Autophosphorylation of Type I Phosphatidylinositol Phosphate Kinase Regulates Its Lipid Kinase Activity*

Phosphatidylinositol phosphate kinases (PIPKs) have important roles in the production of various phosphoinositides. For type I PIP5Ks (PIP5KI), a broad substrate specificity is known. They phosphorylate phosphatidylinositol 4-phosphate most effectively but also phosphorylate phosphatidylinositol (PI), phosphatidylinositol 3-phosphate, and phosphatidylinositol (3,4)-bisphos-phate (PI(3,4)P 2 ), resulting in the production of phos- phatidylinositol (4,5)-bisphosphate (PI(4,5)P 2 ), phos- phatidylinositol 3-phosphate, phosphatidylinositol (3,4)-bisphosphate (PI(3,4)P 2 ), phosphatidylinositol (3,5)- bisphosphate (PI(3,5)P 2 ), and phosphatidylinositol (3,4,5)-trisphosphate. We show here that PIP5KIs have also protein kinase activities. When each isozyme of a was immunoprecipitated from the lysate of overexpressing COS-7 cells. The immunoprecipitates were washed first with lysis buffer and then with alkaline phosphatase buffer (50 m M Tris-HCl (pH 8.2), 50 m M NaCl, 1 m M MgCl 2 , 1 m M dithiothreitol, 1 m M phenylmethylsulfonyl fluoride), after which 2 units of calf intestine alkaline phosphatase (CIAP) (Takara Shuzo Co., Ltd.) or storage buffer for CIAP (10 m M Tris-HCl (pH 8.0), 1 m M MgCl 2 , 50 m M KCl, 0.1 m M ZnCl 2 , 50% glycerol) was added. The reaction was carried out at 30 °C for 60 min. To evaluate the effect of CIAP treatment on lipid kinase activity of Myc-PIPKI a , the immunoprecipitates were washed five times with kinase buffer before lipid kinase assay.

The intracellular multifunctions of phosphoinositides (1)(2)(3) are regulated by a series of their metabolizing enzymes including lipases, kinases, and phosphatases (4 -7). Phosphatidylinositol kinase and phosphatidylinositol phosphate kinase (PIPK) 1 in particular are thought to be important for the spa-tiotemporal production of each phosphoinositide that directly controls a variety of functions.
The lipid kinases, which commit at the final step in the synthetic pathway of PI(4,5)P 2 , PIPKs have been identified and characterized. From their biochemical characteristics, they have been divided into two subtypes (type I and type II) (8). Primary sequences of type I and II PIPK revealed that these lipid kinases are conserved from yeast to mammal and form a family distinct from other lipid kinases (9 -11). To date, three isoforms for each PIPK subtype (PIPKI␣, -␤, and -␥ and PIP-KII␣, -␤, and -␥) in mammal, and two PIPK homologs in budding yeast (MSS4 and FAB1) have been identified (9 -16). A FAB1 homolog in mammal (p235 PIKfyve) has also been reported recently (17). In data bases, sequences that seem to belong to this lipid kinase family are found also in fission yeast, nematode, and fruit fly, as well as in higher plants. A comparison of the primary sequences of types I and II PIPK revealed that these two subtypes are not so closely related (28 -33%), whereas isoforms of the same subtype are highly homologous to each other (66 -78%). In addition, type II PIPK shows relatively low enzymatic activity for PIP isolated from natural phospholipids as a substrate. A recent study has succeeded in explaining these differences between types I and II PIPKs. Rameh et al. (18) showed that type II PIPK is a PI5P 4-kinase (PIP4K), and type I isoform is exactly a PI4P 5-kinase (PIP5K). They also proved the existence of PI5P in vivo, a novel phosphoinositide that had not been identified because of its trace amount in the cell and of its close elution time with PI4P in a separation system by high pressure liquid chromatography (18). Furthermore, a broad substrate specificity of PIPK in vitro has also been reported. Type II PIP4K is able to phosphorylate PI3P to produce PI(3,4)P 2 . Type I PIP5K produces both PI(3,4)P 2 and PI(3,5)P 2 from PI3P and PI(3,4,5)P 3 from PI(3,4)P 2 (18 -20). Moreover, type I PIP5K is also shown to phosphorylate PI at the D-5 position to produce PI5P (20), suggesting that type I PIP5K is a candidate for the kinase that produces PI5P in vivo.
It is already known that there are three classes of PI3K, each of which differs in substrate specificity (5). Class I PI3Ks has a broad substrate specificity and is able to phosphorylate PI, PI4P, and PI(4,5)P 2 at the D-3 position, whereas class II PI3Ks phosphorylate PI and PI4P but not PI(4,5)P 2 . Class III PI3Ks consisting of yeast VPS34p and its mammalian homolog phos-phorylate only PI and thus produce only PI3P. Furthermore, PI3Ks are known to have a Mn 2ϩ -dependent protein kinase activity. One class I PI3K, p110␣, phosphorylates p85␣ regulatory subunit at Ser-608 in the presence of Mn 2ϩ (21,22). In the case of p110␥ and -␦, autophosphorylation of the catalytic subunit occurs most predominantly (23,24). Upon treatment with wortmannin or a point mutation within p110 which diminishes its lipid kinase activity, the protein kinase activity is also lost, suggesting the latter activity is based on almost the same mechanism as the former (22)(23)(24). Upon autophosphorylation, lipid kinase activities of p110␣ and p110␦ were strongly suppressed, indicating a mechanism for down-regulation through phosphorylation (21,22,24). A class III PI3K, VPS34p, has also been shown to have a Mn 2ϩ -dependent protein kinase activity (25). This activity is also diminished in a lipid kinasenegative mutant. The lipid kinase activity of VPS34p is not affected by autophosphorylation (23,25).
Evidence has been presented for the protein phosphorylation of the PIPK family. In platelets, PIP4KII␣ has been shown to be phosphorylated, and a relationship between its phosphorylation state and lipid kinase activity has been suggested (26). It has also been shown that translocation of PIP4KII␣ to the cytoskeletal fraction of platelets in response to thrombin is inhibited by treatment with okadaic acid, suggesting that protein dephosphorylation is involved in this process (27). PIP4KII␥ has been shown to be phosphorylated in vivo, resulting in a shift of electrophoretic mobility (16). The level of phosphorylation is elevated in response to mitogenic stimulation such as that by serum, epidermal growth factor, or platelet-derived growth factor (16). Furthermore, type II PI4K and type I PIP5K activities are co-immunoprecipitated with protein kinase C in COS-7 cells (28). The association is dependent on the protein kinase activity of protein kinase C. These results show a close relationship between both subtypes of PIPK and the protein phosphorylation-dephosphorylation event and also suggest that protein phosphorylation controls the intracellular localization of PIPK as well as its lipid kinase activity.
Here, we show that PIP5KI has a protein kinase activity. PIP5KI isoforms expressed both in COS-7 cells and in Escherichia coli autophosphorylate in vitro. The autophosphorylation level is enhanced specifically in the presence of PI. Lipid kinase activity of PIP5KI is strongly suppressed after the autophosphorylation in the presence of PI. These results suggest that the enzymatic activity of PIP5KI is regulated by its intrinsic protein kinase activity.

EXPERIMENTAL PROCEDURES
Materials-PIP and PIP 2 were purified from bovine spinal cord as described (29) and were used as Ͼ99% pure PI4P and PI(4,5)P 2 , respectively. PA, PI, PC, and PS were purchased from Doosan Serdary Research Laboratories. PI3P was purchased from Matreya, Inc. Synthetic PI5P, PI(3,4)P 2 , and PI(3,4,5)P 3 were generously donated by Dr. Watanabe (Ehime University). [␥-32 P]ATP was purchased from NEN Life Science Products. The polyvinylidene difluoride membranes used for Western blot analysis were from Nihon Eido. Ni 2ϩ -nitrilotriacetic acidagarose was from Qiagen. The thin layer chromatography silica plates and cellulose plates were from Merck. Phosphoamino acid standards (Ser(P), Thr(P), and Tyr(P)) were from Sigma. Monoclonal anti-Myc antibody and anti-penta-His antibody were purchased from Santa Cruz Biotechnology and Qiagen, respectively.
Expression Vectors-His-tagged forms of mouse PIP5KI␤ and rat PIP4KII␣ were constructed by insertion of each cDNA into SalI-BamHI site of pQE32 and SalI-PstI site of pQE31 vector, respectively. GST fusion proteins of mouse PIP5KI␣ and PIP5KI␣(K138A) were constructed by insertion of each cDNA into XhoI-NotI site of pGEX4T-3 vector. Myc-tagged forms of mouse PIP5KI␣, -␤, -␥, and rat PIP4KII␣ were constructed by insertion into SalI-BamHI site of pCMV-Myc. Myc-PIP4KII␤ and -␥ were constructed as described (16). Expression of recombinant proteins in E. coli and purification in the native condition were as described previously (15). Transfection into COS-7 cells and immunoprecipitation of overexpressed proteins were also described previously (16).
In Vitro Kinase Reaction-Purified proteins expressed in E. coli or immunoprecipitated proteins from cell lysate were incubated in a reaction buffer (50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 0.1 mM ATP, and 1 Ci of [␥-32 P]ATP) at room temperature for 30 min. When the effect of the phospholipids on the protein kinase activity of PIPKs was studied, 50 M (final concentration) of each phospholipid was added to the reaction mixture. After the incubation, the reaction mixture was subjected to SDS-polyacrylamide gel electrophoresis, and the protein phosphorylation was observed by autoradiography or quantified by an image analyzer BAS2000 (Fuji).
In Vivo 32 P-Labeling of COS-7 Cells and Phosphoamino Acid Analysis-Myc-PIP5KIs were transfected into COS-7 cells and cultured for 24 h. Then the culture medium was changed to phosphate-free Dulbecco's modified Eagle's medium, and cells were cultured for 30 min. [ 32 P]Orthophosphate (0.2 mCi/ml) was then added, and the cells were labeled for 24 h. Labeled cells were lysed in lysis buffer (20 mM Hepes (pH 7.2), 50 mM NaCl, 30 mM sodium pyrophosphate, 1% Nonidet P-40, 1 mM EGTA, 25 mM NaF, 0.1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride), and Myc-PIP5KIs were immunoprecipitated with anti-Myc antibody and transferred to a polyvinylidene difluoride membrane. The bands corresponding to Myc-PIP5KIs were cut out and hydrolyzed in 6 N HCl for 1 h at 110°C. The resulting amino acids, together with standard phosphoamino acids, were spotted on TLC plates and separated by electrophoresis in pH 1.9 buffer (2.2% formic acid, 7.8% acetic acid) for the first dimension and pH 3.5 buffer (5% acetic acid, 0.5% pyridine) for the second dimension. The labeled phosphoamino acids were detected by autoradiography. The positions of the standard phosphoamino acids were detected by ninhydrin staining.
Lipid Kinase Assay-The lipid kinase reaction and detection of phosphorylated products were described previously (16). Briefly, PIPKs and substrate lipids (50 M) were incubated in a kinase buffer (50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 1 mM EGTA, 0.1 mM ATP, and 1 Ci of [␥-32 P]ATP) at room temperature. After the lipids had been extracted by addition of 1 N HCl and chloroform/methanol (2:1), phosphorylated lipids were separated by TLC and observed by autoradiography.
Alkaline Phosphatase Treatment-Myc-tagged form of PIPKI␣ was immunoprecipitated from the lysate of overexpressing COS-7 cells. The immunoprecipitates were washed first with lysis buffer and then with alkaline phosphatase buffer (50 mM Tris-HCl (pH 8.2), 50 mM NaCl, 1 mM MgCl 2 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), after which 2 units of calf intestine alkaline phosphatase (CIAP) (Takara Shuzo Co., Ltd.) or storage buffer for CIAP (10 mM Tris-HCl (pH 8.0), 1 mM MgCl 2 , 50 mM KCl, 0.1 mM ZnCl 2 , 50% glycerol) was added. The reaction was carried out at 30°C for 60 min. To evaluate the effect of CIAP treatment on lipid kinase activity of Myc-PIPKI␣, the immunoprecipitates were washed five times with kinase buffer before lipid kinase assay.

Type I PIP Kinases
Autophosphorylate in Vitro-To establish whether PIPK has any protein kinase activity, we carried out an in vitro kinase assay using each isoform of the PIPK members. Myc-tagged PIP5KI␣, -␤, and -␥ or PIP4KII␣, -␤, and -␥ were transfected in COS-7 cells, respectively, immunoprecipitated with anti-Myc antibody, and incubated with [␥-32 P]ATP in the presence of Mg 2ϩ . All of the PIPK isoforms tested were phosphorylated, whereas no other phosphorylated protein was detected (Fig. 1). Furthermore, this phosphorylation was also observed when His-tagged PIP5KI␣, -␤, and -␥ expressed in E. coli were used as enzyme sources ( Fig. 2A), showing that no other co-immunoprecipitated protein kinase is involved. From these results, we concluded that type I isoforms of PIPK possess a protein kinase activity and autophosphorylate by themselves. On the other hand, we did not detect any phosphorylation when His-tagged type II PIP4Ks (PIP4KII␣, -␤, and -␥) expressed in E. coli were used (data not shown). Thus we conclude that the phosphorylation of PIP4KIIs observed in Fig.  1 is due to some co-immunoprecipitated protein kinase present in amounts below the detectable limit. Thus, the autophosphorylating activity seems to be a unique characteristic of the type I PIPK isoform.
The Protein Kinase Activity of Type I PIP Kinase Is Enhanced Specifically by PI-As the lipid kinase activity of PIP5KI is known to be activated by PA (30), we next examined the possibility that the protein kinase activity of PIP5KI is also modified by any phospholipid. The His-tagged form of PIP5KI␤ was subjected to an in vitro kinase assay in the absence or presence of 50 M PA, PC, PI and PS, respectively. PI strongly stimulated the activity for autophosphorylation ( Fig. 2A), whereas the other phospholipids including PA had no effect. Furthermore, this activation of autophosphorylation is highly specific to PI as phosphatidylinositol phosphates such as PI3P, PI4P, PI5P (not shown), PI(3,4)P 2 , PI(4,5)P 2 , and PI(3,4,5)P 3 did not have such an effect (Fig. 2B). Finally, we found that the autophosphorylation was stimulated by PI in a dose-dependent manner, most effectively at 10 M PI (Fig. 2C). These results show that PIP5KI has a PI-dependent protein kinase activity, a quite unique dependence on phospholipid different from that of any other protein kinase. All Isoforms of PIP5KI Subtype Autophosphorylate in the Presence of PI-We next tried to determine whether the PI-dependent activation of autophosphorylation is a common characteristic of all PIP5KI isoforms. Myc-tagged forms of PIP5KI␣, -␤, and -␥ were immunoprecipitated with anti-Myc antibody and then subjected to an in vitro kinase reaction in the presence or absence of 50 M PI. As shown in Fig. 3, all type I isoforms were revealed to have PI dependence for their autophosphorylating activity. This result indicates that PI-dependent protein kinase activity is a specific characteristic common to all PIP5KI isoforms.
PI-dependent Autophosphorylation of Type I PIP Kinase Correlates with Its Lipid Kinase Activity-PI 3-kinases have also been reported to possess a protein kinase activity (21)(22)(23)(24). In these cases, substitutions of the amino acids essential for the lipid kinase activity diminish the protein kinase activity as well. To test whether the catalytic residue of PIP5KI is also involved in the protein kinase activity, a lipid kinase-negative mutant (K138A) of PIP5KI␣ was tested. Lys-138 in PIP5KI␣ is a conserved amino acid corresponding to the Lys that binds the ␣-phosphate of ATP in protein kinases such as cAMP-dependent protein kinase. A substitution of this residue with Ala results in complete loss of lipid kinase activity (14).
GST fusion protein of wild type and the mutant (K138A) PIP5KI␣ were expressed in E. coli and purified by glutathione-Sepharose and then subjected to in vitro kinase assay. GST-PIP5KI␣ (K138A) lost almost all its lipid kinase activity (both PI 5-and PIP 5-kinase) (Fig. 4A, and Ref. 14). At the same time, the lipid kinase-negative mutant also lost PI-dependent protein kinase activity (Fig. 4B). This indicates that the lipid kinase and the protein kinase activity of PIP5KI are based on the same structural mechanism for catalysis.
Cation Dependence of the PI-dependent Autophosphorylation of PIP5KI-We next studied the divalent cation dependence of the PI-dependent autophosphorylation of PIP5KI. An in vitro kinase reaction was started in the presence of various divalent cations. As shown in Fig. 5A, PIP5KI preferred Mg 2ϩ most, but it also utilized Mn 2ϩ for autophosphorylation (Fig. 5A). Under the same conditions, lipid kinase activities were also measured. The PIP 5-kinase activity of PIP5KI for PI(4,5)P 2 production was exclusively dependent on Mg 2ϩ , whereas the PI 5-kinase activity for PI5P production was dependent on both Mg 2ϩ and Mn 2ϩ (Fig. 5B). The same cation dependence of PI 5-kinase and PI-dependent autophosphorylation may indicate that both activities are based on a similar catalytic mechanism.
PI-dependent Autophosphorylation Suppresses Both PI5-and PIP5-kinase Activity of Type I PIP Kinase-We further investigated the effect of PI-dependent autophosphorylation on lipid kinase activity of PIP5KI. His-PIP5KI␤, which was immobilized on beads by an immunoprecipitation with anti-penta-His antibody, was subjected to in vitro kinase reaction with or without PI/PIP. After the reaction, the excess of ATP and PI/PIP was washed away, and lipid kinase activities (PI 5-and PIP 5-kinase activities) were measured. Fig. 6A shows that both lipid kinase activities were almost completely lost after autophosphorylation induced by PI. In contrast, the in vitro kinase reaction with ATP and PIP, which failed to enhance the autophosphorylation of PIP5KI, resulted in no change in the lipid kinase activities (Fig. 6, A and B). This indicates that the marked decrease in lipid kinase activity is not due to any denaturation of PIP5KI enzymatic activity during the lipid kinase reaction. Finally, the time course experiment showed that there is a negative relationship between the lipid kinase activity and the degree of PI-dependent autophosphorylation (Fig. 6C). These results suggest that the PI-dependent autophosphorylation strongly down-regulates the lipid kinase activities of PIP5KI.
Autophosphorylation Occurs on Serine and Threonine Residues in Vitro and in Vivo-To characterize further the protein kinase activity of PIP5KI, we carried out a phosphoamino acid analysis using Myc-PIP5KI␣. We examined whether PIP5KI is phosphorylated in vivo by a metabolic 32 P labeling of COS-7 cells that were transfected with Myc-PIP5KI␣. Myc-PIP5KI␣ was immunoprecipitated and subjected to SDS-polyacrylamide gel electrophoresis (Fig. 7A). The result showed that PIP5KI␣ is a phosphoprotein in vivo. Myc-PIP5KI␤ and I␥ were also revealed to be phosphorylated in vivo by the same experiment (not shown). We also found treatment with alkaline phosphatase restored the lipid kinase activity of PIP5KI␣ (Fig. 7B). This together with the data in Fig. 6 suggest that a portion of PIP5KI␣ is phosphorylated and down-regulated in vivo through autophosphorylation. Next, we cut out the phosphorylated band in Fig. 7A and performed a phosphoamino acid analysis for Myc-PIP5KI␣. Fig. 7C shows that the phosphorylation in vivo occurred mainly on serine residues. The same assay after PI-dependent autophosphorylation of Myc-PIP5KI␣ in vitro revealed that the phosphorylation occurred on serine and, to a lesser extent, on threonine residues but not on tyrosine residues (Fig. 7C). These results suggest that PIP5KI possesses a serine/threonine protein kinase activity in vitro, and its phosphorylation in vivo may be explained partly by the autophosphorylation of PIP5KI.

DISCUSSION
The PIP kinase family has been reported to have a broad substrate specificity in vitro. PIP5KI is able to phosphorylate the D-5 position on not only PI4P but also PI, PI3P, and PI(3,4)P 2 to produce PI5P, PI(3,4)P 2 , PI(3,5)P 2 , and PI(3,4,5)P 3 (19,20). The broad substrate specificity of PIP5KI is similar to that of class I PI 3-kinases that also phosphorylate the D-3 position on PI, PI4P, PI5P, and PI(4,5)P 2 . Moreover, PI 3-kinase, including the class I subfamily, has been reported to possess a Mn 2ϩ -dependent protein kinase activity (21)(22)(23)(24). Thus, we tried to elucidate whether PIP5KI also possesses protein kinase activity. As we have demonstrated in this study, type I PIP5K transfected to COS-7 cells or produced by the E. coli expression system was autophosphorylated in vitro, showing a protein kinase activity in addition to lipid kinase activity. These findings expand the paradigm of dual-specific kinases capable of phosphorylating both protein and lipid.
In addition to the case with PI 3-kinase and PIP kinase in this study, there are some reports showing that the dual specificity toward protein and lipid substrates could be applicable to phosphatases as well. PTEN/MMAC1 is a putative tumor suppressor gene product homologous to protein tyrosine phosphatases such as CDC14, PTP-IV1, and CPTPH (31). Interestingly, it has been reported that PTEN/MMAC1 dephosphorylates the D-3 position of PI(3,4,5)P 3 (32) as well as tyrosinephosphorylated protein. These results indicate a close evolutionary relationship between protein-and lipid-kinases/ phosphatases. However, there is no evidence that other lipid kinases, such as PI 4-kinase or diacylglycerol kinase, have protein kinase activity. Unlike PI 3-kinase or PTEN/MMAC1, the PIP kinases are dissimilar to any known protein kinase in primary structure. This still does not rule out the possibility that the PIP kinases are related to some protein kinase family members. Future work may reveal a close relationship between PIPK and other protein kinases in their tertiary structure.
Furthermore, we observed that the autophosphorylation of PIP5KI was stimulated strongly and specifically by PI. The stimulation by PI was highly specific, and other polyphosphoinositides such as PIP, PIP 2 , and PIP 3 did not have such an effect. Although the structural mechanism for PI-dependent autophosphorylation is unclear, the mechanism behind the down-regulation may be anticipated from the crystal structure of PIP4KII␤ reported by Hurley and co-workers (33). PIP4KII␤ forms a flattened surface for interaction with PI5P in the lipid bilayer, and certain positively charged amino acids seem to be involved in the interaction with the phosphate group of the substrate phospholipid. When PI-induced autophosphorylation occurs at serine/threonine adjacent to those positively charged residues, the interaction between PIP5KI and phosphoinositides is interrupted, down-regulating the lipid kinase reaction.
The protein kinase activity is only detected as an autophosphorylation of PIP5KI, and none of the protein substrates for this activity are currently unknown. By using His-PIP5KI␤, we did not observe any significant phosphorylation of protein substrate such as myelin basic protein or histone H1 (not shown). Recently, it was reported that the protein kinase activity of p110␥ PI3-kinase is involved in the activation of the mitogenactivated protein kinase pathway (34). This shows that the protein kinase activity of PI 3-kinase has roles for not only down-regulation of lipid kinase activity but also phosphorylation of any downstream target protein to transduce signals. Future work will answer the question about the existence of protein substrates for PIP5KI.
We have shown that PIP5KI is phosphorylated (Fig. 7A) and down-regulated (Fig. 7B) in vivo in the resting cells. This phosphorylation is possibly caused by endogenous PI. Therefore, some phosphatases may be activated in response to extracellular stimuli and then dephosphorylate PIP5KIs. Indeed, we FIG. 5. Cation dependence of the PIdependent autophosphorylation and lipid kinase activity of PIP5KI. A, PIdependent autophosphorylation of His-PIP5KI␤ in the absence (Ϫ) or presence of various divalent cations (Mg 2ϩ , Mn 2ϩ , Ca 2ϩ , and Zn 2ϩ ) at 5 mM. 50 M PI was added in all assays. B, His-PIP5KI␤ was subjected to PI 5-or PIP 5-kinase assay in the same condition as in A. The level of phosphorylation is indicated in arbitrary units. A typical result corresponding to three independent assays is shown.
FIG. 6. The effect of PI-dependent autophosphorylation on lipid kinase activity of PIP5KI␣. A, His-PIP5KI␤ was immobilized on beads by immunoprecipitation with anti-penta-His antibody and subjected to an in vitro kinase reaction in the presence of 50 M PI (or PIP) and 1 mM ATP. Components in the in vitro kinase reaction are indicated below each lane. After the reaction, beads were washed three times with kinase buffer (see "Experimental Procedures"), and then a PI-or PIP kinase assay was carried out (PI 5-kinase and PIP 5-kinase). Note that PIP5KIs produce PI(4,5)P 2 from PI in a concerted reaction (20). B, a quantitative representation of the PIP 5-kinase assay as in A. C, time course experiment for the negative correlation between PI-dependent autophosphorylation and PIP 5-kinase activity. The level of phosphorylation is indicated in arbitrary units.

FIG. 7. Phosphoamino acid analysis.
A, pCMV-Myc-PIP5KI␣ (I␣) or control vector (control) was transfected into COS-7 cells and labeled with [ 32 P]orthophosphate in vivo. Then Myc-PIP5KI␣ was immunoprecipitated and detected by immunoblotting (IB) with anti-Myc antibody or autoradiography. B, Myc-PIP5KI␣ was expressed in COS-7 cells and immunoprecipitated and then incubated with (؉) or without (Ϫ) calf intestine alkaline phosphatase (CIAP). After the treatment, the immunoprecipitates were washed extensively and subjected to PIP kinase assay. C, Myc-PIPKI␣ metabolically labeled as in A (in vivo) and Myc-PIP5KI␣ subjected to autophosphorylation in the presence of 50 M PI (in vitro) were digested with 6 N HCl and separated by two-dimensional TLC by electrophoresis in pH 1.9 and then pH 3.5 buffer. The position of standard phosphoamino acids (pS, phosphoserine; pT, phosphothreonine; pY, phosphotyrosine) is indicated by dotted lines.
found that PIP5KI is dephosphorylated in response to lysophosphatidic acid, a typical agonist that induces inositol phospholipid turnover. 2 Subsequently, PIP5KI activities are increased, resulting in the synthesis of PI(4,5)P 2 . This down-and up-regulatory mechanism possibly functions in vivo.
In summary, we found that type I PIPKs have protein kinase activities and autophosphorylate in a PI-dependent manner, and this phosphorylation down-regulates the lipid kinase activity of type I PIPK. These results show the general physiological mechanism by which lipid kinase is regulated through protein phosphorylation. At the same time, our results also show a possible regulation of type I PIPK activity that plays critical roles in inositol lipid-signaling systems.