Probing the stereoselectivity of OleD-catalyzed glycosylation of cardiotonic steroids?

Article abstract of DOI:10.1039/c7ra11979h

The glycosyltransferase OleD variant as a catalyst for the glycosylation of four pairs of epimers of cardiotonic steroids (CTS) are assessed. The results of this study demonstrated that the OleD-catalyze glycosylation of CTS is significantly influenced by

Full text of DOI:10.1039/c7ra11979h

RSC Advances
View Article Online
View Journal | View Issue
PAPER
Probing the stereoselectivity of OleD-catalyzed
glycosylation of cardiotonic steroids†
Cite this: RSC Adv., 2018, 8, 5071
Xue-Lin Zhu, ‡a Chao Wen,‡a Qing-Mei Ye,‡b Wei Xu,a Deng-Lang Zou,a
a
a
a
a
*
Guang-Ping Liang, Fan Zhang, Wan-Na Chen and Ren-Wang Jiang
The glycosyltransferase OleD variant as a catalyst for the glycosylation of four pairs of epimers of
cardiotonic steroids (CTS) are assessed. The results of this study demonstrated that the OleD-catalyze
glycosylation of CTS is significantly influenced by the configuration at C-3 and the A/B fusion mode. 3b-
OH and A/B ring cis fusion are favoured by OleD (ASP). An epoxide ring at C-14 and C-15 further
increases the bioconversion rate; while an acetyl group at C-16 and lactone ring type at C-17 did not
influence the biotransformation. A high conversion rate corresponded to a low Km value. A molecular
docking simulation showed that filling of hydrophobic pocket II and interaction with residue Tyr115 may
play an important role in the glycosylation reactions catalyzed by OleD glycosyltransferases.
Furthermore, the glycosylation products showed a stronger inhibitory activity for Na+, K+-ATPase than
the corresponding aglycones. This study provides the first stereoselective properties for OleD (ASP)
catalyzed glycosylation.
Received 1st November 2017
Accepted 16th January 2018
DOI: 10.1039/c7ra11979h
rsc.li/rsc-advances
bufadienolides. Cardenolides, such as ouabain and digoxin,
possess a ve-membered lactone ring at position C-17b of the
Introduction
steroidal skeleton. The natural cardenolide can be aglycone or
glycosides with one or more sugar groups at C-3.16,17 Bufadie-
nolides, isolated from many plants and animals such as Bufo
bufo gargarizans, bears a six-membered lactone ring at C-17b.
The bufadienolides can also be aglycone or glycosides in plants
but only aglycone in animals.18–21 For both bufadienolides and
cardenolides, there is a hydroxy group at C-3, either b- or a-
conguration. Normally the CTS with a 3b-OH shows more
remarkable activities than the 3a-diastereomer.22 Especially, the
sugar groups of C-3 of CTS play a key role for the inhibitory
activity of NKA.23
In this paper, we use OleD (ASP) to catalyze the glycosylation
of four pairs of epimers of cardiotonic steroids. Km values of two
pairs of epimers representing the bufadienolides and carde-
nolides were determined. Molecular docking was used to
simulate the interactions between the substrate and enzyme.
Finally, the inhibitory activities against NKA of CTS aglycone
and the corresponding glycosides were compared.
The glycosyltransferase OleD from Streptomyces antibioticus
catalyzes the glycosylation of oleandomycin using UDP-D-
glucose (UDPG) as the glycosyl donor.1–4 Recent studies revealed
an enhanced triple mutant (A242V/S132F/P67T, OleD (ASP)) that
displayed marked improvement in prociency and substrate
promiscuity. The OleD (ASP) was found to show highly
permissive properties and is capable of glycosylation on 100
diverse acceptors, e.g. erythromycin,5 avones,6 indole alka-
loids7 and steroids.8 Nowadays, 56% of the drugs in clinics are
chiral compounds; however 88% of them are marketed as
racemates consisting of an equimolar mixture of two enantio-
mers.8 Although OleD (ASP) was found to catalyze the glycosyl-
ation of cardenolide and bufadienolide aglycone with a bias
toward the desired C-3 regiospecicity,7 the stereospecicity of
this enzyme remains unreported.
Cardiotonic steroids (CTS)9,10 such as digitoxin, digoxin and
proscillaridin are clinically important drugs for treatment of
congestive heart failure for more than two centuries due to their
potent inhibition of Na+, K+-ATPase (NKA), which is an integral
membrane protein maintaining ionic gradients in all superior
eukaryotic cells.11–15 Natural CTS includes cardenolides and
Experimental section
General experimental procedures
aGuangdong Province Key Laboratory of Pharmacodynamic Constituents of TCM and
New Drugs Research, College of Pharmacy, Jinan University, Guangzhou 510632, P.
R. China. E-mail: trwjiang@jnu.edu.cn
bDepartment of Pharmacy, Hainan General Hospital, Haikou 570311, P. R. China
† Electronic supplementary information (ESI) available: Spectroscopic data (NMR,
MS) for compounds. See DOI: 10.1039/c7ra11979h
All chemicals and reagents were purchased from Sigma unless
otherwise stated. The NMR spectra were recorded on a Bruker
AV-400 spectrophotometer (Bruker, Germany) with TMS as
internal standard. Chemical shis (d) were expressed in ppm
with reference to the solvent signals. HR-ESI-MS were deter-
mined on a Micromass Q-TOF mass spectrometer (Waters,
‡ These authors contribute equally to this work.
This journal is © The Royal Society of Chemistry 2018
RSC Adv., 2018, 8, 5071–5078 | 5071

View Article Online
RSC Advances
Paper
USA). Analytical HPLC is run on Agilent 1200 system (Agilent,
1a: 1H NMR (CD3OD, 400 MHz) d: 8.01 (1H, dd, J ¼ 9.7,
USA) with a Phenomenex Luna C-18 column (250 mm ꢀ 2.5 Hz, H-22), 7.45 (1H, d, J ¼ 2.4 Hz, H-21), 6.30 (1H, d, J ¼
4.6 mm, 5 mm, USA). Preparative HPLC was performed on a Wu 9.8 Hz, H-23), 3.58 (1H, m, H-3), 2.58 (1H, m, H-17), 2.26–2.11
Feng HPLC system (Shanghai, China) equipped with a prepara- (2H, m), 1.92–1.62 (11H, m), 1.55–1.36 (9H, m), 1.29–1.19 (1H,
tive reverse phase column (20 ꢀ 250 mm, 5 mm).
m), 1.09–1.01 (2H, m), 0.95 (3H, s, H-19), 0.73 (3H, s, H-18); 13
C
NMR (CD3OD, 100 MHz) d: 164.8 (C-24), 150.5 (C-21), 149.37 (C-
22), 125.0 (C-20), 115.4 (C-23), 86.1 (C-14), 72.3 (C-3), 52.3 (C-17),
49.8 (C-13), 43.2 (C-9), 43.1 (C-5), 41.9 (C-12), 37.7 (C-8), 37.0 (C-
4), 36.3 (C-1), 35.9 (C-10), 33.2 (C-15), 31.3 (C-2), 29.9 (C-16), 28.4
Cloning and expression of variant OleD (ASP)
glycosyltransferase
The OleD (ASP) glycosyltransferase gene was generated by (C-6), 23.8 (C-19), 22.8 (C-7), 22.5 (C-11), 17.3 (C-18); HR-ESI-MS:
Genscript Biotechnology (Nanjing, China) and was cloned into m/z 387.2576 [M + H]+ (calcd for C24H34O4, 387.2534).
DH5a (Takara, Japan) and pET28a (Novagen, USA) expression
2a: 1H NMR (CD3OD, 400 MHz) d: 7.91 (1H, dd, J ¼ 9.8,
vector. Single colony of Escherichia coli BL21 (DE3) pLysS 2.4 Hz, H-22), 7.47 (1H, d, J ¼ 2.4 Hz, H-21), 6.28 (1H d, J ¼
(Tiangen, China) transformed with pET28a/OleD vector was 9.8 Hz, H-23), 3.62 (1H, s, H-15), 3.57 (1H, m, H-3), 2.61 (1H, d, J
inoculated in Luria Bertani (LB) medium (3 mL) supplemented ¼ 10.0 Hz, H-10), 2.44 (1H, d, J ¼ 10.4 Hz, H-16), 2.04–1.63 (9H,
with 50 mg mLꢁ1 kanamycin. The medium was cultured over- m), 1.63–1.28 (9H, m), 0.99 (3H, s), 0.79 (3H, s); 13C NMR
night at 37 ꢂC with shaking (200 rpm). The entire starter culture (CD3OD, 100 MHz) d: 164.5 (C-24), 151.8 (C-21), 149.6 (C-22),
was then transferred to 1 L LB medium supplemented 124.5 (C-20), 115.3 (C-23), 75.8 (C-14), 72.2 (C-3), 61.1 (C-15),
with 50 mg mLꢁ1 kanamycin and grown at 37 ꢂC with 48.6 (C-17), 46.3 (C-13), 43.1 (C-5), 41.3 (C-9), 40.1 (C-12), 36.9
shaking (200 rpm) until the OD600 reached 0.6. Isopropyl (C-4), 36.1 (C-1), 36.0 (C-10), 35.2 (C-8), 33.2 (C-16), 31.3 (C-2),
b-D-thiogalactoside (IPTG) was subsequently added (nal 27.5 (C-6), 23.8 (C-19), 22.1 (C-11), 22.0 (C-7), 17.1 (C-18); HR-
concentration to 0.4 mM) and the culture was incubated at ESI-MS: m/z 385.2363 [M + H]+ (calcd for C24H32O4, 385.2407).
3a: 1H NMR (CD3OD, 400 MHz) d: 8.04 (1H, d, J ¼ 9.6 Hz, H-
ꢂ
18 C for 18 h. Then the cell pellets were collected by centrifu-
ꢂ
gation at 10 000g at 4 C for 20 min and the supernatant was 22), 7.39 (1H, s, H-21), 6.26 (1H, d, J ¼ 9.6 Hz, H-23), 5.51 (1H, d,
discarded.24 Pellets were resuspended in 10 mL PBS buffer J ¼ 10.3 Hz, H-16), 3.76 (1H, s, H-15), 3.58 (1H, m, H-3), 2.95 (1H,
(20 mM phosphate buffer, 0.2 M NaCl, 2 mM KCl, pH 7.4) and d, J ¼ 9.3 Hz, H-17), 2.20–2.01 (1H, td, H-8), 1.81 (3H, s, COCH3),
then lysed by sonication. Cell debris was removed by centrifu- 1.79–1.04 (15H, m), 0.99 (3H, s, H-19), 0.83 (3H, s, H-18); 13C
ꢂ
gation at 10 000g at 4 C for 20 min and the clear supernatant NMR (CD3OD, 100 MHz) d: 171.6 (COCH3), 164.0 (C-24), 153.5
was immediately submitted to His60 Ni super ow resin (C-22), 150.9 (C-21), 118.4 (C-20), 114.1 (C-23), 76.6 (C-16), 73.4
(Clontech, USA). The resin was balanced by the equilibration (C-14), 72.2 (C-3), 60.8 (C-15), 51.4 (C-17), 46.3 (C-13), 43.0 (C-5),
buffer (50 mM sodium phosphate buffer, 0.3 M NaCl, 20 mM 41.1 (C-9), 40.7 (C-12), 36.9 (C-4), 36.1 (C-1), 36.0 (C-10), 34.7 (C-
imidazole, pH 7.4). The enzyme was allowed to bind for 1 h at 8), 31.2 (C-2), 27.4 (C-6), 23.7 (C-19), 21.9 (C-7), 21.9 (C-11), 20.4
4 ꢂC with gentle agitation, and the resin was washed with 10 mL (COCH3), 17.5 (C-18); HR-ESI-MS: m/z 443.2445 [M + H]+ (calcd
mild buffer (20 mM phosphate buffer, 0.5 M NaCl, 50 mM for C26H34O6, 443.2314).
imidazole, pH 7.4). Finally, the protein was eluted by with 20 mL
strong buffer (50 mM sodium phosphate buffer, 0.3 M NaCl,
General pilot-scale reaction
300 mM imidazole, pH 7.4). The enzyme aliquots were imme-
ꢂ
diately frozen in liquid nitrogen and stored at ꢁ80 C. Protein In vitro glycosylation reactions were carried out in 500 mL reac-
purity was conrmed by SDS-PAGE to be >95% and protein tion buffer (50 mM Tris–HCl, pH 8.0) containing 250 mg of
concentration for all studies was determined using the Bradford enzyme, 2.5 mM of UDPG, 1 mM aglycon and 5 mM MgCl2. The
ꢂ
Protein Assay Kit from Bio-Rad (TransGen Biotech, China).
mixture was incubated at 37 C for 16 h. The reaction mixture
was subsequently frozen and lyophilized, and the residue was
resuspended in 500 mL ice cold methanol and ltered. One
portion of each claried reaction mixture was analyzed by
Preparation of epimers of bufadienolides
CTS (1b, 2b and 3b, 10.7 mM) was dissolved in CH2Cl2 (35 mL) analytical reverse-phase HPLC equipped with a Phenomenex
in a round bottom ask. Pyridinium chloride hydrochloride Luna-C18 column (250 mm ꢀ 4.6 mm, 5 mm). The ow rate was
(PCC) (21.4 mM) was added and stirred constantly to dissolve. 1 mL minꢁ1. The gradient was consisted of solvent A (0.1%
The solution was stirred for 2 h at room temperature. Then the triuoroacetic acid/H2O) and B (100% acetonitrile): (a) 0–
solvent was removed under reduced pressure and the residue 20 min, 10–75% B; (b) 20–21 min, 75–100% B; (c) 21–26 min,
was dissolved in anhydrous tetrahydrofuran (5 mL) in a round 100% B; (d) 26–29 min, 100–10% B; and (e) 29–35 min, 10% B.
bottom ask. Sodium borohydride (NaBH4) was added and Detections were set at 220 and 296 nm. Conversion rate was
stirred constantly to dissolve. The solution was stirred for 1 h at calculated by the corresponding HPLC peak area percentage
room temperature. Then 5 mL of water was added slowly to the using the Agilent Chromatography Workstation Soware. The
reaction solution. Ethyl acetate (10 mL) was used to extract the LC-ESI-MS analysis was accomplished using standard C-18
mixture and the solvent was removed under reduced pressure. reversed-phase chromatography with diode array detection
The nal residue was puried by preparative HPLC eluting with wherein 5% of the ow was diverted to time-of-ight (TOF) mass
acetonitrile to yield C-3a CTS (1a, 2a, 3a).
spectrometer.
5072 | RSC Adv., 2018, 8, 5071–5078
This journal is © The Royal Society of Chemistry 2018

View Article Online
Paper
RSC Advances
(C-2), 27.4 (C-6), 23.7 (C-19), 21.9 (C-7), 21.9 (C-11), 20.4
(COCH3), 17.5 (C-18). HR-ESI-MS: m/z 605.2996 [M + H]+ (calcd
for C32H44O11, 605.2938).
Preparative scale glycosylation reaction
Aglycons (10 mg) were dissolved in 5% of DMSO and transferred
to pH 8 buffer solution (50 mM Tris–HCl, 5 mM MgCl2). UDPG
was added followed by OleD (ASP) catalyst. Aer 16 h incubation
at 37 ꢂC, the reaction was stopped with equal volume of ice cold
methanol. Then the reaction mixture was centrifuged at 10 000g
for 30 min and the supernatant was concentrated under
reduced pressure, and the debris was resuspended in 5 mL of
ice-cold methanol and ltered with 0.22 mM membrane. The
ltrate was subjected to preparative HPLC (20 ꢀ 250 mm, 5 mm;
ow rate: 5 mL minꢁ1; A296 and A220 detection) using water/
acetonitrile as the eluent to afford the corresponding of gluco-
sides. The compound was then characterized using high reso-
4b-glu. 1H NMR (CD3OD, 400 MHz) d: 5.93 (1H, s), 5.07 (1H,
d, J ¼ 20, 18.4 Hz), 4.95 (1H, d, J ¼ 19.8 Hz), 4.35 (1H, d, J ¼
7.8 Hz, sugar H-1), 4.11 (1H, s, H-3), 3.89 (1H, d, J ¼ 2.1 Hz), 3.69
(1H, dd, J ¼ 11.8, 5.4 Hz), 3.40–3.21 (5H, m), 2.95–2.79 (1H, m),
2.35–2.12 (2H, m), 2.00–1.43 (19H, m), 1.40–1.22 (4H, m), 1.00
(3H, s), 0.92 (3H, s); 13C NMR (CD3OD, 100 MHz) d: 178.5 (C-23),
177.3 (C-20), 117.8 (C-22), 102.7 (sugar C-1), 86.5 (C-14), 78.2
(sugar C-5), 77.8 (sugar C-2), 75.4 (sugar C-4), 75.3 (sugar C-3),
75.2 (C-31), 71.7 (C-3), 62.8 (sugar C-6), 52.2 (C-13), 51.1 (C-
17), 42.7 (C-9), 41.0 (C-8), 37.5 (C-5), 36.9 (C-12), 36.3 (C-10),
33.4 (C-4), 31.2 (C-15), 30.9 (C-1), 28.1 (C-2), 27.8 (C-6), 27.5
(C-16), 24.1 (C-7), 22.6 (C-7), 22.4 (C-18), 16.4 (C-19); HR-ESI-MS:
m/z 537.8049 [M + H]+ (calcd for C29H44O9, 537.8021).
1
1
lution MS, D and 2D NMR, including H, 13C and HSQC.
1
1b-glu. H NMR (CD3OD, 400 MHz) d: 8.03 (1H, dd, J ¼ 9.7,
2.3 Hz, H-22), 7.45 (1H, d, J ¼ 1.8 Hz, H-21), 6.30 (1H, d, J ¼
9.6 Hz, H-23), 4.37 (1H, d, J ¼ 7.8 Hz, sugar H-1), 4.12 (1H, m, H-
3), 3.88 (1H, d, J ¼ 11.9, 1.8 Hz), 3.71 (1H, dd, J ¼ 5.4 Hz), 3.40–
3.20 (4H, m), 2.56–2.52 (1H, m, H-17), 2.26–2.11 (2H, m), 1.92–
1.22 (23H, m), 1.09–1.01 (2H, m), 0.95 (3H, s, H-19), 0.73 (3H, s,
H-18); 13C NMR (CD3OD, 100 MHz) d: 164.8 (C-24), 150.5 (C-21),
149.37 (C-22), 125.0 (C-20), 115.4 (C-23), 102.7 (sugar C-1), 86.1
(C-14), 78.2 (sugar C-5), 77.84 (sugar C-2), 75.6 (sugar C-4), 75.2
(sugar C-3), 71.8 (C-3), 61.7 (sugar C-6), 52.3 (C-17), 49.8 (C-13),
43.2 (C-9), 43.1 (C-5), 41.9 (C-12), 37.7 (C-8), 37.0 (C-4), 36.3 (C-
1), 35.9 (C-10), 33.2 (C-15), 31.3 (C-2), 29.9 (C-16), 28.4 (C-6), 23.8
(C-19), 22.8 (C-7), 22.5 (C-11), 17.3 (C-18); HR-ESI-MS: m/z
549.3035 [M + H]+ (calcd for C30H44O9, 549.3021).
Determination of kinetic parameters
Assays were performed in a nal volume of 200 mL 50 mM Tris–
HCl (pH 8.0), and contained constant concentrations of OleD
(ASP) (40 mg) and UDPG (2.5 mM) while varying the concentra-
tion (0.01–1.2 mM) of 2b, 2a, 4b and 4a. Aliquots (100 mL) were
removed every 15 min, mixed with an equal volume of ice cold
methanol, and centrifuged at 10 000g for 10 min. Supernatants
were analyzed by analytical reverse-phase HPLC. Conversion
rate is calculated by the corresponding HPLC peak area
percentage using the Agilent Chromatography Workstation
Soware.25 All experiments were performed in triplicate. Initial
velocities were tted to the Michaelis–Menten equation using
Origin Pro 7.0 soware.
1
2b-glu. H NMR (CD3OD, 400 MHz) d: 7.91 (1H, dd, J ¼ 9.8,
2.4 Hz, H-22), 7.46 (1H, d, J ¼ 2.4 Hz, H-21), 6.27 (1H d, J ¼
9.8 Hz, H-23), 4.32 (1H, d, J ¼ 7.8 Hz, sugar H-1), 4.08 (1H, m, H-
3), 3.86 (1H, d, J ¼ 11.9 Hz), 3.67 (1H, dd, J ¼ 11.9, 5.4 Hz), 3.62
(1H, s, H-15), 3.39–3.17 (7H, m), 2.61 (1H, d, J ¼ 10.0 Hz, H-10),
2.44 (1H, d, J ¼ 10.4 Hz, H-16), 2.04–1.43 (15H, m), 1.40–1.05
(4H, m), 0.99 (3H, s), 0.79 (3H, s); 13C NMR (CD3OD, 100 MHz) d:
164.5 (C-24), 151.8 (C-21), 149.6 (C-22), 124.5 (C-20), 115.3 (C-
23), 102.7 (sugar C-1), 78.2 (sugar C-5), 77.84 (sugar C-2), 75.8
(C-14), 75.4 (sugar C-4), 75.2 (sugar C-3), 71.7 (C-3), 62.8 (sugar
C-6), 61.2 (C-15), 48.6 (C-17), 46.3 (C-13), 40.7 (C-12), 40.1 (C-9),
37.4 (C-10), 36.3 (C-5), 35.0 (C-4), 33.2 (C-8), 31.1 (C-1), 30.8 (C-
16), 27.5 (C-2), 27.0 (C-6), 24.1 (C-19), 22.2 (C-11), 21.7 (C-7), 17.2
(C-18); HR-ESI-MS: m/z 547.2902 [M + H]+ (calcd for C30H42O9,
547.2938).
Molecular docking
The program Autodock Vina was used for docking simulations.
For docking purpose, the crystal structure of OleD (PDBID
4M60) was retrieved from Protein Data Bank. To create receptor
and ligand structures for docking, the following procedure was
conducted. Firstly, the 3D structures of the ligands were
prepared using the Gaussian 09 program at the B3LYP/6-31G(d)
level. Harmonic vibration frequencies were calculated to
conrm the stability of these conformers. Then the receptor and
optimized structure of the ligands were converted to required
pdbqt format using Autodock Tools 1.5.4. The Autodock Vina
1
3b-glu. H NMR (CD3OD, 400 MHz) d: 7.91 (1H, dd, J ¼ 9.8,
2.4 Hz, H-22), 7.47 (1H, d, J ¼ 2.4 Hz, H-21), 6.28 (1H d, J ¼
9.8 Hz, H-23), 4.34 (1H, d, J ¼ 7.8 Hz, sugar H-1), 4.10 (1H, m, H-
3), 3.87 (1H, dd, J ¼ 11.8, 1.8 Hz), 3.76 (1H, s, H-15), 3.69 (1H,
dd, J ¼ 11.8, 5.4 Hz), 3.58 (1H, m, H-3), 3.45–3.17 (5H, m), 2.95
(1H, d, J ¼ 9.3 Hz, H-17), 2.15–2.01 (1H, td, H-8), 1.85 (3H, s,
COCH3), 1.81–1.04 (19H, m), 0.99 (3H, s, H-19), 0.83 (3H, s, H-
18); 13C NMR (CD3OD, 100 MHz) d: 171.6 (COCH3), 164.0 (C-
24), 153.5 (C-22), 150.9 (C-21), 118.4 (C-20), 114.1 (C-23), 102.7
(sugar C-1), 78.2 (sugar C-5), 77.84 (sugar C-2), 76.6 (C-16), 75.4
(sugar C-4), 75.2 (sugar C-3), 73.4 (C-14), 71.7 (C-3), 62.8 (sugar
C-6), 60.8 (C-15), 51.4 (C-17), 46.3 (C-13), 40.5 (C-5), 40.7 (C-9),
37.3 (C-12), 36.3 (C-4), 36.1 (C-1), 36.0 (C-10), 31.7 (C-8), 31.2
ꢀ
parameters were set as follow, box size: 15 ꢀ 15 ꢀ 15 A, the
center of box: x ¼ 38.96, y ¼ 61.05, z ¼ 13.83, the exhaustive-
ness: 100, and number of output conformations was set to 20.
The calculated geometries were ranked in terms of free energy
of binding and the best poses were selected for further analysis.
All molecular visualizations were carried out in PyMOL
soware.
Assessment inhibition NKA activity of CTS
The inhibitory effects of CTS on NKA were determined essen-
tially as previously reported method.11,26,27
This journal is © The Royal Society of Chemistry 2018
RSC Adv., 2018, 8, 5071–5078 | 5073

View Article Online
RSC Advances
Paper
However, it has serious toxicity because there is no distinct
selectivity for a1 and a2 subunits of NKA. Fortunately, a lot of
bufalin analogues including glycosides have been prepared by
chemical and biological transformation. Recently, our group
reported synthesis of 3b-N-methoxy-N-b-D-glucoside of bufalin
which could enhance its inhibition on NKA.22 The latest crystal
structure demonstrated that the level of glycosylation affect the
depth of CTS binding and that the steroid core substituents ne
tune the conguration of transmembrane helices aM1-2.27
Furthermore, the sugar unit could enhance the selectivity on the
alpha2 isoform of NKA; however, chemical synthesis of bufalin
glycoside was laborious and consumed a lot of toxic reagents. It
is necessary to develop a green method for glycosylation of
bufadienolides.
Results and discussion
Preparation of epimers of bufadienolides
Three epimers, i.e. a-bufalin (1a), a-resibufogenin (2a) and a-
cinobufagin (3a) were synthesized from b-bufalin (1b), b-resi-
bufogenin (3b) and b-cinobufagin (3b), respectively, by an
oxidation with pyridinium chloride hydrochloride followed by
a reduction with sodium borohydride. Uzarigenin (4a) and
digitoxigenin (4b), two natural cardenolides with a trans and cis
A/B fusion mode (4a, 4b Fig. 1), respectively, were identied
from the whole herb of Asclepias curassavica28 and the roots of
Streptocaulon juventas,29 respectively. These probes are suitable
for us to investigate whether the congurations at C-3 and the
fusion mode of A/B ring inuence the OleD-catalyze
glycosylation.
The pilot biotransformation of 1b and 1a were carried out
using UDPG as the sugar donor and OleD (ASP) as the catalyst
under the standard conditions (0.5 mM UDPG, 0.1 mM aglycon,
16 h).5 The result of LC-MS showed that OleD (ASP) could
catalyze the generation of monoglycoside of 1b with a conver-
sion rate 30%; while glycosylation product of 1a was not
detected (Fig. 2). Thus the bioconversion of 1b and 1a catalyzed
by OleD (ASP) was inuenced by the C-3 conguration. The
conversion rates of different isomers were compared in Table 1.
To maximize the production of bufalin-3-O-b-D-glucoside
(1b-glu), a 20 h reaction was carried out at 37 ꢂC. Compound 1b
(10 mg, 40 mM) was dissolved in DMSO (0.625 mL) and diluted
with buffer solution (50 mM Tris–HCl, 5 mM MgCl2, pH 8.0,
25 mL total volume). UDPG (38 mg, 50 mM) was added along
with OleD (ASP) (13 mg). The reaction was stopped with 25 mL
of ice cold methanol. Then the reaction mixture was centrifuged
at 10 000g for 30 min and supernatant was concentrated under
reduced pressure. The residue was dissolved in 3 mL methanol,
and ltered with 0.22 mM membrane. The ltrate was subjected
to preparative HPLC using water/acetonitrile as the eluent.
Glycosylation of 1b and 1a
Both 3a- and 3b-hydroxylated bufalin are endogenous in Bufo
bufo gargarizans. Though only 3b-hydroxylated bufalin is
present as a toxic chemical defence in toad venom, both
epimers were found to occur at a 2 : 3 ratio in the heart and
a 1 : 2 ratio in the blood.22 3b-Hydroxylated bufalin is the major
active component of the toad venom and exhibits potent
cardiotonic activity.30,31 Its structural core includes a cis–trans–
cis fused steroid core with two hydroxyl groups at C-3 and C-14.
Fig. 2 HPLC chromatograms of the enzymatic reactions. (A) 1b
without UDPG and OleD (ASP) (before); 1b with UDPG and OleD (ASP)
Fig. 1 Structures of cardiotonic steroid substrates and the corre- (after); (B) 1a without UDPG and OleD (ASP) (before) (1); 1a with UDPG
sponding glycosylation products.
and OleD (ASP) (after).
5074 | RSC Adv., 2018, 8, 5071–5078
This journal is © The Royal Society of Chemistry 2018

View Article Online
Paper
RSC Advances
Table 1 Conversion rates of individual compounds
Compounds
Products
Conversion rate (%)
1a
1b
2a
2b
1a-glu
1b-glu
2a-glu
2b-diglu
2b-glu
3a-glu
3b-diglu
3b-glu
4a-glu
4b-glu
N. Ta
30
1
1
79
2
5
70
2
3a
3b
4a
4b
25
a
N. T: not detected.
Finally, 3.1 mg of 1b-glu was obtained with a conversion rate of
1
21%. The structure of 1b-glu was determined by D NMR and
HR-ESI-MS analysis. The pseudo molecular ion m/z 549.3035 [M
Fig. 3 HPLC chromatograms of the enzymatic reactions. (A) 2b
without UDPG and OleD (ASP) (before); 2b with UDPG and OleD (ASP)
(after); (B) 2a without UDPG and OleD (ASP) (before); 2a with UDPG
and OleD (ASP) (after).
+ H]+ in the HR-ESI-MS was corresponding to a formula
C
30H44O9 (calcd for 549.3021), which was 162 unit larger than
the parent compound bufalin, suggesting the formation of
monoglucoside. Compared with 1b, the chemical shi dC-3 of
1b-glu was shied to downeld by 3.9 ppm and the dH-3 was
shied to downeld by 0.02 ppm; while C-14 was unchanged,
conrming the glycosylation site was at 3-OH. The b-congu-
ration was determined by the large coupling constant of the
anomeric proton (dH 4.37, d, J ¼ 7.8 Hz). Based on the above
analysis, it was concluded that OleD (ASP) could catalyze the
transformation of 1b to its monoglucoside 1b-glu at 3-OH.
To maximize the production of resibufogenin-3-O-b-D-
glucoside (2b-glu), a preparative scale experiment was under-
taken. Using the same condition as 1b, while 2b (10 mg, 40 mM)
was served as substrate, we obtain 8.4 mg 2b-glu with
a conversion rate 65%. The identication of compounds 2b-glu
was determined by 1D NMR and HR-ESI-MS analysis. The
pseudo-molecular ion m/z 547.2902 [M + H]+ in the HR-ESI-MS
corresponded to a formula C30H42O9 (calcd for 547.2938)
which was 162 unit larger than the parent compound 2b.
Compared with 2b, the chemical shi C-3 of 2b-glu was shied
to downeld by 3.9 ppm and the chemical shi H-3 was shied
to downeld by 0.02 ppm, conrming the glycosylation site was
also at 3-OH. The b-conguration was determined by the large
coupling constant of the anomeric proton (dH 4.32, d,
J ¼ 7.8 Hz). Accordingly, it was concluded that OleD (ASP) could
catalyze the transformation of 2b to its glucoside 2b-glu with
a much higher rate than 1b. It is noteworthy that bioconversion
of 2b also lead to the generation of a diglucoside (2b-diglu)
which was conrm by the HR-MS m/z 709.3421 [M + H]+. Due to
the small amount, it was not determined by NMR.
Glycosylation of 2b and 2a
Resibufogenin (2b) is another type of bufadienolides from the
venom of Bufo bufo gargarizans with an 14,15-epoxide ring in
contrast to the 14-OH in bufalin. Compound 2b has been re-
ported to exhibit a wide range of activities such as cardiotonic,
renal sodium excretion, blood pressure stimulating, antitumor
activity and immunoregulatory activity.32 However, the appli-
cation of resibufogenin is restricted because of its strong
toxicity.33 In the past several years, several structure modica-
tions on 2b was performed which generate more than 30
derivatives.23 In order to obtain the glycosylation derivative of
2b and compare the NKA inhibitory activities of these deriva-
tives with different congurations, OleD (ASP) catalysed
biotransformation on 2b and 2a were carried out.
Glycosylation of 3b and 3a
The pilot reaction conditions are the same as those Cinobufagin is also a major bufadienolides (4–6% dry weight)
described for 1b and 1a. According to the result of LC-MS from the venom of Bufo bufo gargarizans with an 14,15-epoxide
analysis, OleD (ASP) catalyzed the formation of monogluco- ring and an acetyl group at C-16.34 Cinobufagin was found to
side (major) and diglucoside (minor) of 2b with a total conver- show potent cardiotonic, blood pressure-stimulating, antiviral,
sion rate 80% which was much higher than that of 1b; while the local anesthetic and antineoplastic activities.34,35 However, the
conversion rate of monoglucoside 2a-glu is about 1% in poor water solubility restricted its clinical use. Though a series
contrast to the absence of 1a glucoside (Fig. 3). Thus similar to of analogues of cinobufagin has been generated by chemical
1b and 1a, the conguration at C-3 was also important for synthesis and cell suspension cultures,33,36 the glycosylation
conversion rate of 2b and 2a. The much high conversion rate of method is rarely reported. Furthermore, difference between the
2b and 2a might be due to the 14,15-epoxide moiety as glycosidation prole on the 3b and 3a isomers of cinobufagin is
compared to the 14-OH for 1b and 1a.
unclear.
This journal is © The Royal Society of Chemistry 2018
RSC Adv., 2018, 8, 5071–5078 | 5075

View Article Online
RSC Advances
Paper
Using the same reaction conditions as described for 2b and which are cis and trans, respectively. Compound 4b was the
2a, OleD (ASP) was found to catalyze the glycosylation of 3b aglycone of digoxin, a commonly used cardiotonic drug.37
leading to both monoglucoside (3b-glu) and diglucoside (3b- Recently, it is reported that introduction of a sugar unit at C-3 of
diglu) of 3b with a total conversion rate of 75%; while at the cardenolides could improves NKA isoform selectivity (a2/a3
same condition as 3b, only the monoglucoside of (3a-glu) was over a1).38,39 In order to compare the glycosylation proles of 4b
detected with a low conversion rate 2% (Fig. 4). Similarly, this and 4a with different A/B ring fusion mode, the OleD (ASP)
phenomenon further conrmed the importance of congura- catalyzed bioconversion was carried out.
tion at C-3 for the conversion rate. It is noteworthy that with an
Using the same reaction conditions as described for the
additional acetyl group at C-16 compound 3b showed similar bufadienolides, OleD (ASP) catalyzed the generation of mono-
conversion rate as 2b. Similarly, 3b also generate a diglucoside glucoside of 4b-glu with a conversion rate of 26%; while the
(3b-diglu) which was conrm by the HR-ESI-MS m/z 767.3479 [M conversion rate for 4a is only 2% (Fig. 5). Thus the A/B ring
+ H]+. Due to the small amount, it was not determined by NMR. fusion mode was also important for the conversion rate, and the
Using the same conditions as 2b, we prepared 6.9 mg 3b-glu lactone rings (either six-membered or ve membered) at posi-
with a conversion rate 50% (10 mg 3b was served as substrate). tion C-17 did not affect the biotransformation.
The identication of the product 3b-glu was also conrmed by
We prepared the monoglucoside of 4b (4b-glu) for structural
1D and 2D NMR and HR-ESI-MS analysis. HR-ESI-MS showed analysis. Compound 4b-glu (3.6 mg) was obtained by prepara-
a pseudo molecular ion m/z 605.2996 [M + H]+ corresponding to tive HPLC with conversion rate 25% (10 mg 4b of was used as
a formula C32H44O11 (calcd for 605.2938) which was 162 unit the substrate). Similar to the bufadienolide glycosides 1b-glu,
larger than the parent compound 3b. Similar to 2b-glu, the 2b-glu and 3b-glu, the monoglucoside structure of 4b-glu was
chemical shis of C-3 and H-3 of 3b-glu were shied to down- conrmed by 1D and 2D NMR and HR-ESI-MS analysis (m/z
eld by 3.9 ppm and 0.02 ppm, respectively, conrming the 537.8049 [M + H]+ C29H44O9, calcd for 537.8021). The glycosyl-
glycosylation site at 3-OH. The b-conguration was also ation site was again determined at 3-OH by comparison of
conrmed by the large coupling constant of the anomeric chemical shis of C-3 (downeld 3.9 ppm) and H-3 (downeld
proton (dH 4.32, d, J ¼ 7.8 Hz). Accordingly, it was concluded 0.02 ppm) of 4b-glu with those of the parent compound 4b and
that OleD (ASP) could catalyze the transformation of 3b into 3b- the b-conguration of the glycosidic bond was conrmed by the
glu.
large coupling constant of the anomeric proton (dH 4.32, d, J ¼
7.8 Hz). Accordingly, it was concluded that OleD (ASP) could
catalyze the transformation of 4b into 4b-glu.
Glycosylation of 4b and 4a
As compared to the three pairs of bufadienolides 1b and 1a, 2b
and 2a, 3b and 3a, digitoxigenin (4b) and uzarigenin (4a) were
two cardenolides with the same substitution pattern. The only
difference between 4b and 4a is the A/B ring fusion modes,
The determination of Km for the cardiotonic steroid epimers
As we can see that the substrate structure and stereochemistry
of cardiotonic steroid epimers signicantly affect the nal
Fig. 4 HPLC chromatograms of the enzymatic reactions. (A) 3b Fig. 5 HPLC chromatograms of the enzymatic reactions. (A) 4b
without UDPG and OleD (ASP) (before); 3b with UDPG and OleD (ASP) without UDPG and OleD (ASP) (before); 4b with UDPG and OleD (ASP)
(after); (B) 3a without UDPG and OleD (ASP) (before); 3a with UDPG (after); (B) 4a without UDPG and OleD (ASP) (before); 4a with UDPG
and OleD (ASP) (after).
and OleD (ASP) (after).
5076 | RSC Adv., 2018, 8, 5071–5078
This journal is © The Royal Society of Chemistry 2018

View Article Online
Paper
RSC Advances
Table 2 Conversion rate and Km
Table 3 Inhibition Na+, K+-ATPase activity
Compound
Conversiona (%)
Km (mM)
Compound
IC50 (mM)
Compound
IC50 (mM)
2b
2a
4b
4a
80
1
26
2
0.28 + 0.10 1b
>100 2b
0.81 + 0.14 3b
>100 4b
1.15 ꢃ 0.08
5.44 ꢃ 0.48
3.43 ꢃ 0.36
1.31 ꢃ 0.02
1b-glu
2b-glu
3b-glu
4b-glu
0.32 ꢃ 0.03
1.94 ꢃ 0.10
1.22 ꢃ 0.10
0.78 ꢃ 0.05
a
Conversion: calculated by the corresponding HPLC peak area
percentage.
when the conversion rate is less than 2%, Km value is toward
a large value more than 100 mM (Table 2).
glycosylation rate. In order to compare the dynamic process, Km
value was measured.
Molecular modeling
Epimers 2b and 2a representing the high conversion bufa-
dienolides and 4b and 4a representing the cardenolides were
chosen in this study. A series of concentrations of the substrates
(0.01–1.2 mM) were incubated with the OleD (ASP) enzyme and
UDPG, and the conversion rate was calculated by the corre-
sponding HPLC peak area using the Agilent Chromatography
Workstation Soware. Km value was determined based on the
Michaelis–Menten equation.25 The result indicated that a high
conversion rate corresponded to a low Km value. Particularly,
Molecular docking studies shed new light on the mechanism of
stereoselective glycosylation of cardiotonic steroids derivatives.
Similar for the Km study, we selected the high conversion
enantiomers 2a and 2b to explore the binding mode in the
enzyme substrate complex. As can be seen from Fig. 6A, the 2H-
pyran-2-one ring of compound 2a penetrated deeply into the
hydrophobic region I dened by His20, Phe85 and Trp74 resi-
dues. The epoxide contacted via van der Waals interactions with
residues lle112 and Val82. In comparison to 2a binding to OleD
(ASP), compound 2b showed an ‘inverse’ binding pose that C-3
aliphatic hydroxyl was placed in the hydrophobic pocket I and
the 2H-pyran-2-one moiety bound in a hydrophobic cavity II
(lle112, Ser184, Fig. 6B). Moreover, the oxygen atom of epoxide
ꢀ
group formed a hydrogen bond (2.5 A) with critical residue
Tyr115. It can be inferred from docking results that the lling of
hydrophobic pocket II and interacted with residue Tyr115 may
play an important role in the O-linked glycosylation reactions
catalyzed by OleD (ASP) glycosyltransferase, since the conver-
sion efficiency is 1% for 2a and 80% for 2b, respectively.
Inhibition of NKA activity
To explore the inhibitory activity of glycosylated products on
NKA, the inhibition activity of CTS (1b, 2b, 3b, 4b) and their
glycosylation products (1b-glu, 2b-glu, 3b-glu, 4b-glu) were
determined using previous reported method.11,27 As seen in
Table 3, glycosylation products showed a stronger inhibitory
activity for NKA than the corresponding aglycones.
Conclusions
In summary, this study demonstrated that the OleD-catalyze
glycosylation of cardiotonic steroids are signicantly inu-
enced by the conguration at C-3 and the A/B fusion mode. 3b-
OH and A/B ring cis fusion are favoured by OleD (ASP), while an
acetyl group at C-16 and lactone ring type at C-17 did not
inuence the biotransformation. A high conversion rate corre-
sponded to a low Km value. Molecular docking simulation
showed that lling the hydrophobic pocket II and interaction
with residue Tyr115 may play an important role in the glyco-
sylation reactions catalyzed by OleD (ASP) glycosyltransferase.
Furthermore, the glycosylation products showed a stronger
inhibitory activity for NKA than the corresponding aglycones.
This study provided the rst stereoselective properties for OleD
Fig. 6 (A) Selected docking poses of compound 2a (depicted in
yellow) and 2b (depicted in magenta) into the substrate cavity of OleD
(ASP). The two ligands are shown in stick representation. The receptors
are shown in cartoon representation with cyan alpha helices, green
beta sheets and magenta loops. (B) OleD (ASP) (surface)–ligand (stick)
complex.
This journal is © The Royal Society of Chemistry 2018
RSC Adv., 2018, 8, 5071–5078 | 5077

View Article Online
RSC Advances
Paper
(ASP) catalyzed glycosylation. It is noteworthy that glycosyl- 18 H. Gao, R. Popescu, B. Kopp and Z. Wang, Nat. Prod. Rep.,
transferase has been used for the glycosylation of a wide variety
2011, 28, 953–969.
of natural products;40,41 however, glycosyltransferase for the 19 L. Krenn and B. Kopp, Phytochemistry, 1998, 48, 1–29.
cardiotonic steroids is rare. Results of this study rstly revealed 20 P. S. Steyn and F. R. van Heerden, Nat. Prod. Rep., 1998, 15,
the stereo-selectivity of OleD-catalyzed glycosylation of cardio-
tonic steroids.
397–413.
21 H. Y. Tian, S. L. Luo, J. S. Liu, L. Wang, Y. Wang,
D. M. Zhang, X. Q. Zhang, R. W. Jiang and W. C. Ye, J. Nat.
Prod., 2013, 76, 1842–1847.
22 H. J. Tang, L. J. Ruan, H. Y. Tian, G. P. Liang, W. C. Ye,
E. Hughes, M. Esmann, N. U. Fedosova, T. Y. Chung and
J. T. Tzen, Sci. Rep., 2016, 6, 29155.
Conflicts of interest
There are no conicts to declare.
23 M. Ye, J. Han, D. An, G. Tu and D. Guo, Tetrahedron, 2005, 61,
8947–8955.
24 G. J. Williams and J. S. Thorson, Nat. Protoc., 2008, 3, 357–
362.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (No. 81573315), the Natural Science
´
˜
25 L. M. Quiros, R. J. Carbajo, A. F. Brana and J. A. Salas, J. Biol.
Chem., 2000, 275, 11713–11720.
Foundation
of
Guangdong
Province,
China
(No.
2015A030313313), the Guangzhou Industry-University Collabo-
rative Innovation Major Projects, China (No. 201508030016) and
the Natural Science Foundation of Hainan Province, China (No.
817307).
26 I. Klodos, M. Esmann and R. L. Post, Kidney Int., 2002, 62,
2097–2100.
27 M. Laursen, J. L. Gregersen, L. Yatime, P. Nissen and
N. U. Fedosova, Proc. Natl. Acad. Sci. U. S. A., 2015, 112,
1755–1760.
28 R. R. Zhang, H. Y. Tian, Y. F. Tan, T. Y. Chung, X. H. Sun,
X. Xia, W. C. Ye, D. A. Middleton, N. Fedosova,
M. Esmann, J. T. C. Tzen and R. W. Jiang, Org. Biomol.
Chem., 2014, 12, 8919–8929.
29 Y. H. Xu, J. Xu, Z. H. Chen, Z. J. Xie, R. W. Jiang and F. Feng,
Chin. J. Struct. Chem., 2016, 35, 1024–1030.
30 R. J. Bick, B. J. Poindexter, R. R. Sweney and A. Dasgupta, Life
Sci., 2002, 72, 699–709.
31 S. Q. Li, W. Schoner and Y. F. Fu, J. Biochem. Mol. Biol., 2002,
18, 250–253.
32 S. K. Wong, S. K. Tsui and S. K. Kwan, J. Pharm. Biomed.
Anal., 2002, 30, 161–170.
33 Y. Kamano, A. Kotake, H. Hashima, M. Inoue, H. Morita,
K. Takeya, H. Itokawa, N. Nandachi, T. Segawa and
A. Yukita, Bioorg. Med. Chem., 1998, 6, 1103–1115.
34 T. Nogawa, Y. Kamano, A. Yamashita and G. R. Pettit, J. Nat.
Prod., 2001, 64, 1148–1152.
35 Q. Chen, Research Methods in Pharmacology of Chinese
Materia Medica, People's Medical Publishing House, 2011,
pp. 330–331.
Notes and references
´
´
1 C. Hernandez, C. Olano, C. Mendez and J. Salas, Gene, 1993,
134, 139–140.
´
2 C. Olano, A. M. Rodriguez, C. Mendez and J. A. Salas,
Tetrahedron, 1995, 16, 333–343.
3 L. M. Quiros, I. Aguirrezabalaga, C. Olano, C. Mendez and
´
J. A. Salas, Mol. Microbiol., 1998, 28, 1177–1185.
4 D. N. Bolam, S. Roberts, M. R. Proctor, J. P. Turkenburg,
E. J. Dodson, C. Martinez-Fleites, M. Yang, B. G. Davis,
G. J. Davies and H. J. Gilbert, Proc. Natl. Acad. Sci. U. S. A.,
2007, 104, 5336–5341.
5 R. W. Gantt, R. D. Goff, G. J. Williams and J. S. Thorson,
Angew. Chem., Int. Ed., 2008, 47, 8889–8892.
6 G. J. Williams, C. Zhang and J. S. Thorson, Nat. Chem. Biol.,
2007, 3, 657–662.
7 M. Zhou, Y. Hou, A. Hamza, C.-G. Zhan, T. S. Bugni and
J. S. Thorson, Org. Lett., 2012, 14, 5424–5427.
8 M. Zhou and J. S. Thorson, Org. Lett., 2011, 13, 2786–2788.
9 J. B. Lingrel, Annu. Rev. Physiol., 2010, 72, 395–412.
10 K. Haustein, Proceedings of the European Society of Toxicology
Meeting, Kuopio, 1985, vol. 6, pp. 16–19.
36 M. Ye, J. Dai, H. Guo, Y. Cui and D. Guo, Tetrahedron Lett.,
2002, 43, 8535–8538.
11 D. A. Middleton, S. Rankin, M. Esmann and A. Watts, Proc.
Natl. Acad. Sci. U. S. A., 2000, 97, 13602–13607.
12 T. Mijatovic, F. Dufrasne and R. Kiss, Curr. Med. Chem.,
2012, 19, 627–646.
13 H. Y. Tian, L. Wang and G. A. O'Doherty, Expert Opin. Ther.
Pat., 2012, 22, 587–605.
37 S. R. Krim, R. P. Vivo, J. Perez, M. Inklab, T. Tenner and
J. Hodgson, Am. J. Med. Sci., 2008, 336, 423–428.
38 A. Katz, Y. Lifshitz, E. Bab-Dinitz, E. Kapri-Pardes,
R. Goldshleger, D. M. Tal and S. J. Karlish, J. Biol. Chem.,
2010, 285, 19582–19592.
39 J. M. Langenhan, N. R. Peters, I. A. Guzei, F. M. Hoffmann
and J. S. Thorson, Proc. Natl. Acad. Sci. U. S. A., 2005, 102,
12305–12310.
40 X. Z. He, W. S. Li, J. W. Blount and R. A. Dixon, Appl.
Microbiol. Biotechnol., 2008, 80, 253–260.
41 E. K. Lim, D. A. Ashford and B. Hou, Biotechnol. Bioeng.,
2004, 87, 623–631.
´
14 O. Aizman, P. Uhlen, M. Lal, H. Brismar and A. Aperia, Proc.
Natl. Acad. Sci. U. S. A., 2001, 98, 13420–13424.
15 T. Akera and T. M. Brody, Life Sci., 1976, 18, 135–141.
16 G. Koren, C. Woodland and S. Ito, Vet. Hum. Toxicol., 1998,
40, 45–46.
17 R. A. Newman, P. Yang, A. D. Pawlus and K. I. Block, Mol.
Interventions, 2008, 8, 36.
5078 | RSC Adv., 2018, 8, 5071–5078
This journal is © The Royal Society of Chemistry 2018