.
Angewandte
Communications
the expected further increase of the enan-
tioselectivity up to 83% ee for the forma-
tion of product 1a (Scheme 3).[15]
Surprisingly, during investigation of the
synthetic process the adjustment of three
more parameters, which are often consid-
ered to be less important in asymmetric
biocatalysis, led to further remarkable
improvements. Thus, the enantioselectivity
could be increased by using the purified
recombinant ene reductase at decreased
reaction temperatures of 5–158C. Here it
should be noted that in contrast to chemo-
catalytic processes the increase of the
Scheme 2. Synthetic concept for the construction of chiral amines of type (R)- or (S)-2 based
on an enantioselective catalytic reduction of nitroalkenes 5.
reductases, this enzyme, which we first used in the form of
a crude extract typical for biotransformations, turned out to
be very suitable already in the initial experiments for the
reduction of trans-nitroalkene 5a. For instance, for the
reduction of 5a the formation of product (R)-1a[11] with an
enantioselectivity of 78% ee (which is already high for this
type of reaction) was observed (Scheme 3). However, we also
found that the enantioselectivity changed with the increasing
enantioselectivity of biotransformations by lowering the
reaction temperature is known to a much lesser extent.[16]
However, the observed beneficial effect of a decreased
reaction temperature of 5–158C on the ee value can also be
caused by a decelerated racemization of nitroalkane 1a.
Furthermore, shortening the reaction time also improved the
enantioselectivity. Interestingly, we could exclude a racemiza-
tion of the formed nitroalkane 1a caused by the reaction
medium, the glucose dehydrogenase, and the inactivated ene
reductase (for the corresponding experiments, see the Sup-
porting Information). To ensure a high conversion and high
robustness of the synthetic process, ultrasound treatment of
the mixture of the nitroalkenes in buffer (prior to the addition
of enzymes) and further ultrasound treatment of the reaction
mixture after a reaction time of 5 h turned out to be
advantageous. This could be due to the improved mixing of
the heterogeneous reaction mixture containing the nitro-
alkene 5, which is hardly soluble in water. Subsequently, these
optimized reaction conditions were applied to the reduction
of a range of nitroalkenes 5 on an increased 100 mL scale
(Table 1). The desired stereoselective enzymatic reductions of
the m-halogen-substituted nitroalkenes 5a–c, which were
used initially as model substrates, proceeded with excellent
(overall) conversions of 96 to > 99% and very good
enantioselectivities of 93–95% ee (Table 1, entries 1–3). The
workup used for entries 2 and 3 led to the products (R)-1b,c,
in each case in a high yield of 84%. The reaction of the m-
methoxy-substituted nitroalkene 5d also proceeded with high
(overall) conversion (95%) and a high enantioselectivity of
93% ee, which is excellent for this kind of reaction (Table 1,
entry 4). This indicates—as desired with respect to a broad
application—that the enzymatic process proceeds with high
enantioselectivity independent of the type of substituent in
meta position. The reduction of the p-substituted nitroalkenes
5e,f, however, gave decreased enantioselectivities of 81% and
66% ee, respectively (Table 1, entries 5 and 6). In turn we
were pleased to find a high enantioselectivity of 90% ee for
the reaction of nitroalkene 5g bearing a nonsubstituted
phenyl group (Table 1, entry 7). Thus, with this practical and
easy-to-conduct enzymatic reduction process, a-methyl-sub-
stituted nitroalkenes of type 5 can be reduced enzymatically
for the first time with enantioselectivities exceeding 90% ee.
An additional focus of our work was on the integration of
the developed process into an improved total synthesis of
nitroalkenes 5 starting from aldehydes as substrates. Here, the
Scheme 3. Dependence of the ene reductase catalyzed reductions on
the purity of the biocatalyst. GDH=glucose dehydrogenase.
storage time of the enzyme.[12] We assumed that this surprising
result arises from an undesired “background activity” caused
by a further, non-enantioselective or minor enantioselective
ene reductase.[13] On examination of the known genome of
E. coli we assigned this background activity to the following
enzymes found in E. coli: nemA (accession no. NP_416167,
“Old Yellow Enzyme” family)[14] and NfsB (YP_002998379,
nitroreductase family). This explanation prompted us to study
the activity of the crude extract originating from the host
organism E. coli (without expression of the ene reductase
from G. oxydans). The use of this crude extract from E. coli
then led to the expected reduction of 5a to give 1a with low
enantiomeric excess (< 3% ee), which indicates a nearly non-
enantioselective reaction course. In order to avoid this
undesired background activity of the E. coli host organism,
we were interested in using the recombinant enzyme compo-
nent, which is overexpressed in E. coli, in purified form.
Toward this end, the ene reductase from G. oxydans was
heterogeneously overexpressed as an N-terminal hexahisti-
dine fusion construct in E. coli so that the enzyme could be
purified simply by immobilized metal ion affinity chromatog-
raphy. The use of this purified form of the enzyme resulted in
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 9323 –9326