First asymmetric total synthesis of tetrodotoxin

Article abstract of DOI:10.1021/ja0342998

Tetrodotoxin, a toxic principle of puffer fish poisoning, is one of the most famous marine natural products because of the complex structure having many functional groups and its potent biological activity leading to death. Since the structure elucidation in 1964, this toxin has been recognized as a formidable target molecule for total synthesis. We have recently achieved the first asymmetric total synthesis from 2-acetoxy-tri-O-acetyl-D-glucal as a chiral starting material. The highly hydroxylated cyclohexane ring was constructed by Claisen rearrangement and regioselective hydroxylations of an acetone moiety and an intramolecular directed aldol condensation of the precursor having methyl ketone with dihydroxyacetone, which was synthesized through Sonogashira coupling. Installation of nitrogen functionality was unsuccessful through an attempted Overman rearrangement. We, therefore, employed a new intramolecular conjugate addition strategy between the carbamate and unsaturated ester groups. The α-hydroxyl lactone moiety was synthesized through an intramolecular epoxide opening by the Z-enolate of aldehyde, which was followed by oxidation - reduction of the resulting cyclic vinyl ether. The lactone was then converted to a protected ortho ester, and then gunanidinylation was followed by cleavage of the 1,2-glycol to give the fully protected tetrodotoxin. Selection of the protective groups has finally led us to accomplish the total synthesis of tetrodotoxin in an enantiomerically pure form. All the stereogenic centers were controlled with high selectivity, and the hydroxyl groups were differently protected to discriminate for the future analogue synthesis of a bioorganic program. The synthetic tetrodotoxin was purifed by ion exchange chromatography and characterized to be identifical with the natural compound.

Full text of DOI:10.1021/ja0342998

Published on Web 06/28/2003
First Asymmetric Total Synthesis of Tetrodotoxin
Norio Ohyabu, Toshio Nishikawa, and Minoru Isobe*
Contribution from the Laboratory of Organic Chemistry, School of Bioagricultural Sciences,
Nagoya UniVersity, Furocho, Chikusa, Nagoya 464-8601, Japan
Received January 23, 2003; E-mail: isobem@agr.nagoya-u.ac.jp
Abstract: Tetrodotoxin, a toxic principle of puffer fish poisoning, is one of the most famous marine natural
products because of the complex structure having many functional groups and its potent biological activity
leading to death. Since the structure elucidation in 1964, this toxin has been recognized as a formidable
target molecule for total synthesis. We have recently achieved the first asymmetric total synthesis from
2-acetoxy-tri-O-acetyl-D-glucal as a chiral starting material. The highly hydroxylated cyclohexane ring was
constructed by Claisen rearrangement and regioselective hydroxylations of an acetone moiety and an
intramolecular directed aldol condensation of the precursor having methyl ketone with dihydroxyacetone,
which was synthesized through Sonogashira coupling. Installation of nitrogen functionality was unsuccessful
through an attempted Overman rearrangement. We, therefore, employed a new intramolecular conjugate
addition strategy between the carbamate and unsaturated ester groups. The R-hydroxyl lactone moiety
was synthesized through an intramolecular epoxide opening by the Z-enolate of aldehyde, which was
followed by oxidation-reduction of the resulting cyclic vinyl ether. The lactone was then converted to a
protected ortho ester, and then gunanidinylation was followed by cleavage of the 1,2-glycol to give the
fully protected tetrodotoxin. Selection of the protective groups has finally led us to accomplish the total
synthesis of tetrodotoxin in an enantiomerically pure form. All the stereogenic centers were controlled with
high selectivity, and the hydroxyl groups were differently protected to discriminate for the future analogue
synthesis of a bioorganic program. The synthetic tetrodotoxin was purifed by ion exchange chromatography
and characterized to be identifical with the natural compound.
Introduction
by hydroxyl groups, an ortho ester showing acidity (pKa ) 8.7),
and cyclic guanidine with hemiaminal. Tetrodotoxin exists as
an equilibrium mixture among the ortho ester (1), anhydride
(2), and lactone (3) forms, as shown in Figure 1.
Puffer fish poisoning has been a very famous yet serious
problem, especially in Japan, where the puffer fish has been
long time recognized as one of the most delicious seafoods.1
The toxic principle was first isolated from the ovaries of the
puffer fish (Spheroides rubripes) in 1909 and named as
tetrodotoxin (TTX) after the puffer fish family “Tetraodon-
tidae.”2 Despite the small molecule, however, the structure
elucidation had been extremely difficult at the time, because of
the unusual chemical properties and the unprecedented structure.
After extensive efforts, three groups including Hirata-Goto,3
Tsuda,4 and Woodward5 independently arrived at the same
structure 1 (Figure 1) for tetrodotoxin.6 Absolute stereochemistry
of tetrodotoxin 1 was determined unambiguously by X-ray
crystallographic analysis of its derivative.7 The structural features
are an unprecedented dioxa-adamantane skeleton functionalized
Since the toxicity of tetrodotoxin was revealed in the 1960s
to be attributed to a specific blockage of sodium ion influx
through sodium channel proteins, tetrodotoxin has been widely
employed in pharmacological studies.8 And further extensive
studies culminated in the identification and isolation of the ion-
channel protein and determination of its amino acid sequence.9
Concerning the total synthesis of this fascinating and challenging
natural product, Kishi-Goto and their co-workers achieved the
racemic total synthesis in 1972.10 Since then, there have been
extensive synthetic efforts devoted to this small natural prod-
uct,11 yet the Kishi-Goto total synthesis had still stood as the
only synthesis of tetrodotoxin 1. The bioorganic studies were
reported through a difficult derivatization from natural tetro-
(1) Review: Tetrodotoxin, Saxitoxin, and the Molecular Biology of the Sodium
Channel. Kao, C. Y.; Lovinson, S. Eds. Ann. N. Y. Acad. Sci. 1986, 479.
(2) (a) Tahara, Y. J. Pharm. Soc. Jpn. (in Japanese) 1909, 29, 587-625. (b)
Tahara, Y. Biochem. Z. 1911, 30, 255-275.
(8) (a) Narahashi, T.; Moore, J. W.; Scott, W. R. J. Gen. Physiol. 1964, 47,
965-974. (b) Hucho, F. Angew. Chem, Int. Ed. Engl. 1995, 34, 39-50.
(9) Review: Numa, S.; Noda, M. Ann. N.Y. Acad. Sci. 1986, 479, 338-355.
(10) (a) Kishi, Y.; Aratani, M.; Fukuyama, T.; Nakatsubo, F.; Goto, T.; Inoue,
S.; Tanino, H.; Sugiura, S.; Kakoi, H. J. Am. Chem. Soc. 1972, 94, 9217-
9219. (b) Kishi, Y.; Fukuyama, T.; Aratani, M.; Nakatsubo, F.; Goto, T.;
Inoue, S.; Tanino, H.; Sugiura, S.; Kakoi, H. J. Am. Chem. Soc. 1972, 94,
9219-9221.
(11) For leading references from other laboratories, see: (a) Itoh, T.; Watanabe,
M.; Fukuyama, T. Synlett 2002, 1323-1325. (b) Noya, B.; Paredes, M.
D.; Ozores, L.; Alonso, R. J. Org. Chem. 2000, 65, 5960-5968. (c) Fraser-
Reid, B.; Burgey, C. S.; Vollerthun, R. Pure Appl. Chem. 1998, 61, 285-
288. (d) Sato, K.; Kajihara, Y.; Nakamura, Y.; Yoshimura, J. Chem. Lett.
1991, 1559-1562. (e) Nachman, R. J.; Ho¨nel, M.; Williams, T. M.; Halaska,
(3) Goto, T.; Kishi, Y.; Takahashi, S.; Hirata, Y. Tetrahedron 1965, 21, 2059-
2088.
(4) Tsuda, K.; Ikuma, S.; Kawamura, M.; Tachikawa, R.; Sakai, K.; Tamura,
C.; Amakasu, O. Chem. Pharm. Bull. 1964, 12, 1357-1374.
(5) Woodward, R. B. Pure. Appl. Chem. 1964, 9, 49-74.
(6) The structure was finally confirmed by X-ray analysis of tetrodotoxin
hydrobromide. See: Furusaki, A.; Tomie, Y.; Nitta, I. Bull. Chem. Soc.
Jpn. 1970, 43, 3332-3341.
(7) Furusaki, A.; Tomie, Y.; Nitta, I. Bull. Chem. Soc. Jpn. 1970, 43, 3325-
3331.
9
8798
J. AM. CHEM. SOC. 2003, 125, 8798-8805
10.1021/ja0342998 CCC: $25.00 © 2003 American Chemical Society

Asymmetric Total Synthesis of Tetrodotoxin
A R T I C L E S
Figure 1. Equilibrium of tetrodotoxin.
Scheme 1. Retrosynthesis and Synthetic Plan
dotoxin.12 Therefore, we need analogous compounds through
chemical synthesis.
moiety with hemiaminal has been prepared from a guanidine
group at the C-8a position and an aldehyde at the C-4 position.16
An intermediate 4 bearing ortho ester and amino aldehyde was
envisaged as the synthetic equivalent of tetrodotoxin (Scheme
1).17 A similar ortho ester intermediate was, in fact, crucial for
successful synthesis of 8,11-dideoxytetrodotoxin.15 Employment
of the ortho ester instead of δ-hydroxyl lactone seemed to be
more advantageous not only for the protecting the carboxylic
acid with two hydroxyl groups at the C-5 and C-7 positions
but also for preventing the C-5 hydroxyl function from
â-elimination.18 The ortho ester form would also inhibit epimer-
ization at the C-9 position.19 The amino group, a pivotal function
for guanidine installation, would be protected as trichloro-
acetamide, while the aldehyde could be prepared from 1,2-glycol
protected as acetonide. These analyses led us to find the lactone
structures (5, 6 as three- and two-dimensional expressions) as
an important intermediate. According to the successful syntheses
of deoxytetrodotoxins in our laboratory, the lactone 6 was then
simplified into vinyl epoxide 7. It was anticipated that nitrogen
functionality would be introduced through Overman rearrange-
ment20 from exo-allylic alcohol 8, in which the stereochemical
course should be controlled by allylic strain between acetonide
and allylic trichloroacetimidate.21 Consequently, one of the main
issues in the current plan was stereocontrolled synthesis of the
In our project directed toward the total syntheses of tetro-
dotoxin and its natural and unnatural analogues, we have
recently completed the syntheses of 5,11-dideoxytetrodotoxin,13
11-deoxytetrodotoxin,14 and 8,11-dideoxytetrodotoxin15 in enan-
tiomerically pure form on the basis of a Diels-Alder strategy
from levoglucosenone. Among them, 11-deoxytetrodotoxin was
the first total synthesis of a naturally occurring tetrodotoxin
analogue. The full details of the total synthesis of tetrodotoxin
(1) are described here on a different basic strategy aiming at a
higher oxidation stage in earlier stages of the synthesis.
Retrosynthesis and Synthetic Plan
For the asymmetric synthesis toward 1, 2, and 3, we had
aimed at retrosynthesis ultimately into a readily available sugar
derivative through the chemistry which we have learned during
the course of the TTX-analogue syntheses. The cyclic guanidine
R. C.; Mosher, H. S. J. Org. Chem. 1986, 51, 4802-4806. (f) Keana, J. F.
W.; Bland, J. S.; Boyle, P. J.; Erion, M.; Hartling, R.; Husman, J. R.;
Roman, R. B. J. Org. Chem. 1983, 48, 3621-3631. (g) Speslacis, J. Ph.D.
Thesis, Harvard University, 1975.
(12) (a) Balerna, M.; Lombet, A.; Chicheportiche, R.; Romey, G.; Lazdunski,
M. Biochim. Biophys. Acta. 1981, 644, 219-225. (b) Bontemps, J.;
Cantineau, R.; Grandfils, C.; Leprince, P.; Dandrifosse, G.; Schoffeniels,
E. Anal. Biochem. 1984, 139, 149-157. (c) Wu, B. Q.; Yang, L.; Kao, C.
Y.; Levinson, S. R.; Yotsu-Yamashita, M.; Yasumoto, T. Toxicon 1996,
34, 407-416. (d) Nakayama, H.; Hatanaka, Y.; Yoshida, E.; Oka, K.;
Takanohashi, M.; Amano, Y.; Kanaoka, Y. Biochem. Biophys. Res.
Commun. 1992, 184, 900-907.
(13) (a) Nishikawa, T.; Asai, M.; Ohyabu, N.; Yamamoto, N.; Isobe, M. Angew.
Chem., Int. Ed. 1999, 38, 3081-3084. (b) Asai, M.; Nishikawa, T.; Ohyabu,
N.; Yamamoto, N.; Isobe, M. Tetrahedron 2001, 57, 4543-4558.
(14) Nishikawa, T.; Asai, M.; Isobe, M. J. Am. Chem. Soc. 2002, 124, 7847-
7852.
(16) The numbering used in this paper corresponds to that of tetrodotoxin.
(17) The compound 5 is an ortho ester form of a tetrodamine intermediate named
by Kishi. See ref 10.
(18) This problem was mentioned in Kishi’s total synthesis; see ref 10.
(19) Since the problem was encountered in the synthesis of (-)-5,11-dideox-
ytetrodotoxin, we devised intramolecular acetal between the hydroxyl group
at the C-9 position and aldehyde at the C-4 position to protect the labile
C-9 position. See ref 13.
(15) Nishikawa, T.; Urabe, D.; Yoshida, K.; Iwabuchi, T.; Asai, M.; Isobe, M.
Org. Lett. 2002, 4, 2679-2682.
9
J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003 8799

A R T I C L E S
Ohyabu et al.
oxygenated precursor 8 for Overman rearrangement. In our
previous retrosynthesis toward the deoxy tetrodotoxin analogues,
a Diels-Alder cycloaddition strategy was employed for the
cyclohexene ring construction.22 An alternative strategy was
based on electrocyclization for the cyclohexene moiety.23
Besides these two basic strategies, we added the third strategy
for the synthesis of the oxygenated cyclohexane skeleton as 8
via a different route. Exoolefin 8 could be prepared from dienone
9 through a directed aldol cyclization of 10 and Claisen
rearrangement of 11 as conceptual key steps. The enone was
envisaged to derive from an intramolecular aldol condensation
of 10a; methyl ketone could be prepared by hydration of
acetylene, while the dihydroxyacetone moiety might need regio-
and stereoselective hydroxylation to the acetone moiety in 10b.
The acetone at the C-4a position would be obtainable via Claisen
rearrangement of allyl isopropenyl ether 11. The ethynyl group
would be introduced via Sonogashira coupling of 12 (X )
halogen) with an acetylene. We have accessed the readily
available 2-acetoxy-tri-O-acetyl-D-glucal 13 as the chiral starting
material.
presence of potassium carbonate30 to afford 19 as the only
isolable product in high yield.
Scheme 2. Sonogashira Coupling and Claisen Rearrangement
Strategy
Synthesis of the Cyclohexane Skeleton
Synthesis started with allylic alcohol 14, readily prepared in
two steps from 2-acetoxy-tri-O-acetyl-D-glucal 13 (Scheme 2).24
The primary alcohol of 14 was protected with tert-butyldi-
methylsilyl chloride (TBS-Cl) in the presence of triethylamine,
while the secondary alcohol was oxidized with SO3•Py, DMSO,
and Et3N (Parikh-Doering oxidation)25 to give enone 15.
Iodination of 15 was carried out with iodine in pyridine and
CH2Cl226 to give R-iodoenone, which was subsequently reduced
under Luche’s conditions27 to 16 as a single product. Sono-
gashira coupling28 of iodide 16 with (trimethylsilyl)acetylene
gave eneyne 17 in quantitative yield. Acetone moiety was
installed to the C-4a position of 17 via Claisen rearrangement
of allyl-2-propenyl ether 18. Thus, 17 was treated with 2-meth-
oxypropene in the presence of pyridinium p-toluenesulfonate
(PPTS) to give a mixture of unstable propenyl ether 18, methyl
acetal, and the starting alcohol 17, which were separated on
silica gel (neutral) column chromatography to give 18 in 56%
yield.29 The recovered materials could be converted to 18 under
the same conditions. After three cycles, an overall yield of 89%
was obtained for 18. The Claisen rearrangement of 18 proceeded
upon heating at 150 °C in o-dichlorobenzene (o-DCB) in the
Hydroxylations of both neighboring sides of the acetone
moiety were carried out stepwise by regioselective enolization
followed by oxygenation (Scheme 3). The ketone 19 was first
treated with TBSOTf and triethylamine to give exclusively the
corresponding vinyl silyl ether at the terminal position, which
was then oxidized with lead tetraacetate (LTA).31 The resulting
crude products, presumably an acetyl silyl acetal, were exposed
to n-Bu4NF (TBAF) to give acetoxyacetone 20, which was
deacetylated to afford hydroxyacetone 21 in excellent overall
yield. Hydroxylation of the internal position was difficult due
to little enolization toward the C-5 position, so 21 was once
converted to R-aldehyde ketone, and then it was tried for the
enolization due to dipole-dipole interaction. Thus, upon
Parikh-Doering oxidation of 21, the resulting R-aldehyde
ketone spontaneously enolized to 22 as an unstable product.
After protection of the enol as methoxymethyl (MOM) ether,
the aldehyde was reduced under the Luche’s reductions to give
23. Geometry of the enol ether was established to be Z from
NOESY correlation as shown in structure 23. Epoxidation of
23 with m-chloroperbenzoic acid (MCPBA) in the presence of
potassium carbonate32 gave unstable products, which were
directly hydrolyzed with Amberlite 15 ion-exchange resin (H+
form) to give dihydroxylacetone 24 as a 7:1 mixture of the
diastereomers; thus, the major product was separated and
isolated in 74% yield along with the minor isomer in about 10%
yield. As it turned out later, the configuration of C-5 was
revealed to be S. The stable conformation of enol ether 23 is
shown in Figure 2, which was assigned from two large coupling
constants (10.5 Hz both) between H-4 and H-4a and H-4a and
H-5 in the 1H NMR spectra. This diastereomeric selectivity (ca.
7:1) might be rationalized by the cooperative effect33 of
hydrogen bonds of MCPBA with the allylic alcohol and oxygen
(20) For reviews of the Overman rearrangement, see: (a) Overman, L. E. Acc.
Chem. Res. 1980, 13, 218-224. (b) Ritter, K. In Houben-Weyl. Stereose-
lectiVe Synthesis. E 21; Helmchen, G., Hoffmann, R. W., Mulzer, J.,
Schaumann, E., Eds.; Thieme: Stuttgart, Germany, 1996; Vol. 9, pp 5677-
5699.
(21) Isobe, M.; Fukuda, Y.; Nishikawa, T.; Chabert, P.; Kawai, T.; Goto, T.
Tetrahedron Lett. 1990, 31, 3327-3330.
(22) Isobe, M.; Nishikawa, T.; Pikul, S.; Goto, T. Tetrahedron Lett. 1987, 28,
6485-6488. (b) Isobe, M.; Yamamoto, N.; Nishikawa, T. In LeVoglucose-
none and LeVoglucosans, Chemistry and Applications. Witczak, Z. J. Ed.;
ATL Press: Mount Prospect, IL, 1994; pp 99-118.
(23) Bamba, M.; Nishikawa, T.; Isobe, M. Tetrahedron 1998, 54, 6639-6650.
(24) Ichikawa, Y.; Isobe, M.; Bai, D.-L.; Goto, T. Tetrahedron 1987, 43, 4737-
4776.
(25) Parikh, J. R.; Doering, W. von E. J. Am. Chem. Soc. 1967, 89, 5505.
(26) Johnson, C. R.; Adams, J. P.; Braun, M. P.; Senanayke, C. B. W.;
Wovkulich, P. M.; Uskokovic, M. R. Tetrahedron Lett. 1992, 33, 917-
918.
(27) Germal, A. L.; Luche, J. L. J. Am. Chem. Soc. 1981, 103, 5454-5459.
(28) Sonogashira, K.; Tohda, Y.; Hagiwara, N. Tetrahedron Lett. 1975, 16,
4467-4470.
(29) Since the product 18 was sensitive to acid as silica gel, neutral silica gel
with eluent containing a small amount of triethylamine was employed for
this chromatography. For details, see experimental procedure in the
Supporting Information.
(30) In the absence of K2CO2, some degree of hydrolysis to 17 was observed.
(31) (a) Rubottom, G. M.; Gruber, J. M.; Kincaid, K. Synth. Commun. 1976, 6,
59-62. (b) Rubottom, G. M.; Gruber, J. M. J. Org. Chem. 1976, 41, 1673-
1674.
(32) In the presence of Na2HPO4 instead of K2CO3, the resulting epoxides were
partially incorporated with the residual m-chlorobenzoic acid to give a
mixture of 24 and the byproducts, which upon hydrolysis (K2CO3, MeOH)
were converted to 24 in ca. 30% yield from 23.
9
8800 J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003

Asymmetric Total Synthesis of Tetrodotoxin
A R T I C L E S
Scheme 3. Synthesis of the Cyclohexenone
Figure 2. (a) Conformation of 23 and (b) a possible interaction with
MCPBA.
Figure 3. Proof of the stereochemistry at the C5 position.
Scheme 4. Exoolefin Synthesis
which was further silylated with TBS-Cl in the presence of
imidazole to afford 27,35 a precursor for directed aldol conden-
sation. Intramolecular aldol reaction was carried out by n-Bu4-
NF (TBAF),36 and subsequent dehydration with trichloroacetyl
chloride afforded enone 28 in good overall yield. The large
coupling constant, 9 Hz between H-5 and H-4a, confirmed the
configuration of the C-5 position as well as the trans diaxial
relationship of these two protons (Figure 3b).
of the siloxy group of 23 as depicted in Figure 2b rather than
steric hindrance of the TBS group. Although the configuration
was opposite for tetrodotoxin, the major product 24 was
employed for further studies with success.
Selective protection of the primary alcohol of 24 with a tert-
butyldiphenylsilyl (TBDPS) group and subsequent benzoylation
of the secondary alcohol gave 25. Acetylene 25 was successively
treated with sulfuric acid and a catalytic amount of mercuric
oxide to afford methyl ketone 26. Thus, with treatment with
sulfuric acid in methanol, the TBS group was immediately
removed to form intramolecular acetal, and then iso-propyl
acetal was replaced with methanol. When mercuric oxide was
added to the resulting mixture, acetylene underwent hydration
to methyl ketone 26 in 82% overall yield. With this bicyclic
compound, the configuration of the C-5 position was established
to be S by a small coupling constant of 2 Hz between protons
at C-5 and C-4a in the 1H NMR spectra, indicating axial
arrangement of the C-5 benzoyloxy group (Figure 3a).34 The
methyl ketone of 26 was then treated with TBSOTf in the
presence of Et3N and 2,6-lutidine to give alkenyl silyl ether,
(34) Interestingly, configuration of the C-5 position affected hydration of
acetylene; the reaction of the R isomer under the identical conditions did
not give the corresponding product. The hydration of the acetylene 25
without the hydroxyl group at the C-5 position necessitated harsh conditions
(2 equiv of HgO in 1.26 M aq H2SO4:acetone ) 1:1, under reflux for 30
min) to give the corresponding ketone in 78% yield. These results imply
the benzoate ester group of S-isomer might participate the acetylene moiety
in the hydration as shown below.
(35) Two step silylation was necessary for high yield of the product 27. Under
the first conditions, silylation of the primary alcohol was not effective to
give a much lower yield of 27.
(36) Because of intramolecular reaction, water did not interfere with the fluoride
induced aldol reaction.
(33) (a) Johnson, M. R.; Kishi, Y. Tetrahedron Lett. 1979, 4347-4350. (b) Isobe,
M.; Kitamura, M.; Mio, S.; Goto, T. Tetrahedron Lett. 1982, 23, 221-
224.
9
J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003 8801

A R T I C L E S
Ohyabu et al.
Thus, imidate 36 was prepared by treatment of 32 with 1,8-
diazabicylco[5.4.0]undec-7-ene (DBU) and trichloroacetonitrile
(Scheme 6). Considering the allylic strain observed in 34, we
expected the conformation of 32 to be suitable for the rear-
rangement, as shown in Scheme 6.42 To our surprise, however,
the attempted Overman rearrangement of 36 even under the
improved conditions (at xylene reflux in the presence of K2-
CO3) did not give the desired product 37 but gave 1,3-shift
product 38 as an only isolable product.43 In this case, the
benzoyloxy substituent occupied the axial position of the C-5,44
which might be steric hindrance against incoming nitrogen of
imidate 36. However, the corresponding epimer at the C-5
position45 also gave 38 under the same conditions. These
unsuccessful results enforced us to explore an alternative route
for installation of nitrogen functionality to the highly congested
C-8a position.
Figure 4. Proof of the configuration of the C-8 position.
The enone of 28 was reduced under Luche’s conditions to
give R-alcohol as a single product (Scheme 4). The configuration
of the resulting alcohol was determined by observing the
NOESY correlation between the protons of the C-4a and C-8
positions of the corresponding BOM (benzyloxymethyl) ether
29 (Figure 4). Further conversion to exo-allylic alcohol 32, a
precursor for the Overman rearrangement, proved to be difficult
because of facile intramolecular acetal (1,6-anhydro ring in sugar
numbering) formation and instability of the hydroxyl group at
the double allylic C-8 position. In this specific case, the resulting
anhydro byproduct could not be converted to 32 due to the
presence of a labile alkoxy group at the C-8 position. Thus, the
BOM group was carefully chosen as the protective group for
the C-8 hydroxyl group because of the low leaving ability, high
compatibility with the further transformation, and the mild
deprotection conditons. The TBS ether of 29 was converted to
the corresponding acetate 30 through desilylation with dl-
camphorsulfonic acid (CSA) in methanol followed by acetyla-
tion. Protection of the primary alcohol was necessary to prevent
1,6-anhydro ring formation under the next reaction. Hydrolysis
of the acetal 30 was best carried out by mercuric oxide,37 and
further treatment with PPTS in aqueous acetone and subsequent
mild deacetylation with Mg(OEt)238 afforded 31. The hemiacetal
of 31 was easily reduced with sodium borohydride, and the
resulting triol was protected as acetonide to furnish 32 in good
overall yield.39 This is now set for the Overman rearrangement.
Scheme 6. Problem of Overman Rearrangement
To overcome this critical problem, we probed alternative
intramolecular processes for this purpose. As the hydroxyl group
of the C-5 position in 32 was located at the axial position, a
possible intramolecular SN2′ reaction to an allyl function (39a)
or conjugate addition to an unsaturated ester (39b) appeared to
be the candidates (Scheme 7) for the nitrogen nucleophile
tethering to the hydroxyl group at the C-5 position.
Introduction of Nitrogen Functionality
In the synthesis of the common intermediate 35a aiming at
a variety of analogues of tetrodotoxin (Scheme 5), we developed
an improved condition (K2CO3 in xylene, reflux) for the
Overman rearrangement of simple exo-allylic alcohol 33a to
give 35a in high yield with reproducibility.40 The improved
condition enabled us to effect the Overman rearrangement of
hydroxylated substrate 33b to afford 35b.41 These successful
results encouraged us to employ the more hydroxylated substrate
32 as a precursor for the Overman rearrangement.
After some exploratory experiments, we found that conjugate
addition of a carbamate was feasible as a nitrogen nucleophile
(37) This hydrolysis was accelerated with mercuric oxide. In the absence of
HgO, harsh conditions (e.g., CSA (100 mg/mL) in aqueous acetone, room
temp, 1 day) were necessary to hydrolyze the methyl acetal to give a mixture
of the desired product and the 1,6-anhydro product in 53% and 30% yield,
respectively.
Scheme 5. Previous Examples of Overman Rearrangement
(38) Xu, Y.-C.; Bizuneh, A.; Walker, C. Tetrahedron Lett. 1996, 37, 455-458.
(39) In sharp contrast, harsh conditions with LiAlH4 in refluxing Et2O were
required for the reduction of the simpler hemiacetal for the synthesis of
33a (Scheme 5), while NaBH4 at rt did not reduce the same hemiacetal.
(40) Nishikawa, T.; Asai, M.; Ohyabu, N.; Yamamoto, N.; Fukuda, Y.; Isobe,
M. Tetrahedron 2001, 57, 3875-3883.
(41) Nishikawa, T.; Asai, M.; Ohyabu, N.; Isobe, M. J. Org. Chem. 1998, 63,
188-192.
(42) This analysis was supported by a coupling constant (0 Hz) between H5
and H4a of 1H NMR spectra of 32.
(43) Similar types of the undesired products in the Overman rearrangement of
highly oxygenated allylic imidate were reported; see: (a) Gonda, J.;
Bednarikova, M. Tetrahedron Lett. 1997, 38, 5569-5572. (b) Eguchi, T.;
Kakinuma, K. Yuki Gosei Kagaku Kyokaishi 1997, 55, 814-823.
(44) The axial alignment was supported by a very small coupling constant
between H5 and H-4a.
(45) The epimer was prepared from the epimer of 32, which was synthesized
from 29 in 11 steps including reductive removal of benzoate with DIBAL-H
followed by inversion of the configuration through oxidation-reduction
and the similar transformation as in the case of 32.
9
8802 J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003

Asymmetric Total Synthesis of Tetrodotoxin
A R T I C L E S
Scheme 7. Intramolecular Conjugate Addition for Nitrogen
Introduction
conformation as shown in 42, which was suitable for the
intramolecular conjugate addition.51 Alkaline cleavage of cyclic
carbamate 43 through the N-Boc intermediate gave rise to the
retro Michael reaction with concomitant aromatization of the
cyclohexane ring. Consequently, the ester group of 43 was first
reduced with LiBH4, and the resulting alcohol was protected
as p-methoxyphenyldiphenylmethyl (MMTr) ether 44a. The
cyclic carbamate of 44a was then cleaved through N-Boc
intermediate 44b followed by hydrolysis with LiOH52 to afford
45. Under the alkaline conditions, the TBDPS group was also
removed. Here, the Boc-protected amino group, an important
function for guanidinylation, was introduced at the C-8a position.
Scheme 8. Installation of the Nitrogen Functionality
Stereoselective Synthesis of the Lactone Ring
Allylic alcohol 45 was epoxidized with MCPBA and then
the primary alcohol was protected as benzoate 46 (Scheme 9).
The configuration of the C-5 position was inverted by a two-
step sequence involving oxidation with Albright-Goldman
procedure (acetic anhydride and DMSO)53 and reduction with
NaBH4 to give a single product, which was protected as acetate
47 in high overall yield. As 47 was equipped with no vinyl
group that had been utilized for synthesis of R-hydroxy lactone
in the previous syntheses of deoxy analogues of tetrodotoxin,
new methodology for this purpose became necessary.54 We
initially planned to synthesize R-hydroxylactone such as 50
through lactone 49, which would be obtained from 47. Thus,
deprotection of MMTr with TFA was followed by two step
oxidation of the resulting alcohol with o-iodoxybenzoic acid
(IBX) in DMSO55 and then sodium chlorite to give lactone 49.
However, all attempts of hydroxylation at the R-position were
unsuccessful. In the event, we fortunately found that heating
aldehyde 48 with DBU in o-DCB at 130 °C provided cyclic
vinyl ether 52 via Z-enolate 51.56 The selective formation of
Z-enol (51) might be attributed to the abstraction of the axial
proton at the C-9 position through a chelated intermediate as
in Figure 5, and formation of the resulting cyclic vinyl ether
52 was under the stereoelectronic control by opening the quasi
equatorial epoxide bond. It was then oxidized successively with
OsO4-NMO57 and IBX55 to give R-ketolactone 53 in 68% yield
from 48. Reduction of 53 with NaBH4 furnished R-hydroxy-
lactone 50 quantitatively as a single product.58 The configuration
of the C-9 position was confirmed by observing the long-range
coupling between the H-9 and H-4a. The high stereoselectivity
should be due to the severe steric hindrance of the axial acetoxy
group at the C-5 position. The product 50 possesses a fully
to unsaturated ester (Scheme 8).46 Thus, the benzoate of 32 was
removed with diisobutylaluminum hydride (DIBAL-H), while
the primary alcohol was selectively oxidized to carboxylic acid
in two steps including 2,2,6,6,-tetramethyl-1-piperdinyloxy, free
radical (TEMPO) oxidation47 followed by sodium chlorite
oxidation.48 Methylation of the resulting carboxylic acid with
(trimethylsilyl)diazomethane (TMSCHN2)49 gave methyl ester
41 in good overall yield from 32. The remaining hydroxyl group
at the C-5 position was then transformed into carbamate 42 in
two steps.50 Intramolecular conjugate addition of 42 was best
carried out with potassium t-butoxide in THF at -78 to -15
°C to give cyclic carbamate 43 in 90% yield. High yield of the
reaction at low temperatures might be due to the dominant
(51) In addition, the stereoelectronic effect (antiperiplanar effect) of the BOM
protected hydroxyl group might operate to promote the conjugate addition.
(52) (a) Ishizuka, T.; Kunieda, T. Tetrahedron Lett. 1987, 28, 4185-4188. (b)
Flynn, D. L.; Zelle, R. E.; Grieco, P. A. J. Org. Chem. 1983, 48, 2424-
2426.
(53) (a) Albright, J. D.; Goldman, L. J. Am. Chem. Soc. 1965, 87, 4214-4216.
(b) Albright, J. D.; Goldman, L. J. Am. Chem. Soc. 1967, 89, 1994, 2416-
2423.
(54) The route through the intermediate bearing vinyl group at the C-8a position
prepared from 44 was fruitless.
(46) (a) Hirama, M.; Shigemoto, T.; Yamazaki, Y.; Ito, S. J. Am. Chem. Soc.
1985, 107, 1797-1798. (b) Hirama, M.; Ito, S. Heterocycles 1989, 28,
295-313.
(55) (a) Frigerio, M.; Santagostino, M.; Sputore, S.; Palmisano, G. J. Org. Chem.
1995, 60, 7272-7276. (b) Frigerio, M.; Santagostino, M. Tetrahedron Lett.
1994, 35, 8019-8022.
(47) Einhorn, J.; Einhorn, C.; Ratajczak, F.; Pierre, J.-L. J. Org. Chem. 1996,
61, 7452-7454.
(56) Preferential formation of Z-enolate 51 was supported by the following
experiments; aldehyde 48 was treated with TIPS-OTf and DBU to give
Z-alkenyl silyl ether, exclusively. The geometry was established by the
coupling constant (J ) 7.5 Hz) of vinyl protons.
(48) (a) Kraus, G. A.; Taschner, M. J. J. Org. Chem. 1980, 45, 1175-1176. (b)
Kraus, G. A.; Roth, B. J. Org. Chem. 1980, 45, 4825-4830. (c) Lindgren,
B. O.; Nilsson, T. Acta Chem. Scand. 1973, 27, 888-890.
(49) Hashimoto, N.; Aoyama, T.; Shioiri, T. Chem. Pharm. Bull. 1981, 29,
1475-1478.
(57) VanRheenen, V.; Cha, D. Y.; Hartley, W. M. Org. Synth. Coll. 1988, 6,
342-348.
(58) For a similar synthesis of R-hydroxyl lactone, see: Corey, E. J.; Ghosh,
A. K. Tetrahedron Lett. 1988, 26, 3205-3206.
(50) Kocovsky, P. Tetrahedron Lett. 1986, 27, 5521-5524.
9
J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003 8803

A R T I C L E S
Ohyabu et al.
Scheme 9. Construction of the Bicyclic Compounds
ylisothiourea63 in the presence of mercuric chloride64 to furnish
56 in good overall yield. Cleavage of the 1,2-diol with sodium
periodate provided aldehyde,65 which was further treated with
methanolic TFA to give cyclic guanidine 57 with a hemiaminal
structure.66 The remaining Boc group was further deprotected
with aqueous 4 M hydrochloric acid to give the desired cyclic
guanidine 58a.67 To protect the labile hemiaminal moiety,68 58a
was treated with 4 M HCl-dioxane in methanol to give methyl
aminal 58b, which was isolated as the peracetate 59 in 50%
overall yield from 57. The product 57 was the last intermediate
in the synthesis that could be readily purified in a preparative
scale by silica gel chromatography. Toward the goal, all the
acyl protective groups were removed with triethylamine in
aqueous methanol to furnish a 4:1 mixture of 4-methoxytetro-
dotoxin (60) and 4,9-anhydro-4-epitetrodotoxin (2)69 in 85%
Figure 5. Conformation of the enol precursor 48.
functionalized cyclohexane with the correct stereogenic centers
for tetrodotoxin.
Introduction of Guanidine and Completion of the Total
Synthesis
We now focused on a final stage for the cyclic guanidine
moiety, one of the unique structural features of tetrodotoxin.
According to the synthetic plan, the lactone of 50 was protected
as ortho ester with a hydroxyl group at the C-5 position (Scheme
10). Deacylation59 of 50 was followed by benzoylation60 and
subsequent acetylation to provide ortho ester 54. Prior to
guanidinylation with sterically crowded di-Boc-protected isothio-
urea, the steric congestion was diminished around the amino
group of 54 by transforming the BOM group to the correspond-
ing acetate in two steps involving hydrogenolysis and acetyla-
tion.61 Attempts to deprotect both the Boc and acetonide of 55
under conventional acidic conditions failed to give six-
membered cyclic carbamate which formed by participation of
the secondary hydroxyl group of the 1,2-diol to the carbonyl
group of Boc. Extensive experiments led us to find that two
step deprotection was necessary; thus, acid hydrolysis of the
acetonide of 55 was followed by removal of the Boc group with
ceric ammonium nitrate (CAN) in aqueous acetonitrile.62 The
resulting amine was directly treated with N,N′-diBoc-S-meth-
1
yield. The H and 13C NMR spectra of 60 were fully assigned
by HSQC,70 while 1H NMR spectra of 2 were in good agreement
with that of the natural product.71 As transformation of these
two tetrodotoxin derivatives to tetrodotoxin (1) under acidic
conditions were reported,5,6 a formal total synthesis of tetro-
dotoxin was completed at this stage. In our experiments,
4-methoxytetrodotoxin (60) was further treated with 2% TFA-d
in deuterium oxide to give a mixture of tetrodotoxin (1) and
4,9-anhydrotetrodotoxin (2) in 65% and 15% yield, respectively.
4,9-Anhydrotetrodotoxin (2) was hydrolyzed under the same
condition to furnish tetrodotoxin (1) in 63% yield along with a
(63) Bergeron, R. J.; McManis, J. S. J. Org. Chem. 1987, 52, 1700-1703.
(64) Kim, K. S.; Qian, L. Tetrahedron Lett. 1993, 34, 7677-7680.
(65) The aldehyde was stable as anticipated. No â-elimination was observed
under the conditions.
(66) The product 57 provided different 1H NMR spectra in each measurement,
presumably because 57 might exist as a mixture of guanidine and
guanidinium forms.
(67) Deprotection of all the acyl groups of 58a proved very difficult; for example,
aqueous ammonia in methanol caused complete decomposition, probably
because of extreme instability of the hemiaminal moiety under basic
conditions. For the stability of tetrodotoxin, see: Mosher, H. S. Ann. N.Y.
Acad. Sci. 1986, 479, 32-43.
(59) The reaction temperature below 15 °C was crucial to prevent from
epimerization at the C-9 position.
(60) Triethylamine was required for benzoylation of the ortho ester. In the
absence of the amine, only the primary alcohol at the C-11 position was
benzoylated.
(61) Protection of the hydroxyl group at the C-8 position prevented from
participation of the hydroxyl group to the proximate Boc group, leading to
cyclic carbamate.
(62) Hwu, J. R.; Jain, M.; Tsay, S.-C.; Hakimelahi, G. H. Tetrahedron Lett.
1996, 37, 2035-2038.
(68) Nishikawa, T.; Ohyabu, N.; Yamamoto, N.; Isobe, M. Tetrahedron 2001,
55, 4325-4340.
(69) Upon exposure of 60 to the conditions, 2 was not detected.
(70) 1H NMR chart (60 MHz) of 60 was reported in ref 4; however, it is difficult
to compare with that of the synthetic 60, because the literature did not
provide the values of the chemical shifts and coupling constants.
(71) Nakamura, M.; Yasumoto, T. Toxicon 1985, 23, 271-276.
9
8804 J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003

Asymmetric Total Synthesis of Tetrodotoxin
A R T I C L E S
Scheme 10. Introduction of Guanidine and the Total Synthesis
small amount of recovered 2. The mixture was purified with
an ion exchange column by eluting with 0.05 N AcOH to give
almost pure 1, which was in slow exquilibrium to be purified.
Synthetic tetrodotoxin was found to be identical with natural
tetrodotoxin by comparison of spectroscopic data (1H, 13C
NMR,72 and HR-FABMS), and specific rotation of the
at the construction of hydroxylated key cyclohexane derivatives
through Sonogashira coupling, Claisen rearrangement, dipole-
driven enolization and oxidation, and directed aldol condensa-
tion. Installation of the nitrogen atom was achieved through an
intramolecular conjugate addition. The cage structure was
constructed by the intramolecular Z-enol attack to the epoxide,
and stereoselective oxidation provided the protected ortho ester
structure. A guanidine group was introduced with retention of
the solubility, and finally, a safer deprotection route provided
tetrodotoxin.
synthetic 1 was [R]28 +1 (c 0.08, 0.05 N AcOH-H2O)(error
D
being ( 2), which was nearly zero but different from the value
[R]D -8.64 (dissolved 855 mg of TTX in 10 mL of dil AcOH)
reported in 1953.73 It may be because of a different ratio of the
components in the equilibrium mixture under different acid
concentrations; thus, no practical component change was
observed for a week at 0.05 N AcOH-H2O. Remeasurement
Acknowledgment. This work was financially supported by
a Grant-in-Aid for Scientific Research from the Ministry of
Education, Science, Sports, and Culture of Japan, PRESTO of
Japan Science and Technology Corporation (JST), a Pfizer’s
Award in Synthetic Organic Chemistry, the Naito Foundation
and the 21st Century COE grant. The authors thank JSPS for a
DC scholarship and the Suzuki scholarship to N.O. Elemental
analyses and measurements of 600 MHz NMR spectra were
performed by Mr. S. Kitamura and Mr. K. Koga (analytical
laboratory in this school), whom we thank. We also thank Prof.
F. Horio (in this school) for help in the bioassay of synthetic
tetrodotoxin.
of the natural 1 showed [R]28 -3.0 (c 0.31, 0.05 N AcOH)
D
after chromatographic separation from 2 with 15% of 3. The
rotation value of 1 is concluded to be nearly zero. Thus, we
have concluded the first asymmetric total synthesis of tetrodo-
toxin.
Conclusion
The first asymmetric total synthesis of tetrodotoxin (1) has
been accomplished via a new route from 2-acetoxy-tri-O-acetyl-
D-glucal as the chiral starting material. The synthesis first aimed
(72) For high-resolution NMR data of 1, see: Yasumoto, T.; Yotsu, M.; Murata,
M.; Naoki, H. J. Am. Chem. Soc. 1988, 110, 2344-2345.
(73) The [R]D value of natural 1 was reported to be -8.64 (c 8.55 in dilute
AcOH) from very old literature, which was published before the structure
elucidation (Tsuda, K.; Kawamura, M. Pharm. Bull. Japan 1953, 1, 112-
113). The opposite sign may be due to component ratio differences in the
equilibrium mixtures under different concentrations of 1 and acid, which
is not clearly reported. As a reference, the plus sign of [R]D was observed
with that of 11-deoxytetrodotoxin (2); thus, the natural 2 lit.19 [R]D +5.37
(c 0.34, 0.05 N AcOH-H2O); synthetic 2 lit.14 [R]D +13 (c 0.065, 0.05 N
AcOH-H2O). Synthetic 1 showed a fatal toxicity with a mouse (ip) test.
Supporting Information Available: Experimental procedures
and analytical and spectral characterization data for all com-
pounds (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
JA0342998
9
J. AM. CHEM. SOC. VOL. 125, NO. 29, 2003 8805