Asymmetric Total Synthesis of Dendrobatid Alkaloids: Preparation of Indolizidine 251F and Its 3-Desmethyl Analogue Using an Intramolecular Schmidt Reaction Strategy

Article abstract of DOI:10.1021/ja0320018

Total syntheses of alkaloid 251F (1), a natural product detected from the skin extracts of the dendrobatid frog species Minyobates bombetes, and its racemic 3-desmethyl derivative (2) are reported. A Diels-Alder reaction initiated both syntheses and established four consecutive stereogenic centers. Important to the synthesis of 2 was a first-generation ozonolysis/olefination/aldol strategy to convert a [2.2.1] bicyclic acid to the [3.3.0]bicyclooctane diquinane 4b. Further elaboration to an appropriate keto azide allowed for a key intramolecular Schmidt reaction to deliver the tricyclic core of the target molecule. In a second-generation approach, a tandem ring-opening/ring-closing metathesis reaction effected an overall [2.2.1] → [3.3.0] skeletal rearrangement to deliver diquinane 4a. In similar fashion, 4a was manipulated to an appropriate keto azide, and an intramolecular Schmidt reaction generated the core cyclic architecture of 251F.

Full text of DOI:10.1021/ja0320018

Published on Web 04/09/2004
Asymmetric Total Synthesis of Dendrobatid Alkaloids:
Preparation of Indolizidine 251F and Its 3-Desmethyl
Analogue Using an Intramolecular Schmidt Reaction Strategy
Aaron Wrobleski, Kiran Sahasrabudhe, and Jeffrey Aube´*
Contribution from the Department of Medicinal Chemistry, UniVersity of Kansas,
Room 4070, Malott Hall, 1251 Wescoe Hall DriVe, Lawrence, Kansas 66045-7582
Received December 30, 2003; E-mail: jaube@ku.edu
Abstract: Total syntheses of alkaloid 251F (1), a natural product detected from the skin extracts of the
dendrobatid frog species Minyobates bombetes, and its racemic 3-desmethyl derivative (2) are reported.
A Diels-Alder reaction initiated both syntheses and established four consecutive stereogenic centers.
Important to the synthesis of 2 was a first-generation ozonolysis/olefination/aldol strategy to convert a [2.2.1]
bicyclic acid to the [3.3.0]bicyclooctane diquinane 4b. Further elaboration to an appropriate keto azide
allowed for a key intramolecular Schmidt reaction to deliver the tricyclic core of the target molecule. In a
second-generation approach, a tandem ring-opening/ring-closing metathesis reaction effected an overall
[2.2.1] f [3.3.0] skeletal rearrangement to deliver diquinane 4a. In similar fashion, 4a was manipulated to
an appropriate keto azide, and an intramolecular Schmidt reaction generated the core cyclic architecture
of 251F.
in 1992.7 The skin extract from M. bombetes caused severe
Introduction
locomotor difficulties, muscle spasms, and convulsions upon
For decades, amphibious sources have provided a rich array
of natural product targets, particularly alkaloids, unprecedented
in other biological systems.1 Most likely used in chemical
defense mechanisms, these alkaloids cover a wide spectrum of
structural complexity and biological activity. Daly and co-
workers have shown the dendrobatid family of frogs to be a
highly prolific source of alkaloids.1-3 To date, over 500 of these
alkaloids have been detected in skin extracts of the dendrobatid
frogs,4 and nearly two dozen structural classes of alkaloids have
been identified.3 Some well-represented structural classes include
the batrachotoxins (the class of alkaloids used as dart poisons),
the histrionicotoxins and pumiliotoxin-A classes (a large family
of bicyclic alkaloids), and the pyridine alkaloids (most notably,
epibatidine).1 Many of the individual dendrobatid alkaloids have
been found in multiple frog species. Thus, a single alkaloid may
be widely distributed among numerous species of dendrobatid
frogs. The cyclopenta[b]quinolizidine alkaloids, however, have
thus far been detected in only one species. Originally described
as Dendrobates bombetes,5 the taxonomic classification of this
small, poisonous Colombian frog species was then changed to
Minyobates bombetes (M. bombetes).6 Alkaloid 251F (1) was
characterized as the major alkaloid of the skin extract of M.
bombetes, and its structure was elucidated primarily via mass
spectroscopy and NMR studies as reported by Daly and Spande
injection into mice.3 It is quite possible the observed biological
effects were due to other alkaloids present in the skin extract,
so nothing is definitively known regarding the biological activity
of 251F.
With undetermined biological activity and a highly intriguing
chemical structure including three fused rings and seven
stereogenic centers (six on contiguous carbons), 251F presents
a distinct challenge for total synthesis. In 1995, Taber and You
reported the first synthesis of 251F utilizing a highly diaste-
reoselective rhodium-catalyzed construction of a key cyclopen-
tane intermediate.8 For over a decade, our laboratory has
investigated the intramolecular Schmidt reaction as a method
for the construction of heterocycles bearing nitrogen at ring
fusion positions.9,10 However, this reaction has only recently
been applied to the synthesis of complex alkaloids.11-14
Because of its structure and (as yet uncertain) potential
biological activity, we identified alkaloid 251F as a target for
total synthesis.15 Retrosynthetically, a key step would be the
intramolecular Schmidt reaction that would entail the conversion
of keto azides such as 3 to the lactam derivative of the target 1
(7) Spande, T. F.; Garraffo, H. M.; Yeh, H. J. C.; Pu, Q.-L.; Pannell, L. K.;
Daly, J. W. J. Nat. Prod. 1992, 55, 707-722.
(8) Taber, D. F.; You, K. K. J. Am. Chem. Soc. 1995, 117, 5757-5762.
(9) Aube´, J.; Milligan, G. L. J. Am. Chem. Soc. 1991, 113, 8965-8966.
(10) Milligan, G. L.; Mossman, C. J.; Aube´, J. J. Am. Chem. Soc. 1995, 117,
10449-10459.
(1) Daly, J. W. The Alkaloids; 1998; pp 141-169.
(2) Daly, J. W. Chemical Ecology: The Chemistry of Biotic Interaction; 1995;
pp 17-28.
(11) Aube´, J.; Rafferty, P. S.; Milligan, G. L. Heterocycles 1993, 35, 1141-
1147.
(3) Daly, J. W.; Garraffo, H. M.; Spande, T. F. The Alkaloids; Academic
Press: New York, 1993; pp 185-288.
(12) Iyengar, R.; Schildknegt, K.; Aube´, J. Org. Lett. 2000, 2, 1625-1627.
(13) Smith, B. T.; Wendt, J. A.; Aube´, J. Org. Lett. 2002, 4, 2577-2580.
(14) Golden, J. E.; Aube´, J. Angew. Chem., Int. Ed. 2002, 41, 4316-4318.
(15) Wrobleski, A.; Sahasrabudhe, K.; Aube´, J. J. Am. Chem. Soc. 2002, 124,
9974-9975.
(4) Daly, J. W. J. Med. Chem. 2003, 46, 445.
(5) Myers, C. W.; Daly, J. W. Am. Mus. NoVit. 1980, 1-23.
(6) Myers, C. W. Pape´is AVulsos Zool. San Paolo 1987, 36, 301-306.
9
10.1021/ja0320018 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 5475-5481
5475

A R T I C L E S
Wrobleski et al.
Scheme 1
Scheme 2
(Scheme 1). However, the complexity of ketones 3 would add
another challenging element to the proposed synthesis. This
paper fully discloses a set of studies of the problem of 251F
synthesis that not only provided advanced examples of the
intramolecular Schmidt reaction but also established a new route
to [3.3.0] bicyclic ketones related to 3. In this work, two distinct
total syntheses were carried out, one leading to (()-desmethyl
251F 2 and the other affording 1 in enantiomerically enriched
form.
Retrosynthetic Analysis
Scheme 3
As noted above, the targeted intramolecular Schmidt reaction
required the intermediacy of keto azides 3a and 3b, respectively.
These intermediates were proposed to result from the appropriate
derivatization of a bicyclic enone 4a or 4b (X represents a
hydroxymethylene equivalent). This sequence presents an
important stereochemical issue in that the two groups being
introduced onto the enone (an exo â methyl group and an endo
R azido side chain) must occupy the indicated trans relationship
relative to one another to complete the stereocontrolled instal-
lation of the six consecutive stereocenters. Plenty of precedent
suggested that conjugate addition would result in exo methy-
lation of the diquinane; however, the stereochemical outcome
of installing the adjacent azidobutyl side chain was less secure.
Although the desired product would result from placing this
group trans to the adjacent methyl group, doing so would force
it onto the endo face of the [3.3.0] bicyclooctane system (which
is less favored relative to the exo orientation due to the cuplike
ring system). To address this, MM2 calculations were performed
on model substrates bearing substituents at the R and â centers
with both cis and trans configurations (Scheme 2). Calculations
indicated that the energy of isomer B with both groups exo and
cis to one another was 2.4 kcal/mol greater than that of the
desired epimer A. Minimally, this suggested that thermodynam-
ics might work in our favor even if a kinetic addition of the
desired group failed to afford the desired relative stereochem-
istry.
sion of the acyloxazolidinone portion of 8 to the corresponding
methyl ketone followed by oxidative cleavage (ozonolysis, for
example) of the endocyclic olefin would generate dialdehyde
7. An intramolecular aldol reaction of 7 could in principle
provide bicyclic enone 6. A variant of this ozonolysis/aldol
strategy was previously explored by Sternbach et al. in an
approach to hirsutene16 and the total synthesis of silphinene17
(Scheme 4).
In other work, Grubbs published a route for converting [2.2.1]-
bicycloheptene ring systems to the corresponding [3.2.0]heptane
enol ethers using titanocene alkylidene complexes (Scheme
4).18,19 This served as a key reaction in a synthesis toward
capnellene. These titanium-mediated transformations are clearly
related to the now-common ruthenium alkylidene-catalyzed
metathesis reactions. Several groups have utilized ring-opening/
ring-closing versions of this latter transformation to carry out
skeletal rearrangements of the type needed here.20-27 We
(16) Sternbach, D. D.; Ensinger, C. L. J. Org. Chem. 1990, 55, 2725-2736.
(17) Sternbach, D. D.; Hughes, J. W.; Burdi, D. F.; Banks, B. A. J. Am. Chem.
Soc. 1985, 107, 2149-2153.
Stereoselective syntheses of diquinanes 4a,b were required
to realize the above strategy. An attractive solution toward this
problem was suggested by the recognition that the four
stereogenic centers in compounds 4 could be directly mapped
onto 5 as shown (Scheme 3), coupled with the fact the latter
should be readily generated from a garden variety Diels-Alder
reaction of cyclopentadiene and a crotonic acid equivalent.
Key precedent for our approach to enone 6 was found in the
literature of triquinane natural product synthesis. Thus, conver-
(18) Stille, J. R.; Santarsiero, B. D.; Grubbs, R. H. J. Org. Chem. 1990, 55,
843-862.
(19) Stille, J. R.; Grubbs, R. H. J. Am. Chem. Soc. 1986, 108, 855-856.
(20) Zuercher, W. J.; Hashimoto, M.; Grubbs, R. H. J. Am. Chem. Soc. 1996,
118, 6634-6640.
(21) Minger, T. L.; Phillips, A. J. Tetrahedron Lett. 2002, 43, 5357-5359.
(22) Stragies, R.; Blechert, S. Synlett 1998, 169-170.
(23) Arjona, O.; Csaky, A. G.; Medel, R.; Plumet, J. J. Org. Chem. 2002, 67,
1380-1383.
(24) Hagiwara, H.; Katsumi, T.; Endou, S.; Hoshi, T.; Suzuki, T. Tetrahedron
2002, 58, 6651-6654.
9
5476 J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004

Total Synthesis of Dendrobatid Alkaloids
A R T I C L E S
Scheme 4
Scheme 6
Scheme 5
acidification spontaneously lactonized to alcohol 13. This
allowed for ready differentiation of the two alcohols. Subsequent
protection yielded lactone 14 in 70% overall yield for the three-
step sequence from 5. Addition of varying methyl nucleophiles
(MeLi, MeMgBr, etc.) gave complex mixtures of products, so
an olefination of lactone 14 was planned to effect the homolo-
gation to the methyl ketone. Initial experiments focused on using
the Tebbe reagent29 to convert 14 to the analogous enol ether;
however, no positive results were obtained via this avenue.
Fortunately, Petasis’ reagent30 (Cp2TiCl2) in place of Tebbe’s
provided the desired enol ether 15. After some experimentation,
it was discovered that carrying out the olefination according to
the procedure developed by Howell31 proved most effective.
The resulting enol ether was subsequently hydrolyzed to
hemiacetal 16 and immediately oxidized with PCC to generate
the aldol precursor 17. A number of conditions were examined
to carry out the intramolecular aldol conversion of 17 to 4b but
failed to produce the aldol adduct in acceptable yields. Eventu-
ally, reasonably efficient conditions were uncovered. Thus, KOH
and Bu4NOH in refluxing THF/water (THF, tetrahydrofuran)
provided aldol adduct 4b in 52% yield.
investigated, and here compare, both aldol and metathesis
approaches to the synthesis of the desired enones.
Synthesis of (()-3-Desmethyl 251F (2)
To permit us to concentrate on the coupled problems of [3.3.0]
bicyclooctane synthesis and relative stereocontrol about the ring
system, we chose to first prepare (()-3-desmethyl 251F lacking
the isolated methyl group. Since all the remaining stereocenters
were destined to arrive from relative stereocontrol techniques,
this allowed us to work in the racemic series at the outset.
Starting from a racemic derivative of 5, an extremely direct
intramolecular aldol route to 4b was briefly investigated.
Accordingly, olefin 928 (Scheme 5) was ozonized to yield the
corresponding dialdehyde 10; however, under no conditions
examined could an intramolecular aldol reaction be induced to
afford 11. It is likely that the abundance of acidic protons and
subsequent retroaldol opportunities available in 10 rendered the
desired aldol process ineffective. A variety of epimerization
reactions may have also competed with the desired cyclization
reaction.
Though the route presented in Scheme 6 afforded quantities
of 4b in excess of 2 g, numerous drawbacks using this strategy
soon became apparent. The Petasis olefination product 15 was
obtained cleanly and could be used in subsequent transforma-
tions without purification (though some titanocene byproduct
could be seen in the NMR spectrum of the crude product).
However, only after the aldol step could product be purified
chromatographically. The four-step sequence from 14 to 4b was
routinely carried out with crude reaction mixtures, analyzing
Guided by the hypothesis that reduction of the extraneous
aldehyde in 10 would eliminate some of the untoward pathways,
a second attempt involved pursuing an aldol precursor via a
lactone derivative of acid 5 (Scheme 6). Ozonolysis of olefin 5
followed by borohydride reduction gave diol 12, which upon
(25) Weatherhead, G. S.; Ford, J. G.; Alexanian, E. J.; Schrock, R. R.; Hoveyda,
A. H. J. Am. Chem. Soc. 2000, 122, 1828-1829.
(29) Tebbe, F. N.; Parshall, G. W.; Reddy, G. S. J. Am. Chem. Soc. 1978, 100,
3611-3613.
(30) Petasis, N. A.; Bzowej, E. I. J. Am. Chem. Soc. 1990, 112, 6392-6394.
(31) Dollinger, L. M.; Ndakala, A. J.; Hashemzadeh, M.; Wang, G.; Wang, Y.;
Martinez, I.; Arcari, J. T.; Galluzzo, D. J.; Howell, A. R. J. Org. Chem.
1999, 64, 7074-7080.
(26) Zuercher, W. J.; Scholl, M.; Grubbs, R. H. J. Org. Chem. 1998, 63, 4291-
4298.
(27) Lee, D.; Sello, J. K.; Schreiber, S. L. Org. Lett. 2000, 2, 709-712.
(28) Available in two steps (Weinreb amide formation then methyl Grignard
addition) from acid 5.
9
J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004 5477

A R T I C L E S
Wrobleski et al.
Scheme 7
Scheme 8
the intermediate products only with 1H NMR to ensure complete
reactant consumption. Overall, the seven-step sequence from 5
could be carried out in 27% overall yield. At seven steps, this
sequence was sufficient for carrying out preliminary experi-
ments; however, more efficient means would ultimately be
sought to streamline the synthesis. The ozonolysis/olefination/
aldol strategy was certainly lengthier than the initially proposed
route, and we felt that a more rapid and efficient enone synthesis
would help ensure a more attractive synthesis of 251F in the
end. Despite this, we decided to use enone 4b obtained via this
route to investigate the addition of the two remaining side chains
and the following Schmidt reaction.
Considering the cup-shaped geometry of a cis-fused 5,5-
bicyclic system, it was expected that methyl cuprate addition
would occur from the desired exo position and that quenching
the resultant enolate with an appropriate electrophile would
stereoselectively install a precursor to the endo azido side chain
(trans to the methyl group; see Scheme 2). However, conjugate
addition of Me2CuLi followed by quenching with 1-chloro-4-
iodobutane resulted only in installation of a methyl group at
the â position of the enone. Under a variety of conditions
examined using even highly reactive alkyl halides such as
iodomethane and allyl bromide as electrophiles, no successful
instances of incorporation of R substituents were observed.
Attempts at two-pot sequences in which the â-substituted ketone
was treated with a variety of bases and subsequently treated
with electrophiles resulted only in starting material.
Given these problems, we proposed that a one-pot conjugate
addition/aldol strategy could serve as an effective means for
introducing both substituents. Thus, to our delight, aldehydes
proved to be efficient coupling partners (Scheme 7). Accord-
ingly, conjugate addition of Me2CuLi from the most exposed
face of 4b resulted in enolate 18, which, when quenched with
4-benzyloxybutanal, led to a single olefinic isomer of 19.
Scheme 7 indicates the E stereochemistry for the aldol dehydra-
tion, but it was not until the asymmetric synthesis of 251F was
completed that the olefin geometry was rigorously proved (see
below).
stereocenter. However, this point was muted as it was recognized
that a dissolving metal reduction could serve two functions: (1)
cleavage of the benzyl ether of 19 to the corresponding alcohol
and (2) reduction of the enone moiety to the saturated analogue
placing the azido side chain precursor in the desired endo
position (Scheme 7). The observed side chain stereochemistry
can be rationalized in two ways. On one hand, the calculations
described above suggested that thermodynamic protonation of
the intermediate enolate (20) would favor placing the side chain
trans to the existing adjacent methyl group. Alternatively, kinetic
protonation might favor quenching the enolate from the less
hindered exo face. In practice, the dissolving metal reduction
of 19 proceeded to yield a single diastereomer of 21. However,
the stereochemistry of this newly formed center was not verified
until a later point in the synthesis due to the lack of diagnostic
NMR evidence from 21 itself.
Progress toward the azido ketone Schmidt precursor was made
via azidation of primary alcohol 21 under Mitsunobu condi-
tions32 to yield azide 22 in ca. 50% overall yield from 19 (2
steps, Scheme 8). With 22 in hand, the stage was set for carrying
out the critical intramolecular Schmidt reaction. Thus, treatment
of 22 with excess triflic acid (TfOH) smoothly converted the
azido ketone to the ring-expanded lactam 23 while simulta-
neously removing the TIPS group to reveal the primary alcohol
in 52% yield. At this time, extensive 2D NMR analysis was
used to corroborate the stereochemistry of all six stereogenic
centers. The most revealing piece of evidence came from the
observation of an NOE between the C10 methine proton and
the adjacent methyl group, indicating a cis orientation between
the two or, alternatively, a trans relationship between the methyl
group and side chain. Four other NOEs were observed and
further corroborated that the Schmidt product had stereochem-
istry corresponding to that of the natural product (Scheme 8).
Completion of the 3-desmethyl 251F synthesis required only
the reduction of the lactam to the corresponding tertiary amine.
The in situ dehydration of the aldol adduct to exocyclic enone
19 precluded the direct formation of the sixth consecutive
(32) Viaud, M. C.; Rollin, P. Synthesis 1990, 130-132.
9
5478 J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004

Total Synthesis of Dendrobatid Alkaloids
A R T I C L E S
Scheme 9
Thus, treatment of 23 with LAH in ether afforded the desired
amine 2 (confirmed from IR and MS data only due to isolation
difficulties; Scheme 8).
Asymmetric Total Synthesis of 251F (1)
Although the model synthesis of 251F helped develop a great
deal of the chemistry needed to formulate a route toward the
natural product, a number of issues remained unresolved. First,
the issue of carrying out an asymmetric synthesis had to be
considered. This issue was readily addressed as numerous
efficient routes toward both enantiomers of acid 5 have been
developed by other workers.33-39 A more pressing task,
however, was the desire to produce enone 4b, or its equivalent,
in a more direct route. This goal ultimately found its realization
with the utilization of metathesis chemistry. In addition, the
3-methyl group had to be incorporated with the proper stereo-
chemistry. Finally, a major goal of this synthesis was to produce
meaningful (decigram) quantities of the dendrobatid natural
product.
Two of the above issues were resolved via the dual utilization
of a common chiral auxiliary to control the formation of five
of the seven stereogenic centers of alkaloid 251F. Thus, an
asymmetric Diels-Alder reaction using 25 set the four consecu-
tive stereocenters of acid 5 in both a relative and absolute sense
(Scheme 9). In addition, the same auxiliary, (4R,5S)-(+)-4-
methyl-5-phenyl-2-oxazolidinone, can be used to control the
outcome of an asymmetric alkylation reaction to set the C3
stereocenter (Scheme 9).
The synthesis of enantiopure aldehyde 30 commenced with
conversion of auxiliary 24 to 27 via deprotonation (n-butyl-
lithium, THF) and quenching with propionyl chloride. The
enolate of 27 was generated with NaHMDS (sodium bis(tri-
methylsilyl)amide) and subsequently reacted with allyl iodide
to provide allylated 28 as a white solid that could be recrystal-
lized to g95% diastereomeric purity. Cleavage of the chiral
auxiliary (LAH) afforded an enantiopure alcohol40 which was
immediately benzylated to ether 29. Finally, oxidative cleavage
of the olefin afforded the desired aldehyde 30 in 36% yield
over the five-step sequence. Conversion of 25 to 26 was effected
by treating a mixture of 25 and cyclopentadiene in methylene
chloride at -100 °C with diethyl aluminum chloride. Diels-
Alder adduct 26 was isolated in ca. 93-95% diastereomeric
purity on a 2 g scale.37
With protocols in hand for the asymmetric synthesis of acid
5 and provisions met for the stereocontrolled installation of the
3-methyl group, attention was then directed at retooling the route
toward the key bicyclic enone intermediate 4b, previously
prepared via an ozonolysis/aldol method. As noted earlier, it
was felt that the use of metal alkylidene complexes (i.e.,
ruthenium-catalyzed ring-opening/ring-closing metathesis) could
be utilized to catalyze the desired [2.2.1] f [3.3.0] skeletal
rearrangement. To examine this strategy it was recognized that
the acid portion of 5 would need to be converted to an olefinic
derivative.
In building this substrate, Diels-Alder adduct 26 was cleaved
to enantiomerically enriched acid 5 with LiOH/H2O2.41 Thus,
conversion of 5 to the Weinreb amide42 followed by addition
of vinyl Grignard afforded 31 in 85% yield for the 2-step
sequence (Scheme 10). At this stage it became possible to
explore the use of Grubbs’s ruthenium catalysts to effect the
desired ring-opening/ring-closing metathesis reactions. Early
attempts using standard conditions20 afforded the desired enone
4a, albeit in poor yield. Examination of these results suggested
that a good deal of the reaction products were oligomerized
materials. Though only a 30% yield of metathesis product was
isolated, it was felt that this yield could be increased with the
identification of more fitting conditions.
(33) Vandewalle, M.; Van der Eycken, J.; Oppolzer, W.; Vullioud, C.
Tetrahedron 1986, 42, 4035-4043.
Remarkable increases in reaction efficiencies were realized,
however, when the tandem ring-opening/ring-closing metathesis
was carried out in ethylene-saturated methylene chloride kept
under an atmosphere of ethylene.43,44 With a minimal amount
(34) Fukuzawa, S.; Matsuzawa, H.; Metoki, K. Synlett 2001, 709-711.
(35) Krotz, A.; Helmchen, G. Tetrahedron: Asymmetry 1990, 1, 537-540.
(36) Evans, D. A.; Barnes, D. M.; Johnson, J. S.; Lectka, T.; von Matt, P.; Miller,
S. J.; Murry, J. A.; Norcross, R. D.; Shaughnessy, E. A.; Campos, K. R. J.
Am. Chem. Soc. 1999, 121, 7582-7594.
(37) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc. 1988, 110,
1238-1256.
(38) Evans, D. A.; Miller, S. J.; Lectka, T. J. Am. Chem. Soc. 1993, 115, 6460-
(41) Evans, D. A.; Britton, T. C.; Ellman, J. A. Tetrahedron Lett. 1987, 28,
6141-6144.
6461.
(39) Evans, D. A.; Miller, S. J.; Lectka, T.; von Matt, P. J. Am. Chem. Soc.
1999, 121, 7559-7573.
(42) Nahm, S.; Weinreb, S. M. Tetrahedron Lett. 1981, 22, 3815-3818.
(43) Mori, M.; Sakakibara, N.; Kinoshita, A. J. Org. Chem. 1998, 63, 6082-
6083.
(40) Evans, D. A.; Bender, S. L.; Morris, J. J. Am. Chem. Soc. 1988, 110, 2506-
2526.
(44) Giessert, A. J.; Brazis, N. J.; Diver, S. T. Org. Lett. 2003, 5, 3819-3822.
9
J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004 5479

A R T I C L E S
Wrobleski et al.
Scheme 10
Scheme 11
using the ozonolysis/aldol strategy). Multigram quantities of 4a
were readily produced with relative ease.
of optimization, enone 4a could be isolated in 93% yield using
5 mol % ruthenium catalyst 32 on multigram scale reactions.
Scheme 10 (path a) presents a possible blueprint for this
reaction. Metallocycloaddition of 32 onto the endocyclic olefin
of 31 generates metallocyclobutane 33, which upon retro-
cycloaddition provides alkylidene 34. A second metallocycload-
dition onto the enone olefin provides metallocyclobutane 35,
which again undergoes retro-cycloaddition to provide the
targeted enone 4a. The yield of 4a is better than tripled when
changing the saturation of the solvent and reaction atmosphere
from argon to ethylene. Apparently having an excess of ethylene
present in the reaction allows for the recycling of unproductive
intermediates (i.e., those leading to polymers or other byprod-
ucts). Scheme 10 (path b) provides an example of one untoward
possibility. In addition, dimeric or oligomeric materials can be
reverted back to monomeric alkylidene intermediates capable
of entering the pathway depicted in Scheme 10 en route to
desired 4a. This complex equilibrium of intermediates ultimately
proceeds toward what is likely the most thermodynamically
stable component, 4a.
Completion of Indolizidine 251F Synthesis
As developed in the model study, it was envisioned that a
one-pot conjugate addition/aldol reaction would install the exo
methyl group and precursor to the azido side chain. However,
the aldehyde component of the aldol reaction now had to
incorporate the 3-methyl substituent with the proper stereo-
chemistry. Thus, enone 4a was reacted with Me2CuLi to afford
enolate 40, which was quenched with chiral aldehyde 30 to
provide dehydrated aldol adduct 41 (Scheme 11). The aldol
adduct was isolated as a single olefinic isomer in 65% yield.
The geometry of the newly formed olefin was investigated using
2D NMR techniques. With a NOESY experiment, it was found
that an NOE existed between the newly installed exo methyl
group and the allylic enone protons (Scheme 11). For these
groups of protons to be within NOE proximity requires that
the exocyclic enone olefin have the E configuration shown.
Following the protocol developed above, 41 was converted
to the azido ketone Schmidt precursor 42 (Scheme 11). The
multipurpose Na/NH3 reduction converted 41 to the correspond-
ing reduced alcohol (not shown) as a ca. 4:1 mixture of
inseparable diastereomers epimeric at the R position. The major
isomer was presumed to be that derived from exo protonation,
placing the side chain in an endo orientation. A subsequent
The successful incorporation of a tandem ring-opening/ring-
closing metathesis in place of an ozonolysis/aldol route proved
to increase the efficiency and ease of bicyclic enone synthesis
tremendously. In only three steps, 5 was converted to 4a in ca.
80% overall yield (compared to seven steps, 27% overall yield
9
5480 J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004

Total Synthesis of Dendrobatid Alkaloids
A R T I C L E S
Scheme 12
expanded Schmidt product 45 in 79% yield (Scheme 12).
Lactam 45 was crystallized, and X-ray analysis confirmed that
the tricyclic structure and relative stereochemistry of this
compound reflected those of 251F.
Reduction of lactam 45 to the corresponding amine repre-
sented the final step of the synthetic sequence, and as expected,
LAH reduction generated alkaloid 1 in 86-100% yield (Scheme
12). Synthetic 251F was identical in all respects to published
spectra.7,8
Summary
In conclusion, a study ending in a racemic synthesis of
3-desmethyl 251F allowed for the development of chemistry
that ultimately culminated in an asymmetric total synthesis of
alkaloid 251F. Each synthesis was initiated with a stereoselective
Diels-Alder reaction that established four consecutive stereo-
genic centers. Integral to both syntheses was the efficient
construction of a key bicyclic enone intermediate. Preliminary
studies focused on an ozonolysis/olefination/aldol method, and
a more streamlined second-generation route was developed
utilizing a tandem metathesis strategy. At seven steps and 27%
overall yield, the ozonolysis/aldol strategy toward enone 4b
proved satisfactory. However, replacing this route with a domino
metathesis strategy better than halved the number of steps (three
total) and tripled the yield (80%) in route to the analogous enone
4a. In general, this skeletal rearrangement reaction could prove
to be a highly useful tool in the synthesis of substituted
diquinane substrates from readily available [2.2.1] bicyclic
systems. Key to both syntheses was the use of an intramolecular
Schmidt reaction, which delivered the core tricyclic system, on
advanced intermediates. Overall, the asymmetric total synthesis
of 251F proceeded in 13 steps with an overall yield of ca.
5-8%. Approximately 100 mg of the natural product was
produced in this waysthis amount is roughly equivalent to the
amount that would have been isolated using current techniques
from 30 000 frogs! Given this healthy supply of 251F, the
unknown biological properties of the alkaloid are currently under
delineation.
Mitsunobu reaction32 transformed the primary alcohol to the
corresponding azide 42 (50% yield, two steps), here obtained
as an inseparable 4:1 mixture of diastereomers. With the
necessary functional groups in place, the key intramolecular
Schmidt reaction was then examined. Treating azido ketone 42
with either TfOH or TiCl4 resulted only in apparent decomposi-
tion of the olefin functionality. None of the desired cyclized
lactam product was detected.
Because the olefin was serving as a masked hydroxymeth-
ylene equivalent and given the earlier success in transforming
an -OTIPS protected analogue into the lactam (Scheme 8), it
was proposed that conversion to its oxidized analogue might
be more consistent with a successful azido-Schmidt reaction.
Toward this end, olefin 42 was reacted with ozone followed by
dimethyl sulfide (DMS) to afford aldehyde 43 (ca. 70% yield).
At this juncture, a few experiments were conducted on 43 to
determine whether this substrate was a viable Schmidt substrate.
Treatment of 43 with TfOH resulted in the detection of a lactam
product (1H and 13C NMR); however, the reaction resulted in a
variety of unidentified impurities, and attention was then directed
at carrying out the Schmidt reaction on alcohol 44. Caution was
exercised with this reduction, though, as once 43 was produced,
three functional groups capable of undergoing reduction were
present. Reduction of either the azide or ketone eliminates one
of the functional groups necessary for the Schmidt reaction.
Fortunately, addition of 1 equiv of NaBH4 chemoselectively
reduced the aldehyde in the presence of both ketone and azide
to afford alcohol 44 (50-55% yield for the two-step ozonolysis/
reduction sequence). In addition, conversion of the 4:1 diaster-
eomeric mixture of olefin 42 to alcohol 44 allowed for facile
chromatographic separation of the diastereomers.
Acknowledgment. We thank the National Institutes of Health
(Grant GM-49093) for support of this research. A.W. addition-
ally thanks the American Chemical Society Division of Organic
Chemistry and Abbott Laboratories for support through a
graduate fellowship. We are grateful to the following scientists
for their constructive comments and suggestions during the
course of this project: John Daly, Carol Ensinger, Robert
Grubbs, and Amy Howell.
Supporting Information Available: Experimental procedures
and compound characterizations (PDF and CIF). This material
is available free of charge via the Internet at http://pubs.acs.org.
Once again, examination of the key intramolecular Schmidt
reaction, this time with alcohol 44, commenced. Treatment of
44 with excess TfOH proceeded smoothly and yielded the ring-
JA0320018
9
J. AM. CHEM. SOC. VOL. 126, NO. 17, 2004 5481