anticipated, the monosubstituted cyclopropene olefin in 1,
the stability of which is ensured by the vicinal gem-dimethyl
group, underwent regioselective electrophilic activation by
[Rh2(OAc)4] to generate the secondary cyclopropyl cation 3,
which has the rhodium atom at the less-hindered site. Ring
opening of cation 3 would then generate the rhodium
carbenoid 4 that promotes the intramolecular cyclopropana-
tion of the alkene (Scheme 4). Rationalization of the
observed high diastereoselectivity deserves further investiga-
tion but the gem-dimethyl substitution of the olefin in the
rhodium carbenoid 4 is presumably a key control element,
owing to minimization of the A1,3 strain in the cyclic transition
state, as previously observed in the gold-catalyzed cyclo-
isomerization of cyclopropenes H.[15]
Scheme 3. Gold-catalyzed cycloisomerization of cyclopropene-ene com-
pounds. Ts=p-toluenesulfonyl.
pene-enes that does not involve diazo compounds. This
reaction proceeds by an intramolecular olefin cyclopropana-
tion to afford functionalized unusual heterocyclic scaffolds,
which possess a [6.1.0] bicyclic ring system fused to an
aromatic ring.
The conversion of cyclopropene 1 into benzoxocane 2 is
clearly distinct from the previously reported rhodium-cata-
lyzed cycloisomerizations that involve cyclopropenes and
proceed by a ring-opening reaction.[24] Notably the donor-
substituted rhodium(II) carbenoid 4 displays an unprece-
dented high reactivity toward the olefin cyclopropanation,
despite its lower electrophilicity compared with rhodium
carbenoids that are generated from a-diazo carbonyl com-
pounds that bear at least one electron-withdrawing group.[1,13]
The scope of this new rhodium-catalyzed cycloisomeriza-
tion involving cyclopropenes was first evaluated with (2-
allyloxyphenyl)cyclopropenyl carbinols 5–11, which bear a
variety of substituents on the aromatic ring, and were
synthesized in two steps from the corresponding substituted
salicylaldehydes.[25] Unsurprisingly, when a methyl or a
methoxy group was present on the aromatic ring the
corresponding cycloisomerization products 12 and 13 were
isolated in 99% and 97% yield (Table 1, entries 1 and 2). The
5-fluoro-, 5-bromo-, or 3-bromo-substituted analogues 7–9,
also underwent a clean rhodium-catalyzed cycloisomerization
to provide the corresponding eight-membered oxygen hetero-
cycles 14–16, in high yields (Table 1, entries 3–5). The
presence of a bromide substituent, which could be potentially
useful for further functionalization, was tolerated, and a nitro
group was also compatible, as exemplified by the cyclo-
isomerization of cyclopropenyl carbinol 10 leading to com-
pound 17 (Table 1, entry 6). The alkynyl substituent in the
cyclopropenyl carbinol 11 did not interfere with the reaction
and the corresponding cycloisomerization product 18 was
isolated in almost quantitative yield (Table 1, entry 7). In all
To cycloisomerize cyclopropene-enes into medium-sized
heterocycles, the reactivity of cyclopropenyl carbinol 1
(Scheme 4), which is readily available by condensation of
3,3-dimethylcyclopropenyl lithium to O-allyl salicylalde-
hyde,[18] was investigated in the presence of gold catalysts.
Whereas AuCl3 (5 mol%) led to the complete conversion of 1
within 1 hour (CH2Cl2, RT), the reaction did not go to
completion with AuCl (2/1 = 60:40 after 12 h); in both cases,
the expected cycloisomerized product 2 was accompanied by
several unidentified by-products that could not be easily
removed by flash chromatography on silica gel. The catalysts
[(Ph3P)AuSbF6] and [(Ph3P)AuNTf2] (Tf = trifluoromethane-
sulfonyl)[19] were even less satisfactory and led to complex
mixtures that contained either none ([(Ph3P)AuSbF6]) or
trace amounts of compound 2 ([(Ph3P)AuNTf2]). The use of
rhodium catalysts was then considered, and [Rh2(OAc)4]
(0.5 mol%) was found to smoothly catalyze the cycloisom-
erization of cyclopropenyl carbinol 1 into the desired eight-
membered oxygen heterocycle 2, which possesses a [6.1.0]
bicyclic ring system, in almost quantitative yield (99%) and
with excellent diastereoselectivity (d.r. ꢀ 97:3).[20] The relative
configuration of 2 was unambiguously determined by single-
crystal X-ray diffraction of the corresponding p-nitroben-
zoate derivative.[21]
Other rhodium(I) and rhodium(II) catalysts were also
screened but provided inferior results.[22] Notably the cyclo-
isomerization of the optically enriched (S)-1 (94% ee) led to 2
with the same optical purity, thereby indicating that the
hydroxy-substituted stereocenter was not affected.[23] As
1
cases, only a single diastereomer was detected by H NMR
spectroscopy (d.r. ꢀ 96:4).
The scope of the cycloisomerization was then evaluated
with (2-allyloxyphenyl)cyclopropenyl carbinols 19–23, which
bear substituted allylic ethers (Table 2).[25] Methallyl ether 19
gave the eight-membered oxygen heterocycle 24 (95%),
which possesses a trisubstituted cyclopropane, as a single
detectable diastereomer (Table 2, entry 1). An iodide sub-
stituent was tolerated, as shown for cyclopropene 20, which
led to the benzoxocane 25 in 96% yield (Table 2, entry 2). The
cinnamyl ether 21 was converted into the oxygen heterocycle
26 (99%), which bears an additional stereocenter on the
cyclopropane ring, as a single diasteromer (Table 2, entry 3).
As anticipated, the cycloisomerization proceeds with the
stereospecific cyclopropanation of the alkene; this stereo-
Scheme 4. Rhodium-catalyzed cycloisomerization of cyclopropenyl car-
binol 1.
Angew. Chem. Int. Ed. 2011, 50, 5932 –5937
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5933