Molecular Basis of Phospholipase A₂ Activity toward Phospholipids with sn-1 Substitutions

Abstract

We studied secretory phospholipase A₂ type IIA (sPLA₂) activity toward phospholipids that are derivatized in the sn-1 position of the glycerol backbone. We explored what type of side group (small versus bulky groups, hydrophobic versus polar groups) can be introduced at the sn-1 position of the glycerol backbone of glycerophospholipids and at the same time be hydrolyzed by sPLA₂. The biophysical characterization revealed that the modified phospholipids can form multilamellar vesicles, and several of the synthesized sn-1 functionalized phospholipids were hydrolyzed by sPLA₂. Molecular dynamics simulations provided detailed insight on an atomic level that can explain the observed sPLA₂ activity toward the different phospholipid analogs. The simulations revealed that, depending on the nature of the side chain located at the sn-1 position, the group may interfere with an incoming water molecule that acts as the nucleophile in the enzymatic reaction. The simulation results are in agreement with the experimentally observed sPLA₂ activity toward the different phospholipid analogs.

INTRODUCTION

Phospholipase A₂ (PLA₂; EC 3.1.1.4) comprises a diverse superfamily of lipolytic enzymes that specifically cleave the sn-2 acyl ester bond of glycerophospholipids to produce free fatty acids and lysophospholipids (1–6). Over the past two decades, numerous PLA₂s have been identified and characterized (7–9). According to their biochemical features, such as cellular localization, requirement of Ca²⁺, substrate specificity, and primary structure, these PLA₂s are classified into several families, including low-molecular-weight secretory PLA₂ (sPLA₂), Ca²⁺-sensitive arachidonoyl-specific 85-kDa cytosolic PLA₂ (cPLA₂), and Ca²⁺-independent PLA₂ (iPLA₂). Secretory phospholipases are small enzymes (14–19 kDa) that are classified into at least four subgroups based on structural differences (10,11). They have several features that distinguish them from other PLA₂ families, such as high-disulfide-bond content, a requirement for millimolar concentration of Ca²⁺ for catalysis, and a broad specificity for phospholipids with different polar headgroups and fatty acid chains (12). The activity of these lipolytic enzymes increases substantially when adsorbed onto a lipid-water interface (13–17)—a phenomenon known as interfacial activation (18). However, there are dramatically different affinities of the various sPLA₂s for membrane surfaces containing different phospholipids (19,20). It has been suggested that this affinity is controlled by the difference in the amino acids located on the interfacial binding surface (interfacial specificity) rather than the substrate affinity being controlled by the active site residues (catalytic site specificity) (21–23). The interfacial binding specificity of sPLA₂s has important physiological consequences. sPLA₂ generally prefers membranes containing anionic lipids (20,24). Human sPLA₂-IIA binds several orders of magnitude more tightly to anionic phospholipid membranes than to a membrane composed of zwitterionic (charge-neutral) phosphatidylcholine lipids (19,20,25). The specificity of the enzyme toward anionic membranes is driven by electrostatic forces (24,26,27), since the enzyme has a relatively large number of cationic residues, resulting in an isoelectric point for the enzyme that is larger than 10.5 (26,28). This electrostatic field provides nonspecific electrostatic interactions between the enzyme and membrane surface that consequently enhance binding to the anionic surface (1,16,18,27,29). Conversely, other sPLA₂ types, such as those isolated from snake venom, lack the large number of cationic residues found in sPLA₂-IIA (1,20). These species are able to bind and hydrolyze both zwitterionic and anionic lipid membranes, and do not exhibit the surface charge specificity of sPLA₂-IIA (20,30). The ability of snake venom sPLA₂ to also hydrolyze zwitterionic membranes has been connected to the presence of aromatic residues in the membrane putative interfacial binding surface (31). These aromatic residues are essential for the nonpolar interaction with the zwitterionic membrane (24,32), and they may interact with membranes in distinct modes that determine membrane-binding affinities and hence catalytic efficiency (32,33).

Members of the group II sPLA₂s have drawn increased attention, since these enzymes have been associated with a wide variety of immune-mediated inflammatory pathologies in humans, ranging from systemic and acute inflammatory conditions to cancer (9,28,34–39). For instance, high levels of sPLA₂ are associated with the onset of rheumatoid arthritis (40,41), allergic rhitinis (42,43), and septic shock (44,45). More recently, it was suggested that sPLA₂-IIA has a central role in both tumor development and progression (46–49), since sPLA₂-IIA occurs at elevated levels in cancer tissue. This observation provides an avenue for site-specific drug delivery by liposomal carriers, where the liposomes are degraded specifically in the tumor tissue by sPLA₂ releasing antitumor drugs at the target site (48–50). Additionally, the fatty acid and lysophospholipid hydrolysis products generated by sPLA₂ have been shown to display a synergistic effect as permeability enhancers for drug transport (2,48).

To further investigate the viability of using sPLA₂-IIA as a trigger of targeted drug delivery systems, we are interested in new phospholipids with functionality in the sn-1 position. These lipid molecules should be hydrolyzed effectively by sPLA₂-IIA and potentially serve as prodrugs. We demonstrated earlier that a relatively small hydrophobic side chain located at the sn-1 position is efficiently hydrolyzed by sPLA₂-IIA (51).

The aim of this study was to explore further which type of side group (small versus bulky groups, hydrophobic versus polar) can be introduced at the sn-1 position of the glycerol backbone of glycerophospholipids. The insight gained from this study is essential for determining the structural criteria that allow sPLA₂ to accommodate and tolerate phospholipids with sn-1 substitutions. Furthermore, this knowledge provides the basis for using the sn-1 position as a possible anchor point for drug conjugates or fluorescence probes. The latter can be used to probe interfacial properties of membranes. We have chosen an interdisciplinary approach that combines organic synthesis, biophysical characterization, and computer simulations to obtain a detailed understanding of structural properties in the substrate that determine sPLA₂ activity. Here, we report the results of our biophysical and computational studies. The biophysical characterization was performed to determine whether the modified lipids can form multilamellar vesicles (MLVs) and be hydrolyzed by sPLA₂. The experimentally determined sPLA₂ activities toward the different phospholipids vary significantly and strongly depend on the type of substitution at the sn-1 position. These results were further supported by computer simulations that provide further insight into substrate recognition on a molecular level. Our results generally indicate that, independent of the polarity of the group substituted at the sn-1 position, limited space is available in the binding cleft of sPLA₂. We could identify two essential contributions that may explain (in part) the difference in the experimentally observed sPLA₂ activities against the synthesized substrates. First, specific lipid-protein interactions impair the catalytic reaction; and second, steric hindrance interferes with an incoming water molecule that acts as a nucleophile in the enzymatic reaction.

METHODS

Preparation of vesicles

Phospholipids were hydrated in an aqueous buffer (150 mM KCl, 10 mM HEPES, 30 μM CaCl₂, 10 μM EDTA, pH = 7.5). MLVs formed spontaneously. To ensure complete hydration, the lipids were hydrated for 1 h at 65°C. During the hydration, the lipids were vortexed every 15 min. When small unilamellar vesicles (SUVs) were needed, the MLVs were sonicated for 1 h at 5°C above the lipid main-phase transition temperature. The total concentration of lipids in the buffer solutions was 2 mM.

Differential scanning calorimetry (DSC) and dynamic light scattering (DLS)

Differential scanning calorimetry (DSC) of 2 mM MLVs was performed by using a Microcal MC-2 (Northhampton, MA) ultrasensitive power compensating scanning calorimeter equipped with a nanovoltmeter. The scans were performed in the upscan mode at a scan rate of 10°C/h. An appropriate baseline was subtracted from the calorimeter curves to afford the melting enthalpies of the MLVs. Dynamic light scattering of the SUVs was performed on a Zetasizer Nano particle analyzer (ZS ZEN3600, Malvern Instruments, Westborough, MA).

Activity measurements

The conditions used to perform the sPLA₂ activity measurements were as follows: 0.15 mM phospholipid as SUVs, 150 nM sPLA₂, 0.15 M KCl, 30 μM CaCl₂, 10 μM EDTA, 10 μM HEPES (pH 7.5). The catalytic reaction was initiated by addition of 8.9 μL of a 42-μM Agkistrodon piscivorus piscivorus snake venom sPLA₂ stock solution to a 2.5-mL SUV suspension thermostated at the main phase-transition temperature of the vesicles before the addition of the enzyme. The time-dependent activity of sPLA₂ was monitored from the changes in the 90° static light scattering giving information of changes in the lipid morphology as non-bilayer-forming lysophopholipids and fatty acids are generated. High-performance liquid chromatography (HPLC) quantification of the products from the enzymatic reaction was performed with a 5-μm diol column. Two different mobile phases (hereafter referred to as mobile phase A and mobile phase B) were used. Mobile phase A was a chloroform/methanol/25% ammonium hydroxide (800:195:5) solution, whereas mobile phase B was a chloroform/methanol/water/25% ammonium hydroxide (600:340:50:5) solution. The following gradients were used at a flow rate of 1 mL/min: 0–14 min, 100% A to 100% B using a linear gradient; 14–25 min, 100% B; 25–30 min, 100% B to 100% A using a linear gradient; 30–45 min, 100% A to regenerate. An evaporative light scattering detector was used for detection. Samples were collected at different time intervals by collecting 100 μL of the reaction mixture and rapidly mixing with a 0.5-mL chloroform/methanol/acetic acid (2:4:1) solution to quench the reaction. The solution was washed with 0.5 mL of water, and 50 μL of the organic phase was used for HPLC. Fluorescence measurements were performed using a DMX-1100 spectrofluorometer (SLM, Urbana, IL). Purified snake venom PLA₂ was a generous gift from Dr. R. L. Biltonen (University of Virginia, Charlottesville, VA).

Molecular dynamics simulations

The crystal structures of European honeybee (Apis Mellifera) venoms sPLA₂ complexed with the transition-state analog, 1-O-octyl-2-heptylphosphonyl-sn-glycero-3-phosphoethanolamine (diC₈(2Ph)PE), resolved to 2.0 Å (15,52) and human PLA₂-IIA complexed with 6-phenyl-4(R)-(7-phenyl-heptanoylamino)-hexanoic acid and resolved to 2.1 Å (53) were obtained from the Protein Data Bank (entry codes 1poc and 1kqu, respectively) (54).

The initial modeling step involved placing diC₈(2Ph)PE into the binding cleft of hPLA₂-IIA, which was done by 1), deleting the inhibitor in 1kqu (keeping the two calcium ions and water molecules in the structure); 2), deleting calcium ions and water molecules in 1poc; 3), aligning 1poc with 1kqu; and, finally, 4) deleting the European honeybee venom PLA₂ structure. The structures for the different phospholipids (Fig. 1) were built from diC₈(2Ph)PE using SPARTAN Version 1.0.2 (Wavefunction, Irvine, CA). The modified lipids were introduced in the protein structures by aligning the modified phospholipids with diC₈(2Ph)PE in the enzyme-diC₈(2Ph)PE structures and subsequently deleting diC₈(2Ph)PE. Missing distance, angle, and torsion parameters for the sn-1 side chains were obtained from the CHARMM27 parameter set (55) describing similar atom types. The structures were solvated using the program SOLVATE (56). Eighteen water molecules were randomly replaced with chloride ions to neutralize the systems. The final systems contained ∼4900 water molecules, and the simulation cell dimensions were 52.7 × 51.7 × 67.3 ų. For the molecular dynamics (MD) simulations, the software NAMD (57) was used with the Charmm27 all-hydrogens parameter set and with the TIP3 water model (55). Each complex was simulated at least three times starting from different initial conditions, which were obtained by different steps of energy minimization. The first energy minimization of the systems involved 1000 steps. For each subsequent simulation of a certain complex, the number of steps for energy minimization was increased by 250; that is, 1000, 1250, and 1500 steps were used for simulations of a complex that was repeated three times. The minimization procedure was followed by 100 ps of heating of the systems to 300 K. Each simulation was carried out for 10 ns in the NPT ensemble. A time step of 1 fs was used throughout all simulations. An isotropic constant ambient pressure of 1 atm was imposed using the Langevin piston method (58) with a damping coefficient of 5 ps⁻¹, a piston period of 200 fs, and a decay of 500 fs. The particle mesh Ewald method was used for computation of the electrostatic forces (59,60). The grid spacing applied was ∼1.0 Å, and a fourth-order spline was used for the interpolation. The long-range part of the electrostatic forces was evaluated every fourth femtosecond. The van der Waals interactions were cut off at 12 Å using a switching function starting at 10 Å. Periodic boundary conditions were applied in x, y, and z directions. Examinations of the molecular structures and analyses of the trajectories were carried out using tcl-scripting within the graphical program Visualization Molecular Dynamics (61).

FIGURE 1.

Chemical structure of compounds A–H.

RESULTS AND DISCUSSION

The ability of secretory phospholipase A₂ to hydrolyze a broad range of phospholipid analogs has been explored in developing new drug delivery systems (48,50). We previously demonstrated that sPLA₂ can hydrolyze phospholipids with their headgroup in the sn-2 position and phospholipids with small substitutions in the sn-1 position (51,62). It is intriguing that sPLA₂ is able to hydrolyze phospholipids with substitutions in the sn-1 position, since there is only limited space available in the binding cleft of sPLA₂. To understand the structural restrictions for substitutions in the sn-1 position at a molecular level, we present results that are based on biophysical characterization of phospholipids and computer simulation of sPLA₂-phospholipid analog complexes. The biophysical characterizations were performed to ensure that the phospholipid analogs formed vesicles and were hydrolyzed by sPLA₂, whereas computer simulations were carried out to provide structural insight on an atomic level. We studied phospholipid analogs with sn-1 substitutes having different properties, i.e., the side chains have different size (bulkiness) and polarity. The chemical structures of the synthesized compounds are shown in Fig. 1, and the phospholipids contain hydroxy (compound A) methoxy (compound B), benzyloxy (compound C), allyl (compounds D and E), and ester (compounds F–H) groups. All compounds except compound B were synthesized, and all compounds were investigated in the computational study. The results obtained from simulations of sPLA₂ in complex with its natural substrate (63) (hereafter referred to as “native”) are also included in the following discussion as a reference to the phospholipid analogs.

Differential scanning calorimetry

DSC scans were performed to verify that the phospholipid analogs formed MLVs and to ensure that the vesicles used in the activity measurement assay were at the main phase transition. All tested compounds showed a single peak in the thermograph with a main phase transition above 35°C. However, T_1/2, which is a measure of the cooperativity of the phase transition, varies between the compounds (Table 1). For compounds C and E, T_1/2 < 1°C, whereas for compounds A, D, and F–H, T_1/2 > 1°C, with relatively broad tails indicating that this series of compounds may not form structural homogeneous MLVs. In particular, compounds A (T_1/2 ≈ 3.7°C) and F (T_1/2 ≈ 3.0°C) have relatively broad main phase-transition peaks, where the phase-transition enthalpy (ΔH) for compound F is approximately half the value observed for compounds C and E. These compounds have comparable ΔH (9.3–11.6 kJ/mole) and the same T_m (43.9°C) (Table 1). Larger variations are observed for the remaining phospholipid analogs. DSC of the SUVs showed the usually observed broadening of the main phase transition when MLVs are transformed into SUVs (64) (Fig. 2 D). Dynamic light scattering measurements of the different SUVs showed that the SUVs had an average diameter of 40–70 nm.

TABLE 1.

Differential calorimetry data of MLVs in the main phase transition for compounds A and C–H

Compound	ΔH (kJ/mole)	Tm (°C)	T½ (°C)
A	7.3	57.0	3.65
C	11.6	43.9	0.43
D	10.3	46.9	1.82
E	9.3	43.9	0.48
F	4.6	48.2	2.99
G	8.6	39.0	1.52
H	7.6	48.9	1.84

ΔH, melting enthalpy; T_½, half-width; T_m, peak position.

FIGURE 2.

The typical appearance of a fluorescence measurement on (A) SUVs composed of phospholipids that are fairly good substrates for sPLA₂, and (B) SUVs composed of phospholipids that are poor substrates for sPLA₂. The upper lines show the intrinsic fluorescence from the enzyme, whereas the lower lines show the morphology changes of the systems. (C) Typical HPLC chromatogram illustrating the effect of sPLA₂-catalyzed hydrolysis of a phospholipid from SUVs that is a fairly good substrate. The chromatograms show the amount of phospholipid before the addition of sPLA₂ (solid line) and 1000 s (dashed line) and 6000 s (dotted line) after the addition of sPLA₂. (D) Typical DSC scans for MLVs and SUVs. Scans are representative of all phospholipids studied.

Activity measurements

We used 90° light scattering to qualitatively determine sPLA₂ activity toward the different compounds. This approach is only an indirect indication for sPLA₂ activity, since it detects the morphology changes of the liposomes. These changes cannot be induced by the adsorption of the enzyme to the liposome surface, since the concentration of sPLA₂ is significantly lower than the concentration of the liposomes. Hence, observed fluorescence changes are induced by hydrolysis. In this assay, SUVs consisting of compounds C, E, and G showed no effect upon sPLA₂ addition, whereas morphology changes were detected for SUVs composed of compounds A, D, F, and H (Table 2). Fig. 2 A shows a typical appearance of a fluorescence measurement that is representative for SUVs composed of compounds A, D, F, or H (sPLA₂ was added at 500 s to secure full equilibration of the systems). The lower curve monitoring the morphology changes of the system shows a period without any changes in the system followed by a sudden increase (“burst”) in fluorescence, indicating the breakdown of the SUVs due to sPLA₂ hydrolysis.

TABLE 2.

Overview of the results obtained from fluorescence measurements and HPLC studies for compounds A and C–H

Compound	Morphology changes observed by fluorescence	Activity observed by HPLC
A	+	+
C	−	−
D	+	+
E	−	−
F	+	+
G	−	−
H	+	+

The intrinsic fluorescence from tryptophan in the enzyme shows a sudden sharp increase after addition of sPLA₂ (Fig. 2 A, upper curve), which might be a result of sPLA₂ immediately attaching to the surface of the negatively charged vesicles. During the experiment, the intrinsic fluorescence decreases, which is expected, since sPLA₂ hydrolysis leads to the breakdown of SUVs, and consequently results in small aggregates to which sPLA₂ binds less efficiently. Fig. 2 B shows a typical appearance of the fluorescence measurements representative for SUVs composed of compounds C, E, or G. No morphology changes of the systems (lower curve) are observed, indicating that these compounds are poor substrates for sPLA₂. Furthermore, the intrinsic fluorescence from tryptophan of the enzyme (upper curve) shows no significant change after the addition of the enzyme. Again, the initial sharp increase might be a sign of instantaneous sPLA₂ attachment to the surface of the SUVs. The fluorescence measurements were followed by HPLC studies to demonstrate the link between morphology changes and sPLA₂ activity. Fig. 2 C shows an example of a typical HPLC measurement on systems in which fluorescence measurements indicate morphology changes. A substantial decrease in the amount of phospholipid in the sample before addition of sPLA₂ (solid line) and 1000 s (dashed line) and 6000 s (dotted line) after addition of the enzyme is observed. Clearly, HPLC results confirm that the indirectly observed morphology changes are induced by sPLA₂ hydrolysis of the phospholipids in the SUVs. The results obtained from the fluorescence measurements and the HPLC studies on compounds A and C–H are summarized in Table 2.

MD simulations

The biophysical characterization revealed that although the tested phospholipids can form MLVs, only some of the phospholipid analogs are relatively good substrates for sPLA₂. To understand, on a structural level, the observed different sPLA₂ activity toward the phospholipids, we applied computational methods. Initially, we applied simple docking of the phospholipid analogs to sPLA₂. These results were not conclusive, and they indicated that all phospholipids fit into the binding cleft of sPLA₂. We therefore applied MD simulations (65,66) to gain further insight into the mechanism(s) that could explain the experimentally observed activities. Enzymatic reaction depends on several factors, including binding of the enzyme to the membrane surface, membrane properties, formation of the Michaelis-Menten complex, etc. (18,66,67). Here, we only consider one step in the process, and have confined ourselves to a study, using classical mechanics, of the dynamics of substrate binding before the formation of the transition state. Our approach was motivated by the work of, e.g., Cho and co-workers (68) and Menger (69), who have challenged the classical hypothesis (70) that enzyme catalysis can only be improved by further stabilizing the transition state. The authors demonstrated that it may be at least as advantageous to consider ways of engineering the substrate, which could improve substrate binding, and/or product clearance (removal of product inhibition effects) (68–74).

It is well established that membrane properties affect enzyme activity (19), and that binding of some sPLA₂ species to the membrane surface is governed by the interfacial charge distribution of the membrane (19,20,63). For instance, snake venom sPLA₂ can hydrolyze both membranes consisting of zwitterionic phosphatidylcholine and those consisting of negatively charged phosphatidylglycerol (PG) phospholipids, whereas human sPLA₂-IIA shows only activity toward membranes containing PG (1,16,19,24,27,29). This has been correlated to the binding surface of the enzyme. sPLA₂-IIA has a relatively high positively charged binding surface (i-face) that makes critical contacts to negatively charged phospholipids (23,75). However, once bound to the surface, sPLA₂-IIA can hydrolyze phosphocholine and PG phospholipids (20), indicating a more generic catalytic mechanism, which is not strongly dependent on the headgroup. Hence, it is expected that the results obtained in this study are relevant for snake venom sPLA₂ as well as for human sPLA₂-IIA. The series of compounds used in this study is only modified in the sn-1 position of PG, which means that measuring no activity cannot be due to the lack of binding of sPLA₂ to the membrane surface, but has to be related to the catalytic reaction.

The catalytic mechanism of sPLA₂ is schematically shown in Fig. 3, where the numbering corresponds to the numbering used for sPLA₂-IIA. The catalytic site consists essentially of four components: a histidine (His⁴⁷), an aspartate (Asp⁹¹), a calcium ion (cofactor), and a water molecule that acts as the nucleophile (15,76). The catalytic reaction is initiated by forming a Michaelis-Menten complex between the substrate and sPLA₂. The complex is stabilized by Ca²⁺ that coordinates to the substrate via the carbonyl oxygen (O22) and phosphonate oxygen (O4), as well as to the protein via the backbone oxygen of Gly²⁹ (O) and the two carboxylate oxygens of Asp⁴⁸ (OD1 and OD2) (Fig. 4). Additionally, Lys⁶², which is highly conserved in the group II sPLA₂₅, interacts with the sn-3 phosphate oxygen (1,77), thereby contributing to the stabilization of the complex (1,77). The reactive carbonyl carbon atom of the ester bond (Fig. 3, C21) is then attacked by a nucleophilic water molecule, leading to the formation of a tetrahedral intermediate (close to the transition state), which is stabilized by the backbone hydrogen of the NH group of Gly²⁹ (oxyanion hole) and the calcium ion (15,76,78). During the formation of the intermediate phase, the hydrogen atom of the water molecule is transferred to His⁴⁷, and thereby, the histidine imidazole ring becomes positively charged. The protonated His⁴⁷ is stabilized by Asp⁹¹, which hydrogen-bonds to Tyr⁵¹ and Tyr⁶⁶ (Fig. 3). The proton abstracted from His⁴⁷ is ideally positioned to protonate the sn-2 oxygen in concert with the productive collapse of the tetrahedral intermediate (3,15,79). Molecular dynamics simulation is a classical approach that cannot account for transition state or processes involving bond breaking (66,71). We therefore focused on the Michaelis-Menten complex of the different compounds. One of the requirements for catalysis is that a stable complex can be formed and that the key distances (involving the Michaelis-Menten complex) are maintained. We have used these criteria to study the stability of the different complexes and to investigate whether that stability can be correlated to the observed sPLA₂ activity toward the different compounds.

FIGURE 3.

Schematic representation of the catalytic mechanism of sPLA₂ and key interactions that stabilize the Michaelis-Menten complex. Atom types given in parentheses refer to the Protein Data Bank nomenclature.

FIGURE 4.

RMSDs of Cα atoms with respect to the initial sPLA₂ structure as a function of simulation time. (Inset) Comparison of the average residue fluctuations based on Cα atoms and averaged over the last 5 ns of the simulation (—) with the RMSD derived from the crystallograhic B_factor: (·–·) (53,89). The data were extracted from a simulation of sPLA₂ in complex with compound D and are representative for all simulations.

MD—protein stability

Simulations were performed at least three times for each complex, to verify the significance of the results (Table 3). For each simulation, the root mean-square deviation (RMSD) of Cα atoms with respect to the protein structure after minimization was calculated as a function of time to check the stability of the simulation. In Fig. 4, the time evolution of RMSD extracted from an sPLA₂-compound D simulation is shown as an example. RMSD approaches a constant value of ∼1.7 Å within 3 ns. Similar results were obtained for the other simulations, and the average RMSDs calculated from the different simulations are summarized in Table 3. Variations are within the statistical uncertainties and the overall RMSD averaged over all RMSDs is 1.5 ± 0.1 Å. We also compared the fluctuations of Cα atoms with the crystallographically determined B-factors (53) (Fig. 4, inset). The fluctuations extracted from simulations resemble the B-factor, but have larger magnitudes in loop regions that are exposed to solvent. These deviations are not surprising, since these regions may be affected by crystal packing (71,80).

TABLE 3.

Summary of the simulations

Compound	Number of simulations	Length of simulations (ns)	Root mean-square deviation mean ± SD (Å)
A	3	each 10	1.5 ± 0.1
B	3	each 10	1.4 ± 0.1
C	3	each 10	1.4 ± 0.1
D	3	each 10	1.7 ± 0.1
E	4	each 10	1.3 ± 0.1
F	3	each 10	1.2 ± 0.1
G	4	each 10	1.5 ± 0.1
H	3	each 10	1.4 ± 0.1
native	6	each 10	1.3 ± 0.1

The last column lists the average RMSD of Cα atoms with respect to the protein structure after minimization. Means and their standard deviations (SD) are based on a series of simulations of a particular complex: “Native” refers to the natural substrate of sPLA₂ ((R)-1,2-dipalmitoyl-glycero-3-phosphocholine) (63).

MD—key distances

For the catalytic reaction to occur, a stable Michaelis-Menten complex has to form, which implies that distances stabilizing the complex are maintained (65,81). We therefore monitored distances involving the cofactor calcium (Ca²⁺ to Asp⁴⁸(OD1), Asp⁴⁸(OD2), Gly²⁹(O), S(O22), S(O4) (Fig. 5, where atom types and S refer to the Protein Data Bank nomenclature and substrate, respectively) or residues in the vicinity of the substrate (Asp⁹¹(OD1)-Tyr⁵¹(OH), Asp⁹¹(OD2)-Tyr⁶⁶(OH), His⁴⁷(ND1)-S(C21), His⁴⁷(HE2)-Asp⁹¹(OD1), and Gly²⁹(O)-S(O22) (Fig. 6)). The data were extracted from an sPLA₂-compound D simulation, and similar results were observed in all other simulations. To compare those, we have calculated mean distances for each complex averaged over time (the last 5 ns of the simulations) and over simulations of that complex. The average distances are listed in Table 4. For reference, the results for the sPLA₂-native simulations are also included in the last column of Table 4. Distances extracted from simulations of different sPLA₂-phospholipid complexes are within the statistical uncertainties and compare well to distances determined from the sPLA₂-native simulations. Exceptions are particularly observed for distances Asp⁹¹–Tyr⁵¹ and Asp⁹¹–Tyr⁶⁶, which are ∼1.2 Å shorter than those monitored in the sPLA₂-native complex, indicating structural effects caused by the sn-1 substitute. The two tyrosines are not required to maintain activity, since mutants in which these two residues were mutated to phenylalanine had virtually unaltered catalytic performance (78). For the single mutant Asp→Asn, significant catalytic activity was retained. Both observations suggest that the hydrogen-bonding network plays only a minor role for the catalytic turnover. Variations in distance are also observed for His⁴⁷(ND1)-S(C21) and S(O22)-Gly²⁹(HN) in the different complexes. However, as shown in Fig. 7, these variations are within the statistical uncertainty when compared to the native phospholipid, and hence, we cannot attribute the lack of sPLA₂ activity toward some of the phospholipid analogs to these variations.

FIGURE 5.

Time evolution of selected distances involving the cofactor calcium. Distances are given in the individual panels (A–D). S, substrate; atom types correspond to Protein Data Bank nomenclature (see also Fig. 4). The distances were extracted from a simulation of sPLA₂ in complex with compound D and are representative for all simulations.

FIGURE 6.

Time evolution of selected distances between atoms that are close to the substrate. Distances are given in the individual panels. See Fig. 5 legend for more details.

TABLE 4.

Summary of the calculated distances between selected atoms that are involved in stabilization of the substrate or part of the catalytic device

Compound	A	B	C	D	E	F	G	H	Native
Distances	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)	Mean ± SD (Å)
D91(OD1)–Y51(OH)	1.7 ± 0.2	1.7 ± 0.1	1.8 ± 0.2	1.8 ± 0.2	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.1	3.1 ± 0.3
D91(OD2)–Y66(OH)	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.2	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.1	1.7 ± 0.1	2.7 ± 0.1
H47(HE2)–D91(OD1)	1.9 ± 0.2	1.9 ± 0.1	1.8 ± 0.2	1.8 ± 0.1	1.8 ± 0.2	1.9 ± 0.2	1.9 ± 0.2	1.9 ± 0.3	1.9 ± 0.3
H47(ND1)–S(C21)	5.2 ± 0.4	4.4 ± 0.3	4.5 ± 0.4	4.2 ± 0.3	4.8 ± 0.2	4.1 ± 0.4	4.8 ± 0.2	4.8 ± 0.4	4.5 ± 0.4
Ca2+–D48(OD1)	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.1	2.3 ± 0.3	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.2
Ca2+–D48(OD2)	2.4 ± 0.3	2.3 ± 0.2	2.3 ± 0.1	2.2 ± 0.1	2.5 ± 0.3	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.1	2.2 ± 0.1
Ca2+–G29(O)	2.3 ± 0.1	2.3 ± 0.1	2.3 ± 0.1	2.4 ± 0.1	2.9 ± 0.3	2.3 ± 0.1	2.3 ± 0.1	2.3 ± 0.1	2.4 ± 0.2
S(O22)–G29(HN)	3.8 ± 0.9	3.0 ± 0.5	3.0 ± 0.5	3.6 ± 0.8	3.4 ± 0.7	2.6 ± 0.4	3.1 ± 0.5	2.6 ± 0.4	2.7 ± 0.5
S(O22)–Ca2+	2.9 ± 0.5	2.4 ± 0.2	2.3 ± 0.2	2.3 ± 0.1	2.3 ± 0.1	2.4 ± 0.1	2.3 ± 0.1	2.3 ± 0.1	2.4 ± 0.2
S(O4)–Ca2+	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1	2.1 ± 0.1

See Scott et al. (15) and Janssen et al. (76). Protein Data Bank nomenclature (see also Fig. 4) has been used for atom types. See Table 3 note for more details.

FIGURE 7.

Average distances of His⁴⁷(ND1)-S(C21) and S(O22)-Gly²⁹(HN) extracted from the simulations of the different sPLA₂-phospholipid complexes. Averages were calculated from a series of simulations of a particular complex, and only the last 5 ns of each simulation was used in the calculations. Dashed and dotted lines indicate the upper and lower error, based on the native substrate for His⁴⁷(ND1)-S(C21) and S(O22)-Gly²⁹(HN)), respectively. See Fig. 5 legend for more details.

MD—nucleophilic water molecule

We previously showed that for phospholipid analogs with their headgroup located in the sn-2 position, the enzymatic reaction is affected by the probability that a water molecule can reach the catalytic site (63). We speculated that a similar effect could be the reason for our observations in this study. We therefore monitored the movements of water molecules and registered when a water molecule reached the vicinity of His⁴⁷(ND1) and S(C21) (Fig. 3) (63). These two residues will hereafter be referred to as H-S. An example of a water molecule trajectory monitored in an sPLA₂-native simulation is shown in Fig. 8 A. Fig. 8 B displays images taken from the trajectory at different time intervals (t₁ < t₂ < t₃) and shows the entering of a water molecule into the H-S region. From the data, we calculated the probability of water molecules being at a certain distance from H-S; the average relative water counts extracted from the simulations are shown in Fig. 9, and the data are provided in Table 5. There are significant differences and the probability of observing a water molecule at a certain distance from H-S depends on the compound. Within the time interval of the simulations, water molecules within 3.5 Å of H-S could be detected for compounds A, D, E, G, and the natural substrate. Water molecules within 4.5 Å of H-S could be observed for compounds B, F, and H, whereas no water molecules could be observed for compound C within this range. These findings relate partly to the observed sPLA₂ activity, where we found that compounds A, D, F, and H serve as fairly good substrates for sPLA₂. Our simulation results would suggest that compounds A, D, E, and G are relatively good substrates for sPLA₂, whereas compounds B, F, and H are less efficiently hydrolyzed by sPLA₂. Based on the relative water count, the activity can be ranked as follows: natural substrate ≈ A > G ≈ E > D > H ≈ F ≈ B. Besides compounds E and G, our findings agree well with the experimental data. To further elucidate the discrepancy observed for compounds E and G, we have followed the trajectories of water molecules and analyzed with which residues/atoms the water molecules interact as they approach H-S.

FIGURE 8.

(A) Trajectory of a water molecule monitored in an sPLA₂-Native simulation. Distances are calculated with respect to the substrate S(C21) (upper) and His⁴⁷(ND1) (lower). (B) Snapshots taken from the trajectory displayed in A. Images are taken at different time intervals (t₁ < t₂ < t₃) and show the entrance of a water molecule into the H-S region. Note that, for clarity, only the water molecule that is entering the H-S region is shown.

FIGURE 9.

Relative water count extracted from the different simulations and averaged over a series of simulations of a particular complex. Before calculating the averages, the data extracted from each simulation were arbitrarily normalized by the water count at 6 Å. The numbers given in the graph belong to the third column and represent the ratio of the relative water count at 4.5 Å to the relative water count at 5.5 Å.

TABLE 5.

Relative water count extracted from the different simulations

Compound	Relative water count at 3.5 Å (mean ± SD)	Relative water count at 4.5 Å (mean ± SD)	Relative water count at 5.5 Å (mean ± SD)
A	0.02 ± 0.02	0.4 ± 0.3	0.6 ± 0.2
B	0	0.002 ± 0.003	0.3 ± 0.2
C	0	0	0.0002 ± 0.0005
D	0.0002 ± 0.0003	0.2 ± 0.3	0.6 ± 0.5
E	0.01 ± 0.02	0.2 ± 0.3	0.4 ± 0.3
F	0	0.004 ± 0.007	0.2 ± 0.3
G	0.1 ± 0.1	0.2 ± 0.3	0.3 ± 0.3
H	0	0.01 ± 0.02	0.3 ± 0.4
Native	0.1 ± 0.1	0.4 ± 0.2	0.7 ± 0.1

Before calculating the averages, the data were arbitrarily normalized by the water count at 6 Å, since there were significant differences in the water count within a series of simulations of a particular complex. See Table 3 note for more details.

MD—key structural interference

Despite the relatively high water count for compounds E and G, these phospholipid analogs are not hydrolyzed by sPLA₂. Further analysis of the water trajectories provided additional information about the effect of the sn-1 side chain on the hydrolysis. The simulation revealed that the two methyl groups located at the first carbon of the sn-1 side chain cause steric hindrance such that water molecules cannot freely reach the H-S site. Fig. 10 displays snapshots of the active site of sPLA₂ taken from sPLA₂-compound D, sPLA₂-compound E, and sPLA₂-compound G simulations. The image taken from the sPLA₂-compound D simulation is included as a reference. This compound is a relatively good substrate for sPLA₂ and, as shown in Fig. 10, a water molecule can freely approach the H-S region. In contrast, for compounds E and G (Fig. 10, configuration 1), a water molecule can only be detected in the vicinity of His⁴⁷ and cannot reach the H-S region. A methyl group located at the first carbon of the sn-1 chain blocks the water molecule from approaching the correct position (i.e., H-S region) such that it can act as the nucleophile in the enzymatic reaction. Furthermore, His⁴⁷ cannot perfectly align in the dyad, due to the bulkiness of the methyl group. For compound G, a second configuration is observed where the carbonyl oxygen of the ester group coordinates to a water molecule that has entered the H-S region (Fig. 10, configuration 2). Here, strong coordination of the water molecule by the ester group prevents activation of the water molecule for the nucleophilic attack.

FIGURE 10.

Snapshots of the active site of sPLA₂ taken from sPLA₂-compound D, sPLA₂-compound E, and sPLA₂-compound G simulations. Secondary protein structure of the enzyme is shown in the ribbon mode. His⁴⁸, the substrate, and a water molecule are shown in the liquorice and are colored according to atom type (H, white; N, blue; O, red; and P, gold). The cofactor calcium is shown in van der Waals and is colored blue.

CONCLUSION

We used biophysical techniques and MD simulations to understand, on a molecular level, the structural restrictions for substitutions in the sn-1 position. We studied phospholipid analogs with sn-1 substitutes of different properties, where size (i.e., bulkiness) and polarity are varied (Fig. 1). The biophysical characterization revealed that all phospholipid analogs form MLVs, but only compounds A, D, F, and H are potential good substrates for sPLA₂.

Molecular dynamics simulations provided structural insight on an atomic level. Notwithstanding the clearly demonstrated importance of transition state studies (82–88), we suggest that the steps before the transition state can also critically determine the outcome of the reaction. Our simulations revealed that the difference observed in sPLA₂ activity is caused by less efficient access of water molecules to the active site. We monitored the number of water molecules that enter the region between His⁴⁷(ND1) and S(C21) (Fig. 3) for the different sPLA₂-phospholipid complexes and found that the probability of a water molecule reaching the correct position such that hydrolysis can occur varies among the different phospholipid analogs. Based on the relative water count, the activity can be ranked as follows: natural substrate > A > D > H ≈ F. In agreement with experimental results, sPLA₂ shows no activity toward compounds E and G, since the methyl group located at the first carbon of the sn-1 substitute (compound E) or the ester group (compound G) blocks water molecules from reaching the right position that would allow hydrolysis. The future plan for this work is to use the MD simulations to predict the ability of sPLA₂ to hydrolyze a given substrate with more complicated residues in the sn-1 position. Consequently, our approach may be used in the design and development of new phospholipids with anticancer drugs linked to the sn-1 position.