Enzymatic Dynamic Reductive Kinetic Resolution Towards 115 g/L (S)-2-Phenylpropanol

Published biocatalytic routes towards chiral 2-phenylpropanol by oxidoreductases showed product concentrations of maximally 80 mM. Enzyme deactivation turned out as one major limitation and was attributed to adduct formation of the aldehyde substrate with the catalytic reductase. Results A Candida tenuis xylose reductase single-point mutant (CtXR D51A) with very high catalytic eciency (43·10 3 s -1 M -1 ) for (S)-2-phenylpropanal was identied. The enzyme showed high enantioselectivity for the (S)-enantiomer but was deactivated by 0.5 mM substrate within 2 h. A whole-cell biocatalyst based on the engineered reductase and a yeast formate dehydrogenase for NADH-recycling provided substantial stabilization of the reductase. The relatively slow in situ racemization of 2-phenylpropanal and the still limited biocatalyst stability required a subtle adjustment of the substrate-to-catalyst ratio. A value of 3.4 g substrate /g cell-dry-weight turned out as compromise between product enantiopurity and conversion. A catalyst loading of 40 g cell-dry-weight was used to convert 1 M racemic 2-phenylpropanal to (S)-phenylpropanol in 93.1 % e.e. Mainly hydrolases have been exploited for the production of profenols at industrial scale so far. The herein established bioreduction presents an alternative route towards profenols that is competitive to hydrolase-catalyzed kinetic resolutions. substrate-to-biocatalyst product profenols by oxidoreductases show markedly lower product concentrations of maximally 80 mM 2-aryl-1-propanols (4, 8, 9, 10, 11, 12). However, oxidoreductases generally outperform hydrolases in terms of enantioselectivity. Here, we report on an enzymatic dynamic reductive kinetic resolution towards (S)-2-phenylpropanol. A single point mutant of the xylose reductase from Candida tenuis (CtXR, superfamily of aldo-keto reductases) with high catalytic activity and excellent enantioselectivity for (S)-2-phenylpropanal was used in E. coli whole-cell reductions. Reaction optimization with the aim to achieve high enantioselectivity and product concentration at full conversion was accomplished. The established enzymatic dynamic reductive kinetic resolution (DYRKR) is compatible with lipase-based processes in terms of product concentration and optical purity. enzymatic reduction bioconversions phenylpropanal, shrunk to < 3 min. In the case of o-chloroacetophenone and its reaction product the unfavourable log P-values (~ 2) were held responsible for fast enzyme deactivation. Use of the enzyme as catalytic oxidoreductase in whole-cell catalysts (E. coli, S. cerevisiae, C. tenuis) had previously shown to stabilize the enzyme substantially and product concentrations were improved > 10-fold (34). Reactive aldehydes are known to form adducts with proteins at the lysine, histidine and cysteine side chains (25). Enzyme deactivation by 2-phenylpropanal might hence follow different mechanisms compared to o-chloroacetophenone. Use of the whole cells provided an extreme case of catalyst stabilization in the presence of 2-phenylpropanal: The isolated enzyme was deactivated by 0.5 mM aldehyde whereas the whole-cell catalyst and the supernatant thereof were able to tolerate and convert 1 M of the substrate. Stabilization of the catalytic enzyme by whole-cells and cell debris was previously reported for the synthesis of (R)-phenylacetylcarbinol from benzaldehyde and pyruvate by a Candida utilis pyruvate decarboxylase (35). The stabilization was ascribed to membrane components that form a microenvironment around the enzyme and thereby decrease aldehyde transfer to the enzyme and protect the enzyme from deactivation at the aqueous/organic interphase. enzymes), all of which were either HLADH or enzymes from thermophilic hosts. The low number of reported enzymes and the low concentrations of obtained product stressed a general deactivating effect of the reactive aldehyde on enzymes. Whole-cell level. The whole-cell catalyst based on D51A CtXR showed, much to our delight, high stability in the presence of 1 M phenylpropanal. Our results indicate a strong protecting effect by the whole-cell catalyst (and the supernatant thereof). We used lyophilized and rehydrated E. coli co-expressing D51A CtXR and yeast formate dehydrogenase as catalyst.


Ct XR mutants
The substrate-binding cavity of aldo-keto reductases is mainly formed by residues from three large and exible loops (15,16). Loop exibilities provide the structural basis for relaxed substrate speci cities but complicate rational engineering (17). A binding mode of the natural substrate D-xylose (open chain form) was previously modelled with C-1 of xylose within hydride-transfer distance above the nicotinamide C-4 and the carbonyl oxygen hydrogenbonded to the general acid catalyst Tyr-52. Therein, the aldehyde hydrogen pointed towards the indole ring of Trp-24, the C2 hydroxyl interacted with Asn-310 and C3, C4 and C5 hydroxyls with Asp-51 (18). Here, we probed CtXR wild-type and single-point mutants of the main substrate recognition residues Trp-24, Asn-310 and Asp-51 as catalysts for 2-phenylpropanal reduction. The replacement of Trp-24 by smaller phenylalanine and tyrosine increased activities on bulky ketone-substrates (13,19). Asp-51 contributes the most to the relative polarity of the binding site; substitution by alanine led to improved selectivity for the aromatic ketone o-chloroacetophenone (14).
Kinetic constants of 2-phenylpropanal reduction by CtXR variants Table 1 summarizes results of a steady-state kinetic analysis of NADH-dependent reduction of racemic and (S)-2-phenylpropanal by CtXR wild-type and mutants.
Racemic substrate. The wild-type enzyme showed a speci city constant (k cat /K m,rac ) of 130 s − 1 M − 1 re ected by a K m,rac value of 350 µM and a k cat value of 0.05 s − 1 . The mutants W24F and W24Y displayed only 8 to 10 % of the wild-type activity. While the N310A mutant had a speci city constant similar to the wild-type, the N310D mutant showed no activity. CtXR D51A stood out with a k cat /K m,rac of 28·10 3 s − 1 M − 1 composed of a K m,rac value of 170 µM and a k cat value of 4.8 s − 1 . Replacement of the charged aspartic acid by alanine led to 215-fold higher catalytic e ciency compared to the wildtype. Likewise, introduction of an additional aspartic acid in the substrate binding pocket (N310D mutant) abolished the activity with 2-phenylpropanal.
Note that used CtXR mutants showed 35 to 100-fold reduced catalytic activity towards the natural substrate xylose (19). Enlargement of the substrate binding pocket by replacement of the bulky Trp24 decreased the enzyme's activity towards 2-phenylpropanal.
(S)-2-phenylpropanal. Kinetic parameters obtained with the racemic substrate and the (S)-2-phenylpropanal were compared and the ratio of the speci city constants formed (k cat /K m,S )/(k cat /K m , rac ) ( Table 1). For the extreme case of sole activity with the S-enantiomer, a ratio of ~ 2, for equal acceptance of Sand R-enantiomer a ratio of 1 and for preference of the R-enantiomer ratios < 1 were expected. The wild-type showed a ratio of 1. 23 and preference for the S-enantiomer. The D51A mutant showed a ratio of 1.54 and hence a stronger preference for the S-enantiomer. The W24F, W24Y and N310A mutants, however, displayed ratios < 1 and therefore preference for the R-enantiomer.   (Table 3).
NAD + concentration. We added higher concentrations of the co-enzyme NAD + to further push the reaction towards full conversion. At 12 and 14 mM of NAD + , conversions up to 99% were reached, again at e.e. values of 92-93 % (Fig. 2, Table 3).
Fed-batch. The step-wise addition of substrate at 0, 2 and 4 h to a total substrate concentration of 1 M led to a 10 % increase of conversion (Table 3).
Recovery, isolated yield, reproducibility and by-products Leis et al. (20) previously suggested that hydrophobic substrates and products remain in the cell sludge of the used biocatalyst. Here, a high excess of ethyl acetate was required for product extraction prior to analysis. Obtained product concentrations in bioreductions of 1 and 2 M were between 27 and 84 % as shown in Table 3.
Recovery. Substrate/product loss in the biomass was found to be < 15% under the conditions used.
Reproducibility. Reaction replicates (N = 7) of bioreductions with 40 g CDW /L and 6 mM NAD + showed high reproducibility with a mean value of 62 % product and a standard deviation of 4 %. The enantiomeric excess was 93.3 ± 1.1 % e.e. The formation of broad peaks prevented quanti cation of the aldehyde substrate by chiral, reversed-phase HPLC. We therefore analyzed bioreduction samples additionally by chiral GC-FID. By-products. The high reactivity of the substrate 2-phenylpropanal prompted investigation of possible by-products from chemical or bio-chemical reactions. It has been previously shown that acetophenone forms by oxygen-catalyzed degradation of rac-2-phenylpropanal (21). We found 7% of acetophenone to be formed maximally and only trace amounts of its enzymatic reduction product in bioconversions of 1 M phenylpropanal, using 40 g CDW /L cells and 6 mM NAD + (22). The substrate is also in a chemical equilibrium between rac-2-phenylpropanal and its corresponding hydrates. The previously reported enzymatic oxidation of 2-phenylpropanal hydrates to the corresponding carboxylic acids was not observed (12,23). No substraterelated enol or aldol was found in detectable amounts (for data of NMR analyses see the Supplementary data).
Isolated yield. To con rm the identity of the obtained product, hydrophobic compounds were extracted from two reaction mixtures containing 1 M 2phenylpropanol (reaction volume 2 mL, 40 g CDW /L catalyst, 6 mM NAD + ). An analytical yield of 78 % was determined by HPLC (product concentration).

Literature survey
Reported enzymatic reductive kinetic resolution of rac-2-phenylpropanal studies are summarized in Table 4. Bioreductions of rac-2-phenylpropanal have been accomplished with free-oating enzymes ( Table 4, entries 1-5, 8-10,12) and immobilized enzymes (entries 6,7,11). Previous studies aimed at probing bioreduction catalysts (free and immobilized oxidoreductases) in the kinetic resolution of rac-2-phenylpropanal (entries 4-11). Most enantioselective enzymes preferred the (S)-aldehyde (entries 1-8, 10). Rocha-Martín et al. (24) reported on an anti-prelog speci c ADH from Thermus thermophilus HB27 (entry 11). Dong et al. (8) accomplished the evolution of ADHs for the formation of (S)-and (R)-alcohols by directed evolution of an ADH from Thermoanaerobacter brockii that displayed moderate prelog-type selectivity (entries 8, 9). The used enzymes had to display not only high enantioselectivities but also su cient stabilities in the presence of the substrate that can form adducts with groups on the enzymes (25). HLADH was used in most studies as it turned out to be enantioselective, stable in the presence of substrate up to a concentration of 165 mM, and useful in coupled substrate strategies (oxidation of cheap alcohols for NADH-recycling). All other examples of selective ADHs stem from thermophilic organisms and display intrinsically high stabilities towards adverse effects of the reaction media. Remarkably, the often-used host E. coli shows native activity towards 2-phenylpropanal (entry 13). Buffered solutions containing water-soluble co-solvents (also used as sacri cial substrate for NADH-recycling) were used frequently. The aqueous phase was required for the racemisation of the substrate. Grunwald et al. (11) tested HLADH in organic solvent and used isopropylether with 0.5 % buffer as reaction medium. A product concentration of 46 mM with 95 % e.e. was obtained, however at a conversion of 15 % (entry 7). Others used bi-phasic solvents (entries 4 and 6). The highest published product concentration of 82 mM was achieved in a buffer/isopropylether mixture (47:63) (entry 4).

Enantioselectivities of CtXR variants
The enantioselectivity of an enzyme is de ned by the ratio of catalytic e ciencies for the two enantiomers (E = (k cat /K m,S )/(k cat /K m , R ), 32). In the present case, the determination of the catalytic e ciencies of enantiopure (S) and (R)-aldehydes was compromised by in situ substrate racemization.
Published racemization velocities of 2-phenylpropanal were 75·10 − 6 s − 1 (k rac ), equal to half-lives of ~ 2 h (t 1/2 ) (27). The relatively slow racemization should generally enable determination of k cat /K m,enantiomer with enzymatic assays lasting 5 minutes. We have, however, experienced slow racemization of the pure enantiomers during freeze-storage (-18°C). Hence, enantioselectivities expressed as (k cat /K m,S )/(k cat /K m,rac ) in Table 1 show approximate values that are still useful to guide enzyme selection and reaction optimization. The wild-type showed preference for the (S)-aldehyde (ratio of 1.23).
Asp-51, Asn-310. The D51A mutant showed a stronger preference for the S-enantiomer (a ratio of 1.54) and the N310A for the R-enantiomer (ratio of 0.77). Asp-51 and Asn-310 are on opposite sides of the substrate binding pocket (Fig. 3). Asp-51 is suggested to interact with the natural substrate xylose at C3, C4 and C5 hydroxyls, Asn-310 with the C2 hydroxyl (18). A docking of wild-type CtXR (in complex with NAD + ; 33) with xylose and (S)-and (R)-2-phenylpropanal was made (Fig. 3AB). Replacement of Asp-51 by alanine might facilitate interaction between the alanine and the phenyl-ring of (S)-2-phenylpropanal (Fig. 3C). After replacement of Asn-310 by alanine, interaction between alanine and the phenyl-ring chain of (R)-2-phenylpropanal becomes plausible for N310A (Fig. 3D). Docking results supported ndings from the kinetic studies: D51A mutation improved transition state stabilization of the (S)-aldehyde while N310A mutation improved stabilization of the (R)-aldehyde, respectively.
Trp-24. W24F and W24Y mutants showed (k cat /K m,S )/(k cat /K m , rac ) ratios of ~ 0.91 (Table 1) along with approximately ≥ 10-fold reduction in catalytic e ciencies. Lower activities (with higher K m -values) and preference for the (R)-aldehyde might indicate poorer interaction between the aromatic rings of phenylalanine or tyrosine and the aldehyde proton of the (S)-aldehyde. In conclusion, the D51A turned out as CtXR variant with improved catalytic e ciency and enantioselectivity.

Catalyst stabilization
CtXR is generally known as a relative labile enzyme and half-lives in the presence of 5 to 10 mM o-chloroacetophenone and 1-(2-chlorophenyl)ethanol shrunk to < 3 min. In the case of o-chloroacetophenone and its reaction product the unfavourable log P-values (~ 2) were held responsible for fast enzyme deactivation. Use of the enzyme as catalytic oxidoreductase in whole-cell catalysts (E. coli, S. cerevisiae, C. tenuis) had previously shown to stabilize the enzyme substantially and product concentrations were improved > 10-fold (34). Reactive aldehydes are known to form adducts with proteins at the lysine, histidine and cysteine side chains (25). Enzyme deactivation by 2-phenylpropanal might hence follow different mechanisms compared to o-chloroacetophenone. Use of the whole cells provided an extreme case of catalyst stabilization in the presence of 2-phenylpropanal: The isolated enzyme was deactivated by 0.5 mM aldehyde whereas the whole-cell catalyst and the supernatant thereof were able to tolerate and convert 1 M of the substrate. Stabilization of the catalytic enzyme by whole-cells and cell debris was previously reported for the synthesis of (R)phenylacetylcarbinol from benzaldehyde and pyruvate by a Candida utilis pyruvate decarboxylase (35). The stabilization was ascribed to membrane components that form a microenvironment around the enzyme and thereby decrease aldehyde transfer to the enzyme and protect the enzyme from deactivation at the aqueous/organic interphase.

Product-enantiopurity and concentration
In the case of an enzyme that converts one enantiomer much faster than the other, the rst percentages of product will show high enantiopurity, whereas a decrease is expected in the course of the reaction (expressed in the Chen equation, 32). A dependence of product enantiopurity on reaction progress, especially at high enzyme loading, was experienced by us and others despite DKR conditions. Substrate racemization velocity was identi ed as main limitation towards high product enantiopurity (12,28). Here, substrate-to-catalyst ratio turned out as main factor determining product enantiopurity. The e.e. value of the product showed a strong dependence on substrate-to-catalyst ratio below 3.4 g substrate /g CDW . At 3.4 g substrate /g CDW product enantiopurities of ~ 94 % were obtained at 100 mM and 1 M 2-phenylpropanal concentrations and e.e. values were only slightly increasing at higher substrate-to-catalyst ratios (Fig. 4).
We previously stated a maximal catalyst loading of 40 to 50 g CDW /L in bioreductions to minimize product loss of hydrophobic substances in the biomass fraction during downstream processing (36). The minimally applicable substrate-to-biocatalyst ratio of 3.4 g substrate /g CDW gives a substrate loading of 1 to 1.25 M. Higher substrate-to-biocatalyst ratios (2 M substrate at 40 g CDW /L, or 1 M substrate at 20-30 g CDW /L) led to lower product formation. The catalyst was fully deactivated during the reaction time of 48 h (14). Elevated NAD + concentrations of 6-14 mM were used to fully exploit the catalytic activity of the coupled oxidoreductase system. In the present case, a compromise between enantiopurity and conversion had to be found. Limiting factors are substrate racemization velocity, catalyst loading and catalyst stability.

Conclusion
Optimization of an enzymatic dynamic reductive kinetic resolution of racemic 2-phenylpropanal towards 843 mM (S)-2-phenylpropanol was accomplished. The multilevel engineering included engineering of the wild-type CtXR for improved enzyme activity and enantioselectivity, use of an E. coli whole-cell catalyst for enzyme stabilization and coenzyme recycling and optimization of the substrate-to-catalyst ratio. Whole-cell level. The whole-cell catalyst based on D51A CtXR showed, much to our delight, high stability in the presence of 1 M phenylpropanal. Our results indicate a strong protecting effect by the whole-cell catalyst (and the supernatant thereof). We used lyophilized and rehydrated E. coli coexpressing D51A CtXR and yeast formate dehydrogenase as catalyst.
Reaction level. The most important factor to obtain high enantiopurities and product concentrations was the ratio of substrate to catalyst (Fig. 1, Fig. 4, Table 3). The amount of NAD + was the second most important factor. Catalyst and co-enzyme concentration directly affected reduction rate. High reduction velocity led to high conversions but also lower product enantiopurities. The conversion with a catalyst loading of 40 g CDW , 10 mM NAD + and 1 M substrate represented a compromise to obtain (S)-phenylpropanol in ~ 93 % e.e. enantiopurity and 843 mM product concentration (Table 4, entry 14).
We explain higher e.e. -values obtained with the whole biomass as compared to the supernatant by the presence of cells and cell debris (Fig. 1). The substrate was added in concentrations far above its solubility limit of 0.5 mM and hence formed a second phase. The presence of cellular components led to a very ne dispersion of the rac-2-phenylpropanal in form of an oil-in-water emulsion (37). Organic/aqueous phase mass transfer was facilitated and led to a scenario where used substrate (mainly S-form) was replenished fast by the racemic substrate. As a result, the (S)-aldehyde was constantly supplied and consumed, the (R)-aldehyde supplied and (slowly) racemized. Use of a stable and selective enzyme (D51A CtXR) enabled the synthesis of a highly enantiopure product under these conditions.

Materials And Methods
Chemicals
Total enzyme activities of the biomass (extracellular and intracellular enzymes). Activities of the whole biomass (extracellular and intracellular enzymes), measured after cell lysis and protein extraction for xylose reductase and formate dehydrogenase were 1590 U/g CDW and 154 U/g CDW , respectively (14).
Enzyme activities of the supernatant (extracellular enzymes). The rehydrated biomass was centrifuged and the activities of xylose reductase and formate dehydrogenase that had leaked out of the cells were determined to 512 U/g CDW and 72 U/g CDW , respectively, in the supernatant. The procedures for activity determination were described earlier (14). Kinetic parameters of NADH-dependent 2-phenylpropanal reduction by CtXR variants were determined spectrophotometrically as described earlier (19). Solubility of 2-phenylpropanal in water was elevated to 0.5 mM by addition of 25 % v/v DMSO. Substrate solutions were freshly prepared and immediately used to avoid non-enzymatic decomposition and, with (S)-2-phenylpropanal, racemization in aqueous solution. A typical measurement period was 5 minutes. Non-speci c background activity was considered by measuring blank activities. The added DMSO had no effect on the enzyme's activity with the natural substrate D-xylose.

Bioreductions of racemic 2-phenylpropanal
Reduction by the isolated D51A CtXR 25 %. The substrate ( nal concentration 0.5 mM) was incubated at 25°C in the presence of 0.1 mM excess [NADH] and D51A CtXR for 2 hours. For timecourse analysis, samples (100 µL) were taken from the reaction mixtures ( nal reaction volume 1.5 mL) at speci ed timepoints. All samples were diluted 1:1 with acetonitrile and centrifuged prior to analysis by chiral HPLC.

Whole-cell bioreduction
Lyophilized cells were re-hydrated in potassium phosphate buffer (100 mM, pH 6.2, re-hydration volume ≤ 50 % v/v of the total bioreduction mixture) in the presence of NAD + (3-14 mM) and sodium formate (50 mM excess on [substrate]) using 2 mL Eppendorf tubes. The tubes were placed on a thermomixer for 30 min at 25°C and 1400 rpm. In case of HBC-aided conversions, cyclodextrins were weighed out separately in Eppendorf tubes followed by adding rac-2-phenylpropanal and 50 µL buffer. Tubes were vortexed vigorously for complexation of substrate and HBC. Afterwards, rehydrated cells were combined with HBC/substrate complexes and lled up to a total working volume of 1 mL. Alternatively, substrate was added directly to the re-hydrated cells if no HBC was applied. Eppendorf tubes were sealed with para lm and vortexed until emulsi cation was reached. The mixtures were reacted for 24 or 48 h at room temperature using an end-over-end rotator (30 rpm). All samples were prepared in duplicates.

Analytical methods
Sample preparation from whole-cell or cell-free bioreductions  NMR analysis 1 H-NMR spectra of isolated substrate/product from biotransformations (78 % conversion) were recorded using a 300 MHz Bruker NMR unit (300 MHz for 1 H) at 300 K. Chemical shifts (δ) were depicted in ppm relative to the resonance of the solvent (MeOD or DMSO-d 6 ).

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.
Availability of data and materials The datasets used and/or analysed during the current study are available in the Supplementary data and from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.    Table 3).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementarydataphprop1712.docx