Synthesis of (+)-1-epiaustraline

Article abstract of DOI:10.1021/jo0101510

A highly efficient total synthesis of (+)-1-epiaustraline ((+)-1), a tetrahydroxypyrrolizidine alkaloid of the alexine/australine subclass, is described. The key step is a tandem intramolecular [4 + 2]/intermolecular [3 + 2] nitroalkene cycloaddition involving dienylsilyloxy nitroalkene 3 and chiral vinyl ether 4, which establishes four of the five stereocenters present. The final center was installed by a diastereoselective dihydroxylation. Hydrogenolytic unmasking of the nitroso acetal tosylate 17 containing the silyl ether linkage was thwarted by a slow alkylation and an undesired Peterson-type elimination. Prior removal of the silicon moiety by Tamao-Fleming oxidation proceeded in excellent yield and provided a substrate suitable for hydrogenolysis and deprotection. The complete synthesis required only 10 steps to deliver the (+)-1-epiaustraline in 7.0% overall yield.

Full text of DOI:10.1021/jo0101510

4276
J . Org. Chem. 2001, 66, 4276-4284
Syn th esis of (+)-1-Ep ia u str a lin e
Scott E. Denmark* and J eromy J . Cottell
Roger Adams Laboratory, Department of Chemistry, University of Illinois, Urbana, Illinois 61801
denmark@scs.uiuc.edu
Received February 7, 2001
A highly efficient total synthesis of (+)-1-epiaustraline ((+)-1), a tetrahydroxypyrrolizidine alkaloid
of the alexine/australine subclass, is described. The key step is a tandem intramolecular
[4 + 2]/intermolecular [3 + 2] nitroalkene cycloaddition involving dienylsilyloxy nitroalkene 3 and
chiral vinyl ether 4, which establishes four of the five stereocenters present. The final center was
installed by a diastereoselective dihydroxylation. Hydrogenolytic unmasking of the nitroso acetal
tosylate 17 containing the silyl ether linkage was thwarted by a slow alkylation and an undesired
Peterson-type elimination. Prior removal of the silicon moiety by Tamao-Fleming oxidation pro-
ceeded in excellent yield and provided a substrate suitable for hydrogenolysis and deprotection.
The complete synthesis required only 10 steps to deliver the (+)-1-epiaustraline in 7.0% overall
yield.
Ch a r t 1
In tr od u ction
1-Epiaustraline (1,7a-diepialexine, 1) is one of several
alkaloids produced by the plant Castanospermum Aus-
tralae (commonly known as the Moreton Bay Chestnut).
Its isolation and structural elucidation, accomplished
independently in the laboratories of Fleet1 and Harris,2
proved 1 to be an epimer of the alkaloid alexine (2).
Alexine-type alkaloids differ from the more common
pyrrolizidines of the necine family by the position of the
substituents on the pyrrolizidine ring. Whereas alexines
possess a hydroxymethyl group at the C(3) position, the
necines bear this substituent at the C(1) position of the
pyrrolizidine and usually also contain an acidic ansa side
chain (Chart 1).
As with other alexine alkaloids, 1-epiaustraline is an
inhibitor of several glycosidases. The highest inhibition
with 1 was observed against the R-glucosidase amylo-
glucosidase (50% inhibition at 2.6 × 10-5 M) and yeast
R-glucosidase (50% inhibition at 2.7 × 10-4 M), while
lower inhibition was obtained with mammalian digestive
glycosidases.1,2 The mode of inhibition is believed to arise
from the highly oxygenated structure and the ability of
the protonated pyrrolizidine to bind as a transition state
analogue.3 Sugar-mimic alkaloids have been investigated
as both anti-cancer and anti-viral treatments due to their
glycosidase inhibition properties.
laboratories of Fleet5 and Ikota.6 The synthesis by Fleet
involves the elaboration of a mannose derivative in 14
steps and 7.2% overall yield. During the course of the
synthesis, all of the stereocenters of mannose are utilized
to provide 1-epiaustraline. In the Ikota synthesis, 1-epi-
australine is synthesized from a glutamine derivative in
15 steps and 4.7% overall yield. The resident stereocenter
of gluatamine controls the installation of the remaining
stereocenters in the target. For our purposes, a de novo
synthesis of 1-epiaustraline provided an opportunity to
further develop the tandem [4 + 2]/[3 + 2] nitroalkene
cycloaddition and its application to pyrrolizidine alka-
loids.
Ba ck gr ou n d
Recent publications from these laboratories have amply
demonstrated the power of the tandem [4 + 2]/[3 + 2]
nitroalkene cycloadditions for the construction of densely
functionalized of nitroso acetals with high levels of
stereocontrol (Scheme 1).7 Subsequent reductive cleavage,
in the presence of an ester at the C(2) position of the
nitroso acetal, then results in the cyclization of the in
situ generated amine to form pyrrolizidinone-type com-
pounds. This approach has been successfully employed
in the synthesis of several necine bases, including ros-
marinecine,8 playtnecine,9 crotanecine,10 and hastane-
cine.11
Although there has been considerable interest in the
synthesis of alexine alkaloids,4 there are only two re-
ported syntheses of 1-epiaustraline, accomplished by the
(1) Nash, R. J .; Fellows, L. E.; Dring, J . V.; Fleet, G. W. J .; Girdhar,
A.; Ramsden, L. E.; Peach, J . M.; Hegarty, M. P.; Scofield, A. M.
Phytochemistry 1990, 29, 111.
(2) Harris, C. M.; Harris, T. M.; Molyneux, R. J .; Tropea, J . E.;
Elbein, A. D. Tetrahedron Lett. 1989, 30, 5685.
(3) (a) Asano, N.; Nash, R. J .; Molyneux, R. J .; Fleet. G. W. J .
Tetrahedron: Asymmetry 2000, 11, 1645. (b) Fellows, L. E.; Nash, R.
J . Sci. Prog. 1990, 74, 245. (c) Winkler, D. A.; Holan, G. J . Med. Chem.
1989, 32, 2084. (d) Kajimoto, T.; Liu, K. K.-C.; Pederson, R. L.; Zhong,
Z.; Ichikawa, Y.; Porco, J . A. J r.; Wong, C. H. J . Am. Chem. Soc. 1991,
113, 6187.
(4) (a) Liddell, J . R. Nat. Prod. Rep. 1999, 16, 499. (b) Casiraghi,
G.; Zanardi, F.; Rassu, G.; Pinna, L. Org. Prep. Proced. Int. 1996, 28,
641.
(5) 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.
(6) Ikota, N. Tetrahedron Lett. 1992, 33, 2553.
(7) For a review on the [4 + 2]/[3 + 2] tandem cycloaddition, see:
Denmark, S. E.; Thorarensen, A. Chem. Rev. 1996, 96, 137.
(8) Denmark, S. E.; Thorarensen, A.; Middleton, D. S. J . Am. Chem.
Soc. 1996, 118, 8266.
10.1021/jo0101510 CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/12/2001

Synthesis of (+)-1-Epiaustraline
J . Org. Chem., Vol. 66, No. 12, 2001 4277
Sch em e 1
Sch em e 2
Apart from the hydrogenolytic cleavage of the nitroso
acetal system, the chemistry of this functional group has
not been intensively investigated.12 As part of total
synthesis endeavors in these laboratories, nitroso acetals
have shown compatibility with common chemical trans-
formations, including oxidations, reductions, and sulfon-
ylations. These manipulations have allowed access to
additional alkaloid structures by the installation of an
internal alkylating agent after the tandem cycloaddition.
Thus, during the reductive cleavage of the nitroso acetal,
the cyclization can proceed through an alkylation event,
instead of an acylation. This process was a critical part
of the strategy that lead to the successful syntheses of
the indolizidine castanospermine13 and the pyrrolizidines
australine13 and casuarine.14
terminal alkene in ii, followed by a selective protection
and activation of the hydroxyl groups to give i. All of
these transformations have been carried out successfully
and selectively on nitroso acetals in previous syntheses.13
Our understanding of the tandem [4 + 2]/[3 + 2]
cycloaddition provides for good predictability in the
stereochemical course of the process. Examination of the
nitroso acetal ii allows for a straightforward retrosyn-
thetic analysis that thus identifies the requisite coupling
partners. First, the trans relationship between the sub-
stituents at C(2) and C(3) of ii necessitates the use of a
trans dipolarophile for the [3 + 2] cycloaddition. However,
to create the cis/cis relationship between the hydrogens
on C(3), C(3a), and C(4), the dipolarophile must approach
the nitronate in an endo mode with respect to the C(3)
substituent and from the same side as the C(4) substitu-
ent. Previous experience has taught us that the use of a
tether between the nitronate and the dipolarophile will
not only control the facial selectivity, but can also
influence the mode of approach of the dipolarophile.15
Specifically, when a two-atom linker is used, only the
endo approach of the dipolarophile is observed, whereas
with longer tethers, an exo approach becomes preferred.
For the synthesis of 1-epiaustraline, a two-atom tether
is needed to provide the proper relative configuration;
however, both atoms of the tether must be converted to
hydroxyl substituents in the final product. By using a
silicon-oxygen tether, the two hydroxyls can be revealed
at a late stage of the synthesis by the agency of a Tamao-
Fleming oxidation.16
Syn th etic Design
The highly oxygenated pyrrolizidine structure of 1-epi-
australine provides the main challenge to its synthesis.
Specifically, this involves the stereocontrolled installation
of hydroxyl groups at four of the eight carbons of the
alkaloid. This necessitates the use of appropriately
oxygenated components in the tandem [4 + 2]/[3 + 2]
nitroalkene cycloaddition. This approach is outlined in
Scheme 2.
We envisioned the construction of the pyrrolizidine core
by alkylation during the reductive cleavage of the nitroso
acetal i. To incorporate the hydroxymethyl substituent
at the C(3) position of 1, the installation of the leaving
group for the alkylation at C(3) in the appropriate
configuration was required. Because this center is not
created as part of the tandem cycloaddition, we needed
to make use of either resident stereodirecting functions
in the nitroso acetal, asymmetric reagents, or both. The
chemical process chosen to introduce the center and the
requisite functional group was the dihydroxylation of a
Under the direction of the tether, the facial approach
of the dipolarophile is then dependent on the configura-
tion of the C(4) substituent of the nitronate iii, which is
set during the [4 + 2] cycloaddition. This requires the
use of a chiral vinyl ether dienophile that will approach
(9) Denmark, S. E.; Parker, D. L.; Dixon, J . A. J . Org. Chem. 1997,
62, 435.
(10) Denmark, S. E.; Thorarensen, A. J . Org. Chem. 1994, 59, 5672.
(11) Denmark, S. E.; Throarensen, A. J . Am. Chem. Soc. 1997, 119,
125.
(12) For a recent review, see: Rudchenko, V. F. Chem. Rev. 1993,
93, 725.
(13) Denmark, S. E.; Martinborough, E. A. J . Am. Chem. Soc. 1999,
121, 3046.
(15) Denmark, S. E.; Moon, Y.-C.; Senanayake, C. B. W. J . Am.
Chem. Soc. 1990, 112, 311.
(16) For a recent review see: J ones, G. R.; Landais, Y. Tetrahedron
1996, 52, 7599.
(14) Denmark, S. E.; Hurd, A. R. J . Org. Chem. 2000, 65, 2875.

4278 J . Org. Chem., Vol. 66, No. 12, 2001
Denmark and Cottell
the Re face of the â-oxygenated nitroalkene to establish
the correct configuration. This can be accomplished in
either exo or endo mode [4 + 2] cycloadditions with the
appropriate configuration of auxiliary because the ste-
reocenter at C(6) of the nitroso acetal is destroyed during
the reductive cleavage of i. However, the configuration
of this acetal does influence the approach of the dipo-
larophile during the [3 + 2] cycloaddition favoring
reaction on the face opposite to that of the C(6) substitu-
ent.17 This preference can even manifest itself when the
dipolarophile is tethered to the nitronate.13 In view of
these facts, we opted to set the R configuration at the
C(6) center of iii by an exo approach of the dienophile.
Our previous investigations have clearly established a
strong preference for the exo approach of the dienophile
when MAPh (methylaluminum bis(2,6-diphenylphenox-
ide)) is used as the Lewis acid. The absolute configuration
of the nitronate is dependent on the choice of the chiral
vinyl ether 4 used as the dienophile. The vinyl ether
derived from trans-phenylcyclohexanol is known to afford
synthetically useful diastereoselectivities in combination
with MAPh,18 such that the resulting nitronate bears a
1,3-ul relationship.19 Therefore, the use of (1S,2R)-
phenylcyclohexanol is necessary to provide the correct
absolute configuration.
Resu lts
P r ep a r a tion of Nitr oa lk en e 3. 1-Lithiobutadiene
was generated from 1-bromobutadiene (5) by the bromine-
lithium exchange with tert-butyllithium in Et2O at -75
°C. The choice of silicon electrophile was guided by the
need for a sufficiently reactive group that would also be
stable enough for handling and isolation. Orienting
studies showed that the use of tert-butyl groups pre-
vented the reaction between the dichlorosilane and
1-lithiobutadiene.21 On the other hand, reaction of
1-lithiobutadiene with dimethyldichlorosilane provided
a product very labile to hydrolysis. Ultimately, we found
that isopropyl groups provided the best balance between
reactivity and stability. Initial attempts at the silylation
of 1-lithiobutadiene with dichlorodiisopropylsilane were
plagued by low yields and oligomeric byproducts. At best,
a 35% yield of the desired chlorosilane was obtained
(Scheme 3). We surmised that 9 may not be stable under
the reaction conditions, leading to the polymerization.
Analogous silylation of vinyllithium showed consistently
higher yields for monochlorosilanes over their dichloro
counterparts.22 Accordingly, the addition of 1-lithiobuta-
diene to chlorodiisopropylsilane proceeded in a gratifying
73% yield to provide 10 under similar conditions.
Sch em e 3
The main component in the tandem inter[4 + 2]/intra-
[3 + 2] cycloaddition is the nitroalkene 3, Scheme 2. The
synthesis of this nitroalkene was envisioned to follow
from the addition of potassium nitroacetaldehyde to a
butadienylsilyl chloride, thereby creating the tether. The
silyl chloride could in turn be made from addition of a
metallobutadiene to a dichlorosilane, e.g., 6. Selection of
the spectator groups on the silicon tether is important
as they will influence both the reactivity of the dichlo-
rosilane toward the metallobutadiene and the stability
of the nitroalkene.
The key question raised by this strategy is the ability
of the nitroso acetal to undergo clean reductive hydro-
genation to a pyrrolidine that can then suffer intramo-
lecular alkylation when constrained by the two atom
tether. In two previous syntheses,8,20 the conforma-
tional restraint imposed by the two atom tether has led
to problems in the hydrogenolysis of the particular
nitroso acetals (vide infra). In both of these cases, the
pyrrolizidine systems were to be formed by acylation of
the intermediate pyrrolidine. The proposed synthesis of
1-epiaustraline involves an alkylative cyclization for
the closure that in our experience is generally slower
than acylation. Thus, cognizant of a serious potential
pitfall, we nonetheless embarked on the synthesis of
1-epiaustraline. As detailed below, these concerns were
justified and necessitated a revision of the synthetic
strategy that not only delivered the target compound but
also expanded our understanding of the chemistry of
nitroso acetals.
The use of chlorodiisopropylsilane obviated the problem
of selective C-Si bond formation, but delivered a silyl-
diene 10 devoid of a leaving group necessary for the
construction of the nitroalkene. This was easily remedied
by smooth oxidation of the silicon unit with a solution of
chlorine in CCl4 to afford the chlorosilane 9 in 89% yield.
Interestingly, the chlorination takes place cleanly at the
silicon center in preference to the diene. However, the
use of excess chlorine or extended reaction times did lead
to over-chlorination and the addition of HCl to the diene.
With the silicon center suitably activated, the forma-
tion of the Si-O linkage proceeded readily upon admix-
ture of 9 with a slurry of the potassium salt of nitro-
acetadehyde (7) at 0 °C, over the course of 5 h. Although
(17) Denmark, S. E.; Seierstad, M.; Herbert, B. J . Org. Chem. 1999,
64, 884.
(18) Denmark, S. E.; Seierstad, M. J . Org. Chem. 1999, 64, 1610.
(19) This terminology refers to the relationship between the con-
figuration of C(1) of the auxiliary and the configuration of C(6) of the
resulting nitroso acetal. In this case, the auxiliary possessing an R
configuration will typically give a product with an S configuration at
the acetal center. Hence, this is designated an 1,3-unlike (ul) relation-
ship.18
1
the nitroalkene 3 could be identified by H NMR spec-
troscopy, the product was not amenable to purification
by either chromatography or distillation (affording the
(21) Gregory Morris is thanked for early investigations into the
synthesis of nitroalkene 3.
(22) Auner, N. Z. Anorg. Allg. Chem. 1988, 558, 87.
(20) Denmark, S. E.; Hurd, A. R.; Sacha. H. J . Org. Chem. 1997,
62, 1668.

Synthesis of (+)-1-Epiaustraline
J . Org. Chem., Vol. 66, No. 12, 2001 4279
dienylsilanol 11) and was therefore used directly in the
tandem cycloaddition.
structure of the nitroso acetal, the six-membered ring is
expected to adopt a ground-state boat conformation.23 In
12, the signal for (HC(6)) appeared as a triplet with a
coupling value of 7.3 Hz, which suggested that the
auxiliary is in an pseudoaxial position, by analogy to
previously studied nitroso acetals.13,20 The signal for HC-
(7a) showed very small couplings to the protons at C(7)
(3.6, 2.2 Hz). Therefore, HC(7a) must be gauche to the
two protons of C(7), which may be accomplished by
positioning the oxygen substituent at C(7a) in a pseudo-
equatorial position.
Ela bor a tion of Nitr oso Aceta l 12. The next stage
of the synthetic plan required the functionalization of the
pendant alkene to provide for the formation of the
pyrrolizidine and creation of the C(3) hydroxymethyl
group. Initial attempts at the diastereoselective dihy-
droxylation of 12 with K2OsO4‚2H2O and K3Fe(CN)6
proceeded slowly to give a 3/1 mixture of diols (Scheme
4). The major isomer (13a ) was assigned the R configu-
ration on the basis of related precedent in these labora-
tories13 and others24 for the dihydroxylations of terminal
olefins bearing allylic heteroatoms. This assignment was
later confirmed by X-ray analysis of an advanced inter-
mediate, 17.
Ta n d em Cycloa d d ition . With the desired nitroalk-
ene 3 in hand, the tandem cycloaddition with the chiral
vinyl ether 4 was investigated. In the presence of MAPh
in toluene solvent, the cycloaddition proceeded over 14
h with gradual warming from -70 to 0 °C, to provide a
60% yield of a single cycloadduct (entry 1, Table 1).
Optimization involved changes in solvent and tempera-
ture, as well as the stoichiometry of the Lewis acid and
vinyl ether. The use of methylene chloride (an alternative
solvent for MAPh-promoted cycloadditions) under the
same conditions led to a decreased yield of the cycload-
duct (Table 1, entry 2). To increase the rate of reaction,
changes in the reaction temperature were investigated.
Interestingly, by raising the temperature of the reaction,
the yields again decreased. The only beneficial change
in the reaction profile was to maintain the temperature
at -75 °C over the course of 14 h, which provided a 66%
yield of the cycloadduct (Table 1, entry 5).
Ta ble 1. Op tim iza tion of Cycloa d d ition of 3 a n d 4
Ta ble 2. Selectivities in th e Ca ta lyzed Dih yd r oxyla tion
of 12
entry
ligand
none
selectivity 13a /13b
1
2
3
4
5
6
7
8
9
3/1
(DHQD)2PHAL
(DHQD)2AQN
(DHQD)2PYR
DHQD-MEQ
DHQD-IND
DHQD-PHN
DHQD-PHN
DHQ-CLB
4.3/1
2.2/1
1.9/1
1/1
1/2.6
1/2.8
1/2.6a
2.7/1
4.1/1
1.8/1
entry 4 (equiv) MAPh (equiv) solvent
T, °C
yield, %a
1
2
3
4
5
6
7
8
9
2
2
2
2
2
3
1.25
1.25
1.5
2
2
2
2
2
2
3
2
1.5
1.5
2
toluene -70 f 0
CH2Cl2 -70 f 0
toluene -55
60
40
53
55
66
33
47
16
41
49b
toluene -70 f 22
toluene -70
toluene -70
toluene -70
10
11
(DHQ)2PYR
(DHQ)2PHAL
toluene -70
toluene -70
a
Preparative scale.
10
toluene -70
a
b
Yield for two steps. Preparative scale.
Clearly, a 3/1 ratio favoring the undesired isomer is
not satisfactory. To reverse this selectivity and produce
the desired diol in synthetically useful amounts, we made
recourse to the asymmetric dihydroxylation reaction.25
This strategy had been successfully implemented in the
previous synthesis of australine and 3-epiaustraline in
these laboratories.13 On the basis of the general mne-
monic developed by Sharpless for the asymmetric dihy-
droxylation, the dihydroquinidine (DHQD) family of
ligands were predicted to provide the diol in the desired
S configuration (13b). Thus, a variety of DHQD-based
ligands26 were surveyed (Table 2). Unfortunately, most
of the ligands exerted at best a modest influence on the
selectivity of the reaction. Some of the DHQD ligands
even enhanced the production of the undesired diaste-
reomer. In view of this fact, several dihydroquinine
(DHQ) ligands were investigated as well; however, they
too provided little or no selectivity. The best ligand found
was a DHQD ligand with an phenanthracene modifier.27
This additive was able to reverse the selectivity, resulting
in a 2.6/1 ratio favoring 13b. The diols 13a and 13b could
The initial stoichiometry (2 equiv of 4 and 2 equiv of
MAPh per nitroalkene) was chosen from previous experi-
ence with related tandem cycloadditions.20 Decreasing the
relative amounts of Lewis acid and vinyl ether were
investigated to increase the ease of workup for the
cycloaddition. However, lowering the amount of either
the Lewis acid or the chiral vinyl ether was detrimental
(Table 1, entries 7-9). Lower yields were also obtained
when the amount of the Lewis acid was increased (Table
1, entry 6). Unfortunately, increasing the scale of the
cycloaddition also lead to lower yields due to complica-
tions associated with chromatographic purification (cf.
Table 1, entries 5 and 10).
The structure of cycloadduct 12 was assured by com-
plete spectroscopic characterization. The full stereostruc-
ture, assumed by analogy to related systems,20 was
ultimately established by single-crystal X-ray analysis
of an advanced intermediate (vide infra). The presence
of three vinyl signals, and the absence of dienyl internal
proton signals, indicated that the [3 + 2] cycloaddition
had taken place spontaneously before purification. The
relative configuration of the cycloadduct was tentatively
(23) X-ray analysis of 17 subsequently bore this out.
(24) Cha, J . K.; Christ, W.; Kishi, Y. Tetrahedron 1984, 40, 2247.
(25) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem.
Rev. 1994, 94, 2483.
(26) Table 2 only contains a partial list of ligands employed. For
the complete listing, refer to the Supporting Information.
1
assigned to be that of 12 by comparison of the H NMR
coupling values of the acetal proton (HC(6)) and the
proton at C(7a) with those at C(7). Because of the tricyclic

4280 J . Org. Chem., Vol. 66, No. 12, 2001
Denmark and Cottell
Sch em e 4
be separated via preparative MPLC; however, they were
taken on as a mixture and separated at the next stage.
Selective protection of the primary hydroxyl group in
13 proceeded readily with TBSOTf in pyridine to provide
the TBS ether (14) in 75% yield, along with a small
amount of the bis TBS ether. The two diastereomers
could be easily separated by silica gel column chroma-
tography. Activation of the secondary alcohol was ac-
complished by the action of methanesulfonic anhydride
in pyridine to provide the mesylate (15) in 80% yield.
Hyd r ogen olysis a n d Com p letion of th e Syn th esis.
With the secondary center in 14 now suitably activated,
the stage was set for the critical unmasking of the nitroso
acetal to form the pyrrolizidine core structure. Initial
hydrogenation of 15 was performed in the presence of
activated Raney nickel (RaNi), under a hydrogen pres-
sure of 160 psi in methanol at room temperature.
Unfortunately, a complex mixture of products was ob-
tained, from which only the chiral auxiliary could be
isolated and identified, Scheme 5. This outcome was
independent of the reaction time or hydrogen pressure;
however, the yield of auxiliary ranged from moderate to
high. This suggested that the N-O bonds were cleaved
during the course of the reaction, but the cyclization was
not proceeding as envisioned.
Last, the cleavage of the N-O bonds reveals a â-hydroxy
silicon unit. This provides the opportunity for a Peterson-
type olefination28 that would cleave the silyl tether,
leaving an allylic mesylate.
To address the potential problems associated with the
nucleofuge during the ring closure, several other leaving
groups were investigated. The corresponding triflate (16)
and tosylate (17) were made from their respective
anhydrides in pyridine. Interestingly, the tosylate solidi-
fied after column chromatography. Recrystallization from
MeOH provided X-ray-quality crystals that revealed that
the absolute and relative configuration were assigned
correctly.29
Hydrogenation of 16 in the presence of Raney nickel
provided the same result as the mesylate, where no
pyrrolizidine-containing products are observed, but the
chiral auxiliary was recovered in good yield. However the
hydrogenation of 17 behaved differently. In addition to
the recovered chiral auxiliary, a second product was
isolated in small amounts as well. The 1H NMR spectrum
of this product showed distinct olefinic peaks, and the
structure was tentatively assigned as the pyrrolizidine
18, Scheme 5. This suggested that the Peterson-type
olefination process was interfering with the desired
cyclization pathway.
To circumvent the problems observed during hydroge-
nation, it seemed prudent to cleave the silicon-oxygen
linkage. Excision of this bond would help alleviate
conformational restrictions that the tether may impose
on the ring closure. This could then lead to a preference
of the cyclization over the olefination pathway. By careful
consideration of the overall strategy, an attractive alter-
native was identified that would introduce no additional
steps. If we could perform the Tamao-Fleming oxidation
on the nitroso acetal, then the strain arising from the
Si-O linkage would be lost along with the possibility of
Peterson olefination during hydrogenolysis. Although
simple in concept, this idea presented unprecedented
chemical compatibility challenges for the delicate nitroso
acetal.
Sch em e 5
The failure to obtain pyrrolizidine-derived products
could be caused by several problems. First, the leaving
group may not be compatible or active enough to effect
the closure of the pyrrolizidine ring. Second, the ring
closure may fail due to an inability to access a conforma-
tion appropriate for the displacement of the mesylate by
the nitrogen, due to the presence of the two atom tether.
To avoid an elaborate protection scheme, the oxidation
of the nitroso acetal would have to be performed with
the leaving group already installed. Of the leaving groups
previously utilized, it was believed that the tosylate had
the best chance of surviving the oxidation conditions. The
(28) Ager, D. J . Org. React. 1990, 38, 1.
(29) The crystallographic coordinates of 17 have been deposited with
the Cambridge Crystallographic Data Centre; deposition no. CCDC
157533.
(27) The superiority of the PHN ligand for asymmetric dihydroxy-
lation of vinyl groups bearing allylic oxygen substituents has been
noted. Oi, R.; Sharpless, K. B. Tetrahedron Lett. 1992, 33, 2095.

Synthesis of (+)-1-Epiaustraline
J . Org. Chem., Vol. 66, No. 12, 2001 4281
tosylate 17 was therefore subjected to the normal Tamao-
Fleming oxidation conditions of KF, H2O2, and KHCO3
in a 1/1 MeOH/THF mixture. We were pleased to find
that the silicon group was effectively removed while
leaving the nitroso acetal intact. However, under these
conditions the TBS groups was also removed, resulting
in the formation of the terminal epoxide. Previous studies
by Tamao have shown that the offending reagent, KF, is
not, in fact, necessary to activate the silicon unit of an
oxasilacyclopentane.30 In practice, the oxidation of 17
could be successfully carried out in the absence of KF
(with mild heating) and cleanly afforded the targeted diol
without cleavage of the TBS group (Scheme 6). This
transformation was somewhat capricious, which we
ascribed to changes in concentration during the lengthy
reaction times. However, attempts to maintain a constant
concentration by carrying out the reaction in a sealed
tube led to no improvement. Interestingly, in a normal
round-bottom flask equipped with a condenser, the
reaction proceeded smoothly over 48 h to provide 19 in a
90% yield. Attempts to accelerate the process at higher
temperatures or with stronger bases led only to decom-
position.
natural or synthetic material (see the Supporting Infor-
mation). Interestingly, the optical rotation of our final
product ([R]D ) 13.7 (c ) 1.72, H2O) more closely matched
the rotation for synthetic material ([R]D ) 13.3 (c ) 0.1,
H2O);5 ([R]D ) 12.5 (c ) 0.6, H2O)6) than the natural
product (([R]D ) 8.5 (c ) 0.41, H2O);1 ([R]D ) 12.0 (c )
1.17, H2O)2).
Discu ssion
Ta n d em [4 + 2]/[3 + 2] Cycloa d d ition . Although not
unexpected in view of our extensive experience with this
reaction, it was nonetheless very satisfying that the
tandem cycloaddition of 3 and 4 provided a single nitroso
acetal (12) with the correct relative and absolute config-
uration. To produce this specific diastereomer, (1S,2R)-
2-phenylcyclohexyl vinyl ether (4) most likely adopts a
reactive s-trans conformation. This is in accordance with
computational studies,32 as well as previous experimental
results.18 In this conformation, the phenyl group shields
the Si face of the vinyl ether, Figure 1. Therefore, the
[4 + 2] cycloaddition will take place on the Re face of the
vinyl ether. It is now well established that MAPh enforces
an exo transition structure in these cycloadditions, which
when combined with the Re face preference for the
dienophile results in attack on the Si face of the nitroalk-
ene, thus establishing the desired S absolute configura-
tion at C(7).
Sch em e 6
After decomplexation of the nitronate upon the addi-
tion of MeOH, the intramolecular dipolar cycloaddition
occurs spontaneously to yield the nitroso acetal 12. The
dipolarophile must approach from the same side to which
it is tethered, thus controlling the facial selectivity of the
[3 + 2] cycloaddition and therefrom the relative config-
uration of C(7) and C(1) in 1-epiaustraline. The use of a
two atom tether also restricts the dipolarophile to ap-
proach in an endo mode (defined by the silicon substitu-
ent) as well as eliminates the regio-reversed cycloaddi-
tion. Therefore, the relative configuration of C(7) and
C(7a) is also set by the dipolar cycloaddition.
The yield for the three steps in this sequence ((1) silyl
chloride displacement, (2) [4 + 2] cycloaddition, and (3)
[3 + 2] cycloaddition) was lower than expected. This can
primarily be attributed to the lability of the nitroalkene
3, which is believed to slowly decompose at a rate
competitive with the cycloaddition.33 Unfortunately, at-
tempts to selectively increase the rate of the cycloaddi-
tion by increasing the reaction temperature did not lead
to increased yields of 12. Lower yields were also obtained
when the reaction was conducted with less than 2 equiv
of the Lewis acid or the chiral vinyl ether. Here, the rate
of the tandem cycloaddition decreased, while the rate of
decomposition of 3 remained constant, leading to yields
of 12 as low as 16%. The pathway of decomposition can
be envisioned as either attack by a nucleophile at the
silicon unit or in a Michael-type addition on the nitro
alkene portion, followed by elimination of the silanol.
Water is a likely culprit in this decomposition; however,
the silanol may be competent as well, and the process
can be initiated by very small amounts of water. This
would lead to oligomeric butadienylsilicon species that
can be hydrolyzed during workup of the reaction.
The hydrogenation of 19 was performed in the presence
of Raney nickel for 48 h under a pressure of 250 psi of
hydrogen. With the silicon tether removed, the hydroge-
nation provided a mixture of the desired pyrrolizidine 17
and an intermediate pyrrolidine 16. After filtration of the
catalyst, the mixture was warmed to 50 °C in MeOH over
5 h to complete the conversion of 20 to 21. Purification
of 21 by silica gel chromatography provided the penul-
timate intermediate. Deprotection of 21 with HF/MeOH
provided 1-epiaustraline in a 55% yield over the two
1
steps. Comparison of the H NMR data of 1 to those of
an authentic sample of the natural product provided by
Nash31 were nearly superimposable, and all other spec-
troscopic and physical data matched those of either
(30) Tamao, K.; Ishido, N.; Kumada, M. J . Org. Chem. 1983, 48,
2120.
(31) The sample of 1-epiaustraline provided by Nash was initially
believed to be australine. Subsequent reassignment proved this
material to be 1-epiaustraline.
(32) Liu, J .; Niwayama, S.; Houk, K. N. J . Org. Chem. 1998, 63,
1064.
(33) The observation of the silanol 11 in the 1H NMR spectrum of
the crude cycloadduct provides support for this contention.

4282 J . Org. Chem., Vol. 66, No. 12, 2001
Denmark and Cottell
F igu r e 1. Proposed transition structures for the tandem cycloaddition.
Dih yd r oxyla t ion s. The diastereoselectivities ob-
served in the simple dihydroxylation of 12 were disap-
pointing, but not unexpected. In diastereoselective reac-
tions of terminal alkenes bearing a heteroatom at an
allylic position, the general preference is for the formation
of the “erythro” isomer.24 This was also observed in the
dihydroxylation of a similar nitroso acetal in a previous
study.11 The diastereoselection was explained by reaction
through the conformer in which the allylic proton was
in the plane of the double bond, thus minimizing allylic
strain. Approach of the oxidant is then directed to the
face opposite to the heteroatom.
The weak influence of the tricyclic nitroso acetal on
the asymmetric induction event can be rationalized by
examining the calculated ground-state conformations of
the cycloadduct. An MM2 conformational search provided
eight conformations within 1 kcal/mol of the energy
minimum and 120 conformations within 2 kcal/mol. The
lowest eight conformations are overlaid in Figure 2.
Though many of these conformations varied in the
position of the isopropyl groups and the chiral auxiliary,
there was a high degree of rotational freedom observed
for the vinyl group. This suggests that many reactive
conformations are available for the vinyl group during
the dihydroxylation, and therefore, the core of the nitroso
acetal will provide little bias.
in the selectivities with nitroso acetal 12. In this case,
the monomeric ligands consistently gave higher propor-
tions of the desired diol 13b. This suggests that the
nitroso acetal may be unable to fit properly into the
reactive pocket created by the OsO4-complexed dimeric
ligands.
Hyd r ogen a tion s. The hydrogenolytic cleavage of ni-
troso acetals is a very powerful, albeit inadequately
understood, transformation. This process allows for the
conversion of the nitroso acetal to several azacyclic
structures applicable to the synthesis of alkaloids. From
the products, and in some cases the byproducts, it is
assumed that this transformation proceeds through the
amino alcohol generated upon N-O bond cleavage.
Reductive amination of the hemiacetal and either acy-
lation or alkylation then provide the observed bicyclic
products. However, the timing of the each N-O bond
cleavage and the subsequent cyclizations is not known.
It is the believed that the dependence on hydrogen
pressure observed in some cases is due to the availability
of several different pathways. Not suprisingly, the hy-
drogenation of nitroso acetals has been found to be highly
substrate dependent.
Sch em e 7
Although the hydrogenation of nitroso acetals generally
proceeds smoothly to the desired product, this is not
always the case, especially when the nitroso acetal is
fused to another ring. This problem was first noticed in
early studies of the hydrogenation of 22 in which the
nitroso acetal is trans fused to six-membered carbocycle
(Scheme 7).15 Hydrogenation of the analogous nitroso
acetal 25 (with a cis fusion to a five-membered ring)
proceeded uneventfully at atmospheric pressure to the
tricyclic lactam 26. However, similar treatment of 22 left
the oxazine ring intact. Elevated hydrogen pressures (160
psi) were necessary to cleave both N-O bonds to afford
the intermediate 23, where the acylation had not taken
F igu r e 2. Overlay of the eight lowest minimum energy
conformations calculated for 12.
The low selectivity observed in the presence of chiral
ligands was again disappointing in light of our previous
success,13 but overall, not surprising. Terminal alkenes
commonly afford the lowest selectivities in the asym-
metric dihydroxylation reaction25 and occasionally pro-
vide selectivities contrary to the sense predicted by the
empirical model.34 There is an interesting trend, however,
(34) Boger, D. L.; McKie, J . A.; Nishi, T.; Ogiku, T. J . Am. Chem.
Soc. 1997, 119, 311.

Synthesis of (+)-1-Epiaustraline
J . Org. Chem., Vol. 66, No. 12, 2001 4283
Sch em e 10
place. Conversion to the lactam required heating in
refluxing toluene for 42 h. The difference in reactivity
was attributed to the formation of a trans ring fusion in
the case of the three-atom tether.
Other instances of problems during the hydrogenation
step were encountered on route to the synthesis of
rosmarinecine11 and detoxinine.20 In the case of rosmari-
necine, hydrogenation of the cycloadduct 27 containing
a fused lactone ring led to a complex mixture of products
devoid of the desired tricyclic lactam 28 (Scheme 8). It
was surmised that the conformational restriction imposed
by the sp2-hybridized carbon and short C-O bond lengths
prevented the acylation process. This was alleviated by
prior reduction of the carbonyl group to a hemiacetal.
Hydrogenation of the less strained nitroso acetal 29
proceeded in 64% yield.
Sch em e 11
Sch em e 8
In the synthesis of detoxinine, hydrogenation of the
intermediate 31 proceeded to give the desired lactam 32,
along with unsaturated side product 33 (Scheme 9). The
ratio of 32 to 33 was found to be independent of changes
in several reaction conditions. The byproduct 33 is
believed to be a result of a Peterson-type olefination.35
However by resubjecting the tricyclic lactam to the
hydrogenation conditions, it was found that the olefina-
tion does not occur after lactam formation. Thus, the
elimination must be a competing process during the
hydrogenation. Here again, the fused ring (this time a
siloxane) hampers the acylation event and provides an
undesirable pathway of destruction.
Sch em e 9
In all the cases discussed above, the final cyclization
event is the formation of a lactam, whereas in the case
at hand, the second ring closure requires an intra-
molecular alkylation. Typically, alkylations (especially
with secondary centers) proceed more slowly than the
corresponding acylations, as was observed in the syn-
thesis of australine (Scheme 10).13 Hydrogenation of the
nitroso acetal 34, in which a six-membered silyl acetal
is embedded, provided the intermediate 35. This inter-
mediate did not access a secondary reaction pathway, and
closure of the pyrrolizidine ring was effected upon heating
in refluxing acetonitrile for 16 h.
The problems encountered during the synthesis of
1-epiaustraline described herein embody various aspects
of the problems described in the foregoing examples
(Scheme 11). In the hydrogenation of 17 (15 or 16),
recovery of the chiral auxiliary indicated that the N-O
(35) A similar Peterson-type olefination has been observed upon
hydrogenolysis of other silicon-bridged nitroso acetals. Righi, P.;
Marotta, E.; Rosini, G. Chem. Eur. J . 1998, 4, 2501.

4284 J . Org. Chem., Vol. 66, No. 12, 2001
Denmark and Cottell
bonds of the nitroso acetal were being cleaved. The
resulting amino aldehyde likely underwent intramolecu-
lar reductive amination to provide the intermediate iv;
however, the alkylation did not proceed. Presumably, the
amine cannot access a conformation favorable for the
displacement of the leaving group due to the presence of
the silicon-oxygen linkage. However, iv can access a
secondary pathway where the alcohol formed from hy-
drogenolysis can engage in a Peterson olefination. The
result is an allylic sulfonate that presumably proceeded
on to unidentifiable products. By installing a better
leaving group (triflate 15), it was hoped that N-alkylation
could occur in preference to elimination, but this was not
the case. The use of a less active leaving group (tosylate
17) allowed the intermediate allylic tosylate to survive
long enough to proceed to the unsaturated pyrrolizidine
18.
According to this analysis the silyl ether tether plays
a dual role in thwarting the success of the hydrogenolysis.
First, it conformationally restricts the tosylate from
achieving a proper alignment for displacement and
second, it provides an facile and deleterious shunt to
undesired products. Thus, removal of the silicon was the
ideal solution as it eliminated the germ of destruction
and added no steps to the synthesis. The remarkably
high-yielding Tamao-Fleming oxidation revealed the
oxidative and hydrolytic stability of the nitroso acetal
function and allowed for the synthesis to be completed.36
Indeed, hydrogenolysis of 19 did provide the desired
product 21 along with amino triol 20 thereby verifying
that the intramolecular alkylation was slow even in the
absence of the silicon tether. The sluggishness of the
alkylation can be understood on the basis of the well-
known rate retarding effects of electronegative â-sub-
stituents on SN2 reactions.37 It is also conceivable that
intramolecular hydrogen bonding restricts access to
conformations with a suitable alignment of the groups.
Con clu sion
1-Epiaustraline was synthesized in 10 steps from
1-bromobutadiene in a 7.0% overall yield. This synthesis
further demonstrates the applicability of the tandem
[4 + 2]/[3 + 2] nitroalkene cycloaddition to the synthesis
of alkaloids. All but one of the stereocenters present in
1-epiaustraline are created in the tandem cycloaddition
step, with very high stereocontrol. The remaining ste-
reocenter is introduced via a diastereoselective dihy-
droxylation. The demonstration that the Tamao-Fleming
oxidation can be done prior to the hydrogenation step
further expands repertoire of transformations tolerated
by nitroso acetals and allows for more flexible synthetic
planning.
Ack n ow led gm en t. We are grateful to the National
Institutes of Health (GM 30938) for generous financial
support. J .J .C. thanks the Procter and Gamble Co. for
a graduate fellowship. We are also grateful to Professor
Robert Nash (Aberswyth) for providing a sample of
natural 1-epiaustraline and to Dr. Esther Martinbor-
ough for the purification and characterization of this
sample.
Su p p or tin g In for m a tion Ava ila ble: Complete experi-
mental details with full spectroscopic and analytical data for
all new compounds along with comparison 1H NMR spectra
of synthetic and natural (+)-1-epiaustraline. A full listing of
ligand structures and dihydroxylation selectivities is provided.
This material is available free of charge via the Internet at
http://pubs.acs.org.
J O0101510
(36) For a previous example of Tamao-Fleming type oxidation in
the presence of a nitroso acetal see ref 35.
(37) (a) Bunton, C. A. Nucleophilic Substitution at a Saturated
Carbon Atom, Elsevier: Amsterdam, 1963. (b) Parker, R. E.; Isaacs,
N. S. Chem. Rev. 1958, 58, 737.