INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Phospholipase A₂ (PLA₂) catalyzes the hydrolysis of the sn-2 ester bond of glycerophospholipids and generates free fatty acids and lysophospholipids that serve as precursors for lipid-derived mediators with a wide range of biological activities (Gelb et al., 1995, 1999; Tischfield, 1997; Dennis, 2000). Many fatty acids themselves act as bioactive mediators (Goodfriend and Egan, 1997; Forest et al., 1997). Eicosanoids, the oxygenated metabolites of arachidonic acid, play key roles in normal and pathological cell functions including cell signaling, inflammation, allergy, apoptosis, and tumorigenesis (for recent review see Heller et al., 1998; Dennis, 2000). The other product of PLA₂, lysophospholipid, may be metabolized either to platelet-activating factor, which is known as a potent inflammatory and allergic mediator (Kume and Shimizu, 1997; Jackson et al., 1998), or to lysophosphatidic acid, a signaling molecule with mitogenic activities (Fourcade et al., 1998; Gennaro et al., 1999).

Secretory PLA₂s constitute a large family of structurally and mechanistically related enzymes with relative molecular masses of 13-16 kDa. They are widespread in various mammalian cells and tissues, as well as in snake, lizard, and insect venom, and are divided into several groups and subgroups based on their amino acid sequences, disulfide bonding patterns, tissue distribution, and functional properties (Heinrikson, 1991; Tischfield, 1997; Maxey and MacDonald, 1998; Dennis, 1997, 2000). These enzymes perform phospholipid hydrolysis using a His-Asp doublet plus a conserved water molecule as a nucleophile and a Ca²⁺ ion as a cofactor. Secretory PLA₂s undergo a substantial increase in their catalytic activity upon binding to the surface of phospholipid membranes or micelles (Pieterson et al., 1974; Verger and de Haas, 1976; Jain and Berg, 1989; Gelb et al., 1995, 1999). Studies on the molecular mechanism of interfacial activation of PLA₂s led to conceptually diverse interpretations of this effect. According to one interpretation (the substrate hypothesis), the physical properties of the membrane, including membrane fluidity, curvature, surface charge, and others were considered as major determinants of the activation of PLA₂ at the membrane surface. The other (enzyme hypothesis) was that conformational changes in PLA₂ are primarily responsible for the interfacial activation of the enzyme. Indeed, unequivocal evidence has been provided for the importance of the physical state of the aggregated substrate in the activation of secretory PLA₂s (Verger and de Haas, 1976; Thuren et al., 1984; Jain and Berg, 1989; Burack and Biltonen, 1994; Burack et al., 1993, 1995; 1997; Gelb et al., 1995, 1999; Berg et al., 1997). The abrupt increase of PLA₂ activity in the presence of zwitterionic phospholipid vesicles or micelles is preceded by a dormant period that can be reduced, or abolished, by modifying the physical properties of the aggregated substrate, for example, by increasing the anionic surface charge of the membranes (Jain et al., 1982, 1986, 1989; Apitz-Castro et al., 1982; Volwerk et al., 1986; Burack and Biltonen, 1994). Strong effects of non-ionized fatty acids, lysophosphatidylcholine, cholesterol, diacylglycerol, or phosphatidylethanolamine on PLA₂ activity suggested that, apart from electrostatic effects, perturbations of the membrane structure by these agents are crucial for PLA₂ activation (Jain and de Haas, 1983; Bell and Biltonen, 1992; Bell et al., 1996; Henshaw et al., 1998; Liu and Chong, 1999).

Considerable efforts have been directed to the characterization of structural changes in PLA₂ involved in the enzyme activation. X-ray crystallography revealed similar structures of secretory PLA₂s with and without bound monomeric substrate analogs (Brunie et al., 1985; Scott et al., 1990a,b; White et al., 1990; Thunnissen et al., 1990; Cha et al., 1996; Sekar et al., 1997), which was considered as evidence against structural changes in the enzyme during its interfacial activation (Scott and Sigler, 1994). Small (~1 Å) structural changes in PLA₂s upon binding of monomeric inhibitors have been detected by several x-ray studies (Scott et al., 1991; Tomoo et al., 1994; Schevitz et al., 1995); however, these changes were considered by others as insignificant (Cha et al., 1996; Sekar et al., 1997). Interpretation of x-ray results in the context of interfacial activation of PLA₂s is not straightforward because PLA₂s are activated by binding to the surface of phospholipid bilayers or micelles but not upon binding of the monomeric substrate; the binding of the substrate to the active site of the membrane-bound enzyme might have quite different structural consequences. NMR experiments revealed that in porcine pancreatic PLA₂ (group IB) the N-terminal helix and the catalytically important residues His⁴⁸ and Asp⁹⁹ adopt a fixed conformation only in a ternary complex of the enzyme with an inhibitory substrate analog and dodecylphosphocholine micelles (Peters et al., 1992; van den Berg et al., 1995), which might be implicated in more productive enzyme-substrate complex formation (Yu et al., 1999). Evidence for a possible allosteric coupling between the interfacial adsorption and catalytic machinery of PLA₂s has also been provided by fluorescence spectroscopy. Distinct shifts in the intrinsic Trp fluorescence of PLA₂ has been detected during the activation of the membrane-bound enzyme (Jain and Maliwal, 1993; Bell and Biltonen, 1989; Burack and Biltonen, 1994; Burack et al., 1995). Combined site-directed mutagenesis and spectroscopic studies showed that substitutions of residues in the interfacial adsorption surface (i-face) of a pancreatic PLA₂ affect both the enzyme-substrate interaction constant (K_S allostery) and the rate constant of the catalytic turnover (k^*_cat allostery), implying an allosteric effect that propagates from the i-face to the catalytic residues of PLA₂ (Rogers et al., 1998; Yu et al., 1999). Although these data provide evidence that interfacial activation of PLA₂ may involve conformational changes in the enzyme, the nature of these conformational changes and their relation to the physical properties of membranes are not well understood.

Our earlier attenuated total reflection Fourier transform infrared (ATR-FTIR) studies identified modification of the -helices in a group IIA PLA₂ upon binding to lipid bilayers (Tatulian et al., 1997). In this work, the advantages of ATR-FTIR spectroscopy have been further exploited to establish a relationship between the surface properties of membranes and membrane-induced structural changes in PLA₂. The data indicate that both the strength and cooperativity of PLA₂ binding to membranes, as well as PLA₂ activity, increase at higher negative surface potentials of membranes. Phospholipid hydrolysis by PLA₂ is followed by preferential removal of the lysophospholipid and accumulation of the fatty acid in the membrane that could modulate the enzyme activation either through increasing negative electrostatic potential at the membrane surface or by affecting the membrane morphology and stability. When PLA₂ was applied to bilayers composed of an equimolar mixture of dipalmitoylphosphatidylglycerol (DPPG) and dipalmitoylphosphatidylcholine with fully deuterated acyl chains [DP(d₆₂)PC], both lipids were hydrolyzed at similar efficiencies, indicating that membrane surface electrostatics, rather than specific recognition of acidic lipids by the enzyme, plays a major role in increased activity of PLA₂ toward negatively charged membranes. A correlation has been established between the induction of previously described modified helices in PLA₂ during interfacial activation (Tatulian et al., 1997) and negative surface charge density of membranes. These findings delineate a reciprocal relationship between membrane electrostatic properties, membrane binding strength of PLA₂, and membrane-induced structural changes in the enzyme that contribute to PLA₂ activation in a synergistic manner.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials

The secretory PLA₂ has been purified from the venom of the snake Agkistrodon piscivorus piscivorus according to Maraganore et al. (1984) and was kindly supplied by Dr. R. L. Biltonen of the Department of Pharmacology of the University of Virginia School of Medicine. The lipids were purchased from Avanti Polar Lipids (Alabaster, AL) and the other chemicals from Sigma (St. Louis, MO).

Preparation of supported membranes

Supported lipid bilayers for ATR-FTIR experiments were prepared on a 1 × 20 × 50 mm³ germanium internal reflection plate (Spectral Systems, Irvington, NY) using two different techniques. The plate was washed by chloroform and methanol and processed in an argon plasma cleaner (Harrick, Ossining, NY) immediately before use. The first technique involved preparation of a monolayer of phosphatidylcholine at the surface of an aqueous buffer (10 mM Tris/acetic acid, pH 5) in a Langmuir trough (model 611, Nima, Coventry, UK). The monolayer was deposited onto the germanium plate by slowly (~2 mm/min) withdrawing the plate from the aqueous phase vertical to the air/water interface. The plate with the monolayer was assembled in a perfusable liquid ATR cell. Vesicles of desired lipid composition were prepared either by sonication, using a Branson tip sonicator, or by extrusion through 100-nm pore size polycarbonate membranes using a Liposofast extruder (Avestin, Ottawa, Canada). Vesicles were injected into the ATR cell that contained the germanium plate with the monolayer and incubated for ~1.5 h to allow the vesicles to spread on the lipid monolayer and yield supported bilayers. This was followed by gently flushing the ATR cell with buffer and washing the excess lipid out of the cell. The advantage of this method is that it can be used for preparation of either symmetric or asymmetric membranes, depending on the choice of the lipids for preparation of the monolayer and the vesicles. Its disadvantage is that the membrane leaflet facing the plate cannot include acidic lipids because in that case the monolayer does not efficiently adsorb to the germanium plate, probably due to electrostatic effects. The second procedure that has been employed in this study permitted preparation of supported bilayers containing acidic lipids in both leaflets. According to this procedure, sonicated phospholipid vesicles are prepared that contain 20% anionic lipid (e.g., phosphatidylglycerol) in a buffer containing ~5 mM CaCl₂. When the vesicles are injected into the ATR cell that contains a bare germanium plate and are incubated for ~1 h, a lipid bilayer is formed at the surface of the germanium plate that is presumably stabilized by Ca²⁺ bridges between the acidic lipids and the germanium plate, which is hydrophilized by argon ion plasma processing. After preparation of supported bilayers, PLA₂ was injected into the ATR cell and allowed to adsorb to the supported membranes for 5-10 min, followed by recording of ATR-FTIR spectra. Protein concentration was measured by the Bradford assay (Bradford, 1976).

ATR-FTIR experiments

ATR-FTIR experiments were carried out on a Nicolet 740 infrared spectrometer (Nicolet Analytical Instruments, Madison, WI) using a liquid-nitrogen-cooled mercury/cadmium/telluride detector at a nominal spectral resolution of 2 cm¹. A four-mirror model 57 single-beam ATR system was used (Buck Scientific, East Norwalk, CT). Normally, 1000 scans were co-added to achieve a reasonably good signal-to-noise ratio of the spectra. The incident infrared light was polarized using a gold grid polarizer (Perkin-Elmer, Beaconfield, UK). To obtain spectra including both the lipid and the protein components in the sample, the single-beam spectra of the buffer in the ATR cell with the germanium plate were used as reference. The absorbance spectra of the membrane-bound protein in the pure form were obtained by using as reference the single-beam spectra of the supported membranes that were measured before injection of the protein. These latter spectra were free of any contributions of the lipid to the spectral regions of the protein absorbance bands. The measurements were preceded by extensive purging of the instrument with dry air to remove humidity (H₂O vapors) and CO₂ and to minimize their interference with the spectra.

Data analysis

The exponentially decaying evanescent field that is created at the germanium/membrane interface at each internal reflection of the infrared beam makes it possible to detect all membrane components, including the membrane-bound protein, while the molecules far from the membrane do not contribute to the ATR-FTIR spectra (Fig. 1). This makes the ATR-FTIR spectroscopy a uniquely well suited technique for quantitative characterization of protein binding to supported membranes, the enzymatic activity of PLA₂, selective hydrolysis of different lipid components in membranes, and dissociation of lipid hydrolysis products from the membrane.

View larger version (50K): [in this window] [in a new window]   FIGURE 1   Schematic depiction of an ATR sample cell with an internal reflection plate (the yellow trapezoid in the center) that has lipid bilayers at both surfaces. The protein molecules are shown as red ellipsoids. The infrared beam is shown to enter the plate, perform several internal reflections, and exit the plate. The exponentially decaying evanescent field that is created at each internal reflection illuminates the membrane and the membrane-bound proteins whereas the molecules far from the membrane are "invisible" and do not contribute to the absorption spectrum. The outlet of one-half of the ATR sample cell is connected to the inlet of the other half, allowing for simultaneous perfusion of the whole cell.

The activity of PLA₂ toward the supported lipid membranes was evaluated based on a PLA₂-concentration-dependent decrease in the intensity of lipid absorbance bands, which was shown to result from the partial removal of the lipid hydrolysis products from the membrane. The methylene symmetric stretching bands were integrated between 2878 and 2830 cm¹ (or between 2111 and 2071 cm¹ for deuterated lipid acyl chains) and plotted as a function of PLA₂ concentration. The less intense symmetric methylene band was used because it, unlike its more intense asymmetric counterpart, is generated by an isolated vibrational mode and is free of Fermi resonance contributions (Rana et al., 1993). The removal from the supported membranes of the free fatty acid or the lysophospholipid were determined from changes in the olefinic CH stretching bands, which were integrated between 3023 and 2996 cm¹ (for lipids containing unsaturated sn-2 chains), and of the phosphate symmetric stretching bands integrated between 1106 and 1078 cm¹, respectively. To estimate the removal from the membranes of PLA₂-generated free fatty acid for lipids with fully saturated hydrocarbon chains, selectively sn-1-chain deuterated lipids were used.

The protein/lipid (P/L) molar ratios in supported membranes were determined using the ratio of integrated intensities of the protein amide I and the lipid methylene stretching bands at perpendicular polarization of the infrared radiation, A_P and A_L, which was corrected for corresponding molar extinction coefficients, _P and _L, and for the orientation factors, _i:

(1)

In Eq. 1, _i = (S_isin²_i)/2 + (1  S_i)/3, where i = P or L, S_i is the corresponding orientational order parameter, and _i is the angle between the corresponding transition dipole moment and the molecular director. The subscripts P and L signify protein and lipid, respectively, n_P is the number of peptide bonds in the protein and n_L is the number of methylene groups in the lipid hydrocarbon chains. A value of _L = 4.7 × 10⁶ cm/mol per CH₂ group of the lipid has been used (Fringeli et al., 1989). The amide I molar extinction coefficients of proteins depend on their secondary structure. A weighted average of _P = 5.7 × 10⁷ cm/mol per peptide bond of PLA₂ was found assuming that the protein secondary structure incorporates 50% , 10% , and 40% irregular structure (Arni and Ward, 1996; Han et al., 1997) and using the corresponding integrated molar extinction coefficients (Venyaminov and Kalnin, 1990). The number of protein molecules per unit area of the membrane was determined using the protein/lipid molar ratio as

(2)

where A is the cross-sectional area per lipid molecule; A = 50 Å² was used for dipalmitoylphosphatidylcholine (DPPC) and DPPG (Seddon, 1993). PLA₂ binding to supported membranes was quantitatively characterized by plotting n against PLA₂ concentration and by describing these plots using a Langmuir-type adsorption isotherm supplemented with the Hill cooperativity coefficient:

(3)

where N is the number of binding sites per unit area, C is the PLA₂ concentration, K is the apparent binding constant, and _H is the Hill coefficient. The values of N were found from extrapolated intersections of the n/C versus n plots with the n/C = 0 line. The Hill coefficients and the dissociation constants (1/K) were determined respectively as the inverted slopes of the Ln(N/n  1) versus LnC plots and the PLA₂ concentrations corresponding to their intersections with the Ln(N/n  1) = 0 line, i.e., when n = N/2.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Quantitative characterization of membrane binding of PLA₂

Adsorption isotherms characterizing the binding of PLA₂ to supported bilayers containing DPPC in the lower (facing the germanium plate) leaflet and a 3:2 mixture of DPPC and DPPG in the upper leaflet at different ionic strengths were obtained by measuring the surface density of membrane-bound PLA₂, n, as a function of PLA₂ concentration (Fig. 2). As shown in Fig. 3 A, at low ionic strengths the n/C versus n dependencies were concave downward, indicating positive cooperativity in PLA₂ binding to negatively charged membranes (Cantor and Schimmel, 1980). The binding parameters K, N, and _H were determined as described above and were used to calculate the theoretical curves of Fig. 2 using Eq. 3. The data of Table 1 and the curves presented in Fig. 2 demonstrate that at low ionic strengths the enzyme binding to membranes is saturable and cooperative. All three binding parameters, i.e., the binding constant, the density of binding sites, and the Hill coefficient, decrease at higher ionic strengths.

View larger version (23K): [in this window] [in a new window]   FIGURE 2   Binding of PLA2 to supported membranes composed of DPPC at the lower leaflet and a 3:2 mixture of DPPC and DPPG at the upper leaflet. The buffer contained 5 mM Tris (pH 8.2), 0.5 mM EGTA, 1 mM NaN3 plus 0, 0.01, 0.1, or 1 M NaCl (curves 1-4, respectively). The curves are simulated by Eq. 3 using the parameters summarized in Table 1.

View larger version (20K): [in this window] [in a new window]   FIGURE 3   Scatchard plots that have been used to evaluate the parameters describing the binding of PLA2 to supported bilayers containing 60 mol % DPPC and 40 mol % DPPG. Curves 1-4 correspond to NaCl concentrations 0, 0.01, 0.1, or 1 M added to the buffer: 5 mM Tris (pH 8.2), 0.5 mM EGTA, 1 mM NaN3. The numbers of binding sites per unit area (N) were found from extrapolated intersections of the curves of A with the n/C = 0 line. The Hill coefficients (H) and the dissociation constants (1/K) were determined as the inverted slopes of the lines of B and the PLA2 concentrations corresponding to the intersections with the line Ln(N/n  1) = 0.

Determination of PLA₂ activity by ATR-FTIR spectroscopy

Supported membranes that contained 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) in the lower (facing the germanium plate) leaflet and a 4:1 mixture of POPC and 1-palmitoyl-2-oleoylphosphatidylglycerol (POPG) in the upper leaflet were prepared and flushed several times with buffer. After each flush, polarized ATR spectra were recorded as control measurements. This was followed by injection of PLA₂ that resulted in the appearance of a prominent amide I absorbance band, indicating binding of the enzyme to the membrane (Fig. 4). The intensity of the lipid methylene stretching band gradually decreased at each flush with buffer and then was stabilized, reflecting removal of excess lipid from the membrane. Binding of PLA₂ to the supported membrane was accompanied by a concomitant abrupt decrease in the intensity of the lipid signal (Fig. 5). This result is interpreted in terms of PLA₂-catalyzed lipid hydrolysis and dissociation of a fraction of the reaction products from the membrane. Several lines of evidence confirm this suggestion. First, before injection of PLA₂ the membrane was flushed with the buffer until the lipid signal was stabilized; i.e., additional flushes without PLA₂ did not affect the lipid signal (Fig. 5). Second, when PLA₂ was inhibited by EGTA, or when nonhydrolyzable lipids were used to prepare supported bilayers, such as dipalmitoyl-glycerol (DPG), dihexadecylphosphatidylcholine (DHPC) in combination with cardiolipin (CL) or arachidic acid (AA), PLA₂ did not cause any significant decrease in the lipid signal (Fig. 6). Third, partial inhibition of PLA₂ by ZnCl₂ (Mezna et al., 1994; Yu et al., 1998) substantially reduced the effect of PLA₂ on the lipid methylene band intensity (cf. Fig. 6 E and Fig. 7 B). These experiments demonstrate that the decrease in the lipid methylene stretching band intensity reflects PLA₂ activity that can be measured by ATR-FTIR spectroscopy.

View larger version (34K): [in this window] [in a new window]   FIGURE 4   ATR-FTIR spectra of a supported bilayer containing POPC in the lower leaflet and a 4:1 mixture of POPC and POPG in the upper leaflet subjected to several flushes with buffer followed by injection of 5 µM PLA2. The buffer contained 5 mM Hepes (pH 8.2), 100 mM NaCl, 15 mM KCl, 2 mM CaCl2. The lipid methylene and carbonyl stretching bands and the protein amide I band are marked. Note a decrease in the lipid signal and appearance of a strong protein amide I signal following injection of PLA2, indicating binding of the enzyme to the membranes and lipid hydrolysis.

View larger version (25K): [in this window] [in a new window]   FIGURE 5   Dependence of the integrated intensities of the lipid methylene and the protein amide I bands on flushes with buffer and injection of 5 µM PLA2, as described in Fig. 4.

View larger version (30K): [in this window] [in a new window]   FIGURE 6   Lipid methylene stretching bands at different concentrations of PLA2. The lipid composition of the asymmetric supported bilayers was as follows. (A) DPPC/(DPPC + DPPG) (4:1); (B) DPG/(DPG + CL) (9:1); (C) DPG/(DPG + AA) (4:1); (D) DHPC/(DHPC + AA) (4:1); (E) POPC/(POPC + POPG) (4:1). The buffer contained 5 mM Hepes (pH 8.2), 100 mM NaCl, and 15 mM KCl with the following additions: 0.5 mM EGTA (A), 2 mM CaCl2 (B-D), and 2 mM CaCl2 plus 0.5 mM ZnCl2 (E). In each of five families of spectra, PLA2 concentration was increased from 0 to 50 µM (from top to bottom). The spectra of DHPC/(DHPC + AA) (D) were more intense and were reduced by a factor of 2 to maintain proportionality; the others are presented as measured.

View larger version (38K): [in this window] [in a new window]   FIGURE 7   Lipid methylene (and phosphate, in the inset) stretching bands at different concentrations of PLA2. The lipid composition of the asymmetric supported bilayers was as follows. (A) PLPC/(PLPC + PLPG); (B) POPC/(POPC + POPG); (C) DPPC/(DPPC + DPPG). In all three cases, the fraction of the acidic lipid in the upper leaflet is 20 mol %. The buffer contained 5 mM Hepes (pH 8.2), 100 mM NaCl, 15 mM KCl, 2 mM CaCl2. In each of the three families of spectra, PLA2 concentration was increased from 0 to 50 µM (from top to bottom).

Differential removal from the membrane of phospholipid hydrolysis products

Because the existing experimental evidence suggests that both products of phospholipid hydrolysis by PLA₂, i.e., the free fatty acid and the lysophospholipid, contribute to the activation of PLA₂ at the membrane surface, it was interesting to quantitatively determine whether one of the two products preferentially accumulates in the membrane and plays a dominant role in the enzyme activation. Although partial removal of PLA₂ reaction products from phospholipid monolayers and bilayers has been demonstrated (Gericke and Hühnerfuss, 1994; Speijer et al., 1996; Callisen and Talmon, 1998), this question has not yet been answered.

The results of the action of PLA₂ on supported membranes of three different lipid compositions are presented in Fig. 7. In all three cases, the lipids contain a palmitic acid residue at the sn-1 position, but the sn-2 position is esterified by linoleic, oleic, and palmitic acids that contain two, one, and zero unsaturated olefinic (---HC==CH---) groups, respectively. The fractions of the total lipid components (i.e., the free fatty acid and the lysophospholipid) that remained in the membrane at each PLA₂ concentration, A_total, were determined based on the integrated intensity of the symmetric CH₂ stretching band normalized relative to the corresponding intensity in the absence of PLA₂. To determine the differential removal from the membrane of the fatty acid and the lysophospholipid that results from phospholipid hydrolysis by PLA₂, the olefinic CH stretching band at 3005-3010 cm¹ was used as a marker for the fatty acid liberated from the sn-2 position of lipids containing unsaturated sn-2 chains (Fig. 7), whereas the phosphate PO₂ symmetric stretching band at ~1090 cm¹ was used as a marker for the lysophospholipids (inset in Fig. 7). The normalized integrated intensities of respective absorbance bands were used as the fractions of retained sn-2 and sn-1 chains following phospholipid hydrolysis, i.e., A_sn-2 and A_sn-1. For lipids with unsaturated sn-2 chains, the following relationship was fulfilled: A_sn-2 + A_sn-1 = 2A_total. Therefore, for the membranes composed of DPPC and DPPG that lack olefinic groups, A_sn-2 was calculated as 2A_total  A_sn-1. Data presented in Fig. 8 demonstrate that although the lipid hydrolysis is followed by partial removal from the membrane of both the liberated fatty acid and the lysophospholipid, the fraction of lysophospholipid that is removed from the membrane significantly exceeds that of the fatty acid, implying a predominant accumulation of the fatty acid in the membrane.

View larger version (28K): [in this window] [in a new window]   FIGURE 8   Normalized integrated intensities due to the total methylene stretching mode (), sn-1 acyl chains (), and sn-2 acyl chains () of the bilayers composed of PLPC/(PLPC + PLPG) in A, POPC/(POPC + POPG) in B, and DPPC/(DPPC + DPPG) in C as a function of PLA2 concentration under conditions described in Fig. 7. For details, see the text.

Because the olefinic stretching mode is not present in lipids with fully saturated hydrocarbon chains, and its intensity is low even in lipids containing chains with one or two double bonds, a second method has been used to assess the relative depletion of the fatty acid and the lysophospholipid resulting from lipid hydrolysis. Supported bilayers were prepared using 1-palmitoyl(d₃₁)-2-palmitoylphosphatidylcholine in which the sn-1 chain is fully deuterated while the sn-2 chain is not. Substitution of methylene hydrogens by deuterium results in a >700-cm¹ shift of the methylene stretching modes toward lower frequencies, because of the heavier nuclear mass of deuterium (Fig. 9). Also, the CD₂ stretching mode is broader and approximately twofold weaker than the CH₂ mode due to the lower extinction coefficient of the former vibrational mode (Rana et al., 1993). The plots of the normalized integrated areas of the CH₂ and CD₂ symmetric stretching bands as a function of PLA₂ concentration showed that lipid hydrolysis is followed by a preferential removal of the sn-1 chain of DPPC (i.e., the lysophospholipid) whereas the sn-2 chain, which belongs to the free fatty acid, tends to stay in the membrane (Fig. 10). This result is consistent with the suggestion of the above experiments that the free fatty acid predominantly contributes to interfacial activation of PLA₂ by 1) increasing negative electrostatic potential at the membrane surface and/or 2) affecting the membrane morphology.

View larger version (21K): [in this window] [in a new window]   FIGURE 9   Methylene stretching bands of the sn-1 (right) and sn-2 (left) acyl chains of 1-palmitoyl(d31)-2-palmitoylphosphatidylcholine.

View larger version (22K): [in this window] [in a new window]   FIGURE 10   Normalized integrated methylene stretching intensities of the deuterated sn-1 and unlabeled sn-2 acyl chains of 1-palmitoyl(d31)-2- palmitoylphosphatidylcholine in supported membranes as a function of PLA2 concentration in the presence of 2 mM CaCl2.

Effect of the acidic lipid on PLA₂ activity

The effect of the acidic lipid in supported membranes on the activity of PLA₂ was studied by using bilayers composed of a mixture of POPC and POPG in which the fraction of POPG was increased form 0 to 0.5. The plots of the methylene stretching intensities as a function of PLA₂ concentration indicated that PLA₂ exhibited higher activity toward membranes with higher fractions of the acidic lipid POPG (Fig. 11). To determine whether the correlation between PLA₂ activity and the membrane negative surface charge density is due to stronger electrostatic attraction between the cationic PLA₂ and negatively charged membranes or whether this effect results from stronger affinity of the acidic lipid to the enzyme active center, experiments were conducted on supported bilayers composed of 50% DP(d₆₂)PC with deuterated acyl chains and 50% unlabeled DPPG (Fig. 12). These membranes were prepared by using the method of direct spreading of sonicated vesicles onto the bare germanium plate, which ensured an equimolar content of the zwitterionic and acidic lipids in the membranes. Dependencies of integrated intensities of CH₂ and CD₂ symmetric stretching bands on PLA₂ concentration showed that both lipids were hydrolyzed at similar efficiencies (Fig. 13), indicating that the acidic lipid is not preferentially hydrolyzed by PLA₂. Instead, the greater activity of the enzyme toward membranes containing higher fractions of acidic lipids results from stronger binding of the enzyme to the surface of membranes with higher anionic surface charge.

View larger version (23K): [in this window] [in a new window]   FIGURE 11   Normalized integrated intensities of methylene stretching bands of lipids in supported bilayers composed of POPC plus 0, 5, 20, and 50 mol % POPG (curves 1-4, respectively) as a function of PLA2 concentration in the presence of 2 mM CaCl2.

View larger version (41K): [in this window] [in a new window]   FIGURE 12   ATR-FTIR spectra of a supported membrane composed of an equimolar mixture of DPPG and DP(d62)PC as a function of PLA2 concentration, as indicated, in the presence of 2 mM CaCl2. The CH2 stretching bands of DPPG and the CD2 stretching bands of DPPC are indicated by thin rectangles.

View larger version (21K): [in this window] [in a new window]   FIGURE 13   Normalized integrated intensities of methylene stretching bands of DPPG () and DP(d62)PC () as a function of PLA2 concentration, calculated from the data of an experiment described in Fig. 12.

Correlation between membrane surface properties and membrane-induced structural changes in PLA₂

Our earlier studies identified significant differences between the amide I bands of free and membrane-bound PLA₂ (Tatulian et al., 1997). The second-derivative spectrum of the free enzyme demonstrated a major component at ~1650 cm¹, indicating a predominantly -helical structure for the protein (Mendelsohn and Mantsch, 1986; Arrondo et al., 1993; Jackson and Mantsch, 1995), whereas the -helical signal of the membrane-bound protein was split into two subcomponents at ~1658 and ~1650 cm¹. Less stable -helices are characterized by stronger carbonyl stretching force constants because of weaker helical hydrogen bonding and, consequently, their amide I vibrational mode occurs at higher frequencies (Dwivedi and Krimm, 1984). Therefore, the appearance of the higher-frequency signal in the -helical region of the amide I band of PLA₂ is interpreted in terms of increased flexibility of the -helices of membrane-bound PLA₂. The resolution-enhanced (second-derivative) amide I spectra of PLA₂ bound to supported membranes of POPC containing 0, 5, 20, and 50% POPG indicated a clear correlation between the intensity of the component at 1658 cm¹ and the fraction of the acidic lipid in the membrane (Fig. 14).

View larger version (28K): [in this window] [in a new window]   FIGURE 14   Second derivatives of the ATR-FTIR spectra in the amide I region of PLA2 bound to supported membranes composed of POPC plus 0, 5, 20, and 50 mol % POPG, as indicated, in a buffer containing 5 mM Hepes (pH 8.2), 100 mM NaCl, 15 mM KCl, 2 mM CaCl2. Note the increase in the intensity of the component at ~1658 cm1 at higher fractions of the acidic lipid, POPG, in the membrane.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strength and cooperativity of PLA₂-membrane interactions

Very high binding affinities have been reported for association of secretory PLA₂s with anionic phosphatidylglycerol surfaces, i.e., K  10⁹ M¹ and K  5 × 10⁷ M¹ for human group IIA PLA₂ and AppD49, respectively (Han et al., 1997; Snitko et al., 1997). Electrostatic effects at least partly determine the high affinities of these PLA₂s for negatively charged membranes. Consistent with this, much lower binding constants (<10³ M¹) have been measured for the binding of both enzymes to zwitterionic phosphatidylcholine vesicles (Han et al., 1997; Bayburt et al., 1993). The data presented in Figs. 2 and 3 and in Table 1 demonstrate that not only the apparent binding constant of AppD49 for anionic membranes but also the density of binding sites and binding cooperativity decrease when surface electrostatics is suppressed by high ionic strengths. Higher apparent binding constants at low ionic strengths are evidently due to electrostatic attraction between the cationic PLA₂ and negatively charged membranes. Binding of Na⁺ ions to the acidic lipids in the membrane, which are probably involved in the creation of binding sites, may account for the decrease in the binding site density at high NaCl concentrations (Tatulian, 1993, 1999; Tatulian and Biltonen, 1997). Increased binding cooperativity at low ionic strengths can be explained by hypothesizing that the enzyme forms dimers at the membrane surface; i.e., each membrane-bound enzyme induces the binding of another one for dimer formation. Dimerization of PLA₂ at the membrane surface may be facilitated by decreased electrostatic repulsion between the cationic enzyme molecules because of the negative surface potential of the membrane, an effect that would be more efficient at lower ionic strengths. Interestingly, the dimeric and monomeric isoforms of AppD49 are structurally similar to each other (Scott et al., 1994), but the dimeric isoform is acidic (excess charge at neutral pH is 1) whereas the monomeric form is strongly basic (excess charge is +6). This agrees with the hypothesis that electrostatic effects exerted by negatively charged membranes may facilitate dimerization of the monomeric enzyme at the membrane surface (see also Welches et al., 1993).

Role of phospholipid hydrolysis products in PLA₂ activation

The role of phospholipid hydrolysis products, the free fatty acid and the lysophospholipid, in PLA₂ activation is important for understanding 1) the mechanism of PLA₂ activation at the membrane surface in general and 2) the factors that make cell membranes susceptible to the action of PLA₂. It has been shown that phospholipid vesicles maintained their structural integrity upon complete hydrolysis of the lipid in their outer leaflet by PLA₂ (Jain et al., 1986; Berg et al., 1991; Bayburt et al., 1993), indirectly implying that most, if not all, reaction product stays in the membrane following phospholipid hydrolysis. The lipid degradation products in the membrane were further shown to promote PLA₂ activation by modifying the membrane structure and strengthening PLA₂-membrane interactions (Jain et al., 1982, 1986; Jain and de Haas, 1983; Apitz-Castro et al., 1982; Bayburt et al., 1993; Burack and Biltonen, 1994; Burack et al., 1997). For example, the binding affinity of human group IIA PLA₂ for phosphatidylcholine vesicles increased by three orders of magnitude in the presence of 18% reaction products in the membrane (Bayburt et al., 1993).

On the other hand, removal of a significant fraction of lipid hydrolysis products has been demonstrated by ellipsometry for bilayers supported on silicon discs (Speijer et al., 1996), by external reflection FTIR spectroscopy for monolayers at the air/water interface (Gericke and Hühnerfuss, 1994), and by cryo-transmission electron microscopy for unilamellar vesicles (Callisen and Talmon, 1998). The present results indicate that phospholipid degradation by PLA₂ is followed by dissociation from the membrane of a fraction of both the free fatty acid and the lysophospholipid and that the lysophospholipid is removed from the membrane to a significantly larger extent than the fatty acid (Figs. 8 and 10). This leads to the accumulation of the free fatty acid in the membrane, which would modulate membrane binding and activity of PLA₂ through electrostatic and/or morphological effects. In fact, an increase in the negative surface potential of both lipid vesicles and planar membranes has been observed in the presence of PLA₂ under catalytic conditions (Cherny et al., 1990, 1992). Addition of fatty acid, but not lysophosphatidylcholine, to lipid vesicles increased their negative zeta potential, suggesting that the PLA₂-induced negative surface potential may result from the accumulation of the fatty acid in the vesicle membranes (Cherny et al., 1992). It should be noted that although most of the liberated fatty acid stays in the membrane, a fraction of it partitions into the aqueous phase. Probably only an optimal amount of the fatty acid in the membrane is required for efficient lipolysis by PLA₂. Formation of 2:1 fatty acid/phosphatidylcholine complexes has been observed by several studies (Cevc et al., 1988, and references therein). At moderate fractions of PLA₂-generated fatty acid, complexes between intact phospholipid and fatty acid may form and serve as PLA₂ binding sites that are characterized by local negative curvature and increased anionic charge, although very high fractions of the fatty acid may inhibit PLA₂ activity by laterally segregating into negatively charged patches and electrostatically sequestering PLA₂ from its substrate.

The other PLA₂ reaction product, the lysophospholipid, is also able to activate PLA₂ (Jain and de Haas, 1983; Bell et al., 1996; Henshaw et al., 1998). It should be noted that the present data indicate predominant, but not complete, removal of the lysophospholipid from the membrane following lipid hydrolysis. Disproportional removal from the membrane of the fatty acid and the lysophospholipid is likely to perturb the membrane structure and stimulate PLA₂ activation to a greater extent than in the case of proportional removal or preservation of both products. This is consistent with the observation that exogenous lysophosphatidylcholine reduces the ability of fatty acid to enhance interfacial activation of PLA₂ (Henshaw et al., 1998), probably by repairing the fatty-acid-induced structural irregularities in phospholipid membranes. These conclusions, which are drawn from the studies on model membranes, are consistent with the results obtained on cell cultures suggesting that proinflammatory cytokines render the membranes of the affected cells susceptible to the action of PLA₂ by modifying the structure of cell membranes (Murakami et al., 1998).

Role of acidic lipid in PLA₂ activity

Higher PLA₂ activity toward membranes with increased anionic surface charge has been observed by several earlier studies. Thus, porcine pancreatic PLA₂ had a two- to threefold preference for anionic lipids (Ghomashchi et al., 1991). AppD49, which was used in this study, exhibited a three- to fivefold preference for catalysis of anionic versus zwitterionic lipids (Han et al., 1997). When human group IIA PLA₂ was applied to polymerized phosphatidylglycerol vesicles containing 1 mol % pyrene-phosphatidylglycerol or pyrene-phosphatidylethanolamine, the anionic lipid was hydrolyzed 10 times faster (Snitko et al., 1997). On the other hand, no significant discrimination by this enzyme between acidic and zwitterionic lipids was detected by the double-radiolabel technique (Bayburt et al., 1993). Consistent with these latter results, the present data indicate that although the activity of AppD49 increased with increasing mole fraction of the acidic lipid in the membrane (Fig. 11), the enzyme did not demonstrate intrinsic preference for the anionic lipid (Fig. 13).

Correlation between membrane surface charge and membrane-induced structural changes in PLA₂

As described in the Introduction, studies on the interfacial activation of PLA₂ have been focused either on the role of the membrane surface properties or, in fewer cases, on the structural changes in PLA₂ caused by membrane binding, leading to the conceptually different substrate and enzyme hypotheses. The results of this work identify a correlation between the membrane surface electrostatics, the strength and cooperativity of membrane binding of PLA₂, membrane-induced structural changes in PLA₂, and PLA₂ activity. The data suggest that conformational changes do occur in PLA₂ during its interactions with membranes and that the membrane surface properties and structural changes in the enzyme contribute synergistically to PLA₂ activation. This synergistic mechanism of the interfacial activation of PLA₂ implies that the factors controlling membrane binding of PLA₂ determine structural changes in the enzyme that result in the activation of the enzyme. It should be emphasized that, as described in Tatulian et al. (1997), the structural changes upon membrane binding of PLA₂ occur under both catalytic and noncatalytic conditions. Therefore, these structural changes are likely to take place during the membrane-binding step of the complex process of interfacial activation of PLA₂, independent of the substrate binding to the active center. However, they are a prerequisite for the activation of PLA₂ at the membrane surface, provided there is calcium in the aqueous phase and a hydrolyzable lipid in the membrane. This is consistent with the notion that although the membrane binding of PLA₂ and the catalytic turnover are temporally dissociated and involve different residues, there is a close structural and functional coupling between them.

CONCLUDING REMARKS

Secretory PLA₂s are perhaps the most extensively studied enzymes that catalyze reactions at the lipid/water interfaces (Jain and Berg, 1989; Scott and Sigler, 1994; Mukherjee et al., 1994; Arni and Ward, 1996; Gelb et al., 1995, 1999; Dennis, 1997, 2000). However, certain aspects of interfacial activation of these enzymes, including the structural changes in the enzyme upon membrane binding and their correlation with the membrane physical properties, are still not well understood. The present study demonstrates that ATR-FTIR spectroscopy is uniquely well suited for investigating a wide range of problems pertaining to the activation of PLA₂ at the membrane surface. The data indicate a reciprocal relationship between the membrane surface properties, membrane binding strength of PLA₂, structural changes in the enzyme, and PLA₂ activity. This finding unifies the substrate and enzyme hypotheses of interfacial activation of PLA₂ and implies that both the membrane and enzyme factors are complementary and synergistic determinants of the activation of membrane-bound PLA₂. The surface properties of the membrane are indeed important for PLA₂ activation. But they are only a prerequisite for binding of PLA₂ to the membrane surface in a proper way, probably including the strength of binding, the depth of membrane insertion, and the orientation, which is required for the induction of the conformational changes in PLA₂ that ultimately activate the enzyme.