Cloning and characterization of a wortmannin-sensitive human phosphatidylinositol 4-kinase.

Phosphatidylinositol (PtdIns) 4-kinases catalyze the synthesis of PtdIns-4-P, the immediate precursor of PtdIns-4,5-P2. Here we report the cloning of a novel, ubiquitously expressed PtdIns 4-kinase (PI4Kβ). The 2.4-kilobase pair cDNA encodes a putative translation product of 801 amino acids which shows greatest homology to the yeast PIK1 gene. The recombinant protein exhibits lipid kinase activity when expressed in Escherichia coli, and specific antibodies recognize a 110-kDa PtdIns 4-kinase in cell lysates. The biochemical properties of PI4Kβ are characteristic of a type III enzyme. Interestingly, both recombinant PI4Kβ and the endogenous protein are inhibited by 150 nM wortmannin, suggesting that we have cloned the previously described PtdIns 4-kinase that is responsible for regulating the synthesis of agonist-sensitive pools of polyphosphoinositides (Nakanishi, S., Catt, J. K., and Balla, T. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 5317-5321).

The metabolism of phosphoinositides has long been acknowledged to play a central role in the transduction of signals triggered by a variety of growth factors and hormones. Both the enzymes and their product phosphoinositides are present in virtually all eukaryotic organisms and tissues that have been studied. Over the past several years the complexity of phosphoinositide metabolism has become better appreciated. In the classically defined phosphatidylinositol (PtdIns) 1 turnover pathway, sequential phosphorylation of the 4 and 5 positions yields PtdIns-4-P and PtdIns-4,5-P 2 , the latter of which acts as a substrate for phospholipase C producing inositol 1,4,5trisphosphate, a stimulator of intracellular Ca 2ϩ release (2), and diacylglycerol, a stimulator of certain protein kinase C isoforms (3). More recently PtdIns-4-P and PtdIns-4,5-P 2 have been shown to regulate cytoskeletal rearrangement through the association with a variety of actin binding proteins (4,5). PtdIns-4,5-P 2 has also been shown to stimulate both phospholipase D (6,7) and ␤-adrenergic receptor kinase (8). Finally, all of these lipids are substrates of PtdIns 3-kinase, yielding an array of 3-phosphorylated products (9). It is now clear that the synthesis of a variety of polyphosphoinositides from the start-ing substrate PtdIns is catalyzed by at least three types of PtdIns kinases (10,11).
PtdIns 3-kinase (a type I enzyme) catalyzes the phosphorylation of PtdIns at the D3 position of the inositol ring. This enzyme was initially identified through its association with viral oncoproteins and a number of growth factor receptors (12). More recently several additional classes of PtdIns 3-kinases have been identified including a G protein-activated enzyme (13) and VPS 34p, a protein involved in protein trafficking in yeast (14).
PtdIns 4-kinases catalyze the phosphorylation of PtdIns at the D4 position of the inositol ring and have been divided into two types (II and III) based on their size and sensitivity to various compounds (11). The type II enzymes were initially characterized as membrane-associated 55-kDa proteins whose lipid kinase activity is highly stimulated by detergent and inhibited by both adenosine and the monoclonal antibody 4C5G (11,15). The type III enzymes are membrane-associated proteins predicted to be Ͼ200 kDa in size that are less stimulated by detergent and are not inhibited by adenosine or 4C5G antibodies. The PtdIns 4-kinases are highly abundant and have been identified in a large number of membrane structures (reviewed Ref. 16).
Recently several PtdIns 4-kinases have been cloned and found to be homologous to PtdIns 3-kinases. They all contain both a lipid kinase unique domain and a C-terminal catalytic domain with distant homology to protein kinases. In yeast, the PIK1 gene encodes a 125-kDa protein that is indispensable for cell growth and plays a role in cytokinesis (17). It contains the lipid kinase unique domain at its far N terminus and the catalytic domain in the characteristic C-terminal position. Although it is intermediate in size, its biochemical properties suggest that it is more similar to the type III enzyme (18). In Dictyostelium discoideum, a putative PtdIns 4-kinase has recently been cloned, whose domain structure is similar to PIK1, extending the identification of these proteins across several species (19). A second yeast gene, STT4, encodes a 200-kDa protein that is dispensable for growth in the presence of osmotic stabilizers and has been implicated in the protein kinase C pathway through its isolation in a screen for mutants sensitive to the protein kinase C inhibitor staurosporine (20). Finally, the first PtdIns 4-kinase from higher eukaryotes, PI4K␣, was cloned and shown to encode a 100-kDa protein with significant homology to STT4 and biochemical properties of a type II enzyme (21). This protein, as well as STT4, contains adjacent lipid kinase unique and catalytic domains at its C terminus. An alternative splice of the PI4K␣ gene that generates a 230-kDa protein has also been recently reported (22).
These three types of PtdIns kinases all show homology to an ever expanding family of protein kinases whose substrates have not yet been identified. This family includes the TOR/ FRAP proteins that are the cellular targets of the FK506-binding protein-rapamycin complex and are involved in cellular signaling and cell cycle control (23)(24)(25)(26)(27). It is interesting to note that although yeast TOR2 and mammalian FRAP/RAFT1 have associated PtdIns 4-kinase activities, these activities are probably not endogenous to the protein kinase catalytic site (27). Other members of this extended family include the ATM/ MEC1/DNA-PK proteins that are involved in both cell cycle progression and checkpoint control and chromosomal maintenance and repair (28 -30). All these proteins share a conserved C-terminal catalytic domain found in both lipid and protein kinases.
Within this conserved domain are specific amino acid stretches that distinguish the subfamily of PtdIns 4-kinases from PtdIns 3-kinases and the other family members. We have taken advantage of this subfamily specificity to design degenerate PCR primers for the use in cloning novel PtdIns 4-kinases. We have identified and cloned one such gene and analyzed the biochemical properties of the encoded protein, which we call PI4K␤. Interestingly, PI4K␤ is wortmannin-sensitive and shows great similarity to a recently described wortmannininhibitable PtdIns 4-kinase that was partially purified from bovine adrenal cortex (1). Nakanishi et al. (1) demonstrate that this enzyme is responsible for regulating the hormone-sensitive pools of inositol phospholipids. Recent studies in which the effects of 100 nM to 1 M wortmannin have been used to implicate phosphatidylinositol 3-kinase in membrane trafficking, cytoskeletal rearrangement, and signal transduction must be reconsidered in view of the nearly ubiquitous expression of wortmannin-sensitive PI4K␤.

EXPERIMENTAL PROCEDURES
Materials-Human placenta and heart cDNA libraries and the TA cloning kit were purchased from Clontech. Taq polymerase was purchased from Perkin-Elmer. Expand PCR kit was purchased from Boehringer Mannheim. PtdIns was purchased from Avanti, [␥-32 P]ATP from DuPont NEN, silica plates from Merck, and wortmannin was purchased from Sigma. Random prime labeling kit was purchased from Pharmacia Biotech Inc.
Cloning and Sequencing-A 32-fold degenerate primer (GGIGA(T/ C)GA(T/C)TG(T/C)(C/A)GICA(A/G)GA), corresponding to the sense orientation of the conserved sequence GDD(C/L)RQ(D/E), as well as a 64-fold degenerated primer (AT(A/G)TTICC(A/G)TT(A/G)TGIC(G/T)(A/ G)TCT/CTT) corresponding to the antisense orientation of the conserved sequence KDRHNGNI were used in PCR reactions containing ϳ1 ϫ 10 7 plaque-forming units of a human placenta cDNA library in GT10. Reaction conditions were 30 cycles of 94°C for 1 min, 55°C for 1 min, 72°C for 1 min and then a 10-min extension at 72°C. The 312-bp product was digested with SmaI to eliminate PI4K␣ clones from the population of PCR products, reamplified as indicated above, and redigested with SmaI. Individual clones were sequenced following subcloning into a TA cloning vector (Clontech) using M13 forward and reverse primers.
The fragment, corresponding to a novel putative PtdIns 4-kinase, was random prime-labeled with [␣-32 P]CTP and used to screen a human heart cDNA library in GT10, under standard procedures (32). ϳ5 ϫ 10 5 plaques were screened, and 8 positive clones were obtained. Each was subcloned into Bluescript pKS-(Strategene) at the EcoRI site, under standard procedures (32). The longest clone (3.2) was 1.5 kb, contained 1.2 kb of coding sequence, and no in-frame stop at the 5Ј end. To extend the sequence, a second overlapping clone (13.1) was digested with EcoRI/PstI, and a 350-bp fragment corresponding to the 5Ј-most end of the sequences obtained was labeled and used to rescreen the library as described above, resulting in one additional clone that extended the coding sequence in the 5Ј direction by 700 bp.
To obtain the remainder of the full-length cDNA, 5Ј-RACE PCR was performed using human placenta cDNA supplied by Clontech in their 5Ј-RACE PCR kit, under the manufacturer's suggested conditions. PCR was performed using adapter primer 1 (Clontech) and an antisense primer recognizing nucleotides 820 -840 in PI4K␤ in reactions containing the Expand PCR Enzyme mix (Boehringer Mannheim). Individual clones were sequenced following subcloning into TA cloning vector as described above. The consensus of five independent clones confirms the sequence of the 5Ј end of PI4K␤.
Northern Analysis-A multiple human tissue blot (Clontech) was probed with the randomly primed 312-nucleotide PCR product from the original placenta cDNA library, as per manufacturer's instructions.
Bacterial Expression and Antibody Production-A GST-fusion protein was generated by PCR using oligonucleotide primers recognizing amino acids (aa) 410 -414 (STRSV) in the sense orientation and aa 534 -538 (PYGHL) in the antisense orientation, both tailed with appropriate restriction enzyme recognition sites for subcloning into pGEX4T2 (Pharmacia). Recombinant clones were screened by SDS-polyacrylamide gel electrophoresis of Escherichia coli protein lysates after isopropyl-1-thio-␤-galactopyranoside induction, and the fusion protein (GST4K␤5Ј) was purified using glutathione-agarose affinity chromatography using standard procedures (33). The purified 45-kDa GST fusion protein was injected into rabbits, and antiserum was collected using standard procedures (Charles River PharmServices). Affinity purified antibodies were prepared by first incubating 5 ml of crude serum (diluted 1:10 in 10 mm Tris, pH 7.5), with 500 l of a 2 mg/ml GST affinity column, for 2 h at 4°C. Unbound antibody was then chromatographed over a 2 mg/ml GST4K␤5Ј affinity column, and specifically bound antibodies were eluted under both acidic and basic conditions as described elsewhere (34). To prepare GST-cleared blotting antibodies, a 1:10,000 dilution of crude serum in TBST was blotted against a membrane containing 40 g of GST, and the supernatant was collected. The removal of GST antibodies was confirmed by Western blotting the supernatant against unrelated GST fusion proteins as described below.
The N-terminally deleted 4K␤L and the full-length 4K␤ cDNAs were prepared by fusing three DNA fragments together as follows. First the backbone was prepared by digesting clone 3.2 in pKSϪ with StuI/ HindIII. Second, clone 8.1 was PCR-amplified using a 5Ј sense primer containing a HindIII site and a 3Ј antisense primer containing a StuI site, digested, and ligated into the 3.2 vector prepared above, yielding pKS4K␤60. pKS4K␤60 was then fused to 5Ј-RACE products to yield 4K␤L and 4K␤ as follows. 5Ј-RACE product p3 was amplified using Clontech sense adapter primer (AP1) and an antisense primer recognizing aa 274 -280, yielding a 1.5-kb fragment. pKS4K␤60 was PCRamplified using a sense primer recognizing aa 239 -245 and a T7 antisense primer, yielding a 2-kb fragment. To generate 4K␤L, these two fragments were PCR-amplified using a sense primer recognizing aa 83-87 and an antisense primer recognizing aa 797-801 both containing appropriate restriction site for subcloning. To generate 4K␤, these two fragments were PCR-amplified using a sense primer recognizing aa 1-5 and the same antisense primer. The 2.3-kb (PI4K␤L) and 2.5-kb (PI4K␤) amplified products were digested and subcloned into pGEX4T-2. Recombinant clones were screened by SDS-polyacrylamide gel electrophoresis, and the fusion proteins were purified as described above. To obtain reasonable amounts of active proteins, we used lower concentrations of isopropyl-1-thio-␤-galactopyranoside (0.1 mM) and overnight induction at 25°C.
Western blots were performed under standard procedures (Promega) using TBSTϩ milk (10 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20, 5% nonfat dry milk) for blocking and TBST for antibody incubations and washes. Signal was detected using horseradish peroxidase-coupled secondary antibodies and chemiluminescence as described by the manufacturer (DuPont NEN).

Cloning of PI4K␤ cDNA and Relationship to the PtdIns 4-Ki-
nase Family-To isolate novel PtdIns 4-kinases, DNA from a human placenta cDNA library was used as a template in PCRs primed with degenerate oligonucleotides derived from two regions highly conserved among PI4K␣, PIK1, and STT4. The 312-bp product generated by PCR was SmaI-digested to eliminate PI4K␣ products from the population, and the reamplified PCR product was subcloned and sequenced to reveal a novel DNA fragment with homology to the family of PtdIns 4-kinases. This fragment was random prime-labeled and used for Northern blot analysis of human tissues (see below). Based on tissue distribution, it was used to screen a human heart cDNA library. A number of overlapping clones were isolated, and the longest clone (3.2) was further analyzed. The 1.5-kb insert contained a 1.2-kb open reading frame, a 3Ј stop, and 0.3-kb of 3Ј-untranslated region. Since no 5Ј stop was detected, the library was rescreened with a probe from the 5Ј end of clone 3.2 yielding clone 8.1 which contained 750 bp of additional open reading frame and still no 5Ј stop codon. To obtain the remainder of the open reading frame 5Ј-RACE was performed on human placenta cDNA, yielding a number of overlapping clones, five of which were analyzed and shown to be identical over Ͼ80% of their length. The clones differed slightly at the 5Ј end but all contained an identical open reading frame contiguous with that of clone 8.1 and all had identical stop codons in all three reading frames 5Ј of a potential initiating methionine.
The 2.4-kb full-length cDNA predicts a protein of 801 aa with a predicted size of ϳ90-kDa, starting from an initiation codon with a favorable Kozak consensus sequence for translation initiation (Fig. 1) (37). A second potential initiating methionine would result in a protein 104 amino acids shorter of predicted size ϳ80 kDa. Both initiating methionines are followed by glycines suggesting myristoylation of the N terminus (Fig. 1). The predicted protein, PI4K␤, contains an N-terminal domain (lipid kinase unique domain) ( Fig. 2A) that is shared by members of both the PtdIns 3-and PtdIns 4-kinase families (22). Additionally, a C-terminal catalytic domain (Fig. 2B) defines this protein as a member of a much larger family of protein/ lipid kinases that includes the PtdIns 3-and PtdIns 4-kinases, the TOR proteins, ATM, DNA-PK, MEC1/RAD3, and MEI41 whose members are involved in such diverse functions as mitogenic signaling, cell cycle regulation, and DNA repair (reviewed in Ref. 38). Interestingly, the functionally related PtdIns 4P5-kinase family appears to share no significant sequence homology in either the lipid kinase unique domain or the catalytic domain (39,40).
PI4K␤ shares most significant sequence homology with yeast PIK1 (42% identity in the catalytic domain and 17% in the lipid kinase unique domain) and with the newly described D. discoideum gene DdPIK4 (45% in the catalytic domain and 19% in the lipid kinase unique domain). This is consistent with the conserved domain structure among these three proteins.
Tissue Distribution-Northern blot analysis was performed on multiple human tissues and revealed a single ϳ4-kb message in a variety of tissues (Fig. 3). Although PI4K␤ is ubiquitously expressed, the precise distribution is distinct from that of the other human PtdIns 4-kinase, PI4K␣ (21), suggesting nonredundant functions for these two enzymes. The highest level of expression was detected in heart, pancreas, and skeletal muscle.
Lipid Kinase Activity of Recombinant and Endogenous PI4K␤-To confirm that PI4K␤ encodes an active PtdIns 4-kinase, several GST fusion constructs were generated, and recombinant protein was expressed in E. coli. One such construct, lacking the N-terminal 82 aa of PI4K␤ (GST4K␤L) was expressed, purified, and assayed for lipid kinase activity in reactions containing PtdIns as a substrate (Fig. 4A). In contrast to the control (C) GST fusion protein that lacked the catalytic domain, GST4K␤L generated significant amounts of PtdIns-P at both concentrations tested. The full-length PI4K␤ gave identical results (data not shown) confirming that the N-terminal nonconserved portion of PI4K␤ was not required for lipid kinase activity. Since E. coli lacks endogenous PtdIns kinases, these results confirm that PI4K␤ encodes a PtdIns kinase.
To identify the lipid products generated, we performed HPLC analysis on the deacylated products of the PtdIns kinase assay. A single peak, precisely comigrating with [ 3 H]glycerophosphorylinositol 4-phosphate standard, was observed (Fig.  4B), supporting the classification of PI4K␤ as a PtdIns 4-kinase. We observed only modest inhibition by 500 M adenosine and the type II-specific inhibitory monoclonal antibody 4C5G (15), suggesting that PI4K␤ is not a type II enzyme (data not shown). To further explore the enzymatic properties of PI4K␤ we next examined its inhibition by wortmannin, a fungal metabolite that inhibits PtdIns 3-kinase at nanomolar concentrations (41,42). Although the PI4K␣ and PIK1 enzymes were resistant to micromolar concentrations of this drug, a partially purified PtdIns 4-kinase from bovine adrenal cortex was shown to be inhibited by 100 nM wortmannin (1). We assayed GST4K␤ activity in the presence of various concentrations of wortmannin and observed concentration-dependent inhibition with an IC 50 of ϳ120 nM (Fig. 4C). This suggests that PI4K␤ may be the same enzyme as was previously shown to be wortmanninsensitive (1).
This wortmannin-sensitive enzyme was identified in several cell types including the human Jurkat T-cell line. We therefore generated PI4K␤-specific antibodies to investigate the properties of endogenous PI4K␤ immunoprecipitated from Jurkat cells. Antibodies were raised against a GST fusion of a 100-aa partial clone of PI4K␤. To demonstrate the specificity of these antibodies for PI4K␤, the GST-cleared polyclonal serum was used to blot protein lysates from wild type DH5␣ cells or DH5␣ cells expressing GST4K␤L (Fig. 5A). A signal corresponding to GST4K␤L and several breakdown products were detected only in lysates from transformed bacteria (lane 2). Identical results were obtained when protein lysates were blotted with the crude PI4K␤ antibodies (data not shown) suggesting that the observed signal was generated by PI4K␤-specific antibodies and not by the GST antibodies also present in this crude serum.
Using either preimmune or affinity purified immune serum, we immunoprecipitated PI4K␤ from detergent-solubilized cell lysates of Jurkat cells. The proteins were separated by SDS- polyacrylamide gel electrophoresis and analyzed by Western blot using the PI4K␤ antibody. A single protein migrating at ϳ110 kDa was observed only in precipitations using immune serum (Fig. 5B). This protein has a mobility slightly slower than that predicted by the 801-aa PI4K␤. It is likely that post-translational modification, such as myristoylation (see FIG. 4. PtdIns-kinase activity of recombinant PI4K␤. A, the control (C) GST fusion protein (500 ng) or the N-terminally deleted GST4K␤L (ϳ10 ng/l slurry) were bound to glutathione beads and assayed for PtdIns-P production. The extracted lipid products were analyzed by TLC as described under "Experimental Procedures." Numbers above the lanes indicate the volume of bead slurry added. The control (C) GST fusion protein GST4KB5Ј contains only 100 aa of PI4K␤ and does not include the catalytic domain (see "Experimental Procedures"). B, HPLC analysis of the 32 P-lipids. GST4K␤ was assayed for PtdIns-P production as in A. Lipid products were purified and deacylated as under "Experimental Procedures." The migration of 3 H standards PtdIns-4-P and PtdIns-4,5-P 2 as well as the migration of PtdIns-3-P are indicated. C, wortmannin sensitivity of GST4K␤. Immobilized GST4K␤ was assayed for PtdIns production in the presence of the indicated concentrations of wortmannin. Lipid products were extracted and analyzed as in A. Results are presented as a percent inhibition of the activity in the absence of drug. as well as the protein A beads from immunoprecipitations with either preimmune serum (Pre) or affinity purified immune serum (Im) were immunoblotted with crude PI4K␤ antiserum. C, immunoprecipitates of B were assayed for lipid kinase activity in the presence of PtdIns. Phospholipids were extracted and analyzed by TLC as described in Fig. 4. D, wortmannin sensitivity of PI4K␤. Jurkat lysates were immunoprecipitated with crude PI4K␤ antiserum, and PtdIns kinase assays were performed in the presence of the indicated concentrations of Wortmannin. Lipid products were extracted and analyzed by TLC as above.
Results are presented as a percent inhibition of the activity in the absence of drug. Fig. 1), accounts for the decreased mobility. Aliquots of these same immunoprecipitates were also assayed for lipid kinase activity in the presence of PtdIns (Fig. 5C). Whereas very low levels of PtdIns-P could be detected in assays containing preimmune serum (Pre), a significant amount of PtdIns-P (10 -90fold higher than preimmune) was routinely produced in assays containing affinity purified antiserum (Im) or crude immune serum (data not shown). When similar assays were performed using either PtdIns-4-P or PtdIns-4,5-P 2 as substrates, no phosphorylated products were generated (data not shown). As expected, HPLC analysis of the lipid products confirmed the immunoprecipitation of a PtdIns 4-kinase (data not shown). Additionally, PI4K␤ was unable to phosphorylate PtdIns-3-P (data not shown) suggesting that it is distinct from the previously characterized PtdIns-3-P 4-kinase (43,44). PI4K␤ lipid kinase activity was only modestly affected by non-ionic detergents, adenosine and 4C5G (data not shown), but was strongly inhibited by wortmannin (Fig. 5D), with an IC 50 of 140 nM. Taken together, these results strongly suggest that the 110-kDa PtdIns kinase immunoprecipitated from Jurkat cell lysates is PI4K␤. DISCUSSION We have identified and characterized PI4K␤, a novel PtdIns 4-kinase that is widely expressed in a variety of tissues. The cDNA encodes an 801-aa protein that exhibits lipid kinase activity when expressed in E. coli. Both the bacterially expressed and the endogenous proteins exhibit properties consistent with the characterization of PI4K␤ as a type III enzyme. Antibodies raised against PI4K␤ detect a ϳ110-kDa protein in a number of cell types across several species. Interestingly, PI4K␤ is the first cloned PtdIns 4-kinase that is inhibitable by wortmannin, potentially implicating PI4K␤ in a variety of wortmannin-sensitive cellular pathways.
Sequence analysis of PI4K␤ places it within the PtdIns 4-kinase family and more generally places it in the larger family of lipid/protein kinases. It contains a conserved C-terminal catalytic domain with distant homology to protein kinases as well as strong homology to the dual specificity kinases such as PtdIns 3-kinase. Within this conserved domain is lysine 549 which, based on homology to PtdIns 3-kinase, is the likely site of wortmannin reactivity (45). All members of this lipid/protein kinase family contain this conserved lysine, yet many, including PIK1 and PI4K␣, are not inhibited by micromolar concentrations of the drug, suggesting that additional residues within the active site confer wortmannin sensitivity.
Members of this extended family have diverse cellular functions. For example, the yeast protein MEC1 and its Drosophila homologue MEI41 are checkpoint control genes which appear to monitor the state of the genome at the G 1 /S and G 2 /M transitions (28,46,47). Another family member, DNA-PK, was originally identified as a DNA-dependent protein kinase (48) but was subsequently shown to function in immunoglobulin gene rearrangement and DNA repair (29).
PI4K␤ has properties similar to the wortmannin-sensitive PtdIns 4-kinase described by Nakanishi et al. (1). The partially purified enzyme was inhibited by wortmannin with an IC 50 of ϳ50 nM, not dissimilar to the 120 -140 nM IC 50 observed for PI4K␤. The wortmannin-sensitive enzyme had an apparent molecular mass of 125 kDa, as judged by gel filtration, in agreement with the 110-kDa molecular mass observed for PI4K␤. The reported enzymatic properties of this protein are also very similar to those of PI4K␤. It is likely that we have cloned the PtdIns 4-kinase that regulates the formation of agonist-sensitive inositol phospholipids that are required for intracellular signaling in some cells.
It should be noted that a PtdIns 4-kinase from the particu-late fraction of Schizosaccharomyces pombe has been observed to be sensitive to the wortmannin analogue demethoxyviridin (49). Curiously, this enzyme was not inhibited by wortmannin. Additionally, attempts to isolate a drug-sensitive PtdIns 4-kinase from rat brain particulate fractions were unsuccessful (49). Although we have detected PI4K␤ in rat brain, both our experiments and those of Nakanishi et al. (1) suggest that it is only loosely associated with the membrane. Furthermore, PI4K␤ is not the major PtdIns 4-kinase present in membrane fractions, and therefore lipid kinase assays on these crude fractions would not be expected to show wortmannin sensitivity. Taken together, these data suggest that we need to reevaluate the interpretation of experiments employing wortmannin as an inhibitor in biological assays. For example, recent experiments have demonstrated that wortmannin inhibits the proper targeting of the lysosomal enzyme procathepsin D in a variety of cell types (50,51). The concentration of wortmannin used was as high as 1 M with an estimated IC 50 of ϳ100 nM. Clearly, these elevated levels of wortmannin could be inhibiting PI4K␤ thereby implicating it in protein trafficking. Furthermore, both wortmannin and demethoxyviridin have been reported to inhibit phospholipase D, PtdIns-phospholipase C, and phospholipase A 2 in vivo (41,52). It is likely that this inhibition is a downstream effect of the inhibition of PtdIns 3-kinase and possibly PI4K␤ in these cells, as little direct inhibition of these enzymes was observed in vitro at M concentrations of wortmannin. The assumption that PtdIns 3-kinase is a critical mediator of all the myriad pathways inhibited by wortmannin is likely to be an oversimplification.