Total synthesis of candicanoside A, a potent antitumor saponin with a rearranged steroid side chain

Article abstract of DOI:10.1002/anie.200604761

(Chemical Equation Presented) The fused-ring scaffold of candicanoside A (1), a potent antitumor agent with a unique differential pattern, has now been synthesized from inexpensive starting materials. Compound 1 has the potential to act as a lead compound

Full text of DOI:10.1002/anie.200604761

Angewandte
Chemie
DOI: 10.1002/anie.200604761
Natural Products Synthesis
Total Synthesis of Candicanoside A, a Potent Antitumor Saponin with a
Rearranged Steroid Side Chain**
Pingping Tang and Biao Yu*
Since the disclosure of the exceptionally potent antitumor
activity of OSW saponins,[1] much attention has been devoted
to cholestan glycosides of this type. These compounds occur
in the bulbs of Ornithogalum saudersiae and taxonomically
related plants, which are members of a lily family native to
southern Africa.[2] Continuous research on these plants led to
the discovery of minor cholestan glycosides of a new type with
a rearranged steroidal side chain.[3,4] Candicanoside A( 1),
isolated in 2000 from the bulbs of Galtonia candicans
(0.00076% of the fresh weight), is such a compound.[3] It
differs from the other congeners through its unique fused-ring
scaffold created by acetal formation between the aldehyde
group at C23 and the diol at C16,18. This compound shows
potent antitumor activity (e.g., IC50 = 32 nm against HL-60
human promyelocytic leukemia cells) comparable to that of
the clinically applied anticancer agents etoposide and metho-
trexate, whereas the congener in which C16 and C23 form
part of a hemiacetal group was found to be inactive.[4d] More
Scheme 1. Retrosynthetic analysis of candicanoside A. AZMB=2-(azi-
domethyl)benzoyl, Bz=benzoyl.
interestingly, candicanoside Adisplays differential cytotoxic-
ities against various tumor cell lines, and its antitumor profile
does not correlate with that of any other antitumor com-
pound.[3] Thus, candicanoside Amay have a unique mode of
action and the potential to act as a lead for the synthesis of a
new type of antitumor agents. Herein we report the first total
synthesis of this interesting natural product.
rhamnosyl trichloroacetimidates 3 and 4[8] were selected as
precursors to the desired diasaccharide. Selective removal of
the 2-O-AZMB group in 3 would release a hydroxy group at
this position for subsequent coupling with 4.[7] Finally, the
acetyl and benzoyl groups remaining in the resulting dis-
accharide could be removed under mild basic conditions to
provide candicanoside A( 1).
Agreat synthetic challenge was presented by the novel
scaffold of the steroid aglycone 2. We planned to use the
readily available and cheap steroid dehydroisoandrosterone
(5) as the starting material. The stereoselective introduction
of the diol at C16,18, installation of the rearranged side chain
at C17, and formation of the acetal at C23 would then be
required to provide 2.
A20-OH group in the steroids could facilitate the
functionalization of the proximal 18-Me group.[9] Therefore,
compound 7, with a hydroxy group at C20(R)[10] and 3,5-cyclo-
6-methoxy protection of the 3-hydroxy-5,6-ene functionality
in the AB ring, was prepared from alkene 6 by hydroboration
and oxidation; compound 6 was obtained conveniently from 5
in three steps (62%).[11] Irradiation of 7 in the presence of
iodine and DIB in cyclohexane with a 300-W tungsten lamp
for approximately 30 min, followed immediately by oxidation
with PCC, provided the desired ketone 8 with an iodo
substituent at C18 in a satisfactory yield (50%).[9] Hydrolysis
of the iodide 8 with silver acetate in aqueous dioxane resulted
in the formation of the hemiketal 9 (92%).[12] The reduction
of 9 with NaBH4 yielded a pair of C18,20(R/S) diols, which
was subjected directly to selective acetylation to provide the
The assembly of the target disaccharide saponin 1 would
demand a mild and stereoselective approach to the glyco-
sylation of the cholestan aglycone 2, which contains two
double bonds and an acetal functionality (Scheme 1).[5]
Glycosyl imidates with a benzoyl group at the 2-O-position
have been shown to be ideal donors, as only a catalytic
amount of TMSOTf is required to promote their coupling to
steroids and triterpenoids. The 2-O-benzoyl group controls
the construction of the 1,2-trans glycosidic bond by neighbor-
ing-group participation while minimizing the formation of the
corresponding orthoester.[6,7] Therefore, the d-glucosyl and l-
[*] P. Tang, Prof. B. Yu
State Key Laboratory of Bioorganic and
Natural Products Chemistry
Shanghai Institute of Organic Chemistry
Chinese Academy of Sciences
354 Fenglin Road, Shanghai 200032 (China)
Fax: (+86)21-6416-6128
E-mail: byu@mail.sioc.ac.cn
[**] This research was supported by the National Natural Science
Foundation of China (20321202), the Chinese Academy of Sciences
(KGCX2-SW-213), and the Committee of Science and Technology of
Shanghai (04DZ14901).
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 2527 –2530
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2527

Communications
acetoxy alcohol 10 with the R configuration at C20 as the
Following the successful installation of the hydroxy
groups at C16 and C18, we were only able to introduce the
rest of the side chain at C20 through a multistep sequence.
Thus, subjection of the allylic alcohol 12 to the standard
conditions for Johnson–Claisen rearrangement afforded the
ester 13 as a single isomer in 87% yield.[16] The R configura-
tion at C20 was confirmed by X-ray diffraction analysis of a
later derivative.[17] The acetate group at C18 and ester group
at C23 in 13 were found to be detrimental to the
readily isolable major product (50% over two steps).[13] Next,
ꢀ
the required Z alkene at C17 C20 was regenerated by
dehydration under mild conditions to give 11 (79%).[14] The
stereoselective introduction of the allylic hydroxy group at
C16 with selenium dioxide and tert-butylhydroperoxide then
led to the desired alcohol 12 (82%; Scheme 2).[15]
hydroboration reaction of the alkene to intro-
duce the hydroxy group at C16. Therefore, the
acetyl protecting group was exchanged for a
TBS protecting group. The ester group in the
resulting compound 14 was reduced to the
corresponding alcohol with LiAlH4. Subsequent
treatment of the resulting alkene with diborane
followed by H2O2 afforded the desired 16a,23-
dihydroxy compound 15 in a stereoselective
manner and 80% yield. Swern oxidation of 15,
followed by further oxidation of the resulting
[18]
aldehyde to the carboxylic acid with NaClO2
and subsequent formation of the corresponding
methyl ester with diazomethane provided the
keto methyl ester 16 in good yield (87%).
LiAlH(OtBu)3 was found to be the best reagent
for the selective reduction of the keto group in
16. The resulting alcohol underwent concom-
itant cyclization to generate the lactone 17. The
use of other reducing agents, such as NaBH4/
CeCl3 and l-selectride, led to considerable
amounts of the products of overreduction.
The remaining portion of the side chain was
introduced successfully at C22 (in lactone 17)
through an aldol condensation. Thus, the treat-
ment of 17 with LDAin THF, followed by the
addition of isobutyraldehyde at ꢀ788C, gave a
diastereoisomeric mixture of alcohols 18, which
was subjected directly to dehydration with
POCl3 in pyridine to provide the a,b-unsatu-
rated lactone 19 in 50% yield. No improvement
in the yield of the aldol reaction was found
when KHMDS (potassium hexamethyldisila-
zide) was used as the base,[19] when HMPA
(hexamethyl phosphoramide) was present in
Scheme 2. Synthesis of the steroid aglycone 2: a) 9-BBN, THF, room temperature; then
H2O2, NaOH, room temperature, 96%; b) DIB, I2, cyclohexane, room temperature, hn;
c) PCC, NaOAc, CH2Cl2, 4-Š MS, room temperature, 50% (two steps); d) AgOAc, 1,4-
dioxane/H2O (9:1), 658C, 92%; e) NaBH4, CH2Cl2/CH3OH (1:1), room temperature;
f) AcCl, pyridine, ꢀ208C, 50% (two steps); g) POCl3, pyridine, ꢀ208C!RT, 79%;
h) SeO2, tBuOOH, CH2Cl2/H2O, room temperature, 82%; i) CH3C(OCH3)3, propanoic
acid (cat.), 1408C, 87%; j) NaOMe, MeOH/THF (1:1), 08C; k) TBSOTf, 2,6-lutidine,
room temperature, 90% (two steps); l) LiAlH4, THF, room temperature; m) BH3, THF,
08C; then H2O2, NaOH, 08C, 80% (two steps); n) Swern oxidation; NaClO2, Na2HPO4,
2-methyl-2-butene, tBuOH, room temperature; then CH2N2, ether, room temperature,
87%; o) LiAlH(OtBu)3, THF, room temperature, 81%; p) LDA, THF, ꢀ788C; then
isobutyraldehyde, ꢀ788C; q) POCl3, pyridine, ꢀ208C!RT, 50% (two steps); r) hn
(l=254 nm), MeOH, room temperature, 90%; s) l-selectride, ꢀ788C; then Ac2O,
DMAP, Et3N, room temperature, 97%; t) HF (40%), CH3CN, room temperature, 85%.
9-BBN=9-borabicyclo[3.3.1]nonane, DIB=(diacetoxyiodo)benzene, DMAP=4-
dimethylaminopyridine, LDA=lithium diisopropylamide, MS=molecular sieves,
PCC=pyridinium chlorochromate, TBS=tert-butyldimethylsilyl, Tf=trifluoromethane-
sulfonyl.
the solvent, or with Bu2BOTf/diisopropylethyl-
amine.[20] Fortunately, the rearrangement of the
conjugated double bond in the lactone 19 to
ꢀ
give a nonconjugated double bond at C24 25
was promoted by irradiation with UV light[21] to
provide 20 as a single isomer at C22 in excellent
yield. NOE correlations between 22-H (d =
2.89 ppm, dd, J = 12.3, 9.8 Hz) and 16-H (d =
4.75 ppm, m), and between 24-H (d = 5.06 ppm,
d, J = 9.6 Hz) and 20-H (d = 2.35–2.27 ppm, m)
were found, but not between 22-H and 20-H, as
evidence for the desired R configuration at C22.
The lactone 20 was then reduced with l-
selectride; the resulting deprotonated lactol was
trapped in situ with acetic anhydride to give the
2528
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 2527 –2530

Angewandte
Chemie
acetate 21 in 97% yield.[22] Finally, HF (40%) was used to
remove the 18-O-TBS group in 21. The desired acetal in the
target compound was formed simultaneously, and the hy-
ꢀ
droxy group at C3 and alkene at C5 C6 were also recovered
from the 3,5-cyclo-6-methoxy protection, which had remained
intact since the beginning of the synthesis. Thus, the aglycone
2 was elaborated successfully in 23 steps and in 1.5% overall
yield from the commercially available steroid dehydroisoan-
drosterone (5).
The allyl glucopyranoside 22 with its 2-OH group
protected with AZMB had already been prepared in our
laboratory.[7] The preparation of the imidate 3 took four more
steps: removal of the benzylidene protecting group, acetyla-
tion of the hydroxy groups at C3,C4, and C6, cleavage of the
allyl group at the anomeric position, and subsequent forma-
tion of the trichloroacetimidate (Scheme 3). Thus, the desired
glucosyl donor 3 was prepared from glucose in seven steps
and 26% overall yield. Another three steps were required for
the preparation of the rhamnosyl donor 4 (in 76% yield).[8]
Scheme 4. Completion of the synthesis of candicanoside A (1):
a) TfOH (0.2 equiv), CH2Cl2, 4-Š MS, room temperature, 96%; b) PBu3,
THF/H2O, room temperature; c) TfOH (0.2 equiv), toluene, 4-Š MS,
room temperature, 81% (two steps); d) NaOMe, THF/MeOH (1:1),
room temperature, 90%.
activity with a unique differential pattern, become projects
for further research.
Received: November 23, 2006
Published online: February 20, 2007
Scheme 3. Synthesis of the glucose donor 3: a) AcOH (70%), 708C;
b) Ac2O, Et3N, DMAP, CH2Cl2, room temperature, 86% (two steps);
c) PdCl2, CH2Cl2/MeOH (1:1), room temperature, 85%; d) CNCCl3,
DBU, CH2Cl2, room temperature, 90%. DBU=1,8-diazabicyclo-
[5.4.0]undec-7-ene.
Keywords: antitumor agents · glycosylation · natural products ·
.
steroids · total synthesis
[1] a) Y. Mimaki, M. Kuroda, A. Kameyama, Y. Sashida, T. Hirano,
K. Oka, R. Maekawa, T. Wada, K. Sugita, J. A. Beutler, Bioorg.
Med. Chem. Lett. 1997, 7, 633; b) A. M. Rouhi, Chem. Eng. News
1995, 73, 28.
[2] B. Shi, P. Tang, X. Hu, J. O. Liu, B. Yu, J. Org. Chem. 2005, 70,
10354, and references therein.
As expected, glycosylation of the steroid 2 with the
glucosyl imidate 3 in the presence of TfOH (0.2 equiv)
afforded the b glycoside 25 in excellent yield (96%;
Scheme 4). The AZMB group was then removed selectively
in the presence of acetyl groups with PBu3 to provide 26. The
unmasked hydroxy group in 26 is sterically hindered; none-
theless, coupling with the benzoyl-protected rhamnosyl
imidate 4 in the presence of TfOH (0.2 equiv) produced the
desired disaccharide 27 in 81% yield (over two steps). Finally,
the acetyl and benzoyl groups on the sugar residue in 27 were
removed with NaOMe in MeOH/THF at room temperature
to furnish the target candicanoside A( 1) in 90% yield. The
analytical data of 1 are in good agreement with those reported
in the literature.[3,17]
[3] Y. Mimaki, M. Kuroda, Y. Sashida, T. Yamori, T. Tsuruo, Helv.
Chim. Acta 2000, 83, 2698.
[4] a) M. Kuroda, Y. Mimaki, Y. Sashida, T. Nikaido, T. Ohmoto,
Tetrahedron Lett. 1993, 34, 6073; b) Y. Mimaki, M. Kuroda, A.
Kameyama, Y. Sashida, T. Hirano, K. Oka, A. Dobashi, K.
Koike, T. Nikaido, Bioorg. Med. Chem. Lett. 1996, 6, 2635; c) M.
Kuroda, Y. Mimaki, Y. Sashida, Phytochemistry 1999, 52, 435;
d) M. Kuroda, Y. Mimaki, A. Yokosuka, Y. Sashida, Chem.
Pharm. Bull. 2001, 49, 1042.
[5] R. Suhr, P. Pfefferkorn, S. Weingarten, J. Thiem, Org. Biomol.
Chem. 2003, 1, 4373.
In summary, candicanoside A( 1), a steroid disaccharide
with a novel fused-ring steroid scaffold, has been synthesized
for the first time. The synthesis requires a total of 37 steps
from cheap starting materials (dehydroisoandrosterone, d-
glucose, and l-rhamnose), with a longest linear sequence of
27 steps and a 1.0% overall yield of 1. Improvement of the
synthesis and, more importantly, studies on the structure–
activity relationships and mechanism of action of this novel
natural lead compound, which shows potent antitumor
[6] S. Deng, B. Yu, J. Xie, Y. Hui, J. Org. Chem. 1999, 64, 7265.
[7] W. Peng, X. Han, B. Yu, Synthesis 2004, 1641.
[8] T. Ziegler, F. Bien, C. Jurisch, Tetrahedron: Asymmetry 1998, 9,
765.
[9] P. de Armas, J. I. Concepcion, C. G. Francisco, R. Hernandez,
J. A. Salazar, E. Suarez, J. Chem. Soc. Perkin Trans. 1 1989, 405.
[10] D. J. Vanderah, C. Djerassi, J. Org. Chem. 1978, 43, 1442.
[11] N. R. Schmuff, B. M. Trost, J. Org. Chem. 1983, 48, 1404.
[12] Z. Yang, J. Meinwald, Tetrahedron Lett. 1998, 39, 3425.
[13] J. F. Weet, G. R. Lenz, J. Med. Chem. 1985, 28, 233.
Angew. Chem. Int. Ed. 2007, 46, 2527 –2530
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2529

Communications
[14] B. G. Hazra, P. L. Joshi, B. B. Bahule, N. P. Argade, V. S. Pore,
[18] B. S. Bal, W. E. Childers, Jr., H. W. Pinnick, Tetrahedron 1981,
37, 2091.
[19] E. Bourque, P. Deslongchamps, Y. L. Dory, J. Org. Chem. 2003,
68, 2390.
M. D. Chordia, Tetrahedron 1994, 50, 2523.
[15] M. A. Umbreit, K. B. Sharpless, J. Am. Chem. Soc. 1977, 99,
5526.
[20] T. Inoue, J. Liu, D. C. Buske, A. Abiko, J. Org. Chem. 2002, 67,
5250.
[21] a) R. Mortezaei, D. Awandi, F. Henin, J. Muzart, J.-P. Pete, J.
Am. Chem. Soc. 1988, 110, 4824; b) O. Piva, D. Caramelle,
Tetrahedron: Asymmetry 1995, 6, 831.
[16] W. S. Johnson, L. Werthemann, W. R. Bartlett, T. J. Brocksom,
T. T. Li, D. J. Faulkner, M. R. Petersen, J. Am. Chem. Soc. 1970,
92, 741.
[17] For the preparation and characterization of all compounds
mentioned in the text, see the Supporting Information.
[22] V. Dahanukar, S. Rychnovsky, J. Org. Chem. 1996, 61, 8317.
2530
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 2527 –2530