Biocatalytic Syntheses of Antiplatelet Metabolites of the Thienopyridines Clopidogrel and Prasugrel Using Fungal Peroxygenases

Antithrombotic thienopyridines, such as clopidogrel and prasugrel, are prodrugs that undergo a metabolic two-step bioactivation for their pharmacological efficacy. In the first step, a thiolactone is formed, which is then converted by cytochrome P450-dependent oxidation via sulfenic acids to the active thiol metabolites. These metabolites are the active compounds that inhibit the platelet P2Y12 receptor and thereby prevent atherothrombotic events. Thus far, described biocatalytic and chemical synthesis approaches to obtain active thienopyridine metabolites are rather complex and suffer from low yields. In the present study, several unspecific peroxygenases (UPOs, EC 1.11.2.1) known to efficiently mimic P450 reactions in vitro—but requiring only hydroperoxide as oxidant—were tested for biocatalytic one-pot syntheses. In the course of the reaction optimization, various parameters such as pH and reductant, as well as organic solvent and amount were varied. The best results for the conversion of 1 mM thienopyridine were achieved using 2 U mL−1 of a UPO from agaric fungus Marasmius rotula (MroUPO) in a phosphate-buffered system (pH 7) containing 5 mM ascorbate, 2 mM h−1 H2O2 and 20% acetone. The preparation of the active metabolite of clopidogrel was successful via a two-step oxidation with an overall yield of 25%. In the case of prasugrel, a cascade of porcine liver esterase (PLE) and MroUPO was applied, resulting in a yield of 44%. The two metabolites were isolated with high purity, and their structures were confirmed by MS and MS2 spectrometry as well as NMR spectroscopy. The findings broaden the scope of UPO applications again and demonstrate that they can be effectively used for the selective synthesis of metabolites and late-state diversification of organic molecules, circumventing complex multistage chemical syntheses and providing sufficient material for structural elucidation, reference material, or cellular assays.


Introduction
Clopidogrel (Plavix ® , Iscover ® ) and prasugrel (Efient ® ) are antithrombotic prodrugs of the thienopyridine family that ignite their intrinsic effect after metabolic bioactivation and irreversibly inhibit the platelet P2Y 12 receptor resulting in prevention of atherothrombotic events [1][2][3]. The two-step enzymatic bioactivation required for both prodrugs differs in its initial reactions: In the case of clopidogrel, the hydroxylation of the thiophene ring, required for the spontaneous formation of the thiolactone metabolite, 2-oxo-clopidogrel prodrugs differs in its initial reactions: In the case of clopidogrel, the hydroxylation of the thiophene ring, required for the spontaneous formation of the thiolactone metabolite, 2oxo-clopidogrel (Scheme 1), is catalyzed by cytochrome P450 (CYP) monooxygenases [4,5]. In case of prasugrel, the corresponsive thiolactone (R-95913) is formed by hydrolysis of its ester functionality, which seems to be mainly catalyzed by human carboxylesterases hCE2 [6]. The final transformation of these thiolactone derivatives into the respective active thiol metabolites is formally a thioester hydrolysis, since the oxidation state remains unchanged. In fact, thioesterases such as paraoxonase-1 (PON-1) are thought to accomplish this kind of hydrolysis, but the reactions studied led to the biologically inactive endo-isomer, in which the double bond has migrated into the piperidine ring [7,8]. Recent studies disclosed the formation of the active metabolite as a multistep biochemical process that also involves CYP enzymes, while the exact chemical nature of the intermediates is still under discussion. The most plausible metabolic process describes the formation of thiolactone sulfoxides during the oxidative activation, the hydrolytic opening of which leads to sulfenic acids that are efficiently reduced by ascorbate, phosphines, reductases or glutathione to give the active thiol metabolites (Scheme 1) [9][10][11][12]. Scheme 1. Proposed mechanism of bioactivation and conversion of clopidogrel and prasugrel into their active thiol metabolites, CAM and PAM, by cytochrome P450 enzymes via formation of thiolactone, thiolactone sulfoxides and sulfenic acids as intermediates. An alternative inactivation pathway has been described to proceed via hydrolysis catalyzed by thioesterases (PON-1) leading to the formation of the endo-thiol metabolite.
The bioactivation of clopidogrel to form 2-oxo-clopidogrel and the active thiol metabolite can be catalyzed by several CYP enzymes. Kinetic and inhibition studies, however, revealed that just CYP2C19 contributes substantially to both oxidative steps and that CYP3A4 is essential for the formation of the active metabolite. Furthermore, CYP1A2, Scheme 1. Proposed mechanism of bioactivation and conversion of clopidogrel and prasugrel into their active thiol metabolites, CAM and PAM, by cytochrome P450 enzymes via formation of thiolactone, thiolactone sulfoxides and sulfenic acids as intermediates. An alternative inactivation pathway has been described to proceed via hydrolysis catalyzed by thioesterases (PON-1) leading to the formation of the endo-thiol metabolite.
The bioactivation of clopidogrel to form 2-oxo-clopidogrel and the active thiol metabolite can be catalyzed by several CYP enzymes. Kinetic and inhibition studies, however, revealed that just CYP2C19 contributes substantially to both oxidative steps and that CYP3A4 is essential for the formation of the active metabolite. Furthermore, CYP1A2, CYP2B6 and CYP2C9 may be involved in the oxidative cascade [5]. Several studies revealed that polymorphism in the gene encoding CYP2C19 contributes to the variability of platelet response to clopidogrel in patients. Individuals who carry a reduced-function CYP2C19 allele had significantly lower levels of the active metabolite of clopidogrel, diminished platelet inhibition and a higher rate of major adverse cardiovascular events [13][14][15]. In addition, clopidogrel undergoes an extensive first-pass metabolism in the human liver where most of the prodrug is hydrolyzed to inactive clopidogrel carboxylic acid by carboxylesterase hCE1, followed by glucuronidation and renal excretion [16]. To address these causes for the low plasma level of active metabolites, prasugrel was developed replacing the ester functionality with a metabolically stable ketone and the introduction of an ester group in the thiophene 2-position that switched the first-step activation from CYP2C19 to carboxylesterase hCE2 [17].
In vitro metabolism studies showed that metabolic activation of clopidogrel results in a mixture of four stereoisomers (H1-H4, Figure 1). The H1 and H2 isomers are trans isomers, whereas the exocyclic double bond between C3 and C16 in H3 and H4 appears in cis configuration. Analysis of clinical samples demonstrated that only H3 and H4 are formed in vivo, with only H4 being active. Therefore, the configuration of the thiol group at C4 seems to be crucial for antiplatelet activity [18]. The reactive free thiol group of the active metabolites leads to high instability in plasma samples, which is why they were stabilized by derivatization with 2-bromo-3'-methoxyacetophenone to be subsequently quantified by HPLC-MS [19]. For the unambiguous structural elucidation of metabolites H1-H4, syntheses for both the stabilized and free thiol metabolites have been described in the literature with yields below 3% [20,21]. Synthesis of active metabolites using human liver microsomes is hampered by low yield as well and by complicated purification protocols, and hence is not suitable for use on a preparative scale [18]. The active metabolite of prasugrel (PAM, R-138727) found in human plasma samples consists of four stereoisomers, (R, S)-, (R, R)-, (S, S)and (S, R)-isomers (the first letter stands for the configuration of the thiol group and the second for that at the benzylic position), with 84% found as (R, S)and (R, R)-, representing the most potent isomers. This underlines the importance of the R-configuration of the thiol group for bioactivity [22][23][24]. A chemical synthesis of PAM (R-138727) has not been reported so far. CYP2B6 and CYP2C9 may be involved in the oxidative cascade [5]. Several studies revealed that polymorphism in the gene encoding CYP2C19 contributes to the variability of platelet response to clopidogrel in patients. Individuals who carry a reduced-function CYP2C19 allele had significantly lower levels of the active metabolite of clopidogrel, diminished platelet inhibition and a higher rate of major adverse cardiovascular events [13][14][15]. In addition, clopidogrel undergoes an extensive first-pass metabolism in the human liver where most of the prodrug is hydrolyzed to inactive clopidogrel carboxylic acid by carboxylesterase hCE1, followed by glucuronidation and renal excretion [16]. To address these causes for the low plasma level of active metabolites, prasugrel was developed replacing the ester functionality with a metabolically stable ketone and the introduction of an ester group in the thiophene 2-position that switched the first-step activation from CYP2C19 to carboxylesterase hCE2 [17].
In vitro metabolism studies showed that metabolic activation of clopidogrel results in a mixture of four stereoisomers (H1-H4, Figure 1). The H1 and H2 isomers are trans isomers, whereas the exocyclic double bond between C3 and C16 in H3 and H4 appears in cis configuration. Analysis of clinical samples demonstrated that only H3 and H4 are formed in vivo, with only H4 being active. Therefore, the configuration of the thiol group at C4 seems to be crucial for antiplatelet activity [18]. The reactive free thiol group of the active metabolites leads to high instability in plasma samples, which is why they were stabilized by derivatization with 2-bromo-3'-methoxyacetophenone to be subsequently quantified by HPLC-MS [19]. For the unambiguous structural elucidation of metabolites H1-H4, syntheses for both the stabilized and free thiol metabolites have been described in the literature with yields below 3% [20,21]. Synthesis of active metabolites using human liver microsomes is hampered by low yield as well and by complicated purification protocols, and hence is not suitable for use on a preparative scale [18]. The active metabolite of prasugrel (PAM, R-138727) found in human plasma samples consists of four stereoisomers, (R, S)-, (R, R)-, (S, S)-and (S, R)-isomers (the first letter stands for the configuration of the thiol group and the second for that at the benzylic position), with 84% found as (R, S)-and (R, R)-, representing the most potent isomers. This underlines the importance of the R-configuration of the thiol group for bioactivity [22][23][24]. A chemical synthesis of PAM (R-138727) has not been reported so far. Herein, we report biocatalytic syntheses of the bioactive metabolites of clopidogrel and prasugrel by unspecific peroxygenases (UPOs, EC 1.11.2.1) secreted by fungi. UPOs have arisen to 'dream catalysts' for oxyfunctionalization reactions, since they incorporate selectively oxygen into nonactivated hydrocarbons generating hydroxylations and epoxidations that are difficult to attain by chemical methods [25][26][27]. These extracellular Herein, we report biocatalytic syntheses of the bioactive metabolites of clopidogrel and prasugrel by unspecific peroxygenases (UPOs, EC 1.11.2.1) secreted by fungi. UPOs have arisen to 'dream catalysts' for oxyfunctionalization reactions, since they incorporate selectively oxygen into nonactivated hydrocarbons generating hydroxylations and epoxidations that are difficult to attain by chemical methods [25][26][27]. These extracellular glycosylated heme-thiolate proteins are activated by hydrogen peroxide and work independently of electron donors, transport proteins and additional cofactors. Their catalytic cycle combines characteristics of heme peroxidases and cytochrome P450s, which is why UPOs and P450 enzymes have partly overlapping reaction portfolios [28,29]. UPOs have already been used for the conversion of various pharmaceuticals, e.g., volixibat (N-dealkylation), cyclophosphamide (aliphatic hydroxylation), propranolol (aromatic hydroxylation), testosterone (epoxidation) and corticosteroids (C-C fission) [30][31][32][33][34]. In this context, it was found that, depending on the catalyzed reactions, the most relevant parameters in respect to reaction optimization were pH, H 2 O 2 concentration and dosing rate as well as organic solvents and their content. The effective generation of authentic samples of drug metabolites is important for their structural confirmation and for LC-MS recovery, for evaluation of their potential safety risk, for investigation of drug-drug interactions, for pharmacological testing and for detailed pharmacokinetic and pharmacodynamic analysis [26,30,31].
We demonstrate here the selective oxidation of thienopyridines and the subsequent ring opening of intermediary formed thiolactones into the bioactive 4-mercapto-3piperidinylidene acetic acid derivatives, CAM and PAM, by UPOs.
Porcine liver esterase (PLE) was purchased from Merck Chemicals (Darmstadt, Germany) with a specific activity of 18 U mg −1 , where one unit represents the hydrolysis of 1 µmol ethyl butyrate in 1 min at pH 8.
All enzyme samples were diluted or dissolved in 10 mM KP i buffer (pH 7 or 5.5) to obtain a volume activity of 20 U mL −1 before usage in small-scale reactions (500 µL).

Enzymatic Reactions
Typical UPO reactions (500 µL final volume) contained 1 mM of the respective thienopyridine, 10% acetone, 20 mM KP i buffer (pH 7 or 5.5), 5 mM ascorbate and 2 U mL −1 UPO. The reaction mixture was tempered to 25 • C and the reaction kept running by adding aqueous H 2 O 2 (0.25 µmol) in pulses every 15 min over a period of 2 h to a final concentration of 4 mM. The reaction mixtures with prasugrel (PSG) were preincubated with 2 U mL −1 PLE for 30 min before ascorbate, UPO and H 2 O 2 were added. The influence of the organic cosolvent was investigated with acetone in the concentration range 0-40% (v/v) and by substitution of 10% acetone with acetonitrile (MeCN), methanol (MeOH), dimethyl sulfoxide (DMSO) or dimethyl formamide (DMF). As a reductant, glutathione (5 mM) was tested as an alternative to ascorbate, as well as the absence of such. All reaction setups were shaken at 700 rpm on a thermo shaker (TurboShaker 3500, Scienova GmbH, Jena, Germany) and stopped by adding 500 µL of ice-cold acetonitrile (−20 • C). Subsequently, the samples were centrifuged at 14,000× g for 10 min and the supernatants were analyzed by HPLC and MS. All reaction mixtures were performed in duplicate. Control experiments for every setup were carried out without UPO or without H 2 O 2 .

Semi-Preparative Synthesis of Thienopyridine Metabolites
Clopidogrel bisulfate (CPG, 84 mg, 0.2 mmol) was dissolved in 40 mL acetone, 140 mL deionized water and 20 mL KP i buffer (200 mM, pH 7) in a round-bottom flask. To this suspension, 350 mg ascorbate and 600 U MroUPO (50 mg) were added, and the mixture was stirred at 25 • C and 200 rpm on a magnetic stirrer. Using a syringe pump, a 400 mM hydrogen peroxide solution was continuously added at a rate of 1 mL h −1 (equivalent to 2 mM h −1 ). Every 15 min, a 200 µL sample was taken, mixed with 200 µL of acetonitrile (−20 • C), centrifuged and analyzed by HPLC to monitor the reaction progress. Another 200 µL sample was taken to measure residual enzyme activity. The reaction was stopped after two hours by extracting the aqueous mixture with dichloromethane, and the organic layer was dried on anhydrous sodium sulfate and filtered. The organic solvent was removed under reduced pressure, the crude product was recovered as brownish oil (approx. 45 mg) and purified by preparative liquid chromatography. The clopidogrel active metabolite (CAM) was isolated as a mixture of isomers (molar ratio 42:58) in 13% yield (9 mg, 0.025 mmol) with a purity above 95% ( 13  Prasugrel (PSG, 150 mg, 0.4 mmol) was dissolved in 40 mL acetone, 140 mL deionized water and 20 mL KP i buffer (200 mM, pH 7). Then, 22 mg PLE was added, and the suspension was stirred at 25 • C and 200 rpm. Thin-layer chromatography (TLC, ethyl acetate/n-hexane 1:1) indicated a complete conversion after 60 min. The reaction mixture was extracted with dichloromethane, and the organic phase was dried with sodium sulfate. Evaporation in vacuo gave 135 mg of a yellow pale oil that was further purified by preparative liquid chromatography; yield: 110 mg of isomers (molar ratio 53:47) of 2-oxo-PSG (83%, >98% purity  Hydrogen peroxide (400 mM) was continuously added via a syringe pump with a flow rate of 1 mL h −1 (corresponding to 2 mM h −1 ), and the reaction mixture was stirred for an additional 2.5 h. Samples were taken every 15 min as described above. Then the mixture was extracted three times with dichloromethane. The combined organic phases were dried with sodium sulfate, filtered and evaporated to dryness under vacuum resulting in 54 mg of brownish resin. The crude product was purified by preparative liquid chromatography to give 17 mg of a bright yellow mixture of PAM isomers (24% yield, >95% purity, molar  13
MS and MS 2 spectra were obtained using a Thermo Scientific Q Exactive Plus Quadrupole-Orbitrap mass spectrometer (Thermo Electron, Waltham, MA, USA) coupled with a heated electrospray ionization source in positive mode. The tune operating parameters were as follows: the rate of sheath gas flow and auxiliary gas flow were 60 and 15 (arbitrary unit), respectively; spray voltage, 4.0 kV; the temperatures of capillary and auxiliary gas heater were 320 and 400 • C, respectively; high-resolution MS was operated at full scan mode with a mass range of m/z 150-1500 at a resolution of 70,000 (m/z 200). The MS 2 data at a resolution of 35,000 were obtained by parallel reaction monitoring mode triggered by inclusion ions list, which was built by the molecule predicted. The collision energy varied between CE10 and CE50 and is indicated with the respective results.

NMR Studies
1 H-and 13 C NMR spectra (500 and 126 MHz, respectively) were recorded on a Bruker Advance neo 500 spectrometer (Billerica, MA, USA) in the solvents indicated. Chemical shifts (δ) are given as ppm relative to (CH 3 ) 4 Si.

Results
In order to elucidate the feasibility of the oxidation of antithrombotic thienopyridines with UPOs, the studies were first limited to clopidogrel as a substrate because of the twostep oxidation cascade. Incubation of clopidogrel with several UPOs was performed in the presence of an organic co-solvent in a phosphate-buffered system. The oxidant hydrogen peroxide was added intermittently over the reaction period. Ascorbate was added to prevent coupling reactions, which may occur due to the radical-forming peroxidase activity of UPOs [29]. Five homologously produced UPOs (wild-type enzymes) were tested: AaeUPO, CraUPO, MroUPO, CglUPO and MweUPO (entry 1-7, Table 1). With the two UPOs of the genus Marasmius (MroUPO, MweUPO), the reaction was additionally carried out at pH 5.5 (entry 4 and 7, Table 1), as this pH corresponds to their reaction optimum [34,37]. HPLC-MS studies of the incubated samples revealed the incipient formation of two pairs of metabolites, each at the same ratio and, to some extent, of partially oxygenated byproducts. The dominant pair was identified as 2-oxo-clopidogrel, as could be deduced from the identical MS 2 spectra that were obtained for the parent ion (m/z = 338.0612, C 16 H 17 ClNO 3 S + ) with the associated fragments (m/z 183.0208 (100%) and m/z 155.0258 (87%), CE20) when using the appropriate reference standard. The second metabolite pair appears to be a stable dimer of oxygenated clopidogrel, which was not further converted by UPOs (m/z 675.1150, C 32 H 33 Cl 2 N 2 O 6 S 2 + ). Experiments with an increased concentration of radical scavengers did not lead to decreased dimer formation (data not shown). The ratio of 2-oxo-CPG and oxo-CPG-dimer was approx. 2:1 in the case of MroUPO and MweUPO. In contrast, the ratio obtained with AaeUPO was about 1:1 and with CraUPO and CglUPO even 1:3. A third less dominant pair of metabolites exhibited molecular masses of m/z 356/358 in the isotopic ratio typical for the presence of a chlorine atom. Fragmentation of the parent ion (m/z 356.0712 ( 35 Cl), C 16 H 19 ClNO 4 S + , CE20) gave clear indication that these are two diastereomers of CAM ( Figure 2). The highest yields under the conditions described above were obtained with the UPOs from Marasmius spp., i.e., MroUPO (8.6%) and MweUPO (6.4%) at pH 7.0 (entries 3 and 6, Table 1). A representative chromatogram with the three metabolite pairs and the MS and MS 2 spectra of the dimer are given in the Supplementary Material (Figures S1 and S2).
C16H17ClNO3S + ) with the associated fragments (m/z 183.0208 (100%) and m/z 155.0258 (87%), CE20) when using the appropriate reference standard. The second metabolite pair appears to be a stable dimer of oxygenated clopidogrel, which was not further converted by UPOs (m/z 675.1150, C32H33Cl2N2O6S2 + ). Experiments with an increased concentration of radical scavengers did not lead to decreased dimer formation (data not shown). The ratio of 2-oxo-CPG and oxo-CPG-dimer was approx. 2:1 in the case of MroUPO and MweUPO. In contrast, the ratio obtained with AaeUPO was about 1:1 and with CraUPO and CglUPO even 1:3. A third less dominant pair of metabolites exhibited molecular masses of m/z 356/358 in the isotopic ratio typical for the presence of a chlorine atom. Fragmentation of the parent ion (m/z 356.0712 ( 35 Cl), C16H19ClNO4S + , CE20) gave clear indication that these are two diastereomers of CAM ( Figure 2). The highest yields under the conditions described above were obtained with the UPOs from Marasmius spp., i.e., MroUPO (8.6%) and MweUPO (6.4%) at pH 7.0 (entries 3 and 6, Table 1). A representative chromatogram with the three metabolite pairs and the MS and MS 2 spectra of the dimer are given in the Supplementary Material (Figures S1 and S2).  The conversion of prasugrel (PSG) requires hydrolysis of the acetic acid ester in the first step. In order to carry out the entire reaction sequence in one reaction batch, hydrolysis with porcine liver esterase (PLE) was investigated. For this purpose, 1 mM substrate was dissolved in 10% acetone in a phosphate-buffered system (pH 7.0) and incubated with 2 U mL −1 PLE. PSG was completely converted within 30 min to 2-oxo-prasugrel, which was also identified as a double peak (ratio 1:1) of two diastereomers (m/z 332.1115, C 18 H 19 FNO 2 S + ). Subsequently, ascorbate, UPO and successively hydrogen peroxide was added over a period of 2 h. HPLC-MS analyses of incubated samples revealed that MroUPO (34%, entry 10) and MweUPO (31%, entry 12) particularly catalyzed the formation of a pair of metabolites, each with m/z 350 (pair of diastereomers in ratio 1:1). Again, MS (m/z 350.1219, C 18 H 21 FNO 3 S + ) and MS 2 data (from m/z 350.1219, CE25) gave clear hints for the formation of the active prasugrel metabolite PAM (Figure 3). Incubation with CglUPO also resulted in a high yield of PAM, but numerous byproducts emerged (entry 11, Table 1). the formation of a pair of metabolites, each with m/z 350 (pair of diastereomers in ratio 1:1). Again, MS (m/z 350.1219, C18H21FNO3S + ) and MS 2 data (from m/z 350.1219, CE25) gave clear hints for the formation of the active prasugrel metabolite PAM (Figure 3). Incubation with CglUPO also resulted in a high yield of PAM, but numerous byproducts emerged (entry 11, Table 1).
The diastereomeric pairs are likely due to the configuration of the thiol group at C4 or the cis/trans configuration at C3. To obtain precise insights into the chemical nature of the products, CAM and PAM (presumably identical to human metabolites), NMR studies were performed. For this, the reactions had to be optimized and upscaled in order to isolate the metabolites.  The diastereomeric pairs are likely due to the configuration of the thiol group at C4 or the cis/trans configuration at C3. To obtain precise insights into the chemical nature of the products, CAM and PAM (presumably identical to human metabolites), NMR studies were performed. For this, the reactions had to be optimized and upscaled in order to isolate the metabolites.
With the aim of optimizing the reaction conditions and increasing the yield of active metabolites, especially in the case of clopidogrel, both reductant and organic solvents as well as their concentration were varied in the approach with MroUPO (Table 2). It turned out that the presence of a reductant was particularly necessary. The absence of ascorbate or glutathione resulted in a variety of oxidation and coupling products, with only small amounts of the desired compounds being detected with LC-MS. When glutathione replaced ascorbate, the formation of CAM was significantly reduced (entry 14, Table 2), and a mixture of four isomers with the appropriate molecular mass of m/z 356/358 was identified in traces. Replacing acetone with other water-miscible solvents such as acetonitrile, methanol, dimethyl sulfoxide or dimethyl formamide did not increase the yields of the active metabolite. Interestingly, the reaction with methanol as co-solvent gave a good yield of 2-oxo-CPG (31%) but almost no formation of the active metabolite. Increasing the amount of acetone to 20% increased the CAM yield to about 10%, which was the highest yield that could be achieved (entry 21, Table 2). Further extensive attempts such as exchanging buffer systems, varying the substrate-catalyst ratio or the amount and dosing rate of the co-substrate hydrogen peroxide as well as the use of other hydroperoxides or hydrogen peroxide producing systems (e.g., glucose/glucose oxidase, carbamide peroxide) did not significantly improve the yields (data not shown). The optimal reaction conditions (entry 21, Table 2) were chosen to perform the syntheses of CAM and PAM at a 100 mg scale. The hydrogen peroxide feed was changed from intermittent addition to continuous feed by a syringe pump to ensure a permanent low hydrogen peroxide concentration and thus an optimal UPO performance. During the reaction, the concentrations of substrate, 2-oxo metabolite and active metabolite as well as the residual activity of the catalyst were monitored. Clopidogrel was almost completely converted within one hour, with the formation of 2-oxo-CPG predominantly at the beginning and, to a somewhat minor extent, the previously mentioned dimerization product. After about 45 min, the concentration of 2-oxo-CPG reached a maximum of about 500 µM and declined afterwards as the thiolactone was further converted mostly to the active metabolite ( Figure 4A). Here, an isomer of 2-oxo-CPG was preferentially transformed by MroUPO ( Figure S1B). After 2 h, the residual activity of MroUPO was below 5%, and 2-oxo-CPG was almost completely converted. The final concentration of active metabolite (pair of stereoisomers) was about 250 µM, and both isomers were present in equal amounts. The metabolite pair was successfully isolated and purified by preparative HPLC. NMR spectra were difficult to interpret because of the presence of isomers. Therefore, spin-spin coupling constants could not be determined in some cases due to the overlapping of the signals of both compounds. The ratio of isomers in the isolate was fortunately about 2:3 so that the individual signals could be assigned to the metabolites with the use of the correlation spectra (COSY, HMBC, HSQC). The singlets of the 1 H NMR spectra at 5.56 ppm (60%) and 5.45 (40%) were allocated to H-16 and confirmed the cis configuration at C3-C16 (CAM-H3 and CAM-H4), which were in agreement with those in the literature [20]. The metabolite, which was initially formed and preferably by MroUPO, eluted earlier from the HPLC column and was identified as CAM-H3. In principle, the isolation of 2-oxo-CPG would also be possible by a kinetically controlled approach. However, the isolation of the metabolites will be challenging due to the presence of the oxo-CPG dimer.
preparative HPLC. NMR spectra were difficult to interpret because of the presence of isomers. Therefore, spin-spin coupling constants could not be determined in some cases due to the overlapping of the signals of both compounds. The ratio of isomers in the isolate was fortunately about 2:3 so that the individual signals could be assigned to the metabolites with the use of the correlation spectra (COSY, HMBC, HSQC). The singlets of the 1 H NMR spectra at 5.56 ppm (60%) and 5.45 (40%) were allocated to H-16 and confirmed the cis configuration at C3-C16 (CAM-H3 and CAM-H4), which were in agreement with those in the literature [20]. The metabolite, which was initially formed and preferably by MroUPO, eluted earlier from the HPLC column and was identified as CAM-H3. In principle, the isolation of 2-oxo-CPG would also be possible by a kinetically controlled approach. However, the isolation of the metabolites will be challenging due to the presence of the oxo-CPG dimer.  Incubation of 150 mg prasugrel with PLE at 25 • C and pH 7 resulted in complete conversion after 60 min as verified by TLC. MS and MS 2 data confirmed the exclusive formation of 2-oxo-clopidogrel. For structural confirmation, the metabolite pair was isolated by preparative HPLC. In a second batch, 75 mg prasugrel was incubated analogously, with a complete turnover after 30 min as analyzed by TLC and HPLC. After subsequent addition of MroUPO, reductant and feeding of hydrogen peroxide, 2-oxoprasugrel was continuously converted in an almost linear way to the stereoisomer pair PAM ( Figure 4B). In this case, no preferential conversion of an isomer by UPO was observed. After 180 min total reaction time, the yield of PAM was about 44%. NMR analyses confirmed the formation of the active metabolites in cis configuration analogously to CAM. HPLC-ELSD elution profiles, 1 H and 13 C NMR spectra as well as two-dimensional correlation spectra (COSY, HMBC, HSQC) of isolated CAM, PAM and 2-oxo-PSG as well as the assignment of the signals are given in Supplementary Material.

Discussion
The aforementioned results show that UPOs can be used for efficient syntheses of active thienopyridine metabolites. The crucial step in the case of clopidogrel seems to be the formation of the 2-oxo metabolite. Analysis of the reaction kinetics revealed that a pair of isomers was formed in parallel with 2-oxo-CPG, which may be a dimer of oxygenated CPG based on the molar mass. Such dimer formation has not been described in the metabolism of CPG by P450s. However, in the reaction described here, these dimers represent about 30% of the products formed (using MroUPO) and, thus, significantly reduce the final yield of CAM. Since the focus of this study was on the syntheses of CAM and PAM, the dimer mixture was not isolated and further analyzed by NMR. Nevertheless, the formation of a dimer of CPG S-oxide can be assumed. The oxidation of aromatic rings by both P450s and UPOs is thought to proceed not via H-atom abstraction (as in the case of the hydroxylation of sp 3 -carbons), but via the formation of arene oxides (epoxides), which subsequently rearrange under an NIH shift to phenols [38][39][40]. Similar arene oxide formation has been described for the oxygenation of thiophenes resulting in thiophen-2-ones, e.g., in the case of CPG metabolism. As an alternative to arene oxide formation, UPO-catalyzed oxygenation may also occur at the heteroatoms to form the corresponding S-oxides [27,41]. Thiophene-S-oxides can undergo Diels-Alder reaction to form stable dimers [42][43][44]. Such a reaction type is also conceivable with clopidogrel and reflects the distribution between aromatic hydroxylation and S-oxidation in the first reaction step (Scheme 2). PAM ( Figure 4B). In this case, no preferential conversion of an isomer by UPO was observed. After 180 min total reaction time, the yield of PAM was about 44%. NMR analyses confirmed the formation of the active metabolites in cis configuration analogously to CAM. HPLC-ELSD elution profiles, 1 H and 13 C NMR spectra as well as two-dimensional correlation spectra (COSY, HMBC, HSQC) of isolated CAM, PAM and 2-oxo-PSG as well as the assignment of the signals are given in Supplementary Material.

Discussion
The aforementioned results show that UPOs can be used for efficient syntheses of active thienopyridine metabolites. The crucial step in the case of clopidogrel seems to be the formation of the 2-oxo metabolite. Analysis of the reaction kinetics revealed that a pair of isomers was formed in parallel with 2-oxo-CPG, which may be a dimer of oxygenated CPG based on the molar mass. Such dimer formation has not been described in the metabolism of CPG by P450s. However, in the reaction described here, these dimers represent about 30% of the products formed (using MroUPO) and, thus, significantly reduce the final yield of CAM. Since the focus of this study was on the syntheses of CAM and PAM, the dimer mixture was not isolated and further analyzed by NMR. Nevertheless, the formation of a dimer of CPG S-oxide can be assumed. The oxidation of aromatic rings by both P450s and UPOs is thought to proceed not via H-atom abstraction (as in the case of the hydroxylation of sp 3 -carbons), but via the formation of arene oxides (epoxides), which subsequently rearrange under an NIH shift to phenols [38][39][40]. Similar arene oxide formation has been described for the oxygenation of thiophenes resulting in thiophen-2-ones, e.g., in the case of CPG metabolism. As an alternative to arene oxide formation, UPO-catalyzed oxygenation may also occur at the heteroatoms to form the corresponding S-oxides [27,41]. Thiophene-S-oxides can undergo Diels-Alder reaction to form stable dimers [42][43][44]. Such a reaction type is also conceivable with clopidogrel and reflects the distribution between aromatic hydroxylation and S-oxidation in the first reaction step (Scheme 2). Scheme 2. Proposed mechanism for product formation in the UPO-catalyzed oxygenation of clopidogrel. Scheme 2. Proposed mechanism for product formation in the UPO-catalyzed oxygenation of clopidogrel.
This also provides a plausible explanation for the different distribution of 2-oxo-CPG and CPG S-oxide dimers when using different UPOs. A product shift to the 2oxo metabolite might be achieved by protein engineering. Such attempts have resulted in improved UPO variants ensuring higher yields, for example, in the synthesis of 5hydroxypropranolol and dextromethorphan [32,45]. In our context, it would be advisable to optimize UPOs by directed evolution both toward arene oxide formation and S-oxidation. In this way, a two-step reaction setup with optimal yields could be established. According to our results, AaeUPO would be a good candidate for arene oxide formation (epoxidation), since the distribution was about 1:1, and S-oxidation of 2-oxo-CPG was not observed. CglUPO, on the other hand, would be a good candidate for improving S-oxidation, which was shown in the reactions with 2-oxo-PSG. However, the best turnover in the two-step reaction with the wild-type enzymes was obtained with MroUPO in this study. The continuous loss of activity of MroUPO that was observed during the reaction is a known phenomenon for some heme-containing enzymes and might be caused by heme bleaching in consequence of increasing peroxide concentration [46].
UPOs' ability to oxygenate heteroatoms is particularly evident in the second reaction step. The S-oxidation desired constitutes the 'triggering sequence' for ring opening to form the active metabolite. Over the years, Dansette et al. have conducted extensive studies on the metabolism of clopidogrel, partly adapted it to prasugrel and elucidated the intermediates of the complex mechanism as well as of influences on the position of the double bonds [7][8][9][10]12,47]. The effects of ascorbate and glutathione observed in experiments with human liver microsomes (HLMs) were the same as with UPOs in this study. In the presence of ascorbate, only the cis thiol metabolites were formed, whereas cis and trans thiol metabolites emerged in the presence of glutathione as reductant [47]. Given the very similar behavior of clopidogrel and prasugrel in the UPO reactions described above, and in comparison to the P450 reactions in HLMs reported in the literature, this example demonstrates again that UPOs can make a crucial contribution to the elucidation of human biotransformation processes. The reaction setups are much easier to perform, less expensive and linearly scalable. Of course, they cannot replace studies with human or animal cell material, but for the synthesis of metabolites as reference material or to support structural elucidation, UPOs can be a useful, primarily time-saving tool. It is worth mentioning that only a handful of UPO proteins are available to date, but thousands of putative UPO protein sequences have been identified in genomes widely distributed in the fungal kingdom, that bear an infinite pool of catalytically diverse oxyfunctionalizing biocatalysts [28].
The large-scale application of peroxygenases still needs to overcome a number of bottlenecks, especially regarding their availability. The heterologous production of UPOs has proven to be difficult and has so far only been (really) successful with satisfactory titers for rAaeUPO expressed in Saccharomyces cerevisiae and Pichia pastoris [48]. This expression system has also been used to create numerous interesting mutants [32,45,49,50]. Successful expression of some UPOs was also achieved by NOVOZYMES A/S using Aspergillus oryzae as the expression host. These enzymes (rCciUPO, rHinUPO) were tested for several applications, but the exact scales and titers are unknown [33,[51][52][53]. Recently, the expression of UPOs in Escherichia coli has been successful at small scale, which allowed the generation of purposefully optimized mutants [54][55][56]. However, large-scale UPO production is still limited to rAaeUPO, so far [57].
Reaction technology challenges such as optimal peroxide delivery (where sensorcoupled pump systems will be the most elegant solution), stability in the presence of organic solvents and enzyme reusability (immobilization) have also been addressed in several studies but solutions for large-scale syntheses are still missing [27]. Nevertheless, the findings of the last two decades, especially the versatility in the modification of bio-based raw materials (e.g., fats, terpenes, steroids) [58], conversion of environmental pollutants (e.g., PAHs) [59] and synthesis of drug metabolites [26] have shown that the way is open for the application of UPOs in the small-scale chemical segment, such as syntheses of specialty chemicals, sensor technology or for the elucidation of metabolite structures in the pharmaceutical sector.

Conclusions
We have demonstrated a biocatalytic method for the syntheses of the active metabolites CAM (25% yield) and PAM (44% yield) starting from the parental drugs clopidogrel and prasugrel by using fungal unspecific peroxygenases (UPOs). Both metabolites were isolated and their structure elucidated by mass spectrometry and NMR spectroscopy. This provides easy access to these compounds to be used, e.g., in pharmacological assays or as reference material. Moreover, this UPO-based reaction sequence, starting with the formation of a thiolactone and followed by the cleavage of the thieno ring via a thiolactone sulfoxide intermediate, has been described for the first time and can be adapted to other chemical problems. In general, UPOs could contribute to the elucidation of biotransformation mechanisms catalyzed by hepatic cytochrome P450 enzymes, as there are numerous examples of mimicking intrinsic P450 activity. On the other hand, UPOs can become valuable catalysts for the synthesis of active metabolites of prodrugs for those cases where polymorphism or other deficiencies do not allow a targeted in vivo transformation and personalized therapeutic solutions are required.