Biocatalytic stereoinversion of d-: Para -bromophenylalanine in a one-pot three-enzyme reaction

Article abstract of DOI:10.1039/c6gc01922f

Halogenated derivatives of phenylalanine can be used as cross-coupling reagents for making drug-like molecules, and pure enantiomers of these precursors are therefore highly desirable. In our exploration of enzymatic routes to simplify the deracemisation process, the application of two enzymes, d-amino acid transaminase and phenylalanine dehydrogenase, both from Lysinibacillus sphaericus, has given promising results for the stereo-inversion of d-enantiomers of para-bromophenylalanine as the model substrate and also p-chloro/fluorophenylalanine and tyrosine. The addition of a coenzyme recycling system using ethanol and alcohol dehydrogenase reduced the amount of coenzyme needed for the reaction catalysed by phenylalanine dehydrogenase, reducing cost and permitting efficient and complete conversion of the racemic amino acids to the l-enantiomer. Relative proportions of the enzymes were optimized. The high purity of the l-enantiomer, with an ee over 99%, and the ease of the process make it an ideal alternative for deracemisation of the studied compounds.

Full text of DOI:10.1039/c6gc01922f

View Article Online
View Journal
Green
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: F. Khorsand, C. D.
Murphy, A. J. Whitehead and P. Engel, Green Chem., 2016, DOI: 10.1039/C6GC01922F.
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the Ethical guidelines still
apply. In no event shall the Royal Society of Chemistry be held
responsible for any errors or omissions in this Accepted Manuscript
or any consequences arising from the use of any information it
contains.
www.rsc.org/greenchem

Page 1 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Biocatalytic stereoinversion of D-para-bromophenylalanine in a one-pot three-
enzyme reaction
Fahimeh Khorsand1, Cormac D. Murphy1, Andrew J. Whitehead2 and Paul C. Engel1*
1School of Biomolecular and Biomedical Science, Conway Institute, University
College Dublin, Belfield, Dublin 4, Ireland
and
2GlaxoSmithKline, Currabinny, Carrigaline, Co Cork, Ireland
* Corresponding author: Address as above
Eꢀmail: paul.engel@ucd.ie
Phone: 00353 1 7166764
Key words: Phenylalanine dehydrogenase, Dꢀamino acid transaminase, paraꢀ
bromophenylalanine, deracemisation

Green Chemistry
Page 2 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Abstract
Halogenated derivatives of phenylalanine can be used as crossꢀcoupling reagents for
making drugꢀlike molecules, and pure enantiomers of these precursors are therefore
highly desirable. In our exploration of enzymatic routes to simplify the
deracemisation process, the application of two enzymes, Dꢀamino acid transaminase
and phenylalanine dehydrogenase, both from Lysinibacillus sphaericus, has given
promising results for the stereoꢀinversion of Dꢀenantiomers of paraꢀ
bromophenylalanine as the model substrate and also pꢀchloro/fluorophenylalanine and
tyrosine. The addition of a coenzyme recycling system using ethanol and alcohol
dehydrogenase reduced the amount of coenzyme needed for the reaction catalysed by
phenylalanine dehydrogenase, reducing cost and permitting efficient and complete
conversion of the racemic amino acids to the Lꢀenantiomer. Relative proportions of
the enzymes were optimized. The high purity of the Lꢀenantiomer, with an e.e. over
99%, and the ease of the process make it an ideal alternative for deracemisation of the
studied compounds.

Page 3 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Introduction
Nonꢀnatural amino acids play an important role in the drug industry as core
components of many drug molecules or as essential precursors in chemical synthesis.
The synthesis, for example, of Lꢀbiarylalanines, found in many pharmaceuticals such
as inhibitors of viral 3C protease and endothelinꢀconverting enzyme 1, 2, is dependent
on the reaction and incorporation of phenylalanine derivatives. The production of
highly enantiopure derivatives of Lꢀphenylalanine is therefore one of the initial steps
in making Lꢀbiarylalanines and the drug molecules containing these compounds.
Since the synthesis of racemic amino acids is fairly well developed, effective methods
for obtaining one single enantiomer from a racemic mixture of amino acids are of
critical importance. This can be done through either dynamic kinetic resolution or
3
stereoꢀinversion . The application of the hydantoinaseꢀcarbamoylase system for the
synthesis of a broad range of Dꢀamino acids as well as Lꢀamino acids 4, 5, and also
6
7
8
proteaseꢀ , lipaseꢀ and acylaseꢀ mediated reactions are examples of enzymatic
kinetic resolution. In stereoꢀinversion, one enantiomer is converted to the target
enantiomer or product while the other starting enantiomer remains unaffected. A wellꢀ
known example is the stereoꢀinversion of DLꢀαꢀamino acids by porcine kidney Dꢀ
9
amino acid oxidase (DAAO) and a hydride reducing agent (NaCNBH3) . However,
there are many approaches focusing on the application of two or more enzymes with
different enantioselectivities without any need for chemical reagents. The conversion
of DLꢀmethionine to Lꢀmethionine by Nakajima et al 10 used four enzymes including
DAAO, catalase, leucine dehydrogenase and a formate dehydrogenase. The industrial
production of a number of Dꢀamino acids such as Dꢀphenylalanine, Dꢀtyrosine, Dꢀ
glutamate and Dꢀalanine has been successfully performed by using recombinant
bacteria expressing Lysinibacillus sphaericus Dꢀamino acid transaminase and a

Green Chemistry
Page 4 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Proteus myxofaciens or a Proteus mirabilis Lꢀamino acid deaminase 11. The use of
multiꢀenzyme deracemisation systems is an attractive approach applicable under both
in vitro and in vivo conditions and has been extensively studied over recent years.
The recent advances in protein engineering and metabolic engineering allow
modification of existing enzymes to take part in reactions with different substrates,
widening their industrial application12. The enzymatic deracemisation of the
halogenated para derivatives of phenylalanine has been investigated in this study.
These compounds are not normally considered as substrates for wildꢀtype enzymes,
which react efficiently with natural amino acids. Nevertheless, some amino acid
metabolising enzymes do show activity with these nonꢀnatural amino acids owing to
the similarity of their structure with those of phenylalanine and tyrosine. The
incorporation/application of D and Lꢀenantiomers of paraꢀbromophenylalanine in
making D and L biarylalanines through chemoenzymatic reaction has been recently
13
studied by Ahmed et al . They used phenylalanine ammonia lyase and recombinant
Dꢀamino acid dehydrogenase to prepare the pure enantiomers of L and D paraꢀ
bromophenylalanine. The preparation of the Lꢀenantiomer of 4ꢀchlorophenylalanine
from its DL mixture has also been performed by a twoꢀenzyme system in E. coli cells
expressing Dꢀamino acid dehydrogenase and branchedꢀchain amino acid
aminotransferase 14.
In the current project, we have used a tandem reaction of two enzymes, Dꢀamino acid
aminotransferase (DAAT) and Lꢀphenylalanine dehydrogenase (PheDH), both from
Lysinibacillus sphaericus, to efficiently deracemise the para bromo, chloro and fluoro
derivatives of phenylalanine, as well as tyrosine, via stereoꢀinversion of the Dꢀ
enantiomers to L.

Page 5 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Since the predominant form of amino acids in Nature is L, Dꢀamino acidꢀspecific
enzymes are relatively rare. Nevertheless, some do exist. L. sphaericus Dꢀamino acid
aminotransferase (DAAT) catalyses the transfer of an amino group from an amino
15
acid donor to a prochiral carbon in a keto acid yielding an oxoꢀproduct and a new
amino acid. This enzyme is strictly specific for Dꢀamino acids. In the present study,
the activity of this enzyme was checked with paraꢀbromoꢀDꢀphenylalanine and good
activity was detected. With αꢀketoglutarate as amino acceptor, therefore, this enzyme
converts paraꢀbromoꢀDꢀphenylalanine to paraꢀbromophenylpyruvate according to
Equation 1.
pꢀbromoꢀDꢀphenylalanine + αꢀketoglutarate ↔ pꢀbromophenylpyruvate + Dꢀglutamate
Equation 1
The second reaction is carried out by the L. sphaericus phenylalanine dehydrogenase,
an NAD+/NADH dependent oxidoreductase which catalyses the reversible reductive
amination of a keto acid to the corresponding Lꢀamino acid. Although the wildꢀtype
enzyme shows broad substrate specificity, the activity is not very high with some nonꢀ
natural amino acids that are of interest for chemical synthesis. This has encouraged
the use of mutagenesis to modify substrate specificity and, in our group, a number of
variants with altered specificity have been made both by siteꢀdirected mutagenesis
based on structural studies (homology modelling) 16, 17 and by random mutagenesis 18.
Among these variants, mutant N145A had already shown good activity with some
para derivatives of phenylalanine 19. Here we have screened the wildꢀtype enzyme
and 19 mutant variants to find the most effective catalyst with paraꢀbromoꢀLꢀ
phenylalanine as the model substrate (Equation 2). The selected biocatalyst was used

Green Chemistry
Page 6 of 32
View Article Online
DOI: 10.1039/C6GC01922F
to carry out the second reaction in our tandem process for stereoinversion (Equation
2).
pꢀbromophenylpyruvate + NADH + NH3 + H+ ↔ pꢀbromoꢀLꢀphenylalanine + NAD+ + H2O
Equation 2
The use of a dehydrogenase entails deployment of the costly nicotinamide cofactor
NADH, which is stoichiometrically consumed in Equation 2. This is not an
economically viable proposition and, as in all such biocatalytic processes, it is
necessary to find a way of recycling the cofactor, reconverting NAD+ to NADH for
reuse. Different approaches, utilising enzymatic, chemical, electrochemical or
photochemical methods, have been developed to regenerate the nicotinamide
20
cofactors . The enzymatic approach for regeneration of NADH generally requires a
21
third oxidoreductase, ideally one with a highly soluble and cheap substrate
.
Alcohols like ethanol and 2ꢀpropanol, sugars such as glucose, glucoseꢀ6ꢀphosphate
(G6P), and glucoseꢀ6ꢀsulfate, formic acid, amino acids such as glutamic acid and also
dihydrogen have all been used as the hydrogen source in such recycling reactions.
Glutamate dehydrogenase and glutamate, for example, have been remarkably useful
in the reduction of NAD+ to NADH and were employed to regenerate NADH for a
multiꢀenzymatic system reducing carbon dioxide to methanol 22. Formate
dehydrogenase has also found favour with a number of authors, offering the
advantage that the product, CO2, is lost to the atmosphere 23. In our own earlier
studies, using PheDH for chiral synthesis, we found the alcohol dehydrogenase
reaction very suitable 24, 25. 5% ethanol not only drives the equilibrium to the right but
also helps to solubilize oxoacid substrates with limited aqueous solubility. From the

Page 7 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
standpoint of equilibrium, propanꢀ2ꢀol is a more favourable option as equilibrium for
the oxidation of a secondary alcohol lies further towards the carbonyl product.
However, with the readily available and cheap alcohol dehydrogenase from yeast, the
rate of oxidation of propanꢀ2ꢀol is very slow compared to that for ethanol.
Accordingly in this study, the first two enzymes, DAAT and PheDH, have been
combined with yeast alcohol dehydrogenase, using ethanol to recycle the
nicotinamide coenzyme and providing us with a oneꢀpot, threeꢀenzyme process
offering easy and costꢀeffective deracemisation of our target compound. Following
initial proof of principle with paraꢀbromophenylalanine, the process has been tried
out also with the corresponding fluoro, chloro and hydroxy derivatives.
Experimental
Chemicals
Analytical grade reagents were used in most cases. αꢀketoglutarate and D/Lꢀamino
acids were purchased from Sigma Chemical Co. (St Louis, MO, USA). Fluka (Buchs,
Switzerland) supplied pyridoxal phosphate (PalP). NAD+ and NADH were from
Apollo (Manchester, UK). Yeast alcohol dehydrogenase and rabbit muscle lactate
dehydrogenase (LDH) were purchased from Sigma. Diaphorase was obtained from
Enzolve Technologies (Dublin, Ireland)

Green Chemistry
Page 8 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Bacteria and plasmids
Recombinant E. coli TG1 cells harbouring the plasmid ptac85, containing the gene
encoding L. sphaericus PheDH were used to express the phenylalanine
dehydrogenases. The Dꢀamino acid transaminase (DAAT) gene inserted in plasmid
ING 2024 and cloned in E. coli BW25113 ꢀdad A was a kind gift from Ingenza Ltd.
Expression and purification of phenylalanine dehydrogenases
PheDH and mutant variants were prepared by a simplified version of the procedure
described by Seah et al.16,17 . Recombinant E. coli TG1 cells containing the ptac85
plasmid encoding either the wildꢀtype L. sphaericus PheDH gene or a mutated variant
were cultivated in LB medium containing ampicillin (100 ꢁg/mL) at 37°C with
shaking. Enzyme expression was induced by adding 1M isopropyl βꢀDꢀ
thiogalactopyranoside (IPTG) after the OD600 of the culture reached 0.6. The cells
were cultivated for a further 6 hours, harvested (11,000 × g, 40 min) and suspended in
10 mL 10 mM potassium phosphate (pH 7.8), followed by ultrasonication for 10 min
at 4°C and centrifugation at 40,000 × g for 20 min to remove cell debris. The clear
solution, brought to 30% saturation with (NH4)2SO4, was centrifuged at 20,000 × g for
20 min. The supernatant was adjusted to 60% (NH4)2SO4 saturation. After
centrifugation, the new precipitate was dissolved in buffer A (0.01 M potassium
phosphate buffer (pH 7.8) containing 1 mM EDTA and 5 mM 2ꢀmercaptoethanol) and
dialysed against the same buffer at 4°C to remove (NH4)2SO4. The enzyme solution
was loaded on a Procion RedꢀP3BN Sepharose CLꢀ6B column (6 x 2 cm), unbound
materials were washed off with buffer A and the enzyme was eluted with 0.5 M KCl
in the same buffer. Enzyme purity was checked by SDSꢀPAGE (12%) at each step.

Page 9 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Purified PheDH was stored at 4°C. and its concentration was calculated using the
absorbance at 280 nm and an extinction coefficient of 1.17 x 10ꢀ3 gꢀ1 mL cmꢀ1 17
.
Expression and purification of D-amino acid transaminase
The recombinant DAAT was overexpressed in E. coli BW25113 cells cultivated and
harvested as described above for E. coli TG1 but using LB medium containing 50
mg/mL kanamycin. Harvested cells were suspended in 10 mL 20 mM TrisꢀHCl
(pH 8), followed by ultrasonication and centrifugation as above. The precipitate from
35% ammonium sulfate was resuspended in 5 mL 20 mM TrisꢀHCl buffer pH 8 and
dialysed against the same buffer to remove the salt. The enzyme solution was loaded
on a Q Sepharose column equilibrated with 20 mM TrisꢀHCl buffer pH 8 and eluted
with a 0 to 0.5 M NaCl gradient. The concentration of purified DAAT was
determined using the Bradford Protein Assay Kit (BioꢀRad, Hercules, CA, USA) and
the purity was checked by SDSꢀPAGE.
Assay of phenylalanine dehydrogenase activity
PheDH activity was assayed using a Cary 50 ultraviolet–visible spectrophotometer
(Agilent Technologies, Cork, Ireland) at 25 °C in 1ꢀcm light path cuvettes. The
increase in A340 caused by reduction of NAD+ to NADH (ε: extinction coefficient is
6220 M−1 cm−1) was monitored after adding enzyme to a solution containing 10 mM
paraꢀbromoꢀLꢀphenylalanine, 2 mM NAD+ and 100 mM KCl in 50 mM GlyꢀKOH
buffer, pH 10.4.

Green Chemistry
Page 10 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Coupled assay of D-amino acid transaminase
The standard assay for DAAT 26 measures the production of pyruvate from Dꢀalanine.
Reaction mixtures (1 mL in 1 cm lightꢀpath cuvettes at 37 °C) contained 0.15 mM
NADH, 15 mM αꢀketoglutarate, 5 mM Dꢀalanine in 100 mM Tris/HCl (pH 8.5) and
25 Unit lactate dehydrogenase as the coupling enzyme.
The activity of the enzyme towards paraꢀbromoꢀDꢀphenylalanine was measured with
a similar coupled assay using phenylalanine dehydrogenase (10 U, approx. 160 µg) in
place of lactate dehydrogenase.
One-pot stereo-inversion of para-bromo-D-phenylalanine
The relative amounts of DAAT and PheDH were optimised by varying the
concentrations of both. The reaction solution (1.5 mL) contained 6.5 mM paraꢀ
bromoꢀDꢀphenylalanine, 15 mM αꢀketoglutarate, 10 mM NADH, 20 ꢁM PalP, 100
mM KCl and 400 mM NH4Cl in 0.1 M TrisꢀHCl, pH 8.5. In the absence of recycling,
the NADH was supplied at a sufficiently high concentration to avoid coenzyme
concentration becoming the limiting factor. Reaction at 37oC in 15 mL Falcon tubes
was initiated by adding both enzymes. To stop the reaction for HPLC analysis, 700
ꢁL of the mixture was transferred to a microtube and incubated at 95oC for 10 min to
denature the enzymes. Denatured protein was precipitated by 10 min centrifugation at
20000 x g. The supernatant was adjusted to a pH between 4 and 7, suitable for the
Astec CHIROBIOTIC T column. The sample was diluted fiveꢀfold with the mobile
phase and filtered for HPLC analysis. Percentage conversion was calculated from
integrated peak areas for the two enantiomers.

Page 11 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
With an optimal ratio of DAAT and PheDH established, a recycling system was
introduced to explore how far the coenzyme concentration could be reduced. The
initial reaction solution contained 6.5 mM paraꢀbromoꢀDꢀphenylalanine and 15 mM
αꢀketoglutarate, 20 ꢁM PalP, 0.2 mg DAAT, 4 ꢁg PheDH (~0.25 U), 5 ꢁL ethanol, 1
mg ADH (300 U). Starting at 2 mM, the concentration of NADH was progressively
reduced to find the lowest required amount. Reactions were stopped after two hours
and the product was checked by chiral HPLC.
Enzyme recycling and scale-up
The Amicon Ultraꢀ15 centrifugal filter unit (30K) was used to separate the reaction
solution from the enzyme molecules, which are not able to pass through the 30K filter
in the tube. The 6 mL reaction solution contained 6.5 mM DLꢀparaꢀ
bromophenylalanine, 15 mM αꢀketoglutarate, 20 ꢁM PalP, 20 ꢁL ethanol, 0.2 mM
NADH, 88 ꢁg PheDH, 2.3 mg DAAT and 14 mg ADH. After 2 hours, the tube was
centrifuged at 2000 × g to separate the reaction solution from the enzymes and the
production of Lꢀenantiomer was checked by HPLC. The tube containing the enzymes
trapped on the filter was used for more than 10 reaction cycles over 10 days. The tube
was stored on ice when not in use.
Authentication of the product
Reactions were monitored throughout by chiral HPLC and, in view of the known
specificity of the enzymes used, the nature of the product was not in serious doubt.
However, the collected material from the multiꢀcycle scaleꢀup was subjected to NMR
analysis. The product was obtained as a white solid after butanol extraction of 20 mL
1
of reaction solution (29.9 mg, 88% purity). This material was characterised by H

Green Chemistry
Page 12 of 32
View Article Online
DOI: 10.1039/C6GC01922F
NMR (400 MHz, DMSO) δ 2.9 (1H, dd, ꢀCβH), 3.06 (1H, dd, ꢀCβH), 3.68 (1H, dd, ꢀ
CαH) , 7.14ꢀ7.2 (2H, d, 2ArꢀH), 7.42ꢀ7.48 (2H, d, 2ArꢀH) and also 13C NMR (101
MHz, DMSO) δ 172.1 (CO), 136.5, 131.5, 132.5, 121.4 (CBr), 53.5 (Cα), 35.5 (Cβ).
Results and Discussion
Activity of PheDH with para-bromo-L-phenylalanine
An enzymatic approach to racemic resolution inevitably depends on the activity and
selectivity of the chosen enzymes towards the desired substrate(s). Additionally, the
efficiency and stability of the enzymes over repeated reaction cycles must be
considered. Initially, the idea of using PheDH as the sole enzyme was investigated by
testing different mutants. Theoretically, a suitable PheDH could convert the L half in
a racemic mixture of e.g. paraꢀbromophenylalanine to the keto acid. After separation
from the untouched paraꢀbromoꢀDꢀphenylalanine, reductive amination of the keto
acid using the same enzyme could yield the pure Lꢀenantiomer, thus effecting a
quantitative resolution. The maximum yield would be 50% of each enantiomer, but
the expected high purity of the resulting products made this route worth exploring.
To find an enzyme sufficiently active with paraꢀbromoꢀLꢀphenylalanine, twenty
variants of PheDH were screened. Some were not active enough to be studied further.
Fig. 1 shows those with fairly good activity towards paraꢀbromophenylalanine and
also lists their mutations. Three mutants, nos. 3, 10 and 11, showed higher activity
than wildꢀtype PheDH with paraꢀbromoꢀLꢀphenylalanine, and one of these, No. 3, is

Page 13 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
much more active with paraꢀbromoꢀLꢀphenylalanine than with unsubstituted
phenylalanine.
An earlier study showed that inactivity with the Dꢀenantiomer need not imply that
19
this isomer does not bind the active site. The activity of the PheDH variants was
therefore checked with paraꢀbromoꢀLꢀphenylalanine as substrate in the absence and
presence of equimolar paraꢀbromoꢀDꢀphenylalanine in order to assess the risk of
inhibition by the inactive enantiomer (Fig. 1). Although enzyme no. 3, with the
mutation N145A, has the highest activity towards paraꢀbromoꢀLꢀphenylalanine, the
activity drops remarkably in the racemic mixture, reaching only 20% of the activity
with the Lꢀenantiomer alone. The inhibition is similarly more or less present for other
variants. Although, with enantiopure starting substrate, the activity of enzyme no. 3
with paraꢀbromoꢀLꢀphenylalanine is quite high compared to other mutants, it does not
show the same superiority with a racemic substrate. Inhibition of 70 to 80% is already
seen with equimolar enantiomers, and, considering that the ratio of D to L in solution
increases as the reaction progresses, the rate would constantly decrease, making it
difficult to achieve full conversion. Using a coenzyme recycling system with
diaphorase and thiazolyl blue tetrazolium bromide (MTT), the full conversion of
paraꢀbromoꢀLꢀphenylalanine to the keto acid was not achieved after a 24ꢀhour
incubation of enzyme no. 3 with the racemic mixture of substrate (data not shown).
This brings into question the appropriateness of the PheDH mutants as an efficient
tool for the racemic resolution of paraꢀbromophenylalanine by oxidative deamination.
Interestingly, however, the inhibition effect of the Dꢀenantiomer is almost negligible
for the reductive amination reaction (data not shown). This would be of great merit if
we could couple the reaction with a Dꢀamino acidꢀspecific enzyme.

Green Chemistry
Page 14 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Oxidative deamination of para-bromo-D-phenylalanine by DAAT
The Dꢀamino acid transaminase of L. sphaericus (DAAT) is known to be active with
a broad range of substrates. If DAAT is sufficiently active with the Dꢀenantiomer in a
racemic mixture, PheDH could transform the resulting keto acid to the Lꢀenantiomer
resulting in 100% conversion to the Lꢀisomer (Fig. 2).
For DAAT purified in two steps (see above and Fig. S1) the standard coupled activity
assay with Dꢀalanine showed a specific activity of 110 U/mg, consistent with the
previous report.15 Replacing Dꢀalanine with paraꢀbromoꢀDꢀphenylalanine as substrate
necessitated a change from LDH to PheDH as the coupling enzyme. The reaction was
easily monitored by detecting the absorbance change. The specific activity of DAAT
measured with 2 mM paraꢀbromoꢀDꢀphenylalanine was 2 U/mg. It should be noted
that much higher activity could be achieved were it not for the low water solubility of
this substrate. The reaction rate with a racemic mixture of paraꢀbromophenylalanine
was similar to the rate of reaction with Dꢀenantiomer as the only starting material.
Therefore, this activity is not inhibited by the inactive enantiomer.
The bienzymatic reaction for conversion of para-bromo-D-phenylalanine to para-
bromo-L-phenylalanine
Since the conditions required for transamination by DAAT and reductive amination
by PheDH are similar, it was possible to use the same buffer (TrisꢀHCl, pH 8.5) and
temperature (37oC) in a singleꢀpot reaction, avoiding troublesome extraction of the
intermediate to start the second reaction in a new environment.
Using combinations of different concentrations of both enzymes, the lowest amounts
of enzymes giving almost 100% product yield over a twoꢀhour reaction course were

Page 15 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
investigated (Table 1). Under our conditions, in a 1.5 mL incubation, 0.2 mg DAAT
and 4 ꢁg PheDH no. 3 were able to convert all the Dꢀenantiomer to the Lꢀform as
shown by the HPLC results (Fig. S2). Owing to the very high activity of PheDH with
paraꢀbromophenylpyruvate, only a small amount of this enzyme is required. Starting
with equal concentrations of both enantiomers, all the D was consumed leading to a
doubled concentration of Lꢀenantiomer. The chromatograms clearly show the
efficiency of the studied reaction for deracemisation of paraꢀbromophenylalanine.
The fact that PheDH catalyses the reverse reaction (reductive amination) at pH values
less basic than needed for the forward reaction (oxidative deamination) is absolutely
helpful for stereoꢀinversion of paraꢀbromoꢀDꢀphenylalanine to the Lꢀenantiomer. The
presence of both enzymes in the same solution is the main reason for rapid
completion of the reaction. Since the product of the first reaction is simultaneously
removed from the medium by the second reaction, transamination goes quickly to
completion in comparison with the situation with DAAT as the sole enzyme in the
reaction.
The application of ADH for recycling the coenzyme
As mentioned earlier, oxidised nicotinamide coenzymes can be reconverted to the
reduced form by enzymatic methods thus avoiding the need for stoichiometric
addition, and in the present study yeast alcohol dehydrogenase and ethanol were
selected for this purpose. Ethanol, however, like other organic solvents, can inhibit
enzymatic reactions. To ensure that the presence of ethanol does not negatively affect
the progress of reaction, activities of the two enzymes were separately checked with
different concentrations of ethanol. The presence of low concentrations of ethanol did

Green Chemistry
Page 16 of 32
View Article Online
DOI: 10.1039/C6GC01922F
not inhibit the studied reactions and in fact had a slight positive effect in some cases
(Fig. 3). The relevant reaction conditions required only a low ethanol concentration
(0.33%), and so no inhibition was expected. The full stereoꢀinversion of 6.5 mM
paraꢀbromoꢀDꢀphenylalanine in the presence of various concentrations of coenzyme
was analysed by HPLC and the reaction yields were determined (Table 2). The
reaction proceeded to completion with NADH concentrations as low as 0.2 mM and
probably would still have done so with even less. Using even 0.02 mM NADH, a
325:1 molar ratio of substrate to coenzyme, 95% of substrate was transformed to the
product over two hours and full conversion was achieved after 5 hours incubation.
Considering 0.2 mM NADH as the appropriate amount of coenzyme for the twoꢀhour
reaction, the application of ADH/ethanol reduced the required amount of NADH
nearly 30ꢀfold, thus substantially moderating costs. The results unambiguously
confirm the efficiency of this recycling reaction.
Scale-up of the reaction using a centrifugal filter unit
In the present study, the reaction was scaled up with two purposes; firstly, to show the
possibility of reusing enzymes and secondly, to get sufficient product for further
studies. The centrifugal filter unit offered the advantage of easy separation of product
solution from enzymes (Fig. 4). The enzymes remained untouched above the filter
barrier and the small molecules like amino acids and coenzyme passed through the
membrane by centrifugation. The same tube was used for 12 rounds of reaction on
different days and the reaction was monitored to reach completion in each round. The
functionality of the enzymes gradually decreased over the period and the reaction
time became longer so that ultimately complete conversion was not attained within
two hours. The first cycle was complete in under 90 min; in the 10th round four days

Page 17 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
later with the same tube, only 90% of Dꢀenantiomer was transformed to the L within
2 hours. The conversion yield decreased to 84% after 2 hours reaction of the 11th
round on the 11th day. More than 60 mL solution was ultimately obtained,
theoretically containing 6.5 mM pure paraꢀbromoꢀLꢀphenylalanine, which was
extracted from the other components of reaction with 1ꢀbutanol. The solvent was
removed by a rotary evaporator and the final product was obtained as a white solid.
A gradual loss in activity after recycling the enzymes was not unexpected, and no
special methods for stabilising enzymes were employed. However, the enzymes
remained remarkably stable and functional considering the storage time and the
number of reaction cycles. An obvious next stage to explore in the context of
industrial application would be either encapsulation or immobilization of the enzymes
on a solid support.
Application of the method to other para-derivatives of phenylalanine
As well as paraꢀbromophenylalanine other paraꢀderivatives of phenylalanine can be
utilised as intermediates in synthesis. So, it is useful to explore the discussed method
of deracemisation with these other compounds. PheDH no. 3 has already been shown
to be active with the paraꢀchloro and –fluoro derivatives of phenylalanine 19. The
activity of wildꢀtype PheDH and variants mutated at N145 was also fairly high with
para derivatives of phenylpyruvate 27. This could be of particular importance as
PheDH catalyses the reductive deamination of the αꢀketo acids in our proposed
deracemisation method. Wildꢀtype PheDH is strongly active towards Lꢀtyrosine, the
hydroxylated derivative of phenylalanine 17, 28. Therefore the second, PheDH reaction
in the deracemisation procedure was expected to proceed smoothly with the

Green Chemistry
Page 18 of 32
View Article Online
DOI: 10.1039/C6GC01922F
mentioned derivatives of phenylalanine, but the viability of the transamination was
under question. The activity of DAAT with the Dꢀenantiomers of the three amino
acids, paraꢀchlorophenylalanine, paraꢀfluorophenylalanine and tyrosine, was
explored using the assay method based on PheDH as the coupling enzyme. Following
the detection of good activities, the threeꢀenzymatic reaction was used for these three
substrates and resulted in full stereoꢀinversion of their Dꢀenantiomers to the L (data
not shown). In the case of tyrosine the reaction required a longer time (only 97%
conversion after 4 hours).
Conclusion
Owing to different requirements for reaction conditions in chemocatalysis, oneꢀpot
multistep reactions are more challenging compared to enzymatic catalysis. Since the
optimum conditions for reactions catalysed by different enzymes are often similar,
biocatalytic cascades are much more readily applied. These series of reactions may be
performed either in sequential mode, with the next catalyst added only after the
previous step is completed, or in simultaneous mode, with all reactants and enzymes
present from the outset. Performing the reactions of a multistep process in a “oneꢀpot”
fashion, without the isolation of the intermediates, has many benefits since isolation
and purification steps are typically timeꢀconsuming, yieldꢀreducing and require large
amounts of chemicals/solvents. Avoiding these isolation steps therefore reduces
operation time, cost and environmental impact 29.
Considering the objective of this study, which was the efficient deracemisation of a
racemic mixture of amino acids, different methods have been developed.
Stereoselective deamination of an unwanted enantiomer by an amino acid oxidase

Page 19 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
followed by amination of the keto acid back to an amino acid of desired chirality by
asymmetric biocatalysis or to a racemic mixture by nonꢀstereoselective chemical
reduction are wellꢀknown processes. However, the production of H2O2, with a
consequent requirement for addition of catalase, is a substantial obstacle. The
application of two transaminases with opposite stereoselectivities was another
strategy to obtain pure enantiomers of amino acids through oneꢀpot deracemisation 30.
Although the transaminases are not considered specific to a narrow range of
substrates, it is not always easy to find two enzymes targeting different enantiomers
of a substrate efficiently. Here, we have developed a novel method for the
deracemisation of a racemic mixture of paraꢀbromoꢀLꢀphenylalanine and some
similar compounds by using a combination of a transaminase and an amino acid
dehydrogenase.
The mutant variants derived from the wildꢀtype L. sphaericus phenylalanine
dehydrogenase have demonstrated striking potential for biocatalysis. The possibility
of changing the substrate specificity of enzyme in favour of a new substrate has
helped us to assemble the tools not immediately available in Nature. The high activity
of PheDH N145A with paraꢀbromoꢀLꢀphenylalanine was a promising start, but
uexpectedly potent inhibition by the Dꢀenantiomer proved problematic. In principle it
should be possible in the future to eliminate this problem through molecular
modelling and further mutagenesis. Even so, the use of a single enzyme would
inevitably involve two reaction steps under different conditions. The addition of
DAAT to the reaction offered a viable alternative route for conversion of the Dꢀ
enantiomer to the ketoacid, avoiding the inhibition. Crucially the subsequent
conversion to Lꢀparaꢀbromophenylalanine by PheDH, increasing the potential yield
to 100%, can be carried out at the same time in the same reaction vessel.

Green Chemistry
Page 20 of 32
View Article Online
DOI: 10.1039/C6GC01922F
One obvious criticism of this reaction system is that substrate, and therefore product,
concentrations are low. This is a consequence of the low solubility of the substrate in
an aqueous buffer. Our attempts to remedy this by addition of miscible organic
solvents offered little or no improvement (results not shown). A possibility yet to be
explored is the use of a biphasic system. In a previous study of chiral synthesis of
amino acids from ketoacids with poor aqueous solubility 24, we observed efficient
conversion in systems where the enzyme and coenzyme resided in the aqueous phase
whilst the substrate resided mainly in the organic phase, which served effectively as a
highꢀconcentration reservoir. A similar approach might well overcome the solubility
problem here also.
Finally, further work is also now needed to investigate a wider range of substrates,
and to explore the immobilization of these enzymes, which should improve stability
31ꢀ33
and allow many cycles of use
.
In cascade reactions involving sequential
reactions performed in one pot, such as the reactions studied in this paper, coꢀ
immobilisation of enzymes could be ideally employed to improve enzymes’ features
34
and also facilitate their recovery. Heterofunctional supports and the preparation of
combiꢀcatalysts with the possibility of reusing the optimally immobilised enzyme
35
after the inactivation of the less stable one have recently been studied to improve
the coꢀimmobilisation methods. Nevertheless current results already indicate that this
combination of biocatalysts offers a versatile and robust tool for deracemisation of
substituted aromatic amino acids.

Page 21 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Acknowledgement
This work was supported by an Irish Research Council Enterprise Partnership Scheme
scholarship with GlaxoSmithKline as the Enterprise Partner. We thank Dr. Ian
Fotheringham of Ingenza Ltd. for the generous gift of the E. coli clone containing the
DAAT gene.

Green Chemistry
Page 22 of 32
View Article Online
DOI: 10.1039/C6GC01922F
References
1.
A. M. Prior, Y. J. Kim, S. Weerasekara, M. Moroze, K. R. Alliston, R. A. Z.
Uy, W. C. Groutas, K. O. Chang and D. H. Hua, Bioorg Med Chem Lett, 2013,
23, 6317ꢀ6320.
2.
C. A. Fink, M. Moskal, F. Firooznia, D. Hoyer, D. Symonsbergen, D. C. Wei,
Y. Qiao, P. Savage, M. E. Beil, A. J. Trapani and A. Y. Jeng, Bioorg Med
Chem Lett, 2000, 10, 2037ꢀ2039.
3.
4.
5.
6.
7.
8.
9.
10.
S. Servi, D. Tessaro and G. PedrocchiꢀFantoni, Coordin Chem Rev, 2008, 252,
715ꢀ726.
C. Syldatk, O. May, J. Altenbuchner, R. Mattes and M. Siemann, Appl
Microbiol Biotechnol, 1999, 51, 293ꢀ309.
O. May, S. Verseck, A. Bommarius and K. Drauz, Org Process Res Dev,
2002, 6, 452ꢀ457.
E. Agosta, A. Caligiuri, P. D'Arrigo, S. Servi, D. Tessaro and R. Canevotti,
Tetrahedron-Asymmetr, 2006, 17, 1995ꢀ1999.
N. J. Turner, J. R. Winterman, R. Mccague, J. S. Parratt and S. J. C. Taylor,
Tetrahedron Lett, 1995, 36, 1113ꢀ1116.
S. K. Hsu, H. H. Lo, C. H. Kao, D. S. Lee and W. H. Hsu, Biotechnol Progr,
2006, 22, 1578ꢀ1584.
T. M. Beard and N. J. Turner, Chem Commun, 2002, DOI: DOI
10.1039/b107580m, 246ꢀ247.
N. Nakajima, N. Esaki and K. Soda, J Chem Soc Chem Comm, 1990, DOI:
DOI 10.1039/c39900000947, 947ꢀ948.
11.
12.
13.
US Pat., US 5728555 A, 1998.
N. Ran, L. Zhao, Z. Chen and J. Tao, Green Chem, 2008, 10, 361ꢀ372.
S. T. Ahmed, F. Parmeggiani, N. J. Weise, S. L. Flitsch and N. J. Turner, Acs
Catal, 2015, 5, 5410ꢀ5413.
14.
15.
D. I. Kato, K. Miyamoto and H. Ohta, Biocatal Biotransfor, 2005, 23, 375ꢀ
379.
I. G. Fotheringham, S. A. Bledig and P. P. Taylor, J Bacteriol, 1998, 180,
4319ꢀ4323.

Page 23 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
16.
17.
S. Y. K. Seah, K. L. Britton, P. J. Baker, D. W. Rice, Y. Asano and P. C.
Engel, Febs Lett, 1995, 370, 93ꢀ96.
S. Y. K. Seah, K. L. Britton, D. W. Rice, Y. Asano and P. C. Engel,
Biochemistry-Us, 2002, 41, 11390ꢀ11397.
18.
19.
S. H. Chen and P. C. Engel, J Biotechnol, 2009, 142, 127ꢀ134.
F. Paradisi, P. A. Conway, A. R. Maguire and P. C. Engel, Org Biomol Chem,
2008, 6, 3611ꢀ3615.
20.
21.
22.
W. F. Liu and P. Wang, Biotechnol Adv, 2007, 25, 369ꢀ384.
W. Hummel and H. Groger, J Biotechnol, 2014, 191, 22ꢀ31.
T. Matsuda, R. Yamanaka and K. Nakamura, Tetrahedron-Asymmetr, 2009,
20, 513ꢀ557.
23.
24.
25.
26.
27.
A. S. Bommarius, K. Drauz, W. Hummel, M. R. Kula and C. Wandrey,
Biocatalysis, 1994, 10, 37ꢀ47.
G. Cainelli, P. C. Engel, P. Galletti, D. Giacomini, A. Gualandi and F.
Paradisi, Org Biomol Chem, 2005, 3, 4316ꢀ4320.
F. Paradisi, S. Collins, A. R. Maguire and P. C. Engel, J Biotechnol, 2007,
128, 408ꢀ411.
W. M. Jones, T. S. Soper, H. Ueno and J. M. Manning, Method Enzymol,
1985, 113, 108ꢀ113.
P. Busca, F. Paradisi, E. Moynihan, A. R. Maguire and P. C. Engel, Org
Biomol Chem, 2004, 2, 2684ꢀ2691.
28.
29.
Y. Asano, A. Nakazawa and K. Endo, J Biol Chem, 1987, 262, 10346ꢀ10354.
J. Muschiol, C. Peters, N. Oberleitner, M. D. Mihovilovic, U. T. Bornscheuer
and F. Rudroff, Chem Commun, 2015, 51, 5798ꢀ5811.
30.
31.
E. S. Park and J. S. Shin, Adv Synth Catal, 2014, 356, 3505ꢀ3509.
C. Mateo, J. M. Palomo, G. FernandezꢀLorente, J. M. Guisan and R.
FernandezꢀLafuente, Enzyme Microb Technol, 2007, 40, 1451ꢀ1463.
32.
C. GarciaꢀGalan, Á. BerenguerꢀMurcia, R. FernandezꢀLafuente and R. C.
Rodrigues, Adv Synth Catal, 2011, 353, 2885ꢀ2904.

Green Chemistry
Page 24 of 32
View Article Online
DOI: 10.1039/C6GC01922F
33.
34.
R. A. Sheldon and S. van Pelt, Chem Soc Rev, 2013, 42, 6223ꢀ6235.
J. Rocha‐Martín, B. d. l. Rivas, R. Muñoz, J. M. Guisán and F. López‐
Gallego, Chemcatchem, 2012, 4, 1279ꢀ1288.
35.
S. Peirce, J. J. VirgenꢀOrtíz, V. G. TaciasꢀPascacio, N. Rueda, R. Bartolomeꢀ
Cabrero, L. FernandezꢀLopez, M. E. Russo, A. Marzocchella and R.
FernandezꢀLafuente, RSC Adv., 2016, 6, 61707ꢀ61715.

Page 25 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Legends
Fig. 1 The specific activity of PheDH variants with 10 mM Lꢀphenylalanine (blue
bars) and with 10 mM paraꢀbromoꢀLꢀphenylalanine in the absence (red bars) and
presence (green bars) of 10mM paraꢀbromoꢀDꢀphenylalanine. The table shows the
mutations in each variant.
Fig. 2 Enzymatic deracemisation using the possibility of targeting both enantiomers
of bromophenylalanine. PheDH specifically interconverts the Lꢀenantiomer and the
keto acid. An enzyme able to do the same for the Dꢀenantiomer would allow the
complete deracemisation.
Fig. 3 The effect of ethanol concentration on the activity of PheDH N145A (red bars)
and DAAT (black bars).
Fig. 4 Scaleꢀup and recycling of enzymes using an Amicon centrifugal filter unit. The
combination of three reacting enzymes converts the Dꢀenantiomer of the racemic
amino acid substrate quantitatively to the L even with amino acid to coenzyme ratios
in excess of 100. Centrifugation separates the small molecules such as substrates and
products from the enzymes. The enzymes remaining on the filter layer catalyse the
next round of reaction with a new solution of substrates.

Green Chemistry
Page 26 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Table 1
Deracemisation of 6.5 mM DLꢀparaꢀbromophenylalanine in a reaction mixture
containing 15 mM αꢀketoglutarate, 10 mM NADH, 20 µM PalP, 100 mM KCl and
400 mM NH4Cl in 0.1 M TrisꢀHCl, pH 8.5. In a 1.5 mL incubation the total additions
of PheDH and DAAT were varied in order to discover minimal amounts required to
achieve production of 99% Lꢀamino acid within 2h.
DAAT mass (mg)
L-bromophenylalanine%
PheDH mass (ꢁg)
0.05
0.05
0.1
8
16
16
8
85.7
90.1
96.5
97.5
98.4
>99
>99
0.1
0.15
0.2
8
8
0.2
4

Page 27 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Table 2
Effect of NADH concentration on the product yield after 2 hours reaction
NADH Conc. (mM) Yield % in the presence of
NADH recycling system
Yield % in the absence of
NADH recycling system
0.02
0.2
1
95
2.6
4.3
22
>99
>99
>99
2
39

Green Chemistry
Page 28 of 32
View Article Online
DOI: 10.1039/C6GC01922F
Fig. 1

Page 29 of 32
Green Chemistry
View Article Online
DOI: 10.1039/C6GC01922F
Fig. 2