A new, ring closing metathesis-based synthesis of (-)-fumagillol

Article abstract of DOI:10.1021/ol016343z

(Equation presented) A new strategy to access the fumagillin/fumagillol skeleton is proposed. An Evans aldolization and a RCM involving an enone are used for the preparation of a key cyclohexanone intermediate, which was readily converted to fumagillol. The synthesis also features an efficient preparation of isogeraniol and isogeranic acid.

Full text of DOI:10.1021/ol016343z

ORGANIC
LETTERS
2001
Vol. 3, No. 17
2737-2740
A New, Ring Closing Metathesis-Based
Synthesis of (−)-Fumagillol
Jean-Guy Boiteau, Pierre Van de Weghe, and Jacques Eustache*
Laboratoire de Chimie Organique et Bioorganique associe´ au CNRS,
UniVersite´ de Haute-Alsace, Ecole Nationale Supe´rieure de Chimie de Mulhouse,
3, Rue Alfred Werner, F-68093 Mulhouse Cedex, France
j.eustache@uha.fr
Received June 25, 2001
ABSTRACT
A new strategy to access the fumagillin/fumagillol skeleton is proposed. An Evans aldolization and a RCM involving an enone are used for
the preparation of a key cyclohexanone intermediate, which was readily converted to fumagillol. The synthesis also features an efficient
preparation of isogeraniol and isogeranic acid.
Inhibition of angiogenesis has emerged as a promising
approach for the treatment of cancer. Among antiangiogenic
agents, fumagillin and related naturally occurring or semi-
synthetic analogues have received close attention and some
have reached the stage of clinical studies.
studied fumagillin analogues are esters of a highly substituted
cyclohexanol called fumagillol. Preliminary structure-activ-
ity relationships (SAR) studies have shown that the nature
of the acyl group is crucial for biological activity, some
semisynthetic analogues being ca. 1000 times more active
than the parent fumagillin.2 The encouraging results obtained
with currently available fumagillin analogues warrant further
SAR studies, which are however hampered by the difficulty
of obtaining fumagillol in appreciable quantity (until now,
fumagillol has been obtained through degradation of the
naturally occurring fumagillin). For this reason and because
fumagillol represents an attractive synthetic challenge, several
approaches to this molecule (and fumagillin) have been
devised. Following the pioneering work of Corey in 1972,3
this area of research remained dormant for several years,
and it is only recently that two enantioselective syntheses
of (-)-fumagillol/(-)-fumagillin,4,5 two syntheses of racemic
Methionine aminopeptidase 2 (Met-AP2), a ubiquitous
enzyme involved in protein post-translational processing, has
been identified as the likely target for fumagillin. The X-ray
structure of MetAP2-fumagillin adducts is known,1 and
irreversible inhibition has been shown to arise from covalent
modification of the enzyme via opening of a reactive epoxide
by an active site histidine (His231).1 The most extensively
(2) Han, C. K.; Ahn, S. K.; Choi, N. S.; Hong, R. K.; Moon, S. K.;
Chun, H. S.; Lee, S. J.; Kim, J. W.; Hong, C. I.; Kim, D.; Yoon, J. H.; No,
K. T. Bioorg. Med. Chem. Lett. 2000, 10, 39-43.
(3) Corey, E. J.; Snier, B. B. J. Am. Chem. Soc. 1972, 94, 2549-2550.
(4) Kim, D.; Ahn, S. K.; Bae, H.; Choi, W. J.; Kim, H. S. Tetrahedron
Lett. 1997, 38, 4437-4440.
(1) Liu, S.; Widom, J.; Kemp, C. W.; Crews, C. M.; Clardy, J. Science
1998, 282, 1324-1327.
10.1021/ol016343z CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/31/2001

fumagillol/fumagillin,6,22 and an approach to a fumagillin and
ovalicin synthetic intermediate7 have been reported. We
disclose in this paper our own efforts in this research area
and describe an alternative approach which may facilitate
the availability of fumagillol.
precursor of isogeranic acid, had only been obtained by rather
complicated or nonspecific methods.10 This led us to develop
the short, stereoselective synthesis shown in Scheme 1.
Treatment of but-3-yn-1-ol with Cp2ZrCl2 and AlMe3 ac-
cording to Negishi,11 followed by Pd0-catalyzed coupling of
the resulting alane with 1-bromo-3-methyl-but-2-ene afforded
a 9:1 mixture of E-isogeraniol (1) and the positional isomer
(1′). E-Isogeraniol and 1′ can neither be distinguished nor
separated by chromatography or distillation, but the latter
could easily be removed by Wacker oxidation of the crude
reaction mixture. As expected, the terminal double bond in
1′ was quantitatively oxidized, while the more hindered
trisubstituted olefine in isogeraniol was unaffected. Isoge-
raniol and the oxidation product 2 could be easily separated
by chromatography on a short pad of SiO2. Oxidation of
E-isogeraniol to the desired E-isogeranic acid proved to be
troublesome. The high sensitivity of the 1,4-dienic system
toward oxidative conditions precluded the use of most
methods commonly used for alcohol to aldehyde and
carboxylic acid conversion. After much experimentation, we
found that treatment with chromium trioxide in acidic
medium12 cleanly led to the desired E-isogeranic acid in 35%
yield. Although the yield is modest, the desired acid was
obtained in the pure E-form and easily separated from
byproducts. Overall, the above short sequence constitutes a
simple, very short stereoselective method for the synthesis
of E-isogeraniol and E-isogeranic acid. With a convenient
access to E-isogeranic acid in hand, we turned our attention
to the crucial Evans aldolization.
Our strategy, based upon the retrosynthetic analysis
outlined in Scheme 1, features two key steps: an Evans aldol
Scheme 1a
a (a) (i) Cp2ZrCl2 (0.4 equiv), AlMe3 (3 equiv), CH2Cl-CH2Cl,
20 °C, 24 h (ii) 1-bromo-3-methyl-but-2-ene, Pd(PPh3)4 (1 mol %),
20 °C, 24 h; (b) CuCl (2 equiv), O2 (1 atm), PdCl2(CH3CN)2 (10
mol %), DMF/water (9:1), 30 °C, 48 h; (c) separation, CrO3, H2SO4,
H2O/acetone, 0 °C, 10 min.
reaction to secure the relative relationship of the C5- and
C6-linked hydroxyls and the side chain attached at C4 (see
Figure 1 for numbering) and a ring closing metathesis (RCM)
to form the six-membered ring.8
In line with our retrosynthetic analysis, Evans aldolization
using the dibutylboron enolate derived from A (S isomer)
(8) Related Evans aldolization/RCM sequences have been reported,
e.g.: (a) Crimmins, M. T.; Tabet, E. A. J. Org. Chem. 2001, 66, 4012-
4108. (b) Crimmins, M. T.; Zuercher, W. J. Org. Lett. 2000, 2, 1065-
1067. (c) Crimmins, M. T.; King, B. W.; Zuercher, W. J.; Choy, A. L. J.
Org. Chem. 2000, 65, 8449-8559.
(9) The preparation of isogeranic acid (E,Z mixture) has been de-
scribed: Hurst, J. J.; Whitham, G. H. J. Chem. Soc. 1960, 2864-2869
(10) (a) Erman, W. F. J. Am. Chem. Soc. 1967, 89, 3828-3841. (b)
Moiseenkov, A. M.; Polunin, E. V.; Semenovsky, A. V. Angew. Chem.,
Int. Ed. Engl. 1981, 20, 1057-1058. (c) Yamamura, Y.; Umeyama, K.;
Maruoka, K.; Yamamoto, H. Tetrahedron Lett. 1982, 23, 1933-1936. (d)
Moiseenkov, A. M.; Polunin, E. V. Bull. Acad. Sci. USSR DiV. Chem. Sci.
(Engl. Transl.) 1983, 32, 1412-1417; IzV. Akad. Nauk SSSR Ser. Khim
1983, 7, 1562-1568. (e) Badet, B.; Julia, M.; Mallet, J. M.; Schmitz, C.
Tetrahedron 1988, 10, 2913-2924.
(11) Ma, S.; Negishi, E. J. Org. Chem. 1997, 62, 784-785.
(12) Neumann, C.; Boland, W. HelV. Chim. Acta 1990, 73, 754-761
(13) A full account of our results will be published in due course.
(14) Abdel-Magid, A.; Pridgen, L. N.; Eggleston, D. S.; Lantos, I. J.
Am. Chem. Soc. 1986, 108, 4595-4602.
(15) Bonner, M. P.; Thornton, E. R. J. Am. Chem. Soc. 1991, 113, 1299-
1308.
(16) (a) Krikstolaityte´, S.; Hammer, K.; Undheim, K. Tetrahedron Lett.
1998, 39, 7595-7598. (b) Paquette, L. A.; Schloss, J. D.; Efremov, I.; Fabris,
F.; Gallov, F.; Mendez-Andino, J.; Yang, J. Org. Lett. 2000, 2, 1259-
1261. (c) Paquette, L. A.; Efremov, I. J. Am. Chem. Soc. 2001, 123, 4492-
4501.
Figure 1. Fumagillol retrosynthesis.
Our first task was to prepare pure E-isogeranic acid. We
were surprised to find that this compound had never been
described in a pure form9 and that isogeraniol, a potential
(17) Fu¨rstner, A.; Langemann, K. Synthesis 1997, 792-803.
(18) Schwab, P.; Grubbs, R. R.; Ziller, J. W. J. Am. Chem. Soc. 1996,
118, 100-110.
(19) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999, 1,
953-956.
(5) Taber, D. F.; Christos, T. E.; Rheingold, A. L.; Guzei, I. A. J. Am.
Chem. Soc. 1999, 121, 5589-5590.
(20) Barrero, A. F.; Alvarez-Manzaneda, E. J.; Chahboun, R.; Meneses,
R. Synlett 1999, 1663-1666.
(6) Vosburg, D. A.; Weiler, S.; Sorensen, E. J. Angew. Chem., Int. Ed.
1999, 38, 971-974.
(21) Amano, S.; Ogawa, M.; Ohtsuka, M.; Childa, N. Tetrahedron 1999,
55, 2205-2224.
(7) Picoul, W.; Urchegui, R.; Haudrechy, A.; Langlois, Y. Tetrahedron
Lett. 1999, 40, 4797-4800.
(22) A similar obervation has recently been made: Hutchings, M.;
Moffat, D.; Simpkins, N. S. Synlett 2001, 63-64.
2738
Org. Lett., Vol. 3, No. 17, 2001

and aldehyde B (R ) PMB, Y ) -CHdCH2) was envis-
aged; a matter of concern was the ability of the â,γ-
unsaturated aldehyde B to survive the aldolization conditions.
In fact, in model studies, we found that, at low temperature
and under a variety of conditions, no coupling could be
obtained.13 When raising the temperature, we found aldehyde
B to be unstable and prone to base-induced reconjugation.
Therefore, we prepared aldehyde 7 (Scheme 2), in which
Scheme 3a
Scheme 2a
a (a) (i) PhSe-SePh, NaBH4, DMF, 90 °C, 2 h (ii) CH2N2, 20
°C; (b) PMBOC(CCl3)dNH, camphorsulfonic acid (0.2 equiv), 20
°C, 20 h; (c) DIBAH, toluene, -78 °C, 20 min.
a (a) (i) (CH3)3C-C(dO)Cl, NEt3, -78 °C, 15 min, then 0 °C,
15 min (ii) (R)-4-benzyl-2-oxazolidinone lithium salt, -78 °C, 20
min; (b) (i) LDA, -78 °C, 30 min (ii) 7, -78 °C, 90 min then
workup (iii) NBu4IO4, CHCl3, reflux, 2 h; (c) (i) N,O-dimethylhy-
droxylamine (3.5 equiv), AlMe3 (2 M in toluene, 3.5 equiv), THF,
20 °C, 30 min, then 9, 0 f 20 °C, 16 h; (d) (CH3)3SiCl, NEt3,
DMAP (1 equiv) THF, 20 °C, 2 h; (e) vinylmagnesium bromide,
THF, 20 °C, 12 h; (f) Ti(OiPr)4, Grubbs’s “type I” catalyst (20
mol %), CH2Cl2, 55 °C, 24 h; (g) RaNi, THF, 0 °C, 20 min.
the â,γ-olefinic double bond is now masked, but here again,
we could not obtain useful yields of aldol. An additional
problem in this case arises from the sensitivity of the
phenylselenyl moiety toward the oxidative conditions used
for the workup.
Cleavage of the chiral auxiliary by N,O-dimethylhydroxy-
lamine in the presence of trimethylaluminum and treatment
of the resulting amide with vinylmagnesium bromide af-
forded enone 12, ready for metathesis. We are aware of only
two examples of RCM involving an olefine and an R,â-
unsaturated ketone.16 Electron-deficient dienic systems (con-
taining R,â-unsaturated esters or amides) are known to be
relatively poorly reactive in RCM reactions and generally
require Lewis acid catalysis.17 In addition, systems containing
an olefine and an R,â-unsaturated ketone would be expected
to be sensitive to Lewis acids. Therefore, the feasibility of
this particular metathesis was unclear. We found that the
only reaction conditions leading to an acceptable (53%) yield
of the desired cyclohexenone 13 required the Grubbs catalyst
type 118 in the presence of Ti(OiPr)4, using dichloromethane
as a solvent. Switching to benzene or using Grubbs type 2
catalyst19 led to no reaction or formation of polymeric
material, respectively.
Faced with the above problems, we decided to modify our
original plans. Although the Evans aldolization is most often
performed using boron enolates, which generally provide
excellent diastereo- and enantioselectivities, other types of
enolates have also been employed. In particular, lithium
enolates have been successfully used in aldolization reactions
with high degrees of selectivity. The reaction thus performed
does not require an oxidative workup. In comparison to their
boron counterpart, lithium enolates also afford syn aldols
predominantly, but the enantioselectivity is opposite,14,15
which required that we started from R (instead of S as
originally planned) benzyloxazolidinone. A further point of
concern was the possible influence of the asymmetric center
in aldehyde 7 on the enantioselectivity of the reaction; this
effect is known to be important in the case of Evans
aldolizations involving R-alkoxyaldehydes, especially when
using lithium enolates.14 Inspection of the literature suggested
that, in our case, high degrees of enantioselectivity could be
expected as the asymmetric induction of the chiral auxiliary
and that of the aldehyde asymmetric center were “matched”.15
The first part of our synthesis, leading to an advanced
intermediate toward fumagillol is shown in Scheme 3.
The N-acyl oxazolidinone 8 (Scheme 3) was deprotonated
at low temperature and treated with aldehyde 7. The raw
aldol thus obtained was subjected to periodate oxidation.
Concomitant spontaneous elimination of selenenic acid
furnished a single product to which the expected structure 9
was tentatively attributed.
Reduction of the R,â-unsaturated ketone in 13 with Raney
nickel20 cleanly afforded the substituted cyclohexanone 14.
Org. Lett., Vol. 3, No. 17, 2001
2739

Sharpless system Ti(OiPr)4/t-BuOOH (and not VO(acac)2/
t-BuOOH)23 and afforded a mixture of the two, nonseparable
epoxides 17 and 17′ in a 1:1 ratio.24 Thus, although
regioselectivity was complete, as expected, no stereoselec-
tivity was observed under a variety of conditions. This was
surprising in view of the good facial selectivity favoring the
desired isomer reported for the same reaction performed on
a close analogue of 16.6 It may be speculated that the
observed different selectivities result from a “fine-tuning”
of the preferred cyclohexane conformation(s) depending on
the nature of the substituants (e.g., OPMB for 16 vs OH in
the reported example). Any difference at this level may
reasonably be expected to influence the conformationnal
distribution of the C4-linked side chain and the accessibility
of one or the other face of the C1′-C2′ olefinic bond.
Scheme 4a
Finally, methylation worked extremely well by using MeI/
NaH to afford 18 and 18′, which could be readily separated
by chromatography. DDQ removal of the PMB protecting
group in 18 proceeded smoothly to afford (-)-fumagillol,
whose data was identical in all respects with that already
reported,5,6 thereby confirming our proposed structural as-
signment for compound 9. In the same way, 18′ was
converted to “epi”-fumagillol.
a (a) (i) Me3S+OI-, NaH, LiI, DMSO/THF, 20 °C, 30 min; (b)
p-toluenesulfonic acid, H2O/THF, 0.5 h; (c) Ti(OiPr)4 (2 equiv),
t-BuOOH, CH2Cl2, -25 °C, 12 h; (d) MeI (30 equiv), NaH, THF/
DMF (1:1), 20 °C, 8 h; (e) DDQ, CH2Cl2/H2O, 20 °C, 90 min.
The final steps of our synthesis are shown in Scheme 4.
In a case related to ours, in which the trimethylsilyl group
in 14 was replaced by a PMB group and the absolute
configuration at C4 was opposite to that in 14, the ylide
derived from trimethylsulfoxonium iodide had been shown
to afford exclusively the epoxide resulting from attack on
the si face of the ketone.21 In our case, the same observation
was made but the desired epoxide was only formed in low
(14%) yield, the main product of the reaction being a
cyclopropanone resulting from base-induced â-elimination
followed by cyclopropanation of the resulting enone.22 The
situation was greatly improved by adding anhydrous LiI to
the reaction mixture. In this case a good (53%) yield of 15
was obtained. After removal of the silyl protecting group,
hydroxyl-directed epoxidation of the C1′-C2′ double bond
was examined. This operation was best conducted using the
Acknowledgment. We thank Drs. Didier Le Nouen and
Ste´phane Bourg for help with NMR recording and interpre-
tation. We thank the CNRS and Galderma R&D for a
fellowship to J-G.B. Financial support of this project by
Galderma R&D is gratefully aknowledged.
Supporting Information Available: Experimental pro-
cedures and 1H and 13C NMR data. This material is available
free of charge via the Internet at http://pubs.acs.org.
OL016343Z
(23) Sharpless, K. B.; Verhoeven, Y. R. Aldrichimica Acta 1979, 12,
63-74.
(24) For analytical purposes 17 and 17′ could be separated by repeated
TLC. However, the method is useless from a preparative viewpoint.
2740
Org. Lett., Vol. 3, No. 17, 2001