8994 J . Org. Chem., Vol. 66, No. 26, 2001
Guindon et al.
Sch em e 1a
etherification reaction involves a fully developed halo-
nium ion and that the cyclofunctionalization reaction
proceeds through a π-complex before undergoing charge
separation (Figure 1).10 It should also be noted that the
presence of a terminally disubstituted double bond forces
the minimization of allylic 1,3-strain, an important steric
constraint in the transition states proposed.
An analysis of these transition states, based on the
minimization of the different torsional strains, would
suggest that trans-predictive transition state A (Figure
1) is the lowest in energy while trans-predictive transition
state B suffers from the presence of an axial hydroxyl
and a gauche staggering within the ring. Nevertheless,
these two transition states are preferred over cis-predic-
tive transition states C and D, which are destabilized
by two gauche effects and by allylic 1,3-strain, respec-
tively.
Several additional points should be made about the
electronic effects of both the ester (end-substituent of
the olefin) and the allylic hydroxyl. First, the presence
of the ester renders the olefin less reactive toward the
electrophile.2a Second, most of the positive charge (in the
onium or in the π-complex) is found on the olefin
â-carbon. Given this, as well as for stereoelectronic
reasons, an electron-withdrawing allylic hydroxyl at the
equatorial position (transition state A, Figure 1) should
be better aligned for maximizing conjugation of the σ*C-O
orbital with either the π-system11 or the carbonium ion,
which results in a decrease in the rate of cyclization or
an increase in the energy of the carbonium ion, respec-
tively. Having the electron-withdrawing allylic group in
an orthogonal position (i.e., axial as in B) with respect
to the π-system might avoid the rate-retarding effect,12
which would make transition state B the lowest in
energy.2a,13 Since both A and B lead to the same product,
however, additional stereochemical information must be
encoded into the substrates in order to differentiate
between two trans-predictive transition states.
a
Conditions: (a) DMSO, (COCl)2, CH2Cl2, -78 °C, 1.5 h, then
DIAE; (b) MeMgBr, THF, -78 °C, 1 h, 52% in two steps; (c) 1 N
HCl aq, THF, 24 h; (d) TBDPSCl, imidazole, DMF, 15 h, 63% in
two steps; (e) PhCH(OMe)2, p-TsOH, CH2Cl2, 15 h; (f) TBAF, THF,
1 h, 62% in two steps.
When minimization of torsional strain is the controlling
factor in the iodoetherification reaction, the transition
state possessing the equatorial hydroxyl group at the
allylic carbon should be preferred (as in A, Figure 1).
Given the steric interaction in F resulting from the
presence of an axial methyl group, this transition state
should be higher in energy than E, where all of the
substituents are equatorial. In this case, the anti diol 4
will have cyclized faster than the syn diol. If both
torsional strain minimization and stereoelectronic argu-
ments are involved, then the preferred transition states
will be those in which the hydroxyl group is orthogonal
(as in B, Figure 1) to the π-system of the double bond (G
and H, Figure 2). In this scenario, transition state G
should be higher in energy than H due to the develop-
ment of a transannular diaxial interaction, and the syn
diol 6 will have reacted faster than the anti diol 4.
In summary, the anti diol should cyclize faster than
the syn when steric effects are the only controlling factor
in the reaction. If both allylic 1,3-strain and stereoelec-
tronic effects are involved, then the syn diol should react
more rapidly.
P r ep a r a tion a n d Cycliza tion of th e Secon d a r y
Alcoh ols 4 a n d 6. The required substrates were pre-
pared as illustrated in Scheme 1. The protected triol 82a
was oxidized into the corresponding aldehyde, which was
immediately subjected to a Grignard reaction using
methylmagnesium bromide to give a 3:2 mixture of
secondary alcohols 9a and 9b. Hydrolysis of the aceto-
nides followed by selective protection of the primary
alcohol, formation of the 1,3-benzylidene acetals, and
subsequent fluoride treatment led to products 11 and 12,
which were easily separated. The relative configurations
of the benzylidene acetals 11 (anti) and 12 (syn) were
determined by the structures of cyclized products 5 and
7 (see below). Both compounds were subjected to a one-
pot Swern oxidation/Wittig reaction to give, after hy-
drolysis, diols 4 and 6, respectively (Scheme 2). These
compounds were then independently subjected to iodo-
etherification conditions in THF.
The search for the operative trans-predictive iodo-
etherification transition state begins with the following
competition experiments involving anti diol 4 and syn
diol 6 (Figure 2). In such competitions, the difference in
reactivity of these diols is a reflection of the difference
in energy of the respective transition states, which can
in turn indicate whether the allylic hydroxyl substituents
involved are axial or equatorial. The four possible transi-
tion states leading to the cyclized products 5 and 7 from
the respective diols 4 and 6 are illustrated in Figure 2.
(8) In the case of halogen additions, onium intermediates are
generally freely reversible. See: (a) Brown, R. S.; Gedye, R.; Slebocka-
Tilk, H.; Buschek, J . M.; Kopecky, K. R. J . Am. Chem. Soc. 1984, 106,
4515. (b) Slebocka-Tilk, H.; Ball, R. G.; Brown, R. S. J . Am. Chem.
Soc. 1985, 107, 4504. (c) Bellucci, G.; Bianchini, R.; Ambrosetti, R. J .
Am. Chem. Soc. 1985, 107, 2464. (d) Bellucci, G.; Chiappe, C.; Marioni,
F. J . Am. Chem. Soc. 1987, 109, 515. (e) Reitz, A. B.; Nortey, S. O.;
Maryanoff, B. E.; Liotta, D.; Monahan, R., III J . Org. Chem. 1987, 52,
4191.
As seen in Table 1, syn diol 6 gave compound 7 in good
yield after 6 h of reaction at room temperature (entry 1).
The unambiguous proof of structure for 7 was determined
via crystallization and X-ray analysis of the correspond-
ing dinitrobenzoate derivative 13.14 Interestingly, anti
diol 4 seemed much less reactive, giving less than 20%
conversion of the final product 5 after 24 h in THF at
room temperature (Table 1, entry 2). The best yield
(9) In the cyclofunctionalization reaction of an R,â-unsaturated ester,
similar results were obtained for the free allylic alcohol, the OMe
counterpart, and the allylic fluorine. See ref 2a.
(10) Chamberlin, A. R.; Mulholland, R. L., J r.; Kahn, S. D.; Hehre,
W. J . J . Am. Chem. Soc. 1987, 109, 672.
(11) Danishefsky, S. J .; Larson, E.; Springer, J . P. J . Am. Chem.
Soc 1985, 107, 1274.
(12) Endo alkoxy effect: Houk, K. N.; Moses, S. R.; Wu, Y.-D.;
Rondan, N. G.; J ager, V.; Schohe, R.; Fronczek, F. R. J . Am. Chem.
Soc. 1984, 106, 3880.
(13) In support of these orthogonal arguments, see: Kahn, S. D.;
Pau, C. F.; Chamberlin, A. R.; Hehre, W. J . J . Am. Chem. Soc. 1987,
109, 650.
(14) See the Supporting Information.