Furanosteroid studies. Stereoselective synthesis of the A,B,E-ring core of wortmannin

Article abstract of DOI:10.1021/ol071158s

Alkyne oxazole 11c is converted in three steps, and ～45% overall yield, to furanolactone 21α having the A,B,E-ring core of the wortmannin (2) family of furanosteroids. The TiCl₄-catalyzed insertion of EtO ₂C-CH=O between C₃

Full text of DOI:10.1021/ol071158s

ORGANIC
LETTERS
2007
Vol. 9, No. 17
3221-3224
Furanosteroid Studies. Stereoselective
Synthesis of the A,B,E-Ring Core of
Wortmannin
E. Hampton Sessions, Roger T. O’Connor, Jr., and Peter A. Jacobi*
Department of Chemistry, Dartmouth College, HanoVer, New Hampshire 03755
p.jacobi@dartmouth.edu
Received May 17, 2007
ABSTRACT
Alkyne oxazole 11c is converted in three steps, and
(2) family of furanosteroids. The TiCl4-catalyzed insertion of EtO2C
a pathway involving chemoselective lactonization of equilibrating aldol intermediates 23r,â
∼
45% overall yield, to furanolactone 21
r
having the A,B,E-ring core of the wortmannin
−CHd
O between C3 and C10 in furanoacid 14d is >98% stereoselective via
(dynamic kinetic resolution).
The furanosteroids are a class of novel pentacyclic fungal
metabolites, characterized in part by a furan ring bridging
positions 4 and 6 of the steroid skeleton.1 Representative
examples include members of the viridin (1) and wortmannin
(2) families, distinguished by an aromatic ring C in the
former and a strained lactone ring A in the latter (Figure 1).
become better known for their ability to selectively block
certain intracellular signaling pathways, in particular those
associated with cell growth and development.2c,d A better
understanding of such signal disruption might lead to new
therapeutic agents for diseases characterized by rapid cell
proliferation.
The growth inhibitory properties of 1 and 2 stem partly
from their activity as irreversible inhibitors of phospho-
inositide 3-kinase (PI3K),1 a class of enzymes that plays a
key role in important cell signaling processes.2c,d Also, in
an interesting recent development, 2 has been shown to be
a potent inhibitor of the mammalian Polo-like kinase PLK1,
an enzyme vital to cellular growth cycles that offers a new
target in cancer therapy.2e Early on it was postulated that
inhibition occurs by nucleophilic addition of the kinase to
the electron defficient C20-position in 1 and 2,3a,b a hypothesis
that was subsequently confirmed by crystallographic studies
Figure 1. Viridin (1) and wortmannin (2) furanosteroids.
(2) (a) Brian, P. W.; McGowan, J. C. Nature (London) 1945, 156, 144.
(b) MacMillan, J.; Vanstone, A. E.; Yeboah, S. K. Chem. Commun. 1968,
613. (c) Powis, G.; Bonjouklian, R.; Berggren, M. M.; Gallegos, A.;
Abraham, R.; Ashendel, C.; Zalkow, L.; Matter, W. F.; Dodge, J. Cancer
Res. 1994, 54, 2419. (d) Ward, S.; Sotsios, Y.; Dowden, J.; Bruce, I.; Finan,
F. Chem. Biol. 2003, 10, 207, and references cited therein; see also ref 1b.
(e) Liu, Y.; Shreder, K. R.; Gai, W.; Corral, S.; Ferris, D. K.; Rosenblum,
J. S. Chem. Biol. 2005, 12, 5256. Wortmannin crystal structures: (f) Petcher,
J.; Weber, H.-P.; Kis, Z. Chem. Commun. 1972, 1061. (g) Doi, M.; Ishida,
T.; Terashima, A.; Taniguchi, T.; Sasaki, M.; Tanaka, C. Anal. Sci. 1998,
14, 1191.
Viridin (1) was the first member of this class to be isolated
and characterized (1945),2a attracting considerable attention
for its potent anti-inflammatory and antibiotic activity.
Similar properties were noted for 2 following its structural
elucidation in 1968.2b However, in recent years, 1 and 2 have
(1) For reviews, see: (a) Hanson, J. R. Nat. Prod. Rep. 1995, 12, 381.
(b) Wipf, P.; Halter, J. H. Org. Biomol. Chem. 2005, 3, 2053.
10.1021/ol071158s CCC: $37.00
© 2007 American Chemical Society
Published on Web 07/26/2007

on PI3K-bound 2.3c Surprisingly, though, given the impor-
tance of these compounds, synthetic methodology in this area
has been relatively slow to develop.4 Notable achievements
include a very elegant total synthesis of 1 by Sorensen et
al.4a and both formal and total syntheses of 2 by Shibasaki
et al.4b-d
Recently, we described a new synthetic approach to the
viridin (1) class of furanosteroids, involving bond discon-
nection at C1-C10 (cf. 3 f 4 in Scheme 1).5 The pivital
insertion of methoxyacetaldehyde or congenors between the
C3-and C10-positions in furanoacid derivatives of type 7
(Scheme 2).7 However, while in our viridin model studies
Scheme 2. Wortmannin (2) Synthetic Analysis
Scheme 1. Viridin (1) Synthetic Analysis
step in this strategy requires dearomatization of ring B in 3,
raising the possibility that the reverse process 4 f 3 might
be energetically more favorable. However, a search of the
literature revealed that simple retro-aldol products of type 3
are not known in viridin chemistry. Nor, it appears, are
naturally occurring C1-epimers of 1,1-3 although aldol-like
equilibration would provide a straightforward pathway for
isomerization. We took these observations as evidence of
the thermodynamic stability of the C1-C10 bond in 1, the
formation of which relieves strong peri-interactions between
the coplanar C4-, C10-, and C11-substituents in open chain
species of type 3.6 Model studies lent support to this premise.
Thus, on TiCl4-catalyzed ring closure, aldehyde 5 was rapidly
transformed to a 4:1 mixture of aldol products 6-syn/anti in
75% overall yield.5 Under no circumstances could we detect
equilibration between 6-syn and 6-anti.
the correct stereochemistry at C1-C10 is kinetically favored
by a lower energy Burgi-Dunitz trajectory angle,5,8 no such
bias is likely in the intermolecular conversion of 7 to 2.
Conversely, an argument could be made that the desired
anti-stereochemistry at C1-C10 would predominate under
thermodynamic control, notwithstanding the fact that this
geometry corresponds to a 1,2-diaxial orientation (Scheme
2). Taking 11-desacetoxywortmannin (8) as an example,
models clearly show that the “unnatural” syn-isomer 1-epi-8
suffers from steric crowding of two types not found in 8.
One of these is an additional gauche interaction between the
C1-methoxymethyl and C10-methyl groups (curved arrow),
while the other corresponds to a strong boat “1,4-flagpole”
interaction between the C1-methoxymethyl and C11-H groups
(best seen in the stereoviews derived from MM2 minimiza-
tion). Calculations at the AM1 level of computation re-
inforced this analysis, with a 3.6 kcal difference in heat of
formation between 8 and 1-epi-8.9
In principle, a similar strategy might be applied to the
synthesis of wortmannin (2) and analogues, involving formal
(3) (a) Wymann, M. P.; Bulgarelli-Leva, G.; Zvelebil, M. J.; Pirola, L.;
Vanhaesebroeck, B.; Waterfield, M. D.; Panatotou, G. Mol. Cell Biol. 1996,
16, 1722. (b) Norman, B. H.; Shih, C.; Toth, J. E.; Ray, J. E.; Dodge, J. A.;
Johnson, D. W.; Rutherford, P. G.; Schultz, R. M.; Worzalla, J. F.; Vlahos,
C. J. J. Med. Chem. 1996, 39, 1106. (c) Walker, E. H.; Pacoid, M. E.;
Perisic, O.; Stephens, L.; Hawkins, P. T.; Wymann, M. P.; Williams, R. L.
Mol. Cell 2000, 6, 909.
(4) (a) Anderson, E. A.; Alexanian, E. J.; Sorensen, E. J. Angew. Chem.,
Int. Ed. 2004, 43, 1998. (b) Honzawa, S.; Mizutani, T.; Shibasaki, M.
Tetrahedron Lett. 1999, 40, 311. (c) Mizutani, T.; Honzawa, S.; Tosaki,
S.-y.; Shibasaki, M. Angew. Chem., Int. Ed. 2002, 41, 4680. (d) Shigehisa,
H.; Mizutani, T.; Tosaki, S.-y.; Ohshima, T.; Shibasaki, M. Tetrahedron
2005, 61, 5057.
To test this hypothesis, we prepared the model furanoacid
derivatives 14a-d, making use of the alkyne oxazole
(7) The reverse of this process is known for both 2 and 8 but requires
refluxing in 2 N HCl; cf. (a) MacMillan, J.; Simpson, T. J.; Vanstone, A.
E.; Yeboah, S. K. J. Chem. Soc., Perkin Trans. 1 1972, 2892, 2898. (b)
Haefliger, W.; Hauser, D. HelV. Chim. Acta 1973, 56, 2901.
(8) Burgi, H. B.; Dunitz, J. D. Acc. Chem. Res. 1983, 16, 153.
(9) Spartan ’02 v1.0.4e (Wavefunction, Inc., Irvine, CA). We are grateful
to Professor Dennis Wright (UConn) for carrying out these calculations.
(5) Sessions, E. H.; Jacobi, P. A. Org. Lett. 2006, 8, 4125.
(6) For related examples, see: (a) Garduno-Ramirez, M. L.; Trejo, A.;
Navarro, V.; Bye, R.; Linares, E.; Delgrado, G. J. Nat. Prod. 2001, 64,
432. (b) Burgueno-Tapia, E.; Bucio, M. A.; Rivera, A.; Joseph-Nathan, P.
J. Nat. Prod. 2001, 64, 518.
3222
Org. Lett., Vol. 9, No. 17, 2007

methodology we have employed in the synthesis of numerous
naturally occurring furans, butenolides, and lactones (Scheme
3).10 The requisite Diels-Alder substrates 11a-c were
we were pleased to find that TiCl4-catalyzed condensation
of 14a with (CH2O)n gave a 70% yield of the vinylogous
aldol product 15a, which was reasonably stable to retro-
aldol cleavage. Also, we found no evidence for attack at other
electron-rich positions. Unexpectedly, though, lactonization
of 15a proved to be problematic, likely because of steric
crowding in the tetrahedral intermediate leading to 16, as
well as the inherent strain of the furanoisochromene product.1b
Thus, 15a was recovered unchanged at temperatures up to
110 °C and suffered decomposition under strongly acidic or
basic conditions (method A).
Scheme 3. Synthesis of the A,B,E-Ring Core of 2 and 8
Better results were obtained employing the tert-butyl ester
14b, obtained in 67% yield upon brief heating of alkyne
oxazole 11b in o-xylene followed by silylation (method B).
On reaction with (CH2O)n/TiCl4/CH2Cl2, 14b underwent
hydroxymethylation with concomitant ester hydrolysis, af-
fording the corresponding acid derivative 15d in one step.
Due to its instability, purification of 15d at this stage was
not possible. However, treatment of the crude reaction
mixture with POCl3/pyridine at rt gave a ∼25% overall yield
of the desired wortmannin model substrate 16.
Finally, a more efficient synthesis of 15d and 16 made
use of the benzyl ester 14c, obtained in 75-80% yield by
thermolysis of alkyne oxazole 11c. Upon condensation with
(CH2O)n/TiCl4, 14c afforded a 67% yield of the readily
purified hydroxymethyl derivative 15c (R ) Bn), which on
hydrogenolysis gave an 82% yield of acid 15d in a high state
of purity. Treatment of 15d thus prepared with POCl3/
pyridine gave lactone 16 in an improved yield of ∼50%
(method C). Alternatively, the reagent system triisopropyl-
benzenesulfonyl chloride (TIPBSCl)/DMAP in pyridine
afforded 54% of 16 (method D). No further efforts were
made to optimize this transformation.
prepared by carboalkoxylation of the terminal alkyne oxazole
10,11 itself derived in a very efficient six step sequence
beginning with the simple BOC derivative 9.5,12 For illustra-
tion, alkyne oxazole 11a (R ) Et) was prepared in 70% yield
by carboethoxylation of the parent alkyne 10 and gave a 50%
yield of the TBS-protected phenol 14a on thermolysis in
o-xylene followed by in situ silylation. We were now in
position to investigate the vinylogous Mukaiyama-like aldol
condensation initiating ring A formation.13 Ideally, this
transformation could be effected with both regio- and
stereochemical control, each of which was addressed sepa-
rately.
Our initial investigations were carried out with paraform-
aldehyde, (CH2O)n, in order to probe regiochemical control
in the absence of stereochemical complications (14 f 15,
Scheme 3). As in our earlier studies,5 we expected that de-
aromatization of ring B would not present a significant
thermodynamic barrier. However, the intermolecular nature
of this reaction introduced additional ambiguity. In the event,
We focused on two means for achieving stereochemical
control at C1-C10. The most straightforward of these
involved introduction of an epimerizable group at C1, which
on thermodynamic equilibration should greatly favor the
desired anti-isomer 17-anti (method I, Scheme 4). To explore
this route, we carried out the TiCl4-catalyzed condensation
of phenol derivative 14c with ethyl glyoxylate (18), which
afforded a ∼1:1 mixture of vinylogous aldol products 19R,â
as an inseparable mixture. Unlike the case with the simple
unsubstituted adducts 15, these materials were very suscep-
tible to retro-aldol reaction and had to be carried forward to
lactones 21R,â without purification. This was accomplished
in identical fashion to that described in Scheme 3 for 15c,
involving hydrogenolysis to acids 20R,â and lactonization
with POCl3 (29% overall yield from 14c). In each step, the
ratio of diastereomeric products remained ∼1:1, indicating
that no further equilibration had occurred. Judging from the
overall yield, though, it was apparent that significant material
had been lost to retro-adol cleavage in one or more steps.
Also disappointing, we were unable to effect equilibration
of 21R,â without significant decomposition. A contributing
factor in this difficulty may be the extremely hindered
environment of the C1-H. On a more positive note, though,
we did uncover an important clue for further investigation
(vide infra).
(10) For leading references, see: Jacobi, P. A.; Lee, K. J. Am. Chem.
Soc. 2000, 122, 4295. See also ref 5.
(11) (a) Vedejs, E.; Piotrowski, D. W. J. Org. Chem. 1993, 58 (6), 1341.
(b) Izawa, Y.; Shimizu, I.; Yamamato, A. Bull. Chem. Soc. Jpn. 2004, 77,
2033.
(12) Van De Water, R. W.; Magdziak, D. J.; Chau, J. N.; Pettus, T. R.
R. J. Am. Chem. Soc. 2000, 122 (27), 6502.
(13) For a recent review of this topic, see: Denmark, S. E.; Heemstra,
J. R.; Beutner, G. L. Angew. Chem. 2005, 117, 4760.
Org. Lett., Vol. 9, No. 17, 2007
3223

Scheme 4. Incorporation of an Epimerizable Group at C1
Scheme 5. (Top) Chemoselective Formation of Lactone 21R;
(Bottom) Stereoselective Introduction of the C1-Substituent
During our attempts at ring closure of 20R,â, it was nearly
always the case that lactones 21R,â were produced in
essentially identical ratios to the precursor alcohol-acids
20R,â (NMR). However, one divergent result caught our
attention. On treatment of a 1:1 mixture of 20R,â with
(COCl)2/CH2Cl2 in the absence of base, alcohol-acid 20R
underwent chemoselective lactonization to give a modest
yield of lactone 21R as the only identifiable product of ring
closure (Scheme 5, top). The structure of 21R was initially
established by NOE studies, which showed a strong cor-
relation between the C1-H and C11-H as well as the C1-H
and C10-Me group (curved arrows). Subsequent X-ray
analysis confirmed this assignment, the solid-state geometry
of 21R closely approximating that predicted computationally
for desacetoxywortmannin (8) (cf. Scheme 2).14,15 No
evidence could be found for lactonization of 20â, whose fate
remains unknown (a number of decomposition products
indicated that retro-aldol cleavage had occurred). Control
experiments ruled out the possibility of equilibration of 21â
to 21R under the reaction conditions.
by chemoselective lactonization (method II; dynamic kinetic
resolution). Gratifyingly, this was accomplished employing
a straightforward modification of our original route (Scheme
5, bottom). Thus, a solution of furanoacid 14d in CH2Cl2
was treated sequentially with (COCl)2 (14d f 22), followed
by in situ aldol condensation employing a slight excess of
ethyl glyoxylate (18)/TiCl4. On stirring at rt one could
observe by TLC the very clean formation of lactone 21R,
which was isolated in 60% yield as a colorless crystalline
solid (45% overall from alkyne oxazole 11c). Within the
limits of NMR and TLC detection we found no evidence
for formation of the epimeric lactone 21â.
The level of efficiency in the transformation of 11c to 21R
is noteworthy, introducing in three steps what are arguably
the most challenging structural features found in 2 (“a
bisallylic quarternary carbon center and a highly reactive
furanocyclohexadienone lactone unit”).4d The application of
this strategy to the synthesis of wortmannin (2) and other
naturally occurring furanosteroids is under investigation.
An appealing explanation for this observation is that the
thermodynamic instability of 21â is reflected in the transition
state leading from 20â to 21â (cf. Scheme 2). This would
explain why 20â suffered intervening decomposition instead
of ring closure. Based on this hypothesis, we sought to
identify conditions wherein rapid aldol-retro-aldol equilibra-
tion of 20R,â or a related intermediate might be followed
Acknowledgment. E.H.S. (Dartmouth College) was the
recipient of a National Science Foundation Predoctoral
Fellowship during portions of this work. We are grateful to
the National Science Foundation, CHE-0646876, for their
recent support of these studies.
(14) We are grateful to Mr. Benjamin E. Kucera and Dr. Victor G. Young,
Jr., of the X-ray Crystallographic Laboratory of the University of Minnesota,
for performing this analysis and for drawing our attention to the unusual
crystal packing geometry found in racemic 21R.15
(15) Racemic 21R crystallized in a centrosymmetric space group
noteworthy for an unusually short intermolecular contact distance between
the furan oxygen atoms related by an inversion center (2.64 Å, in the same
range as a medium-strength hydrogen bond14). This appears to be the
shortest such furan O-O contact distance reported in the CSD data base
(the corresponding distance in monoclinic C2 symmetric wortmannin (2)
is 2.99 Å1g). The nature of this apparent donor-acceptor interaction will
be described separately.
Supporting Information Available: Experimental and
NMR spectra for all new compounds. X-ray crystal structure
for 21R. This material is available free of charge via the
Internet at http://pubs.acs.org.
OL071158S
3224
Org. Lett., Vol. 9, No. 17, 2007