Fluorogenic kinetic assay for high-throughput discovery of stereoselective ketoreductases relevant to pharmaceutical synthesis

Article abstract of DOI:10.1016/j.bmc.2017.05.024

Enantiomerically pure 1-(6-methoxynaphth-2-yl) and 1-(6-(dimethylamino)naphth-2-yl) carbinols are fluorogenic substrates for aldo/keto reductase (KRED) enzymes, which allow the highly sensitive and reliable determination of activity and kinetic constants of known and unknown enzymes, as well as an immediate enantioselectivity typing. Because of its simplicity in microtiter plate format, the assay qualifies for the discovery of novel KREDs of yet unknown specificity among this vast enzyme superfamily. The suitability of this approach for enzyme typing is illustrated by an exemplary screening of a large collection of short-chain dehydrogenase/reductase (SDR) enzymes arrayed from a metagenomic approach. We believe that this assay format should match well the pharmaceutical industry's demand for acetophenone-type substrates and the continuing interest in new enzymes with broad substrate promiscuity for the synthesis of chiral, non-racemic carbinols.

Full text of DOI:10.1016/j.bmc.2017.05.024

Accepted Manuscript
Fluorogenic Kinetic Assay for High-Throughput Discovery of Stereoselective
Ketoreductases Relevant to Pharmaceutical Synthesis
Yen-Chi Thai, Anna Szekrenyi, Yuyin Qi, Gary W. Black, Simon J. Charnock,
Wolf-Dieter Fessner
PII:
S0968-0896(17)30086-X
http://dx.doi.org/10.1016/j.bmc.2017.05.024
BMC 13742
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
16 January 2017
30 March 2017
11 May 2017
Please cite this article as: Thai, Y-C., Szekrenyi, A., Qi, Y., Black, G.W., Charnock, S.J., Fessner, W-D., Fluorogenic
Kinetic Assay for High-Throughput Discovery of Stereoselective Ketoreductases Relevant to Pharmaceutical
Synthesis, Bioorganic & Medicinal Chemistry (2017), doi: http://dx.doi.org/10.1016/j.bmc.2017.05.024
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Fluorogenic Kinetic Assay for High-Throughput Discovery of
Stereoselective Ketoreductases Relevant to Pharmaceutical Synthesis
Yen-Chi Thai,a Anna Szekrenyi,a Yuyin Qi,b,c Gary W. Black,c Simon J. Charnock,b Wolf-
Dieter Fessner a*
a Institut für Organische Chemie und Biochemie, Technische Universität Darmstadt, Alarich-
Weiss-Str. 4, 64287 Darmstadt, Germany. Email: fessner@tu-darmstadt.de
b Prozomix Ltd, Station Court, Haltwhistle, Northumberland, NE49 9HN, UK
c Department of Applied Sciences, Northumbria University, Faculty of Health and Life
Sciences, Newcastle-upon-Tyne, NE1 8ST, UK
ABSTRACT
Enantiomerically pure 1-(6-methoxynaphth-2-yl) and 1-(6-(dimethylamino)naphth-2-yl)
carbinols are fluorogenic substrates for aldo/keto reductase (KRED) enzymes, which allow
the highly sensitive and reliable determination of activity and kinetic constants of known and
unknown enzymes, as well as an immediate enantioselectivity typing. Because of its
simplicity in microtiter plate format, the assay qualifies for the discovery of novel KREDs of
yet unknown specificity among this vast enzyme superfamily. The suitability of this approach
for enzyme typing is illustrated by an exemplary screening of a large collection of short-chain
dehydrogenase/reductase (SDR) enzymes arrayed from a metagenomic approach. We believe
that this assay format should match well the pharmaceutical industry’s demand for
acetophenone-type substrates and the continuing interest in new enzymes with broad substrate
promiscuity for the synthesis of chiral, non-racemic carbinols.
Keywords: alcohol dehydrogenase, asymmetric synthesis, biocatalysis, chiral alcohols,
enantioselectivity
1. Introduction
Biocatalysis offers excellent opportunities for the synthesis of chiral compounds in a non-
racemic fashion.1 Complementary to chemical synthesis or kinetic resolution of racemates,
enzyme catalyzed direct asymmetric synthesis is a highly efficient technology, for example
for the preparation of chiral alcohols, chiral amines, or chiral ketone derivatives by ketone
reductions using carbonyl reductases,2 reductive aminations using transaminases3 or
carboligations using aldolases4 or transketolases,5 respectively, in up to 100% optical purity
and 100% yield. Particularly, NAD(P)H dependent ketoreductases (KREDs) are a well-
established and reliable source of chiral alcohols with high enantiomeric purity, which are of
importance as chiral building blocks for the production of pharmaceutical intermediates
(Figure 1).6
1

OH
OH
OH
Br
Cl
Cl
O
O
ref[7a]
OH
ref[7b]
ref[7c]
Cl
N
F
OH
F3C
Br
O
ref[7d]
ref[7e]
CF3
NHSO2CH3
Fig. 1. Chiral pharmaceutical intermediates produced on large scale using KRED catalysis7
While the application of baker’s yeast and related whole cell catalysts is attractive for the low
costs,8 most organisms were found to follow Prelog’s rule in producing predominantly the
(S)-configured products.9 In addition, microbial cells typically contain multiple enzymes
having potentially conflicting specificities, thus eroding product yields and enantiomeric
excess. The advent of larger commercial collections of recombinant KREDs significantly
facilitates the identification of an appropriate catalyst for the conversion of a specific
substrate. Discovery of new KREDs from various sources, such as for less common enzymes
with anti-Prelog selectivity,10 can be pursued in various ways. This includes theoretical
approaches in silico such as data mining from recent whole genome sequencing, and
experimental approaches in vitro such as screening in culture collections or in metagenomic
campaigns, as well as rational engineering or directed evolution of known enzyme scaffolds.
A decisive factor for the further development of this platform technology is the ability for
rapid identification of enzymes having high activity and high stereoselectivity for the desired
product configuration. This is usually performed by screening for activity to reduce the ketone
substrate, followed by determination of product alcohol configuration and enantiomeric
excess using chiral chromatographic techniques (GC, HPLC), which can only be performed in
a tedious serial fashion and are expensive. As most ketone reductions do not produce readily
detectable changes in UV/VIS absorption or fluorescence of the product, the conversion is
usually followed using either the associated depletion of reduced nicotinamide cofactor by
UV (340 nm = 6 300 M-1 cm-1), or by coupling redox cofactor recycling to a secondary product
that can be monitored colorimetrically (e.g. INT-formazan; 492 nm = 19 900 M-1 cm-1).11
However, significantly higher sensitivity and specificity are attainable by measurement of
fluorescence emission rather than UV/VIS absorption.12 Fluorogenic enzyme substrates are
convenient tools for screening because the signal produced is directly related to the enzyme-
catalyzed reaction. Thus, a fluorometric assay for the discrimination of human alcohol
dehydrogenase isoenzymes has been developed based on the consumption of fluorescent
naphthaldehydes as reporter substrates.13 An indirect assay has been published by which a
chiral secondary alcohol is oxidized to produce a ketone, which subsequently undergoes -
elimination in the presence of bovine serum albumin to release umbelliferone;14 however,
because the blue fluorescent phenolate anion needs a shift to alkaline pH, the assay is not
amenable to continuous kinetic measurements. To the best of our knowledge, no direct
stereoselective fluorescence assay for measuring KRED activity has been reported.
Here we report on the development of a direct fluorogenic assay format, which allows the
rapid and highly sensitive quantification of enzyme activity in a high-throughput fashion in a
highly parallel microtiter plate format. The assay can be used for simultaneous Prelog
stereospecificity typing as well as the determination of the level of enantioselectivity of
KRED enzymes in one step. The assay concept further allows a systematic modular variation
of the substrate structure to facilitate a substrate morphing approach4a for protein engineering
projects.
2

2. Results and Discussion
2.1 Assay principle
KREDs are nicotinamide cofactor-depending enzymes that reduce ketones into chiral
secondary alcohols and vice versa.15 In the reverse mode, the carbinol functionality reacts by
hydride transfer to NAD(P)+ with formation of NAD(P)H and release of a ketone. Substituted
acetophenones comprise one of the most important substrate classes, for which most KREDs
have been found to be active against, many exhibiting high stereoselectivity.8b When the non-
fluorescent 1-(6-methoxynaphth-2-yl)ethanol 1a (R=CH3) in the appropriate enantiomeric
form acts as a KRED substrate, the corresponding acetonaphthone derivative 2a would be
formed, which shows a bright blue fluorescence emission around 460 nm with quantum yield
of about 0.2 that can be readily detected using a common plate reader (Scheme 2). Recently,
we have developed a similar fluorogenic assay for transaminases based on the corresponding
chiral amine, from which the same ketone is liberated upon transaminase catalysis.16 It is
interesting to note that substrate rac-1a very recently has also been used in a confocal
microscopic study to localize redox enzymes in the yeast Candida parapsilosis by occurrence
of fluorescence upon oxidative generation of ketone 2a.17
OH
O
R
R
MeO
MeO
1
2
KRED
NAD(P)+
NAD(P)H
Scheme 1. Fluorogenic assay principle for KREDs with release of the fluorescent ketone 2 upon
oxidation of a chiral benzylic alcohol 1.
2.2 Substrate Synthesis and Resolution
For the synthesis of racemic alcohols 1 we recently described a modular approach comprising
a Grignard addition to the commercial aldehyde 3 (Scheme 3),16 which directly furnishes the
desired compounds for KRED assays. In addition, the individual carbinol 1a was prepared by
standard borohydride reduction of the commercial ketone 2a.18 This synthetic strategy rapidly
provides access to a family of assay compounds that can be precisely designed by selecting an
appropriate Grignard reagent for introducing a broad variety of structural variations in the R
group, which should facilitate an incremental sculpting of the small binding pocket of a given
target KRED scaffold. Besides being highly efficient because of its modular nature, this
synthesis also avoids the need for tedious separation of regioisomeric 6-methoxy-
acylnaphthalenes, which are obtained by Friedel-Crafts acylation.19
Enantiomers of 1a were kinetically resolved by lipase-catalyzed transacylation in vinyl
acetate using immobilized lipase B from Candida antarctica (CAL-B) according to literature
precedent.20 Both antipodes of the carbinol were obtained with >99% ee (HPLC analysis) in
excellent yield, including the deprotection of the ester (R)-4a. Configurational assignments
follow the general predictions of Kazlauskas rule.21
3

O
O
OH
RMgX
H
R
MeO
MeO
MeO
MeO
3
rac-1a R=Me
b R=Pr
c R=Bn
NaBH4
MeOH
vinyl acetate
24 h, 37¡C
CAL-B
2a
OH
OAc
OH
KOH
+
MeOH
2 h r.t.
MeO
MeO
(R)-1a
(R)-4a
(S)-1a
Scheme 2. Synthesis of substrate alcohols 1a-c and racemate resolution of 1a by lipase catalyzed
transacylation.
2.3 Assay Development
Orienting tests performed with commercial yeast alcohol dehydrogenase (YADH) and horse
liver alcohol dehydrogenase (HLADH) proved negative, apparently the acetonaphthone
moiety being too bulky to enter the substrate-binding sites. However, when racemic 1a was
incubated with the (R)-selective KRED from Lactobacillus brevis in the presence of its
preferred cofactor NADP+ at an optimum pH of 7.5, the development of the bright blue
fluorescence of ketone 2a could be monitored against background, verifying the general assay
concept. These experiments also indicated that the extended naphthyl fluorophore structure is
instrumental in effectively discriminating enzymes that are substrate tolerant (e.g. the KRED
from L. brevis) from those having a rather narrow specificity for smaller substrates (e.g.
YADH and HLADH enzymes).
Indeed, the L. brevis KRED has previously been shown to have a broad substrate tolerance,22
in particular being highly adaptive to bulky substrates including substituted acetophenones.23
This broad scope as well as its high stability and stereoselectivity make the enzyme very
attractive for industrial applications.24 The X-ray protein crystal structure has been solved to
atomic resolution, including a ternary complex with nicotinamide cofactor and substrate (R)-
1-phenyl ethanol bound to the active site.25 Inspection of its structure indicates that the
substrate binding site has a narrow pocket at the bottom of the groove, which binds the small
alkyl moiety of an (R)-configured methyl carbinol, while the arene unit makes several
hydrophobic contacts to side-chains but points towards the entrance of the active site.25b The
latter appears to be wide enough to also accommodate the substituted naphthyl moiety of the
fluorogenic substrate (R)-1a.
Despite its polar hydroxyl function substrate 1a is only sparingly soluble in plain water and
must be applied from concentrated stock solution in DMSO or DMF. Dilution to a final assay
concentration of 15% DMSO gave clear solutions containing up to 1.0 mM substrate 1a or
ketone 2a that generated a consistent data quality in liquid assays. KRED enzymes require
nicotinamide cofactors in stoichiometric quantities or a suitable in situ recycling system. In
direction of oxidation, NAD(P)+ recycling is typically achieved by either substrate-coupling
in the presence of a sacrificial ketone, which thereby is reduced to the corresponding
4

secondary alcohol, or by enzyme-coupling such as with the lactate dehydrogenase/pyruvate
system.26 For several reasons we decided on the substrate-coupling option with acetone as co-
substrate, which takes advantage of the ability of many KREDs to react with small ketone
substrates, and because conversion to isopropanol does not induce pH changes that are
typically observed for enzyme-coupled approaches; also no extra enzyme is needed. An
additional benefit is the fact that acetone can assist as a co-solvent to solubilize the substrate
and usually is well tolerated by enzymes at concentrations of 10% or even higher.27 An
enzyme identified by this new procedure in turn will offer an economical advantage with
respect to the necessity for cofactor recycling when planning preparative applications on
larger scale.28
Considering the fluorescence spectrum of ketone 2a in water, kinetic measurements were
performed with excitation at 330 nm (max = 312 nm) and emission observed at 460 nm (max
= 441 nm).18 The ketone shows a linear fluorescence increase with rising concentration,
which allows continuous quantification of enzyme activity by correlation with a standard
curve. The limit of detection (LOD) and limit of quantification (LOQ) values derived by
linear regression to characterize the performance of the assay were 0.35 nmol/min and 1.2
nmol/min (vmax for 2a), respectively.29 In order to shift the reaction equilibrium towards target
ketone 2a synthesis, assays were run under slightly alkaline conditions (pH 7.5), in the
presence of the natural nicotinamide cofactor mixture present in the cell-free extract.
Resultant time curves for appearance of fluorescence due to formation of ketone 2a showed
excellent linearity, and varying the amount of protein confirmed a good correlation between
signal intensity and enzyme activity (Figs. S3, S4). No fluorescence signal was recorded in
either the absence of the biocatalyst or co-substrate, as long as no additional NAD(P)+
cofactor was added (which could act as a co-substrate) beyond that bound to the enzyme
active site. Measurements were tested in microtiter plate format with good reproducibility
against parallel standard dilutions.
Reaction rates were easily determined even for concentrations of alcohol 1a as low as 100 M
and short reaction times (20 min) with good linearity between substrate concentration and
signal intensity (Fig. S3). When the KRED from L. brevis was exposed to pure substrate
enantiomers, only the (R)-1a isomer gave rise to fluorescence development while the opposite
(S)-1a isomer did not react, even over extended reaction times (<1% activity). This confirms
reports for acetophenone-type reductions that this enzyme has a high discrimination in the
specificity constants kcat/KM for individual enantiomers,22a, 23a or an excellent kinetic E-factor
(or enantioselectivity).30 For (R)-1a the L. brevis KRED showed Michaelis-Menten-type
kinetics with a KM = 0.30 mM in the presence of 15% DMSO (Fig. S6).
Although we have not yet explored the option for measuring the influence of further reaction
parameters that might have an influence on enzyme activity, such as pH, temperature or
addition of other co-solvents, it appears to be a rather simple task to study enzyme properties
by serial variation of any such effectors or physical parameters under the standard assay
conditions. Especially, the replacement of the co-substrate acetone by other (non-fluorescent)
ketones or aldehydes as alternatives for the required in situ cofactor recycling seems a
promising option to further determine the substrate specificity of a given KRED by using the
otherwise identical, highly sensitive detection method.
In order to evaluate and validate the assay for its potential in high-throughput screening
applications, we further determined the Z-factor,31 a dimensionless statistical parameter that
reflects both the assay signal dynamic range and the data variation associated with the signal
measurements. For a substrate concentration of 0.8 mM we determined an excellent Z-factor
of 0.90, whereas even at lower substrate concentration of 0.2 mM a very high Z-factor of 0.72
resulted. Thus, the fluorogenic assay seems to be very appropriate for screening efforts of
variant libraries in the directed evolution of KREDs with improved or altered protein
5

properties, while consuming only very small substrate quantities (e.g., 0.5 mg of 1a per 96-
well microtiter plate at 0.25 mM substrate concentration).
2.4 Typing of novel KRED enzymes
As an exemplary application of the new assay we attempted to identify active KRED hits
among a commercial collection of aldo/ketoreductase enzymes arrayed in 96-well plate
format (Prozomix Ltd) by exposure to the rac-1a substrate under the optimized conditions
(0.25 mM, 5% acetone, 90 min). The enzymes were recruited by a metagenomic approach
employing environmental samples from the UK and are classified as belonging to the short-
chain dehydrogenases/reductases (SDR) type. SDR enzymes constitute one of the largest
protein superfamilies known today, most of which are NAD(P)H-dependent
oxidoreductases.32 SDRs present a common three-dimensional structure featuring an N-
terminal Rossmann-fold motif for nucleotide binding. They form a large, functionally
heterogeneous protein family that displays a wide substrate spectrum, including steroids,
alcohols, sugars, aromatic compounds, and xenobiotics. When we discovered that indeed a
high fraction of biocatalysts in this SDR collection proved active on the sterically demanding
substrate 1a, we conducted an enantioselectivity typing of these enzymes by applying the
individual enantiomers (S)-1a and (R)-1a under the same conditions (Fig. 2). The
corresponding activity data allow calculating a rough estimate for the enantioselectivity of the
KRED panelists before determining exact specificity constants.
Interestingly, while the most active candidate (G03 = KRED363) in the specific collection
examined reached up to 90% total substrate conversion, more than half of the enzymes
(53/96) caused at least 10% relative conversion, and more than one quarter (27/96) even
reached at least 50% conversion, with either substrate enantiomer. Concerning their
enantiopreference, (S)-selective Prelog-type enzymes were by far predominating (42/96;
almost 80% of active catalysts), with very few enzymes showing anti-Prelog (R)-specificity
(2/96; <4% of active catalysts); some fraction of the enzymes (9/96; 17% of active catalysts)
showed only low enantioselectivity because they had ambiguous activity with both substrate
enantiomers.
A)
B)
Fig. 2. A) Exemplary screening of a highly diverse metagenomic short-chain dehydrogenase/reductase
(SDR) array in 96-well plate format (Prozomix Ltd, Cat No.s PRO-KRED(00289-00384)), by using
individual enantiomers (S)-1a and (R)-1a ( and , respectively). Maximum substrate conversion
after 1.5 hours was 90% as determined by comparison against a fluorescence standard curve measured
6

with ketone 2a under identical assay conditions. B) Screening of the same plate by using rac-1b as
substrate ().
For an assay validation, in addition to the (R)-selective KRED from L. brevis one
representative hit of an (S)-selective (plate entry A12 = KRED(00300)) and non-
stereoselective SDR (plate entry B03 = KRED(00303)) were expressed in E. coli, purified by
NTA chromatography and re-evaluated with respect to their enantioselectivity. In each case,
the stereopreference could be unequivocally confirmed. All enzymes showed Michaelis-
Menten-type kinetics with high substrate affinity (e.g., KM = 0.076 mM of KRED(00300) for
substrate (S)-1a and 0.025 mM of KRED(00303) for substrate (R)-1a).
An attempt to correlate the activity and stereoselectivity data with a phylogenetic analysis of
the SDR panel revealed that properties were spread throughout the respective protein family
(PF00106) without any clear pattern (Fig. 3). This result emphasizes the need for novel assays
to rapidly characterize novel KREDs experimentally as their stereo-preference cannot be
predicted by genetic cluster analysis alone, even within the primary sequence boundaries of
an individual pfam.
Fig. 3. Phylogenetic analysis for the novel
and primary sequence diverse metagenomic
SDR array (Fig. 2) showing a correlation of
the enantiospecificity for the most active
enzymes (>40% rel. activity). Highly (S)-
selective and (R)-selective enzymes (> 90%
selectivity) are color coded as in Fig. 2,
while non-stereoselective ones (< 90%
selectivity) are shown in magenta.
We also conducted an exemplary test to evaluate the potential of the fluorogenic assay
method to determine the substrate tolerance of individual KRED enzymes. For this purpose,
the arrayed panel was probed against the homologous propyl carbinol substrate rac-1b.16
When compared to the activity with substrate 1a carrying the smallest methyl substituent,
7

interestingly the carbinol 1b having a larger alkyl moiety was converted by about one fifth of
all enzymes (20/96), or more than one third of the active catalysts (20/53) with a relative
velocity of >10%, indicating that many catalyst in the SDR category will offer an interesting
substrate promiscuity. Still, the majority of the active enzymes in the SDR panel seems to
have a rather small pocket for adapting only a methyl moiety. Particularly, the option for
systematic structural variation of the substrate by altering the R group in the modular
synthesis of carbinols 1 should be a valuable asset to probe and expand the small substrate
pocket of a selected KRED by protein engineering, if directed at residues building up the
small substrate binding pocket. The potential utility of such a biocatalyst engineering
approach is supported by the observation that when challenged with the benzyl derivative rac-
1c, an example of a substrate carrying two very bulky moieties next to the reaction site,
however none of the native SDRs in the selected panel showed detectable activity.
2.5 Sensitivity Improvement
The assay principle is based on the blue fluorescence emission of the methoxy substituted
ketone 2, for which a certain limitation might be seen in the incomplete quantum yield and the
potential interference of the UV irradiation wavelength with other photoactive components in
the assay solution, e.g. when using crude cell extracts. In order to enhance the sensitivity of
signal detection, we also tested the corresponding acedan system (Scheme 3), where the
methoxy substituent of ketone 2a is replaced by a dimethylamino group (7a). The latter not
only causes a shift of the fluorescence emission towards a longer green wavelength (abs
=
362 nm and em = 524 nm)18 but also significantly improves upon the fluorescence brightness
(  F = 2240). The synthesis of 6a was first pursued via the phenol derived from 2a by
using the Bucherer reaction,33 which suffered from incomplete conversion and separation
problems. Much preferred is the direct substitution according to a modified literature
procedure using LiNMe2 in DMPU.34 Following borohydride reduction of ketone 6a, the
resulting carbinol rac-5a could then be easily resolved into its enantiomers (S)-5a and (R)-5a
by CAL-B catalyzed transacylation in analogy to the separation of methoxy compound rac-1a
(Scheme 2).
O
H3CO
2a
blue fluorescence
LiNMe2
DMPU/THF-hexane
NaBH4, MeOH
24 h, r.t.
OH
O
KRED
NAD(P)+
Me2N
Me2N
6a
green fluorescence
rac-5a
Scheme 3. Synthesis of substrate alcohol 5a for KRED activity screening with higher sensitivity due
to longer wavelength fluorescence emission.
Although substrate 5a itself is fluorescent with emission in the blue range of the visible
spectrum, there is no overlap with the long-wavelength green emission of ketone 6a. Signal
development was recorded at 520 nm without background interference. For an exemplary test,
8

compound rac-5a was thus applied to the fluorogenic screening of the same diverse SDR
array in 96-well plate format (Fig. 4), using the same conditions previously developed for the
application of 1a. For (R)-5a the KRED from L. brevis showed a high substrate affinity (KM =
0.22 mM), which is almost identical to that for (R)-1a (Fig. S7).
Fig. 4. Exemplary fluorogenic screening of a KRED array in 96-well plate format (Prozomix Ltd, Cat
No.s PRO-KRED(00289-00384)) against rac-5a.
3. Conclusion and Outlook
The generation of brightly fluorescent acetonaphthone derivatives 2 and 6 from chiral 1-(6-
methoxynaphth-2-yl)carbinols 1 and its 6-dimethylamino analog 5, respectively, is a new
principle to prospect for useful KRED catalysis. We demonstrated that this new assay format
is applicable to the reliable determination of activity and kinetic constants of known and
unknown enzymes. When employing enantiomerically pure individual antipodes of carbinols
1a/5a, the method allows for the rapid and immediate and cost-effective enantioselectivity
typing of enzymes. The assay is characterized by high sensitivity in microtiter plate format,
which qualifies for the discovery of novel KREDs of yet unknown specificity among this vast
enzyme superfamily. The suitability of this approach for enzyme typing was confirmed by an
exemplary screening of a large collection of SDR enzymes arrayed from a metagenomic
approach. We believe that this assay format should match well the pharmaceutical industry’s
demand for acetophenone-type substrates and the continuing interest in new enzymes with
broad substrate promiscuity, particularly those tolerating bulky substituents, for the synthesis
of chiral, non-racemic carbinols. We note that our fluorogenic assay also avoids problems
from an interference of UV absorbing pharmaceutical intermediates with conventional
NAD(P)H dependent measurements.
Its simplicity and low substrate consumption renders the assay particularly suitable for large-
scale liquid screening campaigns in the directed evolution of optimized KREDs. The latter
purpose is facilitated by the modular scheme of substrate synthesis that allows a simple and
rapid generation of broad structural variations in the acyl/carbinol moiety, required in
particular for a targeted adaptation of the small substrate-binding pocket. Not least, such
discovery and optimization of desired KREDs is also important for the integration of redox
conversion modules into advanced reaction cascades along the concept of Systems
Biocatalysis.35
4. Experimental Section
4.1 Fluorogenic assay set-up and determination of kinetic constants
9

The assay was developed in Sarsted 96-well plates using a BMG Optima plate reader
equipped with 330 nm filters for excitation and 460 nm filters for emission. The standard
curve of the assay was determined by using a dilution series of the ketone 2a and 6a in a
concentration range of 0.0–1.0 mM in the corresponding assay buffer. Different enzyme-
substrate pairs were measured as triplicates using a serial dilution of 0.0–3.0 mg mL-1 of the
ADH from Lactobacillus brevis (lyophilized cell free extract in 50 mM phosphate buffer, pH
7.5). Each well contained 30 µL substrate stock in a 2:1 mixture of DMSO and acetone, 40
µL enzyme stock and 130 µL phosphate buffer (50 mM, pH 7.5).
Kinetic constants were calculated by varying the substrate concentration of (R)-1a and (R)-5a
between 0.0 and 1.0 mM using a fixed enzyme concentration. Each well contained 0–30 µL
substrate stock in a 2:1 mixture of DMSO and acetone, 30–0 µL of a DMSO–acetone mixture,
40 µL enzyme stock and 130 µL phosphate buffer (50 mM, pH 7.5). All measurements were
performed in triplicate at 30 °C. Raw data were exported to MS Excel and were processed
with Origin 9.1 for hyperbolic fitting.
LOD and LOQ were determined for both 2a and 6a. Assay reaction was performed using
ADH from L. brevis cell free extract (200 μg per well) for incubation of substrate (R)-1a at
0.5 mM concentration and negative control without substrate. All data were measured as
triplicates. The LOD was defined as LOD = 3.3 σ/S and LOQ was defined as LOQ = 10 σ/S,
where σ is the standard deviation of the control and S is the slope of the standard curve. They
were calculated for LOD = 6 μM and 1.7 μM, LOQ = 18 μM and 5 μM for 2a and 6a,
respectively.
4.2 Microplate screening assays using commercial plate
Preparation of screening plates for KRED enzymes was performed as described.11 Plates were
kept frozen at -20°C for storage until use. Using the cross-hairs in the microplate sealing film
as a guide, 100 µL of an aq 0.25 mM solution of enantiomerically pure (R)-1a or (S)-1a
containing 5% DMSO and 5% acetone were added to each well of the microplate (Prozomix
Ltd, Cat No.s PRO-KRED(00289-00384)) using a multi-channel pipette. The contents were
mixed carefully to dissolve the freeze-dried enzyme pellets. The plates were briefly shaken on
a plate shaker, then pulse centrifuged, followed by removal of the cross-hair film before the
kinetic analysis using a microplate reader.
4.3 Screening for substrate promiscuity
The microplate was tested using the assay conditions as described above but replacing the
substrate with 1-(6-methoxynaphthalen-2-yl)butanol rac-1b or 1-(6-methoxynaphthalen-2-
yl)-2-phenylethanol rac-1c.
4.4 Supplementary data
Supplementary data associated with this article (synthetic procedures, compound
characterization and NMR spectra, HPLC analysis, kinetic charts) can be found, in the online
version, at http://…
Acknowledgements
This project has received funding from the European Union’s Horizon 2020 research and
innovation programme under grant agreement No 635595 (CarbaZymes). We gratefully
acknowledge Lucy Zhao and Katja Habrich for their contributions to the experimental
explorations.
10

References
1. (a) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler, M.; Stuermer, R.; Zelinski,
T., Angew. Chem., Int. Ed. 2004, 43, 788-824; (b) Hollmann, F.; Arends, I. W. C. E.;
Buehler, K.; Schallmey, A.; Buehler, B., Green Chem. 2011, 13, 226-265; (c) Hollmann,
F.; Arends, I. W. C. E.; Holtmann, D., Green Chem. 2011, 13, 2285-2314.
2. (a) Voss, C. V.; Gruber, C. C.; Faber, K.; Knaus, T.; Macheroux, P.; Kroutil, W., J. Am.
Chem. Soc. 2008, 130, 13969-13972; (b) Groeger, H.; Hummel, W.; Borchert, S.;
Krausser, M., Reduction of ketones and aldehydes to alcohols. In: Drauz, K.; Gröger, H.;
May, O., eds. Enzyme Catalysis in Organic Synthesis (3rd Ed.); Wiley-VCH: Weinheim,
2011; Vol. 2, pp 1037-1110; (c) Garcia-Urdiales, E.; Alfonso, I.; Gotor, V., Chem. Rev.
2011, 111, PR110-PR180; (d) Hall, M.; Bommarius, A. S., Chem. Rev. 2011, 111, 4088-
4110; (e) Musa, M. M.; Phillips, R. S., Catal. Sci. Technol. 2011, 1, 1311-1323.
3. (a) Hoehne, M.; Bornscheuer, U. T., Application of transaminases. In: Drauz, K.; Gröger,
H.; May, O., eds. Enzyme Catalysis in Organic Synthesis (3rd Ed.); Wiley-VCH:
Weinheim, 2011; Vol. 2, pp 779-820; (b) Mathew, S.; Yun, H., ACS Catal. 2012, 2, 993-
1001; (c) Simon, R. C.; Richter, N.; Busto, E.; Kroutil, W., ACS Catal. 2014, 4, 129-143.
4. (a) Fessner, W.-D., Aldol reactions. In: Drauz, K.; Gröger, H.; May, O., eds. Enzyme
Catalysis in Organic Synthesis (3rd Ed.); Wiley-VCH: Weinheim, 2011; Vol. 2, pp 857-
917; (b) Brovetto, M.; Gamenara, D.; Saenz Mendez, P.; Seoane, G. A., Chem. Rev.
2011, 111, 4346-4403; (c) Clapes, P.; Garrabou, X., Adv. Synth. Catal. 2011, 353, 2263-
2283; (d) Samland, A. K.; Rale, M.; Sprenger, G. A.; Fessner, W.-D., ChemBioChem
2011, 12, 1454-1474;.
5. (a) Mueller, M.; Sprenger, G. A.; Pohl, M., Curr. Opin. Chem. Biol. 2013, 17, 261-270;
(b) Ranoux, A.; Hanefeld, U., Top. Catal. 2013, 56, 750-764; (c) Abdoul-Zabar, J.; Sorel,
I.; Helaine, V.; Charmantray, F.; Devamani, T.; Yi, D.; de Berardinis, V.; Louis, D.;
Marliere, P.; Fessner, W.-D.; Hecquet, L., Adv. Synth. Catal. 2013, 355, 116-128; (d) Yi,
D.; Saravanan, T.; Devamani, T.; Charmantray, F.; Hecquet, L.; Fessner, W.-D., Chem.
Commun. 2015, 51, 480-483; (e) Saravanan, T.; Reif, M.-L.; Yi, D.; Lorilliere, M.;
Charmantray, F.; Hecquet, L.; Fessner, W.-D., Green Chem. 2017, DOI:
10.1039/C6GC02017H.
6. (a) Patel, R. N., Curr. Opin. Biotechnol. 2001, 12, 587-604; (b) Moore, J. C.; Pollard, D.
J.; Kosjek, B.; Devine, P. N., Acc. Chem. Res. 2007, 40, 1412-1419; (c) Goldberg, K.;
Schroer, K.; Luetz, S.; Liese, A., Appl. Microbiol. Biotechnol. 2007, 76, 237-248; (d)
Goldberg, K.; Schroer, K.; Luetz, S.; Liese, A., Appl. Microbiol. Biotechnol. 2007, 76,
249-255; (e) Huisman, G. W.; Liang, J.; Krebber, A., Curr. Opin. Chem. Biol. 2010, 14,
122-129; (f) Patel, R. N., ACS Catal. 2011, 1, 1056-1074; (g) Simon Robert, C.; Mutti
Francesco, G.; Kroutil, W., Drug Discov. Today Technol. 2013, 10, e37-44.
7. (a) Krulewicz, B.; Tschaen, D.; Devine, P.; Lee, S. S.; Roberge, C.; Greasham, R.;
Chartrain, M., Biocatal. Biotransform. 2001, 19, 267-279; (b) Badland, M.; Burns, M. P.;
Carroll, R. J.; Howard, R. M.; Laity, D.; Wymer, N. J., Green Chem. 2011, 13, 2888-
2894; (c) Patel, R. N.; Goswami, A.; Chu, L.; Donovan, M. J.; Nanduri, V.; Goldberg, S.;
Johnston, R.; Siva, P. J.; Nielsen, B.; Fan, J.; He, W.; Shi, Z.; Wang, K. Y.; Eiring, R.;
Cazzulino, D.; Singh, A.; Mueller, R., Tetrahedron: Asymm. 2004, 15, 1247-1258; (d)
Pollard, D.; Truppo, M.; Pollard, J.; Chen, C.-Y.; Moore, J., Tetrahedron: Asymm. 2006,
17 (4), 554-559; (e) Patel, R. N.; Banerjee, A.; Chu, L.; Brozozowski, D.; Nanduri, V.;
Szarka, L. J., J. Am. Oil Chem. Soc. 1998, 75, 1473-1482.
8. (a) Servi, S., Synthesis 1990, 1-25; (b) Csuk, R.; Glaenzer, B. I., Chem. Rev. 1991, 91,
49-97; (c) Matsuda, T.; Yamanaka, R.; Nakamura, K., Tetrahedron: Asymm. 2009, 20,
513-557.
11

9. (a) Prelog, V., Pure Appl. Chem. 1964, 9, 119-130; (b) Sih, C. J.; Chen, C. S., Angew.
Chem., Int. Ed. Engl. 1984, 23, 570-578; (c) Vanmiddlesworth, F.; Sih, C. J., Biocatal.
Biotransform. 1987, 1, 117-127; (d) Fantin, G.; Fogagnolo, M.; Medici, A.; Pedrini, P.;
Poli, S.; Gardini, F.; Guerzoni, M. E., Tetrahedron: Asymm. 1991, 2, 243-246.
10. Hummel, W., Appl. Microbiol. Biotechnol. 1990, 34, 15-19.
11. Qi, Y.; Bawn, M.; Duncan, R.; Lloyd, R.; Finnigan, J.; Black, G.; Charnock, S., A rapid,
inexpensive and colorimetric high-throughput assay format for the screening of
commercial ketoreductase panels, providing indication of substrate scope, co-factor
specificity and enantioselectivity. In: Whittall, J.; Sutton, P.; Kroutil, W., eds. Practical
Methods for Biocatalysis and Biotransformations, vol. 3. Wiley: West Sussex, 2016, pp
266-276.
12. Theorell, H.; Nygaard, A. P.; Bonnichsen, R., Acta Chem. Scand. 1954, 8, 1490-1491.
13. (a) Wierzchowski, J.; Dafeldecker, W. P.; Holmquist, B.; Vallee, B. L., Anal. Biochem.
1989, 178, 57-62; (b) Wierzchowski, J.; Holmquist, B.; Vallee, B. L., Anal. Chem. 1992,
64, 181-6.
14. Klein, G.; Reymond, J.-L., Bioorg. Med. Chem. Lett. 1998, 8, 1113-1116.
15. Barski, O. A.; Tipparaju, S. M.; Bhatnagar, A., Drug Metabol. Rev. 2008, 40, 553-624.
16. Scheidt, T.; Land, H.; Anderson, M.; Chen, Y.; Berglund, P.; Yi, D.; Fessner, W.-D.,
Adv. Synth. Catal. 2015, 357, 1721-1731.
17. Venkataraman, S.; Narayan, S.; Chadha, A., Sci. Rep. 2016, 6, 34344.
18. Yee, D. J.; Balsanek, V.; Sames, D., J. Am. Chem. Soc. 2004, 126, 2282-2283.
19. (a) Giordano, C.; Villa, M.; Annuziata, R., Synth. Commun. 1990, 20, 383 - 392; (b)
Haworth, R. D.; Sheldrick, G., J. Chem. Soc. 1934, 864-867; (c) Johnson, W., J. Am.
Chem. Soc. 1950, 72, 925-935.
20. (a) Bouzemi, N.; Aribi-Zouioueche, L.; Fiaud, J.-C., Tetrahedron: Asymm. 2006, 17, 797-
800; (b) Zaidi, A.; Merabet-Khelassi, M.; Aribi-Zouioueche, L., Catal. Lett. 2015, 145,
1054-1061.
21. Kazlauskas, R. J.; Weissfloch, A. N. E.; Rappaport, A. T.; Cuccia, L. A., J. Org. Chem.
1991, 56, 2656-2665.
22. (a) Leuchs, S.; Greiner, L., Chem. Biochem. Eng. Quart. 2011, 25, 267-281; (b) Hummel,
W., Adv. Biochem. Eng. Biotechnol. 1997, 58, 145-184.
23. (a) Naik, H. G.; Yeniad, B.; Koning, C. E.; Heise, A., Org. Biomol. Chem. 2012, 10,
4961-4967; (b) Borzecka, W.; Lavandera, I.; Gotor, V., J. Org. Chem. 2013, 78, 7312-
7317; (c) Rodriguez, C.; Borzecka, W.; Sattler, J. H.; Kroutil, W.; Lavandera, I.; Gotor,
V., Org. Biomol. Chem. 2014, 12, 673-681.
24. (a) Ernst, M.; Kaup, B.; Mueller, M.; Bringer-Meyer, S.; Sahm, H., Appl. Microbiol.
Biotechnol. 2005, 66, 629-634; (d) Trivedi, A. H.; Spiess, A. C.; Daussmann, T.; Buechs,
J., Biotechnol. Prog. 2006, 22, 454-458; (b) Hildebrand, F.; Luetz, S., Tetrahedron:
Asymm. 2007, 17, 3219-3225; (c) Mueller, P.; Bangasser, B. L.; Greiner, L.; Na'amnieh,
S.; Baeuerlein, P. S.; Vogt, D.; Mueller, C., Open Catal. J. 2011, 4, 113-116.
25. (a) Niefind, K.; Muller, J.; Riebel, B.; Hummel, W.; Schomburg, D., J. Mol. Biol. 2003,
327, 317-328; (b) Schlieben, N. H.; Niefind, K.; Mueller, J.; Riebel, B.; Hummel, W.;
Schomburg, D., J. Mol. Biol. 2005, 349, 801-813.
26. Chenault, H. K.; Whitesides, G. M., Bioorg. Chem. 1989, 17, 400-409.
27. Edegger, K.; Mang, H.; Faber, K.; Gross, J.; Kroutil, W., J. Mol. Catal. A: Chem. 2006,
251, 66-70.
28. Schumacher, J.; Eckstein, M.; Kragl, U., Biotechnol. J. 2006, 1, 574-581.
29. (a) Analytical Methods Committee, Analyst 1987, 112, 199-204; (b) Long , G. L.;
Winefordner, J. D., Anal. Chem. 1983, 55, 712A-724A.
30. Chen, C. S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J., J. Am. Chem. Soc. 1982, 104, 7294-
7299.
12

31. Zhang, J.-H.; Chung, T. D. Y.; Oldenburg, K. R., J. Biomol. Screen. 1999, 4, 67-73.
32. (a) Kallberg, Y.; Oppermann, U.; Jörnvall, H.; Persson, B., Eur. J. Biochem. 2002, 269,
4409-4417; (b) Filling, C.; Berndt, K. D.; Benach, J.; Knapp, S.; Prozorovski, T.;
Nordling, E.; Ladenstein, R.; Jornvall, H.; Oppermann, U., J. Biol. Chem. 2002, 277,
25677-25684; (c) Persson, B.; Kallberg, Y.; Oppermann, U.; Jörnvall, H., Chem. Biol.
Interact. 2003, 143-144, 271-278; (d) Kavanagh, K. L.; Joernvall, H.; Persson, B.;
Oppermann, U., Cell. Mol. Life Sci. 2008, 65, 3895-3906; (e) Persson, B.; Kallberg, Y.,
Chem.-Biol. Int. 2013, 202, 111-115.
33. Rieche, A.; Seeboth, H., Liebigs Ann. Chem. 1960, 638, 66-75.
34. (a) Svensson, R.; Greno, C.; Johansson, A.-S.; Mannervik, B.; Morgenstern, R., Anal.
Biochem. 2002, 311, 171-178; (b) Chen, H.; Zhong, X.; Wei, J., Molecules 2007, 12,
1170-1182.
35. Fessner, W.-D., New Biotechnol. 2015, 32, 658-664.
Graphical Abstract
13