C2 domain membrane penetration by group IVA cytosolic phospholipase A₂ induces membrane curvature changes.

Group IVA cytosolic phospholipase A(2) (cPLA(2)α) is an 85 kDa enzyme that regulates the release of arachidonic acid (AA) from the sn-2 position of membrane phospholipids. It is well established that cPLA(2)α binds zwitterionic lipids such as phosphatidylcholine in a Ca(2+)-dependent manner through its N-terminal C2 domain, which regulates its translocation to cellular membranes. In addition to its role in AA synthesis, it has been shown that cPLA(2)α promotes tubulation and vesiculation of the Golgi and regulates trafficking of endosomes. Additionally, the isolated C2 domain of cPLA(2)α is able to reconstitute Fc receptor-mediated phagocytosis, suggesting that C2 domain membrane binding is sufficient for phagosome formation. These reported activities of cPLA(2)α and its C2 domain require changes in membrane structure, but the ability of the C2 domain to promote changes in membrane shape has not been reported. Here we demonstrate that the C2 domain of cPLA(2)α is able to induce membrane curvature changes to lipid vesicles, giant unilamellar vesicles, and membrane sheets. Biophysical assays combined with mutagenesis of C2 domain residues involved in membrane penetration demonstrate that membrane insertion by the C2 domain is required for membrane deformation, suggesting that C2 domain-induced membrane structural changes may be an important step in signaling pathways mediated by cPLA(2)α.

been found to induce membrane curvature changes ( 27 ), including the epsin N-terminal homology (ENTH) ( 28 ), bin-amphiphysin-Rvs167 (BAR) ( 29 ), Pleckstrin homology (PH) ( 30 ), Amot coiled-coil homology domain (ACCH) ( 31 ), and C2 domains ( 32 ). Here, we investigate the ability of the isolated C2 domain of cPLA 2 ␣ to induce changes to lipid structure. A number of imaging assays are used, including electron microscopy (EM) of large unilamellar vesicles (LUVs), imaging of giant unilamellar vesicles (GUVs), and imaging of membrane sheets. In addition, biophysical assays, including monolayer penetration analysis and surface plasmon resonance (SPR), were used to correlate membrane penetration and affi nity with membrane remodeling activity. Results provide evidence that Ca 2+ -dependent membrane insertion of CBL1 and -3 of the C2 domain drive membrane curvature changes.
Besides its role in eicosanoid biosynthesis, cPLA 2 ␣ is selectively activated upon Fc receptor (FcR)-mediated phagocytosis in macrophages, where it rapidly translocates to the nascent phagosome ( 23 ). Unexpectedly, however, it was shown that membrane binding by the isolated C2 domain of cPLA 2 ␣ was suffi cient to induce phagosome formation ( 23 ), suggesting that the C2 domain alone has membrane binding activity that regulates phagocytosis. This was further verifi ed with a mutation in the C2 domain, D43N, which abrogates Ca 2+ -binding and could not rescue phagocytosis ( 23 ). cPLA 2 ␣ also plays a role in membrane curvature generation through regulation of aberrant Golgi vesiculation ( 24 ), Golgi tubulation ( 25 ), and vesiculation of cholesterolrich, GPI-anchored, protein-containing endosomes ( 26 ). Although it is speculated the C2 domain may induce changes to membrane structure ( 23 ), direct evidence is lacking. These recent studies suggest that cPLA 2 ␣ and its C2 domain have membrane remodeling activity that is critical to biological signaling pathways.
Recently, a number of peripheral proteins, mainly attributed to their modular lipid binding domains, have mutants and were subsequently imaged after a 15 min incubation with a confocal microscope (Nikon A1R-MP with a 100× 1.4 NA oil objective).

Monolayer penetration
Surface pressure ( ) of solution in a circular Tefl on trough (2 ml) was measured using a wire probe connected to a Kibron MicroTrough X (Kibron, Inc., Helsinki, Finland) as previously described ( 33 ). Phospholipid solution (2-8 l) in hexane/ethanol (9:1 v/v) was spread onto 2 ml of subphase to form a monolayer with a given initial surface pressure ( 0 ). The subphase was stirred continuously at 30 revolutions/min with a small magnetic stir bar. After stabilization of the surface pressure of the monolayer ( ‫ف‬ 5 min), the protein solution (typically 10 l) was injected into the subphase, and the change in surface pressure ( ⌬ ) was measured as a function of time. Generally, the ⌬ value reached a maximum after 20 min. The maximal ⌬ value depends upon the protein concentration and reached saturation at cPLA 2 ␣ -C2 > 1 g/ml, as previously reported ( 34 ). Protein concentrations in the subphase are maintained above such values to ensure that the ⌬ represents a maximal value. The ⌬ versus 0 plots were constructed from these measurements to obtain the x-intercept or critical pressure ( c ) defi ned as the value to which the protein penetrates ( 35 ).

SPR assays
All SPR measurements were performed at 25°C. A detailed protocol for coating the L1 sensor chip has been described elsewhere ( 34 ). Briefl y, after washing the sensor chip surface, 90 l of POPC vesicles were injected at 5 l/min to give a response of 6,500 resonance units. An uncoated fl ow channel was used as a control surface. Under our experimental conditions, no binding was detected to this control surface beyond the refractive index change for the C2 domain of cPLA 2 ␣ as previously reported ( 18,34 ). Each lipid layer was stabilized by injecting 10 l of 50 mM NaOH three times at 100 l/min. SPR measurements were done at the fl ow rate of 5 l/min. To give an association time to reach saturation of binding signal ( R eq ), 50-90 l of protein in 10 mM HEPES (pH 7.4) containing 160 mM KCl, and 50 M Ca 2+ was injected (see Fig. 5C). The saturation responses for wild type (WT) and mutations were normalized where maximum WT saturation response was set to 1 to compare the binding capacity of WT versus mutations. The lipid surface was regenerated using 10 l of 50 mM NaOH. Each of the sensorgrams was corrected for refractive index change by subtracting the control surface response from the binding curve. A minimum of three data sets was collected for each protein at a minimum of fi ve different concentrations for each protein within a 10-fold range of K d . R eq values were then plotted versus protein concentration and the K d value was determined by a nonlinear least-squares analysis of the binding isotherm using the equation R eq = R max /(1 + K d / C ) ( 36 ). Each data set was repeated three times to calculate a standard deviation.

Circular dichroism spectroscopy
To ensure the WT and mutant proteins retained a stable structure, circular dichroism (CD) was used to assess the secondary structure of each recombinant protein used in the study. The spectra were taken on a JASCO 815 CD spectrometer scanned from 195 to 250 nm in a 1 mm quartz spectrophotometer cell

Cloning and protein expression
The QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) was used to introduce mutations into the pET28a vector with a His 6 tag engineered into the N-terminus of the cPLA 2 ␣ C2 domain gene ( 9 ). All mutated constructs were sequenced to ensure the presence of the desired mutation. The C2 domain and respective mutations were expressed and purifi ed from Escherichia coli BL21(DE3) cells as previously described ( 9 ). Protein concentrations were determined by the bicinchoninic acid method, and all protein aliquots were stored in 20 mM HEPES (pH 7.4) containing 160 mM KCl.

Electron microscopy
Two hundred microliters of 1 mg/ml POPC LUVs were prepared as previously described ( 32 ). Briefl y, the lipids were dried under nitrogen gas and resuspended in 20 mM HEPES (pH 7.4) containing 160 mM KCl and either 100 M CaCl 2 or 100 M EGTA. The vesicles were incubated at 37°C and extruded through an 800 nm fi lter (Avanti Polar Lipids, Alabaster, AL). The respective C2 domain and mutations were incubated at a concentration of 10 M with the POPC vesicles for 30 min at 25°C. Samples were then applied to a carbon-formvar-coated copper grid and stained with 2% uranyl acetate. Liposome morphologies were then imaged at 80 kV on a FEI 80-300 D3203 electron microscope at 6,300× magnifi cation.

Giant unilamellar vesicle assay
An aliquot of lipids containing POPC, POPE, and POPS suspended in chloroform were prepared in a 60:20:20 molar ratio. The suspension was dried under nitrogen gas and resuspended in chloroform to a fi nal concentration of 0.4 mg/ml. The lipid suspension was dried onto an indium tin oxide-coated slide and dehydrated under a vacuum for 1 h. The GUV apparatus was assembled, and a 350 mM D-sucrose solution was placed into the reservoir containing the dried lipids. Another glass plate was placed on top to eliminate air from the system, and then a sin wave generator was applied at 3V and 20 Hz for 5 h at 25°C. The GUV solution was collected and stored at 25°C until use. The GUV solution was diluted 20-fold in 20 mM HEPES (pH 7.4) containing 160 mM KCl and 10 M FM® 2-10 lipophilic dye. Samples were prepared with 100 M EGTA, 10 M CaCl 2 , or 500 nM CaCl 2 as necessary for experimental conditions. GUV vesiculation was assessed after a 5 min incubation with 2 M or 500 nM cPLA 2 ␣ -C2, 500 nM full-length cPLA 2 ␣ , or 2 M of respective mutants and was imaged via confocal microscopy (Zeiss LSM 710) on Nunc Lab-Tek I Chambered Cover Glasses (8 well) using a 63× 1.4 NA oil objective. Three replicates for each control or sample were quantifi ed by counting 60-100 GUVs per replicate. The number of vesiculated GUVs was determined separately and compared with the total number of GUVs in each replicate. The degree of vesiculation was then expressed as a percentage, compared with the control, and quantifi ed using an unpaired Student t -test.

Membrane sheets
Two microliters of 10 mM POPC was prepared in chloroform, spotted onto Nunc Lab Tek I (8 well) Chambered Cover Glasses, and dried under a vacuum. The lipid was rehydrated with 20 mM HEPES (pH 7.4) containing 160 mM KCl and 20 M FM® 2-10 and allowed to rehydrate for 15 min. Samples were prepared with 100 M EGTA or 100 M CaCl 2 as necessary for experimental conditions. The experiments contained 2 M cPLA 2 ␣ -C2 or respective calcium-binding loops to penetrate deeply into membranes and monolayers where hydrophobic and aromatic residues in these loops protrude into the hydrocarbon region of the membrane ( 9,16 ). Indeed, F35A/L39A and Y96A displayed a drastic reduction in liposome morphology changes and displayed a lack of long, thin tubules emanating from liposomes as seen for the WT C2 domain. M38A and M98A, which have been shown to have a lesser effect than F35A, L39A, or Y96A on cPLA 2 membrane binding ( 9 ), induced changes in liposome structure, albeit to a lesser extent than WT.

Quantifi cation of membrane curvature changes using GUVs
GUVs have served as an effective platform for monitoring changes to membrane structure because they can be fl uorescently labeled and imaged with confocal microscopy and are more easily quantifi ed than EM experiments. GUVs have been effective in monitoring membrane curvature changes for the ENTH domain ( 37 ) and viral matrix proteins ( 40 ). In addition, they are relatively fl at (mean diameter, ‫ف‬ 30 M) in comparison to LUVs (mean diameter, ‫ف‬ 400 nm), so they can be used to assess if proteins induce membrane curvature changes on different membrane surfaces. GUVs composed of POPC:POPE:POPS (60:20:20) were prepared and used to quantify membrane curvature changes for WT C2 domain and respective mutations. All experiments were performed in triplicate, and at least 60 GUVs were counted in each experiment and assessed for membrane curvature changes in response to C2 domain binding. WT C2 domain induced vesiculation of GUVs in the presence of Ca 2+ , which was not observed in the presence of 100 M EGTA ( Fig. 3A ) . The hydrophobic and aromatic mutations F35A/L39A and Y96A, which greatly reduced alterations to liposome morphology in the EM assays, signifi cantly reduced GUV vesiculation for which their quantitative value was similar to that of the control. M38A and M98A displayed a statistically significant reduction in GUV vesiculation, in line with the EM assays, which detected appreciable changes to liposome structure, albeit to a lesser extent than WT. To assess the ability of the C2 domain to induce membrane curvature changes, we assessed the ability of the C2 domain to induce vesiculation in GUVs with 200 nM WT C2 domain scan. WT and mutations displayed overlapping spectra consistent with ␤ -sheet structure.

Calcium binding assay
To measure the calcium-binding capacity of WT cPLA 2 ␣ -C2 and the point mutants, the calcium detector Bis-Fura-2 (Life Technologies, Carlsbad, CA) was used according to the manufacture's protocol. Briefl y, 2 M protein was incubated for 30 min with control or 10 M Ca 2+ standard in a black fl uorescent plate with a clear bottom (Costar Life Science, Tewksbury, MA). The difference in the unknown Ca 2+ concentration was determined in relation to a standard curve by measuring the ratio of the emission at 510 nm at excitation wavelengths of 350 nm and 380 nm, respectively. Percent binding was normalized to the average WT binding capacity. Measurements were performed in triplicate for WT and mutations to determine the standard deviation (supplementary Fig. IB).

Electron microscopy of liposome morphology changes induced by the C2 domain
EM has been used to effectively characterize changes in liposome morphologies induced by the ENTH ( 28 ), BAR ( 29 ), ACCH ( 31 ), and C2 domains ( 32 ). To assess the ability of the C2 domain of cPLA 2 ␣ to induce changes to liposome morphologies, we used transmission EM (TEM) with negative staining to visualize liposomes before and after incubation with the C2 domain. The C2 domain induced dramatic changes in POPC liposome structures as long tubules were extensively visualized through the grids ( Fig. 2 ) . Moreover, the tubulation of liposomes induced by the C2 domain was Ca 2+ dependent; experiments performed in the presence of 100 M EGTA in place of CaCl 2 did not display detectable changes in liposome morphology ( Fig. 2 ). The ENTH, BAR, and ACCH domains insert into the hydrocarbon region of the membrane bilayer, which is a prerequisite for their ability to induce membrane curvature changes ( 31,(37)(38)(39). To test if hydrophobic and aromatic residues, which typically insert into membranes, were required for liposome morphology changes, we prepared mutations of hydrophobic and aromatic residues in calcium binding loops 1 and 3 of the C2 domain ( Fig. 1 ). Earlier studies have demonstrated the ability of these and after introduction of C2 domain and mutants to observe changes in real time ( Fig. 4A ). In the presence of Ca 2+ , the C2 domain induced rapid fragmentation of POPC membrane sheets ( Fig. 4A ). The specifi c nature of this fi nding was confi rmed by adding the same volume of protein storage buffer to ensure that changes in volume did not induce membrane swelling or membrane fragmentation. Additionally, mutations that greatly reduced membrane structural changes in the EM or GUV assays (F35A/L39A and Y96A) also abolished membrane fragmentation of POPC sheets ( Fig. 4B ). M98A displayed similar membrane fragmentation as WT C2, but M38A did not induce detectable changes in membrane sheet structure up to 25 min after the addition of protein.
Taken together, these data indicate that the C2 domain is able to induce changes to membrane structure for small, highly curved membranes (LUVs) as well as less curved and relatively fl at membranes (GUVs and membrane sheets).
in the presence of 500 nM CaCl 2 . Indeed, under these conditions, which are closer to physiological concentrations of cytoplasmic Ca 2+ , the C2 domain induced substantial GUV vesiculation ( Fig. 3C, D ). We also assessed the ability of full-length cPLA 2 ␣ to induce membrane curvature changes to GUVs at 200 nM protein in the presence of 500 nM CaCl 2 . The full-length enzyme not only induced GUV vesiculation but also prompted extensive tubulation emanating from the GUVs.

The C2 domain induces fragmentation of membrane sheets
Membrane sheets labeled with fl uorescent dye, which represent a relatively fl at membrane surface, have been used to image membrane curvature changes for the PH domain of FAPP1 and -2 ( 30 ). Here we used POPC membrane sheets labeled with FM® 2-10 dye to assess the ability of the C2 domain to induce changes to membrane sheet structure. Membrane sheets were imaged before experiments shown in C. The P value for each protein was determined in comparison to the control in C and D (ns, not signifi cant; * P < 0.001; ** P < 0.0001) using an unpaired Student t -test. Scale bars = 5 m. estimated to be in a range of 30-35 mN/m ( 41 ). Because monolayer penetration studies with WT and mutants are performed at saturating amounts of protein where maximal binding of WT or mutants occurs, this signifi es that, even at saturating conditions of F35A/L39A and Y96A, the proteins are not signifi cantly penetrating into the membrane, whereas M38A and M98A have reduced penetration compared with WT. In the absence of Ca 2+ , the C2 domain did not insert into the monolayer, and the c value of POPC monolayers was essentially undetectable, as previously reported ( 9,34 ). Similarly, the mutation D43N in the C2 domain, which reduces Ca 2+ binding and was unable to reconstitute FcR-mediated phagocytosis ( 23 ), also reduced the c value (20 mN/m). Thus, membrane penetration is necessary to induce membrane curvature changes, as observed in the EM, GUV, and membrane sheet assays. Likewise, phosphoinositides are necessary for the ENTH, PH, and ACCH domains to suffi ciently penetrate membranes and induce membrane deformation ( 31,39,42 ).

Membrane penetration and lipid binding affi nity of C2 domain and mutations
Membrane penetration of C2 domain and mutations into POPC monolayers was detected by injecting protein into the subphase buffer at varying initial surface pressure ( 0 ) values ( Fig. 5A, B ). This allows determination of the critical pressure ( c ), which is the pressure up to which the protein penetrates (x-intercept) ( 35 ). As previously reported ( 9,34 ), the WT C2 domain robustly penetrated a POPC monolayer with a value of ‫ف‬ 36 mN/m. In contrast, F35/L39A and Y96A, which abrogate membrane curvature changes, also signifi cantly reduce the ability of the C2 domain to penetrate POPC monolayers, with c values of 23 and 20 mN/m, respectively. In addition, M38A and M98A, which had slightly reduced membrane-deforming capabilities, slightly reduced membrane penetration, with c values of 30 and 31 mN/m, respectively. These results demonstrate that the C2 domain can effectively penetrate physiological bilayers because the surface pressure of cell membranes and LUVs is WT C2 domain and displayed a spectra indicative of ␤ -sheet with an energy minima at ‫ف‬ 215 nm. To rule out changes in calcium binding for the mutants, we quantifi ed the calcium binding ability of WT, F35A/L39A, M38A, and M98A. Mutations had comparable calcium binding ability to WT, with M38A and M98A displaying slightly reduced binding (not statistically signifi cant) (supplementary Fig. IB). Taken together, our data indicate that membrane affi nity of the C2 domain and respective mutations correlates with the ability to penetrate membranes and induce membrane curvature changes.

DISCUSSION
Traffi cking of membrane vesicles, endocytosis, and lipid-enveloped viral egress are a few of the cellular pathways where major membrane curvature changes are necessary. These are highly dynamic processes that cannot occur spontaneously because a signifi cant energy barrier has to be overcome to shape the lipid bilayer into a highly curved vesicle ( 43,44 ). To overcome this energy barrier, To quantitatively assess the effect of mutations on the ability of the C2 domain to bind POPC vesicles, we performed SPR assays ( Fig. 5C, D ). A blank surface was used as a control because it has been shown previously that the C2 domain of cPLA 2 ␣ does not exhibit nonspecifi c binding to the L1 chip surface ( 32,34 ). The dissociation constants ( K d s), obtained in triplicate, demonstrate that the C2 domain was associated with 21 ± 4 nM affi nity to POPC vesicles at 50 M CaCl 2 , but binding was not detectable up to 5 M in the presence of 100 M EGTA (data not shown). Mutations that abolish membrane-deforming activity (F35A/L39A and Y96A) demonstrate >140-and 57-fold increases in K d ( Fig. 5D ), consistent with their role in membrane penetration and in inducing alterations in membrane structure. Lastly, mutations that slightly reduced membrane curvature changes and membrane penetration (M38A and M98A) increase the K d by 6.2-and 4.4-fold, respectively. To rule out misfolded C2 domain mutations, we used CD to determine the CD spectra of each mutant in comparison to WT. As shown in supplementary Fig. IA, CD spectras from all mutations overlapped with that of the . B: Insertion of the wild-type C2 domain (fi lled circles), Y96A (fi lled squares), or M98A (fi lled triangles). All measurements were performed in the presence of Ca 2+ . C: The normalized saturation response ( R eq ) from WT cPLA 2 ␣ -C2 (fi lled circles), M38A (fi lled circles), or Y96A (fi lled squares) binding at each respective protein concentration was plotted versus C2 to fi t with a nonlinear least squares analysis of the binding isotherm [ R eq = R max /(1 + K d / C )] to determine the K d . D: K d values for WT and respective mutations binding to POPC vesicles. The binding experiments were completed from independent experiments in triplicate and are listed with their respective K d ± SD. and vesicle aggregation ( 51 ), could induce lipid demixing of PS in POPC:POPS vesicles ( 52 ), which is thought to induce positive bilayer curvature changes.
Here we demonstrate that the C2 domain of cPLA 2 ␣ , long appreciated as a high-affi nity target for zwitterionic lipids ( 9 ) with an ability to deeply penetrate the hydrocarbon core of zwitterionic membranes, is able to induce substantial changes to membrane structure. Imaging of liposomes with TEM, GUVs with confocal microscopy, or membrane sheets with confocal microscopy demonstrated dramatic changes in membrane structure induced in a Ca 2+ -dependent manner by the cPLA 2 ␣ C2 domain. Detectable changes in membrane structure correlated strongly with membrane penetrating ability and lipid binding affi nity. Membrane penetration of the C2 domain generated positive membrane curvature, as evidenced in the TEM and confocal assays ( Fig. 6 ) . Positive curvature generation by the C2 domain is consistent with the role of cPLA 2 ␣ in formation of the phagosome, Golgi tubulation, and Golgi vesiculation, which occur by budding into the cytoplasm. The protein concentrations of C2 domain used in the membrane curvature assays were similar or lower than most of the previous studies in the literature ( 28,29,32 ), supporting the specifi c nature of our fi ndings. For the C2 domain, there may be a threshold affi nity and depth or extent of penetration that is responsible for generating curvature because M38A and M98A, which reduced membrane penetration and affi nity, induced statistically significant changes in membrane curvature. In the sheet assay, M38A did not induce detectable membrane fragmentation, whereas M98A did. M38A is located in CBL1 and has lower affi nity than M98A, which is located in CBL3 protein-mediated effects or lipid bilayer asymmetry can mediate curvature changes. Protein-induced changes are often facilitated by insertion of proteins into the membrane bilayer or scaffolding of proteins on the membrane surface through oligomerization ( 45 ). Lipid-mediated changes in bilayer structure can be attributed to lipid asymmetry mediated by cone-shaped and inverted-cone-shaped lipids, where lipid shape can induce positive or negative curvature changes ( 44 ).
Membrane curvature changes induced by lipid binding domains were fi rst appreciated with the discovery of the ENTH domain and its ability to induce changes to liposome structure in a PI(4,5)P 2 -dependent manner ( 28 ). The ENTH domain deeply penetrates membranes with a N-terminal amphipathic ␣ -helix and forms oligomers on the membrane ( 37 ), both of which are necessary for effective membrane tubulation. This activity is essential to endocytosis and clathrin-coated vesicle formation, which requires substantial changes to plasma membrane structure to form highly curved membrane vesicles ( 46 ). Subsequently, the discovery of the BAR domains of amphiphysin ( 29 ) and endophilin ( 47 ) led to the notion that intrinsic curvature from the crescent moon-shaped BAR domains is essential to remodeling membranes. This led to further investigation, which demonstrated that BAR domains form elegant scaffolds on the membrane where mutation of residues that mediate scaffolding abrogates membrane curvature changes ( 48,49 ). Additionally, as with the ENTH domain, some BAR domains have a N-terminal ␣ -helix that can penetrate into the membrane in addition to a second predicted amphipathic alpha-helix that resides on the membrane binding interface ( 38 ). The depth and orientation of this penetration may also be important in regulating membrane curvature changes and membrane fi ssion ( 50 ). For instance, it was recently shown that insertion of the amphipathic ␣ -helix drives vesiculation and thus scission by the ENTH domain. In contrast, an antagonistic relationship between the number and length of amphipathic helices in BAR domains was discovered where membrane fi ssion can be restricted by the BAR domains' crescent shape ( 50 ). Taken together, depth and area of insertion as well as inherent protein scaffold shape play a critical role in the type of membrane curvature generated and whether or not membrane fi ssion will proceed.
PH domains ( 30 ) and C2 domains ( 32 ) have also been shown to induce membrane curvature changes. The FAPP1 and 2 PH domains require insertion of a turret loop adjacent to the PI(4)P binding pocket ( 42 ) to induce membrane remodeling where the inherent shape of FAPP1 or -2 may also play a critical role ( 30 ). However, unlike the ENTH and BAR domain, elegant models of membrane scaffolding and modes of membrane curvature induction for PH and C2 domains have not been investigated. In addition, the relationship between membrane penetration of C2 domains and membrane curvature changes is unknown. Recently, it was shown that the C2B domain alone or the tandem C2AC2B domains of synaptotagmin 1, which can induce membrane tubulation ( 32 ) Fig. 6. Membrane penetration by the C2 domain of cPLA 2 ␣ is suffi cient to induce membrane curvature changes. The hydrophobic residues essential in penetrating the membrane are also key for membrane curvature generation. The deep ‫ف‬ 15 Å penetration of these hydrophobic and aliphatic residues as well as a signifi cant area of insertion ( ‫ف‬ 2110 Å 2 ) are suffi cient to reduce the energetic barrier to bend the membrane as deletion of one of these key residues abolishes this effect, as shown for the F35A/L39A and Y96A mutants. Although the overall mechanism is unknown, our data suggest that membrane penetration of the C2 domain is vital for membrane bending, tubulation, vesiculation, and fragmentation, depending on the initial curvature of the membrane. changes. It was fi rst thought that the N-terminal ␣ -helix insertion for ENTH domains, and to some degree for BAR domains, was the major cause of the membrane curvature induction. However, more recent and sophisticated studies have demonstrated the ability of these proteins to scaffold on the membrane ( 48,49 ). This scaffolding is essential for in vitro and cellular observations of membrane curvature changes ( 48,49 ). Future studies need to be directed toward resolving the molecular details of C2 oligomerization as well as the role of ␤ -groove C1P binding in membrane curvature changes or membrane reorganization. Additionally, the type of membrane curvature generated by the C2 domain as well as full-length cPLA 2 ␣ will require extensive analysis using a combination of biophysical, biochemical, and cellular assays. Investigating how the C2 domain insertion and catalytic domain generation of lysophospholipids contribute to in vitro and cellular membrane curvature changes should solve a number of questions in the fi elds of membrane traffi cking and phagosome formation.
The type of curvature generated by the C2 domain or full-length enzyme may also be key to normal and aberrant physiological processes linked to cPLA 2 ␣ activity. cPLA 2 ␣ has been shown to function in generation of cholesterolrich, GPI protein-containing endosomes ( 26 ), Golgi tubulation and vesiculation ( 24,25 ), and Golgi-to-PM traffi cking ( 60 ). Additionally, cPLA 2 ␣ association with the Golgi has been shown to regulate the function of endothelial cells ( 61,62 ), which serve a barrier function in the luminal surfaces of blood vessels. Proliferation of endothelial cells has been shown to occur for the formation of blood vessels in wound healing and tumor formation while blocking cPLA 2 ␣ activity through an inhibitor or siRNA blocks endothelial cell proliferation and cell cycle entry ( 61 ). In terms of pathophysiological states, cPLA 2 ␣ has been linked to diseases such as asthma ( 4 ), arthritis ( 6 ), and cancers ( 5 ). Thus, up-or down-regulation of cPLA 2 ␣ enzyme levels may alter the transport of vesicles from the Golgi to the PM, modify intra-Golgi transport, or effect endothelial barrier function through the combined generation of fatty acids and lysophospholipids and membrane penetration of the C2 domain of cPLA 2 ␣ . Additionally, because cPLA 2 ␣ enzyme inhibitors act as an allosteric block that reduces or precludes cPLA 2 ␣ membrane translocation, it is diffi cult to rule out the C2 domain-mediated effects. Thus, our data support a model where vesiculation or tubulation of the Golgi may occur in response to C2 domain membrane penetration of the full-length enzyme under conditions of low cPLA 2 ␣ activity. Linkage of biochemical and biophysical studies in vitro and in cells with cellular and disease models that can tease apart the role of C2 domain penetration and cPLA 2 ␣ activity in these processes will be key to unraveling the full mechanism of membrane curvature generation. ( Fig. 1 ). It has been shown that CBL1 penetrates more deeply into the bilayer than CBL3 ( 53 ), which could perhaps play a role in the different observations in the membrane sheet assays. Although the origin of this discrepancy is unknown, it leaves room for extensive biophysical studies of C2 domain parameters required for membrane curvature generation. It also appears that M38A and M98A may cause some vesiculation in the liposome assays, as visualized by EM, as well as differences in the extent of vesiculation in the GUV assays, suggesting that the depth of penetration and/or membrane affi nity may play an important role in the type of membrane curvature or membrane reorganization that is generated. Studying the C2 domain's role in the type of curvature generation in conjunction with membrane fi ssion ( 50 ) will be essential to improve our understanding of the mechanism of curvature generation for this C2 domain. Penetration of hydrophobic residues by the C2 domain occurs into the hydrocarbon layer ( ‫ف‬ 15 Å), reminiscent of the ENTH domain ( 37,39 ), so it is possible that membrane fi ssion and vesiculation ( 50 ) may occur, which is supported by membrane fragmentation in the membrane sheet assays. C2 domains and PH domains may also induce different types of curvature. For instance, in this study, the C2 domain induced membrane fragmentation, whereas studies with PH domains have demonstrated extensive positive curvature induction in the form of tubules in the membrane sheet assays ( 42 ).
Future studies need to consider the role of C2 domain membrane binding and penetration in inducing or contributing to membrane curvature changes in conjunction with cPLA 2 ␣ activity. The fact that the C2 domain alone is able to reconstitute FcR-mediated phagocytosis ( 23 ) suggests that the cPLA 2 ␣ C2 domain has a high membrane remodeling activity that is essential to membrane reorganization. Additionally, inhibition of cPLA 2 ␣ activity in cells with the inhibitor pyrrophenone generated an allosteric block that prevented cPLA 2 ␣ translocation ( 23 ), which does not allow one to account for C2 membrane binding and insertion in assessing the generation of lysophospholipids in curvature generation ( 26 ). Nonetheless, the prior study demonstrated the C2 domain process is Ca 2+ dependent because D43N, which abrogates calcium binding, could not reconstitute phagocytosis. In conjunction with the current study, this strongly suggests that membrane penetration of the C2 domain is necessary for these effects because calcium is required for membrane penetration of the C2 domain ( Fig.  1C and 5A ) ( 9 ). It is also now well established that the cationic ␤ -groove of the C2 domain binds C1P ( 21 ), which is important for cPLA 2 ␣ activity ( 21 ) and cellular translocation ( 54 ). Additionally, the role of ceramide kinase and its product C1P are key players in phagocytosis ( 55,56 ). Thus, it has been hypothesized that C1P has an important role in recruitment of cPLA 2 ␣ in phagocytosis ( 57 ). To this end, it is tempting to speculate that cPLA 2 ␣ may be able to induce reorganization or clustering of membranes harboring C1P.
The surface area of insertion for the C2 domain ( 58 ) is more substantial than that of the ENTH ( 39,50 ) and PH domains ( 59 ), at least for a monomer; however, this alone may not account for the membrane-mediated curvature