Identification and Functional Characterization of a Presqualene Diphosphate Phosphatase*

Presqualene diphosphate (PSDP) is a bioactive lipid that rapidly remodels to presqualene monophosphate (PSMP) upon cell activation (Levy, B. D., Petasis, N. A., and Serhan, C. N. (1997) Nature 389, 985–990). Here, we have identified and characterized a phosphatase that converts PSDP to PSMP. Unlike the related polyisoprenyl phosphate farnesyl diphosphate (FDP), PSDP was not a substrate for type 2 lipid phosphate phosphohydrolases. PSDP phosphatase activity was identified in activated human neutrophil (PMN) extracts and partially purified in the presence of Nonidet P-40 with gel filtration and anion exchange chromatography. Peptide sequencing of a candidate phosphatase was consistent with phosphatidic acid phosphatase domain containing 2 (PPAPDC2), an uncharacterized protein that contains a lipid phosphate phosphohydrolase consensus motif. Recombinant PPAPDC2 displayed diphosphate phosphatase activity with a substrate preference for PSDP > FDP > phosphatidic acid. PPAPDC2 activity was independent of Mg2+ and optimal at pH 7.0 to 8.0. Incubation of [14C]FDP with recombinant human squalene synthase led to [14C]PSDP and [14C]squalene formation, and in the presence of PPAPDC2, [14C]PSMP was generated from [14C]PSDP. PPAPDC2 mRNA was detected in human PMN, and is widely expressed in human tissues. Together, these findings indicate that PPAPDC2 in human PMN is the first lipid phosphate phosphohydrolase identified for PSDP. Regulation of this activity of the enzyme may have important roles for PMN activation in innate immunity.

Neutrophil (PMN) 3 activation plays a central role in diverse responses such as host defense, inflammation, and reperfusion injury (1). In response to inflammatory stimuli, PMN target reactive oxygen species and granule enzymes to phagolysosomes for microbial killing or digestion of foreign materials (2). Anomalous release of these potentially toxic agents often occurs, amplifying inflammation and leading to tissue injury, events that are implicated in a wide range of human diseases (3). To prevent an overexuberant inflammatory response and limit damage to the host, PMN pro-grams are tightly regulated with select bioactive lipids serving as endogenous anti-inflammatory signals (4).
It is now well appreciated that several enzymes of lipid metabolism also serve as signaling modules with their products acting as bioeffectors (5,6). Polyisoprenyl phosphates, specifically the mevalonate-derived product presqualene diphosphate (PSDP), carry biological activity as an intracellular down-regulatory signal in human PMNs (7). PSDP directly inhibits phospholipase D (PLD) and leukocyte superoxide anion generation in vitro and in vivo (8,9). Recently, the hyperimmunoglobulinemia D and periodic fever syndrome were identified as a systemic inflammatory illness stemming from partial deletion of mevalonate kinase and subsequently decreased isoprenoid production (10,11). Together, these observations indicate that in addition to their roles as structural elements in cholesterol biosynthesis, mevalonate-derived products can also display properties of lipid mediators in inflammation.
Cell activation by receptor-mediated agonists leads to rapid and transient polyisoprenyl phosphate remodeling (7,8). PSDP is present in freshly isolated human PMN. When cells are activated, PSDP shifts from perinuclear to granule and microsomal subcellular domains, and within seconds is converted to its monophosphate form, presqualene monophosphate (PSMP) (7,12). As an inhibitor of PLD and reactive oxygen species generation, PSMP is over 100-fold less potent than PSDP (8). These findings suggest the presence of a regulated diphosphate phosphatase. Here, we report the identification of the first PSDP phosphatase in human PMN and characterize its biochemical properties.
Enzymatically active materials in these fractions were concentrated by spin filtration with a 10-kDa limiting membrane (Centriplus YM-10, Amicon Bioseparations, Millipore, Bedford, MA), and applied to DEAE-Sepharose (DFF100, Sigma) for anion-exchange chromatography. The column (MT20 (Bio-Rad), internal diameter 15 mm, length 11.3 cm) was eluted at 1.0 ml/min for 10 ml with Buffer A and then an additional 40 ml with a linear KCl gradient (0 -1.0 M) was constructed in Buffer A, and 1-ml fractions were collected. Eluents were screened for polyisoprenyl phosphate phosphatase activity. Fractions with PSDP phosphatase activity were combined, concentrated, and electrophoresed by SDS-PAGE with 12% polyacrylamide. The gel was stained for protein with Coomassie Brilliant Blue (Bio-Rad), and protein in the expected M r range was excised. After proteolytic digestion, sequence analysis was performed at the Harvard Microchemistry Facility by microcapillary reverse-phase high performance liquid chromatography nano-electrospray tandem mass spectrometry on a Finnigan LCQ quadrapole ion trap mass spectrometer.
Assay for Phosphatase Activity-Phosphatase activity was determined by phosphate release using malachite green reagent (Biomol Green kit, Biomol Research Laboratories, Inc., Plymouth Meeting, PA). Each reaction was performed in buffer containing 50 mM HEPES (pH 7.4), 80 mM KCl, 1 mM dithiothreitol, 3 mM EDTA (100 l total volume). Materials were incubated for 30 min at 37°C. Each reaction was stopped with 200 l of malachite green reagent and incubated an additional 20 min (room temperature). Absorbance at 630 nm was measured with a 96-well microplate reader (Quant, Bio-Tek Instruments, Inc., Winooski, YT).
Expression of Recombinant PPAPDC2-His 6 in Sf21 Cells-The human PPAPDC2 cDNA was generated by RT-PCR from human PMN, inserted into the EcoRI site of pCR4-TOPO (Invitrogen, Palo Alto, CA) and verified by direct sequencing. A DNA fragment containing the coding sequence of the PPAPDC2 gene flanked by BamHI and XhoI was obtained by PCR (forward primer, 5Ј-AGTCCCGGATCCCGGAG-GAGCATGGAG-3Ј; reverse primer, 5Ј-CTCGAGTCAATGGTGAT-GGTGATGATGTCTAGATCGTTGACTCCACAGTAA-3Ј) with Pfu polymerase using a cDNA plasmid encoding PPAPDC2 as a template. To generate an additional His 6 sequence at the carboxyl terminus of PPAPDC2, the reverse primer contained a 21-mer artificial sequence that codes His 6 and a stop codon (indicated as underlined). The PCR product was digested with BamHI and XhoI (New England Biolabs, Inc., Beverly, MA), ligated into the same restriction enzyme sites of pBacPAK9 baculovirus vector, and expressed in insect cells using the BacPAK Baculovirus Expression System (Clontech, San Jose, CA). Sf21 cells were infected with a viral multiplicity of 5 and harvested after 72 h. Cells were suspended in 50 mM HEPES (pH 7.4), 80 mM KCl, 1ϫ protease inhibitor mixture (Complete, Roche), and lysed by sonication (three 10-s pulses). Unbroken cells and cellular debris were removed by centrifugation (1,000 ϫ g, 10 min, 4°C). Membrane protein was then sedimented (10,000 ϫ g, 60 min, 4°C) and pellets were solubilized in 50 mM HEPES (pH 7.4), 80 mM KCl, 20 mM imidazole, and 0.1% Triton X-100. Partially purified enzyme was obtained through a Ni 2ϩ -charged column (Amersham Biosciences). For phosphatase activity, PSDP, PA (C10:0), FDP, or DGPP (C18:1) (10 M) were incubated with rhP-PAPDC2 (2 g). Prior to select incubations, rhPPAPDC2 was boiled (20 min).
Squalene Synthesis Assay-Squalene synthase activity was determined in 100 mM potassium phosphate, 5 mM MgCl 2 , 5 mM CHAPS, 10 mM dithiothreitol, [ 14 C]FDP (0.25 Ci) in 23 M FDP and 2 g of rhSQS protein in a total reaction volume of 100 l. Some reactions were performed in the presence of 2 mM NADPH to promote PSDP conversion to squalene. For determination of polyisoprenyl phosphate phosphatase activity, PAP2a (5 g) or PPAPDC2 (2 g) were co-incubated with rhSQS (2 g) in reaction buffer. Enzyme and substrate were incubated for up to 2 h at 37°C, after which the reactions were stopped by the addition of 10 l of 1 M EDTA (pH 9.2) (15). Each reaction mixture (10 l) was analyzed by thin layer chromatography. Silica Gel TLC plates (Fisher) were developed with chloroform/methanol/water (65:25:4, v/v/ v). The radioactivity of each spot was visualized and quantified with a Bio-imaging analyzer (model 425E; Amersham Biosciences, with ImageQuant software version 3.2). After imaging, densitometry was performed using Scion Image software.
RNA Dot Blot Hybridization of PPAPDC2 RNA-Hybridization of a 600-bp fragment encoding the open reading frame of PPAPDC2 to a human RNA multiple tissue expression array (MTE array, Clontech) was performed following the manufacturer's instructions. Primers used to generate the human PPAPDC2 probe were 5Ј-CCCTTATGAAGCT-GCTGGAG-3Ј (forward primer) and 5Ј-GGCTGCTGGTTTAGGTT-GAC-3Ј (reverse primer). The identity of the probe was confirmed by direct sequencing.
Statistical Analysis-Data were provided as the mean Ϯ S.E. Comparisons of results were conducted by two-way analysis of variance followed by post-hoc Scheffe's test. p Ͻ 0.05 was considered significant.

RESULTS
Activated PMNs display polyisoprenyl phosphate phosphatase activity by converting PSDP to PSMP (Fig. 1A), so freshly isolated human PMNs were exposed to leukotriene B 4 (100 nM, 60s, 37°C), lysed by sonication, and nuclei and remaining intact cells were pelleted. Supernatants were applied to Sephacryl S200-HR for gel filtration and 3-ml eluents were collected and assayed for polyisoprenyl phosphate phosphatase activity using FDP (see "Experimental Procedures"). We chose to screen the column fractions with FDP, because it is closely related to the structure of PSDP and quantities of PSDP were more limited. FDP phosphatase activity eluted with proteins of an estimated molecular mass of ϳ35 kDa. To further purify the PMN proteins with PSDP phosphatase activity, we next pooled the catalytically active column fractions after gel filtration, concentrated the fractions by spin filtration with a 10-kDa limiting membrane, and then applied the materials to DEAE-Sepharose for anionexchange chromatography (see "Experimental Procedures"). The column eluents were again screened for polyisoprenyl phosphate phosphatase activity with FDP and a broad peak of activity was identified over 5 separate 1-ml column fractions (Fig. 1B). For each fraction, we next compared FDP to PSDP phosphatase activity and determined that fractions from the leading edge (fractions 27 and 28) had significantly greater phosphatase activity for FDP than PSDP when assayed in parallel (Fig. 1C). In sharp contrast, fractions 31 and 32 from the trailing edge of the peak displayed a preference for PSDP as a substrate. The materials in fractions 27 and 28 and fractions 31 and 32 were combined, separated by 12% SDS-PAGE, and Coomassie Blue staining revealed a prominent band of protein at ϳ35 kDa (Fig. 1D). Type 2 phosphatidic acid phosphohydrolase (PAP2) isoenzymes are ϳ35 kDa, involved in lipid signaling, and can dephospho-rylate FDP (16). To determine whether isoforms of PAP2 also displayed PSDP phosphatase activity, we first expressed human PAP2a and incubated recombinant protein (30 min, 37°C) with PSDP, PA (C10:0), FDP, and DGPP (C18:1)(10 M). We confirmed the ability of PAP2a to dephosphorylate a broad range of substrates, including PA, FDP, and DGPP (1.91 Ϯ 0.45, 2.72 Ϯ 0.7, and 3.17 Ϯ 0.11 nmol/min/mg, respectively, Fig. 1E). In sharp contrast, phosphate release from PSDP was significantly less efficient (0.20 Ϯ 0.03 nmol/min/mg, p Ͻ 0.01). In separate experiments, rhPAP2b and rhPAP2c were also expressed and similar to PAP2a, these isoforms did not have significant phosphatase activity for PSDP (data not shown). These results indicated that PSDP, unlike shorter isoprenoid diphosphates, was not a substrate for type 2 lipid phosphate phosphohydrolases.
To identify the candidate PSDP phosphatase, enzymatically active materials for PSDP were extracted from the SDS-PAGE gel (Fig. 1D). After proteolytic digestion, peptide sequencing uncovered a partial amino acid sequence, AGPSQSXXXXXXEXXR, that was found in phos- phatidic acid phosphatase type 2 domain containing 2 (PPAPDC2) (GenBank accession number BC038108), an uncharacterized but closely related protein to PAP2A (Fig. 2A). PPAPDC2 encodes a protein of 295 amino acids with an anticipated molecular mass of 31.5 kDa. The PPAPDC2 sequence displayed an overall 17% identity and 18% similarity to human PAP2A ( Fig. 2A). Of note, PPAPDC2 had substantially greater homology with PAP2A in three conserved domains (C1, C2, and C3) that comprise a putative lipid phosphate phosphatase sequence motif, K(X) 6 RP-(X 12-54 )-PSGH-(X 31-54 )-SR(X) 5 H(X) 3 D (Fig. 2, A and B). This motif is shared among several yeast and mammalian lipid phosphate phosphatases (17,18). Kyte-Doolittle analysis of the overall hydropathy profile of the protein predicted a structure for PPAPDC2 similar to PAP2a with six membrane-spanning domains in the encoded protein (data not shown). Because these structural features suggested that PPAPDC2 had the ability to serve as a new member of this class of enzymes, we next determined its relationship by evolutionary hierarchy to PAP2a and other proteins with elements of the conserved PAP2 phosphatase motif (Fig. 2C). The resulting phylogenetic tree suggested that PPAPDC2, like Dolpp1 (19), would have enzymatic properties distinct from PAP2 isoforms. Investigation of cellular extracts from human PMN and HEK 293 cells indicated that these diverse cell types both carried PSDP phosphatase activity, so we next determined which of the lipid phosphate phosphatases in Fig. 2C were expressed in both cell types. In addition to PAP2A, mRNA for only PPAPDC2 and LPPRP2 was detectable by RT-PCR in both cell types (Fig. 2D). Together, these findings indicated that PPAPDC2 was a candidate PSDP phosphatase.
To investigate whether PPAPDC2 or LPPRP2 had PSDP phosphohydrolase activity, Sf21 cells were transfected with baculovirus vectors for expression of recombinant protein. In membrane fractions with rhP-PAPDC2, a band with an apparent molecular mass of 31.5 kDa was specifically detected, and the enzyme was partially purified (Ͼ90%) with a Ni 2ϩ -charged column (Fig. 3A). Recombinant LPPRP2 was similarly expressed and purified (data not shown). PSDP phosphatase activity for PPAPDC2 was dependent on substrate concentration (Fig. 3B). No PSDP phosphatase activity was present with rhLPPRP2 (data not shown). PPAPDC2 demonstrated a marked substrate preference for PSDP over PA (C10:0) (Fig. 3, B and C). In these reactions, PPAPDC2 exhibited ϳ5-fold more phosphate release from PSDP than equimolar amounts (10 M) of the other substrates tested with a rank order of PSDP Ͼ FDP Ͼ PA Ͼ DGPP (Fig. 3C). This structure-activity relationship for PPAPDC2 is in sharp contrast to that observed with rhPAP2a (Fig. 1E).
We next examined the effect of pH, divalent cations, and detergent on PPAPDC2 phosphatase activity. The impact of pH on PSDP phosphatase activity was determined at equimolar concentrations of PSDP (10 M) (Fig. 4A). PPAPDC2 exhibited a pH optimum between 7.0 and 8.0, so activity was routinely measured at pH 7.4. We also determined the impact of divalent cations on PPAPDC2 activity. PSDP phosphatase activity was independent of Mg 2ϩ , Mn 2ϩ , Ca 2ϩ , and Zn 2ϩ concentrations between 1 and 10 mM (data not shown). When PSDP phosphatase activity was determined with PPAPDC2 in the presence of 0.1-1% Triton X-100/phospholipid-mixed micelles, the activity was maximal at 0.1% Triton X-100 (data not shown). Experiments using other detergents, including Nonidet P-40 and ␤-octyl glucoside, at concentrations of 0.1-1% were also performed, but none proved superior to 0.1% Triton X-100.
Because the amounts of FDP present in these incubations exceeded the rate-limiting concentrations for both rhSQS and rhPAP2a, their coincubation did not significantly affect the rates of PSDP or squalene synthesis (lane 7). Although PSDP was generated by rhSQS, no PSMP was detected in the presence of PAP2a. In the absence of rhSQS, PAP2a converted FDP to FMP (18.3%) and farnesol (5.0%) without formation of PSDP, PSMP, or squalene (lane 8). Together, these findings confirmed that PPAPDC2 was a polyisoprenyl phosphate diphosphate phosphatase (Fig. 5B), and that PAP2a did not utilize PSDP as a substrate. In addition, PAP2a also differs in its biochemical properties from PPAPDC2 in its ability to dephosphorylate both polyisoprenyl diphosphates and monophosphates.
To determine the pattern of expression for PPAPDC2 mRNA, we performed RNA blot hybridization with a human tissue RNA expression array. In addition to human PMN and HEK 293 cells, PPAPDC2 was also expressed in most organs, in particular gastrointestinal organs, spleen, placenta, kidney, thymus, and brain (Fig. 6A). Semi-quantitative RT-PCR revealed PPAPDC2 expression in human PMN that was ϳ60% of levels in human liver (Fig. 6B).

DISCUSSION
Exposure of human PMN to receptor-mediated stimuli triggers the rapid conversion of PSDP to PSMP (7), indicating activation of a PSDP diphosphate phosphatase. Polyisoprenyl phosphate remodeling in PMN temporally overlaps intracellular signaling events (e.g. PLD activation) and functional responses (e.g. O 2 . production) (8,12). PSDP is over 100-fold more potent than PSMP as an inhibitor of PLD and PMN NADPH oxidase assembly (7,8), and structural mimetics of PSDP can decrease PMN trafficking and activation in vivo during acute inflammatory responses (9). Together, these findings suggest that decrements in cellular PSDP levels by a PSDP phosphatase would play a pivotal role in regulating leukocyte function.
Here, we have identified PPAPDC2 as the first PSDP phosphatase from human PMN and characterized the biochemical properties of this member of the lipid phosphatase/phosphotransferase (LPT) family. LPTs comprise a large multigene family with at least five distinct groups, namely lipid phosphate phosphatases (LPPs), sphingosine-1-phosphate phosphatases, sphingomyelin synthase, lipid phosphatase-related proteins, and a group that was recently provisionally labeled as candidate sphingomyelin synthases type 2 (CSS2s) because of sequence homology with sphingomyelin synthases (21,22). PPAPDC2 (GenBank accession number BC038108) shares a conserved acid phosphatase domain with LPT family members, and close inspection of its sequence revealed its identity as the newly labeled CSS2␤ (SwissProt/TrEMBL accession number Q8IY26) (21). Similar to the LPPs, PPAPDC2 is predicted by hydropathy analysis to be an integral membrane protein and possess six transmembrane ␣-helices. Despite its name, this protein does not exhibit sphingomyelin synthase activity (21) and up until this report its enzymatic properties have remained incompletely characterized (21,22). CSS2␤ (i.e. PPAPDC2) has an entirely conserved phosphatase motif and here displayed lipid phosphate phosphatase activity in vitro for more than one polyisoprenyl phosphate substrate. Outside the con-served phosphatase motif, PPAPDC2 is structurally distinct from LPPs, in particular it has a unique NH 2 terminus of ϳ100 amino acids. The biochemical properties of PPAPDC2 were also distinct from LPPs, as PPAPDC2 had a substrate preference for PSDP with a rank order of PSDP Ͼ FDP Ͼ PA Ͼ DGPP, differing markedly from PAP2a, PAP2b, and PAP2c. These LPPs did not utilize PSDP as a substrate for phosphate release and displayed phosphatase activity for an otherwise broad range of substrates. In addition, PPAPDC2 carried only polyisoprenyl diphosphate, not monophosphate, phosphatase activity, as opposed to PAP2a that demonstrated properties of both a di-and monophosphate phosphatase. The unique preference of PPAPDC2 for diphosphates also differs from the NH 2 terminus of soluble epoxide hydrolase that, similar to PAP2a, carries phosphatase activity for both polyisoprenyl di-and monophosphates (23).
In addition to its enzymatic activity, PPAPDC2 RNA and protein expression were evident in human PMN. In freshly isolated cells, the largest quantities of PSDP are present in PMN nuclear and granule subcellular fractions (12). Upon exposure to leukotriene B 4 , a PMN chemoattractant and secretagogue, PSDP levels rapidly decrease in nuclear and increase in granule-enriched fractions. Of interest, a subtractive proteomic analysis suggests that PPAPDC2 (i.e. CSS2␤) localizes in part to liver nuclear membranes (24). If also present in the PMN nuclear envelope, this phosphatase would be well positioned to effect the PSDP remodeling observed upon cell activation. Here, the PSDP phosphatase activity was present in PMN and HEK 293 cell 1000 ϫ g supernatants and 10,000 ϫ g pellets, indicating that PPAPDC2 is likely to also be present in microsomal and granule-enriched fractions in these cells, similar to other LPTs (22).
PPAPDC2 displayed a wide tissue distribution. Expression profiles obtained for CSS2␤ from online data of expressed sequence tag transcript abundance similarly predict a broad mammalian tissue distri-  4). B, SQS catalyzes the formation of PSDP from FDP, and in a second reaction, PSDP, in the presence of NADPH, is converted to squalene. PSDP is also a substrate for PPAPDC2 that converts PSDP to PSMP in an NADPH-independent manner. FIGURE 6. Expression of PPAPDC2 in human tissues. A, a dot blot containing mRNA from multiple human tissues was probed with a 600-bp fragment from human PPAPDC2 to determine the distribution of its expression. Relative RNA expression was measured by densitometry. B, semiquantitative RT-PCR for PPAPDC2 and ␤-ACTIN expression in human PMN and liver. bution for this gene (22). In addition, the relative abundance of PPAPDC2 RNA, like other LPTs (22), was highest in gastrointestinal organs, placenta, and brain. Of interest in the regulation of inflammatory responses, there was also significant PPAPDC2 expression in immune organs (namely spleen and thymus) that are critical for myeloid, dendritic cell, and B-and T-lymphocyte development, survival and trafficking. Like PPAPDC2, sphingosine-1-phosphate phosphatases are expressed in the thymus and leukocytes (22), and sphingosine 1-phosphate and its cognate receptors can comprise autocrine and paracrine signaling networks in the regulation of leukocyte responses (25).
Inflammation and metabolic pathways for sphingosine 1-phosphate and other LPT substrates have been linked to sterol metabolism (26). PMN are innate immune effectors that are uniquely unable to generate sterols from endogenous sources of mevalonate (27). These cells lack squalene epoxidase and other mixed function oxidases required for sterol biosynthesis, leading to a biosynthetic block at SQS. Because PSDP forms an enzyme-intermediate complex in the SQS catalyzed two-step conversion of FDP to squalene, we investigated whether this polyisoprenyl phosphate was available for interactions with other proteins by co-incubating rhSQS with rhP-PAPDC2. Our results provide direct evidence for PSDP conversion to PSMP in these incubations. Isoprenoid regulatory proteins have been identified at several steps in cholesterol biosynthesis to regulate flux through the pathway (28,29) or divert biosynthetic intermediates to other functions (30). Our results indicate that PSDP is available in vitro for metabolic fates other than squalene and suggest that PSDP could interact with intracellular protein targets, such as PLD and SH2 domains (31), in the regulation of PMN functional responses.
In summary, this is the first identification of a PSDP phosphatase. Identified by functional characterization and direct sequencing, PPAPDC2 (aka CSS2␤) is an LPT with polyisoprenyl diphosphate phosphatase activity and is expressed in human leukocytes and multiple human tissues. Moreover, this is the first enzymatic activity identified for a member of the CSS2 group of LPTs. Based on its newly identified biochemical properties, conservation of the LPP phosphatase motif and lack of sphingomyelin synthase activity, we propose that this protein be renamed as distinct LPT (e.g. polyisoprenyl diphosphate phosphatase 1). Regulation of the activity of the enzyme is predicted to have important roles in leukocyte function during immune responses.