Synthesis of sominone and its derivatives based on an RCM strategy: Discovery of a novel anti-alzheimer's disease medicine candidate "denosomin"

Article abstract of DOI:10.1021/ol901553w

(Figure Presrnted) Synthesis of sominone was achieved starting from dehydroepiandrosterone on the basis of an RCM strategy for the construction of a δ-lactone side chain. This synthetic protocol was applied for the synthesis of several analogous derivativ

Full text of DOI:10.1021/ol901553w

ORGANIC
LETTERS
2009
Vol. 11, No. 17
3970-3973
Synthesis of Sominone and Its
Derivatives Based on an RCM Strategy:
Discovery of A Novel Anti-Alzheimer’s
Disease Medicine Candidate
“Denosomin”
Yuji Matsuya,*,† Yu-ichiro Yamakawa,† Chihiro Tohda,‡ Kiyoshi Teshigawara,‡
Masashi Yamada,§ and Hideo Nemoto†
Graduate School of Medicine and Pharmaceutical Sciences and Institute of Natural
Medicine, UniVersity of Toyama, 2630 Sugitani, Toyama 930-0194, Japan, and
Research Alliance & Portfolio Strategy, Taiho Pharmaceutical Co. Ltd.,
1-2-4 Utikanda, Chiyoda-ku, Tokyo 101-0047, Japan
matsuya@pha.u-toyama.ac.jp
Received July 8, 2009
ABSTRACT
Synthesis of sominone was achieved starting from dehydroepiandrosterone on the basis of an RCM strategy for the construction of a δ-lactone
side chain. This synthetic protocol was applied for the synthesis of several analogous derivatives including 1-deoxy-24-norsominone (denosomin),
which was revealed to exhibit notable bioactivities for new antidementia chemotherapy, exceeding the original natural compound sominone.
In the present aging society, central nervous disorders such
as dementia, Parkinson’s disease and Alzheimer’s disease
represent serious medical problems, and the development of
novel pharmaceuticals for such disorders has been hastened
in order to improve therapeutic strategies. Enhancing brain
function requires the reinforcement of neuronal networks,
including neurite regeneration and synapse formation; there-
fore, we have been exploring antidementia drugs based on
reconstructing neuronal networks in the damaged brain. We
previously investigated the effects of extracts and constituents
of Ashwagandha (root of Withania somnifera Dunal), an
Ayurvedic tonic medicine, on neurite outgrowth in cultured
neurons.1 In the course of these studies, we isolated witha-
noside IV, a member of natural withanolides (ergostane-type
steroids with a δ-lactone side chain), from methanol extracts
of Ashwagandha, and found that this natural compound
showed antidementia activity after oral administration to an
Alzheimer’s disease mouse model.2 Furthermore, we noted
that withanoside IV had significant function in enhancing
(1) (a) Tohda, C.; Kuboyama, T.; Komatsu, K. Neuroreport 2000, 11,
1981. (b) Zhao, J.; Nakamura, N.; Hattori, M.; Kuboyama, T.; Tohda, C.;
Komatsu, K. Chem. Pharm. Bull. 2002, 50, 760. (c) Kuboyama, T.; Tohda,
C.; Zhao, J.; Nakamura, N.; Hattori, M.; Komatsu, K. Neuroreport 2002,
13, 1715. (d) Kuboyama, T.; Tohda, C.; Komatsu, K. Br. J. Pharmacol.
2005, 144, 961.
† Graduate School of Medicine and Pharmaceutical Sciences, University
of Toyama.
‡ Institute of Natural Medicine, University of Toyama.
§ Taiho Pharmaceutical Co. Ltd.
(2) Kuboyama, T.; Tohda, C.; Komatsu, K. Eur. J. Neurosci. 2006, 23,
1417.
10.1021/ol901553w CCC: $40.75
Published on Web 08/04/2009
 2009 American Chemical Society

the growth activity of neurites from injured and/or spared
neurons and synaptogenesis in the brain.2
Withanoside IV is a steroidal saponin conjugated with two
glucose residues at position C3 on the A-ring, and is
speculated to be metabolized into a sapogenin, sominone (1,
Scheme 1),3 by enterobacterial ꢀ-glucosidases. As expected,
construct the δ-lactone moiety. On the other hand, recent
advances in metathesis chemistry, including RCM, have
allowed for the convenient construction of various ring
systems.6 In particular, continuous efforts exploring the
efficient Ru-based catalysts have brought about significant
improvements in the syntheses of sterically hindered cyclic
compounds containing a tri- or tetra-substituted olefin
structures.7 We believed that a suitable choice of a catalyst
could realize a concise δ-lactone formation of sominone and
its derivatives via the RCM reaction. We prepared the RCM
substrates from steroidal ketone compounds 9 and 10.
The synthesis of sominone (1) and analogous compounds
2-8 commences with a standard transformation sequence
starting from commercially available dehydroepiandrosterone
(10) and its 1R-hydroxy derivative 9 (Scheme 2).8 Three-
Scheme 1
Scheme 2
sominone was identified as the main metabolite in serum
after oral administration of withanoside IV,2 indicating the
potential of withanoside IV as a prodrug for sominone
activity on target organs. We found that sominone, prepared
by enzymatic deglycosylation of withanoside IV, induced
significant axonal and dendritic regeneration and synaptic
reconstruction in cultured rat cortical neurons damaged by
Amyloid ꢀ(25-35).2 In addition, in vivo experiments
indicated that orally administered sominone enhanced memory
and axonal density in the brains of normal mice.4 In light of
its high potential as a reconstructing agent of neuronal
networks, sominone can be considered as an extraordinarily
promising candidate for antidementia therapeutic agents.
To the best of our knowledge, no report on synthetic
studies has focused on sominone itself as a synthetic target,
despite such biological interest. Thus, we planned to develop
an efficient synthetic method of sominone and its analogues,
aiming at extensive SAR studies and in depth in vivo
evaluations. Herein, we describe the synthesis of sominone
utilizing a ring-closing metathesis (RCM) strategy and the
discovery of a closely related analogue, 1-deoxy-24-nor-
sominone (named “denosomin”), possessing much higher
efficacy than sominone as an antidementia agent.
step transformation of 9 and 10 into known alcohols 11 and
12 were achieved via Wittig olefination, followed by the ene
reaction using modification of a reported method.9 Stereo-
selective reduction and PDC oxidation efficiently proceeded
to give the aldehydes 13 and 14 in high yield, respectively.10
Methallylation or allylation of aldehydes 13 and 14 was
investigated to obtain a footing for the RCM substrates.
Thus, methallylation was performed utilizing a Barbier-
type addition reaction to afford alcohols 15a,b and 16a,b
in good yields. Although scarcely any diastereoselectivity
at the 22-position was observed, chromatographic separa-
tion was easily attained to give both diastereomers (a and
b) in pure form. On the other hand, asymmetric allylation
using (+)- or (-)-B-allyldiisopinocampheylborane11 pro-
duced the corresponding allyl adducts 15c and 16c or 15d
and 16d, respectively, with high stereoselectivity. In every
case, only trace amounts of the opposite diastereomers
were detected on TLC. Installation of the other olefin unit
required for RCM, containing a hydroxymethyl substitu-
(6) For recent reviews, see: (a) Gradillas, A.; Perez-Castells, J. Angew.
Chem., Int. Ed. 2006, 45, 6086. (b) Grubbs, R. H. Tetrahedron 2004, 60,
7117. (c) Nicolaou, K. C.; Bulger, P. G.; Sarlah, D. Angew. Chem., Int.
Ed. 2005, 44, 4490. (d) Deiters, A.; Martin, S. F. Chem. ReV. 2004, 104,
2199.
Our synthetic plan for sominone (1) and simplified
derivatives 2-8 is depicted in Scheme 1. The key step of
the synthesis is the construction of a δ-lactone moiety using
the RCM reaction. Although synthetic studies of withanolides
have been reported by several groups,5 many of them
required complex and multistep transformations in order to
(7) For reviews, see: Schrodi, Y.; Pederson, R. L. Aldrichim. Acta 2007,
40, 45.
(8) Dodson, R. M.; Goldkamp, A. H.; Muir, R. D. J. Am. Chem. Soc.
1960, 82, 4026.
(9) (a) Konno, K.; Ojima, K.; Hayashi, T.; Takayama, H. Chem. Pharm.
Bull. 1992, 40, 1120. (b) Deng, L.; Wu, H.; Yu, B.; Jiang, M.; Wu, J. Bioorg.
Med. Chem. Lett. 2004, 14, 2781.
(3) (a) Sahai, M. J. Nat. Prod. 1985, 48, 474. (b) Atta-ur-Rahman; Jamal,
S. A.; Choudhary, M. I. Heterocycles 1992, 34, 689.
(4) Tohda, C.; Joyashiki, E. Br. J. Pharmacol. 2009, web-released.
(5) For example, see: (a) Hirayama, M.; Gamoh, K.; Ikekawa, N. J. Am.
Chem. Soc. 1982, 104, 3735. (b) Perez-Medrano, A.; Grieco, P. J. Am. Chem.
Soc. 1991, 113, 1057. (c) Tsubuki, M.; Kanai, K.; Keino, K.; Kakinuma,
N.; Honda, T. J. Org. Chem. 1992, 57, 2930.
(10) (a) Konno, K.; Ojima, K.; Hayashi, T.; Tanabe, M.; Takayama, H.
Chem. Pharm. Bull. 1997, 45, 185. (b) Corey, E. J.; Grogan, M. J.
Tetrahedron Lett. 1998, 39, 9351.
(11) For a review of allylborane reagents, see: Ramachandran, P. V.
Aldrichim. Acta 2002, 35, 23.
Org. Lett., Vol. 11, No. 17, 2009
3971

ent, was carried out through a two-step sequence; esteri-
fication with PMP-protected acid chloride, followed by
removal of the PMP group by CAN oxidation (ac-
companied by cleavage of the 3-TBS ether on the
A-ring).12 Thus, preparation of the requisite RCM sub-
strates 19a-d and 20a-d from compounds 13 and 14 was
accomplished. These transformations are summarized in
Scheme 3 and Table 1.
Scheme 4
Scheme 3
Less sterically hindered substrates 19c,d and 20c,d (R2 )
H) were first subjected to the RCM reaction using catalyst
D in toluene at 80 °C. Excellent results were obtained for
19c and 20c, possessing a natural-type R configuration at
the 22-position, while non-natural diastereomers 19d and 20d
gave moderate yields and required a somewhat prolonged
reaction time. Although the rationale for the differences in
RCM reactivity between these diastereomers due to the
configurations at the 22-position remained unclear, construc-
tion of the δ-lactone with the trisubstituted olefin structure
was satisfactorily accomplished by means of the RCM
strategy (Scheme 5, Table 2). For the synthesis of sominone,
Table 1. Isolation Yields of the Compounds 15-20
yield (%)
compd
15
16
17
18
19
20
a (R2 ) Me, 22ꢀ-O)
b (R2 ) Me, 22R-O)
c (R2 ) H, 22ꢀ-O)
d (R2 ) H, 22R-O)
42a
32a
61
44b
43b
78
78
76
70
84
84
86
75
88
64
54
57
47
65
60
67
74
Scheme 5
65
73
a These diastereomers could be separated by column chromatography.
b These diastereomers could be separated by column chromatography.
For the purpose of seeking an efficient RCM catalyst
of the present synthesis, an initial survey was performed
using several Ru-based catalysts (A-D)6,7 and O-benzy-
lated substrate 21 (Scheme 4).13 RCM of homoallyl acrylates
to form δ-lactone compounds has been recently reported,
albeit in a much more simple compound system, in which
the second-generation Grubbs’ catalyst (A) most efficiently
gave rise to ring-closure, particularly in the cases of tri- or
tetrasubstituted olefin formations.14 However, cyclization of
substrate 21 did not occur in the presence of catalyst A at
80 °C. In light of a recent report regarding the effects of
N-aryl substituents of NHC ligands on the catalyst activity,15
we next attempted the o-tolyl variant B for the RCM of 21,
leading to the formation of desired δ-lactone 22 in low yield
(15%). However, employment of the second-generation
Hoveyda-Grubbs’ catalyst (C) significantly improved the
cyclization yield (46%), and the corresponding o-tolyl variant
D gave the best result (56% yield). Therefore, catalyst D
was selected for further investigation.
Table 2. RCM of the Substrates 19a-d and 20a-d
substrate
time (h) product yielda (%)
19a
19b
19c
19d
20a
20b
20c
20d
R2 ) Me, 22ꢀ-O
R2 ) Me, 22R-O
R2 ) H, 22ꢀ-O
R2 ) H, 22R-O
R2 ) Me, 22ꢀ-O
R2 ) Me, 22R-O
R2 ) H, 22ꢀ-O
R2 ) H, 22R-O
2
2
2
10
3
3
23a
23b
23c
23d
5
6
7
8
22(92)
24(95)
85(85)
58(70)
17(64)
15(75)
88(98)
52(61)
2
5
a Isolation yields. Yields in parentheses are brsm.
we attempted challenging RCM with the substrate 19a
involving tetrasubstituted olefin formation (R2 ) Me) under
3972
Org. Lett., Vol. 11, No. 17, 2009

the same reaction conditions. Although the reaction was
revealed to be rather sluggish as expected, the required
cyclization afforded δ-lactone 23a in a 22% isolation yield after
2 h without any side reactions (92% yield based on the
recovered starting material). Because the longer reaction time
caused a considerable decrease in substrate recovery with an
almost equal isolation yield of product, a recycling process with
a shorter reaction time (2-3 h) was considered to be better
suited for this RCM transformation. Similarly, substrates 19b
and 20a,b could be transformed into the cyclization products
23b, 5, and 6, respectively, using catalyst D (Scheme 5, Table
2). For compounds 23a-d, the TBS group at the 1-position
was removed upon treatment with HF-pyridine in high yield.
Thus, the synthesis of the biologically important withanolide,
sominone (1) and related compounds 2-8, was accomplished
through a concise and efficient synthetic route. The NMR data
for the synthetic sample of sominone were in good agreement
with those reported in the literature.3
(0.1% DMSO) than in control cells (without Aꢀ treatment).
On the other hand, axon length was significantly longer in
the Aꢀ(1-42)-treated cells treated in combination with
compounds 1, 4-7 (1 µM), or nerve growth factor (NGF)
(100 ng/mL) than cells treated with vehicle alone. In
particular, compound 7 (denosomin) exhibited the most
potent activity for axonal regeneration among the new
derivatives, and the potency reached a level comparable to
that of sominone (1).
Encouraged by these results, we further investigated the
protective effects of sominone (1) and denosomin (7) on
Aꢀ(25-35)-induced cell death. Rat cortical neurons were
treated with various concentrations of each compound or
vehicle(0.1%DMSO),simultaneouslywith20µMAꢀ(25-35),
and cell viability was determined after 2 days. The rate of
cell survival was significantly decreased by Aꢀ(25-35)
treatment compared with controls. As shown in Figure 1
(right), sominone (1) increased cell viability up to control
levels at a dose of 1 µM, but lower doses resulted in a
considerable decrease in the effects. On the other hand, the
protective effects of denosomin (7) on Aꢀ(25-35)-induced
cell death were revealed to be much more striking, with peak
activity seen at 0.01 µM. The effects of denosomin (7) were
more potent than those of NGF, thus suggesting that
denosomin is an attractive target molecule in the development
of novel anti-Alzheimer’s disease medicines.
We then examined the effects of the sominone-related
compounds on axonal extensions in amyloid ꢀ (Aꢀ)-damaged
neurons (Alzheimer’s disease model) as an index of neuronal
network reconstructing activity. Test compounds 1-8 were
administered to rat cortical neurons 3 days after treatment
with Aꢀ(1-42), and axon length (Figure 1, left) was mea-
In this paper, we describe the synthesis of sominone and
related compounds utilizing the RCM reaction to construct
the δ-lactone side chain. During the study, we found a novel
artificial compound, denosomin, to be a highly promising
candidate for new antidementia chemotherapy, with activity
exceeding that of the original natural compound, sominone.
It is important to note that the efficiency of denosomin
synthesis is superior to that of sominone, as (1) availability
of the starting material (dehydroepiandrosterone) is good,
(2) stereochemistry of the 22-position can be completely
controlled using the Ipc2B(allyl) reagent, and (3) the key
RCM reaction proceeds more smoothly due to the lack of
the 24-methyl group. Thus, the synthesis of denosomin was
achieved in 9 steps with 20.2% overall yield starting from
commercially available dehydroepiandrosterone (10). Taking
advantage of the synthetic efficiency, large-scale preparation
and in-depth in vivo studies of denosomin, as well as
elucidation experiments of the mechanism of action, have
been performed in our laboratory, and their results will be
reported in due course.
Figure 1. Effects of compounds 1-8 (1 µM) on axonal extensions
following Aꢀ(1-42)-induced atrophy (left) and protective effects of
compounds 1 and 7 (0.01-10 µM) on Aꢀ(25-35)-induced cell death
(right) in rat cortical neurons. Positive control: NGF (100 ng/mL). The
results are presented as the mean ( SD *P < 0.05 vs Veh/Aꢀ.
sured after an additional 5 days. Axon length was shorter in
the cells treated with 5 µM Aꢀ(1-42) followed by vehicle
(12) Cyclization of the compound 17a (precursor of sominone) did not
proceed under any RCM conditions .
(13) Although this substrate was first prepared prior to 19 and 20,
deprotection of the cyclized product 22 by hydrogenolysis was found to be
troublesome.
Supporting Information Available: Experimental pro-
cedures and compound characterization data for all new
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
(14) D’Annibale, A.; Ciaralli, L.; Bassetti, M.; Pasquini, C. J. Org. Chem.
2007, 72, 6067.
(15) (a) Berlin, J. M.; Campbell, K.; Ritter, T.; Funk, T. W.; Chlenov,
A.; Grubbs, R. H. Org. Lett. 2007, 9, 1339. (b) Stewart, I. C.; Ung, T.;
Pletnev, A. A.; Berlin, J. M.; Grubbs, R. H.; Schrodi, Y. Org. Lett. 2007,
9, 1589.
OL901553W
Org. Lett., Vol. 11, No. 17, 2009
3973