|Home | About | Journals | Submit | Contact Us | Français|
The stability of p21, a cyclin-dependent kinase inhibitor, is highly regulated by various protein molecules through the cell cycle and in response to extracellular signals. One of the p21 regulators is MDMX, which can directly bind to p21 and mediate its proteasomal degradation in an ubiquitination-independent fashion. The fact that 14-3-3γ binds to the MDMX domain adjacent to p21 binding suggests that this 14-3-3γ may affect MDMX-mediated p21 proteasomal turnover. Indeed, we found that overexpression of 14-3-3γ increased the level of both endogenous and exogenous p21 in p53-null cells by extending its half-life, leading to p21-dependent G1 arrest. Also, 14-3-3γ excluded p21 from binding to MDMX in a dose-dependent manner as determined by co-immunroprecipitation in vitro using purified proteins and in cells. In response to DNA damage, the level of the 14-3-3γ-MDMX complex increased whereas that of the MDMX-p21 complex declined as detected by co-immunoprecipitation assays, leading to the induction of p21 in p53-null cells. Knockdown of 14-3-3γ inversely alleviated the induction of p21 levels by DNA damage. Hence, our study as presented here unravels a new role for 14-3-3γ in protecting p21 from MDMX-mediated proteasomal turnover, which may partially account for DNA damage-induced elevation of p21 levels independent of p53.
The cell cycle (defined into G1, S, G2, and M phases) essential for stem cell proliferation, embryogenesis, animal development, and tumor growth is highly regulated by a number of cellular proteins (1,–4). Post-completion of their functions during the cell cycle, these proteins are often disposed through ubiquitin-dependent or independent proteolysis executed by either 26 S or 20 S proteasome (5, 6).
One of these short living cell cycle regulatory proteins is p21 (p21cip1/waf1) which acts to pause cycling cells mostly at G1 and sometime at G2 by directly binding to cyclin-dependent kinases and inhibiting their activity (7,–10). The protein level of p21 is oscillated during the cell cycle (11,–13). Both ubiquitin-dependent and independent mechanisms have been shown to mediate p21 proteasomal turnover during the cell cycle or in response to specific signals (14,–17). On one hand, several E3 ubiquitin ligases, such as Skp2, MKRN1, or CRL4Cdt2 were shown to ubiquitinate p21 and mediate its degradation (14, 18,–20). On the other hand, it has been also shown that p21 can be discarded through ubiquitin-independent proteasomal pathways. For example, it was previously shown that p21 is degraded directly by 20 S proteasome without involvement of ubiquitination (21,–23). Also, two p53s suppressors, MDM2 and MDMX (24, 25), can directly bind to p21 and mediate its ubiquitin-independent proteolysis by 26 S proteasome during the cell cycle (15, 17, 26). Their actions can be independent of, though also cooperative with each other (26). Furthermore, REGγ, a proteasome activator, has been reported to facilitate ubiquitin- and ATP-independent proteasomal degradation of p21 (27, 28). Therefore, it becomes clear that the stability of this critical cell cycle regulator p21 is subjected to the regulation by multiple cellular proteins. Whether the process of its terminal fate requires ubiquitination or not is probably signal-, cell-, or even microenvironment-dependent, or depends upon the balanced expression of p21 and one or more of its aforementioned destroyers. Perhaps, this is also modulated by upstream regulators of the above p21 destroyers.
When studying the regulation of MDMX in response to DNA damage, we and others previously found that MDMX is phosphorylated by Chk1 or Chk2 and this phosphorylation enhances the binding of MDMX to 14-3-3γ or other isoforms of 14-3-3, except 14-3-3σ, leading to inactivation of MDMX and subsequent activation of p53 (29,–31). This pathway was later confirmed by an elegant knock-in study in mice (32). Because both p21 and 14-3-3γ bind to the extended central region of MDMX (26, 29, 33), we speculated that 14-3-3γ might suppress MDMX-mediated p21 degradation by influencing the p21 binding to MDMX. Our initial test indeed confirmed this idea. As detailed below, our further analyses demonstrate that 14-3-3γ can inhibit MDMX-mediated p21 degradation and thus induce p21 level by inhibiting their interaction independent of p53 and in response to DNA damage.
Human embryonic kidney (HEK) epithelial 293 cells, human non-small lung carcinoma H1299 cells (p53-deficicent), mouse embryonic fibroblast (MEFp53−/−) single knock-out (SK)2 and (MEFp53−/−/MDMX−/−) double knock-out (DX) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 units/ml penicillin and 0.1 mg/ml streptomycin at 37 °C in a 5% CO2 humidified atmosphere. The GST-14-3-3γ (K50E) mutant plasmid was generated by mutating Lys-50 to Glu-50 from wild type 14-3-3γ (wt)-pGEX4T-1 plasmid by site-directed mutagenesis. Myc-14-3-3γ was subcloned from Flag-14-3-3γ plasmid using BamHI and XhoI sites. The other plasmids utilized here were previously described (16, 26, 29, 34). Monoclonal anti-Flag, anti-α-tubulin, and β-actin (Sigma), polyclonal anti-GST (Sigma), polyclonal anti-p21 (M19, Santa Cruz Biotechnology), monoclonal anti-c-Myc (9E10, Santa Cruz Biotechnology), polyclonal anti- MDMX (Bethyl Laboratories and H130, Santa Cruz Biotechnology) or monoclonal anti-MDMX (d-4, Santa Cruz Biotechnology and 8C6), GFP (FL, Santa Cruz Biotechnology), polyclonal anti-14-3-3γ (C16, Santa Cruz Biotechnology) were purchased and utilized.
Cells were transfected with plasmids as indicated in figure legends using TransFectin reagents (Bio-Rad), following the manufacturer's protocol. The cells were harvested at 36–48 h post-transfection and lysed in cell lysis buffer consisting of 50 mm Tris/HCl (pH 8.0), 1 mm EDTA, 150 mm NaCl, 0.5% Nonidet P-40 (Nonidet P-40), 2 mm phenylmethylsulfonyl fluoride (PMSF), 1 mm dithiothreitol (DTT) in addition to protease inhibitor mixture (Sigma). WB and co-IP analyses were performed as previously described (26, 35).
siRNA duplex specific to human 14-3-3γ were synthesized by Dharmacon. The target sequence of 14-3-3γ, which was utilized in the previous study (36) is as follows; AAGAGCTATATCCTTAACCAT. H1299 cells were transfected with 40 nm (final concentration) of 14-3-3γ siRNA or scrambled siRNA duplexes using siLentFect lipid reagent (Bio-Rad) and transfected once again 24 h later on. Eight hours after the second transfection, cells were treated with DMSO or 10 μm etoposide and 24 h post-treatment, cells were harvested for WB analysis.
To measure endogenous p21 half-life, MEFp53−/−/MX−/−cells were transiently transfected with empty vector (pcDNA3.1), Flag-tagged wild type 14-3-3γ or mutant 14-3-3γ K50E plasmid with Myc-MDMX, respectively. Thirty-six hours post-transfection, cells were treated with 100 μg/ml cycloheximide and harvested at 30, 60, 120, and 240 min post-treatment. The endogenous p21 was detected by a WB analysis using 50 μg of cell lysates and the intensities of each band were measured by using Adobe Photoshop program. The level of remaining endogenous p21 was quantified by normalization with corresponding β-actin bands and plotted using Microsoft Excel.
MEFp53−/− cells were transfected with plasmids encoding empty vector (pcDNA3.1), Flag-tagged wild type 14-3-3γ (wt), or mutant 14-3-3γ (K50E). Fifty hours post-transfection, cells were harvested and fixed with 70% EtOH (final concentration) at least for 30 min on ice. After washing with 1× PBS, cells were stained with PI staining buffer in PBS (50 μg/ml PI, 0.1 mg/ml RNase A, 0.05% Triton X-100) and incubated for 40 min at 37 °C followed by fluorescence-activated cell sorter (FACS) analysis.
The GST(0), GST-14-3-3γ (wt), and GST-14-3-3γ (K50E) proteins were expressed in Escherichia coli. with IPTG induction and cell lysates were prepared by sonication followed by incubation with immobilized glutathione beads (Thermo Scientific) for 30 min at 4 °C. The beads were washed and eluted with elution buffer (10 mm reduced glutathione, 50 mm Tris-Cl (pH 8.0)) followed by dialysis in BC-100 solution (20 mm Tris (pH8.0), 15% glycerol, 0.1 m KCl, 0.2 mm EDTA, 10 mm β-mercaptoethanol, 0.2 mm PMSF).
293 cells were transiently transfected with Flag-p21 (3 μg) or Myc-MDMX (4 μg) in 60-mm plates and 48 h post-transfection, cell lysates were prepared. After purification of GST proteins as described above, the purified GST proteins with final concentrations of 0, 10 nm, 100 nm, 1 μm, respectively were preincubated with 293 cell lysates overexpressing Myc-MDMX (100 μg/sample) in total 300 μl of cell lysis buffer by rotation for 30 min at room temperature. Then, 293 cell lysates (100 μg/sample) overexpressing Flag-p21 were added for co-incubation by rotation for additional 30 min at RT. The mixtures were co-immunoprecipitated with anti-Myc antibodies (1 μg/sample) for 3 h at 4 °C and subjected to Western blot analysis.
Because 14-3-3γ can disable MDMX from inhibition of p53 (29) and binds to its central region nearby where p21 binds to (26, 33), we thought to test if 14-3-3γ could also inhibit the ability of MDMX to degrade p21. To this end, we transfected p53 and MDMX double null MEF (MEFp53−/−/MX−/−) cells with plasmids encoding Myc-MDMX, Flag-14-3-3γ, or Flag-p21 either together or alone (GFP was used as an internal control for transfection), and then analyzed their levels by WB. As expected, overexpression of MDMX markedly reduced the level of p21 (lane 3 of Fig. 1A). Interestingly, further expression of 14-3-3γ rescued this reduction in a dose-dependent fashion (lanes 4–5 of Fig. 1A). To test if 14-3-3γ could affect the level of endogenous p21, we transfected p53-null MEF (MEFp53−/−) cells with only the Flag-14-3-3γ expression vector and conducted a WB analysis. As shown in Fig. 1B, ectopic expression of 14-3-3γ induced the level of endogenous p21 in a dose-dependent manner. This induction was not at the RNA level (data not shown). Because MDM2 has also been shown to mediate p21 degradation (15, 17, 37), we wanted to determine if the induction of p21 by 14-3-3γ is MDMX-specific or not. To this end, we transfected p53 single (SK) or p53/MDMX (DX) double knock-out MEF cells with the Flag-14-3-3γ expression plasmid for WB. Interestingly, the ectopically expressed 14-3-3γ only induced the level of endogenous p21 in the p53-null, but not p53/MDMX double knock-out MEF cells (compare lane 2 with lane 4 of Fig. 1C), indicating that 14-3-3γ fails to affect the level of p21 in the absence of MDMX. These results indicate that 14-3-3γ is able to overcome the negative effect of MDMX on p21, leading to induction of p21 levels. Because other members of 14-3-3 isoforms are able to bind to MDMX (29, 30), we also tested other 14-3-3 isoforms to see if they affect the level of p21. Different from 14-3-3γ, either 14-3-3ζ or 14-3-3τ showed little effect on the p21 level, indicating the effect of each 14-3-3 isoform on p21 is distinct.
Next, we wanted to determine if 14-3-3γ induction of p21 levels is due to the stabilization of this cell cycle-regulated protein. To do so, we transfected p53/MDMX double knock-out MEF cells with MDMX expression vectors together with an empty vector or a plasmid encoding 14-3-3γ or 14-3-3γ K50E that was previously shown to be defective in MDMX-binding (29, 38). Cells were treated with cycloheximide that inhibits protein biosynthesis and harvested at the indicated different time points immediately post-treatment for WB. As shown in Fig. 2, 14-3-3γ, but not its K50E mutant, extended the half-life of endogenous p21 from ~45 min to almost 2 h. This result, together with results of Fig. 1, firmly demonstrates that 14-3-3γ can protect p21 from proteasomal degradation mediated by MDMX, leading to the augmentation of its level.
The result in Fig. 2 indicates that the K50E mutant of 14-3-3γ, which has been previously shown to be unable to interact with MDMX and thus fail to suppress its function toward inhibition of p53 (29), cannot extend the half-life of p21. To further test if this mutant would affect the level of p21 or not, we performed a transfection-coupled WB assay using the same approach with the same cell lines as those for Fig. 1. As shown in Fig. 3, A and B, this 14-3-3γ mutant was unable to induce the level of both exogenous and endogenous p21, in comparison of wild type 14-3-3γ. These results demonstrate that binding to MDMX is necessary for 14-3-3γ to stabilize the p21 protein and again validate the action of 14-3-3γ on p21 is MDMX-dependent. To determine the functional outcome of 14-3-3γ induction of p21, we transfected p53 null MEF cells with an empty vector or a plasmid encoding 14-3-3γ or K50E for a FACS analysis at ~50 h post-transfection. As shown in Fig. 3C, 14-3-3γ, but not K50E, increased the G1 phase cells by ~25%, which was statistically significant compared with the empty vector or K50E transfected cells. The increase in G1 phase cells was also observed 24 h post-transfection by ~15% (data not shown). Because this experiment was conducted in p53 null cells, this G1 arrest was p53-independent. Together with the above results, this result indicates that ectopically expressed 14-3-3γ can lead to G1 arrest by inhibiting MDMX-mediated degradation of p21 and inducing p21 levels independent of p53.
The results from Figs. 11–3 suggest that binding to MDMX is necessary for 14-3-3γ to inhibit its activity of mediating p21 degradation, but do not offer a mechanistic explanation for this phenomenon. Because both 14-3-3γ and p21 bind to the extended central region of MDMX (26, 33), we suspected that their interactions with MDMX might be mutually exclusive. To test this idea, we first carried out a transfection followed by co-immunoprecipitation (co-IP)-WB assays with different combinations of plasmids that express GFP-MDMX, Flag-p21, Myc-14-3-3γ (wt) or Myc-K50E in 293 cells as described in figure legends. As expected (29, 38), GFP-MDMX was co-immunoprecipated with Myc-14-3-3γ, but not Myc-K50E, when anti-Myc antibody was used for co-IP-WB analyses with cell lysates (middle panels of Fig. 4A). Also as expected (26), GFP-MDMX was co-immunoprecipitated with Flag-p21 when anti-Flag was used for co-IP with cells that expressed these two proteins (first lane of right panels of Fig. 4A). Remarkably, when GFP-MDMX and Flag-p21 were co-expressed with Myc-tagged wild-type 14-3-3γ in cells (middle lane of left panels of Fig. 4A), anti-Myc antibodies were unable to pull-down Flag-p21 with GFP-MDMX (middle lane of central panels of Fig. 4A) and anti-Flag antibodies also failed to co-immunoprecipitate GFP-MDMX or Myc-14-3-3γ (middle lane of right panels of Fig. 4A), indicating that ectopic p21 was completely excluded from the exogenous MDMX-14-3-3γ complex. However, this exclusion effect on p21 was not observed when mutant 14-3-3γ K50E (Myc-K50E) was co-expressed with GFP-MDMX and Flag-p21 (right lane of left panels of Fig. 4A), as GFP-MDMX was pulled down together with Flag-p21 by anti-Flag in the presence of Myc-K50E (right lane of right panels of Fig. 4A), indicating that binding of 14-3-3γ to MDMX is critical for its inhibition of the MDMX-p21 interaction.
To determine if overexpression of 14-3-3γ could affect the formation of endogenous MDMX-p21 complexes, we only transfected 293 cells with Flag-14-3-3γ or Flag-K50E for co-IP-WB assays. Consistent with the result in Fig. 4A, 14-3-3γ, but not its K50E mutant, bound to endogenous MDMX and simultaneously reduced the level of the MDMX-p21 complex as pulled down with anti-MDMX antibodies (Fig. 4B). This inhibitory effect of 14-3-3γ on the MDMX-p21 complex was specific to 14-3-3γ, as overexpression of ectopic ribosomal protein L11, which was previously shown to bind to MDM2 (34, 39), but not MDMX (40, 41), did not apparently affect the formation of the exogenous MDMX-p21 complex as pulled down by anti-Myc antibody (Fig. 4C). Again as shown before (40, 41), L11 was not co-immunoprecipitated with MDMX by the anti-Myc antibody (Fig. 4C). Taken together, the results clearly indicate that 14-3-3γ, but not its mutant K50E defective of MDMX-binding (29, 38), excludes the binding of p21 to MDMX perhaps through a competitive mechanism.
To determine if 14-3-3γ can indeed compete with p21 for binding to MDMX, we performed a protein-protein binding competition assay. To do so, we first purified GST, GST-14-3-3γ and GST-14-3-3γ-K50E fusion proteins from E. coli as shown in Fig. 5A. Then, we pre-incubated each of these purified proteins in different concentrations with cell lysates that contained Myc-MDMX. An equal amount of Flag-p21-containing lysates was added into each of the above mixtures and the cocktails were incubated for 30 min prior to being subjected to co-IP-WB with anti-Myc antibody. As shown in Fig. 5B, the level of the MDMX-p21 complex was inversely proportional to that of the MDMX-14-3-3γ complex (left panels of Fig. 5B), indicating that 14-3-3γ can compete with p21 for interaction with MDMX in vitro. By contrast, neither 14-3-3γ K50E mutant nor GST was able to affect the formation of the MDMX-p21 complex (middle and right panels of Fig. 5B). These results, together with the above results, strongly demonstrate that 14-3-3γ is able to exclude p21 from binding to MDMX in vitro and in cells, inhibiting MDMX-mediated p21 degradation and consequently leading to the induction of p21 level.
It was previously shown that DNA damage signals induce the interaction between 14-3-3 proteins including 14-3-3γ and MDMX (29,–31, 42). We wanted to test whether DNA damage could induce p21 levels independent of p53, and if so, whether DNA damage affects the complex formation between 14-3-3γ and MDMX or p21 and MDMX. For this purpose, two p53-null cell lines, H1299 and MEFp53−/− cells, were treated with 10 μm of etoposide and harvested 24 h post-treatment for co-IP-WB analyses. Remarkably, p21 was induced by etoposide in both of the p53-null cell lines (upper panels of Fig. 6, A and B), indicating this p21 induction is p53-independent. More interestingly, DNA damage caused by etoposide induced the formation of the 14-3-3γ-MDMX, but simultaneously reduced the formation of the p21-MDMX complex as co-immunoprecipitated with anti-MDMX antibodies (lower panels of Fig. 6, A and B) in both of the p53-deficient cells. These results, in line with the results in Figs. 11–5, demonstrate that DNA damage signals can also induce the interaction of 14-3-3γ with MDMX, leading to the inverse reduction of the MDMX-p21 complex by competing with p21 for MDMX binding. As a result, the stress signal increases p21 level and induces G1 arrest. Conversely, when endogenous 14-3-3γ was depleted by siRNA as evident in Fig. 6C, p21 levels decreased in DMSO-treated H1299 cells and more evident in etoposide-treated H1299 cells in comparison with scrambled siRNA, supporting the role for endogenous 14-3-3γ in the protection of p21 from MDMX-mediated proteasomal degradation.
It has been known that p21 proteasomal degradation can be facilitated by multiple cellular proteins including MDM2 and MDMX (15, 17, 22, 26, 37, 43). However, it remains incompletely understood how the p21 destroyers are controlled in order for cells to reactivate p21 when needed. In our current study, we present one example of the regulation. As detailed above, our study reveals that 14-3-3γ, which was previously reported to bind to MDMX and inhibit its function toward p53 in response to DNA damage (29, 30, 44), can stabilize p21 by binding to MDMX, subsequently excluding p21 from MDMX and preventing its degradation mediated by MDMX in response to DNA damage. This effect was completely independent of p53. Although it is not surprising that DNA damage can induce p21 independent of p53 (45, 46), our study as shown here provides a new mechanism that may at least partially account for the p53-independent elevation of p21 protein levels post-DNA damage (Fig. 6D). Thus, together with the previously published studies (30, 31, 42), our data suggest two mechanisms for 14-3-3γ to activate p21 in response to DNA damage, one by p53 activation and consequent induction of p21 transcription (29) and the other by p53-independent stabilization of p21 (Fig. 6D). Both of the mechanisms require the direct interaction of 14-3-3γ with MDMX.
MDM2 has also been shown to mediate p21 degradation independent of p53 (17). Although MDM2 and MDMX work together to inactivate p53 (26), they can facilitate p21 degradation independent of each other. Since 14-3-3γ did not bind to MDM2 (29) and also failed to affect p21 level in the absence of endogenous MDMX, such as in p53/MDMX double knock-out MEF cells (Fig. 1C), it is less likely that 14-3-3γ suppresses p21 degradation by directly affecting MDM2 function. However, it remains possible that 14-3-3γ might alleviate MDM2-mediated p21 degradation by directly negating MDMX since MDMX and MDM2 can cooperate with each other to synergistically facilitate p21 proteasomal turnover (26). In contrast to our finding, one of the 14-3-3 isoforms, 14-3-3τ, was recently shown to facilitate ubiquitination-independent 20 S proteasomal turnover of p21 by forming a complex with S8, p21, and MDM2, and implicated to play a possible oncogenic role in breast cancers (22). Along with our observation in Fig. 1D, which shows that 14-3-3τ fails to induce p21, these results are somehow surprising, as 14-3-3τ has been also shown to directly interact with MDMX in a serine 367 phosphorylation-dependent manner, leading to p53 activation (30), exactly the same way by which 14-3-3γ binds to MDMX (29). How would we reconcile these controversial results regarding the role of the 14-3-3 family members in regulating the proteasomal turnover of p21? Since 14-3-3τ, like 14-3-3γ, can inactivate MDMX and consequently induce p53 (30), it would be predicted that 14-3-3τ should also induce, instead degrades, p21 by directly binding to MDMX. Although this question remains to be further addressed and confirmed, it is clear that 14-3-3γ, unlike 14-3-3τ, would not directly facilitate p21 degradation that is mediated by MDM2, as overexpression of 14-3-3γ induced p21 level in cells that contained the wild type of MDM2 (Fig. 1, A and B) and this induction was cancelled in the absence of MDMX (Fig. 1C). One possible mechanism for 14-3-3τ-mediated degradation of p21 would counteract the function of 14-3-3γ, as 14-3-3τ could form a heterodimer with 14-3-3γ (data not shown). If so, this might also explain why 14-3-3τ degrades p21. Another possibility would be that the inactivation of p21 by 14-3-3τ might be cell-specific. 14-3-3ζ has been shown to play an oncogenic role in breast cancer development and progression (47). Consistent with this role, this 14-3-3 isoform does not appear to induce the level of endogenous p21 significantly (Fig. 1D). In any case, these studies suggest that each individual member of the 14-3-3 family may possess its own independent role in regulating p21 turnover and the cell cycle, and their regulation or interplay is more complicated than what any simple model would predict.
How would 14-3-3γ suppress MDMX-mediated p21 degradation? Although detailed biochemical mechanisms still remain to be further investigated, it is clear that 14-3-3γ is able to bind to the central region encompassing amino acids 340–380 while p21 binds to three domains, including the N-, central, and C-terminal Ring finger domains of MDMX (26, 33). Remarkably, binding of 14-3-3γ to MDMX eliminated the binding of p21 to MDMX (Figs. 4–5), suggesting that 14-3-3γ and p21 cannot simultaneously associate with MDMX. Hence, a simplest mechanism for 14-3-3γ action on p21 would be that 14-3-3γ can remove p21 from MDMX by binding to MDMX, thus slashing the chance for MDMX to degrade p21. Another yet untested mechanism could be that 14-3-3γ may prevent MDMX from the association with the 26 S or 20 S proteasome complex. It is also puzzling that p21 was completely excluded from the MDMX by 14-3-3γ (Figs. 4–5) even though p21 was previously shown to bind to the N- and C-terminal Ring finger domain of MDMX as well (26, 33). It is possible that once 14-3-3γ binds to the domain (amino acids 340–380) of MDMX, this binding might convey a conformational change of this protein, which would alter the structures of the p21-binding domains within MDMX so that p21 is no longer able to recognize its binding sites. Further investigation of this enticing possibility would also offer some clues for our better understanding of the biochemical mechanism(s) underlying the inactivation of MDMX function by 14-3-3γ toward not only p53, but also p21.
We thank Shelya Zeng, Guifen He, and Qi Zhang for technical support and helps, and the members of the Lu laboratory for active discussion.
*This work was supported, in whole or in part, by National Institutes of Health NCI Grants CA095441, CA 079721, and CA129828 (to H. L.).
2The abbreviations used are: