In vivo evaluation and atom-based 3D-QSAR studies on saponins from shells of Xanthoceras sorbifolium Bunge as anti-AD agents

Article abstract of DOI:10.1016/j.bioorg.2019.103412

A series of oleanane type of glycosides (1–19), including nine undescribed ones (1–9), were isolated and identified from Xanthoceras sorbifolium. Subsequently, an atom-based 3D-QSAR model was constructed based on results of the in vivo anti-AD evaluation of the isolates (2–3, 10–17) and nine literature-reported anti-AD oleanane type of glycosides to clarify the structure-anti-AD activity, and under the guidance of which 19 was predicted and proved to elicit a nearly equivalent in vivo anti-AD effect as xanthoceraside and donepezil that were used as positive drugs in the Y maze and Morris water maze test.

Full text of DOI:10.1016/j.bioorg.2019.103412

Journal Pre-proofs
In vivo evaluation and atom-based 3D-QSAR studies on saponins from shells
of Xanthoceras sorbifolium Bunge as anti-AD agents
Gang Chen, Yumeng Xie, Di Zhou, Lei Yang, Libo Zou, Maosheng Cheng,
Yang Liu, Ning Li
PII:
S0045-2068(19)31495-6
DOI:
https://doi.org/10.1016/j.bioorg.2019.103412
YBIOO 103412
Reference:
To appear in:
Bioorganic Chemistry
Received Date:
Revised Date:
Accepted Date:
10 September 2019
14 October 2019
28 October 2019
Please cite this article as: G. Chen, Y. Xie, D. Zhou, L. Yang, L. Zou, M. Cheng, Y. Liu, N. Li, In vivo evaluation
and atom-based 3D-QSAR studies on saponins from shells of Xanthoceras sorbifolium Bunge as anti-AD agents,
Bioorganic Chemistry (2019), doi: https://doi.org/10.1016/j.bioorg.2019.103412
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2019 Published by Elsevier Inc.

In vivo evaluation and atom-based 3D-QSAR studies on
saponins from shells of Xanthoceras sorbifolium Bunge as
anti-AD agents
a
a
a
c
c
b
b*
Gang Chen , Yumeng Xie , Di Zhou , Lei Yang , Libo Zou , Maosheng Cheng , Yang Liu , and
Ning Li a,e *
a
School of Traditional Chinese Materia Medica, Shenyang Pharmaceutical University, Shenyang
1
10016, Liaoning, China.
b
Key Laboratory of Structure-Based Drug Design & Discovery of Ministry of Education, School
of Pharmaceutical Engineering, Shenyang Pharmaceutical University, Shenyang 110016, People’s
Republic of China.
c
School of Life Science and Biopharmaceutics, Shenyang Pharmaceutical University, 103 Wenhua
Road, Shenyang
E-mail: y.liu@syphu.edu.cn. liningsypharm@163.com

Abstract
A series of oleanane type of glycosides (1-19), including nine undescribed ones (1-9), were isolated
and identified from Xanthoceras sorbifolium. Subsequently, an atom-based 3D-QSAR model was
constructed based on results of the in vivo anti-AD evaluation of the isolates (2-3, 10-17) and nine
literature-reported anti-AD oleanane type of glycosides to clarify the structure-anti-AD activity, and
under the guidance of which 19 was predicted and proved to elicit a nearly equivalent in vivo anti-
AD effect as xanthoceraside and donepezil that were used as positive drugs in the Y maze and
Morris water maze test.
Key words: Xanthoceras sorbifolium, oleanane glycosides, anti-AD, 3D-QSAR

1
. Introduction
Alzheimer's disease (AD), a disastrous neurodegenerative disorder with an unrelenting progression
of memory dysfunction and cognition impairment, accounts for around 70 % of dementia. The
astrogliosis, neuronal loss, and nerve cell atrophy can be observed in the affected region of the brain,
where senile plaques and neurofibrillary tangles (NFTs) are formed and extensively distributed. In
the 1980s, it was found that the amyloid β (Aβ) peptide was intensively deposited and thereby
consists of senile plaque[1, 2], which makes Aβ a therapeutic target for anti-AD drug development
for over 20 years [3]. However, to be disappointing, almost all the attempts to develop Aβ-targeting
drugs for the treatment of AD have ended up in failure. On the one hand, these failure of Aβ-
targeting drugs make us in a dreadful situation that we do not have any effective in vitro model for
anti-AD drug discovery due to the complex and so-far obscure pathogenesis of AD, while on the
other, to be encouraging, many natural agents, especially saponins, exhibited potent in vivo anti-AD
or neuroprotective abilities [4-6]. Among those effective anti-AD saponins, xanthoceraside has
attracted extensive interests for its potent therapeutic effects on AD treatment evaluated by in vivo
AD-related models, in which the effective dose of xanthoceraside was as low as 0.32 mg/kg [7, 8].
Xanthoceraside, being isolated from the shells of Xanthoceras sorbifolium, is an oleanane-type
glycoside. Besides xanthoceraside, total saponins of the shells of X. sorbifolium also elicited strong
effects in promoting learning and reducing memory impairments via repressing oxidative stress and
synaptic damage[9]. Of note is that other oleanane glycosides were also reported to be able to protect
memory and cognitive impairment[10,11]. All these results indicated the high potential of
discovering other effective anti-AD saponins from X. sorbifolium. More importantly, as the absence
of effective and rapid in vitro screening models, the structure-anti-AD activity relationship for the
oleanane-type glycosides is not available due to a limited number of in vivo researches, preventing
the structural modification-based optimization of therapeutic effect.
X. sorbifolium, a peculiar tree in China, distributes in the mid-eastern and north parts of China,
including Inner Mongolia, Liaoning, Shanxi, and Hebei provinces. In addition to the aforementioned
in vivo protective effects on learning and memory impairments[10, 11], pentacyclic triterpenoid
saponins were also reported to be the main constituents of X. sorbifolium [12].
Hence, a series of oleanane type of glycosides (1-19), including nine undescribed ones (1-9),
were isolated and identified from X. sorbifolium. Subsequently, compounds 2-3, 10-17 were
subjected to in vivo evaluations for their anti-AD effects to construct an atom-based 3D-QSAR
model along with the in vivo anti-AD data of nine literature-reported anti-AD oleanane-type
glycosides (see table S1). Finally, with the constructed 3D-QSAR model and in vivo evaluation
results, compounds 19 was predicted and proved to elicit an equivalent in vivo anti-AD effect as
xanthoceraside.
2
. Results and discussion
2
.1 Chemistry

Compound 1 was obtained as a white, amorphous powder (methanol). Its molecular formula
was determined to be C H O based on its HRESI-MS data that gave an m/z value of 1179.5941
5
8
92 23
+
(
calcd. 1179.5927 for C H NaO ), corresponding to [M+Na] . Seven angular methyl groups and
58 92 23
a double bond were elucidated by the methyl signals of singlet multiplicity at δ 0.86 (Me-25), 1.03
H
2
(
Me-26), 1.12 (Me-29), 1.18 (Me-24), 1.29 (Me-23), 1.33 (Me-30), 1.85 (Me-27) and the sp proton
1
2
signal at δ 5.51 (1H, br s, H-12) in the H NMR spectra. All these methyl and sp proton signals
H
1
13
indicated that the aglycone of 1 was an oleanane-type triterpene. Then all the H and C NMR data
were summarized as shown (Tables 1 and 2) based on the results provided by the HSQC spectra. In
addition to those aforementioned typical oleanane-type signals, most of the NMR signals assignable
to the aglycone, especially six oxygen-substituted carbon/proton signals at δ 63.5 (C-28)/δ 3.48,
C
H
3
.75 (H-28), δ 67.9 (C-15)/δ 4.21 (H-15), δ 73.5 (C-16)/δ 4.40 (H-16), δ 73.8 (C-22)/δ 6.26
C H C H C H
(
H-22), δ 79.1 (C-21)/δ 6.68 (H-21), δ 90.4 (C-3)/δ 3.24 (H-3), resembled those of R -
C H C H 1
barrigenol [12], except for those belonged to the additional angeloyl [δ 21.6 (C-5'), 16.2 (C-4'),
C
1
3
39.1 (C-3'), 129.0 (C-2'), 167.9 (C-1')] and 2-methylbutyryl [δ 12.2 (C-4''), 17.4 (C-5''), 26.9 (C-
C
''), 42.1 (C-2''), 176.8 (C-1'')] groups which were linked to C-21 and C-22 respectively as
determined by the HMBC correlations from δ 6.68 (H-21) to δ 167.9 (C-1') and from δ 6.26 (H-
H
C
H
2
2) to δ 176.8 (C-1'') (Fig. 2). Thus, the aglycone of 1 was established as 21-O-angeloyl-22-O-(2-
C
methyl) butyryl-R -barrigenol. Moreover, three anomeric carbon signals at δ 105.5 (C-1'''), 105.2
1
C
(C-1''''), and 111.5 (C-1''''') suggested the presence of three sugar moieties in 1. Through comparing
the NMR data with those in the literatures [12, 13], the NMR signals ascribable to the sugar moieties
were attributed to the 6-O-methyl-β-D-glucuronyl, β-D-galactosyl, and α-L-arabinosyl respectively.
Acid hydrolysis experiment of 1 showed the presence of D-galactose, L-arabinose, D-glucuronic
acid. Combined with the HMBC correlations of H-1'''/C-3, H-1''''/C-2''', and H-1'''''/C-3''' (Fig. 2),
the structure of 1 was finally determined as 3-O-[α-L-arabinofuranosyl(1→3)]-[β-D-
galactopyranosyl(1→2)]-β-D-6-O-methylglucuronopyranosyl-21-O-angeloyl-22-O-(2-methyl)
butyryl-R -barrigenol.
1
Compound 2 was obtained as white needle crystal (MeOH) with the molecular formula of
+
C H O as determined by the HRESI-MS data of [M+Na] m/z : 855.4890 (calcd. 855.4871 for
4
6
72 13
C H NaO ). All the NMR data that belonged to the aglycone of 2 were almost identical to those
4
6
72
13
of 1, except for the presence of an additional group of angeloyl signals and the absence of signals
assignable to the 2-methylbutyryl group that was linked to C-22 of 1. Thus, the aglycone of 2 was
indicated to be 21, 22-di-O-angeloyl-R -barrigenol, which was further confimed by the HMBC
1
correlation from δ 5.81 (H-21) to δ 166.5 (C-1') and from δ 5.55 (H-22) to δ 166.6 (C-1'') (Fig.
H
C
H
C
2
). The remaining NMR data of 2 were assigned to a glucopyranosyl as the acid hydrolysis
experiment verified the presence of D-glucose. HSQC spectra gave the correlation between the
anoreic carbon (δ 96.0, C-1''') and proton (δ 4.77, d, J = 2.7 Hz, H-1''') signals. The relative
C
H
configuration of the D-glucose was determined to be α due to the coupling constant of H-1''' (J =
2
2
.7 Hz). In conjunction of the HMBC correlation from H-1''' to C-3 (δ 82.4) of the aglycone (Fig.
C
), the structure of 2 was determined as 3-O-α-D-glucopyranosyl-21,22-di-O-angeloyl-R1-

barrigenol.
Compound 3 was obtained as a white, amorphous powder (methanol). The quasi-molecular ion
+
peak of [M+Na] at m/z: 871.4838 indicated the molecular formula of C H O (calcd. 871.4820
4
6
72 14
for C H NaO ). By comparing the NMR data of 3 with those of 1 and 2, the skeleton of the
4
6
72
14
aglycone of 1 was also determined as the R -barrigenol. In addition to the carbon signals of an
1
angeloyl group [δ 20.4 (C-5'), 15.4 (C-4'), 137.5 (C-3'), 127.6 (C-2'), 166.4 (C-1')], a group of
C
1
3
isobutyryl signals [18.9 (C-4''), 18.7 (C-3''), 33.6 (C-2''), 175.4 (C-1'')] were also detected in the C
spectra of 3. HMBC correlations from δ 5.77 (H-21) to δ 166.4 (C-1') and from δ 5.44 (H-22) to
H
C
H
δC 175.4 (C-1'') (Fig. 2) allowed the elucidation of the agylcone as 21-O-angeloy-22-O-isobutyryl-
R -barrigenol. Based on the result of the HMBC correlation from δ 3.66 (-OCH ) to δ 167.9 (C-
1
H
3
C
6
''') (Fig. 2) and the NMR data in the literature [14], the remaining NMR data were attributed to a
β-D-6-O-methyl-glucuronopyranosyl group that was linked to the C-3 of the aglycone as determined
by the HMBC correlation from δ 4.29 (H-1''') to δ 88.1 (C-3) and the result of the acid hydrolysis
H
C
experiment. Therefore, the structure of
3 was established as 3-O-β-D-6-O-methyl-
glucuronopyranosyl-21-O-angeloy-22-O-isobutyryl-R -barrigenol.
1
Compound 4 was also obtained as a white, amorphous powder (methanol). The HR-ESI MS
+
gave the quasi-molecular ion peak at m/z: 1253.6307 [M+Na] , suggesting a molecular formula of
C H O (calcd. 1253. 6295 for C H NaO ). Apart from the additional acetyl-produced signals
6
1
98 25
61 98
25
[δC 171.5 and 21.5], all the typical NMR resonances of 4, including those of seven angular methyl
groups [δ 1.25 (H-23), 0.99 (H-24), 0.90 (H-25), 1.10 (H-26), 1.23 (H-27), 1.04 (H-29), and 1.09
H
(
H-30)], oxygen-substituted C-22 (δ 68.5), four anomeric carbons (δ 100.7, 103.7, 105.9, and
C
C
1
2
07.2), and the angeloyl group [δ 168.4 (C-1''''''), 129.3 (C-2''''''), 137.3 (C-3''''''), 16.3 (C-4''''''),
C
1.1 (C-5'''''')], were almost identical to those of 3-O-[β-D-glucopyranosyl(1→6)]-
[angeloyl(1→3)]-β-D-glucopyranosyl-28-O-[α-L-rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-
1
6-deoxybarringtogenol C, with an evident difference in the chemical shift of C-21 that was shifted
to the lower field (δ 79.9) [15], suggesting that the additional acetyl was located at C-21, which
C
was in consistence with the HMBC result showing the correlation from δ 5.38 (H-21) to δ 171.5
H
C
(C-1') (Fig. 2). The absolute configurations of the sugar moiety were further examined via the acid
hydrolysis experiment, which revealed the presence of D-glucose and L-rhamnose. Thus, 4 was
elucidated as 21-acetyl-3-O-[β-D-glucopyranosyl(1→6)]-[angeloyl(1→3)]-β-D-glucopyranosyl-
2
8-O-[α-L-rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-16-deoxy-barringtogenol C.
Compound 5, a white, amorphous powder (methanol) with the molecular formula of
C H100O26 as its HRESI-MS gave the m/z of 1295.6401 which was attributed to [M+Na]+
6
3
(
calcd.1295.6401 for C H NaO ), suggesting the presence of an additional acetyl compared with
63 100 26
4
regarding the molecular weight. The NMR data of 5 were similar to those of 4, except for those
(
δ 1.99/δ 170.5 and 21.1) that belonged to the aforementioned additional acetyl. HMBC spectra
H C
analysis gave the correlation from 5.48 (H-4'') of the inner glucopyranosyl to the 170.5 (C-1''''''') of
the addition acetyl group (Fig. 2), indicating the location of the acetyl. The acid hydrolysis
experiment of 5 revealed the same results as those of 4, verifying the presence of D-glucose and L-

rhamnose. Thus, the structure of 5 was elucidated as 21-O-acetyl-3-O-[β-D-glucopyranosyl(1→6)]-
angeloyl(1→3)]-β-D-4-O-acetyl-glucopyranosyl-28-O-[α-L-rhamnopyranosyl(1→2)]-β-D-
[
glucopyranosyl-16-deoxybarringtogenol C.
Being a white, amorphous powder (methanol), compound 6 has the same molecular formula
+
as 4 by giving the m/z of 1253.6286 [M+Na] in its HRESI-MS spectra. A detailed comparison
between the NMR resonance of 4 and 6 led to the conclusion that 6 possessed the same aglycone
1
3
moiety and a very similar saccharide moiety as those of 4. In the C NMR spectra of 6, the signals
assignable to the 28-O-[α-rhamnopyranosyl(1→2)]-β-glucopyranosyl, 6''-O-β-glucopyranosyl, and
the angeloyl moieties were identical to those of 4, suggesting that 6 had a different substitution
pattern in the angeloyl-substituted 3-O-β-glucopyranosyl moiety comparing to 4. The HMBC
correlation from H-4'' (δ 5.59) of the 3-O-β-glucopyranosyl to C-1'''''' (δ 168.2) of the angeloyl
H
C
group determined the substitution pattern of the angeloyl-substituted 3-O-β-glucopyranosyl moiety.
Combined with the HMBC correlations shown in figure 2, 6 was elucidated as 21-acetyl-3-O-[β-
glucopyranosyl(1→6)]-[angeloyl(1→4)]-β-glucopyranosyl-28-O-[α-rhamnopyranosyl (1→2)]-β-
glucopyranosyl-16-deoxy-barringtogenol C. Subsequently, the acid hydrolysis experiment was
performed, showing the presence of D-glucose and L-rhamnose. Thus, 6 was determined as 21-
acetyl-3-O-[β-D-glucopyranosyl(1→6)]-[angeloyl(1→4)]-β-D-glucopyranosyl-28-O-[α-L-
rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-16-deoxy-barringtogenol C.
Compound 7 was also obtained as a white, amorphous powder (methanol) whose molecular
+
formula was determined as C H O by the [M+Na] m/z: 701.4238 (calcd. 701.4241 for
3
8
62 10
1
3
C H NaO ). In the C NMR spectra, an anomeric carbon signal at δ 105.1 was observed,
3
8
62
10
C
suggesting the existence of a monosaccharide. The signals of the sugar moiety [δ 105.1 (C-1'), 70.8
C
(C-2'), 73.4 (C-3'), 70.9 (C-4'), 70.1 (C-5'), 16.4 (C-6')] were in consistent with those of the β-D-
fucopyranosyl in the literature [13], including the anomeric proton δ 4.08 (d, 7.4) that suggested
H
1
3
the β configuration. Apart from the C NMR signals ascribable to an acetyl [δ 170.1 (C-1'') and
C
2
0.9 (C-2'')] and the aforementioned β-D-fucopyranosyl, all the remaining carbon signals were
identical to those of barringtogenol C in the literature [13]. HMBC correlations from δ 3.92 (H-
H
2
2
1) to δ 105.1 (C-1') and from δ 3.74/3.64 (H-28) to δ 170.1 (C-1'') allowed the location of the
C H C
1-O-D-fucopyranosyl and 28-O-acetyl. Thus, 7 was elucidated as 28-O-acetyl-21-O-β-D-
fucopyranosyl barrigenol C.
-
Compound 8 showed a quasi-molecular peak of [M-H] m/z: 815.4953, indicating a molecular
formula of C H O (calcd. 815.4946 for C H O ). Compound 8 was also suggested to possess
4
6
72 12
46 71 12
the R -barrigenol as the aglycone by comparing the NMR data with those of 1-3 and those in the
1
literature [13]. The remaining NMR data were identical to those of the 3,4-O-diangeloyl-β-D-
fucopyranosyl in the literature [12]. The HMBC correlation from H-21 (δ 3.99) to C-1' (δ 104.4)
H
C
of the 3,4-O-diangeloyl-β-D-fucopyranosyl determined the location of the sugar moiety. Thus, 8
was established as 21-O-(3,4-di-O-angeloyl)-β-D-fucopyranosyl-R -barrigenol.
1
1
13
Compound 9 was obtained as a white, amorphous powder (methanol). The H and C NMR
data of 9 were very similar to those of compound 2, except for those assignable to the sugar moiety.

The NMR signals ascribable to the sugar moiety of 9 resembled those of the glucuronopyranosyl in
the literature[16]. In conjunction with the result of the acid hydrolysis revealing the presence of D-
glucuronic acid, the structure of 9 was finally deduced as 3-O-α-D-glucuronopyranosyl-21,22-di-
O-angeloyl-R -barrigenol.
1
The structures of compounds 10-19 were identified by comparing the NMR data with those in
the literatures [12, 17] (Fig. S1).
2
.2 In vivo evaluation of anti-AD activities
In order to discover more anti-AD saponins and thereby summarize the structure-activity
relationship accordingly, compounds 2-3, 10-17 were subjected to in vivo evaluations with the
methods of Y-maze and Moris water maze test. As shown in figure 3, treatments with compound
1
3 (0.32 mg/kg), or positive drug donepezil (1.2 mg/kg) significantly attenuated the impairments of
spontaneous alternation behaviors in Aβ1-42 injected mice (Fig. 3A), while compounds 2, 3, 10-12,
4-17 elicited moderate or weak protective effects. There was no significant difference in the total
1
number of arm entries, indicating that both of the operation and the vehicle treatment exhibited no
evident effect on the central excitability and motor function (Fig. 3B).
Subsequently, the Moris water maze test was adopted to further examine the effects of 10-13
that showed stronger activities in the Y-maze test on spatial learning and memory impairment. The
results showed that compared with the QA injected model group, 10-13 (0.64 mg/kg) obviously
decreased the escape latency to find the platform with significance on the 3rd and 4th day (Fig. 4).
It is noteworthy that 13 elicited a nearly equivalent effect as the positive drug donepezil (Fig. 4).
2
.3 Atom-based 3D-QSAR study
To explore the structure-anti-AD activity relationship for oleanane-type glycosides, nine
literature-reported saponins with in vivo neuroprotective or anti-AD effects, along with compounds
-3, 10-17 were used to construct an atom-based 3D-QSAR model[18,19]. Before the
2
implementation of the 3D-QSAR study, a preliminary analysis on the structures in connection with
the anti-AD activities showed that sugar chains played a crucial role and could even dominate the
ultimate anti-AD effect. And for a sugar chain that composed of two or even more than three
monosaccharides, there would be a few of conformers for the sugar moiety. So which conformer
should be used for the alignment in the 3D-QSAR study? As we know that being different from
other small molecules, the accurate alignment of sugar chains of different monosaccharides is very
difficult, which means we will easily get a false positive or a false negative result for a saponin in a
3
D-QSAR study due to the difference in the initial conformations that are used in a 3D-QSAR model.
To have this issues addressed, each saponin was subjected to a conformational search to generate
low-energy conformers in the first place, and one to seven representative conformers, depending on
the ΔG value and the similarity of each conformer (for those that were conformationally similar,
only one conformer was adopted), were aligned (Figure S2) and subsequently utilized in the atom-
based 3D-QSAR model (Tables S1-S3). As shown in table S2, the constructed 3D-QSAR model
2
2
possessed the R value of 0.827 and the Q value of 0.681, suggesting a good predictive ability. The

results of the atom type fractions showed that the hydrophobic/non-polar and electron-withdrawing
fraction accounted for 0.499 and 0.397 of the total fractions respectively. As shown in figure 5A,
the favorable hydrophobic/non-polar fraction (yellow cubes) in position B would contribute to the
anti-AD effect, suggesting that the C-21 and C-22 angeloyls were favorable to the in vivo activity
of xanthoceraside. Moreover, the favorable hydrophobic/non-polar fraction in position A (Fig 5A)
located at the position of the second monosaccharide at -OH (C-3), indicating that the sugar chain
linked to C-3 should be composed of at least two monosaccharides. A similar indication was given
at position D and E (Fig 5A) at which the second or third monosaccharide located, suggesting that
a longer sugar chain (more than two monosaccharides) of C-28 was required for the anti-AD effect.
In addition, the angeloyl was also located at the yellow hydrophobic/non-polar fraction (position C
of figure 5A), which also contributed to the potent anti-AD effect of compound 13. The favorable
electron-withdrawing fraction was shown in figure 5B as green cubes, and those of which located
at position F and G revealed that hydroxyls of C-21 and C-22 were favorable for the anti-AD activity.
Additionally, hydroxyls of position H and I also indicated that sugar chains of C-3 or C-28 were
supposed to be composed of at least two monosaccharides (Fig 5B).
2
.4 In vivo verification of the atom-based 3D-QSAR model
To further verify the proposed structure-anti-AD activity relationship by the atom-based 3D-
QSAR model, compounds 9 and 19, whose structures were almost identical except for the sugar-
chains of C-3 (monosaccharide sugar chain for 9 vs disaccharide sugar chain for 19), were subjected
to further in vivo evaluations through Y-maze test and Moris water maze test. According to the
structure-anti-AD activity relationship derived from the aforementioned atom-based 3D-QSAR
model, 9 was predicted to be inactive (predicted activity of 3.231) while 19 was predicted to be
active (predicted activity of 5.122) (Table S4). Using the donepezil and xanthoceraside which
possesses a trisaccharide sugar chain as positive drugs, 19 (0.64 mg/kg) elicited strong in vivo anti-
AD effects while 9 (0.64 mg/kg) was proved to be almost inactive as predicted (Fig 6 and 7). For
compounds 1, 4-8, 18, due to the low amount of them, their inhibitory effects on NO production
were evaluated in LPS-activated BV2 cells to further illustrate the accuracy of results derived from
the atom-based 3D-QSAR model. As shown in table S4, except for 1 that was predicted to possess
moderate activity, 4-8 and 18 that were predicted to be inactive also exhibited no inhibitory effect
on NO production in LPS-activated BV2 cells.
Based on the results of the atom-based 3D-QSAR study and in vivo verification of 9 and 19.
Structure-anti-AD activity relationships can be concluded as follows:
(1) For sugar chains of C-3 or C-28: more than two monosaccharides (n ≥ 2) were required for
at least one of the two sugar chains;
(
2) C-21 and C-22 hydroxyls were favorable to the in vivo anti-AD activity;
3) C-21 and C-22 angeloyls were crucial to the in vivo anti-AD activity.
(
Interestingly, other types of saponins, such as pseudoginsenoside-F11 [20], ginsenoside Rg1
[21], ginsenoside Rb1 [22], and ginsenoside Re [23], also elicited in vivo neuroprotective effects

against cognitive impairment, and to be noted, all these effective saponins [20-23] consist of at least
two monosaccharides, exhibiting a similar structure-activity relationship compared with that of the
aforementioned oleanane glycosides. Thus, for oleanane glycosides, in addition to preserving the
C-21 and C-22 angeloyls, the structural modification on the C-3/28 disaccharide is a feasible
direction for the discovery of more potent anti-AD agents.
3
. Conclusion
Herein we reported nine undescribed (1-9) and ten known (10-19) oleanane-type glycosides
identified from X. sorbifolium. On the basis of in vivo anti-AD evaluation results of these saponins
by both Y-maze and Moris water maze tests and in vivo data of nine literature-reported anti-AD
oleanane-type glycosides, an atom-based 3D-QSAR model was constructed and thereby the
structure-anti-AD activity relationships were concluded. Additionally, through the studies in this
paper, two potent anti-AD saponins (13 and 19) that elicited almost equivalent in vivo anti-AD
effects as xanthoceraside were also reported for the first time. As almost all the synthetic anti-AD
candidates were reported failing the phase III clinical trial, our results also suggested that the natural
saponins such as xanthoceraside, compounds 13 and 19 were good anti-AD candidates to be
subjected to further structural modifications.
4
. Experimental section
4
.1 Chemistry
NMR spectra were recorded on Bruker ARX-400 and 600M AVIII spectrometers, using TMS
as an internal standard. HRESI-MS was acquired on a Bruker micro TOF-Q mass spectrometer in
m/z mode. Silica gel (200-300 mesh) for chromatography was produced by Qingdao Ocean
Chemical Group Company (Qingdao, China). Sephadex LH-20 was purchased from Pharmacia
Company (Uppsala, Switzerland). ODS (50 μm) was afforded by YMC Company (Ltd) (Japan).
HPLC separations were performed on a YMC-pack Prep-ODS column (250×20 mm) equipped with
a Shimadzu RID-20AUV detector and a Shimadzu LC-6AD series pumping system (Tokyo, Japan).
Pyridine-d5 were obtained from Sigma-Aldrich Company (St. Louis, USA). All the
chromatographic and analytical grade reagents were obtained from Tianjin DaMao Chemical
Company (Tianjin, China).
4
.2 Plant materials
The shell of Xanthoceras sorbifolium Bunge was collected in Fuxin city, Liaoning province in
September (autumn), 2018. The material was identified by Professor Yingni Pan (School of
Traditional Chinese Materia Medica, Shenyang Pharmaceutical University). A voucher specimen
(No.20180921) is deposited in School of Traditional Chinese Materia Medica, Shenyang
Pharmaceutical University.
4
.3 Extraction and isolation
The shell of X. sorbifolium (200 kg) were extracted with 70% ethanol under reflux for 3×2 h.
The filtrate was concentrated under reduced pressure, and afforded the crude extract (200 kg).

Ethanol extract was fractioned on macroporous absorptive resin (D101) column, eluting with H O,
2
2
0% ethanol/H O (Fr. A), 50% ethanol/H O (Fr. B), 70% ethanol/H O (Fr. C), 90% ethanol/H O
2 2 2 2
(Fr. D), successively. Fr. C (1750 g) was chromatographed on silica gel, eluting with CH Cl -MeOH
2 2
(0:100-100:0), to give six fractions (Fr. D1-D7). Fr. D2 was subjected to a ODS column washing
with MeOH/H O (0: 100-100: 0) to get five fractions. The sub-fraction DD-1 was isolated by HPLC
2
with MeOH/H O (70: 30) as eluent to afford compounds 1 (3.1 mg), 2 (13.0 mg), 3 (23.1 mg), 7
2
(3.1 mg), 8 (17.1 mg), 9 (21.1 mg), 10 (12.0 mg); sub-fraction DD-2 was isolated by HPLC with
MeOH/H O (65: 35) as eluent to afford compounds 4 (16.1 mg), 12 (18.6 mg), 14 (19.1 mg), 17
2
(
14.3 mg), 19 (19.1 mg); sub-fraction DD-3 was isolated by HPLC with MeOH/H O (55: 45) as
2
eluent to afford compounds 5 (3.8 mg), 6 (17.7 mg), 11 (13.1 mg), 13 (23.3 mg), 15 (16.1 mg), 16
19.7 mg), 18 (4.3 mg).
(
4
.3.1 3-O-[α-L-arabinofuranosyl(1→3)]-[β-D-galactopyranosyl(1→2)]-β-D-6-O-methylglucuro-
nopyranosyl-21-O-angeloyl-22-O-(2-methyl)butyryl-R -barrigenol.
1
+
1
White powder. HRESI-MS [M+Na] m/z 1179.5941 (calcd. 1179.5927 for C H NaO ); H-
5
8
92
23
1
3
NMR and C-NMR data see Tables 1 and 2.
4
.3.2 3-O-α-D-glucopyranosyl-21,22-di-O-angeloyl-R -barrigenol.
1
+
1
White powder. HRESI-MS [M+Na] m/z 855.4890 (calcd. 855.4871 for C H NaO ); H-
4
6
72
13
1
3
NMR and C-NMR data see Tables 1 and 2.
4
.3.3 3-O-β-D-6-O-methylglucuronopyranosyl-21-O-angeloy-22-O-isobutyryl-R1-barrigenol
+
1
White powder. HRESI-MS [M+Na] m/z 871.4838 (calcd. 871.4820 for C H NaO ); H-
4
6
72
14
1
3
NMR and C-NMR data see Tables 1 and 2.
.3.4 21-acetyl-3-O-[β-D-glucopyranosyl(1→6)]-[angeloyl(1→3)]-β-D-glucopyranosyl-28-O-[α-
4
L-rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-16-deoxy-barringtogenol C
+
1
White powder. HRESI-MS [M+Na] m/z 1253.6307 (calcd. 1253. 6295 for C H NaO ); H-
6
1
98
25
1
3
NMR and C-NMR data see Tables 1 and 2.
.3.5 21-O-acetyl-3-O-[β-D-glucopyranosyl(1→6)]-[angeloyl(1→3)]-β-D-4-O-acetyl-glucopyra-
4
nosyl-28-O-[α-L-rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-16-deoxybarringtogenol C.
+
1
White powder. HRESI-MS [M+Na] m/z 1295.6401 (calcd. 1295.6401 for C H NaO ); H-
6
3
100
26
1
3
NMR and C-NMR data see Tables 1 and 2.
.3.6 21-acetyl-3-O-[β-D-glucopyranosyl(1→6)]-[angeloyl(1→4)]-β-D-glucopyranosyl-28-O-[α-
4
L-rhamnopyranosyl(1→2)]-β-D-glucopyranosyl-16-deoxy-barringtogenol C.
+
1
White powder. HRESI-MS [M+Na] m/z 1253.6286 (calcd. 1253.6295 for C H NaO ); H-
6
1
98
25
1
3
NMR and C-NMR data see Tables 2 and 3.
4
.3.7 28-O-acetyl-21-O-β-D-fucopyranosyl barrigenol C.
+
1
White powder. HRESI-MS [M+Na] m/z 701.4238 (calcd. 701.4241 for C H NaO ); H-
3
8
62
10
1
3
NMR and C-NMR data see Tables 2 and 3.

4
.3.8 21-O-(3,4-di-O-angeloyl)-β-D-fucopyranosyl-R -barrigenol.
1
-
1
White powder. HRESI-MS [M-H] m/z 815.4953 (calcd. 815.4946 for C H O ); H-NMR
4
6
71 12
1
3
and C-NMR data see Tables 2 and 3.
4
.3.9 3-O-α-D-glucuronopyranosyl-21,22-di-O-angeloyl-R -barrigenol.
1
-
1
White powder. HRESI-MS [M-H] m/z 845.4661 (calcd. 845.4687 for C H O ); H-NMR
4
6
69 14
1
3
and C-NMR data see Tables 2 and 3.
.4 Acid hydrolysis
Compounds (each 2.0 mg) were dissolved in 4 M HCl (2 mL) and then heated in a H O bath
4
2
at 90 °C for 3 h to dryness, respectively. Mixture was extracted by CH Cl and get CH Cl extract.
2
2
2
2
The water layer was analyzed by HPLC to detect tR (min): D-galactose: (+) 11.42 min, L-arabinose:
+) 9.25 min, D-glucose: (+) 10.67 min, L-rhamnoose: (-) 7.33 min. The HPLC analysis was carried
(
using the colum of Kaseisorb LC NH -60-5 (4.6 mmI.D. × 250 mm) at room temperature with a
2
flow rate of 0.8 mL/min. Mobile phase was 75 % CH CN/H O. Optional detector: Shodex OR-2
3
2
(Range: 64; Atten: 2). The glucuronic acid was identified using GC method. Column: Supeluco
STBTM-1(30 m × 0.25 mmi.d.). Injetor temp.: 230. Detector temp.: 230. Column temp.:230. He
flow rate: 15 ml/min (methyl-D-glucuronic acid: 20.16 min). For compounds 1-2, 4-6, an alkaline
hydrolysis experiment was performed prior to the aforementioned acid hydrolysis to remove either
the methyl or the angeloyl group in the sugars (dissolved in 1.5 ml 5% KOH: Dioxane 1.5 ml, stirred
at 37°C for 1 h).
4
.5 Y-maze test in mice
In Y-maze test was carried out with the method as previously described[9]. The testing period
was 5 min. The spontaneous alternation behavior was calculated as the formula followed, T =
numbers of successful alternation/(the number of arm entries - 2)]×100 %.
[
4
.6 Morris water maze test in rats
The Morris water maze test was carried out with the reported methods[9]. Briefly, rats were
trained two trials a day for five consecutive days in a round water pool at 22 °C ± 1 °C. The escape
latency (spending time to find the platform) was recorded. On the 3th or 4th day, the rats were given
a probe test. The swimming time and the swimming distance in the 4th quadrant were calculated.
4
.7 Measurement of NO production
Accumulation of nitrite (NO
-), an indicator of NO synthase activity, in culture medium was
2
4
measured using the Griess reaction. Cells (5 × 10 cells/well) were plated on 96-well microtiter plates
and treated with each compound (0.1, 1, 10, 50 μM) in presence of bacterial lipopolysacchride (LPS; 100
ng/ml) for 24 h. Fifty microliter culture medium supernatants were mixed with 50 μl Griess reagent (part
I: 1% sulfanilamide; part II: 0.1% naphthylethylene diamide dihydrochloride and 2% phosphoric acid)
at room temperature. Fifteen minutes later, the absorbance was measured at 540 nm.
4
.8 Atom-based 3D-QSAR

The dataset (Table S1) was divided into training (80 %) and test (20 %) sets randomly (Table
S3) using the “Automated Random Selection” option of the PHASE software in the Schrodinger
2
3
013 package suite. The cleaning steps includes hydrogens attachment, conversion of 2D into
D, stereoisomers generation, and optically, charged structures neutralization or determination
of the most probable ionization state at a user-defined pH were performed using LigPrep with
OPLS 2005 force field incorporated in PHASE. One to seven representative conformers were
adopted depending on the ΔG value and the similarity of each conformer (for those that were
conformationally similar, only one conformer was adopted). The atom-based QSAR protocol was
used to construct the 3D-QSAR model with the default setting.
Acknowledgements
This work was supported partially by the National Natural Science Foundation of China
(
81872768, 81673323), LiaoNing Revitalization Talents Program (XLYC1807118), LiaoNing
BaiQianWan Talents Program (2018), Program for Liaoning Innovative Talents in University
LR2017043).
(
Disclosure statement
The authors declare no competing financial interest.
Reference:
[
1] G.G. Glenner, C.W. Wong, Alzheimer's disease: initial report of the purification and characterization
of a novel cerebrovascular amyloid protein, Biochem. Biophys. Res. Commun., 120 (1984) 885-890.
2] C.L. Masters, G. Simms, N.A. Weinman, G. Multhaup, B.L. McDonald, K. Beyreuther, Amyloid
plaque core protein in Alzheimer disease and Down syndrome, Proc. Natl. Acad. Sci. U. S. A., 82 (1985)
245-4249.
3] M. Menendez-Gonzalez, H.S. Padilla-Zambrano, G. Alvarez, E. Capetillo-Zarate, C. Tomas-Zapico,
[
4
[
A. Costa, Targeting Beta-Amyloid at the CSF: A New Therapeutic Strategy in Alzheimer's Disease,
Front. Aging Neurosci., 10 (2018) 100.
[4] S.Z. Liu, W. Cheng, J.W. Shao, Y.F. Gu, Y.Y. Zhu, Q.J. Dong, S.Y. Bai, P. Wang, L. Lin,
Notoginseng Saponin Rg1 Prevents Cognitive Impairment through Modulating APP Processing in
Abeta1-42-injected Rats, Curr Med Sci, 39 (2019) 196-203.
[
5] Z. Zhang, J. Yang, C. Liu, J. Xie, S. Qiu, X. Yang, C. Wu, Pseudoginsenoside-F11 alleviates cognitive
deficits and Alzheimer's disease-type pathologies in SAMP8 mice, Pharmacol. Res., 139 (2019) 512-
23.
6] E.H. Joh, I.A. Lee, D.H. Kim, Kalopanaxsaponins A and B isolated from Kalopanax pictus ameliorate
memory deficits in mice, Phytother. Res., 26 (2012) 546-551.
7] G. Jin, L. Zhu, P. Liu, Q. Xu, Y. Qi, X. Zhou, J. Xu, X. Ji, T. Chi, L. Zou, Xanthoceraside prevented
synaptic loss and reversed learning-memory deficits in APP/PS1 transgenic mice, J. Physiol. Sci., 69
5
[
[
(2019) 477-488.
[8] Y. Qi, X.F. Ji, T.Y. Chi, P. Liu, G. Jin, Q. Xu, Q. Jiao, L.H. Wang, L.B. Zou, Xanthoceraside
attenuates amyloid beta peptide1-42-induced memory impairments by reducing neuroinflammatory
responses in mice, Eur. J. Pharmacol., 820 (2018) 18-30.

[9] X.F. Ji, T.Y. Chi, P. Liu, L.Y. Li, J.K. Xu, Q. Xu, L.B. Zou, D.L. Meng, The total triterpenoid
saponins of Xanthoceras sorbifolia improve learning and memory impairments through against oxidative
stress and synaptic damage, Phytomedicine, 25 (2017) 15-24.
[
10] G. Li, J. Yu, L. Zhang, Y. Wang, C. Wang, Q. Chen, Onjisaponin B prevents cognitive impairment
in a rat model of D-galactose-induced aging, Biomed. Pharmacother., 99 (2018) 113-120.
11] S.W. Hong, D.H. Yoo, J.Y. Woo, J.J. Jeong, J.H. Yang, D.H. Kim, Soyasaponins Ab and Bb prevent
[
scopolamine-induced memory impairment in mice without the inhibition of acetylcholinesterase, J. Agric.
Food Chem., 62 (2014) 2062-2068.
[
12] W. Xiao, Y. Wang, P. Zhang, N. Li, S. Jiang, J.H. Wang, J. Huang, X. Li, Bioactive barrigenol type
triterpenoids from the leaves of Xanthoceras sorbifolia Bunge, Eur. J. Med. Chem., 60 (2013) 263-270.
13] D. Wang, D. Su, B. Yu, C. Chen, L. Cheng, X. Li, R. Xi, H. Gao, X. Wang, Novel anti-tumour
[
barringenol-like triterpenoids from the husks of Xanthoceras sorbifolia Bunge and their three
dimensional quantitative structure activity relationships analysis, Fitoterapia, 116 (2017) 51-60.
[14] S. Nakamura, G. Chen, S. Nakashima, H. Matsuda, Y. Pei, M. Yoshikawa, Brazilian Natural
Medicines. IV. New Noroleanane-Type Triterpene and Ecdysterone-Type Sterol Glycosides and
Melanogenesis Inhibitors from the Roots of Pfaffia glomerata, Chem. Pharm. Bull. (Tokyo), 58 (2010)
9
0-695.
15] W. Li, X. Li, J. Yang, D.L. Meng, N. Li, Two new triterpenoid saponins from the carpophore of
Xanthoceras sorbifolia Bunge, J. Asian Nat. Prod. Res., 10 (2008) 260-264.
16] S. Nakamura, G. Chen, S. Nakashima, H. Matsuda, Y. Pei, M. Yoshikawa, Brazilian natural
[
[
medicines. IV. New noroleanane-type triterpene and ecdysterone-type sterol glycosides and
melanogenesis inhibitors from the roots of Pfaffia glomerata, Chem. Pharm. Bull. (Tokyo), 58 (2010)
6
90-695.
17] D. Wang, D. Su, X.Z. Li, D. Liu, R.G. Xi, H.Y. Gao, X.B. Wang. Barrigenol triterpenes from
the husks of Xanthoceras sorbifolia Bunge and their antitumor activities. RSC Adv.6 (2016) 27434-
7446.
18] K.S. Bhadoriya, M.C. Sharma, S.V. Jain, Pharmacophore modeling and atom-based 3D-QSAR
[
2
[
studies on amino derivatives of indole as potent isoprenylcysteine carboxyl methyltransferase (Icmt)
inhibitors, J. Mol. Struct., 1081 (2015) 466-476.
[19] U.A. Shah, H.S. Deokar, S.S. Kadam, V.M. Kulkarni, Pharmacophore generation and atom-based
3
5
D-QSAR of novel 2-(4-methylsulfonylphenyl)pyrimidines as COX-2 inhibitors, Mol. Divers., 14 (2010)
59-568.
[
20] Z. Zhang, J. Yang, C. Liu, J. Xie, S. Qiu, X. Yang, C. Wu, Pseudoginsenoside-F11 alleviates
cognitive deficits and Alzheimer's disease-type pathologies in SAMP8 mice, Pharmacol. Res., 139 (2019)
12-523.
21] S.Z. Liu, W. Cheng, J.W. Shao, Y.F. Gu, Y.Y. Zhu, Q.J. Dong, S.Y. Bai, P. Wang, L. Lin,
5
[
Notoginseng Saponin Rg1 Prevents Cognitive Impairment through Modulating APP Processing in
Abeta1-42-injected Rats, Current medical science, 39 (2019) 196-203.
[22] Y. Guo, L.P. Wang, C. Li, Y.X. Xiong, Y.T. Yan, L.Q. Zhao, S.D. Li, J. Sun, H.Y. Luo, C.J. Xian,
Effects of Ginsenoside Rb1 on Expressions of Phosphorylation Akt/Phosphorylation
mTOR/Phosphorylation PTEN in Artificial Abnormal Hippocampal Microenvironment in Rats,
Neurochem. Res., 43 (2018) 1927-1937.
[23] J.M. Kim, C.H. Park, S.K. Park, T.W. Seung, J.Y. Kang, J.S. Ha, D.S. Lee, U. Lee, D.O. Kim, H.J.
Heo, Ginsenoside Re Ameliorates Brain Insulin Resistance and Cognitive Dysfunction in High Fat Diet-

Induced C57BL/6 Mice, J. Agric. Food Chem., 65 (2017) 2719-2729.
Figure 1. The structures of previous undescribed compounds 1-9 from X. sorbifolium. Ang
(angeloyl).
Figure 2. HMBC correlations of compounds 1-9.
Figure 3. Protective effects of compounds 2, 3, 10-17 against Aβ1-42-induced working memory
impairment in Y-maze test in mice. (A) Alternation behavior (%). (B)The number of arm entries.
#
∗
The data are presented as means ± SD. (n = 10-12). P < 0.05 compared with sham group; P
0.05 compared with model group.
<
Figure 4. Protective effects of compounds 2, 3, 10-17 against QA-induced spatial memory
impairment in learning period of water maze test in rats. The data are presented as means ± SD.
#
∗
(
n = 10-12). P < 0.05 compared with sham group; P < 0.05 compared with model group.
Figure 5. Results of the atom-based 3D-QSAR study. (A) Favorable hydrophobic/non-polar
features represented as yellow cubes. (B) Favorable electro-withdrawing features represented
as green cubes.
Figure 6. Protective effects of compounds 9 and 19 against Aβ1-42-induced working memory
impairment in Y-maze test in mice. Don:donepezil; Xan:xanthoceraside. (A) Alternation
behavior (%). (B)The number of arm entries. The data are presented as means ± SD. (n = 10-
#
∗
1
2). P < 0.05 compared with sham group; P < 0.05 compared with model group.
Figure 7. Protective effects of compounds 9 and 19 against QA-induced spatial memory
#
impairment. (A) Escape laterncy. The data are presented as means ± SD. (n = 10-12). P < 0.05
∗
compared with sham group; P < 0.05 compared with model group. (B) Swimming time in the
fourth quadrant where the platform has been locate. Don:donepezil; Xan:xanthoceraside. The
#
∗
data are presented as means ± SD. (n = 10-12). P < 0.05 compared with sham group; P < 0.05
compared with model group. (C) Characteristic swimming trails of the rats in probe test.

1
6
Table 1. H NMR data of 1-5 (DMSO-d )
No.
1
2
3
4
5
1
2
3
5
6
7
9
0.90 m/1.47 m
1.68 m
3.24 (dd, 4.6, 11.6)
0.88 (d, 8.4)
1.61 m
0.86 m/1.58 m
1.39 m/1.69 m
3.09 m
0.72 (d, 11.5)
1.46 m
1.52 m/1.88 m
1.53 m/1.64 m
3.06 m
0.70 (d, 11.8)
1.28 m/1.43 m
1.59 m/1.68 m
1.47 m
1.02 m/1.54 m
1.60 m
3.31 (dd, 4.2, 11.6) 3.36 (dd, 4.0, 11.5)
0.73 (d, 11.9)
1.32 m/1.41 m
1.26 m/1.44 m
1.62 m
1.03 m/1.54 m
1.61 m
0.76 (d, 11.6)
1.34 m/1.48 m
1.27 m/1.45 m
1.62 m
2.18 m
1.69 m
1.60 m/1.70 m
1.48 m
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
3
1
2
5
6
8
9
1
2
3
4
5
6
7
8
9
0
1.89 m
5.51 br s
1.83 m
5.37 br s
3.63 m
3.73 m
3.03 m
1.13 m/2.53 m
5.81 (d, 10.1)
5.55 (d, 10.1)
0.96 s
1.81 m
5.35 br s
3.62 m
3.72 m
3.06 m
1.12 m/2.52 m
5.77 (d, 10.0)
5.44 (d, 10.0)
0.97 s
1.90 m/2.08 m
5.60 br s
1.89 m/2.08 m
5.60 br s
4.21 overlap
4.40 overlap
3.08 m
1.45 m/3.09 m
6.68 (d, 10.3)
6.26 (d, 10.3)
1.29 s
1.23 m
2.01 m
1.24 m
2.02 m
2.91 (dd, 3.9, 13.7) 2.93 (dd, 3.9, 13.6)
1.29 m/2.15 m
5.38 (d, 10.3)
4.42 (d, 10.3)
1.25 s
1.30 m/2.16 m
5.39 (d, 10.3)
4.42 m
1.26 s
1.18 s
0.86 s
1.03 s
1.85 s
0.75 s
0.91 s
17.0
20.3
62.2
29.1
19.6
0.75 s
0.88 s
0.89 s
1.30 s
0.99 s
0.90 s
1.10 s
1.23 s
3.95 m
1.04 s
1.09 s
0.96 s
0.89 s
1.10 s
1.23 s
3.95 m
1.04 s
1.10 s
3.48 m/3.75 m
1.12 s
2.87 m/3.09 m
0.77 s
1.33 s
0.97 s
2
1-Ang
21-Ang
21-Ang
21-Ac
21-Ac
2
3
4
5
'
'
'
'
2.22 s
2.22 s
6.09 (br q, 7.1)
2.19 (br d, 7.1)
2.06 br s
6.01 (br q, 7.2) 6.07 (br q, 7.1)
1.81 (br d, 7.2) 1.88 (br d, 7.1)
1.76 br s
22-Ang
1.76 br s
22-IB
2
2-MB
3-O-Glc
4.88 (d, 7.6)
4.00 m
5.93 (t,9.4)
4.15 m
3-O-Glc
4.93 (d, 7.7)
4.07 m
5.88 m
5.48 (t, 9.8)
4.28 m
1
2
3
4
5
6
''
''
''
''
''
''
2.11 m
1.85 m
0.76 (t, 7.0)
0.98 (d, 7.0)
2.36 m
5.94 (br q, 7.2)
1.84 (br d, 7.2)
1.76 br s
1.01 (d,7.0)
0.96 (d,7.0)
4.15 m
4.37 m
4.12 m
3
-O-Glc A
3-O-Glc
4.77 (d, 2.7)
3.72 m
3.73 m
3.52 m
3-O-Glc A
4.29 (d, 7.7)
3.00 m
3.16 m
3.28 m
3.72 m
-
6''-O-glc
5.10 (d, 7.7)
4.06 (t.7.6)
3.93 m
4.23 m
4.03 m
6''-O-glc
4.99 (d, 7.6)
4.04 m
3.94 m
4.24 m
1
2
3
4
5
6
'''
'''
'''
'''
'''
'''
4.91 (d, 7.5)
4.33 m
4.34 m
4.27 m
4.33 m
-
3.58 m
3.39 m/3.49 m
4.04 m
4.52 m
4.52 m
OMe
3.74 s
-
3.66 s
-
-
2
'''-O-Gal
28-O-Glc
4.63 (d, 7.7)
4.27 m
28-O-Glc
4.64 (d, 7.9)
4.28 m
1
2
3
4
5
6
''''
''''
''''
''''
''''
''''
5.34 (d, 7.8)
4.41 m
4.11 m
4.56 m
4.44 m
4.44 m
4.35 m
4.23 m
3.73 m
4.39 m
4.36 m
4.24 m
3.74 m
4.40 m
3
'''-O-Ara
2''''-O-Rha
6.64 br s
4.62 m
4.68 brs
4.35 m
2''''-O-Rha
6.64 br s
4.63 m
4.69 brs
4.36 m
1
2
3
4
5
6
'''''
'''''
'''''
'''''
'''''
'''''
6.03 br s
4.96 m
4.78 m
4.80 m
4.14 m
-
4.85 m
1.87 (d, 6.1)
4.86 m
1.88 (d, 6.1)
3
''-Ang
3''-Ang
3
4
5
''''''
''''''
''''''
5.83 (br q, 7.2)
1.95 (br d, 7.2)
1.83 br s
5.89 (br q, 7.2)
1.96 (br d, 7.2)
1.88 br s
4
''-Ac
2
'''''''
1.99 s

Table 2. The 13C NMR data of 1-9 (DMSO-d
)
6
No.
1
2
3
4
5
6
7
8
1
2
38.1
21.9
82.4
38.0
54.9
18.1
35.7
40.6
46.3
36.6
23.2
124.5
142.6
46.6
66.3
72.8
46.6
39.4
46.1
35.4
77.9
71.9
28.4
16.8
15.5
17.0
20.3
62.2
29.1
19.6
21-Ang
166.5
127.8
136.7
15.3
20.3
-
3
38.4
25.7
88.1
38.7
54.7
18.0
35.6
40.6
46.2
36.3
23.1
124.4
142.6
46.6
66.2
72.2
46.6
39.4
46.0
35.4
77.6
71.9
27.5
16.5
15.5
17.0
20.3
62.3
29.0
19.6
21-Ang
166.4
127.6
137.5
15.4
20.4
-
4
39.8
26.0
89.7
39.1
56.0
18.8
33.1
40.7
48.2
37.2
24.5
124.5
142.8
42.1
26.6
19.3
43.8
41.0
46.2
36.5
79.9
68.5
28.5
17.3
16.1
17.0
26.9
73.9
30.0
20.6
21-Ac
171.5
21.5
-
5
39.8
26.0
90.0
39.1
56.1
18.9
33.1
40.7
48.2
37.2
24.5
124.5
142.8
42.1
26.6
19.3
43.8
41.0
46.2
36.5
79.9
68.5
28.4
17.3
16.1
17.0
26.9
73.8
30.0
20.6
21-Ac
171.5
21.6
-
6
39.9
26.0
89.7
39.1
56.0
18.9
33.1
40.7
48.2
37.2
24.4
124.6
142.8
42.1
26.9
18.9
43.8
41.0
46.2
36.5
79.9
68.6
28.5
17.4
16.1
17.0
26.6
73.8
30.0
20.6
21-Ac
171.5
21.6
-
7
8
9
38.2
24.5
87.1
38.3
54.5
18.1
35.6
40.6
46.3
36.3
23.1
124.5
142.5
46.6
66.3
72.8
46.6
39.8
46.1
35.4
77.9
71.9
27.9
16.5
15.5
17.0
20.3
62.2
29.0
19.5
21-Ang
166.5
127.8
136.7
15.3
20.3
39.3
26.9
90.4
40.0
56.0
19.2
37.1
41.8
47.5
37.3
24.4
125.8
144.1
48.1
67.9
73.5
48.9
41.3
47.2
36.8
79.1
73.8
28.3
17.1
16.5
17.9
21.5
63.5
29.9
20.6
1-Ang
167.9
129.0
139.1
16.2
21.6
-
2-MB
176.8
42.1
26.9
12.2
17.4
-
38.3
27.0
76.8
38.4
54.8
18.0
32.3
40.0
46.1
36.5
23.0
122.7
142.2
40.9
33.7
66.5
45.2
39.3
46.8
38.3
90.0
69.7
20.3
16.1
15.4
16.4
28.2
65.0
29.2
19.7
38.5
27.1
76.8
38.3
54.6
18.3
35.7
40.6
46.4
36.6
23.1
123.7
143.2
46.7
66.4
72.0
46.8
39.5
46.9
36.1
90.5
69.3
28.2
16.1
15.6
17.0
20.3
62.8
29.2
19.5
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
21-O-Fuc 21-O-Fuc
105.1
70.8
73.4
70.9
70.1
16.4
28-Ac
170.1
20.9
1
2
3
4
5
6
'
'
'
'
'
'
104.4
68.8
72.9
70.3
68.1
15.9
3'-Ang
166.2
127.2
137.9
15.4
19.9
-
4'-Ang
166.6
126.9
138.3
15.5
20.3
-
-
-
-
-
-
-
-
-
-
2
22-Ang
166.6
128.2
135.7
15.2
20.3
-
22-IB
175.4
33.6
18.7
18.9
-
3-O-Glc
107.2
73.8
79.6
70.0
77.1
70.4
6''-O-glc
105.9
75.5
78.9
72.0
78.8
63.0
-
3-O-Glc
106.9
73.4
76.4
71.1
74.0
69.6
6''-O-glc
105.6
75.5
78.9
72.2
78.8
63.0
-
3-O-Glc
107.2
76.3
76.1
72.9
75.0
69.9
6''-O-glc
105.3
75.6
78.8
72.2
78.8
63.0
-
22-Ang
166.6
128.2
135.7
15.2
1
2
3
4
5
6
''
''
''
''
''
''
20.3
-
3
-O-Glc A 3-O-Glc 3-O-Glc A
105.5
78.8
86.4
71.9
76.7
170.5
52.6
3-O-Glc A
111.6
78.6
82.4
77.1
1
2
3
4
5
6
'''
'''
'''
'''
'''
'''
96.0
68.8
71.6
69.6
68.2
60.4
-
105.5
73.7
75.9
71.7
75.3
169.7
51.9
69.2
174.7
OMe
-
2
'''-O-Gal
105.2
73.3
75.5
70.1
28-O-Glc
103.7
80.7
74.6
72.2
78.0
63.1
2''''-O-Rha
100.7
72.8
73.0
74.7
68.9
19.1
28-O-Glc
103.7
80.8
74.7
72.0
78.0
63.1
28-O-Glc
103.7
80.7
75.6
71.9
78.0
63.1
1
2
3
4
5
6
''''
''''
''''
''''
''''
''''
77.1
62.2
3
'''-O-Ara
111.5
84.0
78.0
85.8
62.7
-
2''''-O-Rha 2''''-O-Rha
1
2
3
4
5
6
'''''
'''''
'''''
'''''
'''''
'''''
100.7
72.8
73.0
74.7
68.9
100.7
72.8
73.0
74.7
68.9
19.1
19.3
3
''-Ang
3''-Ang
168.0
128.7
137.9
16.2
4''-Ang
168.2
128.7
138.3
16.3
1
2
3
4
5
''''''
''''''
''''''
''''''
''''''
168.4
129.3
137.3
16.3
21.1
21.0
21.1
4
''-Ac
1
2
'''''''
'''''''
170.5
21.1

1
6
Table 3. H NMR data of 6-9 (DMSO-d )
No.
6
7
8
9
1
2
3
5
6
7
9
1.07 m/1.54 m
1.06 m/1.67 m
3.26 (dd, 4.4, 11.5)
0.72 (d, 11.9)
1.35 m/1.48 m
1.31 m/1.45 m
1.60 m
0.89 m/1.50 m
1.44 m
0.89 m/1.52 m
1.44 m
0.86/1.50 m
1.36 m
2.99 m
0.67 m
2.99 m
0.66 m
2.99 (dd, 4.3, 11.6)
0.70 (d, 11.9)
1.28 m/1.41 m
1.59 m/1.69 m
1.46 m
1.82 m
5.37 br s
3.62 m
3.72 m
2.40 m
1.13 m/1.52 m
5.81 (d, 10.2)
5.53 (d, 10.2)
0.88 s
1.31 m/1.47 m
1.25 m/1.47 m
1.53 (d, 8.9)
1.79 (dd, 3.2, 8.9)
5.21 br s
1.61 m/1.22 m
4.08 m
2.34 (dd, 4.3, 14.2)
0.95 (d, 4.3)/2.50 m
3.92 (d, 9.3)
3.60 (d, 9.3)
1.36 s
1.29 m/1.43 m
1.59 m/1.68 m
1.53
1.80 (dd, 5.6)
5.29 br s
3.6 br s
3.8 br s
2.40 m
0.96 m/2.41m
3.99 (d, 9.5 )
3.75 (d, 9.5 )
0.89 s
0.67 s
1.05 s
0.90 s
1.26 s
3.04/3.17
0.96 s
0.91 s
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
3
1
2
5
6
8
9
1
2
3
4
5
6
7
8
9
0
1.94 m/2.10 m
5.60 br s
1.20 m
1.87 m
2.92 (dd, 3.5, 13.7)
1.30 m/2.15 m
5.38 (d, 10.3)
4.43 m
1.25 s
1.01 s
0.87 s
1.10 s
1.25 s
3.95 m
1.04 s
1.10 s
0.68 s
0.87 s
0.84 s
0.90 s
0.63 s
0.85 s
0.89s
1.31 s
3.74/3.64 (d, 10.6)
0.97 s
2.89 /3.11(d, 3.6, 10.2)
0.78 s
0.92 s
1.00 s
2
1-Ac
-
21-O-Fuc
4.08 (d, 7.4)
3.31 m
21-O-Fuc
4.40
3.55 m
21-Ang
1
2
3
4
5
6
'
'
'
'
'
'
-
-
2.23 s
-
-
-
-
3.31 m
3.40 br s
3.59 m
1.10 (d, 6.4)
28-Ac
5.04
5.94 (br q, 7.0)
0.85 (br d, 7.0)
1.76 s
5.22 (d, 3.6)
4.08 (6.4, 12.3)
1.04 (d, 6.4)
3'-Ang
-
3
-O-Glc
4.96 (d, 7.7)
4.00 m
22-Ang
1
2
3
4
5
6
''
''
''
''
''
''
-
-
-
-
-
1.99 s
4.06 m
5.59 (t, 9.7)
4.30 m
-
-
-
-
6.08
1.92
1.75 br s
-
6.01 (dq, 7.2)
1.83 (dq, 7.2)
1.76 s
4.47 m
6
''-O-glc
4'-Ang
-
-
3-O-Glc A
5.00 br s
4.68 (t, 4.1)
4.66 (dd, 4.1)
4.05 m
4.46 m
-
-
1
2
3
4
5
6
'''
'''
'''
'''
'''
'''
5.04 (d, 7.7)
4.06 m
3.93 m
4.24 m
4.03 m
4.47 m
-
6.19 (br q, 7.3)
1.89 (br d, 7.3)
1.87 br s
-
OMe
-
2
8-O-Glc
1
2
3
4
5
6
''''
''''
''''
''''
''''
''''
4.64 (d, 7.1)
4.28 m
4.40 m
4.24 m
3.73 m
4.40 m
2
''''-O-Rha
1
2
3
4
5
6
'''''
'''''
'''''
'''''
'''''
'''''
6.64 br s
4.63 m
4.70 brs
4.36 m
4.86 m
1.80 (d, 6.0)
4
''-Ang
3
4
5
''''''
''''''
''''''
5.88 (br q, 7.1)
2.01 (br d, 7.1)
1.90 br s


Nine undescribed (1-9) triterpene saponins (10-19) were isolated from Xanthoceras
sorbifolium



In vivo evaluations were conducted regarding the anti-AD effects
The atom based 3D QSAR model was established for the anti-AD effects
19 was predicted and proved to be equivalently effective as xanthoceraside and donepezil

Declaration of interests
☒
The authors declare that they have no known competing financial interests or personal
relationships that could have appeared to influence the work reported in this paper.
☐The authors declare the following financial interests/personal relationships which may be
considered as potential competing interests: