Organic Letters
Letter
precursor 15 would come from an aldol condensation and
subsequent reduction between aldehyde 12 and benzofuran-
3(2H)-one (3-coumaranone, 13). Aldehyde 12 would be
rapidly accessed from commercially available β-keto ester 9.
In the forward sense, aldehyde 12 was prepared in four steps
in high overall yields from β-keto ester 9 through a sequence of
methylation,2 triflation of the ketone, DIBAL-H reduction, and
Dess−Martin oxidation (Scheme 2). The aldol condensation
Table 1. Optimization of the Palladium-Catalyzed
Intramolecular Alkenylation of Ketone 15
a
Scheme 2. Preparation of Intermediate 15
entry
base (1.5 equiv)
solvent
yield of 7 (%)
1
2
3
4
5
6
7
LiHMDS
KHMDS
NaHMDS
LiOtBu
KOtBu
NaOtBu
Cs2CO3
LiOtBu
LiOtBu
LiOtBu
LiOtBu
LiOtBu
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.05 M)
THF (0.10 M)
THF (0.20 M)
THF (0.05 M)
PhMe (0.05 M)
21
34
20
44
25
14
35
b
8
60
b
9
51 (2)
39 (12)
65
b
10
between 12 and 13 was found to be most efficient in the
presence of basic alumina,3 which provided enone 14 in 84%
yield with complete Z selectivity. Selective reduction of the
trisubstituted enone olefin in 14 in the presence of a vinyl
triflate moiety turned out to be challenging. A number of
copper or manganese hydride-mediated conjugate reduction
methods4 failed, likely because the α-oxygen substitution
reduces the electrophilicity of the enone CC bond. Attempts
to use homogeneous palladium-catalyzed reduction5 were also
unfruitful, probably as a result of the competing reactivity of the
vinyl triflate toward oxidative addition with a Pd(0) species.
Eventually, Pd/C-catalyzed hydrogenation (with a H2 balloon)
was found to chemoselectively reduce the enone olefin to
compound 15 in 91% yield when toluene/DCM (50:1) was
used as the optimal solvent combination [see the Supporting
bc
,
11
bc
,
12
trace
a
Reactions were conducted on a 0.05 mmol scale in a 4 mL vial sealed
with a PTFE-lined cap. Unless otherwise noted, 10 mol % Pd(OAc)2
and 20 mol % QPhos were used. Yields were determined by 1H NMR
analysis using 1,1,2,2-tetrachloroethane as the internal standard.
Recovery of starting material is noted in parentheses. Vials were
flame-dried. 5 mol % Pd(OAc)2 and 10 mol % QPhos were used.
b
c
QPhos, and 1.5 equiv of LiOtBu in THF, which provided 7 in
65% (57% isolated) yield (entry 11). The structure of 7 was
unambiguously confirmed by X-ray crystallography of the
corresponding 2,4-dinitrophenyl (DNP)-hydrazone derivative
(16) (eq 1).
With compound 15 in hand, the stage was set to explore the
key intramolecular alkenylation to construct the 4,5-spirocycle.
While palladium-catalyzed ketone alkenylations have been well-
established,6a the use of such a reaction in an intramolecular
setting to access four-membered rings remains elusive.6b The
challenge is anticipated to come from the difficulty of forming a
highly strained ring via reductive elimination. It is known that
reductive elimination benefits from using sterically hindered
ligands. For example, Helquist and co-workers have demon-
strated that various tert-butyl-substituted phosphine ligands
promote the palladium-catalyzed intermolecular alkenylation of
ketones, with QPhos being the most efficient.7 Hence, we
began exploring the intramolecular alkenylation of compound
15 using QPhos as the ligand (Table 1). To our delight, under
Helquist’s original conditions (LiHMDS, THF, rt), the desired
4,5-spirocycle 7 was obtained, albeit in merely 21% yield (entry
1). The choice of base proved to be critical, and LiOtBu was
later found to be the best (entry 4) among all the bases
examined (entries 1−7). Removing residual water in the
reaction vessel by flame drying significantly improved the yield
to 60% (entry 8). Reactions carried out at higher concen-
trations resulted in incomplete conversions of 15 (entries 9 and
10). Using toluene instead of THF shut down the product
formation at room temperature (entry 12). The optimal
conditions were established to be 5 mol % Pd(OAc)2, 10 mol %
It is noteworthy that the other possible diastereomer (7′)
was not observed during the intramolecular alkenylation. To
understand the origin of this diastereoselectivity, computational
studies using density functional theory (DFT) were carried out.
It was suggested that during the reductive elimination step, the
transition state that forms the desired diastereomer is 6.9 kcal/
mol lower in energy than the undesired one (Figure 2). The
stabilization is likely due to a favorable coordinative interaction
between the carbonyl π-bond electrons and the palladium (see
To examine the influence of adjacent stereocenters and
fused-ring structures on the 4,5-spirocycle formation, we sought
B
Org. Lett. XXXX, XXX, XXX−XXX