Biochemistry 2003, 42, 7195-7204
7195
Mechanism, Mutagenesis, and Chemical Rescue of a β-Mannosidase from Cellulomonas fimi† David L. Zechel,‡ Stephen P. Reid,‡ Dominik Stoll,§ Oyekanmi Nashiru,§ R. Antony J. Warren,§ and Stephen G. Withers*,‡ PENCE and Departments of Chemistry and Microbiology, UniVersity of British Columbia, VancouVer, British Columbia, Canada V6T 1Z1 ReceiVed February 27, 2003; ReVised Manuscript ReceiVed April 28, 2003
The chemical mechanism of a retaining β-mannosidase from Cellulomonas fimi has been characterized through steady-state kinetic analyses with a range of substrates, coupled with chemical rescue studies on both the wild-type enzyme and mutants in which active site carboxyl groups have been replaced. Studies with a series of aryl β-mannosides of vastly different reactivities (pKalg ) 4-10) allowed kinetic isolation of the glycosylation and deglycosylation steps. Substrate inhibition was observed for all but the least reactive of these substrates. Brønsted analysis of kcat revealed a downward breaking plot (βlg ) -0.54 ( 0.05) that is consistent with a change in rate-determining step (glycosylation to deglycosylation), and this was confirmed by partitioning studies with ethylene glycol. The pH dependence of kcat/Km follows an apparent single ionization of a group of pKa ) 7.65 that must be protonated for catalysis. The tentative assignment of E429 as the acid-base catalyst of Man2A on the basis of sequence alignments with other family 2 glycosidases was confirmed by the increased turnover rate observed for the mutant E429A in the presence of azide and fluoride, leading to the production of β-mannosyl azide and β-mannosyl fluoride, respectively. A pH-dependent chemical rescue of E429A activity is also observed with citrate. Substantial oxocarbenium ion character at the transition state was demonstrated by the R-deuterium kinetic isotope effect for Man2A E429A of R-D(V) ) 1.12 ( 0.01. Surprisingly, this isotope effect was substantially greater in the presence of azide (R-D(V) ) 1.166 ( 0.009). Likely involvement of acid/base catalysis was revealed by the pH dependence of kcat for Man2A E429A, which follows a bell-shaped profile described by pKa values of 6.1 and 8.4, substantially different from that of the wild-type enzyme. The glycosidic bond cleaving activity of Man2A E519A and E519S nucleophile mutants is restored with azide and fluoride and appears to correlate with the corresponding “glycosynthase” activities. The contribution of the substrate 2-hydroxyl to stabilization of the Man2A glycosylation transition state (∆∆Gq ) 5.1 kcal mol-1) was probed using a 2-deoxymannose substrate. This value, surprisingly, is comparable to that found from equivalent studies with β-glucosidases despite the geometric differences at C-2 and the importance of hydrogen bonding at that position. Modes of stabilizing the mannosidase transition state are discussed. ABSTRACT:
A significant component of enzymatic catalysis is derived from noncovalent enzyme-substrate interactions that are optimized in the transition state. Therefore, to understand enzyme catalysis, it is important to quantify contributions from this source. Good enzyme systems for such studies are glycosidases, which act upon polyhydroxylated substrates. Interactions with the hydroxyls must be important and can be probed, individually, by measuring kinetic parameters for a series of substrates that have been, individually, deoxygenated at each position. Of particular interest in this regard are interactions with the hydroxyl group adjacent to the anomeric carbon (the 2-hydroxyl), which appear to be universally strong in retaining β-glycosidases that act upon † This work was supported by the Protein Engineering Network of Centres of Excellence of Canada (PENCE) and the Natural Sciences and Engineering Research Council of Canada (NSERC). D.L.Z. thanks the Izaak Walton Killam Foundation and NSERC for fellowships. * Corresponding author. E-mail:
[email protected]. ‡ PENCE and Department of Chemistry, University of British Columbia. § PENCE and Department of Microbiology, University of British Columbia.
glycosides with equatorial 2-hydroxyls (e.g., β-glucosidases, β-galactosidases, cellulases). The 2-hydroxyl typically contributes 5-10 kcal mol-1 to catalysis in these enzymes, whereas the other sugar ring hydroxyls contribute 1-2 kcal mol-1 (1-4). Furthermore, close interaction (∼2.5 Å) of the nucleophile carboxyl oxygen and the 2-hydroxyl (or fluorine) has been observed by crystallographic analysis of trapped covalent intermediates (5, 6). From these observations it is believed that a strong hydrogen bond forms between the 2-hydroxyl and the nonreacting oxygen of the nucleophilic carboxylate in the oxocarbenium ion transition state (Figure 1a). The close approach of the 2-hydroxyl and nucleophile would be encouraged in such a transition state as the pyranose ring adopts a flattened half-chair (4H3) conformation and the nucleophile begins to attack anomeric carbon. Likewise, the development of positive charge at the anomeric carbon can be expected to transiently acidify the 2-hydroxyl (pKa ∼12), making it a better hydrogen bond donor to the nucleophile (pKa ∼4-5). The importance of the 2-hydroxyl is underscored by the potency of novel transition state analogues that exploit this interaction (7, 8).
10.1021/bi034329j CCC: $25.00 © 2003 American Chemical Society Published on Web 05/20/2003
7196 Biochemistry, Vol. 42, No. 23, 2003
Zechel et al. consumption of the substrate. Substrate concentrations were varied typically from one-fifth to five times the final Km value, whenever possible, for Michaelian kinetics. Higher substrate concentrations were assayed when substrate inhibition was observed. Steady-state kinetic parameters were derived by fits of the data to the Michaelis-Menten equation or the substrate inhibition equation (eq 1) using GraFit (12).
FIGURE 1: Comparison of pyranoside ring conformations and interaction of the catalytic nucleophile with the substrate 2-hydroxyl in a hypothetical β-glucosidase transition state (a) and β-mannosidase transition state (b).
It would appear that the opposite stereochemical configuration at C-2 in a β-mannoside substrate would exclude β-mannosidases and β-mannanases from utilizing the catalytic nucleophile in this fashion. Nevertheless, a recent crystallographic description of a β-mannanase reaction coordinate suggests that a unique B2,5 mannoside ring conformation may in fact allow close approach of the mannose 2-hydroxyl to the nucleophile (Figure 1b) (9). These observations have prompted us to explore the reaction mechanism of a family 2 β-mannosidase from Cellulomonas fimi (Man2A)1 (10). The active site nucleophile has previously been identified as E519 by labeling with a mechanismbased inactivator (11), and sequence homology with other family 2 glycosidases suggests that E429 is the general acidbase catalyst (10). In this study the reaction mechanism of Man2A has been characterized by Brønsted correlations, kinetic isotope effects, site-directed mutagenesis, and chemical rescue with fluoride and azide. The contribution of the 2-hydroxyl to catalysis has also been examined. MATERIALS AND METHODS General. The synthesis and characterization of the aryl mannoside substrates used in this study will be published elsewhere. Syntheses of other substrates are provided in the Supplementary Information. The cloning, expression, and purification of C. fimi β-mannosidase have been described previously (10). Steady-State Kinetic Analyses of Man2A and Man2AE429A. Steady-state kinetic analyses were performed on Unicam UV-4 or Unicam 8700 UV-vis spectrometers equipped with thermoequilibrated cell blocks. All reactions were performed at 25 °C, pH 7, in acrylic cuvettes unless otherwise noted. A typical reaction cuvette contained substrate, 50 mM sodium phosphate (pH 7), and 1 mg/mL BSA in a total volume of 750 µL. After the cuvette was preequilibrated at 25 °C, a small aliquot of appropriately diluted Man2A (5-10 µL in 1 mg/mL BSA) was added and mixed briefly. Release of the phenol was monitored continuously at the appropriate wavelength. Initial rates (Vi) were determined from linear fits of these plots in regions corresponding to 5-15% 1
Abbreviations: Man2A, Cellulomonas fimi β-mannosidase; BSA, bovine serum albumin; PE, petroleum ether (30-60 °C); EtOAc, ethyl acetate; MES, 2-(N-morpholino)ethanesulfonic acid; AMPSO, 3-[(1,1dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulfonic acid; R-DKIE, R deuterium kinetic isotope effect; R-D(V), R deuterium kinetic isotope effect on Vmax; 2,4DNPMan, 2,4-dinitrophenyl β-mannoside; 2,5DNPMan, 2,5-dinitrophenyl β-mannoside; 4NPMan, 4-nitrophenyl β-mannoside; 4NPC, 4-nitrophenyl β-cellobioside; PhMan, phenyl β-mannoside; R-manF, R-mannosyl fluoride.
Vi kcat[S] ) Et K + [S] + [S]2/K m i
(1)
The wavelengths and extinction coefficients used for assays of aryl β-mannosides and calculation of kcat are the same as those reported previously (13). The substrate 4-methylumbelliferyl β-mannoside was assayed at 365 nm (∆ ) 5136 M-1‚cm-1) (14). Assays with 4-chlorophenyl and phenyl β-mannoside were performed in quartz cuvettes (500 µL total volume). Stock solutions of 2,4DNPMan were prepared immediately before use in an acidic buffer (pH 5-6), and concentrated aliquots (5-50 µL) of the substrate were used to initiate reaction with Man2A in pH 7 buffer. In neutral solutions, 2,4DNPMan slowly rearranges to a new species, presumably through a migration of the 2,4-dinitrophenyl group to the 2-hydroxyl. This migration is very rapid above pH 7. The rearranged species can be observed by TLC (UV and acid charring): Rf for 2,4DNPMan (7:2:1 EtOAc/MeOH/H2O) ) 0.65; Rf for the migration product ) 0.74. Concentrations of Man2A and Man2A E429A stock solutions were determined by absorbance at 280 nm using the extinction coefficient 210000 M-1‚cm-1 (calculated from the amino acid sequence) (15). pH-Rate Studies. The pH dependence of kcat/Km for wildtype Man2A was determined by generating first-order rate curves at low substrate concentrations ([S] , Km). Cuvettes were charged with the appropriate buffer, 1 mg/mL BSA, and a concentration of substrate that was one-tenth or less of the corresponding Km value (9.3 µM 4NPMan, 205 µM PhMan). After equilibration at 25 °C an aliquot of Man2A (10 µL) was added to afford a final concentration of enzyme sufficient to generate a first-order rate of phenol release within 10-20 min (0.12 µM Man2A for 4NPMan, 2.1 µM Man2A for PhMan). Phenol release was monitored continuously on a Unicam UV-4 spectrometer (λ ) 400 nm for 4NPMan, 280 nm for PhMan) until a limiting absorbance was reached. The first-order rate curves generated at each pH value were fit using GraFit to the first-order rate equation (eq 2) to determine the rate constant k, which corresponds
At ) A∞(1 - e-kt) + offset
(2)
to Vmax/Km. The following buffers (50 mM) were used for the following pH ranges: citric acid, pH 6-6.8; sodium phosphate, pH 6.8-8.4. Man2A retained at least 90% activity in these buffers at each pH over a period of 30 min. Rates were obtained at overlapping buffer pH values, and buffer concentrations were varied to reveal buffer effects, none of which were observed. The kcat/Km values determined at each pH were fit with a function describing a single ionization
C. fimi β-Mannosidase: Mechanistic Studies
Biochemistry, Vol. 42, No. 23, 2003 7197
using GraFit (eq 3).
kcat/Km )
limit1 + limit2 × 10pH-pKa 10pH-pKa + 1
(3)
In the case of Man2A E429A the pH dependence of kcat/ Km for the reaction with 2,5DNPMan could not be determined owing to the very low Km value (6 mM) modest inhibition was observed (Figure 2b). Table 1 lists the kinetic parameters determined for the reaction of Man2A with aryl β-mannosides using eq 1.
A Brønsted plot of log kcat versus the pKa of the leaving group reveals an apparent downward break in the correlation occurring at pKa ∼7 (Figure 3a). The slope of the leaving group-dependent section of the Brønsted plot yields a βlg value of -0.54 ( 0.05. The corresponding Brønsted plot for kcat/Km also shows a nonlinear, downward breaking correlation (Figure 3b). A complementary experiment to the Brønsted analysis above is to examine the effect of an external neutral nucleophile on the steady-state rate (16). The rate of reaction of Man2A with the reactive substrate 2,5DNPMan increased linearly with the addition of ethylene glycol (Figure 4a). In contrast, ethylene glycol had essentially no effect on the rate of reaction with PhMan (Figure 4b). The slope of the plot for the 2,5DNPMan data yields an apparent second-order -1 mM-1 for the rate constant kapp nuc ) 0.0203 ((0.0005) s reaction with ethylene glycol, whereas kapp nuc ) 0.0009 ((0.0003) s-1 mM-1 for the reaction with PhMan. A similar apparent increase in reaction rate has been observed for the analogous reaction with Escherichia coli (lacZ) β-galactosidase (17, 18). pH-Rate Dependence of Man2A. The pH dependence of kcat/Km for the reaction of Man2A with two different substrates was examined (Figure 5). In each case kcat/Km shows an apparent dependence on a single ionization event. Essentially identical pKa values were observed with 4NPMan (pKa ) 7.6 ( 0.1) and PhMan (pKa ) 7.7 ( 0.1). The pH behavior of Man2A below pH 5 could not be examined due to instability of the enzyme in this pH range. Identification of the General Acid-Base Catalyst in Man2A and Chemical Rescue. Alignment of the sequence of Man2A with those of other family 2 glycosidases, including two for which three-dimensional structures are available, indicated that E429 was most likely the acidbase catalyst (Figure 6). The corresponding mutant E429A was found to hydrolyze only the most reactive of substrates (2,4DNPMan and 2,5DNPMan). Kinetic parameters of kcat ) 12 min-1 and Km < 1 µM were determined for the reaction with 2,5DNPMan. Saturating kinetic behavior was observed for the reaction of 2,5DNPMan with Man2A E429A, although at high substrate concentrations (3-4 mM) very weak substrate inhibition was perceptible (data not shown). This contrasts sharply with the severe substrate inhibition observed with the wild-type enzyme (Figure 2a). The dependence of kcat on pH for the reaction of 2,5DNPMan with Man2A E429A was described by two ionizations
FIGURE 4: Dependence of the reaction rate of Man2A wt with saturating substrate and added ethylene glycol: (a) 2,5-dinitrophenyl β-mannoside (3.8 mM); (b) phenyl β-mannoside (10.6 mM). Lines are linear regressions of the data.
C. fimi β-Mannosidase: Mechanistic Studies
Biochemistry, Vol. 42, No. 23, 2003 7199
Table 2: Kinetic Parameters for Glycosylation Reactions Catalyzed by Man2A Nucleophile Mutants
Man2A wta +2 M KFa Man2A E519A a +1 M azidea +2 M KFa +R-manF (20 mM 4NPC)b +4NPC (50 mM R-manF)b Man2A E519Sa +1 M azidea +2 M KFa +R-manF (29 mM 4NPC)b +4NPC (50 mM R-manF)b Man2A E429Ac +0.1 M azidec +1 M KFc c
kcat (min-1)
Km (mM)
27000 7900
