Phosphorylation of Phosphatase Inhibitor-2 at Centrosomes during Mitosis*

Inhibitor-2 (I-2) is a regulator of protein phosphatase type-1 (PP1), known to be phosphorylated in vitro by multiple kinases. In particular Thr 72 is a Thr-Pro phosphorylation site conserved from yeast to human, but there is no evidence that this phosphorylation responds to any physiological signals. Here, we used electro-phoretic mobility shift and immunoblotting with a site-specific phospho-Thr 72 antibody to establish Thr 72 phosphorylation in HeLa cells and show a 25-fold increase in phosphorylation during mitosis. Mass spectrometry demonstrated I-2 in actively growing HeLa cells was also phosphorylated at three other sites, Ser 120 , Ser 121 , and an additional Ser located between residues 70 and 90. In vitro kinase assays using recombinant I-2 as a substrate showed that the Thr 72 kinase(s) was activated during mitosis, and sensitivity to kinase inhibitors indicated that the principal I-2 Thr 72 kinase was not GSK3 but instead a member of the cyclin-dependent protein kinase family. Immunocytochemistry confirmed Thr 72 phosphorylation of I-2 during mitosis, with peak intensity at prophase, and revealed subcellular concentration of the phospho-Thr 72 I-2 at centrosomes. Together, the data show dynamic changes in I-2 phosphorylation during mitosis and acetic acid microspray data-dependent routine with the sequence quired the cycle (cid:1) tandem mass spectrometry spectra of peptides ranging in over several orders of magnitude.

Inhibitor-2 (I-2) is a regulator of protein phosphatase type-1 (PP1), known to be phosphorylated in vitro by multiple kinases. In particular Thr 72 is a Thr-Pro phosphorylation site conserved from yeast to human, but there is no evidence that this phosphorylation responds to any physiological signals. Here, we used electrophoretic mobility shift and immunoblotting with a sitespecific phospho-Thr 72 antibody to establish Thr 72 phosphorylation in HeLa cells and show a 25-fold increase in phosphorylation during mitosis. Mass spectrometry demonstrated I-2 in actively growing HeLa cells was also phosphorylated at three other sites, Ser 120 , Ser 121 , and an additional Ser located between residues 70 and 90. In vitro kinase assays using recombinant I-2 as a substrate showed that the Thr 72 kinase(s) was activated during mitosis, and sensitivity to kinase inhibitors indicated that the principal I-2 Thr 72 kinase was not GSK3 but instead a member of the cyclin-dependent protein kinase family. Immunocytochemistry confirmed Thr 72 phosphorylation of I-2 during mitosis, with peak intensity at prophase, and revealed subcellular concentration of the phospho-Thr 72 I-2 at centrosomes. Together, the data show dynamic changes in I-2 phosphorylation during mitosis and localization of phosphorylated I-2 at centrosomes, suggesting involvement in mammalian cell division.
Inhibitor-2 (I-2) 1 was first identified as a heat stable protein capable of selectively inhibiting the phosphorylase phosphatase activity of the PP1 catalytic subunit (1). Subsequent studies purified a heterodimer containing PP1 and I-2, termed MgATP-dependent phosphatase, and showed that I-2 phosphorylation at Thr 72 by GSK3 resulted in transient activation of this complex (2)(3)(4). Phosphorylation of I-2 at Ser 86 , Ser 120 , and Ser 121 by casein kinase II (CKII) promoted phosphorylation at Thr 72 by GSK3, but CKII phosphorylation alone had little effect on the activity of the complex. More recent studies sug-gest that I-2 is subject to phosphorylation by multiple protein kinases, including CKII, CKI, cdc2, mitogen-activated protein kinase, and GSK3 (2)(3)(4)(5)(6)(7). A subset of these kinases, also known as Pro-directed kinases, namely GSK3, cdc2, and mitogenactivated protein kinase, specifically phosphorylate I-2 in the conserved Thr 72 -Pro site and therefore may modulate PP1 function in cells.
Phosphorylation of I-2 in animal tissues has been studied. Depaoli-Roach and Lee (8) immunoprecipitated I-2 from 32 P metabolically labeled mouse diaphragm. Phospho-amino acid analysis showed that I-2 was phosphorylated on both serine and threonine, with ϳnine times more [ 32 P]phosphoserine than [ 32 P]phosphothreonine. Holmes et al. (9) purified I-2 from rabbit skeletal muscle and using mass spectrometry showed that Ser 86 (0.7 mol/mol), Ser 120 , and Ser 121 (0.3 mol/mol each) were phosphorylated, but no phosphorylation of Thr 72 was detected. Lawrence et al. (10) immunoprecipitated I-2 from 32 P-labeled mouse fat cells and showed the presence of both [ 32 P]phosphoserine and [ 32 P]phosphothreonine, with the phosphothreonine level being less than 10% of phosphoserine. Moreover, despite the fact that GSK3 is inactivated in response to insulin, threonine phosphorylation of I-2 was unchanged by insulin treatment of the fat cells (10). This suggested that phosphorylation of Thr 72 in vivo was most likely not regulated by GSK3. Physiological regulation of Thr 72 phosphorylation in I-2 has not been demonstrated.
To analyze the phosphorylation of I-2 in living cells we generated a phospho-site-specific antibody that uniquely recognized I-2 phosphorylated at Thr 72 . Using this antibody, we show here that I-2 is indeed phosphorylated on Thr 72 in cells, and this modification is enhanced nearly 25-fold during mitosis. Using kinase inhibitors we tested whether Thr 72 phosphorylation of I-2 in mitotic cells was catalyzed by GSK3 or cyclindependent protein kinases. Immunostaining showed that phosphorylation of I-2 at Thr 72 in mitotic cells was concentrated at centrosomes. The data suggested that reversible phosphorylation of I-2 may control centrosome-associated PP1 activity, which could play a role in mammalian cell division.
Cell Synchronization and Cell Lysate Preparation-HeLa cells were synchronized by incubating with 400 M mimosine for 24 h or with 200 ng/ml nocodazole for 16 h. Cells were rinsed with PBS and scraped into SDS-containing sample buffer.
Inhibitor 2 Immunoprecipitation-HeLa cells were rinsed with room temperature PBS, and the dish was placed on a mixture of acetone and dry ice, and 1 ml of ice-cold 5% trichloroacetic acid was added for 10 min. Cells were scraped from the dish using a Costar cell lifter. The slurry was transferred to a 1.7-ml microcentrifuge tube and centrifuged for 10 min at 10,000 ϫ g. The supernatant was discarded, and the pellet was washed twice with cold acetone and resuspended in urea resolubilization buffer (9 M urea, 20 mM Tris, 23 mM glycine, 1 mM EDTA, and 10 mM ␤-glycerophosphate). This suspension was incubated at 30°C for ϳ1 h until the protein pellet was dissolved. The solution was diluted 10-fold with immunoprecipitation buffer (1% Nonidet P-40, 200 mM NaCl, 50 mM Tris, pH 8, 1 mM EDTA, 1 mM EGTA, 0.5% deoxycholate, 20 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, 0.1% 2-mercaptoethanol, 0.4 mM Pefabloc, 1 M Microcystin-LR). After centrifugation for 10 min at 10,000 ϫ g the supernatant was used for I-2 immunoprecipitation. Briefly, the lysate was incubated at 4°C for 2 h with I-2 antibody coupled to protein G beads. The beads were pelleted and washed with wash buffer (0.1% Nonidet P-40, 50 mM MOPS, pH 7.4, 150 mM NaCl, 5 mM EGTA, 5 mM EDTA, 1 M Microcystin-LR, 20 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, 0.1% 2-mercaptoethanol) three times. The beads were resuspended in SDS-containing sample buffer, and the proteins were resolved by SDS-PAGE.
Mass Spectrometric Analysis-The protein band corresponding to I-2 was excised from the gel and destained with 0.5 ml of 50% methanol/5% acetic acid overnight at room temperature before dehydration in 200 l of acetonitrile and drying in a vacuum centrifuge. The proteins were reduced by addition of 50 l of 10 mM dithiothreitol and alkylated by addition of 50 l of 100 mM iodoacetamide (each for 30 min at room temperature). The gel slices were dehydrated in 200 l of acetonitrile, hydrated in 200 l of 100 mM ammonium bicarbonate, and dehydrated again with 200 l of acetonitrile. The dehydrated gel pieces were again dried in a vacuum centrifuge and rehydrated in 50 l of 20 ng/l ice-cold, sequencing-grade modified porcine trypsin (Promega) for 5 min on ice. Excess trypsin was removed, and digestion was carried out overnight at 37°C. The peptides released by the digestion were collected by successive extractions with 50 l of 50 mM ammonium bicarbonate and 50 l of 50% acetonitrile/5% formic acid (2ϫ), combining the extracts in a siliconized 0.6-ml microcentrifuge tube rinsed previously with ethanol, water, and ethanol. The combined extract was concentrated in a vacuum centrifuge to 20 l prior to analysis on a Finnigan LCQ (ThermoQuest) ion-trap mass spectrometer with a Protana nanospray ion source that was interfaced to a self-packed 8-cm ϫ 75-m inner diameter Phenomenex Jupiter 10-m C18 reverse-phase capillary column. The peptides (0.5 l) were injected and eluted from the column with an acetonitrile/0.1 M acetic acid gradient (2-85% acetonitrile in 30 min) at a flow rate of 0.25 l min Ϫ1 . The microspray ion source was set at 2.8 kV, and the data were analyzed using a full data-dependent acquisition routine with the amino acid sequence acquired in the four scans before the cycle repeats. This produced ϳ500 tandem mass spectrometry spectra of peptides ranging in abundance over several orders of magnitude.
Antibody Production and Purification-The synthetic peptide, MKIDEPTPYH, with the threonine phosphorylated was conjugated to either hemocyanin or thyroglobulin. Antisera were produced by subcutaneous injection of this peptide into rabbits. The serum was precleared over unphosphorylated recombinant I-2. The unbound fraction was incubated with I-2 phosphorylated by both CKII and GSK3␤ and coupled to AffiGel-15 (Bio-Rad). After several washes with PBS, the column was washed with 50 mM NaPO 4 , pH 5.5. The antibody that remained bound to the phosphorylated I-2 was eluted with 0.1 M glycine, pH 1.8 in 10% dioxane. The antibody solution was neutralized and concentrated. This antibody is referred to as phospho-Thr 72 (9).
Immunocytochemistry-HeLa cells were cultured on fibronectincoated coverslips. Coverslips were washed once with PBS followed by incubation in cold 10% trichloroacetic acid for 10 min. Cells were rinsed three times with PBS and permeabilized with 0.1% Triton-X-100 for 15 min. Specimens were blocked with 3% bovine serum albumin/1% ovalbumin and incubated with a purified sheep antibody to I-2, the rabbit antibody to phospho-Thr 72 I-2, or both. A goat anti-sheep IgG antibody-conjugated rhodamine (Amersham Biosciences) or a goat antirabbit IgG conjugated to Alexafluor was used for immunofluorescent imaging. Coverslips were mounted onto glass slides with Vectorshield mounting medium (Vector Laboratories). Cells were visualized using a Nikon Microphot-5A microscope with ϫ20 objective, and images were captured with a Hamamatsu C4742 digital camera operated by Open-Lab software (Improvision) and processed in Adobe Photoshop.

RESULTS
Specificity of Phospho-site Antibody against P-Thr 72 of I-2-To examine I-2 phosphorylation at Thr 72 in cells we developed an phospho-site-specific antibody using a phosphorylated peptide corresponding to the amino acid sequence surrounding this site in human I-2. The antibody was affinity purified using differential binding between unphospho-and phospho-I-2. Specificity was confirmed with recombinant I-2 (rI-2) that had been phosphorylated in vitro using purified protein kinases. As seen in Fig. 1, the antibody did not crossreact with the unphosphorylated rI-2 or rI-2 phosphorylated by CKII alone (i.e. phosphorylated at Ser 86 , Ser 120 , and Ser 121 ). However, the phospho-Thr 72 antibody recognized rI-2 that had been phosphorylated by the combination of GSK3 and CKII, which is well known to produce I-2 phosphorylation at Thr 72 , in addition to Ser 86 , Ser 120 , and Ser 121 . Thus, the antibody only recognized I-2 that was phosphorylated at Thr 72 .
Phosphorylation of I-2 in Living Cells-HeLa cells were prepared at different phases of the cell cycle, either unsynchronized (U, mostly interphase; see Fig. 2) or synchronized by addition of mimosine (to block at S phase) or nocodazole (to block at mitosis). Cell lysates were subjected to SDS-PAGE and compared by Western blotting (Fig. 2A). I-2 migrated as a single band in either unsynchronized or S-phase cells and was recognized by anti-I-2 antibody. This form of I-2 was not reactive with phospho-Thr 72 antibody, suggesting that Thr 72 in I-2 was predominantly unphosphorylated in these cells. However, the I-2 recovered from nocodazole-treated cells displayed an additional band of I-2 with decreased mobility. This lower mobility form of I-2 was readily recognized by the anti-phospho-Thr 72 antibody, showing phosphorylation of I-2 at Thr 72 in cells during mitosis.
We immunoblotted total cell proteins from unsynchronized and mitotic HeLa cells with the phospho-Thr 72 site-specific antibody (Fig. 2B). In proteins from mitotic cells only one band corresponding to phospho-Thr 72 I-2 was specifically stained near 35 kDa, and this was not detected in unsynchronized cells. One nonspecific band at 50 kDa was stained in both samples with equal intensity. These results further demonstrated the high degree of specificity of the phospho-Thr 72 antibodies and confirmed that phosphorylation of I-2 at the Thr 72 site occurred in particular during mitosis.
To directly determine the phosphorylation state of all the other sites in I-2 in unsynchronized HeLa cells, we employed the trichloroacetic acid fixation method. This protocol was first developed for immunohistochemistry, to preserve the phosphorylation state of proteins that were otherwise rapidly dephosphorylated (12). We adapted this method to purify phosphorylated I-2. Cells were directly treated with trichloroacetic acid, scraped, and centrifuged, and the trichloroacetic acid-precipitated proteins were resuspended in a saturated urea solution. Following the dilution of the urea solution, I-2 was immunoprecipitated, resolved by SDS-PAGE, and visualized by Coomassie blue staining. The I-2 protein was excised from the gel and digested with trypsin. Mass spectrometry revealed that Ser 120 and Ser 121 were phosphorylated to a stoichiometry of 0.9 mol/mol each. An additional phosphate was also detected in a peptide corresponding to residues 70 -90 (which contains a single serine residue, Ser 86 ), but this could not be further analyzed because of low yield of this peptide (Fig. 3). Alkaline phosphatase treatment of HeLa cell lysates resulted in conversion of I-2 to a single band that migrated faster than all of the forms seen in Fig. 2A, consistent with Ser phosphorylation of I-2 in cells throughout the cell cycle (data not shown).
Thr 72 Phosphorylation of I-2 during Mitosis-To examine in more detail the phosphorylation of I-2 on Thr 72 during mitosis, we employed immunoblotting with the phospho-Thr 72 site-specific antibody using unsynchronized (U) and mitotic cells (M) (see Fig. 4) isolated by shaking off cells from nocodazole-treated cultures. The shake-off procedure greatly enriched the population of mitotic cells, relative to nocodazole treatment alone. Using approximately the same amount of total I-2 protein in the samples, immunoblotting with anti-phospho-Thr 72 demonstrated a 25-fold increase in I-2 phosphorylation at Thr 72 in mitotic HeLa cells (Fig. 4).
We prepared lysates from unsynchronized, mimosine-synchronized, or nocodazole-synchronized HeLa cells to determine whether increased I-2 (Thr 72 ) phosphorylation during mitosis corresponded to an increase in kinase specific activity. These lysates were used as source of I-2 kinase activity. The same amount of lysate protein was incubated with recombinant I-2 plus MgATP. The results of the phosphorylation reactions were analyzed by Western blotting with the anti-phospho-Thr 72 antibody (Fig. 5A). Unsynchronized cells displayed a low level of I-2 (Thr 72 ) kinase activity. In contrast, little or no I-2 (Thr 72 ) kinase activity was observed in lysates of mimosine-synchronized cells. Lysates from nocodazole-treated cells displayed significantly higher I-2 (Thr 72 ) kinase activity, indicating that increased I-2 (Thr 72 ) phosphorylation observed during mitosis resulted from the activation of a mitotic I-2 (Thr 72 ) kinase.
Analysis of I-2 (Thr 72 ) Phosphorylation in Cells-HeLa cells were stained using immunohistochemistry with the phospho-Thr 72 site-specific I-2 antibody (Fig. 6A). We demonstrated the specificity of this antibody in this protocol, in addition to the results seen in Fig. 2. Pre-mixing the antibody with 1 M phospho-Thr 72 I-2 abolished all the immunostaining (Fig. 6C), in contrast with pre-mixing with 1 M unphosphorylated I-2, which had little effect on staining (Fig. 6E). DNA was stained using DAPI (shown in Fig. 6, B, D, and F) to show the alignment of condensed chromosomes at the metaphase plate, establishing that the stained cells were at the same stage of mitosis. Staining of HeLa cells with an I-2 antibody (Fig. 7A) and with the phospho-Thr 72 antibody (Fig. 7B) revealed the most intense staining for phospho-Thr 72 occurred at the centrosomes. This is highlighted in the merged images of total plus phospho-I-2 (panel C) and in the phospho-Thr 72 minus total I-2 images (panel D). Quantitation of the phospho-Thr 72 staining relative to total I-2 revealed that I-2 (Thr 72 ) phosphorylation peaked in prophase cells and decreased as cells progressed through mitosis (Fig. 7). DISCUSSION Mammalian PP1 activity is controlled by multiple regulatory subunits and inhibitor proteins. The most ancient of the PP1 regulators is I-2, called Glc8 in yeast, and the most highly conserved region of the protein surrounds the Thr-Pro phosphorylation site, with the sequence IDEPXTPY. Mutation of the corresponding residue Thr 118 in yeast Glc8 prevents rescue of temperature-sensitive mutants of Ipl1 protein kinase, indicating that this phosphorylation site is important for function (13). Numerous biochemical studies established that I-2 forms a 1:1 complex with the PP1 catalytic subunit to form a heterodimer termed MgATP-dependent phosphatase (2,3). The activity of this latent heterodimer was transiently increased following the phosphorylation of I-2 at Thr 72 by GSK3 (2)(3)(4). This proposed regulatory mechanism has remained controversial in part because previous studies have failed to demonstrate regulation of I-2 (Thr 72 ) phosphorylation in vivo.
In this study we produced a new experimental tool, namely an anti-phospho-Thr 72 antibody, and showed I-2 becomes phos-phorylated on Thr 72 in living cells during mitosis. We also utilized gel mobility shifts and mass spectrometry to demonstrate that I-2 was phosphorylated throughout the cell cycle on three Ser residues. Our tests of kinase inhibitors indicate that GSK3, the best known I-2 kinase, contributes little to phosphorylation of I-2 during mitosis. Instead, the data point to other mitotic protein kinases. Roscovitine, a broad spectrum CDK inhibitor, blocked 60% of the I-2 (Thr 72 ) kinase activity when used at a dose more than 10 times the IC 50 for the purified cyclinB⅐cdc2 kinase complex. Thus, we conclude that one or more CDKs account for the majority of the mitotic I-2 (Thr 72 ) kinase activity. Consistent with this finding, recent biochemical and genetic studies in yeast implicate PHO85, a cyclin-dependent protein kinase homologous to mammalian CDK5, as the sole kinase that phosphorylates Glc8 at the Thr 118 site that is homologous to Thr 72 (14). In addition, in HeLa cells there is another fraction of mitotic I-2 (Thr 72 ) kinase that is resistant to roscovitine and as such is presumably distinct from the CDK family. Further work is required to establish the identity of this mitotic I-2 kinase. FIG. 4. Thr 72 in I-2 is phosphorylated during mitosis. HeLa cells were left unsynchronized (U) or synchronized with nocodazole, and mitotic cells (M) were collected by shake-off. Lysates prepared from these cells were subjected to immunoblotting with an I-2 antibody (lower panel) or with the phospho-Thr 72 I-2 site-specific antibody (upper panel). The ratio of phospho-Thr 72 staining intensity to total I-2 intensity was calculated, and the value for unsynchronized cells was set to unity (histogram). The increase in Thr 72 phosphorylation (28.2 Ϯ 8.1, n ϭ 3) was statistically significant by Student's t test (p Ͻ 0.01).
FIG. 5. The major I-2 Thr 72 kinase activated during mitosis is not GSK3 but a CDK. Panel A, total cell lysates were prepared from unsynchronized (U), mimosine-treated (S), or nocodazole-treated (M) HeLa cells. Identical amounts of protein from these lysates were used in a kinase assay with recombinant I-2 and immunoblotting using the phospho-Thr 72 site-specific antibody. Panel B, quantitation of I-2 Thr 72 phosphorylation with a mitotic cell lysate with carrier (Control, set to 100%), 10 M GSK3 inhibitor (GSK3Inh), or 10 M roscovitine added. The average of three independent experiments is shown with error bars representing the S.E.
Timely changes in phosphorylation of mitotic substrates require precise coordination of protein kinase and phosphatase activities. Genetic evidence with yeast, other fungi, and fruit flies points to a critical role for PP1 in various aspects of mitosis (15)(16)(17)(18)(19). Inhibition of PP1 activity leads to premature mitotic entry (20 -22). In fungi and in mammals, PP1 activity appears particularly important for the metaphase-anaphase transition (16,23,24). However, the mechanisms that regulate PP1 activity during various stages of mitosis remain largely unknown.
Interestingly, we found phosphorylation of I-2 (Thr 72 ) rose and fell during stages of mitosis with a peak at prophase. This suggests a highly dynamic regulation of I-2 and its associated PP1. These results are consistent with prior studies that showed PP1 activity is suppressed at entry into mitosis but required for the metaphase-anaphase transition (24). PP1 activity in mammalian cells also seems to be suppressed during telophase, to promote cell spreading (24). Thus, the cell cycledependent phosphorylation and dephosphorylation of I-2 on Thr 72 may play a role in fluctuations in PP1 activity during the cell division cycle (2, 3, 16, 20 -23, 25). Measuring the activity of any one of the many forms of PP1 present in cells (associated with I-2 or not) is a daunting task that would require physical separation, as well as specific substrates.
The physiological function(s) of I-2 in mammalian cells is not well understood, and the presence of multiple pseudogenes has made gene deletion problematic. By demonstrating that I-2 (Thr 72 ) phosphorylation at mitosis is highly focused at centrosomes, our data at least begin to point to a potential physiological role for this conserved modification. Our own recent evidence showed that I-2 is concentrated at centrosomes and associates in a complex with PP1 and the kinase Nek2. Overexpression of I-2 produced premature centrosome separation, consistent with activation of the Nek2 kinase (26). Nek2 is of specific interest, because it is localized at centrosomes, and prior studies have implicated a role for the NEK2⅐PP1 complex in the coordinate control of the phosphorylation of the associated substrate, cNAP (27). However, this is not the only target for I-2 action in cells, because I-2 also associates with neurabin/ FIG. 6. Immunofluorescent localization of phospho-Thr 72 -I-2 in mitotic HeLa cells. HeLa cells were stained with the phospho-Thr 72 I-2 site-specific antibody plus Hoechst dye, as described under "Experimental Procedures." Antibody was pre-incubated with buffer alone as control (panels A and B), 1 M phospho-Thr 72 I-2 to neutralize specific antibody (panels C and D), or 1 M unphosphorylated I-2 as a negative control (panels E and F). spinophilin, as well as a novel transmembrane Ser/Thr kinase called KPI-2 (28,29). Work of a decade ago (30) documented changes in the levels of I-2 during the cell cycle, with peak expression during mitosis, indicating a possible role for I-2 in mitosis. We speculate that phosphorylation of I-2 at Thr 72 during mitosis plays a role in the control of centrosomes and maybe in other events at this critical stage of mammalian cell division.