Journal of the American Chemical Society
Article
8
decalone, a widely used derivative of the venerable Wieland−
nucleophilic side chains could be affixed through simple
olefination chemistry (Figure 2a). We also believed that this
furan could be used to close the C-ring on a suitable allylic
substrate such as 15 through one of a number of two-electron
processes including intramolecular Friedel−Crafts reaction,
transition-metal-catalyzed allylic substitution, Nazarov-type 4π
9
Miescher ketone, would be easily accessible on a large scale
and serve as a launchpad for our annulation studies. However,
it quickly became apparent that, although it trivial to make, this
compound would require nine synthetic steps from commer-
cial 1,3-cyclohexanedione (10) or six steps from the
commercial Wieland−Miescher ketone itself, which was not
an attractive or practical solution on the scales that we
required, given its low monetary- and step-economy. We
therefore envisioned a more efficient, alternative route to this
material from the readily available terpene geranylacetone (11)
through a strategic radical polyene cyclization, as decalin-
forming polyene cyclizations commonly give the trans
stereochemistry in both radical and cationic modes.
9c
electrocyclization, or solvolytic S 2′ substitution.
N
The synthesis of 15, the precursor for these polar reactions,
proceeded from TIPS-decalone 16 through the Barton
15
iodination protocol, followed by lithium-halogen exchange
Synthesis of trans-Decalone 9. The first objective was to
install a suitable radical initiator and acceptor, which was
accomplished in a racemic fashion through a van Leusen
10
reductive cyanation of 11, followed by a chemoselective
epoxidation of the terminal olefin via an intermediate
11
bromohydrin (Figure 2b). Titanium(III)-promoted reductive
12
radical polyene cyclization of 12 proceeded smoothly to
furnish the trans-decalone as an inconsequential 5:1 mixture of
diastereomers at C8. Protection of the liberated C3 alcohol
with TBSOTf delivered protected bicycle 9. This simple, 4-
step sequence could be performed on a 20 g scale and provided
ample material for subsequent studies concerning the assembly
of the C-ring.
Despite its attractive step-economy, we recognized that, in
order for this process to compete with the conventional
Wieland−Miescher pathway, it was necessary to render it
enantioselective. Initial attempts at a Sharpless asymmetric
1
3
dihydroxylation on geranylacetone (11) with AD-mix-α were
promising, delivering optically pure diol 13 with high
regioselectivity (r.r. = 7.5:1) and excellent enantioselectivity
Figure 3. Attempted polar reactions to close the C-ring.
(
96% ee, Figure 2c). Somewhat slow conversion resulted in
modest yields for this reaction, but the starting material was
easily recovered. Given the commercial availability of both
starting material and catalyst, the overall process was quite
synthetically useful. With an interest in improving the
conversion of dihydroxylation, our attention turned toward
the underutilized Corey−Zhang ligand 14, finely tuned for
dihydroxylation of the terminal olefin in extended polyprenoids
such as squalene. We were very pleased to find that application
of ligand 14 afforded the desired diol 13 in 97% yield and
anticipated that the resultant C10 benzylic allylic alcohol
would be quite labile, which would be of some benefit as a
nucleofuge in several of the envisioned modes of cyclization; in
practice, however, the sensitivity of this hydroxyl substituent
was so great as to interfere with all further manipulations of
this compound. Efforts to promote Friedel−Crafts or 4π
electrocyclization reactions via Lewis acid-induced ionization
only resulted in elimination to form conjugated diene 20. This
is in accordance with examples from the literature that have
demonstrated furan’s reluctance toward Nazarov and Friedel−
14
9
7.5% ee (Figure 2c). Completion of the radical cyclization
precursor 12 was accomplished by a standard protocol of
mesylation and elimination to form the terminal epoxide and
an identical van Leusen cyanation as above. With multigram
quantities of optically pure 9, we approached the first major
challenge in the synthesis, involving construction of the C-ring.
Initial Efforts to Construct the C-Ring. We attempted to
build the C-ring through a number of different strategies that,
while ultimately unsuccessful, were illustrative of the obstacles
the trans-decalone system placed in front of this difficult
transformation, namely, the severe steric hindrance of the
ketone and the resultant twist-boat conformation of the B-ring
in the product, necessitating a high-energy ring-flip in the
transition state and producing the thermodynamically
disfavored tricyclic product. An early retrosynthetic proposal
identified tetracyclic furan 8 as an attractive intermediate
toward the isomalabaricanes, wherein the oxidative cleavage of
the furan with singlet oxygen was seen as a key step that
generated an exposed enal to which a diverse set of
16
Crafts reactions at the 3-position, as well as with a report of
an attempted aromatic Nazarov cyclization on a similar system
17
that also yielded dienes. Attempts to convert the hydroxyl
into an acetate, mesylate, or tosylate nucleofuge for solvolysis
and allylic substitution yielded a similar outcome.
The reactivity mismatch between an allylic electrophile and
the poorly nucleophilic C3 position of furan prompted us to
explore single-electron processes, where the emerging C−C
bond might be formed in the opposite direction; thus, the
furan was exchanged with an electrophilic γ-butenolide, the
allyl electrophile was reimagined as an allyl nucleophile, and
lactone 26 became the target molecule. We speculated that
decomposition of the alcohol to its allylic radical might result
in a Giese reaction to form the desired carbon−carbon bond at
the β-position. To this end, we synthesized enal 21 through a
Nozaki−Yamamoto homologation with dichloromethyllithium
18
(Figure 4).
2
140
J. Am. Chem. Soc. 2021, 143, 2138−2155