Production of α-Ketoisocaproate and α-Keto-β-Methylvalerate by Engineered L-Amino Acid Deaminase

Article abstract of DOI:10.1002/cctc.201900259

This study aimed to develop an efficient enzymatic strategy for industrial production of α-ketoisocaproate (α-KIC) and α-keto-β-methylvalerate (α-KMV) from L-leucine and L-isoleucine, respectively. L-amino acid deaminase from Proteus mirabilis (PmLAAD) was heterologously expressed in E. coli BL21(DE3) and modified to increase its catalytic efficiency by engineering the PmLAAD substrate-binding cavity and entrance tunnel. Four essential residues (Q92, M440, T436, and W438) were identified from structural analysis and molecular dynamics simulations. Residue Q92 was mutated to alanine, and the volume of the binding cavity, enzyme activity, and the k_cat/K_m value of mutant PmLAAD ^Q92A increased to 994.2 ?³, 191.36 U mg^?1, and 1.23 mM^?1 min^?1, respectively; consequently, the titer and conversion rate of α-KIC from L-leucine were 107.1 g L^?1 and 98.1 %, respectively. For mutant PmLAAD^T436/W438A, the entrance tunnel, enzyme activity, and the k_cat/K_m value increased to 1.71 ?, 170.12 U mg^?1, and 0.70 mM^?1 min^?1, respectively; consequently, the titer and conversion rate of α-KMV from L-isoleucine were 98.9 g L^?1 and 99.7 %, respectively. Therefore, augmentation of the substrate-binding cavity and entrance tunnel of PmLAAD can facilitate efficient industrial synthesis of α-KIC and α-KMV.

Full text of DOI:10.1002/cctc.201900259

Accepted Article
Title: Production of α-ketoisocaproate and α-keto-β-methylvalerate by
engineered L-amino acid deaminase
Authors: Yuxiang Yuan, Wei Song, Jia Liu, Xiulai Chen, Qiuling Luo,
and Liming Liu
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.201900259
Link to VoR: http://dx.doi.org/10.1002/cctc.201900259
A Journal of
www.chemcatchem.org

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
Production of α-ketoisocaproate and α-keto-β-methylvalerate by
engineered L-amino acid deaminase
Yuxiang Yuan,[a,b] Wei Song,[a,b] Jia Liu,[a,b] Xiulai Chen,[a,b] Qiuling Luo,[a,b] and Liming Liu*[a,b]
to their respective α-keto acids.[3] For example, PPA can be
Abstract: This study aimed to develop an efficient enzymatic strategy
for industrial production of α-ketoisocaproate (α-KIC) and α-keto-β-
methylvalerate (α-KMV) from L-leucine and L-isoleucine, respectively.
L-amino acid deaminase from Proteus mirabilis (PmLAAD) was
heterologously expressed in E. coli BL21(DE3) and modified to
increase its catalytic efficiency by engineering the PmLAAD substrate-
binding cavity and entrance tunnel. Four essential residues (Q92,
M440, T436, and W438) were identified from structural analysis and
molecular dynamics simulations. Residue Q92 was mutated to
alanine, and the volume of the binding cavity, enzyme activity, and the
kcat/Km value of mutant PmLAADQ92A increased to 994.2 Å3, 191.36 U
mg-1, and 1.23 mM-1 min-1, respectively; consequently, the titer and
conversion rate of α-KIC from L-leucine were 107.1 g L-1 and 98.1%,
respectively. For mutant PmLAAD T436/W438A, the entrance tunnel,
enzyme activity, and the kcat/Km value increased to 1.71 Å, 170.12 U
mg-1, and 0.70 mM-1 min-1, respectively; consequently, the titer and
conversion rate of α-KMV from L-isoleucine were 98.9 g L-1 and 99.7%,
respectively. Therefore, augmentation of the substrate-binding cavity
and entrance tunnel of PmLAAD can facilitate efficient industrial
synthesis of α-KIC and α-KMV.
synthesized through hydrolysis and acidification of benzylidene
hydantoin, an intermediate derived from the reaction between
hydantoin and benzaldehyde.[4] However, owing to the high cost
of precursors, multistep chemical reaction, harsh conditions, and
production of toxic intermediate,[5] researchers adopted the (2)
fermentation process: α-keto acids can be produced through
microbial fermentation, e.g., Corynebacterium glutamicum has
been used to synthesize α-KIC[6] and PPA.[7] Although microbial
fermentation is associated with lesser environmental pollution
than chemical synthesis, it requires
a long duration of
fermentation with low productivity.[6] To further decrease the
production cost of α-keto acids, the (3) biocatalytic method with
amino acids as the staring material was developed.[1b]
Biocatalytic technologies to synthesize α-keto acids have been
developed in recent years, using amino acid oxidase (AAO, EC
1.4.3.3),[8] amino acid dehydrogenase (ADH, EC 1.4.1.5),[9]
aminotransferase (AT, EC 2.1.1.X),[10] and L-amino acid
deaminase (L-AAD, EC 1.4.3.2).[1b] AAO is a flavoenzyme
catalyzing the stereospecific oxidative deamination of amino
acids to α-keto acids and produces ammonia and hydrogen
peroxide. For example, PPA production through D-phenylalanine
oxidation by AAO from the porcine kidney has been reported,[11]
with similar results achieved through catalysis of D-methionine to
α-keto-γ-methylthiobutyric acid (α-KMTB) by AAO from
Trigonopsis variabilis CBS 409.[12] Furthermore, recent studies
have focused on recombinant strains expressing enzymes such
as ADH.[13] In the case of ADH, a coenzyme recycling system is
required owing to the dependence on the either NADH or NADPH
to yield reducing equivalents; hence, such reactions are not cost-
effective.[14] A recent study used L-phenylalanine dehydrogenase
derived from Rhodococcus sp. M4 to synthesize PPA.[15] AT
catalyzes the reversible oxidative deamination of amino acids to
α-keto acids and ammonia, the reverse reaction involving the
asymmetric transfer of amino groups to keto acids.[10] In this
process, other keto acids as essential amino acceptors are
necessary; however, because of the low reaction equilibrium
constant of AT, the reversibility of the AT-catalyzed reactions
often decreases the conversion rates.[10] For instance, E. coli
tyrosine aminotransferase (EC 2.6.1.5) was previously used to
synthesize α-ketoglutaric acid (α-KG) from L-glutamate with a
yield of only approximately 50%.[16] Owing to these limitations,
these three enzymes are not suitable for industrial production of
α-keto acids. Hence, numerous studies have focused on enzyme
L-AAD, a flavin adenine dinucleotide (FAD)-dependent amino
acid oxidase, which catalyzes the oxidative deamination of
several L-amino acids to the corresponding α-keto acid.[17] Owing
to the numerous advantages of L-AAD, such as non-requirement
of an extra cofactor, amino acceptor, and non-production of
hydrogen peroxide,[17] it has been used extensively in
synthesizing various α-keto acids, such as PPA,[18] α-KIV,[19] α-
KG,[20] α-KMTB,[21] and α-KIC.[22] To further increase the titer and
Introduction
α-keto acids, organic acids containing a carboxyl group and a
carbonyl group, are widely applied in foods, feeds,
pharmaceuticals, and chemical synthesis.[1] A prominent example
being compound α-keto acid tablets, a drug containing four α-keto
acids, viz., α-ketoisocaproate (α-KIC), α-keto-β-methylvalerate
(α-KMV), α-ketoisovalerate (α-KIV), and phenylpyruvate (PPA),
as the primary components, which potentially reduces the
accumulation of nitrogen waste in patients.[1b] α-keto acids are
synthesized through three different methods, such as (1)
chemical synthesis: in the past few decades, the hydantoin
process has become one of the primary methods for the
preparation of a large amount of α-keto acids, wherein hydantoin
is made to react with carbonyl-containing compounds (such as
isobutyraldehyde, benzaldehyde, etc.) to generate corresponding
hydantoin intermediates,[2] followed by hydrolysis and acidification
[a]
[b]
Y. Yuan, W. Song, J, Liu, X. Chen, Q, Luo, Prof. L. Liu
State Key Laboratory of Food Science and Technology
Jiangnan University
Wuxi 214122, China
E-mail: mingll@jiangnan.edu.cn
Y. Yuan, W. Song, J, Liu, X. Chen, Q, Luo, Prof. L. Liu
Key Laboratory of Industrial Biotechnology, Ministry of Education
Jiangnan University
Wuxi 214122, China
Supporting information for this article is given via a link at the end of
the document
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
conversion rate of α-keto acids, L-AAD activity was increased
through protein engineering strategies,[18, 21, 23] e.g., L-AAD from
Proteus myxofaciens was modified through site-directed
saturation mutagenesis to increase enzymatic efficiency;
consequently, the synthesis rate of α-KIV by the double-mutant
strain increased from 2.01 g L-1 to 8.19 g L-1.[19]
In the present study, a type-Ⅰ L-AAD (PmLAAD) from Proteus
mirabilis was selected owing to its broad substrate specificity and
heterologously expressed in E. coli BL21(DE3). The PmLAAD
substrate-binding cavity and entrance tunnel were then
engineered for to improve catalytic activity toward L-leucine and
L-isoleucine at high substrate concentrations, via site-directed
mutagenesis based on the rational design. Finally, the PmLAAD
mutants were used to synthesize α-KIC and α-KMV, using a 3-L
scale fermenter.
Figure 1. The conversion of PmLAADWT toward the substrate spectrum.
Results
convert 1a, 4a, 7a, and 8a to their corresponding keto-acids. For
amino acids 2a and 3a, the conversion rate was less than 10%.
For aromatic amino acids 5a and 6a, the conversion rate
increased to 10%–30%; however, the conversion rate was
increased to 90% for hydrophobic aliphatic amino acids 9a and
10a. These results indicate that E. coli BL21 (pET28a (+)-
PmLAAD) could efficiently convert L-leucine and L-isoleucine to
their corresponding keto-acids α-KIC and α-KMV.
The effect of E. coli BL21 (pET28a (+)-PmLAAD) concentration
(5-30 g L-1, wet cell weight) on α-KIC production was investigated
with 65 g L-1 L-leucine as substrate (Figure 2a), and the α-KIC titer
peaked at 62.9 g L-1 with 20 g L-1 wet intact cells, with a 97.5%
conversion rate. Furthermore, with 20 g L-1 wet intact cells as
catalysts, the conversion rate decreased from 97.5% to 57.3%,
and the L-leucine content increased from 65 g L-1 to 110 g L-1
(Figure 2b), as only 62.6 g L-1 α-KIC was produced. Similarly, with
L-isoleucine as the substrate to convert to α-KMV by E. coli BL21
(pET28a (+)-PmLAAD), 49.4 g L-1 of α-KMV was obtained from
55 g L-1 L-isoleucine and 20 g L-1 wet intact cells, the conversion
rate approaching 90.5% (Figure 2c) and then decreasing with an
increase in L-isoleucine content (55 to 75 g L-1). consequently, the
α-KMV titer remained at approximately 49.4 g L-1 (Figure 2c).
These results indicate that the catalytic efficiency of E. coli BL21
(pET28a (+)-PmLAAD) is a bottleneck for further increasing the
α-KIC or α-KMV titers.
Production of α-KIC and α-KMV via heterologous expression
of PmLAAD from Proteus mirabilis
As shown in Scheme 1, the amino group of L-amino acids were
oxidized to a carbonyl group; consequently, L-amino acids were
converted to their corresponding α-keto acids. According to the
KEGG database (https://www.genome.jp/kegg/), this reaction can
be catalyzed by L-AAD, which is extensively present in
prokaryotes and eukaryotes.[9] The characteristics of L-AAD from
different microorganisms are enlisted and compared in
Supplementary Tables 1 and 2, type-Ⅰ L-AADs from Proteus
mirabilis (P. mirabilis) exhibited broad substrate specificity toward
aliphatic and aromatic-amino acids, typically hydrophobic amino
acids. Furthermore, type- Ⅰ L-AADs encoding aad from P.
mirabilis was cloned and overexpressed in E. coli BL21(DE3), and
the recombinant strain E. coli BL21 (pET28a (+)-PmLAAD) was
obtained.
The transformation efficiency of E. coli BL21 (pET28a (+)-
PmLAAD) was evaluated for ten different L-amino acids (Scheme
1 and Figure 1). E. coli BL21 (pET28a (+)-PmLAAD) could not
Engineering the substrate-binding cavity and entrance
tunnel of PmLAAD
To increase the catalytic efficiency of E. coli BL21 (pET28a (+)-
PmLAAD), the crystal structure of L-AAD from Proteus
myxofaciens (PDB entry: 5fjm, 93.74% identity) was selected as
a template through a PDB BLAST search, and a theoretical
structure of PmLAAD was constructed using SWISS-MODEL. A
DALI analysis indicated that the overall structural features of
PmLAAD were highly consistent with those of L-AAD from the
template (DALI Z-score of 69.9), with a root mean square
deviation of 0.1 Å (426 to 426 Cα atoms). As illustrated in
Scheme 1. L-amino acid deaminase catalyzing the deamination of L-amino
acids to α-keto acids.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
binding at the active site, and the other residue M440, which was
located proximal to the substrate-binding pocket, toward the
carboxyl group of the substrate, potentially results in steric clash
between its side chain and FAD. In addition, the backbone of L-
leucine is stacked with the FAD isoalloxazine ring, at
approximately 3 Å distance (Figure 3a), thus preventing stable L-
leucine binding. Therefore, we speculated that Q92 and M440 are
key residues affecting the activity of PmLAAD, and hence
residues Q92 and M440 were selected for alanine mutagenesis
to alter the substrate binding cavity, thereby decreasing the
spatial barrier and the steric clash during substrate binding.
As shown by the PmLAAD-L-leucine complex structure, the
broad entrance of the active site enables binding with large
hydrophobic substrates, whereas the substrate-binding pocket is
deep (over 20 Å), and a narrow cavity enclosing the active center
displays steric hindrance during substrate entry (Figure 3b).
Hence, we speculated a substrate entrance tunnel in the narrow
cavity, which potentially regulated substrate entry into the active
site, thereby affecting enzyme activity. We then focused on the
composition of the entrance tunnel to the active site. CAVER 3.01
was used to identify tunnels present in PmLAAD. As shown in
Figure 3c, an 18.88 Å substrate tunnel distance existed in the
narrow cavity; however, an approximately 1.02 Å-wide bottleneck
was observed in this entrance tunnel. Thereafter, six amino acid
residues (S412, P413, T414, T436, V437, and W438) were
identified, which potentially form the bottleneck at the substrate
entrance tunnel (Figure 3d). We assumed that large amino acid
side chains present at the entrance tunnel forms the bottleneck
and may partially block the entrance tunnel of PmLAAD, thus
deterring substrate entry at the catalytic center. To verify whether
these six residues are responsible for the formation of the
entrance tunnel bottleneck, alanine-scanning mutagenesis was
performed at the tunnel.
Alanine scanning is a common method in protein evolution
engineering. Herein, we substituted the aforementioned eight
candidate residues in the wild type enzyme with smaller alanine
residues to enlarge the bottleneck radius of entrance tunnel and
change the binding cavity for efficient substrate access and
binding. The wild type PmLAAD (PmLAADWT) and eight mutants
thus generated (PmLAADQ92A, PmLAADM440A, PmLAADS412A
,
PmLAADP413A, PmLAADT414A, PmLAADT436A, PmLAADV437A, and
PmLAADW438A) were evaluated computationally via molecular
dynamics (MD) simulations to analyze the substrate entrance
tunnel geometry and changes in the binding cavity. Both of
mutants PmLAADQ92A and PmLAADM440A displayed a significant
improvement in the volume of the binding cavity compared to
PmLAADWT (Figure 4). Along with reconfiguration, the bottleneck
radius of the substrate entrance tunnel was significantly increased
in mutants PmLAADT436A and PmLAADW438A (Figure 5a and
Supplementary Table 3), but with no significant changes in the
four other mutants (Supplementary Figure 2). In particular, mutant
PmLAADQ92A showed the largest increase of approximately
59.0% in the volume of the binding cavity compared that of
PmLAADWT. Mutant PmLAADW438A exhibited the largest increase
in the bottleneck radius of nearly 0.51 Å compared with that of
mutant PmLAADWT. Thus, mutants PmLAADQ92A and
Figure 2. (a) Different concentration of intact cells for α-KIC production; (b)
Effect of different L-leucine concentrations on α-KIC production by PmLAADWT
;
(c) Effect of different L-isoleucine concentrations on α-KMV production by
PmLAADWT
.
Supplementary Figure 1, the PmLAAD model comprises a core
domain (with the α+β additional subdomain) and a V-shaped cap
domain.
To determine the key residues of the substrate-binding cavity,
L-leucine as a substrate and FAD as an important cofactor were
docked into the active site of PmLAAD. As illustrated by the
binding mode (Figure 3a), L-leucine binds directly at the binding
cavity with relatively appropriate positioning and the side chain of
Q92 partially occupied the binding sites of L-leucine isopropyl
moiety, thus causing steric hindrance, which blocks substrate
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
Figure 3. (a) The predicted hot spots, Q92 and M440, and highlighted as green sticks; (b) A close-up view of the PmLAAD active site in the complex with the most
favorable docked position of L-leucine (cyan stick) and FAD (yellow stick); (c) The substrate entrance tunnel of PmLAAD calculated using CAVER 3.01, showed
with green spheres; (d) Bottleneck residues of the substrate entrance tunnel (green spheres model) selected for mutagenesis are showed with yellow stick.
PmLAADM440A, while mutants PmLAADT436A and PmLAADW438A
can be expected to increase the volume of the binding cavity and
bottleneck radius of the entrance tunnel, respectively. Therefore,
mutants
PmLAADQ92A
,
PmLAADM440A
,
PmLAADT436A
,
PmLAADW438A were selected and expected as the optimal
mutants to improve the transformation efficiency at a high
substrate load by expanding the volume of the binding cavity and
increasing the bottleneck radius of the entrance tunnel.
Table 1. Results of site-directed mutagenesis (alanine scanning) experiments
Enzyme
Specific activity (U mg-1 protein)
L-leucine
61.73 ± 2.23
191.36 ± 8.32
L-isoleucine
50.12 ± 1.90
55.34 ± 2.78
PmLAADWT
PmLAADQ92A
PmLAADM440A
PmLAADQ92A/M440A
PmLAADT436A
PmLAADW438A
PmLAADT436A/W438A
152.47 ± 5.34
58.35 ± 1.12
65.34 ± 2.65
67.12 ± 1.98
70.03 ± 3.54
48.27 ± 3.64
50.42 ± 2.67
100.45 ± 5.56
134.23 ± 6.12
170.12 ± 4.43
Construction and experimental evaluation of the mutants
Figure 4. Comparison of the substrate binding pocket volumes of PmLAADWT
and mutants PmLAADQ92A and PmLAADM440A computed using Accelrys
Discovery Studio software 4.0 to determine enzyme cavities.
Mutants PmLAADQ92A
,
PmLAADM440A
,
PmLAADQ92A/M440A
,
PmLAADT436A, PmLAADW438A, and PmLAADT436A/W438A were
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
Table 2. Kinetic parameters of the wild type and its mutants
L-leucine
L-isoleucine
kcat (min-1)
3.20 ± 0.43
5.84 ± 0.48
4.62 ± 0.35
4.10 ± 0.25
3.01 ± 0.05
3.22 ± 0.16
3.50 ± 0.32
kcat/Km (mM-1 min-1)
Km (mM)
kcat (min-1)
kcat/Km (mM-1 min-1)
Enzyme
Km (mM)
PmLAADWT
PmLAADQ92A
9.41 ± 0.34
4.72 ± 0.28
5.12 ± 0.45
19.55 ± 1.04
8.14 ± 0.12
9.32 ± 0.56
8.42 ± 0.43
0.34
1.23
0.90
0.21
0.37
0.35
0.41
12.32 ± 0.12
12.02 ± 0.22
12.98 ± 0.58
28.63 ± 0.82
10.16 ± 0.52
10.15 ± 0.28
10.58 ± 0.45
2.46 ± 0.13
2.64 ± 0.05
2.33 ± 0.41
3.15 ± 0.48
3.66 ± 0.26
4.88 ± 0.23
7.41 ± 0.68
0.20
0.22
0.21
0.11
0.36
0.48
0.70
PmLAADM440A
PmLAADQ92A/M440A
PmLAADT436A
PmLAADW438A
PmLAADT436A/W438A
Figure 5. Time evolution of the tunnel bottleneck radius (BR) in 40 ns long MD simulations. (a) The bottleneck radius of PmLAADT436A (yellow) with 1.42 Å and
PmLAADW438A (blue) with 1.53 Å; (b) The bottleneck radius increased from (1.02) Å in PmLAADWT (red) to (1.71) Å in PmLAADT436A/W438A (green). The tunnels were
analyzed using CAVER 3.01
generated via site-directed mutagenesis, and their specific
activities and steady-state kinetic parameters and those of
PmLAADWT for L- leucine and L-isoleucine were detected and
enlisted in Tables 1 and 2, respectively.
increased to 100.45 U mg-1, 134.23 U mg-1, and 170.12 U mg-1,
respectively. However, mutations in the binding pocket
(PmLAADQ92A
,
PmLAADQ92A
,
and PmLAADQ92A/M440A
)
were
unaccompanied by significant changes in the specific activity of
L-isoleucine. The data for the kcat and kcat/Km is summarized in
Table 2. The double mutant (PmLAADT436A/W438A) showed a 3- and
3.5-fold improvement in kcat and kcat/Km, respectively. These
As shown in Table 1, the specific activities of mutants
PmLAADQ92A, PmLAADM440A, PmLAADQ92A/M440A, PmLAADT436A
,
PmLAADW438A, and PmLAADT436A/W438A on L-leucine were 191.36
U mg-1, 152.47 U mg-1, 58.35 U mg-1, 65.34 U mg-1, 67.12 U mg-
1, and 70.03 U mg-1, respectively. Among them, only the specific
activities of mutants PmLAADQ92A, PmLAADM440A increased
significantly in comparison with those of PmLAADWT (61.73 U mg-
1). Hence, the kinetic parameters of mutants PmLAADQ92A and
PmLAADM440A were assessed and enlisted in Table 2. The kcat and
Km values of mutant PmLAADQ92A and PmLAADM440A increased to
182.5% and 144.4%, decreased to 199.4% and 183.8%, resulting
in a 361.8% and 264.7% increase in kcat/Km, respectively,
compared with that of the PmLAADWT. Mutations A92 and A440
are located at the binding cavity, indicating that the increases in
the volume of binding cavity may led to an increase in catalytic
efficiency of enzyme to L-leucine, however, the mutations at the
entrance tunnel did not significantly increase the kcat/Km.
results emulate the increased entrance tunnel
of
PmLAADT436A/W438A with the largest bottleneck radius of 1.71 Å
(Figure 5b and Supplementary Table 3). However, the double
mutant (PmLAADQ92A/M440A) also exhibited a significant reduction
in kcat/Km in comparison with PmLAADWT, concurrent with the
results observed with L-leucine. These results indicate that the
bottleneck radius of the entrance tunnel is the key limiting factor
influencing the entry of L-isoleucine, rather than the broadening
the binding cavity. Furthermore, expression levels of
PmLAADQ92A and PmLAADT436A/W438A mutants were consistent
with the PmLAADWT (Supplementary Figure 3).
We further examined the activity of PmLAADQ92A and
PmLAADT436A/W438A mutants compared with that of previous
substrates 5a and 6a. As shown in Supplementary Table 5,
PmLAADWT displayed very low activity towards 5a and 6a. In
contrast, its two mutants PmLAADQ92A and PmLAADT436A/W438A
with a larger substrate-binding cavity and entrance tunnel,
respectively, displaying improved activity towards bulky
Similarly, the specific activities of mutants PmLAADQ92A
PmLAADM440A, PmLAADQ92A/M440A, PmLAADT436A, PmLAADW438A
,
,
and PmLAADT436A/W438A to L-isoleucine were also determined.
With mutations in the entrance tunnel (PmLAADT436A
PmLAADW438A, and PmLAADT436A/W438A), their specific activities
,
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
Table 3. Production of α-KIC by the wild type and its mutants on 3-L scale
65 (g L-1)
Conversion (%)
110 (g L-1)
Substrate
L-leucine
Enzyme
Titer ( g L-1)
62.7
63.4
63.8
45.5
Titer ( g L-1)
62.8
107.1
93.6
Conversion (%)
pmLAADWT
97.2
98.3
98.9
70.5
57.5
98.1
85.8
39.2
pmLAADQ92A
pmLAADM440A
pmLAADQ92A/M440A
42.8
Table 4. Production of α-KMV by the wild type and its mutants on 3-L scale
55 (g L-1)
Conversion (%)
100 (g L-1)
Substrate
Enzyme
Titer (g L-1)
49.3
Titer (g L-1)
50
74.2
82.5
98.9
Conversion (%)
L-isoleucine
pmLAADWT
90.3
99.8
99.1
99.7
50.4
74.8
83.2
99.7
pmLAADT436A
pmLAADW438A
pmLAADT436A/W438A
54.5
54.5
54.4
substrates. Particularly for PmLAADT436A/W438A
,
wherein the
relative high activity with both L-leucine and L-isoleucine and was
overexpressed in E.coli BL21(DE3), the catalytic efficiency was
further improved through protein engineering. Upon reshaping the
substrate binding cavity for L-leucine in mutant PmLAADQ92A, the
titer and conversion rate of -KIC increased to 107.1 g L-1 and
98.1%, respectively. Similarly, an increase in the average
bottleneck radius of the entrance tunnel in mutant
PmLAADT436A/W438A, the titer, and conversion rate of -KMV from
L-isoleucine increased to 98.9 g L-1 and 99.7%, respectively.
For PmLAAD, the binding sites of the isopropyl moiety of L-
leucine occupied by side chain of residues Q92 was presumed as
a spatial barrier preventing the binding of L-leucine, presenting
severe steric clash between the backbone of L-leucine and the
FAD isoalloxazine moiety (Figure 3a), leading to only 57.5%
conversion of PmLAADWT with 110 g L-1 L-leucine (Figure 2b and
Table 3). With the PmLAADQ92A mutant, the smaller side-chain of
A92 generated a larger binding cavity and increased the volume
of the substrate-binding cavity of mutant PmLAADQ92A by 59.0%,
with the distance between L-leucine and FAD increased from 3 Å
to 3.6 Å, as shown in Supplementary Figure 4a. Consequently,
the substrate bound more efficiently at the cavity with reduced
steric clash between L-leucine backbone amide moiety and the
FAD isoalloxazine moiety compared to the PmLAADWT enzyme.
Consequently, its conformation was significantly altered, and the
kcat/Km value for L-leucine was increased 3.6-fold. As shown in
Supplementary Table 4, the substrate binding energy in the
PmLAADWT was -3.4 kcal mol-1, and that of the mutant
PmLAADQ92A was -4.6 kcal mol-1, consitent with the increased
affinity. The residues located at the binding cavity often exhibited
important effects on substrate binding[24] because these key
residues determine the volume, shape and conformational
flexibility of the substrate binding cavity.[25] Numerous studies
reported that replacement of these key residues can alter the
volume, geometry, and conformational flexibility of the substrate
binding cavity, thereby altering the catalytic profile of the enzyme,
such as substrate specificity and stereoselectivity.[26] For instance,
alcohol dehydrogenase variants engineered by replacing bulky
residues constituting the active site altered the volume of the
substrate binding cavity, thus accommodating a broader range of
substrates.[25] Reconfiguration of the active pocket of amine
dehydrogenases by fine-tuning two hotspots at the distal end of
the substrate binding cavity can significantly expand the scope for
substrate binding.[27] Similarly, conformational dynamics of the
enlarged entrance tunnel exhibited specific activities of 15.03 U
mg-1 and 12.12 U mg-1 toward 5a and 6a compared with the
PmLAADWT (10.91 U mg-1 and 5.12 U mg-1 for 5a and 6a,
respectively).
Synthesis of α-KIC and α-KMV on a 3-L scale
As shown in Table 3, the α-KIC titer of mutants PmLAADQ92A
,
PmLAADM440A, and PmLAADQ92A/M440A, with 65 g L-1 and 110 g L-1
L-leucine, were compared on the 3-L scale. When 65 g L-1 L-
leucine was introduced in the transformation broth, α-KIC levels
and conversion rate were 63.4 g L-1, 63.8 g L-1 and 45.5 g L-1 for
mutants PmLAADQ92A, PmLAADM440A, and PmLAADQ92A/M440
respectively, and the corresponding conversion rate was 98.3%,
,
98.9%. and 70.5%, respectively. Furthermore, the α-KIC titer of
the
PmLAADWT
,
PmLAADQ92A
,
PmLAADM440A
,
and
PmLAADQ92A/M440A at 110 g L-1 L-leucine were 62.8 g L-1 107.1 g
L-1, 93.6 g L-1, and 42.8 g L-1, respectively, with corresponding
conversion rates of PmLAADQ92A and PmLAADM440A mutants
being 40.6% and 28.3% higher than the corresponding value of
the PmLAADWT (with 57.5% conversion), respectively.
Furthermore, mutant PmLAADQ92A displayed a high space-time-
yield (STY) of 214.2 g L-1 d-1.
The oxidative deamination catalyzed by the PmLAADWT
,
mutants PmLAADT436A, PmLAADW438A, and PmLAADT436A/W438A
were investigated at a 3-L scale with 55 g L-1 and 100 g L-1 L-
isoleucine (Table 4). With 55 g L-1 L-isoleucine as the substrate,
the conversion rate of mutants PmLAADT436A, PmLAADW438A, and
PmLAADT436A/W438A was greater than 99%. Upon treatment with
100 g L-1 L-isoleucine, the α-KMV titer, the conversion rate, STY
of the PmLAADWT were 50 g L-1, 50.4%, and 100 g L-1 d-1,
respectively. For the PmLAADT436A/W438A mutant, the α-KMV titer,
the conversion rate, and the STY were 98.9 g L-1, 99.7%, and
197.6 g L-1 d-1, respectively. As expected, increased catalytic
efficiencies were achieved with the most active mutants
(PmLAADQ92A and PmLAADT436A/W438A) at 110 g L-1 L-leucine and
100 g L-1 L-isoleucine, consistent with results from MD simulations.
Discussion
In this study, PmLAAD, derived from P. mirabilis, exhibited
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
substrate-binding site of lipase was engineered, thereby
increasing the high R-enantioselectivity from 8% to 99%.[24]
The substrate entrance tunnel also significantly affect the
catalytic activity, substrate specificity, and stereoselectivity of the
enzyme.[28] In this study, threonine and tryptophan residues in the
entrance tunnel were identified using MD simulations were
responsible for the narrow bottleneck radius; consequently, only
50.4% conversion rate of PmLAADWT at a high L-isoleucine
concentration of 100 g L-1. When residues T436 and W438 were
simultaneously mutated to a smaller non-polar alanine, mutant
PmLAADT436A/W438A exhibited 0.69 Å increases in the bottleneck
radius of the entrance tunnel. Positive mutations toward L-
isoleucine were located proximal to the substrate entrance tunnel
of the structure. As shown in Supplementary Figure 4b, residues
T436 and W438 may constitute the entrance of the tunnel,
comprising the binding site for the L-isoleucine side chain. The
short distances between the L-isoleucine and T436 and W438
were 3.2 Å and 3.0 Å, respectively, resulting in steric hindrance at
the narrowest section of the entrance tunnel. In addition, the
indole side chain of W438 would result in steric clash with the FAD
isoalloxazine moiety (Supplementary Figure 4b). Therefore, both
residues played an important role in L-isoleucine entry and
binding and influenced enzyme activity. As shown in
Supplementary Figure 4c and Table 3, the entrance tunnel of the
mutant PmLAADT436A/W438A was obviously larger than that of the
PmLAADWT, and the distances between the L-isoleucine and
A436 and A438 increased to 3.7 Å and 3.5 Å, respectively,
allowing for substrate L-isoleucine to accept a productive
conformation more easily and for increased kcat, in consistent with
the determined kinetic paramenters. Consequently, the kcat/Km
value, the titer, and the conversion rate of PmLAADT436A/W438A
increased markedly, although the small side chain of alanine led
to weak substrate affinity. Furthermore, binding energies of
mutant PmLAADT436A/W438A and PmLAADWT were -3.4 kcal mol-1
and -3.2 kcal mol-1, respectively, consistent with the
experimentally determined Km values. These results indicate that
engineering the substrate entrance tunnel may be an effective
strategy to increase the catalytic efficiency of the enzyme.[29]
Numerous examples indicate that modifying the size or
physicochemical properties of the substrate entrance tunnel can
significantly increase the catalytic activity of the enzyme.[28, 30] For
example, residues F128 and M145, located in the tunnel of
epoxide hydrolase BmEH, which reportedly restrict product
release, were mutated to alanine. Two mutants, F128A and
M145A, expanded the potential site of product release, displaying
42- and 25-fold greater activity than the wild type enzyme.[31]
Engineering of key residues surrounding the entrance tunnel of
monoamine oxidase MAO-N decreased polarity in the entrance
tunnel, which was more advantageous for the entry and release
of hydrophobic substrates and products, thereby increasing the
catalytic activity of the enzyme.[28]
industrial production. As an alternative, Corynebacterium
glutamicum was genetically engineered to synthesize α-KIC,
yielding a titer of only 9.23 g L-1.[6] Recently, 69.1 g L-1 α-KIC has
been efficiently synthesized at a 50.3% conversion rate by
expressing L-AAD derived from Proteus vulgaris in E. coli BL21,
[22a] followed by an attempt to optimize protein expression levels;
however, α-KIC titer only increased to 86.55 g L-1 within 20 h.[22b]
In addition, although the procedure for synthesizing α-KMV using
amino acid oxidase has been briefly outlined, the yield is very low
(less than 10 g L-1).[32] In this study, two mutants PmLAADQ92A and
PmLAADT436A/W438A of PmLAAD were used to convert L-leucine
and L-isoleucine to α-KIC (107.1 g L-1) and α-KMV (98.9 g L-1) in
the 3-L scale fermenter within 12 h. Compared with the titers
reported in the aforementioned studies, the production of α-KIC
and α-KMV were increased in the present study. In conclusion,
the present results indicate that augmentation of the substrate-
binding cavity and entrance tunnel of PmLAAD can facilitate
effective industrial synthesis of α-KIC and α-KMV.
Experimental Section
Materials
The expression plasmid pET-28a (+) and the host strain E.coli
BL21 (DE3) were purchased from Novegen (Madison, WI). The
restriction enzymes (BamHⅠ, XhoⅠ, DpnⅠ, and AVrⅡ), T4
DNA ligase, primerSTAR polymerase, plasmid miniprep kit, and
agarose gel DNA purification kit were supplied by TaKaRa (Dalian,
China). All other chemicals and solvents were obtained
commercially.
Molecular modeling and MD simulations
The 3D structural models of PmLAAD and its mutants were
constructed based on X-ray crystal structures of the L-AAD from
[34]
Proteus myxofaciens (PDB ID: 5fjm) by homology modeling
.
The 3D structure of L-leucine, L-isoleucine and FAD were
downloaded from the ChemSpider (http://www.chemspider.com/).
The Dali server (http://ekhidna2.biocenter.helsinki.fi/dali/) was
used for a protein structure similarity comparison. All structures
were prepared for analysis by removing ligands and water
molecules. Auto Dock Vina then was used to obtain the staring
structure of the PmLAAD in complex with the substrate L-
leucine.[35] Missing atoms and hydrogen were added to the
enzyme using the GROMACS 4.5.5 simulation package
(http://www.gromacs.org/). The whole system was immersed in
an explicit TIP 3P water box and extended with a thickness of at
least 10 Å from the dissolved atoms in each dimension.[36] Sodium
ions were added to neutralize each system and the protonation
state of residues was set according to pH 7.0. MD simulations
were performed by using the GROMACS 4.5.5 and an AMBER03
force filed. Initially, each system was subjected to a five-round
energy minimization with the steepest descents method. Each
round was relaxed for 1000 steps. After that, each the system was
heated from the 0 to 300 K under an NVT ensemble by Langevin
dynamics. Subsequently, an NPT ensemble was applied for the
α-KIC and α-KMV as vital α-keto acids are extensively used in
feeds and pharmaceuticals.[1b] Studies have attempted to
synthesize them through chemical synthesis,[5] microbial
fermentation,[6] and enzymatic transformation.[22a, 32] Among these,
chemical methods not only need a multistep process but also use
harsh reagents and produce toxic wastes,[5, 33] thus deterring their
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
next 1 ns, followed by 40 ns production MD in 2 fs steps for date
collection. All simulations were performed individually for both the
complexes of PmLAAD and its mutants. Substrate cavity volumes
were calculated in Accelrys Discovery Studio 4.1 by searching for
cavities with no substrate bound. The MD trajectories were further
analyzed to identify relevant substrate binding poses, and
substrate entrance structures that were sampled and calculated
in the simulation using CAVER 3.01. The Min. probe radius was
0.9 Å and shell depth was 4 Å. The initial starting points for
calculation were defined by active site residues. All images of the
structures shown here were generated using PyMol and CAVER.
collected by centrifugation at 14,000 × g for 1 h at 4 ℃. The crude
enzyme solution was used in subsequent analyses.
Activity assay
To measure the enzyme activities of wild type PmLAAD and its
mutants, the reaction mixture containing 30 mM L-amino acids
and a certain enzyme was incubated with 50 mM Tris–HCl buffer
(pH 8.0) in a final volume of 2 mL at 30 ℃ for 15 min. The reaction
was stopped with centrifugation at 12,000 × g for 5 min, samples
were analyzed by HPLC. One unit of enzyme is defined as the
amount of enzyme producing 1 µmol of α-keto acids per minute.
The protein concentration was determined by the Bradford
protocol,[37] with bovine serum albumin as the standard. All
experiments were conducted in triplicate.
Mutants construction
The primers used for gene cloning and PmLAAD plasmid
construction are summarized in Supplementary Table 6.
PmLAAD encoding gene aad (GenBank accession number
U35383) from P. mirabilis was inserted into the pET-28a (+) using
the restriction sites BamHⅠ and XhoⅠ for expression in E.coli
BL21 (DE3). Site-directed mutagenesis was carried by PCR using
mutagenic primers (listed in Supplementary Table 4) and plasmid
pET-28a (+)-PmLAAD as template. DpnⅠ (1 μL of 10 U μL-1) was
added to 25 μL of the PCR reaction mixture and incubated for 3 h
at 37 ℃ to eliminate the template plasmid. The digested product
was transformed into Escherichia coli JM109 cells for
amplification and DNA sequencing until all designed mutants
were obtained. Subsequently, plasmid pET-28a mutants were
extracted and transformed into E.coli BL21 (DE3) for target
protein expression.
Determination of kinetic parameters
The kinetic parameters were determined by measuring the initial
rate of enzymatic reaction at different concentrations of substrate
(2-100 mM) in Tris–HCl buffer (50 mM, pH 8.0) at 30 ℃. The
samples were withdrawn, extracted, and analyzed by HPLC, and
Km and kcat then calculated by nonlinear regression according to
the Michaelis-Menten equation using Origin software. All
experiments were performed in triplicate.
Fed-batch fermentation
Seed cultures were grown overnight in 50 mL of LB medium and
then sub-cultured in 150 mL of LB medium at 37°C for 10 h. Cells
were harvested to inoculate 3 L of TB medium (24 g L-1 yeast
extract, 12 g L-1 peptone, 4 g L-1 glycerol, 2.31 g L-1 KH2PO4, 16.42
g L-1 K2HPO4·3H2O) for fed-batch fermentation in a 7.5-L INFORS
fermenter. The temperature, pH, agitation rate, and aeration rate
were adjusted to 37 °C, 7.0, 600 rpm, and 1.0 vvm, respectively.
Induction of enzyme expression was performed when the culture
reached an OD600 of 3-5, by adding optimized lactose and
changing the cultivation temperature to 30 °C. When the
dissolved oxygen (DO) level increased rapidly (demonstrating
glycerol in the medium was completely exhausted), the system
was supplied with 100 g L-1 yeast extract, 25 g L-1 peptone, and
400 g L-1 glycerol at a rate of 16 mL h-1. The strain was harvested
as a whole cell catalyst by centrifugation at 6,000 x g for 8 minutes.
Analytical methods
The optical density at 600 nm (OD600) was measured using a
spectrophotometer. The α-keto acids concentrations were
measured by HPLC (Agilent 1100 series, Santa Clara, CA, USA)
using a VWD detector. Samples were separated on an Aminex
HPX-87H (7.8 × 300 mm, Bio-Rad Laboratories Inc., Hercules,
CA, USA) column eluted with 0.005 mol L-1 H2SO4 at a constant
flow rate of 0.6 mL min-1 (35 °C) and detected by monitoring
absorbance at 210 nm. The mobile phase was filtered through a
0.22-μm membrane and ultrasonically degassed before use.
Samples were centrifuged at 12,000 × g for 5 min and then the
supernatants were filtered using 0.22-μm filter membrane.
Protein expression and purification
Biocatalytic oxidative deamination reaction on analytical
scale
E.coli BL21 (DE3) cells carrying the recombined plasmid were
cultivated in 2 mL of LB medium containing Kanamycin (100 μg
mL-1) at 37 ℃ and 200 rpm for 12 h. The overnight culture was
inoculated into 100 mL LB medium and grown at 37 ℃ and 200
rpm in a rotary shaker. When the optical density (OD600) of culture
reached 0.6-0.8, 0.4 mM IPTG was added to induce the enzyme
expression at 25 ℃ for an additional 12 h. The cells were
harvested by centrifugation at 6,000 × g for 5 min at 4 ℃ and
resuspended in Tris-HCl buffer (50 mM, pH 8.0). The cells were
lysed by ultrasonication in an ice bath, and the supernatant was
A 1-mL scale biocatalytic reaction mixture containing lyophilized
whole cell catalyst PmLAAD (4 mg), substrate (50 mM), and Tris–
HCl buffer (50 mM, pH 8.0) was incubated at 30 ºC and 200 rpm
for 24 h. Then the reaction mixture was stopped with
centrifugation at 12,000 × g for 5 min. The conversion rate was
measured by HPLC. 20-mL scale biocatalytic reactions were also
conducted to optimize transformation reactions of PmLAAD with
L-leucine and L-isoleucine as the substrates. A 20 mL reaction
mixture contained wet whole cell catalysts PmLAAD (0.1-0.6 g),
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
[20]
[21]
[22]
G. S. Hossain, J. Li, H. D. Shin, L. Liu, M. Wang, G. Du, J. Chen, J.
Biotechnol. 2014, 187, 71-77.
G. S. Hossain, J. Li, H. D. Shin, G. Du, M. Wang, L. Liu, J. Chen,
PLoS One. 2014, 9, e114291.
a) Y. Song, J. Li, H. D. Shin, G. Du, L. Liu, J. Chen, Sci. Rep. 2015,
5, 12641; b) Y. Song, J. Li, H. D. Shin, L. Liu, G. Du, J. Chen, PLoS
One. 2017, 12, e0179229.
L-leucine (1.3-2.2 g) or L-isoleucine (0.3-1.5 g), and Tris–HCl
buffer (50 mM, pH 8.0) was incubated at 30ºC and 200 rpm for 20
h. 500 μL of sample was taken timely after vigorous shaking,
centrifuged at 12,000 × g for 5 min before analysis by HPLC.
[23]
[24]
[25]
[26]
[27]
[28]
L. Liu, G. S. Hossain, H. D. Shin, J. Li, G. Du, J. Chen, J. Biotechnol.
2013, 164, 97-104.
B. Yang, H. Wang, W. Song, X. Chen, J. Liu, Q. Luo, L. Liu, ACS
Catal. 2017, 7, 7593-7599.
Y. Nie, S. Wang, Y. Xu, S. Luo, Y. L. Zhao, R. Xiao, G. T. Montelione,
J. F. Hunt, T. Szyperski, ACS Catal. 2018, 8, 5145-5152.
S. Jemli, D. Ayadi-Zouari, H. B. Hlima, S. Bejar, Crit. Rev.
Biotechnol. 2016, 36, 246-258.
F. F. Chen, G. W. Zheng, L. Liu, H. Li, Q. Chen, F. L. Li, C. X. Li, J.
H. Xu, ACS Catal. 2018, 8, 2622-2628.
G. Y. Li, P. Y. Yao, R. Gong, J. L. Li, P. Liu, R. Lonsdale, Q. Q. Wu,
J. P. Lin, D. M. Zhu, M. T. Reetz, Chem. Sci. 2017, 8, 4093-4099.
L. J. Kingsley, M. A. Lill, Proteins. 2015, 83, 599-611.
Z. J. Luan, F. L. Li, S. Dou, Q. Chen, X. D. Kong, J. Zhou, H. L. Yu,
J. H. Xu, Catal. Sci. Technol. 2015, 5, 2622-2629.
X. D. Kong, S. Yuan, L. Li, S. Chen, J. H. Xu, J. Zhou, Proc. Natl.
Acad. Sci. U. S. A. 2014, 111, 15717-15722.
Production of α-KIC and α-KMV on a 3-L scale
For large scale production of α-KIC and α-KMV, L-leucine (65,
110 g L-1) or L-isoleucine (55, 100 g L-1) was dissolved in a 3 L
Tris-HCl buffer (50 mM, pH 8.0). Wet whole cell catalysts
PmLAAD mutants (20 g L-1) were added to initiate the reaction.
The pH was automatically maintained at 8.0 by titrating with 4 M
NaOH. The reaction mixture in a 7.5-L INFORS fermenter was
stirred at 30 ºC and 500 rpm. The concentrations of α-KIC and α-
KMV were determined using the HPLC method as described
above.
[29]
[30]
[31]
[32]
[33]
A. Meister, Methods Enzymol. 1957, 3, 404-414.
D. Balducci, P. A. Conway, G. Sapuppo, H. Müller-Bunz, F. Paradisi,
Tetrahedron. 2012, 68, 7374-7379.
Acknowledgements
[34]
A. Waterhouse, M. Bertoni, S. Bienert, G. Studer, G. Tauriello, R.
Gumienny, F. T. Heer, T. A. P. de Beer, C. Rempfer, L. Bordoli, R.
Lepore, T. Schwede, Nucleic Acids Res. 2018, 46, W296-W303.
O. Trott, A. J. Olson, J. Comput. Chem. 2010, 31, 455-461.
W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, M.
L. Klein, J. Chem. Phys. 1983, 79, 926-935.
This work was financially supported by the National First-class
Discipline Program of Light Industry Technology and Engineering
(LITE2018-08), the Key Technologies R & D Program of Jiangsu
Province (BE2018623), and the Scientific and Technological
Innovation Leading Talents of National “Ten Thousand Talent
Program” (2018).
[35]
[36]
[37]
N. J. Kruger, The bradford method for protein quantitation, Humana
Press, New Jersey, 2002.
Keywords: catalytic efficiency • L-amino acid deaminase • protein
engineering • α-ketoisocaproate • α-keto-β-methylvalerate
[1]
a) S. Atsumi, T. Hanai, J. C. Liao, Nature. 2008, 451, 86-89; b) Y.
Song, J. H. Li, H. D. Shin, L. Liu, G. C. Du, J. Chen, Bioresource
Technol. 2016, 219, 716-724.
[2]
[3]
[4]
[5]
A. Schouteeten, Y. Christidis, 1985, European patent 0132201.
K. C. Kyowa Hakko, 1984, Japanese patent 59199652.
S. Alain, C. Yani, 1985, US patent 4518800.
A. J. Cooper, J. Z. Ginos, A. Meister, Chem. Rev. 1983, 83, 321-
358.
[6]
[7]
[8]
V. Buckle-Vallant, F. S. Krause, S. Messerschmidt, B. J. Eikmanns,
Appl. Microbiol. Biotechnol. 2014, 98, 297-311.
H. B. Coban, A. Demirci, P. H. Patterson, R. J. Elias, Bioproc Biosyst
Eng. 2014, 37, 2343-2352.
L. Pollegioni, P. Motta, G. Molla, Appl. Microbiol. Biotechnol. 2013,
97, 9323-9341.
[9]
[10]
G. Molla, R. Melis, L. Pollegioni, Biotechnol. Adv. 2017, 35, 657-668.
Y. P. Xue, C. H. Cao, Y. G. Zheng, Chem. Soc. Rev. 2018, 47, 1516-
1561.
[11]
[12]
[13]
M. Yoshimoto, M. Okamoto, K. Ujihashi, T. Okita, Langmuir. 2014,
30, 6180-6186.
M. García-García, I. Martínez-Martínez, A. Sánchez-Ferrer, F.
García-Carmona, Biotechnol. Prog. 2008, 24, 187-191.
a) P. Ödman, W. B. Wellbornb, A. S. Bommariusb, Tetrahedron:
Asymmetry. 2004, 15, 2933-2937; b) W. Greschner, C. Lanzerath,
T. Reß, K. Tenbrink, S. Borchert, A. Mix, W. Hummel, H. Gröger, J
Mol. Catal. B-Enzym. 2014, 103, 10-15.
[14]
[15]
[16]
[17]
[18]
[19]
A. L. Ahmad, E. M. Low, S. R. Abd Shukor, J Mol. Catal. B-Enzym.
2013, 88, 26-31.
N. M. W. Brunhuber, J. B. Thoden, J. S. Blanchard, J. L. Vanhooke,
J. Biochem. 2000, 39, 9174-9187.
T. Li, A. B. Kootstra, I. G. Fotheringham, Org. Process Res. Dev.
2002, 6, 533-538.
Y. Ju, S. Tong, Y. Gao, W. Zhao, Q. Liu, Q. Gu, J. Xu, L. Niu, M.
Teng, H. Zhou, J. Struct. Biol. 2016, 195, 306-315.
Y. Hou, G. S. Hossain, J. Li, H. D. Shin, G. Du, L. Liu, Appl. Microbiol.
Biotechnol. 2016, 100, 2183-2191.
R. X. Li, H. G. Sakir, J. Li, H. D. Shin, G. Du, J. Chen, L. Liu, RSC
Adv. 2017, 7, 6615-6621.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900259
ChemCatChem
FULL PAPER
Entry for the Table of Contents
FULL PAPER
In this study, we developed an efficient enzymatic
strategy for industrial production of α-KIC and α-
KMV from L-leucine and L-isoleucine, respectively,
employing enzyme PmLAAD. Poor productivity in
this process due to the insufficient catalytic
efficiency of PmLAAD. In an effort to increase the
catalytic efficiency of wild type PmLAAD for the
improvement ofα-KIC and α-KMV production, we
engineered the PmLAAD substrate binding cavity
and entrance tunnel.
Yuxiang Yuan, Wei Song, Jia Liu,
Xiulai Chen, Qiuling Luo, and
Liming Liu*
Page No. – Page No.
Production of α-ketoisocaproate
and α-keto-β-methylvalerate by
engineered L-amino acid
deaminase
This article is protected by copyright. All rights reserved.