Distinct Roles for PP1 and PP2A in Phosphorylation of the Retinoblastoma Protein

The function of the retinoblastoma protein (pRB) in controlling the G1 to S transition is regulated by phosphorylation and dephosphorylation on serine and threonine residues. While the roles of cyclin-dependent kinases in phosphorylating and inactivating pRB have been characterized in detail, the roles of protein phosphatases in regulating the G1/S transition are not as well understood. We used cell-permeable inhibitors of protein phosphatases 1 and 2A to assess the contributions of these phosphatases in regulating cyclin-dependent kinase activity and pRB phosphorylation. Treating asynchronously growing Balb/c 3T3 cells with PP2A-selective concentrations of either okadaic acid or calyculin A caused a time- and dose-dependent decrease in pRB phosphorylation. Okadaic acid and calyculin A had no effect on pRB phosphatase activity even though PP2A was completely inhibited. The decrease in pRB phosphorylation correlated with inhibitor-induced suppression of G1cyclin-dependent kinases including CDK2, CDK4, and CDK6. The inhibitors also caused decreases in the levels of cyclin D2 and cyclin E, and induction of the cyclin-dependent kinase inhibitors p21Cip1 and p27Kip1. The decrease in cyclin-dependent kinase activities were not dependent on induction of cyclin-dependent kinase inhibitors since CDK inhibition still occurred in the presence of actinomycin D or cycloheximide. In contrast, selective inhibition of protein phosphatase 1 with tautomycin inhibited pRB phosphatase activity and maintained pRB in a highly phosphorylated state. The results show that protein phosphatase 1 and protein phosphatase 2A, or 2A-like phosphatases, play distinct roles in regulating pRB function. Protein phosphatase 1 is associated with the direct dephosphorylation of pRB while protein phosphatase 2A is involved in pathways regulating G1cyclin-dependent kinase activity.


From the Department of Pharmacology, The University of Texas Southwestern Medical Center, Dallas, Texas 75235-9041
The function of the retinoblastoma protein (pRB) in controlling the G 1 to S transition is regulated by phosphorylation and dephosphorylation on serine and threonine residues. While the roles of cyclin-dependent kinases in phosphorylating and inactivating pRB have been characterized in detail, the roles of protein phosphatases in regulating the G 1 /S transition are not as well understood. We used cell-permeable inhibitors of protein phosphatases 1 and 2A to assess the contributions of these phosphatases in regulating cyclin-dependent kinase activity and pRB phosphorylation. Treating asynchronously growing Balb/c 3T3 cells with PP2A-selective concentrations of either okadaic acid or calyculin A caused a time-and dose-dependent decrease in pRB phosphorylation. Okadaic acid and calyculin A had no effect on pRB phosphatase activity even though PP2A was completely inhibited. The decrease in pRB phosphorylation correlated with inhibitor-induced suppression of G 1 cyclin-dependent kinases including CDK2, CDK4, and CDK6. The inhibitors also caused decreases in the levels of cyclin D2 and cyclin E, and induction of the cyclin-dependent kinase inhibitors p21 Cip1 and p27 Kip1 . The decrease in cyclin-dependent kinase activities were not dependent on induction of cyclin-dependent kinase inhibitors since CDK inhibition still occurred in the presence of actinomycin D or cycloheximide. In contrast, selective inhibition of protein phosphatase 1 with tautomycin inhibited pRB phosphatase activity and maintained pRB in a highly phosphorylated state. The results show that protein phosphatase 1 and protein phosphatase 2A, or 2A-like phosphatases, play distinct roles in regulating pRB function. Protein phosphatase 1 is associated with the direct dephosphorylation of pRB while protein phosphatase 2A is involved in pathways regulating G 1 cyclin-dependent kinase activity.
The retinoblastoma tumor suppressor gene encodes a nuclear phosphoprotein (pRB) 1 that regulates the G 1 /S transition of the cell cycle. The active form of pRB binds and inactivates transcription factors, including members of the E2F family, whose target genes are necessary for S phase (1)(2)(3). The activity of pRB is regulated by phosphorylation and dephosphorylation of serine and threonine residues. pRB is dephosphorylated during mitosis, and the active, hypophosphorylated form inhibits cell cycle progression during early and mid G 1 (4). pRB accumulates in the inactive, hyperphosphorylated state in late G 1 and phosphorylation is maintained during S and G 2 . Hyperphosphorylation causes dissociation of pRB from E2F, induction of gene transcription, and progression into S phase (5). Because of its central role in progression through G 1 , pRB serves as an important point of integration for numerous signaling pathways that influence the cell cycle (2). pRB is phosphorylated by members of the cyclin-dependent family of serine/threonine kinases whose active forms consist of a catalytic subunit (CDK) complexed with a cyclin partner (2,3). The major kinases that phosphorylate pRB during G 1 include cyclin E-CDK2, cyclin D-CDK4, and cyclin D-CDK6. pRB phosphorylation is initiated in a growth factor-dependent manner by assembly of cyclin D with CDK4. Phosphorylation is then accelerated during late G 1 by cyclin E-CDK2 (6). Maintenance of the phosphorylated state during S and G 2 is due to the actions of cyclin A-and cyclin B-CDK complexes. Cyclin-CDK complexes phosphorylate multiple proline-directed consensus sites on pRB (7). The activities of G 1 CDKs are regulated not only by the availability of cyclins, but also by activating (8) and inhibitory (9,10) phosphorylation of the CDK catalytic subunit. CDK activity is down-regulated by cyclin-dependent kinase inhibitors. Inhibitors of G 1 CDKs include p21 Cip1 , p27 Kip1 , and p57 Kip2 . Cyclin D-CDKs are also inhibited by a set of specific CDK inhibitors termed INK4 proteins (11)(12)(13).
pRB is reactivated by dephosphorylation at the end of mitosis. Protein phosphatase 1 has been implicated as the major pRB phosphatase in vivo. Dephosphorylation of pRB by PP1 is thought to play a critical role in controlling the G 1 /S transition (14). The phosphorylation state (15) and activity (16,17) of PP1 vary during the cell cycle. Dephosphorylation of pRB by mitotic cell extracts is sensitive to inhibitors of PP1 (16), and a high molecular weight form of PP1 has been isolated as a pRB phosphatase (18). pRB can associate with the catalytic subunit of PP1 (19), and a constitutively active form of PP1 induces dephosphorylation of pRB and pRB-dependent cell cycle arrest (20).
Protein phosphatase 2A has been implicated in the regulation of many cellular functions including the cell cycle (21)(22)(23). Studies in Xenopus extracts (24) and fission yeast (25,26) have demonstrated that PP2A plays a role in the G 2 to M transition. The G 2 /M function of PP2A is likely to involve regulation of the activation or activity of the cyclin B-CDC2 protein kinase (27,28). There is also evidence that PP2A functions in the G 1 /S transition. Exposure of mammalian cells to the phosphatase inhibitor okadaic acid causes dephosphorylation of pRB, inhibition of DNA synthesis, and G 1 arrest (29 -33). Since PP2A and PP2A-like phosphatases are more sensitive to okadaic acid than PP1, these results imply that the effects of this toxin in G 1 may be due to suppression of PP2A. They also suggest that PP2A activity is required for entry into S phase. The dephosphorylation of pRB in response to okadaic acid is somewhat paradoxical and indicates that an okadaic acid-insensitive phosphatase dephosphorylates pRB. While these results demonstrate that protein phosphatases are intimately involved in the control of the G 1 /S transition and the dynamic regulation of pRB phosphorylation, the sites of action of individual enzymes are not known.
Several toxins are highly specific inhibitors of members of the PPP family of serine/threonine phosphatases (34,35). The differential sensitivities to these inhibitors has provided methods to identify and quantitate the levels of PP1 and PP2A in cell and tissue extracts (36). Several of these inhibitors including okadaic acid, calyculin A, and tautomycin are membranepermeable and potently inhibit phosphatase activity in intact cells. Due to their differential affinities for PP1 and PP2A and their distinct permeation properties, these three inhibitors can inhibit PP1 and PP2A in a highly selective manner (37). We used these inhibitors to identify the roles and potential mechanisms of PP1 and PP2A in regulating pRB phosphorylation. The results demonstrate that both phosphatases are involved in controlling the level of pRB phosphorylation but that they act through distinct mechanisms. PP2A, or a PP2A-like phosphatase, is crucial for activation or maintenance of G 1 cyclindependent kinase activity, while PP1 directly dephosphorylates pRB.

EXPERIMENTAL PROCEDURES
Cell Culture and Inhibitor Treatment-Mouse Balb/c 3T3 cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum in an atmosphere of 5% CO 2 . Log-phase cells (1 ϫ 10 6 cells per 100-mm dish) were incubated in media containing calyculin A (CL-A), okadaic acid (OA), or tautomycin (TAU) (LC Services Corp.). CL-A and OA were prepared by dissolving in dimethyl sulfoxide, and TAU was solubilized in methanol. Control incubations included the same amounts of vehicle alone (final concentration, 0.05%).
Antibodies and Recombinant Proteins-Anti-pRB antibodies included mouse monoclonal antibodies AB-1 (Oncogene Science) and G3-245 (Pharmingen), and C-15, an affinity-purified rabbit polyclonal antibody (Santa Cruz Biotechnology). These anti-pRB antibodies recognize both the phosphorylated and non-phosphorylated forms of pRB from a number of species. Antibodies against p16 (M-156), p21 (C- 19), and p27 (C-19) were affinity purified rabbit IgG fractions (Santa Cruz Biotechnology). The antibody against cyclin D (H295) was a rabbit polyclonal IgG that reacts with cyclins D1, D2, and D3 from a number of species (Santa Cruz Biotechnology). Antibodies against cyclin D1 (R-124) (Santa Cruz Biotechnology), cyclin D2 (PharMingen), and cyclin D3 (PharMingen) were mouse monoclonal antibodies that specifically recognize the individual cyclin isoforms. Anti-cyclin E antibodies included (M-20)-R, a rabbit polyclonal IgG fraction, and (M-20)-G, a goat polyclonal IgG fraction (Santa Cruz Biotechnology). Monoclonal antibody PAb419, directed against SV40 tumor antigens, was used as a negative control (38). Antibody (M2) is a rabbit polyclonal IgG that reacts with the C terminus of CDK2 (Santa Cruz Biotechnology). Anti-CDK4 antibodies included (C-22)-R, a rabbit polyclonal IgG fraction, and (C-22)-G, a goat polyclonal IgG fraction (Santa Cruz Biotechnology). Antibody (C-21) is a rabbit polyclonal IgG that recognizes CDK6 (Santa Cruz Biotechnology). The pRB protein used as substrate in the kinase and phosphatase assays was a glutathione S-transferase fusion with the C-terminal domain of mouse pRB (Santa Cruz Biotechnology).
Immunoprecipitation and Western Blotting-Labeling cells with [ 35 S]methionine, immunoprecipitations, and immunoblotting were performed as described previously (38) except that the lysis buffer for CDK immunoprecipitations was 25 mM HEPES, pH 7.4, 0.4% Triton X-100, 300 mM NaCl, 1 mM EDTA, 1.5 mM MgCl 2 , 1 mM sodium pyrophosphate, 50 mM sodium fluoride, 1 mM sodium orthovanadate, and 20 mM ␤-glycerophosphate. Immunoprecipitated CDKs were assayed for kinase activity or separated by SDS-polyacrylamide gel electrophoresis and detected by immunoblotting with anti-CDK antibodies. The phosphorylation state of pRB was monitored by the differential mobility of the hyper-and hypophosphorylated forms during SDS-polyacrylamide gel electrophoresis (39).
CDK Kinase Assay-Balb/c 3T3 cells were treated with CL-A at the doses and times indicated in the figure legends. Cells were lysed and aliquots of lysates containing equal amounts of protein were immunoprecipitated with anti-CDK antibodies. The immune complexes were isolated with protein A-agarose, washed three times with wash buffer (20 mM HEPES, pH 7.4, 50 mM NaCl, 2.5 mM MgCl 2 , 0.1 mM EDTA, 0.1% Triton X-100, and 0.1 mM sodium orthovanadate), and twice with kinase buffer (20 mM HEPES, pH 7.4; 10 mM MgCl 2 ; 1 mM EDTA, 1 mM EGTA, 20 mM p-nitrophenyl phosphate, 20 mM ␤-glycerophosphate, 0.1 mM sodium orthovanadate, and 1 mM DTT). Resuspended immune complexes were incubated with 1 g of GST-RB, 40 M ATP, and 5 Ci of [␥-32 P]ATP (6000 Ci/mmol) for 30 min at 30°C. The kinase reactions were stopped by adding 20 l of 5 ϫ Laemmli SDS sample buffer (40) and boiling for 3 min. The products were separated by electrophoresis on a 10% SDS-polyacrylamide gel and exposed to x-ray film or analyzed with a PhosphorImager.
Preparation of 32 P-Labeled Substrates-Bovine cardiac myosin light chain and glycogen phosphorylase were labeled with [␥-32 P]ATP as described previously (41). pRB was phosphorylated with the cyclin D-CDK4 complex immunoprecipitated from lysates of 1 ϫ 10 7 Balb/c 3T3 cells with (C22)-R antibody. GST-RB (10 g) was incubated in kinase buffer with 40 M ATP, 100 Ci of [␥-32 P]ATP, and 100 l of resuspended (50%, v/v) CDK4-protein A-Sepharose beads. The kinase reaction was incubated for 60 min at 30°C with constant mixing. CDK4-protein A-Sepharose was removed by centrifugation (3000 ϫ g for 5 min) and the supernatant fraction applied to a Sephadex G-25 Quick Spin Protein Column (Roche Molecular Biochemicals) to remove unincorporated nucleotides. The G-25 column had been equilibrated with 50 mM Tris-HCl, pH 7.6, 50 mM NaCl, 0.1 mM EDTA, and 1 mM DTT.
Preparation of Cell Extracts and Phosphatase Assays-Cell extracts used for phosphatase assays were prepared using a modification of the method described by Cohen and colleagues (36). Balb/c 3T3 cells were incubated for 4 h with OA or CL-A, or for 15 h with TAU at the concentrations indicated in the figure legends. Cells were harvested by scraping and washed with phosphate-buffered saline. Washed cell pellets were lysed for 10 min on ice in 3 volumes (v/v) of extraction buffer (50 mM Tris-HCl, pH 7.0, 0.1 mM EDTA, 0.1 mM EGTA, 0.5% Triton X-100, 1 mM DTT, 5 g/ml leupeptin, 5 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol). The crude extract was passed through a 21-gauge needle several times to facilitate lysis. Insoluble material was removed by centrifugation (3,000 ϫ g for 5 min). The soluble fraction was passed through a Sephadex G-50 spin column (Roche Molecular Biochemicals) equilibrated with storage buffer (50 mM Tris-HCl, pH 7.0, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 5 g/ml leupeptin, 5 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, and 20% glycerol) to remove low molecular weight substances that may interfere with the protein phosphatase assays. The cell extracts were aliquoted and stored at Ϫ80°C. Each aliquot was only thawed once for phosphatase assay.
The activities of protein phosphatases 1 and 2A in cell extracts were determined by measuring the release of 32 P i from 32 P-labeled myosin light chain or phosphorylase (41). The activities of PP1 and PP2A were distinguished by their differential sensitivities to OA (42). PP1 activity was defined as the activity that was insensitive to 5 nM OA. Protein phosphatase 2A, and PP2A-like activities, were defined as the activity that was inhibited by 5 nM OA. As a control, protein phosphatase activity was also measured in the presence of 1 M OA, which completely inhibits both PP1 and PP2A. Unless indicated otherwise, the cell extracts were diluted 400-fold. Due to the extremely high affinity of OA for PP2A, a high dilution of the extracts is necessary in order to accurately measure PP1 and PP2A levels (42). All assays were carried out under conditions where the release of 32 P i was linear with time (less than 20% of the substrate consumed) and directly dependent on the amount of extract protein.
pRB-directed phosphatase activity was assayed as described previously (16). Assays were carried out in a final volume of 50 l in phosphatase assay buffer (50 mM Tris-HCl, pH 7.0, 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 0.7 mg/ml bovine serum albumin) with 1 g of 32 P-labeled GST-RB and 40 g of cell extract. The reactions were terminated by adding 20 l of 5 ϫ Laemmli SDS sample buffer and boiling for 5 min. The samples were separated by electrophoresis on 10% SDS-polyacrylamide gels. The gels were fixed, dried, and exposed to x-ray film or PhosphorImager. When OA (5 nM or 1 M) was included in the assays, it was mixed with the cell extracts for 15 min on ice prior to the addition of [ 32 P]GST-RB.

Selective Inhibition of PP2A and PP1 in Balb/c 3T3 Cells-
Okadaic acid, calyculin A, and tautomycin can selectively inhibit PP1 and PP2A in MCF-7 and other cell types (37). The utility of these inhibitors in assessing the contributions of PP1 and PP2A to the regulation of pRB phosphorylation was determined by measuring their effects on PP1 and PP2A activity in intact cells. Balb/c 3T3 cells were incubated with the inhibitors and the activities of PP1 and PP2A were determined in cell extracts. Extracts were assayed with both 32 P-labeled phosphorylase, a substrate for both PP1 and PP2A (36), and with myosin light chain, a highly selective substrate for PP2A (43). Dose-response experiments demonstrated that complete inhibition of PP2A in Balb/c 3T3 cells required 1 M OA and 50 nM CL-A (not shown). The high concentration of OA required for complete inhibition of PP2A was similar to previous reports and was probably due to the 100-fold higher concentration required for membrane permeation rates equivalent to CL-A (37).
Treatment of Balb/c 3T3 cells with OA or CL-A caused highly selective inhibition of PP2A. Logarithmically growing Balb/c 3T3 cells were incubated with 1 M OA or 50 nM CL-A for 4 h. Cells were harvested and soluble extracts assayed for phosphatase activity using phosphorylase and myosin light chain as substrates. Previously characterized methods were used to quantitate the relative activities of PP1 and PP2A in the cell extracts (36). Accordingly, phosphatase assays were carried out in the absence of added inhibitor, in the presence of 5 nM OA to specifically inhibit PP2A, or in the presence of 1 M OA to completely inhibit both PP2A and PP1. Under these conditions, the portion of total phosphatase activity attributable to PP2A corresponds to the activity inhibited by 5 nM OA while the portion attributable to PP1 corresponded to the activity resistant to 5 nM OA but inhibited by 1 M OA. Addition of 5 nM OA to extracts of untreated cells inhibited phosphorylase activity by 30% (Fig. 1A), indicating that PP1 accounted for the majority of the phosphorylase activity in Balb/c 3T3 cells. As expected, addition of 1 M OA resulted in nearly complete inhibition of phosphorylase phosphatase activity. Since PP2B/ calcineurin and PP2C are not inhibited by 1 M OA (36), this result indicated that PP1 and PP2A (or PP2A-like enzymes) were the major activities toward phosphorylase in Balb/c 3T3 cells. In contrast, addition of 5 nM OA caused complete inhibition of phosphatase activity toward myosin light chain indicating that PP2A and PP2A-like enzymes account for all of the activity toward this substrate (Fig. 1B).
Incubation of Balb/c 3T3 cells with either OA or CL-A caused a 35-40% reduction in total phosphorylase phosphatase activity measured in the absence of added inhibitor (Fig. 1A). Addition of 5 nM OA to lysates from treated cells had no significant effect indicating that the remaining phosphorylase phosphatase activity was due to PP1. The fact that the activity remaining in lysates from treated cells was nearly identical to the PP1 activity in lysates from untreated cells indicated that OA and CL-A inhibited nearly all of the PP2A activity in Balb/c 3T3 cells but had very modest effects on PP1. The mean PP1 activity in lysates from OA-or CL-A-treated cells were 7-10% lower than the PP1 activity from untreated cells but the differences were not statistically significant. The ability of OA and CL-A to cause complete inhibition of PP2A was confirmed in assays using myosin light chain. Phosphatase activity was completely abolished by treatment with either inhibitor (Fig. 1B). Our conclusions from these experiments were that treatment of Balb/c 3T3 cells with 1 M OA or 50 nM CL-A caused complete inhibition of PP2A activity and very modest, if any, effect on PP1 activity.
Unlike OA and CL-A which have higher affinity for PP2A, TAU has a 100-fold higher affinity for PP1 than PP2A (44) and inhibits PP1 preferentially in intact cells (37). Based on previous work showing that complete inhibition of PP1 required a high concentration of TAU and long incubation times, Balb/c 3T3 cells were incubated with 10 M TAU for 15 h. Phosphatase assays of lysates from treated cells showed that this treatment resulted in a dramatic decrease in total phosphorylase activity and a nearly complete loss of PP1. There was very little activity that was insensitive to 5 nM OA (Fig. 1A, right-hand bars). There was also a decrease in PP2A since the phosphorylase phosphatase activity that was sensitive to 5 nM OA was also reduced. A significant decrease in PP2A activity was also observed using myosin light chain as the substrate (Fig. 1B). The mean PP2A activity from TAU-treated cells was reduced to 53% of that from untreated cells. These experiments showed that incubation of Balb/c 3T3 cells with 10 M TAU resulted in complete inhibition of PP1 activity and a partial decrease in PP2A activity.
The cell lysates were diluted 400-fold prior to phosphatase assay. Due to the extremely high affinity of OA for PP2A, this high level of dilution was necessary in order to accurately quantitate PP1 and PP2A activity (36). It was possible that the inhibitors bound to PP1 and PP2A following treatment of intact cells dissociated following cell extraction or dilution. If this were the case, the activities from treated cells would increase relative to those from untreated cells with increasing dilution of the lysates. This possibility was tested by comparing PP1 and PP2A activities at different dilutions of the lysates with both phosphorylase and myosin light chain. The relative phos- phatase activities in extracts from treated and untreated cells did not vary significantly when the assays were done at dilutions varying from 10-to 400-fold (data not shown). These results indicated that there was very little dissociation of the inhibitors from either PP1 or PP2A during dilution and storage prior to assay. The lack of dissociation of OA and CL-A from PP1 and PP2A was consistent with the very high affinity of these compounds (44) and with data showing that release of OA and CL-A from PP1 and PP2A requires repeated precipitation with ethanol (37).
OA and CL-A, but Not TAU, Caused Dephosphorylation of pRB-The effects of OA, CL-A, and TAU on phosphorylation of the retinoblastoma protein were determined in Balb/c 3T3 cells growing asynchronously in 10% serum. Lysates of [ 35 S]methionine-labeled cells were immunoprecipitated with anti-pRB antibody and changes in pRB phosphorylation detected by altered mobility during SDS-gel electrophoresis. pRB was present primarily in the hyperphosphorylated state (upper band) in control cells or in cells treated with vehicle alone. The identities of the two major bands of 35 S-labeled protein as the phospho and dephospho forms of pRB were confirmed by immunoblotting the immunoprecipitates with monoclonal antibody G3-245 (not shown). The fastest migrating band was a nonspecifically precipitated protein since it was also detected in immunoprecipitates obtained with the control antibody (Fig. 2, upper  panel, lanes 1 and 2). Similar results were observed when pRB was immunoprecipitated with a different anti-pRB antibody (C-15).
Incubation of cells with high concentrations of OA caused dephosphorylation of pRB (Fig. 2, upper panel, lanes 4 -9). CL-A also caused dephosphorylation of pRB but the concentrations of inhibitor required were lower than those for OA. Significant dephosphorylation was seen at 50 nM CL-A, and druginduced dephosphorylation was nearly complete at 500 nM (Fig.  2, middle panel, lanes 2-8). In each case, detectable dephosphorylation of pRB coincided with the inhibitor concentrations required for complete inhibition of PP2A. CL-A appeared to cause an increase in the amount of 35 S-labeled pRB recovered in the immunoprecipitates. The significance of the increase in the amount of pRB recovered is not clear. We did not observe an increase in pRB by Western blotting of immunoprecipitates from unlabeled cells following CL-A treatment even though pRB was completely dephosphorylated (not shown). In contrast to OA and CL-A, TAU had no effect on pRB phosphorylation (Fig. 2, bottom panel). pRB was maintained in the highly phosphorylated form after incubation of Balb/c 3T3 cells with 10 M TAU, a concentration that caused complete inhibition of PP1. CL-A induced rapid dephosphorylation of pRB. In most experiments, dephosphorylation was nearly complete within 1 h (Fig. 4A). Dephosphorylation of pRB was not dependent on new mRNA or protein synthesis since preincubation with either actinomycin D or cycloheximide had no effect on CL-A-mediated dephosphorylation (Fig. 7A).
pRB Phosphatase Activity in Balb/c 3T3 Cells Was Inhibited by TAU but Not by CL-A or OA-Previous studies indicated that pRB is dephosphorylated by a PP1-type phosphatase. As shown above, treatment of Balb/c 3T3 cells with CL-A or OA caused dephosphorylation of pRB while TAU treatment maintained the hyperphosphorylated form of pRB. It seemed likely that the paradoxical dephosphorylation of pRB induced by OA and CL-A was due to mechanisms other than direct dephosphorylation of pRB, and stemmed from the selective inhibition of PP2A by these inhibitors. If this were the case, a pRB phosphatase should remain active when cells were treated with OA and CL-A but not with TAU. This possibility was tested by determining phosphatase activity in cell lysates using pRB as the substrate.
Extracts from Balb/c 3T3 cells treated with vehicle alone, 10 M TAU, 50 nM CL-A, or 1 M OA were incubated with 32 Plabeled pRB and phosphatase activity determined by SDSpolyacrylamide gel electrophoresis. Extracts from cells treated with vehicle alone rapidly dephosphorylated pRB (Fig. 3, lane  2). The pRB phosphatase activity was primarily due to PP1 since it was resistant to 5 nM OA but was blocked by 1 M OA (lanes 3 and 4). Treating Balb/c 3T3 cells with either CL-A or OA for 4 h had no effect on pRB phosphatase activity. pRB was rapidly dephosphorylated by lysates from either CL-A- (Fig. 3,  lane 8) or OA-treated (lane 11) cells. In each case the pRB phosphatase activity remaining after drug treatment corresponded to PP1 since it was resistant to 5 nM OA and sensitive to 1 M OA (lanes 9 -10, and 12, and 13). In contrast, there was a reduction in pRB phosphatase activity in lysates of cells treated with TAU (Fig. 3, lane 5). The residual pRB phosphatase activity was insensitive to 5 nM OA (lane 6) and inhibited by 1 M OA (lane 7) indicating it was PP1. Fig. 3 shows the level of pRB phosphorylation after dephosphorylation by the control lysate had gone to completion. A significant reduction in pRB phosphatase activity in lysates from TAU-treated cells was also observed at shorter incubation times where dephosphorylation  11-13) for 4 h. Cells were extracted and lysates assayed for phosphatase activity using 32 P-labeled GST-RB as substrate. Prior to the phosphatase assays, cell extracts were incubated with vehicle, 5 nM, or 1 M OA for 15 min at 4°C as indicated. The labeled GST-pRB was resolved by 8% SDS-polyacrylamide gels and 32 P visualized by autoradiography. The intensity of 32 P-labeled GST-RB substrate incubated in the absence of cell lysate is indicated in lane 1.
was linearly dependent on time. Under these conditions, the inhibition of pRB phosphatase activity in lysates from TAUtreated cells averaged around 60% (not shown). The observation that pRB phosphatase activity corresponded to PP1 was consistent with previous studies indicating that PP1 dephosphorylated pRB in intact cells (14). These results supported the conclusions that CL-A and OA induced dephosphorylation of pRB in intact Balb/c 3T3 by inhibiting PP2A but had little or no effect on the direct dephosphorylation of pRB by PP1.
Inhibition of PP2A Inactivates pRB Kinases-To investigate the mechanisms involved in dephosphorylation of pRB following selective inhibition of PP2A, we examined the effects of CL-A on pRB kinase activities. Although CL-A is less selective than OA for inhibition of PP2A, CL-A was used due to its ability to penetrate cells more rapidly than OA (37). Because of their essential roles in G 1 phosphorylation of pRB, we focused on the G 1 cyclins and on the G 1 cyclin-dependent kinases CDK4, CDK6, and CDK2. Protein kinase activities were measured in anti-CDK immune complexes using GST-RB as substrate. The CDK activities from cells growing in 10% serum were generally 6-fold higher than those from serum-starved cells (not shown). Incubation of cells growing in 10% serum with CL-A caused a time-dependent decrease in the activities of all three pRB kinases (Fig. 4, B and C). CDK4 activity was decreased by 75% 1 h after drug addition. After 3 h, CDK4 activity was reduced to 10% of that in control cells and was lower than the CDK4 activity in serum-starved cells. The decrease in CDK6 activity was more gradual and reached a minimum 3 h after addition of CL-A. CDK6 represented a minor component of the total pRB kinase activity in Balb/c 3T3 cells. Immunoprecipitated CDK6 complexes had 400-fold lower pRB kinase activity than CDK4 or CDK2.
CDK2 kinase activity was also inhibited 2 h after CL-A addition. CDK2 activity was reduced to 22% of control levels after 5 h (Fig. 4C). The rapid dephosphorylation of pRB that occurs within the first hour (Fig. 4A) correlated most closely with the decrease in CDK4 activity. The decrease in kinase activities were not due to changes in the amount of CDK proteins. Immunoblotting showed that similar amounts of CDKs were recovered in immunoprecipitates from treated and non-treated cells (Fig. 5, B and C). Direct immunoblotting of the lysates also confirmed that the levels of CDK proteins were not altered by CL-A treatment (not shown).
The levels of both cyclin D and cyclin E were reduced by CL-A treatment. A decrease in cyclin D was observed 1 h after the addition of CL-A and cyclin E levels were reduced after 3 h (Fig. 4). Cyclin D and E levels in cells treated for 3 h were lower than those present in serum-starved cells (not shown). To determine whether the decreases in cyclin D and E were sufficient to cause a decrease in CDK activity, we determined the amounts of cyclin D and E associated with CDK4 and CDK2. CDK4 and CDK2 immunoprecipitates were prepared from CL-A-treated cells at the time points indicated in Fig. 4. The immunoprecipitates were Western blotted with anti-cyclin D and anti-cyclin E antibodies. The same amounts of cyclin D and E were present in each immunoprecipitate (not shown). This result showed that even though CL-A caused a significant decrease in the levels of cyclin D and E, the levels remaining in these asynchronously growing cells were sufficient to saturate CDK4 and CDK2.
CL-A also caused a change in the mobility of cyclin E. A more slowly migrating form of cyclin E that was not apparent in control cells appeared after drug treatment (Fig. 4C, lower  panel). The altered mobility of this band was similar to the cell cycle-dependent mobility shift observed previously (45) and was probably due to enhanced phosphorylation of cyclin E (46,47).
Since Balb/c 3T3 cells express all three D-type cyclins (48), we wanted to determine which isoforms were affected by CL-A. Each of the D-type cyclins (D1, D2, and D3) was detected using isoform-specific antibodies (Fig. 5). Incubation of cells with 50 nM CL-A caused a decrease in total cyclin D levels. The levels of

FIG. 4. CL-A-induced dephosphorylation of pRB correlates with the decreases in CDK activity. A, [ 35 S]methionine-labeled
Balb/c 3T3 cells were treated with 50 nM CL-A for the times indicated at the top. pRB was immunoprecipitated with antibody AB-1, or as a control, pAb419 (Control), and analyzed as described in the legend to Fig. 2. B, at the times indicated, CDK4 and CDK6 were immunoprecipitated and the immunocomplexes assayed for kinase activity using GST-RB as substrate (CDK4 activity and CDK6 activity). The amounts of CDK4 or CDK6 protein in the immunoprecipitates were quantitated by immunoblotting (CDK4 IP Western and CDK6 IP Western). Cyclin D levels were detected by immunoblotting lysates with an anti-cyclin D antibody (Cyclin D). C, CDK2 was immunoprecipitated and assayed for kinase activity as described above (CDK2 Activity). CDK2 protein in the anti-CDK2 immunoprecipitates (CDK2 IP Western) and cyclin E (Cyclin E) levels in lysates were determined by immunoblotting. FIG. 5. CL-A treatment specifically decreases cyclin D2. Balb/c 3T3 cells were treated with or without 50 nM CL-A for 4 h. Fifty g of cell lysate protein were separated with 12% SDS-polyacrylamide gels and immunoblotted for total cyclin D, cyclin D1, cyclin D2, and cyclin D3 using specific antibodies. cyclin D2 were significantly decreased by CL-A whereas the levels of cyclins D1 and D3 were similar in control and treated cells. This specific effect of CL-A on cyclin D2 indicated that PP2A-dependent mechanisms regulating the levels of cyclin D2 were different from those regulating D1 and D3.
In order to correlate effects on cyclin levels and CDK activity with inhibition of protein phosphatases, a dose-response for CL-A effects on cyclin D-CDK4 was obtained. CL-A caused dose-dependent decreases in cyclin D levels and CDK4 kinase activity. The decrease in cyclin D levels was not observed until the CL-A concentration was 50 nM, the concentration where PP2A is completely inhibited (Fig. 6C). No further decrease in cyclin D was observed at higher concentrations. The decrease in CDK4 kinase activity was more sensitive to CL-A and occurred at concentrations that had no effect on cyclin D levels. CDK4 activity was significantly reduced at 10 nM CL-A and reached a minimum by 50 nM. The dose-response relationships for the decrease in cyclin D and CDK4 activity paralleled inhibition of PP2A and not PP1. The observations that CDK4 activity was inhibited by CL-A concentrations that had no effect on total cyclin D levels, and that there was no decrease in CDK4-associated cyclin D, indicated that the CL-A-mediated decrease in cyclin D was not involved in the inhibition of CDK4 activity.
The results of the time course and dose-response experiments show that CL-A causes decreases in cyclin D2 and cyclin E, and inhibition of CDK4/6 and CDK2 kinase activities in Balb/c 3T3 cells. Similar reductions in cyclin and CDK activity were observed in cells treated with 1 M OA (not shown). The correlation between decreases in CDK activity and pRB dephosphorylation indicates that inactivation of cyclin D-CDK4, cyclin E-CDK2, and to a lesser extent cyclin D-CDK6, are important components of phosphatase inhibitor-mediated decreases in pRB phosphorylation. The loss of cyclin D, the decrease in CDK4 activity, and dephosphorylation of pRB were maximal at concentrations of OA and CL-A that selectively inhibit PP2A. Therefore, PP2A, or a PP2A-like phosphatase, appears to be an important component of pathways that regulate the levels of cyclin D2 and cyclin E, and the activities of CDK4/6 and CDK2.
Phosphatase Inhibitors Induce the Expression of p21 Cip1 and p27 Kip1 -In addition to regulation by the availability of cyclins, cyclin D-, E-, and A-dependent kinases are negatively regulated by CDK inhibitors (3). Induction of CDK inhibitors could also contribute to phosphatase inhibitor-mediated inactivation of pRB kinases. This possibility was tested by determining the effects of CL-A and OA on expression of p21 Cip1 , p27 Kip1 , and p16 Ink4a . Balb/c 3T3 cells were treated with 50 nM CL-A and cell lysates were immunoblotted with antibodies against p21, p27, and p16. In logarithmically growing cells, p21 and p27 were present at low levels (Fig. 6A, lane 1). CL-A treatment caused a dramatic increase in both p21 and p27 (Fig. 6A, lanes  2-8). The time courses for induction of p21 and p27 were delayed relative to dephosphorylation of pRB. pRB was nearly completely dephosphorylated 1 h after CL-A addition (Fig. 4A, lane 2), a time when there was no detectable increase in p21 or p27 (Fig. 6A, lane 3). Pretreatment of cells with either actinomycin D, to block transcription, or cycloheximide, to block translation, completely inhibited the CL-A-induced increases in p21 and p27 (Fig. 7B, lanes 3 and 4). In contrast, neither actinomycin D nor cycloheximide had an effect on pRB phosphorylation (Fig. 7A, lanes 2 and 3). This result indicated that the increases in p21 and p27 were due to increased expression, but that their induction was not required for pRB dephosphorylation. In contrast to p21 and p27, the levels of immunoreactive p16 were not affected by CL-A treatment (not shown). Treating cells with 1 M OA caused increases in p21 and p27 similar to those seen with 50 nM CL-A and, also like CL-A, OA had no effect on the levels of p16 (not shown).
The induction of p21 and p27 by CL-A was dose-dependent. Increases in p21 were detectable at 10 nM CL-A and reached a maximum at 50 nM (Fig. 6B). As was the case with decreases in cyclin D and inactivation of CDK4, the concentrations of CL-A that induced expression of p21 and p27 cause selective inhibition of PP2A. The observations indicate that PP2A, or a PP2Alike phosphatase, is involved in pathways that regulate expression of p21 and p27.

DISCUSSION
The results of this study show that both PP1 and PP2A are involved in regulation of pRB phosphorylation but that they play very distinct roles in this process. Our data provide further support to the proposal that PP1 is the relevant pRB phosphatase in vivo while PP2A functions to regulate the activities of G 1 cyclin-dependent kinases. The actions of the two phosphatases were differentiated using inhibitors that selectively suppress either PP1 or PP2A. Assays of PP1 and PP2A activity in lysates from inhibitor-treated cells verified the selectivity of the inhibitors. Okadaic acid has a 4,000-fold lower K i for PP2A than for PP1 and is the most selective of the inhibitors (44). Accordingly, concentrations of OA as high as 1 M caused complete inhibition of PP2A with very little effect on PP1. Calyculin A has less selectivity (9-fold preference for PP2A). Although treating Balb/c 3T3 cells with higher concentrations of CL-A caused inhibition of PP1, we found that a concentration of 50 nM for 3-4 h completely inhibited PP2A but had little or no significant effect on PP1. Tautomycin has an 800-fold lower K i for PP1 than PP2A. Treating cells with 10 M TAU was associated with nearly complete inhibition of PP1, while PP2A was reduced to 50% of control values. Although the K i of TAU for inhibition of PP1 is 0.4 nM in vitro, the weak ability of the drug to permeate cells necessitated the use of high concentrations and long incubation times (37). Several recently identified members of the PPP family, including PP4, PP5, and PP6, are also inhibited by low concentrations of OA in vitro (49). In the absence of information regarding their sensitivities to other inhibitors, our results cannot distinguish between functions of PP2A and these novel PP2A-like phosphatases. It should be noted, however, that these novel enzymes are generally present at much lower levels than PP2A. The effects of OA, CL-A, and TAU on phosphatase activities in Balb/c 3T3 cells are very similar to their effects on PP1 and PP2A in MCF-7 cells (37).
The effects of phosphatase inhibitors on Balb/c 3T3 cells support a major role for PP1 in catalyzing dephosphorylation of pRB in intact cells. Selective inhibition of PP2A with either OA or CL-A induced dephosphorylation of pRB, indicating that a pRB phosphatase remains active under these conditions. Consistent with this observation, PP2A selective concentrations of these inhibitors had no effect on pRB phosphatase activity in lysates from treated cells. The pRB phosphatase in lysates from OA-or CL-A-treated cells could be classified as PP1 based on its insensitivity to 5 nM OA. In contrast to PP2A selective drugs, the PP1-selective inhibitor TAU resulted in maintenance of pRB in the highly phosphorylated state and inhibited pRB phosphatase activity. Although TAU was not as selective for PP1 as OA and CL-A were for PP2A in intact cells, the ability of TAU treatment to reduce pRB phosphatase activity was consistent with its classification as PP1. The fact that concentrations of OA and CL-A that completely inhibited PP2A had no effect on pRB phosphatase made it unlikely that PP2A played a significant role in the direct dephosphorylation of pRB.
PP1 has been identified as a major activity that dephosphorylates pRB during mitosis (16,50) and plays an important role in the G 1 /S transition. The PP1 catalytic subunit interacts directly with pRB (19). Biochemical fractionation of mitotic extracts has shown that an active form of pRB phosphatase is a high molecular weight complex composed of the PP1 catalytic subunit and a 110-kDa interacting protein (18). This oligomeric form of PP1 may be the same as a nuclear PP1 complex identified previously that contains the catalytic subunit complexed to an inhibitory protein termed R111 (51). The 110-kDa regulatory subunit of the pRB phosphatase may be identical to two homologous PP1 interacting proteins termed PNUTS (52) and p99 (53). Support for a role of PP1 in regulating the G 1 /S transition stems from experiments using a constitutively active mutant of PP1. Introduction of this mutant PP1 caused an arrest in late G 1 that was dependent on the maintenance of pRB in the active, dephosphorylated state (20). In addition, there is a peak of PP1 activity associated with early G 1 (17), which could serve to maintain pRB in the dephosphorylated state.
Our results indicate that PP2A also plays an important role in the G 1 /S transition. Selective inhibition of PP2A, or PP2Alike phosphatases, causes decreases in cyclin D2 and cyclin E, induction of cyclin-dependent kinase inhibitors, and inhibition of CDK4, CDK6, and CDK2 protein kinase activity. These data indicate that PP2A activity is required for the normal regulation of G 1 CDKs. Our results are consistent with previous studies in NIH 3T3 cells showing that okadaic acid causes a G 1 arrest that correlates with decreased pRB phosphorylation, decreased CDK activity, and a block in cyclin A, CDC2, and CDK2 expression (31). Consistent with a role in transcriptional regulation of CDC2, transient expression of the PP2A catalytic subunit can positively regulate CDC2 gene expression (54). However, the effects of OA may depend on the cell type as OA caused induction rather than repression of cyclin A, cyclin B, and CDC2 in transformed cells (55). We show here that PP2A selective concentrations of OA or CL-A cause transcription-dependent up-regulation of p21 and p27. Induction of p21 by OA has been observed previously but it was not clear whether this effect was due to inhibition of PP1 or PP2A (56,57). The effects of inhibiting PP2A on G 1 CDKs support the hypothesis that PP2A is a positive regulator of the G 1 /S transition. The positive role for PP2A at G 1 /S contrasts with its negative role in the G 2 /M transition. Both genetic (25,26) and biochemical (27,28,58) evidence have shown that PP2A negatively regulates activation of the cyclin B-CDC2 kinase. The contrasting actions of PP2A in G 1 and G 2 /M are likely to be due to the actions of distinct PP2A holoenzymes that are targeted to G 1 versus G 2 /M substrates (22,59).
Although selective inhibition of PP2A caused the induction of multiple mechanisms that suppress CDK activity, inhibition of cyclin D-CDK4 appeared to be a major component of the reduced pRB phosphorylation. CDK6 was only present at very low levels in Balb/c 3T3 cells and was unlikely to account for a significant portion of the pRB kinase activity. Suppression of cyclin E-CDK2 is also likely to be an important mechanism in the dephosphorylation of pRB caused by PP2A-selective phosphatase inhibitors. However, this mechanism may be secondary to CDK4 inactivation since cyclin E expression is itself responsive to E2F released following pRB phosphorylation (3). In addition to reduction in cyclin D and induction of CDK inhibitors, the reduction in CDK4 kinase activity induced by inhibition of PP2A may involve additional mechanisms. Induction of CDK inhibitors was not necessary since CDK4 was inhibited to the same extent in the presence of actinomycin D or cycloheximide, which completely blocked induction of p21 and p27. The rapid reduction of CDK4 activity was sensitive to concentrations of CL-A (10 nM) that had no effect on the level of cyclin D. In addition, the amount of cyclin D present in immunoprecipitated CDK4 complexes was the same in control cells and cells treated with phosphatase inhibitors. This latter result indicated that even though cyclin D levels decreased, the levels present after drug treatment were still sufficient to form complexes with CDK4. Additional mechanisms are therefore likely to be involved in PP2A-mediated down-regulation of CDK4 activity. PP2A could be necessary for the activity of CDK activating kinase which phosphorylates Thr 172 in the activation loop of CDK4 (6). PP2A could also directly dephosphorylate the activating Thr 172 phosphorylation site. Other possibilities include changes in inhibitory tyrosine phosphorylation of CDK4 (9) or tyrosine dephosphorylation catalyzed by the Cdc25A dual-specificity phosphatase (10).