FR901464: Total synthesis, proof of structure, and evaluation of synthetic analogues

Article abstract of DOI:10.1021/ja016615t

The natural product FR901464 (1) was isolated by the Fujisawa Pharmaceutical Co. and shown to have intriguing biological properties including impressive antitumor activity. In this paper we describe the first total synthesis of 1 in full detail. A chiral building block synthetic strategy was used to assemble the target: optically active components were generated using asymmetric catalytic reactions, and these fragments were coupled together at a late stage in a convergent synthesis. In particular, a versatile, asymmetric hetero-Diels-Alder (HDA) reaction was developed in the context of this synthesis and used with great effectiveness for the preparation of the two densely functionalized pyran rings. The flexible nature of the synthetic route also allowed us to prepare a series of analogues of 1. These compounds were used to prove the relative stereochemistry of the natural product as well as to probe the importance of certain structural features of FR901464 with regard to biological activity.

Full text of DOI:10.1021/ja016615t

9974
J. Am. Chem. Soc. 2001, 123, 9974-9983
FR901464: Total Synthesis, Proof of Structure, and Evaluation of
Synthetic Analogues
Christopher F. Thompson, Timothy F. Jamison, and Eric N. Jacobsen*
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138
ReceiVed July 13, 2001
Abstract: The natural product FR901464 (1) was isolated by the Fujisawa Pharmaceutical Co. and shown to
have intriguing biological properties including impressive antitumor activity. In this paper we describe the
first total synthesis of 1 in full detail. A chiral building block synthetic strategy was used to assemble the
target: optically active components were generated using asymmetric catalytic reactions, and these fragments
were coupled together at a late stage in a convergent synthesis. In particular, a versatile, asymmetric hetero-
Diels-Alder (HDA) reaction was developed in the context of this synthesis and used with great effectiveness
for the preparation of the two densely functionalized pyran rings. The flexible nature of the synthetic route
also allowed us to prepare a series of analogues of 1. These compounds were used to prove the relative
stereochemistry of the natural product as well as to probe the importance of certain structural features of
FR901464 with regard to biological activity.
Background and Retrosynthetic Analysis
In 1996, the Fujisawa Pharmaceutical Co. reported the
isolation of three novel compounds produced by a bacterium
obtained from a Japanese soil sample.1 The three were closely
related structurally (Figure 1), differing only in the substitution
pattern about the right-hand pyran ring, and were designated
FR901463, FR901464, and FR901465.1 These natural products
were discovered as part of a program to identify small molecules
that induce transcriptional regulation, following the hypothesis
that such compounds might hold potential as antitumor agents.
All three were found to enhance the activity of the promoter of
the SV40 DNA tumor virus in M-8 cells.1,2 Upon further
investigation, the Fujisawa researchers found that these com-
pounds displayed potent cytotoxicity against a number of
different human solid tumor cell lines, with IC50 values on the
order of 1 ng/mL.1 A study evaluating the ability of these small
molecules to prolong the life of tumor-bearing mice revealed
that FR901464 (1) was most active, and as a result this
compound was studied further in mechanistic experiments.3
FR901464 was shown to effect G1 and G2/M cell cycle arrest
and also to induce DNA fragmentation as well as cell shrinkage
during the process of causing cell death.3 Some of the
phenotypic changes observed in M-8 cells treated with FR901464
were found to be similar to those of known histone deacetylase
inhibitors such as trichostatin A.3-5 However, analysis of the
acetylation state of the histones isolated from M-8 cells indicated
that 1 was in fact not an inhibitor of histone deacetylase activity,
but instead appeared to have a distinct, unknown, and potentially
very interesting mode of action.4
Figure 1. Structures of FR901463, FR901464, and FR901465.
We were inspired to pursue a total synthesis of FR901464
because of the interesting biological properties outlined above
coupled with the limited supply of natural material,6 and also
as a result of the interesting structural characteristics of the
molecule.7 FR901464 is composed of three discrete chiral
fragments joined by conjugated, unsaturated linkages. As such,
we considered it a potential target for a chiral building block
synthetic strategy, wherein the target is assembled in a
convergent manner using effective and general fragment cou-
pling strategies from optically active components generated
independently by asymmetric catalysis.8 Furthermore, the rela-
tive stereochemistry of 1 had been assigned only tentatively:
the absolute stereochemistry at C4′ had been established by
(1) Nakajima, H.; Sato, B.; Fujita, T.; Takase, S.; Terano, H.; Okuhara,
M. J. Antibiot. 1996, 49, 1196-1203.
(2) M-8 cells are a human mammary adenocarcinoma cell line stably
transfected with an SV40 promoter-driven CAT reporter gene. See ref 1.
(3) Nakajima, H.; Hori, Y.; Terano, H.; Okuhara, M.; Manda, T.;
Matsumoto, S.; Shimomura, K. J. Antibiot. 1996, 49, 1204-1211.
(4) Nakajima, H.; Kim, Y. B.; Terano, H.; Yoshida, M.; Horinouchi, S.
Exp. Cell Res. 1998, 241, 126-133.
(6) Samples of natural FR901464 were not available from the Fujisawa
Pharmaceutical Co. nor from any other sources.
(7) For a preliminary communication of this work, see: Thompson, C.
F.; Jamison, T. F.; Jacobsen, E. N. J. Am. Chem. Soc. 2000, 122, 10482-
10483. For other synthetic efforts toward the synthesis of FR901464, see:
Horigome, M.; Watanabe, K.; Kitahara, T. Tennen Yuki Kagobutsu Toronkai
Koen Yoshishu 1999, 41, 73-78; Chem. Abstr. 2000, 132, 676.
(5) Taunton, J.; Hassig, C. A.; Schreiber, S. L. Science 1996, 272, 408-
411.
10.1021/ja016615t CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/25/2001

Study of the Natural Product FR901464
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9975
Scheme 1. Preliminary Retrosynthetic Analysis
fragments of FR901464 using asymmetric catalytic methods.
We anticipated taking advantage of recent advances in asym-
metric reduction methodologies to introduce the lone stereo-
center in fragment 2. The two densely functionalized pyran rings,
each of which contains four stereocenters, clearly presented a
more substantial synthetic challenge. We recognized that an
endo-selective asymmetric hetero-Diels-Alder (HDA) reaction
between a hexadiene derivative such as 9 and a suitably
functionalized aldehyde (e.g., 10) would provide a most efficient
construction of the pyran ring framework of the central fragment
with simultaneous introduction of three out of the four stereo-
centers, as in 8. A similar synthetic strategy might allow
enantioselective access to dihydropyran derivative 12, a possible
intermediate in the preparation of the right-hand fragment.
However, at the outset of this project, no such enantioselective
HDA reactions were known between unactivated aldehydes and
dienes bearing a single electron-donating substituent.10 Recog-
nizing the potential utility of such methodology not only for
the synthesis of 1 but also as a general synthetic method, we
undertook a focused effort aimed toward the discovery of
catalysts for the asymmetric HDA reaction between 9 and 10.
This led to the discovery of the novel tridentate chromium
catalyst 14, which provides the desired mono-oxygenated
dihydropyran adduct 8 with excellent enantio- and diastereo-
selectivity (eq 1).11 In addition, the new catalytic method
degradation studies, but only the relative stereochemistry within
each of the pyran rings had been ascertained by NMR.9 Thus,
the relative stereochemistry between the three fragments had
not been established.9a In principle, the use of asymmetric
catalytic methods would allow the construction of either
enantiomer of each fragment in a straightforward manner using
the same chemistry. Coupling these chiral fragments would
provide access to diastereomers that differ only in the relative
stereochemical relationship between the fragments, thereby
allowing unambiguous assignment of the complete stereochem-
istry of FR901464. Finally, a convergent and flexible route to
1 would allow straightforward access to synthetic analogues of
1 of varying complexity. These may serve to help establish
which elements of the complex structure are important for
biological activity and ultimately may lead to the development
of biological tools to help elucidate the mechanism of action
of FR901464.
Our synthetic plan was devised taking all of the objectives
outlined above into account. FR901464 contains two pyran rings
joined via a diene moiety as well as an acyclic side chain
attached to the central pyran ring by an R,â-unsaturated amide
bond. Disconnection of the amide and the diene leads to
fragments 2-4 (Scheme 1). We envisioned joining these
components at the latest possible stage in the synthesis using a
cross-coupling reaction to generate the diene and standard
peptide coupling methods to form the amide. Our principal focus
then became the highly efficient construction of the three chiral
exhibited promising scope with respect to both diene and
dienophile, with a variety of reacting partners undergoing endo-
selective cycloaddition with high ee. In particular, reaction of
aldehyde 10 with pentadiene derivative 13 afforded dihydro-
pyran 12 in 78% yield and 98% ee (>95% de) in the presence
of catalyst (1R,2S)-14a (eq 2).11
(8) For other work from this group illustrating this chiral building block
approach, see: (a) Schaus, S. E.; Brånalt, J. E.; Jacobsen, E. N. J. Org.
Chem. 1998, 63, 4876-4877. (b) Lebel, H.; Jacobsen, E. N. J. Org. Chem.
1998, 63, 9624-9625. (c) Chavez, D. E.; Jacobsen, E. N. Angew. Chem.,
Int. Ed., in press.
(9) (a) Nakajima, H.; Takase, S.; Terano, H.; Tanaka, H. J. Antibiot.
1997, 50, 96-99. (b) Nakajima, H. Private communication.
(10) The only asymmetric HDA reactions able to employ a diene
containing less than two oxygens required very reactive, electron-deficient
aldehydes such as glyoxylate derivatives. For a recent review of the HDA
reaction, see: Jorgensen, K. A. Angew. Chem., Int. Ed. 2000, 39, 3558-
3588.
(11) Dossetter, A. G.; Jamison, T. F.; Jacobsen, E. N. Angew. Chem.,
Int. Ed. 1999, 38, 2398-2400.

9976 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Thompson et al.
Scheme 2. Synthesis of the Left-Hand Side Chain
With the new asymmetric catalytic HDA reaction providing
convenient access to the key building blocks 8 and 12, the total
synthesis of FR901464 required the development of effective
routes to the fully elaborated tetrahydropyran units as well as a
successful fragment coupling strategy. As described in this
paper, both tasks presented significant challenges, ultimately
requiring us to redesign our approach to the initial chiral building
blocks. This exercise inspired the extension of the HDA
methodology in interesting new directions, and it also led us to
a substantially more efficient overall synthesis of 1 than that
which was anticipated in the initial plan. The final route proved
to be very flexible, providing straightforward access to diaster-
eomers in order to confirm the absolute stereochemistry of 1,
as well as to novel analogues for preliminary biological studies.
Results and Discussion
(I) Fragment Synthesis. Although the left-hand side chain
2 is a relatively simple, low-molecular-weight compound, its
efficient synthesis presents interesting challenges nonetheless.
We anticipated that the Z-substituted acrylic acid derivative
might be sensitive to isomerization and other decomposition
pathways, so we endeavored to introduce the alkene as late as
possible in the synthesis by means of partial hydrogenation of
the appropriate propargylic alcohol derivative. Of the various
valuable and highly effective methods devised recently for the
preparation of enantioenriched propargylic alcohols,12 the Noyori
Ru-catalyzed transfer hydrogenation of ynones13 has proven to
be particularly versatile.8b,c Its application to the reduction of
commercially available 4-(trimethylsilyl)-3-butyn-2-one 6 af-
forded propargylic alcohol 5 in excellent yield and with high
enantioselectivity (Scheme 2). Following desilylation, the
terminal alkyne was carboxylated by sequential deprotonation
and reaction with CO2 to furnish hydroxy acid 15 upon
workup.14 Acetylation of the free hydroxyl group and Lindlar
reduction provided the desired cis allylic acetate, thereby
completing a short and efficient synthesis of fragment 2.
Scheme 3. Initial Approach to the Central Fragment
The initial approach taken to the synthesis of the central
fragment began with the asymmetric catalytic HDA reaction
between commercially available aldehyde 10 and diene 9, which
afforded adduct 8 with outstanding enantio- and diastereose-
lectivity (eq 1).11 Oxidation15 of the silyl enol ether functionality
provided R-hydroxy ketone 7 in good yield and with high
diastereoselectivity (Scheme 3). Following tosylhydrazone
formation, a one-pot reduction/fragmentation reaction led to the
tetrasubstituted tetrahydropyran 16 in modest yield.16 After a
straightforward protection/deprotection sequence afforded 17,
installation of the alkyne was investigated. A wide variety of
leaving groups, counterions, and reaction conditions were
examined, yet this seemingly straightforward substitution proved
extremely problematic. It was found that SN2 displacement could
be effected on the triflate derivative of 17 with propynyllithium
in THF/HMPA to furnish 18, but this procedure was difficult
to effect reproducibly and afforded low yields in the best of
cases.
While the route outlined in Scheme 3 did provide access to
the central pyran fragment, the late-stage incorporation of the
alkyne unit in the conversion of 17 to 18 imposed a severe
penalty with respect to yield and efficiency. In retrospect, the
fact that the alkyne was being introduced in this manner was
an artifact of the selection of silyloxyacetaldeyde 10 as the
dienophile in the critical asymmetric HDA reaction. Indeed, it
became evident that intermediate 8 was ill-suited as a precursor
to the central fragment, a problem that was rendered particularly
ironic by the fact that we had designed the asymmetric catalytic
reaction producing 8 specifically for the purposes of this
synthesis.
In reevaluating the key HDA reaction, we considered three
approaches that would allow introduction of the alkyne moiety
directly as a component of the dienophile in the HDA reaction.
The first and most straightforward plan employed aldehyde 19,
wherein the triple bond is positioned correctly with respect to
(12) For leading recent examples, see: (a) Brown, H. C.; Ramachandran,
V. P. Acc. Chem. Res. 1992, 25, 16-24. (b) Corey, E. J.; Cimprich, K. A.
J. Am. Chem. Soc. 1994, 116, 3151. (c) Helal, C. J.; Magriotis, P. A.; Corey,
E. J. J. Am. Chem. Soc. 1996, 118, 10938-10939. (d) Nakamura, M.; Hirai,
A.; Sogi, M.; Nakamura, E. J. Am. Chem. Soc. 1998, 120, 5846-5847. (e)
Tao, B.; Ruble, C.; Hoic, D.; Fu, G. C. J. Am. Chem. Soc. 1998, 120, 5091-
5092. (f) Li, Z.; Upadhyay, V.; DeCamp, A. E.; DiMichele, L.; Reider, P.
J. Synthesis 1999, 1453-1458. (g) Frantz, D. E.; Fa¨ssler, R.; Carreira, E.
M. J. Am. Chem. Soc. 2000, 122, 1806-1807.
(13) Matsumura, K.; Hashiguchi, S.; Ikariya, T.; Noyori, R. J. Am. Chem.
Soc. 1997, 119, 8378-8379.
(14) Getty, S.; Berson, J. J. Am. Chem. Soc. 1991, 113, 4607-4621.
(15) Yang, D.; Wong, M.; Yip, Y. J. Org. Chem. 1995, 60, 3887-3889.
(16) Kim, S.; Oh, C. H.; Ko, J. S.; Ahn, K. H.; Kim, Y. J. J. Org. Chem.
1985, 50, 1927-1932.

Study of the Natural Product FR901464
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9977
Scheme 5. Modified Route to the Central Fragment
Scheme 4. New Synthetic Approaches to the Central
Fragment
(1R,2S)-14b catalyzed the reaction in good yield, the SbF6
complex afforded substantially higher ee (95 vs 86%). As before,
three of the four stereogenic centers found in the central pyran
ring of 1 were thus established in a single reaction. Rubottom
oxidation of 22 to 23 proceeded diastereoselectively21 and in
good yield using a biphasic system buffered with NaHCO3 and
toluene as the organic solvent. A two-step procedure was applied
for reduction and fragmentation of the tosylhydrazone inter-
mediate to provide 24 in good yield.22 Following desilylation,
the critical isomerization of the terminal alkyne to the desired
internal position was effected with KOtBu in DMSO.18 This
precedented, yet rarely used methodology proved extraordinarily
effective, with the reaction proceeding in virtually quantitative
yield within 1 min.
target fragment 3 (Scheme 4A). However, efforts to use
aldehyde 19 in HDA reactions were unsuccessful, as this
aldehyde proved to be highly unstable and decomposed under
the reaction conditions, presumably through isomerization to
highly reactive allenic intermediates. Attempts to effect HDA
reactions of 19 protected as a dicobalt hexacarbonyl complex
also led to complex product mixtures. The second strategy
employed conjugated 4-carbon ynal 20 and relied on the well-
precedented tendency of internal alkynes to migrate to a terminal
position in the presence of strong base (Scheme 4B).17 Comple-
tion of middle fragment 3 would then be attained by methylation
of the terminal acetylide. The third approach relied on cycload-
dition of the 5-carbon terminal ynal 21 and would require a far
more unusual tactic involving selective isomerization of the
terminal alkyne to an internal position18 (Scheme 4C). This
strategy was deemed the more attractive of the latter two, not
only by virtue of it being more concise but also because it would
involve establishing the entire carbon framework of 3 in the
initial HDA reaction.
Conversion of alcohol 25a to the key intermediate 3 was
accomplished in low yield by means of invertive displacement
with NaN3 on the corresponding mesylate. We had anticipated
from the outset that such substitution processes on the central
fragment would be challenging due to the sterically congested
nature of the tetrasubstituted tetrahydropyran ring. Furthermore,
we recognized that the azide functionality might prove incom-
patible with the conditions to be used in late-stage coupling
reactions, and that therefore this substitution would have to be
timed carefully in the overall synthetic sequence. As a result,
we decided to postpone optimization of the nitrogen installation
until the setting for coupling of the two pyran rings was more
clearly defined.
The asymmetric catalytic HDA reaction between 9 and
aldehyde 21b afforded adduct 22 with outstanding enantio- and
diastereoselectivity (Scheme 5).19,20 While both (1R,2S)-14a and
Our first approach to the highly substituted right-hand pyran
ring of 1 began with a HDA reaction between diene 13 and 10
(17) Brown, C. A.; Yamashita, A. J. Am. Chem. Soc. 1975, 97, 891-
892.
(18) Takano, S.; Shimazaki, Y.; Iwabuchi, Y.; Ogasawara, K. Tetrahe-
dron Lett. 1990, 31, 3619-3622.
(19) For the method of synthesis of aldehyde 21a, see: Cruciani, P.;
Stammler, R.; Aubert, C.; Malacria, M. J. Org. Chem. 1996, 61, 2699-
2708.
(20) Aldehyde 21a underwent cycloaddition with similar ee, but its
reactivity was greatly diminished. The absolute stereochemistry of HDA
adducts was assigned by analogy to assignments made in ref 11.
(21) The undesired epimer of the R-hydroxy ketone was generated in
10-15% yield but was removed readily by flash chromatography.
(22) Nair, V.; Sinhababu, A. J. Org. Chem. 1978, 43, 5013-5017.

9978 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Thompson et al.
Scheme 8. Modified Route to the Right-Hand Fragment
Scheme 6. Initial Approaches to the Right-Hand Fragment
Scheme 7. New Retrosynthetic Analysis of the Right-Hand
Fragment
(eq 2). Cycloadduct 12 was subjected to Saegusa oxidation23
to give dihydropyranone 26 in good yield (Scheme 6). However,
efforts aimed toward elaboration of 26 to the fully functionalized
right-hand fragment by enolization-oxidation of the ketone in
26 or by prior hydration to 28 were all unsuccessful. As in the
case of the middle fragment, the problems encountered in the
synthesis of the fully functionalized tetrahydropyran framework
were tied to the original selection of reacting partners in the
HDA reaction. Therefore, we sought an alternative route to the
right-hand fragment that would utilize the key cyclization
strategy to greater effect.
As illustrated in the retrosynthetic analysis in Scheme 7,
installation of the C4 hydroxyl group as in 31 could be rendered
straightforward if the enol ether generated in the HDA cycliza-
tion bore the regiochemistry shown in 30 rather than that of
12. This would require reversal of the roles of the HDA reacting
partners, such that the aldehyde dienophile would contribute
the C1 carbon of the pyran ring. In this context, we were
particularly intrigued by the possibility of using the novel
dienyne 29 in a HDA reaction with 10 to begin the synthesis
of the right-hand pyran (Scheme 7). If successful, the reaction
would afford cycloadduct 30, which bears the entire carbon
framework of the right-hand fragment with the exception of
the epoxide methylene unit. Also, the C4 hydroxyl could be
installed directly from the silyl enol ether intermediate using a
Rubottom oxidation, thereby avoiding the difficulties outlined
above in manipulations of 12. Finally, the C17 silyl ether would
provide a functional handle to incorporate the hemiketal through
an elimination/hydration sequence.
Reaction between the dienyne 2924 and aldehyde 10 in the
presence of catalyst (1R,2S)-14a led to formation of the highly
functionalized pyran 30 in >99% ee (Scheme 8). However, this
transformation proceeded slowly (>40 h), and 30 was isolated
in <50% yield.25 A switch to SbF6 catalyst (1R,2S)-14b led to
improved reactivity and product yield while maintaining high
ee (98%). As anticipated, Rubottom oxidation of 30 afforded
the R-hydroxy ketone directly. A mixture of epimers was
obtained initially, but this could be subjected to equilibration
with K2CO3 in MeOH to afford the desired isomer exclusively.26
It was somewhat surprising and certainly fortunate that no
transposition of the ketone and hydroxyl group was observed
during this equilibration process. Following protection of the
free hydroxyl group and methylenation of the ketone, selective
reaction of the terminal alkyne with Schwartz’s reagent and
(24) Dienyne 30 was synthesized from trimethylsilyl propynal in three
steps with an overall yield of 50%. See Supporting Information.
(25) For most cases examined, nearly quantitative yields are observed
in HDA reactions with the tridentate Schiff base-chromium catalysts.
However, both the dienyne 30 and HDA adduct 31 displayed only marginal
stability, and this fact, combined with the sluggishness of this particular
reaction, accounts for the modest yield in this reaction. No other products
were observed. Results similar to those obtained with SbF6 catalyst (1R,2S)-
14b could be produced with the chloride catalyst (1R,2S)-14a at ∼50 °C
(50-60% isolated yields, 96-98% ee).
(26) On a preparative scale (5-10 mmol), best results were obtained by
subjecting the HDA reaction mixture to quick filtration through a pad of
silica gel to remove sieves and catalyst. The crude HDA adduct was then
subjected to Rubottom oxidation and equilibration with K2CO3. Using this
procedure, a 30% yield for the three-step sequence could be obtained
reproducibly.
(23) (a) Larock, R. C.; Hightower, T. R.; Kraus, G. A.; Hahn, P.; Zheng,
D. Tetrahedron Lett. 1995, 36, 2423-2426. (b) Ito, Y.; Hirao, T.; Saegusa,
T. J. Org. Chem. 1978, 43, 1011-1013.

Study of the Natural Product FR901464
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9979
Scheme 9. Unsuccessful Coupling Attempts
trapping of the resulting vinylzirconium intermediate with NIS
afforded vinyl iodide 33.
At this stage, an elimination/hydration sequence was at-
tempted in order to install the hemiketal unit. The primary
hydroxyl group was deprotected and converted to the iodide.
However, prolonged treatment with DBU to effect elimination
to the exocyclic enol led to spontaneous migration of the double
bond to the endocyclic position to provide the undesired isomer
35. Since enol ether isomerization was most likely driven by
formation of the conjugated diene system, we evaluated whether
this problem might be circumvented by installation of the
epoxide functionality prior to the elimination event. This strategy
would require that the potentially sensitive epoxide be carried
through several steps of the synthesis, including the critical
fragment-coupling events. On the other hand, if successful, this
rather risky approach would allow a very direct route to the
right-hand fragment. Deprotection of the TIPS group in 34 and
directed epoxidation27 of the alkene provided the epoxide as a
single diastereomer. The free hydroxyl group was reprotected
as a TES ether to provide 36, and this compound was subjected
to reaction with DBU in order to effect elimination of the alkyl
iodide. The epoxide withstood the lengthy reaction time (48 h)
that was required, and no isomerization of the enol ether to the
endocyclic position was observed. Intermediate 37 was isolated
in good yield and was readily purified by chromatography on
silica gel and fully characterized. Deprotection of the epoxy
alcohol and hydration of the enol ether with aqueous p-
toluenesulfonic acid provided 4, completing the synthesis of
the fully functionalized right-hand pyran fragment. Although
the compatibility of the epoxide with the conditions necessary
for fragment coupling had not yet been addressed, this func-
tionality had already displayed impressive robustness toward
both basic and acidic reaction conditions, and its early instal-
lation had facilitated the challenging elimination/hydration
sequence to introduce the hemiketal.
(II) Fragment Couplings and Completion of the Synthesis
of FR901464. With synthetic access to all three fragments of
FR901464 in hand, our efforts turned toward examination of
coupling strategies. The relatively high degree of substitution
in the conjugated dienyl unit of 1 would likely require highly
reactive partners for the cross-coupling reaction. On the other
hand, if the fragment assembly were to be executed at a late
stage in the synthesis in order to maximize convergency, a high
degree of functionalization in both pyran rings would have to
be tolerated in the diene synthesis. Palladium-catalyzed Negishi
coupling protocols generally display a balance of high reactivity
and good functional group tolerance, and these were identified
as most promising for linking the two pyran fragments.
Hydrozirconation of the central fragment alkyne with Schwartz’s
reagent under equilibrating conditions was expected to proceed
both stereospecifically and regioselectively to provide the desired
vinylzirconium species (eq 3).28 In initial studies it was found,
Scheme 10. Stereo- and Regioselective Synthesis of the
Central Fragment as Vinyl Iodide Derivative 38
tempted coupling to model system 33 under standard Negishi-
type conditions did not yield any desired product, even with
the hydroxyl group of 25 protected (Scheme 9). This was quite
unexpected in light of the fact that Theodorakis et al. had carried
out a hydrozirconation/Negishi coupling sequence successfully
on a closely related system in the context of their synthesis of
reveromycin B.29
These findings prompted us to redesign the cross-coupling
reaction by introducing the central fragment as a vinyl iodide
and the right-hand pyran as the vinylzinc component. This
reversal of coupling partners was anticipated to have two
significant advantages. First, hydrozirconation of the terminal
alkyne in 32 (Scheme 8), a precursor to the right-hand pyran
fragment, had been shown to proceed very cleanly under mild
conditions. A high-yielding hydrometalation such as this was
clearly desirable for the first step in an intricate coupling
sequence involving multiple transformations. In contrast, treat-
ment of the central fragment with Schwartz’s reagent under more
forcing conditions was found to generate several detectable
byproducts. These impurities were possibly responsible for the
failure of the coupling attempts outlined in Scheme 9. Formation
of an isolable vinyl iodide intermediate (e.g., 38, Scheme 10)
would allow removal of these byproducts in a purification step.
Second, because the Cp2Zr(H)Cl would be consumed by the
point in the reaction at which the vinyl iodide was added, the
central fragment could be elaborated to azide 39 prior to the
cross-coupling event. This would entail a more convergent
approach to the synthesis of FR901464.
Hydrozirconation of 25a with excess Cp2Zr(H)Cl under
equilibrating conditions followed by trapping the vinylzirconium
intermediate with I2 afforded the desired vinyl iodide 38 in 65%
isolated yield (Scheme 10).30 As anticipated, installation of the
azide required careful optimization of the leaving group,
nucleophile, and reaction conditions. It was found that the azide
substitution product 39 could be accessed in 55% yield by
conversion of alcohol 37 to the monochloromethanesulfonate
not unexpectely, that the azide did not survive treatment with
excess Schwartz’s reagent in THF at elevated temperatures. For
this reason, we endeavored to link the two fragments prior to
azide installation. However, hydrozirconation of 25 and at-
(29) Drouet, K. E.; Theodorakis, E. A. J. Am. Chem. Soc. 1999, 121,
456-457.
(30) None of the undesired vinyl iodide regioisomer was obtained. Some
more polar byproducts were isolated, but their structure was not completely
determined.
(27) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. ReV. 1993, 93,
1307-1370.
(28) Hu, T.; Panek, J. S. J. Org. Chem. 1997, 62, 4912-4913.

9980 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Thompson et al.
Scheme 11. Attempted Fragment Coupling/Epoxidation
Sequence
Figure 2. Open-chain form of FR901464.
Scheme 12. Cross-Coupling of the Fully Functionalized
Pyran Units
Scheme 13. Completion of the Synthesis
derivative,31 followed by displacement with LiN3 in DMPU.32
Cross-coupling was then accomplished by hydrozirconation of
32, transmetalation to Zn, and reaction with vinyl iodide 39 in
the presence of Pd(PPh3)4 (Scheme 11). The reaction proceeded
smoothly to produce 41 in an unoptimized yield of 46%.
However, attempted installation of the spiro epoxide via directed
epoxidation reaction proved unsuccessful. Competing oxidation
of the homoallylic conjugated double bond led to side products,
and the desired allylic epoxide was generated only in low yield.
Faced with this difficulty in installing the epoxide after
fragment coupling, we ventured to attempt the coupling reaction
with the epoxide already in place. Although we could find no
precedent for selective hydrozirconation of an alkyne in the
presence of an epoxide, we relied on the very high reactivity of
Schwartz’s reagent toward terminal alkynes. Indeed, hydrozir-
conation of 42, followed by coupling to vinyl iodide 38 under
standard Negishi conditions, provided the desired highly func-
tionalized diene 43 in excellent yield (Scheme 12).
At this point, completion of the synthesis of FR901464
required only installation of the hemiketal and azide reduction
followed by acylation of the resulting amine with side chain 2.
The ordering of these events was critical, and it proved
preferable to introduce the side chain prior to hemiketal
formation. The concern with such a sequence was that epimer-
ization of the C4′ stereocenter or acetate elimination might occur
during the lengthy reaction time required for iodide elimination.
Fortunately, model system 44 was recovered in nearly quantita-
tive yield with no deterioration of enantiopurity when subjected
to the reaction conditions used for iodide elimination. With this
information in hand, completion of the synthesis of FR901464
proved straightforward (Scheme 13). Reduction of azide 43 with
Me3P33 followed by amide bond formation using standard
coupling conditions provided 45 in good yield. An elimination/
deprotection/hydration sequence was then used to install the
hemiketal and complete the synthesis of FR901464. 1H and 13
C
NMR spectra of this synthetic material matched those of the
natural product provided by Dr. H. Nakajima (Fujisawa). HRMS
and optical rotation data also corresponded with those of the
1
natural product. It was interesting to note that the H NMR
spectra of both synthetic and natural FR901464 displayed a
series of minor peaks (∼10%), consistent with the presence of
a minor isomer of the natural product.34 Evidence was obtained
subsequently (vide infra) that this minor isomer is the open-
chain ketone form of the molecule (1′, Figure 2).
(31) Shimizu, T.; Ohzeki, T.; Hiramoto, K.; Hori, N.; Nakata, T. Synthesis
1999, 1373-1385.
(III) Confirmation of the Structure of FR901464. Despite
the fact that NMR and optical rotation data for synthetic 1
(32) The use of mesylate as a leaving group provided only traces of the
desired product. The triflate was unstable and afforded 20-30% yield of
the desired product. Low yields were obtained with the nosylate leaving
group. LiN3 was superior to NaN3. DMPU was superior to move traditional
solvents for azide displacements such as DMF and DMSO, which gave
very low yields of the desired products. Elimination (35%) was a competing
side reaction.
(33) Knapp, S.; Jaramillo, C.; Freeman, B. J. Org. Chem. 1994, 59,
4800-4804. Me3P was used because reduction of the sterically hindered
azide did not reach completion after several days in the presence of Ph3P.
1
(34) The presence of ∼10% of a minor isomer was also seen in the H
NMR of 4.

Study of the Natural Product FR901464
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9981
Table 1. Proof of Structure
a
compound
1H NMR spectrum
[R]23
D
1
46
47
matches natural product
differs from natural product
matches natural product
-13.0
+12.7
- 69.4
Figure 3. Analogues of FR901464 synthesized and assayed.
a All measurements made in CH2Cl2. Lit.1 [R]23 for FR901464 )
(IV) Synthesis and Evaluation of Analogues of FR901464.
In light of the impressive biological activity and intriguing mode
of action reported for 1, we decided to synthesize several
nonisomeric analogues of the natural product and evaluate them
in a biological assay along with the natural and unnatural
diastereomers of FR901464. In particular, compounds 48-50
(Figure 3) were chosen as targets for synthesis in order to probe
the importance of certain structural features of FR901464. We
hoped that the information gathered in studying analogues of 1
would lay the groundwork for preparation of simpler compounds
that retain the biological activity of FR901464 and can be used
as biological tools. Also, we envision future studies to discern
the as yet unidentified cellular target of FR901464 using affinity
chromatography, or by attachment of a fluorescent or photoaf-
finity label to the natural product. To attempt these types of
experiments, it is important to find a region of FR901464 that
can be modified without causing a deleterious effect on
biological activity.
Compound 48 was synthesized in the same manner as
FR901464, except that following azide reduction the amine was
simply acetylated with Ac2O.37 This compound was designed
to test the importance of the side chain for biological activity.
Clearly, if the side chain proved unnecessary, the amine would
provide a convenient functional handle for use in biological
assays. Deshydroxy compound 49 was synthesized by the route
depicted in Scheme 14, the main deviation from the synthesis
of 1 being the reduction of the iodide in 51 to a methyl group
using NaBH4 in DMPU. Compound 49 was of interest to probe
the importance of the hemiketal, which allows natural FR901464
to access minor isomers (e.g., open chain or furanose) that may
be of biological importance. Also, 49 was expected to circum-
vent the instability observed in FR901464 that is attributable
to the hemiketal.38 Finally, compound 50 was synthesized to
probe the role of the epoxide (Scheme 15). The cyclopropyl
group was installed via a Simmons-Smith reaction, and the
remainder of the synthesis paralleled that of FR901464. Upon
completion of the synthesis of 50, we were surprised to find
that the ratio of 50 to the minor isomer with which it was in
equilibrium was ∼1.5:1 in CD2Cl2 rather than the ∼10:1 ratio
observed for FR901464 (vide supra). On the basis of spectro-
scopic evidence, it appears that the minor component is the
open-chain ketone form of the molecule.39 The difference in
the ratio of isomers was surprising, as we had expected the
cyclopropane to enforce a conformation in the right-hand pyran
D
-12 (c 0.5, CH2Cl2).
matched those of natural FR901464, the possibility remained
that we had in fact prepared a diastereomer of natural product
comprised of fragments with the same relative but different
absolute stereochemistry. As noted in the Introduction, in the
original structural elucidation of FR901464 the absolute ster-
eochemistry at C4′ had been established with certainty by
degradation of the natural product, but only the relative
stereochemistry was reported for the pyran rings.9a,b Because
the three chiral components of FR901464 are isolated each from
the other by planar, conjugated linker elements, we could not
rule out that diastereomers containing fragments of different
absolute but identical relative stereochemistries might give rise
to indistinguishable NMR spectra. To ascertain that we had
indeed prepared FR901464 and with the goal of making a
corresponding, unambiguous stereochemical assignment of the
natural product, we prepared compounds 46 and 47 (Table 1).35
Although compound 46 bears the unnatural stereochemistry at
C4′, and therefore could not have been the correct stereoisomer,
it was examined in order to assess whether FR901464 might
have the structure of ent-46. Although the optical rotation of
46 matched that of the natural product (with opposite sign), the
1H NMR of 46 displayed significant differences from that of
the natural product. In particular, the NH proton was shifted
downfield by 0.15 ppm while the C4′ proton was shifted upfield
by 0.15 ppm relative to their positions in the spectra of natural
FR901464. From this information, it was clear that synthetic 1
had the correct relative stereochemical relationship between the
left-hand and central fragments. It remained to be shown that
the right-hand pyran of synthetic 1 had the correct absolute
1
stereochemistry.36 The H NMR spectrum of compound 47
matched that of FR901464, but the observed optical rotation
differed significantly from that of the natural product. This
established that 47 is not FR901464 and allowed us to verify
that synthetic 1, prepared as described, is the same compound
as FR901464 reported by the Fujisawa Pharmaceutical Co. in
1996.
(35) These compounds were prepared using the chemistry described for
the synthesis of FR901464 but employing enantiomeric catalysts for the
HDA or Noyori transfer hydrogenation as needed.
(36) We did not evaluate the diastereomer of 1 bearing only a central
fragment of opposite absolute stereochemistry (i.e., ent-3) as that assigned
to FR901464, but it can be ruled out as a possibility, nonetheless. This is
due to the fact that the relatiVe stereochemistry between the left-hand and
central fragments is established by evaluation of 47. In order for the structure
bearing ent-3 to be correct, it would require that the chemical shift changes
observed in 47 for the NH and C4′ protons relative to FR901464 be canceled
out exactly by switching the configuration of the right-hand fragment.
(37) See Supporting Information for details.
(38) FR901464 could be isolated by purification on preparative TLC,
but significant decomposition was observed when purification by flash
chromatography was attempted. FR901464 is unstable in MeOH and in
CHCl3.

9982 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Scheme 14. Synthesis of 49
Thompson et al.
Table 2. Activity of 1 and Synthetic Analogues in Cell-Killing
Experimentsa
compound
IC50 (ng/mL)
IC50 (nM)
1
46
47
48
49
50
1 ( 0.2
15 ( 5
200 ( 100
700 ( 200
0.8 ( 0.2
>2000
2
30
400
1700
1.6
>4000
a Compounds were administered as THF solutions and assayed for
their ability to inhibit the growth of Tag Jurkat cells. See text and
Supporting Information for details.
cells were treated with THF solutions of the compounds shown
in Table 2, while control Tag Jurkat cells were treated with
THF alone; cells treated with compounds and control cells were
counted after 48 h. IC50 values were determined by treating the
cells with solutions of the compounds being tested over a range
of concentrations.41 It is important to note that this assay
involves the ability of these compounds to inhibit the growth
of whole cells and that changes in biological activity may be
unrelated to the ability of the compounds to interact with the
target of FR901464. Other factors, such as solubility, ease of
entry into the cells, stability, and even changes in the mechanism
of action, may be responsible for observed gains or losses in
activity with respect to the natural product. Such factors are
hard to assess in such a general assay, and, although interesting,
the biological data presented here must be interpreted with
caution. The IC50 of FR901464 was determined to be 1 ng/mL,
a value which corresponds closely to IC50 values reported by
Fujisawa scientists for FR901464 against several cancer cell
lines.1 Diastereomers 46 and 47 are less active than 1 by 1 and
2 orders of magnitude, respectively. This drop in biological
activity further serves to confirm the correct structure of
FR901464. In fact, the biological activity observed for 47 is
very possibly due to contamination with a small amount of the
natural diastereomer 1, because the right-hand ring used in the
synthesis of this compound is only of 98-99% ee.
Scheme 15. Synthesis of 50
The side chain of FR901464 does appear to be key for
biological activity, as a nearly 3 orders of magnitude drop in
the ability to inhibit cell growth was observed when the side
chain was replaced by an acetate group (compound 48). Also,
replacement of the epoxide with a cyclopropyl moiety (com-
pound 50) led to a complete loss in biological activity at the
highest concentrations tested, demonstrating the importance of
the epoxide functionality in the biological activity of 1.42 In
(39) The model system depicted below was synthesized and also existed
as a ∼1.5:1 ratio of isomers, with the major one being the one drawn. The
IR spectrum contained a band at 1705 cm-1, indicating the presence of a
carbonyl group. Also, the 1H NMR of this compound displayed signals for
the C-17 methyl group and both C-2 protons that were shifted downfield
by 0.8 ppm in the minor isomer. These facts are consistent with the minor
isomer being the open-chain ketone form.
(40) We were unable to obtain the M-8 cell line from Fujisawa and were
unsuccessful in developing a more specific assay based on up-regulation
of the SV40 gene using transient transfection (see ref 4). Tag Jurkat cells
are a human T cell lymphoma expressing the T antigen (Tag). See:
Northrop, J. P.; Ulman, K. S.; Crabtree, G. R. J. Biol. Chem. 1993, 268,
2917-2923.
similar to that seen with the epoxide. The shift in equilibrium
could possibly be explained by an intramolecular H-bond
between the hydroxyl group of the hemiketal and the epoxide,
and this stabilization may help to favor the cyclic structure.
The results of a biological assay testing the effect of
FR901464, its diastereomers, and the other synthetic analogues
on Tag Jurkat cells are summarized in Table 2.40 Tag Jurkat
(41) The IC50 is the minimum concentration of compound that causes
cells to grow at e50% the rate of control cells.
(42) FR901463 bears a chlorohydrin in place of the epoxide (ref 1) and
may be functionally equivalent.

Study of the Natural Product FR901464
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9983
contrast, desoxy analogue 49 retained fully the activity of the
parent molecule. This result establishes that the closed tetrahy-
dropyran ring is sufficient for activity, and it suggests that the
minor, open-chain ketone isomer is not responsible for the
cytotoxicity of FR901464. The fact that the hemiketal is not a
critical feature of FR901464 points to this terminus of the
molecule as a site for coupling to a resin or attachment of a
fluorescent tag or photoaffinity label.
either enantiomer of each chiral fragment allowed us to
synthesize a series of compounds that proved the relative and
absolute stereochemistry of 1. Removal of the hemiketal
functionality of FR901464 resulted in a new compound (49)
that is more stable than 1, exists as a single isomeric form, and
retains the biological activity of the natural product. Thus, the
right-hand terminus of FR901464 has been identified as a site
for the types of modifications that may prove useful in future
biological studies. We hope this will enable the discovery of
the cellular target of 1, a worthwhile goal given the potent
biological activity and striking phenotypic changes induced by
FR901464.
Conclusions
The total synthesis of FR901464 showcases the power of
asymmetric catalytic reactions as a core strategy in natural
products synthesis, and also illustrates how a target can give
rise to an entirely new enantioselective reaction methodology.
The discovery of novel tridentate chromium catalyst 14 was
inspired and guided by the quest for an efficient route to the
central fragment of 1. In the end, asymmetric hetero-Diels-
Alder reactions with highly functionalized, alkyne-containing
reaction partners led to efficient syntheses of both the central
and right-hand pyran fragments of FR901464. Finally, high-
yielding coupling reactions allowed completion of the conver-
gent synthesis.
Acknowledgment. This work was supported by the NIH
(GM-59316) and by a postdoctoral fellowship to T.F.J. from
the Cancer Research Fund of the Damon Runyon-Walter
Winchell Foundation (DRG-1431). We thank Dr. H. Nakajima
(Fujisawa) for generously providing spectra of 1 and D. Schmidt
(Harvard) for help with biological assays.
Supporting Information Available: Complete experimental
procedures and characterization data, and NMR spectra of
synthetic and natural FR901464 (1) and 46-50 (PDF). This
material is available via the Internet at http://pubs.acs.org.
The power of the chiral building block strategy and the
flexibility of this particular synthetic route were further dem-
onstrated in the preparation of analogues for proof of the
FR901464 structure and for biological studies. Ready access to
JA016615T