Investigation of the substrate specificity of a b-glycosidase from Spodoptera frugiperda using site-directed mutagenesis and bioenergetics analysis Sandro R. Marana, Eduardo H. P. Andrade, Cle ´ lia Ferreira and Walter R. Terra Departamento de Bioquı ´ mica, Instituto de Quı ´ mica, Universidade de Sa ˜ o Paulo, Sa ˜ o Paulo, Brazil The specificity of the Spodoptera frugiperda digestive b-glycosidase (Sfbgly50) for fucosides, g lucosides and galactosides is determined by noncovalent interactions of glycone 6-OH and glycone 4-OH with the active-site residues Q39 a nd E451.Site-directed mu tagenesis and enzyme steady- state kinetics were describe d, showing that replacement of E451 wit h glutamine increased the preference of Sfbgly50 for glucosides in comparison to g alactosides, whereas replacing E451 with serine had the opposite effect. In contrast, the replacement of E451 with a spartate did not change Sfbgly50 specificity. Th e energy of the interactions formed by these different residues with the axial and equa- torial glycone 4-OH were also measured, showing that the increase in preference for galactosides resulted from a larger energy decrease in the interaction with equatorial 4-OH than with axial 4-OH (22.6 vs. 13.9 kJÆmol )1 ), whereas the increase in preference for glucosides was caused by an energy reduction in the interaction with the axial 4 -OH (5.1 kJÆmol )1 ). The introduction of glutamine at position 451 or of asparagine at position 39 i ncreased the preference of Sfbgly50 for fucosides in comparison to galactosides, whereas the presence of aspartate or serine at position 451 had less effect on this preferen ce. The hydrolysis of fucosides was favored because glutamine at position 451 increased a steric hindrance with 6-OH of 7.1 kJÆmo l )1 and asparagine at position 39 disrupted a favorable interaction with this same hydroxyl. In conclusion, it is proposed that the spe- cificity of new b-glycosidase mutants can be predicted by combining and adding energy of the enzyme–substrate interactions evaluated in the present study. Keywords: b-glycosidase; b ioen ergetics analysis; e nz yme specificity; glycoside hydrolase; site-directed mutagenesis. b-glycosidases from glycoside hydrolase family 1 are enzymes t hat r emove monosaccharides from the nonred uc- ing e nd of di- and/or oligosaccharides. The nonreducing monosaccharide residue binds at the glycone subsite (subsite )1), w hereas the r emaining part of the substrate is accommodated by the aglycone subsite 2 , which may actually be composed of several subsites (+1, +2, +3, etc.). According to the CAZy database, family 1 currently comprises 427 b-glycosidases, with 3D structural data being available for 12 [1]. All f amily 1 b-glycosidases share the same tertiary structure [(b/a) 8 barrel]; they are configur- ation-retaining glycosidases, the catalytic activi ty of which depends on two glutamic acid residues, one positioned after the b strand 4 (catalytic proton donor) and the o ther after the b strand 7 (catalytic nucleophile). Family 1 comprises enzymes with 14 d ifferent EC numbers 3 , catalyzing hydro- lysis of substrates presenting a variety of glycones (mono- saccharides such a s glucose, galactose, fucose, mannose, 6-phosphoglucose and 6-phosphogalactose) and aglycones (monosaccharides, oligosaccharides, alkyl and aryl m oiet- ies) [1]. This broad substrate specificity makes family 1 an interesting model for using to study the molecular basis of the enzymatic specificity. Having a better understanding of the m olecular basis of b-glycosidase specificity would result in an i mproved knowledge of the physiological role of these e nzymes, as well as contributing t o t he design of b-glycosidases w ith novel specificities. Enzymatic specificity mostly relies on noncovalent inter- actions between amino acid residues within enzyme active sites and groups of the reactant transition state (S à )[2];the strength of the n oncovalent interaction between enzyme active sites and S à determines the stability of the enzyme– transition state (ES à ) complex and the rate of the reaction. Thus, in theory, modifications of the enzymatic specificity could be accomplished by changing a ctive-site residues. The role of the noncovalent interactions between amino acid residues of the active site and glycone hydroxyls have already been studied in family 1 b-glycosidases from Correspondence to S. R. Marana, Departamento de Bioquı ´ mica, Instituto de Quı ´ mica, Universidade de Sa ˜ o Paulo, CP 26077, Sa ˜ o Paulo, 05513-970, Brazil. Fax: +55 1 1 38182186, 1 Tel.: +55 11 30913810, E-mail: srmarana@iq.usp.br Abbreviations: ES, enzyme–transition state complex; DG, activation energy of the glycosylation step; DDG, differences in the activation energy of the glycosylation steps; 3-OH, glycone hydroxyl 3; 4-OH, glycone hydroxyl 4; 6-OH, glycone hydroxyl 6; NP, p-nitrophenyl; S, reactant transition state. Enzymes:digestiveb-glycosidase (b- D -glucoside glucohydrolase) from Spodoptera frugiperda (EC 3.2.1.21) GenBank access no.: A F052729. (Received 2 8 March 2004, revised 24 August 2004, accepted 3 September 2004) Eur. J. Biochem. 271, 4169–4177 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04354.x Agrobacterium sp. [3], in Pyrococcus furiosus [4], in lamb (lactase-phlorizin hydrolase) [5], in guinea pig (citosolic b-glucosidase) [ 6] and in Spod optera frugiperda [7]. Nevertheless, how the substitution of active-site amino acids would affect the energy of noncovalent interactions with the substrate in t he ES à complex is not easily predictable. Consequently, there are only a few published studies that report changing the specificity o f family 1 b-glycosidases b y m utagenesis [8–12 ]. In these studies, residues interacting with the substrate were replaced with a single amino acid. The criterion applied when choosing the amino acid residue to be introduced in the mutant enzyme was b ased either on a sequence comparison with o ther b-glycosidases presenting the desired s pecificity [10,11], or on the hydrogen-bonding properties of the mutant residue [12]. Although these studies were successful in changing substrate specificity, they do not provide us w ith a significant basis for further e xperiments on changing the specificity o f family 1 b-glycosidases by mutagenesis, as only one type of mutation was evaluated in those and, further- more, its possible effects on the energy levels of noncovalent interactions with the substrate were not included. Noncovalent interactions in the ES à complex between two amino acid residues (E451 and Q39) and the glycone hydroxyls 4 (4-OH) and 6 (6-OH) determine the preference of the d igestive b-glycosidase from S. frugiperda (Sfbgly50) for three substrates: fucosides, glucosides and galactosides [7]. It was shown that E451 forms s tronger hydrogen bonds with equatorial and axial 4-OH than Q39. E451 presents a steric hindrance, whereas Q39 has a weak interaction with 6-OH [7]. The purpose of the present study was to substitute residues Q39 and E451 of Sfbgly50, through site-directed mutagenesis, for r esidues presenting hydrogen bond-form- ing side-chains. Steady-state kinetic data for the hydrolysis of different substrates by Sfbgly50 mutants were then used to calculate the energy of noncovalent interactions in the ES à complex between different residues at positions 39 or 451 and glycone 4-OH or 6-OH. The resulting data were used to clarify the specificity of family 1 b-glycosidases, as well as to predict the s pecificity of new Sfbgly50 mutants. Materials and methods All reagents, unless otherwise specified, were purchased from Sigma or Merck. Site-directed mutagenesis Site-directed mutagenesis was performed using a plasmid pT7-7 [13] containing an inserted DNA fragment co ding for the mature Sfbgly50 (pT7b50) [14]. Experiments were carried out according to the instructions included with the ÔQuikChange site-directed mutagenesisÕ kit ( Stratagene). The sequence of t he mutagenic p rimers contained a common segment (underlined) and mutated codon (NNN, in bold). Thus, the primer sequence used on mutations at position 3 9 was 5 ¢- CGCTACAGCCTCCTACNNNAT CGAAGGTGCTTGG-3¢, w ith AAC and GAG as the mutated codons for Q39N and Q39E, respectively. For mutation at position E451, the primer sequence was 5¢- GGAGTCTAATGGACAACTTTNNNTGGATGGA GGGTTATATTGAGCG-3¢,withGAC,CAAandTCA as mutated codons for E 451D, E451Q and E451S, respectively. DNA sequencing w as used to confirm t he incorporation of the mutated c odon in the p T7b50. Expression of the mutant Sfbgly50 The Sfbgly50 mutants were expressed in Novablue DE3Ò cells (Novagen), as described previously [7]. Purification of the recombinant Sfbgly50 The mutant Sfbgly50 was purified by hydrophobic chro- matography [7] followed by a second step of ion-exchange chromatography introduced as an additional polishing step. The fractions containing b-glycosidase activity eluted in the first chromato graphy were pooled, dialyzed in 20 m M triethanolamine buffer, pH 8.0, for 16 h at 4 °Candthen loaded onto a ResourceQ column (Amersham Bioscience). The nonretained proteins were washed out in the d ialysis buffer, while the retained proteins were eluted f rom the column using an NaCl gradient prepared in the same buffer. The presen ce of the recombina nt Sfbgly50 was detected using NPbglc [15] and its purity ascertained by SDS/PAGE followed by silver staining [16,17]. Protein determination was performed spectrophoto- metrically (absorbance at 280 nm) using e 280 ¼ 117 200 M )1 Æcm )1 [18]. Kinetic analysis All assays were performed at 30 °Cin50m M citrate- phosphate buffer, pH 6.0, and initial rate data were measured. The hydrolysis of NPbglycosides was followed by the release of p-nitrophenyl (NP) [15]. The kinetic parameters (k cat and K m ) were determined by employing 10 different substrate concentrations, and the data were fitted to a Michaelis–Menten equation by using the ENZFITTER software (Elsevier-Biosoft, Cambridge, UK). Calculation of the energy of noncovalent interactions in the ES à complex Different Sfbgly50 mutants (Q39E, Q39N, E451Q, E451D and E451S) were prepared. N oncovalent i nteractions involving any residue at position 39 were designated by /, while noncovalent interactions involving any residue at position 451 were designated by g. Glycone hydroxyls 4 and 6 were designated 4 and 6, where ÔeÕ stands for an equatorial hydroxyl and ÔaÕ for a n axial one. Therefore, the interaction between any residue at position 39 and an equatorial 4-OH was called /4e, and between any residue at position 451 and an equatorial 4-OH was called g4e. Interactions involving these same residues and an axial 4-OH were named /4a and g4a, respectively. Likewise, interactions between any residue at position 39 and 6-OH were called /6 and between any residue at position 4 51 and 6 -OH was des ignated g6 (Fig. 1). X4a and X6 correspond to noncovalent inter- actions between any amino acid residue other than 39 and 451 with axial 4-OH and 6-OH, respectively. Interaction energy may be measured by using the method described previously [7]. This method assumes that 4170 S. R. Marana et al.(Eur. J. Biochem. 271) Ó FEBS 2004 noncovalent interactions formed by residues 3 9 and 451 are independent and also compares the energy of ES à complexes containing different sets o f noncovalent i nteractions be- tween E and S à . The steps used to calculate the energy of individual noncovalent interactions are described below. The r eaction mechanism of family 1 b-glycosida ses has a glycosylation (from E + S to E–G) and a deglycosylation (from E–G to E + G) step, each with an ES à complex, as described by the following reaction [19]: E þ S  ! k 1 k À1 ES À! k 2 E-G þ Ag ! k 3 E þ G where S is formed by a glycone (G) covalently bound to an aglycone (Ag), ES represents the Michaelis complex and E–G the covalent intermediary ( glycosyl-enzyme). The f ollowing kinetic parameters are valid for t hat reaction: k cat ¼ k 2 k 3 k 2 þ k 3 K m ¼ ðk À1 þ k 2 Þk 3 k 1 ðk 2 þ k 3 Þ k cat =K m ¼ k 1 k 2 k 2 þ k À1 where k cat /K m is the rate constant for the glycosylation step and can be used to calculate its activation energy (DG à ). DG à represents the difference of energy between the glycosylation ES à complex and the E + S ground state. Thus, as different substrates (NP bglycosides) and enzymes (mutant or wild- type) should present s imilar ground state energy, DDG à for the glycosylation step represents the difference of the energy between two ES à complexes. Moreover, as the energy of an ES à complex d epends on the noncovalent interactions between E and S à , DD G à values represent the sum of all noncovalent interactions that differ between the two ES à complexes. Therefore, difference s i n t he activation energy of t he glycosylation steps (DDG à ), corresponding to a pair of different substrates hydrolyzed by the same enzyme, or to a pair of different enzymes hydrolyzing the same substrate (Figs 2 and 3), were calculated by the following equation [2]: DDG z ¼ RT lnðc 1 =c 2 Þ where c 1 and c 2 are k cat /K m ratios determined from the hydrolysis of two different substrates by the same enzyme or from the hydrolysis of t he same substrate by two different enzymes (R ¼ 8.3144 JÆK )1 Æmol )1 and T ¼ 303 K). As detailed in Figs 2 and 3, all DDG à resulting from those comparisons are a sum of DG à .EachDG à value represents the disruption of a noncovalent i nteraction owing to the substitution of residues 39 or 451 for an alanine residue or because o f t he lack of a glycone hydroxyl. Thus, those DG à values actually represent the energy of noncovalent inter- actions in the ES à complex. The energy of noncovalent interaction s /4e, /4a, /6, g4e , g4a and g6 can be calculated by isolating each DG à through the subtraction of different DDG à values, as shown in Table 1. For example, the interaction g6(inter- action between residue 451 and 6-OH) can be calculated by subtracting DDG z 2 from DDG z 7 , a s detailed below: DDG z 7 À DDG z 2 ¼ DG z g6 þ DG z /6 þ DG z X6 ÀðDG z /6 þ DG z X6 Þ DDG z 7 À DDG z 2 ¼ DG z g6 It should be noted that, by using this method, it was not possible to i solate the DG à corresponding to the energy of the hydrogen bond between residue 39 and 3-OH (/3), which was added to t he energy of /4e and /4a interactions. Results and Discussion The induction of recombinant enzymes (mutants Q39E, Q39N, E451Q, E451D, E451S) was confirmed by SDS/ PAGE. Then, the soluble m aterial of the induced bacteria was used in the purification of the m utant enzymes through hydrophobic and ion-exchange chromatography. Isolation of the mutant enzymes was then confirmed by SDS/PAGE. This procedure y ielded % 0.5 mg of m utants E451S and E451Q, 0.3 mg of mutant Q39E and 0.15 mg of m utants Q39N and E451D, from 0.5 L o f bacterial culture. Steady-state kinetic parameters for different Sfbgly50 mutants hydrolyzing several s ubstrates were determined (Table 2). The replacement of residues E451 or Q39 w ith other residues mostly affects k cat , irrespective of the substrate used, suggesting that the m easured e nergy differences mainly r eflect differences in the E S à complex. This finding corroborates the hypothesis that noncovalent interactions formed by glycone hydroxyls with the b-glycosidase active sites are stronger in ES à than in the E S complex [3]. All mutants have lower k cat and k cat /K m values than wild- type Sfbgly50 (Table 2). Moreover, all mutants at position Fig. 1. Schematic diagram showing the noncovalent interactions (dotted lines) involved in the substrate glycone (glc) b inding w ith the w ild-type Spodoptera frugiperda digestive b-glycosidase (Sfbgly50) active site. The interactions are identified as g6, g4e, /4e , /6and/3. Residue 451 is represented by g,and/ represents residue 39, e denotes an equatorial hydroxyl, and 6, 4 a nd 3 indicate the glycone 6 -OH, 4-OH and 3-OH, respectively. Interactions with the axial 4-OH (g4a, /4a) were not r epresented. L etters in side the b oxes indicate residues introduced through site-directed mu tagenesis at positions 39 and 451. The diagram was b ased on the structure of the Sinapis alba my ros ina se [24]. Ó FEBS 2004 Molecular basis for specificity of a b-glycosidase (Eur. J. Biochem. 271) 4171 451 present higher k cat and k cat /K m values than m utant E451A. T he same is true between mutants at position 39 and mutant Q 39A. As the r ate c onstants are invers ely proportional to t he ES à energy level [ 2], t he results indicate that, a lthough the different residues i ntroduced at positions 451 and 39 of Sfbgly50 are l ess effective o n the stabilization of the ES à complex t han w ild-type residues, they still retain the ability to interact with S à . The k cat /K m ratios were used to calculate DDG à within pairs of different substrates hydrolyzed by the same enzyme, or pairs of different enzymes hydrolyzing the same substrate (Tables 3 and 4). These DDG à values a re a sum of all noncovalent i nteractions that differ between two ES à complexes of the glycosylation step, each interaction corresponding to a DG à (Figs 2 and 3). The energy of all these noncovalent interactions between glycone hydroxyls Fig. 2. Free energy changes b etween enzyme–transition state (ES à ) compl exes fo rm ed b y Spo doptera frugiperda digestiv e b-glycosidase (Sfbgly50) mutants at residue 39. ES à complexes, represented in th e same c olumn, were formed with the same su bstrate (glc, gal and fuc), whereas ES à complexes represented in th e same r ow were fo rmed with th e s ame enzyme ( X39 or A39). X39 indicates mutants Q39E and Q39N. A39 indicates an alanine at p osition 39. Amino a cid side-chains and s ubstrates were represented as a simplified outline based on the structure presented in F ig. 1. Only noncovalent interactions (dotted lines) involving the residue 39 and E451 were represented. 4-OH is represented in the equatorial (4e; pointing towards left) and axial (4a; pointing upwards) positions. 3-OH is always in the equatorial position, and a missing 6-OH indicates a 6-deoxy monosaccharide (fucose; third column). Fig. 3. Free energy changes b etween enzyme–transition state (ES à ) compl exes fo rm ed b y Spo doptera frugiperda digestiv e b-glycosidase (Sfbgly50) mutants at residue 451. ES à complexes are represented as described in Fig. 2. X451 indicates mutants E451Q, E451D and E451S. A451 indicates an alanine at position 451. For simplification p urposes, only p art of the E451 side-ch ain (wild-type) was represented, alt hough in mutants E451Q, E451D and E451S, the i nteraction s geometry might be d i fferent. Only the noncovalent interactions (dotted lines) i nvolving residue 451 and Q39 were represented. Other details are as described in the legend t o Fig. 2. 4172 S. R. Marana et al.(Eur. J. Biochem. 271) Ó FEBS 2004 (4 and 6 ) and different active-site residues at positions 39 and 451 ( /4e, /4a, /6, g4e, g4a, g6; Tables 5 a nd 6) were calculated by isolating each DG à , as described in the Materials and methods. Interactions involving residue 39 As seen in Fig. 1, the Q 39 side-chain (N e atom)ofthewild- type Sf bgly50 is a donor in a hydrogen bond with 4-OH (interactions /4e and /4a). Simultaneously, 4-OH is a donor in a h ydrogen bond with the E451 side-chain (interaction g4e and g4a). Also, t he Q39 s ide-chain (O e atom) is an acceptor in t he interaction /3. Two different residues, E and N, were introduced at position 39, originally occupied by Q. Glutam ine and asparagine present s imilar side-chains, containing hydrogen bond donor and acceptor a toms (N d ,N e and O d ,O e , respectively), in s pite of the asparagine side-chain bei ng shorter [20]. Thus, the drastic decrease of the /4e and /4a energy (68 ± 5% and 78 ± 6%, respectively; Table 5) may result exclusively from the l ength increase of these interactions. This also indicates that the length decrease of the residue 39 side-chain affects /4e and /4a interactions equally. Replacement of Q39 with E is not a conservative exchange, as the glutamine s ide-chain i s s imultaneously a hydrogen bond dono r and acceptor, whereas t he glutamate side-chain is only a hydrogen bond acceptor [20]. Thus, in mutant Q39E, the 4-OH may interact with E451, whereas the E39 side-chain (O e atom) may form the /3 interaction. E39 O e and Q39 O e atoms may form similar /3 interactions, as glutamate O e and glutamine O e atoms have similar partial c harges ()0.8188e and )0.8086e, respectively) [21]. Nevertheless, in mutant Q39E, t he replacement of Q with E disrupts the interactions /4e and /4a, because E is not a hydrogen bond donor. Considering t hat t he /3energywas included i nto /4e and /4a energies during calculations (see the Materials and methods), the values obtained for /4e and /4a (8.9 an d 8.2 kJ Æmol )1 , respectively) when E is a t position 39 may actually correspond to the /3 interaction (Table 5). Based on t hose e stimates of /3energy,/4e a nd /4a energy in the wild-type Sf bgly50 should be 10.4 kJÆmol )1 (19.3–8.9) and 5 .9 kJÆmol )1 (14.1–8.2), respectively (Table 5). In the E S à complex of the wild-type S fbgly50, 4-OH has been showntohaveastrongerinteractionwithE451thanwith Q39 (g4e ¼ 33.2 kJÆmol )1 vs. /4e ¼ 19.3 kJÆmol )1 ; g4a ¼ 18.7 kJÆmol )1 vs. /4a ¼ 14.1 kJÆmol )1 ) [7]. Thus, the esti- mates of /4e and /4a energy presented here ( 10.4 kJÆmol )1 and 5 .9 kJÆmol )1 , r espectively) show that this energy differ- ence is actually higher than previously calculated, strengthening t he importance of E451 in determining Sfbgly50 specificity. The large length of /6 i nteraction is incompatible with a hydrogen bond. In addition, the replacement o f Q39 with E (which altered the hydrogen bonding properties of the residue 39 side-chain, while maintaining its length) did not affect the energy of that interaction (Ta ble 5). More over, the replacement of Q39 with N, which sh ortened the length of the residue 39 side-chain, disrupted /6, as indicated by its negative value (Table 5). Table 1. Mathematical expressions used to calculate the e nergy of noncovalent interactions between residues at positions 39 or 451 and 4-OH or 6-OH in the enzyme–tr ansition s tate (ES à )complex./, residue at position 39; g, residue at position 451; 4, glycone hydroxyl 4; 6, glycone hydroxyl 6; a, axial position; e, equatorial position. DG à , activation energy of the glycosylation step; DDG à , differences in the activation energy of the g lycosylation steps. Noncovalent interactions involvingamino acid residues at position 39 Noncovalent interactions involving amino acid residues at position 451 Noncovalent interaction Expression for calculation Noncovalent interaction Expression for calculation /4e a DDG à 3 – DG à /6 g4e DDG à 3 – DG à g6 /4a a DDG à 5 g4a DDG à 5 /6 DDG à 2 – DDG à 7 g6 DDG à 7 – DDG à 2 a The energy value of /3 is included in these interactions. Table 2. Steady-state kinetic parameters for hydrolysis of NPbglycosides by the recombinant wild-type Spodoptera frugiperda digestive b-glycosidase (Sfbgly50) (wt) a nd Sf bgly50 mutants. Experiments w e re c arried ou t a t 10 different substrate con centrations a nd parameters w ere calcu lated b y using the ENZFITTER software. Standard errors were less than 12% of the me an values. NPbglc NPbgal NPbfuc k cat (s )1 ) K m (m M ) k cat /K m (s )1 Æm M )1 ) k cat (s )1 ) K m (m M ) k cat /K m (s )1 Æm M )1 ) k cat (s )1 ) K m (m M ) k cat /K m (s )1 Æm M )1 ) wt a 1.09 0.45 2.4 0.35 2.0 0.17 3.3 0.49 6.7 E451Q 0.145 1.48 0.098 0.0050 3.7 0.0013 1.40 1.60 0.87 E451D 0.0135 1.7 0.0079 0.00014 1.7 0.00087 0.0233 0.25 0.094 E451S 0.00061 2.5 0.00024 0.00093 1.7 0.00053 0.0186 0.68 0.027 E451A a 0.000082 3.1 0.000026 0.000285 0.50 0.00057 0.00061 1.53 0.0039 Q39E 0.051 1.61 0.0321 0.0377 2.8 0.0134 1.49 2.3 0.64 Q39N 0.0092 6.0 0.0015 0.00038 1.5 0.00025 0.0220 0.26 0.084 Q39A a 0.00067 2.3 0.00029 0.00031 1.9 0.00016 0.0032 1.31 0.024 a Data from [7]. Ó FEBS 2004 Molecular basis for specificity of a b-glycosidase (Eur. J. Biochem. 271) 4173 Interactions involving residue 451 Inthewild-typeSfbgly50, the 4-OH i s a donor in a hydrogen bond with the E451 side-chain (O e atom) and an acceptor of a hydrogen bond with the Q39 side-chain (N e atom) (Fig. 1). Three different residues (Q, D and S) were introduced at position 451, originally occupied by E. Based on the hydrogen-bonding properties of Q, D and S side-chains (all of which are hydrogen bond acceptors), they are also expected to interact with 4-OH. In the mutant E451Q, the residue Q451 side-chain is a hydrogen bond acceptor (O e atom) and presents the same length of E side-chain. Thus, the g4e energy values observed when E a nd Q are at position 451 (33.2 kJÆmol )1 and 32.2 kJÆmol )1 , respectively; Table 6 ) indicate that this interaction i s not affected by the exchange of a charged participant ( carboxyl group of the E side-chain) for an uncharged participant (amide g roup of the Q side-chain). Otherwise, the replacement o f E451 with D and S decreases the energy of g4e by 36% and 68%, respectively (calculated from Table 6). As D and S s ide-chains are shorter than those of E and Q, the energy decreases probably result from the l engthening of the interaction g4e. For t he m utant E451S, one cannot rule out that the chemical property changes of the residue 451 side-chain (E to S; polar charged to polar uncharged) may also have affected the properties of the g4e microenvironment, thus contributing to t he observed energy decreases. For mutant E451D, a s D and E side-chains are similar, the e nergy decrease probably results from the change in the l ength of g4e. Thus, t he g4e interactions are c loser to a n optimum length when either E or Q are at position 451, having the highest possible energy. In this situation, the interaction is not affected by the e xchange b etween charged and un- charged participants. However, small increments in the interaction length result in energy alteration (as when D is at position 451), whereas larger interaction length (as when S is at position 451) results in interaction disruption. A similar behavior is observed f or the g4a interaction. In this case, as the axial 4-OH is positioned farther from the residue 451 side-chain, the length of g4a is always larger than of g4e, resulting in lower e nergies (Table 6). E451 replacement with Q and D resulted in higher decreases (27% and 57% , respectively) in g4a energy than those observed for g4e (3 % and 36%, respectively) (calculated from Table 6), whereas changing E451 for S resulted in similar decreases o f g4a and g4e energies (74% and 68%, respectively). This suggests t hat g4a interaction does not have an optimum length, even when E and Q are at position 451. Consequently, exchange between charged and un- charged p articipants affects this interaction, and small increments in the interaction length result in high energy decreases. g6 energy has consistently shown negative values, regardless of which residue (E, Q, D and S) is at position 451 (Table 6), indicating that the i nteraction between 6-OH and any of those residues has consistently been unfavorable to the formation of ES à . This effect, a possible steric hindrance, was not decreased even with the substitution of E451 by amino acids with shorter side-chains, such as D and S. Nevertheless, the replacement of E451 with Q increased that effect, probably owing to the fact t hat glutamine is bulkier than glut amic acid (11 4 A ˚ 3 and 109 A ˚ 3 , respect- ively). These data suggest that 6-OH and the side-chain of residue 451 are very close in t he ES à complex. Table 3. Changes in the transition state energies of the glycosylation step for different Spodoptera frugiperda digestive b-glycosidase (Sfbgly50) mutants resulting from deletion of the residue 39 side-chain o r hydroxyl changes in the substrate. The comparisons used to calculate these DDG à values are shown in Fig. 2. The DDG à values (representing the difference in the activation energy of the glycosylation steps) w ere calculated using k cat /K m data, as discussed in the Materials and methods. Depending on t he column, X39 indicates m utants Q39E or Q39N. A39 indicates m utant Q39A. Data for this mutant were obtained from a previous publication [7]. Comparison Q39E DDG à (kJÆmol )1 ) Q39N DDG à (kJÆmol )1 ) [X39glc] à fi [X39gal] à 2.2 ± 0.1 4.5 ± 0.2 [X39gal] à fi [X39fuc] à )9.7 ± 0.1 )14.6 ± 0.2 [X39glc] à fi [A 39 glc] à 11.8 ± 0.1 4.1 ± 0.1 [X39gal] à fi [A39gal] à 11.1 ± 0.1 1.1 ± 0.2 [X39fuc] à fi [A39fuc] à 8.2 ± 0.2 3.1 ± 0.2 [A39glc] à fi [A39gal] à 1.4 ± 0.1 1.4 ± 0.1 [A39gal] à fi [A39fuc] à )1.2 ± 0.2 )1.2 ± 0.2 Table 4. Changes in the transition state e nergies of the glycosylation step for different Spodoptera frugiperda digestive b-glycosidase (Sfbgly50) mutants resulting from deletion of the residue 451 side-chain or hydroxyl changes in the substrate. The comparisons used to calcu late the DDG à values (i.e. the difference in the activation energy of the glycosylation steps) are shown in Fig. 3 . The DDG à values were calculated using k cat /K m data, as discussed in the Mat erials a nd methods. Depending on t he column, X 451 r epresents mutants E451Q, E4 51D or E451S. A451 indicates mutant E451A. Data for t his mutant were obtained from a previous pu blication [7]. Comparison E451Q DDG à (kJÆmol )1 ) E451D DDG à (kJÆmol )1 ) E451S DDG à (kJÆmol )1 ) [A451glc] à fi [A451gal] à )7.7 ± 0.4 )7.7 ± 0.4 )7.7 ± 0.4 [A451gal] à fi [A451fuc] à )4.8 ± 0.2 )4.8 ± 0.2 )4.8 ± 0.2 [X451glc] à fi [A451glc] à 20.7 ± 0.3 14.3 ± 0.4 5.5 ± 0.4 [X451gal] à fi [A451gal] à 2.0 ± 0.3 1.0 ± 0.3 )0.1 ± 0.1 [X451fuc] à fi [A451fuc] à 13.6 ± 0.2 8.0 ± 0.1 4.8 ± 0.1 [X451glc] à fi [X451gal] à 10.8 ± 0.1 5.5 ± 0.3 )1.9 ± 0.2 [X451gal] à fi [X451fuc] à )16.3 ± 0.2 )11.7 ± 0.2 )9.9 ± 0.1 4174 S. R. Marana et al.(Eur. J. Biochem. 271) Ó FEBS 2004 Effect of replacement of residues 451 and 39 on the specificity of Sfbgly50 E451 replacement with Q increase d the Sfbgly50 preference for glucosides to the detriment of galactosides b y % five- fold (ðk cat =K mNPbglc Þ=ðk cat =K mNPbgal Þ¼14 f or wild-type Sfbgly50, whereas ðk cat =K mNPbglc Þ=ðk cat =K mNPbgal Þ¼75 for mutant E451Q) (Table 2). This change in specificity was caused by a significant d ecrease in t he g4a energy (5.1 kJÆmol )1 ), whereas g4e was not affected (Table 6). Thus, the active site of the E451Q mutant is better designed to stabilize the ES à complex of glucosides (which p resents equatorial 4-OH) than that of the galactosides (which presents axial 4-OH). Otherwise, k cat /K m ratios indicate that replacement of E451 with S decreased the Sfbgly50 preference for glucosides compared to g alactosides by % 35-fold (ðk cat = K mNPbglc Þ=ðk cat =K mNPbgal Þ¼14 for the wild-type Sfbgly50, whereas ðk cat =K mNPbglc Þ=ðk cat =K mNPbgal Þ¼0.4 for the mutant E451S) (Table 2). This m utation d ecreased g4e energy by 22.6 kJ mol )1 ,whereasg4a energy was decreased by 1 3.9 kJÆmol )1 (Table 6), r esulting in an active site that is better designed to stabilize the ES à complex of galactosides in comparison to glucosides. A previous study showed that the r eplacement (with a serine) of a glutamate equivalent to E451 in the b-glucosidase (CelB) from P. furiosus (E417), did not change the preference for glucosides vs. galactosides [11]. Moreover, in c ontrast to the present study, the mutation E417S had a stronger e ffect on K m (increment of 20–100 fold) than on k cat (increment of two-to fourfold). However, the preference for glucosides vs. galactosides was also affected b y mutating a residue equivalent to E451 in th e Sulfolobus solfataricus b-glycosidase (E432) [12]. In this case, the replacement of E432 with C decreased the preference for glucosides by 1.5-fold in comparison to galactosides. The replacement of E451 w ith D did not significantly modify the p reference o f Sf bgly50 for glucosides a nd galactosides (ðk cat =K mNPbglc Þ=ðk cat =K mNPbgal Þ¼14 for wild- type Sfbgly50, whereas ðk cat =K mNPbglc Þ=ðk cat =K mNPbgal Þ¼9 for the mutant E451D) (Table 2) because g4e and g4a are equally affected by this mutation, which resulted in a decrease of % 11 kJÆmol )1 in their energies ( Table 6). In summary, the p reference for glucosides was 14-fold higher than for galactosides i n t he wild-t ype Sfbgly50 (E at position 451), whereas this ratio decreased to 9 in the E451D mutant and to 1.3 in the E432C mutant of S. solfataricus b-glycosidase [12], finally reach ing a value of 0.4 in the E451S mutant, in which the preference f or galactosides is 2.5-fold higher than for glucosides. It was concluded that modification in the ratio between g4e and g4a energy is a key factor in changing the Sfbgly50 preference for glucosides v s. galactosides. Moreover, as the length of the residue 451 side-chain determines this ratio, the mutation series E fi D fi S may be useful in changing the preference for glucosides and galactosides of other family 1 b-glycosidases. The preference for fucosides was also affected by the replacement of E 451 with other residues. Introduction of Q at this position increased the preference of Sfbgli50 for f ucosides by 17-fo ld c ompared with galactosides (ðk cat =K mNPbfuc Þ=ðk cat =K mNPbgal Þ¼39 for t he wild-t ype Sfbgly50, whereas ðk cat =K mNPbfuc Þ=ðk cat =K mNPbgal Þ¼669 for m utant E 451Q) (Table 2). T his s pecificity change resulted from an increase in the steric h indrance between the 6-OH and residue 451 (interaction g6; Table 6). As galactosides and fucosides differ only in t he 6-OH, which is lacking in fucosides, the ES à complexes formed with galactosides is less stable than that formed with fucosides. Introduction of D and S at position 451 had a less significant effect on the Sfbgli50 preference f or fuco- sides vs. galactosides (ðk cat =K mNPbfuc Þ=ðk cat =K mNPbgal Þ¼ 39 for the wild-type Sfbgly50, w hereas ðk cat =K mNPbfuc Þ= ðk cat =K mNPbgal Þ¼108 for mutant E451D and 50 for mutant E451S) (Table 2) because these replacements did not significantly change g6 interaction (Table 6). The effect on the preference for fucosides by mutation E432C in the S. solfataricus b-glycosidase [12] is similar t o t hat o f t he mutations E451S and E451D in Sfbgly50, suggesting that a Table 5. Energies of noncovalent interactions between different residues at position 39 a nd glycone hy droxyls 4 and 6 in the enzyme–transition state (ES à )complex.4, glycone hydroxyl 4; 6, glycone hydroxyl 6; e, equatorial position; a, axial position; /4e, nonconvalent interaction between residue 39 and equatorial 4-OH; /4a, nonconvalent interac- tion between r esidue 39 and axia l 4-OH; /6, no nc ovalent interaction between residue 39 and 6 -OH. The r esults are expressed as DG à (kJÆmol )1 ). DG à values (activation energy of the glycosylation step) were c alculated referring to the disru ption of the interaction, thus positive values indic ate an interaction that stabilizes the ES à complex; the opposite is true fo r negative values. Noncovalent interaction Amino acid residue at position 39 Q a EN /4e b 19.3 ± 0.6 8.9 ± 0.3 6.1 ± 0.3 /4a b 14.1 ± 0.3 8.2 ± 0.3 3.1 ± 0.2 /6 3.3 ± 0.4 2.8 ± 0.3 ) 2.0 ± 0.3 a Q is the residue present in wild-type Sfbgly50, data were obtained from a previous publication [7]. b These values include /3 (inter- action between residue 451 and 3-OH). Table 6. Energies of noncovalent interactions between different residues at position 451 and the glycone hydroxyls 4 and 6. 4, glyc one hydroxyl 4; 6, glycone hydroxyl 6; e, equatorial position; a, axial position; g4e, noncovalent interaction between r esidue 451 and equato rial 4-OH; g4a, noncovalent interaction between residue 4 51 and axial 4-OH; g6, noncovalent i nterac tion b etween r esidue 451 and 6-OH. The results are expressed as DG à (kJÆmol )1 ). The DG à values (activation energy of the glycosylation step) were calculated referring to the disruption of the interaction, thus positive values i nd icate an inte raction that st ab ilizes the enzyme–transition state (E S à ) complex; the opposite is true for negative values. Noncovalent interaction Residue at position 451 E a QDS g4e 33.2 ± 0.7 32.2 ± 0.5 21.3 ± 0.6 10.6 ± 0.5 g4a 18.7 ± 0.2 13.6 ± 0.2 8.0 ± 0.1 4.8 ± 0.1 g6 )4.4 ± 0.4 )11.5 ± 0.3 )6.9 ± 0.3 )5.0 ± 0.3 a Data from a previous publication [7]. Ó FEBS 2004 Molecular basis for specificity of a b-glycosidase (Eur. J. Biochem. 271) 4175 steric hindrance with the 6-OH is also present in the S. solfataricus enzyme. Additionally, the mutation E417S in the P. furiosus b-glucosidase increased the rate of hydrolysis of 6-phospho b-galactosides, indicating that shortening the residue at position 417 (which is equivalent t o E451 i n Sfbgly50) set more room in the a ctive s ite t o t he binding of groups attached to the 6-OH [11]. Accordingly, 6-phospho b-glycosidases have a serine at a positio n corresponding to E451. Therefore, steric hindrance between the 6-OH and E451 seems to be a widespread phenomenon among family 1 b-glycosidases. The Sfbgly50 preference for fucosides is also affected by replacements at position 39. Indeed the introduction of N at this position increased by eightfold, the preference for this s ubstrate ( ðk cat =K mNPbfuc Þ=ðk cat =K mNPbgal Þ¼39 for the wild-type Sfbgly50, whereas ðk cat =K mNPbfuc Þ= ðk cat =K mNPbgal Þ¼336 f or Q39N mutant) ( Table 2), a specificity change resulting from /6 disruption (Table 6). Thus, the ES à complex with galactosides is destabilized, whereas a complex with fucosides, which lack 6-OH, is not affected. In summary, all mutations at position 39 and 451 resulted in an increase in the preference of Sfbgly50 for fucosides, because both of these mutations disrupt /6 (an interaction that stabilizes the ES à complex w ith galactosides a nd glucosides) or increase the steric hindrance with 6-OH (g6, a interaction that destabilizes the E S à complex with galactosides and glucosides). Design of b-glycosidases and modification in specificity The interaction energies involving substrates and three different r esidues (Q, E and N) at position 3 9 and four residues at position 451 (E, Q, D and S) in Sfbgly50 were considered here. Using the wild-type and mutant residues at those p ositions, w e can identify 12 different Sfbgly50 active sites. Six of these mutants – the wild-type enzyme (Q39E451) and five mutants (E39E451, N39E451, Q39D451, Q39N451 and Q 39S451) – had already been characterized, but six double mutants (E39D451, E39N451, E39S451, N 39D451, N39N451 and N39S451) remain to be studied. However, it is possible to predict the specificity of these Sfbgly50 double mutants, as long as we assume the tested noncovalent interactions to remain independent (replacing one residue does not affect the interactions formed by another) and ES à complex energy to b e determined by the sum o f the energy values of all noncovalent interactions formed by their active site residues with 4 -OH and 6-OH of different substrates (glucosides, galactosides and fucosides). The preferred substrate (presenting the highest rate of hydrolysis) would b e that w hich forms the noncovalent interactions set with the highest possible energy. T his strategy of designing and predicting the specificity o f new Sfbgly50 mutants is currently being tested in our lab oratory. Results with Sfbgly50 mutants s howed that changes in enzyme specificity resulted from destabilization of the ES à complex, which caused Sfbgly50 mutants to be less active than the in wild-type counterparts. Sim ilar r esults were observed in other studies aiming to change b-glycosidase specificity through mutagenesis [10–12,22]. This s uggests that simultaneous mutations of other residues, perhaps even outside the active site, are required for a fine tuning of the active site structure, in order to increase the catalytic activity of the mutant b-glycosidases. However, the identification of these residues is not an easy task owing to the large number of amino acid residues that require to be tested. The utilization of in vitro evolution could help us to solve this problem [8,9,23]. Therefore, the strategy suggested here, to p roduce efficient b-glycosidases with planned specificity, would be to perform selection of the appropriate residues at positions 39 and 451, followed by in v itro-directed ev olution. Acknowledgements This project is supported by FAPESP (Fundac¸ a ˜ odeAmparoa ` Pesquisa do Estado de Sa ˜ o Paulo) and CNPq (Conselho Nacional d e Desen- volvimento Cientı ´ fico e Tecnolo ´ gico).We than k A. Wu and R. Dillon for critically reading our ma nuscript. S.R.M., W.R.T. and C.F. are staf members of the Biochemistry Department an d research fellows of CNPq. E.H.P.A. is an undergraduate student and a research fellow of CNPq. References 1. Coutinho, P.M. & Henrissat, B. (1999) Carbohydrate-active enzymes: an integrated da tabase approach. In Recent Advances in Carbohydrate Bioengineering (Gilbert, H.J., Davies, G., Henrissat, B. & S vensson, B ., e ds), pp . 3–12. The Royal Society o f C hemi stry, Cambridge. 2. Fersht, A. (1999) Structure and Mechanism i n Protein Scie nce: a Guide to Enzyme Catalysis and Protein Folding.W.H.Freeman, 4 New York. 3. Namchuk, M.N. & Withers, S.G. (1995) Mechanism of Agro- bacterium b-glucosidase: kinetic a nalysis of the r ole of nonconva- lent enzyme/substrate interactions. Biochemistry 34, 16194–16202. 4. Bauer, M.W. & Kelly, R.M. (1998) The family 1 b-glucosidases from Pyrococcus furiosus and Agrobacterium f aecalis shar e a common catalytic mechanism. Biochemistry 37 , 17170–17178. 5. Paloma, F., Can ˜ ada, F.J., Barbero-Jime ´ nez, J . & M artı ´ n-Lomas, M. (1995) Su bstrate sp ecificity of small-intestinal lactase: stud y of the steric e ffects and hydrogen bonds involved in enzyme–sub- strate interaction. Carbohydrate Res. 271, 31–42. 6. Hays, S.W., VanderJagt, D.J., Bidisha, B., Serianni, A.S. & Glew, R.H. (1998) Catalytic mechanism and specificity for hydrolysis and transglycosylation reactions of cytosolic b-glucosidase from guinea pig liver. J. Biol. Chem. 273, 34941–34948. 7. Marana, S.R., Terra, W.R. & Ferreira, C. (2002) The role of amino acid residues Q39 and E 451 i n the determination o f sub- strate specificity of the Spodo ptera frugiperda b-glycosidase. Eur. J. Biochem. 269, 3705–3714. 8. Ly, H.D. & Withers, S.G. (1999) Mutagenesis of g lycosidases. Annu. Rev. Biochem. 68, 487–522. 9. Rye, C.S. & Withers, S.G. (2000) Glycosidase mechanisms. Curr. Opin. Chem. Biol. 4, 573–580. 10. Schulte, D. & H engstenber, W. (2000) Engineering the active center of the 6-phospho-b-galactosidase fro m La ctoc occu s la ctis. Protein Eng. 13 , 515–518. 11. Kaper, T., Lebb ink, J.H.G., P ouwe ls, J., Ko pp, J., S chul z, G.E., van der Oost, J. & de Vos, W.M. (2000) Comp arative structural analysis and substrate specificity engineering of the hyper- thermostable b-glucosidase CelB from Py rococcus fur iosus. Biochemistry 39, 4963–4970. 12. Corbett, K., Fordham-Skelton, A.P., Gatehouse, J.A. & Davis, B.G. (2001) Tailoring the substrate specificity of the b-glycosidase from the thermophilic archeon Sulfolobus solfataricus. FEBS Lett. 509, 355–360. 4176 S. R. Marana et al.(Eur. J. Biochem. 271) Ó FEBS 2004 13. Tabor, S. & Richardson, C.C. (1985) A bacteriophage T7 RNA polymerase/promoter system for controlled exclusive expression of specific genes. Pr oc. Natl Acad. Sci. USA 82, 1074–1078. 14. Marana, S.R., Jacobs-Lorena, M., T erra, W.R. & Ferreira, C. (2001) Amino acid residues involved in substrate binding and catalysis in an insect digestive b-glycosidase. Biochim. Biophys. Acta 1545, 41–52. 15. Terra, W.R., F erreira, C. & Bianchi, A.G. (1979) D istribution of digestive enzymes amon g the en do and e ctoperitho phic sp aces and midgut cells o f Rhynchosciara a mericana and its physiological significanc e. J. Insect Physiol. 25, 487–494. 16. Laemmli U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 17. Blum, H., Beir, H. & Gross, H. (1987) Improved silver staining of plant proteins, RNA and DNA in polyacrylamide gels. Electro- phoresis 8, 93–99. 18. Gill, S.C. & von Hippel, P.H. (1989) Calculation of p rotein extinction coefficients from amino acid sequence data. Anal. Bio- chem. 182, 319–326. 19. Kempton, J.B. & Withers, S.G. (19 92) Mechanism of Agro- bacterium b-glucosidase: kinetic studies. Biochemistry 31, 9961– 9969. 20. Ippolito, J.A., Alexander, R.S. & Christianson, D. W. (1990) Hydrogen bond stereochemistry in protein structure and function. J. Mol. Biol. 215, 457–471. 21. Cornell, W.D., Cieplak, P., Bayly, C.I., Gould, I.R., Merz, K.M., Ferguson, D.M., Spellmeyer, D.C., Fox, T., Caldwell, J.W. & Kollman, P.A. (1995) A second generation force field for the simulation of proteins, nucleic ac ids and organic compounds. J. Am. C hem. Soc. 117 , 5179–5197. 22. Lebbink, J.H.G., Kaper, T., Kengen, S.W.M., van der Oost, J. & de Vos, W.M. (2001) b-glucosidase CelB from Pyrococcus furio sus: production by Escherichia coli, purification and in vitro evolution. Methods Enzymol. 330, 364–379. 23. Zhang, J., Dawes, G. & Stemmer, W.P.C. (1997) Directed evo- lution of a fucosidase from a galactosidase by DNA shuffling and screening. Proc.NatlAcad.Sci.USA94, 4504–4509. 24. Burmeister,W.P.,Cottaz,S.,Driguez,H.,Iori,R.,Palmieri,S.& Henrissat, B. (1997) The crystal structures of Sinapis alba my ro- sinase and a covalent glycosyl-enzyme intermediate provide insights into the substrate recognition and active-site machinery of an S-glycosidase. Structure 15, 663–675. Supplementary material The following material is available from http://www. blackwellpublishing.com/products/journals/suppmat/EJB/ EJB4354/EJB4354sm.htm Fig. S1. Purfication of the recombinant mutant Sfbgly50. Ó FEBS 2004 Molecular basis for specificity of a b-glycosidase (Eur. J. Biochem. 271) 4177 . Investigation of the substrate specificity of a b-glycosidase from Spodoptera frugiperda using site-directed mutagenesis and bioenergetics analysis Sandro. was 5¢- GGAGTCTAATGGACAACTTTNNNTGGATGGA GGGTTATATTGAGCG-3¢,withGAC,CAAandTCA as mutated codons for E 451D, E451Q and E451S, respectively. DNA sequencing w as