New chemical synthesis of the promising cancer chemotherapeutic agent 12,13-desoxyepothilone B: Discovery of a surprising long-range effect on the diastereoselectivity of an aldol condensation

Article abstract of DOI:10.1021/ja991189l

The epothilones are naturally occurring cytotoxic molecules that possess the remarkable ability to arrest cell division through the stabilization of microtubule assemblies. Our in vivo studies with 12,13-desoxyepothilone B (dEpoB), have established that the desoxy compound is well tolerated and virtually curative against a variety of sensitive and resistant xenograft tumors in animal models. In light of these discoveries, we sought a chemical synthesis of dEpoB that would be able to support a serious and substantial discovery research program directed toward the clinical development of this molecule. The overall strategy for this endeavor assumed the ability to synthesize dEpoB from three constructs which include an achiral β,δ-diketo ester construct A, an (S)-2-methylpentenal moiety B, and the thiazoyl-containing vinyl iodide moiety C. We envisioned that a diastereoselective aldol condensation between an achiral C5-C6 (Z)-metalloenolate derived from construct A and an (S)-2-methylalkanal fragment, B, would generate the desired C6-C7 bond. Second, a B-alkyl Suzuki coupling between the vinyl iodide construct C and an alkyl borane would form the C11-C12 bond. Finally, a late-stage reduction of the C3 ketone to the requisite C3 alcohol with high asymmetric induction would permit us to introduce the β,δ-diketo ester fragment A, into the synthesis as a readily accessible achiral building block. The governing concepts for our new synthesis are described herein.

Full text of DOI:10.1021/ja991189l

7050
J. Am. Chem. Soc. 1999, 121, 7050-7062
New Chemical Synthesis of the Promising Cancer Chemotherapeutic
Agent 12,13-Desoxyepothilone B: Discovery of a Surprising
Long-Range Effect on the Diastereoselectivity of an Aldol
Condensation
Christina R. Harris,† Scott D. Kuduk,† Aaron Balog,†,§ Ken Savin,†,| Peter W. Glunz,† and
Samuel J. Danishefsky*,†,‡
Contribution from the Laboratory for Bioorganic Chemistry, 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 April 14, 1999
Abstract: The epothilones are naturally occurring cytotoxic molecules that possess the remarkable ability to
arrest cell division through the stabilization of microtubule assemblies. Our in ViVo studies with 12,13-
desoxyepothilone B (dEpoB), have established that the desoxy compound is well tolerated and virtually curative
against a variety of sensitive and resistant xenograft tumors in animal models. In light of these discoveries, we
sought a chemical synthesis of dEpoB that would be able to support a serious and substantial discovery research
program directed toward the clinical development of this molecule. The overall strategy for this endeavor
assumed the ability to synthesize dEpoB from three constructs which include an achiral â,δ-diketo ester construct
A, an (S)-2-methylpentenal moiety B, and the thiazoyl-containing vinyl iodide moiety C. We envisioned that
a diastereoselective aldol condensation between an achiral C5-C6 (Z)-metalloenolate derived from construct
A and an (S)-2-methylalkanal fragment, B, would generate the desired C6-C7 bond. Second, a B-alkyl Suzuki
coupling between the vinyl iodide construct C and an alkyl borane would form the C11-C12 bond. Finally,
a late-stage reduction of the C3 ketone to the requisite C3 alcohol with high asymmetric induction would
permit us to introduce the â,δ-diketo ester fragment A, into the synthesis as a readily accessible achiral building
block. The governing concepts for our new synthesis are described herein.
Introduction
and presumably the epothilones, is the microtubule which is
vital for mitosis, motility, secretion, and proliferation.8 Recently,
it was established that the taxanes also induce several early
response and cytokine genes, and it has been demonstrated that
gene expression is dramatically influenced by taxoid chemical
structure.9 Consequently, some of the pharmacological effects
of paclitaxel may originate because of its effect on gene
expression and not exclusively from stabilization of microtubule
assemblies. The observed anticancer effects of the epothilones
are also attributed to their ability to stabilize microtubule
assemblies although minimal data on the physiological effects
of the epothilones on gene expression has been reported in the
literature.9
The taxoids and epothilones demonstrate analogous effects
in their ability to arrest cellular mitosis and to induce the
formation of hyperstable tubulin polymers in both cultured cells
and microtubule protein.10 Thus epothilones A and B competi-
tively inhibited the binding of [3H]paclitaxel to tubulin poly-
mers.11 This result was interpreted to suggest binding of the
Epothilones A (1, EpoA) and B (2, EpoB), Scheme 1, are
cytotoxic macrolide natural products1,2 which exhibit biological
effects that parallel those of Taxol (paclitaxel) and Taxotere
(docetaxel).3-6 The epothilones were first isolated from the
mycobacterium Sorangium cellulosum that were harvested off
the shores of the Zambezi River in South Africa.1,2 Taxol
(paclitaxel) and Taxotere (docetaxel) are clinically effective
antineoplastic agents7 whose chemical structure is significantly
different than the epothilones. The primary target of the taxanes,
† The Sloan-Kettering Institute for Cancer Research.
‡ Columbia University.
§ Present address: Bristol-Myers Squibb Company, Princeton, New
Jersey 08543.
| Present address: Eli Lilly Corporate Research Center, Indianapolis,
Indiana 46285.
(1) Gerth, K.; Bedorf, N.; Ho¨fle, G.; Irschik, H.; Reichenbach, H. J.
Antibiot. 1996, 49, 560.
(2) Ho¨fle, G.; Bedorf, N.; Steinmetz, H.; Schomburg, D.; Gerth, K.;
Reichenbach, H. Angew. Chem., Int. Ed. Engl. 1996, 35, 1567.
(3) Rowinsky, E. K.; Eisenhauer, E. A.; Chaudhry, V.; Arbuck. S. G.;
Donehower, R. C. Semin. Oncol. 1993, 20, 1-15.
(4) Cortez, J. E.; Pazdur, R. J. Clin. Oncol. 1995, 13, 2643-2655.
(5) Rose, W. C. Anti-Cancer Drugs 1992, 3, 311-321.
(6) Cass, C. E.; Janowska-Wieczorek, A.; Lynch, M. A.; Sheinin, H.;
Hindenburg, A. A.; Beck, W. T. Cancer Res. 1989, 49, 5798-5804.
(7) (a) Holmes, F. A.; Walters, R. S.; Theriault, R. L.; Forman, A. D.;
Newton, L. K.; Raber, M. N.; Buzdar, A. U.; Frye, D. K.; Hortobagyi, G.
N. J. Natl. Cancer Inst. 1991, 83, 1797. (b) McGuire, W. P.; Rowinsky, E.
K.; Rosenshein, N. B.; Grumbine, F. C.; Ettinger, D. S.; Armstrong, D. K.;
Donehower, R. C. Ann. Intern. Med. 1989, 111, 273. (c) Rowinsky, E. K.;
Donehover, R. C. N. Engl. J. Med. 1995, 332, 1004.
(8) (a) Horwitz, S. B.; Fant, J.; Schiff, P. B. Nature, 1979, 277, 665. (b)
Hyams, J. F.; Loyd, C. W. Microtubules; Wiley-Liss: New York, 1993.
(9) (a) Moos, P. J.; Fitzpatrick, F. A. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 3896. (b) Kirikae, T.; Ojima, I.; Kirikae, F.; Ma, Z.; Kuduk, S. D.;
Slater, J. C.; Takeuchi, C. S.; Bounaud, P.-Y.; Nakano, M. Biochem. Biophy.
Res. Commun. 1996, 227, 227. (c) Lee, L. F.; Haskill, J. S.; Mukaida, N.;
Natsushima, K.; Ting, J. P. Mol. Cell. Biol. 1997, 17, 5097.
(10) 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.
10.1021/ja991189l CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/20/1999

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7051
epothilones and taxanes at the same site, although binding in
an overlapping or allosteric site could not be ruled out through
these experiments.
Scheme 1. Retrosynthetic Disconnections of the New
Synthesis of the Epothilones through the Merger of
Constructs A, B, and C
Perhaps the most intriguing discovery encountered during an
early investigation of the epothilones is that EpoB is notably
more potent (both in Vitro and in ViVo) than paclitaxel in
inhibiting cell growth.10 Moreover, while the cellular mechanism
of the epothilones and the taxanes is apparently the same, the
epothilones stand out in their ability to retain activity against
multidrug-resistant (MDR) cell lines and tumors where paclitaxel
and other major chemotherapeutic agents fail.9,10
Recognizing that the epothilones, or suitably modified deriva-
tives, might find status as cancer chemotherapy agents, we
embarked in a multidisciplinary pursuit to investigate these
compounds. Included in our objectives was the goal of total
synthesis of the epothilones and their analogues, which was
accomplished first by us12 and subsequently by others.13 While
the epothilones are available by fermentation,1,2 the more active
epothilone B is particularly scarce, and thus we sought chemical
synthesis as a means for our laboratory to gain access to this
series of compounds and their analogues for in Vitro and in ViVo
analysis.
This account begins with a brief synopsis of our early
synthetic efforts toward this end. In the following contribution,
we shall enumerate our findings regarding our recently
accomplished practical chemical synthesis of epothilone B and
its congeners.14
Background
From the outset, it was readily apparent that unlike the
structurally complex taxanes, the less forbidding structure of
the epothilones could be better suited for rapid development of
structure-activity comparisons (SAR) through chemical syn-
thesis. Consequently, the initial goal of our synthetic program
was to realize a total chemical synthesis of the epothilones (and
their analogues) that would provide material for preliminary
biological testing, both in Vitro and in ViVo. We were hopeful
that the SAR data would identify congeners of the naturally
occurring epothilones with enhanced chemotherapeutic efficacy.
In addition to considerations of potency, we were particularly
interested in the MDR-reversing ability of our synthetic targets
in comparison to their MDR-susceptible taxoid counterparts.10
Notwithstanding the less intimidating architecture of the
epothilones relative to the taxoids, we foresaw several synthetic
issues concerning this venture that would require careful
attention and much experimentation. The structures of the
epothilones invite retrosynthetic dissection into two domains.
Thus the polypropionate domain (C1-C8) and the “O-alkyl”
sector (C12-C15) of the lactone are joined through a third
achiral “hinge” region (C9-C11) which consists of three
contiguous methylene groups (Scheme 1). The polypropionate
sector (C1-C8) comprises four discontiguous stereocenters
flanked by two relatively labile â-hydroxy carbonyl moieties.
The O-alkyl sector contains the C12-C13 cis-epoxide moiety
that is insulated by a single methylene group from C15. The
latter bears an allylic alcohol and an R,â-unsaturated thiazole
linkage.
Scheme 1 outlines the retrosynthetic strategy that was
employed for our newly developed chemical synthesis of the
epothilones. Although, the structural issues addressed above
were not insurmountable, several significant obstacles were
encountered during the course of our pursuit.
The synthesis of the C1-C11 domain of the epothilones has
proven most challenging, and indeed, this region had been the
scene of many variations in strategy and synthetic design.13a,15
In our first synthesis of the C1-C11 polypropionate domain,
we chose to contain the stereocenters at C6, C7, and C8 in a
rigid cyclic template with the general structure 5, Scheme 2.16,17
The requisite cyclic template was synthesized as outlined in
Scheme 2 via a Lewis Acid catalyzed diene-aldehyde cyclo-
condensation (LACDAC)18 reaction which enabled us to
accomplish the central goal of stereospecificity by translating
the absolute stereochemistry of the template to C6, C7, and C8
of epothilone.12,16,17,19 Through a series of steps, the resultant
dihydropyran-4-one, 5, could be converted to the desired
(11) Kowalski, R. J.; Giannakakou, P.; Hamel, E. J. Biol. Chem. 1997,
272, 2534.
(12) (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, 2093-2096. (b) Balog, A.; Meng, D.; Kamenecka, T.;
Bertinato, P.; Su, D.-S.; Sorenson, E. J.; Danishefsky, S. J. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 2801.
(13) (a) Nicolaou, K. C.; Roschanger, F.; Vourloumis, D. Angew. Chem.,
Int. Ed. Engl. 1998, 37, 2014-2045 and references therein. For recent
chemical syntheses of epothilone B, see: (b) Mulzer, J.; Mantoulaidis, A.;
Ohler, E. Tetrahedron Lett. 1998, 39, 8633. (c) Taylor, R. E.; Galvin, G.
M.; Hilfiker, K. A.; Chen, Y. J. Org. Chem. 1998, 63, 9580. (d) May, S.
A.; Grieco, P. A. J. Chem. Soc., Chem. Commun. 1998, 1597. (e) Schinzer,
D.; Bauer, A.; Schieber, J. Synlett 1998, 861. (f) White, J. D.; Carter, R.
G.; Sundermann, K. F. J. Org. Chem. 1999, 64, 684.
(15) Harris, C. R.; Balog, A.; Savin, K.; Danishefsky, S. J.; Chou, T.-
C.; Zhang, X.-G. Societe de Chimie Therapeutique, manuscript submitted
for publication.
(16) Bertinato, P.; Sorensen, E. J.; Meng, D.; Danishefsky, S. J. J. Org.
Chem. 1996, 61, 8000.
(17) Meng, D.; Sorensen, E. J.; Bertinato, P.; Danishefsky, S. J. J. Org.
Chem. 1996, 61, 7998.
(18) (a) Danishefsky, S. J. Aldrichim. Acta 1986, 19, 59. (b) Danishefsky,
S. J. Chemtracts 1989, 2, 273.
(14) For an initial communication regarding this work please see the
following reference: Balog, A.; Harris, C.; Savin, K.; Zhang, X. G.; Chou,
T.-C.; Danishefsky, S. J. Angew. Chem., Int. Ed. Engl. 1998, 2675-2678.
(19) Heathcock, C. H.; Young, S. D.; Hagen, J. P.; Pirrung, M. C.; White,
C. T.; VanDerveer, D. J. Org. Chem. 1980, 45, 3846.

7052 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
Scheme 2. First Generation Synthesis of the Epothilones
Using Lewis Acid Catalyzed Diene-Aldehyde
Cyclocondensation (LACDAC) Chemistry
dEpoB was clearly superior to paclitaxel.20 In light of these
findings, we sought a chemical synthesis of dEpoB that would
be able to support a serious and substantial discovery research
program directed toward the clinical development of this
apparently useful compound. Toward this end, we undertook a
major departure from our original academic syntheses. The
governing concepts for our new synthesis are detailed below.
Our overall strategy for this endeavor assumed the ability to
synthesize dEpoB from the three constructs A, B, and C through
the key steps outlined in Scheme 1. We envisioned two possible
routes in confronting this challenge: First, a B-alkyl Suzuki
merger of constructs B and C, followed by a post-Suzuki aldol
condensation between the Suzuki-coupled product (B + C) and
the enolate of A, {e.g., [A + (B + C)]} (Scheme 1). Our second
itinerary projected a key aldol condensation between constructs
A and B, followed by Suzuki coupling to effect the merger of
the aldolate (A + B) with the vinyl iodide, C, {e.g., [(A + B)
+ C]} (Scheme 1).
Consequently, our new synthetic strategy hinged on three vital
requirements, each of which was crucial to the success of the
venture. Our stipulations assumed that the troublesome C1-
C11 polypropionate domain could be assembled through an aldol
condensation of the C5-C6 (Z)-metalloenolate construct A (or
its equivalent) with the chiral aldehyde B or the post-Suzuki
aldehyde (B + C). Fulfilling this goal would require a high
level of asymmetric control in the aldol reaction as well as
excellent regiocontrol in the addition of chiral aldehyde B or
the post-Suzuki aldehyde (B + C) to enolate A. Our second
tenet assumed the ability to perform successfully a B-alkyl
Suzuki merger (Vide infra) between a construct of the form B
or the elaborate aldol condensation product (A + B) and vinyl
iodide C. Finally, a late-stage reduction of the C3 ketone to the
requisite C3 alcohol with high asymmetric induction would be
essential. The ability to successfully control the desired C3
stereochemistry through catalytic asymmetric reduction was a
necessary element as it permitted us to introduce the C1-C7
fragment, A, into the synthesis as a readily accessible achiral
building block.
epothilone precursors, 9 or 10, through a B-alkyl Suzuki
coupling with an appropriate thiazole-containing vinyl iodide,
8. Although this route proved rather lengthy, the first successful
synthesis of both epothilones A and B, performed in these
laboratories, was achieved through this strategy, utilizing either
macroaldolization or macrolactonization reactions.
Dianion Equivalents Corresponding to the Polypropionate
Domain. Two tactical approaches emerged during the evolution
of our synthesis. Initially the aldol reaction was performed
directly between aldehyde 13a21 and the dianion 12 derived from
tricarbonyl 11. In this way, it was indeed possible to generate
the (Z)-lithium enolate of 11, which underwent successful aldol
condensation as shown in Scheme 3.22 Unfortunately, the
resultant C7 â-hydroxyl functionality tended to cyclize to the
C3 carbonyl group, thereby affording a rather unmanageable
mixture of hydroxy ketone 14a and lactol 14b products. Lactol
formation could be reversed following treatment of the crude
aldol product under the conditions shown (Scheme 3); however,
under these conditions an inseparable 3-4:1 mixture of dia-
stereomeric products, 15(a or b):16 (a or b), was obtained.
Progress along this avenue was further impeded when it later
became apparent that neither C7-acetate nor C7-TES-protecting
groups were compatible with the remainder of the synthesis.
Attempts to introduce more durable blocking groups at the C7
hydroxyl center, by diversion of the desired hydroxy ketone
14a, were unsuccessful.
A New Synthesis of Epothilone B. While our initial synthetic
enterprise in the epothilone field afforded fully synthetic
epothilone B in quantities necessary for in Vitro cytoxicity and
tubulin binding studies, it was only through a tenacious effort
that suitable amounts could be prepared for extensive in ViVo
studies against xenografts of human tumors in nude mice. Our
initial in ViVo studies of the parent agent suggested some
potentially serious toxicity problems and an all too narrow
window of possible therapeutic advantage.20b-d Fortunately
however, experiments with 12,13-desoxyepothilone B (dEpoB)
were more promising (Vide infra).20a,b Surprisingly, these studies
revealed that although the in ViVo and in Vitro potency of dEpoB
is lower than 2, the 12,13-desoxy system dramatically outper-
forms paclitaxel in ViVo. We were further enticed to pursue this
direction of research when our preliminary in ViVo studies
established that the desoxy compounds were well tolerated and
virtually curative against a variety of xenograft tumors.20
Moreover, in parallel studies with drug resistant xenografts,
(20) (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.; Savin, K.; Bertino, J. R.; Danishefsky, S. J. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 9642. (c) Su, D.-S.; Meng, D.; Bertinato, P.; Balog,
A.; Sorensen, E. J.; Danishefsky, S. J.; Zheng, Y.-H.; Chou, T.-C.; He, L.;
Horwitz, S. B. Angew. Chem., Int. Ed. Engl. 1997, 36, 757. (d) Balog, A.;
Bertinato, P.; Su, D.-S.; Meng, D.; Sorensen, E. J.; Danishefsky, S. J.; Zheng,
Y.-H.; Chou, T.-C.; He, L.; Horwitz, S. B. Tetrahedron Lett. 1997, 38,
4529.
(21) Aldehyde 13a was readily prepared according to the procedure
outlined in: Lin, N.-H.; Overman, L. E.; Rabinowitz, M. H.; Robinson, L.
A.; Sharp, M. J.; Zablocki, J. J. Am Chem. Soc. 1996, 118, 9062.
(22) (a) Mori, I.; Ishihara, K.; Heathcock, C. H. J. Org. Chem. 1990,
55, 1114. (b) Heathcock, C. H. In Asymmetric Synthesis; Morrison, J. D.,
Ed.; Academic: New York, 1984; Vol. 3, p 111. (c) Evans, D. A. In
Asymmetric Synthesis; Morrison, J. D., Ed.; Academic: New York, 1984;
Vol. 3, p 2.

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7053
Scheme 3a
Scheme 4a
a (a) 2.2 equiv LDA, -50 °C, (S)-2-methyl-4-pentenal (13a), (ca.
60%); (b) Ac2O, DMAP, Et3N, CH2Cl2, 18 h, (60%, two steps); (c)
TESCl, imidazole, DMF, 18 h, (60%, two steps).
Although our initial attempts in this regard proved unavailing,
we were able to confirm that the critical aldol reaction between
the dienolate 12 and (S)-aldehyde 13a did indeed provide the
desired C6-C7 syn stereochemistry concurrent with C7-C8
anti relationship (by anti-Felkin-Anh addition) as the major
diastereomer. Initially, we sought to confirm the resident
stereochemistry at C6, C7, and C8 by converging the protected
aldolate products 15a or 15b with compound 19 which could
be prepared independently (Scheme 4) from previously char-
acterized intermediates in our LACDAC route.12,20,23 Unfortu-
nately, it was not possible to prepare the C7 TBS-protected
compound, 19, using our aldol chemistry due to complications
arising from hemiacetal formation (Vida supra). Likewise,
several attempts to hydrolyze the C7-TBS of 19 resulted in
decomposition of the tricarbonyl system. However, comparison
of the 1H NMR spectra of compounds 15b (C7-TES-protected,
aldol chemistry) and 19 (C7-TBS-protected, LACDAC chem-
istry) revealed similar chemical shifts and multiplicities of
coupling constants which suggested the likelihood of corres-
pondence at the level of stereochemistry. This finding encour-
aged us to pursue our course of research. However, to ascertain
a (a) Dess Martin, (86-88%); (b) HF‚pyridine, (90%); (c) dimethyl
sulfate, K2CO3, reflux (50%); (d) NaH, TESOTf, -30 °C (73%); (e)
TMSCHN2, i-Pr2NEt, MeOH, CH3CN, (74%); (f) (S)-2-methyl-4-
pentenal (13a), LDA, -120 °C, (50-60% of major diastereomer); (g)
Ac2O, DMAP, Et3N, (98%); (h) TBSOTf, 2,6-lutidine, (87%).
unambiguously that the major diasteroemer in the aldol reaction
between 11 and 13 was indeed the desired compound would
require a more rigorously supported argument.
Realizing that the major impediment in this regard was the
untoward lactolization of 14a to the hemiacetal, 14b, we
considered the possibility of simply engaging the C3-carbonyl
group of the nucleophile in another functional arrangement such
as an enol ether (Scheme 4). In this way, unwanted cyclization
would not occur, and the C7-hydroxyl group could be readily
protected with a more robust moiety. Initially, methyl enol ether
(21a) and triethylsilyl (TES) enol ether (21b)14 protecting groups
proved useful for our purposes (Scheme 4). Rewardingly, 21a
and 21b each successfully underwent the aldol coupling with
(S)-aldehyde 13a to afford a 1:5-6 mixture of diastereomers
that were separable by flash column chromatography. Given
the circumvention of the difficulties encountered by undesired
engagement of the C7-hydroxyl moiety, the latter could be
readily protected as triethylsilyl (TES), acetate (Ac), or (tert-
butyldimethylsilyl) TBS ethers. As a result, we were able to
converge the polypropionate domain (prepared through our aldol
(23) (a) Meng, D.; Bertinato, P.; Balog, A.; Su, D.-S.; Kamenecka, T.;
Sorensen, E. J.; Danishefsky, S. J. J. Am. Chem. Soc. 1997, 119, 10073.
(b) Meng, D.; Su, D.-S.; Balog, A.; Bertinato, P.; Sorensen, E. J.;
Danishefsky, S. J.; Zheng, Y.-H.; Chou, T.-C.; He, L.; Horwitz, S. B. J.
Am. Chem. Soc. 1997, 119, 2733.

7054 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
chemistry) with compound 20, prepared from previously
characterized intermediates in our LACDAC synthesis of the
polypropionate region (Scheme 4).12,20 In the event, the tricar-
bonyl system 19, obtained from the previously described
LACDAC chemistry, was successfully converted to the methyl
Scheme 5a
1
enol ether 20, and this compound displayed an identical H
NMR with the TBS-protected aldolate isolated from the aldol
condensation of methyl enol ether 21a and aldehyde 13a
(Scheme 4). Similarly, the C7-protected aldolate product,
isolated from the aldol condensation of the methyl enol ether
21a and aldehyde 13a, exhibited identical 1H NMR with methyl
enol ether 23a derived from C7 acetate 15a (obtained from the
aldol reaction of tricarbonyl 11 and aldehyde 13a). These results
demonstrated conclusively that the aldol reactions provided both
good diastereoselectivity and the requisite syn connectivity
between the stereocenters at C6-C7 and the necessary anti
relationship relative to the resident chirality at C8. We shall
digress at a later point to discuss the apparent reason for this
unexpected anti-Felkin-Anh diastereoselection.
Aldol Condensation Following Suzuki Coupling. Having
confirmed the ability to successfully perform the coupling of
enolate A and chiral aldehyde B, generating the required C7-
C8 stereochemical relationship, we sought to address the
prospect of performing the aldol condensation on the highly
elaborate chiral aldehyde formed from the merger of constructs
B and C, Scheme 1. In the event, we were able to prepare the
requisite aldehyde beginning with the C1 TES-protected alcohol,
24 (Scheme 5). The TES ether 2424 readily underwent hydro-
boration with 9-BBN and subsequent B-alkyl Suzuki coupling
with vinyl iodide C25 in good yield and with good stereochem-
ical integrity about the olefin to afford 25a or 25b. Selective
TBAF26 deprotection of the C1 TES-protecting group and
subsequent Swern oxidation27 afforded the desired aldehyde,
26a or 26b. Aldol condensation of aldehyde 26 with the lithium
enolate of either 21a or 21b afforded consistently a 2.5:1 mixture
of diastereomeric products (C7-C8, anti:syn) in 62-68%
combined yield. Attempts to lower reaction temperature and
modify the solvent and/or base proved ineffective at improving
the apparent diastereoselectivity of the coupling step. Given the
poor diastereoselectivity observed in the coupling reaction of
12 and 26a/b, we redirected our attention toward the more
successful coupling observed between enol ethers 21a/b and
the pre-Suzuki aldehyde 13a.
a (a) 9-BBN, 24; then C, Cs2CO3, Pd(dppf)2Cl2, Ph3As, H2O; (b)
TBAF, 0 °C, (82% (two steps)); (c) oxalyl chloride, DMSO; Et3N,
(87%); (d) LDA, 21a or 21b, -100 °C; R ) Me, (62%, 2.5:1 mixture
of diastereomers); R ) TES, (68%, 2.5:1 mixture of diastereomers),
major diastereomer shown.
As a result of this apparent failure, we investigated the
usefulness of the methyl enol ether linkage discussed above.28
We were happy to discover that the requisite methyl enol ether,
21a, could be readily prepared in multigram quantities from
trimethylsilyl diazomethane, TMSCHN2, with Hunig’s base in
high yield (Scheme 4).29 Likewise, hydrolysis of the C3 methyl
enol ether could be effected, without epimerization at C6, by
the use of p-toluenesulfonic acid (p-TSA) in acetone at room
temperature (Scheme 6).
At this point, we directed our attention to an appropriate
protecting group for the C7 alcohol. We were concerned that
the highly acidic nature of the Noyori reduction protocol (Vide
infra) would not be compatible with the survival of silyl (TES
or TBS) functionality at C7. Therefore, we considered a more
robust functionality at C7 such as acetate; however, on the basis
of preliminary studies, we foresaw difficulty in successfully
removing the C7 acetate from the lactonized dEpoB precursor
without concomitant destruction of a highly advanced macro-
cycle. Hence, we sought a more dischargeable functionality to
protect the C7 alcohol. At this point, we directed our attention
toward slightly more labile protecting groups that would still
be expected to survive the acidic conditions of the asymmetric
Noyori reduction. We investigated the use of both chloroacetyl
Synthesis of C1-C11 Polypropionate Domain. With the
desired C1-C11 domain, 22a/b, in hand, we initially attempted
to effect the synthesis of dEpoB with the C3 TES enol ether,
22b (Scheme 4). We felt that a C3 TES enol ether intermediate
would be advantageous as this group could be readily removed
under mild conditions. While we did accomplish our first total
synthesis of dEpoB with material prepared via this route, our
initial success in this regard did not translate well upon attempted
scale-up (>1.0 g). On larger scales, the aldol reaction between
21b and aldehyde 13a proved to be troublesome. In time we
learned that the resultant C3 TES enol ether in 22b was prone
to decomposition under the very basic conditions of the aldol
reaction and, moreover, was quite sensitive to silica gel
chromatography.
(24) The C-1 TES ether was readily prepared from the Overman alcohol
described in ref 21 by the use of TESOTf and 2,6-lutidine in CH2Cl2 at
-78 °C for 30 min.
(25) See Supporting Information for the synthesis of the previously
characterized vinyl iodide and its precursors.
(26) Corey, E. J.; Snider, B. B. J. Am. Chem. Soc. 1972, 94, 2549.
(27) Roush, W. R.; Palkowitz, A. D.; Ando, K. J. Am. Chem. Soc. 1990,
112, 6348.
(28) Harris, C. R.; Kuduk, S. D.; Savin, K.; Balog, A.; Danishefsky, S.
J. Tetrahedron Lett. 1999, 40, 2263.
(29) Aoyama, T.; Terasawa, S.; Sudo, K.; Shiori, T. Chem. Pharm. Bull.
1984, 32, 3759.

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7055
Scheme 6a
arrangement might be sensitive to the hydroboration conditions
necessary for the preparation of the organoborane at C11.
However, our concerns were allayed as the coupling step was
accomplished without difficulty. The crude B-alkyl Suzuki-
coupled product was best characterized as the C15 alcohol after
hydrolysis of the TBS-protecting group to afford the requisite
C15-hydroxy ester 32. With this elaborate acyclic precursor (32)
in hand, we now directed our thoughts to the asymmetric
reduction of the C3 ketone.
Stereoselective Noyori Reduction. Indeed, the selective
asymmetric reduction of the C3 ketone now loomed as the key
issue on which the completion of the synthesis of dEpoB hinged.
The choice of a suitable reducing agent for this transformation
would be determined with the goal of optimizing diastereose-
lectivity and chemoselectivity. Selective asymmetric reduction
of the C3 ketone without concomitant reduction of the ketone
at C5 would be imperative. Furthermore, concurrent reduction
of the two olefinic moieties present in the acyclic precursor 32
must be avoided.
We soon directed our focus toward the excellent chiral
recognition displayed by the atropisomeric [2,2′-bis(diarylphos-
phino)-1,1′binaphthyl]ruthenium {Ru(BINAP)} species which
has been used occasionally in the total synthesis of several
naturally occurring molecules.34 Noyori35 and others36 have
demonstrated that these Ru(BINAP) derivatives possess the
remarkable ability to catalyze the asymmetric reduction of
â-keto esters. Because of the inherent ability of the ruthenium
species to chelate the 1,3-dicarbonyl arrangement, it is readily
evident that â-keto esters that contain functionality in the
δ-position could affect the stereo- and regiocontrol of the
catalyst upon reduction due to competing coordination modes
and multiple ligation possibilities with Ru(BINAP) in a system
such as ours. We were encouraged to discover that in the
reduction of â,δ-diketo esters, the C3 ketone is at first selectively
hydrogenated to give the 3-hydroxy ester and then subsequently
the C5-carbonyl group is hydrogenated to afford the dihydroxy
ester.37 We were hopeful that the presence of the gem-dimethyl
functionality in 32 would preclude subsequent reduction of the
C5 ketone under the Noyori reduction conditions.
a (a) TrocCl, pyridine, CH2Cl2, 0 °C; (b) p-TSA, acetone, rt, 6 h,
(87%, two steps); (c) 9-BBN, 30; Cs2CO3, Pd(dppf)2Cl2, Ph3As, DMF,
31(C), H2O, (ca. 75%); (d) 0.5 M HCl/MeOH, (85%).
and methoxyacetyl moieties; however, these too proved inef-
fective as they were difficult to remove without concomitant
destruction of the macrocycle by ring-opening at the lactone or
retro-aldol of the â-hydroxy carbonyl moieties.
We finally turned our attention to a 2,2,2-trichlorethyoxy-
carbonate (Troc)-protecting group at C7. Indeed a similar
protective moiety had been utilized by Evans, et al.30 in the
total synthesis of the macrolide antibiotic cytovaricin. We were
pleased to discover that the Troc function could readily be
introduced through the action of compound 22a (or 22b) with
TrocCl and pyridine in CH2Cl2 (Scheme 6). However, we
foresaw difficulties in the standard reductive removal of the
Troc-protecting group (Zn/HOAc/heat), given the possibility of
concurrent reduction of the C5 ketone under the strongly
reducing conditions. However, the Evans report demonstrates
the ability of the mild reducing agent samarium(II) iodide,
SmI2,31 to cleanly remove a 2,2,2-trichloroethoxycarbamate, and
we hoped to extend this protocol to the reductive removal of
the C7 Troc-protected alcohol of epothilone B. We anticipated
that with these mild reaction conditions (SmI2, -78 °C) the
sterically encumbered C5 ketone would not be reduced in the
absence of an added proton source. With the C7 issue addressed,
we could direct our focus toward the merger of the highly
elaborate C1-C11 domain with the vinyl iodide moiety, C,
through a B-alkyl Suzuki coupling.
Recognizing that the conditions of the Noyori reduction may
also effect the facile reduction of isolated olefinic moieties
present in the molecule, we were concerned that asymmetric
reduction of the C3 ketone might be complicated by undesired
reduction of the C12-C13 and C15 allylic olefin species.
However, we were pleased to find that under mild conditions,
the modified Noyori catalyst [RuCl2(BINAP)]2‚NEt3], prepared
by heating Ru(COD)Cl2, BINAP, and Et3N at 140 °C in toluene,
would catalyze the hydrogenation of the 3-oxo group in our
substrate without simultaneous reduction of remote di- and
trisubstituted olefins.38 Encouraged by these previous examples,
we set out to attempt the asymmetric reduction on our â,δ-
diketo ester system.
B-Alkyl Suzuki Merger. The second fundamental component
of our new total synthesis of epothilone B involved a very
ambitious B-alkyl Suzuki32 merger, Scheme 6. The hope was
to couple the previously described vinyl iodide12,20,33 31(C) and
the rather elaborate tricarbonyl 30 to gain access to the requisite
C15 hydroxy ester in preparation for end-game macrolacton-
ization. We were somewhat apprehensive that the tricarbonyl
(34) (a) Taber, D. F.; Deker, P. B.; Silverberg, L. J. J. Org. Chem. 1992,
57, 5990. (b) Taber, D. F.; Silverberg, L. J.; Robinson, E. D. J. Am. Chem.
Soc. 1991, 113, 6639. (c) Schreiber, S. L.; Kelly, S. E.; Porco, J. A.;
Sammakia, T.; Suh, E. M. J. Am. Chem. Soc. 1988, 110, 6210. (d) Taber,
D. F.; Wang, Y. J. Am. Chem. Soc. 1997, 119, 22.
(30) Evans, D. A.; Kaldor, S. W.; Jones, T. K.; Clardy, J.; Stout, T. J. J.
Am. Chem. Soc. 1990, 112, 7001.
(31) Molander, G. A. In Organic Reactions; Paquette, L. A., Ed.; John
Wiley and Sons: New York, 1994; Vol. 46, p 211.
(32) (a) Miyaura, N.; Ishiyama, T.; Sasaki, H.; Ishikawa, M.; Satoh, M.;
Suzuki, A. J. Am. Chem. Soc. 1989, 111, 314. (b) For an excellent review,
see: Miyaura, N.; Suzuki, A. Chem. ReV. 1995, 95, 2457. (c) Johnson, C.
R.; Braun, M. P. J. Am. Chem. Soc. 1993, 115, 11014.
(35) Noyori, R. Tetrahedron 1994, 50, 4259 and references therein.
(36) Ager, D. J.; Laneman, S. A. Tetrahedron: Asymmetry 1997, 8, 3327
and references therein.
(37) (a) Shao, L.; Seki, T.; Kawano, H.; Saburi, M. Tetrahedron Lett.
1991, 32, 7699. (b) Shao, L.; Kawano, H.; Saburi, M.; Uchida, Y.
Tetrahedron 1993, 49, 1997.
(33) Su, D.-S.; Meng, D.; Bertinato, P.; Balog, A.; Sorensen, E. J.;
Danishefsky, S. J.; Zheng, Y.-H.; Chou, T.-C.; He, L.; Horwitz, S. B. Angew.
Chem., Int. Ed. Engl. 1997, 36, 757.
(38) (a) Taber, D. F.; Silverberg, L. J. Tetrahedron Lett. 1991, 32, 4227.
(b) Taber, D. F.; Silverberg, L. J.; Robinson, E. D. J. Am. Chem. Soc. 1991,
113, 6639.

7056 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
Development of Conditions and Substrate. It was soon
apparent that the asymmetric Noyori reduction of various diketo
esters in this series was critically dependent upon the amount
of acid present in the reaction. Thus, the presence of stoichio-
metric acid was required for reduction of the C3 ketone with
both good diastereo- and chemoselectivity. In the absence of
acid, no reduction of the ketone was observed. However,
complete reduction of both olefinic moieties was apparent upon
inspection of the 1H NMR spectrum of the crude reaction
product. In the presence of only 0.5 equiv of HCl, the
diastereoselectivity of the asymmetric reduction was quite poor,
resulting in a reaction mixture comprising a 4:1 mixture of
diastereomeric products at C3 along with an appreciable amount
of product resulting from hydrogenation of the C11-C12 double
bond.
Scheme 7. Competition Experiment of â,δ-Diketo Ester and
â-Keto Ester Substrates Demonstrate Necessity of C5-Ketone
in the Noyori Reduction of These Systems
There are numerous reports citing the dramatic dependence
of the Noyori reduction upon the presence of added strong
acid.34c,36,39 Moreover, it is reported that the addition of Et3N
may retard the hydrogenation, but upon re-acidification, the
original reaction rates are reestablished.36 These failures have
been attributed to the presence of low-level basic impurities in
the substrate.36,39 In our hands, the presence of stoichiometric
HCl was absolutely required for both reduction at C3 and good
diastereoselectivity in the reduction. We suspect that the
presence of the basic thiazolium nitrogen perhaps coordinates
the ruthenium catalyst, resulting in slow reaction rates and poor
levels of asymmetric reduction. However, upon the addition of
stoichiometric HCl, the thiazole moiety is perhaps protonated,
resulting in reestablishment of the appropriate catalytic cycle.
At 25 °C and pressures greater than 1300 psi, the chemo-
selectivity of the reduction was poor, resulting in the reduction
of both alkene groups. Pressures below 1000 psi (and ambient
temperature) resulted in dramatically slower rates of reduction
along with poorer diastereoselectivity in the reduction at C3.
While there are reports of asymmetric Noyori reductions at
elevated temperatures (ca. 80 °C) occurring at even lower
pressures (50 psi),36 we encountered decomposition of the
epothilone substrates at these elevated temperatures. As a result,
our standard Noyori asymmetric reductions were performed at
25 °C and 1200 psi.
Scheme 8a
We were surprised to discover that the nature of the substrate
played a significant role in the observed chemo- and stereo-
selectivity. However, since it is the BINAP ligand that provides
the chiral environment and induces enantioselectivity through
differentiation of the possible diastereomeric transition states,
it should not be surprising that the observed stereo- and
chemoselectivity in the reduction of these chiral ketonic
substrates is markedly affected by the preexisting steric require-
ments and stereogenic centers present in the substrate. Indeed,
we were fascinated to discover that the carbonyl group at C5
was never reduced under any of the attempted reaction condi-
tions but was absolutely necessary for the reduction of the C3-
carbonyl. We attribute this result to the presence of the gem-
dimethyl functionality bridging the adjoining carbonyl groups.
Thus, in a competition study between the two â-ketoesters, 33
and 34, tricarbonyl 33 was completely reduced at C3, whereas
the â-ketoester 34, with C5 in the alcohol oxidation state, was
unwavering toward reduction (Scheme 7).
a (a) (S)-5,5-dimethyl-4-pentenal (13b), LDA, (60%, major diaste-
reomer); (b) TBSOTf, 2,6-lutidine, CH2Cl2, -78 °C, (95%); (c) 0.1 M
HCl/MeOH, (87%); (d) R-(BINAP)RuCl2, 1200 psi, 1.1 equiv HCl,
MeOH, rt, 7-8 h, (87-88%); (e) O3, DMS, (86%); (f) Ph3PdCH2,
(96%).
aldolate product 15a (eq 1); however, as anticipated, the terminal
olefin in 15a was rapidly reduced even under mild reaction
conditions. Since there have been reports of chemoselective
reduction of â-ketoesters in the presence of trisubstituted olefinic
species,34b,38a we prepared the analogous substrate 38 containing
a trisubstituted double bond (Scheme 8). Thus, the aldol
condensation of 5,5-methyl-4-pentenal40 with 21b, afforded a
4.5:1 mixture of diastereomeric aldol products (38, major
diastereomer) which could be separated by flash column
chromatography. After protection of the C7 alcohol as the TBS
ether, the resultant trisubstituted olefin, 39, could be cleanly
reduced under our standard Noyori reduction conditions to
Thus, with these preliminary results in hand, we were
prepared to examine the asymmetric Noyori reduction in the
context of the total synthesis of dEpoB. Initially, we sought to
perform the asymmetric Noyori reduction on the pre-Suzuki
1
provide a single diastereomer 40 (by H NMR) in good yield.
(39) King, S. A.; Thompson, A. S.; King, A. O.; Verhoeven, T. R. J.
Org. Chem. 1992, 57, 6689.
(40) Aldehyde 13b was readily prepared according to the procedure
outlined in ref 21.

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7057
Protection of the C3 alcohol as the TBS ether and subsequent
ozonolysis and Wittig reaction afforded the previously charac-
terized 4112,20 (Scheme 8) which could be subjected to B-alkyl
Suzuki coupling and constituted a formal total synthesis of
epothilone B.
Scheme 9a
Although, the asymmetric Noyori reduction on the pre-Suzuki
aldolate product proved effective for the formal total synthesis
of dEpoB in Scheme 8, we felt that the synthesis could be made
more concise by performing the asymmetric reduction on the
post-Suzuki substrate, 32. Indeed, the Noyori reduction of ketone
32 with the dimerized Noyori catalyst [RuCl2((R)-BINAP)]2-
[NEt3] (42) afforded the desired diol 43 with excellent diaste-
1
reoselectivity (>95:5, no minor diastereomer observed by H
NMR). An additional benefit of avoiding the ancillary ozonolysis
and Wittig transformations (compare Schemes 8 and 9) was
also realized. Having successfully controlled the C3 stereo-
chemistry by reduction of the late-stage intermediate 32, we
were in a position to continue with the completion of the
synthesis of dEpoB (and EpoB).
Completion of the dEpoB (EpoB) Synthesis. With these
critical issues addressed, no significant obstacles remained in
the total synthesis and the conversion of 43 to dEpoB (47),
Scheme 9. Thus, using previously developed methods, the
dihydroxy ester 43 was protected at C3 and C15 as the
triethylsilyl ether and the tert-butyl ester successfully hydrolyzed
to the triethylsilyl ester through the agency of triethylsilyl
trifluoromethanesulfonate (TESOTf) and 2,6-lutidine.20,23 The
resultant C3,C15-bis(triethylsilyl)-protected diol was selectively
hydrolyzed to the requisite C15-hydroxy acid 44. Subsequent
macrolactonization of 44 under Yamaguchi conditions41 afforded
the fully protected macrolactone 45. We were pleased to find
that samarium (II) iodide deprotection of the 2,2,2-trichloro-
ethoxy carbonate (Troc) 45 at C7 proceeded smoothly to afford
the desired C7 alcohol 46 in excellent yield.42 Finally, standard
HF‚pyridine deprotection of the C3 triethylsilyl group afforded
the desired dEpoB 47. The C12-C13 olefin could be selectively
epoxidized20 to afford the natural product epothilone B (2) with
the usual high degree of chemo- and stereoselectivity. For our
purposes, however, dEpoB was the target molecule.
Remarkable Long-Range Effects on the Diastereoface
Selectivity Observed in Aldol Condensations. Traditional
models for diastereoface selectivity were first proposed by Cram
and later by Felkin for predicting the stereochemical outcome
of aldol reactions occurring between an enolate and chiral
aldehyde.43 During our investigations directed toward a practical
synthesis of dEpoB, we were puzzled to discover an unforeseen
bias in the relative diastereoface selectivity observed in the aldol
condensation between the (Z)-lithium enolate B and R-methyl
aldehyde C, Schemes 3 and 4. The aldol reaction proceeds with
a (a) [RuCl2((R)-BINAP)]2[NEt3] (42), H2, 1200 psi, MeOH, HCl,
25 °C, 7 h, (82-88%); (b) TESOTf, 2,6-lutidine, CH2Cl2, -78 °C to
rt; (c) then 0.1 N HCl/MeOH (70-77%, two steps); (d) 2,4,6-
trichlorobenzoyl chloride, TEA, 4-DMAP, PhCH3, (78%); (e) SmI2,
cat. NiI2, -78 °C, THF, (90-95%); (f) HF‚pyridine, THF, 0 °C, (98%);
(g) DMDO, CH2Cl2 (80%).
the expected simple diastereoselectivity providing exclusively
the C6-C7 syn stereochemistry, however, the C7-C8 relation-
ship of the principal product was anti (Schemes 3 and 4).44
Indeed, the diastereoselectivity exhibited in the aldol condensa-
tion between A and B (Scheme 1) occurred with the opposite
sense of diastereoface addition to that predicted, according to
the models for relative face diastereoselection encompassed in
the Felkin rules.45 We were prompted to investigate the cause
for this unanticipated but fortunate occurrence.
A survey of the previous literature revealed that Evans,46
Hoffman,47 Heathcock,22c,44 and Roush48 have examined similar
systems in the recent past. Their studies demonstrated that the
Felkin-Anh rules for diastereoselection fail to adequately
rationalize the results of many reactions involving (Z)-O-enolates
(41) (a) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M.
Bull. Chem. Soc. Jpn. 1979, 52, 1989. (b) Mulzer, J.; Mareski, P. A.;
Buschmann, J.; Luger, P. Synthesis 1992, 215.
(42) (a) Molander, G. A.; Harris, C. R. J. Org. Chem. 1997, 62, 7418.
(b) Machrouhi, F.; Namy, J.-L.; Kagan, H. B. Synlett 1996, 633.
(43) Lodge, E. P.; Heathcock, C. H. J. Am. Chem. Soc. 1987, 109, 2819
and references therein.
(44) Heathcock, C. H. In Asymmetric Synthesis; Morrison, J. D., Ed.;
Academic Press: New York, 1984; Vol. 3, pp 111-213.
(45) (a) Mukaiyama, T. Org. React. 1982, 28, 203. (b) Masamune, S.;
Choy, W.; Petersen, J. S.; Sita, L. R. Angew. Chem., Int. Ed. Engl. 1985,
24, 1.

7058 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
Table 1. Diastereoselectivities Observed in the Aldol Reaction of
Enolate 48 (R′ ) Me or TES) with Aldehyde 49
Figure 1. Proposed transition-state models for the aldol condensation
of aldehydes 49 and enolate 48 in Table 1.
((Z)-crotylboronates)47a,b and allylboron reagents47,48 with
R-branched chiral aldehydes. Examination of the Felkin-Anh
transition state demonstrated the presence of a destabilizing
gauche-pentane interaction which causes the anti-Felkin-Anh
product to be formed preferentially.46
Further examination revealed that our data were reconcilable
when viewed in the above context. In the early 1980s, Evans46
proposed a transition-state model based on the above findings
which is supported by our recent discoveries. However, recently,
Roush has proposed a more refined model to account for attrition
in anti selectivity with R-methyl chiral aldehydes.49 The Roush
model proposes that in the reacting Curtin-Hammett conformer,
the R group (larger group) of the aldehyde is distanced from
the R′ group of the enolate (I) to avoid an unfavorable,
potentially serious syn-pentane interaction (Figure 1). Thus, the
basis of the Roush model focuses on minimization of steric
interactions between the largest functional groups of the enolate
and the R-branched aldehyde in the reacting ensemble. Conse-
quently, the observed anti-Felkin-Anh selectivity displayed in
these reactions is typical for such “R-methyl” aldehydes. When
the R group is approximately equal to the methyl center in its
effective A-value, the selectivity for the anti-product must
correspondingly deteriorate.
Our discoveries using aldehyde 13a and related congeners
were unique in that our substrate aldehydes lack the typical
resident protected alcohol derivative that is usually involved in
fashioning anti diastereoface selectivity (Table 1).50 Rather, the
conformational bias in our substrates is apparently the result of
a very particular relationship between the formyl moiety and
the unsaturation in the pendant side chain. As a result of our
studies, we are prompted to speculate that the presence of
unsaturation at C4-C5 in the aldehyde moiety (present in both
13a and 13b, Table 1, entries a and b) provides a subtle
stabilizing, nonbonded interaction between the unsaturation in
the aldehyde and the carbonyl of the enolate (Figure 1, II).51
This interaction can be envisioned to stabilize the transition state
leading to the observed major anti-Felkin diastereomer 51.
Indeed previous examination of aldol reactions of chiral
R-methyl aldehydes has suggested that nonbonded interactions
play an important role in determining aldehyde diastereofacial
selectivities in the reaction of (Z)-enolates.52
The results of several aldol reactions between the enolate 48
and various other aldehydes (49) are depicted in Table 1. Our
extensive studies using various substrates identified a trend in
the consequences of varying the tether length between the formyl
group and the site of unsaturation. Amazingly, reduction of the
double bond of the side chain led to sharply diminished
selectivity affording a 1.3:1 mixture of diastereomeric products
(entry c). Likewise, lengthening of the tether beyond that found
in entry a or b, led to a 2:1 ratio of diastereomers (entry d).
Given these discoveries, the diminished diastereoselectivities
observed in the aldol condensation of 26a or 26b and 21a or
21b (Scheme 5), containing unsaturation in a similar position,
it is not surprising. By contrast, shortening the tether (entry f)
gave strong syn diastereoface selectivity consistent with previous
findings with particular aldehyde 49f.53 The results of entry g,
in which similar steric factors are virtually equivalent (propyl
(46) Evans, D. A.; Nelson, J. V.; Taber, T. R. Top. Stereochem. 1982,
13, 1.
(47) (a) Hoffmann, R. W.; Zeiss, H.-J. Angew. Chem., Int. Ed. Engl.
1980, 19, 218. (b) Hoffmann, R. W.; Weidmann, U. Chem. Ber. 1985, 118,
3966. (c) Hoffmann, R. W.; Brinkmann, H.; Frenking, G. Chem. Ber. 1990,
123, 2387. (d) Brinkmann, H.; Hoffmann, R. W. Chem. Ber. 1990, 123,
2395. (e) Hoffmann, R. W. Angew. Chem., Int. Ed. Engl. 1982, 21, 555.
(48) (a) Roush, W. R.; Adam, M. A.; Walts, A. E.; Harris, D. J. J. Am.
Chem. Soc. 1986, 108, 3422. (b) Roush, W. R.; Adam, M. A.; Harris, D.
J. J. Org. Chem. 1985, 50, 2000. Roush, W. R.; Palkoitz, A. D.; Ando, K.
J. Am. Chem. Soc. 1990, 112, 6348. (c) Roush, W. R. In ComprehensiVe
Organic Synthesis, Heathcock, C. H., Ed.; Pergamon Press: Oxford, 1991;
Vol. 2, p 1.
(49) Roush, W. R. J. Org. Chem. 1991, 56, 4151-4157.
(50) Heathcock, C. H.; Buse, C. T.; Kleschick, W. A.; Pirrung, M. C.;
Sohn, J. E.; Lampe, J. J. Org. Chem. 1980, 45, 1066.
(51) Harris, C. R.; Kuduk, S. D.; Balog, A.; Savin, K.; Danishefsky, S.
J. Tetrahedron Lett. 1999, 40, 2267.
(52) Gennari, C.; Vieth, S.; Comotti, A.; Vulpetti, A.; Goodman, J. M.;
Paterson, I. Tetrahedron 1992, 48, 4439.

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7059
neoplasms.57 As previously discussed, although the epothilones
are chemically dissimilar to the taxanes, they function with a
paclitaxel-like mechanism. Both families of agents apparently
lead to the arrest of cell division and cell death by stabilizing
cellular microtubule assemblies; however, the epothilones have
demonstrated the capacity to retain activity against multidrug-
resistant (MDR) cell lines and tumors where paclitaxel fails.
The multidrug-resistance (MDR) phenotype is characterized by
broad-spectrum resistance to structurally and mechanistically
diverse anticancer agents.11
Figure 2. Potential transition state for diastereoselective aldol reaction
involving a potential lithium cation bridging the olefin (I) or aryl moiety
(II).
versus 2-propenyl at the branching site) demonstrate a small,
but clear preference for the C7-C8 anti product, presumably
reflecting the special effect of the olefin-aldehyde interaction.
The Roche aldehyde,54 (entry h) a substrate well-known for its
tendency to favor the anti-Felkin-Anh adduct,55 performs as
expected to afford a 4:1 mixture of anti:syn diasteromeric
products (entry h).
Acquired resistance to paclitaxel and other commonly used
cancer chemotherapy agents may be mediated by multiple
mechanisms including mutations in microtubule proteins9,11 and
overexpression of the energy-dependent drug transport protein,
P-glycoprotein (PgP).58 Although certain MDR reversal agents
appear promising when co-administered with the anticancer
agent,59 a search for paclitaxel analogues with improved
performance in Vitro and in ViVo has met with only limited
success.3,5
We next wondered whether the unsaturation site could be
encompassed in the context of a properly positioned benzenoid
linkage (Figure 1, III). We were intrigued to discover that
excellent diastereoface selectivity was obtained in the aldol
condensation of the (Z)-lithium enolate with the benzyl substi-
tuted formyl moiety, entry i. Recognizing that functional group
substitution about the aromatic ring would dramatically affect
the donating ability of the ring, we examined the effects of para-
positioned functional groups on the resultant C7-C8 relationship
(entries m-p, Table 1). Indeed, some minor slippage in the anti:
syn ratio is seen in the electron deficient p-bromo substrate
(entry m). The benchmark ratio, seen in entry i, is restored with
the electron rich p-methoxy substrate (entry n), while a small
improvement was realized with the p-dimethylamino derivative
(entry o). Moreover, in the strongly electron-deficient p-
nitrophenyl substrate (entry p), the C7-C8 anti selectivity is
severely abrogated. Clearly, the “aryl effect” is closely coupled
to the electron-donating ability of the ring.
Our ongoing research program in this field focuses on the
relationship between epothilone and paclitaxel in the context
of tumors with the MDR resistance phenomena. Toward this
end, we have applied our rather more practical synthesis toward
the preparation of both dEpoB and similarly promising epothilone
congeners in an effort to provide these compounds in quantities
suitable for both in Vitro and more importantly in ViVo studies.
The synthesis of one such novel analogue was made possible
by using a modified route of the total synthesis described above
(Scheme 4, 6, and 9). The novel analogue 58, which bears a
fused phenyl ring at C12-C13 was prepared as outlined in
Scheme 10. The synthesis of this analogue centered on the
preparation of the aryl iodide moiety, 56, and required little
modification from our newly described synthesis. Thus, zinc-
mediated nucleophilic addition60 of 2-iodobenzylbromide 5361
with aldehyde 5212,20 afforded racemic 54 in 67% yield.
Oxidation of 54 followed by asymmetric reduction to 56
afforded only modest enantioselectivities. The highest enantio-
meric excess observed for the reduction of 55 by methods that
are generally effective using (R)-2-methyl-CBS-oxazaboroli-
dine62 was 60%. While far from ideal, this enantioselectivity
margin was sufficient for our purposes.
By contrast with the reconcilable data observed with para-
substituted substrates, a range of ortho substituents (entries j-l,
Table 1) all resulted in significant weakening of the C7-C8
anti selectivity. We take these data to suggest that ortho
substitution results in some steric inhibition of the rotamer in
which the faces of the aromatic ring and formyl group are
parallel (Figure 1, III).
To interpret these experiments, we suggest that our data points
to a stabilizing through-space interaction of a donor olefinic
linkage with the formyl function as the likely source of
preference of conformers II and III, leading to the sense of
attack anticipated by the Roush model.48
An alternative hypothesis for the enhanced anti-Felkin-Anh
diastereoselectivity that is also supported by these experiments
has been suggested, wherein a Li cation bridges the olefin (or
π system of the aryl moiety) and the enolate (Figure 2).56 Thus,
intramolecular bridging in the transition state could indeed
enhance the facial approach of the enolate that leads to the C7-
C8 anti diastereomer.
Protection of the resultant alcohol as the TBS enol ether
provided the desired thiazole moiety, 56, suitable for Suzuki
coupling. Palladium-mediated Suzuki coupling of the aryl iodide
56 and polypropionate 30 afforded the hydroxy ester 57 which
could be hydrolyzed to the desired C15-hydroxy acid for
Yamaguchi macrolactonization (not shown). The diastereomeric
products (4:1 mixture of diastereomers, epimers at C15 resulting
from incomplete enantioselective reduction) could be separated
by flash column chromatography after macrolactonization to
afford, as the major product, the analogue containing the desired
(57) Landino, L. M.; MacDonald, T. L. In The Chemistry and Pharma-
cology of Taxol and Its DeriVatiVes; Favin, V., Ed.; Elsevier: New York,
1995; Chapter 7, p 301.
(58) Giannakakou, P.; Sackett, D. L.; Kang, Y.-K.; Zhan, Z.; Buters, J.
T. M.; Fojo, T.; Poruchynsky, M. S. J. Biol. Chem. 1997, 272, 17118-
17125 and references therein.
Synthesis of Epothilone Analogues. Taxol (paclitaxel) is
currently a front line therapeutic agent against a variety of solid
forms of cancer including ovarian, breast, colon, lung, and liver
(53) The stereochemistry of the major diastereomer (50f) obtained in
the aldol condensation of 48 with 49f was confirmed by comparing the
aldol product 50f to the identical compound prepared through an independent
synthesis with known stereochemistry.
(54) Roush, W. R.; Hoong, L. K.; Palmer, M. A. J.; Straub, J. A.;
Palkowitz, A. D. J. Org. Chem. 1990, 55, 4117.
(55) Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1984, 23, 556.
(56) (a) Posner, G. H.; Lentz, C. M. J. Am. Chem. Soc. 1979, 101, 934.
(b) Ghera, E.; Kleinman, V.; Hassner, A. J. Org. Chem. 1999, 64, 8.
(59) Ojima, I.; Bounaud, P.-Y.; Bernacki, R. J. CHEMTECH 1998, 28,
31-36.
(60) (a) Berk, S.; Knochel, P.; Yeh, M. C. P. J. Org. Chem. 1988, 53,
5789. (b) Yeh, M. C. P.; Knochel, P.; Santa, L. E. Tetrahedron Lett. 1988,
29, 3887.
(61) L’abbe, G.; Leurs, S.; Sannen, I.; Dahaen, W. Tetrahedron 1993,
49, 4439.
(62) For an excellent review of this catalyst, see: Corey, E. J.; Helal, C.
J. Angew. Chem., Int. Ed. 1998, 37, 1986.

7060 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
Scheme 10a
compound. As a result of these studies, we have designated
dEpoB as a leading candidate for clinical application. Further
development of an epothilone as a potential clinical candidate
continues to remain the main focus of our research.
Experimental Section
General. All commercial materials were used without further
purification unless otherwise noted. The following solvents were
obtained from a dry solvent system (passed through a column of
alumina) and used without further drying: THF, diethyl ether (Et2O),
CH2Cl2, toluene, and benzene. All reaction were performed under an
atmosphere of prepurified dry Ar(g). NMR (1H, 13C) spectra were
recorded on Bruker AMX-400 MHz, Bruker Advance DRX-500 MHz,
reference to TMS (1H NMR, δ 0.00) or CDCl3 (13C NMR, δ 77.0)
peaks unless otherwise stated. LB ) 1.0 Hz was used before Fourier
transformation for all the 13C NMR. IR spectra were recorded with a
Perkin-Elmer 1600 series-FTIR spectrometer, and optical rotations were
measured with a Jasco DIP-370 digital polarimeter using 10-cm path
length cell. Low-resolution mass spectral analysis were performed with
a JEOL JMS-DX-303 HF mass spectrometer. Analytical thin-layer
chromatography was performed on E. Merck silica gel 60 F254 plates
(0.25 mm). Compounds were visualized by dipping the plates in a
cerium sulfate-ammonium molybdate solution or p-anisaldehyde
solution followed by heating. Flash column chromatography was
performed using the indicated solvent on E. Merck silica gel 60 (40-
63 mm) or Sigma H-Type silica gel (10-40 mm).
tert-Butyl (6R,7S,8S,12Z,15S,16E)-3,5-Dioxo-15-hydroxy-17-(2-
methylthiazol-4-yl)-4,4,6,8,12,16-hexamethyl-(2,2,2-trichloroethoxy-
carbonyl)pentadeca-12,16-dieno-ate (32). 9-BBN dimer (1.58 g, 6.50
mmol) was dissolved in 20 mL of dry THF. Then, tricarbonyl 30 (4.46
g, 8.64 mmol) in THF (15 mL) was added in small portions over a 45
min period to the solution of 9-BBN at 25 °C. After 2 h, TLC analysis
revealed the complete consumption of the starting olefin.
In a separate flask, containing the vinyl iodide 31 (4.0 g, 8.64 mmol)
and DMF (60 mL), were added successively and with vigorous
stirring: Cs2CO3 (5.6 g, 17.3 mmol), AsPh3 (0.53 g, 1.71 mmol), Pd-
(dppf)2Cl2 (1.40 g, 1.71 mmol), and H2O (4.6 mL, 0.25 mol). Then,
H2O (1 mL, 56 mmol) was added to the borane solution, prepared
above, to quench the excess 9-BBN. Then, the solution of alkyl borane
was added rapidly by syringe to the vigorously stirred solution
containing the vinyl iodide. After 2 h, the reaction TLC analysis
revealed that the reaction was complete. The reaction mixture was
poured into Et2O (300 mL), washed with H2O (3 × 200 mL) and brine
(1 × 50 mL), and dried over anhydrous MgSO4. This crude product
was purified by flash column chromatography on SiO2, eluting with
hexanes/ethyl acetate (18:1 to 13:1 to 10:1) to afford the TBS-protected
coupled product as an impure mixture which was taken onto the next
step without further purification.
The crude TBS-protected coupled product was dissolved in 0.5 M
HCl in MeOH (100 mL) at 25 °C. The reaction was monitored by
TLC for completion, and after 3.5 h (disappearance of starting TBS
ether), the mixture was poured into a solution of saturated aqueous
NaHCO3 and extracted with CHCl3 (4 × 60 mL). The combined organic
layers were washed once with brine (50 mL) and dried over anhydrous
MgSO4. The diol was purified by flash column chromatography on
SiO2, eluting with hexanes/ethyl acetate (4:1 to 3:1 to 2:1) to give the
pure product 32 as a clear oil (3.96 g, 5.4 mmol, 62% for two steps):
1H NMR (400 MHz, CDCl3) δ 6.96 (s, 1H), 6.56 (s, 1H), 5.16 (t, J )
6.9 Hz, 1H), 4.83 (d, J ) 11.9 Hz, 1H), 4.75 (dd, J ) 3.4, 8.0 Hz,
1H), 4.70 (d, J ) 11.9 Hz, 1H), 4.14 (t, J ) 6.4 Hz, 1H), 3.45 (q, J )
13.2 Hz, 2H), 3.32 (m, 1H), 2.72 (s, 3H), 2.32 (t, J ) 6.5 Hz, 2H),
2.04 (s, 3H), 2.01 (m, 2H), 1.74 (m, 2H), 1.69 (s, 3H), 1.45 (s, 9H),
1.38 (s, 6H), 1.09 (d, J ) 6.9 Hz, 3H), 0.93 (d, J ) 6.9 Hz, 3H); 13C
NMR (100 MHz, CDCl3) δ 209.51, 203.04, 166.15, 164.39, 154.14,
152.72, 141.71, 138.24, 120.70, 118.76, 115.28, 94.54, 81.85, 77.31,
76.57, 63.41, 54.16, 46.47, 41.48, 34.56, 33.95, 31.98, 31.53, 27.85,
24.85, 23.45, 21.47, 20.75, 19.04, 15.60, 14.33, 11.35; IR (neat) 3546,
3395, 1756, 1717, 1699, 1644, 1621, 1506, 1456, 1251 cm-1; LR-
FAB calcd for C34H50Cl3NO8S 737.2; found 738.5 [M + H]+.
tert-Butyl (3S,6R,7S,8S,12Z,15S,16E)-5-Oxo-3,15-dihydroxy-17-
(2-methylthiazol-4-yl)-4,4,6,8,12,16-hexamethyl-(2,2,2-trichloro-
a (a) Zn, CuCN, LiCl, THF, then BF3‚OEt2, 52, (67%); (b) Swern
oxidation (90%); (c) CBS reduction, 60% ee, (89%); (d) TBSOTf, 2,6-
lutidine, CH2Cl2, (90%); (e) 9-BBN, THF, 2 h, then 56, Pd(dppf)2Cl2,
Cs2CO3, Ph3As, H2O, DMF, rt, 6 h, (70%); (f) 0.5 M HCl/MeOH, THF,
rt, 6 h (85%).
“natural” stereochemistry at C15.63 The remainder of the
synthesis was performed according to the previous protocols
described in Scheme 9 to obtain analogue 58.
Conclusions
We have described a practical total synthesis of dEpoB
(EpoB) that has supported the synthesis of both this agent and
several epothilone analogues in sufficient quantity to perform
pertinent in ViVo studies on several human xenograft tumors in
mice, the results of which have been published elsewhere.14,20,23
Our fully chemical synthesis comprises three key components,
each critical to the completion of the synthesis. These include
a diastereoselective aldol reaction, B-alkyl Suzuki reaction, and
a modified asymmetric Noyori reaction. Surprisingly, the aldol
reaction developed in this total synthesis exhibits enhanced
diastereofacial selectivity for formation of the anti-Felkin-Anh
diastereomer. Continued examination of this novel aldol reaction
is ongoing in our laboratories. These studies are being conducted
in the context of a continued program intended to discover the
optimal epothilone derivative for full-scale clinical development.
With this accomplished, the total synthesis effort will be
conducted on multigram scales.
The chemical synthesis described herein has proved useful
for the discovery of the biological attributes of this promising
(63) The stereochemistry of the C-15 stereocenter resulting from the (R)-
2-methyl-CBS-oxazaborolidine reduction of ketone 55 was not proven
rigorously but is predicted on the basis of enantioselective reductions of
these systems: Smith, D. B.; Waltos, A. M.; Loughhead, D. G.; Weikert,
R. J.; Morgans, D. J.; Rohloft, J. C.; Link, J. O.; Zhu, R. J. Org. Chem.
1996, 61, 2236.

Synthesis of 12,13-Desoxyepothilone B
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7061
ethoxycarbonyl)pentadeca-12,16-dienoate (43). The diketone 32 (3.12
g, 4.22 mmol) was dissolved in 0.15 N HCl in MeOH (36 mL, 1.3
equiv) at 25 °C. The (R)-RuBINAP catalyst40 (0.045 M in THF, 8.0
mL, 0.36 mmol) was then added and the mixture transferred to a Parr
apparatus. The vessel was purged with H2 for 5 min and then pressurized
to 1200 psi. After 12-14 h at 25 °C, the reaction was returned to
atmospheric pressure and poured into a saturated solution of NaHCO3.
This mixture was extracted with CHCl3 (4 × 50 mL), and the combined
organic layers were dried over anhydrous MgSO4. The product was
purified by flash column chromatography on silica gel, eluting with
hexanes/ethyl acetate (4:1 to 2:1) to give 2.53 g (81%) of the hydroxy
2H), 1.16 (s, 3H), 1.13 (s, 3H), 1.09 (d, J ) 6.7 Hz, 3H), 0.99 (d, J )
6.7 Hz, 3H), 0.95 (t, J ) 7.9 Hz, 9H), 0.64 (dq, J ) 2.3, 7.9 Hz, 6H);
13C NMR (100 MHz, CDCl3) δ 215.11, 176.00 (165.10, 154.18, 152.35,
142.24, 138.55, 120.74, 118.21, 115.02, 94.76, 81.91, 76.86, 76.63,
73.95, 54.08, 41.28, 39.64, 34.73, 34.16, 32.02, 31.67, 24.71, 23.41,
22.49, 19.17, 18.62, 15.71, 14.86, 11.20, 6.93, 5.05); IR (neat) 3400-
2390, 1755.9, 1703.8, 1250.4, 735.4 cm-1; LR-FAB calcd for C36H58-
Cl3NO8SSi 797.3; found 800.6 [M + H]+.
12,13-Deoxy-7-(2,2,2-trichloroethoxycarbonyl)-3-(triethylsilyloxy)-
epothilone B (45). Triethylamine (0.61 g, 6.03 mmol) and 2,4,6-
trichlorobenzoyl chloride (1.2 g, 5.0 mmol) were added to a solution
of the hydroxy acid 44 (780 mg, 1.0 mmol) in 14 mL of THF. The
reaction mixture was stirred for 15 min at room temperature and then
diluted with 25 mL of dry toluene. The resultant solution was added
slowly dropwise, via syringe pump, over 3 h to a previously prepared,
stirred solution of DMAP (1.3 g, 10.6 mmol) in 700 mL of dry toluene.
After the addition of the substrate was complete, the reaction was stirred
for an additional 0.5 h and then concentrated in vacuo. Flash column
chromatography of the crude product with 10% EtOAc/hexanes afforded
the desired macrolactone 45 (464 mg, 0.60 mmol) in 60% yield: [R]D
) -2.6° (c 0.80, CHCl3); 1H NMR (400 MHz, CDCl3) δ 6.96 (s, 1H),
6.53 (s, 1H), 5.20 (m, 2H), 5.04 (d, J ) 10.2 Hz, 1H), 4.84 (d, J )
12.0 Hz, 1H), 4.78 (d, J ) 12.0 Hz, 1H), 4.07 (m, 1H), 3.32 (m, 1H),
2.86-2.63 (m, 3H), 2.70 (s, 3H), 2.48 (m, 1H), 2.11 (s, 3H), 2.04 (dd,
J ) 6.17, 14.7 Hz, 1H), 1.73 (m, 4H), 1.66 (s, 3H), 1.25 (m, 2H), 1.19
(s, 3H), 1.15 (s, 3H), 1.12 (d, J ) 6.68 Hz, 3H), 1.01 (d, J ) 6.83 Hz,
3H), 0.89 (t, J ) 8.00 Hz, 9H), 0.58 (q, J ) 7.83 Hz, 6H); 13C NMR
(100 MHz, CDCl3) δ 212.75, 170.66, 164.62, 154.60, 152.52, 140.29,
138.44, 119.81, 119.38, 116.28, 94.84, 86.44, 80.14, 76.59, 76.10, 53.55,
45.89, 39.23, 35.47, 32.39, 31.69, 31.57, 31.16, 29.68, 27.41, 25.00,
23.44, 22.94, 19.23, 18.66, 16.28, 14.83, 6.89, 5.22; IR (neat) 1760.5,
1742.6, 1698.0, 1378.8, 1246.2, 1106.0, 729.8 cm-1; LR-FAB calcd
for C36H57Cl3NO7SSi 780.3; found 780.5.
1
ester 43 as a green foam: [R]D ) -32.4° (c 0.36, CHCl3); H NMR
(400 MHz, CDCl3) δ 6.94 (s, 1H), 6.55 (s, 1H), 5.15 (t, J ) 6.9 Hz,
1H), 4.85 (t, J ) 5.3 Hz, 1H), 4.81 (d, J ) 12.0 Hz, 1H), 4.71 (d, J )
12.0 Hz, 1H), 4.12 (m, 2H), 3.43 (m, 2H), 2.70 (s, 3H), 2.37 (dd, J )
2.2, 6.2 Hz, 1H), 2.30 (t, J ) 6.7 Hz, 2H), 2.24 (dd, J ) 10.6, 16.2
Hz, 1H), 2.03 (s, 3H), 1.99 (m, 2H), 1.68 (s, 3H), 1.44 (s, 9H), 1.18 (s,
3H), 1.16 (s, 3H), 1.09 (d, J ) 6.8 Hz, 3H), 0.94 (d, J ) 6.8 Hz, 3H);
13C NMR (100 MHz, CDCl3): δ 215.95, 172.39, 164.39, 154.21,
152.74, 141.70, 138.33, 120.59, 118.77, 115.27, 94.64, 82.98, 81.26,
76.51, 72.78, 51.82, 41.40, 37.36, 34.66, 33.96, 32.08, 31.10, 30.20,
27.96, 25.06, 23.45, 21.73, 21.07, 19.17, 19.01, 16.12, 15.16, 14.33,
12.17; IR (neat) 3434.0, 1757.5, 1704.5, 1249.9, 1152.8 cm-1; LR-
FAB calcd for C38H52Cl3NO8S 739.2; found 740.5 [M + H]+.
(3S,6R,7S,8S,12Z,15S,16E)-5-Oxo-3,15-bis(triethylsilyloxy)-17-(2-
methylthiazol-4-yl)-4,4,6,8,12,16-hexamethyl-(2,2,2-trichloroethoxy-
carbonyl)pentadeca-12,16-dienoic acid. 2,6-Lutidine (3.8 g, 35.6
mmol) and TESOTf (4.7 g, 17.8 mmol) were added successively to a
cooled solution of the diol 43 (4.4 g, 5.90 mmol) in CH2Cl2 (50 mL)
at -78 °C. The reaction mixture was stirred at -78 °C for 5 min and
then warmed to room temperature and stirred for 1 h. Then 2,6-lutidine
(8.9 g, 83.2 mmol) and TESOTf (10.9 g, 41.6 mmol) were added
successively to a -78 °C cooled solution. The reaction was stirred at
room temperature for 8 h and then quenched with saturated aqueous
NH4Cl and subjected to an aqueous workup. The crude product was
concentrated in vacuo and the 2,6-lutidine removed on high vacuum
pump and then subjected directly to the next set of reaction condi-
tions: 1H NMR (400 MHz, CDCl3) δ 6.96 (s, 1H), 6.66 (s, 1H), 5.04
(t, J ) 6.93 Hz, 1H), 4.90 (d, J ) 12.0 Hz, 1H), 4.77 (dd, J ) 7.99,
3.21 Hz, 1H), 4.66 (d, J ) 12.0 Hz, 1H), 4.46 (m, 1H), 4.10 (dq, J )
12.3, 7.11 Hz, 2H), 3.42 (m, 1H), 2.70 (s, 3H), 2.60 (dd, J ) 16.7,
2.34 Hz, 1H), 2.34 (dd, J ) 16.7, 7.94 Hz, 1H), 2.27 (dd, J ) 14.0,
6.97 Hz, 1H), 2.18 (m, 1H), 2.09 (m, 1H), 2.04 (s, 1H), 1.95 (s, 3H),
1.82 (m, 2H), 1.61 (s, 3H), 1.44 (m, 2H), 1.27-1.22 (m, 4H), 1.14 (d,
J ) 8.45 Hz, 3H), 1.11 (d, J ) 6.81 Hz, 2H), 1.04 (d, J ) 6.88 Hz,
2H), 1.15-1.01 (m, 2H), 0.94 (t, J ) 7.92 Hz, 18H), 0.65-0.57 (m,
12H); 13C NMR (100 MHz, CDCl3) δ 215.11, 175.34, 165.00, 154.14,
152.80, 142.60, 136.84, 121.31, 118.79, 114.60, 94.77, 81.60, 79.06,
76.64, 73.87, 54.19, 41.18, 39.56, 35.09, 34.52, 32.29, 31.95, 24.76,
23.62, 22.55, 18.95, 18.64, 15.87, 13.69, 11.33, 6.94, 6.83, 5.07, 4.76;
12,13-Deoxy-3-(triethylsilyloxy)epothilone B (46). Samarium metal
(4.2 g, 28.2 mmol) and iodine (5.2 g, 20.5 mmol) in 250 mL of dry
THF were stirred together vigorously at reflux for 2.5 h. During this
period of time, the reaction mixture progressed from a dark orange to
an olive green to deep blue color. The resultant deep blue solution of
SmI2 was used directly in the following reaction. A catalytic amount
of NiI2 (0.61 g) was added in one portion to the vigorously stirred
solution of SmI2. The reaction mixture was stirred 5 min at room
temperature and then cooled to -78 °C in a a dry ice/acetone bath.
Then, the macrolactone 45 (3.0 g, 3.8 mmol), in 40 mL of dry THF,
was added over 1 min to the rapidly stirred, cold solution of SmI2/
NiI2.44 The resultant deep blue solution was maintained at -78 °C with
continued vigorous stirring for 2 h. TLC analysis at this time revealed
the complete consumption of the starting material and formation of a
single, lower Rf product. The reaction mixture was quenched with
saturated aqueous NaHCO3 and subjected to an aqueous workup. Flash
column chromatography with 25% EtOAc/hexanes afforded the desired
C-7 alcohol, 46, (1.88 g, 3.12 mmol) in 82% yield: [R]D ) -64.5° (c
IR (neat) 3100-2390, 1756.8, 1708.8, 1459.3, 1250.6, 816.1 cm-1
.
1
0.75, CHCl3); H NMR (400 MHz, CDCl3) δ 6.95 (s, 1H), 6.54 (s,
(3S,6R,7S,8S,12Z,15S,16E)-5-Oxo-15-hydroxy-3-(triethylsilyloxy)-
17-(2-methylthiazol-4-yl)-4,4,6,8,12,16-hexamethyl-(2,2,2-trichloro-
ethoxycarbonyl)pentadeca-12,16-dienoic acid (44). The crude bis-
(triethylsilyl)ether (above) was dissolved in 30 mL of THF and then
cooled to 0 °C. Then, 20 mL of 0.12 M HCl/MeOH was added. The
reaction mixture was stirred at 0 °C for 3 min and maintained at 0 °C
for the duration. The reaction was monitored closely by TLC analysis.
Methanolic HCl (0.12 M) was added in small portions (5 mL), and
roughly 1.3 equiv of 0.12 M HCl was required for the hydrolysis of
the C-15 TBS ether (approximately 65 mL, total). The reaction was
complete in approximately 30 min. The reaction was quenched by
pouring into a solution of saturated aqueous NaHCO3 and subjected to
an aqueous workup. Flash column chromatography with 40% EtOAc/
hexanes afforded the desired carboxylic acid 44 (3.61 g, 4.51 mmol)
1H), 5.15 (m, 1H), 5.05 (d, J ) 10.15 Hz, 1H), 4.08 (dd, J ) 10.1,
2.66 Hz, 1H), 3.87 (m, 1H), 3.01 (s, 1H), 3.06 (m, 1H), 2.83-2.65 (m,
3H), 2.70 (s, 3H), 2.44 (m, 1H), 2.10 (s, 3H), 2.07 (m, 1H), 1.83 (m,
1H), 1.77 (m, 1H), 1.71 (m, 1H), 1.64 (s, 3H), 1.60 (s, 1H), 1.37 (m,
1H), 1.31 (m, 1H), 1.20 (m, 1H), 1.15 (s, 3H), 1.14 (m, 5H), 1.02 (d,
J ) 7.02 Hz, 3H), 0.89 (t, J ) 7.97 Hz, 9H), 0.64-0.52 (m, 6H); 13
C
NMR (100 MHz, CDCl3) δ 218.34, 170.73, 164.59, 152.46, 139.07,
138.49, 120.48, 119.54, 116.00, 79.31, 75.81, 73.48, 53.62, 42.98, 39.48,
39.01, 32.85, 32.41, 31.20, 26.12, 24.26, 22.01, 22.46, 19.18, 16.44,
15.30, 13.99, 6.98 (3), 5.27 (3); IR (neat) 3524.0, 1740.3, 1693.4,
1457.2, 1378.4, 733.2 cm-1; LR-FAB calcd for C33H56NO5SSi 606.4;
found 606.5 [M + H]+.
12,13-Deoxyepothilone B (47). The C-3 TES-protected alcohol 46
(2.04 g, 3.37 mmol) was dissolved in 60 mL of THF in a plastic reaction
vessel and cooled to 0 °C in an ice bath. The resultant solution was
treated with 14 mL of HF‚pyridine. The reaction mixture was stirred
for 5 h at 0 °C and then quenched by being poured into a saturated
aqueous solution of NaHCO3. An aqueous workup followed by flash
column chromatography with 10% EtOAc/hexanes afforded the desired
1
in 76% yield (two steps): [R]D ) -22.4° (c 0.58, CHCl3); H NMR
(400 MHz, CDCl3) δ 6.96 (s, 1H), 6.69 (1, s), 5.11 (t, J ) 6.9 Hz,
1H), 4.91 (d, J ) 12.0 Hz, 1H), 4.71 (dd, J ) 3.1, 8.2 Hz, 1H), 4.64
(d, J ) 12.0 Hz, 1H), 4.42 (d, J ) 5.9 Hz, 1H), 4.10 (m, 1H), 3.43 (m,
1H), 2.71 (s, 3H), 2.57 (dd, J ) 2.1, 10.5 Hz, 1H), 2.25 (m, 3H), 2.11
(m, 1H), 1.98 (s, 3H), 1.95 (m, 2H), 1.72 (m 1), 1.67 (s, 3H), 1.45 (m,

7062 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Harris et al.
desoxyepothilone B 47 (1.5 g, 3.05 mmol) in 91% yield. The resultant
product exhibited a H NMR spectrum identical to desoxyepothilone
B prepared previously in this laboratory.
and NMR analysis. Dr. Xiu-Guo Zhang and Dr. Ting-Chao
Chou are gratefully acknowledged for biological studies.
1
Supporting Information Available: Experimental proce-
dures for the preparation of tert-butyl 4-methyl-3-oxopentanoate,
11, 15, 16, 18-23, 29-31, 38, 41, 49a-p, 50/51a-p, and 54-
58 as well as characterization data are included (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
Acknowledgment. This research was supported by the
National Institutes of Health S.J.D. (Grant Nos. CA-28824 and
CA-08748). Postdoctoral Fellowship support is gratefully
acknowledged by: C.R.H. (American Cancer Society, PF-98-
173-001), S.D.K. (U.S. Army Breast Cancer Research Fellow-
ship, DAMD 17-98-1-1854), and P.W.G. (American Cancer
Society, PF-98-026). The authors thank Dr. George Sukenick
(NMR Core Facility, Sloan-Kettering Institute) for mass spectral
JA991189L