Synthesis of (+)-Miyakolide
J. Am. Chem. Soc., Vol. 121, No. 29, 1999 6817
Figure 1. Proposed miyakolide biosynthetic precursors 2a and 2b. The 2b-Model structure, less the metal ion M, minimized using the AMBER
forcefield.13 C1-C9 and C21-C27 not shown.
rolactonization indeed precedes the indicated intramolecular
aldol construction, macrocyclic precursors such as 2 (Figure 1)
could well be found along the biosynthetic pathway. If this
reaction is to be integrated into a synthesis plan, the desired
aldol adduct constitutes one of four possible product diastereo-
mers, and while this process might be enzymatically mediated,
it could also be simply controlled by the conformation of the
macrocycle.3ef,12,13 In implementing this strategy, it was felt that
macrocycle 2b might provide more conformational ordering in
the aldol step than its ring-chain tautomer 2a. To assess the
probability that the desired aldol macrocyclic stereocontrol might
be possible in 2b, a multiconformational search of the crucial
enol ketone intermediate was undertaken using the AMBER
force field, restricting the C18-C13 atom distance to a maximum
of 4.5 Å.14 Figure 1 depicts the lowest energy structure generated
by this search, wherein the C18 and C13 diastereofaces are
disposed to deliver the desired stereochemistry following an
intramolecular aldol reaction. While a generic metal ion, M,
has been incorporated into the 2b-Model illustration, this does
not imply that the metal ion was part of the calculation. In this
conformation, the chair transition state for the aldol addition is
accessible. The other low-energy conformation of 2b, differing
by only 0.1 kcal/mol, presents the C17-C19 (si) enol diastereo-
face opposite to that of the C13 carbonyl moiety; however, the
resulting aldol reaction must proceed via a boat transition state.
Since a spontaneous transannular aldol addition was antici-
pated when the three carbonyl groups at C13, C17, and C19 were
revealed, we felt it was important to also have the C11 alcohol
in its unprotected state prior to this bond construction. The aldol
adduct would thus undergo immediate hemiketalization, masking
the C17-C19 diketone moiety and suppressing elimination of
the C13 hydroxyl moiety. We then elected to mask the C17-C19
diketone as its derived isoxazole,15 which might undergo
spontaneous ring closure upon reduction of the N-O bond.16
While two possible isoxazole structures were entertained, we
chose to employ isoxazole 3 bearing nitrogen at C19 since the
reduction product of 3 might be easily hydrolyzed with
assistance of the C11 alcohol following the aldol reaction
(Scheme 1).17 In the event that the enaminone failed to
participate in oxydecalin formation, this functionality might be
hydrolyzed under conditions likely to effect the transannular
aldol step.18,19
(11) For recent reviews on the biosynthesis of polyketides, see: (a)
Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.; Leadlay, P. F.
Nature 1990, 348, 176-178. (b) Donadio, S.; Staver, M. J.; McAlpine, J.
B.; Swanson, S. J.; Katz, L. Science 1991, 252, 675-679. (c) Malpartida,
F.; Hopwood, D. A. Nature 1984, 309, 462-464. (d) O’Hagan, D. Nat.
Prod. Rep. 1995, 1-33.
(12) Transannular reactions have been postulated in a number of
biosynthetic pathways. The dolabellanes are postulated to be biosynthetically
converted to the clavularanes and dolastanes via transannular ring-
contracting reactions: (a) Look, S. A.; Fenical, W. J. Org. Chem. 1982,
47, 4129-4134. Dactylol is postulated to be biosynthesized from humulene,
via a ring-contracting cationic olefin cyclization followed by cyclopropyl
cation rearrangement and solvolysis: (b) Schmitz, F. J.; Hollenbeak, K.
H.; Vanderah, D. J. Tetrahedron 1978, 34, 2719-2722. (c) Hayasaka, K.;
Ohtsuka, T.; Shirahama, H. Tetrahedron Lett. 1985, 26, 873-876. The
endiandric acids are postulated to be biosynthesized via a cascade of
electrocyclic reactions, including a ring-contracting cyclization: (d) Ban-
daranayake, W. M.; Banfield, J. E.; Black, D. St. C. J. Chem. Soc., Chem.
Commun. 1980, 902-903.
(13) Macrocyclic conformation has been employed as a control element
in synthesis in several instances: (a) Still, W. C.; Romero, A. G. J. Am.
Chem. Soc. 1986, 108, 2105-2106. (b) Schreiber, S. L.; Sammakia, T.;
Hulin, B.; Schulte, G. J. Am. Chem. Soc. 1986, 108, 2106-2108. (c) Vedejs,
E.; Gapinski, D. M. J. Am. Chem. Soc. 1983, 105, 5058-5061. Macrocyclic
ring contractions have been used with success to control diastereoselectivity
of the contracted ring-forming reaction. (d) Myers, A. G.; Condroski, K.
R. J. Am. Chem. Soc. 1993, 115, 7926-7927.
(15) For a review on the use of isoxazoles in synthesis, see: (a) Baraldi,
P. G.; Barco, A.; Benetti, S.; Pollini, G. P.; Simoni, D. Synthesis 1987,
857-869. (b) Little, R. D. In ComprehensiVe Organic Synthesis; Trost, B.
M., Fleming, I., Eds.; Pergamon Press: New York, 1991; Vol. 5 pp 239-
270. (c) Torssell, K. B. G. Nitrile Oxides, Nitrones, and Nitronates in
Organic Synthesis; VCH Publishers: New York, 1988. (d) Caramella, P.;
Grunanger, P. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A., Ed.;
John Wiley & Sons: New York, 1984; Vol. 1, pp 291-392.
(16) Although deprotonation of an enaminone (NaOH) was required to
promote an aldol reaction (Yuste, F.; Sanchez-Obregon, R. J. Org. Chem.
1982, 47, 3665-3668), we hoped that the intramolecularity of our
transformation would force the reacting partners together and facilitate a
reaction under milder conditions.
(17) The regioisomeric isoxazole containing nitrogen at C17 was deemed
an inferior intermediate, as the aldol adduct would contain a C17 imine that
could tautomerize, epimerizing the C16 stereocenter.
(18) (a) Kato, N.; Hamada, Y.; Shioiri, T. Chem. Pharm. Bull. 1984,
32, 1679-1682. (b) Eiden, F.; Patzelt, G. Arch. Pharm. (Weinheim, Ger.)
1986, 319, 242-251. (c) Auricchio, S.; Ricca, A.; DePava, O. V. Gazz.
Chim. Ital. 1980, 110, 567-570. (d) Kobuke, Y.; Kokubo, K.; Munakata,
M. J. Am. Chem. Soc. 1995, 117, 12751-12758. (e) Kashima, C.; Mukai,
N.; Tsuda, Y. Chem. Lett. 1973, 539-540.
(19) Mineral acids are most commonly employed to achieve this
transformation (see ref 18), but model studies on 3-amino-5-oxo-1-
phenyloct-3-ene demonstrated that this hydrolysis could be achieved under
milder conditions such as 4:4:1 AcOH/THF/water; PPTS in THF/water; or
CuX2 (X ) Cl, OTf, BF4) and water in a variety of organic solvents.
(14) All calculations were performed using the AMBER force field on
structures generated by a Monte Carlo multiconformer search using
MacroModel (Version 5.0) provided by Professor W. Clark Still, Columbia
University. The dielectric coefficient (ELE) in the force field was set at 60
to simulate a polar solvent. Only structures with a C18 to C13 atom distance
of 4.5 Å or less that were generated three or more times during the search
(out of 150, 000 structures generated) were considered. The AMBER force
field was selected because it generated a minimized structure of miyakolide
that more closely fit the X-ray crystal structure than structures generated
using the MM2 and MM3 force fields.