Structure- and computational-aided engineering of an oxidase to produce isoeugenol from a lignin-derived compound

Lignin is an abundant, renewable source of phenolic compounds offering a sustainable alternative to petroleum-derived aromatics, yet its complexity hampers valorization. Reductive catalytic fractionation (RCF) efficiently produces para-substituted phenols, with 4-n-propylguaiacol often a major product. This phenol closely resembles isoeugenol (lacking only an alkene), which is an important flavor/fragrance and precursor for vanillin, fine chemicals, polymers, and epoxy resins. A one-step selective dehydrogenation of 4-n-propylguaiacol to isoeugenol would enable lignin-based production of this valuable molecule. VAO-type flavoproteins act on 4-alkylphenols, but fungal VAO prefers hydroxylation over dehydrogenation and is less suitable for expression and computation. The bacterial eugenol oxidase (EUGO) from Rhodococcus jostii RHA1 is well expressed, dimeric, solvent-tolerant, and structurally characterized, but shows poor activity on 4-alkylphenols (kcat 0.0026 s−1 for 4-ethylphenol; 0.008 s−1 for 4-n-propylguaiacol). The study aims to engineer EUGO to selectively dehydrogenate 4-n-propylguaiacol to isoeugenol with high activity and chemoselectivity while maintaining or improving thermostability.

Background work on VAO-type enzymes has established their mechanisms and product outcomes with 4-alkylphenols, where hydride transfer to FAD forms a p-quinone methide that can either tautomerize to an alkene (dehydrogenation) or be hydroxylated to an alcohol; water accessibility and substrate orientation control selectivity. VAO (fungal) has been extensively studied but is less suitable for E. coli expression and tends toward hydroxylation; VAO is octameric, complicating computation. EUGO (bacterial) is a homolog with covalently bound FAD, good expression, solvent tolerance (up to 10% DMSO), and available crystal structures, but low activity on 4-alkylphenols. Prior studies also reported covalent substrate–flavin adducts in VAO with small alkylphenols that limit turnover. These insights guide redesign of EUGO for improved stability, activity, and dehydrogenation selectivity on 4-n-propylguaiacol.

• Thermostabilization by computation (FRESCO): Starting from EUGO structure (PDB 5FXD; isoeugenol-bound), 9196 allowed single mutants (excluding residues within 5 Å of FAD) were evaluated by FoldX and Rosetta ΔΔG; 496 passing mutants (ΔΔG < −5 kJ/mol) underwent short MD simulations and visual inspection, yielding 72 predicted stabilizing mutations for experimental testing. Expression in 96-well format, purification by Ni-affinity, and Thermofluor assays determined Tm shifts. Activity was checked with vanillyl alcohol to deselect destabilizing/compromising mutations. Five activity-retaining stabilizing mutations were combined iteratively to create EUGO5X, improving Tm by 13.5 °C.
• Active-site redesign for chemoselectivity: 4-n-propylguaiacol (4PG) was docked into EUGO (PDB 5FXD) using AutoDock Vina. Rosetta Coupled Moves sampled mutations at positions within 5 Å of 4PG (non-catalytic residues; allowed substitutions from MSA population >2%). Sixteen single-site variants were selected and introduced into WT EUGO and EUGO5X. Cell-free extracts were first screened by HPLC for conversion and product distribution (isoeugenol, alcohol, ketone). Best hits were purified and reassessed. S394 variants notably improved dehydrogenation selectivity; S394V in the thermostable background (S394V-EUGO5X) gave IEUG as main product (>95%).
• Structural analysis and mechanistic diagnosis: Crystallized S394V-EUGO5X with 4PG (2.9 Å). Observed both non-covalent and covalent 4PG–FAD N5 adducts in different subunits; the covalent adduct correlated with reduced flavin absorbance during turnover and low kcat, indicating a slowly decaying inhibitory intermediate. V394 sits at the cavity entrance, likely restricting water access and reducing hydroxylation.
• Structure-guided mitigation of covalent adduct: Targeted residues near FAD N5 (D151 and Q425). Generated a panel of Q425 mutants in S394V-EUGO5X; several increased conversion but sometimes reduced chemoselectivity. Introduced D151E (inspired by 4-ethylphenol oxidase) into S394V-EUGO5X, which preserved conversion and improved chemoselectivity. Combined D151E and Q425S with S394V on the EUGO5X background to yield D151E/S394V/Q425S-EUGO5X (named PROGO).
• Kinetics and stability measurements: Determined steady-state kinetics by UV–vis (vanillyl alcohol at 340 nm; 4PG at 300 nm). Tm measured by Thermofluor. Compared WT, S394V-EUGO5X, D151E/S394V-EUGO5X, and PROGO.
• Preparative-scale bioconversions: Isolated enzyme reaction: 0.5 g (3.0 mmol) 4PG with 18.5 mg (0.30 μmol) PROGO in 60 mL, 10% DMSO, 50 mM KPi pH 7.5 at 25 °C for 48 h; product extraction and silica gel purification. Whole-cell reaction: 1.27 g (7.6 mmol) 4PG with E. coli expressing PROGO (OD600=29) in 125 mL buffer with 10% DMSO at 25 °C for 48 h; extraction and purification. HPLC monitored conversion.

• Thermostability: Combining five stabilizing mutations yielded EUGO5X with +13.5 °C Tm increase relative to WT (WT Tm 66.5 °C; S394V-EUGO5X Tm 78.5 °C; PROGO Tm 81.6 °C).
• Chemoselectivity: S394 mutations, especially S394V in EUGO5X, strongly favored dehydrogenation, producing >95% isoeugenol with high conversion (≈80% in 24 h at 5 mM 4PG, 10 μM enzyme).
• Mechanism of low activity: S394V-EUGO5X forms a covalent FAD N5–substrate adduct observed crystallographically and spectroscopically (loss of oxidized flavin absorbance during turnover), explaining low kcat.
• Activity improvements (steady-state kinetics on 4-n-propylguaiacol): WT EUGO kcat 0.008 s−1, KM 3.3 μM; S394V-EUGO5X kcat 0.028 s−1, KM 3.4 μM; D151E/S394V-EUGO5X kcat 0.080 s−1, KM 3.0 μM; PROGO (D151E/S394V/Q425S-EUGO5X) kcat 0.43 s−1, KM 7.4 μM. Thus, PROGO improves kcat by ~54-fold vs WT and ~15-fold vs S394V-EUGO5X.
• Selectivity retained in optimized variant: D151E increased chemoselectivity; Q425S relieved steric constraints to reduce covalent adduct formation; the combined PROGO variant gave 97% chemoselectivity to isoeugenol.
• Preparative-scale outcomes: Isolated enzyme (0.5 g scale): nearly complete conversion in 48 h; 328 mg isoeugenol isolated (66% yield). Whole cells (1.27 g scale): nearly full conversion in 48 h; 524 mg isolated (42% yield), lower due to product retention by cell debris.
• Structural insights: V394 narrows cavity near the subunit interface, likely limiting water access; D151E side chain points toward substrate Cα/Cβ region, potentially shielding the quinone methide from water; Q425S loosens packing above the side chain, allowing substrate orientation away from FAD N5 and reducing covalent adduct formation.

The study addressed the challenge of converting a lignin-derived 4-n-propylguaiacol selectively to isoeugenol by engineering EUGO for stability, activity, and chemoselectivity. Computational stabilization (FRESCO) provided a robust scaffold (EUGO5X) enabling accumulation of active-site mutations without loss of fold/function. A focused, computation-guided active-site library identified S394 as a key determinant of chemoselectivity; S394V restricts water access to the quinone methide intermediate, suppressing hydroxylation. However, low turnover was traced to a covalent N5 adduct, verified by crystallography and flavin spectroscopy. Structure-guided mutations proximal to FAD (D151E, Q425S) minimized adduct formation and maintained selectivity, greatly increasing kcat. The final variant, PROGO, efficiently and selectively oxidizes 4-n-propylguaiacol to isoeugenol and functions at gram scale using both purified enzyme and whole cells. These findings validate a stepwise, mechanism-informed, and computation-aided approach to reprogram VAO-type oxidases for non-natural substrates relevant to lignin valorization.

The work delivers PROGO, an engineered EUGO variant with eight mutations that combine high thermostability (Tm 81.6 °C), markedly improved catalytic activity on 4-n-propylguaiacol (kcat 0.43 s−1), and high chemoselectivity (97%) for isoeugenol. Structural studies uncovered a rate-limiting covalent FAD–substrate adduct in an intermediate variant and guided residue changes (D151E, Q425S) to suppress it while maintaining the dehydrogenation pathway favored by S394V. Preparative-scale reactions confirmed practical applicability. Only 104 single-site mutations were tested to arrive at the optimized variant, underscoring the efficiency of the computationally guided, small-library approach. Future work could further enhance turnover (activity on vanillyl alcohol remains higher), explore broader substrate scope among lignin-derived phenols, optimize whole-cell product recovery, and adapt the strategy to other VAO-type enzymes to expand lignin valorization routes.

• Although PROGO shows a ~54-fold kcat improvement over WT on 4-n-propylguaiacol, activity on vanillyl alcohol remains higher (kcat 2.8 s−1), indicating potential for further optimization.
• Whole-cell conversions gave lower isolated yields (42%) due to product retention by biomass, suggesting downstream processing and extraction require optimization.
• Reactions required 10% DMSO to solubilize substrate, which may limit certain process configurations.
• Covalent adduct formation was mitigated but may still occur under some conditions; long-term stability under continuous operation was not reported.