Role of COOH-terminal phosphorylation in the regulation of casein kinase I delta.

Casein kinase Iδ is a member of the casein kinase I (CKI) family, a group of second messenger independent protein kinases. We present evidence that the COOH-terminal domain of CKIδ has regulatory properties. CKIδ expressed in Escherichia coli was activated by heparin, as found previously, and by treatment with the catalytic subunit of type-1 protein phosphatase (CS1). Concomitant with activation by CS1, there was a reduction in the apparent molecular weight of CKIδ from 55,000 to 49,000 as judged by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. Truncation of CKIδ by removal of the COOH-terminal 110 amino acids eliminated the ability of CS1 to activate or to increase electrophoretic mobility. Casein kinase I α, a 37-kDa isoform that lacks an extended COOH-terminal domain, was not activated by CS1 or the presence of heparin. However, a chimeric enzyme consisting of CKIα fused to the COOH-terminal domain of CKIδ was activated by both heparin and CS1. Analysis of the effects of CS1 on a series of CKIδ COOH-terminal truncation mutants identified an inhibitory region between His317 and Pro342, which contained six potential phosphorylation sites. From analysis of the specific activites of these truncation mutants, removal of the same region resulted in enzyme with a specific activity nearly 10-fold greater than wild-type. Thus, CKIδ activity can be regulated by phosphorylation of its COOH terminus, which may serve to create an autoinhibitory domain. This mechanism of regulation could have important consequences in vivo.

Casein kinase I␦ is a member of the casein kinase I (CKI) family, a group of second messenger independent protein kinases. We present evidence that the COOHterminal domain of CKI␦ has regulatory properties. CKI␦ expressed in Escherichia coli was activated by heparin, as found previously, and by treatment with the catalytic subunit of type-1 protein phosphatase (CS1). Concomitant with activation by CS1, there was a reduction in the apparent molecular weight of CKI␦ from 55,000 to 49,000 as judged by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. Truncation of CKI␦ by removal of the COOH-terminal 110 amino acids eliminated the ability of CS1 to activate or to increase electrophoretic mobility. Casein kinase I ␣, a 37-kDa isoform that lacks an extended COOH-terminal domain, was not activated by CS1 or the presence of heparin. However, a chimeric enzyme consisting of CKI␣ fused to the COOH-terminal domain of CKI␦ was activated by both heparin and CS1. Analysis of the effects of CS1 on a series of CKI␦ COOH-terminal truncation mutants identified an inhibitory region between His 317 and Pro 342 , which contained six potential phosphorylation sites. From analysis of the specific activites of these truncation mutants, removal of the same region resulted in enzyme with a specific activity nearly 10-fold greater than wild-type. Thus, CKI␦ activity can be regulated by phosphorylation of its COOH terminus, which may serve to create an autoinhibitory domain. This mechanism of regulation could have important consequences in vivo.
Casein kinase I (CKI) 1 is one of two classes of protein kinase discovered over 20 years ago for their ability to phosphorylate of casein in vitro (for reviews, see Refs. 1 and 2). It is now known that CKI and casein kinase II, the other major casein kinase form, are distinct gene products and, by amino acid sequence, are no more related than other protein kinases. CKI has been detected in numerous animal and plant cells including those of mammals, yeast, broccoli (3), Dictyostelium (4), and Paramecium (5). The enzyme is found in cytosolic, membrane, mitochondrial, and nuclear fractions (1). Cytosolic CKI is a monomeric polypeptide of ϳ34 -37 kDa, but nuclear CKI has been reported with molecular masses from 23 to 55 kDa (1). The heterogeneity in size earlier reported for CKI can be explained, at least in part, by the presence of multiple isoforms. In mammals, at least six different genes are known, encoding isoforms termed CKI␣ and -␤ of 37 and 38 kDa (6 -8), CKI␥1, -␥2, and -␥3 of 45, 47, and 51 kDa (9), and CKI␦ of 49 kDa (10). Saccharomyces cerevisiae has four different genes, HRR25 (11), YCK1 and YCK2 (12,13), and YKS1, 2 encoding proteins ranging in size from 57 to 62 kDa. In Schizosaccharomyces pombe, four genes have also been identified, CKi1 and CKi2 (14) and Hhp1 and Hhp2 (15,16), encoding proteins from 42 to 50 kDa. HRR25, Hhp1, and Hhp2 have been implicated in DNA strand break repair, and yeast defective in these genes show hypersensitivity to DNA damaging agents (11,15). Interestingly, CKI␦ shows 73% identity to Hhp1, the highest identity between any mammalian and yeast CKI isoform, raising the possibility that mammalian CKI␦ might have a role in DNA repair. All CKI isoforms have a similar structural architecture consisting of a conserved catalytic domain of ϳ300 amino acids and differently sized NH 2 -and COOH-terminal extensions. The COOH-terminal noncatalytic domains range in size from 13 to 188 amino acids and in general show little amino acid identity between isoforms.
Unlike most protein Ser/Thr kinases, CKI recognizes acidic amino acids in its substrates and has therefore been termed an "acidotropic" protein kinase (26). Synthetic peptide substrates that contain several acidic amino acids NH 2 -terminal to the target residue, such as the D4 peptide (DDDDVASLPGLRRR), are relatively specific CKI substrates (27). These substrates are, however, usually much less efficient than phosphorylated substrates in which the phosphate group is found in the sequence motif (Ser(P)/Thr(P))-Xaa-Xaa-(Ser/Thr)- (26,28). CKI's ability to utilize a phosphate group as a recognition determinant could link its activity to that of other protein kinases, which in turn could be regulated by classical second messengers (29). Some effective CKI substrates do not require prior phosphorylation. The clearest example here is inhibitor-2 of protein phosphatase 1 (30) which contains a cluster of acidic residues NH 2 -terminal to the target site.
Relatively little is known about the control of CKI enzymes. In this report, we define an autoinhibitory region in the COOH terminus of CKI␦ through which activity can be modulated by phosphorylation.
Expression Vector Construction-All COOH-terminal truncation mutants of CKI␦ were sublconed into the pET-8c (32) expression vector as described previously (10). To create the ␣-␦ chimera, mutant CKI␣ cDNA was digested with NdeI and DraI to produce a 974-bp coding region fragment. This fragment encoded the entire amino acid sequence of CKI␣ (amino acids 1-325). To create the ␦ portion of the chimera, the polymerase chain reaction (GeneAMP, Invitrogen) was used to amplify a product of 365 bp from 1251 to 1616 bp of the CKI␦ cDNA (10), according to the manufacturer's instructions. The oligonucleotides 5Ј-CGCTTCGCGAAATCCAGCCACTCGTGGC and 5Ј-CGCTGGATCCTG-GCTGGAAGCCCGGTCA were used for the polymerase chain reaction with the underlined sequence indicating the NruI and BamHI sites of the sense and antisense primers, respectively. This polymerase chain reaction product, which encoded a portion of the COOH-terminal region of CKI␦ (amino acids 319 -428), was digested with NruI and BamHI and ligated to the 974-bp CKI␣ fragment producing a fragment of 1339 bp. The 1339-bp fragment was then ligated to the pET-3c expression vector previously cut with NdeI and BamHI. The entire polymerase chain reaction product and all junctions formed were confirmed by nucleotide sequence analysis (31).
Expression and Purification of CKI␦-The CKI␦ COOH-terminal truncation mutants and the ␣-␦ chimera were expressed, harvested, and lysed as described previously for wild-type CKI␦ (10). Wild-type CKI␦, the ␣-␦ chimera, and the COOH-terminal truncation mutants were purified using a combination of S Sepharose (Pharmacia Biotech Inc.) and ␥-ATP-agarose chromatography, an affinity resin for the purification of protein kinases (33). Supernatant from Escherichia coli cell lysate (25-50 ml) was batch-absorbed for 2 h at 4°C with 2-3 ml of S Sepharose previously equilibrated with buffer A (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 0.1 mM TLCK, and 1 mM dithiothrietol). The resin was collected by centrifugation at 5000 ϫ g for 10 min, transferred to a 5-ml plastic column (Whatman), and washed with 5 column volumes of buffer A. Enzyme was step-eluted using 0.6 M NaCl in buffer A, and fractions containing CKI activity were pooled and diluted to Ͻ0.1 M NaCl. ␥-ATP-agarose was equilibrated with buffer B (50 mM Tris-HCl, pH 7.5, 1.5 mM EDTA, 20 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 0.1 mM TLCK, and 1 mM dithiothreitol), followed by 5 column volumes of 2.5 mg/ml bovine serum albumin in buffer B to block nonspecific binding. The column was washed with buffer B to remove unbound bovine serum albumin, and the pool from the S Sepharose column was applied and recirculated several times to promote binding. After washing the resin with 5 column volumes of buffer B and 5 column volumes of 250 mM NaCl in buffer B, CKI was eluted with 3 column volumes of 5 mM ATP in buffer B. To remove ATP, the eluate, of ϳ3 ml, was reapplied to a 1-ml S Sepharose column and step-eluted with 0.6 M NaCl in buffer A. Fractions of 0.25 ml were collected and assayed for CKI activity. Fractions containing CKI activity were subjected to 12% SDS-PAGE according to the method of Laemmli (34), and protein concentration was determined using the Bradford assay (35). All enzymes were purified close to homogeneity as judged by SDS-PAGE (see Fig. 7).
Protein Kinase Assays-The standard CKI assay, in 25 l, contained 75 mM Tris-HCl, pH 7.5, 6 mM Mg(CH 3 COO) 2 , 1 mM EDTA, 0.4 mM EGTA, 1 mM ␤-mercaptoethanol, and 250 M [␥-32 P]ATP (300 -1500 cpm/pmol). Under standard conditions, the synthetic peptide DDDD-VASLPGLRRR (D4 peptide) was used at 1 mM in the assay (27). ␣-Casein was used at 2 mg/ml. Blank reactions, lacking protein kinase or substrate, were also run. Reactions were allowed to proceed at 30°C for the desired time and then terminated by the addition of EDTA and adenosine to final concentrations of 20 and 1.5 mM, respectively. 32 P incorporation into the peptide was determined by binding to P-81 papers (36) and into ␣-casein by trichloroacetic acid precipitation on Whatman 31-ET paper (37). Microcystin-LR (Sigma), an inhibitor of type 1 and 2 protein phosphatases (38), was included at 1 M in the assay if ␣-casein was used as substrate.
Protein Phosphatase Reactions-CKI␦ was treated with the recombinant ␣ isoform of the catalytic subunit of type 1 serine/threonine protein phosphatase (CS1) (38). The standard phosphatase reaction contained 1 g/ml CS1␣, 50 mM Tris-HCl, pH 7.5, 2 mM MnCl 2 , and 1 mM dithiothreitol at 30°C. To determine the effect of CS1 treatment on CKI␦ activity, recombinant CKI␦ was added to the reaction mix above, and aliquots were removed at the indicated times and assayed for protein kinase activity as described previously. Control reactions were identical except lacking in CS1. To evaluate the phosphorylation state of enzyme treated with CS1, aliquots of the phosphatase and kinase reaction were removed and added to SDS sample buffer (final composition, 62.5 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 5% 2-mercaptoethanol, and 0.25% bromphenol blue). Samples (ϳ0.5 g) were resolved by SDS-PAGE on a 12% gel and detected by staining with Coomassie Blue (34).
Autophosphorylation Reactions-Autophosphorylation of CKI␦ was performed using conditions as described for kinase assays except for the addition of 50 M [␥-32 P]ATP (1000 -5000 cpm/pmol). For rephosphorylation of CKI␦ following CS1 treatment, CKI␦ was incubated at 30°C for 1 h in the kinase assay buffer plus 5 mM ATP and 1 M microcystin-LR. Autophosphorylation reactions were terminated by removal of an aliquot of the reaction and addition of SDS sample buffer as described above. 32 P-Labeled protein was resolved by 12% SDS-PAGE and detected by staining with Coomassie Blue and autoradiography. For quantitation of phosphorylation, 32 P-labeled protein was excised from the gel and subjected to Cerenkov scintillation counting.
Miscellaneous Methods-For phosphoamino acid analysis, CKI␦ was allowed to autophosphorylate in the presence of 50 M [␥-32 P]ATP (1000 -5000 cpm/pmol) as described above. 32 P-Labeled protein was subjected to 12% SDS-PAGE, transferred to a polyvinylidine difluoride membrane (Millipore), and detected by staining with Coomassie Blue and autoradiography. Radioactive protein bands were excised, and the protein was hydrolyzed in 5.7 N HCl for 90 min at 110°C. Phosphoamino acids were resolved by thin layer electrophoresis for 67 min at 500 V in 5% acetic acid, 0.5% pyridine, pH 3.5, and detected by 0.5% ninhydrin and autoradiography as described previously (39). To localize autophosphorylation sites, CKI␦ was allowed to autophosphorylate in the presence of 50 M [␥-32 P]ATP (1000 -5000 cpm/pmol), and 32 Plabeled protein was digested for ϳ12 h with endolysine C at a ratio of 1:20 in 75 mM Tris-HCl, pH 8.0. The mixture was subjected to 20% SDS-PAGE, transferred to polyvinylidine difluoride membrane and autoradiography performed. 32 P-Labeled species were excised and subjected to the Edman degradation reaction for NH 2 -terminal protein sequencing (40).
Enzymes and Materials-Purified CKI␣ (7) was supplied by Kenton Longenecker. Recombinant ␣ isoform of type 1 protein phosphatase (38) was a gift from Dr. A. A. DePaoli-Roach. ␥-ATP-agarose was a gift from Dr. L. Graves, University of North Carolina. Restriction enzymes and all molecular biology reagents were from New England Biolabs and Life Technologies, Inc. The Sculptor mutagenesis kit and the oligonucleotide-directed in vitro mutagenesis system were obtained from Amersham Corp. DNA sequencing kits were from U. S. Biochemical Corp. Oligonucleotides were synthesized on an Applied Biosystems DNA synthesizer model 380A, and peptides were synthesized and purified as described previously (26). Protein sequence was determined by automated Edman degradation on a Porton Instruments model 2090 integrated microsequencing system. Radioactive nucleotides were from Du-Pont NEN, and 125 I-labeled protein A was from ICN. Protease inhibitors were purchased from Boehringer Mannheim, and S Sepharose was from Pharmacia. Casein, phosvitin, and heparin were from Sigma.

RESULTS
Autophosphorylation of Casein Kinase I␦-CKI␦ can be expressed as a soluble, active enzyme in E. coli (10). However, CKI␦ expressed in E. coli had an apparent molecular weight (M r ) of ϳ55,000 on SDS-PAGE, significantly larger than the M r of 49,121 predicted from its amino acid sequence (Fig. 1, lane  1). Since phosphorylation of proteins is known to retard electrophoretic mobility, we treated CKI␦ with CS1. The apparent M r of CKI␦ was reduced from ϳ55,000 to ϳ49,000 (Fig. 1, lane  2). Moreover, if CKI␦ was subsequently incubated with ATP, Mg 2ϩ , and microcystin to inhibit the phosphatase, its electrophoretic mobility decreased yielding a polypeptide apparent M r 55,000 (Fig. 1, lane 3). Phosphoamino acid analysis of this autophosphorylated CKI␦ revealed phosphoserine and phosphothreonine but no phosphotyrosine (data not shown). Furthermore, expression of a "kinase-dead" mutant of CKI␦ (K 38 N) resulted in inactive protein with apparent M r ϳ49,000, 3 suggesting that phosphorylations affecting mobility were due to autophosphorylation within the E. coli cells. We conclude that recombinant CKI␦ is a phosphoprotein and that its phosphorylation state affects its migration on SDS-PAGE. At earlier stages of the CKI␦ phosphorylation reaction, several intermediate species could be identified with M r values in the range of 49,000 -55,000, suggesting phosphorylation at multiple sites (data not shown). To determine whether the phosphorylation state of CKI␦ affected its activity toward exogenous substrates, recombinant CKI␦ was treated with CS1 and protein kinase activity was measured as described under "Experimental Procedures." Incubation with CS1 led to progressive activation of the enzyme with maximal activation of 2-3-fold after 1 h (Fig.  2). This activation correlated with dephosphorylation of CKI␦ as judged by its conversion to the 49-kDa species (data not shown). We conclude that the autophosphorylation sites involved in altering electrophoretic mobility and activity were already fully phosphorylated, presumably inside the E. coli cells.
However, even though the 55-kDa form of CKI␦ was already phosphorylated, it could still undergo additional autophosphorylation in vitro, to a stoichiometery of 2-3 mol of phosphate/ mol of protein, again exclusively at Ser and Thr residues (data not shown). This latter autophosphorylation did not change the electrophoretic mobility of the enzyme or significantly affect its activity (data not shown). Enzyme allowed to autophosphorylate in the presence of [␥-32 P]ATP was digested with endolysine C, and the resulting peptides were resolved on SDS-PAGE. After transfer to a polyvinylidine difluoride membrane, the predominant 32 P-labeled species was subjected to protein sequencing. A unique CKI␦ sequence, FGA, was obtained, which corresponds to amino acids Phe 295 -Ala 297 . The next lysine residue occurs at Lys 368 , thus localizing these in vitro autophosphorylation sites to the region from residue 295 to 368. In other experiments, the initial rate of CKI␦ autophosphorylation was shown to be first order with respect to protein concentration over a 20-fold range of enzyme (2.5-50 g/ml) (data not shown). Thus, the specific activity is independent of concentration, suggesting that in vitro autophosphorylation of CKI␦ occurs by an intramolecular mechanism.
Regulatory Role of the COOH Terminus of CKI␦-Previously, we reported that CKI␦ activity toward certain substrates, such as the D4 peptide, could be increased 4 -5-fold in the presence of heparin, whereas a truncated form of the enzyme (CKI␦⌬317), lacking the COOH-terminal 111 amino acids, was unaffected (10). Similarly, D4 peptide phosphorylation by CKI␣, an isoform with a minimal COOH-terminal extension, was not activated by heparin and in fact was inhibited by ϳ50% (7). Therefore, we constructed an ␣-␦ chimeric enzyme, CKI␣-␦, consisting of the entire 325 amino acids of CKI␣ fused to the COOH-terminal 110 amino acids of CKI␦. The chimeric enzyme had a calculated M r of 49,609, but, like CKI␦, it had an apparent M r of ϳ55,000 kDa on SDS-PAGE, suggesting that it had undergone autophosphorylation in E. coli (data not shown). Using the D4 peptide as substrate, CKI activity was measured either in the presence of heparin or after treatment with CS1 (Fig. 3). CKI␣ was inhibited by heparin and showed only 39% activation by CS1. CKI␦⌬317 was not activated by heparin or CS1. CKI␦, in contrast, was activated 3-4-fold by both heparin and CS1 as was CKI␣-␦. This result indicates that the COOHterminal region of CKI␦ is necessary and sufficient to confer heparin and CS1 activation to the enzyme.
Since there was a correlation between heparin and CS1 activation of CKI␦ (Fig. 3), we determined whether the phosphorylation state of CKI␦ affected its activation by heparin. CKI␦ was treated with CS1, and then protein kinase activity was measured with increasing concentrations of heparin. Once CKI␦ was activated by CS1, it could no longer be activated by heparin (Fig. 4). This finding suggests that heparin and CS1 activate CKI␦, at least in part, via a common mechanism.
Localization of Autophosphorylation Sites and an Autoinhibitory Domain in CKI␦-To localize the sites already phosphorylated in CKI␦ purified from E. coli, three COOH-terminal truncations were created by site-directed mutagenesis (Fig. 5). In addition, we already had the CKI␦⌬317 construct (10). These mutants, designated CKI␦⌬317, CKI␦⌬327, CKI␦⌬342, and 3 P. R. Graves and P. J. Roach, unpublished results. CKI␦⌬363, were individually expressed and purified from E. coli (Fig. 6). Each mutant protein was treated with CS1 and subjected to SDS-PAGE. While CKI␦⌬317 showed no change in electrophoretic mobility after CS1 treatment, CKI␦⌬327, CKI␦⌬342, and CKI␦⌬363 all showed increased mobility (Fig.  6). Thus, sites contributing to the anamolous migration of CKI␦ on polyacryamide gels are distributed throughout the COOH terminus. The effects of CS1 on activity were also monitored. CKI␦⌬317, as shown previously in Fig. 3, was not activated by CS1 (Fig. 7A). As the COOH terminus was progressively extended, a greater degree of activation was observed, 2-fold in CKI␦⌬327 and 4-fold in CKI␦⌬342. However, the addition of further COOH-terminal sequences, in CKI␦⌬363 and full-length CKI␦, correlated with no greater activation (Fig. 7A). These results suggest that there are minimally two inhibitory autophosphorylation sites, one between residues His 317 and Pro 327 , and another between residues Pro 327 and Pro 342 . There are six candidate sites in these regions, Ser 318 , Thr 323 , Ser 328 , Thr 329 , Ser 331 , and Thr 337 (Fig. 5). Also, not all of the sites that influence electrophoretic mobility affect activity.
COOH-terminal deletion itself affected the activity of CKI␦ as judged by the specific activities of the purified truncated proteins (Fig. 7B). Thus, CKI␦⌬317, in which essentially all of the COOH-terminal tail was removed, had a specific activity nearly 10-fold greater than that of wild-type enzyme. CKI␦⌬327 had a specific activity 5-fold greater than wild-type. However, tuncations up to residue 342 did not alter the specific activity compared with wild-type enzyme. Thus, CKI␦ contains an autoinhibitory region between His 317 and ProP 342 (Fig. 5), the same region identified as inhibitory from CS1 treatment (Fig. 7A). DISCUSSION Casein kinase I has, for many years, been considered to be a constitutively active enzyme since it is spontaneously active after isolation from native tissues or after expression of recombinant enzyme in a prokaryotic system. Moreover, this spontaneous activity is not generally found to be affected by second messengers or by association with any known proteins. There is a report of the inhibition of a 37-kDa CKI species by phosphatidylinositol bisphosphate (41), although this result has been questioned by others (42). In the present study, we provide evidence that the CKI␦ isoform is a regulatable enzyme whose COOH terminus acts as an autoinhibitory domain in which phosphorylation at Ser and Thr residues causes inactivation of the enzyme.
The CKI family consists of isoforms with a conserved catalytic domain and variably sized NH 2 -and COOH-terminal extensions. The NH 2 -terminal extensions are less than 15 amino acids, except in the case of the recently identified ␥ isoforms (9), with lengths of ϳ43 amino acids, and the yeast Yck1p and Yck2p enzymes (12,13), with ϳ73-amino acid extensions. The COOH-terminal domains of most CKI isoforms are considerably larger. For example, two of the mammalian (9, 10) and six of the yeast (11-16) CKI isoforms have COOH-terminal tails of at least 100 amino acids. CKI␣ and -␤, which are most likely the commonly studied 37-kDa forms of CKI, have COOH-terminal domains of only 13-25 amino acids (6).
The first indication that the COOH-terminal region of CKI␦ might be regulatory came from earlier studies with heparin. Effects of heparin on CKI isoforms are complicated and substrate-specific (see Ref. 10). However, we observed that fulllength CKI␦ was activated by heparin when the D4 peptide was used as a substrate, whereas mutant CKI␦ lacking the COOH terminus was insensitive (10). Furthermore, the CKI␣ isoform, with its minimal COOH-terminal extension, was not activated by heparin and in fact was partially inhibited (Fig. 3; Ref. 7). Therefore, we asked if the COOH-terminal domain of CKI␦ conferred the heparin activation and tested the hypothesis by creating a chimeric kinase consisting of CKI␣ fused to the portion of the ␦ isoform removed by truncation. Heparin now activated the ␣-␦ chimeric enzyme. Whatever heparin does mechanistically, the COOH-terminal domain of CKI␦ is clearly critical, being both necessary and sufficient for heparin activation.
That CKI, like most protein kinases, can autophosphorylate has been known for many years but had never been linked with any change in enzyme activity (43,44). We previously reported that recombinant CKI␦ purified from E. coli could autophos-phorylate without change in electrophoretic mobility (10). It is now apparent that CKI␦ does contain autophosphorylation sites that affect both electrophoretic mobility and, more importantly, activity toward exogenous substrates. The key observation was that treatment of purified recombinant CKI␦ with type 1 protein phosphatase activated the enzyme and reduced its apparent M r from 55,000 to its predicted value of 49,000. Thus, the sites that influence both activity and mobility are already modified, the phosphorylation presumably occurring inside the E. coli cells. This phosphorylation could be due to autophosphorylation or the action of E. coli protein kinases. Arguing against the latter possibility is the observation that a kinase-dead mutant of CKI␦ (K 38 N) had apparent M r 49,000, indicative of the protein not being phosphorylated. Interestingly, a nuclear form of CKI (N1) was reported to have an apparent M r of 55,000 on SDS-PAGE (45) and not to autophosphorylate, which could be possible if the enzyme were already fully modified.
Using a series of COOH-terminal truncation mutants, we were able to localize the phosphorylation sites responsible for anomalous electrophoretic migration as being COOH-terminal to His 317 , and several different sites within this region contribute to this behavior. Similar results were observed with a completely different isoform, S. pombe CKi1, in which anomalous migration on SDS-PAGE was attributed to the COOHterminal domain (46). By analyzing the ability of protein phosphatase to activate the COOH-terminally truncated forms of CKI␦, the inhibitory phosphorylation was localized to a region between His 317 and Pro 342 , which contains six potential phosphorylation sites. Although we did not determine which of these six sites has the greatest inhibitory affect on the enzyme, multiple autophosphorylation sites are involved. Exactly the same region, His 317 to Pro 342 , was inferred to be autoinhibitory from consideration of the specific activities of the truncation mutants. Thus, activation of the enzyme by truncation may result from removal of inhibitory autophosphorylation sites. Heparin activates CKI␦ toward the D4 peptide by a mechanism that requires the COOH-terminal region. Possibly, heparin interacts with the COOH terminus, which carries a high positive charge, to cause an overall conformational change that mimics dephosphorylation. Heparin is unlikely to be a physiological regulator of the enzyme, but we cannot exclude the possibility that some other compound interacts with the regulatory COOH terminus. Likewise, it is not known in vivo whether the inhibitory phosphorylations discussed above would result from autophosphorylation or the action of a separate protein kinase. There is a parallel in the extracellular signal-regulated/mitogen-activated protein kinase enzymes, which can activate by autophosphorylation (47) even though enzymes of the MAPK or ERK family are thought to be responsible physiologically (48,49).
Other CKI enzymes may also be regulated via their COOHterminal domains, even though these have no sequence similarity whatsoever. For example, CKI␥3 can be activated 3-4fold by type-1 protein phosphatase concomitant with a shift in its apparent molecular weight from 60 to 55 kDa on SDS-PAGE. 4 Similarly, the S. pombe CKi1 isoform undergoes an inhibitory autophosphorylation that was localized to its COOH terminus (46). Truncation of the enzyme and removal of the COOH-terminal domain resulted in a 3-fold activation in the catalytic rate of the enzyme. Autophosphorylation, the majority of which was localized to the COOH-terminal domain, resulted in a 4-fold decrease in the affinity for protein substrate (46). Inhibitory COOH-terminal phosphorylation could therefore be 4 L. Zhai, P. R. Graves, and P. J. Roach, unpublished results.

FIG. 7. Localization of an autoinhibitory region in the COOHterminal domain of CKI␦.
A, the indicated CKI␦ COOH-terminal truncation mutants or wild-type CKI␦ (WT) were treated with CS1 for 60 min, and their activities toward the D4 peptide were measured. The activation is normalized to the activity measured without dephosphorylation. B, the specific activity of the untreated samples using the D4 peptide as substrate. a common regulatory mechanism for CKI isoforms.
Given the unique substrate recognition characteristics of CKI, as discussed in the Introduction, it is of interest to survey the sequences surrounding the potential autophosphorylation sites between His 317 and Pro 342 (Fig. 5). None is preceded by a cluster of acidic residues, precluding this modality for recognition. The other motif recognized by CKI requires prior phosphorylation and so cannot account for the initial autophosphorylation. In any case, only one such site is available in this region: phosphorylation at Ser 328 would create a potential site at Ser 331 . We conclude, therefore, that different substrate recognition constraints must be operative for the autophosphorylation reaction, which is likely to be an intramolecular process, than when the enzyme acts on exogenous substrates. Interestingly, introduction of phosphate at five of the sites between His 317 and Pro 342 would generate a sequence (Ser(P)/Thr(P))-Xaa-Xaa-Yaa, (where Yaa is not Ser or Thr) that could then act as a pseudosubstrate. Better understanding of the interactions between the catalytic and regulatory domains of CKI will have to await the solution of the three-dimensional structure of full-length CKI.