A PTEN-like phosphatase with a novel substrate specificity.

We show that a novel PTEN-like phosphatase (PLIP) exhibits a unique preference for phosphatidylinositol 5-phosphate (PI(5)P) as a substrate in vitro. PI(5)P is the least characterized member of the phosphoinositide (PI) family of lipid signaling molecules. Recent studies suggest a role for PI(5)P in a variety of cellular events, such as tumor suppression, and in response to bacterial invasion. Determining the means by which PI(5)P levels are regulated is therefore key to understanding these cellular processes. PLIP is highly enriched in testis tissue and, similar to other PI phosphatases, exhibits poor activity against several proteinaceous substrates. Despite a recent report suggesting a role for PI(5)P in the regulation of Akt, the overexpression of wild-type or catalytically inactive PLIP in Chinese hamster ovary-insulin receptor cells or a dsRNA-mediated knockdown of PLIP mRNA levels in Drosophila S2 cells does not alter Akt activity or phosphorylation. The unique in vitro catalytic activity and detailed biochemical and kinetic analyses reported here will be of great value in our continued efforts to identify in vivo substrate(s) for this highly conserved phosphatase.

We show that a novel PTEN-like phosphatase (PLIP) exhibits a unique preference for phosphatidylinositol 5-phosphate (PI(5)P) as a substrate in vitro. PI(5)P is the least characterized member of the phosphoinositide (PI) family of lipid signaling molecules. Recent studies suggest a role for PI(5)P in a variety of cellular events, such as tumor suppression, and in response to bacterial invasion. Determining the means by which PI(5)P levels are regulated is therefore key to understanding these cellular processes. PLIP is highly enriched in testis tissue and, similar to other PI phosphatases, exhibits poor activity against several proteinaceous substrates. Despite a recent report suggesting a role for PI(5)P in the regulation of Akt, the overexpression of wild-type or catalytically inactive PLIP in Chinese hamster ovary-insulin receptor cells or a dsRNA-mediated knockdown of PLIP mRNA levels in Drosophila S2 cells does not alter Akt activity or phosphorylation. The unique in vitro catalytic activity and detailed biochemical and kinetic analyses reported here will be of great value in our continued efforts to identify in vivo substrate(s) for this highly conserved phosphatase.
Protein-tyrosine phosphatases (PTPs) 1 are a family of ϳ100 phosphatases characterized by their highly conserved CX 5 R catalytic motif. Once thought to dephosphorylate only phosphotyrosine residues, a subset of PTPs are now known to dephosphorylate phosphoserine-and phosphothreonine-containing proteins, as well as to use RNA and phosphoinositides as substrates in vitro and in vivo (1)(2)(3)(4)(5). A structural study of the tumor suppressor phosphatase PTEN revealed a wider active site cleft and suggested positions of key basic amino acids in the P-loop (CKAGKGR) among the reasons for its ability to use the phosphoinositide PI(3,4,5)P 3 as its preferred substrate (6). Consistent with this observation, other PTPs possessing highly similar or identical active site motifs, including the PTEN homologs PTEN 2 and TPIP, bacterial effector phosphatases SopB and IpgD, and inositol polyphosphate-4 phosphatases 1 and 2, have also now been shown to possess activity against phosphoinositide substrates (7-13) (see Fig. 1A).
In our bioinformatic searches of the NCBI databases for other PTEN-like PTPs, we identified a highly conserved phosphatase with over 60 orthologs throughout the Animalia, Plantae, Protista, and Eubacteria kingdoms. Here, we report that the murine ortholog of this PTEN-like phosphatase (PLIP), 2 possesses an in vitro activity against a rare phosphoinositide, PI(5)P.
Phosphatidylinositol, an abundant membrane phospholipid, is capable of being phosphorylated on the 3, 4, and 5 positions of its inositol ring to form seven unique lipid signaling molecules collectively termed phosphoinositides (PIs) (14). PIs regulate critical cellular functions, including apoptosis, membrane trafficking, cytoskeletal rearrangement, metabolism, growth, and differentiation by altering the subcellular location, state of aggregation, and activity of a variety of cellular enzymes. PI regulation by lipid kinases, phosphatases, and lipases is therefore critical in achieving proper cellular responses to outside stress (14,15). PI(5)P is the least characterized PI, having only recently been identified as an endogenous lipid (16). Recent studies (7,17,18) report changes in intracellular PI(5)P levels during cell cycle progression, as well as upon thrombin treatment and osmotic stress. Furthermore, the plant homeodomain-containing ING2 protein, a candidate tumor suppressor, was recently shown to act as a nuclear PI(5)P receptor, a function that regulates its ability to activate p53 (19). PI(5)P has also been tied to tumor suppression via its potential regulation of Akt. It was recently demonstrated that the loss of PI(5)P, via a conversion to PI(4,5)P 2 by the phosphoinositide kinase PIP4K II, resulted in a decrease in Akt activity (20). Lastly, PI(5)P has been shown to enhance the activity of various myotubularin phosphatases (MTM1, MTMR3, and MTMR6) toward their preferred substrate PI(3,5)P 2 presumably through allosteric regulation (21). Together these studies stress the importance of PI(5)P as a bona fide signaling molecule, and not merely a metabolic precursor to other PIs as once proposed.
Here, we report the first example of a mammalian phosphatase utilizing PI(5)P as its preferred substrate. PLIP exhibits a highly selective in vitro activity against PI(5)P at a level comparable with that of PTEN against its preferred substrate, PI(3,4,5)P 3 . To date, we have not demonstrated the ability of PLIP to alter PI(5)P levels in vivo. Thus, despite having poor activity toward several proteinaceous substrates, PLIP may still possess a highly specific proteinaceous substrate not assessed in this study. The continued analysis of endogenous * This work was supported by National Institutes of Health Grant 18024 and funds from the Walther Cancer Institute (to J. E. D.) and by National Institutes of Health Pharmacology Training Grant NIH 2 T32 GM07752-25 (to D. J. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  "PROT" denotes sequences derived from characterized proteins or predicted proteins from fully sequenced genomes, "EST" denotes sequences derived from expressed sequence tags. NCBI accession numbers are given at right. (Note: cladogram not to scale.) C, primary sequence of murine PLIP. Shaded region, beginning with green-boxed-start methionine, indicates the sequence of the clone used in this study. The catalytic P-loop region is highlighted in red.
substrates for PLIP should be greatly aided by the extensive biochemical characterization of this enzyme and its highly specific activity against PI(5)P.

EXPERIMENTAL PROCEDURES
Expression and Purification of PLIP-Recombinant murine PLIP was expressed as a fusion protein with N-terminal GST and His 6 tags in Escherichia coli BL21 (DE3) CodonPlus RIL cells (Stratagene). The expression construct was assembled by ligating a PCR product encoding full-length PLIP into the 5Ј-EcoRI and 3Ј-XhoI sites of a modified pET41 vector (pSJ6), kindly provided by Zhaohui Xu (University of Michigan, Ann Arbor, MI). One-liter bacterial cultures were grown at 37°C in 2ϫ YT medium (8 g of tryptone, 5 g of yeast, 5 g of NaCl/liter) possessing chloramphenicol (34 g/ml) and kanamycin (50 g/ml) to an optical density of 0.5-0.7. Cultures were then chilled on ice for 20 min, supplied with fresh antibiotics, induced with isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM, and allowed to grow overnight at 25°C as described previously (22). All subsequent steps were performed at 4°C. The cultures were pelleted by spinning at 6000 rpm for 10 min in a Sorvall GSA rotor, resuspended in 30 ml of buffer (20 mM imidazole, 50 mM Tris, pH 7.4, 300 mM NaCl) with complete protease inhibitors (Roche Applied Science), and lysed via French press.

FIG. 2. Analysis of PLIP activity against phosphoinositides.
A, wild-type (WT) or catalytically inactive (C132S) PLIP was tested against a panel of water-soluble BODIPY-tagged di-C 6 phosphoinositides. Conversion of PI(5)P to PI by wild-type PLIP is indicated with a red arrow. Migration distances of the various PI derivatives are indicated on the right. B and C, activity of PLIP or PTEN against di-C 8 PIs incorporated into a lipid bilayer with phosphatidylserine carrier lipid. D, activity of PLIP against water-soluble PI(5)P derivatives of varying acyl chain length.
Insoluble material was removed by centrifugation at 50,000 rpm for 30 min. The fusion protein was purified from the soluble supernatant first over Ni 2ϩ -agarose beads and then by successive separation on Superdex 75 and 200 gel filtration columns (Amersham Biosciences). The resulting protein was concentrated, flash frozen in liquid nitrogen, and stored at Ϫ80°C in 25% glycerol, 2 mM EDTA, and 2 mM dithiothreitol until use.
Fluorescent Phosphoinositide Assays-Fluorescent PI, TLC, and malachite green assays were performed as described recently (22). Eighteen microliters of assay buffer (50 mM ammonium acetate, pH 5.5, 0.1% (v:v) 2-mercaptoethanol (Sigma)) containing 1 g of a Di-C 6 -NBD6phosphoinositide (Echelon) were prewarmed at 37°C for 5 min. Reactions were initiated by the addition of 2 l of 0.1 g/l GST-PLIP diluted in assay buffered containing 1.0 mg/ml gelatin. Assays were quenched after 2.0 min by the addition of 100 l of acetone and were dried in a Speed Vac at medium heat.
Thin Layer Chromatography-Whatman K6 silica gel plates (Fisher) were soaked in a 1.2% solution of potassium oxalate, air-dried in a fume hood for 10 min, and placed in a baking oven for 1 h at 180°F under vacuum conditions. The dried products of the fluorescent phosphoinositide assays were resuspended in 20 l of MeOH/2-propanol/glacial acetic acid (5:5:2) and spotted onto the TLC plate. The plate was dried for 10 min in a fume hood and developed in CHCl 3 /MeOH/acetone/glacial acetic acid/water (70:50:20:20:20). The plate was again dried in a fume hood and visualized on a Bio-Rad DNA gel UV illuminator.
Malachite Green Assays-Di-C 8 phosphoinositides and dioleoyl-phosphatidylserine (Sigma) were dried together in a Speed Vac and resuspended via sonication in 18 l of assay buffer (100 mM sodium acetate, 50 mM bis-Tris, 50 mM Tris, pH 5.5, 10 mM dithiothreitol) to final concentrations of 50 and 500 M, respectively. After prewarming at 37°C for 5 min, reactions were initiated by the addition of 20 -2000 ng of GST-PLIP diluted in assay buffer containing 1.0 mg/ml gelatin. Reactions were quenched after 5-30 min by the addition of 20 l of 0.1 M N-ethylmaleimide and spun at 18,000 ϫ g for 10 min to sediment the lipid aggregates. Twenty-five microliters of the supernatant was added to 50 l of malachite green reagent and vortexed. Samples were allowed to sit for 40 min for color development before measuring absorbance at 620 nm. Inorganic phosphate release was quantitated by comparison to a standard curve of KH 2 PO 4 in distilled H 2 O.
Tissue Expression-Tissue distribution of PLIP was analyzed both with a FirstChoice Northern blot System (Ambion) using a randomprimed 32 P-labeled probe as well by PCR analysis using a mouse multiple tissue cDNA panel (Clontech) with the following primers, 5Ј-CC-ACCGCATCGACCACACGGTTCTGC-3Ј (forward) and 5Ј-CCTCCTCT-GGGCTCCAGTTGTGTACCTGAATCAG-3Ј (reverse).
Analysis of Akt Activity in Chinese Hamster Ovary-Insulin Receptor Cells-Chinese hamster ovary-insulin receptor cells were maintained at 37°C and 5% CO 2 in ␣-minimum essential medium (Invitrogen) supplemented with 10% fetal bovine serum, 50 units/ml each of penicillin and streptomycin, and 50 g/ml Geneticin. Cells were transfected with FLAG-Akt (kindly provided by Dr. Anne Vojtek, University of Michigan, Ann Arbor, MI) and vector, wild-type PLIP-V5, or C132S PLIP-V5 in a 1:10 ratio using FuGENE (Roche Applied Science) according to the manufacturer's recommended protocol. Twenty-four-hours posttransfection, cells were serum-starved overnight followed by treatment with Ϯ10 nM insulin for 5 min. Cells were then lysed, and the Akt was immunoprecipitated using anti-FLAG-agarose (Sigma). Samples for all six conditions were Western blotted for Akt (anti-FLAG), phospho S473 Akt (Cell Signaling Technology), or PLIP (anti-V5 (Invitrogen)).
RNAi in Drosophila S2 Cells-S2 cells were grown in 1ϫ Schneider's Drosophila medium (Invitrogen) supplemented with 10% fetal bovine serum and 50 g/ml streptomycin in 75-cm 2 T-flasks (Starstedt) at room temperature. dsRNA for RNAi experiments was produced and used to treat S2 cells as previously reported in detail (24). Briefly, dsRNA was added to a final concentration of ϳ37 nM to 1 ϫ 10 6 cells in each well of a six-well plate followed by vigorous agitation. Following a 30-min incubation at 25°C, 2 ml of 1ϫ Schneider's medium with fetal bovine serum was added. After 3 days at 25°C, the mRNA was isolated using a mRNA capture kit from Roche Applied Science and was analyzed by gel electrophoresis. Akt immunoprecipitations and activity assays were performed as described previously (25).

Identification of PLIP as a PTEN-like Phosphatase-Eluci-
dation of the crystal structure of PTEN revealed the importance of the conserved active site lysine residues (Lys-125, Lys-128) in establishing a negatively charged catalytic pocket conducive to the binding of its preferred substrate, PI(3,4,5)P3 (6). In an attempt to identify additional phosphatases possessing this motif, we conducted PSI-BLAST analyses using the PTEN phosphatase domain as a query (26,27). We subsequently identified a predicted protein (NCBI accession, XP_374879) that possessed the PTEN-like active site CK-AGRSR. PTENs from all sources have basic amino acids in the highlighted positions. Using the murine ortholog of this protein, we conducted multiple PSI-BLAST and TBLASTN searches against the non-redundant and EST databases. Through this analysis, we identified what we believe are over 60 orthologs of a protein we now call PLIP. Included in these results is an ortholog from the eubacterium Pirellula sp. strain 1 whose defining characteristics include an intracellular membrane and various eukaryotic-like lipids (28,29). The active site region of PLIP shows remarkable evolutionary conservation, exhibiting greater than 70% identity/80% similarity in orthologs from four different phylogenetic kingdoms (Fig. 1B). Such conservation supports the notion that residues within this region are critical for proper enzyme function, most likely in establishing substrate specificity.
PLIP Exhibits PI(5)P Phosphatase Activity-The murine ortholog of PLIP (Fig. 1C), was cloned into a GST bacterial expression vector, expressed, and purified to near homogeneity. The predicted amino acid sequence of this ortholog (NCBI accession number gi:23956130) has an extended N terminus (Fig. 1C, residues 1-69). However, as this region was not found in any of the more than 60 other PLIP orthologs, it was omitted from the final GST-PLIP construct used for the biochemical characterization of PLIP.
A pH/rate profile for PLIP performed with various substrates revealed this enzyme to be most efficient at pH 5.5 (data not shown). Therefore, all further assays were performed at this pH. Because PLIP had active site residues similar to those found in PTEN, we examined its activity toward a panel of fluorescently labeled di-C 6 phosphoinositides. A TLC analysis of the reaction products, shown in Fig. 2A, revealed that PLIP exhibits a highly selective substrate specificity for PI(5)P. This substrate preference is shared by the Dictyostelium ortholog of PLIP (30). As expected, the mutation of the predicted catalytic cysteine residue to serine (C132S) nullified this activity. To further confirm this result, PLIP was tested against a panel of di-C 8 PIs presented in a lipid bilayer with phosphatidylserine  carrier lipid. PLIP again demonstrated robust activity against PI(5)P, 44-fold greater than against PI(3,5)P 2 , its second most preferred substrate (Fig. 2B). This is a notable enhancement in specificity compared with PTEN, which exhibits only a 1.8-fold preference for PI(3,4,5)P 3 over PI(3,5)P 2 (Fig. 2, B and C). PLIP assayed against di-C 16 PIs revealed highly similar results (data not shown).
We also tested water-soluble PI(5)P of multiple acyl chain lengths as substrates for PLIP, as this variable has been shown to affect the level of enzymatic activity for other PI phosphatases (31,32). PLIP demonstrated a 10-fold increase in activity against di-C 4 PI(5)P versus the inositol head group, inositol (1,5)P 2 , and a second 10-fold increase against di-C 8 PI(5)P versus di-C 4 PI(5)P (Fig. 2D) suggesting that this lipid moiety is also an important factor for PLIP activity.
PI(5)P Activity Is Not a General Feature of PTPs-Because of the only recent emergence of PI(5)P as a known signaling molecule, few PTPs have been tested for activity against this substrate. To ensure that activity against PI(5)P is not a common feature of PTPs, membrane-bound di-C 8 PI(5)P was assayed using a known tyrosine-specific PTP (T-cell-PTP) and a known dual-specific PTP (VH-1-related) (33)(34)(35)(36)(37). Both enzymes yielded activity that was barely detectable above background indicating that PI(5)P activity is not a general feature of the various classes of PTPs (Table I). PTEN and MTM1, known PI phosphatases, both possess the ability to dephosphorylate PI(5)P although with 17-and 200-fold less efficiency than toward their preferred substrates, respectively.
PLIP Exhibits Poor Protein Phosphatase Activity-A characteristic of all PI phosphatases to date has been their extremely poor activity toward proteinaceous substrates (32). To determine whether PLIP is consistent with this trend, its activity was measured against a radiolabeled myelin basic protein and casein side-by-side with the aforementioned phosphatases. The activity of PLIP against these substrates was very poor, most closely resembling that of PTEN and myotubularin, and was ϳ1000-fold lower than the activity of PLIP against PI(5)P (Table I).
To further explore the propensity of PLIP to act as a protein phosphatase, we performed a kinetic analysis of its activity toward the artificial phosphotyrosine analog, pNPP, along side with di-C 8 PI (5)  against this substrate, more than 150-fold higher than the K m of 37.5 M seen for di-C 8 PI(5)P (Fig. 3).
Thus, PLIP demonstrated a clear preference for PI(5)P over any other proteinaceous or lipid substrate tested. Although our efforts to demonstrate the effect of PLIP on endogenous PI(5)P levels are ongoing, we note that in vitro observations have often served as indicators of in vivo function for PI phosphatases (10, 38 -40).
PLIP Is a Testis-enriched Phosphatase-The relative tissue distribution of PLIP was determined by a Northern blot assay using a 32 P-labeled, random-primed PLIP cDNA probe. A major band was detected in the testis lane consistent with the pre-  (20). B, wild-type (WT) PLIP, catalytically inactive (C132S) PLIP, or vector control were transfected along with FLAG-Akt into Chinese hamster ovary-insulin receptor cells and treated with (ϩ) or without (Ϫ) insulin. Following immunoprecipitation with anti-FLAG-agarose beads, Akt and phospho-Akt levels were probed via Western blot using anti-FLAG and anti-phosphoserine 473-Akt-specific antibodies, respectively. A third blot was performed on the cell lysates using anti-V5 (PLIP) to demonstrate the presence of wild-type and C132S PLIP. Note that because of a relatively higher level of Akt expression in the vector control cells, lanes 1 and 2 are derived from lower film exposure times. Results are representative of two separate experiments. C, Drosophila S2 cells were treated with buffer (U) or PLIP or PTEN dsRNA. Following treatment, endogenous Akt was immunoprecipitated and assayed for activity. The effectiveness of the dsRNA treatment, as assessed by mRNA levels, is shown in the blot below. dicted size of the PLIP transcript (Fig 4A). Glyceraldehyde-3phosphate dehydrogenase, probed as a positive control, showed strong expression in all tissues. This result was confirmed by a highly sensitive PCR analysis of cDNA from various tissue lysates that detected PLIP in testis after 26 cycles (Fig. 4B). Relatively high amounts of PLIP cDNA could also be detected in the brain, liver, skeletal muscle, and uterus. After 38 PCR cycles, all lanes were positive indicating that a small amount of message is present in all tissues tested (Fig. 4). The relative enhancement of PLIP in testis tissue is interesting as it follows similar results for two other new PTEN-like phosphatases, PTEN2 and TPIP (8,9).
PLIP Is Not a Regulator of Akt Activation-Recently, Carricaburu et al. (20) proposed that PI(5)P may control Akt activation by regulating the activity of a PI(3,4,5)P 3 phosphatase. In their study, they report that depletion of intracellular PI(5)P by a PI(5)P kinase results in abbreviation of Akt activation (Fig.  5A). To test whether PLIP could modulate Akt activity, we overexpressed wild-type PLIP, catalytically inactive C132S PLIP, or vector control along with FLAG-tagged Akt into Chinese hamster ovary-insulin receptor cells. After serum starvation and with or without insulin treatment (see "Experimental Procedures"), Akt was immunoprecipitated and analyzed by Western blotting. Full activation of Akt requires that it is phosphorylated on serine 473 on its C terminus. As seen in Fig.  5, the level of Akt phosphorylation on serine 473 is not significantly affected by overexpression of wild-type or C132S PLIP, with or without insulin treatment. To ensure that this result was not due to an artifact of overexpression, we used RNAi to knock down the mRNA levels of PLIP in Drosophila S2 cells, which have previously been shown to possess the members of the Drosophila Akt signaling pathway (25,41,42). S2 cells were treated with buffer (negative control), PLIP, or PTEN dsRNA, followed by Akt immunoprecipitation and activity assays. Despite a major reduction in mRNA levels, Akt activity was unaffected by PLIP RNAi treatment, whereas the PTEN knock-out exhibited the previously reported enhancement of Akt activity (Fig. 5) (25). From these data, we concluded that PLIP is not a modulator of Akt function. Furthermore, unless PLIP, despite being overexpressed, is not spatially poised to alter the necessary PI(5)P pools in vivo, these data weaken the idea that this lipid is capable of regulating Akt activity.
The characterization of PLIP as a highly specific PI(5)P phosphatase is a potentially important contribution to our understanding of PI regulation and adds a powerful tool for further analysis of PI signaling pathways within the cell. Until we achieve a clear demonstration of the ability of PLIP to alter PI(5)P levels in vivo, however, we cannot rule out the possibility that it possesses a distinct endogenous substrate not assessed in this study. Even if PLIP utilizes a highly specific proteinaceous substrate in vivo, the fact that it possesses this unique and well characterized preference for PI(5)P should make it possible to selectively measure PLIP activity under a variety of experimental conditions. Detailed investigations into the subcellular localization, in vivo enzymatic activity, and overall cellular function of PLIP are ongoing in our laboratory.