Fig. 1. Ancestral sequence reconstruction and heterologous expression of UPOs. (a) The subtree containing CglUPO (shown in red). All other tips are labelled using their strains. Ancestral nodes selected for gene synthesis are labelled in purple. (b) Expression of ancestral UPO and CglUPO in Pichia pastoris supernatant. (c) Reaction conditions: purified UPO (2.5 μM), substrate 1 mM estra-4,9-diene-3,17-dione (1) (1% v/v DMSO), KPi buffer (100 mM, pH = 7.0), 50 mM glucose, 50 U mL-1 glucose oxidase, Vfinal = 1 mL, 30 °C, 250 rpm for 4 h. All experiments were performed in triplicates. The conversion of 1 to 11β-hydroxylated product 1a was determined by HPLC based on peak area or calibration with commercial product standards.

Fig. 2. Rational mutations guided by molecular dynamics simulations. (a) MD simulations revealed that the bulky side chain of L232 (yellow) introduces steric hindrance near the substrate estra-4,9-diene-3,17-dione (1) (blue), pushing it away from the catalytic center. This spatial constraint causes the substrate to shift toward F83 (yellow), where F83 stabilizes a non-productive binding pose through hydrophobic anchoring. As a result, the substrate is mispositioned and unable to effectively access the reactive iron-oxo species (FeIV=O) of Compound I, thereby impairing catalytic efficiency. The heme group is shown in purple, with the iron atom of Compound I represented as a red sphere. (b, c) Turnover numbers (TON) of enzyme variants using concentrated supernatant were measured under standardized conditions: 1 μM concentrated supernatant, substrate 1 (1% v/v DMSO), KPi buffer (100 mM, pH = 7.0), 50 mM glucose, 50 U mL-1 glucose oxidase, Vfinal = 1 mL, 30 °C, 250 rpm for 4 h. (d) TON of purified enzyme (N1-WT, N1-F83G, N1-L232V and N1-F83GL232V) in the hydroxylation of substrate 1. Reaction conditions: 1 μM concentrated supernatant, substrate 1 (1% v/v DMSO), KPi buffer (100 mM, pH = 8.0), 100 mM ascorbic acid (AscA) [49], Vfinal = 1 mL, 30 °C, 250 rpm for 24 h. (e) Steady-state kinetic analysis of N1 and its variants with substrate 1. Reaction conditions: 0.4 μM (WT and F83G) or 0.1 μM (L232V and F83G/L232V), substrate 1 (0-3 mM), glucose 5 mM, glucose oxidase 5 U mL-1, 100 mM potassium phosphate pH = 7.0, Vfinal = 500 μL. Reactions were carried out in a metal bath at 30?°C, 800 rpm for 10 min (L232V and F83GL232V) or 30 min (WT and F83G). All experiments were performed in triplicates. Fold-change in catalytic efficiency was calculated as the ratio of kcat/Km for each variant relative to the WT.

Fig. 3. MD simulations reveal the molecular basis for enhanced 11β-hydroxylation activity in N1 variants. (a) Two-dimensional density plot of the NAC distance (substrate C11β-Compound I O1) and angle (H11β-C11β-O1) for the WT enzyme, showing a predominant unstable (U) and a minor stable (S) substrate conformation. (b) Representative snapshot of the WT active site illustrating L232 pushing the substrate away and F83 stabilizing a non-productive conformation, collectively impairing catalytic efficiency. (c) NAC density map for the F83G variant showing a reduced U state population. (d) Representative snapshot of F83G showing reduced hydrophobic anchoring due to F83 removal, while steric hindrance from L232 persists. (e) NAC density map for the L232V variant showing a significantly decreased U state population. (f) Representative snapshot of L232V with improved substrate binding due to the shorter side chain at position 232. (g) NAC density map for the F83G/L232V double mutant showing exclusive adoption of the productive S state and complete loss of the U state. (h) Representative snapshot of F83G/L232V demonstrating synergistic effects of the shortened L232 side chain and removal of the F83 hydrophobic anchor, collectively enabling optimal productive substrate binding. Color scheme in all structural panels (b, d, f, h): Substrate cyan carbons; Compound I (Cpd I) porphyrin slate blue; residues V232 and F83 yellow; Fe=O red sphere.

Fig. 4. Regio- and stereoselective hydroxylation of steroids with N1 F83G/L232V. (a) Steroid substrates: testosterone (2), nandrolone (3), 1,4-diene-3,17-dione (4) and pregna-1,4,9(11),16-tetraene-3,20-dione (5). Reaction was in 400 mL of phosphate buffer (pH 8.0, 100 mM) containing 100 mM AscA, 4 μM N1 F83G/L232V, 50 mg substrate dissolved in 20 mL Acetone, 30 °C, 250 rpm for 24 h. (see the Supporting Information for product isolation) (b,c) Representative MD snapshots showing substrates 2/3 (cyan carbons) adopting a flat orientation stabilized by hydrogen bonding between their hydroxyl groups and the backbone oxygen of G183 (blue dashed line). This positions C16α (white sphere) at optimal geometry for Cpd I (Fe=O, red sphere) attack. (d-e) Representative MD snapshots showing substrates 4/5 (cyan carbons) lacking hydrogen bonds to G183, resulting in an upright orientation that positions C6β (white sphere) within optimal Cpd I attack distance/angle. Cpd I porphyrin shown in slate blue; residue V232 in yellow.

Fig. 5. Scale-up in fermenter for converting 1 to 1a with supernatant of N1 F83GL232V. (a) 5 L of fermenter with 1 L of reaction mixture for 1 bioconversion. (b) Time course for supernatant of N1 F83GL232V-catalyzed 11β-OH-estra-4,9-diene-3,17-dione conversion. (c) Typical HPLC chromatograms for supernatant of N1 F83GL232V-catalyzed fermenter conversion of 1 to 1a. (d) Scaled-up for enzymatic synthesis of 1a. (e) Scaled-up for enzymatic synthesis of 3a. (f) Scaled-up for enzymatic synthesis of 4a.