Purification, Properties, Kinetics, and Mechanism of 8-N-Acetylglucosamidase from Aspergillus niger*

8-N-Acetylglucosaminidase has been purified from an acetone extract of Aspergillus niger. The protein has a M,. = 149,000. It contains neither Mn”, Zn2+, nor cysteine and exhibits no cation requirement for activity. Isoelectric focusing separates two isozymes; the major isoenzyme has a pl = 4.4. Both isozymes exhibit 8-N-acetylgalactosaminidase and 8-glucosidase, as well as glucosaminidase activity. The mechanism of action of this enzyme has been studied in detail using a variety of substrate structure/ activity and kinetic experiments. Rate data plotted uer-SUB pH depends on the following ionization constants, respectively: for pK,,,, 2.95; for log R,,, 7.6; and for log &/Xm, 2.95 and 8.25. The kcat value of H20/D20 for p-nitrophenyL/3-N-acetylglucosaminide hydrolysis is 1.27 at pH 4.6 and 1.00 at pH 7.0. The p value for the hydrolysis of pare-substituted phenylglucosaminides is +0.36; p* for the hydrolysis of fluoro-substituted N-acetyl derivatives is -1.41. Two sulfur-containing sub- strate analogues, the 1-thioglucosaminide, and the N-thioacetyl derivative, exhibit either no or little sub- strate activity. The hydrolysis of the 2,4-dinitrophenyl-glucosaminide Stopped flow studies were carried out on a modified Durrum Instrument interfaced with a PDP-8 computer system. 2,4-DiNP-PGlcNAc was used as substrate for these studies since the production of 2,4-dinitrophenol (as the anion) could be followed continuously at 380 nm. The hydrolysis was carried out in 50 m~ citrate, pH 4.6, at 25°C. The stopped flow enzymatic rates were determined for the fust 200 ms of the reaction, and were corrected for spontaneous hydrolysis of substrate.

general acid catalysis which apparently promotes formation of a glycopyranosyl carbonium ion intermediate (1-3) via a typical SN1 process. The retention of configuration observed also has been associated with the classical double displacement (ping-pong) mechanism in which the oxocarbonium ion is trapped by an enzyme nucleophile generating a glycosylenzyme intermediate (4). However, whether the putative enzyme nucleophile is covalently linked to the glycosyl unit or is simply part of an ion pair appears to remain an unresolved question (2, 5, 6). Irrespective of this ambiguity, there is no evidence for the reactions catalyzed by glycosidases to be SN2 and single displacement (sequential) in nature.
An additional factor in the hydrolysis of 2-acetamido-2deoxyglycosides is potential anchimeric assistance by the acyl side chain (1, 7-9). Such a role for the acetamido group is attractive €or two reasons: 1) it can explain the large difference in reactivity between glycosides (which do not hydrolyze spontaneously) and the corresponding glycosaminides (9), and 2) it is consistent with the retention of configuration observed (1, 10). In parallel with mechanisms employing an enzyme nucleophile as above, proposals focusing on the acetamido group failed to distinguish between covalent participation (an SNi displacement process leading to an oxazoline intermediate) or simple ion pairing between the oxocarbonium ion and the partial negative charge on the acyl oxygen. P-Glucosaminidase is an exoglycosidase which removes Nacetylglucosamine from glycosides and polysaccharides with retention of configuration as has been shown to be the case for many other similar exoglycosidases (1, 11). The enzyme is specific for the p configuration and is fairly specific for the sugar moiety; it exhibits very little specificity for the aglycone (12). Glycosyl transfer of the sugar moiety to hydroxylic acceptors other than water has been demonstrated (13), which suggests the presence of an enzyme-glycosyl intermediate of some type and would appear to exclude from consideration a single displacement, SN2-type reaction mechanism. However, while lysozyme, another glucosaminide hydrolase, has been the subject of much research, P-glucosaminidase has been much less well characterized and less mechanistic information on this enzyme has been published. Available data (13-16), while useful, have enabled only qualitative aspects of the reaction mechanism to be established. Therefore, we undertook a detailed mechanistic study of P-N-acetylglucosaminidase (from Aspergillus niger) in an attempt to establish the mechanism of action of this enzyme type. PGalNAc, phenyl-N-acetyl-8-D-glucosaminide, GlcN. HCl, GlcNAc, monofluoroacetate, and bovine serum albumin were all obtained from Sigma. p-Nitrophenyl-P-D-xyloside and p-Np-PGlcN-TAc were gdts from Dr. 0. P. Bahl (State University of New York at Buffalo). p-Nitrophenol, p-methoxyphenol, p-chlorophenol, 2,4-dinitrophenol, trifluoroacetic acid, difluoroacetic acid, and bromoacetyl bromide came from Aldrich. An acetone powder extract (Rhozyme HP-150) of A. niger was obtained from Rohm and Haas. All deuterated solvents were purchased from J. T. Baker. All other reagents were of the highest purity available from commercial sources.

Analytical Methods
Specific rotations were obtained on a Hitachi-Perkin Elmer 140 automatic polarimeter. NMR spectra were recorded using either Varian A60 or XLl00 instruments and IR spectra using a Beckman IR33 Spectrophotometer. Melting points, obtained on a Fisher-Johns Mel-Temp apparatus, are uncorrected. Thin layer chromatography was performed using Silica Gel 6 plates with fluorescent indicator (Eastman Kodak). Elemental analyses of all substrates were done by Galbraith Laboratories to ensure purity and were satisfactory in all cases. Protein analytical methods are described in the figure legends. Unless noted otherwise, all operations were performed at ambient temperature (-22OC).
2,4-Dinitrophenyl-2-acetamido-2deoxy-3,4,6-tri-O-acetyl-~-~-glucopyranoside (0.45 g) was stirred in a mixture of 30 ml of dry MeOH and 20 ml of dry CHCL. After the gradual addition of 10 ml of dry methanolic-HC1 solution (16%, w/w), the mixture was stirred for 3 h. The resulting solution was flash evaporated at 20°C to an oil. The addition of 20 ml of water caused crystallization of the product, which was immediately filtered and washed with anhydrous ether. The resulting compound was dispersed in ether, heated, and filtered at least five times to remove contaminating 2,4-dinitrophenol. This yielded 0.12 g (36%) of product, m.p. 121-123°C; literature, 124-125°C (20). TLC in benzene/methanol ( 7 1) revealed one component. Normal recrystallization procedures using such solvents as methanol, ethanol, or water caused decomposition and contamination of the product with 2,4-dinitrophenol.
Enzyme Assays 8-N-Acetylglucosaminidase, p-galactosidase, a-galactosidase, P-Nacetylgalactosaminidase, p-glucosidase, a-glucosidase, p-mannosidase, a-mannosidase, and P-xylosidase specific activities were all determined by measuring the liberation of p-nitrophenol (as the phenolate anion) from the respective glycosides. In the standard assay, 100 pl of 50 rxm sodium citrate (pH 4.6) and 100 pl of a 5 mM were mixed together and equilibrated at 37°C. To this were added 20 solution of the appropriate p-nitrophenyl glycoside in the same buffer pl of an appropriate dilution of enzyme, and the reaction allowed to proceed at 37'C for either 5 or 10 min, then stopped by the addition of I ml of 0.2 M Na2C03. The resulting phenolate anion was measured at 420 nm on an Hitachi-Perkin Elmer model 139 spectrophotometer equipped with digital readout.
Enzyme assay using the para-substituted phenyl-2-acetamido-2deoxy-P-D-glucopyranoside substrate analogs proceeded 88 above ex-cept that the reaction was stopped by addition of 1.0 ml of a solution containing 98 ml of 0.2% NaZC03 in 0.1 N NaOH, 1 ml of 1% potassium tartrate, and 1 ml of 0.5% CuS04. It was allowed to stand for 10 min, and was then vortexed during the addition of 100 pl of 50% Folin-Ciocalteu reagent. After 90 min, the blue-colored Folin-phenol complex was measured spectrophotometrically at 740 nm. The Folin-Ciocalteu reagent has been used in assaying for a variety of phenols (26). Standard curves for each of the phenols were constructed as were color production versus time curves.

Kinetic Analysis
The kinetic constants ( K , and VmJ were determined by Lineweaver-Burk analysis and by computer analysis of the best fit hyperbola to the u versus S kinetic data (27). The computer programs were obtained from Professor W. W. Cleland, University of Wisconsin, and modified slightly to expedite usage with the Cyber 173 computer. The computer-determined kinetic constants did not differ significantly from those obtained by the Lineweaver-Burk method.
Inhibition studies were analyzed primarily by computer programs obtained from Dr. Cleland, but were also analyzed by the Lineweaver-Burk method; both methods yielded comparable results. The enzymatic rate data were normally obtained with two inhibitor concentrations.
pH-rate kinetic experiments were carried out in a phosphate-citrate McIlvaine constant ionic strength (0.1 M) buffer system (28). Enzyme stability was established over the pH range used. The kinetic constants, obtained as above, were plotted as log k,,,, pK,, and log k& KO, versus pH. On the basis of the observed variations of these parameters with pH, the data were fit to appropriate pH functions which yielded the pH-independent values of the kinetic constants and the residual least squares (variance) of the fit (27). The theoretical curves in the figures were subsequently constructed using these values.
Stopped flow studies were carried out on a modified Durrum Instrument interfaced with a PDP-8 computer system. 2,4-DiNP-PGlcNAc was used as substrate for these studies since the production of 2,4-dinitrophenol (as the anion) could be followed continuously at 380 nm. The hydrolysis was carried out in 50 m~ citrate, pH 4.6, at 25°C. The stopped flow enzymatic rates were determined for the fust 200 ms of the reaction, and were corrected for spontaneous hydrolysis of substrate.

Purification of /?-N-acetylglucosaminidase
Crude Extract-Rhozyme HB-150 (Rohm and Haas), an acetone powder extract of A. niger (100 g), was suspended in 650 ml of distilled water, stirred for 1 h, then centrifuged at 32,000 X g. The supernatant was saved, the pellet was reextracted with an additional 200 ml of distilled water, and the supernatants were combined ("crude extract").
Ammonium Sulfate Precipitation-To 850 ml of crude extract were added 476 g of (NH4)2S04. The suspension was stirred for 15 min and centrifuged at 31,000 X g for 1 h. The resulting pellet was redissolved in 1.6 liters of 50 mM citrate, pH 4.6. To this were added 1230 g of (NH4)$04; the suspension was stirred for 30 min and then centrifuged as above. The supernatant was discarded and the pellet was redissolved in a minimum volume (-600 d) of 50 mM citrate, pH 4.6 ("100% ammonium sulfate precipitation").
Batch DEAE-The 600 ml of 100% ammonium sulfate precipitation were diluted to 6 liters with 3 liters of distilled water and 2.4 liters of 50 mM phosphate, pH 4.5; the pH was adjusted to 4.8 with 5 N NaOH. Enough 20% NaCl was added to bring the conductivity from 7.8 to 12.0 mmho. To this solution were added 475 g (suction filtered wet weight) of DEAE-cellulose previously equilibrated with 50 mM phosphate, pH 4.8. The suspension was mixed thoroughly for 15 min, then suction filtered. The filter cake was resuspended in 500 ml of 50 mM phosphate, pH 4.8, which contained 0.15 M NaCl and was suction filtered again. The filtrates ("batch DEAE") were saved.
Batch Hydroxylapatite-The 8 liters of batch DEAE were diluted with distilled water until a conductivity of 2.0 mmho was attained (to -12 liters). T o this were added 1.2 liters of swelled and settled hydroxylapatite HTP (-400 g of dry weight), the suspension stirred for 20 min, and then suction filtered. The fitrate was discarded and the filter cake was resuspended in 2 liters of 0.2 M phosphate, pH 4.8, stirred for 20 min, and suction filtered again. The fitrate was saved ("batch HA") following a n additional washing ( 5 0 0 ml) of the hydroxylapatite.
Sephadex G-150-The column (2.5 X 100 cm) was packed and equilibrated with 50 m~ phosphate, pH 4.8. A portion (25 m l ) of PMlO ultrafiitrate was loaded and eluted with 50 m~ phosphate, pH 4.8, at a flow rate of 9.5 ml/h. Fractions of 5 ml were collected after an initial 150 ml had eluted. Fractions 5 through 25, which contained the major peak of enzymatic activity, were pooled ("Sephadex G-150").
PMlO Ultrafiltration 11-The Sephadex G-150 fraction was dialyzed against distilled water (4 liters for 2 h), and the pH of the solution was lowered to 3.9 by addition of 5 m~ Following concentration to 50 ml (Amicon PMlO), the solution obtained ("PM10 ultrafiltrate 11") was in 2 mM phosphate, pH 3.9.
Sulfopropyl-Sephadex-The column (2.5 X 100 cm) was packed and equilibrated with 5 mM citrate, pH 3.9. The PMlO ultrafiitrate I1 (25 ml) was loaded and the column eluted with a stepwise pH gradient, as shown in Fig. 1. The fractions indicated in Fig. 1, which contained the most highly purified /?-N-acetylglucosaminidase, were pooled and then dialyzed against 50 m~ citrate, pH 4.6, for storage. This was the fraction used for the characterization and kinetic studies described in this paper. These purification steps are summarized in Table I Ion exchange chromatography of partially purified 8-glucosaminidase on Sulfopropyl-Sephadex. Fifty milliliters of concentrate of the appropriate fractions from the Sephadex G-150 column (see text) were applied to the column (2.5 X 95 cm). The column was eluted stepwise with 5 m sodium citrate buffers, pH 3.9, 4.55, and 4.75, at a flow rate of -30 ml/h. Five-milliliter fractions were collected. M , P-glucosaminidase activity; W, absorbance at 280 nm. The pooled fractions noted in the figure were used for the studies described herein. activity (Fig. 2). The major component had an isoelectric point of 4.4. The ,&glucosidase (and P-galactosaminidase) activity (see also Table I) comigrated with P-glucosaminidase activity (Fig. 2). The purified enzyme was tested for other glycosidase activity by assaying with the appropriate p-nitrophenyl glycosides (Table 11). These specific activities were not changed when the protein was re-chromatographed on an affinity column prepared with p-aminophenyl-1-thio-N-acetylglucosaminide linked to a spacer arm consisting of succinylated 3,3'-aminobispropylamine (30). A single protein and activity peak eluted from this column.
Analysis of the sedimentation behavior of this protein yielded a linear In c versus R'/2 plot. Plots of l/UN and l/uw

FIG.
2. Isoelectric focusing of purified P-glucosaminidase on polyacrylamide disc gels using a modification of the Baumann and Chrumbach method (29) employing N,N'-diallyltartardiamide with acrylamide. Enzyme, 150 pg, was applied to the top of photo-polymerized gels which contained 60% 4 to 6 ampholines, and 30% 3 to 10 ampholines. Focusing proceeded at constant voltage (200 V) for 20 h. One gel was stained for protein after fixing in 10% trichloroacetic acid for 3 h, and the other was sliced and assayed for H , P-glucosidase activity; and M , pH. pH and enzymatic activity. M , P-glucosaminidase activity; versus c intersected on the l/uapp axis, which is also consistent with homogeneity (Fig. 3B). The negative slopes of these plots suggest a self-associating system. Furthermore, in a two species plot (Fig. 3A), the data points fell on a straight line between the theoretical monomer-dimer and monomer-trimer behavior. Similar results were obtained at two enzyme concentrations (0.5 and 1.0 mg/ml).

Composition
The amino acid composition of purified /3-N-acetylglucosaminidase is given in Table 111. The analysis showed a low mole per cent of Lys, His, and Arg, and a high mole per cent of Asp and Glu. This is consistent with the low p1 and pH optimum of this enzyme. No cysteine or cysteic acid was detected in any of the sample runs.
The carbohydrate composition of this enzyme is shown in Table IV. No Sia or GalNAc was detected. There was a large  Roark and Yphantis (31). In B, the two standard plots of concentration in the cell versus l/uaPp are presented. The lines are least squares fit to the data points. In A, a two species plot (31, 32) is presented as l / u~ uersus UW. The theoretical 2 X M , and 3 X M, are represented as (--) with the data as closed circles. The experimental line is a least squares fit to the data points. percentage of Man and smaller, but significant, levels of GlcNAc. Small amounts of Gal and Fuc were present as well. Glc was detected, but was considered to be a contaminant arising from the Sephadex columns used in the purification.
A M , = 148,000 f 2,500 was determined using the sedimentation data. AM, = 149,000 was calculated based on the amino acid composition.
Effect of Metals a n d Other Inhibitors Sodium chloride up to 0.5 M showed no inhibitory effect on purified P-N-acetylglucosaminidase. At 100 mM, Zn2+, Ca2+, Cu2+, Hg2+, and Ag' all significantly inhibited the enzyme; Ag+ and Hg' were most effective. Neither iodoacetamide nor p-chloromercuribenzenesulfonic acid caused significant inhibition of enzymatic activity. Atomic absorption spectrophotometric analysis indicated that the purified enzyme contained neither zinc nor magnesium.

Kinetic Analysis
The kinetic behavior of P-N-acetylglucosaminidase toward a variety of substrates and substrate analogs is presented in Table V. The enzyme was inhibited by substrate concentrations >3 mM. Note thatp-Np-PGlcN andp-Np-PGlcNBz were neither substrates nor inhibitors at the concentrations employed, while Glc was without effect on the hydrolysis of glycosaminides.
pH Dependence Detailed pH-rate studies were performed using p-Np-PGlcNAc over a pH range, 2.5 to 8.0. The K, and kcat values for the enzymatic hydrolysis at these pH values were determined as described under "Experimental Procedures." These data, expressed as log kcat, pK,, and log kCat/K,,, versus pH, are presented in Fig. 4. The theoretical curves in Fig. 4 were derived using the  In addition, Glc, p-Np-PGlcN, and p-Np-PGlcNBz were without effect as inhibitors and the latter two were not substrates.  following pK, values: for log kc,, versus pH, 7.6; for pK, uersus pH, 2.95; and for log kCat/K, uersus pH, 2.95 and 8.25. The instability of the enzyme below and above those pH values noted in the figure precluded an analysis of the pH-rate behavior outside of these limits.

Kinetic Solvent Isotope Effect
Rates of hydrolysis of p-Np-PGlcNAc were determined in DzO at pH (pD) 4.6 and 7.0. The kcat values of H20/D20 found were 1.27 and 1.00, respectively.

Substituent Effects
Hammett studies were carried out utilizing p-Cl, p-NOz, p-H, and p-OCH3-substituted phenyl glucoside derivatives of GlcNAc. The rate of enzymatic hydrolysis of these derivatives is plotted as log V,.,/K, uersus up (36) in Fig. 5A. The slope of the line, equivalent to the p value for the reaction, is +0.36. Taft-type studies were carried out also utilizing p-nitrophenyl-2-deoxy-N-trifluoroacetyl, N-difluoroacetyl, and Nacetyl-P-D-ghcosaminide substrate analogs. The rate of enzymatic hydrolysis of these derivatives is plotted as log V,,,/ K, uersus u* (36) in Fig. 5B. The slope of the line, the p * value for the reaction, is -1.41. There were only minor vari-ations in the K , values for the members of these two series of substrates.

Stopped Flow Studies
In an effort to demonstrate a "burst" phenomenon in the enzymatic hydrolysis, stopped flow studies were conducted using the substrate analogue, 2,CdiNp-PGlcNAc. The hydrolysis of this substrate was followed continuously at pH 4.6 by measuring the production of the 2,4-dinitrophenolate anion.
The kcat obtained from the stopped flow experiments (1958 +-150 min") was compared to the kcat determined in steady state studies (2300 k 250 min") with the same substrate analogue; the two rates were found to be indistinguishable within experimental error. Steady state kinetic studies established a K,,, value of 0.27 mM for this substrate (Table V), thus demonstrating a binding comparable to that observed for the p-nitrophenyl derivative.

DISCUSSION
The isolation of A. niger P-N-acetylglucosanunidase resulted in an overall purification similar to that reported for other fungal hexosaminidases (37-39). The purified protein exhibited apparent homogeneity on polyacrylamide gel electrophoresis, with the only protein band coinciding with enzymatic activity. The linearity of the In c versus R2/2 plot and the intersecting l/uN and l / o~ plots on the y axis in the l/uaPp versus c plot also demonstrated the homogeneity of the preparation. That two different enzyme concentrations yielded similar results provided additional evidence for the purity of the preparation. The fact that the l / u~ and l/uw plots had negative slopes and were nonsuperimposable indicated that the protein exhibits some self-association. The self-associating behavior (and homogeneity) was further indicated by the two species plot of uw versus l/uN. The data points fell on a straight line for both concentrations of enzyme, although the line lay between those expected for a monomer-dimer, and monomer-trimer system, respectively. This system, therefore, appears to consist primarily of monomer-dimer with some trimer.
The purity of the protein was also demonstrated by isoelectric focusing experiments, which revealed only one major (PI 4.4) and one minor protein band. Both possessed enzymatic activity. Although the preparation exhibited a considerable amount of P-N-acetylgalactosaminidase activity, the P-N-acetylglucosaminidase and P-N-acetylgalactosaminidase activities co-purified, indicating that both were associated with the same protein. This phenomenon apparently is characteristic of P-N-acetylglucosaminidases from various sources (38-41). This enzyme preparation also possessed /I-glucosidase as well as a small amount of /I-xylosidase activity; neither activity has been detected in other purified P-N-acetylglucosaminidases. The P-glucosidase-type activity co-migrated with the major and minor P-N-acetylglucosaminidase activity bands on isoelectric focusing. This suggested that the P-glucosidase activity was associated with the same protein as weU. Further evidence for this conclusion was obtained when purified enzyme was passed through an affinity column for P-N-acetylglucosaminidase. As noted, the ratio of /.?-glucosidase to /3-N-acetylglucosaminidase activity was not affected, nor was any further purification obtained. P-N-Acetylglucosarninidase from A. niger resembles this enzyme from other sources. Its molecular weight falls within the range characteristic of these enzymes (42, 43). Like the major isozymes of other /3-N-acetylglucosaminidases (42,44), it has a p1 close to 4.4. The amino acid composition we report is very similar to that reported for P-N-acetylglucosaminidases isolated from various sources (42,45) except that the A. niger protein apparently lacks cysteine.
Although other P-N-acetylglucosaminidases lack GalNAc, most contain Sia (42,45); both are apparently absent in the A. nzger protein. The absence of GalNAc suggests that the carbohydrate attachment to the protein is not through Ser or Thr. Since a considerable mole per cent of GlcNAc was present, Asp or Glu may be involved in the conjugation.
The nature of functional groups in this enzyme remains unresolved. Neither pCMBS nor iodoacetamide affected enzyme activity. The lack of cysteine indicated by the amino acid analysis was consistent with these results. Electrophilic catalysis by a Lewis acid, e.g. Mg2' as in ,&galactosidase (46) or Zn2+ as in a-mannosidase (47), appears lacking in /I-Nacetylglucosamidase from A. niger. The enzyme did not contain either metal nor did it require either for activity. A catalytic role(s) for a carboxyl, imidazole, or tyrosyl side chain cannot be ruled out (see below).
The kinetic studies provide some insight into the mechanism of action of this enzyme. The two mechanistic questions posed by this enzyme type are: 1) is the reaction kinetically sequential or ping-pong (Schemes I and 11 below); and 2) is the reaction chemically SNI, SN2, or, perhaps, SNj in nature? The lack of a burst of 2,4-dinitrophenolate anion (aglycone) in the stopped flow studies using 2,4-diNp-,@GlcNAc and the dependence of the rate of substrate hydrolysis on the nature of the aglycone ( p = +0.36) shows that if Scheme I1 obtains, kz < k3. However, the interaction between the acetamido group and the C1 carbon indicated by this work would appear to preclude simple mechanisms involving an enzyme nucleophile. In such a mechanism, the side chain and enzymic group would presumably approach the glycosidic carbon from the same a or axial direction in competition with one another. Consequently, these inferences most reasonably support Scheme I for the hydrolysis of N-acetylglucosaminides, but, by themselves, do not indicate whether the reaction is a SNI, SN2, or SNi process.
That the acetamido group is involved in the reaction is indicated by: 1) p * = -1.41 for the hydrolysis of the fluorosubstituted N-acyl derivatives; and 2) the weak substrate activity of the thioacetamido derivative. These substrates exhibited Michaelis constants comparable to that for p-Np-PGlcNAc, thus, their poor substrate activity was not due to large differences in binding. These data suggest that 1) the acetamido group is needed, and 2) the basicity of the carbonyl oxygen (or sulfur) modulates catalysis. Replacement of oxygen by sulfur diminishes the basicity of this structural atom as does the increasing electron withdrawal caused by fluoro substitution in the acetamido group. Based on relative velocities, Yamamoto has calculated a p* value of "0.6 for the effect of acetamido side chain substitution in the substrates for Taka-N-acetyl-P-glucosaminidase (15). Steric factors may also reduce the activity of the thiocarbonyl derivative, although such a large effect is most easily rationalized on the basis of a geometrically precise function for the carbonyl group, e.g. oxazoline formation. Although oxazoline formation requires an unfavorable C1"C2 transdiaxial conformation of the substrate, this conformation could be stabilized by the effective ion pairing provided by the ring structure (1). Furthermore, acid-catalyzed hydrolysis of derivatives of GlcNAc has been shown to proceed through an oxazoline intermediate (10). We propose, therefore, that the acetamido group provides anchimeric assistance to glycoside bond cleavage by electrostatic stabilization of the oxocarbonium ion or, perhaps, by facilitating intermediate oxazoline formation. Thus, the overall reaction may be described as a kinetically sequential process (Scheme I) which involves an SNI or SNi displacement in one or a series of steps within a single central complex.
In contrast to the effects of acetamido group alteration, aglycone substitution was without large effect ( p = +0.36).
Also, the kcat value of HzO/D20, 1.27, indicated that proton transfer(s) did not appear to be an important component of the rate-determining step. The interpretation of these data is conceptually simple in the context of the general scheme for acid-catalyzed acetal (ketal) hydrolysis outlined by Jencks The general acid-catalyzed hydrolysis of glycosides by glycosidases is described here as an SN, reaction, apparently proceeding via the path A + D generating the oxocarbonium ion. This model represents a chemically concerted process over a saddle point in the free energy contour linking the four ground states. For the contour to have such a saddle point, AGOAmust be similar to AGOA + C; if this condition is not fulfilled, the reaction will proceed exclusively through either state B or C (3). Based on the evidence discussed above for P-N-acetylglucosaminidase, this condition is achieved by the anchimeric assistance provided by the acetamido group. That is, the stabilization by this group of the oxocarbonium ion, state D, lowers the free energy of states B and D relative to C, and thus generates a concerted path (via a saddle point) between A and D. This model, by itself, does not deal with the fate of the oxocarbonium ion, i.e. whether or not it is trapped by HzO, an enzyme nucleophile, or the acetamido group.
The other data can be interpreted in terms of where this saddle point is relative to states B and C (and A and D). For example, the small effect of aglycone substitution ( p = +0.36) can be due to the opposing effects of electron withdrawal (basicity) on protonation (C) and bond cleavage (B). The substrate p-Np-1-T-PGlcNAc suffers a similar fate and consequently is not hydrolyzed. The sulfur is a poorer base than is oxygen and, thus, would not be protonated readily, destabilizing state C. A t the same time, the p-nitrothiophenolate moiety is a poorer leaving group that the p-nitrophenolate itself; thus, state B is also destabilized. Consequently, the free energy of the saddle point is raised and no hydrolysis is observed.
The solvent isotope effect at pH 4.6 was consistent with the results of Hammett studies. The latter indicate that the transition state looks somewhat more like state C than B. This suggestion is based on the fact that p for specific acid catalysis (through state C) is small and negative (-0.06 to -0.6), while for alkaline and spontaneous hydrolysis, where the phenolate anion is fully formed as in B, p values range from +2.5 to +2.8 (1, 10). Therefore, the proton transfer in the concerted process as indicated by the data appears to be nearly complete in the transition state. Under these circumstances, a small solvent isotope effect, such as is observed here, would be expected. The lack of a solvent isotope effect at pH 7.0 may be explained on the basis of a shift in the saddle point toward state B. This could be caused by the greater stability of the conjugate base of the aglycone (the nitrophen- olate anion) at higher pH which reduced the contribution of proton transfer in the rate-determining step. Under these conditions, proton transfer would appear to occur after this step. This implies that the p value for the hydrolysis of the glucosaminides studied would be pH dependent and would be larger at neutral and alkaline pH. This possible behavior was not explored. Such dependence might also indicate whether the values of p reflected only one of a number of partially ratedetermining steps (1).
The pH-rate studies suggested the presence in the free enzyme of a general acid group(s), pK, = 8.2. This ionizing function cold be either an imidazole, or a carboxyl with an unusual pK,; since amino acid analysis showed this enzyme to lack cysteine, this pK, cannot be associated with sulfhydryl group dissociation. An attractive alternative is to assign this pK, to a Tyr residue as Sinnott has suggested for P-galactosidase (6).
The kinetic pK, of this group appears to be 7.6 in the Michaelis complex. If, as has been suggested, proton transfer to the aglycone occurs early in the reaction, such a transfer is likely to be reflected in the Michaelis complex to some extent. The resulting group pK, in this complex might be taken as some weighted average of the enzymic pK,, 8.2, and that of the leaving group itself. It is tempting to relate the pH dependence of K, to this model, too. The ionization of an enzymic group (pK, = 2.95) enhanced substrate binding. This could be explained in terms of an electrostatic interaction between this conjugate base (perhaps a carboxylate group) and the protonated reaction center. Again, an analogy could be drawn with P-galactosidase (6).
There are certain factors which indicate a greater level of complexity in this enzyme's reaction mechanism than that apparent in the previous discussion. In particular is the fact that p-Np-Glc is hydrolyzed at -20% of the rate of the Nacetyl derivative (Table 11). This suggests that the acetamido group is not, in fact, involved in a rate-determining step. However, other data indicate that this substrate may bind at a different locus and be hydrolyzed by a somewhat different mechanism than for glucosaminides. This is suggested by: 1) the lack of inhibition by Glc of glucosaminide hydrolysis; and 2) the substrate inhibition noted for GlcNAc substrates. That GlcNAc and not Glc is a inhibitor of N-acetylglucosaminidase activity suggests that for the hydrolysis of glucosides a different site is utilized. This site may be nonproductive for glucosaminide hydrolysis, thus causing the substrate inhibition observed with these substrates. The presence of such a nonproductive binding mode may also perturb the protein ionization in the Michaelis complex indicated by the log kcat versus pH profie. Consequently, the difference in this pK, from that in the free enzyme may not be related to the reaction mechanism per se.
Whether or not acetamido group participation could be important in glucosaminidase hydrolysis of oligosaccharides was not established by this work. In lysozyme catalysis, the relative importance of geometric strain, electrostatic stabilization (by Asp-52), and acetamido participation may vary with the nature of the substrate. Levitt has suggested that distortion is not a major element in the activation of substrate oligosaccharides (50). In this regard, the K, value for GlcLNAc (0.41 m), which was not different from the K,,, value for p -Np-PGlcNAc, suggests that this inhibitor is not a true transition state analog for &glucosaminidase. On the other hand, the crystallographic evidence for lysozyme is that an oxazoline-type structure precludes effective stabilization of the nascent oxocarbonium ion by Asp-52 (51), or general base catalysis by Asp-52 of oxazoline formation (8). However, the model studies of Piszkiewicz and Bruice show clearly that the acetamido group can provide significant anchimeric assistance to glycoside hydrolysis and that this effect is not general-base assisted (8). Significantly, the pH uersus log kcat profiie for p-N-acetylglucosaminidase indicated a lack of general base catalysis in the enzymic reaction.
Furthermore, the binding of small substrates to lysozyme is a two-step process involving an isomerization of the ES complex (52). What the significance of this is to catalysis, or to differences in the binding of small sugars and oligosaccharides, is not clear. Inasmuch as the crystallographic studies have yet to elucidate the nature of acetamido group-enzyme interactions by direct measurement, the involvement of this group in the hydrolysis of, at least, some substrates cannot be ruled out. Importantly, substrates lacking this side chain are hydrolyzed by lysozyme as well (53), thus, its presence is not obligatory; however, as with P-N-acetylglucosaminidase, substrate effectiveness is enhanced in the acetamido derivatives.
Indeed, as shown in Table V, either the absence of or gross substitution in the acyl group effectively abolished binding of glucosaminides.
The significance of the acetamido group as a catalytic element becomes clearer upon comparing glycosaminidases (lysozyme, N-acetylglucosaminidase) with glycosidases (P-palactosidase). P-Galactosidase from Escherichia coli is thought to catalyze the formation of a galactosyl enzyme intermediate (6,54). However, based on kinetic isotope experiments and an analysis of the p-galactosidase-catalyzed isomerization of lactose to allolactose, Sinnott (6) has suggested that this galactosyl enzyme is a mixture of two species. One is the galactosyl enzyme, i.e. one in which a covalent bond exists between galactose and an enzyme nucleophile, while the other species is an ion pair consisting of the nucleophile and oxocarbonium ion. In the presence of good nucleophiles, e.g. methanol, the latter species can be trapped before it collapses into a covalent bond. In a sense, this behavior may be similar to that of p-Nacetylglucosaminidase in which the role of the putative enzyme anion (nucleophile) might be played by the substrate acetamido group. However, as in the ,&galactosidase reaction, the degree to which this potentially nucleophilic group participates in covalent bond formation remains an open question.