Inhibitory Phosphorylation of PP1α Catalytic Subunit during the G1/S Transition*

We have shown earlier that, in cells expressing the retinoblastoma protein (pRB), a protein phosphatase (PP) 1α mutant (T320A) resistant to inhibitory phosphorylation by cyclin-dependent kinases (Cdks) causes G1arrest. In this study, we examined the cell cycle-dependent phosphorylation of PP1α in vivo using three different antibodies. PP1α was phosphorylated at Thr-320 during M-phase and again in late G1- through early S-phase. Inhibition of Cdk2 led to a small increase in PP1 activity and also prevented PP1α phosphorylation. In vitro, PP1α was a substrate for Cdk2 but not Cdk4. In pRB-deficient cells, phosphorylation of PP1α occurred in M-phase but not at G1/S. G1/S phosphorylation was at least partially restored after reintroduction of pRB into these cells. Consistent with this result, PP1α phosphorylated at Thr-320 co-precipitated with pRB during G1/S but was found in extracts immunodepleted of pRB in M-phase. In conjunction with earlier studies, these results indicate that PP1α may control pRB function throughout the cell cycle. In addition, our new results suggest that different subpopulations of PP1α regulate the G1/S and G2/M transitions and that PP1α complexed to pRB requires inhibitory phosphorylation by G1-specific Cdks in order to prevent untimely reactivation of pRB and permit transition from G1- to S-phase and/or complete S-phase.

In mammals, three genes encode four isozymes of serine/ threonine-specific protein phosphatase 1 designated PP1␣, PP1␥1, PP1␥2, and PP1␦. Except for PP1␥2, which is found only in testes, these isoforms are expressed in all tissues and cellular compartments and are playing important roles in many different aspects of cellular activities. Although the free catalytic subunit of PP1 1 can dephosphorylate multiple proteins in vitro, it is thought that due to its complex regulation PP1 is nonetheless capable of executing specific reactions upon receiving appropriate signals. This regulation typically involves interaction with inhibitory proteins or so-called targeting subunits that direct PP1 toward distinct cellular locales or even substrates (reviewed in Refs. 1 and 2). Both yeast twohybrid screens (3,4) and affinity chromatography on microcystin-Sepharose (5,6) have recently led to the discovery of a number of proteins that appear to specifically bind PP1. Thus, PP1 is more adequately portrayed as an enzyme system rather than a single enzyme. PP1 activity is crucial for cell cycle regulation (reviewed in Refs. 7 and 8). Genetic studies have shown that PP1 is required for progression through or exit from mitosis in Aspergillus nidulans (9) and fission yeast (10,11), as well as in Drosophila (12). This was confirmed biochemically for mammalian cells in that microinjection of antibodies to PP1 causes mitotic arrest (13). However, essential mitotic substrates for PP1 have not unambiguously been identified. PP1 undergoes inhibitory phosphorylation by cyclin-dependent kinases (Cdks), both in vitro (14,15) and in vivo during M-phase in Schizosaccharomyces pombe (15) and human cells (16,17). Studies in S. pombe suggest that the phosphorylation of PP1 may be necessary to permit entry into M-phase (15). We have mapped the site of phosphorylation to Thr-320 in PP1␣ (14). PP1 activity also oscillates during the mammalian cell cycle. Cytoplasmic activity is maximal in quiescent cells and reduced up to 4-fold in the remaining phases of the cell cycle, whereas nuclear or chromatin-associated PP1 shows two similarly sized peaks of activity during G 0 /G 1 and mitosis (14).
Another important role for PP1 in the cell cycle of mammalian cells may be the dephosphorylation of the retinoblastoma protein, pRB. Many pathways that regulate G 1 progression and the transition to S-phase converge on pRB (reviewed in Refs. 18 -20). Ultimately, to permit passage through G 1 /S, pRB has to be inactivated by Cdk phosphorylation in late G 1 (21). A variety of approaches has shown that PP1 interacts with and dephosphorylates pRB (3,(22)(23)(24). These findings raised the question whether PP1␣ controls the G 1 /S transition and whether this putative function is linked to phosphorylation of PP1␣. Following introduction of recombinant PP1␣ into synchronized cells by electrotransfer, we found that, unlike wildtype PP1␣, a constitutively active mutant of PP1␣ that is resistant to Cdk phosphorylation in Thr-320 prevents cells from entering S-phase, provided they express functional pRB (25). This finding suggested that phosphorylation of PP1␣ somewhere in G 1 might be required to allow phosphorylation of pRB and initiation of S-phase. However, direct evidence for such a reaction was still missing.
The above findings prompted us to re-investigate cell cycledependent phosphorylation of PP1␣ with emphasis on the G 1phase. In this study, using novel antibodies that are capable of reacting with Cdk-phosphorylated PP1␣, we demonstrate that PP1␣ is indeed phosphorylated in vivo before and after the G 1 /S transition. As we have seen with PP1␣-mediated G 1 arrest (25), G 1 /S phosphorylation of PP1␣ was not detectable in the absence of pRB. Consistent with this finding, most if not all of the G 1 /S phosphorylation in pRB-expressing cells affected PP1␣ that was physically associated with pRB. The implications of these findings will be discussed.

EXPERIMENTAL PROCEDURES
Materials-Olomoucine was from LC Laboratories, Woburn, MA, and histone H1 was from Sigma. Saos2 cells reconstituted with wildtype pRB were generated and provided by Yuen-Kai Fung (Childrens Hospital Los Angeles) (26). Sources of other cells and all other chemicals were given previously (25). The recombinant catalytic subunit of PP1␣ was expressed in Escherichia coli and purified as described previously (27). Recombinant baculoviruses encoding human Cdks and cyclins were kindly provided by David O. Morgan, University of California San Francisco (Cdk2 and cyclin A), and Charles J. Sherr, St. Judes Childrens Research Hospital, Memphis, TN (Cdk4, cyclin D1, D2, and E). Phosphorylase b and phosphorylase kinase were kindly provided by Balwant S. Khatra, California State University Long Beach, CA. Three different antipeptide antibodies specifically recognizing the C terminus of PP1␣ were used in this study. The first of these (Ab1), against residues 316 -330 (28), only immunoprecipitates the dephosphorylated form of PP1␣ and weakly recognizes Thr-320-phosphorylated PP1␣ after denaturation in SDS-PAGE sample buffer (24). The second antibody (Ab2) was against residues 294 -309 (29) and reacts with PP1␣ whether phosphorylated or not (17). Antibodies specific for PP1␣ phosphorylated at Thr-320 (Ab3) were obtained from Angus C. Nairn (Rockefeller University, New York). Antibodies to pRB (sc-102) and non-neutralizing antibodies to Cdk2 (sc-163G) were purchased from Santa Cruz Biotechnology, Santa Cruz, CA.
In Vitro Phosphorylation of PP1␣-To phosphorylate PP1␣, we used cytosolic lysates prepared from synchronized MG63 cells or Cdk-cyclin complexes that were expressed and reconstituted from recombinant baculoviruses (see above) in Sf9 insect cells as described (30). Serial dilutions of insect cell lysates were then used to determine their ability to label pRB with 32 P. Aliquots corresponding to equal pRB phosphorylation were then used in the subsequent PP1␣ phosphorylation reactions. These were carried out under conditions described earlier (14) in the presence of [␥-32 P]ATP (specific radioactivity Ն2,000 cpm/pmol).
Immunoprecipitation from Cells or in Vitro Phosphorylation Reactions-Cells were washed in phosphate-buffered saline and lysed with RIPA buffer (25). Each sample was adjusted to 1 mg/ml total protein by adding bovine serum albumin. Cell lysates were pre-cleared by a 30min incubation on ice with 50 l of protein A-Sepharose suspension (equal volumes of beads and buffer) which was removed by centrifugation. The supernatants were then mixed with 50 l of protein A-Sepharose suspension and the appropriate antibody. The amounts of antibody used were as follows: 8 g of Ab2 or 4 g of Ab3 (for PP1␣), 5 g of sc-163G (for Cdk2), and 10 g of sc-102 (for immunoprecipitation or immunodepletion of pRB). The mixtures were rotated at 4°C for 2 h and processed further as described for pRB (25). To determine the activity status of cyclin E-and cyclin A-dependent Cdk2, immunoprecipitated Cdk2 was further incubated with histone H1 and 0.1 mM [␥-32 P]ATP (specific radioactivity ϳ10,000 cpm/pmol) as described previously (31). 32 P-Labeled histone, PP1␣, pRB immune complexes, or pRB-immunodepleted supernatants were separated on 12% SDS-polyacrylamide gels.
Other Methods-Cell synchronization in various phases of the cell cycle (14,25), metabolic labeling of cells with 32 P i , flow cytometry, lysis of cells for immunoprecipitation or enzyme assays, electrophoresis, and assays for protein phosphatase 1 activity were performed by following standard procedures as described previously (25). Proteins to be analyzed by Western blotting were transferred to Immobilon® membranes, incubated with 5% non-fat dry milk overnight, and incubated with primary antibodies for 2 h at room temperature. Protein bands were visualized with the ECL system according to Amersham Pharmacia Biotech.

RESULTS
Cdks Phosphorylate PP1␣ Twice during the Cell Cycle-In order to test whether the observed cell cycle-dependent activity changes in PP1 (14) are due to regulated expression, we sub-jected cytoplasmic and nuclear lysates from synchronized MG63 cells to Western blotting with two antibodies specific for PP1␣. This isoform is of special interest as it physically associates with pRB in a yeast two-hybrid screen and in co-immunoprecipitation experiments from G 0 -to S-phase (3). By using Ab2 for detection, the amounts of PP1␣ protein were apparently constant during all phases of the cell cycle ( Fig. 1); this antibody recognizes residues 294 -309 of PP1␣ (29) and therefore reacts with both unphosphorylated and phosphorylated forms of PP1␣ (17). This result indicated that PP1 activity during the cell cycle is more likely to be modified by (i) interaction with regulatory subunits and/or (ii) phosphorylation. The latter possibility appeared particularly intriguing as we have demonstrated that Cdks inhibit PP1␣ through phosphorylation at Thr-320 in vitro (14). To investigate the in vivo phosphorylation of PP1␣ was initially not possible as the PP1␣specific antibody we have generated (28) is unable to immunoprecipitate CDK-phosphorylated PP1␣ (24). This behavior of Ab1 is most likely due to the fact that it was raised against a peptide derived from a C-terminal sequence harboring Thr-320 (28); upon phosphorylation, the C terminus folds back to mask the catalytic center (32,33). However, two recently described antibodies to PP1␣ put us in a position to address the question whether phosphorylation of PP1␣ occurs in vivo in a cell cycledependent manner. One of these antibodies (Ab2) was described above, and the second one (Ab3) was raised against a peptide comprising residues 316 -323 that were phosphorylated in Thr-320 (16). Both of these antibodies could immunoprecipitate PP1␣ that was phosphorylated by Cdks in vitro (data not shown). Initially, we performed two experiments to examine the cell cycle-dependent phosphorylation of PP1␣. First, recombinant PP1␣ was mixed with cellular extracts derived from MG63 cells synchronized in different phases of the cell cycle and then immunoprecipitated with Ab2. As shown in Fig. 2, only mitotic extracts were able to phosphorylate PP1␣. Next, to examine the in vivo phosphorylation of PP1␣, MG63 cells were synchronized in different stages of the cell cycle and labeled with 32 P i , followed by immunoprecipitation with Ab3. Thr-320 was modified during two periods as follows: the first phosphorylation lasted from late G 1 through early S-phase, and the second throughout M-phase (Fig. 3). Flow cytometry of identically treated sister cultures revealed that the cells were synchronized in the phases indicated (data not shown). Densitometric scanning of the individual bands showed that the level of phosphorylation during mitosis was approximately 2-fold higher than at G 1 /S. Similar results were obtained with Ab2 suggesting that Thr-320 is the only site that undergoes cell cycle-dependent phosphorylation. Based on these results and the previous characterization of Ab2 (16), in all subsequent experiments involving Ab2, 32 P labeling was omitted. By using known amounts of phosphorylated and non-phosphorylated PP1␣ for comparison, we estimated that in late G 1 -early S-FIG. 1. Expression levels of PP1␣ during the cell cycle. Cytosolic (C) and nuclear (N) lysates were prepared from synchronized MG63 cells (14), separated on a 12% SDS-polyacrylamide gel, transferred to Immobilon® membrane, and then probed with Ab2 (see "Experimental Procedures") to determine the expression level of PP1␣. A, asynchronous cells; S, S-phase; M, M-phase. phase perhaps 10 -15% and in mitosis 25-30% of the cellular PP1␣ were phosphorylated.
To address the question whether phosphorylation affects the activity of PP1 in vivo, we exposed MG63 cells synchronized in late G 1 or at the G 1 /S boundary to olomoucine, an inhibitor of Cdk1 and Cdk2 (but not Cdk4) that causes cell cycle arrest (34,35). This experiment revealed that inhibition of Cdks is associated with a small increase of PP1 activity (up to ϳ30% higher than untreated control cells) and a decrease of phosphorylation in Thr-320 (Fig. 4, A and B). The time course of activation correlated well with the phosphorylation pattern presented in Fig. 3.
Phosphorylation of PP1␣ at Thr-320 was first established with Cdk1/cyclin A in vitro (14). Compared with this enzyme, PP1␣ is a very poor substrate for the mitotic Cdk1/cyclin B (24). To investigate further whether other Cdks involved in the G 1 /S transition can regulate PP1␣, we incubated recombinant PP1␣ (27) with Cdk-cyclin complexes that were expressed and reconstituted in the baculovirus system. All reactions were carried out under standard assay conditions (see "Experimental Procedures"). Prior to PP1␣ phosphorylation, the amount of Cdk/ cyclin to be used for each experiment was normalized for their ability to phosphorylate recombinant pRB. This experiment showed that PP1␣ was readily phosphorylated by Cdk2/cyclin A and Cdk2/cyclin E but apparently not by Cdk4/cyclin D1 or D2 (Fig. 5).
Phosphorylation of PP1␣ at G 1 /S, but Not at G 2 /M, Depends on pRB-Earlier we had reported that a phosphorylation-resistant mutant of PP1␣ causes G 1 arrest in pRB-positive MG63 cells, but not in pRB-negative Saos2 cells (25), suggesting that the PP1␣-mediated G 1 arrest depends on functional pRB (25).
Thus, it was important to examine whether the inhibitory phosphorylation of PP1␣ was equally dependent on pRB. We therefore compared the cell cycle-dependent phosphorylation pattern of Thr-320 in MG63 cells with that in Saos2 cells as well as Saos2 cells that had been stably transfected with wildtype pRB. This experiment was conducted in a manner analogous to that described in Fig. 3. With regard to MG63 cells, this experiment confirmed that Thr-320 was increasingly phosphorylated during the G 1 /S transition and M-phase when compared with cells in G 0 /G 1 . However, in cells lacking pRB, only mitotic phosphorylation could be detected, whereas stable re-  3. Cell cycle-dependent phosphorylation of PP1␣ at Thr-320. MG63 cells were synchronized at various stages of the cell cycle corresponding to G 0 /G 1 , late G 1 , G 1 /S boundary, and G 2 /M as described (14,25) and released from drug-induced cell cycle arrest for up to 7.5 h (the last 2.5 h in the presence of 32 P i ) and lysed with RIPA buffer. PP1␣ phosphorylated at Thr-320 was immunoprecipitated with Ab3, separated by 12% SDS-PAGE, and visualized by autoradiography. The amount of cellular protein used for immunoprecipitation and loaded per lane was equal within each group, 0.6 mg for G 0 , late G 1 , and G 1 /S and 0.3 mg for G 2 /M. The position of PP1␣ is given by the arrow.

FIG. 4. Effects of olomoucine on PP1 during the G 1 /S transition.
MG63 cells were synchronized in late G 1 or at the G 1 /S transition and then released for up to 6 h. 1 h before the scheduled time of harvest, the cells were exposed to 200 M olomoucine. The cells were washed repeatedly and lysed under non-denaturing conditions (25). A, PP1 activity in untreated (white columns) and olomoucine-treated cells (gray columns) was determined as described (14). Each column represents the mean of three independent experiments Ϯ S.E. B, the phosphorylation state of PP1␣ at Thr-320 in response to olomoucine was estimated by Western blotting with Ab3, with each lane containing 10 5 cells.

FIG. 5. In vitro phosphorylation of PP1␣ by different Cdkcyclin complexes. Recombinant PP1␣s expressed in E. coli (27) and
Cdk-cyclin complexes reconstituted from baculovirus expression were incubated in the presence of 0.2 mM [␥-32 P]ATP (approximately 2,000 cpm/pmol) for 30 min at 30°C. As in Fig. 2, PP1␣ was immunoprecipitated from the reaction mixture, separated by 12% SDS-PAGE, and analyzed by autoradiography. The arrow indicates the position of PP1␣.
introduction of pRB into Saos2 cells restored phosphorylation of PP1␣ in S-phase (Fig. 6A). Considering that both cyclin E (36) and cyclin A (37) in concert with Cdk2 are required for the G 1 /S transition and that Cdk2 is a prime candidate for phosphorylating PP1␣ (compare Figs. 4 and 5), we determined the histone H1 kinase activity of cyclin E-and cyclin A-associated Cdk2. This was quite similar in all three cell lines (Fig. 6B), suggesting that the lack of phosphorylation at G 1 /S in the absence of pRB was not due to erratic activation of Cdk2.
To see whether phosphorylation was affecting or, perhaps, limited to PP1␣ that was associated with pRB, we immunoprecipitated pRB from cells in late G 1 -through M-phase, and we examined the distribution of PP1␣ and PP1␣ phosphorylated at Thr-320 in both pRB immunocomplexes and pRB-immunodepleted extracts. Although it is known that PP1␣ is associated with pRB from G 0 -until S-phase (3), our experiment revealed that, during the G 1 /S transition, phosphorylation at Thr-320 occurred when PP1␣ was associated with pRB ( Fig. 7). At the arrest point of nocodazole, the onset of mitosis, neither PP1␣ nor phospho-PP1␣ was found in pRB immunocomplexes. As expected, extracts depleted of pRB revealed similar amounts of PP1␣ from late G 1 -through M-phase; however, PP1␣ phosphorylated at Thr-320 could only be detected in M-phase. This phosphorylation (estimated to affect approximately 25-30% of PP1␣, see above) may explain the small but discernible decrease in signal obtained for PP1␣ in M-phase (lane 5, center right panel). As the amount of cellular protein used for the immunodepletion experiments (50 g or approximately 10 5 cells) was identical to that used in the straight Western blot experiments shown in Fig. 6A, these data suggest that most if not all of PP1␣ that was associated with pRB from late G 1 -until S-phase was undergoing inhibitory phosphorylation and, vice versa, that inhibitory phosphorylation of PP1␣ predominantly occurs at G 1 /S, when it is in a complex with pRB, and at the beginning of M-phase, when it is not in a complex with pRB. DISCUSSION G 1 /S Phosphorylation-In this paper, we demonstrate that PP1␣ catalytic subunit is phosphorylated at Thr-320, a Cdk consensus site, as cells approach and traverse S-phase. We had proposed such a mechanism following our previous study showing that a phosphorylation-resistant, constitutively active PP1␣ causes G 1 arrest (25). This PP1␣-mediated G 1 arrest is dependent on the presence of functional pRB (25). Two lines of evidence suggested that phosphorylation of Thr-320 during the G 1 /S transition depends on functional pRB as well. First, phosphorylation of this site could not be detected in pRB-deficient cells but was at least partially restored in the same cell line that was stably transfected with pRB. Second, G 1 /S phosphorylation of Thr-320 was apparent in PP1␣ that was in a complex with pRB but was virtually absent in cell extracts immunodepleted of pRB (see Figs. 6 and 7). The co-immunoprecipitation/immunodepletion experiments depicted in Fig. 7 may also explain the initially puzzling failure of G 1 /S cell extracts to phosphorylate the free catalytic subunit of PP1␣ (see Fig. 2); Cdks active during the G 1 /S transition might preferably or exclusively recognize PP1␣ when associated with pRB. Considering our findings together with earlier studies from this and other laboratories, our observations suggest the following interpretations: (i) in G 0 until mid-G 1 , PP1␣ is associated with pRB (3) and functions to maintain pRB in active form (24); and (ii) in late G 1 through S-phase, PP1␣ remains associated with pRB and, like pRB itself, undergoes inhibitory phosphorylation (this study), which may be required for cells to initiate and maintain S-phase (25). Thus, the action of pRB-phosphorylating kinases on pRB alone may not be sufficient to compromise the FIG. 6. Phosphorylation of PP1␣ at Thr-320 in pRB ؉ and pRB ؊ cells. MG63 cells (pRB ϩ ), Saos2 cells (pRB Ϫ ), or Saos2 cells reconstituted with pRB (26) were synchronized as in Fig. 2. The cells were then lysed with RIPA buffer (25) and subjected to different analyses. A, the phosphorylation state at Thr-320 was analyzed by Western blotting with Ab3. Each lane contained 10 5 cells. Note that a longer observation period was chosen for Saos2 cells released from G 1 /S arrest, as they take much longer to traverse S-phase than MG63 cells (25). B, the histone H1 kinase activity of Cdk2 was determined as described (31). Shown here are the histone H1 bands stained with Coomassie Blue R (lefthand panels) or visualized by autoradiography (right-hand panels).

FIG. 7.
Thr-320 phosphorylation of PP1␣ in association with pRB. Synchronized MG63 cells were lysed in RIPA buffer, and pRB was immunoprecipitated (IP) from 2 ϫ 10 6 cells with 10 l of sc-102 as described (25). To remove pRB, lysates corresponding to 4 ϫ 10 5 cells were mixed with the same amount of sc-102. The resulting immunoprecipitates (left-hand panels) or immunodepleted extracts (right-hand panels) were then separated by 12% SDS-PAGE, transferred to Immo-bilon® membranes, and probed with sc-102, Ab1, or Ab3. The membranes from immunoprecipitations were exposed to film for 30 s and those from immunodepletions for 2 min. WB, Western blot. growth-suppressing function of pRB. The persistent PP1␣-pRB association might have a dual purpose. (i) Before PP1 kinases inactivate PP1␣, it may be crucial for maintaining cells in G 1 -phase. (ii) After PP1 kinases have inactivated PP1␣, it may enable cells that are now committed to initiate or complete S-phase to re-activate PP1␣ which in turn could re-activate pRB. This would allow cells to quickly respond to signals (such as DNA damage) that require a temporary cell cycle arrest.
Evidence from other laboratories has suggested that pRB has anti-apoptotic function (38), and indeed, we have observed earlier that, in pRB-positive cells, PP1␣T320A induces G 1 arrest and then cell death (25) by apoptosis, 2 whereas in pRBnegative cells, both wild-type PP1␣ and PP1␣T320A can trigger cell death without prior G 1 arrest (25). Thus, as has been proposed by others (39), it may be the phosphorylated form of pRB rather than pRB per se that protects cells from apoptosis. The lack of Thr-320 phosphorylation at G 1 /S in the absence of pRB reported here suggests that this phosphorylation event is primarily related to the cell cycle function of PP1␣. Our study does not preclude the possibility that PP1␣ also controls the phosphorylation state of pRB indirectly, e.g. by acting on the Cdk inhibitors p21 cip1 or p27 kip1 . Both of these proteins have recently been shown to be subject to phosphorylation, which possibly triggers their proteolytic degradation (40 -42). Dephosphorylation of these two proteins would be predicted to have a stabilizing effect, thus perpetuating Cdk inhibition and preventing pRB phosphorylation.
M-phase Phosphorylation-Cell cycle-dependent phosphorylation of PP1 has been demonstrated before in S. pombe, where it occurs only at the onset of mitosis (15). Subsequently, mitotic phosphorylation of human PP1 has also been established (16,17), although the reasons for this are less clear. It may be required to support the phosphorylation of multiple proteins that usually accompanies mitosis (15,16). In the present paper, we confirmed that Thr-320 is phosphorylated during mitosis, regardless of whether pRB is present. Phosphorylation occurred in pRB-positive and pRB-negative cells, and in pRB-free cell lysates (see Figs. 6 and 7). This indicates that the phosphorylation of PP1␣ permitting S-phase or M-phase entry is likely to involve two different subpopulations of the enzyme. As PP1 starts to dephosphorylate pRB in mid-to-late mitosis (22), at a time when PP1␣ is phosphorylated as well (see Fig. 3), even in cells lacking pRB (see Fig. 6), it is likely that the pRB-directed activity of PP1␣ and the phosphorylated PP1␣ represent distinct subpopulations of the enzyme or, alternatively, that mitotic dephosphorylation of pRB is catalyzed by another PP1 isoform. The latter explanation would be favored by the results of our co-immunoprecipitation and immunodepletion experiments (compare Fig. 7). At the onset of Mphase, neither PP1␣ nor phosphorylated PP1␣ could be detected in a complex with pRB.
Stoichiometry of Phosphorylation-Both G 1 /S and mitotic phosphorylation events affected only a minor portion of the total PP1␣ present. This is in agreement with our previous study showing that, following electrotransfer of recombinant PP1␣ into synchronized cells, the activity of wild-type PP1␣ decreases only slightly when compared with constitutively active PP1␣ (25). Furthermore, the degree of phosphorylation seen at G 1 /S, as well as the PP1 activity increase upon inhibition of Cdks acting in late G 1 (see Fig. 4A), is smaller than the drop in PP1 activity we reported earlier (14). This could be explained by several possibilities as follows: (i) the activity measured earlier did not distinguish between PP1 isoforms, and (ii) PP1 is down-regulated not only by phosphorylation of the catalytic subunit but also by interaction with one or more inhibitory proteins. In particular, inhibitor-2 oscillates during the cell cycle (43) and translocates to the nucleus at the G 1 /S transition, thereby providing another means to inhibit nuclear PP1 (44). Our results are in agreement with the current model that recognizes PP1 as an enzyme that performs multiple tasks in cells (see the Introduction); this means that only a subpopulation of PP1 can perform functions related to the cell cycle (8). If this model is true, then small and even statistically insignificant changes of the overall activity can trigger significant effects on a particular process. Therefore, the low in vivo stoichiometry notwithstanding, phosphorylation of Thr-320 is expected to be physiologically relevant.
Do pRB Kinases Also Phosphorylate PP1␣?-To inactivate pRB requires the activity of both cyclin D-and then cyclin E-Cdk complexes (45)(46)(47); furthermore, continued phosphorylation of pRB by Cdk2/cyclin A appears to be required during S-phase (48). Exposure of cells to olomoucine before and at the G 1 /S transition slightly increased PP1 activity and prevented Thr-320 phosphorylation (see Fig. 4), indicating that PP1␣ may be inhibited by Cdk2-mediated phosphorylation in vivo during the G 1 /S transition. Cdk2 is known to associate with cyclin E and cyclin A (18 -21). Both Cdk2/cyclin E and Cdk2/cyclin A were able to phosphorylate PP1␣ in vitro (see Fig. 5), in addition to Cdk1/cyclin A (14). The only physiologically relevant substrate for cyclin D-dependent kinases that has been identified is pRB (49). Our observation that Cdk4/cyclin D apparently does not phosphorylate PP1␣ in vitro is consistent with this model. Cdk2/cyclin E is also required to fully inactivate pRB (47); however, this enzyme is required for the G 1 /S transition even in the absence of pRB (36) implying that Cdk2/cyclin E must be involved in the phosphorylation of other substrates. Although PP1␣ may be phosphorylated by Cdk2/cyclin E in vivo, it cannot be the only critical one, as the phosphorylation of PP1␣ was only detectable in pRB-expressing cells. Nonetheless, with the possible exception of Cdk4/cyclin D, the kinases phosphorylating pRB may be the same that phosphorylate PP1␣.
In conclusion, this work demonstrates for the first time that PP1␣ is down-regulated by Cdk phosphorylation shortly before and during the G 1 /S transition. In conjunction with our previous study, these data suggest that a subpopulation of PP1␣ plays a pivotal role in inhibiting the G 1 /S transition via controlling the pRB activity status. As the pRB pathway is malfunctioning in virtually every human cancer studied (50,51), further investigations will be important to characterize the interaction between the different PP1 isozymes, putative regulatory subunits of PP1, and pRB or upstream regulators of pRB.