The stereoselectivity observed might well be ascribed to
equatorial attack by the bulky Cp2TiCl species against 8 to
give 9a (Scheme 2). The intermediate 9a should then undergo
any case, the minor amount of 1314 obtained (8%) indicated
that the participation of a carbocationic process was con-
siderably restricted.15 Thus, 16 becomes the first example
of a new type of reagent that is both compatible with
epoxides and capable of regenerating Cp2TiCl2 from Cp2-
Ti(Cl)H and (as we will see later) acetoxy-titanium deriva-
tives. The above results reveal the versatile nature of the
catalytic procedure, which can be controlled to afford either
reduction products or alkenes by adding or excluding water
and using the additives 15 or 16, respectively. We subse-
quently diminished the catalyst proportions to determine the
minimum required to guarantee the radical nature of the
process. With 0.05 equiv (entry 5), 11a was still the main
product, while in the complete absence of Cp2TiCl2 (entry
6), the mixture containing diol 1414 obtained was character-
istic of a purely carbocationic process, confirming the crucial
role played by the titanium catalyst.
Scheme 2. Proposed Mechanism for the Formation of 10a and
12a
a R ) OTi(Cl)Cp2; (a) hydrolysis, (b) â-hydrogen elimination.
To explore the synthetic usefulness of the method, we
chose as targets two eudesmanolides belonging to different
subfamilies: (+)-9â-hydroxyreynosin (17) and (+)-â-cy-
clopyrethrosin (2). 12,6-Eudesmanolide 17 and its derivative
19 were isolated from Inula heterolepis and Artemisia herba-
alba, respectively, and their structures were established by
spectroscopic techniques.16 We started their synthesis with
5, which, after selective hydrogenation and treatment with
MCPBA, afforded oxirane 18 (Scheme 3). Subsequent
hydrolysis to produce 10a, which would in turn give 11a
after acidic quenching. The alkyl-TiIV complex 9a could
also account for the minor quantity of alkene 12 derived from
a relatively slow â-hydrogen elimination or any other process
generating Cp2Ti(Cl)H.11 This possibility led us to take
advantage of this process to synthesize eudesmanolides with
an exocyclic ∆4,15 double bond.12 Treatment of 7 in the
absence of water, however, gave a mixture containing a
relatively high proportion (60%) of the reduction product
11a (Table 1, entry 2), presumably due to the protonolysis
of the C-Ti bond of 9a by the hydrochloride 15 employed.13
Therefore, we assayed a novel reagent, the aprotic derivative
16 (prepared in situ by mixing collidine and Me3SiCl), to
regenerate Cp2TiCl2 and prevent protonolysis.
Scheme 3. Biomimetic Synthesis of 17 and 19
Treatment of 7 under these new conditions (entry 3) gave
a substantially increased proportion of the exocyclic alkene
12 (74%) and a significant decrease (down to 18%) in the
hydrolysis product 11a (which might derive from adventi-
tious water). The regeneration of the catalyst might be
rationalized by reaction between 16 and the â-elimination
product Cp2Ti(Cl)H, giving Cp2TiCl2 together with collidine
and gaseous Me3SiH, but in the absence of experimental
evidence, alternative explanations cannot be discarded. In
(9) (a) Fu¨rstner, A.; Hupperts, A. J. Am. Chem. Soc. 1995, 117, 4468-
4475. (b) Gansa¨uer, A.; Bluhm, H.; Pierobon, M. J. Am. Chem. Soc. 1998,
120, 12849-12859. (c) Hansen, T.; Daasbjerg, K.; Skrydstrup, T. Tetra-
hedron Lett. 2000, 41, 8645-8649.
titanocene-catalyzed cyclization of 18 gave alkene 19. In this
case, the â-elimination of the C-15 hydroxyl group was much
faster than â-hydrogen elimination from 9a, and neither
anhydrous conditions nor 16 were required. Finally, the
conjugated double bond was restored in a single-pot reaction
applying a slight modification of Grieco’s method.17 Thus,
we obtained 17 in four steps at an overall yield of 31%. The
spectroscopic properties of synthetic compounds 17 and 19
were in agreement with those of the natural products.16
The 12,8-eudesmanolide 2 was isolated from pyrethrum
flowers, and its structure was established by spectroscopic
(10) Ogura, M.; Cordell, G. A.; Farnsworth, N. R. Phytochemistry 1978,
17, 957-961.
(11) Despite the fact that we could not isolate Cp2Ti(Cl)H, previous
results suggest that it might be formed from certain alkyl-TiIV complexes
and possibly displays reductive properties similar to those of Cp2Zr(Cl)H
(the Schwartz reagent).4b For a theoretical study on the reactivity of Cp2-
Ti(Cl)H, see: Sakai, S. J. Mol. Struct.: THEOCHEM 2001, 540, 157-
169.
(12) There are more than 170 natural eudesmanolides described contain-
ing a ∆4,15 double bond.1
(13) As 15 is more acidic than H2O, the different results summarized in
entries 1 and 2 of Table 1 might seem somewhat anomalous. To rationalize
these results and the counterintuitive sequence given in Scheme 1, one of
the referees suggested that water could act as a very good ligand for Ti
that would increase the reduction potential and diminish the tendency toward
â-elimination.
(16) (a) Bohlmann, F.; Ates, N.; Grenz, M. Phytochemistry 1982, 21,
1166-1168. (b) Ahmed, A. A.; Abou-El-Ela, M.; Jakupovic, J.; Seif-El-
Din, A. A.; Sabri, N. Phytochemistry 1990, 29, 3661-3663.
(17) For the two-step original method, see: Grieco, P. A.; Nishizawa,
M. J. Org. Chem. 1977, 42, 1717-1720.
(14) Ando, M.; Takase, K. Tetrahedron 1977, 33, 2785-2789.
(15) Compound 13 derives from the acid-induced cyclization of 7.4b
Org. Lett., Vol. 5, No. 11, 2003
1937