2850
J. Michaux et al. / Tetrahedron Letters 55 (2014) 2849–2852
difficult as the triple bond is very electrophilic, thus leading to con-
jugate addition products rather than the expected esters.6 Indeed,
all our attempts to synthesize ADA esters of suitably functionalized
phenols failed.
We then imagined installing the carboxyl functions after the
cycloaddition. We thus readily prepared triyne ether
2 by
condensation of ortho-ethynylphenol7 with 1,4-dichlorobutyne
and submitted it to the cobalt catalyzed cycloaddition under clas-
sical reaction conditions5 (light irradiation in refluxing toluene) to
access the expected pentacyclic bis-benzochromene 3 in 62% yield
(Scheme 2). This first result was very interesting as it gave an easy
access to the highly substituted aromatic nucleus needed.
Unfortunately, all our attempts to fully oxidize 3 to the bis-chro-
mene-6-one 5 failed. In the best case, using conditions reported
for the oxidation of parent mono-benzochromene8 only traces of
bis-lactone 5 were obtained, in 36% yield of the mono-oxidized
product 4.
As our first two strategies failed, we envisaged to prepare fuma-
royl esters of type 6 and engage them in the [2+2+2] cycloaddition
(Scheme 3). But although the total intramolecular cycloaddition of
acyclic triynes9 and ene–diynes10 is rather well known, the use of
the linear yne–ene–yne sequence is much less common.11
We have thus prepared yne–ene–yne 6a by condensation of
ortho-ethynylphenol with fumaroyl chloride (Scheme 3). This lin-
ear substrate was then subjected to the classical Co(I) cycloaddi-
tion in refluxing toluene, under visible light irradiation.12 In this
case, a stoichiometric amount of cobalt catalyst is necessary, as
the intermediate cyclohexadiene is a very good ligand of Co(I).
After the oxidative decomplexation/aromatization of the cyclo-
hexadienyl–cobalt(I) intermediate, we were pleased to obtain
bis-lactone 5a in an encouraging 25% yield for two steps and the
formation of three CAC bonds. After many attempts, a maximum
of 30% yield could be obtained.13 This yield can seem low but it ap-
pears that bis-lactone 5a is rather unstable during chromato-
graphic purification and thus, part of the product is lost during
purification.14 The X-ray diagram of 5a15 shows a very strained
structure: as the two lactone carbonyl groups are pointing toward
each other, steric and electronic repulsion implies an out of plane
deformation of the central aromatic nucleus. The dihedral angle
of the planes containing the two C@O is 40° and their OAO
Scheme 3. Synthesis and [2+2+2] cycloaddition of yne–ene–yne 6a.
distance is 2.78 Å, indicating a trans conformation of the two lac-
tones as expected.
After this first successful reaction, we wanted to explore the
scope of this new methodology, and started with variation of the
R substitution at the alkynes (Table 1). Replacement of the phenyl
terminal group by a TMS or iodine (entry 2 and 3) does not lead to
any cyclized product but to extensive decomposition. A first possi-
ble explanation based on the steric hindrance was ruled out by the
fact that the terminal alkyne (R = H) does not give any cyclized
product either (entry 4), whereas a butyl chain led to 5% of the cor-
responding dilactone 5e.
These first results indicating the need for a terminal aromatic
group, we then prepared enediynes 6f and 6g, whose terminal aro-
matic is substituted respectively by a para-methoxy and a para-tri-
fluoromethyl group. In the case of the electronically enriched 6f,
the cylotrimerization proceeds in a better 40% yield (entry 6)
compared to the nonsubstituted case, and in only 5% yield for
electron poor 6g (entry 7).
Our next questioning was whether the geometry of the central
double bond plays an important role in the outcome of the
Table 1
Cyclization of yne–ene–yne 6
Entry
R
Isolated yield
1
2
3
4
5
6
7
6a, Ph
6b, TMS
6c, I
6d, H
6e, Bu
5a, 30%
—
—
—
5e, 5%
5f, 40%
5g, 5%
6f, p-MeOPh
6g, p-CF3Ph
Scheme 2. Synthesis and [2+2+2] cycloaddition of triyne ether 2.