A Central Kinase Domain of Type I Phosphatidylinositol Phosphate Kinases Is Sufficient to Prime Exocytosis

Exocytosis, a critical process for neuronal communication and hormonal regulation, involves several distinct steps including MgATP-dependent priming (which involves the synthesis of phosphatidylinositol 4,5-bisphosphate). Type I phosphatidylinositol phosphate kinases (PIPKIs) were purified biochemically as a priming factor. PIPKI consists of three domains: the N-terminal region, the central kinase domain, and the C-terminal region. Three isoforms (α, β, and γ) of PIPKI have been identified, and each is alternatively spliced at the C-terminal region. In the present study, we conducted a structure/function analysis of PIPKIs in the priming of exocytosis, and we found that recombinant PIPKIα and PIPKIγ had priming activity. However, an unexpected finding of these results was that PIPKIβ did not prime exocytosis. The N- or C-terminal region of PIPKIα and PIPKIγ was not required for priming, which indicates that the central kinase domain is sufficient for this process. Alternative splicing in each isoform did not affect the isoform specificity in priming. Priming activity by isoforms is strongly correlated with their phosphatidylinositol phosphate kinase activity because PIPKIα and PIPKIγ had higher kinase activity than PIPKIβ. These results suggest that PIPKIα and PIPKIγ are the critical priming factors for exocytosis; it also suggests that the levels of phosphatidylinositol phosphate kinase activity in producing phosphatidylinositol 4,5-bisphosphate specify the function of PIPKI isoforms in priming.

identified as a triggering factor (5). The recognition of a role for PIPKI and phosphatidylinositol transfer protein has led researchers to hypothesize that PIP2 generation by PIPKI and phosphatidylinositol transfer protein, along with membraneassociated phosphatidylinositol 4-kinase (15), is a key component of exocytosis.
Molecular cloning has identified three isoforms of PIPKI: ␣, ␤, and ␥ (16 -18). Two different research groups, working independently, have simultaneously given the same isoforms different names ("PIPKI␣" and "PIPKI␤") (16,17). In this article, we will follow the nomenclature of Loijens and Anderson (17). PIPKI␥ is expressed primarily in the brain (18), where it is concentrated at the synapse of the neurons (19); PIPKI␣ and PIPKI␤, by contrast, are expressed ubiquitously (16,17). The in vivo function of PIPKI␥ has been investigated recently through the generation of PIPKI␥ knock-out mice. The mice die after birth, and their vesicle trafficking at the synapse is severely disrupted (20).
Each PIPKI isoform is alternatively spliced at the C-terminal region (Refs. 16 and 17 and the present study). The splicing sequence (residues 636 -661) in PIPKI␥ is critical for binding to talin (21,22), which is a component of focal adhesion plaques. It was found that only the longer isoform of PIPKI␥ concentrates at focal adhesion points of the cell via a talin link, where it enhances the PIP kinase activity of PIPKI␥. What has not been examined, however, is whether this alternative splicing also regulates the priming activity of PIPKI.
Overexpression of PIPKI isoforms in COS7 cells results in massive actin polymerization (16,23); this suggests that PIP-KIs may also be involved in the regulation of actin cytoskeleton, in addition to their role in membrane trafficking. Overexpression of PIPKI␣, but not of PIPKI␤, has effects on the endocytosis of the epidermal growth factor receptor (24). Similar isoform specificity may exist in PIPKI-mediated priming: Aikawa and Martin (25) found that the transfection of PC12 cells with PIPKI␣ or PIPKI␥, but not with PIPKI␤, reversed the ARF6 transfection-mediated inhibition of priming. However, the authors of that study could not draw a conclusion about this isoform specificity, given that they found that the expression level of transfected PIPKI␤ in PC12 cells was lower than it was in other isoforms (25). It is unclear what determines these potential isoform specificities. In this article, we show that bacterially expressed recombinant PIPKIs exhibit priming activity. Using these recombinant PIPKIs, we have addressed the isoform specificity as well as the mechanism that underlies it. We also attempted to determine whether alternative splicing in PIPKIs affects their priming activity, and whether the central kinase domain of PIPKIs is sufficient for the priming of exocytosis.

EXPERIMENTAL PROCEDURES
PC12 Secretion Assay-PC12 cells were maintained in 10-cm dishes with 8 ml of Dulbecco's modified Eagle's medium containing 5% calf serum (Hyclone), 5% horse serum (Hyclone), and 100 units/ml penicillin and streptomycin (Sigma) at 37°C in 9.5% CO 2 . The secretion assay followed the protocols of previously published work (4). PC12 cells were labeled for 12-20 h with 4 l of [ 3 H]norepinephrine (NE; 56.4 Ci/mmol; PerkinElmer Life Sciences) in the presence of 0.5 mM ascorbic acid. After washing, the cells were harvested in KGlu buffer (20 mM HEPES, pH 7.2, 120 mM potassium glutamate, 20 mM potassium acetate, and 2 mM EGTA) with 0.1% bovine serum albumin, permeabilized with a ball homogenizer (H&Y Enterprise), and incubated for 1-3 h on ice, in the presence of 10 mM EGTA, in order to extract the cytosolic proteins. Two-stage secretion assays were performed in KGlu buffer with 0.05-0.1% bovine serum albumin. Thirty-minute priming incubations at 30°C contained permeabilized PC12 cells, 2 mM MgATP, and recombinant proteins (or 1.0 mg/ml rat brain cytosol). The cells were recovered by centrifugation, washed with KGlu buffer with 0.1% bovine serum albumin, and used for 5-min triggering incubations at 30°C that contained Ca 2ϩ (1.72 mM; free Ca 2ϩ concentrations are estimated to be ϳ1-10 M) and 0.5 mg/ml rat brain cytosol, which provides the Ca 2ϩdependent activator protein for secretion required for triggering (5).
Expression and Purification of GST Fusion Proteins-Proteins were expressed in Origami B (DE3) strain (Novagen) or Origami B (DE3) harboring pG-Tf2 (Takara) and purified with glutathione-agarose (Sigma) (9,14), which in some cases included MgATP washing. Proteins were eluted with an elution buffer (50 mM Tris, pH 8.0, 300 mM NaCl, 1 mM EDTA, and 10 mM glutathione), and the eluted proteins were then concentrated by centrifugation with Nanosep 30 (Pall Gelman). Concentrated proteins (ϳ0.2 mg/ml) were dialyzed with KGlu buffer. In some cases, GST fusion proteins were cleaved by thrombin (Roche Applied Science) in a cleavage buffer (50 mM Tris, pH 8.0, 150 mM NaCl, and 2.5 mM CaCl 2 ). Cleaved proteins were then processed for the secretion assay, similar to the eluted proteins.
Kinase Activity Assays-PIP kinase activity was measured in 50-l reactions, performed for 11 min at room temperature in a final concentration of 50 mM Tris-HCl, pH 7.5, 1 mM EGTA, 10 mM MgCl 2 , 80 M phosphatidylinositol 4-phosphate (Sigma), 50 M ATP, and 10 Ci of [␥-32 P]ATP. The amounts of recombinant proteins used were 0.4 g for GST and GST-PIPKIs. The reaction was stopped by the addition of 100 l of 1 N HCl, and lipids were then extracted with 200 l of chloroform/ methanol (1:1). This was followed by further extraction with 80 l of methanol/1 N HCl (1:1). The extracted lipids were separated using thin layer chromatography plates (Silica Gel 60; Merck) in chroloform/methanol/15 N NH 4 OH/distilled H 2 O (90:90:7:22), and the labeled products were detected by autoradiography (16,17).

Recombinant PIPKI␣ Primes
Exocytosis-First, we replicated the two-stage assay using permeabilized PC12 cells (4) (Fig. 1). As previously described, incubation of the PC12 cell ghosts with brain cytosol and MgATP at the priming stage enhanced Ca 2ϩ -dependent neurotransmitter exocytosis at the triggering stage. Omission of either brain cytosol or MgATP during the priming stage resulted in poor Ca 2ϩ -dependent exocytosis during the triggering stage (Fig. 1). Incubation with brain cytosol plus MgATP at the priming stage resulted in an ϳ3-fold increase in exocytosis at triggering, compared with incubation with MgATP alone at the same stage ( Fig. 1) (t 10 ϭ 11.3, p Ͻ 0.0001). This significant effect provided sufficient resolution to allow us to examine the effect of an individual priming factor.
We expressed full-length mouse PIPKI␣ as a GST fusion protein (GST-mPIPKI␣-1) in Escherichia coli. The purified protein was eluted with glutathione and dialyzed with the same buffer (KGlu buffer) used in the secretion assay (see "Experimental Procedures") ( Fig. 2A). We questioned whether the recombinant PIPKI␣ alone (i.e. without brain cytosol) could significantly prime exocytosis. Purified GST-mPIPKI␣-1 and MgATP were introduced to permeabilized PC12 cells at the priming stage, and their effects on exocytosis at the triggering stage were examined. Recombinant PIPKI␣ exhibited dose-dependent priming activity and could prime up to 60% above the control (Fig. 2B), whereas purified GST had no effect. The dose required for priming was similar to the required dose of partially purified native PIPKI from erythrocytes (7), suggesting FIG. 1. Establishment of two-stage assay using permeabilized PC12 cells. Both MgATP and brain cytosol are required for efficient priming for NE release. The permeabilized PC12 cell ghosts were primed by incubation for 30 min with MgATP (Ⅺ), MgATP plus brain cytosol (f), or brain cytosol alone ( ). The primed cells were washed once and incubated for 5 min with brain cytosol Ϯ Ca 2ϩ at the triggering stage. Error bars indicate S.E. (n ϭ 6).

FIG. 2. Recombinant PIPKI␣ primes neurotransmitter exocytosis.
A, analysis of purified GST (ϳ2 g) and full-length mouse PIPKI␣ fused with GST (GST-mPIPKI␣-1) by SDS-PAGE and Coomassie Blue staining. In this gel, the amount of full-length product of GST-PIPKI␣-1 was estimated to be ϳ2 g. B, GST-PIPKI␣-1 (but not GST) primed NE release. At the priming stage, PC12 cells were incubated with MgATP and the indicated amounts of either GST or GST-mPIPKI␣-1. The primed cells were washed and incubated with Ca 2ϩ and brain cytosol to trigger NE release. As a result of incubation with only MgATP during priming, the average NE release was set at 100% for each experiment. Error bars indicate S.E. (n ϭ 4). that recombinant PIPKI␣ and native PIPKI are similarly active in priming. We were able to conclude that PIPKI␣ is indeed an important factor for priming and that recombinant PIPKIs are useful tools for the study of the structure/function relationship of PIPKI in priming of neurotransmitter exocytosis.
Isoform Specificity in Priming of Exocytosis and the Role of Alternative Splicing-Molecular cloning revealed three isoforms of PIPKIs, and each isoform is alternatively spliced. In PIPKI␥, the C terminus is alternatively spliced (shown as D in Fig. 3A) (18), and this splicing regulates binding to the focal adhesion protein, talin (21,22). This binding in turn augments the PIP kinase activity of PIPKI␥. The C terminus of PIPKI␤ is similarly spliced (shown as C in Fig. 3A) (17). In contrast, human PIPKI␣ is alternatively spliced at the region in the immediate proximity of the central kinase domain (shown as A in Fig. 3A). We found a nearly identical splicing in mouse PIPKI␣ by searching the mouse expressed sequence tag data base (Fig. 3B). In addition, we found a novel splicing site in mouse PIPKI␤ (shown as B in Fig. 3A). We compared splicing site A in (human and mouse) PIPKI␣ and splicing site B in PIPKI␤ by aligning the sequences of PIPKI␣ and PIPKI␤; we found that these two splicing sites exist at similar positions (Fig. 3B). Thus, the region in the immediate proximity of the central kinase domains of both PIPKI␣ and PIPKI␤ is alternatively spliced by a similar mechanism; this finding suggests that there is a functional significance.
PIPKI␤ has a shorter N-terminal sequence than PIPKI␣ and PIPKI␥ (Fig. 3A), raising the possibility that the N-terminallinked GST might hinder PIPKI␤ action. We cleaved GST from both GST-mPIPKI␤-4 and GST-mPIPKI␣-1 using thrombin, and then we tested the cleaved PIPK␣ and PIPKI␤ in priming (Fig. 5). Isoform specificity remained because PIPKI␣, but not PIPKI␤, was still able to exhibit priming. From this, we concluded that isoform specificity is inherent to PIPKIs and is not solely due to a GST artifact.
A Central Kinase Domain Is Sufficient to Prime Exocytosis-We also attempted to determine the critical domains of PIPKIs for priming. Previous deletion analysis of PIPKIs has determined the minimal kinase domains of PIPKIs (18), and all PIPKI isoform kinase domains are located in the central region (Fig. 3A). PIPKIs contain N-terminal and C-terminal regions that may be critical for protein-protein interactions (21,22), as well as other functions. We suspected that the N-terminal region might contribute to the observed isoform specificity because, as noted above, PIPKI␣ and PIPKI␥ have a longer Nterminal region than PIPKI␤ does. To test this possibility, we generated the N-terminal truncation protein of PIPKI␥ (Fig.  6A). The truncated PIPKI␥ (GST-mPIPKI␥-2) exhibited robust priming that was comparable to the priming of the full-length PIPKI␥ (Fig. 6B). We also generated a C-terminal truncation for PIPKI␣ (Fig. 6A), and we were able to identify strong priming by the truncated PIPKI␣ (GST-mPIPKI␣-3, Fig. 6C). Furthermore, the truncation of both N-and C-terminal regions of PIPKI␣ (GST-PIPKI␣-4) did not reduce priming, indicating that the kinase domain of PIPKI␣, and perhaps PIPKI␥ as well, is sufficient for priming (Fig. 6C).
Isoform Specificity Is Determined by PIP Kinase Activity-Our finding of unexpected isoform specificity in priming raised two possibilities. First, the generation of PIP2 alone may be an inadequate explanation for the priming actions of PIPKIs (assuming that they all have comparable PIP kinase activity). The second possibility is that differences in the degree of kinase activity among PIPKI isoforms may underlie the differences in priming. To further examine these two possibilities, we compared the kinase activity of the recombinant PIPKI isoforms using phosphatidylinositol 4-phosphate as a substrate. Thin layer chromatography was used to separate the PIPKI phos- pholipid products. We found dramatic differences in PIP kinase activity among isoforms of PIPKI (Fig. 7). Both GST-mPIPKI␣-1 and GST-mPIPKI␥-3 exhibited strong kinase activity. In contrast, GST-mPIPKI␤-4 showed much lower kinase activity. Control experiments using GST alone did not show any kinase activity. These results indicate a strong correlation between PIP kinase activity and the priming of exocytosis; this suggests that the lack of priming activity by PIPKI␤ is due to its relatively poor kinase activity. Our results also support the hypothesis that PIP2 generation is the critical element for priming.

DISCUSSION
Two PIPKIs of 68 and 90 kDa were partially purified as priming factors for neurotransmitter exocytosis in permeabilized PC12 cells, with the 90-kDa protein having the major activity (7). Molecular cloning of PIPKI (16 -18) followed this finding; it is worth recalling that priming by recombinant PIPKI has never been demonstrated. In addition, a structure/ function analysis of PIPKIs has been unavailable. The present study has succeeded in demonstrating the priming activity of recombinant PIPKI␣ and PIPKI␥. Our results suggest that the partially purified 68-kDa protein corresponds to PIPKI␣, whereas the 90-kDa protein corresponds to PIPKI␥. The results also imply that the predominant activity of the 90-kDa PIPKI, in the course of purification from brain cytosol (7), is due to the fact that the presence of PIPKI␥ in the brain is greater than that of PIPKI␣, rather than the higher priming activity of PIPKI␥. Furthermore, we have shown that the kinase domain of PIPKI␣ (and probably PIPKI␥) alone is sufficient for priming. Unexpectedly, we found that PIPKI␤ had little or no priming activity, compared with PIPKI␣. This result could be attributed to the lower PIP kinase activity of PIPKI␤, compared with PIPKI␣ and PIPKI␥ (Fig. 7).
Similar isoform specificity has been suggested in other func- tions of PIPKIs. In particular, the effects of PIPKI␣ overexpression on the endocytosis of epidermal growth factor receptor may be another such function (24). Because epidermal growth factor receptor endocytosis requires the kinase activity of PIPKI␣, the lack of an effect upon the overexpression of PIPKI␤ may be due to its lower kinase activity. In contrast, all the isoforms of PIPKI were able to induce similar levels of actin polymerization upon overexpression (18,23). Unexpectedly, actin polymerization by PIPKI seems to be independent of kinase activity (18,23). Thus, the function of PIPKI isoforms may be differentially regulated (depending on the requirements of their PIP kinase activity), although the molecular mechanisms defining the isoform specificity in PIP kinase activity require further study.
It has been found that the C-terminal regions of PIPKI isoforms are alternatively spliced. We found a novel splicing site just in the immediate proximity of the central kinase domain in PIPKI␤, which is nearly identical to the site in PIPKI␣ (Fig. 3). We tested the function of these splicing sites in PIPKI␣ and PIPKI␤, as well as splicing in the C terminus of PIPKI␥, in the regulation of their priming activities. However, our permeabilized secretion assays did not reveal any such regulation by alternative splicing. This result supports our finding that the central kinase domain alone is sufficient to prime exocytosis. But this does not exclude the possibility that the splicing may be critical for secretion from intact cells.
Our results support the hypothesis that PIP2 generation is the critical step in the priming of exocytosis. The generated PIP2 may serve as the signal for neurotransmitter release by interacting with the potential Ca 2ϩ sensors for exocytosis (such as synaptotagmin (9 -12, 26) and Ca 2ϩ -activating proteins (5,27)). The localization of generation of PIP2, however, remains to be determined. The original hypothesis was that PIP2 is generated at the vesicles where major phosphatidylinositol phosphate 4-kinase activity is observed (15). However, recent experiments using the GFP-PH (pleckstrin homology) domain, which specifically binds to PIP2, suggest that PIP2 is located mainly at the plasma membrane (25,28). The hypothesis that PIP2 is generated at the plasma membrane is supported by the revelation that transfected PIPKI attaches to the plasma membrane (29). Our active recombinant PIPKIs will be a useful tool in future investigations of the localization of PIP2 generation in permeabilized PC12 cells.  7. PIP kinase activity of PIPKI␤ is dramatically lower than that of PIPKI␣ and PIPKI␥. Autoradiography of thin layer chromatography from the PIP kinase assay demonstrates the differences in PIP kinase activity among PIPKI isoforms. Each recombinant protein tested was 0.4 g, in a 50-l total reaction. GST was used as a negative control.