The 14-3-3 interactome. A, A VisANT graph showing overlaps among the lists of proteins identified for their affinity for 14-3-3s in the proteomics screens listed in Table I, with data collated in supplemental Table S1. Each paper is assigned a node in gray and lines connect the articles to the proteins, which are assigned as green nodes. Gold standard proteins for which 14-3-3-binding sites were reported in previous low-throughput studies (1) are in red, whereas in orange are proteins for which direct binding to 14-3-3 and relevant binding sites were characterized in this study. Isoforms of kinesin heavy chain are in blue to signify that these proteins bind to 14-3-3s indirectly via phosphorylated kinesin light chains (see text). The interactive version of this graph is available under 14-3-3 partners at our laboratory web page via http://www.ppu.mrc.ac.uk/research. B, The 14-3-3 binding proteome was clustered by molecular function or cellular localization as listed in supplemental Table S3. Note that the MitoMiner database does not distinguish cytoplasmic proteins that interact with mitochondria from intrinsic mitochondrial proteins.

Phosphorylated sites whose mutation prevents 14-3-3 binding. GFP-MFF, GFP-DBNL, ISCU-GFP, and REEP4-GFP, and the indicated single and double alanine mutants of these proteins, were immunoprecipitated on GFP-TRAP® beads from transfected cells, and tested for their ability to bind directly to 14-3-3s in Far Western assays, and by copurification with endogenous 14-3-3s from the cell extracts. For ISCU, the upper GFP signal is the intact protein (which runs as a doublet in Fig. 2) and a fainter lower band is the processed form that lacks the N-terminal signal sequence. For GFP-DBNL, the immunoprecipitated proteins were also examined for phosphorylation of Ser269 and Thr291 using phosphospecific antibodies, as characterized in supplementary Fig. S2.

The phosphorylation-dependent binding of 14-3-3s to kinesin light chain isoforms. A, To identify in vitro AMPK-phosphorylated sites on GST-KLC2 that promote binding to 14-3-3s, GST-KLC2 was phosphorylated by AMPK in vitro with Mg[γ³²P]ATP. A sample was checked for binding to 14-3-3 in a Far Western assay (similar to Fig. 5B. WT), and the remainder digested with trypsin and run on HPLC with the indicated gradient of acetonitrile in 0.1% trifluoroacetic acid. The ³²P trace shows three phosphorylated peptides P1, P2, and P3, which were identified by mass spectrometry to be singly phosphorylated peptides with the amino acid sequences shown. The recovery of ³²P from HPLC was 87%. The phosphorylated residue in P3 was not pinpointed in this experiment, but other data (see text) are consistent with it being Ser582. The bottom panel shows that a release of radioactive inorganic phosphate at the third cycle of solid-phase Edman degradation of P1 revealed Ser545 as the phosphorylated residue in that peptide (performed as in (49)). B, 14-3-3 binding to GST-KLC2 phosphorylated in vitro with AMPK and the effects of Ala and Asp substitutions of Ser residues. Proteins were incubated with AMPK with and without MgATP, and subject to 14-3-3 Far Western binding assay, and immunoblotting with antiphosphoSer545 and antiphosphoSer582, with anti-GST as loading control. In addition to the phosphoSer545 and phosphoSer582 14-3-3 binding sites identified here (see text), mutations of the adjacent Ser581 and also Ser428 (reported to be phosphorylated (http://www.phosphosite.org)) and within a Rxx(pS)xP sequence that looks like a 14-3-3-binding motif) were tested, giving no indication that these sites are involved in 14-3-3 binding. As an aside, note that in contrast to 14-3-3, the pSer582 antibody recognizes KLC2 with Asp as a phosphomimetic in place of pSer582. In contrast, the antiphosphoSer545 antibody does not recognize Asp545. Also, mutation of the neighboring residue Ser581 to Ala reduces the ability of the pSer582 antibody, but not 14-3-3, to recognize pSer582. C, HA-KLC2 was isolated from transfected HEK293 cells that were serum deprived for 5 h, incubated with or without H-89 (30 μm for 30 min), and stimulated with the adenylate cyclase activator forskolin (20 μm for 30 min). The HA-KLC2 was probed for binding to antiphosphoSer545-KLC2, antiphosphoSer582-KLC2 and DIG-14-3-3. The PKA-phosphorylation of Ser157 on VASP was monitored to assess the activation of PKA. D, Wild-type and Ser545Ala/Ser582Ala forms of GFP-KLC2 were isolated from lysates of transfected HEK293 cells in the presence and absence of 2.5 mg/ml dithiobis[succinimidyl propionate] (DSP) to stabilize any labile interactions and assayed for direct binding to 14-3-3 in a Far Western, and for the presence of associated proteins with the indicated antibodies. E, C-terminal sequences of KLC2 from various species aligned, together with human KLC3 and KLC4 and the C-terminal splice variants of human KLC1 (KNS2 gene; Ensembl ENST00000303439) as reported previously (50). The asterisks align with the positions shaded in gray of Ser545 and Ser582 in human KLC2. Residues in lowercase letters have been shown to be phosphorylated here and elsewhere (http://www.phosphosite.org). Underlined residues (Ser611, Ser615) on mouse KLC2 were proposed as GSK3 sites (51); with the preceding bold S proposed as a priming site, though we note that Ser615 has not been shown to be phosphorylated and our attempts to phosphorylate KLC2 with GSK3 in vitro and in vivo, in experiments using priming kinases and GSK3 inhibitors, were unsuccessful (not shown, including experiments with Adam Cole, University of Dundee). Accession numbers are: hKLC1 Q07866 (human), hKLC2 Q9H0B6 (human), hKLC3 Q6P597 (human), hKLC4 Q9NSK0 (human), mKLC2 O88448 (mouse) and dKLC1 Q7ZVT4 (Zebra fish, Danio rerio). F, Wild type and mutant HA-tagged KLC1 and KLC2 proteins were immunoprecipitated from HEK293 cells probed with antiphosphoSer545-KLC2, antiphosphoSer582-KLC2 and by Far Western with DIG-14-3-3.

Direct phosphorylation-dependent interactions of 14-3-3s with selected proteins. Tagged recombinant proteins were isolated from lysates of transfected HEK293 cells that were grown in medium containing serum, and treated or not with calyculin A (CA) as indicated. Proteins were dephosphorylated with lambda phosphatase, or not, when the phosphatase was inhibited with EDTA, and analyzed for retention of co-purified endogenous 14-3-3 proteins (K19 pan-14-3-3 antibody). The ability of recombinant proteins to bind directly to 14-3-3s was examined by Far Western assay. DNBL, KLC2 and SMAUG2 were also analyzed by Western blotting with the indicated phospho-specific antibodies, which were characterized in supplemental Fig. S2.

The phosphorylation-dependent binding of 14-3-3s to SMAUG2. A, Analysis of the 14-3-3 binding of truncated SMAUG2 proteins. HEK293 cells were transfected to express the truncated HA-SMAUG2 proteins indicated. Immunoprecipitations were performed with anti-HA antibodies and 14-3-3 binding was assessed using the DIG-14-3-3 Far Western overlay assay. B, 14-3-3 binding of alanine mutants of HA-SMAUG2. Recombinant proteins were immunoprecipitated from 1 mg lysate of transfected HEK293 cells and endogenous SMAUG2 from 10 mg untransfected cell lysate, and assayed for their ability to bind to 14-3-3s in a Far Western assay. Copurifying endogenous 14-3-3s were detected with the K19 pan-14-3-3 antibody. PACS2 protein that binds 14-3-3s and mutant PACS2 that does not bind 14-3-3s (1) were positive and negative controls, respectively. C, Contribution of phosphorylated Ser642 of SMAUG2 to 14-3-3 binding. GFP-SMAUG2 proteins isolated from 1 mg lysate of transfected HEK293 cells were analyzed with a pSer642-SMAUG2 phosphospecific antibody, 14-3-3 Far Western assay and for coprecipitation of endogenous 14-3-3s. D, Tandem affinity purification of GFP-TAP-SMAUG2 proteins. Stable cell lines were generated to express the GFP-TAP proteins at close to endogenous levels when induced with tetracycline (30 ng/ml) for 16 h. Tandem affinity-purified proteins were separated by SDS-PAGE, stained with Coomassie colloidal blue, excised and digested with trypsin for mass fingerprint identification. The doublet running at ∼30 kDa with wild type, but missing in the T484A/S642A SMAUG2 mutant purification, was identified as 14-3-3. E, Mass spectrometric identification of phosphorylated Ser642 in recombinant SMAUG2. The MS/MS spectrum shows b- and y- ions resulting from CID fragmentation of THSLPVHSSPQAILMFPP. These data suggest phosphorylation of the third serine in the peptide (Ser642 in SMAUG2) with an ion score of 55 where individual ion scores >33 indicate identity or extensive homology (p < 0.05).