Unique Membrane Interaction Mode of Group IIF Phospholipase A2*

The mechanisms by which secretory phospholipases A2 (PLA2s) exert cellular effects are not fully understood. Group IIF PLA2 (gIIFPLA2) is a structurally unique secretory PLA2 with a long C-terminal extension. Homology modeling suggests that the membrane-binding surface of this acidic PLA2 contains hydrophobic residues clustered near the C-terminal extension. Vesicle leakage and monolayer penetration measurements showed that gIIFPLA2 had a unique ability to penetrate and disrupt compactly packed monolayers and bilayers whose lipid composition recapitulates that of the outer plasma membrane of mammalian cells. Fluorescence imaging showed that gIIFPLA2 could also readily enter and deform plasma membrane-mimicking giant unilamellar vesicles. Mutation analysis indicates that hydrophobic residues (Tyr115, Phe116, Val118, and Tyr119) near the C-terminal extension are responsible for these activities. When gIIFPLA2 was exogenously added to HEK293 cells, it initially bound to the plasma membrane and then rapidly entered the cells in an endocytosis-independent manner, but the cell entry did not lead to a significant degree of phospholipid hydrolysis. GIIFPLA2 mRNA was detected endogenously in human CD4+ helper T cells after in vitro stimulation and exogenously added gIIFPLA2 inhibited the proliferation of a T cell line, which was not seen with group IIA PLA2. Collectively, these data suggest that unique membrane-binding properties of gIIFPLA2 may confer special functionality on this secretory PLA2 under certain physiological conditions.

10 sPLA 2 s (groups IB, IIA, IIC, IID, IIE, IIF, III, V, X, and XII) have been identified in mammals so far (1,10). Many sPLA 2 s have been shown to induce or augment cellular AA release and eicosanoid biosynthesis when overexpressed in or exogenously added to mammalian cells. However, it is not clear whether or not these sPLA 2 s are directly involved in AA production and inflammation under physiological conditions. Among various sPLA 2 s, group V PLA 2 (gVPLA 2 ) has been implicated in inflammation by a recent gene knock-out study (11).
It has been reported that sPLA 2 s can exert cellular effects through different mechanisms (12). Based on the earlier finding that the level of sPLA 2 was elevated in inflammatory exudates (13,14), it was generally thought that sPLA 2 s are released to the extracellular medium in response to specific stimuli and act on different target cells by a transcellular or paracrine mechanism. However, Kudo and co-workers found that many basic sPLA 2 s, * This work was supported by National Institutes of Health Grants GM52598 (to W. C.), GM66147 (to D. M.), and AG24234 (to D. S. U.). 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. 1 (2,4-dinitrophenyl)amino)hexanoyl)-1-hexadecanoyl-2-(4,4difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-pentanoyl)sn-glycero-3-phosphoethanolamine triethylammonium salt; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; POPG, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol; POPS, 1-palmitoyl-2oleoyl-sn-glycero-3-phosphoserine; pyrene-PG, 1-hexadecanoyl-2-(1pyrenyldecanoyl)-sn-glycero-3-phosphoglycerol; SM, brain sphingomylein; sPLA 2  including group IIA PLA 2 (gIIAPLA 2 ) and gVPLA 2 , remained bound to their parent cells after secretion due to their high affinity for cell surface heparan sulfate proteoglycans (HSPG) and were reinternalized to augment the stimulus-dependent AA release (15)(16)(17)(18)(19)(20)(21). This HSPG affinity has been shown to be important for the entry of different types of sPLA 2 s into mammalian cells (12,22). More recently, it was reported that gIIAPLA 2 and group X (gXPLA 2 ) could also induce the cellular AA release during the secretory process (23). Among known sPLA 2 s, gVPLA 2 (24,25) and gXPLA 2 (26,27) can effectively bind and hydrolyze zwitterionic phosphatidylcholine (PC) that is rich in the external leaflet of mammalian plasma membranes. As a result, these sPLA 2 s are able to directly act on mammalian cells and catalyze the hydrolysis of cell surface phospholipids. Lastly, some sPLA 2 s have been reported to exert cellular effects through the binding to cell surface receptors (28). Group IIF PLA 2 (gIIFPLA 2 ) is unique among sPLA 2 s in two respects. First, gIIFPLA 2 was shown to induce or augment the cellular AA and eicosanoid formation when overexpressed in mammalian cells despite having extremely low HSPG affinity (29) and low activity on PC vesicles (2). This suggests that gIIFPLA 2 might have a unique mode of cellular action. Second, gIIFPLA 2 is structurally unique in that it has an unusually long, proline-rich C-terminal extension (30, 31) (see Fig. 1A). To elucidate the mechanism by which gIIFPLA 2 acts on mammalian cells, we built a model tertiary structure of gIIFPLA 2 by homology modeling and measured the interactions of wild type and selected mutants of gIIFPLA 2 with various model membranes and mammalian cells. Results show that due to its unique structural and membrane binding properties, gIIFPLA 2 has an unprecedented ability to traverse the plasma membrane of mammalian cells, which is independent of binding to cell surface HSPG or phospholipid hydrolysis on the outer plasma membrane. These unique properties of gIIFPLA 2 may allow this sPLA 2 to perform some unusual functions under certain physiological conditions.
Mutagenesis and Protein Expression-The cDNA of fulllength mouse gIIFPLA 2 was cloned from the mouse testis cDNA library (Clontech) and subcloned into the pET-21a(ϩ) vector (Novagen, Madison, WI) between the restriction sites NdeI and XhoI. Site-directed mutagenesis was carried out by the overlap extension PCR. All mutant constructs were transformed into DH5␣ cells for plasmid isolation, and their DNA sequences were verified. E. coli strain BL21 (DE3) was used as a host for the protein expression. 4 liters of Luria broth medium containing 100 g/ml ampicillin was inoculated with 100 ml of the overnight culture from a freshly transformed single colony. The culture was grown at 37°C. When the optical density of the culture at 600 nm reached 0.8 -1.0, the culture was induced by 1 mM isopropyl-1-thio-␤-D-galactopyranoside (Research Products, Mount Prospect, IL). After incubation for 4 h at 37°C, cells were harvested at 5000 ϫ g for 10 min at 4°C and frozen at Ϫ20°C. The cells were resuspended in the CelLytic B-11 (Sigma) bacterial cell lysis extraction reagent (5 ml/g of cell paste), and deoxyribonuclease was added to a final concentration of 5 g/ml to reduce the viscosity of the suspension. The extraction suspension was shaken at room temperature for 15 min and centrifuged at 25,000 ϫ g for 15 min. Pellets were dissolved in CelLytic B-11 diluted 20-fold in water, incubated, and centrifuged as described above, and these steps were repeated twice to obtain clear inclusion body pellets. Inclusion bodies were solubilized in 10 ml of 50 mM Tris buffer, pH 8.0, containing 6 M guanidinium chloride, 1 mM EDTA and stirred overnight at 4°C. Any insoluble matter was removed by centrifugation at 50,000 ϫ g for 40 min at 4°C. The supernatant was loaded to a Superdex G-200 column (Amersham Biosciences) equilibrated with 50 mM Tris buffer, pH 8.0, containing 3 M guanidinium chloride and 5 mM EDTA. Fractions corresponding to the protein peak were pooled and added dropwise to 50 ml of 50 mM Tris, pH 8.0, containing 5 mM EDTA, 20 mM reduced glutathione, and 10 mM oxidized glutathione over 3 h. The solution was kept at room temperature for 20 h. The refolded protein solution was dialyzed against 4 liters of 25 mM Tris buffer, pH 8.0, containing 1 M urea for 4 h at 4°C and against 4 liters of 25 mM Tris buffer, pH 8.0, containing 0.5 M urea, 0.1 mM dithiothreitol for 2 h at 4°C, and finally against 25 mM Tris, pH 8.0, containing 0.5 M urea for 4 h at 4°C. The protein solution was centrifuged at 50,000 ϫ g for 40 min to remove insoluble matter, and the clear solution was loaded to a phenyl-Sepharose column (Amersham Biosciences) that was attached to an AKTA FPLC system (Amersham Biosciences) and equilibrated with 25 mM Tris, pH 7.4, containing 1 M ammonium sulfate. The column was eluted with a linear gradient of ammonium sulfate from 1 to 0 M and then with a linear gradient of 0 -30% (v/v) acetonitrile in the same buffer. Fractions corresponding to the major protein peak were pooled and dialyzed against 25 mM Tris, pH 7.4, containing 160 mM NaCl, and stored at 4°C. The purity of protein (Ͼ90%) was confirmed by SDS-PAGE. Protein concentration was determined by the bicinchoninic acid method (Pierce) using bovine serum albumin (BSA) as a standard.
PLA 2 Activity Assay-The PLA 2 -catalyzed hydrolysis of polymerized mixed vesicles (0.1 M pyrene-PG inserted in 9.9 M BLPC or BLPG) was carried out at 37°C in 2 ml of 10 mM Tris buffer, pH 7.4, containing 0.16 M KCl, 1 mM CaCl 2 , and 2 M BSA (32,33). The progress of hydrolysis was monitored as an increase in fluorescence emission at 378 nm using a Hitachi F4500 Fluorescence spectrophotometer with excitation wavelength set at 345 nm, and spectral bandwidth was set at 10 nm for both excitation and emission. The PLA 2 -catalyzed hydrolysis of PED6 in the mixed vesicles of POPS/cholesterol/POPG/ PED6 (107:31:20:1) was carried out at 37°C in 2 ml of 10 mM HEPES, pH 7.4, containing 0.16 M KCl, 1 mM Ca 2ϩ . The progress of hydrolysis was monitored as an increase in fluorescence emission at 520 nm with the excitation wavelength set at 488 nm. Spectral bandwidth was set at 10 nm for both excitation and emission. Values of specific activity were determined from the initial rates of hydrolysis.
Surface Plasmon Resonance Analysis-Kinetics of vesicle-PLA 2 binding was measured by the surface plasmon resonance (SPR) analysis using a BIAcore X biosensor system (Biacore AB) and the L1 chip as described previously (36). All measurements were performed at 23°C in 5 mM HEPES buffer, pH 7.4, containing 160 mM NaCl and 0.1 mM EDTA. The first flow cell was used as a control cell and was coated with 5400 resonance units of BSA. The second flow cell contained the surface coated with vesicles with varying lipid compositions at 5400 resonance units. After lipid coating, 30 l of 50 mM NaOH was injected at 100 l/min three times to wash out loosely bound lipids. Typically, no further decrease in SPR signal was observed after one wash cycle. After coating, the drift in signal was allowed to stabilize below 0.3 resonance units/min before any binding measurements, which were performed with a flow rate of 30 l/min. 90 l of protein sample was injected for an association time of 3 min, and the dissociation was then monitored for 10 min in running buffer. After each measurement, the lipid surface was typically regenerated with a 10-l pulse of 50 mM NaOH. The regeneration solution was passed over the immobilized vesicle surface until the SPR signal reached the initial background value before protein injection. When needed, the entire lipid surface was removed with a 5-min injection of 40 mM CHAPS followed by a 5-min injection of 40 mM octyl glucoside at 5 l/min, and the sensor chip was recoated for the next set of measurements. All data were analyzed using BIAevaluation 3.0 software (Biacore).
Vesicle Leakage Experiments-Appropriate amounts of lipids in chloroform were mixed, and the solvent was gently evaporated under a steam of dry N 2 to obtain the thin lipid film at bottom of a small thick-walled glass tube. To the dry lipid samples, 500 l of 5 mM HEPES buffer, pH 7.4, containing 50 mM 5-carboxyfluorescein was added, and the mixture was vortexed. Large unilamellar vesicles (LUVs) were prepared by repeated extrusion through 100-nm polycarbonate filters using a Liposofast extruder (Avestin, Ottawa, Canada). Vesicles were separated from nonencapsulated 5-carboxyfluorescein by gel filtration using a Sephadex G-50 column eluted with 5 mM HEPES buffer, pH 7.4, containing 160 mM NaCl and 0.1 mM EDTA. 150 nM (final concentration) sPLA 2 proteins were added to 300 nM (final concentration) 5-carboxyfluorescein-containing vesicles in 2.0 ml of 5 mM HEPES buffer, pH 7.4, containing 160 mM NaCl and 0.1 mM EDTA, and the release of 5-carboxyfluorescein was measured using a Hitachi F4500 spectrofluorometer with excitation and emission wavelengths set at 430 and 520 nm, respectively. After each leakage measurement, 20 l of Triton X-100 (Pierce) was added to the mixture to achieve 100% release of 5-carboxyfluorescein. The percentage of leakage was calculated as (F Ϫ F 0 )/(F max Ϫ F 0 ) ϫ 100, where F 0 is the fluorescence emission intensity before adding sPLA 2 , and F and F max represent the final fluorescence values after adding sPLA 2 and Triton X-100, respectively. All measurements were performed at 25°C.
Monolayer Measurements-Surface pressure () of solution in a circular Teflon trough (4-cm diameter ϫ 1-cm depth) was measured using a Wilhelmy plate attached to a computer-controlled Cahn electrobalance (model C-32) as described previously (37). 5-10 l of phospholipid solution in ethanol/hexane (1:9 (v/v)) was spread onto 10 ml of subphase (25 mM Tris, pH 7.4, containing 0.16 M KCl and either 0.1 mM EGTA or 0.1 mM CaCl 2 ) to form a monolayer with a given initial surface pressure ( 0 ). Once the surface pressure reading of monolayer had been stabilized (after ϳ5 min), the protein solution (typically 40 l) was injected into the subphase through a small hole drilled at an angle through the wall of the trough, and the change in surface pressure (⌬) was measured as a function of time at 23°C. Typically, the ⌬ value reached a maximum after 30 min. The maximal ⌬ value at a given 0 depended on the protein concentration and reached a saturation value when [sPLA 2 ] was Ն2 g/ml. Protein concentrations in the subphase were therefore maintained above such values to ensure that the observed ⌬ represented a maximal value. The critical surface pressure ( c ) was determined by extrapolating the ⌬ versus 0 plot to the x axis (38).
Fluorescence Labeling of sPLA 2 s-Purified mouse gIIFPLA 2 wild type and mutant proteins (H47Q and Y115A/F116A/ V118A/Y119A) were dialyzed against 25 mM Tris, pH 7.2, containing 0.5 M guanidinium chloride for 4 h at 4°C. Proteins were treated with a 10-fold molar excess of Texas Red TM C 2 -maleimide for 3 h at room temperature. The reaction was quenched by incubating the mixture with an excess amount (10-fold excess of maleimide) of cysteine for 30 min. The solution of labeled protein was dialyzed against 25 mM Tris, pH 7.2, containing 15% ammonium sulfate for 2 h at 4°C to remove excess reagents. The labeled proteins were purified using a phenyl-Sepharose column (Amersham Biosciences) as described above. Labeled protein fractions were collected and dialyzed against 25 mM Tris, pH 7.4, 160 mM NaCl for 24 h at 4°C and then stored at Ϫ20°C. W79C human gVPLA 2 was purified and labeled as described previously (22,39).
Microscopy Measurements on Giant Unilamellar Vesicles (GUVs)-GUVs were prepared by the electroformation method using a home-built device as described previously (40,41). Briefly, GUVs were grown in deionized water at 60°C for 30 min by spreading ϳ3 l of the lipid stock with various compositions on platinum wires. During GUV growth, the platinum wires were connected to a function generator (Hewlett-Pack-ard, Santa Clara, CA) for 30 min, and a low frequency AC field (sinusoidal wave function with a frequency of 10 Hz and an amplitude of 3 V) was applied. After 45 min, the temperature was lowered to 40°C, and the frequency generator was switched off after the system attained this temperature. All subsequent measurements were carried out at 40°C in deionized water.
All microscopy measurements were carried out using a custom-built combination laser-scanning and multiphoton microscope that was described previously (42). Briefly, a 920-nm ultrafast pulsed beam from a tunable Tsunami laser, set up for femtosecond operation (Spectra Physics, Mountain View, CA) was spatially filtered and launched into the scan head. The beam was directed toward the primary dichroic mirror (Chroma Technology, Brattleboro, VT) and then toward the XY scan mirrors (model 6350, Cambridge Technologies, Cambridge, MA). A Prairie Technologies scan lens (Middleton, WI) was used to focus the laser light, collimated by the ϫ1 Zeiss tube lens and directed toward a ϫ40 water-corrected 1.2 numerical aperture Zeiss objective, mounted on a Zeiss 200 M platform (Carl Zeiss Inc., Thornwood, NY). Light excited by a 920-nm ultrafast pulse was collected on a nondescanned pathway by the Peltier-cooled 1477P style Hamamatsu photomultiplier tubes. The light was reflected and filtered using appropriate optics. Instrument control was accomplished with the help of ISS amplifiers, an ISS three-axis scanning card (Champaign, IL), and two ISS 200-kHz analog lifetime cards. All of the microscopic experiments were controlled by a data acquisition program, SimFCS, kindly provided by Dr. Enrico Gratton.
Microscopy Measurements of sPLA 2 Internalization and Activity-The labeling of cell membranes by PED6 was performed as described previously (39). A mixture of POPS/cholesterol/POPG/PED6 (107:31:20:1 in molar ratio, 300 nmol total) in chloroform was dried under N 2 and resuspended in ethanol (10 l), followed by the addition of DMEM (10 l). The solution was dried again under N 2 until the volume was reduced to ϳ7 l to ensure that most of ethanol was evaporated. An additional 10 l of DMEM was added to the mixture, and vesicles were prepared by sonication of the mixture on ice (20 min). Vesicles were incubated with HEK293 cells (25-50 min at 37°C; 10 l in each well) that had been placed into each of eight wells on a sterile Nunc TM chambered cover glass and incubated for 24 h at 37°C with 5% CO 2 in the DMEM medium supplemented with 10% FBS and 250 g/ml Zeocin TM . Vesicletreated HEK293 cells were rinsed five times with phosphatebuffered saline to remove the unincorporated dye. These cells were then treated with 250 nM Texas Red-labeled sPLA 2 , and imaging was performed with a Zeiss LSM510 laser-scanning confocal microscope with the detector gain adjusted to eliminate the background autofluorescence. The fluorescence signal from Texas Red-labeled protein was monitored with a 568-nm argon/krypton laser and a 650-nm line pass filter, whereas the BODIPY TM signal from the hydrolyzed PED6 was monitored with a 488-nm argon/krypton laser and a 530-nm band pass filter. A ϫ63 (1.2 numerical aperture) water immersion objective was used for all experiments. Images were analyzed using the analysis tool provided in Zeiss biophysical software package. For cholesterol depletion, HEK293 cells (1 ϫ 10 6 cells/ml) were washed with the phosphate-buffered saline and incubated for 10 -30 min at 37°C with serum-free DMEM containing 5 mM methyl-␤-cyclodextrin (Sigma). After incubation, the medium was removed, and cells were washed with phosphatebuffered saline to remove methyl-␤-cyclodextrin. These cells were then treated with 200 nM Texas Red-labeled gIIFPLA 2 , and imaging was performed as described above.
AA Release from HEK293 Cells-Radiolabeling of HEK293 cells was achieved by incubating the cells (10 6 ) (10 6 ) were resuspended in 160 l of DMEM and 0.2% BSA and were stimulated with sPLA 2 . The reaction was quenched by adding 0.3 ml of ice-cold DMEM. The cell and the medium were separated by centrifugation, and then the radioactivity of pellet and supernatant, respectively, was measured by liquid scintillation.
Effects of gIIFPLA 2 on Cell Growth-The effects of sPLA 2 on cell proliferation and cell death were measured with the DO11.10 T cell hybridoma, using microscopic and cytofluorimetric readouts. Cells were cultured at 37°C in RPMI 1640 medium (Mediatech, Herndon, VA) supplemented with heatinactivated FBS (10% (v/v); HyClone Laboratories, Logan, UT), 2 mM L-glutamine, and 50 M 2-mercaptoethanol, as described (43). Physiological cell death (apoptosis) was induced by treatment with the macromolecular synthesis inhibitor actinomycin D (200 ng/ml, 12 h) (44). Dose-dependent effects on proliferation and viability were measured by seeding cells (5 ϫ 10 4 /well) into wells of a 24-well plate containing serial 2-fold dilutions of wild type or mutant sPLA 2 . Viable and dead cells (excluding and including trypan blue, respectively) were enumerated after varying periods of incubation. Viability also was confirmed cytofluorimetrically (FACSCaliber instrument and CellQuest software; BD Biosciences) by propidium iodide exclusion (1 g/ml; excitation ϭ 488 nm, emission ϭ 610 nm). Also the absence of externalized phosphatidylserine, as probed with fluorescein isothiocyanate-conjugated annexin V (BD PharMingen; excitation ϭ 488 nm, emission ϭ 525 nm), and the light scatter properties of cells, relative to the characteristic profiles of viable and apoptotic cells, were assessed simultaneously (43).
Expression of gIIFPLA 2 in CD4 ϩ T Cells-Mononuclear cells from peripheral blood from healthy volunteers (with approval by the ethical committee of Showa University) were obtained using Lymphoprep TM (NYCOMED) and were suspended in 5 ml of phosphate-buffered saline, pH 7.2, containing 5% FBS and 2 mM EDTA (MACS buffer). The cells were subjected to isolation of CD4 ϩ T cells through negative selection using the MACS CD4 ϩ T Cell Isolation Kit II (Milteny Biotech). Briefly, 10 7 cells were incubated with biotin/antibody mixture for 10 min and then with anti-biotin microbeads for 15 min on ice in 50 l of MACS buffer. The cells were resuspended in 500 l of MACS buffer and were applied to a MACS separator with LS column to obtain a CD4 ϩ T cell-enriched fraction. These preparations were then applied to MACS CD25 Microbeads (Miltency Biotech) in a similar way to separate CD25 high and CD25 Ϫ T cells. Live CD4 ϩ CD25 Ϫ cells (helper T cells) thus obtained were 90 -95% pure as assessed by flow cytometry (EPICS ELITE (version 4), Beckman Coulter) and were used in subsequent studies.
The CD4 ϩ CD25 Ϫ T cells (5 ϫ 10 5 ) were cultured in 500 l of RPMI1640 medium containing 10% FBS with or without 500 ng/ml anti-CD3 and anti-CD28 antibodies (BD Pharmingen) or 1 g/ml phytohemagglutinin at 37°C in a CO 2 incubator with 5% CO 2 . After 24 h, total RNA was extracted from these cells using TRIzol (Invitrogen), and an aliquot (500 ng) was subjected to a reverse transcriptase reaction with Rever Tra Ace (TOYOBO) at 42°C for 30 min and then 99°C for 5 min. The resulting cDNA was subjected to PCR with a set of 23-bp oligonucleotide primers corresponding to the 5Ј-and 3Ј-nucleotide sequences of the open reading frame of human gIIFPLA 2 using exTaq polymerase (Takara). The PCR conditions were 94°C for 30 s and then 35 cycles of amplification at 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s, as described (45). The reaction products were applied to 1% agarose gel electrophoresis with ethidium bromide.
Molecular Modeling of gIIFPLA 2 -The homology model of the mouse gIIFPLA 2 was built with the Nest (46) and Modeler (47) programs using the crystal structure of Group II sPLA2 from Agkistrodon halys pallas (Protein Data Bank accession number 1JIA) as the template and the alignment obtained with BLAST as a guide. The sequence identity between these sPLA 2 s is very high (41%). The structure of C-terminal extension was predicted by the ab initio modeling method in Nest. A calcium ion was added to the model by structurally aligning it to the template. The quality of the model is tested using Verify3D (48).

RESULTS
Model Structure of gIIFPLA 2 -To gain structural insight into the in vitro and cellular properties of gIIFPLA 2 , we built a homology model of the mouse gIIFPLA 2 . Three protein threading programs, which test the compatibility of the mouse gIIFPLA 2 sequence with structures in the Protein Data Bank, identified a group IIB sPLA 2 from A. h. pallas as the best structural template for modeling gIIFPLA 2 . The sequence alignment between query (gIIFPLA 2 ) and template (A. h. pallas PLA 2 ) that was used to construct the homology model in Nest is depicted in Fig. 1A. Since the sequence identity is high (41%), construct- Mutated residues of gIIFPLA 2 (i.e. His 47 and hydrophobic residues) are colored green. Basic residues near the C-terminal extension are colored blue, and the C-terminal extension (residues 128 -150) that is deleted in the ⌬C 128 -150 mutant is underlined. B, a model structure of mouse gIIFPLA 2 is shown in a ribbon diagram with its putative membrane-binding surface pointing upward. Mutated residues (green), including active site His 47 and near-C terminus hydrophobic residues, and near-C terminus basic residues (blue) are shown in stick representations and labeled. The C-terminal extension is colored red. C, an electrostatic potential surface of gIIFPLA 2 . The molecular orientation is the same as in B. Red and blue grids indicate negative and positive electrostatic potential surfaces, respectively. Mutated hydrophobic residues (green), His 47 (yellow), and Lys 111 and Arg 113 that generate a local cationic patch (blue) are shown in a space-filling representation. A Ca 2ϩ ion is shown in magenta.
Group IIF sPLA 2 OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 ing the alignment was straightforward. The predicted secondary structure composition of gIIFPLA 2 matches well with the observed secondary structure assignments of the template (data not shown). However, as seen in Fig. 1A, gIIFPLA 2 has a unique C-terminal extension. We were unable to find a suitable structural template for this region by searching the Protein Data Bank. Secondary prediction programs predicted that the region had random coil conformation. In addition, programs that detect coiled-coil sequences predicted that the C-terminal region of gIIFPLA 2 had no coiled-coil propensity. Modeling this region using the ab initio modeling method in Nest gave the best result compared with other loop prediction programs according to Verify3D (data not shown).
A resulting model structure of gIIFPLA 2 is shown in Fig. 1B with its putative membrane binding surface pointing upward. Typically, the membrane-binding surface of sPLA 2 surrounding the active site cavity contains basic and aromatic (and aliphatic) residues (49 -51). However, gIIFPLA 2 does not have those amino acids in the putative membrane-binding surface near the active site cavity. Also, the C-terminal extension (residues 128 -150) is rich in proline and acidic residues, and thus this part is not expected to be directly involved in membrane interactions. Interestingly, the stretch of residues between 108 and 119 (Fig. 1, A and B), which precedes the C-terminal extension and is located near the membrane binding surface, is rich in basic and aromatic residues and may, thus, play a role in membrane binding of gIIFPLA 2 .
Electrostatic potential calculation for the homology model for gIIFPLA 2 (see Fig. 1C) also shows its unique properties. Unlike most basic sPLA 2 s, such as gIIAPLA 2 and gVPLA 2 , that have predominant cationic patches, gIIFPLA 2 has a largely negative electrostatic profile with smaller cationic patches. Two most notable cationic patches are found near the Ca 2ϩ -binding loop and near the C-terminal extension (i.e. around Arg 109 , Lys 111 , and Arg 113 ), respectively. These cationic patches on or near the putative membrane binding surface may account for the reported ability of gIIFPLA 2 to bind and hydrolyze anionic phospholipids, such as phosphatidylglycerol (PG) (2). On the other hand, the overall negative electrostatic property of gIIFPLA 2 may be attributed to its low HSPG affinity (29).
Vesicle Binding Properties of gIIFPLA 2 -To investigate how gIIFPLA 2 interacts with membranes, we employed POPC, POPG, and POPC/SM/cholesterol (1:1:2 molar ratio) LUVs for binding measurements. POPC/SM/cholesterol (1:1:2) was used as a mimetic of the outer plasma membrane of mammalian cells, because we were interested in investigating how gIIFPLA 2 interacts with the outer plasma membrane. We measured the interaction of gIIFPLA 2 with these vesicles by SPR analysis in the absence of Ca 2ϩ to circumvent potential phospholipid hydrolysis during binding measurements. For comparison, binding of gVPLA 2 to these vesicles was also measured under the same conditions. As shown in Fig. 2, gVPLA 2 showed typical sensorgrams with association and dissociation phases for three types of vesicles. However, gIIFPLA 2 exhibited highly anomalous sensorgrams. For all three types of vesicles used, SPR signals steadily decreased to the base line during the association phase, indicating that lipid vesicles were detached from the sensor chip. This suggests that gIIFPLA 2 may disrupt the integrity of vesicles upon binding. This activity was not linked to the lipolytic activity of the protein, because Ca 2ϩ was absent in the mixture. Due to its potential bilayer-disrupting activity, binding of gIIFPLA 2 to lipid vesicles could not be quantified by SPR and other conventional methods.
Vesicle Leakage Caused by gIIFPLA 2 -We therefore performed vesicle leakage experiments using LUVs encapsulating 5-carboxyfluorescein to assess the bilayer-disrupting activity of gIIFPLA 2 . We first measured the release of 5-carboxyfluorescein from POPC/SM/cholesterol (1:1:2) LUV by gIIAPLA 2 , gIIFPLA 2 , and gVPLA 2 in the absence of Ca 2ϩ . The vesicle leakage was monitored in terms of the increase in 5-carboxyfluorescein fluorescence emission due to the relief of self-quenching. The addition of gIIFPLA 2 to the vesicles caused the rapid release of 5-carboxyfluorescein from the vesicles in a concentration-dependent manner (see Fig. 3A). Under the same conditions, gIIAPLA 2 induced little leakage, whereas gVPLA 2 showed 30% of the gIIFPLA 2 activity (Fig. 3B).
To understand the molecular basis of the unique vesicle-disrupting activity of gIIFPLA 2 , we then varied the lipid composition of 5-carboxyfluorescein-containing vesicles. Fig. 3C shows that gIIFPLA 2 induced a significantly weaker leakage with POPC/SM/cholesterol (1:1:1) LUV than with POPC/SM/cholesterol (1:1:2) LUV. Furthermore, gIIFPLA 2 was not able to cause any leakage with POPC/SM (1:1) and POPC LUVs. These data thus indicate that the presence of cholesterol is essential for the unique ability of gIIFPLA 2 to disrupt the neutral lipid bilayers. To elucidate the structural determinant of its unique membrane binding properties, we prepared a panel of gIIFPLA 2 mutants and measured the vesicle leakage using POPC/SM/ cholesterol (1:1:2) LUV encapsulating 5-carboxyfluorescein by wild type and mutants under the same conditions. In particular, we mutated aromatic/aliphatic residues near the C-terminal extension of gIIFPLA 2 , Y115A/F116A and Y115A/F116A/ V118A/Y119A, and generated a C-terminal deletion mutant (⌬ 128 -150 ). H47Q was prepared to confirm that the lipolytic activity is not involved in its vesicle-leaking activity. As shown in Fig. 3D, Y115A/F116A and Y115A/F116A/V118A/Y119A caused no detectable vesicle leakage, indicating that these aromatic and aliphatic residues (Tyr 115 , Phe 116 , Val 118 , and Tyr 119 ) are essential for the vesicle-disrupting activity of gIIFPLA 2 . In contrast, ⌬ 128 -150 behaved essentially the same as the wild type, suggesting that this region plays no direct role in membrane interaction. As expected, H47Q behaved similarly to the wild type with respect to the vesicle leakage. The effects of basic residues, Lys 111 and Arg 113 , on the interactions of gIIFPLA 2 with various membranes were not investigated in this study because of low stability of corresponding mutants (e.g. K111A, K111E, R113A, and R113E).
Enzyme Activity of gIIFPLA 2 and Mutants-To investigate the role of the clustered hydrophobic residues in the interfacial catalysis of gIIFPLA 2 , we also measured the activities of wild type and mutants in the presence of 1 mM Ca 2ϩ toward polymerized mixed vesicles that have been used for substrates for many sPLA 2 s (32,33,52). In this model membrane system, a pyrene-labeled phospholipid (i.e. pyrene-PG) incorporated in the inert polymerized matrix of BLPG (or BLPC) is selectively hydrolyzed by sPLA 2 , which can be spectrofluorometrically monitored. This system was particularly useful for gIIFPLA 2 , because polymerized vesicles would not be easily disrupted by gIIFPLA 2 during the activity assay. GIIFPLA 2 had high specific activity for pyrene-PG incorporated in anionic BLPG vesicles (see Fig. 4). However, it showed much lower activity for pyrene-PG incorporated in zwitterionic BLPC vesicles (data not shown), showing that gIIFPLA 2 prefers anionic to zwitterionic membranes in the absence of cholesterol. This finding is also consistent with our model structure (see Fig. 1C) showing the presence of cationic patches on the putative membrane-binding surface. Among gIIFPLA 2 mutants, a good correlation between vesicle-disrupting activity and interfacial enzymatic activity was observed. In other words, ⌬ 128 -150 with the wild type-like vesicle-disrupting activity had essentially the same enzyme activity toward pyrene-PG/BLPG polymerized mixed vesicles as wild type, whereas Y115A/F116A and Y115A/ F116A/V118A/Y119A with greatly reduced vesicle-disrupting activities showed drastically reduced enzymatic activity. Thus, the clustered hydrophobic residues seem to play an important role in the interfacial activity of gIIFPLA 2 , presumably by enhancing its membrane binding. As expected, H47Q showed no activity for any polymerized mixed vesicles.
Monolayer Penetration of gIIFPLA 2 -To understand the mechanism by which gIIFPLA 2 causes the vesicle leakage and  the roles of the above residues in membrane binding of gIIFPLA 2 , we measured the interactions of gIIFPLA 2 and mutants with various lipid monolayers at the air-water interface. This system has been used to measure the membranepenetrating activity of a wide variety of proteins (38). The phospholipid monolayer was spread at constant area and the change in surface pressure (⌬) was monitored after the injection of protein into the subphase. In general, ⌬ is inversely proportional to 0 of the lipid monolayer, and an extrapolation of the ⌬ versus 0 plot yields the critical surface pressure ( c ), which specifies the upper limit of 0 of a monolayer that a protein can penetrate into (38,53). Because the surface pressure of cell membranes has been estimated to be in the range of 30 -35 dynes/cm (54 -56), the c value for a protein that penetrates cell membranes should be above 30 dynes/cm. We first measured the penetration of wild type gIIFPLA 2 into various monolayers (see Fig. 5A). Again, Ca 2ϩ was removed from the subphase in most measurements to circumvent potential hydrolysis during monolayer measurements. gIIFPLA 2 showed high penetrating activity for the monolayer comprising POPC/SM/cholesterol (1:1:2) with c slightly above 30 dynes/ cm. This is consistent with the unique ability of gIIFPLA 2 to cause leakage from the POPC/SM/cholesterol (1:1:2) LUV. Furthermore, the monolayer-penetrating activity of gIIFPLA 2 greatly decreased when cholesterol was removed from the monolayer. The c value was reduced to Ͻ25 dynes/cm for POPC and POPC/SM monolayers. We also used a nonhydrolyzable PC analog, DHPC, instead of POPC in the PC/SM/cholesterol (1:1:2) monolayer and measured its interaction with gIIFPLA 2 in the presence of 1 mM Ca 2ϩ in the subphase. As shown in Fig. 5A, gIIFPLA 2 showed essentially the same affinity for the POPC/SM/cholesterol (1:1:2) monolayer without Ca 2ϩ and for the DHPC/SM/cholesterol (1:1:2) monolayer with 1 mM Ca 2ϩ , validating our approach of measuring membrane interactions of gIIFPLA 2 in the absence of Ca 2ϩ to circumvent the hydrolysis. DHPC was not used in vesicle leakage studies, because vesicles formed in the presence of this lipid had a high background leakage in the absence of gIIFPLA 2 . It should be noted that LUVs used in our vesicle leakage measurements are known to have the surface pressure above 30 dynes/cm (54 -56). This explains why gIIFPLA 2 did not cause a leakage with POPC or POPC/SM vesicles, although it had some affinity for POPC and POPC/SM monolayers with the initial pressure below 25 dynes/cm.
We then compared the interactions of gIIFPLA 2 , gIIAPLA 2 , and gVPLA 2 with the POPC/SM/cholesterol (1:1:2) monolayer. Fig. 5B illustrates that gIIAPLA 2 has significantly lower monolayer-penetrating activity than gIIFPLA 2 . gVPLA 2 was more active than gIIAPLA 2 but was less active than gIIFPLA 2 , with c below 30 dynes/cm. These results are consistent with the differential activities of the three enzymes to induce the vesicle leakage (see Fig. 3B). We then measured the penetration of gIIFPLA 2 mutants into the same monolayer. As shown in Fig.  5C, Y115A/F116A and Y115A/F116A/V118A/Y119A with little vesicle-leaking activity had c values below 25 dynes/cm. In contrast, ⌬ 128 -150 and H47Q with wild type-like vesicle-disrupting activity showed monolayer penetration that was comparable with that of the wild type. Collectively, our monolayer measurements indicate that gIIFPLA 2 has a unique ability to penetrate the compactly packed zwitterionic lipid monolayers and bilayers, which accounts for its capability of inducing vesicle leakage, and that the presence of cholesterol is important for this activity.
Interaction of gIIFPLA 2 with GUV-GUVs (diameter Ͼ10 m) are an excellent model system for cell membranes that allows direct visualization of various membrane processes, including structural changes of membranes (57). Since GUVs are devoid of proteins and carbohydrates, they allow for investigating if and how gIIFPLA 2 crosses the lipid bilayer based solely on its lipid-binding properties. We prepared GUVs composed of PC/SM/cholesterol/NBD-cholesterol (1:1:2:0.04), and the Texas Red-labeled gIIFPLA 2 was added to these GUVs in the absence of Ca 2ϩ . As shown in Fig. 6, Texas Red-labeled gIIFPLA 2 initially bound to the surface of GUVs and then entered the vesicles and was accumulated in high concentration inside the vesicles within 10 min, which led to the dramatic contraction and disruption of the vesicles. Essentially the same pattern was observed for Ͼ90% of GUVs characterized under the same conditions. Neither the Y115A/F116A/V118A/ Y119A mutant gIIFPLA 2 nor gVPLA 2 caused a similar disruption of GUV under the same conditions (data not shown). These results thus confirm that gIIFPLA 2 has a unique lipolytic activity-independent ability to go across the cholesterol-and SM-rich neutral lipid bilayer. This also suggests that this sPLA 2 may be able to enter mammalian cells by directly crossing the plasma membrane without having to rely on any endocytic protein machinery.
Action of gIIFPLA 2 on HEK293 Cells-Unusual membrane binding properties of gIIFPLA 2 suggest that this sPLA 2 may be able to traverse the plasma membrane in a lipolysis-, HSPG-, and endocytosis-independent manner. To explore this possibility, we chemically labeled gIIFPLA 2 with Texas Red and exogenously added the fluorescently labeled gIIFPLA 2 to HEK293 cells whose membranes are separately labeled with a fluorogenic PLA 2 substrate, PED6. PED6 has been shown to be randomly distributed among various cellular membranes, including both leaflets of the plasma membrane, and display a large increase in fluorescence emission upon hydrolysis (39,58). This approach, which has been successfully employed for the cell studies of gVPLA 2 (12,39), allowed simultaneous real time monitoring of the spatiotemporal dynamics and lipolytic activity of gIIFPLA 2 .
Prior to cell studies, we first measured the in vitro specific activities of fluorescently labeled and unlabeled proteins using POPS/cholesterol/POPG/PED6 (107:31:20:1) vesicles as a substrate. As listed in Table 1, gIIFPLA 2 had 3-fold lower activity than gVPLA 2 W79C (this mutant is essentially identical to the wild type gVPLA 2 in all respects) (39) for the substrate at 1 mM Ca 2ϩ , and the difference was far greater at lower Ca 2ϩ concentrations. This is because gVPLA 2 has a lower Ca 2ϩ requirement than gIIFPLA 2 (2). Thus, gIIFPLA 2 is expected to have much lower activity for PED6 than gVPLA 2 in the cytosol, where Ca 2ϩ is present in a submicromolar concentration. The presence of a single free cysteine (Cys 137 ) in the C-terminal extension (see Fig. 1A) made it possible to specifically incorporate a single fluorescence probe into gIIFPLA 2 . The purified Texas Red-labeled gIIFPLA 2 proteins had the same enzymatic activity toward PED6 (see Table 1) and vesicle-disrupting activity toward POPC/SM/cholesterol (1:1:2) vesicles (data not shown) as unlabeled proteins.
As shown in Fig. 7A, wild type gIIFPLA 2 initially bound the plasma membrane and then readily entered HEK293 cells and accumulated on various intracellular locations, including the perinuclear region (see the red panel). A separate study using HEK293 cells expressing enhanced green fluorescent proteintagged EEA1 protein showed that gIIFPLA 2 was not located in endosomal structures (data not shown). Also, incubation of HEK293 cells on ice for 30 min before adding gIIFPLA 2 did not have a significant effect on the cellular entry of gIIFPLA 2 (Fig.  7B). These results suggest that gIIFPLA 2 enters the cells by an endocytosis-independent mechanism. This notion is also consistent with our GUV data. As far as the rate of entry into  HEK293 cells was concerned, gIIFPLA 2 was only slightly less effective than the Texas Red-labeled W79C-gVPLA 2 that was shown to rapidly enter HEK293 and other mammalian cells due to its high HSPG affinity and PC activity (39) (see also Fig. 7D). However, unlike gVPLA 2 , which caused strong intracellular signals of PED6 hydrolysis, which were colocalized with protein signals, gIIFPLA 2 did not induce extensive PED6 hydrolysis (see the green panel of Fig. 7A). gIIFPLA 2 caused a minor degree of PED6 hydrolysis at the plasma membrane, but little PED6 hydrolysis was seen intracellularly. This is not unexpected, given the low enzymatic activity of gIIFPLA 2 toward PED6, particularly at low Ca 2ϩ . To explore the possibility that a rise in intracellular Ca 2ϩ may enhance the activity of internalized gIIFPLA 2 , we treated HEK293 cells with 10 M ionomycin after gIIFPLA 2 entered the cells. However, little increase in PDE6 hydrolysis was seen even after 30 min (data not shown), suggesting that gIIFPLA 2 has low cellular activity even at an elevated level of Ca 2ϩ . To see if gIIFPLA 2 also shows much lower activity than gVPLA 2 toward natural phospholipid substrates, we labeled HEK293 cells with [ 3 H]AA and measured the release of [ 3 H]AA from the cells after incubation with exogenously added gIIFPLA 2 and gVPLA 2 , respectively. Fig. 8 shows that gIIFPLA 2 has less than 5% of the gVPLA 2 activity. Collectively, these results indicate that although gIIFPLA 2 can effectively enter HEK293 cells, it does not induce a significant degree of lipid hydrolysis due to its lower enzyme activity toward phospholipids in the plasma membrane and intracellular membranes.
Unlike the wild type gIIFPLA 2 , Y115A/F116A/V118A/ Y119A with much lower membrane-penetrating activities showed neither cellular entry nor PED6 hydrolysis when exogenously added to HEK293 cells under the same conditions (Fig.  7E). This indicates that the membrane-disrupting activity of gIIFPLA 2 is responsible for its unique cell membrane-traversing capability. A catalytically inactive mutant H47Q did enter HEK293 cells as well as the wild type gIIFPLA 2 without causing any hydrolysis (Fig. 7F), showing that lipolytic activity is not essential for the cellular entry of gIIFPLA 2 .
Lastly, to measure the effect of cholesterol on the cellular entry of gIIFPLA 2 , we treated HEK293 cells with 5 mM methyl-␤-cyclodextrin. Although cells started to lose integrity after elongated incubation as a consequence of cholesterol depletion, about a half-population of cells still maintained the integrity within 20 min of incubation. Interestingly, when Texas Red-labeled gIIFPLA 2 was exogenously added to these methyl-␤-cyclodextrin-treated HEK293 cells, it accumulated on the surfaces of the cells but did not enter the cells (see Fig. 7C). Essentially all red signals were removed from the cells when they were washed with the medium (data not shown). The same pattern was seen with virtually all methyl-␤-cyclodextrintreated cells. This result again underscores that the presence of cholesterol is essential for cell membrane-traversing activity of gIIFPLA 2 .
Effects of gIIFPLA 2 on Cell Proliferation-The unique membrane-disrupting and -translocating properties of gIIFPLA 2 suggested that this sPLA 2 might impact mammalian cell viability or even trigger cell death. To address this possibility, we looked for a particular cell type that intrinsically expresses gIIFPLA 2 . By means of reverse transcription-PCR, we found that CD4 ϩ CD25 Ϫ helper T cells isolated from peripheral bloods of human volunteers express gIIFPLA 2 mRNA (Fig.  9A). The expression of gIIFPLA 2 was markedly increased in CD4 ϩ CD25 Ϫ T cells after immunologic (anti-CD3 ϩ anti-CD28 antibodies) and nonimmunologic (phytohemagglutinin) stimulation, revealing a marked stimulus-coupled inducibility of this enzyme.
Based on this finding, we chose a murine T cell line, DO11.10, which proliferates rapidly and is particularly susceptible to the induction of apoptotic cell death, as a sensitive cell with which to test the effects of sPLA 2 . Over 40 h, the untreated cells underwent 2.5 population doublings (i.e. the doubling time is about 16 h). As shown in Fig. 9B, the wild-type gIIFPLA 2 significantly inhibited this proliferation; 50% inhibition was observed at ϳ0.3 M. No appreciable cell death was triggered, although  much higher doses of sPLA 2 (Ͼ1 M) were toxic. These results were confirmed by flow cytometric analysis. With regard to forward and side angle light scatter, viable cells (Fig. 10A) were heterogeneous in size (because they were proliferating asynchronously) and showed low side scatter, whereas apoptotic cells (Fig. 10B) were smaller (less forward angle light scatter) and more granular (higher side angle scatter) and included cells that failed to exclude propidium iodide (data not shown). The population of cells treated with 0.25 M wild-type gIIFPLA 2 ( Fig. 10C) included fewer large, blasting cells but no appreciable numbers of apoptotic cells, consistent with an inhibition of proliferation and an absence of cell death. It is notable that the side angle scatter of these cells is elevated, suggesting that they may have some nonlethal membrane irregularity, perhaps as a consequence of membrane disruption by the enzyme. Under the same conditions, the Y115A/F116A/V118A/Y119A mutant of gIIFPLA 2 was much less effective than the wild type in inhibiting cell proliferation. Even at 1 M, the mutant caused less than 40% inhibition (Fig. 9B), and the dose-dependent extent of apparent membrane irregularity was greatly reduced (Fig. 10E). Finally, gIIAPLA 2 had no effect on proliferation or cell integrity (Figs. 9B and 10D).

DISCUSSION
Despite intensive studies on sPLA 2 s in the past decade, their physiological functions and the mechanisms by which these enzymes exert cellular effects still remain unknown and will require further genetic and cell studies. Recent studies on various sPLA 2 s over the years have highlighted a good correlation between their biochemical/biophysical properties and their cellular actions. For instance, high activity of gVPLA 2 (24,25) and gXPLA 2 (26,27) toward PC membranes is well correlated with their unique capability to act directly on mammalian cells, the outer plasma membrane of which is rich in PC. Also, high heparin affinity of many basic sPLA 2 s is linked to their binding to cell surface HSPG and cellular uptake (12,(15)(16)(17)(18)(19)(20)(21)(22). Therefore, characterization of biochemical and biophysical properties of different sPLA 2 isoforms may provide an important new clue to their physiological functions and the mechanism of their cellular actions. The present study was performed to characterize biochemical and biophysical properties of gIIFPLA 2 with an aim of gaining new insight into its physiological functions. Our structural modeling suggests that gIIFPLA 2 is a largely negatively charged molecule with smaller cationic patches on its putative membrane-binding surface (see Fig. 1C). gIIFPLA 2 prefers anionic PG membranes to zwitterionic PC membranes due to the presence of surface cationic patches, including one composed of Arg 109 , Lys 111 , and Arg 113 near the C-terminal extension. Although the role of clustered cationic residues could not be assessed in this study due to the low stability of corresponding mutants, they are expected to play a role in binding to anionic membranes. Relatively weak interaction of gIIFPLA 2 with PC membranes is dramatically enhanced in the presence of cholesterol. In particular, gIIFPLA 2 demonstrates novel nonhydrolytic membrane-penetrating and -disrupting activities on cholesterol-