Structural and enzymatic characterization of Os3BGlu6, a rice β -glucosidase hydrolyzing hydrophobic glycosides and (1 → 3)- and (1 → 2)-linked disaccharides

β -glucosidases play roles in many processes in plants, such as chemical defense, alkaloid metabolism, hydrolysis of cell wall-derived oligosaccharides, phytohormone regulation, and lignification. However, the functions of most of the 34 GH1 gene products in rice ( Oryza sativa L.) are unknown. phylogenetic cluster of GH1, was produced in recombinant Escherichia coli. Os3BGlu6 hydrolyzed p -nitrophenyl ( p NP) β - D -fucoside ( k cat /K m = 67 mM -1 s -1 ), p NP-β - D -glucoside ( k cat /K m = 6.2 mM -1 s -1 ), and p NP- β - D -galactoside ( k cat /K m = 1.6 mM -1 s -1 ) efficiently, but had little activity toward other p NP-glycosides. It also had high activity toward n - octyl- β - D -glucoside and disaccharides, and was able to hydrolyze apigenin β -glucoside and several other natural glycosides. Crystal structures of Os3BGlu6 and its complexes with a covalent intermediate, 2-deoxy-2-fluoroglucoside, and a nonhydrolyzable substrate analogue, n -octyl- β - D -thioglucopyranoside, were solved at 1.83, 1.81 and 1.80 Å resolution, respectively. The position of the covalently trapped 2-F-glucosyl residue in the enzyme was similar to that in a 2-F-glucosyl intermediate complex of Os3BGlu7 (rice BGlu1). The side chain of a Met251 in the mouth of the active site appeared to block , etch (w/w) fusion tag and protease and m and m M activity pooled, to m M pool Protein concentrations protein


INTRODUCTION
β-Glucosidases (3.2.1.21) have a wide range of functions in plants, including acting in cell wall remodeling, lignification, chemical defense, plant-microbe interactions, phytohormone activation, activation of metabolic intermediates, and release of volatiles from their glycosides (Esen, 1993). They fulfill these roles by hydrolyzing the glycosidic bond at the nonreducing terminal glucosyl residue of a glycoside or an oligosaccharide, thereby releasing glucose and an aglycone or a shortened carbohydrate. The aglycone released from the glycoside may be a monolignol, a toxic compound or a compound that further reacts to release a toxic component, an active phytohormone, a reactive metabolic intermediate, or a volatile scent compound (Dharmawardhama et al., 1995;Morant et al., 2008;Lee et al., 2006;Brzobohatý et al., 1993;Barleben et al., 2007;Reuveni et al., 1999). Indeed, the wide range of glucosides of undocumented functions found in plants suggests many βglucosidase functions may remain to be discovered.
Aglycone specificity of GH1 β-glucosidases ranges from rather broad to absolutely specific for one substrate and is not obvious from sequence similarity. For 8 The activity of the expressed Os3BGlu6 was highest at pH 4 to 5, dropped by 50% at pH 2.5 and 6.0, and was negligible from pH 7.0 upward. The enzyme had highest activity over the temperature range of 40°C to 55°C, but was most stable at 20°C to 30°C and began to lose activity after 40 min at 40°C or higher. Therefore, standard assays were conducted at 30°C, which is physiologically relevant for rice.

Substrate specificity of Os3BGlu6
The ability of OsBGlu6 to hydrolyze p-nitrophenyl (pNP) glycosides was tested to assess its glycone specificity (Table I). Among the pNP glycosides, Os3BGlu6 preferred pNP-β-D-fucopyranoside (pNPFuc) with a k cat /K m of 67 mM -1 s -1 , due to its relatively low K m of 0.50 mM, followed by pNP-β-D-glucoside (pNPGlc) and pNP-β-D-galactoside (pNPGal) with k cat /K m values of 6.2 and 1.6 mM -1 s -1 , respectively. Hydrolysis of α -D-glucoside was negligible, but α-L-arabinoside was hydrolyzed at a low rate.

Overall structure of Os3BGlu6
To understand the structural basis for the substrate specificity of Os3BGlu6, its 3D structures alone and in complexes with 2-deoxy-2-fluoroglucoside (a slowly hydrolyzed covalent intermediate) and n-octyl-β-D-thioglucopyranoside (a nonhydrolyzable substrate analogue) were determined by protein crystallography. The 1 0 confirmed to bind by its competitive inhibition of Os3BGlu6 with a K i of 5.1± 0.2 mM. Six glycerol molecules from the cryoprotectant were also seen on the surface of the protein in the native structure. Four of these were also seen in the Os3BGlu6/noctyl-β-D-thioglucopyranoside complex and two in the Os3BGlu6/G2F complex. the native structure, but the crystal unit cell expanded with n-octyl-β-Dthioglucopyranoside, as previously seen in crystals with n-octyl-β-D-glucopyranoside (McPherson et al., 1986). An increase of 9.5 Å was seen in the c side of the unit cell, due to a displacement of the protein molecules related by symmetry along the z-axis by 4.5 to 5 Å in their water-mediated interactions, but the only significant change in the monomer structure was the conformation of loop C, which is not involved in these interactions.

1
In the intermediate complex, the density for the 2-deoxy-2-fluoroglucosyl residue in a relaxed 4 C 1 chair conformation covalently bound to the catalytic nucleophile was clearly evident in the glycone-binding subsite, surrounded by the universally conserved plant GH1 residues Gln31, His132, Asn177, Glu178, Glu394, Glu451 and Trp452 (Fig. 3 A and B). The residues Glu451, Gln31, Tyr321, and complex. In addition, the indole ring of Trp452 in the G2F complex was shifted slightly inward compared to the native structure to allow its nitrogen to hydrogen bond with O3 of the sugar ring at 2.84 Ǻ .
In the complex of Os3BGlu6 with the nonhydrolysable substrate analogue noctyl-β-D-thioglucopyranoside, the inhibitor was located deep in the active site slot ( Fig. 2 B, 3C and 4 A). The density for the sugar was well defined (Fig 3C). For the aglycone moiety, the density was strongest for C1', C4' and C8', and was lower for C2', C3', C5', C6' and C7', likely due to the flexibility of the carbon chain, which appeared to be able to bind in multiple positions. Although both the relaxed 4 C 1 chair and less energetically favorable 1 S 3 skew boat conformations fit the density well, the distance of Oε1 of the acid-base catalyst Glu178 to the sulfur in the 4 C 1 chair was 2.13 Ǻ , which was unacceptably close. In contrast, this distance was 3.07 Ǻ in the 1 S 3 conformation putting the sulfur within distance to accept a proton from the catalytic acid. The glycone residue was in a similar position to that of G2F in its complex and formed similar hydrogen bonding interactions, except that interactions of the O2 hydroxyl with His132, Asn177 and Glu394 were also evident ( Fig. 3 B and C). The aglycone was flanked by hydrophobic residues and a few polar residues. The n-octyl chain was bound by hydrophobic contacts from Tyr321 and Trp366 on one side, and Val250 and Met251, which also form the entrance to the active site, on the other side of the chain. The polar residues Thr322, Thr181, Asp249 His320, Asn319, and Gln273 also lay on either side of the aglycone.

Os3BGlu6 substrate glycone specificity
Os3BGlu6 showed a strong preference for β-D-fucoside, followed by β-Dglucoside, and then by β-D-galactoside, as was seen for several other plant β-Dglucosidases (Babcock and Esen, 1994;Srisomsap et al., 1996;Opassiri et al., 2004Opassiri et al., , 2006. Like maize β-glucosidase (Babcock and Esen, 1994) Os3BGlu6 shows a 10fold higher k cat /K m for pNPFuc than for pNPGlc. In maize and rice Os3BGlu7 βglucosidases, the preference for pNPFuc over pNPGlc is due to a higher k cat for pNPFuc, (Opassiri et al., 2004), whereas for Os3BGlu6, the difference is due to a 10fold lower K m for pNPFuc than pNPGlc, similar to Dalbergia cochinchinensis βglucosidase, which had K m values of 0.54 mM for pNPFuc and 5.4 mM for pNPGlc (Srisomsap et al., 1996). In Os3BGlu6, this appears to be partly due to a higher tolerance for the equatorial position of the 4' hydroxyl group in Os3BGlu6, since it had a K m for pNPGal similar to that for pNPGlc, whereas Os3BGlu7 had a K m over 10-fold higher for pNPGal than pNPGlc. In contrast, Os3BGlu6 had a much lower efficiency (k cat /K m of 0.06 mM -1 s -1 ) for hydrolysis of pNP-β-D-mannopyranoside than did Os3BGlu7 (k cat /K m of 1.01 mM -1 s -1 ), indicating the equatorial position of the 2' hydroxyl group is more critical to Os3BGlu6. Among natural substrates, Os3BGlu6 would be expected to hydrolyze β-D-glucosides, since β-D-fucosides are generally not found in plants.
In reported GH1 complex structures, the glutamate corresponding to Glu451 hydrogen bonds to both the C4 and C6 hydroxyls of the sugar residue in the -1 site (Burmeister et al., 1997;Sanz-Aparacio et al., 1998;Czjzek et al., 2000), as seen in the Os3BGlu6 complex with G2F. The flexibility of Glu451 in the native structure may help explain the relatively low K m for pNPFuc, since it may more readily bind to the axial 4-hydroxyl and the loss of entropy that occurs upon its hydrogen bonding to the 4-and 6-hydroxyls may decrease its energetic contribution to binding. In this case, the absence of hydrogen bonding to the 6-hydroxyl in pNPFuc would have less effect than in Os3BGlu7, in which the corresponding Glu440 residue maintains a similar position in the presence and absence of glucoside (Chuenchor et al., 2008).

Os3BGlu6 Aglycone Specificity
The hydrolysis of glucooligosaccharides and glucosides followed a rather unique pattern in Os3BGlu6, although the K m values of over 3 mM, suggest the substrates tested would only be hydrolyzed efficiently at high concentrations. The hydrolysis of β-(1→3)and β-(1→2)-linked disaccharides in preference to longer oligosaccharides reflects the relatively short binding cleft, as will be discussed below in comparison to Os3BGlu7. The hydrolysis of the n-octyl-β-D-glucoside with high efficiency suggests that Os3BGlu6 may act on hydrophobic glycosides in the plant, rather than or in addition to disaccharides. Among the natural glycosides tested, the flavonoid glycoside apigenin-7-O-β-D-glucoside was hydrolyzed most rapidly, followed by 7-O-β-linked isoflavonoids, which suggests that a flavonoid-7-Oglucoside could serve as a hydrophobic substrate in the plant. The apoplastic or vacuolar location of action suggested by its low pH optimum and secretory signal sequence is consistent with its action on either disaccharides or flavonoids or other hydrophobic glucosides, which may be found in these compartments.
The residues that directly interact with the aglycone in ZmGlu1 include Trp378 on one side and three phenylalanines, Phe198, Phe205 and Phe466, on the opposite side of the active-site slot (Czjzek et al., 2000). Although the tryptophan was conserved in Os3BGlu6 (Trp366), the opposing amino acid residues were replaced by Gln185, Gln192 and Ala453 in Os3BGlu6, and the aglycone of n-octyl-β-D- thioglucopyranoside is located on the opposite side of the active site, where it makes hydrophobic contacts with other residues. Among the residues noted to indirectly affect aglycone binding in ZmGlu1, Ala467 is conserved (Ala454 in Os3BGlu6), while Tyr473, which orients the plane of Trp378 for stacking interaction with the DIMBOA substrate, is replaced by Phe460, which has no hydroxyl to hydrogen bond to the indole ring of Trp366 in Os3BGlu6. Thus, Trp366 is twisted away by 1.13Ǻ for its interaction with the n-octyl chain of Os3BGlu6. These differences in aglycone binding may in turn affect the position and orientation of the glycone, as noted by Verdoucq et al. (2004).

Comparison of substrate binding between Os3BGlu6 and Os3BGlu7
Rice Os3BGlu6   appears to form hydrophobic interactions with the n-octyl chain of n-octyl-β-Dthioglucopyranoside, replaces Asn245 of Os3BGlu7, which appears to interact with cellotriose, based on the 40-fold increase in K m in the Os3BGlu7 N245V mutation (Cheunchor et al., 2008). Aside from the Met251, many differences were observed in the residues involved in the aglycone site, including Trp133, Thr181, Ile184, Gln185, Gln192, and Ala453 in Os3BGlu6 in place of Tyr131 Ile179, Leu182, Leu183, Asn190, and Leu442 in Os3BGlu7, respectively. These differences may help explain the fact that Os3BGlu6 prefers hydrophobic β-glucosides and short oligosaccharides, while Os3BGlu7 prefers longer β -(1→4)-linked oligosaccharides.

CONCLUSION
As plant GH1 enzymes exhibit a wide variety of specificities and release products with a wide range of physiological functions, the structural basis for this functional diversity is a significant problem. By extension, the small differences in the active sites of plant GH1 enzymes and their impact on the substrate specificity can help explain the expansion of this family to fulfill a range of functional roles in the plant.

Cloning and PCR amplification of cDNA encoding mature Os3BGlu6
To produce a cDNA library, 7-day-old rice cultivar Yukihikari seedlings were chilled at 5°C for a further 4 days, and then the RNA was extracted by the method of Bachem et al. (1996). Poly-A RNA was isolated from the total RNA with magnetic poly-T beads (Dynal). The poly-A RNA was reverse transcribed and used to produce a library with the λZAP-cDNA/Gigapack cloning kit (Stratagene

Os3BGlu6 pH and temperature optimum & stability studies
The optimum pH was determined by incubating 1 μg enzyme in a reaction volume of 140 μL containing 100 mM universal buffer (citric acid-disodium hydrogen phosphate), pH 2-11, at 0.5 pH unit increments with 1 mM pNPGlc for 10 min at 30ºC. The reaction was stopped with 100 μl of 2 M sodium carbonate. The amount of p-nitrophenol (pNP) released was determined from its absorbance at 405 nm. The enzyme's pH stability was studied by incubating 25 μg enzyme in 20 μL universal buffer at the same pH range for time periods of 10 min, 1, 2, 6, 12 and 24 h at 30ºC.
The enzyme was diluted 400 fold in 100 mM buffer at the optimum pH, and the activity was determined as described above. The temperature optimum was determined by incubating 1 μg enzyme in 100 mM sodium acetate, pH 5.0, over a temperature range of 5 to 100ºC in increments of 5 degrees for 10 min and then pNPGlc was added to the reaction to 1 mM final concentration and incubated for 10 min. Temperature stability was determined by preincubating 1 μg enzyme over a temperature range of 5 to 70ºC. At 10 min increments from 0 to 60 min, aliquots of the samples were placed on ice and the enzyme activity at 30ºC was determined as described above.

Substrate specificity and enzyme kinetics
Os3BGlu6 was tested for hydrolysis of glycosides of pNP and commercially available natural and synthetic substrates (listed in Tables I and II) by measuring release of pNP or glucose from the substrates and by thin layer chromatography (TLC). Activity was assayed in triplicate in 100 mM sodium acetate, pH 5.0, at 30ºC.
For TLC, the reactions were stopped after overnight digest by boiling 5 min, and products from the reactions were spotted on silica gel F 254 plates (Merck) and chromatographed with solvents of ethyl acetate:acetic acid:water (2:1:1, v/v) for the oligosaccharides and ethyl acetate:acetic acid:methanol:water (15:2:1:2, v/v) for natural glycosides. The products were visualized under UV light at 254 nm, and then by coating the plates with 10% (v/v) sulfuric acid in methanol and heating at 120°C until the spots were visible. For oligosaccharides and glucosides other than pNPGlc that were detectably hydrolyzed on TLC, glucose release was measured by the glucose-oxidase/peroxidase coupled assay, as previously described (Opassiri et al., 2003 Kinetic parameters were determined for substrates of interest, based on the initial relative activity assays. The initial velocities (V 0 ) were determined at time points and enzyme concentrations where the reaction rate was linear and the absorbance value was in the range of 0.1 to1.0, and used to calculate the kinetic constants. Substrate concentrations over a range of approximately 1/5 to 5 times the apparent K m were included. The k cat , K m , and k cat /K m were calculated by nonlinear regression of Michaelis-Menten plots with Grafit 5.0 (Erithacus Software).
Competitive K i values were determined by incubating 1 μg enzyme with eight different concentrations of the inhibitors (0 to 30 mM) in 100 mM sodium acetate buffer, pH 5.0, in the presence of 0 to 20 mM pNPGlc substrate under the reaction conditions described above. Lineweaver-Burk and Dixon plots were used to calculate the inhibition constants.

Data Collection, processing and structure refinement
The datasets for native Os3BGlu6, the Os3BGlu6/G2F complex and the Os3BGlu6/n- wavelength x-ray beam and an ADSC Quantum 315 CCD detector. The crystals were maintained at 110K with a cold stream of nitrogen (CryoJet, Oxford Instrument, Oxford, UK) throughout diffraction. All datasets were indexed, integrated and scaled and with the HKL-2000 package (Otwinowski & Minor, 1997). The crystals belonged to space group P2 1 2 1 2 1 with the unit cell parameters shown in Table III

N-terminal amino acid sequencing of Os3BGlu6
Os3BGlu6 crystal clusters were dissolved in sample buffer and separated on a 12% polyacrylamide gel by the method of Laemmli (1970 2006), respectively. One or two example proteins from each plant are given for each of the eight clusters of genes shared by Arabidopsis (At) and rice (Os), and the Arabidopsis-specific clusters At I (β-glucosidases) and At II (myrosinases), with the number of Arabidopsis or rice enzymes in each cluster given in parentheses. These sequences were aligned with all the At and Os sequences in Clustalx (Thompson et al., 1997), the alignment manually edited, all but representative sequences were removed, and the tree was calculated by neighbor joining method and bootstrapped with 1000 trials, then drawn with TreeView (Page, 1996). The grass plastid β-glucosidases, which are not represented in Arabidopsis and rice, are marked in the group marked "Plastid." Percent bootstrap reproducibility values are shown on internal branches where they are greater than 60%. Except those marked by asterisks, all external branches represent groups with 100% bootstrap reproducibility. *To avoid excess complexity, those groups of sequences marked with an asterisk are not monophyletic and represent more branches within the designated cluster than are shown. For a complete phylogenetic analysis of Arabidopsis and rice GH1 proteins, see Opassiri et al. (2006).  Os3BGlu6. The catalytic residues Glu394 and Glu178 and one molecule of Tris in two conformations are shown in ball and stick representation. The α-helices are colored purple, βstrands green and loops cyan. B, Surface view of the Os3BGlu6 structure showing the active site cleft with the ligand n-octyl-β-D-thioglucopyranoside. The surface is colored by electrostatic potential in the online version, with positively charged, negatively charged and neutral regions colored blue, red and white respectively. C, Close-up of the electron density of Tris in the active site in stereo view. The Fo-Fc omit electron density map for the Tris is shown as a mesh contoured at I/σ = 3. The sidechains of the surrounding amino acids are represented by sticks, and the Tris ligand is represented by balls and sticks. In all frames, oxygen atoms are shown in red, nitrogen in blue, sulfur in dark yellow, protein carbons in yellow, Tris carbons in green or pink and n-octyl-β-D-thioglucopyranoside carbons in green.  A, Stereoview of the superimposition of active site residues at the -1 subsite of the native Os3BGlu6 and Os3BGlu6/G2F complex structures. The residues surrounding the -1 subsite are represented by sticks colored with carbons in blue for native Os3BGlu6 and in yellow for Os3BGlu6/G2F. The G2F moiety bound to the catalytic nucleophile, Glu394, is represented by balls and sticks with carbon in pink. B, Active site of Os3BGlu6/G2F complex showing the electron density omit map of G2F and protein-ligand binding interactions. Hydrogen bonding interactions between G2F and amino acid residues are shown as black dashed lines, while the Fo-Fc omit map of G2F contoured at I/σ = 3.0 is shown as blue mesh. C, Glycone and aglycone interactions of n-octyl-β-D-thioglucopyranoside with Os3BGlu6 residues. Representation of the protein and ligand are as in B, with the Fo-Fc omit map for the ligand contoured at I/σ = 2.5. D, Two views of the superimposition of the G2F and n-octyl-β-D-thioglucopyranoside ligands, showing the glucose ring is in nearly the same position, but is shaped as a relaxed 4 C 1 chair in G2F (green carbons) and a 1 S 3 skew boat in n-octyl-β-D-thioglucopyranoside (pink carbons). In all frames, oxygen is in red, nitrogen in blue, fluorine in cyan, and sulfur in yellow.   In all frames, active site amino acid residues are shown as sticks behind the grey transparent surface. The ligands are shown in ball and stick representations. In the online version, protein carbons are shown in green, n-octyl-β-Dthioglucopyranoside carbons in pink and DIMBOAGlc carbons in blue, and cellotriose carbons in yellow, oxygen in red, nitrogen in dark blue, and sulfur in yellow.