An Optimized System for the Study of Rieske Oxygenase‐catalyzed Hydroxylation Reactions In vitro

Rieske non‐heme iron oxygenases (ROs) are primarily known for their ability to catalyze the stereoselective formation of vicinal cis‐diols in a single step, endowing valuable products for pharmaceutical and chemical applications. In addition, ROs can catalyze several other oxidation reactions with high regio‐ and stereoselectivity and typically broad substrate scope. Owing to their dependence on multicomponent electron transfer, the majority of synthetic applications of ROs relies on recombinant whole‐cell catalysts. In this context, important properties of the multicomponent system that determine the catalytic efficiency, including electron transfer via redox partner proteins, stability and uncoupling, have been investigated to a lesser extent in recent years. Here, we show for one of the most prominent ROs, the cumene dioxygenase from Pseudomonas fluorescens IP01 (CDO) that by developing and optimizing an efficient in vitro system, high catalytic activities can be achieved. In addition, we highlight that an efficient and continuous supplementation of electrons to the oxygenase is required to sustain their catalytic activity, while uncoupling can be a major limitation in CDO efficiency and stability.


Introduction
Developing strategies for the selective oxidation of (nonactivated) CÀ H bonds utilizing abundant molecular oxygen (O 2 ) is considered substantial in organic synthesis. As conventional organic synthesis routes are mainly restricted to activated or pre-oxidized functionalities within the molecule framework, the demand for more versatile and sustainable strategies is high.
Here, oxidoreductases such as Rieske non-heme iron oxygenases (ROs), which catalyze various oxidation reactions, could provide access to diverse synthons for the production of fine chemicals, pharmaceuticals and bioactive compounds. [1] ROs are the only enzymes known to catalyze the stereoselective formation of vicinal cis-diols in one step. These enzymes not only display a large substrate scope, but also catalyze various oxidation reactions, ranging from mono-and dihydroxylations, [2] sulfoxidations, [3] dealkylations, [4] desaturations, [5] to oxidative cyclizations, [6] making them particularly interesting for manifold synthetically useful reactions. [7] These enzymes are soluble multi-component systems that retrieve two of the four electrons required for the reduction of oxygen from NAD(P)H oxidation. The reduction equivalents are then supplied through electron transfer proteins to the terminal oxygenase (Oxy) where the dioxygen activation occurs. [8] The first event in the electron transfer chain (ETC) is a hydride (H À ) transfer from NAD(P)H to a flavoprotein (ETC, Figure 1). Flavoenzymes commonly annotated as reductases (Reds) play a crucial role in separating the reducing power of H À into two separate electrons yielding H + . [9] Depending on the prosthetic group of the enzyme, reduced flavin adenine dinucleotide (FADH 2 ) or flavin mononucleotide (FMNH 2 ) are subsequently oxidized via the radical semiquinone intermediate to FAD or FMN, respectively. The electrons are then shuttled either directly (two-component ETC) or indirectly (three-component ETC) via an iron-sulfur (Fe-S) protein (Fd) to the terminal Oxy component harboring the catalytic Fe(II) site. [10] The Oxy comprises the catalytically active α-subunits which either form an (α 3 ) homotrimer, [11] or additionally consist of an equivalent number of associated β-subunits with structural function that together form an (α x β x ) heteromer. [12] The typically larger α-subunits harbor a distinctive Rieske [2Fe-2S] cluster closely located to a catalytic non-heme iron center. [13] This configuration ensures the directed transfer of electrons from the Red or indirectly via the soluble ferredoxin (Fd). [14] In ROs, the mononuclear catalytic Fe is coordinated by a facial triad consisting of two histidines and 1-carboxylate residue. [9b] This motif enables the on-site binding of O 2 to the mononuclear iron and thus a variety of reactions in the hydrophobic active side of ROs. [15] Despite the fact that ROs are already known for decades, surprisingly little is known about the electron transfer (ET) from the Red to the Oxy, and with that how modulating the ETC influences the catalytic behavior. This is in stark contrast to the characterization and engineering efforts of P450s in view of redox partner-mediated ET. [16] In particular, the limited knowledge on how ROs can be efficiently applied in in vitro systems has thus far hampered a comprehensive analysis of how the interplay between redox partners ensures efficient electron transfer to sustain high catalytic rates. Due to the similar multicomponent structure, underlying principles identified for P450s can potentially be transferred to RO catalysis. [17] For instance, it could be shown for P450s that an efficient and continuous supplementation of electrons to the Oxy is required to sustain their catalytic activity. [16c,18] Furthermore, it has been proposed that the correctly timed electron delivery to the heme-domain is fundamental for high coupling rates. [19] This is particularly important as the inefficient transfer of electrons between the individual proteins results in non-productive consumption of NAD(P)H and thus uncoupling. [19a,20] The characterization of ROs in an in vitro system is thus crucial to gain a better understanding of the underlying ET principles. Cumene dioxygenase (CDO) from Pseudomonas fluorescens IP01 is a typical three-component RO. [13a,21] It comprises the Oxy (CumA1 and CumA2), a Fd (CumA3) and a Red (CumA4) containing FAD as prosthetic group. [22] CDO has been applied in various oxidative in vivo biotransformations (Scheme 1a), while the enzyme itself has been engineered to alter important aspects such as substrate scope, activity and selectivity. [23] In a recent example by Heinemann and coworkers, flexible loop regions within the Oxy of CDO prove to be powerful engineering hot-spots affecting activity, regio-as well as stereoselectivities. [24] The versatility of this engineering strategy could be successfully demonstrated for SxtT and GxtA, two enzymes involved in the biosynthesis of paralytic shellfish toxins. [25] As a prototypical RO, CDO was selected in this study as an excellent candidate for in vitro characterization and to elucidate important properties of the catalytic system that determine catalytic efficiency.
Here, we demonstrate the general feasibility of a robust in vitro system for RO-catalyzed hydroxylations, and characterize several parameters that determine the catalytic efficiency and stability of CDO, exemplified by the conversion of indene (Scheme 1b). In addition, we compared the conversion of typical CDO substrates in vivo and in vitro in view of overall conversion, as well as chemo-and enantioselectivity. Therewith, we aim to contribute to a better understanding of the underlying limitations in RO catalysis and the factors that govern the robustness of the in vitro reaction system.

Results and Discussion
To develop the desired in vitro RO system, we first established a suitable expression and purification protocol for all individual CDO components. While for typical in vivo biotransformations all CDO components were expressed from an operon-like genetic construct, [26] the in vitro system required the separate expression and purification of the Oxy, Red and Fd. According to the hexameric structure of the Oxy, we expected that only one N-terminal His-tag at the α-subunit would be sufficient to obtain the active Oxy comprising both subunits, α and β, respectively. As such, CumA1 and CumA2 were expressed from one vector construct (Figure 2a and S3), and the one-step purification confirmed that the Oxy was obtained in its functional hexameric structure (Figure 2b and Figure S4a). Moreover, the UV/VIS spectrum of the purified Oxy in its oxidized state showed absorption maxima at 323 nm, 447 nm and 549 nm, respectively, a reported characteristic for the functional Rieske [2Fe-2S] cluster in its oxidized form (Figure S5), [27] thus making downstream in vitro reconstitution, as suggested for other ROs, [28] redundant.
For P450s, it has been proposed that the correctly timed electron delivery to the heme-domain is fundamental for high coupling rates. [16c,18,29] Therefore, we speculated that an efficient Scheme 1. RO-catalyzed mono-or cis-dihydroxylation of olefins; a) in vivo approach; and b) conversion of indene 1 into 1H-indenol 1 a and 1,2indandiol 1 b catalyzed by CDO in vitro (this work). and continuous supplementation of electrons to the Oxy is also crucial for ROs to achieve high catalytic efficiencies.
To steer full conversion to the desired products, we attempted to finetune the ETC by adjusting the amounts of respective redox partners involved. By supplementing the reactions with catalase and dithiothreitol (DTT) to protect CDO from oxidative damage caused by the formation of reactive oxygen species (ROS), 5 mM 1 was completely converted within 24 h by using an adjusted ratio of Oxy : Fd : Red ( Figure 3a). The highest product concentration was obtained with a molar ratio between Oxy, Fd and Red of 1 : 7.5 : 2.5, corresponding to total protein concentrations of 4 μM, 30 μM and 10 μM. Furthermore, the concentrations of the individual CDO components in the reaction system significantly impacted the observed product formation, ranging from 1.6 mM to 5 mM.
On the one hand, an insufficient supply of electrons transported via the ETC hinders the formation of the activation complex at the non-heme iron center, leading to reduced enzyme activity. [14] On the other hand, an excess of single electrons favors the generation of unwanted ROS, which could lead to enzyme inactivation. [30] The latter effect, defined as O 2 uncoupling, [20d-f31] has been studied for a handful of ROs, [20f,32] showing that up to 65 % unproductive activation of O 2 occurs in the presence of substrate. [20e] Quantification of O 2 uncoupling in the reaction can provide information on the efficiency of the electron transfer within the multicomponent enzyme system when referenced to NAD(P)H turnover. It has been stated that O 2 can also be consumed in reactions of NAD(P)H with Red and Fd alone, [20e] indicating that O 2 uncoupling can also occur at the Red component of the system. Indeed, in NADH depletion assays with purified CDO-Red, the presence of O 2 resulted in the consumption of NADH, whereas no conversion was detected under anoxic conditions ( Figure S6). We observed that O 2 uncoupling at the Red coincided with the formation of H 2 O 2 . The amount of H 2 O 2 quantified by a horseradish-peroxidase-based assay corresponded to the generation of 35 μM H 2 O 2 within 24 h in the presence of 10 μM Red and 5 mM reduced NADH ( Figure S8).
When performing the reactions with ascending concentrations of the optimized CDO system (ratio 1 : 7.5 : 2.5 Oxy : Fd : Red), concentrations exceeding 1.9 mg ml À 1 resulted in full conversion of 1 (Figure 3b). Encouraged by these promising results, analog reactions with 20 mM 1 have been performed. However, relatively low substrate conversions (up to 17 %) were observed after 24 h, presumably caused by enzyme inhibition or deactivation at higher substrate loadings ( Figure S18).
Another important criteria for RO applicability in vitro is the stability in the presence of various cosolvents (Figure 4a), an essential criteria for the conversion of hydrophobic substrates. Therefore, reactions in the presence of dimethyl sulfoxide (DMSO), dimethylformamide (DMF), tert-butyl methyl ether (TBME) as well as acetone were conducted at different concentrations. As expected, elevated concentrations of organic solvent have a negative impact on the enzyme stability and hence yield in lower overall product formation. Already 1 vol % of acetone in the reaction mixture lowered the obtained product concentrations to under 60 % compared to the substrate conversion (100 %) determined in the control without cosolvent. Thus, acetone is the most unfavorable solvent for the proposed CDO reaction system. In comparison, reactions in the presence of 2 vol % DMSO exceeded the maximum product concentration determined in the solvent-free reaction by 14 %. This result emphasizes that DMSO as cosolvent enhances the overall solubility of 1 in the reaction mixture by increasing its accessibility for the enzyme catalyzed hydroxylation.
To evaluate the activity of the CDO reaction system at different temperatures, the product concentrations were determined at regular intervals over a period of 100 min ( Figure S21), and specific activities were determined (Figure 4b). In accordance with previous in vivo studies on ROs, [24b,33] an incubation temperature of 30°C turned out to be most suitable for the  CDO in vitro system, revealing a specific activity of about 50 mU mg À 1 . At incubation temperatures exceeding 40°C, no product formation was detected anymore.
Steady-state kinetic parameters of the CDO in vitro system for the conversion of 1 ( Figure S20) were determined based on time-course experiments at different substrate concentrations. These experiments reveal an apparent K M of 0.05 mM and an apparent maximum reaction rate (V max ) of 0.98 μM min À 1 . In addition, the catalytic efficiency, determined by the turnover number k cat , was calculated as the ratio of V max and the applied enzyme amount, taking into consideration the number of active sites (n = 3) per hetero-hexamer of the Oxy. By that, an apparent k cat of 0.3 min À 1 was determined, which corresponds to the number of molecules converted per active site as a function of time. While a turnover rate of approximately one molecule of 1 per minute catalyzed by one α 3 β 3 hexamer of Oxy seems rather slow for enzymatic reactions, the obtained values must be interpreted with care. The catalytic efficiency of the CDO in the in vitro system was evaluated based on steady-state kinetics. Besides facing the problem of limited solubility of 1 in the aqueous reaction mixture, the performance of CDO is further dependent on the supply of NADH, O 2 and electrons provided by the ETC. It is assumed that the transfer of electrons to the active site of the Oxy can be apparently enhanced by fine-tuning the amounts of redox partners (Figure 3). However, this assumption does not imply the loss of electrons in the ETC due to uncoupling. Besides lowering the electron transfer yield of the ETC, considering O 2 exclusively as cosubstrate for the kinetic characterization of the CDO is not fully applicable by neglecting the effect of enzyme instability caused by oxidative stress. In addition, attempts to assess O 2 consumption using an O 2sensitive electrode failed due to the low V max of CDO which do not match the sensitivity of the oximeter. Therefore, the kinetic data obtained here should be treated as apparent values, taking into account that the environment plays a crucial role in the activity of CDO.
Finally, the chemo-and enantioselectivity of the in vitro CDO system was evaluated and compared with previously reported in vivo approaches. [24b,c] For this, diverse arene-substituted alkenes (1-2), terpenes (3) as well as cycloalkenes (4-5), reported to be accepted by ROs, have been investigated (Table 1). [24c,33] For product identification, a previously reported variant of CDO (M232A), [24c] was used together with the wild-type.
In both reactions systems, CDO favors the formation of 1 a over 1 b, while comparable diastereomeric ratios have been obtained. In terms of the asymmetric dihydroxylation of 1 to 1 b, CDO shows in both cases a preference for the formation of the (1S, 2R)-cis-1 b. For both, in vitro and in vivo reactions, 2 b was identified as main product in the conversion of 2, with overall > 99 % conversion. Consistent with previous studies for in vivo reactions, CDO also favors dihydroxylation of the aromatic ring structure over the vinylic functional group of 2 in the cell-free system, yielding > 99 of 2 b. While performed in vivo reactions with CDO revealed a conversion of 73 � 6 % of 3 into 3 b, the in vitro system resulted in almost full conversion (96 � 1 %) to (1R,5S)-3 b. In addition, reactions performed with 4 and 5 reached full conversion with both systems after 24 hours, while an ee of > 99 % (R) was obtained for 4 a in vitro, and a slightly lower ee of 91 % of 4 a was obtained in vivo (Table 1). Thus, obtained conversions in both reaction systems are highly similar, although a direct comparison between the two systems is challenging due to differences in protein amount and stability. Interestingly, reactions performed with 20 mM 3 resulted in almost full conversion (97 � 1 %) with the in vitro system and only 39 % with the in vivo system. It is therefore suggested that the interplay between mass transport effects across the cell membrane, unbalanced ETC and possible biotoxicity of the substrate in the in vivo approach contribute to the lower product formation. This assumption is further supported by the fact that the observed chemo-and enantioselectivity of CDO is independent of the reaction environment. The control reactions performed confirmed the overall functionality of the ETC (Table S11) and the absence of unwanted background reactions caused by components in the reaction mixture.
These results demonstrate that an efficient CDO in vitro system for asymmetric hydroxylation reactions has been established, with substrate ranges and selectivities consistent with previously reported whole-cell studies. [24c] Furthermore, conversions of up to 20 mM 3 in vitro also indicate that the established cell-free reaction system may operate more efficiently compared to the in vivo system for certain substrates.

Conclusions
Driven by the potential of ROs for applications in organic synthesis, gradually more and more studies focus on expanding the substrate and reaction scope of these enzymes. [23,24b,c] However, due to their assumed instability in in vitro systems and the need for a multicomponent ETC, most of these studies aim at implementing whole-cells harboring heterologous expressed ROs as biocatalysts. While the features of ROs are very similar to those of P450s, ROs are not well understood, especially in view of the electron transfer from the redox partner proteins to the enzymes' active site, determining their catalytic efficiency in in vitro systems. We herein demonstrate that the electron transfer from NADH to the active site of the Oxy can be finetuned and significantly improved by carefully adapting the ratio of all components of the system. Thereby, an optimum molar ratio of 1 : 7.5 : 2.5 (Oxy : Fd : Red) turned out to be most efficient, enabling the full conversion of 10 mM 1 into 1 a and 1 b in vitro. Interestingly, previous studies of other ROs typically report a ratio of 1 : 12 : 1 (Oxy : Fd : Red, normalized to 1 μM).
[20e,f] However, we were able to show that efficient electron transfer from NADH can be achieved with a smaller excess of the electron transfer protein Fd. While an excess of Fd is beneficial to ensure efficient electron transfer, a non-optimal Oxy : Fd : Red ratio has several consequences for the overall catalytic efficiency of the RO system. In particular, O 2 uncoupling and possibly the formation of ROS in the form of H 2 O 2 not only reduces the efficiency of substrate hydroxylation, but could also damage the enzyme system by hydroxylation of amino acid residues and subsequent enzyme inactivation, a wellknown phenomenon in non-heme ferrous iron oxygenases. [20d,f,30b] While the formation of H 2 O 2 due to O 2 uncoupling at the Oxy has been quantified and consequences on substrate hydroxylation have been assessed, [20e,f,32b] our data also suggests that uncoupling at the reductase may also significantly contribute to H 2 O 2 formation within the RO system as a consequence of non-optimal Oxy : Fd : Red ratios. Taken together, we highlight that an efficient and continuous supplementation of electrons to the Oxy is required to sustain their catalytic activity, while uncoupling can be a major limitation in CDO efficiency and stability.
Finally, the substrate scope of CDO in vitro was compared to previously reported in vivo approaches, revealing no significant differences in CDO selectivity. However, the herein developed in vitro system seems to be beneficial for substrates with high biotoxicity or low uptake rate by the cells. Overall, using the protype RO CDO as an example, we successfully demonstrate the development, optimization, and application of a threecomponent in vitro reaction system. While this is the first step in broadening our understanding about the electron transfer in ROs, we believe that the insights gained herein will inspire further advances for optimizing the ETC and with that catalytic efficiencies of ROs for diverse applications in organic synthesis. Ultimately, we expect protein engineering to be used in a similar way to P450s to fully exploit the biotechnological potential of these complex oxidoreductases for the sustainable synthesis of various high-value chemicals and pharmaceuticals.

Experimental Section
Heterologous expression and purification of CDO: Heterologous expressions of single CDO components were performed in E. coli JM109 (DE3) cells harboring pET-28a(+)-based plasmid constructs containing genes for CumA1 and CumA2, CumA3 or CumA4 (see Table S2). Precultures of respective E. coli JM109(DE3) cells were prepared by inoculating 50 mL LB liquid medium supplemented with 50 μg mL À 1 of kanamycin (Kan) in 250 mL baffled Erlenmeyer flasks. After overnight (~16 h) incubation at 37°C and 200 rpm, obtained high cell density precultures were used to inoculate 400 mL of TB liquid medium supplemented with 50 μg mL À 1 of Kan to adjust an initial cell density of OD 600 0.1. After incubation at 37°C at 150 rpm for approximately 1.75 h, protein expression was induced at a cell density of~1 by adding isopropyl β-D-1-thiogalactopyranoside (IPTG) at a final concentration of 50 μM. Protein expression was performed at 20°C and 150 rpm for 19 h. Afterward, cells were harvested by centrifugation (3428 g at 4°C for 30 min). After washing with 50 mL of Equilibration Buffer (25 mM Na 2 HPO 4 , 25 mM NaH 2 PO 4 , 300 mM NaCl, 10 % glycerol, 30 mM imidazole, pH 7.2), cells were subsequently pelleted via centrifugation at 3428 g and 4°C for 45 min. After resuspending with 30 mL Equilibration Buffer, a spatula tip of lyophilized lysozyme from chicken egg white (protein 90 %, 40,000 U mg À 1 , SIGMA®, L6876-10G) was added. After 15 min incubation at a rocking platform on ice, cells were disrupted via sonification using Branson 450 Analog Sonifier. Therefore, a duty cycle of 50 % and output control of 7 were set. The sonification was performed over 3.75 min, keeping pulse rates of 15 sec on and 15 sec off. The lysates obtained after 45 min centrifugation at 38465 g and 4°C were subsequently filtered using 0.45 μM syringe filters (FP 30/0.45 CAS,0.45 μm). The purification-and desalting steps were carried out constantly in the cold room at 5°C. The obtained clear lysates were subsequently transferred into 2x 15 mL gravity columns packed with 1 mL (1 CV) of Ni-sepharose histidine-tagged protein purification resin each and incubated for 1 h. After withdrawing the flowthrough, the columns were washed with 25 CVs of Equilibration Buffer. Elution was performed by adding 3 times 2 CVs of Elution Buffer (25 mM Na 2 HPO 4 , 25 mM NaH 2 PO 4 , 300 mM NaCl, 10 % glycerol, 400 mM imidazole, pH 7.2). Subsequent buffer exchange from Elution Buffer to Desalting Buffer (25 mM Na 2 HPO 4 , 25 mM NaH 2 PO 4 , pH 7.2) was performed using PD-10 desalting columns purchased from Cytiva (Danaher, Massachusetts, US) according to the gravity flow protocol provided by the manufacturer. Macrosept Advance Centrifugal Device (Pall Corporation, New York, US) concentrator columns (10 K) were used to concentrate the obtained protein samples. Protein samples were directly used in in vitro biotransformation studies.

Spectrophotometric quantification of purified protein:
For determining protein concentrations within aqueous buffer solutions, a colorimetric Coomassie G-250-based assay was applied. Therefore, desired protein samples were resuspended with a Coomassie Protein Assay Reagent following the supplier's manual (Thermo Scientific, MA, US). Sample preparation was performed within 96well plates and the resulting color change was detected via spectrophotometric measurement using a plate reader at 595 nm. To evaluate the total protein concentrations of respective samples, diluted albumin standards within a working range of 20-2.000 μg mL À 1 were used for establishing a calibration curve. Measurements were performed in triplicates.

Preparation of in vitro reactions using purified CDO:
In vitro reactions performed throughout this study were prepared in duplicates using freshly purified CDO. As a buffer system, 50 mM sodium phosphate buffer (25 mM Na 2 HPO 4 , 25 mM NaH 2 PO 4 , pH 7.2) was used supplemented with reaction components as indicated in the respective experimental descriptions. The total reaction volume was set to 1 mL using 20 mL airtight-sealed glass vials as reaction vessel. For maintaining the reaction conditions, biotransformations were conducted using incubation shakers.

Heterologous expression of CDO and CDO M232A for in vivo studies:
For in vivo biotransformation studies, resting E. coli JM109(DE3) cells were used as biocatalysts containing overexpressed CDO. For that, 50 mL of LB liquid media containing 100 μg mL À 1 of ampicillin (Amp) were initially inoculated by using a single-colony of E. coli JM109(DE3) harboring CDO plasmid constructs (Table S2), respectively. After overnight incubation for 16 h at 200 rpm and 37°C, the precultures were used for inoculating 400 mL of TB liquid media to an initial OD 600 of 0.1.
Main cultures for the heterologous expression of CDO and CDO M232A variant were cultivated in the presence of Amp (100 μg mL À 1 ). Cultivation was performed at 37°C and 150 rpm until a cell density of OD 600 0.6-0.8 was observed. Following protein expression was induced by adding IPTG at a final concentration of 200 μM. After incubating the main cultures at 30°C and 150 rpm for 2 h, the cells were harvested using a precooled (4°C) high-speed centrifuge (3428 g for 15 min). The resting cells were directly used as whole-cell biocatalysts in in vivo reactions.
Preparation of in vivo reactions using resting E. coli JM109(DE3) cells containing CDO and CDOM232A: In vivo reactions performed throughout this study were prepared in duplicates using resting E. coli JM109(DE3), harboring overexpressed CDO, as biocatalysts. Therefore, 50 mM sodium phosphate buffer (25 mM Na 2 HPO 4 , 25 mM NaH 2 PO 4 , pH 7.2) was used as buffer system. Cells were applied at a final concentration of 200 mg mL À 1 within the reaction mixture which was set to 1 mL throughout the project. The reaction mixture was supplemented with 20 mM of D-glucose as substrate for NADH cofactor regeneration. The reactions were initiated by adding respective amounts of substrates directly to the reaction mixtures (total reaction volume 1 mL) prepared in 20 mL airtightsealed glass vials. To guarantee consistent reaction conditions, biotransformations were conducted within incubation shakers.

Quantification of organic components via GC-FID and GC-MS:
The quantification of organic compounds including substrates and products obtained after biotransformations was conducted via GC-FID or GC-MS. Therefore, liquid-liquid extraction was performed using dichloromethane (containing 2 mM of acetophenone as internal standard) as solvent. To increase phase separation, the aqueous reaction mixture was saturated with NaCl. Extraction was performed at a ratio of 1 : 1 with a total volume of 1 mL. Prior injection for GC measurement, the organic phase was dried over anhydrous MgSO 4 . For further details on applied GC-FID and GC-MS parameters please see supporting information. Non-chiral GC-FID measurement was performed using a HP-5 column (30 m, 0.32 mm, 0.25 μm) purchased from Hewlett Packard, (California, USA). Chiral GC-FID analysis was performed using Astec® CHIRALDEX™ G-TA column (30 m, 0.25 mm, 0.12 μm) from Sigma-Aldrich (St. Louis, USA) or Hydrodex β-6TBDM column from Macherey-Nagel™ (Düren, Germany). For non-chiral quantification via GC-MS, a HP-5 ms column purchased by Agilent J&W (Santa Clara, USA) was used. For further details and applied parameters see Table S7 (non-chiral GC-FID), Table S8 and Table S9 (chiral GC-FID) and Table S10 (non-chiral GC-MS) in the supporting information.

Quantification of H 2 O 2 via colorimetric HRP-ABTS assay:
To determine the concentration of H 2 O 2 within the aqueous reaction solution, a colorimetric assay based on the conversion of 2,2'-azinobis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) by a peroxidase from horseradish (HRP) was performed. In the presence of H 2 O 2 the HRP converts ABTS into a green product with an absorbance maximum at 414 nm. Assay reactions were performed in 96-well microtiter plates and the absorbance of the colorimetric product recorded with a plate reader (SPECTROstar Omega from BMG LABTECH, Ortenberg in Germany). Reactions were performed at a total reaction volume of 250 μL. The reaction mixture for H 2 O 2 determination in aqueous sample typically contains 50 μM of ABTS (ABTS diammonium salt purchased from Sigma-Aldrich, St. Louis, USA) and 10 μg HRP (150-250 U mg À 1 , lyophilized powder purchased from Sigma-Aldrich, St. Louis, USA) within 100 mM SPB (pH 6.0).