60
Can. J. Chem. Vol. 92, 2014
Scheme 2. Intermediates en route to transannular cyclization.
BF ·OEt conditions. However, it is notable that effective con-
Table 3. Nazarov cyclization and exo-elimination of trienones 3 and
dienone 7.
3
2
version of 3a to 4a could be accomplished with only catalytic
quantities of Sc(OTf)3.16 Diethylaluminum chloride also fur-
nished 4a (Table 2, entry 5), but with diminished yields and
the formation of multiple side-products. Finally, two Brønsted
acids, formic acid and trifluoroacetic acid, were examined
(Table 2, entries 6 and 7) but gave either intractable mixtures or
unconsumed starting material.
Failure to observe any triquinane formation from 3a under all
conditions surveyed indicates that the transannulation step is
difficult. We envision two possible explanations (Scheme 2). First,
the cyclooctene ring fused to the initially formed cyclopentenyl
cation A must assume a tub-like conformation (A=) to bring C-1
and C-5 within bonding distance, which may be disfavored in the
conformational equilibrium. Furthermore, cyclization of A= to
triquinane B would afford a relatively high-energy secondary car-
bocation at C-4, which is likely to raise the barrier for trans-
annulation. Notably, previous successful intramolecular trapping
reactions of the Nazarov intermediate by olefins proceeded
1
2
a
Entry
Substrate
R
R
Product (% yield)
1
3a
3b
3c
3d
3e
7
Me
Me
H
Ph
Me
Ph
H
4a (72)
5
,6
through tertiary carbocations. Moreover, Mehta and Rao found
that it was necessary to use a trisubstituted alkene to see efficient
b
2
3
4
5
6
a
4b (61)
c
4c (64)
8
transannulation in their triquinane strategy. With this in mind,
d
Me
O
it is unlikely that the desired process can be effected without
substantial redesign of the substrates. However, clean formation
of 4a under several conditions with no regio- or stereoisomers
offers a valuable synthetic tool. The bicyclo[6.3.0]undecane skele-
ton is found in a variety of diverse terpenoid natural products,
e
−(CH ) −
4e (80)
2
4
Me
Ph
8 (80)
Yields are for isolated product after chromatographic purification.
A 4:1 mixture of anti:syn diastereomers at R was obtained.
b
c
1
Reaction done at 0 °C.
1
7
18
19
including precapnelladiene, dactylol, pleuromutilin, and
d
Reaction formed a complex mixture of products.
R and R (cyclohexano ring methylene carbons) are syn in the case of 4e.
variecolin.2 With this in mind, we examined all of the substrates
0
e
1
2
(3a−3e and 7) under the optimal conditions to evaluate the gener-
ality of this useful transformation (Table 3).
whether a cyclooctene conformation favoring rapid elimination to-
Like 3a, substrates 3b and 3c also underwent conversion to the
corresponding bicyclo[6.3.0]undecadienones 4b and 4c (Table 3,
entries 2 and 3). However, dimethyl product 4b was not formed
with complete diastereoselectivity, furnishing a 4:1 mixture of
anti and syn isomers at C-9/C-10. Substrate 3c, lacking an ␣-
substituent on the dienone moiety, reacted cleanly but required
higher temperature (0 °C) compared with the other cases. Re-
moval of the -substituent, as in the case of 3d, was far more
damaging to the success of the transformation (Table 3, entry 4).
In this case, a complex mixture from which no characterizable
product could be isolated was obtained. Cyclohexenyl ketone 3e
cyclized to the tricyclic product 4e in very good yield (Table 3,
entry 5), although in contrast with the other examples, it provided
22
wards C-2 might predominate. We imagined that the remote alk-
ene might be exerting an effect on the conformational preferences,
and to evaluate this possibility, we subjected dienone 7 lacking that
unsaturation to the standard conditions (Table 3, entry 6). In the
event, this substrate was found to undergo efficient electrocycliza-
tion and exclusive exocyclic elimination to give 8 in high yield. Fur-
ther computational studies to evaluate transition state energies for
the various elimination pathways may help elucidate this high selec-
tivity. Regardless of its origin, this general regioselectivity is wel-
come, especially as it preserves the two adjacent stereocenters at C-8
and C-9, set in the electrocyclization, in contrast with the outcome
had C or D predominated.
2
1
exclusively the 9,10-syn diastereomer. In this case, we presume
that the conformational restrictions of a somewhat strained
hydrindanone skeleton exerted a strong preference for a syn rela-
tionship at the two bridgehead carbons. In all cases, we observed
the expected 8,9-anti relative configuration, established in the
initial 4 conrotatory electrocyclization.
Conclusion
We have described above an efficient and succinct route to gener-
ate fused bicyclic 8-5 ring systems, a common substructure in a vari-
ety of natural products. This method generates three new
stereocenters, two of which form stereospecifically and the other
stereoselectively. The cyclization is effected by several common
Lewis acids. Finally, products 4a−4e and 8 portend synthetic utility,
as the elimination occurs with high regioselectivity and the remain-
ing olefins are situated for further elaboration to more complex
natural products. The origins of this regioselectivity are under cur-
rent investigation. Domino transannulation to afford triquinane sys-
tems was not observed for 3a−3e, suggesting the need for structural
augmentation of the proposed trapping alkene.
There are at least three regiochemical options for the elimination
event: (i) endocyclic (relative to the newly formed cyclopentenyl cation)
towards C-8 to afford the ring-fusing alkene isomer C, (ii) endocyclic
towards C-9 to give cyclopentenone D, or (iii) exocyclic into the cy-
clooctene ring, the observed pathway that provides 4a−4c and 4e
(Scheme 3). Exocyclic elimination from C-10 is also an option for
all but 3c, giving alkylidene cyclopentenone E. The high regioselec-
tivity seen in these cases was surprising, prompting us to consider
Published by NRC Research Press