The role of residues R97 and Y331 in modulating the pH optimum of an insect β-glycosidase of family 1


  • Enzyme: digestive β-glycosidase from Spodoptera frugiperda (β-d-glucoside glucohydrolase; EC; GenBank accession no. AF052729).

S. R. Marana, Departamento de Bioquímica, Instituto de Química, USP, CP 26077, São Paulo, 05513–970, Brazil.
Fax: +55 11 30912186,


The activity of the digestive β-glycosidase from Spodoptera frugiperda (Sfβgly50, pH optimum 6.2) depends on E399 (pKa = 4.9; catalytic nucleophile) and E187 (pKa = 7.5; catalytic proton donor). Homology modelling of the Sfβgly50 active site confirms that R97 and Y331 form hydrogen bonds with E399. Site-directed mutagenesis showed that the substitution of R97 by methionine or lysine increased the E399 pKa by 0.6 or 0.8 units, respectively, shifting the pH optima of these mutants to 6.5. The substitution of Y331 by phenylalanine increased the pKa of E399 and E187 by 0.7 and 1.6 units, respectively, and displaced the pH optimum to 7.0. From the observed ΔpKa it was calculated that R97 and Y331 contribute 3.4 and 4.0 kJ·mol−1, respectively, to stabilization of the charged E399, thus enabling it to be the catalytic nucleophile. The substitution of E187 by D decreased the pKa of residue 187 by 0.5 units and shifted the pH optimum to 5.8, suggesting that an electrostatic repulsion between the deprotonated E399 and E187 may increase the pKa of E187, which then becomes the catalytic proton donor. In short the data showed that a network of noncovalent interactions among R97, Y331, E399 and E187 controls the Sfβgly50 pH optimum. As those residues are conserved among the family 1 β-glycosidases, it is proposed here that similar interactions modulate the pH optimum of all family 1 β-glycosidases.




4-methylumbelliferyl β-d-glucopyranoside






digestive β-glycosidase (Mr 50 000) from Spodoptera frugiperda


enzyme substrate

The β-glycosidases from glycoside hydrolase family 1 are enzymes that remove monosaccharides from the nonreducing end of di- and/or oligosaccharides. According to the CAZy website this family comprises 422 sequenced β-glycosidases, of which the tertiary structure of 12 has been determined. Together with families 2, 10, 17, 26, 30, 35, 39, 42, 51, 53, 59, 72, 79 and 86 family 1 forms clan A, a group of families that shares structural and catalytic similarities [1]. All β-glycosidases of family 1 present the same tertiary structure [the (β/α)8 barrel], they are configuration-retaining glycosidases and their catalytic activity depends on two glutamic acid residues, one positioned after β strand 4 and the other after β strand 7 [1]. These glutamic acids are very close inside the active site (about 4.5 Å apart) [2], and during the reaction the first glutamic acid acts as proton donor, and the second acts as a nucleophile. The catalytic nucleophile pKa is around 5.0 and the catalytic proton donor pKa is around 7.0 [3–7].

A plot of β-glycosidase activity vs. pH presents a bell shape, indicating that in the pH optimum the catalytic nucleophile is deprotonated and the catalytic proton donor is protonated. Hence the branch of the curve below the pH optimum is determined mainly by the ionization of the catalytic nucleophile, whereas the catalytic ionization of the proton donor determines the branch above the pH optimum.

As the β-glycosidase activity depends on the finely tuned ionization of the catalytic nucleophile and proton donor, it is necessary to understand how the ionization of these residues is modulated in order to determine how the pH optimum is controlled. Such data are lacking for family 1. In family 11 xylanases, it was proposed that the negatively charged nucleophile electrostatically destabilizes the proton donor ionization, increasing its pKa[8]. Thus, one may hypothesize that the ‘electrostatic coupling’ between the catalytic glutamates could also operate in family 1, despite the fact that the family 11 and 1 belong to different clans. However, even this interaction is not enough to determine which of the glutamic acid residues will be the catalytic nucleophile or the proton donor. This is because for β-glycosidases it is not known how the ionization of each catalytic glutamate is modulated and therefore there is no model to explain how the pH optimum is determined.

In this work a digestive β-glycosidase from Spodoptera frugiperda (Sfβgly50) was used as an experimental model to fill those gaps. This enzyme had been previously sequenced (GenBank accession number AF052729) and it was classified in the glycoside hydrolase family 1. Catalytic activity of Sfβgly50 depends on residues E399 (catalytic nucleophile, pKa = 4.9) and E187 (catalytic proton donor, pKa = 7.5). The effect of pH on Sfβgly50 activity is a typical bell-shaped curve and the pH optimum is 6.2 [9].

In the active site of the family 1 β-glycosidases the catalytic nucleophile and proton donor are very close (4.5 Å apart). Hence differences in their ionization and resulting catalytic roles should rely on noncovalent interactions that are specific for each residue. Structural data on different family 1 β-glycosidases showed out that the catalytic nucleophile interacts with an arginine and a tyrosine, putatively being stabilized by them [10–13]. Homology modelling the Sfβgly50 active site confirms that Y331 and R97 are positioned very close to E399 (less than 3 Å apart), suggesting that these residues are hydrogen bonded with the catalytic nucleophile (Fig. 1). Therefore, these interactions may affect the E399 ionization so that this residue becomes the catalytic nucleophile, whereas an electrostatic repulsion between E399 and E187 cause this last one to function as the catalytic proton donor. Thus, interactions between E399 and Y331 or R97 may be key elements in the determination of Sfβgly50 pH optimum.

Figure 1.

Schematic representation of the Sfβgly50 active site. E399 is the nucleophile and E187 is the proton donor. Y331 and R97 form hydrogen bonds (dotted lines) with E399 (Y331 Oη atom to E399 Oε1 atom = 2.69 Å and R97 Nη1 atom to E399 Oε2 atom = 2.77 Å). The distance between E399 and E187 side chains is 4.5 Å.

The role of arginine in the modulation of the β-glycosidase pH optimum had not been studied before. The substitution of Y298 for phenylalanine in Agrobacterium sp. β-glucosidase affected the rate constant of the glycosylation step and also changed the enzyme pH optimum [14]. Despite that, the effect of the tyrosine on the pKa values of the β-glycosidase catalytic glutamates still remains to be determined and quantified.

To fill these gaps, residues Y331, R97 and E187 of Sfβgly50 were replaced through site-directed mutagenesis by phenylalanine (Y331F), methionine (R97M), lysine (R97K) and aspartate (E187D). The effect of pH on the activity of the recombinant enzymes were then determined.

Materials and methods


All reagents, unless otherwise specified, were purchased from Sigma or Merck.

Site-directed mutagenesis

Site-directed mutagenesis was performed using as template the plasmid pT7-7 [15] containing as insert a DNA fragment that encodes the mature Sfβgly50 (pT7β50) [9]. The experiments were carried out following the instructions of the QuikChange site-directed mutagenesis kit (Stratagene). Primers used were: R97K, 5′-GCCTGGACGCTTACAAGTTCTCCCTCTCCTG-3′ and 5′-CAGGAGAGGGAGAACTTGTAAGCGTCCAGGC-3′; R97M, 5′-GCCTGGACGCTTACATGTTCTCCCTCTCCTG-3′ and 5′-CAGGAGAGGGAGAACATGTAAGCGTCCAGGC-3′; Y331F, 5′-GATCGGAGTGAACCACTTCACAGCATTCCTGGTATC-3′ and 5′-GATACCAGGAATGCTGTGAAGTGGTTCACTCCGATC-3′; E187D, 5′-GTTCATCACTTTCAACGATCCTAGAGAGATTTGCTTTGAG-3′ and 5′-CTCAAAGCAAATCTCTCTAGGATCGTTGAAAGTGATGAAC-3′. Codons in bold show the mutations. The incorporation of the mutated codon in the pT7β50 was checked through DNA sequencing.

Expression of recombinant Sfβgly50

NovablueDE3 (Novagen) cells were cotransformed with pT7β50 (encoding wild-type or mutant type Sfβgly50) and pT-GroE, a plasmid encoding the chaperone GroELS under the control of the T7 RNA polymerase promoter. pT-GroE increases the Gro-ELS concentration inside the cells and consequently it favours the folding of coexpressed proteins [16]. The transformed bacteria were cultivated (37 °C, 250 r.p.m.) in Luria–Bertani broth containing carbenicillin (50 µg·mL−1) and chloramphenicol (17 µg·mL−1) until D600 = 0.6–1.0 was reached. The bacteria were then induced using 1 mm isopropyl thio-β-d-galactoside for 3 h (25 °C, 250 r.p.m.) and harvested at 7000 g for 20 min at 4 °C. The pellets were stored at −80 °C. Samples of induced and noninduced cells were analysed by SDS/PAGE [17] to detect the expression of the recombinant β-glycosidases.

Lysis of induced bacteria

Induced bacteria were suspended in 50 mm Hepes buffer pH 7.0 containing 150 mm NaCl, 0.02% (w/v) lysozyme (chicken egg white) and 0.1% (v/v) Triton X-100. The suspension was incubated at 4 °C with slow shaking (3 r.p.m.). After 30 min, the cells in the suspension were disrupted using a sonicator adapted with a micro tip (five pulses of 30 s at output 4 in a Branson 250 sonicator) and harvested at 7000 g for 20 min at 4 °C. The supernatant was stored at −20 °C and used as source of recombinant β-glycosidase.

Purification of the recombinant Sfβgly50

Soluble material from the induced cells containing the wild-type or mutant recombinant Sfβgly50 was loaded onto a phenylSuperose HR 10/10 column (Pharmacia Biotech). The nonretained proteins were washed out with 50 mm phosphate buffer pH 7.0 containing 1.27 m (NH4)2SO4, and the retained proteins were then eluted using a gradient of (NH4)2SO4 prepared in 50 mm phosphate buffer pH 7.0. The presence of the recombinant Sfβgly50 was detected by enzymatic assay using NPβglc (p-nitrophenyl β-d-glucopyranoside) as substrate [18]. Fractions containing β-glycosidase activity were pooled and dialysed in 20 mm triethanolamine buffer pH 8.0, and the dialysed material was loaded onto a Resource Q column (Amersham Bioscience). Nonretained proteins were washed out with 20 mm triethanolamine buffer pH 8.0 and retained proteins were eluted using a gradient of NaCl prepared in the same initial buffer. The presence of the recombinant Sfβgly50 was detected as above and its purity ascertained by SDS/PAGE followed by silver staining [19].

Protein determinations were performed spectrophotometrically (absorbance in 280 nm) using ε280 = 117 200m−1·cm−1[20]. The same protocol was used to purify the wild-type and mutant Sfβgly50.

The native Sfβgly50 was purified from the S. frugiperda midgut following the procedures described previously [21].

Kinetic analysis

All assays were performed at 30 °C in 50 mm citrate/phosphate buffer pH 6.0 and initial rate data measured. Hydrolysis of MUβglc (4-methylumbelliferyl β-d-glucopyranoside) was followed by MU fluorescence [22]. Kinetic parameters (kcat and Km) were determined by using nine different substrate concentrations (0.1–8 mm); enzyme concentrations were 0.13 µm for R97M, 0.09 µm for Y331F and 0.62 µm for E187D. The data were fitted to Michaelis–Menten equation using the enzfitter (Elsevier-Biosoft).

Chemical modification with phenylglyoxal

Arginine modification was performed using different concentrations of phenylglyoxal (1, 3, 4 and 5 mm) prepared in 20 mm phosphate buffer pH 8 at 30 °C. In this pH phenylglyoxal reacts specifically with arginine residues [23,24]. Wild-type (0.49 µm) or mutant Sfβgly50 (0.13 µm) samples were incubated with the modifying agent in the absence or presence of high concentration of NPβglc (> 4 × Km). Samples were collected after different periods of time and 10 mm arginine in 20 mm phosphate pH 8.0 was added. This material was used to determine the remaining enzymatic activity using 4 mm MUβglc as substrate in 50 mm citrate/phosphate buffer pH 6.0. Then, the rate constants (kobs) for the Sfβgly50 inactivation in different phenylglyoxal concentrations were calculated.

pH effect on the Sfβgly50 activity

Sfβgly50 enzymatic activity on 8 mm MUβglc was determined in different buffers ranging from pH 5.0 to 8.5 (50 mm citrate/phosphate buffer, pH 5.0–7.0; 50 mm phosphate buffer, pH 7.0–8.0; 50 mm Bicine buffer 7.0–8.5). The pH stability of Sfβgly50 was checked by incubation in the same buffers for a time equal to the assay time and then determining the activity remaining at the optimum pH. To correct the pH shifts due to differences in temperature, pH of the assay media was measured in substrate/buffer mixtures at 30 °C. The enzyme concentration was 0.13 µm for mutant R97M, 0.09 µm for mutant Y331F and 0.37 µm for mutant E187D.

At 8 mm MUβglc, Sfβgly50 is approaching saturation by the substrate. Hence, relative activity is a good approximation of the relative maximum velocity (Vmax app) under these conditions. Thus the pKas in the enzyme–substrate (ES) complex of the catalytically active groups of Sfβgly50 (pKES1 and pKES2) were determined by fitting the Vmax app of the MUβglc hydrolysis at each pH to Eqn (1)[25].


Vmax app is the relative Vmax determined at each pH, [H+] is the proton concentration and KES1 and KES2 are the ionization constants of the two catalytically essential groups in the ES complex. Vmax app was expressed as a percentage of the highest Vmax observed, and fitting was done using the software enzfitter.

Ionization constants in the free enzyme (pKE1 and pKE2) were calculated using Eqn (2)[25].


kcat/Km app is the relative kcat/Km determined at each pH, [H+] is the proton concentration and KE1 and KE2 are the ionization constants of the two catalytically essential groups in the free enzyme. kcat/Km app may be calculated from the enzymatic activity determined using low substrate concentration (0.25 mm) at different pH values and relative kcat/Km app and [H+] were fitted in the above equation using enzfitter.

The data were enough to fit simultaneously the two KES and KE. In the E187D mutant, the determination of pKES1 was less accurate than that of pKES2, because many more points above the pH optimum were obtained. However, it was not possible to go below pH 5.0 because Sfβgly50 becomes unstable.

Homology modelling

The three-dimensional structure of Sfβgly50 was modelled according to structural data for Bacillus polymyxaβ-glucosidase A (1BGA, 1BGG, 1TR1), Trifolium repensβ-glucosidase 2 (1CBG) and Lactococcus lactis 6-phospho β-galactosidase (1PBG). Modelling was performed in the Swiss Model server and the result was visualized by pdbviewer[26].

Sequence alignment and structural comparison

Amino acid sequences of family 1 β-glycosidases were retrieved from the CAZy website [1] and aligned using the software clustalx[27]. The spatial coordinates of family 1 β-glycosidases were retrieved from the PDB website and visualized by pdbviewer[26].


The expression of recombinant wild-type and mutant Sfβgly50 was checked by SDS/PAGE (Fig. 2A). Recombinant Sfβgly50 was purified by a combination of hydrophobic and anion-exchange chromatography (Fig. 2B), resulting in a 50% recovery and a yield of about 0.2 mg purified Sfβgly50 from 0.5 L bacterial culture (Fig. 2C).

Figure 2.

Induction and purification of the recombinant Sfβgly50. (A) SDS/PAGE of proteins from NovablueDE3 cells transformed with plasmid pT7-7 containing Sfβgly50. Lane 1, Noninduced cells; lanes 2, 3, 4 and 5, cells induced to produce the mutants R97M, R97K, Y331F and E187D, respectively. The arrow indicates the recombinant Sfβgly50. The gel (10% polyacrylamide) was stained with Coomassie blue R. (B) The soluble material from the bacteria producing the R97M Sfβgly50 was loaded onto a Phenyl Superose HR 10/10 column eluted with a decreasing gradient of (NH4)2SO4, prepared in 50 mm phosphate buffer pH 7.0. β-Glycosidase activity (◆) was detected using 2 mm NPβglc prepared in 50 mm citrate/phosphate buffer pH 6.0. The two most active fractions were pooled. (C) Ion-exchange chromatography in Resource Q column of the β-glycosidase activity recovered in (B). Elution produced using a gradient of NaCl prepared in 20 mm triethanolamine buffer pH 8.0. β-Glycosidase activity (◆) was detected using NPβglc. The three most active fractions were pooled and analysed by SDS/PAGE. (D) SDS/PAGE of the purified Sfβgly50. The gel (10% polyacrylamide) was silver stained. The same procedure was used to purify the wild-type and mutant (R97K, Y331F and E187D) Sfβgly50. As the gels are not the same size the band positions are not directly comparable.

The kinetic parameters (kcat and Km) for the hydrolysis of MUβglc were determined for the Sfβgly50 mutants (R97M, Y331F and E187D) and compared with those of the wild-type enzyme (Table 1). All mutations resulted in a large activity decrease, indicating that R97 and Y331 do influence catalysis.

Table 1. Steady-state kinetic parameters for hydrolysis of MUβglucoside by recombinant wild-type and mutant Sfβgly50. The experiments were carried out at nine different substrate concentrations and the parameters were calculated using enzfitter.
Wild-type2.3 ± 0.11.73 ± 0.090.75 ± 0.06100
R97M1.9 ± 0.30.0030 ± 0.00020.0015 ± 0.00050.2
Y331F2.0 ± 0.50.0070 ± 0.00050.003 ± 0.0010.45
E187D4.4 ± 0.10.00147 ± 0.000020.00033 ± 0.000010.044

The pH-dependent activity profile is similar for the native and recombinant wild-type Sfβgly50 (Fig. 3). Hence the recombinant wild-type Sfβgly50 is useful as a control in comparisons with the Sfβgly50 mutants. Moreover, wild-type and mutant Sfβgly50 are stable in the pH range 5.0–9.0 (Fig. 3).

Figure 3.

Effect of pH on the activity of native (▪) and recombinant (○) wild-type Sfβgly50. The buffers used were 50 mm citrate/phosphate (pH 4.7–7.0), 50 mm phosphate (pH 7.0–8.0) and 50 mm bicine (pH 8.0–8.5). Each point is the average of four Sfβgly50 activity determinations using 4 mm MUβglc as substrate. The pH stability of the recombinant Sfβgly50 (▵) was checked by incubating the enzyme in the same buffers for an equal length of time and determining the remaining activity in the pH optimum.

As the kinetic data showed that R97 and Y331 influence catalysis, the function of residue R97 was investigated by performing a chemical inactivation of wild-type Sfβgly50 with phenylglyoxal. The reaction order was 1.7 in relation to phenylglyoxal and the inactivation was halted by the presence of saturating concentrations of substrate (Fig. 4). In contrast, phenylglyoxal did not inactivate R97M and R97K mutants (data not shown).

Figure 4.

Inactivation of the Sfβgly50 with phenylglyoxal. Effect of phenylglyoxal concentration (◆, 1 mm; ▪, 3 mm; ▴, 4 mm; •, 5 mm) on the inactivation rate of the wild-type recombinant Sfβgly50. Phenylglyoxal was prepared in 20 mm phosphate buffer pH 8.0. The inactivation order is 1.7 with phenylglyoxal as calculated from the insert. Enzymatic activity was detected using as substrate 4 mm MUβglc in 50 mm citrate/phosphate pH 6.0.

Enzyme activity–pH data showed that R97K and R97M Sfβgly50 mutants presented curves narrower and pH optimum (6.5) slightly higher than wild-type Sfβgly50 (6.2) (Fig. 5A,B). The mutant Y331F presented an activity–pH profile wider and a pH optimum (7.0) higher than the wild-type Sfβgly50, whereas the E187D mutant had a pH optimum (5.8) lower than the recombinant wild-type enzyme (Fig. 5C,D).

Figure 5.

Effect of pH on the relative maximum velocity(Vmaxapp) of the wild-type(○) and mutant Sfβgly50(▪). (A) R97K; (B) R97M; (C) Y331F; (D) E187D. The buffers used were 50 mm citrate/phosphate (pH 4.7–7.0), 50 mm phosphate (pH 7.0–8.0) and 50 mm bicine (pH 8.0–8.5). Each point is the average of four Sfβgly50 activity determinations using 8 mm MUβglc as substrate. The enzymes are stable in this pH range. Based on these data, pKES1 and pKES2 values were calculated as described in the Material and methods.

As the pKa values of the catalytic residues determine the pH optima, the ionization constants (pKa) of the catalytic glutamates (E187 and E399) in the ES complex and in the free enzyme (E) of Sfβgly50 mutants were determined. The pKES and pKE are very similar, thus only the pKES values are presented (Table 2). The differences observed in pKa values between the wild-type and mutant enzymes result from the modifications in the interactions between the catalytic glutamates and the residues R97 and Y331, indicating that these residues play a role in the modulation of the Sfβgly50 pH optimum.

Table 2. pKavalues of the catalytic groups in the ES complex of the wild-type and mutants Sfβgly50.
Wild-type4.8 ± 0.17.4 ± 0.1
R97K5.6 ± 0.17.7 ± 0.1+0.8+0.3
R97M5.4 ± 0.17.5 ± 0.1+0.6+0.1
Y331F5.5 ± 0.19.0 ± 0.1+0.7+1.6
E187D4.5 ± 0.16.9 ± 0.1−0.3−0.5

Sequence alignments and structural comparisons of family 1 β-glycosidases showed that R97 and Y331 are totally conserved and that these residues plus the nucleophile (E399) have the same spatial positioning (Fig. 6). Nevertheless, the distance between arginine and glutamate varies from 2.59 to 3.64 Å and the distance between the tyrosine and glutamate varies from 2.59 to 2.98 Å.

Figure 6.

Sequence alignment and structural comparison of family 1 glycoside hydrolases. (A) Sequence alignment of the regions containing the residue R97 and Y331 (Sfβgly50 numbering). The aligned β-glycosidases are from Actinomyces naeslundii AAK33123.1, Agrobacterium sp. AAA220851, Arabidopsis thaliana Q9SE50, Bacillus circulans Q03506, Bacillus polymyxa P22073, Brassica napus Q42618, Catharanthus roseus AAF28800.1, Cavia porcellus P97265, Clostridium longisporum Q46130, Escherichia coli K12 P11988, Lactobacillus caseii P14696, Lactococcus lactis P11546, Prunus serotina AAL06338.1, Pyrococcus woesei O52626, Sinapis alba P29092, Spodoptera frugiperda AF052729, Staphylococcus aureus P11175, Sulfolobus solfataricus P22498, Thermus nonproteolyticus AAF36392.1, Trifolium repens P26205, Zea mays P49235. An asterisk marks identical residues, a colon indicates strongly conserved residues and a period denotes weakly conserved residues. (B) The spatial position of the residues corresponding to Y331, E399 and R97 (Sfβgly50 numbering) in different glycoside hydrolases was superimposed. The spatial coordinates were retrieved from PDB: 1BGG, β-glycosidase from Bacillus polymyxa (black; R77, Y296 and E352); 2MYR, myrosinase from Sinapis alba (green; R95, Y330 and E409); 1CBG, cyanogenic β-glycosidase from Trifolium repens (orange; R91, Y326 and E397); 1PBG, phospho β-galactosidase from Lactococcus lactis (red; R72, Y299 and E374).


An inspection of the three-dimensional structure of some family 1 β-glycosidases [10–13] and of the structural model of the Sfβgly50 active site show that an arginine (R97) and a tyrosine (Y331) are very close (2.69 Å and 2.77 Å, respectively) and form hydrogen bonds with the side chain of E399. The hydrogen bond between R97 and E399 probably has a strong electrostatic component. However, determination of the relative contribution of each component in this interaction is not simple. On the other hand, E399 is also close to E187 side chain (4.5 Å) and these residues may interact electrostatically (Fig. 1). These noncovalent interactions may modulate the E187 and E399 ionization state and consequently determine the Sfβgly50 pH optimum. In order to test this hypothesis, some mutants (R97M, R97K, Y331F and E187D) were expressed as recombinant proteins in Escherichia coli. The general shape and volume of residues 97 and 331 side chains are conserved in the mutants R97M and Y331F, but the hydrogen bond-forming atoms have been removed. In mutant E187D, the distance between the catalytic nucleophile and the catalytic proton donor has been increased because the side chain of aspartic acid is shorter than that of glutamic acid.

The kinetic parameters for MUβglc hydrolysis (Table 1) show that the substitution of R97 and Y331 results mostly in a decrease in kcat, whereas Km is affected less, suggesting that these residues influence catalysis. As a member of the glycoside hydrolase family 1, the Sfβgly50 probably follows a double displacement mechanism with a glycosyl-enzyme intermediate. However, as it is not known which step (glycosylation or deglycosylation) of the hydrolysis of MUβglc by the mutant enzymes is rate-limiting, no hypothesis on the effect of R97 and Y331 on the rate constant of each step can be advanced.

The influence of R97 on catalysis is confirmed by the phenylglyoxal inactivation (Fig. 4), which is abolished by substrate and is not observed with the R97M and R97K mutants. The reaction order relative to phenylglyoxal (1.7) indicates that the enzyme is inactivated by the reaction of two phenylglyoxal molecules with one arginine residue. Despite the fact that the reaction mechanism is not clear (a dimer or two phenylglyoxal molecules may react with one arginine residue), the reaction order (1.7) is in agreement with the proportion found in reactions between phenylglyoxal and polypeptides (2 : 1) [23]. The structure of the putative reaction product [23] indicates that the modified R97 side chain is unable to hydrogen bond with E399, probably causing wild-type Sfβgly50 inactivation. Moreover, the addition of a bulky group (diphenylglyoxal) in the active site probably hinders substrate binding.

The substitution of R97 by M results in a 500-fold decrease in kcat, whereas the replacement of Y331 by F results in a 250-fold decrease (compare kcat for the wild-type and mutant Sfβgly50 in Table 1). As the extent of kcat decrease is similar in both cases, R97 and Y331 may have a similar influence on catalysis. In a β-glycosidase from Agrobacterium sp. (glycoside hydrolase family 1), the replacement of the residue equivalent to Y331 (Y298, Agrobacterium numbering) for a phenylalanine also results in a large kcat decrease (500-fold) [14].

One possible effect of R97 and Y331 on catalysis is to position E399. Thus, the kcat decrease could result from an incorrect positioning of the catalytic nucleophile, but the data presented here are not enough to support this hypothesis.

In the case of the Y331F mutant part of the decrease in kcat may result from destabilization of the ES complex, because the tyrosine residue is thought to interact with the oxygen of the glycone ring in that complex [10,12]. Nevertheless, it is not possible to assume that the same is occurring in the mutant R97M, because there is no data on the interactions between the substrate and the arginine residue.

Finally, another aspect of the R97 and Y331 influence on catalysis that must be taken into account is the modulation of the ionization state of the catalytic glutamates. Indeed, the substitution of R97 by M, which disrupts the hydrogen bond between residue 97 and E399, shifted the E399 pKa by +0.6 pH units (from 4.8 to 5.4), but had no effect on the E187 pKa. As a consequence of the higher pKa of E399, the mutant R97M has a pH optimum (6.5) slightly higher than that of wild-type Sfβgly50 (6.2).

Although the introduction of a methionine residue at position 97 could have changed the dielectric constant of the active site, the deletion of the hydrogen bond between R97 and E399 is probably a major cause of the shift in the pKa of E399. Therefore, R97 facilitates the ionization of the catalytic nucleophile by stabilizing its charged state.

The observed ΔpKa are directly related to the differences in the free energy change (ΔΔG° = 2.303RTΔpKa) of ionization of the groups in wild-type and mutant enzyme. This ionization differs because of the stabilizing effect provided by R97, which is lacking in the mutant R97M. Hence, the ΔΔG° is equal to the energy of the stabilizing effect provided by R97. Thus, based on the ΔpKa of E399 between the wild-type and R97M Sfβgly50, it was calculated that R97 contributes 3.4 ± 0.4 kJ·mol−1 to stabilize the charge of E399.

In the R97K mutant the pKa of E399 is shifted by +0.8 pH units, whereas pKa of E187 changed by +0.3 pH units (Table 2). Taking into account the experimental errors, the pKa of E187 remained the same, but the pKa of E399 clearly increased. Actually, the substitution of R97 by methionine or lysine resulted in the same increment in the pKa of E399 (Table 2). Therefore, M97 and K97 are equally effective in stabilizing the charged E399. That is unexpected, because as arginine and lysine side chains are positively charged and hydrogen bond donors, they should interact similarly with E399. However, it should be considered that structural data of high resolution protein structures indicate that the geometry of the hydrogen bonds formed by those residues are very different [28] and that lysine side chain is shorter. Thus, the present results suggest that in spite of the fact that K97 could form a hydrogen bond with E399, this bond is weakened or disrupted because of unfavourable spatial positioning of interacting atoms. Alterations in the active site structure could also contribute to the observed result.

In the Y331F mutant the replacement of Y331 for phenylalanine shifted the pKa of E399 by +0.7 pH units (from 4.8 to 5.5) and the pKa of E187 by +1.6 pH units (7.4–9.0). As a result, the pH optimum of the Y331F mutant (7.0) is higher than that of the wild-type Sfβgly50 (6.2) (Fig. 5). The effect of this mutation on the E399 ionization is the same as observed for the mutation R97M (ΔpKES1 = + 0.6; Table 2), indicating that Y331 also stabilizes the charged E399. Part of this effect may result from an alteration of the dielectric constant of the active site, although the deletion of the hydrogen bond between Y331 and E399 is probably a major component of that pKa increment. Based on the ΔpKa of E399 between the wild-type and Y331F Sfβgly50 it was calculated that the Y331 contributes 4.0 ± 0.4 kJ·mol−1 to the stabilization of the charged E399, the same value observed for R97. Therefore, R97 and Y331 together contribute 7.4 kJ·mol−1 to stabilization of the charged E399. Hence, if these two residues were removed, the ionization of E399 would be less favourable and the pKa of E399 would be higher, probably around 6.0.

In the Y331F mutant, the pKa of E187 was increased by +1.6 pH units. This modification is not a result of a direct interaction between E187 and Y331 as these residues are far apart from each other. A possible explanation is that the increase in the pKa of E399 makes more difficult the ionization of E187 due to an increment in electrostatic repulsion between these glutamates. However, this repulsion is not enough to completely explain the ΔpKa of E187 in the Y331F mutant, because in the R97K and R97M mutants the same increment in the pKa of E399 did not change significantly the pKa of E187. This suggests that in the Y331F mutant, the charged side chain of E399 may have moved to a new position closer to E187, in order to minimize unfavourable interactions with the apolar side chain of F331. Thus, the increment in the pKa of E399, in addition to it being closer to E187, may have resulted in a large shift in the pKa of E187. Changes in the dielectric constant due to F331 may further increase the pKa of E187.

This unexpected shift in the pKa of E187 cannot be directly compared with data from other β-glycosidases, but the effect of the Y331F mutation in the pH-dependent activity profile is very similar to that observed in the mutant Y298F of the Agrobacteriumβ-glycosidase [14]. The results here obtained are similar to those found for a family 11 xylanase interaction between a tyrosine and a charged glutamate. The deletion of a hydrogen bond between a tyrosine and the catalytic nucleophile (glutamate) in the xylanase also resulted in a large shift (+1.6 pH units) in proton donor pKa[29]. However, this comparison is not strong because family 11 does not belong to clan A [1], implying in different active site structure and composition.

The mutation E187D also resulted in a large decrease in kcat for MUβglc hydrolysis (Table 1), which is explained by the aspartic acid being less efficient as a catalytic proton donor. That happens because the short D187 side chain is not as close to the glycoside bond as is the E187 side chain does and so proton donation to the leaving group (aglycone) is more difficult. In this mutant the proton donor pKa (residue 187) had a shift of −0.5 pH units (from 7.4 to 6.9) whereas, considering the errors, the nucleophile (E399) had no pKa change (Table 2). A possible explanation for this result is that there is an electrostatic repulsion between the charged E399 and E187. Thus, in the E187D mutant, this repulsion was reduced because carboxyl groups are farther apart. Consequently the proton donor ionizes more easily (pKa decreased) and the pH optimum was shifted to a value (5.8) lower than that of the wild-type Sfβgly50 (Fig. 5). Otherwise, the ΔpKa of the catalytic proton donor mirrors the pKa difference between free glutamic and aspartic acids (0.4 units), suggesting that an electrostatic interaction does not have any influence on the ΔpKa. However, the pKa values of free aspartic and glutamic acids side chains were determined in water, thus they do not have necessarily the same properties in an environment hidden from the solvent like that of the active site (less than 5% of the E187 area is exposed to the solvent). In these conditions the ionization of aspartic and glutamic acids may be equally unfavourable.

Thus, if this hypothesis is correct, the ΔpKa of the catalytic proton donor may really indicate the presence of an electrostatic repulsion between residues E187 and E399. This hypothesis is further supported by the results obtained with β-glucosidase from Agrobacterium sp. (family 1) [4]. Here, the replacement of the catalytic nucleophile (E358) by an aspartic acid resulted in a decrease of 0.9 pH units in the pKa of the catalytic proton donor – a result also interpreted as an indication of an electrostatic repulsion between the catalytic glutamates [4]. An electrostatic repulsion between the catalytic glutamates was already described in a xylanase from family 11 [8], although one must be cautious with this comparison, as noted above.

In conclusion, the combination of these results shows that residues R97 and Y331 modulate the ionization of residue E399 by stabilizing its charge and reducing its pKa, thus enabling it to function as a nucleophile. An electrostatic repulsion between ionized E399 and E187 may make E187 ionization more difficult, increasing its pKa and favouring a role as catalytic proton donor. Finally, as the pH optimum of the wild-type and mutant Sfβgly50 is an average of the E399 and E187 pKa values, it is concluded that the pH optimum of Sfβgly50 is determined by a noncovalent bonds network among R97, Y331, E399 and E187.

Residues Y331 and R97 are totally conserved in different family 1 β-glycosidases from very different organisms (Fig. 6). The conservation of these residues suggests that, in spite of the difference in the physiological role and large evolutionary distance between the enzyme sources, those residues have the same essential function in all those β-glycosidases. Further support for this conclusion comes from a comparison between the available structures (10) of family 1 β-glycosidases. In all of these enzymes the tyrosine, arginine and glutamate (nucleophile) residues occupy a similar spatial position (in order to facilitate visualization only four are shown in Fig. 6). The distances between these residues are always in the range compatible with a hydrogen bond (3 Å), except in the Zea maysβ-glycosidase, where the distance between the arginine and glutamate is 3.64 Å. But even in this case, a small movement in the flexible arginine side chain would that distance shorter without any steric hindrance.

This structural conservation suggests that the same noncovalent interactions are formed in all family 1 β-glycosidases. On this basis it is proposed that the noncovalent interactions network that modulates the Sfβgly50 pH optimum is probably operating in all family 1 β-glycosidases.


This project is supported by FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo) and CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico). E.H.P.A. and L.M.F.M. are undergraduate student fellows of CNPq. S.R.M., W.R.T. and C.F. are staff members of the Biochemistry Department (IQUSP) and research fellows of CNPq.