Cutinase from Fusarium solani is a lipolytic enzyme that hydrolyses triglycerides efficiently. All the inhibited forms of lipolytic enzymes described so far are based on the use of small organophosphate and organophosphonate inhibitors, which bear little resemblance to a natural triglyceride substrate. In this article we describe the crystal structure of cutinase covalently inhibited by (R)-1,2-dibutyl-carbamoylglycero-3-O-p-nitrophenylbutyl-phos phonate, a triglyceride analogue mimicking the first tetrahedral intermediate along the reaction pathway. The structure, which has been solved at 2.3 A, reveals that in both the protein molecules of the asymmetric unit the inhibitor is almost completely embedded in the active site crevice. The overall shape of the inhibitor is that of a fork: the two dibutyl-carbamoyl chains point towards the surface of the protein, whereas the butyl chain bound to the phosphorous atom is roughly perpendicular to the sn-1 and sn-2 chains. The sn-3 chain is accommodated in a rather small pocket at the bottom of the active site crevice, thus providing a structural explanation for the preference of cutinase for short acyl chain substrates.

Introduction Lipases are lipolytic enzymes originating from a wide variety of sources, such as plants, microorganisms and higher animals. The most distinctive property of lipases is their rapid hydrolysis of ester bonds at an oil-water interface, whereas activity on monomeric substrates is generally low [l]. The first structures of lipases [2,3] already gave insight into their general architecture and mechanism of action. A common feature of all lipases is their a/p-hydrolase fold [4], and their active sites contain a 'classical' Ser-His-Asp (or Glu) catalytic triad together with an oxyanion hole, comprising backbone NH groups and sometimes side chain NHand OHgroups [S]. Hence the catalytic mechanism of lipases is similar to that of serine proteases in many respects. Organophosphonates are known to block serine hydrolases by reacting with the active-site serine, and the covalent product has been suggested to mimic the transition state of the acylation step in ester hydrolysis [6]. The X-ray structures of all phosphonate-inhibited lipases show that the phosphoryl oxygen is positioned in the oxyanion hole [7-111, in agreement with the proposed mechanism of ester hydrolysis. However, until now no kinetic evidence for the importance of oxyanion stabilization in the inhibition reaction of lipases with phosphonates has been established. Recently, we reported the synthesis of triacylglycerol analogues in which two ester bonds were replaced by a carbamoyl function, while the third ester bond was replaced by an O-p-nitrophenylalkyl phosphonate [ 121. These organophosphonates appeared to be good lipase inhibitors. In addition to the stereoselectivity for the glycerol backbone, cutinase was also found to be highly stereoselective at the phosphorus chiral centre of these inhibitors. The latter stereoselectivity is high when a diacylglycerol substituent is attached to the phosphorus but is considerably lower for a small methoxy substituent. Such enantioselectivity at phosphorus

Based on the strong inhibitory properties of (R)-2-decanoylamino-octanol-1-phosphocholine and its phosphoglycol analogue for porcine pancreatic phospholipase A2, the corresponding 2-decanoyloxy derivatives have been synthesised in both enantiomeric forms and their substrate properties for the enzyme were analysed. The high aqueous solubility in the absence of detergents, combined with low critical micelle concentrations of both the amide- and ester phospholipids allowed the estimation of the interfacial dissociation constants of the enzyme-substrate and enzyme-inhibitor complexes by kinetic and direct binding techniques.

1,2-Dioctylcarbamoylglycero-3-O-p-nitrophenyl alkylphosphonates, with alkyl being methyl or octyl, were synthesised and tested as irreversible inhibitors of cutinase from Fusarium solani pisi and Staphylococcus hyicus lipase. Rapid inactivation of these enzymes occurred with a concomitant release of one mole of p-nitrophenol per mole of enzyme. With both lipases a higher reactivity was observed when the alkyl substituent on the phosphonate is a methyl rather than an octyl chain, Both lipases are highly selective for the chirality of these compounds at glycerol and at phosphorus. Rapid inactivation at an inhibitor concentration of 0.1 mol% in 100 mM NaTDOC (t(1/2) < 60 min.) occurred when the glycerol moiety had the (R) configuration, while inhibitors of the (S) configuration react 4-10-fold more slowly. The isomer with the p-nitrophenyl octylphosphonate attached to the secondary hydroxyl group of glycerol hardly inhibited (t(1/2) > 1 day) the lipases, These results reflect the known positional- and stereopreference of these enzymes which preferentially release the fatty acid at sn-3 of natural triacylglycerols. The enzymes appeared to be even more selective for the chirality at phosphorus, the differences in reactivity of the faster and slower reacting isomers being as high as about 250-fold for the methylphosphonates and about 60-fold for the octylphosphonates. These phosphonates can be regarded as true active site-directed inhibitors. The inhibited enzymes can be considered as analogues of the tetrahedral intermediate in the acylation step that occurs during triacylglycerol hydrolysis.

It has long been proposed that the higher activity of phospholipase A(2) (PLA(2)) for substrates presented as multimolecular aggregates compared to dispersed molecules (interfacial activation) arises due to a conformational change in the enzyme, X-ray studies have, however, failed to identify any such change. Here we report the solution structures of porcine pancreatic PLA(2) both free and as a ternary complex with micelles and a competitive inhibitor, Important differences between these structures indicate that conformational changes may play an important role in the mechanism of interfacial activation in PLA(2)s.

Staphylococcus hyicus lipase is a serine hydrolase. In order to identify the active site histidine of S. hyicus lipase we have chemically modified S. hyicus lipase with 1-bromo-octan-2-one. The enzyme is rapidly inactivated by this inhibitor with a half-time of 578 s at pH 6.5 and 30 degrees C. Addition of the enzyme's cofactor calcium increases the inactivation rate approx. 2-fold. When n-hexadecylphosphocholine, a non-hydrolysable substrate analogue, is added the inactivation rate decreases about 3-fold, suggesting that a residue in the active site of S. hyicus lipase is involved in the inactivation reaction. Inactivation of S. hyicus lipase with 14C-labelled 1-bromo-octan-2-one shows that 1.4 moles of inhibitor per mole of lipase are incorporated. The results of an electrospray mass spectrometric study of the inactivated enzyme are consistent with this finding. In order to identify the modified residue, both the inactivated and the unmodified lipase were digested with cyanogen bromide followed by trypsin. The resulting peptides were analysed using HPLC and fast atom bombardment mass spectrometry. The results allow the modified residue to be assigned to the peptide Gly597-Lys612. Collision induced dissociation mass spectrometry allowed the modified residue to be identified as His-600. From these results we conclude that this residue forms part of the catalytic triad of S. hyicus lipase.

Triglyceride analogues were synthesized in which one of the primary acyl ester functions has been replaced by an alkyl group and the secondary acyl ester bond has been replaced by an acyl amino bond. The chain length at either position was varied, and both (R)- and (S)-enantiomers of each compound were synthesized. These pseudo triglycerides contain only one hydrolyzable ester bond, and they are ideally suited to studying the influence of the chain length at the 1-, 2-, and 3-position on lipase activity and on stereopreference. These substrates were used to characterize cutinase from Fusarium solani pisi. Our results show that the activity of cutinase is very sensitive to the length and distribution of the acyl chains and that the highest activities are found when the chains at positions 1 and 3 contain three or four carbon atoms. The enzyme preferentially hydrolyzes the (R)-enantiomers, but this preference is strongly dependent on the acyl chain length distribution, with (R) over (S) activity ratios varying from about 30 to 1. This enantioselectivity was found in three different assay systems: a mixed micellar, a reverse micellar, and a monolayer study, Our data suggest that at least two alkyl chains of the pseudo triglycerides must be fixed during hydrolysis. Therefore, these substrates were used to characterize mutants of cutinase with mutations in putative lipid binding domains. Two mutants (A85F and A85W) have increased activities. The results obtained with these mutants suggest an interaction of the acyl chain of the scissile ester bond with a surface loop, comprising residues 80-90, in the enzyme-substrate complex.

The Staphylococcus hyicus lipase is an acyl hydrolase with broad substrate specificity including neutral glycerides and phospholipids. To obtain further insight into the mechanism of action of this enzyme, we tested several sulfonyl fluorides as active site-directed inhibitors. The enzyme is resistant to the well-known serine protease/esterase inhibitor phenylmethanesulfonyl fluoride (PMSF), but is rapidly inactivated by hexadecylsulfonyl fluoride. The kinetics of inactivation were studied in Triton X-100 micelles. Inactivation is fast and the rate of inactivation is constant over the pH range where this lipase is active. Metal ions like Ca2+ and Sr2+ do not appreciably influence the rate of inactivation, although the enzymatic activity is significantly increased, suggesting a structural role for these ions. The S.hyicus lipase contains a consensus sequence G-H/Y-S-X-G. Substitution by site-directed mutagenesis of this serine (Ser369) by a cysteine resulted in a mutant with only 0.2% residual activity. The activity of this mutant could not be inhibited with water-soluble sulfhydryl reagents either in the presence or absence of Triton X-100 micelles. In the presence of Triton X-100 micelles, inactivation of the mutant occurred with 4-nitrophenylhexadecyl disulfide (t1/2 = 125 min) while the wild-type enzyme does not react at all. We conclude that Ser369 is the active site residue and that in water this residue is inaccessible. Only after interfacial activation Ser369 (or Cys369) becomes exposed and reacts with irreversible inhibitors.

The inhibitory power (Z) of a number of (R)-1-alkyl-2-acylamino phospholipid analogues was determined for three mammalian phospholipases A2 from pig, ox and horse pancreas. All three enzymes display a clear preference for anionic (phosphoglycol) inhibitors over the zwitterionic (phosphocholine) derivatives; this effect is most pronounced for the bovine enzyme. Upon variation of the 1-alkyl chain length, the bovine and equine phospholipases, like the porcine enzyme in previous studies, show an optimum in Z for a six-carbon alkyl group. The introduction of a double bond in the 2-acylamino group generally improves the inhibitory power as compared with a fully saturated acyl chain. For the horse enzyme, the presence of an (R)-2-undecenoylamino group in the phosphocholine- and phosphoglycol-containing inhibitors resulted in affinities which are nearly 4 and 5 orders of magnitude higher, respectively, than for the substrate molecule. Direct determination of the dissociation constant Ki* of several inhibitors incorporated in a host lipid/water interface of non-inhibitory n-octadecenylphosphocholine micelles, was performed by ultraviolet difference spectroscopy. The progressive binding of a single inhibitor molecule into the active site of the three enzymes was followed quantitatively by an increasing tyrosine perturbation. With moderately strong competitive inhibitors (Z values ranging from about 50 to 10,000), quantitative values for Ki* were obtained. Extrapolation of the experimentally found linear relationship between Z and 1/Ki* yields predicted Ki* numbers for the much stronger inhibitors with Z values between 10,000 and 100,000.

The X-ray structure of a mutant porcine pancreatic phospholipase A2 inhibitor complex [Thunnissen et al. (1990) Nature 347, 689-691] has been determined. This structure shows several interactions between the sn-2-acyl chain and the phosphate moiety of the inhibitor at sn-3 and the protein. The interactions of the remaining part of the polar head group are less clear. Because Arg53 is in close proximity to the head group, we tested the importance of charge at position 53 on enzymatic activity and specificity. Arg53 has been replaced by a glutamine and a glutamic acid in mutants R53Q and R53E, respectively. The effects of the mutations were tested with both zwitterionic and anionic substrates. With monomeric, zwitterionic, (R,S)-1,2-dihexanoyldithiopropyl-3-phosphocholine as substrate, the mutants R53Q and R53E display twofold and sevenfold, respectively, increased kcat/Km values, composed of increased kcat and decreased Km values. Tested on micelles of zwitterionic (R)-1,2-dioctanoylglycero-3-phosphocholine the mutants R53Q and R53E are more active than the native enzyme, whereas these mutations have an opposite effect on the activity on anionic (R)-1,2-dioctanoylglycero-3-phosphoglycol. Thus, whereas the native enzyme is 0.3 times as active on zwitterionic as on the anionic substrate, these ratios are 1.0 (R53Q) and 1.7 (R53E) for the mutants. No changes in activity were observed with the anionic substrate (R)-1,2-dioctanoylglycero-3-sulfate. Binding studies with substrate-derived inhibitors confirmed the increased affinity for zwitterionic phospholipids and the reduced affinity for anionic phospholipids. The kinetic and binding data indicate the involvement of the charge of residue 53 in head-group specificity and suggest a position of residue 53 closer to the choline or glycol than to the phosphate.

We have determined some kinetic parameters of rat platelet phospholipase A2, such as surface pressure dependency and substrate specificity, using the monomolecular film technique. We found that rat platelet phospholipase A2 is very specific for phospholipids having a negatively charged headgroup, no activity was detected when using zwitterionic phospholipids such as phosphatidylcholine. Furthermore, the interfacial pressure window which permits enzyme activity is very narrow as compared to pancreatic phospholipase A2. Maximal enzyme activity is found at 22 mN/m when using 1,2-dilauroylphosphatidylglycerol as substrate. Studies of the competitive inhibition of mixed films containing 2-acylaminophosphatidylglycol show that platelet phospholipase A2 is less sensitive than pancreatic and intestinal phospholipase A2. These results imply that, despite the high degree of sequence similarity, one must be very cautious in extrapolating inhibition data from one phospholipase A2 to similar enzymes from other origins.

Several 2-acylaminophospholipid analogues have been demonstrated to behave as potent competitive inhibitors of porcine pancreatic phospholipase A2 (De Haas, G.H., Dijkman, R., Ransac, S. and Verger, R. (1990) Biochim. Biophys. Acta 1046, 249-257). Their inhibitory power appeared to be strictly controlled by the stereoconfiguration around the chiral C-2 atom and effective inhibition of the enzyme was observed only when incorporated into a micellar substrate-water interface. In the present study various direct binding techniques were applied to investigate the interaction of the enzyme with pure micelles of the stereoisomeric forms of 2-tetradecyl-amino-hexanol-1-phosphocholine (R-C-14-PN and S-C-14-PN). Upon equilibrium gel filtration of the enzyme (monomeric molecular mass = 14 kDa) on calibrated Superdex columns running in micellar solutions of R-C-14-PN, the phospholipase eluted as a lipid-protein complex of 74 kDa. Under identical conditions, micellar solutions of S-C-14-PN did not give rise to high-molecular mass aggregates and the enzyme eluted at its normal 14 kDa position. Light scattering experiments, ultrasedimentation and time-resolved fluorescence spectroscopy studies confirmed the formation of a high-molecular mass aggregate between enzyme and R-C-14-PN micelles. The ultimate complex was shown to consist of four protein and about ten inhibitor molecules. Using time-resolved fluorescence spectrosopy the interaction was studied between the active site of phospholipase A2 and R-C-14-PN molecules, both incorporated in an inert lipid matrix.

For the first time, we have shown that a stereospecific interaction occurs between porcine pancreatic phospholipase A2 and a monomolecular film of amidophospholipid used as inhibitor. Direct binding experiments, using radiolabelled phospholipase A2, showed that 13 times more enzyme was bound to phospholipid films of the L series by comparison with films of the D series. These results were confirmed by indirect binding studies using re-spreading experiments. Kinetic studies of the porcine pancreatic PLA2, using enantiomeric acyl-amino phospholipid analogues, have shown that: (1) inhibitors of the L series are more potent than inhibitors of the D series, (2) inhibitors having a negative charge are more potent than zwitterionic inhibitors, (3) inhibitory power values are greater when evaluated in micellar system than in a the monolayer system, (4) the inhibitory power increases continuously with surface pressure.

The competitive inhibition of human pancreatic and a mutant human platelet phospholipase A2 (PLA2) was investigated using acylamino phospholipid analogues, which are potent competitive inhibitors of porcine pancreatic PLA2 [De Haas et al. (1990) Biochim. Biophys. Acta 1046, 249-257]. Both the mutant platelet PLA2 and the human pancreatic PLA2 are effectively inhibited by these compounds. The enzyme from platelets is most strongly inhibited by compounds with a negatively charged phosphoglycol headgroup. Compounds with a neutral phosphocholine headgroup are only weak inhibitors, whereas an inhibitor with a phosphoethanolamine headgroup shows an intermediate inhibitory capacity. The platelet PLA2 is most effectively inhibited by negatively charged inhibitors having a relatively short (four or more carbon atoms) alkylchain on position one and a acylamino chain of 14 carbon atoms on position two. For the pancreatic enzyme an inhibitor with a phosphoethanolamine headgroup was more effective than inhibitors with either a phosphocholine or a phosphoglycol headgroup. The chainlength preference of the pancreatic enzyme resembles that of the platelet PLA2. The largest discrimination in inhibition between the human platelet and the human pancreatic PLA2 is obtained with inhibitors with a negatively charged phosphoglycol headgroup, an alkyl chain of four carbon atoms on position one and a long acylamino chain of 14-16 carbon atoms on position two. Because the platelet PLA2 is thought to have several biological functions, specific inhibitors of this enzyme could have important implications in the design of pharmaceutically interesting compounds.

In a recent series of kinetic studies (De Haas et al. (1990) Biochim. Biophys. Acta 1046, 249-257 and references therein) we have demonstrated that synthetic (R)-phospholipid analogues containing a 2-acylaminogroup instead of the 2-acyloxy function found in natural phospholipids, behave as strong competitive inhibitors of porcine pancreatic phospholipase A2 (PLA2). We also showed that these analogues strongly bind to the active site of the enzyme but only after their incorporation into a micellar substrate/water interface. In the present study we investigated the interaction of native PLA2 and of an inactive PLA2 in which the active site residue His-48 has been modified by alkylation with 1-bromo-2-octanone, with pure micelles of several of these inhibitors in both enantiomeric forms by means of ultraviolet difference absorption spectroscopy. Our results show that the first interaction step between native or modified enzyme and micellar lipid/water interfaces probably consists of a low-affinity Langmuir-type adsorption characterized by signals arising from the perturbation of the single Trp-3 residue. Once present at the interface the native enzyme is able to bind, in a second step, a single inhibitor molecule of the (R)-configuration in its active site, whereas the (S)-enantiomer is not bound in the active site. The overall dissociation constant of the interfacial phospholipase-inhibitor complex is three orders of magnitude lower for micelles composed of the (R)-isomer than those of the (S)-isomer. The modified PLA2 still adsorbs to micellar lipid/water interfaces but cannot bind either of the two enantiomers into its active site and similar dissociation constants were found for lipid-protein complexes with micelles of either the (R) or the (S) inhibitors. After blanking the ultraviolet signals due to the perturbation of Trp-3 in the initial adsorption step of the enzyme to a micellar surface of a non-inhibitory phospholipid analogue, the progressive binding of a single (R)-inhibitor molecule into the active site could be followed quantitatively by a tyrosine perturbation. These titrations yielded numerical values for the dissociation constants in the interface and provide a possible explanation for the large difference in overall dissociation constants of the complexes between enzyme and micelles of (R)-and (S)-inhibitors. With the use of PLA2 mutants in which each time a single tyrosine was replaced by phenylalanine, the tyrosine residues involved in binding of the monomeric inhibitor molecule were identified as Tyr-69 and Tyr-52.

Tyr52 and Tyr73 are conserved amino acid residues throughout all vertebrate phospholipases A2. They are part of an extended hydrogen bonding system that links the N-terminal alpha-NH3(+)-group to the catalytic residues His48 and Asp99. These tyrosines were replaced by phenylalanines in a porcine pancreatic phospholipase A2 mutant, in which residues 62-66 had been deleted (delta 62-66PLA2). The mutations did not affect the catalytic properties of the enzyme, nor the folding kinetics. The stability against denaturation by guanidine hydrochloride was decreased, however. To analyse how the enzyme compensates for the loss of the tyrosine hydroxyl group, the X-ray structures of the delta Y52F and delta Y73F mutants were determined. After crystallographic refinement the final crystallographic R-factors were 18.1% for the delta Y52F mutant (data between 7 and 2.3 A resolution) and 19.1% for the delta Y73F mutant (data between 7 and 2.4 A resolution). No conformational changes occurred in the mutants compared with the delta 62-66PLA2, but an empty cavity formed at the site of the hydroxyl group of the former tyrosine. In both mutants the Asp99 side chain loses one of its hydrogen bonds and this might explain the observed destabilization.

An NMR study has been made of porcine pancreatic phospholipase A2 (PLA) in three environments: free in solution, in a binary complex with dodecylphosphocholine micelles, and in a ternary complex with a micelle and the substrate-like inhibitor (R)-1-octyl-2-(N-dodecanoylamino)-2-deoxyglycero-3-phosphoglycol. H-1 and N-15 chemical shifts, amide exchange rates, and NOE intensities are compared for the enzyme in different environments. From these data, structural differences are found for the N-terminal part, the end of the surface loop at residues Tyr69 and Thr70, and the active site residue His48, and also for the Ca-binding loop (residues 28-32). Specifically, when binding to a micelle, the side chains of residues Ala1, Trp3, and Tyr69, as well as all protons of Thr70, are found to be closer together. After subsequent introduction of the competitive inhibitor, further changes are found for these residues. The N-terminus is flexible in PLA free in solution, in contrast with the crystal structures where it adopts an alpha-helical conformation. According to the NMR data, this helix is rigidly formed only in the ternary complex. Furthermore, in the ternary complex, the N-terminal amino group and the exchangeable hydrogen at N3 of the ring of His48 are observed. We propose that PLA is activated in two steps. An initial conformational change occurs upon binding to a micellar interface. The catalytically active conformation of the enzyme, which has an extensive network of hydrogen bonds, is formed only when binding a substrate or competitive inhibitor at a lipid-water interface.

The effects of Ca2+ and substrate analogue binding on the conformational dynamics of porcine pancreas phospholipase A2 (PLA2) in different regions was explored by combining site-directed mutagenesis and time-resolved fluorescence measurements. The single tryptophan residue (Trp-3) of the wild-type protein (W3), in the alpha-helix A, was replaced by a phenylalanine residue (W3F), whereafter Trp was substituted either for leucine-31 (W31), located in the calcium binding loop, or for phenylalanine-94 (W94), located at the "back side" of the enzyme. Furthermore, mutants lacking the 62-66 sequence were constructed with the Trp at position 3 (delta W3) or 31 (delta W31). The total fluorescence intensity decays of Trp in each protein, in the protein-calcium and the protein-calcium-substrate analogue complexes, analyzed by the maximum entropy method (MEM) can be interpreted as distributions of separated lifetime classes. In the case of the W94 mutant, a major short-lived excited-state population (tau approximately 50 ps) is observed, probably deactivated by the interaction with two proximate disulfide bridges via a radiationless process. For the four other mutants, the respective barycenters of the four lifetime classes display comparable values, but the amplitude distributions are different for Trp-3 and Trp-31. The rotational mobility of the Trp residue varies along the peptide chain. Trp-3 experiences only a fast hindered motion. Trp-31 is sensitive to an additional local flexibility that is absent in the N-terminal part of the protein. The largest wobbling angle is observed at position 94. No effect of calcium binding occurs on the lifetime distribution of the Trp-3 and Trp-94 residues. Their mobilities are not affected. In contrast, calcium binding displays a strong influence on the excited-state population distribution of Trp-31. A major population decaying with the longest lifetime is selected in the W31 protein and contributes to approximately 50% of the decay. The local flexibility and the amplitude of motion of Trp-31 is wider in the protein-calcium complex than in the unliganded protein. Binding of the monomeric substrate analogue n-dodecylphosphocholine (C12PN) in the presence of calcium slightly affects the Trp-3 excited-state population distribution and its mobility. Trp-31 is more sensitive to this binding. In particular, a more restricted rotation of the Trp-31 residue and a decrease of the peptide local flexibility as protein-calcium complexes are observed in both the W31 and delta W31 mutants.(ABSTRACT TRUNCATED AT 400 WORDS)

The nearly conserved glycine-30 in porcine pancreatic phospholipase A2 has been replaced by serine. The resulting mutant G30S was expressed in Escherichia coli, purified and characterized. The mutation caused a significant drop in enzymatic activity towards monomeric and aggregated substrates, but had a limited effect on substrate binding. In contrast the affinity for calcium ions, the essential cofactor, was reduced 10-fold. The reduced enzymatic activity is attributed to a reduced stabilization of the transition state. The results are discussed in view of naturally occurring inactive phospholipase A2 homologues from snake venom.