Total synthesis of polycavernoside A, a lethal toxin of the red alga Polycavernosa tsudai

Article abstract of DOI:10.1021/jo0503862

Two approaches to the synthesis of the aglycon 120 of polycavernoside A (1) were developed, only one of which was completed. The successful "second-generation" route assembled the aglycon seco acids 102 and 106 via Nozaki-Hiyama-Kishi coupling of aldehyde 70, prepared from methyl (S)-3-hydroxy-2-methylpropionate (72) and (S)-pantolactone (73), with vinyl bromide 71. The latter was obtained from a sequence which commenced from the silyl ether 24 of 3-hydroxypropionaldehyde and entailed cyclization of (Z)-ζ-hydroxy-α,β-unsaturated ester 82. Regioselective Yamaguchi lactonization of trihydroxycarboxylic acids 102 and 106 and subsequent functional-group adjustments led to macrolactone 120, to which the fucopyranosylxylopyranoside moiety was attached. Stille coupling of the glycosidated aglycon 128 with dienylstannane 129 furnished polycavernoside A in a synthesis for which the longest linear sequence was 25 steps. The overall yield to lactone 120 was 4.7%.

Full text of DOI:10.1021/jo0503862

Total Synthesis of Polycavernoside A, A Lethal Toxin of the Red
Alga Polycavernosa tsudai
Paul R. Blakemore, Cindy C. Browder, Jian Hong, Christopher M. Lincoln,
Pavel A. Nagornyy, Lonnie A. Robarge, Duncan J. Wardrop, and James D. White*
Department of Chemistry, Oregon State University, Corvallis, Oregon 97331-4003
james.white@oregonstate.edu
Received March 1, 2005
Two approaches to the synthesis of the aglycon 120 of polycavernoside A (1) were developed, only
one of which was completed. The successful “second-generation” route assembled the aglycon seco
acids 102 and 106 via Nozaki-Hiyama-Kishi coupling of aldehyde 70, prepared from methyl (S)-
3-hydroxy-2-methylpropionate (72) and (S)-pantolactone (73), with vinyl bromide 71. The latter
was obtained from a sequence which commenced from the silyl ether 24 of 3-hydroxypropionaldehyde
and entailed cyclization of (Z)-ú-hydroxy-R,â-unsaturated ester 82. Regioselective Yamaguchi
lactonization of trihydroxycarboxylic acids 102 and 106 and subsequent functional-group adjust-
ments led to macrolactone 120, to which the fucopyranosylxylopyranoside moiety was attached.
Stille coupling of the glycosidated aglycon 128 with dienylstannane 129 furnished polycavernoside
A in a synthesis for which the longest linear sequence was 25 steps. The overall yield to lactone
120 was 4.7%.
Introduction
episode.2 Skillful structural analysis of the very small
quantity of available polycavernoside A (1) by Yasu-
moto and co-workers resulted in determination of its
planar structure and enabled a tentative stereochem-
ical assignment to be made.2,3 A second collection of
P. tsudai from Tanguisson Beach by Yasumoto in 1992
produced a sufficient quantity of polycavernoside B and
three additional congeners to establish that these sub-
stances all shared the same aglycon with 1 and differed
only in the disaccharide moiety.4 However, subsequent
extractions of P. tsudai yielded no polycavernosides
at all, and it now appears that production of these
metabolites by the seaweed was a transient phenom-
enon.
In early 1991, 13 people were unexpectedly poisoned
after consuming an edible seaweed purchased from a food
market near Tanguisson Beach on the island of Guam.1
The seaweed responsible for this incident was identified
as the red alga Polycavernosa tsudai (formerly Gracilaria
edulis), which is widely consumed in the area. Curiously,
no previous reports of toxicity associated with this food
source had been recorded. The individuals poisoned
by this alga exhibited severe neuropathic symptoms,
and three of them died after suffering cardiac arrest.
Preliminary examination of the ingested alga revealed
the presence of a virulent toxin, and a collection of P.
tsudai made soon after the outbreak led to the isolation
of two substances, polycavernosides A and B, which
elicited very similar symptoms in a rodent bioassay to
those seen in the human victims of the Tanguisson Beach
(2) Yotsu-Yamashita, M.; Haddock, R. L.; Yasumoto, T. J. Am.
Chem. Soc. 1993, 115, 1147.
(3) Yotsu-Yamashita, M.; Yasumoto, T.; Haddock, R. L. In 34th
Symposium on the Chemistry of Natural Products; Tokyo: 1992; p 612.
(4) Yotsu-Yamashita, M.; Seki, T.; Paul, V. J.; Naoki, H.; Yasumoto,
T. Tetrahedron Lett. 1995, 36, 5563.
(1) Haddock, R. L.; Cruz, O. L. T. Lancet 1991, 338, 195.
10.1021/jo0503862 CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/14/2005
J. Org. Chem. 2005, 70, 5449-5460
5449

Blakemore et al.
SCHEME 1
With the natural source of polycavernosides having
vanished, toxicological studies of 1 and its congeners
remain incomplete, and the precise cause of the human
fatalities in Guam is uncertain. This situation presents
an irresistible invitation to synthesis, and the challenge
has been met by two elegant syntheses of polycavernoside
A, the first by Murai5 and the second by Paquette.6,7 Our
own approach to 1 differs significantly from both of these
syntheses and has been disclosed in preliminary form.8
We now describe details of our approach which led to the
successful asymmetric synthesis of polycavernoside A and
which confirmed its absolute configuration as represented
by 1.
The synthesis plan for polycavernoside A as initially
conceived is shown in Scheme 1, but in the event, this
blueprint served only to guide us toward the major
subunits, i.e., 2, 3, and 4, and left the critical issue of
uniting them to improvisation. Attachment of the disac-
charide appendage to the aglycon core of polycavernoside
A appeared to be the most straightforward connection
and was envisioned by glycosidation at C-5 with an
activated form of a suitably protected fucopyranosylxy-
lopyranoside 2. A late-stage elaboration of the all-trans
triene side chain was projected via Stille coupling using
a metallodiene 3, and the aglycon itself would be as-
sembled by macrolactonization of seco acid 4 in a process
which we hoped could be carried out with much of the
polycavernoside functionality already in place. A key
question with 4 was how the bond between C-9 and C-10
would be forged since this was the locus at which we
planned to connect northern and southern fragments of
the seco acid. Two methods for this bond construction
were examined, only one of which was successful.
on the aglycon of polycavernoside A became focused on
the enantiomeric version of the structure now known to
be correct (Scheme 2). The two subunits initially pro-
grammed for assembling seco acid 5 were tetrahydropy-
ran 6 and dithiane 7, the key C-9,10 bond formation being
performed via nucleophilic attack at the aldehyde of 6
by the R-thianyl anion of 7. The latter was foreseen as
the product of two primary subunits, sulfone 8 and
aldehyde 9.
The Southern Subunit. Our first plan for creating
the tetrahydropyran of 6 was based on a reaction, the
intramolecular Pd(II)-mediated alkoxycarbonylation of a
6-hydroxy alkene,9 which had resulted in a favorable
outcome for us in another context10 and for this we
needed to prepare a protected 4,6,8-trihydroxy-3-methyl-
octene such as 10. The starting point for this objective
was isobutyl acetoacetate (11). Alkylation of the dianion
of 11 with chloromethyl benzyl ether afforded 12, which
was hydrogenated with Noyori’s Ru(II)-(S)-BINAP cata-
lyst11,12 to give (S)-hydroxy ester 13 in excellent yield and
very high enantiomeric excess (Scheme 3).After silylation
Results
First-Generation Approach. At the outset of this
study, the absolute configuration of the aglycon of poly-
cavernoside A was not known and even the configuration
of the disaccharide portion was not settled with certainty.
An early attempt to correlate the configuration of the
sugar with that of the aglycon through an NMR experi-
ment was less than conclusive and forced us to make an
arbitrary selection of aglycon absolute configuration for
the purpose of synthesis. In consequence, our first attack
(5) Fujiwara, K.; Murai, A.; Yotsu-Yamashita, M.; Yasumoto, T. J.
Am. Chem. Soc. 1998, 120, 10770.
(6) Paquette, L. A.; Barriault, L.; Pissarnitski, D.; Johnston, J. N.
J. Am. Chem. Soc. 2000, 122, 619.
(9) Semmelhack, M. F.; Kim, C.; Zhang, N.; Bodurow, C.; Sanner,
M.; Dobler, W.; Meier, M. Pure Appl. Chem. 1990, 62, 2035.
(10) White, J. D.; Kranemann, C. L.; Kuntiyong, P. Org. Lett. 2001,
3, 4003.
(11) Noyori, R.; Ohkuma, T.; Kitamura, M.; Takaya, H.; Sayo, N.;
Kumobayashi, H.; Akutagawa, S. J. Am. Chem. Soc. 1987, 109, 5856.
(7) Paquette, L. A. Chemtracts 2002, 15, 345.
(8) White, J. D.; Blakemore, P. R.; Browder, C. C.; Hong, J.; Lincoln,
C. M.; Nagornyy, P. A.; Robarge, L. A.; Wardrop, D. J. J. Am. Chem.
Soc. 2001, 123, 8593.
5450 J. Org. Chem., Vol. 70, No. 14, 2005

Total Synthesis of Polycavernoside A
SCHEME 2
SCHEME 4
attempts to effect intramolecular alkoxycarbonylation of
18 were unsuccessful, and no tetrahydropyran resem-
bling 19 could be detected in the complex mixture of
products.
This result was puzzling in view of the fact that the
substituents in 18 were aligned in a manner which would
have resulted in an all-equatorial arrangement of the four
substituents in tetrahydropyran 19. Suspicious that there
was opposition to the cyclization of 18 due to the
configuration of its substituents, we prepared its dia-
stereomer 22 in which the substituents at C-3 and C-4
were reversed in configuration from 18 (Scheme 4). The
hydroxy alkene 22 obtained from 15 via 20 and 21
underwent smooth intramolecular alkoxy carbonylation
mediated by palladium acetate in methanol-acetonitrile
to afford tetrahydropyran 23, in which two of the tet-
rahydropyran substituents must be axial. Although the
contrasting results with 18 and 22 appear counterintui-
tive, a possible explanation for the failure of 18 to
undergo cyclization may lie in an unfavorable steric
interaction in the transition state between the initially
formed palladium-complexed alkene and the adjacent
methyl substituent.14 This rationale suggested that re-
locating the methyl group to a site more remote from the
alkene would remove the encumbrance associated with
alkoxy carbonylation of 18. To test this hypothesis, the
preparation of a new substrate which reversed the
direction of cyclization was carried out. This is shown in
Scheme 5.
SCHEME 3
The known aldehyde 2415 was reacted with Brown’s
(E)-crotyldiisopinocampheylborane,13 and the resulting
alcohol 25 was esterified with benzoyl chloride to give
anti benzoate 26 in high diastereomeric excess and
enantiomeric purity. After ozonolysis of 26, the aldehyde
27 was subjected to asymmetric allylation and the
homoallylic hydroxyl group of 28 was protected as its silyl
ether 29. Reductive cleavage of the benzoate from 29
afforded a hydroxy alkene 30, which we had hoped would
be a suitable substrate for alkoxy carbonylative cycliza-
tion. However, exposure of this substance to PdCl2, CO,
and MeOH gave a complex mixture in which the silyl
groups appeared to have migrated and/or were lost. An
attempted cyclization of the diol 31 from 30 in which both
of alcohol 13, the ester of 14 was reduced to aldehyde
15, which was subjected to Brown asymmetric crotyla-
tion13 to furnish homoallylic alcohol 16 as the predomi-
nant diastereomer. Protection of this alcohol as its
triisopropylsilyl (TIPS) ether 17 was followed by selective
removal of the tert-butyldimethylsilyl (TBS) ether to
produce the 1-octene derivative 18. Unfortunately, all
(12) Loubinoux, B.; Sinnes, J.-L.; O’Sullivan, A. C.; Winkler, T.
Tetrahedron 1995, 51, 3549.
(13) Brown, H. C.; Bhat, K. S. J. Am. Chem. Soc. 1986, 108, 5919.
(14) White, J. D.; Hong, J.; Robarge, L. A. Tetrahedron Lett. 1999,
40, 1463.
(15) Kozikowski, A. P.; Stein, P. D. J. Org. Chem. 1984, 49, 2301.
J. Org. Chem, Vol. 70, No. 14, 2005 5451

Blakemore et al.
SCHEME 5
SCHEME 7
(S)-3-propionyl-4-benzyloxazolidin-2-one (33)16 gave 34,
which was converted to Weinreb amide 35.17 Silylation
of the secondary alcohol yielded 36, and reduction of the
amide afforded aldehyde 37, which was subjected to
condensation with the Still-Gennari phosphonate 3818 to
give (Z)-R,â-unsaturated ester 39 in excellent yield.
Selective cleavage of the TBS ether from 39 furnished
cyclization substrate 40, and when this hydroxy ester was
exposed to potassium tert-butoxide in THF at low tem-
perature it underwent exceptionally clean cyclization to
produce tetrahydropyran 19 as the sole detectable ster-
eoisomer. Cyclization of the (E)-R,â-unsaturated ester
isomeric with 40 was much less satisfactory in terms of
yield and stereoselectivity, an observation in concert with
a result we have noted previously.19
A proof of structure of tetrahydropyran 19 was initially
sought through its conversion to alcohol 41, obtained by
hydride reduction of the ester appendage (Scheme 7).
However, neither 41 nor diol 42, prepared by hydro-
genolysis of the benzyl ether 41, afforded crystals suit-
able for X-ray analysis. Fortunately, when benzyl ether
19 was subjected to hydrogenolysis and alcohol 43
was oxidized,20 it produced the crystalline aldehyde 44.
X-ray crystallographic analysis fully confirmed the ste-
reostructure of 44, which now became the de facto
subunit representing the southern portion of polycaver-
noside A.
the benzoate and TBS group had been removed also failed
to yield tetrahydropyran 32.
The difficulty associated with preparing a tetrahydro-
pyran subunit corresponding to 6 bearing four correctly
placed substituents in the desired relative configuration
for polycavernoside A was exacerbated by the fact that,
even if 32 could be acquired, oxidation states of side
chains at C-3 and C-7 would need to be reversed in order
to reach 6. These shortcomings persuaded us to explore
a different strategy to the southern portion of polycav-
ernoside A, which relied upon a more conventional
cyclization of a ú-hydroxy-R,â-unsaturated ester for con-
structing the tetrahydropyran nucleus. This sequence is
shown in Scheme 6.
An Evans aldol coupling between the previously pre-
pared aldehyde 15 (Scheme 3) and the enol borinate of
The Northern Subunit. Our strategy for constructing
the northern segment of polycavernoside A aglycon
SCHEME 6
5452 J. Org. Chem., Vol. 70, No. 14, 2005

Total Synthesis of Polycavernoside A
SCHEME 8
SCHEME 10
SCHEME 9
SCHEME 11
obtained by reduction of 46, condensed with p-methoxy-
benzaldehyde dimethyl acetal to give the six-membered
cyclic acetal 53 exclusively.24 The latter underwent Swern
oxidation23 to aldehyde 54, and subsequent Wittig meth-
ylenation afforded alkene 55. Reductive cleavage of acetal
55 with diisobutylaluminum hydride occurred with com-
plete regioselectivity25 to deliver primary alcohol 56,
which was oxidized to aldehyde 9 under Swern condi-
tions.
Addition of the lithio anion from 8 to aldehyde 9
produced hydroxy sulfone 57 as a mixture of four dia-
stereomers in disappointingly low yield (Scheme 11), and
our intended advance from this intermediate, already
complicated by inefficiency, met with another obstacle
resulting from the fact we could find no way to oxidize
the sterically encumbered alcohol 57 to keto sulfone 58
without causing oxidation at sulfur. The problem was
overcome by returning to sulfone 49, the anion of which
coupled smoothly with 9 to give 59, again as a mixture
of all stereoisomers (Scheme 12). In this case, oxidation
of 59 to keto sulfone 60 was straightforward, as was
subsequent reductive cleavage of the sulfonyl group with
samarium diiodide,26 which yielded ketone 61. Cleavage
of the p-methoxybenzyl ether followed by directed reduc-
tion of the resulting â-hydroxy ketone 62 with tetra-
methylammonium triacetoxyborohydride27 furnished
antidiol 63, which was subsequently protected as its
acetonide 64. The silyl ether was cleaved from 64, and
hinged upon coupling of sulfone 8 with aldehyde 9 and
is shown in Scheme 2. An attractive feature of this plan
is that 8 can be envisioned as a “double-headed” nucleo-
phile, first as a sulfonyl anion for linkage with 9 to form
the C12-C13 bond and then as a dithianyl anion for
connection of the northern subunit 7 to aldehyde 6. A
direct route to 8 was foreseen from methyl (R)-3-hydroxy-
2-methylpropionate (45), and access to 9 appeared to be
feasible from (R)-(-)-pantolactone (46) (Scheme 8).
The synthesis of 8 commenced with protection of 45
as its tert-butyldiphenylsilyl (BPS) ether, after which the
ester was reduced with calcium borohydride21 to yield
alcohol 47 (Scheme 9). Replacement of the hydroxyl
function by a phenylthio group and then treatment of
sulfide 48 with oxone gave sulfone 49.22 Cleavage of the
silyl ether from 49 followed by Swern oxidation23 of
alcohol 50 afforded aldehyde 51, which was advanced to
dithiane 8 by reaction with 1,3-propanedithiol in the
presence of boron trifluoride etherate.
Aldehyde 9 was prepared in six steps from (-)-
pantolactone (46) as shown in Scheme 10. The triol 52,
(16) Gage, J. R.; Evans, D. A. In Organic Syntheses; John Wiley &
Sons: New York, 1993; Collect. Vol. 8, p 339.
(17) Levin, J. I.; Turos, E.; Weinreb, S. M. Synth. Commun. 1982,
12, 989.
(18) Still, W. C.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405.
(19) White, J. D.; Blakemore, P. R.; Green, N. J.; Hauser, E. B.;
Holoboski, M. A.; Keown, L. E.; Nylund Kolz, C. S.; Phillips, B. W. J.
Org. Chem. 2002, 67, 7750-7760.
(20) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277.
(21) Brown, H. C.; Narasimha, S.; Choi, Y. M. J. Org. Chem. 1982,
47, 4702.
(24) Lavall_e, P.; Ruel, R.; Grenier, L.; Bissonnette, M. Tetrahedron
Lett. 1986, 27, 679.
(25) Nakatsuka, M.; Ragan, J. A.; Sammakia, T.; Smith, D. B.;
Vehling, D. E.; Schreiber, S. L. J. Am. Chem. Soc. 1990, 112, 5583.
(26) Molander, G. A.; Hahn, G. J. Org. Chem. 1986, 51, 1135.
(27) Evans, D. A.; Chapman, K. T.; Carreira, E. M. J. Am. Chem.
Soc. 1988, 110, 3560.
(22) Walker, K. A. M. Tetrahedron Lett. 1977, 4475.
(23) Mancuso, A. J.; Huang, S.-L.; Swern, D. J. Org. Chem. 1978,
43, 2480.
J. Org. Chem, Vol. 70, No. 14, 2005 5453

Blakemore et al.
SCHEME 12
SCHEME 13
the disaccharide derived from L-fucose and D-xylose
1
matched well in a comparison of its H NMR spectrum
with the corresponding spectral region of the natural
material.29 Enantiomers of a truncated C1-C9 segment
of polycavernoside A aglycon30 were then separately
glycosidated with L-fucosyl-D-xylopyranoside, and of the
two diastereomeric products, the NMR spectrum of one
was found to closely resemble the spectrum of the
glycosidated macrolide.31 Finally, the spatial relationship
between the tetrahydrofuran moiety of the northern
portion and the tetrahydropyran segment in the southern
subunit was established by a nOe experiment, which
linked the methine proton at C11 with one of the
diastereotopic protons at C8. These results led Murai to
propose the absolute configuration shown in 1 for poly-
cavernoside A, a postulate he later proved correct by
synthesis.5
Revision of our original plan for the synthesis of
polycavernoside A now required not only a new tactic for
uniting the northern and southern portions of the aglycon
but also demanded construction of these two segments
in their antipodal form. The latter stipulation could be
accommodated without difficulty since we had adopted
an approach from the outset in which starting materials
and stereocontrolling reagents were available in both
enantiomeric forms. A new plan also gave us an op-
portunity to solve the north-south unification problem
the resulting primary alcohol 65 was oxidized to aldehyde
66. The latter was reacted with the bistrimethylsilyl
thioether of 1,3-propanedithiol in the presence of zinc
iodide to afford dithiane 67.28
Here, unfortunately, our first approach to polycaver-
noside A met its demise; for although dithiane 67 could
be deprotonated, the lithio derivative proved to be
surprisingly unreactive. In particular, no coupling prod-
uct 68 could be coaxed from a mixture containing the
anion of 67 and aldehyde 44. A speculative rationale for
the unreactive nature of the lithio anion of 67 is that an
exceptionally stable six-membered chelate is formed
between the metal and the proximal ketal oxygen atom
in this structure. In any event, we were now confronted
with an impasse, which necessitated a major revision of
our original strategy for connecting the northern and
southern portions of polycavernoside A.
Second-Generation Approach. In 1995, the uncer-
tainty surrounding the absolute configuration of poly-
cavernoside A began to clear, thanks to seminal studies
by Murai and co-workers. First, it was shown that only
(29) Fujiwara, K.; Amano, S.; Murai, A. Chem. Lett. 1995, 191.
(30) Fujiwara, K.; Amano, S.; Oka, T.; Murai, A. Chem. Lett. 1994,
2147.
(28) Evans, D. A.; Truesdale, L. K.; Grimm, K. G.; Nesbitt, S. L. J.
Am. Chem. Soc. 1977, 99, 5009.
(31) Fujiwara, K.; Amano, S.; Murai, A. Chem. Lett. 1995, 855.
5454 J. Org. Chem., Vol. 70, No. 14, 2005

Total Synthesis of Polycavernoside A
SCHEME 14
that had defeated the initial design outlined in Scheme
2. Specifically, we decided to reverse the roles of electro-
phile and nucleophile in the connection of the northern
and southern subunits of the putative aglycon precursor
69. As outlined in Scheme 13, our second generation
strategy takes aldehyde 70, the enantiomer of previously
synthesized 66, as the northern partner for coupling with
bromo alkene 71 in a putative Nozaki-Hiyama-Kishi
(NHK) reaction.32,33
Aldehyde 70 was obtained from methyl (S)-3-hydroxy-
2-methylpropionate (72) and (S)-(+)-pantolactone (73)
exactly as shown in Schemes 9, 10, and 12 but in the
antipodal sequence. For the tetrahydropyran 71, how-
ever, we were able to devise a more efficient route to its
direct precursor, the enantiomer of 44, than that used
previously and shown in Schemes 3, 6, and 7. In
particular, we wanted to avoid steps 11 f 12 f 13 at
the beginning of the sequence which we found difficult
to scale up, and we therefore elected to commence from
aldehyde 24.
Silylation of 78 gave 79, which was reduced to aldehyde
80, and olefination of the latter with the phosphonate
3818 furnished R,â-unsaturated ester 81. Cleavage of both
TBS ethers of 81 was accomplished as before, leaving the
TIPS ether intact, and exposure of the resulting diol 82
to potassium tert-butoxide at -45 °C in THF furnished
tetrahydropyran 83 in 73% yield for the two steps from
81. Alternatively, 83 could be obtained directly from 81
in 70% yield by adding solid potassium carbonate to the
reaction vessel following acid-catalyzed desilylation. This
one-pot process obviated oligomerization of hydroxy ester
83, a problem which sometimes plagued the potassium
tert-butoxide mediated cyclization. Oxidation of the
primary alcohol of 83 and reaction of the derived alde-
hyde 84 with the Ohira-Bestmann reagent34 in the
presence of base yielded terminal alkyne 85, which was
treated with B-bromo-9-borabicyclo[3.3.1]nonane35 to
furnish 71 in an overall yield of 13% from 24.
Assembly of Subunits. The critical union of bro-
moalkene 71 and aldehyde 70 proceeded smoothly under
NHK conditions32,33 if a large excess of CrCl2 and a
substoichiometric quantity of NiCl2 was employed (Scheme
15). The resulting 1:1 mixture of C-10 allylic alcohols 86
was oxidized to a single ketone 87, which we had
expected to yield a diol upon acid-catalyzed methanolysis
of the acetonide. Instead, the internal ketal 88 was
Asymmetric allylation of 24 afforded homoallylic alco-
hol 74, which was first silylated to give 75 and then
ozonized to produce aldehyde 76 (Scheme 14). A sequence
similar to that used previously (Scheme 6) led to N-
acyloxazolidinone 77 and then to Weinreb amide 78.
(32) Jin, H.; Uenishi, J.-I.; Christ, W. J.; Kishi, Y. J. Am. Chem.
Soc. 1986, 108, 5644.
(33) Takai, K.; Tagashira, M.; Kuroda, T.; Oshima, K.; Utimoto, K.;
Nozaki, H. J. Am. Chem. Soc. 1986, 108, 6048.
(34) Ohira, S. Synth. Commun. 1989, 19, 561.
(35) Hara, S.; Dojo, H.; Takinami, S.; Suzuki, A. Tetrahedron Lett.
1983, 24, 731.
J. Org. Chem, Vol. 70, No. 14, 2005 5455

Blakemore et al.
SCHEME 15
SCHEME 16
produced, leading us into a cul-de-sac from which we were
unable to find an exit.
and treatment of 90 with benzaldehyde yielded the
expected hydroxy ester 91. Benzoate 91 was advanced
to a prospective coupling partner for 71 by protection of
the secondary alcohol as its methoxymethyl (MOM) ether
92 followed by cleavage of the silyl ether to yield primary
alcohol 93. However, oxidation of this alcohol was com-
plicated by loss of the MOM group and subsequent
oxidation of the liberated secondary alcohol. A second
option with 91 was protection of the secondary alcohol
as its p-methoxybenzyl ether 94, and this was followed
by cleavage of the primary silyl ether to yield 95.
Our failure to unmask acetonide 87 in a manner which
would release the C15 hydroxyl function for future
macrolactonization persuaded us to return to a precursor
of the northern segment which would permit differenti-
ated protection of the C-13 and C-15 hydroxyl groups.
Our starting point for this purpose was ketone 89, the
enantiomer of 61, which was obtained by a sequence
antipodal to that described in Scheme 12 from the
enantiomers of 9 and 49. Cleavage of the p-methoxyben-
zyl ether from 89 gave â-hydroxy ketone 90, which now
became the focus of studies designed to furnish a selec-
tively protected 1,3-anti diol (Scheme 16). The Evans-
Tishchenko reaction36 seemed well suited to this purpose,
Swern oxidation23 of 95 gave aldehyde 96 in excellent
yield, but this aldehyde performed poorly in the NHK
reaction with 71, affording no more than 30% of the
TABLE 1. Calculated Energies (∆E) of Prelactonization Complexes and Results of Lactonization of 102 and 106 under
Yamaguchi Conditions
5456 J. Org. Chem., Vol. 70, No. 14, 2005

Total Synthesis of Polycavernoside A
SCHEME 17
compliant substrate for NHK coupling and to that end
we first removed the silyl group from 97 with a fluoride
source expecting to obtain a primary alcohol (Scheme 17).
Instead, diol 98 bearing two secondary alcohols was
produced, evidently the result of a base-catalyzed acetate
migration driven by the sterically hindered environment
of the secondary neopentyl-type ester. Confirmation that
this rearrangement had occurred was obtained when 98
was converted to bis-silyl ether 99, then selectively
deprotected to allylic alcohol 100, which yielded vinyl
ketone 101 upon oxidation.
At this juncture, the more obvious methods for dif-
ferentiating the hydroxyl groups at C-13 and C-15
appeared to have been exhausted, leaving only the easiest
but riskiest option of completely eschewing the use of
protecting groups in the pivotal macrolactonization step.
The risk was compounded in the present case by the fact
that, if hydroxyl groups at C-13 and C-15 were left
unprotected, the formation of 88 from 87 had already
demonstrated that functionality at C-10 must remain at
the hydroxyl oxidation level in order to avoid formation
of an internal ketal. In this fully exposed scenario, there
would therefore be three hydroxyl groups in the seco acid
with which the carboxyl could potentially form a lactone.
It seemed improbable that the three contending lactones
would be equally favored, but hoping to put intuition on
a firmer footing, a computational study was carried out
coupled product. The source of the problem was traced
to the allylic benzoate of 96, which appeared to have been
reduced by excess Cr(II) present in the NHK medium.
We hoped that hydroxy acetate 97, prepared by an
Evans-Tishchenko reaction36 of 90 with acetaldehyde in
the presence of samarium diiodide, would be a more
SCHEME 18
J. Org. Chem, Vol. 70, No. 14, 2005 5457

Blakemore et al.
SCHEME 19
in order to obtain an estimate of the energetics associated
with each of the three lactonization pathways.
With this encouraging portent, the separated epimers
of 86 were each advanced to the corresponding seco acids
102 and 106, as shown in Scheme 18, and each of these
acids was subjected to lactonization with 2,4,6-trichlo-
robenzoyl chloride and DMAP in toluene at high dilution.
The lactonization yields are shown in Table 1. The (10S)-
trihydroxy acid 102 gave a crystalline macrolactone in
good yield whose structure was established by X-ray
analysis as 103. The crystal structure of 103 shows good
correspondence with the conformation of the energy-
minimized prelactonization complex, suggesting rela-
tively little atomic displacement is required to close 102
to lactone 103. No other lactones could be detected in the
reaction mixture. Lactonization of the (10R) alcohol 106
was less clean, the 16-membered lactone 107 being
formed in modest yield accompanied by a significant
quantity of the 14-membered lactone 108. Again, none
of the 11-membered lactone was found in the mixture.
With lactone 103 in hand, the steps to polycavernoside
A aglycon now appeared straightforward, and to this end,
103 was oxidized with barium manganate39 to give
hydroxy ketone 110. We had expected to see this product
in the form of a cyclic hemiketal, as it appears in
polycavernoside A itself, but conjugation apparently
stabilizes the ketone sufficiently to prevent cyclization.
Since we intended to cleave both alkenes of 110 simul-
taneously to generate an R-diketo aldehyde in which we
believed the formation of a cyclic five-membered hemi-
ketal with the proximal ketone would now be favored, a
reductive ozonolysis of this diene was carried out. An
unexpected result confounded us again, and only an
unstable hemiacetal believed to be 111 could be isolated
from this reaction. This diversion foiled our plan for
elaborating functionality at C16 of 110 for attachment
of the metallodiene 3, and once more we were forced to
revise our end-game strategy.
Computational Results and Macrolactonization
of a Trihydroxy Carboxylic Acid. A consequence of
86 being obtained as a 1:1 mixture of C10 epimeric
alcohols from the NHK coupling of 70 and 71 was that
both stereoisomers would need to be considered in a
macrolactonization event even though the configuration
at C10 becomes inconsequential when this center evolves
as the hemiketal carbon of 1. In practice, the two epimers
of 86 were easily separable by chromatography, so that
a modeling exercise with each stereoisomer became
meaningful. The model chosen for this purpose was the
acyl DMAP adduct considered to be the active species in
the Yamaguchi lactonization protocol,37 and three prelac-
tonization conformations were examined corresponding
to ring closure at C-10, C-13, and C-15 hydroxyl groups.
These conformations assumed that attack by each of the
three hydroxyl groups on the acyl-DMAP complex would
occur along a Bu¨rgi-Dunitz trajectory.38 The relative
energies of the three conformations were calculated at
the PM3 level, and the results are shown in Table 1 for
both C-10 epimers of trihydroxy acid 86. On the assump-
tion that the relative energies of the prelactonization
conformations provide an approximate guide to the value
of ∆∆G† between the three lactonization pathways avail-
able to 102 and 106, our calculation suggested there
should be a preference for the formation of a 16-
membered lactone, i.e., 103 from 102 and 107 from 106,
over 14-membered lactones 104 and 108. Although
formation of 11-membered lactone 105 was not expected
and, indeed, was not observed, lactone 109 appeared to
be a viable candidate from lactonization of seco acid 106.
However, we found no evidence for the formation of 109.
(36) Evans, D. A.; Hoveyda, A. H. J. Am. Chem. Soc. 1990, 112, 6447.
(37) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M.
Bull. Chem. Soc. Jpn. 1979, 52, 1989.
(38) Bu¨rgi, H. B.; Dunitz, J. D.; Lehn, J. M.; Wipff, G. Tetrahedron
1974, 30, 1563.
(39) Firouzabadi, H.; Ghaderi, E. Tetrahedron Lett. 1978, 839.
5458 J. Org. Chem., Vol. 70, No. 14, 2005

Total Synthesis of Polycavernoside A
SCHEME 20
SCHEME 21
The problem was solved by returning to diol 103,
converting both hydroxyl groups to TES ethers, and
carrying out ozonolysis on the protected diene 112. Keto
aldehyde 114 was produced in excellent yield by this
process, enabling a Takai reaction40 with iodoform and
chromous chloride to transform the aldehyde function
into trans iodoalkene 116. Selective silyl ether cleavage
of 116 yielded R-hydroxy ketone 118, which was oxidized
to an R-diketone with Dess-Martin periodinane.20 Ex-
posure of this diketone to fluoride ion cleaved both of the
remaining silyl ethers, and the liberated C13 alcohol then
spontaneously hemiketalized with the C10 ketone to
produce 120. The same six-step sequence from 107 also
led to 120 via 113, 115, 117, and 119 in a similar 45%
overall yield, thus enabling the synthesis to make use of
both stereoisomers of 86 produced in the NHK coupling.
It happens that 120 was a key compound in both the
Murai5 and Paquette6 routes to polycavernoside A, and
we were able to show through comparison of IR and NMR
spectral data that our 120 was identical to their lactone.
TABLE 2. Comparison of Synthetic Routes to 120
22
Our optical rotation, [R]D -36 (c 0.09, CHCl3), also
Murai Paquette this work
matched well with the value reported by Murai,5 [R]D
22
no. steps to C1-C9 subunit
no. steps to C10-C16 subunit
longest linear sequence (steps)
overall yield (%)
21
21
32
16
18
29
14
14
22
-38 (c 0.12, CHCl3), for 120. This correspondence,
together with the fact that the structure of 103 had been
secured by X-ray crystallographic analysis, lent assur-
ance to Murai’s assignment of absolute configuration to
the aglycon which had relied on nOe data.
2.6
1.5
4.7
(Scheme 21). The benzyl ether of 127 was cleaved with
DDQ, and the iodo alkene 128 was subjected to Stille
coupling with the known dienylstannane 1296 to furnish
polycavernoside A (1). Confirmation that our synthesized
material was identical with natural polycavernoside A
was obtained by comparison of their proton NMR spectra.
In summary, we have completed a total synthesis of
the algal toxin polycavernoside A (1) via the aglycon pre-
cursor 120. The latter was obtained using a convergent
pathway in which the longest linear sequence was 22
steps, and the overall yield was 4.7%. This compares
favorably with other routes to this compound (see Table
2).
The final stages of our synthesis of 1 from 120 adhered
closely to the pathway laid down by Murai5 and Paquette6
and employed the known protected fucopyranosylxylopy-
ranoside 121,30 prepared as shown in Scheme 20, for
glycosidation of 120. Thus, the C4 hydroxyl group of the
fucopyranose derivative 122 was benzylated, and benzyl
ether 123 was converted to an anomeric mixture of
fluorides 124 as carried out by Murai.31 This mixture was
coupled to the known xylopyranoside 12541 to give
disaccharide 126 as the major anomer (R:â 4:1). Saponi-
fication of the two acetates of 126 and subsequent
O-methylation was followed by chromatographic purifica-
tion to furnish Murai’s R anomer 121 in good yield.
Glycosidation of 120 with 121 was carried out with
NBS in acetonitrile and gave 127, albeit in low yield
Acknowledgment. We are grateful to Professor Alex
Yokochi, Department of Chemical Engineering, Oregon
State University, for the X-ray crystal structures of 44
and 103. Financial support was provided by the Na-
tional Institute of General Medical Sciences (GM50574).
(40) Takai, K.; Nitta, K.; Utimoto, K. J. Am. Chem. Soc. 1986, 108,
7408.
(41) Johnston, J. N.; Paquette, L. A. Tetrahedron Lett. 1995, 36,
4341.
J. Org. Chem, Vol. 70, No. 14, 2005 5459

Blakemore et al.
Supporting Information Available: General experimen-
tal techniques, experimental procedures and characterization
data for new compounds, crystallographic data and ORTEP
plots for 44 and 103, molecular modeling data for lactonization
of 102 and 106, and copies of 1H and 13C NMR spectra of new
purified compounds. This material is available free of charge
via the Internet at http://pubs.acs.org.
JO0503862
5460 J. Org. Chem., Vol. 70, No. 14, 2005