The Yeast Inositol Polyphosphate 5-Phosphatase Inp54p Localizes to the Endoplasmic Reticulum via a C-terminal Hydrophobic Anchoring Tail

The budding yeast Saccharomyces cerevisiae has four inositol polyphosphate 5-phosphatase (5-phosphatase) genes, INP51, INP52, INP53, andINP54, all of which hydrolyze phosphatidylinositol (4,5)-bisphosphate. INP54 encodes a protein of 44 kDa which consists of a 5-phosphatase domain and a C-terminal leucine-rich tail, but lacks the N-terminal SacI domain and proline-rich region found in the other three yeast 5-phosphatases. We report that Inp54p belongs to the family of tail-anchored proteins and is localized to the endoplasmic reticulum via a C-terminal hydrophobic tail. The hydrophobic tail comprises the last 13 amino acids of the protein and is sufficient to target green fluorescent protein to the endoplasmic reticulum. Protease protection assays demonstrated that the N terminus of Inp54p is oriented toward the cytoplasm of the cell, with the C terminus of the protein also exposed to the cytosol. Null mutation ofINP54 resulted in a 2-fold increase in secretion of a reporter protein, compared with wild-type yeast or cells deleted for any of the SacI domain-containing 5-phosphatases. We propose that Inp54p plays a role in regulating secretion, possibly by modulating the levels of phosphatidylinositol (4,5)-bisphosphate on the cytoplasmic surface of the endoplasmic reticulum membrane.

Phosphoinositides are ubiquitous membrane components of various intracellular compartments, which regulate many diverse cellular functions including membrane trafficking events, secretion, actin cytoskeletal organization, cellular proliferation, and inhibition of apoptosis (reviewed in Refs. [1][2][3][4]. Many of these functions are mediated by binding and recruiting signaling proteins which contain specific phosphoinositidebinding domains such as SH2 domains, pleckstrin homology domains, FYVE domains, C2 domains or polybasic domains, thereby localizing these effector proteins to specific membranes (reviewed in Refs. 3 and 4).
Inp54p is a PtdIns(4,5)P 2 5-phosphatase (17), as are all the yeast 5-phosphatases. Therefore it is not surprising that single null mutation of INP54 is not lethal (18). However, further characterization of the phenotype of this mutant has not been reported. In this study we demonstrate Inp54p is a C-terminal tail-anchored protein that localizes to the cytosolic face of the endoplasmic reticulum. This localization is mediated by a short 13-amino acid leucine-rich region at the extreme C terminus of the protein. Null mutation of INP54, but not any of the SacI domain-containing 5-phosphatases, results in increased levels of secretion from the endoplasmic reticulum. We propose the enzyme regulates PtdIns(4,5)P 2 levels on the cytoplasmic surface of the endoplasmic reticulum and thereby regulates secretion.

EXPERIMENTAL PROCEDURES
Materials-Restriction and DNA modifying enzymes were supplied by New England Biolabs, Fermentas, or Promega. Biomol Green TM Reagent for phosphate detection was obtained from Biomol. Oligonucleotides were obtained from Bresatec, Australia, and the Department of Microbiology, Monash University, Australia. The Big Dye Terminator Cycle Sequencing kit was from PE Applied Systems (Foster City, CA). All other reagents were from Sigma unless otherwise stated. The plasmid pPS1303 for GFP expression was a kind gift from Professor Pamela Silver, Dana Farber Cancer Institute, Harvard University, the pRS416 vector from Dr. Mark Prescott, Monash University, Australia, the plasmid pJJ250 was a gift from Dr. Doris Germain, Peter MacCallum Cancer Institute, Australia. Bovine pancreatic trypsin inhibitor (BPTI) expression plasmid pEB316U was a kind gift from Professor Dane Wittrup, MIT. Yeast strains used in the study are listed in Table I. Yeast strains were cultured at 30°C in standard YPD media or complete minimal media lacking specific amino acids to maintain selection of markers where appropriate.
Disruption of INP54 in S. cerevisiae-The genomic sequence containing the INP54 ORF (open reading frame), 1600 bp upstream of the start codon and 700 bp downstream of the stop codon was amplified by PCR (incorporating a XhoI site at the 5Ј end and a NotI site at the 3Ј end) and cloned into the XhoI/NotI site of Bluescript. The construct was digested with PstI/SpeI to release a 1.7-kilobase pair fragment containing the full sequence of INP54, which was replaced with a LEU2 expression cassette obtained by digesting the plasmid pJJ250 with XbaI/PstI (19). The LEU2 gene flanked by the sequence upstream and downstream of INP54 ORF was recovered from the Bluescript vector by XhoI/NotI digestion. The DNA fragment was then transformed into W303 diploid cells by electroporation as described previously (20). Transformants were selected by their ability to grow on complete minimal media lacking leucine, and were subsequently screened for homologous integration of the LEU2 marker by PCR. Null mutations of the SacI domain-containing 5-phosphatases has been described previously (21).
Cloning of INP54 -INP54 was amplified from SEY6210 genomic DNA by PCR using synthetic oligonucleotide primers as described in Table II (GenBank accession number Z74807). The primers amplified the complete coding region of the 5-phosphatase from nucleotides 1 to 1156 and incorporated an EcoRI site at the 5Ј end and a BamHI site at the 3Ј end. The PCR product was ligated into the pCRBlunt vector (Invitrogen), released by EcoRI restriction digest, and subcloned into the EcoRI site of the pTrcHisB vector to give the construct pTrcHisB-INP54. The identity of the PCR product was confirmed as INP54 by dideoxy sequence analysis. INP54 331 was amplified and cloned as above with the 3Ј oligonucleotide incorporating a stop codon plus an EcoRI site after nucleotide 1093 (Table II).
Expression and Purification of Recombinant (His) 6 -Inp54p-Two ϫ 100-ml cultures of pTrcHisB-INP54 were grown at 37°C to an A 600 of 0.5-0.6 prior to induction with 0.1 mM isopropylthio-␤-D-galactoside for 2 h at 23°C. Following induction, cells were pelleted and soluble proteins were extracted in 1/10 volume of buffer B (20 mM Tris, pH 8, 250 mM NaCl) supplemented with 12 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotonin, 2 g/ml leupeptin, 1 mM benzamidine, and 1% Triton X-100 at 4°C overnight with gentle agitation. Triton extracts were centrifuged at 15,000 ϫ g for 15 min then the 20-ml supernatant was incubated with 2.5 ml of Talon resin (CLON-TECH) with gentle agitation in a 50-ml tube for 4 h at 4°C. Following incubation, the resin was poured into a column and washed with 20 column volumes of buffer B. Bound proteins were eluted with 4 column volumes of buffer B at pH 6.5 supplemented with 100 mM imidazole and 700-l fractions were collected. 50 l of the starting material, flowthrough, and the eluted fractions were analyzed by 12% SDS-PAGE (22), and either visualized by Coomassie Brilliant Blue staining or transferred to polyvinylidene fluoride membranes according to standard protocols (23) and immunoblotted using monoclonal antibodies to the (His) 6 -tag (Silenus). Immunoblots were developed using enhanced chemiluminescence (ECL) reagent (PerkinElmer Life Sciences) according to the manufacturer's instructions. The protein concentration in Coomassie-stained gels was quantitated using densitometry by comparison with a standard amount of protein loaded on the gel.
PtdIns(4,5)P 2 5-Phosphatase Enzyme Assays-PtdIns(4,5)P 2 substrate was a mixture of 33.3 M PtdIns(4,5)P 2 and 3 g of phosphatidylserine, dried under nitrogen, resuspended in 50 l of lipid resuspension buffer (20 mM Hepes, pH 7.5, 1 mM MgCl 2 , 1 mM EGTA), and sonicated for 5 min. The recombinant Inp54p was incubated with the substrate in the presence of kinase buffer (20 mM Hepes, pH 7.4, 1 mM EGTA, 5 mM MgCl 2 ) in a 100-l total reaction volume. Assays were performed at 37°C for 30 min using two linear protein concentrations in duplicate. The reaction was terminated by incubating with 1 ml of Biomol Green TM reagent at room temperature for 30 min. The absorbance at 620 nm was measured and the amount of phosphate released calculated by comparison with known standards supplied in the Biomol kit.
Expression of GFP-tagged Inp54p in S. cerevisiae-All INP54 constructs were amplified by PCR using appropriate primers listed in Table II, from SEY6210 genomic DNA. The primers incorporated a BamHI site for cloning INP54 into the pPS1303 vector, in-frame with the C-terminal GFP. The PCR product was ligated into pCRBlunt, released by BamHI digestion, and ligated to the compatible BglII site in pPS1303. GFP-tagged LRD13, 12, 11, and 10 were amplified from pPS1303-INP54 using oligonucleotides listed in Table II and a 3Ј primer specific for the 3Ј end of the GFP ORF. The resulting LRD-GFP products were cloned into pCRBlunt, excised using BamHI and cloned into the BglII site of pPS1303H. pPS1303H lacked the GFP ORF and was constructed by excising GFP from pPS1303 via HindIII digest followed by religation of the vector. The INP54-pPS1303 constructs were transformed into an inp54 null mutant strain using the S. cerevisiae Easy-Comp Transformation kit (Invitrogen) and the transformants were selected on complete minimal media agar plates lacking uracil. The identity of all constructs were confirmed using dideoxy sequencing analysis.
Expression of Inp54-GFP under Its Native Promoter in S. cerevisiae-Full-length INP54 was amplified together with 968 bp upstream of the open reading frame using oligonucleotides listed in Table II. The PCR product was cloned into the BamHI site of pRS416-GFP (21) in-frame with the C-terminal GFP. The construct was then transformed into an inp54 null mutant strain using the S. cerevisiae EasyComp Transformation kit (Invitrogen) and the resulting transformants selected on complete minimal media lacking uracil. Single colonies were grown to mid-log phase in minimal media lacking uracil at 30°C, fixed, and stained with an anti-GFP antibody (CLONTECH) to amplify the signal, and analyzed using confocal microscopy as described above.
Expression of HA-tagged Inp54p in S. cerevisiae-The HA-tagged Inp54p was cloned into the pPS1303H expression vector lacking the GFP coding region. The primers used to amplify INP54 incorporated a HA tag at either the 5Ј or 3Ј end, with a BamHI site (see Table II), this facilitated cloning into the BglII site of pPS1303H vector. The resulting constructs were transformed into inp54 null mutants as described above.
Analysis of GFP and HA-tagged Inp54p in S. cerevisiae by Confocal Microscopy-Yeast cells were grown overnight in complete minimal media ϩ 2% glucose, diluted 1/200 in 2% raffinose, and induced the next day in 2% galactose for 4 h. Cells were fixed according to Franzusoff et al. (24) and stained with either an anti-HA antibody (diluted 1/1000) to visualize the HA-tagged proteins or an anti-BiP antibody (diluted 1/25) to co-localize the GFP-tagged proteins with the ER. A polyclonal antibody to BiP (binding protein) was raised in rabbits against the S. cerevisiae endoplasmic reticulum protein Kar2p (BiP) by standard protocols using mature Kar2p recombinant protein as an antigen. The primary antibodies were counterstained with anti-rabbit tetramethylrhodamine B isothiocyanate for BiP (1/200), or anti-mouse tetramethylrhodamine B isothiocyanate (1/200) for HA. For nuclear staining, spheroplasts were pretreated with RNase (200 g/ml) at 37°C for 1 h, then stained with propidium iodide (2 g/ml) for 20 min. Cells were attached onto the microscope slides with poly-L-lysine (Sigma), coverslips were mounted on the slides using SlowFade (Molecular Probes). The cells were then analyzed using a Leica TCS-NT confocal microscope with an ArKr triple line laser, with green fluorescence collected in channel 1 (488 nm excitation, 530 Ϯ 30 nm emission) and red fluorescence in channel 2 (568 nm excitation, LPS90 nm).
Extraction of Recombinant GFP-and HA-Inp54p from Yeast-This extraction method is a slight modification of that as described in Seedorf and Silver (25). Yeast cells expressing Inp54p tagged with GFP or HA were harvested and resuspended in PBSM buffer (phosphate-buffered saline, 5 mM MgCl 2 ), 0.5 mM phenylmethylsulfonyl fluoride, and 3 g/ml each of leupeptin, aprotonin, pepstatin, and chymostatin. Glass bead lysis was performed by vortexing and incubating on ice at 30-s intervals for 12 cycles, respectively. The lysate was centrifuged for 10 min at 4°C, the supernatant represented cytosolic fraction, the pellet was resuspended in PBSM buffer with 1% Triton. After overnight Triton extraction at 4°C, the lysate was spun for 10 min to separate the Triton-soluble and Triton-insoluble fractions. These were separated on 10% SDS-PAGE and transferred to polyvinylidene fluoride membranes according to standard protocols (23), and immunoblotted using an antibody specific for GFP (P. Silver) or HA (Covance). To determine the 5-phosphatase activity of Inp54p tagged with GFP or HA, the fusion protein was immunoprecipitated from the Triton-soluble fraction using 50 l (50% v/v) of protein A-Sepharose, 0.8 g of anti-GFP antibody (Roche Molecular Biochemicals) or 6 g of anti-HA antibody (Covance), and 7 g of rabbit anti-mouse immunoglobulin (DAKO) which served as a linker. Immunoprecipitation was performed at 4°C overnight with gentle agitation. The protein A-Sepharose pellet was washed 6 times with ice-cold Tris saline (20 mM Tris, pH 7.2, 150 mM NaCl), and the pellet used in PtdIns(4,5)P 2 5-phosphatase enzyme assays as described previously.
Protease Protection Assays on Yeast Expressing Recombinant HA-or GFP-Inp54p-Microsomes were extracted from yeast by differential centrifugation fractionation according to Parlati et al. (26) and Paddon et al. (27). The medium-speed microsomal pellet, which was found to be enriched in ER markers, was treated with proteinase K according to the methods described by Bascom et al. (28). The control fractions (untreated) and proteinase K-treated fractions were separated by 10% SDS-PAGE, and immunoblotted with either an anti-HA antibody or anti-GFP antibody. The same fractions were immunoblotted with anti-BiP antibody as a control to ensure that the microsomal membranes were intact.
BPTI Secretion Assay-W303␣, inp51, inp52, and inp53 null mutant strains were transformed with YEplac181-GalBPTI plasmid containing a leucine nutritional marker. This was constructed by amplifying the Gal promoter and BPTI coding regions from pEB316U, using oligonu- cleotides listed in Table II and cloning the PCR product into the EcoRI/ HindIII site of YEplac181 vector (29). W303␣ and the inp54 mutant strain were transformed with a BPTI expression plasmid, pEB316U, containing a uracil nutritional marker. Yeast cells expressing BPTI were grown at 30°C overnight in complete minimal media lacking either leucine or uracil, supplemented with 2% glucose. The next day, cultures were diluted 1/200 in raffinose containing media, grown overnight until they reached 10 7 cells per ml, and induced with galactose for the specified time frames. The amount of BPTI secreted in the culture media was quantified according to the method described by Parekh et al. (30). A rescue experiment was done to investigate whether ⌬inp54 mutant could remove a sufficient amount of BPTI from the intracellular space and maintain viability when grown continuously on galactosesupplemented media. Since accumulation of high levels of BPTI is toxic, continuous induction would be lethal to the cells. Raffinose cultures of yeast cells were spotted in 10-fold serial dilutions in 5-l aliquots onto complete minimal agar plates supplemented with either 2% glucose or galactose.

Inp54p Sequence Predicts for a C-terminal Tail-anchor-
Inp54p is one of four inositol polyphosphate 5-phosphatases (5-phosphatases) found in the yeast S. cerevisiae. Unlike the other previously characterized yeast 5-phosphatases, Inp54p predicts for a smaller molecular mass, 44 kDa, comprising a central 300-amino acid 5-phosphatase domain and no other defined signaling motif. However, a hydropathic profile of Inp54p (scale ϭ Kyte-Doolittle ϫ Ϫ1) indicates that the region spanning amino acids 372-384 is strongly hydrophobic, and is highly rich in leucine residues (Fig. 1A). To identify other 5-phosphatases containing a similar C-terminal hydrophobic tail we performed tBLASTn and PSI-BLAST (31) searches of the nonredundant protein and nucleotide data bases at the NCBI. In all searches the filter excluding low compositional complexity regions was turned off. We were unable to identify any 5-phosphatase in any other organism that contained a similar hydrophobic anchor sequence in its C terminus. However, we noted that the C terminus of Inp54p strongly resembled that found in many C-terminal tail-anchored proteins (Fig.  1B). Tail-anchored proteins are a class of integral membrane proteins which lack any N-terminal targeting sequence and insert into membranes via a single C-terminal hydrophobic sequence (32)(33)(34)(35)(36). These proteins, which include cytochrome b 5 , UBC6, Bcl2, and numerous SNARE proteins have been identified in both mammalian cells and S. cerevisiae (32,37). A comparison of the C-terminal hydrophobic domain of Inp54p versus previously characterized tail-anchored proteins is shown in Fig. 1B. Tail-anchored proteins are predominantly localized on the membranes of the endoplasmic reticulum and/or mitochondria, with the bulk of the protein oriented toward the cytoplasm (33, 34, 36 -38).
Inp54p Localizes to the ER of the Cell-To characterize the intracellular localization of Inp54p, the protein was expressed as a fusion protein with GFP under the control of a galactoseinducible promoter in inp54 null mutant cells. Cells cultured in the presence of raffinose, which does not induce production of the fusion protein, demonstrated no fluoresence ( Fig. 2A). Incubation of cells in media supplemented with galactose resulted in expression of Inp54p-GFP fluorescence in the perinuclear region, as shown by co-localization with propidium iodide, which stains the nucleus (Fig. 2B). Co-localization using an antibody directed against the yeast protein Kar2p or BiP which localizes to the ER, demonstrated co-localization with Inp54p-GFP expression. Peripheral and reticular staining was also detected by both Inp54p-GFP fluorescence and the anti-BiP antibody. To confirm that GFP did not play any role in targeting the fusion protein to the ER, cells expressing the GFP alone were induced and analyzed. GFP was detected throughout the entire cell (Fig. 2B).
To ensure that the ER localization of Inp54p-GFP did not result from overexpression, which may cause proteins to distribute aberrantly, Inp54p-GFP was expressed under the control of its native promoter. The entire INP54 open reading frame and 968 bp upstream of the initiating ATG was amplified by PCR and cloned into a single copy yeast expression vector pRS416-GFP, which lacks a promoter. The construct was transformed into inp54 null mutant cells, which were grown to mid-log phase and analyzed by confocal microscopy. As the expression level was very low (results not shown) the signal was amplified using an antibody specific for GFP. Inp54p-GFP expressed under the Inp54p native promoter localized to a perinuclear ring-like structure, comparable to that observed in Inp54p overexpressing cells (Fig. 2C), indicating that the ER localization of Inp54p is not affected by the level of expression of the recombinant protein.
To further validate the intracellular distribution of Inp54p in yeast cells and demonstrate that the position of the tag did not influence the localization of the protein, a hemagglutinin (HA) tag was fused at either the N or C terminus of Inp54p. Inp54p was amplified by PCR as described under "Experimental Procedures," with the HA tag incorporated in either the 5Ј or 3Ј oligonucleotide primer, and cloned into the pPS1303H vector from which the GFP sequence had been removed. Following induction, cells were fixed and stained with an antibody specific for HA to visualize the fusion proteins. Fig. 2D shows that Inp54p tagged with HA was expressed in a perinuclear distribution consistent with ER localization, as observed with the GFP-tagged protein (Fig. 2B). In addition, these results demonstrate that the position of the tag at the N or C terminus did not affect the localization of the Inp54 protein, as has been shown previously for other ER tail-anchored proteins such as UBC6 (34) and cytochrome b 5 (33).
To confirm that Inp54p-GFP and HA-Inp54p were functional proteins, following induction in galactose-containing media, recombinant fusion proteins were extracted from yeast cells with Triton X-100, immunoprecipitated with antibodies specific for GFP or HA, and assayed for PtdIns(4,5)P 2 5-phosphatase activity as described under "Experimental Procedures." Immunoprecipitated GFP displayed no activity against PtdIns(4,5)P 2 . Both Inp54p-GFP and HA-Inp54p were able to hydrolyze PtdIns(4,5)P 2 confirming that the constructs encoded functional proteins (results not shown).

Inp54p Attaches to the Endoplasmic Reticulum Membrane with the Bulk of the Protein Oriented toward the Cytoplasm-
The C terminus of tail-anchored proteins inserts into the cytoplasmic surface of the endoplasmic reticulum or mitochondrial membrane with the N-terminal domain located in the cytoplasm (33, 37, 39 -41). Inp54p lacks a signal sequence suggesting the enzyme does not represent an integral endoplasmic reticulum protein, but post-translationally associates with this compartment. The membrane orientation of Inp54p in the endoplasmic reticulum was determined by a protease protection assay. Cells expressing HA-Inp54p, or Inp54p-GFP, were subjected to differential centrifugation to isolate the ER-enriched microsomal membrane fraction (26). Recombinant Inp54p association with the membrane fraction was sensitive to alkaline treatment using 0.1 M NaCO 3 , pH 11.5 (results not shown), therefore it is not an integral membrane protein. Inp54p associates tightly with the membrane, however, since treatment with 0.5 and 1 M NaCl failed to release recombinant Inp54p from membranes (results not shown). ER-enriched microsomes were incubated in the presence or absence of Proteinase K for 1 or 2 h at 4°C, followed by centrifugation to isolate the microsomal pellet. Proteins were separated by 10% SDS-PAGE and immunoblotted using an antibody specific for either HA or GFP. Following a 1-h incubation with Proteinase K, virtually all of the Inp54p was degraded (Fig. 3, A and B) indicating that the bulk of the protein was oriented toward the cytoplasm of the cell. To confirm the integrity of the microsomal membranes, the same fractions were blotted with an antibody directed against the ER lumenal protein BiP. Proteinase K treatment had no effect on the integrity of BiP indicating that the microsomal membranes were intact (Fig. 3, A and B). It was noteworthy that similar results were observed for Proteinase K treatment of Inp54p regardless of whether the tag was at the N terminus (HA) or the C terminus (GFP) of the protein suggesting that the C-terminal tail of Inp54p may loop back into the cytosol. Previous studies of the tail-anchored protein cytochrome b 5 have demonstrated that the addition of a C-terminal tag does not prevent the transmembrane tail from inserting correctly into the membrane (33).
The C-terminal Hydrophobic Domain of Inp54p Mediates Localization to the ER-To determine whether the C-terminal hydrophobic domain mediates ER localization a series of Cterminal Inp54p truncation mutants were constructed and expressed in inp54 null mutant yeast. Previous studies have demonstrated removal of the hydrophobic tail of tail-anchored proteins results in the relocalization of the truncated protein to the cytoplasm (42-47).
Mutant Inp54p 331 contained the 5-phosphatase domain from amino acids 1-331 and lacked the entire hydrophobic C-termi-nal 53 amino acids. The second mutant, Inp54p 353 , lacked amino acids 354 -384. These residues are predicted to have a transmembrane topology by a Kyte-Doolittle hydrophobicity plot analysis. Inp54p 371 lacks the leucine-rich area spanning the last 13 amino acids (residues 372-384), which is the minimal hydrophobic domain in the C terminus. All mutant sequences were amplified by PCR using appropriate primers as described under "Experimental Procedures," and cloned into the pPS1303 vector. Expression of the mutant fusion proteins was induced by incubating the cells in galactose-containing media for 4 h and cells were examined live by confocal microscopy. All three mutant proteins were expressed in the cytosol although Inp54p 371 -GFP still showed some ER staining, but to a much lesser degree than that observed in the full-length protein (Fig. 4A). This indicates that the C-terminal tail of Inp54p mediates ER localization and that the last 13 amino acids are required to maintain this localization.
To confirm the cytosolic localization of the mutant fusion proteins, yeast cells were fractionated into cytosol, Triton-soluble and Triton-insoluble fractions as outlined under "Experimental Procedures." Proteins were separated by SDS-PAGE and immunoblotted with an antibody specific for GFP (Fig. 4B). Full-length Inp54p 384 -GFP which localized to the ER, was present intact, as a 71-kDa protein consistent with the predicted molecular mass of the recombinant protein, predominantly in the Triton-soluble fraction, with only a very small amount in the Triton-insoluble fraction. In contrast, the three Inp54p truncation mutants were expressed predominantly in the cytosol. Occasionally, a small, variable proportion of the mutant protein was detected in the Triton-soluble and Triton-insoluble fractions. Minimal proteolysis of recombinant mutant proteins was detected. These studies collectively indicate that the Inp54p C-terminal domain is critical for mediating membrane association, specifically to the endoplasmic reticulum.
To confirm that the 5-phosphatase catalytic function is not regulated by the C-terminal hydrophobic domain, the activity of purified recombinant full-length Inp54p 384 versus mutant Inp54p 331 which lacks the entire C-terminal hydrophobic domain was determined. Recombinant wild-type versus C-terminal mutant (His) 6 -tagged Inp54p were expressed in Escherichia coli, and purified using Talon metal affinity chromatography and assayed for PtdIns(4,5)P 2 5-phosphatase activity. No significant difference in PtdIns(4,5)P 2 5-phosphatase activity was detected between full-length Inp54p 384 or mutant Inp54p 331 (Fig. 4C) confirming that the C-terminal leucine-rich domain does not play a role in regulating enzyme activity.
The Last 13 Amino Acids of Inp54p Constitute the ER Localizing Sequence-Previous studies have demonstrated that Cterminal hydrophobic domains are sufficient to direct heterologous proteins to specific subcellular compartments (34,38,48). To determine whether the last 13 amino acids of Inp54p (372-384) comprise the ER targeting region, a series of leucinerich domain (LRD) mutant constructs were generated, encoding the C-terminal 13, 12, 11, and 10 amino acids of Inp54p fused to GFP (designated LRD13, LRD12, LRD11, and LRD10, respectively). Following induction of the recombinant fusion proteins, expressing cells were analyzed by confocal microscopy. LRD13 was sufficient to localize GFP to the ER (Fig. 5). In contrast, LRD12, LRD11, and LRD10 fused to GFP predominantly demonstrated diffuse cytoplasmic staining, although minimal ER staining was still detected. This indicates that the last 13 amino acids of Inp54p constitute the minimum ER targeting motif, and while the shorter sequences contain some targeting information, it is not as efficient as LRD13.

Substitution of a Single Charged Amino Acid in the Tailanchor Domain Has No Effect on Localization-Several models
for the insertion of the C-terminal targeting domain of tailanchored proteins to the membrane have been proposed. It has been suggested that the absolute sequence of the tail is not important, rather, the amino acid composition in terms of hydrophobicity and length of the domain dictate the intracellular localization (34,49,50). Furthermore, several studies indicate that charged amino acids within the tail-anchor are the determinants of specific subcellular location to the endoplasmic reticulum versus mitochondria. Kuroda et al. (48) demonstrated that charged amino acids within the last 10 amino acids of two isoforms of cytochrome b 5 (outer mitochondrial membrane b 5 and ER b 5 ) determined their localization in the cell. Mutation of the positively charged amino acids (Arg 137 and Lys 144 ) to noncharged residues altered the localization of outer mitochondrial membrane b 5 from the mitochondria to the ER. Mutation of the negatively charged residue (Asp 134 ) to a neutral Ala or a positive Lys in ER b 5 redirected this protein from the ER to the mitochondria.
Examination of the hydrophobic tail of Inp54p, in particular the last 13 amino acids at the extreme C terminus, revealed a single charged residue (Lys) at position 382, and no other full-length or mutant Inp54p-GFP were fractionated into cytosol, Triton-soluble, and Triton-insoluble fractions as described under "Experimental Procedures." The fractions were separated by 10% SDS-PAGE and immunoblotted with an anti-GFP antibody. C, E. coli cultures harboring recombinant full-length (His) 6 -Inp54p 384 or mutant (His) 6 -Inp54p 331 were induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside for 2 h at 23°C. The induced cultures were extracted with 1% Triton X-100 overnight, and the soluble extract was purified by Talon resin immobilized metal affinity chromatography as described under "Experimental Procedures." 50 l of eluted fractions were separated by 10% SDS-PAGE and stained with Coomassie Brilliant Blue. Peak eluted fractions 6 and 7 for each recombinant wild-type or mutant Inp54p were assayed for PtdIns(4,5)P 2 5-phosphatase activity in triplicate.

FIG. 4. The C terminus of Inp54p is required for ER localization.
A, mutant constructs of INP54 with C-terminal deletions were generated as described under "Experimental Procedures." A schematic representation of each construct is shown on the left with the full-length Inp54p 384 included for comparison. The subscript on each construct refers to the last amino acid in the Inp54p mutant proteins. Expression of Inp54p-GFP C-terminal deletion mutants was induced by incubation with galactose for 4 h and following fixation the cells were analyzed by confocal microscopy. Bar indicates 5 m. B, yeast cells expressing charged amino acids. To investigate whether this residue plays a role in ER targeting, Lys 382 was mutated to a neutral alanine. An HA tag was fused to the N terminus of Inp54p(K382A) and the mutant recombinant 5-phosphatase expressed in the inp54 null mutant strain. As can be seen from Fig. 6, mutation of Lys 382 to Ala did not affect the ER localization of the protein.
The ER isoform of cytochrome b 5 has 2 charged residues at the C-terminal tail, an arginine and an aspartic acid. The negatively charged aspartic acid (Asp 134 ) mediates ER localization, whereas the positively charged arginine (Arg 128 ) demonstrated no role in localization (48). Since Lys 382 is the only charged residue in the tail of Inp54p, the "charged amino acids as determinant of intacellular localization" model cannot be applied in this case. A more detailed analysis of the 13-residue tail-anchor of Inp54p has to be performed to delineate the targeting information, and is the subject of ongoing studies.
Null Mutation of inp54 Results in Increased Secretory Capacity Compared with Wild-type Cells-In mammalian cells PtdIns(4,5)P 2 recruits ADP-ribosylation factor to the Golgi membrane where GTP-bound ADP-ribosylation factor engages coatomer proteins and stimulates phospholipase D, followed by vesicle budding (2). PtdIns(4)P and PtdIns(4,5)P 2 have been proposed to act as receptors for the coatomer complex COPII, which coats vesicles budding from the ER (51). PtdIns(4,5)P 2 is also required for fusion of secretory granules with the plasma membrane in exocytosis and clathrin-mediated endocytosis (4). To date, PtdIns(4,5)P 2 has not been shown to be directly involved in regulating vesicular transport from the ER in yeast. However, its precursor PtdIns(4)P is required for normal secretion in S. cerevisiae. Mutation of the PtdIns 4-kinase Pik1p results in impaired secretion from the Golgi and abnormal Golgi morphology (52,53).
Defects in endocytosis have been noted in the yeast SacI domain-containing 5-phosphatases, Inp51p, Inp52p, and Inp53p null mutants, however, secretion was reported to be normal (16). The functional role of Inp54p in yeast is as yet unclear and although null mutation of inp54 is not lethal, no other phenotype has been described. inp54 null mutant was generated as described under "Experimental Procedures" and the morphology of the null mutant strain compared with wildtype. Electron microscopy analysis of inp54 null mutant did not reveal any significant phenotype that differed from the wildtype (results not shown). The growth of inp54 null mutant was tested on media supplemented with either 0.9 M NaCl or 1.4 M sorbitol. The degree of viability on hyperosmotic media was the same for wild-type and null mutant strains (results not shown). Unlike double null mutants of the SacI domain-containing 5-phosphatases which have abnormal actin and chitin organization (16), inp54 null mutant demonstrated normal actin patches and chitin deposition as compared with the wild-type, assessed by staining cells with phalloidin or calcofluor, respectively, and analysis by confocal microscopy (results not shown).
Inp54p is a PtdIns(4,5)P 2 5-phosphatase which localizes on the cytosolic face of the ER. PtdIns(4,5)P 2 and PtdIns(4)P have been demonstrated to enhance binding of coatomer complexes to liposomes suggesting that these phosphoinositides are important in protein transport from the ER, possibly by promoting vesicle budding (51). Since proteins that are to be secreted have to be synthesized in the ER and transported further along the secretory pathway, Inp54p may also contribute indirectly to the regulation of secretion. Therefore, we investigated the role of Inp54p in secretion from the ER. We utilized a reporter protein assay which measures the amount of BPTI secreted from the cell (30,54). Wild-type yeast and cells with null mutation of inp51, inp52, inp53, or inp54 were transformed with an expression plasmid containing BPTI under the control of a galactose promoter. Cells were grown in the presence of galactose for 0, 16, 24, and 48 h at 30°C and the media containing secreted BPTI was collected. The level of BPTI present in the media was determined by measuring the inhibition of trypsin cleavage of a synthetic substrate L-BAPNA (N ␣ -benzoylarginine-p-nitroanilide), as previously described (30). Similar levels of BPTI were secreted into the media at all time points assayed for wild-type cells, the inp51, inp52, and inp53 single null mutant strains (Fig. 7). The results are consistent with studies by Singer-Kruger et al. (16) who showed no abnormalities in secretion in the inp51, inp52, and inp53 mutants. In contrast, deletion of INP54 resulted in a marked increase in secretion, ϳ2-3-fold higher after 48 h induction (Fig. 7), suggesting that Inp54p plays a role in the negative regulation of secretion.
Since accumulation of high levels of BPTI within the cell reduces its viability, we tested the viability of inp54 null mutant when BPTI was expressed continuously in the cell. A previous report has shown that the mammalian 180-kDa ribosome receptor that induces membrane proliferation in yeast could rescue cells expressing high amounts of BPTI by increasing secretion up to 4-fold (54). Wild-type and inp54 null mutant cells transformed with the BPTI plasmid were plated in 10-fold serial dilutions onto complete minimal media supplemented with either glucose or galactose and incubated at 30°C for 2 days. Both wild-type and inp54 null mutant strains grew inefficiently (up to 10 Ϫ4 dilution) on galactose plates, when BPTI was produced constitutively. Viability was 10 3 -fold less compared with cells grown on glucose plates (up to 10 Ϫ7 dilution), in which BPTI synthesis was repressed. There was no significant difference in viability between the wild-type and inp54 mutant when BPTI was expressed continuously (data not shown), indicating that the increased secretion in the inp54 null mutant strain was not sufficient to rescue the cells from BPTI-induced growth arrest. DISCUSSION The results of the studies described here demonstrate that the yeast PtdIns(4,5)P 2 5-phosphatase Inp54p is a C-terminal tail-anchored protein that localizes to the cytoplasmic surface of the ER where it may play a role in the regulation of secretion. Evidence for the ER localization of this novel 5-phosphatase is suggested by the presence of the C-terminal hydrophobic domain, which is consistent with a tail-anchor motif comprising a stretch of hydrophobic residues at the extreme C terminus.
Many studies indicate C-terminal tail-anchored proteins localize to either the mitochondria or the endoplasmic reticulum (reviewed in Ref. 36). We have demonstrated that recombinant Inp54p expressed as a fusion protein with either GFP or HA co-localizes with the ER-specific protein BiP, both when the protein is overexpressed, or expressed under the control of its native promoter. We have shown inp54 null mutant cells demonstrate increased secretion, consistent with a functional role for Inp54p in the ER.
All four yeast 5-phosphatases hydrolyze PtdIns(4,5)P 2 forming PtdIns(4)P. The necessity for four 5-phosphatases in yeast to regulate cellular concentrations of PtdIns(4,5)P 2 may in part be explained by the localization of each 5-phosphatase to spe-cific membrane compartments. However, this has yet to be shown. Recently our laboratory has demonstrated that Inp52p and Inp53p are localized diffusely throughout the cell excluding the nucleus and translocate to cortical actin patches following hyperosmotic stress (21). The intracellular localization of Inp51p has not been delineated, as the enzyme is extensively proteolyzed when expressed as a recombinant protein (21,55). The localization of Inp54p to the ER is the first evidence of the targeting of a yeast 5-phosphatase to a specific subcellular compartment in the resting cell. To date no mammalian 5-phosphatase has been localized to the ER, although the Lowe's protein has been identified in both the Golgi and the lysosomal compartment (56,57), and a novel 72-kDa 5-phosphatase has been localized to the cytoplasmic surface of the Golgi, where the enzyme is proposed to regulate phosphoinositide-mediated vesicular trafficking (58).
Inp54p localizes to the ER of yeast cells by the C-terminal hydrophobic tail comprising the last 13 amino acids of the protein. Deletion of the Inp54p hydrophobic tail resulted in a cytoplasmic distribution of the protein, consistent with previous observations that the C terminus of tail-anchored proteins contains the membrane targeting information (38,45,47,48,59,60). The last 13 amino acids of Inp54p also direct GFP to the ER. The length of the hydrophobic tail is a critical factor in determining intracellular localization. While shorter hydrophobic domains of 10 -12 amino acids have been demonstrated to target synaptobrevin and cytochrome b 5 to the ER (59, 61), domains of greater than 21 amino acids generally result in plasma membrane targeting (34,50). Lengthening the hydrophobic tail of the ER protein Ufe1p from 16 to 24 amino acids, by the insertion of eight hydrophobic residues, resulted in the protein being mislocalized to the plasma membrane (50). Increasing the length of the C-terminal membrane anchor of UBC6 from 17 to 21 amino acids caused mistargeting of the protein from the ER to the Golgi, whereas a further increase to 26 amino acids resulted in expression of the protein at the plasma membrane (34). Based on these previous studies, the short 13-amino acid targeting sequence of Inp54p would be consistent with the ER localization of the protein.
The exact mechanism by which C-terminal anchored proteins are targeted to specific membranes remains unclear. Several studies have shown that the targeting of various isoforms of synaptobrevin to the ER requires ATP, but is independent of the Sec61p/SRP pathway, relying instead on an as yet uncharacterized receptor system (37,40,62). The C-terminal tails of VAMP-2 and Ufe1p form an amphipathic helix critical for localization of the proteins to the ER (50,63). Mutation of the residues forming the polar face of the helix of Ufe1p to leucine abolished ER targeting of the protein (50). Mapping of the C-terminal hydrophobic sequence of Inp54p to a helical wheel FIG. 8. Possible models of Inp54p topology on the ER membrane. The C-terminal hydrophobic tail may "dip" into the membrane and loop back into the cytoplasm (A). Alternatively, the tail may associate tightly with a receptor/binding protein, where it binds noncovalently in a "hairpin loop" mode (B) (adapted from Frangioni et al. (47)).

FIG. 7. Secretion of a reporter protein is increased in inp54
null mutant cells. A plasmid containing BPTI under the control of a galactose-inducible promoter was transformed into wild-type, inp51, inp52, inp53, or inp54 null mutant strains. Yeast cells were incubated for 0, 16, 24, and 48 h in galactose-containing media to induce production of BPTI secreted into the media. The amount of BPTI at each time point was calculated in triplicate and averaged as described under "Experimental Procedures." Results shown are one representative experiment of three. f, wild-type strain; Ⅺ, inp51 null mutant; q, inp52 null mutant; E, inp53 null mutant; and OE, inp54 null mutant. reveals that the lysine, tyrosine, and serine are predicted to line one face of the helix, however, this would not be predicted to form an amphipathic face. Mutation of the positively charged lysine residue at position 382 had no effect on the localization of the protein, as has been shown for mutation of the arginine residue within the C terminus of cytochrome b 5 (48).
We predict Inp54p is orientated on the ER membrane with the bulk of the protein located on the cytoplasmic side of the ER membrane. Location of the tag did not affect the cytoplasmic orientation of Inp54p, or its intracellular localization. Previous studies have shown that C-terminal tags attached to cytochrome b 5 and synaptobrevin did not affect their insertion or orientation in the membrane (33,37,41). The topology of tailanchored proteins in the membrane is predicted to mimic classical transmembrane segments, whereby the tail spans the width of the membrane bilayer, with only a few residues on the other side of the membrane (32). At only 13 amino acids in length, the hydrophobic region of Inp54p is too short to span the membrane as typically integral membrane proteins contain a hydrophobic span of 20 amino acids (64). We have also demonstrated that although Inp54p membrane association is resistant to salt extraction, it is sensitive to alkaline treatment, indicating that it is not an integral membrane protein. Since a GFP tag fused to the extreme C terminus of Inp54p was degraded by proteinase K, it is most likely that the C terminus of Inp54p is also exposed to the cytoplasm. Thus, we propose the hydrophobic tail of Inp54p dips into the membrane to form a hairpin loop configuration whereby both the N and C termini are located in the cytosol (Fig. 8). As Inp54p membrane association is sensitive to high salt treatment, it is also possible that Inp54p could form a tight noncovalent interaction with an ER membrane protein. Protein-tyrosine phosphatase 1B has been proposed to adopt such a membrane topology (47).
The localization of Inp54p on the cytoplasmic surface of the ER would place the enzyme in an optimal position to regulate PtdIns(4,5)P 2 levels on vesicles budding from the ER. In mammalian cells PtdIns(4,5)P 2 binds ADP-ribosylation factor which results in the recruitment of coatomer proteins and activation of phospholipase D, followed by vesicle budding from the Golgi membrane (65)(66)(67). In yeast, the PtdIns 4-kinase Pik1p is essential for normal Golgi morphology and Golgi-plasma membrane trafficking (52,53,68). Pik1p-dependent generation of PtdIns(4)P regulates the transport of vesicles destined for the cell surface and normal secretion (52). Pik1p kinase activity is also required for normal endocytic transport of Ste6p from the plasma membrane to the vacuole (53). In mammalian cells PtdIns 4-kinase plays a role in secretion of granules from chromaffin cells (69). As pik1 null mutants demonstrate decreases in both PtdIns(4)P and PtdIns(4,5)P 2 , a role for PtdIns(4,5)P 2 in regulating secretion cannot be excluded (53). While PtdIns(4,5)P 2 has not been shown to play a definitive role in vesicular trafficking from the ER, our results would suggest the increased secretion observed in inp54 null mutants is mediated by the regulation of PtdIns(4,5)P 2 in this compartment.