Biocatalytic asymmetric synthesis of L-phosphinothricin using a one-pot three enzyme system and a continuous substrate fed-batch strategy

Article abstract of DOI:10.1016/j.apcata.2019.117239

Transamination catalyzed by an aminotransferase is becoming a key tool for the production of chiral amine pharmaceuticals and agrochemicals owing to its excellent enantioselectivity and green credentials. To overcome the unfavorable thermodynamic equilibrium and the inhibition by the byproduct α-ketoglutaric acid in the transamination of 2-oxo-4-[(hydroxy)(methyl)phosphinoyl]butyric acid (PPO) to form the promising herbicide L-phosphinothricin (L-PPT), a tri-enzymatic cascade reaction system was developed by combining a robust glutamate dehydrogenase to recycle the byproduct to the amino donor L-glutamate in situ, together with a cofactor recycling process catalyzed by an alcohol dehydrogenase. Moreover, a continuous substrate fed-batch strategy was employed to alleviate the decomposition of PPO and applied to scale-up the cascade reaction to 90 L, yielding 111.4 g/L (615.4 mM) L-PPT in 99.7% yield and >99.9% ee with an productivity of 15.9 g/L?h. This combination of improved biocatalyst system and process engineering should prove to be economically competitive for industrial applications.

Full text of DOI:10.1016/j.apcata.2019.117239

Journal Pre-proof
Biocatalytic asymmetric synthesis of L-phosphinothricin using a one-pot
three enzyme system and a continuous substrate fed-batch strategy
Haisheng Zhou, Lijun Meng, Xinjian Yin, Yayun Liu, Jianping Wu,
Gang Xu, Mianbin Wu, Lirong Yang
PII:
S0926-860X(19)30394-1
DOI:
https://doi.org/10.1016/j.apcata.2019.117239
APCATA 117239
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
9 June 2019
6 August 2019
4 September 2019
Please cite this article as: Zhou H, Meng L, Yin X, Liu Y, Wu J, Xu G, Wu M, Yang L,
Biocatalytic asymmetric synthesis of L-phosphinothricin using a one-pot three enzyme system
and a continuous substrate fed-batch strategy, Applied Catalysis A, General (2019),
doi: https://doi.org/10.1016/j.apcata.2019.117239
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
©
2019 Published by Elsevier.

Title page
Article title
Biocatalytic asymmetric synthesis of L-phosphinothricin using a one-pot three enzyme
system and a continuous substrate fed-batch strategy
Author names and affiliations
Haisheng Zhou, Lijun Meng, Xinjian Yin, Yayun Liu, Jianping Wu, Gang Xu, Mianbin
Wu, Lirong Yang*
Institute of Bioengineering, College of Chemical and Biological Engineering, Zhejiang
University, 310027 Hangzhou, PR China

Corresponding author Tel: +86-571-8795-2363; Fax: +86-571-8795-2363; E-mail:
lryang@zju.edu.cn (L. R. Yang)
Graphical abstract
1

Highlights

A three-enzyme cascade reaction was developed for synthesis of L-
phosphinothricin

A continuous substrate fed-batch process was used to scale-up the reaction to 90
L


615.4 mM L-Phosphinothricin was produced in 99.7% yield and >99.9% ee
The method could be used for asymmetric synthesis of non-proteinogenic amino
acids
Abstract
Transamination catalyzed by an aminotransferase is becoming a key tool for the
production of chiral amine pharmaceuticals and agrochemicals owing to its excellent
enantioselectivity and green credentials. To overcome the unfavorable thermodynamic
equilibrium and the inhibition by the byproduct α-ketoglutaric acid in the transamination
of 2-oxo-4-[(hydroxy)(methyl)phosphinoyl]butyric acid (PPO) to form the promising
herbicide L-phosphinothricin (L-PPT), a tri-enzymatic cascade reaction system was
developed by combining a robust glutamate dehydrogenase to recycle the byproduct to
the amino donor L-glutamate in situ, together with a cofactor recycling process catalyzed
by an alcohol dehydrogenase. Moreover, a continuous substrate fed-batch strategy was
employed to alleviate the decomposition of PPO and applied to scale-up the cascade
2

reaction to 90 L, yielding 111.4 g/L (615.4 mM) L-PPT in 99.7% yield and >99.9% ee
with an productivity of 15.9 g/L•h. This combination of improved biocatalyst system and
process engineering should prove to be economically competitive for industrial
applications.
Keywords: asymmetric synthesis; aminotransferase; continuous substrate fed-batch;
enzymatic catalyze; herbicide; L-phosphinothricin.
3

1
. Introduction
L-Phosphinothricin (L-PPT; L-2-amino-4-(hydroxy(methyl)-phosphonoyl)butanoic acid)
is the active ingredient of many commercial herbicides used worldwide such as Basta,
Buster, Challenge, Finale, Harvest, and Ignite [1]. As L-PPT resembles L-glutamic acid
(
L-Glu), it functions as a false substrate and irreversibly blocks glutamine synthetase, a
vital enzyme on the nitrogen metabolic pathway of plants, and thus it possesses a broad-
spectrum, non-selective herbicidal activity [2-4]. Because of its low mammalian toxicity
and ecological compatibility [1], it has the potential for huge market share growth with
the development and promotion of genetically modified PPT-tolerant crops and the
shrinking market of its main competitors, glyphosate [5, 6] and paraquat [7]. Because
almost all commercially available formulations consist of racemic D, L-PPT as the
component and only the L-enantiomer has herbicidal efficacy [8], there is a persistent
need for efficient methods to obtain enantiopure L-PPT.
However, as L-PPT is a type of non-proteinogenic amino acid, which is rarely extracted
from natural sources or produced through direct microbial fermentation [9], it is difficult
to prepare at a low cost. Nevertheless, numerous attempts have been made to synthesize
L-PPT including via chemical asymmetric synthesis using chiral auxiliaries [10, 11] or L-
natural amino acids obtained from a chiral pool [12, 13], enzymatic optical resolution of
D, L-PPT derivatives [14-17] and enzymatic amination of the precursor PPO [18-23].
Because of the harsh operating conditions, complicated process, low yield, and
insufficient enantiopurity, few chemical routes for industrial applications have been
4

reported. Resolution of racemic phosphinothricin or its derivatives with the employment
of enzymes such as deacetylase [14, 15], amidase [16], and nitrilase [17] is a feasible way
to produce L-PPT with a high ee (enantiomeric excess) value, but the reaction is
undermined by the theoretical maximum yield not exceeding 50% in a one reaction batch
and the subsequent difficulties of recirculation of the racemate. Enzymatic amination of
the corresponding keto acid PPO, including transamination catalyzed by an
aminotransferase (transaminase, TA, EC 2.6.1.X) and reductive amination catalyzed by
an amino acid dehydrogenase (AADH, EC 1.4.1X), are promising asymmetric synthetic
routes for manufacture of L-PPT based on the easy process and green credentials.
Because transamination to PPO is the major degradation product of naturally occurring
L-PPT [24, 25], it was concluded that there are naturally occurring evolutionary TAs with
robust activity and enantioselectivity for the synthesis of L-PPT from PPO. Indeed, most
researchers have focused on transamination for the asymmetric synthesis of L-PPT [18-
2
1]. However, one of the frequently cited drawbacks of transamination is its unfavorable
thermodynamic equilibrium, that is, reactions catalyzed by TAs are reversible; although
using a stoichiometric excess of the amino donor can easily shift the equilibrium. For
example, Schulz et al. [18] used a four-fold molar excess of the amino donor L-Glu in the
presence of PPO to produce L-PPT to achieve more than 90% yield, whereas the
remaining PPO, excess L-Glu, and the byproduct α-KG made downstream purification
extremely difficult and greatly elevated the process cost. Others have attempted to address
this issue by coupling two transaminases, one utilized L-Glu as the amino donor, while
5

the second recycled the byproduct α-KG back to L-Glu using aspartic acid as the amino
donor, although only a slight excess of amino donors were used and a lower conversion
of 85% was obtained [20]. Another ingenious approach was established by decomposing
α-KG into ethylene and carbon dioxide in situ using an ethylene-forming enzyme with
only an equivalent amount of the L-Glu amino donor to increase the conversion of PPO
from <60% to >99% [21]. All these strategies effectively shifted the equilibrium to some
extent, but all still required at least a stoichiometric amount of amino donor along with
the inevitable creation of some secondary byproducts, compromising the process facility
and atom economy.
Reductive amination of PPO catalyzed by an AADH is considered another potential
asymmetric synthetic route for production of L-PPT with advantages, such as complete
conversion and no need for an organic amino donor because ammonium is used instead.
Owing to the structural similarity between L-PPT and L-Glu, AADHs for reductive
amination of PPO have been identified to be a glutamate dehydrogenase (GluDH) [22,
2
3]. However, this application is limited by the strict substrate specificity of natural
GluDHs and plagued by their low activities and stereoselectivities toward PPO [22].
Although a newly engineered NADPH-specific GluDH with an activity of 0.63 U/mg-
protein toward PPO was recently used for a reductive amination process with a conversion
of 99% (0.1 M PPO) and an L-PPT productivity of 1.35 g/h·L [23], extensive efforts are
needed to enhance its activity and substitute the cofactor with the cheaper NADH.
Therefore, a tri-enzymatic cascade reaction (Fig. 1) was constructed for asymmetric
6

synthesis of L-PPT, which combined three naturally available enzymes, a TA that
transfers the amino group from L-Glu to PPO to form L-PPT, a GluDH that recycles L-
Glu from α-KG by consuming ammonium and NADH, and an alcohol dehydrogenase
(ADH, EC 1.1.1.X) that recycles NADH using the cheap reducing agent isopropanol. In
contrast to previously reported transaminations that used at least a stoichiometric amount
of amino donor, this cascade reaction system only requires a catalytic or sub-catalytic
amount of amino donor, while shifting the reaction equilibrium to obtain a complete
conversion. Additionally, because it uses robust enzymes, this tri-enzymatic cascade
reaction should be efficient and versatile for the synthesis of other non-proteinogenic
amino acids. Finally, by applying strategies of process engineering such as continuous
substrate feeding, the cascade reaction was scaled up and the asymmetric synthesis of L-
PPT was achieved with high yield and productivity.
2
. Materials and Methods
2
.1. Strains, plasmids, enzymes, and chemicals
E. coli W3110 was maintained in our laboratory and used to amplify the aminotransferase
genes gamma-aminobutyrate aminotransferase (GABAT, GenBank: WP_001087611.1),
branched-chain amino acid aminotransferase (BCAT, GenBank: CAQ34114.1), valine-
pyruvate aminotransferase (ValAT, GenBank: AM946981.2), and aspartate
aminotransferase (AspAT, GenBank: 945553). E. coli BL21 (DE3) was used for gene
expression (Novagen, California, USA).
The vector pET-28a (+) (Novagen, California, USA) was used to construct recombinant
7

vectors for gene expression. PrimeSTAR Max DNA polymerase and restriction
endonucleases were purchased from Takara (Dalian, China). Yeast extract, tryptone,
kanamycin, and isopropyl β-D-1-thiogalactopyranoside (IPTG) were purchased from
Sangon Biotech Co., Ltd. (Shanghai, China). 2-Oxopentanoic acid and trimethylpyruvic
acid were purchased from TCI (Shanghai) Development Co., Ltd. (Shanghai, China). 2-
Oxo-4-[(hydroxy)(methyl)phosphinoyl]-butyric acid (PPO) was synthesized and purified
in our laboratory in accordance with previously reported methods [26] and characterized
by our colleague X. J. Yin [23]. Nicotinamide adenine dinucleotide derivatives
(
NAD(P)H, NAD(P)+) were purchased from Bontac Bio-engineering Co., Ltd.
Shenzhen, China). All other chemicals and reagents were chemically pure and obtained
(
commercially.
.2. Construction of recombinant vectors
2
The recombinant vectors GABAT and BCAT were previously constructed by our
laboratory [21]. The genes ValAT and AspAT were amplified by polymerase chain
reaction (PCR) using the genomic DNA of E. coli W3110 as the template. The primers
and
restriction
endonucleases
were
I)
5′-
5′-
CCGGAATTCATGACATTCTCCCTTTTTGGTGAC-3′
(EcoR
and
CCGCTCGAGTTAGTGACTTTCAGCCCAGGC-3′ (Xhol I) for ValAT, and 5′-
CCGGAATTCATGTTTGAGAACATTACCGCCGCTCC-3′ (EcoR I) and 5′-
CCCAAGCTTTTACAGCACTGCCACAATCGCTTCGC-3′ (Hind III) for AspAT. The
PCR products were analyzed using 1% (w/v) agarose gel electrophoresis, and then were
8

digested using their respective restriction endonucleases. The digested DNA fragments
were inserted into the same restriction sites of the pET-28a (+) expression vector under
the control of the T7 promoter.
All of the other enzyme genes employed in this work were synthesized and inserted
between the EcoR I and Hind III restriction sites of the vector pET-28a (+), including
aromatic aminotransferase from Klebsiella michiganensis E718 (AroAT, GenBank:
NC_018106.1) [27], glutamate dehydrogenase from Amphibacillus xylanus DSM 6626
(
GluDH, GenBank: NP_391659.2) [28], and alcohol dehydrogenase from Geobacillus
stearothermophilus (BsADH, GenBank: KR611715) [29].
.3 Expression of recombinant enzymes
2
Recombinant E. coli BL21 (DE3) was cultivated at 37 °C in Luria–Bertani medium with
kanamycin (50 μg/mL). The cells were induced for 16 h using IPTG (0.5 mM) at 25 °C,
after which time OD600 had reached approximately 0.6. The cells were then harvested
by centrifugation (12,000 × g, 5 min, 4 °C), washed with phosphate buffer (50 mM, pH
8
.0), and resuspended in half the volume of buffer for ultrasonic cell disruption to obtain
the crude enzyme extract.
.4 Purification of recombinant enzymes
The collected cells were washed twice and resuspended in 50 mM phosphate buffer (pH
8.0) at room temperature and subsequently disrupted through sonication (Sonicator 400,
2
=
Misonix, USA) in an ice bath. The crude cell extract was centrifuged at 12,000 × g for 30
min to remove the insoluble debris. The obtained supernatant was loaded onto a nickel
9

chelate affinity column (Ni-NTA Resin, Bio Basic) to purify the recombinant enzymes.
Fractions containing the eluted product were identified by sodium dodecyl sulfate-
polyacrylamide gel electrophoresis (SDS-PAGE), and then were desalted and replaced
with phosphate buffer (50 mM, pH 8.0) by ultrafiltration. Finally, 20% (w/w) glycerol
was added to the purified enzyme solution and it was stored at -20 °C.
2
.5. SDS-PAGE and protein concentration assays
The expression and purification of the recombinant enzymes were analyzed by SDS-
PAGE. The gels were stained with Coomassie Brilliant Blue G-250. Molecular weight
protein markers (PageRuler Prestained Protein Ladder, 10 to 170 kDa, Thermo Scientific,
USA) were used as references. Protein concentrations were determined using a Bradford
protein assay kit (Quick Start™, Bio-Rad, California, USA).
2
.6. Enzyme assays
The assay solution for TA activity measurements contained Tris-HCl buffer (100 mM, pH
.0), pyridoxal 5′-phosphate (PLP) (0.1 mM), amino donor (100 mM, adjusted to pH 9.0
9
with NaOH), and PPO (20 mM, adjusted to pH 9.0 with NaOH; Scheme 1). The reaction
was initiated by adding TA enzyme solution and it was allowed to react for 10 min at
3
5 °C. The enzymatic reaction was then quenched by adding 1 M HCl and the amount of
L-PPT was detected by HPLC. The TA activity was defined as the amount of enzyme that
catalyzed the formation of 1 μmol of L-PPT per min
GluDH and BsADH activity were determined at 35 °C by monitoring at 340 nm with
a TU-1810PC recording spectrophotometer equipped with a thermostated cell
10

compartment (Beijing Puhua General Instrument Co., Ltd., China). The reaction mixtures
for the GluDH assay contained Tris-HCl buffer (100 mM, pH 8.0), NADH (0.25 mM), α-
KG (20 mM, adjusted to pH 8.0 with ammonia), and 25 μL of enzyme solution in a total
volume of 1 mL. The reaction mixtures for the BsADH assay contained Tris-HCl buffer
+
(
100 mM, pH 8.0), NAD (0.25 mM), isopropanol (100 mM), and 25 μL of enzyme
solution in a total volume of 1 mL. Nonenzymatic reaction rates served as controls and
were subtracted from the reaction rate in the presence of enzyme. One unit of enzyme
activity was defined as the amount of enzyme that catalyzed generation or degradation of
-
1
-1
1
μmol of NADH per min with a molar absorption coefficient of 6.22 mM •cm under
the above experimental conditions.
.7. Initial velocity measurements
The initial velocity (v ) of the reaction catalyzed by GABAT was measured in 100 mM
2
0
Tris-HCl buffer (pH 8.0) in a 10 mL reactor placed in a thermostatic water bath under
magnetic stirring at 35 °C. For measurement of the inhibition of GABAT by PPO, the 5
mL reaction mixture comprised 1000 mM L-Glu, 0.1 mM PLP, and a various
concentration of PPO. To measure the inhibition of GABAT by α-KG, the reaction
mixture was comprised of 1000 mM L-Glu, 10 mM PPO, 0.1 mM PLP, and a various
concentration of α-KG. All the reactions were initiated by adding 25 μL of pure GABAT
enzyme solution (0.1 U/mL).
2
.8. Asymmetric synthesis of L-phosphinothricin using transamination catalyzed by
gamma-aminobutyrate aminotransferase
11

The reactions were performed in 10 mL EP tubes in a thermostatic water bath under
magnetic stirring at 35 °C. For reactions with Glu/PPO >1, the reaction mixture was
comprised of 1 U/mL GABAT crude extract, 1500 mM L-Glu (adjusted to pH 7.5 with
NaOH beforehand), 0.1 mM PLP, and a various concentration of PPO (the mixture was
neutralized by NaOH to pH 7.5). For reactions with Glu/PPO ≤1, the reaction mixture
comprised 1 U/mL GABAT, 500 mM PPO (adjusted to pH 7.5 with NaOH beforehand),
0
.1 mM PLP, and a various concentration of L-Glu (the mixture was neutralized by NaOH
to pH 7.5).
.9. Asymmetric synthesis of L-phosphinothricin using the tri-enzymatic reaction
2
The reaction was performed in a 250 mL or 1000 mL round bottom flask reactor with a
DG-150 industrial pH detector and controller (Hangzhou Lohand Biological Co., Ltd.,
China) in a thermostatic water bath under magnetic stirring. For optimization of the tri-
enzymatic reaction, the reaction volume was restricted to 100 mL and a 210 mM PPO
aqueous solution adjusted to the desired pH with ammonia was added, and then 0.1 mM
+
PLP, 0.1 mM NAD , and 250 mM isopropanol were added into the reactor followed by
the addition of 15 mM L-Glu, which was neutralized by ammonia to the desired pH value
beforehand (during optimization of the amino donor/acceptor ratio, the concentration of
L-Glu was varied from 60 mM to 0 mM). Finally, 1 U/mL crude enzyme extract was
added to initiate the reaction (during optimization of the amino donor/acceptor ratio, 2
U/mL was used). The pH of the reaction mixture was monitored using a pH detector and
controller using ammonia (about 15%, v/v). In the batch reaction mode, the reaction
12

+
mixture was composed of about 543 mM PPO, 0.1 mM PLP, 0.1 mM NAD , 660 mM
isopropanol, 25 mM L-Glu, and 10 U/mL crude enzyme extract using the optimized
reaction pH and temperature, respectively. In the intermittent fed-batch reaction mode,
the reaction mixture was composed of 0.1 mM PLP, 0.1 mM NAD+, 600 mM isopropanol,
and 25 mM L-Glu. PPO was added into the reactor as a solid, the other conditions were
as same as for the batch reaction mode, and the pH was controlled as above using 15%
ammonia. In the continuous fed-batch reaction mode, the reaction mixture was composed
of 0.1 mM PLP, 0.1 mM NAD+, and 25 mM L-Glu. The 500 mL feeding solution of 0.60
mol isopropanol, about 0.51 mol PPO, and a certain amount of pure water was
continuously added into the reactor, while the other conditions were the same as for the
intermittent fed-batch reaction mode.
2
.10. Scale-up of the continuous fed-batch reaction
The reaction was performed in a 100 L stainless steel bio-reactor equipped with a pH
detector and controller and a temperature detector and controller (Shanghai Baoxing
biological equipment Engineering Co., Ltd., China). The crude enzyme extract and co-
enzyme usage were 10 U/mL GABAT, 10 U/mL GluDH, 10 U/mL ADH, 0.1 mM PLP,
+
and 0.1 mM NAD . A total of 255 g L-Glu was neutralized with ammonia to pH 8.0 and
then was added into the reactor together with the enzyme solution. The 50 L substrate
feed solution was composed of 5 L isopropanol, 10.204 kg PPO (98% purity), and pure
water and the feeding rate was manually monitored. The reaction pH was automatically
controlled using ammonia (15%, v/v) at 8.0±0.1. The reaction temperature was
13

automatically controlled using circulating cooling water at 35±0.5 °C.
.11. Analysis
2
The reactants and products were monitored using analytical HPLC (HP-1100 HPLC
system; Hewlett-Packard Development Company, L.P., Houston, USA) equipped with a
reverse-phase C18 column (Agilent Pursuit 5 C18, 5 μm, 4.6 × 250 mm). For glutamate
acid, phosphinothricin, α-KG, and PPO detection, the flow rate was maintained at 1.0
mL/min with isocratic elution of 90% buffer A (50 mM (NH
4
)
2
HPO
4
and 0.1% (w/v)
tetrabutylammonium hydroxide adjusted to pH 3.6 with 50% (w/v) phosphoric acid) and
1
0% acetonitrile at a column temperature of 40 °C and detection at 205 nm.
The enantiomers D, L-PPT were identified by pre-column derivatization with o-
phthalaldehyde and N-acetyl-L-cysteine [30]. The flow rate was maintained at 0.8
mL/min with detection at 334 nm, and the column temperature was constant at 30 °C. D,
L-PPT was eluted with buffer B (50 mM sodium acetate)/methanol (90:10, v/v).
3
. Results and Discussion
3
.1. Screening, over-expression, and characterization of aminotransferases for
asymmetric synthesis of L-phosphinothricin
First, the feasibility of the tri-enzymatic cascade reaction for the asymmetric synthesis of
L-PPT in one-pot was assessed based on the robust activity and biochemical
characteristics of TAs toward PPO. Therefore, five representatives from all four
subgroups of TAs [31] were included in the investigation. Among them, GABAT, BCAT,
ValAT, and AspAT were cloned from E coli W3110 and expressed in E. coli BL21 (DE3).
14

Additionally, the whole gene of an aromatic aminotransferase (AroAT) from Klebsiella
michiganensis E718 was synthesized and also expressed in E. coli BL21 (DE3). All of
these five TAs were successfully cloned and overexpressed (Fig. S1). Then the
transamination activity of these TAs toward PPO with different amino donors were
determined (Table 1). The results showed that GABAT possessed the best activity
compared with the other TAs using L-Glu as the amino donor. Interestingly, BCAT also
exhibited excellent catalytic activity toward PPO using L-Glu as the amino donor,
indicating this enzyme also has the potency for catalyzing the asymmetric synthesis of L-
PPT.
Nevertheless, GABAT was demonstrated to be the best catalyst and purified
recombinant GABAT was biochemically characterized in more detail before using it in
the cascade reaction. The effect of pH and temperature on the transamination of PPO and
L-Glu was examined by using a standard enzyme assay as described herein. The
recombinant GABAT was most active in the alkalescent region, with a maximum activity
at pH 9.0, and almost no transaminase activity was detected below pH 6 or above pH 12
(Fig. 2A). The reaction velocity dramatically increased with temperature and exhibited a
maximum at 60 to 65 °C (Fig. 2B). The pH stability of the recombinant protein was
examined at 35 °C. A pH of 8.0 was found to be the best for maintaining the stabilization
of the recombinant GABAT (Fig. 2C). The enzyme was found to inactive at 45 °C but
more stable below 35 °C (Fig. 2D) and possessed a half-life of 2.81 h at 35 °C (Table 2),
indicating its characteristics were sufficient for industrial applications.
15

3
.2 Substrate and product inhibition of the gamma-aminobutyrate aminotransferase
(GABAT)
To examine the PPO substrate inhibition of GABAT, the initial velocity of the
transamination was measured at various concentrations of PPO with saturated
concentration of L-glutamate in the presence of recombinant enzyme. As depicted in Fig.
3
A, no obvious substrate inhibition was observed up to 35 mM, and then the substrate
inhibition increased as the concentration of PPO increased from 35 mM to 65 mM, finally
reaching a plateau with about a 31.5% decrease compared with the maximum initial
velocity. Because common TAs are subject to product inhibition from the ketone
byproduct [32], the product inhibition by α-KG for the GABAT was also examined at
invariant concentration of L-Glu and PPO but various concentrations of α-KG. The initial
velocity continuously decreased as the concentration of α-KG increased, and lost up to
6
8.2% at 10 mM α-KG (Fig. 3B), suggesting that product inhibition by α-KG was severe
and must be overcome for a reaction using a high concentration of substrate.
.3 Asymmetric synthesis of L-phosphinothricin (L-PPT) using the gamma-
3
aminobutyrate aminotransferase (GABAT)
With stable and active biocatalysts identified, the asymmetric synthesis of L-PPT
catalyzed only by GABAT at 35 °C was investigated. Accordingly for the transamination
process, a 10-mL EP tube reactor was packed with the same amount of the recombinant
enzyme crude extract, and various ratios of the substrate PPO and the amino donor L-Glu.
When the molar ratio of L-Glu to PPO was varied from 15/1 to 3/1, the L-Glu
16

concentration was held constant at 1500 mM and the PPO concentration was varied from
1
1
00 mM to 500 mM. Conversely, when the molar ratio of L-Glu to PPO was varied from
/1 to 1/15, the L-Glu concentration was varied from 500 mM to 33 mM and the PPO
concentration was held constant at 500 mM. The results (Fig. 4A) showed that only a 15-
fold molar excess of L-Glu with respect to PPO could drive the reaction to absolute
completion. The conversion rate dropped as the ratio of L-Glu/PPO decreased and was
nearly zero when the molar ratio of L-Glu/PPO was 1/15. That is, the asymmetric
synthesis of L-PPT using GABAT as the only catalyst alone did not occur when the
proportion of the amino donor L-Glu to all substrates was lower than 6.3% (mol/mol).
After the yields of L-PPT under the above reaction conditions were determined, the
results as shown in Fig. 4B indicated that almost all the yields of the product L-PPT did
not match the conversion rates of the substrate PPO, and that deviations of the yield from
the conversion rate increased as the concentration of PPO increased (when the molar ratio
of L-Glu to PPO ranged from 15/1 to 3/1). This may be attributed to the employment of
crude cell lysates as the catalysts, which always contains unknown background activities
resulting in undesired side reactions (e.g., the reduction of a keto acid to a hydroxy acid
by hydrogenases). However, the deviation still increased when the concentration of PPO
was fixed at 500 mM and the molar ratio of L-Glu/PPO was varied from 1/1 to 1/15.
Considering that the same dosage of crude cell lysates was used in all reactions, and
similar results were obtained by using pure recombinant GABAT, it is likely that PPO
probably underwent a chemical degradation to other chemicals under the experimental
17

condition.
.4. Stability of 2-oxo-4-[(hydroxy)(methyl)phosphinoyl]butyric acid (PPO) in the
3
biocatalytic environment
The results obtained from the examination of the asymmetric synthesis of L-PPT using
GABAT indicated that the stability of PPO in a biocatalytic environment (mild
temperature and neutral or alkalescent pH) was essential to obtain a high yield of L-PPT
whenever one or more enzymes were used in the reaction mixture, especially when a high
concentration of product was desired. Therefore, at a constant temperature of 35 °C, two
influencing factors, namely the solution pH and the initial concentration of PPO, were
investigated. Fig. 5 shows that PPO became more unstable as the pH increased to the
extent that only 17.4% remained after 12 h at pH 9.5 starting from an initial concentration
of 900 mM. PPO was definitely stable for 12 h from pH 7.0 to 7.5 at the lowest initial
concentration of 300 mM.
Moreover, addition of any other common alkaline species to neutralize PPO for the
bio-reaction would also result in a considerable loss of PPO (Fig. S2-S5). All of these
discoveries indicated that the bio-synthesis of L-PPT would be difficult, especially to
obtain a high concentration of L-PPT with an acceptable yield for industrial
manufacturing. Nevertheless, the experimental results also indicated that PPO had
sufficient stability over an appropriate reaction time at an initial concentration lower than
3
00 mM and near a neutral pH value.
3
.5. Construction of a tri-enzymatic cascade system for asymmetric synthesis of L-
18

phosphinothricin
As suggested by the results from the synthesis of L-PPT using GABAT alone, it was
necessary to construct the cascade reaction as shown in Fig. 1 to improve the manufacture
of L-PPT by overcoming the disadvantages of the transamination. Thus, a GluDH and an
ADH (BsADH) were both needed in addition to GABAT to fulfill the reaction scheme
depicted in Fig. 1. Before optimization of the conditions for the cascade reaction, the
enzymatic properties of the three recombinant enzymes were characterized (Table 3). The
results showed that both the selected GluDH and ADH were more active and stable than
GABAT. Additionally, their similar optimal temperatures and pH stabilities supported the
compatibility of these three enzymes in a single-pot reaction, indicating the great potential
of this three-enzyme one-pot system for industrial application.
3
.6. Optimization of the tri-enzymatic reaction for asymmetric synthesis of L-
phosphinothricin
Although the three enzymes were carefully selected and should have great compatibility,
the cascade reaction conditions still needed to be optimized to demonstrate its feasibility
and confirm the high conversion of the substrate PPO. Thus, the pH and temperature of
the reaction mixture as well as the ratio of amino donor/amino acceptor (Glu/PPO) were
investigated. As shown in Fig. 6A, the tri-enzymatic reaction accelerated as the reaction
broth pH increased from 7.0 to 9.0. However, when the pH was 9.5, the reaction velocity
dropped a little during the first hour compared with the maximum at pH 9.0 and then
continually declined until the reaction stopped. Conversely, the yield of L-PPT dwindled
19

as the reaction pH increased, except for at pH 7.0 where the reaction was not yet complete
within the experimental time. Based on these results, the variance in the reaction velocity
with the reaction pH may be attributed to the pH profile of GABAT (Fig. 2A), while the
variations in the yield probably resulted from the instability of PPO (described in Section
3
.4). Considering the reaction velocity and the yield of the product, the optimal pH of the
cascade reaction was 8.0, at which a yield of 98.5% was achieved within 5 h, although a
slightly higher yield of 99.1% was obtained at pH 7.5 but with a reaction time of 12 h.
Next, the temperatures of the cascade reaction at pH 8.0 were examined to find the
optimal one for synthesis of L-PPT. As shown in Fig. 6B, the reaction velocity before 1.0
h increased as the reaction temperature increased and was attributed to the temperature
profile of GABAT (Fig. 2B). After 1 h, the variation of the reaction velocity with the
temperature became more complicated because of the combined effect of the temperature
profile and thermal stability of GABAT. Therefore, the conversion rates of PPO at 40 °C
and 45 °C were 74.4% and 90.9%, respectively, while the other reactions at 25 °C, 30 °C,
and 35 °C reached complete conversion. Consequently, 35 °C was chosen as the best
temperature for the cascade reaction.
Interestingly, the properties of GluDH and ADH exhibited nearly no influence on the
optimal conditions of the tri-enzymatic cascade reaction, perhaps because of their better
thermal stabilities (Table 3) compared with GABAT. That is, the stability GABAT was a
limiting factor in this tri-enzymatic reaction system. Additionally, 210 mM PPO was
transformed using only 15 mM of the amino donor L-Glu to yield complete conversion
20

at the optimal pH and temperature demonstrating the feasibility of the cascade reaction
for the synthesis of L-PPT. To reduce the dosage of L-Glu to a catalytic amount, different
reactions with various ratios of L-Glu/PPO were carried out at the same pH and
temperature. The results indicated that the reactions with a ratio of Glu/PPO greater than
or equal to 1/15 could achieve complete conversion and the reaction velocity accelerated
rapidly as the ratio increased (Fig. 6C). Moreover, no α-KG was detected because of the
robust catalytic capacity of GluDH and ADH, which immediately recycled α-KG to L-
Glu after it was produced. This system was far more productive than the situation with
GABAT as the solo catalyst, in which the reaction was almost inert at a ratio of Glu/PPO
equal to 1/15. Although the reaction conversion at a Glu/PPO ratio of 1/35 was only
8
8.1%, a conversion near 100% using less amino donor might be expected by increasing
the dosage of GABAT.
.7. Mode of the tri-enzymatic cascade reaction
3
Use of the tri-enzymatic cascade reaction eliminated the issues of the unfavorable
thermodynamic equilibrium, the obligatory excessive amino donor, and the inhibition by
α-KG. However, the usefulness of the system for industrial applications was still limited
by inhibition from the substrate PPO and its instability. As shown in Fig. 7A, in a batch
reaction of this tri-enzymatic system for transformation of 543.6 mM PPO, 491.7 mM L-
PPT was formed and 26.1 mM PPO remained with a conversion rate of 95%, but the yield
of L-PPT was only 90%. Therefore, the traditional batch reaction mode is not suitable for
the biotransformation of PPO to L-PPT, and thus, the supply or delivery of substrates for
21

the biocatalysis must be investigated.
First, an intermittent fed-batch reaction was carried out (Fig. 7B). Based on the result
from the stability research of PPO in Section 3.4, the initial concentration of PPO was set
as 100 mM. When the residual PPO in the medium was nearly 0, the feed PPO (solid)
was added until the PPO concentration reached 100 mM. The conversion rate decreased
when the next feeding was added. After 5 feedings, 500 mM PPO was added to the
reaction mixture and a complete conversion was reached, forming 485 mM L-PPT with
a 97% yield and a productivity of 8.8 g/L/h. Compared with the batch mode, the
intermittent fed-batch mode remarkably improved the substrate conversion and product
yield. However, the reaction time decreased only a little from 12 h to 10 h. Additionally,
this intermittent feeding of a strongly acidic solid substrate to the biocatalysis system can
easily result in inactivation of the biocatalyst and bring about operating difficulties,
impeding its application for large-scale industrial manufacturing.
Therefore, a continuous fed-batch reaction mode was proposed in which PPO was
continuously fed into the reaction mixture in an aqueous solution and the feeding rate was
monitored by controlling the concentration of PPO to be lower than 80 mM. As shown in
Fig. 7C, 508.8 mM PPO was completely converted to L-PPT in 6 h using the continuous
fed-batch mode, and the product yield was 99.8% with a productivity of 15.3 g/L/h, nearly
two-fold more than the intermittent fed-batch mode. Compared with a previously
described two coupled transaminase process [20] where 500 mM PPO was consumed in
2
4 h when a final reaction broth was composed of 413.5 mM L-PPT, 86 mM glutamic
22

acid, 93.5 mM aspartic acid, 62.5 mM alanine, and 55 mM PPO, the system described
herein was more advanced.
3
.8. Scale-up of the tri-enzymatic cascade reaction using continuous fed-batch mode
Finally, a continuous fed-batch reaction system as depicted in Fig. 8A was developed to
scale-up the cascade reaction to verify its potency for large-scale industrial manufacturing.
Based on the results obtained above, a pilot-scale of 90 L was performed in a 100 L
stainless steel tank with continuously feeding of 10.204 kg PPO in aqueous solution. This
allowed a low concentration of PPO to be maintained in the reaction broth to avoid
substrate inhibition and minimize the degradation of PPO. Fig. 8B shows the result of this
pilot-scale reaction in detail. A product concentration of 615.4 mM (111.4 g/L) was
achieved with a yield of 99.7% and a complete conversion in 7 h. Thus, a productivity of
1
5.9 g/L•h was obtained, which was enhanced by more than an order of magnitude
compared with the productivity (1.35 g/L•h) in the study reported by X. J. Yin and
colleagues [23] using a new engineered NADPH-specific GluDH and a glucose
dehydrogenase. Additionally, only 21.2 mM L-Glu remained in the reaction mixture at
the end of the reaction, accounting for a proportion of 2.7% (wt/wt) in the mixture of L-
Glu and L-PPT. Although a small amount of the excessive co-substrate isopropanol and
byproduct acetone might have remained in the mixture, which can be conveniently
removed by evaporation in vacuo, no other compounds requiring complicated separation
procedures existed in the reaction broth. Taken together, the asymmetric synthesis of L-
PPT proposed herein was demonstrated to be practical for industrial applications, and its
23

process simplicity and economic value were tested and verified by the pilot-scale
experiment.
3
.9. Application of this method to other non-proteinogenic amino acids
With the successful synthesis of L-PPT and considering that the aminotransferase BCAT
was already demonstrated to have a broad substrate spectrum [33] and was used to
synthesize a series of optical pure amino acids using L-Glu as amino donor [21], it was
feasible to test the versatility of the developed method by employing BCAT instead of
GABAT to asymmetrically synthesize other compounds. Among them, L-norvaline,
which is a vital intermediate of Perindopril and is mainly produced by chemical synthesis
with low purity [34], and L-tert-leucine, which is a component of several pharmaceutical
development projects focused on tumor-fighting agents and HIV protease inhibitors [35],
are valuable targets and therefore were selected for this study. Except for PPO, for which
the activity of BCAT is shown in Table 1, two α-keto acids, namely 2-oxopentanoic acid
and trimethylpyruvic acid, were identified as substrates of BCAT. Therefore, the three-
enzymatic reaction was used to convert 2-oxopentanoic acid and trimethylpyruvic acid to
produce L-norvaline and L-tert-leucine, respectively, using the continuous fed-batch
reaction mode. A complete conversion of the 500 mM α-keto acid substrate (limited by
resources, these reactions were performed only at a 1000 mL level) with a product yield
and ee value over 99% was obtained.
4
. Conclusion
In summary, we have developed an irreversible transamination procedure that can be
24

applied as an efficient and economic process for the industrial manufacture of optically
pure non-proteinogenic amino acids. This approach is based on using a tri-enzymatic
cascade reaction system to solve the unfavorable thermodynamic equilibrium and to
eliminate the inhibition of the byproduct (α-KG) that commonly plagues transamination
catalyzed by an aminotransferase. Compared with the conventional procedure, which uses
at least a stoichiometric amount of amino donor for the transamination, only a sub-
catalytic amount of amino donor is required and no byproducts requiring complicated
separation procedures are formed. Furthermore, despite the fact that it remains unknown
what mechanism is involved in the decomposition of PPO under biocatalytic conditions,
a continuous substrate fed-batch strategy was validated to alleviate the decomposition of
PPO and was successfully applied to the scale-up of the tri-enzymatic cascade reaction
with complete conversion of PPO to L-PPT in 99.7% yield and >99.9% ee.
Further work is needed to investigate a wider range of products by selecting appropriate
enzymes for the cascade. In fact, the direct asymmetric synthesis of optically pure amines
from the corresponding keto compounds by transamination has emerged as a particularly
promising process [36]. Considering this and the unstable properties of keto substrates
[
37], this cascade reaction using a continuous substrate fed-batch strategy may provide
the engineering tools necessary to move this technology toward a rational approach for
developing large-scale applications of aminotransferases.
25

Acknowledgments
This work was financially supported by the National Natural Science Foundation of China
(No. 21476199, 21676240).
26

References
[
[
[
1] G. Hoerlein, in: George W. Ware (Eds.), Reviews of Environmental Contamination
and Toxicology, volume 138, Springer, 1994, pp. 73-145.
2] E. Bayer, K.H. Gugel, K. Haegele, H. Hagenmaier, S. Jessipow, W.A. Koenig, H
Zaehner, Helv. Chim. Acta. 55 (1972) 224-239.
3] W.R. Ullrich, C.I. Ullrich-Eberius, H. Koecher, Pestic. Biochem. Physiol. 37 (1990)
1
-11.
[
[
4] C. Wendler, M. Barniske, A. Wild, Photosynth. Res. 24 (1990) 55-61.
5] C. Gillezeau, M. van Gerwen, R.M. Shaffer, I. Rana, L. Zhang, L. Sheppard, and E.
Taioli, Environ. Health. (2019) 18:2.
[
[
6] C.M. Benbrook, Environ. Sci. Eur. (2016) 28:3.
7] T. Baltazar, R.J. Dinis-Oliveira, J.A. Duarte, M. de Lourdes Bastos, F. Carvalho, Br.
J. Pharmacol. 168 (2013) 44-45.
[
[
8] L. Maier, P.J. Lea, Phosphorus. Sulfur. 17 (1983) 1-19.
9] R.P. Hicks, A.L. Russell in: L. Pollegioni, S. Servi (Eds.), Unnatural Amino Acids,
Springer, 2012, pp. 135-167.
[
[
[
[
[
10]N. Minowa, M. Hirayama, S. Fukatsu, Bull. Chem. Soc. Jpn. 60 (1987) 1761-1766.
11]H.J. Zeiss, Tetrahedron Lett. 28 (1987) 1255-1258.
12]H.J. Zeiss, Tetrahedron 48 (1992) 8263-8270.
13]M.G. Hoffmann, H.J. Zeiss, Tetrahedron Lett. 33 (1992) 2669-2672.
14]S. Grabley, K. Sauber, US Patent 4389488 (1982), to Hoechst AG.
27

[
[
[
15]K. Bartsch, US Patent 6686181 (2004), to Aventis Corporation Gmbh.
16]L. Willms, G. Fuelling, R. Keller, Europe Patent 382113 (1989), to Hoechst AG.
17]K.F. Albizati, S. Kambourakis, A. Grubbs, B.C. Borer, US Patent 8981142 (2015),
to Strategic Enzyme Application, Inc.
[
[
18]A. Schulz, P. Taggeselle, D. Tripier and K. Bartsch, Appl. Environ. Microbiol. 56
(1990) 1-6.
19]K. Bartsch, R. Dichmann, P. Schmitt, E. Uhlmann and A. Schulz, Appl. Environ.
Microbiol. 56 (1990) 7-12.
[
[
20]K. Bartsch, A. Schulz, Appl. Environ. Microbiol. 62 (1996) 3794-3799.
21]L.J. Meng, Y.Y. Liu, H.S. Zhou, X.J. Yin, J.P. Wu, M.B. Wu, G. Xu, L.R. Yang,
Catal. Lett. 148 (2018) 3309-3314.
[
22]J.M. Fang, C.H. Lin, C.W. Bradshaw, J. Chem. Soc. Perkin Trans. 1 (1995) 967-
9
78.
[
[
[
[
23]X.J. Yin, J.P. Wu, L.R. Yang, Appl. Microbiol. Biotechnol. 102 (2018) 4425–4433.
24]K. Bartsch, C.C. Tebbe, Appl. Environ. Microbiol. 55 (1989) 711-716.
25]C.C. Tebbe, H.H. Reber, Appl. Microbiol. Biotechnol. 29 (1988) 103-105.
26]N. Minowa, N. Nakanishi, M. Mitomi, US Patent 8017797 (2011), to Meiji Seika
Kaisha Ltd.
[
27]T.L. Liao, A.C. Lin, E. Chen, T.W. Huang, Y.M. Liu, Y.H. Chang, J.F. Lai, T.L.
Lauderdale, J.T. Wang, S.C. Chang, S.F. Tsai, and Y.T. Chen, J. Bacteriol. 194
(2012) 5454.
28

[
[
28]T. Jahns, Antonie van Leeuwenhoek. 70 (1996) 89-95.
29]L. Kirmair, D.L. Seiler, A. Skerra, Appl. Microbiol. Biotechnol. 99 (2015) 10501-
1
0513.
[
[
[
[
[
30]H.J. Zeiss, Pest Manag. Sci. 41 (1994) 269-277.
31]P.K. Mehta, T.I. Hale, and P. Christen, Eur. J. Biochem. 214(1993) 549-561.
32]I. Slabu, J.L. Galman, R.C. Lloyd, and N.J. Turner, ACS Catal. 7 (2017) 8263-8284.
33]X.J. Yu, X.G. Wang, P.C. Engel, FEBS J. 281 (2014) 391-400.
34]Y. Qi, T. Yang, J. Zhou, J. Zheng, M. Xu, X. Zhang, Z. Rao, S.T. Yang, Process
Biochem. 55 (2017) 104-109.
[
[
[
35]P. Lehr, A. Billich, B. Charpiot, P. Ettmayer, D. Scholz, B. Rosenwirth, H. Gstach,
J. Med. Chem. 39 (1996) 2060–2067.
36]P. Tufvesson, J. Lima-Ramos, J.S. Jensen, N. Al-Haque, W. Neto, J.M. Woodley,
Biotechnol. Bioeng. 108 (2011) 1479-1493.
37]A.J.L. Cooper, J.Z. Ginos, and A. Meister, Chem. Rev. 83 (1983) 321-358.
29