A potentially quite useful route to the trimethyldecalin unit
of diterpenes, sesterterpenes, and even triterpenes would be
the direct Diels-Alder cycloaddition of a suitably function-
alized diene having the trimethylcyclohexane moiety of the
A ring with a simple or complex dienophile thereby gen-
erating the B ring in one convergent process. There are very
few examples of such a strategy being successful, partly
because of the steric hindrance involved in such a cycload-
dition, but the existence of new methods of carrying out
Diels-Alder or Diels-Alder-type reactions argues that this
route should be reexamined. The diene for the key cycload-
dition, 2-(1-tert-butyl-dimethylsilyloxyethenyl)-1,3,3-trimeth-
ylcyclohexene 2, was readily prepared from 2,6-dimethyl-
cyclohexanone 6 in several steps (Scheme 2). Methylation
Knowing that the direct [4+2] cycloaddition was very
difficult, we decided to see if a stepwise cycloaddition
process could be induced to occur, namely an initial [2+2]
cycloaddition followed by the [3,3]-sigmatropic rearrange-
ment to produce overall the same product. This is similar to
our earlier work4 where we specifically prepared the cy-
clobutane from a silyloxydiene and an allene carboxylate.
Thus, heating a neat mixture of the allene carboxylate 3
(prepared in 58% yield by reaction of acetyl chloride with
ethoxycarbonylmethylenetriphenylphosphorane)6 with the
diene 2 at 110 °C for 14 days gave a separable mixture of
three products, the [2+2] cycloadduct 5 in 9% yield, the
desired exo [4+2] cycloadduct 4x in 23.3% yield, and the
endo adduct 4n in 11.7% yield (Scheme 3). This is yet
Scheme 2
Scheme 3
gave the trimethyl ketone (92%) which was then subjected
to a Barton vinyl iodination procedure, namely formation
of the hydrazone and treatment with iodine and base to give
the vinyl iodide 7 in 73% yield for the final two steps.
Formation of the vinyllithium with tert-butyllithium followed
by trapping with acetaldehyde, TPAP/NMO oxidation of the
resulting alcohol to the acetyl group, and kinetic silyl enol
ether formation afforded the diene 2 in 88% yield from 7.
Heating a neat mixture of the diene 2 and ethyl acrylate 8
afforded none of the desired Diels-Alder adduct giving back
only the starting diene and polymeric material. This lack of
normal [4+2] cycloaddition reactivity is not at all surprising
because theoretical calculations indicate that the diene 2
exists mainly in the noncoplanar conformation 2B rather than
in the cis-planar conformation 2A required for a concerted
[4+2] cycloaddition (Figure 1). Both molecular mechanics
another example of the use of allenes in difficult cycload-
ditions.7 In addition to the three cycloadducts, 31% of the
starting diene 2 was recovered. The structures of the two
[4+2] cycloadducts were determined by NOE experiments
on the corresponding alcohols 9xn (prepared by DIBAL
reduction). We believe that the [4+2] cycloadducts 4xn are
formed in a stepwise fashion, namely via the initial formation
of 5, for the following reasons: (1) monitoring the appear-
(3) (a) Jung, M. E.; Davidov, P. Angew. Chem. 2002, 41, 4125-4128.
(b) Jung, M. E.; Ho, D.; Chu, H. V. Org. Lett. 2005, 7, 1649-1651.
(4) Jung, M. E.; Nishimura, N.; Novack, A. R. J. Am. Chem. Soc. 2005,
127, 11206-11207.
(5) MM2 calculations using Macromodel indicate a difference of about
14 kcal/mol favoring 2B (the cis planar conformation 2A was constrained
and held for these calculations). DFT calculations using B3LYP/6-31.G*//
HF/6--1.G* indicate a 9.2 kcal/mol difference favoring 2B, and those using
just HF/6-31.G* gave a 12.8 kcal/mol difference.
Figure 1. Conformations of diene 2.
(6) Paik, Y. H.; Dowd, P. J. Org. Chem. 1986, 51, 2910.
(7) For allene cycloadditions, see: (a) Jung, M. E.; Min, S.-J. J. Am.
Chem. Soc. 2005, 127, 10834-10835. (b) Jung, M. E.; Nishimura, N. J.
Am. Chem. Soc. 1999, 121, 3529-3530. (c) Kanematsu, K. ReV. Heteroatom
Chem. 1993, 9, 231-259. (d) Aso, M.; Kanematsu, K. Trends Org. Chem.
1995, 5, 157-169. (e) Gras, J. L.; Galledou, B. S.; Bertrand, M. Bull. Soc.
Chim. Fr. 1988, 4, 757-767.
and DFT methods indicate that the ground-state twisted
conformation, e.g., 2B, is 9-14 kcal/mol more stable than
the cis-coplanar conformation 2A.5
5858
Org. Lett., Vol. 8, No. 25, 2006