Calyculin A-induced Vimentin Phosphorylation Sequesters 14-3-3 and Displaces Other 14-3-3 Partners in Vivo

14-3-3 proteins bind their targets through a specific serine/threonine-phosphorylated motif present on the target protein. This binding is a crucial step in the phos-phorylation-dependent regulation of various key proteins involved in signal transduction and cell cycle control. We report that treatment of COS-7 cells with the phosphatase inhibitor calyculin A induces association of 14-3-3 with a 55-kDa protein, identified as the intermediate filament protein vimentin. Association of vimentin with 14-3-3 depends on vimentin phosphorylation and requires the phosphopeptide-binding domain of 14-3-3. The region necessary for binding to 14-3-3 is confined to the vimentin amino-terminal head domain (amino acids 1–96). Monomeric forms of 14-3-3 do not bind vimentin in vivo or in vitro , indicating that a stable complex requires the binding of a 14-3-3 dimer to two sites on a single vimentin polypeptide. The calyculin A-induced association of vimentin with 14-3-3 in vivo results in the displacement of most other 14-3-3 partners, including the protooncogene Raf, which neverthe-less remain capable of binding 14-3-3 in vitro. Concomitant with 14-3-3 displacement, calyculin

The 14-3-3 protein family consists of seven highly conserved proteins of 28 -31 kDa that naturally assemble as homodimers or heterodimers and bind to a wide array of intracellular proteins (1). Among the proteins known to associate with 14-3-3 in vivo are proteins that function in cell cycle control (e.g. Cdc25 (2) and Wee1 (3)), signal transduction (e.g. Raf (4), phosphatidylinositol 3-kinase (5), and BAD (6)), transcriptional regulation (e.g. FKHRL1 (7)), and cell structure (e.g. keratins (8)). Genetic alterations in 14-3-3 proteins have been shown to impair a variety of biologic functions, such as cell cycle timing (9) and cell cycle arrest in response to DNA damage in yeast (10), photoreceptor development in the Drosophila eye (11), and the Ras/Raf signaling pathway in various organisms (4,11). The ability of 14-3-3 to bind target proteins has been shown in many instances to depend on the phosphorylation of specific sites on the target protein (1), and studies with synthetic phosphopeptides have defined two preferred 14-3-3 binding motifs as RSXpSXP or RXXXpSXP (where pS stands for phosphoserine) (12,13). The existence of additional variations in the sequence of 14-3-3 binding sites is probable inasmuch as 14-3-3 partners such as phosphatidylinositol 3-kinase (5), tyrosine hydroxylase (1), and keratins (14) do not contain either of these motifs. Association with 14-3-3 may positively or negatively regulate the function of the partner; thus, the binding of 14-3-3 results in activation of tyrosine hydroxylase (1) but inhibition of phosphatidylinositol 3-kinase (5) and Cdc25 (15); a continuous association with 14-3-3 is necessary for physiologic regulation of Raf kinase activity (12,16,17). The biochemical mechanisms by which 14-3-3 proteins modify the function of the target protein may vary for each partner; for example, in the case of yeast Cdc25 (18) and FKHRL1 (7) the association with 14-3-3 (Rad24 in yeast) results in their exclusion from the nucleus. The ability of the proapoptotic protein BAD to bind and sequester the antiapoptotic factors BCL 2 and BCLx L is inhibited by the phosphorylation of BAD and subsequent binding of 14-3-3 (6). The 14-3-3 proteins have multiple functions in the regulation of c-Raf-1; the binding of 14-3-3 to Raf stabilizes the inactive state in unstimulated cells but is also required to maintain the active conformation of Raf generated after growth factor stimulation (17).
The intermediate filament (IF) 1 network provides a rigid framework that supports cell architecture during interphase (19,20). IF proteins undergo disassembly and major reorganization during cell cycle progression (21). Inasmuch as IF proteins constitute at least 1% of total cell protein, IF disassembly creates a large pool of soluble IF proteins. The assembly/disassembly of IFs is regulated by dephosphorylation/phosphorylation of the filaments on their exposed N-and/or C-terminal domains (22). A variety of protein (Ser/Thr) kinases likely to phosphorylate IFs in vivo have been proposed, including protein kinase Cs, protein kinase A, CaMKII, and Cdc2 kinase. In BHK-21 cells, the IF proteins vimentin and desmin undergo a major increase in phosphorylation (to a stoichiometry of approximately 2 mol of phosphate/mol of protein) concomitant with filament disassembly that occurs as the cells move from G 2 to M (23,24). Incubation of these cells with the protein phosphatase inhibitor calyculin A results in a rapid disassem-bly of IFs, in parallel with their hyperphosphorylation, both at sites already partially phosphorylated in interphase and at mitosis-specific sites (25). The epithelial IF protein keratin K18 associates with 14-3-3 proteins in a cell cycle-and phosphorylation-dependent manner, reaching a peak during late S-G 2 /M phase (8,14).

14-3-3 and Raf cDNA Constructs-GST-wt-14-3-3-
Cell Culture and Transfection-COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. For transient expression of proteins, cells were transfected either with the DEAE-dextran method using 5-10 g of DNA/10-cm dish or using LipofectAMINE (Life Technologies, Inc.) with the indicated amounts of DNA, following the manufacturer's instructions. For cell stimulation, 24 -48 h after transfection cells were serum-deprived for 24 h prior to the addition of the stimulator.
Metabolic Labeling-Serum-deprived cells were washed once with medium lacking methionine/cysteine for [ 35 S]Met labeling or lacking phosphate for 32 P labeling and incubated for 30 min in the same media to deplete intracellular sources. Cells were labeled by incubation in the presence of 0.5 mCi/ml 35 S or 32 P for 4 or 2 h, respectively.
Cell Extraction and Protein Purification-Cells were lysed in ice-cold extraction buffer containing 50 mM Tris-CL, pH 7.5, 100 mM NaCl, 1% Triton X-100, 1 mM dithiothreitol, 1 mM EDTA, 1 mM EGTA, 2 mM Na 3 VO 4 , 50 mM ␤-glycerophosphate, and a mixture of proteinase inhibitors (Amersham Pharmacia Biotech). Cell lysates were incubated at 4°C for 90 min with the appropriate antibody precoupled to protein A/G-agarose beads (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for immunoprecipitation or with GSH-Sepharose beads (Amersham Pharmacia Biotech) for GSH affinity purification. The beads were washed twice with the extraction buffer, twice with extraction buffer containing 0.5 M LiCl, and twice with buffer A (40 mM Tris-Cl, pH 7.5, 0.1 mM EDTA, 5 mM MgCl 2 , and 2 mM dithiothreitol). The proteins were either eluted directly in SDS-sample buffer or served for additional experiments.
Dissociation of 14-3-3-Vimentin Complexes in Vitro-Cell extracts were incubated with 300 M synthetic phosphopeptide corresponding to c-Raf-1 amino acids 613-627 containing a canonical 14-3-3 binding motif, LPKINRSApSEPSLHR, or with a 300 M concentration of the same peptide in its unphosphorylated form for 30 min at 4°C prior to protein purification.
In Vitro Binding Assays-Prokaryotic recombinant GST or GST-14-3-3 fusion variants precoupled to GSH-Sepharose beads were incubated with the various cell extracts or with the lysyl C digest of 32 P-labeled vimentin at 4°C for 30 -60 min in the presence or absence of phosphopeptide as indicated. The beads were washed twice with extraction buffer, twice with extraction buffer containing 0.5 M LiCl, and twice with buffer A. The associated proteins were eluted either with GSH or with SDS-sample buffer as indicated. For the phosphatase/kinasetreated vimentin binding assays, GST-14-3-3 fusions were labeled using the biotinylation kit from Amersham Pharmacia Biotech.
Raf Coupled Kinase Assay-Following anti-Myc immunoprecipitation, the Myc-Raf-containing beads were incubated in buffer A (100 l final volume) supplemented with 100 M ATP, 10 Ci of [␥-32 P]ATP, and 0.3 g of prokaryotic recombinant GST-MEK-1 for 20 min at 30°C and for an additional 30 min after the addition of 2 g of prokaryotic recombinant kinase-inactive ERK-2 (K52R). Samples were separated on SDS-PAGE and transferred to PVDF membranes. The 32 P content in ERK-2 was quantified by PhosphorImager analysis (Molecular Dynamics, Sunnyvale, CA) (data are presented as PhosphorImager units ϫ 10 3 ).

Association of Vimentin with 14-3-3 Depends on Vimentin Phosphorylation and Requires 14-3-3 Dimerization and Its
Phosphopeptide-binding Domain-The ability of the phosphatase inhibitor calyculin A to induce the association of vimentin with 14-3-3 suggested that vimentin phosphorylation is involved; however, since vimentin does not contain a "perfect" 14-3-3 binding motif (19), we tested the role of vimentin phosphorylation in 14-3-3 binding. Incubation of vimentin extracted from calyculin A-treated cells with alkaline phosphatase or protein phosphatase 2A completely prevents vimentin binding to 14-3-3 in vitro ( Fig. 2A, compare lane 2 with lanes 3 and 5). The specificity of the inhibition was demonstrated by the ability of the alkaline phosphatase inhibitor p-nitrophenylphosphate (PNPP) to prevent the inhibition ( Fig. 2A, lane 4). In support of this finding, phosphorylation of vimentin extracted from control, untreated cells with protein kinase A/AKT combination (two kinases previously shown to phosphorylate several proteins on 14-3-3 recognition sites (1, 6, 7)) induced 14-3-3 binding to vimentin ( Fig. 2A, compare lane 9 with lane 10).
These results demonstrate that vimentin phosphorylation mediates and is required for 14-3-3 binding. To further support our findings and to demonstrate that 14-3-3 binding to vimentin is mediated by the 14-3-3 phosphopeptide-binding region, we used a synthetic phosphopeptide corresponding to c-Raf-1 amino acids 613-627, a canonical 14-3-3 binding motif previously shown to dissociate Raf 14-3-3 complexes (12,17), to dissociate 14-3-3/vimentin complexes in vitro (Fig. 2B). It is notable that only the phosphopeptide is able to dissociate 14-3-3-vimentin complexes, whereas the same peptide in the unphosphorylated state has no effect (Fig. 2B). Yaffe et al. (13) demonstrated that the presence of two or more phosphoserinecontaining 14-3-3 binding motifs on a single synthetic phosphopeptide increases the apparent affinity for 14-3-3 by 30-fold over the same peptide containing a single phosphoserine. In view of the absence of "optimal" 14-3-3 binding motifs on vimentin and the inability of the amino-terminally truncated, dimerization-deficient GST-ct-14-3-3-(139 -245) polypeptide to bind vimentin (Fig. 1A), we hypothesized that the coligation of two phosphoserines on a single vimentin polypeptide by a 14-3-3 dimer might be necessary to confer sufficient avidity for a stable association. Consistent with this idea, several forms of dimerization-deficient 14-3-3, although able to bind c-Raf-1 efficiently in vivo ( Fig. 1A and Refs. 17 and 26) were unable to bind vimentin in vivo (Fig. 2C) or in vitro (Fig. 2, A and D). The requirement for a 14-3-3 dimer for vimentin binding is thus consistent with the view that phosphovimentin in calyculin A-treated cells contains two or more low affinity 14-3-3 binding sites, neither of which is sufficient by itself to bind 14-3-3, whereas their coligation by a 14-3-3 dimer enables stable association. To identify regions in vimentin required for 14-3-3 binding, a lysyl C digest of 32 P-vimentin, isolated from calyculin A-treated COS cells, was adsorbed to immobilized prokaryotic recombinant GST-14-3-3 or dimerization-deficient (dm) 14-3-3 mutant (Fig. 2D). A single 32 P-labeled peptide fragment corresponding to approximately 25% of the applied 32 P was retained by the wild type 14-3-3, whereas no 32 P-peptide was retained by the monomeric 14-3-3 (Fig. 2D, compare lanes 1  and 2); retention of the vimentin 32 P-peptide was inhibited by the synthetic Raf phosphopeptide (Fig. 2D, lane 3). The lysyl C

14-3-3 Association with Vimentin Results in Its Dissociation from Raf and from Most of Its Other Target Proteins-We
noticed that the progressive increase in GST-14-3-3/vimentin association in vivo that occurs with prolonged exposure of cells to calyculin A (300 nM, 60 min) is accompanied by the displacement of most other 14-3-3 partners (Fig. 3, A and B), when low levels of GST-14-3-3 are expressed, but not with high level GST-14-3-3 expression (Fig. 3A). Inasmuch as the identity of specific endogenous 14-3-3-associated polypeptide bands is not known, Myc-Raf was coexpressed with GST-14-3-3 to serve as a reporter (Fig. 3B). Calyculin A pretreatment abrogates completely the association of Myc-Raf with 14-3-3 (Fig. 3B, com-pare lane 1 with lane 4 and lane 2 with lane 5); a similar inhibition is observed for the association of endogenous c-Raf-1 with GST 14-3-3 in calyculin A-treated cells (Fig. 3C, compare  lane 2 with lane 4). The loss of the 14-3-3/Raf association and 14-3-3 association with other targets caused by calyculin A in vivo is not due to a modification of c-Raf-1 or the other target proteins that impair their ability to bind 14-3-3, since Raf and the other targets from extracts of untreated or calyculin Atreated cells bind equally well to prokaryotic recombinant GST 14-3-3 in vitro (Fig. 3, D and E). Thus, a more feasible explanation for the calyculin A-induced inhibition of 14-3-3 association with Raf and other target proteins in vivo is that the phosphovimentin that accumulates in the calyculin A-treated cells simply outcompetes and displaces Raf and the other 14-3-3 partners.

FIG. 2. Association of vimentin with 14-3-3 depends on vimentin phosphorylation and requires 14-3-3 dimerization and its phosphopeptide-binding domain. A, vimentin immunoprecipitated from COS cells treated for 45 min with carrier (lanes
Raf kinase activity (16,17). This, together with the ability of calyculin A treatment to disrupt 14-3-3/Raf complexes in vivo, implied that the regulation of Raf kinase activity in calyculin A-treated cells would be impaired. Thus, we decided to use Raf kinase activity as a functional read-out assay for 14-3-3 availability in the cell. Incubation of serum-deprived COS cells with calyculin A caused a slight increase in Raf kinase activity but inhibited by 95% the EGF-induced activation of Raf (Fig. 4A). To confirm that this low activity was due to 14-3-3 limitation in vivo, we examined the ability of purified prokaryotic recombinant 14-3-3 to restore the activity of Raf extracted from calyculin A-treated cells (Fig. 4B). Consistent with earlier findings, the addition of recombinant 14-3-3 to Raf from unstimulated cells (Fig. 4B, compare lane 2 with lane 8) or EGF-treated cells (Fig. 4B, compare lanes 3 and 4 with lane 9) had no effect on Raf kinase activity. In contrast, the addition of 14-3-3 to Raf ex-tracted from calyculin A-treated cells greatly increased Raf kinase activity (Fig. 4B, lanes 11 and 12). The ability of 14-3-3 to restore Raf activity in extracts of calyculin A-treated cells was completely blocked by the Raf synthetic phosphopeptide (Fig. 4B, lane 13), demonstrating that 14-3-3 ability to reactivate Raf requires the 14-3-3 phosphopeptide binding domain. Moreover, a dimerization-deficient 14-3-3 mutant, which is unable to reactivate Raf deactivated in vitro after phosphopeptide-induced displacement of native 14-3-3 (17), is also unable to activate Raf extracted from calyculin A-treated cells (Fig. 4B, lane 14). To further ascertain whether the inability of EGF to activate Raf in calyculin A-treated cells is due to a limitation in the availability of endogenous 14-3-3, increasing amounts of GST-14-3-3 were cotransfected together with Myc-Raf, and the effect of calyculin A pretreatment on EGF-stimu-  Fig. 1A and treated with carrier (lanes 2, 6, 10, 14, and 18) or with 200 nM calyculin A for the indicated times (lanes 3-5, 7-9, 11-13, and 15-17). Cells were extracted in the absence (lanes 1-17) or presence of 1 mM synthetic phosphopeptide corresponding to the 14-3-3 binding sequence of Raf (PP, lane 18), and the GSH-Sepharose-associated proteins were analyzed as in Fig. 1A on a 7-14% continuous gradient SDS-PAGE. An autoradiogram of the 35 S-labeled proteins is shown. The array of bands migrating between 35 and 50 kDa represents degradation products of GST-14-3-3 that bound to the GSH-Sepharose beads. These products are also present in GST-14-3-3 produced in bacteria (see Fig. 3C, lower panel) and are not present in mock-transfected COS cells (Fig. 3C, upper panel). Moreover, these bands were not displaced by the synthetic phosphopeptide (Fig. 3A, lane 18). B, serum-deprived COS cells co-expressing GST-14-3-3 and Myc-Raf-1 were 32 P-labeled and treated with vehicle (lanes 1-3) or calyculin A (300 nM; lanes 4 and 5) for 45 min followed by carrier (lanes 1 and 4) or EGF treatment (100 ng/ml; lanes 2, 3, and 5) for 15 min. GST-14-3-3 was recovered on GHS-Sepharose beads, and the samples were left untreated (lanes 1, 2, 4, and 5) or treated with protein phosphatase 1␥ for 15 min at room temperature (1 milliunit/ml; lane 3). 14-3-3-associated proteins were resolved on a 5-15% gradient SDS-PAGE and transferred to PVDF membrane. An autoradiogram of the 32 2 and 4). Cell extracts were isolated on GSH-Sepharose beads (lanes 3 and 4) or tested for in vitro binding to GSH-Sepharose-immobilized prokaryotic GST-14-3-3 (lanes 1 and 2). lated Raf activation was assayed (Fig. 4C). Transfection of as little as 10 ng/plate of pEBG-GST-14-3-3 DNA was able to substantially overcome the calyculin A-induced inhibition of Raf activation, and at 160 ng/plate, transfection resulted in complete restoration of Raf activity to levels seen with EGF stimulation (Fig. 4C). The requirement for wild type 14-3-3 is shown by the inability of the pEBG vector (i.e. GST alone; Fig.  4C, lanes 2-4) as well as the pEBG-encoded dimerizationdeficient mutant 14-3-3 (Fig. 4C, lanes 17-19), each added at 640 ng/plate, to increase Raf activity in calyculin A-treated cells. It is notable that the expression of recombinant 14-3-3 does not alter Raf activity in cells not exposed to calyculin A, whether unstimulated (Fig. 4C, compare lane 2 with lanes 5, 8, 11, and 14) or treated with EGF (Fig. 4C, compare lane 3 with lanes 6, 9, 12, and 13), confirming that 14-3-3 overexpression does not activate Raf per se. Rather, 14-3-3 overexpression specifically overcomes the inhibition caused by calyculin A pretreatment. These results demonstrate that the calyculin A-induced vimentin binding to 14-3-3 results in limitation in 14-3-3 availability to Raf and in altered regulation of its kinase activity. DISCUSSION 14-3-3 proteins have been found to specifically associate with and participate in the regulation of many proteins including several key proteins involved in cellular signaling and cell cycle control. The present work identifies vimentin as an additional, regulated, 14-3-3 interactor. Vimentin undergoes extensive phosphorylation during cell cycle progression, and the inhibition of protein phosphatase 1 and 2A by calyculin A causes a rapid increase in vimentin phosphorylation comparable in magnitude with that occurring during mitosis, although with a somewhat different site specificity (25). The ability of calyculin A treatment to promote sufficient phosphorylation of endogenous vimentin to enable it to bind a very large fraction of cellular 14-3-3 raises two intriguing possibilities. First, it suggests that 14-3-3 may contribute to the regulation of vimentin phosphorylation/dephosphorylation and/or to the ability of vimentin to participate in filament formation. Second, the high stoichiometry of cellular vimentin phosphorylation achieved during specific phases of the cell cycle may bind sufficient cellular 14-3-3 to limit its availability, thereby modulating the function of other 14-3-3-dependent proteins.
Previous work has shown that the binding of 14-3-3 to phosphoproteins, such as Raf, can diminish their susceptibility to protein (Ser/Thr) phosphatase action (28). Thus, the association of 14-3-3 with phosphovimentin may alter vimentin dephosphorylation. As regards the impact of 14-3-3 on intermediate filament assembly, Liao et al. (8) demonstrated that
binding of 14-3-3 to keratin K18 could interfere with its assembly into filaments in vitro. 14-3-3 binding to K18 depends on keratin Ser 33 phosphorylation; however, this phosphorylation is not sufficient by itself for conferring association (14). Since the sequence around keratin Ser 33 does not represent a strong 14-3-3 motif, it seems that another 14-3-3 binding site(s) on K18 may collaborate with the Ser 33 site to support 14-3-3 binding. This possibility is strengthened by our finding that vimentin, which, like K18, does not contain an optimal high affinity 14-3-3 biding site, can form a stable association only with dimeric 14-3-3 and not with various 14-3-3 monomers. The importance of 14-3-3 dimerization in 14-3-3 function has not gained much attention; the functional role usually proposed for the 14-3-3 dimer is to allow the bridging of two different proteins (29 -31). For proteins that lack high affinity 14-3-3 binding sites, such as vimentin, keratin K18, and Wee1 (3), the coligation of two lower affinity sites by the 14-3-3 dimer may be necessary to enable stable target protein recognition and binding. The role of 14-3-3 dimerization extends, however, beyond the ability to bridge two different proteins and enable binding to single proteins with only low affinity sites. We recently showed that a 14-3-3 dimer is necessary to support Raf activity (17) despite the fact that Raf contains two high affinity binding sites for 14-3-3, each of which is independently sufficient to stably bind 14-3-3 (17,32). The present report extends those findings by demonstrating that not only is the dimerizationdeficient 14-3-3 unable to support the reactivation of Raf inactivated by 14-3-3 displacement in vitro, it is also unable to support an active configuration of Raf achieved in vivo in calyculin A-treated cells. Along similar lines, we recently observed that the ability of 14-3-3 to modulate the DNA binding of the Caenorhabditis elegans transcription factor DAF-16 requires a dimeric 14-3-3 and two independent 14-3-3 binding sites on the DAF-16 polypeptide. 2 The finding that vimentin phosphorylation and its concomitant binding to 14-3-3 proteins results in disassembly of Raf-14-3-3 complexes and other 14-3-3-protein complexes suggests that the association of phosphovimentin with 14-3-3 may have significance in the regulation of 14-3-3 availability besides a role in vimentin dephosphorylation/disaggregation. This intriguing possibility implies that phosphorylation of vimentin and perhaps other IF polypeptides (e.g. keratin K18) may affect various intracellular signaling and cell cycle control pathways by modifying 14-3-3 availability, as observed with Raf in this paper. Further evaluation of the cellular distribution of the different 14-3-3 isoforms under varying physiologic conditions and during different stages of the cell cycle and the effect of IF phosphorylation on this distribution is needed to ascertain whether variation in the availability of 14-3-3 by IFs serves as a regulatory factor in cellular signaling and cell cycle control.