Multisite Phosphorylation and the Nuclear Localization of Phosphatase Inhibitor 2-Green Fluorescent Protein Fusion Protein during S Phase of the Cell Growth Cycle*

Human phosphatase inhibitor 2 (Inh2) is a phosphoprotein that complexes with type 1 protein phosphatase, and its expression peaks during S phase and mitosis during the cell cycle. Localization of Inh2 was visualized in HS68 human fibroblasts by fusing Inh2 to green fluorescent protein (GFP). During G1phase, Inh2-GFP was localized in the cytoplasm, and as cells progressed into S phase Inh2-GFP accumulated in the nucleus. Known phosphorylation sites of Inh2 at Thr-72, Ser-86, and Ser-120/121 were each replaced with alanine. None of the mutated Inh2-GFP proteins accumulated in the nucleus during S phase, indicating that all of these phosphorylation sites were required. Mutation of two lysine residues in a putative nuclear localization sequence in Inh2 also prevented the Inh2-GFP fusion protein from accumulating in the nucleus during S phase. Recombinant Inh2 was phosphorylated by kinases in cytosols prepared from G1 and S phase cells. The amount of Inh2 kinase attributed to casein kinase 2, based on inhibition by heparin, increased 2.6-fold from G1 to S phase. In addition, kinases in G1 versus S phase cytosols produced distinct Inh2 phosphopeptides. The results indicate that changes in phosphorylation of Inh2 are involved in intracellular redistribution of the protein during the cell cycle.

Human phosphatase inhibitor-2 (Inh2) 1 is a 23-kDa heatstable protein first identified as a phosphatase inhibitor protein in rabbit skeletal muscle (1). Rabbit Inh2 was purified to homogeneity, (2,3) and the amino acid sequence was determined as 204 residues (4). The cDNA sequences of rabbit skeletal muscle Inh2 (5,6), rabbit liver Inh2 (5), and human Inh2 (7) have been determined. These amino acid sequences are over 90% identical. The GLC8 gene in Saccharomyces cerevisiae encodes a protein with some similarity to Inh2, but only 25% of the amino acid sequence is identical (8 -10).
The function of Inh2 and the role of its phosphorylation in living cells remains unsettled. One proposal is that MgATP-dependent phosphatase is a significant cytoplasmic form of PP1. Results using PP1 expressed in bacteria led to another proposal that Inh2 functions as a molecular "chaperone" for folding newly synthesized PP1 into a biologically active conformation (26,27). On the other hand, the amount of Inh2 protein and heat-stable inhibitory activity against PP1 were found to oscillate during the cell cycle in rat embryo fibroblasts, peaking during S phase and mitosis (28,29), suggesting a possible role for Inh2 in the cell division cycle. A similar cell cycle-dependent change in the amount of the yeast GLC8 protein has been reported (8).
The green fluorescent protein (GFP) of jellyfish Aequorea victoria (30,31) has emerged as a unique tool for examining intracellular phenomena in living cells. Because GFP possesses an intrinsic fluorescence that does not require other cofactors and because the S65T mutant of GFP shows an enhanced brightness compared with the wild type protein, fusion proteins with GFPS65T provide a powerful system to analyze protein expression and distribution in living cells (32)(33)(34). In the present study, we used a Inh2-GFP fusion protein to examine intracellular localization of Inh2 during G 1 and S phases of the cell cycle.

MATERIALS AND METHODS
Cell Culture-Human foreskin fibroblasts (HS68 cells) and SV40transformed African green monkey kidney cells (COS-7) were obtained from ATCC and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) plus 2 mM glutamine. Cell growth was synchronized in serum-free DMEM for 36 -48 h, followed by addition of 10% FCS. Addition of 1-2 mM hydroxyurea was used to arrest cells near G 1 /S. For fluorescence microscopy to detect GFP protein, cells were cultured in phenol red-free DMEM.
Analysis of Growth Synchronization-HS68 cells were seeded at a density of 1.5 ϫ 10 4 cells/35-mm dish. To monitor the induction of DNA synthesis [methyl-3 H]thymidine (Amersham Corp.) was added in pulses of 1 Ci/2 ml of medium for 4 h. At the indicated time points, cells were released from the dish with 200 l of 0.25% trypsin, 0.02% EDTA and treated with 500 l of 1% Triton X-100 followed by addition of 700 l of 10% trichloracetic acid. The precipitate was collected by vacuum filtration on GF/C filters and washed with 5% TCA. The [ 3 H]thymidine in DNA was quantified by liquid scintillation counting.
Distribution of cells in S phase was analyzed by fluorescence-activated cell sorting with ModFit software (Verity Software) using a Power Mac 7600. Cells were pelleted by centrifugation (200 ϫ g for 5 min) and resuspended in 600 l of propidium iodide solution containing 0.1% sodium citrate, 0.3% Nonidet P-40, 100 g of RNase A/ml, and 50 g of propidium iodide/ml and then subjected to FACScan analysis (Becton Dickinson).
We analyzed whether transfected cells expressing GFP proteins entered S phase at the same time as non-transfected cells. Cells growing on a 100-mm dish were transfected and then synchronized in S phase by hydroxyurea block/release. Cells were sorted by FACS Vantage (Becton Dickinson) based on sorting gates set up around GFP-positive and GFP-negative populations; these gates were based on S65TGFP fluorescence (excitation, 488 nm; emission, 530 nm). GFP-positive cells were recovered (purity, Ͼ95%) and then subjected to analysis using propidium iodide staining.
Plasmid Constructs-Vector pNAssCMVGFPS65T was constructed from pNAssCMV (35) (the cDNA encoding GFPS65T was linked downstream of the CMV promoter) and was kindly provided by Dr. Richard Day (University of Virginia). Human Inh2 cDNA was generated using reverse transcription-polymerase chain reaction, in which total RNA was extracted from HS68 cells using the acid-guanidinium thiocyanatephenol-chloroform extraction method (36). Primers used for polymerase chain reaction contained EcoRI sites 5Ј-CGAATTCCAATGGCGGCC TCGACGG-3Ј and 5Ј-TTGAATTCCTGTTGGTCACTTGGAG-3Ј, creating a human Inh2 cDNA fragment lacking the C-terminal seven amino acids. To make a Inh2-GFPS65T fusion gene, the Inh2 cDNA fragment was cloned into EcoRI sites in the pNAssCMVGFPS65T vector, which directs a synthesis of the fusion protein, tagged with GFPS65T at the C terminus of Inh2. The four Inh2 mutants, T72A, S86A, S120/121A, and K143/145A, were generated by megaprimer polymerase chain reaction (37,38) and cloned into EcoRI sites in the pNAssCMVGFPS65T vector.
To generate FLAG-tagged Inh2 containing the DYKDDDDK epitope, the Inh2 cDNA fragment was cloned into the EcoRI site of FLAG-fused pcDNA3 vector (Invitrogen) to create a fusion protein tagged with FLAG at the N terminus of Inh2. To make FLAG-Inh2-GFP, the GFPS65T insert with SV40poly(A) was excised from pNAssCMVGF-PS65T by digestion with EcoRI and XbaI and then cloned into EcoRI and XbaI sites of FLAG-fused pcDNA3 vector. Inh2 cDNA was inserted into the EcoRI site between FLAG and GFPS65T, creating the FLAG-Inh2-GFP fusion gene. All plasmid constructs were confirmed by dideoxy sequencing (39).
Microscopic Analyses of Cells Expressing GFP Fusion Proteins-Expression vectors encoding the Inh2-GFPS65T fusion proteins were transfected into HS68 cells using liposome-mediated gene transfer carried out according to instructions for the LipofectACE product (Life Technologies, Inc.). Cells (5 ϫ 10 3 ) were grown on a 1.5-mm-thick coverglass (18 ϫ 18 mm) in 35-mm culture dishes and maintained for 36 -40 h with 2 ml of phenol red-free and serum-free DMEM to arrest growth. Quiescent cells were treated with 2-5 g of plasmid DNA in 30 l of liposome reagent and then cultured in a 5% CO 2 incubator at 37°C for 6 -8 h. After incubation, cells were washed with PBS, the medium was replaced with phenol red-free DMEM containing 10% FCS, and then living cells were examined at the indicated times using a Nikon Microphot-FXA/SA fluorescence microscope with a fluorescein filter (Chroma).
Nuclear accumulation of GFP fusion proteins was assayed by counting and scoring all of the fluorescent green cells on the coverglass, typically about 80 -100 cells out of 5 ϫ 10 3 . Cells with higher fluorescence intensity in the nucleus compared with the cytoplasm were scored as positive. Cells with higher intensity in the cytoplasm were scored as negative. The fraction of positive cells of the total green fluorescent cells was defined as the nuclear/cytoplasmic index (N/C index). The results were replicated in 5-6 independent experiments, and the data were used to produce mean Ϯ S.D. values that are plotted in the figures. Photographs of individual cells (Fig. 1C) were taken with Kodak Ektachrome 400 color slide film through an oil immersion ϫ 60 objective lens. The slides were scanned by a Nikon LS-3510AF scanner and processed through Adobe Photoshop, version 3.0.
Analysis of Inh2 Phosphorylation in Cells-Transient expression of FLAG-Inh2 or FLAG-Inh2-GFP fusion proteins in COS-7 cells (2-5 ϫ 10 6 ) used transfection in 100-mm dishes with lipofection (5-8 g of pcDNA3 plasmid). Cells were grown in DMEM supplemented with 10% FCS, incubated with 2 mM hydroxyurea for 20 h, and released from hydroxyurea; 2.5 h later, cells were radiolabeled with 0.5 mCi/ml of [ 32 P]orthophosphate (NEN Life Science Products) for 1 h at 37°C in phosphate-free DMEM supplemented with 10% FCS that had been dialyzed against 0.8% NaCl. Cells were washed with PBS, trypsinized from the dish, and collected by centrifugation. The cell pellet was resuspended in 1 ml of cell lysis buffer (10 mM Tris-HCl, pH 7.8, 0.15 M NaCl, 5 mM EDTA, 5 mM EGTA, 1% Nonidet P-40, 1 mM DTT, 1 mM sodium orthovanadate, 1 M microcystin Leu-Arg, 20 mM sodium pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 0.3 mg/ml benzamidine, and 10 g/ml lima bean trypsin inhibitor) and then kept at 4°C for 45 min. After the lysate was centrifuged at 13,000 ϫ g for 10 min, the supernatant was used as cell extract. Fusion proteins were immunoprecipitated with anti-FLAG M2 affinity gel (Eastman Kodak Co.). After resolution of the immunoprecipitates by SDS-PAGE (10% gel), the proteins were transferred to Immobilon P membrane (Millipore). The phosphorylated FLAG-Inh2 and FLAG-Inh2-GFP fusion proteins were detected by autoradiography. Western blotting was performed using anti-FLAG M2 monoclonal antibody, followed by the use of the Renaissance™ detection system (NEN Life Science Products).
Phosphorylation and Peptide Mapping of Recombinant Inh2-Recombinant rabbit Inh2 (rInh2) was expressed in bacteria and purified to homogeneity (40). Samples of 2 g of rInh2 were incubated for 30 -60 min at 30°C in 60 l of cytosol prepared from HS68 cells (5 g of protein), which was made in buffer including 50 mM Hepes, pH 7.5, 10 g/ml aprotinin, 10 g/ml leupeptin, 10 mM MgCl 2 , 10 mM ␤-glycerophosphate, 1 M microcystin LR, 1 mM sodium orthovanadate, 1 mM DTT, and 1 mM phenylmethylsulfonyl fluoride using mechanical disruption generated by a Dounce homogenizer. Phosphorylation reaction was performed in the presence of 5 Ci of [␥ 32 P]ATP, 200 M ATP, and 10 mM MgCl 2 . After phosphorylation, the reaction mixture was analyzed by SDS-PAGE followed by autoradiography. For two-dimensional phosphopeptide mapping, the corresponding bands were excised and digested with trypsin (0.2 mg/ml) overnight at 30°C, and then the tryptic phosphopeptides were loaded onto thin layer cellulose plates (Selecto Scientific flexible TLC plate) and resolved first by electrophoresis in a pH 1.9 buffer and then by ascending chromatography in an isobutyric acid solvent (41).

Intracellular Localization of Inh2-GFP during G 1 and S
Phase-The intracellular localization of a Inh2-GFP fusion protein was examined in HS68 cells made quiescent by serum deprivation for 36 -48 h and then stimulated into synchronous growth by serum refeeding. To define the time of S phase, cells were pulse-labeled for 4-h intervals with [ 3 H]thymidine. Fig.  1A shows a sharp peak of DNA synthesis at 21 h. As an alternative method for synchronization of cell growth, we used hydroxyurea to block cells near G 1 /S. After washout of hydroxyurea, incorporation of [ 3 H]thymidine could be detected as early as 30 min, and it reached a peak at 2 h ( Fig. 2A). At the peak of S phase, more than 70% of the cells were immunostained for incorporation of bromodeoxyuridine using either method of synchronization (data not shown). The results of both analyses were consistent and established two alternative methods for the synchronization of HS68 cells in S phase.
GFP fluorescence was detected in approximately 5% of cells transfected with an Inh2-GFP expression vector. Transfection efficiency for HS68 cells was limited by the serum starvation protocol used to provide synchronous growth of the cells. After various trials, it was found that transfection was best done during the time period just prior to readdition of serum. After serum stimulation, Inh2-GFP protein was predominantly cytosolic up to 12 h, corresponding to the G 1 phase of growth (Fig.  1C, panel a). However, Inh2-GFP protein accumulated in the nucleus as these cells progressed from G 1 to S phase (Fig. 1C,  panel b). At 21 h, when 70% of the cells were in S phase as shown by bromodeoxyuridine staining, nuclear translocation-positive cells reached 68 Ϯ 15%, expressed as a N/C index of 0.68 (Fig. 1B, closed circles). Most of the cells transfected with GFP alone as a control exhibited green fluorescence uniformly throughout the cell (Fig. 1C, panel c), and these cells had a low N/C index during S phase (Fig. 1B, open circles).
In separate experiments, at 2.5 h after hydroxyurea release (that is, during S phase ( Fig. 2A)), Inh2-GFP protein accumulated in the nucleus, again expressed as a N/C index Ͼ 0.6 ( Fig.  2B, closed circles), whereas control cells expressing GFP itself did not show accumulation of GFP in the nucleus (Fig. 2B, open  circles). The results show that the Inh2-GFP protein was predominantly cytoplasmic during G 1 phase and that it accumulated in the nucleus during S phase. Both methods for synchro- nization of cell growth gave the same percentage of green fluorescent cells exhibiting nuclear localization of the Inh2-GFP fusion protein (N/C index), and this agreed exactly with the percentage of cells in the entire culture that were positive for bromodeoxyuridine staining.
As an additional control experiment, we examined whether transient expression of Inh2-GFP or GFP itself affected entry into S phase. FACScan analysis by propidium iodide staining showed the distribution of mock-transfected cells (Fig. 3A) during hydroxyurea blockade (broken line) and 2.5 h after release by washout (solid line). For this analysis, cells expressing GFP (Fig. 3B) or Inh2-GFP (Fig. 3C) were sorted by GFP fluorescence and then analyzed by propidium iodide staining. The results demonstrate that expression of Inh2-GFP or GFP itself did not affect cell cycle progression into S phase after hydroxyurea blockade.
Mutations to Define Requirements for Nuclear Accumulation of Inh2 during S Phase-The known phosphorylation sites in Inh2 were mutated to alanines using megaprimer polymerase chain reaction, creating T72A-Inh2, S86A-Inh2, and S120A/ 121A-Inh2, which were expressed as GFP fusion proteins in HS68 cells. In contrast to the wild type Inh2-GFP (Fig. 4,  column 2), none of the point mutants accumulated in the nucleus during S phase (Fig. 4, columns 3 -5). The N/C index Ͻ0.2 was the same as that for GFP itself (Fig. 4, column 1). The total number of green fluorescent cells was the same for GFP, for mutated Inh2-GFP, and for wild type Inh2-GFP, and the results were replicated in three to five independent experiments. Thus, mutation of any one of the known phosphorylation sites in Inh2 altered localization of the Inh2-GFP fusion protein in living cells during S phase.
Inh2 has two clusters of basic amino acid residues within the sequence 134 -147 (REKKRQFEMKRKLH. The lysine residues at positions 143 and 145 were mutated to alanines, giving a K143A/K145A mutant of Inh2 (Inh2-KK/AA). rInh2 with lysines 143 and 145 mutated to non-basic amino acids had inhibitory specific activity toward PP1 that was identical to the wild type Inh2 protein (data not shown). Inh2-KK/AA-GFP fusion protein expressed in HS68 cells did not become concentrated within the nucleus; instead, it localized predominantly in the cytoplasm during both G 1 and S phases (Fig. 4, column 6). There was no difference in the number of cells expressing Inh2-KK/AA-GFP, wild type Inh2-GFP, or GFP itself. Mutation Phosphorylation of Inh2 and Inh2-GFP in Living Cells-Inh2 is phosphorylated by multiple protein kinases, providing a potential mechanism for regulating its localization. Antibodies against Inh2 or GFP were not effective at immunoprecipitation in our hands, preventing analysis of phosphorylation by direct recovery of endogenous Inh2 or transiently expressed Inh2-GFP. As an alternative, to produce enough protein for analysis of phosphorylation, COS-7 cells were transfected with plasmids encoding epitope-tagged FLAG-GFP, FLAG-Inh2, or FLAG-Inh2-GFP. Cells were synchronized into S phase by hydroxyurea block/release and metabolically labeled with 32 P. As shown in Fig. 5A, FLAG-Inh2-GFP (lane 1), FLAG-GFP (lane 2), and FLAG-Inh2 (lane 3) were immunoprecipitated as 70-, 31-, and 41-kDa proteins, respectively, based on anti-FLAG immunoblotting. Both FLAG-Inh2-GFP and FLAG-Inh2 proteins were 32 P-labeled (Fig. 5B, lanes 1 and 3), whereas the FLAG-GFP protein was not. FLAG-Inh2-GFP and FLAG-Inh2 were phosphorylated to approximately the same specific radioactivity (FLAG-Inh2-GFP/FLAG-Inh2 ϭ 0.8), calculated by densitometry. We concluded that the sites in the Inh2 portion, not in the GFP portion, of the fusion protein were phosphorylated in living cells. Presumably, the same sites were phosphorylated in FLAG-Inh2-GFP and FLAG-Inh2. Acid hydrolysis and phosphoamino acid analysis revealed only 32 P-labeled Ser in these proteins (data not shown).
Phosphorylation of Thr-72 in Inh2 in Living Cells during S Phase-To provide evidence that Thr-72, which was required for nuclear accumulation during S phase (Fig. 4), was phosphorylated in living cells, COS-7 cells were transfected with plasmids for FLAG-Inh2, FLAG-T72A-Inh2, FLAG-S86A⅐S120A⅐ S121A-Inh2 (triple mutant), and FLAG-T72A⅐S86A⅐S120A⅐ S121A-Inh2 (quadruple mutant). Cells were synchronized into S phase and metabolically labeled with 32 PO 4 as described above. Immunoprecipitates from these cells using anti-FLAG M2 antibody were subjected to SDS-PAGE, Coomassie Blue staining, and autoradiography. As shown in Fig. 6A, FLAG-Inh2 and FLAG-T72A-Inh2 were robustly phosphorylated (lanes 2 and 3). In comparison, the triple mutant FLAG-S86A⅐S120A⅐S121A-Inh2 was also 32 P-labeled, but at a much lower level (lane 4). The quadruple mutant FLAG-T72A⅐S86A⅐ S120A⅐S121A-Inh2 had 32 P labeling even lower than that of the triple mutant FLAG-S86A⅐S120A⅐S121A-Inh2. This difference in 32 P labeling between the triple and quadruple mutant proteins was attributed to phosphorylation of T72. Interestingly, even the quadruple mutant Inh2 was radiolabeled (Fig. 6A,  lane 5), showing that there are other sites in Inh2 phosphorylated in these cells. Coomassie Blue staining was used to measure the recovery of various FLAG-Inh2 proteins (Fig. 6B). Although the same amount of antibody was used for each sample, as seen from staining for the heavy chain at the top of each lane, there were different amounts of FLAG-Inh2 recovered in each sample. In the Coomassie-stained gel, differences in electrophoretic mobility of the various Inh2 proteins were evident. The single mutation of T72 resulted in increase in mobility relative to wild type (Fig. 6B, lane 2 versus lane 3). The triple and quadruple mutants migrated in the same way (Fig. 6B,  lanes 4 and 5), and this was different from the wild type and T72A forms of Inh2. The mobility in SDS-PAGE is another indication of phosphorylation of these sites in Inh2. The results are further evidence for multisite phosphorylation of Inh2 at T72, S86, S120, and S121 in living cells during S phase.
Differential Phosphorylation of Recombinant Inh2 by Kinases in the Cytosols from G 1 and S Phase Cells-Inh2 kinase activity in cytosols prepared from G 1 versus S phase cells was assayed with rInh2 as an exogenous substrate. As shown in Fig. 7A, rInh2 was phosphorylated by kinases present in the cytosol of HS68 fibroblasts. Cytosols prepared from S phase cells (Fig. 7A, lane 4) had a specific activity ( 32 P labeling/g of cytosol protein) that was 1.6 times higher than that of cytosols from G 1 phase cells (Fig. 7A, lane 1). Rabbit Inh2 is known to be phosphorylated at S86, S120, and S121 by CK2, so we added 1 g of heparin, an inhibitor of CK2 (42), to the 60-l reaction mixtures. At this dosage, heparin strongly inhibited the phosphorylation of rInh2 (Fig. 7A, compare lanes 1 versus 2 and  lanes 4 versus 5), reducing labeling by 40% using G 1 or by 60% using S phase cytosol as a source of kinase (quantitation done by scintillation counting of the excised proteins). We calculated a 2.6-fold increase from G 1 to S phase in the heparin-sensitive Inh2 kinase activity, attributed to CK2.
To examine the sites of phosphorylation in rInh2, the radiolabeled samples shown in Fig. 7A, lanes 1 and 4, were excised from the gel and subjected to two-dimensional tryptic phosphopeptide mapping as described under "Materials and Meth- ods." As shown in Fig. 7B, panel a, there was one predominant tryptic phosphopeptide (peptide 1) plus several other phosphopeptides (peptides 2, 4, 5, and 6) recovered from rInh2 after labeling with G 1 cytosol as a source of kinase. In contrast, rInh2 phosphorylated using cytosol from S phase cells (Fig. 7B,  panel b) had only a trace of phosphopeptide 1 and instead was labeled with about equal intensity in five other phosphopeptides (peptides 2-6). Phosphopeptides 2, 4, and 5 matched those obtained from rInh2 phosphorylated with purified CK2 (Fig. 5B, panel c). Therefore, in S phase cells, CK2 accounted for most of the rInh2 kinase activity. There were two notable differences between the G 1 and S phase phosphopeptide patterns. First, there was a kinase in G 1 phase cytosols that produced phosphopeptide 1. This kinase and the site in Inh2 both remain unidentified. Second, both phosphopeptides 3 and 6 were produced by a cytosolic kinase(s) from S phase cells but not by purified CK2. These analyses show that (a) there are multiple Inh2 kinases in cytosols; (b) the level of heparinsensitive Inh2 kinase activity increased from G 1 to S phase; and (c) the sites phosphorylated in Inh2 changed between G 1 and S phase. DISCUSSION This study examined the intracellular localization of Inh2 in living cells during G 1 and S phases of the cell cycle using an Inh2-GFP fusion protein. Accumulation of Inh2-GFP from the cytoplasm into the nucleus occurred during S phase. GFP itself, expressed as a control, remained uniformly distributed in the cell throughout G 1 and S phase. Mutations to eliminate phosphorylation of Inh2 at either Thr-72, or Ser-86, or Ser-120 and Ser-121 abolished nuclear localization during S phase. Differences in amino acid residues at positions immediately adjacent to a phosphorylation site (e.g. C85 in human versus Y85 in rabbit) do not affect nuclear import of Inh2, 2 supporting the idea that the multisite phosphorylation of Inh2 is required for its cell cycle-dependent localization.
Phosphorylation of Inh2 is cooperative, or "synergistic," because reaction with CK2 at Ser-86 potentiated phosphorylation of Thr-72 by glycogen synthase kinase 3, making phosphorylation of Thr-72 highly sensitive to the extent of prior phosphorylation by CK2. Different proline-directed protein kinases, namely glycogen synthase kinase 3, CDK2, and mitogen-activated protein kinase, can phosphorylate Thr-72. Therefore, it is possible that mutation of Ser-86 prevented nuclear localization indirectly by interfering with efficient phosphorylation of Thr-72. We present evidence that Thr-72 was phosphorylated in living cells even in the triple mutant S(86/120/121)A. One might suspect, therefore, that the failure of the S86A mutant of Inh2-GFP to localize in the nucleus indicates a function for this site beyond simply promoting phosphorylation of Thr-72. In addition, the CK2 phosphorylation sites at S120/121 themselves might facilitate nuclear accumulation, parallel to the case of SV40 T antigen (see below). It seems that all of these phosphorylation sites in Inh2 contribute to and are required for nuclear localization during S phase.
Cell cycle-dependent nuclear localization of Inh2 also depended on a basic sequence resembling a NLS. Because the molecular size of Inh2-GFP is 63 kDa, it seems unlikely that the distribution of this fusion protein was a result of simple diffusion between cytoplasm and nucleus (43,44). At residues 134 -147 in Inh2, there is a sequence of two clusters of basic amino acids in a region of Inh2 that is related in sequence to c-fos (5,22). Mutation of lysines 143 and 145 abolished nuclear localization of Inh2 during S phase. As a transcription factor, c-fos surely is imported into the nucleus, but its NLS has not been functionally defined. Maybe Inh2 and c-fos are similar in sequence in the region used as their NLS.
There are several examples of nuclear import of proteins mediated by a NLS and regulated by phosphorylation. Undoubtedly the best known example is that of SV40 T antigen, which has a NLS with the sequence PKKKRKV (45). When this sequence is fused or conjugated to other proteins, it results in their accumulation in the nucleus. However, it has not been well appreciated that the rate of nuclear import mediated by this NLS is controlled by phosphorylation of two serine sites approximately 20 residues to the N-terminal side (46). Phosphorylation of these serines by CK2 accelerated by 30-fold the rate of nuclear import of a SV40 T antigen-␤ galactosidase fusion protein (47). The spacing of the dual CK2 Ser phosphorylation sites and basic residues of the putative NLS in Inh2 matches that in large T antigen, so similar mechanisms may govern localization of these proteins. Phosphorylation of cyclin B also results in nuclear localization, which is critical for its function at G 2 /M in the cell cycle (48). The sites in cyclin B under cell cycle control are thought to be phosphorylated by CK2 and mitogen-activated protein kinase, an interesting par- allel to Inh2. Other examples of phosphorylation-regulated nuclear import are the transcription factor SWI5 of S. cerevisiae and the v-Jun oncoprotein. Both proteins are localized in nuclei in their unphosphorylated forms. Phosphorylation of SWI5 by CDC28 kinase results in its displacement into the cytoplasm (49). Phosphorylation of v-Jun also is correlated to loss of nuclear localization (50). Thus, phosphorylation can be employed to enhance or to eliminate nuclear localization.
The Inh2 kinase activity in cytosols increased between G 1 and S phase. Most of this activity was potently inhibited by heparin, a characteristic of CK2. The results fit together with the previous data showing that CK2 activity oscillates during the cell cycle, peaking at S phase (51). Inh2 phosphorylated by these S phase cytosols or by purified CK2 gave several of the same major phosphopeptides. This is consistent with Ser-86, Ser-120, and Ser-121, the sites in Inh2 phosphorylated by CK2, being phosphorylated during S phase. Peptides 3 and 6 were not produced by purified CK2, but were prominent in Inh2 phosphorylated by cytosols from S phase cells. The PhosPep-Sort2 program (GCG Computer Group) predicts this map location for phosphopeptides containing Thr-72, and because peptide 6 was produced by both G 1 and S phase cytosols, we suspect that Thr-72 is phosphorylated during throughout G 1 and S phases. A remaining issue is the identity of the site in phosphopeptide 1, which was the major site phosphorylated by G 1 phase cytosol. The change in localization of Inh2 at G 1 /S may involve inactivation of a kinase and the dephosphorylation of the site in phosphopeptide 1. Overall, the results contribute to the concept that CK2 is part of a signaling system that regulates the nuclear import of multiple proteins at G 1 /S (reviewed in Refs. 52 and 53).
What might be the biological function of the nuclear accumulation of Inh2 at G 1 /S? One possibility is that it inhibits PP1, which dephosphorylates the retinoblastoma protein Rb in the nucleus throughout the G 1 period (54). Recent evidence shows Rb-dependent cell cycle arrest at G 1 /S in cells loaded with a PP1 mutant that lacks an inhibitory phosphorylation site at T320 (55). Inhibition of nuclear PP1 by an influx of Inh2 plus CDK phosphorylation of T320 could form a double shut-off system for PP1. Inhibition of PP1 coupled to activation of cyclin D:CDK4 would be a potent means to produce abrupt phosphorylation of Rb, which is recognized as a key step for entry into S phase (reviewed in Ref. 56).