Synthesis of (-)-7-Epiaustraline and (-)-1-Epicastanospermine

Article abstract of DOI:10.1021/jo991990d

Highly efficient and selective syntheses of the title compounds are described. The cornerstone of the synthetic plan is the tandem inter [4 + 2]/inter [3 + 2] cycloaddition process. These syntheses differ from previous applications of this strategy in that they incorporate an alkylation in the hydrogenolysis step to close the second ring of the azabicyclic systems. Notable features of the sequence are (1) the highly regio- and stereoselective [3 + 2] cycloaddition of nitronate 15 with siloxymethyl (Z)-β-silylvinyl ketone (Z)-22b and (2) the highly selective reduction of the resulting ketone 24a with L-Selectride. A single-crystal X-ray structure analysis of synthetic (-)-7-epiaustraline confirmed that the targeted structure was successfully synthesized. This stimulated a reexamination of the structural assignment of the natural product. (-)-1-Epicastanospermine was synthesized in four steps from the common intermediate 27a. The absolute configuration of (-)-1-epicastanospermine was assured by single-crystal X-ray structure analysis of intermediate (-)-27a. Thus, the sign of the optical rotation had to be revised. The overall efficiency of these syntheses were 9 steps and 23% yield for (-)-7-epiaustraline and 10 steps and 20% yield for (-)-1-epicastanospermine.

Full text of DOI:10.1021/jo991990d

J . Org. Chem. 2000, 65, 2887-2896
2887
Syn th esis of (-)-7-Ep ia u str a lin e a n d (-)-1-Ep ica sta n osp er m in e
Scott E. Denmark* and B. Herbert
Roger Adams Laboratory, Department of Chemistry, University of Illinois, Urbana, Illinois 61801
Received December 23, 1999
Highly efficient and selective syntheses of the title compounds are described. The cornerstone of
the synthetic plan is the tandem inter [4 + 2]/inter [3 + 2] cycloaddition process. These syntheses
differ from previous applications of this strategy in that they incorporate an alkylation in the
hydrogenolysis step to close the second ring of the azabicyclic systems. Notable features of the
sequence are (1) the highly regio- and stereoselective [3 + 2] cycloaddition of nitronate 15 with
siloxymethyl (Z)-â-silylvinyl ketone (Z)-22b and (2) the highly selective reduction of the resulting
ketone 24a with L-Selectride. A single-crystal X-ray structure analysis of synthetic (-)-7-
epiaustraline confirmed that the targeted structure was successfully synthesized. This stimulated
a reexamination of the structural assignment of the natural product. (-)-1-Epicastanospermine
was synthesized in four steps from the common intermediate 27a . The absolute configuration of
(-)-1-epicastanospermine was assured by single-crystal X-ray structure analysis of intermediate
(-)-27a . Thus, the sign of the optical rotation had to be revised. The overall efficiency of these
syntheses were 9 steps and 23% yield for (-)-7-epiaustraline and 10 steps and 20% yield for (-)-
1-epicastanospermine
In tr od u ction
Alexin es a n d Au str a lin es. In 1988, a colorless, cubic
solid was isolated from the aqueous ethanolic extracts
of finely ground, dried, pod of Alexa leiopetala Sandwith.1
Single-crystal X-ray analysis revealed that the isolated
compound, (+)-alexine ((+)-1), was a pyrrolizidine bear-
ing a hydroxymethyl group adjacent to the ring nitrogen
[C(3)], Figure 1. The isolation of alexine was a significant
departure from the broad class of pyrrolizidine alkaloids
which bear carbon substituents at C(1).2 Furthermore,
the structural similarity to DMDP (11) suggested that
alexine may also possess glycosidase inhibitory activity.
Inhibition studies using â-glucosidase and â-galactosi-
dase enzymes from mouse small intestines and â-glu-
cosidase from Penicillium expansum revealed that alex-
ine was indeed a glucosidase inhibitor, although weaker
than DMDP.3 Alexine was also shown to inhibit thioglu-
cosidase enzymes isolated from Brevicoryne brassiae.11
(1) Nash, R. J .; Fellows, L. E.; Dring, J . V.; Fleet, G. W. J .; Derome,
A. E.; Hamor, T. A.; Scofield, A. M.; Watkin, D. J . Tetrahedron Lett.
1988, 29, 2487.
F igu r e 1. (+)-Alexine, (+)-australine, and related alkaloids.
(2) (a) Wrobel, J . T. Pyrrolizidine Alkaloids. In The Alkaloids; Brossi,
A., Ed.; Academic Press: New York, 1985; Chapter 7. (b) Robins, D. J .
Pyrrolizidine Alkaloids. Prog. Chem. Org. Nat. Prod. 1982, 41, 115.
(c) Dai, W.-M.; Nagao, Y.; Fujita, E. Heterocycles 1990, 30, 1231. (d)
Robins, D. J . Nat. Prod. Rep. 1993, 487 and references to earlier
reviews. (e) Neuberger, A.; Tatum E. L. Frontiers of Biology: The
Pyrrolizidine Alkaloids; North-Holland Publishing Company: Amster-
dam, 1968; Vol. 9. (f) Casiraghi, G.; Zanardi, F.; Rassu, G.; Pinna, L.
Org. Prep. Proc. Int. 1996, 643. (g) For a review of the reported
syntheses of hydroxylated pyrrolizidine alkaloids see: Giovanni, C.;
Zanardi, F.; Rassu, G.; Pinna, L. in Advances in the Stereoselective
Synthesis of Hydroxylated Pyrrolizidines.
Shortly after the isolation of alexine, a second isomer
was isolated from the seeds of Castanospermum aus-
trale.4 (+)-Australine ((+)-2) was found to be a C(7a)
epimer of alexine by X-ray crystallographic analysis.
Enzyme inhibition studies indicated that the alkaloid was
a potent inhibitor of amyloglucosidase4 and the glyco-
protein-processing enzyme glucosidase I but a weak
inhibitor of glucosidase II.5 No inhibition of â-glucosidase,
(3) Scofield, A. M.; Rossiter, J . T.; Witham, P.; Kite, G. C.; Nash, R.
J .; Fellows, L. E. Phytochemistry 1990, 29, 107.
(4) Molyneux, R. J .; Benson, M.; Wong, R. Y.; Tropea, J . E.; Elbein,
A. D. J . Nat. Prod. 1988, 51, 1198.
(5) Tropea, J . E.; Molyneux, R. J .; Kaushal, G. P.; Pan, Y. T.;
Mitchell, M.; Elbein, A. D. Biochemistry 1989, 28, 2027.
(6) Nash, R. J .; Fellows, L. E.; Plant, A. C.; Fleet, G. W. J .; Derome,
A. E.; Baird, P. D.; Hegarty, M. P.; Scofield, A. M. Tetrahedron 1988,
44, 5959.
(7) Nash, R. J .; Fellows, L. E.; Dring, J . V.; Fleet, G. W. J .; Girdhar,
A.; Ramsden, N. G.; Peach, J . M.; Hegarty, M. P.; Scofield, A. M.
Phytochemistry 1990, 29, 111.
(8) Harris, C. M.; Harris, T. M.; Molyneux, R. J .; Tropea, J . E.;
Elbein, A. D. Tetrahedron Lett. 1989, 30, 5685.
(9) Nash, R. J .; Thomas, P. I.; Waigh, R. D.; Fleet, G. W. J .; Wormald,
M. R.; Lilley, P. M.de Q.; Watkin, D. J . Tetrahedron Lett. 1994, 35,
7849.
(10) Pereira, A. C. de S.; Kaplan, M. A. C.; Maia, J . G. S.; Gottlieb,
O. R.; Nash, R. J .; Fleet, G. W. J .; Watkin, D. J .; Scofield, A. M.
Tetrahedron Lett. 1991, 47, 5637.
(11) Fleet, G. W. J .; Haraldsson, M.; Nash, R. J .; Fellows, L. E.
Tetrahedron Lett. 1988, 29, 5441.
10.1021/jo991990d CCC: $19.00 © 2000 American Chemical Society
Published on Web 03/23/2000

2888 J . Org. Chem., Vol. 65, No. 10, 2000
Denmark and Herbert
R- or â-mannosidase, and R- or â-galactosidase was
observed.5 In the next few years, several additional
isomers of alexine and australine were isolated. Among
them were (+)-3-epiaustraline ((+)-3),6 (+)-7-epiaustra-
line ((+)-4),7 (+)-1-epiaustraline ((+)-5),7,8 (+)-casuarine
((+)-6),9 and the related amino acid 7a-epialexaflorine
(7).10 During the course of studies on the synthesis of
several of those natural products, additional epimers
were created as a consequence of the synthetic routes
chosen. Among them, synthetic 3-epialexine (8),11 7-epi-
alexine (9),12 and 1,7-diepiaustraline (10)13 were pro-
duced. The most profound property that all of the
members of this class of alkaloids possess is glucosidase
inhibition to varying degrees. For example, a comparative
study of inhibition of the glucan 1,4 R-glucosidase hy-
drolysis of potato amylose among the naturally occurring
pyrrolizidine alkaloids 1-5 indicated that each isomer
was a potent inhibitor.7 Relative to castanospermine (IC50
) 1.5 × 10-6 M), a potent glucosidase inhibitor, (+)-
alexine ((+)-1) was found to be the weakest (IC50 ) 1.1
× 10-5 M) and 7-epiaustraline (4) the strongest (IC50
)
1.3 × 10-7 M).
F igu r e 2. Castanospermine and related diastereoisomers.
To date, seven of the alexine epimers have been
prepared synthetically by a variety of approaches. 7-Epi-
alexine (9) and (+)-7-epiaustraline ((+)-4) have been
prepared from L-xylose by the reductive cyclization of
azido epoxides.12 The formation of alkaloid (+)-4 was
surprising since (+)-australine ((+)-2) was the initial
target. Since comparison of spectra for synthetic 2
matched only that reported for isolated (+)-4, the authors
were forced to conclude that an unusual inversion at
C(7) had occurred in the epoxide opening. (+)-1-Epiaus-
traline ((+)-5) and 1,7-diepiaustraline (10) were synthe-
sized from heptonolactones by the reductive cyclization
of azido epoxides.13 Australines (+)-5 and 10 were also
synthesized from pyroglutamic acid.14 Alexine ((+)-1),
3-epialexine (8), and 7-epialexine (9) were synthesized
from D-glucose through the intermediacy of methyl
2-azido-3-O-benzyl-2-deoxy-R-mannofuranoside.11 Through-
out those approaches, two main synthetic strategies are
common; the formation of pyrrolidine rings by reductive
cyclizations of azido epoxides and annulation by N-al-
kylation. Common to all syntheses was the use of sugars
or sugar-derived building blocks as starting materials.
Ca sta n osp er m in e a n d Rela ted Dia ster eoisom er s.
Castanospermine was first isolated as a cubic, crystalline
solid from the aqueous ethanolic extracts of finely ground
immature seeds of Castanospermum australe.15 Extensive
2-D NMR analysis revealed that the isolated compound
was a tetrahydroxyindolizidine. The finding was con-
firmed by an X-ray crystal analysis which also estab-
lished the relative configuration of all five stereogenic
centers in (+)-castanospermine ((+)-12). The structural
elucidation of castanospermine also revealed that the
alkaloid bore a strong resemblance to both (+)-australine
((+)-2) and D-glucose. Not surprisingly, the absolute
configuration was determined by a synthesis of the
alkaloid from D-glucose to be as depicted in Figure 2.16
Since then, numerous syntheses of natural, unnatural,
and racemic castanospermine have been completed.17
Castanospermine displays an impressive breadth of
biological activity including R- and â-glucosidase inhibi-
tion and carcinoma, viral, and retroviral suppression.18
Those properties, combined with the functional density
of castanospermine prompted the search for additional
natural and unnatural analogues.17b Among those, (+)-
1-epicastanospermine ((+)-14),19 a synthetic diastereo-
mer, provided the impetus to explore the application of
the tandem cycloaddition chemistry of nitroalkenes.
Ta n d em [4 + 2]/[3 + 2] Cycloa d d ition s of Nitr o-
a lk en es. Over the past nine years we have described the
development and application of the tandem [4 + 2]/
[3 + 2] cycloaddition of nitroalkenes as a powerful
strategy for heterocycle synthesis.20 The tandem cyclo-
addition process has been successfully applied to the
synthesis of a variety of alkaloids including platynecine,20b
rosmarinecine,20c crotanecine,20d mesembrine,20e detoxinin20f
castanospermine,17a 6-epicastanospermine,17a australine,17a
and 3-epiaustraline.17a All of these compounds have been
synthesized by use of the intermolecular [4 + 2] cyclo-
addition followed by intramolecular [3 + 2] cycloaddition
strategy. In addition, the double-intermolecular tandem
sequence has been successfully used in the synthesis of
hastanecine,20g macronecine,20h and casuarine.20i,j
We undertook the synthesis of (+)-7-epiaustraline ((+)-
4) as our entree into the alexine/australine class of
pyrrolizidines. The placement of the hydroxymethyl
(16) Bernotas, R. C.; Ganem, B. Tetrahedron Lett. 1984, 25, 165.
(17) (a) Denmark, S. E.; Martinborough, E. A. J . Am. Chem. Soc.
1999, 121, 3046 and references cited. Reviews on castanospermine and
diastereomers: (b) Burgess, K.; Henderson, I. Tetrahedron 1992, 48,
4045. (c) Michael, J . P. Nat. Prod. Rep. 1997, 21.
(18) Winchester, B. G.; Cenci di Bello, I.; Richardson, A. C.; Nash,
R. J .; Fellows, L. E.; Ramsden, N. G.; Fleet, G. W. J . Biochem. J . 1990,
269, 227.
(19) Ina, H.; Kibayashi, C. J . Org. Chem. 1993, 58, 52 and referenced
cited therein.
(20) (a) Denmark, S. E.; Thorarensen, A. Chem. Rev. 1996, 96, 137.
(b) Denmark, S. E.; Parker, D. L.; Dixon, J . A. J . Org. Chem. 1997, 62,
435. (c) Denmark, S. E.; Thorarensen, A.; Middleton, D. S. J . Am.
Chem. Soc. 1996, 118, 8266. (d) Denmark, S. E.; Thorarensen, A. J .
Am. Chem. Soc. 1997, 119, 125. (e) Denmark, S. E.; Marcin, L. R. J .
Org. Chem. 1997, 62, 1675. (f) Denmark, S. E.; Hurd, A. R.; Sacha, H.
J . J . Org. Chem. 1997, 62, 1668. (g) Denmark, S. E.; Thorarensen, A.
J . Org. Chem. 1994, 59, 5672. (h) Denmark, S. E.; Hurd, A. R. J . Org.
Chem. 1998, 63, 3045. (i) Denmark, S. E.; Hurd, A. R. Org. Lett. 1999,
1, 1311. (j) Denmark, S. E.; Hurd, A. R. J . Org. Chem. 2000, 65, 2875.
(12) Pearson, W. H.; Hines, J . V. Tetrahedron Lett. 1991, 32, 5513.
(13) Choi, S.; Bruce, I.; Fairbanks, A. J .; Fleet, G. W. J .; J ones, A.
H.; Nash, R. J .; Fellows, L. E. Tetrahedron Lett. 1991, 32, 5517.
(14) Ikota, N. Tetrahedron Lett. 1992, 33, 2553.
(15) Hohenshutz, L. D.; Bell, E. A.; J ewess, P. J .; Leworthy, D. P.;
Pryce, R. J .; Arnold, E.; Clardy, J . Phytochemistry 1981, 20, 811.

Synthesis of (-)-7-Epiaustraline
J . Org. Chem., Vol. 65, No. 10, 2000 2889
group at C(3) and the additional oxygen substituent at
C(1) mandated an alternative strategy to our previously
employed syntheses of necine bases. Once the strategy
was formulated (see below) it was apparent that the
indolizidine skeleton was also readily accessed by simple
functional group manipulations, thus allowing the syn-
thesis of 1-epicastanospermine from the same intermedi-
ate. We detail below the successful realization of this
strategy which indeed culminated in the synthesis of (+)-
7-epiaustraline21 and 1-epicastanospermine. However, it
has been found that (+)-7-epiaustraline is not a natural
product and thus has stimulated a reevaluation of the
assignment of various pyrrolizidine structures.22
Several issues embodied in this strategy required
further consideration. Foremost was the influence of an
oxygen substituent at the â-terminus of the dipolarophile
on the regiochemical outcome of the [3 + 2] cycloaddition.
From previous studies on the effects of substituent on
the [3 + 2] cycloaddition, we knew that the dipolarophile
had to contain a silicon moiety as a surrogate for oxygen
to obtain the correct regiochemical result in the [3 + 2]
cycloaddition.23 Our second concern was the facial dia-
stereoselectivity in the [3 + 2] cycloaddition. On the basis
of the foregoing studies, we anticipated this process to
occur selectively in an exo fashion in an approach to the
dipole from the face opposite the C(4) benzoate.23 Less
certain, however, was the ability to stereoselectively
reduce the carbonyl group in the nitroso acetal and
selectively activate either the primary or secondary
alcohols.
Syn th etic Str a tegy: Selective Activa tion -Alk yl-
a tion . The strategy for the synthesis of both targets is
depicted in Scheme 1. We envisioned the formation of
bicyclic amines (+)-4 and (+)-14 by the alkylation of the
pyrrolidine liberated from the hydrogenolysis of a suit-
ably functionalized nitroso acetal (iii). This flexibility in
this strategy is apparent. The hydroxymethyl group of
7-epiaustraline could be introduced by C-N bond forma-
tion in an SN2 displacement reaction at an activated
secondary carbinol such as i. The indolizidine skeleton
could be created by an SN2 displacement at the activated
primary carbinol center in ii. The success of the strategy
thus hinged on the ability to differentiate between and
activate either the primary or secondary alcohol at will.
The formation of the secondary alcohol was planned as
a selective reduction of the carbonyl group needed to
activate the dipolarophile iv in the [3 + 2] cycloaddition.
The 1,2-cis relationship of the hydrogen atoms at C(2)
and C(3) in iii augured for the use of a Z-configured
dipolarophile. The trans relationship between C(3) and
C(3a) mandates an exo selective [3 + 2] cycloaddition
from the face of the nitronate opposite the C(4) substitu-
ent. Furthermore, the 1,2-trans relationship between the
hydrogen atoms at C(3a) and C(4) of the nitroso acetal
would arise from an intermolecular [3 + 2] cycloaddition
between the dipolarophile bearing an oxygen function at
the terminus and a nitronate bearing an oxygen function
at C(4). Taken in together, the nitroso acetal iii would
be created by the cycloaddition between vinyl ketone iv
and nitronate (+)-15.20g
Resu lts a n d Discu ssion
Dip ola r op h ile Syn th esis. Consideration of the de-
sign elements outlined above led to the formulation of
â-silylvinyl ketone (Z)-22 (Scheme 3) as the ideal di-
polarophile for the [3 + 2] cycloaddition. However,
because selection of the specific protecting group could
only be dictated by reactions after cycloaddition, three
choices were carried forward simultaneously. Synthesis
of 22 began with the preparation of protected R-alkoxy-
acetaldehydes 17 in two steps from 2-butene-1,4-diol or
allyl chloride by straightforward diprotection and ozonol-
ysis (Scheme 2).
Sch em e 2
Installation of the cis-alkene for the dipolarophile was
accomplished as outlined in Scheme 3. Addition of the
bromomagnesium derivative of silyl acetylene 1924 to
aldehyde 17a provided the propargyl alcohol 20a in 85%
yield. Semi-hydrogenation of the alkynes 20a -c25 with
a myriad of catalysts was plagued by irreproducibility
and over-reduction.26 Alternative protocols aimed at
increasing the reduction selectivity were also examined
without success, including polymer-bound nickel boride27
(variable selectivities), DIBAL-H in THF at reflux28 (only
31% conversion), DIBAL-H‚Et3N29 (8/1 Z/E but incom-
plete reaction), and diisoamylborane in THF30 (failed to
provide the alkene). However, reaction of 20b with 2
Sch em e 1
(21) A preliminary disclosure has already appeared. Denmark, S.
E.; Herbert, B. J . Am. Chem. Soc. 1998, 120, 7357.
(22) Wormald, M. R.; Nash R. J .; Hrnicar, J . D.; White, J . D.;
Molyneaux, R. J .; Fleet, G. W. Tetrahedron: Asymmetry 1998, 9, 2549.
(23) Denmark, S. E.; Seierstad, M. J .; Herbert, B. J . Org. Chem.
1999, 64, 4, 884.
(24) Fleming, I.; Takai, K.; Thomas, A. P. J . Chem. Soc., Perkin
Trans. 1 1987, 2269.
(25) Reviews: (a) Seigel, S. In Comprehensive Organic Synthesis;
Trost, B. M., Fleming, I., Ed.; Pergamon Press: Oxford, 1991; Vol. 8,
ch. 3.1, p 417. (b) Marshall, E. N.; Li, T. Synthesis 1973, 457.

2890 J . Org. Chem., Vol. 65, No. 10, 2000
Denmark and Herbert
equiv of dicyclohexylborane,30 followed by addition of
acetic acid resulted in a clean conversion to the cis-alkene
with no other olefinic material detectable. Upon purifica-
tion by chromatography and distillation, (Z)-21b was
secured in 71% yield.
Ta ble 1. [3 + 2] Cycloa d d ition s of 15 w ith
Dip ola r op h iles 22
Sch em e 3
dipolarophile
R
product
dra
yield, %
22a
22b
22c
Bn
TDS
DMB
23
24
25
32/1
26/1
32/1
98
97
100
a
Ratios determined by 1H NMR (500 MHz) analysis.
Ta ble 2. Selected NMR Da ta for Ma jor a n d Min or
Nitr oso Aceta ls 23
The Swern oxidation of alcohols 21a -c proceeded
uneventfully to provide the (Z)-â-silylvinyl ketones 22a-c
in 80-99% yield. In all cases the vinyl ketones could be
purified by chromatography on both alumina (neutral III)
and silica gel without isomerization. Final purification
by distillation of the liquid ketones was not possible.
Extensive isomerization to the trans alkene was noted
even at 10-5 mmHg vacuum.
major
minor
carbon 1H (ppm) 13C (ppm) carbon 1H (ppm) 13C (ppm)
C(2)
C(3)
C(3a)
5.06
2.31
3.71
88.6
32.1
74.6
C(2)
C(3)
C(3a)
4.70
3.81
3.87
78.8
52.8
78.6
minor isomer was located 9.8 ppm upfield from the major
isomer. Furthermore, comparison of the 1H and 13C NMR
chemical shifts at C(3) was instrumental in deciphering
[3 + 2] Cycloa d d ition . The thermal cycloaddition
reaction between nitronate 15 and dipolarophiles 22a -c
was a facile process. The diastereomeric ratio was
considerably higher than for monosubstituted dipolaro-
philes,23 as were the isolated yields, Table 1. For example,
the reaction between 15 and (Z)-22a provided the corre-
sponding nitroso acetal (23) in 98% yield as a 32/1 ratio
of diastereomers. The reaction was essentially complete
after 4 h at room temperature. Comparison of the 1H and
13C NMR chemical shifts of HC(2), HC(3), and HC(3a)
supported the assignment that the major nitroso acetal
was formed as a head-to-head regioisomer, Table 2. In
1
the identity of the isomers. The H NMR resonance of
C(3) in the major isomer was found at 1.5 ppm upfield
from that of the minor, suggesting that the silicon atom
was directly attached there. The carbon resonances also
supported that assignment. The resonance for C(3) of the
major nitroso acetal was located at 32.1 ppm while the
resonance for C(3) of the minor isomer was located at
52.8 ppm. The positional differences clearly indicate that
an electron-withdrawing group was attached at C(3) of
the minor isomer. Comparison with data for similar Z-â-
silyl-substituted dipolarophile cycloadditions23 was also
consistent with the identity of the minor isomer as a
regioisomer.
In accord with our [3 + 2] cycloaddition studies with
nitronate 15, both isomers were assigned structures
consistent with the approach of the dipolarophile from
the face opposite the C(4) benzoate.20g The issues sur-
rounding stereoselectivity (endo-exo and diastereofacial)
have been discussed previously.23 It is sufficient to state
that once again an extremely high tendency toward
approach of the dipolarophile to the face of the nitronate
opposite the C(4) substituent is a combined result of the
kinetic anomeric effect and the conformation of the
nitronate.
Keton e Red u ction Stu d ies. Creation of the second-
ary carbinol stereocenter of appropriate configuration
was planned as a selective reduction of the carbonyl
group under the influence of the adjacent stereocenter
of the nitroso acetal. Whereas models for the sense of
stereoinduction could be constructed, their predictive
value with regard to the actual course of reduction was
such that a sampling of reagents thought to respond to
both Felkin-type control31 and a chelation control32 were
tested. From the results of a survey of reagents in the
1
the minor isomer, the H NMR resonance of HC(2) was
shifted upfield by 0.36 ppm relative to the major isomer.
While not entirely convincing, that difference was clearly
within the range noted in the cycloaddition studies on
simple, monosubstituted dipolarophiles where the major
and minor isomers were determined to be endo-exo
isomers.23 Comparison of the 13C chemical shifts were not
consistent with the identity of nitroso acetals 23a and
23b as endo-exo isomers.23 The resonance for C(2) in the
(26) For example, hydrogenation of alkyne 20a over 5% palladium
on calcium carbonate in the presence of quinoline (6 mol %) provided
a 5/1 mixture of Z- (42%) and E- (7)% allylic alcohols ((Z)-21a and (E)-
21a ) in addition to the saturated product (18%). The use of 1 mol equiv
of quinoline (relative to substrate) failed to suppress over-reduction.
The use of 5% palladium on calcium carbonate poisoned with lead
(Lindlar catalyst) offered no advantages. The reduction times were
significantly extended (11 h vs 3 h) and resulted in lower (2-4/1) Z/E
selectivities. Hydrogenation over palladium on barium sulfate also
resulted in low and variable selectivity.
(27) (a) Choi, J .; Yoon, N. M. Tetrahedron Lett. 1996, 37, 1057. (b)
Yoon, N. M.; Park, K. B.; Lee, H. J .; Choi, J . Tetrahedron Lett. 1996,
37, 8527.
(28) Review: Pasto, D. J . In Comprehensive Organic Synthesis;
Trost, B. M., Fleming, I., Ed.; Pergamon Press: Oxford, 1991; Vol. 8,
Ch 3.3, p 471.
(29) Eisch, J . J .; Foxton, M. W. J . Org. Chem. 1971, 36, 3520.
(30) Brown, H. C. Organic Synthesis via Boranes; J ohn Wiley &
Sons: New York, 1975; pp 29 and 100.

Synthesis of (-)-7-Epiaustraline
J . Org. Chem., Vol. 65, No. 10, 2000 2891
Ta ble 3. L-Selectr id e Red u ction Selectivities
atom and to minimize dipole interactions with the
adjacent nitroso acetal C-O bond. In conformer viii, the
carbonyl is oriented in the opposite direction, toward the
nitroso acetal ring. While the relative energy of these
conformers in the ground state without the reagent is
not necessarily relevant, it is noteworthy that in both of
these conformers the phenyl ring of the phenyldimeth-
ylsilyl group is shielding one face of the carbonyl pref-
erentially. The conformer (vii) that leads to the major
reduction product corresponds to the Cornforth35 model
for carbonyl reductions.
ketone
alcohol
dra
yield, %
(23a ) Bn
(24a ) TDS
(25a ) DMB
26
27
28
>20//1
14/1
>20/1
83
87
99
Ratios determined by 1H NMR (500 MHz) analysis.
a
Activa tion of th e Secon d a r y Alcoh ol. With all five
contiguous stereogenic centers required for the synthesis
of both targets installed with high selectivity we could
now turn our attention to the activation of the secondary
carbinol for the critical ring closure. The transformation
of the alcohol to a mesylate or a tosylate proved more
challenging than expected. Reaction of alcohols 26 and
27 with equimolar quantities of methanesulfonyl chloride
or 4-toluenesulfonyl chloride in the presence of a variety
of bases (triethylamine, Hu¨nig base, pyridine, 2,6-luti-
dine, sodium hydride, methyllithium) in methylene chlo-
ride or THF failed to provide the corresponding sulfonate
esters. To our delight, methanesulfonic anhydride36 (6
equiv) was found to react cleanly with alcohol 27 in 5
min when pyridine was used as the solvent (Scheme 4).
Although stable indefinitely in vacuo, mesylate 29 de-
composed gradually upon exposure to air and rapidly in
solution.
reduction of simpler model ketones33 it was clear that
maximal selectivity was obtained when L-Selectride was
used as the hydride source and the R-alkoxymethyl
ketone was protected as a benzyl ether. Not surprisingly,
reduction of the ketone 23a with L-Selectride at -74 °C
provided the pair of epimeric alcohols 26 in excellent
diastereoselectivity (>20/1, 83%), Table 3.
1
Examination of the H NMR coupling constants of the
newly created stereogenic center [C(1iv)] and at C(2) of
the nitroso acetal revealed that those protons did not
couple. The inference was that the protons were aligned
at a 90° angle and that no information regarding the
sense of stereoselection could be obtained. As expected,
similar reduction selectivity was observed for DMB-
protected ketone 25a with L-Selectride. However, we
were gratified to discover that silyl protected ketone 24a
also underwent reduction with L-Selectride with high
stereoselectivity. The diastereomeric alcohols 27 were
obtained in a 14/1 ratio. The protons at the newly created
stereogenic center [C(1iv)] and at C(2) of the nitroso acetal
did not couple, suggesting that the sense of stereochem-
ical induction was the same as with ketones 23 and 25.
Fortunately, both major reduction products 26a and 27a
were obtained as solids. Crystals of each carbinol suitable
for X-ray analysis were obtained by slow crystallization.
The analysis of carbinol 26a confirmed that the reduction
had proceeded to afford the S-configuration at the newly
created center. Thus, the relative configuration of all five
stereogenic centers required for the proposed syntheses
of (+)-7-epiaustraline and (-)-1-epicastanospermine were
correct. Much to our delight, the X-ray crystal structure
analysis of carbinol 27a confirmed our suspicion that the
Selectride reduction of ketone 24a occurred in the same
stereochemical sense as in ketones 23 and 25 (see
Supporting Information). Since carbinol 27a was pre-
pared from nitronate (+)-15 containing the (1S,2R)-
phenylcyclohexyloxy substituent, the relative and abso-
lute configurations were secured.
Sch em e 4
Unfortunately benzyl protected carbinol 26 resisted
activation under these conditions as well. We then
resorted to trapping the reduction product from 23 and
25 in situ with various triflating and sulfonylating
agents, none of which proved successful.37,38 At this stage
the use of the benzyl ether type protecting groups was
abandoned.
These difficulties prompted us to consider liberating
the primary alcohol in order to deliver the activating
group to the secondary alcohol in an intramolecular pro-
cess. The intermediate diol would also allow access to the
doubly activated substrates such as cyclic sulfate or epox-
ide which would make intriguing precursors for closure.
The same sense of stereochemical induction for benzyl-
and thexyldimethylsilyl-protected R-alkoxymethyl ke-
tones is inconsistent with the delivery of the hydride from
a species coordinated to the ether oxygen. Consequently,
an alternative control element must be operative. The
remarkable selectivity and sense of induction can be
rationalized with the aid of ground-state energy mini-
mization calculations34 of ketone 24 which provided two
low energy conformers, Figure 3. In conformer vii, the
carbonyl is oriented away from the nitroso acetal ring
system in such a way as to minimize through-space
interactions with the phenyl ring attached to the silicon
(33) Reduction of model ketones v and vi with a variety of reducing
agents provided initial leads. For v L-Selectride gave 19/1 selectivity
but no reagent was found to give good selectivities with vi.
(34) Molecular Mechanics calculations executed with the Spartan
calculation package on a CAChe machine.
(35) Cornforth, J . W.; Cornforth, R. H.; Mathew, K. K. J . Chem. Soc.
1959, 112.
(36) Field, L.; Settlage, P. H. J . Am. Chem. Soc. 1954, 76, 1222.
(37) Comins, D. L.; Dehghani, A.; Foti, C. J .; J oseph, S. P. Org.
Synth. 1996, 74, 77, and references therein.
(31) Che´rest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1968,
(38) Introduction of the mesylate by [3 + 2] cycloaddition with the
dipolarophile bearing the activating group (allylic mesylate) was also
unsuccessful. The nitronate decomposed prior to reacting with the
dipolarophile at both ambient temperature and at 80 °C.
2199.
(32) Hoveyda, A. H.; Evans. D. A.; Fu, G. C. Chem. Rev. 1993, 93,
1307.

2892 J . Org. Chem., Vol. 65, No. 10, 2000
Denmark and Herbert
F igu r e 3. Ketone 24a conformations calculated using molecular mechanics.
The reaction of silyl ether 27 with 1.3 equiv of TBAF
resulted in the rapid removal of the silyl protecting group
to afford the diol 30 in 72% yield, Scheme 5. The diol
was allowed to react with an excess of thionyl diimida-
zole39 in THF to furnish the cyclic sulfite in 85% yield as
a 1.2/1 mixture of sulfur epimers. Oxidation of the sulfite
to the sulfate occurred smoothly and quantitatively using
ruthenium (III) chloride and sodium periodate.40 Sulfate
31 was subsequently found to be somewhat unstable,
surviving rapid chromatography with a polar solvent but
not recrystallization.
29. At 160 psi of hydrogen in the presence of Raney
nickel, 29 furnished pyrrolizidine 34 in 77% yield with
98% recovery of the auxiliary, Scheme 7. This reduction
was insensitive to hydrogen pressure as evidenced by
comparable yields at 160, 260, and 360 psi (72-74%).
Removal of the silyl ether protecting group afforded the
familiar primary alcohol 33 in 72% yield which was
identical to material obtained from the hydrogenolysis
of the cyclic sulfate.
Sch em e 7
Sch em e 5
The isolation of pyrrolizidine 34 in high yield from
simple hydrogenolysis validated our expectation that the
hydrogenolytic cascade could be terminated successfully
by an alkylation in addition to the many acylations that
had been described. The lack of isolable intermediates
along the hydrogenolysis pathway suggests that the
closure is rapid, even at such a congested center. How-
ever, unlike the acylating closures, it is conceivable
that the alkylation takes place prior to the reductive
amination.42
Syn th esis of 7-Ep ia u str a lin e. Completion of the
synthesis of 7-epiaustraline required three operations;
the removal of the silyl ether protecting group, the
hydrolysis of the benzoate, and the Si to O conversion.
All three steps were conveniently accomplished with the
Tamao-Fleming oxidation,43 Scheme 8. Amine 34 was
first protected by protonation with methanesulfonic acid
and then treated with mercury (II) acetate and a 32%
solution of peroxyacetic acid in acetic acid and heated
for 14 h at 40 °C.44 After purification by ion exchange
chromatography, the N-oxide was reduced by hydro-
genolysis over 10% Pd-C in methanol/acetic acid. Final
purification by ion exchange chromatography provided
Hyd r ogen olysis Stu d ies. With a number of activated
substrates in hand, we could evaluate the crucial hydro-
genolysis/ring closure construction of the azabicyclic
nucleus. We began with the cyclic sulfate 31 to ascertain
the relative preference for closure to a pyrrolizidine
versus an indolizidine. Hydrogenation of 31 with Raney
nickel in methanol at 160 psi for 36 h provided a 4/1
mixture of sulfate salts in 72% yield, Scheme 6. Libera-
tion of the free base was accomplished in 90% yield by
the addition of 1 equiv of concentrated H2SO4 to a THF
suspension of the salt mixture.41 NMR (COSY) analysis
indicated that the major product possessed the indolizi-
dine skeleton (32) and that the minor product was the
pyrrolizidine isomer (33).
Sch em e 6
(39) Denmark, S. E. J . Org. Chem. 1981, 46, 3144.
(40) (a) Ramaswamy, S.; Prasaid, K.; Repic, O. J . Org. Chem. 1992,
57, 6344. (b) Review of cyclic sulfate chemistry: Lohray, B. B. Synthesis
1992, 1035.
(41) Kim, B. M.; Sharpless, K. B. Tetrahedron Lett. 1989, 30, 655.
(42) For an in depth discussion of he consequences of staging the
hydrogenolysis, see ref 17a.
(43) Fleming, I. Chemtracts: Org. Chem. 1996, 9, 1.
(44) Rehders, F.; Hoppe, D. Synthesis 1992, 865.
With authentic samples of both azabicyclic compounds
in hand, we next studied the hydrogenation of mesylate

Synthesis of (-)-7-Epiaustraline
J . Org. Chem., Vol. 65, No. 10, 2000 2893
a solid in 19-25% yield. Substitution of mercury(II)
trifluoroacetate resulted in a dramatic increase in yield.
Amine 34 was allowed to react with Hg(O2CCF3)2 in a
mixture of trifluoroacetic acid and acetic acid for 1 h to
effect protiodephenylation.45,46 The reaction mixture was
diluted with acetic/peroxyacetic acid at 0 °C and then was
heated at 50 °C for 24 h. The N-oxide was isolated by
ion exchange chromatography and subsequently reduced
(H2, 10% Pd-C, MeOH/HOAc). Final purification on
analytical grade Dowex (H-form) provided 7-epiaustraline
(4) in 74% yield. Recrystallization from MeOH-CHCl3
provided an analytically pure sample in 61% yield.47 The
tetrol was peracetylated and the enantiomeric excess
determined by CSP supercritical fluid chromatography
to be 97% ee.48
the sample of 7-epiaustraline (4) provided by Pearson12,50
revealed several interesting differences, Table 4. Exami-
nation of the coupling constants between HC(1), HC(2),
and HC(3) was particularly informative. The coupling
constants for those protons from the Pearson sample were
similar to those of the compound prepared by our
synthetic route. Values of J ) 8-9 Hz were consistent
with an all-trans relationship between those protons.
An all-trans relationship in the fully substituted five-
membered ring can be satisfied by four of the known
australine/alexine isomers: alexine (1), 7-epialexine (9),
australine (2), and 7-epiaustraline (4). The coupling
between HC(1) and HC(7a) for material prepared by
Pearson and ourselves was 7.6-7.8 Hz, suggesting a
trans-relationship between those protons. A trans-
relationship between HC(1) and HC(7a) precludes the
alexines as possible isomers. That leaves australine and
7-epiaustraline as potential candidates for the identity
of the material synthesized by Pearson12 and isolated by
Nash et al.7 The remaining coupling constant, HC(7a)-
HC(7), was determined to be 2 Hz in the synthetic
material prepared herein. A value of 4.4 Hz was found
in the data supplied by Pearson.50 In australine, the value
of the same proton coupling was reported as 5 Hz.7
Comparison of all three values points to a closer match
between Pearson’s sample and australine. In fact, aus-
traline was the synthetic target identified by Pearson.
When compared to the Nash data, those for Pearson’s
synthetic material matched that of isolated 7-epiaustra-
line. They were forced to invoke a curious inversion of
an epoxide stereogenic center during the crucial epoxide
opening reaction to explain the synthesis of 7-epiaustra-
line instead of australine. A subsequent reevaluation of
the published data for australine indeed confirm that the
material originally assigned the structure of 7-epiaus-
traline is indeed australine itself.22 Thus, Pearson et al.
did in fact succeed in synthesizing their intended target,
but chose not to challenge the assignment when the data
did not match. In sum, 7-epiaustraline is not (yet) a
naturally occurring pyrrolizidine.
Syn th esis of 1-Ep ica sta n osp er m in e. The synthetic
strategy detailed in Scheme 1 requires a selective activa-
tion of the primary alcohol of diol 30 to provide the
requisite precursor for 1-epicastanospermine. Treatment
of the diol 30 with 4-toluenesulfonyl chloride resulted in
the slow consumption of the starting material and the
formation of the primary tosylate 35, Scheme 9. Substi-
tution of 2,6-lutidine for pyridine improved the yield of
35 to 95%. Replacing 4-toluenesulfonyl chloride with
the anhydride decreased the reaction times and simpli-
fied purification. The tosylate was unstable in its pure
form and had to be hydrogenolyzed in the presence of
residual 2,6-lutidine. We were able to secure the unstable
epoxide 36 from 35 through the action of potassium tert-
butoxide.
Although the cyclic sulfate 31 had provided the in-
dolizidine selectively (4/1), we hoped to find a precursor
that exclusively produced this ring system. Thus, the
epoxy nitroso acetal 36 was first examined. Hydrogenol-
ysis of 36 provided a low yield of pyrrolizidine 33 whose
1H NMR spectrum was identical to the material from the
hydrogenolysis of the cyclic sulfate 31 and mesylate 29,
Sch em e 8
1
Str u ctu r a l Ver ifica tion a n d Assign m en t. The H
NMR spectrum of the final product did not match
the data reported in the literature.7 Moreover, the 1H
NMR spectrum of synthetic (-)-4 did not match the
data reported for any of the known alexine isomers. In
addition, the product from our synthesis was a levorota-
tory solid (mp 161 °C) whereas the substance reported
as 7-epiaustraline is described as a dextrorotatory oil.7
Repeated purification of a sample of (-)-4 afforded
crystals suitable for X-ray analysis. The crystal structure
(see the Supporting Information) left no doubt that the
synthetic plan was successfully executed and indeed
delivered 7-epiaustraline. Thus, although the synthesis
of the target structure was complete, we had unambigu-
ously established that (1R,2R,3R,7R,7aR)-hexahydro-3-
(hydroxymethyl)-1H-pyrrolizine-1,2,7-triol ((-)-4) was not
the material isolated by Nash et al.49 While outside our
purview, we were nonetheless mystified as to the nature
of the material reported in the literature as 7-epiaustra-
line.22
Comparison of the 1H NMR data for the isolated
material with data obtained for our compound and for
(45) Kolb, H. C.; Ley, S. V.; Slawin, A. M. Z.; Williams, D. J . Perkin
Trans. 1 1992, 2735.
(46) A 1H NMR study indicated that the dephenylation required
<20 min.
(47) The procedure was found to be reproducible within the 0.06-
0.25 mmol range resulting in the isolation of 7-epiaustraline in 68-
77% yield. Application of the exact conditions on a 1 mmol scale gave
a much lower yield (31%). The N-oxide eluted over a considerable
length of time during the ion exchange purification suggesting that
binding the presence of mercury salts was a source of the problem.
Reducing the quantity of Hg(O2CCF3)2 from 7.5 equiv to 1.5 equiv. did
not affect the yield of the process but did slightly improve the elution
of the N-oxide.
(48) Analytical CSP supercritical fluid chromatography was per-
formed by Robert Stavenger.
(49) GC-MS analysis of the persilylated product (-)-4 clearly showed
that it was different from the natural material and also different from
(+)-australine. This ruled out the possibility that the spectroscopic
discrepancies resulted from the formation of salts (carbonates, hy-
drates, or silicates) in the isolation of the natural product. We are
grateful to Dr. R. J . Nash (Aberystwyth) for graciously performing
those analyses.
(50) We are grateful to Professor William Pearson (University of
Michigan) for kindly supplying spectral data for his synthetic 7-epi-
australine.

2894 J . Org. Chem., Vol. 65, No. 10, 2000
Ta ble 4. 1H a n d 13C Da ta for 7-Ep ia u str a lin e (4) a n d Au str a lin e (2)
Denmark and Herbert
nucleus isolateda
Pearsonb
this workc
this workd
3.58 (t, J ) 7.8)
3.71 (dd, J ) 8.5, 7.8)
2.68 (ddd, J ) 9.0, 6.1, 3.4)
3.24 (m)
australinee
HC(1)
HC(2)
HC(3)
HC(5)
4.02 (dd) 4.06 (t, J ) 7.7)
3.70 (dd) 3.73 (t, J ) 8.8)
2.50 (m) 2.58 (m)
2.97 (m) 3.00 (m)
2.50 (m) 2.58 (m)
1.85 (m) 1.85 (m)
1.73 (m) 1.79 (m)
4.19 (m) 4.21 (m)
3.54 (t, J ) 8.1)
3.58 (t, J ) 8.5)
2.50 (ddd, J ) 9.3, 6.3, 3.9)
2.90 (ddd, J ) 11.5, 10.3, 5.9)
2.70 (ddd, J ) 11.0, 7.1, 3.4)
1.90 (m)
4.41 (dd, J ) 5)
3.92 (dd, J ) 9, 5)
3.03 (m)
3.18 (m)
2.93 (m)
2.98 (ddd, J ) 11.2, 7.1, 2.7)
2.77 (ddt, J d ) 13.2, 5.4; J t ) 2.7) 2.02 (m)
HC(6)
HC(7)
1.60 (ddt, J d ) 12.7, 6.1; J t ) 3.4) 2.02 (m)
4.18 (dt, J d ) 4.6; J t ) 2.7) 4.28 (dt, J d ) 4.4; J t ) 2.2)
4.58 (dd, J ) 9, 5)
3.46 (dd, J ) 5)
3.81 (dd, J ) 12, 4)
3.64 (dd, J ) 12, 6.5)
HC(7a) 2.98 (dd) 3.05 (dd, J ) 7.6, 4.4) 2.84 (dd, J ) 7.8, 2.0)
3.16 (dd, J ) 7.8, 1.5)
3.75 (dd, J ) 11.2, 3.4)
3.60 (dd, J ) 11.5, 6.1)
HC(8)
3.59 (dd) 3.63 (dd, J ) 11.8, 3.4) 3.60 (dd, J ) 11.7, 3.9)
3.42 (dd) 3.45 (dd, J ) 11.9, 6.6) 3.46 (dd, J ) 11.7, 6.3)
C(1)
C(2)
C(3)
C(5)
C(6)
C(7)
C(7a)
C(8)
73.8
71.4
71.2
52.5
35.8
79.7
70.2
63.5
73.9
71.7
71.4
52.7
35.9
79.5
70.3
63.1
77.5
75.9
67.9
51.4
33.1
74.8
73.5
62.3
79.8
77.4
71.3
53.5
33.1
77.2
76.1
63.1
74.9
77.3
72.9
54.9
38.1
75.8
68.9
65.6
a
b
Spectra taken in D2O and assigned by COSY.7 These data have subsequently been shown to be those for austraine.22 Spectra taken
in D2O and assigned by comparison with data for isolated material.12 c Spectra taken in D2O and assigned by COSY.4 Spectra taken
d
in CD3OD and assigned by COSY. e Spectra taken in D2O and assigned by COSY. These data have subsequently been shown to be those
for 1-epiaustraline.22
Sch em e 9
minimized by washing the Raney nickel thoroughly with
H2O. To maximize reproducibility, several washings with
pH 7 buffer were included. With these precautions, a
14/1 ratio (1H NMR) of 32 to 33 could be obtained. The
hydrogenolysis of tosylate 35 exhibited a modest pres-
sure effect. For example, at 160 psi of H2, the amines 32
and 33 were isolated in a 8/1 ratio in 79% yield. At
260 and 360 psi, a 12/1 ratio of the amines was ob-
served, albeit in variable yields (75% at 260 psi and
66% at 360 psi). Under the final optimized procedure,
diol 30 was transformed into a mixture of amines 32
and 33 (8/1 by 1H NMR analysis) in addition to re-
covered auxiliary (94%). The amines were separated and
isolated in 64% and 4% yields, respectively. Analysis of
3
the J H,H coupling constants of the indolizidine 32 was
consistent with an all trans-relationship in the six-
membered ring.
Scheme 10. The major product was unidentifiable, but
most certainly not 32.
The divergence of selectivity in the opening of the cyclic
sulfate 31 and epoxide 36 merits comment. In both cases
the formation of the pyrrolizidine nucleus involves a
5-exo-tet closure. However, the indolizidine formation
requires a 9-endo-tet closure from 31 as compared to a
7-endo-tet closure from 36.52 According to Baldwin,51
5-exo-tet is preferred over 6-endo-tet pathway, but the
preference over higher order endocyclic closures is not
established. Since the basis of the Eschenmoser/Baldwin
postulate is the overlap of the approaching nucleophile
with the σ* orbital of the carbon bearing the leaving
group, the two components should ideally achieve an
alignment approximating 180°. In the cyclic sulfate, the
trajectory of displacement at the terminus (9-endo-tet)
is nearer to that angle than at the internal carbon (the
Sch em e 10
We next turned to primary tosylate 35 which should
be capable of producing only the indolizidine skeleton.
Surprisingly, hydrogenation of 35 at 260 psi over Raney
nickel afforded a 1/4 mixture of indolizidine 32 and
pyrrolizidine 33, Scheme 11. The formation of 33 sug-
gested the intervention of a competing pathway in-
volving a base-mediated epoxide formation followed by
5-exo-tet ring opening.51 The amount of 33 could be
(52) Rigorously, to compare endocyclic and exocyclic SN2 reactions
the nucleofuge must be counted as part of the endocyclic process.
Tenud, L.; Farooq, S.; Seibl, J .; Eschenmoser, A. Helv. Chim. Acta 1970,
53, 2059.
(51) (a) Baldwin, J . E. J . Chem. Soc., Chem. Commun. 1976, 734.
(b) Baldwin, J . E.; Lusch, M. J . Tetrahedron 1982, 38, 2939.

Synthesis of (-)-7-Epiaustraline
J . Org. Chem., Vol. 65, No. 10, 2000 2895
Sch em e 11
Ta ble 5. Rota tion Da ta for 1-Ep ica sta n osp er m in e (14)
accessibility of a primary versus a secondary center
notwithstanding). In contrast, the reverse is true for the
epoxide. The net effect in the nucleophilic opening of
cyclic electrophiles is that the size of the ring containing
the nucleofuge dramatically affects the orientation of the
interacting orbitals and thus alters the selectivity.
Completion of the synthesis of 1-epicastanospermine
required the same three operations as were mandated
in the synthesis of 7-epiaustraline. Again, all three steps
were conveniently accomplished with the Tamao-Flem-
ing oxidation, Scheme 12. Thus, amine 32 was allowed
to react with Hg(O2CCF3)2 in trifluoroacetic acid for 1 h
to effect dephenylation of the silane.44 The reaction
mixture was diluted with acetic/peroxyacetic acid and
was heated at 50 °C for 14 h. The N-oxide was isolated
by ion exchange chromatography and subsequently re-
duced (H2, 10% Pd-C, MeOH-HOAc). Final purification
on analytical grade Dowex (H form) followed by chroma-
tography on Reverse Phase silica gel (C-18) (CHCl3/
MeOH/NH4OH, 1/0/0, 90/9/1 then 20/5/1) provided 1-ep-
icastanospermine (14) as an oil in 77% yield. An analytical
sample of the amine was obtained by conversion to its
HCl salt. Recrystallization of the salt from MeOH-
Et2O and MeOH-CHCl3 afforded an analytically pure
hydrochloride, 14‚HCl. The free base was then obtained
from the analytical sample by ion exchange chromatog-
raphy (Dowex OH-form) for the purposes of comparison
with reported data. The 1H NMR spectrum of the
synthetic material was identical to that reported in the
literature.19
starting material
14 [R]D
reference
D-mannose
D-xylose
D-glucose
D-glucofuranurono- (-)
6,3-lactone
dimethyl L-tartrate +3.8° (c ) 0.5, MeOH)
D-glucofuranurono- -5° (c ) 0.1, H2O)
6,3-lactone
-39.1° (c ) 0.4, H2O)
Hashimoto53a
-4° (c ) 0.3, MeOH) [ent-14] Mulzer53b
+6° (c ) 0.45, H2O)
Ganem53c
Anzeveno53d
Kibayashi19
Stu¨tz53e
and sign of the rotation is immediately apparent. Mulzer
has effectively demonstrated the dependence of optical
rotation on pH by measuring the rotation of ent-14 at
pH 7.1, 6, and 5. At those pH levels, [R]2D0 values of -21,
-10, and +4 were obtained. In addition, the authors
determined the rotation in MeOH ([R]2D0 -4), presum-
ably to minimize pH effects. Kibayashi also measured the
rotation of 1-epicastanospermine in MeOH and found the
value to be +3.8.19
Due to the small magnitude of R, it is likely that trace
quantities of impurities or salts will affect both the sign
and magnitude of the rotation.54 Our sample of 14 had
the following rotation values: [R]2D3 +18.25° (c ) 1.23,
H2O) for the HCl salt, [R]23 -5.31° (c ) 0.83, H2O, pH
D
9.7) and [R]2D3 -3.10° (c ) 1.86, MeOH) for the free base.
The absolute configuration of our compound was secured
from the single-crystal X-ray analysis of nitroso acetal
(-)-27a which leads to (1R,6S,7R,8R,8aR)-14. The purity
of (-)-1-epicastanospermine was assured through conver-
sion to the HCl salt. The free amine was obtained from
the analytically pure salt by passage through a Dowex
column (OH-form) eluting with distilled H2O.55 The pH
of the aqueous solution of 14 used to measure the rotation
was determined to be 9.7, a clear indication that the
amine was in the free base form. As a result there is no
Sch em e 12
(53) (a) Setoi, H.; Takeno, H.; Hashimoto, M. Tetrahedron Lett. 1985,
26, 4617. (b) Mulzer, J .; Dehmlow, H.; Buschmann, J .; Luger, P. J .
Org. Chem. 1992, 57, 3194. (c) Bernotas, R. C.; Ganem, B. Tetrahedron
Lett. 1984, 25, 165. (d) Anzeveno, P. B.; Angell, P. T.; Creemer, L. J .;
Whalon, M. R. Tetrahedron Lett. 1990, 31, 4321. (e) Graâberger, V.;
Berger, A.; Dax, K.; Fechter, M.; Gradnig, G.; Stu¨tz, A. E. Liebigs Ann.
Chem. 1993, 379.
(54) For discussion of the use of polarimetry, see (a) Lyle, G. G.;
Lyle, R. E. In Asymmetric Synthesis; Morrison, J . D., Ed.; Academic
Press: New York, 1983; Vol. 1; Chapter 2. (b) Eliel, E. L.; Wilen, S. H.
Stereochemistry of Carbon Compounds; Wiley: New York, 1994; pp
1071-1080.
Op t ica l R ot a t ion . While the spectroscopic data
matched, the optical rotation data for synthetic 14 was
clearly at variance with the literature values. The optical
rotations and individual starting materials for the re-
ported syntheses of 1-epicastanospermine are compiled
in Table 5. With the exception of Mulzer53b (who syn-
thesized ent-14), all the authors claimed the success-
ful preparation of the targeted (1R,6S,7R,8R,8aR)-
octahydro-1,6,7,8-indolizinetetrol (14) in addition to
(1S,6S,7R,8R,8aR)-octahydro-1,6,7,8-indolizinetetrol [(+)-
castanospermine (13)]. The huge variation in magnitude
(55) The discrepancy with the rotation of 6-epicastanospermine was
solved in a similar manner. The rotation was measured on a sample
of the analytically pure HCl salt in 0.1 M NaOH in MeOH: Gerspacher,
M.; Rapoport, H. J . Org. Chem. 1991, 56, 3700.

2896 J . Org. Chem., Vol. 65, No. 10, 2000
Denmark and Herbert
question that the rotation value of 14 under the condi-
tions indicated is [R]D23 -5.31° (c ) 0.83, H2O, pH 9.7).
optical rotation of (1R,6S,7R,8R,8aR)-14 has been re-
evaluated to be levorotatory on the basis of this
work.
Con clu sion s
Ack n ow led gm en t. We are grateful to the National
Institutes of Health (GM-30938) for generous financial
support. We also acknowledge Drs. G. W. J . Fleet, M.
A. Wormald (Oxford), R. J . Nash (Aberystwyth), and
Prof. W. H. Pearson (Michigan) for helpful exchange of
information.
The successful preparation of (-)-7-epiaustraline (9
steps, 23% yield, 97% ee) and 1-epicastanospermine (10
steps, 20% yield, and 97% ee) was accomplished by
extending the tandem [4 + 2]/[3 + 2] cycloaddition
strategy to include an alkylation in the hydrogenolysis
step. The strategy was divergent and allowed for the
preparation of two classes of natural products from a
common intermediate. Central to the strategy was the
highly regio- and stereoselective [3 + 2] cycloaddition of
nitronate (+)-15 with alkoxymethyl (Z)-â-silyl-
vinyl ketones and the highly selective reduction of the
resulting ketone 24a with L-Selectride. The structure
of (-)-7-epiaustraline was secured by single crystal
X-ray analysis which served to initiate a reevaluation
of the structures of naturally isolated australines. The
Su p p or tin g In for m a tion Ava ila ble: The general experi-
mental methods and full experimental, spectroscopic, and
analytical details (1H NMR, 13C NMR, IR, MS, TLC, [R]D) for
all new compounds along with 1H and 13C NMR spectra for
(-)-7-epiaustraline and (-)-1-epicastanospermine; crystallo-
graphic parameters, atomic coordinates, bond lengths and
angles for (-)-4, (-)-26a , (-)-27a . This material is available
free of charge via the Internet at http://pubs.acs.org.
J O991990D