Phospho-specific binding of 14-3-3 proteins to phosphatidylinositol 4-kinase III β protects from dephosphorylation and stabilizes lipid kinase activity

Phosphoinositides are involved in many signal transduction pathways in eukaryotic cells (Clarke, 2003). Phosphatidylinositol 4-phosphate [PtdIns(4)P], a lipid that is essential for secretion in yeast (Walch-Solimena and Novick, 1999) is generated from phosphatidylinositol-4,5-biphosphate [PtdIns(4,5)P₂] by the activity of a PI(4,5)P₂-5-phosphatase or from phosphatidylinositol (PtdIns) by the activity of phosphatidylinositol-4 kinases (PI4-kinases) (De Matteis and Godi, 2004a). PI4-kinases are grouped into type II and type III kinases. Type III PI4-kinases are inhibited by Wortmannin, whereas type II PI4-kinases are inhibited by adenosine (Downing et al., 1996). There are three known PI4-kinases in yeast: Pik1p and Stt4p, both of which are type III PI4-kinases, and LSB6, a type II PI4-kinase (Flanagan and Thorner, 1992; Flanagan et al., 1993; Han et al., 2002; Shelton et al., 2003; Audhya et al., 2000). Mammalian cells have two type III PI4-kinases, PI4KIIIβ and PI4KIIIα, and two type II PI4-kinases, named PI4KIIα and β (Minogue et al., 2001; Barylko et al., 2001; Balla et al., 2002; Wei et al., 2002). At least three of these kinases, PI4KIIα, PI4KIIIα and PI4KIIIβ are localized at the Golgi complex (Wang et al., 2003; Wong et al., 1997; Godi et al., 1999), where they contribute to the local production of PtdIns(4)P (Balla et al., 2005). The best characterized kinase so far is PI4KIIIβ, which is recruited to Golgi membranes by active GTP-bound ARF1 and seems to control the structural integrity of the whole Golgi complex (Godi et al., 1999). However, ARF1 itself has only a small direct effect on lipid kinase activity (Haynes et al., 2005). A noted activator of PI4KIIIβ is the neuronal calcium sensor-1 (NCS-1), also known as frequenin, which can exert bidirectional effects on PI4KIIIβ by either directly activating the lipid kinase or inhibiting the activation by ARF1 (Zhao et al., 2001; Haynes et al., 2005). However, the ability of endogenous frequenin to interact and stimulate PI4KIIIβ remains to be defined. In mammalian cells, regulation of PI4KIIIβ lipid kinase activity at the Golgi complex is controlled by members of the Protein Kinase D (PKD) family of serine/threonine kinases (Hausser et al., 2005). PKD localizes to specific TGN domains and plays a crucial role in the formation of post-Golgi transport carriers, possibly by controlling the fission of vesicles (Liljedahl et al., 2001). Both, PKD1 and PKD2 activate PI4KIIIβ by phosphorylation of Ser294. The inhibition of PKD-mediated phosphorylation at this residue significantly reduces PI4KIIIβ lipid kinase activity and affects transport of secretory proteins to the plasma membrane (Hausser et al., 2005). Therefore, among other mechanisms, PKD might control the fission of vesicles by the regulation of the PtdIns(4)P levels at the Golgi compartment. Golgi-localized PtdIns(4)P selectively recruits adaptor proteins, such as FAPP1 and FAPP2, which appear essential for control of constitutive Golgi-to-cell-surface membrane traffic (Godi et al., 2004). Moreover, PI4KIIIβ was reported to be required for recruitment of active rab11 to the TGN (de Graaf et al., 2004). Taken together, there is substantial experimental evidence for PI4KIIIβ being an important player in inositol signalling and regulation of secretory transport processes at the Golgi compartment. Besides this function, recent studies postulate an essential role for PI4KIIIβ in nuclear inositol signalling, too: Upon inhibition of nuclear export with leptomycin B, PI4KIIIβ accumulated in the nucleus, suggesting that the lipid kinase rapidly shuttles between the nucleus and the cytoplasm in a Crm1-dependent process. The presence of two nuclear localization sequences (NLS) and one nuclear export sequence (NES) located in the N-terminal part of the protein are the molecular basis for the observed nucleo-cytoplasmic shuttling (de Graaf et al., 2002; Strahl et al., 2005).

Here we report a novel interaction between human PI4KIIIβ and the multifunctional 14-3-3 proteins. These proteins are small, acidic, ubiquitous molecules that recognize phosphorylated serine/threonine residues in a context-specific manner (Dougherty and Morrison, 2004). In mammals, there are seven highly homologous family members designated with Greek letters (β, ϵ, γ, η, σ, τ, ζ), that bind to many different types of proteins, including cell cycle regulators, transcription factors, and proteins involved in signalling and apoptosis (Bridges and Moorhead, 2004). In this study, we provide evidence that PKD-mediated phosphorylation of PI4KIIIβ at Ser294 induces 14-3-3 binding to this site, whereby lipid kinase activity is maintained through 14-3-3-mediated protection from dephosphorylation.

We recently demonstrated that PKD1 and PKD2 phosphorylate PI4KIIIβ, a lipid kinase localized at the Golgi, at Ser294 to regulate lipid kinase activity (Hausser et al., 2005). The PKD phosphorylation motif LXRXpS²⁹⁴xP in PI4KIIIβ is highly conserved among a wide variety of species (Fig. 1A). Interestingly, PKD substrate phosphorylation has been shown to promote binding of 14-3-3 proteins (Wang et al., 2002). Interaction of 14-3-3 proteins with target proteins is generally mediated through RSXpS/TXP or RXXXpS/TXP sequences, in which pS/T is phosphorylated (Yaffe, 2004). We noted that the PKD phosphorylation residue Ser294 is part of a similar motif in PI4KIIIβ. To test for a possible interaction between the lipid kinase and 14-3-3 proteins, we transiently expressed Flag-PI4KIIIβ in HEK293 cells, prepared whole cell lysates, and performed pulldown experiments using GST-14-3-3-Sepharose beads. Flag-PI4KIIIβ associated with all of the seven GST-14-3-3 isoforms, but not with GST alone (Fig. 1B). For all further experiments, we selected the τ and ζ isoforms as representative members of the 14-3-3 family. To examine whether Ser294 contributes to the interaction between 14-3-3 proteins and PI4KIIIβ, Flag-PI4KIIIβ-transfected HEK293 cells were treated with okadaic acid (OA), a selective inhibitor of the serine/threonine phosphatase PP2A. Treatment with OA caused a substantial increase of Ser294 phosphorylation in PI4KIIIβ, indicated by the enhanced detection of the lipid kinase with the PKD pMOTIF antibody (Fig. 1C, left panel), which specifically recognizes the phosphorylated Ser294 residue (Hausser et al., 2005). Moreover, a shifted band in SDS-PAGE typical for phosphorylated proteins was observed after a 2-hour treatment (Fig. 1C, left panel). Next, we transiently expressed wildtype Flag-PI4KIIIβ and Flag-PI4KIIIβ_S294A mutant, treated cells with OA and performed pulldown experiments using GST-14-3-3τ-Sepharose beads. Treatment of cells with OA increased the interaction of Flag-PI4KIIIβ with 14-3-3τ proteins; in contrast the interaction between the S/A mutant and GST-14-3-3τ was strongly reduced and increased only slightly upon OA treatment, indicating that Ser294 in PI4KIIIβ contributes to 14-3-3 binding (Fig. 1D). The Ser294-dependent interaction was further verified by co-immunoprecipitation of wildtype Flag-PI4KIIIβ with 14-3-3τ, 14-3-3γ and 14-3-3ζ isoforms, whereas the S/A-mutated Flag-PI4KIIIβ showed a strongly reduced binding to 14-3-3τ and γ and no binding to ζ (Fig. 1E, left panel). Moreover, we precipitated endogenous 14-3-3 proteins and analysed their capacity to interact with transiently expressed wildtype or S/A-mutated Flag-PI4KIIIβ. Only the wildtype Flag-PI4KIIIβ protein coprecipitated with endogenous 14-3-3 proteins, whereas the S/A mutant protein failed to bind (Fig. 1E, right panel). In the reciprocal experiment (shown below), endogeneous PI4KIIIβ was precipitated with HA-14-3-3 proteins (Fig. 2C, right panel). Unfortunately, we were unable to detect coprecipitation of the two endogenous proteins under the applied experimental conditions. It is possible that the stoichiometry of basal phosphorylation is low in the cells studied here, making a coprecipitation of the endogenous proteins difficult to detect. The above data nevertheless clearly show that the PKD phosphorylation site Ser294 in PI4KIIIβ is involved in the binding of 14-3-3 proteins. However, both GST-pulldown and co-immunoprecipitation, revealed a residual binding of 14-3-3 to PI4KIIIβ_S294A, suggesting the existence of an additional binding site for these proteins in PI4KIIIβ.

Binding of 14-3-3 protein dimers to a target protein can have different effects: It may lead to a conformational change of the target protein, mask a specific region on the target or act as a scaffold to bring two proteins into close proximity (Bridges and Moorhead, 2004). Like its wildtype counterpart, PI4KIIIβ_S294A still localizes to the Golgi, but displays reduced lipid kinase activity (Hausser et al., 2005). To assess a potential functional link between 14-3-3 binding to the S294 site and lipid kinase activity, we constructed dominant negative forms of 14-3-3 that are deficient in their ability to bind to phosphoserine-containing proteins. We used the double arginine mutant form (R56A and R60A) of human 14-3-3τ. The proposed mechanism for dominant negative action of these 14-3-3 mutants is the formation of inactive heterodimers with wildtype 14-3-3 monomers (Xing et al., 2000). To determine whether DN-14-3-3τ proteins indeed form dimers with wildtype 14-3-3 proteins, we performed BiFC. BiFC detects predominantly stable protein-protein interactions and is based on the fact that non-fluorescent halves of a fluorescent protein can complement each other to form the intact fluorophore. For complementation to occur, an interaction time of the halves is required that is in the range of several seconds (Hu et al., 2002). Wildtype 14-3-3τ and DN-14-3-3τ were fused to the N-terminal (YN-14-3-3τ) or the C-terminal half (YC-14-3-3τ) of YFP. We cotransfected HEK293 cells with the wildtype 14-3-3 constructs (YN-14-3-3τ and YC-14-3-3τ), wildtype and DN-14-3-3 constructs (YC-14-3-3τ and DN-YN-14-3-3τ) and measured BiFC by fluorescence flow cytometry analysis (Fig. 2A). Expression of the two wildtype proteins as well as coexpression of wildtype and dominant negative 14-3-3τ resulted in a distinct population of BiFC-positive cells that could not be detected in the untransfected control (Fig. 2A, left panel). The percentage of cells with YFP fluorescence above the level of untransfected cells was 42% for WT-14-3-3τ-YN and -YC, and 40% for WT-14-3-3τ-YN and DN-14-3-3τ-YC. Confocal microscopy of BiFC-positive COS7 cells revealed a fluorescence signal in the cytoplasm and the nucleus for both pairs, with the DN-14-3-3 proteins displaying a strong accumulation in the nucleus (Fig. 2A, right panel). These results show that dimerisation of dominant negative and wildtype 14-3-3 is equally efficient as homodimerisation of two wildtype proteins in HEK293 cells. However, the mutated HA-14-3-3τ proteins displayed only marginal interaction with transiently expressed Flag-PI4KIIIβ or with endogenous PI4KIIIβ, as shown by GST-pulldown experiments and co-immunoprecipitation assays, confirming its proposed dominant negative action (Fig. 2B,C).

We also analysed the interaction between PI4KIIIβ and 14-3-3τ proteins with BiFC. PI4KIIIβ wildtype was fused to the N-terminal half of YFP (YN-PI4KIIIβ). Cells were cotransfected with PI4KIIIβ and 14-3-3τ constructs and analysed with fluorescence flow cytometry. We detected specific BiFC fluorescence in 50% of all YN-PI4KIIIβ and YC-14-3-3τ cotransfected HEK293 cells, whereas only 22% of YN-PI4KIIIβ and DN-YC-14-3-3τ cotransfected cells were BiFC-positive (Fig. 3A). This clearly demonstrates an interaction of both proteins in intact cells. Next, we analysed the spatial distribution of BiFC triggered by PI4KIIIβ and 14-3-3τ interaction by confocal microscopy in COS7 cells. The YFP fluorescence resulting from complementation was detectable in the cytoplasm and clearly enriched at a perinuclear structure, which colocalizes with the trans-Golgi protein p230 revealing the interaction of both proteins in the cytoplasm and at the Golgi compartment (Fig. 3B).

Recent studies demonstrated that PI4KIIIβ shuttles between the nucleus and the cytoplasm, with a regulated export in a Crm1-dependent manner (de Graaf et al., 2002; Strahl et al., 2005). It is recognized that 14-3-3 proteins regulate subcellular localization of target proteins (Dougherty and Morrison, 2004). Although 14-3-3 proteins do not seem to be involved in the localization of PI4KIIIβ to the Golgi complex (Fig. 4A), their binding to Ser294 might be involved in the regulation of the nuclear-cytoplasmic shuttling of PI4KIIIβ. We transfected COS7 cells with wildtype or S294A Flag-tagged PI4KIIIβ and analyzed the subcellular localization of these proteins in immunofluorescence microscopy. In untreated cells, both the wildtype and the S294A Flag-PI4KIIIβ are cytoplasmic proteins, which were found enriched at the Golgi compartment (Fig. 4A). No nuclear staining was detectable under steady-state conditions. To visualize shuttling through the nucleus, cells were treated with Leptomycin B, an inhibitor of Crm1-dependent nuclear export. In accordance with published data (de Graaf et al., 2002), inhibition of the nuclear-export machinery led to an accumulation of the wildtype Flag-PI4KIIIβ in the nucleus. Likewise, the S294A Flag-PI4KIIIβ was trapped in the nucleus, too, demonstrating that 14-3-3 binding to Ser294 was not important for nuclear shuttling of PI4KIIIβ (Fig. 4A). This was further supported by immunofluorescence studies with DN-HA-14-3-3τ proteins. Endogenous 14-3-3 proteins were localized in the cytosol, to some part in the nucleus and at a perinuclear region. DN-HA-14-3-3τ proteins, which cannot bind to target proteins, were reported to be predominantly localized in the nucleus (Brunet et al., 2002). However, these mutated 14-3-3 proteins did not influence the localization of Flag-PI4KIIIβ at the Golgi complex (Fig. 4B).

We next investigated whether overexpression of DN-14-3-3τ proteins versus wildtype 14-3-3τ proteins influenced the phosphorylation state of PI4KIIIβ at Ser294. We transfected HEK293 cells with the respective plasmids and performed Western blot analysis with the PKD pMOTIF antibody. Expression of DN-14-3-3τ strongly decreased the phosphorylation at Ser294 compared with wildtype 14-3-3τ and the vector control (Fig. 5A).

The loss in Ser294 phosphorylation by expression of DN-14-3-3τ proteins can be a result of the action of phosphatases or an indirect effect by decreased activity of the upstream kinase PKD. To clarify this, we analysed the influence of DN-HA-14-3-3τ expression on PKD kinase activity. The expression of DN-14-3-3τ proteins increased PKD1 kinase activity, demonstrated by enhanced phosphorylation of the substrate aldolase in an in vitro kinase assay (Fig. 5B). This is in accordance with previous results, which showed a negative regulatory role for 14-3-3 proteins on PKD substrate phosphorylation (Hausser et al., 1999). Therefore, a direct role of 14-3-3 in regulating dephosphorylation of PI4KIIIβ appeared probable. We reasoned that 14-3-3 binding to phosphorylated PI4KIIIβ may subsequently protect against dephosphorylation by serine phosphatases such as PP2A. To investigate this, we performed in vitro dephosphorylation assays with purified protein. Fig. 5C shows that the preincubation of purified Flag-PI4KIIIβ with wildtype GST-14-3-3τ substantially blocked λ-phosphatase-mediated dephosphorylation at Ser294, whereas GST or GST-DN-14-3-3τ had no effect. These results provide biochemical evidence that 14-3-3-bound phospho-PI4KIIIβ is specifically protected against phosphatase attack. In accordance with an essential role of Ser294 phosphorylation for PI4KIIIβ activation (Hausser et al., 2005), reduction of Ser294 phosphorylation because of coexpression of DN-HA-14-3-3τ was accompanied with a strong inhibition of lipid kinase activity (50% reduction compared with vector control) (Fig. 6), whereas OA treatment increased phosphorylation at Ser294 and lipid kinase activity (180% compared with vector control) (Fig. 6), providing direct evidence for a role for 14-3-3 in positively regulating lipid kinase activity by protection from dephosphorylation.

In this study, we show a phosphorylation site-specific association between 14-3-3 proteins and the Golgi-associated lipid kinase PI4KIIIβ and uncover the functional consequences of this interaction. The binding of 14-3-3 proteins to PI4KIIIβ was dependent on PKD1- and PKD2-mediated phosphorylation of Ser294, which is required for catalytic activity of the lipid kinase (Hausser et al., 2005). Binding of 14-3-3 proteins to this site stabilized Ser294 phosphorylation by protecting from phosphatases and stabilizing lipid kinase activity. Our data suggest that the Ser294-phosphorylation-dependent association between 14-3-3 proteins and PI4KIIIβ is important to maintain a continuous supply of PtdIns(4)P at the Golgi complex.

14-3-3 proteins are ubiquitously expressed phosphoserine/phosphothreonine-binding proteins that are members of a large family of isoforms (Wilker and Yaffe, 2004). In general, 14-3-3 proteins bind to their ligands through RSXpS/TXP or RXXXpS/TXP sequences in which pS/T is phosphorylated (Yaffe, 2004). Binding of 14-3-3 proteins to PI4KIIIβ was largely dependent on phosphorylated Ser294, which is part of such a motif. This is of particular interest, because phosphorylation of this site is critically involved in regulating lipid kinase activity (Hausser et al., 2005). Interaction of 14-3-3 with its various target proteins can have multiple functional consequences (Bridges and Moorhead, 2004). In the specific case of PI4KIIIβ, our data indicate that binding of 14-3-3 protected the lipid kinase from dephosphorylation at Ser294. This reasoning is in accordance with previously reported functions of 14-3-3 proteins. Several examples exist, in which the binding of 14-3-3 directly protects from dephosphorylation, thereby influencing enzymatic activity or protein stability (Gohla and Bokoch, 2002; Margolis et al., 2003). Here we could show that the functional consequence of 14-3-3 binding to Ser294 in PI4KIIIβ was to keep the lipid kinase in an active state. This was evident from decreased lipid kinase activity upon expression of DN-14-3-3 proteins. Although PI4KIIIβ activity can be discerned in cells without external stimulation, the intracellular signals that regulate and maintain this apparently constitutive activity begin to be uncovered. On the one hand, PI4KIIIβ activity is stimulated by PKD-mediated phosphorylation of Ser294 (Hausser et al., 2005). On the other hand, we now show that PI4KIIIβ activity is subjected to negative regulation by OA-sensitive phosphatases and that 14-3-3 is a modulator of phosphatase sensitivity of PI4KIIIβ. At the concentrations applied, OA is known to be a highly selective inhibitor of PP2A (Millward et al., 1999), suggesting that this ubiquitously expressed phosphatase counteracts PKD activation of PI4KIIIβ. PKD has been recognized as the bottleneck in constitutive transport of basolateral cargo from the TGN to the plasma membrane (Liljedahl et al., 2001). According to a recent model, PKD activity at Golgi membranes is triggered by basolateral cargo in a Gβγ protein-mediated pathway and promotes vesicle fission from the TGN via phosphorylation-dependent activation of several target proteins at Golgi membranes (Diaz Anel and Malhotra, 2005). Our previous studies have identified PI4KIIIβ as one of these important targets in secretory transport (Hausser et al., 2005), placing PI4KIIIβ downstream of PKD within this cargo-triggered signal pathway. We show that 14-3-3 binding subsequent to phosphorylation of the PKD site is an important mechanism to protect the PI4KIIIβ phosphorylation state at Ser294. Therefore, we propose that 14-3-3 binding to Ser294-phosphorylated PI4KIIIβ shifts the balance towards an activated state of the lipid kinase, explaining the observed `constitutive' activity. The maintenance of the phosphorylated state and thus of lipid kinase activity may ensure the continuous production of PtdIns(4)P at Golgi membranes. This is supported by a recent publication by Szivak and co-workers demonstrating that Ser294-phosphorylated PI4KIIIβ is exclusively associated with the Golgi compartment (Szivak et al., 2006). PtdIns(4)P is an important lipid mediator with multiple functional roles, among which its essential involvement in the functional organisation of Golgi membranes and secretory transport processes, for example via recruitment of effector proteins such as FAPP, appears prominent (Walch-Solimena and Novick, 1999; Godi et al., 2004; De Matteis and Godi, 2004b). Although four PI4-kinases exist in mammalian cells and thus should contribute to steady state PtdIns(4)P levels, a different topology and distinct functional activities of the three PI4K isoforms at the Golgi complex have been suggested, but at least partially redundant functions are probable and have not been ruled out (Weixel et al., 2005). In accordance with our data, a functional role of the 14-3-3 homologues in yeast, Bmh1p and Bmh2p in vesicular transport has been proposed (Gelperin et al., 1995). As the PI4K homologue of yeast, Pik1p, is known to be essential for secretion (Walch-Solimena and Novick, 1999), it will be exciting to find out whether a similar regulatory mechanism of PI4-kinase activity as described here exists in yeast. Considering the multitude of known 14-3-3 binding partners (Pozuelo et al., 2004), it is not surprising that 14-3-3 proteins also have bridging functions, mediating interaction between two different binding partners. Whether this is also the case for PI4KIIIβ has to be further investigated.

M. Olayioye (University of Stuttgart, Germany) generously provided HA-tagged 14-3-3τ cDNA in a pEF vector, pRC-14-3-3ζ and Glu-Glu-tagged 14-3-3γ expression constructs. The pGEX-14-3-3 constructs GST-14-3-3β, GST-14-3-3ϵ, GST-14-3-3γ, GST-14-3-3η and GST-14-3-3σ were obtained from C. Walker and J. Bergeron (University of Texas, USA), GST-14-3-3τ and GST-14-3-3ζ were described previously (Hausser et al., 1999). Generation of dominant negative (DN) HA-14-3-3τ_R56/60A or GST-14-3-3τ_R56/60A (Xing et al., 2000) was performed using site-directed mutagenesis according to the manufacturer's instructions (Stratagene, The Netherlands). Integrity of the constructs was verified by sequencing. Plasmids encoding Flag-tagged PI4KIIIβ wildtype and the S294A mutant as well as PKD1-GFP and PKD1-KD-GFP were described previously (Hausser et al., 2005). Commercially available antibodies used were: HA-specific mouse monoclonal antibody (NEB, Germany), anti-Flag M2 mouse monoclonal antibody and anti-HA rabbit polyclonal (Sigma-Aldrich, Germany), anti-PI4KIIIβ mouse monoclonal, anti-p230 mouse monoclonal antibody (BD Biosciences, Germany), anti-PI4KIIIβ rabbit polyclonal (Upstate, MA, USA), anti-14-3-3β K19 rabbit polyclonal (Santa Cruz, Germany), anti-GST polyclonal goat (GE Healthcare, Germany) and anti-GFP mouse monoclonal antibody (Roche Applied Science, Germany). The rabbit polyclonal antibody anti-PKD pMOTIF was described previously (Hausser et al., 2005; Doppler et al., 2005). Secondary antibodies used were goat anti-mouse IgG, Alexa488 coupled, and goat anti-rabbit IgG, Alexa546 coupled (Invitrogen, Germany), goat anti-mouse and anti-rabbit IgG HRP coupled, as well as goat anti-mouse IgG Cy5 coupled (Dianova, Germany) and goat anti-mouse and anti-rabbit and donkey anti-goat IgG AP coupled (Sigma-Aldrich, Germany).

HEK293 cells and COS7 cells were maintained in RPMI 1640 medium supplemented with 5% fetal calf serum. For transient transfections, HEK293 cells were transfected with TransIT293 (Mirus Bio Corporation, WI, USA) according to the manufacturer's instructions. COS7 cells were transfected with Lipofectamine 2000 (Invitrogen, Germany). Treatment of HEK293 cells with okadaic acid (Merck Biosciences GmbH, Germany) was at 100 nM, treatment of COS7 cells with leptomycin B (Biomol, Germany) was at 10 ng/ml for 16 hours. To support BiFC, 24 hours after transfection the incubation temperature was decreased from 37°C to 30°C. After 18 hours of incubation, cells were further analysed.

Transfected COS7 cells were grown on coverslips, washed with PBS, fixed in 4% paraformaldehyde at room temperature (RT) for 20 minutes, washed, permeabilized with 0.1% Triton X-100 (5 minutes, RT) and blocked with blocking buffer (5% normal goat serum and 0.05% Tween 20 in PBS) for 30 minutes. The cells were incubated with the primary antibodies diluted in blocking buffer (1 μg/ml) for 2 hours, washed, incubated with secondary antibodies diluted in blocking buffer for 1 hour, washed, mounted in Fluormount G (Southern Biotechnology, AL) and analyzed on a Confocal Laser Scanning Microscope (TCS SP2, Leica, Germany). GFP and Alexa488 were excited with the 488 nm line of the argon laser and fluorescence was detected at 500-550 nm. EYFP fluorescence, after bi-molecular complementation of its two separately expressed fractions, was excited with the 514 nm line of the argon laser and detected at 525-620 nm. The dyes Alexa546 and Cy5 were excited with the 543 nm and the 633 nm line of the helium-neon laser and detected at 555-620 nm and 640-750 nm, respectively. Cells were imaged with a 40.0×/1.25 HCX PL APO or a 100×/1.3 HCX PL APO objective lens. Images were processed with Adobe Photoshop.

Induction of GST-14-3-3 protein production and purification of the fusion protein were performed as described by Hausser et al. (Hausser et al., 1999). In brief, GST-14-3-3 protein expression was induced with 0.1 mM isopropyl-β-D-1-thiogalactopyranoside for 4 hours. Following induction, the bacterial cultures were harvested and the pellets were washed in ice-cold phosphate-buffered saline, resuspended in 10 ml PBS and sonicated for 30 seconds on ice. Afterwards, Triton X-100 was added to a final concentration of 1% and the suspension was centrifuged for 30 minutes at 8000 g. Purification of GST-14-3-3 was performed by mixing with glutathione sepharose 4B (Amersham Biosciences, Germany) for 1 hour at 4°C. The sepharose was washed three times with PBS and the amount of bound GST-14-3-3 was then determined by SDS-PAGE and Coomassie staining. Elution of GST-14-3-3 proteins was performed in 10 mM Glutathione/50 mM Tris, pH 8, at RT for 10 minutes.

Whole cell extracts were obtained by solubilizing cells in lysis buffer [20 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100 or NP-40, 5 mM MgCl₂, 10 mM sodium fluoride, 20 mM β-glycerophosphate plus complete protease inhibitors (Roche Applied Science, Germany)]. Lysates were clarified by centrifugation at 13,000 g for 10 minutes.

Pulldowns were performed by incubation of protein lysates with 5 μg GST or GST-14-3-3 coupled to glutathione sepharose beads for 2 hours at 4°C. Beads were washed three times with lysis buffer (see above). For immunoprecipitation equal amounts of proteins were incubated with specific antibodies for 1.5 hours at 4°C. Immune complexes were collected with protein G-Sepharose (GE Healthcare) and washed three times with lysis buffer. Precipitated proteins were released by boiling in sample buffer and were subjected to SDS-PAGE. The proteins were blotted onto nitrocellulose membranes (Pall, Germany). After blocking with 1% blocking reagent (Roche Applied Science, Germany), filters were probed with specific antibodies. Proteins were visualized with AP-coupled secondary antibodies using NBT/BCIP as substrate or HRP-coupled secondary antibodies using ECL.

Lysates of HEK293 cells expressing Flag-PI4KIIIβ were immunoprecipitated with Flag-M2-Agarose (Sigma-Aldrich, Germany) according to the manufacturer's instructions. Immunoprecipitates were washed three times with lysis buffer, three times with PBS, and Flag-PI4KIIIβ was eluted from the beads with 100 nM Glycin, pH 2.5, and immediately neutralized using 1 M Tris, pH 8.8. For phosphatase treatment, purified Flag-PI4KIIIβ was incubated on ice with PBS or 1-4 μg of GST or GST-14-3-3 fusion proteins in phosphatase buffer overnight. Dephosphorylation was performed for 30 minutes at 30°C, with or without 10 units of λ-phosphatase (NEB, Germany). The protein complexes were subjected to SDS-PAGE and further analyzed by Western blotting.

After immunoprecipitation (with anti-GFP antibodies for transfected PKD), the kinase reaction was performed for 15 minutes at 37°C in 30 μl kinase buffer (50 mM Tris, pH 7.4, 10 mM MgCl₂ and 2 mM DTT). Reaction was started by addition of 10 μl of a kinase buffer mixture containing 2 μCi [γ-³²P]-ATP and 5 μg aldolase as substrate. To terminate reaction, 10 μl of 5× SDS-sample buffer was added and the samples were resolved by SDS-PAGE, blotted onto nitrocellulose and analyzed on a phosphoimager (GE Healthcare). Quantification was done with ImageQuant software (GE Healthcare).

The activity of Flag-tagged PI4KIIIβ was measured as incorporation of radioactivity from [γ-³²P]-ATP into organic solvent-extractable material. HEK293 cells grown on Petri dishes were transfected with vectors encoding Flag-tagged PI4KIIIβ. The cells were cultured for 48 hours, harvested, lysed and Flag-tagged PI4KIIIβ was immunoprecipitated using anti-Flag M2 antibody. The standard reaction mixture for PI4KIIIβ (100 μl volume) contained 100 mM MgCl₂, 10 mM HEPES, pH 7, 4 μg phosphatidylinositol (Biomol, Germany) and 10 μCi [γ-³²P]-ATP. Reactions were started by addition of [γ-³²P]-ATP and terminated after 10 minutes by the addition of 25 μl 5 M HCl. Lipids were extracted by adding 160 μl Chloroform/Methanol followed by vigorously mixing and high-speed centrifugation. The bottom phase containing the lipids was transferred to a new tube, 20 μl at a time were spotted onto a potassium oxalate-coated TLC plate, and lipids were separated using 0.7 M acetic acid in N-Propanol as solvent. TLC was exposed to phosphoimager (GE Healthcare) overnight. Density of spots representing PtdIns(4)P was quantified using ImageQuant software (GE Healthcare).

The BiFC fragments of EYFP (YN: residues 1-155; YC: residues 156-239) were selected according to Hu et al. (Hu et al., 2002). Sequences encoding PI4KIIIβ, 14-3-3τ and DN-14-3-3τ were fused to sequences encoding YN or YC to yield the constructs pEYN-C1-PI4KIIIβ, pEYC-C1-14-3-3τ, pEYN-C1-14-3-3τ, pEYN-C1-DN-14-3-3τ and pEYC-C1-DN-14-3-3τ. After complementation of YC-14-3-3τ and YN-14-3-3τ or YN-PI4KIIIβ and YC-14-3-3τ, the reconstituted yellow fluorescent protein was imaged with the same parameters as common EYFP (as described under Immunofluorescence and confocal microscopy). BiFC was also analysed by quantitative fluorescence flow cytometry. After fluorescence complementation, cells were washed with PBS and resuspended in PBS containing 5% FCS and 0.05% sodium azide on ice. The analysis was performed with an Epics XL-MCL (Beckman-Coulter, Germany) with FL1 set on logarithmic scale and reflects cell fluorescence of 10,000 cells after gating out debris and dead cells.

We thank Peter Storz (Mayo Clinic, Jacksonville, FL) and Alex Toker (Harvard Medical School, Boston, MA) for supplying us with the PKD pMOTIF antibody. We are grateful to Monilola A. Olayioye for thoughtful discussions. This work was supported by grants of the Deutsche Forschungsgemeinschaft (SFB 495 to K.P.) and the Landesstiftung Baden-Württemberg (to A.H.). The authors declare that they have no competing financial interests.