The retrieval and handling of protein interaction data is challenging. Despite general efforts in data organization and sharing (40
), much information is spread across databases and it is often difficult for the researcher to identify and retrieve a subset of protein-protein interactions from the literature to compare with their own data. Here, we organized published data on interactions of 14-3-3s with individual proteins for submission to the MINT database, as a resource for the community interested in this family of proteins.
In addition, published high-throughput proteomics data on 14-3-3-binding proteins were drawn together from multiple sources and visually represented in VisANT graphs to facilitate its exploration. The VisANT visualization shows that each experiment identifies new proteins, but also misses proteins identified in other studies, even comparing studies using same cell type and similar experimental procedures. To some extent, this situation is an understandable consequence of the state-of-the-art in mass spectrometry. A solution to this problem could be to make inclusion lists of validated 14-3-3-binding proteins to instruct instruments to specifically monitor ion masses for the corresponding peptides, and exclusion lists for the machine to ignore. Reiterations of biochemical validation and high-throughput experiments (in both “inclusion list” and open “discovery” modes) will improve the robustness of mass spectrometry for tracking the dynamics of the 14-3-3-phosphoproteome in response to extracellular stimuli, drugs, and diseases.
However, the VisANT graphs starkly highlighted the sparsity of validated proteins that could be used to make such inclusion lists. To improve this situation, we therefore made a major push to extend the validated data sets. Recent advances that helped these validation experiments are the databases of experimentally identified phosphorylated residues (http://phospho.elm.eu.org/
), from which to identify sites of potential interest guided by the specificities identified from published 14-3-3-binding sites (1
), and the GFP-Trap reagent for clean isolation of GFP-tagged proteins from lysates of transfected cells is also useful. Nevertheless, validation is still laborious, requiring a variety of strategies depending on the characteristics of individual phosphorylation sites and proteins. For example, phosphorylated Ser642 of SMAUG2 was identified only by using a phospho-specific antibody and when a trypsin/AspN double digest generated a phosphopeptide within the mass range for analysis.
As well as enhancing datasets to underpin 14-3-3-phosphoproteomics, our findings also point to interesting biology for future discovery. For example, concerns that 14-3-3-affinity purified proteins might be contaminated with mitochondria and ER proteins were not borne out with those proteins tested. 14-3-3 binds to a site in the cytoplasmic side of the ER-tethered protein REEP4 that is conserved in the REEP1 protein, whose mutation causes hereditary spastic paraplegia 31 (9
). 14-3-3 must bind to ISCU in the cytoplasm, which is compatible with the cytoplasmic localization of 14-3-3 proteins, because the phosphorylated binding site(s) are in the mitochondrial import sequence of ISCU that is cleaved off when the protein enters the mitochondria. Also, 14-3-3s bind to a site in the cytoplasmic part of mitochondrial fission factor (). Given the central roles of 14-3-3s in upregulating glucose uptake and glycolysis (41
), it will be interesting to determine how 14-3-3 binding to mitochondrial proteins influences the balance between mitochondrial and cytoplasmic metabolism, which is commonly deregulated in cancers and other diseases. Indeed, mRNA processing defects in ISCU cause a myopathy with exercise intolerance and lactic acidosis, showing how mitochondrial defects can have secondary effects on cytoplasmic metabolism (43
Pairs of 14-3-3-binding sites sometimes lie either side of a functional domain on a target protein, and interestingly Thr484 and Ser642 of SMAUG2 flank what is predicted to be a globular domain by GlobPlot analysis (http://globplot.embl.de/
), but whose function is unknown (). Thr484 and Ser642 are conserved across the vertebrate forms of SMAUG2, but not the sequences from invertebrates such as Drosophila
where the function of Smaug is better defined. Other than 14-3-3s no other proteins were identified to bind to SMAUG2 however, and one possibility is that 14-3-3 binding modulates SMAUG2 binding to mRNAs. For DBNL (actin-binding protein 1), the 14-3-3 binding sites flank a site of proteolysis by calpain 2, which may influence formation of cellular dorsal ruffles (46
Kinesin is a motor complex that transports proteins along microtubules in the anterograde direction toward the peripheries of cells. The motor action of the paired KHCs “walking” along microtubules has been dissected in exquisite detail, and although KLCs are less well understood they inhibit the KHC motor in the absence of cargo and contribute to cargo loading, with different KLC variants specifying which cargoes are loaded up for trafficking (47
). Our finding that 14-3-3s bind to phosphorylated residues on KLC2, KLC3, KLC4 and isoforms of KLC1 containing sequences expressed from exon 16 and/or exon 17, suggest that 14-3-3s may be involved in specifying cargo loading/unloading. A KLC1 variant that is enriched on mitochondria and rough endoplasmic reticulum, KLC1B, has neither of the 14-3-3 binding sites identified in the present study, whereas variants located on the Golgi apparatus, KLC1D and KLC1E lack Ser582 but contain Ser545 (47
). The specific cargoes of the KLC1 isoforms that contain both 14-3-3-binding sites have not been defined however, and one possibility is that the 14-3-3s are themselves the cargoes, contributing to the roles of 14-3-3s in cell polarity (48