Enantioselective synthesis of (-)-terpestacin and structural revision of siccanol using catalytic stereoselective fragment couplings and macrocyclizations

Article abstract of DOI:10.1021/ja0470968

(-)-Terpestacin (1, naturally occurring enantiomer) and (+)-11-epi-terpestacin (2) were prepared using catalyst-controlled, stereoselective, intermolecular reductive coupling reactions of alkyne 9 and aldehyde 10, affording allylic alcohols 42 or 11-epi-42 in a 3:1 ratio (or 1:3 depending on the enantiomer of ligand 41a used). These stereoselective fragment couplings were instrumental in confirming that "siccanol" is not 11-epi-terpestacin but, in fact, is (-)-terpestacin itself. Several intramolecular alkyne-aldehyde reductive coupling approaches to 1 and 2 were also investigated and are discussed herein.

Full text of DOI:10.1021/ja0470968

Published on Web 08/03/2004
Enantioselective Synthesis of (-)-Terpestacin and Structural
Revision of Siccanol Using Catalytic Stereoselective
Fragment Couplings and Macrocyclizations
Johann Chan and Timothy F. Jamison*
Contribution from the Massachusetts Institute of Technology, Department of Chemistry,
Cambridge, Massachusetts 02139
Received May 18, 2004; E-mail: tfj@mit.edu
Abstract: (-)-Terpestacin (1, naturally occurring enantiomer) and (+)-11-epi-terpestacin (2) were prepared
using catalyst-controlled, stereoselective, intermolecular reductive coupling reactions of alkyne 9 and
aldehyde 10, affording allylic alcohols 42 or 11-epi-42 in a 3:1 ratio (or 1:3 depending on the enantiomer
of ligand 41a used). These stereoselective fragment couplings were instrumental in confirming that “siccanol”
is not 11-epi-terpestacin but, in fact, is (-)-terpestacin itself. Several intramolecular alkyne-aldehyde
reductive coupling approaches to 1 and 2 were also investigated and are discussed herein.
Introduction
Over the past 10 years, several scientific communities have
been drawn to the sesterterpene terpestacin (1), (Figure 1) and
structurally related compounds (2, 3) because of their unique
structural architectures and biological activities. Originally
isolated from the fungal strain Arthrinium sp. and studied
collaboratively by Oka and Bristol-Myers Squibb, terpestacin
was found to inhibit the formation of syncytia, giant-multi-
nucleated cells that arise from expression of gp120 on cell
surfaces in the course of HIV infection.1 Recently, a study of
the effects of terpestacin on bovine aortic endothelial cells
(BAECs) and chorioallantoic membrane (CAM) from growing
chick embryos has determined that this natural product also
inhibits angiogenesis.2
Figure 1. Terpestacin (1) and related molecules.
Scheme 1. Previous Enantioselective Syntheses of Terpestacin
(1) and Fusaproliferin (3)
A structurally related compound, proliferin, was discovered
shortly after terpestacin and was renamed “fusaproliferin” (3)
since a prolactin-related protein had already been named
“proliferin”.3 In a subsequent report, the absolute configuration
of fusaproliferin was initially assigned as the enantiomer of 23-
epi-3 (terpestacin numbering) based on statistical differences
in R and Rw X-ray data between two enantiomorphs.4
Both terpestacin and fusaproliferin have been prepared by
total synthesis (Scheme 1),5 and Tatsuta was the first to prepare
the racemate of the former in 1998. Later that same year, his
group completed an enantiospecific, 38-step synthesis beginning
(1) Oka, M.; Iimura, S.; Tenmyo, O.; Sawada, Y.; Sugawara, M.; Ohkusa, N.;
Yamamoto, H.; Kawano, K.; Hu, S.-L.; Fukagawa, Y.; Oki, T. J. Antibiotics
1993, 46, 367-373.
(2) Jung, H. J.; Lee, H. B.; Kim, C. J.; Rho, J.-R.; Shin, J.; Kwon, H. J. J.
Antibiotics 2003, 56, 492-496.
(3) Lee, S. J.; Nathans, D. J. Biol. Chem. 1988, 263 (7), 3521-3527.
(4) (a) Randazzo, G.; Fogliano, V.; Ritieni, A.; Mannina, L.; Rossi, E.; Scarallo,
A.; Segre, A. L. Tetrahedron 1993, 49, 10883-10896. (b) Manetti, C.;
Fogliano, V.; Ritieni, A.; Santini, A.; Randazzo, G.; Logrieco, A.; Mannina,
L.; Segre, A. L. Struct. Chem. 1995, 6, 183-189. (c) Santini, A.; Ritieni,
A.; Fogliano, V.; Randazzo, G.; Mannina, L.; Logrieco, A.; Benedetti, E.
J. Nat. Prod. 1996, 59, 109-112.
(5) Approaches toward the synthesis of terpestacin: (a) Takeda, K.; Nakajima,
A.; Yoshii, E. Synlett 1995, 249-250. (b) Mermet-Mouttet, M.-P.; Gabriel,
K.; Heissler, D. Tetrahedron Lett. 1999, 40, 843-846.
with tri-O-acetyl-D-galactal 4 that took advantage of a highly
selective Horner-Wadsworth-Emmons reaction of 6 to solve
a central structural challenge presented by these compounds,
the 15-membered carbocycle. The specific rotation measured
for synthetic terpestacin ([R]D ) +27, c 0.22, CHCl3) was
consistent with that obtained by Oka ([R]D ) +26, c 0.5,
9
10682
J. AM. CHEM. SOC. 2004, 126, 10682-10691
10.1021/ja0470968 CCC: $27.50 © 2004 American Chemical Society

Enantioselective Synthesis of (−)-Terpestacin
A R T I C L E S
Scheme 2. Structural Assignment Timeline of 1, 2, and 3
CHCl3), and so it appeared that the absolute configuration of
terpestacin had been confirmed by means of chemical synthesis.6
Nearly coincident with the Myers’ report of the syntheses of
terpestacin and fusaproliferin, Miyagawa reported the isolation
of another natural product, siccanol (2) from Bipolaris soro-
kiniana, a fungal strain commonly found in decayed ryegrass
leaves. Their structural determination led them to conclude that
siccanol was diastereomeric to terpestacin at the allylic carbinol
in the 15-membered ring (C11), i.e., 11-epi-terpestacin.9 Dif-
fering only by the configuration about the C11 allylic carbinol,
1 and 2 provided an attractive context to pursue our ongoing
interest in the stereoselective formation of allylic alcohols.10
Herein, we provide a detailed account of our syntheses of (-)-
terpestacin and (+)-11-epi-terpestacin that utilizes our recently
developed intermolecular, stereoselective methods of nickel-
catalyzed reductive coupling of alkynes and aldehydes. One
outcome of these studies is the discovery that “siccanol” is not
11-epi-terpestacin but, in fact, is (-)-terpestacin itself.
Retrosynthetic Analysis. A central component of our ap-
proach to 1 and 2 was a disconnection at the C11-C12 allylic
alcohol, such that we could examine the suitability of catalytic
reductive coupling reactions of alkynes and aldehydes, related
to methodology developed in our laboratory and by Montgom-
ery.11 Two contrasting approaches using this method were
devised (Scheme 3). The first utilized an intramolecular
reductive coupling of an alkyne and aldehyde to install the C11
carbinol stereocenter, construct the 15-membered carbocyclic
ring, and establish the (E)-geometry at one of the three alkenes
of this macrocycle. An alternate approach planned on controlling
the C11 configuration via an intermolecular alkyne-aldehyde
reductive coupling of 9 and 10, with subsequent formation of
However, this conclusion was brought into question as a result
of an unusual chain of events (Scheme 2). In 2001, Gra¨fe and
co-workers isolated a molecule from Ulocladium sp. that was
believed to be the enantiomer of terpestacin,7 since it was
spectroscopically identical to Oka’s material, differing only in
the sign of its specific rotation. Interestingly, both “(+)-
terpestacin” (Oka) and “(-)-terpestacin” (Gra¨fe) inhibited
syncytium formation. In March 2002, Myers reported enantio-
selective syntheses (Scheme 1) of terpestacin (1) and fusapro-
liferin (3) that began with an amide derived from pseudoephe-
drine (5) and should have lead to (+)-terpestacin, as reported
by Oka and corroborated by Tatsuta. However, having unex-
pectedly obtained (-)-terpestacin at the end of the synthesis,
Myers initiated a thorough series of investigations that ultimately
revealed that exposure of (-)-terpestacin to chloroform stored
over potassium carbonate gave rise to a chloroetherification
product possessing a dextrorotatory sense of specific rotation,
the same as that obtained previously for both natural (Oka) and
synthetic (Tatsuta) “(+)-terpestacin”.8 Therefore, it is likely that
the same enantiomer of terpestacin had been isolated by Oka
and Gra¨fe but that Oka’s material underwent a chloroetherifi-
cation in the CHCl3 (stored over K2CO3) used to obtain the
specific rotation. In other words, only (-)-terpestacin has been
isolated to date from natural sources. By acetylation of (-)-1,
Myers obtained (-)-3, confirming that naturally occurring
terpestacin (1) and fusaproliferin (3) formed a homochiral
structural series. In other words, natural fusaproliferin was
simply an acetate ester of natural terpestacin, not the enantio-
meric C23 epimer, as was originally reported by Santini.
(9) Nihashi, Y.; Lim, C.-H.; Tanaka, C.; Miyagawa, H.; Ueno, T. Biosci.
Biotechnol. Biochem. 2002, 66, 685-688.
(10) (a) Huang, W.-S.; Chan, J.; Jamison, T. F. Org. Lett. 2000, 2, 4221-4223.
(b) Chan, J.; Jamison, T. F. J. Am. Chem. Soc. 2003, 125, 11514-11515.
(c) Colby, E. A.; Jamison, T. F. J. Org. Chem. 2003, 68, 156-166. (d)
Miller, K. M.; Huang, W.-S.; Jamison, T. F. J. Am. Chem. Soc. 2003, 125,
3442-3443.
(6) Tatsuta, K.; Masuda, N. J. Antibiotics 1998, 51, 602-606.
(7) Schlegel, B.; Schmidtke, M.; Do¨rfelt, H.; Kleinwa¨chter, P.; Gra¨fe, U. J.
Basic Microbiol. 2001, 41, 179-183.
(8) Myers, A. G.; Siu, M.; Ren, F. J. Am. Chem. Soc. 2002, 124, 4230-4232.
(11) Montgomery, J. Acc. Chem. Res. 2000, 33, 467-473.
9
J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004 10683

A R T I C L E S
Chan and Jamison
Scheme 3. Nickel-Catalyzed Alkyne-Aldehyde Coupling
Strategies
in order to investigate the feasibility of this macrocylization
strategy (Scheme 4). Although the model lacked several features,
Scheme 4
the macrocycle using an intramolecular allylation of a ketone
enolate.
namely, the cis-fused 5,5-ring system and its corresponding
functional groups, a 1.5:1 ratio was nevertheless obtained,
favoring the formation of the desired regioisomer (13), dem-
onstrating that forming a 15-membered macrocycle with three
(E)-trisubstituted double bonds was possible using this approach.
In both approaches, the synthesis of the reductive coupling
precursors required the protection of two otherwise interfering
functional groups, a latent 1,2-diketone located at (C16-C17)
and the primary hydroxyl group at C24. We envisioned
accomplishing both tasks by, in effect, lowering the oxidation
state of C17 and attaching to it the C24 oxygen, thus forming
a tetrahydrofuran that we planned to unravel at a late stage in
the synthesis through an enolate hydroxylation reaction.
The tetrahydrofuran also addressed challenges common to
both strategies, namely, establishing the remaining three ste-
reogenic centers (Figure 2). A conjugate addition to oxabicyclo-
Results and Discussion
Intramolecular Alkyne-Aldehyde Reductive Coupling
Approach. Beginning with commercially available â-methallyl
alcohol, mole-scale rhodium-catalyzed hydroformylation, fol-
lowed by dehydration under acidic conditions, afforded multi-
hundred-gram quantities of racemic dihydrofuran 15 in 72%
yield13 that was resolved to >95% ee at 55% conversion using
(+)-(Ipc)2BH (Scheme 5).14 Initial attempts to promote the
Scheme 5
Figure 2. Proposed basis of stereoselectivity in conjugate addition-
alkylation sequences of 11a and 11b.
[3.3.0]octenone 11a or 11b would relay the configuration of
C23 to both the quaternary carbon center (C1) and its neighbor
(C15) that together also comprise the junction of the five- and
fifteen-membered rings. The first event was expected to occur
on the convex face, but prediction of the major diastereomer in
the subsequent alkylation was less clear (desired approach
shown; E ) MeI), since a substituent on this face, adjacent to
the site of alkylation, might strongly influence the stereochem-
ical course of the reaction.
Also unclear in both approaches a priori was the degree of
influence of the existing stereocenters in determining the
selectivity in the formation of the carbinol center at C11 using
the nickel-catalyzed alkyne-aldehyde reductive coupling dis-
cussed above. In both cases, the nearest stereocenter would
reside at C15, two bonds away from the site of asymmetric
induction. An additional uncertainty particular to the intramo-
lecular reductive coupling approach was the effect of the
conformation of the nascent 15-membered ring not only on
diastereoselectivity but also on regioselectivity (15-membered
ring vs 14-membered ring).
intermolecular Pauson-Khand reaction between 15 and cobalt
cluster 16 involved the use of NMO,15 cyclohexylamine,16 and
(12) Model substrate 12 was synthesized through the DCC coupling of 40a and
pent-3-ynoic acid.
(13) Botteghi, C.; Consiglio, G.; Ceccarelli, G.; Stefani, A. J. Org. Chem. 1972,
37, 1835-1837.
Accordingly, a model substrate (12) was synthesized12 and
subjected to Et3B and catalytic amounts of Ni(cod)2 and PBu3
(14) Brown, H. C.; Prasad, J. V. N. Vara J. Am. Chem. Soc. 1986, 108, 2049-
2054.
9
10684 J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004

Enantioselective Synthesis of (−)-Terpestacin
A R T I C L E S
a phosphine sulfide17 or simply heating a solution of these two
compounds. The greatest success was achieved using a sulfide-
promoted intermolecular Pauson-Khand reaction,18 furnishing
the desired oxabicyclo[3.3.0]octenone (11a) in 40-60% yield.
Notably, no other diastereomers, or any of the three other
possible regioisomers, could be detected (1H NMR).19
In 1990, Haruta and co-workers demonstrated that allenylt-
riphenylstannanes in TiCl4-mediated conjugate additions to a
variety of cyclic enones, afforded 1,5-ynones.20 Unfortunately,
our attempts to install the 2-butynyl moiety at C15 (terpestacin
numbering) using this method with the corresponding allenyl
stannane 17 were unsuccessful. In our studies we observed that
enone 11a underwent either decomposition or no reaction,
despite evaluation of several Lewis acids of varying nature and
strength (TiCl4, BF3‚OEt2, SnCl4, Et2AlCl, MgBr2, Me3SiCl,
ZnCl2, Ti(OiPr)4, Ti(OiPr)3Cl, and Yb(OTf)3) (Scheme 6).
addition of 100 mol % water to a mixture of sodium hydride,
the ketone, and methyl iodide in toluene afforded 21 in 90%
yield with a diastereoselectivity of 93:7 in favor of the desired
isomer. The assignment of the newly formed quaternary center
was based on an NOE experiment and the observation that the
1H NMR resonance corresponding to the quaternary methyl
group in the major product resided upfield relative to that in
the undesired diastereomer (δ 0.92 ppm vs δ 1.06 ppm), likely
due to magnetic anisotropy imparted by the triple bond. Selective
cleavage of the terminal isopropylidene unit was effected in a
two-step sequence that involved a catalytic dihydroxylation
using Sharpless’ (DHQD)2PHAL ligand23 and sodium periodate
cleavage of the resulting diol, affording 8 in 25% yield for the
two steps.
Intramolecular, nickel-catalyzed reductive cyclization of 8
afforded, unfortunately, the undesired 14-membered ring
regioisomer (22) in 45% yield with no detectable trace of the
desired 15-membered ring (Scheme 7). Formation of the
undesired regioisomer was surprising, given our results with
the model system 12. The structural differences between 8 and
12, as highlighted by the ovals in Figure 3, were thought to be
Scheme 6
Consequently, an alternate route was devised (Scheme 7). A
highly diastereoselective conjugate addition of a lithium cuprate
(19) to 11a provided 20 in 72% yield and >95:5 diastereo-
selectivity. Notably, attempts at the use of a trimethylsilyl group
in place of triisopropylsilyl on the alkyne gave rise to lower
yields, possibly due to unimolecular or bimolecular decomposi-
tion of the nucleophile.21 Following reduction, deprotection,
isomerization of the terminal alkyne with KOtBu in DMSO,22
and a TPAP/NMO oxidation, 18 was delivered in a four-step
sequence in 57% yield overall. The reduction/oxidation tandem
(steps 1 and 4) was necessary, since ketones such as 18 were
not compatible with KOtBu in DMSO solutions.
Our studies of the methylation of an enolate derived from
ketone 18 included a survey of a variety of bases and conditions.
Nearly all conditions resulted in either decomposition of the
starting material or the formation of an O-methyl vinyl ether
as the primary product. Exclusively successful was the use of
sodium hydride in benzene, effecting a reaction that proceeded
with good conversion and site selectivity on a small scale (<10
mg). However, upon increasing the scale of the reaction (100
mg), slow decomposition of 10 was observed with no conversion
to the desired product. We reasoned that adventitious water
might be present in the small scale reaction, implying that finely
dispersed sodium hydroxide was the operative base. Indeed, the
Figure 3. Comparison of 8 to a cyclization model compound (12); (E)-
ester conformer shown for clarity.
responsible for the divergent behavior in the macrocyclizations.
Specifically, the critical structural features that may have
affected the regioselectivities in the cyclization were the 5,15-
ring junction and the stereogenic centers in 8, which were not
present in 12. Since changing the C15 stereocenter would require
major modifications of the existing route, we first focused on
altering the nature of C1 through removal of the C19 quaternary
methyl group in order to remove steric interactions with the
2-butynyl group. By this approach, more conformations of the
alkyne might be accessible, possibly affecting regioselectivity
in the macrocyclization.
Intramolecular Alkyne-Aldehyde Approach: Cyclization
in the Absence of C19. To test these hypotheses, alkynal 23
was synthesized in an analogous fashion to the synthesis of 8.
Treatment with Ni(cod)2, PBu3, and BEt3 in toluene afforded
the undesired regioisomer (25) as well as 24, a product derived
from the intramolecular reductive coupling of the alkyne and,
surprisingly, the ketone (Scheme 8). Acetone can be used as a
solvent in intermolecular nickel-catalyzed reductive couplings,
and alkyne-acetone coupling products have never been de-
tected, suggesting that the proximity of the alkyne to the ketone
may explain the formation of 24.
(15) Shambayati, S.; Crowe, W. E.; Schreiber, S. L. Tetrahedron Lett. 1990,
31, 5289-5292.
(16) Sugihara, T.; Yamada, M.; Ban, H.; Yamaguchi, M.; Kaneko, C. Angew.
Chem., Int. Ed. Engl. 1997, 36, 2801-2804.
(17) Hayashi, M.; Hashimoto, Y.; Yamamoto, Y.; Usuki, J.; Saigo, K. Angew.
Chem., Int. Ed. 2000, 39, 631.
(18) Sugihara, T.; Yamada, M.; Yamguchi, M.; Nishizawa, M. Synlett 1999, 6,
771-773.
The results obtained from these experiments suggested that
the ketone might participate in the catalytic reaction, possibly
through interaction with nickel. Accordingly, the next course
of action was to alter this functional group. Reduction of the
(19) Kerr, W. J.; McLaughlin, M.; Pauson, P. L.; Robertson, S. M. J. Organomet.
Chem. 2001, 630, 104-117.
(20) Haruta, J.; Nishi, K.; Matsuda, S.; Akai, S.; Tamura, Y.; Kita, Y. J. Org.
Chem. 1990, 55, 4853-4859.
(21) A 1H NMR spectrum of the unpurified reaction mixture displayed a large
number of signals between 5.0 and 6.0 ppm when the trimethyl silyl
protecting group was used in place of triisopropylsilyl.
(22) (a) Takano, S.; Sekiguchi, Y.; Sato, N.; Ogasawara, K. Synthesis 1987,
139-141. (b) Recent application in total synthesis: Thompson, C. F.;
Jamison, T. F.; Jacobsen, E. N. J. Am. Chem. Soc. 2000, 122, 10482-
10483.
(23) (a) Crispino, G. A.; Sharpless, K. B. Tetrahedron Lett. 1992, 33, 4273-
4274. (b) Crispino, G. A.; Ho, P. T.; Sharpless, K. B. Science 1993, 259,
64-66. (c) Corey, E. J.; Noe, M. C.; Lin, S. Tetrahedron Lett. 1995, 36,
8741-8744. (d) Corey, E. J.; Zhang, J. Org. Lett. 2001, 3, 3211-3214.
9
J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004 10685

A R T I C L E S
Chan and Jamison
Scheme 7
ketone with NaBH4 and protection as its tert-butyldimethylsilyl
ether accomplished this task, but unfortunately, cyclization of
group.10 Therefore, we targeted alkynylsilane 35 in hopes of
observing the same regioselectivity in the catalytic macro-
cyclization similar to that seen in intermolecular cases.
An approach to making use of a KAPA-mediated isomer-
ization25 (KAPA ) potassium 3-aminopropylamide) was un-
dertaken (Scheme 10). Beginning with 28, a 1,4-addition of
Scheme 8
Scheme 10
26 also afforded exclusively the 14-membered ring (27) in 71%
yield (Scheme 9).
Scheme 9
1-lithio-1-propyne-AlMe3 “ate” complex26 afforded 29. Fol-
lowing deprotection with TBAF, alkylation of 30 afforded 31
with greater than 95:5 diastereoselectivity, presumably due to
the low steric demand of the propynyl group residing on the
convex face of the bicyclic system.27 Reduction by NaBH4
furnished the desired product 29. Unfortunately, attempts to
utilize KAPA-mediated isomerizations of either 29 or 32,
followed by treatment with trimethylsilyl chloride, did not afford
the desired alkynylsilanes. Fortunately, a remarkably functional
group-tolerant alkyne cross metathesis28 of 33, using a catalyst
Intramolecular Alkyne-Aldehyde Reductive Coupling
Approach: Cyclization of an Alkynylsilane. Since changing
nearby functional groups had no effect upon the cyclization
regioselectivity, we next investigated the effects of altering the
nature of the alkyne itself. In general, catalytic additions to
internal acetylenes of the type RCH2-CâC-Me, i.e., with
substitutents nearly identical in electronic nature and steric
demand, are typically nonregioselective.24 In contrast, acetylenes
of the type R-CâC-SiMe3 couple with complete regioselec-
tivity, favoring C-C bond formation adjacent to the SiMe3
(25) (a) Brown, C. A.; Yamashita, A. J. Am. Chem. Soc. 1975, 97, 891-892.
(b) Midland, M. M.; Halterman, R. L. Tetrahedron Lett. 1981, 22, 4171-
4172.
(24) Notable exceptions: (a) Et vs Me (2:1): Larock, R. C.; Yum, E. K. J. Am.
Chem. Soc. 1991, 113, 6689-6690. (b) i-Bu vs Me (7.3:1): Molander, G.
A.; Retsch, W. H. Organometallics 1995, 14, 4570-4575. (c) n-undecyl
vs Me (2.4:1): Trost, B. M.; Ball, Z. T. J. Am. Chem. Soc. 2001, 123,
12726-12728.
(26) Kim, S.; Park, J. K. Synlett 1995, 163-164.
(27) We have observed that the diastereoselectivity of quaternary methylation
reaction increases as the size of the â-substitutent decreases.
9
10686 J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004

Enantioselective Synthesis of (−)-Terpestacin
A R T I C L E S
Scheme 11
developed by Cummins,29 afforded the desired product (34) in
36% yield with 39% as recovered 33. While removal of the
acetonide and periodate cleavage of the resulting diol to 35 were
straightforward, attempts at nickel-catalyzed reductive cycliza-
tion, even under forcing conditions (refluxing toluene), afforded
only an aldol self-condensation product (36) and recovered
starting material (Scheme 11).
an oxametallacycle pathway, whereas regioisomeric intermedi-
ates B/D may form as a result of hydrometalation across the
alkyne to form two possible alkenyl nickel intermediates.
This observation can be attributed to both the steric and
electronic nature of the alkyne. We have observed in our
investigations of alkyne scope in intermolecular couplings
that alkynylsilanes generally exhibited reduced reactivity relative
to internal acetylenes. In general, as the steric demand of the
alkyne increases, as in the case of RCH2-≡-Me versus
t-Bu-≡-Me, reactivity decreases, and the electronic nature of
the trimethylsilyl group likely reduces reactivity as well. For
instance, (3-methyl-but-1-ynyl)benzene (Ph-â-i-Pr) has been
found to couple readily, whereas sterically similar phenyl-
ethynyltrimethylsilane (Ph-â-SiMe3) exhibits markedly lower
reactivity with various aldehydes.
From the previous cyclization experiments, it appears that
the unusual regioselectivities observed may be rationalized
through mechanistic considerations.30 A recent report by
Montgomery31 suggests that the Ni(cod)2/PBu3 catalyst system
can proceed through either an oxametallacyclopentene or
through an alkenyl nickel intermediate (Figure 4). Taking into
account these two contrasting mechanisms, four different
transition states leading to the two macrocycles would be
possible, as illustrated by intermediates A-D. Intermediates A/B
lead to the 14-membered ring, while the C/D pair lead to the
desired 15-membered ring. Intermediates A/C proceed through
Figure 4. Putative intermediates in the formation of 14-membered and
15-membered rings.
Our experiments would suggest that putative intermediates
C/D leading to the 15-membered ring were highly disfavored
relative to A/B. This bias toward A/B was originally postulated
to have been caused by either the steric congestion of the C19
quaternary methyl group and/or the trans relationship of the C1/
C15 stereocenters at the 5,15-ring junction. However, with no
detectable formation of the 15-membered ring upon cyclization
in the absence of C19, it appears more likely that the high barrier
to C/D formation may be linked to the latter, which was not a
factor included in model substrate 12. If the cyclization proceeds
via an oxametallacyclepentene (A vs C), the bicyclo[12.2.1]
system in intermediate C, possessing a bridgehead olefin, may
not form as readily as the bicyclo[13.3.0] found in A. Likewise,
under the hydrometalation pathway (B vs D), incorporation of
(28) (a) Fu¨rstner, A.; Mathes, C.; Lehmann, C. W. J. Am. Chem. Soc. 1999,
121, 9453-9454. (b) Fu¨rstner, A.; Mathes, C. Org. Lett. 2001, 3, 221-
223.
(29) (a) Laplaza, C. E.; Cummins, C. C. Science 1995, 268, 861-863. (b)
Laplaza, C. E.; Odom, A. L.; Davis, W. M.; Cummins, C. C. J. Am. Chem.
Soc. 1995, 117, 4999-5000. (c) Laplaza, C. E.; Johnson, A. R.; Cummins,
C. C. J. Am. Chem. Soc. 1996, 118, 709-710. (d) Laplaza, C. E.; Johnson,
M. J. A.; Peters, J. C.; Odom, A. L.; Kim, E.; Cummins, C. C.; George, G.
N.; Pickering, I. J. J. Am. Chem. Soc. 1996, 118, 8623-8638.
(30) Oblinger, E.; Montgomery, J. J. Am. Chem. Soc. 1997, 119, 9065-9066.
(31) Mahandru, G. M.; Liu, G.; Montgomery, J. J. Am. Chem. Soc. 2004, 126,
3698-3699.
9
J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004 10687

A R T I C L E S
Chan and Jamison
an extra (E)-double bond in D may be disfavored over B
(exocyclic olefin) upon macrocycle formation. Finally, the added
strain created by the trans relationship of the C1/C15 stereo-
centers may also bias the reaction to exclusive 14-membered
ring formation. Unfortunately, this hypothesis could not be tested
directly, since attempts to synthesize substrates with a cis
relationship of the C1/C15 stereocenters led to epimerization
to the thermodynamic trans product under all conditions
attempted.
Intermolecular Alkyne-Aldehyde Reductive Coupling
Approach. Since the intramolecular alkyne-aldehyde coupling
approach consistently afforded 14-membered ring products, we
aimed to overcome this regioselectivity problem through the
assembly of the allylic alcohol by way of an intermolecular
reductive coupling, using methods related to those we had
developed in our laboratory. The synthesis of the alkyne frag-
ment began with an NMO-promoted, intermolecular Pauson-
Khand reaction between dihydrofuran 15 and the hexacarbon-
yldicobalt complex of trimethylsilylacetylene (37) to afford
oxabicyclo[3.3.0]octenone 11b in 51% yield (Scheme 12). As
In the intramolecular reductive coupling approach described
above, we had hoped for diastereocontrol of the C11 carbinol
through a conformational bias during ring formation and/or
ligand control. The absence of such a conformational bias in
the intermolecular fragment coupling, however, predicted that
diastereocontrol would be low, since the closest stereocenter
would be two carbons removed from the site of reaction. In
addition, little regiocontrol would also be expected as the
catalytic process would have to differentiate between a methyl
group vs an isobutyl-like group on the alkyne. Not surprisingly,
the use of PBu3 offered neither diastereoselectivity (1:1) nor
regioselectivity (1.5:1).
We were pleased, however, to discover that our recently
developed P-chiral ferrocenyl phosphine ligands10 afforded a
good level of control at the C11 stereocenter (Table 1). After
evaluating a variety of these ligands, we found that a catalyst
incorporating (R)-41e favored the diastereomer (2:1) and re-
gioisomer (2.6:1) corresponding to (-)-terpestacin in a com-
bined yield of 85% (Table 1, 42a-d). Therefore, the desired
allylic alcohol 42a was obtained in an overall yield of 41%
2
(85% × /3 × 2.6/3.6). The diastereomer corresponding to 11-
Scheme 12
epi-terpestacin (42b) (terpestacin numbering) was obtained with
equal regioselectivity and equal and opposite diastereoselectivity
simply by using (S)-41e, the enantiomer of the ligand used in
the terpestacin-series coupling. Notably, the use of (R)-41a
provided the best diastereoselectivity (3:1) of all ligands
examined. However, (R)-41a also gave reduced regiocontrol
(2:1) and chemical yield (70%), corresponding to an overall
3
2
35% (70% × /4 × /3) yield of desired 42a.
The choice of protective group on the newly formed allylic
alcohol would prove crucial into the completion of the synthesis
(Scheme 14). The main factors considered were balancing
Scheme 14
observed in the Pauson-Khand reaction involving the farnesyl
derived alkyne hexacobaltdicarbonyl complex (16, Scheme 5),
no other diastereomers nor any of three other possible regio-
isomers could be detected (1H NMR). Conjugate addition of a
lithium dialkyl cuprate (19) afforded 38 which occurred with
complete diastereoselectivity, as observed before. To place the
triple bond in the proper position required for the catalytic
reductive coupling, terminal acetylene 39 was isomerized with
KOt-Bu in DMSO. Including protection of the secondary alcohol
as a trimethylsilyl ether, 9 was provided in 77% yield over two
steps.
Synthesis of the aldehyde coupling partner (10) commenced
with diol 40, obtained from site selective catalytic dihydroxyla-
tion of farnesyl acetate.21 The acetate ester was cleaved quanti-
tatively under basic conditions, and a sodium periodate cleavage
of the unpurified triol afforded an aldehyde that was protected
as a TBS ether (10) in 52% overall yield (Scheme 13).
Scheme 13
functional group compatibility with ease of removal following
the late stage alkylation (installation of C19). While a pivaloyl
ester of the secondary allylic alcohol initially appeared to be
an attractive choice because of its ease of introduction and
9
10688 J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004

Enantioselective Synthesis of (−)-Terpestacin
A R T I C L E S
Table 1. Evaluation of Phosphine Ligands in Catalytic, Stereoselective, and Regioselective Fragment Coupling (Alkyne 9 + Aldehyde 10)
temperature
yielda
(%)
diastereoselectivityc
(C11)
entry
ligand
solvent
toluene
(°C)
regioselectivityb
1
2
Bu3P
41f
23
23
23
23
0
0
0
0
0
0
0
0
0
70
68
77d
81
70
18
46
35
65
65
75
83
85
52
57
1.5:1
1:1.5
n/a
2:1
2:1
2:1
1.8:1
2:1
2:1
2:1
1:1
2:1
EtOAc
3
4
5
6
7
8
9
10
11
12
13
14
15
41f
EtOAc/DMI (1:1)
EtOAc
EtOAc
EtOAc/DMI (1:1)
THF
toluene
acetone
EtOAc
EtOAc
EtOAc
41a
41a
41a
41a
41a
41a
41b
41c
41d
41e
41e
41d
2.5:1
3:1
2:1
3:1
3:1
3:1
2:1
2:1
2:1
2:1
3:1
3:1
2:1
2.5:1
2.6:1
2.9:1
3:1
EtOAc
EtOAc
EtOAc
-12
-12
a Combined yield of all allylic alcohol products (42a-d). b Regioselectivity (42a + 42b/42c + 42d) determined by 1H NMR. c Diastereoselectivity (42a
1
+ 42c/42b + 42d) estimated by H NMR. d “Alkylative coupling” (Et at C13 instead of H).
Scheme 15
removal, we found that this protective group was not compatible
under the conditions required for the installation of the
quaternary methyl group.32
Therefore, the triisopropylsilyl ethers of 42a-d were pre-
pared, and the trimethylsilyl group was removed using 5%
sodium hydroxide in methanol, giving 43a-d in 80% yield over
the two steps. At this stage, the regioisomers formed in the
fragment coupling reaction were separated, and the desired
regioisomers (43a and 44b) were oxidized under Ley’s condi-
tions to afford the desired ketones in 89% yield. After removal
of the primary TBS groups (1 M HCl/THF 1:1) in 84% yield,
44a and 44b were converted to the allylic iodides and treated
with LiHMDS to afford a separable mixture of macrocycles
45a and 45b via a ketone enolate alkylation in a combined, an
overall two-step yield of 32% (22% overall of 45a).
As we had observed in the intramolecular reductive coupling
approach, installation of the critical quaternary methyl group
(C19) at C1 was best accomplished using methyl iodide,
(presumably) sodium hydroxide generated from sodium hydride
(300%), and H2O (200 mol % relative to 45a) giving the desired
(46) in > 95:5 dr (NOE), with no trace of methylation at the
5,5-ring junction (Scheme 15).
Completion of the synthesis of (-)-terpestacin thus required
alcohol (47). Treatment of 47 with 300 mol % of potassium
three further transformations. The TIPS protective group was
hexamethyldisilazane and P(OEt)3 under an atmosphere of O2
smoothly cleaved with TBAF to afford the desired secondary
effected the enolate hydroxylation,33 and opening of the
hemiketal was best effected with potassium carbonate in
methanol, furnishing (-)-terpestacin (1). Synthesis of 11-epi-
(32) Attempted alkylation of the pivaloyl-ester resulted in the isolation of multiple
unidentifiable products.
9
J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004 10689

A R T I C L E S
Chan and Jamison
terpestacin (2) was prepared utilizing the analogous method.
Overall, preparation of 1 and 2 each required 17 steps from 15
(longest linear sequence).34
Comparison of 1 and 2 to Natural Samples of Terpestacin
and Siccanol. The spectroscopic data we obtained for our
synthetic (-)-terpestacin were identical in all respects to those
previously reported for natural and synthetic material, and
comparison of (-)-terpestacin differed from our synthetic 11-
epi-terpestacin, particularly in the chemical shifts of the protons
3, 13, 15, and 19, as illustrated in Table 2. Remarkably, the
did not agree with our data for synthetic 11-epi-terpestacin and,
to our surprise, was indistinguishable from (-)-terpestacin,
confirming a structural reassignment: “siccanol” is (-)-ter-
pestacin, not 11-epi-terpestacin (Table 2).
The sequence of events that led Miyagawa to his original
assignment of “siccanol” as 11-epi-terpestacin can be explained
by his original structural elucidation studies and an unusual chain
of events, beginning with Oka’s discovery of terpestacin nearly
10 years before that. As Oka had done with natural terpestacin,
Miyagawa’s structural determination of “siccanol” began with
differentiating the protons on C14 (Ha and Hb) based on their
coupling constants with H15 (Scheme 16). An NOE between
Table 2. Selected 1H NMR Data for Terpestacin (1),
11-epi-Terpestacin (2), and Siccanol
Scheme 16. Structural Elucidation of Terpestacin (Oka) and
Siccanol (Miyagawa)
terpestacin
11-epi-terpestacin
siccanol
carbon
(Oka, Myers, Jamison)
(Jamison)
(Miyagawa)
2
3
1.68-1.80, 2.40
5.25
2.05-2.27
5.34
1.75, 2.36
5.25
5
6
7
1.90-2.04, 2.22-2.30
1.98, 2.05-2.27
2.01, 2.24
2.11, 2.26
5.13
2.09-2.12, 2.22-2.30
2.05-2.27
5.14
5.13
9
1.68-1.80, 2.09-2.12
1.70-1.88, 2.05-2.27
1.78, 2.18
1.70, 1.75
4.07
5.38
1.92, 2.44
2.72
0.99
2.66
3.80, 3.85
1.29
10
11
13
14
15
19
23
24
25
1.68-1.80
1.70-1.88
4.06
5.41
1.90-2.04, 2.45
2.72
1.01
2.68
3.83, 3.90
1.29
4.06
5.50
1.70-1.88, 2.50
2.57
1.13
2.7
3.83, 3.89
1.30
Ha and H19 indicated a trans relationship between H15 and H19.
Following acid-catalyzed elimination, methylation of the enol
with diazomethane and reduction to afford 48, an NOE between
H15 and H25 established the relatiVe relationships of stereo-
centers C1, C15, and C23. The absolute configuration of C18
was determined through the chiral exciton method after forma-
tion of the benzoyl ester of 48, and from the presence of a nOe
between H18 and H19, the absolute configurations of the
remaining stereocenters were deduced, except for C11. A
Mosher ester analysis led to the assignment that the C11
stereocenter was (R), consistent with 11-epi-terpestacin, whereas
Oka assigned the C11 stereocenter as (S), also based on a
Mosher analysis. Oka then proceeded further to obtain an X-ray
crystal structure of terpestacin to confirm all relative stereo-
centers, but Miyagawa did not characterize “siccanol” crystal-
lographically.
chemical shift of the proton at C11 was identical for 11-epi-
terpestacin and (-)-terpestacin, while the biggest difference was
observed for C19 (δ 0.99 vs δ 1.13), initially suggesting to us
that the C19 diastereomer might have been obtained in the
alkylation of ketone 45b. To eliminate this possibility, separate
NMR experiments conducted with 46 and 11-epi-46 (Figure 5)
Figure 5. Both 46 and 11-epi-46 exhibit NOE between H17 and H19.
showed an NOE between H17 and H19 in both cases, consistent
with the conclusion that both compounds have the same relative
configurations about C19.
Having established that 11-epi-47 leads to 11-epi-terpestacin,
a comparison of siccanol provided to us from Prof. Miyagawa
Given that “siccanol” was identical in every respect to natural
terpestacin, including the absolute configurations of C1, C15,
and C23, but possessed a specific rotation of opposite sign,
it was certainly reasonable for Miyagawa to conclude that
they had isolated the diastereomer of terpestacin at C11. Since
Myers’ total syntheses of terpestacin and fusaproliferin were
published at about the same time as Miyagawa’s report of
“siccanol”, it is likely that neither group would be aware of
each other’s findings through the chemical literature. Conse-
(33) (a) Gardner, J. N.; Carlon, F. E.; Gnoj, O. J. Org. Chem. 1968, 33, 3294-
3297. (b) Harwig, W.; Born, L. J. Org. Chem. 1987, 52, 4352-4358. (c)
Belletire, J. L.; Fry, D. F. J. Org. Chem. 1988, 53, 4724-4279.
(34) epi-terpestacin (1b) was prepared by the same sequence from 11-epi-36.
The same NOE shown for 37 was observed in 11-epi-37.
9
10690 J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004

Enantioselective Synthesis of (−)-Terpestacin
A R T I C L E S
quently, Miyagawa would not have known that the specific
rotation of terpestacin originally reported by Oka was an artifact.
Thus the only remaining difference between Oka’s determi-
nation of terpestacin and Miyagawa’s of “siccanol” is the result
of a Mosher ester analysis for the assignment of the stereo-
chemistry of C11. No experimental details are provided in
Miyagawa’s paper, but the following excerpt suggests a plausible
explanation for this discrepancy: “...a set of (R)-MTPA and
(S)-MTPA esters (at C11) was prepared ...(from)... the respectiVe
MTPA chlorides.” (Italics added for emphasis.) The Cahn-
Ingold-Prelog convention for assigning absolute configuration
indicates that the (R)-MTPA chloride leads to the (S)-MTPA
ester. Therefore, if by “respective” Miyagawa meant that the
(R)-acid chloride was used to (incorrectly) prepare the (R)-
MTPA ester, then the assignment of the stereochemistry of C11
would be the opposite of the actual configuration. Professor
Miyagawa, in addition to graciously providing us with a copy
of the original notebook pages describing his Mosher ester
determination, has corroborated this hypothesis.35
In short, the combination of the originally reported (arti-
factual) specific rotation of natural terpestacin and a Mosher
ester analysis led to the original assignment of “siccanol” as
11-epi-terpestacin. However, our syntheses of both diastereomers
by way of catalyst-controlled, stereoselective fragment couplings
of late-stage alkyne and aldehyde intermediates indicate that
11-epi-terpestacin (2) has yet to be isolated from natural sources.
That is, the natural product “siccanol” is not diastereomeric to
naturally occurring terpestacin ((-)-1) but, in fact, is (-)-
terpestacin itself.
Acknowledgment. We are grateful to Professor Hisashi
Miyagawa, Kyoto University, both for a generous supply of
natural siccanol, as well as a copy of the primary data obtained
in his Mosher ester determination of the configuration of C11
in this natural product. We thank Ryan K. Zeidan (Pfizer SURF)
and Jianlong Tan for experimental assistance, as well as Fran
Stephens and Professor Christopher C. Cummins for a generous
donation of the tris(amido)molybdenum (III) complex used in
the alkyne cross metathesis study. This work was supported by
the National Institute of General Medical Sciences (GM-063755)
and by Bristol Myers-Squibb and Boehringer-Ingelheim (re-
search assistantships to J.C.). We also thank the NSF (CAREER
CHE-0134704), Amgen, GlaxoSmithKline, Johnson & Johnson,
Merck Research Laboratories, the Sloan Foundation, 3M, and
MIT for generous financial support. The NSF (CHE-9809061
and DBI-9729592) and NIH (1S10RR13886-01) provide partial
support for the MIT Department of Chemistry Instrumentation
Facility.
Supporting Information Available: Experimental procedures
and data for 1, 2, 8-16, 18, 20-35, 38-40, and 42-47. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(35) Miyagawa, H. Personal communication.
JA0470968
9
J. AM. CHEM. SOC. VOL. 126, NO. 34, 2004 10691