buffer containing 20 M EGTA, the S100B⌬Ct/IQGAP1 interaction is almost undetectable (lane 2). If cell extract containing 20 M EGTA is supplemented with Zn 2ϩ (40 M) (lane 3) or Zn 2ϩ plus Ca 2ϩ (lane 5), binding of IQGAP1 to S100B⌬Ct is rescued. However, in contrast to the full-length S100B (Fig.  5D, lane 3), addition of Ca 2ϩ (300 M) alone also stimulates S100B⌬Ct/IQGAP1 interaction, but to a much lower extent than Zn 2ϩ or Zn 2ϩ plus Ca 2ϩ (Fig. 6B, lane 4). All together these data confirm that, in solution, Zn 2ϩ is sufficient to promote the S100B/IQGAP1 interaction and that Ca 2ϩ binding to S100B might contribute to strengthen the interaction via the C terminus of S100B.
Mapping the Minimal Interaction Domain for S100B on IQ-GAP1-To investigate which domains of IQGAP1 are responsible for S100B binding, the full-length protein and the indicated mutants of IQGAP1 were produced in rabbit reticulocytes (Fig. 7A), and their interaction with S100B-and CaM-Sepharose beads compared (Fig. 7B), as described under "Materials and Methods." There is no difference between the full-length IQGAP1 and the N-terminal domain of IQGAP1 in binding S100B and CaM beads. All fusion proteins containing the IQ domain (IQ, CHD-IQ, and IR-IQ) also bind to S100B and CaM in the presence of EGTA/EDTA. These findings suggest that IQ motifs are essential for interactions between S100B and IQ-GAP1. They are also consistent with previous data showing that the high affinity CaM binding region on IQGAP1 corresponds to its IQ domains (29).
CaM for binding with IQGAP1, we next performed competition assays. Purified S100B was mixed with 35 S-labeled recombinant IQGAP1 together with CaM-Sepharose beads. The amount of IQGAP1 bound to calmodulin, in the presence of various concentrations of divalent ions and of S100B, was quantified by measuring the radioactivity associated with the CaM-beads pellet (Fig. 7C). In the absence of divalent ion, S100B interfered with the association of IQGAP1 with CaM only at high concentration. In the presence of Zn 2ϩ , or Zn 2ϩ plus Ca 2ϩ , S100B produced a dose-dependent inhibition of binding of IQGAP1 to CaM. The S100B concentration-dependent inhibition curves show that, in the presence of Zn 2ϩ , or Zn 2ϩ plus Ca 2ϩ , half inhibition occurs with a S100B concentration below the estimated CaM-Sepharose concentration in the assay (1.8 M), suggesting that, in its Zn 2ϩ -or Ca 2ϩ /Zn 2ϩbound conformations S100B may have higher affinity for IQ-GAP1 than CaM. As expected, when purified bovine brain CaM was used in competition with CaM-Sepharose, in the presence of Ca 2ϩ and Zn 2ϩ , for binding IQGAP1, the inhibition curve is shifted to a higher competitor protein concentration (Fig. 7C). Inversely, when the purified competitor proteins, CaM and S100B, were mixed with 35 S-labeled recombinant IQGAP1 together with S100B-Sepharose beads, CaM antagonized IQ-GAP1 binding to S100B-Sepharose at a much higher concentration than S100B (Fig. 7D) confirming that in its Ca 2ϩ /Zn 2ϩbound conformations S100B has a higher affinity for in vitro translated IQGAP1 than has CaM. S100B Co-localizes with IQGAP1 at Ruffling Membranes but Not at Sites of Cell-Cell Contact in U373 Cells-The specificity of interaction between S100B and IQGAP1 in U373 cells was then investigated at the cellular level by indirect double im-  2 and 7) or monoclonal anti-IQGAP1 AF4 antibody (lanes 3-6 and 8 -11). Lane 1 is total cell extract used for immunoprecipitation. Protein complexes were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and probed with monoclonal anti-IQGAP1 or with polyclonal anti-S100B antibodies. Immune complexes were visualized using an ECL kit. B, [ 35 S]methionine-labeled recombinant IQGAP1 produced in rabbit reticulocyte was mixed with soluble purified recombinant S100B (2 M), then S100B was immunoprecipitated with monoclonal anti-S100B S16 antibody (1 g) (lanes [3][4][5] or with S100B-Sepharose beads (2 M) (lanes 7-9) in buffer containing 5 mM EDTA and 5 mM EGTA (lanes 3 and 7), 10 M ZnSO 4 (lanes 4 and 8), or 0.3 mM CaCl 2 and 10 M ZnSO 4 (lanes 5 and 9). Protein complexes were resolved by SDS-PAGE and autoradiography. Lanes 1 and 6 correspond to total reticulocyte lysate. Lane 2 is immunoprecipitation with control anti MyoD antibody. C, quantitative analysis of the radioactivity associated with S100B-Sepharose beads as shown in B.
Results are the average of one typical experiment done in triplicate. Lane C corresponds to S100A11-Sepharose beads used as control. D, mouse NIH 3T3 cells were transfected with S100B pcDNA, and S100B was immunoprecipitated with monoclonal S16 antibody in buffer  4). Protein complexes were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and probed with monoclonal anti-IQGAP1 or with polyclonal anti-S100B antibodies. Immune complexes were visualized using an ECL kit.
FIG. 6. Mechanism of Zn 2؉ -dependent interaction of S100B with IQGAP1. A, NIH 3T3 fibroblasts were transfected with empty (control) plasmid (lanes 1-3), or with wild-type (wt) S100B (lanes 4 -6) or mutant (⌬Ct) S100B⌬Ct (lanes 7-9) plasmids for 36 h by the calcium phosphate method. Cell extracts were immunoprecipitated using the monoclonal anti-S100B S16 antibody (lanes 2-3, 5-6, and 8 -9). Immunoprecipitations were performed in the absence (lanes 2, 5, and 8) or in the presence (lanes 3, 6, and 9) of Ca 2ϩ /Zn 2ϩ . Lanes 1, 4, and 7 are total cell extracts used for immunoprecipitation. B, NIH 3T3 cells were transfected with control plasmid (lane 1) or with S100B⌬Ct plasmid (lanes 2-5) as in A. S100B⌬Ct was immunoprecipitated with monoclonal S16 antibody in buffer containing 20 M EGTA (lanes 1 and 2) In A and B, protein complexes were resolved by SDS-PAGE, transferred to nitrocellulose membrane, and probed with monoclonal anti-IQGAP1, polyclonal anti-AHNAK, and polyclonal rabbit anti-S100B. Immune complexes were visualized by an ECL kit. munofluorescence studies (Fig. 8). In sub-confluent U373 MG cells, S100B expression is down-regulated (Fig. 1) and is hardly detectable with the monoclonal S16 anti S100B antibody. Sparse U373 cells were transfected with the S100B cDNA and the sub-cellular localization of both ectopically expressed S100B and IQGAP1 were evaluated with anti S100B monoclonal antibody and rabbit anti-IQGAP1 antibody (Fig. 8). The majority of ectopically expressed S100B protein translocates and accumulates within cell nuclei. Nevertheless, a substantial amount of S100B remains specifically sequestered within the cytoplasm. In the cytoplasm, S100B accumulates and co-localizes with IQGAP1 in a region that resembles the polarized leading edge and area of membrane ruffling (Fig. 8A, see arrow). Membrane ruffles were characterized by intense staining with fluorescein isothiocyanate-phalloidin (not shown). Several groups have shown previously in different cell models that IQGAP1 accumulates at the polarized leading edge and areas of membrane ruffling (27,34). At the polarized leading edge, IQGAP1 forms a tripartite complex with CLIP-170 and Rac1/ Cdc42 to regulate linkage of microtubules to the cortical region (27). In transfected U373 cells, S100B immunoreactivity is totally excluded at the sites of cell-cell contacts where IQGAP1 also localizes (Fig. 8B, see arrowheads). At the cell-cell junctions, IQGAP1 may associate with Cdc42 or ␤-catenin for regulation of cadherin-based cell adhesion (22)(23)(24)(25). S100B and IQGAP1 Relocate in a Ca 2ϩ -dependent Manner within Secretion-associated Vesicle-like Structures in Human Astrocytoma U87 Cells-Relocation of S100B from the perinuclear area toward the periphery of the cell in the form of vesicle-like structures in response to intracellular Ca 2ϩ increase has been recently reported in the human astrocytoma U87 cell line transfected with S100B-GFP fusion protein (12). In this cell model, Ca 2ϩ -dependent relocation of S100B-GFP has been linked to the S100B secretion pathway (12). To evaluate whether IQGAP1 could be implicated in the S100B-secre-tion pathway, we studied the effect of intracellular Ca 2ϩ increase on S100B and IQGAP1 localization in U87 cells. U87 cells were transiently transfected with the S100B cDNA and the sub-cellular localization of S100B, and IQGAP1 were evaluated by immunocytochemistry (Fig. 9). In U87 cells grown in normal medium, ectopically expressed S100B protein localizes FIG. 8. Cytoplasmic S100B protein co-localizes with IQGAP1 protein at membrane ruffling but not at sites of cell-cell contact in U373 MG cells. Sparse U373 MG cells transfected with the S100B expression plasmid were fixed and double-stained with monoclonal anti-S100B and polyclonal anti-IQGAP1 as indicated. In sub-confluent U373 MG cells, S100B expression is down-regulated and is hardly detectable with the monoclonal S16 anti-S100B antibody. Only transfected cells with S100B plasmid are strongly immunostained. Arrows show co-localization of S100B with IQGAP1 in the region that resembles the polarized leading edge and area of membrane ruffling. Arrowheads point to the site of cell-cell contact where IQGAP1 but not S100B also localizes. FIG. 7. Mapping of the S100B-binding domains on IQGAP1. A, domain diagram of IQGAP1. IQGAP1 has a variety of domains, including a calponin homologous domain (CHD), six internal repeats (IR) of yet unknown function, four IQ repeats, and a Ras GAP-related domain (GRD). B, in vitro translated IQGAP1 and different N-terminal domains as indicated were tested for interaction with S100B- Sepharose (lanes 2-3) or CaM-Sepharose (lanes 4 -5) in buffer supplemented with 5 mM EDTA/EGTA (lanes 2 and 4) or 0.3 mM Ca 2ϩ plus 10 M Zn 2ϩ (lanes 3 and 5). Lane 1 corresponds to total reticulocyte lysate that has been precipitated by 20% trichloric acid (TCA). to membrane ruffling, the perinuclear area, and within the cell nuclei (Fig. 9A). Ionomycin stimulation of U87 cells induces rapid relocation of S100B within vesicle-like structures that have the appearance of membrane blebs (Fig. 9B). As previously noticed (12), these vesicle-like structures are often located toward the periphery of the cells. When considering IQ-GAP1 in non-stimulated U87 cells, IQGAP1 is diffusely located within the cytoplasm and accumulates at membrane ruffling (Fig. 9A). Increase in the intracellular Ca 2ϩ concentration causes relocation of IQGAP1 to the newly formed membrane blebs where it co-localizes with S100B (Fig. 9B). IQGAP1 relocation did not depend on the expression levels of S100B, because it is also observed in non-transfected cells.
Ca 2ϩ Regulates Both Relocation and Stable Association of S100B and IQGAP1 with Membrane Vesicles in MCF7 Cells-As observed in most epithelial cells, in MCF7 human breast epithelial cells, IQGAP1 accumulates at the cell-cell junction (Ref. 24 and Fig. 10A). With MCF7 cells stably transfected with the S100B gene (S100B-MCF7) and grown in complete culture medium, IQGAP1 remains concentrated at the cell-cell junction, whereas the ectopically expressed S100B protein accumulates within the cell nuclei (Fig. 10A). When S100B-MCF7 cells are cultured in serum-free medium, intracellular calcium elevation mediated by calcium ionophore ionomycin produced a rapid relocation of both S100B and IQGAP1 within the cell cytoplasm (Fig. 10B). When ionomycin-containing medium was replaced by fresh complete medium, both S100B and IQGAP1 relocated to the cell nuclei and the cell-cell junctions, respectively, indicating that Ca 2ϩ -dependent translocation is reversible. However, in many S100B-MCF7 cells, a co-localization of S100B and IQGAP1 persisted within newly formed membrane blebs located toward the periphery of the cells (Fig. 10C).

DISCUSSION
Characterization of the Molecular Interaction between S100B and IQGAP1-In solution, S100B associates as a non-covalent dimer. In its dimeric form, S100B interacts not only with Ca 2ϩ but also with Zn 2ϩ ions (2). S100B binds Zn 2ϩ ions with affinity in the nanomolar range (2). In contrast, the S100B dimer affinity for calcium is rather weak compared with other EFhand calcium-binding proteins and is not within the range of physiological intracellular calcium concentrations (2). In the presence of Zn 2ϩ , or upon alkylation of Zn 2ϩ ligand Cys84, the S100B adopts a "Ca 2ϩ -bound-like" conformation (2,31). That conformation is associated with destabilization of the quaternary protein structure, exposure of the calcium-binding sites to solvent, and increased apparent Ca 2ϩ affinity compatible with local intracellular calcium concentration (2,31). The capacity of S100B to bind and release Zn 2ϩ without denaturation suggests that Zn 2ϩ may not only play a structural role but might be involved, together with Ca 2ϩ , in concerted regulation of S100B interaction with target proteins. In this study, we identified for the first time a Zn 2ϩ -dependent S100B target protein, IQGAP1. Zn 2ϩ -bound S100B co-immunoprecipitates with IQGAP1 present in cell extract or expressed in rabbit reticulocyte. In contrast to immunoprecipitation assay, pull-down assay using FIG. 9. Ca 2؉ -dependent relocation of S100B and IQGAP1 within secretion associated vesicle-like structure in U87 cells. Transiently transfected U87 cells with S100B-pcDNA were incubated 5 min in complete medium plus Me 2 SO (A) or 5 M ionomycin in Me 2 SO (B). Cells were fixed and double-stained with polyclonal anti-S100B or with monoclonal anti-IQGAP1 as indicated.
FIG. 10. Ca 2؉ regulates relocation of S100B and IQGAP1 in MCF7 cells. A, stably transfected MCF7 cells with S100B-pcDNA were grown in complete medium. B, S100B-MCF7 cells were left 20 h in serum-free medium and stimulated 5 min with 5 M ionomycin prior to fixation. C, S100B-MCF7 cells were treated as in B. After 5 min culture medium was changed to fresh complete medium and cells were left 1 h prior to fixation. In A, B, and C, cells were double-stained with polyclonal anti-S100B or with monoclonal anti-IQGAP1 as indicated. S100B cross-linked onto Sepharose beads revealed that the S100B/IQGAP1 interaction can also be detected in the presence of EGTA/EDTA in binding buffer. We have shown that these discrepancies probably result from differences in conformation between the soluble S100B and the S100B cross-linked onto Sepharose beads (Fig. 5B). This was further confirmed by competition experiments using CaM-Sepharose. Only in its Zn 2ϩor Zn 2ϩ /Ca 2ϩ -bound conformations is S100B capable of substantially antagonizing the binding of recombinant IQGAP1 to CaM-Sepharose beads (Fig. 7C). The mainly CaM-binding domains on IQGAP1 correspond to IQGAP1-IQ motifs (29). It is therefore likely that Zn 2ϩ -dependent interactions of S100B with IQGAP1 also require IQGAP1-IQ motifs. This was confirmed by mapping the Ca 2ϩ -independent S100B-binding domain on IQGAP1 using a pull-down assay (Fig. 7, A and B). The IQ motif was initially identified in brain-specific protein kinase C substrates neuromodulin (GAP43) and neurogranin as part of Ca 2ϩ -independent CaM binding and protein kinase C phosphorylation site domain (35,36). IQ motifs have since been identified as binding sites for CaM in a variety of proteins (37). We provide here evidence that the IQ motifs are not specific to CaM and can also be targeted by Zn 2ϩ -bound S100B. The Zn 2ϩ -dependent interaction of S100B with IQGAP1 is unique among the S100B-target proteins so far identified. With conventional S100B target proteins, Zn 2ϩ does not promote direct interaction but modulate S100B Ca 2ϩ affinity (28). Although Zn 2ϩ binding to S100B is sufficient to promote interaction of S100B with IQGAP1, quantitative comparison of binding of full length recombinant IQGAP1 to S100B revealed that addition of Ca 2ϩ to the binding buffer significantly potentiates the interaction between the two proteins ( Fig 5B). A similar effect of Ca 2ϩ was reported for CaM/IQGAP1 interaction (26,30). It has been proposed that differences in CaM binding to IQGAP1 in the absence and in the presence of Ca 2ϩ is attributable to difference between IQ motifs, some of which bind Ca 2ϩ -CaM, and others bind Ca 2ϩ -free CaM (26,30). Because Zn 2ϩ -bound S100B prevents the binding of IQGAP1 to Ca 2ϩ -free CaM, it is possible that, in its Zn 2ϩ -bound conformation, S100B interacts with some of the IQ motifs that are capable of binding CaM in the absence of Ca 2ϩ . When complexed to Ca 2ϩ , the S100B would then be able to utilize the four IQ motifs to further strengthen its interaction with IQGAP1. It is also possible that other domains on IQGAP1 could confer Ca 2ϩ sensitivity to the S100B/IQGAP1 interaction. A striking amino acid sequence conservation exists between the IQGAP1-5 repeat motif ( 541 INEALDEGDAQ 550 ) and the Ca 2ϩ -dependent S100B-binding domain on p53 ( 344 LNEALELKDAQ 353 ) (38). The S100Bbinding domain present on p53 is a tetramerization domain that is also implicated in the interaction of p53 with other regulatory proteins (39). Further studies should explore if the interaction of S100B with IQGAP1 could regulate IQGAP1 interactions with partner proteins through its repeat motifs.
IQGAP1 as a Mediator of S100B-induced Regulation of Cell Shape and S100B Secretion Pathway-In post-confluent human glioma U373 MG, the strongest S100B immunoreactivity is found associated with cells that are characterized by a less flattened morphology and long processes (Fig. 2). In these cells, a co-localization of S100B and IQGAP1 is evident at plasma membrane and within growing processes. Transient transfection of S100B in sub-confluent U373 MG cells confirmed that cytoplasmic S100B specifically co-localizes with IQGAP1 at the plasma membrane but not at sites of cell-cell junction (Fig. 8).
Many studies have implicated IQGAP1 as a scaffold to recruit and localize protein complexes involved in actin and microtubule-based cellular functions at the plasma membrane (19,20,27,34). Interaction of Ca 2ϩ /Zn 2ϩ -S100B with IQGAP1 could, therefore, regulate IQGAP1 scaffold function at the plasma membrane in response to incoming signals linked to the reorganization of the actin and microtubule cytoskeleton. This hypothesis is consistent with previous studies showing that selective inhibition of S100B expression by antisense strategies in rat glioma C6 cells resulted in a more flattened cellular morphology and a more organized actin stress fiber staining pattern with less membrane ruffling (6). It is noteworthy that IQGAP1 can also bind to S100A1 (Fig. 3B). S100A1, the closest S100B homologue, is also a potential regulator of cell cytoskeleton and cell morphology (41).
In this study, we have also provided evidence that the interaction between S100B and IQGAP1 might also occur within vesicle-like structures that also have the appearance of membrane blebs in the human astrocytoma U87 and MCF7 cells. In U87 cells, both S100B and IQGAP1 translocate and co-localize within vesicle-like structures in response to increased intracellular calcium. The Ca 2ϩ -dependent translocation of S100B in U87 has been studied in detail elsewhere (12). It has been proposed that these vesicle-like structures, which correspond to cytoplasmic extensions that form and disappear dynamically, are implicated in Ca 2ϩ -dependent S100B secretion process (12). In U87 cells not stimulated with ionomycin, these vesiclelike structures were rarely immunostained with IQGAP1 antibodies, suggesting that Ca 2ϩ regulates targeting of IQGAP1 to these structures. The mechanism that controls Ca 2ϩ -dependent IQGAP1 relocation is independent of S100B, because it is also observed in cells that do not express S100B. Ca 2ϩ -dependent translocation of S100B and IQGAP1 within structures having the appearance of membrane blebs was confirmed with stably transfected S100B-MCF7 cells (Fig. 10). In stably transfected S100B-MCF7, IQGAP1 is naturally targeted to the cellcell junctions and S100B accumulates predominantly within the cell nuclei. Serum deprivation coupled to intracellular calcium elevation mediated by ionomycin induces a relocation of both S100B and IQGAP1 within the cell cytoplasm. This process is maximal within 5 min, supporting the idea that it depends on calcium fluxes. When ionomycin-containing medium is replaced by fresh complete medium, the majority of S100B and IQGAP1 has a tendency to relocated to nuclei and to cell-cell junctions, respectively, indicating that Ca 2ϩ -dependent translocation is reversible. However, in many S100B-MCF7 cells, a co-localization of S100B and IQGAP1 persisted within membrane blebs located toward the periphery of the cells. Hence, Ca 2ϩ is required for both S100B and IQGAP1 relocation and their subsequent stable association with newly formed membrane structures. Taking into account the high affinity interaction between S100B and IQGAP, we propose that IQ-GAP1 might be implicated in recruiting S100B to secretory vesicular structures and might be involved in the Ca 2ϩ -dependent S100B secretion pathway. The observed Ca 2ϩ -dependent association of IQGAP1 with vesicle-like structure is new and might reveal an additional cellular function for the protein.
Whether or not IQGAP1 directly participates in stabilization, turnover, and dynamics of these vesicle-like structures has yet to be evaluated. It is significant that functional disruption of Iqg1p, a yeast homologue of the mammalian IQGAPs, has been directly implicated in vesicle accumulation at the growing bud, suggesting a possible involvement of yeast Iqg1p in secretion or some aspect of vesicle trafficking (40).
The interaction between cytoplasmic S100B and IQGAP1 might also have important implications for understanding the relationship between overexpression of cytoplasmic S100B and the development of a more aggressive cell phenotype during brain tumor progression. In astroglial brain tumors, progression from low grade (astrocytoma) into faster growing, more dysplastic and invasive high grade tumors (glioblastomas), correlates with increased expression of cytoplasmic S100B (42). Although a causal relationship between S100B concentration and malignancy has not been demonstrated, it is hypothesized that increased concentration of S100B may contribute to neoplastic transformation. In these brain tumor cells, overexpressed S100B could interfere with the regulatory function of IQGAP1 during processes of dynamic rearrangement of cell-cell adhesion (23,24) to favor cell motility and metastasis.