Synthesis of withanolide A, biological evaluation of its neuritogenic properties, and studies on secretase inhibition

Article abstract of DOI:10.1002/anie.201101869

Neurons on steroids: A stereoselective synthesis of the neuritogenic steroid lactone withanolide A was achieved by singlet oxygen ene reaction, Wharton transposition, a Corey-Seebach homologation, and a vinylogous aldol reaction. Biological evaluation demonstrated neurite outgrowth, which supports the potential neuritogenic role of this compound in traditional Indian medicine.

Full text of DOI:10.1002/anie.201101869

DOI: 10.1002/anie.201101869
Neurosteroids
Synthesis of Withanolide A, Biological Evaluation of Its Neuritogenic
Properties, and Studies on Secretase Inhibition**
Chandan Kumar Jana, Johannes Hoecker, Tom M. Woods, Henning J. Jessen,
Markus Neuburger, and Karl Gademann*
The enhancement of memory and learning by small molecules
is of significant current interest to society.[1] This approach
could alleviate the course of neurodegenerative diseases such
as Alzheimerꢀs or Parkinsonꢀs.[2] In addition, the use of
cognitive enhancers by healthy subjects for increased intel-
lectual performance is under intense current debate.[1,3] For
both aspects, traditional medicinal plants constitute a val-
uable resource, in particular regarding the identification and
synthesis of biologically active lead structures.[4] Withania
somnifera (“Ashwagandha” in Ayurveda, Indian ginseng) is
considered an important plant in traditional Indian medicine
and is being prescribed for a variety of ailments, including as a
general tonic for the elderly related to antiaging and cognitive
improvement.[5] Molecular pharmacological studies have
associated some of these activities with certain metabolites
in the plant. In particular, withanolide A (1) has been shown
to possess strong neuropharmacological activities in promot-
ing neurite outgrowth, reversing neuritic atrophy, and aiding
synapse reconstruction.[6] These remarkable properties are
further extended by a recent study documenting that with-
anolide A (1) modulates several secretase targets with regard
to neurodegenerative diseases.[7] Chan and co-workers
recently demonstrated that this steroid lactone 1 downregu-
lates BACE1 and upregulates ADAM10 in primary rat
cortical neurons.[7] These combined reports provide a strong
pharmacological rationale for further investigation of with-
anolide A (1), for which no preparation from simple steroid
precursors has yet been reported.[8]
genic properties of 1 and of its derivatives and have also
carried out mechanism-of-action studies using a series of
enzymatic assays.
The steroid lactone withanolide A (1) displays function-
alities in both the A and the B ring, as well as the
functionalized side-chain lactone, which raise strategic ram-
ifications concerning the sequential order of their installation.
In particular, the oxidation pattern at C5–C7 as well as the
tertiary alcohol at C20 combined with the adjacent dihydro-
pyrone were identified in the beginning as key challenges for
the synthesis. During semi synthesis studies to derivatize
withanolide A, it became clear that both the side chain as well
as the hydroxy epoxide in the B ring are relatively stable
under a variety of conditions. Surprisingly, after extensive
experimentation both on model systems as well as on the
natural product, we realized that the chemically most
sensitive functionality was the a,b-unsaturated ketone in the
A ring. By building on this chemical knowledge and hoping to
exploit the inherent reactivity pattern of withanolide A (1),
we planned to minimize the use of protecting groups towards
the end of the synthesis. Therefore, we decided to stereose-
lectively install the dihydropyrone carbinol group first, with
subsequent oxidation of the B ring and finally installation of
the enone in the A ring.
The synthesis of withanolide A (1) started with the
protection of the hydroxy group of pregnenolone as a TBS-
ether; the keto group was then homologated using the Corey–
Seebach umpolung methodology to provide known[9] dithiane
2 in 82% yield over two steps (Scheme 1). The oxidative
cleavage of the 1,3-dithiane by N-chlorosuccinimide pro-
ceeded smoothly to provide the corresponding C22 hydroxy
aldehyde, of which the tertiary C20-OH group was protected
as MOM ether (94%). Following a procedure developed by
Ikekawa,[8d] the resulting aldehyde 3 reacted with the vinyl-
ogous enolate (from ethyl 2,3-dimethylbut-2-enoate[10] and
LiHMDS) in a stereoselective vinylogous aldol reaction to
provide the lactone 4 in very good yield and stereoselectivity
(87%, d.r. = 93:7). The configuration of both C20 with the
tertiary hydroxy group and the newly formed stereogenic
Herein, we report the first successful synthesis of with-
anolide A (1) from pregnenolone featuring a singlet O2 ene
reaction, a vinylogous aldol reaction, and a Wharton carbonyl
transposition. In addition, we have re-evaluated the neurito-
[*] Dr. C. K. Jana, J. Hoecker, Dr. H. J. Jessen, Prof. Dr. K. Gademann
Department of Chemistry, University of Basel
St. Johanns-Ring 19, 4056 Basel (Switzerland)
E-mail: karl.gademann@unibas.ch
Homepage: http://www.chemie.unibas.ch/~gademann
Dr. T. M. Woods
Department of Chemistry, The University of Auckland
23 Symonds Street, Auckland (New Zealand)
M. Neuburger
Laboratory for chemical crystallography, University of Basel
St. Johanns-Ring 19, 4056 Basel (Switzerland)
[**] K.G. is a European Young Investigator (EURYI). We thank Dr.
Solange Meyer, Actelion Pharmaceuticals Ltd, for carrying out the
enzymatic assays. We gratefully acknowledge financial support by
the SNF (PE002-117136/1) and the DFG (Project JE 572/1-1).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201101869.
Angew. Chem. Int. Ed. 2011, 50, 8407 –8411
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8407

Communications
Scheme 1. Total synthesis of withanolide A (1): a) TBSCl, Imd, THF, RT, 98%; b) dithiane, BuLi, THF, À788C to RT, 84%; c) NCS, CH2Cl2, RT,
73%; d) MOMCl, NaI, DIPEA, DME, reflux, 94%; e) ethyl-2,3-dimethylbut-2-enoate, LiHMDS, THF/DMPU, À788C to RT, 87% (93:7); f) HCl,
THF/H2O, RT, 87%; g) O2, TPP, Na light, pyridine, PPh3, RT, 61%; h) 3-chloroperbenzoic acid, CH2Cl2, 08C to RT, 96% (96:4); i) HCl, THF/H2O,
RT, 80%; j) TPAP, NMO, CH2Cl2, RT, 95%; k) IBX, MPO, DMSO, 408C, 81%; l) H2O2, Triton B, THF, 08C, 60%; m) N2H5Cl, Et3N, 08C to RT, 62%;
n) PDC, CH2Cl2, RT, 80%. TBSCl=tert-butyldimethylsilyl chloride, NCS=N-chlorosuccinimide, MOMCl=(chloromethyl)methyl ether, DIPEA =
diisopropylethylamine, LiHMDS=lithium hexamethyldisilazide, DMPU=N,N’-dimethylpropyleneurea, TPP=meso-tetraphenylporphyrin, TPAP =
tetrapropylammonium perruthenate, NMO=N-methylmorpholine N-oxide, IBX=2-iodoxybenzoic acid, MPO=4-methoxypyridine-N-oxide, Tri-
ton B=benzyltrimethylammonium hydroxide, PDC=pyridinium dichromate.
center were confirmed by X-ray crystal structure analysis of
deprotected lactone 5 (Figure 1).
We then planned to install the correct oxidation pattern in
the B ring of withanolide A (1). The challenge consisted of
the regioselective installation of the B-ring epoxyalcohol
from a simple olefin in 4 in the presence of the unsaturated
lactone. After a thorough experimental investigation of
various possibilities, we identified the singlet-oxygen-medi-
ated photooxygenative olefin migration as a straightforward
method[11] for the synthesis of the allylic tertiary alcohol 6.
Accordingly, the olefin 4 was allowed to react with singlet
oxygen, generated in situ from O2, in the presence of meso-
tetraphenylporphyrin as sensitizer and under irradiation of a
Na lamp, affording the allylic alcohol 6 in good yield (61%).
3-Chloroperbenzoic acid mediated stereoselective directed
epoxidation of allylic alcohol 6 proceeded cleanly to provide
the desired epoxyalcohol in excellent yield, which was further
converted to triol 7 by treatment with aqueous HCl. The
configuration of intermediate 7 was unambiguously estab-
lished by X-ray crystal structure analysis (Figure 1).
After having established the correct functionalization of
both the B ring and the dihydropyrone carbinol, we next
addressed the elaboration of the A ring. In particular, the
final steps from 7 were carried out without the use of
protecting groups, thus exploiting the inherent reactivity
pattern. Therefore, the challenge consisted of the correct
Figure 1. X-ray crystal structures of compounds 5, 7 and 10.[19]
8408 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 8407 –8411

transformation of the A ring to the enone in the presence of
two tertiary hydroxy groups. Oxidation of the secondary
hydroxy group of triol 7 by tetrapropylammonium perruthen-
ate (TPAP) and N-methyl morpholine N-oxide (NMO)
provided the corresponding ketone. Further oxidation of
this ketone to the enone 8 (81%) was mediated by 2-
iodoxybenzoic acid (IBX) under the optimized conditions
initially reported by Nicolaou, Montagnon, and Baran.[12]
Other methods such as the Saegusa oxidation provided
lower yields, and dichlorodicyanobenzoquinone did not
result in the desired product. The enone 8 was epoxidized
by aqueous hydrogen peroxide in the presence of triton B,[13]
affording epoxy ketone 9 (60%) and setting the stage for the
final key step of the synthesis: a Wharton carbonyl trans-
position[14] of 9. Hence, compound 9 was treated with
hydrazine hydrochloride in the presence of base to give the
rearranged allylic alcohol. This reaction proceeded smoothly
to furnish, after subsequent oxidation by pyridinium dichro-
mate, withanolide A in good yield (50% over two steps). The
analytical data (1H and 13C NMR spectroscopy, MS, UV
spectroscopy, melting point, HPLC analyses) for synthetic
withanolide A (1) were found to be in full agreement with
those of an isolated natural sample.[15] In addition, the identity
of natural and synthetic samples was confirmed by mixing
natural and synthetic samples and analyzing them by NMR
spectroscopy and HPLC.
identical to that obtained after the Wharton reaction.
Acetylation of compound 5 under standard conditions led
to the protected lactone 11, which demonstrates the possi-
bility for functionalization at this secondary OH group.
Withanolide A (1) was evaluated with regard to its
neuritogenic properties in the human SH-SY5Y neuroblas-
toma cell line according to a published method.[6a,17] Briefly,
cells were grown on 24-well plates in minimal essential
medium (MEM) supplemented with 5% fetal bovine serum
(FBS) in the presence of the corresponding compounds (1 mm,
0.1% DMSO) in a humidified atmosphere. After six days,
cells were examined under a phase-contrast microscope.
Those having at least one neurite with a length of more than
50 mm were counted as positives. In control experiments, cells
were treated with vehicle (DMSO 0.1%, negative blank) and
all-trans retinoic acid[18] (1 mm, positive blank). Pictures were
taken from six random areas of the wells and evaluated for
neurite positive cells. All assays were conducted in triplicate,
and at least 300 cells were counted in each experiment. Error
bars denote the standard error of the mean (SEM). To obtain
more accurate results for withanolide A, more than 1500 cells
were evaluated (n = 18–21). Interestingly, when we used
conditions published by Tohda et al.,[6a,17] we were able to
reproduce these results (DMSO vehicle 12% differentiated
cells vs. withanolide A 22%, data not shown).
We then examined the same compounds on collagen-
coated 24-well plates to support adhesion of the cells under
otherwise identical conditions. The vehicle control experi-
ments (0.1% DMSO) displayed a strikingly different pheno-
type compared to cells treated with withanolide A or retinoic
acid (Figure 2). In the former case, we almost exclusively
observed large cell aggregate formation with only few isolated
Derivatives of 1 were obtained either by semi-synthesis of
the natural product (extracted and purified from the dried
roots of Withania somnifera)[15,6a] or by further transforma-
tions of synthetic intermediates (Scheme 2). The challenge
Figure 2. Neurite outgrowth induced by withanolide A (1) and negative
(DMSO) and positive (retinoic acid) controls in human SH-SY5Y cells
in minimal essential medium (MEM) and Dulbecco’s modified Eagle’s
medium (DMEM). Error bars denote SEM. Representative micrographs
are given in the Supporting Information. For MEM, the phenotypes
induced by natural and synthetic withanolide A are in full agreement.
Scheme 2. Synthesis of A-ring-modified derivatives. DMAP=4-(dime-
thylamino)pyridine.
with regard to semi-synthesis of 1 resides in its low solubility
and the high reactivity of the A ring, which led to decom-
position and undesired side reactions under a variety of
conditions. After considerable experimentation, regioselec-
tive 1,2-reduction of the a,b-unsaturated ketone to the allylic
alcohol 10 by Luche reduction[16] could be achieved in good
yields. The configuration was established by X-ray crystal
structure analysis (Figure 1). Interestingly, the configuration
of the C1 stereogenic center in compound 10 was found to be
cells. Cells treated with withanolide A formed aggregates to a
lesser extent, with many viable isolated cells that were often
found to be differentiated. Both synthetic and isolated
material displayed phenotypes that are in full agreement.
Finally, treatment with retinoic acid completely suppressed
aggregate formation and led to fully differentiated pheno-
types.
Angew. Chem. Int. Ed. 2011, 50, 8407 –8411
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 8409

Communications
We then incubated the cells in Dulbeccoꢀs modified
binding of withanolide A (1) to secretase targets as suggested
earlier[7] appears unlikely, as shown by enzyme assays. These
results on the synthesis, derivatization, and neuritogenic
activity open the way for more detailed studies on the
mechanism of action of withanolide A (1), which will be
carried out in our laboratories.
Eagleꢀs medium (DMEM, 10% FBS, antibiotics; higher
nutrient concentration in the medium) in collagen-coated
wells with the compounds and controls. In these cases, the
large difference obtained for negative control experiments in
MEM compared to withanolide A (14% vs. 46%) was
reduced to a nonsignificant difference (28% vs. 32%). Also
in rat cortical neurons, a reassessment of the neuritogenic
activity has been proposed recently.[7] In the presence of
DMEM, isolated non-differentiated cells appear to be more
viable, thus leveling the observed ratios. These experiments
therefore suggest that the neuritogenic phenotype observed
in SY5Y cells appears to be conditional related to medium
(MEM vs. DMEM) or coating of the wells. Further studies
should thus address the question of whether withanolide A
exerts a genuine neuritogenic effect.
Concerning the mechanism of action of withanolide A (1),
a recent study by Chan and co-workers[7] reported that this
steroid lactone is able to modulate several secretase targets
relevant to neurodegeneration. Furthermore, this study
reported the docking of withanolide A into beta-secretase 1
(BACE1) suggesting enzyme inhibition by binding.[7] We have
evaluated the binding of withanolide A (1) against several
proteases of potential relevance to neurodegenerative dis-
eases (Table 1); however, in these assays withanolide A and
derivatives 5 and 11 were found to be inactive up to a
Received: March 16, 2011
Revised: May 1, 2011
Published online: July 15, 2011
Keywords: chemical biology · natural products ·
.
neurochemistry · total synthesis · steroids
[1] Selected articles: a) J. Harris, R. C. Kessler, M. Gazzaniga, P.
Campbell, M. J. Farah, Nature 2008, 456, 702 – 705; b) M. J.
Farah, J. Illes, R. Cook-Deegan, H. Gardner, E. Kandel, P. King,
E. Parens, B. Sahakian, P. R. Wolpe, Nat. Rev. Neurosci. 2004, 5,
421 – 425.
[2] Reviews: a) P. Williams, A. Sorribas, M.-J. R. Howes, Nat. Prod.
Rep. 2011, 28, 48 – 77; b) R. M. Wilson, S. J. Danishefsky, Acc.
Chem. Res. 2006, 39, 539 – 549; c) C. Tohda, T. Kuboyama, K.
Komatsu, Neurosignals 2005, 14, 34 – 45.
[3] a) S. E. Hyman, Neuron 2011, 69, 595 – 598; b) J. C. Lucke, S. K.
Bell, B. J. Partridge, W. D. Hall, EMBO Rep. 2011, 12, 197 – 201.
[4] Selected recent examples: a) H. J. Jessen, A. Schuhmacher, T.
Shaw, A. Pfaltz, K. Gademann, Angew. Chem. 2011, 123, 4308 –
4312; Angew. Chem. Int. Ed. 2011, 50, 4222 – 4226; b) H. J.
Jessen, D. Barbaras, M. Hamburger, K. Gademann, Org. Lett.
2009, 11, 3446 – 3449; c) C. Yuan, C.-T. Chang, A. Axelrod, D.
Siegel, J. Am. Chem. Soc. 2010, 132, 5924 – 5925; d) A. P.-J.
Chen, C. C. Mꢁller, H. M. Cooper, C. M. Williams, Org. Lett.
2009, 11, 3758 – 3761; e) S.-J. Min, S. J. Danishefsky, Angew.
Chem. 2007, 119, 2249 – 2252; Angew. Chem. Int. Ed. 2007, 46,
2199 – 2202.
Table 1: Determined IC50 values [mm] of compounds 1, 5, and 11 against
targets potentially relevant to neurodegenerative diseases (BACE1 and
cathepsin) and control enzymes (plasmepsin).
BACE1
Cath D[a]
Cath E[a]
PM I[b]
PM II[b]
PM IV[b]
1
5
11
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
67Æ4
46Æ14
>100
>100
52Æ5
>100
>100
31Æ7
[5] Review: S. K. Kulkarni, A. Dhir, Prog. Neuro-Psychopharmacol.
Biol. Psychiatry 2008, 32, 1093 – 1105.
[6] a) J. Zhao, N. Nakamura, M. Hattori, T. Kuboyama, C. Tohda, K.
Komatsu, Chem. Pharm. Bull. 2002, 50, 760 – 765; b) T.
Kuboyama, C. Tohda, K. Komatsu, Br. J. Pharmacol. 2005, 144,
961 – 971.
[a] cath=human cathepsin. [b] PM=plasmepsin.
concentration of 100 mm. Only very weak activity was found
against plasmepsin I (compound 5) and plasmepsin I, II, and
IV (compound 11). While it remains feasible that with-
anolide A is able to modulate the expression of BACE1, as
demonstrated by Chan and co-workers,[7] direct interaction of
1 with BACE1 as suggested[7] appears very unlikely based on
the values reported herein.
In conclusion, we have reported the first successful
preparation of the pharmacologically important steroid
lactone withanolide A (1) from a simple precursor and have
evaluated its neuritogenic properties and mechanism of
action related to secretase targets. Notable features of this
synthesis include: 1) a highly diastereoselective singlet-
oxygen-mediated photooxygenative olefin migration, 2) a
strategy that minimizes the use of protecting groups and
exploits the inherent reactivity pattern in the endgame, and
3) a Wharton transposition for the establishment of the
A ring. In addition, biological studies of neurite outgrowth in
human SH-SY5Y cells demonstrated the dependence of
neuritogenic properties on parameters such as medium and
surface coating. Concerning the mechanism of action, direct
[7] S. P. Patil, S. Maki, S. A. Khedkar, A. C. Rigby, C. Chan, J. Nat.
Prod. 2010, 73, 1196 – 1202.
[8] Reviews on withanolide structures, bioactivities and synthetic
approaches: a) L.-X. Chen, H. He, F. Qiu, Nat. Prod. Rep. 2011,
28, 705 – 740; b) I. Kirson, E. Glotter, J. Nat. Prod. 1981, 44, 633 –
647; c) N. V. Kovganko, Z. N. Kahkan, Chem. Nat. Compd. 1997,
33, 133 – 145; d) K. Gamoh, M. Hirayama, N. Ikekawa, J. Chem.
Soc. Perkin Trans. 1 1984, 449 – 454; e) A. Perez-Medrano, P. A.
Grieco, J. Am. Chem. Soc. 1991, 113, 1057 – 1059; f) E. Glotter, S.
Kumar, M. Sahai, A. Goldman, I. Kirson, M. Medelovici, J.
Chem. Soc. Perkin Trans. 1 1991, 739 – 745; g) M. Ishiguro, A.
Kajikawa, T. Haruyama, Y. Ogura, M. Okubayashi, M. Morisaki,
N. Ikekawa, J. Chem. Soc. Perkin Trans. 1 1975, 2295 – 2302; h) P.
Neogi, M. Kawai, Y. Butsugan, Y. Mori, M. Suzuki, Bull. Chem.
Soc. Jpn. 1988, 61, 4479 – 4481.
[9] B. B. Shingate, B. G. Hazra, V. S. Pore, R. G. Gonnade, M.
Bhadbhade, Tetrahedron 2007, 63, 5622 – 5635.
[10] R. Huston, M. Rey, A. S. Dreiding, Helv. Chim. Acta 1982, 65,
1563 – 1575.
[11] a) W. Adam, E. Staab, Liebigs Ann. Chem. 1988, 757 – 759;
b) review: M. Prein, W. Adam, Angew. Chem. 1996, 108, 519 –
538; Angew. Chem. Int. Ed. 1996, 35, 477 – 494.
[12] K. C. Nicolaou, T. Montagnon, P. S. Baran, Angew. Chem. 2002,
114, 1035 – 1038; Angew. Chem. Int. Ed. 2002, 41, 993 – 996.
8410 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 8407 –8411

[13] M. T. Barros, C. D. Maycock, M. R. Ventura, Tetrahedron 1999,
55, 3233 – 3244.
[17] T. Kuboyama, C. Tohda, J. Zhao, N. Nakamura, M. Hattori, K.
Komatsu, Neuroreport 2002, 13, 1715 – 1720.
[18] Review: M. Clagett-Dame, E. M. McNeill, P. D. Muley, J.
Neurobiol. 2006, 66, 739 – 756.
[19] CCDC 816947 (5), 816948 (7), 816949 (10) contain the supple-
mentary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[14] C. Dupuy, J. L. Luche, Tetrahedron 1989, 45, 3437 – 3444.
[15] S. S. Subramanian, P. D. Sethi, E. Glotter, I. Kirson, D. Lavie,
Phytochemistry 1971, 10, 685 – 688.
[16] A. L. Gemal, J.-L. Luche, J. Am. Chem. Soc. 1981, 103, 5454 –
5459.
Angew. Chem. Int. Ed. 2011, 50, 8407 –8411
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8411