September 2007 Biol. Pharm. Bull.

30(9) 1727—1731 (2007) 1727

Effect of Polygala tenuifolia Root Extract on Scopolamine-Induced

Impairment of Rat Spatial Cognition in an Eight-Arm Radial Maze Task
Xiu-Lan SUN,a Hideyuki ITO,a Takayoshi MASUOKA,b Chiaki KAMEI,b and Tsutomu HATANO*,a
a
Department of Pharmacognosy, Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical
Sciences; Tsushima, Okayama 700–8530, Japan: and b Department of Medicinal Pharmacology, Okayama University
Graduate School of Medicine, Dentistry and Pharmaceutical Sciences; Tsushima, Okayama 700–8530, Japan.
Received March 17, 2007; accepted June 5, 2007; published online June 11, 2007

The effects of Polygala tenuifolia root fractions and the acyl groups of its constituents on the retrieval
process of spatial cognition in rats were studied using an eight-arm radial maze task. Oral administration of a
precipitate fraction (PTB) obtained by concentration of the n-BuOH-soluble portion from the extract of the roots
significantly decreased the number of total errors (TEs) and that of working memory errors (WMEs) at doses of
100 mg/kg and 200 mg/kg. However, it caused no significant decrease in the number of reference memory errors
(RMEs). In addition, the saponin-rich fraction (PTBM) obtained by purification of PTB also showed significant
decreases in TEs and WMEs at a dose of 100 mg/kg. Among the cinnamic acid derivatives present as the acyl
groups in the P. tenuifolia constituents, sinapic acid (SNPA) significantly decreased TEs and WMEs at doses of 10
to 100 mg/kg. These results indicated that P. tenuifolia extracts, PTB and PTBM, and SNPA had a beneficial
effect on the memory impairment induced by dysfunction of the cholinergic system in the brain. The memory
improvement in the scopolamine-induced memory impairment seen in the radial maze performance was due to
improvement in the short-term memory. A contribution of some constituents other than SNPA to the memory
improvement was also suggested.
Key words Polygala tenuifolia; spatial cognition; eight-arm radial maze task; sinapic acid; saponin

Since senile dementia of the Alzheimer type and multi-in-
farct dementia accompanying the increase in aging popula-
tions has become a social problem, the development of effec-
tive drugs for treating these conditions is vital. Among the
traditional Chinese natural medicines, Polygala root (Japan-
ese name: onji)—the root of Polygala tenuifolia WILLDENOW
(family Polygalaceae)—may improve memory, since some
Chinese prescriptions containing Polygala root have been
used to treat amnesia, neurasthenia, palpitation, and insom-
nia.1) Recent studies revealed that P. tenuifolia roots are ef-
fective for memory impairment in rats and mice, and scopol-
amine-induced impairment of passive avoidance response
was improved by oral administration of water extract of Poly-
gala root in mice.2) Park et al. also reported that ethanol ex-
tract from Polygala root (BT-11) was effective against scopol-
amine-induced cognitive impairments in passive avoidance
response and water maze tests.3) Tenuifoliside B, one of the
saccharides containing acyl groups isolated from P.
tenuifolia, had an ameliorative effect on scopolamine-in-
duced impairment in the performance of passive avoidance
tasks in rats.4)
The eight-arm radial maze task is regarded as more useful
for studying the effects of drugs on the learning process and
memory than other methods such as passive and active
avoidance tasks5—7) because it estimates the process of spa-
tial cognition. Moreover, drugs used clinically to treat de-
mentia also display effectiveness in the radial maze task.8)
Therefore, we studied the effects of a precipitate fraction
(PTB) obtained from an n-BuOH extract of P. tenuifolia on
the retrieval process of spatial cognition in rats using the
eight-arm radial maze with four-arms baited. We also investi-
gated the effects of cinnamic acid derivatives, which are
present as the acyl groups of the constituents of PTB, and
a purified saponin-rich fraction (PTBM) on scopolamine-
induced memory impairment in this experimental system.
∗ To whom correspondence should be addressed. e-mail: hatano@pharm.okayama-u.ac.jp
MATERIALS AND METHODS
Animals Male Wistar rats, 7 weeks old (body mass,
200—220 g), were purchased from Japan SLC, Shizuoka,
Japan. Animals were maintained in an air-conditioned room
with controlled temperature (24
2 °C) and humidity (55%

15%). They were housed in plastic cages with sawdust and
kept under a light–dark cycle (lights on from 07:00 to 19:00).
The animals were allowed free access to food and water ex-
cept during the experiments. All procedures involving ani-
mals were carried out in accordance with Guidelines of the
Animal Experiments at Okayama University Advanced Sci-
ence Research Center.
Eight-Arm Radial Maze Test The apparatus and the
procedure used have been described in previous reports.9—11)
Rats received one daily habituation session for 3 d before
training to familiarize them with the radial maze. Food pel-
lets (45 mg each, Bio-Serv, A Holton Industries, Frenchtown,
NJ, U.S.A.) were scattered over the entire maze surface, and
three or four rats were simultaneously placed in the radial
maze and allowed to take them freely. A pellet was then
placed in the food cup in each of the eight arms, and each of
the rats was allowed to explore it until it had taken all the
pellets during the next 2 d. All rats were trained with one trial
per day.
Four of the eight arms were baited for the trial and the se-
quence of the arms was not changed throughout the test. The
center platform on which the rat was placed was closed off
by a door. The door was opened after 20 s and the rat was al-
lowed to choose an arm to obtain food pellets until all four
pellets had been taken or 10 min had elapsed. Eighteen rats
were trained in this way until achieving a criterion of no
more than one error per trial for five successive trials. It took
30—50 days to achieve the criterion. After reaching this cri-
terion, scopolamine was injected intraperitoneally on the fol-
© 2007 Pharmaceutical Society of Japan
1728
Fig. 1. Extraction of Polygala tenuifolia Roots and Fractionation of the
Extract
lowing day. The test drugs were administered 30 min before
the scopolamine treatment, and test trials were undertaken 30
min posttreatment.
The total error (TE) was defined as the number of entries
into the unbaited arms. The reference memory error (RME)
and working memory error (WME) were defined as the first
entry to never-baited arms, and reentry to arms where the
pellet had already been taken, respectively.
Extraction and Fractionation of P. tenuifolia Roots
Dried roots (500 g) of P. tenuifolia purchased from Tochi-
moto-tenkai-do, Osaka, Japan, were pulverized and extracted
three times with MeOH at room temperature. The combined
crude extract (141 g) was dissolved in H2O and successively
extracted with n-hexane, EtOAc, and n-BuOH (Fig. 1). These
solutions were concentrated to give the n-hexane extract
(10.2 g), EtOAc extract (5.5 g), and a water-soluble portion
(63 g). The n-BuOH solution was concentrated to give PTB
as the precipitate and the precipitate (27 g) was separated
from the supernatant (9.6 g) by centrifugation (300g for 5
min).
HPLC Analysis of the Hydrolysate of PTB The PTB
fraction (1 g) was dissolved in 5% NaOH (15 ml) and the so-
lution was refluxed for 1.5 h (98 °C).12) The reaction mixture
was then analyzed by HPLC [column, YMC-pack ODS A-
302 (internal diameterlength, 4.6 mm150 mm); mobile
phase, CH3CN–H2O–HCOOH (30 : 70 : 1), or 0.01 M H3PO4–
0.01 M KH2PO4–MeOH (40 : 40 : 20); column temp., 40 °C;
flow rate, 1.0 ml/min; detection, UV 205 nm or 230 nm; in-
strument, model LC-10AD (Shimadzu, Kyoto, Japan)] to de-
tect the presence of 3,4,5-trimethoxycinnamic acid (TMCA),
p-methoxycinnamic acid (PMCA), sinapic acid (SNPA), and
tenuifolin (= presenegenin 3b -O-glucoside), the major prosa-
pogenin in the saponins of P. tenuifolia roots (Fig. 2, Table
1). The presence of tenuifolin was further substantiated by
isolation from the reaction mixture by preparative TLC [Sil-
ica gel 60F254 (Merck, Darmstadt, Germany)] with the lower
phase of CHCl3–MeOH–H2O (7 : 3 : 1), and the identity was
confirmed by 1H-NMR spectroscopy.
Vol. 30, No. 9
Fig. 2. Chemical Structures of SNPA, TMCA , PMCA, and Tenuifolin
Table 1. Contentsa) of SNPA, PMCA, TMCA, and Tenuifolin in PTB and
PTBM
Compounds PTB PTBM
SNPA 1.2 —b)
PMCA 3.4 4.2
TMCA 4.0 5.0
Tenuifolin 3.3 4.7
a) w/w % based on the PTB or PTBM fraction. b) Not detected.
Separation of PTB Part of the PTB fraction (8 g) was
chromatographed on Diaion HP-20 with increasing concen-
trations of MeOH in water (0%→20%→40%→60%→
100%) to give PTB0, PTB20, PTB40, PTB60, and PTBM,
respectively. A portion (40 mg) of the PTB60 (716.1 mg) was
further purified by preparative TLC [Silica gel 60F254
(Merck)] with the upper layer of n-BuOH–AcOH–H2O (4 :
1 : 5) to produce a compound (5.6 mg), which was identified
as 3,6-di-O-sinapoylsucrose by comparing its 1H- and 13C-
NMR spectral data with those in the literature.12) The identity
was confirmed by the electrospray-ionization mass spec-
troscopic data, and the heteronuclear multiple-bond con-
nectivity (HMBC), nuclear Overhauser effect spectrosco-
py (NOESY), heteronuclear single quantum correlation
(HSQC), and 1H–1H correlation spectroscopy (1H–1H COSY)
analyses. The 100% MeOH elute from the Diaion column
was concentrated to give a saponin-rich fraction (PTBM, 4.8
g). The PTBM fraction was treated with alkali as in the case
of PTB, and the reaction mixture was analyzed by HPLC to
show the presence of TMCA, PMCA, and tenuifolin (Table
1).
Administration of Polygala Root Fractions and Hy-
droxycinnamic Acid Derivatives The PTB and PTBM
fractions were dissolved in 0.9% physiological saline and ad-
ministered orally (p.o.) 60 min before each session. Scopol-
amine hydrobromide (Nacalai Tesque, Kyoto, Japan) was dis-
solved in 0.9% physiological saline and 0.5 mg/kg scopol-
amine was injected intraperitoneally (i.p.) 30 min before the
session. TMCA (Sigma, St. Louis, MO, U.S.A.), PMCA
(Nacalai Tesque), and SNPA (Tokyo Kasei, Tokyo, Japan)
were suspended in 5% gum arabic and administered p.o. 60
min before each session.
Data Analysis Each value is shown as the mean

S.E.M. One-way analysis of variance (ANOVA) with Dun-
nett’s test or the Mann–Whitney U-test was used for statisti-
September 2007 1729

cal analysis of the results of radial maze performance.

RESULTS AND DISCUSSION

Scopolamine-Induced Memory Impairment and Effect

of PTB on Radial Maze Performance The polar PTB pre-
cipitate, obtained by concentration of an n-BuOH extract of
P. tenuifolia, was considered to be a glycoside-rich fraction
(Fig. 1), since P. tenuifolia contains large amounts of polar
glycosides.4,14) The presence of glycosides was confirmed by
hydrolysis of PTB to yield tenuifolin.
Scopolamine administration (0.5 mg/kg i.p.) increased
both RMEs and WMEs in the radial maze task with four-
arms baited. Oral administration of PTB improved TEs and
WMEs and the decreases in these errors were dose-depend-
ent. At doses of 100 mg/kg and 200 mg/kg, PTB significantly
decreased TEs and WMEs, while PTB at a dose of 50 mg/kg
produced no significant decreases in TEs and WMEs. More-
over, it showed no significant effect on RMEs at any dose.
Since WMEs and RMEs correspond to short-term and
long-term memory, respectively,15) the results indicated that
PTB has an ameliorating effect on short-term memory, but
not long-term memory (Fig. 3). This is the first report on the
improvement effect of a glycoside-rich fraction from P.
tenuifolia on the retrieval process in spatial cognition using
eight-arm radial maze assay systems. The effects were lim-
ited to short-term memory in the experimental system used.
Although the ameliorating effects of the crude extract of P.
tenuifolia on memory in a passive avoidance response exper-
imental system and in a water maze system have been previ-
ously reported,2,3) these systems did not discriminate between
these two types of errors.
The Analysis of the Constituent Acyl Groups of PTB Fig. 3. Effect of PTB on Spatial Memory Deficits Induced by Scopol-
Since PTB improved WMEs and TEs in the radial maze test, amine in the Eight-Arm Radial Maze
this fraction was analyzed after hydrolysis to determine its The effect was assessed using the TE (A), RME (B), and WME (C). Scopolamine
(0.5 mg/kg, i.p.) was administrated 30 min before and PTB was administered p.o. 1 h
constituent acyl groups. HPLC analysis of the hydrolysate before the test trial at the doses indicated at the bottom of each column. Each value
showed the presence of TMCA, PMCA, and SNPA, in addi- represents the mean
S.E.M. (n12). ## p0.01: compared to the vehicle (saline);
∗ p0.05; ∗∗ p0.01: significant differences compared to the scopolamine-treated
tion to tenuifolin in the reaction mixture (Table 1). The pres- group.
ence of SNPA in PTB was further verified by isolation of
3,6-di-O-sinapoylsucrose from PTB. Since these cinnamic marked decrease in TEs, while the improvement at a dose of
acid derivatives revealed various pharmacological effects on 100 mg/kg was weaker. The bell-shaped effects have often
the central nervous system (CNS),16—18) we further inves- been observed in CNS experimental systems after admin-
tigated the effects of these compounds on scopolamine- istration of natural medicines, e.g., Angelica acutiloba
induced memory impairment. (SIEBOLD & ZUCCARINI) KITAGAWA (family Umbelliferae)20)
Effects of Cinnamic Acid Derivatives: the Constituents and ginsenosides.21) Some cholinergic drugs have also shown
of P. tenuifolia on Radial Maze Performance As shown similar effects.22—24)
in Fig. 4, a significant improvement on scopolamine-induced Some CNS effects, such as the prolongation of hexobarbi-
memory impairment was observed for SNPA under a wide tal sleeping time in mice,17) were also reported for the cin-
range of doses: SNPA decreased TEs and WMEs at doses of namic acid derivatives PMCA and TMCA. However, these
10 to 100 mg/kg. These results show that SNPA participated compounds caused no improvement in the cognitive impair-
in the cognitive improvement of PTB on TEs and WMEs. ment in the present experiment systems: PMCA slightly de-
SNPA may enhance neurotransmitters in the central creased TEs and WMEs in radial maze performance at doses
cholinergic nervous system, since an increase in choline of 20 to 100 mg/kg, but the decreases were not statistically
acetyl transferase activity was reported18) in the cerebral cor- significant. TMCA produced no significant effect on TEs and
tex of basal forebrain-lesioned rats from the long-term ad- WMEs at doses of 20 to 100 mg/kg.
ministration of SNPA. Further investigation into the relation- Effect of PTBM Obtained by Purification of PTB on
ship between the levels of these substances and improvement the Radial Maze Performance Since tenuifolin was de-
of spatial cognition is needed.19) tected in the hydrolysate of PTB, saponin (triterpene glyco-
The relationships between the doses of SNPA and the im- side) structures or their aglycone residues other than the acyl
provement in TEs and WMEs showed an inverse bell-shaped groups (i.e., cinnamic acid derivatives) may also contribute
form: SNPA at a dose of 20 mg/kg produced the most to the improvements produced by PTB. PTB was therefore
1730
Fig. 4. Effects of SNPA, TMCA, and PMCA on Spatial Memory Deficits
Induced by Scopolamine in the Eight-Arm Radial Maze
The effect was assessed using the TE, RME, and WME. Scopolamine (0.5 mg/kg,
i.p.) was administrated 30 min before and the SNPA (A), TMCA (B), and PMCA (C)
were administered p.o. 1 h before the test trial at the doses indicated at the bottom of
each column. Each value represents the mean
S.E.M. (n10—12). ## p0.01: com-
pared to the vehicle (gum arabic); ∗ p0.05: significant differences compared to the
scopolamine-treated group. , TE; , RME; , WME.
further purified in a Diaion HP-20 column to give PTBM as
the MeOH eluate. The quantitative analysis of the reaction
mixture obtained by alkaline hydrolysis of PTBM indicated
that the content of tenuifolin was higher than that in the hy-
drolysate of PTB (Table 1).
The improvement in spatial cognition with PTBM was
more pronounced than that observed for PTB: PTBM caused
a significant decrease in TEs and WMEs at a dose of 100
mg/kg (p0.01). By contrast, PTBM showed no significant
effect on RMEs at any dose. Analysis of the hydrolysate of
PTBM indicated that the major constituent acids were
PMCA and TMCA. Therefore, the cognitive improving ac-
tivity of PTB could be due to a residue or residues other than
these acyl groups. However, the effect of PTBM at a dose of
200 mg/kg was weaker than that observed for a dose of 100
mg/kg (Fig. 5).
Vol. 30, No. 9
Fig. 5. Effect of PTBM on Spatial Memory Deficits Induced by Scopol-
amine in the Eight-Arm Radial Maze
The effect was assessed using the TE (A), RME (B), and WME (C). Scopolamine
(0.5 mg/kg, i.p.) was administrated 30 min before and PTBM was administered p.o. 1 h
before the test trial at the doses indicated at the bottom of each column. Each value
represents the mean
S.E.M. (n12). ## p0.01: compared to the vehicle (saline);
∗∗ p0.01: significant differences compared to the scopolamine-treated group.
Effect of PTB, PTBM, and Cinnamic Acid Derivatives
on the Running Time per Choice of Rats in Radial Maze
Performance In this study, the running time per choice
was measured as an index of the locomotor activity. As show
in Table 2, PTBM significantly suppressed the increase in
running time per choice induced by scopolamine. Con-
versely, PTB and the cinnamic acid derivatives did not affect
the running time per choice significantly compared to the
scopolamine-treated group. These results suggest that the ef-
fects of PTB and SNPA observed in memory improvement
tests were not directly related to the effects on the locomotor
activity.
CONCLUSIONS
In this study, PTB from the n-BuOH fraction of P. tenuifo-
lia had a significant ameliorating effect on scopolamine-
induced memory impairment in the spatial cognition process
of rats. The effect was especially pronounced in short-term
memory, but not in long-term memory. The HPLC analysis
of hydrolysate of PTB suggests that the effect of PTB might
be due to highly polar compounds. SNPA, which is present
September 2007 1731

Table 2. Effects of Polygala tenuifolia Fractions and Compounds Struc-
turally Related to the Constituents on the Running Time per Choice of Rats
in Radial Maze Performance
(A) Effect of PTB and PTBM on the Running Time per Choice of Rats in
Radial Maze Performance
was due to the improvement of the short-term memory, but
did not involve long-term memory. Therefore, P. tenuifolia
may represent a potential treatment for the early phase of
Alzheimer’s disease.

Running time REFERENCES AND NOTES

Treatment Dose (mg/kg) Scopolamine
per choice (s)
1) Chang H.-M., But P. P.-H., “Pharmacology and Applications of Chi-
Saline  6.0
0.5 nese Material Medica,” Vol. 1, ed. by Word Scientific Publishing, Sin-
Control (Saline)  18.8
2.0# gapore, 1986, pp. 551—553.
PTB 50  13.5
2.6 2) Egashira N., Yuzurihara M., Hattori N., Sakakibara I., Ishige A., Phy-
PTB 100  13.2
1.2 tomedicine, 10, 467—473 (2003).
PTB 200  13.2
1.1 3) Park C. H., Choi S. H., Koo J. W., Seo J. H., Kim H. S., Jeong S. J.,
PTBM 50  12.5
2.0* Suh Y. H., J. Neurosci. Res., 70, 484—492 (2002).
PTBM 100  11.2
1.0** 4) Ikeya Y., Takeda S., Tunakawa M., Karakida H., Toda K., Yamaguchi
PTBM 200  11.8
1.4* T., Aburada M., Biol. Pharm. Bull., 27, 1081—1085 (2004).
5) Levin E. D., Neurosci. Biobehav. Rev., 12, 169—175 (1988).
Each value represents the mean
S.E.M. (n12). # p0.01 compared to the non- 6) Okada M., Tamura A., Urae A., Nakagomi T., Kirino T., Mine K.,
scopolamine-treated group. ∗ p0.05; ∗∗ p0.01: compared to the control (scopol- Fujiwara M., J. Cereb. Blood Flow Metab., 15, 505—512 (1995).
amine-treated group) value.
7) Olton D. S., Becer J. T., Behav. Brain Sci., 2, 313—365 (1979).
(B) Effect of Cinnamic Acid Derivatives on the Running Time per Choice of 8) Ogawa N., Haba K., Sora Y. H., Higashida A., Sato H., Ogawa S.,
Rats in Radial Maze Performance Clin. Therapeutics, 11, 704—711 (1988).
9) Chen Z., Sugimoto Y., Kamei C., Pharmacol. Biochem. Behav., 51,
Dose Running time 513—518 (1999).
Treatment Scopolamine 10) Taga C., Sugimoto Y., Nishiga M., Fujii Y., Kamei C., Eur. J. Pharma-
(mg/kg) per choice (s)
col., 423, 167—170 (2001).
Gum Arabic  5.9
1.7 11) Nishiga M., Kamei C., Psychopharmacology, 166, 360—365 (2003).
Control (Gum Arabic)  17.0
4.9# 12) Pelletier S. W., Nakamura S., Soman R., Tetrahedron, 27, 4417—4427
SNPA 100  11.6
1.8 (1971).
TMCA 100  12.4
2.0 13) Ikeya Y., Sugama K., Okada M., Mitsuhashi H., Chem. Pharm. Bull.,
PMCA 100  10.4
1.1 39, 2600—2605 (1991).
14) Wang S. S., Saito K., Owada E., Kano Y., J. Trad. Med., 12, 102—108
Each value represents the mean
S.E.M. (n12). # p0.01: compared to the non- (1995).
scopolamine-treated group. 15) Kinoshameg S. E., Persinger M. A., Psychol. Rep., 91, 729—730
(2002).
16) Kawashima K., Miyako D., Ishino Y., Makino T., Saito K., Kano Y.,
as the sinapoyl moiety in the glycosides,25) showed a signifi- Biol. Pharm. Bull., 27, 1317—1319 (2004).
17) Wang S. S., Kozuka K., Saito K., Kano Y., J. Trad. Med., 11, 44—49
cant improvement in TEs and WMEs. The participation of (1994).
other parts of the glycosides such as tenuifolin in these im- 18) Abe T., Iizuka S., Komatsu Y., Yamada H., Phytomedicine, 4, 199—
provements was also suggested. 205 (1997).
Since the mechanism underlying the memory-enhancing 19) The effect of SNPA on the CNS was also reported very recently.
effect of P. tenuifolia root extracts was correlated with the in- Karakida F., Ikeya Y., Tstunakawa M., Yamaguchi T., Ikarashi Y.,
Takeda S., Aburada M., Biol. Pharm. Bull., 30, 514—519 (2007).
duction of choline acetyl transferase activity18) and the inhi- 20) Hatip-Al-Khatib I., Egashira N., Mishima K., Iwasaki K., Iwasaki K.,
bition of acetylcholine esterase activity,3) the improving ef- Kurauchi K., Inui K., Ikeya T., Fujiwara M., J. Pharmacol. Sci., 96,
fect of PTB on scopolamine-induced memory impairment 33—41 (2004).
shown here may be correlated with enhancement of the cen- 21) Yamaguchi Y., Haruta K., Kobayashi H., Psychoneuroendocrinology,
tral cholinergic system. In the early phase of Alzheimer’s dis- 20, 645—653 (1995).
22) Flood J. F., Cherkin A., Behav. Neural Biol., 45, 169—184 (1986).
ease, the memory deficits are thought to involve short-term, 23) Matsuoka N., Maeda N., Ohkubo Y., Yamaguchi I., Brain Res., 559,
but not long-term memory,26) and the spatial cognitive im- 233—240 (1991).
pairment induced by anticholinergic agents corresponds to 24) Yamamoto Y., Adachi A., Fujii Y., Kamei C., Brain Res., 1129, 161—
dementia of the Alzheimer type. Our results indicate that the 165 (2007).
25) Ikeya T., Ueno A., Shoyakugaku Zasshi, 47, 267—278 (1993).
memory improvement of PTB in the scopolamine-induced
26) Carmela G., Glynda J., Neuropsychol. Rev., 15, 1—10 (2005).
memory impairment seen in the radial maze performance