Arginine 127 Stabilizes the Transition State in Carboxypeptidase*

Crystallographic studies suggest that Arg-127 is a key amino acid in the hydrolysis of peptides and esters by carboxypeptidase A. The guanidinium group of Arg- 127 is hypothesized to stabilize the oxyanion of the tetrahedral intermediate formed by the attack of water on the scissile carbonyl bond. We have replaced this amino acid in rat carboxypeptidase lysine methionine in order to define the role of Arg-127 in carboxypeptidase catalyzed hydrolysis. The wild-type and mutant enzymes were expressed in yeast and purified.


Pancreatic
CPAJ is a Zn'+-dependent exopeptidase which cleaves carboxyl-terminal aromatic or aliphatic amino acids from peptides. CPA has been extensively studied by a wide range of techniques that include crystallography, spectroscopy, kinetics, and site-directed mutagenesis (l-3, 8, 11, 19).
The key functional groups positioned near the scissile bond are Zn*+, Glu-270, and Arg-127 ( Fig. 1) (1). Zn*+ is coordinated to the enzyme by residues  ( Fig.  1) (1). A hydrophobic cleft forms the binding pocket of the terminal hydrophobic side chain and provides substrate specificity. The terminal carboxylate forms a bifurcated hydrogen bond to Arg-145 and a single hydrogen bond to Tyr-248 (1). Additional contacts with Arg-71, Tyr-198, and Phe-279 form an extended binding site for larger substrates (1).
The proposed mechanism of CPA hydrolysis is a Zn*+promoted attack of water on the scissile carbonyl bond with Glu-270 assisting as a general base ( Fig. 2) (1). The enzyme must additionally stabilize the oxyanion in the tetrahedral intermediate.
This role has classically been assigned to Zn*+. However, Christianson and Lipscomb (1) argue that interaction of Zn"+ with the scissile carbonyl oxygen would decrease the nucleophilicity of the Zn2+ bound water; therefore, Zn*+ would not be a good candidate for the electrophilic catalyst. Instead, they suggest that the positive charge of Arg-127 may stabilize the oxyanion (4, 5). Crystallographic studies show that Arg-127 forms a hydrogen bond to the tetrahedral oxygen of the transition state analogs Bop (the ketone analog of Bz-Gly-Phe) and Zgp (the phosphonamidate analog of Cbz-Gly-Phe) (Fig. 1). Chemical modification of an unidentified arginine residue abolished peptide hydrolysis without affecting ester hydrolysis (6). Peptides can bind to the apoenzyme, whereas esters cannot (7), and it has been hypothesized that a binding mode involving Arg-127 exists for peptides but not esters. These results suggest that Arg-127 will have a more important function in the hydrolysis of peptides than for esters( 1).
To elucidate the role of Arg-127 in peptide and ester hydrolysis, we have replaced Arg-127 with Lys (R127K), Met (R127M), and Ala (R127A) by site-directed mutagenesis of the rat enzyme. The wild-type and mutant enzymes were expressed in yeast, purified, and evaluated by kinetic methods. These studies reveal an important catalytic role for Arg-127.  Fig. 2). The calculation to obtain the contribution of Arg-127 at any particular site was done as described (21). All modeling and electrostatic calculations were based on the structure of cow CPA because a structure for rat CPA is unavailable.
However, based on the observation that rat and cow CPA have strong sequence similarity (79 % of the amino acid residues are similar (24)) and all catalytically important residues are invariant, we make the assumption that the analysis will be valid for rat CPA as well. This assumption is valid in the comparison of the active site electrostatic potentials for cow and rat trypsin (21).

Mutagenesis, Expression,
and Purification of CPA-Arg-127 was replaced by lysine (R127K), methionine (R127M), and alanine (R127A) as described under "Experimental Procedures." The wild-type and mutant CPA genes were cloned into the ADH/GAPDH yeast expression vector (Fig. 3) and purified as described under "Experimental Procedures." The yield of proCPA is approximately 10 mg/liter of yeast culture. This represents a 25-fold increase in CPA expression over our previous system and will provide sufficient material for structural characterization of CPA and its mutants (11). Unlike the wild-type enzyme, the purified Arg-127 mutants were unstable when stored for prolonged periods (days to weeks) in the absence of ZnCl*; the mutant enzymes were subsequently stored in the presence of 10e4 M ZnC&. CPA purity was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis (data not shown) and by isoelectric focusing. Fig. 4 shows an isoelectric focusing gel of wild-type CPA, R127K, R127M, and R127A. The gel was stained with Coomassie Blue (Fig. 4A) and for CPA activity (Fig. 4B). Wild-type CPA and R127K migrated with the same mobility in the gel (p1 6.1), whereas R127M and R127A both shifted toward the anode (p1 5.6). The band observed by the activity staining co-migrated with the major Coomassie Blue staining protein band in each case and was observed only in lanes containing CPA. Several minor bands were observed in all lanes. These bands may represent deamidation (they have more acidic p1 values) or alternate cleavage sites for propeptide removal and it is not surprising that they also show activity.
The Effect of At-g-127 Mutagenesis on kc,, and K,,-The effect of Arg-127 mutagenesis on CPA catalyzed hydrolysis of peptides and esters varied with the substrate as shown in Table I. The k,,, values for R127K hydrolysis of Bz-Gly-Phe, Bz-Gly-OPhe and Cbz-Gly-Gly-Phe are decreased on average 150-fold, whereas the K,,, values are increased on average lofold. The k,,, values for the hydrolysis of these substrates by R127M or R127A are 1500-and 3000-fold lower than for wildtype CPA, respectively, whereas the & values are increased about lo-fold.
In contrast to the three substrates described above, hydrolysis of Cbz-Gly-Phe and CLCPL by R127K was only mildly affected. The k,,, values were reduced by 3-and 4-fold respectively, whereas the K,,, values were increased 3-fold for both substrates (Table I). For the hydrolysis of Cbz-Gly-Phe and CLCPL by R127M, the k,,, values were reduced by 25-and 50-fold, respectively (Table I).
In contrast, the mutant enzymes have a reversed preference with the ratio of the specificity constants averaging 0.1. These trends are also observed for Cbz-Gly-Gly-Tyr and Cbz-Gly-Tyr (data not shown). Even more surprisingly, the ratio of the specificity constants for the structurally similar substrates, Cbz-Gly-Phe and Bz-Gly-Phe, has gone from 1.4 for the wild-type enzyme to an average of 100 for the mutants. Electrostatic Calculations-The value of the electrostatic potential generated by all charged amino acids in Bop . CPA was calculated by the Delphi program (21-23). Charges were assigned as described under "Experimental Procedures." Fig.  5A displays the +3.6 kcal/mol contours of the electrostatic potential for the Bop. CPA structure and Fig. 5B displays the +3.6 kcal/mol contours for the Bop.CPA structure in which Arg-127 was replaced by Met. The active site has a prominent region of positive potential generated by arginines 145, 127, 71, and 124. An additional region of positive potential parallels the positive active site potential and is generated by Lys-128 and Arg-130. Interestingly, all 6 of these residues are conserved among the pancreatic carboxypeptidases of known sequence (24). The replacement of Arg-127 with Met disrupts the positive potential field within the active site (Fig. 5 A and  B). To calculate the electrostatic potential of Arg-127 alone, each guanidinium nitrogen was assigned a charge of 0.5, and all other charges were set at 0 for the Delphi calculation.
The electrostatic effect of Arg-127 could then be sampled at any position within the structure. The hydrogen atom, derived from the dissociation of water in the catalytic attack, was built into the Bop and Zgp structures halfway between the oxygen of Glu-270 and the O(1) of the inhibitor.
The results are summarized in Table II. The large difference in potential at the Zn*+ atom found between the Bop. CPA complex and the Zgp.CPA complex is due to a difference in the position of the inhibitor and Arg-127. Solvent Isotope Effects-Deuterium solvent isotope effects were measured for the wild-type and mutant enzymes. The results observed for the wild-type CPA hydrolysis of CLCPL (kH/kD = 2.3 f 0.1) are consistent with the previously reported values (8, 25). The results for the wild-type CPA hydrolysis of Cbz-Gly-Phe (kH/kD = 1.2 + 0.1) and Cbz-Gly-Gly-Phe (kH/ko = 1.2 + 0.1) by wild-type CPA are likewise in agreement with the reported values for peptide substrates (25). In addition, we measured an isotope effect for the hydrolysis of Bz-Gly-OPhe, by wild-type CPA, that was equal in magnitude to that reported for CLCPL (kH/kD = 2.5 f 0.1). Deuterium solvent isotope effects similar to those observed for wild-type CPA were observed for the hydrolysis of both ester and peptide substrates by R127K and R127M. These data imply that no change in the rate-limiting step has occurred for the hydrolysis of the ester substrates. No information could be obtained from this study regarding the rate-limiting step for the peptide substrates, because a significant isotope effect was not observed.
The Effect of Arg-127 Mutation on the Binding of Inhibitors--The effects on the binding of three competitive inhibitors of CPA hydrolysis were tested for the Arg-127 mutant enzymes; BzSA and PC1 are ground state inhibitors and Cbz-Phe-Ala(P)-OAla is a transition state analog. BzSA has a KI of 0.2 jtM for cow CPA and is believed to of Bz-Gly-OPhe and Cbz-Gly-Gly-Phe hydrolysis. Cbz-Phe-Ala(P)-OAla, the phosphonate analog to the substrate Cbz-Phe-Ala-OAla, has been shown previously to be a transition state analog of CPA catalyzed hydrolysis (28). The phosphonate inhibitors are the tightest binding class of inhibitors to CPA known; Cbz-Phe-Ala(P)-OAla is a relatively weak binding member of this class with a K, of 56 nM for cow CPA (28). The K, for the inhibition of Cbz-Gly-Phe hydrolysis by wild-type rat CPA is 30 nM. In contrast, the binding affinity of the R127K and R127M for Cbz-Phe-Ala(P)-OAla is greatly reduced; the K, values for the inhibition of Cbz-Gly-Phe hydrolysis by R127K and R127M are 75 and 300 pM, respectively (Fig. 6) Criteria for Mutant Purity-A potential problem when evaluating mutants with such low levels of activity is contamination from wild-type enzyme or second site revertants caused by transcriptional or other procedural errors (29). The evidence indicates that the hydrolytic activity of R127K, R127M, and R127A must result from catalysis by these enzymes and not by a more active contaminant.
The major Coomassie Blue staining protein band on isoelectric focusing gels also stains with the CPA activity reagents for both wild-type CPA and the Arg-127 mutants. Furthermore, the K, for inhibition of the Arg-127 mutant enzymes by PC1 and Cbz-Phe-Ala(P)-OAla are over 1000 times higher than that of the wild-type enzyme; therefore, the mutant enzymes are still active at concentrations of inhibitors that completely inactivate wildtype CPA. Additionally, the K,,, values of the mutant enzymes are higher than the wild-type K,,, values for the same substrates and a change in substrate specificity has occurred. Although much is known about CPA, the mechanism of hydrolysis is still not well understood.
Hydrolysis is believed to proceed by attack of Zr?+-bound water on the peptide bond to form a tetrahedral intermediate, but the exact nature and function of the amino acids involved is still unclear. The enzyme must stabilize the developing oxyanion to facilitate formation of the tetrahedral intermediate.
Arg-127 was replaced with Lys, Met, and Ala by site-directed mutagenesis. These three amino acids were chosen to evaluate the contribution to catalysis of the size, shape, and charge of Arg-127.
The data collected for the three Arg-127 mutant enzymes, R127K, R127M, and R127A, indicate that Arg-127 functions primarily by stabilizing the rate-limiting step. The loss in binding energy (AAG) of the rate-limiting transition state is substantial, ranging from 4.1 kcal/mol for R127K to 6.0 kcal/ mol for R127M and R127A hydrolysis of Cbz-Gly-Gly-Phe, Bz-Gly-Phe, and Bz-Gly-OPhe. Peptide and ester substrates are equally dependent on the contributions of Arg-127 to catalysis, as illustrated by the similar reduction in kcat for the hydrolysis of the matched peptide-ester pair Bz-Gly-Phe and Bz-Gly-OPhe.
Therefore, the initial prediction that Arg-127 would only be important for peptide hydrolysis is not supported by our experimental data. However, mutation of Arg-127 does not affect all substrates equally. The kcat values are decreased substantially for hydrolysis of Cbz-Gly-Gly-Phe, Bz-Gly-Phe, and Bz-Gly-OPhe by the mutant CPAs, whereas only a mild reduction in the kc,, values for the hydrolysis of Zgp. This interaction may account for the unusual strength of the binding interaction of phosphates to CPA (5). The binding affinity of these inhibitors to the Arg-127 mutants would be expected to decrease in parallel with k,,,/K,,,; the loss in binding energy for Cbz-Phe-Ala(P)-OAla ranges from 4.6 kcal/mol for R127K to 5.4 kcal/mol for R127M and is in excellent agreement with the observed decreases in k,,JK,,,. In contrast, the loss in binding energy for the ground state inhibitor, BzSA, ranges from 0.8 kcal/mol for R127K to 1.7 kcal/mol for R127M, similar to the changes observed for substrate binding.
Electrostatic calculations also suggest that the main role of Arg-127 should be transition state stabilization. The theoretical contribution of Arg-127 to transition state stabilization was calculated by totaling the electrostatic potentials generated by Arg-127 on the oxyanion (O(2)) and on the attacking water molecule (the positive charge was positioned either on 0( 1) or the modeled proton, Fig. 2 Table II), whereas the observed decrease in binding energy for Cbz-Phe-Ala(P)-OAla is only 5.4 kcal/mol. However, these calculations have not accounted for the desolvation energy of Arg-127 upon inhibitor binding.
This factor will reduce the contribution of the positive charge on Arg-127 to the binding energy of the phosphonate and phosphamidate inhibitors.
Since desolvation occurs in the ground state, it is irrelevant to transition state stabilization.
Having established that the positive charge of Arg-127 is important to CPA-catalyzed hydrolysis, can we learn anything further about the essential nature of arginine in this position? In addition to the positive charge, arginine has other important properties not shared by Lys, Met, or Ala. 1) It can form spatially separate hydrogen bonds, and 2) it disperses the charge over the entire guanidinium group. If Lys occupied the same position as arginine, the positive charge would be centered at the position of CZ (the carbon within the guanidinium group) of arginine and electrostatic calculations predict that substitution of Arg with Lys would only destabilize the rate determining transition state by 1 kcal/mol; a decrease of up to 4 kcal/mol is actually observed. Why then is the positive charge on Lys unable to mimic that of Arg? When substrate is bound, Arg-127 forms two hydrogen bonds, one to the scissile carbonyl and the second to Asp-142. Modeling of Lys into the space vacated by Arg in the Bop .CPA structure, predicts that Lys can form a hydrogen bond to Asp-142, but not to the scissile carbonyl. This second hydrogen bond may be important for proper orientation of the substrate relative to other important catalytic residues. It is, therefore, likely that hydrogen bonds formed by Arg-127 in the active site and its size and hydrophobicity contribute to the optimal positioning of the charge relative to the scissile carbonyl bond and to stabilizing the active site geometry. In the absence of these additional factors, proper transition state stabilization cannot be achieved. The actual position of Lys in the active site cannot be predicted, and structural characterization of the mutant enzymes will be required to completely understand these observations.
The binding affinity of PC1 toward the Arg-127 mutant enzymes has been significantly altered when compared with wild-type CPA; the loss in PC1 binding energy for the mutant enzymes is 4.1 kcal/mol.  (Table  II). The structure of Zgp (transition state analog of Cbz-Gly-Phe) bound to wild-type CPA shows that it binds abnormally when compared with Bop (transition state analog of Bz-Gly-Phe) and several other inhibitors (Fig. 1, Ref. 5). A likely explanation for the insensitivity of Cbz-Gly-Phe hydrolysis to Arg-127 replacement is the existence of yet another alternate binding mode for this substrate on the mutant enzymes. This is supported by the BzSA inhibition data, which shows that noncompetitive inhibition is observed for hydrolysis of Cbz-Gly-Phe by R127K but not for any other tested substrate. Within this binding mode the oxyanion may be partially stabilized by another active site residue. A change in the ratelimiting step for Cbz-Gly-Phe hydrolysis by the mutant enzymes could also account for these differences.
The structural data on the inhibitor analogs Bop and Zgp suggests an additional possibility.
The tetrahedral oxygen (O(2)) of Zgp, the transition state analog of Cbz-Gly-Phe, is closer to Arg-127 (Figs. 1 and 2, dis$ance 2.6 A) than to Zn2+ (Figs. 1 and Fig. 2, distance 3.5 A), while the tetrahedral oxygen (O(2)) of Bop, the transition state analog of Bz-Gly-Phe, is closer to Zn'+ (Figs. 1 and 2, distance 2.5 A) than Arg-127 (Figs. 1 and 2, distance 3.2 A). By inference Cbz-Gly-Phe would have the strongest interaction with Arg-127 and Bz-Gly-Phe the weakest, yet hydrolysis of Cbz-Gly-Phe is the least affected by the loss of Arg-127. Thus, one role of Arg-127 may be to prevent the exclusive interaction of the scissile carbonyl oxygen with Zn'+. Such an interaction could decrease the acidity of the Zn2+ bound water and hinder nucleophilic attack of water on the peptide bond. This function may be more essential for those substrates that prefer the putative Zn'+-O(2) binding mode over the Arg-127-O(2) binding mode. Supporting this possibility, studies on model compounds indicate that the pK, of Zn2+ bound water can vary 2-3 pK, units depending on the type of Zn2+ ligands and the coordination environment.
Furthermore, rate enhancement of peptide bond cleavage was dependent on the formation of Zn'+bound hydroxide as the pH rate profiles contained an inflection point which corresponded to titration of Zn'+-bound water (32, 33). However, it is also possible that the different binding modes of the two inhibitors do not reflect the actual substrate binding modes; alternatively they may reflect differences between a phosphonamidate (Zgp) and a ketone (BoP).
In summary, the evidence indicates that Arg-127 stabilizes the rate determining transition state by 6 kcal/mol, whereas it plays a more minor role in binding of the ground state structure. Arg-127 apparently functions as an electrophilic catalyst through interaction of the positively charged guanidinium group with the oxyanion. The importance of Arg-127 in CPA catalyzed hydrolysis is further underscored by comparison to the effect observed for mutation of Glu-270; reductions in k,,, for the Glu-270 mutants were of similar magnitude to that observed for the Arg-127 mutants.' Structural data on rat CPA and the Arg-127 mutant enzymes will be required to ' S. Gardell, personal communication. provide further insight into the observations made in these studies.