Autoinhibition of Casein Kinase I ε (CKIε) Is Relieved by Protein Phosphatases and Limited Proteolysis*

Casein kinase I ε (CKIε) is a member of the CKI gene family, members of which are involved in the control of SV40 DNA replication, DNA repair, and cell metabolism. The mechanisms that regulate CKIε activity and substrate specificity are not well understood. We report that CKIε, which contains a highly phosphorylated 123-amino acid carboxyl-terminal extension not present in CKIα, is substantially less active than CKIα in phosphorylating a number of substrates including SV40 large T antigen and is unable to inhibit the initiation of SV40 DNA replication. Two mechanisms for the activation of CKIε have been identified. First, limited tryptic digestion of CKIε produces a protease-resistant amino-terminal 39-kDa core kinase with several-fold enhanced activity. Second, phosphatase treatment of CKIε activates CKIε 5–20-fold toward T antigen. Similar treatment of a truncated form of CKIε produced only a 2-fold activation. Notably, this activation was transient; reautophosphorylation led to a rapid down-regulation of the kinase within 5 min. Phosphatase treatment also activated CKIε toward the novel substrates IκBα and Ets-1. These mechanisms may serve to regulate CKIε and related forms of CKI in the cell, perhaps in response to DNA damage.

The casein kinase I (CKI) 1 gene family encompasses an increasing number of genes expressed in eukaryotes including yeast Caenorhabditis elegans and mammals. Two subgroups of the CKI family have been separated by functional analysis and complementation of mutations in yeast. One group encoding nuclear kinases appears in yeast to be involved in the response to DNA damage. Mutations in these genes, including HRR25 and YCK3 in Saccharomyces cerevisiae and hhp1 and hhp2 in Schizosaccharomyces pombe, lead to sensitivity to DNA damaging agents such as x-rays and methyl methanesulfonate (1)(2)(3)(4). The mammalian genes encoding CKI␦ and CKI⑀ complement HRR25-deleted yeast, suggesting they too may be involved in DNA repair pathways (5). A second group in S. cerevisiae encodes prenylated membrane-bound isoforms involved in bud growth and includes YCK1 and YCK2 (6); deletions in these genes are complemented by the mammalian genes encoding CKI␥ (7).
The structure of the CKI family suggests several potential mechanisms for the regulation of activity. All family members contain a short amino-terminal domain of 9 -76 amino acids, a highly conserved kinase domain of 284 amino acids, and a highly variable carboxyl-terminal domain that ranges from 24 to more than 200 amino acids in length. The carboxyl terminus of a CKI isoform may serve several functions, including regulation of substrate recognition, modulation of catalytic activity, and/or determination of kinase subcellular localization. Prenylation of the tail of the YCK1/YCK2 family has been shown to be of functional importance in yeast (3,8,9). Phosphorylation of CKI may also be an important regulatory mechanism. Most of the CKI proteins are phosphoproteins, and several of the yeast kinases can autophosphorylate on serine, threonine, and tyrosine residues (10). Studies using synthetic substrates and the artificial substrate casein have indicated that these phosphoryl residues may inhibit kinase activity (11,12), although the location of these inhibitory groups remains unclear. Kuret and co-workers (11) found that an unphosphorylated truncation of Cki1 containing only the kinase domain was twice as active as the phosphorylated form, whereas Graves et al. (12) mapped a phosphorylation-dependent inhibitory domain in CKI␦ to a 26-amino acid domain in the carboxyl-terminal tail. Polyanions such as heparin can regulate kinase activity; activation by heparin appears dependent on the presence or absence of the carboxyl-terminal domain (12,13). Additionally, membrane bound forms of the kinase may be regulated by phosphatidylinositol-4,5-bisphosphate (14).
One in vitro functional assay of CKI activity is its ability to phosphorylate simian virus 40 (SV40) large T antigen. We have previously shown that CKI␣ purified from HeLa cell extracts phosphorylates T antigen on physiologic sites and inhibits the initiation of viral DNA replication (15)(16)(17). As SV40 DNA replication is regulated in the cell cycle and by DNA damage (18 -21), it was of interest to determine whether forms of CKI implicated in DNA repair pathways could also regulate in vitro DNA replication.
We have now overexpressed and purified active CKI⑀. The data show that this form of CKI, although active on peptide substrates, is markedly hindered in its ability to phosphorylate and inhibit the origin unwinding function of T antigen. This decreased activity of CKI⑀ is apparently due to an inhibitory effect of the carboxyl-terminal domain not present in CKI␣, since limited tryptic digestion of CKI⑀ released a catalytically active amino-terminal 39-kDa fragment able to both phospho-rylate T antigen and inhibit its replication initiation function.
We also report that CKI⑀ is activated by dephosphorylation in a tail-dependent manner. Treatment of recombinant CKI⑀ with the catalytic subunits of PP1, PP2A, or PP2B (calcineurin) leads to as much as a 20-fold increase in activity toward T antigen, casein, and two novel substrates, the Ets-1 transcription factor and recombinant IB␣. Activation of the kinase by phosphatases was transient and self-limited; reautophosphorylation of the kinase led to inactivation within 5 min. Activation was dependent on the presence of the carboxyl-terminal domain of the kinase; a truncation mutant of CKI⑀ was activated by phosphatase 4-fold less than was full-length kinase. These findings support the hypothesis that the carboxyl-terminal domain of CKI⑀ inhibits its activity on key protein substrates and suggests that in the cell, CKI⑀ may be regulated in a self-limited manner by phosphatases and in a more sustained manner by intracellular proteolysis.

MATERIALS AND METHODS
Ni 2ϩ -nitrilotriacetate-agarose was obtained from Qiagen. Trypsin (T8642), calcineurin, and calmodulin were from Sigma. PP1 c and inhibitor 2 were from New England Biolabs. Okadaic acid was from Life Technologies, Inc. Restriction enzymes were from Life Technologies, Inc. and New England Biolabs. Plasmids expressing IB␣ (22) were the generous gift of John Hiscott.
Cloning and Escherichia coli Expression of CKI⑀-The cDNA encoding wild type human CKI⑀ was isolated as a NcoI/SalI 1333-base pair fragment and ligated into NcoI/XhoI-digested pET16b (a T7-based expression vector from Novagen) as described previously (5). This construct (pKF115) removes the hexahistidine sequence present in the pET16b vector downstream from the NcoI site. The 1317-base pair NcoI/HindIII fragment from pKF115 was ligated into the same sites in pRSET-B (a T7-based expression vector from Invitrogen). This construct (pV71) encodes CKI⑀ with a 41-amino acid amino-terminal fusion that contains a hexahistidine tag and an enterokinase cleavage site. The d305 truncation mutant was created by the introduction of a stop codon after amino acid 305 in a derivative of pV71 (pKF162) by sitedirected mutagenesis. All recombinant proteins were expressed in BL21(DE3) cells (23). Bacteria were grown in Luria broth containing 50 g/ml carbenicillin at 37°C to an A 600 of 0.3 and induced overnight at 28°C with 1 mM isopropyl-1-thio-␤-D-galactopyranoside.
Purification and Partial Proteolysis of Recombinant CKI⑀-Hexahistidine-tagged CKI⑀ was expressed at low levels in BL21(DE3) cells. Lysates in 20 mM Hepes, pH 7.5, 25 mM NaCl, 1 mM DTT, 1 mM EDTA, 0.02% Nonidet P-40, 10% sucrose (buffer B) with 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 1 g/ml pepstatin were applied to DEAE-cellulose and batch-eluted with the same buffer containing 200 mM NaCl. Kinase-containing fractions were dialyzed into buffer B with 10 mM NaCl and then applied to a S-Sepharose column equilibrated with the same buffer. Kinase activity was batch-eluted with buffer B with 250 mM NaCl (without EDTA or DTT) and loaded directly onto Ni 2ϩ -nitrilotriacetate-agarose from which it was eluted with buffer B (without EDTA or DTT) containing 80 mM imidazole. Untagged CKI⑀ was purified by a similar procedure except that the Ni 2ϩ -nitrilotriacetate-agarose step was omitted. The typical yield was 100 g of CKI⑀ from a 2-liter culture. Hexahistidine-tagged CKI⑀ was used in all assays except where specifically indicated. Histidine-tagged and untagged kinase were found to behave identically in all assays tested.
Proteolysis of purified wild type and hexahistidine-tagged CKI⑀ was performed in the standard kinase reaction buffer (30 mM Hepes, pH 7.5, 7 mM MgCl 2 , 0.5 mM DTT for 30°C for 15 min either without or with previous autophosphorylation of substrate. Trypsin was then inhibited by soybean trypsin inhibitor added to a final concentration of 10 g/ml. Proteolysis with trypsin at 1 g/ml gave almost quantitative scission of CKI⑀ into its major digestion products; this concentration of trypsin was routinely used for the production of the active tryptic fragment. Immunoblot Analysis-For immunoblot analysis, proteins and trypsin-generated peptides were separated by SDS-PAGE on a 17% gel and then transferred to nitrocellulose membrane in 12.5 mM Tris, 86 mM glycine, pH 8.3, 0.1% SDS, 20% methanol. After a 15-min fixation in 0.5% glutaraldehyde in phosphate-buffered saline, the membrane was blocked with 3% bovine serum albumin and then incubated with a 1:500 dilution of UT31 antiserum raised against the amino-terminal CKI⑀ peptide MELRVGNKYRLGC (5). Immunoreactive peptides were de-tected using an alkaline phosphatase-conjugated goat-anti-rabbit IgG (Bio-Rad) followed by incubation with bromo-chloro-indolylphosphate and nitroblue tetrazolium (24).
Kinase and Phosphatase Assays-Kinase reactions were performed in buffer containing 30 mM Hepes, pH 7.5, 7 mM MgCl 2 , 0.5 mM DTT, 100 g/ml bovine serum albumin, 150 or 250 M ATP, and 1-5 Ci of [␥-32 P]ATP at 37°C or as indicated. Reactions were stopped by the addition of SDS-PAGE sample buffer and analyzed by SDS-PAGE and autoradiography as described previously (15,16).
Peptide phosphorylation reactions contained the synthetic 15-mer phosphopeptide substrate AHALS(P)VASLPGLKKK (termed 5P) that was synthesized with the serine in position 5 phosphorylated. This peptide contains a CKI consensus site and is phosphorylated by both CKI␣ and CKI⑀ with a K m of approximately 200 M. 2 Peptide kinase assays were terminated by the addition of 50 l of 30% acetic acid/10 l assay and quantitated by spotting the reaction mixture onto P81 phosphocellulose filters as described (25). All phosphatase reactions were performed in 30 mM Hepes, pH 7.5, 7 mM MgCl 2 , and 100 g/ml bovine serum albumin unless otherwise noted. Protein concentration was determined by the method of Bradford using bovine serum albumin as a standard (26).

RESULTS
Substrate Specificity-In previous studies we purified CKI␣ based on its ability to phosphorylate SV40 large T antigen and thus inhibit the initiation of SV40 DNA replication (15,16). In the process of characterizing human CKI genes, we cloned a novel member of the CKI family, CKI⑀, that encodes a 48-kDa kinase (5). CKI⑀ was of interest because, unlike CKI␣, it was able to functionally complement yeast deleted for the DNA repair-related kinase HRR25. In preliminary studies on recombinant CKI⑀, we found virtually identical specific activity to CKI␣ when tested on peptide substrates (data not shown). However, although CKI␣ was found to readily phosphorylate T antigen, CKI⑀ was approximately 9-fold less active on this protein substrate (Fig. 1). Similar results were obtained when a partially purified non-tagged form of CKI⑀ was tested (data not shown), indicating the histidine tag has no significant effect on the ability of CKI⑀ to phosphorylate T antigen. Interestingly, CKI⑀ autophosphorylation was also consistently stimulated severalfold by the addition of T antigen (Fig. 1, lane 4 compared with lanes 5 and 6), suggesting a specific proteinprotein interaction between the carboxyl-terminal domain and the substrate. Effect of Carboxyl-terminal Tail on CKI⑀ Activity-CKI⑀ and CKI␣ are closely related in the kinase domain (86% similar and 74% identical) but differ in the length of the carboxyl-terminal tail. CKI␣ has only a 24-amino acid extension beyond the kinase domain, whereas CKI⑀ has a 123-amino acid tail. Recent studies of carboxyl-terminal domains in the related kinases CKI␦ and Cki1 have indicated that they perform a regulatory function (11,27). The differing activities of the CKI␣ and CKI⑀ isoforms suggested that the carboxyl-terminal domain of CKI⑀ was responsible for the decreased activity of the kinase on T antigen and secondly, that this region might form a discrete structure separable from the kinase domain by limited proteolysis. To test this, intact autophosphorylated CKI⑀ was subjected to limited digestion with various concentrations (0-5 g/ml) of trypsin. As demonstrated in Fig. 2, digestion of 32 P-autophosphorylated CKI⑀ with increasing amounts of trypsin results in the production of a series of higher mobility species that are further digested to a relatively stable 39-kDa phosphoprotein that finally disappears during incubation at the highest trypsin concentration (5 g/ml). This 39-kDa product appears to be activated CKI⑀ since (a) its appearance coincided with a 3-fold increase in kinase activity toward T antigen, (b) it reacts with polyclonal antibody UT31 that recognizes the amino terminus of CKI⑀, and (c) the purified 39-kDa polypeptide retains kinase activity ( Fig. 3 and data not shown). The 3-fold activation achieved by trypsin treatment may under-represent the actual activation since (a) there may have been partial loss of the kinase domain as well, and (b) proteolytic tail fragments may retain some inhibitory activity (data not shown). These results suggest that CKI⑀ contains a discrete inhibitory domain that blocks specific substrate phosphorylation.
Functional Activity of the 39-kDa Amino-terminal Fragment of CKI⑀ Protein-T antigen catalyzes the initial steps in SV40 DNA replication, the unwinding of the duplex origin of replication to single-stranded DNA. CKI␣ inhibits T antigen function by phosphorylating it on at least two inhibitory sites, serines 120 and 123 (15,16). The finding that proteolytic cleavage of CKI⑀ stimulated its activity on T antigen led to the further functional characterization of the 39-kDa tryptic fragment. Specifically, we wished to determine whether its increased protein kinase activity translated into an increased ability to inhibit T antigen-catalyzed SV40 origin unwinding. CKI␣, CKI⑀, and the 39-kDa CKI⑀ fragment (CKI⑀-39) were pre-incubated with T antigen and ATP, and then T antigen-dependent unwinding of the SV40 minimal origin of replication was assayed (Fig. 3). As previously demonstrated, phosphorylation of T antigen by CKI␣ inhibited its origin unwinding activity with 50% inhibition occurring at about 0.4 pmol of kinase/reaction (40 nM in the preincubation mixture). In contrast, CKI⑀ very inefficiently inhibited the origin unwinding activity of T antigen with 50% inhibition occurring at about 7 pmol of kinase/reaction (700 nM in preincubation mixture). The 39-kDa fragment of CKI⑀ was substantially more active than full-length CKI⑀, producing 50% inhibition of T antigen origin unwinding activity at about 0.9 pmol of kinase/reaction (90 nM in preincubation mixture, Fig. 3B). These results indicate that the catalytic core of CKI⑀ is in fact similar in activity to CKI␣ and that proteolytic removal of the carboxyl terminus of CKI⑀ partially activates it to phosphorylate the same or similar inhibitory sites on T antigen.
Diverse Serine/Threonine Phosphatases Activate CKI⑀ toward T Antigen-Recent studies on the related CKI isoform CKI␦ have demonstrated that the full-length kinase can be activated 2-5-fold toward casein or a synthetic peptide (D4) by treatment with the catalytic subunit of protein phosphatase 1 (PP1 c ) (27). We tested whether removal of phosphoryl groups from CKI⑀ stimulated its kinase activity toward SV40 large T antigen (Fig.4, A and B). Purified CKI⑀ was incubated with increasing amounts of the catalytic subunit of either PP1 or PP2A. At the end of the dephosphorylation reaction, okadaic acid was added to a final concentration of 250 nM, and the activity of the kinase on T antigen was assessed. As demonstrated in Fig. 4, A and B, pretreatment of CKI⑀ with either phosphatase stimulates its activity on T antigen up to 20-fold, with half-maximal activation occurring with less than 2 ng (ϳ4 nM) of PP2A c . The stimulation requires phosphatase catalytic activity, since inclusion of okadaic acid in the preincubation fully blocks the effect of PP2A c (Fig. 4A, lane 7). The activation by 16 ng (ϳ30 nM) of PP1 c is only partially blocked by 250 nM okadaic acid, a not unexpected result since PP1 c is 10 -100-fold less sensitive than PP2A c to inhibition by okadaic acid. Of note, alterations in the phosphorylation state of CKI⑀ had a significant effect on kinase mobility (Fig. 4A). This appears to be due to tail phosphorylation, as truncated CKI⑀ can autophosphorylate without a significant change in electrophoretic mobility ( Fig. 7A and data not shown).
Autophosphorylation Rapidly Inactivates CKI⑀-CKI⑀ is capable of autophosphorylation of up to approximately 12 mol/ mol ( Fig. 4C and see below). Since phosphatase treatment of the bacterially expressed kinase leads to marked activation toward protein substrates, we asked how rapidly reautophosphorylation of the kinase occurs and whether this reautophosphorylation correlates with a decrease in protein kinase activity. CKI⑀ was dephosphorylated with PP2A c and then allowed to reautophosphorylate in the presence of okadaic acid. T antigen was either added at the same time as [␥-32 P]ATP (time 0) or at the indicated times after ATP addition. As Fig. 5 demonstrates, CKI⑀ reautophosphorylation was complete within 5 min, coincident with the re-suppression of kinase activity. Thus, CKI⑀ appears to re-autophosphorylate and autoinhibit itself rapidly in the absence of phosphatase activity. Dilution studies and experiments with CKI⑀ kinase dead mutants indicate that CKI⑀ autophosphorylation is entirely intramolecular (data not shown). Of note, neither activation nor autoinhibition of kinase activity toward a peptide substrate was seen in similar time course reactions (data not shown), suggesting autoinhibition is effective toward protein but not peptide substrates.
Bacterially-produced CKI⑀ Is Heavily Autophosphorylated-Dephosphorylation of bacterially expressed CKI⑀ had a significant activating effect on the kinase. To determine the number of phosphates on CKI⑀ that contribute to kinase inhibition, CKI⑀ autophosphorylation was quantitated after pretreatment with increasing amounts of PP2A c or PP1 c (Fig. 4C). Untreated kinase was able to add about 2.5 mol of phosphate/mol of kinase, whereas PP2A c -treated kinase added about 12 mol/mol. These data suggest that the kinase was close to maximally autophosphorylated at approximately 10 mol/mol when it was purified from E. coli. Half-maximal activation of the kinase toward T antigen was seen when the kinase added approximately 8 mol/mol, suggesting that 4 mol/mol of phosphate had not been removed by the pretreatment. Similar results were seen with PP1 c .
Phosphoamino acid analysis performed on CKI⑀ pretreated with PP2A c and autophosphorylated in the presence of [␥-32 P]ATP indicates that autophosphorylation occurs on both serine and threonine residues (data not shown). Although CKI family members have been shown to be able to autophosphorylate on tyrosine as well as on serine and threonine (10), the autophosphorylation experiments in this study address only serine and threonine phosphorylation as the phosphatases used are serine-/threonine-specific. The data indicate that removal of phosphoryl groups from serine and threonine but not tyrosine residues is responsible for the activation of the kinase.
PP2A Activates CKI⑀ Toward an Array of Protein Substrates-To determine whether dephosphorylation of CKI⑀ activated it toward multiple protein substrates, the kinase was  Fig. 6A, CKI⑀ is activated toward T antigen, casein, and Ets-1 but not replication protein A (RP-A) by PP2A c treatment. Similarly, treatment of CKI⑀ with PP1, PP2A, or PP2B catalytic subunit activated the kinase toward GST-IB␣ (Fig. 6B). The activation of CKI⑀ by phosphatases is therefore a general phenomenon. Whereas CKI substrate specificity on peptides can be determined by phosphoryl groups, we note that the increase in activity occurs toward unphosphorylated (bacterially produced Ets-1 and GST-IB␣) as well as phosphorylated (T antigen) proteins.
CKI⑀ Phosphorylates the Carboxyl Terminus of IB␣-Although genetic studies have suggested a role for CKI⑀-related proteins in DNA damage response pathways, few physiologic substrates of CKI⑀ have been identified. As IB␣ is phosphorylated and degraded in response to many signals including DNA damage and it contains sequences in the amino and carboxyl termini similar to CKI phosphorylation consensus sites found in peptide substrates, we tested whether CKI⑀ could be activated to phosphorylate IB␣. As shown in Fig. 6B, CKI⑀ phosphorylates GST-IB␣ in a phosphatase-activable manner.
The serine/threonine phosphatases PP2A c , PP1 c , and PP2B/ calcineurin were all able to activate CKI⑀ to phosphorylate recombinant IB␣.
Two regions of phosphorylation of IB␣ have been identified; serines 32 and 36 in the amino terminus are essential for the ubiquitin-mediated degradation of IB␣ bound to NF-B, whereas phosphorylation of extreme carboxyl-terminal residues in a PEST region regulates proteolysis of free IB␣ (22,28,29). To determine whether CKI⑀ phosphorylated the carboxyl terminus of IB␣, the ability of activated kinase to phosphorylate full-length and carboxyl-terminal truncation mutants (the generous gift of John Hiscott) was assayed. As shown in Fig. 6C, CKI⑀ phosphorylated full-length GST-IB␣ (lanes 1 and 5) but had no activity on mutants lacking the carboxyl-  2-4, 6 -8) (22). Note that although CKI⑀ phosphorylates IB␣ on both serine and threonine (data not shown), the nonphosphorylated ⌬4 truncation removes two threonine but no serine residues. Studies with point mutants suggest serines 288 and 293 are phosphorylated by CKI⑀ (data not shown). The ⌬4 mutant (truncated at amino acid 296) has been shown to be a substrate for the unrelated kinase, casein kinase II, indicating that the truncation does not lead to denatured protein (22). The data therefore suggest that removal of the carboxyl-terminal 22 amino acids disrupts the local structure enough to interfere with CKI⑀ but not CKII activity on IB␣. These results are consistent with our previous data (15) indicating that CKI activity on protein (but not peptide) substrates is highly dependent on the intact tertiary structure of the substrate.
Full Activation by Phosphatases Requires the Carboxyl Terminus of CKI⑀-One model to explain both the inhibitory effect of the carboxyl-terminal tail and the stimulatory effect of dephosphorylation is that phosphate groups on the carboxyl terminus of CKI⑀ interact with the kinase domain, leading to inhibition of protein substrate binding. This model suggests that a tail-less CKI⑀ should not be activated by phosphatases. To test this, a stop codon was introduced after amino acid 305 in CKI⑀, and the truncated histidine-tagged protein (denoted CKI⑀-d305) was expressed in E. coli and purified on Ni 2ϩnitrilotriacetate-agarose. Roughly equal amounts of full-length and truncated CKI⑀ were used to phosphorylate T antigen, either without or with prior phosphatase treatment. As shown in Fig. 7, CKI⑀ was activated 9-fold by PP2A c , whereas CKI⑀-d305 was activated only 2-fold. Similarly, CKI␣ (76% identical to CKI⑀ over the kinase domain and lacking a carboxyl-terminal tail) was not activated by PP2A c (data not shown). The data suggest that much of the observed autoinhibition of CKI⑀ requires the carboxyl-terminal tail, but that an inhibitory phosphoryl group on the CKI⑀ kinase domain also contributes to autoinhibition. These results are consistent with those seen by Kuret and co-workers (11) on the S. pombe homolog Cki1-d298 but differ slightly from those of Graves and Roach (27) who found that truncated CKI␦ was not activated by phosphatase treatment.
Previous work demonstrated that the activity of CKI⑀ toward peptide substrates was not significantly influenced by the presence of the carboxyl-terminal domain. We next tested whether phosphatase activation of the kinase had an effect on activity toward peptide, as opposed to protein, substrates. As shown in Fig. 7C, PP2A c selectively activated CKI⑀ toward T antigen but not the 5P peptide substrate. DISCUSSION The casein kinase I family is characterized by a conserved core kinase domain and a series of variable carboxyl-terminal extensions. This study indicates that one major function of the carboxyl terminus is to regulate the activity of CKI on protein substrates. The ability of the carboxyl terminus to inhibit protein but not peptide phosphorylation suggests that this tail region interacts with the substrate binding face of the kinase, as illustrated in Fig. 8, rather than directly in the catalytic cleft. The data suggest CKI autoinhibition may be relieved by at least three mechanisms; (a) proteolytic cleavage of the tail (this study), (b) dephosphorylation of the kinase (this study and Refs. 11 and 27), and (c) binding of heparin (and presumably other polyanions) to the tail (27). Which, if any, of these mechanisms function in vivo is the subject of ongoing investigation. One reason for the diversity in CKI carboxyl-terminal domains may be to allow distinct activation mechanisms for the different family members. Alternatively, the function of the tail may be to constitutively restrict access to the catalytic cleft to all but a limited number of substrates, and no further regulation of the kinase may occur in vivo. Differentiation between these mechanisms will require the demonstration and characterization of CKI activation in vivo. Given the functional similarity between CKI⑀ and HRR25, one possibility is that agents that trigger the DNA repair response will lead to CKI⑀ activation. Activation of forms of CKI lacking a carboxyl-terminal extension by dephosphorylation suggests that at least one inhibitory phosphoryl group is present on the core kinase domain and may interact directly with the kinase tail. Additionally, the finding that multiple phosphatases can activate CKI⑀ raises the possibility that the kinase may be activated by diverse signal transduction pathways. Since the inhibitory phosphoryl groups are placed in an intramolecular autophosphorylation reaction, kinase activation by dephosphorylation allows a very tight temporal control over the duration of kinase activity. As we have demonstrated, once the activating phosphatase is removed, autophosphorylation leads to rapid kinase inactivation. Such self-limited bursts of kinase activity allow rapid downregulation of kinase activity, a mechanism that may be required to turn off a pathway once its inciting agent is removed. Alternatively, if CKI⑀ itself activates a phosphatase, for example by CKI phosphorylation of the protein phosphatase 1 inhibitor 2 (data not shown and Ref. 30), then a positive feedback loop may be established that can further increase the activity of CKI⑀.
The observation that trypsin can activate CKI⑀ supports the model (shown in Fig. 8) that the inhibitory tail is much more accessible to proteolysis than is the kinase domain itself. Cleavage of CKI⑀ by regulated intracellular proteolysis would lead to a long lived activation of CKI⑀ in response to the appropriate stimulus. Proteolytic activation has been demonstrated for an increasing number of intracellular proteins, including protein kinase C, mitogen-activated protein kinase kinase kinase 1 (MEKK1), and NF-B (31)(32)(33). DNA damage resulting in protease activation might lead to CKI⑀ cleavage. Notably, caspases with DEVD sequence specificity (34) may be activated during apoptosis and cleave in the acidic region of CKI⑀ at the beginning of the carboxyl terminus. Such cleavage might either activate CKI⑀ or release it from its normal anchoring site.
Previous studies have suggested that CKI recognizes peptide and protein substrates differently. CKI phosphorylates acidic peptides, with the best peptide substrates containing either phosphorylated or acidic residues amino-terminal to the target site (35,36). In this case, peptide recognition requires interaction of the phosphorylated region with the catalytic fold (37,38). However, this localized charge interaction does not appear operative in the case of T antigen phosphorylation. We have previously shown that (a) CKI␣ can phosphorylate bacterially expressed (and hence unphosphorylated) T antigen and (b) CKI␣ does not recognize its target site in the amino terminus of T antigen unless the carboxyl-terminal residues of T antigen are present (15). Thus, CKI␣ could not phosphorylate a T antigen-derived peptide, an amino-terminal tryptic fragment of T antigen, nor T259, an active recombinant amino-terminal fragment of T antigen that contains the target sites and the DNA binding domain. Similarly, in the current study we found that removal of the carboxyl terminus of IB␣ blocks phosphorylation of upstream residues. These results are consistent with a more complex picture of kinase-substrate interactions, where the tertiary structure of the substrate interacts with multiple surface features of the kinase. Therefore, additional structural elements of the kinase may interfere with docking of substrate proteins at sites distant from the target residues. One function of the CKI⑀ tail may be to restrict the access of protein substrates to the active site until the kinase is activated by transient dephosphorylation or limited proteolysis. However, these results do not exclude the possibility that in vivo, the inhibitory effect of the tail observed in vitro could also be overcome by constitutive active dephosphorylation or additional mechanisms such as noncovalent binding of regulatory molecules.
Identification of potential cellular substrates of CKI has been problematic. CKI isoforms can phosphorylate SV40 large T antigen (15,16), p53 (39), inhibitor-2 (30), and glycogen synthase, among others. The present study extends the list by identifying the carboxyl terminus of IB␣ as an in vitro substrate. Whether CKI is as important as CKII in the phosphorylation of IB␣ in vivo is unclear. In very few cases has a clear functional role for the mammalian CKI family been defined. One approach to this question has been to overexpress kinases in cells or organisms and examine them for changes in phosphorylation patterns. An implication of this current study is that overexpression of CKI⑀ and related CKI family members may produce inactive, autoinhibited kinase with no phenotype or effect on in vivo substrates unless point mutations, truncations, or activating conditions are first introduced. One possible route to the identification of additional in vivo-specific substrates of CKI⑀ is therefore the development of autophosphorylation site point mutants that retain an intact regulatory domain but are constitutively active. FIG. 8. Model. Regulation of CKI⑀ activity by autophosphorylation and an inhibitory tail. The tail of autophosphorylated CKI⑀ interacts with the kinase domain to block access to the active site of protein but not peptide substrates. Full autoinhibition requires the presence of both the tail and inhibitory phosphoryl groups that are accessible to multiple phosphatases. The data suggest that at least one inhibitory phosphoryl group is on the body of the kinase.