Exploiting the catalytic power of enzymes for oxy-and amino-functionalization reactions

Multi-enzymatic cascades exploiting engineered enzymes are a powerful tool for the tailor-made synthesis of complex molecules from simple inexpensive building blocks. In this work, we engineered the promiscuous enzyme 4oxalocrotonate tautomerase (4-OT) into an effective aldolase with 160-fold increased activity compared to 4-OT wild type. Subsequently, we applied the evolved 4-OT variant to perform an aldol condensation, followed by an epoxidation reaction catalyzed by a previously engineered 4-OT mutant, in a one-pot two-step cascade for the synthesis of enantioenriched epoxides (up to 98% ee) from biomass-derived starting materials. For three chosen substrates, the reaction was performed at mil-ligram scale with product yields up to 68% and remarkably high enantioselectivity. Furthermore, we developed a three-step enzymatic cascade involving an epoxide hydrolase for the production of chiral aromatic 1,2,3-prim , sec , sec -triols with high enantiopurity and good isolated yields. The reported one-pot, three-step cascade, with no intermediate isolation and being completely cofactor-less, provides an attractive route for the synthesis of chiral aromatic triols from biomass-based synthons.


Introduction
In recent years, biocatalytic cascades have gained a surge in interest in chemical synthesis due to their low environmental impact and their time and cost efficiency. [1,2]Artificial multi-enzymatic cascade reactions do not require isolation of intermediates and the whole process can be performed under environmentally benign conditions, such as the use of aqueous media and mild temperatures.In addition, akin to biosynthetic pathways, the overall thermodynamic parameters are favorable, thus the equilibrium is usually shifted towards the desired product.Another feature in common with biosynthetic pathways is the exquisite selectivity.Enzymes are generally highly selective catalysts with remarkable regio-, chemoand stereoselectivity.Therefore, unwanted side reactions are usually avoided, and the desired molecule is typically obtained in high enantiopurity.][6][7][8][9][10] The enzyme 4-oxalocrotonate tautomerase (4-OT) is a versatile and powerful catalyst.[18][19] Although previous engineering efforts to improve the aldolase activity of 4-OT provided more active variants, these variants presented some drawbacks.For example, the best 4-OT variant M45T/F50A (TA), which has greatly enhanced aldolase activity, displays significantly reduced substrate selectivity. [19]Therefore, in this work, we further explored 4-OT as an artifical aldolase with the aim to improve its activity and restore the substrate selectivity towards benzaldehyde.In addition to its promiscuous aldolase activity, it was shown recently that 4-OT exhibits catalytic activity for the enantioselective epoxidation of α,β-unsaturated aldehydes using different peroxides, H2O2 or tBuOOH, making 4 OT a cofactor-independent peroxyzyme. [20]The promiscuous peroxygenase-like activity of 4 OT has been strongly improved by the directed evolution of a tandem-fused variant of 4 OT, [21] increasing the epoxidation activity and enabling the gram-scale synthesis of α,β-epoxy aldehydes with excellent enantiopurity (up to 98% ee) applying H2O2 as oxidant. [22]n this study, the promiscuous character of 4-OT, functioning both as a nonnatural aldolase and a peroxyzyme, was exploited in a one-pot, two-step enzymatic cascade for the production of enantioenriched epoxides as valuable synthons for the preparation of more complex molecules.Furthermore, we coupled the twostep cascade with a third step involving a specifically chosen epoxide hydrolase, enabling the synthesis of chiral aromatic 1,2,3-prim,sec,sec-triols, otherwise named aryl-glycerols (AGs). [23]The structural motif of 1,2,3-prim,sec,sec-triols, with one primary and two secondary alcohol groups and four possible stereoisomers, is very common in nature.Moreover, 1,2,3-prim,sec,sec-triol derivatives are used as building blocks for the production of several pharmaceuticals and biologically active molecules, [24][25][26][27][28] as it was recently shown for the microtubule-stabilizing agent zampanolide. [29]While several examples of the chemical synthesis of AGs are present in literature, [24,27,28] the multi-step enzymatic cascade developed in this work represents an environmentally friendly alternative methodology.Our procedure involves three cofactor-independent enzymes, which provides the advantage that no expensive cofactors nor cofactor recycling systems are needed.Furthermore, acetaldehyde can be produced from the degradation of biomass as a pyrolysis product of cellulose, while benzaldehyde derivatives can be obtained from the degradation of lignin. [30,31]Hence, the one-pot multi-enzymatic cascades presented in this work, operating under mild conditions without the use of cofactors and harsh chemicals, provide greener and more sustainable methodologies for the synthesis of chiral aromatic α,β-epoxy aldehydes and 1,2,3-prim,sec,sec-triols.

Results and Discussion
[18][19] This unique feature makes 4-OT an exceptionally useful enzyme that can be applied to build molecular complexity out of simple building blocks such as benzaldehyde and its derivatives, acetaldehyde and longer aliphatic aldehydes.We have previously shown that 4-OT-catalyzed aldol reactions can also be coupled with subsequent chemical or enzymatic transformations to obtain a variety of products, such as γ-aminobutyric acids and antidepressant precursors. [11,19]Mutability landscape-guided engineering allowed to increase significantly the catalytic efficiency (k cat /K M ) of the aldolase activity of 4-OT. [33]The best 4-OT variant, M45T/F50A (TA), displayed an improvement in the catalytic efficiency of >3000-fold compared to wild-type 4-OT.However, the 80-fold increased accumulation of cinnamaldehyde 3a (Fig. 1 -A) in the reaction catalyzed by TA (0.5 mg/mL enzyme, 100 mM acetaldehyde, 2 mM Scheme 1. Schematic representation of the three-step enzymatic cascade for the synthesis of enantioenriched 1,2,3-prim,sec,sec-triols, involving two engineered 4-OT variants (R3M1 and P8a), [22] one specifically selected epoxide hydrolase [32] and one final reduction step using NaBH 4 .

Results and Discussion
[18][19] This unique feature makes 4-OT an exceptionally useful enzyme that can be applied to build molecular complexity out of simple building blocks such as benzaldehyde and its derivatives, acetaldehyde and longer aliphatic aldehydes.We have previously shown that 4-OTcatalyzed aldol reactions can also be coupled with subsequent chemical or enzymatic transformations to obtain a variety of products, such as γ-aminobutyric acids and antidepressant precursors. [11,19]tability landscape-guided engineering allowed to increase significantly the catalytic efficiency (kcat/KM) of the aldolase activity of 4-OT. [33]The best 4-OT variant, M45T/F50A (TA), displayed an improvement in the catalytic efficiency of >3000-fold compared to wild-type 4-OT.However, the 80-fold increased accumulation of cinnamaldehyde 3a (Fig. 1 -A) in the reaction catalyzed by TA (0.5 mg/mL enzyme, 100 mM acetaldehyde, 2 mM benzaldehyde) compared to 4-OT wild type comes at the cost of the substrate selectivity.After 24 h incubation of the reaction mixture, TA mainly produces the side product 3aa (Fig. 1 -A) with formation of only 27% 3a (Fig. 1 -B).TA catalyzes the 1,2-addition of acetaldehyde to the freshly formed cinnamaldehyde 3a, while a relatively large amount of benzaldehyde 1a is not consumed (Fig. S1).For this reason, we decided to further engineer the TA enzyme, using triple-site saturation mutagenesis followed by two rounds of error-prone PCR, to obtain a 4-OT variant with better substrate selectivity towards benzaldehyde.To reduce the benzaldehyde) compared to 4-OT wild type comes at the cost of the substrate selectivity.After 24 h incubation of the reaction mixture, TA mainly produces the side product 3aa (Fig. 1 -A) with formation of only 27% 3a (Fig. 1 -B).TA catalyzes the 1,2-addition of acetaldehyde to the freshly formed cinnamaldehyde 3a, while a relatively large amount of benzaldehyde 1a is not consumed (Fig. S1).For this reason, we decided to further engineer the TA enzyme, using triple-site saturation mutagenesis followed by two rounds of error-prone PCR, to obtain a 4-OT variant with better substrate selectivity towards benzaldehyde.To reduce the screening effort, we developed a solid-phase iminium-ion activated prescreening method based on a previously reported methodology. [34]4-OT variants catalyzing the aldol condensation of salycilaldehyde and acetaldehyde could be detected upon the formation of a red-colored merocyanine-dye-like enzyme-bound complex between the reaction product 2-hydroxy-cinnamaldehyde and Pro-1 of the active 4-OT mutant.In the first round of the directed evolution trajectory, the pre-screening allowed a ~40-fold reduction in the screening effort (about 2.5% of the colonies turned red).The red colonies were selected and the corresponding 4-OT mutants screened for aldolase activity at two different concentrations of benzaldehyde (0.4 mM and 2 mM).We reasoned that to obtain a 4-OT variant with a better substrate selectivity towards benzaldehyde, we should screen for a variant with a low K m for benzaldehyde.Interestingly, after three rounds of directed evolution, we were able to retrieve almost completely the substrate selectivity observed for 4-OT wild type (Fig. 1 -B, formation of 88% of the desired product (3a/3a+3aa = 0.88) for the reaction catalyzed by 4-OT variant R3M1), while enhancing the cinnamaldehyde accumulation by 160-fold.The substrate scope of the best 4-OT mutant found in each round of the directed evolution procedure, named respectively R1M1, R2M13 and R3M1, was investigated (Scheme 1 -first step highlighted in orange) and compared with the activity of the parental variant TA. [19] The reaction was followed and analyzed by GC-MS after 6 h (Fig. S2) and 24 h (Table 1).Overall, R3M1 showed the best conversions and a broad substrate scope, although TA showed slightly better conversions with the p-F-benzaldehyde and the p-Br-benzaldehyde.Importantly, R3M1 has an increased substrate selectivity producing only neglectable amounts of side product 3aa (Fig. S1).For several more reactive substrates, such as the o-Cl-, m-Cl-and o-F-benzaldehyde, the unwanted side product is still formed in slight [a] dehydrated 1,2-aldol adduct from o-Cl-cinnamaldehyde.
amount (4-17%).Notably, for the o-Cl-benzaldehyde almost full conversion to the respective cinnamaldehyde is reached already after 6 h in the presence of R3M1 (Fig. S2).Hence, the formation of the side product is observed only after longer incubation times, decreasing the overall amount of o-Cl-cinnamaldehyde in the reaction mixture from 98% after 6 h to 82% after 24 h.Depending on the starting substrate, either R3M1 or TA was chosen as the catalyst for the first step of the anticipated multiple-step enzymatic cascades.
For the second step of the cascade, the expoxidation of 3 to yield 4 (Scheme 1), the following 4-OT variants present in our laboratory were tested: YIA, the first 4-OT mutant reported as a promiscuous cofactor-independent peroxyzyme, [20] P8a, a tandem-fused 4-OT variant specifically evolved for the epoxidation reaction using hydrogen peroxide, [22] and the best 4-OT mutants from each round of directed evolution for enhanced and selective aldol condensation activity (R1M1, R2M13 and R3M1, this work).Overall, the 4-OT variant P8a showed the best epoxidation activity against a series of cinnamaldehydes (Table 2).Interestingly, R2M13 and R3M1 also exhibited epoxidation activity with different cinnamaldehyde derivatives, although the conversions achieved were generally lower compared to those obtained for the reactions catalyzed by P8a.Progress curve analysis indeed confirmed that P8a has somewhat better activity than R3M1 (Fig. 2).Because P8a showed satisfactory conversion of various cinnamaldehydes and displayed good diastereoselectivity and excellent enantioselectivity, [22] it was chosen as catalyst for the epoxidation step of the designed cascade.Having evolved an improved 4-OT variant for the aldol condensation between acetaldehyde and benzaldehydes with high substrate specificity and selected a suitable catalyst for the epoxidation of cinnamaldehydes, we set out to establish a two-step enzymatic cascade for the synthesis of enantioenriched epoxides.The aldol condensation catalyzed by TA or R3M1 and the epoxidation reaction catalyzed by P8a were combined in a one-pot cascade manner, either in a continuous or stepwise mode.
Both approaches led to good conversions of the starting substrates into the corresponding epoxides.Nevertheless, the stepwise mode was preferred based on comparison of the overall conversions (Figs.S4 and S5).Accordingly, eight substrates were tested for the one-pot, two-step cascade in a stepwise mode.Good to excellent conversions were achieved, yielding the desired epoxides 4 in good to excellent diastereo-and enantiopurity (Table 3).To further demonstrate the preparative usefulness of this two-step biocatalytic cascade, we selected three substrates (1a, 1j and 1k) for preparative-scale (10 mg) reactions, yielding the corresponding enantioenriched epoxide products in 47-68% crude yield (Table 3).
Since R3M1 was also shown to be capable of catalyzing the epoxidation step (Fig. 2), we performed a mg-scale reaction with substrate 1a using R3M1 as single multifunctional catalyst (2 mol%) for both the aldol condensation and the epoxidation step.Excellent conversion (89% in 27 h), good isolated product yield (55%), and excellent stereopurity (d.r.= 93:7; e.r.= 98:2) were achieved.For comparison, when applying R3M1 (1 mol%) for the aldol condensation and P8a (0.5 mol%) for the epoxidation step, 91% conversion, 59% isolated product yield, and an d.r. of 93:7 and e.r. of 99:1 were realized.These results demonstrate that for selected substrates, the two-step cascade can also be performed with only one 4-OT variant (R3M1) as the sole catalyst for two different reactions, giving very good conversion and pleasingly high enantioselectivity.Having developed a two-step enzymatic cascade for the production of different chiral α,β-epoxy aldehydes, we focused our attention on the synthesis of chiral vicinal 1,2,3-prim,sec,sec-triols via a three-step enzymatic cascade (followed by a chemical reduction step).The synthesis of these vicinal triols can be performed by the coupling of the two 4-OT variants, previously evolved and used in the two-step cascade, with an epoxide hydrolase (Scheme 1 -third step highlighted in yellow).We selected six epoxide hydrolases from literature based on their broad substrate scope and convenient protein purification method (affinity chromatography applying a His-tag) (Table S2).
][37] Monitoring substrate consumption via HPLC, preliminary enzyme activity assays demonstrated that SibeEH and CH65, which belong to the α,β-hydrolase superfamily, were active towards several ((2R,3R)-3-phenyloxiran-2-yl)methanol derivatives.Therefore, SibeEH and CH65 were further investigated for their ability to promote the final step of the cascade.The reaction conditions for the epoxide ring opening were first optimized on analytical scale with compound 4a as racemic mixture and compound 6 as commercially available chiral source (Scheme S1), using SibeEH as the catalyst.For compound 6 the reaction was conducted on a c] e.r.milligram-scale to verify the presence of the final desired compound 7 by NMR analysis (Fig. S6).Finally, several selected aromatic 1,2,3-prim,sec,sec-triols (5a, 5j and 5k, Table 4) were prepared via the one-pot, three-step enzymatic cascade (followed by chemical reduction), applying the 4-OT variants R3M1 and P8a and the epoxide hydrolases SibeEH (substrates 4a and 4k) and CH65 (substrate 4j).
Moderate to good isolated product yields (22-33%) were achieved after 24 h from the initiation of the third step, with an overall reaction time of 51 h.The assignment of the absolute configuration for the 1,2,3-prim,sec,sec-triols is rather challenging.In a recent study, Zhang and co-workers developed an interesting approach based on NMR analysis for such compounds. [23]They could distinguish among all four isomers and between threo and erythro configurations in different deuterated solvents.In this work, we tentatively assigned the absolute configuration of the three chemoenzymatic products (5a, 5j and 5k) based on chiral HPLC analysis, using chemically synthesized racemic and chiral reference compounds.Following the general rules of epoxide ring opening, which can occur under either acidic or basic conditions (Scheme S2), and the data collected, we found that the chemoenzymatic product corresponds to the chiral reference compound generated from epoxide ring opening under acidic conditions (0.5 N H 2 SO 4 , SI -paragraph 7).We therefore tentatively assign the chemoenzymatically produced triols the 1S,2R configuration.Table 4.One pot, three-step enzymatic cascade, followed by chemical reduction (Scheme 1).Isolated yield [%], d.r.(threo:erythro) and e.r. of the final 1,2,3-prim,sec,sec-triols from the one-pot, three-step enzymatic cascade.

Entry
a] 1

Conclusion
In conclusion, we first developed an artificial aldolase, 4-OT R3M1, with 160-fold improved activity compared to 4-OT wild type and enhanced selectivity for benzaldehydes compared to previously evolved 4-OT variants. [19]We then applied this artificial aldolase in combination with a recently engineered artificial peroxyzyme [22] to construct a one-pot, two-step enzymatic cascade for the synthesis of highly enantioenriched oxiranes (e.r. up to 99:1) from three simple building blocks, namely acetaldehyde, benzaldehyde and H 2 O 2 .This cascade, employing tailor-made enzymes for each reaction step, was conducted at semi-preparative scale for selected substrates producing the respective enantioenriched epoxides in good yield (up to 68%).Interestingly, the tailored aldolase R3M1 also showed promiscuous peroxygenase activity, enabling the catalysis of both reaction steps (C-C and C-O bond-formation) by a single multifunctional enzyme.Furthermore, a three-step enzymatic cascade was developed for the synthesis of aryl-glycerols, achieving excellent enantiopurity (up to 98% ee) and respectable yields.Given that the isolation of 1,2,3-prim,sec,sec-triols from aqueous medium is challenging, enhanced product yields may be achieved by further improved work-up and purification procedures.The combination of two tailor-made non-natural biocatalysts and a specifically chosen natural epoxide hydrolase, all independent from any cofactor, in a one-pot, three-step cascade provides an environmentally friendly strategy to synthesize vicinal aromatic triols, achieving the desired molecular complexity from simple biomass-derived starting materials.

Experimental Section Materials
Chemicals were purchased from Sigma-Aldrich Chemical Co.(St.Louis, MO), TCI Europe N.V., Thermo Fisher Scientific (Geel, Belgium) or acbr (Germany) and used without further purification.The cinnamaldehyde derivatives were prepared using a previously reported method. [11]Synthetic genes were purchased from Invitrogen (Thermo Fisher Scientific).Proteins were analyzed by SDS-PAGE on precast gels (NuPAGE TM 12% Bis-Tris protein gels).ESI-MS analysis of purified enzyme variants was performed by the Mass Spectrometry core facility of the University of Groningen.High performance liquid chromatography (HPLC) was performed with a Shimadzu LC-10AT HPLC with a Shimadzu SPD-M20A diode array detector.GC-MS analysis was conducted with a Shimadzu GC-MS-QP2010 SE.NMR spectra were recorded on a Bruker 500 MHz NMR machine at the Drug Design laboratory of the University of Groningen.Chemical shifts (δ) are reported in parts per million (ppm).

Methods
Construction of mutant libraries and pre-screening.
In the first round of directed evolution a triple-site library, 4-OT Q4X/L8X/A33X/ M45T/F50A, was constructed by PCR with degenerate primers (Table S1), by means of QuikChange PCR.pDNA harboring the gene coding for 4-OT M45T/ F50A was used as template.During the second and third round of directed evolution, error prone PCR libraries were built using the gene encoding the best mutant from the previous round as template (1 ng/µL-200 pg/µL; for primers see Table S1).The PCR reactions were set up using DreamTaq DNA polymerase (ThermoFisher Scientific), 0.5 mM Mn 2+ , 7 mM Mg 2+ and dNTPs in the following concentrations: dCTP 1 mM, dTTP 1 mM, dATP 0.2 mM, dGTP 0.2 mM.Following, 40 cycles of PCR amplifications were performed.The amplified gene was then cloned in an empty pET20b vector using T4 ligase after digestion with NdeI and BamHI restriction enzymes.The ligation mixture was transformed into electrocompetent E. coli DH5α cells.The overnight culture was harvested, and the plasmids were isolated and used for transformation of E. coli BL21 DE3 cells and screening.The constructed libraries were assayed for active clones based on a slightly modified iminium-ion activated pre-screening method. [34]Accordingly, the BL21 (DE3) transformation mixture (100 µL), diluted 10x, was plated on LB agar media, supplemented with ampicillin (100 µg/mL) and 0.2% lactose, and was incubated overnight at 37°C and subsequently at RT for 1.5 h.Subsequently, a filter sterilized solution of 0.6% agarose in 20 mM NaPi pH 7.3 with 5 mM salycilaldehyde was poured on the surface of the agar plates covering it with a thin layer.The agarose solidified at RT in 30 minutes.A second solution of 20 mM NaPi pH 7.3 and 2 M acetaldehyde was pipetted on the gel under sterile conditions.The agar plate was sealed with parafilm and incubated at RT. Colonies producing enzyme variants capable of performing the aldol condensation reaction between salycilaldehyde and acetaldehyde turned red upon formation of a merocyanin-like enzyme-bound complex between the reaction product 2-hydroxy-cinnamaldehyde and the Pro-1 residue of the 4-OT mutant.Colonies that showed clear red staining were picked with sterile toothpicks and were used for enzyme activity screening as described below.

Library screening for aldol condensation
The colonies picked were used to inoculate 1 mL of LB medium supplemented with ampicillin (100 μg/mL) and lactose (0.2% w/v) in 96-deep-well plates (Greiner Bio-one, 96-well Masterblock).The plates were sealed with sterile gas-permeable seals (Greiner Bio-one, BREATHseal) and incubated at 37°C for 18 h at 250 rpm.
To prepare glycerol stocks, 50 µL of the culture was mixed with 30 μL of a sterile solution of 80% glycerol in H 2 O and stored at −80°C.The overnight culture was harvested, and the pellets were lysed for 30 min with 200 μL of BugBuster solution (Novagen).The cell free extract (CFE) was obtained by centrifugation.The final screening mixture consisted of: 50% CFE, 0.4 mM benzaldehyde, 50 mM acetaldehyde in NaPi buffer 20 mM pH 7.3.The reaction was performed in a 96-well plate (Greiner Bio-one, UV-star F-bottom microplate) and monitored in a plate reader for depletion of benzaldehyde at 250 nm and formation of cinnamaldehyde at 290 nm for 18 h.Mutant enzymes with enhanced aldolase activity were selected and further characterized by DNA sequencing and activity testing as purified enzymes.

Expression, purification and preliminary activity assay of the newly evolved 4-OT variants
4-OT enzymes were expressed and purified according to a protocol reported previously with slight modifications. [38]5 mL of LB medium (pre-culture) was inoculated with colonies from E. coli BL21 (DE3) cells carrying the appropriate expression plamid and incubated at 37°C for 18 h. 1 L of LB medium (containing 100 mg/mL ampicillin) was inoculated with the pre-culture and incubated at 37°C until an OD 600 of 0.6-0.8 was reached.Then IPTG (0.75 mM) was added to induce the expression and the cells were incubated at 18°C for additional 18-20 h.The cells were harvested by centrifugation and further used for protein purification as previously reported. [38]The concentration of the purified enzymes was determined using the Waddel method. [39]The mass of the purified enzymes and the removal of the initial methionine were confirmed by ESI-MS.The enzymes were snap-frozen by liquid nitrogen and stored at −20°C until further usage.Reaction conditions for the aldol condensation were set as following: purified 4-OT variant 0.5 mg/mL, 2 mM or 10 mM benzaldehyde 1a, and 100 mM acetaldehyde in 20 mM NaPi buffer pH 7.3 at room temperature.The accumulation of cinnamaldehyde 3a and side product 3aa (Fig. 1 -A) were assessed via reverse phase HPLC with a gradient mixture of ACN/water as the mobile phase at different time points (24 h and 48 h) after dilution of the reaction mixture 1:1 in ACN (1mL/min, linear gradient of 8% to 90% ACN/20 min, Phenomenex-kinetex xb-c18 column).The area of the peaks corresponding to 3a and to 3aa were used to assess relative product accumulation and substrate selectivity (Fig. S1).

Substrate screening -aldol condensation (first step)
The aldol condensation activity assay for four 4-OT variants (R1M1, R2M13, R3M1 and TA) was performed in 600 µL reaction volume.Fresh stock solutions of acetaldehyde (1 M in 20 mM NaPi buffer pH 6.5), benzaldehyde derivatives 1a-j (100 mM in EtOH) and purified 4-OT variants (in buffer, concentration varies) were prepared.The final reaction mixture consisted of acetaldehyde (100 mM), 1a-j (5 mM), and 4-OT (1 mol%) in 20 mM NaPi buffer (pH 6.5) with 5% (v/v) EtOH.The reactions were initiated by the addition of acetaldehyde and run at RT without shaking.After 6 h and 24 h from the initiation of the reaction a sample of 200 µL from each mixture was withdrawn and extracted with 200 µL of EtOAc (1x).The organic phase was analyzed by GC-MS to measure the conversion [%] of benzaldehyde (1a-j) to cinnamaldehyde (3a-j).The results for the analysis after 6 h incubation time are summarized in Fig. S2.

Substrate screening -epoxidation (second step)
The epoxidation activity assay for five For the one-pot two-step continuous cascade mode all the components were added from the beginning and the reaction mixture was incubated for 24 h at RT without shaking.The mixture was then extracted with EtOAc (1x) and the organic phase was analyzed by GC-MS.The results for the one-pot two-step stepwise and one-pot twostep continuous cascade mode are summarized in Fig. S4 and Fig. S5 respectively.

Two-step enzymatic cascade -semi-preparative scale
General procedure.The two-step enzymatic cascade was performed on a semi-preparative scale (20 mg) as follows: 1a, 1j or 1k (5 mM), acetaldehyde (100 mM), R3M1 (1 mol%), P8a (1 mol%), H 2 O 2 (50 mM) in 20 mM NaPi buffer (pH 6.5) with 5% (v/v) EtOH, using a total volume of 30 or 50 mL depending on the substrate.The first step of the cascade was initiated by the addition of the substrate 1a, 1j or 1k (5 mM), acetaldehyde (100 mM) and R3M1 (1 mol%) in 30 or 50 mL NaPi buffer (20 mM, pH 6.5) in a 50 mL falcon tube and stirred at 60 rpm at 25°C.After 24 h the second enzyme, P8a (0.5 mol%), and H 2 O 2 (50 mM) were added and the reaction mixture was left at 25°C, 60 rpm for 3 h.Before product isolation, 0.005 mg/mL of catalase was added to the reaction mixture to quench the excess of hydrogen peroxide.The reaction mixture was left at room temperature for 1 h and then extracted 3x with EtOAc (30-50 mL).The organic layer was dried over MgSO 4 , filtered and the solvent was removed under vacuum.The crude product was isolated and analyses by 1 H NMR without further purification.To perform the chiral HPLC analysis, ~1 mg of the isolated aldehyde was reduced to the corresponding alcohol by the addition of NaBH4 (3 mg) and left at RT for 1 h.50 µL of a saturated solution of NH 4 Cl was added and then the mixture was extracted with EtOAc (1x).The EtOAc was then evaporated and the isolated product re-suspended in 200 µL ACN.

Cloning, expression and purification of epoxide hydrolases
][37] The cloning procedures were slightly modified from previously reported protocols.The gene sequences of the selected epoxide hydrolases were obtained from the NCBI GenBank and codon-optimized for E. coli using the programme GeneArt ® .The restriction sites for NdeI and XhoI were added to the 5'-end and the 3'-end of the gene sequence to insert the genes into the respective vector in frame with the vector-encoded His-tag.The synthetic genes were cloned in pET28a (BmEH, Lsd19 and AuaI) or in pET20b (SibeEH and CH65).The final reaction mixture for the restriction of the genes contained (30 µL total volume): 200 ng DNA, 1 µL NdeI, 1µL XhoI, Fast Digest buffer and MilliQ water.The mixture was incubated at 37°C for 4 h.The final reaction mixture for the vectors, pET28a and pET20b, contained (100 µL total volume): 2 µg vector, 5 µL NdeI, 5 µL XhoI, Fast Digest buffer and MilliQ water.The mixture was incubated for 4 h at 37°C and then the vector DNA was dephosphorylated with FastAP (30 min digestion at 37°C).DpnI (2.5 µL) was added in case the vector was amplified by PCR before usage.The cut and purified genes were ligated with the respective cut and purified vector, with a vector to insert ration of 1:3.The ligation mixture (20 µL total volume) resulted in: 80 ng of vector, approx.20 ng DNA of the respective gene, 1 µL ligase (1 U), ligase buffer and MilliQ water incubated at room temperature for 1 h.The successful insertion of the corresponding genes into the respective vector was confirmed by DNA sequencing.The plasmids containing the cloned genes were transformed in chemo-competent E. coli BL21 cells and the genes were expressed at 18°C for 18 h by induction with IPTG (0.75 mM for 1 L culture at an OD 600 of 0.6-0.8).The cells were disrupted by sonication and the epoxide hydrolase enzymes purified by Histag affinity chromatography (Ni-Sepharose NTA resin).The fractions containing pure protein were analyzed by SDS-PAGE (Fig. S3), combined, and applied on a PD10 desalting column for imidazole removal.Then, the collected fractions were concentrated by Vivaspin (30K cutoff), snap-frozen in liquid nitrogen and stored at −20°C.The protein concentration was calculated by Nanodrop spectrophotometric analysis using the corresponding extinction coefficient at 280 nm for each protein.

Activity assay of epoxide hydrolases -analytical and semipreparative scale
The enzymatic epoxide ring opening reaction was initially analyzed as follows: 4 (5 mM) and epoxide hydrolase (0.1 mol%) in 20 mM NaPi buffer (pH 6.5) with 5% (v/v) EtOH.The reaction progress was followed by reverse phase HPLC.To verify the presence of the desired product, a milligram-scale reaction (10 mL final volume) was performed with substrate 6 and SibeEH as the catalyst.Fresh stock solutions of 6 (200 mM in EtOH) and SibeEH (in buffer, concentration varies) were prepared.The final reaction mixture consisted of 6 (10 mM) and SibEH (0.1 mol%) in 20 mM NaPi buffer, pH 6.5.The reaction was run at room temperature without shaking.After 48 h, the mixture was extracted with EtOAc (5x).The combined organic layers were washed with brine and dried over Na 2 SO 4 .The solvent was removed under vacuum and the crude product was analyzed by 1 H NMR.

Three-step enzymatic cascade synthesis of vicinal chiral 1,2,3-prim,sec,sec-triols -semi-preparative scale
The three-step enzymatic cascade was performed on a semi-preparative scale (20 mg) using the following conditions: 1a, 1j or 1k (5 mM), acetaldehyde (100 mM), R3M1 (1 mol%), P8a (1 mol%), H 2 O 2 (50 mM), SibeEH or CH65 (0.1 mol %) in 20 mM NaPi buffer (pH 6.5) with 5% (v/v) EtOH (total volume of 2 75 30 mL).The first step of the cascade was initiated by the addition of the substrate 1a, 1j or 1k, acetaldehyde, and R3M1 in 30 mL NaPi buffer (20 mM, pH 6.5) in a 50 mL falcon tube and stirred at 60 rpm at 25°C.After 24 h, the second enzyme P8a and H 2 O 2 were added and the reaction mixture was left at 25°C and 60 rpm for 3 h.Before the addition of the third enzyme, 0.005 mg/mL of catalase was added to the reaction mixture.The reaction mixture was incubated for 1 h at RT to remove the excess of H 2 O 2 .Afterwards, the respective epoxide hydrolase was added, SibeEH for the substrates 1a and 1k and CH65 for substrate 1j.The reaction mixture was incubated at 25°C and 60 rpm for additional 24 h.After completion of the reaction, 40% ACN were added to allow denaturation and precipitation of the enzymes.Then, NaBH 4 (85 mg, 15 eq of 1) was added at 0°C and the mixture was incubated at 0°C for 30 min.The mixture was stirred at 60 rpm at room temperature for an additional hour and quenched with a saturated solution of NH 4 Cl (15 mL) followed by lyophilization for 18 h.The lyophilized mixture was treated with EtOAc and sonicated for 10 min followed by filtration on a pad of celite.The celite was washed with 3x15 mL of EtOAc and the solvent was removed under vacuum to provide the crude product.The crude product was further purified by flash chromatography (DCM/MeOH gradient of eluent from 2 to 20% MeOH).For compound 5a the two diastereomers could be separated.

DNA and Amino Acid Sequences of New 4-OT Variants
Amino acid sequence: in blue are highlighted the mutations present in the template and in red the new mutations characteristic for that specific mutant.

Two-Step Enzymatic Cascade Synthesis -Analytical Scale
Figure S4.One-pot two-step cascade, stepwise mode.Product distribution (%) in the reaction mixture after 24 h from the initiation of the second step by the addition of P8a and H 2 O 2 .Reaction followed and analysed by GC-MS.  .One-pot two-step cascade, continuous mode.Product distribution (%) in the reaction mixture after 24 h from the addition of all components to the reaction mixture.Reaction followed and analysed by GC-MS.

Synthesis of reference compounds
General procedure for synthesis of racemic reference compounds.
Racemic (3-phenyloxiran-2-yl)methanol, and derivatives (67 mg, 0.45 mmol), [9] was added to a mixture of sulfuric acid (0.5 N, 2 mL) and ethyl acetate (0.3 mL) and vigorously stirred for 6 h at room temperature.The reaction was followed by Thin-Layer Chromatography (TLC) (DCM/MeOH 10%).The organic layer was removed under vacuum.The remaining water mixture was transferred to an SPE column (Supelclean ENVI-18, 500 mg, 6 mL, 57064 Merck).The SPE column was first washed with MeOH (2 × 5mL) and demi-water (2 × 5 mL).Then, it was rinsed 4x with 1 mL of water.Each rinsing step was collected separately.Subsequently, to collect either the side products or the starting material, in every following rinsing step, water was mixed with MeOH (50 µL) to give a 1:1 mixture.The rinsing water steps were collected and neutralized with saturated NaHCO 3 solution followed by lyophilization.The solid mixture was washed with ethyl acetate by filtration over celite until all the desired compound was eluted from the celite pad and no further product was observed by TLC.
Each rinsing step was collected separately.Subsequently, to collect the side products or the starting material, in every following rinsing step water was mixed with MeOH (50 µL) giving a 1:1 mixture.The rinsing water steps were collected and neutralized with saturated NaHCO3 solution followed by lyophilization.The solid mixture was washed with ethyl acetate by filtration over celite until all the desired compound was eluted from the celite pad and no further product was observed by TLC.1H). 1 H NMR data were in agreement with literature. [6,7]eneral procedure for synthesis of enantiomers as reference compounds.
Each rinsing step was collected separately.Subsequently, to collect the side products or the starting material, in every following rinsing step water was mixed with MeOH (50 µL) giving a 1:1 mixture.The rinsing water steps were collected and neutralized with saturated NaHCO3 solution followed by lyophilization.The solid mixture was washed with ethyl acetate by filtration over celite until all the desired compound was eluted from the celite pad and no further product was observed by TLC.
steps were collected and neutralized with saturated NaHCO 3 solution followed by lyophilization.The solid mixture was washed with ethyl acetate by filtration over celite until all the desired compound was eluted from the celite pad and no further product was observed by TLC.1H). 1 H NMR data were in agreement with literature. [6,7]S,2R)-

III. Chiral HPLC Analysis
Entry Compound Column [b] Mobile phase [c] Retention time [d] (min)

Figure 1 .
Figure 1.A) Schematic representation of the aldol condensation between benzaldehyde (1a) and acetaldehyde catalyzed by a 4-OT variant and further addition of acetaldehyde to cinnamaldehyde (3a) with the formation of an undesired aldol adduct (3aa); B) Enhanced cinnamaldehyde accumulation relative to 4-OT wild type (green bars), and improved substrate selectivity (dark red curve) from 4-OT wt to R3M1 variant showing the amount of cinnamaldehyde formed as fraction of the total product (3a/3a+3aa).Reaction conditions: purified 4-OT variant 0.5 mg/mL, 2 mM (dark red curve) or 10 mM (green bars) benzaldehyde 1a, and 100 mM acetaldehyde in 20 mM NaPi buffer pH 7.3 at room temperature (24 h).
[a] determined by GC-MS; [b] determined by NMR of the reaction mixture on 10 mg-scale after an overall reaction time of 27 h; [c] determined by chiral HPLC.In brackets, the d.r. and e.r. of 10 mg-scale reactions are shown; [d] TA used as catalyst; [e] d.r. and e.r.calculated after an overall reaction time of 27 h; [f] yield calculated after an overall reaction time of 42 h.
calculated by chiral HPLC; [b] CH65 used as catalyst.

Figure S3 .
Figure S3.SDS-PAGE analysis of the purified epoxide hydrolase (EH) fractions.(SibeEH was purified in a separate batch).

Figure S3 .
Figure S3.SDS-PAGE analysis of the purified epoxide hydrolase (EH) fractions.(SibeEH was purified in a separate batch).

Figure S4 .
Figure S4.One-pot two-step cascade, stepwise mode.Product distribution (%) in the reaction mixture after 24 h from the initiation of the second step by the addition of P8a and H2O2.Reaction followed and analysed by GC-MS.
Figure S5.One-pot two-step cascade, continuous mode.Product distribution (%) in the reaction mixture after 24 h from the addition of all components to the reaction mixture.Reaction followed and analysed by GC-MS.

Figure S4 .
Figure S4.One-pot two-step cascade, stepwise mode.Product distribution (%) in the reaction mixture after h from the initiation of the second step by the addition of P8a and H2O2.Reaction followed and analysed by GC-MS.

Figure S5 .
Figure S5.One-pot two-step cascade, continuous mode.Product distribution (%) in the reaction mixture after h from the addition of all components to the reaction mixture.Reaction followed and analysed by GC-MS.

Table 1 .
The aldol condensation of 1 (5 mM) and 2 (100 mM) to yield 3 catalyzed by different 4-OT variants (1 mol%).The formation of the side product, produced by the 1,2-addition of acetaldehyde to cinnamaldehyde, is shown in brackets.Conversions were determined by GC-MS (reaction time is 24 h).

Table 2 .
Conversion of 3 to 4 [%] catalyzed by selected 4-OT variants and d.r. of formed product 4 (syn:anti), determined by GC-MS after two hours reaction time.

Table 3 .
One-pot, two-step cascade.Conversion of 1 to 4 [%] catalyzed by 4-OT variants in two sequential steps in one pot on analytical scale, yield of 4 [%] shown in brackets, and d.r.(syn:anti) and e.r (R:S) of 4 after 48 h (overall reaction time).

4k). Two-step enzymatic cascade -analytical scale General procedure
4-OT variants (P8a, YIA, R1M1, R2M13 and R3M1) was performed in 600 µL reaction volume.Fresh stock solutions of cinnamaldehdye derivatives 3a-f, 3h and 3k (100 mM in EtOH) and purified 4-OT variants (in buffer, concentration varies) were prepared.H 2 O 2 was directly added from the main stock (10 M in H 2 O).The final reaction mixture consisted of H 2 O mM in EtOH) and purified 4-OT variants (in buffer, concentration varies) were prepared.H 2 O 2 was directly added from the main stock (10 M in H 2 O) either in a one-pot two-step continuous cascade mode or a one-pot two-step stepwise cascade mode.P8a and H 2 O 2 were added in the same pot and the mixture was left to incubate for an additional 24 h at RT without shaking.

Table S1 .
Primers used in the construction of mutant libraries

Table S2 .
Overview of the epoxide hydrolases cloned, expressed, purified and tested.