A toolbox approach for the rapid evaluation of multi-step enzymatic syntheses comprising a 'mix and match' E. coli expression system with microscale experimentation

Article abstract of DOI:10.3109/10242422.2011.609589

This work describes an experimental 'toolbox' for the rapid evaluation and optimisation of multi-step enzymatic syntheses comprising a 'mix and match' E. coli-based expression system and automated microwell scale experimentation. The approach is illustrated with a de novo designed pathway for the synthesis of optically pure amino alcohols using the enzymes transketolase (TK) and transaminase (TAm) to catalyze asymmetric carbon-carbon bond formation and selective chiral amine group addition respectively. The E. coli expression system, based on two compatible plasmids, enables pairs of enzymes from previously engineered and cloned TK and TAm libraries to be evaluated for the sequential conversion of different initial substrates. This is complemented by the microwell experimentation which enables efficient investigation of different biocatalyst forms, use of different amine donors and substrate feeding strategies. Using this experimental 'toolbox', one-pot syntheses of the diastereoisomers (2S,3S)-2-aminopentane-1,3-diol (APD) and (2S,3R)-2-amino-1,3, 4-butanetriol (ABT) were designed and performed, which gave final product yields of 90% mol/mol for APD and 87% mol/mol for ABT (relative to the initial TK substrates) within 25 hours. For the synthesis of APD, the E coli TK mutant D469E was paired with the TAm from Chromobacterium violaceum 2025 while for ABT synthesis the wild-type E. coli TK exhibited the highest specific activity and ee( enantiomeric excess) of >95%. For both reactions, whole-cell forms of the TK-TAm biocatalyst performed better than cell lysates while isopropylamine (IPA) was a preferable amine donor than methylbenzylamine (MBA) since side reactions with the initial TK substrates were avoided. The available libraries of TK and TAm enzymes and scalable nature of the microwell data suggest this 'toolbox' provides an efficient approach to early stage bioconversion process design in the chemical and pharmaceutical sectors.

Full text of DOI:10.3109/10242422.2011.609589

Biocatalysis and Biotransformation, September–October 2011; 29(5): 192–203
ORIGINAL ARTICLE
A toolbox approach for the rapid evaluation of multi-step enzymatic
syntheses comprising a ‘mix and match’ E. coli expression system
with microscale experimentation
L. RIOS-SOLIS1, M. HALIM1, A. CÁZARES2, P. MORRIS1, J. M. WARD3, H. C. HAILES2,
P. A. DALBY1, F. BAGANZ1 & G. J. LYE1
1Department of Biochemical Engineering, University College London,Torrington Place, London,WC1E 7JE, UK,
2Department of Chemistry, University College London, 20 Gordon Street, London,WC1H 0AJ, UK,
3Institute of Structural and Molecular Biology, University College London, Gower Street, London,WC1E 6BT, UK
Abstract
This work describes an experimental ‘toolbox’ for the rapid evaluation and optimisation of multi-step enzymatic syntheses
comprising a ‘mix and match’ E.coli-based expression system and automated microwell scale experimentation.The approach
is illustrated with a de novo designed pathway for the synthesis of optically pure amino alcohols using the enzymes tran-
sketolase (TK) and transaminase (TAm) to catalyze asymmetric carbon-carbon bond formation and selective chiral amine
group addition respectively.The E. coli expression system, based on two compatible plasmids, enables pairs of enzymes from
previously engineered and cloned TK and TAm libraries to be evaluated for the sequential conversion of different initial
substrates. This is complemented by the microwell experimentation which enables efficient investigation of different bio-
catalyst forms, use of different amine donors and substrate feeding strategies. Using this experimental ‘toolbox’, one-pot
syntheses of the diastereoisomers (2S,3S)-2-aminopentane-1,3-diol (APD) and (2S,3R)-2-amino-1,3,4-butanetriol (ABT)
were designed and performed, which gave final product yields of 90% mol/mol for APD and 87% mol/mol for ABT
(relative to the initial TK substrates) within 25 hours. For the synthesis of APD, the E coli TK mutant D469E was paired
with the TAm from Chromobacterium violaceum 2025 while for ABT synthesis the wild-type E. coli TK exhibited the high-
est specific activity and ee( enantiomeric excess) of Ͼ95%. For both reactions, whole-cell forms of theTK-TAm biocatalyst
performed better than cell lysates while isopropylamine (IPA) was a preferable amine donor than methylbenzylamine
(MBA) since side reactions with the initial TK substrates were avoided. The available libraries of TK and TAm enzymes
and scalable nature of the microwell data suggest this ‘toolbox’ provides an efficient approach to early stage bioconversion
process design in the chemical and pharmaceutical sectors.
Keywords: transketolase, transaminase, biocatalyst, multi-step enzymatic synthesis, chiral amino alcohols, de novo pathway
Introduction
producing chiral compounds difficult to obtain by
existing biosynthetic pathways or chemical synthesis
(Lye et al. 2002, Burkart 2003, Roessner and Scott
2003, Prather and Martin 2008 Dalby et al. 2009).
These studies have led to the rapid expansion of
molecular pathway construction for the synthesis of
a range of commodity fine chemicals and pharma-
ceutical intermediates (Wilkinson and Bachmann
2006).
Biological catalysts have repeatedly demonstrated
their usefulness for chemical organic synthesis due
to their moderate reaction conditions, reduced
environmental impact and high regio- and stereo
selectivity (Whitesides andWong 2003).This has led
to the successful investigation of multi-step, meta-
bolically engineered pathways toward application as
pharmaceutical production platforms (Wilkinson
and Bachmann 2006). It is now possible to design
de novo non-native pathways in heterologous
hosts, to carry out specific non-natural bioconversions
The de novo design of synthetic pathways has
recently been applied for the sequential two-step
reaction of carbon–carbon bond formation using a
Correspondence: Professor Gary Lye, University College London, Biochemical Engineering,Torrington Place, London,WC1E 7JE United Kingdom. E-mail:
g.lye@ucl.ac.uk
(Received 14 December 2010; revised 9 May 2011; accepted 28 July 2011)
ISSN 1024-2422 print/ISSN 1029-2446 online © 2011 Informa UK, Ltd.
DOI: 10.3109/10242422.2011.609589

Rapid evaluation of multi-step enzymatic syntheses 193
O
-
HO
O
O
NH2
R'
O
CO2
HPA
R
R
R'
OH
OH
O
TK
TAm
OH
OH
OH
HO
HO
H
O
GA
NH2
(2S,3R)-ABT
L-ERY
Scheme 1. Reaction scheme of the de-novoTK-TAm pathway for the synthesis of chiral amino-alcohol (2S,3R)-2-amino-1,3,4-butanetriol
(ABT), from achiral substrates glycolaldehyde (GA) and hydroxypyruvate (HPA).
transketolase (TK) enzyme, followed by a transami-
nase (TAm) to create chiral amino alcohols from
achiral substrates (Ingram et al. 2007). The amino
alcohol moiety is a useful industrial synthon, neces-
sary in the synthesis of several protease inhibitors
(Kwon and Ko 2002), or broad spectrum antibiotics
like chloramphenicol and thiamphenicol (Bhaskar
et al. 2004, Boruwa et al. 2005). It also serves as
chiral starting material for the synthesis of optically
active molecules such as (S)-amphetamine (Rozwa-
dowska 1993). The standard chemical synthesis
route to produce optically pure amino alchohols is
usually complex, requiring many steps in order to
obtain the final product, resulting in low overall pro-
ductivities (Hailes et al. 2009, Smithies et al.
2009).
Using erythrulose (ERY) and glycolaldehyde
(GA) as achiral substrates, the one-pot synthesis of
the single syn-diastereoisomer (2S,3R)-2-amino-1,
3,4-butanetriol (ABT) was previously demonstrated
using E. coli whole cells expressing both wild type
TK and a β-alanine-pyruvate TAm as shown in
Scheme 1 (Ingram et al. 2007). ABT is an important
building block which can be found in the core of the
protease inhibitor drug NelfinavirTM (Kaldor et al.
1997).The main limitation of this two step synthesis
was the low activity of the β-alanine-pyruvate TAm
causing a low overall yield of the reaction. Conse-
quently we have constructed libraries of engineered
TKs and novel TAms, which could be used to pro-
duce a range of the desired amino alcohols with
higher yields and stereoselectivities (Hibbert et al.
2007, Kaulmann et al. 2007, Hibbert et al. 2008,
Smith et al. 2008). Subsequent work has shown that
the synthesis of (2S,3S)-2-aminopentane-1,3-diol
(APD) (Scheme 2) could be achieved in two sepa-
rate TK and TAm bioconversions, where the
enzymes selected were the E. coliTK mutant D469T
and the Chromobacterium violaceum CV2025 TAm,
reaching more than 90% mol/mol yield in each of
the reaction steps (Smith et al. 2010). Compounds
like APD containing the chiral 2-amino-1,3-diol
moiety are generally an important class of pharma-
ceutical building blocks with their motif present in
antibiotics, antiviral glycosidase inhibitors, and
sphingo-lipids (Hailes et al. 2010). In this case, the
required isolation of the TK product (3S)-1,3-dihy-
droxypentan-2-one (PKD) to perform theTAm bio-
conversion, led to a loss of more than 30% mol/mol
of the intermediate.This could be potentially avoided
by using a one-pot synthesis strategy, where the
activities of the enzymes could be matched leading
to higher yields of final product (Chen et al.
2006).
The aim of this work is to demonstrate a toolbox
approach for the rapid evaluation of different vari-
ants of TK and TAm for the one-pot synthesis of a
range of amino alcohols.The toolbox is based on an
E. coli host where it is possible to systematically eval-
uate different pairs of the enzymes, screening them
against different ketol donors, acceptors and amino
donors. All the experiments were performed in
O
O–
HO
NH2
R'
O
O
CO2
R
R
R'
HPA
OH
OH
O
TK
TAm
HO
HO
H
O
NH2
(2S,3S )-APD
PA
(S)-PKD
Scheme 2. Reaction scheme of the de-novo TK-TAm pathway for the synthesis of chiral amino-alcohol (2S,3S)-2-aminopentane-1,3-diol
(APD) from achiral substrates propionaldehyde (PA) and hydroxypyruvate (HPA).

194
L. Rios-Solis et al.
parallel at the microwell scale and could be readily
automated and potentially scaled up following the
procedure described in our previous works for fer-
mentation and bioconversion (Micheletti and Lye
2006, Islam et al. 2008). The one-pot syntheses of
ABT (Scheme 1) and (APD) (Scheme 2) were suc-
cessfully demonstrated at 87 and 90 % mol/mol
yields respectively, using the most appropriate forms
of TK-TAm expressing biocatalysts.
E. coli tktA was constitutive because of the presence
of its native promoter (Ingram et al. 2007). Site
directed mutagenesis of theTK gene was performed
as described previously (Hibbert et al. 2007).
Transaminase Plasmid. Plasmid pQR801 contained
the complete Chromobacterium violaceum 2025 TAm
gene with an N-Terminal His6-tag (GenBank acces-
sion no. NP_901695). Plasmid pQR801 was con-
structedusingtheexpressionvectorpET29(a)ϩ(5.3kb),
which contains an inducible T7 promoter, the Lac
repressor and codes for resistance to kanamycin
(Kaulmann et al. 2007).
Materials and methods
Materials
Molecular biology enzymes were obtained from New
England Bio-laboratories (NEB, Hitchin, UK).
Nutrient broth and nutrient agar were obtained from
Fisher Scientific (Leicestershire, UK). Competent
E. coli BL21-Gold (DE3) cells were obtained from
Stratagene (Amsterdam, NL). All other reagents
were obtained from Sigma-Aldrich (Gillingham,
UK) unless noted otherwise, and were of the highest
purity available.
Biocatalysts
TK whole cell and lysate preparation. Competent E.
coli BL21-Gold (DE3) cells were transformed with
the plasmid pQR412 using the heat shock technique
described by the supplier (Stratagene, Amsterdam,
NL). An overnight culture of the transformed cells
was obtained in a 100 ml shake flask (10 ml working
volume) of LB-glycerol broth (10 g lϪ1 tryptone,
5 g lϪ1 yeast extract, 10 g lϪ1 NaCl and 10 g lϪ1
glycerol) containing 150 μg mlϪ1 ampicillin. Growth
was performed at 37 oC with orbital shaking at 250
rpm using an SI 50 orbital shaker (Stuart Scientific,
Redhill, UK). The total volume of this culture was
used to inoculate a 1 litre shake flask (100 ml work-
ing volume) which was left to grow for 8 h.The cells
were harvested and stored at –20oC following the
removal of broth by centrifugation.When aTK lysate
was needed, the cells were resuspended in 50 mM
TRIS buffer, at pH 7.5 and sonicated with a Soni-
prep 150 sonicator (MSE, Sanyo, Japan).The lysate
suspension was centrifuged at 5000 rpm in Falcon
tubes for 5 min and then stored at –20oC and used
within 1 month. Prior to the initiation of a biocon-
version, the frozen lysates (300 μl) were thawed in a
water bath at 30oC for typically 10 min and then
immediately incubated with cofactors.
Synthesis of substrates and products
Hydroxypyruvate (HPA) was synthesized by reacting
bromopyruvic acid with LiOH following a previously
described method (Morris et al. 1996). PKD was syn-
thesized in a 100 ml scale bioconversion with 300 mM
HPA, 300 mM PA, 9 mM MgCl2, 2.4 mM thiamine
pyrophosphate (TPP), at pH 7.0 and 30% v/v of
D469E TK lysate (final TK concentration of 0.3 mg
mlϪ1). The reaction was stirred for 10 h at room
temperature in a sealed flask and the pH was main-
tained at 7.0 using a 718 STAT Titrino pH control-
ler (Metrohm Ion Analysis, Switzerland). The
solution was dried on silica and purified by column
chromatography (ethyl acetate: hexane, 1:1) to yield
PKD as a colorless oil that crystallized on standing.
ABT and APD products were prepared in a multi-
step chemical synthesis described elsewhere (Ingram
et al. 2007, Smith et al. 2010).
TAm whole cell and TK-TAm biocatalyst preparation.
Transformation of E. coli BL21-Gold (DE3) cells
with the plasmid pQR801 and inoculum preparation
were performed in the same way as described in Sec-
tion 2.4.1, except that 150 μg mlϪ1 of kanamycin
were used for the single transformed cells and 50 μg
mlϪ1 of both kanamycin and ampicillin were used for
the double transformed strain.To construct the dou-
ble transformed biocatalyst, 50 μl of competent E.
coli BL21-Gold (DE3) cells were double transformed
with a 1 μl solution of plasmids pQR412 and pQR801
(50 μg mlϪ1 for each plasmid) using the heat shock
technique described by the supplier (Stratagene,
Amsterdam, NL). For whole cell bioconversions,
Plasmids
Transketolase Plasmid and Mutagenesis. Plasmid
pQR412 contained the complete E. coli TK gene,
tktA, with its native promoter and an N-terminal
His6-tag. It was constructed using the expression
vector pMMB67HE (8.8 kb), which has a RSF1010
origin of replication, the tac promoter, the Lac
repressor and codes for resistance to ampicillin
(Fürste et al. 1986). Even if the plasmid contained
the inducible tac promoter, the expression of the

Rapid evaluation of multi-step enzymatic syntheses 195
fresh cells were always used and they were produced
as follows: after inoculation of a 1 litre shake flask
(100 ml working volume), when the OD600 reached
a value of 1.5-2, isopropylthiogalactopyranoside
(IPTG) was added to a final concentration of 0.2 mM.
After 4 hr induction, the cells were harvested and
following the removal of broth by centrifugation,
they were resuspended in 50 mM HEPES buffer, at
pH 7.5 and used for whole cell bioconversions.
When lysates were needed, after the resuspension,
the cells were sonicated and stored using the same
procedure described in the previous Section.
time intervals and quenched with 380 μl of a 0.1%
v/v trifluoroacetic acid (TFA) solution. They were
then centrifuged for 5 min at 5000 rpm and trans-
ferred into an HPLC vial for further analysis. All
experiments were performed in triplicate. The spe-
cific activities were determined as the amount of
PKD, ERY, acetophenone (AP), APD and ABT
formed per unit of time normalized by the amount
of enzyme used in the reaction. The specific activity
was based on the measured mass of TAm or TK
present in each bioconversion. For whole cell exper-
iments, it was calculated based on 50% of the dry
cell weight of the cells being protein (Watson 1972),
combined with the measured result of the percentage
of TK and TAm by SDS-PAGE analysis.
Bioconversion kinetics
Microscale experimental platform. All bioconversions
were performed in a glass 96-well, flat-bottomed
microtiter plate with individual wells having a diam-
eter of 7.6 mm and height of 12 mm (Radleys Dis-
coveryTechnologies, Essex, UK).The microplate was
covered with a thermo plastic elastomer cap designed
to work with automated equipment (Micronic, Lelys-
tad, Netherlands). All the bioconversions were per-
formed using 300 μl total volume at 30oC, at pH 7.5
unless noted otherwise, and shaking was provided at
400 rpm with aThermomixer Comfort shaker (shak-
ing diameter of 3 mm, Eppendorf, Cambridge, UK).
TK single step reactions were performed in 50 mM
TRIS buffer and the concentration of cofactors
MgCl2 and thiamine pyrophosphate (TTP) were
9 mM and 2.4 mM respectively for all reactions.
TAm single step reactions and all the two step syn-
theses were carried out in 50 mM HEPES buffer,
and the concentration of TAm cofactor pyridoxal-
5-phosphate (PLP) was 0.2 mM in all cases. (S)-
(α)-Methyl benzylamine (MBA) or isopropylamine
(IPA) were added as amino donors as indicated in
the results section at final concentrations of 10 mM
and 100 mM respectively unless noted otherwise. A
final concentration of 0.3 mg mlϪ1 of TK or TAm
was used in all the single enzyme reactions. For
whole cell bioconversion, 1.5 and 2.3 gDCW lϪ1
(where DCW is dry cell weight)were used for the
single TAm assays and for the one-pot syntheses
respectively.The whole cell suspension or lysate with
the cofactor solutions were always added first in the
well and left to incubate for 20 min at 30oC, prior
to initiation of the reaction by addition of the sub-
strate solutions. Preliminary experiments showed
that the initial cofactor incubation helped to ensure
that all expressedTK orTAm was in the active holo-
form. This practice also enabled more consistent
measurement of specific activity results, by avoiding
initial non linear reaction kinetics believed to be
caused in some cases by the binding of the enzyme
and cofactors. Aliquots of 20 μl were taken at various
Analytical methods
Biomass concentration was measured as optical
density at 600 nm (OD600) using a spectrophotom-
eter (Thermo Spectronic, Cambridge, UK) and con-
verted to dry cell weight (DCW) using a calibration
curve where 1 OD600 ϭ 0.5 gDCW lϪ1. Protein con-
centrations of the lysates were obtained using the
Bradford assay and SDS-PAGE as described previ-
ously (Kaulmann et al. 2007). A Dionex HPLC sys-
tem (Camberley, UK) with a Bio-Rad Aminex
HPX-87H reverse phase column (300 ϫ 7.8 mm,
Bio-Rad Labs., Richmond, CA, USA), controlled by
Chromeleon client 6.60 software was used for the
separation and analysis of PKD, ERY and HPA.The
system comprised a GP50 gradient pump, a FAMOS
autosampler, an LC30 chromatography column
oven and an AD20 UV/Vis absorbance detector.The
HPLC method used has been described previously
(Chen et al. 2008).To quantify MBA, AP, APD and
ABT, an integrated Dionex ultimate 3000 HPLC
system (Camberley, UK) with an ACE 5 C18 reverse
phase column (150mm ϫ 4.6 mm, 5 μm particle
size; Advance Chromatography Technologies,
Aberdeen, UK) controlled by Chromeleon client
6.60 software was employed. The HPLC method
has been reported elsewhere (Kaulmann et al. 2007).
To analyse ABT and APD, the samples were deriva-
tized by addition of an excess of 6-aminoquinolyl-N-
hydroxysuccinimidyl carbamate. The derivatizing
reagent was made in house following the protocol of
Cohen and Michaud (1992), and the HPLC method
used has been described previously (Ingram et al.
2007). Previously characterized by MS and NMR,
the enantiomeric excess (ee) of the ketodiols was
determined by derivatization via dibenzoylation for
satisfactory peak resolution by chiral HPLC.This was
performed on a Varian Prostar instrument equipped
with a Chiracel AD chiral column (Daicel; Chiral
Technologies Europe, France, 25 cm ϫ 0.46 cm) and

196
L. Rios-Solis et al.
a mobile phase of n-hexane/2-propanol solvent mix-
ture (90:10 v/v) was used. The method has been
described in more detail elsewhere (Cázares et al.
2010). The ee of ABT and APD was determined by
first derivatizing the amino alcohol to the respective
benzoate form.The assay was performed against the
four diastereomer samples of the benzoate synthe-
sized as described elsewhere (Smith et al. 2010)
using chiral HPLC: Chiracel-OD column (Daicel);
mobile phase, isopropanol/hexane (5:95); flow rate,
0.8 mL/min, detection, UV 210 nm. Figure 1 shows
examples of HPLC profiles for the chiral separation
and ee determination of (3S)-1,3-dihydroxypentan-2
-one (S-PKD) and (3R)-1,3-dihydroxypentan-2-one
(R-PKD), as well as the HPLC profiles for the detec-
tion and quantification (non chiral) of APD, HPA,
ERY, PKD, AP and MBA.
Figure 2. Shake flask fermentation kinetics for E. coli BL21-Gold
(DE3) grown in LB-glycerol medium at 37oC with (᭢)
untransformed cells, and cells transformed with (᭝) pQR412
Results and discussion
(TK), ( ) pQR801 (CV2025 TAm), (᭺) double transformed
•
uninduced strain with pQR801 and pQR412, (ᮀ) double
transformed strain induced with 0.2 mM IPTG after 6 h of cell
growth. Error bars represent one standard deviation about the
mean (n ϭ 3).
Biocatalyst Production
In order to separately regulate expression of the TK
and TAm enzymes in E. coli, a dual plasmid system
was selected to facilitate the “mix and match”
approach first proposed by Hussain andWard (2003).
This is based on two plasmids with different origins
of replication and antibiotic resistance genes that
would enable the rapid evaluation of different vari-
ants ofTK andTAm.To establish the metabolic bur-
den of the two plasmids on the host cells, five
different whole cell biocatalysts were initially gener-
ated and evaluated in shake flask fermentations.
Growth kinetics were obtained with cultures of E.
coli BL21-Gold (DE3) containing either no plasmid,
the single TK or TAm gene expressing plasmids, or
both the TK and TAm plasmids, in each case either
induced or non induced (Figure 2).
inducing at a biomass concentration of 1 gDCW lϪ1
for 4 hs, the final enzyme expression of TK and
TAm represented 8% and 36% w/w of the total pro-
tein. Because of the significantly lower activity of the
TAm compared to theTK (Chen et al. 2006, Ingram
et al. 2007), this method of induced expression of
TK and TAm was considered acceptable and was
used to prepare all the biocatalysts for the later two-
step bioconversions. Related work in our group is
focusing on fine control of the expression levels of
the two enzymes, so that the optimum proportions
can be obtained for specific bioconversions.
Single transformed strains with plasmid pQR412
(TK) or pQR801 (TAm) reached biomass concen-
trations of 2.1 and 1.8 gDCW lϪ1 respectively. These
were 12.5% and 25% less than for the non trans-
formed strain respectively. The double transformed
and induced strain obtained a final biomass concen-
tration of 1.2 gDCW lϪ1, representing nearly a 50%
decrease compared to the non transformed strain.
This decrease in growth, however, was compensated
by the increase in enzyme synthesis which was
expected due to the metabolic burden caused by the
maintenance and expression of two plasmids.
In terms of enzyme expression levels, TK and
TAm (induced with IPTG) in the single transformed
cells were found to represent 20% and 40% w/w of
the total protein respectively. For the double trans-
formed strain, the TK concentration diminished
compared to the single TK plasmid strain, and after
Considerations for two-step TK-TAm syntheses
The use of multi-step enzymatic syntheses necessar-
ily requires some compromises to be made regarding
the optimum reaction conditions for each enzyme.
This is particularly acute with isolated enzymes but
can be alleviated, to some extent, by the use of whole
cell biocatalysts. Optimum conditions reported for
TK-catalyzed ketodiol syntheses were pH 7.0 using
TRIS buffer at 25oC (Chen et al. 2006, Hibbert
et al. 2007, Cázares et al. 2010).The best conditions
for the CV2025 TAm to catalyze syntheses of vari-
ous amino alcohols were at pH 7.5 using HEPES
buffer at 37oC (Kaulmann et al. 2007, Smithies
et al. 2009). In order to evaluate the process com-
promises to be made, the synthesis of 10 mM ABT
using MBA as amino donor was used as a reference

Rapid evaluation of multi-step enzymatic syntheses 197
Figure 1. HPLC profiles from different reaction mixtures obtained as described in Section 2.5 for (a) the chiral separation and ee
determination of (3S)-1,3-dihydroxypentan-2-one (S-PKD) and (3R)-1,3-dihydroxypentan-2-one (R-PKD), and for the detection and
quantification (non chiral) of (b) APD , (c) HPA and ERY, (d) HPA and PKD, (e) AP and (f) MBA. HPLC conditions and details about
the samples preparation are described in Section 2.6.
obtained for the effect of TPP and Mg2ϩ on the
TAm bioconversion. Reactions ofTK carried at a pH
of 7.5 showed a decrease ofTK activity of 30% (data
not shown), nevertheless this pH was selected for
reaction (Scheme 1). Control TK reactions in the
presence of the working concentration of the TAm
cofactor PLP demonstrated that this did not have
any effect on theTK reaction.The same finding was

198
L. Rios-Solis et al.
subsequent studies considering the generally higher
activity of TK compared to TAm as described
above.
In terms of reaction temperature, a control reac-
tion using a lysate of untransformed E. coli BL21-
Gold (DE3) showed that at 37oC up to 50% of the
HPA substrate was consumed by side reactions after
2 h, while at 30oC, less than 5% of HPA was con-
sumed over the same time (data not shown). Con-
sequently the TK-TAm reactions were performed at
30oC to avoid HPA metabolism.With regards to buf-
fer selection, control experiments using TRIS and
phosphate buffer inhibited the CV2025 TAm. We
have previously shown that HEPES buffer catalyses
several side reactions between HPA and aldehydes
at 37oC (Smith et al. 2006b), nevertheless at 30oC,
control experiments showed that less than 5% of
HPA was lost by the biomimetic activity of 50 mM
HEPES after 2 h. Based on all these considerations,
the reaction conditions selected for the in vitro eval-
uation of the differentTK-TAm variants were 30 oC
and pH 7.5 in 50 mM HEPES buffer.
Figure 3. Bioconversion kinetics showing ketodiol synthesis using
different TK lysates and aldehyde acceptors: ( ) ERY with wild
•
type TK, (᭢) ERY with D469T TK, (᭿) ERY with D469E TK,
(᭝) PKD with D469TTK, (ᮀ) PKD with D469ETK, (᭺) PKD
with wild type TK. Reaction conditions: 10 mM HPA and GA,
2.4 mM TTP, 9 mM Mg2ϩ, 0.3 mg mlϪ1 TK, pH 7.5 in 50 mM
TRIS, 30oC. Error bars represent one standard deviation about
the mean (n ϭ 3).
Bioconversions with TK
Both the TK and TAm bioconversions were initially
studied individually under conditions suitable for
later two-step reactions established in the previous
section. Information on specific activities, ee and
final yields gathered from those experiments will
then enable us to successfully “mix and match” the
best pair of enzymes to synthesize specific amino
alcohols. Initial TK kinetics experiments showed
that the activities and final yields obtained with E.
coliTK lysates and whole cells were similar (data not
shown), therefore data reported here forTK biocon-
versions were performed only with lysates containing
the different TK variants (Table 1).
From TK mutant libraries developed previously
(Hibbert et al. 2008, Smith et al. 2008), theTK vari-
ants D469E, D469T and the wild type TK were
selected as candidates due to their catalytic potential
for the synthesis of ERY and/or PKD. Using HPA,
the TK reactions were irreversible due to the release
of CO2 as a side product, and yields of more than
90% mol/mol were obtained using concentrations up
to 300 mM of HPA and the aldehyde GA or PA (data
not shown). In contrast, TAm reaction high yields
were limited to around 10 mM using MBA as amino
donor, hence initial TK bioconversions studies were
performed at this lower concentration.
presented the highest activity and ee of 6.6 μmol
minϪ1 mgϪ1 and Ͼ95% respectively. Gyamerah and
Willetts (1997) obtained an activity of 13.2 μmol
minϪ1 mgϪ1 for the same HPA and GA reaction
using wild type E. coli TK.This is in agreement with
the work described here considering the differences
in concentration of substrates used. For the synthe-
sis of PKD, the highest activity was 1.98 μmol minϪ1
mgϪ1 using theTK mutant D469T, while no activity
after 70 min of reaction was detected using the wild
type enzyme. The ee of PKD obtained with TK
D469E was Ͼ90%, compared to a 64% ee of D469T.
Due to the importance of the ee in the synthesis of
optically pure amino alcohols, theTK mutant D469E
Table I. Measured TK specific activities, product yields and ee
for the lysate catalyzed bioconversions shown in Figure 3.
TK
variant
Specific activitya,c
Final yieldb
Aldehyde (μmol minϪ1 mgϪ1) ee (%) (% mol/mol)
Wild type
Wild type
D469T
D469T
D469E
GA
PA
GA
PA
GA
PA
100
Ͻ1
95
98
100
90
6.62 Ϯ 0.13
Ͻ0.1
5.13 Ϯ 0.43
1.98 Ϯ 0.27
4.96 Ϯ 0.39
1.49 Ϯ 0.26
Ͼ95
ndd
85
64
nd
D469E
Ͼ90
aThe specific activity is based on the measured mass of TK in
each bioconversion.
Figure 3 shows the kinetics of ketodiol produc-
tion for different combinations ofTKs and aldehyde
acceptors. The quantified activities, yields and mea-
sured ee values are summarised in Table I. The wild
type TK was selected for ERY synthesis because it
bYield was determined at t ϭ 70 min.
cThe error of the specific activity represents one standard deviation
about the mean (n ϭ 3).
dNd ϭ not determined.

Rapid evaluation of multi-step enzymatic syntheses 199
was selected, even though it had a 25% lower
activity than D469T. The D469E mutation is in
the pyrimidine binding domain of TK, and is
assumed to enable activity towards PA because it
replaces the original interaction of Asp with the
C-2 hydroxyl group of GA, by creating a specific
interaction between the Glu and the methyl group
of the new substrate (Hibbert et al. 2008, Cázares
et al. 2010).
Bioconversions with TAm
Single step TAm bioconversions focused on the
CV2025TAm as it had already been shown to be the
best candidate for amino alcohol synthesis among
several other transaminases cloned by us in previous
work (Kaulmann et al. 2007). The CV2025 TAm
showed a 4 fold increase in yield and a 40 fold
improvement in specific activity (Figure IV and
Table II) compared to initial work with a β-alanine-
pyruvate transaminase (Ingram et al. 2007). The
comparison of in vitro and in vivo catalytic activity
of CV2025 TAm has not previously been reported,
so it is investigated here where a whole cell TAm
biocatalyst is compared with a lysate form for syn-
theses of a number of ketodiols.
Figure 4. Bioconversion kinetics showing APD synthesis using
CV2015 TAm in either lysate or whole cell form with different
amino donors: ( ) whole cell with MBA, (᭡) whole cell with IPA,
•
(᭺) lysate with MBA, (᭝) lysate with IPA. Reaction conditions :
10 mM MBA or 100 mM IPA, 10 mM PKD, 0.2 mM PLP, 0.3
mg mlϪ1 TAm, pH 7.5 in 50 mM HEPES, 30oC. Error bars
represent one standard deviation about the mean (n ϭ 3).
possible solution as has been shown by others
(Truppo et al. 2010).
Figure 4 shows product formation kinetics using
either MBA or IPA as amino donors and PKD as
amino acceptor. Due to the volatility of IPA, an
excess concentration of 100 mM IPA was initially
used instead of 10 mM MBA to compensate for any
loss due to evaporation. The results of activities and
final yields of both amino alcohols are summarised
in Table II. The final yield of ABT of 92 and 86%
mol/mol using whole cell and lysates respectively
improved our previously reported values of 14-22%
mol/mol (Kaulman et al. 2007). The quantitative
conversion of the reaction observed in Figure 4 and
Table II is in agreement with the equilibrium con-
stants obtained for a similar transamination of pyru-
vate with MBA using the ω-transaminase from
Bacillus thuringiensis JS64, showing that the forward
reaction was thermodynamically more favorable
than the reverse reaction (Shin and Kim 1998).
Smith et al. (2010) obtained a 93% mol/mol yield
for the synthesis of APD with the CV2025 TAm
using an excess of PKD (molar ratio of 1.5 PKD/
IPA). In this work, the use of whole cells enabled
us to reach an almost complete conversion of the
substrates with a final yield of 96% mol/mol without
having to use an excess of the chiral ketodiol.
Nevertheless we observed that working with higher
concentrations severely inhibited the bioconversion;
in this case the use of in situ product removal by
ion exchange resins or coupled reactions could be a
Similar yields were obtained using ERY or PKD
as substrates, but the specific activities using ERY
were between 17% and 42% higher than those of
PKD. For synthesis of both amino alcohol products,
the whole cell biocatalysts performed better than the
lysates demonstrating the beneficial effect of using
TAm in vivo for these bioconversions.This phenom-
enon has also been reported in the literature, where
higher yields using whole cell TAm biocatalysts have
Table II. Measured specific activities and product yields for the
TAm catalyzed conversion of PKD and ERY.
Biocatalyst Amino Amino Specific activitya,c Final yieldb
form
donor acceptor (μmol minϪ1 mgϪ1) (% mol/mol)
Whole cell MBA
Lysate MBA
Whole cell IPA
Lysate IPA
Whole cell MBA
Lysate MBA
Whole cell IPA
Lysate IPA
PKD
PKD
PKD
PKD
ERY
ERY
ERY
ERY
95
83
96
75
92
86
91
71
0.074 Ϯ 0.006
0.034 Ϯ 0.003
0.029 Ϯ 0.0035
0.015 Ϯ 0.003
0.096 Ϯ 0.005
0.045 Ϯ 0.003
0.035 Ϯ 0.05
0.026 Ϯ 0.04
aThe specific activity was based on the measured mass of TAm
present in each bioconversion. For whole cell experiments, it was
calculated based on 50% of the dry cell weight of the cells being
protein (Watson 1972), and SDS-PAGE analysis that showed that
40% of the protein was TAm.
bYield was determined at t ϭ 30 h.
cThe error of the specific activity represents one standard deviation
about the mean (n ϭ 3).

200
L. Rios-Solis et al.
been attributed to better stability of the enzyme
inside the cell (Shin and Kim 1997), to metabolism
consumption of inhibitory compounds (Shin and
Kim 1999), or to the effect of the cell membrane in
partitioning inhibitory components between the
inside and outside of the cell (Yun et al. 2004). In
this work, the inhibitory product acetophenone (AP)
was not further metabolised by the whole cells and
MBA and AP where found to be more inhibitory
than IPA and acetone (Figure 5). In contrast, when
using TAm lysates, lower yields were obtained using
IPA (75% and 71% mol/mol for APD and ABT
respectively) compared to MBA (83% and 86% mol/
mol respectively). Addition of more enzyme and
PLP after the initial reaction had stopped, partially
restored the biocatalytic activity, suggesting that the
improved yield using whole cells could be attributed
to an increase in stability of the enzyme. Figure 5
shows the kinetics ofTAm production of APD using
initial equimolar concentration of substrates of
50 mM PKD with IPA or MBA as amino donors.
After 40 h, the final yields were 52 and 23% using
IPA and MBA respectively, underlying the strong
inhibition caused by MBA. This inhibitory effect of
MBA has also been reported in the literature with
transaminases from Klebsiella pneumonia and Vibrio
fluvialis (Shin and Kim 1997,Yun et al. 2004).
Based on the results in Table II, the whole cell
TAm biocatalyst was selected for further study
because of its superior performance compared to the
lysate. IPA appears preferable as amine donor
because of its advantages in terms of economy,
downstream processing and reduced inhibition com-
pared to MBA. However, both amine donors were
evaluated for one-pot synthesis reactions in the next
section.
One-Pot synthesis of amino alchohols
In order to achieve the one-pot synthesis of amino
alcohols, two whole cell biocatalysts were con-
structed both with plasmid pQR801 expressing the
CV2025 TAm, and with either pQR412 expressing
the TK mutant D469E or the wild type variant for
APD and ABT syntheses respectively.The first one-
pot synthesis of APD using MBA as amino donor
gave a poor yield (15% mol/mol) of amino alcohol
after 20 h but another product was detected by
HPLC analysis. This compound was identified as
serine, which was possibly produced by the transam-
ination of HPA. To overcome this problem, MBA
was subsequently added in fed batch mode after the
TK reaction was complete (Figure 6). The volume
of the solution of MBA added was equal to the vol-
ume of the reaction mixture, hence the original con-
centration of the reactants was diluted 2-fold after
MBA addition. The first step comprising the TK
reaction reached completion after 4 h with a specific
activity of 2.68 μmol minϪ1 mgϪ1. The second step
TAm reaction,reaction had a specific activity of
0.067 μmol minϪ1 mgϪ1 and reached a maximum
APD yield of 90% mol/mol after 21 h of reaction
Figure 6. Typical bioconversion kinetics for the one-pot, whole
cell TK-TAm catalytic synthesis of APD with addition of amine
Figure 5. Bioconversion kinetics showing APD synthesis using
CV2015 TAm in whole cell form with different amine donors:
donor (MBA) after 4 h : (᭢) PKD, ( ) HPA, (Δ) APD, (᭛) MBA
•
and (᭺) AP. Reaction conditions (TK step): 20 mM HPA and
PA, 0.4 mM PLP, 2.4 mM TTP, 9 mM Mg2ϩ, 0.095 mg mlϪ1
TK, 0.43 mg mlϪ1 TAm, pH 7.5 in 50 mM HEPES, 30oC.
TAm step was initiated by adding MBA solution which resulted
in a 2 fold dilution of reactants.
(
) 50 mM PKD and MBA and (Δ) 50 mM PKD and IPA.
•
Reaction conditions: 0.2 mM PLP, 0.3 mg mlϪ1 TAm, pH 7.5 in
50 mM HEPES, 30oC. Error bars represent one standard
deviation about the mean (n ϭ 3).

Rapid evaluation of multi-step enzymatic syntheses 201
with MBA. This measured activity is in agreement
with the one found for the single TAm reaction
(Table II).
IPA was also seen to be a preferable amine donor
compared to MBA, and the whole cell biocatalyst
performed better than the lysate.The 87% mol/mol
yield for ABT synthesis represented an improve-
ment of 4-fold in yield in a quarter of the reaction
time compared to our previous work (Ingram et al.
2007). The 90% mol/mol yield for APD repre-
sented a 2-fold improvement yield in a third of the
reaction time compared to our previous two-step
synthesis (Smith et al. 2010).
When using IPA as amine donor at the compro-
mise temperature of 30oC, the TAm did not show
any noticeable activity towards HPA, GA or PA
over 5 h (data not shown). This was not the case at
37oC, where activity towards TK substrates was
detected in that period of time.This suggested that
it was possible to perform the bioconversion of
APD using IPA in a true one-pot synthesis at 30oC
as shown in Figure 7, were the TK reaction was
completed after 5 h, leading to a 90% mol/mol yield
of APD after 25 h.The specific activities of TK and
TAm in the one-pot synthesis were 1.46 and 0.022
μmol minϪ1 mgϪ1 respectively, which were also in
agreement with the corresponding TK and TAm
specific activities for the single bioconversions
(Table I and II). Similar results were obtained for
the one-pot synthesis of ABT using IPA as amine
donor, reaching a final yield of 87% mol/mol after
25 h (Figure 8). The TK reaction for the ABT syn-
thesis was completed in 1.5 h instead of the 5 h
required for the APD synthesis. This was in agree-
ment with the higher activity of the wild type TK
enzyme towards ERY synthesis in comparison to
the activity of the D469E variant towards PKD syn-
thesis (Table I).
Conclusions
The microscale ‘toolbox’ described in this work has
been shown to enable the efficient evaluation and
“mix and match” of pairs of TK and TAm enzymes
for the one-pot synthesis of optically pure amino
alcohols. In particular the CV2025 TAm was found
to display better catalytic activity when used in vivo
leading to the selection of whole cell biocatalysts for
the synthesis of the single diastereoisomers of both
ABT and APD. IPA was identified as a preferable
amine donor compared to MBA for the one-pot syn-
theses because side reactions with the initialTK sub-
strates (catalysed byTAm) were negligible and higher
yields could be obtained at higher concentrations.
Our current work is seeking to increase the size of
the available TK and TAm libraries and to expand
the available microscale methods for evaluation
of subsequent product recovery and purification
operations.
Taking all these results together demonstrated
the utility of the ‘mix and match’ microscale exper-
imental system for rapidly evaluating pairs of
enzymes and the interaction of reaction conditions.
Figure 7. Typical bioconversion kinetics for the one-pot, whole
cell TK-TAm catalytic synthesis of APD using IPA as amine
Figure 8. Typical bioconversion kinetics for the one-pot, whole
cell TK-TAm catalytic synthesis of ABT using IPA as amine
donor: (᭢) PKD, ( ) HPA, and (᭝) APD. Reaction conditions:
donor: (᭢) ABT, ( ) HPA, and (᭺) ERY. Reaction conditions:
•
•
10 mM HPA and PA, 100 mM IPA, 0.2 mM PLP, 2.4 mM TTP,
9 mM Mg2ϩ, 0.095 mg mlϪ1 TK , 0.42 mg mlϪ1 TAm, pH 7.5
in 50 mM HEPES, 30oC.
10 mM HPA and GA, 100 mM IPA, 0.2 mM PLP, 2.4 mMTTP,
9 mM Mg2ϩ, 0.095 mg mlϪ1 TK, 0.42 mg mlϪ1 TAm, pH 7.5 in
50 mM HEPES, 30oC.

202
L. Rios-Solis et al.
Gyamerah M, Willetts AJ. 1997. Kinetics of overexpressed
transketolase from Escherichia coli JM 107/pQR 700. Enzyme
Microbial Technol 20:127–134.
Hailes HC, Dalby PA, Lye GJ, Ward, JM. 2009. Biocatalytic
approaches to ketodiols and aminodiols. Chim Oggi 27:
28–31.
Hibbert EG, Senussi T, Costelloe SJ, Lei W, Smith MEB,
Ward JM, et al. 2007. Directed evolution of transketolase
activity on non-phosphorylated substrates. J Biotechnol 131:
425–432.
Hibbert EG, Senussi T, Smith MEB, Costelloe SJ, Ward JM,
Hailes HC, Dalby PA. 2008. Directed evolution of transketo-
lase substrate specificity towards an aliphatic aldehyde. J Bio-
technol 134:240–245.
Acknowledgements
The Mexican National Council for Science and
Technology (CONACYT),The Mexican Ministry of
Public Education (SEP), The Ministry of Higher
Education of Malaysia (MOHE) andThe RoyalThai
Government are acknowledged for the support to
the authors of this work. The UK Engineering and
Physical Sciences Research Council (EPSRC) is
thanked for the support of the multidisciplinary Bio-
catalysis Integrated with Chemistry and Engineering
(BiCE) programme (GR/S62505/01) at University
College London (London, UK). The help from the
thirteen industrial partners supporting the BiCE
programme is also acknowledged.
Hussain HA, Ward JM. (2003) Enhanced Heterologous Expres-
sion of Two Streptomyces griseolus Cytochrome P450s and
Streptomyces coelicolor Ferredoxin Reductase as Potentially
Efficient Hydroxylation Catalysts. Appl Environ Microbiol 69:
373–382.
Ingram CU, Bommer M, Smith MEB, Dalby PA, Ward J. M.,
Hailes HC, Lye GJ. 2007. One-pot synthesis of amino-alcohols
using a de-novo transketolase and β-alanine:pyruvate transami-
nase pathway in Escherichiacoli. Biotechnol Bioeng 96:
559–569.
Islam RS, Tisi D, Levy MS, Lye GJ. (2008) Scale-up of E. coli
growth and recombinant protein expression conditions from
microwell to laboratory and pilot scale based on matched kLa.
Biothecnol Bioeng 99:1128–1139.
Kaldor WK, Kalish VJ, Davies JF, Shetty BV, Fritz JE, Appelt K,
et al. 1997. Viracept (Nefinavir Mesylate, AG1343): A potent,
orally available inhibitor of HIV-1 protease. J Med Chem 40:
3979–3985.
Kaulmann U, Smithies K, Smith MEB, Hailes HC, Ward JM.
2007. Substrate spectrum of ω-transaminase from Chromobac-
terium violaceum DSM30191 and its potential for biocatalysis.
Enz Microb Tech 41:628–637.
Declaration of interest: The authors report no
conflicts of interest. The authors alone are respon-
sible for the content and writing of the paper. The
studentships was supported by the Mexican National
Council for Science and Technology (CONACYT),
the Mexican Ministry of Public Education (SEP),
the Ministry of Higher Education of Malaysia
(MOHE) and he Royal Thai Government.More-
over, financial support was provided bythirteen
industrial partners supporting the BiCE pro-
gramme.
References
Kwon SJ, Ko SY. 2002. Synthesis of statine employing a general
syn-amino alcohol building block. Tetra Lett 43:639–641.
Lye GJ, Dalby PA, Woodley J M. 2002. Better biocatalytic proc-
esses faster: new tools for the implementation of biocatalysis in
organic synthesis. Org Proc Res Dev 6:434–440.
Bhaskar G, Satish KV, Venkateswara RB, 2004. A short stereose-
lective synthesis of (–)chloramphenicol and (ϩ)-thiamphenicol.
Tetra Asym 15:1279–1283.
Boruwa J, Borah JC, Gogoi S, Barua NC. 2005. A short asym-
metric total synthesis of chloramphenicol using a selectively
protected 1,2-diol. Tetra Lett 46:1743–1746.
Micheletti M, Lye GJ. 2006. Microscale bioprocess optimization.
Curr Opin Biotechnol 17:611–618.
Morris KG, Smith MEB,Turner NJ, Lilly MD, Mitra RK,Wood-
ley JM. 1996. Transketolase from Escherichia coli: A practical
procedure for using the biocatalyst for asymmetric carbon-
carbon bond synthesis. Tetra Asym 7:2185–2188.
Prather KL, Martin CH. 2008. De novo biosynthetic pathways:
rational design of microbial chemical factories Curr Opin Bio-
technol 19:468–474.
Roessner CA, Scott AI .2003. Achieving natural product synthe-
sis and diversity via catalytic networking ex vivo. Chem Biol 3:
325–330.
Rozwadowska MD. 1993. An efficient synthesis of (S)-(ϩ)-
amphetamine. Tetra Asymm 4:1619–162.
Shin JS, Kim BG. 1997. Kinetic resolution of α-methylben-
zylamine with w-transaminase screened from soil microorgan-
isms: application of a biphasic system to overcome product
inhibition. Biotechnol Bioeng 55:348–358.
Shin JS, Kim BG. 1998. Kinetic modeling of ω-transamination
for enzymatic kinetic resolution of α-methylbenzylamine. Bio-
technol Bioeng 60:534–40.
Shin JS, Kim BG. 1999. Asymmetric synthesis of chiral amines
with ω-transaminase. Biotechnol Bioeng 65:206–211.
Smith MEB, Chen BH, Hibbert EG, Kaulmann, U, Smithies K,
Galman JL, et al. .2010. A multidisciplinary approach toward
the rapid and preparative-scale biocatalytic synthesis of chiral
Burkart MD. 2003. Metabolic engineering-a genetic toolbox
for small molecule organic synthesis. Org Biomol Chem; 1:
1–4.
Cázares A, Galman JL, Crago LG, Smith MEB, Strafford J, Rıos-
Solıs L, et al.2010. Non-a-hydroxylated aldehydes with evolved
transketolase enzymes. Org Biomol Chem 8:1301–1309.
Chen BH, Hibbert EG, Dalby PA. 2008. A new approach to
bioconversion reaction kinetic parameter identification. AIChE
J 58:2155–2163.
Chen BH, Sayar A, Kaulmann U, Dalby PA, Ward JM, Woodley
JM. 2006. Reaction modelling and simulation to assess the inte-
grated use of transketolase and ω-transaminase for the synthe-
sis of an aminotriol. Biocat Biotrans 24:449–457.
Cohen SA, Michaud DP.1993. Synthesis of a fluorescent derivatiz-
ing reagent, 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate,
and its application for the analysis of hydrolysate amino acids via
high-performance liquid chromatography. Anal Biochem 211:
279–287.
Dalby PA, Baganz F, Lye GJ, Ward JM. 2009. Protein and
pathway engineering in biocatalysis. Chim Ogg 27:18–22.
Fürste JP, PansegrauW, Frank R, Blöcker H, Scholz P, Bagdasar-
ian M, Lanka E. 1986. Molecular cloning of the plasmid RP4
primase region in a multi-host-range tacP expression vector.
Gene 48:119–131.

Rapid evaluation of multi-step enzymatic syntheses 203
amino alcohols: a concise transketolase-/ω-transaminase-medi-
ated synthesis of (2S,3S)-2-aminopentane-1,3-diol. Org Proc
Res Dev 14:99–107.
Smith MEB, Hibbert EG, Jones AB, Dalby PA, Hailes HC. 2008.
Enhancing and reversing the stereoselectivity of Escherichia coli
transketolase via single-point mutations. Adv Syn Cat 350:
2631–2638.
mediated amination of 1,3-dihydroxy-1-phenylpropane-2-one.
Tetrahedron: Asymm 20:570–574.
Truppo MD, Rozzell JD,Turner NJ. 2010. Efficient production of
enantiomerically pure chiral amines at concentrations of 50g/L
using transaminases. Org Process Res Dev 14: 234–237.
Watson JD. 1972 .Molecular Biology of the Gene. 2nd ed., Phil-
adelphia, PA: Saunders.
Smith MEB, Kaulmann U,Ward JM, Hailes HC. 2006a. A color-
imetric assay for screening transketolase activity. Bioorg Med
Chem 14:7062 –7065.
Smith MEB, Smithies K, Senussi T, Dalby PA, Hailes HC.
2006b. The first mimetic of the transketolase reaction. Eur J
Org Chem 5:1121–1123.
Yun H, Cho BK, Kim BG. 2004. Kinetic resolution of (R,S)-
secbutylamine using omega-transaminase from Vibrio fluvialis
JS17 under reduced pressure. Biotechnol Bioeng 87: 772–777.
Whitesides GN, Wong CH. 2003. Enzymes as catalysts in syn-
thetic organic chemistry. Angew Chem Int Ed 24:617–638.
Wilkinson B, Bachmann BO. 2006. Biocatalysis in pharmaceuti-
cal preparation and alteration. Curr Opin Chem Biol 10:
169–176.
Smithies K, Smith MEB, Kaulmann U, Galman JL, Ward JM,
Hailes HC. 2009. Stereoselectivity of an ω-transaminase-