Impaired Binding of 14-3-3 to C-RAF in Noonan Syndrome Suggests New Approaches in Diseases with Increased Ras Signaling

Protein expression and purification from E. coli.cDNA clones of human 14-3-3ζ (NCBI accession no. BC083508) and 14-3-3σ (NCBI accession no. BC000329) were purchased from OpenBiosystems (Huntsville, AL) and subcloned into the expression vector pProEx HTb (Invitrogen, Karlsruhe, Germany). Protein expression was induced in Escherichia coli BL21(DE3) by addition of 0.4 mM IPTG (isopropyl-β-d-thiogalactopyranoside) at 18°C for 12 h. Purification of His₆-tagged 14-3-3ζ and 14-3-3σ protein was carried out using standard procedures. After the protein was concentrated to 100 mg/ml, it was dialyzed against buffer consisting of 20 mM HEPES-NaOH (pH 7.5), 2 mM MgCl₂, and 2 mM dithiothreitol (DTT) and stored at −80°C. The C-RAFpSer²⁵⁹ peptides (H₂N-²⁵⁵QRSTpSTPNVH²⁶⁴-COOH and H₂N-²⁵⁵QRSTpST²⁶⁰-COOH) were synthesized by Biosyntan (Berlin, Germany) and resuspended in distilled water to a final concentration of 10 mM. For the Biacore measurement, GST-C-RAF_220-268S259D (C-RAF cDNA; NCBI accession no. BC018119) was cloned into the pGEX4T-1 vector (GE Healthcare, Freiburg, Germany). The cDNA template was ordered from OpenBiosystems (Huntsville, AL). Protein expression of GST-C-RAF_220-268S259D and glutathionine S-transferase (GST) was induced in E. coli BL21(DE3) by the addition of 0.4 mM IPTG at 18°C for 12 h. Purification of GST-tagged C-RAF_220-268 and GST was carried out using standard procedures. After the protein was concentrated to 35 mg/ml, it was dialyzed against the mixture consisting of 20 mM HEPES-NaOH (pH 7.5), 2 mM MgCl₂, and 2 mM DTT and stored at −80°C.

Crystallization.Prior to crystallization, 14-3-3ζ was diluted to 30 mg/ml and mixed with the C-RAFpSer²⁵⁹ peptide at a 1:1.5 molar ratio in the mixture consisting of 20 mM HEPES-NaOH (pH 7.5), 2 mM MgCl₂, and 2 mM DTT. Crystals appeared after 3 to 4 weeks at 20°C in 0.1 M PCB buffer (sodium propionate, sodium cacodylate, BIS-Tris [N,N-methylenebisacrylamide-Tris] propane [molar ratios, 2:1:2])-27% polyethylene glycol (PEG) 1500-2 mM DTT and grew within 3 to 4 days to dimensions of 200 by 100 by 100 μm. For flash-cooling, the crystals were incubated in the mother liquor with 35% PEG 1500 and transferred to liquid nitrogen. For crystallization of the 14-3-3σ/C-RAFpSer²⁵⁹ peptide complexes (10 mg/ml), protein and peptides were mixed at a 1:1.5 molar ratio in the reaction mixture consisting of 20 mM HEPES-NaOH (pH 7.5), 2 mM MgCl₂, and 2 mM DTT and set up for crystallization in a reaction mixture consisting of 0.1 M HEPES-NaOH (pH 7.5), 0.2 M CaCl₂, 28% PEG 400, 5% glycerol, and 2 mM DTT at 4°C. Crystals grew within a week to dimensions of 400 by 200 by 200 μm and were directly flash-cooled in liquid nitrogen. The ternary complex with fusicoccin was obtained via a soaking procedure. A 1-μl volume of a 1.5 mM fusicoccin (Sigma-Aldrich, Munich, Germany) solution in ethanol was allowed to dry on a coverslide. After the addition of 2 μl of reservoir solution to the dried fusicoccin, crystals of the binary 14-3-3σ/QRSTpST-COOH complex were transferred to the drop and incubated for 24 h. Fragment soaking was done by incubating binary complex crystals in reservoir solution containing a 50 mM concentration of the single fragments followed by direct transfer into liquid nitrogen.

Data collection, structure determination, and refinement.Data collection was performed at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland, and data were processed with XDS software (21). Molecular replacement was carried out using MOLREP software with the high-resolution structure of 14-3-3ζ (Protein Data Bank [PDB] code 1QJB) or 14-3-3σ (PDB code 1YWT) used as the search model. The obtained model was subjected to iterative rounds of model building and refinement using the programs COOT and REFMAC (10, 26). Figures were prepared with PYMOL (www.pymol.org ).

ITC.All isothermal titration calorimetry (ITC) experiments were carried out with a VP-ITC instrument (MicroCal, Northampton, MA) using buffer containing 20 mM MES (morpholineethanesulfonic acid) (pH 6.5) and 2 mM MgCl₂. A solution of 50 μM 14-3-3 protein was placed in the sample cell and subjected to stepwise titration with 8-μl aliquots of a 500 μM solution of the different C-RAF peptides by the use of a 300-μl syringe for a total of 35 injections. The equilibration time between pairs of injections was 120 s. The heating power per injection was observed throughout the duration of the reaction until equilibrium was reached. The association constant K_a (K_a = 1/K_d), molar binding stoichiometry (N), and molar binding enthalpy (ΔH°) were determined by fitting the binding isotherm by the use of a single-binding-site model and Origin7 software (MicroCal). The listed K_d values represent the averaged results of at least three independent measurements.

Cloning of green fluorescent protein (GFP)-tagged C-RAF.cDNA of C-RAF (NCBI accession no. BC018119) was obtained from OpenBiosystems (Huntsville, AL) and cloned into pEGFP-C1 (Clontech) (GenBank accession no. U55763) by the use of primers with restriction sites for BglII (sense primer) and EcoRI (antisense primer). Correct ligations and orientations were verified by sequencing.

Cloning of BiFC constructs of C-RAF and 14-3-3ζ.For bimolecular fluorescence complementation (BiFC) experiments, pcDNA3 vector (Invitrogen, Karlsruhe, Germany) was used. As templates for the PCRs, cDNA of C-RAF (NCBI accession no. BC018119), a cDNA clone of 14-3-3ζ (NCBI accession no. BC083508), and enhanced yellow fluorescent protein (eYFP) (a generous gift from T. K. Kerppola) were used. First, the PCR product of the N-terminal part of eYFP (amino acids [aa] 1 to 154) was cloned into pcDNA3 vector by the use of a KpnI-BamHI restriction site. The C-terminal part of eYFP (aa 155 to 238) was cloned into pcDNA3 vector by the use of a EcoRI-XhoI restriction site. After confirming the correct cloning of the YFP fragment, 14-3-3ζ was subcloned as C-terminal (EcoRI-XhoI) and N-terminal (KpnI-BamHI) fusions to the split eYFP to get the 14-3-3-BiFC constructs. The C-RAF-BiFC construct (aa 1 to 330) was subcloned as C- and N-terminal fusions. All constructs were confirmed by sequencing.

Site-directed mutagenesis of C-RAF constructs.Site-directed mutagenesis was performed using a Stratagene QuikChange kit (Agilent Technologies Sales, Waldbronn, Germany). All mutated constructs were confirmed by sequencing.

Cell culture and transfection.HEK293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (Gibco BRL, Grand Island, NY) and 0.3% antibiotics at 37°C in a 7.5% CO₂ atmosphere. Experiments were performed with 5 to 20 cell passages. HEK293 cells were plated on glass coverslips in 24-well plates and transiently transfected according to the manufacturer's instructions with 0.25 μg of plasmid DNA and 5 μl of Qiagen's Effectene transfection reagent (Qiagen, Hilden, Germany) in 700 μl of growing medium.

Live-cell imaging.HEK293 cells were examined 24 h after transfection using a Leica TCS SP2 confocal microscope equipped with an argon ion laser that was used to excite GFP fluorescence at 488 nm. HEK293 cells were cotransfected with pEGFP-C1 C-RAF (WT or mutant) and pCMV-HA HRas (WT or G12V) for GFP-C-RAF localization studies. Complementation studies were performed using an argon ion laser excitation at 514 nm to excite YFP fluorescence. For these studies, HEK293 cells were transiently cotransfected with pcDNA3 vector containing the split YFP fragments and interacting proteins (14-3-3ζ, HRas, and C-RAF).

Immunostaining of cells.HEK293 cells were seeded on poly-l-lysine-coated 10-mm coverslips and transfected with GFP-C-RAF_1-330 and Effectene transfection reagent (Qiagen, Hilden, Germany) according to the manufacturer's recommendations. For immunostaining, cells were treated using standard procedures. To detect endogenous C-RAF, sc-133 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), a C-RAF antibody that specifically detects the C-terminal 12 amino acids of C-RAF and is coupled to Alexa Fluor 647, was used at a concentration of 1:50. Cells were directly examined in Hank's balanced salt solution (HBSS) in a Leica life-imaging chamber or glued onto slides with Mowiol and dried for 24 h before examination. GFP was excited at 488 nm and detected at 525 to 545 nm, whereas Alexa Fluor 647 was excited at 633 nm and detected at 650 to 700 nm with a Leica confocal microscope (model TCS SP2).

Immunoblotting.For analysis of ERK phosphorylation, HEK293T cells were transiently transfected using GFP-C-RAF and a mammalian transfection kit (Stratagene, La Jolla, CA) following the manufacturer's recommendations. At 8 h after transfection, cells were serum starved and, after a further 16 h, lysed in buffer containing phosphatase inhibitor cocktail 1 and 2 at a ratio of 1:100 (Sigma-Aldrich, Munich, Germany). Immunoblotting was performed using standard procedures. Transfection efficiency was monitored with a rabbit monoclonal antibody against GFP (ab32146; Abcam, Cambridge, MA). Basal ERK1/2 levels and ERK1/2 phosphorylation were detected with rabbit monoclonal antibodies against ERK1/2 and phospho-ERK1/2, respectively (catalog no. 4695 and 9101; Cell Signaling, Boston, MA). Signals were detected with a secondary antibody against rabbit IgG coupled to alkaline phosphatase (ab67221; Abcam, Cambridge, MA) and BCIP (5-bromo-4-chloro-3-indolylphosphate)/NBT (Nitro Blue Tetrazolium) (Merck, Darmstadt, Germany).

Isolation of GST-fused C-RAF from COS7 cells.For preparation of GST-fused C-RAFWT and mutants, 4.5 × 10⁶ COS7 cells were seeded in 14-cm-long plates and transfected with 25 μg of DNA from each construct in the presence of 50 μl of jetPEI transfection reagent (Biomol) after 24 h. At 24 h posttransfection, cells were washed twice with 1× phosphate-buffered saline (PBS) buffer and lysed with the buffer containing 25 mM Tris-HCl (pH 7.6), 150 mM NaCl, 10 mM sodium pyrophosphate, 10% glycerol, 25 mM β-glycerophosphate, 25 mM NaF, 0.1% β-mercaptoethanol, 1 mM Na₃VO₄, 0.75% Nonidet P-40, and protease inhibitor cocktail (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The clarified lysates were incubated with 200 μl of glutathione Sepharose 4B beads (GE Healthcare) for 2 h at 4°C. The beads were then washed three times with 1 ml of buffer containing 25 mM Tris-HCl (pH 7.6), 300 mM NaCl, 10 mM sodium pyrophosphate, 10% glycerol, 25 mM β-glycerophosphate, 25 mM NaF, 0.1% β-mercaptoethanol, 1 mM Na₃VO₄, 0.2% Nonidet P-40, and protease inhibitor cocktail. One-third of the beads were used for an in vitro kinase assay. Two-thirds of the beads were mixed with Laemmli buffer. The precipitated proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on a 10% gel, blotted, and detected by the use of anti-C-RAF antibody (catalog no. sc-133; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-pS338 antibody (catalog no. 9427; Cell Signaling), anti-pS259 antibody (catalog no. 9421; Cell Signaling), and anti-pS621 antibody (no. 6B4; in-house production). For purification of the GST-C-RAF proteins, the same procedure as that described above was employed, with the exception that the bound GST-C-RAF proteins were eluted by buffer containing 50 mM Tris-HCl (pH 8.0), 20 mM glutathione, and protease inhibitor cocktail.

In vitro kinase assay.For an in vitro kinase assay, the GST-C-RAF coupled beads were washed with 1 ml of buffer containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 25 mM β-glycerophosphate, and 1 mM Na₃VO₄. The kinase assay was performed using recombinant MEK and ERK2 as substrates in a mixture containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 25 mM β-glycerophosphate, 10 mM MgCl₂, 1 mM dithiothreitol, 1 mM Na₃VO₄, and 1 mM ATP buffer (50-μl final volume). The assay reaction mixtures were incubated for 30 min at 30°C. The kinase assay was stopped by addition of 20 μl of 5× Laemmli buffer, and the reaction mixture was boiled for 5 min at 100°C. The proteins were separated by SDS-PAGE on 10% gel, blotted, and detected by anti-pERK and anti-ERK antibodies.

Biosensor measurements.To measure interactions between C-RAF and 14-3-3 proteins, the surface plasmon resonance (SPR) technique was applied. The measurements were carried out at 25°C essentially as previously described (13). Briefly, a CM5 biosensor chip was loaded with anti-GST antibody by the use of covalent derivation according to the manufacturer's instructions. Purified and GST-tagged C-RAF proteins were immobilized in biosensor buffer (10 mM HEPES [pH 7.6], 150 mM NaCl, and 0.05% Nonidet P-40) at a flow rate of 10 μl/min, which resulted in protein deposition equivalent to approximately 300 response units (RU). Next, the purified 14-3-3 proteins were injected at the indicated concentration. The values for nonspecific bindings measured in the reference cell were subtracted.

For the experiments with GST-C-RAF_220-268S259D (8 μg/ml [negative-control GST]), a T100 Biacore system (Biacore AB, Uppsala, Sweden) was used. The proteins were coupled to the matrix of a GST antibody-coated dextran chip following the manufacturer's protocol. Binding of 14-3-3ζ (10 μM) in the presence of different compounds and concentrations was measured at 10°C in a reaction mixture containing 10 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM MgCl₂, and NP-40 at 1:2,000. Binding and dissociations curves were fitted to the sum of single exponential and linear functions.

Accession numbers.The atomic coordinates and structure factors of 14-3-3ζ/C-RAFpSer²⁵⁹ and the 14-3-3σ/C-RAFpSer²⁵⁹ complex have been deposited in the Protein Data Bank under identification codes 3NKX and 3IQJ; the deposition codes for the complexes of 14-3-3σ and the C-terminally truncated C-RAFpSer²⁵⁹ phosphopeptide with and without fusicoccin are 3IQV and 3IQU, respectively. The 14-3-3σ/C-RAFpSer²⁵⁹ structure in complex with the fusicoccin-derived fragment has been deposited under identification code 3O8I.

In vitro effects of C-RAF mutations.To study the effect of the reported C-RAF mutations associated with the Noonan and LEOPARD syndromes (Fig. 1A) on the interaction with 14-3-3 in vitro, we used synthetic C-RAFpSer²⁵⁹ peptides bearing the corresponding amino acid substitutions. Isothermal titration calorimetry (ITC) was used to quantify the affinities of binding of these mutated phosphopeptides to 14-3-3ζ (Fig. 1B). The wild-type C-RAFpSer²⁵⁹ peptide (H₂N-²⁵⁵QRSTpSTPNVH²⁶⁴-COOH) bound to 14-3-3ζ with an apparent K_d value of 7.5 μM, whereas the mutated peptides—with the exception of the V263A substitution peptide—displayed significantly lower affinity levels or could not be measured at all. The only mutation found in the Ser²⁵⁹ motif of C-RAF whose in vivo effects (34, 38) cannot be unambiguously explained by the ITC data is V263A. We observed a binding affinity of the corresponding peptide that was at least as strong as that of the wild-type peptide (K_d of 6.5 μM). Since V263 is near the extreme C-terminal end of the peptides used, we further analyzed this position with a longer peptide (H₂N-²⁵⁵QRSTpSTPN²⁶³AHMVSTTLPVD²⁷³-COOH). Indeed, in this longer peptide the V263A mutation resulted in a measurable weaker affinity of binding to 14-3-3ζ (27.1 μM versus 14.1 μM for the WT peptide [Fig. 1C]).

The 14-3-3ζ/C-RAFpSer²⁵⁹ peptide complex.In order to provide a mechanistic rationale for the consequences of the C-RAF mutations and to corroborate our biophysical data, we solved the structure of a C-RAF-derived phosphopeptide (H₂N-²⁵⁵QRSTpSTPNVH²⁶⁴-COOH) in complex with the ζ isoform of human 14-3-3 to 2.4 Å. Although the details of a similar structure have been published before (36), due to its resolution of 3.6 Å, this structure is not as appropriate for the analysis of the side-chain interactions that define the effects of single amino acid substitutions.

We obtained crystals of the 14-3-3ζ/C-RAFpSer²⁵⁹ peptide complex by mixing the synthetic phosphopeptide with 14-3-3ζ prior to crystallization setups. The resulting crystals diffracted to 2.4 Å, and the structure was solved by molecular replacement using phases from the 14-3-3ζ apo structure (39). The relevant statistics are summarized in Table 1. The 14-3-3ζ dimer displays a characteristic W-like shape, with each monomer constituting an amphipathic groove that harbors the C-RAFpSer²⁵⁹ peptide (Fig. 1D). The peptide binds in the groove in an elongated conformation and occupies approximately 70% of the total length of this ligand-binding channel (Fig. 1E and F). Analysis of the interactions between the mutated amino acid positions (Fig. 1G [red boxes]) and 14-3-3ζ reveals essential functionalities of the respective side chains and explains the effects that these mutations show in vitro and in vivo.

In vivo effects of C-RAF mutations.To study the effect of the reported C-RAF mutations in vivo, we transfected HEK293 cells with a green fluorescent protein (GFP)-labeled N-terminal C-RAF construct (GFP-C-RAF_1-330). C-RAF is activated following recruitment by active Ras to the plasma membrane (24, 47). This process is counteracted by 14-3-3 proteins that bind to the N terminus of C-RAF, thereby retaining it in the cytoplasm (8, 23). The two inhibitory 14-3-3 binding sites in C-RAF are Ser²³³ and Ser²⁵⁹ (8, 41), whereas a third site, Ser⁶²¹, seems to be activated upon 14-3-3 binding (47, 52). Therefore, the activity status of C-RAF is reflected by the formation of a complex of its N terminus with Ras or 14-3-3. We chose a C-RAF construct comprising the first 330 amino acids, which include the Ras binding domain (RBD) and both N-terminal 14-3-3 interaction sites, to analyze the effect of the reported mutations in the C-RAFSer²⁵⁹ region on inhibitory 14-3-3 binding. Since we assumed that the endogenous levels of Ras proteins responsible for the plasma membrane recruitment of C-RAF were too low in comparison to the expressed GFP-C-RAF constructs for our purposes, we cotransfected HRas together with GFP-C-RAF_1-330WT or the corresponding mutants to obtain clearer results to reveal the effects of the C-RAF mutations on subcellular localization. When HEK293 cells were cotransfected with HRasWT and GFP-C-RAF_1-330WT, diffuse labeling, theoretically due to the formation of a complex of the C-RAF construct with endogenous 14-3-3 proteins, was observed throughout the cytoplasm (Fig. 2A). In contrast to GFP-C-RAFWT, the GFP-C-RAF protein bearing the mutations (e.g., P261L) displayed plasma membrane and Golgi localization (Fig. 2B). In our experiments, we used HRas, which is known to localize to the plasma membrane and the Golgi apparatus (4, 41). Therefore, the observed subcellular distribution of the mutated GFP-C-RAF_1-330 P261L construct suggests an enhanced HRas association as a consequence of the impaired binding to 14-3-3 proteins.

• Open in new tab
• Download powerpoint

FIG. 2.

Complete set of C-RAF mutants found in patients with Noonan syndrome and LEOPARD syndrome that influence the subcellular localization of C-RAF_1-330 in HEK293 cells. (A) Cells cotransfected with HRas and GFP-C-RAF_1-330WT. (B) Cells cotransfected with HRas and the 14-3-3 binding-impaired construct GFP-C-RAF_1-330P261L. (C) Cells cotransfected with HRas and the complete set of C-RAF 14-3-3 binding mutants introduced into GFP-C-RAF_1-330. (D) Cells cotransfected with HRas and GFP-C-RAF_1-330 constructs additionally bearing the R89L Ras binding mutation. (E) Bimolecular fluorescence complementation. Cells cotransfected with C-RAF_1-330 N-terminally fused to YFP_1-154 (YN-C-RAF_1-330) and 14-3-3ζ or HRas N-terminally fused to YFP_155-238 (YC-14-3-3ζ or YC-HRas, respectively) are shown. Bars, 5 μm. C, cytoplasm; G, Golgi apparatus; N, nucleus; PM, plasma membrane.

The eight remaining mutations demonstrated the same effects as P261L; GFP labeling was found at the plasma membrane and the Golgi apparatus (Fig. 2C). This included the V263A mutation, as expected from previously reported genetic findings (34, 38).

In order to support the hypothesis that the observed subcellular localization of the GFP-C-RAF_1-330 fusion proteins was due to Ras-mediated recruiting, we additionally introduced the R89L mutation, which is known to abrogate the Ras-C-RAF interaction (3), into the GFP-C-RAF_1-330 constructs. These double-mutated proteins exhibited predominantly nuclear localization, with weaker labeling of the cytoplasm and no visible localization to the plasma membrane or the Golgi apparatus (Fig. 2D). This result is in line with the previously reported tendency of a GFP-C-RAF-RBD (Ras binding domain; residues 51 to 131) fusion protein to be preferentially localized in the nucleus in addition to the cytoplasm (4). To verify that the observed different localization of the GFP-C-RAF_1-330 constructs was due to its interaction with 14-3-3 proteins or HRas, we conducted bimolecular fluorescence complementation (BiFC) studies (Fig. 2E). HEK293 cells were cotransfected with C-RAF_1-330 fused to the N-terminal part of YFP (YN-C-RAF_1-330) and either 14-3-3ζ or HRas fused to the C-terminal part of YFP (YC-HRas or YC-14-3-3ζ). The YN-C-RAFWT construct could interact with both YC-14-3-3ζ in the cytoplasm and YC-HRas at the plasma membrane and the Golgi apparatus. Mutation of Ser²⁵⁹ to phenylalanine (S259F) abrogated the interaction with 14-3-3ζ but did not interfere with binding of the same construct to HRas. On the other hand, introducing a mutation that diminished the interaction of C-RAF with HRas (R89L) had no effect on the ability of C-RAF to bind to 14-3-3ζ. Finally, a C-RAF construct bearing both mutations bound neither to 14-3-3ζ nor to HRas (data not shown). These in vivo interaction results support the interpretation that the observed different locations of the GFP-C-RAF constructs shown in Fig. 2 can indeed be explained by the interaction with 14-3-3 proteins or HRas. The other mutations showed the same effect as S259F, revealing abrogation of the interaction with 14-3-3ζ, whereas the binding with HRas remained unaffected (data not shown).

For the purpose of simultaneously monitoring the localization of endogenous C-RAF and the mutated N-terminal GFP-C-RAF fusions, we probed HEK293 cells cotransfected with the HRasWT and different GFP-C-RAF constructs with an antibody against the C terminus of C-RAF. The wild-type GFP-C-RAF_1-330 construct showed the same localization as endogenous C-RAF, which was mainly cytoplasmic. In contrast, the PM and Golgi localization of the GFP-C-RAF_1-330 mutants S259F, T260R, and V263A did not coincide with localization of endogenous C-RAF in the same cells (Fig. 3A). In order to address the issue of whether the Noonan syndrome C-RAF alleles were also causing higher levels of MAPK activation, we transfected HEK293 cells with GFP-C-RAF full-length constructs and measured ERK phosphorylation in serum-starved cells. Whereas transfection with GFP-C-RAFWT did not result in phosphorylation of ERK in these unstimulated cells, the S259F, T260R, and (to a lesser extent) V263A mutations rendered C-RAF capable of constitutively activating the MAPK pathway (Fig. 3B). Furthermore, we investigated the 14-3-3 binding properties of C-RAF mutants used in the experiments represented in Fig. 3A and B by the SPR technique. For this assay, GST-C-RAF full-length proteins were purified from COS7 cells and 14-3-3 proteins were isolated from E. coli. As shown in Fig. 3C, the biosensor analysis revealed considerably lower binding of 14-3-3 to C-RAF mutants compared to the WT results. This finding is consistent with the colocalization and MAPK activation results. In addition, our previous data demonstrated that the replacement of C-RAF serine 259 by alanine led to a pronounced decrease of the 14-3-3 association as well (17).

• Open in new tab
• Download powerpoint

FIG. 3.

Colocalization of C-RAFWT and C-RAF mutants and MAPK activation. (A) HEK293 cells were cotransfected with HRasWT and GFP-C-RAF_1-330 and probed with anti-C-RAF (CT antibody). Whereas the GFP-C-RAFWT colocalizes with endogenous (WT) C-RAF, the mutant variants (S259F, T260R, and V263A) show a different distribution pattern, localizing to the plasma membrane and the Golgi apparatus. Bars, 5 μm. (B) Immunodetection of GFP, ERK, and pERK from lysates of HEK293 cells transfected with full-length GFP-C-RAFWT or mutant. (C) Noonan and LEOPARD syndrome-associated C-RAF mutants S259F, T260R, and V263A bind with lower affinity to 14-3-3ζ compared to the C-RAF wild type. GST-tagged C-RAFWT wild type and indicated C-RAF mutants were transfected in COS7 cells and purified by the use of glutathione Sepharose. To measure binding of recombinant 14-3-3ζ to purified C-RAF proteins, the SPR technique was utilized. To this end, the biosensor chip was first loaded with anti-GST antibody and C-RAF proteins were injected, resulting in a deposition of approximately 300 response units. Next, purified 14-3-3ζ (1 μM) was applied and the association-dissociation curves were monitored. No binding between purified GST and 14-3-3ζ (GST control curve) was measured.

Kinase activity of C-RAF mutants correlates with S259 phosphorylation.Next, we examined the phosphorylation status of the regulatory sites S259, S621, and S338 in all of C-RAF mutants used for ITC analysis (Fig. 1B) in conjunction with their kinase activities. Upon precipitation of GST-C-RAF proteins from COS7 cell lysates, an in vitro kinase assay was carried out using recombinant MEK and ERK as substrates. As shown in Fig. 4 A, the catalytic activity of the mutants was significantly increased compared to C-RAFWT results. These values correlate with the loss of phosphorylation at S259. In contrast, the phosphorylation levels of S338 and S621 were mostly unchanged.

• Open in new tab
• Download powerpoint

FIG. 4.

Characterization of the full-length C-RAF proteins containing Noonan and LEOPARD syndrome-associated mutations. (A) Decreased phosphorylation of serine 259 within the internal 14-3-3 binding domain of the indicated C-RAF mutants correlates with enhanced kinase activity. COS7 cells were transfected with GST-tagged WT or indicated C-RAF mutants. The cells were lysed, and C-RAF proteins were precipitated by the use of glutathione-Sepharose. Subsequently, kinase activities were determined using recombinant MEK and ERK as substrates. The assessment of C-RAF activities and degrees of phosphorylation of serines 259, 338, and 621 was carried out by the use of the appropriate phosphospecific antibodies. (B) Impaired binding of 14-3-3 to C-RAF mutants S257L, P261A, P261L, P261S, and V263A leads to dephosphorylation of pS259 in vivo. COS7 cells were transfected with GST-tagged C-RAFWT or indicated C-RAF mutants and treated with either okadaic acid (1 μM) or solvent (ethanol) for 2 h. The cells were lysed, and C-RAF proteins were precipitated by the use of glutathione-Sepharose. The degree of serine 259 phosphorylation was detected by the use of a phosphospecific antibody. The amount of the precipitated C-RAF proteins was visualized by the use of anti-C-RAF antibody. IB, immunoblot.

To explain the reason for the loss of phosphorylation at S259, we treated the GST-C-RAF-transfected COS7 cells with okadaic acid, which is a potent phosphatase inhibitor. In general, two possibilities could be considered. On the one hand, the indicated mutations would inhibit per se the phosphorylation in these positions. On the other hand, it is possible that phosphorylation occurs but that, due to the impaired 14-3-3 binding, the unprotected pS259 is accessible for dephosphorylation by phosphatases. Our results presented in Fig. 4B support the second possibility for C-RAF mutants S257L, P261A, P261L, P261S, and V263A, whereas the mutants R256S, S259F, T260I, and T260R were shown not to be affected by okadaic acid.

Possible stabilization of the C-RAF/14-3-3 interaction by small molecules.We previously reported the structure of a plant 14-3-3 protein complexed to a phosphopeptide derived from plant H⁺-ATPase PMA2 with the protein complex stabilized by the fungal toxin fusicoccin (53). The 14-3-3 binding site in PMA2 is a so-called mode III motif where the polypeptide chain ends at position +1 with respect to the C terminal of the phosphorylated serine or threonine (QSYpTV-COOH). Therefore, the binding groove of a 14-3-3 monomer is occupied by this peptide for only about two-thirds of its length. The fusicoccin fungal toxin fills the remaining gap in the protein-protein interface, simultaneously contacting both protein partners. Since the C-RAFSer²⁵⁹ phosphopeptide displays a mode I motif (39) that continues in the C-terminal direction from the phosphorylated serine, the fusicoccin binding site in 14-3-3 is partly occupied by this peptide (Fig. 1E). This is in line with our finding in an ITC experiment that fusicoccin does not bind to the binary 14-3-3/C-RAFSer²⁵⁹ (QRSTpSTPNVH-COOH) complex (Fig. 5 B). In contrast, when a C-terminally shortened C-RAF phosphopeptide (QRSTpST-COOH) that resembles a mode III motif was used, its binary complex with 14-3-3 bound fusicoccin with a K_d of 6.6 μM (Fig. 5E). As a control, the two phosphopeptides bound to 14-3-3 with comparable levels of affinity (Fig. 5A and D). These results indicate that, for sterical reasons, fusicoccin cannot bind to and stabilize the 14-3-3/C-RAF complex. To corroborate these findings, we elucidated the structure of 14-3-3σ complexed with the truncated C-RAF phosphopeptide (QRSTpST-COOH) and fusicoccin A, as well as the binary complexes 14-3-3/QRSTpST-COOH (not shown) and 14-3-3/QRSTpSTPNVH-COOH, to a resolution of 1.20 Å, 1.05 Å, and 1.15 Å, respectively. We chose 14-3-3σ for these structural biology experiments, since in our hands that isoform yielded the best diffracting crystals by far, with the higher resolution being especially beneficial for analysis of small-molecule-protein interactions and bioinformatics approaches. Furthermore, the robustness of that isoform rendered it ideally suited for soaking experiments, which can put considerable physical strain on protein crystals.

• Open in new tab
• Download powerpoint

FIG. 5.

Binding of fusicoccin A (FC) to binary 14-3-3σ/C-RAF-phosphopeptide complexes. (A and B) ITC analysis of C-RAF peptide QRSTpS²⁵⁹TPNVH-COOH binding to 14-3-3σ (A) and subsequent binding of fusicoccin A to the binary complex (B). (C) The 1.10-Å crystal structure of C-RAF-pSer²⁵⁹ (QRSTpS²⁵⁹TPNVH-COOH [mode I-like]) phosphopeptide (cyan sticks) bound to 14-3-3σ (wheat-colored surface). The 2F_O-F_C electron density map contoured at 1σ is shown in blue. (D and E) ITC analysis of C-RAF peptide QRSTpS²⁵⁹T-COOH binding to 14-3-3σ (D) and subsequent binding of fusicoccin A to the binary complex (E). (F) The 1.20-Å crystal structure of C-terminally truncated C-RAF-pSer²⁵⁹ (QRSTpS²⁵⁹T-COOH [mode III-like]) phosphopeptide (magenta sticks) and fusicoccin A (yellow sticks) bound to 14-3-3σ (wheat-colored surface).

In the binary complex of 14-3-3 with the longer C-RAF QRSTpSTPNVH-COOH phosphopeptide, the binding pocket of fusicoccin is partially occupied by the C-RAF peptide (Fig. 5C). In contrast to that finding and in similarity to the results seen with PMA2, the truncated C-RAF phosphopeptide and fusicoccin together almost completely filled the amphipathic groove of the 14-3-3 monomer (Fig. 5F). Interestingly, there were no significant structural differences between the empty and the occupied fusicoccin binding sites (data not shown). When bound with the longer C-RAF phosphopeptide, a significant portion of the fusicoccin binding site was occupied by residues in a C-terminal orientation from the phosphorylation serine (Fig. 6A ). However, the space occupied by the main part of the 8-membered (B) central ring and the complete lower 5-membered ring (A) remained available in this binary structure also (Fig. 6B and C). Starting from this structural information, we generated a small library of fusicoccin A-ring fragments (Fig. 6D) and soaked crystals of the binary 14-3-3/QRSTpSTPNVH-COOH complex in crystallization buffer containing a 50 mM concentration of the individual fragments. One of these fragments bound to the site occupied by fusicoccin in the 14-3-3/PMA2 complex. Importantly, this fragment is coordinated very similarly to its fusicoccin template (Fig. 6E). Next, we tested the potential of these fragments to actually stabilize the C-RAF/14-3-3 interaction, employing surface plasmon resonance (SPR). To this end, an N-terminal S259D phosphomimicry mutant of C-RAF (residues 220 to 268) fused to GST (GST-C-RAF_220-268S259D) was immobilized on a GST antibody-coated dextran matrix of a Biacore chip, and binding of the 14-3-3 protein was measured in the presence of 25 mM concentrations of the different fragments (Fig. 6F). Most of the fragments were not able to stabilize binding of the 14-3-3 protein to the C-RAF constructs, but fragments 1, 2, and 3 (Fig. 6D) conferred an association for both proteins. Very interestingly, fragment 1, the most potent of those fragments, was the one we were able to identify in our crystal-soaking experiments. Analysis of the concentration-dependent stabilization induced by this fragment (Fig. 6G) yielded a dissociation constant in the low millimolar range. To verify the findings obtained with fragment 1 and the phosphomimicry construct (GST-C-RAF_220-268S259D), we performed an analogous experiment using full-length C-RAF purified from COS7 cells. A finding that was consistent with the data shown in Fig. 6F and G was that, in the presence of fusicoccin A fragment 1, the association of 14-3-3 protein with the C-RAFWT was considerably enhanced, whereas the dissociation rates did not reveal substantial differences. As demonstrated in Fig. 6H, stabilization of the C-RAF/14-3-3 interaction had already been induced by the presence of a 100 μM concentration of fragment 1.

• Open in new tab
• Download powerpoint

FIG. 6.

Fusicoccin A-derived molecule fragments binding to the binary 14-3-3/C-RAF complex. (A) Overlay of C-RAFSer²⁵⁹ phosphopeptides QRSTpS²⁵⁹TPNVH-COOH (green) and QRSTpS²⁵⁹T-COOH (magenta) with fusicoccin A (FC) (yellow) bound to 14-3-3σ (wheat-colored surface). (B) Structure of fusicoccin A with parts of the molecule (indicated in green) that would occupy space already claimed by the QRSTpS²⁵⁹TPNVH-COOH peptide in the binary 14-3-3/C-RAF complex. (C) Two-dimensional structure of fusicoccin as described for panel B. (D) Fragments derived from the lower part of the fusicoccin molecule (ring A). (E) Stereo view of the 2.0-Å crystal structure of a fusicoccin-derived fragment (blue sticks) binding to the binary 14-3-3σ/QRSTpSTPNVH-COOH complex. The 2F_O-F_C electron density map contoured at 1σ for the fragment is shown in black, and fusicoccin bound to 14-3-3σ/QRSTpST-COOH is superimposed. (F) Surface plasmon resonance (SPR) analysis of the fragments described for panel D with respect to their ability to stabilize the GST-C-RAF_220-268S259D/14-3-3σ complex. The association or dissociation of 14-3-3 with respect to the immobilized GST-C-RAF constructs was measured in the presence of 20 mM concentrations of the different fragments. Binding curves of the active fragments are labeled with numbers corresponding to those shown in panel D. (G) SPR analysis of the concentration-dependent activity of fragment 1. (H) Fusicoccin A-derived fragment 1 (as shown in panel D) stabilizes interaction between 14-3-3 proteins and full-length C-RAF. The effects of fusicoccin A-derived fragment 1 on C-RAF/14-3-3 interactions were measured by the SPR technique. To this end, purified and GST-tagged C-RAF was first immobilized on the GST antibody-coated surface of the biosensor chip, resulting in a deposition of approximately 300 response units. Next, purified 14-3-3ζ (1 μM) was applied in the absence and presence (0.1 and 1 mM) of fusicoccin A-derived fragment 1. Prior to injection, the 14-3-3 proteins were preincubated with fragment 1 for 10 min at room temperature. The association-dissociation curves indicate increased association rates in the presence of the fragment 1. No binding between purified GST and 14-3-3ζ (GST control curve) was measured. Injection of fragment 1 alone (without 14-3-3) did not lead to any change for immobilized GST-C-RAF (data not shown).

14-3-3 proteins are structurally highly conserved adapter proteins that undergo little conformational change upon binding to their partner proteins. For example, 14-3-3ζ complexed to the C-RAFpS259 phosphopeptide and tobacco 14-3-3 protein T14-3c complexed to PMA2 and fusicoccin (32) superposes with a root-mean-square deviation (rmsd) of 0.9 Å. Comparison of 14-3-3ζ in its apo form (PDB code no. 1A4O) with the form bound to arylalkylamine-N-acetyltransferase (AANAT; PDB code no. 1IB1) (29) reveals a high level of rigidity of the 14-3-3 protein, as indicated by a low rmsd of only 0.8 Å. These values are in line with the “molecular anvil” hypothesis of Yaffe (54), who observed that 14-3-3 proteins seem to force their protein partners into certain conformations without changing their own structure. In this context, the characteristic cup-like 14-3-3 dimers seem to be predestined to form complexes with their protein partners that display relatively deep pockets in their interfaces, a feature that is not necessarily common in other protein-protein complexes.

A cavity at the interface of the two proteins can also be found in the 14-3-3/C-RAFpSer²⁵⁹ structure formed by the polypeptide chain of the 14-3-3 partner exiting the binding groove via the Pro²⁶¹-induced kink (Fig. 6A and Fig. 7A). Since this complex structure displays just a small part of the 14-3-3 partner protein, we included AANAT, the only 14-3-3 structure with a (nearly) full-length protein partner, in our analysis (Fig. 7B). We investigated whether the region of the protein-protein interface corresponding to the fusicoccin site would display small-molecule binding pockets in the complexes of 14-3-3 with C-RAFpSer²⁵⁹ or AANAT. To evaluate this possibility, we analyzed the corresponding structures by the use of SCREEN software (28). This program predicts the “druggability” of potential binding sites by assigning a score between 0 and 1, with 1 being the optimal value. As expected, the binding pocket of fusicoccin in the T14-3c/PMA2-CT52 complex (Fig. 7C) showed a very high value (0.85), but the corresponding site in the 14-3-3ζ/AANAT (Fig. 7B) complex displayed an only slightly lower score (0.81). Interestingly, the potential binding pocket generated in the 14-3-3ζ/C-RAFpSer²⁵⁹ complex reached a value of 0.71 (Fig. 7A). For comparison, the binding pocket of the protein kinase active site inhibitor Gleevec bound to BCR-ABL kinase (PDB code no. 1IEP) achieved a value of 0.86. This suggests the possibility that, in contrast to most other protein complexes, the protein-protein interfaces of 14-3-3 proteins bound to their targets might be “druggable” in a manner similar to that seen with active sites of enzymes.

• Open in new tab
• Download powerpoint

FIG. 7.

Structural comparisons of binding interfaces of 14-3-3/target protein complexes. (A) Structure of human 14-3-3ζ complexed to the C-RAFSer²⁵⁹ phosphopeptide (data from this study; PDB code no. 3NKX). Left panel, stereo view of the interface region of 14-3-3ζ (brown surface) and C-RAFSer²⁵⁹ (green surface) that corresponds to the fusicoccin binding site shown in panel C. Right panel, complete binary complex. (B) Structure of human 14-3-3ζ bound to serotonin N-acetyltransferase (NAT; PDB code no. 1ib1). Left panel, stereo view of the interface region of 14-3-3ζ (light green surface) and NAT (dark red surface) that corresponds to the fusicoccin binding site shown in panel C. Right panel, complete binary complex. (C) Structure of tobacco 14-3-3 isoform c (T14-3c) in complex with the last 52 C-terminal amino acids of the plasma membrane H⁺-ATPase PMA2 (PMA2-CT52) stabilized by the fungal toxin fusicoccin (PDB code no. 2o98). Left panel, stereo view of the fusicoccin binding site, with fusicoccin shown as a stick model in yellow (carbon) and red (oxygen), T14-3c in green, and PMA2-CT52 as a blue surface. Right panel, complete ternary complex.

Role of 14-3-3 proteins in C-RAF regulation.An important feature of 14-3-3 proteins is their ability to regulate the activity of many of their partner proteins by modulating their subcellular distribution. For example, one of the most important regulatory mechanisms for the cell-cycle phosphatase Cdc25C is its sequestration in the cytoplasm induced by 14-3-3 proteins (5, 35). The HDAC4 class II histone deacetylase is regulated in a similar manner by binding to 14-3-3 proteins, resulting in its export from the nucleus and, consequently, its inactivation (51). Besides C-RAF, other protein kinases such as Chk1 (9) and c-Abl (56) are also regulated by 14-3-3-mediated subcellular localization. Finally, plasma membrane recruitment of the adapter protein KSR, another component of the Ras signaling pathway, is inhibited by 14-3-3 binding in a manner similar to that seen with C-RAF (25). These examples document the role of 14-3-3 proteins as widespread and important allocators of many different proteins in the cell. As recently demonstrated (6), binding of 14-3-3 to the C terminus of C-RAF (Ser⁶²¹) has an activating effect on this kinase. Thus, 14-3-3 proteins play opposing roles in C-RAF regulation, depending on the site of interaction. This raises the issue of how to obtain site-specific modulation of 14-3-3 binding by pharmacological intervention with small molecules stabilizing protein-protein interactions. Importantly, the primary 14-3-3 binding sites in C-RAF differ significantly in the +1 position after the phosphorylated serine. At the N-terminal Ser²⁵⁹ site, a threonine is located in the +1 position (QRSTpSTPNV) whereas a glutamate can be found at the C-terminal Ser⁶²¹ site (NRSApSEPSL). Consequently, binding of fusicoccin or an identified fragment to a 14-3-3/C-RAFpSer⁶²¹ complex is hindered due to sterical and electrostatic clashes. Thus, the pS/T +1 position in 14-3-3 partners acts as a discriminator for the binding of specific small-molecule stabilizers. Therefore, rational molecule design starting from structural fragments of fusicoccin as well as de novo compound library screening employing the N terminus of C-RAF might indeed yield specific stabilizers for the interaction of 14-3-3 with the N but not the C terminus of C-RAF. As we have recently shown by high-throughput screening, it is indeed possible to identify small-molecule stabilizers of a specific 14-3-3 protein-protein interaction (43).

Modulation of the Ras-RAF-MAPK pathway by targeting the C-RAF/14-3-3 complex.In addition to displaying developmental disorders, patients with Noonan or LEOPARD syndrome also show an increased risk for several malignancies (44). This is not surprising, given the fact that activating mutations in the Ras-RAF-MAPK pathway are present in 30% of all human cancers (7). For C-RAF, some mutations have been reported in conjunction with human cancers (11) and the S259A mutation is one of the two C-RAF mutations found in the COSMIC database (Catalogue of Somatic Mutations in Cancer [www.sanger.ac.uk/genetics/CGP/cosmic/ ]).

Our report supports the hypothesis that binding of 14-3-3 proteins to the N terminus of C-RAF attenuates the Ras-RAF-MAPK pathway by sequestering C-RAF in the cytoplasm. The Ras binding domain (RBD) is also situated at the N terminus of C-RAF (amino acids 51 to 131), and the importance of the Ras/C-RAF interaction for downstream signaling has very recently been confirmed (22). Since binding of Ras and binding of 14-3-3 to the N terminus of C-RAF have opposite functional effects, associations with these C-RAF protein partners can be expected to be mutually exclusive. As a consequence, the competition of Ras and 14-3-3 proteins for C-RAF leads to activation or attenuation of the Ras-RAF-MAPK pathway. Our cellular localization analyses (Fig. 2 to 3) suggest that an impairment of the 14-3-3 interaction with C-RAF results in enforced plasma membrane recruitment by Ras that is independent of external stimuli and illustrates the significance of the 14-3-3 proteins in the regulation of the C-RAF protein kinase. Importantly, the observed impairment of 14-3-3 binding to the N terminus of C-RAF results in the actual activation of the pathway (Fig. 3B and 4A). In addition to providing an explanation of how the C-RAF/14-3-3 binding mutations activated the Ras-RAF-MAPK pathway in a subset of patients with Noonan or LEOPARD syndrome, the conclusions of this work indicate potential therapeutic interventions for such diseases involving targeting the corresponding regulatory protein complexes. As our structural analyses of the effect of the fungal toxin fusicoccin on a regulatory 14-3-3 protein complex have previously shown (32, 53), a molecule stabilizing 14-3-3 protein-protein interactions can induce a dramatic physiological response. Of interest in this regard is the fact that fusicoccin is also able to stabilize the 14-3-3 protein-protein interaction in the absence of phosphorylation of the 14-3-3 target sequence (32). Therefore, these molecules have the potential to be efficient in physiological settings, where, due to deregulated kinase signaling, 14-3-3 target phosphorylation is impaired. In the case of C-RAF, stabilization of 14-3-3 binding to the N terminus of the kinase would lead to increased cytoplasmic sequestration of C-RAF, resulting in a decreased ability of activated Ras to recruit C-RAF to the plasma membrane and thus in attenuated Ras-RAF-MAPK signaling (Fig. 8B). One crucial question, however, is the “druggability” of this 14-3-3/target protein complex. As shown in Fig. 7, cavities at the interface of 14-3-3 proteins with the C-RAFSer²⁵⁹ peptide and with AANAT can be identified at the site of fusicoccin binding in the 14-3-3/PMA2 complex. Since these cavities are considerably smaller than the fusicoccin pocket, it remains to be determined whether a molecule can be found or designed that can specifically occupy these cavities and stabilize the 14-3-3 protein-protein complex. In this regard, fragment-based approaches employing protein crystallography will greatly profit from the new 14-3-3σ/C-RAFpSer²⁵⁹ crystals that allow elucidation of structures in atomic detail. By demonstrating the binding and stabilizing effect of a fusicoccin-derived molecular fragment (Fig. 6), the principle feasibility of this approach is exemplified. This strategy might facilitate the development of stabilizing molecules in the future, leading to the discovery of therapeutical agents that can be used to address developmental disorders such as Noonan or LEOPARD syndrome as well as human cancers that exhibit an activated Ras-RAF-MAPK pathway.

• Open in new tab
• Download powerpoint

FIG. 8.

Model proposing the attenuation of the Ras-RAF-MEK-ERK pathway by small molecules that stabilize the inhibitory RAF/14-3-3 complex. (A) In unstimulated cells, RAF is sequestered by 14-3-3 proteins in the cytoplasm in an inactive conformation. After dephosphorylation of two serine residues (Ser²³³ and Ser²⁵⁹ in C-RAF), 14-3-3 dissociates from RAF, which can then be recruited to the plasma membrane by activated Ras. At the plasma membrane, RAF is itself activated and phosphorylates MEK. Via activating phosphorylation of ERK transcription factors such as Elk1, Ets and Sp1 stimulate genes whose expression leads to cell proliferation and, in cancers, to metastasis and angiogenesis. Furthermore, as the examples of Noonan syndrome and LEOPARD syndrome show, an overstimulation of this pathway can result in developmental defects. (B) The overactivation of the Ras-RAF-MEK-ERK pathway observed in many cancers and in developmental disorders can be neutralized by small molecules that stabilize the inhibitory RAF/14-3-3 complex, thereby preventing RAF plasma membrane recruitment by activated Ras.