The isolation and characterization of a cDNA encoding phospholipid-specific inositol polyphosphate 5-phosphatase.

We report the cDNA cloning and characterization of a novel human inositol polyphosphate 5-phosphatase (5-phosphatase) that has substrate specificity unlike previously described members of this large gene family. All previously described members hydrolyze water soluble inositol phosphates. This enzyme hydrolyzes only lipid substrates, phosphatidylinositol 3,4,5-trisphosphate and phosphatidylinositol 4,5-bisphosphate. The cDNA isolated comprises 3110 base pairs and predicts a protein product of 644 amino acids and M(r) = 70,023. We designate this 5-phosphatase as type IV. It is a highly basic protein (pI = 8.8) and has the greatest affinity toward phosphatidylinositol 3,4,5-trisphosphate of known 5-phosphatases. The K(m) is 0.65 micrometer, 1/10 that of SHIP (5.95 micrometer), another 5-phosphatase that hydrolyzes phosphatidylinositol 3,4,5-trisphosphate. The activity of 5-phosphatase type IV is sensitive to the presence of detergents in the in vitro assay. Thus the enzyme hydrolyzes lipid substrates in the absence of detergents or in the presence of n-octyl beta-glucopyranoside or Triton X-100, but not in the presence of cetyltriethylammonium bromide, the detergent that has been used in other studies of the hydrolysis of phosphatidylinositol 4,5-bisphosphate. Remarkably SHIP, a 5-phosphatase previously characterized as hydrolyzing only substrates with d-3 phosphates, also readily hydrolyzed phosphatidylinositol 4,5-bisphosphate in the presence of n-octyl beta-glucopyranoside but not cetyltriethylammonium bromide. We used antibodies prepared against a peptide predicted by the cDNA to identify the 5-phosphatase type IV enzyme in human tissues and find that it is highly expressed in the brain as determined by Western blotting. We also performed Western blotting of mouse tissues and found high levels of expression in the brain, testes, and heart with lower levels of expression in other tissues. mRNA was detected in many tissues and cell lines as determined by Northern blotting.

Inositol polyphosphate 5-phosphatases (5-phosphatases) 1 comprise a large family of enzymes that cleave the 5 position phosphate of several inositol phosphates and lipids including inositol 1,4,5-trisphosphate, inositol 1,3,4,5-tetrakisphosphate, phosphatidylinositol 3,4,5-trisphosphate (PtdIns 3,4,5-P 3 ) and phosphatidylinositol 4,5-bisphosphate (PtdIns 4,5-P 2 ) (1, 2). There are eight mammalian 5-phosphatases known to date, four yeast enzymes, and two putative 5-phosphatases in Caenorhabditis elegans. Drosophila and arabidopsis genes have also been identified. Deletions of mammalian 5-phosphatases in mice suggest that the enzymes serve mainly nonredundant functions in signaling reactions. Thus deletion of SHIP, a 5-phosphatase expressed mainly in hematopoietic cells, causes a severe myeloid proliferation and infiltration of the lungs (3). Deletion of synaptojanin is accompanied by defects in neuronal signaling with a failure to nurse and early demise (4). Deletion of OCRL, the gene mutated in Lowe syndrome, has no phenotype. However, when these mice were bred with type II 5-phosphatase-deficient mice, the double knockout resulted in embryonic lethality (5). The mammalian 5-phosphatases have been previously classified into four groups based on substrate specificity (1,2). Group I hydrolyzes only the water soluble inositol phosphate substrates, group II hydrolyzes all four known 5-phosphatase substrates, group III was previously characterized by hydrolysis only of substrates with phosphate in the D-3 position, and group IV was represented as an activity that was specific for hydrolysis of only PtdIns 3,4,5-P 3 . We now report the cloning and characterization of a type IV enzyme. Initially we found that the enzyme only hydrolyzed PtdIns 3,4,5-P 3 ; however, in further study of recombinant enzyme we find that under some conditions it will also utilize PtdIns 4,5-P 2 . Because multiple roles for inositol lipids in cell signaling have been reported, this enzyme represents a candidate enzyme for controlling these reactions (6,7).

EXPERIMENTAL PROCEDURES
cDNA Cloning-The consensus sequences that define 5-phosphatases were used to search the sequence data bases using the BLAST algorithm. An EST from human infant brain, accession number H10559, was identified as a putative 5-phosphatase. The EST was obtained from ATCC and sequenced. The 1.8-kb EST insert was used to screen a human fetal brain zapII cDNA library (Stratagene) to obtain an additional sequence. 5Ј-RACE of human fetal brain Marathon-Ready cDNA (CLONTECH) was performed according to the manufacturer's instructions to obtain the 5Ј-sequence. However, the 5Ј-region of 5-phosphatase type IV contains a very high percentage of GC residues, including long stretches of multiple GC repeats, leading to secondary struc-* This work was supported by Grants HL55772 and HL 16634 from the National Institutes of Health. 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.
Northern Blotting-Multiple tissue Northern blots (CLONTECH) were probed with the XmaI fragment corresponding to nucleotides 1834 -2921, encoding the very C terminus of 5-phosphatase type IV protein and 3Ј-untranslated region. A gel-purified fragment was labeled by random priming with a rediprime DNA-labeling system (Amersham Pharmacia Biotech). ␤-Actin probe was used as a control.
Expression and Purification-The Bac-to-Bac expression system (Life Technologies, Inc.) was used for high level expression of 5-phosphatase type IV in cultured insect cells. cDNA (nucleotides 1-2135) was subcloned into the pFastBac1 vector to express full-length 5-phosphatase type IV protein. For purification, 5-phosphatase type IV was subcloned into the pFastBac HT vector. Cell extracts were harvested 48 h postinfection, and the protein was purified on Ni 2ϩ -NTA resin according to the manufacturer's instructions, with the addition of 5 mM MgCl 2 in all buffers.
TLC-Silica Gel 60 TLC plates were treated with a solution of 1% potassium oxalate in 50% ethanol. The plates were baked at 90°C overnight. TLC plates were developed using a solvent mixture of chloroform/acetone/methanol/glacial acetic acid/water (80:30:26:24:14 (v/v)). Phospholipid standards were stained with iodine. Radiolabeled phospholipids were detected by a PhosphoImager or autoradiography. Scrapings of radiolabeled phospholipids were counted in a Beckman liquid scintillation counter.
For enzyme assay, radiolabeled lipids were dried under nitrogen and resuspended in buffer containing 50 mM Tris-HCl, pH 7.5, 3 mM MgCl 2 , and 0.03% n-octyl ␤-glucopyranoside and briefly sonicated on ice. The assay of activity was performed as described previously (10,11).
Proof of Product-Deacylation of inositol lipids was as described previously (12). GroPIns derivatives were separated by HPLC on a PartiSphere SAX column (Whatman) with the gradient of 0 -1.25 M NaH 2 PO 4 , pH 4.5. The gradient consisted of a 0 -5-min linear rise to 29% for pump B, a 5-35-min linear rise to 60% B, and a 35-45-min linear rise to 100% B followed by a 5-min wash with 100% B. Radiolabeled deacylated phospholipids were detected by a ␤-RAM Flow-Through System (In/US Systems, Inc., Tampa, Florida).

RESULTS AND DISCUSSION
cDNA Cloning of Type IV Phosphatidylinositol Polyphosphate 5-Phosphatase-Inositol polyphosphate 5-phosphatases are defined by two essential domains, with the consensus sequences, FWXGDXN(F/Y)R and R/NXP(S/A)W(C/T)DR(I/V)L. Our method for screening for additional members of the family was a search of the human EST data base with the consensus sequences. One of the clones identified in the GenBank TM data base with the BLAST search was clone H10559, with a reported sequence FWFGDFNFR that corresponded to the consensus sequence of the 5-phosphatase domain I. The clone was sequenced and shown to encode a partial open reading frame that contained both 5-phosphatase domains. The 1.8-kb EST insert was used to screen a human fetal brain zapII cDNA library (Stratagene). However, all clones isolated contained only partial coding sequence. To obtain ad-ditional N-terminal sequence, we used a 5Ј-RACE protocol and human fetal brain Marathon-ready cDNA (CLONTECH). However, because of a very high GC content of the cDNA, the protocol was modified as described under "Experimental Procedures" (8,9). An additional 300 base pairs were isolated containing a methionine codon. Although the full-length mRNA is 3.4 kb according to the Northern blot (see below), we believe that we have identified the starting methionine, because the putative initiation site is in the context of a good Kozak consensus sequence (13) and there is an in-frame stop codon 5Ј of it.
The composite cDNA of 3110 base pairs encodes an open reading frame of 644 amino acids (Fig. 1A). It contains two 5-phosphatase domains, FWFGDFNFR and KQRTPSYT-DRVLY (invariant consensus residues are in bold). The sequence of domain II is unusual; it encodes tyrosine instead of an invariable tryptophane residue in the consensus motif. Only one other putative 5-phosphatase gene from C. elegans has the same tryptophane to tyrosine substitution. This gene has not been characterized at the protein level. The four C-terminal residues CSVS represent an S-farnesylation signal. Of other mammalian 5-phosphatases, types I and II also have posttranslational lipid modifications at their C termini. 5-Phosphatase type I has a CVVQ C-terminal isoprenylation site and is a substrate for purified farnesyltransferase (14). 5ptase type II has a CNPL prenylation motif and is likely to be geranyl-geranylated (15,16).
5-Phosphatase type IV has relatively low similarity to other known mammalian 5-phosphatases (Fig. 1B). There is only 10 -20% amino acid identity with the closest identity of 19% to SHIP. There are two mouse EST clones found in the sequence data base that encode the partial sequence of the 5-phosphatase type IV homologue. We sequenced both of them, and the deduced amino acid sequences are 87% identical to the human protein in the coding portion of the clones. The 3Ј-untranslated regions have no similarity.
5-Phosphatase type IV has a very high content of proline residues (9.5%), particularly in the N-terminal portion (19%). The N-terminal region of the enzyme contains a Pro-rich domain class I (17) and thirteen PXXP sequences, a potential SH3 binding core. Surprisingly, 5-phosphatase type IV also contains a sequence YVLLSSAAHGVLYMSL that corresponds to YXXL X (6 -8) YXXL consensus, which is the immunoreceptor tyrosine-based activation motif (ITAM). ITAM modules function to link cell surface receptors to signaling effector molecules and are found in the cytoplasmic tails of T-and B-cell receptors, CD3, high affinity Fc receptors for IgE and IgA, and Fc␥RIIA. This is the first case where an ITAM sequence has been reported in a nonreceptor molecule. One of the clones isolated in the screen of the human brain library had a 102-nucleotide deletion spanning this region. It is likely to represent an alternatively spliced transcript, resulting in deletion of 34 amino acids including the putative ITAM domain. It will be of interest to study the functional role of this protein motif in 5-phosphatase type IV. Another member of the 5-phosphatase family, SHIP, is known to interact with the immunoreceptor tyrosinebased inhibition motif of Fc␥RIIB in B cells. SHIP 1 is recruited to the phosphorylated immunoreceptor tyrosine-based inhibition motif via its SH2 domain, providing the inactivation of the signal presumably by hydrolysis of PtdIns 3,4,5-P 3 and inositol 1,3,4,5-tetrakisphosphate-signaling molecules (18). Because 5-phosphatase type IV is also a PtdIns 3,4,5-P 3 phosphatase, it may serve as a "double inhibitor," both hydrolyzing phosphatidylinositol signaling molecules and competing with immunoreceptors as a scaffolding base.
Tissue Distribution of Phosphatidylinositol 5-Phosphatase Type IV-Expression of 5-phosphatase type IV was analyzed by Northern blot analysis. In human tissues, 5-phosphatase type IV mRNA was present at highest levels in brain, heart, spleen, pancreas, and testis (Fig. 2, A and B). A major band of 3.6 kb was detected in all tissues; however, additional species of 4.9 and 9.5 kb are present as well and are most pronounced in testis. Other members of the 5-phosphatase family are known to exist as multiple spliced variants. Multiple splice forms of SHIP, SHIP2, synaptojanin, and 5-phosphatase type II have been reported (16,19). Different splice variants may have different tissue and/or intracellular distribution as in the case of 5-phosphatase type II (16) and synaptojanin 1 and synaptoja-nin 2 (20,21). They may also lack a protein module, SH2 domain in the case of SIP110, a splice variant of SHIP (19). A splice variant of synaptojanin 1 without an N-terminal SACI domain has been reported also (22). Both synaptojanin 1 and synaptojanin 2 exist as multiple splice forms. The majority of synaptojanin 1 in brain is a 145-kDa isoform, whereas a 175-kDa isoform is ubiquitously expressed. The 175-kDa isoform is tightly bound to membranes compared with the 145-kDa isoform. Interestingly, these two isoforms arise from a small (27 nucleotide) deletion that deletes the stop codon and results in splicing of a long open reading frame found in the 3Ј-untrans- lated region of synaptojanin (20). The same pattern was observed in the case of synaptojanin 2, even though there is no homology between the alternatively spliced C-terminal portions of synaptojanin 1 and synaptojanin 2 (21). The 3Ј-untranslated region of 5-phosphatase type IV also contains a long open reading frame immediately following the stop codon. It is possible that alternatively spliced isoforms of 5-phosphatase type IV may arise by a mechanism similar to that found with synaptojanin genes. We did not encounter any isoforms with such an alternatively spliced C terminus in our clones of 5-phosphatase type IV or in the EST data base. However, a cDNA entry in GenBank TM (accession number U 45974) has a small deletion spanning the stop codon (TG versus TTTCTT-GAAG) that would result in the addition of 50 amino acids to the C terminus of the protein.
5-Phosphatase type IV has a similar tissue distribution in mouse, with the highest level of expression in the testis, heart, and brain based on Northern blot analysis (Fig. 2D). We also tested several human cancer cell lines. The 5-phosphatase type IV mRNA is present in all cell lines tested in various amounts (Fig. 2C). A very high level of 5-phosphatase type IV mRNA was detected in MOLT-4, a lymphoblastic leukemia cell line. This is of interest, given the presence of an ITAM module in the 5-phosphatase type IV, found typically in immune cell receptors.
Expression in Human and Mouse Tissues and Sf9 Cells-Four human tissues were examined for the presence of 5-phosphatase type IV protein: brain, liver, spleen, and kidney by Western blotting. A single band of apparent molecular mass of 68 kDa was detected in the brain (Fig. 3B) but not in the liver, spleen, or kidney. We also performed Western blots of several mouse tissues and found high levels of expression in brain, heart, and testes with lower levels expressed in thymus and lung (Fig. 3A). After prolonged exposure of the Western blot faint bands were seen in kidney, spleen and liver. The observed molecular mass correlates well with the predicted mass of 70 kDa based on the cDNA sequence. The staining was specific and disappeared when primary antibodies were preincubated with the peptide antigen prior to staining (Fig. 3B, lane brain ϩ peptide).
We expressed the protein in Sf9 cells to study the catalytic activity of the enzyme. First, the full-length construct was expressed to confirm that the starting methionine supports initiation of protein synthesis. In accordance with the predicted molecular mass, a band of 70 kDa was detected in insect cells 48 h following infection with 5-phosphatase type IV baculovirus (Fig. 3C, lane 3). To obtain a highly purified enzyme, we expressed His-tagged 5-phosphatase type IV followed by affinity purification on a Ni 2ϩ -NTA-agarose column. (Fig. 3C, lanes  1 and 2). The purified enzyme was used in subsequent experiments to study its enzymatic activity.
PtdIns 3,4,5-P 3 Hydrolysis and the Effect of Detergents-Known inositol polyphosphate 5-phosphatases can potentially utilize four different substrates, soluble inositol 1,4,5-trisphosphate and inositol 1,3,4,5-tetrakisphosphate, as well as lipids, PtdIns 4,5-P 2 and PtdIns 3,4,5-P 3 . Substrate preference is the basis of classification of known 5-phosphatases into four groups as described above. We examined the ability of the type IV enzyme to hydrolyze all four potential substrates. It was unable to hydrolyze either of the soluble substrates, making it the first known 5-phosphatase that is not capable of metabolizing soluble substrates. In these experiments we used 10 -50 times

FIG. 3. Expression of 5-phosphatase type IV protein. A and B,
Western blot analysis of the tissue distribution of 5-phosphatase type IV protein. Soluble fractions from indicated mouse (A) and human (B) tissues (80 g of total protein) were analyzed by 10% reduced SDSpolyacrylamide gel electrophoresis, followed by immunostaining with affinity purified peptide-specific rabbit polyclonal antibodies. Specificity of the band was confirmed by preincubation of the primary antibodies with an excess amount of peptide (lane marked brain ϩ peptide). The position of molecular mass markers (in kilodaltons) are shown on the left. b, brain; lu, lung; li, liver; k, kidney; s, spleen; th, thymus; te, testis; h, heart; c, colon. C, expression of recombinant 5-phosphatase type IV in Sf9 cells. Lane 1, His-tagged 5-phosphatase type IV protein after affinity purification on Ni 2ϩ -NTA resin, silver staining. Lane 2, immunoblotting of the same sample with 5-phosphatase-specific polyclonal antibodies. Lane 3, immunoblotting of Sf9 cell soluble extract 48 h postinfection with the virus encoding full-length human 5-phosphatase type IV (not His-tagged).
as much recombinant enzyme as in the experiments using lipid substrates, incubated for up to 40 min, and found no hydrolysis. Thus if the enzyme were able to hydrolyze at even 1% of the rates seen with lipids, we would have detected it. When lipid substrates were used, type IV enzyme readily utilized PtdIns 3,4,5-P 3 converting it to PtdIns-P 2 in a time-and concentrationdependent manner as shown below. To identify the product of the reaction, the substrate and the product of the reaction were deacylated and analyzed on HPLC (Fig. 4). The GroPIns-P 2 product migrated in the position corresponding to PtdIns 3,4-P 2 , confirming that the novel enzyme is an inositol 5-phosphatase.
Of the known 5-phosphatases, SHIP is the most efficient enzyme that hydrolyzes PtdIns 3,4,5-P 3 . It is thought to be the primary enzyme that controls the level of PtdIns 3,4,5-P 3 in cells during activation of tyrosine kinase receptors. We compared the enzymatic activity of 5-phosphatase type IV and SHIP in hydrolyzing PtdIns 3,4,5-P 3 (Fig. 5B). Based on K m values, 5-phosphatase type IV has almost 10 times greater affinity for the substrate than SHIP (0.65 versus 5.95 M, respectively). 5-Phosphatase type IV has a lower V max (0.1145 mol/mg/min) compared with a V max of 0.458 mol/mg/min for SHIP. However, the catalytic efficiency (V max /K m ) of the two enzymes is similar, taking into account that molecular weight of SHIP is twice as high as 5-phosphatase type IV. We compared the first order rate constants of three 5-phosphatases: 5-phosphatase type IV, SHIP, and OCRL, that represent three different types of 5-phosphatases that are capable of PtdIns 3,4,5-P 3 hydrolysis (Fig. 6). SHIP has the highest Kapp, 35.4 ϩ 4.02 min Ϫ1 g Ϫ1 ; Kapp value for 5-phosphatase type IV is 16.00 ϩ 2.87 min Ϫ1 g Ϫ1 ; and OCRL has the lowest Kapp value, 6.29 ϩ 0.18 min Ϫ1 g Ϫ1 .
The ability of type IV enzyme to hydrolyze PtdIns 3,4,5-P 3 is very sensitive to the presence of detergents in the assay. Whereas the enzyme is capable of hydrolyzing PtdIns 3,4,5-P 3 presented in mixed vesicles without added detergent and in the presence of 0.03% n-octyl ␤-glucopyranoside or Triton X-100, its activity was completely inhibited in the presence of 2 mM CTAB (Fig. 7A). The later detergent is present in standard assays for hydrolysis of PtdIns 4,5-P 2 . Thus, we re-evaluated the ability of type IV enzyme to hydrolyze PtdIns 4,5-P 2 under different assay conditions. PtdIns 4,5-P 2 Hydrolysis-We compared the ability of three enzymes, OCRL, SHIP, and 5-phosphatase type IV to hydrolyze PtdIns 4,5-P 2 with either n-octyl ␤-glucopyranoside or CTAB in the assay. Under standard assay conditions, with 2 mM CTAB, only OCRL effectively converted PtdIns 4,5-P 2 to PtdIns 4-P, with no activity observed using either SHIP or type IV enzyme (Fig. 7C). However, both type IV enzyme and, surprisingly, SHIP were capable of PtdIns 4,5-P 2 hydrolysis when n-octyl ␤-glucopyranoside but not CTAB was added to the assay (Fig.  7B). The ability of SHIP to utilize PtdIns 4,5-P 2 was unexpected, because in all previously reported studies SHIP was able to utilize only the D-3 phosphate-containing substrates. We compared the three 5-phosphatases for their ability to hydrolyze PtdIns 4,5-P 2 . The first order rate constants of OCRL, 5-phosphatase type IV, and SHIP were 63.05 ϩ 10.54 min Ϫ1 g Ϫ1 , 41.99 ϩ 25.07 min Ϫ1 g Ϫ1 , and 29.9 ϩ 18.75 min Ϫ1 g Ϫ1 . Thus the classification of SHIP as an enzyme specific for D-3 phosphate containing substrates is incorrect as it readily cleaves PtdIns 4,5-P 2 under appropriate conditions. Which substrates are most important for either SHIP or type IV enzymes in vivo cannot be determined by these in vitro assays.
Selective sensitivity to detergents has been reported for yeast inositol polyphosphate 5-phosphatases (23). PtdIns 4,5-P 2 -hydrolyzing activity of Inp51p is completely abolished in the presence of 0.2% Triton X-100, whereas 5-phosphatase activity of the two other gene products, Inp52p and Inp53p, is detergent-insensitive. At the same time, polyphosphoinositide 3/4/5phosphatase activities associated with the SACI-like domain of Inp52p and Inp53p is inhibited by detergent. It is possible that detergent sensitivity reflects the sensitivity of the enzymes to their lipid micro-environment in vivo.
Recent years mark tremendous progress in understanding the important role that phosphatidylinositol polyphosphates play in cellular signaling. A role for PtdIns 4,5-P 2 in the regulation of the actin cytoskeleton and vesicular trafficking is well established (reviewed in Ref. 7). There is also a growing body of evidence that PtdIns 3,4,5-P 3 is an important second messenger (reviewed in Ref. 6). Although practically absent in quiescent cells, PtdIns 3,4,5-P 3 is produced in response to cell stimulation predominately by phosphoinositide 3-kinase from PtdIns 4,5-P 2 , or by recently characterized phosphatidylinositol 4/5-kinase from PtdIns 3-P by a concerted reaction (24,25).
PtdIns 3,4,5-P 3 binds with high affinity to PH domains of several proteins, including ARNO, a guanine nucleotide exchange factor for ARF, Grp1 and cytohesin-1, Bruton's tyrosine kinase (Btk), PLC␥, Akt/PTB and it's kinase PDK. PH domains of proteins have varying degrees of binding preferences between PtdIns 3,4,5-P 3 , PtdIns 4,5-P 2 and, in case of Akt/PTB, PtdIns 3,4-P 2. The PH domain of Btk is the most PtdIns 3,4,5-P 3 -specific and binds PtdIns 3,4,5-P 3 with a K d of less than 1 M, about 10 times less than PtdIns 4,5-P 2 . (26). This preferential binding to PtdIns 3,4,5-P 3 must be physiologically important, because a mutation in the Btk PH domain that results in a significant reduction of the selectivity for PtdIns 3,4,5-P 3 binding also causes agammaglobulinemia. An intact PH do- main of Btk and PtdIns 3,4,5-P 3 production enhance phosphorylation and activation of Btk. One of the proposed downstream targets of Btk, PLC␥, also preferentially binds PtdIns 3,4,5-P 3 over PtdIns 4,5-P 2 via its PH domain. Thus, PtdIns 3,4,5-P 3 binding by PH domains of both Btk and PLC␥ may lead to simultaneous translocation of the proteins to the plasma membrane and activation of PLC␥. A somewhat similar dual role of PtdIns 3,4,5-P 3 has been proposed in the activation of Akt in response to mitogenic stimuli (27). PtdIns 3,4,5-P 3 binding to PH domains of Akt and PDK1 brings both proteins to the plasma membrane. In addition, PtdIns 3,4,5-P 3 binding to the PH domain of Akt exposes the Thr-308 residue, which is critical for phosphorylation and activation of Akt.
As with all signaling molecules, PtdIns 4,5-P 2 and PtdIns 3,4,5-P 3 need to be present in the right place at a precise time, so it is of great interest to identify enzymes that control their intracellular levels. One way of control is removal of phosphate at the D-5 position by members of the inositol polyphosphate 5-phosphatase family of enzymes. Synaptojanin, a group II 5-phosphatase, is the key enzyme that mediates vesicular trafficking at the synapses via control of the PtdIns 4,5-P 2 and, possibly, PtdIns 3,4,5-P 3 levels. SHIP, a group III 5-phosphatase, is a negative regulator of signaling initiated by growth factors in hematopoietic cells and Fc␥RIIB negative signaling in B cells, where it regulates the PtdIns 3,4,5-P 3 levels. SHIP2, which is ubiquitously expressed, is likely to play a similar role in nonhematopoietic cells. Cloning and identification of a novel 5-phosphatase with a unique substrate specificity toward phosphatidylinositol polyphosphates will help to elucidate complex regulation of PtdIns 3,4,5-P 3 and PtdIns 4,5-P 2 levels in the cell.