10.1002/anie.201807709
Angewandte Chemie International Edition
COMMUNICATION
Scheme 3. Cyclobutane expansion via DeMayo fragmentation-aldol addition sequence and synthesis of the cyclic skeleton of epicolactone. Reagents and
conditions: a) OsO4 (30 mol%), NMO (3.0 equiv), CH2Cl2, 0 °C to 25 °C, then Pb(OAc)4 (2.5 equiv), CH2Cl2, 0 °C, 99% b) O-phenyl chlorothionoformate (3.0 equiv),
NEt3 (1.5 equiv), DMAP (5 mol%), 25 °C, 94%; c) AIBN (35 mol%), Bu3SnH (4.0 equiv), PhMe, 80 °C, 90%; d) LDA (1.05 equiv), 17, then PhNTf2 (2.4 equiv),
THF/HMPA, –78 °C, 51%; e) 19 (5.0 equiv), Pd(PPh3)2Cl2 (25 mol%), dioxane, 85 °C, 90%; f) TFA, CH2Cl2, 25 °C, quant. NMO = N-methylmorpholine N-oxide,
DMAP = 4-dimethylaminopyridine, AIBN = 2,2’-azobisisobutyronitrile, LDA = lithium diisopropyl amide, THF = tetrahydrofuran, HMPA = hexamethylphosphoramide,
TFA = trifluoroacetic acid.
With a robust route that allowed for the preparation of
multigram quantities of 11 in hand, we examined the ring
expansion required to access the central five-membered ring of
epicolactone via the proposed acyloin rearrangement. Such a
structural reorganization of the carbon skeleton was observed
other four electrophilic positions in 17 by exploiting its enolizable
-protons.
The formation of enol derivatives of 17 using amine bases
and Tf2O/NfF or employing an established two-step procedure[16]
failed. We then turned our attention to the triflation of the alkali
upon treatment of 11 with refluxing TFA. To our surprise, however, metal enolates of 17. Strict control of the reaction time turned out
the angularly fused tetracycle 14 was obtained in 93% yield.[13]
This product is likely formed by migration of the C(9)-C(14) bond,
suggesting that the underlying mechanism for the ring expansion
resembles that of a retroaldol-aldol sequence, similar to the
to be crucial for the successful outcome of the reaction.
Employing an almost equimolar quantity of LDA in THF/HMPA
followed by trapping of the enolate with PhNTf2 gave triflate 18 in
51% yield. In search of a side chain which could be installed by
cross coupling, we speculated that a dioxene such as 20 would
enable an intramolecular nucleophilic attack on the C(14) ketone.
Its synthesis was effected by coupling known stannane 19[17] to
18 in 90% yield. It should be noted that the sidechain bears the
correct oxidation pattern found in epicolactone. While dioxenes
have found use in carbonyl addition reactions through lithiation of
the vinylic C–H bond,[18] we found that exposure of 20 to TFA in
CH2Cl2 led to quantitative formation of hexacycle 21 via an acid-
mediated aldol addition. This transformation successfully closed
the last carbocycle of epicolactone without revealing its -
hydroxyenone motif. With 21 in hand, we sought to install the
methyl group at C(10) via selective addition to the carbonyl of the
vinylogous ester in presence of its conjugated position and the -
lactone. Unfortunately, exposure of 21 to various C1-
organometallic reagents resulted mainly in decomposition of the
substrate and/or poor regioselectivities. We speculated that O–H
deprotonation by basic organometallic reagents preceded
nucleophilic addition to C(10), leading to an unreactive alkoxide.
We envisioned that this innate tendency to resist attack by
external nucleophiles could be circumvented by employing a
silicon tether strategy.[19] Since the tertiary hydroxyl group in 21 is
positioned above the plane of the vinylogous ester, we speculated
that the C–I bond in silyl ether 22 could be converted into an
organometallic nucleophile poised for an intramolecular carbonyl
addition (Scheme 4). Upon treatment of 22 with 2.1 equiv of SmI2
in THF at –78 °C and subsequent acidic workup, clean formation
of enone 24 was observed in high yield.[20]
DeMayo fragmentation, rather than
a concerted 1,2-shift
(Scheme 3). Since the carbonyl functionality which promotes this
undesired pathway in 11 is part of a vinylogous ester motif, we
envisioned that conversion of the 1,1-disubstituted olefin in 11 to
the corresponding ketone 12 would facilitate a pathway involving
the C(1)-C(14) bond rearrangement to outcompete the undesired
C(9)-C(14) counterpart. Exposure of olefin 11 to OsO4 and NMO
in CH2Cl2 followed by one-pot oxidative diol cleavage with lead
tetraacetate gave ketone 12 in near quantitative yield. Treatment
of 12 with TFA or Me2AlCl analogous to 11 led to quantitative
starting material recovery and decomposition, respectively. The
use of BF3-based Lewis acids turned out to be crucial for the
successful outcome of the desired retroaldol-aldol sequence.
While treatment of 12 with BF3·OEt2 led only to trace amounts of
desired product, exposure of 12 to excess BF3·2HOAc in refluxing
CH2Cl2 produced the ring-expanded cyclopentanol 16 as the sole
product in 55% yield.[14],[15] The tertiary alcohol in 16 was
subsequently transformed to its O-thiocarbonate derivative and
subjected to radical deoxygenation to generate 17 in 85% yield
over the two steps. Having the central cyclopentane core structure
assembled in 17, we sought to attach a suitable C2 fragment to its
methyl ketone moiety to close left hand side carbocycle of
epicolactone. Unfortunately, the selective addition of nucleophiles
to the methyl ketone in 17 failed. This was due to the tendency of
nucleophiles to add preferentially to the C(14) carbonyl. We hence
envisioned differentiating the methyl ketone carbonyl from the
This article is protected by copyright. All rights reserved.