HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Scott, M. T. Articles by Ball, K. L. Search for Related Content PubMed PubMed Citation Articles by Scott, M. T. Articles by Ball, K. L. Social Bookmarking                      What's this?

J Biol Chem, Vol. 275, Issue 15, 11529-11537, April 14, 2000

Reversible Phosphorylation at the C-terminal Regulatory Domain of p21^Waf1/Cip1 Modulates Proliferating Cell Nuclear Antigen Binding^*

Mary T. Scott, Nick Morrice^§, and Kathryn L. Ball^¶

From the  Cancer Research Campaign Laboratories, University of Dundee Medical School, Dundee DD1 9SY and ^§ MRC Protein Phosphorylation Unit, Department of Biochemistry, University of Dundee, Dundee DD1 5EH, United Kingdom

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The p53-inducible gene product p21^WAF1/CIP1 plays a critical role in regulating the rate of tumor incidence, and identifying mechanisms of its post-translational regulation will define key pathways that link growth control to p21-dependent tumor suppression. A eukaryotic cell model system has been developed to determine whether protein kinase signaling pathways that phosphorylate human p21 exist in vivo and whether such pathways regulate the binding of p21 to one of its key target proteins, proliferating cell nuclear antigen (PCNA). Although human p21 expressed in Sf9 cells is able to form a complex with human PCNA, the inclusion of cell-permeable phosphatase inhibitors renders p21 protein inactive for PCNA binding. The treatment of this inactive isoform of p21 with alkaline phosphatase restores its binding to PCNA, suggesting that p21 expressed in Sf9 cells is subject to reversible phosphorylation at a key regulatory site(s). A biochemical approach was subsequently used to map the phosphorylation sites within p21, whose modification in vitro can inhibit p21-PCNA complex formation, to the C-terminal domain at residues Thr¹⁴⁵ or Ser¹⁴⁶. A phospho-specific antibody was developed that only bound to full-length p21 protein after phosphorylation in vitro at Ser¹⁴⁶, and this reagent was further used to demonstrate that the inactive isoform of p21 recovered from Sf9 cells treated with phosphatase inhibitors had been phosphorylated in vivo at Ser¹⁴⁶. These data identify the first phosphorylation site within the C-terminal regulatory domain of p21 whose modification in vivo modulates p21-PCNA interactions and define a eukaryotic cell model that can be used to study post-translational signaling pathways that regulate p21.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cell cycle progression is driven by the cyclin-dependent protein kinase (CDK)¹ family, which is tightly regulated by post-translational mechanisms including phosphorylation, degradation, and the action of small molecular weight kinase inhibitors. One of these kinase inhibitors the p21^WAF1/CIP1 protein (p21) was originally identified as a cyclin-dependent kinase and PCNA-binding protein (1) that was able to inhibit CDK catalytic activity (2) and as a gene whose expression was induced by the tumor suppressor protein p53 (3). It has now become clear that p21 is capable of contributing to the regulation of cell division on several different levels. These include mediation of negative growth signals, functions in differentiation and senescence, and the more recently defined roles for p21 as a modulator of the apoptotic response (4, 5) and an activator of certain cyclin-CDKs in response to mitogenic signals (6, 7).

The most well defined function of p21 linked to its activity as a growth inhibitor is as a cyclin-dependent kinase inhibitor (CKI). During the response to DNA damage, p21 levels increase in a p53-dependent manner leading to the inhibition of Cdk2 catalytic activity (6). The N-terminal half of p21 has been shown to be sufficient for both cyclin-CDK binding and inhibition as it contains a cyclin binding motif (aa 16-24) and a CDK interaction site (aa 45-65) (7-11). A second cyclin binding motif lies at the extreme C terminus of p21, and although this domain is sufficient to bind and inhibit some cyclin-CDKs, its function within the full-length protein remains unclear (12, 13). The mechanism by which p21 inhibits cyclin-CDK activity is controversial. However, recent developments showing that p21 is not a general G₁-CDK inhibitor but that it can differentially regulate individual cyclin-CDK combinations begins to explain some of the discrepancies. Whereas previously it was reported that p21 could be associated with both active and inactive G₁ cyclin-CDK complexes (14, 15) and that changes in the stoichiometry of p21 regulated this transition, recent studies have established that a single molecule of p21 is sufficient to inhibit completely the catalytic activity of cyclin A-Cdk2 (16). This correlates with studies showing that all the p21-containing Cdk2 complexes in cells are catalytically inactive (17). In contrast, all the cyclin D-Cdk4-pRb kinase activity in proliferating cells is in complex with either p21 or its close relative p27 (18, 19), and at a molar ratio of 1:1 p21 does not efficiently inhibit cyclin D-Cdk4 activity (18, 20). In fact, it appears from in vitro studies that p21 can stimulate the assembly of catalytically active cyclin D-Cdk4 by stabilizing the pre-formed complex (18), and in cells p21 appears to assemble and target the nuclear localization of cyclin D1-Cdk4 (19). Thus, whereas p21 is a potent inhibitor of Cdk2-containing complexes, it is a positive modulator of Cdk4 activity.

In addition to a second cyclin-binding site, the C terminus of p21 also contains a region that interacts with the replication and repair protein proliferating cell nuclear antigen (PCNA) (aa 144-151), a putative nuclear localization sequence (aa 140-157), and two cleavage sites for the apoptosis-associated protease, caspase 3 (21). It was originally proposed that p21 and PCNA were both components of a quaternary cyclin D-Cdk4 complex in proliferation cells (1, 14), and it has subsequently been shown that p21 and PCNA can also bind directly to each other in the absence of cyclin-CDKs (22). Binding of p21 to PCNA blocks the ability of PCNA to act as a processivity factor for DNA polymerase  and , modulating the primer template recognition complex and inhibiting DNA replication in vitro (23). Although there is clear evidence that p21-PCNA complexes form in response to DNA damage, it has proved difficult to show a significant effect of p21 on DNA replication in cells. However, recent data suggest that the interaction with PCNA is important in preventing endoreduplication (24), inhibiting S-phase progression (25) and promoting DNA repair (26, 27). The data describing a role for p21 in nucleotide excision repair (NER) suggests that although p21 can inhibit PCNA-dependent gap filling activity, it can only do this when gap filling is uncoupled from incision, and therefore cellular NER is refractive to inhibition (28). On the other hand, p21-null human colon cancer cells (HCT116) are defective in their ability to repair certain types of DNA damage, and efficient repair is restored following expression of exogenous wild type p21 but not a mutant form of p21 missing the PCNA-binding site (27). Other studies have suggested that p21 is required to aid efficient disassembly of PCNA from repair sites leading to an increase in the rate at which damaged DNA is repaired (26). Most recently, PCNA has been shown to play a role in the regulation of p21 activity by modulating the rate at which p21 is degraded (29).

To date the majority of studies involving p21 have concentrated on its transcriptional regulation by p53-dependent and -independent mechanisms (30). Evidence is now beginning to emerge that p21 can also be regulated by post-translational mechanisms. The best characterized of these is C-terminal cleavage of p21 by a member of the caspase family of proteases (21, 31-33). This appears to be an early event during DNA damage-induced apoptosis that affects both PCNA binding (31, 33) and cellular localization (32) of p21. In addition, evidence implicating reversible phosphorylation as a mechanism for regulating p21 comes from several studies. The strongest data to date, however, is a study of p21 during cellular differentiation (34). Stimulation of PC12 cells to differentiate using nerve growth factor led to loss of epitope binding by a C terminus-specific anti-p21 monoclonal antibody, although there appeared to be no overall change in p21 protein levels. Phosphatase treatment of the lysates resulted in recovery of the epitope, suggesting that reversible phosphorylation within the extreme C-terminal region of p21 occurs in response to nerve growth factor-stimulated signaling pathways. However, potential phosphorylation sites on p21 were not identified in this cell system nor were possible biochemical effects of this phosphorylation on p21 activity defined. The main difficulty in mapping such regulatory phosphorylation sites on p21 protein is the relatively low level expression of p21 protein in cultured cells that precludes an accurate biophysical study of post-translational regulatory mechanisms.

In our study, a eukaryotic cell model system was developed using recombinant human p21 protein expressed in Sf9 cells to acquire relatively large amounts of p21 in an extensively phosphorylated state for biochemical characterization. The PCNA binding activity of p21 protein synthesized in this cell system can be modulated by protein phosphatase inhibitors or phosphatase treatment, suggesting that in vivo phosphorylation of p21 can in fact modulate its activity. A biochemical approach was subsequently used to map the phosphorylation sites within p21 whose modification in vitro can inhibit p21-PCNA complex formation to the C-terminal domain at residue Thr¹⁴⁵, Ser¹⁴⁶, and Ser¹⁶⁰. An antibody specific for the phospho-Ser¹⁴⁶ p21 epitope was developed and used to demonstrate that the inactive isoform of p21 recovered from Sf9 cells treated with phosphatase inhibitors had been phosphorylated in vivo at Ser¹⁴⁶. These data identify the first phosphorylation site within the C-terminal regulatory domain of p21 whose modification in vivo modulates p21-PCNA interactions and define a eukaryotic cell model that can be used to study post-translational signaling pathways that regulate the p21 protein.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Peptides, Proteins, and Antibodies-- All biotinylated peptides were synthesized by Chiron Mimotopes, Peptide Systems (Clayton, Australia). Each peptide had a biotin-SGSG spacer at the C terminus and a free N terminus. The peptides were dissolved in Me₂SO at 5 mg/ml and stored at 20 °C. Untagged human p21 and PCNA were expressed in Escherichia coli using a pT7-7 expression vector and purified by the methods previously described (34, 35). PCNA for ELISA was expressed in Sf9 insect cells infected with baculovirus containing a human PCNA construct. The cells were harvested 2 days post-infection by low speed centrifugation, and the pellet was lysed in an equal volume of extraction buffer containing 50 mM Hepes, pH 7.4, 50 mM NaCl, 1 mM EDTA, 5 mM DTT, 1% (v/v) Nonidet P-40, 10 mM NaF, 10 µg/ml leupeptin, 4 µg/ml aprotonin, 2 µg/ml pepstatin, 1.2 mM benzamidine, 10 µg/ml soybean trypsin inhibitor, and 400 µg/ml Pefabloc before centrifuging at 14,000 rpm for 15 min. For generation of the p21-phospho-Ser¹⁴⁶ serum, phosphopeptide, KRRQTS[PO₃]MTDFYHSK (synthesized by Dr. G. Bloomberg, University of Bristol), was conjugated to keyhole limpet hemocyanin and used to immunize a rabbit by standard methods (36). For affinity purification of the rabbit sera, the same phosphopeptide, KRRQTS[PO₃]MTDFYHSK, was coupled to Reacti-Gel beads (Pierce) at a concentration of 0.8 mg of peptide per ml of beads following the manufacturer's instructions. After coupling excess reactive groups were blocked by adding 1 M Tris, pH 8.0. The phosphopeptide coupled beads were added to the rabbit sera (250 µl of beads per 10 ml of sera) and rotated for 1 h at 4 °C. The phosphatase inhibitors NaF and okadaic acid had previously been added to the rabbit sera at concentrations of 50 mM and 50 nM, respectively, to prevent endogenous phosphatases removing the phosphate group. Following incubation the beads were pelleted and washed three times with PBS containing 0.1% Tween 20. To elute the bound antibody, 0.5 ml of 0.1 M glycine, pH 2.5, was added, and the mixture was rotated at 4 °C for 30 min. The beads were pelleted; the supernatant containing the eluted antibody was collected, and the pH was neutralized using 0.1 M Tris, pH 8.0.

p21 Phosphorylation-- p21 purified from E. coli (1 µg) was incubated for 20 min at 30 °C in a final volume of 30 µl of kinase assay buffer (50 mM Hepes, pH 7.4, 10 mM MgCl₂, 0.8 mM EDTA, 0.8 mM DTT, 100 µM ATP with [-³²P]ATP (250 cpm/pmol)) containing 0.5 milliunits of protein kinase A or 0.1 milliunits of casein kinase 2 or 0.06 milliunits of protein kinase C plus 1 mM CaCl₂, 100 µg/ml phosphatidylserine, and 20 µg/ml diacylglycerol. Following incubation the reactions were stopped either by (i) removing a 10-µl sample of the reaction mixture, adding to Laemmli sample buffer, and subjecting to SDS-PAGE prior to drying and autoradiography or (ii) by adding a 20-µl sample of the reaction mixture to 0.5 ml of ice-cold 25% (w/v) trichloroacetic acid and leaving on ice for 1 h to precipitate the p21. The precipitated protein was pelleted at 10,000 rpm for 10 min, and the supernatant was removed. The pellet was washed twice with 0.5 ml of ice-cold 10% (w/v) trichloroacetic acid and twice with 0.5 ml of H₂O followed by a final rinse with acetone. The incorporation of radioactivity into the precipitated p21 was determined by Cerenkov counting. Alternatively, p21 phosphorylated at Ser¹⁴⁶ in the absence of radiolabel was detected using p21-phospho-Ser¹⁴⁶ IgG by Western blot analysis following the method of Ball and Lane (37), using the affinity purified p21-phospho-Ser¹⁴⁶ IgG and mAb 118 at 1 µg/ml.

Mapping of p21 Phosphorylation Sites-- p21 was phosphorylated by either PKA or PKC as described above, and the reactions were stopped by adding urea to a final concentration of 6 M. After reducing and alkylating any cysteines (38), the p21 was precipitated with trichloroacetic acid as described above. The precipitated pellets were dissolved in 250 mM ammonium bicarbonate containing 1% n-octyl--glucopyranoside. After reducing the detergent concentration to 0.25% with 20 mM ammonium bicarbonate, the phosphorylated p21 was digested by incubating with either trypsin or Asp-N (1:40, protease:total protein by weight) overnight at 30 °C. The resulting peptide mixtures were then separated on a C-18 reverse phase HPLC column with a 0-70% acetonitrile gradient. The peptide peaks containing radiolabeled phosphate were collected, and the phosphorylation site(s) for each enzyme were mapped using a combination of mass spectrometry and solid phase Edman degradation.

ELISA-- Peptide ELISAs for determination of PCNA binding were performed as described previously (39) using Sf9 cell lysates containing human PCNA (see above). Binding of p21-phospho-Ser¹⁴⁶ IgG to immobilized peptides was determined by ELISA using the method of Craig et al. (45). To measure the interaction of full-length p21 with PCNA, microtiter plates were coated for 16 h at 4 °C with 0.5 µg/well purified recombinant human p21 diluted in 0.1 M CO₃/HCO₃, pH 9.0. After extensive washing with PBS plus 0.2% Tween 20 (PBST), the wells were blocked with 5% non-fat milk powder in PBST (milk-PBST) for 2 h at room temperature. Increasing amounts of Sf9 cell lysate from cells expressing human PCNA diluted in 0.1% milk/PBST was added to the wells and incubated for 2 h at room temperature. The plates were incubated with anti-PCNA polyclonal antibody 3009 (37) diluted 1:1000 in 2% milk/PBST for 1 h at room temperature. This was followed by horseradish peroxidase-conjugated swine anti-rabbit IgG (Dako) diluted 1:1000 in 2% milk/PBST for 1 h at room temperature. Extensive washing with PBST was carried out between each incubation. Finally, chromogenic substrate (50 µl) complete 3,3',5,5'-tetramethylbenzidine (Sigma), prepared in accordance with the manufacturer's instructions, was added per well. The reaction was stopped by addition of 50 µl/well 1 M H₂SO₄. The plates were analyzed using a Dynatech 5000 ELISA plate reader at 450 nm.

BIAcore-- Surface plasmon resonance measurements were performed using BIAcore. Human PCNA purified from E. coli (35) was captured on a CM5 sensor chip surface by amine coupling as described in the manufacturer's instructions (Amersham Pharmacia Biotech). Full-length purified p21 was passed over the surface of the chip, and the binding characteristics of different phosphorylated forms of p21 were observed. SPR response was measured in resonance units (RU), for most proteins 1000 RU corresponds to a surface concentration of approximately 1 ng/mm². HBS (10 mM Hepes, pH 7.6, 150 mM NaCl, 3.4 mM EDTA, 0.05% BIAcore surfactant P20) was used as a buffer in this system at a flow rate of 5 µl/min.

Insect Cell Expression of Human p21-- Sf9 cells grown in Ex-Cell 400 (JRH Bioscience) + 5% FCS were infected with baculovirus carrying a full-length untagged p21 construct following the method described by Hupp and Lane (40). After 48 h the cells were treated with okadaic acid dissolved in ethanol (or ethanol alone) at a final concentration of 10 nM and incubated for a further 90 min. The cells were washed in PBS and lysed on ice in an equal volume of lysis buffer (50 mM Hepes, pH 7.6, containing 0.2% Triton X-100, 0.1 mM EDTA, 2 mM DTT, 2 mM MgCl₂, 150 mM NaCl, 10 µg/ml leupeptin, 4 µg/ml aprotonin, 2 µg/ml pepstatin, 1.2 mM benzamidine, 10 µg/ml soybean trypsin inhibitor, 400 µg/ml Pefabloc, 50 mM NaF, and 1 nM okadaic acid) or under denaturing conditions in an equal volume of urea buffer (50 mM Hepes, pH 7.6, containing 8 M urea, 1% (v/v) Triton X-100, and 100 mM DTT), centrifuged at 14,000 rpm for 15 min, and the supernatant removed.

PCNA Binding and Phosphatase Treatment-- The Sf9 cell supernatant (extracted under non-denaturing conditions, see above) was diluted 1:10 in lysis buffer containing pure recombinant PCNA at 5 µg/ml, and immunoprecipitation (41) was carried out using the anti-p21 monoclonal antibody AB1 (Oncogene Sciences) linked to protein G. Alternatively, the supernatant was diluted 1:2 in lysis buffer with and without 0.2 units/ml alkaline phosphatase (Roche Molecular Biochemicals) and incubated at 30 °C for 10 min. They were then diluted 1:5 in lysis buffer containing 2.5 µg/ml PCNA, and immunoprecipitation was carried out with AB1 as above. The immunoprecipitated complexes were separated by SDS-PAGE (42), transferred to nitrocellulose, and detected using AB1 (p21) and the polyclonal serum 3009 (PCNA (37)) following the method described in Ball and Lane (37).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Modulation of p21 Binding Using the Phosphatase Inhibitor Okadaic Acid-- Although there is increasing evidence to suggest that the biochemical activity of p21 is regulated at the post-translational level, the sites of modification have not been mapped, and the effect of modification on the biochemical activity of p21 has not been determined. We were interested in whether targeting of p21 by protein kinase signaling pathways had the potential to regulate its activity and more particularly PCNA binding. Sf9 cells expressing recombinant human p21 were used as a model system to determine whether kinase signaling pathways that target p21 exist in cells. Insect cell-viral expression systems have been previously used as a unique model to identify novel regulatory phosphorylation sites in cyclin-dependent protein kinases (43), SV40 viral T-antigen (44), and the tumor suppressor protein p53 (40, 45). The advantage that this expression system has in identifying novel signaling pathways that target a protein of interest is as follows: (i) relatively large amounts of a target protein can be purified and used for biophysical study; (ii) the target protein is not degraded, which is a common problem with cell cycle proteins normally expressed at very low levels; and (iii) eukaryotic signaling pathways are highly conserved and functioning in this cell system (46). Various growth conditions were tested in the p21 expression system, and the majority of the p21 produced in insect cells was competent for binding to E. coli expressed recombinant human PCNA (Fig. 1B, untreated). However, when the cell-permeable phosphatase inhibitor okadaic acid (OA) was added to the cell media 90 min prior to harvesting, the p21 was severely compromised in its ability to bind to PCNA (Fig. 1B, OA treated), despite the fact that 10 nM OA had no effect on the amount of p21 protein expressed (Fig. 1A). When the lysates containing p21 from OA-treated cells were incubated with alkaline phosphatase, PCNA binding to p21 was restored (Fig. 1C). These data suggest that p21 is subject to rapid reversible phosphorylation and that the addition of OA "traps" p21 in a phosphorylated form that is unable to bind to PCNA. Incubation with alkaline phosphatase removes the phosphate, relieving inhibition of binding and allowing p21 to interact with PCNA. Thus, the data indicate the possible existence of kinase/phosphatase pathways that target p21 in eukaryotic cells and can modulate PCNA binding.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Sf9 cells as a model to study p21 regulation. A, human p21 expression levels in Sf9 cells. Human p21 protein was expressed in Sf9 insect cells and left untreated or incubated in the presence of 10 nM OA for 90 min prior to harvesting. Following lysis the extracts were immunoblotted to quantitate the levels of p21 protein using the p21-specific mAb 118. B, OA treatment induces a form of p21 inactive for PCNA binding. p21-containing lysates from untreated or OA-treated cells were incubated in buffer containing bacterially expressed human PCNA. The amount of p21 or PCNA immunoprecipitated with the anti-p21 monoclonal antibody AB-1 was quantitated using either mAb 118 or 3009, respectively. C, phosphatase (PPase) treatment restores PCNA binding to OA-treated p21. p21 containing lysate from OA-treated cells were incubated in the presence or absence of alkaline phosphatase and then added to buffer containing PCNA. p21-PCNA complexes were detected by immunoprecipitation as described for B.

Phosphorylation of a p21 Peptide Inhibits PCNA Binding-- Previous studies aimed at determining the structure of p21, and its close relative p27, have shown that this family of CKIs displays little ordered structure when free in solution and that only when bound to cyclin-CDKs or PCNA is a more ridged conformation adopted (47, 48). Given this we reasoned that the best way to regulate the activity of p21 would be to sterically inhibit protein-protein interactions by phosphorylation in, or close to, the binding motif, rather than inducing conformational changes by phosphorylation at a distant site. Thus, putative phospho-acceptor sites that lie within the PCNA binding motif on p21 were identified (Fig. 2). A cluster of serine and threonine residues is present within the C-terminal domain of p21 that binds to PCNA, and a "motif" search identified three of these residues as lying within possible consensus phosphorylation sites.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   Functional and regulatory domains of p21. The primary amino acid sequence of human p21 is depicted. Highlighted are putative kinase phosphorylation consensus sites, and underlined are the known binding sites for cyclin, CDK, and PCNA.

We and others (37, 39) have shown previously that a 20-amino acid peptide based on residues 140-160 of p21 is sufficient to bind and form a high affinity interaction with PCNA that can be detected by peptide precipitation and ELISA. A quantitative ELISA was developed to compare PCNA binding to p21-derived phosphopeptides with PCNA binding to the unphosphorylated p21-based peptide. The synthetic peptides contained the core amino acid sequence ¹KRRQTSMTDFYHSKRRLIFSKRKP²⁴, derived from the sequence of human p21 protein where position 1 is equivalent to Lys¹⁴¹ and position 24 is Pro¹⁶⁴. Increasing amounts of PCNA-containing lysate was titrated into ELISA wells pre-coated with the indicated biotinylated synthetic peptide, and the amount of PCNA bound was quantitated using a polyclonal antibody specific for the protein (Fig. 3). Incorporation of a phosphate at positions 5 (Thr¹⁴⁵) and 6 (Ser¹⁴⁶) decreases PCNA binding more efficiently than phosphorylation at positions 13 (Ser¹⁵³) with phosphorylation at position 20 (Ser¹⁶⁰) also giving a significant decrease in PCNA binding (Fig. 3). This indicates that phosphorylation at key residues can, as predicted, reduce the stability of the PCNA-p21 peptide complex.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Phosphorylation of p21-derived peptides within the C-terminal regulatory domain reduces PCNA binding. Biotinylated peptides based on the C-terminal domain of p21 (residues 141-164; KRRQTSMTDFYHSKRRLIFSKRKP) were used in an ELISA format (39) to determine if phosphorylation at Thr145 (), Ser146 (), Ser153 (), and Ser160 () affected the ability of the peptide to form a high affinity interaction with full-length PCNA from Sf9 cell lysates and were compared with binding to the unphosphorylated peptide (). PCNA binding was detected using the anti-PCNA sera 3009 (37). The data represent the mean of three experiments.

PKA and PKC Phosphorylate p21 Protein-- As phosphorylation of the PCNA-binding peptide derived from p21 reduced PCNA binding, it was important to determine whether phosphorylation of full-length p21 within this region also affected PCNA binding. The motif search suggested that the three serine residues whose phosphorylation reduced PCNA binding were potential sites for PKA (cAMP-dependent protein kinase), CK2 (casein kinase 2), and PKC (protein kinase C) (see Fig. 2), and the ability of these three kinases to phosphorylate untagged full-length p21 purified from a bacterial expression system was examined (Fig. 4). Incorporation of radioactive phosphate into p21 in the presence of each protein kinase was determined by SDS-PAGE/autoradiography (Fig. 4A) and quantitated by trichloroacetic acid precipitation assays (Fig. 4B). When p21 was incubated with either PKA or PKC for 20 min at 30 °C in the presence of [-³²P]ATP, the stoichiometry of phosphorylation was calculated to be 0.5 mol of phosphate/mol of p21 for PKA and 0.7 mol of phosphate/mol of p21 for PKC. On the other hand, p21 was not a substrate for CK2 (Fig. 4) under conditions where p53 was phosphorylated efficiently by this enzyme (data not shown). These results suggest that full-length p21 could be phosphorylated with a high stoichiometry by PKA and PKC and prompted us to map the phosphorylation sites.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Phosphorylation of full-length p21 by PKA and PKC. The incorporation of [32P]phosphate into full-length p21 was determined following incubation with PKC, PKA, or CK2, as indicated. A, an autoradiograph of the samples analyzed by SDS-PAGE. B, data obtained by scintillation counting showing the rate of phosphate incorporation (pmol/min/µg p21) and the stoichiometry defined in moles of phosphate/mol of protein.

PKA and PKC Phosphorylation within the C terminus of Full-length p21 Is Similar but Distinct from That Observed within the p21 Peptide-- Full-length untagged recombinant p21 purified from a bacterial expression system was phosphorylated by PKA and PKC using radiolabeled ATP, and phosphorylation sites were mapped by standard procedures. Urea was used to stop the kinase reactions, and following reduction and alkylation of the cysteine residues, the reaction products were precipitated with trichloroacetic acid thus removing unincorporated [³²P]ATP. The resuspended protein was proteolyzed using either trypsin or Asp-N, and the resultant peptides were separated by reverse phase HPLC on a C-18 column. The column was developed with a linear gradient of 0-70% acetonitrile, and those polypeptide peaks containing radiolabeled phosphate were analyzed by mass spectrometry and subjected to Edman degradation to identify the phosphorylated residues. Based on these methods, PKA was found to phosphorylate full-length p21 at Thr¹⁴⁵ (Fig. 5, C and D). In addition, PKC phosphorylated two sites on full-length p21 at Ser¹⁴⁶ and Ser¹⁶⁰ (Fig. 5, A, B, and D). Interestingly, when similar experiments were performed using the p21 C-terminal peptide as a substrate (data not shown) the preferential phosphorylation site for PKA was Ser¹⁴⁶ and not the Thr¹⁴⁵ site mapped on the full-length protein. As it is becoming increasingly clear that many protein kinases require determinants other than the defined minimal substrate motif for efficient phosphorylation, this difference suggests that another region(s) of the full-length protein may influence PKA specificity. This highlights the importance of using native protein, as well as peptide substrates when screening for physiologically relevant kinases during the fractionation of cell lysates.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Mapping the PKA and PKC phosphorylation sites on full-length p21 protein. p21 was phosphorylated with either PKC or PKA using [-32P]ATP and digested with either trypsin or Asp-N. The resulting peptide mixtures were separated on a C-18 reverse phase HPLC column using a 0-70% acetonitrile gradient. The peptide peaks containing radiolabeled phosphate were collected, and the phosphorylation sites for each enzyme were mapped using a combination of mass spectrometry and solid phase Edman degradation. Two radiolabeled peptides were identified after phosphorylation with PKC (A and B) and one peptide after phosphorylation with PKA (C). The data from the Edman degradation show the released of 32P (cpm) per residue, and the sites phosphorylated by the two enzymes are summarized in the table (D).

Phosphorylation of Full-length p21 Protein in Vitro by Either PKA or PKC Inhibits PCNA Binding-- To determine the effect of phosphorylating full-length p21 at either residue Thr¹⁴⁵ or Ser¹⁴⁶ and Ser¹⁶⁰ on PCNA binding, an ELISA was used to quantitate the binding reaction (Figs. 6, A and B). The ELISA wells were coated with either unphosphorylated p21 or with p21 that had been phosphorylated in the presence of either PKA or PKC. The binding of p21 alone to the ELISA wells and of PCNA to p21 was quantitated using the anti-p21 monoclonal antibody AB-1 or an anti-peptide sera to the C-terminal 15 amino acids of PCNA (3009 (37)). Phosphorylation of p21 by either PKA or PKC did not affect binding of p21 to the wells (data not shown), whereas incorporation of phosphate at Thr¹⁴⁵ by PKA and dual phosphorylation at Ser¹⁴⁶ and Ser¹⁶⁰ by PKC prevents PCNA binding (Fig. 6, A and B). Although this assay clearly demonstrates an effect of phosphorylation on the ability of p21 to form a stable complex with PCNA, it does not determine whether this is because phosphorylation blocks p21-PCNA complex formation or whether it destabilizes the complex once it has formed.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Phosphorylation of full-length p21 in vitro inhibits PCNA binding. The ability of phosphorylated and unphosphorylated p21 to bind to PCNA was determined by ELISA (A and B) and by surface plasmon resonance using BIAcore (C and D). A and B, p21 incubated in the absence or presence of PKA (A) or PKC (B) was adsorbed to microtiter wells and incubated with increasing amounts of PCNA containing Sf9 cell lysate. The amount of p21 protein bound to the well was unaffected by phosphorylation (data not shown). PCNA binding was quantitated using the anti-PCNA sera 3009. The data are the means ± S.E. of four experiments with the OD being read at 450 nm. C and D, p21 incubated in the absence or presence of PKA (C) or PKC (D) was passed over PCNA bound to a sensor chip, and the SPR response was measured in RUs (resonance units) as a function of time (in seconds).

To address this problem and determine whether phosphorylation had an effect on the association rate or the dissociation rate of the complex, BIAcore was used to measure surface plasmon resonance (SPR) and determine whether there was a difference in the real time binding of unphosphorylated and phosphorylated p21 to PCNA. PCNA purified from a bacterial expression system was captured on a CM5 sensor chip and either untreated, PKA-phosphorylated p21, or PKC-phosphorylated p21 was then passed over the surface of the chip. The resulting binding curves clearly show that phosphorylation of p21 by both PKA and PKC strongly inhibits real-time binding of the protein to PCNA (Fig. 6, C and D). In addition, as the dissociation rate of any bound p21 is unaffected by phosphorylation, the data suggest that phosphorylation prevents the formation of stable complexes by reducing the association of phospho-p21 and PCNA. These data also suggest that the interaction between p21 and PCNA could be regulated in vivo by protein kinases that modify residues within the PCNA binding motif of p21 and prompted us to develop reagents that could be used to determine whether the isoform of p21 we can recover from cell lysates, which is inactive for PCNA binding, is phosphorylated within the C-terminal domain.

The Development of a Quantitative Immunochemical Assay to Detect Ser¹⁴⁶ Phosphorylation of p21-- Recent studies have demonstrated that the conventional method for mapping and quantifying phosphorylation events in vivo which relies on metabolic labeling with [³²P]orthophosphate can result in DNA damage and induction of p53-dependent stress-activated growth arrest pathways (49). The use of immunochemical reagents that are specific for phosphorylated substrates circumvents these problems and allows the study of protein phosphorylation under non-invasive conditions. Thus, we developed -phosphopeptide rabbit sera (p21-phospho-Ser¹⁴⁶) that is specific for the Ser¹⁴⁶ phospho-epitope using both p21-based peptides and full-length p21. When the C-terminal p21 peptide series was used to determine the specificity of affinity purified p21-phospho-Ser¹⁴⁶ IgG (Fig. 7A), the IgG bound to the phospho-Ser¹⁴⁶ peptide but failed to bind to both the unphosphorylated peptide and peptides phosphorylated at position Thr¹⁴⁵, Ser¹⁵³, and Ser¹⁶⁰, showing absolute specificity under these conditions for the Ser¹⁴⁶ phosphopeptide. A titration of PKC-phosphorylated full-length p21 protein was compared with unphosphorylated full-length p21 by Western blot analysis, where it was demonstrated that the p21-phospho-Ser¹⁴⁶ IgG was at least 100-fold more efficient at binding to p21 phosphorylated by PKC than to unphosphorylated protein (Fig. 7B). In addition, to show that the antibody was specific for phosphorylation at Ser¹⁴⁶, the p21-phospho-Ser¹⁴⁶ IgG did not bind to full-length p21 phosphorylated at Thr¹⁴⁵ by PKA (Fig. 7C).

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Characterization of the phospho-specific antibody, p21-phospo-Ser146 IgG. A, binding of p21-phospho-Ser146 IgG to immobilized synthetic peptides was determined by ELISA (45). The peptide (KRRQTSMTDFYHSKRRLIFSKRKP) was either unphosphorylated () or phosphorylated at one of the following positions: Thr145 (), Ser146 (), Ser153 (), or Ser160 (). B, the affinity of the p21-phospho-Ser146 sera for bacterially expressed full-length p21 phosphorylated at Ser146 compared with unphosphorylated p21 was determined by Western blot analysis. The upper panel shows p21 that was phosphorylated with PKC, and the bottom two panels are of unphosphorylated p21. The anti-p21 monoclonal antibody mAb 118 is used to detect total p21 protein. C, equal amounts of bacterially expressed p21 protein phosphorylated by PKA (lane 1), PKC (lane 2), or unphosphorylated (lane 3) were analyzed by Western blot analysis. In the bottom panel, total p21 protein was detected using mAb 118, and in the top panel p21-phospho-Ser146 was shown to be specific for p21 phosphorylated at Ser146.

The PCNA-binding Inactive Isoform of p21 Expressed in Sf9 Cells Is Phosphorylated at Ser¹⁴⁶-- As one of the key modifications that can inhibit PCNA binding is phosphorylation of p21 at Ser¹⁴⁶ (Figs. 3 and 6), we tested the hypothesis that the inactive isoform of p21 (lysed from Sf9 cells treated with phosphatase inhibitors) was inactivate for PCNA binding due to phosphorylation at Ser¹⁴⁶. When Sf9 cells expressing p21 protein were lysed in either Nonidet P-40 lysis buffer or urea lysis buffer, the PCNA-binding competent form of p21 (Fig. 1) had a very low level or undetectable level of Ser¹⁴⁶ phosphorylation (Fig. 8, upper panel). Strikingly, when Sf9 cells expressing p21 protein were pretreated with phosphatase inhibitors (+ OA, Fig. 8, upper panel), this PCNA-binding inactive form of p21 (Fig. 1) had significantly higher levels of Ser¹⁴⁶ phosphorylation (Fig. 8, upper panel). As a control, an identical gel was transferred to nitrocellulose and probed with the p21-specific monoclonal antibody 118 to show that the increase in Ser¹⁴⁶ phosphorylation observed in the presence of OA occurred without an increase in p21 protein levels (Fig. 8, lower panel). Although we cannot rule out the possibility that other in vivo phosphorylation events contribute to the inactivation of p21 for PCNA binding (for example at Thr¹⁴⁵), there is a strict correlation between the loss of PCNA binding activity and an increase in Ser¹⁴⁶ phosphorylation of p21. Thus, we have demonstrated that phosphorylation at Ser¹⁴⁶ can be stabilized by OA and contributes to the loss of the PCNA binding function of p21. In addition, the data show that signaling pathways that target Ser¹⁴⁶ of p21 protein exist in eukaryotic cells.

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Ser146 of p21 is phosphorylated in vivo. Sf9 cells expressing human p21 were incubated in the absence or presence of the phosphatase inhibitor OA (as indicated) for 90 min prior to harvesting. Control and treated cells were extracted in lysis buffer containing Nonidet P-40 and phosphatase inhibitors (see under "Experimental Procedures") or under denaturing conditions using urea buffer. The lysates were subjected to Western blot analysis with bacterially expressed full-length p21 phosphorylated with PKC (lane 1) or unphosphorylated p21 (lanes 2 and 3) being included as controls. The bottom panel shows total p21 protein detected by the anti-p21 mAb 118, and the upper panel shows Ser146 phosphorylated p21 detected using p21-phospho-Ser146 IgG.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Recent studies have made it apparent that the absolute levels of the cyclin-dependent kinase inhibitor protein p21 are important in determining the rate of tumor incidence (50, 51) and that levels of p21 protein can determine the sensitivity of cells within a solid phase tumor to radiation-induced damage (52). Thus, enzymes that modify the specific activity of p21 as a growth suppressor may be expected to impact on growth control mechanisms and the rate of tumorigenesis. We have initiated studies to determine if post-translational mechanisms contribute to the control of p21 activity as a PCNA-binding protein.

p21-PCNA protein complexes form in cells in response to DNA damage (53) and appear to be involved in multiple damage responses. First, although it has previously proven difficult to demonstrate that p21 inhibits DNA replication in cells, two recent studies suggest that p21 inhibits S-phase progression and DNA synthesis in a PCNA-dependent manner (25, 54). Second, p21 appears to stimulate PCNA-dependent NER, and p21-deficient cells are unable to repair efficiently certain types of damaged DNA (26, 27). Third, it has been proposed that the ability of p21 to arrest cells in the G₂ phase of the cell cycle (54-57) is also a PCNA-dependent process, at least in some cell types (54). Finally, steady state levels of p21 are regulated by both its rate of protein synthesis and degradation, with recent studies demonstrating that PCNA plays a critical role in determining the rate of p21 turnover. When p21 is bound to PCNA it is degraded more slowly leading to a decrease in its rate of turnover. This is in contrast to cyclin-CDK binding which increases the rate of p21 degradation (29).

By using a human cell line we observed that p21 and PCNA could form a complex after very low doses of UVC, and the interaction occurred in the absence of protein synthesis, implicating a post-translational mechanism for regulating p21-PCNA complex assembly.² However, human cell lines express relatively small amounts of p21, and we have found that the protein is associated with a number of different protein complexes making quantitative immunoprecipitation difficult. In addition, given that (i) many regulatory post-translational modifications might be short lived in cells and (ii) labeling with [³²P]orthophosphate leads to activation of p53-dependent pathways (49), it is difficult to dissect accurately post-translational signaling to p21 in mammalian cells using conventional methods. In the current study these difficulties have been overcome by first developing a eukaryotic cell system to study post-translational regulation of p21 protein and second by generating immunochemical reagents to study p21 phosphorylation, bypassing the requirement for metabolic labeling with ³²P.

The evolutionary conservation of signaling pathways that regulate key cell cycle and checkpoint proteins in Sf9 cells is now well documented. Examples of how such cells can be utilized to study post-translational modification are provided by elegant studies on the regulation of p53 and cyclin-CDKs by reversible phosphorylation. Insect cells expressing human p53 have revealed a UV- and serum-dependent signaling pathway that induces phosphorylation of p53 within the C-terminal negative regulatory domain (40), and most recently, novel regulatory phosphorylation sites within the Mdm2-binding site of p53 have been identified in this system (45). In both cases subsequent research has shown that these pathways also exist in human cells (58, 59). In addition, studies on the regulation of human cyclin-CDK catalytic activity in insect cell lysates lead to the discovery of the cyclin-dependent kinase activating kinase (43), an essential activator of the cell cycle-regulated CDKs.

The advantage of the insect cell expression system for studying p21 modification is that relatively large amounts of full-length p21 protein can be produced in a soluble form for biophysical study. Our initial data using this system suggested that agents that inhibit protein phosphatase activity could be used to modulate the PCNA binding activity of p21; however, whether this reflected the direct modification of p21 or an indirect mechanism was unclear. The approach taken to address this issue was to identify putative in vitro phosphorylation sites with the potential to disrupt the high affinity interaction normally observed between p21 and PCNA (22, 37, 39). By using this approach we identified three putative phosphorylation sites within the C-terminal domain of p21 which significantly decreased its ability to bind to PCNA by steric hindrance. In order to determine if these sites were phosphorylated in Sf9 cells and if phosphorylation was increased by the addition of phosphatase inhibitors, we chose to make antisera to synthetic phosphopeptides representing the C-terminal phosphorylation sites. Rabbit sera produced to a Ser¹⁴⁶-phosphorylated peptide had a strict requirement for phosphorylation at Ser¹⁴⁶ of full-length p21 with the affinity purified IgG failing to bind to unphosphorylated p21 or p21 phosphorylated at Thr¹⁴⁵. By using this reagent, it was demonstrated that the effect OA had on PCNA binding activity of p21 was accompanied by stabilization of Ser¹⁴⁶ phosphorylation within the C terminus of p21. A site that we also showed could strongly inhibit PCNA binding.

In addition to the fact that this immunochemical approach allowed us to study p21 modification in the absence of DNA-damaging radioisotopes, it has also provided a valuable reagent that can be used to develop a site-specific assay for the Ser¹⁴⁶ kinase using purified full-length p21 as a substrate. Although in the current study PKC and PKA were used as tools to phosphorylate p21 in vitro for biochemical analysis, there is no evidence that these kinases are relevant to the physiological regulation of p21 in cells. Thus, we are currently using the p21-phospho-Ser¹⁴⁶ IgG to develop a sensitive ELISA-based assay to screen for endogenous kinase activity during fractionation of cell lysates by column chromatography.

The region of p21 involved in binding to PCNA is found within the C-terminal domain of p21 with residues 140-160 being sufficient to form a stable interaction with PCNA (37, 39, 60) and to inhibit its function in DNA replication (39). The crystal structure of human PCNA complexes with a C-terminal peptide containing the PCNA binding motif has been solved (60). The structure shows that p21 interacts with the interdomain connector loop of PCNA and is therefore likely to prevent the interaction of PCNA with other replication factors. p21 appears to be anchored to PCNA via hydrogen bonds. In addition, an anti-parallel -strand structure is formed with the connector, and basic regions of the peptide form electrostatic interactions with PCNA, and hydrophobic residues in two regions of the peptide lie in complementary hydrophobic cavities on the surface of PCNA. If the residues that we have identified as phosphorylation sites on p21 are compared with the structure of the PCNA-p21 peptide complex, it is possible to see why phosphorylation is able to disrupt the interaction of p21 with PCNA, as all three of the residues are involved in direct interactions with PCNA. The hydroxyl group of Ser¹⁴⁶ contributes to an intramolecular hydrogen bond stabilizing a 3₁₀ helix, and this amino acid is also part of a hydrophobic cluster that fits into a corresponding hydrophobic pocket on PCNA. Thus, phosphorylation at this residue would interfere with secondary structure as well as introducing a negative charge into a hydrophobic region. Phosphorylation at Thr¹⁴⁵ would similarly disrupt the formation of an intermolecular hydrogen bond with Pro²⁵³ of PCNA, whereas the hydroxyl group of Ser¹⁶⁰ is involved in hydrogen bonds with both Gly⁶⁹ and Met¹¹⁹ of PCNA.

Our current data suggest a model where post-translational mechanisms may contribute to the regulation of p21 by preventing formation of some complexes and promoting others. Thus, in dividing cells we suggest that phosphorylation of p21 may suppress formation of a linear complex between p21 and PCNA and that following exposure to damaging agents dephosphorylation of the C-terminal regulatory domain would be required to promote a complex between these two proteins. Additional properties of p21 that could potentially be subject to regulation by C-terminal phosphorylation include steady state levels, nuclear localization, and Cdk4 targeting. As the rate of p21 protein degradation can be influenced by PCNA binding, C-terminal phosphorylation may be expected to decrease the absolute level of p21 protein in a cell. The nuclear localization of p21 and its ability to target cyclin D-Cdk4 complexes to the nucleus both require an intact C-terminal regulatory domain. A putative bipartite nuclear localization sequence lies between residues 140 and 157, and the C-terminal domain of p21 has been shown by LaBaer et al. (18) to be essential for the nuclear localization of cyclin D-Cdk4. Truncation mutants missing the extreme C terminus disperse throughout the cell, whereas N-terminal deletion constructs that retain the nuclear localization sequence show strong nuclear staining (25). Recent reports highlight the importance of p21 subcellular localization for physiological outcome following DNA damage. Thus, localization of p21 could determine the response to apoptotic signals with cytoplasmic accumulation of full-length p21 providing protection against apoptotic cell death (4).

	ACKNOWLEDGEMENT

We thank Eleanor Ramsay for valuable technical assistance.

	FOOTNOTES

^* This work was supported by the Cancer Research Campaign.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ Cancer Research Campaign Senior Cancer Research Fellow. To whom correspondence should be addressed. Tel.: 44-1382-425582; Fax: 44-1382-669993; E-mail: k.l.ball@dundee.ac.uk.

² M. T. Scott and K. L. Ball, manuscript in preparation.

	ABBREVIATIONS

The abbreviations used are: CDK, cyclin-dependent protein kinase; PCNA, proliferating cell nuclear antigen; aa, amino acid; NER, nucleotide excision repair; ELISA, enzyme-linked immunosorbent assay; DTT, dithiothreitol; SPR, surface plasmon resonance; RU, resonance units; PBS, phosphate-buffered saline; PAGE, polyacrylamide gel electrophoresis; OA, okadaic acid; HPLC, high pressure liquid chromatography; PKC, protein kinase C; PKA, cAMP-dependent protein kinase.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES