Complex Formation between S100B Protein and the p90 Ribosomal S6 Kinase (RSK) in Malignant Melanoma Is Calcium-dependent and Inhibits Extracellular Signal-regulated Kinase (ERK)-mediated Phosphorylation of RSK*

Background: S100B is overexpressed in malignant melanoma and contributes to cancer progression. Results: The S100B-RSK complex was found to be Ca2+-dependent, block phosphorylation of RSK at Thr-573, and sequester RSK to the cytosol. Conclusion: The Ca2+-dependent S100B-RSK complex provides a new link between the MAPK and Ca2+ signaling pathways. Significance: S100B inhibitors may restore normal MAPK and Ca2+ signaling in malignant melanoma. S100B is a prognostic marker for malignant melanoma. Increasing S100B levels are predictive of advancing disease stage, increased recurrence, and low overall survival in malignant melanoma patients. Using S100B overexpression and shRNAS100B knockdown studies in melanoma cell lines, elevated S100B was found to enhance cell viability and modulate MAPK signaling by binding directly to the p90 ribosomal S6 kinase (RSK). S100B-RSK complex formation was shown to be Ca2+-dependent and to block ERK-dependent phosphorylation of RSK, at Thr-573, in its C-terminal kinase domain. Additionally, the overexpression of S100B sequesters RSK into the cytosol and prevents it from acting on nuclear targets. Thus, elevated S100B contributes to abnormal ERK/RSK signaling and increased cell survival in malignant melanoma.

The S100 protein family consists of more than 20 members expressed in a tissue-and cell type-specific manner. One of the earliest discovered S100 family members, S100B, is a 21.5-kDa symmetric noncovalant homodimer found in melanocytes, glial cells, chondrocytes, and adipocytes (1,2), with higher than normal levels observed in malignant melanoma (3) and in several other cancers (4,5). Clinically, S100B is a prognostic marker for melanoma (6), with increasing serum levels predictive of disease stage, increased cancer recurrence, and low patient survival (7,8).
MAPK is another growth signal activated by S100B, although this effect is indirect and via receptor-mediated processes (19,22). The role of intracellular S100B on MAPK signaling is not well defined, and no such information is available for melanoma (23)(24)(25)(26)(27)(28)(29). Nonetheless, ERK is overly active in 90% of melanoma patients because of activating mutations in upstream kinases such as BRAF (mutations in 50 -70% of melanomas) and NRAS (mutations in 15-30% of melanomas), so a myriad of ERK1/2 targets involved in cell proliferation, differentiation, increased survival, and the reduction of apoptosis are impacted in melanoma (22, 30 -32). In this study, the role of S100B on MAPK signaling pathways was evaluated and found to directly affect RSK phosphorylation.
The RSKs 4 (90-kDa ribosomal S6 kinases) represent one important set of ERK1/2 substrates with four family members (RSK1-4), each containing six highly conserved phosphorylation sites (Ser-221, Thr-359, Ser-363, Thr-573, Ser-380, and Ser-749) (Fig. 4A). In quiescent cells, ERK1/2 associates with a docking domain in the extreme C terminus of RSK (consensus sequence LXPXXXSXLAXRRXXK) (33). Upon phorbol esteror growth factor-dependent activation, phosphorylation of RSK is achieved by ERK1/2, PDK1, and two autophosphorylation events (34). In the canonical RSK activation pathway, ERK phosphorylates RSK in its linker region (Thr-359, Ser-363) and at Thr-573 in the C-terminal kinase domain (CTKD). The ERKdependent phosphorylation of RSK at Thr-573 enables autophosphorylation at Ser-380 via residues from the CTKD. Phosphorylated Ser-380 on RSK also creates a docking site for PDK1 that, in turn, phosphorylates Ser-221 in the RSK N-terminal kinase domain. The N-terminal kinase domain of RSK is then able to trans activate a variety of downstream protein targets necessary for cellular function. As part of a feedback loop, the activated N-terminal kinase domain catalyzes a second autophosphorylation event within RSK at Ser-749 to dissociate the ERK-RSK complex and disengage the RSK signaling cascade (35).
To better understand how elevated S100B affects ERK/RSKmediated signaling in malignant melanoma, S100B knockdown and overexpression studies were completed using the 501mel (low S100B) and WM115 (elevated S100B) melanoma cell lines. It is generally assumed that active ERK inevitably results in the phosphorylation and activation of its downstream target p90 RSKs (34,36). However, surprisingly, inhibition of a critical ERK phosphorylation site on RSK, at Thr-573, was observed when S100B levels are elevated. Furthermore, phosphorylation of RSK Thr-573 was blocked via a direct Ca 2ϩ -dependent interaction between S100B and the CTKD of RSK and promoted RSK sequestration to the cytoplasm. Thus, in addition to repressing the p53-tumor suppressor pathway (9 -11), elevated S100B alters MAPK signaling in malignant melanoma via a direct and Ca 2ϩ -dependent interaction with RSK.

EXPERIMENTAL PROCEDURES
Cell Lines and Cell Culture-WM115 (ATCC) malignant melanoma cells were cultured in minimal essential medium (Invitrogen) supplemented with 10% heat-inactivated FBS and 100 units/ml penicillin/streptomycin. 501mel (Dr. Ruth Halaban, Yale University) malignant melanoma cells were cultured in RPMI medium (Invitrogen) supplemented with 10% heatinactivated FBS and 100 units/ml penicillin/streptomycin. Where indicated, assay medium was composed of minimal essential medium or RPMI medium with 1% charcoal-stripped, dextran-treated FBS (Hyclone) and 100 units/ml penicillin/ streptomycin. All cells were maintained in a 37°C incubator with 5% CO 2. Both cell lines were sequenced to establish BRAF status. The WM115 cells harbor the activating BRAF V600D mutation and wild-type NRAS, and the 501mel cells have WT BRAF and the activating NRAS G12D mutation (37). As shown here by Western blot analysis, the WM115 cells have elevated levels of endogenous S100B protein, and the 501mel cells have little, if any, detectable S100B protein, which is rare in malignant melanoma.
Lentiviral shRNA Particle Infections-WM115 cells were seeded in triplicate at 1 ϫ 104 cells/well in 96-well plates in normal growth medium and allowed to recover overnight. The cells were infected with SMARTvector 2.0 lentiviral particles containing either non-targeting scrambled or anti-S100B shRNA according to the recommendations of the manufacturer (Thermo Scientific Dharmacon). The following day, the medium containing lentivirus was removed, the cells were washed twice with PBS and trypsinized, and each well was expanded into a 24-well plate containing growth medium supplemented with puromycin (0.5 g/ml). Upon confluence, the wells were trypsinized and single cell-diluted into 96-well plates. Positive clones, having significantly reduced S100B expression, were maintained in puromycincontaining medium.
Site-directed mutagenesis-Human S100B cDNA was purchased from the ATCC, subcloned into the mammalian expression vector pcDNA3.1(ϩ) (Invitrogen), and confirmed by sequence analysis. Using the QuikChange II site-directed mutagenesis kit (Agilent), two successive point mutations were introduced into the calcium-binding domains of the S100B cDNA. First, the glutamate at position 31 was changed to an alanine, and then the glutamate at position 72 was changed to an alanine, creating the E31A/E72A double mutant of S100B, and the mutations were verified by sequencing. The E31A/ E72A mutations did not affect protein structure but abolished detectable calcium-binding activity (18).
Transfections-501mel cells were seeded in 6-well plates at 2.5 ϫ 10 5 cells/well and allowed to recover overnight. The 501mel cells were then transfected with the pcDNA3.1(ϩ) vector alone, with the same vector containing human wild-type S100B, or with the vector harboring the E31A/E72A double mutant of S100B, which no longer binds Ca 2ϩ ions. The transfections were completed using the Mirus TransIT-LT1 reagent (Mirus Bio), following the protocol of the manufacturer. After 48 h of incubation at 37°C, the transfected cells were transferred to growth medium containing neomycin (0.5 mg/ml, Invitrogen) and single cell-diluted into 96-well plates. Positive clones overexpressing wild-type or mutant S100B were maintained in neomycin-containing medium.
Cell Viability Assays-Cells were seeded in triplicate at 3 ϫ 10 3 cells/well in 96-well plates in assay medium overnight. The medium was then changed to fresh normal growth medium (day 0). At each time point, 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (Sigma) was added for 3 h. The resultant formazan crystals were solubilized by the addition of a 1:10 solution of 0.1 M glycine (pH 10.5):dimethyl sulfoxide overnight at room temperature, and the absorbance (450 nm) was measured. Absorbance readings at day 0 were taken to establish starting cell viability, from which the percentage of cell viability over initial cells plated was calculated.
Preparation of Recombinant Proteins-The pET11b expression vector (Novagen) containing the wild-type rat S100B gene or the E31A/E72A double mutant rat S100B gene was used to produce recombinant S100B in HMS174 (DE3) cells (Novagen). S100B was prepared and purified (Ͼ99%) under reducing conditions using procedures similar to those described previously (14,39), except that DTT was used as a reducing agent instead of ␤-mercaptoethanol. The cDNA (Addgene) for the CTKD (residues 410 -735) of human RSK1 was subcloned into the pET-41 Ek/LIC expression vector (Novagen) according to the recommendations of the manufacturer, and its sequence was verified. The recombinant GST-RSK1 410 -735 fusion protein was expressed in Rosetta (DE3) cells (Novagen) and isolated using glutathione-Sepharose 4B beads (GE Healthcare Bioscience) batch purification according to the instructions of the vendor. The same protocol was used to purify GST for use as a control. The concentrations of the protein stock solutions were determined using a protein assay (Bio-Rad) using wildtype S100B of known concentration as the standard. The concentration of this S100B standard was determined by quantitative amino acid analysis (BioSynthesis).
Kinase Assays-Active ERK2 (2 ng, Millipore) was incubated for 15 min with 0.5 g of purified GST-RSK1 410 -735 (Ͼ99%) at 30°C in a buffer optimized for protein kinase reactions (New England Biolabs, termed NE buffer) in the presence and absence of 0.5 g of pure S100B (Ͼ99%). The final conditions in the reaction buffer were 20 M ATP, 50 mM Tris-HCl, 10 mM MgCl 2 , 0.1 mM EDTA, 2.0 mM DTT, 0.01% Brij 35 (pH 7.5), and, where indicated, 1.0 mM CaCl 2 . The kinase reactions were stopped by adding an equal volume of 4ϫ-concentrated SDS-PAGE sample buffer prior to analysis by Western blot.
Pulldown Assays-S100B or BSA was covalently attached to magnetic Dynabeads according to the recommendations of the manufacturer (Invitrogen). An aliquot of protein-bound beads was washed three times in radioimmune precipitation assay buffer (40 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 0.3% Triton X-100) supplemented with 5 mM EDTA or 5 mM calcium chloride, mixed with 500 g of WM115 cell lysate for 1 h at 4°C, and then washed again three times with radioimmune precipitation assay buffer. Target proteins were identified using Western blot analysis after boiling the beads in SDS-PAGE buffer. GST pulldown experiments were completed at 4°C by first washing glutathione-Sepharose 4B beads (100 l, GE Healthcare Bioscience) three times with TBS containing either 5 mM EDTA or 5 mM calcium chloride and then incubating the prewashed beads with 200 g of GST or GST-RSK1 410 -735 for 30 min. Next, the protein-bead complexes were washed three times and blocked with 1% BSA in TBS for 30 min, and then purified S100B protein (500 g) or the E31A/E72A mutant S100B protein (500 g) was added, rocked for 2 h at 4°C, and washed five times. The beads were boiled in SDS-PAGE buffer, and the eluted proteins were analyzed by Western blot analysis.
Subcellular Fractionation-Cells were grown and harvested at subconfluence, cell pellets were prepared for cytoplasmic and nuclear extraction using the NucBuster kit (Novagen) according to the recommendations of the manufacturer, and the subcellular fractions were analyzed by Western blot analysis. Primary antibodies for MEK1/2 (Cell Signaling Technology) and p84 (Abcam) served as controls for cytoplasmic and nuclear fractions, respectively, and were prepared using dilutions recommended by the manufacturer.

Elevated S100B Levels Positively Correlate with Increased
Melanoma Cell Viability-The role of S100B in melanoma progression has been established previously (10,11). However, all melanoma cells tested already expressed elevated S100B. Therefore, to elucidate whether S100B promotes melanoma viability, it was important to examine a melanoma cell line with low levels of S100B and determine the effect of adding S100B. This is particularly important because patients with low levels of S100B generally have a better prognosis in the clinic and often benefit from malignant melanoma immunotherapy approaches. The 501mel cells were critical for these studies because they do not express detectable levels of S100B, which is a rarity for human malignant melanoma (40,41). For these studies, separate single cell-derived S100B overexpression lines were derived from 501mel cells, and elevated S100B protein expression was confirmed by Western blot analyses (Fig. 1B). After 7 days, a significant increase in cell viability was observed for the S100B-expressing cells compared with cells containing the empty vector alone. In complementary experiments, stable shRNA expression was used to block S100B expression in WM115 melanoma cells that express high levels of endogenous S100B (Fig. 1A). For these studies, WM115 cells were infected with lentivirus containing human anti-S100B shRNA, and single cell clones were derived. Multiple clones were examined by Western blot analysis, and significant knockdown of S100B protein was observed (Ͼ99%). As a control, WM115 cells were infected with lentivirus containing scrambled non-targeting shRNA. After verifying significant S100B expression knockdown, cell viability was determined by 3-(4,5-dimethylthiazol-S100B-dependent Regulation of RSK 2-yl)-2,5-diphenyltetrazolium bromide assays. Both S100B knockdown clones 1 and 2 had decreased cell viability after 7 days compared with control cells with elevated S100B (Fig. 1A). The S100B knockdown and overexpression studies demonstrate that higher levels of S100B, as found in most malignant melanoma patients, directly correlate with increased cellular viability. The cell viability results from these cells lines are in agreement with previous S100B knockdown studies performed in human C8146A cells that were derived from patients with malignant melanoma (9 -11). S100B Inhibits ERK-dependent Phosphorylation of RSK at Residue Thr-573-One potential mechanism for affecting cell viability is an increased activation of the MAPK signaling cas-cade. Although the effects of S100B on phosphorylation and activation of ERK (i.e. to pERK) via cell-surface receptors is well established in several cell types (23)(24)(25)(26)(27)(28)(29), little is known about how intracellular S100B affects MAPK signaling, particularly in malignant melanoma where S100B levels are typically elevated. Therefore, the effect S100B expression has in 501mel or its inhibition in WM115 cells was examined first with respect to ERK activity. When intracellular S100B expression was blocked in WM115 cells, either no change or a small decrease in pERK was observed, whereas an increase in pERK was never seen (Fig.  2A). Consistent with the knockdown experiments but unlike robust increases in phosphorylated ERK (pERK) from receptormediated responses, a large increase in activated ERK was not observed with increased S100B in 501mel malignant melanoma  1 and 2). Cellular viability of the knockdown cells compared with that of the non-targeting scrambled cells was assessed at several time points over the course of 7 days using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide colorimetric assay to determine the change in cell number. The scrambled control cells (S), S100B knockdown clone 1 (1), and clone 2 (2) are also represented as in the inset. Shown is a compilation of three different experiments, each performed in triplicate (n ϭ 9). B, Western blot analysis was performed on total cell lysate (25 g) to confirm expression of S100B in the 501mel cells. Shown are 501mel cells containing vector alone (V) and the S100B clonal lines A and B (inset). Cellular viability of the 501mel S100B-expressing clonal lines was compared with that of the vector control cells at several time points measured over a period of 7 days. The vector control cells (V), S100B-expressing clone A (A), and clone B (B) are represented as in the inset. Shown is a compilation of three different experiments, each performed in triplicate (n ϭ 9). FIGURE 2. S100B suppresses RSK Thr-573 phosphorylation by ERK. A, Western blot analysis (25 g) of WM115 cells with the non-targeting scrambled vector (S) and two stable S100B knockdown clonal cell lines derived from WM115 cells (1 and 2). B, Western blot analysis (20 g) showing the effects of S100B expression in 501mel cells (first and second lanes) compared with an empty vector control (third and fourth lanes) in the absence (first and third lanes) or presence (second and fourth lanes) of the MEK1/2 inhibitor (5 g/ml) U0126.

S100B-dependent Regulation of RSK
cells (Fig. 3B). Thus, no reductions of phosphorylated ERK were observed, and the apparent slight increases in some experiments were likely experimental variability (Figs. 3B and 5A). These data indicate that intracellular S100B levels increase minimally or have no effect on ERK phosphorylation in these two melanoma cell lines.
However, despite finding that intracellular S100B minimally affects pERK levels, if at all, we discovered that phosphorylation of RSK, a downstream target of ERK, was inhibited significantly when S100B levels were increased (Fig. 6). Specifically, the stable introduction of S100B into 501mel malignant melanoma cells inhibited the phosphorylation of RSK at Thr-573 (Figs. 2B and 3B) without affecting the phosphorylation of other ERKdependent sites on RSK such as Thr-359/Ser-363 (Figs. 2B and 3B). The ability to block one ERK site on RSK suggests that S100B interacts with RSK directly. Previously, S100B has been shown to interact with other kinase substrates, rather than the kinase itself, to block phosphorylation via steric and/or allosteric mechanisms (16). A S100B binding site within the CTKD of RSK, sterically blocking Thr-573, would be consistent with this mechanism.
To confirm that ERK is indeed responsible for the phosphorylation of RSK in these cells, MEK1/2 was inhibited with U0126 (Fig. 2B). The addition of U0126 was found to abolish ERK phosphorylation in the 50lmel cells. Likewise, a drastic decrease in RSK phosphorylation occurred at Thr-573, Thr-359/Ser-363, and Ser-380 and demonstrated that ERK is indeed responsible for the majority of RSK phosphorylation in 501mel cells.
As an aside, a very low level of ERK-independent phosphorylation of RSK was observed (Fig. 3), but this residual phosphorylation was not affected by S100B addition (data not shown). In a complementary experiment, EGF was added to serum-starved WM115. However, pERK levels did not decrease with serum starvation, and EGF addition did not significantly increase pERK (data not shown). Thus, changing pERK levels in WM115 cells could not be easily achieved, which is not too surprising because WM115 cells already have activated ERK from the oncogenic BRAF mutation (i.e. V600E). As a result, the dynamic range for pERK levels is quite small, and the resulting effects from exogenous growth factors are minimal, if observed at all, as described previously (42).
The S100B Inhibitory Effect on RSK Phosphorylation Is Ca 2ϩ -dependent-To evaluate whether S100B protein alone is sufficient to inhibit ERK-dependent phosphorylation of RSK, cell-free protein kinase assays were completed. In these studies, ERK2-mediated phosphorylation of the RSK C-terminal kinase domain (RSK1 410 -735 ) at Thr-573 was monitored in the absence and presence of S100B (Fig. 3A). ERK2 successfully catalyzed the phosphorylation of RSK1 410 -735 within 15 min of incubation. In the presence of S100B and Ca 2ϩ , RSK phosphorylation by ERK2 was decreased significantly. However, this effect was abolished in the presence of a Ca 2ϩ chelator, EDTA (Fig. 3A). These data show that S100B inhibits ERK-dependent phosphorylation of RSK at Thr-573 in a Ca 2ϩ -dependent manner.
To further prove whether Ca 2ϩ was necessary for S100B to exert its inhibitory effect on RSK Thr-573 phosphorylation in cells, an E31A/E72A double mutant of S100B was tested, which is incapable of Ca 2ϩ binding. A comparison of protein levels over time using Western blot analysis confirmed that the stability of S100B within 501mel cells was not affected by the two mutations (data not shown). The S100B double mutant was similarly found to have no effect on activated ERK levels, and it had no effect on ERK-dependent phosphorylation of RSK at Thr-573 (Fig. 3B). These data provided additional evidence that the Ca 2ϩ -binding properties of S100B are necessary for its modulatory effects on RSK. Importantly, for the first time, these data also show the Ca 2ϩ requirement for an S100B function within cells. S100B Binds to RSK in a Ca 2ϩ -dependent Manner-Proteinprotein interactions involving S100B require a Ca 2ϩ -dependent conformational change to expose its protein target-binding site (14). Therefore, binding experiments were performed in vitro to determine whether S100B binds to ERK and/or to RSK directly (Fig. 4). For these studies, S100B or BSA was covalently attached to magnetic Dynabeads and mixed with WM115 cell lysate in the absence (EDTA-treated) or presence FIGURE 3. Inhibition of RSK phosphorylation by S100B is Ca 2؉ -dependent. A, in vitro kinase assays examining the ERK-mediated phosphorylation of RSK Thr-573 in the absence and presence of S100B and 1 mM CaCl 2 . Western blot analysis was employed following a 15-min incubation period, and changes in pRSK at Thr-573 were observed. B, Western blot analysis of 25 g of 501mel cell lysate with untreated cells (first lane) or cells transfected with empty vector (second lane), a vector with wild-type S100B (third lane), or a vector with the E31A/E72A double mutant of S100B (fourth lane), which is an S100B construct incapable of binding calcium (18). These data confirm that S100B must bind calcium ions to bind and inhibit RSK phosphorylation at Thr-573.

S100B-dependent Regulation of RSK
of Ca 2ϩ . Western blot analysis of the eluates revealed that fulllength cellular RSK, but not full-length ERK, bound to S100B in a Ca 2ϩ -dependent manner (Fig. 4B). Because S100B blocked the phosphorylation of several protein kinase C substrates by sterically preventing kinase access to the phosphorylation site, rather than by inhibiting kinase enzymatic activity (16,20), the ability of S100B to bind to the CTKD of RSK, containing Thr-573, was then tested. These binding experiments were performed with a GST-tagged RSK 410 -735 construct with either purified recombinant WT S100B or the Ca 2ϩ -binding double mutant (E31A/E72A, Fig. 4C). It is important to note that the E31A/E72A double mutant was sufficient to fully abolish Ca 2ϩbinding (K D Ͼ 500 mM) without affecting the overall structure of the protein, as evaluated by protein NMR (18). In this series of binding experiments, analysis of the eluates showed that WT S100B, but not the double mutant, bound to GST-RSK1 410 -735 . To further prove that the binding of S100B to the CTKD of RSK is Ca 2ϩ -dependent, binding was evaluated in the absence and presence of the Ca 2ϩ chelator EDTA (Fig. 4). Together, these data demonstrated that S100B binds directly within the CTKD of RSK and that S100B binding to the CTKD of RSK is Ca 2ϩ -dependent. Because recombinant proteins were used to study this S100B-RSK1 410 -735 complex, no posttranslational modifications were necessary for this protein-protein interaction.
Elevated S100B Inhibits the Nuclear Localization of RSK-RSK resides in the cytoplasm until mitogenic activation causes its translocation into the nucleus via an unknown mechanism (43). As a result, RSK is capable of phosphorylating different protein targets, depending on its cellular localization (36). We next determined whether S100B affected RSK cellular localization. Cytoplasmic and nuclear protein fractions were isolated from WM115 cells expressing non-targeting scrambled or anti-S100B shRNA. Western blot analyses showed an increase in nuclear RSK and phosphorylated Thr-573 RSK (pRSK573) in the S100B knockdown clone in comparison with control cells (Fig. 5A). Immunofluorescence studies showed that pRSK573 is enriched in the nuclei of S100B knockdown cells and that control vectors with elevated S100B showed diffuse staining of pRSK573 in the cytoplasm (Fig. 5B). To be certain, confocal XYZ image stacks established that pRSK573 is cytoplasmic and is above or below the nucleus. Areas within the nuclear boundaries that are devoid of DNA staining are typical of nucleoli, and the exclusion of pRSK573 from these areas is consistent with both a nuclear localization for pRSK573 and previous observations that RSK does not typically accumulate in nucleoli (Fig.  5C) (44). In summary, elevated S100B contributes to diffuse cytoplasmic staining of RSK, whereas blocking S100B expression allows for RSK translocations into the nucleus as necessary for the nuclear component of its biological activities (Fig. 6). These studies indicate that S100B is able to block RSK nuclear localization.

DISCUSSION
Increased S100B levels contribute to cancer cell growth and survival as shown here (Fig. 1) and elsewhere (19). One potential mechanism for S100B-dependent cell growth is its ability to activate ERK in an indirect manner via cell-surface receptors such as RAGE (23)(24)(25)(26)(27)(28)(29). However, other explanations can be provided. These include intracellular S100B-target interactions such as those involved in inhibiting p53 activities (i.e. S100B-p53, S100B-hdm2, and S100B-hdm4) (9 -11) and/or from other S100B-target interactions reviewed elsewhere (19). In this study, we examined, for the first time, whether intracellular S100B had a direct effect on MAPK signaling in malignant melanoma. Although no direct interaction was found between S100B and ERK, a Ca 2ϩ -dependent interaction was detected between S100B and a downstream ERK target, RSK (Fig. 4). As a result of this complex, S100B uniquely modulated a downstream ERK signal by blocking the phosphorylation of RSK at Thr-573 and preventing its nuclear localization (Figs. 2, 3, and 5). Such an effect would thereby inhibit the effects of RSK on its nuclear targets, and possibly increase its activity toward cytoplasmic targets (Fig. 6). The binding of S100B and RSK were also shown to be dependent on Ca 2ϩ , and, thus, S100B links two important signaling pathways involved in regulating cell growth/survival (i.e. MAPK and Ca 2ϩ signaling). Finally, the results presented here answer important but previously unresolved questions about the ability of S100B to bind intracellular FIGURE 4. S100B directly binds to RSK and a construct of RSK with the CTKD but not to ERK. A, schematic of RSK1 (residues  showing the N-terminal kinase domain (NTKD, residues 62-321), the CTKD (residues 418 -675), the ERK binding site at the C terminus of RSK1 (residues 722-735, underlined), and the phosphorylation sites reported to be necessary for RSK activation are labeled P (43). B, Western blot analysis of RSK and ERK protein levels from BSA or S100B pulldown eluates performed in the presence of either 5 mM EDTA (first through third lanes) or 5 mM CaCl 2 (fourth through sixth lanes). The first and fourth lanes contain total WM115 cell lysate antibody controls (CT), whereas eluates of BSA and lysate are shown in the second and fifth lanes, and S100B and lysate are shown in the third and sixth lanes. C, Western blot analysis of GST or GST-RSK1 386 -752 cell-free pulldowns supplemented with either 5 mM EDTA (second through fifth lanes) or 5 mM calcium (sixth through ninth lanes). Eluates of GST with S100B (second and sixth lanes), GST-RSK1 386 -752 with S100B (third and seventh lanes), GST with the E31A/E72A mutant (fourth and eighth lanes), and GST-RSK1 386 -752 with the E31A/E72A mutant (fifth and ninth lanes) are shown. S100B protein was loaded in the first lane as an antibody control. MAY 2, 2014 • VOLUME 289 • NUMBER 18 FIGURE 5. RSK nuclear localization is inhibited by S100B. A, Western blot analysis comparing the cytoplasmic (C) and nuclear (N) protein levels of S100B knockdown clone 1 (1) to those of the non-targeting scrambled cell line (S). Shown is an example experiment that was repeated in triplicate. The nuclear matrix protein p84 and MEK were used as loading controls for the nuclear and cytoplasmic fractions, respectively. B, immunofluorescence studies showing the localization of pRSK Thr-573 in both the S100B knockdown clone 1 and non-targeting scrambled cell lines (representative of three different fields of vision are shown). C, immunofluorescence studies focusing on a single XYZ image to show the localization of pRSK Thr-573 as well as nucleolar exclusion for each cell line. Planes are indicated by dotted lines and examples of nucleolar exclusion by arrows (representative of numerous XYZ images). FIGURE 6. Schematic representation of the Ca 2؉ -dependent effects of S100B on the MAPK signaling cascade. S100B-Ca 2ϩ directly interacts with RSK, inhibiting phosphorylation at Thr-573 by ERK and reducing subsequent translocation of RSK to the nucleus, thus allowing it to act on its cytoplasmic targets but not on nuclear targets. Phosphorylation sites on RSK are labeled with the letter P.

S100B-dependent Regulation of RSK
Ca 2ϩ ions and challenge current models for ERK-dependent activation of RSK.
The S100B-dependent inhibition of RSK Thr-573 phosphorylation by ERK in the cell does not occur with the Ca 2ϩ -binding mutant (Figs. 3 and 4), demonstrating that S100B undoubtedly binds Ca 2ϩ ions inside the cell. This is important because the Ca 2ϩ -binding affinities reported in vitro for the "typical" EFhand and the "pseudo" EF-hand of S100B are K D ϳ20 M and Ͼ200 M, respectively (45). Therefore, it has been questioned by many as to whether S100B could be an active signaling protein inside the cell where physiological Ca 2ϩ ion concentrations are generally very low (0.1-2 M) (46). It is possible that local Ca 2ϩ concentration gradients exist and/or increased Ca 2ϩ levels occur in cancer cells because of aberrant regulation allowing S100B to be activated (47). However, another explanation is that the Ca 2ϩ binding affinity of S100 proteins can also be increased upon binding other metals and/or their physiologically relevant protein target(s). For example, it is well established in vitro that the affinity of S100B and other S100 proteins for Ca 2ϩ is increased after Zn 2ϩ binding (48,49), redox modification of critical cysteine residues (50), and/or by target binding, although the mechanism by which Ca 2ϩ binding is increased by as much as 200-fold upon target binding to S100B is still under investigation (18,(51)(52)(53).
Another finding reported here is that increased S100B inhibited ERK-dependent phosphorylation of RSK at residue Thr-573 via a direct and Ca 2ϩ -dependent interaction (Fig. 6). In the same study, S100B had no effect on the phosphorylation of residues Thr-359/Ser-363 or Ser-380 of RSK (Figs. 2B and 3B). This is in conflict with the current model of RSK activation by sequential phosphorylation (43), where it is thought that ERK phosphorylation of Thr-573 occurs first, followed by ERK phosphorylation of Thr-359/Ser-363 in the linker region and the CTKD-mediated autophosphorylation of Ser-380. pSer-380 is said to create a binding site for PDK1, which phosphorylates RSK Ser-221, to fully activate RSK. However, because Thr-573 phosphorylation is blocked by S100B, we have shown that phosphorylation of Thr-573 is not required for RSK CTKD autophosphorylation of Ser-380. The ERK-dependent phosphorylation of RSK Thr-359/Ser-363 is sufficient to phosphorylate the remaining sequential sites on RSK, although it remains possible that an alternative kinase is responsible. The effects of S100B may not be limited to RSK, so additional studies are ongoing to determine whether S100B selectively inhibits the phosphorylation of other substrates of ERK.
The Ca 2ϩ -dependent binding of S100B to RSK was next found to block RSK nuclear localization (Fig. 5). Because RSK has specific functionality in the cytoplasm and the nucleus, its restriction to the cytosol in melanoma cells may preferentially drive the activation of only a subset of RSK targets. Likewise, the sequestering of RSK in the cytosol would prevent it from performing any necessary nuclear functions (Fig. 6). Although we have already shown that S100B inhibition of p53 plays a role in increased melanoma survival (9,10,15,16,20), modification in cytosolic and nuclear signaling because of S100B binding of RSK and the corresponding change in RSK localization could also contribute to increased cell proliferation and survival. For example, in the cytoplasm, RSK could contribute to increased cell growth by inducing the degradation of the NF-B inhibitor IB␣ (54 -56), suppressing BAD-mediated apoptosis (57,58) and/or inactivating tumor suppressors such as DAPK and TSC2 (43,59). Also, a recent study showed that MAPK-activated RSK promotes melanoma growth by increasing the activity of another cytoplasmic target, mammalian target of rapamycin complex 1, which is known to regulate cell growth, cell proliferation, cell motility, cell survival, protein synthesis, and transcription (35,60). In contrast, the sequestration of S100B of RSK to the cytoplasm would decrease the activity of numerous target transcription factors, several of which promote differentiation (35,61). Future experimentation is certainly required to investigate these and numerous other hypotheses stemming from these findings involving S100B-dependent regulation of MAPK signaling.
How S100B sequesters RSK in the cytoplasm is not known. It may sterically block its entry into the nucleus, mask a nuclear localization signal on RSK, and/or require Thr-573 phosphorylation for nuclear localization. Similarly, the small death effector domain protein PEA-15 (phosphoprotein enriched in astrocytes, 15 kDa) binds RSK2 and prevents it from entering the nucleus (62). The authors of the RSK2 study did not note any change in the phosphorylation state of RSK2, suggesting simply that its interaction with PEA-15 alone was enough to sequester it to the cytoplasm. Unlike PEA-15, the S100B interaction with RSK is Ca 2ϩ -regulated and, thus, links Ca 2ϩ and RSK signaling. A possible mechanism linking MAPK and Ca 2ϩ signaling is that at elevated Ca 2ϩ levels, RSK binds Ca 2ϩ -S100B and is sequestered to the cytoplasm, where it is able to phosphorylate a certain subset of target substrates, whereas, at lower levels of Ca 2ϩ , RSK is not bound to S100B and can freely translocate to the nucleus, modulating its nuclear targets (Fig. 6).
Like other Ca 2ϩ -signaling proteins (i.e. calmodulin, troponin C, etc.), S100 proteins regulate multiple biological activities. However, unlike calmodulin, which is ubiquitously expressed, the 24 S100 family members regulate individual biological activities in a cell-specific manner (12,13,19,63,64). It has also been established that S100B levels are highly elevated in malignant melanoma (3). However, our understanding of how S100B contributes to the melanoma phenotype is not fully understood. In this study, the effects of varying S100B levels on MAPK signaling were examined, providing more evidence that S100B is a direct mediator of tumor survival and not just a prognostic marker in this deadly cancer.