Enzymatic reduction of acetophenone derivatives with a benzil reductase from Pichia glucozyma (KRED1-Pglu): electronic and steric effects on activity and enantioselectivity

Article abstract of DOI:10.1039/c6ob00047a

A recombinant ketoreductase from Pichia glucozyma (KRED1-Pglu) was used for the enantioselective reduction of various mono-substituted acetophenones. Reaction rates of meta- and para-derivatives were consistent with the electronic effects described by σ-Hammett coefficients; on the other hand, enantioselectivity was determined by an opposite orientation of the substrate in the binding pocket. Reduction of ortho-derivatives occurred only with substrates bearing substituents with low steric impact (i.e., F and CN). Reactivity was controlled by stereoelectronic features (C=O length and charge, shape of LUMO frontier molecular orbitals), which can be theoretically calculated.

Full text of DOI:10.1039/c6ob00047a

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
Enzymatic reduction of acetophenone derivatives
with a benzil reductase from Pichia glucozyma
(KRED1-Pglu): electronic and steric effects on
activity and enantioselectivity†
Cite this: DOI: 10.1039/c6ob00047a
Martina L. Contente,a Immacolata Serra,a Luca Palazzolo,b Chiara Parravicini,b
Elisabetta Gianazza,b Ivano Eberini,b Andrea Pinto,c Benedetta Guidi,d
Francesco Molinaria and Diego Romano*a
A recombinant ketoreductase from Pichia glucozyma (KRED1-Pglu) was used for the enantioselective
reduction of various mono-substituted acetophenones. Reaction rates of meta- and para-derivatives
were consistent with the electronic effects described by σ-Hammett coefficients; on the other hand,
enantioselectivity was determined by an opposite orientation of the substrate in the binding pocket.
Reduction of ortho-derivatives occurred only with substrates bearing substituents with low steric impact
(i.e., F and CN). Reactivity was controlled by stereoelectronic features (CvO length and charge, shape of
LUMO frontier molecular orbitals), which can be theoretically calculated.
Received 8th January 2016,
Accepted 29th February 2016
DOI: 10.1039/c6ob00047a
www.rsc.org/obc
cular dynamics simulations suggest that low reactivity of ortho-
derivatives is mostly due to unfavourable intramolecular inter-
Introduction
Substituted 1-phenylethanols comprise an important struc- actions between the ortho carbons of the aromatic group and
tural class of compounds for synthetic purposes, and their the nucleophilic nitrogen, as well as between the stereocenter
obtainment by enantioselective reduction of the corresponding and the enzymatic oxyanion. Moreover, the analysis of the
acetophenones has been thoroughly studied using chemo- solvent accessible surfaces of the hydrogens of the phenyl ring
catalysis1 and biocatalysis.2,3 Various microbial alcohol explains why para and meta substituents are good substrates,
dehydrogenases (ADHs), carbonyl reductases (CRs), and whereas ortho-substituted substrates were not accepted.5
ketoreductases (KRs) are able to reduce acetophenone deriva-
Acetophenone derivatives are generally reduced with oppo-
tives giving access to both the enantiomers of the corres- site enantioselectivity depending on the enzyme employed. An
ponding 1-phenylethanols.4 Reactivity and stereoselectivity of ADH from Rhodococcus erythropolis proved efficient in the
the enzymatic reduction of acetophenones depend on the reduction of a number of mono-, di- and tri-substituted aceto-
steric and electronic properties of the substituent on the aro- phenones, giving always the corresponding (S)-1-phenyletha-
matic ring, which will largely affect the rate of the reaction; nols (Prelog rule);6 the same stereopreference was observed
other specific interactions (e.g. hydrogen bonding, electrostatic with other ADHs, i.e., from Ralstonia sp.7 and from Thermoa-
interactions) between the substituents and the enzyme may defi- naerobacter sp.8 A general preference for the Prelog rule was
nitely contribute to the overall reactivity. The latter effect has also detected in a screening with 24 recombinant KRs tested
been reported for other enzymatic reactions, such as the lipase- on thirteen mono-substituted acetophenones.9 Conversely,
catalysed acylation of 2-aryl-1,3-propanediamines, where mole- optically pure (R)-1-phenylethanol can be prepared by the
reduction of unsubstituted acetophenone using specific
enzymes, such as LK-ADH from Lactobacillus kefir,10 PF-ADH
aDepartment of Food, Nutritional and Environmental Sciences (DeFENS), University
from Pseudomonas fluorescens11 and carbonyl reductase from
of Milan, Via Mangiagalli 25, 20133 Milano, Italy. E-mail: diego.romano@unimi.it
bDepartment of Pharmacological and Biomolecular Sciences (DiSFeB), University of
Candida magnoliae (CMCR).12
Systematic studies on the effect of ring-substituents on
enzymatic catalytic efficiency are quite rare. In fact, steric hin-
drance and electronic properties of acetophenone derivatives
are largely dependent on the type and position of the substi-
tuent, thus affecting their overall reactivity, no matter if a
chemo- or a bio-catalyst is involved. A number of mono-substi-
Milan, Via Balzaretti 9, 20133 Milano, Italy
cDepartment of Pharmaceutical Sciences (DISFARM), University of Milan, Via
Mangiagalli 25, 20133 Milano, Italy
dDepartment of Medical Biotechnology and Translational Medicine (BIOMETRA),
University of Milan, Via Vanvitelli 32, 20129 Milano, Italy
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c6ob00047a
This journal is © The Royal Society of Chemistry 2016
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
tuted acetophenone derivatives were studied using Candida
tenuis xylose reductase (CtXR), which showed, in all the cases,
preference for the formation of (S)-1-phenylethanols.13 For this
enzyme, σ-Hammett coefficients of para- and meta-substituents
were in good correlation with the reaction velocities, indicating
again that enzyme–substrate binding has a subordinate influ-
ence on the CtXR catalytic efficiency on these substrates. Inter-
estingly, ortho-derivatives were sometimes reduced with higher
rates, indicating the occurrence of a beneficial mode of
binding with favourable electronic interactions between the
substrate and the enzyme. An interesting effect of the nature
of the para-substituents was observed with ADH from Lacto-
bacillus kefir and Thermoanaerobacter sp., by which aceto-
phenones with neutral para-substituents were efficiently and
selectively reduced, whereas those with ionizable substituents
could not be reduced.14 A recombinant short-chain dehydro-
genase from Saccharomyces cerevisiae showed excellent (S)-
enantioselectivity with a series of mono-substituted aceto-
phenones, irrespective of the nature of the substituent. Notably,
in the reduction of ethyl 2-oxo-2-phenylacetate, the enantio-
selectivity of this enzyme was strongly affected by the presence
of substituents on the aromatic ring: with any substituent on
the phenyl ring, the (R)-alcohol was produced, whereas the
unsubstituted ethyl 2-oxo-2-phenylacetate gave the (S)-
alcohol.15 These results show that not only reactivity, but also
enantioselectivity may be influenced by the presence of substi-
tuents on the aromatic ring.
Table 1 Reduction of para-substituted acetophenones with KRED1-
Pglu
Entry
Substrate
X
Rel. rate (%)23
Yield (%)24
ee (%)
1
2
3
4
5
6
7
8
1a
1b
1c
1d
1e
1f
NH2
OCH3
CH3
H
F
CF3
CN
0
16
23
22
35
51
68
100
0
20
30
40
40
42
70
100
—
60 S
>98 S
95 R
93 S
95 S
97 S
>98 S
1g
1h
NO2
substituted acetophenones (Table 2), having the same substi-
tuents of the para-acetophenones previously tested (1i–1o). In
this case, all the substrates did react, including the amino
derivative 1i. All the meta-derivatives were reduced to the
corresponding (S)-1-phenylethanols with high-to-excellent
enantioselectivity. Notably, 3′-methoxyacetophenone 1j was
reduced to enantiopure (S)-2j, whereas the corresponding para-
derivative 1b showed moderate enantioselectivity.
The σ-Hammett coefficients were used to assess the influ-
ence of the nature of the para- and meta-substituents on the
kinetics of the hydride transfer to acetophenone. The initial
rates were plotted against the σ-values of the substituents; σpara
In this study we have systematically investigated the effect
of the steric and electronic properties of the substituent on the
reduction of different mono-substituted acetophenones cata-
lysed by a new recombinant ketoreductase (KRED1-Pglu).16,17
and σmeta were independently correlated (Fig. 1). σpara
-
Hammett coefficients were in excellent correlation with the
reaction rates, showing that the activity of KRED1-Pglu was
predominantly dependent on the electronic properties, while
the substrate size and/or hydrophobicity had negligible effects.
The reaction rate was accelerated by the electron withdrawing
groups (ρ > 0), as a consequence of the negative charge in the
transition state caused by the attack of the hydride on the
carbonyl group.
Results and discussion
The recombinant benzil ketoreductase KRED1-Pglu was iso-
lated from the non-conventional yeast Pichia glucozyma;16
whole cells of P. glucozyma enantioselectively reduced different
aromatic ketones,18–20 including acetophenone, with a marked
preference for the formation of (S)-1-phenylethanol;21,22
interestingly, KRED1-Pglu was isolated from the same yeast
reduced acetophenone as a recombinant protein with opposite
enantioselectivity. KRED1-Pglu was initially investigated using
different para-substituted acetophenone derivatives (1a–1c and
1e–1h); acetophenone (1d) was considered as a reference
(Table 1). The substrates were transformed with very different
rates, except for 4′-aminoacetophenone 1a (entry 1, Table 1),
which was the only substrate not to be processed. Reduction
followed the Prelog rule for all the substituted acetophenones,
giving the corresponding (S)-1-phenylethanols; most of the
biotransformations occurred with high enantioselectivity, with
the exception of 4′-methoxyacetophenone 1b (entry 2, Table 1),
which was reduced with only 60% ee. The observed stereo-
preference is opposed to the one obtained with acetophenone 1d,
which was reduced to (R)-1-phenylethanol 2d with 95% ee.
KRED1-Pglu was also used for the reduction of a set of meta-
Table 2 Reduction of meta-substituted acetophenones with KRED1-
Pglu
Entry
Substrate
X
Rel. rate (%)23
Yield (%)24
ee (%)
1
2
3
4
5
6
7
8
1i
1j
NH2
OCH3
CH3
H
F
CF3
CN
7
30
15
22
61
31
46
53
10
67
18
40
85
25
49
78
—
>98 S
>98 S
95 R
>98 S
94 S
95 S
>98 S
1k
1d
1l
1m
1n
1o
NO2
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 1 Correlation between activity and the Hammett σ-values.
In order to evaluate the electronic features of the tested sub- also correlated with the electronic charge density distribution
strates, a geometry optimization in the gas phase with an of the CvO bond (Fig. 2) and CvO length (Fig. 3). These
approach based on density functional theory (DFT/B3LYP/6- factors have been previously used as parameters for correlating
31G*) was performed, resulting in prediction of the CvO the activity of ADHs on different mono-substituted aceto-
lengths and of the Mulliken derived charges for the C atom phenones.12,13 A high correlation between activity and electronic
involved in the hydride attack of the substrates. A further features of the substituents was found with para-derivatives,
characterization was carried out through the analysis of the whereas a lower correlation could be detected with meta-
lowest unoccupied molecular orbitals (LUMO) in the investi- derivatives, as also seen with other enzymes.12 This behaviour
gated substrates. The activity of para- and meta-derivatives was is different from what was observed with ADHs from
Fig. 2 Correlation between activity and charge on the carbon of the CvO bond.
Fig. 3 Correlation between activity and CvO length.
This journal is © The Royal Society of Chemistry 2016
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Lactobacillus brevis and Thermoanaerobacter sp., with which the
electron withdrawing or donating properties of the substi-
tuents did not exert a significant effect.13
All these collected data indicate that para-derivatives are
reduced with reaction rates apparently not depending on steric
interactions of the substrates with the enzyme, whereas steric
effects have a (minor) impact only on the reactivity of meta-
derivatives. The co-occurrence of steric and electronic effects
for meta-derivatives was corroborated by the observation that
the preferred substrate was 3′-fluoroacetophenone 1l (bearing
the less sterically hindered substituent still being an electron
withdrawing group).
The occurrence of any substituent in the para- or meta-posi-
tion favours the delivery of the hydride to the Si face of the
ketone, whereas acetophenone 1d predominantly gives (R)-1-
phenylethanol as a result of an attack on the opposite prostereo-
genic face. These results are similar to those observed in the
reduction of ethyl 2-oxo-2-phenylacetate derivatives with the
recombinant alcohol dehydrogenase YMR226c from Saccharo-
myces cerevisiae: in this case, any type of substituent on the
phenyl ring inverted the substrate stereorecognition in the
enzyme active site in comparison with the unsubstituted
substrate.14
Fig. 4 LUMO density (panels a, b, c) and DFT-based geometry optimi-
zation studies (panels d, e, f) for 1d, 1u and 1v molecules.
Finally, with our tests, we assayed ortho-benzil derivatives
(Table 3). ortho-Substituents introduce significant steric effects
in the reactivity of acetophenone derivatives, and therefore a
simple Hammett’s approach to quantify the substituent effects
on the reaction center cannot be applied. KRED1-Pglu showed
activity only towards 2′-fluoroacetophenone 1s and 2′-cyano-
acetophenone 1u, the substrates bearing electron withdrawing
substituents with the lowest steric impact. Low reactivity of
sterically demanding ortho-substituted acetophenones is gene-
rally observed with a number of recombinant ketoreduc-
tases.9,12 A notable exception is the Candida tenuis xylose
reductase (CtXR), which reduces ortho-substituted aceto-
phenones better than the correspondent para- and meta-deriva-
tives, provided the substituent had both low steric impact and
electron withdrawing effect (i.e., F, Cl, Br);13 CtXR has a wide
binding pocket, able to accommodate a wide variety of sub-
strates, including ortho-substituted acetophenones.25
In order to explain the different reactivity for the CvO
group of 2′-cyanoacetophenone 1u and 2′-nitroacetophenone
1v, we computed the LUMO frontier molecular orbitals, as
shown in Fig. 4, panels b (1u) and c (1v). The LUMO density in
1u (Fig. 4, panel b) is localized on the reactive CvO, whereas
no electronic density correlated with LUMO can be observed,
in the same region, for 1v (Fig. 4, panel c).
In panels e and f, Fig. 4 also shows the results of a DFT-
based geometry optimization on 2′-cyanoacetophenone 1u and
2′-nitroacetophenone 1v. The almost perfect co-planarity for
the 1u conjugated system (Fig. 4, panel e), not conserved in 1v
(Fig. 4, panel f) because of the steric hindrance between the
NO2 and the CvO group, is a further explanation of the
different reactivity of these two substrates. Data for aceto-
phenone 1d, are shown in Fig. 4, panels a and d, as a reference.
The LUMO frontier molecular orbitals for all the investigated
substrates are shown in Fig. S1†. It should be remarked that
substrates (such as 1q or 1v), which have an intrinsically low
reactivity due to electronic features, were poorly reduced even
with the “ortho-specialized” ketoreductase CtXR.13
Table 3 Reduction of ortho-substituted acetophenones with KRED1-Pglu
Entry
Substrate
X
Rel. rate (%)23
Yield (%)24
ee (%)
Conclusions
1
2
3
4
5
6
7
8
1p
1q
1r
1d
1s
1t
NH2
OCH3
CH3
H
F
CF3
CN
0
0
0
22
76
0
82
0
0
0
0
40
97
0
98
0
—
—
—
95 R
94 S
—
>98 S
—
In conclusion, with this series of comparative tests we showed
that a benzil ketoreductase from Pichia glucozyma (KRED1-
Pglu) is an efficient biocatalyst for the enantioselective
reduction of various mono-substituted acetophenones. Reac-
tion rates of meta- and para-derivatives were consistent with
the electronic effects described by σ-Hammett coefficients,
1u
1v
NO2
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Organic & Biomolecular Chemistry
Paper
with the DFT-Mulliken partial charges, and also with the CvO 11 P. Hildebrandt, A. Musidlowska, U. T. Bornscheuer and
length measured on substrates after DFT-geometry optimi-
J. Altenbuchner, Appl. Microbiol. Biotechnol., 2002, 59,
zation. Hence, the electronic properties of the meta- and para-
483.
substituent affected the rate of the reaction, showing that the 12 D. Zhu, Y. Yang and L. Hua, J. Org. Chem., 2006, 26, 4202.
outcome of the reaction can be predicted and analysed simi- 13 M. Vogl, R. Kratzer, B. Nidetzky and L. Breck, Org. Biomol.
larly to any other chemical reaction. Reduction of ortho-deriva-
tives occurred only with substrates bearing substituents with 14 H. G. Naik, B. Yeniad, C. E. Koning and A. Heise, Org.
low steric impact (i.e., CN vs. NO2). DFT-based geometry Biomol. Chem., 2012, 10, 4961.
optimization and LUMO frontier orbital calculations 15 Y. Yang, D. Zhu, T. J. Piegat and L. Hua, Tetrahedron: Asym-
accounted for the different reactivity of the CvO group in the metry, 2007, 18, 1799.
ortho-substituted acetophenone series. (S)-Enantioselectivity 16 M. L. Contente, I. Serra, M. Brambilla, I. Eberini,
Chem., 2011, 9, 5863.
with any substrate bearing a substituent on the aromatic ring
was observed; notably, only acetophenone gave (R)-1-phenyl-
ethanol with high enantioselectivity.
E. Gianazza, V. De Vitis, F. Molinari, P. Zambelli and
D. Romano, Appl. Microbiol. Biotechnol., 2016, 100, 193.
17 M. L. Contente, F. Molinari, I. Serra, A. Pinto and
D. Romano, Eur. J. Org. Chem., 2016, DOI: 10.1002/
ejoc.201501557.
18 P. Hoyos, G. Sansottera, M. Ferna, M. F. Molinari,
J. V. Sinisterra and A. R. Alcántara, Tetrahedron, 2008, 64,
7929.
19 R. Gandolfi, E. Cesarotti, F. Molinari and D. Romano, Tetra-
hedron, 2009, 20, 411.
Notes and references
1 R. Noyori, Angew. Chem., Int. Ed., 2002, 41, 2008.
2 J. C. Moore, D. J. Pollard, B. Kosjek and P. N. Devine, Acc.
Chem. Res., 2007, 40, 1412.
20 M. C. Fragnelli, P. Hoyos, D. Romano, R. Gandolfi,
A. R. Alcántara and F. Molinari, Tetrahedron, 2012, 68,
523.
21 C. Forzato, R. Gandolfi, F. Molinari, P. Nitti, G. Pitacco and
E. Valentin, Tetrahedron, 2001, 12, 1039.
22 M. L. Contente, F. Molinari, P. Zambelli, V. De Vitis,
R. Gandolfi, A. Pinto and D. Romano, Tetrahedron Lett.,
2014, 55, 7051.
23 Initial rates were calculated from initial progress curves;
100% corresponds to the activity of the highest reacting
substrate (1h).
3 B. H. Hoff and E. Sundby, Bioorg. Chem., 2013, 51, 31.
4 M. J. Sorgedrager, F. van Rantwijk, G. W. Huisman and
R. A. Sheldon, Adv. Synth. Catal., 2008, 350, 2322.
5 E. Garcia-Urdiales, E. Busto, N. Rios-Lombardia, V. Gotor-
Fernandez and V. Gotor, ChemCatChem, 2009, 10, 2875.
6 H. Groger, W. Hummel, C. Rollmann, F. Chamouleau,
H. Husken, H. Werner, C. Wunderlich, K. Abokitse,
K. Drauz and S. Buchholz, Tetrahedron, 2004, 60, 633.
7 I. Lavandera, A. Kern, B. Ferreira-Silva, A. Glieder, S. de
Wildeman and W. Kroutil, J. Org. Chem., 2008, 73, 6003.
8 Z. Findrik, D. Vasic-Racki, S. Lutz, T. Daussmann and
C. Wandrey, Biotechnol. Lett., 2005, 27, 1087.
24 Analytical yields were determined after 24 h by GC or HPLC
(see the ESI† for details).
25 R. Kratzer, D. K. Wilson and B. Nidetzky, IUBMB Life, 2006,
58, 499.
9 D. Zhu, B. E. Rios, J. D. Rozzell and L. Hua, Tetrahedron:
Asymmetry, 2005, 16, 1541.
10 W. Hummel, Appl. Microbiol. Biotechnol., 1990, 34, 15.
This journal is © The Royal Society of Chemistry 2016
Org. Biomol. Chem.