A multidisciplinary approach toward the rapid and preparative-scale biocatalytic synthesis of chiral amino alcohols: A concise transketolase-/ ω-transaminase-mediated synthesis of (25,35)-2-aminopentane-1,3-diol

Article abstract of DOI:10.1021/op900190y

Chiral amino alcohols represent an important class of value-added biochemicals and pharmaceutical intermediates. Chemical routes to such compounds are generally step intensive, requiring environmentally unfriendly catalysts and solvents. This work describes a multidisciplinary approach to the rapid establishment of bio- catalytic routes to chiral aminodiols taking the original synthesis of (2S,3S)-2-aminopentane-1,3-diol as a specific example. An engineered variant of Escherichi coli transketolase (D469T) was used for the initial asymmetric ynthesis of (3S)-1,3-dihydroxypen- tan-2-one from the achiral substrates propanal and hydroxypyru- vate. A bioinformatics led strategy was then used to identify and clone an ε-transaminase from Chromobacterium violaceum (DSM30191) capable of converting the product of the transketo- lase-catalysed step to the required (2S,3S)-2-aminopentane-1,3-diol using isopropylamine as an inexpensive amine donor. Experiments to characterize, optimize and model the kinetics of each reaction step were performed at the 1 mL scale using previously established automated microwell processing techniques. The microwell results provided excellent predictions of the reaction kinetics when the bioconversions were subsequently scaled up to preparative scales in batch stirred-tank reactors. The microwell methods thus provide process chemists and engineers with a valuable tool for the rapid and early evaluation of potential synthetic strategies. Overall, this work describes a concise and efficient biocatalytic route to chiral amino alcohols and illustrates an integrated multidisciplinary approach to bioconversion process design and scale-up.

Full text of DOI:10.1021/op900190y

Organic Process Research & Development 2010, 14, 99–107
A Multidisciplinary Approach Toward the Rapid and Preparative-Scale Biocatalytic
Synthesis of Chiral Amino Alcohols: A Concise Transketolase-/
ω-Transaminase-Mediated Synthesis of (2S,3S)-2-Aminopentane-1,3-diol
Mark E. B. Smith,† Bing H. Chen,‡,¶ Edward G. Hibbert,‡,0 Ursula Kaulmann,§,9 Kirsty Smithies,† James L. Galman,†
Frank Baganz,‡ Paul A. Dalby,‡ Helen C. Hailes,† Gary J. Lye,‡ John M. Ward,§ John M. Woodley,⊥ and Martina Micheletti*,‡
Department of Chemistry, UniVersity College London, 20 Gordon Street, London WC1H 0AJ, U.K., Department of Biochemical
Engineering, UniVersity College London, Torrington Place, London WC1E 7JE, U.K., Institute of Structural and Molecular
Biology, UniVersity College London, Gower Street, London WC1E 6BT, U.K., and Department of Chemical and Biochemical
Engineering, Technical UniVersity of Denmark, 2800 Lyngby, Denmark
Abstract:
drug discovery to the final manufacturing process as rapidly
and cost-effectively as possible. Rising costs related to the
development of increasingly complex pharmaceuticals (with
multiple chiral centers and functional groups) are compounded
by increasingly stringent environmental legislation and the drive
toward sustainable processes. Working together with industry
these are the technologies and challenges that the multidisci-
plinary UCL Bioconversion - Chemistry - Engineering interface
(BiCE) programme aims to address.
While there are extensive synthetic transformations for which
chemical (catalytic) conversions are most appropriate, biocata-
lytic (enzyme and microbial) strategies have been increasingly
adopted where mild conditions and high regio- or stereoselec-
tivity are desired.1,2 The majority of biocatalytic processes
reported to date have involved the use of a single isolated
enzyme. However, recent developments in metabolic engineer-
ing and synthetic biology have highlighted the possibility of
using multienzyme synthetic pathways to perform more com-
plex syntheses.3,4 Furthermore, many of the enzymes that could
be used in series do not necessarily exist naturally together in
known metabolic pathways. This raises the possibility of
creating de noVo pathways in engineered microorganisms using
existing powerful tools such as rDNA technology. As a
demonstration of such an approach we have previously devised
a synthetic scheme using a transketolase (TK) and a transami-
nase (TAm) to create an optically enriched 2-amino-1,3-diol
using an engineered Escherichia coli strain.5
Chiral amino alcohols represent an important class of value-added
biochemicals and pharmaceutical intermediates. Chemical routes
to such compounds are generally step intensive, requiring envi-
ronmentally unfriendly catalysts and solvents. This work describes
a multidisciplinary approach to the rapid establishment of bio-
catalytic routes to chiral aminodiols taking the original synthesis
of (2S,3S)-2-aminopentane-1,3-diol as a specific example. An
engineered variant of Escherichi coli transketolase (D469T) was
used for the initial asymmetric ynthesis of (3S)-1,3-dihydroxypen-
tan-2-one from the achiral substrates propanal and hydroxypyru-
vate. A bioinformatics led strategy was then used to identify and
clone an ω-transaminase from Chromobacterium Wiolaceum
(DSM30191) capable of converting the product of the transketo-
lase-catalysed step to the required (2S,3S)-2-aminopentane-1,3-diol
using isopropylamine as an inexpensive amine donor. Experiments
to characterize, optimize and model the kinetics of each reaction
step were performed at the 1 mL scale using previously established
automated microwell processing techniques. The microwell results
provided excellent predictions of the reaction kinetics when the
bioconversions were subsequently scaled up to preparative scales
in batch stirred-tank reactors. The microwell methods thus provide
process chemists and engineers with a valuable tool for the rapid
and early evaluation of potential synthetic strategies. Overall, this
work describes a concise and efficient biocatalytic route to chiral
amino alcohols and illustrates an integrated multidisciplinary
approach to bioconversion process design and scale-up.
Chiral 2-amino-1,3-diols are an important class of pharma-
ceutically relevant compounds, and their motif is present in
antibiotics,6-10 antiviral glycosidase inhibitors,11,12 and sphingo-
1. Introduction
The pharmaceutical industry today faces significant chal-
lenges in bringing new drugs to the market. A key area of
interest is in harnessing new technologies to progress from initial
(1) Pollard, D. J.; Woodley, J. M. Trends Biotechnol. 2007, 25, 66.
(2) Woodley, J. M. Trends Biotechnol. 2008, 26, 321.
(3) Ro, D. K.; Paradise, E. M.; Ouellet, M.; Fisher, K. J.; Newman, K. L.;
Ndungu, J. M.; Ho, K. A.; Eachus, R. A.; Ham, T. S.; Kirby, J.; Chang,
M. C.; Withers, S. T.; Shiba, Y.; Sarpong, R.; Keasling, J. D. Nature
2006, 440, 940.
* Corresponding author. E-mail: m.micheletti@ucl.ac.uk.
† Department of Chemistry, University College London.
‡ Department of Biochemical Engineering, University College London.
§ Institute of Structural and Molecular Biology, University College London.
⊥ Department of Chemical and Biochemical Engineering, Technical University
of Denmark.
(4) Chang, M. C.; Eachus, R. A.; Trieu, W.; Ro, D. K.; Keasling, J. D.
Nat. Chem. Biol. 2007, 3, 274.
(5) Ingram, C. U.; Bommer, M.; Smith, M. E.; Dalby, P. A.; Ward, J. M.;
Hailes, H. C.; Lye, G. J. Biotechnol. Bioeng. 2006, 96, 559.
(6) Controulis, J.; Rebstock, M. C.; Crooks, H. M. J. Am. Chem. Soc.
1949, 71, 2463.
¶ Current address: Department of Chemical and Biochemical Engineering,
College of Chemistry and Chemical Engineering, Xiamen University, Xiamen
361005, China.
(7) Wu, G.; Schumacher, D. P.; Tormos, W.; Clark, J. E.; Murphy, B. L.
J. Org. Chem. 1997, 62, 2996.
0 Current address: Illumina Cambridge Ltd., Chesterford Research Park,
Cambridge CB10 1XL, UK.
(8) Veeresa, G.; Datta, A. Tetrahedron Lett. 1998, 39, 8503.
(9) Boruwa, J.; Borah, J. C.; Gogoi, S.; Barua, N. C. Tetrahedron Lett.
2005, 46, 1743.
9 Current address: Biotica Ltd., Chesterford Research Park, Cambridge CB10
1XL, UK.
10.1021/op900190y  2010 American Chemical Society
Published on Web 12/29/2009
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
99

lipids.13,14 A number of chemical methods exist to produce this
class of amino alcohols including those starting from the chiral
pool materials serine8,15,16 and glucose,17 Sharpless asymmetric
dihydroxylation with subsequent regioselective azide substitu-
tion18,19 and nucleophilic attack of R,ꢀ-epoxy carboxylic esters
with azide followed by azide and ester reduction.20 However,
these procedures are either step-intensive and/or make use of
toxic catalysts. We have recently developed the first nonenzy-
matic one-pot synthesis of R,R′-dihydroxyketones via a mimetic
of the transketolase reaction.21 However, no asymmetric variant
of this synthesis has been developed to date, and thus reductive
amination of the racemic products obtained from this reaction
can only be manipulated into mixtures of diastereoisomers.
Alternatively, using TK in combination with a TAm potentially
offers a highly concise, stereospecific, and benign biocatalytic
route to this key class of synthons.
In ViVo TK catalyses the transfer of a two-carbon ketol unit
from D-xylulose-5-phosphate, to either D-ribose-5-phosphate or
D-erythrose-4-phosphate.22 The enantioselective carbon-carbon
bond-forming ability of TK, together with the ability to yield
an irreversible reaction when using ꢀ-hydroxypyruvate (HPA)
as the ketol donor, makes it very attractive as a biocatalyst in
industrial synthesis.23,24 Although nonhydroxylated aliphatic
aldehydes can be accepted by TK, the activity is typically very
low. Using active-site targeted saturation mutagenesis we have
recently identified several mutants with improved activities,
notably a novel TK mutant, D469T, with a nearly 5-fold
increase in specific activity when screened towards the non-
hydroxylated aldehyde acceptor substrate, propionaldehyde
(PA), for the production of 1,3-dihydroxypentan-2-one (DHP).25
Other mutants with improved, and even reversed, enantiose-
lectivity have also been described.26 ω-Transaminases, such as
that isolated from Vibrio fluVialis, have been shown to aminate
a wide range of aldehyde and ketone substrates. This is not
true for R-transaminases which typically have a strong prefer-
ence for either R-ketoacids or R-amino acids as substrates.27
However, several transaminases which efficiently aminate an
aromatic R,R′-dihydroxyketone have been described.28 Recently
we used the published sequence of the ω-TAm isolated from
V. fluVialis to conduct a bioinformatics-based search of genome
homologues, thereby facilitating the recruitment and charac-
terization of novel ω-TAms.28 The production of chiral aliphatic
2-amino-1,3-diols from chiral ketodiols using the ω-TAm
recruitedfromChromobacteriumViolaceumDSM30191(CV2025)
would demonstrate the significant commercial potential of this
enzyme and class on bioconversion.
From an engineering perspective, there are several factors
contributing to the successful implementation of a multienzy-
matic process such as a linked TK-ω-TAm bioconversion. In
order to achieve a competitive final product yield the reaction
rates and initial substrate loading(s) need to be maximised,
whilst at the same time overcoming issues of substrate or
product inhibition of each enzyme. The identification of the
“best” biocatalyst at an early stage is crucial since recent
advances in protein engineering have enabled the subsequent
modification of enzymes to achieve greater activity, enhanced
stability and enantioselectivity, and wider substrate range.29
Ultimately, experiments performed early during development
need to provide insight into the most suitable strategies for
process optimization and scale-up in order to maximize the yield
of product on substrate and catalyst. In this regard we have
recently established a range of automated and microscale
(100-1000 µL) experimental techniques to successfully mimic
key bioprocess unit operations. An understanding of the
engineering fundamentals governing experimentation at this
scale underpins the ability to obtain quantitative results capable
of predicting larger-scale process performance.30-32
The aim of this work is to demonstrate a challenging and
novel example of a two-step stereoselective biocatalytic syn-
thesis of 2-amino-1,3-diols using a multidisciplinary approach
that integrates (1) enzymes obtained either by directed evolution
or bioinformatics-based cloning strategies, (2) microscale
experimentation and robotics for rapid process optimisation and
scale-up, (3) advanced high-throughput analytical techniques
for small-molecule detection and analysis, and (4) establishment
of an enzymatic kinetic model from microscale data for the
rapid design of the larger-scale bioconversion process. Specif-
ically, as shown in Scheme 1, we have focused on the TK
D469T-catalysed conversion of propanal (1) and hydroxypyru-
vate (HPA) (2) to (3S)-1,3-dihydroxypentan-2-one (DHP) (3)
and the subsequent CV2025 ω-TAm-catalysed conversion of
DHP to (2S,3S)-2-aminopentane-1,3-diol (APD) (4).
(10) Hajra, S.; Karmakar, A.; Maji, T.; Medda, A. K. Tetrahedron 2006,
62, 8959.
(11) Hughes, A. B.; Rudge, A. J. J. Nat. Prod. Rep. 1994, 11, 135.
(12) Liang, P. H.; Cheng, W. C.; Lee, Y. L.; Yu, H. P.; Wu, Y. T.; Lin,
Y. L.; Wong, C. H. Chem. Biochem. 2006, 7, 165.
(13) Nakamura, T.; Shiozaki, M. Tetrahedron 2001, 57, 9087.
(14) Hannun, Y. A.; Obeid, L. M. J. Biol. Chem. 2002, 277, 25847.
(15) Ndakala, A. J.; Hashemzadeh, M.; So, R. C.; Howell, A. R. Org. Lett.
2002, 4, 1719.
(16) Azuma, H.; Takao, R.; Niiro, H.; Shikata, K.; Tamagaki, S.; Tachibana,
T.; Ogino, K. J. Org. Chem. 2003, 68, 2790.
(17) Chaudhari, V. D.; Kumar, K. S. A.; Dhavale, D. D. Org. Lett. 2005,
7, 5805.
2. Results and Discussion
(18) He, L.; Byun, H. S.; Bittman, R. J. Org. Chem. 2000, 65, 7627.
(19) Smithies, K.; Smith, M. E. B.; Kaulmann, U.; Galman, J. L.; Ward,
J. M.; Hailes, H. C. Tetrahedron: Asymmetry 2009, 20, 570.
(20) Takanami, T.; Tokoro, H.; Kato, D. I.; Nishiyama, S.; Sugai, T.
Tetrahedron Lett. 2005, 46, 3291.
2.1. Multidisciplinary Approach and Strategy. Taking
into account the lack of kinetic information available on the
TK D469T and CV2025 ω-TAm-catalyzed reactions, a detailed
(21) Smith, M. E. B.; Smithies, K.; Senussi, T.; Dalby, P. A.; Hailes, H. C.
Eur. J. Org. Chem. 2006, 1121.
(27) Okada, K.; Hirotsu, K.; Hayashi, H.; Kagamiyama, H. Biochemistry
2001, 40, 7453.
(22) Sprenger, G. A.; Schorken, U.; Sprenger, G.; Sahm, H. Eur. J. Bio-
chem. 1995, 2, 525.
(28) Kaulmann, U.; Smithies, K.; Smith, M. E. B.; Hailes, H. C.; Ward,
J. M. Enzyme Microb. Technol. 2007, 41, 628.
(23) Hobbs, G. R.; Lilly, M. D.; Turner, N. J.; Ward, J. M.; Willetts, A. J.;
Woodley, J. M. J. Chem. Soc., Perkin Trans. I 1993, 165.
(24) Hibbert, E. G.; Dalby, P. A. Microb. Cell Fact. 2005, 4, 29.
(25) Hibbert, E. G.; Senussi, T.; Smith, M. E. B.; Costelloe, S. J.; Ward,
J. M.; Hailes, H. C.; Dalby, P. A. J. Biotechnol. 2008, 134, 240.
(26) Smith, M. E. B.; Hibbert, E. G.; Jones, A. B.; Dalby, P. A.; Hailes,
H. C. AdV. Synth. Catal. 2008, 350, 2631.
(29) Dalby, P. A. Curr. Opin. Struct. Biol. 2003, 13, 500.
(30) Doig, S. D.; Pickering, S. C. R.; Lye, G. J.; Woodley, J. M. Biotechnol.
Bioeng. 2002, 80, 42.
(31) Micheletti, M.; Lye, G. J. Curr. Opin. Biotechnol. 2006, 17, 611.
(32) Micheletti, M.; Barrett, T.; Doig, S. D.; Baganz, F.; Levy, M. S.;
Woodley, J. M.; Lye, G. J. Chem. Eng. Sci. 2006, 61, 2939.
100
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Scheme 1. Two-step biocatalytic synthesis of 2-amino-1,3-diols using transketolase (TK) and transaminase (TAm)
initial characterization was considered necessary for both
enzymes in order to determine the optimal reaction conditions
in terms of substrate concentration ratios, enzyme loading, and
in the case of the ω-TAm-catalysed step, potential use of a
cosolvent. This lack of initial kinetic information is a common
feature at the early stages of bioconversion process evaluation
and so is typically the starting point for experimental studies.
Automated microscale experimentation was initially used for
this purpose, and experiments were planned with a view to
subsequent integration of the two steps and ultimately process
scale-up. Development of a successful and scalable enzymatic
synthesis of APD (4) involved specifically: (i) evaluation of
the TK D469T-catalysed step (section 2.2), (ii) evaluation of
the CV2025 ω-TAm-catalysed step (section 2.3), and (iii)
integration of the two enzymatic steps such that a concise
synthesis of APD (4) could be successfully modeled and then
verified experimentally at preparative scale (section 2.4).
Underpinning this approach was the establishment of high-
throughput assays to characterize the synthetic potential of the
enzymes and to quantitatively determine the influence of process
conditions on biocatalyst performance.
2.2. TK Evaluation. Mutants of Escherichia coli transke-
tolase with improved substrate specificity towards the non-
natural aldehyde substrate propanal have been obtained by
directed evolution in a previous study.25 In particular mutant
D469T was identified from an active site library of TK mutants
as being a likely catalyst, showing a 4.9-fold increase in specific
activity with this substrate, compared to wild-type TK. The
D469 residue was previously identified as important for
substrate binding by Scho¨rken who carried out a kinetic analysis
of several E. coli transketolase mutants obtained by using site-
directed mutagenesis.33,34 This mutant enzyme was therefore
selected for use in the proposed biocatalytic process demonstra-
tion (Scheme 1). The availability of an analytical method for
the study of the TK-catalyzed reaction was crucial in order to
quantify the reaction kinetics. Several TK assays have been
reported for product quantification;23,35-37 however, an HPLC
assay was established25 in order to detect both HPA (2) starting
material and the product of the propanal reaction, DHP (3). The
assay time and sample loading needed were minimized to
increase the throughput in accord with the microscale experi-
mentation strategy. In-house chemical synthesis21 of HPA (2)
and racemic DHP (3) provided samples for calibration purposes
to ensure complete quantitative analysis of this reaction. As
propanal is a rather volatile substrate, control experiments were
carried out to confirm that no loss was occurring during
sampling and that all mass balances were closed.
The TK D469T-catalyzed reaction required optimization in
order to achieve the highest possible product yield and reaction
rate. For simplicity of operation, a batch reaction approach to
performing the bioconversion was adopted. The final concentra-
tion of both propanal (1) and HPA (2) were dictated to a degree
by the necessary requirement to preincubate the enzyme with
its cofactors, ThDP and Mg2+, prior to the introduction of the
substrates. It has been previously documented that aldehyde
substrates react with primary amines forming a Schiff base and
may therefore react with enzymes exhibiting these functional
groups in accessible positions, thus affecting enzyme activity.35
A significant effect of glycolaldehyde on holo- and apo-
transketolase activity has been shown at a range of aldehyde
concentrations.35 For this reason a strategy of working at an
excess of propanal was immediately ruled out. Taking into
account the economic implications of working at high HPA
concentrations, it was decided on the basis of these observations
and requirements to optimize the TK-catalysed step using
equimolar and stoichiometric concentrations of propanal and
HPA.
All TK D469T evaluation experiments were carried out at
1 mL scale and at room temperature using previously estab-
lished automated microwell methods.31 At this scale the reaction
pH was controlled at pH ) 7 using appropriate buffers and
concentrations. Initial experiments were carried out at 50, 100,
200, and 300 mM concentrations of reagents in stoichiometric
amounts with 30% v/v cell-free lysate (Figure 1a). At 200 mM
substrate concentration the reaction went to completion in 8-10
h, with reaction rates increasing from 22.2 mmol L-1 h-1 of
DHP (3) to 57 mmol L-1 h-1 at 50 and 200 mM initial substrate
concentrations, respectively. When the initial substrate concen-
tration was set at 300 mM, an initial rate of 45.3 mmol L-1
h-1 was obtained, indicating the probable effects of aldehyde
inhibition and/or toxicity on the enzyme. Increasing the amount
of enzyme used increased the final product yield (Figure 1b).
Increasing the lysate concentration from 30% to 50% v/v, whilst
maintaining the substrate concentration at 300 mM, allowed
complete conversion to DHP (3) to be achieved in 5-10 h and
with an initial rate of 92.5 mmol L-1 h-1. This set of microwell
reaction conditions was therefore selected as the optimum for
the TK D469T-catalysed synthesis of the intermediate product
(3S)-1,3-dihydroxypentan-2-one (DHP) (3) and was subse-
quently adopted for the larger-scale reaction (section 2.4).
2.3. TAm Evaluation. The value of the well-known V.
fluVialis ω-TAm JS17 for synthetic applications has already
(33) Bisswanger, H., Schellenberger, A., Eds. Biochemistry and Physiology
of Thiamin Diphosphate Enzymes; A.u.C. Intemann, Wissenschaftlicher
Verlag: Prien, Germany, 1996.
(34) Scho¨rken, U. InVestigation of Structure and Function of Transketolase
and Transaldolase from Escherichia coli and Biochemical Charac-
terization of the Enzymes; Institut fu¨r Biotechnologie, Ju¨lich, Germany,
1997.
(35) Mitra, R. K.; Woodley, J. M.; Lilly, M. D. Enzyme Microb. Technol.
1998, 22, 64.
(36) Sevestre, A.; He´laine, V.; Guyot, G.; Martin, C.; Hecquet, L.
Tetrahedron Lett. 2003, 44, 827.
(37) Smith, M. E. B.; Kaulmann, U.; Ward, J. M.; Hailes, H. C. Bioorg.
Med. Chem. 2006, 14, 7062.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
101

Figure 1. Characterization of the TK D469T reaction step at
1 mL scale. (a) Effect of substrate, propanal and HPA,
concentration on product formation kinetics at stoichiometric
conditions with 30% v/v lysate. (b) Effect of enzyme concentra-
tion on product formation kinetics at stoichiometric conditions
and 300 mM substrate concentration.
Figure 2. Characterization of the TAm reaction step at 1 mL
scale. (a) Effect of cosolvent and type on product formation
kinetics using the CV2025 ω-TAm. (b) Effect of substrate, IPA,
and DHP ratio on product formation kinetics.
envisaged as the likely method for isolation of the intermediate
product DHP (3) from the initial TK D469T-catalysed reaction,
an appraisal of solvent tolerance was carried out for the C.
Violaceum ω-TAm. All evaluation experiments were carried
out at the 1 mL microwell scale using buffers for pH control
as in the TK studies. Ethyl acetate (EtOAc) and toluene were
initially evaluated as solvents with differing polarities and
physicochemical properties. The influence of these solvents on
the kinetics of the CV2025 ω-TAm-catalysed conversion of
DHP (3) to APD (4) using isopropylamine (IPA) as the amine
donor is illustrated in Figure 2a. When the transamination of
DHP (3) in EtOAc was carried out, no aminodiol product was
formed. Similarly, no product was detected when the reaction
was carried out in a mixture of EtOAc and water. Using toluene
as a cosolvent proved to be more positive with an excellent
initial rate of reaction observed. However, after a relatively short
period of time the reaction ceased, suggesting that the solvent
may be having an adverse effect on the CV2025 ω-TAm over
time. It was therefore concluded that use of a cosolvent was
not appropriate in the TAm-catalysed step. Consequently, it
would be necessary to isolate the intermediate DHP product
from the TK D469T-catalysed reaction by solvent extraction
prior to the second CV2025 ω-TAm step.
been widely recognized. Both the ketol 3-hydroxy-2-butanone
(acetoin) and the aromatic amino alcohol 2-phenylglycinol are
reported to be accepted by the enzyme as substrates.38 V.
fluVialis and related ω-TAms therefore have the potential to be
used as biocatalysts in the preparation of chiral 2-amino-1,3-
diols. Since the V. fluVialis ω-TAm is not publicly available, a
detailed search for homologues of this enzyme was carried out
in the genome databases of fully sequenced bacteria using the
published protein sequence of the enzyme. This led to the
identification, cloning into E. coli, and partial characterization
of 12e ω-TAms from a variety of bacteria. The most efficient
ω-Tam28 for aminodiol synthesis among all those recruited was
that from C. Violaceum DSM30191 (CV2025). Initial experi-
ments using the CV2025 enzyme with DHP (3) indicated it
was converted to APD (4), and so further experiments were
performed using this enzyme.
Preliminary experiments to investigate the potential of
running a one-pot reaction found that the CV2025 ω-TAm could
act upon the TK substrates in the presence of an amine donor
(data not shown). Consequently, it was decided to run the
synthesis as two sequential steps. Since solvent extraction was
(38) Shin, J. S.; Kim, B. G. J. Org. Chem. 2002, 67, 2848.
102
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Table 1. Characterization of the CV2025 ω-TAm reaction at
1 mL scale: impact of the amount of cell lysate on the extent
of APD (4) formation over 24 h based on initial substrate
concentrations of 40 mM of IPA and 60 mM of DHP
lysate (enzyme)
concentration (% v/v)
[APD] after
6 h (mM)
[APD] after
24 h (mM)
30
40
50
60
70
80
8.7
11.8
12.8
13.5
14.3
18.1
22.7
22.5
19.6
21.5
20.0
22.0
Given that the product yield from the ω-TAm step will
depend upon the equilibrium position of the reaction, another
crucial variable with regard to process optimization is the ratio
of the amine donor to DHP (3). Initial experiments to investigate
this were conducted with a number of amine donors (data not
shown), systematically varying each of the substrate concentra-
tions while maintaining the other constant. The aim of these
experiments was to identify donors with minimal inhibition
effects on the enzyme. Preliminary studies conducted with (S)-
(R)-methylbenzylamine, which has been used in other trans-
aminase systems,39 showed the detrimental effect of using an
excess of this amine donor, probably due to its toxicity towards
the enzyme. Attempts were made to reduce this effect by
carrying out the reaction in the presence of a nonpolar organic
solvent, the aim being to diminish the effective concentration
of the amine in the ω-TAm-containing aqueous phase whilst
ensuring that the DHP (3) remains in the reactive phase. The
use of hexadecane allowed for higher concentrations of amine
to be present in the reactor; however, the maximum working
concentration of (S)-(R)-methylbenzylamine was still limited
to 40 mM. Alternative amine donors were therefore examined,
and similar conversions were achieved with a cheaper, more
volatile, and more water-soluble substrate, isopropylamine
(IPA). At comparable reaction concentrations the quantity of
product 4 formed was similar when using (S)-(R)-methylben-
zylamine and IPA without the need for additional organic
solvent. In addition, any excess IPA (and its product, acetone)
could be more easily removed downstream compared to the
aromatic amine.
Figure 2a shows the results obtained at different substrate
ratios of the selected amine donor, IPA, and DHP (3) using
30% v/v lysate. The highest concentration of APD (4) achieved
was 30 mM, obtained with 40 mM of IPA and 60 mM DHP.
In order to further increase the reaction rate, increased quantities
of cell lysate were used that allowed for the same amount of
product to be formed in shorter periods of time. The results of
these experiments are shown in Table 1. Approximately the
same amount of product formed using 30% v/v cell lysate in
48 h could be obtained in 24 h using 80% v/v cell lysate. The
enzyme activity values, based on initial rates, were 10.8 and
15.6 U g-1TAm for 30% and 80% v/v cell-free lysate concentra-
tions, respectively. A compromise is therefore needed between
the cost of preparing an increased amount of cell lysate and
the overall reaction time when operating conditions for scale-
up are selected. It is noteworthy that the reaction could be
Figure 3. Comparison of TK D469T reaction kinetics at
microwell (open symbols) and 100 mL (solid symbols) scales
and with the kinetic model predictions (lines). TK reaction
performed at 300 mM stoichiometric substrate concentration
and 40% v/v lysate. Model predictions are based on eq 1 using
the kinetic parameters shown in Table 2 determined from
microscale experiments.
potentially driven to completion by feeding of one of the two
substrates or by enzymatic44 or in situ product removal.45
2.4. TK-TAm Reaction Scale-up, Modeling, and Integra-
tion. Once the optimal conditions for each biocatalytic step had
been assessed, the reactions optimized at microwell scale were
scaled to 100 mL scale for the TK D469T reaction and 333
mL for the ω-TAm reaction. Both reactions were performed
in laboratory-scale batch stirred-tank reactors. At the 100 mL
scale, the pH of the TK-catalysed reaction was maintained at
pH 7.0 by the use of automated pH stat addition of 1 M HCl
(aq) to minimize the salt concentration in the reactor outlet and
waste stream. The kinetic profiles for the TK D469T reactions
conducted at both 100 and 1 mL scale, using 300 mM of both
HPA and PA and 40% v/v cell lysate, are shown in Figure 3.
The results of the microwell optimized conditions have scaled
excellently, showing very similar kinetic profiles after a 100-
fold scale-up. Calculated values of the initial reaction rates
varied by just 15%, confirming the utility of the automated
microwell experiments for rapid process optimization and scale-
up. The product DHP (3) was extracted from the reaction via
multiple solvent extractions into EtOAc. Subsequent removal
of the solvent under vacuum gave the product as an oil in 68%
yield. The ee of the product was determined by diacetylation
and chiral GC to be comparable to that previously reported of
64%.26 No reduction in ee was thus observed during the
transition to preparative-scale operation. This crude DHP
material was taken forward without further purification as
substrate for the following ω-TAm step of the synthesis.
At this stage in process development a reliable enzyme
kinetic model for the TK D469T mutant would be desirable in
order to predict reaction rate and yield over a range of bioreactor
operating conditions. The general TK reaction mechanism is
well-known and leads to the following kinetic expression based
on competitive substrate inhibition:
(39) Shin, J. S.; Kim, B. G. Biosci. Biotechnol. Biochem. 2001, 65, 1782.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
103

(4) whilst allowing all other impurities to be washed away. The
APD was then released from the column by elution with 4 M
ammonia in methanol, giving the product as a solid in 26%
isolated yield. Reports using (S)-(R)-methylbenzylamine as the
amine donor with ω-TAms and our studies using the C.
Violaceum ω-TAm with aromatic R,R′-dihydroxyketones have
indicated selectivity for formation of the S-amine.19,39 The ee
of the product 4 at C-2 was determined to be >98% and
postulated from previous work to be the 2S-isomer. The de of
4 was 61% by chiral HPLC, the impurity being exclusively
(2S,3R)-2-aminopentane-1,3-diol. This indicated that the C.
Violaceum ω-TAm was able to accept both (R)- and (S)-isomers
of 3, highlighting its potential versatility, and demonstrated
excellent stereoselectivity in catalyzing the conversion of the
ketone DHP (3) to the amine moiety in APD (4).
Table 3 provides a summary of the reaction and recovery
yields of each step of the TK D469T and CV2025 ω-TAm-
catalysed route to APD (4). For the TK reaction there is
particularly good agreement between the reaction rates (as
previously discussed) and yields achieved at the two scales. The
isolated yield of the final product at 26% is acceptable at this
stage since little attention has been directed toward optimization
of the recovery process. This two-step biocatalytic synthesis
represents an extremely concise and atom-efficient route to such
aminodiols. Indeed, a recent study focused on compound
screening for fungicidal activity reported the nine-step synthesis
of the anti-2S,3R-isomer of APD.42 The biocatalytic route
reported here thus represents a significant improvement. On the
basis of the process design information obtained at the microwell
scale (sections 2.2 and 2.3) and subsequently verified here at
the preparative scale, Figure 4 shows a feasible process
flowsheet of how the complete two-step biocatalytic synthesis
might proceed at a manufacturing scale. Standard geometry
stirred tank bioreactor configurations would be best suited for
the two bioconversion steps since for the TK reaction at least
mass transfer limitations have been reported when performed
without mixing or shaking at volumes higher than 10 mL.43
The product of the first biocatalytic reaction, an aqueous mixture
of DHP and other lysate components, can be followed by a
suitable enzyme recovery step, such as ultrafiltration, from
where the TK enzyme could be recycled to the bioreactor. A
solvent extraction step using EtOAc in a simple mixer/settler
unit is necessary to separate the DHP (3) from the other
components. After removal of the EtOAc under vacuum, the
DHP would then be redissolved in water to the optimal
concentration for the following CV2025 ω-TAm reaction.
Following enzyme separation the product mixture from this
reaction containing ADP (4) and unreacted substrates would
d[Q]
dt
)
kcatEi[A][B]
(1)
[A]
Kia
[B]
Kib
Kb[A] 1 +
+ Ka[B] 1 +
+ [A][B] +
(
)
(
)
K
KaK
a [B][Q] +
ib[Q]
Kiq
Kiq
Solutions to such models are, however, difficult to establish
due to the large number of experiments required to determine
the specific kinetic constants for a particular enzyme mutant
and set of reaction substrates. We have previously reported a
model-driven approach to kinetic parameter identification that
combines automated microscale experimentation with numerical
methods for kinetic parameter estimation from a minimum
number of experiments.40,41 This approach was used here to
rapidly determine the kinetic parameters for the new TK D469T
mutant for DHP (3) synthesis from propanal (1) and HPA (2).
The specific reaction kinetic parameters obtained are pre-
sented in Table 2. Comparing the model parameters it can be
seen that there is a major difference in reaction rate between
use of propanal (1) as substrate compared to the kinetic
parameters previously reported for glycolaldehyde.42 Another
key difference is that the Michaelis constant for propanal is 6
times that of glycolaldehyde. The consequence of this for
process design is that significantly more TK D469T will be
required for the propanal/HPA reaction in order to achieve the
same bioconversion rate. Using eq 1 and the kinetic parameters
listed in Table 2 from microscale experimentation, Figure 3
also shows the predicted preparative-scale bioconversion kinet-
ics (solid and dashed lines). The good agreement between the
experimentally determined preparative-scale reaction kinetics
and the model predictions confirms the utility of the modeling
approach. Such modeling tools can build confidence for process
chemists and engineers in process scale-up, reactor sizing, and
first approximations of the process costs.
Following the TK D469T-catalysed synthesis of DHP (4),
the subsequent CV2025 ω-TAm reaction was performed using
40 mM IPA and 60 mM DHP together with cofactor PLP and
30% v/v clarified lysate in HEPES buffer. After 24 h, an
Amberlite IRA-410 resin was added with stirring to the final
product mixture to remove any HEPES from solution. The
filtrate was then passed through an Isolute SCX-2 (Biotage)
ion-exchange column which retained the desired APD product
Table 2. Enzyme kinetic parameters for the TK
D469T-catalysed conversion of propanal, PROP (1), and
HPA (2) into (3S)-1,3-dihydroxypentan-2-one (3)a
kinetic parameters
rate constant: Kcat (min-1
Michaelis constant for HPA: Ka (mM)
Michaelis constant for PROP: Kb (mM)
inhibition constant for HPA: Kia (mM)
inhibition constant for PROP: Kib (mM)
inhibition constant for DHP: Kiq (mM)
value
(40) Chen, B. H.; Hibbert, E. G.; Dalby, P. A.; Woodley, J. M. A.I.Ch.E.J.
2008, 54, 2155.
)
501
12
98
43
625
681
(41) Chen, B. H.; Micheletti, M.; Baganz, F.; Woodley, J. M.; Lye, G. J.
Chem. Eng. Sci. 2009, 64, 403.
(42) Thevissen, K.; Hillaert, U.; Meert, E. M. K.; Chow, K. K.; Cammuea,
B. P. A.; Van Calenbergh, S.; Franc¸ois, I. E. J. A. Bioorg. Med. Chem.
Lett. 2008, 18, 3728.
(43) Matosevic, S.; Micheletti, M.; Woodley, J. M.; Lye, G. J.; Baganz, F.
Biotechnol. Lett. 2008, 30, 995.
a Kinetic parameters determined using the method described in section 2.4 from
34 microscale (1 mL) experiments at
concentrations.
(44) Koszelewski, D.; Lavandera, I.; Clay, D.; Rozzell, D.; Kroutil, W.
AdV. Synth. Catal. 2008, 350, 2761.
a range of substrate and enzyme
(45) Lye, G. J.; Woodley, J. M. Trends Biotechnol. 1999, 17, 395.
104
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Table 3. Comparison of microscale and preparative-scale yield data for the TK D469T (batch reactor 1) and CV2025 ω-TAm
(batch reactor 2) conversions showing reaction conditions, product concentrations, and isolated product yields (N/D ) not
determined)
batch reactor R1
[D469T, 300 mM HPA/PA, TK 40% v/v]
batch reactor R2
[CV2025, 40 mM IPA/60 mM DHP, TAm 30% v/v]
reaction scale (mL)
product [DHP] (mM)
isolated yield (%)
product [APD] (mM)
isolated yield (%)
1
100
300
300
N/D
68
22.7
37.8
N/D
26
undergo a simple resin-based purification step, as described
above, prior to a final evaporation step to obtain the dry solid
product.
and used without further purification. 1H NMR and 13C NMR
spectra were recorded at the field indicated using a Bruker
AMX300 MHz and AMX500 MHz machines. Coupling
constants are measured in hertz (Hz) unless otherwise specified,
NMR spectra were recorded at 298 K. Mass spectra were
recorded on a Thermo Finnegan MAT 900XP spectrometer.
LC-MS was performed on a Finnigan LTQ mass spectrometer
and an Agilent 1100 HPLC (G1322A degasser, G1311A
quaternary pump, and a G1367A autosampler). The data were
processed using Xcalibur Quan browser. Lithium hydroxypyru-
vate was synthesised as previously described.46 The TK D469T
mutant was engineered, overexpressed in E. coli, and obtained
as a clarified lysate as reported elsewhere.25 Similarly the
CV2025 ω-TAm from C. Violaceum was cloned, overexpressed
in E. coli, and obtained as a clarified lysate as reported
previously.28
Automated Microscale Bioconversion Methods. TK D469T
and CV2025 ω-TAm bioconversions were carried out in 96-
deep square well (96-DSW) and 96-standard round well (96-
SRW) microplate formats over a range of experimental con-
ditions. Automated microscale process sequences involving
bioconversion setup and operation, sample acquisition, and
processing were established for both reaction steps shown in
Scheme 1. Automation was achieved using a Genesis Worksta-
tion (Tecan, Berkshire, UK) equipped with a liquid handling
arm and a RoMa arm for plate manipulation. Disposable tips
were used for reagent addition. The accuracy and precision of
pipetting were determined for the low viscosity liquids and
dispense volumes used in this work, and the error was less than
5% in all cases.32 All the experiments were carried out in a
Class II robotic containment cabinet.
3. Conclusion
The two-step biocatalytic preparative synthesis of (2S,3S)-
2-aminopentane-1,3-diol from the achiral starting substrates,
propanal and hydroxypyruvate, and the inexpensive amine
donor, isopropylamine, represents a concise and efficient route
to this class of compound. The synthesis was facilitated by the
ability to engineer the TK enzyme to accept the non-natural
substrate propanal and the use of bioinformatics-led strategies
to identify a suitable ω-TAm able to accept the intermediate
TK product and convert it into the final desired synthon. The
use of automated microscale processing techniques facilitated
rapid optimisation and initial scale-up of the two biocatalytic
steps. Such microscale approaches are particularly useful at the
early stages of process development where quantities of key
intermediates and information of their reaction kinetics are
scarce.
While the TK D469T mutant displays desirable reaction rates
and yields, alternative recently identified mutants could be
used,25 or the D469T mutant could be further engineered to
specifically improve the ee of the product of this reaction step.
Similarly, now that a suitable ω-TAm has been cloned and
overexpressed, strategies to increase the rate and yield of this
step could be explored including directed evolution of the
enzyme or the application of techniques to shift the reaction
equilibrium to product formation such as enzymatic byproduct
removal44 or in situ product removal.45
4. Experimental Section
Materials. Unless otherwise noted, solvents and reagents
were reagent grade from commercial suppliers (Sigma-Aldrich)
Analytical Methods. Progress of the TK D469T-cataly-
sed synthesis of (S)-1,3-dihydroxypentan-2-one was moni-
tored by HPLC: 100% of 0.1% TFA in water, 0.6 mL/
Figure 4. Proposed manufacturing process flow sheet for the large-scale, two-step TK D469T (batch reactor 1) and CV2025 ω-TAm
(batch reactor 2) biocatalytic synthesis and purification of (2S,3S)-2-aminopentane-1,3-diol.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
105

min, a 15 cm C18 column, 0.1% TFA (aq), 0.6 mL/min,
UV detection at 210 nm, HPA 2 8.95 min, DHP (3) 15.55
min, propanal 22.53 min.25
the pH was adjusted to 7.0 using 1 M NaOH (aq). Following
the 20 min enzyme incubation, the solution of hydroxypyruvate
and propanal was added to the enzyme solution with stirring,
and the reaction commenced. The pH of the reaction was
maintained at 7.0 using a pH stat (Metrohm Stat Titrino) Via
addition of 1 M HCl (aq) and the progress of the reaction
followed by HPLC. After 22 h, the crude reaction mixture was
extracted with EtOAc (10 × 100 mL), and the combined
organics were dried over MgSO4. Filtration and concentration
of the organics yielded the desired compound (2.40 g, 68%).
The ee of the product was determined by chiral GC to be 61%
(S-isomer). Found (+HRCI) MH+, 119.07043; C5H11O3 re-
quires 119.07082. 1H NMR (300 MHz; D2O): δ 0.82 (3H, t, J
7.5, CH3), 1.49-1.76 (2H, m, CH2CH3), 4.22 (1H, dd, J 7.7
and 4.4, CHOH), 4.37 (1H, d, J 19.4, CH2OH), 4.46 (1H, d, J
19.4, CH2OH); 13C NMR (75 MHz; H2O): δ 8.7, 26.5, 65.3,
76.1, 214.5.
Preparation of (2S,3S)-2-Aminopentane-1,3-diol. (S)-1,3-
Dihydroxypentan-2-one (2.36 g; 0.5 M in water, 40 mL)
was added to a solution of isopropylamine (1.25 mL; 6.7
M in water, 4.54 mL, pH 7.5) in 1 M HEPES (23.3.mL,
pH 7.5) containing and pyridoxal 5′-phosphate hydrate
(17.1 mg, 69 µmol) To this mixture was further added
water (165.5 mL) and clarified CV2025 ω-TAm lysate
(100 mL). Lysate was obtained in 100 mM HEPES, pH
7.5, containing 0.2 mM PLP, 0.5 mM phenylmethyl
sulfonyl fluoride (PMSF), and 8 U of benzonase (Novagen)/
mL of cell extract. The reaction was shaken in a 500 mL
Erlenmeyer flask at 100 rpm and 37 °C for 48 h and
monitored exclusively by LC-MS. The enzyme was spun
down (4500 rpm, at 4 °C for 10 min), and the mixture
was additionally filtered through a 0.2 µm sterile filter.
Amberlite IRA-410 resin (50 g) was added to the reaction
mixture. After 1 h, the resin was removed by filtration
and washed with H2O (300 mL). The aqueous mixture
was then passed through an Isolute SCX-2 (Biotage) ion-
exchange column (40 g of resin) and the column washed
with MeOH (2 × 200 mL). The column was then eluted
with 4 M NH3 in MeOH (3 × 200 mL) and the eluent
concentrated to yield the desired product as a solid (410
mg, 26%). The ee of the (2S,3S)-product was determined
to be >98% (S-isomer at C-2) and the de determined as
61% by chiral HPLC, the impurity being exclusively
(2S,3R)-2-aminopentane-1,3-diol. Found (+HRCI) MH+,
120.10250; C5H14NO2 requires 120.10245; 1H NMR (500
MHz; D2O): δ 0.93 (2.4H, t, J 7.5, CH3), 0.94 (0.6H, t, J
6.5, CH3), 1.40-1.60 (2H, m, CH2CH3), 2.76 (0.8H, ddd,
J 6.3, 5.1, and 4.6, CHNH2), 2.81 (0.4H, ddd, J 7.5, 5.9,
and 4.1, CHNH2), 3.47-3.76 (3H, m, CH2OH and
CHOH);13C NMR (75 MHz; H2O): δ (2S,3S) 10.1, 26.3,
55.8, 64.1,73.7; (2S,3R) 9.9, 25.6, 56.2, 63.7, 74.7.
The ee of the (S)-1,3-dihydroxypentan-2-one was determined
by first derivatising the ketodiol to the diacetate as previously
described.26 The assay was performed against a racemic sample
of the diacetate using chiral GC: ꢀ-Dex column (Supelco, 30 m
× 0.25 mm); injection volume, 1 µL; carrier gas, He; carrier
gas pressure, 15 psi; injector temperature, 250 °C; oven
temperature, 60 °C and then increased at 10 °C/min to 160 °C
and held; detector temperature, 300 °C; detection, flame-ionised
detector (FID). Retention times: (R)-isomer, 13.7 min; (S)-
isomer, 13.9 min.
Progress of the ω-TAm-catalysed synthesis of (2S,3S)-2-
aminopentane-1,3-diol was monitored by LC-MS. Ten micro-
liters of the aminodiol biotransformation reaction mixture was
dissolved in 990 µL acetonitrile. The solutions were then
homogenised by vortex for 10 s in a 2 mL centrifuge tube.
This was followed by centrifugation at 13000 rpm for 5 min.
The supernatant was transferred to a glass autosampler vial and
1 µL injected into the chromatographic system. For screening
the reaction, separation between DHP and APD was achieved
using isocratic conditions of 40% 10 mmol NH4OAc and 60%
acetonitrile at a flow rate of 0.2 mL min-1, and the analytical
column was a ZIC-HILIC column (50 mm × 2.1 mm, i.d., 5
µm particle size) maintained at 20 °C. Operating conditions of
the ESI interface in positive ion mode: capillary temperature
245 °C, capillary voltage 2 kV, spray voltage 4 kV, sheath gas
30, auxiliary gas 0, sweep gas 10 arbitrary units. For LC-MS
calibration purposes stock standard solutions (10 mmol) of APD
were prepared in water. Six calibration standards were prepared
(0.6-12 ng/mL) covering the possible concentration range of
the reaction (0-10 mmol conversion) and six point calibration
curves were generated. Over the concentration ranges linear
regression of observed peak areas against concentration gave
correlation coefficients from 0.9935 to 0.9996. Limit of quan-
titation (LOQ) for APD from the low-level spiked matrix with
1 µL injected on a 2.1 mm i.d. column was determined (S/N )
3) to be 0.6 ng/mL.
The ee and de of the (2S,3S)-2-aminopentane-1,3-diol (4)
was determined by first derivitising the aminodiol to the
tribenzoate. The assay was performed against a four diastereo-
isomer sample of the tribenzoate (synthesised as described in
the Supporting Information) using chiral HPLC: Chiracel-OD
column (Daicel); mobile phase, isopropanol/hexane (5:95); flow
rate, 0.8 mL/min, detection, UV 214 nm. Retention times; 26.7
min, 31.5 min, 40.6 min, 47.8 min for the (2R,3S)-, (2S,3S)-,
(2S,3R)-, and (2R,3R)-isomers, respectively.
Preparation of (S)-1,3-Dihydroxypentan-2-one. Thiamine
diphosphate (111 mg, 240 µmol) and MgCl2 ·6H2O (183 mg,
900 µmol) were added to a reaction vessel containing H2O (6.7
mL), and the pH was adjusted to 7.0 using 0.1 M NaOH (aq).
To the stirred solution was added clarified TK D469T lysate
(40 mL, XL10:pQR412(D469T), containing 1.37 mg/mL of TK
by densitometry), and the mixture stirred for 20 min. In another
flask, lithium hydroxypyruvate (3.30 g, 300 mmol) and propanal
(2.16 mL, 30 mmol) were dissolved in H2O (53.3 mL), and
Acknowledgment
We thank the UK Engineering and Physical Sciences
Research Council (EPSRC) for support of the multidisciplinary
Bioconversion Integrated with Chemistry and Engineering
(BiCE) programme (GR/S62505/01). For further information
see http://www.ucl.ac.uk/biochemeng/industry/bice. Financial
(46) Dickens, F. Biochim. Prep. 1962, 9, 86.
106
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

ThDP
TK
thiamine diphosphate
transketolase
support from the 13 industrial partners supporting the BiCE
programme is also acknowledged.
NOMENCLATURE
Supporting Information Available
APD
DHP
EtOAc
HPA
IPA
(2S,3S)-2-aminopentane-1,3-diol
1,3-dihydroxypentan-2-one
ethylacetate
Further detail of experimental methods including synthesis
and analysis of the four diastereomer mixture of 2-aminopen-
tane-1,3-diol hydrochloride and the chiral HPLC assay for the
determination of ee and de of transaminase-derived 2-amino-
pentane-1,3-diol and synthesis of related standards. This material
is available free of charge via the Internet at http://pubs.acs.org.
ꢀ-hydroxypyruvate
isopropylamine
PA
propionaldehyde
PLP
pyridoxal 5′-phosphate
transaminase
Received for review July 24, 2009.
OP900190Y
TAm
TFA
trifluoroacetic acid
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
107