Use of directed mutagenesis to probe the role of tyrosine 198 in the catalytic mechanism of carboxypeptidase A.

Derivatization of Tyr198 in carboxypeptidase A (CPA) results in lowered catalytic activity toward peptide substrates (Cueni, L., and Riordan, J.F. (1978) Biochemistry 17, 1834-1842). We have synthesized via directed mutagenesis a rat CPA variant [Phe198] CPA containing a Tyr198-to-Phe substitution in order to test whether the phenolic hydroxyl plays a critical role in catalysis. A double mutant [Phe193, Phe248]CPA in which both Tyr198 and Tyr248 have been replaced by phenylalanine has also been engineered. Enzymatic characterization of [Phe198]CPA indicates that the Tyr198 hydroxyl is not obligatory for the hydrolysis of peptide and ester substrates. Furthermore, parallel studies with [Phe198, Phe248]CPA show that simultaneous removal of both the Tyr198 and Tyr248 hydroxyls does not abolish catalytic activity. Analysis of the acetylated derivatives of [Phe198]CPA, [Phe248]CPA, and [Phe198, Phe248]CPA establishes that Tyr198 and Tyr248 are the active site tyrosines which are modified by N-acetylimidazole. In addition, the perturbations of enzymatic activity which accompany acetylation of native CPA can be largely assigned to derivatization of Tyr248. The changes in the kinetic constants of substrate hydrolysis due to the Tyr198-to-Phe substitution are manifested as small decreases in the kcat values, but the Km values are essentially unaffected. This exclusive effect on the kcat values suggests that the Tyr198 hydroxyl participates in catalysis by stabilizing the rate-determining transition-state complex.

events which occur during catalysis are not fully understood.
We are attempting to clarify the roles of putative functional residues in CPA by directed mutagenesis of the rat CPA cDNA (Quint0 et al., 1982). A CPA variant in which Ty?@ was replaced by phenylalanine [Phe2@]CPA was recently synthesized (Gardell et al., 1985) in order to analyze the possible role of this residue as a general acid catalyst, particularly in the hydrolysis of peptide substrates. These studies demonstrated that the Tyr"' hydroxyl is not obligatory for substrate hydrolysis although it plays a significant role in ligand binding. In the present investigation we test the functional significance of the Tyr"' hydroxyl by analyzing the effects on catalytic activity due to a Tyrlg8-to-Phe replacement; a double mutant in which both Tyr'" and T y P have been changed to phenylalanine has also been characterized. X-ray diffraction analysis of bovine CPA indicates that Tyr"' is situated in the active site, but it is too far from the substrate's scissile bond to be implicated directly in the cleavage reaction (Quiocho and Lipscomb, 1971). Model building (Quiocho and Lipscomb, 1971) and the structure of bovine CPA complexed with the potato carboxypeptidase inhibitor (Rees and Lipscomb, 1982) show that Tyrl* occupies a position in the secondary recognition site involved in the binding of extended substrates. Chemical modification studies suggest that this residue nevertheless plays a significant role in catalysis. Nitration of Tyr'" has no effect on the esterase activity but decreases the peptidase activity by 10-fold in standard assays (Cueni and Riordan, 1978). In addition, Tyrlg8 apparently exhibits marked reactivity toward the acetylating reagents, N-acetylimidazole (Simpson et al., 1963) and acetic anhydride . The consequences on activity specifically due to acetylation of T y P could not be established due to the accompanying derivatization of Ty13*' in these experiments.
We have previously shown that T y P is not required for the hydrolysis of peptide or ester substrates (Gardell et al., 1985;Hilvert et al., 1986). We wished to test whether Tyr'" might fulfill that function, perhaps by a structural rearrangement which is not evident from the crystal structure or via the intercession of water molecules. The selective inhibition of peptidase activity due to nitration of Tyr'" (Cueni and Riordan, 1978) could be consistent with this residue acting as the proton donor in light of the relative stabilities of the leaving groups generated during the turnover of ester and peptide substrates.
In this paper, we characterize the enzymatic activities of the [Phe'*]CPA and [Phe'98,Phe2@]CPA mutant enzymes in order to evaluate the importance of the Tyr"' hydroxyl in the catalytic mechanism of CPA. Furthermore, these CPA variants and [Phe2@]CPA enable the elucidation of the consequences on activity which accompany specific acetylation of either Tyr'" or Ty?' and thereby provide an additional avenue for illuminating the events which occur during catalysis.

EXPERIMENTAL PROCEDURES
Materials-Cbz-Gly-Gly-Phe and the sodium salt of Bz-Gly-OPhe were purchased from Sigma and used without further purification. ClCPL was prepared according to the published procedure (Suh and   (Gardell et al., 1985). Yeast strain BJ1994 (MATa, leu2, trpl, prbl-1122, pep4-3) was transformed (It0 et d., 1983) with either pYaCPA-Phel98 or pYaCPA-Phel98Phe248, and leucine prototrophs were selected. Transformants were grown in containing 50 mM sodium succinate (pH 5.8) and 0.1 mM ZnClz. synthetic complete medium lacking leucine (Sherman et al., 1981), Large-scale cultures (6 X 16 liters) were grown to saturation ( A m -4.4) at room temperature in carboys which were magnetically stirred. Conditioned media were collected with the Pellicon Cassette System (Millipore) using the Durapore coarse cassette. Concentration by ultrafiltration and buffer equilibration in 20 mM Tris-HC1, 20 mM NaCl (pH 8.0) was also achieved with the Pellicon system using the polysulfone (PTGC) cassette. The retentate was clarified by centrifugation and applied to a diethylaminoethyl cellulose (Whatman) column (6 cm X 4.9 cmz bed volume) equilibrated in 20 mM Tris-HC1, 20 mM NaCl (pH 8.0). Protein was eluted with a linear NaCl gradient (20-250 mM). Fractions were screened for carboxypeptidase activity by treating an aliquot (10 pl) with trypsin (100 ng), incubating for 15 min at 37 "C, and assaying with the ester substrate Bz-Gly-OPhe. Fractions containing activity were pooled and dialyzed exhaustively against 20 mM MES, 0.1 M NaCl (pH 6.0) at 4 'C. Preparative zymogen activation was carried out by the addition of trypsin (-1 mg) and subsequent incubation at 37 "C. When maximum CPA activity was achieved, phenylmethylsulfonyl fluoride was added to a concentration of 0.5 mM, and the solution was applied to an affinity column containing immobilized glycyl-L-tyrosyl-azo-benzylsuccinate (Cueni et al., 1980) (Pierce). Protein was eluted with 20 mM Tris-HCl, 0.2 M NaCl (pH 8.0) and monitored for Bz-Gly-OPhe hydrolyzing activity. The final purification step for [Phe198]CPA or [Phelm,PheUS]CPA involved affinity chromatography using the carboxypeptidase inhibitor isolated from potatoes (Hass and Ryan, 1981) (Sigma) which was immobilized on CNBr-activated Sepharose 4B (Pharmacia). The CPA variants were eluted with 0.1 M Na2CO3, 0.5 M NaCl (pH 11.4) and dialyzed against 20 mM Tris-HC1,0.5 M NaCl (pH 8.0). Proteins were homogeneous as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis; [Phe'"]CPA and [Phe'%,PheUS]CPA comigrated during sodium dodecyl sulfate-polyacrylamlde gel electrophoresis with CPA isolated from rat pancreas as well as CPA-WT and [PheUB]CPA. Enzyme concentrations were determined spectrophotometrically using the value of 6.42 X lo' M" cm" as the extinction coefficient at 278 nm (Simpson et al., 1963).
Acetylation of CPA-WT and its variants was carried out by incubating the enzymes (1 mg/ml; 300 pg) with 5 mM N-acetylimidazole for 60 min at room temperature in 20 mM HEPES, 1 M NaCl (pH 7.5). Excess reagent was removed by dialysis against two changes of buffer at 4 "C. Protein concentration was estimated by the absorbance at 280 nm (c = 5.92 X lo' M" cm") (Whitaker et al., 1966). The extent of derivatization of tyrosyl residues was determined by adding 20 mM hydroxylamine HCl to the acetylated protein and monitoring the increase in Az7s (Simpson et al., 1963). A Ac = 1210 M" cm" at 278 nm for the deacetylation of 0-acetyltyrosine (Myers and Glazer, 1971) was used for this analysis.
The rate of hydrolysis of each substrate was measured as described elsewhere (Hilvert et al., 1986). The experimental data for the pH dependence of kat and kJK, for the hydrolysis of Cbz-Gly-Gly-Phe were fitted to Equations 1 and 2, respectively, by an iterative curvefitting computer program (Yamaoka et al., 1981).
The pH dependence of k.JKm for the hydrolysis of ClCPL was also fitted to Equation 2. The acidic limb of the Lt uersus pH profile for the hydrolysis of ClCPL was fitted to Equation 3.
The buffer for the solvent kinetic isotope effect study was prepared by lyophilizing a known volume of 50 mM Tris-HC1, 0.5 M NaCl (pH 7.5), adding DzO to the original volume, and titrating to pD 8.0 with DCl (pD = meter reading + 0.40 (Glasoe and Long, 1960); pD 8.0 is equivalent to pH 7.5 in terms of the ionization of oxygen and nitrogen bases (Schowen, 1977)). DzO solutions were stored under argon.

Directed Mutagenesis and Heterologous
Expression in Yeast-Conversion of the Tyr'% codon (TAT) to phenylalanine (TTT) in the cDNAs for CPA-WT and [PheZa]CPA was accomplished by oligonucleotide-directed mutagenesis (Zoller and Smith, 1984;Craik, 1985). To each template was annealed an oligonucleotide, 5'AGAGCAGCAGCTGGGAwGCTG-TGGATG3', which introduces an A+T transversion in the Tyr'% codon, and in addition, a "silent" C-G transversion in order to facilitate screening for the mutant template. The cDNAs coding for [PhelN]CPA and [Phe'98,Phe2a]CPA were sequenced in their entirety by the dideoxy termination method (Sanger et al., 1977) to confirm that only the desired nucleotide changes had been introduced. Expression of these CPA variants was carried out in yeast via the CY factor system (Brake et al., 1984) as described previously for CPA-WT and [PheZa]CPA (Gardell et al., 1985). This system directs the synthesis and secretion of the proenzyme forms of the various carboxypeptidases. [ (Hilvert et al., 1986). In contrast, the K,,, values of Cbz-Gly-Gly-Phe and ClCPL for [Phe'98,Phe2a]CPA are elevated 6-and 10-fold, respectively; similar increases in the K , values of these substrates were demonstrated for [PheZa]CPA (Hilvert et al., 1986).
A deuterium oxide kinetic solvent isotope effect is observed for the hydrolysis of ClCPL by CPA-WT. The 2-fold decreases in kcat and kJKm due to the change in solvent from H,O (pH 7.5) to DzO (pD 8.0) are identical to those exhibited by the bovine enzyme for this ester (Kaiser and Kaiser, 1969 The pH dependencies of kc, and kcat/Km for the hydrolysis of Cbz-Gly-Gly-Phe by [Phe'%]CPA ( Fig. 1) are similar to those previously demonstrated for CPA-WT (Hilvert et al., 1986). The various ionization constants are summarized in Table 11. The kat uersw pH profile for ClCPL hydrolysis by [Phe'g8]CPA (Fig. 2 A ) does not display the steeply ascending basic limb which is characteristic of CPA-WT (Hilvert et al., 1986). The suppression of the basic limb of the kcat profile  al., 1986). The pKm+, pKmZ, and pKm values for the hydrolysis of ClCPL are only slightly shifted as a result of the T~r"~-to-Phe replacement (Fig. 2, Table 11). The Consequences of Acetylation of T y F or T~r~~~-T h e number of tyrosines which are derivatized following treatment of CPA with N-acetylimidazole can be determined by treating the modified enzyme with hydroxylamine and monitoring the deacetylation of 0-acetyltyrosine at 278 nm (Simpson et al., 1963). Previous studies on bovine CPA demonstrated that four tyrosines are modified by N-acetylimidazole (Simpson et al., 1963); protection experiments with p-phenylpropionate indicate that two of these are situated in the active site. The rate of deacetylation of the modified active-site tyrosines is faster than the other two; hence a biphasic deacetylation profile is observed. The rapid increase in A278 during treatment of AcCPA-WT with hydroxylamine ( Fig. 3) corresponds to the deacetylation of two modified tyrosines. The burst phase during deacetylation of either [Phe'g8]AcCPA or [Phe248] AcCPA reveals the presence of a single 0-acetyltyrosine. Treatment of [Phe'98,Phe248]A~CPA with hydroxylamine does not cause a rapid increase in A278. If CPA-WT is treated with N-acetylimidazole in the presence of ,&phenylpropionate, subsequent incubation with hydroxylamine results in a deacetylation profile identical to that exhibited by [Phe'98,Phe248] AcCPA (data not shown). The increases in A278 displayed by AcCPA-WT, [Phe'98]AcCPA, and [Phe248]AcCPA following treatment with hydroxylamine are accompanied by a return to activity characteristic of the unmodified enzymes.
The changes in the kinetic constants of [Phelg8]CPA due to treatment with N-acetylimidazole (Table I) are similar to those observed following acetylation of CPA-WT (Hilvert et al., 1986). For example, [Phe'98]AcCPA exhibits a 3.4-fold decrease in the kcat value and a 4-fold increase in the K, value for Cbz-Gly-Gly-Phe when compared to the corresponding values of [Phelg8]CPA. In contrast, acetylation of [Phe248] CPA resulted in only a small decrease in kcat with no corresponding effect on K, (Hilvert et al., 1986). The ability of the double mutant [Phe198,Phe248]CPA to hydrolyze Cbz-Gly-Gly-Phe was unaffected by treatment with N-acetylimidazole (Table I). The kcat values exhibited by CPA-WT or its variants for the hydrolysis of Bz-Gly-OPhe were essentially unchanged following acetylation. Nevertheless, both AcCPA-WT (Hilvert et al., 1986) (Hilvert et al., 1986) are similarly affected by elevated levels of this substrate. It was previously demonstrated that nitration (Riordan et al., 1967) or acetylation (Whitaker et al., 1966) of bovine CPA shifts the onset of substrate inhibition to higher Bz-Gly-OPhe concentrations. Likewise, acetylation of CPA-WT (Hilvert et al., 1986) or [Phe'98]CPA (Fig. 4.4) by N-acetylimidazole abolishes substrate inhibition at concentrations of Bz-Gly-OPhe up to 2 mM. In contrast, treatment of either [PheZa]CPA (Hilvert et al., 1986) or [Phe'98,Phe248]CPA ( Fig. 4B) with Nacetylimidazole has no effect on substrate inhibition by Bz-Gly-OPhe. Another kinetic anomaly which we investigated was substrate inhibition by elevated levels of Cbz-Gly-Gly-Phe (Auld and Vallee, 1970). The absence of the Tyrlg8 phenolic hydroxyl does not change the onset of this kinetic anomaly (Fig. 5A) from that observed with CPA-WT (Hilvert et al., 1986). Nevertheless, [Phe'98,PheZ48]CPA (Fig. 5B) is similar to [PheZa]CPA (Hilvert et al., 1986) in that substrate inhibition is not exhibited even at 2 mM Cbz-Gly-Gly-Phe. As previously shown for CPA-WT (Hilvert et al., 1986), treatment of [Phelg8]CPA with N-acetylimidazole suppresses the onset of substrate inhibition (Fig. 5A). Similar treatment of [Phe198,Phe248]CPA does not affect the hydrolysis of Cbz-Gly-Gly-Phe (Fig. 5B). DISCUSSION We have used directed mutagenesis of the rat CPA cDNA and heterologous expression in yeast to synthesize a CPA variant containing a Tyrlg8-to-Phe replacement and a double mutant in which both Tyr"' and TyrZa have been substituted by phenylalanine.
[Phelg8]CPA and [Phe'98,Phe248]CPA were purified to homogeneity and characterized enzymatically in order to investigate the possibility suggested by chemical modification studies that TyrIg8 may play an important role in catalysis (Cueni and Riordan, 1978).
The catalytic activity of CPA toward peptide and ester substrates is not abolished by the TyF"to-Phe replacement; thus Tyr'= does not play a crucial catalytic role such as mediating general acid catalysis. This same conclusion was reported previously for Tyr248 (Gardell et al., 1985;Hilvert et al., 1986). The ability of [Phe'98,Phe248]CPA to catalyze substrate hydrolysis eliminates the possibility that proton dona- tion to the leaving group can be mediated independently by either Tyr19' or TyrZ4'.

Directed Mutagenesis
The shape of the kcat/K,,, uersus pH profile for the hydrolysis of the ester substrate, ClCPL, is essentially unchanged in spite of the Tyr'=-to-Phe substitution. Hence, the ionization of Tyr"' does not govern the basic limb of this pH-rate profile as suggested by Makinen and co-workers (Makinen et al., 1985). The pH dependencies of both kcat and k,,/K,,, for the hydrolysis of Cbz-Gly-Gly-Phe are also unaffected by the Tyr"'-to-Phe substitution. In contrast to CPA-WT, the kc, uemw pH profile for the hydrolysis of ClCPL by CPA-Phe'98 does not exhibit a steeply ascending arm at pH values greater than 9. It is unlikely that ionization of Tyrlg8 is responsible of Carboxypeptidase A 579

Parameters for pH dependence of hydrolysis of Cbz-Gly-Gly-Phe and ClCPL by CPA-WT and [Phe's]CPA
Measurements were at 25 "C in aqueous buffer, ionic strength 0.5. The parameters were determined from Eauations 2 and 3 as described in the text. The errors are at the 99% confidence level. for the increase in kcat in this pH range since the Tyr248-to-Phe substitution has a similar effect (Hilvert et al., 1986). The free energy changes associated with the binding of the rate-determining transition state in the hydrolysis of Cbz-Gly-Gly-Phe suggest that there is no coupling between the Tyrlgs and Tyr2@ mutations (Fig. 6). Specifically, the TyP@ hydroxyl of CPA-WT or [Phelg8]CPA each contribute 1.6 kcal/mol to binding of the rate-determining transition state. Similarly, the Tyrlg8 hydroxyl contributes 0.4 kcal/mol to transition-state binding regardless of whether tyrosine or phenylalanine is in position 248. In addition, the decrease in the binding free enzyme exhibited by the double mutant, 2.0 kcal/mol, is the algebraic sum of the changes of the two single mutants. This comparison suggests that the replacement of either Tyr"' or TyP@ with phenylalanine does not result in an extensive perturbation of enzyme structure.

Cbz-Gly-Gly-Phe
The present study establishes that TyrIg8 and TyP4' exhibit a marked reactivity toward acetylation and that the effects on substrate hydrolysis due to acetylation of CPA-WT are largely a result of TyrZ4' modification. The ability of the [Tyr248,Phe'98]A~ derivative to hydrolyze peptide and ester substrates supports our previous conclusion that the Ty?@ hydroxyl is not required for catalysis (Gardell et al., 1985;Hilvert et al., 1986). Acetylation of Tyr2@ or removal of the TyrZ4' hydroxyl (Gardell et al., 1985;Hilvert et al., 1986) results in similar changes in the kcat and K, values for Cbz-Gly-Gly-Phe. The comparable effect on peptidase activity due to these different T~l . 2~~ alterations suggests that the hydroxyl of Tyr248 may be involved in hydrogen-bonding interactions that contribute to the hydrolysis of peptide substrates. In contrast, the TyrZ4'-to-Phe replacement and acetylation of TyP4' have dissimilar effects on the hydrolysis of the ester substrate, Bz-Gly-OPhe. The K,,, value for Bz-Gly-OPhe is unaffected by removal of the Tyr2@ hydroxyl but increases approximately 4-fold by acetylation.
The effect of acetylation of CPA-WT on the suppression of substrate inhibition by Bz-Gly-OPhe and Cbz-Gly-Gly-Phe is also attributed to the derivatization of TyP@ by parallel studies with [Phelg8]CPA and [PheZ4']CPA. This conclusion is consistent with previous studies on bovine CPA which showed that TyP@-specific chemical modifications displace the onset of substrate inhibition by Bz-Gly-OPhe to higher substrate concentrations (Riordan et al., 1967;Urdea and Legg, 1979). Removal of either the Tyr"' or the Tyr248 hydroxyl does not affect the kinetic anomaly observed at elevated concentrations of Bz-Gly-OPhe. In contrast, removal of the Tyr2@ hydroxyl mimics the effect of acetylation in suppressing the substrate inhibition observed at elevated concentrations of Cbz-Gly-Gly-Phe. The alternative substrate binding mode of Cbz-Gly-Gly-Phe responsible for this kinetic anomaly thus appears to involve hydrogen-bonding interactions with the phenolic hydroxyl of T y P . Although the T y P hydroxyl is not crucial for catalysis, its removal results in significant decreases in the kcat values of Cbz-Gly-Gly-Phe and ClCPL (Table I). Interestingly, there are no accompanying changes in their K, values. This selective effect on the kc, values is consistent with the conclusions from a previous investigation of the influence of substrate structure on the Michaelis-Menten parameters exhibited by bovine CPA (Abramowitz et al., 1967). It was shown that the K,,, value of a peptide substrate is largely determined by the structure of its COOH-terminal moiety. In contrast, the kcat value is largely determined by the steric bulk of that part of the substrate which interacts with the S, subsite. Model building of extended substrates onto the CPA structure indicates that the side chain of Tyrlg8 constitutes part of the S, subsite (Quiocho and Lipscomb, 1971). Hence the removal of the Tyrlg8 phenolic hydroxyl may alter the complementarity between enzyme and substrate in this critical region. The exclusive effect of the Tyrlg8-to-Phe substitution on FIG. 6. The Tyrlss-to-Phe and Tyrs4'-to-Phe substitutions in CPA are independent mutations. The binding energies (kcal/ mol) of the T y P or Tyl" phenolic hydroxyls with the rate-limiting transition-state structure of Cbz-Gly-Gly-Phe were calculated from the kJK, values (M-' 8-l) of CPA-WT or its variants. the kc,, values of Cbz-Gly-Gly-Phe and ClCPL indicates that the binding energy derived from the Tyrl= hydroxyl may be utilized to increase the catalytic rate. The use of binding energy to facilitate catalysis can, in principle, be mediated by several mechanisms (Pauling, 1946;Jencks, 1975;Fersht, 1985). For example, the T y P hydroxyl would increase kc, by preferentially binding to the rate-determining transition state structures of these substrates, thereby decreasing the free energy difference between the enzyme-substrate complex and the transition-state complex. The rate-determining transition state for the hydrolysis of ClCPL appears to be the same for both CPA-WT and [Phe'=]CPA since they exhibit similar kinetic solvent isotope effects. Previous workers had proposed that the rate-determining step for the turnover of ClCPL was the hydrolytic breakdown of a mixed-anhydride intermediate (Makinen et al., 1976). The free energy contributed by the TyrlSS hydroxyl to transition-state binding, 0.8 kcal/mol, may be manifested as stabilization of the transition state which results during the hydrolysis of the putative acylenzyme intermediate. Interestingly, TyrlS8 was recently implicated on the basis of magnetic resonance and molecular modeling techniques to be a determinant of the strain which is imposed upon a spin-label substrate, TEPOPL, following its interaction with the extended binding cleft of CPA (Kuo et al., 1983).
The present study shows that the Tyr'= phenolic hydroxyl is not obligatory for the hydrolysis of peptide or ester substrates. Simultaneous removal of the Tyr'= and TyrZ4' phenolic hydroxyls also does not abolish catalytic activity. Nevertheless, in spite of the apparent distance of Tyrlg8 from the site of bond scission, it appears to play a role in catalysis as measured by the specific effect on Lt due to the Tyr'=-to-Phe substitution. The Tyr'" hydroxyl may facilitate substrate hydrolysis by participating in the stabilization of the ratedetermining transition state. In addition, this study illustrates the utility of directed mutagenesis to clarify the effects of chemical modification and hence to increase the effectiveness of chemical modification as a probe of protein structure and function.