14-3-3 Is Phosphorylated by Casein Kinase I on Residue 233

14-3-3 proteins mediate interactions between proteins involved in signal transduction and cell cycle regulation. Phosphorylation of target proteins as well as 14-3-3 are important for protein-protein interactions. Here, we describe the purification of a protein kinase from porcine brain that phosphorylates 14-3-3 ζ on Thr-233. This protein kinase has been identified as casein kinase Iα (CKIα) by peptide mapping analysis and sequencing. Among mammalian 14-3-3, only 14-3-3 τ possesses a phosphorylatable residue at the same position (Ser-233), and we show that this residue is also phosphorylated by CKI. In addition, we show that 14-3-3 ζ is exclusively phosphorylated on Thr-233 in human embryonic kidney 293 cells. The residue 233 is located within a region shown to be important for the association of 14-3-3 to target proteins. We showed previously that, in 293 cells, only the unphosphorylated form of 14-3-3 ζ associates with the regulatory domain of c-Raf. We have now shown thatin vivo phosphorylation of 14-3-3 ζ at the CKIα site (Thr-233) negatively regulates its binding to c-Raf, and may be important in Raf-mediated signal transduction.

The name 14-3-3 was given to an abundant mammalian brain protein family due to its particular migration pattern on two-dimensional DEAE-cellulose chromatography and starch gel electrophoresis (1). The proteins were subsequently named by Greek letters according to their respective elution positions on HPLC. 1 Seven mammalian forms of 14-3-3 (␤, ␥, ⑀, , , , and ) have been found, and two are specifically expressed in T cells () and epithelial cells (). The 14-3-3 family is highly conserved, and individual proteins are either identical or contain a few conservative substitutions over a wide range of mammalian species. All are dimeric proteins with a pI around 4.5 and a subunit mass of 30 -33 kDa. Homologues of 14-3-3 proteins have also been found in a broad range of eukaryotic organisms.
Although the exact function of 14-3-3 is not known, various biological activities have been ascribed for 14-3-3: activation of tyrosine and tryptophan hydroxylases (2), regulation of protein kinase C (3), stimulation of calcium-dependent exocytosis (4), cofactor activity for ADP-ribosylation by Pseudomonas aeruginosa exoenzyme S (5), and a role in cell cycle control (6). New findings have suggested many additional roles for the 14-3-3 family, in particular mediating interactions between components involved in intracellular signal transduction (7). The discovery of the interaction of specific 14-3-3 proteins with Raf (3,8,9) generated much interest in the 14-3-3 family. Whether 14-3-3 directly activates Raf is still controversial, and activation of Raf by 14-3-3 may in fact be due to stabilization rather than stimulation of Raf activity (10). However, it has been shown that dimerization may provide a mechanism for Raf activation (11,12), and 14-3-3 may be involved in this process (13). 14-3-3 have also been shown to interact with other important signaling proteins including polyoma middle T antigen (14), Cdc25 phosphatases (15), protein kinase C (16), Cbl (17), PI 3-kinase (18), Bcr and Bcr-Abl (19), KSR (20), and insulin-like growth factor I and insulin receptor substrate I (21). 14-3-3 proteins form homo-and heterodimers in cells (22). Since different signaling proteins have been shown to associate with distinct 14-3-3 isoforms, heterodimeric 14-3-3 could act as a scaffold protein to mediate the formation of protein complexes. Indeed, it has been shown that Raf can form a complex with Bcr (23) or A20 (24), which is mediated in both cases by 14-3-3. The crystal structures of 14-3-3 (25) and (26) showed they are highly helical proteins, and the dimer forms a large negatively charged channel, the interior of which contains residues that are almost invariant throughout the family. The specificity of interaction of each 14-3-3 protein with diverse target proteins may involve the outer surface of the protein. 14-3-3 dimerization has been shown to be essential for target binding (17,27).
It has been reported that target protein phosphorylation is important for 14-3-3 binding to tryptophan hydroxylase (28), nitrate reductase (29), keratin (30), BAD protein (31), Cbl (32), and insulin-like growth factor I receptor and insulin receptor substrate I (21). In addition, phosphatase treatment of Raf-1 and Bcr inhibits their associations with 14-3-3 in vitro (33). Analysis of the major phosphorylation site of Raf has led to the identification of a novel sequence motif RSXS P XP (where S P is phosphoserine) that may represent a conserved interaction sequence within 14-3-3-binding proteins (34).
14-3-3 was shown to be phosphorylated in human embryonic kidney cells, and only the unphosphorylated form bound to the N-terminal regulatory domain of Raf (35). Therefore, the phosphorylation of 14-3-3 may also play an important role in the regulation of protein complex formation, and therefore in signal transduction. Other 14-3-3 isoforms have been shown to be phosphorylated. 14-3-3 is phosphorylated on Ser residues and on Ser/Tyr residues in vivo by the kinase activities of Bcr and Bcr-Abl, respectively (19). 14-3-3 also binds to Bcr, but is not phosphorylated (19). 14-3-3 ␤, , and are phosphorylated in vitro by a sphingosine-dependent kinase (36). In all cases described above, the phosphorylation sites in 14-3-3 were not identified. In addition, some 14-3-3 forms are phosphorylated on Ser-64 by protein kinase C at a low stoichiometry (37). Moreover, 14-3-3 ␤ and are highly phosphorylated in brain on Ser-185 in an SPEK motif, which is a motif unique to these two isoforms (38).
In conclusion, the regulation of 14-3-3-mediated protein complex formation may be regulated by the ratio of homo-and heterodimers in cells, and by the phosphorylation of 14-3-3 targets as well as the phosphorylation of 14-3-3 itself. Therefore, the identification of the protein kinases that phosphorylate 14-3-3 proteins is important in the study of the role of 14-3-3 in signal transduction. 14-3-3 Recombinant Proteins-14-3-3 ⑀ was purified as a maltosebinding protein as described (39).
The cDNA corresponding to 14-3-3 ␥ has been cloned from a mouse brain cDNA library. 2 14-3-3 cDNA from this clone was amplified by PCR using 2 oligonucleotides (5Ј-GGCGGATCCATGGTGGACCGC-GAGCAACTAGTGC-3Ј) and (5Ј-GGCGAATTCTTAGTTGTTGCCT-TCGCCGCCGTGGTC-3Ј) to create a 5Ј BamHI site and a 3Ј EcoRI site (both are underlined in sequences). Amplified cDNA was inserted in a pGEX-2T vector (Pharmacia) at BamHI/EcoRI restriction sites to express it as a GST fusion protein.
Bacteria carrying 14-3-3 constructs were grown overnight at 37°C in LB medium containing 50 g/ml ampicillin and were diluted the following day (1/10) in the same buffer. Culture was then continued until the optical density of the bacterial growth reached 0.8. Expression of GST-14-3-3 was induced with 0.5 mM isopropyl ␤-D-thiogalactopyranoside for 3 h, and the fusion protein was purified by using glutathione-Sepharose beads (Sigma). GST was removed by digestion with thrombin (Sigma) for 1 h at room temperature, and 14-3-3 was further purified by FPLC on a MonoQ column (Pharmacia).
Kinase Assay-5 l aliquots of column fractions or purified CKI were added to a solution of 20 mM HEPES (pH 7.2), 1 mM DTT, 10 mM MgCl 2 , 20 M [␥-32 P]ATP (8 -10 Ci/mmol) containing 1.5 g of purified 14-3-3, to a final volume of 30 l. After incubation at 30°C for 20 min, the reaction was stopped by the addition of electrophoresis sample buffer and analyzed on 12.5% SDS-PAGE. Gels were stained with Coomassie Blue and autoradiographed.
Photolabeling of the 14-3-3 Kinase-The labeling of the 14-3-3 kinase with 8-azido-[␥-32 P]ATP was performed as described (41) with some modifications. An aliquot containing the 14-3-3 kinase was incubated in 50 l of buffer A (pH 7.5) containing 1 mM DTT, 5 Ci of 8-azido-[␥-32 P]ATP (ICN) (with or without 300 M 8-azido-ATP; Sigma) for 5 min at room temperature prior to irradiation with an ultraviolet lamp at 254 nm. Samples were then analyzed on 12.5% SDS-PAGE after the addition of electrophoresis sample buffer. Gels were stained with Coomassie Blue and autoradiographed.
In Gel Digestion of Phosphorylated Proteins-Phosphorylated 14-3-3 (10 -20 g) was run on 12.5% SDS-PAGE mini-gels. Gel were stained for 5-10 min and then destained for the minimum time. The stained bands were excised and subjected to trypsin (Worthington, TPCKtreated) digestion. The extracts were then dried in a SpeedVac vacuum centrifuge, and made up to the injection volume with water for on-line liquid chromatography MS.
Mass Spectrometry (MS) of Phosphorylated Peptides-Electrospray MS of in gel digested phosphoprotein and solid phase sequencing on arylamine membrane were carried out as described (42).

RESULTS
The brain-specific phosphorylation of 14-3-3 ␤ and at the SPEK motif (38) could be due to a proline-directed kinase such as cyclin-dependent kinase or mitogen activated kinase. Therefore, initial experiments were designed to phosphorylate 14-3-3 by different proline-directed kinases as well as other kinases. Experiments were performed using purified protein kinases or by immunoprecipitation with specific antibodies. None of them was able to significantly phosphorylate 14-3-3 in vitro (Table  I). We then attempted to purify the protein kinase from mammalian brain and we found that 14-3-3 is phosphorylated in vitro on Thr-233 by a protein kinase (termed "T233 kinase") present in mammalian brain (42). In the present study, we identify this kinase as casein kinase I␣.
Purification of the T233 Kinase from Pig Brain-No kinase activity toward recombinant 14-3-3 was detectable in crude extract of pig brain (Fig. 1A, lane T), possibly due to the presence of inhibitors as already shown for several other kinases. The extract (lane T) was loaded on a SP-Sepharose column, and bound proteins were eluted using a salt gradient. Under these conditions, we recovered a kinase activity eluting in fractions 16 -18 at 0.35 M NaCl, which phosphorylated 14-3-3 (Fig. 1A). The kinase activity from the SP-Sepharose was eluted from an Affi-Gel Blue column in fractions 32-42 between 0.6 and 0.8 M NaCl (Fig. 1B). Fractions containing the kinase activity were pooled and chromatographed on a Mono S column. The kinase eluted in fractions 12-14 at 0.55 M NaCl (Fig. 1C) and was subjected to gel filtration chromatography (Fig. 1D). The absolute levels of kinase activity at each step of chromatography are not known, and no specific activity can be measured because of the high sensitivity of the kinase to salt concentration as shown in Fig. 1, B and C, before (lane b) and after (lane a) dialysis to remove salt.
The T233 Kinase Is a Protein of 38 -40 kDa-The chromatographic procedures did not lead to a complete purification of the T233 kinase, and a few protein bands remained. Nevertheless, one major protein of 38 kDa was visualized with silver staining (Fig. 2A). Photoaffinity labeling of the kinase using 8-azido-[␥-32 P]ATP was performed to establish its molecular weight. This revealed the presence of one radiolabeled band around 40 kDa (Fig. 2B). The M r of the kinase after gel filtration was 30 -35 kDa (Fig. 1D). When gel filtration was performed at low (50 mM) NaCl, the kinase eluted from the column with a lower apparent M r , suggesting that some nonspecific interactions with the column occurred (data not shown). SDS-PAGE analysis of the kinase activity eluting at an apparent M r of 30 -35 kDa from the gel filtration column revealed a phosphorylated band of 40 kDa, corresponding possibly to the autophosphorylated form of the kinase (Fig. 1, B and D). We therefore concluded that the kinase has an M r of 38 -40 kDa.
Identification of the 38-kDa Protein as Casein Kinase I␣-The fractions containing the T233 kinase activity from the gel filtration column were then separated on SDS-PAGE ( Fig. 2A). The 38-kDa protein band was digested in gel with trypsin, and the mass of each peptide was measured by electrospray mass spectrometry. Analysis of the peptide mass map using the "Peptidesearch" program identified the 38-kDa protein as casein kinase I␣ (CKI␣) (Fig. 3A).
Is CKI Identical to the T233 Kinase?-To be certain that the T233 kinase was CKI, and not a copurifying protein kinase, in vitro kinase assays were performed using purified mammalian CKI and recombinant yeast CKI (Fig. 3B). Both kinases phosphorylated 14-3-3 as well as the positive control proteins, i.e. casein and phosvitin. Moreover, the T233 kinase phosphorylated both the CKI substrates. The residue in 14-3-3 phosphorylated by mammalian and yeast CKI was in both cases identified as Thr-233, since the mutant T233A was not phosphorylated.

14-3-3 Is Phosphorylated in Vivo Exclusively on Thr-233 in 293
Cells-In human embryonic kidney 293 cells, only one tryptic peptide corresponding to the C terminus of 14-3-3 was phosphorylated (Fig. 4A). Sequencing of the phosphopeptide revealed that only Thr-233 is phosphorylated (Fig. 4B). We have previously shown that in these cells only the unphosphorylated form of 14-3-3 bound to Raf (35). Therefore, together with the present results, we conclude that in vivo, when phosphorylated on Thr-233, 14-3-3 does not bind to the N-terminal regulatory domain of c-Raf.

DISCUSSION
Proteins of the 14-3-3 family bind in a phosphorylation-dependent manner to several target proteins. This may involve a novel consensus sequence RSXS P XP, where S P is phosphoserine (34). Indeed, most 14-3-3-binding proteins contain a putative consensus motif (7,34), and it has been shown that for association with 14-3-3, this sequence must be phosphorylated in nitrate reductase (29) and BAD (31). However, the requirement for phosphorylation at this motif is still controversial for Raf because recombinant Raf purified from bacteria, which has been shown by electrospray MS not to be phosphorylated, 3 is still able to bind 14-3-3. However, the phosphorylation of the motif may increase the affinity between Raf and 14-3-3.
The domain in 14-3-3 which is involved in target binding is not yet well defined. However, using deletion mutants of 14-3-3, Luo et al. (27) showed that the C-terminal 65 residues (176 -245) of 14-3-3 were sufficient to interact with Raf. With a similar approach, Ichimura et al. (44) found that the last C-terminal 76 residues (170 -246) in 14-3-3 were essential for the binding to phosphorylated tryptophan hydroxylase, in particular the amino acids 171-213. Using the yeast two-hybrid system, the ligand-activated glucocorticoid receptor has been shown to interact to the C terminus (residues 187-246) of 14-3-3 (45). Liu et al. (17) showed that the last 15 residues (230 -245) of 14-3-3 were required for efficient binding to Cbl, Raf, and PI 3-kinase. Therefore, the C terminus of 14-3-3 is essential for interaction with target proteins. As shown in the crystal structure of 14-3-3 (25) and (26), this region is not highly ordered.
In this study, we have shown that 14-3-3 is phosphorylated on Thr-233 in the C terminus. This site is accessible at the surface of the dimer, and we found that the dimer is phosphorylated in vitro (data not shown). The protein kinase responsi-3 S. Howell and A. Aitken, unpublished results.

TABLE I In vitro phosphorylation of 14-3-3 by different protein kinases
Purified protein kinases (p) were used to phosphorylate in vitro recombinant 14-3-3 . In some cases, protein kinases were immunoprecipitated (IP) from rat brain before doing an in vitro kinase assay with recombinant 14-3-3 as a substrate. No (Ϫ) or a weak (ϩ/Ϫ) 14-3-3 phosphorylation was detected with all the kinases tested. All kinases were assayed with appropriate control substrates.

Kinase
Method Phosphorylation ble has been identify as CKI by the following criteria: 1) mass spectrometric and sequencing analysis of the purified protein revealed that the kinase was CKI␣, 2) this kinase phosphorylated specific substrates of CKI such as casein and phosvitin, and 3) CKI from mammalian and from S. pombe phosphorylated 14-3-3 and at residue 233. CKI are a family of ubiquitous monomeric Ser/Thr protein kinases ranging in size from 25 to 55 kDa (46). They were FIG. 1. Purification of the T233 kinase from pig brain. A, the extract (lane T) from pig brains was loaded on a 100-ml FF SP-Sepharose column (Pharmacia) pre-equilibrated with buffer A. The bound proteins were eluted using a FPLC system (Pharmacia) with a gradient of 0 -1 M NaCl, and 13 ml fractions were collected (top). The flow-through (lane FT) and the fractions were assayed for their phosphorylating activity toward recombinant 14-3-3 (bottom). B, the peak fractions (80 ml) corresponding to the 14-3-3 kinase activity (lane b) were dialyzed into buffer A (lane a) and loaded on a 35 ml Affi-Gel-Blue column (Bio-Rad) pre-equilibrated with buffer A. The bound proteins were eluted with a gradient of 0 -1 M NaCl (top), and fractions of 5 ml were collected and assayed for their ability to phosphorylate 14-3-3 (bottom). A 40-kDa protein co-eluted with the T233 kinase activity and may represent the autophosphorylated form of the protein kinase (kinase?). C, active fractions (55 ml) were pooled (lane b), dialyzed into buffer A (lane a), and loaded onto a 1-ml Mono S column (Pharmacia) pre-equilibrated with buffer A. The bound proteins were eluted with a gradient of 0 -1 M NaCl (top). Fractions of 0.5 ml were collected and tested for 14-3-3 phosphorylation (bottom). D, active fractions were pooled (2.5 ml) and directly loaded on a Sephacryl S-100 column (Pharmacia) equilibrated with buffer A containing 0.2 M NaCl. Standard molecular sizes indicated in kilodaltons (kDa) are for albumin (66), carbonic anhydrase (29), and cytochrome c (12.4) (top). Fractions of 1 ml were collected and aliquots were assayed for their ability to phosphorylate 14-3-3 (bottom). The protein of 40 kDa may be the autophosphorylated form of the T233 kinase (kinase?). originally described as preferring acidic substrates such as casein and phosvitin. CKI is found in the cytosol, in the nucleus, and associated with the membrane. In vitro CKI substrates include cytosolic, cytoskeletal, and membrane-associated proteins. Components implicated in protein synthesis are also CKI substrates. There is evidence for in vivo phosphorylation of simian virus 40 large T antigen (47), p53 (48), the p75 tumor necrosis factor receptor (49), DARPP32 (a dopamineand cAMP-regulated phosphoprotein; Ref. 50), and the yeast plasma membrane H ϩ -ATPase (51).
CKI homologues in yeast may regulate aspects of cellular DNA metabolism and are implicated in DNA repair (60,61). It is interesting to note that the 14-3-3 homologues in S. pombe, rad24 and rad25, are required for the DNA damage checkpoint (6). We also showed that 14-3-3 from Saccharomyces cerevisiae, BMH1 and BMH2, are phosphorylated at a site equivalent to residue 233 (42). Interestingly, the existence of a link between 14-3-3 and CKI in a cell-cycle checkpoint and growth control has been reported (62).
CKI can be negatively regulated by phosphatidylinositol 4,5bisphosphate (57,63), which suggests a novel mechanism for regulating formation of signaling complexes mediated by 14-3-3 proteins through agonists that regulate phosphatidylinositol turnover. In this context, it is interesting to note that 14-3-3 binds to PI 3-kinase and inhibits its activity (18).
Of all the kinases tested (Table I), none phosphorylated 14-3-3 . Some of those (Mos and Raf) contain a putative motif for 14-3-3 binding (34). Moreover, PCTAIRE 1 and 2 (64) were also tested because they both contain a putative RSXS P XP motif: RSSSMP (477-482) and RNSSYP (504 -509), respectively. CKI␣ (52) contains a potential interaction site for 14-3-3, RTSLP (216 -220), and the possibility of a complex between CKI␣ and 14-3-3 is currently being investigated. CKI␤ and ␥ (52) contain the sequence RGSLP at an equivalent position, and CKI␦ and ⑀ (53,54) LGSLP. We will also investigate the ability of the different forms of CKI to phosphorylate 14-3-3 to establish a substrate specificity for CKI isoforms. In addition, the alternatively spliced forms of CKI␣ have been shown to have different substrate specificity (55). A recent report showed that CKI associates with Nck (65), an adaptor molecule recruited to receptor tyrosine kinases that probably initiates signal transduction cascade. This implies possibly a role for CKI in the signal transduction of receptor tyrosine kinases in which 14-3-3 may be implicated as a component essential for protein complex formation.
The phosphorylation site in 14-3-3 is TSDTQ where the underlined residue is phosphorylated. Among other 14-3-3 proteins, only 14-3-3 has a putative phosphorylation site on the residue 233 (Table II), and we showed that this residue (Ser-233) is phosphorylated in vitro by CKI. The TSDTQ site could belong to the atypical group of proteins which are phosphorylated by CKI (50). These authors showed that CKI isoforms phosphorylated two other motifs; the atypical group includes glycogen synthase and SV40 large T antigen, which are known to be phosphorylated in vivo. The site in 14-3-3 (TSDTQ) is particularly similar to that in SV40 large T antigen (TADSQ) ( Table II).
Residue 233 in 14-3-3 is located in a region that is not FIG. 3. The T233 kinase is identical to CKI. A, the active fractions from the gel filtration column were pooled and loaded on SDS-polyacrylamide gel. The band corresponding to the 38-kDa protein ( Fig. 2A) was excised, subjected to trypsin digestion, and electrospray mass spectrometry. The 38-kDa protein was shown to be identical to CKI␣ by analyzing with the "Peptidesearch" program from the group of Matthias Mann at EMBL (Heidelberg) using the non-redundant data base of protein sequences (43). The boxes correspond to the peptides recovered after trypsin digestion of the 38-kDa protein, and match unequivocally with the sequence of CKI␣. Some peptides were Edman sequenced and also matched (Ͼ) with CKI␣. B, T233 kinase, purified mammalian CKI and S. pombe CKI (Upstate Biotechnology Inc) were tested for their ability to phosphorylate wild-type (WT) and T233A 14-3-3 , as well as the positive control substrates casein (C) and phosvitin (P). The first lane (/) represents the negative control without substrate. highly ordered in the crystal structure (25,26), but which has been shown to be required for efficient binding of 14-3-3 proteins to target proteins (17,27,44,45). Therefore, the phosphorylation of 14-3-3 at the CKI site may inhibit its association not only with Raf, but also with other proteins. Homo-oligomerization of Raf has been shown to regulate its activation, and 14-3-3 proteins have been proposed to be involved in this process (13). The phosphorylation of 14-3-3 on Thr-233 would therefore potentially affect Raf activity. The control of 14-3-3 phosphorylation at the CKI site will be investigated to provide further insights for its function in cells. In conclusion, phosphorylation of specific isoforms of 14-3-3 may play an important role in the regulation of protein complex formation in signal transduction.
FIG. 4. 14-3-3 is phosphorylated in vivo on Thr-233. 293 cells were metabolically labeled with [ 32 P]orthophosphate. Immunoprecipitated 14-3-3 and 10 g of cold 14-3-3 as a carrier were digested with trypsin. A, a typical HPLC trace is shown at the top. Below is the 32 P associated with the phosphorylated peptide from 14-3-3 (open bars) and from 14-3-3 and Ha-Ras(V12) (hatched bars) transfected cells. The phosphopeptides were identified by microbore HPLC electrospray mass spectrometry. The phosphopeptide in both cases corresponds to the C terminus of 14-3-3, and is indicated by an arrow in the top panel. B, solid phase sequencing on arylamine membrane revealed that only Thr-233 was phosphorylated.

TABLE II
Alignment of all mammalian 14-3-3 proteins around residue 233 All mammalian 14-3-3 proteins are aligned around residue 233 phosphorylated in vitro by CKI and in vivo in 14-3-3 . Only 14-3-3 has a potential phosphorylatable serine at the same position as 14-3-3 . We also showed that CKI phosphorylates in vitro this residue. As a comparison, we aligned the sequence of SV40 large antigen (SV40 T Ag) which has been shown to be phosphorylated in vitro and in vivo by CKI (47). S. pombe CKI was tested for their ability to phosphorylate 2 g of 14-3-3 (wild-type (WT) and T233A mutant (T233)), , ␥, or ⑀ as well as the positive control substrates casein (C) and phosvitin (P). The first lane (/) represents the negative control without substrate. B, recombinant wildtype (WT) or S233A mutant (S233A) 14-3-3 (2 g) were phosphorylated by purified mammalian CKI.