Organic Letters
Letter
catalyzed by copper−phosphoramidite complexes, and the
alkyl zirconocenes are generated in situ from alkenes and
Cp2ZrHCl via hydrometalation. Alkylzirconium reagents, when
used in this fashion, can mitigate the aforementioned
shortcomings of traditional nucleophiles because they are
more tolerant toward functional groups. We hoped to use an
appropriate fragment in the synthesis of the taxol core, but the
precise choice of the precursor alkene was not immediately
obvious. Also, it is more common than not that ACAs are
sensitive to the solvent, temperature, concentration, method of
addition, and presence of additives.5
both been used extensively in similar transformations6a,7 but
gave poor enantioselectivities for our system. To our delight,
L3,6a developed in our group and used in other challenging
transformations,8e,11−13 was found to give the best yield/ee
combination and is easily prepared. The investigation of
different counterions (BF4, SbF6, OTf, PF6, ClO4) indicated
that NTf2 gave superior yields and enantioselectivity. Previous
experiments in the group showed that using dichloromethane
as a solvent was often beneficial. The use of a chlorinated
cosolvent gave better enantioselectivity but lower yields. With
hydrocarbon or ethereal cosolvents, higher yields but lower ee
values were observed.
Upon scale-up (to 3 mmol of 6), the reaction performed
better, and the product could be obtained in 96% yield with
88% ee. ACAs of alkylzirconium nucleophiles often work better
when scaled up; this effect is attributed to the fact that it is
easier to measure and mix the reaction components on larger
scales, and larger scales minimize the impact of trace air and
moisture.
In an approach that would have closely follow Baran’s route
to the core, conjugated triene 59 was added to 6 (Scheme 2).
Scheme 2. Preparation of Intermediate 7 Using ACA of
a
Triene 5
Having demonstrated that 9 could reliably be added to 6, we
next turned our attention to the preparation of a suitable
intermediate for the synthesis of the desired product. The
trapping of zirconium enolates is a challenging problem,14 and
we recently reported trapping reactions using the Vilsmeier−
Haack reagent to give β-chloroaldehydes from our ACA
zirconium enolates.15 It was found that this trapping protocol
also worked for our substrate; using 16.5 mmol of 6, 3.6 g of
11 (corresponding to 69% yield) could be obtained with 92%
ee, requiring minimal modification (Scheme 3, steps a and b).
a
(a) Vinylmagnesium bromide (1.3 equiv), ZnBr2 (2.0 equiv), PdCl2·
dppf (0.020 equiv), THF, 0 °C to rt, 6.5 h, 49%. (b) ee determined by
SFC or HPLC. Abbreviations: dppf = 1,1′-bis(diphenylphosphino)-
ferrocene.
Despite extensive optimization, 7 could be obtained in only
14% yield with 80% ee. Although disappointing, no side
products that would result from hydrozirconation at the di- or
tetrasubstituted olefins were observed, and we postulated that
the low yield was due to the size of the polyene.
In our search for an alkene with a smaller steric profile, we
encountered bromodiene 910 first reported by Takahashi and
coworkers. 9 was prepared from 3-methylcrotonaldehyde via
bromination to the corresponding bromoaldehyde followed by
Peterson olefination and could easily be made on a decagram
The ACA of 9 to 6 was examined, and it was found that the
phosphoramidite ligand has a tremendous effect on the
enantioselectivity of this reaction (Table 1). L1 and L2 have
a
Scheme 3. Completion of the Synthesis
Table 1. Optimization of Key ACA Conjugate Addition
Step
a
a
(a) Cp2ZrHCl (1.9 equiv), CuCl/L3 (0.08 equiv), DCE/CH2Cl2, rt,
17 h; then, POCl3/DMF (10 equiv), DCE, 60 °C, 1 h, 69%, 92% ee.
(b) Vinylmagnesium bromide (1.2 equiv), Et2O, 0 °C, 1 h, 87%. (c)
Isopropenylboronic acid pinacol ester (1.1 equiv), PdCl2(PPh3)2/
dppf (0.10 equiv), K3PO4 (3.0 equiv), DMF, 65 °C, 17 h, 59%. (d)
Dess−Martin periodinane (1.2 equiv), H2O (0.50 equiv), CH2Cl2, rt,
30 min, undesired/desired 1:10, 90%. (e) TiCl4 (1.1 equiv), slow
addition, CH2Cl2, −35 °C, 6 h, 35%, 1:1 d.r., 92% ee. Abbreviations:
DCE = 1,2-dichloroethane, dppf = 1,1′-bis(diphenylphosphino)-
ferrocene.
b
c
To complete the synthesis, allylic alcohol 12 was obtained
from the 1,2-addition of vinylmagnesium bromide to 11. Then,
standard Suzuki−Miyaura conditions with isopropenyl boronic
acid pinacol ester gave triene 13 in 59% yield.16 The final two
steps of the synthesis proved to be problematic. Triene 13 was
found to be unstable under various oxidation conditions. (See
use of Dess−Martin periodinane with one equivalent of water
gave good results. Without water, an allylic transposition
product was observed in appreciable quantities.17
entry
L
cosolvent
yield (%)
ee (%)
1
2
3
4
5
1
2
3
3
3
MTBE
MTBE
MTBE
DCE
70
63
81
45
71
6
18
82
90
88
PhMe
a
Reactions were performed on a 0.5 mmol scale using 2.4 equiv of 9.
Isolated yields. ee values were determined by SFC or HPLC analysis
b
c
using a chiral nonracemic stationary phase.
B
Org. Lett. XXXX, XXX, XXX−XXX