p38α and p38γ Mediate Oncogenic ras-induced Senescence through Differential Mechanisms*

Oncogene-induced senescence is a tumor-suppressive defense mechanism triggered upon activation of certain oncogenes in normal cells. Recently, the senescence response to oncogene activation has been shown to act as a bona fide barrier to cancer development in vivo. Multiple previous studies have implicated the importance of the p38 MAPK pathway in oncogene-induced senescence. However, the contribution of each of the four p38 isoforms (encoded by different genes) to senescence induction is unclear. In the current study, we demonstrated that p38α and p38γ, but not p38β, play an essential role in oncogenic ras-induced senescence. Both p38α and p38γ are expressed in primary human fibroblasts and are activated upon transduction of oncogenic ras. Small hairpin RNA-mediated silencing of p38α or p38γ expression abrogated ras-induced senescence, whereas constitutive activation of p38α and p38γ caused premature senescence. Furthermore, upon activation by oncogenic ras, p38γ stimulated the transcriptional activity of p53 by phosphorylating p53 at Ser33, suggesting that the ability of p38γ to mediate senescence is at least partly achieved through p53. However, p38α contributed to ras-inducted senescence via a p53-indepdendent mechanism in cells by mediating ras-induced expression of p16INK4A, another key senescence effector. These findings have identified p38α and p38γ as essential components of the signaling pathway that regulates the tumor-suppressing senescence response, providing insights into the molecular mechanisms underlying the differential involvement of the p38 isoforms in senescence induction.

The ras proto-oncogenes encode small GTP-binding proteins that transduce growth signals from cell surface (1)(2)(3). Aberrant activation of ras is a crucial step in tumor formation. Constitutive activation of ras genes, either through point mutations or overexpression, is associated with a wide variety of human tumors at high frequency and contributes to the initiation and maintenance of multiple tumorigenic phenotypes in these cancers (4 -11). However, in early-passage primary human and rodent cells, activated ras causes a permanent pro-liferative arrest known as premature senescence, because of its phenotypic similarities to replicative senescence observed in late-passage cells (12). Other oncogenes, such as E2F1 and raf, or inactivation of certain tumor suppressor genes, also induce senescence in normal human cells (13)(14)(15). The existence of the premature senescence response to oncogene activation implies that like apoptosis, oncogene-induced senescence serves as an anti-tumorigenic defense mechanism. Indeed, it has been well documented that cellular transformation by ras requires cooperation from immortalizing oncogenes that overcome the senescence response, such as those inactivating p53 (8,16,17). Recent studies have also demonstrated that senescent cells can be detected in early-stage premalignant lesions of lung, pancreas, skin, and prostate in both human cancer patients and mouse tumor models and that disruption of senescence accelerates the development of malignant tumors (18 -23). These findings indicate that oncogene-induced senescence occurs in vivo and serves as a barrier to tumorigenesis.
Although the downstream effectors of the oncogenic activity of ras have been studied extensively, relatively little is known about the signaling pathways that mediate the ras-induced senescence response. Studies have indicated that the ability of ras to induce senescence depends on activation of the Raf/ MEK/ERK MAPK pathway (13,24) and is accompanied by upregulation of several inhibitors of cell proliferation, including p16 INK4A , p53, p14/p19 ARF , and p21 WAF1 (12,25), and silencing of E2F target genes (26). In some cells, senescence is triggered as a result of ras-induced production of reactive oxygen species (27). In addition, it has been reported that oncogene induced senescence is mediated by DNA damage responses generated by aberrant DNA replication (28,29). Recently, studies from our laboratory and others have shown that ras-induced senescence relies on activation of the p38 MAPK 3 (30 -33). p38 and its upstream MAPK kinases MKK3 and MKK6 (34,35) are activated by oncogenic ras as a result of persistent MEK/ERK (mitogen-activated protein kinase/extracellular signal-regulated kinase kinase/extracellular signal-regulated kinase) activation in senescent cells. Constitutive activation of p38 causes premature senescence, whereas pharmacological inhibition of p38 prevents ras-induced senescence (30). * This work was supported, in whole or in part, by Grant CA106768 from the The requirement of p38 for oncogene-induced senescence suggests that the p38 pathway has a tumor-suppressing function, in addition to its previously known roles in inflammatory and stress responses (36 -38). Indeed, target deletion of p38␣ or PRAK, a downstream substrate kinase of p38, accelerates cancer development in mouse models (23,39,40). Moreover, deletion of Wip1, a p38 phosphatase frequently amplified in human breast tumors, leads to p38 activation and reduced mammary tumorigenesis in mice (41,42). Therefore, the p38 pathway is likely to play an important role in tumor suppression by mediating the senescence response to oncogene activation.
Four mammalian isoforms of p38 (␣, ␤, ␦, and ␥), each encoded by a different gene, have been identified; they differ in tissue-specific expression and affinity for the upstream regulatory MAPK kinases (43)(44)(45)(46)(47)(48)(49). Among these isoforms, only p38␣ has been shown essential for inflammatory and stress responses by genetic analysis in murine models (50), whereas the physiological roles of the other p38 isoforms in inflammation or other cellular functions are still unclear (51,52). We have shown previously that SB203580, a chemical compound that inhibits p38␣ and p38␤, prevents ras-induced senescence in primary cells (30), indicating that p38␣/␤ or both might be required for senescence induction. However, this compound also inhibits the activity of other p38 isoforms and other protein kinases although with lower affinity. The specific involvement of each p38 isoform in senescence has never been investigated. In the current study, we examined the role of the p38 isoforms in oncogenic ras-induced senescence in primary human cells. Our data demonstrate that p38␣ and p38␥, but not p38␤, are essential components of the signaling pathway that mediates rasinduced senescence and that p38␣ and p38␥ contribute to senescence induction through different mechanisms. Whereas p38␥ mediates ras-induced senescence at least partly by stimulating the transcriptional activity of p53 through direct phosphorylation, p38␣ appears to regulate senescence in a p53-independent, p16 INK4A -dependent manner.

EXPERIMENTAL PROCEDURES
Cell Culture-BJ human foreskin fibroblasts were maintained in minimum essential medium supplemented with 10% fetal calf serum, non-essential amino acids, glutamine, and antibiotics. WI38 and IMR90 human fibroblasts and LinX-A retroviral packaging cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, glutamine, and antibiotics.
Analysis of Senescence-This was performed in cell cultures by measuring the rate of proliferation and the expression of the senescence-associated ␤-galactosidase (SA-␤-gal) senescence marker as described previously (30). Population doublings (PD) were calculated with the formula PD ϭ log(N2/N1)/log2, where N1 is the number of cells seeded and N2 is the number of cells recovered (59). To quantify SA-␤-gal positives, at least 200 cells were counted in random fields in each of the duplicated wells. Each experiment was performed in triplicates or duplicates.
p53 Reporter Assays-BJ cells were stably transduced with a retroviral luciferase reporter driven by a promoter containing multiple copies of a functional p53-binding sites (PG-Luc) or a mutant p53-binding site (MG-Luc) (57). These cells were transduced with shRNA for GFP, p38␣, or p38␥ at PD28 -32 and subsequently with Ha-RasV12 or vector at PD30 -34. Cells were split into 12-well plates on day 7 or 8 post-ras transduction and lysed on day 8 or 9. Luciferase activity was determined using a luciferase assay system (Promega) according to the manufacturer's instructions and normalized to protein concentrations as determined by the Bradford assay. Each experiment was performed in triplicates or duplicates.
Immunoprecipitation-coupled Kinase Assays for p38-BJ cells were lysed at PD30 -40 on day 6 -8 post-ras/MKK3/6E transduction in a buffer containing 50 mM HEPES, pH 7.5, 2.5 mM EGTA, 1 mM EDTA, 1% Triton X-100, 150 mM NaCl, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 50 mM NaF, 1 mM sodium vanadate, 1 mM ␤-glycerophosphate, 1 mM dithiothreitol, and Complete protease inhibitors. 100 -300 g of lysate were incubated with 60 l of agarose-conjugated anti-p38 Isoforms and Oncogene-induced Senescence FLAG antibody M2 (Sigma) at 4°C for 2 h. The beads were washed three times with 1 ml of lysis buffer and three times with 1ϫ kinase buffer (50 mM HEPES, pH 7.5, 0.5 mM EGTA, 10 mM MgCl 2 , 0.1 mM phenylmethylsulfonyl fluoride, 1 mM NaF, 0.1 mM sodium vanadate, 0.1 mM ␤-glycerophosphate, and 1 mM dithiothreitol). The reactions were performed in 20 l of 1ϫ kinase buffer (above) with 10 M ATP, 0.5 l of [␥ 32 P]ATP, and 10 g of hp53 (1-61) or 2 g of GST-ATF2 at 30°C for 45 min. The reactions were stopped by 7 l of 4ϫ Laemmli buffer, heated at 95°C, and separated by SDS-PAGE. Radioactive signals were detected by using a PhosphorImager. Part of the immunoprecipitates, as well as the total protein lysates, were subjected to Western blot analysis to ensure equal efficiency of immunoprecipitation and equal input of proteins.
Kinase Assays with Recombinant p38-Assays with recombinant kinases were performed in two sequential steps, with the first step being the phosphorylation of His-p38 by MKK6E and the second the phosphorylation of substrates by p38. The first step was carried out at 30°C for 10 min in 14 l of 1ϫ kinase buffer (20 mM Tris-HCl, pH 7.5, 20 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 20 M cold ATP, and 1 mM NaF) containing 0.4 g of His-p38 with or without 50 ng of GST-MKK6. Subsequently, 6 l of substrate mix in 1ϫ kinase buffer (same as above) containing 20 g of hp53 (1-61) (wild type or S33A or S46A mutant) or 10 g of MBP and 2 Ci of [␥-32 P]ATP was added to each reaction. The resulting 20 l of reaction was incubated at 30°C for 30 min, stopped by the addition of 7 l of 4ϫ Laemmli buffer, and heated at 95C°for 10 min. The reactions were separated on 4 -20% gradient SDS-polyacrylamide gels. Radioactive signals were detected by PhosphorImager.

Expression and Activation of the p38 Isoforms during Oncogene-induced Senescence in Primary Human Fibroblasts-We
showed previously that oncogenic ras fails to induce senescence in primary BJ human fibroblast cells treated with the p38␣and p38␤-specific inhibitor SB203580 (30), suggesting that at least one of these two isoforms might be required for senescence induction. However, this compound also inhibits other p38 isoforms and even other protein kinases, although with a lower affinity. To investigate the specific involvement of each p38 isoforms, in the present work we initially examined the expression and activity of these isoform in senescent cells. Although all four p38 isoforms contain similar numbers of amino acid residues, they displayed distinct rates of mobility on SDS-PAGE. Whereas the other isoforms had apparent molecular mass of 38 -42 kDa, p38␥ migrated closely to the 49-kDa marker (Fig. 1A). p38␣ had a slightly faster mobility than p38␤ and p38␦. These differences allowed us to differentiate these isoforms from each other.
Upon transduction of oncogenic ras (HarasV12) or active mutant of MKK3 (MKK3E), the phosphorylation of ectopically expressed p38␣, p38␤, p38␥, and p38␦ in their activation loop was greatly enhanced in BJ cells as detected by a phospho-specific antibody (Fig. 1B). These p38 isoforms also displayed increased protein kinase activity toward ATF2 in vitro when immunoprecipitated from BJ cells transduced with oncogenic ras as compared with those from control cells (Fig. 1C). These results indicate that all four of the p38 isoforms can be activated during ras-induced senescence. Using isoform-specific antibodies, we were able to show that p38␣, p38␤, and p38␥ were expressed in primary BJ human fibroblasts (Fig. 1D). In addi- , or vector, detecting phospho-p38, p38␣, p38␤, and p38␥ using specific antibodies. Cells were lysed on day 8 post-Ras transduction at PD25. Identical sets of lysates, each set containing lysates from BJ cells transduced with vector, Ha-RasV12, or MKK3E were resolved side-by-side on the same SDS-polyacrylamide gel and transferred to a nitrocellulose membrane. The membrane was cut into pieces, each containing one set of lysates. These pieces of membrane were then hybridized to the antibody against phospho-p38, -p38␣, -p38␤, and -p38␥, respectively. The chemiluminescence signals were captured after the membranes were re-aligned into the original position. The positions of p38␣, p38␤, and p38␥ are marked by arrows.
p38 Isoforms and Oncogene-induced Senescence APRIL 24, 2009 • VOLUME 284 • NUMBER 17 tion, both Ras and MKK3E induced the activating phosphorylation of the endogenous p38 proteins that co-migrated with p38␣ and p38␥ (Fig. 1D). We confirmed that the protein bands detected by the phospho-specific antibody and co-migrating with p38␣ and p38␥ indeed represent the phosphorylated p38␣ and p38␥ isoforms, respectively, because these bands were abolished in cells expressing p38␣ ( Fig. 2A) or p38␥ shRNA (Fig. 3A). Therefore, these findings indicate that oncogenic ras activates not only the endogenous p38␣, as we demonstrated previously (30), but also the endogenous p38␥ during senescence induction. Interestingly, in BJ human fibroblasts, oncogenic ras activated p38␥ through phosphorylation without altering its expression level (Fig. 1D). This is in contrast to a previous finding that oncogenic ras induces p38␥ expression but not its phosphorylation in rat intestinal epithelial cells (IEC-6) (63). This raises the possibility that Ras may stimulate the activity of p38␥ through different mechanisms in a speciesor cell type-dependent manner.
Although p38␤ was expressed in primary human fibroblasts, we failed to detect an obvious phospho-p38 band co-migrating with p38␤ (Fig. 1D). However, based on the induction of phosphorylation and kinase activity of ectopically expressed p38␤ by oncogenic ras (Fig. 1, B and C), we reasoned that p38␤ was activated during ras-induced senescence. Because the mobility of p38␤ was only slightly slower than that of p38␣ on SDS-PAGE, it is possible that the signal for phospho-p38␤ was obscured by that of phospho-p38␣ because of the relatively lower abundance of p38␤ as compared with p38␣. Moreover, although the ectopically expressed p38␦ could be activated by both oncogenic ras and MKK3E (Fig. 1, B and C), the level of p38␦ was barely detectable in primary human fibroblasts (data not shown). Thus, our study focused on p38␣, -␤, and -␥ isoforms.
When stably expressed in BJ cells, the p38␣-shRNAs (shp38␣-756 and -758), which efficiently knocked down p38␣, prevented oncogenic ras-induced growth arrest (Fig. 2B) and accumulation of SA-␤-gal, a biomarker for senescence (Fig. 2C). In contrast, shRNA for GFP or the p38␣ shRNA (shp38␣-577), which failed to silence p38␣ expression, had no effect on senescence induction (Fig. 2, B and C). Furthermore, two shRNAs that silenced p38␥ expression (shp38␥-550 and -1023) also blocked the ability of ras to induce growth arrest (Fig. 3B) and greatly inhibited ras-induced expression of SA-␤-gal (Fig. 3C) as compared with the GFP shRNA. On the other hand, the p38␤ shRNA that effectively silenced p38␤ expression did not disrupt oncogenic ras-induced senescence (data not shown). These results were reproduced in WI38 primary human fibroblast cells derived from normal embryonic lung tissue, in which ras-induced senescence was inhibited by shRNA for p38␣ and p38␥ but not by shRNA for p38␤ (supplemental Fig. S1). ShRNA for p38␥ also delayed the onset of ras-induced senescence in the IMR90 primary human lung fibroblasts (data no shown). These results demonstrate that both p38␣ and p38␥ are essential for rasinduced senescence, whereas p38␤ is dispensable for senescence induction.
Constitutive Activation of p38␣ or p38␥, but Not p38␤, Leads to Premature Senescence-To analyze the effect of p38 activation on senescence induction, we took advantage of the intrinsically active mutants of the p38 isoforms recently constructed in Drs. Engelberg and Livnah's groups (Hebrew University of Jerusalem (53,53)). These mutants have acquired spontaneous protein kinase activity in vitro and in vivo and maintain specificity toward substrates and inhibitors similar to that of the wild p38 Isoforms and Oncogene-induced Senescence type p38 isoforms. When transduced into BJ cells via retrovirus, the active form of p38␣ (p38␣-D176A), p38␤ (p38␤-D176A), or p38␥ (p38␥-D179A) and its wild type counterpart were expressed at comparable levels (Fig. 4, A and B). However, the active mutants displayed a much higher level of autophosphorylation in the activation loop than the corresponding wild type proteins as detected by the phospho-specific antibody in Western blot analysis (Fig. 4, A and B). These results were consistent with previous reports in other cell lines and indicated that these mutants of p38 isoforms were indeed constitutively active.
Moreover, similar to oncogenic ras, expression of p38␣-D176A or p38␥-D179A led to growth arrest (Fig. 4C, top and middle panels) and accumulation of the SA-␤-gal marker (Fig.  4D) in BJ cells, whereas wild type p38␣ and p38␥ did not significantly inhibit cell proliferation or induce SA-␤-gal expression. In contrast, the active mutant of p38␤ (p38␤-D176A) did not cause inhibition of proliferation in primary BJ fibroblasts (Fig,  4C, bottom panel). Therefore, constitutive activation of p38␣ or p38␥, but not p38␤, is sufficient to induce premature senescence in primary human fibroblasts.
p38␥, but Not p38␣, Is Required for ras-induced Activation of p53 in Senescent Cells-In an attempt to investigate the molecular mechanism underlying the essential roles of p38␣ and p38␥ in senescence, we examined the ability of these isoforms to regulate the activity of p53, a key effector of rasinduced senescence. For this purpose, we used a retrovirusbased, stable, luciferase reporter system for p53 (57). In BJ cells stably transduced with this stable p53 reporter (PG-Luc), luciferase activity was stimulated significantly (3-4-fold) in the presence of Ha-RasV12 (Fig. 5, A and B, top panels), confirming the induction of p53 transcriptional activity during senescence. However, ras-induced p53 activity was greatly reduced in BJ cells expressing the effective p38␥ shRNA (shp38␥-550 and -1023) (Fig. 5B, top panel), whereas the p38␣ shRNA had no obvious effect on the p53-dependent luciferase activity in senescent cells. Neither ras nor the p38␣ or p38␥ shRNA significantly altered the transcription of MG-Luc, a control reporter containing mutant p53-binding sites (57) (Fig. 5A and  B, bottom panels), indicating that the effects we observed were specific for p53. These data indicate that p38␥, but not p38␣, is required for the activation of p53 during ras-induced senescence.
Consistent with the essential role of p38␥ in p53 activation and in senescence induction, shRNA for p38␥ inhibited rasmediated induction of p21 WAF1 , an endogenous transcriptional target of p53 and a key effector of senescence (Fig. 5C). By contrast, although the effective p38␣ shRNA (shp38␣-756 and -758) blocked ras-induced senescence (Fig. 2, B and C), it did not reduce the p21 WAF1 induction by ras, as compared with the GFP shRNA (Fig. 5D). A similar observation was made in IMR90 cells (Fig. 5E). Moreover, although the constitutively active mutants of both p38␣ (p38␣-D176A) and p38␥ (p38␥-D179A) induced senescence (Fig. 4, C and D), only p38␥-D179A, but not p38␣-D176A, increased the expression of p21 WAF1 (Fig. 5F). The active mutant of p38␤ also failed to stimulate the level of p21 WAF1 (Fig. 5F), consistent with the inability of this mutant to induce premature senescence. Taken together with the results from the luciferase reporter assays, these data demonstrate that although the role of p38␥ in rasinduced senescence correlates with its ability to stimulate p53 activity and subsequently to induce the expression of p21 WAF1 , p38␣ mediates senescence induction through a p53/p21 WAF1independent mechanism.
p38␥, but Not p38␣, Mediates Phosphorylation of p53 in Vivo during Senescence Induction by Oncogenic ras-The activity of p53 is regulated through phosphorylation of its N-terminal transcriptional activation domain. The differential effects of p38␣ and p38␥ on ras-induced p53 activity suggest that these isoforms may play different roles in p53 phosphorylation during senescence induction. It has been shown that Ser 33 and Ser 46 of p53 are direct substrates of p38␣ in vitro and that phosphorylation of these sites contributes to the activation of p53 upon DNA damage (64). The ability of the other p38 isoforms to phosphorylate p53 was unknown. We found that in vitro, recombinant p38␥ phosphorylated the N-terminal transcriptional activation domain of p53, as well as the positive control, MBP, in an MKK6E-dependent manner (Fig. 6A), indicating that p53 is a substrate of activated p38␥. Mutation of Ser 33 to Ala essentially abolished the phosphorylation of p53 by p38␥, whereas mutation of Ser 46 to Ala had little or no effect on p53 phosphorylation (Fig. 6B). In contrast, mutation of either Ser 33 or Ser 46 to Ala greatly diminished phosphorylation of p53 by p38␣ (Fig. 6B). Therefore, p38␥ phosphorylates p53 mainly at Ser 33 in vitro, whereas p38␣ phosphorylates p53 at both Ser 33 and Ser 46 , as demonstrated previously.  APRIL 24, 2009 • VOLUME 284 • NUMBER 17

JOURNAL OF BIOLOGICAL CHEMISTRY 11241
To investigate the phosphorylation of p53 by p38␥ during senescence induction, we compared the p53 kinase activities of p38␣ and p38␥ immunoprecipitated from control and senescent BJ cells. Correlating with its increased phosphorylation in the activation loop (Fig. 1B), p38␥ immunoprecipitated from Ras-and MKK3E-expressing, senescent BJ cells phosphorylated p53 at much higher levels as compared with that from control cells (Fig. 6C, right panel). On the other hand, although the phosphorylation of p38␣ in the activation loop was induced in Ras-or MKK3E-expressing cells to levels comparable with those of p38␥ (Fig. 1B), p38␣ barely phosphorylated p53 when immunoprecipitated from these cells (Fig. 6C, compare left and right panels, which were derived from the same exposure). The phosphorylation of p53 by p38␣ immunoprecipitated from Ras-and MKK3Eexpressing cells could still be detected upon overexposure, but the signals were almost negligible when compared with the signals derived from p38␥ on the same exposure (Fig. 6C, 32 P-p53 overexposure (Over Exp)). These results demonstrate that whereas oncogenic Ras and MKK3E activate both p38␣ and p38␥ during senescence, only activated p38␥, but not p38␣, is able to phosphorylate p53.
The Ras-induced kinase activity of p38␥ toward p53 was almost completely abolished when Ser 33 was mutated, but it was unaltered by the mutation of Ser 15 or Ser 46 (Fig. 6D). Thus, upon activation in senescent cells, p38␥ phosphorylates p53 mainly at Ser 33 , consistent with the results obtained with recombinant p38␥ activated by MKK6E in vitro (Fig. 6B).
We further investigated the effect of p38␣ and p38␥ on phosphorylation of the endogenous p53 protein at Ser 33 in vivo during senescence induction by ras. As with activated ras, intrinsically active p38␥ (p38␥-D179A) induced phosphorylation of p53 at Ser 33 in BJ cells, whereas active p38␣ (p38␣-D176A) did not increase p53-Ser 33 phosphorylation, although it induced senescence (Fig. 5F). Wild type p38␣ and p38␥ had little effect on p53-Ser 33 phosphorylation. Wild type and constitutively active p38␤ also failed to cause phosphorylation of p53-Ser 33 (Fig. 5F), which correlates with their inability to induce senescence. Furthermore, the effective p38␥ shRNA (shp38␥-550 and -1023) greatly diminished ras-induced phosphorylation of p53 at Ser 33 but not the phosphorylation of p53-Ser 15 , a site that is not a p38 substrate (Fig. 5C). However, the effective p38␣ shRNA (shp38␣-756 and -758) blocked ras-induced senescence but not p53-Ser 33 phosphorylation (Fig. 5D). Consistent with these findings in BJ cells, oncogenic ras-induced phosphorylation of p53-Ser 33 and the increase in p21 WAF1 expression were also inhibited by the p38␥ shRNA, but not by the p38␣ shRNA, in IMR90 primary human fibroblast cells (Fig. 5E). Taken together, our data indicate that oncogenic ras activates both p38␣ and p38␥, which in turn mediate senescence induction through different mechanisms. Upon activation by oncogenic ras, p38␥ induces p53 activity by directly phosphorylating Ser 33 , a residue that is required for p53 to mediate ras-induced senescence (23). Activation of p53 by p38␥ leads to increased expression of a key senescence effector, p21 WAF1 . By contrast, p38␣ contributes to senescence induction through a mechanism independent of p53.
p38␣, but Not p38␥, Is Essential for ras-induced Expression of p16 INK4A -We demonstrated previously that oncogenic ras stimulates the transcript level of p16 INK4A , another major effector of senescence, through activation of the p38 pathway (30). To gain insights into the p53-independent role of p38␣ in senescence, we examined the requirement of p38␣ and p38␥ for p16 INK4A expression in senescent cells. As shown previously (30), oncogenic ras induces a 3-fold increase in the mRNA level of p16 INK4A in BJ cells (Fig. 7). However, this induction was abolished by the p38␣ shRNA but not the p38␥ shRNA. Thus, p38␣ and p38␥ mediate oncogene-induced senescence by inducing two major senescence effectors, p53 and p16 INK4A , respectively.

DISCUSSION
The existence of multiple p38 isoforms with differences in tissue distribution and affinity for upstream regulators suggests that these isoforms may have distinct functions. Although the p38␣ isoform has been shown to be required for inflammatory and stress responses in vivo (50,(65)(66)(67), the physiology roles of the other p38 isoforms have been unclear. Using SB203580, a pharmacological inhibitor with relatively higher affinity for p38␣ and p38␤ as compared with p38␥ and p38␦, we were able to demonstrate a key role of p38 in oncogene-induced senescence. However, the relative contribution of the specific p38 isoforms to senescence had never been defined. In the present study, we have shown that p38␣ and p38␥, but not p38␤, mediate senescence induction by oncogenic ras. These studies have identified a novel function of p38␥ in the regulation of oncogene-induced senescence. Taking these findings together with our previous report demonstrating the involvement of p38␥ in ␥-radiation-induced G 2 cell cycle arrest and DNA damage checkpoint control (68), we conclude that a major function of p38␥ may be to suppress tumorigenesis and maintain genome stability.
The requirement of both p38␣ and p38␥ suggests that the functions of these p38 isoforms are not redundant during senescence induction and that p38␣ and p38␥ may target different downstream substrates in the senescence pathway. Indeed, our data demonstrate that p38␣ and p38␥ contribute to senescence induction through different mechanisms, with p38␥ transducing the senescence signal via the p53-p21 WAF1 pathway and p38␣ via a p53-independent, but p16 INK4A -dependent, route. The p53-p21 WAF1 circuit is one of the key effector pathways known to be essential for almost all types of senescence. The ability of p38␥ shRNA to disrupt ras-induced p53-Ser 33 phosphorylation, p53 transcriptional activity, and p21 WAF1 expression, as well as the ability of constitutively active p38␥ to mimic ras in the induction of p53 phosphorylation and activity, indicates that p38␥ mediates senescence induction at least partly by regulating the p53-p21 WAF1 pathway. The p16 INK4A -Rb pathway is the other major effector of senescence. We previously demonstrated that constitutive activation of p38 by active MKK3 or MKK6 leads to  APRIL 24, 2009 • VOLUME 284 • NUMBER 17 increased expression of p16 INK4A at both protein and mRNA levels (30). The results from our current study indicate that ras-induced increase in p16 INK4A expression is mediated by p38␣.

p38 Isoforms and Oncogene-induced Senescence
It has been reported that recombinant p38␣ phosphorylates p53-Ser 33 in vitro (64). We confirmed this finding and further demonstrated that activated recombinant p38␥ also phosphorylated p53 at Ser 33 in vitro as efficiently as p38␣. However, when immunoprecipitated from senescent cells, only p38␥, but not p38␣, could phosphorylate p53-Ser 33 . In addition, ras-induced phosphorylation of p53-Ser 33 in senescent cells was greatly reduced by p38␥ shRNA but not by p38␣ shRNA. Constitutively active p38␥, but not active p38␣, consistently induced p53-Ser 33 phosphorylation in cells. These findings indicate that in cells, phosphorylation of p53-Ser 33 is mainly mediated by p38␥, but not p38␣, upon senescence induction. The mechanism underlying this discrepancy between the in vitro and in vivo activities of p38 toward p53 is currently unknown. It is possible that during ras-induced senescence in vivo, the kinase activity of p38␣ toward p53 is repressed as a result of posttranslational modification or binding to an inhibitory protein. Alternatively, the p53 kinase activity of p38␥ may be enhanced by posttranslational modification or an associated protein in senescent cells.
In vitro, recombinant p38␣ and p38␥ seem to have different affinities for the substrate sites on a same protein. Whereas p38␣ phosphorylates both Ser 33 and Ser 46 of p53, p38␥ phosphorylates only Ser 33 . It has been shown previously that p38␣ and p38␥ have different substrate selectivity in vitro. MAP-KAPK2, MAPKAPK3, and PRAK are preferred substrates of p38␣ over p38␥, whereas p38␥ has higher kinase activity toward the microtubule-associated protein Tau and scaffold proteins SAP90 and SAP97 than does p38␣ (60,69). These differences in substrate selectivity for p53 and other proteins are consistent with the fact that these two isoforms belong to different subgroups within the p38 MAPK family (35,69). P38␥ shares lower identity in amino acid sequence with p38␣ than other isoforms, and the structure of the ATP-binding pocket differs between the ␣ and ␥ isoforms.
Although our study has suggested the importance of the p38␥/p53-Ser 33 /p21 WAF1 cascade in ras-induced senescence, there are almost certainly other pathways that act in a parallel or partially overlapping fashion to mediate senescence induction. Supporting this notion, we found that constitutively active p38␥ increased phosphorylation of p53-Ser 33 to a significantly higher level than oncogenic ras but that the p21 level was induced more robustly by ras as compared with the active p38␥ (Fig. 4A). It is highly likely that besides phosphorylation of Ser 33 by p38␥, oncogenic ras induces additional posttranslational modifications on p53, leading to a further increase in the p53 activity and p21 WAF1 expression. Oncogenic ras may also FIGURE 6. Phosphorylation of p53 by recombinant or immunoprecipitated p38␣ and p38␥ in vitro. A, recombinant p38␥ phosphorylates p53. His-p38␥ was first incubated with GST-MKK6E (ϩ) or buffer (Ϫ) and cold ATP and then with the substrate MBP or p53  in the presence of [␥-32 P]ATP. B, recombinant p38␣ phosphorylates p53 at Ser 33 and Ser 46 , whereas recombinant p38␥ phosphorylates Ser 33 only. His-p38␣ or -p38␥ was first incubated with GST-MKK6E and cold ATP and then with p53 (1-61) (WT, wild type) or p53 (1-61) carrying the S33A or S46A mutation in the presence of [␥-32 P]ATP. C, p38␥ immunoprecipitated from senescent cells displays much higher kinase activity toward p53 than p38␣ does. FLAG-p38␣ or -p38␥ was immunoprecipitated from BJ cells transduced with FLAG-p38␣ or -p38␥ and Ha-RasV12 (Ras), MKK3E, or vector (WH) at PD30 on day 8 post-Ras transduction using an agarose-conjugated anti-FLAG M2 antibody and incubated with p53  in the presence of [␥-32 P]ATP. The same cell lysates as described in the legend for Fig. 1B were used for immunoprecipitation. Part of the immunoprecipitates was subjected to Western blot analysis to detect FLAG-p38. D, p38␥ immunoprecipitated from senescent cells phosphorylates p53 at Ser 33 . FLAG-p38␥ immunoprecipitates from control (WH) or Ras-expressing BJ cells (Ras), as described in C, were incubated with wild type or mutant (S15D, S33D, or S46D) p53  in the presence of [␥-32 P]ATP. A-D, the reactions were separated by SDS-PAGE. Phosphorylated MBP, p53, and p38 were detected by using a PhosphorImager. The input of the substrates was determined by staining with Coomassie Brilliant Blue R. FIGURE 7. p38␣, but not p38␥, is essential for oncogenic ras-induced increase in p16 INK4A mRNA levels. Total RNA was isolated from BJ cells transduced with shGFP, shp38␣-756, or shp38␥-550 and Ha-RasV12 (Ras) or vector (WH) on day 8 after transduction with Ras, separated on an agarose gel, transferred to nylon membrane, and hybridized to a human p16 INK4A cDNA probe labeled by random priming. The signals were visualized and quantified with a PhosphorImager. The numbers represent the relative intensities of p16 INK4A signals from Ras cells after being normalized to the signals from vector control cells.
p38 Isoforms and Oncogene-induced Senescence induce p53-independent signaling pathways that contribute to increased p21 WAF1 expression. It has been shown that senescence induction is accompanied by phosphorylation of other sites on p53 in addition to Ser 33 , such as Ser 15 and Ser 37 (23,25). We found that all of these sites (Ser 15 , Ser 33 , and Ser 37 ) are required for p53 to be able to mediate senescence (23) and for the ras-induced activation of p53 (data not shown). Therefore, it is likely that p53 needs to be phosphorylated at multiple sites to be fully activated during senescence and to function as a senescence effector.
The key role of the p38 pathway in inflammation has prompted efforts to develop anti-inflammatory drugs targeting this pathway. Most such drug candidates currently under development inhibit p38␣. However, the essential role of p38␣ in the tumor-suppressing senescence response to activated oncogenes, as demonstrated in this study, suggests that these drugs would potentially increase the risk of initiating cancer. It is thus imperative to determine the functional specificity of the signaling components of the p38 pathway so that the anti-inflammatory drugs can be designed to target the signaling molecules that are specifically involved in inflammation but not in tumor suppression.