3022
S. Jarosz, K. Szewczyk / Tetrahedron Letters 42 (2001) 3021–3024
readily isomerized to the R–CHꢁCH2SnR3 derivatives
even with very mild Lewis acids.5 We expected, there-
fore, that compound 5 could be converted into 2. Of
course, Lewis acids cannot be used, since they cause
rearrangement of 2 to dienoaldehyde 4.
Its thermal decomposition (at 140°C) provided the cis-
dienoaldehyde 6b (75%), characterized as alcohol 7b
(obtained in 70% yield from 6b)9 (Scheme 2).
Such high-temperature reactions strongly suggest a
rearrangement involving radicals formed by homolytic
cleavage of the CꢀSn bond. The radical 10 would then
need to rearrange with breaking of the CꢀO bond, a
process which is very unlikely (route a in Scheme 3).
We decided, therefore, to perform this rearrangement at
high temperature; the results are summarized in Scheme
2. Heating the
D
-gluco-configured compound 5a4 (a
single isomer of unknown configuration) at 140°C for 4
h did not give the desired primary allyltin 2a, but a
product which did not contain the tin moiety. This
compound underwent facile reduction with NaBH4
providing a compound which was identified as the
dienoalcohol6 7a with the cis-configuration (J5,6=10.5
Hz) at the internal double bond.
To check out if this unprobable mechanism is possible,
we performed the reaction of 5b in the presence of
tributyltin hydride and separately hexabutyldistannane.
The first process should afford the reduction products
11 (from radical 10) and, the regioisomer 2b. The
reaction of 5b with the tributyltin radical (generated by
homolytic cleavage of the SnꢀSn bond in Bu3Sn–SnBu3)
should lead to the more stable regioisomer 2b. How-
ever, in both reactions only the product of decomposi-
tion of 5b—dienoaldehyde 6b—was formed. These
results excluded the possibility of decomposition of 2b
via a radical mechanism.
We reasoned that this unusual process (6) might
provide an easy access to carbobicyclic derivatives such
as those shown in Scheme 1, but with the opposite
stereochemistry at the ring junction. Thermal decompo-
sition (boiling xylene) of the allyltin 5a in the presence
of the simplest stabilized ylide (Ph3PꢁCH–CO2Me) was
expected to induce the tandem Wittig/Diels–Alder pro-
cess which should provide bicyclo[4.3.0]nonene deriva-
tive(s). Indeed, the formation of compound 9a—as the
single stereoisomer—was noted under these conditions.
The configuration of this cycloadduct was assigned by
The alternative mechanism had, therefore, to be consid-
ered. Tin atoms are only slightly more electropositive
than carbon (organostannanes, because of the weak
SnꢀC bond, exhibit reactivity as carbanions or
radicals10) and, therefore, tetraalkylstannanes might be
regarded as extremely mild Lewis acids. Complexation
of such tin derivatives (Bu3Sn–Sug route b in Scheme
3), although very weak, is possible. At high temperature
(boiling xylene, 140°C)11 decomposition of 5b according
to the ionic mechanism could occur with elimination of
the tributyltin cation. This species is a much stronger
Lewis acid and complexes the methoxy group of 5b
inducing again, decomposition and providing
dienoaldehyde 6b (Scheme 3).
1
careful NMR experiments: COSY H–1H and NOESY
correlations.7 The very high stereoselectivity of this
reaction can be rationalized assuming the endo transi-
tion state for the cycloaddition. The attack of the
dienophile part from ‘the top’ A results in formation of
9a. Transition state B (attack from ‘the bottom’) should
be prevented by the severe steric interaction of the
benzyloxy group at the C-6 position with the diene part
of the molecule (Fig. 1).
The
D
-manno-configured compound 5b8 (obtained from
We do not have an explanation for the stereospecific
formation of the cis-dienoaldehydes (6a and 6b) in the
the bromide 3b according to Ref. 4) behaved similarly.
Scheme 2. (i) Xylene, reflux 4 h (80%); (ii) NaBH4 (85%); (iii) Ph3PꢁCHCO2Me, xylene, reflux (75% from 5a).
Figure 1. The endo transition states leading to bicyclic adducts.