Discovery of New Carbonyl Reductases Using Functional Metagenomics and Applications in Biocatalysis

Article abstract of DOI:10.1002/adsc.202100199

Enzyme discovery for use in the manufacture of chemicals, requiring high stereoselectivities, continues to be an important avenue of research. Here, a sequence directed metagenomics approach is described to identify short chain carbonyl reductases. PCR from a metagenomic template generated 37 enzymes, with an average 25% sequence identity, twelve of which showed interesting activities in initial screens. Six of the most productive enzymes were then tested against a panel of 21 substrates, including bulkier substrates that have been noted as challenging in biocatalytic reductions. Two enzymes were selected for further studies with the Wieland Miescher ketone. Notably, enzyme SDR-17, when co-expressed with a co-factor recycling system produced the anti-(4aR,5S) isomer in excellent isolated yields of 89% and 99% e.e. These results demonstrate the viability of a sequence directed metagenomics approach for the identification of multiple homologous sequences with low similarity, that can yield highly stereoselective enzymes with applicability in industrial biocatalysis. (Figure presented.).

Full text of DOI:10.1002/adsc.202100199

FULL PAPER
DOI: 10.1002/adsc.202100199
Discovery of New Carbonyl Reductases Using Functional
Metagenomics and Applications in Biocatalysis
a
b
c, d
b,
Sophie A. Newgas, Jack W. E. Jeffries, Thomas S. Moody, John M. Ward, *
a,
and Helen C. Hailes *
a
Department of Chemistry, University College London,
2
0 Gordon Street, London, WC1H 0AJ, U.K.
E-mail: h.c.hailes@ucl.ac.uk
Department of Biochemical Engineering, Bernard Katz Building,
University College London,
London, WC1E 6BT, U.K.
E-mail: j.ward@ucl.ac.uk
Almac Sciences, Department of Biocatalysis and Isotope Chemistry,
Almac House, 20 Seagoe Industrial Estate,
Craigavon, BT63 5QD,
Northern Ireland, U.K.
Arran Chemical Company, Unit1 Monksland Industrial Estate,
Athlone, N37 DN24, Co. Roscommon, Ireland.
b
c
d
Manuscript received: February 11, 2021; Revised manuscript received: March 22, 2021;
Version of record online: ■■■, ■■■■
Supporting information for this article is available on the WWW under https://doi.org/10.1002/adsc.202100199
©
2021 The Authors. Advanced Synthesis & Catalysis published by Wiley-VCH GmbH. This is an open access article under the
terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
Abstract: Enzyme discovery for use in the manufacture of chemicals, requiring high stereoselectivities,
continues to be an important avenue of research. Here, a sequence directed metagenomics approach is
described to identify short chain carbonyl reductases. PCR from a metagenomic template generated 37
enzymes, with an average 25% sequence identity, twelve of which showed interesting activities in initial
screens. Six of the most productive enzymes were then tested against a panel of 21 substrates, including
bulkier substrates that have been noted as challenging in biocatalytic reductions. Two enzymes were selected
for further studies with the Wieland Miescher ketone. Notably, enzyme SDR-17, when co-expressed with a co-
factor recycling system produced the anti-(4aR,5S) isomer in excellent isolated yields of 89% and 99% e.e.
These results demonstrate the viability of a sequence directed metagenomics approach for the identification of
multiple homologous sequences with low similarity, that can yield highly stereoselective enzymes with
applicability in industrial biocatalysis.
Keywords: biocatalysis; carbonyl reductases; short-chain reductases; Wieland-Meischer ketone; functional
metagenomics
Introduction
not require toxic transition metals, organic solvents or
extreme temperatures and can achieve excellent che-
In recent years the use of biocatalysts has continued to mo- and stereoselectivities. Biocatalysts are now
grow and play an important role in the fine chemical, recognised industrially to provide a sustainable alter-
[1]
pharmaceutical, food, biofuel and waste industries. native to conventional chemical catalysts. It is for
Compared to traditional organic chemistry strategies, these reasons that the market in industrial enzyme use
[2]
biocatalysts typically use mild reaction conditions, do is predicted to grow to £10 billion by 2024.
Adv. Synth. Catal. 2021, 363, 1–10
1
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
Carbonyl reductases, ketoreductases, and alcohol assay enzymes expressed from their environmental hosts
dehydrogenases (ADHs) (EC 1.1.1) cover a large or as a recombinant protein in a more readily culturable
[16–18]
group of enzymes that reduce a ketone or an aldehyde host.
The enzyme functions are then experimentally
group to an alcohol, or perform the reverse reaction, determined, which requires high throughput assays which
using either NADH or NADPH as a cofactor. One can be time consuming and costly. For example, in an
class of ADHs known as short chain dehydrogenases/ interesting recent publication by Popovic et al. the
reductases (SDRs) form a diverse family of proteins, activity of over 1 million fosmid and Lambda-ZAP
which all display similar α/β folding patterns with a clones from 16 different environments was screened,
[3]
central β-sheet, typical of the Rossmann-fold. With generating 714 positive hits, 80 of which displayed
[19]
oxidation and reduction accounting for the second validated esterase activity. Another commonly used
largest number of studies on biocatalysis, SDRs are an strategy in metagenomic studies uses degenerate primers
[4]
important area of focus. More importantly SDRs also to amplify gene sequences of a target enzyme family
provide an efficient route to single enantiomer from environmental samples. This can generate a larger
[2]
alcohols. In the pharmaceutical industry the produc- number of enzymes per clones screened, however the
tion of single isomer drugs is becoming ever more enzymes often exhibit high sequence similarity to each
[
20]
important, in 2015 all chiral drugs approved for sale by other.
the FDA bar one, were single enantiomers.
[5,6]
The approach presented here follows our previously
One of the limitations of using SDRs is a lack of described metagenomics strategy, combining in silico
enzymes that can accept large lipophilic substrates or protein function prediction with sequence specific
carbonyl moieties flanked by sterically challenging PCR. This methodology increases the rate of positive
groups, as well as the use of the expensive cofactor hits that display the desired functionality, whilst at the
NAD(P)H. Most SDRs preferentially generate the (S)- same time delivering hits with comparatively low
alcohol in line with Prelog’s rule and the limited sequence similarity, which increases the chance of
number of anti-Prelog enzymes, which will generate finding enzymes with different characteristics. Follow-
[7,8]
the (R)-alcohol, is another stumbling block.
In ing this approach, 37 enzymes were expressed with
addition, non-engineered biocatalysts can suffer from low average sequence identity of 20–40%. Active
low organic solvent tolerances, narrow working pH enzymes were investigated for their substrate selectiv-
ranges, and low thermostabilities. The discovery of ity towards twenty substrates, including the Wieland
new enzymes is a key strategy to overcome these Miescher ketone (WMK), a precursor to a wide
issues. Identifying enzymes from metagenomic sour- selection of pharmaceutically relevant products. Se-
ces, where DNA is extracted from an environmental lected enzymes from the metagenome were then used
sample and analysed using high throughput sequencing with the bicyclic WMK and co-expressed with a
technologies, has proved to be a valuable method. cofactor recycling enzyme, glucose-6-phosphate dehy-
With the increasing availability and the decreasing cost drogenase, to stereoselectively reduce the (R)-WMK,
of sequencing methods, there has been a rise in studies in excellent yields.
[9–11]
using functional metagenomics.
In recent work, we have utilised this approach by
collecting metagenomic material from the human oral Results and Discussion
cavity and a domestic drain, sequencing the DNA and
assembling the shorter reads into larger contiguous
Metagenome Mining and Isolation of SDR Genes
stretches of DNA (contigs), creating an in silico To identify potential SDR genes an in silico tongue
library. Using this library, open reading frames, metagenomic library was mined for enzymes using a
[
12,13]
operons and enzymes can be identified and this method we have recently described.
approach has been used to discover new transketolases, libraries are a viable resource for the identification and
Such in silico
[9,10,12,13]
ene-reductases and transaminases.
After using retrieval of enzymes using both BLAST and Pfam
[21,22]
PCR to amplify identified DNA, the enzymes were based search methodologies.
The Pfam ID for
cloned, and assayed for their functionality. Metage- adh_short_C2 (PF13561.6) used to search the meta-
nomics has now emerged as a very effective tool to genome was generated when querying the Pfam data-
mine for non-redundant genes, creating panels of base using amino acid sequences of SDRs from
enzymes from the same families with low sequence Lactobacillus brevis (Uniprot ID Q84EX5), Lactoba-
[14,15]
homologies.
cillus kefiri (Uniprot ID Q6WVP7), and Weissella
A sequence directed strategy increases the emphasis thailandensis (Uniprot ID G0UH95). These were
on in silico screening compared to other methods, using identified as sequences of interest as they exhibit
[23,24]
bioinformatics and enzyme functionality prediction to preference for (R)-alcohols.
The sequences were
enhance the likelihood of the desired activity being also used in a BLAST search of the metagenomic
displayed in the expressed enzymes. Alternative ap- library. In total 139 open reading frames (ORFs) were
proaches incorporating metagenomic strategies directly identified.
Adv. Synth. Catal. 2021, 363, 1–10
2
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
From these ORFs, sequences were selected if they described as NAD(P)H dependant oxidoreductases and
had an initiator methionine and stop codon and were at again as oxidoreductases. The clade in purple, contains
least 230–250 amino acids in length. They were only enzymes which were identified as NAD(P)H
clustered based on similarity and one from each cluster dependant oxidoreductases or oxidoreductases.
was chosen, giving 38 non redundant sequences, that
Three enzymes were given unique assignments,
were taken forward for cloning using Gibson assembly. SDR-4, 6 and 30 labelled in red. SDR-4 was identified
All but one sequence, SDR-14 were successfully as an acetoin reductase, SDR-6 as a glucose-1-
amplified by PCR. Multiple sequence alignments of dehydrogenase and SDR-30 as an enoyl ACP reduc-
the 37 successfully retrieved SDRs showed sequence tase. Enzymes in the same clade as SDR-30, in orange,
identity ranging from 11% to 98%, with most enzymes were identified as oxidoreductases. It is possible that
having identity to each other of between 20–40% enzymes in this clade with high similarity, SDR-25 and
(Figure 1). Some groups of enzymes had above 27 could also be enoyl ACP reductases. Other enzymes
average similarity suggesting that they may have in the clade but with lower similarity could similarly
similar properties such as: SDR-10 and SDR-13, 51%; be enoyl ACP reductases, such as SDR-35 or act on
SDR-5, 18 and 37, ~60%; SDR-1 and SDR-21, 67%; substrates with similar structure such as SDR-4 the
SDR-25, 27 and 30, ~70%; SDR-3, SDR-29, 81% and acetoin reductase (Figure 1). Taxonomic assignments
SDR-20 and SDR-36, 98%.
came from multiple genera, all matched to known
The protein and DNA sequences of the SDRs were human oral commensals. Streptococcus, Porphyromo-
searched against the NCBI database by BLAST. This nas, Actinomyces, Haemophilus, Neiserria, Prevotella,
generated predicted functions and taxonomic assign- Rothia and Veillonella species were the source of
ments. Enzyme functional assignments broadly fell multiple enzymes. Named species from these genera
into 3 groups, 3-oxoacyl acyl carrier protein (ACP) were the source of multiple enzymes, however of these
reductases, NAD(P)H dependant oxidoreductases and multiple sequences, no more than one was mapped to
oxidoreductases: SDR numbers are labelled in green, the clade in green from any single species. Rothia
purple and blue respectively in Figure 1. The clade in mucilaginosa was the source of 4 enzymes SDR-9, 10,
green contains all the enzymes identified as 3-oxoacyl 11 and 16: SDR-10 and SDR-11 map to the clade in
ACP reductases but also those identified as oxidor- purple while SDR-16 maps to the clade in green. This
eductases, SDR-7, 15, 31 and a NAD(P)H dependant pattern is repeated for S. parasanguinis with SDR-5
oxidoreductase SDR-28. It is possible that these SDRs and SDR-24 mapping to the purple clade and SDR-17
are also 3-oxoacyl ACP reductases. This is possible as to the green clade (Figure 1). Atopobium parvulum,
the three assignments are nested terms, that is, 3- Megasphaera micronuciformis, Oribacterium sinus
oxoacyl ACP reductases can be less specifically and Mycobacteriodes abscessus were the source of a
Figure 1. A heat map and phylogenetic tree showing the relationships and sequence identity of the metagenome derived SDRs.
SDR numbers on the left are coloured to reflect their functional identification, SDR numbers in green were identified as 3-oxo acyl
ACP reductases, purple as NAD(P)H dependant oxidoreductases, blue as oxidoreductases and red were given unique functional
assignments. Branches belonging to the same clade were coloured orange, green and purple.
Adv. Synth. Catal. 2021, 363, 1–10
3
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
single enzyme. The full taxonomic and functional with the same co-factor. With NADPH, 1 was accepted
assignments along with DNA and amino acid sequen- by SDR-31 and SDR-37, 2 predominantly by SDRs-3,
ces are given in the supplementary information (SI, 11, 17, 23 and 37, and 3 by SDRs-11, 16, 17 31 and
Tables S1-3).
SDR-37. The most active enzymes displayed low
sequence identity to each other in the range 20–40%
(
Figure 1).
Initial Screening of the SDR Enzymes
Overall, from the 37 enzymes investigated, the
The 37 successfully cloned SDRs were screened as success rate of finding an active enzyme towards the
clarified cell lysates against acetophenone 1, methyl compounds tested was approximately 20%. Compared
acetoacetate 2 and cyclohexanone 3, using an NAD(P) to related studies this was a higher hit rate and the
[25]
H spectrophotometric assay (Scheme 1. A), to high- retrieved enzymes were less homologous. For example,
light initial activities. Notably, the SDRs had a clear Itoh et al. used degenerate primers to identify ADHs
preference for either NADPH or NADH. For example, from metagenomic material generated from 17 soil
acetophenone 1 was well accepted by SDR-4 and samples. They identified 240 putative ADH-positive
SDR-24 with NADH, while methyl acetoacetate 2 had clones via colony PCR from 2800 colonies and only
low levels of acceptance by SDR-4 and SDR-22, and 10% functional ADH genes were collected that were
cyclohexanone 3 was accepted by SDR-4 and SDR-36
Scheme 1. A. Spectrophotometric carbonyl reductase assay monitoring the consumption of NAD(P)H at 340 nm. The change in
absorption (340 nm) after a 100 min incubation of substrates 1–3 with the 37 SDRs using either NADH (top) or NADPH (bottom)
as co-factor. Reaction conditions: substrate 1, 2 or 3 (5 mM) NAD(P)H (1 mM) (200 μL), clarified cell lysate (0.4 mg/mL), KPi
(
100 mM, pH 7.2), DMSO (10%, v/v). The reactions were shaken for 100 min, at 25°C, and quantified using the spectrophotometer
at 340 nm. B. The change in absorption (340 nm) after a 100 min incubation of substrates 1–21 with SDRs-3, 4, 11, 17, 31 and 37
using either NADPH or NADH as co-factor. Reaction conditions: As for A and reactions were performed in triplicate, with an
average standard deviation of �0.05%.
Adv. Synth. Catal. 2021, 363, 1–10
4
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
very similar at the amino acid sequence level residue considered to be important. The TGxxxGxG
[20]
>
90%.
motif important in cofactor binding was also present
SI Figure S1). Sequence analysis identified SDR-4 as
(
an acetoin reductase originating from M. abseccus, as
well as having 99% sequence identity to a reported
acetoin reductase from several Streptococcus species.
Screening of Selected SDRs for Biocatalytic Appli-
cations
From the panel of enzymes, based upon the prelimi- Such reductases have been shown in the literature to
nary successful activity data and levels of expression, reduce diacetyl and 2,3-pentane-dione as well as
[26]
SDRs-3, 11, 17, 31 and SDR-37 were then screened acetoin. However, they have not been reported to
against a larger substrate set, 1–21, with NADPH and readily accept other linear ketones such as 2-heptanone
SDR-4 with NADH, again with the spectrophotometric 21 or aromatic or cyclic ketones as noted here. SDR-
assay as described above. The results are summarised 17 was identified as a 3-oxoacyl-ACP reductase
in Scheme 1B.
involved in fatty acid synthesis and SDR-31 as an
Reductases SDR-17, SDR-31 and SDR-37 were the oxidoreductase, although it was placed in the same
best performing enzymes, demonstrating high conver- clade as SDR-17 in the phylogenetic tree constructed
sions on many of the substrates (up to 69%). The from all 37 sequences (Figure 1), suggesting it too may
cyclohexanone ring systems (3-9), were readily ac- be a 3-oxoacyl-ACP reductase. 3-Oxoacyl-ACP reduc-
cepted by SDR-17, SDR-31, SDR-37 and SDR-4. tases have been shown to exhibit a wide specificity
Interestingly, the addition of a methyl group at C-2 or with respect to the chain length of the β-ketoacyl
[27]
C-3 in 4 and 5 increased the acceptance by SDR-4 and group. SDR-17 and SDR-31 also had wide substrate
SDR-17 compared to cyclohexanone 3. For example, promiscuity, although no β-ketoacyl-ACP compounds
conversions increased from very low levels with 3 to were screened in this study. SDR-37 although being
4
0% and 26% with SDR-4 and 4 and 5 respectively. identified as an NAD(P)H oxidoreductase and having
For SDR-17, conversions increased from 11% with 3 similar substrate profiles to SDR-17 and 31, has a low
to ~55% with 4 and 5. None of the enzymes screened sequence identity of ~24% with these enzymes and
showed much activity towards 2-cyclohexen-1-one 6, was placed in a different clade in the phylogenetic tree
suggesting that they were poor acceptors of α,β- (Figure 1).
unsaturated carbonyl functionalisation. Piperidone sys-
Considering the enzymes, in general SDR-3 ac-
tems 7, 8, and 9, were generally well accepted by the cepted aromatic ketones, but surprisingly the cyclic
SDRs enzymes with up to 65% conversions. For the ketone 8 at high levels. SDR-3 was also the only
aromatic ketones, 1 and 10–15, 1 was only readily enzyme to accept substrates 16 and 20. SDR-11, was
accepted by SDR-4 as indicated above, with 10 also active towards fewer compounds but tolerated
accepted by SDR-31, and ketone 11 only readily some of the substrates that were poorly accepted by
accepted by SDR-11. Substrates 12–14 were however, the other tested enzymes, such as 2 and 11. SDR-17,
generally well accepted by SDRs-3, 17, 31, 37 and SDR-31 and SDR-37 had the best substrate tolerances
SDR-4 (>50% for SDRs-17, 31 and SDR-37), towards the panel of compounds, while SDR-4
although 13 had the lowest activity. While these all accepted approximately half of the substrates.
possess aryl ketone-CF groups, 13 has a para-F
3
moiety which may result in unfavourable interactions
Development of the WMK Substrate with Selected
in the active site. In 15, the electron donating para-N-
SDRs
propyl group will make the carbonyl carbon less
electron deficient, reducing its reactivity, however The WMK 18 is a precursor to many natural products
unfavourable steric effects may have also lowered and analogues which have medical applications as
conversions with all the enzymes. For the bicyclic anti-inflammatory,
anti-fungal
and
anti-cancer
[28]
ketones, 1-indanone 16 was only accepted at low levels treatments. The regioselective reduction at the C-1
by SDR-3, while 17 was not accepted at all. Following carbonyl in 18 has been reported using sodium or zinc
[29,30]
these results, it was particularly interesting that the borohydride.
For example, the reduction of S-18
rac-WMK 18 was well accepted by SDRs-17, 31 and using sodium borohydride was incorporated into the
SDR-37. Of the aldehydes tested, benzaldehyde 19 total synthesis of baccatin III to give (4aS,5S)-22 with
showed similar conversions to the aromatic ketones a stereoselective reduction at the least hindered face of
with SDR-17, SDR-31, and SDR-37, but phenyl- the bicycle, resulting in a syn-relationship between the
[31]
acetaldehyde 20 was poorly accepted. For the linear methyl group at C-4a and C-5 OH. There have been
substrate 21 it was accepted most by SDR-11 as was 2. limited investigations into the use of biocatalysts to
For these six SDRs showing the most interesting perform the reduction of 18 with varying yields and
activities, the average sequence identity is 28%. stereoselectivities. Notably, whole cell experiments
Sequence alignment confirmed that the active site with the yeasts Torulaspora delbrueckii which prefer-
catalytic triad Ser-Tyr-Lys was conserved plus a fourth entially reduced (S)-18 (70% e.e.) to give syn-(4aS,5S)-
Adv. Synth. Catal. 2021, 363, 1–10
5
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
2
2 in 26% yield, while Candida melibiosica reduced displayed lower activity towards (S)-18, while SDR-31
[32]
(
R)-18 to give anti-(4aR,5S)-22 in 31% yield.
Baker’s yeast has also been utilised to reduce 18 and ence for the (S)-isomer.
gave syn-(4aS,5S)-22 as the major product in 32% When performing non-whole cell biocatalytic reac-
and tions that require co-factors either these must be
readily accepted both (S)- and (R)-18, with a prefer-
[33]
yield together with 4% of anti-(4aR,5S)-22,
using Coryneum betulinum or Didymosphaeria ignia- supplied directly, which is possible on small scales, or
ria cultures also preferentially converted (S)-18 into co-factor recycling systems are required. These can be
syn-(4aS,5S)-22: reduction of (R)-18 was slow and produced separately, however co-expression can en-
required 4–6 days to give anti-(4aR,5S)-22 in only hance the ease of use. Glucose-6-phosphate dehydro-
[34]
moderate diastereoselectivities (46-64% d.e.). Other genase (G-6-PDH) was selected as a co-factor recy-
+
approaches have involved enzymatic resolutions of cling system due to its high activity towards NADP .
[35]
acetylated-22 with commercial lipases. It was hoped Co-expression was successful with both SDR-17 and
that following the promising levels of reactivity of SDR-31 and using rac-18 quantitative conversions
+
SDR-17 and SDR-31 towards 18, the metagenomic- were achieved with NADP at 37°C after 24 h.
derived recombinant enzymes could be used to Notably, SDR-17 was very stereoselective and only
selectively produce isomers of 22 (Scheme 2), partic- produced the S-enantiomer at C-5, syn-(4aS,5S)-22 and
ularly those that are more difficult to access.
anti-(4aR,5S)-22, and the time course is shown in
Initially, studies were performed using SDR-17 and Scheme 2. This revealed the more rapid consumption
SDR-31 with either (S)-18 or (R)-18 at different pHs of (R)-18. Moreover, under the same conditions SDR-
using the spectrophotometric assay. Both SDR-17 and 31 fully reduced rac-18 after 24 h, with the (S)-isomer
SDR-31 were more active at pH 7–8 and this was used converted into syn-(4aS,5S)-22 only and the (R)-isomer
in later reactions (SI Figure S2). In addition, SDR-17 generating mostly anti-(4aR,5S)-22 with a small
amount of syn-(4aR,5R)-22 in a ratio of ~6:5:1,
respectively. Reaction stereoselectivities were deter-
mined by the reduction of (R)-18 and (S)-18 using
sodium borohydride to give (predominantly) the syn-
[30]
products as previously described. A combination of
chiral HPLC and NMR spectroscopy confirmed the
absolute stereochemistry in the products.
To explore the utility of the enzyme system further,
SDR-17 co-expressed with G-6-PDH (as a crude cell
lysate) was used to perform a preparative enzyme scale
reaction using (R)-18 (20 mM substrate, 50 mL reac-
tion) which generated anti-(4aR,5S)-22 in 89% isolated
yield after 24 h and >99% e.e. (at C-5). Notably, the
product was readily isolated in high purity without the
use of chromatographic methods, by extraction with
ethyl acetate. The reaction was highly stereoselective
giving the anti-product which is less readily produced
using traditional borohydride reducing reagents. Since
the SDR has been successfully co-expressed with a co-
factor recycling system, further scale-up has the
potential for delivery of an economical and efficient
bioreduction process for multiple carbonyl containing
substrates yielding the subsequent alcohol.
Conclusion
The metagenomic approach utilized here is an effective
tool for searching for new enzymes. Using bioinfor-
matics to predict reductase functionality from a
Scheme 2. Consumption of rac-18 with SDR-17 co-expressed
with G-6-PDH and production of 22. Reaction conditions:
(
500 μL volume): rac-18 (12 mM), clarified cell lysate (0.4 mg/
+
mL), NADP (3 mM), glucose-6-phosphate (G-6-P) (100 mM), searchable metagenomic library, yielded 37 enzymes
KPi (100 mM, pH 7.2), DMSO (10%, v/v). The reaction was of high sequence diversity (20–40% sequence identity).
shaken for 24 h at 37°C. Reactions were performed in duplicate. Compared with other metagenomic methods this is a
Red (S)-18, purple (R)-18, products blue syn-(4aS,5S)-22, black higher hit rate for active enzymes, whilst also
anti-(4aR,5S)-22, green anti-(4aS,5R)-22 and brown syn-
generating sequences with much lower overall
[19,20]
(4aR,5R)-22 were formed.
similarity.
Adv. Synth. Catal. 2021, 363, 1–10
6
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
From the 37 retrieved enzymes, expression and was used. Chemicals, media and apparatus were autoclaved
(
Priorclave) at 121°C for 30 min where required.
screening highlighted several (~20%) with interesting
activity profiles. The panel of 6 enzymes taken through
for further investigation showed good coverage against
the 21 substrates tested, including bulkier and more
lipophilic substrates that have been noted as challenging
for biocatalytic reductions. SDR-17 and SDR-31
showed interesting activities toward monocyclic ketones
and for the bicyclic ketone 18, a useful synthetic
precursor, the pH preference of the enzymes was also
established and the timecourse for the reaction monitored.
SDR-17 was then used for the reduction of (R)-18 in
high yields and excellent stereoselectivity at a 50 mL
scale. Moreover, the product anti-(4aR,5S)-22 was readily
isolated by extraction from the reaction media.
Cloning of SDR Sequences
Primers matching the 38 target sequences were designed and,
appended to these primers, were sequences identical to the
DNA sequence of the expression vector pET29a immediately
upstream and downstream of its multiple cloning site. The
pET29a sequences appended to the forward primers were
chosen to facilitate the insertion of the amplified genes with the
initiator methionine at the translation start site in the expression
vector. The reverse primers were designed to remove the
endogenous stop codon and allow read through of the hexa-
histidine tag encoded by the expression vector. Gibson assembly
(HiFi assembly NEB) was used to include the PCR products of
[7]
In this work one enzyme family, the SDRs were the SDR genes directly into pET29a plasmids. Amplifications
investigated, however there are other potential of genes from the metagenome and cloning into the vector were
successful for 37 out of 38 target sequences. Primers and
adapters for Gibson cloning are given in in the SI Table S4. The
recombinant vectors were then transformed into a chemically
competent cloning strain E. coli Top10 (Invitrogen) and a
chemically competent expression strain E. coli BL21*(DE3)
pLysS (Invitrogen) and stored as glycerol stocks.
carbonyl reducing enzyme families which could be
investigated, such as aldo keto reductases, alcohol
dehydrogenases, aldehyde dehydrogenases and iron
and zinc containing alcohol dehydrogenases. Expand-
ing the scope of enzyme families targeted and applying
the technique to more diverse environmental metage-
nomes, has the potential to generate many more
enzymes of interest.
SDR Growth, Purification and Desalting
Glycerol stocks of the expression strains were used to inoculate
Terrific Broth (TB) medium (Merck, 500 μL/well in a 96 deep-
square well plate) containing kanamycin (50 μg/mL) and
chloramphenicol (30 μg/mL) and incubated for 16 h, 37°C,
Experimental Section
General experimental
4
00 rpm (sealed with a microporous breathable membrane).
All reagents were obtained from commercial sources (Sigma
Aldrich, Fisher, Alfa Aesar) and used as received unless
otherwise stated. Silica column chromatography was performed
Then 100 μL of this starter culture was added to TB media
(
10 mL) containing kanamycin (50 μg/mL) and chlorampheni-
col (30 μg/mL). This was incubated for 6–8 h, until an OD600 of
.6–0.8 was achieved. IPTG (1 mM final concentration) was
®
using Geduran Si 60 Silica (43–60 μM). Thin layer chromatog-
0
raphy was performed using plates with a silica gel matrix on an
aluminium support. Infrared spectroscopy was carried out using
a Spectrum 100 FTIR spectrometer (Perkin-Elmer). H and
NMR spectra were obtained using an Avance 600 (Bruker)
spectrometer. Chemical shifts specified are relative to trimeth-
ylsilane (set at 0 ppm) and referenced to the residual, protonated
added and the mixture incubated for 16 h at 25°C at 400 rpm.
Cells were harvested by centrifugation (12000 rpm, 4°C, 5 min)
and resuspended in lysis buffer (1 mL) containing BugBuster
protein extraction solution (Novagen), DNAse I (20 μL/mL)
and lysozyme (1 μg/mL). The cell lysate was aliquotted into
20 μL volumes and flash frozen at À 80°C for storage. Where
applicable, NTAÀ Ni spin columns (QAIGEN) were used to
purify the protein, following the manufacturer’s instructions.
The protein was buffer exchanged into 0.1 M sodium phosphate
buffer pH 7.2 using Zebra Spin Desalting plates (Thermoscien-
tific) following the manufacturer’s instructions.
1
13
C
1
NMR solvent. Coupling constants in H NMR spectra (J) are
given in Hertz (Hz) and described as singlet (s), doublet (d),
multiplet (m). Mass spectroscopy was carried out using a
VG70-SE mass spectrometer Trace 1310 Gas Chromatograph
(
(
Thermoscientific) connected to a ISQ single quadrapole MS
Thermoscientific). Melting points were determined using
IA9000 Series melting point apparatus (Electrothermal). Ana-
lytical HPLC analysis was carried out using a Series 1100
SDS-PAGE Procedures
(Hewlett Packard) or 1260 (Aligent Technologies) instruments
with a HiChrom ACE C18 column (250 mm×4.6 mm) or
Chiralcel OJ column (250×4.6 mm).
The protein composition of induced cells was analysed by
sodium dodecyl sulfate-12% polyacrylamide gel electrophoresis
(
SDS-PAGE) using Mini-Protean TGX Gels (Bio-Rad). Protein
Sonication of cell lysates was performed using a probe Soniprep
samples were prepared by heating for 5 min at 95°C in the
presence of sample buffer (1:1 dilution of 2x Laemmli Sample
Buffer (Bio-Rad) and 100 mM dithiothreitol:protein). A broad
range protein marker (10–25 kDa, New England Biolabs) was
used to estimate the molecular mass of the proteins. Imperial
Protein Stain (Thermoscientific) was used to stain the protein
150 (MSE) sonicator. The following centrifuges were used:
Allegras x-15R centrifuge (Beckman Coulter), Centrifuge
5
5
415R (Eppendorf), Centrifuge 5810R (Eppendorf), Centrifuge
430R (Eppendorf). A ShakerX ClimoShaker ISFI-X (Kuhner),
Innova 44 (New Brunswick Scientific) or Mixing Block MB-
02 (BIOER) shaker was used. A GENios plate reader (Tecan)
1
(
SI Figure S3).
Adv. Synth. Catal. 2021, 363, 1–10
7
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
Spectrophotometric SDR Assay
4aR,5R:4aR,5S). R
ether); νmax (neat) 3407, 2932, 1719, 1614 cm ; H NMR
600 MHz; CDCl ) δ 5.80 (1H, d, J=1.9 Hz, 1-H), 3.44 (1H, br
f
0.30 (30% EtOAc in 40–60 petroleum
À 1
1
Activity data on the SDR reduction of ketones was determined
at room temperature by following the oxidation of NAD(P)H
using a GENios microplate reader (Tecan) at 340 nm over 100
cycles of 57 seconds with a shake duration of 1 s and a shake
settle time between cycles of 1 s. Each reaction mixture
(
3
d, J=11.7 Hz, 5-H), 2.31–2.50 (3H, m, 3-H , 8-HH), 2.16–2.26
2
(2H, m, 4-HH, 8-HH), 1.81–1.94 (3H, m, 4-HH, 6-HH, 7-HH),
1
.66–1.76 (1H, m, 6-HH), 1.37–1.48 (1H, m, 7-HH),1.21 (3H,
13
s, CH ); C NMR (150 MHz; CDCl ) δ 199.6, 168.3, 125.7,
3
3
(
5
(
200 μL) contained 10 mM sodium phosphate buffer (pH 7.2),
mM substrate, DMSO (10%, v/v) and the enzyme lysate
20 μL). The reaction was initiated by the addition of the
7
8.5, 41.7, 34.4, 33.8, 32.1, 30.4, 23.3, 15.4; m/z (EI) 180
+
([M] ). Analytical chiral HPLC retention time 61.2 min
Chiralcel OJ, propanol/hexane at 0.5 mL/min; SI Figure S4).
(
substrate. Protein concentrations were determined using a
Nanodrop 2000c spectrophotometer (Thermoscientific).
(4aR,5S)-5-Hydroxy-4a-methyl-4,4a,5,6,7,
8
-hexahydronaphthalen-2(3H)-one, (4aR,5S)-22
Co-Expression of SDRs and G-6-PDH
Freeze-dried cells of SDR-17 (25 mg) were suspended in buffer
Competent E. coli BL21 DE3 cells were transformed using
standard protocols with the SDR pET29a plasmid (3 μL) and G-
(
NaPi, 100 mM, 1 mL) and sonicated (10×15 s pulses) at 0°C
before being centrifuged (10 min, 3000 g, 4°C). The reaction was
6
-PDH from Saccharomyces cerevisiae SF838 (Uniprot
performed using a total volume of 50 mL, containing buffer (NaPi,
P11412) in a pACYCDuet-1 plasmid (3 μL). The SDR and G-6-
PDH were then expressed as previously described above.
100 mM, pH 7.2), (R)-18 (20 mM), clarified cell lysate (0.4 mg/
+
mL), NADP (3 mM), DMSO (10%, v/v) and G-6-P (100 mM),
and shaken for 24 h, 37°C, at 500 rpm. The reaction was stopped
with the addition of TFA (0.5% v/v), and the product was extracted
with EtOAc (5×15 mL) and washed with sat. NaCl solution (5×
Stereoselectivities and the WMK Reactions
Reaction conversion yields and enantiomeric excesses at the
reduced centre were determined by chiral HPLC using a
Chiracel OJ column. For 200 μL scale reactions, the assay
procedure above was followed, using flash-frozen clarified cell
lysate. For 500 μL scale reactions using co- expressed SDR-17
and SDR-31 with G-6-PDH, enzymes were expressed as
described above, and freeze-dried. Each reaction mixture
1
>
5 mL) to afford (4aR,5S)-22 as a yellow solid (0.160 g, 89%,
[37]
99% e.e. by HPLC analysis). M.p. 75–79 C [lit. 88–90 C ];
°
°
25
R 0.61 (33% EtOAc in 40–60 petroleum ether); [α] À 112 (c
f
D
25
[38]
1
.3, toluene) [lit. [α] À 111 (c 1.3, benzene)]; νmax (neat) 3427,
D
À 1
1
2963, 1638, 1600 cm ; H NMR (600 MHz; CDCl ) δ 5.87 (1H,
3
d, J=1.9 Hz, 1-H), 3.65 (1H, t, J=2.7 Hz, 5-H), 2.56–2.62 (1H,
m, 4-HH), 2.44–2.52 (2H, m, 3-H ), 2.39–2.44 (1H, m, 8-HH),
2
(500 μL) contained 10 mM sodium phosphate buffer (pH 7.2),
2.24–2.32 (1H, m, 8-HH), 2.01–2.09 (1H, m, 6-HH), 1.82–1.93
1
0 mM (R)- or (S)-18), DMSO (10%, v/v) and the 0.4 mg/mL
(1H, m, 7-HH), 1.76–1.82 (1H, m, 6-HH), 1.67–1.75 (1H, m, 7-
13
SDR enzyme. The reaction was initiated by the addition of the
substrate. The reaction was shaken (250 rpm, 24 h, 25°C). The
reaction was stopped by the addition of TFA (0.5% v/v) and
denatured protein was removed by centrifugation. Diethyl ether
HH), 1.47–1.53 (1H, m, 4-HH), 1.24 (3H, s, CH3); C NMR
150 MHz; CDCl ) δ 199.6, 168.0, 127.2, 75.4, 41.0, 34.1, 31.8,
(
3
+
3
0.9, 28.8, 21.9, 19.9; m/z (ES+) 181 ([M+H] ). Analytical
HPLC retention time 54.0 min (Chiralcel OJ, propanol/hexane at
.5 mL/min; SI Figure S6).
(
1 mL) was then added to the supernatant and the mixture
vortexed for 30 s. The organic layer was separated, dried
Na SO ) and evaporated. EtOH (200 μL) was then added and
0
(
2
4
the mixture analysed by analytical HPLC (Chiralcel OJ (4aS,5S)-5-Hydroxy-4a-methyl-4,4a,5,6,7,
column). Concentrations of substrates and products were 8-hexahydronaphthalen-2(3H)-one, (4aS,5S)-22
determined using calibration curves against standards, SI Fig-
[
39]
To a stirred solution of (S)-18 (0.030 g, 0.17 mmol) in EtOH (500
ure S4, (solvents: 6% 2-propanol/hexane at 0.5 mL/min flow-
rate, detection at 230 nm and a run time of 120 min). Retention
times: (R)-18 68.2 min; (S)-18 78.3 min; (4aS,5S)-22 85.9 min;
μL) at 0°C, NaBH (2.0 mg, 0.057 mmol) was added and the
4
reaction stirred for 15 min. The solvents were evaporated in vacuo,
and the mixture extracted with EtOAc (3×10 mL) and washed
with sat. NaCl solution (2×10 mL). The organic layer was dried
(4aR,5R)-22 61.2 min; (4aR,5S)-22 54.0 min.
(
MgSO ) and concentrated in vacuo. The product was purified via
4
flash silica column chromatography (50% EtOAc in 40–60
petroleum ether) to give (4aS,5S)-22 as a colourless oil (0.019 g,
(
8
4aR,5R)-5-Hydroxy-4a-methyl-4,4a,5,6,7,
-hexahydronaphthalen-2(3H)-one, (4aR,5R)-22
[36]
63%), 96% d.e. by chiral HPLC analysis (ratio 98:2 (4aS,5-
To a stirred solution of (R)-18 (0.020 g, 0.12 mmol) in EtOH
S):(4aS,5R)). R 0.32 (50% EtOAc in 40–60 petroleum ether);
f
(
500 μL) at 0°C, NaBH (5.5 mg, 0.11 mmol) was added and
25
25
[39]
4
[α] +115 (c 0.18, CHCl ) [lit. [α] +123 (c 0.34, CHCl )];
D 3 D 3
νmax (neat) 3320, 2971, 1655 cm ; H NMR (600 MHz; CDCl ) δ
À 1
1
the reaction stirred for 15 min. Acetic acid (50 μL) was added
and the reaction stirred for 15 min at 0°C. The solvents were
evaporated, and the remaining mixture was extracted with
EtOAc (3×10 mL) and washed with sat. NaCl solution (2×
3
5
2
1
.79 (1H, d, J=1.8 Hz, 1-H), 3.44 (1H, dd, J=11.7, 4.2 Hz, 5-H),
.30–2.49 (3H, m, 3-H , 8-HH), 2.15–2.26 (2H, m, 4-HH, 8-HH),
2
.81–1.94 (3H, m, 4-HH, 6-HH, 7-HH), 1.63–1.76 (2H, m, 6-HH,
13
1
0 mL). The organic layer was dried (MgSO ) and concentrated
4
OH), 1.37–1.48 (1H, m, 7-HH), 1.20 (3H, s, CH3); C NMR
150 MHz; CDCl ) δ 199.7, 168.4, 125.7, 78.5, 41.7, 34.4, 33.8,
in vacuo. The crude product was purified using a silica plug and
washed with EtOAc (10 mL) to give (4aR,5R)-22 as an oil
(
3
+
32.1, 30.4, 23.3, 15.4; m/z (EI) 180 ([M] ). Analytical chiral
(0.019 g, 96%), 98% d.e. by chiral HPLC analysis (ratio 99:1,
Adv. Synth. Catal. 2021, 363, 1–10
8
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
asc.wiley-vch.de
HPLC retention time 85.9 min (Chiralcel OJ, propanol/hexane at
.5 mL/min; SI Figure S5).
[18] T. R. Ngara, H. Zhang, Genomics Proteomics Bioinf.
2018, 16, 405–415.
0
[
19] A. Popovic, T. Hai, A. Tchigvintsev, M. Hajighasemi, B.
Nocek, A. N. Khusnutdinova, G. Brown, J. Glinos, R.
Flick, T. Skarina, T. N. Chernikova, V. Yim, T. Brüls,
D. L. Paslier, M. M. Yakimov, A. Joachimiak, M. Ferrer,
O. V. Golyshina, A. Savchenko, P. N. Golyshin, A. F.
Yakunin, Sci. Rep. 2017, 7, 44103.
Acknowledgements
Funding from the Department of Chemistry UCL for S.A.N. and
the Biotechnology and Biological Sciences Research Council
(
BBSRC) and ALMAC for J.W.E.J. (BB/J013005/1) is gratefully
[
20] N. Itoh, K. Isotani, Y. Makino, M. Kato, K. Kitayama, T.
Ishimota, Enzyme Microb. Technol. 2014, 55, 140–150.
21] S. El-Gebali, J. Mistry, A. Bateman, S. R. Eddy, A.
Luciani, S. C. Potter, M. Qureshi, L. J. Richardson, G. A.
Salazar, A. Smart, E. L. L. Sonnhammer, L. Hirsh, L.
Paladin, D. Piovesan, S. C. E. Tosatto, R. D. Finn,
Nucleic Acids Res. 2019, 47, D427-D432.
acknowledged. We also thank the UCL NMR and Mass
Spectrometry facilities in the Department of Chemistry.
[
References
[
[
[
1] R. Singh, M. Kumar, A. Mittal, P. K. Mehta, 3 Biotech
2
016, 6, 174.
[22] C. Camacho, G. Coulouris, V. Avagyan, N. Ma, J.
Papadopoulos, K. Bealer, T. L. Madden, BMC Bioinf.
2009, 10, 1–9.
2] E. M. M. Abdelraheem, H. Busch, U. Hanefeld, F. Tonin,
React. Chem. Eng. 2019, 4, 1878–1894.
3] U. Oppermann, C. Filling, M. Hult, N. Shafqat, X. Wu, [23] S. Leuchs, L. Greiner, Chem. Biochem. Eng 2011, 25,
M. Lindh, J. Shafqat, E. Nordling, Y. Kallberg, B.
267–281.
Persson, H. Jörnvall, Chem.-Biol. Interact. 2003, 143– [24] E. L. Noey, N. Tibrewal, G. Jiménez-Osés, S. Osuna, J.
1
44, 247–253.
Park, C. M. Bond, D. Cascio, J. Liang, X. Zhang, G. W.
Huisman, Y. Tang, K. N. Houk, Proc. Natl. Acad. Sci.
USA 2015, 112, E706.
[
4] S. Matsuda, R. Yamanaka, K. Nakamura, Tetrahedron:
Asymmetry 2009, 20, 513–557.
[
[
5] A. M. Thayer, Chem. Eng. News 2007, 85, 11–19.
6] A. Calcaterra, I. D. Acquarica, J. Pharm. Biomed. Anal.
[25] J. T. Wu, L. H. Wu, J. A. Knight, Clin. Chem. 1986, 32,
314–319.
2
018, 147, 323–340.
[26] Y. Takusagawa, M. Otagiri, S. Ui, T. Ohtsuki, a Mimura,
M. Ohkuma, T. Kudo, Biosci. Biotechnol. Biochem.
2001, 65, 1876–1878.
[
[
[
7] W. Kroutil, H. Mang, K. Edegger, K. Faber, Curr. Opin.
Chem. Biol. 2004, 8, 120–126.
8] K. Goldberg, K. Schroer, S. Lütz, A. Liese, Appl. [27] E. Toomey, J. Wakil, Biochim. Biophys. Acta 1966, 116,
Microbiol. Biotechnol. 2007, 76, 237–248.
189–197.
9] L. Leipold, D. Dobrijevic, J. W. E. Jeffries, M. Bawn, [28] B. Bradshaw, J. Bonjoch, Synlett 2012, 337–356.
T. S. Moody, J. M. Ward, H. C. Hailes, Green Chem. [29] D. C. Sarkar, A. R. Das, B. C. Ranu, J. Org. Chem. 1990,
2
019, 21, 75–86.
10] D. Dobrijevic, L. Benhamou, A. E. Aliev, D. Méndez- [30] D. E. Ward, C. K. Rhee, W. M. Zoghaib, Tetrahedron
Sánchez, N. Dawson, D. Baud, N. Tappertzhofen, T. S.
Lett. 1988, 29, 517–520.
Moody, C. A. Orengo, H. C. Hailes, J. M. Ward, RSC [31] S. J. Danishefsky, J. J. Masters, W. B. Young, J. T. Link,
55, 5799–5801.
[
Adv. 2019, 9, 36608–36614.
11] D. D. Roumpeka, R. J. Wallace, F. Escalettes, I. Fother-
ingham, M. Watson, Front. Genet. 2017, 8, 23, doi:
L. B. Snyder, T. V. Magee, D. K. Jung, R. C. A. Isaacs,
W. G. Bornmann, C. A. Alaimo, C. A. Coburn, M. J.
Di Grandi, J. Am. Chem. Soc. 1996, 118, 2843–2859.
[32] K. I. Fuhshuku, N. Funa, T. Akeboshi, H. Ohta, H.
Hosomi, S. Ohba, T. Sugai, J. Org. Chem. 2000, 65,
129–135.
[
1
0.3389/fgene.2017.00023.
[
12] J. W. E. Jeffries, N. Dawson, C. Orengo, T. S. Moody,
D. J. Quinn, H. C. Hailes, J. M. Ward, ChemistrySelect
2
016, 1, 2217–2220.
[33] H. Hioki, T. Hashimoto, M. Kodama, Tetrahedron:
Asymmetry 2000, 11, 829–834.
[34] T. Janeczko, J. Dmochowska-Gładysz, E. Kostrzewa-
Susłow, World J. Microbiol. Biotechnol. 2010, 26, 2047–
2051.
[
13] D. Baud, J. W. E. Jeffries, T. S. Moody, J. M. Ward,
H. C. Hailes, Green Chem. 2017, 19, 1134–1143.
14] E. E. Ferrandi, C. Sayer, S. A. De Rose, E. Guazzelli, C.
Marchesi, V. Saneei, M. N. Isupov, J. A. Littlechild, D.
Monti, Front. Bioeng. Biotechnol. 2018, 6, 144.
[
[35] N. Shimizu, H. Akita, T. Kawamata, Tetrahedron 2002,
13, 2123–2131.
[
15] R. Wohlgemuth, J. Littlechild, D. Monti, K. Schnorr, T.
van Rossum, B. Siebers, P. Menzel, I. V. Kublanov, A. G. [36] K. Ma, C. Zhang, M. Liu, Y. Chu, L. Zhou, C. Hu, D.
Rike, G. S. Skretas, Z. Szabo, X. Peng, M. J. Young,
Biotechnol. Adv. 2018, 36, 2077–2100.
16] O. López-López, M. E. Cerdán, M. I. González-Siso,
Curr. Protein Pept. Sci. 2014, 15, 445–455.
Ye, Tetrahedron Lett. 2010, 51, 1870–1872.
[37] A. Hammoumi, G. Revial, J. D’Angelo, J. P. Girault, R.
Azerad, Tetrahedron Lett. 1991, 32, 651–654.
[38] P. Renouf, J. Poirier, P. Duhamel, J. Chem. Soc. Perkin
Trans. 1 1997, 1739–1745.
[
[
17] M. Martin, S. Biver, S. Steels, T. Barbeyron, M. Jam, D.
Portetelle, G. Michel, M. Vandenbol, Appl. Environ. [39] W. S. Cheung, H. N. C. Wong, Tetrahedron 1999, 55,
Microbiol. 2014, 80, 4958–4967.
11001–11016.
Adv. Synth. Catal. 2021, 363, 1–10
9
© 2021 The Authors. Advanced Synthesis & Catalysis
published by Wiley-VCH GmbH
�
�
These are not the final page numbers!

FULL PAPER
Discovery of New Carbonyl Reductases Using Functional
Metagenomics and Applications in Biocatalysis
Adv. Synth. Catal. 2021, 363, 1–10
S. A. Newgas, J. W. E. Jeffries, T. S. Moody, J. M. Ward*,
H. C. Hailes*