The expression-engineered variant rAaeUPO (PaDa-I mutant) was typically used as concentrated supernatant or purified enzyme derived from a 2,500 L pilot-scale cultivation of recombinant Pichia pastoris X-33 expressing rAaeUPO (Molina-Espeja et al., 2014; Tonin et al., 2021). Purification was similarly performed according to previous studies (Molina-Espeja et al., 2014). rAaeUPO concentrations were determined in CO-difference spectra (see below) or via their absorption at 420 nm using the extinction coefficient (ε ₄₂₀ = 115 mM⁻¹ cm⁻¹) (Fernández-Fueyo et al., 2016).

The amount of rAaeUPO in concentrated supernatants was determined using CO-difference spectra. 950 µL of protein sample, diluted in 100 mM KPi-buffer if necessary, were filled into plastic cuvettes and placed in a photometer. After recording a blank (base subtraction), the sample was exposed with CO for a few seconds. Next, 50 µL of a 1 M sodium dithionite stock solution was added, and a difference spectrum between 400 and 500 nm was recorded. The measurements were continued until a constant absorption maximum was obtained. All measurements were performed in duplicates. The concentration of rAaeUPO was calculated based on the absorption values at 445 and 490 nm using the molar extinction coefficient of ε₄₄₅ = 107 mM⁻¹ cm⁻¹ (Tieves et al., 2019a).

The peroxidase activity of rAaeUPO was determined in a continuous photometric assay against the substrate 2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) at 420 nm (ε ₄₂₀ = 36.0 mM⁻¹ cm⁻¹). AaeUPO (50 µL) in appropriate dilution (typically 50- or 100-fold) was mixed with 0.5 mM ABTS in 100 mM sodium citrate buffer, pH 4.4 in 2 mL plastic cuvettes. After short incubation at 25°C, measurements were started by the addition of 2 mM H₂O₂. The increase of absorption at 420 nm caused by ABTS formation was tracked for 90 s at 25°C in a Cary 60 UV–VIS spectrophotometer (Agilent). The initial slope (ΔA₄₂₀/min) between 20–80 s was linear and used to calculate the volumetric activity [U/mL]. 1 U is defined as the amount of enzyme which is needed to convert 1 µmol substrate in 1 min under assay conditions. All measurements were performed in triplicates.

The activity of rAaeUPO against the primary alcohols methanol and ethanol was examined using the purpald assay (Dickinson and Jacobsen, 1970; Hopps, 2000). Reaction mixtures contained 1 µM rAaeUPO (concentrated supernatant or purified enzyme), 20 vol% co-solvent, and H₂O₂, which was added with a dosing rate of 4 mM h⁻¹. The 200 mM H₂O₂-stock solution was freshly prepared from commercially available H₂O₂ (50 wt% solution, Merck). For analysis, 5–50 µL sample was diluted in 100 mM potassium phosphate buffer, pH 7.5 to a final volume of 200 µL in a 96-well plate. Afterward, 50 µL purpald (0.16 M in 2 M NaOH) was added, and samples were properly mixed. The samples were incubated at room temperature for at least 1 h, allowing the formation of a purple dye in the presence of aldehydes and oxygen.

rAaeUPO-mediated oxidation reactions were typically performed in 0.5 mL 100 mM potassium phosphate buffer, pH 6.0 in 2 mL glass vials. Reaction mixtures contained 1 µM rAaeUPO (concentrated supernatant), 10 mM ethylbenzene or cis-β-methyl styrene (in 10–97 vol% of acetonitrile), and H₂O₂, which was added with a dosing rate of 4 mM h⁻¹. The 200 mM H₂O₂-stock solution was freshly prepared from commercially available H₂O₂ (50 wt% solution, Merck). The samples were incubated at 25°C and 600 min⁻¹ for 0–3 h in an Eppendorf shaker.

In order to track the substrate oxidation with automated H₂O₂ addition, reactions in a 10 mL scale were prepared in glass vessels (Supplementary Figure S5). The glass devices were placed in an Eppendorf shaker equipped with a 50 mL falcon attachment, which was fixed in such a way that proper shaking and correct placement of the glass vials were enabled as it would the case for 50 mL falcons. Springe pumps containing a stock solution of 800 mM H₂O₂ (freshly prepared from a 50 wt% H₂O₂) were connected and automatically pumped 100 µL of H₂O₂ every hour (dosing rate: 8 mM h⁻¹ H₂O₂).

Ethyl acetate (500 µL) containing 5 mM of the internal standard n-dodecane was used for extraction. After addition, samples were vortexed and centrifuged at 13.000 x g for 3 min at room temperature. The organic phase was transferred to a plastic vial with inlet and used for GC analysis.

The kinetic parameters K_m and k_cat for the rAaeUPO substrates ethylbenzene and 1-phenylethanol were determined in the presence of different concentrations of the co-solvent acetonitrile. Reaction mixtures (8 mL) in 10 mM KPi buffer (pH 7) contained 8.4 nM rAaeUPO (concentrated supernatant), 1.7 mM H₂O₂ (freshly diluted from a 50 wt% H₂O₂ stock solution), and the substrates ethylbenzene or 1-phenylethanol at concentrations ranging from 0.03–2 mM. Acetonitrile concentrations varied from 0–20 vol%. For reactions without acetonitrile, the maximum concentration of ethylbenzene was set at the solubility limit of 1.2 mM (150 mg/L, IPCS Inchem.). Samples were incubated at room temperature for 15 s, and then the reaction was stopped by adding 1 mL 50 vol% trifluoroacetic acid. Afterward, the samples were extracted with ethyl acetate containing 5 mM of the internal standard 1-octanol, dried over magnesium sulfate, and analyzed by GC. K_m and v_max values were obtained by fitting the reaction rate (µM product per min) vs. the substrate concentration according to the equation of Michaelis–Menten by non-linear regression using Prism GraphPad, version 5 (Supplementary Figures S9–S11). The turnover number k_cat was calculated by dividing v_max by the total UPO concentration.

The rAaeUPO-mediated hydroxylation of ethylbenzene on a 100 mL scale was performed in a SYSTAG jacketed lab reactor (250 mL operational volume) at 25°C and 300 rpm shaking speed. Reaction solution (100 mL) contained 100 mM KPi buffer (pH 6), 50 vol% acetonitrile, 10 µM rAaeUPO (concentrated supernatant), and 300 mM ethylbenzene. H₂O₂ was added continuously from a 4.5 M stock solution with a H₂O₂-dosing rate of 50 mM h⁻¹ and was freshly prepared from a commercial 50 wt% H₂O₂ solution (Merck) prior to the experiment. At different time points, samples from the aqueous phase were withdrawn, extracted with ethyl acetate containing 5 mM n-dodecane (IS), and analyzed via chiral GC.

For GC analysis, a Shimadzu GC-14A/FID or Shimadzu GC-2010 plus/FID system was used. Achiral GC-analysis was performed on an Agilent CP-Sil 5 CB column (50 m × 0.53 mm × 1.0 μm) with N₂ as carrier gas. Kinetic studies were analyzed on a CP-Wax 52 CB column (25 m × 0.25 mm × 1.2 µm) using N₂ as carrier gas. Chiral GC-analysis for cis-β-methyl styrene and products was performed according to Rauch et al. on a Lipodex E column (Macherney Nagel) using helium as carrier gas (Rauch et al., 2019). Ethylbenzene and oxidized products thereof were analyzed on a CP-Chirasil-Dex CB (Agilent) (25 m × 0.32 mm × 0.25 μm) with helium as carrier gas (Tieves et al., 2019b). Substrate and product concentrations were calculated based on calibration curves. Because an authentic standard for (1R,2S)-cis-methyl styrene oxide was not available, the response factor between calibration curves of styrene and styrene epoxide was used to calculate the concentration of (1R,2S)-cis-methyl styrene oxide relative to cis-β-methyl styrene.

Temperature programs are described in key points as follows:

To investigate the effect of water-soluble organic co-solvents on the activity of rAaeUPO, we evaluated a selection of commonly used solvents ranging from dimethyl sulfoxide (DMSO) (logP = −1.35) to isopropanol (logP = 0.05) at a substrate concentration of 10 mM. As model reactions, we chose the hydroxylation of ethylbenzene and the epoxidation of cis-β-methyl styrene (Figure 1) in the presence of 20 vol% of the respective solvents. It is worth mentioning that the enantioselectivity of both reactions was not significantly impaired by the co-solvent (Supplementary Figures S1, S2).

Both model reactions were influenced by the presence of the solvents in a similar fashion. There was no obvious correlation between the polarities of the solvent (as expressed by their logP value). Specially for the alcohol-based solvents methanol (MeOH) and ethanol (EtOH), we suspected that possibly alcohol oxidation may compete with the model reaction investigated. Note that the molar surplus of alcohol solvent over the starting material was approximately 250-fold. In fact, using the purpald assay (Dickinson and Jacobsen, 1970; Hopps, 2000), we could qualitatively confirm the formation of formaldehyde and acetaldehyde from the corresponding alcohol oxidations catalyzed by rAaeUPO (Figure 2). We assume that also isopropanol at least to some extend can be converted by rAaeUPO but an analytical validation is yet pending. Also in case of DMSO, which is an inhibitor of peroxygenase activity, GC/MS analysis confirmed rAaeUPO-conversion of the solvent to the sulfone (Supplementary Figure S6). For acetone and acetonitrile, we did not observe any indication for co-solvent oxidation in our experiments and nearly full conversion (≥90%) of 10 mM substrate was achieved under the experimental conditions chosen.

We next investigated the influence of the acetonitrile concentration on the rAaeUPO-catalyzed hydroxylation reaction (Figure 3). Increasing the acetonitrile concentration up to approximately 50 vol% had a beneficial effect on product formation. Notably, even in the presence of very high acetonitrile concentrations of 90 vol% significant activity was detected. Possibly, the increasing product accumulation up to 50 vol% can be attributed to the increasing solubility of the reactants in the reaction medium thereby also minimizing the undesired evaporation. With increasing acetonitrile concentrations, also a larger amount of the alcohol product (1-phenylethanol) compared to the double oxidation product acetophenone was formed. We decided to investigate this effect by further determining the kinetic parameters K_m and k_cat for the substrates ethylbenzene and 1-phenylethanol at different acetonitrile concentrations (Table 1). It was found that increasing acetonitrile concentrations did not significantly affect the kinetic parameters for the substrate ethylbenzene. In contrary, higher K_m and higher k_cat values for the substrate 1-phenylethanol were observed when the acetonitrile amount was increased. Currently, we are lacking a full understanding of this behavior as the better solubility of the reactants as well as the solvent presence in the active site might contribute to these changes in kinetic parameters and the product concentrations, respectively. Thus, further investigations in this regard need to be performed. Nevertheless, the use of increased acetonitrile concentrations seems attractive in order to avoid over oxidation to the ketone and obtain an accumulation of the desired hydroxylation product.

Increasing the co-solvent concentration beyond 50 vol% most likely has a negative impact on the structural integrity of the biocatalyst explaining the decreasing product accumulation (Figure 3). Similar observations were also made for the epoxidation of cis-β-methyl styrene, indicating that these effects are independent on the tested substrate (Supplementary Figure S7). To further investigate the influence of acetonitrile on the stability of rAaeUPO, we performed a range of pre-incubation experiments exposing the enzyme to varying concentrations of acetonitrile and determining at intervals the residual activity (Figure 4). While the enzyme, incubated in buffer without any acetonitrile, showed no significant decrease in catalytic activity over several days, a sharp drop was observed in the presence of 50 and 80 vol% acetonitrile, resulting in approximately 40 and 1% of residual enzyme activity after 1 h. In the presence of 50 vol% acetonitrile, rAaeUPO activity did not decrease further for several days. The dramatic decrease in catalytic activity was accompanied by a green precipitate evolving in the incubation mixture. We assume that the dramatic activity losses in the first few hours of incubation can be attributed to precipitation of the biocatalyst. Activity measurements of the recovered precipitate from 50 vol% acetonitrile resuspended in buffer indicated that the precipitated UPO was still active (Supplementary Figure S12). Thus, the loss of activity in 50 vol% acetonitrile is most likely not due to enzyme inactivation but rather to a decreased solubility of the enzyme in the acetonitrile/buffer mixture. Consequently, future studies might provide exciting possibilities to use acetonitrile for UPO precipitation to easily obtain active enzyme immobilisates and compare them to previous immobilization procedures (Rauch et al., 2019; Hobisch et al., 2020b).

Complementary to the stability measurements, we followed the rAaeUPO-catalyzed oxidation of the model substrate ethylbenzene over time in the presence of high concentrations of acetonitrile (Figure 5). We observed a different effect of the tested acetonitrile concentration of 50 and 80 vol%. The product concentration increased linearly within 7 h in the presence of 50 vol% acetonitrile, whereas the reaction stopped earlier in the presence of 80 vol% acetonitrile. This resulted in a maximum target product concentration of 57 mM (R)-1-phenylethanol in 80 vol% acetonitrile. Looking at the stability measurements (Figure 4), the lower product formation in the presence of 80 vol% acetonitrile can be attributed to insufficient enzyme activity and stability in the presence of co-solvent. In this regard, it is still astonishing that over 50 mM (R)-1-phenylethanol are formed though the enzyme activity was less than 1% compared to its initial value according to our stability measurements. Using 50 vol% acetonitrile for the UPO-mediated oxidation of cis-β-methyl styrene resulted in similar trends, indicating that these conditions can also be applied for other UPO substrates (Supplementary Figure S8).

Encouraged by these promising results, we decided to demonstrate the preparative-scale application using an ethylbenzene starting concentration of 300 mM (Figure 6). The reaction produced smoothly accumulating enantiomerically pure (R)-1-phenylethanol until nearly all starting material was consumed after 6 h. Only then, over oxidation to acetophenone occurred. Within 6 h, 188 mM of enantiomerically pure (R)-1-phenylethanol has been formed corresponding to a turnover number (TN = mol_Product × mol_rAaeUPO ⁻¹) of 18.800 and a turnover frequency (TOF) of 0.9 s⁻¹ (over 6 h). These turnover numbers were lower compared to previous results from our group, where for the same substrate and enzyme in a two-liquid phase system TNs of 340.000 were achieved (Tonin et al., 2021). However, it should be pointed out that the achieved concentration of 120 mM (R)-1-phenylethanol was obtained after 7 days of incubation resulting in a low product accumulation rate of 0.7 mM h⁻¹. The initial (R)-1-phenylethanol accumulation rate in our experiments was more than 31 mM h⁻¹ (3.8 g_product h⁻¹ L⁻¹) and thus represent to the best of our knowledge one of the highest productivities reported for a biocatalytic oxyfunctionalization reaction to date.

Overall 216 mM of acetophenone (double oxidation product) were obtained corresponding to 25.9 g L⁻¹ and 73% yield on the starting material used. The product to the catalyst ratio was 70 g_product g⁻¹ _AaeUPO. As no starting material and only minor (R)-1-phenylethanol was detected after 24 h, we conclude that still evaporation of the reagents (especially the starting material) occurred. Nevertheless, these numbers bring peroxygenase-catalysis close to industrial requirements (Tufvesson et al., 2011).

From an environmental point of view, we believe that the application of acetonitrile represents an interesting alternative to pure aqueous media. A simple E-factor analysis (Sheldon, 2017; 2018) (Figure 7) revealed that in the experiment shown in Figure 6 approximately 35.2 g_waste g⁻¹ _product have been generated. 98% of this waste has been caused by the solvents (19.3 g_water g⁻¹ _product and 15.2 g_acetonitrile g⁻¹ _product). The non-reacted starting material and phosphate buffer contributed roughly 1 g g⁻¹ _product each, whereas the contribution of the biocatalyst was almost negligible (<0.02 g _AaeUPO g⁻¹ _product). Hence, the overall E-factor compared to similar reactions could be reduced more than 7-fold (Xu et al., 2022). In this regard, the direction for further improvements of the reaction’s environmental footprint is also clear by further increasing the product concentration.

Obviously, a simple E-factor analysis does not replace a more detailed environmental impact analysis taking the history of the reagents used and energy contributions into account (Tieves et al., 2019a). Nevertheless, this analysis already demonstrates the impact of increasing product concentrations.