Phosphatidylinositol-4-phosphate 5-Kinase Isozymes Catalyze the Synthesis of 3-Phosphate-containing Phosphatidylinositol Signaling Molecules*

Phosphatidylinositol-4-phosphate 5-kinases (PIP5Ks) utilize phosphatidylinositols containing D-3-position phosphates as substrates to form phosphatidylinositol 3,4-bisphosphate. In addition, type I PIP5Ks phosphorylate phosphatidylinositol 3,4-bisphosphate to phosphatidylinositol 3,4,5-trisphosphate, while type II kinases have less activity toward this substrate. Remarkably, these kinases can convert phosphatidylinositol 3-phosphate to phosphatidylinositol 3,4,5-trisphosphate in a concerted reaction. Kinase activities toward the 3-position phosphoinositides are comparable with those seen with phosphatidylinositol 4-phosphate as the substrate. Therefore, the PIP5Ks can synthesize phosphatidylinositol 4,5-bisphosphate and two 3-phosphate-containing polyphosphoinositides. These unexpected activities position the PIP5Ks as potential participants in the generation of all polyphosphoinositide signaling molecules.


and 4).
A pathway for the formation of D-3-phosphatidylinositols, proposed based on kinetic studies of intact human neutrophils, is through phosphorylation of the D-3 position of the myoinositol ring of phosphatidylinositol 4,5-bisphosphate (PtdIns 4,5-P 2 ) by a PtdIns 4,5-P 2 3-kinase and subsequent dephosphorylation of PtdIns 3,4,5-P 3 to produce PtdIns 3,4-P 2 (5). This pathway has been supported by the existence of the extensively characterized PtdIns 3-kinase enzyme family, which can catalyze in vitro phosphorylation of phosphatidylinositol 4-phosphate (PtdIns 4-P) and PtdIns 4,5-P 2 . Evidence for a different pathway for the formation of D-3 phosphatidylinositols has been found in human platelets, NIH 3T3 cells, and plants in which phosphorylation of the D-3-position of PtdIns to form PtdIns 3-P is followed by phosphorylation of the D-4-position to give PtdIns 3,4-P 2 and then of the D-5-position to form PtdIns 3,4,5-P 3 (6 -10). The importance of these various routes of synthesis has been disputed. Indeed, until now, enzymes that catalyze the direct phosphorylation of PtdIns 3-P and PtdIns 3,4-P 2 have not been clearly identified.
Several phosphatidylinositol-4-phosphate 5-kinases (PIP5Ks) have been discovered and characterized as enzymes synthesizing PtdIns 4,5-P 2 (reviewed in Ref. 11). The best characterized isoforms are the type I and type II kinases that differ both biochemically and immunologically (12)(13)(14). The sequence of a type II PIP5K (PIP5KII␣) has established that the PIP5K enzymes belong to a novel family (15). More recently, cDNAs encoding type I PIP5K (PIP5KI) and an additional PIP5KII isoform have been isolated (16 -19). The translated sequences of PIP5KI and PIP5KII enzymes have only 35% amino acid identity in their kinase homology domains, further establishing the distinctiveness of these two subfamilies (16). The PIP5KI␤ cDNA is identical to a product of the gene reported to be mutated in Friedreich's ataxia, a common hereditary autosomal recessive disease (16,20,21). All four recombinant type I and II PIP5Ks have PtdIns 4-P 5-kinase activity (15-17, 19, 21).
None of the PIP5K isozymes have been examined for alternative substrates except phosphatidylinositol for which there was no detectable activity (12)(13)(14). We report here that the type I and II PIP5K isozymes also utilize the 3-phosphate-containing phosphatidylinositides, forming PtdIns 3,4-P 2 and PtdIns 3,4,5-P 3 . This supports the existence of an additional pathway for the synthesis of 3-phosphate-containing phosphatidylinositol polyphosphates.

EXPERIMENTAL PROCEDURES
Materials-PtdIns and PtdIns 4-P were from Boehringer Mannheim or Sigma. PtdIns 3-P and PtdIns 3,4-P 2 dipalmitoyl esters were synthesized (22) according to Chen and Prestwich, 2 and Thum et al. (24), respectively. PtdIns 3-P and PtdIns 3,4-P 2 dipalmitoyl esters were also from Matreya. [ (15,16,19). These proteins were stored in 50 mM Tris, pH 7.5, 1 mM EGTA, 150 mM NaCl, 0.01% sodium azide, and 20% glycerol for the type I isozymes or 50 mM Tris, pH 8.0, and 20% glycerol for the type II isozymes. Polymerase chain reactionbased cloning of the recombinant PIP5KI␤ used in these experiments resulted in two changes from the published sequence: glutamine 300 to arginine and serine 421 to proline. Both changes lie outside of the conserved kinase homology domain (16). The activity of PIP5KII␤ is less stable; thus, larger amounts were used to achieve comparable activity in the kinase assays.
Phosphatidylinositol Phosphate Kinase Activity Assay-PIP5K isozymes were assayed in 50-l reactions containing 50 mM Tris, pH 7.6, 10 mM MgCl 2 , 0.5 mM EGTA, 2-100 M substrate prepared in isotonic KCl solution, and 50 M ATP (4 Ci/nmol). The reactions, at 37 or 22°C, ranged from 5 to 40 min. Immunoprecipitate kinase activities were tested for 10.5 min at 22°C in 50 mM Tris, pH 7.5, 10 mM MgCl 2 , 1 mM EGTA, 50 M substrates prepared in Tris buffer and 50 M ATP (4 Ci/nmol). The reactions were stopped by the addition of 100 l of 1 N HCl, and the lipid products were extracted using 200 l of chloroform/ methanol (1:1). The organic phase was washed at least once in 80 l of methanol, 1 N HCl (1:1). The organic phase was then spotted on TLC plates and run as before (16). After autoradiography for 1-12 h, the spots on the TLC plates corresponding to products were scraped into vials and counted using Cerenkov radioactivity in a Beckman scintillation counter.
Expression of PIP5Ks in COS-7 Cells-PIP5KI␣ and PIP5KII␣ were N-terminally tagged with the FLAG epitope by subcloning their coding regions into pcDNA3-FLAG vectors provided by Dr. Jon Morrow (Yale University). The epitope is recognized by the anti-FLAG m2 monoclonal antibody (Eastman Kodak Co.). Plasmids were introduced into COS-7 cells by lipofectamine-mediated transfection following the manufactur-er's procedures (Life Technologies, Inc.). Transfection of pcDNA3-CAT (Invitrogen) was used as a control. Transfected cells were harvested 17 h after adding the DNA and lysed in 250 l of Nonidet P-40 lysis buffer.

RESULTS
The ability of four different PIP5K isozymes to phosphorylate PtdIns 3-P, PtdIns 3,4-P 2 , and PtdIns 4-P is shown in Fig.  1A. When analyzed by TLC, the PtdIns 3-P phosphorylation product migrated as a phosphatidylinositol bisphosphate (PtdInsP 2 ). The product migrated more slowly than PtdIns 4,5-P 2 , probably because the PtdIns 3-P substrate is a dipalmitoyl synthetic lipid (22). The two additional faint spots above PtdIns 3-P-derived PtdInsP 2 may be phosphorylatable minor contaminants arising during the chemical synthesis of PtdIns 3-P. The PIP5KII enzymes have similar or greater activity toward PtdIns 3-P in comparison with PtdIns 4-P, whereas the PIP5KI isozymes prefer PtdIns 4-P. In addition, the PIP5KI enzymes also have activity toward PtdIns 3,4-P 2 , producing a product that migrates as PtdInsP 3 (Fig. 1A, lanes 5 and 6). Remarkably, both PIP5KI isozymes and PIP5KII␣ produce products from PtdIns 3-P that migrate in the PtdInsP 3 position (Fig.  1A, lanes 1-3).
The initial rate of PtdIns 3-P phosphorylation using PIP5KII␣ was 4-fold greater than that of its previously identified substrate PtdIns 4-P (Fig. 1B). In addition, the phosphorylation of PtdIns 3-P was linear for a longer time interval compared with PtdIns 4-P (Fig. 1B). In a short reaction (5 min), comparison of PIP5KII␣ kinase activity toward PtdIns 3-P and PtdIns 4-P at various concentrations indicated that the activity toward both substrates was dependent on their concentrations (data not shown). The kinetic parameters for PIP5KII␣ phosphorylation of PtdIns 3-P and PtdIns 4-P are shown in Table I. The apparent K m value of PIP5KII␣ for PtdIns 3-P is 3-fold greater than PtdIns 4-P. However, the V m for PtdIns 3-P is also about 3-fold greater than that for PtdIns 4-P. As a result, the catalytic efficiency (V m /K m ) for these two substrates is the same. Little activity toward PtdIns 3,4-P 2 was detected with the type II PIP5Ks (Fig. 1A, lanes 3 and 4).
The kinetic parameters of PIP5KI isozymes with their different substrates are listed in Table I. The time dependence of PIP5KI␣ activity using PtdIns 3-P and PtdIns 4-P is shown in Fig. 1C. The substrate preferences of the type I isozymes are different from the type II isozyme in that the type I PIP5K (␣ and ␤) enzymes phosphorylate PtdIns 4-P with a much greater V m than PtdIns 3-P. However, the K m values of the PIP5KI enzymes using PtdIns 3-P are lower than that of PIP5KII␣.
The products of the PIP5K reactions using PtdIns 3-P and PtdIns 3,4-P 2 as substrates were identified by HPLC analysis (Fig. 2). It was anticipated that the PtdInsP 2 products would be PtdIns 3,5-P 2 , given the specificity of the PIP5K enzymes toward PtdIns 4-P. Surprisingly, HPLC analysis demonstrated that the deacylated product of all of the PIP5K isozymes using PtdIns 3-P as substrate co-chromatographed with GroPIns 3,4-P 2 , as illustrated for PIP5KII␣ in Fig. 2A. This indicated that these kinases synthesize PtdIns 3,4-P 2 . The HPLC analysis also revealed that the product of PIP5KI␣ and PIP5KI␤ activity toward PtdIns 3,4-P 2 was PtdIns 3,4,5-P 3 . This is shown for PIP5KI␣ in Fig. 2B.
These data were obtained using recombinant, E. coli-expressed PIP5K isoforms, but similar results were observed using native PIP5Ks from mammalian cells. PIP5KII purified from erythrocytes had similar activity to the PIP5KII␣ presented above (data not shown). The ability of PIP5KI␣ to phosphorylate the 3-phosphate-containing lipids was validated by FIG. 3. Substrate comparison of native and transiently transfected PIP5Ks immunoprecipitated from COS-7 cells. A, anti-PIP5KI␣ antibodies or control depleted IgG were used to immunoprecipitate from COS-7 cell lysates. Shown is a Western blot with anti-PIP5KI␣ antibody (2 g/ml) of human red blood cell (hRBC) membranes, a COS-7 cell lysate, and the immunoprecipitates. B, the immunoprecipitates or recombinant PIP5KI␣ were assayed for activity toward 50 M PtdIns 3-P, PtdIns 3,4-P 2 , or PtdIns 4-P for 10.5 min. The substrates are numbered based on the positions of the inositol ring already phosphorylated, i.e. 3 for PtdIns 3-P. The positions of the products of the reaction are marked by arrows. The autoradiograph exposures (same TLC plate) were 7 h for the immunoprecipitates and 4 h for the recombinant enzyme. C, the epitope-tagged PIP5KI␣ and PIP5KII␣ were transfected into COS-7 cells and immunoprecipitated with the anti-FLAG m2 antibody. Transfection of untagged chloramphenicol acetyl transferase (CAT) served as a control for these experiments. The lysates of transfected cells and FLAG immunoprecipitates were transferred to an Immobilon-P membrane and Western blotted sequentially for PIP5KI␣ (2 g/ml) and PIP5KII␣ (10 g/ml). The epitopetagged kinases are slightly larger than the native enzymes. D, the anti-FLAG immunoprecipitates were assayed for kinase activity as before. A 5-h autoradiograph exposure is shown.  PtdIns 3,4, Reactions were carried out in 50 l, and the product concentration at the end of the reaction is listed. Reactions were for 10.5 min using 0.2 g of PIP5KI␣ or 8 g of PIP5KII␣. After TLC separation, the products were deacylated, and the glycerophosphorylinositol phosphate derivatives were determined by HPLC. immunoprecipitation of the kinase from COS-7 cells. When COS-7 cell lysates were Western blotted with anti-PIP5KI␣ antibody, a single 68-kDa protein was detected, which was immunoprecipitated with the same antibody (Fig. 3A). The PIP5KI␣ was not immunoprecipitated using an IgG depleted of PIP5KI␣ reactivity (Fig. 3A) or preimmune IgG (data not shown). The native PIP5KI␣ was able to phosphorylate both PtdIns 3-P and PtdIns 3,4-P 2 , and the activity toward the former was only 4-fold lower compared with PtdIns 4-P (Fig.  3B). In addition, the production of PtdIns 3,4,5-P 3 from PtdIns 3-P was also observed. The same pattern of phosphorylation of D3-phosphatidylinositols was observed using recombinant PIP5KI␣ and PIP5KII␣ expressed in COS-7 cells (Fig. 3, C and D). PIP5KI␣ and PIP5KII␣ containing a FLAG epitope at their N termini were transiently transfected into COS-7 cells, immunoprecipitated with an anti-FLAG monoclonal antibody, and assayed for activity using PtdIns 3-P, PtdIns 3,4-P 2 and PtdIns 4-P. No PIP5K immunoreactivity (Fig. 3C) or activity (data not shown) was immunoprecipitated from the control transfected cells. The only difference between the PIP5Ks expressed in E. coli and mammalian cells was that the native and recombinant PIP5KI␣ expressed in COS cells had quantitatively greater activity toward PtdIns 3-P. DISCUSSION Here we report that the PIP5K isozymes are kinases with dual substrate specificity. They can phosphorylate PtdIns 4-P on the adjacent D-5-position. They can also phosphorylate 3-phosphate-containing phosphatidylinositols including PtdIns 3-P and PtdIns 3,4-P 2 on the adjacent D-4-or D-5-positions. None of the other characterized phosphatidylinositol kinases are known to have this ability. The uniqueness of these enzymes is further emphasized by the lack of sequence homology with known phosphatidylinositol, inositol phosphate, or protein kinases (15,16).
Our data demonstrate that the PIP5Ks have the in vivo potential to synthesize the signaling molecules PtdIns 3,4-P 2 and PtdIns 3,4,5-P 3 , in addition to PtdIns 4,5-P 2 , as summarized in Fig. 4. This catalytic capacity was shown for both native and recombinant PIP5K isozymes. The production of both PtdIns 3,4-P 2 and PtdIns 3,4,5-P 3 when PtdIns 3-P is the initial substrate suggests that the product of the first reaction is retained on the enzyme and phosphorylated again in a concerted reaction. This result has important biological implications because it suggests an additional mechanism for PtdIns 3,4,5-P 3 synthesis within cells.
There is evidence in the literature that a platelet PtdIns 3-P 4-kinase activity is stimulated by the thrombin receptor and protein kinase C activation (7,8). PIP5KII␣ is present in platelets (18,31) and, as the only currently identified PtdIns 3-P 4-kinase in platelets, is a likely candidate for that PtdIns 3-P 4-kinase activity. Based on all of the available data, PIP5KII␣ has properties indistinguishable from the platelet PtdIns 3-P 4-kinase (7,8). PIP5KII␤ is activated by its association with the TNF receptor (19) and could generate PtdIns 3,4-P 2 involved in proliferation. The PIP5Ks could also be regulated by receptors that are known to stimulate PtdIns 3,4-P 2 and PtdIns 3,4,5-P 3 production upon agonist activation. As discussed previously, a pathway leading to synthesis of PtdIns 3,4-P 2 and PtdIns 3,4,5-P 3 has been proposed to occur by phosphorylation of PtdIns 4,5-P 2 by an agonist-activated 3-kinase (5). However, these arguments were based upon the observation that this PtdIns 3-kinase will phosphorylate all phosphoinositides lacking a phosphate in the D-3-position and that PtdIns 3,4,5-P 3 appears to be the initial product. With the data presented here, an alternative pathway in which PtdIns is phosphorylated by PtdIns 3-kinase and then phosphorylated by a PIP5K isoform is plausible, and the concerted reaction could explain why PtdIns 3,4,5-P 3 appears first.
The PIP5Ks have the potential to produce three signaling molecules: PtdIns 4,5-P 2 , PtdIns 3,4-P 2 , and PtdIns 3,4,5-P 3 . It will be very interesting to determine if the substrate preferences of these PIP5K isozymes are altered by receptor activation or different regulators such as the small G-proteins Rac and Rho (23,32,33). It is conceivable that modulation of these activities will be both spatially and temporally regulated, and thus the PIP5K enzymes could participate in a plethora of cellular events by generating multiple messengers. Given the expanded substrate repertoire of these enzymes, we propose that they be designated as phosphatidylinositol phosphate 4/5-kinases.