Organic Letters
Letter
this disconnection has involved either the Lewis-acid-mediated
addition of olivetol 9 to allyl alcohol 67a or the combination of
organometallic reagents with electrophiles such as 77b or 8.7c
After considering previous strategies, we believed the best
approach would involve a modification to the work of Pan and
She who exploited the Marino mixed organocuprate method8 to
add aryl nucleophiles to silyl enol ether 8 as a means to access
both (+)-machaeriol D7d and (−)-Δ8-trans-tetrahydrocannabi-
nol7c employing both (+)- and (−)-carvone [(−)-12].
A main drawback of their work, however, is the requirement of
generating a moisture-sensitive nucleophile from 13 via
magnesium−halogen exchange followed by addition of the in
situ-generated Grignard reagent to a suspension of copper(I)
cyanide.7c As initially disclosed by Marino,8 it is believed that
this reaction occurs via an SN2′ mechanism; subsequent β-
hydroxy elimination is effected in a second synthetic operation
using 1 M HCl to generate α-aryl carvone (+)-14 in 51% isolated
yield from 8 [or 40% from (−)-12] (Scheme 2).7e Of note,
by She and Pan (51% over two steps, using olivetol-derived
13).7c To our surprise, ethereal solvents and both titanium and
copper Lewis acids provided good, but somewhat lower, yields
when compared to those of either BF3·OEt2 or Me3SiOTf in
CH2Cl2. Increasing the reaction time to 1 h or separately
employing 2 equiv of Lewis acid did not improve yields, and
allowing the reaction to run at higher temperature (0 °C and rt)
resulted in decreased product formation. Importantly, nucleo-
philic addition occurred without the need to employ air- and
moisture-sensitive organometallic reagents and gave the C5−C6
1
trans product exclusively, as evidenced by H NMR spectros-
copy. Several unidentified byproducts (accounting for the mass
balance) were also isolated and spectroscopically determined to
arise from the degradation of 8 not having incorporated the aryl
group.
Once suitable conditions had been established, our attention
turned to a screen of several different nucleophiles with 1,3-
dimethoxybenzene being the first (Scheme 3). Based on the
Scheme 2. Previous Work and Optimization Studies
Scheme 3. Regiochemical Outcome with Dimethoxybenzene
well-known nucleophilic preferences for substituted aromatics
with electron-donating groups (i.e., ortho/para directing),12 we
anticipated that there might be two distinct modes of addition.
Not surprisingly, when a solution of 8 and dimethoxybenzene
was treated with BF3·OEt2, both possible α-addition products
were isolated after aqueous workup and silica gel chromatog-
raphy [(+)-16a and (+)-17a] (Scheme 3), with the major
product resulting from a preference for addition at C(4) of the
aryl ring and not at C(2), which lies between the two methoxy
substituents.13 Similar results were realized when using
Me3SiOTf as Lewis acid, although the yields of the two
regioisomers were somewhat higher than that of BF3·OEt2. As a
result, all subsequent reactions were screened solely using
Me3SiOTf.
It was at this stage that we began questioning the mechanism
of our Lewis-acid-mediated reaction (vide infra). Previously,
when addition occurred using the Marino mixed organocuprate
method, it was necessary to synthesize 8 so that the epoxide
leaving group could be orientated anti to the incoming
nucleophile (Scheme 2).8b However, based on knowledge
from the literature,14 we believed it was reasonable to assume
that, under Lewis acidic conditions, the epoxide in 8 may first
open to a reveal a 3° allylic carbocation, which is then trapped by
the incoming aryl nucleophile via an SN1′ process (19) (Scheme
4). If this mechanism is operative, it should be possible to access
the same products 16 and 17 using diastereomer 18 (Scheme
4),15 where the approach of the nucleophile would be syn to the
epoxide. If viable, α-addition products of this type could be
achieved in two fewer steps from carvone and three fewer steps
when compared to the Marino method.8 After 18 was
synthesized and subjected to our Lewis-acid-mediated reaction
conditions, (+)-15 was isolated in good overall yield from (−)-3
(Scheme 4), which helped to confirm our hypothesis that an
SN1′ mechanism may be operative.
based on the established literature precedent, for a productive
SN2′ addition to occur, the nucleophile must approach anti to
the face of the epoxide.8b To generate this epoxide from carvone,
it is first necessary to stereoselectively reduce the ketone in
(−)-12 to an alcohol so that epoxidation can occur in a
substrate-directed fashion before oxidation and silyl ether
formation [(−)-12 to 8] (four steps, 78% yield).7c,9
Based on previous studies in our laboratory,10 we believed
that it might also be possible to access related α-arylated
products under Lewis acidic conditions; however, a search of the
literature did not reveal any previous reports documenting the
addition of nucleophiles to cyclohexadiene monoepoxides
bearing a C3 siloxy group in this fashion.11 When a solution of
8 and 1,3,5-trimethoxybenzene in ether was treated with a single
equivalent of BF3·OEt2 at −78 °C, the desired α-arylated
carvone product [(+)-15] (Scheme 2) could be accessed
directly, without the need for an additional β-elimination step.
Whereas the isolated yield for this adduct was extremely low
(15%), we were encouraged by this result and initiated a
solvent/Lewis acid screen. We found that when using either BF3·
OEt2 or Me3SiOTf in CH2Cl2, yields could be improved to 52
and 50%, respectively (Scheme 2 table, red), which is
comparable to the direct nucleophilic addition method used
B
Org. Lett. XXXX, XXX, XXX−XXX