13C and 1H NMR Studies of Ionizations and Hydrogen Bonding in Chymotrypsin-Glyoxal Inhibitor Complexes*

Benzyloxycarbonyl (Z)-Ala-Pro-Phe-glyoxal and Z-Ala-Ala-Phe-glyoxal have both been shown to be inhibitors of α-chymotrypsin with minimal Ki values of 19 and 344 nm, respectively, at neutral pH. These Ki values increased at low and high pH with pKa values of ∼4.0 and ∼10.5, respectively. By using surface plasmon resonance, we show that the apparent association rate constant for Z-Ala-Pro-Phe-glyoxal is much lower than the value expected for a diffusion-controlled reaction. 13C NMR has been used to show that at low pH the glyoxal keto carbon is sp3-hybridized with a chemical shift of ∼100.7 ppm and that the aldehyde carbon is hydrated with a chemical shift of ∼91.6 ppm. The signal at ∼100.7 ppm is assigned to the hemiketal formed between the hydroxy group of serine 195 and the keto carbon of the glyoxal. In a slow exchange process controlled by a pKa of ∼4.5, the aldehyde carbon dehydrates to give a signal at ∼205.5 ppm and the hemiketal forms an oxyanion at ∼107.0 ppm. At higher pH, the re-hydration of the glyoxal aldehyde carbon leads to the signal at 107 ppm being replaced by a signal at 104 ppm (pKa ∼9.2). On binding either Z-Ala-Pro-Phe-glyoxal or Z-Ala-Ala-Phe-glyoxal to α-chymotrypsin at 4 and 25 °C, 1H NMR is used to show that the binding of these glyoxal inhibitors raises the pKa value of the imidazolium ion of histidine 57 to a value of >11 at both 4 and 25 °C. We discuss the mechanistic significance of these results, and we propose that it is ligand binding that raises the pKa value of the imidazolium ring of histidine 57 allowing it to enhance the nucleophilicity of the hydroxy group of the active site serine 195 and lower the pKa value of the oxyanion forming a zwitterionic tetrahedral intermediate during catalysis.

The ␣-keto carbon of the glyoxal inhibitor is expected to occupy the same position as the carbonyl carbon of a substrate, and it has been shown that it is bound as a tetrahedral adduct, which should closely resemble the tetrahedral intermediate formed during substrate catalysis (1). By using 13 C NMR, it has been shown that ␦-chymotrypsin (1) and subtilisin (2) reduce the oxyanion pK a by ϳ6 and ϳ8 pK a units, respectively. It has been estimated that hydrogen bonding in the oxyanion hole will only reduce the oxyanion pK a by ϳ1.3 pK a units (1). This is consistent with the fact that hydrogen bonding is expected to be effective in both water and in the oxyanion hole, and so it should not reduce the oxyanion pK a to a value lower than that expected in water. This has led to the conclusion that hydrogen bonding in the oxyanion hole only has a minor role in lowering the oxyanion pK a (5-7). However, it has been proposed that substrate binding raises the pK a of the imidazolium ion of the active site histidine enabling it to promote ionization of the active serine hydroxyl enhancing its nucleophilicity for catalysis and to also reduce the oxyanion pK a value when a tetrahedral intermediate is formed (5). Binding by inhibitors such as glyoxals is expected to have a similar effect raising the pK a of the imidazolium ion of the active site histidine, which then reduces the oxyanion pK a . However, there is no direct evidence that the pK a of the imidazolium ion of histidine 57 is raised on binding glyoxal inhibitors. Therefore, one of the aims of this work is to examine the ionization state of histidine 57 in the presence of glyoxal inhibitors.
It has been shown that 1 H NMR can be used to observe the N ␦1 and N ⑀2 protons of the active site histidine in chymotrypsin and subtilisin with or without ligands bound (8 -10).
Therefore, in this work we have used 1 H NMR to observe the hydrogen-bonded proton located between the carboxylate group of aspartate 102 and N ␦1 of histidine 57 in both free ␣-chymotrypsin and in ␣-chymotrypsin-glyoxal inhibitor complexes. We have shown that glyoxal inhibitors are bound at alkaline pH values under our NMR conditions ([E] ϳ [I] ϳ1 mM). This has allowed us to determine whether the pK a of the active center histidine residue is raised in ␣-chymotrypsin-glyoxal inhibitor complexes. We have also used 13 C NMR to examine ionizations within the Z-Ala-Pro-Phe-glyoxal-␣-chymotrypsin complex. By using Z-Ala-Ala-Phe-glyoxal, we have been able to assess how proline contributes to binding and whether it affects the observed protonic equilibria within the ␣-chymotrypsin-glyoxal inhibitor complexes.
Enzyme Solutions-␣-Chymotrypsin (crystallized and lyophilized) was obtained from Sigma, and the amount of fully active protein (69%) was determined as described by Finucane et al. (11).
Inhibition of ␣-Chymotrypsin by Z-Ala-Pro-Phe-Glyoxal and Z-Ala-Ala-Phe-Glyoxal-The inhibition of the ␣-chymotrypsin-catalyzed hydrolysis of succinyl-Ala-Ala-Pro-Phe-p-nitroanilide by Z-Ala-Pro-Phe-glyoxal and Z-Ala-Ala-Phe-glyoxal was studied at 25°C in 0.1 M buffers containing 3.3% (v/v) dimethyl sulfoxide. The buffers used were potassium formate (pH 2.9 -4.5), sodium acetate (3.8 -5.6), potassium phosphate (pH 6.2-8.2 and ϳpH 12), and sodium carbonate (pH 9.7-11.0 Surface Plasmon Resonance-Surface plasmon resonance experiments were performed using a Biacore X instrument (Biacore, Uppsala, Sweden). ␣-Chymotrypsin was immobilized on a carboxymethylated sensor chip (CM5). The sensor carboxyl groups were activated by treatment with N-ethyl-N-(dimethylaminopropyl)-carbodiimide and N-hydroxysuccinimide. ␣-Chymotrypsin (10 g/ml) in 10 mM sodium acetate buffer at pH 5.5 was injected over a 7-min time period at a flow rate of 5 l/min. Nonreacted N-hydroxysuccinimide esters were deactivated by injecting with 1 M ethanolamine for 7 min at the same flow rate. Protein inhibitor interactions were studied at pH 7.23 using filtered and degassed 10 mM phosphate buffer. The inhibitor was injected at varying concentrations (6.2-248 nM) over a 2-min time period at a flow rate of 50 l/min. NMR Spectroscopy-NMR spectra at 11.75 tesla were recorded with a Bruker Avance DRX 500 standard-bore spectrometer operating at 125.7716 MHz for 13 C-nuclei. 10-mm-diameter sample tubes were used. The 13 C NMR spectral conditions for the samples of chymotrypsin inhibited by Z-Ala-Pro-[2-13 C]Phe-glyoxal or Z-Ala-Ala-[2-13 C]Phe-glyoxal at 11.75 tesla were as follows: 32,768 time domain data points; spectral width 240 ppm; acquisition time 0.541 s; 8.0-s relaxation delay time; 90°pulse angle; 200 transients were recorded per spectrum. Waltz-16 composite pulse 1 H decoupling with a BLARH100 amplifier was used with 16-db attenuation during the acquisition time, and a 34-db attenuation during the relaxation delay to minimize dielectric heating but maintaining the nuclear Overhauser effect. Spectra were transformed using an exponential weighting factor of 20 Hz. Samples of chymotrypsin inhibited by Z-Ala-Pro-[1-13 C]Phe-glyoxal were examined under the same conditions except that acquisition time was 0.135 s, the relaxation delay was 0.6 s, and 2320 transients were recorded per spectrum. Spectra were transformed using an exponential weighting factor of 40 Hz. 1 H NMR spectra were obtained at 500 MHz using 5-mmdiameter sample tubes. The 1 H NMR spectral conditions for the samples of chymotrypsin inhibited by Z-Ala-Pro-Pheglyoxal or Z-Ala-Ala-Phe-glyoxal at 11.75 tesla were as follows: 32,768 time domain data points; spectral width 40 ppm; acquisition time 0.818 s; 1.0 s relaxation delay time; 90°pulse angle; and 128 transients were recorded per spectrum. Water suppression was achieved using the Watergate W5 pulse sequence with gradients (12). Spectra were transformed using an exponential weighting factor of 50 Hz. 13 C NMR spectra confirming the formation of the ␣-chymotrypsin inhibitor complex with either 1-or 2-13 C-enriched glyoxal inhibitors were obtained prior to 1 H NMR studies.
Both 1 H and 13 C chemical shifts are quoted relative to tetramethylsilane at 0.00 ppm. In aqueous solutions the chemical shift of the ␣-carbon of glycine was used as a chemical reference as described previously (11). For nonaqueous solvents either 10% tetramethylsilane was used as an internal standard or an appropriate solvent signal was used as a secondary reference (13).
All aqueous samples contained 10% (v/v) 2 H 2 O to obtain a deuterium lock signal, as well as 10 mM potassium phosphate buffer to help maintain stable pH values during pH titrations.

RESULTS
Inhibition of the ␣-Chymotrypsin-catalyzed Hydrolysis of Succinyl-Ala-Ala-Pro-Phe-p-Nitroanilide by Glyoxal Inhibitors Z-Ala-Ala-Phe-Glyoxal and Z-Ala-Pro-Phe-Glyoxal-K i values were estimated at 25°C (Table 1) The binding constants for Z-Ala-Ala-Phe-glyoxal increased ϳ46-fold from 0.34 Ϯ 0.02 M to 15.6 Ϯ 0.8 mM at low pH and ϳ5.2-fold to 1.77 Ϯ 0.31 M at high pH according to a pK a of 4.32 Ϯ 0.14 and 10.70 Ϯ 0.44, respectively (Fig. 1a). The binding of Z-Ala-Ala-Phe-glyoxal to ␣-chymotrypsin at pH 7 is ϳ10ϫ less effective than with Z-Ala-Pro-Phe-glyoxal binding to ␦-chymotrypsin at pH 7 (1). The binding constants for Z-Ala-Pro-Phe-glyoxal increased ϳ19-fold from 0.033 Ϯ 0.002 M to 0.63 Ϯ 0.06 M at low pH and ϳ9.1-fold to 0.30 Ϯ 0.03 M at high pH according to a pK a of 4.00 Ϯ 0.21 and 10.41 Ϯ 0.22, respectively (Fig. 1b). In our NMR experiments we use concentrations of enzyme and inhibitor of ϳ1 mM. Therefore, the enzyme and inhibitor concentrations are ϳ3300 and ϳ560-fold greater than the K i values obtained with Z-Ala-Pro-Phe-glyoxal and Z-Ala-Ala-Phe-glyoxal, respectively, confirming that the inhibitor will be bound to the enzyme in our ⌵⌴R experiments even at pH values 3 and 12.
The pK a values of 4.0 -4.3 determined from the increase in K i values at low pH ( Fig. 1) are similar to that assigned to protonation of the hemiketal oxyanion in the enzyme inhibitor complex (Fig. 8, structure E to B). The pK a values determined from the pH dependence of K i or K d values will reflect ionizations within the enzyme-inhibitor complex (14). Therefore, the pK a values of 4.0 -4.3 observed with both inhibitors are assigned to hemiketal oxyanion formation in the enzyme-inhibitor complex (Fig. 8, structure B to E).
The ionization of the isoleucine 16/aspartate 194 ion pair in free ␣-chymotrypsin occurs with a pK a value of 8.8, and it causes a conformational change that reduces catalytic activity (15). The binding of ligands raises the pK a value of the ion pair to ϳ10.5 (16), which is in good agreement with the pK a of 10.4 -10.7 obtained from the pH dependence of the K d values when either of the inhibitors are bound to ␣-chymotrypsin. Therefore, pK a values of 10.4 -10.7 are assigned to a conformational change that decreases the K i values for inhibitor binding to ␣-chymotrypsin.
Examination of Inhibitor Binding Using Surface Plasmon Resonance-Surface plasmon resonance is widely used to determine binding constants and the on (k on ) and off (k off ) rate constants for ligand binding (17)(18)(19). Chymotrypsin was immobilized on the sensor chip by forming amide bonds between its free amino groups and the carboxyl groups of the sensor chip matrix as described under "Experimental Procedures." If the association rate constant (k on ) is diffusion-controlled (ϳ10 8 M Ϫ1 s Ϫ1 ) then separate signals for the free and bound ligands will only be observed for tightly bound inhibitors with apparent K d values of Ͻ ϳ1 M (20). With Z-Ala-Ala-Pheglyoxal, the K d values at low pH were Ͼ10 M. However, separate signals were observed for both free and bound inhibitor (Fig. 2B). Using surface plasmon resonance, the rates of association (k on ) and disassociation (k off ) for Z-Ala-Pro-Phe-glyoxal were estimated to be 4.83 ϫ 10 5 M Ϫ1 s Ϫ1 and 9.30 ϫ 10 Ϫ3 s Ϫ1 , respectively, at pH 7.23. Similar low rates of association were estimated at pH 4.75 from line broadening measurements with subtilisin (2). This explains why signal coalescence (of the free and bound ligand signals) and fast exchange broadening are not significant effects in the ⌵⌴R spectra presented (Fig. 2), even when the apparent disassociation constants approach millimolar values ( Fig. 1) at low and high pH values (20). At ϳpH 7 the disassociation constants for ␣-chymotrypsin and Z-Ala-Pro-Phe-glyoxal determined by inhibition kinetics and surface plasmon resonance were essentially the same (Table 1).
Then there was a further small increase in chemical shift from 104.02 Ϯ 0.02 to 104.26 Ϯ 0.02 ppm with a pK a of 8.83 Ϯ 0.49 (Fig. 3). This pK a is similar to that attributed to the ionization of the isoleucine 16/aspartate 194 ion pair in the free enzyme, which inactivates chymotrypsin when the carboxy group of aspartate 194 moves and protrudes into the active site increasing K i values. However, this pK a is usually raised when ligands bind or if the hydroxy group of serine 195 is chemically modified (11). Analysis of the pH-dependent changes in disassociation constants shows that the efficiency of binding of Z-Ala-Pro-Phe-glyoxal and Z-Ala-Ala-Pheglyoxal by ␣-chymotrypsin decreases with pK a values of 10.43 (Fig. 1b) and 10.67 Ϯ 0.24 (Fig. 1a), respectively. This shows that the ionization of the isoleucine 16/aspartate 194 ion pair must have a pK a Ն10.43-10.67. Therefore, the pK a of ϳ9 cannot be assigned to ionization of the isoleucine 16/aspartate 194 ion pair. The pK a of 5.23 obtained with ␣-chymotrypsin at 25°C is identical to the value obtained for ␦-chymotrypsin (Table 2), and it is assigned as in earlier studies (1) to the ionization of the hydrated aldehyde hydroxy groups of the inhibitor (Fig. 8, structures B and C).
Most 1 H NMR studies of the hydrogen-bonded protons of chymotrypsin have been undertaken at 4°C to minimize exchange broadening. As we have undertaken similar 1 H NMR studies with Z-Ala-Pro-[2-13 C]Phe-glyoxal and ␣-chymotrypsin, we have also undertaken low temperature 13 C ⌵⌴R studies to determine how temperature affects the pK a values. With Z-Ala-Pro-[2-13 C]Phe-glyoxal and ␣-chymotrypsin, lowering the temperature to 4°C produced a small 0.5 pH unit increase in the pK a from 5.23 to 5.71 and a small 0.2 pH unit decrease in the second pK a from 8.83 to 8.59 (Table 2). This   suggests that both ionizing groups have low enthalpies of ionization.
The signal at ϳ100.7 to ϳ104 ppm in the Z-Ala-Pro-[2-13 C]Phe-glyoxal-␦-chymotrypsin complex has been has been shown to be in slow exchange with a signal at ϳ107 ppm, which is formed with a pK a ϳ 4.5 (1). We have obtained similar results with the Z-Ala-Pro-[2-13 C]Phe-glyoxal-␣-chymotrypsin complex ( Fig. 2A). However, on extending our observations to alkaline pH values, it was found that the intensity of the signal at ϳ107 ppm had a bell-shaped pH dependence (Fig. 4a) increasing with a pK 1 of 4.45 Ϯ 0.16 to a maximum intensity at ϳpH 5.7 ( Fig. 2A, spectrum 5) and decreasing in intensity with a pK 2 of 7.81 Ϯ 0.15 at 25°C (Fig. 4a). Similar results were obtained at 4°C (Fig. 4b). However, although there was only a small increase of 0.4 in pK 1 , there was a larger increase of 1.21 in pK 2 suggesting that pK 2 may be due to cationic acid such as an amino group or imidazolium group. The hydration of aldehyde carbonyl groups is subject to general acid-base catalysis (21). Therefore, hydration is usually minimal at neutral pH values but increases at higher or lower pH values, whereas aldehyde concentrations decrease at high or low pH values. The intensities of the signals at ϳ107 ppm showed the expected pH dependence, decreasing in intensity at high or low pH values. This adds additional support to our earlier assignment of the signals at ϳ107 ppm to Fig. 8, structure E. This scheme has been modified from the scheme presented in earlier work (1) to allow for the re-hydration (Fig. 8, structures E to C) of the aldehyde carbonyl (Fig. 8, structure E) at high pH.
The signal ϳ104 ppm reached a maximum intensity at alkaline pH values ( Fig. 2A) with a pK a of 9.2 Ϯ 0.1 (Fig. 5a). This change in signal intensity is assigned to the hydration of the aldehyde carbon at alkaline pH values (Fig. 8, structures E to C).
When ␣-chymotrypsin was inhibited by Z-Ala-Pro-[1-13 C]Phe-glyoxal, a new signal was observed that titrated from 91.6 Ϯ 0.15 ppm at low pH to 97.55 Ϯ 0.11 ppm with increasing pH according to a pK a of 5.26 Ϯ 0.06. This titration behavior is essentially the same as that observed using ␦-chymotrypsin. The intensity of the signal at ϳ205 ppm increased with a pK a of 3.83 Ϯ 0.21 and decreased with a pK a of 8.29 Ϯ 0.21. A similar bell-shaped pH dependence was seen for the signal at 107 ppm, which is consistent with the formation and loss of species E in Fig. 8. However, although the signal at 104 ppm assigned to the quaternary carbon (Fig. 8, structure C) was detected at high pH, the signal at 97.55 ppm was not detected at high pH, presumably because of the larger line width of the aldehyde carbon. The fitted parameters were as follows: a at 25°C, pK a1 ϭ 4.45 Ϯ 0.16, pK a2 ϭ 7.81 Ϯ 0.15, and I (max) ϭ 4.81 Ϯ 0.35; b, at 4°C, pK a1 ϭ 4.86 Ϯ 0.14, pK a2 ϭ 9.02 Ϯ 0.17, and I (max) ϭ 3.88 Ϯ 0.35.

C]Phe-Glyoxal and Z-Ala-Ala-[2-C]
Phe-Glyoxal-Similar results were obtained at 25°C using Z-Ala-Ala-[2-13 C]Pheglyoxal and ␣-chymotrypsin (Fig. 2B) except that for the signal titrating from 100.79 to 104.00 ppm to 104.32 there was a small increase of ϳ0.3 pH units in the first pK a to 5.52 and a larger ϳ0.9 pH unit decrease in the second pK a to 7.89 at 25°C ( Table  3). The intensity of the signal at 107.8 ppm had a similar bellshaped pH dependence, and again at 25°C there was a small increase in the first pK a to 4.63, but there was a larger ϳ1 ppm increase in the second pK a to 8.87 (Table 3). The intensity of the signal at 104.3 ppm again reached a maximum intensity at alkaline pH values (Fig. 2B) with a pK a of 8.92 Ϯ 0.1 (Fig. 5b).
With Z-Ala-Ala-[1-13 C]Phe-glyoxal at 25°C, the intensity of the signal at 205.2 ppm showed a bell-shaped pH dependence ( Table 3) similar to that seen for the signal at ϳ107 ppm. The signal due to the hydrated aldehyde carbon of the Z-Ala-Ala-[1-13 C]Phe-glyoxal inhibitor bound to ␣-chymotrypsin titrated with increasing pH from 90.73 Ϯ 0.04 to 96.69 Ϯ 0.03 according to a pK a ϭ 5.62 Ϯ 0.02, which is similar to the titration results obtained with Z-Ala-Pro-[1-13 C]Phe-glyoxal and ␣or ␦-chymotrypsin (Table 2). But as we observed with the Z-Ala-Pro-[1-13 C]Phe-glyoxal inhibitor complex and ␣-chymotrypsin, the signal at 96.69 ppm (97.55 ppm with Z-Ala-Pro-[2-13 C]Pheglyoxal) did not re-appear at high pH as expected (i.e. signal at 97.4 in Fig. 8, structures C and D).
The formation of the signal at ϳ107 and 205.2 ppm (Fig. 8, structure E) occurred with a pK a value of 4.5-4.9, and the loss of these signals occurred with pK a values of 7.8 and 10.0, respectively. The formation of the signal at 104 ppm occurred with a pK a of 8.9 -9.2. Therefore, the one-step conversion of E to C in Fig. 8 may be a simplification. 1

H NMR of the Hydrogen-bonded Protons of ␣-Chymotrypsin and of Its Complex with Z-Ala-Pro-Phe-Glyoxal-
The hydrogen-bonded proton at ϳ18 ppm in unligated ␣-chymotrypsin has been assigned to the N ␦1 proton of the imidazolium ion of histidine 57 that is hydrogen-bonded to aspartate 102 (8,10,22). At 4°C this proton titrated from 18.0 Ϯ 0.1 at low pH to 14.7 Ϯ 0.1 at high pH with a pK a of 6.7 Ϯ 0.1 at 4°C. This value for ␣-chymotrypsin is lower than the value of 7.5 obtained using ␦-chymotrypsin at 3°C (10). However, it is in good agreement with the value of 6.56 Ϯ 0.13 obtained from the pH dependence of 1/K d for Z-Ala-Ala-Phe-glyoxal binding to ␣-chymotrypsin at 25°C (data not shown). This confirms that under our experimental conditions the active site histidine 57 of free ␣-chymotrypsin has a pK a of ϳ6.7. A signal at 13.0 ppm was detected at low pH which could not be observed at pH values Ͼ5.1. This signal has been observed in chymotrypsinogen and has been assigned to the N ⑀2 proton of histidine 57 (23). It has also been observed before in ␣-chymotrypsin at low pH (24,25) and in an ␣-chymotrypsin inhibitor complex with a trifluoromethyl ketone inhibitor (25). The signal at ϳ13.0 ppm has been assigned to the N ⑀2 proton of the imidazolium ring of histidine 57 of ␣-chymotrypsin (24).
When Z-Ala-Pro-Phe-glyoxal was added to ␣-chymotrypsin at pH 3.42, signals at 18.4 and 12.8 ppm were observed at 4°C (Fig. 6A, spectrum 1). In a complex between Ac-Leu-Phe-trifluoromethyl ketone and ␣-chymotrypsin similar signals were observed at 18.7 and 12.8 ppm as well as signals at 18.0 and 13.1 ppm from free ␣-chymotrypsin (25). The signal at 18.7 ppm was assigned to the N ␦1 proton of histidine 57 in the ␣-chymotryp-  sin-Ac-Leu-Phe-trifluoromethyl ketone inhibitor complex, but the signal at ϳ13 ppm was not assigned (25). We assigned the signals at 18.4 and 12.8 ppm at 4°C in the Z-Ala-Pro-Pheglyoxal-␣-chymotrypsin complex to the N ␦1 and N ⑀2 protons, respectively, of the imidazolium ring of histidine 57. The signals at 18.7 and 13.0 ppm that were observed at a similar pH but at 25°C (Fig. 6B, spectrum 1) were assigned in the same way. These signals confirm that the imidazolium ion of histidine 57 is present at low pH in the Z-Ala-Pro-Phe-glyoxal-␣-chymotrypsin complex at 4°C (Fig. 6A, spectrum 1) and 25°C (Fig. 6B,  spectrum 1). These signals are also present at high pH values with slightly different chemical shifts of 13.3-13.6 and 17.2-17.5 ppm at both 4°C (Fig. 6A, spectra 9 -11) and 25°C (Fig. 6B, spectra 8 -11). This demonstrates that the pK a of the imidazolium ion has been raised to a value Ͼ11 in the Z-Ala-Pro-Pheglyoxal-␣-chymotrypsin complex. At 4°C the ⌵⌴R signals broaden and disappear as the pH is increased to ϳ9.4 (Fig. 6A, spectra 2-8) but re-appear along with additional signals at pH values 10 -11 (Fig. 6A, spectra  9 -11). At 25°C the signals persist over much larger pH ranges (Fig. 6B, spectra 1-11). Robillard and Schulman reported (10) that in chymotrypsin chloromethyl ketone inhibitor adducts, oxyanion formation could be detected by a small titration shift from 17.25 to ϳ16.3 ppm. We have examined the signals at ϳ13 and ϳ18 ppm for similar small pH-dependent changes in chemical shift (Fig. 6), which we hoped could be attributed to ionizations within the enzyme-inhibitor complex. At 25°C with Z-Ala-Pro-Phe-glyoxal, the signals at ϳ13 and ϳ18 ppm were detected over a wider range of pH values (Fig. 6B) than at 4°C (Fig. 6A). The signal at ϳ18 ppm (Fig. 6B) decreased in chemical shift from 18.72 Ϯ 0.04 to 18.32 Ϯ 0.02 and to 17.40 Ϯ 0.01 ppm as the pH was raised (Fig. 7), and these decreases depended on pK a values of 4.17 Ϯ 0.34 and 8.18 Ϯ 0.19, respectively, which appear to reflect the exchange processes converting structures B to E to C in Fig.  8. Therefore it appears that the N ␦1 proton of histidine 57 is being affected by the formation of both the hemiketal oxyanion (Fig. 8, structure E) and the hemiacetal oxyanion (Fig. 8, structure C). The signal at ϳ13 ppm showed smaller increases in chemical shift as the pH increased (Fig. 6), which were too small to be reliably analyzed for pK a values. At 4°C with Z-Ala-Pro-Pheglyoxal, a signal at 14.7 ppm appeared at pH values 10.43 (Fig. 6A, spectrum 11) and 11.03 (Fig. 6A, spectrum 12) because of the N ␦1 proton of the imidazole ring of histidine 57 suggesting that some free ␣-chymotrypsin had been formed at the highest pH values because of the alkali-catalyzed breakdown of the inhibitor (1).

TABLE 3 Titration constants obtained from pH-dependent changes in the intensities of the 13 C-enriched signals in glyoxal inhibitor complexes formed with ␣-chymotrypsin
An additional signal at 16.4 -16.6 ppm was also detected at high pH values in both inhibitor complexes at 4 and 25°C. A similar signal has been observed in chloromethyl ketone derivatives of ␦-chymotrypsin, and it was in slow exchange with a signal at 17.3 ppm at 3°C, but they were in fast exchange at 16°C and titrated with a pK a of 8.4 (10). It was suggested that this signal was because of the imidazole N ␦1 proton, and the pK a resulted from the conformational change resulting from the pH-dependent disassociation of the tetrahedral adduct. Likewise, formation of this signal in our experiments could reflect the conformational change that makes inhibitor binding weaker at high pH. 1

H NMR of the Hydrogen-bonded Protons of ␣-Chymotrypsin and of Its Complex with Z-Ala-Ala-Phe-Glyoxal-When
Z-Ala-Ala-Phe-glyoxal was incubated with ␣-chymotrypsin at 25°C, signals at 18.7 and 12.9 ppm were detected at low pH and signals at 17.4, 16.6, and 13.5 ppm were detected at high pH values (Fig. 6C), as was observed when ␣-chymotrypsin was incubated with Z-Ala-Pro-Phe-glyoxal (Fig. 6B). All these signals (Fig. 6C) underwent similar pH-dependent changes to those observed with Z-Ala-Pro-Phe-glyoxal (Fig. 6B). The chemical shift of the signal at ϳ18 ppm reflected (Table 4) both the formation of both the hemiketal oxyanion (Fig. 8, structure E) and the hemiacetal oxyanion (Fig. 8, structure C).

DISCUSSION
Z-Ala-Pro-Phe-glyoxal can exist in both cis and trans forms with the trans form (91%) predominating over the cis form (9%) in dimethyl sulfoxide (1). The carbon atoms can have different chemical shifts in the cis and trans forms. Therefore, compounds like Z-Ala-Pro-Phe-glyoxal-containing proline may give rise to up to twice as many 13 C NMR signals as compounds such as Z-Ala-Ala-Phe-glyoxal, which do not contain proline. Careful examination of our spectra (e.g. Fig. 2) of ␣-chymotrypsin bound to Z-Ala-Pro[2-13 C]Phe-glyoxal and Z-Ala-Ala-Phe-glyoxal gave no evidence of additional signals because of a mixture of cis/trans isomers. Therefore, we conclude that Z-Ala-Pro-Phe-glyoxal bound predominantly in one form. Because chymotrypsin is specific for the trans isomer of substrates containing X-Ala-Pro-Phe-p-nitroanilide (26) then we propose that the Z-Ala-Pro-Phe-glyoxal inhibitor is bound in the trans form. Substitution of the alanine residue in the S 2 subsite with proline led to a 12-fold improvement in binding. This is consistent with the observation that replacing alanine residues in the S 2 subsite with proline residues enhances substrate catalysis (27).
Surface plasmon resonance studies have shown that the apparent rate of association of glyoxal inhibitors with chymotrypsin is significantly slower (4.8 ϫ 10 5 M Ϫ1 s Ϫ1 ) than the diffusion-controlled limit (10 8 M Ϫ1 s Ϫ1 ). This is because the k on is a two-step process consisting of a diffusion-controlled secondorder association followed by a first-order rate-limiting step leading to the reversible formation of a tetrahedral adduct (Fig.  8, structures B to E). This has greatly facilitated our 13 C NMR studies as there is no fast exchange broadening of the NMR signals.
At 25°C the signal at ϳ107 ppm in the ␣-chymotrypsin-Z-Ala-Pro-Phe-glyoxal complex, assigned to the hemiketal oxyanion (Fig. 8, structure E), is lost in slow exchange processes at low and high pH with pK a values of 4.5 and 7.8, respectively (Table 3). At low pH this loss correlated with the increase in the disassociation constant for the inhibitor (Table 1). This led to the suggestion that oxyanion formation promotes inhibitor binding. However, our current studies show that at alkaline pH   values the disassociation constants for inhibitor binding to ␣-chymotrypsin increase with a pK a of ϳ10.7 with Z-Ala-Ala-Phe-glyoxal and with a pK a of 10.4 with Z-Ala-Pro-Phe-glyoxal (Fig. 1). The decrease in binding with these pK a values is thought to reflect the conformational change associated with the ionization of the isoleucine 16-aspartate 194 ion pair. Inhibitor binding is expected to increase the pK a value of the ion pair from 8.8 in free ␣-chymotrypsin to a value of ϳ10.5 in ␣-chymotrypsin-glyoxal inhibitor complexes. The fact that binding is apparently decreased by protonation of the hemiketal oxyanion but not by formation of the hemiacetal oxyanion (Fig. 8, E or B to C to D) suggests that the hemiacetal oxyanion does not make a significant contribution to inhibitor binding.
The monophasic titration shift of ϳ1 ppm for the N ␦1 proton of histidine 57 in chloromethyl ketone derivatives of chymotrypsin has been attributed to oxyanion formation (10). Therefore, the larger 1.4 ppm biphasic titration shift of this signal at ϳ18 ppm in the glyoxal inhibitor complexes is expected as two oxyanions are formed in these inhibitor complexes (Fig. 8). In the chloromethyl ketone inhibitor complexes, the oxyanion is 5.7 Å from N ␦1 of histidine 57 (28). Using x-ray crystallographic data from Ac-Leu-Phe-CF 3 (29), the equivalent hemiketal oxyanion in the glyoxal inhibitor complexes is estimated to be 7.4 Å from N ␦1 of histidine 57. This shows that the chemical shift of the N ␦1 proton of histidine 57 is affected by electrostatic interactions with groups up to at least ϳ7.5 Å.
Several studies have provided evidence for a low barrier hydrogen bond being formed when specific trifluomethyl ketone inhibitors bind to chymotrypsin (30,31) raising the histidine pK a to 11-12 (32). In this work with ␣-chymotrypsin and in earlier work with both ␦-chymotrypsin (1) and subtilisin (2), we have shown that specific substrate-derived glyoxal inhibitors form tetrahedral adducts analogous to the tetrahedral intermediate formed during catalysis and that the oxyanion pK a is reduced by 6 -8 pK a units. It has been suggested that the primary factor in reducing the pK a of the oxyanion is its electrostatic interaction with the imidazolium ion of histidine 57 (5-7, 33, 34). For these interactions to be effective, the imidazolium ion pK a must be similar (pK a ϳ15) to that of the oxyanion and serine hydroxyl group in the absence of this interaction (31,35). Our results show that the active site histidine residue is fully protonated up to pH 11, and so when the glyoxal inhibitor is bound the imidazolium ion of the active site must have a pK a Ͼ 11. This shows that inhibitor binding raises the pK a of the active site histidine residue. Therefore, substrate binding should also raise the pK a of histidine 57 enabling it to act as an efficient general base catalyst for deprotonation of the hydroxy group of the active site serine residue. This will greatly enhance the nucleophilicity of the active site serine hydroxyl group, promoting tetrahedral intermediate formation and so promoting catalysis.
Raising the pK a of the active site histidine residue by ligand binding will also allow its pK a to approach or exceed that of conjugate acid of the oxyanion, and it will permit optimal electrostatic interaction between the imidazolium ion of histidine 57 and the oxyanion as well as optimal hydrogen bonding between the imidazole group and the conjugate acid of the oxyanion. These interactions should lower the oxyanion pK a and raise the pK a of the imidazolium ion, resulting in the formation of a zwitterionic tetrahedral adduct with glyoxal inhibitors and zwitterionic tetrahedral intermediates during catalysis. Similar increases in the basicity of the active site histidine have also been observed when peptidyl trifluoromethyl ketones (25,32) or peptidyl boronic acid inhibitors (36) form negatively charged tetrahedral adducts with chymotrypsin. The pK a of the active site histidine residue is also raised when subtilisin is covalently modified to give the monoisopropylphosphoryl enzyme (37). The pK a shifts of monoisopropylphosphoryl chymotrypsin and subtilisin could be predicted if a dielectric constant of ϳ4 was used (38).
It has been suggested that the elevation of the histidine 57 pK a on ligand binding is because of the decrease in the effective dielectric constant between the carboxylate group of aspartate 102 and the imidazolium ion of histidine 57, which enhances the electrostatic interaction between them (5). Mutations converting aspartate 102 into the neutral residues alanine (39) and asparagine (40,41) inactivated the enzyme, but converting it to a negatively charged cysteine residue only led to a small decrease in catalytic efficiency (42). This supports the suggestion (5) that the increased electrostatic interaction between aspartate 102 and histidine 57 on binding ligands is an important factor in raising the pK a of histidine 57 allowing it to enhance the nucleophilicity of the hydroxyl of serine 195 and to reduce the pK a of the oxyanion.