A Cycloaddition Cascade Approach to the Total Synthesis of (-)-FR182877

Article abstract of DOI:10.1021/ja037643+

An asymmetric synthesis of the cytotoxic natural product, (-)-FR182877 (1), has been achieved. Chirality for the entire structure was established using two (4R)-4-benzyl-2-oxazolidinone-mediated boron aldol reactions. A 19-membered macrocarbocycle was syn

Full text of DOI:10.1021/ja037643+

Published on Web 10/14/2003
A Cycloaddition Cascade Approach to the Total Synthesis of
(-)-FR182877
David A. Evans* and Jeremy T. Starr
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138
Received July 30, 2003; E-mail: evans@chemistry.harvard.edu
Abstract: An asymmetric synthesis of the cytotoxic natural product, (-)-FR182877 (1), has been achieved.
Chirality for the entire structure was established using two (4R)-4-benzyl-2-oxazolidinone-mediated boron
aldol reactions. A 19-membered macrocarbocycle was synthesized by the coupling of two fragments using
a regioselective Suzuki coupling (17 + 23 f 26; 84%) and macrocyclization of a â-keto ester (30 f 31;
77%). Oxidation of 31 triggered a sequence of stereoselective transannular Diels-Alder reactions (32 f
34; 63%) forming four new rings and seven new stereocenters in the pivotal construction event. This
pentacyclic intermediate was subsequently transformed to (-)-FR182877. Semiempirical calculations of
the transannular Diels-Alder cycloaddition cascade were carried out to determine the origins of asymmetric
induction.
Introduction
Background. In 1998, Sato and co-workers reported a new
cytotoxic natural product (WS9885B), isolated from Strepto-
myces sp.# 9885.1a WS9885B was later renamed FR182877,
and the absolute configuration of this substance was reported
to be enantiomeric to the structure illustrated for 1.1b-d This
error was corrected in a subsequent publication, and 1 represents
what is now the accepted absolute configuration.1e FR182877
possesses a hexacyclic architecture containing 12 stereogenic
centers and a vinylogous carbonate embedded in a fused 6-6-7
ring system.
A closely related natural product, hexacyclinic acid (2), also
isolated from Streptomyces, was reported in 2000 by Zeeck and
co-workers.2 The ring system and carbon connectivity of
hexacyclinic acid are identical to those of FR182877; however,
differences between the two structures exist mainly in the
peripheral functionality and the stereochemistry of the ABC ring
fusions. Specifically, the C8 oxygen is acetylated in 2, the C11
substituent is at the carboxylic acid oxidation state in 2, the
C2-C20 double bond is hydrated relative to that of 1, and the
hexacyclinic acid ring fusions are AB (cis) and BC (trans) as
compared to AB (trans) and BC (cis) in FR182877. Due to the
structural similarity of these natural products, 1 and 2 are
undoubtedly biosynthetically related and have become the
defining members of a new class of potential cancer therapeutics.
FR182877 exhibits potent antitumor activity across a number
of susceptible cell lines including MCF-7, A549, HT-29, Jurkat,
P388, and B16. Results obtained in morphology studies
performed with baby hamster kidney (BHK) cells indicate a
mechanism of action involving microtubule stabilization and
interruption of the cell cycle in metaphase.1d As compared to
the control (untreated BHK cells), FR182877 performed simi-
larly to Taxol in this assay and, thus, holds forth promise as a
new lead structure for the development of antitumor therapeutics
that take advantage of this mechanism.
(1) (a) Muramatsu, H.; Miyauchi, M.; Sato, B.; Yoshimura, S. 40th Symposium
on the Chemistry of Natural Products; Fukuoka, Japan, 1998; Paper 83, p
487. (b) Sato, B.; Muramatsu, H.; Miyauchi, M.; Hori, Y.; Takese, S.; Mino,
M.; Hashimoto, S.; Terano, H. J. Antibiot. 2000, 53, 123. (c) Sato, B.;
Makajima, H.; Hori, Y.; Hino, M.; Hashimoto, S.; Terano, H. J. Antibiot.
2000, 53, 204. (d) Yoshimura, S.; Sato, B.; Kinoshita, T.; Takese, S.;
Terano, H. J. Antibiot. 2000, 53, 615. (e) Yoshimura, S.; Sato, B.; Kinoshita,
T.; Takese, S.; Terano, H. J. Antibiot. 2002, 55, C-1.
Sato documented the pronounced vulnerability of the strained
C2-C20 double bond of 1 to undergo air oxidation to an
(2) Ho¨fs, R.; Walker, M.; Zeeck, A. Angew. Chem., Int. Ed. 2000, 39, 3258.
9
10.1021/ja037643+ CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 13531-13540
13531

A R T I C L E S
Evans and Starr
epoxide, isolated as the diacetate 3 (eq 1). The epoxide
derivative did not appreciably inhibit tumor cell growth, and
thus the strained C2-C20 double bond may be necessary for
the observed antitumor activity. Hexacyclinic acid displays weak
anticancer activity and, in the case of MCF-7 (GI50 ) 14 µmol),
proved to be 2 orders of magnitude less potent, further impli-
cating the C2-C20 double bond as important for the antitumor
activity of this class of compounds. While it is tempting to assign
a biologically important role to the strained bridgehead enol
ether in 1, the available examples do not rule out acetylation of
the ring-A hydroxyls as the cause of diminished activity in the
series.3
Interest in the synthesis of FR182877 has risen sharply since
its introduction due to its important biological activity as well
as its complex polycyclic architecture. Efforts by the Nakada4
and Armstrong5 groups have resulted in publications of partial
syntheses of the AB ring system and a model of the DF ring
system, respectively. The Sorensen group has published two
partial syntheses6a,b and recently a completed total synthesis6c,d
of ent-1 nearly contemporaneous with our own communication7
of the completed synthesis of 1. In addition to these reports,
other research groups are currently investigating syntheses of
FR182877.8
The unifying theme of the published approaches to 1 has been
the disconnection of ring-B using a Diels-Alder cycloaddition
transform. The further disconnection of ring-D through a hetero-
Diels-Alder cycloaddition was recognized both by us and by
Sorensen,6a ultimately leading to our independent development
of similar syntheses of FR182877 based on a stereocontrolled
cascade of transannular Diels-Alder cycloaddition events. In
this Article, we provide a complete account of the design and
realization of a transannular cycloaddition strategy leading to
the completion of a total synthesis of the natural antipode of
FR182877. In addition, we describe our efforts to utilize the
tools of computational modeling to develop an understanding
of the stereochemical outcome of the Diels-Alder cycloaddition
cascade.
The hexacyclic structure of 1 incorporates a bridgehead enol
ether and an embedded cyclohexene ring, suggesting its bio-
synthetic origin from a sequence of intramolecular Diels-Alder
cycloaddition events. This idea was initially proposed by
Sorensen in 1999 and is further supported by the work of Zeeck
who isolated isotopically labeled 2 from Streptomyces cultures
inoculated with 13C-labeled acetate and propionate to establish
the connectivity of a supposed polyketide chain precursor to
Figure 1. Polyketide origin of hexacyclinic acid.
hexacyclinic acid (Figure 1). This allowed mapping of acetate
and propionate subunits to their respective locations in the
skeleton of 2 and proved to be consistent with a biosynthetic
pathway including Diels-Alder cycloadditions for generating
rings B and D.
The structures of 1 and 2 were established by standard
spectroscopic methods coupled with X-ray crystallographic
analysis, and the absolute configurations were determined in
each case by the Mosher ester method. However, on the basis
of our own analysis of the reported data for the FR182877
Mosher esters,9 we suspected that its absolute configuration had
been incorrectly assigned. Consequently, we designed a syn-
thesis targeting the configuration of hexacyclinic acid, confident
that, in the final analysis, it would prove to be homologous with
(-)-FR182877. Our presumption was later confirmed with the
publication of a correction in 2002.1e
Synthesis Plan. The structural relationship between FR182877
and hexacyclinic acid prompted us to develop a synthesis plan
targeting an advanced intermediate bearing the common ster-
eochemical features and peripheral functionality of both natural
products. The pivotal transformations associated with both
syntheses are the sequential transannular Diels-Alder10 and
hetero-Diels-Alder11 cycloadditions forming rings B and D,
respectively (Scheme 1). Because the ring-B Diels-Alder
reaction would be the initiating reaction (see Molecular Model-
ing section), the endo-TS transition state would afford the
FR182877 ring system while the diastereomeric exo-TS transi-
tion state would afford the hexacyclinic acid ring system. We
designed macrocycle 5, carrying a bromine substituent at C11,
so that it might serve as a common intermediate for the synthesis
of either natural product. In so doing, the syntheses of 1 and 2
could be unified under the common advanced intermediate 5.
Macrocycle 5 was envisioned to arise from a â-keto ester
alkylation and Suzuki12 cross coupling of the illustrated
C3-C10 and C11-C20 fragments. Each fragment could in turn
be derived from auxiliary-controlled enolborane syn-aldol
(3) The acetylated pachyclavulariaenones D-F are not cytotoxic, in contrast
to potent pachyclavulariaenone G having all three hydroxyls free. Suscep-
tible cell lines were P-388 and HT-29; however, no comment is made on
mechanism of action. See: Wang, G.-H.; Sheu, J.-H.; Duh, C.-Y.; Chiang,
M. Y. J. Nat. Prod. 2002, 65, 1475. A comparison of MM2 minimized
space-filling models of pachyclavulariaenone G and (-)-FR182877 using
Spartan showed considerable homology in size, shape, and disposition of
polar functionality between the two natural products (Evans and Starr,
unpublished).
(9) Ohtani, I.; Kusumi, T.; Kashman, Y.; Kakisawa, H. J. Am. Chem. Soc.
1991, 113, 4092. See also ref 1d for Mosher analysis data on (-)-FR182877.
(10) (a) For a review of TADA reactions, see: Marsault, E.; Toro, A.; Nowak,
P.; Deslongchamps, P. Tetrahedron 2001, 57, 4243. (b) For a recent
example, see also: Longithorone A: Layton, M. E.; Morales, C. A.; Shair,
M. D. J. Am. Chem. Soc. 2002, 124, 773. (c) Use of a bromodiene in TADA
reaction, see: Roush, W. R.; Koyama, K.; Curtin, M. L.; Moriarty, K. J.
J. Am. Chem. Soc. 1996, 118, 7502.
(4) Suzuki, T.; Nakada, M. Tetrahedron Lett. 2002, 43, 3263.
(5) Armstrong, A.; Goldberg, F. W.; Sandham, D. A. Tetrahedron Lett. 2001,
42, 4585.
(6) (a) Vanderwal, C. D.; Vosberg, D. A.; Weiler, S.; Sorensen, E. J. Org.
Lett. 1999, 1, 645. (b) Vanderwal, C. D.; Vosberg, D. A.; Sorensen, E. J.
Org. Lett. 2001, 3, 4307. (c) Vosburg, D. A.; Vanderwal, C. D.; Sorensen,
E. J. J. Am. Chem. Soc. 2002, 124, 4552. (d) Vanderwal, C. D.; Vosberg,
D. A.; Weiler, S.; Sorensen, E. J. J. Am. Chem. Soc. 2003, 125, 5393.
(7) Evans, D. A.; Starr, J. T. Angew. Chem., Int. Ed. 2002, 41, 1787.
(8) The following investigators have indicated their efforts toward FR182877
via the Internet: Clarke (http://www.nottingham.ac.uk/∼pczpac1/research-
web/synthesis.html), Prunet (http://www.dcso.polytechnique.fr/f1-032equipe-
.htm).
(11) (a) For this type of hetero-[4+2] cycloaddition, see: Shin, K.; Moriya,
M.; Ogasawara, K. Tetrahedron Lett. 1998, 39, 3765. (b) Takano, S.; Satoh,
S.; Ogasawara, K. J. Chem. Soc., Chem. Commun. 1988, 59. (c) Takano,
S.; Satoh, S.; Ogasawara, K.; Aoe, K. Heterocycles 1990, 30, 583. (c)
Yamauchi, M.; Katayama, S.; Baba, O.; Watanabe, T. J. Chem. Soc., Chem.
Commun. 1983, 281.
(12) For a review, see: Miyaura, N.; Suzuki, A. Chem. ReV. 1995, 95, 2457.
9
13532 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Total Synthesis of (−)-FR182877
A R T I C L E S
Scheme 1
reactions13 affording intermediates, designated as Aldol-A and
Aldol-B, from the appropriately functionalized aldehyde precur-
sors.
adduct 13 in 88% yield (Scheme 3).13 Conversion of 13 to the
derived Weinreb amide 14 (Me3Al, MeNH2OMeCl, 96%), sily-
lation of the secondary hydroxyl (TBSCl, 96%), then desilylation
of the primary hydroxyl (1:4 TsOH/nBu4NHSO4, MeOH, 89%)
gave primary alcohol 15. Successive oxidation (DMP, 94%)17
to aldehyde 16 and Corey-Fuchs olefination (CBr4, PPh3,
74%)18 afforded the dibromide fragment 17.
Results and Discussion
The Aldol Fragments. The synthesis began with the syn-
theses of the two obligatory aldehyde fragments (Scheme 2).
Silylation of 3-buten-1-ol (6) (TBSCl, imidazole, DMF, 90%)
was followed by ozonolysis and in situ reduction of the ozonide
with PPh3 to provide an 83% yield of aldehyde 7. Addition of
this aldehyde to a freshly prepared solution of isopropenylmag-
nesium bromide gave the derived allylic alcohol that was
subjected to the Johnson Orthoester Claisen procedure to give
the expected ester. The desired aldehyde 8 was then obtained
by DIBAL reduction (73%, three steps). Aldehyde 11 was
synthesized in two steps from cis-2-buten-1,4-diol (9). Monosi-
lylation of this diol (TBDPSCl, BuLi, THF, 94%)14 afforded
monosilyl ether 10 which underwent Parikh-Doering oxidation
(SO3-Pyr, TEA, DMSO, CH2Cl2, 63%)15 to give the desired
R,â-unsaturated aldehyde 11 after spontaneous cis-trans isomer-
ization during the oxidation procedure.16
Scheme 3 a
Scheme 2 a
a Key: (a) Bu2BOTf, TEA, 88%; (b) MeNHOMe-HCl, Me3Al, THF,
96%; (c) TBSCl, DMF, imid., 96%; (d) TsOH:nBu4NHSO4 (1:4), MeOH,
0 °C, 89%; (e) DMP, NaHCO3, 94%; (f) CBr4, PPh3, CH2Cl2, NaHCO3,
74%.
The synthesis of the C3-C10 “boronic acid fragment” 23
began with the illustrated aldol addition to provide the expected
a Key: (a) TBSCl, DMF, imid., 90%; (b) O3/O2 then PPh3, 83%; (c)
isopropenylmagnesium bromide, THF; (d) triethylorthoacetate, propionic
acid, 150 °C, 75% (two steps); (e) DIBAL, CH2Cl2, -78 °C, 97%; (f) nBuLi,
THF, TBDPSCl, 94%; (g) SO3-Pyr, DMSO, CH2Cl2, TEA, 63%.
(13) Gage, J. R.; Evans, D. A. Org. Synth. 1990, 68, 77; J. Am. Chem. Soc.
1981, 103, 2127.
(14) Roush, W. R.; Reilly, M. L.; Kogama, K.; Brown, B. B. J. Org. Chem.
1997, 62, 8708.
Synthesis of C11-C20 “dibromide fragment” 17 began with
an aldol addition of the enolborane derived from (R)-4-benzyl-
N-propionyl-2-oxazolidinone (12) to aldehyde 8 to give syn aldol
(15) Anderson, J. C.; McDermott, B. P. Tetrahedron Lett. 1999, 40, 7135.
(16) Complete cis-trans isomerization was observed after storing the neat
unpurified reaction product for 24 h.
(17) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13533

A R T I C L E S
Evans and Starr
Scheme 4 a
a Key: (a) Bu2BOTf, TEA, 89%; (b) MeNHOMe-HCl, Me3Al, THF, 97%; (c) HCCMgBr, 77%; (d) DIBAL, THF, -78 °C, 98%, >20:1 dr; (e) TBSCl,
DMF, imid., 98%; (f) Catechol-BH, Cy2BH (0.1 equiv), THF, then 1 N NaOH; 97%.
syn aldol adduct 18 in 89% yield (Scheme 4). The conversion
of 18 to amide 19 (Me3Al, MeNH2OMeCl, 97%) was followed
by ynone formation (HCCMgBr, 77%) to give acetylenic ketone
20. Diastereoselective syn reduction using Kiyooka’s method
(DIBAL, THF, >20:1 dr, 98%)19 afforded the syn diol 21. In
contrast, reduction of 20 with DIBAL in either Et2O or CH2Cl2
afforded a modest 4:1-syn diastereoselectivity. The relative
configuration of 21 at C8 was established by an examination
of the 13C NMR chemical shifts of the acetonide carbons of
derivative 24 (Scheme 4).20 The relevant shifts were within
accepted ranges for the syn diol isomer. The same analysis was
also carried out on the fully saturated analogue 25 to rule out
any influence on the result by the adjacent alkyne substituent.21
Double silylation (TBSCl, 94%) of 21 was followed by alkyne
hydroboration using the Arase procedure (catecholborane,
catatytic Cy2BH) and saponification (1 N NaOH, 97%) to give
the desired boronic acid fragment 23.22
tion of the starting materials 17 and 23. Products of protode-
borylation, oxidation, and elimination of the C8 silyloxane in
23 typically exceeded 20% yield from these reactions. Dibro-
mide 17 also decomposed, although the product(s) of that path-
way could not be identified. Silver bases completely decom-
posed the starting materials with no return of products 26 or
27. Carbonate bases improved selectivity for the desired coup-
ling product 26, but only Tl2CO3 retained a reasonable rate of
reaction at room temperature.27 Ultimately, Tl2CO3 became the
base of choice, and, using the improved conditions, Suzuki coup-
ling of 17 and 23 (Tl2CO3, Pd(PPh3)4) afforded bromodiene 26
in 84% yield.
Scheme 5
Fragment Coupling. Suzuki coupling of the C3-C10 (17)
and C11-C20 (23) fragments according to the TlOEt-mediated
procedure described by Roush23-24 was initially complicated
by the unexpected formation of the doubly coupled product 27
as a substantial side reaction (Scheme 5). With TlOEt, the
desired coupling product 26 and side product 27 were obtained
in a disappointing 2:1 ratio. Interestingly, the regioisomeric
monocoupled product was not observed, suggesting that 27
forms primarily through further reaction of 26 as its concentra-
tion builds over time and not through a parallel pathway
involving the regioisomer. It is well known through the work
of Roush, Kishi,25 and others26 that the course of this cross-
coupling reaction is greatly influenced by the choice of base.
In this reaction, strongly basic conditions (hydroxides or oxides)
or less halophilic cations (Na+, K+, Mg2+, Ba2+) were deleteri-
ous to the relative rates of reaction, and competitive decomposi-
The completion of the synthesis of the pivotal macrocyclic
intermediate 32 is summarized in Scheme 6. Partial reduction
of Weinreb amide 26 afforded the corresponding aldehyde
(DIBAL), which was subjected to a two-carbon Roskamp
homologation (ethyl diazoacetate, SnCl2, 70%, two steps)28 to
give â-keto ester 28. Selective removal of the TBDPS group
(18) Corey, E. J.; Fuchs, P. L. Tetrahedron Lett. 1972, 3769.
(19) Kiyooka, S.-i.; Kuroda, H.; Shimasaki, Y. Tetrahedron Lett. 1986, 27, 3009.
(20) Evans, D. A.; Rieger, D. L.; Gage, J. R. Tetrahedron Lett. 1990, 49, 7099.
(21) Rychnovsky, S. D.; Rogers, B.; Yang, G. J. Org. Chem. 1993, 58, 3511.
(22) Arase, A.; Hoshi, M.; Mijin, A.; Nishi, K. Synth. Commun. 1995, 25, 1957.
(23) Use of TlOEt: Frank, S. A.; Chen, H.; Kunz, R. K.; Schnaderbeck, M. J.;
Roush, W. R. Org. Lett. 2000, 2, 2691.
(24) First Suzuki coupling with vinyldibromide: Roush, W. R.; Riva, R. J. Org.
Chem. 1988, 53, 710.
(25) Uenishi, J.; Beau, J. M.; Armstrong, R. W.; Kishi, Y. J. Am. Chem. Soc.
1987, 109, 4756.
(26) Zhang, H. C.; Chan, K. S. Tetrahedron Lett. 1996, 37, 1043.
(27) Use of Tl2CO3: Marko, I. E.; Murphy, F.; Dolan, S. Tetrahedron Lett.
1996, 37, 2507.
9
13534 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Total Synthesis of (−)-FR182877
A R T I C L E S
Scheme 6 a
a Key: (a) Pd(PPh3)4 (0.05 equiv), Tl2CO3, H2O-THF (1:3), 23 °C, 84%; (b) DIBAL, CH2Cl2, -78 °C; (c) ethyldiazoacetate, SnCl2, 70% (two steps);
(d) TBAF, AcOH, DMF, 92%; (e) I2, PPh3, CH2Cl2; (f) Cs2CO3, THF (0.005 M), 23 °C, 77% (two steps); (g) Ph2Se2O3, SO3-Pyr, TEA, THF, 23 °C then
hexanes, 50 °C, 63%.
(nBu4NF hydrate, HOAc, DMF, 92%)29 provided macrocycliza-
tion precursor 29 which was iodinated (I2, PPh3) to afford the
sensitive allyl iodide 30 that was immediately submitted to
macrocyclization conditions (Cs2CO3, THF, room temperature,
77%, two steps) to give macrocycle 31 as an inconsequential
1:1 mixture of C2 diastereomers.30
It was correctly anticipated that intermediate 31, lacking the
C2-C3 double bond, would not engage in a spontaneous
transannular Diels-Alder reaction. Oxidation of 31 (Ph2Se2O3,
SO3-pyr, TEA, THF, room temperature then 50 °C) initiated
a sequence of transannular cycloadditions culminating in the
formation of 34 as a single diastereomer as the only isolable
product in 63% yield (Scheme 6). Our choice of this particular
reagent for the C2-C3 oxidation was based on Deslongchamps’
observation that high selectivity was observed in the oxidation
of a similar macrocycle.31
only one olefinic methine having a characteristic chemical shift
for a nonconjugated halogenated double bond (δ 5.89 ppm).
The course of this reaction sequence was monitored by thin-
layer chromatography (TLC). During the selenium-based oxida-
tion step, 31 was rapidly transformed (<1 h) to a new species,
believed to be 32 that further reacted to form the product 34.
Attempts to isolate 32 as a discrete entity were unsuccessful;
1
however, the unpurified H NMR spectrum of the isolation
attempt revealed a complex collection of resonances consistent
with a ca. 1:1 mixture of 32 and 32′ (eq 2). Signals consistent
with the tricyclic intermediate 33 were never observed, sug-
gesting the hetero-Diels-Alder cycloaddition is faster than the
normal ring-B Diels-Alder process. Further interpretation of
the spectrum was complicated by the presence of several
equivalents of selenium reagent and associated byproducts as
well as amine salts.
The transformations leading from 32 to 34 are believed to
follow the sequence of events illustrated in Scheme 6. After
oxidation of the C2-C3 bond in macrocycle 31, the targeted
macrocycle 32 spontaneously undergoes a normal electron-
demand ring-B Diels-Alder cycloaddition to form tricycle 33
followed by an inverse electron-demand hetero-Diels-Alder
cycloaddition to give the observed pentacyclic product 34. The
high stereoselectivity of the transannular cycloaddition cascade
became apparent upon comparison of the olefinic regions of
the 1H NMR spectra of the starting material 31 and product 34
of the reaction. While 31 displays six olefinic methine signals,
the oxidation-cycloaddition sequence afforded a product with
The presence of the 2-H pyran 32′ was not unexpected
because vinylogous â-keto esters are known to undergo revers-
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13535

A R T I C L E S
Evans and Starr
Scheme 8 a
ible 6-π cyclization to give the corresponding 2-H pyrans (eq
3). Moorhoff has documented the relationship between substitu-
tion at the X, Y, and Z positions and the observed equilibrium
constant for the formation of the pyran tautomer (eq 3).32 In
general, these equilibria ranged between Keq ) 1-10. If the 32
T 32′ equilibrium is in fact intervening, then the subsequent
heating (50 °C) is sufficient to establish a reversible 6-π
cyclization-ring opening, thus channeling the reaction cascade
through monocyclic intermediate 32.
In an attempt to more thoroughly understand the stereose-
lectivity of the transannular Diels-Alder cascade, a portion of
the post-Suzuki-coupling product 26 (Scheme 6) was diverted
to study the inherent diastereoselectivity of the ring-B Diels-
Alder cycloaddition. Desilylation of 26 followed by oxidation
with Dess-Martin periodinane afforded R,â-unsaturated alde-
hyde 35a.17 Gentle warming (60 °C) of a solution of 35a in
1
CDCl3 was monitored periodically by H NMR spectroscopy
until complete conversion to a mixture of two new species was
observed (ca. 3 h) (Scheme 7). The individual products were
isolated, and their structures were determined to be intramo-
lecular Diels-Alder products (36a and 37a) by a combination
of COSY and NOESY NMR analysis. In this experiment, very
high endo selectivity was obtained with an accompanying poor
diastereoface selection (dr 37:63 ) 36a:37a). After we had
performed this experiment, Sorensen published the results of
his own investigation of an analogous intramolecular Diels-
Alder reaction in which he also observed complete endo
selectivity and poor diastereoface selection.6b In Sorensen’s
experiment, a similar Diels-Alder precursor (ent-35b) under-
went a spontaneous intramolecular cycloaddition to afford a 61:
39 ratio of cycloadducts ent-36b and ent-37b.
a Key: (a) HF-MeCN, 89%; (b) Me3B3O3, Pd(dppf)Cl (0.1 equiv),
Cs2CO3, DMF-H2O (2:1), 100 °C, 71%; (c) TMSOK, THF; (d) Mukaiya-
ma’s reagent, NaHCO3, CH2Cl2, 62%.
result (32 f 34) (Scheme 6) suggests the C6-C8 stereotriad
plays only a muted role in differentiating the reacting π-faces
in the ring-B cycloaddition. We suspected this would be the
case, given that the particular C6-C8 stereotriad present in 32
is symmetric with respect to the proximal double bonds and,
thus, ought to have little differentiating effect upon them. The
transannular Diels-Alder strategy was envisioned to amplify
the intrinsically small diastereoface selection associated with
the C6-C8 triad by transmitting a stereodifferentiating influence
through the ring system from the C18 and C19 stereocenters.
We believe the resulting high stereoselectivity of the cycload-
dition sequence is governed by the stereocenters at C18 and
C19 acting through the rigidified macrocyclic architecture to
favor a single product of the Diels-Alder cascade. The nature
of this influence becomes apparent through computational
analysis using simulated transition state structures and is
discussed below in the Molecular Modeling section.
Scheme 7
COSY and NOESY studies established that the C5 stereo-
center in 34 had the desired configuration relative to the
C6-C8 triad, but lacked a suitably diagnostic NOESY cross-
peak to unambiguously assign the C9 and C12 stereocenters.
Because FR-182877 (1) and hexacyclinic (2) share the C5 ster-
eocenter in common, we could only surmise that 34 must
correspond to one of the natural product configurations for the
ABC ring system. Comparison of the coupling patterns for the
1
C5-C9 protons in the H NMR spectrum of 34 to the corre-
1
sponding signals in the H NMR spectrum of aldehyde 36a
obtained in our IMDA study (Scheme 7) provided qualitative
evidence that C5 and C9 were trans in 34, thus corresponding
to the FR-182877 ring system. Accordingly, the synthesis of
this target was completed by the route outlined in Scheme 8.
Global desilylation of 34 (HF-CH3CN, 89%) afforded triol 38.
This transformation was followed by Suzuki methylation using
the procedure of Gray (Me3B3O3, Pd(dppf)Cl2, Cs2CO3, 71%)
Subsequent comparison of the intramolecular Diels-Alder
result (35a f 36a + 37a) with the transannular Diels-Alder
(28) Holmquist, C. R.; Roskamp, E. J. J. Org. Chem. 1989, 54, 3258.
(29) Higashibayashi, S.; Shinko, K.; Ishizu, T.; Hashimoto, K.; Shirahama, H.;
Nakata, M. Synlett 2000, 1306.
(30) For Cs2CO3-mediated macrocarbocyclization, see: Phoenix, S.; Bourque,
E.; Deslongchamps, P. Org. Lett. 2000, 2, 4149.
(31) Ouellet, L.; Langlois, P.; Deslongchamps, P. Synlett 1997, 689 and
references therein.
(32) Moorhoff, C. M. Synthesis 1997, 685.
9
13536 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Total Synthesis of (−)-FR182877
A R T I C L E S
to give 39.33 Saponification of the ethyl ester (TMSOK, THF)34
yielded seco-acid 40, and lactonization (1-methyl-2-chloropy-
ridinium iodide, NaHCO3, 62%, two steps)35 afforded a white
1
solid whose H NMR and mass spectral characteristics were
identical to those published for the natural product 1. In addition,
the 13C NMR spectral data also agree to within a 2% margin of
error. Finally, synthetic 1 exhibited an optical rotation of
[R2D3] ) -5° as compared to [R2D3] ) -3.5° reported for the
natural sample, leading us to conclude that synthetic 1 was of
the same absolute stereochemistry as natural (-)-FR182877.
Molecular Modeling
Molecular modeling was used as a tool to address two
questions raised during the course of the synthesis described
above: (1) In what order do the two transannular Diels-Alder
cycloadditions occur? (2) What is the origin of the high
diastereoface selection in the normal electron-demand transan-
nular Diels-Alder reaction in the construction of ring-B?
Cycloaddition Order. Simplified macrocycle 32A was used
to calculate the HOMO and LUMO associated with the two
Diels-Alder cycloadditions forming the ABCDE ring system
(Fgure 2). Calculations were performed using Spartan 5.1
running on an SGI Indigo2 computer.36 The calculated Pz-
coefficients of the HOMO and LUMO (for relevant atoms) of
32A are shown in Figure 2. The best HOMO-LUMO orbital
overlap leads to the C4(Pz ) -0.26)-C12(Pz ) -0.48) and
C5(Pz ) 0.51)-C9(Pz ) 0.38) bonds, the σ-bonds formed in
the ring-B normal electron-demand Diels-Alder (NDA) cy-
cloaddition. The C3(Pz ) -0.56)-C14(Pz ) -0.07) and O1-
(Pz ) -0.29)-C15(Pz ) 0.03) combinations associated with
the hetero-Diels-Alder (HDA) cycloaddition are considerably
weaker due to small HOMO Pz coefficients for the C14-C15
double bond. The π-bonding electron density of C14-C15 is,
necessarily, lower lying than that of the C9-C12 diene
(HOMO[-1] is 0.11 kcal/mol lower in energy), consistent with
the generally higher reactivity of dienes toward electrophilic
reagents. This frontier molecular orbital analysis predicts a better
HOMO-LUMO overlap for the normal ring-B Diels-Alder
reaction than for the ring-D hetero-Diels-Alder process in
macrocycle 32, in accord with the sequence in Scheme 6.
Figure 3. Diagramatic definitions of endo-I and endo-II transition
structures.
Alder (TADA) cycloaddition (32 f 33) was next examined
using semiempirical calculations.37 The contrast between the
poorly face-selective intramolecular Diels-Alder (IMDA) cy-
cloaddition (Scheme 7) and the highly face-selective (TADA)
cycloaddition (32 f 33) (Scheme 6) reinforces the conjecture
that the stereocenters at C18 and C19 are involved in creating
the observed TADA face selectivity. This proposal was exam-
ined by considering the two endo TADA transition structures
Endo-1 TS and Endo-2 TS (Figure 3). The diene and dienophile
termini were constrained at specified distances ranging from
2.9 to 2.1 Å, with energy minimizations (PM3) carried out at
0.2 Å intervals for the two endo cycloaddition geometries.
Published Diels-Alder transition state bond distances in the
2.0-2.2 Å range were used in the selection of the internuclear
distance constraints.38 The nine simplified macrocycles (41-
44d, Figure 4) were minimized with the indicated distance
constraints leading either to the observed, natural product
(endo-1 TS) or to its endo diastereomer (endo-2 TS).
For each of the structures 41-44d, five energy-minimized
geometries were obtained, where the interatomic distances
between the dienophile (C4 and C5) diene and (C9 and C12)
termini were constrained to 2.9, 2.7, 2.5, 2.3, and 2.1 Å. From
these data, energy curves (kcal mol-1 Å-1) were generated
representing the increasing energy penalty of approaching the
transition state. For each macrocycle, two curves were gener-
ated: one representing the endo-1 π-face alignment and the
other representing the endo-2 π-face alignment (Figure 3). The
profiles of the two curves generated for each macrocycle were
then compared to qualitatively determine which endo approach
is preferred by that macrocycle in the TADA cycloaddition. This
preference was graphically correlated with the C18 and C19
substituents positioned on the macrocycle.
(33) Gray, M.; Andrews, I. P.; Hook, D. F.; Kitteringham, J.; Voyle, M.
Tetrahedron Lett. 2000, 41, 6237.
(34) Laganis, E. D.; Chenard, B. L. Tetrahedron Lett. 1984, 25, 5831.
(35) Mukaiyama, T.; Usui, M.; Saigo, K. Chem. Lett. 1976, 49.
(36) The conformational geometry of model 32 was optimized at the PM3 level
with the π-conjugated systems constrained to coplanarity using the dihedral
constraints outlined in ref 37. No interatomic distance constraints were
imposed for this experiment. Single point energy and MO coefficients were
computed using ab initio methods at the RHF/STO-3G level on the
geometry optimized model 32. All Px and Py coefficients were 0 ( 0.05
for both the HOMO and the LUMO of the indicated atoms.
(37) Dihedral angle constraints were used for all PM3 calculations as
follows: O(ester CO)dC1-C2dC3, 180°; C2dC3-C4dC5, 180°;
C9dC10-C11dC12, 0°. Interatomic distance contraints were applied to
C5-C9 and C4-C12 at distances indicated in the text for the individual
experiments.
Figure 2. HOMO and LUMO coefficients for macrocycle 32A.
Ring-B Cycloaddition. The origin of the high observed
diastereoface selection in the normal ring-B transannular Diels-
(38) Storer, J. W.; Raimondi, L.; Houk, K. N. J. Am. Chem. Soc. 1994, 116,
9675.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13537

A R T I C L E S
Evans and Starr
Figure 4. Macrocycles employed for molecular modeling studies.
Macrocycles possessing one of the two stereocenters
(42a,b-43a,b) were examined next (Figure 6). The curves
generated from 42a and 42b, having 19R and 19S methyl
configurations, respectively, and no C18 methoxyl substituent,
demonstrate a clear bias imparted by the C19 methyl group.
Here, the 19R (42a) configuration favors the endo-I TS by 4.7-
8.2 kcal/mol, while the 19S (42b) configuration favors endo-II
TS by 5.6-8.1 kcal/mol.
A similar bias is apparent in the curves generated by 43a
and 43b carrying only the C18 methoxyl substituent (Figure
6). The 18S configuration in 43b favors the endo-1 TS by
2.4-5.5 kcal/mol, while the 18R configuration in 43a favors
the endo-2 TS by 3.3-6.5 kcal/mol. This trend suggests that
the combination of the two configurations (18S, 19R), the natural
configurations found in (-)-FR182877, is reinforcing the endo-1
TS π-face alignment observed in the laboratory. Extending this
additivity model further, it would be expected that (18R, 19S)
would favor the endo-II diastereomer while both (18S, 19S)
and (18R, 19R) would be nonreinforcing cases with no clear
diastereoface preference.
Fully substituted cases 44a-44d were next examined (Figure
7). Energy versus distance plots for these four structures predict
that the “natural” (18S, 19R) macrocycle 44a displays a strong
(4.9-9.4 kcal/mol) preference for the endo-1 TS in accord
with the laboratory outcome. The unnatural (18R, 19S) diaster-
eomeric macrocycle 44b exhibits a similarly strong preference
for the endo-2 TS alignment, while macrocycles 44c and 44d,
(18R, 19R) and (18S, 19S), respectively, deliver less clear-cut
results. Macrocycle 44c modestly favors the endo-I TS at
larger π-face separations (2.9-2.7 Å). The curves then cross,
and the endo-II TS is favored as the π-face separation decreases
(2.5-2.1 Å).
Figure 5. Endo diastereoselection control experiments.
Figure 5 shows the energy curves obtained in a control
experiment wherein 41 lacking substitution at both C18 and
C19 is evaluated. Only a slight difference in energy (0.5-1.1
kcal/mol) appears between the curves leading to the endo-1 and
endo-2 transition states. Calculations of the two endo approach
curves in a model of the acyclic system 45 produced a similarly
small energy difference between the two transition state energy
curves, also consistent with the laboratory outcome of the acyclic
IMDA and further confirming the negligent impact of the
C6-C8 stereocenters in this analysis (Scheme 7).
Macrocycle 44d, on the other hand, seems to show no dia-
stereofacial preference for either transition state geometry at
the largest π-face separation (2.9 Å); however, at shorter inter-
Figure 6. Macrocycles substituted at either C18 or C19.
9
13538 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Total Synthesis of (−)-FR182877
A R T I C L E S
Figure 7. Macrocycles substituted at both C18 and C19.
Scheme 9
nuclear distances (2.3-2.1 Å), the two curves diverge to favor
the endo-1 transition state, albeit by a smaller margin than was
seen for the reinforcing cases. It is apparent from this analysis
that the nonreinforcing cases, 44c and 44d, confound the dia-
stereoface selection of the TADA cycloaddition, by bringing
the two possible endo transition states closer in energy for those
diastereomers.
It is noteworthy that the limitation posed by the macrocyclic
architecture is primarily the lack of free rotation around the
C18-C19 and C19-C20 bonds and the opportunity to alleviate
energetically costly interactions across these bonds. In sum-
mary, the computational data and the empirical results impli-
cate both the C18 and the C19 stereocenters as the primary
determinants for the diastereoface selectivity of the TADA cy-
cloaddition leading to ring B with the best selectivities being
observed in the reinforcing cases (18S, 19R) and (18R, 19S)
leading to endo-I and endo-II, respectively. The preferred endo-I
transition structure is illustrated in Figure 8. As stated earlier,37
constraints on dihedral angles, L1, L2, and L3, were imposed
on the calculations, while no constraint was imposed on the
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13539

A R T I C L E S
Evans and Starr
Scheme 10
Figure 8. Preferred ring-B Diels-Alder transition state.
Conclusion
There are numerous examples of complex cycloaddition
reactions that occur during the biosynthesis of a broad spectrum
of natural products.39 Some of these transformations are en-
zymatically mediated, while others need not be so controlled.
Rather, the stage could be set for a complex series of trans-
formations that are mediated by internal control elements such
as resident stereocenters within the molecular architecture. Such
appears to be the case with FR182877.
A related example serves to illustrate this point (Scheme 10).
Our group had previously exploited a transannular C-C bond
forming reaction in the synthesis of miyakolide.40 In this synthe-
sis, three stereocenters on the advanced macrocyclic intermediate
47 preorganized the transition state of the transannular aldol
reaction 47 f 48 to impart a high degree of stereocontrol to
the ensuing reaction. Again, the synthesis activity demonstrates
that enzyme mediation of this aldol step is not obligatory.
intra-annular dihedral angle L4 linking the ketone carbonyl to
the dienophile π-system. Nonetheless, the preferred ring con-
formation permits conjugation of this carbonyl moiety to the
π-system.
The analysis of the hetero-Diels-Alder cycloadditions is
provided in Scheme 9. The tricyclic intermediate, formed as a
consequence of the first cycloaddition, exists in the two low-
energy conformations 33-A and 33-B differing in energy by 2
kcal/mol. The more stable conformer 33-A is the kinetic
conformation that evolves from the first cycloaddition, while
conformation 33-B is the obligatory conformation for the desired
hetero-Diels-Alder process leading to 34. As can be seen by
the three-dimensional structures, the two macrocyclic conforma-
tions only differ in the region C13-C17 where the two
diastereofaces of the ∆15 trisubstituted olefin are oriented
toward the transannular heterodiene. If no conformational
interconversion were possible, the cycloaddition cascade would
have probably led to the undesired adduct 46. Experiment tells
us that conformational interconversion must have occurred.
We next carried out calculations to estimate the relative
energies of the hetero-Diels-Alder cycloaddition steps. In the
first instance, we applied only a single-bond constraint (2.20
Å) between the forming C3-C14 bond. This scenario leads to
a carbocation intermediate. In this instance, the desired cy-
cloaddition mode is favored by 2 kcal/mol. In the more realistic
situation, we applied an additional bond constraint (2.40 Å)
between the C14 and carbonyl component in the heterodiene.
When both transition state bond constraints are applied, the
cycloaddition mode leading to the desired product is favored
by 27 kcal/mol!
Acknowledgment. We are grateful to Dr. Bunji Sato of
Fujisawa Pharmaceutical Co. for providing spectra and helpful
advice regarding (-)-FR182877. Support has been provided by
the National Institutes of Health (GM-33328-18). We also
acknowledge the NIH for postdoctoral fellowship support for
J.T.S.
Supporting Information Available: Experimental details and
analytical data for all new compounds (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
JA037643+
(39) Stocking, E. M.; Williams, R. M. Angew. Chem., Int. Ed. 2003, 42, 3078.
(40) Evans, D. A.; Ripin, D. H. B.; Halstead, D. P.; Campos, K. R. J. Am.
Chem. Soc. 1999, 121, 6816.
9
13540 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003