Transcriptional Activation of the p21 WAF1,CIP1,SDI1 Gene by Interleukin-6 Type Cytokines

The cyclin-dependent kinase inhibitor p21 WAF1,CIP1,SDI1 plays a critical role in cell differentiation, and it has been shown to confer resistance to apoptosis. Based on this, and on evidence that activation of the gp130/signal transducer and activator of transcription (STAT) signal transduction pathway by interleukin (IL)-6 type cytokines promotes differentiation and prevents apoptosis in osteoblastic cells, we have investigated the possibility that p21 is a downstream effector of this signaling pathway in osteoblasts. We report that either oncostatin M (OSM) or IL-6 plus soluble IL-6 receptor increased the levels of p21 mRNA and protein in the osteoblast-like human osteosarcoma cell line MG63 and stimulated the activity of a 2.4-kilobase pair segment of the human p21 gene promoter. Further, nuclear extracts from cytokine-stimulated MG63 cells formed protein-DNA complexes with a 19-base pair nucleotide fragment of the p21 promoter containing a single STAT response element. The identity of the binding proteins as Stat3 and Stat1 was demonstrated with specific antibodies. In addition, and in support of a mediating role of STATs in the activation of the p21 promoter, overexpression of Stat3 potentiated the cytokine effect on the p21 promoter; whereas a dominant negative Stat3, or a mutation of the STAT response element on the promoter, significantly reduced the cytokine effect. Finally, antisense oligonucleotides complementary to p21 mRNA inhibited OSM-induced stimulation of alkaline phosphatase expression and antagonized the protective effect of OSM on anti-Fas-induced apoptosis. These results demonstrate that p21 is a downstream effector of gp130/Stat3 activation and a critical mediator of the pro-differentiating and anti-apoptotic effects of IL-6 type cytokines on human osteoblastic cells.

The cyclin-dependent kinase inhibitor p21 WAF1,CIP1,SDI1 plays a critical role in cell differentiation, and it has been shown to confer resistance to apoptosis. Based on this, and on evidence that activation of the gp130/signal transducer and activator of transcription (STAT) signal transduction pathway by interleukin (IL)-6 type cytokines promotes differentiation and prevents apoptosis in osteoblastic cells, we have investigated the possibility that p21 is a downstream effector of this signaling pathway in osteoblasts. We report that either oncostatin M (OSM) or IL-6 plus soluble IL-6 receptor increased the levels of p21 mRNA and protein in the osteoblast-like human osteosarcoma cell line MG63 and stimulated the activity of a 2.4-kilobase pair segment of the human p21 gene promoter. Further, nuclear extracts from cytokinestimulated MG63 cells formed protein-DNA complexes with a 19-base pair nucleotide fragment of the p21 promoter containing a single STAT response element. The identity of the binding proteins as Stat3 and Stat1 was demonstrated with specific antibodies. In addition, and in support of a mediating role of STATs in the activation of the p21 promoter, overexpression of Stat3 potentiated the cytokine effect on the p21 promoter; whereas a dominant negative Stat3, or a mutation of the STAT response element on the promoter, significantly reduced the cytokine effect. Finally, antisense oligonucleotides complementary to p21 mRNA inhibited OSM-induced stimulation of alkaline phosphatase expression and antagonized the protective effect of OSM on anti-Fasinduced apoptosis. These results demonstrate that p21 is a downstream effector of gp130/Stat3 activation and a critical mediator of the pro-differentiating and anti-apoptotic effects of IL-6 type cytokines on human osteoblastic cells.
Fine orchestration of the activity of cyclins, Cdk, 1 and inhib-itors of such kinases governs the progression of cells from one phase of their division cycle to the next. Increased expression of Cdk inhibitors is now recognized as a general mechanism for the arrest of cell division (1,2). One of these Cdk inhibitors, designated as p21 WAF1,CIP1,SDI1 because of its original identification by three independent groups as wild type p53 activated fragment 1 (WAF1) (3), Cdk-interacting protein 1 (CIP1) (4), or senescence cell-derived inhibitor (SDI1) (5), binds and inactivates several cyclin/Cdk complexes including the ones involved in the transition between G 1 and S phases, and inhibits the replication factor PCNA (proliferating cell nuclear antigen) (6). To date, p21 has been implicated in three different processes: DNA damage repair, differentiation, and apoptosis. In the former, p21 gene expression is stimulated by the tumor suppressor gene p53, and the protein arrests cells in the G 1 phase in order to allow the DNA damage repair process to be performed before cells reenter the S phase (7,8). In the latter two processes, p21 expression increases independently of p53. In this case, the resulting halt of cell division allows cells to progress along their differentiation pathway (9 -14). Interestingly, in differentiating muscle cells, neuroblastoma, and melanoma cells, p21 also protects against programmed cell death (15)(16)(17).
It is now established that cytokines such as IL-6 can influence bone resorption and formation by regulating the production of osteoclasts and osteoblasts (reviewed in Ref. 18). Indeed, IL-6, IL-11, LIF, and OSM are produced by cells of the stromal/ osteoblastic lineage in response to local factors and systemic hormones, and they stimulate osteoclast differentiation from hematopoietic precursors. In addition, they modulate the rate of proliferation and promote the differentiation of osteoblast progenitors and committed osteoblastic cells from rodent and human origin (18 -21). Moreover, IL-6 type cytokines suppress osteoblastic cell apoptosis induced by serum withdrawal, glucocorticoids, tumor necrosis factor, or the Fas pathway (22,23).
The receptors for all these cytokines share a common signaltransducing component known as gp130 (24). Ligand-induced dimerization of gp130 initiates intracellular signaling by activating members of the Janus kinase family of tyrosine kinases (25). This step induces the recruitment, tyrosine phosphoryla-tion, and nuclear translocation of STATs, which activate transcription of cytokine-responsive genes (26,27).
Because of evidence that p21 WAF1,CIP1,SDI1 is induced during cell differentiation and confers resistance to apoptosis in some cell types, we have investigated whether this gene is regulated by IL-6 type cytokines and whether it plays a role in their actions on osteoblasts. For these studies, we utilized an established human osteosarcoma cell line, MG63, which exhibits many features of normal osteoblastic cells and has been used extensively as a model of this cell type (28). MG63 cells express high levels of the OSM receptor ␤ and low levels of the ligand binding (␣) subunit of the IL-6 receptor (29); hence, the gp130 signal pathway can be activated by OSM or by the combination of IL-6 with sIL-6R. We have previously reported that such activation leads to cell cycle arrest and a simultaneous increase in the expression of the phenotypic marker AP (19), and prevents Fas-induced apoptosis (22). Here, we present evidence that activation of the gp130/STAT signal transduction pathway by IL-6 type cytokines leads to the transcriptional activation of the p21 WAF1,CIP1,SDI1 gene, and that p21 is a critical mediator of the differentiation-promoting and anti-apoptotic actions of these cytokines on osteoblastic cells.
Cell Culture Conditions-Human osteosarcoma MG63 cells (32) were cultured in phenol red-free MEM supplemented with 10% FBS, 100 units/ml penicillin, and 100 g/ml streptomycin. Cultures were kept in a humidified atmosphere of 5% CO 2 in air at 37°C.
Western Blot Analysis-Cells were lysed in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10 mM NaF, 1 mM sodium orthovanadate, 5 g/ml leupeptin, 0.14 unit/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, and 1% Triton X-100. Insoluble material was pelleted in a microcentrifuge at 10,000 ϫ g for 10 min. Protein lysates were dissolved in buffer for protein electrophoresis, separated on SDS-polyacrylamide gels, and electrotransferred to polyvinylidene difluoride. Membranes were blocked for 1 h at room temperature in PBS containing 0.1% Tween 20 and 5% nonfat dry milk, and subsequently incubated overnight at 4°C with an antibody to either p21, p53, or ␤-actin, followed by incubation for 1 h with the corresponding secondary antibody conjugated with HRP. Blots were developed by ECL, according to the manufacturer's recommendations. Quantification of the intensity of the bands in the autoradiograms was performed by laser densitometry.
RNA Extraction and Northern Blot Analysis-Total cellular RNA was isolated from semiconfluent control or cytokine-treated cultures, and polyadenylated RNA was selected as described previously (29). Messenger RNA was separated by electrophoresis in 1% agarose formaldehyde gels, transferred to nylon membranes, and fixed by heating at 80°C under vacuum for 2 h. Blots were probed with radiolabeled cDNAs for p21 or for actin, and analyzed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays were performed essentially as described elsewhere (19). Eighty to 90% confluent cultures of MG63 cells were maintained in serum-free medium for 2 h and stimulated with 50 ng/ml OSM for 15 min at 37°C. Cells were rinsed once with PBS, and nuclear extracts were prepared. Cells (4 -7 ϫ 10 7 ) were washed with hypotonic buffer (10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol) and were lysed for 10 min on ice in 30 l/10 7 cells of the same buffer containing 0.1% Nonidet P-40. Lysates were centrifuged at 10,000 ϫ g at 4°C for 10 min. Pelleted nuclei were resuspended in 30 l/10 7 cells of lysis buffer (20 mM Hepes, pH 7.9, 420 mM NaCl, 1 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, and 25% glycerol) and were incubated at 4°C for 15 min. Lysed nuclei were dispersed and centrifuged at 10,000 ϫ g at 4°C for 10 min. Supernatants were collected, snap frozen, and stored at Ϫ70°C. A synthetic double-stranded oligonucleotide corresponding to the most proximal Stat binding element (p21-SIE1: -640 nucleotides from the TATA promoter site) present in the p21 promoter (5Ј-GATCTCCTTC-CCGGAAGCA-3Ј) (3,33) was end-labeled using [␥-32 P]ATP and T4 polynucleotide kinase. The probe (1 ng) was incubated for 20 min at room temperature with 10 g of nuclear proteins in a solution containing 50 g/ml double-stranded salmon sperm DNA, 6% glycerol, 10 mM Hepes, pH 7.5, 80 mM KCl, 1 mM EDTA, and 1 mM EGTA, in the absence or presence of 100 ϫ molar excess unlabeled SIE1 wild type oligonucleotide or unlabeled mutated SIE1 (SIE1M: 5Ј-GATCTCCAAGCTT-GAAGCA-3Ј). Supershift experiments were performed by incubating the nuclear proteins with either non-immune IgG, anti-Stat1␣ antibody, or anti-Stat3 antibody, for 10 min at room temperature prior to the addition of the labeled probe. Transient Transfections, Assay of Reporter Gene Activity, and Oligonucleotide-directed Mutagenesis-Transient transfections of MG63 cells were carried out in 12-well culture plates using SuperFect transfection reagent (Qiagen) and the dual-luciferase reporter assay system (Promega). Cells were transfected with 2.5-3 g of a p21-luc reporter construct containing a 2.4-kb segment of the human p21 gene promoter (or an identical segment in which the putative STAT binding segment was mutated as described below) inserted into the firefly luciferase reporter gene, pGL2-basic (3,11), together with 3 ng of the co-reporter vector containing the Renilla luciferase gene and the CMV immediate early enhancer/promoter region (pRL-CMV). Reporter plasmids were cotransfected with 0.5 g of the expression plasmid for Stat3 or the empty vector (30). Forty-eight hours after transfection, cells were treated with OSM (20 ng/ml) for the indicated times, and dual luciferase activity was analyzed in cell lysates according to the manufacturer's instructions. Light intensity was measured with a Turner luminometer, and firefly luciferase activity was normalized versus Renilla luciferase activity. For oligonucleotide-directed mutagenesis, a 30-base oligonucleotide was synthesized as the reverse complement of the p21 sense strand containing a mutation in the SIE1 sequence (Ϫ660 nucleotides from the TATA promoter site: 5Ј-GTCACATGCTTCAAGCTTGGAGG-GAATTGG-3Ј). This was used to anneal and extend the wild type single-stranded template generated from p21-luc, a plasmid containing a 2.4-kb HindIII segment of the human p21 gene promoter inserted into the firefly luciferase reporter gene, pGL2-basic (11). The mutation incorporated a HindIII restriction site that was used for the selection of mutated plasmids, and it was confirmed by sequencing (34).
Antisense Oligonucleotides-Phosphorothioate oligodeoxynucleotides (15.7 M) and Lipofectin (3 ϫ 10 Ϫ4 mg/ml) were incubated at 37°C for 15 min. The mixture was diluted with serum-containing medium to a final concentration of 1 M oligonucleotides, and added to the cells. Antisense oligonucleotides were based on the p21 WAF1,CIP1,SDI1 coding sequence. AS-IC is complementary to the region around the initiation codon (5Ј-TCC CCA GCC GGT TCT GAC AT-3Ј), AS-MID is complementary to the middle of the coding region (5Ј-CCT CCA GTG GTG TCT CGG TG-3Ј), AS-3Ј is complementary to the 3Ј end (5Ј-TGT CAT GCT GGT CTG CCG CC-3Ј), and the control sense oligonucleotide S-IC is complementary to the AS-IC (5Ј-ATG TCA GAA CCG GCT GGG GA-3Ј).
AP Activity-AP activity was determined as previously reported (19). MG63 cells were suspended in medium containing 10% FBS and cultured in 96-well plates to 85-90% confluence. Medium was removed and replaced with fresh medium containing 5% FBS, and cells were cultured with or without 20 ng/ml OSM in the absence or presence of 1 M phosphorothioate oligonucleotides for 3 days (8 replicas per condition). Media were removed and cells were fixed in 10% formalin PBS buffer. The amount of cells per well was estimated using an assay based on the reduction of the tetrazolium derivative, MTS, coupled to LDH, using a microtiter plate reader. Monolayers were washed once with distilled water and three times with 0. mg/ml MTS, and 2.572 units/ml diaphorase, was added to each well, and readings at 490 nm and at 750 nm were taken at 0, 5, 10, 20, 30, 45, 60, 90, and 120 min, in a plate reader using Time Management Software (Dynatech Laboratories, Chantilly, VA). After MTS/LDH detection, cells were washed three times with 20 mM HEPES, 150 mM NaCl, and 1 mM MgCl 2 , pH 7.4. Subsequently, 75 l of AP buffer and 75 l of AP substrate were added to each well, and readings at 410 and 750 nm were taken at the same time points. Results are expressed as ratios of AP rate (OD/min) to MTS/LDH rate (OD/min).
Apoptosis-Apoptosis of human osteoblast-like MG63 cells was induced as described previously (22). Cells were cultured for 24 h in MEM containing 10% FBS in the absence or presence of 0.1 nM ␥-IFN, alone or in combination with 20 ng/ml OSM. The medium was then replaced with 10% FBS-MEM containing 10 g/ml anti-Fas antibody with or without OSM. Cells were cultured for an additional 16 h, and apoptotic cells were quantified by trypan blue staining. Nonadherent cells were combined with adherent cells released from the culture dish using trypsin-EDTA, and resuspended in PBS. Subsequently, 0.4% trypan blue was added, and the percentage of cells exhibiting both nuclear and cytoplasmic staining was determined using a hemocytometer. At least 200 cells per condition were counted. We have previously demonstrated that the percentage of cells exhibiting trypan blue staining correlates with the percentage of TUNEL-labeled cells, indicating that apoptotic osteoblasts could be reliably quantified by either method (22).
Statistical Analysis-For the statistical analysis of the data of Fig. 7, standardized means and standard errors were calculated using the variance for the ratio of two random variables (35). Approximate ninetyfive percent confidence intervals for the standardized means of the OSM-treated groups were calculated as the standardized mean Ϯ 2 standard errors. Confidence bounds with a lower limit greater than 1 were considered to denote a significant (p Ͻ 0.05) increase over the corresponding control group. For the analysis of the data regarding the effects of OSM on AP in the presence of oligonucleotides (Fig. 8B), standardized OSM AP/MTS values were calculated by dividing each replica of the OSM-treated group by the mean of the corresponding basal group. These values were analyzed by one-way ANOVA, and, subsequently, each oligonucleotide group was compared with the group without oligonucleotide using the Dunnet's test and an experiment-wise significance level of 0.05. In addition, an analysis using Dunnet's test to compare all antisense oligonucleotide groups to the group treated with sense oligonucleotide was also conducted. Data of Fig. 8C were analyzed by one-way ANOVA, and Dunnet's test was utilized to estimate the level of significance of differences between means.

IL-6 Type Cytokines
Increase the Protein Levels of p21 in a p53-independent Fashion-MG63 cells cultured in the presence of OSM exhibited an increase in the levels of p21 protein, as detected by Western blot analysis (Fig. 1A). This effect was evident as early as 2 h of treatment and remained present for at least 24 h. By 48 h, p21 levels had returned to base-line levels. Similar increases in p21 protein expression were found when IL-6 in combination with sIL-6R was used in these experiments instead of OSM (Fig. 1B). However, there was no change in the levels of p53 protein in cells treated with OSM (Fig. 1A). Consistent with the results of the Western blot analysis, p21 protein was detected by immunohistochemistry in MG63 cells maintained under basal conditions and OSM increased the staining (Fig. 2, upper panels). Nuclear staining was abolished when preparations were incubated with the anti-p21 antibody together with the peptide to which the antibody was raised (lower panels).
Because of evidence that some agents that increase p21 protein levels do so by increasing the half-life of the protein (36), we investigated whether OSM influences p21 protein stability. For this purpose, cells were treated with OSM for 6 h, and cycloheximide was subsequently added to stop any further protein synthesis; and cell lysates were obtained from control and treated cultures at different times after cycloheximide addition. The rate of decay of p21 was the same in OSM-treated cells and untreated cells, indicating that OSM does not change the stability of the p21 protein (Fig. 3).
gp130 Activation Increases Steady State p21 mRNA Lev-els-We next examined p21 mRNA expression in control and OSM-treated MG63 cells. As shown in Fig. 4A, OSM increased the abundance of p21 mRNA as early as 1 h after treatment. The effect was maximal at 6 h, decreased at 24 h, and returned to control levels by 48 h. To investigate whether de novo protein synthesis was required, we examined the regulation of p21 mRNA in the presence of cycloheximide. In this experiment, cells were preincubated with cycloheximide for 30 min, and then treated with OSM for 6 h. Cycloheximide did not block the effect of OSM. In fact, both basal and OSM-induced p21 levels were significantly increased in the presence of cycloheximide (Fig. 4B), suggesting that p21 mRNA levels may be posttranscriptionally regulated by labile proteins. The increase of steady state mRNA levels could be due either to an enhanced rate of transcription, to stabilization of previously transcribed mRNA, or to a combination of both mechanisms. To determine whether gp130 activation affected the half-life of the p21 transcript, transcription was blocked by adding actinomycin D to control or OSM-treated cells. Total mRNA was extracted at different times (15 min to 3 h) after addition of actinomycin D, and the amount of p21 mRNA was determined by Northern blot analysis (Fig. 5). The rate of decay of p21 mRNA transcripts in OSM-treated cells was similar to that seen in control cells, suggesting that the stability of p21 FIG. 1. gp130 activation increases p21 protein expression in a p53-independent manner. MG63 cells were cultured in the absence or presence of 20 ng/ml OSM or 20 ng/ml IL-6 in combination with 1 g/ml sIL-6R during the indicated times. Subsequently, cells were lysed, and proteins separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. Western blotting was performed sequentially with goat polyclonal anti-p21 antibody, mouse monoclonal anti-p53 antibody and anti-␤-actin antibody. Blots were incubated with the corresponding secondary antibody conjugated with HRP and developed by ECL. mRNA is not affected by gp130 activation; hence, the OSMinduced increase in p21 mRNA could result from transcriptional activation.
gp130 Activation Increases the Transcriptional Activity of the p21 Promoter, and Stat3 Potentiates This Effect-To determine whether gp130 activation affected the transcriptional activity of the p21 promoter, MG63 cells were transiently transfected with a 2.4-kb fragment of the p21 promoter inserted upstream from the luciferase reporter gene. Cells were then treated with OSM and luciferase activity was determined at different time points following the addition of the cytokine (Fig. 6A). OSM increased luciferase activity, with a maximal effect after 6 h of treatment. The effect of OSM was greater in cells transfected with the promoter/luciferase construct together with a Stat3 expression plasmid. The luciferase activity did not change with time either in the absence or presence of exogenous Stat3 in cells not treated with OSM (not shown).
Because of the above findings and the fact that the 2.4-kb fragment of the p21 promoter used in our experiments contains a potential SBE designated p21-SIE1 (3, 33), we eliminated this element by oligonucleotide-directed mutagenesis and determined the impact of the mutation on the ability of OSM to stimulate the activity of the promoter. Cells were transfected with either the wild type promoter-luciferase construct (p21luc) or with the mutated construct (p21-SIEM-luc). Elimination of the SBE significantly reduced the transcriptional response of the promoter to OSM stimulation (Fig. 6B), indicating that this cis-acting element played a major role in the OSM-induced activation of the promoter.
Identification of the p21-SBE-binding Proteins-Nuclear extracts from cells untreated or treated with OSM for 15 min were obtained, and electrophoretic mobility shift assays were performed using p21-SIE1 as a probe. Nuclear extracts from cells treated with OSM contained a protein(s) that bound the SBE probe and retarded its mobility (Fig. 7A). This binding was effectively competed by a 100-fold molar excess of unlabeled wild type p21-SIE1 oligonucleotide but not by a mutated p21-SIE1 oligonucleotide. An antibody to Stat3, but not a non-FIG. 2. gp130 activation increases p21 protein expression. MG63 cells plated on two-well chamber slides were cultured in the absence (Control) or presence (OSM) of 20 ng/ml OSM for 6 h. Cells were fixed with cold 100% methanol and stained with anti-p21 antibody (upper panels) or with anti-p21 antibody plus a 10-fold excess of the peptide to which the antibody was generated (lower panels). Original magnification, ϫ400.
immune IgG control, reduced the mobility of the protein-DNA complex, indicating the presence of Stat3. An antibody to Stat1␣ did not affect the mobility of the protein-DNA complex; however, it reduced the intensity of the band.
To establish the role of Stat3 in the OSM-induced stimulation of the p21 promoter, we utilized a dominant negative form of the Stat3 protein, generated by mutation of tyrosine 705 to phenylalanine (31). Transfection of MG63 cells with an expression plasmid for dnStat3 significantly reduced the OSM-induced increase in transcriptional activity of the wild type p21 promoter (Fig. 7B). For the purpose of comparison, the dnStat3 expression construct was also co-transfected with a promoterluciferase construct consisting of the SBE from the c-Fos promoter linked to the herpesvirus thymidine kinase promoter (pSIE-TK-luc) (Fig. 7B). The dnStat3 inhibited the OSM-stimulated activity of this construct to the same extent as the p21 promoter, confirming that binding of activated Stat3 to p21-SIE1 increases the transcriptional activity of the p21 promoter.
The Increases in p21 Protein Are Required for the Pro-differentiating and Anti-apoptotic Effects of gp130 Activation on Osteoblasts-The relevance of the stimulation of p21 gene expression by IL-6 type cytokines to their biologic effects was investigated using antisense oligodeoxynucleotides to block the synthesis of p21 protein. As shown by the results of the Western blot analysis depicted in Fig. 8A, the OSM-induced increase in p21 protein levels was largely abrogated in cells exposed to antisense oligonucleotides complementary to a region encompassing the initiation codon (AS-IC). Moreover, the AS-IC oligonucleotide blocked the OSM-induced stimulation of AP activity in MG63 cells (Fig. 8B). Two other antisense oligodeoxynucleotides complementary to the middle of the coding region and the 3Јend of the p21 gene, respectively, exerted similar inhibitory effects on OSM-induced AP activity. As in the case of the effect of OSM on AP, blocking the expression of the p21 protein with the p21 antisense oligonucleotide AS-IC abolished the protective effect of OSM on Fas-induced apoptosis of MG63 cells (Fig. 8C). A sense oligonucleotide (S-IC), used as a negative control in these experiments, had no significant influence on the effects of OSM. DISCUSSION The results of the studies described in this paper demonstrate for the first time that activation of the gp130 signaling pathway and Stat3 by IL-6 type cytokines leads to the transcriptional activation of the p21 WAF1,CIP1,SDI1 gene in human osteoblastic cells. Moreover, our findings strongly suggest that stimulation of p21 expression is required for the pro-differentiation and anti-apoptotic effects of IL-6 type cytokines on this cell type. These observations are in full agreement with a role of p21 as a downstream effector of the biologic actions of cytokines that activate STATs, as indicated by evidence that Stat1 and Stat5 mediate ␥-IFNand thrombopoietin-induced p21 expression, respectively, and that p21 mediates the cell growth arrest effect of ␥-IFN on A431 cells and the promotion of differentiation by thrombopoietin on a megakaryoblastic leukemia cell line (13,33).
The p21 gene promoter contains three SBEs, designated p21-SIE1, p21-SIE2, and p21-SIE3 (33). The 2.4-kb fragment of the p21 promoter used in our studies, however, contained only p21SIE1, the most proximal of these three elements. Therefore, it is possible that we have underestimated the magnitude of the effect of IL-6 type cytokines on the transcriptional activity of this gene. Nonetheless, Stat3 binds preferentially to p21-SIE1, Stat5 binds to p21-SIE2 and p21-SIE3, but not p21-SIE1, and Stat1 binds to all three (13,33). In view of this evidence and the fact that Stat3 is the principal STAT activated by IL-6 type cytokines (37), p21-SIE1 is likely to be the most relevant element of the three for the actions of the IL-6 type cytokines.
We have previously reported that IL-6 type cytokines arrest MG63 cell growth at the G 1 phase (19). These earlier observations, taken together with the present finding that blockade of p21 prevented the cytokine-induced stimulation of AP activity, are consistent with the contention that acquisition of a specific phenotype is associated with cell cycle withdrawal and that p21 may play a critical role during this process. In support of this, up-regulation of p21 expression, in a p53-independent fashion, and cell cycle arrest have been documented during the differentiation of a number of cell types (9 -14, 38, 39). Furthermore, cell cycle arrest in the G 1 phase has been shown to facilitate the differentiation process. Thus, overexpression of the Cdk inhibitors p21 and/or p27 caused U937 myelomonocytic cells to differentiate toward a mature monocyte/macrophage phenotype (11). Likewise, overexpression of the Cdk inhibitors p21 and p16 in myoblasts augmented muscle-specific gene expression (40).
In studies not shown here, we have found that besides IL-6 type cytokines, two steroid hormones known to promote osteoblast differentiation in vitro, namely 1,25-dihydroxy-vitamin D 3 and dexamethasone, also increase p21 levels in human osteoblast-like MG63 cells. Furthermore, we have observed an increase in the expression of p21 in primary cultures of murine bone marrow cells maintained in the presence of ascorbic acid, another established inducer of osteoblast differentiation. Taken together with the evidence presented in this report, these observations suggest strongly that p21 up-regulation may be a general mechanism of stimulating osteoblastic cell differentiation.
Other Cdk inhibitors that are involved in cell differentiation may act in concert with p21. For example, the greatest increase in differentiation markers of myelomonocytic U397 cells are achieved when p21 and p27 are overexpressed simultaneously (11). In line with this evidence, we have observed that besides up-regulating the expression of p21, gp130 activation also induces smaller increases in the levels of p27 and p15 in MG63 cells (data not shown). These phenomena may account for the fact that OSM-induced increase in AP was not completely suppressed by p21 antisense oligonucleotides.
In addition to the evidence for a role of p21 in cell differentiation, it has been shown previously that muscle cells or tumor cells expressing high levels of p21 are protected against apoptosis (15-17). Furthermore, p53-induced apoptosis in embryonic fibroblasts from a p21-deficient mouse could be prevented by overexpression of p21 (17). The finding of the present paper FIG. 7. Identification of the p21-SBE-binding proteins. A, nuclear extracts prepared from unstimulated or OSM-stimulated MG63 cells were incubated with a 32 P-end-labeled double stranded oligonucleotide containing p21-SIE1, in the absence or presence of a 100-fold molar excess of unlabeled wild type p21-SIE1 oligonucleotide (SIE1), or unlabeled mutated p21-SIE1 oligonucleotide (SIE1M). Supershifts were performed by incubating the nuclear proteins with either non immune IgG, anti-Stat3 antibody, or anti-Stat1␣ antibody for 10 min before addition of the probe. The black arrow indicates the position of the Stat3/Stat1␣ heterodimers. The white arrow indicates the complex supershifted by Stat3 antibody. B, MG63 cells were transfected with 2.5 g of either p21-luc or with pSIE-TK-luc, together with 0.5 g of the expression plasmid for Stat3, 3 ng of the Renilla luciferase control plasmid, and the expression vector for the dnStat3 or the empty vector. Forty-eight hours after transfection, cells were treated with OSM for 6 h, and lysates were prepared. Values shown are standardized means of -fold induction (ϩOSM/untreated) of three independent transfections. Error bars indicate the standard error of the mean; *, p Ͻ 0.05 versus control. Similar results were obtained in an additional experiment.
FIG. 8. Increases in p21 protein are required for the pro-differentiating and anti-apoptotic effects of OSM on osteoblasts. A, cells were incubated with the indicated oligonucleotides for 30 min before the addition of 20 ng/ml OSM. After 6 h, cells were lysed, and proteins were separated by SDS-polyacrylamide gel electrophoresis. Western blotting was performed sequentially with anti-p21 and anti-␤actin antibodies, as in Fig. 1. B, cells were cultured with or without 20 ng/ml OSM, and in the presence or absence of the indicated oligonucleotides for 3 days (8 replicas/condition). The standardized OSM AP/ MTS were analyzed by one-way ANOVA and compared with the group without oligonucleotide or with the group with sense oligonucleotide using Dunnet's test. Each antisense oligonucleotide reduced OSM response compared with that of the no oligonucleotide group, or with that of the sense oligonucleotide group (*, p Ͻ 0.05). Similar results were obtained in two additional experiments. C, cells were cultured in the absence or presence of 0.1 nM ␥-IFN, alone or in combination with 20 ng/ml OSM. Apoptosis was induced by addition of anti-Fas antibody, and the percentage of dead cells was determined as indicated under "Experimental Procedures." Results are the averages of triplicate wells, in which more than 200 cells per condition were counted. Bars represent means Ϯ S.D. Differences between means were analyzed by ANOVA. * indicates differences statistically significant (p Ͻ 0.05) versus cells treated with ␥-IFNϩ␣Fas, by Dunnet's test. Two additional experiments showed similar results. that blockade of the OSM-induced increase in p21 prevented the anti-apoptotic effect of the cytokine is consistent with these earlier observations.
That IL-6 type cytokines influence osteoblast differentiation was originally suggested by the observation that mice overexpressing LIF exhibit sclerotic bone formation (41). Conversely, targeted disruption of the LIF receptor ␤ gene results in decreased bone volume in the primary spongiosa of developing bone of fetal mice (42). As in the case of LIF, mice overexpressing OSM exhibit increased bone formation and enlarged limbs (43). Consistent with these in vivo observations, we and others have demonstrated that IL-6 type cytokines induce differentiation of committed osteoblastic cells in vitro (19, 44 -46). In addition, IL-6 plus sIL-6R or LIF stimulate the differentiation of uncommitted mesenchymal progenitors of the bone marrow toward the osteoblast phenotype (21), and IL-6 plus sIL-6R or IL-11 plus soluble IL-11 receptor act on embryonic fibroblasts to promote commitment toward the osteoblast phenotype, without affecting the differentiation toward adipocytes or muscle cells (20).
The relevance of p21 regulation by IL-6 type cytokines to osteoblast differentiation and survival in vivo is speculative at this stage. Mice lacking p21 WAF1,CIP1,SDI1 appear normal, and histological sections of vertebral bones from these animals were reported to be normal (47). Therefore, p21 is not critical for osteoblast differentiation during development; it may be that other Cdk inhibitors substitute for p21 in these animals. Normal bone development in p21-deficient mice does not rule out, however, a potentially significant role for this gene in states of altered bone cell formation, such as sex steroid deficiency. For example, IL-6-deficient mice also undergo normal bone development. However, after gonadectomy these mice do not develop the increase in osteoblast and osteoclast formation that causes the bone loss observed in their gonadectomized wild type littermates (48,49). Thus, although IL-6 is not required for normal bone development, it is absolutely required for the bone loss associated with loss of sex steroids. The work presented in this report demonstrates that p21 is a downstream effector of at least some of the IL-6 type cytokines actions on osteoblastic cells. Therefore, it is possible that p21 is also required for the changes in bone cell formation mediated by IL-6 following loss of sex steroids. The elucidation of the role of p21 in states of altered bone cell formation will require similar studies in p21-deficient mice.
In conclusion, the evidence presented here establishes that p21 is a primary response gene for cytokines that activate the gp130/STAT signaling pathway, and a critical effector of their pro-differentiation and anti-apoptotic properties in osteoblastic cells. Further studies will be required to establish whether the transcriptional activation of the p21 WAF1,CIP1,SDI1 gene is responsible for the well established differentiation-promoting and anti-apoptotic effects of this family of cytokines on their numerous normal and malignant cell targets.