Assay of phospholipases A2 and their inhibitors by kinetic analysis in the scooting mode

Several cellular processes are regulated by interfacial catalysis on biomembrane surfaces. Phospholipases A2 (PLA2) are interesting not only as prototypes for interfacial catalysis, but also because they mobilize precursors for the biosynthesis of eicosanoids and platelet activating factor, and these agents ultimately control a wide range of secretory and inflammatory processes. Since PLA2 carry out their catalytic function at membrane surfaces, the kinetics of these enzymes depends on what the enzyme ‘sees’ at the interface, and thus the observed rate is profoundly influenced by the organization and dynamics of the lipidwater interface (‘quality of the interface’). In this review we elaborate the advantages of monitoring interfacial catalysis in the scooting mode, that is, under the conditions where the enzyme remains bound to vesicles for several thousand catalytic turnover cycles. Such a highly processive catalytic turnover in the scooting mode is useful for a rigorous and quantitative characterization of the kinetics of interfacial catalysis. This analysis is now extended to provide insights into designing strategy for PLA2 assays and screens for their inhibitors.

992 Rapid Communications of Oxford Ltd possibility that the release of arachidonate and lysophospholipids from membrane phospholipids is probably the rate-limiting step in the biosynthesis of eicosanoids (prostaglandins, thromboxanes, leukotrienes, lipoxins, hydroxyeicosatetraenoic acids) and may be involved in the generation of the precursor for platelet activating factor. These regulatory molecules ultimately control a wide range of physiological and pathological processes leading to inflammation, rheumatoid arthritis, asthma, ischaemia, toxic shock, psoriasis, pancreatitis, and burn trauma.
Secretory PEA e: Secretory phospholipases A 2 are ubiquitous in secretory granules of inflammatory cells and in extracellular exudates (pancreatic, ascitis, synovial and peritoneal), and venoms (snakes, bees, or lizards). Their molecular and catalytic properties are similar, which suggests a role for these enzymes in digestion and defence mechanisms, including macrophage functions. The genes coding for type I PLA2 (in pancreas, spleen, gastric mucosa, lung, stomach, kidney, and elapid snakes of the Old World) and type II PLA2 (in venoms of crotalids and viperids of the New World, platelets, macrophages, ascites, neutrophils, placenta, tonsil, kidney, synovial fluid and liver mitochondria) are localized on chromosomes 12 and 1, respectively. 3 These stable and water-soluble enzymes contain about 120 amino acids in a highly conserved single polypeptide chain with a rigid three-dimensional structure stabilized by seven disulphide bridges. 4,s The two types of PLA2 are distinguished by the translocation of Cys from positions 11 and 77 in type I, to 50 and 132 in,type II, which also contains seven additional amino acid residues at the C-terminus. Calcium is an essential cofactor for the binding of the substrate to the active site via Asp-49 and a highly conserved loop. The Kca values for these enzymes are in the 0.1 to 3 mM range. The invariant catalytic triad consists of His-48, Asp-99 and a water molecule. Seceetory PEA2 have relatively broad substrate specificities and apparent catalytic turnover numbers of 50 to 1 000 s -1. Structural analysis by X-ray diffraction ('-9 has revealed a common three-dimensional architecture in which five of the seven disulphide bridges are conserved in PLA2 of types I and II, and these structures are not noticeably altered by the binding of substrate analogues to the active site. Although the bee venom phospholipase A 2 (type III) has a considerably different architecture, the conformations of all the three types of PLA2 in the region of the substrate binding site are essentially the same and involve similar amino acid residues. Cytoplasmic PLA 2: Cytoplasmic phospholipases A2 of higher molecular weight  range) have been reported in the cytoplasm of several cell types. An 86 kDa protein has been cloned 1' and the cDNA inferred amino acid sequence shows no homology with that of the secretory PLA2. Also, calcium and other divalent ions promote binding of cytoplasmic PEA2 to the substrate interface through a specific domain. -3 The properties and substrate specificity of these cytoplasmic PLA2 suggest that they may be primarily involved in the mobilization of arachidonate triggered by submicromolar concentrations of calcium required for the binding of the enzyme to the interface.
Phospholipase A 2 activities with very different characteristics have been reported in a variety of tissues, and their relationships to these two classes of PLA2 is not yet established. Similarly, the role and regulation of PLA2 in the inflammatory processes is not clear. Secretory PEA2 accumulate in certain body fluids during inflammatory responses; however, it is not certain whether phospholipase A 2 causes inflammation or is produced in response to inflammation. In this review we focus on the interfacial kinetic properties of secretory phospholipases A2, primarily because they are well characterized. However, the kinetic behavior of cytoplasmic PLA2 abides by the same general principles of interfacial catalysis. 1 The kinetic paradigm Key features of interfacial catalysis are adequately represented by the minimal kinetic scheme shown in Fig. 1 Within the paradigm of the kinetic scheme in Fig.  1, phospholipase A 2 is distributed between the aqueous phase and the interface, and catalysis is mediated only by the enzyme at the interface (designated as E*). Thus the concentration of the substrate is important in two ways: 14'15 the bulk concentration, most conveniently expressed in units of moles/litre, controls the fraction of the enzyme at the interface. On the other hand, the mole fraction of the substrate in the interface determines the probability of the encounter of the substrate with the enzyme at the interface, that is, it is the concentration that the bound enzyme 'sees' for the formation of the Michaelis-Menten complex (E'S).
Thus the effective rate of hydrolysis is determined not only by the number density of substrate molecules in the interface as related to the mole fraction of the substrate, but also by the E to E* equilibrium that depends upon the bulk concentration of the interface.
The apparent catatic turnover is influenced by the interface" The behaviour of amphipathic molecules in aqueous dispersions and factors governing the organization and dynamics of phospholipids at the interface are now reasonably well understood. [21][22][23][24] The hydrophobic effect provides the driving force for amphipathic molecules in water to form organized structures, such as bilayers, monolayers, micelles and emulsions. In such aggregates, the dynamics of exchange of monomers depends on the shape of the amphiphile, and on the balance between the hydrophobic and hydrophilic forces on the amphiphile in the aqueous environment. The free energy for the removal of long chain phospholipid molecules from a bilayer to the surrounding aqueous phase exceeds 20 kcal/mole. Therefore, the concentration of solitary monomers in the aqueous phase is well below 100 pM. Under such conditions intervesicle exchange of phospholipids has a half-time of several hours. Amphiphiles in micelles undergo faster exchange because of fission and fusion of micelles.
The enzyme at the interface also facilitates the steps involved in the catalytic turnover. 2s '26 In the context of the scheme shown in Fig. 1, the apparent activation of PLA2 at the interface is a necessary consequence of the accessibility and availability of the substrate to the enzyme at the interface. How a phospholipase A 2 molecule at the interface optimizes the catalytic turnover can be operationally viewed as follows. 26 Since the catalytic site of secretory PLA2 is about 1.5 nm away from the surface of the protein, it is likely that the bound enzyme must dislodge the substrate monomer from the interface to the active site without bringing the substrate in contact with the aqueous phase. It appears that this is achieved by desolvation of the microinterface between the enzyme and the interface where the substrate is localized. 25 The'controversies': There is a general consensus on the existence of the E to E* step as a prelude to interfacial catalytic turnover (Fig. 1). Thus, depending on the equilibrium and kinetic properties, this step may or may not influence the observed steady state reaction rate. Many practical and theoretical difficulties arise because, for a quantitative interpretation of interfacial catalysis, one must keep track of the fraction of the enzyme in the interface, its residence time at the interface, and the local concentrations of the interacting species (substrate, products, inhibitor, calcium) that the bound enzyme 'sees' at the interface. Based on such considerations, for the kinetic analysis it is often necessary to have a detailed quantitative knowledge of the distribution and dynamics of the exchange of the enzyme, substrate, products, and other additives such as surface diluents, activators and inhibitors. Under most of the commonly used conditions for the assay and kinetic analysis, such variables are not controlled. Also in most published analyses of interfacial catalysis 23'27 such variables and the constraints of the dynamics of the substrate and products are not explicitly considered. In short, most, if not all, of the observed anomalies in the interfacial kinetics of PLA2 can be resolved within the general paradigm of the scheme in Fig. 1 if proper attention is paid to the factors regulating the E to E* equilibrium. 14'26 Indeed, for such kinetic analyses it is not necessary to make ad hoc assumptions about slow 'penetration' of the enzyme into the interface, nor about dimerization or acylation of PLA2 at the interface. Also it has been experimentally demonstrated that such events do not occur and are not required for the full catalytic activity of PEA2 at the interface. 4,s'8'2a-3 The steady state condition at the interface The scheme in Fig. 1 is deceptively simple. It is a remarkably versatile kinetic representation of interfacial catalysis. 4'1s For example, reaction progress curves with virtually any shape can be generated within the constraints of this scheme. This is because a rigorous description of the overall progress of the reaction requires a consideration of not only what the enzyme 'sees', but also how its local environment changes with time. Such considerations include not only the explicit terms in the scheme, but also the implicit constraints that control the 'local' concentration of the substrate and products that the enzyme 'sees' as the reaction progresses. In short, the organization and dynamics of all the molecular species at the interface control the overall effective rate of interfacial catalysis, that is, not only the factors governing the binding of PEA2 to the substrate interface but also the lateral distribution and the rate of exchange of the substrate, products and the enzyme between the interfaces of the aggregates present in the reaction mixture. Some of these considerations are illustrated by the example discussed next.
Constraints on interfacial catasis on micelles: Mixed micelles of phospholipids and detergents have been used extensively as substrates for PLA2 .5'27 They are attractive in the sense that a linear initial rate of hydrolysis is observed for several minutes. On this basis it has been assumed that the substrate concentration that the enzyme 'sees' corresponds to the total substrate in the reaction tube which remains constant for several minutes. This may be so, however as illustrated in Fig. 2 and elaborated below there is no basis for such an assumption. The major difficulties in the interpretation of the kinetics of hydrolysis of mixed-micelles by PLA2 arise from the fact that the rate of hydrolysis depends not only on the nature of the enzyme and the substrate, but also on the nature and the mole fraction of the detergent which acts as a 'diluent' for the substrate and influences the surface charge density, dispersity, and dynamics of the components. The following kinetic and equilibrium consequences of such factors deserve consideration" (a) In order to compare catalytic activities of phospholipases A 2 from different sources, one must optimize the properties of the interface for each enzyme. In effect, it is very unlikely that a single set of experimental conditions using mixed micelles can be used to compare the rates of different enzymes.
(b) It is often assumed that by increasing the bulk mole fraction of the detergent, the substrate is surface diluted. 27 Even after correction for the intermicellar concentration, this assumption would be valid only under equilibrium conditions and only if the enzyme can bind equally well to micelles of the detergent as well as to the mixed micelles. (c) For the validity of the surface dilution under the kinetic conditions, it is also necessary to assume that the rate of replenishment of the substrate on the enzymecontaining micelles is rapid on the time scale of the hydrolysis of only a small proportion of the substrate molecules in the enzyme containing mixed-micelle, ls'19 Otherwise, the observed steady state rate will depend on the kinetics of phospholipid exchange between mixed micelles.
Since such microscopic constraints for the steady-state condition are not satisfied, it is not possible to interpret the kinetics on mixed-micelles according to the Michaelis-Menten formalism. For example as shown in Fig. 2, consider what the enzyme 'sees' during the course of hydrolysis of the micelle to which it is bound. The steady-state assumption would be valid on the microscopic scale of a micelle, if and only if the rate of replenishment of the substrate and the rate of removal of products is rapid enough for only an insignificant fraction of the substrate molecules present in the enzymecontaining micelle to be hydrolysed. In a micelle of aggregation number 50 that contains a bound phospholipase A 2 with kca of 300 s-l, is hydrolysis of ten substrate molecules would change the surface concentration of substrate by 20% in 30 ms. Thus, a true steady state rate in such enzyme-containing micelles can be achieved only if the rate of replenishment of the substrate and the rate of removal of the product from the micelles is rapid on the time scale of 30 ms.
There are two possible mechanisms for the replenishment of the substrate that the bound enzyme 'sees'. If the rate of desorption of the enzyme from the interface is rapid on the 30 ms time-scale, then it would leave the micelle before a significant fraction of the substrate has been hydrolysed. Direct measurements demonstrate that this is not the case since the rate constant for the desorption of the enzyme from the interface is smaller than 0.0002 8--1,19'28 The other possibility is that the replenishment of the substrate occurs by the intermicellar exchange of phospholipids. For long chain phospholipids, the intermicellar exchange involving passage through the aqueous phase occurs only on the time scale of hours. The rate of transfer of phospholipids by a collisional process involving the fusion--fission events, occurs on the time scale of 0.1 to 30 s depending upon the concentration of the micelles, 31 and this is probably the reason why the rate of hydrolysis on micelles increases with the concentration of micelles. It is also pertinent to note that evidence about the fast intermicellar exchange of phospholipids in mixed micelles should not be based on the measurements of the exchange rate of the detergent, but rather on the exchange rate of phospholipids from the enzyme-containing micelles.
Based on such considerations, most observations on the interfacial kinetics by PLA2 on mixed micelles cannot be used to develop a detailed description of the fundamental catalytic properties. The crux of the problem is that the substrate is present in micelles that have a relatively small aggregation number and low exchange rate compared to the catalytic turnover rate. The kinetic analysis of any system requires a perfect mixing of all reactants so that every enzyme, on the average, 'sees' the same environment of substrate and product as in the bulk concentration at any particular time along the progress curve. In solution enzymology, rapid mixing is normally not a concern because the solitary monomeric reacting species disuse in a common and uniform aqueous environment. However, during interfacial catalysis this is not necessarily the case, and therefore special care has to be taken to ensure that all enzymes operate in a common and uniform environment as the reaction progresses.
Interfacial catalysis within explicitly defined constraints of organization and dynamics on anionic vesicles" Our approach for obtaining a rigorous quantitative description of the kinetics of interfacial catalysis is based on the premise that by studying the action of enzyme on relatively large substrate aggregates, such as vesicles, and by eliminating the inter-aggregate exchange of the interacting species, it is possible to rigorously describe the microenvironment of the bound enzyme. As developed in this section, such constraints on the dynamics of all the interacting species can be accomplished on vesicles of anionic phospholipids, where, under suitable conditions, the exchange of the enzyme, substrate, and products is negligible on the time scale of the entire progress curve. 5-2'32-3s Thus as elaborated below, the environment that the bound enzyme 'sees' can be rigorously described. The molecular organization and dynamics of phospholipids in bilayer vesicles is reasonably well constrained (Fig. 3), which permits an unequivocal description of interfacial catalysis under well characterized experimental boundary conditions, is Thus, imagine using vesicles, containing typicallly 10 000 to 200 000 phospholipid molecules, to which a phospholipase A 2 molecule binds with such high affinity that once bound it does not leave the vesicle. Two scenarios unfold under the conditions where there is at most one enzyme on each enzymecontaining vesicle. On a large vesicle, 10% of the substrate will be hydrolysed in about one minute by a single bound enzyme molecule. Therefore, it is possible to satisfy the steady state reaction condition on the microscopic level for this duration when the mole fraction of substrate that the bound enzyme 'sees' would remain close to one. In contrast, on a small vesicle containing about 5 000 phospholipids in the outer monolayer, due to a relatively rapid decrease in the mole fraction of substrate and accumulation of the reaction products, only a first order reaction progress curve is observed as the steady state concentration of E*S and the resulting steady state rate will decrease at each successive time point in the reaction progress curve. Vesicle of different sizes and narrow size dispersity can be prepared by the extrusion method.
Under such conditions each enzyme-containing vesicle behaves identically in time and all enzymes sample a common environment at all points during the reaction progress, is The system has a kind of phase coherence, and therefore the total product formed as a function of time is obtained by simply adding together the contribution of products from each of the enzyme-containing vesicles. This is true even though each enzyme molecule operates in its own 'isolated world' of a single vesicle. This highly processive reaction in which the enzyme remains bound to the vesicle during thousands of catalytic turnover cycles is termed scooting.
If the enzyme does not bind tightly to the vesicle it may hop from one vesicle to another during the reaction. Such hopping must be eliminated since the kinetic results under these conditions cannot be readily interpreted. For example, after the initiation of the reaction, an enzyme may leave a partially hydrolysed vesicle and bind to a vesicle that has not 'seen' enzyme previously. Alternatively, another enzyme may leave a vesicle shortly after binding to it and bind to a vesicle that has been completely hydrolysed. Clearly, hopping among vesicles gives a scrambling of the time of reaction and invalidates the kinetic analysis. Even if the hopping occurs rapidly, under such conditions one must take into consideration the contribution of the 'on' and 'off' steps for the enzyme to and from the interface in the catalytic turnover cycle, 14'1s which may not necessarily be once per catalytic cycle.
Other attractive features of interfacial catalysis in the scooting mode may also be noted.
(a) The E to E* equilibrium is completely in favour of E*, 14  (c) Binding of PLA2 to the bilayer interface does not require occupancy of the active site of the enzyme by a substrate, inhibitor or calcium, which establishes that the binding of the enzyme to the interface and the catalytic cycle are separate steps. 1<18'26 (d) Over 30 different PLA 2 from a variety of sources also exhibit catalysis in the scooting mode not only on DMPM vesicles 8 but also on covesicles of zwitterionic and anionic phospholipids. 7 This suggests that a highly processive catalytic turnover at the interface is a general property of anionic interfaces.
(e) The gross organization of the bilayer is not altered on binding of PLA 2 to the outer monolayer of anionic phospholipid vesicles. 14,15 (f) In bilayers, the reaction sequence in the aqueous phase is neglected because the concentration of solitary substrate molecules in the aqueous phase is exceedingly small (< 100 pM) compared to the apparent affinity of the enzyme for the monomeric substrate. (g) The half-time for transbilayer movement (flip-flop) of phospholipids is of the order of several hours. 2 (h) The intervesicle exchange of the substrate and the products is negligible on the time scale usually employed for the kinetic studies. 24'3 (i) The rate of lateral diffusion of phospholi-pids in vesicles is at least ten times faster than 14 the rate of catalytic turnover.
(i) The integrity of the vesicles is maintained because under suitably chosen conditions the rate of fusion of vesicles is negligibly small. 32 In addition, the products of hydrolysis of long chain phospholipids also form bilayers. 37 Thus, the integrity of the substrate bilayer is maintained even when all the substrate molecules in the outer monolayer of the target vesicles are hydrolysed, and contents of the inner aqueous compartment of vesicles are not released even when the bilayer surface is substantially covered by PLA2 .15 These observations demonstrate that not only is the bilayer organization retained during and after the hydrolysis of vesicles by PLA2, but within the constraints of the organization and dynamics of phospholipids in bilayer vesicles it is possible to rigorously describe the kinetics of interfacial catalysis in the scooting mode on anionic vesicles, the steady state initial rate of hydrolysis is observed for several minutes under these conditions. 32 Similarly, by a yet unknown mechanism, the contribution of the parallel kinetic processes without compromising the underlying catalytic mechanism for the enzyme.
Lipid transfer between vesicles: Owing to the finite size of vesicles, in the absence of intervesicle exchange of phospholipids, the initial steady state rate in the largest vesicles can be maintained only for about a minute. It can however be extended by promoting intervesicle transfer of phospholipids as accomplished in two different ways for DMPM vesicles. At calcium concentrations greater than 1.5 mM, the half-time for fusion of DMPM vesicles is of the order of a few seconds. Thus, even with small vesicles, the steady state initial rate of hydrolysis is observed for several minutes under these conditions. 32 Similarly, by a yet unknown mechanism, polymyxin B increases the rate of intervesicle transfer of DMPM molecules, such that in the presence of 1/M polymyxin B, the steady state initial rate of hydrolysis is observed for several minutes. 19 The initial enzymatic rate observed under these conditions corresponds to the steady-state rate at mole fraction 1 of the substrate at the interface.
As described later such conditions are very useful for kinetic characterization of PLA2 activity and inhibition because all the enzyme remains at the interface.
Hydro&sis of zwitterionic vesicles: Compared to anionic vesicles, the affinity of most PLA2 for zwitterionic vesicles is poor, and it increases in the presence of anionic additives. For example the dissociation constant, Kd, for pig pancreatic PLA2 from DTPC vesicles is > 10 mM, more than 10-fold larger than the dissociation constant from DTPM or DMPM vesicles (Kd < 1 pM). This difference is apparent in reaction progress curves for the hydrolysis of DMPC vesicles 4 where a lag period is observed. The duration of the lag depends on the presence of lipophilic additives and on the gel-fluid transition properties of the bilayer. The lag is shortest and the percentage hydrolysis at a single time point is highest (apparent activation) at the gel-fluid phase transition temperature of phosphatidylcholine bilayers. Such a lag is not observed in phosphatidylcholine vesicles that contain the products of hydrolysis or other anionic amphiphiles. This is due to an increase in the fraction of the bound enzyme. 36'38 Also the K for E* on DTPC vesicles remains the same (> 10 mM) either at, below, or above the phase transition temperature. 3<39 Thus the shape of the entire progress curve, including the lag period and the apparent activation, can be quantitatively accounted for in terms of the product-dependent shift in the E to E* equilibrium. 4 As fatty acid molecules are formed, the vesicles take on an anionic character and this causes an increase in the fraction of vesicle-bound enzyme with a concomitant increase in the reaction velocity. Even in the presence of the products of hydrolysis the K on the ternary vesicles is of the order of 0.1 mM. Therefore the bound enzyme hops between vesicles. As discussed earlier, under these conditions the 'on' and the 'off' steps contribute to the catalytic turnover period. Such anomalous kinetics observed with zwitterionic vesicles makes them rather unsuitable for assay of PLA2 activity and inhibition.
Uses of interfacial catasis in the scooting mode: The unique features of catalysis in the scooting mode on anionic vesicles permit quantitative characterization of virtually all aspects of interfacial catalysis. Such studies with DMPM vesicles have unequivocally demonstrated that secretory PLA2 from virtually all sources are fully catalytically active as monomers. In addition, the entire reaction progress curve has been characterized to obtain values of the interfacial rate and equilibrium parameters for the pancreatic PLA2 .5 This provides a basis for evaluating substrate specificity, 17 specific competitive inhibitors, <2'3s'4 effect of activators such as polymyxin B 19 and calcium, 1 and the kinetics of covalently modified enzymes. <8 '37 Through such studies, it has also been demonstrated that the effects of many previously reported phospholipase A 2 inhibitors are due to a shift in the E to E* equilibrium toward the enzyme in the aqueous phase. <41 Similarly, the anomalous kinetics at the gel-fluid transition or at isothermal phase transition in zwitterionic bilayer is due to a shift in the E to E* equilibrium. 14'33'3--39 These observations demonstrate that even though the organization and dynamics of the amphiphiles in the interface has a profound influence on the E to E* equilibrium, the primary catalytic parameters for the turnover in the interface are virtually insensitive to the nonspecific organizational perturbations. Through such protocols that unequivocally distinguish the kinetic contribution of the E to E* step from the steps involved in the catalytic turnover cycle, it should be possible to carry out a variety of structure-activity correlations not only with substrates and inhibitors, but also with site-directed mutants designed in an effort to understand the role of specific amino acid residues in the catalytic and interfacial events.
Assay of phospholipase Az activity and inhibition in the scooting mode The major problems encountered in the assay of phospholipase A 2 activity are due to the inability to control the E to E* equilibrium. To some extent this equilibrium depends on the organization and dynamics of the interface, therefore activities of enzymes from different sources cannot be compared readily on certain interfaces. Also the presence of lipophilic impurities can shift the E to E* equilibrium and thus modulate the apparent activity. This difficulty can be minimized to a certain extent by using sufficiently high bulk substrate concentrations so that all the enzyme molecules are bound to the interface. However, at interfaces containing zwitterionic phospholipids the apparent K d for PLA2 is often in the millimolar range. Thus, the problem becomes particularly acute with radiolabelled or fluorescently labelled substrates which are used typically in the submicromolar concentration range. Under such conditions only a fraction of the total enzyme is bound, and this fraction could change significantly in the presence of impurities. 16'4'41 Many phospholipase a 2 assays suffer from this problem, although with proper considerations the micelle-based assays have been used to identify relatively potent specific inhibitors of PLA2 .42 In addition, a major problem with such assays for PLA 2 inhibitors is that not only do they give false-positives 16

Characterization of specific inhibitors
Criteria for competitive inhibition: Within the constraints of the organization and dynamics of the bilayer, interfacial catalysis in the scooting mode can be adopted to obtain quantitative information about the whole Michaelis-Menten space, including the characterization of competitive inhibitors. 14,1s The primary considerations and the basic protocols are outlined below.
Rates of hydrolysis" The ratio of the initial rates of hydrolysis in the scooting mode in the absence (Vo) and in the presence (v) of a competitive inhibitor at mole fraction X at the interface is given by" Here, any significant effect on the size of vesicles obtained from the first order progress curve would be a caution flag against its efficacy as a PLA2 inhibitor. As summarized in Table 1 many of the inhibitors reported in the literature have been shown 16 to be nonspecific because they reduce the rate of hydrolysis by promoting the desorption of the bound enzyme. 94 Mediators of Inflammation. Vol 1992 micelles), and they partition readily and essentially completely into the bilayer. Since these inhibitors probably do not cross the cell membrane, they have had little use as pharmacological agents. However, recently we have successfully delivered M J33 in intact lung by taking advantage of the fact that the liposomes containing inhibitors are taken up by intact cells. MJ33 delivered in this fashion had no effect on the arachidonate cascade, although it selectively blocked the secretory PLA2 activated during ischaemia. 49 Independent methods to measure K*: Besides the kinetic methods described above, two other independent methods have been used to quantitatively characterize the binding of inhibitors to the active sites of PLA2.
Use of ligands: The rate of chemical modification of histidine-48 in the catalytic site of secretory PLA2 by an alkylating agent is modulated in the presence of ligands that bind to the catalytic site. 16 Such ligands include calcium, specific inhibitors, PLA2 reaction products, and substrate analogues. The kinetics of alkylation of phospholipase A 2 bound to a neutral diluent, such as 2-hexadecylglycerophosphocholine, is not modulated because such amphiphiles have poor affinity for the catalytic site. Thus, from the relative rates of alkylation of the enzyme bound to a neutral diluent in the absence and in the presence of a suitable ligand, it is possible to obtain the values of Kca, K*, K,, and K, respectively. if the substrate molecules are dispersed as solitary monomers, during the course of their hydrolysis the pig pancreatic phospholipase A 2 is in a state that resembles the enzyme bound to the interface. One of the ways in which this could happen is that binding of a substrate molecule to the active site promotes (nucleates) aggregation of additional substrate molecules so as to achieve appropriate hydrophil--liphophil balance. Thus the ES complex is stabilized as a larger aggregate with additional amphiphiles present in the medium.
Binding to the interface versus binding to the active site: Micelles of alkylphosphocholines have been used as amphiphile models in physical studies of bound PLA2. Protection from alkylation studies show that these amphiphiles have a modest affinity ,or the active site, K* 0.65 mole fraction. 18 Two processes must be considered. The bulk concentration of the amphiphile modulates the fraction of the enzyme bound to micelles (E*). Since the mole fraction of the amphiphile at the interface is always one, about 60% of the enzyme would be in the E*I form and 40% in the E* form in the presence of an excess of bulk amphiphile. Thus virtually all such results on the physical studies at saturating bulk concentrations of the amphiphile contain information about the E* as well as the E*I form of the enzyme.
Screening strategy for phospholipase Az As discussed in this review, analysis of inhibitors of PLA2 in the scooting mode eliminates the kinetic complications due to the perturbation of the E to E* step. This analysis is not designed to detect inhibitors that bind to the interfacial recognition site of PLA2 in the aqueous phase and prevent its association with the substrate interface. This is because the affinity of the enzyme toward anionic interfaces is very large and therefore the concentration of the enzyme in the aqueous phase is insignificant. Under pharmacological conditions it is conceivable that this may not be the case. General consequences of the possible equilibria at the interface under such conditions are elaborated below.
In principle, a phospholipase a 2 inhibitor could interact either in the membrane phase with E* or in the aqueous phase with E. Thus, in order to develop an optimal strategy to search for specific active site-directed inhibitors of PLA2, it is useful to consider a general scheme in which all species partition between the water and membrane phases. In order to identify inhibitors that are likely to function in vivo, it is also necessary to consider the factors that apply in the screen versus in man. The minimal scheme in Fig. 5  To develop a 'feel' for the interpretation and application of this general equation it is useful to imagine an inhibitor that has a portion that interacts with the active site of the enzyme and another portion that extends away from the enzyme and only interacts with the environment. The first requirement for the design of a potent inhibitor must be to provide maximal interactions in the active site. This part of the inhibitor structure will influence primarily the values of KI* and K.
Changes in the remaining portion, however, will have little or no effect on these constants but can influence the efficiency of the inhibitor by changing the partitioning of I and E1 between the water and amphiphile (A) phases. It is reasonable to suggest that such changes will affect K' and K by the same factor, ft. If we compare two inhibitors Here, fA* is the fraction of A that is at the interface, fa, A*/Aw. When the ratio in equation 8 is less than one, this implies that the case of small / leads to a smaller value of K ff and therefore the best inhibitor will be the one with the smallest K and that partitions mostly into the A phase. Conversely, when this ratio is larger than one, the best inhibitor will be the one with the lowest K* (also the lowest K since the change described by /g does not involve direct enzyme-inhibitor interactions) and that partitions mostly into the aqueous phase. In other words, equation 8 expresses the relative partitioning between the two phases of the enzyme--inhibitor complex and the inhibitor alone, and the phase in which the inhibitor functions most efficiently depends on this ratio.
The ratio in equation 8 is hard to predict and its value will vary from inhibitor to inhibitor. On the one hand, the partitioning of EI into the interface relative to I alone is favoured by the fact that the E1 complex can interact with the interface using both the regions of I that protrude from the enzyme (effected by /3) and the interfacial binding surface of the enzyme. On the other hand, the partitioning of I into the membrane is favoured over EI by two factors. Firstly, the portion of I that is within the confines of the enzyme will help I, but not EI, to partition into the interface. Secondly, the possible advantage of EI partitioning over I will be reduced by the fact that in man the value of fa* is expected to be significantly smaller than in the screen.
The ratio given in equation 8 will in general have a different value for every inhibitor and furthermore there is no a priori way to predict the value of this ratio. Thus, it is not a general solution to the problem of deciding whether to inhibit the PLA2 in the aqueous phase versus the membrane phase. Based on the following arguments, it is probably best to use a screen in which the phospholipase a 2 is operating in the scooting mode.
Firstly, in order to search for inhibitors that act in the aqueous phase, it is necessary to use an assay in which the enzyme is not tightly bound to the interface. However, as discussed previously there is precedence demonstrating that such an assay will pick up compounds that function as nonspecific modulators of the E to E* equilibrium. 16 Secondly, it is difficult to predict a priori whether an inhibitor that binds to E will function in vivo. This is because the degree of inhibition will depend not only on the K] value but also on the affinity of the PLA2 for the various interfaces that exist in vivo, and the magnitude of such affinities is not yet known. The scooting mode analysis detects inhibitors that have a low value of K* and such compounds would almost certainly work in man. This is because regardless of where the enzyme is located most of the time in vivo, it must go to the interface for the lipolysis reaction.
In the scooting mode analysis, E* >> E, and it is also very likely that E*I >> EI since the binding of the enzyme to anionic interfaces occurs with high affinity. With these inequalities, equation 6 (11) The same K has been used in both equations 10 and 11, since this dissociation constant is likely to be the same in both the screen and in man. Also, as indicated in these equations, the value of [A]T is different in the two systems, but this will not influence the search for inhibitors since this will introduce the same difference for all inhibitors. It is clear from these equations that structural modifications to the inhibitor that produce a lower value of K will also lead to a decrease in K ff and such changes will improve the degree of inhibition in both the screen and in man.
What are the consequences of the fact that I/IA is much larger in the screen than in man? Since in the scooting mode PLA2 are confined to the interface, it is the value of X* and its relationship to K* that will determine the fraction of enzyme that is inhibited (equation 1). If the screen is conducted with [A] creen--10-Sm and [I]r creen--10-SM, and if P is 10 3 the value of X* in the screen will be approximately 10 -s. In man, if we use [I] 10-SM and the same P 10 as in the screen, then since [A] is about 0.3 M, X* in man will be about 3 10 -. Thus, it is clear from this calculation that despite the relatively large value of I/IA in the screen, X* in the screen will be large enough so that the inhibition will not be missed. It is also important to note that an inhibitor that functions well in the screen will likely function equally well or even better in man, since a larger fraction of Iv will be in the membrane phase in man. There is one possible exception to this generalization. In man an inhibitor could be substantially partitioned into interfaces that are not available to the enzyme, i.e. fA* is very small.
Under these conditions K ff would increase.
The main consequence of the large value of I/IA in the screen is that K ff is a function of both K and P (equation 10), whereas K ff in man is almost always independent of PI (equation 11). Thus, if structural changes are made to the inhibitor that change the value of K f in the screen (and thus the degree of inhibition), it will not be immediately obvious whether the increase in potency is due to changes in P, K, or both. A simple solution to this problem is to re-run the in vitro analysis using a fiveto ten-fold higher value of [[A-IscreenxlT If K ff changes in proportion to [A] (equation 11), then the change in inhibitor potency is mainly due to a change in K/*. However, if the effect is due only to a change in PI, then equation 11 applies, and since [A]x and VA change by the same factor, the observed K ff will not change when [A]w is increased.
In summary, by using the scooting assay to search for PLA2 inhibitors, water soluble compounds that bind to E and prevent its interracial binding probably will not be found. However, it is hard to imagine a screen that would find such compounds without the interference from numerous nonspecific agents that modulate the E to E* equilibrium. Furthermore, the scooting analysis yields the value of KI* and inhibitors with sufficiently small values of this dissociation constant are guaranteed to function in vivo. The only problem that arises in translating the effectiveness of the inhibitor from in vitro system to the in vivo system is due to partitioning effects; however, this problem is readily remedied as discussed above.

Epilogue
Interfacial enzymology involves the interplay of physical and interfacial processes with biochemical catalysis. Such enzymes must access the substrate at the interface because the concentration of solitary monomeric substrate molecules in the aqueous phase is very low, and also the rate of substrate desorption from the interface to the aqueous phase is very low. The microinterface between the bound enzyme and the organized substrate not only facilitates formation of the enzyme-substrate complex, but the residence time of the enzyme at the interface controls the extent of processivity at the interface. The minimum kinetic model shown in Fig. 1 permits adaptation of the Michaelis-Menten formalism as a basis to accommodate virtually all aspects of interfacial catalysis. The binding of the enzyme to the interface has two extreme kinetic consequences (Fig. 3): during catalysis in the scooting mode, binding of the enzyme to the interface influences only the pre-steady state portion of the progress curve; on the other hand, catalysis in which the enzyme hops among the ensemble of vesicles, the E to E* step, becomes a part of each catalytic cycle in the steady-state. Complex kinetic consequences with different extents of processivity are predicted 14'26 if the enzyme hops after a few catalytic turnovers. The intervesicle exchange of the enzyme, substrate and products creates further time-dependent changes in what the enzyme 'sees' during the progress of the reaction. In the absence of such parallel rate processes, as is the case in the scooting mode, the kinetics of interfacial catalysis is considerably simplified.  The kinetic studies on the characterization of specific competitive inhibitors of PLA2 have also come of age. Kinetic analysis of PLA2 in the scooting mode provides a reliable method for random screening of large compound banks in order to find novel lead compounds. It is also possible to obtain very potent inhibitors and to determine their interracial K values by using three independent methods. With the help of the inhibitors available so far, it is now possible to address detailed mechanistic questions about the catalytic steps and the nature of the transition state, and to begin to ask questions related to the biological role of phospholipases A 2.