Regulation of Casein Kinase I ε and Casein Kinase I δ by an in Vivo Futile Phosphorylation Cycle*

Casein kinase I δ (CKIδ) and casein kinase I ε (CKIε) have been implicated in the response to DNA damage, but the understanding of how these kinases are regulated remains incomplete. In vitro, these kinases rapidly autophosphorylate, predominantly on their carboxyl-terminal extensions, and this autophosphorylation markedly inhibits kinase activity (Cegielska, A., Gietzen, K. F., Rivers, A., and Virshup, D. M. (1998) J. Biol. Chem. 273, 1357–1364). However, we now report that while these kinases are able to autophosphorylatein vivo, they are actively maintained in the dephosphorylated, active state by cellular protein phosphatases. Treatment of cells with the cell-permeable serine/threonine phosphatase inhibitors okadaic acid or calyculin A leads to rapid increases in kinase intramolecular autophosphorylation. Since CKI autophosphorylation decreases kinase activity, this dynamic autophosphorylation/dephosphorylation cycle provides a mechanism for kinase regulation in vivo.

The casein kinase I (CKI) 1 gene family contains two major subgroups that regulate cytoplasmic and nuclear processes, respectively. The nuclear family appears involved in the response to DNA damage in mammals and yeast, whereas the cytoplasmic members are involved in membrane structure and bud morphogenesis (1)(2)(3)(4)(5). The founding member of the nuclear family, HRR25, was first cloned in a screen for budding yeast mutants sensitive to DNA double strand breaks (3). In Schizosaccharomyces pombe, the essential gene pair hhp1 ϩ and hhp2 ϩ performs similar functions (4). The mammalian homologs of these genes are the closely related CKI⑀ and CKI␦; they encode monomeric protein kinases that are 97% identical to each other over the kinase domain and 53% identical over their 124 amino acid carboxyl-terminal extensions (6,7). In fact, the human CKI⑀ gene partially complements Saccharomyces cerevisiae with a deletion of the HRR25 gene (7). A number of the substrates identified to date for the CKI family also support a role for the kinase in regulation of DNA repair and replication. For example, phosphorylation of SV40 large T antigen blocks its ability to unwind the SV40 origin of replica-tion (8 -11). CKI also phosphorylates several isoforms of the rad24 and rad25-related 14 -3-3 proteins, regulating their ability to bind to substrate proteins (12). Purified CKI⑀ can phosphorylate the carboxyl terminus of IB␣, potentially regulating its degradation rate (10,13). Hrr25 protein binds to and phosphorylates the yeast transcription factor Swi6 (14). Most recently, CKI⑀ and CKI␦ have been identified as the cellular kinases that constitutively phosphorylate the extreme amino terminus of p53 (5).
CKI genes appear widely and constitutively expressed, but despite the progress in defining a role for CKI isoforms in the cell, a detailed understanding of the regulation of CKI activity is lacking. Several lines of evidence suggest the carboxyl-terminal tails of CKI␦ and CKI⑀ contain an important autoregulatory domain. First, CKI⑀, with a 124-amino acid carboxylterminal extension, complements hrr25⌬ yeast, while CKI␣, with a 24-amino acid extension, does not (7). CKI␦ and CKI⑀ both display a significant increase in specific activity when the carboxyl terminus is removed (10,15). More strikingly, CKI⑀ rapidly autophosphorylates in vitro in an intramolecular reaction that substantially inhibits kinase activity toward a number of protein substrates. CKI⑀, but not a mutant lacking the carboxyl terminus, can then be activated up to 20-fold by dephosphorylation (10). These studies suggest that changes in the phosphorylation state of CKI␦ and CKI⑀ carboxyl termini in vivo may be an important mechanism for regulation of the kinase activity.
In the present study, we investigated the autoregulatory autophosphorylation of CKI⑀ and CKI␦ in vivo. Unexpectedly, we found that in cultured cells and in tissues, these kinases are actively maintained in the dephosphorylated state. While the kinases are indeed able to rapidly autophosphorylate in vivo as well as in vitro, they are maintained in the dephosphorylated state in vivo by rapid constitutive dephosphorylation by cellular protein phosphatases sensitive to okadaic acid and calyculin A. These results indicate that CKI⑀ and CKI␦ utilize ATP in a futile cycle of autophosphorylation and dephosphorylation. Stimuli that regulate the relevant phosphatases may regulate the activity of these kinases in the cell.

EXPERIMENTAL PROCEDURES
Recombinant histidine tagged and untagged CKI⑀ were produced in Escherichia coli as described (7,10). The monoclonal antibody directed against the CKI⑀ carboxyl-terminal domain, CKI⑀ mAb, was purchased from Transduction Laboratories, Lexington, KY. This antibody preferentially recognizes CKI⑀ although it cross-reacts to a lesser extent with CKI␦. The specific anti-CKI␦ monoclonal antibody 128A was the generous gift of M. Hoekstra, Icos Corp., Bothell, WA. The polyclonal antiserum UT31, raised against the amino terminus of CKI⑀, has been previously described (10). Immunoblots were visualized with the enhanced chemiluminescence (ECL) kit from Amersham Pharmacia Biotech.
Cell Treatments with Phosphatase Inhibitors-NIH 3T3 mouse fibroblasts were plated in 35-mm round dishes and grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum at 37°C in a 5% CO 2 incubator. At 70% confluence, cells were treated with calyculin A or okadaic acid in Me 2 SO at indicated times and concentrations. Me 2 SO was added as control in equal amounts to all untreated cells. At the indicated times, the growth medium was removed, and the cells were lysed with 1ϫ SDS-polyacrylamide gel electrophoresis (PAGE) loading sample buffer (10% glycerol, 50 mM Tris, pH 6.8, 2% SDS, 3% ␤-mercaptoethanol, and 0.1% bromphenol blue), boiled for 3 min, and separated by SDS-PAGE. Proteins were electrophoretically transferred to nitrocellulose and immunoblotted with the indicated antibodies.
Phosphorylation and Dephosphorylation Reactions-HeLa extracts were prepared as described (16). In vitro dephosphorylation or phosphorylation reactions contained 25 g of HeLa extract with 4 mM adenosine triphosphate, in the presence or absence of phosphatase inhibitors (100 nM calyculin A, 250 nM okadaic acid, and 50 mM ␤-glycerol phosphate), where indicated, in 30 mM Hepes, pH 7.5, and 7 mM magnesium chloride and were incubated at 37°C for 30 min. Phosphatase reactions contained 200 ng of protein phosphatase 2A catalytic subunit (PP2A c ).
Rat Tissues-The indicated rat organs were excised, snap-frozen in liquid nitrogen, and then weighed and ground with a chilled mortar and pestle. Tissues were homogenized with a Polytron (Brinkmann) using 3 ml/gm tissue of 50 mM Tris, pH 7.5, 150 mM sodium chloride, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 3 g/ml leupeptin, and 3 g/ml pepstatin, or in a separate experiment, by addition of 1ϫ SDS-PAGE sample buffer. Samples were clarified by centrifugation at 27,000 ϫ g for 30 min at 4°C, and the supernatants were analyzed by SDS-PAGE and immunoblotting.
Transfections-COS-1 (SV40-transformed monkey kidney) cells were grown in DMEM plus 10% supplemented calf serum (Life Technologies, Inc. and Hyclone, respectively). 293 cells (human embryonic kidney cells) were grown in DMEM plus 10% fetal bovine serum (Hyclone). Cells at 50 -90% confluence were transfected with 2 g of plasmid DNA mixed with 6 l of LipofectAMINE reagent (Life Technologies) following the instructions of the manufacturer. Transfection efficiencies were routinely 40 -50% for COS-1 and 293 cells.
Subcellular Fractionation-COS-1 cells were lysed in hypotonic lysis buffer (10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 1 mM EDTA, 0.5 mM dithiothreitol, 1 g/l Ϫ1 leupeptin, 1 g/l Ϫ1 pepstatin, and 1 mM PMSF) 48 h after transfection. Soluble material was clarified by centrifugation at 1,000 ϫ g for 10 min. To the resulting supernatant, a 1/10 volume of 0.3 M Hepes, pH 7.9, 1.4 M KCl, and 30 mM MgCl 2 was added, and the mixture was centrifuged for 15 min at 14,000 ϫ g. The resulting pellet was designated the membrane fraction and the supernatant was the cytoplasmic fraction. The pellet from the 1000 ϫ g centrifugation was extracted with high salt buffer containing 20 mM Hepes, pH 7.9, 420 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 25% glycerol, 1 g/l Ϫ1 leupeptin, 1 g/l Ϫ1 pepstatin, and 1 mM PMSF. The pellet was mixed slowly at 4°C for 30 min and then clarified by centrifugation for 15 min at 14,000 ϫ g. The resulting supernatant is referred to as the nuclear fraction.
Metabolic Labeling-Transiently transfected 293 cells were metabolically labeled for 5 h in 5% dialyzed calf serum (Life Sciences), 2 mCi ml Ϫ1 H 3 32 PO 4 (NEN Life Science Products), and phosphate-free DMEM (Life Sciences) starting at 36 h post-transfection. Calyculin A was not added or added to the transiently transfected cultures at a final concentration of 50 nM during the last 30 min of metabolic labeling. Cultures were harvested by lysis in RIPA buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 1 g/l Ϫ1 leupeptin, 1 g/l Ϫ1 pepstatin, 0.1 mM PMSF, 5 mM Na 3 VO 4 , 20 mM NaF, 250 nM okadaic acid, 20 mM ␤-glycerol phosphate) and clarified by centrifugation at 14,000 ϫ g for 30 min. Soluble extracts containing HA-tagged proteins were immunoprecipitated with 12CA5 mAb and protein Aagarose and analzyed by SDS-PAGE on a 9% gel. Results were visualized by PhosphorImager (Molecular Dynamics).

RESULTS
The phosphorylation state of CKI⑀ was initially assessed by following the shift in electrophoretic mobility of the kinase after SDS-PAGE and immunoblot analysis. As shown in Fig.  1A, recombinant autophosphorylated CKI⑀ (lane 2) displays a marked decrease in electrophoretic mobility compared with PP2A c -treated CKI⑀ (lane 1). When log-phase 3T3 cells were lysed in SDS and the mobility of endogenous CKI⑀ was examined by immunoblot analysis, we likewise found that the immunoreactive kinase was predominantly in the high mobility, dephosphorylated state (Fig. 1A, lanes 3 and 8).
In a recent study, we demonstrated that in vitro purified CKI⑀ is capable of rapid autophosphorylation, with a marked reduction in electrophoretic mobility within 5 min of the addition of ATP (10). How is CKI⑀ maintained in the dephosphorylated state in vivo? Two potential mechanisms are, first, the kinase may be unable to autophosphorylate in vivo or, second, it may be dephosphorylated more rapidly than it autophosphorylates. To evaluate the second possibility, that there are cellular phosphatases that actively dephosphorylate CKI⑀, we treated 3T3 cells with the cell-permeable serine/threonine phosphatase inhibitors okadaic acid and calyculin A and determined their effect on the phosphorylation state of the kinase. As shown in Fig. 1A, lanes 3-13, treatment of cells with either okadaic acid or calyculin A leads to a dose-dependent change in CKI⑀ electrophoretic mobility that reflects a change in phosphorylation. This effect was quite rapid; as seen in Fig. 1B, lane 6, exposure of cells to 100 nM calyculin A for as little as 15 min led to a marked increase in kinase phosphorylation. Note, too, that the effect is not antibody-dependent since the immunoblot in Fig. 1A was probed with anti-CKI⑀ mAb directed against the carboxyl terminus of CKI⑀, and in Fig. 1B, it was probed with UT31, a rabbit polyclonal antibody that recognizes the amino terminus of CKI⑀ and CKI␦ (10). While the CKI⑀ mAb may also recognize CKI␦, the two kinases appear regulated in a coordinate manner since all the immunoreactive kinase behaved similarly both in the absence and presence of phosphatase inhibitors.
To address the question of whether the in vivo dephosphorylation of the kinases was restricted to transformed, log-phase tissue culture cells or was a general phenomenon, the tissue expression pattern of CKI was first investigated (Fig. 2A). Northern blot analysis indicates several forms of CKI, including CKI␣, CKI␦, and CKI⑀ mRNA, are widely and similarly expressed in all human tissues examined, with the highest levels in muscle and brain ( Fig. 2A). Protein extracts were then prepared from flash-frozen rat tissues and analyzed by SDS-PAGE and immunoblotting. As seen in Fig. 2B, CKI⑀ in all rat tissues examined was predominantly in the high mobility, dephosphorylated form. Interestingly, the ratio of CKI mRNA to protein varies widely between tissue (e.g. compare the ratio of mRNA to immunoreactive protein in liver and heart), suggesting additional post-translational regulation. Trace amounts of lower mobility immunoreactive species were seen in several tissues after longer exposure times (data not shown), suggesting some turnover of more highly phosphorylated kinase. The results suggest CKI⑀ dephosphorylation is a widespread phenomenon, and is not limited to tissue culture or transformed cells.
To determine whether CKI was dephosphorylated by nuclear or cytoplasmic phosphatases, the subcellular localization of CKI⑀ (and as a control, CKI␣2) was investigated. COS-1 cells were transiently transfected with hemagglutinin epitopetagged CKI␣2 and CKI⑀ expression constructs. 48 h after transfection, cells were lysed and membrane, soluble, and nuclear fractions were prepared by differential centrifugation as described (see "Experimental Procedures"). The pattern of crossreacting material present in the fractions from untransfected cells (Fig. 2C, lanes 1-3) confirms that the fractions contain distinct proteins. CKI␣2 remained predominantly in the soluble fraction (Fig. 2C, lanes 4 -6), while high levels of CKI⑀ were seen in membrane, soluble, and nuclear fractions (Fig. 2C,  lanes 7-9). A parallel experiment done with kinase-inactive forms of CKI␣2 and CKI⑀ showed an identical pattern of subcellular distribution (data not shown), suggesting that kinase activity is not required for CKI⑀ nuclear localization.
To further characterize the role of cellular phosphatases that catalyze the rapid turnover of phosphoryl groups on CKI⑀, the fate of the kinase in cell extracts was examined. As shown in Fig. 3, recombinant histidine-tagged CKI⑀ fully autophosphorylates when incubated with ATP in the absence of cell extracts (Fig. 3A, lanes 1  and 2). However, when the same amount of kinase is incubated with 25 g of HeLa extract, a small increase in kinase mobility is seen (compare Fig. 3A, lane 1 with lanes 3 and 4), indicating removal of residual phosphoryl groups. If the protein phosphatases present in the HeLa extract are inhibited, then the added CKI⑀ rapidly autophosphorylates. (Fig. 3A, lane 5). Similarly, when HeLa extracts are incubated without added recombinant CKI⑀ with or without ATP (Fig. 3A, lanes 6-9), no change in endogenous CKI⑀ electrophoretic mobility is seen unless protein phosphatase inhibitors are first added (compare Fig. 3A, lanes 6 and 7 with lanes  8 and 9). Note that even in the absence of added ATP, sufficient ATP is present in the cell extract to permit autophosphorylation of endogenous CKI⑀ once phosphatase inhibitors are added (Fig. 3A,  lane 8). The HeLa phosphatases are very efficient at removing phosphates from CKI⑀; Fig. 3B demonstrates that when histidinetagged CKI⑀, autophosphorylated in the presence of [␥-32 P]ATP, is added back to HeLa extract then 50% of the 32 P-labeled phosphate groups are removed in less than 1 min and 90% are removed by 10 min. Note that fully dephosphorylated histidine-tagged CKI⑀ has an electrophoretic mobility of approximately 50 kDa and is not visible in the autoradiograph. This decrease in the 32 P-labeled CKI⑀ is due to dephosphorylation and not proteolysis, as incubation of CKI⑀ in HeLa extract for as long as 30 min has no effect on protein levels (Fig. 3A). This data is consistent with the hypothesis that casein kinase I ⑀ autophosphorylation is ongoing both in vivo and in extracts but is rapidly reversed in vivo and in cell extracts by okadaic acid and calyculin A-sensitive phosphatases.
In order to confirm that CKI⑀ phosphorylation in cells and extracts is indeed due to an intramolecular reaction, rather than phosphorylation in trans by other forms of CKI or other cellular kinases, recombinant wild-type CKI⑀ or a kinase-inactive CKI⑀ mutant (K38R) (7) was incubated with HeLa extract, and its phosphorylation state was assessed by immunoblotting (Fig. 4). While wild-type CKI⑀ readily autophosphorylated Fig.  4, lanes 1-4), the kinase-deficient mutant K38R showed no change in electrophoretic mobility despite a 30-min incubation in the presence of ATP and phosphatase inhibitors (lanes 5-8). This result strongly suggests that CKI⑀ phosphorylation is due to intramolecular autophosphorylation. To confirm this result in vivo, a hemagglutinin-tagged kinase-dead CKI⑀ was transiently expressed in 3T3 cells. No change in electrophoretic mobility (i.e. no significant phosphorylation) of the kinase-dead CKI⑀ occurred even after treatment of the transfected cells with calyculin A (data not shown). These results demonstrate that CKI⑀ autophosphorylates in an intramolecular manner, and additionally suggests that no other cellular kinases present in HeLa extracts phosphorylate CKI⑀ in a manner that contributes to changes in electrophoretic mobility.
Two closely related forms of CKI have been identified in mammals: CKI⑀ and CKI␦. These kinases differ by only eight residues in the kinase domain and are about 53% identical over the 124-amino acid carboxyl-terminal regulatory domain. Since the monoclonal antibody directed against the CKI⑀ tail cross-reacts with CKI␦, it was important to examine whether CKI␦ was specifically maintained in a high mobility, dephosphorylated state in vivo. 3T3 cells were therefore treated with calyculin A or okadaic acid, and cellular extracts were probed with monoclonal antibodies against CKI⑀ and CKI␦. As shown in Fig. 5, CKI␦ showed a similar marked decrease in electrophoretic mobility upon phosphatase inhibitor treatment. Interestingly, CKI␦ displayed a more rapidly and uniform change in electrophoretic mobility, while CKI⑀ appeared to decrease in intensity. While this decrease in CKI⑀ immunoreactivity could be due to phosphorylation-regulated proteolysis, we found no change in signal intensity when cells were pretreated with proteosome or other protease inhibitors (data not shown). The fact that the two antibodies gave similar but not identical results supports the conclusion that the CKI⑀ mAb is relatively specific for CKI⑀.

DISCUSSION
Phosphorylation is a common mode of kinase regulation, acting to both positively and negatively modulate enzyme activity. Phosphorylation is required to turn on a number of protein kinases, including protein kinase A, and members of the mitogen-activated (MAP) and cyclin-dependent (cdk) protein kinase families. Interestingly, CKI is one of a small group of kinases that does not require such an activating catalytic domain phosphorylation. Phosphorylation also negatively regulates a substantial number of protein kinases. For example, phosphorylation of cyclin-dependent kinases on threonine and tyrosine residues in the ATP-binding loop by wee1-related ki- FIG. 3. Phosphatases in HeLa extract actively dephosphorylate CKI⑀. A, recombinant CKI⑀ autophosphorylates in the presence of protein phosphatase inhibitors. In lanes 1-5, recombinant histidinetagged CKI⑀ (rCKI⑀, 200 ng) was incubated for 30 min at 37°C with or without the addition of 4 mM ATP (ATP); a phosphatase inhibitor mixture containing 100 nM calyculin A, 250 nM okadaic acid, and 50 mM ␤-glycerol phosphate (PP inhibitors); and/or 25 g HeLa extract (HeLa extract), as indicated. Reactions were stopped by the addition of SDS-PAGE sample buffer and analyzed by 8% SDS-PAGE and immunoblotting with CKI⑀ mAb. In lanes 6 -9, reactions identical to those in lanes 1-5 were performed except that no recombinant CKI⑀ was added, so only endogenous CKI⑀ was detected. Note untagged CKI⑀ (lanes 6 -9) has a higher electrophoretic mobility than histidine-tagged CKI⑀ (lanes 1 -5). B, CKI⑀ is rapidly dephosphorylated by HeLa phosphatases. Histidine-tagged CKI⑀ was autophosphorylated with [␥-32 P]ATP and then incubated with 25 g of HeLa extract for the indicated number of minutes (0, 1, 3, 5, or 10; lanes [1][2][3][4][5]. Reactions were stopped by the addition of SDS-PAGE sample buffer and analyzed by SDS-PAGE and autoradiography (autorad). nases leads to kinase inactivation, whereas phosphorylation of the Src tyrosine kinase by CRK on a tyrosine residue in the carboxyl terminus maintains the kinase in an inactive conformation (18). We and others have recently demonstrated that, in vitro, intramolecular autophosphorylation of the carboxyl-terminal domain also inhibits kinase activity of CKI␦ and CKI⑀ in a reversible manner. The present study indicates that this auto-inhibition occurs in vivo but is only detected when cellular protein phosphatases are inhibited. When the cellular phosphatases are active, kinase autophosphorylation is rapidly reversed.
The data suggest that the highly related kinases CKI⑀ and CKI␦ are both maintained in the dephosphorylated state in vivo by a futile cycle of autophosphorylation and rapid dephosphorylation. While the mAb raised against the carboxyl terminus of CKI⑀ cross-reacts with CKI␦, we conclude that both kinases are subject to the same phosphorylation cycle. The data supporting this conclusion include the findings that CKI␦ and CKI⑀ antibodies identify similar but not identical bands, that endogenous and HA-tagged CKI⑀ undergoes an electrophoretic mobility shift in transfected cells after phosphatase inhibitor treatment, that CKI␦ and CKI⑀ actively autophosphorylate in vitro, and there is no remaining unshifted immunoreactive kinase after phosphatase inhibitor treatment, suggesting all of the endogenous CKI␦ and CKI⑀ can autophosphorylate.
These findings suggest that in vitro experiments with fulllength CKI⑀ and CKI␦ should be approached with caution since these kinases have the ability to rapidly autophosphorylate and autoinhibit. In cell extracts, the addition of phosphatase inhibitors leads to the autophosphorylation of these kinases even in the absence of added ATP. If the role of the dephosphorylated carboxyl terminus is substrate recognition or kinase targeting, then to study the kinase in vitro may necessitate mapping the inhibitory sites and producing mutant kinases that are unable to autoinhibit their activity. While it is inviting to attempt to determine CKI autophosphorylation sites on the basis of known kinase peptide preferences, analysis of the sequence of the carboxyl terminus of CKI␦ and CKI⑀ show no signature CKI phosphorylation sites. While CKI prefers acidic residues upstream of the target serine or threonine, the isoelectric point of the tails is calculated to be near 12, with no runs of acidic residues next to potential phosphorylation sites. Since CKI can utilize phosphorylated residues as specificity determinants, it is possible that there is a directional step-wise phosphorylation of the tail, with each phosphorylation event creating a new target site.
What might perturb this regulation? CKI⑀ and CKI␦ have been implicated in DNA damage response pathways because of their homology to HRR25, the ability of CKI⑀ to functionally complement HRR25-deleted yeast, and the ability of CKI isoforms to phosphorylate substrates such as p53, SV40 large T antigen, Swi6, and the 14 -3-3 proteins. The finding that CKI⑀ and CKI␦ are actively maintained in a hypophosphorylated state in both cultured cells and in tissues suggests that these kinases are maximally active under basal conditions. Previous studies have suggested that the activity of CKI on a target may be regulated by prior phosphorylation of that target by other kinases, creating a CKI recognition site (19). However, this mechanism does not appear to regulate phosphorylation of the p53, IB␣, ets-1, and 14-3-3 proteins, nor SV40 large T antigen (9,10,12,20). Our results raise the additional possibility that CKI␦ and CKI⑀ may also be regulated by specific intracellular phosphatases. One possible mode of regulation might then be changes in the phosphorylation state and hence the activity of the kinase, due to a change in the activity of the specific phosphatase that normally acts on the kinase. Decreased phos-phatase activity would lead to increased CKI autophosphorylation and decreased CKI kinase activity, perhaps in response to DNA damage. Santos et al. (21) described a Drosophila melanogaster CKI isoform whose activity in immunoprecipitates was markedly stimulated by either alkaline phosphatase treatment or prior ␥-irradiation of embryos, suggesting that this mechanism of CKI control may be widespread. While no specific modulators of global mammalian CKI␦/⑀ phosphorylation state have yet been found, it may be that distinct intracellular subsets of the kinase will be regulated by locally active phosphatases. This mechanism could therefore differentially regulate CKI activity at different subcellular locations. These studies suggest that one level of control of CKI␦ and CKI⑀ activity is mediated by protein phosphatases. The change in CKI⑀ and CKI␦ phosphorylation after treatment of cells with the phosphatase inhibitors okadaic acid or calyculin A suggests that either PP1 or PP2A-like phosphatases act on these kinases in vivo. Since the in vitro studies were performed with purified phosphatase catalytic subunits, we currently have no information on which phosphatase holoenzyme might be active on the kinase. The subcellular localization of these kinases suggests that nuclear as well as cytoplasmic forms of these phosphatases may be involved in this regulation. Multiple mechanisms to regulate these phosphatases have recently been elucidated, including phosphorylation (17,22), polypeptide inhibitors including oncogenes and developmental regulators (23,24), and changes in phosphatase regulatory subunits (17,(22)(23)(24)(25).