Communications
that were screened to convert either 1 or 11 into either 3 or 4
With a method in hand for the desired oxidative
conversion, we turned our attention to stephacidin A (1)
once again. Gribble reduction of synthetic (+)-1 furnished
indoline 16 (Scheme 3) in essentially quantitative yield as a
separable but inconsequential mixture of diastereomers. This
mixture was subjected to Somei oxidation, which unfortu-
nately provided about 20% yield of (+)-3 mixed with some
inseparable impurities. Alternatively, indoline 16 could be
were similarly unsuccessful. These shortcomings forced a
reevaluation of our planned pathway to avrainvillamide.
In 1971, Somei put forth a provocative hypothesis for the
role of 1-hydroxyindoles (tautomers of saturated indolic
nitrones) in the biosynthesis and functionalization of indole
[
12]
alkaloids in nature.
These highly reactive species are
susceptible to nucleophilic attack and dimerization, and
undergo a variety of interesting rearrangements. These
pioneering studies led us to hypothesize that perhaps such a
species would be a viable precursor to 3. As a proof of
principle, model compound 12 was synthesized by a route that
[
15]
treated with catalytic SeO2 and excess H O to provide pure
2
2
(+)-3 in 27% isolated yield along with 50% recovered 16
(spectroscopically identical to the samples obtained from
[
6]
Prof. Zeeck and Prof. Fenical and that reported by Myers;
synthetic (+)-3: [a] = + 11 (c = 0.1, CHCl ); natural (+)-3:
[
13]
paralleled our synthesis of 1.
As shown in Scheme 3,
D
3
chemoselective reduction of the indole C2ꢀC3 p bond with
[a] = + 10.6 (c = 0.17, CHCl ). We speculate that this cas-
D
3
sodium cyanoborohydride in acetic acid (Gribble reduc-
cade oxidation proceeds via the putative intermediate 1-
hydroxyindole 17, which is further oxidized directly to 3 or
perhaps first to 4 (Figure 1) followed by loss of water to form
(+)-3.
[
14]
tion)
gave indoline 13 (53% yield), poised for Somei
[
12]
oxidation. Treatment of 13 with catalytic Na WO ·2H O
2
4
2
and excess H O did not lead to appreciable amounts of the
2
2
expected 1-hydroxyindole 14. Instead, we were pleased to
find that the major constituent in the crude reaction mixture
was the bright yellow a,b-unsaturated nitrone 15 isolated in
approximately 30% yield (unoptimized).
In accord with Herzon and Myersꢀ observations in the
[
6]
unnatural series, synthetic (+)-3 underwent spontaneous
dimerization to (ꢀ)-2 under a variety of conditions, including
exposure to silica gel (during preparative TLC), base
[
6]
(
(
Et N), or even simple evaporation from DMSO (synthetic
3
ꢀ)-2 was spectroscopically identical to a sample obtained
[
2a]
[6]
from BMS and to that reported by Myers; optical rotation
of synthetic (ꢀ)-2: [a] = ꢀ33 (c = 0.1, CDCl ); natural (ꢀ)-2
D
3
(
2
as received from BMS): [a] = ꢀ21.1 (c = 0.19, CDCl ); (+)-
D
3
[
6]
:
[a] = + 91 (c = 0.25, CH CN)). The ease with which the
D 3
dimerization took place actually hampered purification of 3.
Likewise, 2 underwent facile retrodimerization back to a
mixture of 3 and 2 during chromatography. A final issue that
[
2d]
needed to be addressed was the true identity of CJ-17,665 as
slight differences between synthetic 3 and the reported
1
H NMR spectra of CJ-17,665 were observed by both us and
[
6]
Herzon and Myers. Comparison (LC-MS, TLC, NMR
spectroscopy) with an authentic sample from Pfizer confirms
that it is indeed 3, and, perhaps not surprisingly, the sample
contained approximately 20% of stephacidin B (2) as judged
by H NMR spectral analysis and LC-MS. Interestingly, the
sample from Pfizer was provided as a (yellow) solution in
DMSO, whereas the sample from Professor Zeeck (see above)
was a yellow–green powder and contained no stephacidin B (as
1
[16]
1
judged by H NMR spectroscopy), which implies that dimeri-
zation does not occur over time in the solid state.
The spontaneous (and reversible) dimerization of 3 to 2 is
consistent with the known tendency of saturated indolic
nitrones (a tautomeric form of a 1-hydroxyindole) to dimer-
[
12]
ize. Taken together, these findings add further support for
Someiꢀs hypothesis regarding the potentially widespread
significance of fleeting 1-hydroxyindoles in nature. The new
selenium- and tungsten-based protocols reported herein to
chemoselectively generate an unsaturated nitrone group from
an easily accessible indoline should facilitate the synthesis of
avrainvillamide and stephacidin mimics for biological explo-
Scheme 3. Synthesis of simple avrainvillamide model 15 and the
successful conversion of (+)-1 into (+)-3 and (ꢀ)-2. Reagents and
conditions: a) NaBH CN (10 equiv), AcOH, 258C, 12 h, 53%;
3
b) Na WO ·2H O (0.2 equiv), aq. 35% H O (50 equiv), MeOH, H O,
2
4
2
2
2
2
2
58C, 6 h, ca. 30%; c) NaBH CN (50 equiv), AcOH, 258C, 24 h, 93%;
[16]
3
rations.
d) SeO (0.25 equiv), 35% H O (50 equiv), dioxane, 258C, 40 h, 27%
2
2
2
3
with 50% recovered 16; e) Procedure A: Preparative TLC (SiO ,
2
[
6]
EtOAc); Procedure B: Et N (excess), CH CN, 258C, 1 h; Procedure C:
Received: February 22, 2005
Revised: March 30, 2005
Published online: May 18, 2005
3
3
DMSO, then solvent removal, approx. 2:1 mixture of 3 to 2, purified by
preparative TLC. PMB=p-methoxybenzyl; DMSO=dimethyl sulfoxide.
3
894
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 3892 –3895
Baran, Phil S.
