Structure-activity relationship for (+)-taxifolin isolated from silymarin as an inhibitor of amyloid β aggregation

Article abstract of DOI:10.1271/bbb.120925

Silymarin, the seed extract of Silybium marianum, has preventive effects against Alzheimer's disease-like pathogenesis in vivo. We isolated (+)-taxifolin (4) from silymarin as an inhibitor of aggregation of the 42- residue amyloid -protein. Structure-activity relationship studies revealed the 30,40-dihydroxyl groups to be critical to the anti-aggregative ability, whereas the 7-hydroxyl group and the stereochemistry at positions 2 and 3 were not important.

Full text of DOI:10.1271/bbb.120925

Biosci. Biotechnol. Biochem., 77 (5), 1100–1103, 2013
Note
Structure–Activity Relationship for (þ)-Taxifolin Isolated from Silymarin
as an Inhibitor of Amyloid ꢀ Aggregation
Mizuho SATO,1 Kazuma MURAKAMI,1 Mayumi UNO,1 Haruko IKUBO,1 Yu NAKAGAWA,1;2
y
1;
Sumie KATAYAMA,3 Ken-ichi AKAGI,3 and Kazuhiro IRIE
1Division of Food Science and Biotechnology, Graduate School of Agriculture, Kyoto University,
Kyoto 606-8502, Japan
2Synthetic Cellular Chemistry Laboratory, RIKEN Advanced Science Institute, Saitama 351-0198, Japan
3National Institute of Biomedical Innovation, Osaka 567-0085, Japan
Received December 3, 2012; Accepted February 14, 2013; Online Publication, May 7, 2013
[doi:10.1271/bbb.120925]
Silymarin, the seed extract of Silybium marianum, has
preventive effects against Alzheimer’s disease-like
pathogenesis in vivo. We isolated (þ)-taxifolin (4) from
silymarin as an inhibitor of aggregation of the 42-
residue amyloid ꢀ-protein. Structure-activity relation-
ship studies revealed the 30,40-dihydroxyl groups to be
critical to the anti-aggregative ability, whereas the
7-hydroxyl group and the stereochemistry at positions
2 and 3 were not important.
against Aꢀ42 aggregation; this is defined as the change
of the Aꢀ42 monomer into amyloid fibril by way of an
oligomer and protofibril.
Silymarin (lot no. 228–216; LKT Laboratories, St.
Paul, MN, USA) was fractionated by column chroma-
tography, eluting with 5% MeOH/CHCl3 on Wakogel
C-200 (Wako, Osaka, Japan), to give two major
fractions containing flavonoids. The first fraction was
chromatographed by high-performance liquid chroma-
tography (HPLC) in a YMC-Pack ODS-A column
(20 mm i.d. ꢁ 150 mm; YMC, Kyoto, Japan), using
50% MeOH/H2O, to yield silibinin A (1, 16 mg from
240 mg of silymarin, 6.7%)16) and silibinin B (2, 25 mg
from 240 mg of silymarin, 10%),16) and using 40%
MeOH/H2O to yield silydianin (3, 31 mg from 370 mg
of silymarin, 8.4%.17) The second fraction was separated
in a YMC-Pack ODS-AL column (20 mm i.d. ꢁ
150 mm; YMC) using 40% MeOH/H2O to give (þ)-
taxifolin (4, 6.7 mg from 310 mg of silymarin, 2.2%),18)
isosilychristin (5, 3.9 mg from 310 mg of silymarin,
1.3%),19) and silychristin (6, 26 mg from 310 mg of
silymarin, 8.4%)20) (Fig. 1A). The structures of these
Key words: Alzheimer’s disease; amyloid ꢀ; aggrega-
tion; (þ)-taxifolin; silymarin
Alzheimer’s disease (AD) is characterized by amyloid
fibril in senile plaques which mainly consist of 40- and
42-residue amyloid ꢀ-proteins (Aꢀ40 and Aꢀ42).1,2)
Aꢀ42 has been considered a principal cause of AD-like
pathogenesis because of its strong aggregative ability
and neurotoxicity.3) It is widely accepted that a soluble
oligomeric assembly of Aꢀ42 induces neuronal death
and cognitive dysfunction.4,5)
Such polyphenols as curcumin,6,7) resveratrol,8) and
(ꢀ)-epigallocatechin-3-gallate (EGCG)9) have been re-
ported to show preventive effects on the aggregation and
neurotoxicity of Aꢀ42. Some of these compounds are
in clinical or preclinical trials.10) Since polyphenols can
be found in daily foods or supplements,11) they are
promising as preventive medicines or therapeutic agents
for AD.
Silymarin, a seed extract of Silybum marianum
containing flavonolignane diastereomers,12) has long
been used as an anti-hepatotoxic medicine without
notable adverse effects,13) and in particular, is effica-
cious against the damage induced by alcohol and
disturbances in the function of the gastrointestinal
tract.14) Our group has recently reported that silymarin
reduced such AD-like pathologies as senile plaques,
neuroinflammation, behavioral dysfunction, and Aꢀ
oligomer formation in a well-established AD mouse
model (J20 line).15) We report in this paper the
structure–activity relationship for (þ)-taxifolin (4) iso-
lated as one of the active components of silymarin
1
compounds were confirmed by H-NMR (AVANCE III
500, ref. tetramethylsilane, Bruker, Germany),16–20) EI-
MS (JMS-600H, 70 eV, 300 mA, JEOL, Tokyo, Japan),
and specific optical rotation (P-2200, Jasco, Tokyo,
Japan).
The effects of these flavonoids on Aꢀ42 aggregation
were examined by using thioflavin-T (Th-T), a reagent
that fluoresces when bound to Aꢀ aggregates, and
transmission electron microscopy (TEM), as previously
described.21,22) As shown in Fig. 1B, only (þ)-taxifolin
(4) among the isolated flavonoids strongly reduced the
Th-T relative fluorescence induced by Aꢀ42 aggrega-
tion, meaning the potent inhibition of Aꢀ42 aggregation
by 4. The analysis of TEM showed that the fibril
formation of Aꢀ42 was inhibited by 4; shorter or slighter
fibrils (Fig. 2B). (þ)-Taxifolin (4) also disaggregated
the preformed fibrils of Aꢀ42 (Fig. 1C). The inhibitory
effect of 4 on Aꢀ42 aggregation was almost equal to that
of silymarin (Fig. 1B). A quantification analysis by
HPLC revealed that silymarin used in this work
y
To whom correspondence should be addressed. Tel: +81-75-753-6281; Fax: +81-75-753-6284; E-mail: irie@kais.kyoto-u.ac.jp
Abbreviations: Aꢀ, amyloid ꢀ; AD, Alzheimer’s disease; HPLC, high-performance liquid chromatography; Th-T, thioflavin-T; TEM, transmission
electron microscopy

Structure–Activity Relationship for (þ)-Taxifolin
1101
(Fig. 2A). In brief, a 0.20 mM ether/EtOH solution
(35 mL/15 mL) of N-methyl–N-nitroso-p-toluenesulfo-
namide was heated at 70 ꢂC. To the solution was added a
potassium hydroxide solution (one gram of potassium
hydroxide in 15 mL of water) to yield diazomethane
which was condensed in a cold tube as a yellow ether
solution. (þ)-Taxifolin (4, 65 mg, 0.21 mmol) was
dispersed in benzene (1.0 mL) and diethylether
(3.0 mL), to which an aliquot of a diazomethane solution
(12 mL) was added at 0 ꢂC, and the mixture was stood
at the same temperature for 1.5 h. The solution was
evaporated in vacuo, and part of the residue was
separated by preparative thin-layer chromatography
and followed by HPLC in a YMC-Pack ODS-A column
(20 mm i.d. ꢁ 150 mm; YMC) with a linear gradient of
50–100% CH3CN/H2O for 30 min to yield (þ)-7-O-
methyl-(7, 18 mg, 28% yield), (þ)-7,30-di-O-methyl-(8,
6.3 mg, 9.7% yield), (þ)-7,40-di-O-methyl-(9, 7.3 mg,
11% yield), and (þ)-7,30,40-tri-O-methyltaxifolin (10,
2.1 mg, 3.2% yield, Fig. 2A). Their structures were
A
B
120
80
40
0
1
confirmed by H-NMR18) and EI-MS to be identical to
0
8
16
24
32
40
48
those reported previously. The Th-T assay showed that
(þ)-7-O-methyl-taxifolin (7) prevented the aggregation
of Aꢀ42 in a manner similar to 4, whereas (þ)-7,30-di-
O-methyl-(8), (þ)-7,40-di-O-methyl-(9), and (þ)-7,30,40-
tri-O-methyltaxifolin (10) did not (Fig. 2A). The TEM
images of Aꢀ42 fibrils treated with 7, but not with 8,
were similar to those treated with 4 (Fig. 2B). These
results indicate the 30,40-dihydroxyl groups on the B-ring
of 4 to be important to prevent Aꢀ42 aggregation, while
the 7-hydroxyl group was not critical. This is consistent
with the findings that only 4 had a catechol moiety
among the flavonoids isolated from silymarin in this
study. These findings do not contradict the report by
Akaishi et al. that the 30,40-dihydroxyl group, and not the
7-hydroxyl group, was essential to the inhibitory effect
of fisetin (a quercetin analog without the 5-hydroxyl
group) on Aꢀ42 fibril formation.23)
Incubation time (h)
120
80
40
0
C
0
8
16
24
Incubation time (h)
Fig. 1. Identification of (þ)-Taxifolin (4) from Silymarin as One of
the Active Components against Aꢀ42 Aggregation.
A, Structure of the flavonoids isolated from silymarin. EI-MS and
optical rotation data are as follow: silibinin A (1), m=z 482 [M]þ,
½ꢁꢃ þ26:0 (c 0.25, MeOH, 26 ꢂC);19) silibinin B (2), m=z 482 [M]þ,
½ꢁꢃD þ12:0 (c 0.19, MeOH, 26 ꢂC);19) silydianin (3), m=z 482 [M]þ,
½ꢁꢃDD þ231 (c 0.0050, MeOH, 27 ꢂC);19) (þ)-taxifolin (4), m=z 304
[M]þ, ½ꢁꢃD þ22:2 (c 0.12, MeOH, 29 ꢂC);19) isosilychristin (5), m=z
482 [M]þ, ½ꢁꢃD þ207 (c 0.0050, MeOH, 26 ꢂC);19) silychristin (6),
m=z 482 [M]þ, ½ꢁꢃD þ112 (c 0.30, MeOH, 26 ꢂC).28) B, The effect of
each flavonoid on Aꢀ42 aggregation was estimated by the Th-T
method. Aꢀ42 (25 mM) was incubated with or without each flavonoid
(50 mM) in phosphate-buffered saline (PBS, 50 mM sodium phos-
phate, 100 mM NaCl, pH 7.4) at 37 ꢂC for 48 h. Each flavonoid was
dissolved in ethanol at 5.0 mM before use, and diluted with PBS
(50 mM final concentration). The molecular weight of silymarin was
defined as 482, which was that of the main components (silybinin A
and B, silydianin, isosilychristin, and silychristin) in silymarin.
(þ)-Taxifolin (4) was not methylated at position 5 by
diazomethane, implying that the hydroxyl group at
position 5 could not be involved in the intermolecular
interaction. Indeed, the hydroxyl group at position 5 of 4
could have participated in the intramolecular hydrogen
bond with the carbonyl oxygen on the C-ring, this being
1
deduced from the H-NMR chemical shift (11.7 ppm in
(CD3)2CO). The practical implication of this result is
that the hydroxyl group at position 5 did not contribute
to the inhibition of Aꢀ42 aggregation by 4. Although
methylated 4 at position 3 was not also obtained
(Fig. 2), the report23) that luteolin without a hydroxyl
group at position 3 inhibited Aꢀ42 aggregation suggests
that the hydroxyl group at position 3 of 4 would not
participate in the inhibitory activity.
Aꢀ42 without flavonoids;
Aꢀ42 with silymarin; Aꢀ42 with 1;
Aꢀ42 with 2; Aꢀ42 with 3; Aꢀ42 with 4; Aꢀ42 with 5; ꢁ
Aꢀ42 with 6. Data are presented as the mean ꢄ SEM (n ¼ 8). C,
The disaggregation of Aꢀ42 fibrils by (þ)-taxifolin (4) was
estimated by the Th-T method. Aꢀ42 (25 mM) was incubated at
37 ꢂC in PBS (pH 7.4) for 48 h for preparing the Aꢀ42 fibrils, to
which were then added 4 (50 mM) before incubating at 37 ꢂC for 24 h.
Furthermore, to examine the effect of the stereo-
chemistry of the hydroxyl group at position 3 on the
C-ring of (þ)-taxifolin (4), the 2,3-(R,R)-trans form, on
the inhibition of Aꢀ42 aggregation, the (ꢀ)-taxifolin,
2,3-(S,S)-trans form was synthesized basically accord-
ing to the method of Roschek et al.,24) except for using
3,4-trihydroxybenzaldehyde as a substrate. Briefly,
vanillin (0.10 g, 0.63 mmol) dissolved in CH2Cl2 was
demethylated by being treated with 1 M boron tribromide
in dichloromethane (2.6 mL, 2.6 mmol) at 4 ꢂC for 1 h to
quantitatively give 3,4-dihydroxybenzaldehyde. The
Aꢀ42 without flavonoid;
the mean ꢄ SEM (n ¼ 8).
Aꢀ42 with 4. Data are presented as
contained 1.2% (þ)-taxifolin (4) which was slightly less
than the isolated yield (2.2%, purity 99.6%). The low
content rate of 4 does not exclude the presence of other
active components in silymarin.
We identified the hydroxyl groups of (þ)-taxifolin (4)
involved in the inhibitory effect by preparing four O-
methyl derivatives of 4 by a diazomethane treatment

1102
M. SATO et al.
120
80
40
0
A
O-Methyl-(+)-taxifolin
120
80
40
0
0
8
16
24
32
40
48
Incubation time (h)
Fig. 3. Effects of the Stereochemistry at Position 3 of (þ)-Taxifolin
(4) on Aꢀ42 Aggregation.
The effects of (ꢀ)-taxifolin on Aꢀ42 aggregation were evaluated
0
8
16
24
by the Th-T test. Aꢀ42 (25 mM) was incubated with or without each
Incubation time (h)
enantiomer (50 mM) in PBS (pH 7.4) at 37 ꢂC for 48 h.
Aꢀ42
Aꢀ42
without flavonoid;
with (ꢀ)-taxifolin. Data are presented as the mean ꢄ SEM (n ¼ 8).
Aꢀ42 with (þ)-taxifolin (4); and
B
sometimes influenced by several factors; for example,
the outside temperature and batch of Aꢀ42. However,
similar results were obtained by another independent
experiment.
Fig. 2. Structure–Activity Relationships of (þ)-Taxifolin (4).
A, Methylated (þ)-taxifolins and their inhibitory effects on Aꢀ42
aggregation determined by the Th-T assay. Aꢀ42 (25 mM) was
incubated with or without each methylated (þ)-taxifolin (50 mM) in
PBS (pH 7.4) at 37 ꢂC for 48 h. Aꢀ42 without flavonoid; Aꢀ42
In conclusion, we found (þ)-taxifolin (4) with a
catechol moiety on the B-ring from silymarin (a mixture
of flavonoid-related compounds) to be one of the active
components for anti-Aꢀ42 aggregation. To the best of
our knowledge, this is the first report that 4, belonging
to flavanonols containing a single bond between C2 and
C3 on the C-ring, had a preventive effect on Aꢀ42
aggregation. On the other hand, flavonoids (myricetin,
quercetin, etc.), most of which have previously been
reported to inhibit Aꢀ42 aggregation,27) belong to
flavonols containing a double bond between C2 and
C3 on the C-ring. The structure-activity relationship of 4
clarified the requisite moiety (a catechol structure on the
B-ring) for inhibitory activity against Aꢀ42 aggregation.
A further study to clarify its inhibitory mechanism is in
progress in our laboratory.
with 4;
Aꢀ42 with 7;
Aꢀ42 with 8;
Aꢀ42 with 9; and
Aꢀ42 with 10. EI-MS data for the methylated (þ)-taxifolins are as
follow: 7, m=z 318 [M]þ; 8, m=z 332 [M]þ; 9, m=z 332 [M]þ; and
10, m=z 346 [M]þ. Data are presented as the mean ꢄ SEM (n ¼ 8).
B, The TEM analysis of Aꢀ42 fibrils treated with the methylated
(þ)-taxifolins was performed under an H-7650 electron microscope
(Hitachi, Ibaraki, Japan). Scale bar, 200 nm. Left, Aꢀ42 without a
flavonoid; left middle, Aꢀ42 with 4; right middle, Aꢀ42 with 7;
right, Aꢀ42 with 8.
phenolic hydroxyl groups were protected with metho-
xymethyl groups (73% yield). On the other hand, the
phenolic hydroxyl groups of 2,4,6-trihydroxyacetophe-
none were also protected with methoxymethyl groups
(24% yield). A cross-aldol reaction of these two
products in KOH/MeOH (83% yield) and subsequent
treatment with H2O2 under alkaline conditions quanti-
tatively yielded the epoxide which was cyclized and
deprotected under HCl/MeOH to give (ꢄ)-taxifolin
(59% yield). The enantiomers were separated by HPLC
in a CHERALCEL OJ-RH column (10 mm i.d. ꢁ
150 mm; Daicel Corporation, Osaka, Japan) by using
15% CH3CN/H2O containing 0.10% acetic acid.25) The
ratio of the enantiomers was almost 1:1, based on the
isolated yield of each enantiomer: (þ)-taxifolin (4), ½ꢁꢃD
þ17:3 (c 0.10, MeOH, 16 ꢂC, lit.19) ½ꢁꢃD þ19:0, c 0.1,
MeOH); (ꢀ)-taxifolin, ½ꢁꢃD ꢀ16:2 (c 0.10, MeOH,
16 ꢂC). As shown in Fig. 3, the inhibitory ability against
Aꢀ42 aggregation was almost the same between these
two enantiomers. In addition, no difference in the
inhibitory activity against Aꢀ40 or Aꢀ42 aggregation
between (þ)-catechin (2,3-trans form) and (ꢀ)-epicate-
chin (2,3-cis form) has been reported.26) Furthermore,
quercetin with a C2–C3 double bond on the C-ring has
been reported to inhibit Aꢀ42 aggregation.27) These
findings suggest the stereochemistry at positions 2 and 3
not to play an important role in the inhibitory effects of 4
against Aꢀ42 aggregation. Although the inhibition of
Aꢀ42 aggregation by 4 in Fig. 3 seemed slightly
stronger than that in Figs. 1 and 2, the fluorescence
intensity in the Th-T assay between different figures is
Acknowledgments
This study was partly supported by Grants in Aid for
Scientific Research (A) (Grant no. 21248015 to K. I.),
and (C) (no. 22603006 to K. M.), and by a Fund for the
Promotion of Science for Young Scientists (Grant
no. 22.4068 to M. S.) from The Ministry of Education,
Culture, Sports, Science and Technology of Japan, and
by funds from the Asahi Group Foundation (to K. I.) and
from the Kato Memorial Bioscience Foundation (to K.
M.). We thank Prof. Nobutaka Fujii and Dr. Shinya
Oishi from the Graduate School of Pharmaceutical
Sciences at Kyoto University for use of the MALDI-
TOF-MS. M.S. is a Research Fellow of the Japan
Society for the Promotion of Science.
References
1) Glenner GG and Wong CW, Biochem. Biophys. Res. Commun.,
120, 885–890 (1984).
2) Masters CL, Simms G, Weinman NA, Multhaup G, McDonald
BL, and Beyreuther K, Proc. Natl. Acad. Sci. USA, 82, 4245–
4249 (1985).
3) Haass C and Selkoe DJ, Nat. Rev. Mol. Cell Biol., 8, 101–112
(2007).
4) Walsh DM, Klyubin I, Fadeeva JV, Cullen WK, Anwyl R,

Structure–Activity Relationship for (þ)-Taxifolin
1103
Wolfe MS, Rowan MJ, and Selkoe DJ, Nature, 416, 535–539
(2002).
Planta Med., 73, 1214–1216 (2007).
18) Kiehlmann E and Slade PW, J. Nat. Prod., 66, 1562–1566
(2003).
5) Roychaudhuri R, Yang M, Hoshi MM, and Teplow DB, J. Biol.
Chem., 284, 4749–4753 (2009).
19) Kim N-C, Graf TN, Sparacino CM, Wani MC, and Wall ME,
Org. Biomol. Chem., 1, 1684–1689 (2003).
6) Ono K, Hasegawa K, Naiki H, and Yamada M, J. Neurosci.
Res., 75, 742–750 (2004).
20) Tanaka H, Hiroo M, Ichino K, and Ito K, Chem. Pharm. Bull.,
37, 1441–1445 (1989).
7) Yang F, Lim GP, Begum AN, Ubeda OJ, Simmons MR,
Ambegaokar SS, Chen PP, Kayed R, Glabe CG, Frautschy SA,
and Cole GM, J. Biol. Chem., 280, 5892–5901 (2005).
8) Li F, Gong Q, Dong H, and Shi J, Curr. Pharm. Des., 18, 27–33
(2012).
21) Murakami K, Irie K, Morimoto A, Ohigashi H, Shindo M,
Nagao M, Shimizu T, and Shirasawa T, J. Biol. Chem., 278,
46179–46187 (2003).
22) Murakami K, Irie K, Morimoto A, Ohigashi H, Shindo M,
Nagao M, Shimizu T, and Shirasawa T, Biochem. Biophys. Res.
Commun., 294, 5–10 (2002).
9) Rezai-Zadeh K, Arendash GW, Hou H, Fernandez F, Jensen M,
Runfeldt M, Shytle RD, and Tan J, Brain Res., 1214, 177–187
(2008).
23) Akaishi T, Morimoto T, Shibao M, Watanabe S, Sakai-Kato K,
Utsunomiya-Tate N, and Abe K, Neurosci. Lett., 444, 280–285
(2008).
10) Gravitz L, Nature, 475, S9–S11 (2011).
´
´
11) Manach C, Scalbert A, Morand C, Remesy C, and Jimenez L,
´
Am. J. Clin. Nutr., 79, 727–747 (2004).
12) Lee JI, Narayan M, and Barrett JS, J. Chromatogr. B, 845, 95–
103 (2007).
24) Roschek Jr B, Fink RC, McMichael MD, Li D, and Alberte RS,
Phytochemistry, 70, 1255–1261 (2009).
25) Vega-Villa KR, Remsberg CM, Ohgami Y, Yanez JA,
Takemoto JK, Andrews PK, and Davies NM, Biomed. Chro-
matogr., 23, 638–646 (2009).
13) Morazzoni P, Montalbetti A, Malandrino S, and Pifferi G, Eur.
J. Drug Metab. Pharmacokinet., 18, 289–297 (1993).
´
14) Valenzuela A, Barrıa T, Guerra R, and Garrido A, Biochem.
26) Ono K, Yoshiike Y, Takashima A, Hasegawa K, Naiki H, and
Yamada M, J. Neurochem., 87, 172–181 (2003).
27) Ono K, Hamaguchi T, Naiki H, and Yamada M, Biochim.
Biophys. Acta, 1762, 575–586 (2006).
Biophys. Res. Commun., 126, 712–718 (1985).
15) Murata N, Murakami K, Ozawa Y, Kinoshita N, Irie K,
Shirasawa T, and Shimizu T, Biosci. Biotechnol. Biochem., 74,
2299–2306 (2010).
28) Smith WA, Lauren DR, Burgess EJ, Perry NB, and Martin RJ,
Planta Med., 71, 877–880 (2005).
16) Lee DYW and Liu Y, J. Nat. Prod., 66, 1171–1174 (2003).
17) MacKinnon SL, Hodder M, Craft C, and Simmons-Boyce J,