stereoisomers of the 3-methyl-4-silyloxy-2-alkanones. The
non-aldol aldol reaction of the epoxy silyl ether 7a afforded
exclusively the syn aldol product 8a with clean inversion of
stereochemistry at the tertiary epoxide atom (Scheme 2). This
product 15 and the methyl migration product 16 (Scheme
4). The major product 14 is formed via the intermediate C
Scheme 4 a
Scheme 2
a Reaction conditions: (a) TBSOTf, iPr2NEt, CH2Cl2, 4 Å MS,
-40 °C; 14:15:16 ) 1:0.8:0.25.
(Figure 1), which suffers from some nonbonded interaction,
e.g., the methyl-methyl interaction and the steric crowding
of the two large groups oriented cis on the epoxide.
Consequently, considerable amounts of the elimination
product 15 and the product of the methyl migration 16 are
obtained. When the diastereomeric epoxy silyl ether 13a was
subjected to the rearrangement (Scheme 5), the expected anti
reaction presumably proceeds via anti hydride migration in
the silylated epoxonium ion B (Figure 1). Rearrangement
of the enantiomer 7b also gave the syn aldol product 8b.4
The epoxides derived from the (Z)-alkenes were then
prepared starting from 3-hydroxy-2-butanone 9a (Scheme
3). A (Z)-selective Wittig reaction5 of the TBS ether 9b gave
Scheme 5 a
Scheme 3 a
a Reaction conditions: (a) TBSOTf, iPr2NEt, CH2Cl2, 4 Å MS,
-40 °C; 12b:17:16 ) 1:0.1:trace.
product 14 did not result but rather the syn aldol product
8b, with high selectivity. This is a very unusual result and
requires a new mechanistic picture to account for the
retention (and not inversion) of stereochemistry at the tertiary
epoxide carbon. We propose that upon activation of the
epoxide 13a with TBSOTf to give the intermediate D, the
opening of the epoxide occurs to generate the tertiary
carbocation E, and subsequent migration of the hydride to
form 8b is much faster than rotation around the bond (Figure
2). The conformation that would lead to the expected anti
a Reaction conditions: (a) TBSCl, ImH, CH2Cl2, 0 °C, 90 min,
94%. (b) PrPPh3Br, KHMDS, THF, 23 to -78 °C, 9b, -78 to 23
°C, 85%. (c) TBAF, THF, 0 to 23 °C, quant. (d) tBuOOH, Ti(OiPr)4,
D-(-)-DET, CH2Cl2, -15 °C, 24 h. (e) TBSCl, ImH, DMF, 23 °C,
1 h. (f) PPh3, DEAD, p-NO2C6H4CO2H, Et3N; K2CO3, MeOH.
the (Z)-allylic ether, 10a, which was deprotected. The
resulting allylic alcohol 10b was subjected to the Sharpless
asymmetric epoxidation under kinetic resolution conditions
to afford the epoxide 11a and the alkene 12. The other
diastereomer 11b was obtained in variable yield from the
epoxidation but could be prepared via a Mitsunobu reaction
on 11a followed by hydrolysis.6 Silylation of 11ab proceeded
smoothly to give the two diastereomeric silyl ethers 13ab.
Non-aldol aldol rearrangement of 13b occurred as expected
to give the anti product 14 along with some elimination
Figure 2.
(4) Compounds 8a, 8b, and 15 were desilylated, and the NMRs of the
resulting keto alcohols matched those in the literature: Hoffmann, R. W.;
Ditrich, K.; Fro¨ch, S. Liebigs Ann. Chem. 1987, 977-985.
(5) (a) Sreekumar, C.; Darst, K. P.; Still, W. C. J. Org. Chem. 1980, 45,
5, 4260-4262. (b) Jung, M. E.; Fahr, B. T.; D’Amico, D. C. J. Org. Chem.
1998, 63, 2982-2987.
migration F (Figure 2) is presumably too hindered to be
formed in large amounts perhaps due to the severe non-
bonded interaction between the ethyl and methyl groups as
shown.
(6) Mitsunobu, O. Synthesis 1981, 1-28.
3376
Org. Lett., Vol. 5, No. 19, 2003