Total synthesis of halichondrin C

Article abstract of DOI:10.1021/ja2108307

The first total synthesis of halichondrin C has been completed, highlighted by development of the synthetic method to construct the C8-C14 polycycle. Cr-mediated coupling reactions are used seven times to form a new C-C bond. The acid stability of halicho

Full text of DOI:10.1021/ja2108307

Communication
pubs.acs.org/JACS
Total Synthesis of Halichondrin C
Akihiko Yamamoto, Atsushi Ueda, Paul Bremond, Paolo S. Tiseni, and Yoshito Kishi*
́
Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, United States
S
* Supporting Information
halichondrin C, highlighted by development of the synthetic
ABSTRACT: The first total synthesis of halichondrin C
method for construction of the C8−C14 polycycle and effective
has been completed, highlighted by development of the
synthetic method to construct the C8−C14 polycycle. Cr-
mediated coupling reactions are used seven times to form
a new C−C bond. The acid stability of halichondrin C is
studied, demonstrating that the macrolactone stabilizes the
C8−C14 polycycle, relative to the one present in the C1−
C16 model.
use of Cr-mediated coupling reaction seven times to form a
new C−C bond.
In the halichondrin B series, the C8−C14 polycycle was
constructed via an oxy-Michael reaction, followed by
ketalization (Scheme 1).3 The oxy-Michael addition gave the
Scheme 1. Plan for Construction of the C8−C14 Polycycle
alichondrins are polyether macrolides, originally isolated
H
from the marine sponge Halichondria okadai by Uemura,
Hirata, and co-workers.1 Depending on the oxidation state at
C12 and C13 of the C8−C14 polycycle or the length of the
carbon backbone, halichondrins are subgrouped into the A−C
series or the norhalichondrin−halichondrin−homohalichondrin
series. Except halichondrin A, all the subgroup members are
known (Table 1).2
Table 1. Subgroupings of Halichondrins
desired stereoisomer as the major product, along with the
undesired (minor) stereoisomer, which was recycled. Later, an
ion-exchange resin-based device was developed to transform
both stereoisomers to the desired product in one pot.7 In an
analogy, we hoped that the C8−C14 polycycle of both
halichondrins A and C could be constructed from an ynone
(Scheme 1). Specifically, we expected that an oxy-Michael
addition of the C9 alcohol yields the enone, which should serve
for the construction of the C8−C14 polycycle in both
halichondrins A and C. Related to this plan, we should note
our early work: hydrogenation of an enone similar to the one
proposed took place exclusively from a convex face.8
We first studied the feasibility of this approach in a model
system (Scheme 2). Ynone 4 was uneventfully prepared from
the C1−C11 building block reported previously.9 On treatment
with HF−pyridine in MeCN, 4 gave a ca. 4:1 E- and Z- mixture
of 5a (ca. 40% combined yield) and C8/C9-ketal 6 (ca. 30%
yield). Apparently, 6 was formed via oxy-Michael addition of
the C8 hydroxyl to the enone 5a; indeed, DBU treatment of the
In 1992, we reported the total synthesis of halichondrin B,
norhalichondrin B, and homohalichondrin B.3,4 On completion
of the synthesis, we asked the late Dr. Suffness at the National
Cancer Institute (NCI) and Dr. Littlefield at Eisai Research
Institute (ERI) to test the in vitro and in vivo antitumor
activities of the totally synthetic halichondrins, as well as several
synthetic intermediates. The results were sensational: their
experiments clearly demonstrated that the antitumor activities
of halichondrin B reside in the right portion of the molecule,5
which served as the foundation for successful development of
the antitumor drug Halaven (Eribulin, E7389).6 In this
Communication, we report the first total synthesis of
Received: November 17, 2011
Published: December 20, 2011
© 2011 American Chemical Society
893
dx.doi.org/10.1021/ja2108307 | J. Am. Chem.Soc. 2012, 134, 893−896

Journal of the American Chemical Society
Communication
a
Scheme 2. Model Studies for C8−C14 Polycycle Synthesis
Scheme 3. Three Building Blocks and C1−C13 Building
a
Block Synthesis
a
Cr-mediated coupling was used to form the C−C bond at the sites
a
Reagents and conditions: (a) For 4→6, HF-py, MeCN, rt, then DBU,
rt (90%). For 4→5b, HF-py, py, MeCN, rt (70%). (b) Hf(OTf)4, allyl
alcohol, THF, rt (59%). (c) Pd[P(Ph)3]4, dimedone, CH2Cl2, rt
(quant). (d) DMDO, acetone, rt, followed by CSA, wet CH2Cl2, rt,
then HF-py, THF added (74%).
indicated by a red broken line. Reagents and conditions: (a) i. Dess−
Martin oxidation. ii. cat. asymm Ni/Cr-mediated coupling with Br-C
C-TMS with sulfonamide A with R1 = (S)-i-Pr, R2 = PhCl2-3,5/R3 =
OMe (dr: 23:1; overall yield of 11: 75%). (b) i. TFA, wet CH2Cl2, rt.
ii. TBSOTf, 2,6-lutidine, CH2Cl2, 0 °C→rt. iii. NIS, AgNO3, DMF, rt
(94% overall yield).
crude HF−pyridine product furnished 6 in 90% overall yield
from 4. The spectroscopic data fully supported the proposed
structure 6, which was ultimately confirmed by an X-ray
chemistry at C13. The observed stereochemical outcome of
DMDO oxidation is explained by the preferential convex
approach, whereas that of MCPBA oxidation is a result of the
C11-OH directing effect in the concave environment. These
results were encouraging, yet we still need to develop a method
for selectively oxidizing the enone over the exocyclic olefins at
C19 and C26.18
With these results in hand, we began the synthesis of
halichondrin C and decided to follow the synthesis of
halichondrin B as closely as possible.3 In this way, we can
utilize the previously used C14−C38 and C39−C54 building
blocks (Scheme 3).319 We should note that the newly
developed, convergent method was adopted to synthesize 13;
this method allowed stereoselectively to assemble 13 via
catalytic asymmetric Cr-mediated coupling reaction at the sites
indicated with a broken red line. With use of the catalytic
asymmetric Ni/Cr-mediated reaction,20−22 the remaining C1−
C13 building block was synthesized from the known C1−C11
aldehyde9 with good stereoselectivity.
Catalytic Ni/Cr-mediated reaction was used for coupling 12
with 13 (Scheme 4). Two observations are noteworthy. First, it
was not necessary to use the asymmetric process, as both of the
resultant allylic alcohols were oxidized to the same ynone.
However, we noticed that the coupling was significantly more
effective with the Cr catalysts prepared from the chiral
sulfonamide ligand.23 Second, we noticed that a Ni catalysts
was required for this coupling, although activation of
iodoalkynes without a Ni catalysts was recently reported.24
The ynone thus obtained was converted to the seco-acid 15
and then to the macrolactone 16 with Shiina’s reagent.25,26
Upon treatments with TBAF (buffered with imidazole
hydrochloride) and then DBU, macrolactone 16 was cleanly
converted to C8/C9-ketal 17, which was transformed to the
protected C8−C14 polycycle with Hf(OTf)4 in THF
containing allyl alcohol. This transformation took place in a
way amazingly parallel to the model series. Finally, the
functional groups at C35−C38 were adjusted to give 18 for
the next Ni/Cr-mediated coupling.
analysis of its p-nitrophenylcarbamate.10,11
Because of the facile, second oxy-Michael addition, it was
difficult selectively to obtain the enone with a free hydroxyl
group at C8. For this reason, we became interested in the
possibility of converting the C8/C9-ketal into the C8−C14
polycycle. In particular, we speculated that such a trans-
formation could be achieved with an acid in the presence of an
alcohol. Because of their compatibility with protic solvents, we
tested lanthanide triflates; indeed, Sc(OTf)3, La(OTf)3, and
Y(OTf)3 promoted the desired transformation. Among Lewis
acids tested, Hf(OTf)4 was found most effective for this
transformation.12 Methanol, allyl alcohol, and ethanethiol were
effective in trapping the C12 oxonium ion in the C8−C14
polycyclic ring.13 For the synthesis of halichondrin C, we chose
to use allyl alcohol; on treatment with Pd[P(Ph)3]4,14 the allylic
ether was smoothly removed to give the C8−C14 polycycle in
the halichondrin C series.15 The structure of product was
1
established on comparison with the H and 13C NMR data
reported for natural halichondrin C.1
With the C8 hydroxyl group blocked, we sought a selective
formation of the enone 5b from 4 and eventually found that
TBS deprotection with HF−pyridine in MeCN containing
excess pyridine left the C8 axial TBS group intact, to furnish an
11:1 E- and Z-mixture of enone 5b (70% combined yield). E-
and Z-enones were chromatographically separable but gradually
isomerized to a mixture on standing. We hoped that a Michael
addition would allow us to selectively functionalize the enone,
to construct the C8−C14 polycycle in the halichondrin A
series. Unfortunately, we found that the enone exhibited a poor
reactivity against common nucleophiles, including H2O2, under
basic conditions.16 In contrast, oxidation of chromatographi-
cally separated E-enone with DMDO and MCPBA, followed by
acid treatment, gave the desired product as a dr = >30:1
(DMDO) and 1:1 (MCPBA) mixture of C13 diaster-
eomers.10,17 On comparison with the H and 13C NMR data
1
reported for natural homohalichondrin A,1 the major product
obtained with DMDO was assigned as the desired stereo-
894
dx.doi.org/10.1021/ja2108307 | J. Am. Chem.Soc. 2012, 134, 893−896

Journal of the American Chemical Society
Communication
a
Following the protocol used for the synthesis of halichondrin
B,3 we subjected enone 19 to TBAF (buffered with 0.5 equiv of
imidazole hydrochloride), DDQ, and PPTS treatment. This
transformation involved (1) deprotection of the five TBS
groups, (2) hemiketal formation at C44, (3) oxy-Michael
addition of the resultant hemiketal hydroxyl group to the α,β-
unsaturated ketone to form the [6,6]-spiroketal at C44, (4)
deprotection of the C41-MPM group, and (5) formation of
[5,5]-spiroketal at C38. To realize this transformation
selectively, it was necessary to keep the C41-hydroxyl group
protected at the TBAF step to avoid [5,6]-spiroketal formation
at C44.
Scheme 4. Synthesis of the Right Half of Halichondrin C
The three-step transformation introduced three stereogenic
centers at C38, C40, and C44. Based on the following
considerations, we anticipated that it should give the desired
stereoisomer stereoselectively. In the antibiotic (−)-A23187
(calcimycin) synthesis,28 we constructed the [6,6]-spiroketal
system via a route closely related to the present case. This
precedent suggested that the first two steps should give the
desired stereoisomer at C40 and C44 under the kinetically and/
or thermodynamically controlled conditions; indeed, TLC and
1H NMR analyses indicated that it was highly stereoselective.29
In addition, to ensure that the oxy-Michael reaction gave the
desired equatorial adduct at C40, we tested a treatment of the
crude product at the TBAF step with DBU and found that the
DBU treatment did not noticeably affect the overall stereo-
selectivity. Last, the [5,5]-ketalization should yield the desired
stereoisomer under the acidic (thermodynamically controlled)
conditions.27
a
Reagents and conditions: (a) i. cat. asymm Ni/Cr-mediated coupling
with sulfonamide A with R1 = (S)-i-Pr, R2 = Me/R3 = OMe (85%). ii.
Dess−Martin oxidation, rt (94%). iii. DDQ, t-BuOH, CH2Cl2,
phosphate buffer, 0 °C→rt (80%). iv. aq LiOH, THF, rt (87%). (b)
(2,6-Me,NO2PhCO)2O, DMAP, tol, 70 °C (high dilution) (93%). (c)
TBAF-imidazole·HCl (0.5 equiv), THF, rt, then DBU, CH2Cl2, rt
(68%). (d) i. Hf(OTf)4, allyl alcohol, THF, rt (74%). ii. p-
O2NPhCOCl, py, CH2Cl2, rt (87%). iii. TBSOTf, 2,6-lutidine,
CH2Cl2, 0 °C (88%). iv. K2CO3, MeOH−CH2Cl2, rt (93%). v.
Dess−Martin oxidation, rt (92%).
a
Scheme 5. Completion of Synthesis
Preparative TLC purification furnished the C12 allyl-
protected halichondrin C, cf. R = CH2CHCH2 of 18, in
ca. 30% overall yield.30 A small amount of several byproducts
was isolated, but it was not clear whether any of them was a
stereoisomer of halichondrin C.28
The final step of synthesis was to remove the allyl group at
C12, which was accomplished with Pd[P(Ph)3]4,14 to furnish
synthetic halichondrin C (20). The synthetic material was fully
characterized and confirmed to be identical with natural
1
halichondrin C on comparison of H and 13C NMR spectra.
With synthetic halichondrin C (20) and C1−C16 model
compound 7b in hand, we had the opportunity to study their
acid stability, thereby demonstrating a remarkable difference.
Halichondrin C was found stable in CD2Cl2 containing CSA
(0.1 equiv) at room temperature overnight, whereas the model
compound was labile, yielding a ca. 1:1 mixture of two
products.31 Apparently, the macrolactone ring stabilizes the
C8−C14 polycycle of halichondrin C, which is intriguing in
two respects. First, the macrolactone did not affect for
formation of the C8/C9-ketal and its transformation to the
C8−C14 polycycle, yet stabilized the C8−C14 polycycle
relative to the open-form model. Second, the observed stability
of halichondrin C suggests the possibility that, once the
macrolactone is introduced, the C8−C14 polycycle could be
constructed from an open-form precursor under acidic
conditions.
In summary, we completed the first total synthesis of
halichondrin C, highlighted by development of the synthetic
method to construct the C8−C14 polycycle. Cr-mediated
couplings were used seven times\ to form a new C−C bond.
The acid stability of halichondrin C was studied, thereby
demonstrating that the macrolactone stabilizes the C8−C14
polycycle relative to the one present in the C1−C16 model.
a
Reagents and conditions: (a) i. stoich asymm Ni/Cr-mediated
coupling with sulfonamide A with R1 = (S)-i-Pr, R2 = Me/R3 = OMe.
ii. Dess−Martin oxidation, CH2Cl2, rt (75% and 79% overall yield with
a 1:1 and 1:2 mixture of 18 and 14). (b) TBAF-imidazole·HCl (0.5
equiv), AcOMe, DMF, rt. (c) i. DDQ, t-BuOH, CH2Cl2, phosphate
buffer, 0 °C→rt. ii. PPTS, CH2Cl2, rt (30% overall yield from 19). (d)
Pd[P(Ph)3]4, dimedone, CH2Cl2, rt (56%).
Once again, Ni/Cr-mediated reaction was used to couple 14
and 18 (Scheme 5). Because of the scale of this experiment, we
opted to use a stoichiometric protocol and found that the
sulfonamide ligand with R1 = (S)-i-Pr, R2 = Me, and R3 = OMe
greatly improved the coupling efficiency.27 It is worth
mentioning that this coupling gave a satisfactory result, even
with a 1:1 molar ratio of coupling partners (75 and 79% yields
with 1:1 and 1:2 ratios of 18 and 14, respectively). The
coupling product was then oxidized to enone 19.
895
dx.doi.org/10.1021/ja2108307 | J. Am. Chem.Soc. 2012, 134, 893−896

Journal of the American Chemical Society
Communication
(16) Attempted nucleophiles included MeOH, EtSH, MeCO2K, and
p-MeOPhSH.
(17) DMDO oxidation of Z-enone 5b gave the C13 diastereomer of
ASSOCIATED CONTENT
* Supporting Information
Experimental procedures, characterization data, spectral data,
crystallographic data (CIF), and complete ref 6a. This material
is available free of charge via the Internet at http://pubs.acs.org.
■
S
9 selectively.
(18) DMDO and MCPBA oxidation of a 1:1 mixture of 5b and
C14−C23 building block 11 (cited in J. Am. Chem. Soc. 2009, 131,
15636 ) revealed virtually no reactivity difference between the enone
and the C19 exocyclic olefin.
(19) Dong, C.-G.; Henderson, J. A.; Kaburagi, Y.; Sasaki, T.; Kim, D.-
S.; Kim, J. T.; Urabe, D.; Guo, H.; Kishi, Y. J. Am. Chem. Soc. 2009,
131, 15642and references cited therein.
AUTHOR INFORMATION
Corresponding Author
kishi@chemistry.harvard.edu
■
(20) (a) Jin, H.; Uenishi, J.-I.; Christ, W. J.; Kishi, Y. J. Am. Chem. Soc.
1986, 108, 5644. (b) Guo, H.; Dong, C.-G.; Kim, D.-S.; Urabe, D.;
Wang, J.; Kim, J. T.; Liu, X.; Sasaki, T.; Kishi, Y. J. Am. Chem. Soc.
2009, 131, 15387. (c) Liu, X.; Henderson, J. A.; Sasaki, T.; Kishi, Y. J.
Am. Chem. Soc. 2009, 131, 16678and referencescited therein.
(21) Takai, K.; Tagashira, M.; Kuroda, T.; Oshima, K.; Utimoto, K.;
Nozaki, H. J. Am. Chem. Soc. 1986, 108, 6048.
ACKNOWLEDGMENTS
Financial support from the Eisai USA Foundation is gratefully
acknowledged. We thank Prof. Uemura for a copy of the H
NMR spectrum of natural halichondrin C.
■
1
REFERENCES
■
(22) For recent reviews on Cr-mediated carbon−carbon bond-
forming reactions, see reference 3 in ref 20c.
(1) (a) Uemura, D.; Takahashi, K.; Yamamoto, T.; Katayama, C.;
Tanaka, J.; Okumura, Y.; Hirata, Y. J. Am. Chem. Soc. 1985, 107, 4796.
(b) Hirata, Y.; Uemura, D. Pure Appl. Chem. 1986, 58, 701.
(2) For isolation of the halichondrins from different species of
sponges, see: (a) Pettit, G. R.; Tan, R.; Gao, F.; Williams, M. D.;
Doubek, D. L.; Boyd, M. R.; Schmidt, J. M.; Chapuis, J.-C.; Hamel, E.;
Bai, R.; Hooper, J. N. A.; Tackett, L. P. J. Org. Chem. 1993, 58,
2538and references cited therein. (b) Hickford, S. J. H.; Blunt, J. W.;
Munro, M. H. G. Bioorg. Med. Chem. 2009, 17, 2199and references
cited therein.
(23) There was no match/mismatch behavior observed between (S)-
and (R)-sulfonamides for this coupling.
(24) Usanov, D. L.; Yamamoto, H. J. Am. Chem. Soc. 2011, 133, 1286.
(25) Shiina, I.; Katoh, T.; Nagai, S.; Hashizume, M. Chem. Rec. 2009,
9, 305.
(26) Macrolactone 16 was also obtained via esterification between
C1-CO2H and C30-OH, followed by macrocyclization of C13-
iodoacetylene and C14-aldehyde. However, the route reported in
the text gave a significantly better overall efficiency.
(3) Aicher, T. D.; Buszek, K. R.; Fang, F. G.; Forsyth, C. J.; Jung, S.
H.; Kishi, Y.; Matelich, M. C.; Scola, P. M.; Spero, D. M.; Yoon, S. K. J.
Am. Chem. Soc. 1992, 114, 3162and references citedtherein.
(4) For synthetic work by Phillips, Salomon, Burke, and Yonemitsu,
see: (a) Henderson, J. A.; Jackson, K. L.; Phillips, A. J. Org. Lett. 2007,
9, 5299. Jackson, K. L.; Henderson, J. A.; Motoyoshi, H.; Phillips, A. J.
Angew. Chem., Int. Ed. 2009, 48, 2346andreferences cited therein.
(b) Cooper, A. J.; Pan, W.; Salomon, R. G. Tetrahedron Lett. 1993, 34,
8193and references cited therein. (c) Lambert, W. T.; Hanson, G. H.;
Benayoud, F.; Burke, S. D. J. Org. Chem. 2005, 70, 9382andreferences
cited therein. (d) Horita, K.; Hachiya, S.; Nagasawa, M.; Hikota, M.;
Yonemitsu, O. Synlett 1994, 38.
(27) For this coupling, we noticed a match/mismatch behavior, i.e.,
the coupling with the antipode of the sulfonamide ligand was
significantly less satisfactory in terms of yield and coupling rate.
(28) Negri, D. P.; Kishi, Y. Tetrahedron Lett. 1987, 28, 1063.
(29) CSA and PPTS treatments of byproduct did not yield the
desired product (1H NMR).
(30) The overall yields of three separate experiments (8.5, 10.0, and
28 mg scales) were 32, 31, and 29%, respectively.
1
(31) Based on the MS and H NMR spectra, one of the products
appeared to be the C12/C14 β-diketone.
(5) Kishi, Y.; Fang, F. G.; Forsyth, C. J.; Scola, P. M.; Yoon, S. K. U.S.
Patent 5338866, International Patent WO93/17650.
(6) (a) Zheng, W.; et al. Bioorg. Med. Chem. Lett. 2004, 14, 5551.
(b) Yu, M. J.; Kishi, Y.; Littlefield, B. A. In Anticancer Agents from
Natural Products; Cragg, G. M., Kingston, D. G. I., Newman, D. J.,
Eds.; CRC Press: Boca Raton, FL, 2005; p 41.
(7) (a) Namba, K.; Jun, H.-S.; Kishi, Y. J. Am. Chem. Soc. 2004, 126,
7770. (b) Kaburagi, Y.; Kishi, Y. Org. Lett. 2007, 9, 723.
(8) Aicher, T. D.; Kishi, Y. Tetrahedron Lett. 1987, 28, 3463.
(9) (a) Duan, J. J.-W.; Kishi, Y. Tetrahedron Lett. 1993, 34, 7541.
(b) Stamos, D. P.; Kishi, Y. Tetrahedron Lett. 1996, 37, 8643. (c) Choi,
H.; Demeke, D.; Kang, F.-A.; Kishi, Y.; Nakajima, K.; Nowak, P.; Wan,
Z.-K.; Xie, C. Pure Appl. Chem. 2003, 75, 1.
(10) For details, see Supporting Information.
(11) The following abbreviations are used: CSA, camphor-10-
sulfonic acid; DBU, 1,8-diazabicyclo[5,4,0]undec-7-ene; DDQ, 2,3-
dichloro-5,6-dicyano-1,4-benzoquinone; DMAP, 4-dimethylaminopyr-
idine; DMDO, dimethyldioxirane; MPM, p-methoxyphenymethyl;
PPTS, pyridinium p-toluenesulfonate; TBAF, tetra-n-butylammonium
fluoride; TBS, tert-butyldimethylsilyl; TMS, trimethylsilyl.
(12) For a recent review,see: Kobayashi, S.; Sugiura, M.; Kitagawa,
H.; Lam, W. W.-L. Chem. Rev. 2002, 102, 2227.
(13) H2O, MeCO2Na, ClCH2CH2OH, and TMSCH2CH2OH were
not effective.
(14) For a recent review, see: Lipton, M. In Handbook of
Organopalladium Chemistry for Organic Synthesis; Negishi, E., Ed.;
John Wiley & Sons: New York, 2002; Vol. 2, p 1901.
(15) 7b was also obtained from 5b in four steps: (1) NBS, (2) (n-
Bu)3SnH, AIBN, (3) TBAF, and (4) PPTS, in 15% overall yield.
896
dx.doi.org/10.1021/ja2108307 | J. Am. Chem.Soc. 2012, 134, 893−896