Stereospecific formal total synthesis of ecteinascidin 743

Article abstract of DOI:10.1002/anie.200503983

(Chemical Equation Presented) A new strategy was designed for the construction of the pentacyclic ring system of ecteinascidin 743. Key features include highly concise routes to the enantiopure, configurationally matched subunit 1, a novel vinylogous Pict

Full text of DOI:10.1002/anie.200503983

Communications
Natural Products
DOI: 10.1002/anie.200503983
Stereospecific Formal Total Synthesis of
Ecteinascidin 743**
Shengping Zheng, Collin Chan, Takeshi Furuuchi,
Benjamin J. D. Wright, Bishan Zhou, Jinsong Guo, and
Samuel J. Danishefsky*
Dedicated to Professor Ryoji Noyori
In the preceding communication in this issue, we described a
comprehensive study directed at understanding the multi-
faceted stereochemical issues associated with the cascadelike
reaction initiated by cleavage of the urethane function of an
N-methyl-N-Boc ketoaldehyde of the type 3.[1] Even after
extensive experimentation, we could not realize a lynchpin
Mannich cyclization of the type 3!2s (s = syn). Such a
cyclization, had it occurred, would have found ready appli-
cation in the total synthesis of the highly cytotoxic natural
product ecteinascidin 743 (1; Scheme 1).[2] In practice, it was
shown with C1 and C13 matched in their correct relative and
absolute configurations as shown in 3 that the product is
cleanly the 3,11-anti compound 2a (a = anti). At no time could
we pass from a 2a to a 2s compound by epimerization at C3.[3]
To attain the relative and absolute stereochemistry at C3,
C11, and C13 required to reach 1, it would be necessary to
resort to mismatching at C1 (i.e. (S)-C1 rather than (R)-C1
configuration). This would in turn necessitate a late-stage and
problematic epimerization at C1.[4] In summary, at no stage
could we solve the ecteinascidin 743 problem (encompassing
C1, C3, C4, C11, and C13) or the saframycin total synthesis
problem (encompassing C1, C3, C11, and C13) by direct
cyclization of appropriately matched ((R)-C1, (S)-C13)
precursors bearing ketone and aldehyde functions at C4 and
C11, respectively (cf. 3).[5]
[*] Dr. S. Zheng, C. Chan, Dr. T. Furuuchi, B. J. D. Wright, Dr. B. Zhou,
Dr. J. Guo, Prof. S. J. Danishefsky
The Department of Chemistry
Columbia University
Havemeyer Hall, New York, NY 10027 (USA)
Fax: (+1)212-772-8691
E-mail: s-danishefsky@ski.mskcc.org
Prof. S. J. Danishefsky
The Laboratory for Bioorganic Chemistry
Sloan-Kettering Institute for Cancer Research
1275 York Avenue, New York, NY 10021 (USA)
[**] This work was supported by the National Institutes of Health (Grant
HL25848) and by Pharmamar Corporation of Madrid, Spain. We
thank Dr. Atsushi Endo for many helpful dicussions. B.J.D.W.
gratefully acknowledges the NSF for a predoctoral fellowship. We
thank Bristol Myers Squibb for the generous fellowship awarded to
C.C. We also thank Dr. Yasuhiro Itagaki (Columbia University) for
mass spectrometric analyses
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1754
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759

Angewandte
Chemie
(Scheme 3). Given the importance of providing access to a
broad range of biologically active tetrahydroisoquinoline
alkaloids, it seemed prudent to gain reliable and convenient
access to the required matched antipodes.[8] Indeed, as will be
Scheme 1. Lynchpin Mannich strategy en route to ecteinascidin 743
(1). PG=protecting group.
Previously, in connection with our total synthesis of
cribrostatin IV, we charted a viable pathway from a 2a-type
compound to a 3,4-dehydro-type compound (cf. 4).[6] Of
course, in cribrostatin IV, the final target we were pursuing
contains a 3,4-olefinic linkage. Accordingly, we envisioned an
indirect solution to the ecteinascidin 743/saframycin total
synthesis problem by reaching a 2a species that would be
converted into a 3,4-dehydro system (cf. 4).[7] We hoped to
reach a key intermediate (cf. 5) required for the ecteinasci-
din 743 series from 4 through an overall a-face hydration (i.e.
Scheme 3. Construction of the pentacycle through a novel vinylogous Pictet–
Spengler cyclization.
described, a new concise plan was devised to reach the
configurationally defined tetrahydroisoquinoline moiety
((R)-C1). The latter was acylated with the enantiomerically
defined l-aminoacyl CDE precursor, bearing the future C11
aldehyde (see compound 6).[9]
Our synthesis started with a highly regiospecific bromi-
nation of the Borchardt catechol 10 (Scheme 4).[10] Engage-
ment of the catechol hydroxyl groups (by methylenation) set
the stage for successful Baeyer–Villiger oxidation and
hydrolysis. The resultant phenol was protected as its tert-
butyldiphenylsilyl ether 11. Lithiation of the bromine func-
tion followed by coupling of the organometallic agent with
the Weinreb amide of benzyloxyglycolic acid yielded 12.[11]
Asymmetry was introduced by submitting ketone 12 to
Noyori transfer-hydrogenation conditions, thereby providing
homochiral 13.[12] A modified Mitsunobu reaction of alcohol
13 with diphenylphosphorylazide gave rise to 14, with
complete inversion of stereochemistry.[13] At this stage, the
R configuration at C1 was securely fashioned. Catalytic
hydrogenation of 14 to the amine was followed by two-
carbon-atom homologation, and subsequent deprotection/
reprotection of the phenol as its allyl ether gave acetal 15. The
Bobbitt-modified Pomerantz–Fritsch reaction of 15 provided
the required tetrahydroisoquinoline 16, in which the stereo-
chemistry at C1 was correctly defined.[14] That C4 is, at this
stage, presented as a mixture of epimers, is an awkwardness
rather than an impediment (see below).
ꢀ
3-Ha, 4-OHa) of the C3 C4 double bond (see 5, Scheme 2).
On further reflection, a conceptually simpler possibility
presented itself. Perhaps direct cyclization of an ortho-
hydroxystyrene prototype 7 would occur at the styrene
double bond, with its regiochemical sense controlled by the
hydroxy group of the phenol moiety. The cyclization we
envisioned (presumably via the iminium derivative 8, formed
upon cleavage of the Boc group from N12) would produce 9
Coupling of tetrahydroisoquinoline 16 and the previously
described l-tyrosine derivative 17[6] under the agency of
BOPCl afforded amide 6 in excellent yield (Scheme 5). After
oxidative cleavage of the PMB group and dehydration of the
benzylic alcohol,[15] the primary alcohol was oxidized to the
C11 aldehyde.[16] The allyl group was cleaved by treatment
with tributyltin hydride in the presence of catalytic
Scheme 2. Proposed epimerization strategy at C3.
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1755

Communications
screened various acidic conditions for achieving the desired
cyclization to 9. In the event, compound 7 successfully
underwent cyclization to produce pentacycle 9 upon expo-
sure to 30 equivalents of difluoroacetic acid in benzene. For
the moment, the product is obtained in a somewhat
disappointing yield of 42–58%.
With a viable route to pentacyclic alkene 9 in hand, we
ꢀ
were now faced with the need to functionalize the C3 C4
double bond (Scheme 6). Implementation of this task
Scheme 4. a) 1) Br2 (1.01 equiv), NaOAc (1.05 equiv), AcOH, 08C!258C, 12 h;
2) BrCH2Cl (1.5 equiv), Cs2CO3 (1.5 equiv), DMF, 1058C, 12 h, 66%; b) 1) MCPBA
(3 equiv), 258C!648C, 3 h; 2) HCl (0.3 equiv), 08C!258C, 12 h, 78%;
c) TBDPSCl (1.15 equiv), TEA (1.5 equiv), DMAP (0.1 equiv), 258C, 12 h, 89%;
d) 1) nBuLi (1.6m; 1.1 equiv), toluene/THF (9:1), ꢀ788C, 20 min; 2) neat Me-
(MeO)NC(O)CH2OBn (1.5 equiv), ꢀ788C, 50 min, 80%; e) Noyori R,R catalyst
(0.01 equiv), HCO2H (13.2 equiv), TEA (7.8 equiv), DMF, 08C!408C, 24 h, 78%
(95% ee); f) DPPA (2 equiv), DBU (2 equiv), toluene/DMF (9:1), 508C, 24 h, 89%
(95% ee); g) H2 (1 atm), Pd/C (5%), EtOAc, 258C, 15 h, 80%;
h) 1) (MeO)2CHCHO (1 equiv), AcOH (4 equiv), NaCNBH3 (1.3 equiv), MgSO4
(6 equiv), MeOH, 08C!658C, 4 h; 2) TBAF (1.5 equiv), THF, 08C, 30 min, 94%;
i) allyl bromide (1.15 equiv), NaH (2.8 equiv), DMF, 08C!258C, 2 h, 84%; j) 7m
HCl (70 equiv), dioxane, 08C!258C, 72 h, 90%. DMF=N,N-dimethylformamide;
MCPBA=m-chloroperoxybenzoic acid; TBDPS=tert-butyldiphenylsilyl; TEA=tri-
ethylamine; DMAP=4-(dimethylamino)pyridine; Bn=benzyl; DPPA=diphenyl-
phosphoryl azide; DBU=1,8-diazabicyclo[5.4.0]undec-7-ene; TBAF=tetrabutyl-
ammonium fluoride.
ꢀ
Scheme 6. Unsuccessful attempts to refunctionalize the C3 C4
olefinic linkage. NBS=N-bromosuccinimide.
commenced with temporary protection of phenol 9 as its
methyl ether derivative 18. As 18 could be readily obtained in
ample quantities, it was used as a model to probe eventual
management of the synthesis in a fully updated system. In our
ꢀ
first attempts to functionalize the C3 C4 double bond, we
took recourse to hydroboration reagents such as BTHF, BMS,
BH3·Py, Br2BH, and catecholborane. Remarkably, such
attempts uniformly resulted only in recovery of 18.[18] By
contrast, the double bond of 18 could be dihydroxylated with
N-bromosuccinimide in aqueous THF, leading to a diol
assigned as 19 in which the dihydroxylation was presumed
to occur from the less-hindered a face.[19] However,
attempted reductive removal of the now extraneous angular
3-OH group of 19 with a variety of reducing agents resulted in
the recovery of starting material.
Given these results, we concluded that to be successful, a
highly reactive oxidizing agent would be needed to attack the
double bond. We also needed functionality at C3, which
would be more reactive than a simple hydroxy group to allow
introduction of the required 3-Ha group. Accordingly, we first
explored the use of peracid agents for the epoxidation of the
Scheme 5. a) BOPCl (1.1 equiv), TEA (3 equiv), CH2Cl2, 08C!258C,
24 h, 85%; b) DDQ (1.6 equiv), CH2Cl2/pH 7.00 buffer solution (18:1),
258C, 30 min, 90%; c) Cu(OTf)2 (0.2 equiv), benzene, 858C, 15 min,
61%; d) DMP (1.5 equiv), CH2Cl2, 258C, 5 h, 94%; e) [(PPh3)2PdCl2]
(0.2 equiv), Bu3SnH (1.2 equiv), AcOH (5 equiv), 08C!258C, 1 h,
93%; f) CHF2CO2H (30 equiv), MgSO4 (4 equiv), benzene, 1008C,
45 min, 42–58%. BOPCl=bis(2-oxo-3-oxazolidinyl)phosphinic
chloride; DDQ=2,3-dichloro-5,6-dicyano-1,4-benzoquinone;
DMP=Dess–Martin periodinane.
ꢀ
C3 C4 olefin linkage. However, at this stage N-oxide
ꢀ
formation of the N Me tertiary amine group became a
potential problem. Accordingly, we conducted a McCluskey
reaction to afford N-Troc urethane 20.[20] Unfortunately,
attempts at various epoxidizing conditions with peracids were
unsuccessful (possibly owing to the slowness of the desired
reaction in the context of the highly oxidizable aromatic
sectors). With our options rapidly narrowing, it was fortunate
[(Ph3P)2PdCl2] to yield cyclization precursor 7, as hypothe-
sized above.[17] The resulting aldehyde 7 was suited for the
crucial intramolecular Pictet–Spengler cyclization. We
1756
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759

Angewandte
Chemie
that reaction of 20 with 2,2-dimethyldioxirane (DMDO) did
ꢀ
occur, giving rise to material formulated as the C3 C4 a-
epoxide 21 (Scheme 7).[2d] Treatment of 21 with sodium
cyanoborohydride provided the required 3-Ha, 4-OHa hydra-
Scheme 7. a) DMDO (1.5 equiv), CH2Cl2, 08C!258C, 30 min;
b) NaCNBH3 (5 equiv), 10 min, 75%. DMDO=dimethyl dioxirane.
Scheme 8. a) TBSOTf (2.5 equiv), TEA (3 equiv), CH2Cl2, 08C, 30 min,
tion product 22. Greater insight into this fascinating reductive
cleavage of the novel enamide epoxide (cf. 21) was gained in
the MOM ether series (see below, compound 25).
100%; b) TrocCl (20 equiv), TBAI (3 equiv), toluene, 1108C, 3 h, 92%;
c) 1) TBAF (3 equiv), CH2Cl2, 08C, 2 min; 2) MOMCl (3 equiv), Hünig
base (5 equiv), 30 min, 79%; d) DMDO (2 equiv), CH2Cl2, 08C!258C,
1.5 h; e) NaCNBH3 (5 equiv), 10 min, 27/26 60%:10%; f) NaCNBH3
(50 equiv), 10 min, (26) 78%. TB S=tert-butyldimethylsilyl;
On the basis of our experiences in this series, we could not
be confident that a compound such as 22, containing a methyl
ether at C5 could be advanced to conclude the synthesis.
Moreover, in the same vein, it was decided to confirm our
various assignments by connecting our series with a pre-
viously synthesized intermediate, from which ecteinasci-
din 743 could definitely be reached. Accordingly, we defined
32 as our milestone target, which is an advanced intermediate
in the Fukuyama total synthesis of ecteinascidin 743.[2d]
Phenol 9 was first converted into its tert-butyldimethylsilyl
ether derivative (Scheme 8). This step was followed by a
Tf =trifluoromethanesulfonyl; Troc=2,2,2-trichloroethoxycarbonyl;
TBAI=tetrabutylammonium iodide; MOM=methoxymethyl.
ꢀ
McCluskey reaction of the N Me amine, thereby providing
23. Deprotection of the TBS ether with TBAF, followed by
addition of MOMCl and Hünig base, led to 24 in good
yield.[21] Treatment of this compound with DMDO led, as
ꢀ
before (cf. 20), to epoxidation of the C3 C4 double bond, thus
affording the presumed 25. When this compound was treated
with 5 equivalents of sodium cyanoborohydride the major
product obtained was a ketone assigned as shown in 27,
although small amounts of 26 were also obtained. Two
mechanistic hypotheses were entertained to account for the
formation of the ketone (Scheme 9). One could envision a
concerted rearrangement with hydride migration, from C4 to
C3, to afford ketone 27. Alternatively, the nitrogen atom of
the lactam opens the epoxide to produce amidonium alkoxide
28. This intermediate can undergo 1,2-hydride migration to
give 27 or competitive reduction by an external hydride to
provide 26. Apparently this duality is, in fact, operative, since
recourse to a large excess of sodium cyanoborohydride does
afford 26 as the predominant product.[22]
Scheme 9. Mechanistic proposals for the reduction of epoxide 25 to
alcohol 26 and ketone 27.
At this juncture, we faced only the installation of the
cyano function at C21 (Scheme 10). Toward this end, the two
benzyl protecting groups were removed, thereby giving rise to
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1757

Communications
din 743 which are enabled by the formal total synthesis
described above.
Received: November 9, 2005
Keywords: antitumor agents · asymmetric synthesis ·
.
natural products · Pictet–Spengler reaction · total synthesis
[1] See the previous Communication in this issue: C. Chan, S.
Zheng, B. Zhou, J. Guo, R. M. Heid, S. J. Danishefsky, Angew.
Chem. 2006, 118, 1781; Angew. Chem. Int. Ed. 2006, 45, 1749.
[2] Isolation of 1: a) K. L. Rinehart, T. G. Holt, N. L. Fregeau, J. G.
Stroh, P. A. Keifer, F. Sun, L. H. Li, D. G. Martin, J. Org. Chem.
1990, 55, 4512; b) R. Sakai, K. L. Rinehart, Y. Guan, A. H.-J.
Wang, Proc. Natl. Acad. Sci. USA 1992, 89, 11456; total
syntheses of 1: c) E. J. Corey, D. Y. Gin, R. S. Kania, J. Am.
Chem. Soc. 1996, 118, 9202; d) A. Endo, A. Yanagisawa, M. Abe,
S. Tohma, T. Kan, T. Fukuyama, J. Am. Chem. Soc. 2002, 124,
6552; e) J. Chen, X. Chen, M. Bois-Choussy, J. Zhu, J. Am.
Chem. Soc. 2006, 128, 87; semisynthesis of 1: f) C. Cuevas, M.
Pꢀrez, M. J. Martꢁn, J. L. Chicharro, C. Fernꢂndez-Rivas, M.
Flores, A. Francesch, P. Gallego, M. Zarzuelo, F. de la Calle, J.
Garcꢁa, C. Polanco, I. Rodrꢁguez, I. Manzanares, Org. Lett. 2000,
2, 2545.
[3] For an example of base-induced epimerization at C3, see: J. W.
Lane, Y. Chen, R. M. Williams, J. Am. Chem. Soc. 2005, 127,
12684.
[4] W. Jin, S. Metobo, R. M. Williams, Org. Lett. 2003, 5, 2095.
[5] a) B. Zhou, J. Guo, S. J. Danishefsky, Org. Lett. 2002, 4, 43; b) B.
Zhou, S. Edmondson, J. Padron, S. J. Danishefsky, Tetrahedron
Lett. 2000, 41, 2039; c) B. Zhou, J. Guo, S. J. Danishefsky,
Tetrahedron Lett. 2000, 41, 2043.
Scheme 10. a) H2 (1 atm), Pd/C (10%), EtOAc, 258C, 3 h, 77%;
b) DIBAL-H/BuLi (1:1; 40 equiv), THF, 08C, 5 h, 78%; c) allyl bromide
(20 equiv), Hünig base (25 equiv), CH2Cl2, 508C, 16 h, 66%; d) KCN
(9 equiv), AcOH, 258C, 4 h, 79%; e) TFA (1 equiv), CH2Cl2, ꢀ208C,
30 min, 54%. DIBAL-H=diisobutylaluminum hydride; TFA=trifluoro-
acetic acid.
[6] C. Chan, R. Heid, S. Zheng, J. Guo, B. Zhou, T. Furuuchi, S. J.
Danishefsky, J. Am. Chem. Soc. 2005, 127, 4596.
[7] X. Chen, J. Chen, M. De Paolis, J. Zhu, J. Org. Chem. 2005, 70,
4397.
[8] For an excellent review of the tetrahydroisoquinoline alkaloids,
see: J. D. Scott, R. M. Williams, Chem. Rev. 2002, 102, 1669.
[9] For a previous example of the connection of the subunits by
amide bond formation to a secondary amine, see: E. J. Martinez,
E. J. Corey, Org. Lett. 2000, 2, 993.
[10] A. K. Sinhababu, A. K. Ghosh, R. T. Borchardt, J. Med. Chem.
1985, 28, 1273.
[11] R. M. Williams, P. P. Ehrlich, W. Zhai, J. Hendrix, J. Org. Chem.
1987, 52, 2615.
[12] A. Fujii, S. Hashiguchi, N. Uematsu, T. Ikariya, R. Noyori, J. Am.
Chem. Soc. 1996, 118, 2521.
triol 29. The latter, upon treatment with a 1:1 ate complexof
BuLi and DIBAL-H,[23] underwent partial reduction of the
lactam to provide oxazolidine 30. The reduction seemed to be
highly dependent on the reactivity of the ate complex, which
may differ from batch to batch. It was found that if there was a
slight excess of “BuLi” in the ate complex, the Troc protecting
group was easily reduced.[24] Fortunately, the phenol hydroxy
group was selectively protected as its allyl ether by treatment
of oxazolidine 30 with allyl bromide in the presence of Hünig
base. Upon exposure to KCN in acetic acid, 30 underwent
ring opening and simultaneous introduction of the C21 cyano
group to give aminonitrile 31 as a single isomer. The MOM
group was cleaved by treatment of 31 with TFA to give the
goal compound 32. As this compound is an advanced
intermediate in the Fukuyama total synthesis of 1,[2d] our
work described herein constitutes a formal total synthesis of
ecteinascidin 743.
[13] A. S. Thompson, G. R. Humphrey, A. M. DeMarco, D. J.
Mathre, E. J. Grabowski, J. Org. Chem. 1993, 58, 5886.
[14] a) J. M. Bobbitt, J. C. Sih, J. Org. Chem. 1968, 33, 856; b) J. M.
Bobbitt, T. E. Moore, J. Org. Chem. 1968, 33, 2958.
[15] K. Laali, R. J. Gerzina, C. M. Flajnik, C. M. Geric, A. M.
Dombroski, Helv. Chim. Acta 1987, 70, 607.
Having validated our routes, we are now in a position to
redirect our focus to reaching ecteinascidin 743 by novel and
far more concise pathways without necessarily intersecting an
existing pathway. We note that several fascinating pieces of
chemistry have been developed herein: 1) the very straight-
forward routes to the matched subunits 16 and 17, 2) the use
of an unusual o-hydroxystyrene moiety for the vinylogous
Pictet–Spengler cyclization, and 3) exploitation of an unusual
[16] D. B. Dess, J. C. Martin, J. Am. Chem. Soc. 1982, 104, 902.
[17] P. Four, F. Guibe, Tetrahedron Lett. 1982, 23, 1825.
[18] For a related example, see: M. J. Martin-Lopez, F. Bermejo-
Gonzalez, Tetrahedron Lett. 1994, 35, 8843.
[19] E. J. Corey, J. Das, Tetrahedron Lett. 1982, 23, 4217.
[20] J. D. Hobson, J. G. McCluskey, J. Chem. Soc. 1967, 2015.
[21] Direct McCluskey reaction of a MOM derivative of the type 9
failed owing to undesired reactivity at the MOM ether protect-
ing group.
ꢀ
enamide epoxide for the hydration of the C3 C4 double
[22] For a related rearrangement versus selective reduction of an
enamide epoxide, see: L. A. Mitscher, H. Gill, J. A. Filppi, R. L.
Wolgemuth, J. Med. Chem. 1986, 29, 1277.
bond, in the desired sense, through reductive treatment of 25.
We are currently exploring direct syntheses of ecteinasci-
1758
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759

Angewandte
Chemie
[23] S. Kim, K. H. Ahn, J. Org. Chem. 1984, 49, 1717.
[24] The use of LiAlH2(OEt)2, LiAlH3(OEt), and Red-Al gave
intractable reaction products. Along these lines, LiAlH(OtBu)3
reduced lactam 29 to oxazolidine 30, however, in low conversion
(10%), whereas boranes afforded the corresponding amine by
exhaustive reduction of the lactam.
Angew. Chem. Int. Ed. 2006, 45, 1754 –1759
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1759