S100C/A11 is a key mediator of Ca²⁺-induced growth inhibition of human epidermal keratinocytes

Exquisite spatial and temporal control of cell growth and differentiation is a prerequisite for embryonic development and maintenance of fine tissue architecture. The epidermis is a typical tissue in which compartments of growing cells and growth-arrested terminally differentiating cells are clearly demarcated. In normal epidermis, proliferating cells are only observed in the basal layer. On moving to the upper layers, the cells stop growing, progressively follow a terminal differentiation pathway, and finally shed off from the surface. The critical step determining whether the cells withdraw from the growing cell population is regulated by a complex network involving many genes, but the precise molecular mechanisms remain largely unknown.

A number of factors have been reported to trigger differentiation of human and mouse epidermal keratinocytes in culture, i.e., increased extracellular Ca²⁺ concentration (Hennings et al., 1980), TGFβ (Shipley et al., 1986), forced expression of PKC (Ohba et al., 1998), and detachment of cells from the substrate (Zhu and Watt, 1996). Normal human keratinocytes (NHKs) continuously proliferate only in a culture medium with Ca²⁺ of <0.1 mM. An increase in the Ca²⁺ concentration to 1.2–2.0 mM results in termination of cell growth and induction of terminal differentiation phenotypes (Hennings et al., 1980). An increase in the extracellular Ca²⁺ level resulted in a sustained higher intracellular Ca²⁺ concentration (Sharpe et al., 1989). Some other conditions inducing differentiation of keratinocytes also lead to increased intracellular Ca²⁺ levels (Sharpe et al., 1989; Missero et al., 1996). An increasing gradient of Ca²⁺ concentration is present from the basal to the cornified layers of the epidermis in vivo (Menon et al., 1992). Mice lacking the expression of full-length extracellular Ca²⁺-sensing receptors showed deteriorated epidermal differentiation (Komuves et al., 2002). These results indicate that higher Ca²⁺ levels lead to induction of epidermal differentiation not only in culture, but also in vivo.

An elevation in the Ca²⁺ level triggers a number of intracellular signal transductional events, including production of inositol 1,4,5-triphosphate and 1,2-diacylglycerol; activation of calcineurin, PKCs, and Raf/MEK/ERK pathway; and tyrosine phosphorylation of p62 and fyn (Dotto, 1999; Schmidt et al., 2000). On the other hand, it has been shown that p21^CIP1/WAF1 was induced 4 or 8 h after an increase in the extracellular Ca²⁺ level, leading to inhibition of Cdk activity and blockage of cell cycle progression (Missero et al., 1996). p21^CIP1/WAF1 protein has been detected in cells of the suprabasal layers, but not in those cells of the basal layer of the human epidermis (Ponten et al., 1995). An important missing link is how the Ca²⁺-induced initial events lead to the induction of p21^CIP1/WAF1.

In a previous work on density-dependent growth inhibition of normal human fibroblasts, we identified S100C/A11 (calgizzarin), a member of the Ca²⁺-binding S100 protein family, as a key mediator of growth arrest (Sakaguchi et al., 2000). In a confluent state, S100C/A11 was phosphorylated at ¹⁰Thr and translocated into nuclei, and it eventually inhibited DNA synthesis through the induction of p21^CIP1/WAF1 and p16^INK4a. S100C/A11 is comprised in the epidermal differentiation complex (EDC) located on chromosome 1q21 in humans. EDC encodes nearly 30 genes. About half of them are specifically expressed during Ca²⁺-dependent terminal differentiation of keratinocytes (e.g., profilaggrin and loricrin), and the other half are members of the S100 protein family. The S100 family proteins have been assumed to play signal transduction roles in the differentiation of epidermis and other tissues. Some of the S100 protein family members were differentially expressed in normal human skin and melanocytic lesions (Boni et al., 1997). S100C/A11 was reported to be up- or down-regulated in malignant tumors (Van Ginkel et al., 1998). These results prompted us to examine possible involvement of S100C/A11 in growth regulation of epidermal keratinocytes. Here, we show that S100C/A11 is a key mediator of the high Ca²⁺-induced growth arrest in human keratinocytes.

At first, we confirmed that an increase in extracellular Ca²⁺ concentration from 0.03 to 1.5 mM results in inhibition of DNA synthesis of NHK cells in dose- and time-dependent manners (Fig. 1, A and B). HaCaT cells, an immortalized human keratinocyte line (Boukamp et al., 1988), were routinely cultivated in a medium with 1.5 mM Ca²⁺. When the concentration of Ca²⁺ was increased to 5 or 10 mM, DNA synthesis of HaCaT cells was inhibited (Fig. 1, A and B). Under similar conditions, Sharpe et al. (1989) and Gonczi et al. (2002) reported that there was no remarkable difference in intracellular Ca²⁺ level (70–100 nM) of NHK and HaCaT cells, despite different extracellular Ca²⁺ concentrations. Intracellular Ca²⁺ concentrations in NHK and HaCaT cells increased by ∼40% within 30 min after increasing the extracellular Ca²⁺ concentrations from 0.03 to 1.5 mM and from 1.5 to 10 mM, respectively (Fig. 1 C). Overall, the response of HaCaT cells to 10 mM Ca²⁺ was similar to that of NHK cells to 1.5 mM Ca²⁺. No immediate cytotoxicity was noted in HaCaT cells in a medium with 10 mM Ca²⁺ during the observation period. Therefore, throughout the present work, 0.03 and 1.5 mM Ca²⁺ was used as low Ca²⁺ media and 1.5 and 10 mM Ca²⁺ was used as high Ca²⁺ media for NHK and HaCaT cells, respectively.

To examine the possible involvement of S100C/A11 in growth regulation of human keratinocytes, we transfected various constructs of S100C/A11 flanking SV40-derived NLS to exponentially growing HaCaT cells. As shown in Fig. 2 A, incorporation of BrdU was specifically inhibited in the cells transfected with wild-type S100C/A11 (indicated by arrows), whereas GFP alone did not show any effects. DNA synthesis was inhibited by ∼25% by the wild-type S100C/A11 (Fig. 2 B). This reduction is substantial, considering that the transfection efficiency was at most 40%. Introduction of recombinant S100C/A11 protein into HaCaT cells by the aid of the protein transduction domain of TAT protein (YGRKKRRQRRR; Schwarze et al., 1999) efficiently inhibited DNA synthesis (Fig. 2 C). The TAT-flanked protein was transferred to cell nuclei with an efficiency of nearly 100%.

When extracts of NHK and HaCaT cells labeled with [³²P]phosphate were analyzed by a two-dimensional gel electrophoresis, increased phosphorylation of S100C/A11 (as identified by sequencing) was noted in both cell types exposed to high Ca²⁺ (Fig. 3 A). Exposure to high Ca²⁺ increased the immunoprecipitable amount of S100C/A11 and more remarkably the level of phosphorylation in both cell types (Fig. 3 A, bottom), whereas the S100C/A11 protein levels only slightly increased by high Ca²⁺ (Fig. S3).

To determine the phosphorylation sites of S100C/A11, endogenous S100C/A11 protein was purified from ³²P-labeled HaCaT cells and degradated to peptides. As shown in Fig. 3 B (middle and bottom), only NH₂-terminal p1-1 and COOH-terminal p4 were phosphorylated in a totally high Ca²⁺-depending manner. When the synthesized peptides of S100C/A11 were incubated with cell extracts in vitro, NH₂-terminal pepA (amino acid residues 1–23) and COOH-terminal pepF (87–105), but not four other peptides covering the middle part of S100C/A11, were significantly phosphorylated by the extracts prepared from NHK and HaCaT cells cultivated in the high Ca²⁺ media (unpublished data). Thr and Ser residues were phosphorylated in pepA and pepF, respectively (Fig. 3 C). Finally, phosphorylation sites in p1-1 and p4 were determined as ¹⁰Thr and ⁹⁴Ser, respectively, by an amino acid sequencer (Fig. 3 D). In accordance with this, Ala-substituted forms at ¹⁰Thr of pepA and at ⁹⁴Ser of pepF were not phosphorylated by the cell extracts prepared after cultivating in the high Ca²⁺ media (unpublished data).

Increases in Ca²⁺ in the media resulted in partial translocation of endogenous S100C/A11 to nuclei of NHK and HaCaT cells as demonstrated by immunostaining (Fig. 4 A) and by Western blotting and autoradiography (Fig. 4 B). Preincubation with a membrane-permeable Ca²⁺ chelator, BAPTA-AM, inhibited the nuclear translocation of endogenous S100C/A11 by high Ca²⁺ (Fig. 4 A). An S100C/A11 variant protein lacking Ca²⁺-binding capacity was not translocated to nuclei by high Ca²⁺ (Fig. 4 C). Transfection of the wild-type S100C/A11-GFP construct before the shift to high Ca²⁺ resulted in nuclear translocation of the protein in HaCaT cells (Fig. 4 D). This nuclear translocation was not observed when ¹⁰Thr was substituted with Ala (Fig. 4 D) or when the NH₂-terminal 23 amino acids were deleted (unpublished data). ¹⁰Thr-to-Asp–substituted protein (TD) was observed in nuclei even in the low Ca²⁺ medium. On the contrary, ⁹⁴Ser-substituted forms behaved in a manner similar to the wild-type protein. A COOH-terminal truncated form of S100C/A11 lacking the residues from 87 to 105 was transferred to nuclei by the high Ca²⁺ (unpublished data). These results indicate that the phosphorylation of ¹⁰Thr, but not of ⁹⁴Ser, is critical for the Ca²⁺-induced nuclear localization of S100C/A11, and that binding of Ca²⁺ to S100C/A11 is necessary for the translocation.

Because S100C/A11 lacks the canonical NLS sequence, we screened for possible interacting protein(s) using a pepA(TD)-immobilized column (pepA with Asp at the 10th residue, a phosphorylated pepA equivalent). An ∼100-kD protein was identified in extracts of HaCaT cells as showing specific binding (Fig. 5 A, left). Sequencing of the eluted protein gave an amino acid sequence of VKLAKAGKANQGD, corresponding to the NH₂ terminus of human nucleolin. The binding of S100C/A11 and nucleolin was confirmed in vitro (Fig. 5 A, right; Fig. 5 B). The bound amounts of nucleolin depended on the amounts of S100C/A11 protein immobilized, and S100C/A11(TD), a ¹⁰Thr phosphorylation equivalent form, showed higher affinity to nucleolin than did the wild type (Fig. 5 B). Synthesized pepA having phosphothreonine at the 10th residue showed a higher nucleolin-binding capacity similar to pepA(TD) (Fig. S6). Only pepA, and not the other five partial peptides of S100C/A11, binds to nucleolin. Furthermore, neither GST-S100A2 nor GST-S100A6, the other members of S100 family, bound to nucleolin as assayed in vitro (unpublished data).

In the Ca²⁺-stimulated HaCaT cells, S100C/A11 protein that was phosphorylated and bound to nucleolin was predominantly found in the nuclei (Fig. 5 C). Among various GFP-conjugated ¹⁰Thr- or ⁹⁴Ser-substituted forms of S100C/A11 transfected to HaCaT cells, only ¹⁰Thr-to-Asp–substituted forms were coprecipitated with nucleolin in the low Ca²⁺ medium (Fig. 5 D). In the high Ca²⁺ medium, phosphorylation of S100C/A11 was observed when either ¹⁰Thr or ⁹⁴Ser was intact. Nucleolin was coprecipitated with ¹⁰Thr-intact forms in addition to ¹⁰Thr-to-Asp–substituted forms. BAPTA-AM inhibited the phosphorylation and interaction of wild-type S100C/A11 with nucleolin by high Ca²⁺ (Fig. 5 E). The S100C/A11 variant protein lacking Ca²⁺-binding capacity was neither phosphorylated nor coprecipitated with nucleolin by high Ca²⁺ (Fig. 5 E).

When small interfering RNA (siRNA) against nucleolin was applied to HaCaT cells, the protein levels were substantially reduced (Fig. 5 F, top), whereas the control siRNA against GAPDH showed no effects. Increase in Ca²⁺ concentration translocated endogenous S100C/A11 into nuclei in the siRNA against GAPDH-treated cells, but not in the siRNA against nucleolin-treated cells (Fig. 5 F, bottom). The necessity of nucleolin for the Ca²⁺-induced nuclear translocation of S100C/A11 was also confirmed in an in vitro translocation assay using permeabilized HaCaT cells by depleting nucleolin with a specific antibody (Fig. S5). These results indicate that (1) high Ca²⁺ treatment of HaCaT cells induces endogenous activity to phosphorylate S100C/A11 at ¹⁰Thr and ⁹⁴Ser, and eventually enhances the binding of S100C/A11 to nucleolin; (2) S100C/A11 binds to nucleolin through its NH₂-terminal region in a ¹⁰Thr phosphorylation–dependent manner; (3) the binding of S100C/A11 to nucleolin is essential for the translocation of S100C/A11 to nuclei; and (4) the binding of Ca²⁺ to S100C/A11 is necessary for the phosphorylation, binding to nucleolin, and nuclear translocation of S100C/A11.

The next question is how nuclear S100C/A11 exerts its inhibitory action on the growth of keratinocytes. We screened a number of possible candidates and found that p21^CIP1/WAF1 mRNA was remarkably induced in NHK cells either cultivated at 1.5 mM Ca²⁺ or exposed to TAT-S100C/A11 (Fig. 6 A). p16^INK4a was also induced, but to a moderate extent. Similar induction was observed in HaCaT cells in a time-dependent manner (Fig. 6 B).

To confirm the role of S100C/A11 on the induction of p21^CIP1/WAF1, anti-S100C/A11 antibody was introduced into HaCaT cells by the aid of Chariot or polyethyleneimine (PEI). The polyclonal antibody that we used was specific as shown in Fig. 6 C, and was demonstrated to recognize the NH₂-terminal pepA and COOH-terminal pepF of S100C/A11 (unpublished data). Binding of the introduced antibody to endogenous S100C/A11 was demonstrated by immunoprecipitation (Fig. 6 D). Introduction of the antibody inhibited the high Ca²⁺-induced phosphorylation of S100C/A11 (Fig. 6 E). Fab fragments of the anti-S100C/A11 antibody also reduced the phosphorylation of S100C/A11 in HaCaT cells (Fig. S8). The induction of p21^CIP1/WAF1 by TAT-S100C/A11 was dose-dependently abrogated by the antibody introduced into the cells (Fig. 6 F). The decrease in the amount of p21^CIP1/WAF1 protein due to the action of anti-S100C/A11 antibody was associated with the recovery of cdk2 activity as assayed in vitro (Fig. 6 F). Inhibition of cdk2 activity by S100C/A11 introduced into HaCaT cells was dose- and time-dependent. Finally, anti-p21^CIP1/WAF1 antibody (but not anti-p16^INK4a antibody) completely abrogated the growth inhibition by S100C/A11 as assayed for [³H]TdR incorporation or for cdk2 activity (Fig. 6, G and H), indicating that p21^CIP1/WAF1 is the principal effector protein of the S100C/A11-mediated growth inhibition of human keratinocytes. Specificity of anti-p21^CIP1/WAF1 and anti-p16^INK4a antibodies was confirmed as shown in Fig. 6 C.

Next, we addressed the question of how S100C/A11 in the nuclei induces the effector p21^CIP1/WAF1. Screening using Signal Transduction AntibodyArray™ (Hypromatrix) resulted in identification of Sp1 as a protein that binds to nucleolin, and the binding was inhibited by the addition of S100C/A11(TD) (Fig. 7 A). Anti-Sp2 antibody (but not anti-Sp3 antibody) was also on the membrane, but showed a negative result. Binding of Sp1 to nucleolin was confirmed by applying the extract of HaCaT cells onto GST-nucleolin beads (Fig. 7 B). Sp1 did not bind directly to S100C/A11.

To examine whether nucleolin binds to Sp1 in the cells, we performed a coprecipitation assay (Fig. 7 C). Sp1 and S100C/A11 were detected in the precipitates prepared using anti-nucleolin antibody, whereas nucleolin, but not S100C/A11, was detected in the precipitates prepared using anti-Sp1 antibody. These results indicate that binding among the proteins actually takes place in vivo, and that S100C/A11 does not bind to Sp1. In HaCaT cells cultivated in the presence of 1.5 mM Ca²⁺, nucleolin bound to Sp1, but not to S100C/A11. When the Ca²⁺ concentration was increased to 10 mM, S100C/A11 instead of Sp1 was detected in the immunoprecipitates prepared using anti-nucleolin antibody, and the bound S100C/A11 was phosphorylated. When the extract was treated with alkaline phosphatase in advance, the amount of S100C/A11 bound to nucleolin was dramatically reduced and the binding of Sp1 was recovered concomitantly. Immunoprecipitation with anti-Sp1 antibody confirmed that the stimulation with high Ca²⁺ liberated Sp1 from nucleolin.

Regions of nucleolin and Sp1 involved in the interaction were determined using recombinant partial proteins (Fig. 7 E). The COOH-terminal region of Sp1 (segment 3) including a zinc-finger domain was shown to be responsible for the binding to nucleolin (Fig. 7 D, top). In turn, Sp1 in the cell extract bound to segment 4 of nucleolin immobilized on a membrane (Fig. 7 D, middle). Recombinant Sp1 segment 3 also bound to the nucleolin segment 4, indicating direct interaction between the two proteins. S100C/A11 was recovered only from beads conjugated with segment 3 of nucleolin (Fig. 7 D, bottom). When segment 3 of nucleolin was introduced into HaCaT cells, the Ca²⁺-induced p21^CIP1/WAF1 protein level was decreased and the activity of cdk2 was recovered within 8 h, despite continuous cultivation at 10 mM Ca²⁺ (unpublished data). This is possibly due to a dominant-negative effect of the peptide through sequestering S100C/A11 by binding. The interacting regions of the proteins are depicted in Fig. 7 E.

Binding of nucleolin in cell extracts to Sp1 (segment 3) was dose-dependently inhibited by recombinant S100C/A11 (Fig. 7 F). S100C/A11(TD) showed a stronger effect, whereas S100C/A11(TA) did not affect the binding between nucleolin and Sp1 (segment 3).

Because Sp3 is also known to activate the p21^CIP1/WAF1 promoter (Prowse et al., 1997), we examined possible interaction with nucleolin and S100C/A11 and found that (1) Sp3 can bind to nucleolin segment 4 via its COOH-terminal zinc-finger domain; and (2) the binding of Sp3 to nucleolin was inhibited by S100C/A11 (unpublished data).

A gel shift assay was performed using GC-rich regions of the p21^CIP1/WAF1 promoter, the well-known Sp1/Sp3-binding sites (see Materials and methods). As shown in Fig. 7 G, exposure of NHK and HaCaT cells to high Ca²⁺ resulted in an increase in the intensity of the shifted bands by Sp1 and Sp3, which were further retarded by the addition of anti-Sp1 or -Sp3 antibody. Under similar experimental conditions in which HaCaT cells were cultivated in a high Ca²⁺ medium, addition of GST-nucleolin into the incubation mixtures inhibited the extent of the band shift as compared with GST (Fig. 7 G, right). This was possibly due to sequestration of Sp1 by exogenous nucleolin. The inhibition of the band shift by GST-nucleolin was abrogated by the ¹⁰Thr phosphorylation equivalent form S100C/A11(TD), whereas wild-type S100C/A11 and S100A2 showed no effect. We found that the interaction of the proteins were dose-dependent, and addition of an excess amount of wild-type S100C/A11 showed recovering effects similar to S100C/A11(TD) (unpublished data). These results indicate the competitive nature of the binding between S100C/A11 and Sp1/Sp3 to nucleolin. We performed a series of luciferase assays using a plasmid containing the p21^CIP1/WAF1 promoter region (Nakano et al., 1997) and obtained essentially the same results as those obtained by using the gel shift assay (unpublished data). Deletion of the Sp1/Sp3-binding sites in the promoter abrogated the induction of luciferase activity by S100C/A11.

The findings described in the previous paragraphs indicate that S100C/A11 is involved in growth inhibition of human keratinocytes induced by high Ca²⁺ (Fig. 8 A). Under low Ca²⁺ conditions, S100C/A11 is present mostly in the cytoplasm. The unphosphorylated S100C/A11 partly binds to actin fiber via its COOH-terminal region (unpublished data). An increase in intracellular Ca²⁺ results in activation of an as-yet-unidentified protein kinase(s) that phosphorylates S100C/A11 at ¹⁰Thr and ⁹⁴Ser. Phosphorylated S100C/A11 binds to nucleolin via its NH₂-terminal region and is translocated to nuclei. In the nuclei, S100C/A11 competes with Sp1 and Sp3 for binding to nucleolin, and the resulting free Sp1 and Sp3 induce p21^CIP1/WAF1.

A question that arises is to what extent the pathway is biologically relevant. To address this question, anti-S100C/A11 antibody was introduced into NHK and HaCaT cells by PEI and Chariot, respectively. The introduction efficiency was confirmed to be >70% by immunostaining. The introduced antibody inhibited the high Ca²⁺-induced phosphorylation of S100C/A11 (Fig. 6 E). High Ca²⁺ suppressed DNA synthesis to ∼17% of the control level, and the antibody recovered the DNA synthesis to 64 and 72% of the control level in NHK and HaCaT cells, respectively, indicating that S100C/A11 is a principal (if not sole) mediator of growth regulation (Fig. 8 B). The incomplete recovery may be either due to incomplete blockage of S100C/A11 function by the introduced antibody or due to the presence of some additional pathway mediating high Ca²⁺-induced growth inhibition.

S100C/A11 was found to be expressed at various levels in many human tissues (Fig. 8 C), among which skin showed a particularly high level of S100C/A11 expression. When normal human skin tissues were stained with anti-S100C/A11 antibody, S100C/A11 protein was detected in epidermal keratinocytes (Fig. 8 D). Nuclei of cells in the basal layer were consistently negative for S100C/A11 protein, whereas nuclei of cells in the suprabasal layers were positive. p21^CIP1/WAF1 protein was not detected in the basal layer, but appeared in suprabasal cell nuclei in which S100C/A11 was present. Among 21 nuclei counted in the basal layer, 20 and 1 nuclei were negative and positive for both S100C/A11 and p21^CIP1/WAF1, respectively. In the suprabasal layer, 89 nuclei were positive for S100C/A11 and p21^CIP1/WAF1, whereas 11 nuclei were positive for S100C/A11, but negative for p21^CIP1/WAF1. No S100C/A11-negative nuclei expressed p21^CIP1/WAF1. These results indicate that S100C/A11 not only mediates high Ca²⁺-induced growth inhibition of human keratinocytes in culture, but most likely plays a key role in the growth regulation in vivo as well.

We have demonstrated that an S100C/A11-mediated pathway plays a key role in high Ca²⁺-induced growth inhibition of human keratinocytes in culture. This finding has substantial biological significance for the following reasons: (1) high Ca²⁺ is a physiological inducer of growth arrest and differentiation of human keratinocytes in culture, and probably in vivo as well; (2) prior introduction of anti-S100C/A11 antibody resulted in recovery of the high Ca²⁺-induced growth inhibition of human keratinocytes by ∼70%, indicating that S100C/A11 is a key (if not the sole) mediator of growth regulation; and (3) immunostaining of the human epidermis for S100C/A11 and p21^CIP1/WAF1 showed a picture consistent with the model presented as a summary of the results of the present report using cultured human keratinocytes (Fig. 8 A).

Endogenous S100C/A11 was phosphorylated at ¹⁰Thr and ⁹⁴Ser in HaCaT cells exposed to high Ca²⁺ (Fig. 3 D), whereas only ¹⁰Thr was shown to be the major target of phosphorylation in confluent normal human fibroblasts (Sakaguchi et al., 2000). The phosphorylation of S100C/A11 at ¹⁰Thr stimulated binding to nucleolin via its NH₂-terminal region, this being a prerequisite for nuclear translocation of S100C/A11 (Fig. 5, C–F). In human keratinocytes, some of the cytoplasmic S100C/A11 proteins were colocalized with actin fibers. We found that S100C/A11 binds to actin through the COOH-terminal region and that phosphorylation of ⁹⁴Ser reduced the affinity (unpublished data). The liberation of S100C/A11 from actin may facilitate the association with nucleolin and the eventual nuclear translocation of S100C/A11.

Ca²⁺ binds to S100C/A11 and induces its conformational changes (Rety et al., 2000). The binding of Ca²⁺ to S100C/A11 appears to be critical for the S100C/A11-mediated growth inhibition because interference of the binding either by amino acid substitution or with a membrane-permeable Ca²⁺ chelator (BAPTA-AM) inhibited phosphorylation, binding to nucleolin, and nuclear translocation of S100C/A11 (Fig. 4, A and C; Fig. 5 E). Cellular kinase activity for S100C/A11 as assayed in vitro using S100C/A11 or pepA was induced by high Ca²⁺ in NHK and HaCaT cells. The kinase activity was detected in a cytosolic-soluble fraction, but not in an insoluble or nuclear fraction, of HaCaT cells exposed to high Ca²⁺. When the cell extract was applied onto a cation exchange column, the activity was eluted with 160–180 mM KCl (unpublished data). Further characterization of the kinase(s) is now in progress in our laboratory.

Nucleolin was first identified as a protein involved in ribosomal assembly and maturation, but was later found to have diverse biological functions (Srivastava and Pollard, 1999). Although nucleolin is mainly localized in the nucleus, it shuttles between the nucleus and cytoplasm and is even present on the cell surface. When we microinjected wheat germ hemagglutinin into HaCaT cells, which is known to generally inhibit nuclear import of proteins (Finlay and Forbes, 1990), nucleolin was diffusely distributed in the cells (unpublished data). Recently, Shibata et al. (2002) reported that nucleolin mediates a signal from the cell surface to the nucleus. Nucleolin inhibited transcriptional activation by A-Myb through specific binding (Ying et al., 2000). Heat shock or irradiation was shown to mobilize nucleolin from nucleoli to promote the formation of a nucleolin–p53 complex (Daniely et al., 2002). In the present work, we showed that S100C/A11 and Sp1/Sp3 compete for binding to nucleolin. This is the first report of nucleolin providing a site for switching transcriptional activation through protein–protein interaction. Unexpectedly, it was found that S100C/A11 and Sp1/Sp3 bind to different regions of nucleolin (Fig. 7 E). Binding of S100C/A11 may change the conformation of nucleolin so that it can no longer hold Sp1 and Sp3. Although nucleolin is notoriously sticky, the interaction with Sp1/3 and S100C/A11 is unlikely to be nonspecific because (1) only the specific regions of the respective proteins show the binding capacity; and (2) other members of S100, such as S100A2 and S100A6, did not bind to nucleolin.

Sp1/krupple-like factors act as ubiquitous and tissue-restricted transactivators in many different contexts. Sp1 and Sp3 were shown to activate the p21^CIP1/WAF1 promoter (Prowse et al., 1997). In the gel shift assay, the shifted bands shown in Fig. 7 G were further retarded by incubation with anti-Sp1 and anti-Sp3 antibody. We performed a series of luciferase assays using the plasmid containing the human p21^CIP1/WAF1 promoter sequence of 2.4 kb (Nakano et al., 1997). Application of TAT-S100C/A11 induced the activity of the p21^CIP1/WAF1 promoter by more than 10-fold in HaCaT cells in an Sp1/Sp3 site-dependent manner (unpublished data). This induction was abrogated by pretreatment of the cells with TAT-nucleolin segment 3, but not TAT-nucleolin segment 4. These results indicate that p21^CIP1/WAF1, the effector protein of S100C/A11-mediated growth inhibition (Fig. 6 G), is transcriptionally activated by Sp1 and Sp3 in human keratinocytes when exposed to high Ca²⁺. This induction must be p53-independent because HaCaT cells are known to have only mutated p53 genes (Lehman et al., 1993).

There remains little doubt that p21^CIP1/WAF1, when expressed, inhibits the growth of keratinocytes through inactivating Cdks (Gartel et al., 1996). The results of immunostaining for p21^CIP1/WAF1 (Fig. 8 D) suggest that p21^CIP1/WAF1 is involved in growth arrest and differentiation of human epidermal keratinocytes in vivo. Various differentiation factors for human keratinocytes, including high Ca²⁺ (Todd and Reynolds, 1998), TGFβ (Datto et al., 1995), and TPA (Todd and Reynolds, 1998), are known to induce p21^CIP1/WAF1.

In the human epidermis, p21^CIP1/WAF1 protein was detected in the lower and middle suprabasal layers, but not in the uppermost layers (Fig. 8 D; Ponten et al., 1995). This seems reasonable because overexpressed p21^CIP1/WAF1 was shown to inhibit terminal differentiation of mouse keratinocytes at later stages (Di Cunto et al., 1998). Alternatively, p21^CIP1/WAF1 may simply not be needed in irreversibly arrested cells.

TGFβ transduces signals via Smad proteins. In HaCaT cells treated with TGFβ, Smad3 and Smad4 were found to cooperate with Sp1 for transcriptional activation of the p21^CIP1/WAF1 promoter (Pardali et al., 2000). When we treated HaCaT and NHK cells with TGFβ, S100C/A11 was translocated to nuclei (unpublished data). Thus, S100C/A11 may be a partial converging point of high Ca²⁺ and TGFβ.

The proposed S100C/A11-mediated pathway for Ca²⁺-induced growth arrest (Fig. 8 A) has partly revealed the thus-far-unknown link from the initial events triggered by high Ca²⁺ to the induction of Cdk inhibitors, particularly p21^CIP1/WAF1. However, we still have little information on how S100C/A11 is phosphorylated upon stimulation by high Ca²⁺. The specific phosphorylation observed in the present paper in terms of culture conditions and substrates (Fig. 3, B and D) provides an opportunity for the identification of the involved protein kinase(s). Furthermore, the function of S100C/A11 for liberation of Sp1 and Sp3 in nuclei was born by residues 1–23 of the NH₂-terminal peptide. Considering the ubiquitous expressions of nucleolin and Sp1/Sp3 and the large variety of promoters that are regulated by Sp1 and Sp3, further narrowing down of the functional structural domain of the peptide and screening for chemicals that have similar structure may be a promising venture eventually leading to pharmacologically useful substances.

NHK cells (Kurabo Industries, Ltd.) were cultured in Defined Keratinocyte-SFM (GIBCO BRL) with Defined Keratinocyte-SFM Growth Supplement (GIBCO BRL). HaCaT cells, a gift from Dr. Fusenig (German Cancer Research Center, Heidelberg, Germany), were cultured in DME with 10% FBS. For monitoring DNA synthesis, 1 μCi/ml tritiated thymidine ([³H]TdR; American Radiolabeled Chemicals) was added to the cultures 1 h before cell harvest. An anti-BrdU antibody (Neomarkers) was used to stain BrdU-incorporated cells. Subfractionation of cells was performed as described previously (Lindeman et al., 1997). Imaging analysis of the intracellular Ca²⁺ level was performed as reported previously (Yamamoto et al., 2000).

To determine the phosphorylation sites of endogenous S100C/A11, the protein was collected from HaCaT cells cultivated in the low or high Ca²⁺ medium using an affinity column with anti-S100C/A11 antibody. After further purification by PAGE, the protein was cleaved with cyanogen bromide and the peptides were fractionated by HPLC. Four peptides (Fig. 3 D) were identified by partial sequencing. pep1 was further cleaved with endoproteinase Asp-N (Roche). The purified p1-1 and p4 were subjected to NH₂-terminal sequencing on a peptide sequencer (model 491; Applied Biosystems). Phosphorylated amino acid residues were run through the column, and thus disappeared from the expected position of unphosphorylated residues.

Peptides and S100C/A11 protein with relatively low molecular sizes were introduced into cells by the aid of the protein transduction domain of TAT protein (YGRKKRRQRRR) from human immunodeficiency virus (Schwarze et al., 1999). A Chariot kit (active motif) and cationic PEI-conjugated S100C/A11 antibody were used to introduce antibodies into HaCaT and NHK cells, respectively. To functionally inactivate proteins within the cells, rabbit anti–human S100C/A11, mouse anti–human p16^INK4a (Ab-4, clone 16P04; Neomarkers), and mouse anti-p21^CIP1/WAF1 (Ab-11, clone CP74: Neomarkers) antibodies were used. The anti-S100C/A11 antibody detects only a single band in Western blot analysis using cells irrespective to Ca²⁺ levels in the media (Fig. 6 C). Efficacy for neutralization of the target proteins p16^INK4a and p21^CIP1/WAF1 by the antibodies was confirmed by in vitro phosphorylation assay.

Antibodies and plasmid constructs used in the present work are listed in the Online supplemental material section.

An optimal target site of nucleolin mRNA for siRNA was determined as 5′-GACAGTGATGAAGAGGAGG-3′ by a program provided by QIAGEN, and the synthetic siRNA was purchased from the same company. siRNA against GAPDH mRNA (Ambion) was used as a control. The siRNA was transfected to the logarithmically growing HaCaT cells using LipofectAMINE™ 2000 (Invitrogen).

An in vitro cdk2 kinase assay was performed using immunoprecipitates with anti–human cyclin E antibody and histone H1 as a substrate as described previously (Ohashi et al., 1999).

Signal Transduction AntibodyArray™ (Hypromatrix) was used under the conditions indicated by the manufacturer. In brief, the antibody membrane was incubated with 1-mg HaCaT cell extracts, followed by the application of 10-μg full-length recombinant proteins (GST, GST-S100C/A11, GST-nucleolin, or GST-nucleolin + S100C/A11(TD)). After washing, the bound proteins were detected using HRP-conjugated anti-GST antibody (Santa Cruz Biotechnology, Inc.).

A gel mobility shift assay was performed as described by Nakano et al. (1997). We used GC-rich regions located between 120 and 94 bp (A), 93 and 74 bp (B), and 73 and 47 bp (C) of the human p21^CIP1/WAF1 promoter as probes. The ³²P-labeled probes (A + B + C; 10⁴ cpm) were mixed with 4 μg crude nuclear extracts of NHK or HaCaT cells, incubated for 20 min at RT, and electrophoresed in a 5% polyacrylamide gel under nondenaturing conditions. For a super-shift experiment, 1 μg mouse anti–human Sp1 antibody (Santa Cruz Biotechnology, Inc.) or rabbit anti–human Sp3 antibody (Santa Cruz Biotechnology, Inc.) was added to the reaction mixture.

For GFP-labeled vital cells, images were acquired by a fluorescent microscope (IX71–22FL/PH; CCD camera, DP50; objective lens, LCPlan F1 40×; Olympus) and processed using Adobe Photoshop^® 6.0. For fixed cells labeled with FITC, Alexa^® 488, or TRITC, a laser-scanning microscope (Axioplan 2; objective lens, Plan-Apochromat^® 63× 1.4 oil DC and Plan-Neofluar^® 40×/0.75; Carl Zeiss MicroImaging, Inc.) was used.

Fig. S1 shows dose-dependent inhibition of DNA synthesis by TAT-S100C. Fig. S2 shows expression of variant S100C/A11 proteins. Fig. S3 shows effects of high Ca²⁺ on endogenous S100C/A11 levels. Fig. S4 depicts intracellular localization of nucleolin. Fig. S5 shows necessity of nucleolin for nuclear translocation of S100C/A11. Fig. S6 depicts binding of S100C/A11 pepA with nucleolin. Fig. S7 shows proteins recovered by protein G from HaCaT cells applied with anti-S100C antibody or IgG. Fig. S8 shows Fab of anti-S100C/A11 antibody and reduced phosphorylation level of S100C/A11 induced by high Ca²⁺. Fig. S9 depicts induction of p21^CIP1/WAF1 by variant S100C/A11 proteins. Fig. S10 shows interaction among nucleolin, Sp1, and S100C/A11. Fig. S11 shows competition of S100C/A11 peptides with Sp1 and Sp3 for binding to nucleolin. Fig. S12 shows the dominant-negative effect of nucleolin Seg-3 for the induction of p21^CIP1/WAF1 by TAT-S100C/A11. The supplemental material is available.

We are indebted to Dr. T. Sakai (Kyoto Prefectural University of Medicine, Kyoto, Japan) for providing a plasmid.

This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (14370260 to N. Huh) and the Japan Society for the Promotion of Science (0104310 to M. Sakaguchi).