The organic-synthetic potential of recombinant Ene reductases: Substrate-Scope Evaluation and Process Optimization

Article abstract of DOI:10.1002/cctc.201402903

In this study an evaluation of the synthetic potential of a broad range of recombinant ene reductases was performed. In detail, a library of 23 ene reductases was used to screen the C=C reduction of 21 activated alkenes from different compound classes as substrates. The chosen set of substrates comprises nitroalkenes with an aryl substituent at the β-position and a methyl substituent at the α- or β-position, α,β-unsaturated carboxylic acids and their esters with and without substituents at the β-position, a range of cyclic α,β-unsaturated ketones with different ring sizes and substitution patterns and one α,β-unsaturated boronic acid. After we obtained insight into the substrate scope, several biotransformations were prioritised and further investigated in a screening of 41 reaction parameters (which included chaotropic and kosmotropic salts, polyols, buffer solutions, amino acids and organic solvents) towards their impact on the activity and enantioselectivity of the applied ene reductases. Under the optimised conditions, selected reduction processes were performed on an increased lab scale (up to 30 mL) with up to 10% substrate concentration, which led in general to both high conversion and (if chiral products were formed) enantioselectivity. Comprehensive screening: A detailed screening of 23 recombinant ene reductases for the reduction of 21 activated alkenes is performed as well as a subsequent study of the influence of 41 reaction parameters towards the enzyme activity and selectivity of selected reactions. In addition, a range of biocatalytic reductions on an increased laboratory scale are performed with substrate concentrations between 5 and 100 gL^-1. EWG=Electron-withdrawing group.

Full text of DOI:10.1002/cctc.201402903

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
DOI: 10.1002/cctc.201402903
Full Papers
The Organic–Synthetic Potential of Recombinant Ene
Reductases: Substrate-Scope Evaluation and Process
Optimization
Tina Reß,[a, c] Werner Hummel,[a, b] Steven P. Hanlon,[c] Hans Iding,*[c] and Harald Grçger*[a]
In this study an evaluation of the synthetic potential of
a broad range of recombinant ene reductases was performed.
In detail, a library of 23 ene reductases was used to screen the
C=C reduction of 21 activated alkenes from different com-
pound classes as substrates. The chosen set of substrates com-
prises nitroalkenes with an aryl substituent at the b-position
and a methyl substituent at the a- or b-position, a,b-unsaturat-
ed carboxylic acids and their esters with and without substitu-
ents at the b-position, a range of cyclic a,b-unsaturated ke-
tones with different ring sizes and substitution patterns and
one a,b-unsaturated boronic acid. After we obtained insight
into the substrate scope, several biotransformations were pri-
oritised and further investigated in a screening of 41 reaction
parameters (which included chaotropic and kosmotropic salts,
polyols, buffer solutions, amino acids and organic solvents) to-
wards their impact on the activity and enantioselectivity of the
applied ene reductases. Under the optimised conditions, se-
lected reduction processes were performed on an increased
lab scale (up to 30 mL) with up to 10% substrate concentra-
tion, which led in general to both high conversion and (if
chiral products were formed) enantioselectivity.
Introduction
During recent decades, ene reductases have become an effi-
cient tool in organic synthesis for the enantioselective reduc-
tion of activated C=C bonds.[1–16] In the beginning of the stud-
ies on ene reductases the major focus was the identification of
new ene reductases,[1,2] and numerous enzymes have been
successfully identified, characterised and overexpressed recom-
binantly. Although many of those ene reductases have been
applied in organic synthesis already, insight into their substrate
scope mostly remains not thoroughly explored because of the
use of only a narrow range of so-called model substrates.[1,2]
Recently, however, an increased tendency in organic synthesis
to apply ene reductases in target-driven asymmetric synthesis
has been seen.[1,3] In such multi-step syntheses, reductions cat-
alysed by ene reductase represent key steps within new retro-
synthetic approaches to complex chiral molecules, for example,
pharmaceuticals. Representative examples are new approaches
towards b2-amino acids[3a] (for which alternative chemocatalytic
routes are still rare) as well as nitroalkane intermediates[3b] for
the drugs tamsulosin and selegiline. Furthermore, ene reduc-
tases are also applied to an increasing extent in industry as un-
derlined by the recent development of such a process technol-
ogy at BASF[4] and recent process development work by Pfizer
researchers on a route towards pregabalin.[5,6] The current ac-
cessibility of a range of enzymes in recombinant form and an
expected broad synthetic utility (as alkenes activated by vari-
ous electron-withdrawing groups as substituents represent po-
tential substrates) are some of the advantages of ene reduc-
tases, which also make them attractive for applications in
route screening in pharmaceutical research. However, it would
be helpful to have a “synthetic-scope roadmap” of known re-
combinant ene reductases at hand to gain an insight into the
synthetic properties of such enzymes. Such comprehensive
screenings are rare in the literature with some notable excep-
tions from the Faber,[7] Bommarius,[8] and Bornscheuer and
Mihovilovic groups.[9] Herein we report our results on the eval-
uation of 23 ene reductases available in recombinant form for
the reduction of 21 substrates from different types of com-
pound classes. In addition, we also report the process optimi-
sation of selected biotransformations with prioritised ene re-
ductases that result from the screening study.
[a] T. Reß, Prof. Dr. W. Hummel, Prof. Dr. H. Grçger+
Faculty of Chemistry
Bielefeld University
Universitätsstr. 25, 33615 Bielefeld (Germany)
E-mail: harald.groeger@uni-bielefeld.de
[b] Prof. Dr. W. Hummel
Institute for Molecular Enzyme Technology (IMET)
Heinrich-Heine-Universität Düsseldorf
Research Center Jülich, 52425 Jülich (Germany)
[c] T. Reß, Dr. S. P. Hanlon, Dr. H. Iding
Process Research & Development, Biocatalysis
F. Hoffmann-La Roche Ltd.
Grenzacher Str.124, 4070 Basel (Switzerland)
E-mail: hans.iding@roche.com
[+] Temporary address:
Institute of Scientific and Industrial Research (ISIR)
Osaka University
Mihogaoka, Ibaraki-shi, Osaka 567-0047 (Japan)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201402903.
ChemCatChem 2015, 7, 1302 – 1311
1302
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
Results and Discussion
Table 1. Screening enzymes, E. coli was used as host organism.
Entry
Enzyme[a]
Strain
Expression system
Enzyme screening
1
2
3
4
5
6
7
8
9
OPR1
OYE 2.6
OYE3
NCR
DBVPG
Lycopersicon escultentum
Pichia stipitis
pET28 or pQR1[b]
pET28 or pQR1[b]
pET28 or pQR1[b]
pET28 or pQR1[b]
pET28 or pQR1[b]
pET28 or pQR1[b]
pET21a[c]
As one of the exciting features of ene reductases is
their tolerance for a range of electron-withdrawing
substituents at the C=C bond, we decided to test
a variety of types of activated alkenes (21 substrates
in total; Figure 1). In detail, we chose several nitro-
alkenes (1) that bear an aryl substituent at the b-po-
sition and a methyl substituent at the a- or b-posi-
tion (as the resulting nitroalkane products can be
converted by established methods into the corre-
sponding amines, which also represent a class of
products of pharmaceutical interest), a,b-unsaturat-
ed carboxylic acids and their esters with and without
Saccharomyces cerevisiae
Zymomonas mobilis
Kazachstania lodderae
Pseudomonas putida
Gluconobacter oxydans
–
XenB
GOx-ER
Ene-101 to Ene-116
BL21(DE3)
[d]
–
–
[b]
E. coli
[a] all enzymes were overexpressed in E. coli and used as cell-free crude extracts;
[b] preparation performed by F. Hoffmann-La Roche Ltd., Biocatalysis research group;
[c] preparation performed by the Hummel group; [d] purchased from Codexis.
substituents at the b-position (2–5 and 7; which lead, e.g., to
profene-type structures after C=C reduction in the case of 4
and 5), a range of cyclic a,b-unsaturated ketones (8–12) with
different ring sizes and substitution patterns and one boronic
acid (6). An overview of the selected substrates 1–12 is given
in Figure 1.
zymes as in the latter case additional enzyme purification
step(s) are required, which make such a “biocatalyst formula-
tion” less economically attractive. This is especially true if
crude extracts of enzymes with high overexpression data are
used. In addition, we also investigated the activity of the crude
extract from native E. coli cells for each substrate because of
potential “background reactions” of ene reductases from E. coli
(as such an undesired side reaction has been reported[3b]).
In such a case, the use of highly purified enzymes would be
preferred compared to the use of crude extracts to avoid un-
desired side reactions.
We used a library of 23 ene reductases that consist of litera-
ture-known recombinant ene reductases (OPR1,[10] OYE 2.6,[11]
XenB,[12] NCR,[13] DBVPG,[14] OYE 3,[6] GOx-ER;[15] these enzymes
were obtained by overexpression in E. coli; an overview of
these enzymes and the applied plasmid systems is given in
Table 1, entries 1–7 and 9) and 16 commercial ene reductases
purchased from Codexis (Ene-101 to Ene-116;[16] Table 1,
entry 8) as enzymes for the substrate screening. The enzymes
were used as cell-free crude extracts (for overexpression of se-
lected ene reductases, see Supporting Information) as this
offers economic advantages over the use of highly purified en-
The screening was performed on a 0.5 mL scale using
NADPH as a cofactor, glucose as co-substrate and a glucose
dehydrogenase (GDH) to recycle the cofactor in situ. The re-
duction of each substrate was studied with all of the enzymes
listed in Table 1.
The result of the initial screening of the 21 alkene substrates
1–12 with the 23 ene reductases is shown in Figure 2. The
data given in the coloured boxes are related to the conver-
sions with respect to the formation of the desired product,
and the colours indicate the degree of conversion (which
range from green for high conversion to red for no or very low
conversion). A qualitative graphical indication if suitable en-
zymes were found for the corresponding compound classes is
given in Figure 1 (blue: reasonable to high activity; red: no or
low activity).
To start with the reduction of nitroalkenes, most of the en-
zymes accepted the investigated substrates 1a–d. In particular,
a-substituted nitroalkenes 1a–c were suitable substrates for
most of the ene reductases, whereas a stronger dependence
on the type of enzyme was observed if we used a-unsubstitut-
ed and b-disubstituted nitroalkene 1d as a substrate. Next, the
enantioselectivity exemplified for nitroalkene 1a was investi-
gated, which led to the identification of the two promising en-
zymes NCR and Ene-111 that showed both high activity and
enantioselectivity. Notably, the use of NCR and Ene-111 gave
the opposite enantiomers of the corresponding nitroalkane
(71% ee for the R enantiomer if NCR was used, and 84% ee for
the S enantiomer if Ene-111 was used), which is of synthetic in-
terest as both absolute configurations can be formed selective-
ly by this reduction method catalysed by ene reductase.
Figure 1. Overview of the substrates used in the ene reductase screening.
ChemCatChem 2015, 7, 1302 – 1311
www.chemcatchem.org
1303
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
Figure 2. Results of the ene reductase screening. The numbers indicate product-related conversion (%; determined by GC or HPLC, for details, see the Support-
ing Information; defined as the ratio of the amount of product to the sum of the amounts of (remaining) substrate, (formed) product and all side products).
In contrast to the successful reduction of nitroalkenes with
a range of enzymes, cis- and trans-cinnamic acid (2a,b) as well
as trans-ester 3a gave no or a negligible conversion in all
cases (0–1% conversion). The more activated cinnamate deriv-
ative 3b led to very low conversions of 4% at best. At first
glance, it appears that a carboxylic acid function does not suf-
ficiently activate the C=C bond for reduction. However, in con-
trast for the a-phenyl-substituted carboxylic acid and ester an-
alogues 4 and 5, the ene reductase screening yielded promis-
ing candidates for each substrate albeit with a low conversion
of approximately 10–15% in most cases. Two enzymes,
OPR-1 and Ene-108, accept acid 4 as well as ester 5. This result
is noteworthy as an acid (in deprotonated form) and ester
moiety are very different types of functional groups with differ-
ent requirements for binding in the active site of an enzyme.
The highest conversion was found if GOx-ER was used with
54% conversion for the free acid 4, although ester 5 was not
tolerated as a substrate in this case. Interestingly, the enantio-
selectivities were high. For example, the reduction of 5 was
performed with >99% ee using Ene-114, and GOx-ER, which is
the only enzyme that showed activity towards 4, also gave an
excellent >99% ee.
thoxy as a substituent in the b-position of both the cyclic five-
and six-membered ketones (8 and 11) led to no or negligible
activity (0–1% conversion) with all of the applied enzymes, b-
methyl-substituted cyclopentanone 10 was tolerated by some
of them to lead to conversions of up to 49%. In addition, the
use of cyclic enone 9, which bears a methoxy substituent in
the a-position, revealed numerous enzymes as suitable cata-
lysts to lead to the reduced product with up to 89% conver-
sion. Thus, it appears that cyclic ketones that are not substitut-
ed in the b-position are favoured substrates. As an example of
a non-cyclic enone, compound 12 was studied and high con-
versions were achieved with a range of enzymes.
Reaction parameter screening
Next, we prioritised the most promising hits from the enzyme
screening and for each of these biotransformations (which are
based on the use of substrates 1a, 4, 7b and 12) we studied
the influence of 41 reaction parameters on reactivity and enan-
tioselectivity. The biotransformations were performed on
a 1 mL scale using a glucose/GDH system to recycle NADPH
in situ. As reaction parameters different buffer solutions (ar-
ranged in order of pH), chao- and kosmotropic salts (arranged
according to their position in the Hofmeister series; see Sup-
porting Information), polyols, amino acids and organic solvents
(arranged according to their logP value; see Supporting Infor-
mation) were investigated. Additives often have a significant
impact on biotransformations, for example, as demonstrated
by the influence of salt and solvent additives on enzyme activi-
ty studied by Dordick et al.,[17] Klibanov,[18] Halling et al.[19] and
Bommarius et al.[20] For the example, in the latter case the in-
fluence of salt additives along the Hofmeister series as well as
water-immiscible organic solvents on the enzyme activity of
glucose dehydrogenase mutants was studied intensively.[20]
For the a,b-unsaturated boronic acid 6, however, no enzyme
was suitable, whereas for (substituted and non-substituted) a-
cyano cinnamic acid and esters 7a–c a range of enzymes cata-
lysed the desired C=C bond reduction with conversions of up
to 89%. Although only moderate activities were found towards
7a, in general, higher activities were observed if the alkene
bears an ester instead of an acid moiety (7b and 7c). For ex-
ample, for 7b as a model substrate, the highest activities were
obtained with Ene-107 and Ene-114.
In the screening of different types of ketones (8–12)
a strong dependence of the activity of the enzymes on their
substitution pattern was observed. Although a halogen or me-
ChemCatChem 2015, 7, 1302 – 1311
www.chemcatchem.org
1304
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
As a first reaction, we focused on the reduction of nitro-
alkene 1a with the enzyme NCR (Figure 3). If NCR was tested
with 1a as the substrate, the initial “standard experiment” that
was performed in piperazine-N,N’-bis(2-ethanesulfonic acid)
(PIPES) buffer gave 35% conversion and 31% ee with prefer-
ence for the R enantiomer. Among the salt additives, none of
the applied salts led to an increase of the conversion. Al-
though KCl and LiSCN caused the most significant decrease in
activity, the kosmotropic sulfates were able to increase the
enantioselectivity up to 45% ee (K2SO4). With regard to the
tested water-miscible solvents, there was no significant correla-
tion between logP and conversion. In biphasic systems, sol-
vents with highest logP values gave the best results (cyclohex-
ane: logP=3.35, 26% conversion; heptane: logP=4.40, 26%
conversion). In general, however, organic solvents and salt ad-
ditives did not result in a significant improvement (and some-
times caused a loss of activity and/or selectivity). We found
that polyethylene glycol (PEG) had a positive influence on the
reactivity and led to 42% conversion at expense of the enan-
tioselectivity, which decreased to 15% ee. With regard to the
enantioselectivity, a positive influence of 2-amino-2-hydroxy-
methyl-propane-1,3-diol (Tris) buffer was noted (20% conver-
sion, 52% ee). Notably, a 20-fold increase of the amount of
enzyme not only led to a higher conversion but also an in-
creased enantioselectivity of up to 72% ee (Figure 3). The com-
bination of beneficial parameters, namely, Tris buffer and
a high enzyme loading, yielded an improved enantiomeric
excess of 80% ee (Supporting Information).
This enantioselectivity of 80% is an excellent value as it is
known[3b] that there is an E. coli background reactivity caused
by the native E. coli ene reductase NemA in the case of nitro-
alkene reduction. In this study we gained an insight into the
impact of this side reaction with 1b as a substrate (Supporting
Information) and found that the crude cells lysate produced
the opposite enantiomer preferably compared to that if the
ene reductase NCR was used. However, the conversions
achieved with the E. coli crude cell extract were very low so
the background reaction does not have a significant negative
impact on the reduction if the ene reductase NCR is used
(Figure 3). According to the initial screening (Figure 2), the
E. coli background reactivity is negligible for most substrates
except the nitroalkenes and 7c. Furthermore, sodium dodecyl
sulfate (SDS) gels of most applied enzyme formulations show
good overexpression (Supporting Information), which is gener-
ally associated with a low impact of the E. coli background re-
action. However, an effect that has to be considered is the
racemisation of the produced nitroalkane (R)-13 (which was
addressed in the later process optimisation).
The reduction of nitroalkene 1a was also studied with the
ene reductase Ene-111 as the opposite enantiomer of the re-
sulting product 13 is produced with this enzyme. Thus, with
enzymes NCR and Ene-111 enantio-complementary biocatalysts
are available that give selective access to both enantiomeric
forms of nitroalkanes of type 13. In the presence of Ene-111,
the substrate 1a was reduced to (S)-13 with 21% conversion
and 18% ee under the initial standard reaction conditions
(Figure 4). Notably, there was a positive influence of glycerol
and glycine as additives on the enantioselectivity. For example,
the use of 5% of glycine gave a significantly improved enan-
tioselectivity of 34% ee at a similar conversion of 19%. A dra-
matic negative impact on both conversion and enantioselectiv-
ity, however, was observed with a range of salts. The best con-
versions with salt additives were achieved with phosphates (20
and 19% conversion, respectively), but all other salts de-
creased enzyme activity significantly. In addition, the use of
any type of (water-miscible or -immiscible) organic solvent led
to a strong decrease of enantioselectivity and often also to
a very low conversion. For example, acetonitrile gave 28% con-
version and 0% ee, whereas 2-propanol led to 26% conversion
and 3% ee. We found no correlation between logP and conver-
Figure 3. Reaction parameter screening with substrate 1a and ene reductase
NCR (enzyme”10 (20) means the use of a 10 (20)-fold amount of enzyme
compared to the other experiments, see also Experimental Section).
ChemCatChem 2015, 7, 1302 – 1311
www.chemcatchem.org
1305
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim