Stereoselective reduction of aromatic ketones by a new ketoreductase from Pichia glucozyma

Article abstract of DOI:10.1007/s00253-015-6961-y

A new NADPH-dependent benzil reductase (KRED1-Pglu) was identified from the genome of the non-conventional yeast Pichia glucozyma CBS 5766 and overexpressed in E. coli. The new protein was characterised and reaction parameters were optimised for the enantioselective reduction of benzil to (S)-benzoin. A thorough study of the substrate range of KRED1-Pglu was conducted; in contrast to most other known ketoreductases, KRED1-Pglu prefers space-demanding substrates, which are often converted with high stereoselectivity. A molecular modelling study was carried out for understanding the structural determinants involved in the stereorecognition experimentally observed and unpredictable on the basis of steric properties of the substrates. As a result, a new useful catalyst was identified, enabling the enantioselective preparation of different aromatic alcohols and hydroxyketones.

Full text of DOI:10.1007/s00253-015-6961-y

Appl Microbiol Biotechnol
DOI 10.1007/s00253-015-6961-y
BIOTECHNOLOGICALLY RELEVANT ENZYMES AND PROTEINS
Stereoselective reduction of aromatic ketones by a new
ketoreductase from Pichia glucozyma
Martina Letizia Contente1 & Immacolata Serra1 & Marta Brambilla1 & Ivano Eberini1
&
Elisabetta Gianazza2 & Valerio De Vitis1 & Francesco Molinari1 & Paolo Zambelli2
&
Diego Romano1
Received: 25 May 2015 /Revised: 20 July 2015 /Accepted: 23 August 2015
#
Springer-Verlag Berlin Heidelberg 2015
Abstract A new NADPH-dependent benzil reductase
(KRED1-Pglu) was identified from the genome of the
non-conventional yeast Pichia glucozyma CBS 5766 and
overexpressed in E. coli. The new protein was
characterised and reaction parameters were optimised
for the enantioselective reduction of benzil to (S)-ben-
zoin. A thorough study of the substrate range of
KRED1-Pglu was conducted; in contrast to most other
known ketoreductases, KRED1-Pglu prefers space-
demanding substrates, which are often converted with
high stereoselectivity. A molecular modelling study
was carried out for understanding the structural determi-
nants involved in the stereorecognition experimentally
observed and unpredictable on the basis of steric prop-
erties of the substrates. As a result, a new useful cata-
lyst was identified, enabling the enantioselective prepa-
r a t i o n o f d i f f e r e n t a r o m a t i c a l c o h o l s a n d
hydroxyketones.
Introduction
The use of ketoreductases (KREDs, often also described as
carbonyl reductases (CR) or alcohol dehydrogenases (ADH))
is a well-recognised method for the enantioselective reduction
of prochiral ketones into stereodefined chiral alcohols (Hall
and Bommarius 2011; Hollmann et al. 2011). Reduction of
aromatic ketones has been studied with different reductases,
with particular focus on the reduction of sterically hindered
(bulky) substrates, not easily recognised by most of the avail-
able KREDs (Lavandera et al. 2008). Commercially available
or engineered KREDs can reduce structurally different (in-
cluding bulky) ketones and their use has been also scaled-up
to industrial processes (Huisman et al. 2010; Liang et al.
2010). Other alcohol dehydrogenases, such as the ones from
Rhodococcus ruber (ADH-A; Stampfer et al. 2004), Ralstonia
sp. (RasADH; Lavandera et al. 2008), and Lactobacillus
brevis (LBADH; Haberland et al. 2002) have been identified
as suitable biocatalysts for the reduction of different aromatic
bulky substrates. Steric effects of the substrates on the activity
of LBADH towards aromatic ketones (acetophenone deriva-
tives) were accurately studied. In most cases, a tendency for
higher activity with smaller substrates could be detected, but
noticeable deviations from this general trend were observed
due to stereoelectronic effects (Rodríguez et al. 2014).
Understanding how substrate-binding residues of the protein
affect substrate binding orientation and thus sterereocognition
is necessary not only for predicting enantioselectivity, but also
for designing rational or semirational mutations, aimed at in-
creasing or changing the stereobias of a given reduction
(Baerga-Ortiz et al. 2006; O’Hare et al. 2006). This approach
was employed for mutating a carbonyl reductase from
Sporobolomyces salmonicolor AKU4429 (SSCR) and
switching its enantiopreference in the reduction of para-
substituted acetophenones (Zhu et al. 2008). This SSCR
.
.
Keywords Ketoreductase Carbonyl reductase Pichia
.
.
.
glucozyma Enantioselective reduction Biocatalysis
Stereoselective
* Diego Romano
diego.romano@unimi.it
1
Department of Food, Environmental and Nutritional Sciences
(DeFENS), University of Milan, Via Mangiagalli 25,
20133 Milan, Italy
2
Department of Pharmacological and Biomolecular Sciences
(DiSFeB), University of Milan, Via Balzaretti 9, 20133 Milano, Italy

Appl Microbiol Biotechnol
variant has been recently used for the enantioselective reduc-
tion of asymmetric benzophenone derivatives (Li et al. 2009).
Whole cells of the non-conventional yeast Pichia
glucozyma CBS 5766 (now reclassified as Ogataea
glucozyma) have been used for the reductive biotransforma-
tion of various aromatic ketones and ketoesters (including
bulky ones), often showing interesting stereoselectivity
(Forzato et al. 2001; Gandolfi et al. 2009; Rimoldi et al.
2011a; Husain et al. 2011; Rimoldi et al. 2011b; Contente
et al. 2015). The reduction of these substrates frequently oc-
curred with activity and enantioselectivity strongly dependent
on the conditions of growth and biotransformation, suggesting
the possible occurrence of different ketoreductases acting on
aromatic ketones. Notably, whole cells of P. glucozyma carried
out the selective monoreduction of different 1,2-diaryl-
ethanediones (benzils) (Hoyos et al. 2008; Fragnelli et al.
2012). Enzymes with benzil reductase activity, belonging to
the short-chain dehydrogenase/reductase (SDR) family, have
been found in bacteria (Maruyama et al. 2002; Pennacchio
et al. 2013) and yeasts (Johanson et al. 2005). Here, we report
the identification of a benzil reductase from P. glucozyma and
a study of its activity/enantioselectivity towards different aro-
matic ketones using the recombinant protein (KRED1-Pglu)
overexpressed in Escherichia coli.
Production of the recombinant protein
Identification of KRED1-pglu gene
Genome of P. glucozyma CBS 5766 was sequenced by
Baseclear (BaseClear B.V., Leiden, The Netherlands) using
Illumina Casava pipeline version 1.8.0; assembly has been
performed using the Bde novo assembly^ option of the CLC
Genomics Workbench version 5.1. Annotation has been made
on the assembled contig or scaffold sequences using the
GAPS Annotation Pipeline from Progenus (Gembloux,
Belgium). Upon analysis, 311 oxidoreductases, 517 transfer-
ases, 512 hydrolases, 71 lyases, 53 isomerases and 109 ligases
were annotated and, among the oxidoreductases, a unique
benzil reductase gene (named KRED1-Pglu) was found after
BLAST analysis.
Cloning
KRED1-Pglu gene, coding for a putative benzil reductase, was
amplified from the genomic DNA of P. glucozyma by PCR
using the following primers, carrying NdeI and HindIII restric-
tion sites: forward 5′-ATACCATATGACGAAGGTGA
CTGTTGTGAC-3′; reverse 5′-AGAGAAGCTTGG
CGTACTCCTTCAACTCTG-3′.
The amplified gene was then cloned into a pET26b(+)
vector using the NdeI and HindIII restriction sites. With
this cloning strategy, the resulting protein is expressed
with an N-terminal His6 tag. The correct construction of
the expression plasmid was confirmed by direct
sequencing.
Materials and methods
General
Phusion DNA polymerase was from New England
Biolabs (EuroClone S.p.A., Pero (MI), Italy).
pET26b(+) vector was from Novagen (Merck
Millipore, Vimodrone (MI), Italy).
Protein expression
HIS-Select® Nickel Affinity Gel used for protein purifica-
tion was purchased from Sigma Aldrich (Milano, Italy).
Pichia glucozyma CBS 5766 was from CBS-KNAW
Fungal Biodiversity Centre (Utrecht, The Netherlands).
Stocks of the strain were stored at −20 °C in 15 % v/v glycerol
and revitalized liquid YPD (yeast extract 5 g/L, peptone 5 g/L,
glucose 20 g/L, pH 6.0) medium prior use. The bacterial strain
E. coli XL1 blue was used for the propagation of the recom-
binant plasmid. E. coli BL21(DE3)Star (Merck Millipore,
Vimodrone, Italy) was used for expression of KRED1-Pglu.
Both were used according to the instruction of the suppliers.
All the DNA manipulations and bacterial transformations
were carried out according to standard protocols (Sambrook
and Russell 2001) or manufacturers’ instructions, unless oth-
erwise stated.
Cultures of E. coli BL21(DE3)Star transformed with the
resulting plasmid were grown overnight at 37 °C in LB me-
dium supplemented with 25 μg/mL kanamycin. The seed cul-
ture was then diluted into a stirred bioreactor (Biolafitte &
Moritz bioreactor, Pierre Guerin Technologies) contain-
ing 4 L of cultivation medium (Terrific Broth, 12 g/L
bacto-tryptone, 24 g/L yeast extract, 4 g/L glycerol,
2.3 g/L KH2PO4, 9.4 g/L K2HPO4, pH 7.2) to an initial
OD600nm of 0.05. Cultivation was carried out in batch-
mode at 37 °C, 250 rpm stirring and 250 L/h aeration
rate. Cells were grown until OD600nm reached the value
of 0.8. The cultures were induced for 20 h with
isopropyl-β-D-thiogalactopyranoside (IPTG) to a final
concentration of 0.5 mM. Cells were then harvested by
centrifugation at 4500 rpm for 30 min, washed once
with 20 mM phosphate buffer at pH 7.0 and stored at
−20 °C.
Protein concentration was determined with the Bradford
method on a Pharmacia Biotech Ultrospec 1000 using bovine
serum albumin as standard (Bradford 1976).

Appl Microbiol Biotechnol
Protein purification
NADPH, 50 mM Tris–HCl buffer pH 8.0, 25 °C), using
0.5 mM of each compound (in final 0.1 % DMSO); relative
activities were expressed as per cent values versus benzil.
Cells were suspended in 50 mM Tris–HCl, 100 mM NaCl,
6 mM imidazole, pH 8.0 buffer. Proteins were extracted by
sonication (5 cycles of 30 s each, in ice, with 1 min interval)
and cell debris were harvested by centrifugation at 15,000 rpm
for 30 min at 4 °C. The enzyme was purified by affinity chro-
matography with HIS-Select® Nickel Affinity Gel. Briefly, the
column was equilibrated with 50 mM Tris–HCl, 100 mM
NaCl, 6 mM imidazole, pH 8.0 and the crude extract loaded;
column was then washed with 50 mM Tris–HCl, 100 mM
NaCl, 6 mM imidazole, pH 8.0; finally, the adsorbed enzyme
was eluted with 50 mM Tris–HCl, 100 mM NaCl, 250 mM
imidazole, pH 8.0.
Pellets, crude extracts and collected fractions were
analysed by SDS-PAGE. The fractions showing the presence
of a band of the expected size (27 kDa) were pooled, dialysed
against 50 mM Tris–HCl buffer pH 8.0 and stored at −20 °C.
Typically, starting from 4 g of wet cell paste, it was possible to
obtain 70 mg of pure protein.
Determination of kinetic parameters
Kinetic parameters for the reduction of benzil were deter-
mined using the reduction assay (0.07 g/L of KRED1-Pglu,
0.25 mM NAD(P)H, 50 mM Tris–HCl buffer pH 8.0, 30 °C).
Data were fitted to suitable kinetic models with KaleidaGraph
software (Synergy Software Inc., USA); kinetic parameters
were calculated with the same program.
Preparative reductions
Molar conversion and enantioselectivity towards different ar-
omatic ketones and ketoesters were determined by performing
biotransformations at 50-mg scale, using an enzyme-coupled
system (glucose-glucose dehydrogenase (GDH) from Bacillus
megaterium; Bechtold et al. 2012) for cofactor recycling.
Reductions were carried out in 10-mL screw-capped test tubes
with a reaction volume of 5 mL with KRED1-Pglu (20 mU/
mL), GDH (1 U/mL), NADP+ (0.1 mM), substrate (1 g/L),
glucose (4× mmol of substrate) suspended in 50 mM Tris/HCl
buffer pH 8.0. GDH activity and stability in our reaction sys-
tem were checked before its use (data not shown).
Benzil-reductase activity test
Activity measurements were performed spectrophotometrical-
ly at 340 nm by determining the consumption of NAD(P)H at
25 °C in a half-microcuvette (total volume 1 mL) for 5 min.
One unit (U) of activity is defined as the amount of enzyme
which catalyses the consumption of 1 μmol of NAD(P)H per
minute under reference conditions, namely with 0.25 mM
NAD(P)H and 0.47 mM benzil as substrate (added as concen-
trated DMSO solution; final DMSO concentration in cuvette
amounts to 0.1 %), in 50 mM Tris–HCl buffer, pH 8.0.
Analyses 1H-NMR and 13C-NMR spectra were recorded with
a Varian Mercury 300 (300 MHz) spectrometer. Chemical
shifts (δ) are expressed in ppm, and coupling constants (J)
are expressed in Hz.
HPLC analyses were performed with a Jasco Pu-980
equipped with a UV–vis detector Jasco UV-975. The follow-
ing HPLC columns were used: LichroCART (250×4.6 mm),
Chiralcel OD (250×4.6 mm, Daicel), Chiralcel OD-H (250×
4.6 mm, Daicel), Chiralcel OJ (250×4.6 mm, Daicel), Lux
cellulose 2 (250×4.6 Phenomenex) Lux cellulose 3 (250×
4.6 Phenomenex). (R) and (S) isomers were identified by com-
parison of their retention times with those of synthetic
standards.
Temperature and pH activity and stability studies
Temperature and pH optima were studied conducting the
benzil-reductase test at temperatures between 15 and 50 °C
and at pH values between 4 and 12 (using suitable buffers). To
investigate its temperature-dependent stability, the enzyme
was incubated at different temperatures (15, 28, 37, 50 and
60 °C) in 50 mM Tris–HCl buffer pH 8.0; residual activity
was assessed conducting the benzil reductase test under refer-
ence conditions at different time points between 0 and 24 h of
incubation. pH-dependent stability was investigated preincu-
bating the enzyme in different buffers at pH between 5 and
11 at 25 °C. Residual activity was measured at pH 8.0, 25 °C,
at different time points between 0 and 24 h, and compared to
the value before incubation.
Analytical data of product of biotransformations are as
follows:
Benzoin (2a)
[α]D25=+37.1° (c 0.8 Acetone) lit.1 [α]D25=+44.2° (c 1.0
acetone)
1H-NMR (300 MHz, CDCl3): δ 7.88–7.93 (m, 2H), 7.48–
7.53 (m, 1H), 7.23–7.41 (m, 7H), 5.95 (d, J=6 Hz, 1H), 4.55
(d, J=6 Hz, 1H) ppm.
Substrate range
13C-NMR (75 MHz, CDCl3): δ 199.1, 139.2, 134.0, 133.7,
129.3, 129.2, 128.8, 128.7, 127.9, 76.4 ppm.
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD column (n-hexane/i-PrOH 90:10, 0.7 mL/
KRED1-Pglu substrate range was studied carrying out the
reduction assay under reference conditions (0.25 mM

Appl Microbiol Biotechnol
min, 254 nm): tr (S)-benzoin 14.7 min, tr (R)-benzoin
21.6 min).
13C-NMR (75 MHz, CDCl3): δ 145.0, 128.4, 127.5, 125.9,
74.4, 41.2, 19.0, 13.9 ppm.
Diphenylmethanol (2b)
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD column (n-hexane/i-PrOH 99:1, 0.5 mL/
min, 254 nm): tr (R)-1-phenylbutanol 19.4 min, tr (S)-1-
phenylpbutanol 23.2 min.
1H-NMR (300 MHz, CDCl3): δ 2.19 (d, J=4.0 Hz, 1H),
5.86 (d, J=4.0 Hz, 1H), 7.25–7.28, (m, 2H), 7.32–7.35 (m,
4H), 7.38–7.40 (m, 4H) ppm.
13C-NMR (75 MHz, CDCl3): δ 143.8 128.5, 127.6, 126.5,
76.2 ppm.
The molar conversion was determined by HPLC using
LichroCART 250–4,6 MeOH/H2O 65:35, 0.7 mL/min,
220 nm: tr diphenylmethanol 12.5 min, tr benzophenone
20.6 min.
1-Pyridin-3yl-ethanol (2i)
[α]D25=−38.04° (c 0.4 ethanol) lit.4 [α]D20=−39.0° (c 0.93
ethanol).
1H-NMR (300 MHz, CDCl3): δ 8.46–8.33 (m, 2H), 7.75–
7.70 (m, 1H), 7.24 (dd, J1=8.0 Hz, J2=5.0 Hz, 1H), 4.90 (q,
J=6.5 Hz, 1H), 4.5 (br s, 1H, OH), 1.49 (d, J=6.5 Hz, 3H)
ppm.
Ethyl mandelate (2d)
[α]D25=+88.7° (c 0.7 EtOH ) lit.2 [α]D22=+97.33° (c 1.1
EtOH).
13C-NMR (75 MHz, CDCl3): δ 148.0, 147.0, 141.8, 133.5,
123.5, 67.5 25.1 ppm.
1H-NMR (300 MHz, CDCl3): δ 7.35–7.42 (m, 5H), 5.15 (s,
1H), 4.15–4.25 (m, 2H), 1.20 (m, 3H) ppm.
13C-NMR (75 MHz, CDCl3): δ 173.7, 138.5, 128.7, 128.5,
126.6, 73.0, 62.3, 14.2 ppm.
The enantiomeric excess (ee %) was determined by HPLC
using Phenomenex Lux cellulose 2 column (n-hexane/i-PrOH
90:10, 0.7 mL/min, 254 nm): tr (R)-(1-pyridin-3yl)ethanol
20.46 min, tr (S)-(1-pyridin-3yl)ethanol 25.30 min.
1-Pyridin-3yl-propanol (2j)
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD-H column (n-hexane/i-PrOH 90:10,
1.0 mL/min, 210 nm): tr (S)-ethyl mandelate 6.0 min, tr (R)-
ethyl mandelate 10.0 min).
[α]D25=−40.0° (c 0.2 methanol) lit.6 [α]D20=−41.4° (c 2.1
methanol).
1H-NMR (300 MHz, CDCl3): δ 8.31 (m, 1H), 8.25 (m,
1 H), 7.63 (m, 1 H), 7.15 (m, 1 H), 4.90 (s, 1 H), 4.51 (t, J=
6.6 Hz, 1 H), 1.68 (m, 2 H), 0.82 (t, J=7.41 Hz, 3 H) ppm.
13C-NMR (75 MHz, CDCl3,): δ 147.9, 147.4, 140.6, 133.9,
123.3, 72.8, 31.9, 9.8 ppm.
1-Phenylethanol (2e)
[α]D25=+40.0° (c 0.5 chloroform) lit.3 [α]D25=+36.0° (c
1.0 chloroform).
1H-NMR (300 MHz, CDCl3): δ 7.20–7.35 (m, 5H), 4.85
(q, J=6.4 Hz, 1H), 1.46 (d, J=6.4 Hz, 3H) ppm. 13C-NMR
(75 MHz, CDCl3): δ 145.9, 128.5, 127.5, 125.4, 70.4,
25.2 ppm.
The enantiomeric excess (ee %) was determined by
HPLC using Phenomenex Lux cellulose 2 column (n-
hexane/i-PrOH 90:10, 0.7 mL/min, 254 nm): tr (R)-(1-
pyridin-3yl)propanol 18.2 min, tr (S)-(1-pyridin-
3yl)propanol 21.1 min.
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD column (n-hexane/i-PrOH 95:5, 0.7 mL/
min, 254 nm): tr (R)-1-phenylethanol 14.2 min, tr (S)-1-
phenylethanol 18.0 min.
(1-Thiophen-2-yl)ethanol (2k)
[α]D25=−20.80° (c 0.2 chloroform) lit.4 [α]D20=−23.2° (c
0.79 chloroform).
1-Phenylpropanol (2f)
[α]D25=+20.5° (c 0.2 chloroform) lit.4 [α]D20=+22.2° (c
1.0 chloroform).
1H-NMR (300 MHz, CDCl3): δ 7.25–7.18 (m, 1H), 6.95–
6.92 (m, 2H), 5.07 (q, J=6.5 Hz, 1H), 2.54 (br s, 1H, OH),
1.56 (d, J=6.5 Hz, 3H) ppm.
1H-NMR (300 MHz, CDCl3): δ 7.35–7.22 (m, 5H), 4.54 (t,
J=6.5 Hz, 3H), 2.20 (br s, 1H, OH), 1.83–1.68 (m, 2H), 0.89
(t, J=7.5 Hz, 3H) ppm.
13C-NMR (75 MHz, CDCl3): δ 150.0, 126.6, 124.4, 123.2,
66.2, 25.2 ppm.
13C-NMR (75 MHz, CDCl3): δ 144.6, 128.4, 127.5, 126.0,
76.0, 31.9, 10.1 ppm.
The enantiomeric excess (ee %) was determined by HPLC
using Phenomenex Lux cellulose 3 column (n-hexane/i-PrOH
90:10, 0.5 mL/min, 254 nm): tr (S)-(1-thiophen-2-yl)ethanol
13.5 min, tr (R)-(1-thiophen-2-yl)ethanol 15.90 min.
(1-Furan-2-yl)ethanol (2l)
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD-H column (n-hexane/i-PrOH 98:2,
1.0 mL/min, 254 nm): tr (R)-1-phenylpropanol 16.5 min, tr
(S)-1-phenylpropanol 19.5 min.
[α]D25=−20.5° (c 0.2 chloroform) lit.4 [α]D20=−13.6° (c
0.55 chloroform).
1-Phenylbutanol (2g)
[α]D25=−40.5° (c 0.2 chloroform) lit.7 [α]D25=−43.5° (c
1.0 chloroform).
1H-NMR (400 MHz, CDCl3): δ 7.36 (dd, J1=2.0 HzJ2=
1.0 Hz, 1H,), 6.32 (dd, J1=3.0 HzJ2=2.0 Hz, 1H,), 6.22 (d,
J=3.0 Hz, 1H,), 4.86 (q, J=6.5 Hz, 1H,), 2.35 (br s, 1H, OH),
1.52 (d, J=6.5 Hz, 3H,) ppm.
1H-NMR (300 MHz, CDCl3): δ 7.34–7.32 (m, 4H), 7.27–
7.24 (m, 1H), 4.66 (t, J=6.5 Hz, 1H), 1.93 (br s, 1H, OH),
1.80–1.63 (m, 2H), 1.46–1.27 (m, 2H), 0.92 (t, J=7.5 Hz, 3H)
ppm.
13C NMR (100 MHz, CDCl3): δ 157.7, 141.8, 110.1,
105.1, 63.6, 21.2 ppm.

Appl Microbiol Biotechnol
The enantiomeric excess (ee %) was determined by HPLC
using Phenomenex Lux cellulose 3 column (n-hexane/i-PrOH
98:2, 0.5 mL/min, 220 nm): tr (S)-(1-furan-2-yl)ethanol
19.3 min, tr (R)-(1-furan-2-yl)ethanol 21.3 min.
3-Hydroxy-3-phenylpropanenitrile (2m)
Activity and enantioselectivity
The activity of KRED1-Pglu was first investigated for reduc-
tion of benzil, observing that the enzyme is NADPH-depen-
dent. The enzyme exhibited the highest activity at pH between
7.0 and 8.0, being stable at pH between 6.0 and 8.0 (residual
activity of 90 % and 80 %, respectively). KRED1-Pglu
showed the highest activity between 30 °C and 35 °C and
was quite stable up to a temperature of 30 °C, above which
its activity started to decrease. As a compromise between ac-
tivity and stability, the following studies have been carried out
at 30 °C in 50 mM Tris/HCl buffer pH 8.0. Kinetic parameters
for monoreduction of benzil by KRED1-Pglu were deter-
mined using 0.07 mg/ml of protein, NADPH as cofactor, at
30 °C and pH 8.0 (Table 1).
The initial rates of the biotransformation of different aro-
matic ketones and ketoesters were determined using a
substrate-independent assay (Table 2, column 5); relative ac-
tivities were expressed as per cent values versus benzil. Molar
conversion and enantioselectivity were determined by
performing the reduction of the substrates on a 50 mg scale,
using an Benzyme-coupled^ system (glucose/glucose dehy-
drogenase, GDH) (Table 2).
[α]D25=−40.5° (c 0.2 ethanol) lit.8 [α]D20=−57.7° (c 2.6
ethanol).
1H-NMR (300 MHz, CDCl3) δ 7.30–7.45(m, 5H), 5.1 (t,
J=6.0,1H), 2.79 (d, J=6 Hz, 2H), 2.36 (br s, 1H, OH) ppm.
13C-NMR (75 MHz, CDCl3) δ 140.1, 128.9, 128.8, 125.5,
117.6, 70.2, 28.1.
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OJ column (n-hexane/i-PrOH 90:10, 1 mL/
min, 216 nm): tr (S)-3-hydroxy-3-phenylpropanenitrile
21.4 min, tr (R)-3-hydroxy-3-phenylpropanenitrile 25.0 min.
2-Hydroxy-1-phenylpropan-1-one (2v)
[α]D25=−100° (c 0.2 chloroform) lit.8 [α]D27=−91.1° (c
0.3 chloroform).
1H-NMR (300 MHz, CDCl3) δ 7.87 (d, J=8.2 Hz, 2H),
7.57 (t, J=7.42 Hz, 1H), 7.46 (t, J=7.7 Hz, 2H), 5.17 (m,
1H), 3.5 (br s, 1H, OH), 1.45 (d, J=7.0 Hz, 3H) ppm.
13C-NMR (75 MHz, CDCl3) δ 202.3, 133.9, 133.4, 128.7,
128.6, 69.3, 22.2.
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD column (n-hexane/i-PrOH 95:5, 0.4 mL/
min, 254 nm): tr (R)-2-hydroxy-1-phenylpropan-1-one
17.2 min, tr (S)-2-hydroxy-1-phenylpropan-1-one 21.9 min.
3-Hydroxy-1-phenyl-butan-1-one (2w)
Preparative reduction of benzil (1a) resulted in total conver-
sion into (S)-benzoin (ee >98 %) in 2 h; bulky benzophenone
(1b) (which is transformed at low rates) was reduced to
diphenylmethanol in 24 h with good yields (66 %), showing
that a sluggish initial activity does not necessarily imply low
conversion, as already observed for the reduction of aromatic
ketones using LBADH (Rodríguez et al. 2014). Ethyl 2-oxo-2-
phenylacetate (1d) was totally reduced within 30 min, albeit
with mediocre S-enantioselectivity (ee 40 %). No consistent
trend could be detected between bulkiness of the substrates
and initial rates; for instance, the series 1e-h showed an erratic
tendency of activities with respect to the increasing bulkiness of
the substrate, similar to what observed with LBADH (Rodríguez
et al. 2014). KRED1-Pglu gave pronounced anti-Prelog
stereorecognition in the reduction of acetophenone (1e) and
propiophenone (1f) to yield the corresponding (R)-1-aryl-2-
ethanol, whereas other aromatic ketones (1g and 1i-1m) provid-
ed the corresponding (S)-alcohol (Prelog rule) with high
enantioselectivity. This remarkable switch of enantioselectivity
between similar substrates is not common among ketoreductases
able to reduce aromatic ketones (Lavandera et al. 2008; Zhu
et al. 2006). Finally, KRED1-Pglu showed no activity towards
much hindered substrates like β-γ-ketoesters (1n-1u).
[α]D25=+54.2° (c 0.1 chloroform) lit.10 [α]D27=+67.5° (c
1.2 chloroform).
1H-NMR (300 MHz, CDCl3) δ 7.97 (dd, J1=8.5, J2=
2.0 Hz, 2H ), 7.50 (m, 3H), 4.42 (m, 1H), 3.24 (br s, 1H,
OH), 3.10 (d, J=6.0 Hz, 2H) 1.30 (d, J=6.4 Hz 3H) ppm.
13C-NMR (75 MHz, CDCl3) δ 200.95, 136.8, 133.6, 128.8,
128.1, 64.1, 46.5, 22.5 ppm.
The enantiomeric excess (ee %) was determined by HPLC
using Chiralcel OD-H column (n-hexane/i-PrOH 95:5,
0.8 mL/min, 254 nm): tr (R)-3-hydroxy-1-phenyl-butan-1-
one 12.1 min, tr (S)-3-hydroxy-1-phenyl-butan-1-one
13.6 min.
Accession numbers
GenBank accession number of KRED1-Pglu gene is
KR080472.
Results
Analysis of the genome of Pichia glucozyma CBS 5766 re-
vealed the occurrence of a sequence (called KRED1-Pglu)
with high homology with known benzil reductases. The cor-
responding His-tagged protein was successfully expressed in
E. coli, yielding an active enzyme accounting for about 40 %
of the total protein content of the cell extract.
Table 1 Kinetic parameters for monoreduction of benzil by KRED1-
Pglu
kcat (s−1
)
Km (mM)
kcat/Km (s−1 mM−1
)
0.14 0.02
0.64 0.10
0.22

Appl Microbiol Biotechnol
Table 2 Investigation of the substrate range of KRED-Pglu concerning the reduction of different aromatic ketones and ketoesters
Entry
Substrate
Ar
R
Activity (%)a
Yield (%)
ee (%)
Time (h)b
1
1a
1b
1c
1d
1e
1f
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
3-Pyridyl
3-Pyridyl
2-Thienyl
2-Furanyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
COPh
100
15
<5
115
21
25
18
<5
58
60
51
48
55
<5
<5
<5
<5
<5
<5
<5
<5
>95
66
>98 (S)
–
2
2
Phenyl
24
24
0.5
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
3
CH2Ph
<5
<5
4
COOEt
>95
40
40 (S)
95 (R)
97 (R)
97 (S)
<5
5
CH3
6
CH2CH3
60
7
1g
1h
1i
CH2CH2CH3
CH2CH2CH2CH3
CH3
56
8
<5
9
>95
>95
>95
>95
>95
<5
>98 (S)
>98 (S)
>98 (S)
>98 (S)
95 (S)
<5
10
11
12
13
14
15
16
17
18
19
20
21
1j
CH2CH3
1k
1l
CH3
CH3
1m
1n
1o
1p
1q
1r
1s
1t
CH2CN
CH2COOEt
CH2COOtBu
CH(CH3)COOEt
CH(CH3)COOtBu
CH2CH2COOEt
CH2CH2COOtBu
CH2CH(CH3)COOEt
CH2CH(CH3)COOtBu
<5
<5
<5
<5
<5
<5
<5
<5
<5
<5
<5
<5
1u
<5
<5
a Relative activities measured using the benzil-reductase activity test and expressed as percent values referred to benzil
b Referred to the maximum yield
1-Phenylpropane-1,2-dione (1v) and 1-phenylbutane-1,3-
dione (1w) were also tested; both diketones were exclusively
reduced at the less hindered carbonyl group with formation of
(S)-2-hydroxy-1-phenylpropan-1-one (2v) and (S)-3-hy-
droxy-1-phenylbutan-1-one (2w) as the only reaction prod-
ucts, with complete enantioselectivity (Fig. 1).
The observed activities and enantioselectivity cannot be
explained by simply assuming a model of the active site with
two hydrophobic pockets differing in size, where the large
pocket binds the phenyl moiety, while the small one defines
whether the other substituent can be properly accommodated
and thus controls reaction rates and enantioselectivity. Since
the experimental results were unpredictable on the basis of just
the steric properties of the substrates, a molecular modelling
study was carried out for understanding the structural deter-
minants for the observed stereorecognition.
As the starting point for homology modelling, the PDB
database (http://www.rcsb.org/pdb/) was scanned with
BLAST against the target sequence of the KRED1-Pglu while
setting the maximal E-threshold=100; the search retrieved
several items. The top-scoring identified template, an
uncharacterized oxidoreductase YIR035C from
Saccharomyces cerevisiae (UniProt entry P40579), and the
target share 46 % identity; YIR035C has been crystallized in
its apo form (X-ray structure as PDB entry 3KZV). The two
sequences were automatically aligned and manually adjusted,
and a model for the target protein was built by comparative
modelling. The structure so obtained did not allow, however,
for any effective docking of NAD(P)H to it. The second best
match, cis-2,3-dihydrobiphenyl-2,3-diol dehydrogenase from
Comamonas testosteroni (UniProt entry Q46381), shares
31 % identity with the target and has been crystallized in a
complex with NAD (PDB entry 2Y99). After alignment of the
sequences with T-coffee and superposition of the cognate
structures, NAD was imported from the template to the model
and the complex was energy-minimized, which resulted in
minor RMSD overall and a negligible displacement of the
amino acids predicted to be involved in the catalytic process
(annotations to UniProt entry P40579). However, once more,
O
O
KRED1-Pglu
yield > 95%
ee > 98%
buffer pH 8.0 / 30 °C
NADP+/glucose/GDH
O
OH
1v
O
2v
O
O
OH
KRED1-Pglu
yield = 60%
ee > 98%
buffer pH 8.0 / 30 °C
NADP+/glucose/GDH
1w
2w
Fig. 1 Reduction of 1-phenylpropane-1,2-dione (1v) and 1-
phenylbutane-1,3-dione (1w) with KRED1-Pglu

Appl Microbiol Biotechnol
no effective docking of the test substrates could be obtained to
the model•NAD complex.
placements changes depending on the length of the aliphatic
chain, which defines the stereochemistry of the reductive reac-
tion. Figure 2 shows the relative placement of acetophenone
(substrate 1e; colour = green) and butyrophenone (substrate
1h; colour = yellow), and their position with respect to the
reductive hydride (arrow) of NAD(P)H (CPK colours).
Lengthening the aliphatic chain produces steric clashes with
amino acids in the enzyme catalytic site (specifically, W147
and L208) and results in an inversion of the predominant ori-
entation during the interaction. As a result, the stereochemistry
predicted for the stereocenter of the alcohol products is different
in either case and identical to what is obtained in vitro.
Various PDB entries of holo enzymes, crystallized as to
contain both NAD and a substrate/substrate analog/inhibitor
(2JAP, 1NAS, 4NBU), were thus inspected from the previous
BLAST output; clavulanic acid dehydrogenase from
Streptomyces clavuligerus (UniProt entry Q9LCV7; PDB entry
2JAP; 30 % identity with the target) was selected as further
template because of the comparable size of its substrate to our
test substrates. After superposition of the structures, substrate
clavulanic acid was imported from 2JAP to our model•NAD.
Effective docking was possible to the ternary complex
model•NAD•clavulanic acid after energy minimization; we will
refer to the outcome of this procedure as KRED1-Pglu model.
Docking to KRED1-Pglu model was carried out using dif-
ferent protocols (rigid receptor, induced fit), various forcefields
(MMF94x, Amber12:EHT) and different electrostatics setups
(distance-dependent dielectric, reaction field); comparing the
position of the redox centre in the top scoring poses for the test
substrates in each procedure with that of the redox centre of
clavulanic acid taken as reference suggested that the combina-
tion: rigid receptor, Amber12:EHT, and distance-dependent
dielectric performs best. This protocol was then used to dock
to the KRED1-Pglu model all selected substrates.
The procedures of molecular modelling provided insight
into the ways of enzyme-substrate interaction and allowed
computing the affinities (dissociation constant values, Kds)
for different substrates, with an accuracy within one order of
magnitude, as discussed by Eberini et al. (2006). Table 3 lists
the in silico affinities of the enzyme for selected substrates
alongside the in vitro activities (same as in Table 2). Binding
of a substrate to the enzyme active site is necessary, but not
sufficient for catalysis to take place: once this prerequisite is
met, the rate of the process is controlled by the
stereoelectronic property of the substrate, a feature not directly
addressed by the classical docking procedure. Within these
boundaries, the relative order of the computational data is in
agreement with the experimental results and the enzyme is
most active on benzil (entry 1), the substrate with the highest
affinity for the catalytic site.
Discussion
Ketoreductases (KREDs) stereoselectively reduce prochiral ke-
tones giving access to optically pure secondary alcohols (Hall
and Bommarius 2011; Hollmann et al. 2011). Bulky aromatic
ketones can be reduced by specialised KREDs (Haberland et al.
2002; Stampfer et al. 2004; Lavandera et al. 2008; Huisman et al.
2009). Benzil reductases (Maruyama et al. 2002; Johanson et al.
2005; Pennacchio et al. 2013) are specific KREDs with high
affinity towards very hindered 1,2-diphenyl-ethanediones.
Sequence comparison among the benzil reductases from bacteria,
yeasts and mammals showed that conserved amino acid residues
are essential for benzil reductase activity and can be useful for
fishing out their genes from genomes. Whole cells of wild-type
Pichia glucozyma CBS 5766 have been used for a number of
reductions of aromatic ketones, including symmetric and asym-
metric 1,2-diaryl-ethanediones (benzils); the genome of
P. glucozyma CBS 5766 was annotated and a putative gene for
benzil reductase was identified based on sequence homology
with known benzil reductases and cloned in E. coli. The His-
tagged protein (named as KRED1-Pglu) was purified and used
for the reduction of different aromatic ketones; the enzyme
showed a marked preference for NADPH and was used on
When the series of aryl aliphatic ketones is docked into the
enzyme active site, the orientation of the top-scoring
Table 3 In silico affinities of KRED-Pglu for selected substrates
Entry Substrate Ar
R
Activity (%) Ki
1
2
5
6
7
8
1a
1b
1e
1f
Phenyl COPh
100
5
7.94×10−6
Phenyl Phenyl
Phenyl CH3
1.21×10−5
6.15×10−5
3.36×10−5
2.32×10−5
1.38×10−5
Fig. 2 Close-up to the catalytic site in the model of KRED1-Pglu.
NADH is rendered with CPK colors; the hydride atom involved in the
reduction process is marked with a dotted circle. Two substrates,
acetophenone (1e) and butyrophenone (1 h) are rendered in green and
yellow. Their orientation in the enzyme pocket is opposite, leading to
inverted configuration in the alcohol products
11
35
10
Phenyl CH2CH3
Phenyl CH2CH2CH3
1g
1h
Phenyl CH2CH2CH2CH3 <5

Appl Microbiol Biotechnol
preparative scale coupled with a glucose dehydrogenase (GDH)
for cofactor recycling. Benzil was reduced giving (S)-benzoin as
the sole observable product, whereas the use of whole cells
yielded lower enantiomeric excesses and further reduction to
hydrobenzoin (Fragnelli et al. 2012), indicating the occurrence
in P. glucozyma of other KR(s) active not only on benzil but also
on benzoin. Very hindered benzophenone was also reduced, as
previously observed with selected KRs (Truppo et al. 2007).
Aromatic ketoesters were poorly reduced by KRED1-Pglu,
whereas whole cells of P. glucozyma CBS 5766 were very effec-
tive and sometimes highly enantioselective on these substrates,
indicating the occurrence of other enantioselective KRs in this
unconventional yeast (Contente et al. 2014). Acetophenone and
propiophenone were reduced to the corresponding (R)-alcohols
(anti-Prelog’s rule), whereas butyrophenone was converted into
the corresponding alcohol with different stereobias, giving (S)-
phenylbutanol with high enantioselectivity. This shift of
stereopreference was explained by modelling the interaction be-
tween substrate and protein and observing that the length of the
aliphatic chain governs the stereochemistry of the biotransforma-
tion, determining which prostereogenic face of the ketone can be
favourably attacked by the hydride. The key role of the aliphatic
chain can also be seen in the case of valerophenone, where the
longer chain determines steric clashes with amino acids in the
enzyme catalytic site and the ketone is not reduced. KRED1-
Pglu was also employed for the reduction of 1-phenylpropane-
1,2-dione and 1-phenylbutane-1,3-dione leading to the
enantioselective formation of the α-methyl alcohols as the only
product, suggesting that KRED1-Pglu has potential as catalyst
for the preparation of hydroxyketones with high chemical and
optical purity.
References
Baerga-Ortiz A, Popovic B, Siskos AP, O’Hare HM, Spiteller D,
Williams MG, Campillo N, Spencer JB, Leadlay PF (2006)
Directed mutagenesis alters the stereochemistry of catalysis by iso-
lated ketoreductase domains from the erythromycin polyketide syn-
thase. Chem Biol 13:277–285. doi:10.1016/j.chembiol.2006.01.004
Bechtold M, Brenna E, Femmer C, Gatti FG, Panke S, Parmeggiani F,
Sacchetti A (2012) Biotechnological development of a practical syn-
thesis of ethyl(S)-2-ethoxy-3-(p-methoxyphenyl)propanoate (EEHP):
over 100-fold productivity increase from yeast whole cells to recom-
binant isolated enzymes. Org Process Res Dev 16(2):269–276
Bradford MM (1976) A rapid and sensitive method for the quantitation of
microgram quantities of protein utilizing the principle of protein-dye
binding. Anal Biochem 72:248–254
Contente ML, Molinari F, Zambelli P, De Vitis V, Gandolfi R, Pinto A,
Romano D (2014) Biotransformation of aromatic ketones and
ketoesters with the non-conventional yeast Pichia glucozyma.
Tetrahedron Lett 55(51):7051–7053
Contente ML, Zambelli P, Galafassi S, Tamborini L, Pinto A, Conti P,
Molinari F, Romano D (2015) A new chemoenzymatic approach to
the synthesis of Latanoprost and Bimatoprost. J Mol Catal B Enzym.
doi:10.1016/j.molcatb.2014.05.022
Eberini I, Fantucci P, Rocco AG, Gianazza E, Galluccio L, Maggioni D,
Ben ID, Galliano M, Mazzitello R, Gaiji N, Beringhelli T (2006)
Computational and experimental approaches for assessing the interac-
tions between the model calycin β-lactoglobulin and two antibacterial
fluoroquinolones. Proteins 65:555–567. doi:10.1002/prot.21109
Forzato C, Gandolfi R, Molinari F, Nitti P, Pitacco G, Valentin E (2001)
Microbial bioreductions of γ- and δ-ketoacids and their esters.
Tetrahedron 12:1039–1046. doi:10.1016/S0957-4166(01)00184-7
Fragnelli MC, Hoyos P, Romano D, Gandolfi R, Alcántara AR, Molinari
F (2012) Enantioselective reduction and deracemisation using the
non-conventional yeast Pichia glucozyma in water/organic solvent
biphasic systems: preparation of (S)-1,2-diaryl-2-hydroxyethanones
(benzoins). Tetrahedron 68:523–528. doi:10.1016/j.tet.2011.11.014
Gandolfi R, Cesarotti E, Molinari F, Romano D (2009) Asymmetric re-
ductions of ethyl 2-(benzamidomethyl)-3-oxobutanoate by yeasts.
Tetrahedron 20:411–414. doi:10.1016/j.tetasy.2009.02.023
Haberland J, Kriegesmann A, Wolfram E, Hummel W, Liese A (2002)
Diastereoselective synthesis of optically active (2R,5R)- hexanediol.
Appl Microbiol Biotechnol 58:595–599
Hall M, Bommarius AS (2011) Enantioenriched compounds via enzyme-
catalyzed redox reactions. Chem Rev 111:4088–4110. doi:10.1021/
cr200013n
Hollmann F, Arends IWCE, Holtmann D (2011) Enzymatic reductions
for the chemist. Green Chem 13:2285–2314
Hoyos P, Sansottera G, Fernández M, Molinari F, Sinisterra JV, Alcántara AR
(2008) Enantioselective monoreduction of different 1,2-diaryl-1,2-
diketones catalysed by lyophilised whole cells from Pichia glucozyma.
Tetrahedron 64:7929–7936. doi:10.1016/j.tet.2008.06.019
Huisman GW, Liang J, Krebber A (2010) Practical chiral alcohol manu-
facture using ketoreductases. Current Opinion in Chemical Biology
14:122–129. doi:10.1016/j.cbpa.2009.12.003
Husain SM, Stillger T, Dünkelmann P, Lodige M, Walter L, Breitling E,
Pohl M, Bürchner M, Krossing I, Müller M, Romano D, Molinari F
(2011) Stereoselective reduction of 2-hydroxy ketones towards syn-
and anti-1,2-diols. Adv Synth Catal 353:2359–2362. doi:10.1002/
adcs.201100150
Johanson T, Katz M, Gorwa-Grauslund MF (2005) Strain engineering for
stereoselective bioreduction of dicarbonyl compounds by yeast re-
ductases. FEMS Yeast Res 5:513–525. doi:10.1016/j.femsyr.2004.
12.006
In conclusion, a new (bio)catalyst for the preparation of
bulky aromatic chiral alcohols and hydroxyketones with high
enantioselectivity by reduction of the corresponding ketones
has been identified and produced from an unconventional
yeast. The different and erratic stereopreferences towards sim-
ple aromatic ketones were explained on the basis of modelling
studies. Further studies could be useful/help to identify,
generate and analyse KRED1-Pglu mutants, in which the
stereoselectivity of the reductive reaction can be carefully
tuned in order to customize and scale-up specific reactions
with industrial relevance.
Acknowledgments Recombinant GDH from Bacillus megaterium was
kindly provided by Prof. Daniela Monti (Istituto di Chimica del
Riconoscimento Molecolare, C.N.R., Via Mario Bianco 9, 20131 Milano,
Italy).
Compliance with ethical standards This article does not contain any
studies with human participants or animals performed by any of the
authors.
Conflict of interest All the authors declare they have no conflict of
interest.
Lavandera I, Kern A, Ferreira-Silva B, Glieder A, de Wildeman S, Kroutil
W (2008) Stereoselective bioreduction of bulky-bulky ketones by a

Appl Microbiol Biotechnol
novel ADH from Ralstonia sp. J Org Chem 73:6003–6005. doi:10.
1021/jo800849d
Li H, Zhu D, Hua L, Biehl ER (2009) Enantioselective reduction of diaryl
ketones catalyzed by a carbonyl reductase from Sporobolomyces
salmonicolor and its mutant enzymes. Adv Synth Catal 351:583–
588. doi:10.1002/adsc.200900045
catalytic asymmetric hydrogenation or by biotransformation.
Tetrahedron: Asymmetry 22:2110–2116. doi:10.1016/j.tetasy.2011.
03.007
Rodríguez C, Borzęcka W, Sattler JH, Kroutil W, Lavandera I, Gotor V
(2014) Steric vs. electronic effects in the Lactobacillus brevis ADH-
catalyzed bioreduction of ketones. Org Biomol Chem 12:673–681.
doi:10.1039/C3OB42057D
Sambrook J, Russell DW (2001) Molecular Cloning: A
Laboratory Manual, 3rd edn. Cold Spring Harbor
Laboratory, New York
Stampfer W, Edegger K, Kosjek B, Faber K, Kroutil W (2004) Simple
biocatalytic access to enantiopure (S)-1-heteroarylethanols
employing a microbial hydrogen transfer reaction. Adv Synth
Catal 346:57–62
Truppo MD, Pollard D, Devine P (2007) Enzyme-Catalyzed
enantioselective diaryl ketone reductions. Org Lett 9:335–338. doi:
10.1021/ol0627909
Zhu D, Yang Y, Hua L (2006) Stereoselective enzymatic synthesis of
chiral alcohols with the use of a carbonyl reductase from Candida
magnoliae with anti-prelog enantioselectivity. J Org Chem 71:4202–
4205
Zhu D, Yang Y, Majkowicz S, Pan T (2008) Inverting the
enantioselectivity of a carbonyl reductase via substrate−enzyme
docking-guided point mutation. Org Lett 10:525–528. doi:10.
1021/ol702638j
Liang J, Lalonde J, Borup B, Mitchell V, Mundorff E, Trinh N, Kochrekar
DA, Cherat RN, Pai GG (2010) Development of a biocatalytic pro-
cess as an alternative to the (−)-DIP-Cl-mediated asymmetric reduc-
tion of a key intermediate of Montelukast. Org Process Res Dev 14:
193–198. doi:10.1021/op900272d
Maruyama R, Nishizawa M, Itoi Y, Ito S, Inoue M (2002) The enzymes
with benzil reductase activity conserved from bacteria to mammals.
J Biotechnol 94:157–169. doi:10.1016/S0168-1656(01)00426-6
O’Hare HM, Baerga-Ortiz A, Popovic B, Spencer JB, Leadlay PF (2006)
High-throughput mutagenesis to evaluate models of stereochemical
control in ketoreductase domains from the Erythromycin polyketide
synthase. Chem Biol 13:287–296. doi:10.1016/j.chembiol.2006.01.003
Rimoldi I, Cesarotti E, Zerla D, Molinari F, Albanese D, Castellano C,
Gandolfi R (2011a) Chemo- and biocatalytic strategies to obtain
phenylisoserine, a lateral chain of Taxol by asymmetric reduction.
Tetrahedron: Asymmetry 22:597–602. doi:10.1016/j.tetasy.2011.
11.017
Rimoldi I, Pellizzoni M, Facchetti G, Molinari F, Zerla D, Gandolfi R
(2011b) 3-(Hydroxy(phenyl)methyl)azetidin-2-ones obtained via