Total synthesis of (+/-)-epoxysorbicillinol.

Full text of DOI:10.1021/ja0057979

J. Am. Chem. Soc. 2001, 123, 2097-2098
2097
Scheme 1
Total Synthesis of (()-Epoxysorbicillinol
John L. Wood,* Brian D. Thompson, Naeem Yusuff,
Derek A. Pflum, and Mike S. P. Mattha¨us
Sterling Chemistry Laboratory, Department of Chemistry
Yale UniVersity, New HaVen, Connecticut 06520-8107
ReceiVed NoVember 16, 2000
Recent interest in species of the genus Trichoderma as sources
of unique metabolites has led to the isolation of several vertinoid
polyketides. Members of this natural products class have shown
interesting biological activity, including inhibiting TNF-R produc-
tion and DPPH radical scavenging activity.1 Epoxysorbicillinol
(1) represents the first isolated vertinoid polyketide possessing
an epoxide functionality.2 Its densely functionalized chemical
structure and potential for interesting biological activity led us
to undertake a total synthesis of 1.3 Herein we report the details
of this investigation.
Scheme 2
From a retrosynthetic perspective, the known propensity of the
sorbyl side chain (1: C(7-12)) to undergo polymerization led
us to target diene 2 as an advanced intermediate (Scheme 1).4
However, as will be illustrated, it was discovered that such dienes
are exceedingly prone to aromatize via a facile 1,2-methyl shift
(e.g., 2 f 3). As an added challenge, attempts to avoid this
disastrous thermodynamic pitfall via protection of the C(4)
hydroxyl of 2 produced substrates that were recalcitrant toward
epoxidation. These difficulties dictated the development of a
modified approach wherein allylic alcohol 4 served as a key
intermediate. Importantly, quaternization of C(2) in 4 prevented
aromatization and allowed the C(4) hydroxyl to remain free to
direct epoxidation. Fortunately, both approaches (i.e., 4 f 1 and
2 f 1) could be divergently accessed from the versatile common
intermediate 5, where the C(4) tertiary alcohol is masked
intramolecularly as a cyclic acetal. Oxabicycle 5 was envisioned
to arise from a rhodium-catalyzed 1,3-dipolar cycloaddition
between R-diazo ketone 6 and a propiolate ester. Finally, 6 could
be accessed from any number of commercially available meth-
ylmalonates.
In the forward sense, we first explored the critical metal-
catalyzed carbonyl ylide cycloaddition using 8 as a model
substrate (Scheme 2). To this end, deprotonation of dimethyl
methylmalonate (7) with NaH followed by the addition of
phosgene at -78 °C gave the corresponding acid chloride which,
without purification, was added to an ice-cooled solution of
diazoethane in ether to provide R-diazo ketone 8.5 Gratifyingly,
exposure of 8 to catalytic Rh2(pfb)4 (1 mol %) generated
intermediate carbonyl ylide 10, which underwent smooth 1,3-
dipolar cycloaddition with methyl propiolate (9) to afford 11 in
Scheme 3
excellent yield.6 Interestingly, this reaction provided only a single
diastereomer of 11, the structure of which was confirmed via
X-ray crystallography.
Having established a sound method for assembly of the
carbocyclic core, we turned our attention toward the synthesis of
a specific bicyclic substrate that would allow us efficient access
to diene 16. After a challenging search for the proper ester
protecting group, 15 was found to be a rather versatile intermedi-
ate.7 Although the preparation of 15 was practical on a small scale
using the diazotization procedure shown in Scheme 2, a more
efficient synthesis which did not require the use of phosgene or
diazoethane was desired for multigram scale. To this end, diethyl
methylmalonate (12) was transesterified with 2 equiv of â-(tri-
methylsilyl)ethanol (Scheme 3). Deprotonation of the resulting
diester with NaH followed by the addition of pyruvoyl chloride
at -78 °C gave diketone 13.8 Although 13 was unstable to silica
gel chromatography, treatment of the crude solution with 1 equiv
of TsNHNH2 followed by basic alumina gave regioselective
diazotization of the desired ketone to produce R-diazo ketone 14
in excellent overall yield.9 Optimal conditions for the cycload-
dition between 14 and 9 were found using Rh2(OAc)4 in benzene
(1) (a) Warr, G. A.; Veitch, J. A.; Walsh, A. W.; Hesler, G. A.; Pirnik, D.
M.; Leet, J. E.; Lin, P.-F. M.; Medina, I. A.; McBrien, K. D.; Forenza, S.;
Clark, J. M.; Lam, K. S. J. Antibiot. 1996, 49, 234. (b) Abe, N.; Murata, T.;
Hirota, A. Biosci., Biotechnol., Biochem. 1998, 62, 661. (c) Abe, N.; Murata,
T.; Hirota, A. Biosci., Biotechnol., Biochem. 1998, 62, 2120.
(2) Sperry, S.; Samuels, G. J.; Crews, P. J. Org. Chem. 1998, 63, 10011.
(3) For syntheses of related natural products, see: (a) Barnes-Seeman, D.;
Corey, E. J. Org. Lett. 1999, 1, 1503. (b) Nicolaou, K. C.; Vassilikogiannakis,
G.; Simonsen, K. B.; Baran, P. S.; Zhong, Y.-L.; Vidali, V. P.; Pitsinos, E.
N.; Couladouros, E. A. J. Am. Chem. Soc. 2000, 122, 3071. (c) Nicolaou, K.
C.; Simonsen, K. B.; Vassilikogiannakis, G.; Baran, P. S.; Vidali, V. P.;
Pitsinos, E. N.; Couladouros, E. A. Angew. Chem., Int. Ed. 1999, 38, 3555.
(d) Nicolaou, K. C.; Jautelat, R.; Vassilikogiannakis, G.; Baran, P. S.;
Simonsen, K. B. Chem. Eur. J. 1999, 5, 3651. (e) Abe, N.; Sugimoto, O.;
Tanji, K.; Hirota, A. J. Am. Chem. Soc. 2000, 122, 12606.
(6) For recent reviews of various types of carbonyl ylide cycloadditions,
see: (a) Padwa, A.; Weingarten, M. D. Chem. ReV. 1996, 96, 223. (b) Padwa,
A. Acc. Chem. Res. 1991, 24, 22.
(7) The syntheses of several oxabicycles and their deprotection chemistry
was explored and will be reported at a later date.
(8) For a preparation of pyruvoyl chloride, see: Ottenheijm, H. C. J.;
Tijhuis, M. W. Org. Synth. 1983, 61, 1.
(4) Andrade, R.; Ayer, W. A.; Mebe, P. P. Can. J. Chem. 1992, 70, 2526.
(5) A diazoethane solution was prepared by the following method: Short,
R. P.; Revol, J.-M.; Ranu, B. C.; Hudlicky, T. J. Org. Chem. 1983, 48, 4453.
(9) This modified diazotization procedure is much more efficient than the
previously described phosgene/diazoethane route and is amenable to multigram
scale.
10.1021/ja0057979 CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/10/2001

2098 J. Am. Chem. Soc., Vol. 123, No. 9, 2001
Communications to the Editor
Scheme 4
Scheme 6
addition of 1 equiv of 3-pentenyllithium (24) produced ketone
25, a compound possessing the complete carbon framework of
epoxysorbicillinol.13,14 The incorporation of 24 had the benefit
of delaying introduction of the R,â-unsaturation into the side
chain, thus preventing polymerization problems and allowing the
selective epoxidation of the C(5,6) double bond.15 Exposure of
25 to 1.5 equiv of TFA opened the cyclic acetal to furnish an
alcohol which, upon treatment with t-BuOOH and catalytic DBU,
underwent smooth conversion to a single diastereomer of epoxy
alcohol 26.
Scheme 5
At this point only decarboalkoxylation and oxidation of the
side chain were required to complete the synthesis.16 To this end,
treatment of 26 with DDQ effected the desired dehydrogenation
to yield 27 with the completed side chain in place (Scheme 6).
Deprotection of (trimethylsilyl)ethyl ester 27 by treatment with
excess TFA afforded the natural product (1), albeit in very poor
yield by crude NMR. Fortunately, reversing the order of the final
two steps proved to be a much better route to 1. Treatment of 26
with excess TFA provided enol 28 in excellent yield. The presence
of the epoxide appeared to prevent aromatization even under these
highly acidic conditions. Finally, treatment of 28 with DDQ
furnished 1 in 30-40% yield.
In summary, we have completed the first total synthesis of
(()-epoxysorbicillinol (1) in 13 steps, starting from commercially
available diethyl methylmalonate (12). A key step in this synthesis
utilizes a novel 1,3-dipolar cycloaddition between an R-diazo
ketone and a propiolate ester. During the course of this work, a
synthetic route was developed that simultaneously avoided a
disastrous thermodynamically favored aromatization, allowed a
fastidious epoxidation to occur, and prevented polymerization of
the side chain. Efforts toward the completion of an asymmetric
synthesis of 1 using an appropriate enantioenriched rhodium
catalyst for the cycloaddition are currently underway.
at 60 °C, thus furnishing the diastereomerically pure oxabicycle
15 in excellent yield.
With multigram quantities of 15 readily available, a series of
studies investigating the removal of both (trimethylsilyl)ethyl
groups to produce diene 16 was begun (Scheme 4). Although
attempts with fluoride failed, it was eventually discovered that
exposure of 15 to triflic acid effected complete deprotection;
however, the desired diene (16) was unstable, undergoing
aromatization via a 1,2-methyl shift to give 17 as the only isolable
product.10,11 In an effort to more delicately manipulate 15, it was
found that the cyclic acetal could selectively be opened using
1.5 equiv of TFA to provide allylic alcohol 18, a manipulable
intermediate wherein the quaternary center at C(2) prevented
aromatization. To maintain protection against aromatization, the
C(4) tertiary alcohol was masked prior to removal of the C(2)
carboalkoxy group. Although this could be performed in a
standard two-step protection/deprotection sequence (18 f 21),
it was discovered that both manipulations could be performed in
a single step by means of a base-promoted acyl migration that
furnished carbonate 19.
Unfortunately, numerous attempts to epoxidize 19, 20, and 21
failed, and it soon became clear that a successful epoxidation
would require a free C(4) alcohol.12
In an effort to address the remaining epoxidation issue and
facilitate installation of the sorbyl side chain, a revised route was
developed (Scheme 5) wherein allyl propiolate (22) was used as
a dipolarophile in the cycloaddition with 14 to furnish oxabicycle
23. Facile deprotection of allyl ester 23 with catalytic Pd(PPh3)4
and piperidine, coupling to Weinreb’s amine, followed by the
Acknowledgment. We acknowledge the support of this work by
Bristol-Myers Squibb, Eli Lilly, Glaxo-Wellcome, Yamanouchi, and
Zeneca through their Faculty Awards Programs and the Camille and Henry
Dreyfus Foundation for a Teacher-Scholar Award. N.Y. acknowledges
the NIH for a postdoctoral fellowship, and M.S.P.M. acknowledges the
Hermann-Schlosser Foundation for a postdoctoral fellowship. In addition,
we thank Susan DeGala for assistance in securing the X-ray analysis.
Supporting Information Available: Spectral and experimental data
pertaining to all new illustrated compounds and isolable intermediates
generated en route but not illustrated; X-ray data for 11 (PDF). This
material is available free of charge via the Internet at http://pubs.acs.org.
JA0057979
(10) Specifically, tetrabutylammonium fluoride and HF/pyridine failed to
generate 16.
(11) Elix, J. A.; Wardlaw, J. H. Aust. J. Chem. 1996, 49, 539. Spectroscopic
verification of 17 was kindly provided by Professor Elix via personal
communication.
(12) A separate series of experiments (see Supporting Information) showed
that substrates such as 18 with the C(4) alcohol unprotected and the C(2)
quaternary center in place were easily epoxidized to provide diastereomerically
pure epoxides.
(13) For a preparation of 1-iodo-3-pentene from R-methylcyclopropanemeth-
anol, see: Hrubiec, R. T.; Smith, M. B. J. Chem. Soc., Perkin Trans. 1 1984,
109.
(14) Bailey, W. F.; Punzalan, E. R. J. Org. Chem. 1990, 55, 5404.
(15) Products derived from the addition of 24 to the ketone or ester were
not found. This observed selectivity for the amide is likely due to steric factors.
(16) Attempts to isomerize an alkynyl group to produce the sorbyl side
chain were unsuccessful.