Specific interaction of protein tyrosine phosphatase-MEG2 with phosphatidylserine.

Protein tyrosine phosphatase (PTP)-MEG2 is an intracellular tyrosine phosphatase that contains a Sec14 homology domain. We have purified the full-length and truncated forms of the enzyme from recombinant adenovirus-infected human 293 cells. By using lipid-membrane overlay and liposome binding assays, we demonstrated that PTP-MEG2 specifically binds phosphatidylserine among over 20 lipid compounds tested. The binding is mediated by its N-terminal Sec14 domain. In intact cells, the Sec14 domain is responsible for localization of PTP-MEG2 to the perinuclear region, and uploading of PS into the cell membrane causes translocation of PTP-MEG2 to the plasma membrane. Phosphatidylserine is a relatively abundant cell membrane phospholipid non-symmetrically distributed in the outer layer and inner layer of cell membranes. It has recently been defined as an important ligand for clearance of apoptotic cells. By specifically binding phosphatidylserine, PTP-MEG2 may play an important role in regulating signaling processes associated with phagocytosis of apoptotic cells.

Protein tyrosine phosphatases (PTPs) 1 consist of a highly diverse family of enzyme with crucial roles in cell signaling (1)(2)(3). Like their counterpart protein tyrosine kinases, PTPs can be divided into transmembrane receptor-like and intracellular enzymes. The intracellular PTPs contain a single conserved phosphatase domain. Beyond the limit of the catalytic domains are various segments or domains that modulate the activity and/or intracellular localization of the enzymes. Among these is the Sec14 domain found in PTP-MEG2 (4). The Sec14 domain is shared by several lipid-binding proteins such as cellular retinaldehyde-binding protein (CRALBP), ␣-tocopherol transfer protein, and yeast Sec14p and by regulators of G proteins including RhoGAPs, RhoGEFs and the RasGEF, and neurofibromin (5). CRALBP is a water-soluble protein found in the retina and pineal gland. It acts as a carrier protein for 11-cis-retinaldehyde or 11-cis-retinol and modulates the inter-actions of these retinoids with visual cycle enzymes (6). ␣-Tocopherol transfer protein specifically sorts out ␣-tocopherol from all incoming tocopherols for incorporation into plasma lipoproteins (7). Sec14p acts as a phosphatidylinositol transfer protein by catalyzing the transfer of phosphatidylinositol and phosphatidylcholine between membrane bilayers, and it is required for protein transport through the Golgi complex in Saccharomyces cerevisiae (8,9). With significant sequence homology to these lipid-binding proteins, PTP-MEG2 may also bind certain lipid molecules that may regulate its activity and/or localization. A ligand for the Sec14 domain of PTP-MEG2 has not been found. In this study, we examined the interaction of PTP-MEG2 with various known lipid molecules and demonstrated that phosphatidylserine (PS) specifically interacts with PTP-MEG2. PS is a relatively abundant membrane lipid molecule. Recently, it has been defined as an important ligand for clearance of apoptotic cells through the PS receptor (PSR) located on the surface of phagocytes (10 -12). Recognition of surface PS by PSR initiates uptake of the apoptotic cells but does not induce inflammatory responses (12,13). By specifically interacting with PS, PTP-MEG2 may have an important role in regulating cell-signaling pathways in this process.

EXPERIMENTAL PROCEDURES
Materials-Commercial PIP strips spotted with 100 pmol of 16 different phospholipids were purchased from Echelon Research Lab (Salt Lake City, UT). Individual lipid compounds including phosphatidic acid, PC, PE, phosphatidylglycerol, phosphatidylinositol, PS, sphingomyelin, and cholesterol, all from natural sources, were purchased from Avanti Polar Lipids. Aranchidonic acid, oleic acid, linoleic acid, alltrans-retinal, and ␣-tocopherol were from Sigma. A mixture of retinal isomers was made by photoisomerization of all-trans-retinal as previously described (14). Homemade lipid strips were prepared by spotting on nitrocellulose membranes 1 l of individual lipid compounds dissolved in chloroform/methanol (1:1 v/v) at a concentration of 0.01-1 mg/ml. The strips were air-dried and kept at 4°C in the dark. Anti-PTP-MEG2 antibodies 144 and 159 were raised against peptides corresponding to amino acids 575-593 at the C terminus and to amino acids 297-314 in the middle of enzyme, respectively.
Expression and Purification of Full-length and Truncated Forms of PTP-MEG2-Three recombinant PTP-MEG2 proteins were employed in this study (see Fig. 1). They are full-length form PTP-MEG2 (amino acid residues 1-593), N-terminal Sec14 domain-truncated form ⌬NPTP-MEG2 (amino acid residues 283-593), and C-terminal catalytic domain-truncated form ⌬CPTP-MEG2 (amino acid residues 1-329). PTP-MEG2 and ⌬NPTP-MEG2 were purified by using adenovirus and Escherichia coli expression systems, respectively, as previously described (14). ⌬CPTP-MEG2 was also expressed in the adenovirus. It contains a His 6 tag at the C terminus without addition of any other extra amino acid residues. To generate the recombinant adenovirus carrying ⌬CPTP-MEG2, a cDNA fragment encoding amino acids 1-329 of PTP-MEG2 plus a DNA sequence encoding 6 consecutive histidine residues was inserted into adenovirus transfer vector pACCMV.pLpA. Recombinant adenovirus was generated by co-transfection of 293 cells with the pACCMV.pLpA construct and pJM17 adenovirus genome DNA by using FuGENE 6 cell transfection reagent (Roche Applied Science). * This work was supported by Grants CA75218 (to Z. J. Z.), DK-15555 (to S. B. K.), and CA-68485 (to Vanderbilt-Ingram Cancer Center) from the National Institutes of Health and a Veterans Health Administration Merit Review grant (to S. B. K.). 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 resulting recombinant virus was purified by soft agar plaque assays and then amplified in 293 cells according to standard procedures (15). Positive clones were selected based on expression of ⌬CPTP-MEG2 in infected 293 cells as determined by anti-PTP-MEG2 antibody 159. The purified recombinant adenovirus was used to infect 293 cells to express ⌬CPTP-MEG2 following the protocol used for expression of the full-length PTP-MEG2 (14). The His 6 -tagged recombinant protein was purified from the cytosolic extracts of the infected cells by using a nickel-nitrilotriacetic acid column (Qiagen) following the manufacturer's protocol. The purity of the enzyme was about 90% as judged by Coomassie Blue staining. Western blotting analyses with antibody 159 showed single band of expected size (ϳ45 kDa).
Lipid-membrane Overlay Assays-Nitrocellulose membrane strips spotted with various lipid compounds were incubated with 0.5 g/ml full-length or truncated forms of PTP-MEG2 in Buffer A containing 10 mM Tris-HCl (pH 8.0), 0.15 M NaCl, 0.1% Tween 20, and 3% fatty acid-free bovine serum albumin (BSA). After extensive washing with Buffer A, enzymes bound to the membrane strips were probed with appropriate anti-PTP-MEG2 antibodies and then with horseradish peroxidase-conjugated anti-rabbit secondary antibodies. Detection was made by enhanced chemiluminescence reactions followed by exposure to x-ray film. The control experiments were performed without addition of any enzymes under the same conditions.
Liposome Binding and Phosphatase Activity Assays-For liposome preparation, lipid mixtures in chloroform containing 320 g PC and 80 g PE plus 100 g of indicated lipids were dried by using a speed vacuum concentrator and then were sonicated with a water bath sonicator in 0.5 ml of Buffer B containing 25 mM HEPES-NaOH (pH 7.0), 1 mM dithiothreitol, 0.15 M NaCl. The suspensions were centrifuged for 15 min at 14,000 rpm in a microcentrifuge. The pellets were resuspended by in 0.5 ml of Buffer B supplemented with 0.5 mg/ml BSA. For binding assays, 50 l of the lipid vesicles were mixed with 2 g of PTP-MEG2, ⌬NPTP-MEG2, and ⌬CPTP-MEG2 made in 20 l of Buffer B plus 0.5 mg/ml BSA. After 15 min of incubation at room temperature, the samples were centrifuged at 14,000 rpm in a microcentrifuge, and both pellets and supernatants were used to make SDS gel samples. Equal proportions of the supernatants and pellets were subjected to Western blotting analyses with appropriate anti-PTP-MEG2 antibodies. For phosphatase activity assays, 2 mM para-nitrophenyl phosphate (pNPP) was added to the mixture of PTP-MEG2 and lipid vesicles before centrifugation. The reactions were allowed to proceed at room temperature for 20 min and were stopped by addition of 800 l 0.2 M NaOH. Absorbance at 410 nm was measured and contributions by the lipid vesicles (ϳ10 -20%) were subtracted in calculating the phosphatase activity.
Transient Expression of Full-length and Truncated Forms of PTP-MEG2 in HT-1080 Cells and Immunofluorescent Cell Staining-HT-1080 fibrosarcoma cells seeded on glass cover slips were transfected with a pCDNA3 construct carrying N-terminal Sec14 domain-truncated ⌬NPTP-MEG2 in the presence of the FuGENE 6 transfection reagent or infected with adenoviruses carrying full-length PTP-MEG2 or catalytic domain-truncated ⌬CPTP-MEG2. After 48 h, the cells were fixed with 4% formaldehyde, permeabilized with 0.2% Triton X-100, and incubated with anti-PTP-MEG2 antibodies 144 or 159 and then with a Cy3conjugated anti-rabbit secondary antibody. Cover slips were mounted onto glass slides and visualized by using a Zeiss Axiophot microscope with a 100ϫ oil-immersion objective. Cell images were captured with a cooled CCD digital camera.
Loading of PS into Cell Membranes-To verify the interaction of PS with PTP-MEG2 in intact cells, we followed a well established protocol to load excess amounts of PS into cell membranes (16). For this purpose, unilayer lipid vesicles in phosphate-buffered saline containing 5 mg/ml PC (for controls) or a mixture of 2.5 mg/ml PC and 2.5 mg/ml PS, were prepared by sonication with a water bath sonicator as described above. The lipid vesicles were then added to PTP-MEG2 adenovirus-infected HT-1080 cells deprived of culture medium. Following 1 h of incubation at 37°C, free lipid vesicles were removed by washing the cells with serum-containing medium. The cells then were incubated in normal tissue culture medium at 37°C for 2 h to allow translocation of loaded PS from the outer leaflet to the inner leaflet by the aminophospholipid translocase. To detect the localization of PTP-MEG2, cells were fixed directly onto the tissue culture plates with 4% formaldehyde and stained with anti-PTP-MEG2 antibody 144 followed by probing with a Cy3-conjugated anti-rabbit secondary antibodies as described above. Fluorescent cell images were visualized by using an inverted microscope with a 40ϫ objective (Nikon) and captured with a cooled CCD digital camera.

PTP-MEG2 Specifically
Interacts with PS-To study the binding of PTP-MEG2 with known lipids, we employed a lipidmembrane overlay assay that has become a very popular method to analyze lipid-protein interactions in recent years (17)(18)(19)(20)(21). We started with the PIP strips purchased from Echelon Research Lab. The strips were spotted with 100 pmol of 16 different phospholipids. Following the protocol recommended by the manufacturer, we probed the membrane with purified PTP-MEG2 at a final concentration of 0.5 g/ml. As shown in Fig. 2, A and B, among the 16 phospholipids analyzed, PTP-MEG2 bound specifically to PS. In control experiments without PTP-MEG2, no signal was seen with PS despite a few weak and nonspecific signals seen with several other lipids. We further verified the results by using homemade phospholipid strips. In addition to some common phospolipids, we also included fatty acids, cholesterol, ␣-tocopherol, and retinal isomers. The data shown in Fig. 2C further confirmed the specific interaction of PTP-MEG2 with PS and ruled out binding with other lipid molecules tested. Further experiments as illustrated in Fig. 2D revealed that PTP-MEG2 could easily detect 25 ng of PS on the membrane.
The Sec14 Domain Is Responsible for the Interaction of PTP-MEG2 with PS-The Sec14 domain of PTP-MEG2 is presumably involved in the interaction of PTP-MEG2 with PS. To verify this, we analyzed the binding of catalytic domain-truncated and Sec14 domain-truncated enzymes. ⌬NPTP-MEG2 was purified from E. coli cells as previously described (14). However, ⌬CPTP-MEG2 was largely insoluble when expressed in E. coli cells. We thus employed the adenovirus system for expression of the Sec14 domain as described for the full-length PTP-MEG2. The protein was expressed as a His 6 -tagged fusion protein. Over 70% of ⌬CPTP-MEG2 was partitioned in the cytosolic extracts of recombinant adenovirus-infected 293 cells. By using a single nickel-nitrilotriacetic acid agarose column, we were able to enrich the recombinant protein from the cytosolic extracts to over 90% purity as revealed by Coomassie Blue staining, and Western blotting with anti-serum 159 showed a single protein band (Fig. 3). Fig. 4 shows the results of lipid membrane overlay assays obtained with the truncated enzymes. Although ⌬CPTP-MEG2 displayed a specific binding to PS as observed with the full-length form of PTP-MEG2, ⌬NPTP-MEG2 did not show any significant interaction with any of the lipids analyzed. These data indicate that the Sec14 domain mediates the interaction of PTP-MEG2 with PS.
PS-containing Liposomes Bind PTP-MEG2 but Do Not Affect its Enzymatic Activity toward pNPP-The interaction of PS with PTP-MEG2 was also confirmed by performing liposomebinding assays (22,23). We used PC and PE as carriers for the lipids to be analyzed, and lipid vesicles were formed by sonication. For the recombinant PTP-MEG2 proteins, we employed BSA as a carrier, which is necessary to reduce the nonspecific binding of proteins to lipid vesicles. Upon incubation of the full-length and two truncated forms of PTP-MEG2 with the lipid vesicles followed by centrifugation, proteins bound to the lipid vesicles were recovered from the pellets. As shown in Fig.  5A, among the 8 phospholipids analyzed, only PS displayed a specific and strong binding to PTP-MEG2 as indicated by a near total recovery of the enzyme in the lipid vesicle pellet and a near depletion of the enzyme from the supernatant. All the other lipids showed marginal basal binding. As expected, a similar binding with PS was seen with ⌬CPTP-MEG2 but not with ⌬NPTP-MEG2. These data confirm not only the specific interaction of PTP-MEG2 with PS but also reveal the role of its Sec14 domain in the interaction. We further analyzed the effect of PS on the activity of PT-MEG2 with pNPP as a substrate (Fig. 5B), but we did not find any significant effect of the lipid on the catalytic activity of PTP-MEG2. If anything, it caused a slight inhibition, which was also seen with ⌬NPTP-MEG2, which carries only the catalytic domain. Therefore, PS does not affect the activity of PTP-MEG2 at least toward pNPP. These results agree with our previous study in which the assays were performed with lipid micelle containing Triton X-100 (14). However, these data do not rule out the possible effects on PTP-MEG2 activity toward physiological substrates. By using the same procedures, we have analyzed binding and regulatory ability of other lipid molecules as described in Fig. 2. None of them showed any significant binding or regulatory activity (data not shown).
The Sec14 Domain Is Responsible for the Localization of PTP-MEG2 to the Perinuclear Region-We performed immunofluoresecent cell staining to determine the intracellular localization of PTP-MEG2 as shown in Fig. 6. Staining of wild type HT-1080 cells with antibody 144 revealed a moderate expression of endogenous PTP-MEG2 in the perinuclear region resembling the endoplasmic reticulum and Golgi apparatus. This localization is further confirmed by a much more intensive staining of over-expressed PTP-MEG2 in the same region upon infection of the cells with recombinant adenovirus carrying the full-length form of PTP-MEG2. Furthermore, ⌬CPTP-MEG2 that contains the Sec14 domain alone displayed exactly the same staining pattern as seen with full-length PTP-MEG2. In contrast, truncation of the Sce14 domain resulted in the distribution of ⌬NPTP-MEG2 in the entire cytoplasma. Note that ⌬NPTP-MEG2 corresponds to the catalytic domain of PTP-MEG2. These data indicate that the N-terminal Sec14 domain of PTP-MEG2 is necessary and sufficient for the perinuclear localization of the enzyme. We believe that PS present in the perinuclear region may be responsible for such localization.
Loading of PS into the Cell Membrane Alters Localization of PTP-MEG2-Translocation is one of the major mechanisms by which an enzyme or protein is regulated. Although PS does not change the catalytic activity of PTP-MEG2 (toward pNPP at least), it may regulate the enzyme by controlling its intracellular localization. To study the effects of PS on the localization, cells were loaded with lipid vesicles containing 50% PS and 50% PC according to a published protocol (16). Lipid vesicles made of 100% PC were used as control. The data are shown in Fig. 7. In non-treated cells and cells treated with control PC lipid vesicles, PTP-MEG2 was localized exclusively in the perinuclear region. However, loading of cells with PS caused partial localization of PTP-MEG2 to the plasma membrane. This translocation of PTP-MEG2 to the plasma membrane depended on incubation time after treatment with PS. Plasma membrane localization was apparent 2 h after PS treatment but totally disappeared 8 h later. This presumably reflects the redistribution of PS in the cells. Together, the data demonstrate a binding of PTP-MEG2 with PS in intact cells and an important role for PS in controlling localization of PTP-MEG2. It should be pointed out that the intracellular balance of PS may be the determinant factor for the localization of PTP-MEG2. Under normal conditions, PTP-MEG2 may be retained in perinuclear region by PS there, and the amount of PS present in the inner layer of the plasma membrane is apparently not sufficient to cause a translocation of the enzyme. We believe that a perturbation of PS distribution associated with cellular processes should target PTP-MEG2 to specific membrane compartments.

DISCUSSION
It has been well accepted that phospholipids are not only components of cell membranes but also act as important second messengers (24). A major characteristic of lipid second messengers is that they cannot diffuse freely through the aqueous compartments of cells. Once produced in a particular cell membrane, they stay there and often serve as markers for the membrane compartment until they are enzymatically converted or translocated or the membrane buds off or fuses with another. This membrane confinement makes lipid messengers ideal regulators of subcellular localization. Many proteins are localized to specific cellular compartments through conserved domains that specifically recognize particular membrane-  (17)(18)(19)(20)(21)(22)(23)(24). These domains specifically bind phospho-derivatives of phosphatidylinositol. In this study, by analyzing over 20 lipid compounds with lipid-membrane overlay and liposome-binding assays, we defined a specific interaction of PS with the Sec14 domain of PTP-MEG2 that has no identifiable sequence similarities to other known PS-binding proteins. We further demonstrated that changing of PS distribution in cells alters localization of PTP-MEG2. Our study suggests that PS may have an important role in controlling function of PTP-MEG2. Intracellular PTPs contain various domains or segments surrounding the catalytic domain of ϳ230 amino acids. It has been well accepted that these flanking peptide segments or protein domains play regulatory and/or targeting roles. Recent studies (14,25) have shown that the N-terminal Sec14 domain-truncated form of PTP-MEG2 has a significantly higher activity than the full-length enzyme. This suggests a regulatory role of the non-catalytic Sec14 domain. We expected that interaction of PS with the Sec14 domain would stimulate the activity of PTP-MEG2. However, with the artificial substrate pNPP we did not see any significant change in the activity of PTP-MEG2 in the presence of PS. Nevertheless, it should be pointed out that this result does not necessarily rule out the possible regulatory role on the activity toward physiological substrates. In fact, anionic phospholipids stimulate the activities of SHP-1 and SHP-2 toward myelin basic protein but inhibit those toward pNPP (26,27). Although the possible effects on enzymatic activity needs identification of physiological substrates of PTP-MEG2 to clarify this point, the regulatory role of PS in controlling the function of PTP-MEG2 is well represented by PSmediated translocation of the enzyme. It should be noted that an earlier study with recombinant PTP-MEG2 expressed as a GST fusion protein in E. coli cells demonstrated a significant activation of the enzyme by phosphatidylinositol 4,5-diphosphate and other phospho-derivatives of phosphatidylinositol (25). Furthermore, a screening of cell extracts with matrices carrying the tethered homologs of lipids identified PTP-MEG2 as one of the binding proteins of phosphatidylinositol 3,4,5trisphosphate (28). However, with PTP-MEG2 purified from adenovirus-infected 293 cells, we did not see a stimulatory effect of phosphatidylinositol 4,5-diphosphate (14), and in this study, we did not observe a significant binding of PTP-MEG2 to phosphatidylinositol 3,4,5-trisphosphate either. Further work is required to clarify this discrepancy.
PS is a relatively abundant membrane lipid and has been recently implicated as an important ligand for clearance of apoptotic cells (10,11). Like other phospholipids, it is synthesized in the cytosolic leaflet of the endoplasmic reticulum membrane and moves to the plasma membrane through vesicular transport, and it is asymmetrically distributed on endoplasmic reticulum and the plasma membrane with a higher proportion in the inner leaflet of the membrane bilayer (29). When cells undergo apoptosis, however, PS on the plasma membrane is translocated to the outer leaflet by a phospholipid scramblase activated by protein kinase C. This is accompanied by concurrent inactivation of the aminophospholipid translocase thereby preventing PS returning to the inner leaflet and leaving PS expressed on the surface of apoptotic cells. Exposed PS is then recognized by its receptor-designated PSR and thereby initiates uptake of the apoptotic cells by phagocytes (12,13). It is well known that uptake and removal of necrotic or lysed cells involves inflammation and an immune response, but clearance of apoptotic cells does not induce either inflammation or immunity (30,31). The PSR is thought to be the molecular switch that determines the outcome (13). In fact, engagement of PS with PSR causes the production of inflammatory mediators, including transforming growth factor-␤, prostaglandin E2, and interleukin-10 (32,33). However, the exact signaling pathways involved in the apoptotic cell uptake and the anti-inflammation processes have not yet been sorted out. In this study, we have identified a potential intracellular receptor of PS. It is conceivable that when engulfed by phagocytes, a high level of PS in the outer membrane of apoptotic cells may alter the distribution of PTP-MEG2 in the phagocytes. Because PTPs play both positive and negative roles in cell signaling, we postulate that recruitment of PTP-MEG2 by PS accompanying the phagocytosis of apoptotic cells may suppress signal transduction pathways that lead to production of inflammatory factors and/or augment those pathways that stimulate generation of anti-inflammatory mediators. This notion is supported by the recent findings that PTP-MEG2 is localized on phagosomes and secretory vesicles (25,34). Finally, it should be pointed out that like PSR, PTP-MEG2 appears to be widely expressed, and thus its function may not be limited to phagocytes. After all, loss of PS asymmetry is not only found in apoptotic cells but also in other activated cells at least transiently, due to activation of the scramblase (13). Studies of PTP-MEG2 should provide a better understanding of this transient redistribution of PS.