ˇ
´
T. Petrovicova et al.
Molecular Catalysis 502 (2021) 111364
recombinant KREDs produce the (2R,3S) or the (2R,3R) diastereomer
(KREDs 102, 103, 106 and 117 and KREDs 107and 121 respectively) for
the same keto esters 1a and 1b [29]. In particular, in the reduction of
ethyl-2-methyl acetoacetate (1a, Table 2, Entry 1), the optically active
(2S,3S)-hydroxy ester was produced and KRED Hansenula polymorpha
showed higher enantioselectivity (86 % ee) compared to the previously
reported commercially available KRED107 (70 % ee) [30]. Recently,
Wei et al. reported the reduction of methyl acetoacetate for the pro-
duction of methyl (R)-3-hydroxybutyrate with 87 % conversion by using
an anti-Prelog stereospecific carbonyl reductase [25].
stereoselectivity was de 91 % and ee 51 %; however, the reduced
products were not analysed further. As shown in Table 2, no activity was
observed with ketones 14a and 15a (Entries 35, 36, Table 2). Only two
enzymes (KRED 101 and 132) have been reported to show activity to-
wards 14a (1.48 and 2.20 U. mg–1, respectively), but ee was not deter-
mined [84]. In the cases where the KRED showed low substrate
conversions, the absolute configuration of the products was not deter-
mined and therefore it was not possible to compare with literature data.
In order to further determine the absolute configuration of the reduced
products, scale-up of reactions (15 mL) was performed for nine selected
substrates from every group (keto esters, aryl ketones, aliphatic ketones)
with which KRED showed good conversion and enantioselectivity dur-
ing substrate screening. Determination of the absolute configuration of
the optically pure hydroxy products was performed by a modified
Mosher’s method [52–58], and the results are summarised in Table 3.
Interestingly, the KRED showed R-enantiopreference with the
following ketones: para-Me-acetophenone (2b), acetophenone (2d), 2-
heptanone (3a), 2-octanone (3b) and 3-octanone (3d) as well as with
4-phenylbutan-2-one (9a) (Table 3, Entries 4–9). In contrast, in the case
of β-keto ester 1e (Table 3, Entry 3), the KRED showed S-enantiopre-
ference and the methyl (S)-3-hydroxybutanoate was produced
exclusively.
The structure of the screened aryl alkyl ketones varied in terms of the
position of substituents on the aromatic ring: ortho (2f, 2 g, Table 2,
Entries 14, 15) meta (2a, Table 2, Entry 9) and para (2b, 2c, 2e, Table 2,
Entries 10, 11, 13) position. The enzyme showed no activity towards
ortho-substituted acetophenones (o-methyl, 2f, Entry 14 and oꢀ OH, 2 g,
Entry 15, Table 2), most probably due to steric hindrance [25,81]. The
enzyme showed increased activity, with conversions between 74 % and
>99 % and high enantioselectivity (90 %–>99 % ee), for the meta- and
para-substituted acetophenones (2a, 2b, 2c, 2e, Table 2, Entries 9, 10,
11, 13). The lowest enantioselectivity between the group of acetophe-
none and its derivatives was observed for acetophenone 2d (60 % ee,
Table 2, Entry 12). Previous reports had shown increased enzymatic
activity with bromine and other halogens or electron-withdrawing
groups as para substituents [11,25]. In our case, the KRED showed
slightly decreased activity with para-Br-acetophenone (2e; 74 % con-
version, 95 % ee, Table 2, Entry 13). Excellent conversion was derived
with ketones 2h–j (conv. > 99 %, Table 2, Entries 16–18), which con-
tained –Et, ꢀ CH2OH and ꢀ CH2OAc next to the carbonyl group. The
highest enantioselectivities were observed with hydroxy ketone 2i (>99
% ee, Table 2, Entry 17) and metaꢀ OH substituted acetophenone (2a),
(>99 % ee, Table 2, Entry 9). This confirms the positive effect of
electron-withdrawing groups if there are suitable spatial arrangements
in the active centre of the enzyme. According to previously published
data with recombinant protein (KRED1-Pglu) overexpressed in E. coli
[26], longer chains connected to the carbonyl group resulted in
decreased enzymatic activity. Stereoselective reduction of aliphatic ke-
tones with substituents longer than four carbon atoms on both sides of
the carbonyl group or remote aliphatic keto esters still remains chal-
lenging [82,83]. Among the aliphatic ketones, the highest conversion
(>99 %) and excellent enantioselectivities up to >99 % ee were achieved
for acyclic ketones 3a, 3b, 3d (Table 2, Entries 19, 20, 22), and good
enantioselectivity was observed with hydroxy ketone 3c (66 % ee,
Table 2, Entry 21). Moreover, with 4-phenylbutan-2-one 9a and methyl
ketones 8a and 13a, the KRED showed excellent activity and enantio-
selectivity (conv. > 99 % and 88– >99 % ee, Table 2, Entries 29, 30, 34).
This result, compared to the obtained data for acetophenone (2d; 80 %
conv., 60 % ee, Table 2, Entry 12), shows that the KRED can potentially
reduce more efficiently those methyl ketones bearing longer or bulkier
aryl substituents next to the carbonyl group. The enzyme activity
decreased dramatically in the case of the methyl ketone 3e (tridodeca-
n-2-one, Table 2, Entry 23), where there is a long nonpolar carbon chain
of 11 carbon atoms, assuming that spatial limits in the active site of the
enzyme were involved [84]. The position of the methyl group on the
cyclohexanone ring in the cases of 2- and 3-methyl cyclohexanones (4a
and 4b, respectively) was an important factor in their conversion
(Table 2, Entries 24, 25). Conversion of 3-methyl-cyclohexanone (4b)
was 5-fold higher than that of 2-methyl-cyclohexanone (4a) after 24 h
(conv. 99 % compared to conv. 18 %), most probably due to the
increased steric hindrance of the methyl group in 4a. Our further sub-
strate specificity investigations with the KRED from Hansenula polu-
morpha showed high activity with most of the tested substrates (4b,
5a–8a, Table 2, Entries 25–29). The enantioselectivity was excellent
with diketone 6a and ketone 13a (>99 % ee, Table 2, Entries 27, 34).
The diastereoselectivity for the reduced products of the chiral bicyclic
ketone 11a (Table 2, Entry 32) could not be determined because of
insufficient GC separation. For the chiral diketone 7a, the observed
These results suggest anti-Prelog selectivity for the enzyme (Table 3),
which is quite rare with these types of substrates [85]. We would like to
emphasise here that previous attempts to reduce 2-octanone (substrate
3b) with commercial ketoreductases [84] and genetically engineered
KREDs [40] showed that most of these enzymes did not display signif-
icant activities. The successful production of the R-enantiomer was re-
ported for whole-cell biocatalyst Acetobacter pasteurianus GIM1.158 (95
% yield; >99 % ee) [85] and Candida rugosa (92 % yield; 95 % ee) using
ionic liquids as solvents to increase the solubility of the substrate [86].
There are few reports for the enzymatic production of 3-octanol, of
which the most successful was when Lactobacillus brevis ADH was used,
showing 84.6 % conversion and >99 % ee for (R)-3-octanol [87] and
when E. coli cells with ADH from Rastolnia sp. were used to produce
(S)-3-octanol (98 % ee) [36]. 3-octanone (3d) was successfully reduced
using the KRED from Hansenula polymorpha with high conversion and
excellent enantioselectivity (conv. 97 %, >99 % ee, Table 3, Entry 8) and
the 3-octanol produced had R configuration, as determined by chiral GC
analysis and after comparison of the retention time with the corre-
sponding data for the commercially available optically active
(R)-3-octanol.
(R)/(S)-3-octanol and other aliphatic chiral alcohols are bioaroma
compounds and could be used in the fragrance industry. Also, (R)-3-
octanol is a known pheromone that acts as sex attractant for Myrmica
scabrinodis [88].
All the above results indicated that the KRED from Hansenula poly-
morpha is a very versatile enzyme with a broad substrate specificity and
high activity towards carbonyl substrates with various structural fea-
tures (Table 2). Among the 36 carbonyl substrates which were screened
in this study (Table 2), the KRED showed activity with 31 and no activity
was found with only five substrates.
Conclusions
Ketoreductase from Hansenula polymorpha was successfully overex-
pressed in Escherichia coli BL21(DE3) in a high cell density process, with
biomass concentration 49.7 g L–1 and high specific activity (2220.1 ±
16.1 U g–1 DCW). The purified enzyme showed activity towards 31 sub-
strates, including aliphatic and aromatic ketones, acetophenone and
substituted acetophenones, aryl alkyl ketones, diketones and β-keto es-
ters. Interestingly, the KRED from Hansenula polymporpha catalysed the
enantioselective reduction of 2-heptanone, 2-octanone, 3-octanone, p-
Me-acetophenone and 4-phenyl-2-butanone for the production of the
optically active corresponding alcohols, with remarkable conversions
(>99 %) and excellent enantioselectivities (>99 % ee). In these
7