Osmotic Stress Regulates the Stability of Cyclin D1 in a p38SAPK2-dependent Manner*

We report here that different cell stresses regulate the stability of cyclin D1 protein. Exposition of Granta 519 cells to osmotic shock, oxidative stress, and arsenite induced the post-transcriptional down-regulation of cyclin D1. In the case of osmotic shock, this effect was completely reversed by the addition of p38SAPK2-specific inhibitors (SB203580 or SB220025), indicating that this effect is dependent on p38SAPK2activity. Moreover, the use of proteasome inhibitors prevented this down-regulation. Thus, osmotic shock induces proteasomal degradation of cyclin D1 protein by a p38SAPK2-dependent pathway. The effect of p38SAPK2 on cyclin D1 stability might be mediated by direct phosphorylation at specific sites. We found that p38SAPK2 phosphorylates cyclin D1 in vitroat Thr286 and that this phosphorylation triggers the ubiquitination of cyclin D1. These results link for the first time a stress-induced MAP kinase pathway to cyclin D1 protein stability, and they will help to understand the molecular mechanisms by which stress transduction pathways regulate the cell cycle machinery and take control over cell proliferation.

Mammalian cell cycle progression depends on the sequential activation of different members of a family of serine-threonine kinases named cyclin-dependent kinases (CDKs). 1 The activity of these kinases is positively regulated by cyclin binding, and phosphorylation by the CDK-activating kinase. The activity is also modulated negatively by phosphorylation at specific residues of the CDKs and by the binding of CDK inhibitors (1)(2)(3). Progression through G 1 phase is controlled first by CDK4 and CDK6, which bind combinatorially to cyclins D1, D2, and D3; and later on by CDK2, which associates with cyclin E (4). During G 1 these complexes are responsible for the phosphorylation of different members of the pocket proteins family, which includes the retinoblastoma protein, p107, and p130 (5)(6)(7)(8). The hyperphosphorylation of the pocket proteins leads to the transactivation of genes that are necessary for the onset and progression of DNA replication (9 -11).
Quiescent cells contain low levels of D-type cyclins. After growth factor stimulation, their synthesis is induced, and then cyclin D1-CDK4/6 complexes can be formed during G 1 (12,13). Cyclin D1 expression, assembly of cyclin D-CDK complexes, and their nuclear translocation require the activation of Ras, Raf1, MAP kinase kinase 1/2, ERKs, and the transcription factor c-Ets-2 (14 -17). The maintenance of active cyclin D1-CDK4/6 complexes requires persistent mitogenic signaling, and mitogen withdrawal cancels cyclin D1 synthesis and cyclin D1-CDK4 complexes rapidly dissipate (18). Cyclin D1 turnover is regulated by degradation, mediated by phosphorylation of a specific threonine residue (Thr 286 ) located near the carboxyl terminus. This phosphorylation promotes its polyubiquitination and subsequent degradation by the 26 S proteasome (19). Recently, an alternative mechanism of cyclin D1 ubiquitination, independent of Thr 286 phosphorylation, has been described (20). The Thr 286 phosphorylation of cyclin D1 is catalyzed by glycogen synthase kinase 3 ␤ (GSK3␤), which is active in quiescent cells, but its activity is strongly decreased during proliferation (21). Its inactivation is mediated by site specific phosphorylation by c-Akt (also named protein kinase B), which in turn is controlled by a Ras-activated pathway that includes phosphatidylinositol-3-OH kinase (22,23). Thus, Ras-activated pathways increase cyclin D1 stability by inhibiting GSK3␤ activity in proliferating cells (21).
A variety of stresses induce growth arrest in bacteria and yeast but also in mammalians. Hyperosmolarity causes growth arrest in murine kidney cells, although the mechanisms involved in the proliferation blockade remain unknown (24,25). Low level oxidative stress causes a mitotic arrest by selective activation of MAP kinase kinase 3/6 and p38 SAPK stress signaling pathway (26). Other reports indicate that lipopolysaccharide blocks CSF 1-induced cyclin D1 and CDK4 expression and proliferation in macrophages (27,28).
The cellular MAP kinase modules that are responsible for stress signaling transduction are JNK and p38 SAPK families, but their specific implication in the regulation of cell cycle machinery is still poorly understood. The stress-activated serine-threonine kinase p38 SAPK2 promotes the transcriptional down-regulation of cyclin D1, thus having opposite effects to those triggered by the Ras-activated pathways (14). Cells overexpressing p38 SAPK2 have reduced levels of cyclin D1, whereas blocking p38 SAPK2 activity by the specific inhibitor SB203580 enhances cyclin D1 transcription and protein levels (14). Thus, it is clear that p38 SAPK2 negatively regulates the expression of cyclin D1 at the transcriptional level, although it remains to be established whether p38 SAPK2 modulates cyclin D1 post-transcriptionally.
We report here that different cellular stresses regulate the stability of cyclin D1 protein. Likewise, we demonstrate that, in the case of osmotic shock, this down-regulation is mediated by p38 SAPK2 and is dependent on proteasome degradation. The effect of p38 SAPK2 on cyclin D1 stability might be mediated by direct phosphorylation at specific sites. We found that p38 SAPK2 phosphorylates cyclin D1 in vitro at Thr 286 , and this phosphorylation triggers its ubiquitination and targets it for degradation by the proteasome. These results link for the first time a stress-induced MAP kinase pathway to cyclin D1 protein stability, and they will help in understanding the molecular mechanisms by which stress transduction pathways regulate the cell cycle machinery and take control over cell proliferation.

EXPERIMENTAL PROCEDURES
Cell Cultures-Granta 519 cell line was obtained from the German Tissue Bank and was grown in Dulbecco's modified Eagle's medium (Biological Industries) with 10% fetal calf serum under a 10% CO 2 atmosphere. Molt-4 lymphoblastoid cell line was grown in RPMI 1640 (Biological Industries) supplemented with 10% fetal calf serum. Osmotic shock was performed by the addition of 50 mM NaCl, 50 mM CaCl 2 , or 50 mM MgCl 2 to the culture media. Oxidative stress and arsenite treatment were performed by the addition of 500 M of H 2 O 2 or sodium arsenite (NaAsO 2 ) (Sigma), respectively, to the culture media. The specific p38 SAPK2 inhibitors SB203580 and SB220025 (Calbiochem) were used at 40 M final concentration, and the proteasome inhibitors were used at final concentrations of 100 M for acetyl-leucinyl-leucinylnorleucinal (aLLnL) (Sigma), 10 M for lactacystin, and 50 M for MG132.
For the expression and purification of all these recombinant proteins, the resulting plasmids were transformed into Escherichia coli BL21 (DE3) strain carrying the pLysS plasmid. The expression and purification was performed as described in Ref. 32 with minor modifications. After purification, all recombinant proteins were resuspended in Kinase Buffer (25 mM HEPES, pH 7.4, and 10 mM MgCl 2 ).
p38 SAPK2 Activation in Molt-4 Cells-To activate p38 SAPK2 in Molt-4 cells, 0.4 M NaCl was added to the cell culture, and 10 min later cells were harvested and washed once with cold phosphate-buffered saline. The activity of p38 SAPK2 was determined by Western blotting using a specific phospho-p38 antibody (New England Biolabs 9211S) that recognizes only the phosphorylated form of p38␣ SAPK2a (phospho-Thr 180 -Tyr 182 ).
Immunoprecipitation (IP) and Kinase Assays-Cells were lysed in IP Buffer (50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 5 mM EDTA, 50 mM NaF, and 0.1% Triton X-100), containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 5 g/ml leupeptin, 10 mM glycerophosphate, and 1 mM Na 3 VO 4 for 30 min on ice. Lysates (1 mg) were immunoprecipitated with 1 g of rabbit-anti-p38 antibody that recognizes the N-terminal part of p38␣ SAPK2a and to a lesser extent p38␤ SAPK2b (Santa Cruz, SC-728) or 1 g of rabbit-anti-p38 antibody antibody that recognizes the C-terminal part of p38␣ SAPK2a (Upstate Biotechnology Inc., 06-620) overnight at 4°C. Protein A-Sepharose (Pierce) (10 l) was added and the sample incubated for 1 h at 4°C. After that, it was washed three times with IP buffer and one time with Kinase Buffer. The immunoprecipitates were used as a source of active p38 SAPK2 . The kinase assays were performed in Kinase Buffer containing 2 mM dithiothreitol and 30 M unlabeled ATP. The reactions were initiated by the addition of 0.1 Ci/l [␥-32 P]ATP (Amersham Pharmacia Biotech) and 1 M substrate, and then they proceeded during 30 min at 37°C. The samples were then electrophoresed, and the gels were stained with Coomassie Blue, dried, and exposed to x-ray films at Ϫ80°C. Similar experiments were performed using purified active GST-p38␣ SAPK2a , instead of immunoprecipitated kinase. Purified active GST-p38␣ SAPK2a was obtained from Upstate Biotechnology Inc. . In these experiments, 25 ng of GST-p38␣ SAPK2a active kinase were used and the kinase assay was performed exactly as described above.
Phosphoamino Acid Analysis-First, phosphorylation reactions were performed using 200 ng of recombinant active kinase and 2 g of recombinant GST-cyclin D1 as substrate in Kinase Buffer containing 2 mM dithiothreitol. Reactions were initiated by the addition of 2 Ci/l [␥-32 P]ATP and performed during 30 min at 37°C. After stopping the reaction with SDS-sample buffer, the samples were electrophoresed and the proteins transferred to Immobilon membranes. The band corresponding to phosphorylated GST-cyclin D1 was excised and treated with 6 N HCl at 110°C for 1.5 h. The samples were then lyophilized, and the hydrolyzed amino acids were resuspended in the presence of 1 mg/ml phosphoserine, phosphothreonine, and phosphotyrosine (Sigma), which were used as mobility standards. Finally, the samples were subjected to two-dimensional running (1000 V, 80 min, pH 1.9; and 1000 V, 80 min, pH 3.5).
In Vitro Ubiquitination Assay-GST-cyclin D1 was phosphorylated with recombinant p38 SAPK2 as described above but using cold ATP instead of [␥-32 P]ATP. After phosphorylation reaction, 60 l of reticulocyte cell lysate (Promega) were added to the total volume of reaction as a source of ubiquitin and ubiquitination enzymes. The mixture was incubated for 1 h at 37°C, and then the recombinant GST-fusion substrates were re-purified using glutathione-agarose beads (Amersham Pharmacia Biotech) as described by the manufacturer. The GST-fusion substrates were separated by SDS-polyacrylamide gel electrophoresis, and the presence of ubiquitinated cyclin D1 forms was analyzed by Western blotting using an anti-GST antibody (Santa Cruz, SC-138) and an anti-ubiquitin antibody (Sigma U-5379).

Different Cellular Stresses Regulate
Cyclin D1 Stability-To study whether cellular stresses regulate the stability of cyclin D1 protein, the effect of osmotic shock, oxidative stress, and sodium arsenite on cyclin D1 protein levels was analyzed in proliferating Granta 519 cells. This cell line was derived from a case of high grade non-Hodgkin's lymphoma, has a mature B cell immunophenotype, and harbors a translocation t(11; 14)(q13;q32) (33). This translocation is one of the most frequent alteration in mantle cell lymphomas and juxtaposes the bcl-1 (cyclin D1) locus to immunoglobulin gene sequences that lead to deregulation of cyclin D1 transcriptional expression (34). Thus, Granta 519 cells overexpress cyclin D1 in a constitutive manner, being a helpful model to analyze the post-transcriptional regulation of cyclin D1. Osmotic shock (50 mM NaCl, 50 mM CaCl 2 , or 50 mM MgCl 2 ), oxidative stress (500 M H 2 O 2 ), or arsenite (500 M NaAsO 2 ) induced a down-regulation of cyclin D1 protein levels (Fig. 1). Interestingly, cyclin D1 protein decrease induced by osmotic stresses was higher than those observed by oxidative stress or arsenite. To study whether degradation was implicated in these cyclin D1 protein decreases, three proteasome inhibitors (aLLnL, lactacystin, or MG132) were added to stressed Granta 519 cells. In all cases, the proteasome inhibitors prevented the down-regulation of cyclin D1 and even caused different levels of accumulation of this protein (Fig. 1). The different accumulation might be due to the constitutive transcription of cyclin D1 in these cells and to the different specificity of the inhibitors. Thus, these cellular stresses down-regulate cyclin D1 by increasing its degradation by the proteasome.
To analyze the involvement of stress MAP kinase pathways in the down-regulation of cyclin D1, we first determined the activation of p38 SAPK2 in Granta 519 cells subjected to all those stress treatments. Results revealed that all the treatments activated p38 SAPK2 as determined by Western blotting using an anti-phospho-p38 SAPK2 antibody that specifically recognizes the active (dually phosphorylated) form of p38 SAPK2 (data not shown). Then, a time-course analysis of cyclin D1 protein levels in Granta 519 cells subjected to osmotic stress (50 mM NaCl) were performed in the presence or in the absence of the specific p38 SAPK2 inhibitor SB203580. Results indicated that cyclin D1 decrease was very rapid and that, when cells were pre-incubated with SB203580, this effect was completely reversed (Fig.  2). To further demonstrate that this effect was specific for p38 SAPK2 activity, another chemically different and more potent p38 SAPK2 inhibitor SB220025 was used (35,36). Results indicated that this inhibitor also reversed cyclin D1 decrease. The addition of the proteasome inhibitor aLLnL to the incubation medium prevented the osmotic shock-derived cyclin D1 decrease. These results indicate that osmotic shock induces a p38 SAPK2 -dependent cyclin D1 decrease by triggering its proteasomal degradation. Similar experiments performed with the same cells under oxidative stress (500 M H 2 O 2 ) or arsenite treatment (500 M NaAsO 2 ) revealed that SB220025 or SB203580 only produced a very low reversion of cyclin D1 degradation (data not shown).
These results indicate that, in these cases, p38 SAPK2 protein did not significantly mediate cyclin D1 degradation.
p38 SAPK2 Phosphorylates Cyclin D1 in Vitro-Since cyclin D1 stability is regulated by direct phosphorylation at threonine 286 (19, 21), we analyzed whether p38 SAPK2 was able to directly phosphorylate cyclin D1. Thus, we performed an in vitro kinase assay using recombinant GST-cyclin D1 as a substrate. Active p38 SAPK2 was immunoprecipitated from osmotically shocked Molt-4 lysates by using an antibody that specifically recognizes the N-terminal sequence of p38 SAPK2 as described under "Experimental Procedures." The immunoprecipitated p38 SAPK2 was active as checked by Western blotting using an anti-phospho-p38 SAPK2 antibody that specifically recognizes the active (dually phosphorylated) form of p38 SAPK2 (data not shown). In vitro kinase reactions demonstrated that p38 SAPK2 efficiently phosphorylated purified GST-cyclin D1 but not purified GST-CDK4, GST-CDK2, GST-cyclin A, or GST (Fig. 3A). Myelin basic protein that was used as a control substrate was also efficiently phosphorylated by p38 SAPK2 . Similar results were obtained when the immunoprecipitation was performed using another anti-p38 SAPK2 antibody raised against a C-terminal sequence (data not shown). To confirm that no other kinase than p38 SAPK2 was involved in this phosphorylation, kinase reactions were performed using purified active GST-p38␣ SAPK2a instead of that obtained by immunoprecipitation. Purified active GST-p38␣ SAPK2a also phosphorylated GST-cyclin D1 but not GST-CDK4, GST-CDK2, GST-cyclin A, or GST. Under these experimental conditions, cyclin D1 phosphorylation was abolished by the presence of SB203580 (Fig. 3B). The affinity constant of p38 SAPK2 versus cyclin D1 was measured by kinetic analysis. Results showed that the K m was 288 nM (Fig. 4). These results indicate that cyclin D1 is a specific substrate for p38 SAPK2 , and its high affinity, compared with those of p38s to other substrates, suggests that cyclin D1 phosphorylation by this kinase may have physiological relevance (see "Discussion").
Identification of the Cyclin D1 Amino Acid Residues Phosphorylated by p38 SAPK2 -Since p38 SAPK2 is a proline-directed serine/threonine kinase, TP or SP motifs were searched in the primary sequence of D-type cyclins. Three putative phosphorylation sites for proline-directed kinases were found in the sequence of cyclin D1. Two of them are threonines (Thr 156 and Thr 286 ), and the other one is a serine (Ser 219 ). Only one of these three sites (Thr 286 ) is conserved in all D-type cyclins. The Thr 156 site is conserved in both cyclin D1 and cyclin D2, whereas the Ser 219 site is only present in cyclin D1.
To analyze whether these sites may be phosphorylated by p38 SAPK2 kinase assays were performed using two cyclin D1- excluding fragments. One of the fragments, cyclin D1-(1-200), contained one of the three putative phosphorylation sites (Thr 156 ) whereas the other fragment, cyclin D1-(201-295), contained the other two putative sites (Ser 219 and Thr 286 ). Kinase reactions performed in vitro demonstrated that both the immunoprecipitated p38 SAPK2 from Molt4 lysates and the purified active GST-p38␣ SAPK2a efficiently phosphorylated both cyclin D1 fragments (Fig. 5). These results indicate that both cyclin D1 fragments contain at least one phosphorylation site for p38 SAPK2 .
Cyclin D1 amino acid residues phosphorylated by p38 SAPK2 were determined by phosphoamino acid analysis. GST-cyclin D1 was phosphorylated by the purified active GST-p38␣ SAPK2a and subjected to acid hydrolysis. The amino acids were then recovered and subjected to two-dimensional electrophoresis. Results revealed that threonine was highly phosphorylated, whereas only slight phosphorylation of serine was observed (Fig. 6). These results suggested that Thr 156 and Thr 286 are good phosphorylation site candidates. To confirm this possibility, these two threonines were mutated to alanines. Then, we expressed the two fragments of cyclin D1 containing the mentioned mutations (cyclin D1-(1-200)-T156A and cyclin D1-(201-295)-T286A) and in vitro kinase assays were performed.
These kinase reactions revealed that both the immunoprecipitated p38 SAPK2 from Molt4 lysates and the purified active GST-p38␣ SAPK2a efficiently phosphorylated the wild type cyclin D1 fragments but not the mutated fragments (Fig. 7). These results clearly confirm that the sites of p38 SAPK2 phosphorylation were both Thr 156 and Thr 286 .
To determine whether both sites where phosphorylated by p38 SAPK2 in the full-length cyclin D1, we expressed two fulllength cyclin D1 forms, one containing an Ala for Thr 286 substitution and another one containing these substitution plus an additional Ala for Thr 156 substitution. Kinase reactions were performed using these two mutated forms of the full-length cyclin D1. Interestingly, results revealed that both the immunoprecipitated p38 SAPK2 from Molt4 lysates and the purified active GST-p38␣ SAPK2a did not phosphorylate neither of the two mutated forms (cyclin D1-T286A and cyclin D1-T156A-T286A) (Fig. 8). Thus, the substitution of Thr 286 by Ala completely abolished the phosphorylation by p38 SAPK2 , indicating   FIG. 3. In vitro phosphorylation of cyclin D1 by p38 SAPK2 . Kinase assays were performed using p38 SAPK2 obtained by IP from Molt-4 cell lysates (A) or purified active GST-p38␣ SAPK2a (B). A, IPs were performed using a polyclonal anti-p38 SAPK2 antibody (␣-p38) or a normal rabbit serum (NRS) as a control. GST-cyclin D1, GST-CDK4, GST-CDK2, GST-cyclin A, and GST were used as substrates as indicated. Arrows mark phosphorylated substrates, whereas lines mark those not phosphorylated. B, kinase assays using purified active GST-p38␣ SAPK2a as kinase and GST-cyclin D1, GST-CDK4, GST-CDK2, GST-cyclin A, and GST as substrates were performed in the presence or absence of the p38 SAPK2 inhibitor SB203580 (20 M) as indicated. Assays using myelin basic protein (MBP) as a substrate were performed as a control. Note that GST-p38␣ SAPK2a became autophosphorylated. Purified GST-p38␣ SAPK2a was used to extensively phosphorylate purified GST-cyclin D1 in an in vitro phosphorylation assay. Phosphorylated GST-cyclin D1 was subjected to hydrolysis with HCl as described under "Experimental Procedures," and the phosphorylated amino acids separated by two-dimensional electrophoresis. The phosphoamino acids were visualized by autoradiography, and their relative position was determined by ninhydrin staining. that, in the full-length cyclin D1, this is the major site of phosphorylation. Thr 156 is not phosphorylated in the fulllength cyclin D1, although it is a good phosphorylation site when cyclin D1 is fragmented.
Cyclin D1 Phosphorylation by p38 SAPK2 Triggers Its Ubiquitination in Vitro-Phosphorylation of cyclin D1 in Thr 286 by GSK3␤ triggers its ubiquitination and degradation by the 26 S proteasome (21,37). Thus, we analyzed whether phosphorylation of this site by p38 SAPK2 also triggers the ubiquitination of cyclin D1. In vitro kinase reactions were performed using the wild type and T286A-mutant forms of cyclin D1 as substrates and the purified active GST-p38␣ SAPK2a as a source of kinase. Then, an in vitro ubiquitination assay was performed using the total volume of the kinase assay as a source of phosphorylated and unphosphorylated cyclin D1 and a reticulocyte cell lysate as a source of ubiquitin and ubiquitination enzymes (see "Experimental Procedures"). After that, the presence of ubiquitinated cyclin D1 forms was analyzed by Western blotting and results revealed that the phosphorylation of cyclin D1 by p38 SAPK2 promotes its ubiquitination (Fig. 9). This ubiquitination was abolished when SB203580 was added to the kinase reaction or when the mutated form of cyclin D1 (T286A) was used as substrate. Thus, these data revealed that the specific phosphorylation of cyclin D1 at Thr 286 by p38 SAPK2 promotes its ubiquitination in vitro and targets it for degradation by the proteasome.

DISCUSSION
The regulation of the levels of D-type cyclins is a critical step for G 1 progression. Quiescent cells contain low levels of D-type cyclins, and mitogenic stimuli induce their increase by transcriptional and post-transcriptional mechanisms. Studies on the post-transcriptional regulation of cyclin D1 indicate that its phosphorylation has special relevance on the turnover of the protein. During G 1 , cyclin D1 turnover is rapid (t1 ⁄2 ϳ20 -30 min), whereas that of its catalytic subunit CDK4 is relatively slow (t1 ⁄2 ϳ4 h) (19,21,(37)(38)(39). Moreover, cyclin D1-CDK4 complexes exist in a dynamic equilibrium between bound and free cyclin D1 that permits continuous replacement of the cyclin subunit of the complex (38). For those reasons, slight changes in cyclin D1 protein levels can rapidly modulate cyclin D1-CDK4/6 activity; thus, the regulation of cyclin D1 protein stability is a crucial step for G 1 progression.
Cyclin D1 degradation is regulated by GSK3␤, which catalyzes the phosphorylation of a specific threonine residue (Thr 286 ), and this triggers its polyubiquitination and subsequent degradation by the 26 S proteasome (21). The low levels of cyclin D1 in quiescent cells are due to the low rate of synthesis and to the high instability of the protein. This instability is probably due to the high activity of GSK3␤ in cells deprived of growth factors (40). Mitogenic stimuli induce Ras-dependent signaling pathways that increase cyclin D1 transcription and also down-regulate GSK3␤ preventing cyclin D1 phosphorylation and degradation. These two mechanisms give a precise and fast control of the total amount of cyclin D1 protein that correlates with cyclin D1-CDK4/6 kinase activity in the cell.
Results reported here revealed that different cellular stresses regulate cyclin D1 stability in vivo. For these studies proliferating Granta 519 cells were used because they have an abnormal constitutive expression of cyclin D1. This characteristic makes this cell line very useful to analyze the post-transcriptional regulation of cyclin D1. Osmotic shock and, to a FIG. 8. In vitro phosphorylation of mutated forms of cyclin D1 by p38 SAPK2 . Phosphorylation assays were performed using p38 SAPK2 obtained by IP of Molt-4 cell lysates (A) or purified recombinant p38 SAPK2 (B). A, IPs were performed using a polyclonal anti-p38 SAPK2 antibody (␣-p38) or a normal rabbit serum (NRS) as a control. GSTcyclin D1, GST-cyclin D1-T286A, and GST-cyclin D1-T156A-T286A were used as substrates as indicated. B, kinase assays using purified active GST-p38␣ SAPK2a as kinase and GST-cyclin D1, GST-cyclin D1-T286A, and GST-cyclin D1-T156A-T286A as substrates were performed. Experiments in the presence of the p38 SAPK2 inhibitor SB203580 (20 M) were performed as a control. Note that GST-p38␣ SAPK2a became autophosphorylated.
FIG. 9. In vitro ubiquitination assay using phosphorylated and un-phosphorylated forms of cyclin D1 as substrates. GST-cyclin D1 or GST-cyclin D1-T286A were phosphorylated in vitro by purified active GST-p38␣ SAPK2a . Samples were then subjected to a ubiquitination assay as described under "Experimental Procedures" and then analyzed by Western blotting using an anti-GST antibody or an antiubiquitin antibody as indicated. In both panels, lanes 1 correspond to the unphosphorylated GST-cyclin D1; lanes 2, to GST-cyclin D1 phosphorylated by p38 SAPK2 in the presence of SB203580; lanes 3, to GSTcyclin D1 phsophorylated by p38 SAPK2 ; and lanes 4 to the mutant form of cyclin D1 (GST-cyclin D1-T286A) phosphorylated by p38 SAPK2 . lesser extent, oxidative stress or arsenite down-regulated cyclin D1 total protein levels in these cells. This effect was due to the increase of cyclin D1 degradation by proteasome, because the addition of different proteasome-specific inhibitors blocked cyclin D1 decrease.
Although all these different stresses induced the activation of p38 SAPK2 , only in the case of osmotic stress was the downregulation of cyclin D1 clearly mediated by this kinase. This differential response might be due to the fact that the different stresses can activate distinct additional signaling pathways, which can differentially participate in the modulation of cyclin D1 degradation. Thus, it has been reported that oxidative stress also induced the protein kinase B, ERK, and JNK pathways (41)(42)(43). It has also been shown that arsenite activates ERK and JNK pathways in addition to p38 SAPK2 (44). A possible involvement of these different pathways might explain why cyclin D1 degradation was not reversed by p38 SAPK2 inhibitors in these cases. Alternatively, the participation of SB203580insensitive isoforms of p38 (p38␦ or/and p38␥) might also explain the lack of reversion by these inhibitors.
p38 SAPK2 is a member of the MAP kinase family that is activated by environmental stresses such as osmotic shock and oxidative stress, but also by different agents as arsenite, proinflammatory cytokines or UV light (45). p38 SAPK2 phosphorylates different protein substrates as GADD153 (46), transcription factors such as ATF2 (47), and protein kinases as MAP kinase-activated protein kinases 2 and 3 (48 -50).
Taking together previous reports and results reported here, it appears that p38 SAPK2 may be a dual regulator of cyclin D1 levels, because it inhibits cyclin D1 transcription (14), and it triggers cyclin D1 degradation in cells under osmotic stress. Thus, p38 SAPK2 regulates cyclin D1 transcription and degradation in the opposite way to Ras during proliferation. These results suggest that stress activated pathways may counteract the effects of mitogen-induced pathways as has been described in Ref. 51. In fact it is known that a variety of stresses induce growth arrest in bacteria, yeast, and mammalian cells. For instance, osmotic stress blocks proliferation in murine kidney cells (24,25) and lipopolysacharide blocks CSF 1-stimulated macrophage proliferation (27,28). Consistent with this antimitogenic action, lipopolysaccharide inhibits CSF 1-induced cyclin D1 and CDK4 expression and pRB phosphorylation (52,53). Osmotic stress and lipopolysaccharide are potent activators of p38 family members. Thus, it is possible that cell cycle arrest in both cases could be mediated by cyclin D1 degradation induced by p38 SAPK2 .
We have demonstrated that p38 SAPK2 phosphorylates cyclin D1 in vitro by using enzymes obtained from two different sources. On one hand p38 SAPK2 was obtained by immunoprecipitation of Molt4 cell extracts using specific anti-p38 SAPK2 antibodies. On the other hand, a recombinant purified kinase was used. In both cases a high affinity cyclin D1 phosphorylation (K m 288 nM) was observed. This affinity constant was much lower than those reported for other p38 SAPK2 protein and peptide substrates (ranging from 6.2 to 840 M) (54 -56). Cyclin D1 phosphorylation by p38 SAPK2 occurs specifically at Thr 286 as indicated by the fact that a mutant cyclin D1 Thr 286 3 Ala is not phosphorylated by p38 SAPK2 . Interestingly, other members of the MAP kinase family including JNK and ERK2 are not able to phosphorylate cyclin D1 in vitro (21), suggesting that phosphorylation of cyclin D1 by p38 SAPK2 is specific and could have physiological relevance.
p38 SAPK2 phosphorylation of cyclin D1 triggers its ubiquitination in an in vitro ubiquitination assay. Previous reports indicate that phosphorylation of cyclin D1 at Thr 286 promotes its ubiquitination and proteasomal degradation (19). This phos-phorylation is catalyzed by GSK3␤, which triggers in vivo cyclin D1 polyubiquitination and degradation by the proteasome (21). Taken together, these data and results reported here suggest that p38 SAPK2 may regulate the stability of cyclin D1 in vivo by phosphorylation at Thr 286 and subsequent proteasomal degradation.
Results obtained from the in vitro kinase assays, using the full-length cyclin D1, indicate that the only phosphorylated site was Thr 286 . These results are similar to those described in Ref. 21, using GSK3␤. However, another amino acid residue (Thr 156 ) was also phosphorylated when the protein was truncated. These results suggest that the Thr 156 position is masked in the free full-length form of cyclin D1 but it is accessible when protein is truncated by elimination of the carboxyl-terminal fragment containing the last 95 amino acid residues. Thus, Thr 156 seems not to be accessible for p38 SAPK2 in the free form of cyclin D1. However, it remains to be explored whether in other physiological conditions, for instance in cyclin D1-CDK4 complexes, this residue can be unmasked and consequently physiologically phosphorylated by p38 SAPK2 .
In summary, results reported here indicate that different types of stresses down-regulate cyclin D1 levels by proteasomal degradation; in the case of osmotic stress, this degradation is mediated by p38 SAPK2 . The in vitro experimental analysis of cyclin D1 phosphorylation and ubiquitination suggests that this degradation might be induced by direct phosphorylation at Thr 286 of cyclin D1 by p38 SAPK2 . The analysis of the physiological relevance of the phosphorylation of cyclin D1 by p38 SAPK2 is currently under way in our laboratory.