In this study, 6 model substrates have been selected (Fig. 1), including primary and secondary alcohols as well as diols. In addition, isopropanol, pure (R)-1, and pure (S)–1 were used in selected experiments. All screenings were performed in 96-plates (plate layouts and the corresponding screening results for every substrate/enzyme class are reported in the Supplementary Information 1, Sections 1, 5 and 7).
Substrates used in this study.
Ketoreductase screening
NAD(P)H absorbance at 340 nm is commonly used to measure the activity of KREDs18,19. Unfortunately, this approach is generally not suitable for high-throughput methods due to background noise deriving from the cell lysate. Colorimetric assays solve most of these problems and are quite amenable to high-throughput screening. Tetrazolium salts (e.g. nitroblue tetrazolium) were widely used for the determination of NAD(P)H, however, their application is limited by the low solubility of the formazan product20. Using a commercially available kit typically employed for cell viability (CCK-8), a quick screening of the proprietary Roche KRED library was carried out (220 KREDs). All the samples were used as cell-free extracts (CFEs); no enzyme purification was carried out. CCK-8 kit allows a convenient assay based on WST-8 (2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt). WST-8 in the presence of an electron mediator such as 1-methoxy-5-methylphenazium methylsulfate (1-mPMS), is readily reduced by NAD(P)H to produce an orange water-soluble formazan product, which can be detected by monitoring the absorbance at 450 nm (Fig. 2).
Schematic representation of the reaction for the detection of KRED activity using colorimetric WST-8 assay.
Among the advantages, CCK-8 is commercially available as single solution to add directly to the samples, no pre-mixing of components is required. Moreover, the detection sensitivity is higher than any other tetrazolium salts21. Although the potential of this assay was already described by Chamchoy and coworkers for the quantification of the activity of two model NAD(P)+—dependent dehydrogenases21, here the screening was tested with a sizable library of 220 different enzymes. Moreover, whereas the human G6DP (glucose-6-phosphate dehydrogenase) and the SDR (short chain dehydrogenase) from Burkholderia pseudomallei presented by Chamchoy et al.21 were employed in the reduction direction and required an enzyme-coupled approach for NA(P)H recycling, here the test reagents could be directly employed as regeneration system for the NAD(P)+ cofactor involved in the oxidation reactions (Fig. 2), which could therefore be used in low concentration (1:10 cofactor:substrate).
KRED library
220 KREDs previously expressed in E. coli strains were prepared at Roche as CFEs and tested towards the 6 substrates in Fig. 1. The catalysts were divided in 3 groups: NAD+-, NADP+-dependent and Cofactor Undefined. The cofactor undefined plates were screened by adding separately NAD+ or NADP+ to elucidate their preferential cofactor dependency (See Supplementary Information 1, Section 1). Absorbance was monitored at different reaction times; 30 min, 2 h and 24 h, which allowed for the quick identification of the best hits in terms of activity and reaction rate for each substrate. As an example NADP+-dependent KREDs assayed with a racemic mixture of substrate 1 after 24 h incubation time are shown in Fig. 3, illustrating that the most active enzymes can clearly be distinguished. For full results with 1 and with the remaining substrates see Supplementary Information 1, Section 1. A handful of hits were found for all six substrates and the activity could be confirmed by HPLC analysis (See Supplementary Information 1, Section 2). The KRED library was also screened for the oxidation of isopropanol to form acetone. If this reaction is sufficiently reversible, acetone can be used as a sacrificial substrate for the regeneration of the cofactor in preparative-scale oxidation reactions.
KRED assay at 24 h; A5: KP00104, A6: KP00127, A7: KP00128, A8: KP00129; B5: KP00138, B6: KP00139, B7: KP00140, B8: KP00141, C5: KP00150, C6: KP00151, C7: KP00152, C8: KP00153. Increased orange color intensity corresponds to higher absorbance values and consequently to a more active enzyme.
Evaluation of the stereoselectivity using (R)-1 or (S)-1
For the evaluation of the stereoselectivity of the Roche KRED collection (NAD+, NADP+ and Cofactor Undefined enzymes), substrate 1 was used. A comparison between the results obtained by the addition of the racemic substrate 1 and the pure enantiomers ((R)-1 and (S)-) 1) was subsequently carried out for each plate. As an example, the active NADP+-dependent KREDs (Fig. 4) with one of the enantiomers and both of them at 30 min and 24 h are reported in Table 2. While enzymes in position C5, C8, C10, D1 and D4 are selective for (S)-1-phenylethanol (results at 24 h), only C5, C8 and D4 demonstrated a higher reaction rate (increased orange color and absorbance at 30 min). The same comparison was performed for (R)-selective KREDs and enzymes active on both enantiomers. Interestingly, enzymes A5 and C6 showed a stereopreference for (R)–1 and (S)–1 respectively at the beginning of the oxidation reaction, but by 24 h both the enantiomers were converted. B11 is the only enzyme showing high initial activity with both enantiomers. A full overview of KREDs’ screening is reported in the Supplementary Information 1, Section 3.
Comparison at 30 min and 24 h between results obtained by the addition of the racemic substrate 1 and the pure enantiomers ((R)–1 and (S)–1). Increased orange color intensity corresponds to higher absorbance values (reported in the cells) and consequently to a more active enzyme. Ovrflw: absorbance values over 3 (corresponding to the limit of the instrument used for the detection).
Alcohol oxidases
The activity of commercial AlcOXs, 10 from Gecco and 2 from Sigma Aldrich (Supplementary Information 1, Section 5) were assayed using a commercial kit (MAK 311) for the detection of H2O2, formed during alcohol oxidation. It provides a simple and adaptable assay for quantitative determination of H2O2 concentration without any sample pre-treatment. This test utilizes the chromogenic Fe3+—xylenol orange reaction, in which a purple complex is formed when Fe2+ is oxidized to Fe3+ by peroxides present in the sample, generating a colorimetric (585 nm) result, proportional to the level of peroxide present. While this assay was described for the evaluation of para-phenol oxidases, the catalysts were purified22. No data with complex CFEs were reported so far. Moreover, the optimized formulation reduces interference by substances in the crude enzyme samples.
The best conditions were found using GLUO (glucose oxidase from Aspergillus niger) with substrates 2 and 5 (1 mM) giving 32% and 7% molar conversion respectively (Table 3). These data were obtained by fitting the sample absorbance values (duplicate) to a calibration curve previously established following the kit instructions (Supplementary Information 1, Section 6). A reaction with higher substrate loading (10 mM), but unchanged enzyme concentration (1 mg/mL), and reaction time (2 h), was subsequently carried out. Based on the calibration curve, 16% and 9% molar conversion were observed for substrate 2 and 5 respectively.
Due to lack of reactivity of the 10 Gecco enzymes towards the selected substrates, further investigations using the natural substrates eugenol and vanillyl alcohol of EUGO (Eugenol oxidase Rhodococcus jostii) and VAO (Vanillyl alcohol oxidase Penicillium simplicissimum) (Supplementary Information 1, Section 5) were carried out at 1 mM and 10 mM scale. In all cases full conversion was obtained between 2 and 24 h. It was suspected that possible catalase-mediated by-reactions (the Gecco enzymes are sold as E. coli CFEs) could interfere with the kit, rapidly converting the H2O2 produced by the oxidases and therefore making the detection highly unreliable. The reactions (all 6 substrates and 10 Gecco enzymes) at 1 mM scale were then monitored by HPLC (Supplementary Information 1, Section 4). 10% molar conversion was observed with substrate 2 and HMFO (5-Hydroxymethylfurfural oxidase Methylovorus strain) after 2 h (stable after 24 h). For substrate 5, 22% molar conversion was obtained with NICO (6-Hydroxy-d-nicotine oxidase Arthrobacter nicotinovorans). None of the other substrates gave any detectable conversion.
Laccase-mediator systems
For the laccase-mediator systems, a single enzyme laccase C from Trametes sp. was screened with different mediators (Supplementary Information 1, Section 7) and pH values (4.5–6.0). Different reaction conditions were found to be active for the oxidation of substrate 1 (90–93% molar conversion using AZADO as mediator at all pH values), 2 (60% molar conversion using TEMPO, pH 6) and 3 (65% molar conversion using AZADO, pH 4.5), no clear results have been obtained for substrate 4 and 6, probably due over-oxidation phenomena, typically observed with the use of LMSs. No oxidation reaction was observed for substrate 5.
