|Home | About | Journals | Submit | Contact Us | Français|
Loss of Cdc2 activity following Cyclin B degradation is necessary, but not sufficient, for mitotic exit. Proteins phosphorylated by Cdc2 and downstream mitotic kinases must also be dephosphorylated. We report here that protein phosphatase-1 (PP1) is the major catalyst of mitotic phosphoprotein dephosphorylation. Suppression of PP1 during early mitosis is maintained through the dual inhibition of PP1 by Cdc2 phosphorylation and the binding of Inhibitor-1 (I1), which is facilitated by PKA-mediated I1 phosphorylation. As Cdc2 levels drop following Cyclin B degradation, autodephosphorylation of PP1 at the site of Cdc2 phosphorylation (T320) allows partial PP1 activation. This promotes PP1-regulated dephosphorylation of I1 at its activating site (T35), dissociation of the I1-PP1 complex, and full PP1 activation to promote mitotic exit. Thus, Cdc2 both phosphorylates multiple mitotic substrates and inhibits their PP1-mediated dephosphorylation.
Exit from mitosis requires destruction of Cyclin B protein, Cdc2 kinase inactivation, and dephosphorylation of mitotic phosphoproteins. The identity and regulation of the vertebrate phosphatase(s) responsible for this dephosphorylation have not been established1.
In yeast, mitotic phosphoprotein dephosphorylation is catalyzed by the phosphatase Cdc14. Prior to M phase exit, Cdc14 is sequestered within the nucleolus through association with its inhibitor, Net1. Cdc2-mediated Net1 phosphorylation promotes Cdc14 release from the nucleolus to allow dephosphorylation of mitotic phosphoproteins; nucleolar exclusion is then maintained by the mitotic exit network (MEN)2–4. Although Cdc14 homologs exist in vertebrates, they do not seem to play a similar role in these cells, though yeast and vertebrate Cdc14 proteins share a conserved role in cytokinesis.5
In Aspergillus, Drosophila, and S Pombe, genetic analyses have suggested a role for PP1 in controlling mitotic exit6–9. In attempting to purify a phosphatase from interphase Xenopus egg extracts able to dephosphorylate mitotic phosphoproteins, Che et al concluded that the predominant phosphatase was neither PP1 nor protein phosphatase 2A (PP2A), but they noted that this activity might differ from the phosphatase acting at mitotic exit10. Skoufias et al reported a requirement for okadaic acid (OA)-inhibitable phosphatase(s) in dephosphorylating Cdc2 substrates in human cells11. Finally, calcineurin is required for the specialized exit from M phase triggered by Ca2+ in the cytostatic factor (CSF)-arrested egg12, 13. It was also noted that an additional unidentified phosphatase activity was required for full dephosphorylation of M phase phosphoproteins and release from CSF arrest12, 13.
In studying the early embryonic mitotic cycles in Xenopus, we have found that PP1 dephosphorylates mitotic phosphoproteins required for mitotic exit. PP1 is not similarly required for exit from CSF arrest. Furthermore, we have identified a regulatory loop that controls PP1 to promote the timely dephosphorylation of mitotic substrates. As reported previously, PP1 can be inhibited by Cdc2-mediated phosphorylation at T32014, 15. We show that PP1 auto-dephosphorylates T320, but that this activity is inhibited during M phase by the association of PP1 with its inhibitor, I1. Activation of I1 through PKA-mediated phosphorylation of T35 is enhanced in M phase, both because PKA activity is elevated and the rate of T35 dephosphorylation is low. When Cyclin B is destroyed at mitotic exit, Cdc2 activity drops, allowing PP1 auto-dephosphorylation to predominate, promoting partial PP1 activation. PP1-regulated dephosphorylation of T35 then inactivates I1, allowing complete PP1 activation, dephosphorylation of mitotic phosphoproteins, and M phase exit.
To characterize phosphatases involved in dephosphorylation of mitotic phosphoproteins, without having to account for possible effects of phosphatases in controlling the APC and Cyclin degradation, we supplemented interphase egg extracts with recombinant, non-degradable Cyclin B to drive mitotic entry, and then inhibited Cdc2 using the Cdk inhibitor, Roscovitine (Ros). This inhibition resulted in marked dephosphorylation of M phase substrates, including the APC subunit, Cdc27, and multiple mitotic phosphoproteins recognized by the MPM-2 antibody (Fig. 1A). This dephosphorylation was inhibited by OA, consistent with the involvement of PP1 or PP2A-like phosphatases (Fig. 1B)11, 12. Similar results were obtained when we treated CSF-arrested extracts with Ros (S. Fig. 1). OA treatment did not affect Cdc2 kinase activity (Fig. 1B and S. Fig. 1).
As shown in Fig. 1C and 1D, the phosphatase promoting M phase exit was sensitive to 10μM, but not 1μM OA, consistent with the responsible phosphatase being PP1-like, rather than PP2A-like16. We added a PP1-specific inhibitor (either I1 or I2) to Ros-treated extracts and monitored Cdc27 and MPM2 epitope dephosphorylation. As shown in Fig. 2A and 2B, both I1 and I2 inhibited dephosphorylation of mitotic phosphoproteins (but not Cdc2 kinase activity). We then immunodepleted PP1 from either mitotic or CSF-arrested meiotic extracts prior to Ros addition and found that PP1 depletion prevented Cdc27 dephosphorylation in mitotic extracts and largely prevented dephosphorylation in the CSF-arrested extract, without affecting Cdc2 kinase activity (Fig. 2C and 2D). Adding recombinant PP1 back to the depleted extract restored Cdc27 dephosphorylation (Fig. 2D). Cdc27 exhibited partial dephosphorylation even in the presence of PP1-specific inhibitors or after PP1 depletion, reflecting either incomplete PP1 inhibition/depletion or the possible involvement of additional phosphatases.
Although Ros ought to have mimicked the shut off of Cdc2 upon Cyclin degradation, to confirm that PP1 was involved in the dephosphorylation of Cdc2 substrates following physiological Cyclin degradation, we added I1 to cycling extracts of Xenopus eggs (that oscillate between S phase and M phase) immediately prior to mitosis. As shown in Fig 2E, this largely prevented Cdc27 dephosphorylation. The timing of I1 addition was critical because if added too early, it could inhibit mitotic entry, perhaps reflecting the requirement for PP1 in promoting Cdc25 activation17, 18.
To investigate further the physiological relevance of PP1-mediated M phase substrate dephosphorylation, we examined the effect of PP1 inhibition in HeLa cell lysates. Cell lysates prepared from mitotically-arrested HeLa cells exited mitosis spontaneously (through Cyclin B degradation, S. Fig. 2A); both I1 and I2 largely prevented dephosphorylation of M phase phosphoproteins (Fig. 2F and S. Fig. 2B). Cdc2 kinase assays confirmed that the delayed mitotic substrate dephosphorylation was due to PP1 inhibition and not sustained Cdc2 kinase activity. We did not examine mitotic exit following knock-down of PP1 in intact HeLa cells as PP1 is required for mitotic entry17.
Release from CSF arrest requires calcineurin to promote M phase substrate dephosphorylation12, 13. As reported, Cdc27 dephosphorylation following Ca2+ addition to CSF extracts was markedly delayed by the calcineurin inhibitor, cyclosporine A (S. Fig. 2C). It has also been reported that an additional phosphatase cooperates with calcineurin to achieve full dephosphorylation of meiotic phosphoproteins12, 13. Since neither I1 addition nor PP1 depletion altered Cdc27 dephosphorylation following Ca2+ addition to CSF-arrested extracts, PP1 is unlikely to be the additional phosphatase required for CSF release(Fig. 2G and 2H).
Because PP1 seemed to be required for dephosphorylation of M phase phosphoproteins in the mitotic (though not meiotic) cycle, we thought excess PP1 might drive M phase substrate dephosphorylation even in the absence of Ros. However, excess PP1 (added to levels 8-fold greater than those found endogenously (0.5 μM)), did not drive Cdc27 dephosphorylation in mitotic (or meiotic) extracts (Fig. 3A). The PP1 added was active and able to dephosphorylate Cdc25 at a known PP1 site (S. Fig. 3). These data suggested that PP1 might be turned off in M phase to prevent premature dephosphorylation of M phase phosphoproteins. This was interesting in that Skoufias et al11 suggested that Cdc2 activity could inhibit dephosphorylation of Cdc2/Cyclin B substrates.
It is known that Cdc2 can inhibit PP1 and that PP1 phosphorylation at T320 peaks in mitosis in mammalian cells14. However, the physiological consequences of altering this phosphorylation have not been reported. Accordingly, we mutated T320 to A, and added this recombinant PP1 to mitotic egg extracts. As shown in Fig. 3B and 3C, when added at levels 8-fold greater than that found endogenously, the mutant, but not the WT PP1 promoted Cdc27 dephosphorylation, suggesting that PP1 inhibition by Cdc2 can be overcome by mutation of T320. However, at lower levels (~2–4 fold that found endogenously) this mutant protein alone was insufficient to drive dephosphorylation of Cdc2 substrates (Fig. 4K and 4L), raising the possibility that another cell cycle-regulated activity might physiologically control PP1.
Consistent with the idea that PP1-mediated dephosphorylation of mitotic phosphoproteins could be inhibited by Cdc2 phosphorylation of T320, when we added Ros to mitotic egg extract, Cdc27 dephosphorylation and PP1 T320 dephosphorylation were coincident (Fig. 4A). To determine how T320 dephosphorylation might be regulated, we monitored PP1 T320 phosphorylation throughout the cell cycle in a cycling extract. In the extract shown in Fig. 4B, mitosis occurred at 80 min and interphase resumed at ~100 min. T320 of PP1 was noticeably dephosphorylated by 100 min, consistent with phosphorylation of this site inhibiting PP1 function (Fig. 4B, left panel; some interphase phosphorylation may have resulted from the constitutively active Cdk2 in egg extracts). When these experiments were repeated and extracts were supplemented with I1 immediately prior to mitotic entry, T320 dephosphorylation was abrogated (Fig. 4B, right panel), suggesting that PP1 might auto-dephosphorylate T320. Indeed, PP1 could be more robustly phosphorylated in vitro by recombinant Cdc2 when I2 was included in the reaction (Fig. 4C). More importantly, pre-phosphorylated purified PP1 was able to auto-dephosphorylate in vitro in the absence of any co-factors (Fig. 4D).
If PP1 could auto-dephosphorylate to relieve its Cdc2-mediated inhibition, we reasoned that there must be an additional factor restraining PP1 activity during M phase. Specifically, since I1 is controlled by phosphorylation, we speculated that cell cycle regulation of I1 phosphorylation might contribute to control of M phase exit19. PP1 phosphorylated at T320 in CSF extracts could be co-immunoprecipitated with endogenous I1 (Fig. 4E and 4F). Moreover, mitotic PP1 exhibited much higher phosphorylation than interphase PP1 (Fig. 4G). As in Xenopus egg extracts, I1 interacted with PP1 in mitotic HeLa cell lysate (Fig. 4H).
Protein kinase A (PKA) phosphorylates I1 and this phosphorylation is required for I1 activity20. If I1-mediated phosphorylation contributes to M phase restraint of PP1, activation of PKA should strengthen I1-PP1 interactions. Accordingly, the PKA activator, 8-bromo cyclic AMP, enhanced binding of PP1 to I1 (Fig 4I). Moreover, WT I1 bound more strongly to PP1 than T35A mutant I1 lacking the site of PKA phosphorylation; T35A mutant binding could not be enhanced by 8-bromo cyclic AMP (Fig. 4I and 4J). When CSF extracts were treated with the PKA-specific inhibitor PKI, the T320A PP1 mutant, or both, PKI synergized with low levels of the T320A mutant PP1 (4-fold endogenous), allowing dephosphorylation of Cdc2 substrates, even without Ros addition (Fig. 4K). Moreover, Cdc2 substrate dephosphorylation correlated with I1 dephosphorylation (Fig. 4L). Thus, inhibiting I1 activation and preventing Cdc2-mediated PP1 inhibition could swing the Cdc2/PP1 balance in favor of substrate dephosphorylation.
This foregoing suggested that phosphorylation/dephosphorylation of I1 must be under cell cycle control. To assess this, we first examined phosphorylation of mitotic I1. As shown in Fig. 5A, I1 co-precipitated with PP1 was recognized by anti-phospho-T35 I1 antibody. PKA-mediated I1 T35 phosphorylation was, as reported, required for I1’s PP1 inhibitory activity (Fig. 5B). Moreover, PKA immunoprecipitated from mitotic egg extract or HeLa cell lysates phosphorylated I1 much more robustly than PKA from interphase extracts/lysates (Fig. 5C), consistent with a previous report that PKA activity peaks in mitosis21. Equal amounts of PKA were immunoprecipitated from interphase and mitotic extracts (S Fig. 4A). In addition, I1 lost significant activity when PKA was inhibited with PKI (Fig. 5D). Conversely, thio-phosphorylated I1 largely inhibited PP1 (and consequently Cdc27 dephosphorylation) even in the presence of PKI, indicating that I1 is a key mitotic target of PKA in preventing dephosphorylation of M phase substrates.
We next sought to identify the I1-directed phosphatase. As shown in Fig. 5E, I1 radiolabeled in the presence of γ32P ATP and PKA was dephosphorylated in interphase, but not mitotic egg extract, suggesting that the I1-directed phosphatase might be cell cycle regulated (T35 is the only PKA site on I1, S. Fig. 4B). When we treated CSF extracts with Ros, we detected I1 dephosphorylation, suggesting that I1 might be controlled by a Cdc2-inhibited phosphatase (Fig. 5F). Importantly, I1 dephosphorylation was inhibited in PP1-depleted extracts (Fig. 5G). Similar results were obtained in CSF extracts supplemented with calcium (Fig. 5H). Finally, PP1 T320A mutant addition alone (to 8-fold endogenous PP1) dramatically promoted I1 dephosphorylation and this correlated directly with the gel mobility downshift of Cdc27 (Fig. 6I).
In addition to Cyclin B destruction, mitotic exit requires dephosphorylation of mitotic phosphoproteins11, 22. We show here that the appropriately timed PP1-mediated dephosphorylation of these substrates results from the combined effects of Cdc2-mediated PP1 T320 phosphorylation, PP1 autodephosphorylation, and PKA-controlled I1-PP1 binding (Fig. 5J).
In that I1 prevented PP1 dephosphorylation in cycling extracts and I2 enhanced the weak in vitro phosphorylation of PP1 by Cdc2/Cyclin B, it appears that PP1 catalytic activity is required for T320 dephosphorylation. Indeed, prephosphorylated PP1 lost phosphorylation after incubation in the absence of any co-factor, substantiating the notion that PP1 can auto-dephosphorylate.
Although PP1 is off during M phase to ensure accumulation of mitotic phosphoproteins, some PP1 activity is required for Cdc25 activation to promote mitotic entry7. Upon entry into mitosis, PP1 activity can be suppressed through T320 phosphorylation and binding of I1, which is enhanced by the M phase peak in PKA-mediated T35 phosphorylation. Unlike in cardiac myocytes where PP2A and calcineurin appear to control I1 dephosphorylation or renal medulla, where PP2A regulates this dephosphorylation23, 24, PP1 appears to control I1 dephosphorylation in the embryonic mitotic cycles. Though we have shown that PP1 is required for T35 dephosphorylation, based on previous in vitro studies20, PP1 may not directly dephosphorylate T35 of I1. In theory, PP1 might act through its ability to dephosphorylate other sites in I1 that influence the phosphorylation/dephosphorylation of T35 or PP1 might regulate another phosphatase to promote I1 dephosphorylation25.
Though exogenous I2 could control PP1 in M phase extracts, it is not known if I2 is normally present in these extracts and/or if I2 is also cell cycle regulated. I2 does not require phosphorylation for activation and we have not seen cell cycle oscillations in stability or activity of exogenously added I2.
As cells enter mitosis, the phosphorylation of mitotic proteins is precisely timed and spatially regulated. The same is likely true for dephosphorylation of mitotic substrates by PP1, whose regulation by subcellular location remains to be determined. Proteins phosphorylated at mitosis by enzymes other than Cdc2 may be subject to distinct regulation, either by PP1 or other phosphatases.
Although targeting subunits typically confer substrate recognition by PP1, substrate recognition can also occur via direct docking of PP118, 26. It will be interesting to determine if recognition of mitotic phosphoproteins occurs via a uniform mechanism or if different substrates are recognized by different targeting subunits/direct docking. Since cell cycle regulation of PP1 appears to be at the level of the catalytic subunit (thorough phosphorylation and I1 binding), there may be no need to invoke targeting subunit regulation to explain cell cycle-dependent dephosphorylation of mitotic phosphoproteins.
We saw no effect of PP1 depletion on Ca2+ –induced release from CSF arrest. Therefore, PP1 is unlikely to be the phosphatase that cooperates with calcineurin at meiotic exit. Because OA can inhibit full dephosphorylation of meiotic phosphoproteins, and PP1/PP2A represent >95% of the OA-sensitive phosphatase activity in egg extracts, PP2A may be involved.
Following PP1 depletion or inhibition, we still observed partial Cdc27 dephosphorylation at M phase exit. Thus, it remains possible that additional phosphatases contribute to mitotic substrate dephosphorylation. In yeast, Cdc14 is the major phosphatase catalyzing M phase substrate dephosphorylation. Interestingly, negative feedback loops also link Cdc2 and Cdc1427. Even within the same species different phosphatases might be utilized under different circumstances. Moreover, more than one phosphatase may cooperate in substrate dephosphorylation at mitosis, either within a single cell type or in a manner dependent on cell type or developmental stage. Our analysis demonstrates that PP1 makes a major contribution to M phase substrate dephosphorylation in vertebrate mitotic divisions and that a tightly regulated feedback loop ensures that dephosphorylation will occur in a timely manner, well-coordinated with the destruction of Cyclin B.
Xenopus PP1 cDNA was cloned from an oocyte cDNA library using the following primers: forward-GCCGAATTCATGGGGGACGGAGAAAAACTAAA and reverse-GGCTCTAGATCATTTGGACTGTTTGTTTTTG. It was then cloned into pMAL-c2X in frame with maltose-binding protein (MBP) or pGEX-KG in frame with GST. Rabbit His-PP1 was a gift from Dr. David Armstrong (NIEHS).
His-tagged proteins or GST-fusion proteins were expressed in bacteria and purified by using Ni-NTA Agarose (Invitrogen) or glutathione-Sepharose (Amersham) beads respectively. All eluted proteins were dialyzed into XB buffer (100 mM KCl, 50 mM sucrose, 10 mM Hepes, pH 7.7, 1 mM MgCl2, and 0.1 mM CaCl2, 0.5mM EGTA).
Rabbit polyclonal anti-xPP1 was raised against MBP-xPP1 protein and affinity purified using GST-xPP1. Anti-Cdc27 (BD Biosciences, Franklin Lakes, NJ), anti-MPM2 (Upstate, Lake Placid, NY), anti-pCdc2 (Upstate), anti-pPP1T320 (Abcam, Cambridge, MA), anti-Cyclin B1 (Santa Cruz, CA), anti-PKA (Abcam), and anti-pPlk1 (Abcam) antibodies were purchased. Anti-Cyclin B2 antibody was as reported28.
Xenopus cycling extracts, interphase extracts, and CSF extracts were prepared as described previously29. Mitotic extracts were prepared by adding non-degradable Cyclin B1 to interphase extracts.
For immunodepletion of PP1, CSF or mitotic extracts were incubated with purified polyclonal anti-xPP1 antibody coupled to protein A Sepharose beads for 30 min at 4 °C. Following incubation, the beads were removed, and the supernatants were treated again for another two consecutive depletions. The same amount of control rabbit IgG (Jackson Immuno Research) was used for mock depletion.
HeLa cells were treated with 40 ng/mL of Nocodazole for 17 hours to arrest in mitosis. The cells were then washed with PBS twice and cultured in fresh DMEM for another 1 hour. The cells were finally lysed in hypotonic buffer (20 mM HEPES pH 7.5, 5 mM KCl, 1.5 mM MgCl2 and 1 mM DTT) with dounce and cellular supernatants were collected after centrifuge.
To generate prephosphorylated GST-Cdc25 or GST-I1, Glutathione-Sepharose beads coupled with GST fusion proteins were incubated in kinase buffer (10 mM Tris-HCl, pH 7.2, 0.1mM ATP, 2μCi γ-32P ATP, 10mM MgCl2, 1mM DTT, pH7.2) with Chk1 or PKA (New England Biolabs, Ipswich, MA) for 30min at room temperature, then washed. Glutathione-Sepharose beads bound to prephosphorylated proteins were then dipped into phosphatase buffer (10 mM Tris-HCl, pH 7.2, 10mM MgCl2, 1mM DTT) with His-PP1 or the indicated extracts. Aliquots were retrieved, washed with PBS plus 300 mM NaCl and 0.1% Triton, eluted by SDS sample buffer and subjected to SDS-PAGE. The phosphorylation was detected by phosphorimager (Molecular Dynamics).
This work was supported by RO1 GM67225 to SK and DA10044 to ACN.