Rapid construction of the cortistatin pentacyclic core

Article abstract of DOI:10.1002/anie.200802203

(Chemical Equation Presented) Ringing in cortistatin: The pentacyclic core of the cortistatin steroidal alkaloids has been readily assembled in eleven steps from simple building blocks (see scheme; PMB=para-methoxybenzyl, TBS=tert-butyldimethylsilyl). An

Full text of DOI:10.1002/anie.200802203

Communications
DOI: 10.1002/anie.200802203
Natural Products
Rapid Construction of the Cortistatin Pentacyclic Core**
Eric M. Simmons, Alison R. Hardin, Xuelei Guo, and Richmond Sarpong*
Angiogenesis, the formation of new
blood capillaries, is an essential process
for embryonic development and wound
repair. Under normal circumstances, this
process is tightly regulated and is pro-
moted by angiogenic polypeptides such
as vascular endothelial growth factor
(VEGF), basic fibroblast growth factor
[1]
(bFGF)and angiogenin.
However,
angiogenesis also plays a key role in
tumor growth and metastasis, and the
discovery of angiogenesis inhibitors has
generated excitement about their poten-
tial use in the treatment of cancer. When
administered alongside traditional che-
motherapeutics, antiangiogenics sup-
press the recurrence of tumor growth,
leading to curative treatment in some
animal models.[2] In contrast to many
existing chemotherapies, antiangiogenic
therapy typically exhibits low toxicity
and drug resistance does not appear to
be a significant problem.[1b] Research in
this area is in the early stages, but
recently Bevacizumab,
a monoclonal
antibody against VEGF, was approved
for the treatment of colon and breast
cancers.[3] Because of these promising
initial results, there is intense interest in
the discovery and study of novel anti-
angiogenic factors.
Scheme 1. Members of the cortistatin family and their respective IC50 values against HUVECs.
Herein, we report our initial syn-
thetic studies on the antiangiogenic com-
pound cortistatin A (1, Scheme 1). This compound is among
the first members of a new family of rearranged steroidal
alkaloids (1–4)that were isolated from the Indonesian marine
sponge Corticium simplex in 2006 by Kobayashi and co-
workers.[4] All four compounds were found to have significant
antiangiogenic activity, with the most potent being cortista-
tin A (1), displaying an IC50 value of 1.8 nm against human
umbilical vein endothelial cells (HUVECs), which is a
standard model for antiangiogenic activity. Importantly, this
high activity was coupled with an exquisite selectivity (>
3000)relative to other cell types, thus arguing against general
cytotoxicity as a mode of action.
This initial isolation report was followed a year later by
the discovery of seven additional family members, cortista-
tins E–L, from the same source.[5] Although they share the
same pentacyclic core as their previously isolated congeners,
cortistatins E–H incorporate N-methylpiperidine or 3-meth-
ylpyridine in their side chains (see 5–8, Scheme 1)in place of
the isoquinoline fragment. In addition, the degree and
position of unsaturation and oxygenation along the northern
edge of the ABC rings differs relative to cortistatins A–D.[5a]
Cortistatins J–L (9–11)show similar variation in the ABC-
ring substitution pattern, but retain the isoquinoline ring at
C17.[5b] Preliminary structure–activity analysis suggests that
the presence of the isoquinoline moiety is crucial for potent
activity: while cortistatins A–D and J–L were found to have
[*] E. M. Simmons, A. R. Hardin, X. Guo, Prof. R. Sarpong
Department of Chemistry
University of California, Berkeley
Berkeley, CA 94720 (USA)
Fax: (+1)510-642-9675
E-mail: rsarpong@berkeley.edu
Homepage: http://www.cchem.berkeley.edu/rsgrp/
[**] We are grateful to UC Berkeley and GlaxoSmithKline for supporting
this research. E.M.S. acknowledges the ACS Division of Medicinal
Chemistry and Eli Lilly for a predoctoral fellowship.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200802203.
6650
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 6650 –6653

Angewandte
Chemie
IC50 values as low as 1.8 nm, cortistatins E–H were only
weakly active (IC50 value: 350 nm–1.9 mm). Furthermore, the
oxygenation at C16 or C17 of the D ring appears to
significantly attenuate activity (see 1–4).[6]
The cortistatins discovered to date share a unique
rearranged steroid core where the C19 methyl group has
been incorporated into the B ring, and an ether bridge links
C5 and C8. There are many steroids in which a cyclopropane
ring has formed from C9, C10 and the C19 methyl group. In
contrast, the corresponding ring-expanded seco-tetracycle
bearing a seven-membered B ring is relatively rare. This type
of arrangement has only previously been found in other
marine sponges in the Corticium genus,[7] as well as in
terrestrial plant species in the Buxus[8] and Cimicifuga[9]
genera. However, the ether-containing pentacycle that is
characteristic of the cortistatins is unprecedented among
known natural products. The clear synthetic challenge posed
by the cortistatins, coupled with their unique biological
activity has generated significant interest from the synthetic
community, but as yet a total synthesis has not been
reported.[10] Herein, we present a concise approach which
has led to the de novo assembly of the pentacyclic core of the
cortistatins. This approach now sets the stage for a total
synthesis of these intriguing natural products.
Scheme 2. Retrosynthetic analysis. PG=protecting group, TBS=tert-
butyldimethylsilyl.
One could imagine utilizing a commercially available
steroid as a starting point to arrive at a given member of the
cortistatin family through appropriate functional group
manipulations. Indeed, such partial syntheses are common
in the steroid literature,[11] and this approach would likely be
successful in devising a route to a specific cortistatin target, as
evidenced by the recent synthesis of 1 by Baran and co-
workers.[10] In considering our synthetic strategy, we sought a
flexible intermediate that would provide facile access to any
of the cortistatin natural products as well as any desired
synthetic analogues. To that end, we envisioned that penta-
cycle 12 (Scheme 2)could be tailored to a variety of
substitution patterns utilizing the trienone handle in the A
and B rings and the oxygen substituent at C17. The ether
bridge could be constructed by oxidative
Our synthesis of the cortistatin pentacyclic core com-
menced with aldehyde (ꢀ )-19 (Scheme 3), which was readily
prepared in four steps and with 53% overall yield from 18.[14]
Notably, 18 has been previously synthesized in enantiopure
form, thus providing a straightforward avenue for an enan-
tioselective synthesis. Fragment coupling of 19 with indanone
20[15] was achieved by aldol condensation to give enone 21 as a
single diastereomer in 51% yield following recrystallization.
Selective 1,4-reduction of the enone moiety of 21 was
achieved by treatment with K-Selectride[16] to yield indanol
22 (as an inconsequential mixture of diastereomers)and the
corresponding indanone in approximately a 1:2 ratio. The
dearomatization of phenol 13. We were
especially intrigued by the possibility of
arriving at 13 from benzocyclohepta-
diene 14, which we anticipated would
be readily available from alkynyl indene
15 by an enyne cycloisomerization reac-
tion, which was developed in our labo-
ratories and was previously applied to
the total synthesis of several icetexane
diterpenoid natural products.[12] Finally,
15 could be derived from indanone 16
and aldehyde 17, which in turn could be
prepared from the known ester 18.[13] Not
only would this strategy be amenable to
the preparation of synthetic analogues
modified in the A and D rings, but also
changes in the B- and C-ring substitution
Scheme 3. Synthesis of indene 23. Reagents and conditions: a) 20 (1.0 equiv), 19 (1.0 equiv),
KOH (1.0 equiv), EtOH/CH2Cl2 (3:1), RT, 2.5h, 51%; b) K-Selectride (3.0 equiv), THF, ꢁ788C,
2 h; then ꢁ788C!RT, 2 h; c) NaBH4 (1.0 equiv), MeOH/CH2Cl2 (1:1), 08C, 1 h; d) KHSO4
(1.0 equiv), toluene, 508C, 18 h, 65% (over 3 steps).
could easily be accommodated by appro-
priate manipulations prior to the con-
struction of the pentacycle.
Angew. Chem. Int. Ed. 2008, 47, 6650 –6653
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6651

Communications
remaining indanone was smoothly con-
verted into indanol 22 by treatment of the
crude mixture with NaBH4. Subsequent
dehydration of the secondary alcohol was
achieved by gentle heating of 22 with
KHSO4 in toluene to give indene 23 in
65% yield over the three-step sequence.
Notably, while thermally promoted
double-bond isomerization of indenes
has previously been observed in related
systems,[12a] no trace of the isomerization
product of 23 could be detected even after
36 h at 508C.
With indene 23 in hand, we antici-
pated that cycloisomerization could be
achieved with GaCl3 as a catalyst on the
basis of previous observations.[12a,17] How-
ever, preliminary studies indicated that
cleavage of the TBS ether was a compet-
ing side reaction. Gratifyingly, the desired
cycloheptadiene (24, Scheme 4)could be
readily obtained by treatment of 23 with a
catalytic amount of PtCl2.[18] In contrast to
the icetexane substrates from our previ-
ous studies, cycloisomerization of 23 with
PtCl2 occurred without significant forma-
tion of isomeric by-products. We have
also examined the cycloisomerization of
Scheme 5. Completion of the cortistatin core synthesis. Reagents and conditions:
a) TsNHNH2 (8.0 equiv), Et3N (16 equiv), 1,2-dichloroethane, 658C, 24 h, 95% after one
recycle; b) m-CPBA (1.2 equiv), NaHCO3 (2.0 equiv), CH2Cl2, 08C, 2.5h, 66% ( 29a);
c) MgBr2·OEt2 (6.0 equiv), Me2S (20 equiv), CH2Cl2, ꢁ788C!RT, 3 h; d) TESCl (1.5equiv),
imidazole (3.0 equiv), DMF, RT, 3 h; e) m-CPBA (1.2 equiv), NaHCO3 (2.0 equiv), CH2Cl2,
08C, 1.5h, 46% ( 29b over 3 steps from 28); f) nBuLi (5.0 equiv), THF, 08C, 1 h; g) PhI(OAc)2
(1.5equiv), CH 2Cl2/iPrOH/TFE (5:3:2), 0 8C, 30 min, 60% (over 2 steps from 29b). m-
CPBA=meta-chloroperbenzoic acid, DMF=N,N-dimethylformamide, Ts=para-toluenesul-
fonyl.
several
other
indene
substrates
(Scheme 4), which were easily synthe-
sized from the corresponding indanones[12a,19] by a sequence
analogous to that shown in Scheme 3. Pleasingly, each of these
alkynyl indenes underwent smooth cycloisomerization to the
corresponding cycloheptadiene tetracycle (see 25–27)upon
treatment with a catalytic amount of PtCl2.
selectively reduced with diimide to give tetrasubstituted
alkene 28 in excellent yield.[20] Subsequent treatment with
m-CPBA at 08C gave epoxide 29a as a single diastereomer.[21]
Although 29a proved to be exceedingly acid sensitive,
regioselective epoxide opening could be achieved by treat-
ment with n-butyllithum, which proceeded with concomitant
cleavage of the para-methoxybenzyl (PMB)ether to give
phenol 30.[22] Even though this reaction
To elaborate cycloheptadiene 24 (Scheme 5)to the
cortistatin core, the disubstituted double bond was chemo-
proceeded satisfactorily on a small amount
of material (1 mg), scale-up led to a
significant increase in the amount of
side-products and to variable yields of 30.
To circumvent this impediment, the PMB
group was removed under non-oxidative
conditions (MgBr2·OEt2/Me2S),[23] which
prevented oxidation of the electron-rich
benzene ring, and the resulting phenol was
directly converted into the triethylsilyl
[24]
(TES)ether.
Epoxidation of the TES
ether proceeded to give epoxide 29b in
46% yield over the three-step sequence
(from 28).[21] Gratifyingly, base-mediated
epoxide opening of 29b was accompanied
by the clean removal of the silyl group to
give 30 as a single product that did not
require purification.
With a reliable route to 30 in hand, we
turned our attention to the key tandem
Scheme 4. Cycloisomerization products.
oxidative
dearomatization/cyclization,
6652 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 6650 –6653

Angewandte
Chemie
Choudhary, S. Naz, A. Ata, B. Sener, S. Türköz, J. Nat. Prod.
which we expected could be achieved using hypervalent
iodine.[25] Although protic solvents are typically more effec-
tive at stabilizing the cationic intermediates generated under
the oxidation conditions, we posited that competitive solvent
addition at C1 could be problematic. Indeed, treatment of 30
with PhI(OAc)2 in 1:1 CH2Cl2/MeOH led to a 1:1 mixture of
the desired pentacycle 31 and the C1 dimethyl ketal.[26]
Replacement of MeOH with the less nucleophilic, protic
solvent 2,2,2-trifluoroethanol (TFE)led to 31 as the major
product, but this was accompanied by a significant amount of
decomposition products. We hypothesized that a bulky, non-
fluorinated, protic solvent might stabilize the reactive inter-
mediates without acting as a nucleophile. In the event,
treatment of 30 in CH2Cl2/iPrOH with a solution of PhI-
(OAc)2 in TFE gave pentacyclic trienone 31 in 60% yield over
two steps (from 29b), with no detectable products arising
from solvent addition. The structure of 31 was supported by
extensive 2D NMR analysis (COSY, HMQC, and NOESY
spectroscopy). The stereochemistry of the ether bridge was
secured by observation of an nOe interaction between the
C18 methyl group and one of the C7 protons, which is
analogous to that observed for cortistatin A.[4]
In summary, we have reported the construction of the
pentacyclic core of the cortistatin steroidal alkaloids through
a concise strategy featuring a PtCl2-catalyzed enyne cyclo-
isomerization and an oxidative dearomatization/cyclization as
the key steps. The entire sequence proceeds in 11 steps from
aldehyde 19 and is directly amenable to an asymmetric
synthesis given the ready availability of enantiopure 18.
Application of the strategy disclosed here to the total
synthesis of cortistatin A is underway and will be reported
in due course.[27]
1997, 60, 770 – 774; c)Atta-ur-Rahman, F. Noor-e-ain, M. I.
Choudhary, Z. Parveen, S. Türköz, B. Sener, J. Nat. Prod. 1997,
60, 976 – 981.
[9] S. Kadota, J. X. Li, K. Tanaka, T. Namba, Tetrahedron 1995, 51,
1143 – 1166.
[10] During the review of this manuscript, an elegant semi-synthesis
of cortistatin A was reported by Baran and co-workers, see:
R. A. Shenvi, C. A. Guerrero, J. Shi, C.-C. Li, P. S. Baran, J. Am.
Chem. Soc. 2008, 130, 7241 – 7243.
[11] A. B. Turner in The Chemistry ofNatural Products (Ed.: R. H.
Thomson), 2nd ed., Blackie, New York, 1993, Chapter 4.
[12] a)E. M. Simmons, R. Sarpong, Org. Lett. 2006, 8, 2883 – 2886;
b)E. M. Simmons, J. R. Yen, R. Sarpong, Org. Lett. 2007, 9,
2705 – 2708.
[13] W. Oppolzer, D. A. Roberts, Helv. Chim. Acta 1980, 63, 1703 –
1705.
[14] For details, see the Supporting Information.
[15] For the synthesis of 20, see the Supporting Information.
[16] J. M. Fortunato, B. Ganem, J. Org. Chem. 1976, 41, 2194 – 2200.
[17] N. Chatani, H. Inoue, T. Kotsuma, S. Murai, J. Am. Chem. Soc.
2002, 124, 10294 – 10295.
[18] For reviews, see: a)V. Michelet, P. Y. Toullec, J.-P. GenÞt,
Angew. Chem. 2008, 120, 4338 – 4386; Angew. Chem. Int. Ed.
2008, 47, 4268 – 4315; b)A. Fürstner, P. W. Davies, Angew.
Chem. 2007, 119, 3478 – 3519; Angew. Chem. Int. Ed. 2007, 46,
3410 – 3449.
[19] a)K. Ohkata, M. Akiyama, K. Wada, S. Sakaue, Y. Toda, T.
Hanafusa, J. Org. Chem. 1984, 49, 2517 – 2520; b)D. R. Adams,
M. A. J. Duncton, Synth. Commun. 2001, 31, 2029 – 2036.
[20] For a related diene hydrogenation using diimide, see: X. Li,
R. E. Kyne, T. V. Ovaska, Org. Lett. 2006, 8, 5153 – 5156.
[21] The relative configuration of epoxides 29a and 29b was inferred
from the assignment of pentacycle 31.
[22] The mechanism of this PMB cleavage is under active inves-
tigation and will be reported in due course. The success of this
reaction may be dependant on the presence of adventitious O2
for oxidation of the lithiated benzylic carbon of the PMB ether.
For an example of an organolithium/O2-mediated cleavage of a
PMB amide, see: R. M. Williams, T. Glinka, E. Kwast, H.
Coffman, J. K. Stille, J. Am. Chem. Soc. 1990, 112, 808 – 821.
[23] a)T. Onoda, R. Shirai, S. Iwasaki, Tetrahedron Lett. 1997, 38,
1443; b)B. M. Trost, G. B. Dong, J. A. Vance, J. Am. Chem. Soc.
2007, 129, 4540 – 4541.
Received: May 11, 2008
Published online: July 21, 2008
Keywords: alkaloids · angiogenesis · cycloisomerization ·
.
enynes · oxidative dearomatization
[24] Attempted epoxidation of the intermediate phenol led to clean
epoxide formation as judged by TLC analysis, but only decom-
position products were obtained following a standard workup.
[25] For a recent review, see: S. Quideau, L. Pouysegu, D. Deffieux,
Synlett 2008, 467 – 495.
[26] The structure of the C1 dimethyl ketal (I)is shown below.
Although this compound slowly converts into 31 upon standing
in CDCl3, it is accompanied by the formation of decomposition
products.
[1] a)J. Folkman, Y. Shing, J. Biol. Chem. 1992, 267, 10931 – 10934;
b)J. Folkman, Nat. Med. 1995, 1, 27 – 31.
[2] B. A. Teicher, S. A. Holden, G. Ara, E. Alvarez Sotomayor,
Z. D. Huang, Y. N. Chen, H. Brem, Int. J. Cancer 1994, 57, 920 –
925.
[3] T. Shih, C. Lindley, Clin. Ther. 2006, 28, 1779 – 1802.
[4] S. Aoki, Y. Watanabe, M. Sanagawa, A. Setiawan, N. Kotoku, M.
Kobayashi, J. Am. Chem. Soc. 2006, 128, 3148 – 3149.
[5] a)Y. Watanabe, S. Aoki, D. Tanabe, A. Setiawan, M. Kobayashi,
Tetrahedron 2007, 63, 4074 – 4079; b)S. Aoki, Y. Watanabe, D.
Tanabe, A. Setiawan, M. Arai, M. Kobayashi, Tetrahedron Lett.
2007, 48, 4485 – 4488.
[6] S. Aoki, Y. Watanabe, D. Tanabe, M. Arai, H. Suna, K.
Miyamoto, H. Tsujibo, K. Tsujikawa, H. Yamamoto, M.
Kobayashi, Bioorg. Med. Chem. 2007, 15, 6758 – 6762.
[7] S. De Marino, F. Zollo, M. Iorizzi, C. Debitus, Tetrahedron Lett.
1998, 39, 7611 – 7614.
[27] After online publication of our manuscript, we became aware of
a relevant manuscript: M. Dai, S. J. Danishefsky, Heterocycles
2008, DOI: COM-08-S(F)6.
[8] a)M. I. Choudhary, Atta-ur-Rahman, A. J. Freyer, M. Shamma,
Tetrahedron 1986, 42, 5747 – 5752; b)Atta-ur-Rahman, M. I.
Angew. Chem. Int. Ed. 2008, 47, 6650 –6653
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6653