Scheme 4. Total Synthesis of (-)-Kainic Acid Using a Pd-Catalyzed Zn-ene Cyclization of an Allyl Chloride
clized material (e.r. ) 2:1).18 This is likely due to epimer-
ization of the configurationally labile R-amino aldehyde
intermediate obtained after DIBAL-H reduction of ester 5.19
While the synthesis of 1 was accomplished in the formal
sense and the suitability of the key cyclization established,
there are several limitations to this first generation route.
First, while unhindered amines are known to add smoothly
to diene 2,20 D-serine methyl ester does not and an equivalent
of chiral starting material is wasted because it is needed in
excess. Second, the reduction-olefination route to 6 and the
acylation route to 8 are relatively low yielding. Third, while
the cyclization gives the correct relative stereochemistry, 8
was not optically pure. A different strategy was therefore
necessary if the route was to be amenable to a scalable,
enantiopure synthesis. Addressing these limitations, and
motivated by the high diastereoselectivity of the zinc-ene
cyclization, we revised our approach.
Returning to 3, protection was carried out by reductive
benzylation and TBS protection to give 10 quantitatively
(Scheme 4).17 Chemoselective allylation with chloro bromide
11,21 generated in two steps from isoprene, proceeded
smoothly to provide allyl chloride 12. While we originally
set out to use 12 as a substrate for sulfonylation by sodium
benzenesulfinate to obtain the allylic isomer of 5, it was
apparent that an allyl chloride would be suitable in the
cyclization,11a provided it could survive conversion of the
methyl ester of 12 to the terminal olefin. Gratifyingly, the
chloride survived such a manipulation and chemoselective
reduction with LiBH4 provided alcohol 13 in high yield.
Subsequent Swern oxidation and Wittig olefination provided
diene 14 in 85% yield. Unfortunately, while it was possible
to carry out the oxidation and olefination on a large scale
(up to 30 g), significant racemization was again observed.22
This result was disappointing in light of instances where
Swern oxidations of very similar substrates were carried out
with little or no epimerization reported.17,23,24
Although 14 was not optically pure, the ease in accessing
large amounts of the material facilitated investigation of the
key step as well as a revised end game. Thus, with cycliza-
tion substrate 14 in hand, we subjected the allyl chloride to
Pd(PPh3)4 and ZnEt2 in a mixture of diethyl ether and hexane.
After the mixture had been stirred at room temperature for
42 h and quenched with iodine, pyrrolidine 7 was obtained
in 91% yield as a single diastereomer.25
While both the allylic sulfone 6 and allylic chloride 14
undergo cyclization with complete diastereoselectivity, the
yield and mild conditions of the chloride approach provide
a distinct advantage over the sulfone strategy.26 Furthermore,
the cyclization of 14 could be carried out on a 10 g scale
without significant deterioration in stereoselectivity and yield.
(15) Other mechanisms of quenching, such as 1,5 proton transfer to give
an allyllithium, were ruled out by D2O quenching after lithiation. For an
example of such a proton transfer, see: Cheng, D.; Zhu, S.; Liu, X.; Norton,
S.H.; Cohen, T. J. Am. Chem. Soc. 1999, 121, 10241. Standard precautions
to prevent the formation of 9 (e.g., reverse quench) were ineffective. Excess
tBuLi (>5 equiv) improved the yield of 8 slightly, but this measure is
impractical for a scalable process.
(16) Our original plan, designed to both facilitate the cyclization and
save steps, was to use the conjugated unsaturated methyl ester as the
enophile, readily prepared by using the suitable Wittig reagent, instead of
the alkene in 6. However, the result was a surprise. The cyclization was
considerably more sluggish, requiring an elevated temperature, and the major
product had a trans C3-C4 relationship. The sluggish cyclization suggests
that this Zn-ene reaction should be considered an ambiphilic rather than a
nucleophilic addition. The origin of the trans selectivity is not yet fully
understood, but optimizing this transformation provides a route to other
kainoid stereoisomers.
(21) Lee, J.; Jeong, Y.; Ji, M.; Baik, W.; Lee, S.; Koo, S. Synlett 2004,
1937. Apparently, a high E/Z ratio is critical in the synthesis of 12. When
batches of 11 with low E/Z were used, yields diminished to 80-90%. We
suspect that intramolecular ammonium formation of the Z isomer is
responsible for the reduced yield in these cases.
(22) Direct determination of the optical purity of 14 was complicated
by a minor amount of Z olefin. An e.r. of 1.6:1.0 was inferred by chiral
HPLC after the cyclization (see Supporting Information). Interestingly, the
specific rotation of known compound 8, when synthesized through the
chloride route, was nearly identical to the literature value where a Swern
oxidation on a similar amino alcohol substrate was used: [R]D ) -29.1°
(c ) 1.16) vs [R]D ) -27.2° (c ) 0.99) in ref 17. We consider the optical
purity of this material suspect.
(17) Barco, A.; Benetti, S.; Spalluto, G. J. Org. Chem. 1992, 57, 6279.
(18) See Supporting Information for details.
(19) For a review discussing the inherent configurational lability of
R-amino aldehydes, see: Gryko, D.; Chalko, J.; Jurczak, J. Chirality 2003,
15, 514.
(20) (a) Ba¨ckvall, J.-E.; Juntunen, S. J. Am. Chem. Soc. 1987, 109, 6396.
(b) Ref 11d.
(23) Martinez, M. M.; Hoppe, D. Org. Lett. 2004, 6, 3743.
(24) It is conceivable that silyl transfer in 13 contributes to racemization.
However, such a transfer in TBS-protected serinols was not observed in
several similar substrates. See: (a) Novachek, K. A.; Meyers, A. I.
Tetrahedron Lett. 1996, 37, 1743. (b) La¨ıb, T.; Chastanet, J.; Zhu, J. J.
Org. Chem. 1998, 63, 1709. (c) Jurczak, J.; Gryko, D.; Kobrzycka, E.;
Grunza, H.; Prokopowicz, P. Tetrahedron 1998, 54, 6051.
Org. Lett., Vol. 9, No. 19, 2007
3827