Stereoselective total syntheses and reassignment of stereochemistry of the freshwater cyanobacterial hepatotoxins cylindrospermopsin and 7-epicylindrospermopsin

Article abstract of DOI:10.1021/ja020032h

A stereoselective total synthesis of the structure 1 proposed for the freshwater cyanobacterial heptatotoxin cylindrospermopsin has been accomplished in approximately 30 operations starting from commercially available 4-methoxypyridine. Utilizing methodol

Full text of DOI:10.1021/ja020032h

Published on Web 03/22/2002
Stereoselective Total Syntheses and Reassignment of
Stereochemistry of the Freshwater Cyanobacterial
Hepatotoxins Cylindrospermopsin and
7-Epicylindrospermopsin
Geoffrey R. Heintzelman, Wen-Kui Fang, Stephen P. Keen, Grier A. Wallace, and
Steven M. Weinreb*
Contribution from the Department of Chemistry, The PennsylVania State UniVersity,
UniVersity Park, PennsylVania 16802
Received January 7, 2002
Abstract: A stereoselective total synthesis of the structure 1 proposed for the freshwater cyanobacterial
heptatotoxin cylindrospermopsin has been accomplished in approximately 30 operations starting from
commercially available 4-methoxypyridine. Utilizing methodology developed by Comins, the tetrasubstituted
piperidine A-ring unit of the hepatotoxin was efficiently constructed. The two remaining stereocenters in
the natural product were then set by a stereospecific intramolecular N-sulfinylurea Diels-Alder cyclization/
Grignard ring opening/allylic sulfoxide [2,3]-sigmatropic rearrangement sequence previously developed in
these laboratories, leading to key intermediate 29. The stereochemical assignment of alcohol 29, which
contains all six of the stereogenic centers of the natural product, was confirmed by an X-ray crystal structure
determination of a derivative. Installation of the D-ring uracil moiety was effected by using our new
methodology developed for this purpose, and construction of the C-ring guanidine completed the total
synthesis of racemic structure 1. However, the 1H NMR data for this compound do not match that of
cylindrospermopsin, but instead agree with the data reported for 7-epicylindrospermopsin, a minor toxic
metabolite that co-occurs with cylindrospermopsin. Therefore, we propose a revision of the stereochemical
assignments of these natural products such that cylindrospermopsin is now represented as structure 2
and 7-epicylindrospermopsin is 1. This reassignment was further confirmed by Mitsunobu inversion of the
C-7 alcohol 51 to epimer 52, and conversion of this compound to tetracyclic diol 57, which has previously
been transformed to cylindrospermopsin (2).
Introduction
after treatment of the reservoir with copper sulfate to control a
severe algal bloom, thereby causing lysis of the cyanobacterial
Freshwater blue-green algae (cyanobacteria) have long been
known to produce a diverse group of secondary metabolites
which have biological activity as hepatotoxins, cytotoxins, and
cells and presumably the release of a toxin into the water.
Examination of the aqueous extract from laboratory cultured
C. raciborskii revealed the presence of a compound that causes
severe hepatotoxicity in mice and which is strongly implicated
as the causitive agent of the Palm Island outbreak of hepatoen-
teritis in humans.
1
neurotoxins. A number of reliable reports of livestock mortality
due to ingestion of cyanobacterial-contaminated water have
appeared over the years but fewer cases of human poisoning
have been firmly documented. One such instance occurred in
Studies by Moore and co-workers led to the isolation of the
toxin that was named cylindrospermopsin, and was assigned
the novel tetracyclic structure and relative stereochemistry
shown in 1 primarily based upon a series of sophisticated NMR
experiments. It should be noted that an important but rather
speculative premise in the assignment of the stereochemistry
at C-7 originally proposed for cylindrospermopsin was that the
molecule exists in the conformation indicated in 1a, enforced
by a hydrogen bond between an unusual enolic uracil D-ring
tautomer and the guanidine C-ring, which was used to rationalize
1
979 on Palm Island, in northeast tropical Australia, which led
to the hospitalization of about 150 people, the large majority
children, with hepatitus-like symptoms. The cause of the illness
was eventually traced to the cyanobacterium Cylindrospermopsis
raciborskii, which had contaminated a reservoir (Solomon Dam)
3
,4
2
supplying drinking water. This poisoning event occurred shortly
*
Address correspondence to this author. E-mail: smw@chem.psu.edu.
1) For good overviews see: (a) Falconer, I. R. EnViron. Toxicol. 1999, 14, 5.
b) Moore, R. E.; Ohtani, I.; Moore, B. S.; DeKoning, C. B.; Yoshida, W.
Y.; Runnegar, M. T. C.; Carmichael, W. W. Gazz. Chim. Ital. 1993, 123,
29. (c) Burja, A. M.; Banaigs, B.; Abou-Mansour, E.; Burgess, J. G.;
(
(
3
Wright, P. C. Tetrahedron 2001, 57, 9347. (d) Gerwick, W. H.; Tan, L.
T.; Sitachitta, N. Nitrogen-Containing Metabolites from Marine Cyano-
bacteria. In The Alkaloids; Cordell, G. A., Ed.; Academic Press: San Diego,
(3) Ohtani, I.; Moore, R. E.; Runnegar, M. T. C. J. Am. Chem. Soc. 1992,
114, 7941.
(4) For studies on the biosynthesis of cylindrospermopsin see: Burgoyne, D.
L.; Hemscheidt, T. K.; Moore, R. E.; Runnegar, M. T. C. J. Org. Chem.
2000, 65, 152.
2
001; Vol. 57, p 75.
(
2) Hawkins, P. R.; Runnegar, M. T. C.; Jackson, A. R. B.; Falconer, I. R.
Appl. EnViron. Microbiol. 1985, 50, 1292.
10.1021/ja020032h CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 3939-3945
9
3939

A R T I C L E S
Heintzelman et al.
the 4.0 Hz C-7,8 proton-proton coupling constant. The absolute
configuration for this metabolite was not determined at the time
on the natural material but was recently established to be as
shown via enantioselective total synthesis by White and
Hansen.5
Production of these toxins by cyanobacteria in drinking water
is clearly a serious public health problem, particularly in tropical
areas, and has recently been traced to the deaths of livestock in
Australia. Thus, considerable work continues to appear on
development of analytical methods for measuring cylindrosper-
1
4
1
3,15
It has become clear that the occurrence of cylindrospermop-
sin-producing cyanobacteria is widespread in temperate as well
as tropical and subtropical areas. For example, in 1994, Harada
et al. isolated cylindrospermopsin from the alga Umezakia
mopsin levels in water
destroying the compounds in situ.
Due to the fascinating and unusual structures of 1 and 2, as
as well as methodology for potentially
1
6
1
7
6,18,19
well as their significance in public health, we and others
6
natans collected in Lake Mikata (Fukui, Japan). More recently,
have undertaken studies on the synthesis of these hepatotoxins.
Snider has recently reported a total synthesis of racemic
this same toxin was isolated from another cyanobacterium,
Aphanizomenon oValisporum (Forti), found in Lake Kinneret
1
8c
cylindrospermopsin. Although the Snider group’s synthetic
material in fact corresponded to natural cylindrospermopsin, it
was not possible to conclusively establish the C-7 stereochem-
istry of any of their intermediates, and therefore the work did
not independently confirm the original assignment for this
center. In 2001 we described in preliminary form a total syn-
thesis that completely controls all six stereogenic centers of
the putative cylindrospermopsin structure 1, and which proves
conclusively that the stereochemical assignments at C-7 in fact
have been reversed in cylindrospermopsin and the 7-epi
(
Sea of Galilee), a major source of fresh drinking water in
7
Israel. A. oValisporum strains which produce cylindrospermop-
sin have also been identified in Australia. In addition, C.
raciborskii has been found in Europe, Brazil, and the United
8
States. Further investigation of the Israel cyanobacterium led
to the isolation of a second, minor metabolite with toxicity
9
similar to that of cylindrospermopsin. This compound was
assigned the 7-epicylindrospermopsin structure 2, once again
mainly based upon NMR evidence. The C-7,8 coupling constant
of 6.8 Hz for this compound was rationalized by a hydrogen-
bonded conformation 2a, as was proposed for cylindrosper-
mopsin. Finally, a third metabolite, 7-deoxycylindrospermopsin
1
7a
compound. In this paper we now provide the full details of
this work.
Synthetic Plan
(
3), has been isolated from C. raciborskii and interestingly was
10,11
found to be nontoxic to mice.
Several years ago we devised a strategy for synthesis of
cylindrospermopsin based upon a novel variation of N-sulfinyl
dienophile Diels-Alder methodology that we had previously
2
0
developed. We recognized that the C-7,8 relationship in
structure 1 is that of a syn-vicinal amino alcohol derivative of
the type we could access from a stereospecific N-sulfinyl Diels-
Alder reaction of an (E,E)-diene to form a dihydrothiazine oxide,
followed by a stereospecific Grignard ring opening/[2,3]-
(
10) Norris, R. L.; Eaglesham, G. K.; Pierens, G.; Shaw, G. R.; Smith, M. J.;
Chiswell, R. K.; Seawright, A. A.; Moore, M. R. EnViron. Toxicol. 1999,
14, 163.
(
11) There have been some limited structure-activity relationship studies in
this area: (a) Banker, R.; Carmeli, S.; Werman, M.; Telsch, B.; Porat, R.;
Sukenik, A. J. Toxicol. EnViron. Health 2001, 62, 281. (b) Runnegar, M.
T.; Xie, C.; Snider, B. B.; Wallace, G. A.; Weinreb, S. M.; Kuhlenkamp,
J. Toxicol. Sci. 2002, in press.
(
12) Runnegar, M. T.; Kong, S.-M.; Zhong, Y.-Z.; Ge, J.-L.; Lu, S. C. Biochem.
Biophys. Res. Commun. 1994, 201, 235. Runnegar, M. T.; Kong, S.-M.;
Zhong, Y.-Z.; Lu, S. C. Biochem. Pharmacol. 1995, 49, 219.
(
(
(
13) Froscio, S. M.; Humpage, A. R.; Burcham, P. C.; Falconer, I. R. EnViron.
Toxicol. 2001, 16, 408 and references cited therein.
14) Saker, M. L.; Thomas, A. D.; Norton, J. H. EnViron. Toxicol. 1999, 14,
179.
15) See for example: Eaglesham, G. K.; Norris, R. L. G.; Shaw, G. R.; Smith,
M. J.; Chiswell, R. K.; Davis, B. C.; Neville, G. R.; Seawright, A. A.;
Moore, M. R. EnViron. Toxicol. 1999, 14, 151. Norris, R. L. G.; Eaglesham,
G. K.; Shaw, G. R.; Senogles, P.; Chiswell, R. K.; Smith, M. J.; Davis, B.
C.; Seawright, A. A.; Moore, M. R. EnViron. Toxicol. 2001, 16, 391.
16) Chiswell, R. K.; Shaw, G. R.; Eaglesham, G.; Smith, M. J.; Norris, R. L.;
Seawright, A. A.; Moore, M. R. EnViron. Toxicol. 1999, 14, 155. Senogles,
P.; Shaw, G.; Smith, M.; Norris, R.; Chiswell, R.; Mueller, J.; Sadler, R.;
Eaglesham, G. Toxicon 2000, 38, 1203.
(
(
17) For a preliminary account of portions of this work see: (a) Heintzelman,
G. R.; Fang, W.-K.; Keen, S. P.; Wallace, G. A.; Weinreb, S. M. J. Am.
Chem. Soc. 2001, 123, 8851. See also: (b) Heintzelman, G. R.; Parvez,
M.; Weinreb, S. M. Synlett 1993, 551. (c) Heintzelman, G. R.; Weinreb,
S. M.; Parvez, M. J. Org. Chem. 1996, 61, 4594. (d) Keen, S. P.; Weinreb,
S. M. Tetrahedron Lett. 2000, 41, 4307.
It has been postulated that cylindrospermopsin probably exerts
its toxic effects by inhibiting biosynthesis of cell-reduced
glutathione11b,12
and also by inhibition of protein synthesis.13
(
5) White, J. D.; Hansen, J. D. Submitted for publication.
(18) (a) Snider, B. B.; Harvey, T. C. Tetrahedron Lett. 1995, 36, 4587. (b) Snider,
B. B.; Xie, C. Tetrahedron Lett. 1998, 39, 7021. (c) Xie, C.; Runnegar, M.
T. C.; Snider, B. B. J. Am. Chem. Soc. 2000, 122, 5017. (d) McAlpine, I.
J.; Armstrong, R. W. Tetrahedron Lett. 2000, 41, 1849. (e) White, J. D.;
Hansen, J. D. Abstracts of Papers; 219th National Meeting of the American
Chemical Society, San Francisco, CA; American Chemical SOciety:
Washington, DC, 2000; ORGN 812. (f) Djung, J. F.; Hart, D. J.; Young,
E. R. R. J. Org. Chem. 2000, 65, 5668. (g) Looper, R. E.; Williams, R. M.
Tetrahedron Lett. 2001, 42, 769.
(
6) (a) Harada, K.; Ohtani, I.; Iwamoto, K.; Suzuki, M.; Watanabe, M. F.;
Watanabe, M.; Terav, K. Toxicon 1994, 32, 73. (b) Terav, K.; Ohmori, S.;
Igarashi, K.; Ohtani, I.; Watanabe, M. F.; Harada, K. I.; Ito, E.; Watanabe,
M. Toxicon 1994, 32, 833 and references cited therein.
(
(
(
7) Banker, R.; Carmeli, S.; Hadas, O.; Teltsch, B.; Porat, R.; Sukenik, A. J.
Phycol. 1997, 33, 613.
8) For lead references see: Schembri, M. A.; Neilan, B. A.; Saint, C. P.
EnViron. Toxicol. 2001, 16, 413.
9) Banker, R.; Teltsch, B.; Sukenik, A.; Carmeli, S. J. Nat. Prod. 2000, 63,
(19) Synthetic work in this area has recently been reviewed: Murphy, P. J.;
Thomas, C. W. Chem. Soc. ReV. 2001, 30, 303.
3
87.
3940 J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002

Synthesis of Cylindrospermopsin and 7-Epicylindrospermopsin
A R T I C L E S
Scheme 1
sigmatropic rearrangement.20,21 The plan was to establish the
backbone C-7,8,10 stereogenic centers and the attendant func-
tionality of the natural product employing an intramolecular
cycloaddition of an N-sulfinyl urea, a type of N-sulfinyl
dienophile not previously known. This approach was tested in
a simple model system as outlined in Scheme 1.1 Thus, urea
Scheme 2
7b
(E,E)-diene 4 was converted in situ to the N-sulfinyl urea, which
was found to undergo facile cycloaddition to generate a single
Diels-Alder adduct 6 having the correct relative configuration
at the C-8,10 centers for toxins 1 and 2. This result can be
rationalized by assuming that cycloaddition of a Z-N-sulfinyl
urea22 occurs via the conformation shown in 5. The alternative
conformation 8, which would produce the isomeric adduct 9,
would appear from models to be destabilized vs 5 due to subtle
nonbonded interactions. It is difficult to fully rationalize this
result since no information is available as to the structure of an
N-sulfinyl urea, nor is much known regarding the nature of the
transition state for an N-sulfinyl dienophile Diels-Alder reac-
tion.23 Suffice it to say we were pleased to find that N-sulfinyl
ureas are reactive hetero dienophiles, and that remote stereo-
chemistry is controllable in cycloadditions of such a system.
Using our methodology,21 cycloadduct 6 was subsequently
treated with phenylmagnesium bromide, followed by trimethyl
phosphite in methanol, to produce the cyclindrospermopsin
model 7 having the desired C-7,8,10 stereogenicity and func-
tionality.
(
10) with benzyl chloroformate, followed by a hydroxymethyl
25
anion equivalent in the form of Grignard reagent 11, led to
dihydropyridinone 12 (Scheme 2). Enone 12 could be depro-
tonated and cleanly alkylated with methyl iodide to stereose-
lectively afford the desired trans product 13. As anticipated
based upon the work of Comins and our own earlier studies,
conjugate addition of vinyl cuprate to enone 13 was stereose-
lective and led exclusively to the requisite adduct 14. L-
Selectride reduction of ketone 14 was also stereoselective and
produced the desired alcohol 15 having the four correct A-ring
stereogenic centers of the natural products. This alcohol was
protected as the benzyl ether 16, and subsequent Tamao oxi-
dation of the silane led directly to the cyclic carbamate 17 in
Results and Discussion
2
6
Having successfully tested the key N-sulfinyl Diels-Alder
chemistry, we next turned to construction of a suitably sub-
stituted piperidine A-ring system equivalent to model urea diene
26
17c
4
. Although we have previously reported some success in
utilizing an imino dienophile-based Diels-Alder route to such
1
7c
an intermediate, we decided to explore a potentially more
efficient and direct alternative based upon the dihydropyridinone
chemistry of Comins.24 Thus, treatment of 4-methoxypyridine
17c
2
5
very high overall yield from pyridine 10.
(
20) For reviews of N-sulfinyl dienophile Diels-Alder-based methodology,
see: (a) Weinreb, S. M. Acc. Chem. Res. 1988, 21, 313. (b) Boger, D. L.;
Weinreb, S. M. Hetero Diels-Alder Methodology in Organic Synthesis;
Academic Press: San Diego, 1987; Chapter 1. (c) Weinreb, S. M.
Heterodienophile Additions to Dienes. In ComprehensiVe Organic Synthesis;
Trost, B. M., Fleming, I., Eds.; Pergamon: Oxford, 1991; Vol. 5, p 401.
21) Garigipati, R. S.; Freyer, A. J.; Whittle, R. R.; Weinreb, S. M. J. Am. Chem.
Soc. 1984, 106, 7861.
To elaborate the required diene chain, alkene 17 was first
hydroborated to afford primary alcohol 18, which was then
transformed to the corresponding aldehyde by a Swern oxidation
(
Scheme 3). Coupling of this aldehyde with phosphonate 1927
(
(
cleanly afforded the desired (E,E)-diene ester 20. The ester group
22) For comprehensive reviews of N-sulfinyl compounds, see: Bussas, R.;
Kresze, G.; Munsterer, H.; Schwobel, A. Sulfur Rep. 1983, 2, 215. Schubart,
R. In Houben-Weyl, 4th ed.; Klamann, D., Ed.; Thieme, Stuttgart, 1985;
Vol. E11, p 122. The ground-state configuration of N-sulfinyl compounds
is usually Z although in some systems an E/Z equilibrium has been observed.
23) For theoretical studies of the mechanism of N-sulfinyl Diels-Alder reactions
see: Park, Y. S.; Kim, W. K.; Kim, Y. B.; Lee, I. J. Org. Chem. 2000, 65,
(24) For a review see: Comins, D. L. J. Heterocycl. Chem. 1999, 36, 1491.
(25) Tamao, K.; Ishida, N. Tetrahedron Lett. 1984, 25, 4249. The Grignard
reagent 11 was prepared from the corresponding chloride: Connolly, J.
W.; Fryer, P. F. J. Organomet. Chem. 1971, 30, 315.
(26) See for example: Comins, D. L.; LaMunyon, D. H.; Chen, X. J. Org. Chem.
1997, 62, 8182.
(
3
997.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002 3941

A R T I C L E S
Heintzelman et al.
Scheme 3
Scheme 4
Scheme 5
the cycloaddition most likely occurs via the N-sulfinyl urea/
diene conformation shown in 26. Using our methodology,
treatment of cycloadduct 27 with phenylmagnesium bromide,
followed by trimethyl phosphite in methanol, cleanly led to a
single stereoisomeric allylic alcohol 29. This compound now
possesses all six stereogenic centers of the cyclindrospermopsin
of 20 was next reduced with DIBALH to provide the allylic
alcohol 21, which was then protected as the p-methoxybenzyl
ether 22. Basic hydrolysis of the cyclic carbamate functionality
of 22 subsequently provided the amino alcohol 23. Since the
expectation was that we would be able to differentiate between
a primary and secondary alcohol in the latter stages of the
synthesis (vide infra), we decided to protect the alcohol group
of 23 as the benzyl ether, leading to the dibenzyl compound
3
structure 1 proposed by Moore.
The next stage of the project was to be construction of the
D-ring of the toxin utilizing our efficient, newly developed three-
step method for uracil synthesis starting from an R,â-unsaturated
2
4. The amino group of compound 24 appears to be rather
hindered and thus conversion of this compound to the corre-
sponding urea 25 proved problematic. However, after some
effort it was found that a modification of the method of Magnus
17d
ester. We originally opted to first protect the allylic alcohol
group of intermediate 29 as the MOM ether 30 (Scheme 5).
Removal of the PMB group of 30 led to a crystalline alcohol
28
et al. worked reproducibly to generate the (E,E)-diene urea 25.
3
1 whose structure was determined by X-ray analysis, thereby
With intermediate 25 now in hand, we investigated the pivotal
intramolecular hetero Diels-Alder reaction. We were gratified
to find that simply treating the compound with thionyl chloride/
imidazole in methylene chloride initally at -78 °C and warming
slowly to room temperature led to an excellent yield of a single
tricyclic adduct 27 (Scheme 4). The PMB group of this
compound was removed with DDQ to afford the alcohol 28
whose structure and stereochemistry were firmly established by
X-ray crystallography.29 The formation of dihydrothiazine oxide
fully confirming the stereochemical assignment of key inter-
30
31
mediate 29. Using the procedure of Corey, allylic alcohol
1 was then directly converted to the desired R,â-unsaturated
3
methyl ester 32. Unfortunately, despite the structural similarity
between this compound and the model unsaturated ester
17d
employed in our developmental studies, all attempts to effect
the conjugate addition of a nitrogen nucleophile to 32 failed.
It seemed to us that one possible cause for this problem was
that the bulky allylic MOM group was impeding the conjugate
addition step, and we considered that perhaps changing the
protecting group as well as rigidifying the system might alleviate
this difficulty. We therefore converted urea alcohol 29 into the
2
7 is fully in accord with our model studies (cf. Scheme 1) and
(
27) For preparation of this phosphonate see: (a) Houllemare, D.; Outurquin,
F.; Paulmier, C. J. Chem. Soc., Perkin Trans. 1 1997, 1629. (b) Bæckstr o¨ m,
P.; Jacobsson, U.; Norin, T.; Unelius, C. R. Tetrahedron 1988, 44, 2541.
28) Carter, P.; Fitzjohn, S.; Halazy, S.; Magnus, P. J. Am. Chem. Soc. 1987,
(
(
1
09, 2711.
(30) We thank Dr. D. Powell (University of Wisconsin) for the X-ray analysis
of compound 31. Data have been deposited with the Cambridge Crystal-
lographic Data Centre (CDCC 176837).
(31) Corey, E. J.; Gilman, N. W.; Ganem, B. E. J. Am. Chem. Soc. 1968, 90,
5616.
29) We thank Dr. Louis Todaro (Hunter College) for the X-ray determination
of compounds 28, 45, and 54. Data have been deposited with the Cambridge
Crystallographic Data Centre (28: CCDC 175623; 45: CDCC 171477;
5
4: CDCC 175499).
3942 J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002

Synthesis of Cylindrospermopsin and 7-Epicylindrospermopsin
A R T I C L E S
Scheme 6
cyclic acetonide 33 as has been done in a related cylindrosper-
mopsin model system by Hart (Scheme 6).32 Subsequent
removal of the PMB group gave allylic alcohol 34, which was
then transformed by the sequence shown to the desired R,â-
unsaturated methyl ester 35. We were gratified to find that
treatment of enoate 35 with N,O-bis-trimethylsilylhydroxylamine
in the presence of ethanol in fact afforded the desired conjugate
addition product 36 (a single stereoisomer of undetermined
configuration).33 Exposure of hydroxylamine 36 to phenyl
chloroformate, followed by ammonium hydroxide provided the
anticipated N-hydroxydihydrouracil 37. Elimination of water
Scheme 7
from 37 could then be successfully effected with triflic
anhydride1
7d,34
to provide the fully aromatized uracil D-ring
system 38. Both benzyl groups of 38 could then be removed
simultaneously by catalytic hydrogenolysis to yield diol 39.
We next turned to construction of the remaining guanidine
C-ring of the natural product. Toward this end, a number of
attempts were made to activate the primary alcohol group of
39 for displacement by a nitrogen nucleophile but in general
no reaction took place, probably for steric reasons. Similarly,
this compound was unreactive toward Mitsunobu processes. An
alternative approach that we considered was to first convert the
urea functionality of 39 to the corresponding guanidine, and
then to effect an intramolecular Mitsunobu closure to form the
C-ring as was done in a cylindrospermopsin model study by
Armstrong.1 Therefore, alcohol urea 39 was treated with
triphosgene, followed by ammonium acetate (Scheme 7). To
our surprise, however, the product of this reaction was not a
guanidine but the urea monoacetate 42. Careful examination of
the reaction showed that a reasonably stable, isolable intermedi-
ate was formed upon treatment of 39 with triphosgene, and that
subsequent exposure of this compound to ammonium acetate
led to the observed product 42. Although we have not been
able to definitively establish the structure of this intermediate,
we believe it is probably chloride 40 based upon NMR spectral
data, and the fact that in the mass spectrometer it produces a
molecular ion consistent with oxonium structure 41. SN2
displacement at carbon by acetate ion via 41 would lead to the
observed product 42. Such a ring-opening process has good
8d
35
precedent in related amide-derived systems analogous to 41.
It was in fact possible to use this unexpected result to our
advantage, since when we exposed the triphosgene product 40
to sodium azide in DMF the desired azide 43 was formed in
good yield.
With azide 43 in hand, we expended a considerable amount
of time and effort attempting to construct the guanidine ring
directly from this intermediate. However, we eventually con-
cluded that the uracil moiety was probably interfering with our
various attempts to activate the urea functionality for ring
closure. We therefore decided to investigate a variation of this
strategy involving N-protection of the uracil prior to guanidine
formation. Returning to compound 38, it was possible to intro-
(
32) We are grateful to Professor David Hart (Ohio State University) for
suggesting this type of protection and also for providing experimental
details.1
8f
1
7d
(
33) Interestingly, unlike our model system, adduct 36 did not spontaneously
lactonize to produce the corresponding isoxazolidinone. Attempts to effect
this cyclization under various conditions led primarily to a retro-Michael
reaction.
(35) Hunig, S. Angew. Chem., Int. Ed. Engl. 1964, 3, 548. Gracias, V.; Milligan,
G. L.; Aube, J. J. Org. Chem. 1996, 61, 10. Meyer, F.; Uziel, J.; Papini,
A. M.; Juge, S. Tetrahedron Lett. 2001, 42, 3981. We thank Professor Aube
for providing these references.
(
34) Cf.: Hoffman, R. V.; Nayyar, N. K. J. Org. Chem. 1994, 59, 3530.
Hoffman, R. V.; Nayyar, N. K.; Shankweiler, J. M.; Klinekole, B. W., III
Tetrahedron Lett. 1994, 35, 3231.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002 3943

A R T I C L E S
Heintzelman et al.
Scheme 8
Scheme 9
duce MOM protecting groups onto the uracil nitrogens using
groups of 48 could subsequently be cleaved by vigorous acidic
hydrolysis to afford diol guanidinium chloride 49. Surpris-
ingly, diol 49 was found to be different when it was directly
3
6
18c
the procedure of Arias and co-workers to afford 44, which
was then converted to the crystalline diol 45 via catalytic hydro-
genolysis (Scheme 8).37 An X-ray analysis of 45 was conducted
as further verification of the structures of these late-stage
compared to the corresponding intermediate produced in
18c,39
Snider’s total synthesis of cylindrospermopsin.
In particular,
2
9
intermediates. Exposure of diol 45 to triphosgene followed
by sodium azide provided the desired azide 46. Once again,
formation of an intermediate like 40 could be observed in this
reaction. The acetonide protecting group of 46 could then be
selectively hydrolyzed with dilute HCl to produce urea diol 47.
We were pleased to find that urea activation of 47 with
MeOTf, followed by catalytic hydrogenation of the azide, now
proceeded smoothly, leading directly to the tetracyclic guani-
dinium compound 48 (isolated as the formate salt).38 The MOM
our compound had δC7 4.50 (J7,8 ) 6.6 Hz, in D2O) in accord
9
with that reported for 7-epicylindrospermopsin, compared to
δC7 4.70 (J7,8 ) 4.0 Hz) for both the Snider diol and
3
18c
cylindrospermopsin. Using Snider’s protocol our diol 49 was
then selectively converted to the monosulfate 1 (70%) along
with some of the bis-sulfate 50 (25%). Synthetic compound 1
was in fact found to have NMR spectra identical with those of
natural 7-epicylindrospermopsin and substantially different from
40
those of cylindrospermopsin.
However, to further confirm this unanticipated result, we
decided to also prepare the cylindrospermopsin structure 2 from
one of our late synthetic intermediates. Therefore, the acetonide
(
36) Arias, L.; Guzman, A.; Jaime-Figueroa, S.; Lopez, F. J.; Morgans, D. J.,
Jr.; Padilla, F.; Perez-Medrano, A.; Quintero, C.; Romero, M.; Sandoval,
L. Synlett 1997, 1233.
(
37) We have also prepared the N-BOM and N-benzyl uracil protected series
of compounds. In both cases it was possible to successfully construct the
guanidine rings via the methodology used for the MOM series (cf. Scheme
(38) Overman, L. E.; Rabinowitz, M. H.; Renhowe, P. A. J. Am. Chem. Soc.
1995, 117, 2657.
(39) Thanks are due to Professor Barry Snider (Brandeis University) for
providing NMR spectra and a comparison sample of synthetic diol 57, along
with spectra of cylindrospermopsin.
8
) but in these cases the conditions required either for hydrogenolytic
removal of the protecting groups or for azide hydrogenation led to
overreduction of the uracil ring.
3944 J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002

Synthesis of Cylindrospermopsin and 7-Epicylindrospermopsin
A R T I C L E S
functionality of intermediate 38 was first removed by acid
hydrolysis to afford alcohol 51 (Scheme 9). This compound
could be cleanly inverted by a Mitsunobu process4 to give the
desired C-7 alcohol epimer 52. Removal of the two benzyl
groups of 52 by catalytic hydrogenolysis then provided triol
(i.e., cylindrospermopsin is now formulated as 2 and 7-epicy-
clindrospermopsin is 1). Also, it presently seems unlikely that
the toxins exist in the hydrogen-bonded conformations repre-
sented by 1a and 2a, nor is it clear at this point whether the
uracil actually assumes the unusual tautomeric form shown in
1
4
2
53. We were pleased to find that treatment of this triol with
these structures.
triphosgene (for spectral data on the intermediate see the
Supporting Information) followed by sodium azide in DMF
cleanly led to regioselective formation of the requisite mono
azide diol 54, whose structure and stereochemistry were verified
by X-ray crystallography.29 Prior to guanidine formation, the
diol 54 was protected as the diacetate 55 since the diol itself
was not sufficiently soluble in methylene chloride for use in
the next step. Activation of the urea functionality of 55 with
MeOTf followed by hydrogenation of the azide led to the desired
guanidinium salt 56. Finally, acidic hydrolysis of the acetyl and
MOM groups afforded the tetracyclic diol 57, which had spectra
In conclusion, we have accomplished total syntheses of the
two algal hepatotoxins cylindrospermopsin (2) and 7-epicylin-
drospermopsin (1) in approximately 30 steps starting from
4-methoxypyridine, leading to a revision of the C-7 stereo-
chemistry of the natural products. Our approach utilizes a novel
stereospecific intramolecular [4+2]-cycloaddition of an N-
sulfinylurea heterodienophile and application of our new uracil
1
7c
synthesis as pivotal steps.
Acknowledgment. We are grateful to the National Science
Foundation (CHE-9732038 and CHE-0102402) for financial
support of this research and the National Institutes of Health
for a postdoctoral fellowship (1F32GM-20664) to G.A.W. We
also thank Professor J. D. White for a preprint of ref 5.
3
9
identical with those of the Snider synthetic material. This
compound has previously been converted to cylindrospermopsin,
thus verifying that the toxin is actually represented by stereo-
structure 2.1
8c
Supporting Information Available: Full experimental details
and spectral data for new compounds (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
It is evident, therefore, that the original Moore assignment
of C-7 stereochemistry for cylindrospermopsin is incorrect, and
that the structures of the two toxins have in fact been reversed
JA020032H
(
40) We thank Professor Shmuel Carmeli (Tel Aviv University) for copies of
the proton and carbon NMR spectra of natural 7-epicylindrospermopsin.
41) Martin, S. F.; Dodge, J. A. Tetrahedron Lett. 1991, 32, 3017.
(42) For a discussion of uracil tautomers see: Kryachko, E. S.; Nguyen, M. T.;
Zeegers-Huyskens, T. J. Phys. Chem A 2001, 105, 1934 and references
cited therein.
(
J. AM. CHEM. SOC.
9
VOL. 124, NO. 15, 2002 3945