J. Thibonnet, S. Inack Ngi et al.
halogen–carbon bond, and the stereochemistry of the exocy-
clic double bond was found to be exclusively Z with R1 dif-
ferent from a hydrogen atom. The reaction worked well
with various substituents (Table 1). These findings were con-
firmed by X-ray diffraction analysis, performed on buteno-
lides 4a and 4k.[19] When R=H, mixtures of Z/E isomers
were produced in various ratios, but the Z isomer was
always the major product (Table 1, entries 8–10 and 18).
Some additives were tested to accelerate the transforma-
tion and decrease the amount of copper catalyst. Bidendate
ligands, which are known to improve the catalytic reactivity
of copper,[20] were found to be detrimental. In our case, the
use of 2-pyridylimine or 1,3-diketones dramatically de-
creased yields. The conditions were mild and a relatively
short exposure time was needed (4 h average for 5 mmol of
acid) to obtain the desired butenolides. Leaving the reaction
running for more than 24 h, induced a-dehalogenation (up
to 50% for 2a and 10% for 3a as starting materials).
and, by very close proximity, could favour the insertion step
into the b-carbon–halogen bond. We assume that the pres-
ence of the internal carboxylate ligation of copper could
also prevent the b-elimination side reactions.
Having established an easy set of conditions for the for-
mation of a-halo-g-alkylidenebutenolides, we turned our at-
tention to the synthesis of additional derivatives, by exploit-
ing the reactivity of the remaining carbon–halogen bond.
The classical Suzuki cross-coupling reaction[23] was therefore
applied to 4a and 4k, affording a,b-substituted g-alkylidene-
butenolides 5 (5a from 4a and 5b–d from 4k) in good
yields (Scheme 4).
As reported by Rossi,[11b,c] we found that with classic So-
nogashira coupling conditions without protecting the car-
boxylic function, the reaction of our reference system did
not provide the desired product. In fact, with palladium cat-
alysts ([PdACHTUNGTRENNUNG(PPh3)4] or [PdACHUTNTGREN(NGUN CH3CN)2Cl2]) we obtained diphe-
nylacetylene and tetrolic acid, clearly indicating the palladi-
um insertion into one of the two carbon–halogen bonds, fol-
lowed by a b-elimination step. Compared to literature re-
sults,[20,21] we noted no side reaction, such as a Sonogashira-
or Castro–Stephens-like process, on the remaining a-halo-
gen, even with a large excess of alkyne.
Scheme 4. Suzuki and Negishi coupling routes from selected a-halo-g-al-
kylidenebutenolides to a-arylated or a-alkylated g-alkylidenebutenolides.
1
ꢀ
Suzuki coupling: R B(OH)2, Na2CO3, [Pd
EtOH, 70 8C, 5h (4a!5a; 4k!5b, 5c, 5d). Negishi coupling: MeZnBr,
G
PdCl
N
The scope of the reaction was wide and the transforma-
tion tolerated several functions associated with the alkyne,
such as ether and acetal. Various original a-halobutenolides
were prepared and the results are summarised in Table 1. In
each case, no 6-endo-dig cyclisation was detected. Interest-
ingly, with 1,7-octadiyne or 1,4-diethynylbenzene, the reac-
tion was found to be very selective for only one of the triple
bonds (Table 1, entries 2, 10 and 12) and only traces of the
dimeric lactone were detected. Enynes also reacted effi-
ciently (Table 1, entries 3, 4 and 13) without isomerisation of
the internal double bonds. Interestingly, the reaction was
also found to be effective with an alkyne bearing a ferrocen-
yl substituent (Table 1, entry 17). On the other hand, no re-
action occurred with alkynes, functionalised by aliphatic
amines, carboxylic acids or a metal (Bu3Sn, Me3Ge), directly
connected to the triple bond.[16] Experiments are underway
to show the tolerance of this copper-catalysed reaction to-
wards alkynes, bearing a metal distant from the triple bond.
Early results seem to show that this copper-catalysed reac-
tion may lead to organometallic butenolides.[22] When ap-
plied to a,b-dihalobut-2-ynoic acid, the defined conditions
constitute a simple access to a-halobutenolides, as compared
to the routes using Pd-catalysed cross-coupling or (and) Pd-
catalysed oxacyclisation,[11] which needed two additional
steps of protection–deprotection of the carboxylic function.
The greater efficiency of copper with respect to palladium
in this process could be explained by the role of the carbox-
ylate function, which could certainly act as a ligand of CuI
The readiness of this sequence (copper-catalysed reaction
associated with a Pd-catalysed cross-coupling reaction) rep-
resents a new and efficient route to polysubstituted g-alkyli-
denebutenolides and suggests interesting outcomes. To illus-
trate the scope of this methodology, we proposed this se-
quence for the preparation of cyclised analogues of cis-reti-
noic acid (isotretinoin) and natural butter flavour bovolide.
cis-Retinoic acid belongs to the retinoids family that in-
cludes one of the existing vitamin A forms in the human
body. Vitamin A and its derivatives are used as treatments,
in particular in anticancer therapy.[24] We therefore consid-
ered, whether these products could be formed by our
copper-catalysed tandem coupling–heterocyclisation, as
shown in the following retrosynthetic pathway (Scheme 5).
Scheme 5. Retrosynthesis of cyclic analogues of cis-retinoic acid.
13694
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 13692 – 13696