Insights into long-range structural effects on the stereochemistry of aldol condensations: A practical total synthesis of desoxyepothilone F

Article abstract of DOI:10.1021/ja010039j

A processable total synthesis of a potent antitumor agent, desoxyepothilone F (dEpoF, 21-hydroxy-12,13-desoxyepothilone B, 21-hydroxyepothilone D), has been accomplished. The route is highly convergent. The new technology has also been applied to a total

Full text of DOI:10.1021/ja010039j

J. Am. Chem. Soc. 2001, 123, 5249-5259
5249
Insights into Long-Range Structural Effects on the Stereochemistry of
Aldol Condensations: A Practical Total Synthesis of
Desoxyepothilone F
Chul Bom Lee,† Zhicai Wu,§ Fei Zhang,§ Mark D. Chappell,† Shawn J. Stachel,†
Ting-Chao Chou,‡ Yongbiao Guan,‡ and Samuel J. Danishefsky*,†,§
Contribution from The Laboratories for Bioorganic Chemistry, and Preclinical Pharmacology,
The Sloan-Kettering Institute for Cancer Research, 1275 York AVenue, New York, New York 10021, and
The Department of Chemistry, Columbia UniVersity, HaVemeyer Hall, New York, New York 10027
ReceiVed January 3, 2001
Abstract: A processable total synthesis of a potent antitumor agent, desoxyepothilone F (dEpoF, 21-hydroxy-
12,13-desoxyepothilone B, 21-hydroxyepothilone D), has been accomplished. The route is highly convergent.
The new technology has also been applied to a total synthesis of 12,13-desoxyepothilone (dEpoB). The crucial
point of departure from previous syntheses of dEpoB and dEpoF involves presentation of the C1-C11 sector
for Suzuki coupling with C3 in reduced form. Hitherto, the required S stereochemistry at C3 had been
implemented via reduction of a keto function after Suzuki coupling. Whereas that chemistry worked quite
well in a synthesis of dEpoB, it was not transferable to a high-yielding synthesis of dEpoF. The reduction of
the keto group at C3 via a Noyori protocol after Suzuki coupling had proved to be very difficult. In our
current approach, two consecutive aldol reactions are used to fashion the acyl sector. In the first aldol
condensation, C6 becomes attached to C7. Following protection at C7, a two-carbon acetate equivalent is used
to join C2 and C3 with very high asymmetric induction at C3. Only after this center has been implemented is
the Suzuki reaction conducted. This major advance allowed us to synthesize dEpoF in a straightforward fashion.
These findings found ready application in the total synthesis of dEpoB. Another part of the study involved
analysis of the factors associated with aldol condensations joining C6 to C7. In the work described herein, the
consequences of the status of C3 in promoting the C6-C7 aldol coupling are probed in detail. Dramatic
stereochemical long-range effects uncovered during the study are described, and a working model to explain
these effects has emerged.
Introduction
their own risks and management issues.8 Moreover, Taxol is
vulnerable to deactivation through multiple drug resistance
(MDR). Resistance apparently arises through overexpression of
P-glycoprotein (P-gp) and via tubulin mutation at the binding
site.9
Among naturally occurring microtubule modulators, the
recently discovered epothilones have evoked much excitement
among scientists in various disciplines. These cytotoxic mac-
rolides originated from a cellulose-degrading mycobacteria of
the genus Sorangium and were found to possess remarkable
antitumor activities.10 In particular, epothilone B (1b, EpoB)
The clinical successes of paclitaxel (Taxol) and docetaxel
(Taxotere) in the treatment of cancer has spurred searches for
other agents that inhibit the growth of tumors through similar
mechanisms of action.1 The cytotoxicity of Taxol (and possibly
its mode of action at the level of clinical application) arises
from its ability to induce apoptosis via tubulin polymerization
and stabilization of microtubule assemblies.2 The tubulin-based
mechanism of action, originally discovered by Horwitz,3 has
been recognized to be operative in several naturally occurring
nontaxoids such as discodermolide,4 eleuthesides,5 laulimalides,6
and epothilones.7
(4) Gunasakera, S. P.; Gunasakera, M.; Longley, R. E.; Schulte, G. K.
J. Org. Chem. 1990, 55, 4912.
(5) Lindel, T.; Jensen, P. R.; Fenical, W.; Long, B. H.; Casazza, A. M.;
Carboni, J.; Fairchild, C. R. J. Am. Chem. Soc. 1997, 119, 8744.
(6) Moobery, S. L.; Tien, G.; Hernandez, A. H.; Pluburukarn, A.;
Davidson, B. S. Cancer Res. 1999, 59, 653.
(7) (a) Ho¨fle, G. H.; Bedorf, N.; Steinmetz, H.; Schomburg, D.; Gerth,
K.; Reichenbach, H. Angew. Chem., Int. Ed. Engl. 1996, 35, 1567. (b) Gerth,
K.; Bedorf, N.; Ho¨fle, G. H.; Irschik, H.; Reichenbach, H. J. Antibiot. 1996,
49, 560.
(8) (a) Rowinsky, E. K.; Eisenhauer, E. A.; Chaudhry, V.; Arbuck, S.
G.; Donehawer, R. C. Semin. Oncol. 1993, 20, 1. (b) Fletcher, B. S.;
Kujubadu, D. A.; Perrin, D. M.; Herschman, H. R. J. Biol. Chem. 1992,
267, 4338. (c) Tsuji, M.; Dubois, R. N. Cell 1995, 3, 493. (d) Essayan, D.
M.; Kagey-Sobotka, A.; Colarusso, P. J.; Lichtenstein, L. M.; Ozols, R. F.;
King, E. D. J. Allergy Clin. Immunol. 1996, 97, 42.
While of unquestioned clinical value, Taxol is far from being
an ideal drug. Its marginal aqueous solubility necessitates
recourse to formulation vehicles such as cremophores that pose
† Laboratory for Bioorganic Chemistry, The Sloan-Kettering Institute for
Cancer Research.
‡ Laboratory for Preclinical Pharmacology, The Sloan-Kettering Institute
for Cancer Research.
§ Columbia University.
(1) For reviews, see: (a) Taxane Anticancer Agents; Georg, G. I., Chen,
T. T., Ojima, I., Vyas, D. M., Eds.; American Chemical Society: San Diego,
1995. (b) The Chemistry and Pharmacology of Taxol and its DeriVatiVes;
Farin, V., Ed.; Elsevier: New York, 1995. (c) Taxol: Sciences &
Applications; Suffness, M., Ed.; CRC Press: Boca Raton, FL, 1995.
(2) Hyams, J. F.; Loyd, C. W. Microtubules; Wiley-Liss: New York,
1993.
(9) Giannakakou, P.; Sackett, D. L.; Kang, Y.-K.; Zhan, Z.; Buters, J.
T.; Fojo, T.; Poruchynsky, M. S. J. Biol. Chem. 1997, 272, 17118 and
references therein.
(3) Horwitz, S. B.; Fant, J.; Schiff, P. B. Nature 1979, 277, 665.
10.1021/ja010039j CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/10/2001

5250 J. Am. Chem. Soc., Vol. 123, No. 22, 2001
Lee et al.
dissection of epothilone bioactivity patterns into discrete zones
of varying tolerances to structural modification.14
Using totally synthetic derived drug, our laboratory was the
first to conduct in vivo evaluations of epothilone B.15 When it
was found that EpoB itself exhibited a worrisome toxicity profile
(which to our estimation could result in an unacceptably narrow
therapeutic index), we turned to the 12,13-desoxy compound
in the hope of providing greater margins of potential opportunity.
From combined total synthesis and SAR research, it had been
shown that dEpoB is significantly less potent in vitro than EpoB.
It is nonetheless, highly cytotoxic and robust with respect to
MDR deactivation.Our original rationale for pursuing the 12,-
13-deoxy congeners at the time arose from a conjecture that
the enormous cytotoxicity of EpoB itself may be a consequence
of two components. With the epoxide intact, there could well
be a general cytotoxicity component (perhaps arising from the
capacity of the system to function as a DNA alkylating agent).
In addition we theorized that there might be a more specific
mode of action (for instance, tubulin binding) which could be
of potential therapeutic value against transformed cells and
perhaps tumor masses. Our hope was that the nontumor-specific
cytotoxicity, vested in the epoxide linkage of EpoB, would be
substantially abrogated in dEpoB which would nonetheless
retain the mechanistically specific tumor-directed cytotoxicity.
Figure 1. Structure of epothilones.
exhibits significantly higher cytotoxicity than does Taxol against
various cancer cell lines. Furthermore, epothilones are apparently
not susceptible to deactivation through MDR and maintain
activity against resistant tumor cell lines.
We note that epothilones have also recently been constructed
by heterologous expression of the appropriate biosynthesis genes
in productive microorganisms.11 The problem of epothilone
supply may well be resolved by conventional fermentation
augmented by emerging biotechnology. It is also possible that
eventually, other useful analogues could be produced by this
biotransformation protocol (Figure 1).
Given the promising in vitro data associated with epothilones,
given several challenging features in their molecular frameworks
vis-a`-vis chemical synthesis, and given the uncertainties as to
the availability of homogeneous epothilones in bulk via
fermentation, these drugs have attracted significant attention
from the community of synthetic chemists. Indeed, a variety of
total syntheses have been accomplished.12 Our exercises, initially
launched from the perspective of total synthesis,13 enabled a
rather detailed SAR mapping of the drugs and allowed for
Indeed, in vivo experiments based on various mouse models
have consistently demonstrated that dEpoB possesses remarkable
therapeutic potential and is essentially curative against various
sensitive and resistant tumors in xenografts. Due to its impres-
sive in vivo profile, dEpoB has been advanced through
toxicology evaluations in dogs, in expectation of human trials
anticipating its deployment as an anticancer drug.16 The excellent
preclinical in vivo successes realized with dEpoB do not
necessarily prove the validity of our dissection of its cytotoxicity
into a tumor selective component as well as nonspecific toxicity
arising from the epoxide linkage. At the very minimum, this
hypothesis served as a valuable working model in prompting
our interest in the 12,13-desoxy series. The full accounting of
the superior performance of dEpoB remains an intriguing
subject, and clearly merits further investigation.
In building upon the “12,13-desoxy concept” it was appropri-
ate to evaluate the in vivo efficacy of other such agents. In this
spirit we pursued 21-hydroxylated versions of 12,13-desoxy-
epothilones such as dEpoE (1c) and dEpoF (1d).17 As indicated
by in vitro assays, 21-hydroxylated epothilones such as EpoE
and EpoF do not forfeit the activities of the A and B systems.18
Since EpoB is significantly more potent than EpoA, we naturally
placed dEpoF at a higher priority than dEpoE. Furthermore,
we anticipated that the 21-hydroxyl group of dEpoF might
provide advantageous properties relative to the 21-methyl group
of dEpoB in terms of aqueous solubility and might serve as a
handle for further chemical elaborations. While a semi- and total
(10) (a) Bollag, D. M.; McQueney, P. A.; Zhu, J.; Hensens, O.; Koupal,
L.; Liesch, J.; Goetz, M.; Lazarides, E.; Woods, C. M. Cancer Res. 1995,
55, 2325. (b) Kowalski, R. J.; Terhaar, E.; Longley, R. E.; Gunasekera, S.
P.; Lin, C. M.; Day, B. V.; Hamel, E. Mol. Biol. Cell 1995, 6, 2137.
(11) (a) Tang, L.; Shah, S.; Chung, L.; Carney, J.; Katz, L.; Khosla, C.;
Julien, B. Science 2000, 287, 640. (b) Molnar, I.; Schupp, T.; Ono, M.;
Zirkle, R. E.; Milnamow, M.; Nowak-Thompson, B.; Engel, N.; Toupet,
C.; Stratmann, A.; Cyr, D. D.; Gorlach, J.; Mayo, J. M.; Hu, A.; Goff, S.;
Schmid, J.; Ligon, J. M. Chem. Biol. 2000, 7, 97. (c) Julien, B.; Shah, S.;
Ziermann, R.; Goldman, R.; Katz, L.; Khosla, C. Gene 2000, 249, 153.
(12) (a) Yang, Z.; He, Y.; Vourloumis, D.; Vallberg, H.; Nicolaou, K.
C. Angew. Chem., Int. Ed. Engl. 1997, 36, 166. (b) Nicolaou, K. C.; Sarabia,
F.; Ninkovic, S.; Yang, Z. Angew. Chem., Int. Ed. Engl. 1997, 36, 525. (c)
Nicolaou, K. C.; He, Y.; Vourloumis, D.; Vallberg, H.; Roschangar, F.;
Sarabia, F.; Ninkovic, S.; Yang, Z.; Trujillo, J. I. J. Am. Chem. Soc. 1997,
119, 7960. (d) Nicolaou, K. C.; Winssinger, N.; Pastor, J.; Ninkovic, S.;
Sarabia, F.; He, Y.; Vourloumis, D.; Yang, Z.; Li, T.; Giannakakou, P.;
Hamel, E. Nature 1997, 387, 268. (e) Nicolaou, K. C.; He, Y.; Roschangar,
F.; King, N. P.; Vourloumis, D.; Li, T. H. Angew. Chem., Int. Ed. 1998,
37, 84. (f) Schinzer, D.; Limberg, A.; Bauer, A.; Bohm, O. M.; Cordes, M.
Angew. Chem., Int. Ed. Engl. 1997, 36, 523. (g) Schinzer, D.; Bauer, A.;
Bohm, O. M.; Limberg, A.; Cordes, M. Chem. Eur. J. 1999, 5, 2483. (h)
Schinzer, D.; Bauer, A.; Schieber, J. Chem. Eur. J. 1999, 5, 2492. (i) May,
S. A.; Grieco, P. A. Chem. Commun. 1998, 1597. (j) Sawada, D.; Shibasaki,
M. Angew. Chem., Int. Ed. 2000, 39, 209. (k) Sawada, D.; Kanai, M.;
Shibasaki, M. J. Am. Chem. Soc. 2000, 122, 10521. (l) Martin, H. M.;
Drescher, M.; Mulzer, J. Angew. Chem., Int. Ed. 2000, 39, 581. (m) White,
J. D.; Carter, R. G.; Sundermann, K. F. J. Org. Chem. 1999, 64, 684. (n)
White, J. D.; Sundermann, K. F.; Carter, R. G. Org. Lett. 1999, 1, 1431.
(o) Zhu, B.; Panek, J. S. Org. Lett. 2000, 2, 2575.
(14) (a) Su, D.-S.; Balog, A.; Meng, D.; Bertinato, P.; Danishefsky, S.
J.; Zheng, Y.-H.; Chou, T.-C.; He, L.; Horwitz, S. B. Angew. Chem., Int.
Ed. Engl. 1997, 36, 2093. (b) For comprehensive in vitro SAR results, see:
Nicolaou, K. C.; Roschangar, F.; Vourloumis, D. Angew. Chem., Int. Ed.
1998, 37, 2014.
(15) (a) Chou, T. C.; Zhang, X. G.; Harris, C. R.; Kuduk, S. D.; Balog,
A.; Savin, K. A.; Bertino, J. R.; Danishefsky, S. J. Proc. Natl. Acad. Sci.
U.S.A. 1998, 95, 15798. (b) Chou, T. C.; Zhang, X. G.; Balog, A.; Su, D.
S.; Meng, D. F.; Savin, K.; Bertino, J. R.; Danishefsky, S. J. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 9642.
(16) Harris, C. R.; Danishefsky, S. J. J. Org. Chem. 1999, 64, 8434.
(17) (a) Reichenbach, H.; Ho¨fle,; Gerth, K.; Steinmetz, H. (GBF). WO-A
9822461, 1998; Chem. Abstr. 1998, 129, 5346. (b) Ho¨fle, G. In GBF Annual
Report; Walsdorff, J.-H., Ed.; GBF: Braunschweig, 1997.
(18) Ho¨fle, G.; Glaser, N.; Kiffe, M.; Hecht, H. J.; Sasse, F.; Reichenbach,
H. Angew. Chem., Int. Ed. 1999, 38, 1971.
(13) (a) Balog, A.; Meng, D. F.; Kamenecka, T.; Bertinato, P.; Su, D.-
S.; Sorensen, E. J.; Danishefsky, S. J. Angew. Chem., Int. Ed. Engl. 1996,
35, 2801. (b) Su, D.-S.; Meng, D. F.; Bertinato, P.; Balog, A.; Sorensen, E.
J.; Danishefsky, S. J.; Zheng, Y. H.; Chou, T. C.; He, L. F.; Horwitz, S. B.
Angew. Chem., Int. Ed. Engl. 1997, 36, 757. (c) Meng, D. F.; Bertinato, P.;
Balog, A.; Su, D. S.; Kamenecka, T.; Sorensen, E. J.; Danishefsky, S. J. J.
Am. Chem. Soc. 1997, 119, 10073. (d) Balog, A.; Harris, C. R.; Savin, K.;
Zhang, X. G.; Chou, T.-C.; Danishefsky, S. J. Angew. Chem., Int. Ed. 1998,
37, 2675. (e) Harris, C. R.; Kuduk, S. D.; Balog, A.; Savin, K.; Glunz, P.
W.; Danishefsky, S. J. J. Am. Chem. Soc. 1999, 121, 7050.

Synthesis of Desoxyepothilone F
J. Am. Chem. Soc., Vol. 123, No. 22, 2001 5251
Scheme 1. Previous Total Synthesis of dEpoB and dEpoF
synthesis of the epothilones bearing a 21-hydroxy group has
recently been disclosed,18,19 our interests are focused rather
specifically on homogeneous end products with Z-double bonds
in the B (12-methyl) series. Apparently no such product had
been previously synthesized and evaluated as to usefulness in
chemotherapy. Having no access to such compounds through
fermentation, we were obliged to accomplish the preparation
of dEpoF by total synthesis20 if our proposed studies were to
go forward.
In approaching this goal, the starting point of our thinking
was our semi-practical synthesis of dEpoB,13e which has
produced multigrams of the drug, enabling extensive preclinical
studies.15 Naively, we thought that extension of the dEpoB
strategy could readily deliver the desired 21-hydroxy analogue.
However, as will be discussed later, our initial synthesis suffered
a serious setback due to the unexpected behavior of the
structurally altered thiazole in dEpoF. To understand the nature
of the dEpoF problem as it developed, we first focus on the
key B-alkyl Suzuki reaction by which the C11-C12 bond of
dEpoB had been established.
RuCl}2Cl3] as the catalyst in methanol under strongly acidic
conditions (methanolic HCl). Fortunately, the precursor to
dEpoB proved to be stable to these harsh conditions.
Turning to dEpoF, it did indeed prove possible to synthesize
the required 5. However, following deprotection of the TBS
ether, attempted reduction of the C3 ketone under the conditions
described above, led to extensive competitive solvolysis at C15
with formation of the methyl ether 7. The ratio of 6:7 was not
reproducible, and yields of the desired 6 oscillated in the 25-
50% percent range (Scheme 1).
The precise reasons as to why substitution at C21, by an
electron-withdrawing oxygen-based group, exerted such a
profound effect in favoring methanolysis at C15 are still not
clear. Nonetheless, the concurrent solvolysis reaction constituted
a crippling blow to our plans to produce substantial amounts
of dEpoF by total synthesis. This difficulty became a particularly
pressing matter when the dEpoF, which we did produce, turned
out to be a highly promising anticancer agent on the basis of in
vitro assays as well as in the context of in vivo settings in
xenografts.20
The key coupling step conducted between 3 and 4, maintained
the Z-geometry of the C12-C13 linkage. After the required
Suzuki coupling, the required S-steroechemistry at C3 was
implemented by asymmetric reduction of the less hindered C3
ketone using protocols developed by Noyori.21 To achieve the
reduction of the C3 ketone with hydrogen in the dEpoB syn-
thesis, we required recourse to 5% [Et2NH2][{(R-BINAP)-
Accordingly, it emerged that to reach dEpoF by total
synthesis, it would be necessary to present the acyl sector for
Suzuki coupling with C3 already reduced. Needless to say,
attempts to conduct the Noyori reduction at the stage of 4 were
not possible due to reduction of the double bond. This raised
the significant question as to how we could concisely generate
a structure of the type 9 for Suzuki coupling. Potential solutions
to this problem by recourse to substances of appropriate
configuration arising from the “chiral pool” were entertained.
(19) For synthetic work in the simpler A series, see: (a) Nicolaou, K.
C.; He, Y.; Roschangar, F.; King, N. P.; Vourloumis, D.; Li, T. Angew.
Chem., Int. Ed. Engl. 1998, 37, 84. (b) Nicolaou, K. C.; Hepworth, D.;
King, N. P.; Finlay, M. R. V.; Scarpelli, R.; Pereira, M. M. A.; Bollbuck,
B.; Bigot, A.; Werschkun, B.; Winssinger, N. Chem. Eur. J. 2000, 6, 2783.
(20) Lee, C. B.; Chou, T. C.; Zhang, X. G.; Wang, Z. G.; Kuduk, S. D.;
Chappell, M. D.; Stachel, S. J.; Danishefsky, S. J. J. Org. Chem. 2000, 65,
6525.
(21) For the Noyori reduction, see: (a) Noyori, R. Tetrahedron 1994,
50, 4259. (b) Noyori, R.; Ohkuma, T.; Kitamura, M. J. Am. Chem. Soc.
1987, 109, 5856. For the use of acid catalysts, see: (c) King, S. A.;
Thompson, A. S.; King, A. O.; Verhoeven, T. R. J. Org. Chem. 1992, 57,
6689. (d) Taber, D. F.; Silverberg, L. J. Tetrahedron Lett. 1991, 32, 4227.

5252 J. Am. Chem. Soc., Vol. 123, No. 22, 2001
Lee et al.
Scheme 2. New Synthetic Plan for 12,13-Desoxyepothilone F
However, such forays held out, to us, the prospect of being quite
lengthy. In addition to solving the C3 problem, we also hoped
to deal with several other awkward elements arising in our
previous syntheses of dEpoF and dEpoB. An account of our
findings in accomplishing effective total syntheses of both agents
and our newly gained insights in the fashioning of the C2-C3
and C6-C7 bonds in the acyl sector are provided below.
Synthetic Plan. In planning a strategy to circumvent the
serious problems encountered in our earlier synthesis of dEpoF,
primary focus was placed on the development of a sequence in
which carbon 3 of the coupling component, containing the
terminal vinyl group (and subsequently the boryl function)
would be presented for Suzuki coupling with C3 already in
reduced form and of appropriate (S) configuration. In this way,
the problematic Noyori reduction, which results in concurrent
solvolysis at C15, would be avoided. In the context of such an
effort, we also hoped to develop a more convergent and modular
route to both the acyl and alkyl sectors. These two key fragments
would be merged by a B-alkyl Suzuki coupling reaction and
would be advanced to a macrolactone by a Yamaguchi mac-
rocyclization (in a fashion similar to the previous syntheses of
dEpoB13e and dEpoF20). It was first necessary to gain comfort-
able access to the O-alkyl sector 8. In our earlier efforts, the
route to this compound was lengthy and processable only with
considerable difficulty. As an alternative, we envisioned reach-
ing 8 via condensation of a Horner-like reagent (cf. 10) and
ketone 11. While a phosphine oxide of general type 10 (R )
CH3) had been successfully employed for our syntheses of EpoA
and EpoB,22 de novo synthesis of a C21 functionalized version
would now be required to reach dEpoF. Our approach to ketone
10 stemmed from the recognition that the critical iodoalkene
functionality could be accessed by the stereoselective iodination
of a proper alkyne precursor (vide infra) (Scheme 2). In this
way, the serious problems associated with the complicated and
low-yielding (albeit Z-selective) iodoethylenation practiced
earlier would be resolved. Clearly, we would also require a
maximally concise route to ketone 11. It would be particularly
critical to integrate efficient access to both the C12-C13
iodoalkene and the C15 (protected alcohol) functionalities in
to a program which also deals effectively with the substituted
thiazole issue.23
to exploit the chiral pool of starting materials. Instead, we
anticipated recourse to two aldol condensations eventually
joining keto aldehyde 13, acetate 12 and aldehyde 14. As we
contemplated these two aldol condensations, two obvious
sequencing variations presented themselves. In one case, we
would first couple 13 to 14. The only inducing element to
control the C6-C7 stereochemistry would be the chirality of
aldehyde 14. In our recent synthesis of dEpoB,13b it had been
shown that this R-stereogenic center provides remarkable facial
stereochemical guidance in aldol condensation with a related
achiral enolate. Following suitable protection at C7 and depro-
tection at C3, 9 would be in hand. Needless to say, a device
would be necessary to ensure highly stereoselective formation
of the (S)-configured C3, in the course of the second aldol
condensation.
An alternative possibility contemplated coupling of 12 + 13,
first giving rise to 15. Again, this bond-forming sequence would
require a control element to ensure formation of the required
C3(S) configuration in 15. The second stage, aldol condensation,
would be conducted between 15 and aldehyde 14. In considering
this second aldol condensation we leave unspecified whether
C3 would be protected or unprotected. One influencing element
would be the R-methyl stereogenic center at C6 in 14 in
conjunction with its C10-C11 double bond to control the sense
of the C6-C7 bond formation. We also anticipated that the
configuration at C3 could have consequences for the second
aldol condensation. This type of possibility had intervened in
the Schinzer synthesis of epothilones, although only in a specific
case where C1 was also present in fully reduced form.24 The
findings stemming from our total synthesis plans and a detailed
investigation of the consequences of a prebuilt C3 stereogenic
center on the stereochemical sense of C6-C7 bond formation
will also be related.
A New Synthetic Route to the O-Alkyl Fragment. Our new
synthesis commenced with the preparation of methyl ketone 11
by two independent sequences. The first approach employed
an asymmetric catalytic oxygenation method to install the C15
center and iodoboration of an alkyne to implement the (Z)-alkene
geometry (Scheme 3). In practice, propyne (17) reacted with
B-iodo-9-BBN, and the resultant vinyl borane was added to
methyl vinyl ketone to furnish ketone 18.25 Subsequent treatment
of this compound with TMSI/HMDS afforded an 88:12 mixture
For the synthesis of the O-acyl wing 9, the largest issue was
that of gaining access to the required (S)-configured C3, without
benefit of a C3 ketone reduction (via a Noyori protocol) and
without a lengthy excursion which would be required if we were
(24) (a) Schinzer, D.; Limberg, A.; Bauer, A.; Bohm, O. M.; Cordes,
M. Angew. Chem., Int. Ed. Engl. 1997, 36, 523. (b) Schinzer, D.; Bauer,
A.; Bohm, O. M.; Limberg, A.; Cordes, M. Chem. Eur. J. 1999, 5, 2483.
(c) Schinzer, D.; Bauer, A.; Schieber, J. Chem. Eur. J. 1999, 5, 2492. For
related work see: (d) Nicolaou, K. C.; Hepworth, D.; Finlay, M. R. V.;
King, N. P. Werschkun, B.; Bigot, A. Chem. Commun. 1999, 519. (e)
Reference 12(l). (f) Mulzer, J.; Mantouliolis, A.; O¨ hler, E. J. Org. Chem.
2000, 65, 7456. (g) Mulzer, J.; Montoulidids, A. Tetrahedron Lett. 1996,
37, 9179.
(22) Bertinato, P.; Sorensen, E. J.; Meng, D. F.; Danishefsky, S. J. J.
Org. Chem. 1996, 61, 7998.
(23) This strategy has been successfully applied to the plant-scale
synthesis of dEpoB, see: Chappell, M. D.; Stachel, S. J.; Lee, C. B.;
Danishefsky, S. J. Org. Lett. 2000, 2, 1633.

Synthesis of Desoxyepothilone F
J. Am. Chem. Soc., Vol. 123, No. 22, 2001 5253
Scheme 3. Catalytic Asymmetric Route to Ketone 11a
Scheme 5. Attempted Preparation of the O-Alkyl Winga
a Reagents and conditions: (a) TBSCl, imidazole, DMF, 99%; (b)
i) Dibal-H, CH2Cl2, 65%, ii) CCl4, PPh3 84%; (c) i) HPPh2, n-BuLi,
THF, ii) H2O2, 33%; (d) TBAF, THF, rt, 60%.
a Reagents and conditions: (a) i) 9-BBN-I, hexanes, ii) methyl vinyl
ketone, iii) 3 N NaOH, toluene, 100 °C, 65%; (b) TMSI-HMDS,
CH2Cl2, -20 °C to rt; (c) 1 mol % OsO4, AD-mix-R, MeSO2NH2,
t-BuOH-H2O (1:1), 55% for two steps; (d) TESCl, imidazole, DMF,
85%.
O-protecting group.30 Applied to our case, deprotection of a
stable protected alcohol, given the sensitive iodoalkene moiety,
seemed to be questionable. Hence, we examined the feasibility
of using silyl-protected 24 in these alkylations.31 Synthesis of
the silylated glycolate began with the known PMB derivative
24b which was obtained from 23 in three steps.32 It was found
later that 23 could be converted to 24a in multigram scales,
albeit in moderate yields, by a one-flask procedure involving
in situ formation of a mixed anhydride.33 Remarkably, treatment
of lithio 24a with diiodide 22 at -78 °C afforded the desired
25a as a single isomer (>98% de) in good yield. Interestingly,
strong dependence of the yield on the nature of the silyl-
protecting group was observed. For example, only trace
amounts, at best, of the alkylation product could be obtained
using other silyl-protecting groups (e.g., R ) TMS, TBS, or
TBDPS in 24). In these cases, the chiral auxiliary 23 was
cleaved, and a complicated mixture of intractable products was
formed. Thus, the TES function was preferentially used as the
protecting device, and the asymmetric alkylation of 24a could
be routinely performed in multigram scales to give enantiopure
25a. After removal of the TES group, the formation of Weinreb
amide 26a34 was effected by simultaneous detachment of the
chiral auxiliary. Subsequent protection and Grignard addition
then afforded 11.
Scheme 4. Stereoselective Alkylation Route to Ketone 11a
a Reagents and conditions: (a) i) TiCl4, CH2Cl2, DIPEA, 87%, ii)
TESCl, imidazole, DMF, 84%; (b) i) glycolic acid, TESCl, NaH-TEA,
ether, 0 °C, then t-BuCOCl, -78 °C, ii) n-BuLi, -78 °C to rt, 38∼41%;
(c) LHMDS, -78 °C, THF, 81%; (d) AcOH:H2O:THF (3:1:1), 86%;
(e) i) CH3ONHCH3, AlMe3, CH2Cl2 ii) TESCl, imidazole, DMF, 88%;
(g) MeMgBr, 0 °C, 93%.
of two silyl ether regioisomers 19 and 20. Asymmetric dihy-
droxylation of the mixture, using AD-mix-R, generated hydroxy-
ketone 21 in 55% yield and in 87% enantiomeric excess.26 It is
noteworthy that it was possible to sustain the potentially
vulnerable iodoalkene functionality during the osmium-mediated
dihydroxylation.27 Finally, triethylsilylation of 25 produced 11,
completing the sequence in only four steps.
To develop a more enantioselective route, we explored the
possibility of an asymmetric alkylation reaction to establish the
C15 configuration (Scheme 4). Carbon-carbon bond formation,
using a chiral glycolate enolate as a nucleophile, had appeal
since the required (Z)-2,4-diiodo-2-butene (22) could be readily
prepared from 2-butyn-4-ol.28 Previously, asymmetric synthesis
of glycolates had been accomplished by oxygenation of the
“chiral enolate” (i.e., one armed with an auxiliary acyl group)
rather than by alkylation.29 Examples involving the alkylation
of a glycolate of the general type 24 tended to require a robust
With the requisite ketone 11 in hand, we turned our attention
to the preparation of Horner-like compound 10. Using the known
2-substituted thiazole derivative 27,35 phosphine oxide 29a was
uneventfully advanced through the intermediacy of chloride 28
by standard transformations (Scheme 5). Surprisingly, the
condensation between lithio 29a and 11, however, led to the
recovery of both reactants rather than to the formation of the
desired 30. Under a variety of different conditions, including
the employment of dilithio 29b as the nucleophile, the formation
of 30 could not be realized. These results stand in sharp contrast
to those encountered in our synthesis of the corresponding
(29) For example, see: Evans, D. A.; Morrissey, M. M.; Dorow, R. L.
J. Am. Chem. Soc. 1985, 107, 4346.
(30) For examples, see: (a) Jung, J. E.; Ho, H.; Kim, H.-D. Tetrahedron
Lett. 2000, 41, 1793. (b) Paterson, I.; Bower, S.; McLeod, M. D. Tetrahedron
Lett. 1995, 36, 175. (c) Cardillo, G.; Orena, M.; Romero, M.; Sandr, S.
Tetrahedron 1989, 45, 1501. (d) Kelly, T. R.; Arvanitis, A. Tetrahedron
Lett. 1984, 25, 39.
(31) Reference 23. During our study, an independent investigation on
this subject appeared. Crimmins, M. T.; Emmmitte, K. A.; Katz, J. D. Org.
Lett. 2000, 2, 2165.
(32) Evans, D. A.; Kaldor, S. W.; Jones, T. K.; Clardy, J.; Stout, T. J. J.
Am. Chem. Soc. 1990, 112, 7001.
(25) Satoh, Y.; Serizawa, H.; Hara, S.; Suzuki, A. J. Am. Chem. Soc.
1985, 107, 5225.
(26) (a) Sharpless, K. B.; Amberg, W.; Beller, M.; Chen, H.; Hartung,
J.; Kawanami, Y.; Lubben, D.; Manoury, E.; Ogino, Y.; Shibata, T.; Ukita,
T. J. Org. Chem. 1991, 56, 4585. (b) Sharpless, K. B.; Amberg, W.; Bennani,
Y. L.; Crispino, G. A.; Hartung, J.; Jeong, K.-S.; Kwong, H.-L.; Morikawa,
K.; Wang, Z.-M.; Xu, D.; Zhang, X.-L. J. Org. Chem. 1992, 57, 2768. (c)
Hashiyama, T.; Morikawa, K.; Sharpless, K. B. J. Org. Chem. 1992, 57,
5067.
(27) Fattori, D.; de Guchteneere, E.; Vogel, P. Tetrahedron Lett. 1989,
52, 7415.
(28) Gras, J.-L.; Kong Win Chang, Y.-Y.; Bertrand, M. Tetrahedron Lett.
1982, 23, 3571.
(33) Coss´ıo, F. P.; Palomo, C. Tetrahedron Lett. 1985, 26, 4239.
(34) (a) Nahm, S.; Weinreb, S. M. Tetrahedron Lett. 1981, 22, 3815.
(b) Levin, J. I.; Turos, E.; Weinreb, S. M. Synth. Commun. 1982, 12, 989.
(35) Ciufolini, M. A.; Shen, Y. C. J. Org. Chem. 1997, 62, 3804.

5254 J. Am. Chem. Soc., Vol. 123, No. 22, 2001
Lee et al.
Scheme 6. Synthesis of the O-Alkyl Wing via Horner
ester 37 were prepared in good yields with excellent enanti-
oselectivities (entries 4 and 5). The ee and sense of the absolute
stereoselectivity was determined by derivatization of 37 to the
corresponding Mosher’s ester, and corroborated by subsequent
events in our program.38
Condensationa
Having found a method to generate 37 in high enantiomeric
excess, we set out to investigate the double stereo-differentiation
aldol reaction.39 To use protected 37 as the nucleophile, both
enantiomeric aldol adducts were converted to corresponding
TBS ether antipodes 38. Unfortunately, the attempted reaction
between lithio 38 with 14, however, could not be effected in
meaningful yield due to the sensitivity of the â-silyloxy system
to elimination. Thus, we took recourse to a less basic, titanium
enolate of 38 (TiCl4, i-Pr2NEt, CH2Cl2, -78 °C) for the same
aldol reaction (eqs 2 and 3).40 Indeed, both (S)-38 and (R)-38
underwent the aldol condensation, giving mixtures of diaster-
eomers in moderate yields (41b:42b ) 3.5:1, 52% and 43b:
44b ) 1:3.0, 40%). The stereochemical outcome of the major
diastereomers, 41b and 44b, from each reaction indicated that
the configuration of C3 rather than C8 had a larger effect on
the sense of the newly formed C6 and C7 centers. However,
the degree of the selectivity (dr ) 3∼3.5:1) in these reactions
was lower than that of our previous reaction (dr ) 6.5:1),
suggesting that neither enantiomer of 38 benefited from match-
ing chirality in 14. While the explanation for the puzzling lack
of “matching” in these alkylation reactions will require more
revealing experiments, the results must, to some extent, reflect
differences arising from altering the cation (titanium rather than
lithium). The special “γ,δ-unsaturated aldehyde effect” was most
pronounced with lithium as the counterion.41
a Reagents and conditions: (a) toluene, 110 °C, 2 h, 96%; (b)
HOPPh2, Cs2CO3, cat. TBAI, CH2Cl2, rt, 48 h, 82%; (c) LHMDS, THF,
-78 °C, 52%; (d) 2.5 eq Dibal-H, CH2Cl2, 0 °C, 98%; (e)
Cl3CCH2OCOCl, Pyridine, CH2Cl2, 86%.
segment required for EpoB. In that case, excellent yields were
achieved using a 2-methylthiazole derivative (R ) CH3 in
10).22,23
The failure of the needed Horner reaction led us to seek an
alternative approach in which the C21 hydroxy function could
emerge after the olefination reaction (Scheme 6). Accordingly,
ethyl thiooxamate (31) was condensed with 1,3-dichloroacetone
(32) to provide 2,4-disubstituted thiazole 33 in excellent yield.
Subsequent Arbuzov reaction using Ph2POEt gave 34 in
moderate yield. Fortunately, direct P-alkylation with HOPPh3
generated 34 in good yield. Upon treatment with LHMDS,
thiazole 34 and ketone 11 condensed smoothly to furnish 35 as
a single geometric isomer. Finally, reduction of ester 35 with
Dibal-H generated alcohol 36 which was protected with a Troc
group to yield the desired O-alkyl moiety. The new route is
significantly shorter and more convergent than the previous
sequence20 which involved 10 linear transformations.
New Synthetic Route to the O-Acyl Fragment. Having
successfully prepared the O-alkyl building block 8 in quantity,
our attention was turned to the development of a route to the
O-acyl fragment 9. As described earlier, use of the segment
used in our previous synthesis (cf. 4), led to a surprising
solvolysis problem at C15 during the Noyori reduction. Hence,
the major point of improVement was focused on the establish-
ment of the C3 carbinol center prior to the Suzuki merger. Our
investigation commenced with attempts to realize an asymmetric
Reformatsky reaction (eq 1, Table 1).
The difficulty of conducting the aldol reaction with a lithium
enolate of 38 led us to investigate the feasibility of condensation
using dilithio 37 as a nucleophile. Indeed, the reaction of the
dianion of 37 with aldehyde 14 did not induce â-elimination,
and the condensation proceeded smoothly.42 When dilithio (S)-
37 was treated with homochiral 14, there was obtained a 2:3
mixture of diastereomers 41a and 42a in 53% yield. Extensive
Addition reactions of “zincated” tert-butyl bromoacetate to keto
aldehyde 13 were performed in the presence of chiral amino
alcohol 39.36 The reactions gave rise to the expected aldol adduct
in excellent yields. Unfortunately, the enantioselectivities did
not rise to synthetically useful levels (entries 1 and 2). While
decreasing the reaction temperature increased the enantioselec-
tivities significantly, reactions conducted this way suffered from
poor conversions (entry 3).
Accordingly, the classical Reformatsky inspired reactions
were set aside. Fortunately, a high degree of stereocontrol was
achieved by the addition reaction of a chiral titanium enolate
derived from tert-butyl acetate to the aldehyde. Following
literature protocol,37 both enantiomers of â-hydroxy tert-butyl
(38) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512.
(39) For examples of aldol reactions of â-hydroxy ketones, see: (a) Luke,
G. P.; Morris, J. J. Org. Chem. 1995, 60, 3013. (b) Evans, D. A.; Calter,
M. A. Tetrahedron Lett. 1993, 34, 6871. (c) McCarthy, P. A.; Kageyama,
M. J. Org. Chem. 1987, 52, 4681.
(40) Evans, D. A.; Urp´ı, F.; Somers, T. C.; Clark, J. S.; Bilodeau, M. T.
J. Am. Chem. Soc. 1990, 112, 8215.
(41) (a) Harris, C. R.; Kuduk, S. D.; Balog, A.; Savin, K.; Danishefsky,
S. J. Tetrahedron Lett. 1999, 40, 2267. (b) The use of other enlolates (K,
Na, Mg, Al, and B) instead of Li for the aldol reaction gave inferior results
(dr ) 2∼4:1). Harris, C. R.; Kuduk, S. D.; Lee, C. B. Unpublished results
in these laboratories.
(42) Martin, V. A.; Murray, D. H.; Pratt, N. E.; Zhao, Y.-b.; Albizati,
K. F. J. Am. Chem. Soc. 1990, 112, 6965.
(36) (a) Soai, K.; Takahashi, K. J. Chem. Soc., Perkin Trans 1 1994,
1257. (b) Soai, K.; Niwa, S. Chem. ReV. 1992, 92, 833.
(37) Duthaler, R. O.; Herold, P.; Lottenbach, W.; Oretle, K.; Reidiker,
M. Angew. Chem., Int. Ed. Engl. 1989, 28, 495.

Synthesis of Desoxyepothilone F
J. Am. Chem. Soc., Vol. 123, No. 22, 2001 5255
Table 1. Asymmetric Synthesis of C3 Stereogenic Center
a Isolated yields. b er ) enantiomeric ratio determined by 1H NMR integration of appropriate signals of the corresponding Mosher’s ester.
c L-DIPGF ) 1,2:5,6-di-O-isopropylidene-R-L-glucofuranos-3-O-yl. d D-DIPGF ) 1,2:5,6-di-O-isopropylidene-R-D-glucofuranos-3-O-yl.
Figure 2. Factors governing the aldol reaction.
chemical and spectroscopic correlation studies revealed that the
major diastereomer 42a corresponded to a C6-C7 bis-epi
version of 9 (vide infra). Thus, the desired isomer 41a whose
configurations were those required to reach “natural” epothilones
constituted the minor product. We then performed the corre-
sponding reaction using the enantiomeric ketone, (R)-37.
Remarkably, this reaction gave rise to a single diastereomer in
88% yield. Subsequent correlation studies verified that the
stereochemistry of the C6 and C7 centers corresponded to those
of “natural” epothilones. The C3 center was, of course,
configured as R rather than S. Therefore, the use of 43a as a
precursor of the O-acyl wing 9 would necessitate a subsequent
inversion at C3. Attempts to conduct Mitsonobu type inversions
were not rewarding.
be taken into account. We propose that the factors governing
such aldol condensation include; (i) the chairlike transition state
which leads to a syn-aldol product,43 (ii) chelation of the lithium
counterion by the â-alkoxido oxygen, (iii) a syn relationship
between the R-proton of the aldehyde and the optimal presenta-
tion of the C5 methyl group of the enolate to minimize a
developing syn pentane interaction,44 (iv) a preference for the
aldehyde to attack the enolate anti to the large resident R group,
and (v) the special γ,δ-unsaturated aldehyde effect41 which tends
to orient the pendent allyl group for a favorable interaction with
the carbonyl group of the aldehyde, thus guiding the enolate
attack anti to such organized framework of the aldehyde. Note
that factor (v) reinforces the facial selectivity of the aldehyde
influenced by factor (iii).
Despite the undesired outcome of this attempt, the results
from the dianion aldol reaction provided potentially useful
information for understanding the principles underlying double
stereo-differentiation aldol reactions, and the matter was pursued
in some detail. The C3 protected series, such as in Schinzer’s
case,24 required the (S) configuration at C3 to deliver C6(R),-
C7(S) configurations in a related aldol reaction (employing a
C8(S)-aldehyde structurally very similar to 14). In contrast, the
combination of (R) chirality at C3 and (S) chirality at C8 led to
the required C6(R),C7(S) configurations in our dianion aldol
reactions. The disparity between the two systems in terms of a
“matching” and “mismatching” pair, strongly suggested that the
transition state of the aldol reaction of the C3 unprotected
(alkoxide) system is very different from that of the C3 protected
system.
Taking all ofthese preferences into consideration, we analyzed
the structure of transition states leading to the formation of 41a-
44a (Scheme 7).45 While the relative contribution of each factor
to transition state stabilization cannot be weighted, we note the
reaction of (R)-37 with 14 incorporates each of the energy-low-
ering elements we propose in the transition state. Thus, the
reaction between these two compounds constitutes a “matched”
event, giving rise to 43a with complete diastereoselectivity. On
the other hand, the transition state for the formation of the alter-
native (nonobserved) diastereomer 44a, would lack both favor-
able elements (iv) and (v). In a similar manner, the combination
(43) (a) Heathcock, C. H.; Pirrung, M. C.; Lampe, J.; Buse, C. T.; Young,
S. D. J. Org. Chem. 1981, 46, 2290. (b) Frater, G. Tetrahedron Lett. 1981,
22, 425. (c) Seebach, D.; Wasmuth, D. Angew. Chem., Int. Ed. Engl. 1981,
20, 971. (d) Evans, D. A.; Ennis, M. D.; Mathre, D. J. J. Am. Chem. Soc.
1982, 104, 1737. (e) Williard, P.; Salvino, J. M. Tetrahedron Lett. 1985,
26, 3931.
(44) Roush, W. R. J. Org. Chem. 1991, 56, 4151.
(45) For the preliminary communication, see: Wu, Z.; Zhang, F.;
Danishefsky, S. J. Angew. Chem., Int. Ed. 2000, 39, 4505.
To explain the drastic influence of the C3 protection status
on the stereochemical outcome and the remarkable selectivity
in the aldol reaction between (R)-37 and aldehyde 14, various
considerations pertinent to the transition state (Figure 2) must

5256 J. Am. Chem. Soc., Vol. 123, No. 22, 2001
Lee et al.
Scheme 7. Proposed Transition States for Dianion Aldol Reactions
Scheme 8. Schinzer-Type Aldol Reaction
We have not conducted our own experiments with Schinzer-
type aldol condensations where the oxygens at C3 and C1 are
co-protected as a cyclic acetonide (Scheme 8). However, we
note that in the Schinzer C3 protected series,24 it is the C3(S)
configuration in B which is required to induce the C6(R),C7(S)
configurations. This result is in contrast to our C3 unprotected
case, reported above, wherein the C3(R) antipode is the one
which induces the C6(R),C7(S) configurations. We propose that
the C3-OR bond is orientated differently in the two enolates
and that this key divergence accounts for the opposite senses
of long-range asymmetric induction of the protected (cf. B)
versus unprotected enolates (see dilithio S-37). We propose that
in B (as opposed to 37), the C3-OR group is positioned anti-
periplanar to the C4-C5 bond in the transition state C. This
orientation in the C3(S) enolate would favor the attack of the
C8(S) aldehyde A syn to the C3 hydrogen rather than syn to
the large C1-C2 moiety. This bias would give rise to the C6-
(R),C7(S) configuration in the product D. While the orientation
of the C3 alkoxy substituent in the transition state is not
established with certainty, it is well to note that the aldol reaction
of chiral â-alkoxy enolates have been shown to exhibit strong
1,5-asymmetric induction.46
of dilithio (S)-37 with 14 to produce 41a or 42a, although
incorporating the features (i-iii), leads to suboptimal transition
states, lacking either (iv) or (v). To reach 41a, the aldehyde
must attack the enolate syn to the large group, contrary to the
preferred arrangement (factor iv). The alternative facial mode
which leads to 42a would require a “trade off” between features
(iii) and (v), since the transition state either forfeits the favorable
overlap of the terminal alkene with the aldehyde carbonyl or
accepts the unfavorable syn pentane interaction.
To explore the concepts implied in Figure 2 and in Scheme
7, a critical experiment was conducted. Since our interpretation
of the outcome of the results described above clearly indicates
the presence of matched and mismatched events, it seemed
reasonable that there could be measurable kinetic implications.
Accordingly, we performed an aldol reaction in a setting where
the homochiral aldehyde 14 could face internal competition
between the two enantiomeric enolates. When the racemic
enolate corresponding to both antipodes of 37 was treated with
a limiting concentration (0.5 equiv) of homochiral aldehyde 14
at -78 °C, the reaction generated only 43a as the product,
virtually free of other diastereomers (41a, 42a, and 44a) in 70%
yield (eq 4). Therefore, our hypothesis as to the matching-
mismatching issues in the fascinating aldol reaction has been
supported at the kinetic level.
Although the double stereo-differentiating aldol process
proved highly informative, the use of 43a in a total synthesis
would, as noted above, require a nonstraightforward inversion
at the C3 center of the product. Thus, we sought an alternative
sequence to 9 wherein sequential “substrate-directed” and
“reagent-controlled” aldol reactions would establish the requisite
configurations at C6, C7, and C3 (Scheme 9).
(46) For a related issue in the aldol reaction of ketones, see: (a) Paterson,
I.; Gibson, K. R.; Oballa, R. M. Tetrahedron Lett. 1996, 37, 8585. (b) Evans,
D. A.; Coleman, P. J.; Coˆte´, B. J. Org. Chem. 1997, 62, 788. (c) Tanimoto,
N.; Gerritz, S. W.; Sawabe, A.; Takeshi, N.; Filla, S. A.; Masamune, S.
Angew. Chem., Int. Ed. Engl. 1994, 33, 673.

Synthesis of Desoxyepothilone F
J. Am. Chem. Soc., Vol. 123, No. 22, 2001 5257
Scheme 9. New Synthetic Route to the O-Acyl Winga
a Reagents and conditions: (a) CH(Oi-Pr)3, iPrOH, cat. TsOH, 88%; (b) LDA, -78 °C, 85%, 46:47 ) 4:1; (c) TrocCl, Pyridine, CH2Cl2, 0 °C,
99%; (d) H2O/THF, cat. TsOH, 88%; (e) THF, 89%, dr > 20:1; (f) TBSOTf, 2,6-lutidine, CH2Cl2, -78 °C, 81%; (g) TESCl, imidazole, DMF,
96%.
Scheme 10. Application to the Total Synthesis of dEpoBa
a Reagents and conditions: (a) i) 9-BBN-H, THF (ii) PdCl2(dppf), AsPh3, DMF-THF-H2O, rt, 2 h, 72%; (b) TESOTf, 2,6-lutidine, CH2Cl2,
-78 °C to rt, 8 h, (ii) HCl-CH3OH, THF, 0 °C, 69%.
Toward this end, keto aldehyde 13 was protected as a
diisopropyl acetal, and the resultant ketone 45 was subjected
to an aldol reaction with 14. Upon deprotonation and reaction
of 45 with 14, smooth condensation gave rise to a 4:1 mixture
of aldol adducts 46a and 47a. While this ratio was somewhat
lower than that of our previous route (6.5∼5.5:1),13e,16,45 the
major diastereomer 46a was very easily separated by flash
chromatography and protected as a Troc group. Indeed, there
was a significant advantage in this route in that the aldol
condensation, unlike our earlier cases where C3 corresponded
to an enol ether,13e went to completion. Furthermore, the reaction
conditions are much less demanding technologically, since now
the coupling is conducted at -78 °C rather than at -120 °C in
the previous synthesis.
51a, whose TBS ether 51b correlated with the aldol product 43
from the earlier studies (eqs 5 and 6). Similarly, the minor
isomer 47 was correlated. Compound 53, arising from 52,
corresponded to the previously encountered 42.
Hydrolysis of the diisopropyl acetal group of 46b under acid
catalysis gave keto aldehyde 48, setting the stage for the second
aldol reaction. Following the same “titano” tert-butyl ester
method using a glucose derived auxiliary as in eq 1, the desired
C3(S)-49 was obtained in high diastereoselectivity (dr > 20:1).
Protection of the C3 alcohol with a silyl group finally afforded
the O-acyl wing 9 and the TBS derivative 50 whose spectral
and chromatographic properties were identical to previously
obtained material from other programs in these laboratories.13e
For the purpose of correlation studies, keto aldehyde 48 was
treated with an enantiomeric tert-butyl titanoacetate to afford
In addition to these correlation studies, the newly prepared 9
was utilized in the formal total synthesis of dEpoB as illustrated
in Scheme 10. The B-alkyl Suzuki coupling with the O-alkyl
segment for EpoB series 54 proceeded to yield 55. Subsequently,
treatment of 55 with TESOTf followed by selective desilylation
afforded hydoxyacid 56 which had been advanced to dEpoB
(2b) and EpoB (1b) in our previous syntheses. Accordingly,
these studies unambiguously established the stereochemical

5258 J. Am. Chem. Soc., Vol. 123, No. 22, 2001
Lee et al.
Scheme 11. Completion of the Total Synthesis of dEpoFa
a Reagents and conditions: (a) i) 9-BBN-H, THF (ii) PdCl2(dppf), AsPh3, DMF-THF-H2O, rt, 8 h, 89%; (b) TESOTf, 2,6-lutidine, CH2Cl2,
-78 °C to rt, 8 h, (ii) HCl-CH3OH, 0 °C, 78%; (c) 2,4,6-trichlorobenzoyl chloride, (C2H5)3N then 4-DMAP, toluene, slow addition 3 h, 60-70%;
(d) Zn, AcOH-THF, rt, 1 h, 90% (e) HF‚pyridine, THF 0 °C to rt, 91%.
Scheme 12. Synthesis of EpoF and Photoffinity Labeled dEpoF
outcomes of the various aldol reactions and constitute an
alternate total synthesis of dEpoB. The ultimate choice between
the “titanoacetate”-based route to the C3(S) series in the acyl
sector described above relative to the post-Suzuki (C3 keto f
C3(S) hydroxy) route practiced previously will depend on issues
of costing and on susceptibility to scale-up.
Finally, and most importantly, we applied these findings to
the total synthesis of dEpoF (Scheme 11). As explained above,
a much improved total synthesis of this compound was critical
if it were to advance to clinical evaluation. At this stage it
seemed that the key synthetic issues had indeed been favorably
resolved. The proposed O-alkyl and acyl building blocks were
now available in multigram quantities with a clear closing
strategy in place.
Fragments 8 and 9 were conjoined via the B-alkyl Suzuki
protocol to provide the seco ester 58. Conversion of the tert-
butyl ester 58 to the TES ester followed by acid-catalyzed
selective desilylation provided 59, which awaited macrolacton-
ization. At this point, detailed spectral correlations of 59
confirmed its identity with the previously synthesized intermedi-
ate. Following the same sequence as was used in our earlier
synthesis,20 hydroxyacid 59 was cyclized to the fully protected
macrolactone 60. Finally, sequential removal of the Troc and
TES groups afforded 12,13-desoxyepothilone (2d, dEpoF) which
again proved identical in all respects with the previously
synthesized dEpoF.
by epoxidation at the 12,13-alkene (Scheme 12). Treatment of
the synthetic dEpoF with 2,2-dimethyldioxirane (DMDO)
induced the formation of the 12,13-epoxide with natural
stereochemistry (dr > 15:1) to afford EpoF. Spectroscopic data
and the observed [R]D -26.5 (c 0.35, MeOH) corresponded well
to the naturally occurring EpoF, lit.,18 [R]D -27.4 (c 0.5,
MeOH). As we had hoped at the outset, the 21-hydroxyl group
did increase the aqueous solubility of dEpoB by a factor of 2.5
as measured by a HPLC method.47 In addition, its utility as a
staging point for further functionalization was demonstrated.
The condensation of unprotected dEpoF with azidoacid 61 was
performed under the agency of DCC to afford the photoaffinity
labeled dEpoF 62. While the tubulin-binding assay had shown
that dEpoF retains 90% activity of dEpoB, the aroylated
derivative 62 did not induce tubulin polymerization,48 thus
underscoring the subtle nature of the thiazole region in effecting
tubulin binding.
The fully synthetic dEpoF was first tested against various
cell types to evaluate its antitumor potential. As shown in Table
2, dEpoF showed high cytotoxic activities against a broad range
of sensitive and resistant tumor cell lines. In particular, dEpoF
retained high potency and low cross-resistances against MDR
cell lines and consistently outperformed other non-epothilone
anticancer agents such as paclitaxel, vinblastine, etoposide,
actinomycin, and adriamycin. These properties of dEpoF are
(47) Swindell, C. S.; Krauss, N. E.; Horwitz, S. B.; Ringel, I. J. Med.
Chem. 1992, 34, 1176. Also see ref 20.
(48) Horwitz, S. B. Personal communication.
Following completion of the synthesis of dEpoF, we also
accomplished the synthesis of epothilone F (1d, EpoF) itself

Synthesis of Desoxyepothilone F
J. Am. Chem. Soc., Vol. 123, No. 22, 2001 5259
Table 2. Potency of dEpoF, dEpoB, and Taxol against Various Tumor Cell Growth in Vitro
IC50 (µM)a
tumor cell Lines
dEpoF
dEpoB
Taxol
Others
Human T-cell AL Leukemia
CCRF-CEM
0.0027
0.0095
0.0021
0.00063b, 0.0290c
CCRF-CEM/VBL100
CCRF-CEM/VM1
CCRF-CEM/Taxol
0.047 (17.4 x)
0.0049 (1.8 x)
0.0053 (2.0 x)
0.017 (1.8 x)
0.014 (1.5 x)
0.0162 (1.7 x)
4.140 (1971 x)
0.0066 (3.18 x)
0.120 (57 x)
0.332b (527 x)
3.44c(117 x)
Hamster Lung Fibroblasts
DC-3F
DC-3F/ADX
DC-3F/ADII
0.0017
0.0136 (8.0 x)
0.0223 (13.1 x)
0.0019
0.0135
0.583 (43.2 x)
20.19 (1496 x)
0.00025d
0.0073 (3.8 x)
0.0288 (15.2 x)
0.00153d (61.2 x)
0.4092d (1637 x)
Human Promyelocytic Leukemia
0.0031 0.0011
HL-60
K562
PC-3
0.0007
0.0021
0.0119
0.0014
Human CM Leukemia
0.0036
0.0029
Human Prostate Adenocarcinoma
0.0209
0.0280
Human Colon Adenocarcinoma
HT-29
0.0048
0.0016
Human Mammary Adenocarcinoma
MCF-7
0.0069
0.0040
0.0024
0.081e, 0.0094b
MCF-7/Adr
0.0097 (1.4 x)
0.00715 (1.4 x)
0.0135 (5.6 x)
0.280 (3.5 x), 0.025 (26.6 x)
Human Mammary Carcinoma
MX-1
0.0042
0.0221
0.0394
0.00184f
Human Ovary Carcinoma
SK-OV-3
UL-3-C
UL-3-B/Taxol
0.0051
0.0048
0.0067 (1.4 x)
0.0035
0.0021
0.0038
0.0016
0.0016f
0.00037b, 0.00058f
0.00310 (8.4 x)b, 0.00124 (2.1 x)f
0.0070 (3.3 x)
0.0107 (6.7 x)
a Cell growth inhibition was measured by XTT tetrazonium assay after 72 h incubation for cell growth as described previously in ref 15. The
values were determined with six to seven concentrations of each drug using a computer program. The cross-resistances are shown in parentheses.
b Vinblastin (VBL). c Etoposide (VM1, VP-16). d Actinomycin D (AD). e Adriamycin (Adr). f Epothilone B (EpoB).
closely comparable to those of the highly promising antitumor
agent, dEpoB. Striking in vivo comparisons of dEpoF with a
current clinical candidate already in phase I trials will be
presented elsewhere.49
an aldol condensation have been probed at the kinetic level. A
remarkable instance of the consequences of molecular recogni-
tion in covalent bond formation has been demonstrated via
kinetic resolution.
Given the high degree of intolerance found in the SAR map
of the polypropionate region,13b the new O-acyl wing fragment
may serve as a widely applicable intermediate for accessing
various analogues. The in vitro and in vivo evaluations of
dEpoF49 attest to the impressive potential of 12,13-desoxye-
pothilones in the B series.
Summary
A markedly improved total synthesis of the promising
antitumor agent, dEpoF has been accomplished. The new
synthesis features a convergent and modular nature strategy with
strong stereoselectivities at each step. The conciseness of the
syntheses of the key intermediates readily allows for large-scale
preparation and easy structural variation in each synthetic
segment.
For the synthesis of the O-alkyl wing of dEpoF, the Horner-
like condensation conveniently conjoined the thiazole moiety
and the segment possessing the critical C15 stereochemistry and
(Z)-12,13-alkene function. The stereoselective aldol reactions
en route to the O-acyl wing were investigated in some detail. It
was demonstrated that a subtle variation in factors affecting the
transition state has a remarkable influence on the long-range
transmission of stereochemical information. In these studies,
the implications of stereochemical matching of components in
Acknowledgment. This work is dedicated to Professor
Barry M. Trost on the occasion of his 60th birthday. This work
was supported by the National Institutes of Health (Grant
Numbers: CA28824). Postdoctoral Fellowship Support is
gratefully acknowledged by C.B.L. (U.S. Army Grants DAMD
17-97-1-7146 and 17-98-1-8155), M.D.C. (NIH F32GM1997201)
and S.J.S. (NIH F32CA81704). We thank Dr. George Sukenick
(NMR Core Facility, Sloan-Kettering Institute and Grant
CA08748) for mass spectral and NMR analysis.
Supporting Information Available: Procedures for the
synthesis and characterization data of all new compounds (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
(49) (a) Stachel, S. J.; Lee, C.; Bornmann, W. G. Spassova, M.; Chappell,
M. D.; Danishefsky, S. J.; Chou, T.-C.; Guan, Y. J. Org. Chem. In Press;
(b) Chou, T.-C.; Guan, Y.; Zhang, X.-G.; Stachel, S. J.; Lee, C.;
Danishefsky, S. J. Proc. Nat. Acad. Sci. U.S.A., Manuscript submitted.
JA010039J