Efficient Synthesis of D-Phenylalanine from L-Phenylalanine via a Tri-Enzymatic Cascade Pathway

Article abstract of DOI:10.1002/cctc.202100237

D-phenylalanine is an important intermediate in food and pharmaceutical industries. Here, to enable efficient D-phenylalanine biosynthesis from L-phenylalanine, a tri-enzymatic cascade was designed and reconstructed in vivo. The activity of Proteus vulgaris meso-diaminopimelate dehydrogenase (PvDAPDH) toward phenyl pyruvic acid was identified as the limiting step. To overcome, the tension in the phenyl pyruvic acid side-chain, PvDAPDH was engineered, generating PvDAPDH^{W121A/R181S/H227I}, whose catalytic activity of 6.86 U mg^?1 represented an 85-fold increase over PvDAPDH. Introduction of PvDAPDH^{W121A/R181S/H227I}, P. mirabilis L-amino acid deaminase, and Bacillus megaterium glucose dehydrogenase in E. coli enabled the production of 57.8 g L^?1 D-phenylalanine in 30 h, the highest titer to date using 60 g L^?1 L-phenylalanine as starting substrate, which meant a 96.3 % conversion rate and >99 % enantioselectivity on a 3-L scale. The proposed tri-enzymatic cascade provides a novel potential bio-based approach for industrial production of D-phenylalanine from cheap amino acids.

Full text of DOI:10.1002/cctc.202100237

The European Society Journal for Catalysis
Supported by
Accepted Article
Title: Efficient synthesis of D-phenylalanine from L-phenylalanine via a
tri-enzymatic cascade pathway
Authors: Lu Cui, Zhang Sheng, Song Wei, Chen Xiulai, Liu Jia, Liu
Liming, and Wu Jing
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.202100237
Link to VoR: https://doi.org/10.1002/cctc.202100237
01/2020

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
Efficient synthesis of D-phenylalanine from L-phenylalanine via a
tri-enzymatic cascade pathway
Cui Lu,[a,b] Sheng Zhang,[c] Wei Song,[a,b] Jia Liu,[b] Xiulai Chen,[b] Liming Liu,[b] and Jing Wu*[a]
Corynebacterium glutamicum CgDAPDH[10], is approximately 327
Abstract: D-phenylalanine is an important intermediate in the food
and pharmaceutical industry. Here, to enable efficient D-
phenylalanine biosynthesis from L-phenylalanine, a tri-enzymatic
cascade was designed and reconstructed in vivo. The activity of
Proteus vulgaris meso-diaminopimelate dehydrogenase (PvDAPDH)
toward phenylpyruvic acid was identified as the limiting step. To
overcome, the tension in the phenylpyruvic acid side-chain,
amino acids (aa) in length and shows high selectivity for meso-
DAP[9]. Type II, represented by Symbiobacterium thermophilum
StDAPDH[11], encompasses an average of 299 aa and accepts a
wide range of substrates, including meso-DAP and other D-amino
acids[9, 12]. However, the narrow substrate spectrum of type I and
the lower catalysis rate of type II DAPDHs often limit the industrial
production of D-amino acids. Therefore, structure-guided
engineering of the binding pocket and random mutagenesis have
been applied to broaden the substrate spectrum or improve
catalytic efficiency. For example, mutating residues R196, T171,
H245, Q151, and D155 in CgDAPDH to R196M, T171I, H245N,
Q151L, and D155G yielded the variant CgDAPDHBC621, whose
substrate range also included aliphatic and aromatic amino
acids[6]. Similarly, structure-guided site-saturation mutagenesis in
the pocket of StDAPDH yielded StDAPDHW121L/H227I, which
displayed dramatically improved catalytic activity toward bulky 2-
keto acids, such as PPA (70-fold) and 2-oxo-4-phenylbutyric acid
(34-fold)[13]. As NADPH serves as a hydrogen donor during
DAPDH-mediated catalysis, cofactor recycling by glucose
dehydrogenase (GDH)[14] or formate dehydrogenase (FDH)[15] is
required. Hence, one of the protein engineering strategies
employed to DAPDH was to change cofactor preference from
NADPH to NADH. Accordingly, introduction of R35S/R36V/Y76I
mutations in StDAPDH increased the NAD+/NADP activity ratio
PvDAPDH was engineered, generating PvDAPDHW121A/R181S/H227I
,
whose catalytic activity of 6.86 U mg-1 represented an 85-fold increase
over PvDAPDH. Introduction of PvDAPDHW121A/R181S/H227I, P. mirabilis
L-amino acid deaminase, and Bacillus megaterium glucose
dehydrogenase in E. coli enabled the production of 57.8 g L-1 D-
phenylalanine in 30 h, the highest titer to date using 60 g L-1 L-
phenylalanine as starting substrate, which meant a 96.3% conversion
rate and >99% enantioselectivity on a 3-L scale. The proposed tri-
enzymatic cascade provides a novel potential bio-based approach for
industrial production of D-phenylalanine from cheap amino acids.
Introduction
D-phenylalanine (D-Phe), a non-protein amino acid, is used as a
building block for pharmaceuticals and fine chemicals[1]. Currently,
D-Phe is produced mainly via enzymatic methods[2], including (i)
hydantoinase processes[3], (ii) transamination[4], (iii) resolution of
racemate[5], and (iv) reductive amination. The latter, which is
catalyzed by meso-diaminopimelate dehydrogenase (DAPDH,
EC 1.4.1.16), has the greatest potential[6] as it can covert
phenylpyruvic acid (PPA) and NH4Cl to D-Phe in a single step with
high enantioselectivity.
DAPDH, an NADP+-dependent oxidoreductase[7], catalyzes the
reversible oxidative deamination of meso-diaminopimelate
(meso-DAP) in the D-center and stereoselective reductive
amination from the corresponding α-keto acid[8]. Based on
substrate specificity and structural characteristics[9], DAPDH can
be divided in two classes. Type I, represented by
from 0.13 to 1.39[16]
.
In this study, an artificial enzymatic cascade capable of
converting L-Phe to D-Phe was designed and reconstructed in
vivo. A “tension release” strategy affecting the substrate side-
chain was developed to improve the catalytic efficiency of
PvDAPDH toward PPA and increase D-Phe titers. Finally, by
integrating the best catalytic variant, PvDAPDHW121A/R181S/H227I
,
into the cascade, 57.8 g L-1 D-Phe was synthesized with 96.3%
conversion and >99% enantiomeric excess (ee) within 30 h in a
3-L fermentor.
Results
[a]
C. Lu, W. Song, Prof. J. Wu
School of Pharmaceutical Science
Jiangnan University
Design and in vitro reconstruction of a tri-enzymatic cascade
for D-Phe synthesis
Wuxi 214122, China
E-mail: wujing@jiangnan.edu.cn
C. Lu, W. Song, J, Liu, X. Chen, Prof. L. Liu
State Key Laboratory of Food Science and Technology
Jiangnan University
Wuxi 214122, China
S. Zhang
Tianrui Chemical Co.,Ltd
Department of Chemistry
Quzhou 324400, China
As illustrated in Figure 1a, D-Phe synthesis from L-phenylalanine
(L-Phe) occurs in two steps: first, L-Phe undergoes oxidative
deamination by L-amino acid deaminase (LAAD, EC 1.4.3.2) to
produce the prochiral intermediate PPA; second, PPA undergoes
reductive amination by DAPDH to generate D-Phe, with GDH (EC
1.1.1.47) coupling to regenerate NADPH. To reconstruct the
cascade pathway in vitro, Proteus vulgaris PvDAPDH (Table S1),
P. mirabilis PmLAAD variant, and Bacillus megaterium BmGDH[17]
were selected based on their specific enzymatic activities. The
[b]
[c]
Supporting information for this article is given via a link at the end of
the document
1
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
three selected genes were amplified, overexpressed, and purified
(Figure S1, Table S2). To assess the feasibility of the proposed in
vitro reconstruction system, the three enzymes were combined in
an equimolar ratio and reacted with 5 g L-1 L-Phe, 0.5 mM NADP+,
1 g L-1 NH4Cl, and 7 g L-1 glucose for 4 h, after which 1.2 ± 0.1 g
L-1 D-Phe was detected (Figure S3). The identity of the final
product was confirmed using mass spectrometry and NMR
analysis (Figure S5-7). This result demonstrated that the
designed cascade composed of PvDAPDH, PmLAAD, and
BmGDH successfully converted L-Phe to D-Phe.
of the reductive amination reaction catalyzed by PvDAPDH and
BmGDH. A comparison of the activities of PmLAAD (40.5 ± 1.0 U
g-1), PvDAPDH (8.7 ± 1.3 U g-1), and BmGDH (561.2 ± 6.8 U g-1)
revealed a 4.6:1:62 ratio, pointing to PvDAPDH as the bottleneck
in this pathway.
The effect of altering the BmGDH:PvDAPDH activity ratio (from
0.1:1 to 3:1) was investigated with 20 g L-1 PPA and DAPDH
activity fixed at 4 U mL-1. As illustrated in Figure 1b, when the
BmGDH:PvDAPDH activity ratio reached 0.3:1, the D-Phe titer
stabilized at 18.5 ± 0.4 g L-1, with 92.7% conversion rate. Similarly,
when the PmLAAD:PvDAPDH activity ratio was set to 0.6–1.0:1,
the D-Phe titer was 16.5 ± 0.4 g L-1, with 82.7% conversion rate
Figure 2. The construction and evaluation of E. coli 01. (a) Construction of co-
expressed strain E. coli 01 with the first T7 promotor followed by PmLAAD, and
the second T7 promotor followed by PvDAPDH and BmGDH. (b) The
performance of E. coli 01 with L-Phe from 5–25 g L-1, E. coli 01 whole-cell
catalysts 20 g L-1.
(Figure
1c).
Therefore,
the
optimum
PmLAAD:PvDAPDH:BmGDH activity ratio was set to 0.6–
1.0:1:0.3.
Table 1. Activity of PvDAPDH toward other α-keto acids
Substrates
Specific activity (U mg-1)
0.08±0.02
-
ee[a]
99[b]
99[b]
98[b]
99[b]
98[c]
97[c]
98[c]
phenylpyruvic acid
phenylglyoxylic acid
2-oxo-4-phenylbutyric acid
4-hydroxyphenylpyruvic acid
2-oxobutyric acid
0.15±0.02
0.05±0.02
0.03±0.01
0.03±0.01
0.02±0.01
α-ketovaleric acid
4-methyl-2-oxopentanoic acid
[a] The conversions for ee determination were set in 1 mL biotransformation
systems containing purified enzyme PvDAPDH 1 U L-1, BmGDH 1 U L-1, α-keto
acids 5 g L-1, NH4Cl 5 g L-1, glucose 7 g L-1 in Tris-HCl buffer (100 mM, pH 9.0),
30°C for 24 h. [b] The D/L-products were detected by the same analysis
method for D/L-Phe. [c] The D/L-products were detected after FDAA-
Figure 1. D-Phe biosynthesis pathway starting from L-Phe; (a) Designed D-Phe
biosynthesis pathway by a tri-enzyme system containing LAAD, DAPDH and
GDH; (b) Effect of different activity ratio (0.1:1–3:1) of GDH/DAPDH on D-Phe
production, with PPA 20 g L-1, NH4Cl 20 g L-1, glucose 28.0 g L-1 in Tris-HCl
buffer (100 mM, pH 9.0, containing NADP+ 5 mM). (c) Effect of different activity
ratio (0.2:1–3:1) of LAAD/DAPDH on D-Phe production, with L-Phe 20 g L-1,
NH4Cl 4 g L-1, glucose 28.0 g L-1, in Tris-HCl buffer (100 mM, pH 9.0, containing
NADP+ 5 mM, FAD 5 mM and enough E. coli membranes).
derivatization[13]
.
Engineering of PvDAPDH for high catalytic efficiency toward
PPA
In vivo reconstruction of the D-Phe biosynthesis pathway
As shown in Table 1, a study of PvDAPDH substrate specificity
revealed low activity toward keto acids including PPA, but high
stereospecificity (97–99% ee). To improve the catalytic efficiency
of PvDAPDH, a structural model of PvDAPDH was built using
StDAPDH (PDB ID: 3WBF)[18] as template. Docking the cofactor
NADPH and imine intermediate 2-imino-3-phenylpropanoic acid
in the cofactor-free structure yielded outputs (Figure 3a) whose
hydrogen bond interactions in the D-center were almost identical
To construct a highly efficient conversion system for industrial
application, three enzymes were designed to co-express in one
host cell. Then, the genes PmLAAD, PvDAPDH, and BmGDH
were inserted into plasmid pCDFDuet-1 (Figure 2a) and
transformed into Escherichia coli BL21(DE3), resulting in strain E.
coli 01. Expression of the three enzymes was verified using SDS-
PAGE (Figure S1). The conversion performance of E. coli 01 was
investigated with 20 g L-1 wet cells at 30°C (Figure 2c) and 5–25
g L-1 L-Phe. When the L-Phe concentration was increased from 5
g L-1 to 15 g L-1, D-Phe titer increased from 4.2 ± 0.2 g L-1 to 7.2 ±
0.1 g L-1, and the conversation rate decreased from 84.4% to
48.0%. Concurrently, the intermediate PPA increased from 0.5 ±
0.2 g L-1 to 16.2 ± 0.1 g L-1, and it was detected in the
transformation broth (Figure S4). This result pointed to inhibition
to those with the natural meso-DAP substrate (Figure 3b)[18-19]
However, a hydride transfer distance of 4.1 Å (3.6 Å reported[19]
.
)
was unfavorable for catalysis. When investigating the L-center,
T171 and R181 were found to sterically clash with the benzene
ring, thus affecting the binding of bulky substrates. Residue F146
formed a pi-pi interaction with the benzene ring, further pulling the
2
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
imine center from NADPH and hindering catalysis (Figure 3c). We
hypothesized that the poor catalysis toward PPA was determined
by tensions on the benzene ring side-chain. Therefore, a “tension
release” strategy was proposed, whereby creation of enough
space allowed for the binding of bulky substrates and reducing
the pulling effect from the L-center facilitated hydride transfer.
As listed in Table 2, an increase in the kcat and KM values of
the five variants led to a concomitant increase in their kcat /KM and
overall catalytic efficiency. The highest kcat/KM (2.43 s-1 mM)
belonged to PvDAPDHW121A/R181S/H227I, which was chosen for
further studies on large α-keto acids, such as phenylbutanoic acid,
4-hydroxyphenylpyruvic acid, and 3-indolylpyruvic acid. As
illustrated in Figure 4b, these three keto acids could be converted
into the corresponding D-amino acids, with the highest titer of D-
homophenylalanine being 38.5 g L-1 and corresponding to a
96.2% conversion rate.
Figure 4. The catalytic efficiency of PvDAPDH variants toward keto-acids. (a)
Directed evolution of the parent PvDAPDHWT for reductive amination of PPA;
(b) The catalytic efficiency of the PvDAPDHW121A/R181S/H227I toward bulky 2-keto
acids. The conversions were performed in 2 mL Tris-HCl buffer (100 mM, pH
9.0, containing 0.5 mM NADP+, 40 g L-1 NH4Cl and 56 g L-1 glucose) with 10 g
L-1 PvDAPDH and BmGDH wet whole-cell biocatalysts, 40 g L-1 keto acids, 30°C
for 24 h.
Figure 3. Docking analysis of PvDAPDH. (a) The docking pose in D-center with
cofactor NADPH and imine intermediate 2-imino-3-phenylpropanoic acid
docked in generated PvDAPDH model; (b) The docking pose of nature substrate
meso-DAP in D-center; (c) The interaction of phenyl side-chain in imine
intermediate; (d) The selected 5 residues for mutation studies.
Accordingly, five residues in a radius of 5 Å around imine
intermediate were selected in the L-center; they included W121,
F146, T171, R181, and H227 (Figure 3d). The five selected
residues were engineered by semi-saturated mutagenesis using
NBT, which contains small residues, and four single variants were
obtained (Figure 4a). These included W121A, T171L, R181S, and
H227I, whose activities were improved by 15-fold, 8-fold, 9-fold,
and 32-fold, respectively. The best single variant, PvDAPDHH227I
(2.56 U mg-1), raised the D-Phe titer to 25.4 g L-1. Random
recombination of the four single variants yielded a tri-variant,
PvDAPDHW121A/R181S/H227I, whose activity of 6.86 U mg-1 was 85-
fold higher than in PvDAPDHWT, and it produced 37.8 g L-1 D-Phe
(Figure 4a, Figure S8).
Table 2. Kinetic parameters of PvDAPDHWT and its variants toward
phenylpyruvic acid
kcat/KM
Catalyst
KM (mM)
kcat (s-1)
(s-1 mM-1)
0.038
0.35
PvDAPDHWT
PvDAPDHW121A
4.95±1.2
6.19±1.3
6.54±1.9
5.46±1.0
8.3±1.7
0.19±0.01
2.19±0.22
2.35±0.13
3.73±0.15
8.54±0.52
24.8±0.68
Figure 5. The changes in the binding pocket and interactions of PvDAPDH. (a)
The pocket shape of WT; (b) The changed pocket shape and increased volume
of variant PvDAPDHW121A/R181S/H227I; (c) Conformation comparison of imine
intermediate side-chain in WT(magenta) and in variant PvDAPDHW121A/R181S/H227I
(orange); (d) The interaction of imine intermediate and hydride transfer distance
PvDAPDHT171L
0.35
in variant PvDAPDHW121A/R181S/H227I
.
PvDAPDHR181S
0.60
The catalytic mechanism of the variant DAPDHW121A/R181S/H227I
was analyzed from the point of steric hindrance and interactions
in the L-center. Spatially, the pocket changed from an elongated
cavity to an enlarged one (Figure 5a, 5b), resulting in a 244-Å3
PvDAPDHH227I
1.04
PvDAPDHW121A/R181S/H227I
10.2±2.1
2.43
3
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
increase in pocket volume and the possibility to accommodate the
benzene ring. As a result, instead of being projected toward the
outside as in the wild-type, the side-chain of the mutant was
rotated in such way that it became buried deeply inside the pocket
(Figure 5c). Accordingly, the pi-pi interaction of the benzene ring
with F146 was replaced by a weaker pi-alkyl interaction with
W121A and a pi-sigma interaction with H227I (Figure 5d). These
changes allowed the side-chain to release tension and meanwhile
shortened the hydride transfer distance by 0.2 Å (from 4.1 Å to
3.9 Å), allowing the imine intermediate easier access to NADPH
(Figure 5d). The disappeared pi-pi interaction in the L-center
might account for the increased KM (from 4.95 mM to 10.2 mM),
which was in line with the calculated increase in binding energy
from -7.5 kcal mol-1 to -7.1 kcal mol-1. The enlarged binding pocket
and reduced hydride transfer distance greatly elevated the kcat (by
130-fold) and kcat/KM (64-fold). In practice, the reshaped pocket
space and reduced hydride transfer distance were responsible for
corresponding to an enzymatic activity ratio of 0.35:1:4.4;
therefore, PmLAAD acted as the limiting step. To overcome this
bottleneck, PmLAAD gene copy number was increased to two (E.
coli 03) and three (E. coli 04) copies. PmLAAD activity increased
to 58.4
±
1.3
U
g-1 in E. coli 03, bringing the
PmLAAD:PvDAPDH:BmGDH activity ratio to 0.57:1:3.1 (Figure
6b), which closely resembled the in vitro reconstruction result.
Whole-cell catalysis in E. coli 03 produced, 38.7 g L-1 D-Phe from
40 g L-1 L-Phe, corresponding to a 96.8% conversion rate, and
without any L-Phe and PPA in the transformation broth (Figure
5d). Interestingly, D-Phe titer and conversion were lower in E. coli
04 than in E. coli 03. It seems that duplication of LAAD gene
improved its activity meanwhile it also affected the expression of
DAPDH and GDH, making DAPDH insufficient for further
conversion (Figure S2).
The effect of substrate feeding, buffer pH (7.5 to 9.5), and
temperature (20 to 37°C) on D-Phe titer were investigated at 100-
mL scale with 60 g L-1 L-Phe. As shown in Figure 7a, use of 20 g
L-1 wet E. coli 03 for whole-cell catalysis yielded 54.2 ± 1.3 g L-1
D-Phe, amounting to a conversion rate of 90.3% when L-Phe was
fed at 10 g L-1 every 2 h together with 2 g L-1 NH4Cl and 14 g L-1
glucose. Increasing the pH of the Tris-HCl buffer to 8.5 pushed
the D-Phe titer to 56.1 ± 1.2 g L-1 (Figure 7b). Finally, as shown in
Figure 7c, D-Phe titer (57.4 ± 1.4 g L-1) and conversion yield
(95.7%) were highest at a temperature of 25°C. Thus, optimum
transformation conditions (20 g L-1 wet E. coli 03, pH 8.5, 100 mM
Tris-HCl, 25°C), resulted in 57.8 ± 1.1 g L-1 D-Phe from 60 g L-1
L-Phe during 30 h in a 3-L fermentor, corresponding to 96.3%
conversion rate and >99% ee.
improving the catalytic efficiency of PvDAPDHW121A/R181S/H227I
.
One-pot production of D-Phe through
cascade
a tri-enzymatic
PvDAPDH was replaced by PvDAPDHW121A/R181S/H227I in E. coli 01.
The resulting strain, E. coli 02, generated 35.6 g L-1 D-Phe at 40
g L-1 L-Phe and 20 g L-1 wet cells (Figure 6a). However, 3.8 g L-1
L-Phe remained in the transformation broth and no PPA
intermediate was detected. This could be explained by PmLAAD,
PvDAPDH, and BmGDH in E. coli 02 having an activity of 42.3 ±
1.2 U g-1, 120.8 ± 1.9 U g-1, and 532.3 ± 4.5 U g-1, respectively,
Figure 6. Optimization of PmLAAD in co-expressed strain by different copies. (a) D-Phe production by strain E. coli 02 with L-Phe 40 g L-1 and wet whole-cell
biocatalysts 20 g L-1 in 10 mL Tris-HCl buffer (100 mM, pH 9.0, containing 0.5 mM NADP+, 8 g L-1 NH4Cl and 56 g L-1 glucose), 30°C for 30 h; (b) The assay
of activity in co-expressed strain with different copies of PmLAAD. (c) Effect of different PmLAAD copies on D-Phe production with L-Phe 40 g L-1 and wet
whole-cell biocatalysts 20 g L-1, in 10 mL Tris-HCl buffer (100 mM, pH 9.0, containing 0.5 mM NADP+, 8 g L-1 NH4Cl and 56 g L-1 glucose), 30°C for 30 h.
Figure 7. Optimization of conversion process by strain E. coli 03. (a) Effect of feed rate of L-Phe on D-Phe concentration; (b) Effect of conversion pH on D-Phe
concentration; (c) Effect of conversion temperature on D-Phe concentration. The conversion reactions were performed in a 100 mL volume of 100 mM Tris-HCl
buffer, containing NADP+ 0.5 mM, with wet whole-cell biocatalysts, 20 g L-1, NH4Cl 12 g L-1 and glucose 84 g L-1, for 30 h.
4
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
DAPDH is an excellent enzyme to convert bulky aromatic α-
keto acids to the corresponding D-amino acids. The catalytic
activity and substrate specificity of DAPDH toward bulky aromatic
α-keto acids has been a potential bottleneck in industrial
applications. In this study, the low activity of PvDAPDH toward
PPA was the limiting step for efficient D-Phe production. A
“tension release” strategy in the L-center was proposed and led
to PvDAPDHW121A/R181S/H227I, an improved variant with 85-fold
increased specific activity and 64-fold increased kcat/KM. During
the past decade, various rational protein engineering strategies
based on the analysis of structure–function relationships or
catalytic mechanisms have been used to increase the catalytic
Discussion
The aim of the present study was to develop a highly efficient
method for the production of D-Phe. To this end, a tri-enzymatic
cascade, which included PmLAAD, PvDAPDH, BmGDH, and L-
Phe as substrate, was designed and reconstructed in vivo. The
efficiency of this cascade was further improved by increasing the
catalytic activity of PvDAPDH (6.86 U mg-1) up to 85-fold toward
PPA, a limiting step in the cascade. Use of a “tension release”
strategy
led
to
the
best
PvDAPDH
variant,
PvDAPDHW121A/R181S/H227I, which achieved one-pot conversion of
L-Phe to 57.8 g L-1 D-Phe, with 99% ee in 30 h at a 3-L scale.
These values are suitable for industrial-scale production of D-Phe
from a cheap amino acid source such as L-Phe.
efficiency of DAPDH. A double-site variant, StDAPDHW121L/H227I
,
was achieved through structure-guided pocket engineering.
Compared with wild-type StDAPDH, the activity of
StDAPDHW121L/H227I toward 2-oxo-4-phenylbutyric acid was
increased by 34-fold[13]. Substrate specificity of DAPDH could also
be broadened. For example, DAPDH from Clostridium tetani E88
was mutated to CtDAPDHQ154L/T173I/R199M/P248S/H249N/N276S, which
The enzymatic cascade described here represents a chiral
reset process with PPA as intermediate. It benefits from the low
price of substrate (L-Phe) and elevated enantioselectivity. Various
enzymatic methods have been developed to synthesize D-Phe:
(1) In the hydantoinase process, racemic 5-benzylhydantoin acts
as substrate (Figure 8a) for a cascade that involves hydantoin
racemase, D-hydantoinase, and N-carbamoyl-D-amino acid
amidohydrolase. D-Phe is produced at high yield (45.9 g L-1, 98%
molar yield) and 99% enantioselectivity in 48 h[3b]; however, the
low solubility of substrate represents a rate-limiting step and leads
to prolonged conversion times[20]. (2) In transamination processes,
PPA and D-glutamate act as co-substrates (Figure 8b), with PPA
being transaminated to D-Phe via D-transaminase, while D-
glutamate is converted to α-ketoglutarate. During this catalytic
process, NAD+ and formate are fed to the transformation broth to
provide enough NADH. Glutamate dehydrogenase and glutamate
racemase help recycle α-ketoglutarate back to D-glutamate, while
FDH restores the reducing power (NADH). This process has
yielded 58 g L-1 D-Phe with 100% ee in 35 h[4b]. (3) Asymmetric
resolution processes utilize D/L-Phe as substrate and
phenylalanine ammonia-lyase for the resolution from racemate[21]
(Figure 8c). Although enantioselectivity can be quite high (>99%),
the conversion rate of this process is limited to 50%, making
caused the activity toward PPA to increase from 0 to 0.11 U mg-
1[22]
.
Here, analysis of steric hindrance and unfavorable
interactions, whereby increased molecular tension and decreased
the catalytic activity, identified five candidate residues for semi-
saturation
mutagenesis.
The
best
resulting
variant,
PvDAPDHW121A/R181S/H227I, had an expanded binding pocket,
allowing the PPA side-chain to be rotated and buried further in the
cavity of PvDAPDH, thus releasing the tension and shortened the
hydride transfer distance.
By introducing variant PvDAPDHW121A/R181S/H227 into E. coli 03
and then optimizing transformation conditions, 57.8 g L-1 D-Phe
was achieved with 96.3% conversion and >99% ee via a one-pot
transformation. A similar study combining PmLAAD whole-cell
with purified StDAPDHH227V and Burkholderia stabilis BsFDH
yielded 13.2 g L-1 D-Phe in 6 h, with 100% conversion rate
and >99% ee[15]. And more recently, the three enzymes were
assembled in one cell and the titer of D-Phe was improved to 24.7
g L-1 in 24 h by 75 g L-1 whole-cell catalyst and 15 mM NADP+[23]
.
downstream purification more difficult[5b]
.
A quick cost analysis of our final process from materials and
catalyst showed that the cost for D-Phe production is about
$11450 per ton. By contrast, the factory price for D-Phe is as high
as $30500 per ton (Table S4). Therefore, the cascade pathway
developed in this study greatly improves previous D-Phe titers
and provides an attractive strategy for the industrial production of
D-Phe.
(a) Hydantoinase process
O
O
O
O
OH
OH
HRase
DHHase
DCHase
NH
NH
NH2
HN
NH2
HN
HN
recemization
O
O
O
D-Phe
L-5-benzylhydantoin
D-5-benzylhydantoin
N-carbamoyl-D-phenylalanine
(b) Transamination process
O
O
OH
DAAT
OH
D-Glu
NH2
O
+
+
PPA
D-Phe
a-KG
GluDH
L-Glu
NH4
HCOO-
NAD+
Glu rasemase
Experimental Section
NADH
FDH
CO2
Materials
(c) Asymmetric resolution process
O
O
O
OH
+
The expression plasmid pET-28a(+), pCDFDuet-1 and the host
strain E. coli BL21(DE3) were purchased from Novagen (Madison,
WI, USA). PmLAAD variant (patent application number:
202011588760.0) and BmGDH (UniProtKB: A7XZE6_BACME)
were reserved by our Lab. PvDAPDH (NCBI: WP_075674287.1)
was cloned from Proteus vulgaris. Other DAPDH genes were
cloned from corresponding strains (Table S1). The restriction
OH
PAL
OH
NH2
NH2
NH3
D/L-Phe
trans-cinnamic acid (to be removed)
D-Phe (remained)
Figure 8. Three other ways to produce D-Phe. (a) Hydantoinase process; (b)
Transamination process; (c) Asymmetric resolution process.
5
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
enzymes (BamHI, NcoI, EcoRI, HindIII, NotI, NdeI, XhoI, KpnI,
and DpnI), primerSTAR polymerase, T4 Ligase, plasmid miniprep
kit, agarose gel DNA purification kit were supplied by TaKaRa
(Dalian, China) and one step cloning kit was supplied by Vazyme
(Nanjing, China). All other chemicals and solvents were obtained
commercially.
(pH 1.5):acetonitrile (8:2) at a flow rate of 0.2 mL min-1 (25°C) and
detected at 210 nm. Samples for D/L-Phe detection were first
dissolved in 2M NaOH and then centrifuged at 12,000 × g for 20
min. The supernatant was then diluted by HClO4 (pH 1.5) and
filtered using 0.22-μm filter membrane. The concentration of PPA
was also detected by HPLC. The samples for PPA detection were
centrifuged and filtered through 0.22-μm filter membrane.
Samples were seperated by Bio-Rad Aminex HPX-87H column
(300×7.8 mm, 9 μm), eluted with 0.005 M H2SO4 (0.6 mL min-1,
35°C) and detected at 210 nm.
Molecular modeling
The 3D structural models of PvDAPDH and its mutant were
constructed based on X-ray crystal structure of the StDAPDH
from Symbiobacterium thermophilum (PDB ID: 3WBF) by
homology modeling (https://swissmodel.expasy.org/). Cofactor
NADPH/ NADP+ was first docked into PvDAPDH to obtain the
binary complex by Auto Dock Vina，then the imine intermediate
2-imino-3-phenylpropanoic acid or nature substrate meso-
diaminopimelate was further docked into the binding site. The
pocket volume was calculated by binding site detection website
(https://proteins.plus/).
Enzyme expression and purification
PmLAAD, PvDAPDH and BmGDH were individually
overexpressed and purified from E. coli BL21(DE3) with
pET28a(+) plasmid (Table S2). Cells were harvested by
centrifugation at 6,000 × g for 10 min and resuspended in Tris-
HCl buffer. The cell suspensions were ruptured by high pressure
cell crusher at 4°C and centrifuged at 12,000 × g for 20 min. The
recombinant strains were purified by an AKTA pure system (GE
Healthcare Life Science, USA) with a nickel-affinity column (for
PmLAAD and BmGDH) and anion exchange column (for
PvDAPDH).
Construction of the co-expressed strains
Main primers used for constructing co-expressed strains were
summarized in Table S3. PmLAAD gene was firstly inserted into
pCDFDuet-1 after the first T7 promoter using restriction sites
BamHI and HindIII. Gene duplication of PmLAAD (added with
RBS sequence) was further inserted by restriction sites HindIII (2
copies) and NotI (3 copies). Then PvDAPDH gene was inserted
after the second promoter using the restriction sites NdeI and
KpnI, followed by BmGDH gene (added with RBS sequence, by
restriction sites KpnI and XhoI).
Enzyme assay
PmLAAD (EC: 1.4.3.2) activity on L-Phe was assayed by coupling
PPA formation. The reaction system (Tris-HCl buffer, 100 mM, pH
9.0, 30°C) contained 20 g L-1 L-Phe. One unit of PmLAAD activity
(U) was calculated as the amount of enzyme producing 1 µmol of
PPA in 1 min. PvDAPDH (EC: 1.4.1.16) activity on PPA was
assayed by coupling NADPH consumption. The reaction system
contained 50 mM PPA, 200 mM NH4Cl and 2 mM NADPH. One
unit of PvDAPDH activity (U) was calculated as the amount of
enzyme consuming 1 µmol of NADPH in 1 min. BmGDH (EC:
1.1.1.47) activity on glucose was assayed by coupling NADPH
formation. The reaction system contained 50 mM glucose, 2 mM
NADP+. One unit of BmGDH activity (U) was calculated as the
amount of enzyme producing 1 µmol of NADPH in 1 min. The
activity of the recombinant strains expressed single enzyme was
measured using purified enzyme, while the activity of the co-
expressed strains was measured using wet whole-cell catalysts
(for PmLAAD) and the supernatant (for PvDAPDH and BmGDH).
Mutant library construction and screening
The primers used for constructing PvDAPDH and its variants were
summarized in Table S2. PvDAPDH (WP-0756742847.1) from
Proteus vuglaris was inserted into the pET-28a (+) using the
restriction sites NcoI and EcoRI for expression in E. coli BL21
(DE3). Semi-saturation mutagenesis and the recombination of
variants were carried by whole plasmid PCR protocol
(PrimeSTAR) and the PCR product was digested by DpnI and
then transformed into E. coli BL21 (DE3) cells for further
screening and DNA sequencing (Genewiz, China). To high-
efficiently screen the positive variant from 96-well plate, the
supernatants were prepared and used for activity screening by
monitoring the NADPH consumption at 30°C in a reaction system
of Tris-HCl buffer (100 mM, pH 9.0), containing 160 µL crude
enzyme extract, 20 µL substrate mixture (200 mM PPA and 2 M
NH4Cl), and 20 µL of NADPH (10 mM). Variants with 3-fold higher
activity toward PPA than wild-type enzyme were selected.
Determination of kinetic parameters
Kinetic parameters for reductive amination of PPA were
performed at 30°C in Tris-HCl buffer (100 mM, pH 9.0), which
contained NH4Cl 200 mM, NADPH 2 mM, with PPA concentration
varied from 2 to 50 mM. DAPDH was added to start the reaction,
and the absorbance in 340 nm were monitored in the process. All
of the experiments were conducted with three replicates. KM and
kcat were calculated by nonlinear fitting by Origin software.
Analytical methods
D-Phe and L-Phe levels were determined by high-performance
liquid chromatography (HPLC) using a VWD detector with a
CrownPak CR(+) column (4×150 mm, 5 μm), eluted with HClO4
Production of D-Phe from L-Phe
6
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
2016, 28, 495-513.
The strains were harvested by centrifugation at 6,000 × g for 10
min after overexpression. The conversion experiments were
carried out in a 3-L bioreactor with 1-L working volume.
Substrates L-Phe, NH4Cl and glucose were fed to the reaction
system every 2 h. Because the cascade reactions by BmGDH
was pH-decreasing, in a 3-L reactor, 2 M NaOH was used to
control pH automatically. The concentration of D-Phe, L-Phe, and
phenylpyruvic acid was determined using the HPLC method as
described above.
[2]
[3]
a) Y.-P. Xue, C.-H. Cao, Y.-G. Zheng, Chem. Soc. Rev. 2018, 47, 1516-
1561; b) L. Pollegioni, E. Rosini, G. Molla, Int. J. Mol. Sci. 2020, 21, 3206.
a) H. B. Dae, J. J. Song, S.-G. Lee, S. J. Kwon, Y. Asano, M.-H. Sung,
Enzyme Microb. Technol. 2003, 32, 131-139; b) H. Nozaki, Y. Takenaka, I.
Kira, K. Watanabe, K. Yokozeki, J. Mol. Catal. B: Enzym. 2005, 32, 213-
218; c) J. M. Clemente-Jiménez, S. Martínez-Rodríguez, F. Rodríguez-Vico,
F. J. L. Heras-Vázquez, Recent Pat. Biotechnol. 2008, 2, 35-46.
a) A. Galkin, L. Kulakova, H. Yamamoto, K. Tanizawa, H. Tanaka, N. Esaki,
K. Soda, J. Ferment. Bioeng. 1997, 83, 299-300; b) H.-S. Bae, S.-P. Hong,
S.-G. Lee, M.-S. Kwak, N. Esaki, M.-H. Sung, J. Mol. Catal. B: Enzym. 2002,
17, 223-233; c) H.-S. Bae, S.-G. Lee, S.-P. Hong, M.-S. Kwak, N. Esaki, K.
Soda, M.-H. Sung, J. Mol. Catal. B: Enzym. 1999, 6, 241-247; d) H. Zhou,
L. Meng, X. Yin, Y. Liu, G. Xu, J. Wu, M. Wu, L. Yang, Eur. J. Org. Chem.
2019, 38, 6470-6477.
[4]
Isolation Protocols
The D-Phe product in reaction mixture was isolated using a
Dowex 50WX8 cation exchange column. First, the resin was
washed in sequence by NH4OH (2 M, 2×30 ml), HCl (2 M, 2×30
ml) and H2O (4×30 mL). Then, the reaction mixture (centrifuged)
was acidified with 1 M HCl and loaded onto the column. Finally,
the column was washed with HCl (1M, 2×30 mL), H2O (4×30 mL)
and eluted with NH4OH (2 M, 4×30 mL). Fractions containing D-
Phe were combined and lyophilized to remove the water, and then
purified by preparation thin liquid chromatography (PTLC) with
developing solvent: n-BuOH/H2O/HOAc(4:1:1). Silica gel
containing target amino acid was collected and eluted with n-
BuOH/H2O (2:1). After filtration, the organic solvent was
evaporated by nitrogen blowing, and the product was dried
overnight under vacuum. The solid was washed with (EtOH/H2O,
9:1) to afford D-amino acids in high chemical purity.
[5]
a) L. Zhu, L. Zhou, N. Huang, W. Cui, Z. Liu, K. Xiao, Z. Zhou, PloS One
2014, 9, e108586; b) F. Parmeggiani, N. J. Weise, S. T. Ahmed, N. J. Turner,
Chem. Rev. 2018, 118, 73-118.
[6]
[7]
[8]
K. Vedha-Peters, M. Gunawardana, J. D. Rozzell, S. J. Novick, J. Am.
Chem. Soc. 2006, 128, 10923-10929.
H. Misono, H. Togawa, T. Yamamoto, K. Soda, J. Bacteriol. 1979, 137, 22-
27.
a) H. Misono, M. Ogasawara, S. Nagasaki, Agric. Biol. Chem. 1986, 50,
2729-2734; b) H. Misono, H. Togawa, T. Yamamoto, K. Soda, Biochem.
Biophys. Res. Commun. 1976, 72, 89-93.
[9]
X. Gao, Z. Zhang, Y. n. Zhang, Y. Li, H. Zhu, S. Wang, C. Li, Appl. Environ.
Microbiol. 2017, 83, e00476-17.
[10] a) S. G. Reddy, G. Scapin, J. S. Blanchard, Proteins: Struct., Funct., Genet.
1996, 25, 514-516; b) G. Scapin, S. G. Reddy, J. S. Blanchard,
Biochemistry 1996, 35, 13540-13551.
[11] X. Gao, X. Chen, W. Liu, J. Feng, Q. Wu, L. Hua, D. Zhu, Appl. Environ.
Microbiol. 2012, 78, 8595-8600.
[12] a) H. Akita, Y. Nakamichi, T. Morita, A. Matsushika, Microbiologyopen 2020,
e1059; b) H. Akita, Y. Nakamichi, T. Morita, A. Matsushika, Biochim.
Biophys. Acta, Proteins Proteomics 2020, 1868, 140476.
[13] X. Cheng, X. Chen, J. Feng, Q. Wu, D. Zhu, Catal. Sci. Technol. 2018, 8,
4994-5002.
Data availability
All of the mutants described in this manuscript are available from
the corresponding author upon request.
[14] F. Parmeggiani, S. T. Ahmed, M. P. Thompson, N. J. Weise, J. L. Galman,
D. Gahloth, M. S. Dunstan, D. Leys, N. J. Turner, Adv. Synth. Catal. 2016,
358, 3298-3306.
Acknowledgements
[15] D. Zhang, X. Jing, W. Zhang, Y. Nie, Y. Xu, RSC Adv. 2019, 9, 29927-29935.
[16] L. Zhao, W. Liu, X. Chen, M. Wang, J. Feng, Q. Wu, D. Zhu, Chin. J.
Biotechnol. 2015, 31, 1108-1118.
This work was financially supported by the National Natural
Science Foundation of China (grant numbers: 22008089 and
21878126), Provincial Natural Science Foundation of Jiangsu
Province (grant number: BK20200622), the key technologies R &
D Program of Jiangsu Province (grant number: BE2018623), and
the National First-Class Discipline Program of Light Industry
Technology and Engineering (grant number: LITE2018-20).
[17] J. Li, J. Pan, J. Zhang, J.-H. Xu, J. Mol. Catal. B: Enzym. 2014, 105, 11-17.
[18] W. Liu, Z. Li, C.-H. Huang, R.-T. Guo, L. M. Zhao, D. Zhang, X. Chen, Q.
Wu, D. Zhu, Chembiochem 2014, 15, 217-222.
[19] X. Gao, F. Huang, J. Feng, X. Chen, H. Zhang, Z. Wang, Q. Wu, D. Zhu,
Appl. Environ. Microbiol. 2013, 79, 5078-5081.
[20] S.-F. Wang, Y. Huang, F.-Y. Liao, H.-Y. Zeng, G. Xu, Chem. Eng. Des.
Commun. 2019, 45, 164-165.
[21] Y.-Q. Fang, L.-B. Zhu, N. Huang, L. Zhou, Z.-Y. Ding, Y. Liu, Z.-M. Zhou,
Sci. Technol. Food Ind. 2015, 36, 243-246.
Keywords: D-phenylalanine • DAPDH • enzyme cascade •
[22] W. Liu, R.-T. Guo, X. Chen, Z. Li, X. Gao, C.-H. Huang, Q. Wu, J. Feng, D.
Zhu, ChemBioChem 2015, 16, 924-929.
catalytic efficiency • tension release
[23] D.-P. Zhang, X.-R. Jing, L.-J. Wu, A.-W. Fan, Y. Nie, Y. Xu, Microb. Cell Fact.
2021, 20, 11.
[1]
a) M. Friedman, Chem. Biodiversity 2010, 7, 1491-1529; b) J. Liu, J. Liu, L.
Chu, Y. Zhang, H. Xu, D. Kong, Z. Yang, C. Yang, D. Ding, ACS Appl. Mater.
Interfaces 2014, 6, 5558-5565; c) M. Munoz, S. Recio, M. Rosso, M.
Redondo, R. Covenas, J. Physiol. Pharmacol. 2015, 66, 421-426; d) E.
Başaran, A. Karaküçük-Lyidoğan, D. Schols, E. E. Oruç-Emre, Chirality
7
This article is protected by copyright. All rights reserved.

10.1002/cctc.202100237
ChemCatChem
FULL PAPER
Entry for the Table of Contents
FULL PAPER
D-Phe is an important intermediate in food and pharmacy. In this study, we constructed a tri-enzymatic
cascade reaction for the generation of D-Phe from cheap L-Phe, employing the enzymes LAAD, DAPDH
and GDH. Limited by poor catalytic efficiency of DAPDH, PPA intermediate was found accumulated when
producing D-Phe in whole-cell system. We improved its activity towards PPA by protein engineering. By
integrating the best DAPDH variants into the system and further duplication of the LAAD gene, we
approached to the optimal activity ratio among LAAD, DAPDH and GDH and realized high production of
D-Phe.
8
This article is protected by copyright. All rights reserved.