Structural and Functional Characterization of a Ruminal β-Glycosidase Defines a Novel Subfamily of Glycoside Hydrolase Family 3 with Permuted Domain Topology

Abstract

Metagenomics has opened up a vast pool of genes for putative, yet uncharacterized, enzymes. It widens our knowledge on the enzyme diversity world and discloses new families for which a clear classification is still needed, as is exemplified by glycoside hydrolase family-3 (GH3) proteins. Herein, we describe a GH3 enzyme (GlyA₁) from resident microbial communities in strained ruminal fluid. The enzyme is a β-glucosidase/β-xylosidase that also shows β-galactosidase, β-fucosidase, α-arabinofuranosidase, and α-arabinopyranosidase activities. Short cello- and xylo-oligosaccharides, sophorose and gentibiose, are among the preferred substrates, with the large polysaccharide lichenan also being hydrolyzed by GlyA₁ The determination of the crystal structure of the enzyme in combination with deletion and site-directed mutagenesis allowed identification of its unusual domain composition and the active site architecture. Complexes of GlyA₁ with glucose, galactose, and xylose allowed picturing the catalytic pocket and illustrated the molecular basis of the substrate specificity. A hydrophobic platform defined by residues Trp-711 and Trp-106, located in a highly mobile loop, appears able to allocate differently β-linked bioses. GlyA₁ includes an additional C-terminal domain previously unobserved in GH3 members, but crystallization of the full-length enzyme was unsuccessful. Therefore, small angle x-ray experiments have been performed to investigate the molecular flexibility and overall putative shape. This study provided evidence that GlyA₁ defines a new subfamily of GH3 proteins with a novel permuted domain topology. Phylogenetic analysis indicates that this topology is associated with microbes inhabiting the digestive tracts of ruminants and other animals, feeding on chemically diverse plant polymeric materials.

Keywords: GH3 family; X-ray crystallography; enzyme structure; enzymology; glycoside hydrolase; metagenomics; permuted domains topology; phylogenetic analysis; protein structure; β-glycosidase.

FIGURE 1.

Temperature (A) and pH (B) profiles of the purified β-glucosidase GlyA₁. The data represent the relative percentages of specific activity (units/g) compared with the maximum activity using pNPβGlc as substrate (100% in A, 2841 units/g; 100% in B, 3056 units/g). The specific activities were calculated using 0.23 μm protein and 10 mg/ml pNPβGlc as the assay substrate. A, reactions were performed in 50 mm sodium acetate buffer, pH 5.6, at different temperatures. B, reactions were performed at different pH (50 mm BR buffer) and at 40 °C. Standard deviations of the results of assays conducted in triplicate are shown.

FIGURE 2.

Permuted domain composition of GlyA₁. A, comparison of GlyA₁ structure with representative members of multidomain GH3 enzymes. β-glucosidases from K. marxianus, KmβGlu (9) and T. neapolitana, TnβGlu (8), the exo-1,3/1,4-β-glucanase from Pseudoalteromonas sp., PsExoP (10) and the barley β-d-glucan exohydrolase, HvExoI (7) are shown. Domains are named as ABS: (α/β)₆-sandwich; FLD fibronectin-like; ABB (α/β)₈ barrel; PA14, protective antigen PA14 domain. B, folding of GlyA₁. The N-terminal (α/β)₆-sandwich domain (red) is followed by the FnIII domain (beige) and the (α/β)₈ barrel domain (green). Two long segments connect the three domains (gray). A glucose found in the active site is represented by spheres. C, scheme of the GlyA₁ domain organization (left) as compared with that of T. neapolitana β-glucosidase (right) (8). D, superimposition of GlyA₁ (gold) onto T. neapolitana β-glucosidase (blue) coordinates. Both enzymes present a deviation from the canonical (α/β)₈ barrel topology, with their first α-helix missing, which makes strand β2 reversed and antiparallel with the other seven strands. The main difference between both enzymes is the long arm linking the FnIII to the (α/β)₈ domain in GlyA₁, which is missing in T. neapolitana β-glucosidase. Also, small differences in the orientation of some helixes are observed.

FIGURE 3.

GlyA₁ active site architecture. A, detail of the loops surrounding its active site from the (α/β)₈ barrel (green) and the (α/β)₆-sandwich (raspberry) domains, superimposed onto the T. neapolitana β-glucosidase (8) (pale blue). Three glycerol molecules from the cryobuffer found in the GlyA₁ crystals are shown in orange. Asp-709 and Glu-143 are the nucleophile and the acid/base catalyst, respectively. Main features of GlyA₁ are the extended loop containing Asp-709, which includes Trp-711 and the ion pair Arg-717–Glu-447 fixing it to the unique long arm and a highly flexible loop containing Trp-106. Two different conformations found among the crystals at Trp-111 and Phe-147 are highlighted. B, detail of the atomic interactions defining subsite −1. A glucose molecule is shown in gold. Xylose binds in the same relaxed chair conformation, and only interaction of the glucose O6 hydroxyl is missing. Inset, binding mode of galactose in a semi-chair conformation by flattening of the C4 atom that has the axial hydroxyl substituent and keeping the same interaction pattern. C, thiocellobiose (cyan) and thiogentibiose (pink) modeled at the active site by structural superimposition to the previously determined β-d-glucan glucohydrolase barley complexes (PDB entries 1IEX and 3WLP (34)), delineating putative subsite +1. D, molecular surface of the GlyA₁ active site, with relevant residues as sticks. Three different β-1,4/β-1,3-linked tetraglucosides have been manually docked by superposition of their non-reduced end to the experimental glucose: a cellotetraose, as found in PDB entry 2Z1S (green); a Glc-4Glc-3Glc-4Glc (purple), and a Glc-4Glc-4Glc-3Glc (yellow), as built by the on-line carbohydrate-building program GLYCAM (45) and exported in its minimum energy state. E, superposition of GlyA₁-Glc structure (beige) with those reported for T. reesei β-glucosidase (purple) (12) and barley β-d-glucan glucohydrolase complexed with thiocellobiose (cyan) (34).

FIGURE 4.

SAXS analysis of GlyA₁. Six ab initio models were generated for complete GlyA₁ from SAXS data, using the experimental structure of the truncated protein and two different models of the last 120 residues (GlyA₁-Ct). The two templates were obtained from Swiss-Model (red) (48) or CPHmodel (blue) (49) servers, which predict different lengths of the linker attaching this domain to the core protein, 32 or 5 residues, respectively. CORAL (47) modeling of this linker in each run is represented in spheres. The active site pocket is indicated by the galactose found at the crystal (yellow), and the mobile loop (residues 100–113), as observed in the galactose-soaked crystals, is highlighted in green.

FIGURE 5.

GlyA₁ phylogenetic analysis. The unrooted circular Neighbor-Joining tree indicating phylogenetic positions of polypeptide sequences of the GlyA₁ enzyme characterized in present work (red boldface) and reference similar enzymes. GenBank^TM or PDB (in boldface) accession numbers are indicated. The domain architecture (ABB, ABB_ABS, ABB_ABS_FLD, ABB-ABS(PA14)-FLD, and ABS_FLD_ABB) to which each sequence is associated is specifically indicated. Multiple protein alignment was performed using ClustalW program, built into software version 2.1. Phylogenetic analysis was conducted with the Ape package implemented for R programming language. Sequences resembling NagZ (β-N-acetyl-glucosaminidase) are highlighted with pink background. Those encoding GH3 β-glucosidases are indicated in brown; within them, those with GlyA₁-like permuted domain topology are indicated in gray. ABB, (α/β)₈ barrel; ABS (α/β)₆-sandwich; FLD, fibronectin-like type III domain; PA14, protective antigen PA14 domain.