Phosphorylation of peptide substrates for the catalytic subunit of cAMP-dependent protein kinase.

The steady-state kinetic parameters for the phosphorylation of four peptides by the catalytic subunit of cAMP-dependent protein kinase were measured as a function of pH. For peptides containing the minimum consensus sequence, R-R-X-S-hyd (where hyd is a hydrophobic residue), the kcat/Kpeptide profile is bell-shaped with pK values of 6.4 and 9.4. Inhibition studies with the peptide LRRNAI indicate that the lower pK corresponds to an intrinsic pK on the enzyme, whereas the higher pK is perturbed upward by 1 pH unit. Viscosity studies verify that substrate stickiness accounts for the kinetic perturbation of the higher pK in kcat/Kpeptide. Substitution of the P-3 arginine with alanine (where serine is the P-site) yields a kcat/Kpeptide versus pH profile that is also bell-shaped, although both pK values are intrinsic acid dissociation constants of the enzyme. Replacement of the P-2 arginine with alanine removes the lower pK in the pH-rate profile without altering the higher pK. These results indicate that recognition of the P-2 arginine requires the ionization of an enzyme residue. This result implies that if the catalytic subunit mechanism involves general base catalysis, the ionization of this bse is not manifested in the pH-rate profiles.

metabolism, growth, and gene expression (Hanks et al., 1985;Cooper, 1990). Despite the discovery of a large number of protein kinases, to date, CAMP-dependent protein kinase remains unique. Its ability to be easily dissociated from its regulatory constituents provides a simple model for the entire protein kinase family. Detailed physical studies of this enzymatic reaction have been greatly enhanced by two recent breakthroughs: 1) the overexpression and purification of recombinant mammalian catalytic subunit from Escherichia coli (Slice and Taylor, 1989;Yonemoto et al., 1991) and 2) the xray crystallographic structure solution of the catalytic subunit with an inhibitor peptide bound at the active site (Knighton et al., 1991a(Knighton et al., , 1991b. Peptide studies have shown that the C-subunit will preferentially phosphorylate serine and threonine residues in the minimum consensus sequence R-R-X-S/T-hyd, where X is variable and hyd is a hydrophobic residue (Kemp et al., 1977;Zetterqvist et al., 1990). Replacement of one of the arginines with alanine or lysine lowers the specificity constant, V/K, by approximately 2-3 orders of magnitude (for review, see Jarv and Ragnarsson (1991)). Both steady-state kinetic (Cook et al., 1982) and isotope partitioning studies (Kong and Cook, 1988) indicate that the C-subunit binds ATP and substrate randomly, although initial binding of the nucleotide is preferred. The 2 orders of magnitude difference in K,,, compared to Kd (Kd > K,) for the heptapeptide Kemptide, LRRASLG, suggests that a rapid step occurs after ternary complex ( E . ATP.Kemptide) formation. Viscosity studies support a kinetic mechanism involving fast phosphoryl transfer from the y PO, of ATP to the hydroxyl group acceptor of the peptide followed by rate-limiting release of the product, ADP (Adams and Taylor, 1992). Phosphorylation of the peptide considerably weakens its affinity for the C-subunit so that it has no energetic consequence on the transition state for kcat (Whitehouse et al., 1983). Stereochemical studies indicate a direct, in-line attack of the hydroxyl group on the y PO, of ATP (Ho et al., 1988).
The kinetic processing of substrate depends greatly on pH. Yoon and Cook (1988) showed that the binary E .ATP complex can populate three protonation states, of which only one binds peptide and transfers phosphate. This conclusion derives from the bell-shaped pH dependence of V/K and the pH insensitivity of V. It is not clear which residue on the enzyme needs to be protonated for full activity. However, a compelling role for the group with the lower pK is general base catalysis. The C-subunit may efficiently deliver phosphate by providing an ionizable group close to the hydroxyl proton of the serinecontaining peptide. Removal of this proton would increase the nucleophilicity of the attacking group. We studied the pH dependence values of peptide analogs on the steady-state kinetic parameters to ask whether these ionizable residues are important for positioning the peptide for attack or for general base catalysis. Kinetic mechanisms were established 7747 Kinetics of CAMP-dependent Protein Kinase for four peptides that differ in size and overall charge, For two substrates, a neutral residue, alanine, systematically replaced each arginine in the consensus sequence. It is postulated that alterations in the peptide distal from the site of phosphotransfer should not affect the acid-base chemistry of the pH-rate profile if a true, general base mechanism is operative and is manifested in the pH-rate profile.
Peptides and Protein-All peptides were synthesized at the Peptide and Oligonucleotide Facility at the University of California, San Diego. Peptides were purified by reverse phase preparative high performance liquid chromatography. The concentration of some peptides was determined by turnover with the catalytic subunit under conditions of limiting peptide. The recombinant C-subunit was expressed and purified from E. coli according to previously published procedures (Yonemoto et al., 1991). The concentration of enzyme was measured by Am (Ao.lw = 1.2).
Kinetic Assay-The enzymatic activity of the C-subunit was measured spectrophotometrically (Cook et al., 1982). This assay couples the production of ADP with the oxidation of NADH by pyruvate kinase and lactate dehydrogenase. Typically, 2-5 mM ATP was preequilibrated with catalytic subunit in a buffer containing 1 mM phosphoenolpyruvate, 0.3 mM NADH, 12 units of lactate dehydrogenase, and 4 units of pyruvate kinase in a final volume of 1 ml. Reactions were initiated by adding varying amounts of peptide. Inhibition studies were done by preequilibrating 2 mM ATP, C-subunit, and inhibitor before adding substrate. All reactions were done in a buffer containing 50 mM Mes, 25 mM Tris, 25 mM Caps, and 50 mM NaCl (MTCN buffer) either in the presence or absence of glycerol or sucrose. The pH of the buffers was adjusted by adding small volumes of concentrated HC1 or NaOH. All kinetic measurements were performed at 23.0 "C, pH 6-10.8, and 10 mM free M$+. The C-subunit was sufficiently stable at both low and high pH extremes within the time frame of ligand preequilibration and assay. The C-subunit could be preequilibrated in pH 6 MTCN buffer for 10 min with no loss of activity. At pH 10.8, tlI2 for inactivation is approximately 30 min in MTCN buffer.
For peptides 111 and IV (Table I), steady-state kinetic parameters were also determined by a discrete time point assay in addition to the continuous, coupled enzyme assay. In this technique, 0.35-0.53 pM C-subunit, 5 mM ATP, 15 mM MgCl,, and varying amounts of peptide (6-45 mM for IV and 22-90 mM for 111) were mixed in a total volume of 100 p1 of MTCN buffer (pH 6, 7.5, or 8). After 60-240 s of reaction time, the samples were diluted to 1.0 ml in an assay mix containing 1 mM phosphoenolpyruvate, 0.3 mM NADH, 12 units of lactate dehydrogenase, and 4 units of pyruvate kinase in pH 10 MTCN buffer. The initial amount of the C-subunit in the 100-pl reaction sample and the pH of the assay mix were chosen so that, upon dilution, little or no peptide was further phosphorylated. The amount of ADP produced during the reaction time period (before dilution) was measured by the total absorbance change at 340 nm (AA,). The absorbance change due to dilution of the assay mix (Ma) upon adding the reaction sample was subtracted from AA, by recording the absorbance change after adding 100 p1 of water to 900 pl of assay mix. The corrected absorbance change (AA,,, = AA, -A A a ) due solely to ADP production was used to measure reaction velocities as a function of peptide concentration. This method was used for measuring k., values for weakly bound substrates.
Solution Viscosity Measurements-The relative viscosity (f") of buffers containing glycerol or sucrose was measured relative to MTCN buffer at pH 8.0, 23.0 "C, using an Ostwald viscometer (Shoemaker and Garland, 1962). 20% and 29% glycerol buffers (w/w ' %) were used to obtain relative viscosity values of 1.5 and 1.8, respectively. Solutions of 26, 32, and 39% sucrose (w/w %) were used to obtain relative viscosities of 2.0, 2.4, and 2.9, respectively. All measurements of viscosity were performed in triplicate.
Data Analysis-The values of Vmax and Kpepti& were determined from plots of initial velocity versus substrate concentration according to Equation 2. u is the initial velocity, [SI is the concentration of the varied substrate, V, , is the maximal velocity, and Kpeptids is the Michaelis constant.
The maximal velocity was then converted to kat by dividing V, , by the total enzyme concentration. For several peptides, Lt/Kmtib values were obtained by plotting kob. versus peptide concentration and fitting the data by linear regression. Plots of kat/Kmti& ascertained either by Equation 2 or by linear regression were fit to Equations 3, 4, or 5.
y is the observed kst/Kpepti& at a given pH, C and C* are the maximum and minimum values of kst/Kpapti*, respectively, and pK1 and pKz are the lower and higher acid dissociation constants, respectively. Competitive inhibition data were fit by a Dixon plot (Segel, 1975 KI is the pH-independent dissociation constant and pK. and pKb are the lower and higher acid dissociation constants, respectively.

RESULTS
pH-dependent Steady-state Kinetic Parameters-The steady-state kinetic parameters, kcat and k&/Kpeptide, for peptides I and I1 (Table I) were measured as a function of pH under conditions of saturating ATP (2 mM) and 10 mM free Mg2+ using the continuous coupled enzyme assay. As illustrated in Fig. lA, kcat/Kpeptide is bell-shaped and kcat is constant over the pH range 6-10 for peptides I and I1 (kcat data is not shown). The kat/Kpptih data were fit to equation 3 to yield the pH-independent kCat/Kpeptide and the two pK values. Table  I lists the results of these fits. For peptides I11 and IV (Table  I), kcat/Kpeptide values were determined from linear plots of initial reaction velocity uersus substrate concentration. Fig.   1B illustrates the pH dependence values of kcst/Kpptide for these peptides. Like the di-arginine class of peptides (I and 11), peptide I11 has a bell-shaped pH-rate profile. The pH dependence of peptide IV gives the ionization of a single residue. Both peptides I11 and IV show a plateau in ket/KpPtide at high pH (pH > 9). This is in contrast to the di-arginine peptides, I and 11, that apparently approach zero at high pH. Equations 4 and 5 satisfy mathematically the shapes of the pH dependence values of I11 and IV, respectively. Table I compiles the best fits of k,t/Kgtide, kcat/Kk&e, pKl, and ~K z . The pK, values for peptides I, 11, and I11 are statistically identical. Likewise, the pKz values for I and I1 are the same. However, the pK2 values for peptides I11 and IV are statistically lower than those of peptides I and 11.
The kcat values for peptides I11 and IV could not be determined directly from the continuous coupled enzyme assay owing to their high Kpeptide values for the C-subunit (Table I).
Instead, a discrete time point assay was used to take advantage of smaller reaction volumes and to lower the total quantity of peptide needed to saturate the C-subunit in a 1-ml assay. In this modified assay, the C-subunit is combined with ATP and peptide in a small reaction volume (100 pl) for a desired time period before stopping the reaction by a 10-fold dilution into

FIG. 1. Plots of &JK-*
for peptides I-IV as a function A , Ll/Kw* for peptides I (0) and I1 (0). Equation 3 was used to fit these data (see text and Table I). B, kat/Kmti* for peptides 111 (0) and IV (0). These data were determined from the slopes of the initial reaction velocity versus peptide concentration and were fit to Equations 4 and 5, respectively (see text and Table I).
a high pH buffer (pH 10 MTCN buffer) containing the coupling reagents. The total amount of ADP produced was calculated from the corrected total absorbance change at 340 nm. The total production of ADP did not exceed 10% of the initial peptide or ATP concentrations and was linearly dependent on the C-subunit concentration (data not shown). The values for kcat were determined at two pH values for peptides I11 and IV. Table I compiles these results. For both peptides, kcat did not change between pH 6 and 8. The reduced  Table I.
Effect of Viscosity on the Kinetics of Peptide Z and ZV-kat and kat/Kpptide for peptides I and IV were measured at varying solvent viscosity in MTCN buffer under conditions of saturating ATP (2-5 mM), 10 mM free M e , and varying peptide concentrations. Fig. 2 shows the plots of the ratio of kcat and kcat/Kpeptih in the absence and presence of viscosogen as a function of relative solvent viscosity, tn'. Both steady-state kinetic parameters are linearly dependent on solvent viscosity for peptide I and IV. The data for kcat/Kpeptih for peptide I were removed for clarity. The slopes of these lines for peptide I are 1.0 f 0.08 and 0.97 f 0.10 for kat and kat/Kpeptih, respectively. For peptide IV the slope for kcat is 0.30 k 0.06. However, kcat/Kpeptid. for peptide IV is insensitive to solvent viscosity. ,411 kat/&ptide values for peptide I v were determined from the slopes of initial velocity uersus peptide concentration in the absence and presence of viscosogen. All kcat values for peptide IV were determined from the time point assay method. For all values of tn', the observed enzymatic velocities were dependent linearly on the C-subunit concentration, indicating that added viscosogen did not limit the rate at which the coupling enzymes converted ADP (data not shown).
pH-dependent Inhibition Kinetics-Peptide V is a competitive inhibitor of the C-subunit with respect to peptide I (Salerno et al., 1990). The apparent dissociation constant (KYp) for this inhibitor was measured as a function of pH in MTCN buffer. Varying amounts of inhibitor, saturating ATP (2 mM), 12 mM MgClZ, and fixed amounts of C-subunit were preequilibrated before adding a fixed amount of peptide I (80-125 p~) .
Plots of l / v versus inhibitor concentration (Dixon plot) were used to extrapolate to the value of K;"" (Segel, 1975). Fig. 3 illustrates the pH dependence of l/KYpp. The data gave a bell-shaped curve that was fit to Equation 6. Table  I compiles the pH-independent dissociation constant, KI, pK,, and pK6. pK, is statistically identical to pKl for all substrate peptides. Conversely, pKb is identical to pK2 for peptides I11 and IV but not peptides I and 11.

DISCUSSION
Kinetic Mechanism for Peptides I and 11- Yoon and Cook (1987) showed previously that the apparent second order rate constant for ATP, kat/KATP, and the maximal rate constant, kc.,, are pH-independent in the range of 6-10 when using peptide I as a phosphoryl acceptor. Alternatively, the pH dependence of ka,t/Kpeptide is bell-shaped with two defining pK values of 6.2 and 8.5. These data imply that the free and ATP-bound C-subunit exist in three ionization states and that ATP does not discriminate between these forms. In contrast, peptide I binds a single ionization state that supports phosphotransfer. Scheme I illustrates this mechanism at high ATP concentrations. Fully protonated (H2. E . ATP. S ) and fully ionized ( E . ATP. S) ternary complexes are not popu-  Table I lated. Under these conditions the pH dependence in kcat/ Kpptide gives intrinsic pK values for both free and ATP-bound C-subunit. pH-dependent studies of the competitive inhibitor, LRRAALG, confirmed that these pK values are, indeed, intrinsic (Yoon and Cook, 1987).
Our data for peptides I and I1 are consistent with this steady-state kinetic mechanism. kcat is independent of pH, and kcat/Kpeptide is bell-shaped between pH 6 and 10 (Fig. L4). Furthermore, the magnitudes of the steady-state kinetic parameters are equivalent except that pK2 for both peptides is approximately 1 unit higher (9.4 versus 8.5; Table I). We performed pH-dependent inhibition studies to see if this elevated pK value reflects the intrinsic pK of the enzyme (Fig.  3). Although pK, for the inhibition kinetics is the same as pKl for the steady-state phosphorylation kinetics (Table I), pKz is approximately 1 unit higher than pKb. Thus, the basic region of the pH-rate profile for peptides I and I1 does not reflect the intrinsic pK of an ionizing group on the enzyme. It is not clear why our data indicate a perturbed pK. However, the consistency of pK. and pKb compared to that of Yoon and Cook (1987) argues that the buffer components have no different effect on the ionization of the ATP-bound enzyme.2 Scheme I was modified to satisfy the inhibition and steadystate kinetic data. In MTCN buffer, peptides I and I1 must bind to the fully ionized enzyme-ATP complex so that the measured pKz reflects a perturbed pK on the enzyme. Scheme I1 shows this new mechanism. This scheme allows for the perturbation of pKz by substrate stickiness ( i x k3 > k Z ) . We presume that the fully protonated form (Hz. E . ATP) does not bind peptide since pK, in the inhibition kinetics and pKl in the steady-state kinetic studies are identical. Consideration of the steady-state kinetics of the other peptides will confirm this assertion (see "Kinetic Mechanism of Peptides I11 and IV"). Since the steady-state kinetic parameters for both peptides I and I1 are similar, both arginines are important recognition elements for the pH-dependent binding of peptides of 6-7 residues in length.
Interpretation of Viscosity Data-Solvent viscosity affects the phosphorylation of peptide I by the C-subunit (Adams and Taylor, 1992). The individual effects on kat and kcat/ Kpptide are consistent with the Stokes-Einstein equation and are interpreted with the microscopic rate constants for substrate/product binding and phosphoryl transfer. Large effects of viscosity on k, imply that product dissociation limits this parameter, whereas smaller effects on kcat/Kpptik indicate that substrate is in near rapid equilibrium with the enzyme in 100 mM Tris buffer (pH 8). Since the predominant kinetic pathway in Scheme I1 at pH 8 involves the binding and phosphorylation of peptide through the mono-protonated species, the pathways involving the fully protonated and fully ionized species can be ignored. Since the diffusional rate constants, Yoon and Cook (1988) used 100 m M concentrations of Mes for p H 5-6.5, Mops for pH 6.5-7.5, Taps for pH 7.5-8.5, and BPT for p H 8.5-10. k2, k"2, and k4, are indirectly proportional to the intrinsic solvent viscosity of the buffer and k3 is insensitive to viscosity, kc, and k,,/KPptid, are related to the relative viscosity (7"'). In this mechanism, phosphorylated peptide dissociates rapidly owing to its large Kd relative to that of ADP (Whitehouse et al., 1983). Thus, k4 is the dissociation rate constant for ADP alone. Equations 7 and 8 relate the slopes of the plots of keat and kcat/Kpptide in the absence and presence of viscosogen as a function of  (Table I). We have found that the level of substrate stickiness is greatly dependent on buffer conditions. In 100 mM Tris, pH 8, peptide I is no longer a sticky substrate (Adams and Taylor, 1992). Varying buffer conditions, though, have no measurable effects on the sensitivity of kcat to viscosogen. Since both values for (kcat)" and (kCat/Kpptide)' are near the diffusion-controlled limit, all proton transfers needed to achieve the active ternary complex, H . E.ATP. S, are fast and thermodynamically favorable at pH 8. For the mechanism outlined in Scheme 11, Equation 9 relates the displacement of the pK for kcat/Kpeptide (pK,) to the intrinsic pK (p&) and the microscopic rate constants (Cleland, 1977).
Since steady-state kinetic and competitive inhibition studies (Table I)  gives (k&/Kpeptid.)' = 1. This provides an internal check on both experimental methods. A lower limit for k Z can be set (k"P 2 25 s-') from the stickiness term and the lower limit for k3. Although changes in buffer components have effects on the stickiness of peptide I, it is not clear whether these effects are manifested separately on k3 and k-, or both.
Kinetic Mechanism for Peptides 1 1 1 and IV-The discrete time point assay method was used to measure kat values for peptides I11 and IV. Both parameters were unchanged by pH but were lower in value than those of peptides I and 11. Since the rate-determining step in kc, is product release for the diarginine class of peptides, we wondered whether the same is true for these peptides. The effect of solvent viscosity on k, for the phosphorylation of peptide IV is lower compared to the effect on peptide I (Fig. 2). We have shown previously that other di-arginine-containing peptides behave similarly (Adams and Taylor, 1992). The slope of kcac/kcat uersus qre' for peptide IV is intermediate between a diffusion and a nondiffusion-controlled reaction (( kc.,)" = 0.30 f 0.06 for peptide IV). Equation 7, kcat and ( kcst)" were used to calculate the rate constants for phosphoryl transfer (k3) and product release (k4) (k3 = 13 k 1.2 s" and k4 = 30 f 6.1 s-'). For this peptide the chemical rate is similar in value to the product release step. Additionally, the value of 30 s" for ADP release is close to the value of 22 s-l for peptide I. However, the replacement of 1 of the consensus arginines with alanine lowers the rate of phosphoryl transfer more than 19-fold (compare k~ 2 250 s" for peptide I). We presume that replacement of the other arginine with alanine in peptide I11 has a similar reduction in catalysis. Certainly, the lower value for kcat for this peptide compared to those of peptides I and I1 supports this point. The diminution in the rate of phosphotransfer for alaninecontaining peptides underpins the tight relationship between substrate binding and catalysis in this enzyme.
Using the continuous coupled enzyme assay, only kcat/ Kpeptide values could be derived owing to the weak affinities of peptides I11 and IV with the C-subunit. Nevertheless, the kcat/ Kpeptide values listed in Table I, which were determined from the slopes of initial velocity uersus substrate concentration, are consistent with values reported in the literature (Kemp et al., 1977). The shapes of the kcat/Kpept,de uersus pH profiles show three distinct differences from those of peptides I and 11. 1) The higher pK values (pK2) are approximately 1 unit lower than those of the other peptides; 2) peptide IV has no lower pK; 3) both peptides have plateaus in kcat/Kp(peptide at high PH-Since pKb and pK2 for peptides I11 and IV are equal (Table  I), the enzyme residue needed for optimum activity of the diarginine peptides (I and 11; Table I) is the same for the other peptides. It is not clear, however, if this enzyme residue makes a direct or indirect contact with the substrate. The plateau observed in kcat/Kpeptide at high pH implies that the fully ionized ternary complex in Scheme I1 can form and support phosphoryl transfer for peptides I11 and IV. It is also possible that peptides I and I1 may support catalysis from this complex. However, the perturbation of pKb places the plateau in an unobservable pH region (pH > 10). It is likely that the similarity of k,,t/Kp$ide and kccat/Kkttide (Table I) for peptides 111 and I v implies that PKb and pKb* are close in value since kcat is constant and Kpeptide closely represents the true thermodynamic dissociation constants when the substrate is not sticky and k3 does not greatly exceed kq. However, since no measurable plateau was found at high pH in the competitive inhibition studies (Fig. 3), the presence of both arginines in the ternary complex must raise pKb* sufficiently above pKb in Scheme 11. In other words, positively charged residues in the peptide reduce the acidity of the basic enzyme group.
The lack of a lower pK for the phosphorylation of peptide IV argues that a functional group on the enzyme ionizes to interact directly with the P-2 arginine.3 This residue does not, however, contact the P-3 arginine since the lower pK is observed in peptide 111. It is not likely that the residue with the lower pK is still operative in the mechanism for peptide IV but is perturbed to even lower pK values since this peptide binds weakly and offers no means of kinetic perturbation through substrate stickiness. Since solvent viscosity does not affect kcat/Kpptide for peptide IV (Fig. 2), this peptide is in rapid equilibrium with the E. ATP complex (i.e. k-, >> k3 in For all peptides studied in this manuscript, the site of phosphorylation is the P-site. All positions N-terminal to the serine are designated P-1, P-2, P-3, etc. For example, asparagine in peptide I1 is the P-1 asparagine. Scheme 11). Thus, pKl is not shifted out of the studied pH range of 6-10. Since pKl and pKa for peptides I, 11, and I11 are identical and kcat is pH-independent, these peptides do not bind the fully protonated species.
General Base Catalysis- Yoon and Cook (1987) have postulated that the acidic pK. observed in keat/Kpeptide may be due to the ionization of a general base catalyst near the site of chemical transformation. This residue would greatly facilitate phosphotransfer by removal of the hydroxyl group proton of serine-containing peptides. However, the data presented here are not consistent with this hypothesis since substitution of the substrate's P-2 arginine removes pK,. We cannot argue that the replacement of the arginine in peptide IV has local effects on the positioning of the serine relative to a putative general base. Since replacement of the P-3 arginine with alanine in peptide I11 does not remove the acidic pK., the putative base should still be functional for this peptide. However, there is no significant difference in k,JKz$ih or kcat (Table I) for peptides I11 and IV. One would expect large rate reductions for substrates that did not exploit the general base mechanism. Since this is not the case, we conclude that the enzyme does not recruit an amino acid side chain that ionizes with a pK. of 6.4 to remove the hydroxyl group proton. The large rate reduction in phosphoryl transfer incurred upon replacing the P-2 arginine with alanine in peptide IV implies that positioning of the substrate is critical for efficient catalysis.
The C-subunit positions several amino acid side chains near the P-2 and P-3 arginines cf the peptide based on a crystal structure resolved to 2.0 A (Knighton et al., 1993). The carboxyl group of Glu-127 and the phenolic oxygen of Tyr-330 interact with the P-3 arginine. In contrast, the carboxyl groups of Glu-170 and Glu-230 interact with the P-2 arginine. The ionization of either Glu-127 or Tyr-330 is unlikely to influence the pH-rate profile since peptide 111 has no effect on pKl (Table I). Therefore, these residues are not responsible for the pH dependence in kcst/KFptih. Chemical modification studies with hydrophobic and water soluble carbodiimides Taylor, 1988, 1990) suggest that Glu-170 and Glu-328 are exposed to solvent and Glu-230 is partially buried. Since pKa is 2 units higher than the solvent exposed pK of the glutamate side chain, Glu-230 is the likely candidate for pKl in Fig. 1. We are currently testing this hypothesis using site-directed mutagenesis.