Rational engineering ofAcinetobacter tandoiiglutamate dehydrogenase for asymmetric synthesis ofl-homoalanine through biocatalytic cascades

Article abstract of DOI:10.1039/d1cy00376c

l-Homoalanine, a useful building block for the synthesis of several chiral drugs, is generally synthesized through biocascades using natural amino acids as cheap starting reactants. However, the addition of expensive external cofactors and the low efficiency of leucine dehydrogenases towards the intermediate 2-ketobutyric acid are two major challenges in industrial applications. Herein, a dual cofactor-dependent glutamate dehydrogenase fromAcinetobacter tandoii(AtGluDH) was identified to help make full use of the intracellular pool of cofactors when using whole-cell catalysis. Through reconstruction of the hydrophobic network between the enzyme and the terminal methyl group of the substrate 2-ketobutyric acid, the strict substrate specificity ofAtGluDH towards α-ketoglutarate was successfully changed, and the activity obtained by the most effective mutant (K76L/T180C) was 17.2 times higher than that of the wild-type protein. A three-enzyme co-expression system was successfully constructed in order to help release the mass transfer restriction. Using 1 Ml-threonine, which is close to the solubility limit, we obtained a 99.9% yield ofl-homoalanine in only 3.5 h without adding external coenzymes to the cascade, giving 99.9% ee and a 29.2 g L^?1h^?1space-time yield. Additionally, the activities of the engineeredAtGluDH towards some other hydrophobic amino acids were also improved to 1.1-11.2 fold. Therefore, the engineering design of some dual cofactor-dependent GluDHs could not only eliminate the low catalytic activity of unnatural substrates but also enhance the cofactor utilization efficiency of these enzymes in industrial applications.

Full text of DOI:10.1039/d1cy00376c

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Rational engineering of Acinetobacter tandoii
glutamate dehydrogenase for asymmetric
synthesis of L-homoalanine through biocatalytic
cascades†
Cite this: DOI: 10.1039/d1cy00376c
Liuzhu Wang, ‡ Shiqing Diao, ‡ Yangyang Sun,
Shuiqin Jiang,
Yan Liu,
*
*
Hualei Wang
and Dongzhi Wei
L-Homoalanine, a useful building block for the synthesis of several chiral drugs, is generally synthesized
through biocascades using natural amino acids as cheap starting reactants. However, the addition of
expensive external cofactors and the low efficiency of leucine dehydrogenases towards the intermediate
2-ketobutyric acid are two major challenges in industrial applications. Herein, a dual cofactor-dependent
glutamate dehydrogenase from Acinetobacter tandoii (AtGluDH) was identified to help make full use of the
intracellular pool of cofactors when using whole-cell catalysis. Through reconstruction of the hydrophobic
network between the enzyme and the terminal methyl group of the substrate 2-ketobutyric acid, the strict
substrate specificity of AtGluDH towards α-ketoglutarate was successfully changed, and the activity
obtained by the most effective mutant (K76L/T180C) was 17.2 times higher than that of the wild-type
protein. A three-enzyme co-expression system was successfully constructed in order to help release the
mass transfer restriction. Using 1 M L-threonine, which is close to the solubility limit, we obtained a 99.9%
yield of L-homoalanine in only 3.5 h without adding external coenzymes to the cascade, giving 99.9% ee
and a 29.2 g L−1 h−1 space–time yield. Additionally, the activities of the engineered AtGluDH towards some
other hydrophobic amino acids were also improved to 1.1–11.2 fold. Therefore, the engineering design of
some dual cofactor-dependent GluDHs could not only eliminate the low catalytic activity of unnatural
substrates but also enhance the cofactor utilization efficiency of these enzymes in industrial applications.
Received 2nd March 2021,
Accepted 22nd April 2021
DOI: 10.1039/d1cy00376c
rsc.li/catalysis
L-homoalanine residues, may further facilitate the growth of
demand for L-homoalanine.6
Introduction
Optically pure amino acids can be used as active
pharmaceutical ingredients.1,2 In particular, unnatural amino
acids, such as L-homoalanine, D-alanine, L-norvaline,
D-phenylalanine, D-tryptophan, and L-3,4-dihydroxy-Phe, are
more attractive because they can be artificially endowed with
diverse functional groups and are useful chiral intermediates
in the pharmaceutical industry.3 Among these compounds,
L-homoalanine is an important chiral building block widely
used in the synthesis of chiral pharmaceuticals, such as
levetiracetam, brivaracetam, and ethambutol.4,5 In recent
studies focused on cancer treatment, bivalent antagonists of
inhibitors of apoptosis proteins, which contain two
Biocatalysis may be the best approach to realize an
environmentally friendly process and obtain excellent optical
purity of L-homoalanine with high efficiency.7–9 Among
various biocatalysts, transaminases (TAs) and amino acid
dehydrogenases (AADHs) are two potential enzymes applied
in this field. TA reactions usually have high turnover rates,
broad substrate specificity, no requirements for cofactors and
neutral equilibrium constants.10,11 Park et al. synthesized
L-homoalanine by ω-transaminase using benzylamine as an
amino donor and obtained 91% conversion.12 Silva et al.
studied the asymmetric conversion of 2-ketobutyric acid to
either L- or D-homoalanine by aminotransferase.13 However,
their industrial applications are limited by the equilibrium of
the reaction. Amino acid dehydrogenases, which catalyze the
reductive amination of α-keto acids by consuming cheap
ammonia and the equivalent cofactor NAD(P)H, are the most
State Key Laboratory of Bioreactor Engineering, New World Institute of
Biotechnology, East China University of Science and Technology, Shanghai,
200237, China. E-mail: hlwang@ecust.edu.cn, dzhwei@ecust.edu.cn
† Electronic supplementary information (ESI) available: Supporting figures and
tables. See DOI: 10.1039/d1cy00376c
promising
catalysts
for
effective
synthesis
of
L-homoalanine.14,15 α-Keto acids have been reported as initial
‡ Liuzhu Wang and Shiqing Diao contributed equally to this work.
reactants for the production of L-homoalanine in previous
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
studies. For example, Galkin et al. employed a LeuDH from
Thermoactinomyces intermedius (TiLeuDH) and achieved a
yield of 88% and a concentration of 0.35 M within 12 h (3 g
L−1 h−1 space–time yield), indicating the relatively low activity
of TiLeuDH.16 Additionally, currently reported AADHs
generally exhibited low activity towards substrates, such as
2-ketobutyric acid. Moreover, α-keto acids are typically quite
expensive, and the use of 2-ketobutyric acid as a starting
reactant may be impractical, particularly in industrial
applications for which high substrate concentrations are
required.
the efficiency of coenzyme utilization, and a systematically
effective method has not yet been reported.23–26 As the three
most commonly used AADHs, LeuDHs and PheDHs are
strictly NADH dependent, whereas some glutamate
dehydrogenases (GluDHs) are dependent on dual cofactors
and may serve as good candidates for avoiding the
requirement for additional cofactors in whole-cell catalysis.
Accordingly, in this study, we aimed to develop a highly
efficient approach for reductive amination without the
addition of external cofactors using
a dual cofactor-
dependent GluDH from Acinetobacter tandoii (AtGluDH) to
replace LeuDHs that are traditionally used to catalyze α-keto
acids with hydrophobic side chains. Our results provided
important insights into the use of dual cofactor-dependent
GluDHs for industrial applications.
Based on these concepts, an improved approach through
cascades for the production of L-homoalanine using cheaper
natural amino acids as substrates has been developed. First,
prochiral intermediate 2-ketobutyric acid was prepared from
inexpensive
L-threonine
or
L-methionine
through
dehydratases, and this compound was then reductively
aminated to L-homoalanine by L-AADHs. For example, Tao
et al. coupled a threonine dehydratase (TD) with a LeuDH
from Bacillus cereus (BcLeuDH) in a cascade to reduce the
cost of using α-keto acids and improve the efficiency of the
reaction.17 The concentration of the substrate can be as high
as 1 M, with a yield of 97.3%. However, this process required
a long time (approximately 16 h; 6.37 g L−1 h−1 space–time
yield) because of the mismatch in the activities of TD and
BcLeuDH. The first step mediated by TD is generally
conducted with high efficiency, even if the substrate
concentration is high (up to 1 M), and complete conversion
can be obtained within a short time. Indeed, reductive
amination is still a rate-limiting step in cascades owing to
the low activity of LeuDH. Although many researchers have
attempted to improve the activity of LeuDHs towards
2-ketobutyric acid, the results have not been satisfactory. For
example, Zhou et al. and Xu et al. increased the space–time
yields to 6.86 and 8.59 g L−1 h−1, respectively, through
rational or semi-rational engineering.18,19 And the space–time
yield was still low owing to the unsatisfactory catalytic
efficiency caused by LeuDHs; thus, this approach could not
meet the demands of industrial applications.
In addition to the problem described above, the redox
reaction involved in this pathway requires expensive
cofactors, such as NAD(P)H, to provide reducing equivalents.
Even with an efficient coenzyme regenerating system, a
minimum amount of 0.1 mM NAD(P)H is generally required
for high catalytic efficiency in AADH-mediated reactions,
contributing to the high cost of the reaction. However, in
some other redox reactions, external coenzymes are not
required when using whole cells or crude enzyme solutions
as catalysts, such as those catalyzed by ketoreductases.20–22
To achieve high performance of biocatalysis without
additional NAD(P)H, the turnover efficiency of the cofactors
could be improved by modifying the affinity of cofactors with
enzymes or screening related enzymes with both NADH and
NADPH dependence to make full use of the intracellular pool
of cofactors. The former approach has proved to be effective
in most instances, while it requires much effort to change
Results and discussion
Characterization of the enzymes in the cascade reaction
A two-step cascade reaction using cheap and easily accessible
L-threonine as a starting material was employed for the
synthesis of L-homoalanine (Scheme 1).
A
threonine
dehydratase (EcTD) from Escherichia coli with reported high
activity was used for the dehydration of L-threonine to
produce the intermediate 2-ketobutyric acid. In order to
catalyze the reductive amination of 2-ketobutyric acid
efficiently, we screened dual cofactor-dependent GluDHs with
high activity in the library of our laboratory. Among them,
the GluDH (AtGluDH) from Acinetobacter tandoii was chosen
for further study due to its highest specific activity (Tables S1
and S2†). Also, its melting temperature and half-life of
inactivation at 50 °C is 53.7 °C and 20.4 h, respectively,
indicating its good thermostability (Fig. S1†). Coupled with a
glucose dehydrogenase (BmGDH) from Bacillus megaterium,
which is also a dual cofactor dependent on the regeneration
of cofactors in cells, the reaction could recycle NADH and
NADPH simultaneously, thereby improving the efficiency of
the coenzymes.
We then evaluated the efficiency of the two steps in the
cascade. Notably, the specific activities of EcTD, AtGluDH,
and BmGDH were found to be 734.1 28.7, 57.3 9.1, and
695.7 34.5 U mg−1, respectively. Obviously, the activity of
AtGluDH was much lower than the activities of the other two
enzymes, suggesting that this may be the rate-limiting
enzyme in the cascade. Before starting a cascade reaction, we
first studied the effect of pH on the reaction (Fig. S2†), and it
Scheme
1 Cascade reaction catalyzed by threonine dehydratase,
L-amino acid dehydrogenase and glucose dehydrogenase.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
has been found that pH 9.5 was the optimum pH for the
cascade, so 200 mM NH4Cl/NH4OH buffer (pH 9.5) was used
as the reaction buffer. The effects of the substrate
concentrations on the activity of EcTD and AtGluDH were
then studied. For EcTD, at a starting concentration of 1 M
L-threonine, the residual amount of the substrate was not
detectable after 10 min, demonstrating the high activity of
EcTD in the cascade (Fig. S3†). For AtGluDH, when the
substrate concentration reached 600 mM, the conversion was
93% within 24 h (Fig. S4†). This was much lower than the
99% conversion of natural LeuDHs at a concentration of 1 M,
as has been reported previously, suggesting that the second
step mediated by AtGluDH was the rate-limiting step. This
result was consistent with the strict substrate specificity of
GluDHs, which were used as catalysts for reductive
amination. Because the side chains of the two substrates (α-
ketoglutarate and 2-ketobutyric acid) had different
requirements for the polarity of the active pocket, the
efficiency of AtGluDH was low towards 2-ketobutyric acid. In
order to match the efficiencies of the two steps, AtGluDH
should be further engineered to change its substrate
specificity from the native substrate α-ketoglutarate to α-keto
acids with more hydrophobic side chains based on our
understanding of the structure.
Fig. 1 (A) Illustration of the native substrate α-ketoglutarate binding
position in the active site of AtGluDH. Orange-colored and green-
colored dotted lines represent hydrogen bonds with an electrostatic
character and hydrogen bonds without an electrostatic character,
respectively, contributing to the binding of the α-ketoglutarate. (B) The
difference in the binding position between the native substrate
α-ketoglutarate and the docking substrate 2-ketobutyric acid (2-KB) in
AtGluDH. Residues having interactions with the 2-ketobutyric acid
directly or indirectly are shown as sticks. Purple-colored and green-
colored dotted lines represent hydrophobic interactions and hydrogen
bonds, respectively, contributing to the binding of the 2-ketobutyric
acid.
Construction of an interaction network between AtGluDH
and 2-ketobutyric acid
To elucidate the interaction of AtGluDH with its native substrate
or 2-ketobutyric acid, a homology model of AtGluDH was first
constructed based on the crystal structure of GluDH from
Burkholderia thailandensis (PDB ID: 4XGI, resolution: 2.0 Å) in
complex with its NAD+ cofactor and native substrate
α-ketoglutarate with the highest sequence identity (62%)
between the enzymes. The substrate 2-ketobutyric acid was then
docked into AtGluDH using AutoDock (Fig. S5†).27
AADHs have two primary conformations, i.e., the substrate
binding “open” state and the catalytic “closed” state in the
aqueous phase. To study the interactions between the substrate
and the enzyme, we evaluated the structural alignment between
AtGluDH and GluDH from Clostridium symbiosum (PDB ID:
1AUP and 1BGV) and GluDH from Corynebacterium glutamicum
an obvious deviation at the active site and resulting in
decreased catalytic activity (57.3 U mg−1 for 2-ketobutyric acid,
2185.8 U mg−1 for α-ketoglutarate) (Fig. 1B). The analysis of the
affinity between AtGluDH-WT and the natural substrate
α-ketoglutarate or 2-ketobutyrate showed that AtGluDH has a
very strong affinity for α-ketoglutarate (for 2-ketobutyric acid,
Km = 4.12 mM, kcat = 0.53 s−1; for α-ketoglutarate, Km = 0.07
mM, kcat = 16.71 s−1), indicating that the strict substrate
specificity of wild-type glutamate dehydrogenases limited their
catalytic activity on other unnatural substrates.
To obtain high activity towards 2-ketobutyric acid, a more
hydrophobic environment is crucial, and the interaction
network between 2-ketobutyric acid and the enzyme should
be rebuilt. Among the four conserved polar residues in the
active pocket of AtGluDH, R192, which has a hydrogen bond
with the carbonyl group of the nicotinamide of NADH, was
not selected for further engineering. The other three residues
were then mutated to alanine and valine in order to quickly
analyze the effects of different steric hindrances, polarities
and straight or branched chains on the activities. Finally, six
mutants, i.e., K76A/V, S355A/V, and T180A/V, were generated.
(PDB ID: 5IJZ), which showed
a
catalytic closed
conformation.28–31 Residues K76, S355, R192, and T180 were
conserved and played important roles in substrate binding in
these GluDHs (Fig. S6†). The positively charged K76 and R192
stabilized the negatively charged α-ketoglutarate through
charge–charge interactions and cooperated with residue S355 to
modulate the substrate specificity of GluDHs (Fig. 1A). At the
same time, these three residues together with T180 provided a
hydrophilic environment to accommodate the γ-carboxyl of
α-ketoglutarate in the active pocket.32 Compared with
α-ketoglutarate, 2-ketobutyric acid has one less carboxyl group
at the γ-position and harbors a more hydrophobic side chain.
Therefore, the original hydrophilic active pocket was not
suitable for the stable binding of 2-ketobutyric acid, leading to
Analysis of mutant activities
The reductive amination activities of the mutants and wild-
type AtGluDH toward 2-ketobutyric acid were determined
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
(Fig. S7†). Among the selected six mutants, only mutations at
K76 showed beneficial activities, and the activity of K76V
showed the highest improvement (up to 7.4-fold) compared
with that of the wild-type protein. These results may be
attributed to the observation that hydrophobic residues with
longer side chains (like valine compared with alanine) can
form more efficient interactions with 2-ketobutyric acid,
which has one less carboxyl group on the substrate compared
with α-ketoglutarate. Based on these speculations, residue
K76 was mutated to other hydrophobic amino acids with
different side chain dimensions (G, A, V, L, and I) to study
the effects of different side chain lengths on the substrate.
All mutants showed consistent results, as expected. Notably,
K76L (960.6 U mg−1) showed the highest efficiency (increase
of 16.8-fold) compared with the wild-type protein. The
longest side chain (from the leucine residue) at site 76 was
likely to accommodate the structurally small size of
2-ketobutyric acid, resulting in beneficial interactions.
The mutation of S355 and T180 to more hydrophobic
amino acids did not yield the expected results. The S355A
mutant maintained 80% of the activity of the wild-type
protein, whereas the S355V mutant lost all its activity.28 This
may be because the mutations disrupted the interactions
between residues S355 and R192/V352, the latter two of
which directly or indirectly interacted with the substrate to
stabilize the conformations. Accordingly, no further
modifications were made to this site. Both T180A and T180V
showed decreased catalytic activity, i.e., 67.2% and 61.0% of
the activity of the wild-type protein, respectively. These
results could be explained by the observation that mutating
T180, which is located in the hinge region close to the
solvent, to a hydrophobic amino acid may cause loss of the
inherent stability of the hydration structure. Therefore, we
mutated T180 to other polar residues with small side chains
(Ser and Cys) in order to avoid steric hindrance at the bulky
pyrrolidine ring of P151. The mutants T180S and T180C
showed 11.5- and 14.4-fold enhanced activities, respectively,
suggesting that mutation of T180 was involved in the
reconstruction of the interaction network and facilitated
biocatalysis. To explore the potential synergistic effect, a
double mutant K76L/T180C was constructed; the mutant
showed the highest activity (985.7 U mg−1), which was 17.2
times higher than that of the wild-type enzyme. However, it
was 1.03-fold and 1.19-fold higher than the mutants K76L
and T180C, respectively, and there was no obvious synergy
between these two mutations.
Fig.
2
(A) Docking of 2-ketobutyric acid into AtGluDH-WT. (B)
Docking of 2-ketobutyric acid into AtGluDH-K76L. (C) Docking of
2-ketobutyric acid into AtGluDH-T180C. (D) Docking of 2-ketobutyric
acid into AtGluDH-K76L/T180C. Residues having interactions with the
2-ketobutyric acid directly or indirectly are shown as sticks. Purple-
colored and green-colored dotted lines represent hydrophobic
interactions and hydrogen bonds, respectively, contributing to the
binding of the 2-ketobutyric acid.
hydrophobic network to stabilize the substrate in the catalytic
closed conformation. For mutant T180C, substitution of the
threonine residue with a less polar cysteine reduced the
hydrophilicity of the bottom of the active pocket and altered
the hydrophobic microenvironment of the substrate binding
pocket (Fig. 2C). At the same time, mutant T180C contributed
to the stable conformation of the substrate in the active
pocket by indirect interactions with the substrate through
generating hydrophobic forces with residues A150 and S355.
The binding free energy of 2-ketobutyric acid was
estimated through the molecular mechanics Poisson–
Boltzmann (generalized Born) surface area method, which
could explain the enhancement of activity by the double
mutant K76L/T180C. For the wild-type AtGluDH, the binding
free energy was calculated to be −57.56 10.16 kcal mol−1
this value was decreased to −62.44
6.81 kcal mol−1 in
AtGluDH–K76L/T180C (Table S4†). These findings
demonstrated that the mutant K76L/T180C showed
thermodynamically stronger binding forces with the
substrate compared with the wild-type protein, consistent
with the experimentally higher catalytic activity. Moreover,
the distance (dc) between the reactive carbonyl carbon (C2) of
the substrate and the hydride donating/accepting carbon (C4)
of the nicotinamide of the coenzyme was measured in
;
To elucidate the molecular mechanisms of these
beneficial mutations, 2-ketobutyric acid was docked into the
active site of the AtGluDH-WT (Fig. 2A) and mutants (Fig. 2B–
D) through computational simulation. The mutant K76L
provided a more hydrophobic environment at the active site
and generated interactions with the γ-methyl group of
2-ketobutyric acid directly (Fig. 2B). Moreover, K76L also
formed hydrophobic interactions with A150 and V352, which
were the only two residues that interacted with the side chain
molecular dynamics simulations as
a crucial catalytic
distance, as reported for AADHs (Fig. S8†).33–35 The data
showed that the dacvg of K76L/T180C was 0.6 Å shorter than
that of the wild-type protein. The reduction of the catalytic
of the substrate. These interaction forces formed
a
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
crucial distance was conducive to increasing the reaction
efficiency, increasing the catalytic activity of mutant K76L/
T180C towards 2-ketobutyric acid.
Application of AtGluDH mutants for the synthesis of
L-homoalanine by cascades
To verify the application of the mutants (K76L, T180C, and
K76L/T180C) in the cascade, a 20 mL reaction was carried
out using 1 M L-threonine as a substrate owing to its
solubility limit and a low concentration of NADH (0.1 mM) in
an aqueous system. Three enzymes, including 3 g L−1 EcTD
(31.8 U mg−1-DCW), 10 g L−1 AtGluDH (9.2 U mg−1-DCW), and 3
g L−1 BmGDH (36.9 U mg−1-DCW), were added as lyophilized
cells. Compared with the wild-type protein, all three mutants
showed higher catalytic efficiency (Fig. 3). The double mutant
K76L/T180C exhibited the highest efficiency, providing a
yield of 99.9% within 2 h. The mutants K76L and T180C
showed a 99.9% yield within 2.5 and 3 h, respectively. The
space–time yields of the three mutants varied from 34.4 to
51.1 g L−1 h−1, which are the highest level reported in current
studies. However, a high enzyme loading (16 g L−1) was
applied when feeding the three enzymes separately in the
cascade. Thus, mass transfer in different cells may be
challenging, and downstream separation and purification
may be more complex.
Kinetic parameters and substrate specificity analysis
To explicate the reason for the improved activity in
biocatalysis, the kinetic parameters of purified enzymes
towards 2-ketobutyric acid were determined (Table 1). The Km
values of K76L, T180C, and K76L/T180C were all lower than
that of the wild-type protein, indicating that the enzyme had
increased affinity with the substrate. Moreover, the Km value
of the double mutant K76L/T180C was 2.52 0.21 mM, which
was lower than that of the wild-type protein (4.12
0.82
mM), demonstrating its high affinity than that of the wild-
type protein for 2-ketobutyric acid. The catalytic efficiencies
(kcat/Km) of the mutants K76L, T180C, and K76L/T180C were
7.9-, 7.5-, and 9.9-fold higher than that of the wild-type
protein, respectively, demonstrating enhanced activity toward
2-ketobutyric acid.
Next, to elucidate the substrate spectrum of the
engineered enzyme, AtGluDH and its mutants were
characterized using various α-keto acids with different
structures (Table 2). Compared with the wild-type AtGluDH
enzyme, the mutants K76L, T180C, and K76L/T180C showed
decreased activity when employing α-ketoglutarate, the native
substrate of GluDHs. All mutants showed a strong preference
for unbranched aliphatic nonpolar α-keto acids, particularly
2-oxopentanoic acid, which is similar in structure to
2-ketobutyric acid but contains one more carbon atom in the
skeleton. In our analysis of branched aliphatic substrates, the
mutants K76L and K76L/T180C showed higher activity
towards short-branched α-keto acids, such as S5 (2.3- or 2.5-
fold) and S8 (2.8- or 3.3-fold), than the wild-type protein. This
result may be explained by the efficient anchoring of short-
branched α-keto acids in the tailored active pocket by K76L
with a bulky leucine.
To further improve the efficiency of the reaction, EcTD,
AtGluDH, and BmGDH were expressed simultaneously in a
single cell. Using lyophilized cells with the same activity as
AtGluDH, as described above, the co-expression system
reached a yield of 99.9% L-homoalanine within 100 min
(Fig. 3). The co-expression of the three enzymes greatly
decreased fermentation costs and improved catalytic
efficiencies,
suggesting
applications
in
industrial
manufacturing. To investigate the ability of the dual cofactor-
dependent AtGluDH to utilize intracellular cofactors by whole-
cell catalysis, a three-enzyme co-expression strain was used
without the addition of NAD(P)H. The results showed that a
yield of 99.9% L-homoalanine was achieved after 3.5 h without
the addition of NAD(P)H, and an isolated yield of 78.2% was
obtained. Thus, the economic cost of industrial production
was significantly reduced, and the route of reductive amination
for production of L-homoalanine became more competitive.
For aromatic α-keto acids, both the wild-type protein and
mutants exhibited low activity owing to significant
differences between the structures of aromatic amino acids
and the natural substrate of GluDHs. Thus, by reconstructing
Experimental
Materials
a
new hydrophobic interaction network, the substrate
specificity of AtGluDH was successfully switched from
hydrophilic to hydrophobic amino acids.
The chemicals used in the experiment including
2-ketobutyric acid, L-threonine, L-2-aminobutyric acid, NH4Cl,
and NH3·H2O were purchased from Aladdin Reagents
(Shanghai, China). The restriction enzymes XhoI, BamHI and
NdeI were obtained from TaKaRa (Shanghai, China). The
enzymes used for the experiment including PrimeSTAR, KOD-
One and T4 DNA ligase were also purchased from TaKaRa
(Shanghai, China).
Table
1 Kinetic parameters of AtGluDH and its mutants towards
2-ketobutyric acid
Enzyme
kcat (s−1
)
Km (mM)
kcat/Km (mM−1 s−1
)
WT
K76L
T180C
K76L/T180C
0.53 0.08
3.03 0.11
2.81 0.28
3.24 0.13
4.12 0.82
2.96 0.97
2.88 0.83
2.52 0.21
0.13
1.02
0.98
1.29
Gene cloning and site-directed mutagenesis
The genomic DNAs from Acinetobacter tandoii, Escherichia coli
and Bacillus megaterium were extracted using a Bacteria DNA
Kit from Tiangen (Shanghai, China). We designed
Reaction conditions: 200 mM NH4Cl/NH4OH buffer (pH 9.5), 0.1 mM
NADH, 2-ketobutyric acid (0.1–50 mM), and at 30 °C.
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 2 Specific activity of wild type AtGluDH and mutants K76L and T180C and K76L/T180C towards various α-keto acids
Specific activitya (U mg−1
Wild type
)
S
K76L
T180C
K76L/T180C
1
2
3
4
5
6
7
8
2185.8 151.7
57.3 9.1
159.6 13.2
44.1 2.4
34.6 28.5
43.2 17.7
19.3 12.4
9.1 5.3
947.7 88.5
960.6 23.5
196.5 9.9
495.2 32.6
77.9 11.5
38.7 13.2
14.6 7.3
25.5 8.4
15.6 1.8
3.3 0.2
1271.5 132.5
827.5 21.7
114.6 11.3
212.8 13.3
32.7 8.4
50.4 29.6
21.4 13.6
5.3 3.3
864.5 79.8
985.7 18.1
115.8 5.7
327.4 14.5
88.0 44.9
25.1 12.7
17.9 11.2
29.7 17.6
12.2 1.2
9
10
11
12.7 0.9
2.5 0.3
4.1 0.3
11.1 2.5
2.9 0.5
2.5 0.1
3.2 0.3
4.9 0.2
3.2 0.8
a
Specific activity was determined with purified proteins. Reaction conditions: 200 mM NH4Cl/NH4OH buffer (pH 9.5), 0.1 mM NADH,
appropriate substrate concentrations and at 30 °C.
oligonucleotide primers with restriction sites based on the
gene sequence found in the NCBI. The amplified DNA
fragments were double-digested with the corresponding
restriction endonuclease and inserted into the expression
vector of pET-28a plasmid (Novagen, Shanghai, China). The
plasmids constructed were transformed into E. coli BL21
(DE3) cells. Site-directed mutagenesis of AtGluDH was
conducted using a KOD-One site mutation kit from Toyobo
Co., Ltd. (Japan). The primer comprising required codons for
mutations was designed and listed in Table S3.† After
amplification, DpnI was added to specifically remove the
methylated template DNA chain, and then T4 polynucleotide
kinase and Ligation high in the kit were used to enable the
digested products to self-cyclize. After transformation, a
single colony was picked and inoculated into Luria-Bertani
(LB) broth containing kanamycin (50 μg mL−1), and the DNA
sequencing for the mutations was conducted by Tsingke
(Shanghai, China) after growing for 12 h.
Expression and purification of proteins
The recombinant cells were cultured on an agar plate for 12
hours at 37 °C, and then single colonies were picked into
tubes containing 5 ml of LB medium and 50 μg ml−1
kanamycin. After that, the cells were cultured at 37 °C with
200 rpm for 10–12 hours. Then, they were transferred into
flasks containing 200 ml of LB medium and 50 μg ml−1
kanamycin, and cultured at 37 °C with 200 rpm for about 3.5
hours. When the optical density value (λ = 600 nm, OD600)
reached 0.6 to 0.8, 0.1 mM IPTG was added for induction,
and the cells were incubated for 18 hours at 20 °C with 200
rpm. These recombinant cells were then centrifuged at 8000
rpm for 10 minutes at 4 °C, washed with 0.5% NaCl solution
three times and lyophilized using a vacuum freeze dryer or
resuspended in 20 mM PB buffer (pH 8.0) on ice. If
resuspended, the recombinant cells were disrupted by
Fig. 3 Cascade reaction curves of AtGluDH and mutants using 1 M
L-threonine as
a substrate. Enzyme loading of WT-0.1, K76L-0.1,
T180C-0.1 and K76L/T180C-0.1: 3 g L−1 EcTD, 10 g L−1 AtGluDH and 3
g L−1 BmGDH using lyophilized cells. Enzyme loading of the co-
expression system (K76L/T180C-Co-0.1 and K76L/T180C-Co-0): 12 g
L−1 lyophilized cells. The amount of NADH added into the reaction was
0.1 mM (WT-0.1, K76L-0.1, T180C-0.1, K76L/T180C-0.1 and K76L/
T180C-Co-0.1) and 0 mM (K76L/T180C-Co-0). Error bars represent
the standard deviation of three independent experiments.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
ultrasonication and centrifuged at 8000 rpm for 30 minutes
at 4 °C to remove the unbroken cells and cell debris. The
supernatant was loaded onto a Ni-NTA column, and the 6×
His-tagged protein bound to the Ni-NTA column at a flow rate
of 1 ml min−1, and then was eluted with 20 mM, 50 mM, 100
mM, 250 mM, and 500 mM imidazole eluents at the same
flow rate. The protein purity of the collected fractions was
examined by sodium dodecyl sulfate polyacrylamide gel
electrophoresis (SDS-PAGE). The fractions containing the
target proteins were gathered and dialyzed against 20 mM PB
buffer (pH 8.0) for desalting. The enzyme solution was
concentrated, mixed with 20% (v/v) glycerol and stored at −80
°C for further use.
Kinetic analysis
The kinetic parameters were measured at 30 °C and pH 9.5.
Using purified GluDH and its variants, the enzyme activities
at different substrate concentrations (0.1–50 mM
2-ketobutyric acid or 0.01–10 mM α-ketoglutarate) were
measured at
a fixed NADH concentration (0.1 mM).
Importing the obtained data, the relationship curve between
the initial reaction rate (vi) and the substrate concentrations
was drawn, and the Michealis–Menten equation fitting was
used to get the vmax and Km values. The kcat values were
calculated from vmax and the concentration of the enzymes.
Co-expression of the enzymes
The open reading frame (ORF) of GluDH from pET28a-
GluDH was cloned into the backbone of pET28a-GDH in
front of GDH to create pET28a-GluDH-GDH. The pACYC184
vector had been split by HindIII to obtain a linker vector
backbone, and then the ORF of TD from pET28a-TD was
inserted to create pACYC184-TD. Then pET28a-GluDH-GDH
and pACYC184-TD were co-transformed into E. coli
BL21(DE3) to have a three protein co-expression.
Enzyme assay
A 1 mL scale reaction mixture containing a certain amount of
purified EcTD, L-threonine (200 mM) and 200 mM NH4Cl–
NH4OH (200 mM, pH 9.5) was incubated at 30 °C, and then
quenched with 5 M KOH (100 μL) at default time intervals.
The samples were analyzed by HPLC.
The activities of GluDHs were measured at 30 °C. They
were measured indirectly by monitoring the absorbance at
340 nm (ε = 6220 M−1 cm−1) to obtain the concentration
of consumed NADH during the reductive amination
reaction. The reaction mixture consisted of 2-ketobutyric
acid (50 mM), NADH (0.1 mM), and NH4Cl–NH4OH (200
Analytical methods
Analysis of L-homoalanine and L-threonine: HPLC detection
conditions: column Agilent C18 (5 μm × 4.6 mm × 250 mm),
mobile phase: V (acetonitrile) : V (0.02 mol L−1 Na2HPO4
buffer (pH 7.2)) = 15 : 85, UV detection wavelength: 360 nm,
column temperature: 30 °C, flow rate: 1.0 mL min−1. Sample
derivatization conditions: 100 μL of the testing sample
diluted by a certain multiple was taken and mixed with 100
μL of 0.5 mol L−1 NaHCO3 solution and 50 μL of 1%
2,4-dinitrofluorobenzonitrile (DNFB) solution. The samples
were incubated at 60 °C in the dark for 1 h, cooled to room
temperature after the reaction, and finally mixed with 750 μL
NaH2PO4/Na2HPO4 buffer (0.2 mol L−1, pH 7.0).
Analysis of 2-ketobutyric acid: COSMOSIL packed column
PBr (4.6 ID × 250 mm), mobile phase: pure water containing
0.1% H3PO4, UV detection wavelength: 210 nm, column
temperature: 30 °C, flow rate: 0.8 mL min−1.
Chiral analysis of L-homoalanine: column SCAS
SUMICHIRAL OA-5000 L (5 μm × 4.6 mm × 150 mm), mobile
phase: a liter of water containing 3 mL acetonitrile and 0.5 g
CuSO4, UV detection wavelength: 254 nm, column
temperature: 35 °C, flow rate: 0.8 mL min−1.
mM, pH 9.5) buffer, and
enzyme extract or purified enzyme solution was added to
start the reaction.
a certain amount of crude
For BmGDH, the reaction mixture contained a certain
amount of purified BmGDH, glucose (200 mM), NAD+ (0.1
mM) and 200 mM NH4Cl–NH4OH (200 mM, pH 9.5). The
activity of BmGDH was measured at 30 °C by using a similar
method to that of AtGluDH.
One unit of enzyme activity is defined as the amount of
enzyme required to catalyze the conversion of 1 μmol
substrate per minute. The method of Bradford was used to
measure the concentrations of the protein.
Preparative application of AtGluDHs for the synthesis of
L-homoalanine
The reaction system for producing L-homoalanine from
L-threonine through a cascade contained 1 M L-threonine, 1.2
M glucose, NADH (0.1 mM or 0 mM), EcTD (3 g L−1),
AtGluDH (WT or mutants) (10 g L−1) and BmGDH (3 g L−1)
which were prepared as lyophilized cells or a certain amount
of co-expressed cells and 200 mM NH4Cl/NH4OH buffer (pH
9.5). The final volume of the reaction mixture was 20 mL.
The reaction was conducted at 30 °C, and the pH was
automatically adjusted to 9.5 by titrating with NH3·H2O.
Aliquots (100 μL) of the reaction mixture were taken at
different time intervals and quenched with 100 μL KOH (5
M), and then analyzed by HPLC. The product was separated
and purified using the method described by Yin et al.36 and
analyzed by NMR (Fig. S11†).
Conclusions
In summary, to efficiently and economically synthesize
L-homoalanine,
a
novel AtGluDH with dual cofactor
dependency was identified. By reconstructing an interaction
network between the enzyme and the substrate, the strict
substrate specificity of AtGluDH was altered, and a high
catalytic efficiency towards 2-ketobutyric acid was obtained.
Coupled with an efficient EcTD, the adverse effects on the
catalytic efficiency caused by the rate-limiting step in the
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
cascade were successfully alleviated, and a high substrate
loading of up to 1 M and a high space–time yield of 29.2 g
L−1 h−1 were achieved. When combined with the dual
cofactor-dependent BmGDH, the cascade reaction could be
efficiently carried out without the addition of external
cofactors, suggesting great advantages in industrial
production. Furthermore, increased activities toward some
hydrophobic amino acids were obtained, indicating that the
substrate spectrum of AtGluDH was expanded.
16 A. Galkin, L. Kulakova, T. Yoshimura, K. Soda and N. Esaki,
Appl. Environ. Microbiol., 1997, 63, 4651–4656.
17 R. Tao, J. Yu, F. Zhu and Y. Sheng, Biotechnol. Lett., 2014, 36,
835–841.
18 J. M. Xu, F. Cheng, F. T. Fu, H. F. Hu and Y. G. Zheng, Appl.
Biochem. Biotechnol., 2016, 182, 898–909.
19 J. P. Zhou, Y. L. Wang, J. J. Chen, M. J. Xu and T. W. Yang,
Biotechnol. J., 2018, 14, 1800253.
20 H. M. Ma, L. L. Yang, Y. Ni, J. Zhang and C. X. Li, Adv. Synth.
Catal., 2012, 354, 1765–1772.
Conflicts of interest
21 L. Huang, H. M. Ma, H. L. Yu and J. H. Xu, Adv. Synth.
Catal., 2014, 356, 1943–1948.
There are no conflicts to declare.
22 Z. Y. Yang, H. W. Fu, W. J. Ye, Y. Y. Xie, Q. H. Liu, H. L.
Wang and D. Z. Wei, Catal. Sci. Technol., 2020, 10, 70–78.
23 N. H. Schlieben, K. Niefind, J. Muller, B. Riebel, W. Hummel
and D. Schomburg, J. Mol. Biol., 2005, 349, 801–813.
24 D. Cui, L. Zhang, Z. Yao, X. Liu, J. Lin, Y. A. Yuan and D.
Wei, Biotechnol. J., 2013, 167, 386–392.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (No. 21776084).
25 Z. N. You, Q. Chen, S. C. Shi, M. M. Zheng, J. Pan, X. L.
Qian, C. X. Li and J. H. Xu, ACS Catal., 2018, 9, 466–473.
26 S. C. Shi, Z. N. You, K. Zhou, Q. Chen, J. Pan, X. L. Qian,
J. H. Xu and C. X. Li, Adv. Synth. Catal., 2019, 361,
4661–4668.
Notes and references
1 J. Fortunak, Trop. J. Pharm. Res., 2008, 7, 865–867.
2 M. Hoehne and U. T. Bornscheuer, ChemCatChem, 2010, 1,
42–51.
3 M. A. Blaskovich, J. Med. Chem., 2016, 59, 10807.
4 M. A. Martínez-Granero, A. García-Pérez and F. Montañes,
Neuropsychiatr. Dis. Treat., 2010, 6, 309–316.
5 E. Falk, C. Laura and P. Marisa, Br. J. Clin. Pharmacol.,
2014, 77, 612–617.
6 H. L. Perez, C. Chaudhry, S. L. Emanuel, C. Fanslau and
R. M. Borzilleri, J. Med. Chem., 2015, 58, 1556–1562.
7 R. N. Patel, ACS Catal., 2011, 1, 1056–1074.
8 D. M. Solano, P. Hoyos, M. J. Hernáiz, A. R. Alcántara and
J. M. Sánchez-Montero, Bioresour. Technol., 2012, 115, 196–207.
9 A. S. Bommarius, Annu. Rev. Chem. Biomol. Eng., 2015, 6,
319–345.
10 E. Heuson, F. Charmantray, J. L. Petit, V. D. Berardinis and
T. Gefflaut, Adv. Synth. Catal., 2019, 361, 778–785.
11 P. Eul-Soo, D. Joo-Young and S. Jong-Shik, Org. Biomol.
Chem., 2013, 11, 6929–6933.
12 E. Park, M. Kim and J. S. Shin, Adv. Synth. Catal., 2010, 352,
3391–3398.
27 G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K.
Belew, D. S. Goodsell and A. J. Olson, J. Comput. Chem.,
2009, 30, 2785–2791.
28 P. J. Baker, M. L. Waugh, X. G. Wang, T. J. Stillman, A. P.
Turnbull, P. C. Engel and D. W. Rice, Biochemistry, 1997, 36,
16109–16115.
29 M. I. H. Khan, H. Kim, H. Ashida, T. Ishikawa, H. Shibata
and Y. Sawa, Biosci., Biotechnol., Biochem., 2005, 69,
1802–1805.
30 L. L. Chen, J. L. Wang, Y. Hu and B. J. Qian, J. Mol. Model.,
2013, 19, 1919–1927.
31 H. F. Son, I. K. Kim and K. J. Kim, Biochem. Biophys. Res.
Commun., 2015, 459, 387–392.
32 T. J. Stillman, P. J. Baker, K. L. Britton and D. W. Rice,
J. Mol. Biol., 1993, 234, 1131–1139.
33 J. L. Vanhooke, J. B. Thoden, N. M. Brunhuber, J. S.
Blanchard and H. M. Holden, Biochemistry, 1999, 38,
2326–2339.
13 M. Silva, I. Costa, R. De Souza and U. Bornscheuer,
ChemCatChem, 2019, 11, 407–411.
14 N. Krall, C. F. Da, O. Boutureira and G. J. Bernardes, Nat.
Chem., 2016, 8, 103.
34 N. M. Brunhuber, J. B. Thoden, J. S. Blanchard and J. L.
Vanhooke, Biochemistry, 2000, 39, 9174–9187.
35 P. Prakash, N. S. Punekar and P. Bhaumik, J. Biol. Chem.,
2018, 293, 6241–6258.
15 M. Sharma, J. Mangas-Sanchez, N. J. Turner and G. Grogan,
Adv. Synth. Catal., 2017, 359, 2011–2025.
36 X. Yin, Y. Liu, L. Meng, H. Zhou, J. Wu and L. Yang, Adv.
Synth. Catal., 2019, 361, 803–812.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2021