Identification of Inhibitory Autophosphorylation Sites in Casein Kinase I ε*

Casein kinase I ε (CKIε) is a widely expressed protein kinase implicated in the regulation of diverse cellular processes including DNA replication and repair, nuclear trafficking, and circadian rhythm. CKIε and the closely related CKIδ are regulated in part through autophosphorylation of their carboxyl-terminal extensions, resulting in down-regulation of enzyme activity. Treatment of CKIε with any of several serine/threonine phosphatases causes a marked increase in kinase activity that is self-limited. To identify the sites of inhibitory autophosphorylation, a series of carboxyl-terminal deletion mutants was constructed by site-directed mutagenesis. Truncations that eliminated specific phosphopeptides present in the wild-type kinase were used to guide construction of specific serine/threonine to alanine mutants. Amino acids Ser-323, Thr-325, Thr-334, Thr-337, Ser-368, Ser-405, Thr-407, and Ser-408 in the carboxyl-terminal tail of CKIε were identified as probable in vivo autophosphorylation sites. A recombinant CKIε protein with serine and threonine to alanine mutations eliminating these autophosphorylation sites was 8-fold more active than wild-type CKIε using IκBα as a substrate. The identified autophosphorylation sites do not conform to CKI substrate motifs identified in peptide substrates.

Casein kinase I ⑀ (CKI⑀) is a widely expressed protein kinase implicated in the regulation of diverse cellular processes including DNA replication and repair, nuclear trafficking, and circadian rhythm. CKI⑀ and the closely related CKI␦ are regulated in part through autophosphorylation of their carboxyl-terminal extensions, resulting in down-regulation of enzyme activity. Treatment of CKI⑀ with any of several serine/threonine phosphatases causes a marked increase in kinase activity that is self-limited. To identify the sites of inhibitory autophosphorylation, a series of carboxyl-terminal deletion mutants was constructed by site-directed mutagenesis. Truncations that eliminated specific phosphopeptides present in the wild-type kinase were used to guide construction of specific serine/threonine to alanine mutants. Amino acids Ser-323, Thr-325, Thr-334, Thr-337, Ser-368, Ser-405, Thr-407, and Ser-408 in the carboxylterminal tail of CKI⑀ were identified as probable in vivo autophosphorylation sites. A recombinant CKI⑀ protein with serine and threonine to alanine mutations eliminating these autophosphorylation sites was 8-fold more active than wild-type CKI⑀ using IB␣ as a substrate. The identified autophosphorylation sites do not conform to CKI substrate motifs identified in peptide substrates.
Casein kinase I epsilon (CKI⑀) 1 is a member of a family of widely expressed, highly conserved, monomeric, basic protein kinases. Distinct CKI family members are likely to have distinct roles in the cell, as recent studies have defined a role for CKI family members in diverse processes including the regulation of SV40 DNA replication (1), in vivo vesicle trafficking (2), DNA repair in yeast (3,4), circadian rhythm in Drosophila (5), cell cycle progression (6), and nuclear import of NF-AT4 (7) in mammalian cells. Distinct CKI isoforms may be regulated by differences in subcellular localization, substrate specificity, and modes of regulation. For example, the YCK1 and YCK2 isoforms are membrane-bound in yeast because of carboxyl-terminal prenylation, whereas HRR25 is predominantly nuclear (8 -10). CKI isoforms have also been identified in the cytosol and the nucleus and on mitotic spindles (11)(12)(13).
An increasing number of potential physiologic substrates of casein kinase I isoforms have been identified, but how the activity of the CKI family members on those substrates is regulated is generally not known. CKI isoforms in vitro preferentially phosphorylate peptides with acidic or phosphorylated residues N-terminal of the target site (14,15), and therefore prior phosphorylation of the substrate is one potential mechanism for regulation of kinase activity. CKI⑀ and the related kinase CKI␦ phosphorylate N-terminal residues of p53 in vitro and in vivo (16,17); this activity is enhanced by DNA damaging drugs. CKI␣ binding to NF-AT4 may be regulated by MEKK1 (7), whereas a CKI␣ homolog in Drosophila has been reported to change subcellular localization and activity in response to irradiation (18).
One way the activity, localization, and specificity of CKI isoforms may be regulated is through their diverse carboxylterminal domains. CKI family members have a similar primary sequence arrangement consisting of a highly conserved aminoterminal catalytic domain of approximately 283 amino acids and carboxyl-terminal extensions of variable length and sequence. Interestingly, although the kinase domains are highly conserved between species (e.g. human CKI⑀ and yeast HRR25 kinase domains are 64% identical and 81% similar), the carboxyl termini in general have no discernible sequence homology. One exception to this is in mammals, where the 124-amino acid tail of CKI⑀ is 50% identical to the tail of CKI␦. Several lines of evidence suggest the activities of CKI␦ and CKI⑀ are regulated by a carboxyl-terminal phosphorylation-dependent autoinhibitory domain (19,20). Autophosphorylation both inactivates the kinase and leads to the accumulation of up to 8 mol of phosphate/mol of kinase. Removal of the CKI␦ or CKI⑀ carboxyl-terminal domain by mutagenesis or proteolysis reactivates the kinases. Furthermore, Graves and Roach (20) showed that transfer of the CKI␦ tail to CKI␣ conferred autoinhibition on that chimeric kinase as well. Interestingly, in vivo, these kinases also autophosphorylate, but this autophosphorylation is rapidly reversed by endogenous protein phosphatases in a futile autophosphorylation-dephosphorylation cycle (13). The specific function of this futile cycle is not known, but it is potentially a mechanism to regulate either kinase activity or the ability of specific substrates to bind to the kinase.
To further study the function of CKI⑀ in vitro and in vivo, we mapped the regulatory autophosphorylation sites on the CKI⑀ carboxyl terminus. Progressive truncation of CKI⑀ eliminated both its autophosphorylation sites and the ability to activate the kinase by dephosphorylating it with protein phosphatase 2A. Potential phosphorylation sites were then identified by two-dimensional phosphopeptide mapping. Mutation of specific residues to alanine produced a recombinant enzyme with 8-fold higher specific activity. Interestingly, none of the identified CKI autophosphorylation sites conform to the consensus sites determined by studies of synthetic peptides.

MATERIALS AND METHODS
Ni 2ϩ -nitrilotriacetate-agarose was obtained from Qiagen. Trypsin (T8642) and cellulose plates were from Sigma. Okadaic acid and calyculin A were from Life Technologies, Inc. and CalBiochem, respectively. Restriction enzymes, T4 DNA ligase, and T4 DNA polymerase were from Life Technologies, Inc. and New England Biolabs. Anti-CKI⑀ monoclonal antibody was from Transduction Laboratories. Primers and peptides were obtained from the DNA/Peptide Facility at the University of Utah. An expression construct for CKI␦ ⌬317 was the gift of Paul Graves and Peter Roach, and purified CKI␦ ⌬317 protein was graciously provided by Erica Vielhaber.
Metabolic Labeling and Mapping of in Vivo Phosphorylation Sites-The human embryonic kidney cell line 293 was transiently transfected with cytomegalovirus expression constructs pKF182 or pKF183 (13) or with empty vector (pCEP4-lerner). Cells at approximately 80% confluence were transfected with 2 g of plasmid DNA mixed with 6 l of LipofectAMINE reagent (Life Technologies, Inc.) according to the manufacturer's instructions. At 36 h post-transfection, cultures were metabolically labeled for 5 h in 5% dialyzed calf serum, 2 mCi ml Ϫ1 H 3 32 PO 4 , and phosphate-free Dulbecco's modified Eagle's medium (all from NEN Life Science Products). Calyculin A or buffer/solvent control was 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 radioimmune precipitation buffer (1% Nonidet P-40, 150 mM NaCl, 0.1% SDS, 50 mM Tris, pH 8.0, 1 g l Ϫ1 leupeptin, 1 g l Ϫ1 pepstatin, 0.1 mM phenylmethylsulfonyl fluoride, 5 mM Na 3 VO 4 , 20 mM NaF, 250 nM okadaic acid, and 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 monoclonal antibody and protein A-agarose. The immunoprecipitates were eluted from the protein A-agarose and separated by SDS-PAGE on a 9% gel. Results were visualized by Phospho-rImager (Molecular Dynamics). Radiolabeled kinases were isolated in gel slices and subjected to trypsin digestion as described below. As a control, a corresponding region of the gel from the empty vector lanes was excised and processed identically to those for the HA-tagged kinases. Two-dimensional phosphopeptide maps were generated as described below.
Construction of Site-directed Mutagenesis Construct-The construct used for site-directed mutagenesis was a derivative of pV71 (19) that contains the CKI⑀ open reading frame downstream of a hexahistidine tag and an enterokinase cleavage site in the vector pRSET-B (Invitrogen). The modified construct, pKF158, contains a tetracycline resistance gene (tet r ) and a point mutant ampicillin resistance gene (amp m ). The tetracycline resistance cassette was PCR-amplified from pAlter-1 (Promega) using the primers TET1 (5Ј-AACATGTCCGGATTCTCAT-GTTTGACAGCTTATCA) and TET2 (5Ј-GTGCAGTCCGGAGACTTC-CGCGTTTCCAGACTT), each containing engineered BspEI sites. The PCR product was digested with BspEI and inserted at the same site of pV71 in an orientation such that the direction of transcription was away from the polylinker. The tet r amp r construct, pKF152, was then modified by replacing a 1-kilobase pair fragment bordered by AlwNI and ScaI with a similar fragment from pAlter-1 (Promega) containing a nonfunctional ampicillin resistance gene. The pKF158 plasmid was used for both site-directed mutagenesis and overexpression in Escherichia coli.
Site-directed Mutagenesis of CKI⑀-Site-directed mutagenesis was conducted essentially by the Altered Sites method (Promega). Individual primary amp r transformants of E. coli strain 71-18 mutS were screened for either the presence or absence of the restriction site introduced or eliminated with each mutation (see Table I).
Quantitative Immunoblot Analysis-Equal amounts of total protein from partially purified CKI⑀ preparations were run on 10% SDS-PAGE and transferred to supported nitrocellulose (Amersham Pharmacia Biotech) as described previously (19).The membrane was blocked by incubation in TTBS (20 mM Tris, pH 7.5, 500 mM NaCl, 0.05% Tween 20) containing 3% bovine serum albumin and then probed with affinitypurified polyclonal antibody UT31, added at a 1:1000 dilution in blocking solution as the primary antibody (23). The proteins of interest that reacted with UT31 were visualized with a secondary detection step of 125 I-labeled protein A (Amersham Pharmacia Biotech) added at 10 nCi ml Ϫ1 of blocking solution. A standard curve of serial dilutions of a single protein preparation was used to determine the relative concentration of all CKI⑀ preparations tested. The results were visualized and quantitated by PhosphorImager (Molecular Dynamics).
Kinase and Phosphatase Assays-Kinase reactions were performed in buffer containing 100 or 250 M ATP, 30 mM HEPES, pH 7.5, 7 mM MgCl 2 , 0.5 mM dithiothreitol, and 2 Ci of [␥-32 P]ATP in a final volume of 20 l. The reaction mixtures were incubated for 5 min at 37°C, and then the reactions were stopped by the addition of SDS-PAGE sample buffer and analyzed by SDS-PAGE and autoradiography as described previously (1, 24). All assays were performed at least twice with good interassay reproducibility.
All phosphatase reactions were performed for 15 min at 37°C in 30

5Ј-TGGGATCCGGGAGACTTATTACCGCCCAGTGAGGTC-3Ј
a Italics identify sites of mutagenesis. Introduced restriction sites are underlined. All primers are phosphorylated at their 5Ј-ends and represent the anticoding strand of the CKI⑀ gene. mM HEPES, pH 7.5, 7 mM MgCl 2 , and 200 g of ml Ϫ1 bovine serum albumin and contained 8 -12 ng of the catalytic subunit of PP2A unless otherwise noted. Protein concentration was determined by the method of Bradford (25).
In Vitro Autophosphorylation and Two-dimensional Peptide Mapping-Partially purified CKI⑀ proteins were radiolabeled in vitro or, for Fig. 1, immunoprecipitated from 32 P-labeled cells. Approximately 20 -50 g of each kinase were treated with 1 g of PP2A c for 15 min at 37°C. Phosphatase activity was blocked by the addition of 200 nM okadaic acid, and the kinase was allowed to re-autophosphorylate in the presence of [␥-32 P]ATP for 15 min at 37°C. The kinase reactions were resolved by SDS-PAGE on a 10% gel. Protein was stained briefly with Coomassie Brilliant Blue, the gels were dried, and the labeled proteins was visualized by autoradiography. Radiolabeled protein bands were excised and rehydrated in 50 mM ammonium bicarbonate digestion buffer. The gel slices were minced, 10 g of trypsin was added, and digestion was carried out for 20 h at 37°C. The buffer was removed from the gel slices and Cerenkov-counted to determine recovery of tryptic phosphopeptides. The digest was then lyophilized to dryness.
The two-dimensional peptide mapping method of Van Der Geer et al. (26) was used to separate phosphopeptides of CKI⑀. Plastic-backed 100-m cellulose plates were obtained from Sigma. Lyophilized tryptic peptides of CKI⑀ were suspended in 5-10 l of pH 1.9 electrophoresis buffer and spotted onto a cellulose plate. For maps performed in parallel, equal counts were spotted on each plate. Electrophoresis was carried out at 1300 V for 30 min in pH 1.9 buffer containing 2.2% formic acid, and 7.8% acetic acid. Following electrophoresis, the cellulose plates were allowed to dry completely. Dried plates were subjected to ascending chromatography for 3 h in phosphochromatography buffer containing 37.5% n-butanol, 25% pyridine, and 7.5% acetic acid. Phosphopeptides were visualized by PhosphorImager (Molecular Dynamics).

CKI⑀ in Vivo
Phosphorylation-CKI⑀ activity is regulated in vitro by carboxyl-terminal tail autophosphorylation; in vivo the autophosphorylated kinase is rapidly dephosphorylated in a futile cycle of autophosphorylation and dephosphorylation (13). To determine whether CKI⑀ was autophosphorylated in vitro and in vivo on the same sites, two-dimensional phosphopeptide maps were prepared from CKI⑀ autophosphorylated in vivo (Fig. 1, A and B, panels a-f) and in vitro (Fig. 1B, panel g). In vivo autophosphorylated CKI⑀ was immunoprecipitated from transiently transfected human embryonic kidney 293 cells metabolically labeled with H 3 32 PO 4 . The rapid turnover of phos-  (Table I). Predicted trypsin cleavage sites are denoted by filled triangles.

FIG. 1. CKI⑀ is autophosphorylated on similar sites in vivo and in vitro.
A, plasmids encoding 4xHA-tagged CKI⑀, either wild type (WT, lanes b and e) or a kinase-inactive mutant (MUT, lanes c and f), under the control of the cytomegalovirus promoter were transiently transfected into HEK 293 cells. Empty vector (V, lanes a and d) was used as a control. At 2 days post-transfection, cultures were metabolically labeled with 2 mCi/ml H 3 32 PO 4 in phosphate-free medium for 5 h. For the last 30 min of labeling, cultures were either not treated (lanes a-c) or treated (lanes d-f) with the cell-permeable phosphatase inhibitor calyculin A (50 nM). Cells were then lysed in radioimmune precipitation buffer, and extracts were clarified by centrifugation. Expressed proteins were immunoprecipitated from equal amounts (g) of extract with 12CA5 monoclonal antibody and protein A-agarose and separated by SDS-PAGE on a 9% gel. Control experiments demonstrated equal expression of wildtype and mutant CKI⑀ under these conditions (data not shown). Radioactive proteins were visualized by PhosphorImager analysis. The sites of CKI⑀ predicted migration is indicated by an open circle. CKI⑀ with altered mobility is indicated by a filled circle. B, immunoprecipitated in vivo 32 P-lableled CKI⑀ was excised from the lanes shown in A and analyzed by two-dimensional phosphopeptide mapping. Phosphopeptides were visualized by PhosphorImager. Recombinant CKI⑀ autophosphorylated in vitro is shown in panel g. Specific phosphopeptides are indicated with letters (generally above the spot) for clarity. phate on CKI⑀ in vivo autophosphorylation sites (13) was blocked by addition of the cell-permeable phosphatase inhibitor calyculin A to selected cells for the last 30 min of labeling. As  Fig. 1A, lanes c and f, shows, immunoprecipitated kinase-inactive CKI⑀ (K38R) appears minimally phosphorylated in vivo. Phosphopeptide mapping demonstrates that the kinase-inactive CKI⑀ is phosphorylated predominantly on a single peptide (spot f in Fig. 1B) and that phosphorylation is not altered substantially by the addition of the phosphatase inhibitor calyculin A (Fig. 1B, panels c and f).
In vivo, wild-type CKI⑀ is minimally phosphorylated, and on the same peptide f in the absence of calyculin A, with low levels of autophosphorylation on additional peptides i and h (compare Fig. 1B, panels b and c). The addition of calyculin A to transfected cells inhibits a number of endogenous serine/threonine phosphatases and leads to a marked increase in autophosphorylation of CKI⑀ on additional sites (Fig. 1A, lanes b and e, and peptides d, e, g, and m in Fig. 1B, panel e). The phosphorylation of CKI⑀ in vivo in the presence of calyculin A is primarily autophosphorylation, because the phosphorylation of kinaseinactive CKI⑀ is not increased significantly by the phosphatase inhibitor (compare lanes c and f, Fig. 1A). Previous studies have established that the autophosphorylation of CKI⑀ in vitro and in vivo is intramolecular (13). CKI⑀ appears to autophosphorylate on the same peptides in vitro as it does in vivo, as the phosphopeptide maps of the kinase labeled either way are very similar (compare panels e and g, Fig. 1B). In addition, peptide maps prepared from bacterially expressed protein (Fig. 1B,  panel g) demonstrate two additional phosphopeptides, labeled j and k. In vitro labeled protein may contain these extra phosphopeptides because of more extensive autophosphorylation in vitro, or the sites may be phosphorylated in vivo but not detected because they are rapidly dephosphorylated by a cellular phosphatase that is not inhibited by calyculin A. Phosphopeptides a, b, and c appear to be nonspecific, as they appear in the absence of kinase as well (compare Fig. 1B, panel d with panels e and f). Because the CKI⑀ in vivo and in vitro autophosphorylation sites appear to be similar, bacterially expressed in vitro autophosphorylated protein was used for phosphopeptide mapping experiments.
Truncation Mutagenesis of the CKI⑀ Carboxyl-terminal Tail-Autophosphorylated CKI⑀ can be activated up to 20-fold by treatment with active PP2A c . Previous truncation and domain-swap experiments have indicated that autophosphoryla-

FIG. 3. Truncation of the CKI⑀ carboxyl terminus diminishes kinase activation by dephosphorylation.
A, immunoblot of CKI⑀ truncation mutant proteins. His 6 -tagged CKI⑀ truncation mutants were expressed in E. coli and partially purified on Ni 2ϩ -nitrilotriacetateagarose. This recombinant protein was autophosphorylated in E. coli and was not further autophosphorylated in vitro (19). Kinase levels were normalized by quantitative immunoblot using affinity-purified UT31 as the primary antibody and detection using 125 I-labeled protein A. The results were visualized and quantitated by PhosphorImager analysis. Shown are two concentrations each, 500 (even-numbered lanes) and 250 ng (odd-numbered lanes), of full-length (FL) CKI⑀ (lanes 13 and 14) and truncation mutants D305 (lanes 1 and 2), D329 (lanes 3  and 4), D349 (lanes 5 and 6), D360 (lanes 7 and 8), D370 (lanes 9 and 10), and D383 (lanes 11 and 12). B, kinase activation by PP2A. CKI⑀ full-length and truncation mutants D305, D319, D329, D349, D360, D370, and D383 shown in panel A were incubated without or with 8 ng of PP2A c for 15 min at 37°C before the addition of okadaic acid, [␥-32 P]ATP, and SV40 large T antigen for a 3-min kinase reaction. Less than 5% of the substrate was converted to the phosphorylated product under these conditions. Reaction products were separated by SDS-PAGE, quantitated by PhosphorImager analysis, and graphed as the fold activation of CKI⑀ after PP2A c treatment. This assay was repeated with similar results. Removal of residues between amino acids 360 and 349 produce a kinase substantially less responsive to dephosphorylation. tion sites in the carboxyl-terminal tail of CKI⑀ and CKI␦ are responsible for this autophosphorylation-dependent autoinhibition (13,19,20). To determine the specific regions of CKI⑀ required for phosphorylation-dependent autoinhibition, a series of histidine-tagged carboxyl-terminal truncation mutants of CKI⑀ were generated by site-directed mutagenesis ( Fig. 2 and Table II). These truncated active kinases were expressed in E. coli and partially purified by metal-chelate chromatography, and CKI⑀ protein levels were normalized by quantitative immunoblot (Fig. 3A). The kinases as purified from E. coli were substantially autophosphorylated and hence autoinhibited (19).
To determine which regions within the CKI⑀ carboxyl-terminal tail were important for phosphorylation-dependent inhibition, the activity of the recombinant truncated autoinhibited kinases on SV40 large T antigen was determined before and after their activation by PP2A c . As shown in Fig. 3B, the activity of full-length CKI⑀ and truncation mutants D383, D370, and D360 was stimulated up to 15-fold toward T antigen by pre-treatment with PP2A c , whereas truncation mutants D349, D329, D319, and D305 were activated only 3-fold. These results were reproducible using T antigen as a substrate, and similar results were obtained when the CKI⑀ truncation mutants were used to phosphorylate casein (data not shown).
Removal of the inhibitory carboxyl-terminal domain of CKI⑀ by limited trypsin proteolysis has previously been shown to increase the specific activity of CKI⑀ 3-fold (19). To determine whether recombinant truncated forms of CKI⑀ would show a similar increase in activity relative to full-length CKI⑀, fulllength CKI⑀ and truncation mutants D383, D370, D360, D349, D329, and D305 were tested for their ability to phosphorylate IB␣ (19). The truncation mutant D349 was approximately 2.5-fold more active than truncation mutant D360 (Fig. 4). These results suggest that a phosphorylation-dependent inhibitor of CKI⑀ activity lies between or near residues 349 -360 of the CKI⑀ carboxyl-terminal tail.
The simplest explanation for these results is that inhibitory autophosphorylation sites are located between residues 349 and 360. To test this, the putative phosphoacceptor residues (Ser-350, Thr-351, and Ser-354) in this region were mutated (in the background of full-length CKI⑀) to alanine (mutant M3, Fig. 2 and Table II), and the resulting mutant kinase was expressed and tested. However, mutant M3 showed no signif-icant decrease in the ability to be activated by phosphatase (data not shown) nor was there any detectable increase in the specific activity of the enzyme (Fig. 5). Additionally, two-dimensional phosphopeptide mapping of the M3 protein showed no change in phosphopeptides (data not shown). One potential explanation of the data is that this region may inhibit kinase activity or kinase-substrate interaction by interaction with inhibitory phosphoryl groups more proximal to the kinase domain.
Two-dimensional Peptide Maps of CKI⑀-To identify the specific sites of CKI⑀ autophosphorylation, the panel of truncation mutants was further analyzed. Mutant proteins expressed in E. coli and purified as described were first treated with PP2A c to remove phosphoryl groups placed by autophosphorylation during expression and purification. Phosphatase activity was then inhibited by the addition of okadaic acid, and the kinase was allowed to autophosphorylate in the presence of 250 M [␥-32 P]ATP. Autophosphorylated kinases were isolated by SDS-PAGE and phosphopeptide-mapped as described under "Materials and Methods." This procedure allowed the preferential radiolabeling of phosphoacceptor sites sensitive to PP2A c and hence sites implicated in autoinhibition and relief of autoinhibition by phosphatase treatment (19). Truncations that lead to loss of phosphopeptides were further analyzed by introduction of mutations in the implicated region, converting specific serine and threonine residues to alanine. All informative phosphopeptide maps were performed at least twice with similar results. Fig. 6, A and B, illustrates the identification of potential autophosphorylation sites. Truncation mutant D319 lacks a single phosphopeptide (indicated by an arrow) present in truncation mutant D329 and full-length CKI⑀, suggesting there are phosphorylation sites in that interval (Fig. 6A). The two potential phosphorylatable residues, Ser-323 and Thr-325, were therefore mutated to alanine in full-length CKI⑀, and the resultant protein (designated M1) was expressed, autophosphorylated, and phosphopeptide-mapped. As the arrow indicates, mutant M1 lacks a phosphopeptide present in the wild-type protein, strongly suggesting that Ser-323 and/or Thr-325 are autophosphorylation sites.
Phosphopeptide mapping results for truncations between residues 360 and 383 (Fig. 6B) permit the assignment of specific autophosphorylation site, Ser-368. Both D360 and the double mutant S368/377A lacked a specific phosphopeptide (Fig. 6B, solid arrow) present in full-length and D383 CKI⑀. A phosphopeptide not present in D360 is apparent in the D370 mutant, albeit at a faster vertical mobility than in full-length or D383 CKI⑀ (open arrow). Closer examination of the primary amino acid sequence in this region revealed that the D370 stop codon was introduced one amino acid carboxyl-terminal to a potential trypsin cleavage site. Trypsin cleaves inefficiently very close to the ends of polypeptides (27); therefore, the aberrant migration observed for D370 is probably because the trypsin site following residue 369 was not used in the D370 mutant, resulting in a peptide longer by several residues, including a valine residue that increased the hydrophobicity of the peptide. To confirm this hypothesis, digestion of radiolabeled D370 was repeated. A fraction of the phosphopeptide with aberrant migration could be shifted to the position predicted by D383 and full-length CKI⑀ upon digestion with a 10-fold higher trypsin concentration than previously used. These data are consistent with Ser-368 being an autophosphorylation site.
Autophosphorylation sites in the carboxyl-terminal tail of CKI⑀ have been localized to residues Ser-323 and/or Thr-325 (M1, Fig. 6A), Ser-368 (Fig. 6B), Thr-334, and/or Thr-337 (M2), and Ser-405, Thr-407, and/or Ser-408 (M5) (data not shown). In most cases, the site of autophosphorylation was identified as FIG. 7. CKI⑀ compound mutant MM2 lacking carboxyl-terminal autophosphorylation sites has a marked increase in specific activity. A, mutagenesis of autophosphorylation sites in the carboxylterminal tail of CKI⑀ eliminates known phosphopeptides. Mutant and wild-type kinases were autophosphorylated and subjected to phosphopeptide mapping as described. MM2 is full-length CKI⑀ with mutations S323A, T325A, T334A, T337A, S368A, S405A, T407A, and S408A. B, normalization of mutant and wild-type kinase activity against IB␣. Kinase activities of the MM2 mutant or CKI⑀ wild type were normalized using IB␣ as a substrate. MM2 or wild-type CKIe were incubated with 1.5 g of IB␣, and [␥-32 P]ATP for a 5-min kinase reaction. Reaction products were separated by SDS-PAGE and quantitated by Phospho-rImager analysis. Phosphorylated IB␣ is indicated by an arrow. Sizes are indicated in kilodaltons. C, much less MM2 protein is required to achieve the same level of IB␣ kinase activity as wild-type CKI. Fiftyfold more of each kinase than was input to IB␣ kinase reactions shown in B was analyzed by quantitative immunoblot using UT31 as a primary antibody and 125 I-labeled protein A as a secondary detection step.  (Table  II).
Effect of CKI⑀ Phosphoacceptor Site Mutagenesis-Phosphopeptide mapping data were used to choose mutation sites for generating a nonphosphorylatable tail mutant of CKI⑀. Primers M1, M2, M5, and S368A were used for simultaneous site-directed mutagenesis of CKI⑀. The resulting multiple mutant, MM2 (S323A/T325A/T334A/T337A/S368A/S405A/T407A/ S408A), was sequenced to confirm the presence of planned mutations and the absence of adventitious changes, and it was then expressed, partially purified, and autolabeled. A twodimensional peptide map of MM2 indicates that the these mutations lead to a CKI⑀ molecule with markedly reduced autophosphorylation (Fig. 7A).
To determine whether a lack of tail autophosphorylation correlates with an increase in catalytic activity, wild-type CKI⑀ and the MM2 mutant were normalized by their kinase activities against IB␣ (Fig. 7B). However, when the amount of kinase in each reaction was checked by quantitative immunoblot, a dramatic difference was seen between the amount of mutant MM2 CKI⑀ and wild-type CKI⑀. Kinase activity on IB␣ was normalized to the amount of kinase detected by immunoblot (Fig. 7C) and graphed (Fig. 7D). About 8-fold less MM2 kinase is required to achieve the same amount of IB␣ phosphorylation as wild-type CKI⑀. This activation is more pronounced than was seen by complete elimination of the carboxyl-terminal domain (Fig. 4). DISCUSSION CKI⑀ belongs to a family of ubiquitous protein kinases with an emerging role in regulation of transcription, DNA replication and DNA repair. A mechanism for the potential regulation of CKI⑀ and related CKI␦ has been autophosphorylation. In the current study we have mapped the autophosphorylation sites of CKI⑀ in the carboxyl-terminal inhibitory domain and demonstrated that a multiple phosphorylation site mutant has an 8-fold increase in kinase activity on the substrate IB␣. Additionally, a region between amino acids 349 and 360 was identified as a negative regulator of kinase activity. This constitutively active mutant now allows us to test the role of inhibitory autophosphorylation in the potential biologic functions of CKI⑀ including processes such as circadian rhythm and DNA replication.
It was noted previously that truncation and dephosphorylation of CKI⑀ do not produce the same degree of activation (19). However, this discrepancy was attributed to incomplete truncation of the inhibitory domain or the presence of an inhibitory phosphorylated residue in the kinase domain. The latter possibility has not been ruled out, as the CKI⑀ D319 truncation mutant still autophosphorylates (Fig. 6A) and the D305 truncation mutant is still activated two-fold by phosphatase treatment (19). Further supporting the presence of an inhibitory phosphorylation site in the kinase domain, Kuret and co-workers (28) described two forms of recombinant yeast Cki1 kinase domain; one form was autophosphorylated in the kinase domain and had a 4-fold decrease in activity compared with unphosphorylated Cki1. Thus, inhibition of kinase activity via phosphorylation of the kinase domain may be a common fea-ture of the CKI family. It is notable that CKI is one of the few serine/threonine kinases that do not require phosphorylation on their T-loop for full kinase activity (29). The data therefore suggest that there are inhibitory autophosphorylation sites within the kinase domain of several CKI family members.
Using recombinant mutants of CKI⑀, including carboxyl-terminal truncations and point mutations of putative phosphoacceptor residues in the carboxyl-terminal tail region, a twodimensional peptide mapping approach was used to identify sites of autophosphorylation (Fig. 8). In previous peptide phosphorylation experiments, the preference of CKI for an acidic residue or phosphate group three residues amino-terminal to the phosphoacceptor was well characterized (14, 15, 30 -32). Interestingly, the sites mapped in this study do not match consensus CKI sites. Autophosphorylation sites were found scattered throughout the CKI⑀ carboxyl-terminal tail. In two of the mutants altering CKI⑀ autophosphorylation sites, M2 (T334A,T337A) and M5 (S405A,T407A,S408A), at least two phosphoacceptor residues are oriented 3 residues apart. These sites are in regions of high homology to CKI␦. It is possible that phosphorylation on the amino-terminal residue of the series could be a catalyst for phosphorylation of the phosphoacceptor carboxyl-terminal to the first residue. However, if this were the case, the first phosphorylation event would still have taken place without any upstream acidic region to direct it. It may be that the high local concentration of the tail is more important in determining specific phosphorylation sites. Alternatively, it may be that three-dimensional structure of the tail is more important than upstream acidic character. In support of this model, CKI is able to phosphorylate specific residues in the amino terminus of SV40 large T antigen only in the context of full-length protein. The same T antigen residues were not phosphorylated by CKI when present in peptides or in T antigen domains (24).
It appears that mammalian cells go to considerable lengths to ensure that CKI⑀ remains in a dephosphorylated, active form, suggesting that its activity is either required or modulated by some other means. Thus far, no agents except the phosphatase inhibitors okadaic acid and calyculin A have been identified as instigating CKI⑀ autophosphorylation in vivo. The link between the CKI⑀ and DNA damage-responsive pathways in yeast (3), and recently in mammals (17), and the link between CKI⑀ and circadian rhythm in Drosophila suggest that a DNA damage or circadian rhythm-regulated event triggers upor down-regulation of CKI activity. The identification of CKI⑀ autophosphorylation sites described above may provide a means to determine the role of CKI⑀ autophosphorylation on the in vivo regulation of this enzyme.