Substrate recognition by three family 13 yeast a-glucosidases Evaluation of deoxygenated and conformationally biased isomaltosides Torben P. Frandsen 1, *, Monica M. Palcic 2 and Birte Svensson 1 1 Department of Chemistry, Carlsberg Laboratory, Copenhagen Valby, Denmark; 2 Department of Chemistry, University of Alberta, Edmonton, Canada Important hydrogen bonding interactions between s ubstrate OH-groups i n yeas t a-glucosidases a nd oligo-1,6-glucosidase from glycoside hydrolase family 13 have been identi®ed by measuring the rates of hydrolysis of methyl a-isomaltoside and its seven monodeoxygenated analogs. The transition - state stabilization energy, DDGà, contributed by the indi- vidual OH-groups was calculated from the activities for the parent and the deoxy analogs, respectively, according to DDGà  ±RT ln[(V max /K m ) analog /(V max /K m ) parent ]. This analysis of the e nergetics gave DDGà values for all three enzymes r anging from 16.1 to 24 .0 kJámol )1 for O H-2¢,-3¢, -4¢,and-6¢, i.e. t he OH-groups of the nonreducing s ugar ring. These OH-groups interact with enzyme via charged hydro- gen bonds. In contrast, OH-2 and -3 of the reducing sugar contribute to transition-state stabilization, by 5.8 and 4.1 kJámol )1 , respectively, suggesting that these groups participate in neutral hydrogen bonds. The OH-4 group is found to be unimportant in this respect and very little or no contribution is indicated for all OH-groups of the reducing- end ring of the two a-glucosidases, probably re¯ecting their exposure t o bulk solvent. T he stereoch emical course of hydrolysis by these three members of the retaining family 13 was con®rmed by directly monitoring isomaltose hydrolysis using 1 H NMR spectroscopy. Kinetic analysis of the hydrolysis of methyl 6-S-ethyl-a-isomaltoside and its 6-R- diastereoisomer indicates that a-glucosidase has 200-fold higher speci®city for the S-isomer. Substrate molecular rec- ognition by these a-glucosidases are compared to earlier ®ndings for the inverting, exo-acting glucoamylase from Aspergillus niger and a retaining a-glucosidase of glycoside hydrolase family 31, respectively. Keywords: protein-carbohydrate interaction; NMR; glyco- sidase mechanism; substrate analogs; molecular recognition. Strong intermolecular hydrogen bonds are very important in speci®city of enzymes and o ther proteins that metabolize or bind carbohydrates [1±6]. Substrate analogs such as deoxygenated sugars, facilitate identi®cation of critical contacts and enable quanti®cation of the energetics of the protein±carbohydrate binding at the level of individual interacting sugar OH-groups and functional atoms or groups in the protein [4,7±11]. Alternatively, site-speci®c mutants of a protein are useful in evaluation of speci®c protein±carbohydrate i nteractions and further insight has been gained by combining mutant enzymes and analogs [7,9,10]. The binding energy contributed by substrate OH-groups has been determined for only a few carbohy- drate active enzymes. Of these, the starch hydrolase glucoamylase from Aspe rgillus n iger has been the most intensively examined [7,9±13]. Three-dimensional structures of protein±carbohydrate complexes can guide and support protein engineering and molecular r ecognition experiments. For family 13 glycoside hydrolases, there are no c rystal structures for a-glucosidases; however, t he structure o f free Bacillus oligo-1,6-glucosidase has been solved [14]. Furthermore, only a few a-glucosidases are produced by heterologous gene expression, which is a prerequisite for structure±function relationship i nvestiga- tions by site-directed mutagenesis [15±21]. While the yeast genome is known and thus the primary structures of its a-glucosidases, the sequenced strain of Saccharomyces cerevisiae is not necessarily identical to the baker's yeast used as a source of enzymes in the present study and sequences have not been reported for brewer's yeast enzymes. In view of this limited information, use of synthetic substrate analogs is particularly attractive for gaining knowledge of the nature and strength of substrate± a-glucosidase interactions. Thus using deoxy-analogs key polar groups in maltose were i denti®ed for high pI barley a-glucosidase of glycoside hydrolase family 31 to be OH-4¢ and -6¢ with minor contributions for OH-3¢,-2¢,and-3 [13, 22]. Yeast a-glucosidase and oligo-1,6-glucosidase are exo- acting glycoside hydrolases catalyzing release of a- D -glucose from nonreducing ends of various a-linked substrates. The enzymes are further subclassi®ed into type I, hydrolysing heterogeneous substrates like aryl glucosides and sucrose more ef®ciently than maltose; type II being highly active on maltose and isomaltose but of low a ctivity toward aryl glucosides; and type III resembling type II, but hydrolysing Correspondence to B. Svensson, Department of Chemistry, Carlsberg Laboratory, DK-2500 Copenhagen Valby, Denmark; Fax: + 45 33 27 47 08; Tel.: + 45 33 27 53 45; E-mail: bis@crc.dk Enzymes: a-glucosidase (a- D -glucoside glucohydrolase, EC 3.2.1.20); oligo-1,6-glucosidase (dextrin 6-a-glucanohydrolase, EC 3.2.1.10); glucoamylase (a- D -glucan glucohydrolase, EC 3.2.1.3). *Present address: Pantheco, Bùge Alle  , DK 2970 Hùrsholm Denmark. Dedication: this paper is de dicated to Prof. Joachim Thiem on the occasion of his 60 th birthday. (Received 12 O ctober 2001, revised 26 November 2001, accepted 30 November 2001) Eur. J. Biochem. 269, 728±734 (2002) Ó FEBS 2002 di- a nd oligo-saccharides and starch at comparable rates [23,24]. The sequence classi®es a-glucosidases in glycoside hydrolase families 13 and 31 [25±27]. Yeast a-glucosidases and oligo-1,6-glucosidase belong to family 13 and are of type I that prefers p-nitrophenyl-a- D -glucopyranoside [28]. Glycoside hydrolase family 13 (or Ôthe a-amylase familyÕ) currently comprises 28 speci®cities of amylolytic and related enzymes. Several crystal structures of enzyme-inhibitor complexes highlight active sites created by b ® a segments in c atalytic (b/a) 8 barrel domains (reviewed in [29±31]). Because no ligand complex is available of oligo-1,6- glucosidase, the only structure-determined exo-acting a-glucosidase [14], s ide-chains partic ipating in s ubstrate binding and catalysis are solely identi®ed by sequence comparison. Clearly a-glucosidases lack the sequence motif in b ® a loop 4 of family 13 [30] containing residues binding substrate at s ubsite +2 (nomenclature as in [32]) in a-amylases, cyclodextrin glycosyltransferases, and related enzymes [30,33±37]. In this study, seven monodeoxygenated isomaltosides are used to map substrate OH-groups required by yeast a-glucosidases and oligo-1,6-glucosidase in hydrolysis of the a-1,6-glucosidic bond. The energy contributed by each OH-group for transition-state stabilization re¯ects the strength of a speci®c protein±carbohydrate c ontact a nd energy pro®les for the a-glucosidases a nd oligo-1,6-glucosi- dase are compared. 1 H-NMR spectroscopy was used to con®rm that all enzymes hydrolyse isomaltose with reten- tion of anomeric con®guration characteristic of family 13 (reviewedin[38]). The exo-acting glucoamylase s imilarly to the a-glucosid- ases catalyses the releases of glucose from the nonreducing ends of substrates, but with inversion of the anomeric con®guration [39]. While glucoamylase prefers the R-isomer of isomaltose diastereoisomeric analogs [40], a-glucosidase in the p resent study selects methyl 6-S-ethyl-a-isomaltoside in preference to the R-isomer. Molecular recognition of isomaltosides is more similar for the yeast a-glucosidase and oligo-1,6-glucosidase of glycoside of hydrolase family 13 when compared to that of glucoamylase of glycoside hydrolase family 15 or of a type II a-glucosidase from the retaining glycoside hydrolase family 31 [9,10,40,41]. MATERIALS AND METHODS Enzymes and substrates Oligo-1,6-glucosidase from baker's yeast (EC 3.2.1.10; Lot no. 23H8080), and a-glucosidases from brewer's (EC 3.2.1.20; Type VI; Lot no. 21F8 105) and b aker's (EC 3.2.1.20; Type I; Lot no. 122H8000) yeast were obtained from Sigma. After dissolution in 50 m M phosphate pH 6.8 (a-glucosidases) or 50 m M sodium maleate pH 6.8 (oligo-1,6-glucosidase) followed by extensive dialysis at 4 °C against these buffers, the different enzymes (oligo-1,6- glucosidase, 30 UámL )1 ; brewer's yeast a-glucosidase, 200 UámL )1 ; baker's yeast a-glucosidase, 61 UámL )1 )were used without further puri®cation in the kinetic and stereo- chemical studies. One unit i s de®ned as t he amount of enzyme required to liberate 1 lmol of glucose from p-nitrophenyl a- D -glucoside (Sigma) p er min at 30 °C. The synthesized methyl a-isomaltoside, seven monodeoxy- genated methyl a-isomaltosides [42], methyl 6-R-C-ethyl- and methyl 6-S-C-ethyl-a-isomaltoside [41] were generous gifts of U. S pohr and the late R. U. Lemieux, University of Alberta, Edmonton, Canada. Enzyme assays a-Glucosidase activity was d etermined at 30 °Cin0.1 M sodium maleate, pH 6.8 (oligo-1,6-glucosidase) or 50 m M phosphate, pH 6 .8 (a-glucosidases). Glucose [7,10,40,41] was a nalysed f or analogs at the reducing end ring (reaction volume 100 lL), aliquots (15 lL) being transferred at regular time intervals to microtiter plate wells already containing quench solution (200 lL1 M Tris, pH 7.6, 5UámL )1 glucose oxidase (A. niger), 1 UámL )1 peroxidase (horseradish), and 0.21 mgámL )1 o-dianisidine). Absor- bances were read at 450 nm after 1 h incubation at room temperature using a microtiter plate reader (C eres UV900Hdi, Bio-Tek), and quanti®ed using D -glucose as a standard [22,40]. Deoxygenated glucose analogs were ana- lysed essentially as described [10,40,41] with substrate analogs at t he nonreducing e nd sugar ( reaction volume 400 lL) aliquots (100 lL) were transferred to quench buffer containing 60 UámL )1 glucose oxidase, 1 UámL )1 peroxi- dase, and 0.1 mgámL )1 o-dianisidine, and the absorbances were read at 450 nm after 4 h incubation at room temperature, and quanti®ed using the relevant deoxygenated D -glucose as standard. The a-glucosidase catalyzed hydro- lysis was initiated by a ddition of 0.1±91 U enzyme. The limited amounts of deoxygenated analogs available allowed only determination of second-order rate constants, V max /K m (s )1 áU )1 )  v o /E o S o ,wherev o is the initial rate of h ydro- lysis, S o the initial substrate concentration, and E o the amount of enzyme in U. Two S o concentrations of around 0.1 ´ K m were used to ensure that substrate hydrolysis w as linear w ith t ime. The increase in activation e nergy due to substrate d eoxygenation was c alculated by DDGà  ±RTln[(V max /K m ) a /(V max /K m ) b ][43],whereareferstoana- log and b to parent substrate. For the two diastereoisomers, V max and K m were determined by ®tting initial rates at eight different substrate concen trations from 0.1 ´ K m to 4 ´ K m to the Michealis±Menten equation essentially as described previously [40]. Reaction stereochemistry Lyophilized enzymes w ere redissolved in 0.1 M sodium phosphate pH 6.8 in D 2 O and the stereochemistry of isomaltose hydrolys is was determined by 1 HNMRat 310 K using a Bruker AMX-600 spectrometer operated at 600 MHz. After recording the substrate spectrum of 100 m M isomaltose (in 600 lL0.1 M phosphate, pH 6.8, in D 2 O), enzyme was added (oligo-1,6-glucosidase, 40 U; baker's, 135 U and brewer's yeast a-glucosidase, 140 U) and reactions monitored by recording spectra at regular intervals. RESULTS AND DISCUSSION Energetics of deoxy isomaltoside hydrolysis V max /K m values for h ydrolysis of m ethyl a-isomaltoside are comparable for the three enzymes, the a-glucosidases from Ó FEBS 2002 Substrate recognition in yeast a-glucosidases (Eur. J. Biochem. 269) 729 brewer's and baker's yeast showing 31 and 164% of the activity of the oligo-1,6-glucosidase, respectively (Table 1). Furthermore, the activity of the three enzymes was reduced by roughly the same extent, i.e. 440±3400-, 560±2900-, and 1350±8800-fold by substrate deoxygenation at OH-2¢,-3¢, -4¢,or-6¢ (Table 1). The losses in activity compared to the parent substrate for all enzymes were smallest for t he 6¢-deoxy ana log and largest for the 2¢-deoxy analog, while intermediary losses in activity for 3¢-and4¢-deoxy analogs did show small variations among the enzymes (Table 1). For the two a-glucosidases, deoxygenation at the reducing end ring of the substrate had no effect or a very minor effect, the activity varying relat ive to the parent substrate b y factors of 0.84±1.4 a nd 0.42±1.0 for the enzymes f rom brewer's and baker's yeast, respectively. In contrast, for oligo-1,6-glucosidase, the deoxy-2, -3, and -4 analogs showed ninefold, ®vefold, and no reduction in V max /K m , respective ly (Ta ble 1). The DDGà calculated from the V max /K m values deter- mined for a given analog and the parent substrate, respectively, indicated the energy contributed to transition- state stabilization by corresponding the OH-group. Because DDGà for the four deoxy-analogs at t he nonreducing sugar ring, that binds to the enzymes at subsite )1, was in the range 16.1±24.0 kJámol )1 for the three enzymes (Table 1), the r emoval of one of the OH-groups from this ring dramatically affected substrate hydrolysis. These OH-groups can therefore be considered key polar groups and m ost likely interact w ith c harged residues on t he proteins [44] (Fig. 1 ). At the reducing end ring, however, DDGà values of 4±6 kJámol )1 for oligo-1,6-glucosidase (Table 1) were obtained by replacement of the OH-2 and -3 groups, respectively, suggesting that these OH-groups participate in neutral hydrogen bonds with the enzyme (Fig. 1). The OH-4 did not seem important in substrate binding and hydrolysis. The three-dimensional structure of oligo-1,6-glucosidase from Bacillus cereus [14], is currently the only available structure of a ny type of a-glucosidases. This enzyme has an N-terminal (b/a) 8 barrel common to glycoside hydrolase family 13 [30,31], a domain B that protrudes from the barrel b strand 3, and a C-terminal Greek k ey motif. Moreover several extra-barrel secondary structure elements occur in the segments that connect the b strands to the a helices of the (b/a) 8 barrel fold [ 14]. The catalytic site is located at the bottom of a cleft between domain B and several of the b ® a connecting segments [14,30]. The molecular recog- nition of isomaltose analogs described above indicate very strong interaction of the nonreducing substrate ring at the enzyme subsite )1, most probably with charged side chains, as a major driving force for stabilization o f the enzyme± substrate transition-state. Several of the side-chains inter- acting at subsite )1 will belong to the consensus sequence motifs containing catalytic acids, t ransition-state stabilizing histidines, and structurally important arginine and aspartate side chains [30]. While protein±substrate contacts at subsite )1provide major binding energy, t he distribution and strength of intermolecular hydrogen bonds involving the aglycon moiety and subsite +1, as well as subsites beyond subsite +1 in type III a-glucosidases, exhibit substrate speci®city variation among the a-glucosidases. The yeast a-glucosid- ases as reported h ere only show protein±carbohydrate hydrogen bonding involving subsite )1,andnosugar OH-groups associated stabilization energy was critical for accommodation at subsite +1. As shown in Table 2, these a-glucosidases that do not require hydrogen bonding to the Table 1. Speci®city constants and DDGà a (kJámol )1 )fora-glucosidase catalyzed hydrolysis of methyl a-isomaltoside and a series of mono-deoxy- genated analogs. Oligo-1,6-glucosidase b a-glucosidase (brewer's yeast) c a-Glucosidase (baker's yeast) d V max /K m (s )1 áU )1 ) DDGà V max /K m (s )1 áU )1 ) DDGà V max /K m (s )1 áU )1 ) DDGà Methyl-a-isomaltoside 1.4 ´ 10 )4  0.8 ´ 10 )5e ± 4.4 ´ 10 )5  2.9 ´ 10 )6 ± 2.3 ´ 10 )4  3.2 ´ 10 )6 ± 2-Deoxy-methyl-a-isomaltoside 1.6 ´ 10 )5  0.9 ´ 10 )6 5.8 3.7 ´ 10 )5  4.9 ´ 10 )6 0.5 9.7 ´ 10 )5  5.8 ´ 10 )6 2.3 3-Deoxy-methyl-a-isomaltoside 2.9 ´ 10 )5  0.6 ´ 10 )6 4.1 6.0 ´ 10 )5  1.4 ´ 10 )5 )0.8 1.3 ´ 10 )4  1.3 ´ 10 )5 15 4-Deoxy-methyl-a-isomaltoside 1.4 ´ 10 )4  1.1 ´ 10 )5 ± 5.7 ´ 10 )5  0.8 ´ 10 )6 )0.7 2.4 ´ 10 )4  2.0 ´ 10 )6 )0.1 2¢-Deoxy-methyl-a-isomaltoside 4.1 ´ 10 )8  5.5 ´ 10 )9 21.5 1.5 ´ 10 )8  1.2 ´ 10 )9 21.2 2.6 ´ 10 )8  3.0 ´ 10 )9 24.0 3¢-Deoxy-methyl-a-isomaltoside 1.1 ´ 10 )7  6.5 ´ 10 )9 18.9 2.7 ´ 10 )8  5.3 ´ 10 )9 19.6 2.7 ´ 10 )8  3.1 ´ 10 )9 23.9 4¢-Deoxy-methyl-a-isomaltoside 1.0 ´ 10 )7  1.1 ´ 10 )8 19.1 1.5 ´ 10 )8  4.4 ´ 10 )10 21.2 3.2 ´ 10 )8  2.8 ´ 10 )9 23.5 6¢-Deoxy-methyl-a-isomaltoside 3.2 ´ 10 )7  2.1 ´ 10 )8 16.1 7.9 ´ 10 )8  4.7 ´ 10 )9 16.7 1.4 ´ 10 )7  1.5 ´ 10 )8 19.6 a DDGà  )RT ln[(V max /K m ) a /(V max /K m ) b ] [43], where a and b refer to analog and parent substrate, respectively; b At 30 °C using 0.1 M sodium maleate, pH 6.8; c At 30 °C using 50 m M phosphate, pH 6.; d At 30 °C using 50 m M phosphate, pH 6.8; e standard deviation. Fig. 1. Schematic representation of proposed intermolecular hydrogen bond interactions between isomaltose and a-glucosidases from baker's and brewer's yeast and from oligo-1,6-glucosid ase from baker's yeast. a , only f or oligo-1,6-glucosidase. Invariant glycoside hydrolase family 13 side chain candidates of interaction with the four nonreducing substrate ring O H-groups are described in detail in a recent review [30]. 730 T. P. Frandsen et al. (Eur. J. Biochem. 269) Ó FEBS 2002 substrate aglycon also have much higher activity for p-nitrophenyl-a- D -glucopyranoside, which lacks hydrogen bonding groups in the aglycon, than for isomaltose. Due t o effects on b oth k cat and K m yeast a-glucosidase thus has 4500-fold lower k cat /K m for i somaltose than for p-nitrophe- nyl-a- D -glucopyranoside, p-nitrophenol being also a better leaving group than g lucose. Structural e lements of t he nonsugar aglycon, however, were not explored. It is conceivable, however, that such speci®city e xists and could be investigated using a series of synthetic substrates. In contrast, the activity of oligo-1,6-glucosidase signi®cantly depends on aglycon interactions at subsite +1 via neutral hydrogen bonds with glucose OH-2 and -3 (Table 1). That such protein interactions with sugar OH-groups are impor- tant for this enzyme i n contrast to the two a-glucosidases is also emphasized by the 225-fold difference in the value of the relative speci®city p-nitrophenyl a-glucoside/isomaltose, (k cat /K m )/(k cat /K m ), being 4500 for the brewer's a-g lucosi- dase (which was chosen because it has the smallest requirement for OH-groups at subsite +1; see Table 1), and 20 for oligo-1,6-glucosidase (Table 2). Moreover, the 30-fold more favorable speci®city constant (k cat /K m )for isomaltose for the oligo-1,6-glucosidase over the a-glucosi- dase indeed re¯ects t he gen uine speci®city of the former enzyme for exo-action on the a-1,6-linkage. Remarkably, barley a-gluco sidase of glycoside hydrolase family 31 has a completely different pattern for hydrolysis of monodeoxy maltoside analogs which indicated strong protein±substrate interactions at OH-4¢ and -6¢ and weaker, probably neutral hydrogen-bonds with maltose OH-2¢,-3¢, and -3 [22]. An even stronger requirement for protein± isomaltose aglycon interactions was found in isomaltose hydrolysis by glucoamylase, which depended on enzyme± substrate transition-state interactions with OH-4 and -3 of an energy of 16.5 and 8.6 kJámol )1 , respectively (Table 2; [7]). Glucoamylase thus has only sixfold lower k cat /K m for isomaltose than for p-nitrophenyl-a- D -glucopyranoside (Table 2). The substrate speci®city d ifferences and varia- tions in aglycon-protein contacts with the two a-glucosid- ases, the oligo-1,6-glucosidase, and glucoamylase emphasize that these enzymes display different geometry for the binding interactions with polar groups of substrates at subsite +1. This will be investigated further in a study of the diastereo- isomer speci®city of isomaltoside h ydrolysis (see below). Catalytic mechanism One f eature of the d isposition o f substrate relative to enzyme during the various steps of the catalytic events directly relates to the mechanism of catalysis being funda- mentally different for retaining and in verting enzymes [38]. The stereochemistry of isomaltose hydrolysis by yeast oligo- 1,6-glucosidase and a-glucosidases was con®rmed to involve retention of the substrate anomeric con®guration in the product. This is illustrated for baker's yeast a-glucosidase which shows 1 H NMR spectra of isomaltose before (Fig. 2A) and after (Fig. 2B,C) addition of the enzyme. Table 2. Kinetic parameters and DDGà for hydrolysis of isomaltose and p-nitrophenyl-a- D -glucopyranoside, and mono-deoxy analogs of methyl a-isomaltoside at binding subsite +1 by a-glucosidases and glucoamylase. Substrate k cat (s )1 ) K m (m M ) k cat /K m (s )1 ám M )1 ) DDGà a (kJámol )1 ) a-Glucosidase (Brewer's yeast) b Isomaltose 5.2 34.5 0.15 p-Nitrophenyl-a- D -glucopyranoside 135 0.2 677 OH-2 0.5 OH-3 )0.8 OH-4 )0.7 Oligo-1,6-glucosidase a Isomaltose 33.3 6.9 4.8 p-Nitrophenyl-a- D -glucopyranoside 129 1.3 988 OH-2 5.8 OH-3 4.1 OH-4 ± Glucoamylase c Isomaltose 0.41 19.8 0.021 p-Nitrophenyl-a- D -glucopyranoside 0.50 3.7 0.135 OH-2 1.1 OH-3 8.6 OH-4 16.5 a Data from Table 1; b [28]; c [7,51]. Fig. 2. Hydrolysis of isomaltose by baker's yeast a-glucosidase followed by 1 HNMR.(A) before addition o f enzyme; (B) 4 min; and (C) 16 h after addition of enzyme. Ó FEBS 2002 Substrate recognition in yeast a-glucosidases (Eur. J. Biochem. 269) 731 Comparison of these spectra showed the appearance of a doublet centered at 5.22 p.p.m. This was assigned to H-1 of free a-glucose while the resonance at 4.64 p.p.m., which appeared lat er was assigned to H-1 o f b-glucose which stemmed from mutarotation of the initially released a-glucose. The anomer ratio of D -glucose (33% a: 67% b) was deduced from the 1 H NMR spectrum after complete hydrolysis of isomaltose (Fig. 2C) and falls within the range normally found for t he equilibrium mixture. The stereo- chemistry of the products thus con®rmed that the three a-glucosidases catalyze hydrolysis of isomaltose with reten- tion of the anomeric con®guration as is characteristic o f family 13 glycoside hydrolases. Figure 3 shows the widely accepted double displacement mechanism for retaining hydrolases, which is believed to occur through oxacarbeni- um ion transition-states and formation of a covalent intermediate between the catalytic nucleophile and the C-1 of the substrate glycon [18,30,31,45±49]. Further kinetics analyses are not feasible due to the limited amounts of analogs available; we therefore cannot determine the role of a key polar group in the glycosylation or the deglycosylation steps in the mechanism (Fig. 3). However, one can conclude that the discrimination of the diastereoisomer, as this is associated with the V max and not the K m , does not happen in the initial reversible part of substrate complex formation, but in subsequent steps o f the catalytic m echanism [40]. Recognition of diastereoisomeric isomaltoside derivatives Isomaltose is ¯exible due to rotation around the C5±C6 bond. It is possible to block this conformational ¯exibility by alkylation of C6 (Fig. 4). Previously, methyl 6-R-and methyl 6-S-methyl-a-isomaltoside were used to determine the preferred rotational conformer for glycoamylase [40]. Hydrolysis catalyzed by baker's yeast a-glucosidase (this enzyme was chosen as it has the highest activity of the two a-glucosidases; see Table 1) was similarly examined using methyl 6-R-ethyl- and methyl 6-S-ethyl-a-isomaltoside as the pair of conformationally biased substrate analogs (Table 3). While methyl 6-S-ethyl-a-isomaltoside was hydrolyzed with twofold lower V max ,butthesameK m as isomaltose (Table 3), the 6-R enantiomer was a poor substr ate V max being 150-fold lower and K m twofold higher than for isomaltose (Table 3). Baker's yeast a-glucosidase thus preferred the 6-S isomer. In contrast, glucoamylase from A. niger hydrolyzed the 6-R enantiomer with 2 30-fold higher k cat /K m compared to the parent i somaltoside, the difference being essentially in the K m and not in the rate of hydrolysis as for the a-glucosidase [40]. This distinct preference for one of the two diastereoisomers of the C-6 alkyl isomaltose derivatives re¯ects the fact that one of the rotamers adopts a conformation with more favorable Fig. 3. Catalytic mechanism for retaining gly- coside hydrolases including steps of protonation, formation of a cov alent intermediate, and product release, respectively, but not the i nter- mediate two transition-states (see text for details and [30,31,39]). AB Fig. 4. Structure of the conformationally biased diastereosisomer substrates methyl 6-R- ethyl-a-isomaltoside (A) and methyl 6 -S-ethyl- a-isomaltoside (B). Table 3. Kinetic parameters for the hydrolysis of conformationally biased isomaltosides. Substrate V max (m M á s )1 áU )1 ) K m (m M ) V max /K m (s )1 áU )1 ) a-Glucosidase from baker's yeast a Isomaltose 2.8 ´ 10 )3 9.8 2.8 ´ 10 )4 Methyl 6-S-ethyl-a-isomaltoside 1.6 ´ 10 )3 9.6 1.7 ´ 10 )4 Methyl 6-R-ethyl-a-isomaltoside 1.8 ´ 10 )5 19.4 9.3 ´ 10 )7 Glucoamylase from A. niger b k cat (s )1 ) K m (m M ) k cat /K m (s )1 ám M )1 ) Methyl a-isomaltoside 1.04 24.5 0.0042 Methyl 6-S-methyl-a-isomaltoside 1.1 90.0 0.012 Methyl 6-R-methyl-a-isomaltoside 0.68 0.71 0.96 a At 30 °C, using 50 m M phosphate, pH 6.8. b [40]. 732 T. P. Frandsen et al. (Eur. J. Biochem. 269) Ó FEBS 2002 spatial distribution of the groups that play an important role in the enzyme recognition. This ®nding stresses the fundamentally different active site architecture that exists for the inverting glucoamylase and the retaining a-gluco- sidases. Glucoamylase, in c ontrast to a-glucosidase, applies a single d isplacement mechanism and belongs to a different fold family, glycoside hydrolase 15. The speci®c activities and substrate af®nities are similar for these retaining and inverting enzymes, all of which have reasonable capacity i n the g lucose release from the no nreducing end of disac- charides and small substrates. However, the a-glucosidase showed large variation in rate of hydrolysis between the methyl 6-S-and6-R-ethyl a-isomaltosides, with small differences in af®nity for the two distereoisomers, whereas the discrimination b y glucoamylase was associated with the K m [40] and not with the r ate of hydrolysis (Table 3). CONCLUSION The enzyme preparations used in the present analysis are considered valuable representatives of two categories of yeast a-glucosidases. The study strongly demonstrates the advantage offered by enzymes with simple speci®city for application o f substrate analogs in elucidation of the roles of individual substrate groups or atoms in binding and catalysis. This is in contrast to other enzymes of the glycoside hydrolase family 13 catalysing polysacchari de degradation in an endo-like fashion, for which even model substrates would typically be rather large and hence extremely dif®cult, laborious and costly to synthesize. In addition, the option of s everal function al bin ding m odes in the active site cleft in these latter enzymes obscures interpretation using analogs of the impact of speci®c substrate groups on catalysis. The fact that the enzymes used in the present study possess a simple substrate speci®city and belong to the large family 13 representing 28 substrate speci®cities [30] suggests an application of the present ®ndings, ultimately, for rational protein engineering of these and other family members with other speci®cities. Contacts with invariant Arg, His, and Asp residues involved in charged hydrogen bonds to the glucose ring at subsite )1 in family 13 (reviewed in [30]) are thus proposed to be responsible for the reported major role of h ydroxyl groups of this ring in transition-state stabilization. While the invariant Asp plays a r ole in c atalysis [47] and mutation leads to i nactivation, the s ingle mutation to Asn of each of two His interacting at subsite )1 with OH-6 and OH-2 plus OH-3, in case of b arley a-amylase 1, affected transition- state stabilization and reduced activity to 5 and 10%, respectively [50]. Structure guided sequence comparisons, in contrast, do not yet allow tentative identi®cation of speci®c residues that are important for the interactions with the substrate ring at subsite + 1 in the oligo-1,6-glucosidase as well as for controlling the exo-action at th e level of the nonreducing end ring at subsite )1ofallthreea-glucosid- ases included in the present comparison. The ®ndings on substrate key polar groups and preferred isomaltoside diasteroisomers, however, will be valuable in future mod- eling of s ubstrate complexes of the a-glucosidases a nd related enzymes if the structures become available. The data may thus guide protein engineering studies that address the a-1,4 and a-1,6 bond speci®city of these closely related a-glycosidases. ACKNOWLEDGEMENTS Bent O. Petersen is gratefully acknowledged for performing the NMR spectroscopy experiments and Ulrike Spohr and Raymond U. Le mieux are t hanked for t he synthetic substrate analogs. Th is work was supported in part b y funding from the Natural Sciences and Engineering Research Council of Canada (to M. M. P). REFERENCES 1. Bourne, Y., van Tilbeurgh, H. & Cambillau, C. (1993) Protein± carbohydrate interactions. Curr. Opin. Struct. Biol. 3, 681±686. 2. Bundle, D.R. & Young, N.M. (1992) Carbohydrate±protein interactions in antibodies and lectins. Curr. Opin. Struct. Biol. 2, 666±673. 3. Lemieux, R.U . (1996) H ow water provides the impetus for molecular recognition in aqueous solution. Acc. Chem. Res. 29, 373±380. 4. Lis, H. & Sharon, N. (1998) Lectins: carbohydrate-speci®c pro- teins that mediate cellular recognition. Chem. Rev. 98, 637±674. 5. Toone, E.J. (1994) Structure and energetics of protein-carbohy- drate complexes. Curr. Opin. Struct. Biol. 4, 719±728. 6. Vijayan, M. & Chandra, N. (1999) Lectins. Curr. Opin. Struct. Biol. 9, 707±714. 7. Frandsen, T.P., Stoer, B.B., Palcic, M.M., Hof, S. & Svensson, B. (1996) Structure and energetics of the glucoamylase-isomaltose transition-state complex probed by using modeling and deox y- genated substrates coupled with site-directed mutagenesis. J. Mol. Biol. 263, 79±89. 8. Nikrad, P.V., Beierbeck, H. & Lemieux, R.U. (1992) Molecular recognition X. A novel procedure for the detection of the inter- molecular hydrogen bonds present in a protein oligosaccharide complex. Can. J. Chem. 70, 241±253. 9. Sierks, M.R. & Svensson, B. (1992) Kinetic identi®cation of a hydrogen bonding pair in the glucoamylase/maltose transition state complex. Prot. Eng. 5, 185±188. 10. Sierks, M.R. & Svensson, B. (2000) Energetic and mechanistic studies of glucoamylase using molecular rec ognition of maltose OH groups c oupled with site-directed mutagenesis. Biochemistry 39, 8585±8592. 11. Spohr, U., L e, N., Ling, C.C. & Lemieux, R.U. (2001) The syn- theses of 6-C-alkyl derivatives of methyl a-isomalt os ide for a study of the mec hanism of hydrolysis by amyloglucosidase. Can. J. Chem. 79, 238±255. 12. Bock, K. & Pedersen, H. (1988) Synthesis of 6-(S)deuterium- labelled derivatives of maltose and isomaltose. Acta Chem. Scand. Ser. B 42, 190±195. 13. Frandsen, T.P. & Svensson, B. (1998) Plant a-glucosidases of the glucoside hydrolase fam ily 31. M olecular properties, substrate speci®city, r eaction mechanism, a nd comparison with family members of d ierent origin. Plant Mol. Biol. 37, 1±13. 14. Watanabe, K., Hata, Y., Kizaki, H., Katsube, Y. & Suzuki, Y. (1997) The re®ned structure of Bacillus cereus oligo-1,6-glucosi- dase at 2.0 A Ê resolution: structural characterizat ion of p roline- substitution sites for protein thermostabilization. J. Mol. Biol. 269, 142±153. 15. Hu È lseweh, B., D ahlems, U.M., Dohmen, J., S trasser, A.W.M. & Hollenberg, C.P. (1997) Characterization of the active site of Schwanniomyces occidentalis glucoamylase by in vitro mutagenesis. Eur. J. Biochem. 244, 128±133. 16. Inohara-Ochiai, M., Nakayama, T., Goto, R., Nakao, M., Ueda, T. & Shibano, Y. (1997) Altering substrate speci®city of Bacillus sp. SAM1606 a-glucosidase by comparative site-speci®c muta- genesis. J. Biol. Chem. 272, 1601±1607. 17. Inohara-Ochiai, M., Okada, M., Nakayama, T., Hemmi, H., Ueda, T., Iwashita, T., Kan, Y., Shibano, Y., Ashikari, T. & Nishino, T. (2 000) An active-site m utat ion causes e nh anced Ó FEBS 2002 Substrate recognition in yeast a-glucosidases (Eur. J. Biochem. 269) 733 reactivity and altered regiospeci®city of transglucosylation cata- lyzed by t he Bacillus sp. SAM1 606 a-glucosidase. J. Biosci. Bioeng. 89, 431±437. 18. Liiv, L., Parn, P. & Alamae, T. (2001) Cloning of maltase gene from a methylotrophic yeast, Hansenula polymorpha. Gene 265, 77±85. 19. Nakamura,A.,Nishimura,I.,Yokoyama,A.,Lee,D.G.,Hidaka,M., Masaki, H., Kimura, A., Chiba, S. & Uozumi, T. (1997) Cloning and sequencing o f an a-glucosidase gene from Aspergillus niger and its expression in Aspergillus nidulans. J. Biotechn 53, 75±84. 20. Okuyama, M., Okuno, A., Shimizu, N., Mori, H., Kimura, A. & Chiba, S. (2001) Carboxyl group of residue Asp647 as possible proton do nor i n catalytic reaction o f a-glucosidase from Schizo- saccharomyces pombe. Eur. J. Biochem. 268, 2270±2280. 21. Takii, Y., D aimon, K. & Suzuki, Y. (1992) Cloning and expression of a the rmostable e xo-a-1,4-glucosidase gen e from Bacillus Stearothermophilus ATCC12016 in Escherichia coli. Appl. Micro- biol. Biotechnol. 38, 243±274. 22. Frandsen, T.P., Olsen, F.L., Mirgo rodskaya, E., Roepstor, P. & Svensson,B.(2000)HighpIa-glucosidase from barley malt. Puri®cation, enzymic character ization, and nucleo tide sequence. Plant Physiol. 123, 275±286. 23. Chiba, S. (1988) a-Glucosidases. In Handbook of Amylases and Related Enzymes (The Amylase Research Society of Japan, ed.) pp. 104±105. Pergamon Press, Oxford. 24. Chiba, S. (1997) Molecu lar mechanism in a-glucosidase and glu- coamylase. Bio sci. Biot ech. Biochem. 61, 1233±1239. 25. Henrissat, B. (1991) A c lassi®cation of glycosyl hydrolases based on amino acid sequence similarity. Biochem. J. 280, 309±316. 26. Henrissat, B. & Bairoch, A. (1993) New families i n the classi®ca- tion of glycosyl hyd rolases based on amino acid similarities. Bio- chem. J. 293, 781±788. 27. Coutinho, P.M. & Henrissat, B. The Carbo hydrate-active Enzyme Server, available from http://afmb.cnrs-mrs.fr/CAZ Y/ 28. Yoshikawa, K., Yamamoto, K. & Okada, S. (1994) Classi®cation of some a-glucosidases and a-xylosidases on the basis of substrate speci®city. Biosci. Biote chn Biochem . 58, 1392±1398. 29. JanecË ek, S Ï . (1997) a-Amylase family: molecular biology and evo- lution. Prog. Biophys. Mol. Biol. 67, 67±97. 30. MacGregor, E.A., JanecË ek, S Ï . & Svensson, B. (2001) Relationship between structure an d s peci®city i n t he a-amylase family. Biochim. Biophys. Acta 1546, 1±20. 31. Svensson, B. (1994) Protein engineering in the a-amylase fam ily: catalytic mechanism, substrate speci®city, and stability. Plant. Mol. Biol. 25, 141±157. 32. Davies, G.J., Wilson, K.S. & Henrissat, B. (1997) Nomenclature for sugar-binding subsites in glucosyl hydrolases. Biochem. J. 321, 557±559. 33. Brzozowski, A .M. & Davies, G.J. ( 1997) Structure of the Aspergillus oryzae a-amylase complexed with the inhibitor acar- bose at 2.0 A Ê resolution. Biochemistry 36 , 10837±10845. 34. Brzozowski, A.M., Lawson, D.M., Turkenberg, J.P., Bisga Ê rd- Frantzen,H.,Svendsen,A.,Borchert,T.V.,Dauter,Z.,Wilson, K.S. & Davies, G.J. (2000) Biochemistry 39, 9099±9107. 35. Dauter, Z., Dauter, M., Brzozowski, A.M., Christen se n, S., Borchert, T.V., Beier, L., Wilson, K.S. & Davies, G.J. (1999) X-ray structure of Novamyl, the ®ve-domain ÔmaltogenicÕ a-amylase from Bacillus stearothermophilus: maltose and acarbose complexes at 1.7 A Ê resolution. Biochemistry 38 , 8385±8392. 36. Kadziola, A., Sùgaard, M., Svensson, B. & Haser, R . (1998) Molecular structure of a barley a-amylase-inhibitor complex: implications for starch binding and catalysis. J. Mol. Biol. 278, 205±217. 37. Qian, M., H aser, R., Buisson, G., Due  e, E. & Payan, F. (1994) The active cent er of a mammalian a-amylase. Structure of th e co mplex of a pancreatic a-amylase with a carbohydrate inhibitor re®ned to 2.2 A Ê resolution. Biochemi st ry 33, 6284±6294. 38. Ly, H.D. & Withers, S.G. (1999) Mutagenesis of glycosidases. Ann. Rev. Biochem. 68 , 487±522. 39. Weill, C.E., Burch, R.J. & van Dyk, J.W. (1954) An a-amylo- glucosidase that produces b-glucose. Cereal. Chem. 31,150± 158. 40. Palcic, M.M., Skrydstrup, T., Bock, K., Le, N. & Lemieux, R.U. (1993) Substrate recognition by amyloglucosidase: evaluation of conformationally biased isomaltosides. Carbohydr. Res. 250,87± 92. 41. Sierks, M.R., Bock, K., Refn, S. & Svensson, B. (1992) Active site similarities of glucose dehydrogenase, glucose oxidase and gluco- amylase probed by deoxygenated substrates. Biochemistry 31, 8972±8977. 42. Lemieux, R.U., Spohr, U ., Bach, M., Cameron, D.R., Frandsen, T.P., Stoer, B.B., Svensson, B. & Palcic, M.M. (1996) Chem ical mapping of the active site of glucoamylase o f Aspergillus niger. Can. J. Chem. 74, 319±335. 43. Wilkinson, A.J ., Fersht, A.R., Blow, D.M. & Winther, G. (1983) Site-directed mutagenesis as a probe of enzyme struc ture and catalysis: tyrosyl-tRNA synthetase cysteine-35 to glycine-35 mutation. Biochemistry 22, 3581±3586. 44. Fersht, A.R., Shi, J P., K nill-Jones, J., Lowe, D.M., Wilkinson, A.J., Blow, D.M., Brick, P., Carter, P., Waye, M.M.Y. & Winter, G. (1985) Hydrogen bonding and biological speci®city analyzed by protein engineering. Nature 314, 235±238. 45. Davies, G.J., Mackenzie, L., Varrot, A., Dauter, M., Brzozowski, A.M., Schu È lein, M. & Withers, S.G. (1998) Snapshots along an enzymatic-reaction coordinate ± analysis of a retaining b-glucoside hydrolase. Biochemistry 37, 11707±11713. 46. McCarter, J.D. & Withers, S.G. (1994) Mechanisms of enzymatic glycosid e hydro lysi s. Curr. Opin. Struct. Biol. 4, 885±892. 47. Uitdehaag, J.C.M., Mosi, R., Kalk, K.H., van der Veen, B.A., Dijkhuizen, L., Withers, S.G. & Dijkstra, B.W. (1999) X-ray structures along the reaction pathway of cyclodextrin glycosyl- transferase elucidate catalysis i n the a-amylase family. Nat. Struct. Biol. 6, 432±436. 48. van der Veen, B., Uitdehaag, J.C.M., Penninga, D., van Alebeek, G J., Smith, L.M., D ijkstra, B.W. & Dijkhuizen, L . (2000) Rational design of cyclodextrin glycosyltransferase from Bacillus circulans strain 251 to increase a-cyclodextrin produ ction. J. Biol. Chem. 296, 1027±1038. 49. Vocadlo, D.J., Davies, G.J., Laine, R. & Withers, S.G. (2001) Catalysis by hen egg-white lysozyme p roceeds via a covalent interm ed iat e. Nature 412, 835±838. 50. Sùgaard, M., Kadziola, A., Haser, R. & Svensson, B. (1993) Site- directed mutagenesis of histidine 93, aspartic acid 180, glutamic acid 205, histidine 290, and aspartic acid 291 at the active site and tryptophan 279 at the raw starch binding site in barley a-amylase 1. J. Biol. Chem. 268, 22480±22484. 51. Frandsen, T.P., Ch ristensen, T., S toe r, B., L ehmbec k, J., Dupont, C., Honzatko, R.B. & Svensson, B. (1995) Mutational analysis of the roles in catalysis and substrate recognition of arginines 54 and 305, aspartic acid 309, and tryptophan 317 located at subsites 1 and 2 in glucoamylase from Aspergillus niger. Biochemistry 34, 10162±10169. 734 T. P. Frandsen et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Substrate recognition by three family 13 yeast a-glucosidases Evaluation of deoxygenated and conformationally biased isomaltosides Torben P. Frandsen 1, *,. Molecular recognition of isomaltosides is more similar for the yeast a-glucosidase and oligo-1,6-glucosidase of glycoside of hydrolase family 13 when compared