The architecture of human acetylcholinesterase active center probed by interactions with selected organophosphate inhibitors.

The role of the functional architecture of human acetylcholinesterase (HuAChE) active center in facilitating reactions with organophosphorus inhibitors was examined by a combination of site-directed mutagenesis and kinetic studies of phosphorylation with organophosphates differing in size of their alkoxy substituents and in the nature of the leaving group. Replacements of residues Phe-295 and Phe-297, constituting the HuAChE acyl pocket, increase up to 80-fold the reactivity of the enzymes toward diisopropyl phosphorofluoridate, diethyl phosphorofluoridate, and p-nitrophenyl diethyl phosphate (paraoxon), indicating the role of this subsite in accommodating the phosphate alkoxy substituent. On the other hand, a decrease of up to 160-fold in reactivity was observed for enzymes carrying replacements of residues Tyr-133, Glu-202, and Glu-450, which are constituents of the hydrogen bond network in the HuAChE active center, which maintains its unique functional architecture. Replacement of residues Trp-86, Tyr-337, and Phe-338 in the alkoxy pocket affected reactivity toward diisopropyl phosphorofluoridate and paraoxon, but to a lesser extent that toward diethyl phosphorofluoridate, indicating that both the alkoxy substituent and the p-nitrophenoxy leaving group interact with this subsite. In all cases the effects on reactivity toward organophosphates, demonstrated in up to 10,000-fold differences in the values of bimolecular rate constants, were mainly a result of altered affinity of the HuAChE mutants, while the apparent first order rate constants of phosphorylation varied within a narrow range. This finding indicates that the main role of the functional architecture of HuAChE active center in phosphorylation is to facilitate the formation of enzyme-inhibitor Michaelis complexes and that this affinity, rather than the nucleophilic activity of the enzyme catalytic machinery, is a major determinant of HuAChE reactivity toward organophosphates.

The role of the functional architecture of human acetylcholinesterase (HuAChE) active center in facilitating reactions with organophosphorus inhibitors was examined by a combination of site-directed mutagenesis and kinetic studies of phosphorylation with organophosphates differing in size of their alkoxy substituents and in the nature of the leaving group. Replacements of residues Phe-295 and Phe-297, constituting the HuAChE acyl pocket, increase up to 80-fold the reactivity of the enzymes toward diisopropyl phosphorofluoridate, diethyl phosphorofluoridate, and p-nitrophenyl diethyl phosphate (paraoxon), indicating the role of this subsite in accommodating the phosphate alkoxy substituent. On the other hand, a decrease of up to 160-fold in reactivity was observed for enzymes carrying replacements of residues Tyr-133, Glu-202, and Glu-450, which are constituents of the hydrogen bond network in the HuAChE active center, which maintains its unique functional architecture. Replacement of residues Trp-86, Tyr-337, and Phe-338 in the alkoxy pocket affected reactivity toward diisopropyl phosphorofluoridate and paraoxon, but to a lesser extent that toward diethyl phosphorofluoridate, indicating that both the alkoxy substituent and the p-nitrophenoxy leaving group interact with this subsite. In all cases the effects on reactivity toward organophosphates, demonstrated in up to 10,000-fold differences in the values of bimolecular rate constants, were mainly a result of altered affinity of the HuAChE mutants, while the apparent first order rate constants of phosphorylation varied within a narrow range. This finding indicates that the main role of the functional architecture of HuAChE active center in phosphorylation is to facilitate the formation of enzyme-inhibitor Michaelis complexes and that this affinity, rather than the nucleophilic activity of the enzyme catalytic machinery, is a major determinant of HuAChE reactivity toward organophosphates.
Acetylcholinesterase (AChE, 1 EC 3.1.1.7) is among the most efficient enzymes, with a turnover number of over 10 4 s Ϫ1 (Quinn, 1987). Its catalytic power and the high reactivity toward organophosphorus inhibitors are believed to be determined by the unique architecture of the AChE active center, consisting of several subsites. Resolution of the three-dimensional structure of Torpedo AChE (Sussman et al., 1991), sitedirected mutagenesis, and molecular modeling together with kinetic studies of the AChE muteins with substrates and reversible inhibitors (Gibney et al., 1990;Velan et al., 1991aVelan et al., , 1991bShafferman et al., 1992aShafferman et al., , 1992bShafferman et al., , 1992cShafferman et al., , 1993Ordentlich et al., 1993aOrdentlich et al., , 1993bOrdentlich et al., , 1995Radic et al., 1992Radic et al., , 1993Barak et al., 1994;Kronman et al., 1994;Gnatt et al., 1994) (for a recent review see also Taylor and Radic (1994)) are beginning to unveil the functional role of the various active center subsites in the reactivity characteristics of the enzyme: (a) the esteratic site containing the active site serine; (b) the "anionic subsite," Trp-86(84) 2 ; (c) the hydrophobic site for the alkoxy leaving group of the substrate including residues , , and Phe-338(331); and (d) the acyl pocket,  and . Apart from the esteratic subsite, the main contribution of these active center components to the catalytic activity appears to be in the stabilization of the Michaelis-Menten complexes, since most of the structural perturbations of the active center hardly affect the rate of the nucleophilic step (Ordentlich et al., 1993a). In these complexes, the trigonal geometry of the substrates and their limited structural variety allow only partial mapping of the specific interactions with the other elements of the active center. In addition, the Michaelis-Menten constants (K m ) only approximate the true dissociation constants of the noncovalent complexes. The saturation phenomena demonstrated for inhibition of AChE by a variety of organophosphorus compounds indicate the intermediacy of a noncovalent complex in the phosphorylation process. Formation of such Michaelis complexes, in the AChE active center (analogous to the Michaelis-Menten complexes), is followed by nucleophilic reaction of the catalytic serine resulting in stable tetrahedral adducts (Aldridge and Reiner, 1972;Main, 1976). In these reactions, the tetrahedral configuration of the organophosphorus inhibitors affords an additional dimension in probing the spatial organization of the AChE active center. Furthermore, since in phosphorylation a process analogous to deacylation does not occur (or takes place very slowly), the dissociation constant of the enzyme-organophosphate Michaelis complex can be kinetically estimated, provided no other enzyme-inhibitor complexes are formed (e.g. interaction with the peripheral anionic site, Friboulet et al., 1990). In the past most of the structure-activity studies of organophosphate inhibitors were interpreted in terms of the covalent adduct's stabilization (Jä rv, 1984;Benschop and de Jong, 1988;Barak et al., 1992). The contribution of the Michaelis complex formation was omitted mainly due to the lack of structural information essential for definition of the inhibitor binding sites. On the other hand, the reactivity and stereoselectivity of AChE toward various organophosphate derivatives are critically dependent upon the nature of the groups surrounding the tetrahedral phosphorus (Berman and Leonard, 1989).
In the present study, we further explore the functional architecture of AChE active center and the reactivity characteristics of the enzyme by means of site-directed mutagenesis and kinetic studies with three different organophosphate derivatives. Symmetrically substituted phosphate inhibitors have been used in order to avoid chirality at the phosphorus, which would complicate the kinetic analysis. Phosphates bearing different substituents allowed us to probe both the hydrophobic interactions of the alkoxy moieties with the binding environment of the active center and the effects of different leaving groups on the reactivity of AChE. We show that unlike in the case of substrates, the unique organization of the active center contributes predominantly to the formation of the enzymeorganophosphate Michaelis complexes and that stabilization of these complexes is the major determinant of AChE reactivity toward organophosphorus inhibitors.

Recombinant HuAChE and Its Mutants
Expression of recombinant HuAChE and its mutants in a human embryonal kidney-derived 293 cell line (Shafferman et al., 1992a;Velan et al., 1991aVelan et al., , 1991bKronman et al., 1992) and generation of all the mutants were described previously (Shafferman et al., 1992a(Shafferman et al., , 1992b(Shafferman et al., , 1992cOrdentlich et al., 1993aOrdentlich et al., , 1993b. Stable recombinant cell clones expressing high levels of each of the mutants were established according to the procedure described previously .

Kinetic Studies and Analysis of Data
AChE activity was assayed according to Ellman et al. (1961) (in the presence of 0.1 mg/ml bovine serum albumin, 0.3 mM 5,5Ј-dithiobis(2nitrobenzoic acid), 50 mM sodium phosphate buffer, pH 8.0, and various concentrations of ATC), carried out at 27°C and monitored by a Thermomax microplate reader (Molecular Devices). The apparent bimolecular rate constants for the irreversible inhibition of HuAChE enzymes by organophosphates DFP, DEFP, and paraoxon (k i ) were determined by the two following methods.
Method A-Phosphorylation experiments were carried out in at least four different concentrations of organophosphorus inhibitor (PX, see Schemes 1 and 2). The overall inhibitor concentration range was 10 Ϫ8 -10 Ϫ6 M, and for an individual enzyme the concentration span was about a factor of 10. In the first procedure, enzyme residual activity (E) was monitored at various times. The apparent bimolecular phosphorylation rate constants (k i A ) determined under pseudo-first order conditions (Scheme 1) were computed from the plot of slopes of ln(E) versus time at different inhibitor concentrations. Rate constants under second order conditions were determined from plots of ln{E/(PX 0 Ϫ (E 0 Ϫ E))} versus time.
Method B-Determination of the bimolecular rate constants for phosphorylation (k i B ) was carried out following the double-reciprocal method of Hart and O'Brien (1973). Apart from bimolecular rate constants this method allows us to evaluate the apparent dissociation constant for the enzyme-inhibitor Michaelis complex (EPX; K d ϭ k Ϫ1 /k 1 ) and the phosphorylation rate constant of the reaction (k 2 , see Scheme 2). The enzyme is reacted simultaneously with an excess of inhibitor and of substrate that ensure pseudo-first order conditions with respect to both reactions. The kinetic data were analyzed according to the reaction depicted in Scheme 2. Slopes of the tangents at each 20-s interval were obtained as part of fitting the cubic spline curve through the experimental points (Rogers and Adams, 1990). Semilogarithmic plots of these slopes against time resulted in linear correlations for all the concentrations of the inhibitors used. The slopes (⌬ln(v)/⌬t) were determined by linear regression. These values are related to the kinetic parameters of the inhibition process according to the following expression.
While in method A inhibitor concentrations fulfilled the condition [PX] Ͻ Ͻ K d , in method B inhibitor concentrations for most enzymes were in the range of 0.3 ϫ K d up to 10 ϫ K d . In these experiments, for all enzymes except for W86A, saturating concentrations of substrate (e.g. [ATC]/K m Ͼ 3.5) were used.
Molecular Modeling-Building and optimization of three-dimensional models of the HuAChE-organophosphate Michaelis complexes were performed on a Silicon Graphics IRIS 70/GT workstation using SYBYL modeling software (Tripos Inc.). The initial models were constructed by manual docking of the ligands into the active site gorge using the following guidelines. (a) The alkoxy substituents were placed into the acyl pocket with the oxygen positioned according to the location of the methyl group in the model of HuAChE-ATC tetrahedral intermediate (Ordentlich et al., 1993a). (b) The PϭO bond was positioned in a way that minimizes the distance of the phosphorus from the O ␥ -Ser-203, and the distance from the phosphoryl oxygen to amide nitrogens of residues Gly-121, Gly-122, and Ala-204. The resulting structures were optimized by molecular mechanics using the MAXMIN force field (and AMBER charge parameters for the enzyme) and zone-refined, including 127 amino acids (15-Å substructure sphere around Ser-203). Optimization of the initial models included restriction of the distance between the phosphorus and O ␥ -Ser-203 to 2.8 Å, which was relieved in the subsequent refinement.

Selection of Organophosphate Inhibitors and HuAChE
Mutants for Study-For both the Michaelis complex and the tetrahedral adduct we assume that the positioning of the phosphoryl moiety in the AChE active center is determined by the orientation of the PϭO bond and by the P-O ␥ -Ser-203 distance. Therefore, the interactions of the phosphate inhibitor with the elements of its binding site are mainly through the alkoxy substituents and the leaving group. Such interactions can be studied by using DFP and DEFP, which share the same leaving group (fluor) versus paraoxon, in which the leaving group is p-nitrophenoxy (Fig. 1). Similarly, DEFP and paraoxon, having the same alkoxy moieties are compared with DFP. Selection of mutants for this study was guided by SCHEME 1 SCHEME 2 previous mapping of the active center, which identified residues constituting the functional subsites for interaction with substrates and presumably also with covalent inhibitors. These residues include Tyr-337 and Phe-338 (which are part of the hydrophobic alkoxy site); Trp-86 (the "anionic" subsite (Ordentlich et al., 1993a(Ordentlich et al., , 1995; Glu-450, Glu-202, and Tyr-133 key elements maintaining the functional architecture of the active center (Ordentlich et al., 1993b(Ordentlich et al., , 1995; and Phe-295 and Phe-297, which confer specificity of the acyl pocket (Fournier et al., 1992;Harel et al., 1992;Vellom et al., 1993;Ordentlich et al., 1993a;Gnatt et al., 1994).
Kinetic Analysis of Interaction of HuAChE Enzymes-Elucidation of the specific HuAChE-inhibitor interactions in the Michaelis complex and during the subsequent nucleophilic process is essential for understanding the reactivity of AChE toward phosphate derivatives. Towards this goal, the apparent bimolecular rate constants for the irreversible inhibition of the HuAChE enzymes were determined under experimental conditions (in the presence of substrate) that allow the evaluation of the apparent dissociation constants (K d ) as well as the phosphorylation rate constants k 2 (Hart and O'Brien (1973); see Scheme 2). Since in this method the apparent bimolecular rate constants of inhibition (k i B ϭ k 2 /K d , method B; see "Experimental Procedures" and Fig. 2) are determined using high concentrations of inhibitor, the calculated values of K d may include contributions from other possible enzyme-inhibitor complexation processes and consequently may not correspond to the dissociation constants of the Michaelis complexes. Therefore, it was important to confirm the values of k i by an alternative procedure (method A) in which much lower inhibitor concentrations could be used ([PX] Ͻ Ͻ K d ) and where k obs becomes insensitive to k 2 (Gray and Duggleby, 1989;Kovach, 1991). Under these conditions, the values of k i A are determined directly from the residual enzymatic activity after various periods of exposure to the inhibitor, and in all cases a linear correlation of k obs versus inhibitor concentration was observed. This indicates that the reaction involves a single enzyme-inhibitor equilibrium, which is presumably the Michaelis complex (Aldridge and Reiner, 1972). The combination of the two methods allowed examination of the k i values over a 1000-fold range of inhibitor concentrations. At the lower end of the range k obs approximates k i /[PX], while at the high concentrations k obs approaches the limiting value of k 2 . For most HuAChE enzymes, an excellent correlation between the bimolecular rate constant values determined by the two methods was observed. This correlation of k i values, extending over 3 orders of magnitude, is linear irrespective of the nature of mutation or inhibitor with slopes approaching unity (0.945, 0.987, and 0.999 for DFP, DEFP, and paraoxon, respectively; see Fig. 3). The single HuAChE enzyme, which consistently deviates from these correlations is Y337F. However, for reasons not clear to us, an outstanding standard deviation in determinations of k i A was also observed in this case. In the log k i A versus k i B plots (Fig.  3) three major clusters of HuAChE enzymes can be defined. Interestingly, these three clusters correspond to the division according to the functional subsites in the HuAChE active center: mutants of the acyl pocket (F295A, F295L, F297A, F297V, and F295L/F297V) exhibiting the highest k i values of all the mutants tested; mutants of the alkoxy pocket (W86A, Y337A, Y337F, and F338A) having k i values close to that of the wild type enzyme; and mutants of the hydrogen bond network (E202A, E202D, E202Q, E450A, and Y133F) having k i values consistently lower than the wild type enzyme. We note that within each cluster some variations of the k i values can be observed, indicating a direct effect of the mutation on the interactions with the inhibitor. The considerable variability of the k i values reported for AChEs from different species (e.g. for paraoxon the reported range was 1.8 -0.02 M Ϫ1 min Ϫ1 ; Wang and Murphy (1982)) probably originates from subtle structural changes similar to those described here. On the other hand, the k i values for the wild type HuAChE are in good agreement with those reported for primates and some other mammalian AChEs. For inhibition of monkey AChE by paraoxon and DFP, the k i values are 1.22 ϫ 10 6 M Ϫ1 min Ϫ1 and 3.3 ϫ 10 4 M Ϫ1 min Ϫ1 , respectively (Wang and Murphy, 1982).
The inhibitor concentrations used in the kinetic measurements of phosphorylation, according to method B, were for most mutants in the range of 0.3-10-fold or higher of the K d value of the respective HuAChE enzyme derivative. Kinetic measurements (method B) for all the HuAChE mutants, except for the W86A enzyme, were carried out at saturating levels of substrate (e.g. [ATC]/K m Ͼ 3.5). In the case of W86A such levels of substrate concentration are precluded by the high value of K m (Ordentlich et al., 1993a), and therefore the corresponding K d and k 2 values should be regarded as tentative. The calculated values of k 2 for the HuAChE enzymes are low compared with those reported by Hart and O'Brien (1973) and by Main and Iverson (1966) for the bovine erythrocyte and eel AChEs, respectively. However, many of the more recent studies report values closer to those obtained by us (Liu and Tsou, 1986;Kemp and Wallace, 1990). In particular, inhibition studies of mouse AChE mutants by paraoxon resulted in k 2 values in the range of 3.4 -1.8 min Ϫ1 (Radic et al., 1995), while inhibition of Torpedo californica AChE mutants by DFP yielded k 2 values in the range of 0.3-1.6 min Ϫ1 (Radic et al., 1992). Notably, the rate constants of the chemical transition from the enzymephosphate Michaelis complex to the phosphorylated enzyme (k 2 ) varied within a narrow range for almost all of the various HuAChE enzymes as is evident from the similar intercepts in Fig. 2 (see also Fig. 4). The relative invariance of k 2 values, irrespective of the enzyme source or the structure of organophosphorus inhibitors, has been already observed in several cases (Main and Iverson, 1966;Andersen et al., 1977;Forsberg and Puu, 1984;Gray and Dawson, 1987;Kemp and Wallace, 1990). In the present study the limited variability of the k 2 values is evident for a series of enzymes differing only in a single or in one case a double replacement of residues at the active center. This may suggest that perturbations of the active center architecture affect mainly the enzyme affinity toward organophosphate inhibitors as demonstrated in the variability of the dissociation constant (K d ) values over a range of 3 orders of magnitude for each of the inhibitors (Fig. 4).
Effect of Mutation of Acyl Pocket Residues on Interactions with Various Organophosphates-Two phenylalanine residues at positions 295 and 297 were implicated in accommodation of the acyl moieties of various substrates, by site-directed mutagenesis Ordentlich et al., 1993a), by molecular modeling of HuAChE (Ordentlich et al., 1993a) and of TcAChE (Harel et al., 1992), and by comparison with BuChE (Gnatt et al., 1994). We replaced these phenylalanine residues with less bulky amino acids and studied the kinetics of inhibition for the following enzymes: F295A, F295L, F297A, F297V, and F295L/F297V (Table I). Replacements by alanine remove steric hindrance in the HuAChE acyl pocket, while introduction of valine at position 295 or leucine at position 297 follows the corresponding occupation (Lockridge et al., 1987) of these positions in BuChE.
In general, replacements at the acyl pocket enhance the reactivity toward the organophosphate inhibitors, although some differences related to the inhibitor structure can be observed (Table I). Replacement of Phe-295 by alanine brings a larger decrease in the values of K d than the one observed for its replacement by valine, suggesting that phenylalanine at position 295 restricts the accommodation of the ethoxy group at the acyl pocket. Removal of the bulky group from position 297 (F297A) has no effect on the affinity toward DEFP, while introduction of valine (F297V) improves the accommodation of the inhibitor. This observation is somewhat surprising in view of the prediction from previous molecular modeling ( Ordentlich et al., 1993a), which implicated both residues Phe-295 and Phe-297 in steric interference with groups larger than methyl.
The pattern of effects due to replacements at positions 295 and 297 on the affinities toward DFP is somewhat different from that of DEFP, reflecting the influence of the bulkier alkoxy group. Steric hindrance due to Phe-297 is evident from the increase in affinity of DFP compared with DEFP in the F297A enzyme. In addition the hydrophobic interactions in both F295L and F297V are more pronounced in the case of DFP. Both DFP and DEFP show enhanced affinities toward the double mutant F295L/F297V; however, the effect in the case of DFP is more pronounced, indicating again the larger contribution of hydrophobic interactions. As could be expected from the identity of their alkoxy groups, the pattern of affinity changes for paraoxon follows that of DEFP with respect to replacements of Phe-295. However, the lack of affinity enhancement of the double mutant F295L/F297V toward paraoxon may be related to the difference in the leaving groups. Unlike the fluorine in  Tables I-III. DFP and DEFP, the leaving group in paraoxon is capable of forming specific interactions with elements of the active center, which restricts its ability to optimally adjust the positioning of the alkoxy group in the acyl pocket (Fig. 5, compare panels A and B with panel C).
Effect of Mutation of Alkoxy Pocket Residues on Interactions with Various Organophosphates-The aromatic residues Trp-86, Tyr-337, and Phe-338 define a hydrophobic alkoxy pocket for the alkoxy leaving group of the substrates (Ordentlich et al., 1993a). The extent to which any of these residues interact with the substrates depends upon the charge of the substrate leaving group. As shown previously for ATC (Ordentlich et al., 1993a(Ordentlich et al., , 1995, the contribution of Trp-86, the anionic subsite, is of paramount importance for stabilization of the charged tetraalkyl nitrogen, while for the noncharged isosteric substrate 3,3-dimethylbutyl thioacetate the hydrophobic interactions with Tyr-337 and Phe-338 appear to be more significant. One can assume that the same binding pocket can accommodate one of the phosphate alkoxy moieties. However, due to its tetrahedral geometry the phosphate is expected to interact with additional elements in the active center, and the juxtaposition of the alkoxy moiety versus the aromatic side chains of Trp-86, Tyr-337, and Phe-338 should be somewhat different than for the noncharged substrates such as 3,3-dimethylbutyl thioacetate. The affinity of paraoxon toward the W86A enzyme is 12-fold lower than that for the wild type HuAChE. This effect can be attributed mainly to the nature of the paraoxon leaving group, since no comparable effects were observed for the corresponding affinities of DEFP and DFP (Fig. 5, compare panels A and B to panel C). Affinity toward paraoxon is also sensitive to replacement of Phe-338 by alanine (17-fold lower than that of the wild type enzyme). Again, no such effect is observed for either DEFP or DFP, suggesting a specific participation of the p-nitrophenyl group in interaction with Phe-338. Involvement of Phe-338 in aromatic-aromatic interaction was reported previously for the p-nitrophenyl acetate hydrolysis by HuAChE enzymes (Ordentlich et al., 1993a).
The affinities of both DFP and DEFP are not sensitive to replacements at position 337 by either phenylalanine or alanine. The comparable reduction in affinity of paraoxon toward Y337A and Y337F enzymes indicates that the main contribution to the stabilization of the complex is due to the hydroxyl group of Tyr-337 and not due to its aromatic moiety. The picture emerging from these results is that the extent of participation of residues Trp-86, Tyr-337, and Phe-338 in accommodation of the phosphate alkoxy moiety depends mainly upon the nature of the leaving group. The p-nitrophenyl moiety of paraoxon is unlikely to interact directly with all the three aromatic residues, and its indirect effect is probably due to restrictions imposed on the possible reorientation of the inhibitor in the active center (compare the relative binding environment for the leaving groups of DEFP (Fig. 5B) and paraoxon (Fig. 5C).
Although residue Tyr-133 is adjacent to Trp-86, molecular models (Ordentlich et al., 1995) indicate that it does not interact directly with the alkoxy moieties of substrates or organophosphates. The reported major reduction in reactivity of the Y133A enzyme toward organophosphates was explained by steric obstruction of the alkoxy pocket (Ordentlich et al., 1995). The effect of replacement of Tyr-133 by phenylalanine on the affinities toward all of the phosphate inhibitors studied is less pronounced and not uniform with respect to the different inhibitors. While the respective affinities for paraoxon and DEFP decrease about 5-fold, the affinity for DFP is hardly affected. The participation of the hydroxyl group of Tyr-133 in the hydrogen bond network (Fig. 5D) maintaining the relative position of the Glu-202 carboxylate in the HuAChE active center has been suggested before (Ordentlich et al., 1993b). Thus, the decrease in affinity of Y133F mutant enzyme toward paraoxon and DEFP could be attributed to the somewhat altered architecture of the active center due to the disruption of this hydrogen bond network.

Effect of Mutation of Hydrogen Bond Network Residues on Interactions with Various Organophosphates-
The hydrogen bond network in the HuAChE active center was suggested (Ordentlich et al., 1993b) to include two of the three buried acidic residues (Glu-202 and Glu-450), two water molecules that correspond to solvent molecules in the structure of TcAChE (Sussman et al., 1991), Tyr-133, and the backbone amide oxygens of Gly-122 and Gly-448 (Fig. 5D). Replacement of each of the acidic residues (Glu-202 and Glu-450) affected the catalytic activity of the resulting enzymes toward both charged and noncharged substrates (Ordentlich et al., 1993b). This effect was attributed to reorganization of the active center upon replacement of the carboxylates with noncharged moieties. In the case of phosphate inhibitors, replacement of Glu-450 by alanine resulted in a substantial decrease of the affinities toward paraoxon and DEFP (18-and 42-fold, respectively). A similar pattern of effects was observed for E202Q; however, while Glu-202 is adjacent to the catalytic serine (Ser-203), residue Glu-450 is 9 Å away. Replacement of Glu-202 by alanine or by aspartic acid produces only minor effects on the respective affinities toward DFP and DEFP, but for paraoxon the decrease in affinities is more substantial (30-and 10-fold, respectively). Quite surprisingly, the most pronounced effect on affinity is observed for E202Q for which the steric perturbation is minimal. These results indicate that the effects due to replacement of Glu-202 are not related only to charge, since E202A and E202D show comparable affinities for the phosphate inhibitors.

Interactions of Organophosphates with the Acyl Pocket-
Early investigations of AChE and BuChE active center topologies, based on structure-activity studies of several series of substrates and organophosphorus inhibitors, indicated specific binding pockets for the acyl and the alkoxy groups of the substrate (Jä rv, 1984). In addition, these studies suggested that the distinct sizes of the AChE and BuChE acyl pockets are the main structural difference in the ligand binding environments of the two enzymes. Molecular modeling of this subsite in HuAChE , based on the x-ray structure of TcAChE (Sussman et al., 1991), equally implicated residues Phe-295 and Phe-297 in conferring selectivity for ATC by restricting the dimensions of the acyl pocket (Harel et al., 1992;Ordentlich et al., 1993a). However, measurements of K m values of HuAChE enzymes for ATC and butyrylthiocholine (Ordentlich et al., 1993a; Hosea et al., 1995) revealed that while replacements at position 297 had only a minor effect on selec-

TABLE I Effect of mutation of acyl pocket residues on interactions with various organophosphates
The apparent bimolecular rate constant k i A was determined using method A, while k i B , K d , and k 2 were determined using method B (see "Experimental Procedures"). The following K m (mM) values were used for determination of ␣-parameter (Equation 1): wild type, 0.14; F295A, 0.13; F295L, 0.27; F297A, 0.43; F297V, 1.20; F297L/F297V, 0.40.  . 4. Effects of mutations at the HuAChE active center on the values of the dissociation constants K d and the phosphorylation rate constants k 2 . The figure illustrates that for a given organophosphate the variability in K d extends over a factor of 1000, while that in k 2 does not exceed a factor of 4. tivity, those at position 295 lowered the K m value for butyrylthiocholine even below that of BuChE. The absence of similar effects on k cat signified that replacements at position 295 relieve steric interference mainly in the noncovalent (Michaelis-Menten) complex of butyrylthiocholine.
The observed similarity of the overall effects of residue replacements at the acyl pocket on the values of K m for substrates and K d for phosphates provides an important insight into the organization of the respective Michaelis complexes, indicating that the acyl pocket indeed serves an analogous purpose in both cases. However, due to the added dimensionality of the phosphates, variations in the K d values reveal additional effects of the acyl pocket structural modifications. These are exemplified by the opposite effects on affinity, due to replacement of Phe-295 by alanine versus leucine, toward the ethoxy and isopropoxy substituted phosphates, indicating the interplay between the steric and the hydrophobic interactions in the acyl pocket. Replacement of Phe-297 by valine stabilizes the complex of DEFP, while it destabilizes that of paraoxon. This difference may signify the influence of the leaving group, since in the DEFP complex the enzyme-ligand assembly can readjust without affecting the relative juxtaposition of its PϭO bond relative to the active site serine and the oxyanion hole. Such plasticity is less likely in the case of paraoxon due to the additional specific interactions of the p-nitrophenoxy moiety, which prevents the repositioning of the alkoxy group in the acyl pocket (compare the model of the Michaelis complex of paraoxon with that of DEFP in Fig. 5, panels C and B, respectively).
The equivalence of the binding subsites for the substrate acyl moiety and for the phosphate alkoxy group can be further demonstrated by comparing the relative reactivity of the F295L/F297V HuAChE toward DFP with that of BuChE (in which the native acyl pocket configuration is leucine/valine).  . A, HuAChE-DFP complex; distances of the phosphoryl oxygen (PϭO) from the amide nitrogens of 2.82,and 3.84 Å,respectively); distances of the isopropoxy methyl from aromatic C -Phe-297 and C ⑀1 -Phe-295 (3.30 and 3.10 Å, respectively). B, HuAChE-DEFP complex; distances of PϭO-N-amide from 2.72,and 3.90 Å,respectively); distances of the ethoxy methyl from aromatic C -Phe-297 and C ⑀1 -Phe-295 (3.56 and 2.98 Å, respectively). C, HuAChE-paraoxon complex; distances of PϭO-N-amide from Gly-121, 2.69,and 3.80 Å,respectively); distances of the ethoxy methyl from aromatic C -Phe-297 and C ⑀1 -Phe-295 (3.60 and 2.98 Å, respectively); distance of paraoxon nitro oxygen from O -Tyr-337 (2.80 Å); distances of paraoxon aromatic C 2 and C 3 from C -Phe-338 (3.60 and 3.44 Å, respectively). D, the proposed hydrogen bond network across the active center of HuAChE, bridging three of the loops (shown by blue ribbon) that bear, respectively, the oxyanion hole (Gly-121, Gly-122), the catalytic Ser-203, and the catalytic His-447. The hydrogen bond distances, marked by dashed lines, between the active center water molecules (shown as red dots) and relevant residues within the network are as follows: The bimolecular rate constant of DFP phosphorylation (k i ) for the F295L/F297V is 80-fold higher than that for the wild type HuAChE (Table I), and likewise the corresponding rate constant for BuChE is 110-fold higher than that for AChE (Main and Iverson, 1966). In addition, the reactivity of the F295L/ F297V double mutant of HuAChE toward paraoxon is very similar to that of the wild type enzyme ( Table I) and to that of BuChE (Aldridge and Reiner, 1972). For the acylation reaction, the functional equivalence of the acyl pocket subsites in the F295L/F297V HuAChE and in BuChE was already demonstrated through the similarity (less than 2-fold) of the bimolecular rate constants for butyrylthiocholine hydrolysis by the two enzymes (Ordentlich et al., 1993a).
The emerging picture, concerning the roles of residues Phe-295 and Phe-297 as the determinants of specificity in the HuA-ChE active center, is that they form a pocket that can accommodate substituents of ligands for which the primary locus of interaction is the active site of the enzyme. In the case of ATC, accommodation of the methyl group in this pocket together with the oxyanion hole orients the molecule in plane for the incipient nucleophilic attack (Fournier et al., 1992;Harel et al., 1992;Vellom et al., 1993;Ordentlich et al., 1993a;Gnatt et al., 1994;Hosea et al., 1995). In a similar way, interaction of this subsite with organophosphates helps to orient the molecules for the in-line attack by the catalytic serine. In both cases the respective substituent is projected toward Phe-295, and therefore the corresponding complexes are more sensitive to the volume of residue at this position. Steric interactions with the residue at 297 become important for branched alkoxy substituents, which present a larger volume to the acyl pocket (see Fig.  5, A-C).
Accommodation of the Phosphate Leaving Group in the Alkoxy Pocket-As in the case of the acyl pocket, the involvement of the substrate alkoxy binding pocket in the accommodation of organophosphate inhibitors appears to be confined to the stabilization of the Michaelis complexes, since replacements of residues associated with this pocket had no effect on the phosphorylation rate constants k 2 (Table II). Only marginal effects could be observed upon replacement of the aromatic residue Trp-86, Tyr-337, or Phe-338 on the stability of complexes with DFP and DEFP, underscoring the nonspecific nature of the hydrophobic interactions of the alkyl moieties with the enzyme surface. Such an absence of contributions due to specific residues was already observed in the cases of non- charged substrates 3,3-dimethylbutyl thioacetate and propylacetate (Ordentlich et al., 1993a). A somewhat more pronounced dependence on the structure of the alkoxy pocket is evident in the case of paraoxon. However, most of these effects appear to originate from interactions with the p-nitrophenoxy leaving group (e.g. compare the effects of replacements at positions 337 and 338 on values of K d for DEFP and paraoxon (Table II)). Molecular modeling (Fig. 5C) of paraoxon Michaelis complexes with the corresponding HuAChE enzymes (Y337A, Y337F, and Y338A) demonstrates that the effects of replacement of the aromatic residue at position 338 or the hydrogen bond donating function at position 337 on the complex stability are due to interactions with the p-nitrophenyl moiety. Furthermore, the model shows that these interactions induce a somewhat different position of the ethoxy moiety, compared with the DEFP complex, bringing it closer to the indole ring of Trp-86 (Fig. 5, B-C). This proximity, together with the already mentioned limited plasticity of paraoxon complexes is consistent with the larger decrease in affinity of the W86A enzyme toward paraoxon than toward DEFP. In complexes of DEFP and DFP the corresponding fluoro leaving group does not appear to interact with elements of the alkoxy pocket.
Interaction of the p-nitrophenoxy leaving group with residue Phe-338 has already been demonstrated for the tetrahedral intermediate of p-nitrophenyl acetate (Ordentlich et al., 1993a). However, the absence of effect due to residue Tyr-337 indicates that the position of the p-nitrophenoxy moiety in the tetrahedral conjugates of p-nitrophenyl acetate and paraoxon are somewhat different. In addition, since for these two conjugates the positioning of the appropriate substituent in the acyl pocket appears to be equivalent, the locations of the other oxy substituents (ethoxy group in paraoxon or Ser-203 in the case of p-nitrophenyl acetate) have to be quite different. The potential multiplicity of the alkoxy pocket binding elements raises the possibility that both the alkoxy substituent and certain leaving groups are accommodated by this subsite. Thus, it appears that the interactions of AChE with the tetrahedral organophosphate mimic only partially those of the enzymesubstrate tetrahedral intermediate and by analogy those with the transition state of the acylation by carboxylates.

Affinity of HuAChE toward Organophosphates Is a Major Determinant of Its Overall Reactivity in the Phosphorylation
Process-The most consistent characteristics of the HuAChE enzyme's reactivity pattern toward the organophosphates studied here is that structural variations in both the enzyme and the inhibitor affect mainly the stability of the Michaelis complexes. While replacements of selected residues in HuAChE brought about changes of about 2,000-fold in the K d values, the corresponding phosphorylation rate constants (k 2 ) remained essentially unchanged (Fig. 4). Although observations regarding the limited variability of k 2 have been reported for AChEs from numerous sources, including bovine erythrocytes, electric eel, monkey, catfish, and frog brain (Wang and Murphy, 1982;Forsberg and Puu, 1984;Kemp and Wallace, 1990), this is the first study in which these reactivity characteristics of AChE have been systematically examined in a series of enzymes with limited and well defined structural differences. Furthermore, the relative invariance of k 2 was also reported for various organophosphorus inhibitors, using different experimental techniques, including stop flow methods (Forsberg and Puu, 1984;Gray and Dawson, 1987;Ryu et al., 1991). A possible reason for these observations is that while the affinity is influenced by the spatial complementarity of the organophosphorus inhibitor with the AChE active center binding environment, the nucleophilic reaction rate is dependent mainly on the nucleophilicity of the catalytic serine and the electronic properties of the phosphoryl moiety, as has been also suggested by others (Kemp and Wallace, 1990). In addition, perturbation of the hydrogen bond network in the HuAChE active center, through replacement of the participating residue or Glu-450, does not significantly affect the rate constants of the phosphorylation step (Table III). Such limited effect on the k 2 values indicates that, for the organophosphates tested here, the nucleophilic reaction at the phosphorus is not very sensitive to the altered architecture of the active center. Since the same perturbations of the hydrogen bond network were shown to affect the rate constants of the catalytic process (Ordentlich et al., 1993b) we may conclude that the participation of the active center molecular environment in facilitating the formation of covalent bonds is probably different for acylation by carboxyl and by phosphoryl esters.