A Novel Ligand-binding Site in the ζ-Form 14-3-3 Protein Recognizing the Platelet Glycoprotein Ibα and Distinct from the c-Raf-binding Site

We reported previously that the ζ-form 14-3-3 protein (14-3-3ζ) binds to a platelet adhesion receptor, glycoprotein (GP) Ib-IX, and this binding is dependent on the SGHSL sequence at the C terminus of GPIbα. In this study, we have identified a binding site in the helix I region of 14-3-3ζ (residues 202–231) required for binding to GPIb-IX complex and to the cytoplasmic domain of GPIbα. We also show that phosphorylation-dependent binding of c-Raf to 14-3-3ζ requires helix G (residues 163–187) but not helix I. Thus, the GPIbα-binding site is distinct from the binding sites for RSXpSXP motif-dependent ligands. Furthermore, we show that wild type 14-3-3ζ has a higher affinity for GPIb-IX complex than recombinant GPIbα cytoplasmic domain. Deletion of helices A and B (residues 1–32) disrupts 14-3-3ζ dimerization and decreases its affinity for GPIb-IX. Disruption of 14-3-3ζ dimerization, however, does not reduce 14-3-3ζ binding to recombinant GPIbα cytoplasmic domain. This suggests a dual site recognition mechanism in which a 14-3-3ζ dimer interacts with both GPIbα and GPIbβ (known to contain a phosphorylation-dependent binding site), resulting in high affinity binding.

In this study, we have identified a binding site in 14-3-3 in the helix I region of 14-3-3 encompassing residues 202-231 that interact with the GPIb␣ cytoplasmic domain and the intact GPIb-IX complex. We also show in vitro that c-Raf binds to 14-3-3 in a PKA-dependent manner and that this binding requires helix G but not helix I. Thus, the GPIb␣-binding site in 14-3-3 is distinct from the binding site for RSXpSXP-containing ligand c-Raf. Furthermore, we show that 14-3-3 dimerization is required for high affinity binding to GPIb-IX complex, suggesting that a dual site recognition mechanism involving GPIb␣ and ␤ subunits and dimerized 14-3-3.

Recombinant 14-3-3, 14-3-3 Mutants, and Recombinant
GPIb␣ Cytoplasmic Domain-Cloning of the cDNA encoding wild type 14-3-3 was described previously (14). The 14-3-3 cDNA was subcloned into a pmalC2 vector (New England Biolabs, Beverly, MA). The construct (pmal1433) encodes a fusion protein with the N-terminal region corresponding to the Escherichia coli maltose-binding protein (MBP) and C-terminal region corresponding to 14-3-3. Mutagenesis of pmal1433 was performed using PCR techniques (38). In all mutants except T3, stop codons were introduced into reverse primers at designated sites of the 14-3-3. The stop codon in mutant T3 was introduced inadvertently by PCR error. The mutants were subcloned into a pmalC2 vector at the EcoRI and XbaI sites. Correct sequences were verified by automated sequencing. The wild type 14-3-3 and 14-3-3 truncation mutants were purified by affinity chromatography using a cross-linked amylose-Sepharose column (New England Biolabs, Beverly, MA). Equivalent amounts (71 mol of protein/ml Sepharose) of the purified wild type or mutant 14-3-3 or MBP were conjugated onto cyanogen bromide-activated Sepharose 4B (Amersham Pharmacia Biotech), respectively. Coupling efficiencies in all cases were better than 99% as assessed by optical density at 280 nm wave length.
The cDNA encoding GPIb␣ in a pBlueScript vector was a generous gift from Dr. Jerry Ware at the Scripps Research Institute, La Jolla, CA. The cDNA fragment encoding the cytoplasmic domain of GPIb␣ (residues 518 -610) was generated by PCR with EcoRI and XbaI sites incorporated in the forward and reverse primers, respectively. The PCR product was subcloned into the pmalC2 vector. The correct sequence was verified by automated sequencing. The protein was expressed and purified as described previously (38).
Binding of Platelet GPIb-IX and Mal-Ib␣C to 14-3-3-Specific binding of the platelet GPIb-IX complex to 14-3-3-conjugated beads has been described previously (14). Briefly, washed platelets were resuspended in Hepes buffer (137 mM NaCl, 2.7 mM KCl, 1 mM MgCl 2, 5.6 mM D-glucose, 3.3 mM Na 2 HPO 4 , 3.8 mM Hepes, pH 7.35) and solubilized by adding an equal volume of the solubilization buffer (2% Triton X-100, 0.1 M Tris, 0.01 M EGTA, 0.15 M NaCl, and 1 mM dithiothreitol, pH 7.4) containing 0.2 mM E64 (Boehringer Mannheim) and 1 mM phenylmethylsulfonyl fluoride (13). In some experiments, platelets were solubilized in the presence of 1 mM CaCl 2 but in the absence of EGTA and E64 to allow calpain cleavage of GPIb␣ and thus generation of the C-terminal domain of GPIb-IX complex. After centrifugation at 100,000 ϫ g for 30 min, the lysates (200 l) were incubated with 25 l (50% (v/v)) MBPconjugated control beads or 14-3-3-conjugated beads at 4°C for 1 h. The beads were then washed three times in a 1:1 mix of Hepes buffer and solubilization buffer. Bound proteins were extracted with SDS-PAGE sample buffer and analyzed by SDS-PAGE followed by Western blotting with a monoclonal antibody against GPIb␣, WM23 (kindly provided by Dr. Michael C. Berndt, Baker Institute, Melbourne, Australia (39)). In some experiments, GPIb-IX was also detected using a polyclonal anti-peptide antibody against the C-terminal domain of GPIb␣, anti-Ib␣C (14). Reactions with antibodies were visualized using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech), and Kodak X-Omat AR film. In some experiments, reactions were visualized using SuperSignal chemiluminescence substrate (Pierce) and quantitated with a Bio-Rad PhosphorImager and chemiluminescence-sensitive phosphor-screens that have a theoretical linear range of 10 0 -10 4 . Binding of recombinant GPIb␣ cytoplasmic domain fusion protein (Mal-Ib␣C) to 14-3-3 was performed essentially as described above except purified Mal-Ib␣C fusion protein was used. The Mal-Ib␣C binding was performed in the platelet lysate buffer described above or in 0.1 M sodium citrate, pH 5.6. Binding of Mal-Ib␣C protein to beads was detected by Western blotting with a rabbit anti-peptide antibody against the cytoplasmic domain of GPIb␣, anti-Ib␣C (14).
Binding of Recombinant c-Raf to 14-3-3-Human recombinant c-Raf cDNA (a gift from Dr. Michael Karin, University of California at San Diego, La Jolla) was subcloned into a pmalC2 vector by three-fragment ligation (EcoRI, HindIII, and XbaI) and expressed as a fusion protein with maltose-binding protein. The purified recombinant c-Raf was phosphorylated by incubation with protein kinase A catalytic subunit (10 units/100 l) (Sigma) and 1 mM ATP in 15 mM Hepes, 5 mM magnesium acetate, 0.1 mM EGTA, 130 mM KCl, 1 mg/ml bovine serum albumin, 1 mM dithiothreitol, pH 7.4, at 22°C for 30 min. The PKAtreated c-Raf was incubated with 14-3-3-conjugated beads at 4°C for 1 h. After three washes, bead-bound proteins were then analyzed by SDS-PAGE and immunoblotted with an antibody against c-Raf (Santa Cruz Biotechnology).
Analysis of the Molecular Mass of 14-3-3 under Non-denaturing Conditions-Purified MBP-14-3-3 fusion protein and 14-3-3 mutants (200 l, ϳ1.5 mg/ml) were analyzed by gel filtration using a Pharmacia Superdex 200 HR 10/30 column and a Pharmacia Explorer 10 high performance liquid chromatography system at a flow rate of 0.5 ml/min. The column was equilibrated with 0.02 M Tris, 0.15 M NaCl, pH 7.4. The molecular mass was determined by comparing with the elution volumes of the molecular mass standard proteins, IgG (150 kDa), bovine serum albumin (67 kDa), and ovalbumin (43 kDa).

RESULTS
Truncation Mutagenesis of 14-3-3-In order to identify the sequence within 14-3-3 that is responsible for its interaction with the GPIb-IX complex, we made various 14-3-3 truncation mutants (Fig. 1). Mutants T1 to T6 were truncated from the C-terminal end of 14-3-3. Mutants T7-T9 were truncated progressively from the N-terminal end of the protein. T11-(136 -209) encompasses helices G and H, and T12 contains helices H and I. T13-(188 -209) contains a single helix H that was implicated in tryptophan hydroxylase binding (34). T14-(202-231) containing helix I was generated when results obtained with T1-T13 indicated the location of the GPIb␣-binding site (see below). As 14-3-3 is composed of nine anti-parallel ␣-helices (A-I), truncation sites in all these mutants are located between two neighboring ␣-helices as determined by the published crystal structure to avoid disruption of each of these helical structures (Fig. 1). These mutant proteins were expressed as fusion proteins with maltose-binding protein. Equivalent amounts of the purified wild type and mutant 14-3-3 were conjugated to cyanogen bromide-activated Sepharose 4B as described previously (38).
ative control, and the wild type 14-3-3-beads was used as a positive control. Bead-bound GPIb-IX was detected by Western blotting with an anti-GPIb␣ antibody, WM23, and quantitated with a Bio-Rad PhosphorImager and chemiluminescence-sensitive phosphor-screens. Fig. 2 shows that GPIb-IX binds to wild type 14-3-3-but not MBP-conjugated beads. Removal of the C-terminal tail of 14-3-3 by truncation at amino acid residue 231 (mutant T6) did not negatively affect GPIb-IX binding, suggesting that the C-terminal region between residues 231 and 246 is not required. Further truncation at residue 209 (mutant T5), which removes the helix I (the first ␣-helix from the C-terminal end), completely abolished GPIb-IX binding (Fig. 2). Furthermore, none of the truncation mutants that lack helix I bound to GPIb-IX (T5, T11, T13, Fig. 2; T1-T4, data not shown). Thus, the helix I region encompassing residues 209 -231 of 14-3-3 appears to be required for binding to GPIb-IX. Truncation mutants T9-(188 -246) and T12-(188 -231) containing helix H and I interacted with GPIb-IX, indicating that the sequence between residues 188 and 231 of 14-3-3 (helices H and I) contains a binding site for the GPIb-IX complex. Similar results were also obtained when bead-bound GPIb-IX was detected with an antibody against the C-terminal region of GPIb␣, anti-Ib␣C (data not shown).
High Affinity Interaction Requires Dimerized 14-3-3 and More Than One Subunit of GPIb-IX- Fig. 2 also shows that truncation mutants of 14-3-3 lacking the N-terminal domain (T7, T8, T9, and T12) binds significantly less GPIb-IX in comparison with wild type 14-3-3. Since similar amounts of proteins were conjugated to these beads (Fig. 2C), this result indicates that GPIb-IX bound to these mutants with reduced affinity. In particular, the mutant T7 lacking only 33 residues in helices A and B (1-33) showed dramatically reduced binding to GPIb-IX (7%) in comparison with wild type 14-3-3 (Fig. 2), suggesting that the N-terminal helices (A and B) are important to the high affinity interaction between wild type 14-3-3 and GPIb-IX. However, this reduction in GPIb-IX binding affinity is partially "rescued" by further truncations from the N terminus that removed helices A-E (T8, 20% binding compared with wild type 14-3-3) or helices A-G (T9, 50% binding compared with wild type) (Fig. 2). It is not clear why these further truncations rescued the loss of binding affinity. One of the possibilities is that further truncations resulted in increased accessibility of GPIb-IX to the binding site in helix I. Nevertheless, this result suggests that the N-terminal domain, while important for high affinity, is not required for binding to GPIb-IX.
Previously reported crystal structural analysis of 14-3-3 has revealed that the N-terminal helices (A and B) are involved in the formation of 14-3-3 dimers (18,19). Thus, it is possible that the reduced affinity for GPIb-IX is caused by disruption of dimerization. To examine this possibility, we determined the molecular mass of the MBP-14-3-3 fusion proteins by gel filtration chromatography under non-denaturing conditions. The 14-3-3 monomer has a molecular mass of ϳ30 kDa (40). MBP has a molecular mass of ϳ40 kDa. The molecular mass for the recombinant MBP-14-3-3 fusion protein is ϳ70 kDa as analyzed by SDS-PAGE. We found that the MBP-wild type 14-3-3 fusion protein has a molecular mass of ϳ160 kDa as determined by gel filtration chromatography (Fig. 3A). Thus, the recombinant wild type 14-3-3 indeed exists as a dimer even though its N terminus is fused to MBP. In contrast, the fusion proteins encoding 14-3-3 mutants (T7, T9, and T12) lacking the N-terminal domain showed a molecular mass of approximately 70, 55, and 45 kDa by gel filtration (Fig. 3A), indicating that they are monomers. Thus, the N-terminal domain deletion disrupted dimerization of 14-3-3. Taken together, these results suggest that a dimeric structure of 14-3-3 is required for high affinity binding to GPIb-IX.
Previous studies suggest that 14-3-3 interaction with GPIb-IX may involve two binding sites in GPIb-IX located in the C-terminal domain of GPIb␣ and GPIb␤, respectively (14 -16). To examine whether the high affinity binding of dimerized 14-3-3 to GPIb-IX may involve interaction with the cytoplasmic domains of both GPIb␣ and GPIb␤, we compared binding of 14-3-3 to the recombinant GPIb␣ cytoplasmic (C-terminal) domain alone and to the C-terminal domain of the GPIb-IX complex (composed of the C-terminal domain of GPIb␣, GPIb␤ and GPIX (6, 14)) (Fig. 4). As we reported previously (14), the C-terminal domain of GPIb-IX maintained the high affinity 14-3-3 binding function (Fig. 4). In contrast, the recombinant cytoplasmic domain of GPIb␣ bound much more weakly to 14-3-3, indicating that in addition to GPIb␣, other subunits of the GPIb-IX complex (probably GPIb␤) are involved in high affinity binding to 14-3-3 dimer.
Helix I of 14-3-3 (Residues 202-231) Contains a Binding Site for the GPIb␣ Cytoplasmic Domain-We showed previously that the binding of the GPIb-IX complex to 14-3-3 is dependent on the interaction between 14-3-3 and the C-terminal se- GPIb-IX binding to mutants T1-(1-109), T2-(1-162), T3-(1-31), and T4-(1-187) were also examined, but no specific binding was detected. To estimate the amounts of various 14-3-3 mutant MBP fusion proteins conjugated to Sepharose beads, the beads were incubated with a rabbit antibody against MBP at 4°C for 1 h. After washing, the bead-bound antibody was analyzed by SDS-PAGE and Western blotting with antirabbit IgG and enhanced chemiluminescence (C).
quence of the GPIb␣ cytoplasmic domain (14). It is thus possible that the above identified GPIb-IX-binding site of 14-3-3 contains a binding site for GPIb␣. To examine this possibility, we tested the binding of recombinant GPIb␣ C-terminal domain MBP fusion protein (Mal-Ib␣C) to wild type 14-3-3 and various 14-3-3 mutants. Fig. 5 shows that wild type Mal-Ib␣C protein specifically binds to the 14-3-3-conjugated beads. Removal of the C-terminal tail of 14-3-3 by truncation at residue 231 (mutant T6) did not negatively affect Mal-Ib␣C binding, whereas truncation at residue 209 (mutant T5) to remove helix I abolished Mal-Ib␣C binding. In addition, binding of other truncation mutants that lack the helix I to Mal-Ib␣C was also disrupted. Thus, the same helix I region (residues 209 -231) of 14-3-3 that is required for GPIb-IX binding is also required for 14-3-3 binding to the GPIb␣ cytoplasmic domain. In contrast to the results obtained with the GPIb-IX complex, however, the binding of Mal-Ib␣C to the monomeric truncation mutants T7, T8, T9, and T12 (lacking the N-terminal dimerization site) was not significantly reduced but rather a little stronger than the dimeric wild type 14-3-3. This suggests that the dimeric structure of 14-3-3 is not required for binding to the recombinant GPIb␣ cytoplasmic domain. It is not clear why Mal-Ib␣C bound to the small truncation mutants stronger than to wild type 14-3-3. It is unlikely that this is caused by the variation in the amounts of bead-conjugated proteins as the wild type and mutant 14-3-3-MBP fusion protein bound identical amounts of anti-MBP antibody (see Fig. 2C). It is possible that truncations caused increased exposure of the ligand-binding site in the helix I of 14-3-3 to the recombinant GPIb␣ cytoplasmic domain.
To verify further that helix I is indeed a binding site for the GPIb␣ cytoplasmic domain, we made an additional 14-3-3

FIG. 3. Disruption of 14-3-3 dimerization by N-terminal deletion.
Molecular mass of the wild type 14-3-3-MBP fusion protein and the deletion mutants T7, T9, and T12 were analyzed by gel filtration chromatography (A) as described under "Materials and Methods" and by SDS-PAGE (B). The elution volume for wild type 14-3-3 (WT1433) on a Superdex 200HR 10/30 column is similar to that of rabbit IgG, indicating that it has a molecular mass of ϳ160 kDa, in comparison with the molecular mass determined by SDS-PAGE (ϳ70 kDa). Thus, wild type 14-3-3 exists as a dimer under non-denaturing conditions. Elution volume of mutant T7 lacking the N-terminal helices A and B (residues 1-32) indicated that its molecular mass is ϳ70 kDa under non-denaturing conditions (A), similar to its molecular mass determined by SDS-PAGE (67 kDa) (B). Similarly, molecular mass of T9 and that of T12 as determined by gel filtration are comparable to that determined by SDS-PAGE, indicating that the mutants lacking the N-terminal domain exist as monomers. Shown at the bottom of A is the chromatograph of the molecular mass standard proteins IgG (150 kDa), bovine serum albumin (67 kDa), and ovalbumin (45 kDa).

FIG. 4. Comparison between 14-3-3 binding to the C-terminal domain of GPIb-IX complex and to the recombinant C-terminal domain of GPIb␣.
Platelets were solubilized as described previously (14) to allow proteolysis at the protease-sensitive region of GPIb␣, thus generating the C-terminal fragment of GPIb␣ complexed with GPIb␤ and GPIX (see C). The lysates were then incubated with 14-3-3-conjugated beads or control MBP-conjugated beads for 1 h. Comparable amounts of purified recombinant GPIb␣ cytoplasmic domain MBP fusion protein were also incubated with 14-3-3 beads (14-3-3) or control beads (MBP) under identical conditions. The beads were washed three times. Bound proteins were eluted with SDS-PAGE sample buffer, separated by SDS-PAGE, and detected by Western blotting with an antibody against the C terminus of GPIb␣ (anti-Ib␣C). A, a typical picture of Western blots showing that comparable amounts of GPIb-IX C-terminal domain or recombinant GPIb␣ cytoplasmic domain were incubated with 14-3-3 beads (added), but 14-3-3-bound recombinant GPIb␣ cytoplasmic domain (Mal-Ib␣C) was reduced in comparison with the C-terminal domain of GPIb-IX complex (Ib-IX). B, the Western blot was scanned, and optical density of each band from the same blot was quantitated with NIH Image software. Note that the amount of bound GPIb-IX C-terminal domain (Ib-IX) is approximately 3 times that of the recombinant GPIb␣ cytoplasmic domain (Mal-Ib␣C) when comparable amounts were incubated with 14-3-3-conjugated beads. C, a schematic of the GPIb-IX complex. Locations of 14-3-3-binding sites in both GPIb␣ and GPIb␤ as well as the protease-sensitive region are indicated by arrows. truncation mutant, T14-(202-231), containing only helix I. This mutant strongly bound to Mal-Ib␣C (Fig. 5B), indicating that the 30-residue helix I-(202-231) region of 14-3-3 is sufficient for binding to the GPIb␣ cytoplasmic domain.

PKA-dependent c-Raf Binding to 14-3-3 Requires Helix G (Residues 162-187) but Not Helix I (Residues 209 -231)-To
determine whether the GPIb␣-binding site of 14-3-3 is also required for the binding of other 14-3-3 ligands, we examined PKA-dependent binding of c-Raf (a 14-3-3 ligand with the RSX-pSXP binding motif (32)) to the 14-3-3 truncation mutants. Fig. 6A shows that, in the absence of PKA pretreatment, c-Raf-MBP fusion protein bound weakly to 14-3-3-beads, comparable with control MBP-conjugated beads. After PKA-pretreatment, however, c-Raf bound to 14-3-3 strongly. This binding was not affected by truncation of 14-3-3 that removed helix I (T5) or both helices H and I (T4) from the C terminus, indicating that the GPIb␣-binding site in helix I is not required for the interaction of 14-3-3 with c-Raf (Fig. 6B). However, further truncation from the C terminus to residue 162 to remove helix G (T2) abolished specific c-Raf binding. The truncation mutant T8 lacking the N-terminal helices A-E still bound to c-Raf, but the mutant T9, which contains C-terminal helices H and I but lacks helices A-G, lost c-Raf binding capacity (Fig. 6B). Thus, binding of the RSXpSXP motif-containing ligand c-Raf to 14-3-3 requires helix G but not helix I.

DISCUSSION
In this study, we have further characterized the interaction between GPIb-IX and 14-3-3 by analyzing the GPIb-IX binding function of a series of truncation mutants of 14-3-3. We show that a binding site for the GPIb␣ cytoplasmic domain is located in the helix I of 14-3-3 encompassing residues 202-231 in the C-terminal domain (Fig. 5). This binding site is also required for 14-3-3 binding to intact GPIb-IX complex (Fig. 2). The GPIb␣-binding site is distinct from the binding sites for RSXpSXP motif-dependent 14-3-3 ligands, as we show that PKA-dependent binding of c-Raf to 14-3-3 requires helix G but not helix I (Fig. 6). Previous studies also showed that the RSXpSXP-containing ligand tryptophan hydroxylase bound to a fragment of 14-3-3 containing helices G and H but not I (34). Furthermore, crystal structure data (35) and mutagenesis studies (41) suggest that phosphorylation-dependent binding of RSXpSXP motif containing ligands may involve the interaction of phosphoserine (pS) with Lys-49 and Arg-56 in helix C and Arg-127 and Tyr-128 in helix E, which are not required for binding to GPIb␣ cytoplasmic domain (Fig. 5). In fact, the presence of the helices C-E is inhibitory to the interaction between GPIb-IX and monomeric 14-3-3 mutants (Fig. 2). Thus, our data suggest that different types of 14-3-3 ligands preferentially interact with 14-3-3 at different sites in the large ligand binding groove surrounded by helices C, E, G, and I (18,19,35). It is likely that GPIb␣ represents a different class of the 14-3-3 ligands which preferentially recognize the binding site in the helix I of 14-3-3 proteins, whereas RSXpSXP-like ligands preferentially interact with helix G. Andrews et al. (15) aligned several 14-3-3-binding proteins that have sequences similar to the GPIb␣ binding sequence (including Cdc25a, Cdc25b, c-Raf, and c-Cbl). A striking feature in these 14-3-3 ligands is the presence of an HSL tripeptide sequence. It would be interesting to investigate further whether this HSL motif is responsible for the binding of this class of 14-3-3 ligands to helix I of 14-3-3. Interestingly, while our manuscript was in revision, Petosa et al. (42) reported crystal structural data indicating that an unphosphorylated non-RSXpSXP peptide ligand of 14-3-3 (WLDLE) obtained by screening a phage display library binds by amphipathic interaction to sites within the ligand binding groove overlapping with but distinct from c-Raf PSXpSXP peptide-binding sites, supporting the notion that different ligands may interact with different sites in the groove. Our data provide first evidence that the interaction with the helix I region is both required and sufficient for 14-3-3 binding to a physiolog- In all cases, c-Raf was pretreated with PKA as described in A.
We show that the GPIb-IX complex has a higher affinity for 14-3-3 than the recombinant GPIb␣ cytoplasmic domain (Fig.  4), suggesting that other subunits of GPIb-IX complex in addition to GPIb␣ may be involved in the interaction with 14-3-3. This result is consistent with the previous findings that high affinity binding to 14-3-3 involves both GPIb␣ (14,15) and GPIb␤ which contains a phosphorylation-dependent 14-3-3binding site (15,16). Wild type 14-3-3 exists as a homodimer via the interaction between the N-terminal domains (helices A and B) of each monomer (18,19). Thus, we investigated the possibility that high affinity interaction involves simultaneous binding of GPIb␣ and GPIb␤ to a 14-3-3 dimer. We show that the 14-3-3 mutants lacking the dimerization site in helices A and B bound to the GPIb-IX complex with significantly reduced affinity compared with the dimeric wild type 14-3-3 (Fig. 2). Gel filtration chromatography confirmed that deletion of helices A and B disrupted dimerization of 14-3-3 (Fig. 3), suggesting that the high affinity binding of 14-3-3 to GPIb-IX requires 14-3-3 dimerization. Furthermore, since the binding of the recombinant GPIb␣ cytoplasmic domain and the phosphorylated c-Raf to these monomeric 14-3-3 mutants was not decreased compared with dimeric wild type 14-3-3 (Figs. 5 and 6), the reduced binding of GPIb-IX is unlikely to result from the disruption of ligand-binding sites in these monomeric mutants. Thus, 14-3-3 binding to GPIb-IX appears to involve the interaction of 14-3-3 dimer to GPIb␣ and an additional binding site in other GPIb-IX subunits. As GPIb␤ contains a PKA-catalyzed phosphorylation-dependent binding site for 14-3-3, it is likely that high affinity binding of intact GPIb-IX involves interaction of GPIb␣ and GPIb␤ to each monomer of the 14-3-3 dimer. As GPIb␤ phosphorylation has been found to be dynamically regulated in platelets by increase in intracellular c-AMP level (36,37,43), this dual site binding mechanism may enable the 14-3-3-GPIb-IX interaction to be dynamically regulated in platelets and thus may participate in the GPIb-IX-coupled intracellular signaling and cytoskeleton regulation.
We show that helix G of 14-3-3 is critical for binding to the RSXpSXP motif containing ligand c-Raf (Fig. 6). The published crystal structural data, however, indicate that the critical phosphoserine in RSXpSXP motif is in the proximity of the residues Lys-49, Arg-56, Arg-127, and Tyr-128 in helices C and E, suggesting that helix G may not be the recognition site for the phosphoserine in the RSXpSXP motif. While our manuscript is under revision, Petosa et al. (42) and Wang et al. (44) reported crystal structure of 14-3-3 bound to a RSXpSXP peptide derived from c-Raf, suggesting a hydrophobic surface lining the ligand binding groove is proximal to the ligand peptide. This hydrophobic surface includes residues Leu-172, Val-176, Leu-216, Leu-220, and Leu-227. Among these residues, Leu-172 and Val-176 are located in helix G. Replacement of Leu-172 or Val-176 with negatively charged residues abolished the binding of c-Raf suggesting that these hydrophobic residues are critical for the c-Raf binding. Thus, it is possible that RSXpSXP ligands may bind to helix G by hydrophobic interaction. As helix G is required for interaction with phosphorylated c-Raf, we suggest that interaction of the phosphoserine in the ligand peptide with residues in helices C and E is not sufficient to support the binding of c-Raf. Furthermore, as we show that mutant T8 lacking helices C and E binds to phosphorylated c-Raf, these helices are not required for interaction with c-Raf. Thus, it is possible that the interaction between phosphoserine in the RSXpSXP motif and helices C and E serves to regulate the ligand interaction with the helix G (c-Raf) and helix I (GPIb␣) regions in the wild type 14-3-3.
Interestingly, Wang et al. (44) showed that point mutations that replace the hydrophobic residues Leu-216, Leu-220, and Leu-227 in the helix I with negatively charged residue aspartic acid also significantly reduced 14-3-3 binding to c-Raf. However, we show that PKA-phosphorylated c-Raf bound to 14-3-3 mutant lacking helix I (Fig. 6). Our data suggest that, while it is possible that helix I is involved in the interaction between c-Raf and 14-3-3, this region is not required for this interaction. A major difference between Wang et al. (44) and our data is that the recombinant c-Raf used in our assays is in vitro phosphorylated by PKA which is known to phosphorylate the RSXpSXP motif (32), whereas Wang et al. (44) used the yeast two-hybrid system, in vitro translated protein, or cell lysates with unknown phosphorylation status. Furthermore, binding of c-Raf to 14-3-3 in their assays did not appear to be influenced by stimulation of c-Raf phosphorylation (44), which is similar to GPIb-IX binding to 14-3-3 (13,14). Thus, it is possible that in the assay by Wang et al. (44), c-Raf binding to 14-3-3 may involve a GPIb␣-like motif in addition to the RSX-pSXP motifs. An HSL sequence similar to the 14-3-3 binding sequence of GPIb␣ was indeed found in c-Raf (15). Thus, we speculate that c-Raf binding to dimeric 14-3-3 in vivo may be similar to GPIb-IX binding, involving a dual binding site mechanism.