|Home | About | Journals | Submit | Contact Us | Français|
PHLPP1 belongs to a novel family of Ser/Thr protein phosphatases that serve as tumor suppressors by negatively regulating Akt signaling. Our recent studies have demonstrated that loss of PHLPP expression occurs at high frequency in colorectal cancer. In this study, we identified PHLPP1 as a proteolytic target of a β-TrCP-containing Skp-Cullin 1-F-box protein (SCF) complex (SCFβ-TrCP) E3 ubiquitin ligase in a phosphorylation-dependent manner. Overexpression of wild-type but not ΔF-box mutant β-TrCP leads to decreased expression and increased ubiquitination of PHLPP1, whereas knockdown of endogenous β-TrCP has the opposite effect. In addition, we show that the β-TrCP-mediated degradation requires phosphorylation of PHLPP1 by casein kinase I and glycogen synthase kinase 3β (GSK-3β), and activation of the phosphatidylinositol 3-kinase/Akt pathway suppresses the degradation of PHLPP1 by inhibiting the GSK-3β activity. Furthermore, expression of a degradation-deficient PHLPP1 mutant in colon cancer cells results in a more effective dephosphorylation of Akt and inhibition of cell growth. Taken together, our findings demonstrate a key role for β-TrCP in controlling the level of PHLPP1, and activation of Akt negatively regulates this degradation process.
Hyperactivation of phosphatidylinositol 3-kinase/Akt signaling is commonly associated with human cancers (1, 5, 27). Inability to terminate the growth and survival signals mediated by Akt is one of the major mechanisms contributing to the development of cancer (1, 22, 32). The activation of Akt involves two phosphorylation steps: it is first phosphorylated at the activation loop (Thr308) within the kinase core by PDK-1 and subsequently at the hydrophobic motif (Ser473) in the C terminus by the TORC2 complex (22). Since the activity of Akt is tightly controlled by phosphorylation, dephosphorylation of Akt leads to effective signaling termination by inactivating the kinase. Recently, a novel family of Ser/Thr protein phosphatases, PHLPP, has been identified to fulfill the role of a negative regulator for Akt via direct dephosphorylation (3, 14). Two isoforms of PHLPP, namely PHLPP1 and PHLPP2, are found in this phosphatase family. Although the two isoforms of PHLPP share their ability to dephosphorylate Akt, each PHLPP preferentially regulates a subset of Akt isoforms in human lung cancer cells (3). Several lines of evidence suggest that PHLPP functions as a tumor suppressor. For example, overexpression of PHLPP in glioblastoma and colon cancer cells inhibits tumorigenesis in xenografted nude mice (14, 20), while decreased PHLPP expression correlates with increased metastastic potential in breast cancer cells (26). Furthermore, our recent studies have shown that downregulation of both PHLPP isoforms occurs at high frequency in colorectal cancer clinical samples (20). Loss of tumor suppressor expression can be caused by alterations at the gene level such as loss of heterozygosity or gene methylation. However, dysregulation of protein degradation pathways has also been implicated as a reason for downregulation of tumor suppressors (2, 6, 16).
The ubiquitin (Ub) proteasome pathway controls degradation of the majority of eukaryotic proteins (12). β-TrCP belongs to a large family of F-box-containing proteins, and it serves as the substrate recognition subunit in the SCF (Skp1-Cullin 1-F-box protein) Ub-E3 ligase protein complex (4). By regulating the proteolytic process of its substrates, β-TrCP plays an important role in controlling cell cycle and cancer biogenesis (10). It is believed that β-TrCP-mediated ubiquitination requires phosphorylation of its substrates (35). A consensus binding motif with the sequence of DSG(X)2-nS (so-called “phospho-degron”) has been proposed, in which the two serine residues are phosphorylated prior to binding to β-TrCP (4). However, variations of this motif, including replacement of the serine residues with phosphomimetic residues (e.g., Glu or Asp) in the substrate sequence, have been shown to be equally effective in mediating association with β-TrCP (31, 34).
In this study, we report the identification of PHLPP1 as a proteolytic target of β-TrCP. We show that the degradation process of PHLPP1 depends on casein kinase I (CK1)- and glycogen synthase kinase 3 (GSK-3)-mediated phosphorylation, and activation of Akt negatively regulates PHLPP1 turnover. In addition, a PHLPP1 phosphorylation/degradation mutant antagonizes Akt more effectively in colon cancer cells.
The expression plasmids for Myc-tagged E3-ligases including β-TrCP1, ΔF-box/β-TrCP1, FBW2, FBL5, Itch, and Myc-tagged GSK-3β were gifts from Binhua Zhou (University of Kentucky). The Myc-CK1α plasmid was generously provided by Wade Harper (Harvard Medical School). The mammalian expression constructs for glutathione S-transferase (GST)-tagged domains of PHLPP1 and the plasmid for expressing GST-PP2C of PHLPP1 in bacteria were generated as previously described (13, 14). All PHLPP1 mutant plasmids were generated by site-directed mutagenesis (Stratagene QuikChange kit) using pcDNA3-HA-PHLPP1 or GST-PP2C (14) as the templates. The following antibodies were purchased from commercial sources: polyclonal antibodies for PHLPP1 and Myc from Bethyl Laboratory; polyclonal antibodies against Akt, polyclonal phospho-Akt (p473) and phospho-GSK-3α/β(Ser21/9) from Cell Signaling; a monoclonal antibody (MAb) for GSK-3β from Biosource; an anti-hemagglutinin (HA) high-affinity rat MAb from Roche; an anti-myc MAb from Santa Cruz Biotechnology; and anti-γ-tubulin MAb from Sigma. β-TrCP1/2-specific small interfering RNA (siRNA) was synthesized by Invitrogen, and the targeting sequence was reported previously (7). Inhibitors including LY290042, CK1 inhibitor-D4476, GSK-3 inhibitor IX, and MG-132 were purchased from EMD/CalBiochem, and the concentrations of the inhibitors used are as follows: MG-132, 10 μM; CK1-I, 10 μM; GSK3-I, 30 nM; and LY294002, 20 μM. [γ-32P]ATP (3,000 Ci mmol−1) was obtained from PerkinElmer Life Sciences. The recombinant active forms of CK1, GSK-3β, and lambda protein phosphatase (λ-PPase) were obtained from New England Biolabs.
Human colon cancer cells HCT116 and HCT-P1 (HCT116 cells overexpressing wild-type PHLPP1 ) were cultured in McCoy's 5A medium, and Caco2 and 293T cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen). All media were supplemented with 10% fetal bovine serum (HyClone) and 1% penicillin/streptomycin. Transient transfection of all cell types was carried out using Lipofectamine 2000 transfection reagents (Invitrogen). To knockdown β-TrCP1 and β-TrCP2 expression in cells, On-TARGETplus siRNAs specific for each gene were obtained from Dharmacon. A combination of siRNAs for both genes was used at total concentration of 50 nM to transfect cells. The short hairpin RNAs (shRNAs) for human Akt1, GSK-3α, and GSK-3β genes were constructed in pLKO.1-puro vector and purchased from Sigma-Aldrich (30), and the targeting sequences are as follows: Akt1, CGCGTGACCATGAACGAGTTT; GSK-3α, CCAGGACAAGAGGTTCAAGAA; and GSK-3β, CCGATTGCGTTATTTCTTCTA. The lentivirus-mediated delivery of shRNA and selection for stable knockdown cells were carried out as previously described (20). To generate stable cells overexpressing wild-type and mutant PHLPP1, the coding sequences of HA-PHLPP1 and HA-P1/4A were subcloned into pBabe-puro vector (Addgene), and the corresponding retroviruses were subsequently produced as described previously (23). HCT116 cells infected with the PHLPP1 retroviruses were subjected to selection with puromycin (2 μg/ml).
Equal numbers of Caco2 or HCT116 stable cells were seeded onto 12-well plates and allowed to attached for ~18 h. The cells were treated with cycloheximide (CHX; 20 μg/ml) and harvested at indicated time points. The cell lysates were prepared in sodium dodecyl sulfate (SDS) sample buffer directly, and an equal amount of lysates was analyzed by immunoblotting. To examine the effects of different inhibitors, MG-132, CK1-I, or GSK3-I were added 30 min prior to the addition of CHX.
Immunoprecipitation (IP) experiments were performed following procedures described previously (14). Briefly, the cells were lysed in buffer A (50 mM Na2HPO4, 1 mM sodium pyrophosphate, 20 mM NaF, 2 mM EDTA, 2 mM EGTA, 1% Triton X-100, 1 mM dithiothreitol [DTT], 200 μM benzamidine, 40 μg ml−1 leupeptin, 200 μM phenylmethylsulfonyl fluoride), and the detergent-solublized cell lysates were incubated with the indicated antibodies and protein A/G agarose at 4°C for 3 h. The IP beads were washed twice in buffer A and twice in buffer B (buffer A plus 200 mM NaCl). Bound proteins were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotting. The density of Western blot signals was obtained and quantified using a FluoChem digital imaging system (Alpha Innotech).
A mammalian expression constract was generated to express PHLPP1 without any epitope tag. Briefly, the coding sequence of human PHLPP1 was amplified using PCR and subcloned into the EcoRI/XhoI sites on the pcDNA3 vector directly. To examine the ubiquitination of PHLPP1 under different treatments, 293T cells were cotransfected with expression constructs of HA-Ub and untagged PHLPP1. Approximately 40 h posttransfection, the cells were treated with the indicated inhibitors. To compare the ubiquitination of wild-type PHLPP1 with the mutants, HCT116 stable cells overexpressing wild-type PHLPP1 or the mutants were transfected with Myc-Ub expression plasmid. The transfected cells were lysed in buffer C (50 mM Tris, pH 7.4, 2 mM EDTA, 2 mM EGTA, 10% Triton X-100, 1 mM DTT, 200 μM benzamidine, 40 μg ml−1 leupeptin, 200 μM phenylmethylsulfonyl fluoride, and 10 mM N-ethylmaleimide). The detergent-solubilized cell lysate was incubated with the PHLPP1 antibody and protein A/G agarose at 4°C for 3 h. The IP beads were washed three times in buffer D (buffer C plus 250 mM NaCl) and once in buffer C. Bound proteins were analyzed by SDS-PAGE and immunoblotting.
Coomassie-stained SDS-PAGE gel bands containing either HA-PHLPP1 isolated from HCT-P1 cells (pretreated with a phosphatase inhibitor Calyculin A) or GST-PP2C fusion proteins phosphorylated with CK1 and GSK-3β were excised and subjected to in-gel digestion and reversed-phase microcapillary liquid chromatography tandem mass spectrometry (MS/MS) using an LTQ 2D linear-ion-trap mass spectrometer (ThermoScientific). The data were obtained and analyzed at the Beth Israel Deaconess Medical Center Mass Spectrometry Core Facility. The following phospho-peptides were identified (the phosphorylated Ser and Thr residues are indicated by asterisks): PHVQ*SVLL*TPQDEFFILGSK and GLWDSL*SVEEAVEAVR. Mutations in PHLPP1 were created with the Quickchange site-directed mutagenesis kit (Stratagene).
Recombinant GST-PP2C fusion proteins of wild-type and mutant PHLPP1 were expressed in bacteria and purified using glutathione-Sepharose as described previously (14). Purified GST-PP2C proteins were left bound on beads. The beads were washed three times with kinase reaction buffer (20 mM Tris-HCl, pH 7.5, 10 mM MgCl2, and 5 mM DTT). In vitro phosphorylation reactions were carried out by combining 40 μl of glutathione-Sepharose-bound GST-PP2C in 10 μl of kinase reaction buffer containing 10 μCi of [γ-32P]ATP and active CK1 and/or GSK-3β (1 U per reaction). The samples were incubated at room temperature for 30 min, and the reactions were terminated by addition of SDS sample buffer. The phosphorylated proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes for visualization by autoradiography.
To examine the direct binding between β-TrCP1 and GST-PP2C of PHLPP1, the TNT T7 coupled reticulocyte lysate system (Promega) was used to produce 35S-labeled Myc-tagged β-TrCP1 using pcDNA3-Myc-β-TrCP1 as the template. The recombinant GST-PP2C proteins were purified and phosphorylated as described above, and phosphorylation reactions were terminated by adding buffer A. Subsequently, the GST-pull-down assays were carried out by incubating GST-PP2C-bound beads with 35S-labeled Myc-β-TrCP1 at 4°C for 3 h. The beads were washed twice in buffer A and twice in buffer B, and the bound proteins were analyzed by SDS-PAGE and transferred to polyvinylidene difluoride membranes for visualization by autoradiography. The membranes were subsequently blotted with the anti-GST antibody to reveal the amount of GST-PP2C protein.
A phospho-specific antibody for the S847 residue in PHLPP1 was generated against a phosphorylated peptide with the sequence FLHPSVVPRPHVQSVLL (in which the phospho-Ser847 residue is underlined). The phospho-peptide was conjugated to keyhole limpet hemocyanin and injected into rabbits at Bethyl Laboratories. The antibodies were affinity purified against the phospho-peptide: the resulting antibody is termed “P847.” To detect the phosphorylation of PHLPP1, the cells were treated with dimethyl sulfoxide or MG-132 for 4 h followed by calyculin A (100 nM) for 20 min to trap the phosphorylated species. The cells were analyzed as described in the “IP” section above. The immunoprecipitated PHLPP1 proteins were separated by SDS-PAGE and subjected to immunoblotting analysis with P847.
293T cells transfected with HA-PHLPP1 and HA-P1/4A were lysed in buffer A. Two identical IPs were set up for both HA-PHLPP1 and HA-P1/4A, and each IP was carried out by incubating 1 mg of cell lysates with the PHLPP1 antibody and protein A/G-agarose at 4°C for 3 h. The IP beads were washed three times with buffer B and twice with λ-PPase reaction buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM DTT, 0.1 mM EGTA, and 0.01% Brij 35). One set of the IP beads was then incubated with λ-PPase at room temperature for 30 min, while the other set was left untreated. Following the dephosphorylation reaction, the beads were washed three times in buffer A. 293T cells separately transfected with Myc-β-TrCP1 were lysed in buffer A, and equal amounts of the lysates were incubated with the washed IP beads from above at 4°C for 3 h. The IP beads were then washed twice in buffer A and twice in buffer B, and associated proteins were analyzed by SDS-PAGE and immunoblotting.
Equal numbers of HCT116 stable cells were seeded onto 12-well plates (2 × 104 cells/well) and allowed to grow for 4 days. For some cells, LY294002 was added to the media on the second day of the experiments. At the end of the experiments, the cells were fixed and stained with 0.5% crystal violet in 20% methanol for 30 min. After washing with water, the stained cells were dissolved in 1% SDS and A570 was determined.
To elucidate the potential mechanism of PHLPP degradation, we first examined whether PHLPP1 is downregulated via the proteasome pathway. Colon cancer Caco2 cells were treated with a protein synthesis inhibitor, CHX, alone or in combination with the proteasome inhibitor MG-132, and the level of endogenous PHLPP1 was monitored for 8 h. PHLPP1 was effectively degraded in cells, and the degradation of PHLPP1 was blocked by MG-132 (Fig. (Fig.1A).1A). Note that there are two splicing variants of PHLPP1, PHLPP1α (the short form) and PHLPP1β (the long form) (3), expressed in Caco2 cells. PHLPP1β has a slightly faster degradation rate. Our present study focuses on PHLPP1α as the cloning of PHLPP1β cDNA has not been reported to date. However, the two splicing forms of PHLPP1 are likely to share a common degradation route as the majority of the protein sequences overlap between PHLPP1α and PHLPP1β. To keep the nomenclature consistent with previous reports, PHLPP1α is simply referred as PHLPP1 (3, 14). A similar degradation rate was observed for overexpressed PHLPP1 in stable HCT-P1 cells (stable colon cancer HCT116 cells overexpressing PHLPP1 (20)) (Fig. (Fig.1B).1B). The half-life of PHLPP1 is ~4 h in both cell lines. These results provide initial evidence that PHLPP1 is degraded by the proteasome pathway in cells.
Since E3 ligases are known to bind their substrates with relatively high affinity, we screened a set of E3 ligases for their ability to interact with PHLPP1. Results from a representative co-IP experiment showed that PHLPP1 specifically associated with wild-type and ΔF-box mutant β-TrCP1 (a ubiquitination-deficient mutant in which the substrate binding moiety is intact) (Fig. (Fig.1C).1C). As controls for specificity, little interaction was detected between PHLPP1 and other E3 ligases, including FBW2 (another F-box/WD40 repeat containing E3 ligase), FBL5 (an F-box/leucine-rich repeat containing E3 ligase), and Itch (a HECT domain E3-ligase). In addition, we found that β-TrCP2, a protein closely related to β-TrCP1, interacted with PHLPP1 as well (data not shown). Furthermore, the interaction between endogenous PHLPP1 and β-TrCP was detected in Caco2 cells (Fig. (Fig.1D1D).
To examine whether β-TrCP regulates the level of PHLPP1 in cells, we transfected HCT-P1 cells with wild-type or ΔF-box mutant β-TrCP1 and analyzed PHLPP1 expression. Overexpression of wild-type β-TrCP decreased PHLPP1 expression in HCT-P1 cells, while the ΔF-box mutant had the opposite effect (Fig. (Fig.1E).1E). This is consistent with the notion that the ΔF-box mutant may serve as a dominant-negative construct by competing for substrate binding with endogenous β-TrCP (11, 18).
We next addressed whether endogenous β-TrCP plays a role in regulating PHLPP1 expression. Because of the functional redundancy between β-TrCP1 and β-TrCP2 (7, 24), we used a pool of siRNAs targeting both genes. Knockdown of endogenous β-TrCP resulted in a twofold increase in PHLPP1 expression in HCT-P1 cells (Fig. (Fig.2A).2A). A similar increase in endogenous PHLPP1 expression was observed in Caco2 cells transfected with β-TrCP-specific siRNA (Fig. (Fig.2B).2B). As a positive control, the expression of β-catenin, a known substrate of β-TrCP, was increased in the β-TrCP knockdown cells as well (Fig. (Fig.2B).2B). Note that knockdown of either β-TrCP1 or β-TrCP2 by itself had very little effect on PHLPP1 expression (data not shown), confirming the functional redundancy of these two isoforms. Furthermore, we compared the rate of PHLPP1 degradation in cells transfected with β-TrCP-specific siRNA or the ΔF-box mutant to that in the control cells. As shown in Fig. Fig.2C,2C, the half-life of PHLPP1 was increased from ~4 h in the control cells to over 12 h in β-TrCP knockdown cells. Similarly, the rate of PHLPP1 degradation is decreased in cells overexpressing the ΔF-box mutant (Fig. (Fig.2D;2D; the half-lives of PHLPP1 in the control and ΔF-box transfected cells are 4.2 and 8.2 h, respectively). Taken together, our findings demonstrate that the endogenous β-TrCP is involved in promoting the degradation of PHLPP1, and intact E3-ligase function is required for β-TrCP-mediated degradation.
To further assess whether β-TrCP is involved in the ubiquitination of PHLPP1, we performed in vivo ubiquitination experiments in transfected 293T cells. MG-132 was used to enrich the ubiquitinated species in cells. PHLPP1 was basally ubiquitinated, and treatment of cells with MG-132 increased the level of ubiquitination (Fig. (Fig.3A,3A, lanes 1 and 2). Overexpression of wild-type β-TrCP significantly increased the amount of ubiquitination of PHLPP1 both basally and after MG-132 treatment, while the ΔF-box mutant had no effect (Fig. (Fig.3A,3A, lanes 3 to 6). Furthermore, we analyzed the ubiquitination of PHLPP1 in β-TrCP knockdown cells. Since the ubiquitination of endogenous PHLPP1 was below the detection limit in these experiments, the expression constructs for PHLPP1 and HA-ubiquitin were cotransfected with the control or β-TrCP specific siRNA into 293T cells. As shown in Fig. Fig.3B,3B, a slight decrease of basal ubiquitination was observed in β-TrCP knockdown cells (compare lanes 1 and 3), and MG-132-induced increase of ubiquitination was also significantly blocked by depleting β-TrCP expression using RNA interference (compare lanes 2 and 4). Collectively, these results suggest that β-TrCP negatively regulates the stability of PHLPP1 by promoting its ubiquitination.
It has been shown previously that β-TrCP-mediated degradation is triggered by phosphorylation of its substrates (4, 35). Since β-TrCP targets a small phosphorylated motif in its substrates for binding, we first determined which domain of PHLPP1 interacts with β-TrCP in cells. The domain composition of PHLPP1 is shown in Fig. Fig.4A.4A. Different domains of PHLPP1 were constructed as GST-tagged fusion proteins, and GST-pull-down experiments showed that only the PP2C domain of PHLPP1 interacted with β-TrCP (Fig. (Fig.4B,4B, lane 4). To search for potential kinases that may play a role in downregulating PHLPP1, we treated HCT-P1 cells with a panel of protein kinase inhibitors together with CHX for 8 h and monitored the level of PHLPP1 expression. Among all the kinase inhibitors examined, a CK1 inhibitor (CK1-I) and a GSK-3 inhibitor (GSK-3-I) were able to block the degradation of PHLPP1 (data not shown), suggesting that phosphorylation of PHLPP1 by CK1 and GSK-3 is likely required for degradation. Interestingly, both CK1 and GSK-3 are known to phosphorylate other β-TrCP substrates (17, 19, 36).
To identify the phospho-degrons for β-TrCP within the PP2C domain, PHLPP1 proteins were isolated from HCT-P1 cells by IP and subjected to phospho-peptide analysis by MS. In addition, the purified GST-PP2C fusion proteins were phosphorylated by CK1 and GSK-3β in vitro and analyzed by MS as well. Three sites within the phosphatase domain, Ser847, Thr851, and Ser869, were found to be phosphorylated in cells and in vitro (see Materials and Methods). Moreover, we found that Ser867 and Ser869 are redundant in regarding to CK1-mediated phosphorylation. Thus, the Ser867 and Ser869 sites are mutated together in all subsequent experiments. These phosphorylation sites are termed the “N-clusters” and “C-clusters” (containing the Ser847/Thr851 and Ser867/Ser869 sites, respectively). The protein sequences surrounding these phosphorylation sites and the mutants created are shown in Fig. Fig.4A4A.
To dissect the contribution of each phosphorylation site, the purified wild-type and mutant GST-PP2C fusion proteins were subjected to in vitro phosphorylation. The wild-type GST-PP2C was readily phosphorylated by CK1, while GSK-3β was unable to induce any phosphorylation by itself. However, addition of CK1 enabled GSK-3β to increase the total phosphorylation by 30% (Fig. 4C and D). This is consistent with the notion that GSK-3-mediated phosphorylation requires a priming phosphorylation of the substrate by other kinases (9). Interestingly, the amount of CK1-mediated phosphorylation was reduced by 50% in the C2A mutant, whereas addition of GSK-3β together with CK1 resulted in a more pronounced increase over CK1 alone (Fig. (Fig.4D,4D, lanes 4 to 6). Furthermore, mutation of all four phosphorylation sites led to a total loss of phosphorylation for both kinases (Fig. (Fig.4D,4D, lanes 7 to 9). Based on the phosphorylation changes observed, we postulate that the PP2C domain of PHLPP1 is phosphorylated at two sites by CK1 and one additional site by GSK-3β. Because CK1-mediated phosphorylation was reduced by half in the C2A mutant, either Ser867 or Ser869 likely constitutes one of the two CK1 sites. Since GSK-3β requires a priming phosphate at the n + 4 position, Ser847 is most likely to be the GSK-3β site, while CK1-mediated phosphorylation of Thr851 provides the priming phosphate.
Between the two kinases regulating PHLPP1, CK1 is known to be a constitutively active kinase (25), while the activity of GSK-3β is readily modulated in cells. To determine whether PHLPP1 is phosphorylated at the Ser847 site by GSK-3β in cells, we generated a phospho-specific antibody against this site. As shown in Fig. Fig.4E,4E, the phosphorylated wild-type PHLPP1 was detected by the P847 antibody, while the antibody did not react with the S847A mutant. Furthermore, overexpression of GSK-3β increased the amount of phospho-PHLPP1 detected in cells, whereas a GSK-3β inhibitor, LiCl, eliminated phosphorylation of PHLPP1 at Ser847 (Fig. (Fig.4F).4F). Interestingly, treatment of cells with MG-132 enriched the phosphorylated species of PHLPP1. Taken together, these results show that the Ser847 site in PHLPP1 is phosphorylated by GSK-3β in cells, and PHLPP1 is rapidly targeted for proteasome-mediated degradation once phosphorylated.
To further assess whether CK1 and GSK-3 activity is required for PHLPP1 degradation, we examined whether overexpression of CK1α and GSK-3β promotes PHLPP1 degradation in cells. Overexpression of CK1α resulted in a marked decrease in wild-type PHLPP1 expression (Fig. (Fig.5A,5A, lanes 1 and 2). However, this decrease was largely attenuated in both N2A and C2A mutants (lanes 3 to 6) and totally blocked in the 4A mutant (lanes 7 and 8). This is consistent with the in vitro phosphorylation results showing that each of the two clusters contains a CK1 targeting site, and two sites contribute together to increase PHLPP1 degradation. Similarly, overexpression of GSK-3β reduced the expression of wild-type PHLPP1 and the C2A mutant (Fig. (Fig.5B,5B, lanes 1 to 4). However, the addition of another mutation at either the GSK-3β site (Ser847) or the priming site (Thr851) to C2A rendered it resistant to GSK-3β (lanes 5 to 8). In addition, the 4A mutant was entirely insensitive to GSK-3β overexpression (Fig. (Fig.5B,5B, lanes 9 and 10).
To determine whether phosphorylation of Ser847 requires the priming phosphorylation at Thr851, we analyzed the phosphorylation status of 3A-1 and 3A-2 mutants using the P847 antibody. While the wild-type PHLPP1 was readily detected by the P847 antibody, a mutation at either the Ser847 or Thr851 site resulted in a complete loss of phosphorylation (Fig. (Fig.5C).5C). This is consistent with the results shown in Fig. Fig.5B5B that both mutants are resistant to GSK-3β overexpression. Moreover, the N2A mutant is resistant to GSK-3β overexpression as well (data not shown), confirming the requirement of phosphorylation at both Ser847 and Thr851 sits for GSK-3β-mediated degradation of PHLPP1.
To examine whether β-TrCP binds PHLPP1 in a phosphorylation-dependent manner, we performed in vitro GST-pull-down experiments. Purified GST-PP2C was first phosphorylated in vitro by CK1 and GSK-3β, as described in Fig. Fig.4C,4C, and subsequently incubated with 35S-labeled Myc-β-TrCP1 produced by in vitro transcription and translation. No specific interaction was detected between the unphosphorylated GST-PP2C and Myc-β-TrCP1 (Fig. (Fig.6A,6A, lane 2). The amount of β-TrCP bound to GST-PP2C was increased upon phosphorylation by CK1, while addition of GSK-3β alone had no effect (lanes 3 and 4), and a combination of both kinases further enhanced the interaction (lane 5). In contrast, the 4A mutant was unable to bind β-TrCP (lane 6). To further assess the phosphorylation status of PHLPP1 in cells and the effect of phosphorylation on β-TrCP binding, wild-type and mutant PHLPP1 proteins immunoprecipitated from transfected 293T cells were subjected to treatment with a nonselective phosphatase, λ-PPase. The control and phosphatase-treated PHLPP1 proteins were then incubated with cell lysates overexpressing Myc-β-TrCP1 and analyzed for the amount of β-TrCP bound. Treatment of wild-type PHLPP1 with λ-PPase resulted in a marked decrease in β-TrCP binding (Fig. (Fig.6B,6B, lanes 1 and 2). Less β-TrCP coimmunoprecipitated with the 4A mutant in the absence of λ-PPase, and the amount of β-TrCP bound to the 4A mutant was insensitive to the phosphatase treatment (lanes 3 and 4). Collectively, these results suggest that the binding between PHLPP1 and β-TrCP is controlled by the phosphorylation of the PP2C domain.
We next analyzed the expression of wild-type and mutant PHLPP1 in the presence of β-TrCP to determine whether β-TrCP-mediated degradation requires phosphorylation of PHLPP1 by CK1 and GSK-3β. Coexpression of β-TrCP significantly decreased the expression of wild-type PHLPP1 (Fig. (Fig.6C,6C, lanes 1 and 2). In contrast, the N2A and C2A mutants became partially resistant (lanes 3 to 6), while the 4A mutant was entirely insensitive to β-TrCP overexpression (lanes 7 and 8). In addition, we examined the effect of β-TrCP on the expression of phosphomimetic mutations of PHLPP1. Overexpression of β-TrCP resulted in decreased expression of all phosphomimetic mutants (including N2D, C2D, and 4D), and the level of reduction was similar to that observed in wild-type PHLPP1 (data not shown). These results are consistent with the notion that each of the phosphorylation clusters provides a binding site for β-TrCP and elimination of both sites is necessary to fully protect PHLPP1.
To better compare the differences between wild-type PHLPP1 and the mutants, stable cell lines overexpressing wild-type PHLPP1 and the 4A mutant were generated in HCT116 cells. CHX chase experiments showed that the rate of degradation was markedly reduced in the 4A mutant, and no detectable decay of the 4A mutant was observed during the course of the experiment (Fig. (Fig.7A).7A). Thus, the 4A mutant of PHLPP1 is considered degradation resistant. Moreover, the level of ubiquitination in the 4A mutant was reduced basally, and overexpression of β-TrCP was unable to rescue the ubiquitination (Fig. (Fig.7B).7B). These results confirm that the β-TrCP-mediated ubiquitination depends on phosphorylation of PHLPP1.
Since PHLPP1 serves as a tumor suppressor by dephosphorylating Akt (14, 20), we examined whether the 4A mutant functions more effectively at negatively regulating Akt. In the stable HCT116 cells, the basal expression of the 4A mutant was twofold higher than that of wild-type PHLPP1 (Fig. (Fig.7C).7C). Overexpression of wild-type PHLPP1 resulted in a 60% decrease in Akt phosphorylation at the Ser473 residue, while Akt was dephosphorylated to 25% of the control level in cells expressing the 4A mutant (Fig. (Fig.7C).7C). To establish a functional link between PHLPP1 degradation and its ability to inhibit tumor cell growth, the rate of cell proliferation was determined in the stable cells. Our previous studies have shown that overexpression of wild-type PHLPP1 inhibits cell proliferation both basally and synergistically with LY294002 (20). While overexpression of wild-type PHLPP1 resulted in a 20% inhibition of cell growth, the 4A mutant reduced cell growth by 40% under basal conditions (Fig. (Fig.7D).7D). Furthermore, LY294002-induced growth inhibition was further potentiated in cells expressing wild-type PHLPP1 and the 4A mutant. However, the 4A mutant exhibited a twofold-higher efficacy than the wild-type PHLPP1 (Fig. (Fig.7D).7D). Thus, the 4A mutant of PHLPP1 antagonizes Akt more effectively due to higher expression in cells.
We have previously shown that loss of PHLPP1 expression occurs at high frequency in colon cancer clinical specimens (20). To determine whether increased the expression of β-Trcp is related to PHLPP1 downregulation, we analyzed PHLPP1 expression in colon cancer cell lines. Among the five cell lines tested, an inverse correlation between PHLPP1 expression and the level of β-TrCP protein was observed (Fig. (Fig.7E).7E). Taken together, our results suggest that alteration of PHLPP1 expression is tightly associated with β-TrCP levels in cancer cells.
Since the GSK-3β activity is negatively regulated by its upstream kinase Akt, we next examined whether Akt is involved in controlling PHLPP1 degradation. Stable knockdown cells were generated using lentivirus-based shRNA targeting Akt1. To exclude the potential involvement of Akt in modulating the mRNA level of PHLPP1, HCT-P1 cells were used as the parental cells in which the transcription of PHLPP1 is initiated from an exogenous cytomegalovirus promoter. CHX-chase experiments showed that the rate of PHLPP1 degradation was largely accelerated in Akt1 knockdown cells. Specifically, the half-life of PHLPP1 was reduced from ~4 h in the control cells to ~2 h in Akt1 knockdown cells (Fig. (Fig.8A).8A). In addition, while the Akt-mediated inhibitory phosphorylation of GSK-3 was reduced in Akt1 knockdown cells, the level of PHLPP1 expression was also decreased to 25% of the control level (Fig. (Fig.8B).8B). Conversely, overexpression of a constitutively active Akt (myristoylated-Akt1) markedly increased expression of PHLPP1 by promoting the inhibitory phosphorylation on GSK-3β (Fig. (Fig.8C8C).
To further elucidate which GSK-3 isozyme is important in regulating PHLPP1 expression, we infected HCT-P1 cells with lentivirus-based shRNAs targeting either GSK-3α or GSK-3β. The PHLPP1 expression was markedly increased in the GSK-3β stable knockdown cells, while knocking down GSK-3α had no effect (Fig. (Fig.8D),8D), suggesting that GSK-3β is the major isoform responsible for PHLPP1 degradation. Moreover, the amount of PHLPP1 ubiquitination was elevated in Akt1 knockdown cells (Fig. (Fig.8E).8E). Therefore, by negatively regulating GSK-3β activity in cells, Akt plays a critical role in maintaining the stability of PHLPP1.
In summary, our results demonstrate that the degradation of PHLPP1 is controlled by a dual-phosphorylation mechanism (Fig. (Fig.8F).8F). Two kinases, CK1 and GSK-3β, phosphorylate PHLPP1 and provide two targeting sites for β-TrCP-mediated ubiquitination and degradation. In addition, this downregulation of PHLPP1 is negatively regulated by its substrate Akt.
Recent studies have suggested that enhanced protein degradation may contribute to loss of tumor suppressor proteins in cancer (7, 12, 33). In this study, we investigate the mechanism by which PHLPP1, a novel protein phosphatase and tumor suppressor, is downregulated. Our results show that PHLPP1 specifically interacts with β-TrCP, a substrate binding subunit of SCF E3 ubiquitin ligase, and the levels of PHLPP1 expression and ubiquitination are tightly controlled by β-TrCP in cells. This β-TrCP-mediated degradation is promoted by CK1- and GSK-3β-induced phosphorylation within the phosphatase domain of PHLPP1. More importantly, a phosphorylation/degradation-deficient mutant PHLPP1 is expressed at a higher level in colon cancer cells and functions more effectively at inhibiting cell growth. Furthermore, we show that inhibition of Akt activity accelerates PHLPP1 degradation and increases the ubiquitination of PHLPP1. Taken together, our findings demonstrate a key role of Akt in regulating the stability of its negative regulator, PHLPP1.
In this study, we show that PHLPP1 is phosphorylated by CK1 and GSK-3β and that phosphorylated PHLPP1 becomes a substrate targeted by β-TrCP for degradation. There are several lines of evidence indicating that the level of PHLPP1 is controlled by phosphorylation in cells: (i) PHLPP1 expressed in cells interacts with β-TrCP, and the binding decreases upon dephosphorylation of PHLPP1; (ii) the phosphorylation-deficient mutant (4A) is expressed at an elevated level in cells and is no longer regulated by β-TrCP; and (iii) phosphorylation of PHLPP1 by GSK-3β at the Ser847 site is detected by the phospho-specific antibody, and this phosphorylated species is stabilized by the proteasome inhibitor. Our results show that phosphorylation of both the N- and C-clusters is required for β-TrCP-mediated degradation of PHLPP1. Based on the phosphorylation site analysis above, the sequences surrounding the N- and C-clusters (846QpSVLLpT851 and 866LpSVEE872, respectively) likely form the destruction motifs of β-TrCP. Although these two phospho-motifs (in boldface) found in PHLPP1 do not match the canonical consensus, they meet the requirement of two positively charged/phosphorylated residues for binding β-TrCP. Interestingly, several recently identified β-TrCP substrates all lack the canonical consensus sequences (28, 31, 34), suggesting that β-TrCP is capable of recognizing various sequence deviations for degradation. Future studies are needed to depict that β-TrCP is directly involved in the ubiquitination of PHLPP1.
Our results indicate that mutation of either the Ser867 or Ser869 site alone does not decrease CK1-mediated phosphorylation in vitro, and these single-site mutants behave exactly the same as wild-type PHLPP1 with regard to degradation (data not shown). However, the stoichiometric analysis and MS analysis suggest that Ser869 is likely preferred by CK1 in the context of wild-type PHLPP1. It has been shown that CK1-mediated phosphorylation of the adenomatous polyposis coli protein occurs at Ser1510 within a stretch of sequence containing LpSLDE, in which the two acidic residues (underlined) are required for CK1 recognition. This type of sequence is referred to as the noncanonical consensus sequence for CK1 (8). In the case of PHLPP1, the sequences surrounding Thr851 and Ser869, LpTPQDE and LpSVEE, respectively, fit the criteria of the noncanonical consensus sequence. However, if Ser869 is mutated to Ala, Ser867 may become phosphorylated since it is followed by the two acidic residues (DpSLSVEE) as well. Furthermore, the results obtained with the P847 antibody indicated that the level of phosphorylation at Ser847 is determined by the amount of active GSK-3β in cells and a priming phosphorylation at Thr851 is required. Since the CK1-mediated phosphorylation is likely to be constitutive, phosphorylation at Ser847 by GSK-3β provides the switch for controlling the tempo of PHLPP1 degradation.
As the phosphorylation sites are found in PP2C domain of PHLPP1, we analyzed whether the phosphatase activity is affected by the mutations. In vitro phosphatase assays show that the phosphatase activities of GST-PP2C and GST-PP2C/4D are comparable against purified Akt as a substrate. A similar extent of dephosphorylation is achieved with GST-PP2C/4A, although the rate of dephosphorylation is slightly slower than that of the wild-type protein (data not shown). In summary, since the phosphorylated species is rapidly degraded in cells, we believe that the phosphorylation of PHLPP1 does not directly regulate its phosphatase activity. However, the amount of PHLPP1 proteins available in cells can be effectively controlled by signals that regulate the phosphorylation of PHLPP1 in cells.
Our previous studies have shown that PHLPP1 serves as a negative regulator of Akt by directly dephosphorylating the kinase (14). Here, we found that PHLPP1 is regulated by its substrate Akt as well. It is known that Akt activity is modulated by a negative-feedback loop involving p70S6K-mediated downregulation of IRS proteins (15, 21). Our findings indicate that Akt activity can also be self-regulated by controlling the expression of its phosphatase, PHLPP1. In other words, there is a built-in regulatory mechanism in cells to sense the need for PHLPP1 based on the activation level of Akt. In unstimulated cells, basally phosphorylated PHLPP1 is degraded at a relatively fast rate. Upon stimulation by mitogens, the rate of PHLPP1 degradation slows down due to Akt-mediated inhibition of GSK-3β. As a result, increased levels of PHLPP1 proteins become available to dephosphorylate Akt in order to prevent Akt hyperactivation. Thus, our results support another feedback model in which the functions of PHLPP1 and Akt are mutually dependent: while the level of PHLPP1 in cells sets the threshold for Akt signaling, the activity of Akt controls the expression of PHLPP1.
Ubiquitination is one of the most important posttranslational modifications that plays a pivotal role in maintaining the appropriate level of signaling molecules. β-TrCP-mediated ubiquitination has been linked to cancer by its ability to degrade substrates such as IκB and β-catenin (12). Interestingly, it has been shown that β-TrCP expression and activity is elevated in colorectal cancer clinical samples with activated Wnt/β-catenin signaling (29). It is likely that loss of PHLPP expression may occur upon induction of β-TrCP in cancers with elevated Wnt/β-catenin signaling. Indeed, we find an inverse correlation between PHLPP1 expression and the level of β-TrCP protein in colon cancer cell lines (Fig. (Fig.7E).7E). Further studies are needed to determine whether loss of PHLPP1 expression as observed in colon cancer specimens is a result of dysregulation of β-TrCP-mediated ubiquitination pathways. It has been reported recently that PTEN is ubiquitinated by a ubiquitin ligase, NEDD4-1, and increased NEDD4-1 expression correlates with decreased PTEN levels in human bladder cancers (33). Therefore, blocking the ubiquitination process may present a novel strategy to increase the level of tumor suppressor proteins and enhance their function.
We thank Binhua Zhou for providing the wild-type and mutant β-TrCP1 and GSK-3β plasmids, Michele Pagano for other F-box E3 ligase plasmids, and Wade Harper for the CK1 plasmid. We thank Chunming Liu, Jianhang Jia, and Mark Evers for discussion and Pat Gulhati for critical reading of the manuscript. We also thank John Asara at the Beth Israel Deaconess Medical Center Mass Spectrometry Core Facility for assistance with the MS analysis.
This work was supported by NIH K01 CA10209-05 (T.G.), NIH R01 CA133429-01A1 (T.G.), and American Cancer Society RSG0822001TBE (T.G.).
Published ahead of print on 21 September 2009.