Design and directed evolution of noncanonical β-stereoselective metalloglycosidases

The study addresses whether glycosidases can be engineered to use a metal ion directly for hydrolysis—an ability common in many hydrolases but notably absent in natural glycosidases, which typically rely on two acidic residues for nucleophilic attack and general acid/base catalysis. The authors hypothesize that a coordinatively unsaturated Zn site could be installed into a suitable protein scaffold to activate water and promote glycosidic bond cleavage, overcoming the electrostatic challenge of stabilizing an oxocarbenium-like transition state. They choose the outer membrane protein OmpF, a β-barrel with a constriction zone and host-like cavity, as a scaffold amenable to both catalysis and whole-cell display/evolution. The purpose is to create noncanonical, Zn-dependent glycosidases and to explore their mechanism, selectivity, and applicability as whole-cell biocatalysts.

• Hydrolases are widespread and often include metallohydrolases wherein metals generate nucleophiles for bond cleavage. Both metal-dependent and independent forms exist in many subclasses (esterases, peptidases, β-lactamases), but glycosidases are a notable exception where metals, if present, usually aid binding/transition-state stabilization rather than water activation.
• Prior reports describe inorganic complexes, metallopolymers, and metallopeptides with glycosidase-like activity, highlighting the potential for metal-mediated glycoside hydrolysis outside proteins.
• Canonical β-glycosidases (Koshland mechanism) use a pair of acidic residues separated by characteristic distances to effect inverting or retaining mechanisms.
• OmpF’s structure and role as an antibiotic channel suggest intrinsic binding capabilities for small molecules, making it a plausible catalytic host scaffold.
• Related metalloenzymes such as Cu-dependent lytic polysaccharide monooxygenases perform oxidative glycoside cleavage, and a recently characterized Zn-dependent glycosidase featuring a cysteine nucleophile broadens known metallocofactor roles in glycosidases.

• Scaffold selection and design: The constriction zone of the β-barrel OmpF was analyzed to place a mononuclear, coordinatively unsaturated Zn site. Based on geometric constraints from natural Zn sites (Cα–Cα distances 3.8–10.6 Å), triad designs with histidines were introduced by mutagenesis.
• Constructed variants: Two triple mutants were created as parents: OmpF1 (R82H/Y102H/R132H) and OmpF2 (L83H/Y102H/R132H). A derivative reversing Y102H (OmpF1Y) was also made. D113E single mutations were introduced (denoted /E) to probe the role of a nearby acidic residue.
• Protein production and structural validation: Variants were expressed in E. coli, extracted/refolded or expressed directly into the outer membrane, purified, and crystallized with Zn. X-ray structures confirmed Zn sites: OmpF1 formed a 2His/1Glu (E71’ from neighboring protomer) site; OmpF2 formed a 3His site. In both, a fourth coordination position was occupied by a water molecule, leaving the site catalytically poised; the open coordination faced different sides (extracellular vs periplasmic) in OmpF1 vs OmpF2.
• Activity assays (in vitro): Esterase (PNPA) and β-lactamase (nitrocefin) activities were measured for apo vs Zn-bound states to establish Zn-dependent hydrolysis. Glycosidase activity was assayed using 4-methylumbelliferyl-β-D-glucopyranoside (4-β-MUG) by fluorescence and HPLC.
• Mechanism-guided redesign: Considering Koshland distances, nearby acidic residues were evaluated; D113E variants were tested to assess coupling to the Zn site.
• Directed evolution: Libraries targeting residues near the Zn site were constructed (saturation mutagenesis at R42, Y106, G120, A123, and R82 depending on template). Whole-cell screening on M9/agar with Zn used dual selection: growth/color on cellobiose and X-Glu, followed by screening with 4-β-MUG. Iterative rounds yielded improved mutants: OmpF1/E-R4 (R42S/Y106A/G120C/A123N), OmpF1Y/E-R4 (Y106R/R42E/G120S/A123V), and OmpF2/E-R2 (Y106H/R82C).
• Kinetics and characterization: Purified evolved variants were analyzed for 4-β-MUG hydrolysis (0–20 mM) to obtain second-order rate constants; turnover numbers (TONs) were measured from cell lysates. Stereoselectivity was probed using 4-α-MUG. ICP-MS confirmed Zn incorporation in whole-cell OmpF2/E-R2. pH profiles and apparent pKa values were determined. Substrate scope included n-octyl-β-D-glucopyranoside (OG).
• Mechanistic probes: Covalent inhibitor CBE labeling with LC-MS/MS identified adducts at Zn(OH)-His sites and nearby acidic residues (E62, D113E, E117). Azide trapping experiments with a closely related variant (OmpF2/E-R2*, Y106H/R82Y) identified β-1-azido-1-deoxy-glucose by NMR/LC-MS, supporting a retaining mechanism. Docking simulations positioned 4-β-MUG near the Zn site and D113E in catalytically relevant orientations.

• Successful creation of Zn-dependent glycosidases: Installing a coordinatively unsaturated Zn site in OmpF yielded variants (OmpF1, OmpF1Y, OmpF2) with Zn-dependent hydrolysis of PNPA, nitrocefin, and 4-β-MUG; wild-type lacked Zn-dependent activity.
• Structural confirmation: X-ray structures showed OmpF1 forms a 2His/1Glu Zn site (with E71’ from an adjacent protomer) and OmpF2 forms a 3His Zn site; each features a water ligand and an open site oriented differently (extracellular vs periplasmic).
• Role of adjacent acidic residue: D113E modulated glycosidase activity; coupling between the Zn site and D113/E is implicated. Some apo activity emerged in OmpF1/E, but Zn-bound forms were generally more active.
• Directed evolution boosts activity: Evolved mutants OmpF1/E-R4, OmpF1Y/E-R4, and OmpF2/E-R2 showed up to ~100-fold higher TONs and up to two orders of magnitude higher whole-cell initial rates versus parents.
• Kinetic performance: Evolved enzymes achieved second-order rate constants k2 = 5.6–10.4 min⁻¹ M⁻¹ with 4-β-MUG, corresponding to up to 2.8 × 10^6-fold rate enhancement over the uncatalyzed reaction. The abstract reports catalytic proficiency up to 2.8 × 10^3.
• High β-stereoselectivity: Negligible activity on 4-α-MUG; selection on β-linked substrates likely shaped chiral pockets favoring β-glycosides.
• Broadened substrate handling: OmpF2/E-R2 hydrolyzed n-octyl-β-D-glucopyranoside, generating glucose and 1-octanol, showing accommodation of bulky, hydrophobic substrates.
• pH dependence: Bell-shaped profiles with apparent pKa ≈ 7.7 and 8.3; Zn-bound forms optimal at pH 8.0–8.5 (apo at 7.5–8.0), differing from many natural glycosidases.
• Mechanistic evidence: CBE covalent labeling at Zn(OH)-His and acidic residues (E62, D113E, E117) supports a dual-motif mechanism. Azide trapping with a single-site variant produced β-1-azido-1-deoxy-glucose, evidencing a retaining mechanism proceeding via glycosylation and deglycosylation.
• Zn incorporation in cells: ICP-MS confirmed Zn bound to OmpF2/E-R2 in the outer membrane, supporting Zn-dependent whole-cell catalysis.

The work demonstrates that a noncatalytic membrane protein scaffold (OmpF) can be transformed into a Zn-dependent glycosidase by installing a coordinatively unsaturated Zn site and optimizing nearby residues through mechanism-guided redesign and directed evolution. The findings address the central question by showing that metals can directly mediate glycosidic bond hydrolysis in a protein environment when coupled with appropriately positioned acidic residues. High β-stereoselectivity and improved activities after evolution, together with whole-cell function, underscore the scaffold’s suitability and the power of directed evolution. Mechanistic probes (CBE labeling and azide trapping) support a noncanonical retaining mechanism where one member of the classical acid pair is replaced by a Zn–OH2/OH motif that alternately serves as Lewis acid/base and potential nucleophile during glycosylation/deglycosylation. The atypical pH optima likely arise from the Zn-centered chemistry and the unique electrostatics of the OmpF constriction zone. These results broaden the recognized roles of metals in glycoside hydrolysis, suggest possible undiscovered metalloglycosidases in nature, and provide a basis for engineering outer-membrane whole-cell biocatalysts targeting glycosides.

By retrosynthetic active-site construction, structure-guided redesign, and iterative saturation mutagenesis with selection, the authors created β-stereoselective, Zn-dependent glycosidases from OmpF. Crystal structures validated the intended Zn coordination motifs; biochemical assays and mechanistic probes established Zn–acidic residue cooperation and a retaining mechanism in optimized variants. Activities and TONs improved markedly through directed evolution, and whole-cell catalysis was demonstrated with evidence of Zn incorporation in vivo. This expands the chemical repertoire of artificial metalloenzymes and highlights membrane proteins as viable scaffolds for catalysis. Future work could: (i) expand substrate scope to complex polysaccharides and diverse glycosidic linkages, (ii) obtain substrate/inhibitor-bound structures for direct mechanistic visualization, (iii) further evolve kinetics and stability under application-relevant conditions, and (iv) explore alternative metals or ligation patterns for tailored selectivity and mechanisms.

• Substrate scope assessed primarily with model substrates (4-β-MUG, cellobiose, X-Glu) plus a single detergent-like β-glycoside (OG); broader applicability to varied glycosidic linkages/substrates remains to be shown.
• Detailed catalytic parameters (kcat, KM) for glycosidase activity are limited; comparisons to natural glycosidases are not comprehensively provided.
• Mechanistic conclusions are based on inhibitor labeling, azide trapping, and docking; no substrate- or transition state analog-bound crystal structures are reported.
• Some evolved variants show apo activity, complicating exclusive attribution to Zn-dependent pathways and indicating parallel catalytic routes.
• pH optima and membrane environment effects may limit direct translation to certain applications; tuning for physiological conditions may be necessary.