The noncatalytic beta-propeller domain of prolyl oligopeptidase enhances the catalytic capability of the peptidase domain.

Prolyl oligopeptidase, which is involved in memory disorders, is a member of a new family of serine peptidases. In addition to the peptidase domain, the enzyme contains a beta-propeller, which excludes large peptides from the active site. The enzyme is inhibited with thiol reagents, possibly by reacting with Cys-255 located close to the substrate binding site. This assumption was tested with the Cys-255 --> Thr, Cys-255 --> Ala, and Cys-255 --> Ser variants of prolyl oligopeptidase. In contrast to the wild type enzyme, the Cys-255 --> Thr variant was not inhibited with N-ethylmaleimide, indicating that Cys-255, of the 16 free cysteine residues, exclusively accounts for the enzyme inhibition. Unlike the wild type enzyme that showed a doubly bell-shaped pH rate profile, the modified enzyme displayed a single bell-shaped pH dependence with benzyloxycarbonyl-Gly-Pro-naphthylamide. It was the high pH form of the enzyme that virtually disappeared with all three enzyme variants. A substantial reduction was also observed in k(cat)/K(m) for the aminobenzoyl-Ser-Pro-Phe(NO(2))-Ala-OH substrate. The high pK(a) (9.77) of Cys-255 determined by titration with N-ethylmaleimide excluded the possibility that ionization of the thiol group was responsible for generation of the two active enzyme forms. The impaired activity of the enzyme variants could be rationalized in terms of weaker binding, which manifests itself in high K(m) for substrates and high K(i) for inhibitors, like benzyloxycarbonyl-Gly-Pro-OH and aminobenzoyl-Ser-d-Pro-Phe(NO(2))-Ala-OH. It was concluded that, besides selecting substrates by size, the beta-propeller domain containing Cys-255 remarkably contributed to catalysis of the peptidase domain.

Prolyl oligopeptidase is encountered in a variety of organs and species (3). These enzymes, however, differ in their reactivities toward thiol reagents. Thus, p-chloromercuribenzoate prevents catalysis by the porcine enzyme almost completely and instantaneously; N-ethylmaleimide causes about 85% inhibition; and the small iodoacetamide inhibits the activity only 45-50% (14). However, prolyl oligopeptidases isolated from Flavobacterium meningosepticum (15,16) and mushroom (17) are not inhibited with thiol reagents.
We have recently determined the three-dimensional structure of prolyl oligopeptidase (18). The enzyme is composed of two different domains of equal size, one mainly consisting of the C-terminal moiety is a serine peptidase of ␣/␤-hydrolase fold, and the other is a seven-bladed ␤-propeller structure not closed by a "velcro" (see Ref. 18 for explanation). Thanks to this unique feature of the propeller, an opening between the first and last blades may ensure an access to the active site for oligopeptides but not for proteins encountered in the cytosol. The catalytic triad (Ser-554, Asp-641, and His-680), located between the two domains at the base of the large cavity of the propeller domain, displays a different handedness compared with that of the triad of chymotrypsin or subtilisin.
Several differences have been observed between the extensively studied serine peptidases and prolyl oligopeptidase. The former enzymes exhibit a simple pH rate profile, controlled by the catalytic imidazole group with a pK a of 7 as found with subtilisin, whereas the enzymes of the trypsin family have an additional group with a pK a of 9 -10, which is associated with conformational changes (19,20). Also, these enzymes are only slightly, if at all, affected by ionic strength of the medium. In contrast, prolyl oligopeptidase catalysis conforms to a doubly sigmoid pH rate curve and is rather sensitive to ionic strength (3,14). The doubly sigmoid pH rate profile indicates the existence of two pH-dependent enzyme forms, one being predominant at pH 6 and the other at pH 8.
Perusal of the active site region has suggested that Cys-255, which is located close to the S1 and S3 binding subsites, can account for inhibition of the enzyme with bulky thiol-specific reagents. Prolyl oligopeptidase isolated from F. meningosepticum, which is not inhibited with thiol-reacting agents, has a threonine residue in place of Cys-255. Because the implication of other cysteines in inactivating prolyl oligopeptidase cannot be excluded, in particular those located in the N-terminal region (21), in this work the role of Cys-255 is tested by substituting threonine, serine, and alanine for Cys-255. It is expected that kinetic investigations with the enzyme variants will explain as to what extent this residue is responsible for the inactivation of prolyl oligopeptidase by thiol reagents, and, more importantly, whether a thiol group from an attached noncatalytic domain could affect the catalytic activity.

EXPERIMENTAL PROCEDURES
Enzyme Preparation-Prolyl oligopeptidase of porcine brain was expressed in Escherichia coli and purified as described (22). The enzyme does not have any post-translational modification, so its expression in E. coli is warranted. In the absence of disulfide bonds, the enzyme is readily folded and produced in a large quantity. Because of the simpler and faster preparation protocol used with the bacterial starting material, which is rather rich in prolyl oligopeptidase, the specific activity of the recombinant enzyme is even higher than the activity obtained directly from pig (22).
In the first step of PCR, two independent reactions were carried out, using the Vent polymerase, the pTrc/PPO plasmid template digested with the KpnI enzyme, and (i) the SD primer and the antisense primer and (ii) the 3Ј-primer and sense primer, each run in 25 cycles. The reaction mixture (100 l) contained 300 M of each component of dNTP, 0.5 g of the primers, and 1 l of 10 mg/ml bovine serum albumin in the standard 1ϫ ThermoPol polymerase buffer. The denaturing, annealing, and polymerization temperatures were 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min (with the shorter 5Ј product) or 2 min (with the longer 3Ј product), respectively.
The mutated gene was then obtained in two pieces, which were isolated by agarose gel electrophoresis and purified with a QIAquick gel extraction kit (Qiagen GmbH, Hilden, Germany). In the second step the mixture of the two portions of the gene were used as templates. The reaction mixture was heated to 95°C, prior to addition of the Vent polymerase. The temperature was then lowered to 50°C when the complementing ends of the two DNA portions formed double stranded DNA. Heating at 72°C for 5 min resulted in the synthesis of the strands. The complete strands were able to bind the terminal primers (SD primer and 3Ј-primer), which were added to the mixture at 95°C. The amplification was then carried out in 29 cycles at 95, 55, and 72°C for 1, 1, and 3 min, respectively. The PCR product was identified on an agarose gel (1%) and purified with the QIAquick gel extraction kit. The pure product was digested with EcoRI and KpnI restriction enzymes, isolated with agarose gel electrophoresis, purified with the gel extraction kit, and cloned into the EcoRI and KpnI sites of vector pTrc 99A. The resulting mixture was transformed into E. coli DH5␣. Some transformant colonies were grown on LB, to A 600 ϭ 0.6, then induced with isopropyl-1-thio-␤-D-galactopyranoside. After overnight incubation, the cells were harvested by centrifugation, sonicated, and centrifuged, and the supernatant was assayed for prolyl oligopeptidase activity with the standard substrate, Z-Gly-Pro-Nap.
Plasmids were isolated from the active clones, and the purified DNA was digested with BamHI restriction endonuclease. Mutant clones had a distinct digestion pattern, because we introduced an extra restriction site into the gene using the mutant oligonucleotides.
Large scale production of the Cys-255 3 Thr, Cys-255 3 Ala, and Cys-255 3 Ser variants of prolyl oligopeptidase was achieved in E. coli JM 105 and purified as described for the wild type enzyme (22).
Kinetics-The reaction of prolyl oligopeptidase with Z-Gly-Pro-Nap (Bachem Ltd., Bubendorf, Switzerland) was measured fluorometrically, using a Jasco FP 777 spectrofluorometer equipped with a thermostated cell holder. The excitation and emission wavelengths were 340 nm (1.5-nm bandwidth) and 410 nm (5-nm bandwidth), respectively. The gain of the photomultiplier was set to "medium." Cells with excitation and emission path lengths of 1.0 and 0.4 cm, respectively, were used. The substrate with internally quenched fluorescence, Abz-Ser-Pro-Phe(NO 2 )-Ala-OH, was prepared with solid-phase synthesis, and its hydrolysis was followed in the same way as Z-Gly-Pro-Nap, except that the excitation and emission wavelengths were 337 and 420 nm, respectively.
Pseudo-first-order rate constants were measured at substrate concentrations lower than K m and were calculated by nonlinear regression data analysis, using the GraFit software (23). The specificity rate constants (k cat /K m ) were obtained by dividing the first-order rate constant by the total enzyme concentration present in the reaction mixture.
The Michaelis-Menten parameters, k cat and K m , were determined with initial rate measurements, using substrate concentrations in the range of 0.2-5 K m . Because of the poor solubility of Z-Gly-Pro-Nap, in particular in the presence of 0.5 M NaCl, the reactions were carried out in the presence of 0.3% acetonitrile, which caused about 20% inhibition of prolyl oligopeptidase. The kinetic parameters were calculated with nonlinear regression analysis.
Theoretical curves for bell-shaped pH rate profiles were calculated using nonlinear regression analysis, using Eq. 1 and the GraFit software (23). In Eq. 1, k cat /K m (limit) stands for the pH-independent maximum rate constant and K 1 and K 2 are the dissociation constants of a catalytically competent base and acid, respectively. When an additional ionizing group modifies the bell-shaped character of the pH dependence curve, the data were fitted to Eq. 2 (double bell-shaped) or Eq. 3 (sigmoid ϩ bell-shaped) or Eq. 4 (bell-shaped curve with double dissociation at the acidic limb), where the limiting values stand for the pH-independent maximum rate constants for the two active forms of the enzyme and K 1 , K 2 , and K 3 are the dissociation constants of three enzymatic groups whose state of ionization controls the rate constants.
Inhibition of prolyl oligopeptidase with N-ethylmaleimide was measured with 0.14 -0.28 M enzyme and 1.7-6.9 M N-ethylmaleimide in 1.0 ml of reaction mixture containing 25 mM acetic acid, 25 mM Mes, 25 mM glycine, 75 mM Tris, 1 mM EDTA at pH 8.0. Aliquots of 20 -40 l were withdrawn at appropriate times, and the initial rates of the remaining activity were determined with Z-Gly-Pro-Nap (3.4 M, 0.3% acetonitrile) The K i values, the dissociation constants of the enzyme-inhibitor complex, were calculated from Eq. 5 using at least 10 inhibitor concentrations.
where k i and k 0 are pseudo-first-order rate constants determined at substrate concentrations at least 10-fold less than K m in the presence and absence of inhibitor (I), respectively. Fig. 1, N- Fig. 2a illustrates that the wild type enzyme displays a doubly bell-shaped curve, corresponding to two pH-dependent enzyme forms, the high pH form being more active. In contrast, the high pH form of the Cys-255 3 Thr variant is essentially inactive; the activity is restricted to the low pH form. One reason for the loss of activity might be a steric hindrance and/or the less polar environment created by the extra methyl group at the substrate binding site. Therefore, we have examined two other variants, each with a smaller side chain and different polarity, namely Cys-255 3 Ala and Cys-255 3 Ser. At low ionic strength both variants gave single bell-shaped pH dependence curves as found with the Cys-255 3 Thr variant ( Fig. 2a and Table I). This ruled out the implication that the methyl group abolished the activity of the high pH form. It would be interesting to compare the results obtained with the Cys-255 3 Thr variant to those of the native prolyl oligopeptidase from F. meningosepticum, which also has a threonine in place of Cys-255. Unfortunately, a detailed kinetic investigation has not yet been performed with this enzyme. However, it has been reported that the pH optimum of the enzyme from F. meningosepticum is 6.5 with a peptide substrate (24). This suggests that the pH rate profile for the bacterial enzyme is similar to that observed here for the Cys-255 3 Thr variant. There are not any or a very few thiol groups in bacterial proteins. Thus, the F. meningosepticum enzyme contains 2 compared with the 16 cysteines of the mammalian prolyl oligopeptidase. The physiological significance of generation of Cys-255 during evolution is demonstrated by the improved catalytic efficacy.

Cys-255 3 Thr Variant of Prolyl Oligopeptidase Is Not Inhibited with N-Ethylmaleimide-As shown in
The effects of 0.5 M NaCl on the pH rate profiles has also been investigated ( Fig. 2b and Table I). Under these conditions, the two enzyme forms are preserved with the wild type peptidase (not shown in Fig. 2b to avoid compressing the other curve), while the rate constants significantly increased (Table I). With the Cys-255 3 Thr and Cys-255 3 Ala variants, the activity of the high pH form of the enzyme considerably decreased with respect to the low pH form but did not disappear, whereas with the Cys-255 3 Ser variant a simple bell-shaped curve was obtained within the experimental error. In Table I are shown the kinetic parameters extracted from the pH dependence curves. It is apparent that pK 1 is rather low, about 5 in the absence of 0.5 M NaCl, and somewhat higher in the presence of salt. Because only the basic form of the catalytically competent histidine is active, this residue should have a pK a lower than that (ϳ7) found with most serine peptidases. This may be due to the effect of positive charges, such as the nearby guanidinium ion of Arg-643 (18), which can destabilize the protonated histidine. As can be expected, electrostatic shielding by high ionic strength reduces this effect, leading to an enhanced pK 1 value. It is worth noting that in the presence of salt the points at the acidic limb of the pH dependence curve do not precisely fit the theoretical curve. This suggests that at least one additional ionizing group affects the catalysis possibly by causing conformational changes in the protein structure.
We feel bound to emphasize that the pK 1 of ϳ5 may not be the group constant of His-680. It is very likely an apparent molecular constant, which may arise from the increased activity of the protonated enzyme conformation. The pK 2 may also be generated as a consequence of the enhanced activity at low pH, and its value is dependent upon the relative activities of the two enzyme forms, so that it may not be assigned to a definite functional group. The pK 3 is less affected in this way. It is very likely associated with the instability of the protein at high pH. It has previously been shown that the stability of the enzyme decreases around pH 9 as demonstrated by urea denaturation and inactivation at high temperature (25). Importantly, the protein tends to denature at somewhat less alkaline pH in the presence of 0.5 M NaCl, which is consistent with the lower pK 3 for the wild type enzyme at higher ionic strength ( Table I).
The pK a of Cys-255 Is Unusually High-The pK 2 and pK 3 values are not far from the pK a of a normal thiol group (ϳ8.5), and, in a particular protein environment, Cys-255 may adopt either value. Hence, it is possible that the ionization of Cys-255 can contribute to these pK a values. It is known that the reactivity of thiol group depends on its ionization state, where the nondissociated form is virtually inactive. Therefore, we have measured the pH dependence of the rate constants for the reaction of Cys-255 with N-ethylmaleimide in the range of pH 8.0 -9.3 (Fig. 3). Denaturation at higher pH ruled out a correct determination of the rate constants. A pK a of 9.77 Ϯ 0.03 was extracted from the curve. Although this is an extrapolated value, its error is surprisingly small, and this clearly excludes that pK 2 would depend on the ionization of the thiol group of Cys-255. Moreover, it is unlikely that ionization of the thiol group would significantly contribute to a pK 3 of ϳ9. The rather high pK a value of this thiol group is conceivable in light of its immediate environment, which includes the side chains of Phe-476 and Trp-595 at an approximate van der Waals distance from the sulfur atom (18). Indeed, the less polar environment is not favorable for the formation of a thiolate ion, and thus the ionization occurs with an enhanced pK a value.
Several Ionizing Groups Contribute to the Active Site Environment-At or near the active site there are many charged groups, like Arg-643, which participates in substrate binding; Asp-149, which forms a salt bridge with that Arg residue; and Asp-642, Arg-128, and Lys-172, the further constituents of the extended hydrogen bond network. In such a system, the pK a of individual groups cannot be predicted, but alteration in their interactions may elicit a conformational change that is characterized by pK 2 . Structural differences between the two pH-dependent forms have also been detected by intrinsic fluorescence measurements, which clearly indicated that the low pH form is more unfolded (25). Therefore, it appears that the contribution of the nondissociated thiol group of Cys-255 is only essential for the formation of the high pH form of prolyl oligopeptidase. Its absence does not significantly affect the activity of the low pH form (Fig. 2a).
The pH Rate Profile Is Modified with Substrate Variation-To further investigate the distinct effects of Cys-255 on the two active forms of prolyl oligopeptidase, we have also used a fluorogenic tetrapeptide substrate, Abz-Ser-Pro-Phe(NO 2 )-Ala-OH, which is cleaved between the Pro-Phe(NO 2 ) bond. The existence of the two pH-dependent active forms of the wild type enzyme is even more expressively illustrated in this case (Fig.  4a) than with Z-Gly-Pro-Nap. Specifically, the pH rate profile exhibits a minimum between the two forms in the absence of 0.5 M NaCl, rather than a shoulder found with the classic substrate. The acidic limb of the pH dependence curve could not be detected, because the enzyme tends to be denatured at low pH. If the rate increase with lowering pH stems from an ionization event, the shielding of the electrostatic effects at enhanced ionic strength should modify the pH rate profile.  Indeed, the pH rate profile has remarkably changed at low pH in the presence of 0.5 M NaCl so that an approximately bellshaped curve is obtained (Fig. 4a). The acidic limb of the curve, however, is steeper than expected for a single dissociation (broken line) and even for a double dissociation (solid line). It is noteworthy that the salt effect is quite different for the reactions of the classic and the tetrapeptide substrates. In the case of Z-Gly-Pro-Nap the rate constants increase 2.5-to 3.0-fold (Table I), whereas the alteration with Abz-Ser-Pro-Phe(NO 2 )-Ala-OH is rather small (Fig. 4a and Table II). A moderate rate enhancement is only found for the low pH enzyme form.
In the reaction of Cys-255 3 Thr variant with the tetrapeptide, the activity of the high pH form is very much reduced ( Fig.  4b and Table II), as observed with Z-Gly-Pro-Nap (Fig. 2a). The pH rate profiles exhibit doubly bell-shaped characters, independent of the ionic strength (Fig. 4b), but the activity of the low pH form is diminished in the presence of 0.5 M NaCl, which is opposite to that found with the Z-Gly-Pro-Nap substrate. The distinct kinds of behavior of the two substrates suggest that the relative activities of the low and high pH forms are different with the different substrates.
Binding of Substrate Analog Inhibitors to the High pH Enzyme Form Is Impaired in the Cys-255 3 Thr Variant-From the above results the question emerges as to why the activity of the Cys-255 3 Thr variant decreases so dramatically above neutrality. The weaker binding is a reasonable possibility that can be tested by measuring the changes in K m . Although K m is not necessarily a binding constant, it is clearly dependent upon binding, being K m ϭ K s k 3 /(k 2 ϩ k 3 ), where k 2 and k 3 are the first-order acylation and deacylation rate constants, respectively. It is seen in Table III that K m considerably increases with the Cys-255 3 Thr variant at pH 8.0 but only moderately changes at pH 6.0. The increase in K m may, at least in part, be the consequence of impaired binding to the Cys-255 3 Thr variant. The poor solubility of Z-Gly-Pro-Nap precluded the precise determination of K m for the reaction with the Cys-255 3 Thr variant. It was, however, absolutely clear that, in contrast to pH 6, at pH 8 the K m was much higher for the variant than for the wild type enzyme (not shown in Table III).
The above reasoning is strongly supported by measuring true binding constants with peptide inhibitors. One such inhibitor is derived from the tetrapeptide substrate by substituting a D-Pro for Pro at the P1 subsite. The K i values, i.e. the dissociation constants of the enzyme-inhibitor complexes shown in Table  III, indicate that the binding of inhibitor at pH 8.0, similarly to that of the substrates, is stronger to the wild type enzyme than to its Cys-255 3 Thr variant. The effect is similar with the product-like inhibitor, Z-Gly-Pro-OH ( Fig. 5 and Table III), except that the difference in K i is more substantial at pH 8.0. The changes in K i may explain why the high pH form of the Cys-255 variant displays lower k cat /K m than the low pH form. It is unexpected that elimination of the thiol group, although it does not interact with the substrate, still weakens its binding. Hence, the sulfur atom may stabilize the binding site conformation by interacting with nonpolar side chains of its environment. It is noteworthy that the K i value for the peptide containing the D-Pro residue is approximately independent of pH in the case of the wild type enzyme. In contrast, Z-Gly-Pro-OH binds much stronger at pH 6 than at pH 8. Apparently, the carboxylate ion of the P1 proline residue interacts with His-680, and this interaction exhibits greater strength at pH 6, where the imidazole is more extensively protonated than at pH 8. This is supported by the earlier finding that streptogrisin A, a serine peptidase structurally related to chymotrypsin, forms complexes with Ac-Pro-Ala-Pro-Phe-OH and Ac-Pro-Ala-Pro-Tyr-OH, in which the carboxylate ion is bonded to the catalytic histidine but only at low pH (26).
Cys-255 belongs to the fourth blade of the seven-bladed ␤-propeller domain that regulates the flow of substrates to the   active site. This work has provided evidence that, in addition to the filtering effect, the evolutionarily independent propeller domain also offers residues that significantly contribute to the catalytic action of the peptidase domain.