β-Glucosidases and β-mannosidases hydrolyze substrates that differ only in the epimer of the nonreducing terminal sugar moiety, but most such enzymes show a strong preference for one activity or the other. Rice Os3BGlu7 and Os7BGlu26 β-glycosidases show a less strong preference, but Os3BGlu7 and Os7BGlu26 prefer glucosides and mannosides, respectively. Previous studies of crystal structures with glucoimidazole (GIm) and mannoimidazole (MIm) complexes and metadynamic simulations suggested that Os7BGlu26 hydrolyzes mannosides via the B2,5 transition state (TS) conformation preferred for mannosides and glucosides via their preferred 4H3/4E TS conformation. However, MIm is weakly bound by both enzymes. In the present study, we found that MIm was not bound in the active site of crystallized Os3BGlu7, but GIm was tightly bound in the −1 subsite in a 4H3/4E conformation via hydrogen bonds with the surrounding residues. One-microsecond molecular dynamics simulations showed that GIm was stably bound in the Os3BGlu7 active site and the glycone-binding site with little distortion. In contrast, MIm initialized in the B2,5 conformation rapidly relaxed to a E3/4H3 conformation and moved out into a position in the entrance of the active site, where it bound more stably despite making fewer interactions. The lack of MIm binding in the glycone site in protein crystals and simulations implies that the energy required to distort MIm to the B2,5 conformation for optimal active site residue interactions is sufficient to offset the energy of those interactions in Os3BGlu7. This balance between distortion and binding energy may also provide a rationale for glucosidase versus mannosidase specificity in plant β-glycosidases.
Glycoside hydrolases (GH) bind tightly to the sugar moiety at the glycosidic bond being hydrolyzed to stabilize its transition state conformation. We endeavored to assess the importance of glucose-binding residues in GH family 116 (GH116) β-glucosidases, which include human β-glucosylceramidase 2 (GBA2), by mutagenesis followed by kinetic characterization, X-ray crystallography, and ONIOM calculations on Thermoanaerobacterium xylanolyticum TxGH116, the structural model for GH116 enzymes. Mutations of residues that bind at the glucose C3OH and C4OH caused 27–196-fold increases in KM for p-nitrophenyl-β-D-glucoside, and significant decreases in the kcat, up to 5000-fold. At the C6OH binding residues, mutations of E777 decreased the kcat/KM by over 60,000-fold, while R786 mutants increased both the KM (40-fold) and kcat (2–4-fold). The crystal structures of R786A and R786K suggested a larger entrance to the active site could facilitate their faster rates. ONIOM binding energy calculations identified D452, H507, E777, and R786, along with the catalytic residues E441 and D593, as strong electrostatic contributors to glucose binding with predicted interaction energies > 15 kcal mol−1, consistent with the effects of the D452, H507, E777 and R786 mutations on enzyme kinetics. The relative importance of GH116 active site residues in substrate binding and catalysis identified in this work improves the prospects for the design of inhibitors for GBA2 and the engineering of GH116 enzymes for hydrolytic and synthetic applications.