2015 Volume 40 Issue 2 Pages 95-104
Protein phosphorylation plays a key role in regulating nearly all intracellular biological events. However, poorly developed phospho-specific antibodies and low phosphoprotein abundance make it difficult to study phosphoproteins. Cellular protein phosphorylation data have been obtained using phosphoproteomic approaches, but the detection of low-abundance or fast-cycling phosphorylation sites remains a challenge. Enrichment of phosphoproteins together with phosphopeptides may greatly enhance the spectrum of low-abundance but biologically important phosphoproteins. Previously, we used 14-3-3ζ to selectively enrich for HeLa cell lysate phosphoproteins. However, because 14-3-3 does not isolate phosphoproteins lacking the 14-3-3-binding motif, we looked for other domains that could complementarily enrich for phosphoproteins. We here assessed and characterized the phosphoprotein binding domains Pin1-WW, CHEK2-FHA, and DLG1-GK. Using a strategy based on affinity chromatography, phosphoproteins were collected from the lysates of HeLa cells treated with phosphatase inhibitor or cAMP activator. We identified different subsets of phosphoproteins associated with WW or FHA after calyculin A, okadaic acid, or forskolin treatment. Our Kinase-Oriented Substrate Screening (KiOSS) method, which used phosphoprotein-binding domains, showed that WW and FHA are applicable and useful for the identification of novel phospho-substrates for kinases and can therefore be used as biological filters for comprehensive phosphoproteome analysis.
As one of the most common post-translational modifications in eukaryotic cells, protein phosphorylation regulates almost every biological event, including growth, differentiation, cell death, and signal transduction (Hunter, 2000; Ubersax and Ferrell, 2007). Together with phosphatases, protein kinases regulate the extent of phosphorylation. Hence, the identification of proteome-wide phosphorylation dynamics will help in understanding cell signaling events (Aebersold and Mann, 2003; Taylor and Kornev, 2011). Liquid chromatography tandem mass spectrometry (LC-MS/MS)-based phosphoproteomics has become an indispensable tool for identifying global phosphopeptides from complex protein samples (Grønborg et al., 2002; Aebersold and Mann, 2003). However, the transient nature of protein phosphorylation state and the low relative abundance of phosphoproteins compared to the highly abundant structural or housekeeping proteins, make them difficult to identify using LC-MS/MS. The removal of abundant non-phosphoproteins or peptides and the selective enrichment of phosphoproteins and phosphopeptides may enhance the spectrum of phosphoproteome analyses (Paradela and Albar, 2008). Various approaches have been developed to enrich and identify kinase substrates of interest (Grønborg et al., 2002; Johnson and Hunter, 2005; Hamaguchi et al., 2015). Both in vitro and in vivo data are informative in identifying broad classes of the phosphoproteome. However, methods to obtain real substrates still need to be improved. In this regard, we have developed in vivo methods to identify the substrates of various kinases by enriching phosphoproteins using 14-3-3 (Nishioka et al., 2012, 2015). The family of 14-3-3 proteins specifically binds to a group of phosphoproteins and cooperatively mediates intracellular signals. However, certain binding motif specificities of the 14-3-3 protein such as the R(S/R)XpSXP motif (Yaffe et al., 1997; Yang et al., 2006), limits its phosphoprotein enrichment. To detect additional biologically important phosphoproteins, we therefore explored whether the other phosphopeptide binding domains could be used to enrich the phosphoproteins from a complex mixture of proteins such as cell lysates. In addition to 14-3-3, tryptophan-tryptophan (WW) domain, the forkhead associated (FHA) domain, and the guanylate kinase (GK) domain were known to associate with phosphoproteins (Lu et al., 1999; Li et al., 2000; Sudol and Hunter, 2000; Mohammad and Yaffe, 2009; Zhu et al., 2011). Because each of these domains has a different binding specificity for phosphoserine or phosphothreonine (pS/pT)-containing sequences, they are potential tools for identifying various kinase substrates that may provide useful information in deciphering complex signaling networks.
In this study, we characterized the phosphoprotein enrichment potential of several pS/pT-binding domains. We found that, unlike 14-3-3, which primarily isolates cytoplasmic and membrane phosphoproteins, WW and FHA can efficiently enrich nuclear and membrane-enclosed lumen phosphoproteins from HeLa cells treated with phosphatase inhibitors. These domains could enrich specific kinase oriented candidate substrates such as protein kinase A (PKA).
The phosphatase inhibitors calyculin A (PP1 and PP2A inhibitor) and okadaic acid (primarily PP2A inhibitor) were purchased from Wako Chemicals (Osaka, Japan), and LC Laboratories (Woburn, MA, USA), respectively. The Adenylate cyclase activator forskolin (FSK) and the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX) were purchased from Merck Millipore (Billerica, MA, US).
The rabbit monoclonal phospho-motif antibodies of phospho-PKA (#9624), phospho-ATM/ATR (#9607), phospho-MAPK/CDK (#2325), and phospho-Akt substrates (#10001) were obtained from Cell Signaling Technology (Denver, MA, USA). Goat polyclonal TNKS1BP1 antibody was obtained from Santa Cruz Biotechnology (Dallas, USA). Rabbit polyclonal CHAMP1 antibody was purchased from Bethyl Laboratories (Texas, USA).
Plasmids and protein purificationThe cDNA encoding 14-3-3ζ was cloned from a human fetal brain cDNA library (Clontech Laboratories, Inc., California, USA). The cDNAs encoding the human CHEK2 FHA was purchased from Kazusa (Kazusa DNA Res Inst, Chiba, Japan). The cDNAs encoding the human Pin1-WW and DLG1-GK were obtained from a human fetal brain cDNA library and cloned into the pGEX-2T vector. GST-fusion proteins were produced in E. Coli (BL21-pLys) and purified with Glutathione Sepharose 4B (GE Healthcare).
Cell culture and stimulationsHeLa cells (1×106 cells/10 cm dish) were cultured in DMEM with 10% fetal bovine serum. After overnight incubation, cells were treated with the phosphatase inhibitor calyculin A (50 nM until cells started to round up, approximately 9 min), okadaic acid (1 μM for 30 min), or a mixture of the PKA activator FSK (20 μM) and the phosphodiesterase inhibitor IBMX (125 μM) for 30 min. After stimulation for the indicated time, culture medium was aspirated and washed with ice-cold phosphate buffered saline (PBS) on ice. Lysis buffer was applied immediately after washing. Cells were scraped off the culture dish and cellular proteins extracted.
Pull-down assayThe proteins in HeLa cells were extracted using lysis buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 100 mM NaCl, 1 mM dithiothreitol, 20 mM β-glycerophosphate, 20 mM NaF, 100 μM APMSF, 0.5 μg/mL aprotinin, 10 μg/mL leupeptin, and 1.0% NP-40), followed by centrifugation at 17,000 g for 20 min at 4°C. The supernatant was then incubated with GST fusion bait protein-coated beads for an hour at 4°C. After the bound proteins were precipitated with the beads, the beads were washed three times with lysis buffer and three times with wash buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 100 mM NaCl, 1 mM dithiothreitol). To elute bound proteins, samples were boiled in SDS sample buffer and subjected to silver-staining or immunoblot analysis with the indicated antibodies.
Phosphoproteomic analysis and mass spectrometryThe method of sample preparation for mass spectrometry was performed as previously described (Nishioka et al., 2012). Briefly, after pull-down using the WW, FHA, GK, or 14-3-3ζ domains, bound proteins were eluted using guanidine solution (7 M guanidine and 50 mM Tris/HCl), and the eluate was subjected to reduction, alkylation, demineralization, concentration and finally digestion with trypsin. Phosphopeptide enrichment was performed with a Titansphere® Phos-TiO kit (GL Sciences, Tokyo, Japan) according to the manufacturer`s instructions. All LC-MS/MS experiments were performed using Q-Exactive mass spectrometry (Thermo-Fisher Scientific Inc., Waltham, MA) equipped with an HTC-PAL autosampler and the Michrom nano-Advance UHPLC (Michrom BioResources Inc., CA) with a MonoCap C18 Nano-flow (0.1×150 mm) column (GL Science, Tokyo, Japan).
Proteomic data analysisAnalysis of the raw data was performed with MaxQuant software (version 1.3.0.5), integrated with the Andromeda search engine (Cox and Mann, 2008). Database searches were performed against the complete proteome of Homo sapiens in UniProtKB 2013_07 and concatenated with reverse copies of all the sequences (Peng et al., 2003). Fixed modification was set to carbamidomethylation, and variable modifications were set to oxidation of methionine, phosphorylation of serine/threonine/tyrosine, and N-terminal acetylation. A total of three missed trypsin cleavages were allowed. The mass tolerance for a fragment ion was set to 0.5 Da, and the maximum peptide posterior error probability (PEP) that essentially operates as p-value was set to 1. Minimum peptide length for detection was set to 7. The false discovery rate (FDR) for the peptide, protein, and site levels was set to 0.01. Search parameters were set to an initial precursor ion tolerance of 7 ppm. All raw files obtained from an experiment by LC-MS/MS were processed together, and quantified by label free quantification based on peptide extracted ion chromatogram. Data were non-normalized, since the phosphatase inhibitor treated samples substantially increases number and extent of phosphoproteins compared to the control. Phosphorylation sites with a more than five-fold increase in ion intensity in the drug-treated samples versus the control in at least two independent experiments were considered to be candidate kinase substrates.
Immunoprecipitation assayHeLa cells (1×106 cells/10 cm dish) were grown in DMEM with 10% fetal bovine serum and incubated overnight. The cells were treated with PKA inhibitor H-89 (10 μM) for 30 min, and/or a mixture of PKA activator FSK (20 μM) and phosphodiesterase inhibitor IBMX (125 μM) for 30 min. After the indicated time, culture medium was aspirated and washed with ice-cold PBS on ice. Cells were then lysed in 1 mL of lysis buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 1.0% NP-40, 1 mM dithiothreitol, 80 mM β-glycerophosphate, 80 mM NaF, 100 μM APMSF, 1 μg/mL aprotinin, and 4 μg/mL leupeptin) and centrifuged at 17,000 g for 20 min at 4°C. Supernatants were incubated with anti-TNKS1BP1 antibody or anti-CHAMP1 antibody for 1 hr at 4°C with continuous rotation and immobilized for an additional 1 hr with protein G or A sepharose 4 fast flow beads respectively (GE Healthcare). Beads were washed twice with lysis buffer and thrice with wash buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 150 mM NaCl). Finally, bound proteins were eluted by boiling with SDS-sample buffer.
Because the 14-3-3 phospho-motif is limited to binding specific subsets of phosphoproteins, we characterized phosphoprotein-binding domains with additional recognition motifs. The Pin1-WW, and CHEK2-FHA domains recognize primarily (pS/pT)P and pTXX(D/I/S/Y), respectively (Yaffe and Elia, 2001), whereas the DLG1 GK domain recognizes phosphorylated serine with arginine at the -3 position (Zhu et al., 2011). To enrich phosphoproteins, HeLa cells were treated with calyculin A (PP1 and PP2A inhibitor) or okadaic acid (PP2A inhibitor mainly) and subjected to GST affinity chromatography with Pin1-WW, CHEK2-FHA, DLG1-GK, and 14-3-3ζ. The bound proteins were eluted with SDS sample buffer and examined by silver staining (Fig. 1A) and immunoblotting (Fig. 1B). Calyculin A and okadaic acid treatment induced an increase in the number and amount of phosphoproteins recognized by PKA (RRXpS/T), ATM/ATR (pSQ), MAPK/CDK (PXpSP), and AKT (RXXpS/T) motif antibodies (Fig. 1B). CHEK2-FHA and Pin1-WW precipitated different phosphoproteins than 14-3-3ζ, suggesting these two domains are good tools for identifying diverse signaling substrates such as PKA, MAPK/CDK, ATM/ATR, and AKT. Compared to Pin1-WW, CHEK2-FHA, and 14-3-3ζ, DLG1-GK did not demonstrate strong phosphoprotein binding in the drug-treated samples, suggesting this domain is not a useful tool to identify PKA, MAPK/CDK, ATM/ATR, and AKT kinase substrates (Fig. 1B).
Evaluation of phosphoprotein binding by the phosphoprotein binding domains WW, FHA, GK, and 14-3-3ζ. (A) GST-pull down assay, performed from HeLa cell lysates treated with the phosphatase inhibitor calyculin A (50 nM for 9 min) or okadaic acid (1 μM for 30 min) using the baits: Pin1-WW, CHEK2-FHA, DLG1-GK, and 14-3-3ζ (indicated as black arrowheads). Silver staining of the pulled down samples showed many new or enhanced protein bands for different baits. (B) Immunoblot analysis with the indicated antibodies showed enrichment in phosphoproteins from the pulled-down samples of HeLa lysates treated with phosphatase inhibitor. The input amount represented 0.2% of the whole-cell lysates that was used for pull-down analysis, and the pulled-down samples represented 10% of the eluted samples.
The proteins bound to WW, FHA, GK, or 14-3-3ζ were identified by LC-MS/MS analysis. Using WW, FHA, GK, and 14-3-3ζ, we identified 46, 156, 14, and 146 phosphoproteins, respectively, whose phosphopeptide intensities in calyculin A-treated cells were more than five times those in control cells in at least two independent experiments (Fig. 2A). These data indicate that WW, FHA, and 14-3-3ζ bound to a broader range of phosphoproteins than GK. It is also noteworthy that most of the phosphoproteins (~85%) were not detected in control samples, rather detected only in calyculin A treated samples strongly suggesting a phosphorylation dependent binding. The Venn diagram shows that 26, 114, and 107 phosphoproteins were specifically associated with WW, FHA, and 14-3-3ζ, respectively (Fig. 2B). Subcellular localization based on GO classification showed that, in comparison with 14-3-3ζ, WW and FHA preferentially bound to nuclear and membrane-enclosed lumen phosphoproteins (Fig. 2C). Sequence alignment of the phosphopeptides showed that WW preferentially bound to proline-rich (pS/pT)P motifs. FHA bound to the (pS/pT)P motif with preference for E/I/D in the +3 position and basic amino acids in the –3 position, whereas 14-3-3ζ prefers RXXpS (where X is any amino acid) (Fig. 2D). Of these motifs, threonine-phosphorylated sequences are more prominent in WW and FHA-bound phosphoproteins than for 14-3-3. The motif observed for WW-bound phosphopeptides is consistent with previously reported data (Yaffe and Elia, 2001), while the FHA-bound phosphopeptide motif is partially consistent because FHA was reported as phospho-threonine binding modules in cell signaling (Li et al., 2002; Hammet et al., 2003). KEGG pathway analysis from the phosphoproteins bound to these domains showed that both WW and FHA could bind to phosphoproteins involved in the cell cycle and cancer development (Table I). FHA also tends to bind phosphoproteins involved in RNA transport, oocyte meiosis, ubiquitin-mediated proteolysis, and Wnt signaling pathway, indicating a possible role for this domain in multi-signaling complexes (Table I). A similar tendency for binding to phosphoproteins was observed in okadaic acid-treated samples (data not shown). These results suggest that much like 14-3-3ζ, the WW and FHA domains enable the enrichment of specific phosphoproteins subsets.
Summary of the identified candidate phosphoproteins from calyculin A-stimulated HeLa lysates. (A) The number of phosphoproteins and their phosphorylation sites obtained with the WW, FHA, GK, and 14-3-3ζ baits following phosphatase inhibitor calyculin A treatment are shown. (B) The number of phosphoproteins bound to each bait was compared using a Venn diagram (Created by Venny. Oliveros JC, (2007-2015). http://bioinfogp.cnb.csic.es/tools/venny/index.html). (C) Subcellular localization of the candidate phosphoproteins bound to each bait based on GO slim classification of the cellular components (http://bioinfo.vanderbilt.edu/webgestalt/). (D) The sequence alignment of the phosphopeptides bound to each of the WW, FHA, and 14-3-3ζ baits showed different phosphopeptide binding preferences (http://www.phosphosite.org/).
Pathway Name | Gene | Gene symbol | |
---|---|---|---|
WW | RNA transport | 4 | NUP107; NUP88; NUP98; RANBP2 |
Cell cycle | 3 | CDC27; HDAC1; HDAC2 | |
Pathways in cancer | 3 | CCDC6; HDAC1; HDAC2 | |
FHA | RNA transport | 9 | ACIN1; EIF5B; NUP107; NUP133; NUP153; NUP98; RANBP2; SMN1; TPR |
Cell cycle | 8 | ANAPC1; BUB1; BUB1B; CDC23; DBF4; HDAC2; MCM3; RAD21 | |
Pathways in cancer | 8 | APC; DVL1; DVL2; HDAC2; MAPK1; RALGDS; RASSF1; TPR | |
Oocyte meiosis | 5 | ANAPC1; BUB1; CDC23; ITPR3; MAPK1 | |
Ubiquitin mediated proteolysis | 5 | ANAPC1; CDC23; HERC2; UBE2O; UBR5 | |
Wnt signaling pathway | 4 | APC; CHD8; DVL1; DVL2 | |
14-3-3ζ | Regulation of actin cytoskeleton | 5 | APC; ARHGEF7; BAIAP2; GIT1; PAK4 |
MAPK signaling pathway | 5 | CDC25B; DUSP16; MAP3K2; NF1; RAPGEF2 | |
Insulin signaling pathway | 4 | IRS2; PDE3A; PPP1R3D; TSC2 | |
Adherens junction | 4 | BAIAP2; CTNND1; LMO7; PARD3 | |
Endocytosis | 4 | GIT1; IQSEC1; PARD3; RAB11FIP1 |
Data from the phosphatase inhibitor calyculin A showed binding by the WW, FHA, and 14-3-3ζ baits to specific pathway substrates.
Based on the data from the phosphatase inhibitor calyculin A, we employed WW and FHA, in addition to 14-3-3ζ, to enrich PKA signaling substrates. HeLa cells were treated with a mixture of forskolin (FSK) and the phosphodiestarase inhibitor 3-isobutyl-1-methylxanthine (IBMX) to activate endogenous PKA, and phosphoproteins were pulled down using bait from the cell lysates. Immunoblotting data (Fig. 3A) showed that the amount of phosphoproteins containing RRXpS/T motif associated with WW, FHA, and 14-3-3ζ was increased upon FSK/IBMX stimulation. Phosphoproteins eluted from WW, FHA, and 14-3-3ζ, were analyzed by LC-MS/MS, and we identified 26, 46, and 36 phosphoproteins responding to FSK/IBMX associated with these baits, respectively (Fig. 3B). A Venn diagram (Fig. 3C) shows that very few phosphoproteins are common among the baits, and the numbers of unique, non-overlapping phosphoproteins bound to these domains were 19, 38, and 33, respectively. Sequence alignment of the FSK/IBMX-stimulated phosphopeptides bound to WW, FHA, and 14-3-3ζ showed similarity with the PKA phospho-motif sequence (obtained from previously reported data at www.phosphosite.org), suggesting these domains bind PKA signaling substrates.
Identification of candidate substrates for PKA from HeLa lysates. (A) Immunoblotting with phospho-PKA motif antibody was performed from the GST-pull down samples of FSK/IBMX-treated HeLa lysates. Input indicates 0.2% of the whole-cell lysates that was used for pull-down analysis, and the pulled-down samples represented 10% of the eluted samples. (B) The number of phosphoproteins bound to each of the WW, FHA, and 14-3-3ζ baits after FSK/IBMX treatment are shown. (C) The number of phosphoproteins bound to each bait domain was compared using a Venn diagram. (D) The sequence motif obtained from FSK/IBMX-stimulated phosphopeptides data showed preference for the PKA motif by the WW, FHA, and 14-3-3ζ baits.
To examine whether the candidate substrates are phosphorylated by PKA in vivo, we focused on TNKS1BP1, the 182 kDa Tankyrase 1 binding protein 1. TNKS1BP1 is both a nuclear and cytoplasmic protein that is involved in the ionizing radiation (IR)-induced DNA damage response and increases in expression following IR exposure. TNKS1BP1 depletion impairs DNA double-strand break efficiency and increases the sensitivity of cells to IR (Zou et al., 2015).
TNKS1BP1 is a highly phosphorylated protein. From our proteomic data, obtained from HeLa cells treated with the phosphatase inhibitor calyculin A, we found 2 phosphorylation sites (494S, 498S) obtained with WW and 5 phosphorylation sites (429S, 494S, 601S, 691S, 836S) with 14-3-3ζ, whereas FHA failed to detect any phosphorylation sites. From the FSK/IBMX proteomics data, we found only one phosphorylation site (893S) that specifically binds the WW domain (Fig. 4A). Next, we examined whether PKA phosphorylates TNKS1BP1 in vivo. We treated HeLa cells with H-89 and/or a mixture of FSK/IBMX, immunoprecipitated endogenous TNKS1BP1 with anti-TNKS1BP1 antibody, and immunoblotted with PKA phospho-substrates motif antibody. TNKS1BP1 phosphorylation was increased in FSK/IBMX-treated cells and reduced upon H-89 exposure (Fig. 4B). These results suggest that PKA phosphorylates TNKS1BP1 in HeLa cells. We obtained similar results for another PKA candidate substrate CHAMP1 (Chromosome alignment-maintaining phosphoprotein 1) obtained with Pin1 WW (Supplementary Fig. 1).
Phosphorylation of TNKS1BP1 by PKA in vivo. (A) Phosphorylation sites of the TNKS1BP1 protein, identified by WW, FHA, and 14-3-3ζ. (B) Immunoprecipitation analysis with anti-TNKS1BP1 antibody, performed from HeLa cell lysates treated with control, H-89, FSK/IBMX, and H-89+(FSK/IBMX), followed by immunoblotting with phospho-PKA motif antibody.
In our study, we characterized the phosphoprotein binding potential of several domains such as WW, FHA, GK, and 14-3-3. Silver staining data (Fig. 1A) showed clear enhancement in protein levels in phosphatase inhibitor-treated samples in the case of WW, GK, and 14-3-3ζ, suggesting these proteins undergo phosphorylation state-dependent binding to the bait domains. Because the FHA domain bound to so many proteins, we could not distinguish the enhancement of phosphorylation-dependent protein binding (Fig. 1A). However, immunoblotting with various kinase phospho-substrate motif antibodies demonstrated that WW and FHA, together with 14-3-3, bound to many phospho-proteins (Fig. 1B). From our phosphoproteomic data, we found 46, 156, 14, and 146 phosphoproteins bound to WW, FHA, GK, and 14-3-3, respectively, upon calyculin A stimulation (Supplementary Table 1, Fig. 2A). The low number for GK-bound phosphoproteins (consistent with the immunoblotting data in Fig. 1B) limits its usage as a tool for identifying various types of kinase phospho-substrates. Analysis of subcellular localization (Fig. 2B) showed that the WW and FHA domains primarily bound nuclear and luminal membrane phosphoproteins, whereas 14-3-3 bound cytoplasmic and membrane phosphoproteins. These data potentially reflect the subcellular localization of the physiological binding partners since both Pin1 (with a WW domain) and CHEK2 (with a FHA domain) works mainly in nucleus in vivo (Wu et al., 2001; Takahashi et al., 2008). Thus, compared with 14-3-3, WW and FHA differed in their preferred phosphoprotein subset, making these two domains important tools for compartment-targeted phosphoproteomic study (Fig. 2C).
The consensus binding motifs of Pin1 WW and CHEK2 FHA are (pS/pT)P, and pTXX(S/I/D), respectively, whereas that of 14-3-3 is RXXpS (Yaffe and Elia, 2001). Analysis of the motifs from phosphopeptides that have at least a five-fold increase in intensity in calyculin A-treated samples showed a high consistency for WW and 14-3-3 (Fig. 2D). Previous reports suggest that the FHA domain recognizes the pTXX(S/I/D) motif (Yaffe and Elia, 2001; Hammet et al., 2003), but the sequence alignment of the phosphopeptides derived from the FHA precipitant does not completely match with this motif. In our study, the ability of the FHA domain to bind its targets was not dramatically changed by calyculin A or okadaic acid treatment (Fig. 1A), suggesting that most of the phosphorylation sites on the target proteins that bind FHA are constantly phosphorylated under study conditions. On the other hand, upon calyculin A or okadaic acid-stimulation, the target proteins appeared to be phosphorylated at other sites than the FHA-binding sites (Fig. 1B), potentially reflecting differences between the phospho-motifs.
Consistent with a previous report that the WW and FHA domains are crucial regulators of cell cycle progression and the DNA damage response (Reinhardt and Yaffe, 2013), KEGG pathway analysis (Table I) showed that both WW and FHA are capable of precipitating phosphoproteins involved in the cell cycle and cancer development. We also found additional phosphoproteins involved in oocyte meiosis, RNA transport, ubiquitin-mediated proteolysis, and Wnt signaling pathway that were precipitated by FHA. Consistent with previous reports, 14-3-3 precipitated phosphoproteins involved in the MAPK signaling pathway and the regulation of the actin cytoskeleton and the insulin-signaling pathway (Jin et al., 2004; Kleppe et al., 2011; Pozuelo-Rubio, 2012). These data suggest these domains may be used to identify biologically functional phosphoproteins.
To capture PKA signaling molecules, we found 26, 46, and 36 PKA candidate substrates, using the WW domain, FHA domain, and 14-3-3 proteins, respectively (Supplementary Table 2, Fig. 3B). Twelve of the candidate substrates (CAMKK1; CTNNB1; DSP; MARK3; NCOR1; NEDD4L; PDE3A; PDE3B; PFKFB2; RAF1; TPR; TRIM32) obtained with these baits had already been reported as PKA substrates. Alignment of candidate phosphopeptide sequences clearly indicated that the WW and FHA domains as well as the 14-3-3 protein retain the PKA motif after FSK/IBMX stimulation (Fig. 3D). These data illustrate the utility of these domains in identifying substrates of particular kinases such as PKA. Stimulation of cells with various growth factors or kinase activators or inhibitors and precipitating phosphoproteins with WW, FHA, and 14-3-3 presents a viable option for identifying various signaling molecules.
From our proteomic data, we identified TNKS1BP1 as a PKA candidate substrate that was obtained from WW-bound phosphoproteins, following FSK/IBMX stimulation. We examined the phosphorylation of TNKS1BP1 by PKA upon FSK/IBMX stimulation in HeLa cells, and the immunoprecipitation data (Fig. 4B) suggest that TNKS1BP1 is a PKA substrate. TNKS1BP1 is a highly phosphorylated protein, involved in DNA double strand breakage repair (Zou et al., 2015). We hypothesize that PKA first phosphorylates TNKS1BP1, and then WW binds, modulating DNA double strand repair. Further study is needed to identify the physiological meaning of this interaction.
In summary, our approach of identifying kinase oriented substrates screening (KiOSS) using the phosphoprotein-binding domains WW and FHA provided many novel substrates, including known substrates that are mainly nuclear or enclosed in the luminal membrane. Together, these data suggest that the WW and FHA domains serve as biological filters that can be used to identify phosphosignaling downstream of a specific kinase.
We thank all the members of Kaibuchi laboratory, especially Drs. D. Tsuboi, T. Watanabe, K. Kuroda, Y. Funahashi and K. Kato for discussion and technical support. We are grateful to Mss. Y. Kanazawa and T. Ishii for technical and secretarial assistance. This research was supported in part by Grant-in-Aid for Scientific Research (A) (25251021) and (C) (23590357) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT). Part of this study results from “Bioinformatics for brain sciences,” carried out under the Strategic Research Program for Brain Sciences and Grant-in-Aid for Scientific Research on Innovative Areas (Comprehensive Brain Science Network) by MEXT.