Two new triterpenoid saponins cytotoxic to human glioblastoma U251MG cells from Ardisia pusilla

Article abstract of DOI:10.1002/cbdv.200800233

Two new triterpenoid saponins, ardipusillosides IV and V (1 and 2, resp.), together with one known saponin, ardisiacrispin B (3), were isolated from the whole plants of Ardisia pusilla A. DC. Their structures were deduced by extensive spectral analysis an

Full text of DOI:10.1002/cbdv.200800233

CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
1443
Two New Triterpenoid Saponins Cytotoxic to Human Glioblastoma U251MG
Cells from Ardisia pusilla
by Hai-Feng Tang*a), Jun Yunb)1), Hou-Wen Linc), Xiao-Li Chena), Xiao-Juan Wangd), and
Guang Chenge)
a) Department of Pharmacy, Xijing Hospital, Fourth Military Medical University, 15 Changle West Road,
Xiꢀan 710032, P. R. China (phone/fax: þ86-29-84775471; e-mail: tanghaifeng71@163.com)
b) Department of Vascular Surgery, Xijing Hospital, Fourth Military Medical University,
15 Changle West Road, Xiꢀan 710032, P. R. China
c) Department of Pharmacy, Changzheng Hospital, Second Military Medical University,
415 Fengyang Road, Shanghai 200003, P. R. China
d) Department of Pharmacy, School of Stomatology, Fourth Military Medical University, 1 Kangfu Road,
Xiꢀan 710032, P. R. China
e) Department of Neurosurgery, Xijing Hospital, Fourth Military Medical University,
15 Changle West Road, Xiꢀan 710032, P. R. China
Two new triterpenoid saponins, ardipusillosides IV and V (1 and 2, resp.), together with one known
saponin, ardisiacrispin B (3), were isolated from the whole plants of Ardisia pusilla A. DC. Their
structures were deduced by extensive spectral analysis and chemical evidences. Compound 1 contains a
glycosylated glycerol residue which is a very rare structural feature among triterpenoid glycosides and
has been so far found only in the genus Ardisia. All the saponins exhibited significant cytotoxicity against
human glioblastoma U251MG cells, but did not affect the growth of primary cultured human astrocytes.
Introduction. – Ardisia pusilla A. DC (Myrsinaceae) is a widely occurring shrub in
southern China. Its whole plants, also known as ꢁJiu Jie Longꢀ, have been used as an
antidote in Chinese traditional medicine [1]. Other plants of this genus have also been
used and are well-documented in traditional medicine in Southeast Asia [2]. Previous
chemical studies have shown that triterpenoid saponins are the main components in this
genus. To date, four triterpenoid saponins, ardipusillosides I, II, III, and an unnamed
saponin, have been reported from A. pusilla [3]. Ardipusillosides I and II showed
significant antitumor effects in both Lewis pulmonary carcinoma and hepatocarcinoma
[4]. As part of our ongoing investigation on new antitumor glycosides from natural
sources [5], we studied the bioactive triterpenoid saponins of this plant. In this article,
we report the isolation and structure elucidation of two new triterpenoid saponins,
ardipusillosides IVand V (1 and 2, resp.), along with one known saponin, ardisiacrispin
B (3), as well as their potent cytotoxicity against human glioblastoma U251MG cells.
Results and Discussion. – The EtOH extract of A. pusilla was suspended in H2O and
partitioned successively with petroleum ether and BuOH. The BuOH extract was
subjected to several chromatographic purification steps to afford 1–3.
1
)
First co-author.
ꢂ 2009 Verlag Helvetica Chimica Acta AG, Zꢃrich

1444
CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
Ardipusilloside IV (1), a colorless amorphous powder, was positive to Lieber-
mann–Burchard and Molish tests. The positive-ion mode HR-ESI-MS showed
pseudomolecular-ion peak at m/z 1385.5955 ([M þ Na]þ ), which, together with the
pseudomolecular-ion peak at m/z 1339 ([MꢀNa]ꢀ ) in the negative-ion mode ESI-MS
and with the NMR data, led to the molecular formula C62H99NaO31. The fragment-ion
peak at m/z 1187 in the positive-ion mode ESI-MS corresponding to the loss of 198
mass units from 1385, was interpreted as due to the loss of a sodium uronate unit,
suggesting that the uronic acid residue was in the form of a monosodium salt.
Saponin 1 displayed 62 C-atom signals in the 13C-NMR spectrum, of which 30 could
be assigned to the signals of the aglycone. It was evident that 1 was a triterpenoid
saponin related to olean-12-ene skeleton based on the 1H-NMR signals assigned to six
tertiary Me groups at d(H) 0.64 (s, Me(25)), 0.72 (s, Me(26)), 0.79 (s, Me(24)), 0.97 (s,
Me(23)), 1.12 (s, Me(29)), and 1.55 (s, Me(27)), together with the 13C-NMR signals for
olefinic C-atoms at d(C) 122.0 (C(12)) and 143.3 (C(13)). The presence of a C¼O C-
atom, C(30), was deduced from the 13C-NMR resonance at d(C) 177.7, and the long-
range coupling between Me(29) and C(30) in the HMBC experiment. Further, OH
groups at C(3) and C(16) could be attributed by the 13C-NMR resonance at d(C) 88.1

CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
1445
(C(3)) and 72.2 (C(16)). In the NOESY spectrum, the correlations between HꢀC(3)
(d 3.00) and Me(23) (d 0.97), HꢀC(3), and HꢀC(5) (d 0.47), and between HꢀC(16)
(d 4.53) and H-atoms of CH2(28) (d 3.29, 3.55) indicated the b-configuration of the OH
group at C(3) and the a-configuration of the OH group at C(16). The assignments of
the NMR signals associated with the aglycone moiety (Table 1) were derived from
1H,1H-COSY, TOCSY, HSQC, HMBC, and NOESY experiments, and the relative
configurations at all stereogenic centers were established by analysis of the NOESY
spectrum. By comparison of the NMR data with those reported [6], the aglycone of 1
was identified as (3b,16a)-3,16,28-trihydroxy-12-olean-30-oic acid (jacquinic acid).
The glycosidation shifts for the signals due to C(3) (Dd(C) þ10.0) and C(30) (Dd(C)
ꢀ2.7) indicated that 1 is a bisdesmoside of jacquinic acid with glycosyl linkage at both
the OH group at C(3) and the COOH group at C(30).
Alkaline hydrolysis of 1 afforded a prosapogenin a and a side-chain moiety (1b),
which was originally linked to C(30) of the aglycone. Compound a was subjected to
acidic hydrolysis to give l-arabinose (Ara), l-rhamnose (Rha), and d-glucose (Glc) in
a ratio of 1:1:2 based on GC analysis of the corresponding trimethylsilyl derivatives
using an l-Chirasil-Val column [7]. Hydrolysis of the side-chain moiety (1b) with HCl
vapor on TLC plate revealed the presence of glucuronic acid (GluA) by comparison
with an authentic sample. The above evidence suggested that 1 was a pentoside with a
tetrasaccharide moiety linked to C(3) of the aglycone and a side chain including a
sodium glucuronate residue attached to C(30) of the aglycone through an ester bond.
The 1H-NMR spectrum of 1 displayed five signals ascribable to the anomeric H-atoms
with signals at d(H) 4.77, 4.84, 5.11, 5.35, and 5.92, which were correlated in the HSQC
experiment with the corresponding C-atoms with signals at d(C) 103.0, 102.0, 103.2,
99.0, and 100.3, resp. The b-configurations for both glucose units were determined from
their coupling constants of anomeric H-atoms (7.8 Hz). The HꢀC(1) non-splitting
pattern of rhamnose unit and the three-bond HMBC correlations from HꢀC(1) (Rha4)
to C(3) (Rha4) and C(5) (Rha4) indicated that its anomeric H-atom was equatorial,
thus possessing an a-configuration in the 1C4 form [8]. The a-configuration for
glucuronic acid moiety was clear from its small HꢀC(1) coupling constant (3.6 Hz).
The a-configuration for the arabinose residue was evidenced by the correlations
between HꢀC(1) (Ara1) and HꢀC(3) (Ara1), and between HꢀC(1) (Ara1) and
4
HꢀC(5) (Ara1) in the NOESY experiment observed for the C1 form, although its
coupling constant of anomeric H-atom (4.8 Hz) was smaller than methyl-a-l-
arabinopyranoside (8 Hz). This could be explained by the fast conformational
1
4
equilibrium between its C4 and C1 conformers [9]. All H-atom signals due to sugar
moieties were identified by careful analysis of the 1H,1H-COSY, TOCSY, and NOESY
spectra, and the C-atom signals were assigned by HSQC and further confirmed by the
HMBC spectrum. Data from the above experiments (Table 1) indicated that five sugar
residues were in their pyranose forms. The absolute configuration of the sodium
glucuronate residue was assumed to be d based on biogenetic considerations.
After mapping all of the signals for each sugar moiety, a 5-H spin system ascribable
to a glycerol (Gly) residue was evident from the 1H,1H-COSY and TOCSY spectra in
the region from d(H) 3.8 to 4.5 that correlated, in the HSQC experiment, with C-atom
signals at d(C) 64.8 (CH2(1) (Gly)), 67.9 (C(2) (Gly)), and 69.3 (CH2(3) (Gly)). The
glycerol fragment was confirmed by HMBC correlations as shown in the Figure.

1446
CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
Table 1. 1 H- and 13C-NMR Data of 1. At 600 and 150 MHz respectively, in C5D5N/D2O 2 : 1; d in ppm,
J in Hz.
d(H)
d(C)
d(H)
d(C)
CH2(1)
CH2(2)
0.62–0.66 (m, 1 H),
1.32–1.36 (m, 1 H)
1.75–1.80 (m, 1 H),
1.82–1.86 (m, 1 H)
37.7
Ara1(1!C(3)):
HꢀC(11)
HꢀC(21)
HꢀC(31)
HꢀC(41)
CH2(51)
4.77 (d, J¼4.8)
4.29–4.32 (m)
103.0
78.3
70.4
73.0
62.7
25.1
4.33–4.36 (m)
4.37 (br. s)
HꢀC(3) 3.00 (dd, J¼4.4, 11.4)
88.1
38.1
54.6
17.7
31.9
C(4)
–
3.90 (d, J¼10.4, 1 H),
HꢀC(5) 0.47 (d, J¼10.8)
4.27–4.31 (m, 1 H)
CH2(6)
CH2(7)
1.27–1.33 (m)
1.11–1.15 (m, 1 H),
1.37–1.40 (m, 1 H)
–
Glc2(1!2):
HꢀC(12)
HꢀC(22)
HꢀC(32)
HꢀC(42)
HꢀC(52)
CH2(62)
5.11 (d, J¼7.8)
3.82–3.86 (m)
4.09–4.12 (m)
3.95–3.99 (m)
3.78–3.81 (m)
4.08–4.11 (m, 1 H),
4.25 (br. d, J¼11.4, 1 H)
103.2
74.7
75.8
69.8
77.0
61.0
C(8)
38.7
45.9
35.5
HꢀC(9) 1.41–1.46 (m)
C(10)
–
CH2(11) 1.65–1.70 (m)
HꢀC(12) 5.46 (br. t)
22.6
122.0
143.3
40.4
C(13)
C(14)
–
–
CH2(15) 1.50 (dd, J¼4.2, 14.4, 1 H),
33.2
Glc3(1!4):
HꢀC(13)
HꢀC(23)
HꢀC(33)
HꢀC(43)
HꢀC(53)
CH2(62)
2.02 (dd, J¼4.2, 14.4, 1 H)
4.84 (d, J¼7.8)
102.0
76.4
77.2
70.1
76.6
60.8
HꢀC(16) 4.53 (br. s)
72.2
38.9
42.6
3.92–3.95 (m)
C(17)
–
4.16–4.19 (m)
HꢀC(18) 2.25 (dd, J¼4.2, 14.4)
3.75 (dd, J¼9.0, 9.2)
3.65–3.69 (m)
CH2(19) 2.01–2.06 (m, 1 H),
2.49 (dd, J¼13.2, 13.8, 1 H)
43.3
3.96–4.00 (m, 1 H),
4.21 (br. d, J¼11.4, 1 H)
C(20)
–
43.6
32.3
30.8
CH2(21) 2.18–2.22 (m)
CH2(22) 1.97–2.00 (m, 1 H),
2.27–2.30 (m, 1 H)
Rha4(1!2):
HꢀC(14)
HꢀC(24)
HꢀC(34)
HꢀC(44)
HꢀC(54)
Me(64)
5.92 (s)
100.3
71.5
70.6
72.6
68.7
17.3
4.48–4.52 (m)
4.40–4.43 (m)
4.04–4.08 (m)
4.63–4.67 (m)
1.63 (d, J¼6.6)
Me(23)
Me(24)
Me(25)
Me(26)
Me(27)
0.97 (s)
0.79 (s)
0.64 (s)
0.72 (s)
1.55 (s)
27.0
15.5
14.5
15.7
26.1
69.0
CH2(28) 3.29 (d, J¼10.8, 1 H),
GluA5 (1!CH2(3)(Gly)):
3.55 (d, J¼10.8, 1 H)
HꢀC(15)
HꢀC(25 )
HꢀC(35 )
HꢀC(45 )
HꢀC(55 )
C(65 )
5.35 (d, J¼3.6)
3.98–4.02 (m)
4.38–4.42 (m)
4.34–4.37 (m)
4.55 (d, J¼10.2)
–
99.0
71.5
72.4
71.5
71.8
175.7
Me(29)
C(30)
1.12 (s)
27.6
177.7
–
Gly:
CH2(1)
4.44 (br. d, J¼11.4, 1 H),
4.46 (br. d, J¼11.4, 1 H)
64.8
HꢀC(2) 4.30–4.33 (m)
67.9
69.3
CH2(3)
3.80–3.83 (m, 1 H),
4.14 (br. d, J¼11.4, 1 H)

CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
1447
Inspection of HMBC and NOESY spectra led to the determination of the sequence and
binding sites of the oligosaccharide chain. In the HMBC spectrum, a cross-peak
between C(3) of the aglycone and HꢀC(1) of arabinose indicated that Ara1 was
connected to C(3) of the aglycone. The linkage of the terminal glucose at C(2) of Ara1
was indicated by the cross-peak HꢀC(1) (Glc2)/C(2) (Ara1). Similarly, the linkages of
the terminal rhamnose at C(2) of the inner glucose, in turn, linked to C(4) of Ara1 were
indicated by cross-peaks HꢀC(1) (Rha4)/C(2) (Glc3), HꢀC(1) (Glc3)/C(4) (Ara1).
The conclusion was confirmed by NOESYexperiment. It must be noted that, compared
with the corresponding C-atoms of Glc2, the glycosidation shifts for the signals due to
C(2) (Glc3) (Dd(C) þ1.7) and C(3) (Glc3) (Dd(C) þ1.4) caused by the attachment of
Rha4 were too small and unusual to predict an interglycosidic site. It has been reported
that this situation could be correlated with the distortion of the corresponding torsion
angles [10]. The side-chain structure at C(30) of the aglycone was also elucidated by
HMBC correlations between H-atoms of CH2(1) (Gly) and C(30) of aglycone, and
between HꢀC(1) of sodium glucuronate residue and CH2(3) (Gly) (Fig.). The
configuration of the glycerol remains to be determined. On the basis of the above
evidences, the structure of ardipusilloside IV (1) was determined as (3b,16a)-16,28-
dihydroxy-3-({a-l-rhamnopyranosyl-(1!2)-b-d-glucopyranosyl-(1!4)-[b-d-gluco-
pyranosyl-(1!2)]-a-l-arabinopyranosyl}oxy)olean-12-en-30-oic acid 2-hydroxy-3-
[(sodium a-d-glucopyranuronat)yloxy]propyl ester. Ardipusilloside IV contains a
glycosylated glycerol residue, which is a very rare structural feature among triterpenoid
glycosides. To date, only ardisicrenoside E and F, two saponins possessing the same
structural feature previously isolated from A. crenata, have been reported. Compound
1 was the sodium salt of the demethylated derivative of ardisicrenoside E [10].
Figure. Key NOESY and HMBC correlations for saponin 1
Ardipusilloside V (2) was obtained as a colorless amorphous powder with a
molecular formula of C58H94O27 determined by HR-ESI-MS. Fragment-ion peak at m/z
1113 ([MþNaꢀ132]þ ) in ESI-MS (positive-ion mode) indicated the presence of a
terminal pentose residue in 2. Alkaline hydrolysis of 2 afforded a prosapogenin and an
l-arabinose identified by GC analysis of its corresponding trimethylsilyl derivative.

1448
CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
Further subjected to acidic hydrolysis, the prosapogenin yielded the same mono-
saccharides as for a, i.e., l-arabinose, l-rhamnose, and d-glucose (1:1:2), identified by
GC analysis. The 1H-NMR, 13C-NMR, and DEPT spectra of 2 (Table 2) revealed that 2
was also a bisdesmoside, with the same triterpene and the same tetrasaccharide moiety
linked to C(3) of the aglycone as those of 1, but differed in the side-chain fragment
attached to C(30). Extensive NMR studies (1H,1H-COSY, TOCSY, HSQC, HMBC,
and NOESY) indicated that the only difference between 1 and 2 was the replacement
of the glycosyl glycerol moiety in 1 by an a-l-arabinose unit in 2. A cross-peak of long-
range coupling between a H-atom signal at d(H) 6.19 (HꢀC(1) (Ara5)) and a C¼O
C-atom signal at d(C) 177.1 (C(30) of the aglycone) in HMBC spectrum substantiated
that one of the arabinoses was located at C(30) through an ester bond. Consequently,
the structure of 2 was established as (3b,16a)-16,28-dihydroxy-3-({a-l-rhamnopyra-
nosyl-(1!2)-b-d-glucopyranosyl-(1!4)-[b-d-glucopyranosyl-(1!2)]-a-l-arabino-
pyranosyl}oxy)olean-12-en-30-oic acid a-l-arabinopyranosyl ester. To date, glucose is
the only sugar residue bound to the aglycone by a glycosidic ester linkage among
triterpenoid saponins from genus Ardisia reported. Saponin 2 is the first example of an
arabinose residue located at the aglycone through an ester bond.
On the basis of its MS and NMR data compared with literature data, and by acid
hydrolysis, followed by GC analysis of the corresponding trimethylsilylated mono-
saccharides, saponin 3 was identified as ardisiacrispin B, i.e., (3b,13b,16a)-16-hydroxy-
3-({a-l-rhamnopyranosyl-(1!2)-b-d-glucopyranosyl-(1!4)-[b-d-glucopyranosyl-
(1!2)]-a-l-arabinopyranosyl}oxy)-13,28-epoxyoleanan-30-al, previously isolated
from A. crenata [11].
The in vitro cytotoxicity of 1–3 and the prosapogenin a, i.e., (3b,16a)-16,28-
dihydroxy-3-({a-l-rhamnopyranosyl-(1!2)-b-d-glucopyranosyl-(1!4)-[b-d-gluco-
pyranosyl-(1!2)]-a-l-arabinopyranosyl}oxy)olean-12-en-30-oic acid, against human
glioblastoma U251MG cells and primary cultured human astrocytes was evaluated by
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium hydrobromide (MTT) col-
orimetric assay [3d][12]. The IC50 value of each saponin tested was determined on the
basis of cell viability, after 48-h treatment. Nimustine hydrochloride (ACNU), a known
anticancer agent commonly used in clinical treatment for malignant brain tumors, was
used as the positive control. The results were listed in Table 3, and indicated that
saponins 1, 2, and 3 displayed significant cytotoxicity against glioblastoma U251MG
cells, while the prosapogenin a was inactive (IC50 >100 mm).
Saponin 3 and a share the same tetrasaccharide chain, but differ in their oleanane
skeleton. This led to significant difference in cytotoxicity against U251MG cells. Earlier
studies on the cytotoxicity of 3 and a against other three human cancer cell lines have
reached the same result. In addition, ardisiamamilloside F, an analogue of 3 with the
similar structure except that the aldehyde group at C(30) in 3 was replaced by a
carboxyl group, was also inactive [8]. Thus, it seems that the structure of the aglycone
especially the aldehyde group at C(30) play an important role in terms of cytotoxic
activity against tumor cells. However, the bisdesmosides 1 and 2 with the same aglycone
and tetrasaccharide moiety at C(3) exhibited remarkable cytotoxicity. It seems that the
presence of an additional sugar or glycosyl glycerol moiety at C(30) greatly increases
the activity. Thus, our results indicated that the cytotoxic activity of such saponins was
influenced by the structures of both the aglycones and the sugar parts, and very

CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
1449
Table 2. 1H- and 13C-NMR Data of 2. At 600/150 MHz resp., in C5D5N/D2O 2 : 1; d in ppm, J in Hz.
d(H)
d(C)
d(H)
d(C)
CH2(1)
CH2(2)
0.66–0.70 (m, 1 H),
1.35–1.39 (m, 1 H)
1.76–1.81 (m, 1 H),
1.84–1.88 (m, 1 H)
3.05 (dd, J¼4.0, 11.4)
–
0.51 (d, J¼10.8)
1.30–1.36 (m)
1.13–1.17 (m, 1 H),
1.40–1.44 (m, 1 H)
–
37.8 Ara1(1!C(3)):
HꢀC(1)
4.80 (d, J¼4.8)
4.34–4.37 (m)
4.30–4.33 (m)
4.40 (br. s)
3.85–3.89 (m, 1 H),
4.24–4.26 (m, 1 H)
103.1
78.8
70.7
73.4
62.7
25.2 HꢀC(2(1)
HꢀC(3(1)
HꢀC(3)
C(4)
HꢀC(5)
CH2(6)
CH2(7)
88.2 HꢀC(4(1)
38.2 CH2(51 )
54.6
17.7
31.9
Glc2(1!2):
HꢀC(1(2)
5.15 (d, J¼7.8)
3.86–3.90 (m)
4.17–4.20 (m)
3.98–4.02 (m)
3.82–3.85 (m)
4.13–4.17 (m, 1 H),
4.28 (br. d, J¼10.8, 1 H)
103.7
74.9
76.1
70.1
77.1
61.2
C(8)
38.9 HꢀC(2(2)
45.9 HꢀC(3(2)
35.6 HꢀC(4(2)
22.6 HꢀC(5(2)
122.1 CH2(62 )
143.4
HꢀC(9)
1.44–1.49 (m)
–
1.69–1.74 (m)
C(10)
CH2(11)
HꢀC(12) 5.52 (br. t)
C(13)
–
–
C(14)
CH2(15)
40.5
1.53 (dd, J¼4.2, 14.4, 1 H),
33.5 Glc3(1!4):
2.05 (dd, J¼4.2, 14.4, 1 H)
HꢀC(13)
4.88 (d, J¼7.8)
102.0
76.5
77.5
70.4
76.9
60.9
HꢀC(16) 4.58 (br. s)
72.7 HꢀC(23 )
39.1 HꢀC(33 )
3.97–4.00 (m)
C(17)
–
4.14–4.17 (m)
HꢀC(43 )
HꢀC(53 )
CH2(63 )
3.79 (dd, J¼9.0, 9.6)
HꢀC(18) 2.28 (dd, J¼4.2, 14.4)
42.2
43.5
CH2(19)
2.05–2.10 (m, 1 H),
2.54 (dd, J¼13.2, 14.4, 1 H)
–
2.22–2.26 (m)
2.00–2.03 (m, 1 H),
2.31–2.34 (m, 1 H)
1.01 (s)
0.83 (s)
0.69 (s)
0.76 (s)
1.58 (s)
3.68–3.72 (m)
4.01–4.05 (m, 1 H),
4.22 (br. d, J¼10.8, 1 H)
C(20)
CH2(21)
CH2(22)
43.6
32.4
30.5
Rha4(1!2):
HꢀC(14)
HꢀC(24 )
HꢀC(34 )
HꢀC(44 )
HꢀC(54 )
Me(64 )
5.99 (s)
100.4
71.4
70.8
72.8
68.8
17.3
Me(23)
Me(24)
Me(25)
Me(26)
Me(27)
CH2(28)
27.0
15.5
14.6
15.8
26.2
68.9
4.51–4.55 (m)
4.46 (dd, J¼3.2, 9.6)
4.06–4.10 (m)
4.64–4.68 (m)
1.66 (d, J¼6.0)
3.32 (d, J¼10.8, 1 H),
3.57 (d, J¼10.8, 1 H)
1.14 (s)
Ara5(1!C(30)):
HꢀC(15)
HꢀC(25)
HꢀC(35)
HꢀC(45)
CH2(55)
6.19 (d, J¼7.2)
95.7
71.3
73.8
68.0
66.0
Me(29)
C(30)
27.5
177.1
–
4.62 (br. t)
4.08–4.12 (m)
4.20 (br. s)
3.88–3.92 (m, 1 H),
4.30–4.34 (m, 1 H)
sensitive to their precise functionalization. Therefore, more extensive studies are
needed before a clear structure-activity relationship (SAR) can be reached.
It is worth to be mentioned that saponins 1–3 did not affect the growth of primary
cultures of human astrocytes. In fact, exposure of the astrocytes to the highest
concentration of 1–3 (100 mm) for 48 h did not result in any statistically significant

1450
CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
Table 3. Cytotoxicity of Saponins 1, 2, 3, and a against Human Glioblastoma U251MG Cells and Primary
Cultured Astrocytes in vitro
IC50 [mm]a)
1
2
3
a
ACNUb)
0.98ꢂ0.06
>100
U251MG
Astrocytes
1.55ꢂ0.21
>100
2.20ꢂ0.35
>100
3.33ꢂ0.29
>100
>100
>100
a) IC50 Values are means from three independent experiments (meanꢂSD). b) Nimustine hydrochloride
(ACNU) as positive control.
change in cell viability, with the % inhibition ranging from 4.3 to 8.5% (p>0.05).
Meanwhile, viability of astrocytes treated with 100 mm ACNU for 48 h decreased to
69.2% (p<0.05). Glioblastoma multiforme is the most common and lethal primary
brain malignancy and relatively resistant to chemotherapy. Therefore, the development
of effective drugs to reverse its drug resistance and induce apoptosis is critical. We have
reported that ardipusilloside III isolated also from A. pusilla could induce apoptosis of
U251 cells, and both the BAD-mediated intrinsic apoptotic signaling pathway and the
caspase-8-mediated extrinsic apoptotic signaling pathway were involved in the
apoptosis [3d]. Clearly, the promising saponins 1, 2, and 3 also merit further study as
potential anticancer agents.
Experimental Part
General. Column chromatography (CC): silica gel (200–300 mesh and 10–40 mm; Qingdao, China),
Sephadex LH-20 (Pharmacia), and Lobar Lichroprep RP-18 (size B, 40–63 mm; Merck). TLC: precoated
silica gel GF254 plates (10–40 mm; Yantai, China); detection by spraying with 20% aq. H2SO4, followed by
heating. HPLC: Dionex P680 liquid chromatograph equipped with a UV170 UV/VIS detector monitored
at 206 nm; Elite Sino Chrom ODS-BP column (250ꢁ10 mm i.d., 5 mm). M.p.: XT5-XMT apparatus;
uncorrected. Optical rotations: Perkin-Elmer 341 polarimeter. NMR: Bruker AV-600, at 600 (1H) and
150 MHz (13C); d in ppm rel. to Me4Si, J in Hz. ESI- and HR-ESI-MS: Waters Q-TOF Micro YA019 mass
spectrometer; in m/z. GC: Finnigan Voyager GC/MS apparatus with a l-Chirasil-Val column (25 mꢁ
0.32 mm i.d.).
Plant Material. The whole plants of A. pusilla were collected in Jiajiang County, Sichuan Province,
China, in September 2006, and identified by one of the authors, Prof. Xiao-Juan Wang. A voucher
specimen (No. 06-SC03) was deposited with the Department of Pharmacy, School of Stomatology, Fourth
Military Medical University, Xiꢀan, China.
Extraction and Isolation. Dried powder of the whole plants (3 kg) was refluxed with 95% EtOH (3ꢁ
8 l, each for 2 h) and centrifuged. The combined extract was suspended in H2O (1 l) and partitioned
successively with petroleum ether and BuOH (each 3ꢁ1 l). The BuOH-soluble fraction (10 g) was
subjected to CC (SiO2; lower phases of CHCl3/MeOH/H2O 7:3 :1 and 6.5 :3.5 :1): Fr. 1–8. Fr. 3 and 7
mainly contained triterpenoid saponins. Fr. 7 (0.3 g) was subjected to CC (RP-18; MeOH/H2O 1:1):
Fr. 7.1–7.3. Fr. 7.1 (58 mg) was purified by HPLC (Elite Sino Chrom ODS-BP; 38% MeOH, 2.0 ml/min):
1 (12.3 mg, tR 22.0 min). Fr. 7.3 (80 mg) was purified by HPLC (75% MeOH, 2.0 ml/min): 2 (37.3 mg, tR
13.5 min). Fr. 3 (1.2 g) was submitted to CC (Sephadex LH-20; MeOH/H2O 1:1) to remove the pigments
and carbohydrates and finally purified by HPLC (90% MeOH, 2.0 ml/min): 3 (445 mg, tR 9.5 min).
Ardipusilloside IV (¼(3b,16a)-16,28-Dihydroxy-3-({a-l-rhamnopyranosyl-(1!2)-b-d-glucopyra-
nosyl-(1!4)-[b-d-glucopyranosyl-(1!2)]-a-l-arabinopyranosyl}oxy)olean-12-en-30-oic Acid 2-Hy-
droxy-3-[(sodium-a-d-glucopyranuronat)yloxy]propyl Ester; 1). Colorless amorphous powder. M.p.

CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
1451
235–2368 (dec.). [a]2D2 ¼ þ25.2 (c ¼ 0.15, MeOH). 1H- and 13C-NMR: Table 1. Key HMBC and NOESY
correlations: Figure. ESI-MS (pos.): 1385 ([MþNa]þ ), 1187 ([MþNaꢀGluANa]þ ), 1113 ([1187ꢀ
CH2CHOHCH2O]þ ), 1097 ([1187ꢀOCH2CHOHCH2O]þ ). ESI-MS (neg.): 1361 ([MꢀH]ꢀ ), 1339
([MꢀNa]ꢀ ), 1163 ([MꢀGluANaꢀH]ꢀ ), 1089 ([MꢀGluANaꢀCH2CHOHCH2OH]ꢀ ), 1073
([1089ꢀO]ꢀ ). ESI-MS/MS (neg., 1339): 1193 ([1339ꢀRha]ꢀ ), 1177 ([1339ꢀGlc]ꢀ ), 1089, 1073,
1031 ([1177ꢀRha]ꢀ ), 943 ([1089ꢀRha]ꢀ ), 927 ([1089ꢀGlc]ꢀ ), 869 ([1031ꢀGlc]ꢀ ), 781 ([943ꢀ
Glc]ꢀ ), 619 ([781ꢀGlc]ꢀ ), 601 ([AraþRhaþ2ꢁGlcꢀH]ꢀ ), 487 ([aglyconeꢀH]ꢀ ). HR-ESI-MS:
1385.5955 ([M þ Na]þ , C62H99Na2O3þ1 ; calc. 1385.5965).
Ardipusilloside V (¼(3b,16a)-16,28-Dihydroxy-3-({a-l-rhamnopyranosyl-(1!2)-b-d-glucopyrano-
syl-(1!4)-[b-d-glucopyranosyl-(1!2)]-a-l-arabinopyranosyl}oxy)olean-12-en-30-oic Acid a-l-Arabi-
nopyranosyl Ester; 2). Colorless amorphous powder. M.p. 232–2338 (dec.). [a]2D2 ¼ þ12.8 (c ¼ 0.13,
MeOH). 1H- and 13C-NMR: Table 2. ESI-MS (pos.): 1245 ([MþNa]þ ), 1113 ([MþNaꢀAra]þ ). ESI-
MS (neg.): 1221 ([MꢀH]ꢀ ), 1089 ([MꢀHꢀAra]ꢀ ). ESI-MS/MS (neg., 1221): 1089 ([1221ꢀAra]ꢀ ),
1075 ([1221ꢀRha]ꢀ ), 1059 ([1221ꢀGlc]ꢀ ), 943 ([1089ꢀRha]ꢀ ), 927 ([1089ꢀGlc]ꢀ ), 781 ([943ꢀ
Glc]ꢀ ), 619 ([781ꢀGlc]ꢀ ), 487 ([aglyconeꢀH]ꢀ ). HR-ESI-MS: 1245.5887 ([M þ Na]þ , C58H94NaOþ27
;
calc. 1245.5880).
Alkaline Treatment of 1 and 2. Saponin 1 (5 mg) was dissolved in 1m NaOH soln. (2 ml) and heated at
1008 for 1 h. The mixture was neutralized with a cation-exchange resin (Dowex 50W-X2, Hþ ) and
evaporated in vacuo. The residue was subjected to CC (RP-18; H2O and MeOH/H2O 6 :4): 1b (<1 mg)
and a (2.6 mg), resp. 1b was applied to a TLC plate and hydrolyzed with HCl vapor at 808 for 20 min. The
plate was subjected to co-TLC analysis (CHCl3/MeOH/H2O 15 :6 :2 (9 ml of the lower phase and 1 ml
AcOH), detection: aniline phthalate) with authentic sugars. Glucuronic acid was identified (Rf 0.62). By
the same method, saponin 2 (10 mg) furnished a (6.2 mg) and 2b. Compound 2b was dissolved in 1-
(trimethylsilyl)-1H-imidazole and pyridine (0.1 ml). The soln. was stirred at 608 for 5 min and dried with
a stream of N2. The residue was partitioned between CHCl3 and H2O. The org. layer was analyzed by GC,
and the peaks at 8.77 and 9.65 min enabled the identification of l-arabinose. tR Values for authentic
sugars after being treated simultaneously with 1-(trimethylsilyl)-1H-imidazole in pyridine were 8.68 and
9.60 min (d-arabinose), 8.76 and 9.64 min (l-arabinose), 9.39 and 10.28 min (d-rhamnose), 9.31 and
10.21 min (l-rhamnose), 14.56 min (d-glucose), and 14.50 min (l-glucose).
Acid Hydrolysis of 1 and 2 (Prosapogenin a). Each prosapogenin (2 mg) was heated in 2m
CF3COOH (1 ml) at 1208 for 2 h. The mixture was evaporated to dryness, and the residue was
partitioned between CHCl3 and H2O. The aq. phase was concentrated and the residue was treated with 1-
(trimethylsilyl)-1H-imidazole in pyridine by the same procedure performed for 2b. The sugar moieties of
1 and 2 were both determined to be l-arabinose, l-rhamnose, and d-glucose in a ratio of 1:1:2 by
comparing the GC retention times of the corresponding trimethylsilylated hydrolysates with those of the
authentic samples prepared in the same manner.
Bioassays. The cytotoxicity of 1–3 and a against human glioblastoma U251MG cells (originally
obtained from Uppsala, Sweden) and primary cultured human astrocytes was evaluated by MTT (¼ 3-
(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium hydrobromide) colorimetric assay described in
[3d][12], with the anticancer agent nimustine hydrochloride (ACNU, Sigma) as positive control.
Cultured primary astrocytes were obtained from a slightly impaired brain-tissue fragment of a volunteer
with cerebral trauma who consented to the procedure as described in [3d]. Acquisition of the tissue was
approved by the local medical research ethics committee at Xijing Hospital, Fourth Military Medical
University. Target cells were grown to log phase in DMEM medium containing 10% fetal bovine serum
and transferred to serum free medium in 96-well plates at a density of 4ꢁ103 cells/well. Cultures were
preincubated for 24 h in a humidified 5% CO2/95% O2 atmosphere at 378. Then, control or test soln. was
put into each well, and the plates were incubated for an additional 48 h. At the end of exposure, MTT
solved in PBS (¼ phosphate-buffered saline) was added to each well at a final concentration of 5 mg/ml,
and then incubated for 4 h. The H2O-insoluble dark blue formazan crystals formed during MTT cleavage
in actively metabolizing cells were dissolved in DMSO. The optical density of each well was measured
with a Bio-Rad 680 microplate reader at 490 nm. The activities of saponins 1, 2, 3, a, and ACNU were
determined at 100, 10, 1, 0.1, and 0.01 mm (each concentration was tested in triplex wells), resp. Data were
calculated as percentage of inhibition by the formula: % Inhibition¼(100ꢀ(ODt/ODs)ꢁ100), ODt and

1452
CHEMISTRY & BIODIVERSITY – Vol. 6 (2009)
ODs being the mean optical densities of the test compounds and the solvent control, resp. The
concentration inducing a 50% inhibition of cell growth (IC50) was determined graphically for each
experiment using the curve-fitting routine of the computer software Prism 4.0 (GraphPad) and the
equation derived by De Lean et al. [13]. The IC50 value represented the mean of three independent
experiments and was expressed as meanꢂSD using Studentꢀs t test.
REFERENCES
[1] Jiangsu College of New Medicine, ꢁDictionary of Chinese Materia Medicaꢀ, Shanghai Peopleꢀs
Publishing House, Shanghai, 1977, p. 892.
[2] L. M. Perry, J. Metager, ꢁMedicinal Plants of East and South-east Asiaꢀ, MIT Press, Cambridge, MA,
1980.
[3] a) Q. H. Zhang, X. J. Wang, Z. C. Miao, R. Feng, Acta Pharm. Sin. 1993, 28, 673; b) Z. C. Miao, R.
Feng, Y. X. Zhou, G. Y. Li, Chin. J. Magn. Reson. 1999, 16, 395; c) Z. C. Miao, R. Feng, Y. X. Zhou,
G. Y. Li, Q. H. Zhang, Chin. J. Org. Chem. 2000, 20, 361; d) H. Lin, X. Zhang, G. Cheng, H. F. Tang,
W. Zhang, H. N. Zhen, J. X. Cheng, B. L. Liu, W. D. Cao, W. P. Dong, P. Wang, Apoptosis 2008, 13,
247.
[4] X. J. Tao, P. X. Wang, X. J. Yang, H. P. Yao, J. Liu, Y. X. Cao, Zhong Yao Cai (China), 2005, 28, 574.
[5] H. F. Tang, Y. H. Yang, L. Ling, P. Sun, S. Q. Zhang, Y. P. Zhao, J. Nat. Prod. 2005, 68, 337; H. F.
Tang, Y. H. Yang, L. Ling, P. Sun, S. Q. Zhang, Y. P. Zhao, Planta Med. 2005, 71, 458; H. F. Tang,
Y. H. Yi, S. Y. Zhang, P. Sun, L. Li, D. Z. Zhou, Chin. Chem. Lett. 2005, 16, 479; H. F. Tang, Y. H. Yi,
L. Li, P. Sun, S. Q. Zhang, Y. P. Zhao, Fitoterapia 2006, 77, 28; S. Y. Zhang, Y. H. Yi, H. F. Tang,
Chem. Biodiversity 2006, 3, 807; G. Cheng, X. Zhang, H. F. Tang, Y. Zhang, X. H. Zhang, W. D. Cao,
D. K. Gao, X. L. Wang, B. Q. Jin, J. Neuro-Oncol. 2006, 79, 235; S. Y. Zhang, Y. H. Yi, H. F. Tang, J.
Nat. Prod. 2006, 69, 1492; S. Y. Zhang, H. F. Tang, Y. H. Yi, Fitoterapia 2007, 78, 283.
[6] Z. H. Jia, K. Koike, T. Nikaido, T. Ohmoto, M. Y. Ni, Chem. Pharm. Bull. 1994, 42, 2369; K. Koike,
Z. H. Jia, S. Ohura, S. Mochida, T. Nikaido, Chem. Pharm. Bull. 1999, 47, 434; J. Huang, Y. Ogihara,
H. Zhang, N. Shimizu, T. Takeda, Chem. Pharm. Bull. 2000, 48, 1413.
[7] S. De Marino, N. Borbone, M. Iorizzi, G. Esposito, J. B. McClintock, F. Zollo, J. Nat. Prod. 2003, 66,
515.
[8] X. L.Chang, W. Li, Z. H. Jia, T. Satou, S. Fushiya, K. Koike, J. Nat. Prod. 2007, 70, 179.
[9] N. De Tommasi, S. Piacente, F. De Simone, C. Pizza, Z. L. Zhou, J. Nat. Prod. 1993, 56, 1669.
[10] Z. H. Jia, K. Koike, T. Nikaido, T. Ohmoto, Tetrahedron 1994, 50, 11853.
[11] C. Jansakul, H. Baumann, L. Kenne, G. Samuelsson, Planta Med. 1987, 53, 405.
[12] F. Denizot, R. Lang, J. Immunol. Methods 1986, 89, 271.
[13] A. De Lean, P. J. Munson, D. Rodlbard, Am. J. Physiol. 1978, 235, E97.
Received July 4, 2008