JOURNAL PRE-PROOF
2
Tetrahedron Letters
Scheme 1. Initial retrosynthetic analysis of (-)-caribenol A (1)
Conjugated reduction of ,β-unsaturated ketone 5 with L-
selectride in -78 ℃ generated the corresponding lithium enolate,
which was subsequently converted to the enol triflate 9 in 85%
yield by addition of Comins reagent (Scheme 3). Selective
removal of the O-TBS in 9 was then investigated. After several
trials, we found that treatment of 9 with hydrogen fluoride-
pyridine in THF successfully gave the desired alcohol 10. The
intramolecular carbonylation-lactone formation smoothly took
place in the presence of Pd(OAc)2 and dppf to furnish the
bicyclic lactone 11 in 50% yield.11 Desilylation of 11 with K2CO3
in methanol provided the key Pauson-Khand precursor 12 in 65%
yield.
As shown in Scheme 2, our synthesis commenced with the
known compound 2. Owing to the steric hindrance of the
isopropenyl group,3b the aldol condensation of
2
with
methacrolein furnished alcohol 3 in 90% yield with single
configuration at C6 and a pair of separable diastereomers at C5 in
a satisfied stereselectivity (7:1 d.r). The alcohol 3 was then
protected by using TBSOTf and 2,6-lutidine to provide the
hydroxy masked compound 4 in 98% yield. The CuI-catalyzed
substitution reaction of TMS acetylene with allyl chloride 4
afforded the desired enyne 5 in 87% yield.10
With the key cyclization precursor 12 in hand, we proceeded
to strategically construct the 7-5 ring system. Under the classic
Pauson-Khand reaction condition (Co2(CO)8, toluene, 110℃),12
a
single product was obtained in 54% yield. To our surprise,
although the mass spectrum of the product revealed the same
result with our design compound 14, the NMR data of the
1
product looked pretty confused. The H NMR spectrum of the
product showed a signal of β-enone proton instead of α-enone
proton (δ=7.06 vs 6.30-6.47). Fortunately, this compound
provided crystal for analysis by X-ray crystallography. The result
obtained from this determination was quite amazing: the Pauson-
Khand reaction afforded a [5.2.1] bicyclooctane bridged ring
product 13! Here we call this as the type-II intramolecular
Pauson-Khand reaction due to formation of bridged 8/5 ring
product. The widely used Pauson-Khnad reaction to build fused
6/5 or 7/5 ring is named type-I Pauson-Khand reaction.
Scheme 2. Synthesis of compounds 5 and 6.
Considering that enyne 5 could serve as a potential cyclization
precursor, Pauson-Khand reaction of 5 was firstly investigated
(Scheme 2). Unfortunately, although subjecting 5 to Co2(CO)8 in
toluene would generate the corresponding Co2(CO)6 complex,
which would go back to 5 after heating or treated with excess N-
methylmorpholine oxide (NMO). The TMS-deprotected
compound 6 was then prepared by treatment 5 with K2CO3.
However, Pauson-Khand reaction of enyne 6 gave the same
result, which might be due to the far distance between alkene and
alkyne impeded the insertion of alkene to cobalt alkyne complex.
Other metal catalysts and reaction conditions were also
examined, however, all of them failed to give the desired product.
We speculated that the distance between alkene and alkyne might
become closer after the formation of lactone owing to the
conformational restriction. Thus, the construction of lactone was
then pursued.
Then, we performed DFT (B3LYP) calculations to understand
the unusual regioselectivity of the PK reaction of enyne 12
(Figure 1). According to the previous computational studies,13 the
regioselectivity of the PK reaction was plausibly controlled by
the rate-determining alkene insertion step. We computed the two
reaction pathways (type-I and type-II Pauson-Khand reactions)
leading to the experimentally observed cycloadduct 13 and the
anticipated product 14, respectively (see Supporting Information
for more details). First, ligand displacement of CO in
alkyne−dicobalt complex A by the alkene moiety results in the
generation of two isomers B and D. After that, intermediate B
may undergo alkene insertion into the C(distal)−Co bond via
transition state TS1 (the overall Gibbs energy of activation for
this pathway is 32.7 kcal/mol). The resulting cobaltacycle C then
proceeds through CO insertion and reductive elimination to
furnish α-substituted cyclopentenone 13 (type-II Pauson-Khand
pathway).
Alternatively, intermediate D may undergo a similar reaction
pathway to form β-substituted cyclopentenone 14, in which the
rate-determining step is the alkene insertion into the
C(proximal)−Co bond via TS2 (the overall Gibbs energy of
activation is 39.9 kcal/mol). This pathway is expected to give the
traditional type-I Pauson-Khand product. Our DFT calculations
suggested that formation of cycloadduct 13 is the final product
because type-I pathway is disfavored by 7.2 kcal/mol compared
to type-II pathway, which is in good accordance with our
experimental observations.
Previous DFT calculations from several groups found that, the
intermolecular PK reactions of terminal aliphatic alkynes favored
the formation of α-substituted cyclopentenones in view of both
steric and electronic effects (Figure 2).14 In the intramolecular PK
reactions, for 1,6- and 1,7-enynes, the ring strains overrode the
steric and electronic effects so that the type-I PK reactions took
place in most cases (Figure 3). In contrast, for 1,8-enynes such as
12, the ring strains were not so severe. Therefore, the type-II PK
reactions were favored, leading to the generation of α-substituted
cyclopentenone 13.15 The above calculations confirmed these.
Scheme 3. Synthesis of compound 13.