The migrastatin family: Discovery of potent cell migration inhibitors by chemical synthesis

Article abstract of DOI:10.1021/ja048779q

The first asymmetric total synthesis of (+)-migrastatin (1), a macrolide natural product with antimetastatic properties, has been accomplished. Our concise and flexible approach utilized a Lewis acid-catalyzed diene aldehyde condensation (LACDAC) to install the three contiguous stereocenters and the trisubstituted (Z)-alkene of migrastatin (2 + 3 → 21). Construction of the two remaining stereocenters and incorporation of the glutarimide-containing side chain was achieved by an anti-selective aldol addition of propionyl oxazolidinone 28 to angelic aldehyde 27, followed by a Horner-Wadsworth-Emmons (HWE) coupling of 32 with glutarimide aldehyde 5. Finally, the assembly of the macrocycle was realized by a highly (E)-selective ring-closing metathesis (35 → 37). Utilizing the power of diverted total synthesis (DTS), a series of otherwise inaccessible analogues was prepared and evaluated for their potential as tumor cell migration inhibitors in several in vitro assays. These studies revealed a dramatic increase in activity when the natural motif was considerably simplified, presenting macrolactones 45 and 48, as well as macrolactam 55, macroketone 60, and CF₃-alcohol 71 as promising anti-metastatic agents.

Full text of DOI:10.1021/ja048779q

Published on Web 07/30/2004
The Migrastatin Family: Discovery of Potent Cell Migration
Inhibitors by Chemical Synthesis
Christoph Gaul,† Jo´n T. Njardarson,† Dandan Shan,‡ David C. Dorn,§ Kai-Da Wu,§
William P. Tong,| Xin-Yun Huang,‡ Malcolm A. S. Moore,§ and
Samuel J. Danishefsky*,†,
Contribution from the Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for Cancer
Research, 1275 York AVenue, New York, New York 10021, Department of Physiology,
Weill Medical College of Cornell UniVersity, 1300 York AVenue, New York, New York 10021,
Laboratory of DeVelopmental Hematopoiesis, Cell Biology Program, Sloan-Kettering Institute
for Cancer Research, New York, New York 10021, Analytical Core Facility, Sloan-Kettering
Institute for Cancer Research, 1275 York AVenue, New York, New York 10021, and
Department of Chemistry, Columbia UniVersity, HaVemeyer Hall, 3000 Broadway,
New York, New York 10027
Received March 3, 2004; E-mail: s-danishefsky@ski.mskcc.org
Abstract: The first asymmetric total synthesis of (+)-migrastatin (1), a macrolide natural product with anti-
metastatic properties, has been accomplished. Our concise and flexible approach utilized a Lewis acid-
catalyzed diene aldehyde condensation (LACDAC) to install the three contiguous stereocenters and the
trisubstituted (Z)-alkene of migrastatin (2 + 3 f 21). Construction of the two remaining stereocenters and
incorporation of the glutarimide-containing side chain was achieved by an anti-selective aldol addition of
propionyl oxazolidinone 28 to angelic aldehyde 27, followed by a Horner-Wadsworth-Emmons (HWE)
coupling of 32 with glutarimide aldehyde 5. Finally, the assembly of the macrocycle was realized by a
highly (E)-selective ring-closing metathesis (35 f 37). Utilizing the power of diverted total synthesis (DTS),
a series of otherwise inaccessible analogues was prepared and evaluated for their potential as tumor cell
migration inhibitors in several in vitro assays. These studies revealed a dramatic increase in activity when
the natural motif was considerably simplified, presenting macrolactones 45 and 48, as well as macrolactam
55, macroketone 60, and CF3-alcohol 71 as promising anti-metastatic agents.
Introduction
depend on angiogenesis (formation of new blood vessels) to
obtain the necessary oxygen and nutrient supplies for growth
Background. Traditionally, cancer chemotherapy relies on
therapeutic agents with cytotoxic properties that inhibit tumor
cell proliferation and cause cell death. Recently, the idea of
targeting cell migration as an alternative strategy for the
development of anti-cancer therapies has generated considerable
interest.1 Intense research efforts are currently directed to the
exploration of cell shape change and movement and their
underlying mechanisms.2 Cell migration is involved in a number
of physiological processes, including ovulation, embryonic
development, tissue regeneration (wound healing), and inflam-
mation. On the other hand, cell migration is also observed in
pathological conditions such as tumor angiogenesis, cancer cell
invasion, and metastasis.3 It is believed that primary solid tumors
beyond a certain size (ca. 1-2 mm). The transition from a pre-
angiogenic condition to tumor angiogenesis,4 often referred to
as the angiogenic switch, is followed by tumor growth, cancer
cell invasion, and metastasis.5 In principle, it could be possible
to halt (or retard) this procession at different stages with the
help of cell migration inhibitors. Because cell migration under
ordinary physiological conditions in adults is rather infrequent,
its repression might be accompanied by manageable toxicity.
A significant part of our general research program focuses
on the development of novel, natural product-inspired anti-
cancer agents. These efforts have led to the total chemical
synthesis of a number of prominent antitumor natural products,
(4) For research efforts toward anti-angiogenic agents, consult: (a) Brower,
V. Nat. Biotechnol. 1999, 17, 963-968. (b) Klohs, W. D.; Hamby, J. M.
Curr. Opin. Biotechnol. 1999, 10, 544-549. (c) Deplanque, G.; Harris, A.
L. Eur. J. Cancer 2000, 36, 1713-1724. (d) Scappaticci, F. A. J. Clin.
Oncol. 2002, 20, 3906-3927. (e) Cristofanilli, M.; Charnsangavej, C.;
Hortobagyi, G. N. Nat. ReV. Drug DiscoVery 2002, 1, 415-426. (f) Kerbel,
R.; Folkman, J. Nat. ReV. Cancer 2003, 2, 727-739. For early syntheses
of natural product angiogenesis inhibitors, see: (g) Corey, E. J.; Snider,
B. B. J. Am. Chem. Soc. 1972, 94, 2549-2550. (h) Corey, E. J.; Dittami,
J. P. J. Am. Chem. Soc. 1985, 107, 256-257. (i) Corey, E. J.; Guzman-
Perez, A.; Noe, M. C. J. Am. Chem. Soc. 1994, 116, 12109-12110.
(5) Woodhouse, E. C.; Chuaqui, R. F.; Liotta, L. A. Cancer 1997, 80 (S8),
1529-1537.
† Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for
Cancer Research.
‡ Weill Medical College of Cornell University.
§ Laboratory of Developmental Hematopoiesis, Sloan-Kettering Institute
for Cancer Research.
| Analytical Core Facility, Sloan-Kettering Institute for Cancer Research.
Columbia University.
(1) Fenteany, G.; Zhu, S. Curr. Top. Med. Chem. 2003, 3, 593-616.
(2) Lauffenburger, D. A.; Horwitz, A. F. Cell 1996, 84, 359-369.
(3) Carmeliet, P. Nat. Med. 2003, 9, 653-660.
9
11326
J. AM. CHEM. SOC. 2004, 126, 11326-11337
10.1021/ja048779q CCC: $27.50 © 2004 American Chemical Society

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Scheme 1. Structure of Migrastatin and Related Natural Products
such as the epothilones,6 taxol,7 and, most recently, radicicol8
and TMC-95A/B.9 The recent entry of 12,13-desoxyepothilone
B (dEpoB), first prepared by total chemical synthesis, into phase
II clinical trials,10 has been followed by the discovery of a new
generation of highly potent epothilone analogues.11 For the most
part, our endeavors have converged on cytotoxic agents. The
possibility of exploiting natural products as leads for the
development of anti-angiogenic and anti-metastatic agents was
prompted by the recent isolation and synthesis of compounds
such as epoxyquinol A and B,12 trachyspic acid,13 azaspirene,14
evodiamine,15 motuporamines,16 borrelidin,17 and terpestacin.18
In particular, a series of independent reports by Imoto19 and
Kosan Bioscience researchers20 on the discovery of the natural
product migrastatin (1) enhanced our interest in this area
(Scheme 1). It was reported that 1, isolated from a cultured broth
of Streptomyces, has the potential of metastasis suppression
through its ability to inhibit tumor cell migration. Although the
reported activity of migrastatin in a wound-healing assay was
rather modest (IC50 value of 29 µM), we considered it as an
attractive lead compound in the search for other, more potent
agents. The structure of migrastatin (1), determined by X-ray
crystal structure analysis, features a 14-membered macrolactone
(6) For comprehensive reviews, see: (a) Harris, C. R.; Danishefsky, S. J. J.
Org. Chem. 1999, 64, 8434-8456. (b) Stachel, S. J.; Biswas, K.;
Danishefsky, S. J. Curr. Pharm. Des. 2001, 7, 1277-1290.
(14) For the isolation of azaspirene, see: (a) Asami, Y.; Kakeya, H.; Onose,
R.; Yoshida, A.; Matsuzaki, H.; Osada, H. Org. Lett. 2002, 4, 2845-2848.
For the synthesis of azaspirene, see: (b) Hayashi, Y.; Shoji, M.; Yamaguchi,
J.; Sato, K.; Yamaguchi, S.; Mukaiyama, T.; Sakai, K.; Asami, Y.; Kakeya,
H.; Osada, H. J. Am. Chem. Soc. 2002, 124, 12078-12079.
(15) A screening approach revealed evodiamine as a potent anti-invasive and
anti-metastatic agent: (a) Ogasawara, M.; Matsubara, T.; Suzuki, H. Biol.
Pharm. Bull. 2001, 24, 720-723. (b) Ogasawara, M.; Matsubara, T.; Suzuki,
H. Biol. Pharm. Bull. 2001, 24, 917-920. (c) Ogasawara, M.; Matsunaga,
T.; Takahashi, S.; Saiki, I.; Suzuki, H. Biol. Pharm. Bull. 2002, 25, 1491-
1493.
(16) For the isolation, synthesis, and discussion of the anti-angiogenic properties
of the motuporamines, see: (a) Williams, D. E.; Lassota, P.; Andersen, R.
J. J. Org. Chem. 1998, 63, 4838-4841. (b) Roskelley, C. D.; Williams, D.
E.; McHardy, L. M.; Leong, K. G.; Troussard, A.; Karsan, A.; Andersen,
R. J.; Dedhar, S.; Roberge, M. Cancer Res. 2001, 61, 6788-6794. (c)
Williams, D. E.; Craig, K. S.; Patrick, B.; McHardy, L. M.; van Soest, R.;
Roberge, M.; Andersen, R. J. J. Org. Chem. 2002, 67, 245-258.
(17) For the discovery of the anti-angiogenic properties of borrelidin, see: (a)
Wakabayashi, T.; Kageyama, R.; Naruse, N.; Tsukahara, N.; Funahashi,
Y.; Kitoh, K.; Watanabe, Y. J. Antibiot. 1997, 50, 671-676. For the
synthesis of borrelidin, see: (b) Duffey, M. O.; LeTiran, A.; Morken, J. P.
J. Am. Chem. Soc. 2003, 125, 1458-1459.
(18) For the discovery of the anti-angiogenic properties of terpestacin, see: (a)
Jung, H. J.; Lee, H. B.; Kim, C. J.; Rho, J. R.; Shin, J.; Kwon, H. J. J.
Antibiot. 2003, 56, 492-496. For the synthesis of terpestacin, see: (b)
Tatsuta, K.; Masuda, N. J. Antibiot. 1998, 51, 602-606. (c) Myers, A. G.;
Siu, M.; Ren, F. J. Am. Chem. Soc. 2002, 124, 4230-4232. (d) Chan, J.;
Jamison, T. F. J. Am. Chem. Soc. 2003, 125, 11514-11515.
(7) Danishefsky, S. J.; Masters, J. J.; Young, W. B.; Link, J. T.; Snyder, L.
B.; Magee, T. V.; Jung, D. K.; Isaacs, R. C. A.; Bornmann, W. G.; Alaimo,
C. A.; Coburn, C. A.; DiGrandi, M. J. J. Am. Chem. Soc. 1996, 118, 2843-
2859.
(8) For the synthesis and biological evaluation of radicicol and cyclopropyl-
radicicol, see: (a) Garbaccio, R. M.; Stachel, S. J.; Baeschlin, D. K.;
Danishefsky, S. J. J. Am. Chem. Soc. 2001, 123, 10903-10908. (b)
Yamamoto, K.; Garbaccio, R. M.; Stachel, S. J.; Solit, D. B.; Chiosis, G.;
Rosen, N.; Danishefsky, S. J. Angew. Chem., Int. Ed. 2003, 42, 1280-
1284. (c) Yang, Z. Q.; Danishefsky, S. J. J. Am. Chem. Soc. 2003, 125,
9602-9603.
(9) For the synthesis and evaluation of TMC-95A/B, see: (a) Lin, S.;
Danishefsky, S. J. Angew. Chem., Int. Ed. 2001, 40, 512-515. (b) Yang,
Z. Q.; Kwok, B. H.; Lin, S.; Koldobskiy, M. A.; Crews, C. M.; Danishefsky,
S. J. Chembiochem 2003, 6, 508-513.
(10) For more information about clinical trials of dEpoB, visit: www.kosan.com.
(11) For the synthesis and biological evaluation of recent epothilone analogues,
see: (a) Rivkin, A.; Yoshimura, F.; Gabarda, A. E.; Chou, T. C.; Dong,
H.; Tong, W. P.; Danishefsky, S. J. J. Am. Chem. Soc. 2003, 125, 2899-
2901. (b) Yoshimura, F.; Rivkin, A.; Gabarda, A. E.; Chou, T. C.; Dong,
H.; Sukenick, G.; Morel, F. F.; Taylor, R. E.; Danishefsky, S. J. Angew.
Chem., Int. Ed. 2003, 42, 2518-2521.
(12) For the isolation of epoxyquinol A and B, see: (a) Kakeya, H.; Onose, R.;
Koshino, H.; Yoshida, A.; Kobayashi, K.; Kageyama, S. I.; Osada, H. J.
Am. Chem. Soc. 2002, 124, 3496-3497. (b) Kakeya, H.; Onose, R.;
Yoshida, A.; Koshino, H.; Osada, H. J. Antibiot. 2002, 55, 829-831. For
the synthesis of epoxyquinol A and B, see: (c) Shoji, M.; Yamaguchi, J.;
Kakeya, H.; Osada, H.; Hayashi, Y. Angew. Chem., Int. Ed. 2002, 41,
3192-3194. (d) Chaomin, L.; Bardhan, S.; Pace, E. A.; Liang, M. C.;
Gilmore, T. D.; Porco, J. A., Jr. Org. Lett. 2002, 4, 3267-3270. (e) Mehta,
G.; Islam, K. Tetrahedron Lett. 2003, 44, 3569-3572.
(19) (a) Nakae, K.; Yoshimoto, Y.; Sawa, T.; Homma, Y.; Hamada, M.;
Takeuchi, T.; Imoto, M. J. Antibiot. 2000, 53, 1130-1136. (b) Nakae, K.;
Yoshimoto, Y.; Ueda, M.; Sawa, T.; Takahashi, Y.; Naganawa, H.;
Takeuchi, T.; Imoto, M. J. Antibiot. 2000, 53, 1228-1230. (c) Takemoto,
Y.; Nakae, K.; Kawatani, M.; Takahashi, Y.; Naganawa, H.; Imoto, M. J.
Antibiot. 2001, 54, 1104-1107. (d) Nakamura, H.; Takahashi, Y.; Naga-
nawa, H.; Nakae, K.; Imoto, M.; Shiro, M.; Matsumura, K.; Watanabe,
H.; Kitahara, T. J. Antibiot. 2002, 55, 442-444.
(13) For the isolation of trachyspic acid, see: (a) Shiozawa, H.; Takahashi, M.;
Takatsu, T.; Kinoshita, T.; Tanzawa, K.; Hosoya, T.; Furuya, K.; Furihata,
K.; Seto, H. J. Antibiot. 1995, 48, 357-362. For the synthesis of trachyspic
acid, see: (b) Hirai, K.; Ooi, H.; Esumi, T.; Iwabuchi, Y.; Hatakeyama, S.
Org. Lett. 2003, 5, 857-859.
(20) Woo, E. J.; Starks, C. M.; Carney, J. R.; Arslanian, R.; Cadapan, L.; Zavala,
S.; Licari, P. J. Antibiot. 2002, 55, 141-146.
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11327

A R T I C L E S
Gaul et al.
with a characteristic glutarimide-containing side chain. Embed-
ded in the macrocycle are a trisubstituted (Z)-alkene and two
disubstituted (E)-alkenes, as well as three contiguous stereo-
centers. The side chain projecting from the cyclic core is
associated with stereogenic centers at C13 and C14.
NA30424 by researchers at Nippon Kayaku. Furthermore, four
related compounds, derived from oxidation of the thioether to
the sulfoxide, were detected as minor constituents in the cultured
broth and were titled as NK30424AS1-2 and NK30424BS1-2.
The NK compounds are formally derived from isomigrastatin
by conjugate addition of cysteine to the C2-C3 double bond
and hydroxylation at C17. Interestingly, these NK congeners
are reported to be very potent inhibitors of lipopolysaccharide-
induced tumor necrosis factor-R (TNF-R) promoter activity. To
date, migrastatin is the only member of the natural product
family described above in which the relative and absolute
configurations have been determined. Possibly, total chemical
synthesis might aid in deciphering the stereochemistry of other
members of this series.
Our research plan for evaluating the migrastatins as anti-
metastatic agents comprises the following key stages: A total
synthesis of 1 would afford our laboratory with material for an
independent evaluation of the biology of the natural product
and with opportunities through standard medicinal chemistry
for gaining access to a broad range of structural analogues.
Moreover, by diversion of the route of total synthesis, we could
be in a position to explore structural types that could not,
plausibly, be accessible by chemical modification of migrastatin
itself. In vitro screening of migrastatin derivatives in cell
migration assays could lead to an informative structure-activity
relationship (SAR) profile, conceivably assisting in the emer-
gence of compounds with improved biological profiles for
progression to in vivo models. Furthermore, efforts directed at
target identification are expected to yield some insight into the
biological mode of action of migrastatin and its congenitors.
Previously, we have communicated a total synthesis of
migrastatin27 and preliminary SAR studies.28 In this paper, we
provide a full account of the synthetic endeavor, which resulted
in an efficient and flexible total synthesis of 1. In addition, we
detail the preparation and biological evaluation of an extended,
diverse set of migrastatin analogues which led to the discovery
of highly potent cell migration inhibitors. We also describe our
undertakings to address issues of in vivo evaluation (large-scale
preparation of analogues), metabolic stability studies, and target
identification (labeling). In short, we show how the science of
total synthesis can be integrated into the progression of drug
discovery.
Upon reviewing the literature in search of glutarimide-
containing natural products, one encounters prominent examples
such as cycloheximide (CHX),21 streptimidone,22 and thalido-
mide, which has resurfaced recently as an anti-angiogenic agent
despite its controversial history!23 Moreover, a number of
structural homologues of migrastatin, lactimidomycin,24 dorri-
gocin A and B,20,25 isomigrastatin,20 and NK30424A/B,26 have
been discovered (Scheme 1). In 1992, lactimidomycin was
isolated from Streptomyces amphibiosporus and characterized
by researchers at Bristol-Myers Squibb. This unique triene-
containing 12-membered macrolactone antibiotic is highly
cytotoxic in vitro against a number of tumor cell lines and
displays in vivo antitumor activity in mice. In addition,
lactimidomycin exhibits potent anti-fungal properties and acts
as an inhibitor of DNA and protein synthesis. Two years later,
the isolation of dorrigocin A and its allylic isomer dorrigocin
B from Streptomyces platensis was described by researchers at
Abbott Laboratories. The dorrigocins are linear polyketide
carboxylic acids with a functional group arrangement closely
related to migrastatin and isomigrastatin, respectively. They were
found to reverse the morphology of ras-transformed NIH/3T3
cells from a transformed phenotype to a normal one. Dorrigocin
A was also reported to be the first natural product inhibitor of
the carboxyl methyltransferase involved in Ras processing. In
2002, the dorrigocins were again isolated from Streptomyces
platensis by researchers at Kosan Biosciences along with
migrastatin and a new member of the family, isomigrastatin.
Structurally, isomigrastatin can be described as being derived
from migrastatin by an allylic transposition (C13 f C11) and
a concomitant double bond isomerization. Thus, isomigrastatin
is a 12-membered macrolactone with an exocyclic trisubstituted
(E)-alkene. The Kosan researchers have shown that the hy-
drolysis of isomigrastatin leads to dorrigocin B, whereas the
hydrolysis of migrastatin produces a geometric isomer of
dorrigocin A. The biological profile of isomigrastatin has not
been reported to date. The latest members of the glutarimide-
containing macrolide family are the natural products NK30424A
and its stereoisomer NK30424B, isolated from Streptomyces sp.
Synthetic Plan. It was particularly important to devise a
concise, flexible, and readily scaleable synthesis, since it would
remain for synthesis to fuel an aggressive SAR elucidation
program and to provide significant quantities of materials for
in vivo studies. This is in keeping with the notion that total
synthesis was viewed as a first milestone of the project, rather
than as an end-point.
Any synthetic plan directed at migrastatin, and seeking to
meet the standards discussed above, must address several
obvious structural issues. Clearly, accommodation must be
provided for the (E)-configuration of the C2-C3 and C6-C7
double bonds, as well as for the (Z)-configuration of the C11-
C12 double bond of the trienic lactone. Moreover, protocols
are required for maintaining stereocontrol over the dispositions
of the stereogenic centers at C8, C9, C10, and C13. In addition,
(21) Cycloheximide (CHX) is a glutarimide antibiotic that inhibits protein
synthesis. CHX is widely used for studies of cell death and is commercially
available as Ready Made solution by Sigma. For a more recent leading
article, see: Mattson, M. P.; Furukawa, K. Apoptosis 1997, 2, 257-264.
(22) For a recent synthesis of streptimidone, a glutarimide antibiotic, see: Kondo,
H.; Oritani, T.; Kiyota, H. Eur. J. Org. Chem. 2000, 3459-3462.
(23) For the original discovery of the anti-angiogenic properties of thalidomide,
see: (a) D’Amato, R. J.; Loughnan, M. S.; Flynn, E.; Folkman, J. Proc.
Natl. Acad. Sci. U.S.A. 1994, 91, 4082-4085. For recent discussions of
the use of thalidomide in anti-cancer therapy, see: (b) Thomas, D. A.;
Kantarjian, H. M. Curr. Opin. Oncol. 2000, 12, 564-573. (c) Raje, N.;
Anderson, K. C. Curr. Opin. Oncol. 2002, 14, 635-646. (d) Dredge, K.;
Dalgleish, A. G.; Marriott, J. B. Anti-Cancer Drugs 2003, 14, 331-335.
(e) Capitosti, S. M.; Hansen, T. P.; Brown, M. L. Bioorg. Med. Chem.
2004, 12, 327-336.
(24) Sugawara, K.; Nishiyama, Y.; Toda, S.; Komiyama, N.; Hatori, M.;
Moriyama, T.; Sawada, Y.; Kamei, H.; Konishi, M.; Oki, T. J. Antibiot.
1992, 45, 1433-1441.
(25) (a) Karwowski, J. P.; Jackson, M.; Sunga, G.; Sheldon, P.; Poddig, J. B.;
Kohl, W. L.; Adam, S. J. Antibiot. 1994, 47, 862-869. (b) Hochlowski, J.
E.; Whittern, D. N.; Hill, P.; McAlpine, J. B. J. Antibiot. 1994, 47, 870-
874. (c) Kadam, S.; McAlpine, J. B. J. Antibiot. 1994, 47, 875-880.
(26) (a) Takayasu, Y.; Tsuchiya, K.; Aoyama, T.; Sukenaga, Y. J. Antibiot. 2001,
54, 1111-1115. (b) Takayasu, Y.; Tsuchiya, K.; Sukenaga, Y. J. Antibiot.
2002, 55, 337-340.
(27) Gaul, C.; Njardarson, J. T.; Danishefsky, S. J. J. Am. Chem. Soc. 2003,
125, 6042-6043.
(28) Njardarson, J. T.; Gaul, C.; Shan, D.; Huang, X. Y.; Danishefsky, S. J. J.
Am. Chem. Soc. 2004, 126, 1038-1040.
9
11328 J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Scheme 2. Strategy for the Assembly of Migrastatin
Scheme 3. A Model Study: Synthesis of the Migrastatin Core 15a
it is necessary to accomplish an installation of the required side
chain projecting from C13 and the inclusion of the stereocenter
at C14, which is not part of the ring structure. Needless to say,
suitable provision is needed for the incorporation of the C15
keto group as well as the δ-substituted glutarimide at C18.
A comprehensive plan that embraces these structural issues
is suggested in Scheme 2. Our retrosynthetic analysis led us
back to the five components 2, 3, 4, 5, and 6 shown therein.
The (E)-geometric character of the C2-C3 double bond could
be secured via recourse to the known compound 6. A key feature
of this synthesis would be the use of aldehyde 2, bearing the
methoxy-substituted stereogenic center, ultimately to be em-
placed at C8. Another important building block would be diene
3. This type of synergistically activated, dibranched, bisoxy-
genated butadiene was part of our all-carbon Diels-Alder
research in the mid 1970s.29 Indeed, in the 1980s, this type of
diene was used in the context of our LACDAC chemistry to
create dihydropyrans.30 The aldehyde in this case would be the
previously discussed 2. Appropriate disconnection of the pyran
would expose the four-carbon segment comprising C10-C13.
The two methyl-branching elements of 3 would appear at C10
and C12 in migrastatin following appropriate manipulations. At
the outset, the precise nature of the R function in keto building
block 4 awaited specification. The decisive criterion for various
candidate structures that might be interrogated would be their
amenability to linkage to the emerging C13 in the context of
macrolactone formation, while enabling smooth incorporation
of the δ-branched glutarimide. We note that the sum of
fragments 2 and 6 contains two carbons in excess of those
required for formation of the 14-membered macrolactone. Such
a disconnection invited the prospect of establishing this lactone
through a ring-closing metathesis (RCM) reaction with extrusion
of the two seemingly extraneous carbon centers. A more detailed
analysis of the synthetic issues appears in the context of the
next section, in which we describe the implementation of the
broad plan.
a Reagents and conditions: (a) (i) TBDPSCl, imidazole, DMF, room
temperature, (ii) MeI, NaH, THF, room temperature, (iii) H2, Pd(OH)2,
EtOAc, room temperature, (iv) (COCl)2, Et3N, DMSO, CH2Cl2, -78 °C to
room temperature, 66%; (b) (i) TiCl4, CH2Cl2, -78 °C, (ii) TFA, CH2Cl2,
room temperature, 79%; (c) (i) NaBH4, CeCl3‚7H2O, EtOH, 0 °C, (ii) CSA,
H2O, THF, reflux; (d) LiBH4, H2O, THF, room temperature, 55% from 9;
(e) (i) DMAP, CH2Cl2, room temperature, (ii) MOMCl, Bu4NI, i-Pr2NEt,
CH2Cl2, room temperature, 57%; (f) (i) HF‚pyridine, THF, room temper-
ature, (ii) Dess-Martin periodinane, CH2Cl2, room temperature, (iii) Tebbe
reagent, pyridine, THF, -78 to -10 °C, 54%; (g) Grubbs-II catalyst 16
(20 mol %), toluene (0.5 mM), reflux, 50%.
Because we were concerned about the stability, stereochemical
integrity, and potential volatility of the previously unknown
R-methoxy-R-vinyl aldehyde 2 (Scheme 2) contemplated for
the LACDAC reaction, we began, in this testing phase, with
the structurally less challenging heterodienophilic siloxy-
aldehyde 8 (Scheme 3). This compound was prepared from
commercially available (S)-3-benzyloxy-1,2-propanediol 732 in
four steps. The sterically demanding TBDPS protecting group
was chosen with a view toward suppressing a possible â-
chelation pathway relative to the desired R-chelation mode in
the LACDAC sequence. Earlier research of Reetz33 provided a
suggestion that oxygen chelation effects in the control of
diastereofacial reactions are suppressed in bulky silyl ether
settings.
Indeed, as intended, reaction of aldehyde 8 and diene 334
under the influence of TiCl4 yielded the R-chelation controlled
product 9 (Scheme 3).35 Treatment of dihydropyrone 9 with
NaBH4 and CeCl3‚7H2O (Luche reduction)36 led to the corre-
sponding 1,2-reduced compound, which underwent a Ferrier
rearrangement37 in aqueous acidic THF to produce lactol 10,
with the desired (Z)-olefin now in place. Reductive opening of
lactol 10 with LiBH4 afforded diol 11 in 55% overall yield from
dihydropyrone 9. The primary hydroxyl group of 11 was
selectively acylated with 2,6-heptadienoyl chloride 12,38 and,
Results and Discussion
Model Study. It seemed prudent to assess, in the context of
a model study, the feasibility of RCM to construct the
14-membered ring of migrastatin (Scheme 3).31 In this connec-
tion, we would also address the stereoselectivity (geometry of
the C6-C7 double bond) and chemoselectivity (undesired
RCM-participation of the C2-C3 and C11-C12 double bonds)
of the ring-closing reaction. Such questions were to be first
posed in a study directed to the synthesis of the migrastatin
core structure lacking the glutarimide-containing side chain.
(32) S)-3-benzyloxy-1,2-propanediol 7 is commercially available (Fluka, Ald-
rich), but only at a high cost. Compound 7 can be easily prepared from
inexpensive starting materials: (a) Xiang, G.; McLaughlin, L. W. Tetra-
hedron 1998, 54, 375-392. (b) Kitaori, K.; Furukawa, Y.; Yoshimoto, H.;
Otera, J. Tetrahedron 1999, 55, 14381-14390.
(33) Reetz, M. T.; Kessler, K. J. Org. Chem. 1985, 50, 5434-5436.
(34) Danishefsky, S. J.; Yan, C. F.; Singh, R. K.; Gammill, R. B.; McCurry, P.
M., Jr.; Fritsch, N.; Clardy, J. J. Am. Chem. Soc. 1979, 101, 7001-7008.
(35) For chelation-controlled cyclocondensations of R-alkoxy aldehydes with
synergistically activated dienes, see: Danishefsky, S. J.; Pearson, W. H.;
Harvey, D. F.; Maring, C. J.; Springer, J. P. J. Am. Chem. Soc. 1985, 107,
1256-1268.
(36) Luche, J. L.; Gemal, A. L. J. Am. Chem. Soc. 1979, 101, 5848-5849.
(37) Ferrier, R. J. J. Chem. Soc. 1964, 5443-5450.
(38) (a) Katzenellenbogen, J. A.; Crumrine, A. L. J. Am. Chem. Soc. 1976, 98,
4925-4935. (b) Ahmar, M.; Duyck, C.; Fleming, I. J. Chem. Soc., Perkin
Trans. 1 1998, 2721-2732.
(29) Danishefsky, S. J.; Kitahara, T. J. Am. Chem. Soc. 1974, 96, 7807-7808.
(30) (a) Danishefsky, S. J. Aldrichimica Acta 1986, 19, 59-69. (b) Danishefsky,
S. J. Chemtracts 1989, 2, 273-297.
(31) Gaul, C.; Danishefsky, S. J. Tetrahedron Lett. 2002, 43, 9039-9042.
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11329

A R T I C L E S
Gaul et al.
Scheme 4. Synthesis of Dihydropyrone 21 by a
Cyclocondensation (LACDAC)a
Figure 1. Structures of the Grubbs-II (16) and Grubbs-I (17) catalysts.
thereafter, the secondary hydroxyl group in the acylation product
was protected as its MOM ether.
The RCM precursor 14 was reached from 13 through a three-
step sequence, consisting of deprotection, oxidation, and Tebbe
olefination.39 When tetraene 14 was subjected to the action of
Grubbs catalyst 1640 (Figure 1) in refluxing toluene,41 the 14-
membered macrolactone 15 was generated as the desired (E)-
congener in 50% yield. Competitive participation of the electron-
poor C2-C3 double bond and the sterically hindered C11-
C12 double bond in the metathesis step could not be detected.
Interestingly, treatment of 14 with Grubbs-I catalyst 17 (Figure
1) in refluxing CH2Cl2 led exclusively to the dimeric product
derived from cross metathesis of the terminal double bond of
the acyl moiety.
a Reagents and conditions: (a) DIBALH, then ZnCl2, H2CdCHMgBr,
toluene, -78 °C to room temperature, 75% (ds > 90%); (b) (i) MeI, NaH,
DMF, room temperature, (ii) 2 M HCl, MeOH, reflux, 80%; (c) Pb(OAc)4,
Na2CO3, CH2Cl2, 0 °C to room temperature; (d) (i) TiCl4, CH2Cl2, -78
°C, (ii) TFA, CH2Cl2, room temperature, 87% from 20.
Scheme 5. Synthesis of Key Intermediate 26a
Synthesis of Key Intermediate 26. The model study
demonstrated the efficacy of the LACDAC sequence to construct
the three stereocenters C8-C10 and the power of RCM to
establish the macrocyclic system. Encouraged by these early
results, we embarked on the total synthesis of migrastatin. Prior
to facing the unresolved issues of building up the remaining
stereocenters at C13 and C14 in the context of emplacement of
the glutarimide moiety, we addressed an issue of process, asking
the question whether R-methoxy-R-vinyl aldehyde 2 might
indeed serve as a suitable heterodienophile in the reaction with
diene 3 after all. Utilization of aldehyde 2 in the LACDAC
reaction would allow us to streamline the synthesis in a most
useful way by avoiding the chemistry needed to incorporate
the C6-C7 double bond required for RCM.
Happily, we could gain an excellent entry into this type of
aldehyde, starting from commercially available dimethyl 2,3-
O-isopropylidene-L-tartrate 18 (Scheme 4). Toward this end,
tartrate 18 was reduced by DIBALH to the corresponding
dialdehyde, which was then reacted in situ with divinylzinc to
afford carbinol 19 in a highly stereoselective fashion.42 Di-
methylation and cleavage of the acetonide protecting group with
aqueous acid furnished 1,2-diol 20 in excellent yield.43 The
desired R-methoxy-R-vinyl aldehyde 2 emerged following
cleavage of the glycol linkage of 20. Importantly, no attempts
were undertaken to isolate 2 in neat form. Instead, a stock
solution of the aldehyde as obtained from the glycol cleavage
was directly used for the LACDAC sequence. We were rather
encouraged to find that the R-chelation-controlled cyclocon-
densation of 2 with butadiene 3 occurred in very good yield,
producing dihydropyrone 21 as the only detected diastereomer.
a Reagents and conditions: (a) LiBH4, MeOH, THF, -10 °C; (b) CSA,
H2O, THF, reflux; (c) LiBH4, H2O, THF, room temperature, 53% from 21;
(d) (i) TBSOTf, 2,6-lutidine, CH2Cl2, room temperature, (ii) HOAc, H2O,
THF (3:1:1), room temperature, 80%.
Compound 21 not only possesses the three contiguous stereo-
centers of the macrolide, but it also serves as a template for the
construction of the trisubstituted C11-C12 (Z)-alkene (Scheme
5). From a process standpoint, it is noteworthy that only two
chromatographic purifications were needed to obtain pure 21,
rendering the sequence amenable to scale-up.
Transformation of dihydropyrone 21 into open-chain diol 25
was accomplished as described for our model study (Scheme
3) using a reduction-Ferrier-rearrangement-reductive ring-open-
ing protocol (Scheme 5). Initially, we followed the Luche
procedure for the reduction of enones to effect the conversion
of 21 to 22. Subsequently, we found that the addition of cerium
salts was not needed in our case. In fact, all of the reductants
screened led exclusively to 1,2-reduction. In the end, LiBH4
turned out to be the reducing agent of choice based on its
associated ease of handling and workup. When alcohol 22 was
subjected to catalytic amounts of camphorsulfonic acid (CSA)
in refluxing aqueous THF, the desired Ferrier-rearranged product
23 was obtained, together with small amounts of dimeric acetal
24. It is appropriate to note that there are few examples of
aqueous Ferrier rearrangements reported in the literature,44
whereas variants that involve alcohol-based nucleophiles are
widely encountered. Reductive opening of lactol 23 with LiBH4
(39) Tebbe, F. N.; Parshall, G. W.; Reddy, G. S. J. Am. Chem. Soc. 1978, 100,
3611-3613.
(40) For initial reports of Grubbs-II catalyst 16, see: (a) Scholl, M.; Trnka, T.
M.; Morgan, J. P.; Grubbs, R. H. Tetrahedron Lett. 1999, 40, 2247-2250.
(b) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999, 1,
953-956.
(41) For the first report of these new, optimized RCM conditions, see: (a)
Yamamoto, K.; Biswas, K.; Gaul, C.; Danishefsky, S. J. Tetrahedron Lett.
2003, 44, 3297-3299. The same reaction conditions were also applied to
the first total synthesis of epothilone 490: (b) Biswas, K.; Lin, H.;
Njardarson, J. T.; Chappell, M. D.; Chou, T. C.; Guan, Y.; Tong, W. P.;
He, L.; Horwitz, S. B.; Danishefsky, S. J. J. Am. Chem. Soc. 2002, 124,
9825-9832.
(42) Jorgensen, M.; Iversen, E. H.; Paulsen, A. L.; Madsen, R. J. Org. Chem.
2001, 66, 4630-4634.
(43) Lee, W. W.; Chang, S. Tetrahedron: Asymmetry 1999, 10, 4473-4475.
9
11330 J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Scheme 6. Incorporation of the Side Chain by an Anti-Aldol
afforded diol 25 in 53% overall yield (from dihydropyrone 21).
In investigating the preparative aspects of the sequence 21 f
25, we realized that larger amounts of 24 (ca. 15%) were isolated
when the Ferrier rearrangement was conducted at higher
concentrations (0.3 M instead of 0.1 M). Happily, we were able
to obtain single crystals of dimer 24. X-ray analysis led to a
decisive structural verification,45 revealing the relative config-
uration of the three contiguous stereocenters and the geometry
of the double bond to be as predicted on the basis of our
precedents. By extension, this X-ray analysis also confirms the
structure of diol 25.
Reaction and a HWE Couplinga
In the next phase of our synthesis of migrastatin, it was
necessary to differentiate the two hydroxyl groups of 25.
Previously, we had accomplished this sub-goal via a three-step
sequence, acetylation of the primary hydroxyl group, silylation
of the secondary hydroxyl group, and subsequent removal of
the acetate protecting group.27 However, during scale-up efforts,
we observed the formation of considerable amounts of di-
acetylated product. This obstacle complicated purification
procedures and lowered the overall yield of the sequence.
Fortunately, the problem could easily be managed by initial
disilylation, followed by a mild and selective deprotection,
producing allylic alcohol 26 in 80% yield (Scheme 5).
a Reagents and conditions: (a) Dess-Martin periodinane, CH2Cl2, room
temperature; (b) (i) MgCl2, Et3N, TMSCl, EtOAc, room temperature, (ii)
TFA, MeOH, room temperature, 67% from 26; (c) TESCl, imidazole,
CH2Cl2, room temperature; (d) LiBH4, MeOH, THF, room temperature,
83% from 29; (e) (i) Dess-Martin periodinane, CH2Cl2, room temperature,
(ii) dimethyl methylphosphonate, BuLi, THF, -78 to 0 °C, (iii) Dess-
Martin periodinane, CH2Cl2, room temperature; (f) LiCl, DBU, MeCN, room
temperature, 57% from 31; (g) (i) [(Ph3P)CuH]6, toluene, room temperature,
(ii) HOAc, H2O, THF (3:1:1), room temperature, 82%.
Incorporation of the Glutarimide-Containing Side Chain.
A straightforward way to construct the two remaining stereo-
centers at C13 and C14 could, in principle, be accomplished
by an anti-selective aldol reaction between an aldehyde derived
from 26 and an appropriate propionyl fragment. Indeed, Dess-
Martin oxidation46 of 26 generated angelic-type aldehyde 27
(Scheme 6). Fortunately, 27 proved to be notably resistant to
(Z) f (E)-double bond isomerization or vinylogous epimeriza-
tion and, accordingly, could serve as a potential substrate in
the aldol construction. In our early studies, we explored
Masamune’s anti-aldol protocol, which utilizes a boron enolate
of readily available norephedrine derivatives.47 This aldol
reaction indeed worked smoothly with aldehyde 27. Nonetheless,
we were particularly drawn to a mild MgCl2-catalyzed anti-
aldol procedure that had recently been disclosed by Evans.48 In
practice, aldehyde 27 reacted with propionyl oxazolidinone 28
in the presence of MgCl2, triethylamine, and TMSCl to afford,
after treatment with TFA, the desired aldol adduct 29 in 67%
yield as a single diastereomer. Noteworthy, the robust reaction
conditions, which tolerate the use of reagent-grade ethyl acetate
and high substrate concentrations, are attractive features for
scale-up purposes. Because the next step of the total synthesis
was the protection of the C13 hydroxyl group as a TES ether,
we tried to accomplish the anti-aldol joining with TESCl instead
of TMSCl. Unfortunately, the reaction was very slow under
these conditions, and the yields were far from satisfactory.
Hence, we had to protect the secondary hydroxyl group with
TESCl in a separate step (29 f 30) (Scheme 6).
fragment. A Horner-Wadsworth-Emmons (HWE) reaction
between â-ketophosphonate 32 and aldehyde 5 appeared to be
a plausible, attractive solution to this synthetic problem (Scheme
6). Toward this end, we investigated the direct addition of
lithiated dimethyl methylphosphonate to imide 30 to access the
desired phosphonate 32 in a single transformation. Unfortu-
nately, this transformation met with no success, resulting in
recovery of starting material.49 Accordingly, the chiral auxiliary
was removed reductively (30 f 31). Progress continued with a
simple and reliable three-step oxidation-addition-reoxidation
protocol, cleanly affording phosphonate 32. Glutarimide alde-
hyde 5,50 the fourth component in our synthetic plan, was then
treated with phosphonate 32 using the Masamune-Roush
variant of the HWE reaction.51 Enone 33 was obtained as a
single olefin isomer in excellent yield (Scheme 6). Fortunately,
neither this reaction nor any of the subsequent transformations
required protection of the glutarimide nitrogen. Conjugate
reduction of enone 33 with the Stryker reagent52 and cleavage
of the TES protecting group occurred smoothly to give alcohol
(48) Evans, D. A.; Tedrow, J. S.; Shaw, J. T.; Downey, C. W. J. Am. Chem.
Soc. 2002, 124, 392-393.
Having successfully merged three of our five components,
we focused now on attaching glutarimide aldehyde 5 to the main
(49) For examples of the direct addition of lithiated dimethyl methylphosphonate
to esters, see, for example: (a) Edmonds, M. K.; Abell, A. D. J. Org. Chem.
2001, 66, 3747-3752. (b) Smith, A. B., III; Frohn, M. Org. Lett. 2001, 3,
3979-3982.
(44) (a) Danishefsky, S. J.; Kato, N.; Askin, D.; Kerwin, J. F., Jr. J. Am. Chem.
Soc. 1982, 104, 360-362. (b) Eng, H. M.; Myles, D. C. Tetrahedron Lett.
1999, 40, 2279-2282.
(45) Crystallographic data (excluding structural data) for compound 24 have
been deposited with the Cambridge Crystallographic Data Centre (CCDC)
as Deposition No. CCDC 230121.
(46) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277-7287.
(47) Abiko, A.; Liu, J. F.; Masamune, S. J. Am. Chem. Soc. 1997, 119, 2586-
2587.
(50) Egawa, Y.; Suzuki, M.; Okuda, T. Chem. Pharm. Bull. 1963, 11, 589-
595.
(51) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. M.; Masamune,
S.; Roush, W. R.; Sakai, T. Tetrahedron Lett. 1984, 25, 2183-2186.
(52) Mahoney, W. S.; Brestensky, D. M.; Stryker, J. M. J. Am. Chem. Soc.
1988, 110, 291-293. The Stryker reagent provided by Aldrich performed
poorly in the conjugate reduction. The reagent provided by Fluka or
prepared by us led to superior results.
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11331

A R T I C L E S
Gaul et al.
Scheme 7. Acylation of Alcohol 34 by a Modified Yamaguchi
Procedurea
Figure 2. Regions of migrastatin targeted for derivatization.
delivering macrolactone 37 in a highly (E)-selective fashion in
69% yield. This corresponds to an increase in yield by almost
20% as compared to our model studies! Finally, removal of the
TBS protecting group by buffered hydrogen fluoride completed
the total synthesis of (+)-migrastatin (1), whose physical data
(NMR, MS, optical rotation) matched those of migrastatin
isolated from natural sources.
Design, Chemical Synthesis, and Evaluation of Migrasta-
tin Analogues. Having achieved the first key goal, the total
synthesis of migrastatin, we could take full advantage of our
flexible multicomponent synthesis for the subsequent analogue
program. As will be evident, our modular approach served as
an excellent platform from which to quickly explore the SAR
profile of migrastatin and assess the anti-metastatic potential
of the migrastatin family.
a Reagents and conditions: (a) 2,4,6-trichlorobenzoyl chloride, i-Pr2NEt,
pyridine, toluene, room temperature, 67%.
Scheme 8. RCM and Deprotection Leading to (+)-Migrastatin (1)a
Our modus operandi in searching for, preparing, and inter-
rogating migrastatin derivatives as to improved cell migration
inhibition properties was comprised of three distinct steps:
design, chemical synthesis, and biological evaluation. The design
of the analogues was aimed at probing the different regions of
migrastatin for their contributions to biological activity. The
domains of the molecule that we considered important and
accessible by synthesis are highlighted in red as shown in Figure
2. The selection was driven by the following considerations:
The glutarimide moiety is certainly one the most characteristic
functional groups of migrastatin which might be indispensable
for activity. The C2-C3 conjugated double bond is a potential
site for deactivation by 1,4-addition of nucleophiles (e.g., thiols,
confer the natural products NK30424A/B in Scheme 1) or, on
the contrary, could render migrastatin a suicide inhibitor by
covalent bond formation with bionucleophiles present in the
active site of an enzyme. The lactone functionality could
possibly be a target of hydrolysis in living systems. As such,
manipulation of the ester bond might enhance the in vivo
stability of the molecule. Furthermore, the C6-C12 portion of
migrastatin is highly functionalized and, thus, might have
biological relevance. One simple way of exploring this region
is through derivatization of the C9 hydroxyl group.
a Reagents and conditions: (a) Grubbs-II catalyst 16 (20 mol %), toluene
(0.5 mM), reflux, 69%; (b) HF‚pyridine, THF, room temperature, 85%.
34. At this stage, the path was clear for introduction of our last
component, 2,6-heptadienoic acid 6.
Completion of the Total Synthesis of Migrastatin. In our
planning stages, we presumed that acylation of secondary
alcohol 34 with acid 6 would be straightforward. Unexpectedly,
a number of acylation conditions had to be explored to
accomplish the desired transformation effectively. Only after
extensive experimentation did we find that a modified Yamagu-
chi acylation protocol53 (using pyridine instead of DMAP)
provided satisfying yields of acylated product 35 (Scheme 7).
Most other standard ester formation protocols (a, acid chloride
+ DMAP, pyridine, or AgCN; b, acid + EDC or DCC; c, acid
+ Mukaiyama reagent;54 d, Keck coupling55) led to either
decomposition of starting material or an inseparable product
mixture of 35 and â,γ-unsaturated ester 36 (Scheme 7). The
latter presumably arose from acylation of 34 with the vinylketene
derived from 6 upon activation of the acyl group. Attempts to
isomerize the C3-C4 double bond of 36 back into conjugation
resulted in loss of the carboxylic acid fragment, apparently
through a â-elimination pathway.
The chemical synthesis of the analogues was accomplished
in an efficient manner by utilizing the concept of diverted total
synthesis (DTS).28 We put to advantage key intermediates of
the migrastatin synthesis, such as 26 and 34 (Schemes 5 and
6), as branching points to rapidly assemble a diverse, but still
limited, set of migrastatin derivatives. In keeping with the unique
capabilities of diverted total synthesis, we focused on target
structures that would not have been accessible through manipu-
lations of the natural product itself or through biosynthetic
pathways.
With RCM precursor 35 now available, we were positioned
to investigate the cyclization reaction (Scheme 8). In the event,
the ring-closing metathesis conditions employed in our model
system (Scheme 3) also sufficed nicely for the case at hand,
(53) (a) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M. Bull.
Chem. Soc. Jpn. 1979, 52, 1989-1993. For a recent example, see: (b)
Song, F.; Fidanze, S.; Benowitz, A. B.; Kishi, Y. Org. Lett. 2002, 4, 647-
650.
The biological evaluation of our compounds (in terms of their
ability to inhibit cell migration) was accomplished in a Boyden
chamber cell migration assay. In this assay, mouse breast tumor
(54) (a) Mukaiyama, T.; Usui, M.; Shimada, E.; Saigo, K. Chem. Lett. 1975,
1045-1048. (b) Mukaiyama, T. Angew. Chem., Int. Ed. Engl. 1979, 18,
707-721.
(55) Boden, E. P.; Keck, G. E. J. Org. Chem. 1985, 50, 2394-2395.
9
11332 J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Scheme 9. Preparation of Migrastatin Analogues 41 and 42a
Table 1. Chamber Cell Migration Assay with 4T1 Tumor Cells
compound
IC50 (4T1 tumor cells)a
migrastatin (1)
29 µM
2,3-dihydromigrastatin (41)
N-methyl-2,3-dihydromigrastatin (42)
migrastatin core (45)
10 µM
7.0 µM
22 nM
macrolactone (48)
24 nM
acetylated macrolactone (49)
oxidized macrolactone (50)
hydrolyzed core (51)
macrolactam (55)
macroketone (60)
(S)-isopropyl macrolactone (65)
(R)-isopropyl macrolactone (68)
macrocyclic secondary alcohol (69)
macrocyclic tertiary alcohol (70)
macrocyclic CF3-alcohol (71)
macrooxime (72)
192 nM
223 nM
378 nM
255 nM
100 nM
227 µM
146 µM
8.9 µM
3.1 µM
101 nM
2.3 µM
331 nM
26 nM
biotinylated macrohydrazone (73)
epoxyquinol
a Reagents and conditions: (a) 6-heptenoyl chloride 38, DMAP, CH2Cl2,
room temperature, 69%; (b) Grubbs-II catalyst 16 (20 mol %), toluene (0.5
mM), reflux, 79%; (c) HF‚pyridine, THF, room temperature, 81%; (d) MeI,
Cs2CO3, acetone, 85%.
evodiamine
315 nM
a Average of three experiments. Each experiment consists of nine data
points (nine different concentrations).
cells (4T1 cells) or endothelial cells (HUVECs) are seeded on
the upper chamber of a transwell insert. Growth factor-
containing serum is added to the lower chamber. After incuba-
tion for 6-8 h in the presence of different concentrations of
our analogues, cells that migrated from the upper chamber
through the membrane to the lower compartment are counted.
Additionally, some of our more potent compounds were tested
for their effect on cell proliferation and metabolic stability in
mouse plasma. As will be demonstrated below, the program
helped to provide us with a broad SAR picture in the context
of our migrastatin studies.
Scheme 10. Synthesis of Migrastatin Core 45 and Macrolactone
48a
Our search commenced with the synthesis of 2,3-dihydro-
migrastatin 41 and N-methylated 2,3-dihydromigrastatin 42.
Secondary alcohol 34 (Scheme 6), an advanced intermediate
of the migrastatin synthesis, was acylated with 6-heptenoyl
chloride 38 to deliver RCM precursor 39 (Scheme 9). As
expected, acylation proceeded smoothly, without resort to the
special reaction conditions required for the acylation of 34 with
2,6-heptadienoic acid 6 (Scheme 7). Compound 39 was cyclized
to macrolactone 40 by a very efficient (E)-selective RCM.
Cleavage of the TBS ether with HF‚pyridine yielded our first
analogue, 2,3-dihydromigrastatin 41. Alternatively, compound
41 was prepared directly from migrastatin by regioselective
reduction using the Stryker reagent (Scheme 9). The yield of
the direct transformation, however, was compromised by the
formation of a side product that arose from an intramolecular
aldol addition of the transient copper enolate to the C15 ketone.
Methylation of the glutarimide nitrogen was accomplished by
treatment of 41 with MeI and Cs2CO3 in acetone, delivering
methylated 2,3-dihydromigrastatin 42 in excellent yield.
a Reagents and conditions: (a) 2,6-heptadienoic acid 6, 2,4,6-trichloro-
benzoyl chloride, i-Pr2NEt, pyridine, toluene, room temperature, 48%; (b)
Grubbs-II catalyst 16 (20 mol %), toluene (0.5 mM), reflux, 55% (44),
76% (47); (c) HF‚pyridine, THF, room temperature, 66% (45), 94% (48);
(d) 6-heptenoyl chloride 38, DMAP, CH2Cl2, room temperature, 82%.
The small change in inhibitory activity upon alkylation of
the glutarimide moiety encouraged us to undertake a more
drastic structural modification of the migrastatin skeleton.
Toward this end, analogues were synthesized lacking the entire
glutarimide-containing side chain, migrastatin core 45 and the
corresponding reduced version 48 (Scheme 10). Starting from
advanced key intermediate 26 (Scheme 5), derivatives 45 and
48 were quickly assembled via the already established acylation-
RCM-deprotection sequence. While the reaction of 26 with 2,6-
heptadienoic acid 6 produced acylated product 43 in only
moderate (48%) yield, the acylation steps in the “dihydro series”
occurred smoothly, affording 46 in 82% yield. The same trend
was observed for the subsequent transformation, in which the
ring closure was achieved in excellent (76%) yield for the
saturated case (46 f 47). By contrast, the unsaturated core was
delivered in lower (55%) yield (43 f 44). Finally, protecting
group removal delivered macrolactones 45 and 48 without
complications.
The first set of compounds - migrastatin, together with its
analogues 41 and 42 - was then evaluated in the chamber cell
migration assay. The IC50 value for fully synthetic migrastatin
with 4T1 tumor cells was 29 µM (Table 1); this result was in
excellent agreement with that reported by Imoto for migrastatin
obtained from natural sources. Interestingly, reduction of the
C2-C3 double bond and methylation of the glutarimide nitrogen
were well tolerated with respect to maintenance of activity.
Analogues 41 and 42 are actually slightly more potent than
migrastatin itself, with IC50 values of 10 and 7 µM, respectively
(Table 1).
Given our goal of evaluating the efficacy of migrastatin
analogues in animal models, we undertook a brief study of the
RCM parameters for a potential large-scale preparation of the
macrocycles. The original reaction conditions called for 20 mol
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11333

A R T I C L E S
Gaul et al.
Scheme 11. Optimization of the RCM Conditions for Scale-Up
Purposes
Table 2. Metabolic Stability of Selected Compounds in Mouse
Plasma
compound
stability (t1/2, mouse plasma)
migrastatin (1)
stablea
stablea
stablea
20 min
<5 min
stablea
stablea
stablea
stablea
2,3-dihydromigrastatin (41)
N-methyl-2,3-dihydromigrastatin (42)
migrastatin core (45)
macrolactone (48)
macrolactam (55)
macroketone (60)
(S)-isopropyl macrolactone (65)
(R)-isopropyl macrolactone (68)
a Intensity of HPLC signal unchanged over 60 min of incubation.
Scheme 12. Modification of Macrolactone 48 at the C9 Position
and Hydrolysis of 48a
Scheme 13. Synthesis of Macrolactam 55 and Macroketone 60a
a Reagents and conditions: (a) AcCl, DMAP, CH2Cl2, room temperature,
76%; (b) Dess-Martin periodinane, CH2Cl2, room temperature, 72%; (c)
0.5 M NaOH, MeOH, room temperature, 77%.
% catalyst at 0.5 mM concentration, but as illustrated for the
cyclization of 46 to 47 (Scheme 11), the RCM product could
be obtained in just slightly reduced yield by conducting the
reaction with 10 mol % catalyst at 5 mM concentration.
Upon examination of compounds 45 and 48 in the cell
migration assay, we achieved, to our great surprise, a major
breakthrough in potency (Table 1). The IC50 values for mi-
grastatin core 45 and macrolactone 48 were found to be 22 and
24 nM, respectively. This translates into an increase in activity
by 3 orders of magnitude as compared to migrastatin! The assay
outcome clearly shows that the migrastatin side chain is not
required for in vitro inhibition of tumor cell migration, but it
remains to be seen if migrastatin and the core analogues 45
and 48 are indeed directed at the same cellular targets.
Another potential “hot spot” of migrastatin is the C6-C12
region with its two double bonds, three stereocenters, and two
heteroatoms. An easy way of derivatizing this portion of the
molecule was found to be the acylation or oxidation of the C9
hydroxyl group, producing macrolactones 49 and 50, respec-
tively (Scheme 12). As shown in Table 1, the inhibitory activity
of analogues 49 and 50 was reduced by roughly an order of
magnitude, as compared to macrolactone 48, indicating that the
C9 position is sensitive toward modification.
Starting from our lead compound migrastatin, we were able
to reach simplified congeners with drastically improved inhibi-
tory activities, in particular, macrolactones 45 and 48. Anticipat-
ing in vivo evaluation of our compounds, we undertook to
evaluate, in a preliminary way, their metabolic stabilities. Based
on our experiences from the epothilone program,56 the ester bond
of the macrolactones could well be susceptible toward opening
by esterases in mouse (or human) plasma. Such hydrolysis
would lead to deactivation of the compound. From this
perspective, we undertook to determine the mouse blood plasma
stability of migrastatin and the novel analogues 41, 42, 45, and
a Reagents and conditions: (a) DPPA (diphenylphosphoryl azide), DBU,
toluene, room temperature, 87%; (b) (i) PPh3, H2O, THF, 70 °C, (ii)
6-heptenoic acid, EDC, i-Pr2NEt, CH2Cl2, room temperature, 92%; (c)
Grubbs-II catalyst 16 (20 mol %), toluene (0.5 mM), reflux, 60% (54),
81% (59); (d) HF‚pyridine, THF, room temperature, 81% (55), 90% (60);
(e) CBr4, solid supported PPh3, CH2Cl2, room temperature; (f) (i) â-keto-
sulfone 57, DBU, toluene, room temperature, (ii) Na/Hg, Na2HPO4, MeOH,
room temperature, 61% from 26.
48. As summarized in Table 2, migrastatin and the side-chain-
containing derivatives 41 and 42 were completely inert toward
lactone opening over the full test period (1 h). However, the
most active compounds, macrolactones 45 and 48, were
hydrolyzed rapidly (Table 2). These findings are not entirely
surprising, considering that the ester bonds in 45 and 48 are
sterically less congested relative to those in the other analogues.
As a test of the “deactivation hypothesis”, the hydrolysis product
of macrolactone 48 was prepared (Scheme 12) and tested for
its activity against tumor cell migration. Surprisingly, compound
51 was not completely inactive in the chamber assay, but
retained a good part of its activity (IC50 value of 378 nM).
Therefore, compound 48 might be effective in the projected in
vivo models despite its sensitivity toward hydrolysis. Neverthe-
less, the data on the unsatisfactory metabolic stability of 45 and
48 influenced us, when we entered the second phase of our
analogue program. The aspiration of reaching migrastatin
congeners with enhanced plasma stability and retained or
improved activity (as compared to 45 or 48) led to the diverted
total synthesis of macrolactam 55, macroketone 60 (Scheme
13), and the sterically hindered macrolactones 65 and 68
(Scheme 14).
The synthesis of analogues 55, 60, 65, and 68 diverged from
the original route to migrastatin at the stage of the advanced
key intermediate 26. For the preparation of lactam 55, alcohol
26 was subjected to Mitsunobu conditions with DPPA affording
allylic azide 52 in 87% yield (Scheme 13).57 To avoid double
(56) Stachel, S. J.; Lee, C. B.; Spassova, M.; Chappell, M. D.; Bornmann, W.
G.; Danishefsky, S. J. J. Org. Chem. 2001, 66, 4369-4378 (includes an
example for the Mitsunobu-Staudinger sequence).
9
11334 J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Scheme 14. Synthesis of the Diastereomeric Isopropyl
Macrolactones 65 and 68a
Scheme 15. Derivatization of Macroketone 60a
a Reagents and conditions: (a) NaBH4, MeOH, room temperature, 95%;
(b) MeMgBr, THF, 0 °C to room temperature, 95%; (c) TMSCF3, TBAF,
THF, 0 °C to room temperature, 80%; (d) NH2OH‚HCl, pyridine, 45 °C,
70%; (e) Biotin-dPEG4-hydrazide, EtOH, 55 °C, 75%.
a Reagents and conditions: (a) (i) Dess-Martin periodinane, CH2Cl2,
room temperature, (ii) i-PrMgCl, Et2O, -78 °C, 86% (3:2-mixture of 61
and 62); (b) (i) TsCl, pyridine, THF, room temperature, (ii) LiAlH4, Et2O,
room temperature, (iii) HOAc, H2O, THF (3:1:1), room temperature, yield
not determined; (c) 6-heptenoic acid, 2,4,6-trichlorobenzoyl chloride,
i-Pr2NEt, pyridine, toluene, room temperature, 75% (63), 70% (66); (d)
Grubbs-II catalyst 16 (20 mol %), toluene (0.5 mM), reflux; (e) HF‚pyridine,
THF, room temperature, 65% (65 from 63), 66% (68 from 66).
(R)-MPA esters (MPA ) R-methoxyphenylacetic acid) and
analyzed by NMR, leading to the assignment of the newly
created stereocenter:59 Major isomer 61 has the “unnatural” (S)-
configuration, and minor isomer 62 has the “natural” (R)-
configuration. In addition, the results of the NMR experiment
were probed by a degradation study. We were able to transform
compound 31 (Scheme 6), a synthetic intermediate of the total
synthesis of migrastatin, into minor isomer 62, thereby delivering
convincing proof for the correctness of the configurational
assignment. The transformation was accomplished by converting
alcohol 31 into its tosylate, reducing the tosylate with LiAlH4,60
and removing the TES protecting group. For the preparation of
lactones 65 and 68, the addition products 61 and 62 were
separated and independently acylated (Scheme 14). Intermedi-
ates 63 and 66 were then subjected to our RCM conditions,
furnishing the macrocycles 64 and 67 in very good yield.
Deprotection occurred smoothly and provided the diastereomeric
isopropyl lactones 65 and 68.
Indeed, lateral modification of the vulnerable ester bond of
macrolactones 45 and 48 for more robust entities led to the
desired effect of enhanced metabolic stability in all four cases:
Macrolactam 55, macroketone 60, and isopropyl macrolactones
65 and 68 display no sign of degradation in our assay (Table
2). When tested for their ability to inhibit 4T1 cell migration,
compounds 55 and 60 were found to be considerably more active
than the natural product migrastatin (255 and 100 nM, respec-
tively, Table 1), although some loss of potency relative to
lactones 45 and 48 was recorded. Surprisingly, incorporation
of an isopropyl group at C13 proved to be deleterious for
biological function. Isopropyl macrolides 65 and 68 exhibited
only very weak effects on tumor cell migration (Table 1).
As depicted in Scheme 15, our SAR studies were further
diverted on the basis of macroketone 60. The ketone functional-
ity proved to be an attractive handle for additional derivatization.
bond isomerization of the (Z)-allylic system, azide 52 was
immediately reduced following the Staudinger protocol57 and
subsequently joined with 6-heptenoic acid under standard
peptide coupling conditions. The resulting product, amide 53,
was treated with RCM catalyst 16 under our established reaction
conditions, delivering lactam 54 in 60% yield. The latter was
then deprotected with HF‚pyridine to afford lactam 55.
The preparation of ketone 60 required the conversion of
alcohol 26 into allylic bromide 56, which was displaced by
â-ketosulfone 57 (Scheme 13).57 Subsequent reductive removal
of the sulfone group yielded RCM precursor 58. The ring closure
of 58 to the carbocycle was accomplished, again, very efficiently
and selectively by RCM. The desired macroketone 60 was
obtained following deprotection of 59.
The synthesis of the isopropyl macrolactones 65 and 68 also
commenced from alcohol 26 (Scheme 14). Oxidation of 26
generated the corresponding (Z)-enal, which was then treated
with i-PrMgCl. When the nucleophilic addition was carried out
in THF, an equimolar mixture of the desired addition products
61/62 (3:2 ratio) and the reduced product 26 was obtained. It is
well documented in the literature that addition of isopropyl-
Grignard reagents to hindered substrates must compete with
reduction through hydride delivery from the nucleophile.58
Fortunately, the product ratio could be improved by changing
the solvent from THF to Et2O. The reduction pathway was
almost completely suppressed by slow addition of i-PrMgCl to
a solution of the aldehyde in Et2O, while carefully maintaining
the reaction temperature at -78 °C for several hours. The
diastereomers 61 and 62 were derivatized as their (S)-MPA and
(59) (a) Trost, B. M.; Bunt, R. C.; Pulley, S. R. J. Org. Chem. 1994, 59, 4202-
4205. (b) Seco, J. M.; Latypov, S. K.; Quinoa, E.; Riguera, R. Tetrahedron
1997, 53, 8541-8564. (c) Seco, J. M.; Quinoa, E.; Riguera, R. Tetrahedron:
Asymmetry 2000, 11, 2781-2791.
(57) Chun, J.; Li, G.; Byun, H. S.; Bittman, R. J. Org. Chem. 2002, 67, 2600-
2605.
(58) For a recent example, see: Dixon, D. J.; Krause, L.; Ley, S. V. J. Chem.
Soc., Perkin Trans. 1 2001, 2516-2518.
(60) Li, D. R.; Xia, W. J.; Shi, L.; Tu, Y. Q. Synthesis 2003, 41-44.
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11335

A R T I C L E S
Gaul et al.
As discussed above, there are other recently discovered
natural products that are reported to be strong cell migration
inhibitors. In particular, two compounds, epoxyquinol A,12
a
pentaketide dimer with anti-angiogenic activity, and evodi-
amine,15 a potent anti-metastatic and anti-invasive alkaloid,
attracted great interest and are currently under serious investiga-
tion by several research groups (Figure 3). We regarded it as a
valuable experiment to test these natural products side by side
with our migrastatin analogues for the purpose of validating
and calibrating our assay. As shown in Table 1, our simple
macrolactones outperform evodiamine and are comparable to
epoxyquinol A in the chamber assay.
Figure 3. Structures of epoxyquinol A and evodiamine.
Table 3. Chamber Cell Migration Assay with Human Endothelial
Cells (HUVECs)
compound
IC50 (HUVEC)a
migrastatin (1)
65 µM
150 nM
125 nM
18 µM
migrastatin core (45)
macrolactone (48)
macrolactam (55)
macroketone (60)
Due to the significance of endothelial cell migration in the
angiogenesis process, the chamber cell migration assay described
above was also conducted with HUVECs (human umbilical vein
endothelial cells) and used for the evaluation of our most potent
analogues, macrolactones 45 and 48, macrolactam 55, and
macroketone 60, together with migrastatin as a reference. The
IC50 values obtained from this study are listed in Table 3. The
general trend in activity, with the simplified analogues 45, 48,
55, and 60 being significantly more active against 4T1 tumor
cell migration than the parent natural product, was also observed
for endothelial cells. However, some erosion of potency in the
HUVEC determination as compared to the 4T1 cell determi-
nation was observed for all compounds tested.
12 µM
a Average of three experiments. Each experiment consists of nine data
points (nine different concentrations).
We started our explorations by adding various nucleophiles to
the carbonyl functionality accessing analogues 69-71. Simple
NaBH4 reduction of 60 afforded secondary alcohol 69 as a
mixture of diastereomers, while addition of MeMgBr gave the
corresponding tertiary carbinol mixture 70. Following a pro-
cedure by Olah,61 nucleophilic addition of a trifluoromethyl
group to 60 was accomplished using (trifluoromethyl)tri-
methylsilane (TMSCF3) and catalytic amounts of Bu4NF
(TBAF). This treatment produced the TMS-protected alcohol
intermediate, which was transformed into 71 upon prolonged
exposure to TBAF (compound 71 was isolated as a single
diastereomer after chromatography). Traditional functionalities,
such as in oxime 72, could also be easily incorporated starting
from macroketone 60. As a part of our long-term goal of
elucidating the cellular target of migrastatin and our new
migrastatin scaffolds, we condensed commercially available
Biotin-dPEG4-hydrazide with ketone 60 furnishing the biotin-
labeled acyl-hydrazone 73.
Derivatives 69-73 were evaluated for their ability to inhibit
tumor cell migration in the chamber assay. Interestingly, the
substitution of the ketone functionality for a more polar group,
such as an alcohol or oxime function, seems to be detrimental
for the activity. Secondary alcohol 69, tertiary alcohol 70, and
oxime 72 are rather weak cell migration inhibitors, with IC50
values of 8.9, 3.1, and 2.3 µM, respectively (Table 1). It appears
that incorporation of a trifluoromethyl group can compensate
for the loss of activity caused by the hydroxyl group: Macro-
cyclic CF3-alcohol 71 displays the same activity profile as
macroketone 60. Gratifyingly, inhibitory potency is largely
retained in biotinylated hydrazone 73. Therefore, system 73,
although a mixture of geometric isomers, could qualify as a
probe to assist in the target identification process.
To complete the in vitro mouse assay data set for our
analogues, we examined the effect of migrastatin and our front
line cell migration inhibitors 48, 55, and 60 on 4T1 cell
proliferation. Macrolactone 48, macrolactam 55, and macro-
ketone 60 did not have any cytotoxic or anti-proliferative effects
up to 20 µM, whereas migrastatin turned out to be a weak
proliferation inhibitor (IC50 value of 42 µM). Gratifyingly, this
outcome allows one to conclude that cell proliferation inhibition
is not a contributor to the effects observed in the chamber assays,
rendering our migrastatin analogues specific for cell migration
inhibition.
To translate the murine findings into a human cancer setting,
four solid human tumor cell lines and four liquid cancer cell
lines were investigated with respect to the inhibitory potential
of the most potent compounds macrolactone (48), macroketone
(55), and macrolactam (60). Intriguingly, a very high selectivity
between the various tumor cell lines was detected (Table 4 and
Figure 4). (+)-Migrastatin (1) did not respond well to any of
the cell lines tested. As stated above, nontransformed human
cells as HUVEC appear to be largely more resistant to the
compounds. In contrast, three of our tested cell lines (HT 29,
Ovcar3, and RL) clearly demonstrate a very sensitive response
in terms of migration inhibition, without interfering with cell
proliferation. Surprisingly, the highly metastatic human breast
Table 4. Chamber and Wound-Healing Cell Migration Assay for Various Human Cancer Cellsa
HT 29
HCT 116
IGROV
Ovcar3
RL
RPMI8226
CAG
MDA MB435
48
55
60
13 ( 3
+
60 ( 12
+
76 ( 4
+
20 ( 2
+
38 ( 6
n.d.
75
n.d.
85 ( 12
n.d.
88 ( 5
n.d.
82 ( 7
n.d.
92 ( 7
n.d.
89 ( 12
n.d.
96 ( 5
n.d.
94 ( 6
n.d.
68 ( 7
+
112 ( 6
-
106 ( 3
-
165 ( 18
-
101 ( 3
-
55 ( 10
+
95 ( 9
-
104 ( 2
-
91 ( 5
-
65 ( 13
+
98
-
a Note: The first number represents percent (%) of uninhibited cells as determined by a chamber cell migration assay, and the plus/minus sign refers to
a successful (+) or unsuccessful (-) inhibition of cell migration as determined by a wound-healing assay, n.d. ) not done. Concentrations of compounds
employed: 10X IC50, 48 (240 nM), 55 (1000 nM), 60 (2550 nM). None of the compounds responded to the human breast cancer cell line MCF-7.
9
11336 J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004

Discovery of Potent Cell Migration Inhibitors
A R T I C L E S
Figure 4. Wound-healing cell migration assay results for ovarian cancer cell line Ovcar3. Note: Concentrations of compounds employed: 10X IC50, 48
(240 nM), 55 (1000 nM), 60 (2550 nM).
cancer cell line MDA MB435 responded only slightly to 48,
but not at all to 55 and 60.62 This was an unforeseen discovery,
especially in light of the great response found to these
compounds to the 4T1 mouse breast cancer cell line. These
findings suggest a highly specific effect on selective cancerous
cells, primarily ovarian (Ovcar3) and colon (HT 29) cancers,
in comparison to nontransformed cells, indicating a large
window for possible therapeutic approaches. In pursuing these
data, compound 48 emerges as a particularly appropriate
candidate for preclinical development.
easily accessible in large quantities through efficient synthesis.
They exhibit favorable properties in cell migration, proliferation,
and degradation assays. We are now well positioned to study
their efficacy in mouse and human in vivo tumor metastasis
models to learn more about their pharmacological profiles in
living systems. One of our goals will be the evaluation of these
new agents in combination with long-term, low-dose (metro-
nomic) chemotherapy. We anticipate building upon these
findings in the context of exploring the migrastatin-based estate
in clinical settings.
Summary
Acknowledgment. This work was supported by grants from
the NIH (S.J.D. and X.-Y.H.). Postdoctoral support is grate-
fully acknowledged by C.G. (Deutscher Akademischer Aus-
tauschdienst, DAAD) and J.T.N. (General Motors Cancer
Research Program). We thank Dr. Louis Todaro (Hunter College
New York) for X-ray structure analyses and Dr. George
Sukenick, Ms. Sylvi Rusli, and Ms. Anna Dudkina (NMR Core
Facility, Sloan-Kettering Institute, CA-02848) for mass spectral
analyses. We are grateful to Professor John A. Porco, Jr. (Boston
University) for supplying us with a sample of epoxyquinol A
and to Professor Yueming Li (Sloan-Kettering Institute) for
helpful discussions.
In conclusion, the total synthesis of migrastatin has been
accomplished in a highly convergent way. Key to our successes
in the synthesis was the melding of several key reactions. Thus,
the powers of the LACDAC reaction (see 2 + 3 f 21) to control
functionality and stereochemistry were interfaced with the
Ferrier rearrangement (see 22 f 23) to fully exploit the resultant
pyran matrix. This combination of transformations set the stage
for the RCM reaction to establish the macrolactone.
Moreover, recourse to migrastatin as a lead structure, in the
context of the modality of diverted total synthesis, created a
small, but focused collection of migrastatin analogues, leading
to the discovery of a series of highly potent cell migration
inhibitors. Our most promising candidates 45, 48, 55, 60, and
71 are structurally simplified relative to migrastatin and are
Supporting Information Available: Full experimental details
and physical data for all new compounds. Description for the
biological assays employed. This material is available free of
charge via the Internet at http://pubs.acs.org.
(61) Prakash, G. K. S.; Krishnamurti, R.; Olah, G. A. J. Am. Chem. Soc. 1989,
111, 393-395.
(62) MCF-7, another human breast cancer cell line, did not respond at all.
JA048779Q
9
J. AM. CHEM. SOC. VOL. 126, NO. 36, 2004 11337