Report

Greatwall Phosphorylates an Inhibitor of Protein Phosphatase 2Α That Is Essential for Mitosis

See allHide authors and affiliations

Science  17 Dec 2010:
Vol. 330, Issue 6011, pp. 1670-1673
DOI: 10.1126/science.1195689

Abstract

Entry into mitosis in eukaryotes requires the activity of cyclin-dependent kinase 1 (Cdk1). Cdk1 is opposed by protein phosphatases in two ways: They inhibit activation of Cdk1 by dephosphorylating the protein kinases Wee1 and Myt1 and the protein phosphatase Cdc25 (key regulators of Cdk1), and they also antagonize Cdk1’s own phosphorylation of downstream targets. A particular form of protein phosphatase 2A (PP2A) containing a B55δ subunit (PP2A- B55δ) is the major protein phosphatase that acts on model CDK substrates in Xenopus egg extracts and has antimitotic activity. The activity of PP2A-B55δ is high in interphase and low in mitosis, exactly opposite that of Cdk1. We report that inhibition of PP2A-B55δ results from a small protein, known as α-endosulfine (Ensa), that is phosphorylated in mitosis by the protein kinase Greatwall (Gwl). This converts Ensa into a potent and specific inhibitor of PP2A-B55δ. This pathway represents a previously unknown element in the control of mitosis.

Progression from interphase to mitosis requires that a large number of proteins be phosphorylated to bring about processes such as nuclear-envelope breakdown, chromosome condensation, and spindle assembly (1). Cyclin-dependent kinase 1 (Cdk1)–cyclin B, also called maturation-promoting factor (MPF), is the main (but not the sole) protein kinase that catalyzes these modifications (2). The activation of MPF requires that Cdc25, a phosphatase for the phospho-Tyr15 residue of Cdk1, is active, whereas Wee1 and Myt1, the protein kinases that phosphorylate Tyr15, must be turned off. This is achieved by multisite phosphorylation of Cdc25, Wee1, and Myt1 (35), mainly catalyzed by MPF itself (6), and is opposed by protein phosphatase 2A (PP2A)–B55δ (PP2A-B55δ) (7). This form of PP2A is a major phosphatase for model CDK substrates (8), and depletion of PP2A-B55δ accelerates mitotic progression in Xenopus extracts (8), partly by promoting the activity of Cdc25 and the inactivity of Wee1 and Myt1, and partly by reducing antagonizing phosphatase activity against the substrates of CDKs and other mitotic kinases (9). Crucially, the activity of PP2A-B55δ fluctuates during the cell cycle, being high in interphase and low in mitosis (10). This presumably accentuates the switchlike behavior of the system; when a kinase is active, the opposing phosphatase is inactive and vice versa. This behavior also avoids futile cycles. But how PP2A-B55δ phosphatase is regulated was not known.

An important clue to the control of PP2A-B55δ has recently emerged from studies of the protein kinase Greatwall (Gwl) (11). In Xenopus egg extracts, Gwl is activated in mitosis and is essential for both entry into mitosis and maintenance of the mitotic state (11). Gwl apparently promotes mitosis by inactivating PP2A (9, 12). We were unable to detect substantial phosphorylation of the components of PP2A-B55δ in vitro (13); hence, we suspected that Gwl might instead act by phosphorylating and activating some as-yet unidentified PP2A-B55δ–specific inhibitor protein. We therefore searched for Gwl substrates with the KESTREL (kinase substrate tracking and elucidation) method (14) in which active Gwl was incubated with extracts containing potential substrates, under conditions where endogenous protein kinase activity was suppressed. Several proteins in interphase egg extracts were phosphorylated by Gwl (Fig. 1A), and two prominent substrates of small apparent molecular size (arrows in Fig. 1A) were heat stable (15). The boiled extract was fractionated on a Mono-Q column, and fractions containing the major substrate were analyzed by mass spectrometry. We detected about 80 different Xenopus proteins, but no Gwl-dependent phosphorylated peptides emerged from the analysis. Thus, we choose 29 proteins with molecular sizes between 10 and 30 kD from the list and expressed them in bacteria (Fig. 1B). Seventeen of them were soluble after boiling, and these proteins were further tested as substrates for Gwl in vitro. One protein—cyclic adenosine monophosphate (cAMP)–regulated phosphoprotein-19 (Arpp-19)—was a better substrate for Gwl than any of the others (Fig. 1B, lane 22). Arpp-19 is a major substrate for cAMP-activated protein kinase (PKA) in postsynaptic neurons (16). Arpp-19 is a close relative of α-endosulfine (Fig. 1C), which was identified as an endogenous ligand of sulfonylurea receptor K+ channels in the pancreas (17), although this idea seems to have been abandoned by its original proponents (18). Recombinant endosulfine (Ensa), like Arpp-19, was a good substrate for Gwl. Substitution of alanine for all of the potential conserved phosphorylation sites in Arpp-19 and Ensa (11 in the former and 7 in the latter) abolished phosphorylation by Gwl (Fig. 1D and fig. S1). In contrast, mutants of Arpp-19 and of Ensa that retained Ser67 in the highly conserved sequence FDSGDY (19) were phosphorylated by Gwl to a similar degree as the wild-type (WT) proteins. Because these proteins also have conserved CDK (S/T-P-X-K/R) and PKA (K/R-K/R-X-S/T) consensus sites, we tested to see whether they were substrates for these kinases in vitro. Indeed, Ser28 of Arpp-19 and Thr28 of Ensa were strongly phosphorylated by Cdk2-cyclin A (and Thr99 residues of both proteins phosphorylated weakly). Ser109 of both proteins was the major site targeted by PKA (Fig. 1D and fig. S1A) (20).

Fig. 1

Ensa and ARPP-19 are substrates of Gwl. (A) Phosphorylation of two small proteins in Xenopus interphase egg extract to which Gwl was added. Coomassie blue staining (right) and autoradiography (left). The positions of the two candidates and of Gwl are indicated with arrows and an arrowhead, respectively. (B) In vitro phosphorylation of candidate substrates by Gwl. Lanes marked “Mw” are molecular size markers. (C) Sequence alignment of X. laevis Ensa and Arpp-19. Conserved residues are shown in red. (D) In vitro phosphorylation of Ensa by Gwl. WT Ensa and mutants were phosphorylated by Gwl kinase (top), CDK (middle), or PKA (bottom). In the 7A mutant, Ser2, Thr28, Ser51, Ser67, Thr93, Thr99, and Ser109 of Ensa were all substituted by alanine. In the 6A mutants, one of these seven residues was restored to its original state as indicated.

We raised antibodies against Arpp-19 and Ensa in rabbits and used them to measure the concentrations of Arpp-19 and Ensa in Xenopus egg extracts. Arpp-19 was hardly detectable, despite its presence in the original mass spectrometric analysis. On the other hand, Ensa was present at ~150 to 300 nM in egg extracts, in excess of PP2A-B55δ, whose concentration we measured at ~50 to 70 nM.

We tested whether Ensa affected the phosphatase activity of PP2A-B55δ. Although dephosphorylated Ensa did not inhibit any form of PP2A, Ensa phosphorylated by Gwl strongly inhibited the PP2A-B55δ heterotrimer holocomplex (Fig. 2A), but not the monomeric catalytic subunit or the dimer comprising the A scaffold and C catalytic subunits (AC). Phosphorylation of Ensa by CDK or PKA alone did not generate any inhibitory activity against PP2A-B55δ. Similar results were obtained for Arpp-19 (fig. S1B). These results indicate that Arpp-19 and Ensa are converted into inhibitors of PP2A-B55δ by phosphorylation on Ser67.

Fig. 2

Regulation of PP2A-B55δ by Ensa. (A) Phosphatase activities of recombinant PP2A-C monomer, AC dimer, and AB55δC trimer complexes were analyzed with and without Ensa using 32PO4-labeled maltose binding protein fused to a 25-residue peptide containing Ser50 of Fizzy (MBP-Fizzy-Ser50) as a substrate (10). Results are the means ± range for three experiments. The activities are expressed as percentages of controls. Open bars, buffer; gray bars, Ensa; black bars, Ensa thiophosphorylated by Gwl. Error bars indicate the range from three independent experiments. (B) Physical interaction between Ensa and PP2A subunits in Xenopus interphase egg extract. Fractions bound to (lanes 5 to 7) and not retained by (lanes 2 to 4) Ensa beads were analyzed for PP1, PP2A, and PP5 subunits.

We tested the specificity of Ensa’s interactions with various forms of PP2A and other protein phosphatases by making affinity matrices covalently conjugated with Ensa. Gwl-phosphorylated Ensa strongly bound PP2A-B55δ from interphase egg extracts, but no other B subunits, nor the PP1 or PP5 catalytic subunit, were retained by Ensa beads (Fig. 2B). Mock- and PKA-phosphorylated Ensa bound PP2A-B55δ only weakly. These results suggest that Ensa inhibits PP2A-B55δ by a Ser67 phosphorylation-dependent physical interaction, which makes it a highly specific inhibitor of PP2A-B55δ. Neither Ensa nor Arpp-19 inhibited PP1, regardless of their phosphorylation state (fig. S1C).

Antibodies specific for phospho-Ser67 reacted with Ensa in mitotic extracts much more strongly than in interphase extracts (Fig. 3A). Using Phos-Tag acrylamide gels, which specifically retard phosphorylated proteins (21), we confirmed that the majority of Ensa became phosphorylated in mitotic extracts (Fig. 3A) and that Ser67 was required for the observed phosphorylation-dependent shift in migration on polyacrylamide gel electrophoresis (Fig. 3A). These results show that Ensa is quantitatively phosphorylated at its Ser67 residue in mitosis, which leads to the inhibition of PP2A-B55δ. The timing of Ensa phosphorylation in relation to that of a number of other cell-cycle markers is shown in fig. S2. The phosphorylation of anaphase-promoting complex 3 (Apc3), Cdc25, Wee1, and Gwl all took place at the same time at the resolution of this experiment.

Fig. 3

Phosphorylation and function of Ensa in Xenopus egg extracts. (A) Proteins from Xenopus egg extracts arrested in the indicated cell-cycle stages were supplemented with or without His-tagged recombinant WT or S67A mutant Ensa. Recombinant (added) or endogenous Ensa was detected by immunoblotting with antibodies to Ensa (top) or phospho-Ser67 (middle). The same set of samples was also analyzed by SDS–polyacrylamide gel electrophoresis in the presence of Phos-Tag (bottom). The migration positions of dephosphorylated [Ensa-OH, added-OH, or Gwl-phosphorylated (Ensa-PO4, added-PO4] proteins are shown. (B) Endogenous Ensa was immunodepleted (lane 2) from fresh cycling extract and either WT (lane 3) or S67A mutant (lane 4) of recombinant Ensa added back. (C to E) Egg extracts shown in (B) were incubated at 23°C, and aliquots were taken at 7-min intervals for analysis. Apc3 (upper panels), cyclin B2 (middle panels), and phosphorylation of Tyr15 of Cdc2 (lower panels) were detected by immunoblotting. Histone H1 kinase (black squares) and MBP-Fzy-Ser50 phosphatase activities (red circles) were measured and plotted (70 min of H1 kinase and 0 min of Ser50 phosphatase activities of mock extract taken as 100%). (C) Mock depletion; (D) Ensa-depleted; (E) Ensa-depleted with WT Ensa added back. See fig. S3 for additional data from this experiment. Arrows indicate the mobilities of interphase and mitotic Apc3. A.U., arbitrary units.

We tested the effects of Ensa depletion on “cycling” frog egg extracts (22, 23). In this experiment, the control (Fig. 3C) performed two cycles of entry into and exit from mitosis, whereas the Ensa-depleted extract (Fig. 3D) never entered M phase, despite larger amounts of histone H1 kinase activity than in the control. Apc3 was never fully phosphorylated, and mitotic phosphoproteins showed only slightly increased intensity (fig. S3C). Cyclin destruction was not activated in the Ensa-depleted extract. Adding back WT recombinant Ensa to depleted extracts restored induction of mitosis (Fig. 3E), whereas the Ser67→Ala67 (S67A) mutant protein did not (fig. S3A), indicating that residue Ser67 of Ensa is crucial for proper cell-cycle control. Tyr15 of Cdk1 was dephosphorylated in all three conditions, but in the Ensa-depleted extract, Tyr15 remained dephosphorylated for the duration of the incubation, whereas in the controls, Tyr15 was rephosphorylated. This experiment implies that the main effect of not inhibiting PP2A-B55δ is in antagonizing Cdk1 phosphorylation of downstream target proteins, rather than on the control of Cdk1 activity. Somehow, Cdc25 or an equivalent activity that dephosphorylated Tyr15 was activated in the Ensa-depleted extract, yet despite strong histone H1 kinase activity, the extract failed to enter mitosis.

We tested the results of adding thiophosphorylated Ensa to egg extracts. Supplemental fig. S4 shows that when extra Ensa was added to interphase extracts supplemented with cycloheximide and a small amount of recombinant stable cyclin B (not enough to induce mitosis by itself), phosphorylation events typical of mitotic extracts were induced when Gwl-thiophosphorylated Ensa was added. Dephosphorylated Ensa or Ensa thiophosphorylated by PKA did not show any effect. These results are all consistent with the idea that Gwl inhibits PP2A-B55δ by phosphorylating Ensa, thereby both activating Cdk1 (by inhibiting Wee1 and turning on Cdc25) and allowing Cdk1 to efficiently phosphorylate its target proteins in mitosis by suppressing the activity of the main opposing protein phosphatase.

In this paper, we identify Ensa and Arpp-19 as phosphorylation-dependent inhibitors of PP2A-B55δ and physiological substrates of Gwl kinase. In Drosophila, Ensa is required for proper spindle assembly and oocyte maturation; endosulfine-deficient oocytes are unable to pass between prophase of first meiosis into metaphase (2426). Tellingly, these oocytes have high Cdk1 activity yet display little phosphorylation of mitotic substrates, exactly as we observed in Xenopus extracts, consistent with a failure to shut off protein phosphatase(s) upon entry into M phase (25). The Gwl-Arpp-Ensa module appears to be active in human cells, because reduction of Gwl levels by RNA interference blocks cells in G2 phase of the cell cycle (27).

It is still difficult to understand the details of how gradually accumulating cyclin levels are converted into the sharp activation of MPF at the G2-to-M transition. Gwl appears to require activation by MPF (11), and once it is turned on, it phosphorylates Ensa, which, in turn, switches off PP2A-B55δ (Fig. 4). This promotes the activation of MPF by increasing phosphorylation of Wee1, Myt1, and Cdc25, and it also assists entry into mitosis by reducing the dephosphorylation of MPF targets. This is an example of a “coherent feed-forward loop” (28), because Cdk1 phosphorylates its own activation module in a positive feedback loop, it phosphorylates its mitotic target proteins, and it indirectly inactivates the antagonizing protein phosphatase by activating Gwl and the downstream phosphatase inhibitor Ensa. Clearly, once this system is active, the cell switches completely into mitosis in a kind of latch mechanism. Thus, both Gwl and Ensa are essential for the maintenance of the mitotic state, implying that inhibition of protein phosphatases is critically important for this process. How the return to interphase is brought about is unclear. When cyclins are degraded and Cdk1 levels fall, Gwl and Ensa are dephosphorylated, presumably by protein phosphatase(s) other than PP2A-B55δ, such as PP1 (29, 30). The threshold for activation of Gwl and entry into mitosis depends on the balance of activities of these as-yet unidentified phosphatases and that of MPF.

Fig. 4

Diagram of how PP2A-B55δ is inactivated by CDK via the Gwl and Ensa pathway. Pathways in blue are active in interphase; pathways in red denote those active during mitosis.

Supporting Online Material

www.sciencemag.org/cgi/content/full/330/6011/1670/DC1

Methods

Figs. S1 to S4

References and Notes

  1. Methods are available as supporting material on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. We thank H. Mahbubani, J. Kirk, and L. Egbuniwe for care of frogs and other members of the laboratory for advice and reagents, especially J. Gannon for the antibody to doubly phosphorylated Thr14-Tyr15 of Cdk1 (CP 3.2). S.M. was supported by a fellowship from the Japan Society for the Promotion of Science.
View Abstract

Navigate This Article