Overcoming co-product inhibition in the nicotinamide independent asymmetric bioreduction of activated C=C-bonds using flavin-dependent ene-reductases

Eleven flavoproteins from the old yellow enzyme family were found to catalyze the disproportionation (“dismutation”) of conjugated enones. Incomplete conversions, which were attributed to enzyme inhibition by the co-product phenol could be circumvented via in situ co-product removal by scavenging the phenol using the polymeric adsorbent MP-carbonate. The optimized system allowed to reduce an alkene activated by ester groups in a “coupled-substrate” approach via nicotinamide-free hydrogen transfer with >90% conversion and complete stereoselectivity.


Introduction
Ene-reductases from the "old yellow enzyme" family (OYE), which catalyze the asymmetric trans-reduction of alkenes bearing an electron-withdrawing activating group (Stuermer et al., 2007;Toogood et al., 2010;Winkler et al., 2012) became important biocatalysts over the last few years. In the classic approach, the reduced flavin is recycled via a nicotinamide cofactor at the expense of a sacrificial hydrogen-donor cosubstrate, such as glucose, glucose-6-phosphate, formate, 2-propanol, or phosphite (Hollmann et al., 2010). Overall, this so-called "enzyme-coupled" process depends on two enzymes and two cofactors (Faber, 2011;Matsuda et al., 2009;Tauber et al., 2011;Wandrey, 2004). Several attempts were made to reduce the complexity of these systems by cancelling NAD(P)H and its recycling from the system. Direct reduction of the active site flavin was accomplished by an additional flavin-catalyst which in turn was regenerated in a lightmediated reaction by an auxiliary substrate (Grau et al., 2009;Taglieber et al., 2008). Only recently, ene-reductases were successfully employed with molar equivalents of synthetic nicotinamide mimics instead of "natural" NAD(P)H (Paul et al., 2013). We have recently proposed a nicotinamideindependent recycling system for reduced flavins based on the disproportionation (dismutation) of enones , which has been observed as catalytic promiscuity of OYEs ( Fig. 1) (Buckman and Miller, 1998;Karplus et al., 1995;Vaz et al., 1995). During this reaction, an equivalent of [2H] is transferred by a single flavoprotein between two enone substrates (1a) yielding an oxidized (1d) and reduced product (1b) in equimolar amounts. The reductive half-reaction proceeds via the desaturation of enone 1a (Vaz et al., 1995) forming FMNH 2 and cyclohexa-1,4-dienone, which irreversibly tautomerises to phenol (1d), thereby providing a strong driving force for the overall process. The reduced flavin subsequently reduces the second equivalent of enone 1a, which resembles the oxidative halfreaction, and closes the catalytic cycle.
The crosswise diproportionation between two identical enone substrates (1a) could be turned into a directed hydrogen-transfer system by combining two different enone substrates, each serving as distinct hydrogen donor and acceptor, respectively (Fig. 2). Although the proof of principle was shown, the system was practically not applicable due to incomplete conversions (max. 65%). The latter were attributed to inhibition exerted by the co-product phenol.
Electron-rich phenols act as strong inhibitors of OYEs through formation of stable charge-transfer complexes with the electron-deficient flavin in the active site (Abramovitz and Massey, 1976a, b;Buckman and Miller, 1998;Matthews et al., 1975;Spiegelhauer et al., 2010;Stewart and Massey, 1985;Strassner et al., 1999). Prompted by the fact, that complex formation is reversible (Buckman and Miller, 1998), we aimed to overcome inhibition by reaction optimization (pH and temperature) and co-product scavenging using a solid-phase organic resin.

Materials and Methods
General GC-FID analyses were carried out on a Varian 3800 using H 2 as carrier gas (14.5 psi). HPLC analyses were performed by using a Shimadzu system equipped with a Chiracel OD-H column (25 Â 0.46 cm).
General Procedure B for Anaerobic Enzymatic Disproportionation of Cyclohex-2-Enone (1a) An aliquot of isolated enzyme (OYE1, OYE2, CrS, EBP1, NCR, XenA, and YqjM; protein purity >90%, protein content in reaction 100 mg/mL) was added to a screw-top glass vial (2 mL) containing a degassed buffer solution (0.8 mL, 50 mM, Tris-HCl buffer; pH 7.5), cyclohex-2-enone (1a, 20 mM) and (optionally) MP-carbonate (up to 100 mg, 40 eq. loading capacity). The vial was flushed with argon and sealed using a screw cap lined with a teflon septum. The mixture was shaken for 24 h at 30 C and 120 rpm using an Infors Unitron shaker and products were extracted with ethyl acetate (0.7 mL). The organic phase was dried over Na 2 SO 4 and analyzed on GC to determine the conversion. For every test, a control was performed in the absence of enzyme.

Optimization of Reaction Conditions: Buffer-Type and pH, Reaction Time and Temperature
The optimization of reaction conditions was carried out by individual variation of every single parameter of general procedure A. For the optimization of the reaction temperature, the microcentrifuge tubes were shaken at 500 rpm in an Eppendorf thermomixer.

General Procedure C for Anaerobic NAD(P)H-Independent Asymmetric Bioreduction of Activated Alkenes
An aliquot of isolated enzyme (OYE1-2, CrS, EBP1, NCR, XenA, YqjM, NerA, and GkOYE; protein purity >90%, protein content in reaction 100 mg/mL) was added to a screwtop glass vial (2 mL) containing a degassed buffer solution (0.8 mL, 50 mM, Tris-HCl buffer; pH 7.5), the substrate (2a, 3a or 4a, 10 mM), the H-donor (5c or 6c; 10 mM) and (optionally) MP-carbonate (up to 100 mg, 40 eq. loading capacity). The vial was flushed with argon and sealed using a screw cap lined with a teflon septum. The mixture was shaken for 24 h at 30 C and 120 rpm using an Infors Unitron shaker and products were extracted with ethyl acetate (2 Â 0.7 mL). The combined organic phase was dried over Na 2 SO 4 and analyzed on GC to determine the conversion and stereoselectivity. For every test, a control was performed in the absence of enzyme. For the determination of conversion a calibration curve was established for a range of substrate/Hdonor ratios in presence of MP-carbonate to compensate for the different adsorption of substrate and H-donor onto the carrier.

Results and Discussion
The disproportionation-activity of a series of 22 enereductases was evaluated in a screening using cyclohex-2enone (1a) as substrate (Fig. 1). During the course of these tests under standard conditions (pH 7.5, aerobic), the list of previously reported candidate enzymes-OYE1 from S. pastorianus, OYE2 and OYE3 from S. cerevisiae, NCR from Z. mobilis, EBP1 from C. albicans and YqjM from B. subtiliscould be considerably expanded by several OYE-homologs, such as NerA from A. radiobacter (Durchschein et al., 2010), and the thermostable OYE-variants CrS from Thermus scotoductus SA-01, (Opperman et al., 2008) and GkOYE from G. kaustophilus DSM 7263 (Schittmayer et al., 2010), which were recently discovered (Table I). Most remarkably, CrS from T. scotoductus SA-01 was highly active showing 55% conversion. The high dismutase-activity of OYE1, OYE2, and EBP1 was confirmed by conversions of up to 61% (Buckman and Miller, 1998;Stueckler et al., 2010;Vaz et al., 1995). Modest conversions were found using GkOYE and NerA (10%), all other tested enzymes showed only low activities (<7% conversion) ( Table I, column A). An attempt to correlate the disproportionation activity with sequence-data with emphasis on the residues involved in FMN-binding, thereby modulating its redox potential, and the catalytic residues responsible for substrate binding and H þ -donation did not reveal any apparent patterns (Table SI).
In an attempt to overcome incomplete conversions caused by co-product inhibition exerted by phenol (1d), the reaction conditions were optimized in terms of (i) the buffer type and its pH, (ii) the reaction temperature, and (iii) the presence of molecular oxygen.
For the pH-tuning, three different buffer systems (citrate, phosphate, and Tris-HCl) were tested, covering a pH range from 4 to 10 (Fig. 3). Since it was shown that the more electron-rich phenolate-anion dominated over the neutral phenol species in charge-transfer complex formation (Abramovitz and Massey, 1976a;Miller, 1998, 2000a, b), elevated pH values are expected to be unfavorable based on the estimated pKa of 7.3 for phenol (1d) within the active site of EBP1 (Buckman and Miller, 1998). However, this effect is compensated by destabilization of the charge-transfer complex by action of an acidic amino acid residue in the active site (Tyr206 in EBP1, pKa 9.4) (Buckman and Miller, 1998) acting as proton donor/acceptor on Ca, which is deprotonated under basic conditions, thereby repelling the phenolate species. Overall, the latter effect seemed to dominate because endpoint conversions were enhanced at pH 9 with all enzymes (Fig. 3).
Since this catalytic promiscuity is also supported by high pH, 2,3-epoxycyclohexanone was formed between 0% and 5% at pH 9 (Table I, column B). In order to suppress the undesired loss of reduction equivalents, anaerobic conditions were applied (Table I, columns C and D). As expected, the absence of O 2 completely eliminated the competing epoxidation.
Investigation of the disproportionation rate over a temperature range of 20-70 C revealed typical bell-shaped optima between 40 and 50 C for the mesophilic enzymes, whereas the thermophilic candidates, such as GkOYE and CrS showed the highest conversions at 60 and 50 C, respectively (see Supporting Information). Based on these parameters, all further experiments were performed in Tris-HCl buffer at pH 7.5 and pH 9 under anaerobic conditions at 30 C and 24 h.
A search for a suitable phenol-adsorbing polymeric material revealed macroporous polystyrene (MP-)carbonate as a suitable candidate (Lyon and Kercher, 2004;Selwood et al., 2001). The latter possesses positively charged triethylammonium-groups linked to an aromatic styrene moiety, which enables ionic binding of the phenolate anion supported by p-p stacking of both aromatic systems (Fig. 4). The disproportionation of cyclohexenone by OYE1, OYE2, XenA, and CrS was investigated in presence of varying amounts of MP-carbonate. For OYE2 and CrS, a 40-fold loading capacity of adsorbent gave best results by scavenging >90% of phenol, going in hand with considerably enhanced conversions and a near-quantitative value for CrS (c 97%) (Table II).
Based on the optimization of the disproportionation of cyclohex-2-enone (1a), we finally attempted to further increase the performance of CrS with substrates 2a and 3a via reaction engineering. Using the enzyme giving best conversions at given conditions as a starting point, we could push the bioreduction of 2a to full conversion by raising the enzyme amount, and/or temperature and extending the reaction time, albeit with racemisation of 3a (Table IV, entries 2-4). In contrast, larger enzyme amounts improved the conversion of 3a from 47% to 76% (entries 5-7) and by extending the reaction time, a conversion of 92% could be finally reached for (R)-3b with an e.e. of >99% (entry 8).

Conclusion
From a library of 22 flavin-dependent ene-reductases from the OYE family, 13 candidates were shown to possess strong activities in the NAD(P)H-independent disproportionation of conjugated enones. Limited conversions caused by enzyme inhibition by the co-product phenol forming a charge- Table III. Nicotinamide-independent asymmetric bioreduction of activated alkenes (method C).

Entry
Substrate Donor Enzyme pH MP-C (eq.) a Conversion (%) e.e. (%) transfer complex with the flavin cofactor in the active site could be successfully overcome via ISPR employing MPcarbonate as polymeric phenol-scavenger at elevated pH. Although stereochemically labile compounds, such as asubstituted ketones were incompatible due to racemization, chirally stable a-substituted esters could be obtained for the first time with quantitative conversion via a nicotinamideindependent hydrogen-transfer system.

Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site. Figure S1. Optimization of pH in citrate buffer (50 mM) for the disproportionation of cyclohex-2-enone (1a) according to method A. Figure S2. Optimization of pH in phosphate buffer (50 mM) for the disproportionation of cyclohex-2-enone (1a) according to method A. Figure S3. Optimization of pH in Tris-HCl buffer (50 mM) for the disproportionation of cyclohex-2-enone (1a) according to method A. Figure S4. pH-Dependent epoxide (1e) formation in the disproportionation of cyclohex-2-enone (1a) in Tris-HCl buffer according to method A. Figure S5. Temperature profile for the disproportionation of cyclohex-2-enone (1a) according to method A. Table SI. Additional data for NAD(P)H-independent bioreduction of alkenes 2a and 3a using H-donors 5c and 6c in presence of MP-carbonate (40 eq. loading capacity) according to method C. n.d., not determined; n.c., no conversion. Table SII. Sequence alignment of OYEs from the screening for disproportionation activity with cyclohex-2-enone (1a) ( Table I).