Altered Regulation of Cell Cycle Machinery Involved in Interleukin-1-induced G 1 and G 2 Phase Growth Arrest of A375S2 Human Melanoma Cells*

Interleukin-1 (IL-1) inhibits the growth of A375S2 human melanoma cells by arresting them at G 1 and G 2 phases of the cell cycle. The arrests are preceded by a rapid decrease in kinase activities of cyclin E-Cdk2 and cyclin B1-Cdc2, which are critical for G 1 -S and G 2 -M progression, respectively. IL-1 quickly enhances the protein expression of the CDK inhibitor p21 cip1 . The induced p21 binds preferentially to cyclin E-Cdk2, and the increase in p21 binding parallels the decrease in cyclin E-Cdk2 activity. Thus, p21 is likely to be responsible for the inhibition of cyclin E-Cdk2 activity and G 1 arrest. Coinciding with the decrease in cyclin B1-Cdc2 activity, there is an increase in tyrosine phosphorylation of Cdc2, suggesting that an increase in the inactive Tyr-15-phos-phorylated form of Cdc2 is involved in the decrease in cyclin B1-Cdc2 activity and G 2 arrest. Furthermore, we found that IL-1 causes rapid dephosphorylation of p107, but not of pRb or p130, while the total protein levels of p130 are increased. Thus, IL-1 may exert its growth-arresting effects via p107 and p130 pathways rather than through pRb.

Interleukin-1 ␤ (IL-1), 1 originally defined as a monocytederived factor mitogenic for thymocytes, is now known to affect many biological activities, including the ability to alter immunologic, inflammatory, hematopoietic, and homeostatic responses in the host system. In vitro, IL-1 has been shown to inhibit the growth of certain tumor cells (1)(2)(3). A role for IL-1 in host defense against tumors has been further suggested by its ability to augment natural killer cell activity (4), monocytemediated tumor cytotoxicity (5), and T-and B-cell responses (6) and to induce tumor regression in mice (7). These properties have made IL-1 an attractive candidate for potential application for certain solid tumors (8 -12), although several problems including various adverse effects such as fever and hypotension remain unresolved. Elucidation of the molecular mechanisms that mediate the antiproliferative effect of IL-1 on tumor cells would not only be of help in resolving these problems; it would also provide valuable information on how cell proliferation is regulated negatively by extracellular signals.
In cell culture, various types of tumor cells, such as melanoma (4,(13)(14)(15)(16)(17)(18)(19), breast carcinoma (20), myeloid leukemia (21,22), ovarian carcinoma (23), and lung adenocarcinoma (24) cells, have been shown to be susceptible to the antiproliferative action of IL-1. A highly susceptible human melanoma cell line, A375-C6, is commonly used to study the mechanism of the IL-1-mediated growth arrest. The action of IL-1 in A375-C6 cells has been documented to be mediated through specific cell surface receptors (25). The binding of IL-1 to its receptor results in activation of a variety of second-messenger signaling pathways (reviewed in Ref. 13) and a unique primary gene expression program characterized by the induction of a composite set of immediate early genes such as gro-␣, gro-␤, c-jun, nur77/NGF1-B/NAK1, IRG-9/TIS11, and Egr-1 (16,17,19). IL-1 action in A375-C6 cells is also characterized by inhibition of ornithine decarboxylase, a rate-limiting enzyme in polyamine synthesis, leading to inhibition of DNA synthesis (18,26). Although these events are thought to mediate the antiproliferative effect of IL-1, further investigation is required to elucidate the direct mechanisms responsible for initiation and/or maintenance of the growth-arrested state. Since cell growth and proliferation are ultimately regulated by a highly conserved set of cell cycle-regulatory proteins, a fruitful approach to the study of the antiproliferative mode of action of IL-1 is to examine the influence of IL-1 on the regulation of the cell cycle-regulating machinery.
The eukaryotic cell cycle is regulated by the action of the cyclin-dependent kinases (CDKs) and their activating subunits, the cyclins (27,28). In mammalian cells, Cdk6 and Cdk4 are associated with the D-type cyclins and regulate G 1 progression. Cdk2 is associated with E-and A-type cyclins, and the respective complexes are believed to control G 1 -S transition and S phase progression, respectively. Cdc2 is associated with B-type cyclins and regulates G 2 -M phase. The most well studied substrates of the CDKs operating during G 1 and S phases are the retinoblastoma family of proteins (29,30). This family consists of pRb and the related proteins p107 and p130, collectively termed the pocket proteins. Progression from G 1 to S requires inactivation of the retinoblastoma family proteins by phosphorylation and the consequent release of a number of factors including the E2F family of transcription factors. These transcription factors then activate transcription of various genes that promote cell cycle progression. Thus, the phosphorylation state of retinoblastoma family proteins is a critical determinant in the execution of the progression from G 1 to S. Recently, Muthukkumar et al. (15) documented that growth arrest by IL-1 was linked to suppression of pRb phosphorylation and suggested that hypophosphorylated pRb may possibly mediate the action of IL-1. Since pRb is phosphorylated by the action of the CDKs, it is conceivable that the IL-1-induced pRb hypophosphorylation is the result of negative regulation of the CDKs. However, the effect of IL-1 on the activity or the regulation of the CDKs has not been defined as yet. In mammalian cells, regulation of CDKs is achieved by several mechanisms including alteration of CDK levels; changes in the expression of the cyclins with which CDKs interact; activation and inactivation of CDKs by phosphorylation/dephosphorylation events; and the abundance and action of two families of CDK inhibitors, the Cip/Kip family (p21 cip1 , p27 kip1 , and p57 kip2 ) and the Ink4 family (p16 ink4a , p15 ink4b , p18 ink4c , and p19 ink4d ) (reviewed in Refs. 27, 31, and 32). Modulation at any of these levels of regulation could regulate pRb phosphorylation.
The current study was undertaken to investigate the molecular mechanisms, which mediate the antiproliferative effect of IL-1, through a detailed analysis of the IL-1 effects on the cell cycle-regulating molecules, such as CDKs, cyclins, CDK inhibitors, and pRb family proteins. By using A375S2 human melanoma cells (33), which are as highly sensitive to the antiproliferative effect of IL-1 as A375-C6 cells, we have systematically analyzed the changes that occur in kinase activities, expression levels, interactions, and phosphorylation status of the cell cycle-regulating molecules and compared them to the timing of onset and completion of the growtharresting process. We conclude from these experiments that IL-1 arrests A375S2 cells at G 1 and G 2 phases of the cell cycle by inhibiting kinase activities of cyclin E-Cdk2 and cyclin B1-Cdc2 complexes, and the inhibitory mechanism for each complex is different. Furthermore, data presented here suggest that IL-1 exerts its growth-arresting effects via p107 and p130 pathways rather than through pRb.

EXPERIMENTAL PROCEDURES
Cell Culture-A375S2 human melanoma cells (obtained from Otsuka Pharmaceutical Co., Ltd., Cellular Technology Institute, Tokushima, Japan) were cultured in minimum essential medium (Life Technologies, Inc. Oriental Co., Tokyo, Japan) supplemented with 10% heat-inactivated fetal calf serum (Life Technologies) in a humidified incubator at 37°C in 5% CO 2 . Human recombinant IL-1 was purchased from R & D Systems (Minneapolis, MN). To assure exponential cell growth, A375S2 cell cultures were set up 48 h prior to the addition of IL-1; 0.6 ϫ 10 6 cells were plated in 100-mm plastic dishes and incubated for 24 h, the culture medium was then replaced, and the incubation was continued for another 24 h before the addition of IL-1 (1.0 ng/ml, unless otherwise indicated). Cells were harvested at different times after the addition of IL-1 and processed for cyclin-dependent kinase assays, Western blotting, immunoprecipitation, or Northern blotting, as described below. For the cell proliferation assay, cells at 1.2 ϫ 10 5 per 60-mm dish were treated with IL-1 at different concentrations. At various experimental intervals, cells were trypsinized and counted using a cell counter (Sysmex model F-500, Sysmex Co., Kobe, Japan). In a parallel experiment, the number of viable cells was also determined by the Trypan blue dye exclusion test. For cell cycle analysis, trypsinized cells were stained with propidium iodide by using the Cycle TEST PLUS kit (Nippon Becton Dickinson Co., Ltd., Tokyo, Japan) and analyzed for DNA content by using the Becton Dickinson fluorescence-activated cell sorting system (FACSCalibur). Cell cycle distribution was determined using ModFit LT Software (Verity Software House, Popsham, ME). All experiments were repeated at least twice.
Immunoprecipitation and Kinase Assay-Cell monolayers were washed twice with ice-cold phosphate-buffered saline and then scraped into ice-cold Nonidet P-40 lysis buffer (50 mM HEPES (pH 7.5), 250 mM NaCl, 0.1% Nonidet P-40, with 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 10 mM ␤-glycerophosphate, 1 mM NaF, 0.1 mM Na 3 VO 4 , 10 g/ml aprotinin, 10 g/ml leupeptin, and 5 g/ml pepstatin A added just before use). Cell lysates were sonicated and clarified by centrifugation at 19,000 ϫ g for 15 min at 4°C, aliquots were removed for analysis of protein concentration (measured using the DC Protein Assay kit; Nippon Bio-Rad Laboratories, Osaka, Japan), and samples were adjusted to equivalent protein concentrations. After precleaning with protein G-Sepharose beads (Sigma-Aldrich Japan, Tokyo, Japan), equal amounts of protein (150 g) were incubated with the appropriate antibody for 2 h at 4°C. Antibody complexes were recovered on protein G-Sepharose beads and washed four times with Nonidet P-40 lysis buffer and twice with 50 mM HEPES buffer (pH 7.5) containing 1 mM dithiothreitol. The beads were then incubated at 30°C for 20 min in 25 l of reaction buffer (50 mM HEPES (pH 7.5), 10 mM MgCl 2 , 20 M ATP, 1 mM dithiothreitol, 2 mM glutathione, 10 mM ␤-glycerophosphate, 1 mM NaF, 0.1 mM Na 3 VO 4 ) in the presence of either 2.5 g of histone H1 (Roche Molecular Biochemicals) or 1 g of GST-pRb substrate (Santa Cruz Biotechnology) and 5 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech) per reaction. The reactions were stopped by the addition of 2ϫ SDS sample buffer, and samples were analyzed by 12.5% SDS-PAGE followed by autoradiography.
Western Blot Analysis-Lysis buffer and immunoprecipitation were as described above. Whole cell lysates or immunoprecipitates were resolved by 12.5% SDS-PAGE (7.5% in the case of pRb family proteins), the resolved proteins were transferred to nitrocellulose membranes (Nippon Bio-Rad Laboratories) using a semidry transfer apparatus (Nippon Bio-Rad Laboratories), and the membranes were blocked by incubating them overnight in TBST (Tris-buffered saline (pH 7.6) containing 5% nonfat dry milk and 0.05% Tween 20). The blots were then rinsed three times in TBST and incubated with primary antibody for 2 h. After a wash, the nitrocellulose was incubated with a secondary antibody conjugated to horseradish peroxidase (Amersham Pharmacia Biotech) for 30 min. Finally, the protein was visualized by ECL-based autoradiography as recommended by the manufacturer (Amersham Pharmacia Biotech).
Immunodepletion-Whole cell extracts were prepared with Nonidet P-40 lysis buffer and subjected to four rounds of immunodepletion, using either normal rabbit serum or a rabbit polyclonal antibody raised against p21 (15431E; PharMingen) immobilized on protein G-Sepharose beads. Following depletion, supernatants were subjected to Western blot analysis as described above to show the presence of remaining proteins.
Northern Blot Analysis-Total cellular RNA was isolated by using RNAzol (Biotecx Laboratories, Inc., Houston, TX), and RNA samples (10 g) were fractionated by electrophoresis through 1.3% agarose gels containing 6% formaldehyde, 20 mM MOPS, 5 mM sodium acetate, and 1 mM EDTA and blotted onto a synthetic nylon membrane (Biodyne A; Pall BioSupport Corp., East Hills, NY) in 20ϫ SSC (1ϫ SSC: 150 mM NaCl, 15 mM sodium citrate) and baked at 80°C for 2 h. The blots were prehybridized and then hybridized to a 32 P-labeled antisense RNA probe for human p21 at 65°C. The probe was synthesized in the presence of [␣-32 P]UTP (Amersham Pharmacia Biotech) using a Riboprobe Gemini II system (Promega KK, Tokyo, Japan). As a template for the antisense RNA probe, the full-length cDNA of human p21 (kindly provided by Dr. A. Noda, Kobe University, Kobe, Japan) was used. The membrane was washed twice in 2ϫ SSC plus 0.1% SDS for 5 min and then twice in 0.1ϫ SSC plus 0.1% SDS for 30 min at 65°C and exposed to x-ray film at Ϫ80°C using an intensifying screen. After detection of p21 mRNA, the membranes were rehybridized with a 32 P-labeled antisense RNA probe for human glyceraldehyde-3-phosphate dehydogenase (Ambion Inc., Austin, TX) to verify the equal loading of RNA.

RESULTS
Initial Characterization of the Antiproliferative Action of IL-1 on A375S2 Cells-To determine the optimal IL-1 concentration to inhibit A375S2 cell proliferation completely, exponentially growing cells were treated with various concentrations of IL-1 for 96 h, and the total number of live cells was counted. As shown in Fig. 1A, more than 1 ng/ml of IL-1 inhibited the cell proliferation almost completely. A trypan blue dye exclusion test indicated little toxicity associated with this dose of IL-1 (Ͼ90% viability after 96 h). We therefore used 1 ng/ml IL-1 for subsequent experiments. Fig. 1B shows the time course of the antiproliferative effect of IL-1 on A375S2 cells. A slight inhibition of cellular proliferation was apparent as early as 24 h after IL-1 treatment, and thereafter the proliferation was inhibited almost completely, indicating that the action of IL-1 was exerted rapidly. To further characterize the antiproliferative action of IL-1, an analysis of the distribution of cells in the various phases of the cell cycle was performed by flow cytometry as a function of time following IL-1 treatment (Fig. 2). No change in cell cycle distribution was apparent during the first 4 h following exposure to IL-1. A decrease in the percentage of cells in S phase was noticeable between 6 and 8 h. There was a remarkable decrease in the S-phase fraction after 24 h (to 3% or less) accompanied by an increase in the percentage of cells in both G 1 and G 2 /M. By 48 h, the percentage of cells in S phase had decreased to Ͻ1%. In microscopic analysis of Wright-Giemsa-stained cell preparations, no mitotic cell was observed after 48 h of IL-1 treatment (data not shown). These results showed that IL-1 arrested A375S2 cells at G 1 and G 2 phases of the cell cycle. It was also shown that the process resulting in the blockage at two different cell cycle points began to take effect within the first 6 -8 h and was nearly completed by 24 h following IL-1 treatment. We therefore focused on the mechanisms inhibiting cell cycle progression during the early phase (Ͻ6 -8 h) of IL-1 treatment.
Effects of IL-1 on Cyclin-CDK Kinase Activities-Since CDKs complexed with their catalytic subunit cyclins regulate the cell cycle progression in different phases, we first investigated whether the IL-1-induced blockage at two different cell cycle points was associated with changes in the kinase activities of various cyclin-CDK complexes. Results of the in vitro kinase assays using appropriate immunoprecipitates and substrates (GST-pRb or histone H1) are shown in Fig. 3. Cdk6-and cyclin D-associated kinase activities both elevated from 2 to 4 h following IL-1 treatment, peaked at 6 -8 h, and thereafter declined and fell below control levels at 24 -48 h. Cdk4 was expressed at very low levels in all control and IL-1-treated A375S2 cells (see below), and no kinase activities could be detected in Cdk4 immunoprecipitates (data not shown). The levels of cyclin A-associated kinase activity did not change significantly during the first 8 h of IL-1 treatment, but most of the activity had disappeared after 24 h of treatment. The kinase activities associated with the other Cdks and cyclins we tested, namely Cdk2, Cdc2, cyclin E, and cyclin B, all decreased gradually from 4 to 8 h and almost disappeared by 24 h after IL-1 treatment. These results demonstrate that IL-1 treatment results in a temporary increase in cyclin D-Cdk6 activity and consistent decreases of cyclin E-Cdk2 and cyclin B1-Cdc2 activities. It should be noted that the decreases of cyclin E-Cdk2 and cyclin B1-Cdc2 activities were both already detectable after 4 h of IL-1 treatment, before any changes in cell cycle distribution could be detected (Fig. 2). The kinase activities of cyclin E-Cdk2 and cyclin B-Cdc2 are known to be critical for G 1 -S and G 2 -M transition, respectively. Hence, it is likely that down-regulation of cyclin E-Cdk2 and cyclin B1-Cdc2 activities from 4 h is a mechanism of braking the cell cycle at their respective points during the early phase of IL-1 treatment.
Effects of IL-1 on the Protein Levels of Cyclins and CDKs-We next examined whether the early changes in cyclin-CDK kinase activities are attributable to changes in the protein levels of cyclins and CDKs. Results of Western analysis are shown in Fig. 4. IL-1 did not reduce the expression of any of the cyclins or CDKs for at least 8 h, indicating that the changes of cyclin D-CDK6, cyclin E-Cdk2, or cyclin B1-Cdc2 kinase activities observed in the early phase of IL-1 treatment (Fig. 3) were not due to changes in protein expression levels of the respective components. At 24 h, the expression of Cdk6 and cyclin D was decreased significantly. It was further decreased at 48 h. The expression of cyclin A, cyclin B1, and Cdc2 was also decreased markedly at 24 h, and these components had mostly disappeared at 48 h. The decreased expression of these components after 24 h may correspond to the decrease in the respective cyclin-CDK activities during the late phase of IL-1 treatment (Fig. 3). In contrast, the protein expression of Cdk2 and cyclin E remained unchanged up to 48 h despite the complete loss of cyclin E-Cdk2 kinase activity after 24 h of IL-1 treatment (Fig.  3). For Cdk2, however, a slower migrating form was observed to increase at 24 and 48 h. Since the form represents the Thr-160dephosphorylated inactive form of Cdk2 (34), the increased dephosphorylation of Cdk2 was likely to contribute to the down-regulation of Cdk2-associated kinase activities after 24 h of IL-1 treatment. Cdk4 was expressed at very low levels in these cells with no detectable change in the expression following IL-1 treatment.
Further experiments examined the possibility that altered association of cyclins with their CDK partners might contribute to the changes in cyclin-CDK kinase activities during the early phase of IL-1 treatment. In these experiments, cyclins were immunoprecipitated, and the immunoprecipitates were then immunoblotted for the respective CDK partners. As shown in Fig. 5, levels of cyclin D-associated Cdk6 elevated from 2 to 4 h following IL-1 treatment, peaked at 6 -8 h, and thereafter declined. The changes in cyclin D-Cdk6 association were similar in timing and magnitude to the changes in cyclin D-Cdk6 kinase activity (Fig. 3), suggesting that the temporary increase of cyclin D-Cdk6 activity between 2 and 8 h reflected the increased formation of cyclin D-Cdk6 complex. On the other hand, no change was observed in levels of cyclin E-associated Cdk2 over the 48 h of treatment. Levels of cyclin A-associated Cdk2 and cyclin B1-associated Cdc2 were unaltered for the first 8 h following IL-1 treatment, but thereafter both declined to low levels at 24 h, paralleling the temporal changes in total protein levels of cyclin A, cyclin B1, and Cdc2 (Fig. 4).
Effects of IL-1 on the Expression of CDK Inhibitors-As described above, the early decreases in cyclin E-Cdk2 and cyclin B1-Cdc2 activities were not accompanied by decreases in the expression of respective cyclins or CDKs or by decreases in their complex formation (Figs. [3][4][5]. Then, to investigate other possible causes of the decrease in kinase activities, the expression of various CDK inhibitors was examined by Western analysis. Data in Fig. 6 show that IL-1 treatment resulted in a rapid induction of p21. The protein levels of p21 began to increase in the first 2 h of treatment and continued to increase up to 48 h. On the other hand, the protein levels of p27, which was expressed at very low levels in untreated A375S2 cells, remained unchanged during the early phase of IL-1 treatment, and an obvious increase was observed only after 24 h of treatment. The expression of other CDK inhibitors, p57, p15, p16, p18, and p19, was only detected at very low levels in cells that were untreated or treated with IL-1 (Fig. 6).
To determine whether the increase of p21 protein expression is regulated at the transcriptional level, p21 mRNA was quantitated by Northern blot analysis using a full-length antisense RNA probe for human p21. As shown in Fig. 7, a small increase in p21 mRNA levels was detectable within 30 min of IL-1 treatment, and the levels continued to increase up to 48 h, paralleling the changes in p21 protein levels. These results indicated that the IL-1-induced increase in p21 protein levels is regulated transcriptionally at least in part. p21 expression is known to be under the control of both p53-dependent and p53-independent mechanisms. To examine whether p53 protein contributes to the IL-1-mediated induction of p21, the protein levels of p53 were analyzed after treatment of cells with IL-1. As shown in Fig. 7, p53 protein was barely detectable in normal growing A375S2 cells, and no change was observed in its expression levels following IL-1 treatment. On the other hand, when we treated A375S2 cells with camptothecin, a topoisomerase inhibitor that induces double-stranded DNAbreaks, we observed the induction of both p53 and p21 proteins, indicating that the function of p53, required for induction of p21 by DNA damage, is retained in these cells (data not shown). Thus, the present results suggested that IL-1 induces p21 independently of p53.
Association of p21 with Cyclin-CDK Complexes-To determine whether p21 is responsible for the inhibition of cyclin E-Cdk2 and cyclin B-Cdc2 complexes during the early phase of IL-1 treatment, we next investigated whether there was any change in the association of p21 with cyclin-CDK complexes after IL-1 treatment. Cell extracts were immunoprecipitated with antibodies against various cyclins or CDKs and Western blotted for p21 (Fig. 8). We found that a certain amount of p21 was coimmunoprecipitated with anti-Cdk6, -Cdk2, -cyclin D, -cyclin E, and -cyclin A antibodies even from extracts of untreated growing A375S2 cells. The relative amount of p21 associated with cyclin E was increased gradually from 4 to 48 h following treatment with IL-1. Cyclin A-associated p21 was increased slightly from 4 to 8 h, although cyclin A-Cdk2 kinase activity was not changed during the early phase of IL-1 treatment (Fig. 3). Cdk2-associated p21 increased up to 8 h and thereafter decreased gradually, consistent with the observed changes in the p21 association with cyclins E and A, partners of Cdk2. For Cdc2 and cyclin B1, a p21 association was not detectable in untreated growing cells but became detectable faintly at 2-4 h following IL-1 treatment and thereafter tended to increase slightly up to 24 h. These findings indicate that the level of p21 associated with cyclin E-Cdk2, cyclin A-Cdk2, and cyclin B1-Cdc2 was increased during the early phase of IL-1 treatment.
We next asked whether the increased level of p21 associated with these cyclin-CDK complexes was sufficient to inhibit their kinase activities. To answer this question, we determined what fraction of each cyclin-CDK complex was associated with p21 by immunodepletion experiments. Cell extracts from untreated cells or from cells treated with IL-1 for 8 h were immunodepleted by three successive rounds of immunoprecipitation with either anti-p21 antiserum or normal rabbit serum as a control. What remained in the depleted extracts was then examined by immunoblotting with specific antibodies against cyclins and CDKs (Fig. 9). Quantitation of p21 levels indicated that anti-p21 antiserum successfully removed p21 from control or IL-1treated extracts. We found that nearly all of cyclin E was depleted with anti-p21 antiserum not only from IL-1-treated extracts but also from control extracts, indicating that all of the cyclin E-Cdk2 complex was already bound by p21 before treatment with IL-1 in A375S2 cells. This finding was surprising in light of the fact that cyclin E-Cdk2 complex is active in growing cells and that the amount of p21 bound to cyclin E-Cdk2 increased still further following IL-1 treatment (Fig. 8). A possible interpretation of these findings is that p21 can associate with the cyclin E-Cdk2 complex in a noninhibitory mode and that multiple molecules of p21 can associate with the complex. Indeed, it has been proposed that the first molecule of p21 that associates with a cyclin-CDK complex does not inhibit its activity and that the binding of a second p21 molecule is required to inhibit its kinase activity (35,36). For Cdk2, a small proportion of the protein was depleted from both control and IL-1treated extracts, consistent with the fact that Cdk2 is expressed in excess over cyclin E and A in the cell and that the binding affinity of p21 to free Cdk2 is much lower than to cyclin-Cdk2 complex (35,(37)(38)(39). In contrast with the case of cyclin E, no depletion of cyclin A, cyclin B1, or Cdc2 was observed either before or after treatment with IL-1, suggesting that only a minor portion of the cyclin A-Cdk2 or cyclin B-Cdc2 complexes was associated with p21 even after IL-1 treatment. On the basis of these results, we conclude that p21 binds preferentially to cyclin E-Cdk2 complex and that the increased level of p21 after IL-1 treatment is effectual for the inhibition of cyclin E-Cdk2 but has little or no effect on cyclin A-Cdk2 and cyclin B1-Cdc2.
We also found that the relative amount of Cdk6-and cyclin D-associated p21 increased 2-4 h following IL-1 treatment, peaked at 6 -8 h, and thereafter declined (Fig. 8). Immunodepletion experiments performed with anti-p21 antiserum verified that a certain proportion of cyclin D, possibly complexed with Cdk6, was indeed associated with p21 in cells treated with IL-1 for 8 h. These findings were surprising since, as seen in Fig. 3, the kinase activity of cyclin D-Cdk6 increased during the early phase of IL-1 treatment. It is noteworthy that the time kinetics of the increase in the association of p21 with cyclin D-Cdk6 complexes (Fig. 8) coincided with the increase of kinase activity (Fig. 3), resulting probably from the increase of complex formation between cyclin D and Cdk6 (Fig. 5). These findings may be explained by the recent suggestion that p21 acts not only as a CDK inhibitor but also as an assembly factor for cyclin-CDK complex formation (36,40).
IL-1 Induces Tyrosine Phosphorylation of Cdc2-Another important mechanism for controlling CDK activity is phosphorylation of conserved threonine and tyrosine residues of CDKs. The kinase activity of both Cdk2 and Cdc2 is regulated negatively by phosphorylation of Tyr-15 and Thr-14 (34,41). To examine the possibility that such phosphorylation might contribute to the inhibition of Cdk2 and Cdc2 during the early phase of IL-1 treatment, we analyzed the relative content of phosphotyrosine in these two CDKs by immunoblotting with an anti-phosphotyrosine antibody. As shown in Fig. 10, no tyrosine phosphorylation was detected in Cdk2 immunoprecipitated with anti-cyclin E antibodies from either the control or IL-1-treated cell extracts. On the other hand, phosphotyrosine was detected at low levels in cyclin B1-associated Cdc2 before treatment with IL-1, and the levels increased gradually from 4 to 8 h following IL-1 treatment (Fig. 10), coincident with the observed decrease in cyclin B1-Cdc2 kinase activity (Fig. 3). These findings suggest that Tyr-15 phosphorylation is a mechanism that inhibits the activity of cyclin B1-Cdc2 but is not involved in the inhibition of cyclin E-Cdk2 during the early phase of IL-1 treatment.
Analysis of Phosphorylation Status of pRb Family Proteins-As shown above, the kinase activity of cyclin E-Cdk2 started to decrease 4 h following IL-1 treatment (Fig. 3). The best studied G 1 cyclin-CDK substrate is the product of the retinoblastoma tumor suppresser gene, pRb. It has become clear that pRb is a negative regulator that acts in the G 1 phase of the cell cycle, and its activity appears to be modulated by phosphorylation. Then we determined whether pRb underwent dephosphorylation at times compatible with the decrease in cyclin E-Cdk2 activities. The reduction in mobility of pRb on SDS-PAGE and Western blot analysis is a widely accepted indicator of pRb phosphorylation. In asynchronously growing A375S2 cells, pRb was found in both hyper-and hypophosphorylated forms, which migrated as multiple bands in the molecular mass range of 110 -115 kDa ( Fig. 11; detected by an antibody recognizing a pRb C-terminal protein). Contrary to our expectation, no change in the proportions of hyper-and hypophosphorylated pRb was recognized in the early phase (up to 8 h) of IL-1 treatment, and disappearance of the slow-migrating hyperphosphorylated forms of pRb was observed only 24 h after treatment (Fig. 11). Then, we next investigated whether IL-1 affected the phosphorylation state of the two known pRb-related proteins, p107 and p130. As for the p107 protein, a slight downward mobility shift, which seemed to represent the dephosphorylation of the protein, was first seen at 4 h of IL-1 treatment, and a more marked shift was evident at 8 h (Fig. 11). The protein expression level of p107 was not affected during the first 8 h of treatment, but after 24 h, when the vast majority of cells have exited the cell cycle, the p107 protein almost disappeared. On the other hand, p130 was present as both hyper-and hypophosphorylated forms in asynchronously growing control cells. In contrast with p107, the total protein level of p130 was increased by treatment with IL-1. The increase was first seen at 6 h and peaked at 24 h. Interestingly, independent of the increase in the total protein levels, the ratio between the hyper-and hypophosphorylated forms of p130 was constant during the first 8 h, and a decrease in the hyperphosphorylated form with an accumulation of the hypophosphorylated form was observed only at the 24-and 48-h time points. Taken together, these results suggest that IL-1 exerts its growth-inhibitory effects not through pRb but by regulating the phosphorylation state of p107 and the abundance of p130.

DISCUSSION
This study represents an extensive analysis of the potential roles of the cell cycle machinery in IL-1-induced growth arrest of A375S2 human melanoma cells. The first important observation in this study is that IL-1 arrested A375S2 cells at both G 1 and G 2 phases of the cell cycle, suggesting that IL-1 targets a molecule(s) critical for progression through G 1 (or G 1 -S) and G 2 (or G 2 -M). In the previous study (14), in some contrast with our data, A375 cells, which are less sensitive to IL-1 than A375S2 cells (33), were shown to be arrested by IL-1 at G 1 phase but not at G 2 phase, although IL-1 did retard progression of A375 cells through G 2 -M. Another A375-derived IL-1-sensitive clone named A375-C6 has been reported to be arrested by IL-1 at G 0 /G 1 phase, but no evidence for blockage or retardation at G 2 phase was obtained (15). It therefore appears likely that the type of arrest induced by IL-1 varies with the clone of the A375 cell line. The cell cycle analysis of A375S2 cells also showed that IL-1 started to brake cell cycle progression at G 1 and G 2 phases within the first 6 -8 h of treatment, and the blockage at the two phases was nearly completed by 24 h. To elucidate the molecular mechanism of the IL-1 action, it is of importance to distinguish the induced changes in cell cycleregulating molecules causing growth arrest from those that are a consequence of the cell cycle arrest. We therefore focused on the changes and/or modulations of cell cycle-regulating molecules observed during the early phase (Ͻ6 -8 h) of IL-1 treatment before any changes in cell cycle distribution could be detected. We found that the kinase activities of cyclin E-Cdk2 and cyclin B-Cdc2 complexes both started to decrease as early as 4 h after treatment with IL-1, suggesting that these changes were causes rather than consequences of the inhibition of cell cycle progression. Since cyclin E-Cdk2 and cyclin B1-Cdc2 complexes have been shown to have crucial roles in G 1 -S and G 2 -M transition, respectively, it is likely that the IL-1-induced rapid down-regulation of cyclin E-Cdk2 and cyclin B1-Cdc2 activities is a mechanism of braking cell cycle progression at their respective points. On the other hand, the kinase activity of cyclin A-Cdk2, whose function is essential in S phase, showed no change during the first 8 h of IL-1 treatment. Consistent with this, no evidence for blockage in S phase was obtained during the early phase of IL-1 treatment. After 24 h, when the vast majority of cells have exited the cell cycle, the kinase activity of cyclin A-Cdk2, as well as that of cyclin E-Cdk2 and cyclin B1-Cdc2, disappeared completely. The complete loss of these cyclin-CDK activities after 24 h may be a secondary response reflecting the cessation of cell cycle and maintenance of cells at G 1 and G 2 phases. The kinase activity of cyclin-CDK complexes can be regulated in a number of ways (see Ref. 42 and references therein). In this study, examination of potential mechanisms for the IL-1-induced rapid down-regulation of cyclin E-Cdk2 and cyclin B1-Cdc2 activities revealed that the CDK inhibitor p21 is likely to be responsible for the inhibition of cyclin E-Cdk2 activity, whereas an increase in tyrosine phosphorylation of Cdc2 may be involved in the inhibition of cyclin B1-Cdc2 activity. p21 has been shown to inhibit a wide variety of cyclin-CDK complexes including G 1 cyclin/CDK and has been implicated in G 1 arrest following DNA damage (37,43) in response to negative growth factors, such as transforming growth factor-␤ (44,45) or interferon-␣ (46), and in maintenance of terminally differentiated cells in a nonproliferative state (47,48). Our results provide the following evidence to support a possible role for p21 in IL-1-mediated inhibition of cyclin E-Cdk2 activity causing cell cycle arrest at G 1 . First, p21 induction is the first change to be observed, occurring well before the start of cell cycle arrest. In contrast with p21, the p21-related CDK inhibitor p27, which was expressed at very low levels in proliferating A375S2 cells, was induced only after 24 h of IL-1 treatment. Other CDK inhibitors examined, including p57, p15, p16, p18, and p19, were not induced at all by IL-1. Second, the increase in p21 expression levels is paralleled by an increased binding of p21 to cyclin E-Cdk2 complex, coinciding with the decrease of kinase activity of the complex. Although the binding of p21 to cyclin A-Cdk2 and cyclin B1-Cdc2 complexes also increased in parallel with the increase in p21 expression levels, immunodepletion experiments revealed that the vast majority of the cyclin A-Cdk2 and cyclin B1-Cdc2 complexes were still free from p21 after IL-1 treatment, indicating that the increased binding of p21 to these cyclin-CDK complexes after IL-1 treatment is not sufficient in quantity to inhibit the activity of these complexes. In contrast, all of the cyclin E-Cdk2 complex was found to be complexed with p21 after and even before IL-1 treatment. At first glance, this observation appears to be in conflict with the fact that cyclin E-Cdk2 complex is active in growing cells and that the relative amount of p21 bound to the complex increased still further following IL-1 treatment. The apparent paradox, however, is resolved by proposing that p21 can associate with cyclin E-Cdk2 complex in a noninhibitory mode and that multiple molecules of p21 can associate with the complex. This is fully consistent with the idea that the first molecule of p21 that associates with a cyclin-CDK complex does not inhibit its activity and that the binding of a second p21 molecule is required to inhibit its kinase activity (35,36). Based on this idea, it has been proposed that cyclin-CDK complexes become maximally sensitive to increases in p21 levels by the binding of one p21 molecule to each complex (35). This is consistent with our observation that the decrease of cyclin E-Cdk2 activity occurred quickly, coinciding with the rapid increase in p21 expression levels following IL-1 treatment. A similar observation has been reported in normal human fibroblasts, where all of the active cyclin E-Cdk2 complex is associated with p21, and the active complex containing p21 can be inhibited by upregulated p21 following UV-induced DNA damage (37). This phenomenon was also explained based on the stoichiometric inhibitory action of p21.
Although p21 is known as a universal CDK inhibitor, recent reports provide evidence that prime targets of p21 for inhibition are cyclin E/A-Cdk2 complexes in vivo (37,49,50). Consistent with this, our observations suggest that IL-1-induced p21 is a potent inhibitor of cyclin E-Cdk2 but not effective for cyclin D-Cdk6 or cyclin B1-Cdc2. However, our observations FIG. 11. Western blot analysis of phosphorylation status of pRB family proteins. Exponentially growing A375S2 cells were treated at time 0 with 1.0 ng/ml IL-1. At intervals thereafter, whole cell lysates were prepared, subjected to SDS-PAGE, and immunoblotted with antibodies to either pRb (9032), p107 (C-18), or p130 (R27020). The positions of hypophosphorylated proteins (pRb, p107, and p130) and hyperphosphorylated proteins (ppRb, pp107, and pp130) are indicated. Representative data from two independent experiments are shown. also suggest that the induced p21 has little inhibitory effect even on cyclin A-Cdk2. This appears to be associated with the finding that all of the cyclin E-Cdk2 complex, but only a minor portion of the cyclin A-Cdk2 complex, was already associated with p21 before treatment with IL-1. Such a physical association of p21 with active cyclin E-Cdk2 but not with cyclin A-Cdk2 in growing cells is also observed in normal human fibroblasts (37) and in certain cell lines (51,52). These observations suggest that at least in these cells and cell lines, p21 binds to cyclin E-Cdk2 complex with much higher affinity than to cyclin A-Cdk2 complex, and thus p21 is a more effective inhibitor of cyclin E-Cdk2 than of cyclin A-Cdk2 in vivo. It remains to be clarified, however, whether the relative affinity of p21 to cyclin E-and cyclin A-Cdk2 complexes varies with the cell type.
While this work was in progress, Nalca and Rangnekar (53), using another A375-derived IL-1-sensitive clone, A375-C6, showed that IL-1 caused rapid induction of p21 at the mRNA and protein level. However, they have proposed that p21 does not play an important role in the growth-arresting effect of IL-1, since they found that inhibition of p21 expression by the antisense construct resulted in only a marginal rescue from the growth-arresting action of IL-1. This view stands in contrast with the present results suggesting that the rapid induction of p21 is of major importance in the action of IL-1 in A375S2 cells. This discrepancy may be attributable to a difference in the steadiness of the IL-1-induced increase in p21 protein levels between the two clones of A375 cells used. In A375-C6 cells, Nalca and Rangnekar (53) observed a rapid but transient increase (peaked at 3 h) in p21 protein expression levels following IL-1 treatment, in contrast to our data showing that IL-1 caused a rapid and sustained increase (up to 48 h) in p21 levels in A375S2 cells. Thus, in A375-C6 cells, the transient induction of p21 may not be sufficient to cause cell cycle arrest, and other mechanisms that control cell proliferation may be responsible for the arrest. Alternatively, it is conceivable that compensatory regulation mechanisms become operative in the absence of p21. In addition, it is also possible that the regulatory functions of p21 are redundant.
An unexpected finding of this study was the increase in cyclin D-Cdk6 activity during the early phase of IL-1 treatment. The increase was paralleled by the increase in the relative amount of p21-bound cyclin D-Cdk6 complex, coinciding with the increase in cellular p21 levels. A possible interpretation of these findings is that p21 facilitates the association of the complex formation between cyclin D and Cdk6 without hindering the function of the complex. This is consistent with the idea that besides simply inhibiting kinase activity, members of the p21 family can promote the association of CDKs with cyclins and thus stabilize cyclin-CDK complexes (36,40). There is increasing evidence that such a role of p21 is crucial during assembly of cyclin-CDK complexes. For instance, recent data suggest that p21 promotes the association of D-type cyclins with CDKs by counteracting the effects of the Ink4 family of CDK inhibitors (54,55). Thus, it may be concluded that the increase in cyclin D-Cdk6 activity during the early phase of IL-1 treatment mirrored the increased assembly of the active cyclin D-Cdk6-p21 complexes driven by the increase in cellular p21 levels. In this context, the possibility that this novel function of p21 also contributes to the formation of active cyclin E-Cdk2-p21 ternary complexes in growing A375S2 cells may be admitted.
In addition to the inhibition by CDK inhibitors, phosphorylation of conserved threonine and tyrosine residues near the ATP binding sites of CDKs (Thr-14 and Tyr-15 on both Cdk2 and Cdc2) is also an important mechanism employed to keep the CDKs inactive (34,41). Our data showed that there was an increase in phosphotyrosine content in Cdc2 but not in Cdk2 during the early phase of IL-1 treatment. Thus, it is conceivable that the increase in Tyr-15 phosphorylation is a mechanism that inhibits the cyclin B1-Cdc2 but is unrelated to the decrease in cyclin E-Cdk2 activity during the early phase of IL-1 treatment. Phosphorylation of Tyr-15 on Cdc2 is controlled by the opposing activities of the Wee1/Myt1 kinases (56 -58) and Cdc25 phosphatase (59). Thus, the IL-1-induced increase in Tyr-15 phosphorylation on Cdc2 could be due to an increase in Wee1/Myt1 kinase activity and/or due to a decrease in Cdc25 phosphatase activity. Both Myt1 and Cdc25 also control phosphorylation of Thr-14 on Cdc2. It seems possible, therefore, that IL-1 causes an increase in phosphorylation of Thr-14, as well as Tyr-15, on Cdc2, thereby inhibiting cyclin B-Cdc2 activity. Furthermore, we do not exclude the possibility that other post-translational modifications, such as a decrease in Thr-161 phosphorylation on Cdc2, are also involved in the down-regulation of cyclin B1-Cdc2 activity during the early phase of IL-1 treatment.
Recently, it has been reported that growth inhibition of A375-C6 human melanoma cells by IL-1 is mediated at least in part by the suppression of pRb phosphorylation to retain pRb in an unphosphorylated, growth-inhibitory state (15). In the present study using A375S2 cells, however, a shift in electrophoretic mobility of pRb to a hypophosphorylated form was not observed until 24 h after IL-1 treatment, when cell cycle arrest had already been completed. This finding suggests that the observed suppression of pRb phosphorylation at the 24-h time point is the result of IL-1-induced growth arrest, and thus pRb is not a functional mediator of IL-1 action in A375S2 cells. However, since the kinase activity of cyclin E-Cdk2, which is one of the candidates for pRb kinases in vivo, was inhibited in the early phase (Ͻ8 h) of IL-1 treatment, we cannot fully rule out the possibility that the early inhibition of cyclin E-Cdk2 activity caused dephosphorylation of a few phosphorylation sites in pRb that could not appreciably alter the electrophoretic mobility of pRb. In respect to the substrate of cyclin E-Cdk2 kinase, several lines of evidence indicate that cyclin E-Cdk2 catalyzes events that are rate-limiting for the G 1 /S transition that are independent of pRb phosphorylation. For example, cyclin E, but not cyclin D, is essential for entry into S phase in mammalian cells that lack a functional pRb (60 -62). In vitro phosphorylation by Cdk2 kinases had little effect on pRb function in a microinjection-based in vivo cell cycle assay (63). Thus, it is suggested that cyclin E-Cdk2 phosphorylates key substrates other than pRb. Consistent with this hypothesis, here we found that phosphorylation of the pRb-related protein p107 was suppressed from 4 h following IL-1 treatment, coincident with the inhibition of cyclin E-Cdk2 kinase activity, suggesting that the suppression of p107 phosphorylation is mediated through inhibition of cyclin E-Cdk2 activity by IL-1. It is possible, therefore, that p107 acts as a downstream target of the IL-1-induced arrest pathway. We also found that the total protein level of another pRb-related protein, p130, gradually increased in the early phase of IL-1 treatment, whereas the ratio between the hyper-and hypophosphorylated forms of p130 was constant. This finding suggests an additional function of IL-1, up-regulation of the p130 protein level, which also plays an important role in IL-1-induced G 1 arrest.
Both p107 and p130, like pRb, can regulate cell proliferation, and there are a number of cell systems where the retinoblastoma family members have been involved in eliciting cell cycle arrest. The growth arrest mediated by the three pocket proteins are however not identical (64,65), and their relative importance varied with the antiproliferative agent and the cell type employed in the arrest system. For example, it has been shown that pRb plays a central role in cell cycle arrest after DNA damage, whereas p107 and p130 are dispensable for this process (66). Although all of the pRb family members have been reported to be associated with transforming growth factor-␤induced G 1 arrest (67)(68)(69), recent experiments suggest that p130 is the major downstream target in transforming growth factor-␤-regulated growth arrest in gastric-carcinoma cells (45). Muthukkumar et al. (15) have suggested that growth arrest of A375-C6 melanoma cells by tumor necrosis factor-␣ is mediated by a pRb-independent pathway, whereas that by IL-1 is dependent on pRb function. Koudssi et al. (70) have also shown that pRb is involved in IL-1-induced G 1 /S arrest of rat cardiac fibroblasts, although they have not addressed the involvement of p107 or p130. Our data, while in contrast, suggest that IL-1 causes cell cycle arrest not through pRb but by regulating the phosphorylation state of p107 and the abundance of p130. The discrepancy implies that the pRb family proteins involved in IL-1-induced cell cycle arrest could vary with cell lines examined. Although these pRb family proteins share considerable sequence homology, each protein has been shown to have a different temporal profile of interaction with different E2F family members. The p107 and p130 proteins bind specifically to E2F4 and E2F5 (71)(72)(73), whereas pRb interacts with each of the E2F family members (74,75). pRb seems to bind preferentially to E2F in middle to late G 1 and S phases, and p107 forms complexes with E2F predominantly in late G 1 and S phases, while p130-E2F complexes accumulate when cells exit from the cell cycle. Of note, p130 has been suggested to function as the major pocket protein during various differentiation programs (76 -78), and the protein is expressed at high levels in terminally differentiated cells, such as neurons and skeletal muscle (79). It is tempting, therefore, to speculate that the p130 protein induced by IL-1 plays a crucial role in initiation and maintenance of the growth-arrested state. Although the precise role of each individual pRb family protein is not addressed here, the data presented provide a new clue that should help in future investigations of the molecular mechanisms underlying the antiproliferative effect of IL-1.
In conclusion, our findings demonstrate that IL-1 inhibits the growth of A375S2 cells by arresting them at G 1 and G 2 phases of the cell cycle, and the arrests are preceded by a rapid decrease in cyclin E-Cdk2 and cyclin B1-Cdc2 kinase activities. Thus, the rapid down-regulation of these two cyclin-CDK activities is likely to be the mechanism responsible for the cell cycle arrest at G 1 and G 2 phases, respectively. The CDK inhibitor p21 is likely to be responsible for the inhibition of cyclin E-Cdk2 activity, whereas an increase in tyrosine phosphorylation of Cdc2 may be involved in the inhibition of cyclin B1-Cdc2 activity. Furthermore, we found that IL-1 causes rapid dephosphorylation of p107, but not of pRb or p130, while the total protein levels of p130 are increased. Thus, IL-1 may exert its growth-arresting effects not through pRb but by regulating the phosphorylation state of p107 and the abundance of p130. It will be important in our future studies to define the upstream signal transduction pathway mediating the IL-1 action.