Two key amino acids variant of α-l-arabinofuranosidase from bacillus subtilis str. 168 with altered activity for selective conversion ginsenoside rc to rd

Article abstract of DOI:10.3390/molecules26061733

α-L-arabinofuranosidase is a subfamily of glycosidases involved in the hydrolysis of L-arabinofuranosidic bonds, especially in those of the terminal non-reducing arabinofuranosyl residues of glycosides, from which efficient glycoside hydrolases can be screened for the transformation of ginsenosides. In this study, the ginsenoside Rc-hydrolyzing α-L-arabinofuranosidase gene, BsAbfA, was cloned from Bacilus subtilis, and its codons were optimized for efficient expression in E. coli BL21 (DE3). The recombinant protein BsAbfA fused with an N-terminal His-tag was overexpressed and purified, and then subjected to enzymatic characterization. Site-directed mutagenesis of BsAbfA was performed to verify the catalytic site, and the molecular mechanism of BsAbfA catalyzing ginsenoside Rc was analyzed by molecular docking, using the homology model of sequence alignment with other β-glycosidases. The results show that the purified BsAbfA had a specific activity of 32.6 U/mg. Under optimal conditions (pH 5, 40^?C), the kinetic parameters K_m of BsAbfA for pNP-α-Araf and gin-senoside Rc were 0.6 mM and 0.4 mM, while the K_cat /K_m were 181.5 s^?1 mM^?1 and 197.8 s^?1 mM^?1, respectively. More than 90% of ginsenoside Rc could be transformed by 12 U/mL purified BsAbfA at 40^?C and pH 5 in 24 h. The results of molecular docking and site-directed mutagenesis suggested that the E173 and E292 variants for BsAbfA are important in recognizing ginsenoside Rc effectively, and to make it enter the active pocket to hydrolyze the outer arabinofuranosyl moieties at C₂₀ position. These remarkable properties and the catalytic mechanism of BsAbfA provide a good alternative for the effective biotransformation of the major ginsenoside Rc into Rd.

Full text of DOI:10.3390/molecules26061733

molecules
Article
Two Key Amino Acids Variant of α-L-arabinofuranosidase from
Bacillus subtilis Str. 168 with Altered Activity for Selective
Conversion Ginsenoside Rc to Rd
Ru Zhang 1,2,*, Shi Quan Tan 1, Bian Ling Zhang 1, Zi Yu Guo 1, Liang Yu Tian 1, Pei Weng 1 and Zhi Yong Luo 3
1
College of Materials and Chemical Engineering, Hunan Institute of Engineering, Xiangtan 411104, China;
tanshiquan11@163.com (S.Q.T.); blzhang369@163.com (B.L.Z.); g13367365606@163.com (Z.Y.G.);
tianliangyu2021@163.com (L.Y.T.); wp1475963@163.com (P.W.)
Hunan International Joint Laboratory of Animal Intestinal Ecology and Health, Laboratory of Animal
2
Nutrition and Human Health, College of Life Sciences, Hunan Normal University, Changsha 410081, China
Molecular Biology Research Center, School of Life Sciences, Central South University,
3
Changsha 410078, China; luo_zhiyong@hotmail.com
*
Correspondence: zhangru2002@hnie.edu.cn; Tel.: +86-731-5868-3049
Abstract: α-L-arabinofuranosidase is a subfamily of glycosidases involved in the hydrolysis of L-
arabinofuranosidic bonds, especially in those of the terminal non-reducing arabinofuranosyl residues
of glycosides, from which efficient glycoside hydrolases can be screened for the transformation of
ginsenosides. In this study, the ginsenoside Rc-hydrolyzing α-L-arabinofuranosidase gene, BsAbfA,
ꢀꢁꢂꢀꢃꢄꢅꢆꢇ
was cloned from Bacilus subtilis, and its codons were optimized for efficient expression in E. coli BL21
(DE3). The recombinant protein BsAbfA fused with an N-terminal His-tag was overexpressed and
purified, and then subjected to enzymatic characterization. Site-directed mutagenesis of BsAbfA was
performed to verify the catalytic site, and the molecular mechanism of BsAbfA catalyzing ginsenoside
Rc was analyzed by molecular docking, using the homology model of sequence alignment with
ꢀꢁꢂꢃꢄꢅꢆ
Citation: Zhang, R.; Tan, S.Q.; Zhang,
B.L.; Guo, Z.Y.; Tian, L.Y.; Weng, P.;
Luo, Z.Y. Two Key Amino Acids
Variant of α-L-arabinofuranosidase
from Bacillus subtilis Str. 168 with
Altered Activity for Selective
other
β
-glycosidases. The results show that the purified BsAbfA had a specific activity of 32.6 U/mg.
◦
Under optimal conditions (pH 5, 40 C), the kinetic parameters Km of BsAbfA for pNP-
senoside Rc were 0.6 mM and 0.4 mM, while the Kcat/Km were 181.5 s mM and 197.8 s mM ,
α
-Araf and gin-
−1
−1
−1
−1
Conversion Ginsenoside Rc to Rd.
Molecules 2021, 26, 1733. https://
doi.org/10.3390/molecules26061733
respectively. More than 90% of ginsenoside Rc could be transformed by 12 U/mL purified BsAbfA at
◦
40 C and pH 5 in 24 h. The results of molecular docking and site-directed mutagenesis suggested that
the E173 and E292 variants for BsAbfA are important in recognizing ginsenoside Rc effectively, and
to make it enter the active pocket to hydrolyze the outer arabinofuranosyl moieties at C20 position.
These remarkable properties and the catalytic mechanism of BsAbfA provide a good alternative for
the effective biotransformation of the major ginsenoside Rc into Rd.
Academic Editors: Young-won Chin,
Young Hee Choi and Hojun Kim
Received: 9 March 2021
Accepted: 18 March 2021
Published: 19 March 2021
Keywords: Bacilus subtilis; α-L-arabinofuranosidase; ginsenoside Rc; biotransformation; ginsenoside
Rd; site-directed mutagenesis; molecular docking
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
1. Introduction
Ginseng, as a famous traditional herbal medicine, has been used to cure diseases and
promote health in East Asia for thousands of years. In recent decades, the medicinal value
of ginseng has been recognized worldwide [
that ginsenosides play pivotal pharmacological and therapeutic roles [
100 ginsenosides isolated and identified from ginseng, five major ginsenosides, viz., Rb1,
Rb2, Rc, Re, and Rg1, account for more than 80% of all ginsenosides [ ]. Ginsenoside Rd
has been proved to have unique pharmacological activities, such as reducing the prolif-
eration and migration of glioblastoma cells [ ], attenuating breast cancer metastasis [10],
1
,
2
]. A large number of studies have shown
3–5
]. Among over
Copyright:
© 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
6–8
9
stimulating the proliferation of endogenous stem cells [11], improving the blood–brain
barrier in ischemic stroke [12], attenuating mitochondrial dysfunction, and sequential
Molecules 2021, 26, 1733. https://doi.org/10.3390/molecules26061733
https://www.mdpi.com/journal/molecules

Molecules 2021, 26, 1733
2 of 20
apoptosis after transient focal ischemia [13]. Due to the value of ginsenoside Rd in medic-
inal applications, and as a promising medicine candidate, its transformation from other
ginsenosides has been investigated. However, the content of ginsenoside Rd in nature is
much less than that of the major ginsenosides [5,7], which makes it particularly difficult
to obtain ginsenoside Rd from ginseng and other plants. Some researchers have tried to
obtain ginsenoside Rd by chemical synthesis, but found it difficult to succeed, because of
its complex structure [14].
Ginsenosides are tetracyclic triterpenoids with similar structures (Figure 1). Most
ginsenosides belong to the protopanaxadiol type, the main difference lies in the variety and
quantity of C3 and C20 sugar groups. Ginsenoside Rd is structurally similar to Rb1, Rb2,
Rb3, and Rc, but lacks one outer glycoside moiety at position C20 [8]. Therefore, it is feasible
to obtain ginsenoside Rd from Rb1, Rb2, Rb3, and Rc by hydrolyzing the outer monosaccha-
ride residue (i.e., arabinopyranose or arabinofuranose moieties) using a specific glycosidase,
such as
all the major ginsenosides, the content of Rc accounts for about 20% of the total gin-
senosides [ 15]. Therefore, it can serve as an important substrate for ginsenoside Rd
β-glucosidase, α-L-arabinopyranosidase, or α-L-arabinofuranosidase [15–18]. In
7
,
production. Hence, it is essential to screen enzymes and/or reagents that can cleave the
arabinofuranose at the C20 position of ginsenoside Rc with high activity and specificity.
At present, physical and chemical methods have been developed to obtain ginsenoside
Rd from other major ginsenosides, but the efficiency and specificity are not ideal [19].
Glycosidases are widely distributed in almost all organisms, which hydrolyze glycosidic
bonds in various glycosides by endo- or exo-digestion. Compared to the known physical
and chemical methods, the preparation of rare ginsenosides by hydrolysis of glycosidic
bonds using glycosidase has advantages, such as high selectivity, mild reaction condition,
and environmental friendliness [20].
α-L-arabinofuranosidase (EC 3.2.1.55) is a subfam-
ily of glycosidases commonly involved in the hydrolysis of L-arabinofuranosidic bonds,
especially in the hydrolysis of terminal non-reducing arabinofuranosyl residues from
different oligosaccharides and polysaccharides [21,22]. Some α-L-arabinofuranosidases
from Bifidobacterium breve K-110 [23], Rhodanobacter ginsenosidimutans Gsoil 3054T [15],
Caldicellulosiruptor saccharolyticus [24], Sulfolobus solfataricus [25], Thermotoga thermarum
DSM5069 [26], and Geobacillus caldoxylolyticus TK4 [27] have been proved to possess the
ability of transforming ginsenoside Rc into ginsenoside Rd. Although there are several
α
-L-arabinofuranosidases from different species that are known to have the potential
of preparing ginsenoside Rd, there have been no reports on the scaled-up production
of ginsenoside Rd using -L-arabinofuranosidase. This is plausibly due to limitations
α
such as low activity and poor extraction efficiency for natural enzymes, as well as low
expression level, poor substrate specificity, and low transformation ability for recombinant
enzymes. Moreover, the mechanism of ginsenoside hydrolysis by glycosidase is unclear,
enormously hindering further modification and optimization of the enzyme. Therefore,
it is urgent to investigate the efficient and specific hydrolysis of ginsenoside Rc using
α-L-arabinofuranosidase.
In this paper, the ginsenoside Rc-hydrolyzing α-L-arabinofuranosidase gene BsAbfA
was cloned from B. subtilis and its codons were optimized for efficient expression in E. coli.
The optimized recombinant protein, BsAbfA, fused with an N-terminal His-tag was over-
expressed and purified, and then subjected to enzymatical characterization. The catalytic
efficiency of recombinant BsAbfA on biotransformation of the major ginsenoside Rc to Rd
was investigated (Figure 2). We aimed to explore more α-L-arabinofuranosidases for effi-
cient enzymatic production of ginsenoside Rd. We performed a site-directed mutagenesis
of BsAbfA to verify the catalytic site. Furthermore, we analyzed the molecular mechanism
of BsAbfA catalysis on ginsenoside Rc by molecular docking using the homology model
of sequence alignment with other β-glycosidases. The E173 and E292 variants of BsAbfA
are important for the effective recognition of ginsenoside Rc, leading to its entrance to the
active pocket for the hydrolysis of the outer arabinofuranose moieties at C20 position. These

Molecules 2021, 26, 1733
3 of 20
remarkable properties and catalytic mechanism of BsAbfA provide a good alternative for
the effective biotransformation of ginsenoside Rc into Rd.
Figure 1. Chemical structure of ginsenoside Rb1, Rb2, Rc, Rd, Re, Rg1, F2, C-K, C-Mc, and C-Mc1.
Glc, Ap, Af, and Rp are abbreviations of glucopyranosyl, arabinopyranosyl, arabinofuranosyl, and
rhamnopyranosyl, respectively.
Figure 2. Proposed biotransformation pathway of ginsenoside Rc into Rd by BsAbfA.
2. Results and Discussion
2.1. Cloning of BsAbfA Gene and Sequence Analysis
In the structural characterization of BsAbfA (GenBank accession: AL009126.3, 2938330–
2939832) an alignment of the amino acid sequence of BsAbfA with several GH51
α-L-
arabinofuranosidases by using ClustalX indicated the sharing of three conserved motifs.
Studies have shown that arabinofuranosidases are involved in general acid-base catalysis,
and two essential residues are required in the process of cleaving glycosidic bonds. In
most arabinofuranosidases conservative aspartic acid and/or glutamic acid residues are
available and needed [22,28]. Among the two catalytic residues, one acts as a general
acid to provide proton assistance for the departure of glycosidic oxygen, and the other
acts as a general base to activate a water molecule and affect the direct replacement at
the anomeric center [22]. As shown in Figure 3, the first motif with a conserved RYPGG
sequence is regarded as an important motif for stabilizing the structure and confers flex-
ibility to the enzyme [29]. According to sequence similarities with BsAbfA, the motifs

Molecules 2021, 26, 1733
4 of 20
of WCLGN
reserved with the GH51
bold italic letters) are regarded as typical and necessary acid-base and nucleophilic catalytic
residues to hydrolyze the glycosidic bond [22 30 31], respectively. The three conserved mo-
EMDGPWQ (residues 168–179) and D
E
WNVW (residues 291–296) are highly
α
-L-arabinofuranosidases, and the residues E173 and E292 (Red
,
,
tifs of BsAbfA are basically the same as those of AbfA (GenBank accession No. ADM26764)
from R. ginsenosidimutans Gsoil 3054. This AbfA shows preferential substrate specificity
for exo-polyarabinosides or oligoarabinosides, and it only hydrolyzes arabinofuranoside
moieties from ginsenoside Rc and its derivatives, and no other sugar groups [15]. Be-
tween BsAbfA and the α-L-arabinofuranosidase (GenBank accession No. ABP67153) from
T. thermarum DSM5069, we found only one residue difference in the three motifs. It is
worth noting that they also have a high specific ability to biotransform ginsenoside Rc to
Rd through the hydrolysis of the arabinoside bond [26]. Therefore, we speculate that they
may share a similar catalytic mechanism.
2.2. Identification of Enzymatic Properties of Recombinant BsAbfA
The wild-type BsAbfA gene without any optimization was cloned and fused in pET-
28a (+), and the results showed that expression in E. Coli BL21 (DE3) was difficult. There-
fore, the codons of wild-type BsAbfA were optimized and synthesized for expression
efficiency. The optimized and mutated BsAbfAs fused to His-tag were purified using
Ni-NTA magnetic agarose beads from E. Coli BL21 (DE3), followed by the induction of
0.5 mM IPTG at 20 ◦C for 16 h. SDS-PAGE analyses revealed that the molecular masses of
all the optimized and mutated BsAbfAs were similar to those of those predicted according
to amino acid sequences (approximately 60 kDa, data not shown). The results indicated
that the mutated BsAbfAs were correctly expressed and folded. The purified BsAbfA,
with 32.6 U/mg and using p-nitrophenyl-
displayed a higher activity than most of the characterized
as -L-arabinofuranosidase from C. saccharolyticus with 28.2 U/mg [32]. The crude enzyme
and purified recombinant -L-arabinofuranosidase from Cellulosimicrobium aquatile Lyp51
showed a lower activity of 2 U/mg and 15 U/mg, respectively [33]. The activity of T.
petrophila -L-arabinofuranosidase purified from E. Coli BL21 (DE3) toward ginsenoside
Rc was only 10.3 U/mg at the optimal condition [20]. The ginsenoside-hydrolyzing
α
-L-arabinofuranoside (pNP-
α-Af) as substrate,
α
-L-arabinofuranosidases, such
α
α
α
α
-
L-arabinofuranosidase from R. ginsenosidimutans Gsoil 3054T was 14.9 U/mg [15]. To
investigate the importance of the two key residues (i.e., E173 and E292) for hydrolysis of
ginsenoside Rc, each one was independently substituted by an alanine, asparagine, and
glutamine. It was found that the E173A or E292A mutants had no specific activity toward
ginsenoside Rc and pNP-α-Af. Furthermore, the mutants of E173D, E173Q, E292D, and
E292Q led to an obvious decrease of activity, and the extent of activity loss was 73.4%,
46.1%, 76.4%, and 56.1%, respectively, compared to that of BsAbfA using ginsenoside Rc
as a substrate (Table 1). For investigation of the catalytic mechanism, researchers have
attempted to identify the residues of α-L-arabinofuranosidase that are catalytically essential.
They found that all enzymes of this superfamily possess conserved glutamates E173 and
E292 (numbering for BsAbfA based on the alignment), which are catalytically essential
acid-bases and nucleophiles, respectively [22,34].
2.3. Temperature and pH Dependence of Recombinant BsAbfA
To investigate the temperature and pH dependence of BsAbfA, the enzymic activity
of BsAbfA and isosteric mutants E173Q and E292Q were determined using pNP-
α-Af as
substrate at different temperatures and pH environments. As disclosed in Figure 4A, the
◦
◦
recombinant BsAbfA and mutants were active at 30–50 C and relatively stable at 25–45 C.
The optimal temperature was 40 ◦C. When the temperature was higher than 45 ◦C, the
activity decreased sharply. The◦activity loss for all the tested enzymes was more than 60%,
80%, and 90% at 50, 55, and 60 C, respectively. It is known that many glycosidases exhibit
optimal activity at mild temperature conditions. For example, the optimal temperatures of
the glycosidase from Leuconostoc mesenteroides DC102 [35], β-glucosidase from Lactobacillus

Molecules 2021, 26, 1733
5 of 20
brevis [36],
β-D-xylosidase,
α
-L-arabinopyranosidase, and
α
-L-arabinofuranosidase from
Bifidobacterium breve K-110 [23
,37], α-L-arabinofuranosidase from Leuconostoc sp. 22-3 [16],
as well as those of soil deuteromycete Penicillium funiculosum [16] and Cellulosimicrobium
aquatile Lyp51 [33] are 30–45 ◦C for ginsenosides biotransformation.
Figure 3. Multiple sequence alignment of BsAbfA with
α-L-arabinofuranosidases from several
microorganisms: CAA61937 (BsAbfA) from B. subtilis, ADM26764 from R. ginsenosidimutans Gsoil
3054, ABP67153 from T. thermarum DSM5069, and AEH51197 from Pseudothermotoga thermarum DSM
5069. The three motifs with conserved RYPGG (residues 67–73), WCLGNEMDGPWQ (residues
168–179), and DEWNVW (residues 291–296) are highlighted in a box. The residues E173 and E292
(Red bold italic letters) are regarded as typical and necessary acid-base and nucleophilic catalytic
residues for the hydrolysis of the glycosidic bond.

Molecules 2021, 26, 1733
6 of 20
Table 1. Production of ginsenoside Rd from ginsenoside Rc by optimized and mutated BsAbfA.
Relative Activity (%) b
Enzymes
Substrates a
BsAbfA
BsAbfA
E173A
E173D
E173Q
E292A
E292D
E292Q
pNP-α-Af
100.1 ± 6.2
121.6 ± 4.3
ND c
32.3 ± 2.9
65.6 ± 2.7
ND
Ginsenoside Rc
Ginsenoside Rc
Ginsenoside Rc
Ginsenoside Rc
Ginsenoside Rc
Ginsenoside Rc
Ginsenoside Rc
28.7 ± 3.8
53.4 ± 2.2
a
b
Substrate concentration: 1 mM ginsenoside Rc and 10 mM pNP-α-Af. The reaction was performed in 50 mM
citric acid/sodium citrate buffer (pH 5) at 40 ◦C for 24 h, and the amount of enzyme was equivalent to 12 U/mL
c
of BsAbfA. The relative activity of pNP-α-Af was defined as 100%. ND: not detected.
Figure 4. Temperature, pH dependence, and thermal stability of enzymatic activity of BsAbfA. (
A
) Activity measurements
◦
catalyzed by BsAbfA, E173Q, and E292Q mutants at 25-60 C with pH 5. (
B) Activity measurements catalyzed by BsAbfA,
◦
E173Q, and E292Q mutant at 40 C; the buffers in a pH range of 3-9 were citric acid-sodium citrate (50 mM, pH 3-6), sodium
phosphate (100 mM, pH 6-8), and glycine-NaOH (50 mM, pH 9–10). (
50 C, catalyzed by BsAbfA at pH 5. The maximum relative activity of each enzyme was defined as 100% using pNP-
C
) The thermal stabilities were tested at 35, 40, 45, and
◦
α
-Af
as substrate.
As shown in Figure 4B, the pH dependent curves of the activity of BsAbfA and E292Q
mutants display an increase of pH sensitivity, while E173Q is characterized by a marked
insensitivity to pH. Higher activities were obtained at pH 4.5-6 for BsAbfA and E292Q
mutant. In 50 mM citric acid/sodium citrate buffer, the BsAbfA and E292Q mutants
showed maximal activity at pH 5, and activities of above 80% were retained at pH 6. When
the pH was above 6 or below 4.5, the enzyme activity decreased significantly. Interestingly,

Molecules 2021, 26, 1733
7 of 20
the activity of E173Q mutant remained at a high level with the increase of pH until the
pH was higher than 7. The results suggest that the mutation of E173 residue should lead
to significant alteration of pH dependence. In this case, the activity of the mutant E173Q
is consistent with that of a mutant enzyme that is without catalytic acid-base residue, as
reported in the literature [22
result of E173 mutation supports the view that just like other
the residue is an acid-base catalyst [38 40 41]. Therefore, it is reasonable to consider that
the sites of glutamate residues relative to each other, as well as to the substrate should be
critical for efficient catalysis. In this study, no matter if it was pNP- -Af or ginsenoside
,38
,
39]. The main change of the pH dependence curve as a
α
-L-arabinofuranosidases,
,
,
α
Rc that was used as substrate, the temperature or pH conditions for maximum enzyme
activity stayed the same.
The thermal stability of BsAbfA at pH 5 versus incubation time is depicted in Figure 4C
.
The thermodynamic parameters confirm that BsAbfA was stable below 40 ◦C and it had
◦
a half-life of 225.7, 196.9, and 74.6 h at 35, 40, and 45 C, respectively. The enzyme de-
creased significantly in stability above 50 ◦C, and the half-life was only 10.8 h. BsAbfA
has a half-life comparable to those of characterized
species at its optimized temperature. For example, the half-life of
originating from S. solfataricus is 30 h at 85 ◦C [25]. Furthermore, its half-life is higher than
α
-L-arabinofuranosidase from other
α
-L-arabinofuranosidase
some other β
-glucosidases from Fusarium solani (159 min, 65 ◦C) [29] and Alteromonas sp.
L82 (21 min, 40 ◦C) [42]. It should be emphasized that during the 3 months storage of the
BsAbfA (at 4 ◦C), there was an activity loss of about 18%, which could be attributed to the
good stability of BsAbfA during the test. It is obvious that the long half-life and appreciable
thermostability of BsAbfA are properties desirable for practical applications.
2.4. Kinetic Analysis of BsAbfA
The Km, Kcat, and Kcat/Km for pNP-
α-Af were determined under the optimal conditions
for enzymatic reactions catalyzed by BsAbfA and mutants (Table 2). The Km, Kcat, and
−1 −1
cat/Km were 0.6 mM, 108.9 s−1, and 181.5 s mM for BsAbfA, respectively. No activity
K
was detected for E173A and E292A. It is noted that the Km values of the majority of
other mutants, except E173A and E292A, were like those of BsAbfA, whereas the Kcat
and Kcat/Km values of all the mutants decreased significantly. The E173Q mutant had a
highest Kcat/Km that was only 28.8-fold lower than that of BsAbfA, and a Kcat value only
43.6-fold lower. Likewise, the Kcat values of the E173D, E292D, and E292Q mutants decrease
by 217.8-, 121.1-, and 68.1-fold, while the Kcat/Km values decreased by 72.6-, 139.6-, and
78.9-fold, respectively. The Km, Kcat, and Kcat/Km values of BsAbfA for ginsenoside Rc
were 0.4 mM, 79.1 s−1, and 197.8 s−1 mM−1 (data not listed). The catalytic efficiency of
BsAbfA for ginsenoside Rc was higher than that of glycosidase from S. acidocaldarius for
Rb1 (Kcat, 4.8 s−1 mM−1), Rc (Kcat, 4.5 s−1 mM−1), Rd (Kcat, 1 s−1 mM−1), and Rb2 (Kcat
0.8 s−1 mM−1) [43]. The results suggest that BsAbfA is an efficient enzyme for hydrolyzing
ginsenoside Rc.
,
Table 2. Enzymatic parameters for hydrolysis of pNP-α-Af by optimized and mutated BsAbfA a.
−1
−1
−1
)
Enzymes
Km (mM)
Kcat (s
)
Kcat/Km (s mM
BsAbfA
E173A
E173D
E173Q
E292A
E292D
E292Q
0.6 ± 0.08
108.9 ± 8.6
181.5 ± 6.9
ND b
ND
ND
0.2 ± 0.05
0.4 ± 0.07
ND
0.5 ± 3.1
2.5 ± 2.1
ND
2.5 ± 0.06
6.3 ± 0.09
ND
0.7 ± 0.06
0.9 ± 0.05
1.3 ± 0.03
0.7 ± 0.09
1.6 ± 0.04
2.3 ± 0.05
The reaction was performed in citric acid/sodium citrate buffer (pH 5), 12 U/mL enzyme at 40 ◦C for 24 h. The
a
b
released pNP was assayed spectrophotometrically at 405 nm. ND: not detected.
The effects of metal ions and chemicals on the kinetic parameters of BsAbfA are indi-
cated in Table 3. The BsAbfA activity was obviously enhanced by Mn2+, but significantly

Molecules 2021, 26, 1733
8 of 20
inhibited by Hg2+ and Cu2+. The presence of Na+, K+, Ca2+, Mg2+, Fe2+, Zn2+, Ni2+, EDTA,
DDT, or SDS had no significant effect on the enzyme activity. The results demonstrate that
the recombinant BsAbfA had a good catalytic activity and environmental compatibility.
Table 3. Effects of metal ions and chemicals on the activity of BsAbfA.
Relative Activity ± SD (%) a
Metal Ions or Chemicals
1 mM
5 mM
Na+
K+
100.4 ± 2.1
99.8 ± 1.4
98.4 ± 2.7
99.6 ± 1.8
105.7 ± 2.2
119.7 ± 2.4
87.1 ± 1.9
98.8 ± 2.8
36.8 ± 0.9
20.3 ± 1.7
100.2 ± 2.3
98.6 ± 2.7
98.2 ± 2.5
100 ± 1.9
98.3 ± 1.7
93.6 ± 1.5
93.1 ± 1.9
91.5 ± 2.3
95.7 ± 3.1
109.4 ± 2.5
77.8 ± 2.4
91.8 ± 2.5
24.3 ± 1.1
5.3 ± 1.1
Ca2+
Mg2+
Fe2+
Mn2+
Zn2+
Ni2+
Cu2+
Hg2+
EDTA
DTT
99.7 ± 3.1
97.4 ± 2.8
81.2 ± 3.2
100 ± 2.6
SDS
Control
a
Relative activities of BsAbfA were assayed using 10 mM pNP-
α
-Af as substrate in 50 mM citric acid/sodium
citrate buffer (pH 5) with 12 U/mL enzyme at 40 ◦C for 24 h. The relative activity of pNP-
α-Af was defined
as 100%.
2.5. Substrate Specificity of BsAbfA
The substrate specificity of BsAbfA was measured under the optimal conditions using
pNP- -Af, pNP- -L-arabinopyranoside (pNP- -Ap), pNP- -L-rhamnopyranoside (pNP-
-Rp), pNP- -D-glucopyranoside (pNP- -Glc), Rb1, Rb2, Rc, Rd, Re, Rg1, C-Mc1, C-Mc,
gentiobiose, and sophorose as substrates. As revealed in Table 4, BsAbfA displayed a
high hydrolytic activity on ginsenoside Rc and pNP- -Af, having a preference for Rc.
However, BsAbfA had a low activity on C-Mc1 and C-Mc, and no activity on pNP- -Ap,
pNP- -Rp, pNP- -Glc, Rb1, Rb2, Rd, Re, Rg1, gentiobiose, and sophorose. Of the eight
α
α
α
α
α
β
β
α
α
α
β
ginsenosides except for Re and Rg1, all belonged to the protopanaxadiol-type (PPD-type).
In contrast, ginsenoside Rd contains only one glucopyranosyl at C20. The main difference
between Rb1, Rb2, and Rc is that the sugar residues substituted at C20 of aglycone have
sugar moieties (i.e., glucopyranose, arabinopyranose, and arabinofuranose) linking to the
glucopyranosyl at C20 of aglycone [44]. Owing to the structure at C20 like that of Rc, the
minor ginsenosides C-Mc1 and C-Mc were used as controls to test the specificity. The results
suggest that BsAbfA specifically cleaves the outer glucosidic linkage at the C20 position of
ginsenoside Rc, C-M, and C-M1, but does not hydrolyze the inner glucosidic linkage and
glucopyranosyl at C20 of PPD-type ginsenosides. It also cannot hydrolyze any of the outer
or inner sugar moieties, including glucopyranose, arabinopyranose, and rhamnopyranose
at the C3 or C6 (only for the protopanaxatriol-type) position of the ginsenoside skeleton.
In addition, the specific activity of BsAbfA for the ginsenosides follows the order of Rc
> C-M > C-M1. Although the ginsenoside Rc, C-M, and C-M1 have the same glycosidic
bond at C20 position, the recombinant BsAbfA is more active towards Rc and has specific
stereo preference for C3 sugars. The C3 position of ginsenosides C-M and C-M1 contains a
hydrogen or glucopyranose with small steric structures that may not benefit the formation
of hydrogen bonds between Rc and BsAbfA. Therefore, BsAbfA has a high selectivity to
ginsenoside Rc, and it hydrolyzes the glucoside at the C20 position in ginsenosides, whereas
the enzyme does not hydrolyze the glycoside at the C3 and C6 position. That BsAbfA has
the same substrate specificity to the recombinant α-L-arabinofuranosidase CaAraf51 from
C. aquatile has been previously reported [33].

Molecules 2021, 26, 1733
9 of 20
Table 4. Substrate specificity of BsAbfA.
Substrates a
Relative Activity (%) b
pNP-α-Af
pNP-α-Ap
pNP-α-Rp
100 ± 3.9
ND c
ND
pNP-β-Glc
ND
ND
ND
Ginsenoside Rb1
Ginsenoside Rb2
Ginsenoside Rc
Ginsenoside Rd
Ginsenoside Re
Ginsenoside Rg1
Ginsenoside F2
C-K
120.6 ± 2.9
ND
ND
ND
ND
ND
C-Mc1
C-Mc
Gentiobiose
106.2 ± 3.8
108.9 ± 3.5
ND
Sophorose
Substrate concentration: 10 mM pNP-
ND
a
α
-Af, pNP-
α
-Ap, pNP-
α
-Rp, pNP-
β
-Glc; 1 mM Rb1, Rb2, Rc, Rd, Re, Rg1,
b
F2, C-K, C-Mc1, C-Mc, gentiobiose, and sophorose. The reaction was performed in 50 mM citric acid/sodium
citrate buffer (pH 5), 12 U/mL enzyme at 40 ◦C for 24 h. The relative activity of pNP-
α
-Af was defined as 100%.
c
ND: not detected.
2.6. Biotransformation of Ginsenoside Rc by BsAbfA
The effect of BsAbfA on the biotransformation of ginsenoside Rc was investigated
◦
at pH 5 and 40 C by varying the enzyme amount from 0 to 20 U/mL enzyme with
2 20 mg/mL Rc for 24 h. As shown in Figure 5A, the molar conversion rate of Rc reached
−
65% using 4 U/mL enzyme. At 8 U/mL enzyme, ginsenoside Rc was biotransformed to Rd
with a corresponding molar conversion rate of 85%. With the increasing of enzyme activity,
the productivity of ginsenoside Rd gradually increases, and at BsAbfA of 12 U/mL, the
conversion rate of ginsenoside Rc reached the climax and was more than 90%. As shown in
Figure 5B, 4 16 mg ginsenoside Rc/mL was efficiently converted to Rd with a more than
−
80% conversion rate. The conversion rate was still as high as 81% at 16 mg ginsenoside
Rd/mL, but decreased obviously above 20 mg ginsenoside Rc/mL. The results indicate
that BsAbfA has a good relative stability and high substrate tolerance.
Figure 5. Effect of enzyme amount (A) and ginsenoside Rc concentration (B) on the production of Rd by using purified
BsAbfA. (
Rc and 0
(pH 5) containing 2
three experiments, and error bars represent standard deviation.
A
) The reaction was performed in 50 mM citric acid/sodium citrate buffer (pH 5) containing 5 mM ginsenoside
◦
−
20 U/mL enzyme at 40 C for 24 h. (
B
) The reaction was performed in 50 mM citric acid/sodium citrate buffer
◦
−
20 mg/mL ginsenoside Rc and 12 U/mL BsAbfA at 40 C for 0
−
72 h. The data represent the means of

Molecules 2021, 26, 1733
10 of 20
To investigate the transformation mechanism, time-course analysis of the reaction
under optimal conditions was performed. The biotransformed products were analyzed
by LC-MS and ESI-MS/MS techniques. As shown in Figure 6A–E, there was the detection
of ginsenoside Rd using UPLC after 6 48 h, and an obvious decrease of Rc and increase
−
of Rd after 12 h. The results demonstrate that ginsenoside Rc was transformed into Rd in
large amounts. At 24 h the Rd yield became much higher and Rd was the sole product of
biotransformation. To further verify the transformed products, ESI-MS/MS was used for
structural information (Figure 6F). The ESI-MS/MS spectrum of Rc displayed a signal from
[M
from [M
−
H]− at m/z 1077.27. For the transformed product, there was a signa−l at m/z 945.35
−
−
H] generated via the loss of HCOOH (46 Da) from [M + HCOO] at m/z 991.35
(Figure 6G). By comparison, the retention time and ESI-MS/MS fragment patterns of the
transformed product were the same as those of the ginsenoside Rd standard. Therefore,
the sole transformed product was identified as ginsenoside Rd. The BsAbfA exhibited
substrate specificity for ginsenosides Rc with arabinofuranose moieties at the C20 position,
indicating a specific affinity to outer C20 arabinofuranose. The proposed pathway of
biotransformation ginsenoside Rc by BsAbfA was as illustrated in Figure 2.
2.7. Molecular Docking and Examination of BsAbfA Active Site
From the results of the sequence analysis, it is reasonable to conclude that ginsenosides
with arabinofuranose are the likely substrate of BsAbfA. Of the ginsenosides (over 100)
that have been discovered, ginsenosides Rc, C-Mc1, and C-Mc contain arabinofuranoside
bonds at C20 of the ginsenoside skeleton. Among them, Rc is higher in abundance, and
is often used to prepare its derivatives [16,25,45]. In order to understand the molecular
mechanism of the interaction between BsAbfA and ginsenoside Rc, a three-dimensional
(3D) model of BsAbfA was built using alpha-L-arabinofuranosidase Ara51 (PDB code:
5O7Z) from Clostridium thermocellum [21] as a template in the SWISS-MODEL server, and
the sequence identity and similarity were 64.24% and 51%, respectively. To obtain a
final structural model, the collected model of BsAbfA was optimized using an energy
minimization procedure followed by a short molecular dynamics simulation. As shown
in Figure 7, the predicted BsAbfA structure used for docking studies with ginsenoside
Rc displayed the outer part of a structural fold that contains
α
-helices (cyan) and loops
(purple), while the inner part is almost all parallel
active pocket.
β-sheets (red), forming a catalytically
The surface electro-static potential plot at the active site of BsAbfA receptor (A) and
the structure of ginsenoside Rc are shown in Figure 8. The electrostatic surface potential of
the model was used to illustrate the potential interactions between the BsAbfA receptor
and substrate ginsenoside Rc. As shown in Figure 8A, the electrostatic potential of the
hydrolyzed active pocket is a deep negatively charged cavity, suggesting that the glycosidic
bond and sugar ring may be attracted as a result of electrostatic interaction, which is one of
the reasons for the selectivity of ginsenoside Rc (Figure 8B–C) toward the active pocket.
To understand the structural foundation of BsAbfA hydrolysis, the candidate substrate
ginsenoside Rc was docked into the catalytic pocket of BsAbfA. The BsAbfA was held
rigid while the substrate ginsenoside was allowed to flex during the docking process. The
docking results were analyzed based on the energy and efficiency of polar and non-bonded
interactions. It was found that ginsenoside Rc could easily dock into the catalytic pocket of
BsAbfA and in different forms. The selected poses of ginsenoside Rc show binding energy
from
−
8.9 to
−
9.8 kcal/mol, and the dissociation constant was between 6.29
×
104 pM and
2.86 × 105 pM (Table 5).

Molecules 2021, 26, 1733
11 of 20
Figure 6. LC/ESI-MS analysis of metabolites of ginsenoside Rc hydrolysis by BsAbfA. (
F2, and C-K. ( ) The biotransformed products of ginsenoside Rc to Rd for 6, 12, 24, and 48 h respectively. (
ion chromatograms obtained from the biotransformed products of Rc by BsAbfA for 0 h and 48 h, respectively.
A
) Standard ginsenoside Rc, Rd, Re,
B–
E
F–G) Extracted

Molecules 2021, 26, 1733
12 of 20
Figure 7. The 3D model of BsAbfA, with location of the deduced catalytic sites indicated. The 3D
model of BsAbfA in top view (
the crystal structure of -L-arabinofuranosidase Ara51 (PDB code: 5O7Z) from C. thermocellum as
a template in the SWISS-MODEL server. The colored segment of the skeleton structure marks the
position of -sheet (red), -helix (cyan), and loop (purple). The red circles show the active pockets or
sites. The active sites of three residues (i.e., N72, E173, and E292) are shown as a stick. ( ) The top
and surface view are shown in overlap form. (D) The active pocket is shown in a surface view.
A) and surface view (B) was built by comparative modeling using
α
β
α
C
Figure 8.
) geometry of ginsenoside Rc (non-polar hydrogens are implicit). Coloring represents the electrostatic surface potential
(KbT/ec) with a spectrum of red (electronegative) to blue (electropositive). The red circles show the active pocket.
(A) Surface electro-static potential plot at the active site of BsAbfA receptor, (B) structure of ginsenoside Rc, and
(
C

Molecules 2021, 26, 1733
13 of 20
Table 5. Cluster analysis of the docking position (kcal/mol).
Cluster
Members
Energy
Dissociation Constant (pM)
6.29 × 104
2.19 × 105
2.49 × 105
2.64 × 105
2.86 × 105
1
2
3
4
5
3
2
2
2
1
−9.824
−9.158
−9.086
−8.991
−8.927
It is observed that the more negative the docking energy, or the smaller the dissociation
constant, the better the binding ability and stability of the ligand with protein. The lowest
docking score for the ligand–receptor binding indicates that the binding ability is strong.
Therefore, a discussion was conducted based on the minimum and maximum docking
score (cluster 1 and 5). The interactions between ginsenoside Rc and binding pocket of
BsAbfA are shown in Figures 9–11. As shown in Figure 9, ten important amino acid
residues, viz. N72, A96, W97, Q98, N172, S214, N215, H242, E292, and W296 of BsAbfA,
bind to ginsenoside Rc in the form of hydrogen bond or hydrophobic bond. Among the
binding sites, five residues (i.e., N72, N172, S214, H242, and E292) form hydrogen bonds
with the outer arabinofuranose moieties at C20 position, one residue (i.e., N215) binds to
the inner glucopyranose moieties of C20 position, especially the hydrogen bond at N72 site
is crucial to stabilizing the structure, while E173 and E292 are key residues responsible for
substrate catalysis [22,29–31]. In this structure, N172 replaces E173 to form the hydrogen
bond, while the neighbor residue E173 participates in the hydrophobic interaction with
BsAbfA (Figure 10). Furthermore, the hydrophobic interaction of W97 and W296 residues
could, relatively, be as important as that of hydrogen bonding. Despite the limit of space
in the binding pocket for the accommodation of the fatty acid chain upon C20 insertion
into the narrow hydrophilic pocket, there is binding of W97 and W296 residues to the
fatty acid chain with hydrophobic bonds, squeezing the arabinofuranose into the active
pocket as a consequence. There are only two residues (i.e., A96 and Q98) that bind to
the glucopyranose moieties at the C3 position of ginsenoside Rc, which is at the opposite
side of C20. Overall, due to the structure of the covalent glycosyl-enzyme intermediate
of BsAbfA and ginsenoside Rc, the arabinofuranose cycle at C20 easily inserted into the
narrow hydrophilic pocket in a stable manner. Accordingly, the C3 position of ginsenoside
Rc only laid in the groove outside the binding pocket.
Further analysis of the enzyme substrate complexes by ligplot disclosed that one
of the best complexes amongst the first 100 contains 19 acid residues that participate to
form hydrogen bonds or hydrophobic bonds (Figure 11), and having E173 and E292 in the
catalytic cleft or active site. Additionally, the different docking models and experimental
results imply that the two conserved E173 and E292 residues are in close contact with the
substrate to catalyze the reaction [30,31]. Site directed mutagenesis demonstrated that
E173 and E292 mutations of BsAbfA reduce or disable its ability to hydrolyze ginsenoside
Rc, indicating that these two amino acids are essential for recognizing hydrolysis of the
outer arabinofuranosyl moieties at the C20 position. Moreover, the enzyme–substrate
docking data showed conformity with the reports concerning similar Glu-based catalytic
sites of
neighboring amino acids [22
conservedin GH51 -L-arabinofuranosidases [30
with that of the sequence alignment in Figure 3.
α
-L-arabinofuranose that emphasized the active role of E173, along with the other
31]. The location and function of these key residues are
31], and the result is in good agreement
,
α
,
In contrast, if the ginsenoside Rc docked into the BsAbfA in the manner of cluster 5
(Figure 11), there are only seven residues that can bind to ginsenoside Rc, and no bond
is formed between the arabinofuranose moiety and BsAbfA. In the BsAbfA complex, the
entire ginsenoside Rc can form six hydrogen bonds and one hydrophobic bond in a scat-
tering manner. Therefore, it is less tightly bonded in comparison with the cluster 1 case.
Figure 11D shows the two binding forms of ginsenoside Rc and BsAbfA, in which the
green and yellow sticks stand for the binding forms of cluster 1 and cluster 5, respectively.

Molecules 2021, 26, 1733
14 of 20
The results indicate that arabinofuranosyl is located deeper into the active pocket than
glucopyranosyl. Interestingly, according to the Kcat/Km of 197.8 s−1 mM−1, BsAbfA hy-
drolyzes ginsenoside Rc with higher catalytic affinity. These results suggest that the outer
hexose ring at C20 position induces a steric hindrance, which makes BsAbfA protein highly
selective for the recognition of pentose α-L-arabinofuranose.
Figure 9. Molecular docking of ginsenoside Rc in the cluster 1 model of BsAbfA (with the minimum
docking score). (
A
) Stereoview of the predicted binding of BsAbfA by initial docking. (
B) Key residues
interacting with ginsenoside Rc. (C) The binding form and type of key residues with ginsenoside Rc.
The dashed line represents hydrogen bonding, and the green line represents hydrophobic interaction.
Figure 10. Ligplot output for H-bond and hydrophobic interactions of the amino acid residues of
BsAbfA involved in binding with ginsenoside Rc. Hydrogen bonds < 3.1 Å are shown as green
dotted lines.

Molecules 2021, 26, 1733
15 of 20
Figure 11. Molecular docking of ginsenoside Rc in the cluster 5 model of BsAbfA (with the maximum docking score).
(
(
A
C
) Stereoview of the predicted binding of BsAbfA by initial docking. (
) The binding form and type of key residues with ginsenoside Rc. The dashed line represents hydrogen bond, and the
) Docking positions of cluster 5 compared to cluster 1. Ginsenoside
B) Key residues interacting with ginsenoside Rc.
green line represents hydrophobic interaction. (
D
Rc represented as yellow (cluster 5) and green (cluster 1) stick model inside a solvent excluded surface (SES) colored
by elements.

Molecules 2021, 26, 1733
16 of 20
3. Materials and Methods
3.1. Bacterial Strains, Plasmids, and Chemicals
Escherichia coli BL21 (DE3) used in this study was purchased from Beinuo Biotech
(Shanghai, China). The plasmid pET-28a (+) used as an expression vector was purchased
from GenScript (Nanjing, China). The ginsenoside standards of Rb1, Rb2, Rc, Rd, Re, Rg1,
F2, C-K, C-Mc1, and C-Mc purchased from Chengdu Herbpurify (Chengdu, China) were
chromatographic grade. pNP-α-Af, pNP-α-Ap, pNP-α-Rp, pNP-β-Glc, gentiobiose, and
sophorose were purchased from Solarbio Science (Beijing, China). All other reagents were
analytical grade.
3.2. Cloning, Site-directed Mutagenesis, Heterologous Expression, and Protein Purification
The full open reading frame (ORF) of BsAbfA gene (GenBank accession: AL009126.3,
2938330-2939832) encoding 1515 bp was synthesized by GenScript (Nanjing, China) after
codon optimization [46]. The BsAbfA genes separately mutated at E173 and E292 were
synthesized by GenScript using a Site-directed Mutagenesis Kit (Nanjing, China). All
mutations were confirmed by DNA sequencing. The E. coli BL21 harboring BsAbfA or mu-
tated gene was cultivated at 200 rpm in Luria–Bertani (LB) medium containing 50
µ
g/mL
◦
kanamycin at 37 C for 8 h. The cultured bacterium was inoculated to the fresh LB medium
to reach OD600 = 0.4–0.6. To induce BsAbfA expression, 0.5 mM IPTG, as a final concen-
tration, was supplemented in LB medium. After the induction, the culture temperature
and agitation were reduced to 20 ◦C and 150 rpm, respectively, and the cells were further
incubated for 16 h. The induced cells were harvested by centrifugation (12,000
×
g, 10 min)
at 4 ◦C and stored at 20 ◦C for further use. Cells were disrupted by sonication with an
ultrasonic homogenizer in 50 mM citric acid/sodium citrate buffer (pH 5.5) with 1 g/L
lysozyme, as well as EDTA-free protease inhibitor cocktail and 2 mg/L DNase. The debris
were removed by centrifugation at 8000
×
g for 20 min at 4 ◦C. The resulting supernatants
m. The filtrate was loaded onto
were subjected to filtration through a membrane of 0.45
µ
Ni-NTA magnetic agarose beads (Qiagen, Germany) for the enrichment of the recombinant
BsAbfA protein carrying His-tag. The supernatant was removed by a magnetic separator
and washed at least twice with elution buffer. Purified proteins were maintained in 50 mM
◦
citric acid/sodium citrate buffer (pH 5) at 4 C. The concentration of purified recombi-
nant BsAbfA was assayed using Folin-phenol reagent [47]. The expression quantity and
molecular weight of the protein were analyzed by SDS-PAGE.
3.3. Determination of Kinetic Parameters and Substrate Specificity
In order to determine the enzyme properties of the recombinant BsAbfA, p-Nitrophenyl
α
-L-arabinofuranoside (pNP-α-Af) was used as a substrate to assay the enzymatic activi-
ties in 50 mM citric acid/sodium citrate buffer (pH 5) at 40 ◦C. The reaction was ceased
by adding 500 mM sodium carbonate with volume equal to that of the reaction. The
released p-nitrophenol (pNP) was immediately measured at 405 nm. One unit (U) of the
BsAbfA activity was defined as the amount of enzyme required to generate 1 µmol pNP
per minute [32]. When ginsenoside Rc was used as a substrate, the reaction was ceased by
adding n-butanol with a volume 2-fold that of the reaction. The products were assayed
by UPLC. The kinetic parameters of BsAbfA were measured using pNP-Af and Rc as sub-
strate at concentrations ranging from 0.1 to 5 mM. The Km, Kcat, and Vmax were calculated
by fitting the activity data to a linear regression on Lineweaver–Burk double-reciprocal
plots [47]. The substrate specificity of purified recombinant BsAbfA was assayed by using
pNP-α-Af, pNP-α-Ap, pNP-α-Rp, pNP-β-Glc, ginsenoside Rb1, Rb2, Rc, Rd, Re, Rg1, F2,
C-K, C-Mc1, C-Mc, gentiobiose, and sophorose as substrates, individually. All assays were
performed in triplicate.
3.4. Effects of pH, Temperature, Metal Ions, and Chemicals on Stability
The effects of pH, temperature, metal ions, and chemicals on BsAbfA activity were
investigated using pNP-Af as substrate. The buffers used were as follows: citric acid-

Molecules 2021, 26, 1733
17 of 20
sodium citrate buffer (50 mM, pH 3–6), sodium phosphate buffer (100 mM, pH 6–8), a◦nd
glycine-NaOH (50 mM, pH 9–10). The temperature was set ranging from 25 to 70 C.
The relative activity of BsAbfA was defined as 100% at either the optimal pH or optimal
temperature. The effects of metal ions and chemicals on the activity of BsAbfA were
assessed in the presence of NaCl, KCl, CaCl2, MgCl2, FeCl2, MnCl2, ZnCl2, NiCl2, CuCl2,
HgCl2, EDTA, DDT, and SDS at optimal pH buffer and temperature. All assays were
performed in triplicate.
3.5. Biotransformation of Ginsenoside Rc
To investigate the catalytic ability of recombinant BsAbfA for the biotransformation of
ginsenoside Rc, as well as to disclose the reaction pathway, ginsenoside Rc was dissolved
in methanol and incubated in 50 mM citric acid/sodium citrate buffer (pH 5) containing
◦
5 mM Rc and 12 U/mL enzyme at 40 C. The reaction was sampled at regular intervals for
◦
a certain period and ceased by heating the mixture to 80 C for 15 min. The biotransformed
products were subsequently extracted with H2O-saturated n-butanol. After evaporation of
solvents, the products were dissolved in methanol and then subjected to filtration using
0.45
µm microfiltration membranes [48]. UPLC analysis was performed with Shimadzu
LC-MS 8050 with a ACQUITY UPLC BEH Shield RP18 column (1.7
µ
m, 2.1 mm 50 mm).
×
The mobile phase consisted of acetonitrile (A) and 1% formic acid (B) and the elute program
was as follows: A:B (10:90) to (25:75) for 2 min; A:B (25:75) for 2-8 min; A:B (25:75) to (45:55)
for 8-16.5 min; A:B (45:55) for 16.5-21.5 min; A:B (45:55) to (98:2) for 21.5-21.6 min; A:B (98:2)
for 21.6-25 min; A:B (98:2) to (10:90) for 25-25.1 min A:B (10:90) for 25.1-29 min. Rc, Rd, Re,
F2, and C-K were used as standards. The samples were analyzed using a triple quadrupole
mass spectrometer equipped with electrospray ionization source. LC-MS analyses were
acquired in the negative ion mode by full scan. High purity nitrogen was used as a drying
gas, the nebulizing gas flow and heating gas flow were 2 L/min and 10 L/min at 4000 V of
ionspray voltage, respectively. The atomizing temperature was 300 ◦C [49].
3.6. Homology Modeling and Molecular Docking
3.6.1. Homology Modeling
The template crystal structure for BsAbfA was identified through BLAST [50] and
downloaded from the RCSB Protein Data Bank (PDB code: 5O7Z) [51]. Homology modeling
was conducted in the Swiss-model [52]. The target sequence was searched with BLAST
against the primary amino acid sequence contained in the SMTL [53]. Models were built
based on a target–template alignment using ProMod3 [54]. Coordinates which were
conserved between the target and the template were copied from the template to the model.
Insertions and deletions were remodeled using a fragment library. Side chains were then
rebuilt. Finally, the geometry of the resulting model was regularized by using a force field.
3.6.2. Molecular Docking
AutoDock Vina [55] was employed for molecular docking of BsAbfA with candidate
substrate ginsenoside Rc for the prediction of binding affinity and binding sites. The
structure of ginsenoside Rc was acquired from PubChem [56] (ID:12855889) and converted
to 3D by Avogadro [57] through energy minimization. Then, a conformation search was
performed to confirm the stable geometry for the docking preprocessing. As reported
in [21], the binding site was located in a beta sheet barrel surrounded by an alpha helix. The
three glutamine residues were participants of the hydrolase reaction with a docking box
defined at the center of a barrel (50
were ranked by binding energy, and the threshold of cluster analysis was set at 5 angstroms.
Other parameters not mentioned were set as default.
×
50
×
50 in size). All docked poses of ginsenoside Rc
4. Conclusions
The ginsenoside Rc-hydrolyzing
α-L-arabinofuranosidase gene BsAbfA was cloned
from B. subtilis, and the optimized recombinant protein was overexpressed and character-

Molecules 2021, 26, 1733
18 of 20
ized successfully in E. coli. The enzymatic properties of BsAbfA are unique and superior
to the other reported α-L-arabinofuranosidases, exhibiting higher catalytic efficiency and
higher tolerance to metal ions, as well as to organic solvents and detergents. Additionally,
BsAbfA shows a high selectivity to cleave the outer arabinofuranosyl moieties at C20 of
ginsenoside Rc, catalyzing the conversion of ginsenoside Rc to the more pharmacologically
active Rd with high productivity. The 3D structure of BsAbfA of family 51, glycosidase
modeled by comparative modeling was compact and stable. The docking results revealed
that the active site of BsAbfA can accommodate ginsenoside Rc very well. Site-directed
mutagenesis of the E173 and E292 residues confirms that it is important to recognize
ginsenoside Rc effectively and make it enter the active pocket for the hydrolysis of the
outer arabinofuranose moieties at C20 position. Thus, we report the BsAbfA as a promising
enzyme, and efficient for the industrial production of ginsenoside Rd.
Author Contributions: Conceptualization, R.Z.; Methodology, S.Q.T., B.L.Z., Z.Y.G. and Z.Y.L.;
Validation, S.Q.T., L.Y.T. and R.Z.; Data Analysis: R.Z. and B.L.Z.; Investigation, S.Q.T. B.L.Z.,
Z.Y.G., L.Y.T. and P.W.; Writing—review and editing: R.Z. and B.L.Z.; Supervision: R.Z. and B.L.Z.;
Funding acquisition, R.Z. and Z.Y.L.; All authors have read and agreed to the published version of
the manuscript.
Funding: The study was supported by the National Natural Science Foundation of China (81874332,
81973710), the Natural Science Foundation of Hunan Province (2020JJ2012), the Scientific Research
Fund of Hunan Provincial Education Department (18B387), and the Postdoctoral Science Foundation
of China (2017T100601, 2016M590746).
Data Availability Statement: All data supporting the findings of this study are available in the
main text.
Acknowledgments: The critical reading of the manuscript by C.T. Au, College of Materials and
Chemical Engineering, Hunan Institute of Engineering is greatly appreciated.
Conflicts of Interest: The authors declare no conflict of interest.
Sample Availability: Samples of the compounds are available from the corresponding authors.
References
1.
2.
3.
Kim, J.H.; Oh, J.M.; Chun, S.; Park, H.Y.; Im, W.T. Enzymatic biotransformation of ginsenoside Rb2 into Rd by recombinant
alpha-L-arabinopyranosidase from Blastococcus saxobsidens. J. Microbiol. Biotechnol. 2020, 30, 391–397. [CrossRef] [PubMed]
Fan, S.; Zhang, Z.; Su, H.; Xu, P.; Qi, H.; Zhao, D.; Li, X. Panax ginseng clinical trials: Current status and future perspectives.
Biomed. Pharmacother. 2020, 132, 110832. [CrossRef] [PubMed]
Lee, J.O.; Choi, E.; Shin, K.K.; Hong, Y.H.; Kim, H.G.; Jeong, D.; Hossain, M.A.; Kim, H.S.; Yi, Y.S.; Kim, D.; et al. Compound K, a
ginsenoside metabolite, plays an antiinflammatory role in macrophages by targeting the AKT1-mediated signaling pathway. J.
Ginseng Res. 2019, 43, 154–160. [CrossRef] [PubMed]
4.
5.
6.
7.
8.
Xue, Q.; He, N.; Wang, Z.; Fu, X.; Aung, L.H.H.; Liu, Y.; Li, M.; Cho, J.Y.; Yang, Y.; Yu, T. Functional roles and mechanisms of
ginsenosides from Panax ginseng in atherosclerosis. J. Ginseng Res. 2021, 45, 22–31. [CrossRef] [PubMed]
Shin, B.K.; Kwon, S.W.; Park, J.H. Chemical diversity of ginseng saponins from Panax ginseng. J. Ginseng Res. 2015, 39, 287–298.
[CrossRef] [PubMed]
Chen, W.; Balan, P.; Popovich, D.G. Ginsenosides analysis of New Zealand-grown forest Panax ginseng by LC-QTOF-MS/MS. J.
Ginseng Res. 2020, 44, 552–562. [CrossRef] [PubMed]
Sun, B.S.; Gu, L.J.; Fang, Z.M.; Wang, C.Y.; Wang, Z.; Lee, M.R.; Li, Z.; Li, J.J.; Sung, C.K. Simultaneous quantification of 19
ginsenosides in black ginseng developed from Panax ginseng by HPLC-ELSD. J. Pharm. Biomed. Anal. 2009, 50, 15–22. [CrossRef]
Zhang, F.; Tang, S.; Zhao, L.; Yang, X.; Yao, Y.; Hou, Z.; Xue, P. Stem-leaves of Panax as a rich and sustainable source of less-polar
ginsenosides: Comparison of ginsenosides from Panax ginseng, American ginseng and Panax notoginseng prepared by heating and
acid treatment. J. Ginseng Res. 2021, 45, 163–175. [CrossRef]
9.
Liu, G.M.; Lu, T.C.; Sun, M.L.; Jia, W.Y.; Ji, X.; Luo, Y.G. Ginsenoside Rd inhibits glioblastoma cell proliferation by up-regulating
the expression of miR-144-5p. Biol. Pharm. Bull. 2020, 43, 1534–1541. [CrossRef]
10. Wang, P.; Du, X.; Xiong, M.; Cui, J.; Yang, Q.; Wang, W.; Chen, Y.; Zhang, T. Ginsenoside Rd attenuates breast cancer metastasis
implicating derepressing microRNA-18a-regulated Smad2 expression. Sci. Rep. 2016, 6, 33709. [CrossRef]
11. Yang, N.; Liang, G.; Lin, J.; Zhang, S.; Lin, Q.; Ji, X.; Chen, H.; Li, N.; Jin, S. Ginsenoside Rd therapy improves histological and
functional recovery in a rat model of inflammatory bowel disease. Phytother. Res. 2020, 34, 3019–3028. [CrossRef] [PubMed]

Molecules 2021, 26, 1733
19 of 20
12. Zhang, X.; Liu, X.; Hu, G.; Zhang, G.; Zhao, G.; Shi, M. Ginsenoside Rd attenuates blood-brain barrier damage by suppressing
proteasome-mediated signaling after transient forebrain ischemia. Neuroreport 2020, 31, 466–472. [CrossRef] [PubMed]
13. Ye, R.; Zhang, X.; Kong, X.; Han, J.; Yang, Q.; Zhang, Y.; Chen, Y.; Li, P.; Liu, J.; Shi, M.; et al. Ginsenoside Rd attenuates
mitochondrial dysfunction and sequential apoptosis after transient focal ischemia. Neuroscience 2011, 178, 169–180. [CrossRef]
[PubMed]
14. Anufriev, V.P.; Malinovskaya, G.V.; Denisenko, V.A.; Uvarova, N.I.; Elyakov, G.B.; Kim, S.I.; Baek, N.I. Synthesis of ginsenoside
Rg3, a minor constituent of Ginseng radix. Carbohydr. Res. 1997, 304, 179–182. [CrossRef]
15. An, D.S.; Cui, C.H.; Sung, B.H.; Yang, H.C.; Kim, S.C.; Lee, S.T.; Im, W.T.; Kim, S.G. Characterization of a novel ginsenoside-
hydrolyzing alpha-L-arabinofuranosidase, AbfA, from Rhodanobacter ginsenosidimutans Gsoil 3054T. Appl. Microbiol. Biotechnol.
2012, 94, 673–682. [CrossRef] [PubMed]
16. Liu, Q.M.; Jung, H.M.; Cui, C.H.; Sung, B.H.; Kim, J.K.; Kim, S.G.; Lee, S.T.; Kim, S.C.; Im, W.T. Bioconversion of ginsenoside
Rc into Rd by a novel alpha-L-arabinofuranosidase, Abf22-3 from Leuconostoc sp. 22-3: Cloning, expression, and enzyme
characterization. Antonie Leeuwenhoek 2013, 103, 747–754.
17. Renchinkhand, G.; Cho, S.H.; Park, Y.W.; Song, G.Y.; Nam, M.S. Biotransformation of major ginsenoside Rb1 to Rd by Dekkera
anomala YAE-1 from Mongolian fermented milk (Airag). J. Microbiol. Biotechnol. 2020, 30, 1536–1542. [CrossRef]
18. Zhong, F.L.; Ma, R.; Jiang, M.; Dong, W.W.; Jiang, J.; Wu, S.; Li, D.; Quan, L.H. Cloning and characterization of ginsenoside-
hydrolyzing beta-glucosidase from Lactobacillus brevis that transforms ginsenosides Rb1 and F2 into ginsenoside Rd and compound
K. J. Microbiol. Biotechnol. 2016, 26, 1661–1667. [CrossRef] [PubMed]
19. Wang, C.Z.; Zhang, B.; Song, W.X.; Wang, A.; Ni, M.; Luo, X.; Aung, H.H.; Xie, J.T.; Tong, R.; He, T.C.; et al. Steamed American
ginseng berry: Ginsenoside analyses and anticancer activities. J. Agric. Food Chem. 2006, 54, 9936–9942. [CrossRef]
20. Kim, T.H.; Yang, E.J.; Shin, K.C.; Hwang, K.H.; Park, J.S.; Oh, D.K. Enhanced production of
β-D-glycosidase and α-L-
arabinofuranosidase in recombinant Escherichia coli in fed-batch culture for the biotransformation of ginseng leaf extract to
ginsenoside compound K. Biotechnol. Bioproc. E 2018, 23, 183–193. [CrossRef]
21. Taylor, E.J.; Smith, N.L.; Turkenburg, J.P.; D’Souza, S.; Gilbert, H.J.; Davies, G.J. Structural insight into the ligand specificity of
a thermostable family 51 arabinofuranosidase, Araf51, from Clostridium thermocellum. Biochem. J. 2006, 395, 31–37. [CrossRef]
[PubMed]
22. Debeche, T.; Bliard, C.; Debeire, P.; O’Donohue, M.J. Probing the catalytically essential residues of the alpha-L-arabinofuranosidase
from Thermobacillus xylanilyticus. Protein Eng. 2002, 15, 21–28. [CrossRef]
23. Shin, H.Y.; Park, S.Y.; Sung, J.H.; Kim, D.H. Purification and characterization of alpha-L-arabinopyranosidase and alpha-L-
arabinofuranosidase from Bifidobacterium breve K-110, a human intestinal anaerobic bacterium metabolizing ginsenoside Rb2 and
Rc. Appl. Environ. Microbiol. 2003, 69, 7116–7123. [CrossRef] [PubMed]
24. Shin, K.C.; Lee, G.W.; Oh, D.K. Production of ginsenoside Rd from ginsenoside Rc by alpha-L-arabinofuranosidase from
Caldicellulosiruptor saccharolyticus. J. Microbiol. Biotechnol. 2013, 23, 483–488. [CrossRef] [PubMed]
25. Choi, J.H.; Shin, K.C.; Oh, D.K. An L213A variant of beta-glycosidase from Sulfolobus solfataricus with increased alpha-L-
arabinofuranosidase activity converts ginsenoside Rc to compound K. PLoS ONE 2018, 13, e0191018.
26. Xie, J.; Zhao, D.; Zhao, L.; Pei, J.; Xiao, W.; Ding, G.; Wang, Z.; Xu, J. Characterization of a novel arabinose-tolerant alpha-L-
arabinofuranosidase with high ginsenoside Rc to ginsenoside Rd bioconversion productivity. J. Appl. Microbiol. 2016, 120, 647–660.
[CrossRef]
27. Canakci, S.; Belduz, A.O.; Saha, B.C.; Yasar, A.; Ayaz, F.A.; Yayli, N. Purification and characterization of a highly thermostable
alpha-L-arabinofuranosidase from Geobacillus caldoxylolyticus TK4. Appl. Microbiol. Biotechnol. 2007, 75, 813–820. [CrossRef]
28. Pitson, S.M.; Voragen, A.G.; Beldman, G. Stereochemical course of hydrolysis catalyzed by arabinofuranosyl hydrolases. FEBS
Lett. 1996, 398, 7–11. [CrossRef]
29. Pazos, F.; Heredia, P.; Valencia, A.; de las Rivas, J. Threading structural model of the manganese-stabilizing protein PsbO reveals
presence of two possible beta-sandwich domains. Protein Struct. Funct. Genet. 2001, 45, 372–381. [CrossRef]
30. Shallom, D.; Belakhov, V.; Solomon, D.; Shoham, G.; Baasov, T.; Shoham, Y. Detailed kinetic analysis and identification of the
nucleophile in alpha-L-arabinofuranosidase from Geobacillus stearothermophilus T-6, a family 51 glycoside hydrolase. J. Biol. Chem.
2002, 277, 43667–43673. [CrossRef] [PubMed]
31. Shallom, D.; Belakhov, V.; Solomon, D.; Gilead-Gropper, S.; Baasov, T.; Shoham, G.; Shoham, Y. The identification of the acid-base
catalyst of alpha-arabinofuranosidase from Geobacillus stearothermophilus T-6, a family 51 glycoside hydrolase. FEBS Lett. 2002
514, 163–167. [CrossRef]
,
32. Lim, Y.R.; Yoon, R.Y.; Seo, E.S.; Kim, Y.S.; Park, C.S.; Oh, D.K. Hydrolytic properties of a thermostable alpha-L-arabinofuranosidase
from Caldicellulosiruptor saccharolyticus. J. Appl. Microbiol. 2010, 109, 1188–1197. [CrossRef]
33. Zuo, S.S.; Wang, Y.C.; Zhu, L.; Zhao, J.Y.; Li, M.G.; Han, X.L.; Wen, M.L. Cloning and characterization of a ginsenoside-hydrolyzing
alpha-L-arabinofuranosidase, CaAraf51, from Cellulosimicrobium aquatile Lyp51. Curr. Microbiol. 2020, 77, 2783–2791. [CrossRef]
34. Zverlov, V.V.; Liebl, W.; Bachleitner, M.; Schwarz, W.H. Nucleotide sequence of arfB of Clostridium stercorarium, and prediction of
catalytic residues of alpha-L-arabinofuranosidases based on local similarity with several families of glycosyl hydrolases. FEMS
Microbiol. Lett. 1998, 164, 337–343.

Molecules 2021, 26, 1733
20 of 20
35. Quan, L.H.; Piao, J.Y.; Min, J.W.; Kim, H.B.; Kim, S.R.; Yang, D.U.; Yang, D.C. Biotransformation of ginsenoside Rb1 to
prosapogenins, gypenoside XVII, ginsenoside Rd, ginsenoside F2, and compound K by Leuconostoc mesenteroides DC102. J. Ginseng
Res. 2011, 35, 344–351. [CrossRef]
36. Zhong, F.L.; Dong, W.W.; Wu, S.; Jiang, J.; Yang, D.C.; Li, D.; Quan, L.H. Biotransformation of gypenoside XVII to compound K by
a recombinant beta-glucosidase. Biotechnol. Lett. 2016, 38, 1187–1193. [CrossRef]
37. Shin, H.Y.; Lee, J.H.; Lee, J.Y.; Han, Y.O.; Han, M.J.; Kim, D.H. Purification and characterization of ginsenoside Ra-hydrolyzing
beta-D-xylosidase from Bifidobacterium breve K-110, a human intestinal anaerobic bacterium. Biol. Pharm. Bull. 2003, 26, 1170–1173.
[CrossRef] [PubMed]
38. Zechel, D.L.; Withers, S.G. Glycosidase mechanisms: Anatomy of a finely tuned catalyst. Acc. Chem. Res. 2000, 33, 11–18.
[PubMed]
39. Moracci, M.; Capalbo, L.; Ciaramella, M.; Rossi, M. Identification of two glutamic acid residues essential for catalysis in the
beta-glycosidase from the thermoacidophilic archaeon Sulfolobus solfataricus. Protein Eng. 1996, 9, 1191–1195. [CrossRef]
40. Wang, Q.; Trimbur, D.; Graham, R.; Warren, R.A.; Withers, S.G. Identification of the acid/base catalyst in Agrobacterium faecalis
beta-glucosidase by kinetic analysis of mutants. Biochemistry (U. S.) 1995, 34, 14554–14562. [CrossRef] [PubMed]
41. Moracci, M.; Trincone, A.; Perugino, G.; Ciaramella, M.; Rossi, M. Restoration of the activity of active-site mutants of the hyper-
thermophilic beta-glycosidase from Sulfolobus solfataricus: Dependence of the mechanism on the action of external nucleophiles.
Biochemistry (U. S.) 1998, 37, 17262–17270. [CrossRef]
42. Sun, J.; Wang, W.; Yao, C.; Dai, F.; Zhu, X.; Liu, J.; Hao, J. Overexpression and characterization of a novel cold-adapted and
salt-tolerant GH1 beta-glucosidase from the marine bacterium Alteromonas sp. L82. J. Microbiol. 2018, 56, 656–664. [CrossRef]
[PubMed]
43. Noh, K.H.; Oh, D.K. Production of the rare ginsenosides compound K, compound Y, and compound Mc by a thermostable
beta-glycosidase from Sulfolobus acidocaldarius. Biol. Pharm. Bull. 2009, 32, 1830–1835. [CrossRef]
44. Cao, H.; Nuruzzaman, M.; Xiu, H.; Huang, J.; Wu, K.; Chen, X.; Li, J.; Wang, L.; Jeong, J.H.; Park, S.J.; et al. Transcriptome analysis
of methyl jasmonate-elicited Panax ginseng adventitious roots to discover putative ginsenoside biosynthesis and transport genes.
Int. J. Mol. Sci. 2015, 16, 3035–3057. [CrossRef]
45. Upadhyaya, J.; Yoon, M.S.; Kim, M.J.; Ryu, N.S.; Song, Y.E.; Kim, Y.H.; Kim, M.K. Purification and characterization of a novel
ginsenoside Rc-hydrolyzing beta-glucosidase from Armillaria mellea mycelia. AMB Express. 2016, 6, 112. [CrossRef] [PubMed]
46. Morales-Quintana, L.; Beltran, D.; Mendez-Yanez, A.; Valenzuela-Riffo, F.; Herrera, R.; Moya-Leon, M.A. Characterization of
FcXTH2, a novel xyloglucan endotransglycosylase/hydrolase enzyme of Chilean strawberry with hydrolase activity. Int. J. Mol.
Sci. 2020, 21, 3380. [CrossRef]
47. Zhang, R.; Huang, X.M.; Yan, H.J.; Liu, X.Y.; Zhou, Q.; Luo, Z.Y.; Tan, X.N.; Zhang, B.L. Highly selective production of compound
K from ginsenoside Rd by hydrolyzing glucose at C-3 glycoside using beta-glucosidase of Bifidobacterium breve ATCC 15700. J.
Microbiol. Biotechn. 2019, 29, 410–418. [CrossRef]
48. Ji, Q.C.; Harkey, M.R.; Henderson, G.L.; Gershwin, M.E.; Stern, J.S.; Hackman, R.M. Quantitative determination of ginsenosides by
high-performance liquid chromatography-tandem mass spectrometry. Phytochem. Anal. 2001, 12, 320–326. [CrossRef] [PubMed]
49. Liang, W.; Wang, S.; Yao, L.; Wang, J.; Gao, W. Quality evaluation of Panax ginseng adventitious roots based on ginsenoside
constituents, functional genes, and ferric-reducing antioxidant power. J. Food Biochem. 2019, 43, e12901. [CrossRef]
50. Camacho, C.; Coulouris, G.; Avagyan, V.; Ma, N.; Papadopoulos, J.; Bealer, K.; Madden, T.L. BLAST+: Architecture and
applications. BMC Bioinform. 2009, 10, 421. [CrossRef] [PubMed]
51. Berman, H.M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T.N.; Weissig, H.; Shindyalov, I.N.; Bourne, P.E. The Protein Data Bank.
Nucleic Acids Res. 2000, 28, 235–242. [CrossRef] [PubMed]
52. Waterhouse, A.; Bertoni, M.; Bienert, S.; Studer, G.; Tauriello, G.; Gumienny, R.; Heer, F.T.; de Beer, T.A.P.; Rempfer, C.; Bordoli,
L.; et al. SWISS-MODEL: Homology modelling of protein structures and complexes. Nucleic Acids Res. 2018, 46, W296–W303.
[CrossRef] [PubMed]
53. Bienert, S.; Waterhouse, A.; de Beer, T.A.; Tauriello, G.; Studer, G.; Bordoli, L.; Schwede, T. The SWISS-MODEL Repository-new
features and functionality. Nucleic Acids Res. 2017, 45, D313–D319. [CrossRef] [PubMed]
54. Guex, N.; Peitsch, M.C.; Schwede, T. Automated comparative protein structure modeling with SWISS-MODEL and Swiss-
PdbViewer: A historical perspective. Electrophoresis 2009, 30 (Suppl. S1), S162–S173. [CrossRef]
55. Trott, O.; Olson, A.J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient
optimization, and multithreading. J. Comput. Chem. 2009, 31, 455–461. [CrossRef]
56. Kim, S.; Chen, J.; Cheng, T.; Gindulyte, A.; He, J.; He, S.; Li, Q.; Shoemaker, B.A.; Thiessen, P.A.; Yu, B.; et al. PubChem 2019
update: Improved access to chemical data. Nucleic Acids Res. 2019, 47, D1102–D1109. [CrossRef]
57. Hanwell, M.D.; Curtis, D.E.; Lonie, D.C.; Vandermeersch, T.; Zurek, E.; Hutchison, G.R. Avogadro: An advanced semantic
chemical editor, visualization, and analysis platform. J. Cheminform. 2012, 4, 17. [CrossRef] [PubMed]