Post-transcriptional Destabilization of p21cip1 by Protein Kinase C in Fibroblasts*

p21cip1 inhibits S phase entry by binding to cyclin-cdk2 (cyclin-dependent kinase-2) complexes. The levels of p21cip1 are rapidly induced after mitogenic stimulation of quiescent fibroblasts and then down-regulate as the cells reach late G1 phase and activate cyclin E-cdk2. In this study, we have shown that pharmacological inhibition of protein kinase C (PKC), expression of dominant negative PKCδ, or knockdown of PKCδ with small interfering RNA elevates p21cip1 protein levels in mouse embryo fibroblasts. This effect is selective, post-transcriptional, and proteasome-dependent but distinct from previously identified post-transcriptional control mechanisms involving cyclin D1 and Skp2. PKCδ inhibition results in a reduced entry into S phase, and this effect is not detected in p21cip1-null cells. Thus, post-transcriptional destabilization of p21cip1 appears to be a major mitogenic effect of PKCδ in fibroblasts.

p21 cip1 is a dual regulator of the cyclin-dependent kinases (cdks). 4 It was initially described as a cdk inhibitor that could prevent the activation of cyclin-cdk complexes or bind to proliferating cell nuclear antigen and repress DNA synthesis (1)(2)(3). Although the cdk inhibitory role of p21 cip1 is now firmly established, subsequent studies have also described a positive role for p21 cip1 as an assembly factor for cyclin D1-cdk4/6 complexes during G 1 phase (4,5).
One of the best-studied aspects of p21 cip1 is its induction during DNA damage and the activation of its promoter by p53 (6). However, p21 cip1 gene expression is also regulated by many signals that are p53-independent (7,8). In addition to the control of gene expression, p21 cip1 levels can be regulated posttranscriptionally through the proteasome (9,10), although the exact mechanism and role of E3 ubiquitin ligases are still under debate. For instance, p21 cip1 can be ubiquitinated, but this modification is not required for the proteosomal degradation of p21 cip1 (11)(12)(13)(14)(15).
We and others have previously reported that the levels of p21 cip1 oscillate during cell cycle re-entry of nontransformed quiescent fibroblasts (8,16,17). The levels of p21 cip1 mRNA and protein are low in quiescent cells, transiently increase as a consequence of mitogen-stimulated cell cycle reentry, and then decline in late G 1 phase. The early G 1 phase induction of p21 cip1 is thought to play a role in the assembly of cyclin D-cdk4/6 complexes (4,5), whereas the subsequent decline in p21 cip1 contributes to the activation of cyclin E-cdk2 (16). Extracellular signal-regulated kinase (ERK) activity is required for the growth factor-dependent induction of p21 cip1 mRNA in early G 1 phase but does not affect the late G 1 phase down-regulation associated with S phase entry (16). Rho is thought to have the opposite effect, suppressing p21 cip1 gene expression and destabilizing the protein to allow for efficient S phase entry, at least in transformed cells (18 -21).
In addition to ERK and Rho, the PKC family of isozymes has an established role in cell proliferation and as regulators of p21 cip1 . However, the biological outcome of PKC stimulation or inhibition is strongly influenced by the isozyme that is regulated and the cell type in which the effect occurs (22,23). For example, PKC isozymes can stimulate or inhibit cell proliferation depending on the cell type (24). In epithelial cells, several PKC isozymes have been reported to increase p21 cip1 levels to promote cell cycle arrest and initiate differentiation or apoptosis (25)(26)(27). In lung adenocarcinoma cells, the p21 cip1 -dependent G 1 arrest mediated by PKC was attributed specifically to PKC␦ (28). These effects of PKC␦ have been typically accompanied by corresponding changes in the levels of p21 cip1 mRNA.
In this study, we have examined the effect of PKC on p21 cip1 levels and S phase entry in mouse embryonic fibroblasts. Our results show that PKC␦ selectively destabilizes p21 cip1 protein levels, and this effect is required for optimal S phase entry. In addition to identifying a novel post-transcriptional effect of PKC, our results emphasize that a single PKC isoform can stimulate or inhibit S phase entry, even when its target is the same cell cycle regulator.

EXPERIMENTAL PROCEDURES
Cell Culture-GF109203X (bisindolyl-maleimide I), LY294002, and MG132 were purchased from Calbiochem. PP1 was purchased from Biomol. All stocks were made in Me 2 SO. PMA was obtained from LC Laboratories and dissolved in ethanol. Spontaneously immortalized mouse embryonic fibroblasts (MEFs) from wild-type and p21 cip1 -null mice were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Unless noted otherwise, MEFs were grown to near confluence in 5% FBS-DMEM and then serum-starved in DMEM-1 mg/ml heat-inactivated, fatty acidfree BSA (DMEM-BSA) for 48 h. The quiescent cells were trypsinized, suspended in DMEM-BSA (1-1.5 ϫ 10 6 cells/10 ml), plated in a 100-mm dish, and stimulated with 10% FBS. In some experiments, the suspended cells were preincubated in suspension with 25 M LY294002, 10 M PP1, 10 M GF109203X, or 10 M MG132 for 30 min at 37°C. The cells were collected and lysed as described previously (29).
To monitor entry into S phase, quiescent wild-type and p21 cip1 -null MEFs were infected with adenoviruses as described below. At the end of the starvation period, the cells were reseeded at 50% confluence in 6-well dishes containing autoclaved glass coverslips and incubated in 2 ml of DMEM-10% FBS with 3 g/ml bromouridine (BrdUrd) (Amersham Biosciences; 1000-fold dilution). Formaldehyde-fixed cells were incubated with anti-BrdUrd as described previously (30). Multiple fields of view (usually 50 -100 cells in 3-4 fields) were counted per sample to determine the percentage of BrdUrdlabeled nuclei.
Adenoviral Infections and Plasmid RNA Interference Transfections-Adenoviruses expressing dominant negative PKC␦ (DN-PKC␦), dominant negative PKC␣ (DN-PKC␣), LacZ, or green fluorescent protein were generated as described in Refs. 31 and 32 and titered using Adeno-X TM Rapid Titer Kit (BD Biosciences). The dominant negative adenoviruses are kinase-dead due to a mutation from Arg to Lys in the ATPbinding site of the catalytic domain. The Skp2 adenovirus was a generous gift of Keiichi Nakayama. Infections with adenoviruses were performed as follows. Near confluent MEFs in 150-mm dishes were incubated in DMEM-BSA for 8 h. Adenoviruses were then directly added to the culture medium. After overnight infection, the medium was replaced with fresh DMEM-BSA, and the cells were starved for an additional 24 h. MEFs were transiently transfected with cyclin D1 RNA interference in pSUPER (kindly provided by Michael Olson) using 5 g of plasmid and Lipofectamine Plus (Invitrogen) as described previously (29). The infected/transfected cells were trypsinized, reseeded at subconfluence, and stimulated with FBS as described above.
Quantitative Real-time Reverse Transcription-PCR-Isolation of total RNA, cDNA synthesis, and PCR was performed as described previously (33). For mouse p21 cip1 , the real-time PCR reaction contained 900 nM forward primer TCC ACA GCG ATA TCC AGA CAT T, 900 nM reverse primer CGG ACA TCA CCA GGA TTG G, and 250 nM probe 6FAM-AGA GCC ACA GGC ACC-minor groove binder-nonfluorescent quencher (MGB-NFQ). For 18 S rRNA, we used 150 nM forward primer CCT GGT TGA TCC TGC CAG TAG, 150 nM reverse primer CCG TGC GTA CTT AGA CAT GCA, and 100 nM probe VIC-TGC TTG TCT CAA AGA TTA-MGB-NFQ. p21 cip1 mRNA and 18 S rRNA levels were quantified by standard curve using ABI Prism 7000 sequence detection system software. The levels of p21 cip1 mRNA were normalized to 18 S rRNA. Duplicate PCR reactions were run for each sample, and results were plotted as mean Ϯ S.D.
PKC Knockdown with Small Interfering RNA (siRNA)-MEFs (ϳ5 ϫ 10 5 ) were seeded in 6-well plates with antibiotic-free DMEM-10% FBS and incubated overnight. The near confluent cultures were washed three times with OptiMEM before transfection in duplicate with either an irrelevant siRNA (GGUAUUGACAGGGAUCUGAtt; Ambion 156130) or siR-NAs specific to mouse PKC␦). The sequences for the PKC␦ siRNAs were CCGUCGUGGAGCCAUUAAAtt (Ambion 151132; called siRNA-A) and CCAUGUAUCCUGAGUG-GAAtt (Ambion 151130; called siRNA-B). The transfection used Lipofectamine 2000 (Invitrogen; 1 l/25,000 cells) diluted in OptiMEM and generally followed the procedure described by the manufacturer. siRNAs were diluted in OptiMEM and used at a final concentration of 300 nM. The final transfection volume was 1 ml. The transfected cells were washed twice in DMEM (4 -6 h after transfection), incubated in antibiotic-free DMEM-10% FBS for 42-44 h, collected, and analyzed.

Inhibition of PKC Up-regulates p21 cip1 Levels as Fibroblasts
Progress into Late G 1 Phase-Our initial studies used pharmacological inhibitors of several kinases to identify signal transducers other than Rho (see Introduction) that control the down-regulation of p21 cip1 during G 1 phase progression in fibroblasts. Inhibition of Src family kinases (SFKs) with PP1 ( Fig. 1A) reduced the induction of p21 cip1 , but when we inhibited SFKs after the initial induction of p21 cip1 occurred, we found p21 cip1 down-regulation was normal in the SFK-inhibited cells (Fig. 1B). Similarly, inhibition of phosphatidylinositol 3-kinase with LY294002 did not affect the down-regulation of p21 cip1 during late G 1 phase (Fig. 1C). In contrast, inhibition of the PKC family with GF109203X, a pharmacological inhibitor that blocks PKC␣, -␤, -␦, and -⑀ (34), prevented the down-regulation of p21 cip1 protein seen in late G 1 phase (Fig. 1D).
Studies in epithelial cells indicate that phorbol ester-sensitive PKCs usually act as anti-mitogens and that their anti-mitogenic effects are due to the stimulation of p21 cip1 gene expression (Introduction). To determine whether PKCs could stimulate p21 cip1 levels in fibroblasts, we compared the levels of p21 cip1 after quiescent MEFs were treated with the phorbol ester PMA (an activator of conventional and novel PKCs), serum, or the combination of PMA and serum ( Fig. 2A). PMA alone did increase the level of p21 cip1 in early G 1 phase of MEFs ( Fig. 2A, 3 h), but this effect was transient, and p21 cip1 levels were barely detectable by late G 1 phase ( Fig. 2A, 9 -18 h). Serum stimulation also resulted in a transient increase in p21 cip1 levels, and PMA did not enhance the duration of the serum-stimulated signal ( Fig. 2A). Although quantitative real-time reverse transcription-PCR did reveal a small stimulatory effect of PMA on p21 cip1 mRNA relative to serum-treated MEFs (Fig. 2B), PMA (alone or in the presence of serum) did not allow for the sustained increase in p21 cip1 gene expression that is characteristic of PMA-stimulated PKC activation in epithelial cells (28). Taken together, the results shown in Figs. 1 and 2 indicate that (i) a pro-mitogenic PKC is effectively activated by serum and inhibited by GF109203X in MEFs, (ii) activation of phorbol ester-sensitive PKCs can stimulate p21 cip1 gene expression in MEFs, but the effect is very short-lived, and (iii) the dominant PKC effect in MEFs is the late G 1 phase reduction in p21 cip1 .
PKC␦ Mediates the Post-transcriptional Down-regulation of p21 cip1 -To identify the PKC family member required for the down-regulation of p21 cip1 , we infected MEFs with adenoviruses encoding LacZ (control) or DN-PKC␣ and -PKC␦, two of the major PKC isozymes targeted by GF109203X and expressed in MEFs. Despite similar levels of expression, DN-PKC␣ did not affect the levels of p21 cip1 during G 1 phase, whereas DN-PKC␦ strongly inhibited the late G 1 phase down-regulation of p21 cip1 (Fig. 3A). Interestingly, the effect of DN-PKC␦ on p21 cip1 protein was not accompanied by changes in the level of p21 cip1 mRNA (Fig. 3B). These results indicate that PKC␦ is required for the post-transcriptional down-regulation of p21 cip1 in late G 1 phase in MEFs. This post-transcriptional regulation of p21 cip1 by PKC␦ that we see in fibroblasts is notably distinct from the transcriptional control of p21 cip1 that is commonly seen in response to altered PKC␦ activity in epithelial cells (28). Selective knockdown of PKC␦ with two different siRNAs increased p21 cip1 levels 2.2-2.6-fold without comparable effect on p21 cip1 mRNA (not shown), confirming that the ␦ isoform of PKC is responsible for the post-transcriptional down-regulation of p21 cip1 (Fig. 3C).   PKC␦ Regulates p21 cip1 Protein Stability-Because we found that PKC␦ regulates p21 cip1 post-transcriptionally, we asked whether it was affecting p21 cip1 protein stability. The half-life of p21 cip1 in MEFs is ϳ30 min (16), and indeed we found that p21 cip1 levels were almost undetectable 2 h after exposure of MEFs to cycloheximide (Fig. 4A). We then infected MEFs with an adenovirus expressing LacZ or DN-PKC␦, blocked protein synthesis with cycloheximide, and determined the stability of p21 cip1 during the subsequent 2 h (Fig. 4B). In LacZ-infected MEFs, p21 cip1 levels decayed with a halflife of ϳ60 min. In contrast, p21 cip1 levels remained nearly constant in cycloheximide-treated MEFs expressing DN-PKC␦ for at least 120 min. A similar, although somewhat less pronounced, effect was detected after siRNA-mediated knockdown of PKC␦ (Fig. 4C). Thus, PKC␦ destabilizes the p21 cip1 protein in MEFs. Consistent with this conclusion, we found that the pronounced effect of DN-PKC␦ on late G 1 phase p21 cip1 levels (Fig. 4D,  compare lanes 2-4 and 6 -8) was lost when the 20 S proteasome was inhibited with MG132 ( Fig. 4D;  compare lanes 9 -11 and 12-14). PKC Stabilizes p21 cip1 Independently of Skp2 and cyclin D1-Some studies have implicated the E3 ubiquitin ligase SCF Skp2 in the degradation of p21 cip1 (35). We therefore treated MEFs with GF109203X or DN-PKC␦ to assess the effect of PKC inhibition on Skp2 expression, the rate-limiting component of the SCF-Skp2 complex. In some experiments, the expression of Skp2 was partially inhibited by GF109203X (Fig. 5A) or DN-PKC␦ (not shown), whereas in other experiments, these treatments had no effect on Skp2 levels (not shown). However, even when these treatments caused a reduction in Skp2, the magnitude of the effect was insufficient to prevent the down-regulation of p27 kip1 (Fig. 5A), the best established substrate of the SCF Skp2 complex. Moreover, ectopic overexpression of Skp2 did not affect the levels of p21 cip1 during G 1 phase progression in either control or GF109302X-treated MEFs (supplemental Fig. 1). We therefore conclude that the effect of PKC on p21 cip1 half-life is independent of its effect on the expression of Skp2. Additionally, these data show that the effect of GF109203X and DN-PKC␦ on p21 cip1 levels are not due to a nonspecific inhibition of G 1 phase progression, because p27 kip1 down-regulates normally in these PKC-inhibited cells.
Coleman et al. (36) propose that p21 cip1 is stabilized by cyclin D1 in Ras-transformed cells, because cyclin D1 competes with the C8␣ subunit of the 20 S proteasome for binding to p21 cip1 . To determine whether cyclin D1 is required for the PKC-dependent destabilization of p21 cip1 , we transiently transfected MEFs with pSuper vectors encoding cyclin D1 RNA interference. We found that p21 cip1 levels were slightly higher after knockdown of cyclin D1, suggesting that cyclin D1 expression  was not stabilizing p21 cip1 in this system. Additionally, p21 cip1 was down-regulated with similar kinetics in both control and RNA interference-treated cells, and GF109203X inhibited the down-regulation of p21 cip1 in both conditions (Fig. 5B). These results indicate that destabilization of p21 cip1 by PKC is independent of cyclin D1.

PKC-dependent S Phase Entry and Cell Proliferation in MEFs-
We asked whether the effect of PKC␦ on p21 cip1 was functionally significant for S phase entry by infecting quiescent wild-type or p21 cip1 -null MEFs with control (LacZ or green fluorescent protein) adenoviruses or the DN-PKC␦ adenovirus. The infected cells were serum-starved and then incubated with FBS in the presence of BrdUrd. DN-PKC␦ inhibited S phase entry by ϳ50% as compared with the infected controls (Fig. 6A). In contrast, DN-PKC␦ was unable to inhibit S phase entry in p21 cip1 -null MEFs (Fig. 6A). Thus, the inhibitory effect of DN-PKC␦ on S phase entry requires p21 cip1 . Consistent with its effect on S phase entry, DN-PKC␦ inhibited the proliferation of MEFs (Fig. 6B), and the magnitude of the proliferative effect was in general agreement with the magnitude of the DN-PKC␦ effect on S phase entry.

DISCUSSION
The down-regulation of p21 cip1 in late G 1 phase is thought to contribute to the activation of cdk2 and entry into S phase. We now report that PKC is required for the post-transcriptional destabilization of p21 cip1 protein in MEFs. S phase entry is reduced by inhibiting PKC␦, and this effect is lost in p21 cip1null MEFs. Thus, a major mitogenic effect of PKC␦ appears to be its destabilizing effect on p21 cip1 protein. Interestingly, similar results have also been reported for Rho (at least in Rastransformed cells); inhibition of Rho destabilizes p21 cip1 protein (18 -21) and inhibits S phase entry in wild-type but not p21 cip1 -null fibroblasts (19). Thus, the p21 cip1 protein is the major cell cycle target for both Rho and PKC␦. However, the Rho effect is not selective for p21 cip1 protein, as Rho inhibition also results in increased expression of p21 cip1 mRNA (19).
The increased level of p21 cip1 protein seen upon inhibition of PKC␦ was dependent on the proteasome. Although one report linked the degradation of p21 cip1 to ubiquitination by the SCF Skp2 complex (35), Sheaff et al. (12) showed that p21 cip1 can be degraded by the proteasome, even when all of its lysines are mutated to arginines to prevent its ubiquitination. Additionally, the N terminus of p21 cip1 appears to be acetylated in vivo, precluding its use as a site for attachment of ubiquitin (15). Consistent with these findings, we did not find a role for the E3 ubiquitin ligase Skp2 in the regulation of p21 cip1 by PKC␦. Skp2 levels were slightly reduced by GF109203X or DN-PKC␦, but this effect is not likely to be functionally significant, because it was not accompanied by a change in the level of p27 kip1 , the major SCF Skp2 substrate. Additionally, the elevated levels of p21 cip1 seen in PKC␦-inhibited cells were not affected by overexpression of Skp2.
Several studies have identified kinases that phosphorylate p21 cip1 , although on different sites, and regulate its stability (37)(38)(39)(40)(41). For example, Akt-dependent phosphorylation has been linked to the cytoplasmic retention and increased stability of p21 cip1 (37)(38)(39). Our data indicate that PKC has the opposite effect, decreasing p21 cip1 levels and enhancing S phase entry. One study reported that a 3-phosphoinositide-dependent protein kinase-1 (PDK)-dependent activation of PKC leads to increased p21 cip1 degradation (40), as does PKC␦ in our studies. However, PDK is not likely involved in our system, because direct inhibition of phosphatidylinositol 3-kinase (which should inhibit PDK activation) did not up-regulate p21 cip1 in MEFs. Moreover, PKC is not targeted by GF109203X or the other approaches we used to inhibit PKC.
Our data reveal specificity in the PKC isoform regulating p21 cip1 stability; the effect is readily detected by inhibiting PKC␦ but not by inhibiting PKC␣. One difference between the two isoforms is that PKC␦ can be tyrosine-phosphorylated by SFKs (42). However, our data indicate that SFK-dependent phosphorylation is not playing a role in regulating p21 cip1 stability, because direct inhibition of SFKs did not mimic the effect of PKC␦ inhibition and up-regulate p21 cip1 . Similarly, the abil- ity of PKC␦ to bind RasGRP3 and enhance ERK activity (43) is unlikely to underlie its effects on p21 cip1 protein stability, because changes in ERK activity regulate p21 cip1 mRNA levels in MEFs (16), and p21 cip1 mRNA levels are not strongly affected by inhibiting PKC␦. Another difference between PKC␣ and PKC␦ is that the ␦ isoform has a much broader subcellular distribution, being detected in mitochondria and in the perinuclear region as well as in the cytosol and plasma membrane (32,44). We have yet to determine whether redirecting PKC␣ to these noncanonical compartments would allow it to regulate p21 cip1 stability. Unfortunately, detailed analysis of the pathway that mediates PKC-dependent p21 cip1 stability is compromised by the fact that the mechanism of p21 cip1 degradation is not yet established (see above).
Although the results shown here indicate that PKC␦ is responsible for the destabilization of p21 cip1 and stimulation of S phase entry in MEFs, we have recently reported that PKC stimulates p21 cip1 gene expression and inhibits S phase entry in transformed epithelial cells (28). In that study, we also observed that the PKC effect on p21 cip1 was dependent on PKC␦ and not PKC␣. Thus, it appears that PKC␦ can have two opposing effects in cells, it can increase the expression of p21 cip1 mRNA, and it can destabilize p21 cip1 protein. The final proliferative effect of PKC␦ will therefore depend on how cells respond to an activated PKC␦ signal. Interestingly, many of the studies showing the anti-mitogenic effect of PKC have been performed with transformed epithelial cell lines (see Introduction). It will be interesting to determine whether the distinct effects of PKC␦ on p21 cip1 reflect different cell lineages (mesenchymal versus epithelial) or the process of cellular transformation.