Electrostatic effects and binding determinants in the catalysis of prolyl oligopeptidase: Site-specific mutagenesis at the oxyanion binding site*

Prolyl oligopeptidase, a member of a new family of serine peptidases, plays an important role in memory disorders. Earlier x-ray crystallographic investigations indicated that stabilization of the tetrahedral transition state of the reaction involved hydrogen bond formation between the oxyanion of the tetrahedral intermediate and the OH group of Tyr(473). The contribution of the OH group was tested with the Y473F variant using various substrates. The charged succinyl-Gly-Pro-4-nitroanilide was hydrolyzed with a much lower k(cat)/K(m) compared with the neutral benzyloxycarbonyl-G1y-Pro-2-naphthylamide, although the binding modes of the two substrates were similar, as shown by x-ray crystallography. This suggested that electrostatic interactions between Arg(643) and the succinyl group competed with the productive binding mechanism. Unlike most enzyme reactions, catalysis by the wild-type enzyme exhibited positive activation entropy. In contrast, the activation entropy for the Y473F variant was negative, suggesting that the tyrosine OH group is involved in stabilizing both the transition state and the water shell at the active site. Importantly, Tyr(473) is also implicated in the formation of the enzyme-substrate complex. The nonlinear Arrhenius plot suggested a greater significance of the oxyanion binding site at physiological temperature. The results indicated that Tyr(473) was more needed at high pH, at high temperature, and with charged substrates exhibiting "internally competitive inhibition."


INTRODUCTION
Pro-Nap 1 substrate provides a doubly sigmoid pH-k cat /K m profile, and is rather sensitive to ionic strength (23)(24)(25). Kinetic investigations have also shown that the rate-determining step of the reaction is a conformational change (23,26) rather than the chemical step characteristic of the chymotrypsin reactions. Studies of the structural changes induced by pH, temperature, and urea have demonstrated that the denaturation of the enzyme is promoted at 0.5 M NaCl concentration (27). These results suggest that prolyl oligopeptidase is stabilized by a water shell, which is less stable at low pH and high ionic strength.
A major catalytic difference from the trypsin-and subtilisin-type enzymes concerns the stabilization of the negatively charged tetrahedral intermediate. This is achieved by the oxyanion binding site, which provides two hydrogen bonds to the oxyanion. In the trypsintype enzymes the hydrogen bonds are contributed by two main chain NH groups, while in the subtilisin and its homologues one of the two hydrogen bonds originates from the side chain amide of an asparagine residue (cf. 28,29). In prolyl oligopeptidase one hydrogen bond is formed between the oxyanion and the main chain NH group of Asn 555 , adjacent to the catalytic Ser 554 . The other bond comes from the OH group of Tyr 473 (20). We have recently eliminated the tyrosine OH group by replacing the Tyr 473 by a phenylalanine (24). The enzyme variant displayed decreased catalytic activity; the degree of reduction depended on the nature of the substrate. Interestingly, the tyrosine OH group was implicated in utilizing a portion of the binding energy in the chemical reaction step. However, the mechanistic importance of the electrophilic assistance by the oxyanion binding site has not yet been established in sufficient detail. In this work we have demonstrated that, in addition to the 6 The reaction of suc-Gly-Pro-Nan (Bachem, Ltd.) was monitored at 410 nm using a Cary lE spectrophotometer. For calculation of the liberation of 4-nitroanilide a molar extinction coefficient of 8800 was used (30).
Kinetics -The specificity rate constants (k cat /K m ) were determined under first-order conditions; i. e. at substrate concentrations lower than K m . The first-order rate constant, calculated by nonlinear regression analysis, was divided by the total enzyme concentration to provide k cat /K m . The pH dependence of catalysis was measured in a four-component buffer, which consisted of 25 mM glycine, 25 mM acetic acid, 25 mM Mes, 75 mM Tris and contained 1 mM EDTA and 1 mM DTE (standard buffer). The buffer was titrated to the desired pH with HCl or NaOH, while the ionic strength remained fairly constant over a wide pH range. Small changes in the conductivity were adjusted by the addition of NaCl. After the reaction had been completed the pH of each sample was practically identical to the starting value.
Theoretical curves for bell-shaped pH-rate profiles were calculated by nonlinear regression analysis, using Equation 1 and the GraFit software (31). In Equation 1 k stands for k cat /K m , and (limit) refers to the pH independent maximum rate constant. K 1 and K 2 are the dissociation constants of a catalytically competent base and acid, respectively. The pH-rate profiles composed of two bell-shaped curves were fitted to Equation 2 where k 1 and k 2 gave the limiting values of the rate constant for the low-pH and high-pH forms of the enzyme. The curve for suc-Gly-Pro-Nan is described by Equation 3 comprising one sigmoid and one bellshaped term. A simple sigmoid curve is represented by Equation 4, where the pK a reflects the ionization of a general acid. k = k (limit)[l/(l + 10 pK 1 -pH + 10 pH-pK 2 )] (Eq. 1) k = k 1 [l/(l + 10 pK 1 -pH + 10 pH-pK 2 )] + k 2 [l/(l + 10 pK 2 -pH + 10 pH-pK 3 )] (Eq. 2) k = k l /(l + 10 pH-pK a ) (Eq. 4) The dissociation constant of the enzyme-inhibitor complex (K i ) was calculated from Equation 5, where k 1 and k 0 are pseudo-first-order rate constants determined at substrate concentrations at least 10-fold less than K m with and without inhibitor (I), respectively.
Rate-limiting general acid/base catalyses were tested by measuring kinetic deuterium isotope effects in heavy water (99.9%). The deuterium oxide content of the reaction mixture was at least 95%. The p 2 H of the deuterium oxide solutions was obtained from pH meter readings according to the relationship p 2 H = pH(meter reading) + 0.4 (32).
Thermodynamic parameters -The temperature dependence of the hydrolysis of Z-Gly- promoted by a histidine residue, which exhibits a pK a , of about 7 and operates as a general base/acid catalyst (cf. 28,29). The ionization of this residue governs the pH dependence of catalysis, which conforms to a simple dissociation curve with subtilisin, and a bell-shaped curve with chymotrypsin and its homologues, since in these enzymes an additional acidic group modifies the pH-rate profiles through a conformational change at the active site.
Compared with the classic serine peptidases, the reaction of porcine muscle prolyl oligopeptidase with Z-Gly-Pro-Nap has already been shown to exhibit a more complex pH-by guest on March 24, 2020 http://www.jbc.org/ Downloaded from dependence curve, which is composed of two bell-shaped terms (23,25,36). In contrast to the wild-type enzyme, the Y473F variant displays a normal bell-shaped curve for the reactions with both the classical substrate Z-Gly-Pro-Nap (Fig. 1A) and an octapeptide (Fig. 1B), Abz-Gly-Phe-Gly-Pro-Phe-Gly-Phe(NO 2 )-Ala-NH 2 , possessing the genuine peptide bond, Pro-Phe, which is stronger than the Pro-Nap bond. The differences may modify the hydrolytic mechanism. The doubly bell-shaped curve for the wild type enzyme is also shown in Fig. 1B.
Since the reaction of the wild type prolyl oligopeptidase with Z-Gly-Pro-Nap was found to be rather sensitive to the ionic strength (23,25), the rate constants for the enzyme variant were tested at 0.5 M NaCl concentration, too (Table II). Interestingly, the salt effects, which caused about a 2-fold increase in k cat /K m with the wild-type enzyme, were significantly reduced with the Y473F variant. It can be seen from Table II that the effects of ionic strength are less important with the Y473F variant than with the wild-type enzyme. The rate of hydrolysis of Z-Gly-Pro-Nap diminishes to a lesser extent (17-32-fold, respectively, without and with 0.5 M NaCl) than that of the octapeptide (53-81-fold, respectively, without and with 0.5 M NaCl). This indicates that the contribution of the oxyanion binding site is dependent on the nature of the substrate. The pH dependence curves in Fig. 1 show that in the absence of the Tyr473 OH group, the activity of the low-pH form virtually vanishes, which implies the electrophilic catalysis by the Tyr473 OH group is not uniformly operative over the total pH range. The reason for this effect and the nature of the group that perturbs the simple pH-rate profile is not clear. Fig. 1 and Table II here The rate-limiting step is not general base/acid catalysis -The hydrolysis of Z-Gly-Pro-Nap was also conducted in 2 H 2 O (Fig. 1A). It is known that general base/acid catalyzed reactions proceed slower in heavy water by a factor of 2-3 as found with chymotrypsin and subtilisin (cf. 28,29). However, such a high kinetic isotope effect is not associated with the prolyl oligopeptidase reaction, indicating that the rate-limiting catalytic step is not the chemical reaction catalyzed by the active site histidine residue, but rather a conformational change not affected by the medium (3,23). The pH dependence for k cat /K m for the reaction of the Y473F variant with Z-Gly-Pro-Nap is shown in Fig. 1A The use of thiono substrates having a sulfur atom in place of the carbonyl oxygen of the P1 residue has been rewarding for the studies on the oxyanion binding site of serine peptidases (37). Compared with the corresponding oxo substrates, the rate constants for the thiono substrates of chymotrypsin and subtilisin are lower by more than four orders of magnitude, which is about the detection limit (37). Although to a somewhat lesser extent, similar changes have also been observed with prolyl oligopeptidase, using the Z-Gly-Pro t -Nap thiono substrate (38). The low rate may not be attributed to a reduced chemical reactivity, because the reactivities of the oxo and the corresponding thiono compounds are comparable.

Please insert
The effects of steric hindrance and transition state stabilization with a sulfur atom at the oxyanion binding site are difficult to separate. It appears that the oxyanion hole is designed for the precise accommodation of an oxygen atom, and fails to accept the larger sulfur. The steric hindrance to the sulfur atom may be alleviated by removal of the OH group of Tyr 473 .
Indeed, Table III shows that in the case of the wild-type enzyme there is a dramatic reduction in the rate constant for the thiono substrate (2346-fold), whereas with the Y473F variant the decrease is substantially less (2.7-fold). Elimination of the steric hindrance in the Y473F variant increases the catalytic efficacy for the thiono substrate about 50-fold. However, the rate constant of 80 M -1 s -1 is still less by a factor of 45 than that for the wild-type enzyme with the oxo substrate, and this can account for the electrophilic catalysis. These results underline the importance of both the precise stereochemistry and the electrophilic catalysis at the oxyanion binding site.
Please insert Table III here Estimation of the pK a of Tyr 473 -We have recently shown that Z-Gly-Pro-OH is an inhibitor of prolyl oligopeptidase; the carbonyl oxygen of the free carboxyl group binds to the oxyanion binding site, while the other oxygen forms a hydrogen bond with the catalytic imidazole of His 680 (36). It can be expected that the K i for Z-Gly-Pro-OH will increase with the ionization of Tyr 473 at high pH, due to the repulsion by the developing charge on the Tyr 473 oxygen atom. Indeed Fig. 2 shows that above pH 8 the K i strongly increases, indicating weakening of binding. The experimental points perfectly fit to a dissociation curve with a pK a of 10.21±0.05 that was calculated by extrapolation because the enzyme tends to denature close to pH 10. This pK a value is consistent with the ionization of a normal tyrosine. The removal of Tyr 473 OH considerably enhances the K i . A simple pH dependence curve cannot be fitted to the points obtained with the Y473F variant. The weaker binding to the modified enzyme underlines the importance of the oxyanion binding site, and indicates that the change in the K i of the wild-type enzyme is predominantly coupled with the ionization of Tyr 473 rather than with conformational changes at high pH.

Please insert Figs. 2 and 3 here
Electrostatic effects on the binding of suc-Gly-Pro-Nan -The charged suc-Gly-Pro-Nan has been a widely used substrate of prolyl oligopeptidase because of its better solubility compared with the corresponding neutral Z-derivative. The pH-activity profile for suc-Gly-Pro-Nan ( Fig. 3) is very different from the doubly bell-shaped character for the related Z-Gly-Pro-Nap substrate. Thus, k cat /K m is lower by two orders of magnitude, and the succinyl derivative reacts in the absence of NaCl faster at low pH than at high pH. Addition of 0.5 M NaCl remarkably modifies the curve, which then assumes a bell-shaped character (Fig. 3).
The difference between the reactivities of Z-Gly-Pro-Nap and suc-Gly-Pro-Nan can obviously be attributed to the disparate N-terminal acylating groups. Crystal structures of the S554A variant of the enzyme complexed with the acyl products of the above two substrates revealed that the succinyl and the benzyloxycarbonyl groups occupied similar positions. (Fig.   4, and [36]). This finding hardly explains why the succinyl derivative is so much less effective. Apparently, the binding is less efficient because the K s value increases with the succinyl substrate, but the k cat remains virtually unchanged (24). molecules. In the present case, however, inhibition and catalysis take place within the same molecule, so that the enzyme performs a part time job. When compared with the Z-derivative, the formation of the enzyme-substrate complex (k 1 ) decreases (see Table V and the discussion in the next section) and its dissociation (k -1 ) probably increases, with the consequent Please insert Fig. 4 here With Z-Gly-Pro-Nap, as well as with the octapeptide, the wild-type enzyme displays a doubly bell-shaped pH-rate profile, whereas the Y473F variant exhibits bell-shaped curves ( Fig. 1). On the other hand, with suc-Gly-Pro-Nan the two enzymes show similar pH dependencies. For simplicity, only the curves for the wild type enzyme are displayed in Fig 3, but the parameters for both enzymes are shown in Table IV Please insert Fig. 5 here A low temperature optimum, such as found with the Y473F variant (Fig. 5A), is uncommon among enzyme reactions. A similar temperature effect was observed with thrombin and the phenomenon was interpreted in terms of the change in the individual rate constants that compose k cat /K m as defined by Equations 8 and 9 (39,40).
where k 1 and k -1 are the rate constants for binding and dissociation of the substrate, k 2 is the first-order acylation rate constant, EA is the acyl enzyme, and α = k 2 /k -1 measures the stickiness of the substrate (41), which indicates that the substrate dissociates more slowly from its complex formed with the enzyme than it reacts to yield product, i.e. stickiness is high if k -1 < k 2 . It follows from Equation 9 that k cat /K m approximates k 1 whenever α >> 1. The temperature dependence of the rate constants can be obtained from Equation 10 (39).  Tables V-VII:  Please insert Tables V-VII here The parameters for the abnormal temperature dependence (Fig. 5B) are shown in Table V.
It can be seen that the ratio of k 2 /k -1 decreases with the increase in temperature, indicating a great rate enhancement for the dissociation of the enzyme-substrate complex compared to its conversion into acyl-enzyme. At low temperature k 2 >> k -1 , and k cat /K m becomes equal to k 1 .
Because the substrate dissociation has high activation energy, k -1 becomes predominant at It is worthy of note that E 1 values for the wild type enzyme reactions having greater k 1 values are larger than E 1 for the slower Y473F variant (Tables V-VII). This indicates that considerable entropy effects should be associated with the reaction of the native enzyme (cf. next section).
The temperature dependence of the suc-Gly-Pro-Nan reaction was estimated in the absence of added NaCl at pH 6.3, close to the pH optimum (Fig 3). At this pH the temperature dependence was more linear for the Y473F variant. Since at pH 6.3 the enzyme was less stable, the temperature dependence was not probed above 38 °C. The Z-Gly-Pro-Nap reaction was also examined at this pH. It was found that the Arrhenius plots for both the wild type and the Tyr 473 variant displayed better linearity at pH 6.3 than at pH 8.0. In fact, the temperature optimum of the modified enzyme is significantly shifted towards higher values (not shown).
This made the estimation of the parameters for the reaction of the suc-Gly-Pro-Nan with the Y473F variant very uncertain, at least at low temperature. The large error of k 2 /k -1 is also seen in the reaction of the wild type enzyme with Z-Gly-Pro-Nap (Table V), where the maximum is not explicit enough.  Please insert Fig. 6 and Table VIII here The Eyring plots for the reactions of Z-Gly-Pro-Nap with the wild type enzyme and its Y473F variant are shown in Figure 6, and the thermodynamic parameters calculated from the plots are compiled in Table VIII, along with the corresponding parameters of the octapeptide and suc-Gly-Pro-Nan. It is conspicuous that the ∆S* gives a positive value for the wild type enzyme in all cases, while it is negative or exhibits a low positive value for the Y473F variant, as it is common in most enzyme reactions. This is because the transition state becomes more ordered by freezing considerable translational and rotational degrees of freedom of motion of the reactants. The consequent entropy loss is apparently overcompensated in the prolyl oligopeptidase catalysis by the release of the ordered water molecules located around the active site, in particular at the oxyanion binding site. This is consistent with the remarkably positive entropy and the high enthalpy (Table VIII); the latter is required for breaking the hydrogen bonds of the water shell, in addition to breaking the covalent peptide bond.

Estimation of thermodynamic parameters from nonlinear Eyring plots
A water molecule in the oxyanion hole of the free wild-type enzyme is extensively hydrogen-bonded to the Tyr 473 Oη, Ser 554 Oγ, main chain NH group of Asn 555 and other water molecules that make an extensive hydrogen bonding network (see supplementary material).
These, together with other water molecules must be removed upon substrate binding, which enhances both ∆H* and ∆S*. It could be anticipated that the water shell in the Y473F variant is less stable and hence more easily removed, as indicated by the strikingly reduced ∆S*. hydrogen bonds (20). The latter atom is also hydrogen bonded to a water molecule, which is too far away from the oxyanion to form an interaction (Fig. 7A). When the same complex is formed with the Y473F variant, a water molecule moves closer and partly compensates for loss of the Tyr 473 OH group by providing a hydrogen bond to the oxyanion (Fig. 7B). This water molecule is enthalpically and entropically less efficient for stabilization of the oxyanion. Being less acidic, it is a poorer proton donor than the tyrosine OH group. Also, it is more mobile, which is consistent with the significant difference in entropy between the two forms of the enzyme. Tyr 473 displayed a normal ionization and did not affect the rate-limiting step. Elimination of the Tyr 473 OH group from the oxyanion binding site restricted the catalytic activity mainly to the high-pH form of prolyl oligopeptidase, when reacted with Z-Gly-Pro-Nap or an octapeptide. This indicated that the hydrogen bond formation between the tyrosine OH and the oxyanion facilitated the catalysis to a higher degree at lower pH. On the other hand, the pH-rate profile did not significantly change with the succinyl substrate having a lower pH optimum, except that its specificity rate constant more considerably diminished. This was very likely due to unfavorable electrostatic interactions of the succinyl group with Arg 643 . The implication of Arg 252 , another candidate for interacting electrostatically with the succinyl group, was discounted using the R252S variant. It should also be considered that the benzyloxycarbonyl group binds stronger in the hydrophobic environment than the succinyl group can do.
An important corollary of this work is that major kinetic differences are not immediately apparent from the study of crystallographic structures. Thus, the large entropy differences between the reactions of the wild type enzyme and its Y473F variant are not readily seen from the water structure around its transition state analog complexes. Furthermore, the similar binding mode of the charged and neutral substrates (suc-Gly-Pro-Nan and Z-Gly-Pro-Nap) also cannot account for the large difference in the specificity rate constants.
The study with a thiono substrate indicates that the oxyanion binding site is a very    Table V.