A membrane-associated complex composed of the Vpsl5 protein kinase and the Vps34 phosphatidylinositol 3-kinase (Ptdlns 3-kinase) is essential for the delivery of proteins to the yeast vacuole. An active Vpsl5p is required for the recruitment of Vps34p to the membrane and subsequent stimulation of Vps34p Ptdlns 3-kinase activity. Consistent with this, mutations altering highly conserved residues in the lipid kinase domain of Vps34p lead to a dominant-negative phenotype resulting from titration of activating Vpsl5 proteins. In contrast, catalytically inactive Vpsl5p mutants do not produce a dominant mutant phenotype because they are unable to associate with Vps34p in a wild-type manner. These data indicate that an intact Vpsl5p protein kinase domain is necessary for the association with and activation of Vps34p, and they demonstrate that a functional Vps15p-Vps34p complex is absolutely required for the efficient delivery of proteins to the vacuole. Analysis of a temperatureconditional allele of VPS15, in which a shift to the nonpermissive temperature leads to a decrease in cellular Ptdlns(3)P levels, indicates that the loss of Vpsl5p function leads to a defect in activation of Vps34p. In addition, characterization of a temperaturesensitive allele of VPS34 demonstrates that inactivation of Vps34p leads to the immediate missorting of soluble vacuolar proteins (e.g., carboxypeptidase Y) without an apparent defect in the sorting of the vacuolar membrane protein alkaline phosphatase. This rapid block in vacuolar protein sorting appears to be the result of loss of Ptdlns 3-kinase activity since cellular Ptdlns(3)P levels decrease dramatically in vps34 temperature-sensitive mutant cells that have been incubated at the nonpermissive temperature. Finally, analysis of the defects in cellular Ptdlns(3)P levels in various vps15 and vps34 mutant strains has led to additional insights into the importance of PtdIns(3)P intraceUular localization, as well as the roles of Vpsl5p and Vps34p in vacuolar protein sorting. T HE accurate and efficient delivery of proteins to specific intracellular organelles is essential to establish and maintain the functional integrity of these compartments. Proteins destined for the mammalian lysosome or the yeast vacuole are transported through the early stages of the secretory pathway from the endoplasmic reticulum to the Golgi complex (Kornfeld and Mellman, 1989; Klionsky et al., 1990). In a late Golgi compartment, lysosomal/vacuolar proteins are sorted away from proteins headed to the cell surface in a process that requires a functional lysosomal/vacuolar targeting signal. In mammalian cells, lysosomal proteins that contain phosphomannosyl residues are recognized by mannose-6-phosphate receptors, which mediate delivery to the lysosome (Kornfeld and Mellman, 1989; Kornfeld, 1992). The delivery of proteins to the yeast vacuole does not involve modification of carbohydrate residues. Instead, the targeting signal is found within the amino acid sequence of vacuolar proteins (Johnson et al., 1987; Vails et al., 1987, 1990). The recent identification of a transmembrane sorting receptor for the vacuolar hydrolase carboxypeptidase Y (CPY) ~ indicates that, like lysosomal protein sorting, the delivery of proteins to the vacuole in yeast is a receptormediated process (Marcusson et al., 1994). Genetic selections in Saccharomyces cerevisiae have identified a large number of mutants that are specifically defective in vacuolar protein sorting (Bankaitis et al., 1986; Rothman and Stevens, 1986; Robinson et al., 1988; Rothman et al., 1989). Instead of delivering proteins to the vacuole, J. H. Stack and D. B. DeWald both contributed significantly to this work. Address correspondence to Dr. Scott D. Emr, Division of Cellular and Molecular Medicine, Howard Hughes Medical Institute, University of California at San Diego School of Medicine, La Jolla, CA 92093-0668. Tel.: (619) 534-6462. Fax: (619) 534-6414. The present address for Kaoru Takegawa is Department of Bioresouree Science, Faculty of Agriculture, Kagawa University, Kagawa 7614)7, Japan. I. Abbreviations used in this paper: ALP, alkaline phosphatase; CPY, carboxypeptidase Y; DSP, dithiobis(succinimidylpropionate); PI 3-kinase, phosphoinositide 3-kinase (uses Ptdlns, Ptdlns(4)P, and Ptdlns(4,5)P2 as substrates); Ptdlns, phosphatidylinositol; Ptdlns 3-kinase, phosphatidylinositol 3-kinase (uses Pldlns as a substrate); Ptdlns(3)P, phosphatidylinositol 3-phosphate; tsf, temperature sensitive for function; vps, vacuolar protein sorting. © The Rockefeller University Press, 0021-9525/95/04/321/14 $2.00 The Journal of Cell Biology, Volume 129, Number 2, April 1995 321-334 321 these vps (vacuolar protein sorting defective) mutants missort and secrete vacuolar proteins as their Golgi-modified precursors. Characterization of the products of the VPS genes has provided considerable insight into the molecules and mechanisms involved in the signal-mediated delivery of proteins to the vacuole. Analyses of the VPSl5 and VPS34 genes have indicated that they encode homologues of a serine/threonine protein kinase and a phosphatidylinositol 3-kinase (PtdIns 3-kinase), respectively, suggesting that protein and phospholipid phosphorylation events are required for vacuolar protein sorting (Herman and Emr, 1990; Herman et al., 1991a; Hiles et al., 1992). Mutations in the VPS15 gene that alter residues highly conserved among protein kinases result in functional inactivation of the Vpsl5 protein (Vpsl5p). These mutations eliminate VpslSp protein kinase activity, and the mutant strains missort multiple vacuolar proteins (Herman et al., 1991a, 1991b; Stack and Emr, 1994). In addition, truncation of 30 amino acids from the COOH terminus of VpslSp results in a temperature-conditional vacuolar protein sorting defect (Herman et al., 1991b). A shift to the nonpermissive temperature in vpslSAC30 cells causes an immediate but reversible defect in the sorting of soluble vacuolar proteins. The extremely rapid onset of the sorting defect in the vps15AC30 strain indicates that VpslSp is directly involved in the delivery of soluble proteins to the vacuole. The product of the VPS34 gene shares extensive sequence similarity with the pll0 catalytic subunit of mammalian phosphoinositide 3-kinase (PI 3-kinase; Herman and Emr, 1990; Hiles et al., 1992). In mammalian cells, PI 3-kinase phosphorylates membrane PtdIns and its more highly phosphorylated derivatives, PtdIns(4)P and PtdIns(4,5)P2, and the 3'-phosphorylated products have been postulated to serve as second messenger molecules important in regulating cell growth and proliferation (Auger et al., 1989; Cantley et al., 1991; Soltoff et al., 1992). S. cerevisiae has been shown to contain PtdIns 3-kinase activity, and strains deleted for the VPS34 gene are extremely defective for this activity (Auger et al., 1989; Schu et al., 1993). Alteration of conserved residues in the lipid kinase domain of Vps34p results in severe defects in both PtdIns 3-kinase activity and vacuolar protein sorting (Schu et al., 1993). Biochemical characterization of Vps34p has shown that, unlike mammalian pll0, it is only able to use PtdIns as a substrate, and it is inactive toward PtdIns(4)P and PtdIns(4,5)P2 (Stack and Emr, 1994). The substrate specificity and other biochemical properties of its PI 3-kinase activity suggest that Vps34p may be similar to a PtdIns-specific 3-kinase activity recently characterized from mammalian cells (Stack and Emr, 1994; Stephens et al., 1994). On the basis of the role for Vps34p in vacuolar protein sorting, we have proposed that the production of a specific phosphoinositide, PtdIns(3)P, is involved in regulating intracellular protein sorting reactions in eukaryotic cells (Stack and Emr, 1994). VpslSp and Vps34p have been shown by genetic and biochemical criteria to interact as a complex that is associated with the cytoplasmic face of an intracellular membrane fraction, most likely corresponding to a late Golgi compartment (Herman et al., 1991a; Stack et al., 1993). In addition to recruiting Vps34p to the membrane, VpslSp also serves to activate Vps34p since PtdIns 3-kinase activity is defective in vps15 mutant strains (Stack et aLl., 1993). Therefore, VpslSp and Vps34p appear to act within a membrane-associated complex to facilitate the delivery of proteins to the vacuole in yeast. In this study, we took a genetic approach to investigate the regulatory role of Vpsl5p in the activation of Vps34p. It was found that catalytically inactive forms of Vps34p will act in a dominant-negative manner by titrating VpslSp, leading to defects in vacuolar protein sorting and PtdIns 3-kinase activity. Analysis of kinase-defective VpslSp mutants has shown that VpslSp protein kinase activity is required for the association with and subsequent activation of Vps34p. In addition, we have generated a temperatureconditional allele of VPS34 which demonstrates that Vps34p Ptdins 3-kinase activity is directly involved in vacuolar protein sorting. Finally, analysis of the in vivo levels of PtdIns(3)P, the phospholipid product of Vps34p activity, has provided insight into the role for VpslSp in activation of Vps34p and on the functional significance of cellular PtdIns(3)P levels in vacuolar protein sorting. Materials and Methods Strains, Plasmids, Media and Yeast Genetic Methods S. cerevisiae strains used were SEY6210 (MATot leu2-3,112 ura3-52 his3A200 trpl-AgO1 lys2-801 suc2-A9; Robinson et al., 1988), PHY102 (SEY6210 vps34Al::TRPI; Herman and Emr, 1990), BHY10 (SEY6210 leu2-3,112::pBHYll[CPY-Inv LEU2] Horazdovsky et al., 1994), KTY214 (BHY10 vps34Al::HIS3), and PHYII2 (SEY6210 vpslS~d::HlS3; Herman et al., 1991a). Plasmids containing VPS15 and VPS34 point mutations were described previously (Herman et al., 1991a,b; Schu ct al., 1993). Nomenclature of the mutant alleles reflects the original and altered residues in the gene product, i.e., DI65R represents an alteration at amino acid number 165 that changes an aspartic acid to an axginine. Standard yeast (Sherman et al., 1979) and Escherichia coil (Miller, 1972) media were used and supplemented as needed. Standard yeast genetic methods were used throughout (Sherman et al., 1979). Yeast cells were transformed using the alkali cation treatment method (Ito et al., 1983) and transformants were selected on the appropriate synthetic glucose media. PCR Mutagenesis and Screening for vps34 ~s Allele The temperature-conditional allele of VPS34 was generated by random PCR mutagenesis (Muhlrad et al., 1992). A 3' portion of the VPS34 gene was synthesized in the presence of MnC12 and limiting dATP to decrease the fidelity of Taq polymerase (Perkin-Elmer Cetus Corp., Norwalk, CT). The oligonucleotide primers used in this reaction annealed 100 nucleotides upstream of the 5' SpeI site in VPS34 and 100 nucleotides downstream of the 3' SpeI site in VPS34, and they amplified a 750-bp fragment. Standard reaction conditions were used with the modifications of 0.1 mM MnCI2 and 50 mM dATP. An acceptor plasmid was constructed by digesting a low copy number plasmid (CEN URA3 ) containing the VPS34 gene with SpeI to create a deletion slightly smaller than the mutagenized DNA (Fig. 1). Equimolax amounts of the gel-purified acceptor plasmid and mutagenized DNA were cotransformed into KTY214 (Avps34 CPY-Inv). Transformants were selected on minimal yeast plates, replica plated to YP-fructose plates, and incubated at 26°C (permissive temperature) or 37°C (nonpermissive temperature). Screening for vps34 sf mutants was accomplished using an overlay assay to detect extracellular invertase enzymatic activity as the result of mislocalization of a CPY-invertase fusion protein (Paxavicini et al., 1992). Mutants that secreted the CPY-invertase fusion only at the nonpcrmissive temperature were selected. Plasmids containing candidate vps34 ~f alleles were isolated from the strain, and retransformed into PHY102 (Avps34) and rescreened by a pulse-chase experiment to assess CPY sorting at the permissive and nonpermissive temperatures. Cell Labeling and lmmunoprecipitation For analysis of CPY processing, whole yeast cells were labeled essentially as described (Herman and Emr, 1990). Cells were pulse-labeled with Expres35S-label (NEN Research Products, Boston, MA) for l0 min at 30°C, and they were chased for 30 min at 30°C by the addition of methioThe Journal of Cell Biology, Volume 129, 1995 322 nine and cysteine to 2 raM. The media contained bovine serum albumin (1 mg/ml) and ot2-macroglobulin (I0 t~g/rnl; Boehringer Mannheim Corp., Indianapolis, IN) to stabilize secreted proteins. After the chase, an equal volume of cold 2× stop buffer (2 M sorbitol, 50 mM Tris-HCl, pH 7.5, 40 mM NaF, 40 mM NAN3, and 20 mM DTT) was added, and the cultures were incubated on ice for 5 rain. Zymolyase-100T (Seikagaku Kogyo Co., Tokyo, Japan) was added to 20/~g/ml, and the cells were incubated at 300C for 25 rain. The culture was separated into intracellular and extracellular fractions by centrifugation at 4,000 g for 30 s, and the proteins were precipitated by the addition of TCA to a final concentration of 5 %. Immunoprecipitation of CPY and alkaline phosphatase (ALP) was as described previously (Herman and Emr, 1990), and samples were electrophoresed on 9% SDS-polyacrylamide gels. After electrophoresis, the gels were fixed in 40% methanol, 10% acetic acid, treated with 1.0 M sodium salicylate containing 1% glycerol, and were then dried and subjected to autoradiography. In Vivo Labeling and HPLC Analysis of Phosphoinositides For analysis of cellular phosphoinositides, yeast cells were grown for '~16 h at 260C in minimal media lacking inositol and including 5 t~Ci/mt [3H]myo-inositol (18.8 Ci/mmol; Amersham Corp., Arlington Heights, IL). For temperature-shift experiments, the labeled cells were centrifuged and resuspended in YPD media that had been prewarmed to the appropriate temperature. Samples were collected by rapid centrifugation of the cells and resuspended in 0.5 ml 1.0 M HCI. 1.0 ml of chloroform/methanol (1:1) was added, and the cells were lysed by vortexing vigorously in the presence of glass beads. The organic phase was dried down, and the labeled lipids were deacylated essentially as described (Serunian et al., 1991). The pellet was resuspendnd in 1.0 ml methylamine reagent (0.428 ml 25% methylamine, 0.457 ml methanol, and 0.114 rni n-butanol), and was incubated at 53°C for 50 rain. The deacylated lipids were dried down in a Speed-Vac (Savant Instruments, Inc., Farmingdale, NY) and lyophilized several times from water. The resulting pellet was resuspended in 0.3 ml of water and extracted with 0.3 ml of butanol/ether/ethyl formate (20:4:1) to remove the acyl groups. The aqueous phase was dried down and resuspended in 50/~1 of water. The resulting glycerophosphoinositols were separated by HPLC on a Beckman System Gold using a 25-cm Partisil 5 SAX column (Whatman Inc., Clifton, NJ). The column was developed with a gradient of (NI-I4)2/ '04, pH 3.8, generated as follows: 10 mM for 5 min, 10-125 mM over 40 rain, and 125 mM to 1.0 M over 10 rain; the flow rat¢was 1.0 rnl/min. The column was calibrated using 32P-labeled glycerophospboinositols generated in an in vitro PI 3-kinase assay. In addition, each sample was spiked with unlabeled AMP and ADP, and their elution was monitored with a UV absorbance detector to assess column performance. 0.3-ml fractions were collected and counted in a scintillation counter (LC6000IC; Beckman Instruments, Inc., Fullerton, CA) using Cytoscint (ICN Radioehemicals, Irvine, CA) scintillation fluid. Ptdlns 3-Kinase Assays Yeast spheroplasts w e r e resuspended in 0.1 M KCI, 15 mM Hepes, pH 7.5, 3 mM EGTA, and 10% glycerol at 15-20 OD6oo/ml, and they were vortexed in the presence of 0.25 nun glass beads and protease inhibitors. The iysates were centrifuged at 750 g for 5 rain at 4°C to generate a crude lysate. The lysate was frozen in a dry ice-ethanol bath and stored at 80"C until use. Approximately 0.05 OD60o equivalents (<4 ttg protein) were assayed for Ptdlns 3-kinase activity as described (Whitman et al., 1988; Schu et al., 1993). The 50 t~! reactions were performed in 20 mM Hepes, pH 7.5, 10 mM MgC12, 0.2 mg/rnl sonicated PtdIns, 60/~M ATE and 0.2 mCi/ml ~[32p]ATP. The reactions were incubated at 25°C for 5 rain, and they were terminated by the addition ofS0 t~l 1 M HCI. The lipids were extracted with 160/~1 chloroform/methanol (1:1), and the organic phase was dried down and stored at -80"C. Labeled samples dissolved in chloroform were spotted onto Silica gel 60 TLC plates (Merck Sharpe & Dohme, West Point, PA), and they were developed in a borate buffer system (Walsh et al., 1991). Labeled species were detected by autoradiography. Cross-linking of Yeast Cell Extracts Lrnmunopreeipitation and cross-linking of yeast extracts was as previously described (Stack et al., 1993). Yeast strains were grown to midlogarithmic phase, converted to spberoplasts, labeled with Expres35S-label for 30 rain at 30"C, and ehnsed for 1 h at 30"C by adding methionine and cysteine to 2 mM and yeast extract to 0.2%. Labeled spberoplnsts were resuspended in XL lysis buffer (1.2 M sorbitol, 0.1 M KH2PO4, pH 7.5, and 5 mM EDTA) at 10-20 OD6oo U/ml. Cells were lysed by the addition of 4 vol of H:O. All solutions contained the protease inhibitors antipain, leupeptin, chymostatin, pepstatin (all at 2 ~,g/ml), aprotinin (0.1 TIU/ml), phenylmethylsulfonyl fluoride (100/~g/ml), and ~2-macroglobulin (10/~g/ml). DSP [dithiobis(succinimidylpropionate); Pierce Chemical Co., Rockford, ILl cross-linker, dissolved in DMSO, was added to a final concentration of 200 #g/ml. Control samples without cross-linker received DMSO alone. The extracts were incubated at room temperature for 30 rain, after which the reaction was quenched by the addition of 1 M hydroxylamine to a final concentration of 20 raM. Proteins were precipitated by the addition of TCA to 5%. The TCA pellets were resuspended in urea-cracking buffer (50 mM Tris-HCI, pH 7.2, 6 M urea, and 1% SDS) without reducing agent and processed for immunoprecipitation using anti-Vpsl5p antisera. After the first immunoprecipitation, the cross-linked samples were solubilized in urea-cracking buffer with or without 2 % 2-mercaptoethanol, and they were reimmunoprecipitated with the appropriate antisera. Control experiments demonstrated that anti-Vpsl5p antisera was irreversibly denatured by incubation with urea-cracking buffer in the presence or absence of reducing agent (not shown). The final samples were solubilized in urea-cracking buffer containing 2% 2-mercaptoethanol, and they were electrophoresed on 8 %-polyacrylamide gels.