Total synthesis of (-)-sarain A

Article abstract of DOI:10.1021/ja074300t

This article describes the details of our synthetic studies toward the complex marine alkaloid sarain A. Various strategies were conceived, setbacks encountered, and solutions developed, ultimately leading to a successful enantioselective total synthesis. Our route to (-)-sarain A features a number of key steps, including an asymmetric Michael addition to install the C4′-C3′-C7′ stereotriad, an enoxysilane-N-sulfonyliminium ion cyclization to set the C3 quaternary carbon stereocenter, and assemble the diazatricycloundecane core, a ring-closing metathesis to construct the 13-membered ring, an intramolecular Stille coupling to fashion the unsaturated 14-membered macrocycle, and a late-stage installation of the tertiary amine-aldehyde proximity interaction.

Full text of DOI:10.1021/ja074300t

Published on Web 09/12/2007
Total Synthesis of (-)-Sarain A
Michael H. Becker,1a Peter Chua,1b Robert Downham,1c Christopher J. Douglas,1d
Neil K. Garg,1e Sheldon Hiebert,1f Stefan Jaroch,1g Richard T. Matsuoka,1h
Joy A. Middleton,1i Fay W. Ng,1j and Larry E. Overman*
Contribution from the Department of Chemistry, 1102 Natural Sciences II, UniVersity of
California, IrVine, California 92697-2025
Received June 13, 2007; E-mail: leoverma@uci.edu
Abstract: This article describes the details of our synthetic studies toward the complex marine alkaloid
sarain A. Various strategies were conceived, setbacks encountered, and solutions developed, ultimately
leading to a successful enantioselective total synthesis. Our route to (-)-sarain A features a number of
key steps, including an asymmetric Michael addition to install the C4′-C3′-C7′ stereotriad, an enoxysilane-
N-sulfonyliminium ion cyclization to set the C3 quaternary carbon stereocenter, and assemble the
diazatricycloundecane core, a ring-closing metathesis to construct the 13-membered ring, an intramolecular
Stille coupling to fashion the unsaturated 14-membered macrocycle, and a late-stage installation of the
tertiary amine-aldehyde proximity interaction.
Introduction
Complex molecules isolated from marine environments
continue to spark the interest of synthetic chemists. In 1986,
Cimino and co-workers discovered the sarain alkaloids (1-3,
Figure 1) from the marine sponge Reniera sarai, which was
collected in the Bay of Naples.2,3 Although attempts to resolve
the structures of sarains A-C were unsuccessful for several
years, a diacetate derivative of sarain A (i.e., 4) eventually
provided crystals suitable for X-ray diffraction.2c,4 The results
of these studies revealed the impressive architecture of sarain
A, which contains a number of daunting synthetic challenges,
namely: (a) a densely functionalized diazatricycloundecane core,
which contains five contiguous stereogenic centers, one of which
is quaternary (C3); (b) a saturated 13-membered macrocycle;
(1) Current addresses: (a) BASF Aktiengesellschaft, Carl-Bosch-Strasse 38,
67056 Ludwigshafen, Germany. (b) Cylene Pharmaceuticals, 5820 Nancy
Ridge Drive, Suite 200, San Diego, CA 92121. (c) F2G Ltd, P.O. Box 1,
Lankro Way, Eccles, Manchester M30 0BH, United Kingdom. (d)
University of Minnesota, Department of Chemistry, 139 Smith Hall, 207
Pleasant St. SE, Minneapolis, MN 55455-0431. (e) University of
California, Los Angeles, Department of Chemistry and Biochemistry, 607
Charles E. Young Drive East, Box 951569, Los Angeles, CA, 90095-
0515. (f) Bristol-Myers Squibb, 5 Research Parkway, Wallingford, CT
06492. (g) Bayer Schering Pharma AG, Medicinal Chemistry Berlin,
D-13342 Berlin, Germany. (h) GlaxoSmithKline, Chemical Development,
Five Moore Drive/P.O. Box 13398, Research Triangle Park, NC 27709-
3398. (i) Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080.
(j) Columbia University, Department of Chemistry, 3000 Broadway, New
York, NY 10027.
Figure 1. Sarain alkaloids (1-3) and the X-ray structure of diacetate
derivative 4.
(c) a 14-membered ring containing both skipped-triene and
vicinal diol functional groups; and (d) a tertiary amine-aldehyde
proximity interaction, which was not only found to be sensitive
to pH and solvent environment, but also complicated the
purification and characterization of the natural product.5 It should
be noted that related proximity interactions have previously been
studied,6,7 though sarains A-C (1-3) mark the first reported
instance of such an interaction between a trialkylamine and an
aldehyde occurring in a natural product.8 In addition, the
(2) (a) Cimino, G.; De Stefano, S.; Scognamiglio, G.; Sodano, G.; Trivellone,
E. Bull. Soc. Chim. Belg. 1986, 95, 783-800. (b) Cimino, G.; Mattia, C.
A.; Mazzarella, L.; Puliti, R.; Scognamiglio, G.; Spinella, A.; Trivellone,
E. Tetrahedron 1989, 45, 3863-3872. (c) Cimino, G.; Scognamiglio, G.;
Spinella, A.; Trivellone, E. J. Nat. Prod. 1990, 53, 1519-1525. (d) Guo,
Y.; Madaio, A.; Trivellone, E.; Scognamiglio, G.; Cimino, G. Tetrahedron
1996, 52, 8341-8348.
(3) Although referred to as “sarains A-C” in an initial publication,2b these
alkaloids were called “saraines A-C” in later publications from the Cimino
group. These alkaloids have most commonly been termed “sarain” in the
literature.
(5) Chromatographic purifications of sarains A-C were described as seldom
being reproducible, possibly because of the zwitterionic character of these
natural products; see reference 2b.
(4) Cambridge Structural Database (CSD) reference code for diacetate 4:
SAZRAM.
9
10.1021/ja074300t CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 11987-12002
11987

A R T I C L E S
Scheme 1
Becker et al.
absolute configurations of sarains A-C (1-3) have been
proposed on the basis of Mosher’s ester analysis and are depicted
accordingly in Figure 1.2d Sarains A-C are reported to display
modest antibacterial, insecticidal, and antitumor activities.9
Biosynthetically, sarains A-C belong to a larger group of
natural products believed to be derived from bis(dihydropyri-
dine) macrocycles.10 It was initially proposed that, in nature,
sarain A (1) could be accessed from pyridinyl cyclophanes, such
as 5, through a series of unspecified transannular reactions
(Scheme 1).2c,d Subsequently, Marazano and co-workers sug-
gested a possible mechanism for these transformations.11 Pro-
tonation of bis(dihydropyridine) 5 would afford iminium species
6, which could undergo an intramolecular cyclization to provide
enamine 7. Reduction of the R,â-unsaturated iminium of 7 and
reaction of the enamine with an electrophile X+ would deliver
iminium ion intermediate 8. Another intramolecular cyclization
would take place, followed by trapping with water to form
hemiaminal 9. Finally, displacement of leaving group X by the
proximal nitrogen would construct the pyrrolidine ring of sarain
A (1). Although alternative biosynthetic proposals have surfaced
in recent years,12 credibility for the initial hypothesis stems from
the notion that related alkaloids, such as manzamine A and
keramaphidin B, are thought to be biosynthetically derived from
similar precursors.10 These natural products, as well as other
congeners that possess a minimum of two nitrogen atoms and
contain macrocyclic rings of at least eight atoms, are shown in
Figure 2.13
Figure 2. Alkaloids biosynthetically related to sarains A-C (1-3).
efforts toward these remarkable alkaloids. Aside from the work
previously carried out in our laboratory,14 significant progress
has been made by the Weinreb,15 Heathcock,16 Cha,17 and
Marazano12 groups. The details surrounding these endeavors
have recently been reviewed.14d To summarize, successful
preparations of the diazatricycloundecane core and the saturated
macrocycle have been reported;12b, 15c,d,16d,17b,c however, aside
from recent work in our laboratory,14e studies involving the
synthesis of the highly functionalized 14-membered ring have
been limited,16c and those concerning the tertiary amine-
aldehyde proximity interaction have been nonexistent. Herein,
we provide a comprehensive account of our efforts toward (-)-
sarain A, which recently resulted in the first total synthesis of
this intriguing natural product.14e
(13) For sarain-1, see: reference 2a. For keramaphidin B, see: (a) Kobayashi,
J.; Tsuda, M.; Kawasaki, N.; Matsumoto, K.; Adachi, T. Tetrahedron Lett.
1994, 35, 4383-4386. For madangamine A, see: (b) Kong, F.; Andersen,
R. J.; Allen, T. M. J. Am. Chem. Soc. 1994, 116, 6007-6008. For
haliclonacyclamine E, see: (c) Clark, R. J.; Field, K. L.; Charan, R. D.;
Garson, M. J.; Brereton, I. M.; Willis, A. C. Tetrahedron 1998, 54, 8811-
8826. For manzamine A, see: (d) Sakai, R.; Higa, T.; Jefford, C. W.;
Bernardinelli, G. J. Am. Chem. Soc. 1986, 108, 6404-6405. For upenamide,
see: (e) Jime´nez, J. I.; Goetz, G.; Mau, C. M. S.; Yoshida, W. Y.; Scheuer,
P. J.; Williamson, R. T.; Kelly, M. J. Org. Chem. 2000, 65, 8465-8469.
For cyclostellettamine A, see: (f) Fusetani, N.; Asai, N.; Matsugnaga, S.
Tetrahedron Lett. 1994, 35, 3967-3970.
(14) (a) Downham, R.; Ng, F. W.; Overman, L. E. J. Org. Chem. 1998, 63,
8096-8097. (b) Jaroch, S.; Matsuoka, R. T.; Overman, L. E. Tetrahedron
Lett. 1999, 40, 1273-1276. (c) Douglas, C. J.; Hiebert, S.; Overman, L.
E. Org. Lett. 2005, 7, 933-936. (d) Douglas, C. J. Ph.D. Dissertation,
University of California, Irvine, CA, 2005. (e) Garg, N. K.; Hiebert, S.;
Overman, L. E. Angew. Chem., Int. Ed. 2006, 45, 2912-2915.
(15) (a) Sisko, J.; Weinreb, S. M. J. Org. Chem. 1991, 56, 3210-3211. (b)
Sisko, J.; Henry, J. R.; Weinreb, S. M. J. Org. Chem. 1993, 58, 4945-
4951. (c) Irie, O.; Samizu, K.; Henry, J. R.; Weinreb, S. M. J. Org. Chem.
1999, 64, 587-595. (d) Hong, S.; Yang, J.; Weinreb, S. M. J. Org. Chem.
2006, 71, 2078-2089.
(16) (a) Henke, B. R.; Kouklis, A. J.; Heathcock, C. H. J. Org. Chem. 1992,
57, 7056-7066. (b) Griffith, D. A.; Heathcock, C. H. Tetrahedron Lett.
1995, 36, 2381-2384. (c) Heathcock, C. H.; Clasby, M.; Griffith, D. A.;
Henke, B. R.; Sharp, M. J. Synlett 1995, 467-474. (d) Denhart, D. J.;
Griffith, D. A.; Heathcock, C. H. J. Org. Chem. 1998, 63, 9616-9617.
(17) (a) Sung, M. J.; Lee, H. I.; Chong, Y.; Cha, J. K. Org. Lett. 1999, 1, 2017-
2019. (b) Sung, M. J.; Lee, H. I.; Lee, H. B.; Cha, J. K. J. Org. Chem.
2003, 68, 2205-2208. (c) Lee, H. I.; Sung, M. J.; Lee, H. B.; Cha, J. K.
Heterocycles 2004, 62, 407-422.
Soon after the unprecedented structures of sarains A-C (1-
3) were disclosed, laboratories worldwide initiated synthetic
(6) For pioneering studies involving ketone-amine proximity interactions,
see: (a) Leonard, N. J.; Oki, M.; Chiavarelli, S. J. Am. Chem. Soc. 1955,
77, 6234-6237. (b) Bu¨rgi, H. B.; Dunitz, J. D.; Shefter, E. J. Am. Chem.
Soc. 1973, 95, 5065-5067.
(7) For studies involving tertiary amine-aldehyde proximity interactions,
see: (a) McCrindle, R.; McAlees, A. J. J. Chem. Soc., Chem. Commun.
1983, 61-62. (b) Kirby, A. J.; Komarov, I. V.; Bilenko, V. A.; Davies, J.
E.; Rawson, J. M. Chem. Commun. 2002, 2106-2107.
(8) Since the isolation of sarains A-C, another alkaloid that displays a tertiary
amine-aldehyde proximity interaction was isolated; see: Guo, Y.; Triv-
ellone, E.; Scognamiglio, G.; Cimino, G. Tetrahedron 1998, 54, 541-550.
(9) Caprioli, V.; Cimino, G.; De Guilio, A.; Madaio, A.; Scognamiglio, G.;
Trivellone, E. Comp. Biochem. Physiol. B 1992, 103B, 293-296.
(10) (a) Baldwin, J. E.; Whitehead, R. C. Tetrahedron Lett. 1992, 33, 2059-
2062. (b) Baldwin, J. E.; Claridge, T. D. W.; Culshaw, A. J.; Heupel, F.
A.; Lee, V.; Spring, D. R.; Whitehead, R. C.; Boughtflower, R. J.; Mutton,
I. M.; Upton, R. J. Angew. Chem., Int. Ed. 1998, 37, 2661-2663.
(11) Gil, L.; Gateau-Olesker, A.; Marazano, C.; Das, B. C. Tetrahedron Lett.
1995, 36, 707-710.
(12) (a) Hourcade, S.; Ferdenzi, A.; Retailleau, P.; Mons, S.; Marazano, C. Eur.
J. Org. Chem. 2005, 1302-1310. (b) Ge, C. S.; Hourcade, S.; Ferdenzi,
A.; Chiaroni, A.; Mons, S.; Delpech, B.; Marazano, C. Eur. J. Org. Chem.
2006, 4106-4114.
9
11988 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 2
Scheme 3
Scheme 4
Initial Synthesis Planning. In our initial synthesis planning,
we elected to focus efforts on construction of the diazatricy-
cloundecane core and postulated that sarain A (1) could
ultimately be derived from aldehyde 10 (Scheme 2). In turn,
we hoped that aldehyde 10 would be accessed from enoxysilane
12 by intramolecular cyclization onto an N-sulfonyliminium ion
species (e.g., 11). Although Mannich-type transformations
involving N-sulfonyliminium ion species were known,18 no
examples using enoxysilane nucleophiles had been reported in
the literature. If successful, this new transformation would not
only lead to assembly of the sarain A core, but would
simultaneously set the congested C3 quaternary stereocenter.
Enoxysilane 12 could be prepared from pyrrolidinone 13, which
in turn would be accessed from non-allylated precursor 14.
Further disconnection of lactam 14 revealed amino acid 15,
which was recognized as a being a 2,3-disubstituted glutamic
acid derivative. Methods for constructing related, simpler
structural frameworks had been reported in the literature,19 and
so began our total synthesis endeavor.
A C4′-C3′-C7′ stereotriad (ent-18) could be accessible. To
initiate work in this area, oxazoline 16 was prepared by known
methods beginning from L-threonine derivatives.20 Although the
use of this enantiomer of the starting material would provide
access to ent-18, the strategy would later be amenable to the
preparation of Michael adducts 18, en route to the presumed
naturally occurring (-) enantiomer of sarain A (1).2d
To explore the use of enoate electrophiles as partners with
oxazoline 16 in Michael additions, E and Z enoates 24 and 25
were prepared as shown in Scheme 4. Monosilylation of 1,3-
propanediol (19)21 afforded TBS and TBDPS ethers 20 and 21,
respectively, which could be individually oxidized to aldehydes
22 and 23 using Swern conditions.22 Horner-Wadsworth-
Emmons homologation23 of TBS ether 22 delivered (E)-enoate
24, whereas the modified variant disclosed by Still and Gennari24
using TBDPS ether 23 furnished (Z)-enoate 25. These substrates
were considered optimal for initial studies because they were
readily accessible and contained robust silyl ethers, which
potentially could be used as functional group handles for the
eventual installation of an amine substituent.
Results and Discussion
Constructing the C4′-C3′-C7′ Stereotriad. In the 1980s,
Seebach and co-workers introduced a powerful asymmetric
transformation that could potentially be used to prepare 2,3-
disubstituted glutamate derivatives beginning from enantiopure
oxazolines.19a,b As shown in Scheme 3, reaction of oxazoline
16 with lithium diisopropylamide (LDA), followed by quenching
with (E)-1-nitro-1-propene afforded Michael adduct 17 in 80%
yield, with excellent diastereoselectivity (dr > 20:1). Notably,
two new stereogenic centers are generated in this process,
although the stereochemical configuration â to the nitro group
(i.e., C4′) was not determined. If this stereocenter could be
controlled to provide the desired configuration at C4′, and if
alternative Michael acceptors and oxazoline derivatives could
be employed, a variety of useful products possessing the sarain
Our studies of the use of enoates 24 and 25 as Michael
acceptors are summarized in Scheme 5. Upon reaction of
oxazoline 16 with LDA in DME, the corresponding lithium
enolate was generated at low temperature. Subsequent introduc-
tion of either enoate 24 or 25 led to formation of Michael
adducts 26 and 28, respectively, both with excellent diastereo-
selectivity. Although the configuration at C4′ for these species
was not known initially, the structural assignments were inferred
upon derivatization as follows: the Michael addition product
(20) Oxazoline 16 was prepared by reaction of methyl N-benzoyl-(L)-threoninate
with thionyl chloride; see: Elliot, D. F. J. Chem. Soc. 1950, 62-68.
(21) McDougal, P. G.; Rico, J. G.; Oh, Y.-I.; Condon, B. D. J. Org. Chem.
1986, 51, 3388-3390.
(18) For an excellent review of N-sulfonyliminium ion chemistry, see: Weinreb,
S. M. In Topics in Current Chemistry; Springer-Verlag: Berlin, 1997; Vol.
190, pp 131-184.
(19) (a) Seebach, D.; Aebi, J. D. Tetrahedron Lett. 1983, 24, 3311-3314. (b)
Calderari, G.; Seebach, D. HelV. Chim. Acta 1985, 68, 1592-1604. (c)
Masahiko, Y.; Torisu, K.; Minami, T. Chem. Lett. 1990, 377-380.
(22) Swern, D.; Mancuso, A. J. Synthesis 1981, 165-185.
(23) (a) Thompson, S. K.; Heathcock, C. H. J. Org. Chem. 1990, 55, 3386-
3388. (b) For a pertinent review, see: Maryanoff, B. E.; Reitz, A. B. Chem.
ReV. 1989, 89, 863-927.
(24) Still, W. C.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405-4408.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11989

A R T I C L E S
Scheme 5
Becker et al.
Scheme 6
obtained from (E)-enoate 24 was exposed to TFA to provide
hydroxy ester 27, whereas the Michael adduct derived from (Z)-
enoate 25 was desilylated and reacted under the identical acidic
conditions to afford δ-lactone 29. Because the 2-pyrrolidinone
intermediate having a cis relationship between the â-hydroxy-
ethyl and ester functional groups should lactonize more readily
than stereoisomer 27, stereostructures 26 and 28 were assigned
to the initial Michael adducts.25 Thus, it became clear that the
product possessing the desired relative stereochemical config-
uration at C4′-C3′-C7′ (i.e., 28) could be obtained if a (Z)-
enoate was employed in the Michael addition.
to afford Boc-protected sulfonamide 30 (Scheme 6).28 Alter-
natively, sulfonamide 30 could be accessed using a tosylation/
displacement procedure. This latter method was typically
employed to access larger quantities of sulfonamide 30, since
the Mitsunobu approach required the difficult separation of Ph3-
PO and excess DEAD from the desired product. Lithiation of
30 and quenching with methyl chloroformate afforded ester 31,
which in turn was reduced under Lindlar conditions to furnish
(Z)-enoate 32.29 With respect to the oxazoline component, we
chose to synthesize an alkoxymethyl derivative. Beginning from
diethyl L-tartrate ((+)-33), azide 34 was prepared according to
an established procedure.30 Monoreduction of diester 34,31
followed by silylation of the primary alcohol, generated hydroxy
azide 35. After hydrogenation of azide 35, the resulting amino
alcohol was converted to oxazoline (+)-36 in the presence of
methyl benzimidate hydrochloride.32 These routes to enoate 32
and oxazoline (+)-36 proved quite robust and scalable; to date,
several hundred grams of each of these fragments have been
prepared.
With these new enoate and oxazoline coupling partners in
hand, we explored the critical Michael addition with the hope
that all substrate variations would be tolerated. Fortunately,
deprotonation of oxazoline (+)-36 with lithium hexamethyld-
isilazide (LHMDS) in a 2:1 mixture of DME and THF at
-78 °C, in the presence of (Z)-enoate 32, delivered Michael
adduct 37 in 71% yield (Scheme 7). This product possesses the
required relative configuration of the C4′-C3′-C7′ stereotriad
of sarain A (1), as well as sufficient functional group handles
to plausibly access the natural product. Although the enantiomer
of Michael adduct 37 would also be accessible, beginning from
diethyl D-tartrate, we elected to temporarily continue studies in
the less-expensive enantiomeric series.
Figure 3. Plausible transition structures for Michael additions involving
E and Z enoates 24 and 25.
Plausible transition structures E and Z are suggested in Figure
3 to rationalize the stereochemical outcome of the Michael
addition reactions involving enoates 24 and 25.26 The 5-methyl
group of the oxazoline anion shields the R-face, forcing enoate
approach to occur from the â-face of the Li enolate as depicted.19
It is also believed that Li plays a critical organizational role in
the transition state, as supported by the observation that
diastereoselectivity is eroded in the presence of HMPA.14b
After identifying that Z enoates could be used as coupling
partners with oxazoline anions to construct the stereotriad
contained in sarain A, we designed enoate and oxazoline
derivatives that would likely be useful for the total synthesis
effort. Regarding the enoate fragment, a compound containing
a terminal nitrogen substituent, rather than a silyl ether, would
curtail the need for functional group interconversion steps at a
later stage.27 To arrive at the desired enoate, 3-butyn-1-ol was
allowed to react with (Boc)TsNH under Mitsunobu conditions
Initial Assembly of the Diazatricycloundecane Core by
an Enoxysilane-N-Sulfonyliminium Ion Cyclization. Diester
(28) Henry, J. R.; Marcin, L. R.; McIntosh, M. C.; Scola, P. M.; Harris, G. D.,
Jr.; Weinreb, S. M. Tetrahedron Lett. 1989, 30, 5709-5712.
(29) Taschner, M. J.; Rosen, T.; Heathcock, C. H. Org. Synth. 1986, 64, 108-
113.
(25) (a) Molecular mechanics calculations (MMFF) were performed using
Spartan ’02; Copyright 1991-2001; Wavefunction Inc.: Irvine, CA. (b)
cis-fused lactone 29 was calculated to be more stable than its trans-fused
counterpart by 6.4 kcal/mol.
(26) Seebach, D.; Golinski, J. HelV. Chim. Acta 1981, 64, 1413-1423.
(27) Prior to the use of sulfonamide 32, we explored the use of triazinone- and
phthalamido-containing enoates. Although these species were competent
in the Michael addition chemistry, neither the triazinone or phthalamido
groups were chemically inert in later transformations.
(30) Satio, S.; Komada, L.; Moriwake, T. Organic Syntheses; Wiley: New York,
1998; Collect. Vol. IX, p 200.
(31) Sauret-Cladie`re, S.; Jeminet, G. Tetrahedron: Asymmetry 1997, 8, 417-
423.
(32) For a comprehensive review of the chemistry of oxazolines, see: Gant, T.
G.; Meyers, A. I. Tetrahedron 1994, 50, 2297-2360.
9
11990 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 7
Scheme 9
Scheme 8
37 was advanced in the total synthesis according to the approach
depicted in Scheme 8. The Boc group of 37 was removed by
thermolysis28 to afford an intermediate tosylamide, which in
turn was subjected to Me3Al to induce amidation.33 Next, we
hoped to carry out an alkylation of lactam (-)-38 with 3-bromo-
2-methylpropene (39).34 This reagent was strategically chosen
for initial studies because: (a) it would provide all of the
remaining carbon atoms of the sarain A diazatricycloundecane
core, and (b) it would ultimately allow us to test the enoxysilane-
N-sulfonyliminium ion cyclization, with concomitant formation
of the C3 quaternary carbon center, in the absence of a complex
side chain or macrocycle. Upon examining the alkylation
reaction, we were gratified to find that deprotonation of lactam
(-)-38 with LHMDS, followed by quenching with excess
bromide 39 provided lactam 41 as a single isomer in 79% yield.
Although it would later be confirmed, the stereochemical
assignment for 41 was made initially with the assumption that
alkylation should occur from the enolate face opposite the
oxazoline side chain (e.g., 40).35 In the final step shown in
Scheme 8, alkylated compound 41 was allowed to react with
HCl to promote oxazoline ring cleavage,36 followed by trans-
lactamization. After workup, pyrrolidinone 42 was isolated in
75% yield.
upon reaction with K2CO3 in MeOH underwent mono-Boc
deprotection, benzoate cleavage, and lactonization to afford
spirolactone 45. Subsequent reaction of lactone 45 with i-Bu2-
AlH led to partial reduction to the corresponding lactol, at which
point the reaction mixture was quenched with HCl to provide
tricyclic aminal 46 in 78% yield. At this stage, the stereochem-
ical configuration of the methallyl group was confirmed by 1H
NMR NOESY studies. In the next transformation, reaction of
secondary alcohol 46 with sodium hexamethyldisilazide (NaH-
MDS) promoted cyclization to smoothly furnish tetracyclic
oxazolidinone 47. Finally, a three-step sequence involving
hydroboration with oxidative workup, Dess-Martin oxidation,39
and treatment of the resulting aldehyde to TIPSOTf,40 delivered
cyclization precursor 48 as a 3:2 mixture of alkene stereoiso-
mers.
With the appropriate substrate in hand, we attempted the
crucial intramolecular iminium ion cyclization of enoxysilanes
48 (Scheme 9). Although construction of the C3-C4 bond of
sarain A was at the time precedented by the early studies of
Sisko and Weinreb,15b the use of an enoxysilane nucleophile
and the simultaneous introduction of the critical C3 quaternary
carbon stereocenter had not previously been demonstrated. A
number of Lewis acids were examined to induce the desired
cyclization. Whereas many Lewis acids were ineffective (SnCl4,
BF3•OEt2, Me3Al),41 addition of BBr3 to enoxysilanes 48 in CH2-
Cl2 from -78 °C to room-temperature delivered a mixture of
A robust sequence was developed to elaborate pyrrolidinone
42 to a suitable enoxysilane-N-sulfonyliminium ion cyclization
precursor. Boc-protection of 42,37 followed by a standard two-
step reduction of pyrrolidinone 43,38 yielded pyrrolidine 44
(Scheme 9). The TBDPS protecting group of 44 was cleaved
with TBAF to provide an intermediate primary alcohol, which
(33) Basha, A.; Lipton, M.; Weinreb, S. M. Tetrahedron Lett. 1977, 18, 4171-
4172.
(34) Attempts to directly alkylate ester 37 were unsuccessful, as were attempts
to trap the intermediate lithium enolate from the Michael reaction of (+)-
36 and 32.
(38) (a) Hubert, J. C.; Winjberg, J. B. P. A.; Speckamp, W. N. Tetrahedron
1975, 31, 1437-1441. (b) Lenz, G. R. J. Chem. Soc., Perkin Trans. 1990,
1, 33-38.
(35) (a) Liu, L. T.; Hong, P.-C.; Huang, H.-L.; Chen, S.-F.; Wang, C.-L. J.;
Wen, Y.-S. Tetrahedron: Asymmetry 2001, 12, 419-426. (b) Johnson, T.
A.; Jang, D. O.; Slafer, B. W.; Curtis, M. D.; Beak, P. J. Am. Chem. Soc.
2002, 124, 11689-11698.
(39) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277-7287.
(40) Emde, H.; Domsch, D.; Feger, H.; Frick, U.; Go¨tz, A.; Hergott, H. H.;
Hofmann, K.; Kober, W.; Kra¨geloh, K.; Oesterle, T.; Steppan, W.; West,
W.; Simchen, G. Synthesis 1982, 1-26.
(36) Meyers, A. I.; Roth, G. P.; Hoyer, D.; Barner, B. A.; Laucher, D. J. Am.
Chem. Soc. 1988, 110, 4611-4624.
(37) Grehn, L.; Ragnarsson, U. Angew. Chem., Int. Ed. Engl. 1985, 24, 510-
511.
(41) Attempts to cyclize the corresponding TES derivative of 48 led either to
no reaction or substantial non-specific decomposition.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11991

A R T I C L E S
Scheme 10
Becker et al.
Scheme 11
two cyclization products in 81% overall yield. On the basis of
NMR and X-ray diffraction studies,14a,42 these products were
assigned as aldehydes 49 and 50, which formed in a 1:3 ratio
favoring undesired epimer 50. Although disappointed by the
unfavorable diastereoselectivity observed in setting the C3
quaternary stereocenter, we felt that this problem might be
surmounted by slight modification of our fundamental strategy.
At this stage, we reasoned that optimization of the cyclization
would best be deferred to future studies in nonmodel systems.
Strategies for Controlling the C3 Quaternary Carbon
Stereocenter. Having successfully assembled a model system
for the sarain A core (49, Scheme 9), we turned our attention
to accessing related compounds that could ultimately provide
access to the natural product. All future work was carried out
in the postulated natural enantiomeric series, beginning from
diethyl D-tartrate. A major concern from our earlier studies
involved the unfavorable selectivity observed in the formation
of the C3 quaternary carbon stereocenter. To circumvent this
problem, two strategies were conceived and are illustrated in
Scheme 10. In the first approach, the saturated macrocycle of
sarain A would be installed prior to the iminium ion cyclization
(Plan A). Although this strategy would not be amenable to the
N-sulfonyliminium ion cyclizations developed earlier, an N-acyl
iminium ion cyclization of enoxysilane 51 could potentially be
used instead. On the basis of molecular modeling calculations,
it was expected that formation of desired product 52 would be
greatly favored over production of the undesired epimer 53 (R
) H, Erel ) +14.6 kcal/mol).25a Alternatively, the use of a
substrate containing a larger C3 side chain (e.g., 54), together
with further reaction optimization, could potentially lead to
improved selectivity in forming the C3 quaternary carbon
stereocenter (Plan B). In this latter scenario, the disfavorable
syn-pentane interaction between the C3 side chain and the
pyrrolidine ring methylene would be somewhat extenuated in
the transition state leading to undesired epimer 56. This
hypothesis is in accord with molecular modeling calculations,
which suggest that aldehyde 56 is higher in energy than its
epimer 55, by approximately +7.6 kcal/mol if R′ ) n-pentyl.25a
An Initial Attempt to Control the C3 Quaternary Carbon
Stereocenter: An Unproductive Detour. In our initial efforts,
we implemented Plan A as a means to control installation of
the C3 quaternary carbon stereocenter, as shown in Scheme 11.
Following our earlier studies in the unnatural series (see Scheme
6), diethyl D-tartrate ((-)-33) was elaborated to oxazoline (-)-
36 using a five-step sequence. Subsequent lithiation of (-)-36,
followed by addition of enoate 57,43 afforded Michael adduct
58 in 71% yield. Cleavage of the PMB protective group of 58
under oxidative conditions provided an intermediate alcohol,
which in turn underwent acid-catalyzed cyclization to lactone
59. Next, lactone 59 was alkylated with allylic bromide 6043 to
deliver enyne 61. Upon reaction of 61 with dilute aqueous HCl,
oxazoline cleavage36 and amidation occurred. The resulting
intermediate was then converted to azide 62 under Mitsunobu
conditions. Subsequent Boc-protection of the amide (62 f 63),
followed by reduction of the amide carbonyl,38 produced
pyrrolidine 64. This intermediate was allowed to react with DDQ
to furnish alcohol 65 in 84% yield.
Our efforts to elaborate alcohol 65 toward a macrocycle-
containing iminium ion cyclization substrate are depicted in
Scheme 12. With the aim of fashioning the macrocycle by amide
bond formation, alcohol 65 was oxidized to carboxylic acid 66
using a standard two-step procedure. Reduction of azide 66 with
Ph3P then afforded amino acid 67, which resisted macrocy-
clization under a variety of standard conditions. However, by
slowly adding a solution of amino acid 67 to Mukaiyama’s salt44
at high dilution, it was possible to obtain lactam 68 in 58%
yield. Subsequent treatment of lactam 68 with TBAF in THF,
followed by reaction with K2CO3 in MeOH, provided spiro-
lactone 69. Upon reaction with i-Bu2AlH, intermediate 69 was
(42) Cambridge Structural Database (CSD) reference code for aldehyde 50:
(43) See Supporting Information for details.
(44) Bald, E.; Saigo, K.; Mukaiyama, T. Chem. Lett. 1975, 1163-1166.
GOCZON.
9
11992 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 12
Scheme 14
Scheme 8). Lactam (+)-38 was lithiated, then quenched with
allylic bromide 72,43 to afford alkylated product 73 in 76% yield.
The use of bromide 72 as an alkylating reagent was thought to
have two potential benefits: (a) it would lead to installation of
a larger C3 side chain, which would hopefully improve the
cyclization reaction, and (b) it could provide a means to install
the saturated macrocycle of sarain A after the planned iminium
ion cyclization, because it possesses a convenient functional
group handle (i.e., the Bn-protected alcohol). Following the
general approach developed earlier to access enoxysilanes 48
(see Scheme 9), lactam 73 was elaborated to enoxysilanes 74
in 15 steps, with an overall yield an 18%. With enoxysilanes
74 in hand, we explored the key enoxysilane-N-sulfonyliminium
ion cyclization. Upon reaction of enoxysilanes 74 with excess
BBr3 in CH2Cl2 from -78 °C to room-temperature, a mixture
of products was obtained that reflected a preference for forming
the desired isomer (approximately 3:1).45,46 Unfortunately, the
Bn protecting group was not compatible with these BBr3
conditions, and both alcohol-containing products (i.e., 75 and
77) and the related bromide products (i.e., 76 and 78) were
observed. Nonetheless, this result supported the notion that a
larger C3 side chain would lead to improved diastereoselectivity.
We next sought to further increase the yields and selectivity
observed in the critical iminium ion cyclization of enoxysilanes
74. A variety of experimental parameters were tested, with a
particular emphasis on the Lewis acid promoter. Although many
Lewis acids were examined (e.g., TiCl4, SnCl4, TMSOTf,
BF3•OEt2, etc.), only BBr3 and BCl3 promoted the cyclization.
Of these, BCl3 proved advantageous as it exclusively furnished
products containing a free hydroxyl group (i.e., 75 and 77).
Finally, as an additional modification, we elected to carry out
cyclizations in the presence of 2,6-di-tert-butyl-4-methylpyri-
dine,47 which would serve to sequester protic acids and perhaps
prevent premature cleavage of the TIPS group. Under optimal
reaction conditions (BCl3, 2,6-di-tert-butyl-4-methylpyridine,
-78 f -40 °C), cyclization of substrate 74 occurred with
improved diastereoselectivity (dr 6-8:1), with aldehyde 75
being isolated in 83% yield (Scheme 14).46
Scheme 13
reduced to the corresponding lactol, which was worked up under
acidic conditions to afford oxazolidinone 70. At this stage, to
access an iminium ion cyclization precursor, the piperidine ring
would need to be fashioned, conceivably by conversion of
lactam 70 to pentacyclic aminal 71. Unfortunately, it was not
possible to promote this critical dehydration, despite extensive
experimentation involving both protic and Lewis acids (e.g.,
PPTS, TsOH, BF3•OEt2, etc.). Moreover, exhaustive efforts to
cyclize activated lactol derivatives of 70 (e.g., R ) Ac,
trichloroacetate) were also unsuccessful. Thus, we were forced
to abandon this strategy.
Forming the Saturated Macrocycle. Having significantly
improved the enoxysilane-N-sulfonyliminium ion cyclization en
route to sarain A, two viable strategies were explored for
assembly of the saturated macrocycle beginning from diol 75
(Scheme 15). In the first scenario, alcohol 75 was elaborated to
carboxylic acid and ester derivates (i.e., 79) as a prelude to
constructing the macrocycle by intramolecular lactam formation.
Unfortunately, only trace amounts of the desired lactam product
could be obtained from precursors 79 under a variety of
conditions.
A Second Attempt to Control the C3 Quaternary Carbon
Stereocenter. As described earlier, an alternative strategy for
potentially improving selectivity in the formation of the C3
stereocenter involved the use of a larger C3 side chain during
the iminium ion cyclization, concurrent with further reaction
optimization (Scheme 10, Plan B). To initiate studies in this
area, lactam (+)-38 (Scheme 13) was prepared following the
route used to prepare its enantiomer in our earlier studies (see
(45) Attempts to prepare the corresponding TMS or TBS enoxysilanes gave
low yields of contaminated products, which performed poorly in the
subsequent iminium ion cyclization.
1
(46) Stereochemical configuration was determined by H-1H NOESY experi-
ments.
(47) Anderson, A. G.; Stang, P. J. J. Org. Chem. 1976, 41, 3034-3036.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11993

A R T I C L E S
Scheme 15
Becker et al.
Scheme 17
Scheme 16
C3 side chain and the enoxysilane, multiple cyclization precur-
sors 88 would be accessible. As the latter of these routes allowed
us to defer installation of the side chain to a later stage in the
synthesis, it appeared advantageous and became the focus of
our efforts.49
Given our experience, initial forays into the allylation
approach proceeded in a relatively straightforward manner
(Scheme 17). Allylation of lactam (+)-38 delivered substituted
adduct 86 in 92% yield. Using a now standard sequence, 86
was converted to pyrrolidinone 89, which in turn, was reduced
to pyrrolidine 90.38 Next, mono-Boc deprotection, desilylation,
and spirolactone formation with concomitant methanolysis
afforded intermediate 91. Tetracycle 87 was accessed in two
additional steps, by reduction of the lactone (91 f 92), followed
by reaction of â-amino alcohol 92 with NaOMe.50 Although
we would ultimately develop better methods for elaborating
alkene 87, we initially proceeded by carrying out a hydrobo-
ration/oxidation51 sequence to furnish aldehyde 94.
The second approach relied on the use of ring-closing
metathesis (RCM)48 to construct the macrocycle, a strategy that
drew precedent from Weinreb’s studies in this area.15c To pursue
this plan, diene 80 was generated from diol 75 and subjected
to Grubbs’s second generation catalyst 81. Under these condi-
tions, it was possible to obtain macrolactam 82 in 25% yield.
Although the yield was modest, the RCM strategy appeared
promising; thus, our attention turned to optimizing this approach.
Because we believed the tether lengths of the ring-closing
metathesis substrate could play a critical role in the efficiency
of macrocyclization, we wished to define a flexible route to
RCM precursors. Two approaches were considered as depicted
in Scheme 16. In the first, lactam (+)-38 could be alkylated
with any of a number of different reagents 83. By varying the
chain length (i.e., n value), several potentially useful products
84 could be obtained. Each of these could be elaborated further
to dienes 85. If the double bonds of 85 could be readily
differentiated, it would likely be possible to arrive at the desired
enoxysilane substrates 88. In an alternative strategy, lactam (+)-
38 would be allylated to afford intermediate 86, which in turn,
could be elaborated to tetracycle 87. If the terminal alkene could
now be used as a functional group handle for installation of the
The alkylation of aldehyde 94, and various derivatives thereof,
proved to be challenging. After initial efforts to directly alkylate
aldehyde 94 were deemed unsuccessful, we focused on the
alkylation of metalloenamine derivatives.52 Imine 95 was
prepared by condensation of aldehyde 94 with t-butylamine in
the presence of K2CO3 (Scheme 18). However, attempts to
alkylate the derived metalloenamine species to furnish aldehyde
96, under a variety of standard conditions, led to either recovery
of starting material or decomposition. As an alternative, N,N-
dimethylhydrazone 97 was prepared by reaction of aldehyde
(49) The earlier alkylation strategy was briefly pursued where n ) 4; however,
selective hydroboration of diene 85 proved difficult.
(50) In some instances, product 87 was isolated with a minor contaminant, which
arose from cleavage of the Boc group of 92. By reacting this crude mixture
of products with triphosgene, it was possible to funnel the material to
tetracycle 87.
(48) For reviews on RCM reactions in alkaloid synthesis: (a) Deiters, A.; Martin,
S. F. Chem. ReV. 2004, 104, 2199-2238. (b) Felpin, F. -X.; Lebreton, J.
Eur. J. Org. Chem. 2003, 3693-3712. (c) Pandit, U. K.; Overkleeft, H.
S.; Borer, B. C. Biera¨ugel, H. Eur. J. Org. Chem. 1999, 959-968. (d)
Phillips, A. J.; Abell, A. D. Aldrichimica Acta 1999, 32, 75-89.
(51) Ley, S. V.; Norman, J.; Griffith, W. P.; Marsden, S. P. Synthesis 1994,
639-666.
9
11994 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 18
Scheme 20
Scheme 19
sequent reaction of the intermediate enol ether with HF in
MeCN/H2O provided aldehyde 103. Aldehyde 103 was smoothly
converted to TIPS enoxysilanes 104 as a 3:2 mixture of
stereoisomers.40
94 with 1,1-dimethylhydrazine under Dean-Stark conditions.
However, all efforts to alkylate hydrazone 97 led to extensive
decomposition, rather than to desired product 96.
We also examined the alkylation of an ester enolate (Scheme
19). The requisite substrate, 99, was prepared in two steps from
alcohol 93 by oxidation with TEMPO/NaClO/NaClO2 (93 f
98) in buffered MeCN,53 followed by reaction of the carboxylic
acid product with MeI and K2CO3. Fortunately, the desired
alkylated product, 100, could be obtained after low-temperature
deprotonation of ester 99 and quenching with 5-bromopentene,
albeit in low yields when stoichiometric quantities of base and
electrophile were employed. As the mass balance from this
reaction was largely attributed to recovered starting material
99, more forcing conditions were probed that relied on the use
of excess base and excess electrophile or increased temperatures.
Regrettably, these conditions led to the formation of multiple
byproducts, mostly involving alkylation of the tosyl group.54
Given the difficulties encountered in the attempted alkylation
of various carboxylic acid derivatives, we chose to pursue an
alternative method for installation of the alkene-containing side
chain (Scheme 20). Returning to tetracyclic terminal alkene 87
(see Scheme 17), an ozonolysis/Grignard addition/oxidation
sequence successfully delivered ketone 101 in 76% yield over
three steps. With an appropriate side chain now in place,55 we
explored elaboration of the ketone to an enoxysilane substituent.
Although direct conversion of ketone 101 to TIPS enoxysilane
104 by reaction with diethyl (triisopropylsilyloxymethyl)-
phosphonate and LDA was not possible,56 a related homolo-
gation using phosphonium salt 102 and KHMDS to give a
2-(trimethylsilyl)ethyl (TMSE) enol ether was fruitful.57 Sub-
Figure 4. Optimization of enoxysilane-N-sulfonyliminium ion cyclization
of substrate 104.
With a reliable route to enoxysilanes 104 established, we
directed attention to the critical iminium ion cyclization (Figure
4). Use of the previously optimized buffered BCl3 conditions,
unfortunately, resulted in complex reaction products and low
yields of desired aldehyde 105. When the number of equivalents
of BCl3 was reduced from eight to four, fewer side products
were obtained, and the observed diastereoselectivity was ap-
proximately 5:1 in favor of the desired product. However, in
an unexpected result, we found that warmer temperatures
(0 °C f room-temperature) actually led to cleaner reactions,
thereby facilitating the selective formation of aldehyde 105,
which could be isolated in 85% yield.58 Several features
regarding this transformation should be noted: (a) only the
desired diastereomer (i.e., 105) is observed under the reaction
1
conditions, as determined by H NMR analysis of the crude
reaction products; (b) the cyclization is fast, with the starting
material being consumed upon addition of BCl3 at 0 °C; and
(c) these optimal conditions could be used to reliably prepare
gram quantities of key intermediate 105.
(52) For pertinent reviews, see: (a) Hickmott, P. W. Tetrahedron 1982, 38,
3363-3446. (b) Whitesell, J. K.; Whitesell, M. A. Synthesis 1983, 517-
536.
(56) Direct synthesis of enoxysilanes through a Wittig reaction remains an
unsolved problem in organic synthesis. For a pertinent discussion, see: (a)
Kluge, A. F.; Cloudsdale, I. S. J. Org. Chem. 1979, 44, 4847-4852. For
a review regarding one-carbon homologations of ketones, see: (b) Badham,
N. F. Tetrahedron 2004, 60, 11-42.
(57) Scho¨nauer, K.; Zbiral, E. Tetrahedron Lett. 1983, 24, 573-576.
(58) Reaction of the corresponding TMSE enol ether of 104 (see Scheme 20)
under the identical conditions also led to the formation of aldehyde 105,
albeit only in 55% yield.
(53) Zhao, M. M.; Li, J.; Mano, E.; Song, Z. J.; Tschaen, D. M. Org. Synth.
2005, 81, 195-203.
(54) p-Toluenesulfonamides can be deprotonated at the methyl group by bases
as weak as LDA. For a discussion, see: MacNeil, S. L.; Familoni, O. B.;
Snieckus, V. J. Org. Chem. 2001, 66, 3662-3670.
(55) On the basis of Weinreb’s RCM studies on related compounds, a 5-carbon
C3 substituent was thought to be suitable; see reference 15c.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11995

A R T I C L E S
Scheme 21
Becker et al.
Figure 5. TIPS-N,O-acetal 113 is observed in cyclization reactions.
Some mechanistic insight was gleaned by examining saturated
substrate 107 in the enoxysilane-N-sulfonyliminium ion cy-
clization (Scheme 22). Under the optimal cyclization conditions
(0 °C f room-temperature), a 9:1 mixture of aldehyde 108 and
its epimer 109 was observed (entry 1). The fact that epimer
109 was isolated supports the notion that the corresponding
undesired product (i.e., 106) in the real system likely decom-
poses by Prins pathways (see Scheme 21).
Scheme 22
One hypothesis to explain the high (9:1) stereoselectivity
observed in the enoxysilane-N-sulfonyliminium ion cyclization
of substrate 107 is that the cyclization process is reversible and
the products are thereby formed under thermodynamic control.
In this scenario, N-tosyliminium ion 110 could cyclize to afford
oxocarbenium intermediate 111 or epimer 112. The pyridine
buffer could serve to prevent O-Si bond cleavage of these
newly formed TIPS-oxocarbenium intermediates, thereby al-
lowing thermodynamic equilibrium to be reached. Of the two
potential intermediates, 111 should be thermodynamically
favored as it places the smaller of the two C3 substituents in
juxtaposition with C6 of the pyrrolidine ring, thereby minimizing
a syn-pentane interaction.59 This reversible pathway would favor
formation of aldehyde 108, rather than C3 epimer 109, in accord
with empirical observations and theoretical calculations.60
To probe this reversibility hypothesis, two experiments were
carried out. First, at -78 °C, the reaction was found to proceed
with much lower levels of diastereoselectivity (dr 2.6:1). In a
second experiment, the cyclization was initiated at -78 °C, and
the reaction was then allowed to warm to room-temperature. A
low diastereomeric ratio of 2.8:1 was observed, close to that
obtained at -78 °C. Thus, the cyclization is likely not reversible,
as warming the reaction mixture from -78 °C to room-tem-
perature prior to quenching did not lead to enhanced diastereo-
selectivity. It should be noted that, in all cases, we observed
the formation of TIPS-N,O-acetal 113 (Figure 5),61 which
suggests that cyclization of substrate 107 proceeds through the
TIPS-enoxysilane rather than a boron enolate or enol.
Origin of Diastereoselectivity in the Enoxysilane-N-Sul-
fonyliminium Ion Cyclization. Intrigued by the results obtained
during optimization of the enoxysilane-N-sulfonyliminium ion
cyclization of substrate 104, we sought an explanation for the
improvements in yield and stereoselectivity at higher reaction
temperatures. Because the undesired epimer of product was not
observed under these conditions and isolated yields were not
quantitative, we questioned the stability of the undesired C3
aldehyde epimer 106 to our new reaction conditions (Scheme
21). It was plausible that this epimer would be less stable under
the reaction conditions than epimer 105, because the aldehyde
group of 106 would be easier to activate for subsequent trans-
formations as it is in a much less congested environment. When
a sample of aldehyde 106 was subjected to BCl3 and 2,6-di-
tert-butyl-4-methylpyridine in CH2Cl2 at 0 °C, then warmed to
room-temperature, indeed rapid decomposition took place. Thus,
the high levels of diastereoselectivity can partially be attributed
to byproduct instability. As we believed the decomposition of
aldehyde 106 occurred through Prins cyclization pathways
involving the terminal alkene, an alternative cyclization substrate
bearing a saturated side chain was desired for further studies.
The appropriate substrate 107 was readily obtained by catalytic
hydrogenation of alkene 103, followed by formation of the silyl
enol ether.
Having shed doubt on the reversibility hypothesis, we
considered alternative explanations for the high levels of
diastereoselectivity observed in the enoxysilane-N-sulfonylimin-
ium ion cyclization carried out at 0 °C to room-temperature.
An analysis of the four likely transition structures that would
lead to the observed products 108 and 109 is presented in
Scheme 23. As the cyclization experiments are typically carried
out with a mixture of enoxysilane isomers (3:2, E:Z), two
N-sulfonyliminium ion species are thought to form initially,
namely, (E)-110 and (Z)-110. Whereas cyclization of either of
these intermediates in an antiperiplanar fashion would deliver
(59) Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of Organic
Compounds; Wiley-Interscience: New York, 1994.
(60) Epimer 109 was calculated to be approximately 7.6 kcal/mol higher in
energy than the desired product 108; see Scheme 10.
(61) The isolation of N,O-acetal 113 is somewhat dependent on reactions times
and quenching methods.
9
11996 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 23
Figure 6. Impact of enol ether geometry in the enoxysilane-N-sulfo-
nyliminium ion cyclization.
Scheme 25
Scheme 24
After just 3 min, the reaction mixture was quenched with Et3N,
1
followed by MeOH. As determined by H NMR analysis of
the resulting mixture, a 1:3 mixture of (Z)- and (E)-enoxysilanes
114 was obtained, thus supporting the notion that our optimal
reaction conditions would promote rapid isomerization of the
TIPS enoxysilane intermediates.63
To further investigate the enol ether isomerization hypothesis,
the E and Z stereoisomers of 107 were carefully separated using
preparative HPLC, then individually studied in the enoxysilane-
N-sulfonyliminium ion cyclization. As summarized in Figure
6, the E and Z isomers behaved very differently from one
another, thus confirming the importance of enol ether geometry
in the cyclization outcome. Whereas excellent diastereoselec-
tivity (dr . 20:1) was observed in forming tetracyclic aldehydes
108 and 109 when (E)-107 was employed, use of (Z)-
enoxysilane 107 led to these products in much lower diaster-
eomeric ratios. Moreover, in the case of (Z)-107, variations in
temperature significantly influenced the product distribution,
with lower temperatures leading to lower levels of diastereo-
selectivity. These results are consistent with a scenario in which
the rate of double bond isomerization (i.e., (Z)-110 f (E)-110,
Scheme 24) increases with a higher temperature, where upon it
becomes competitive with cyclization.
desired aldehyde 108, the corresponding synclinal cyclization
pathway would afford epimeric product 109. Neither of the two
synclinal transition structures appear particularly favorable, as
both would suffer from destabilizing interactions between the
n-pentyl side chain (i.e., R′) and the pyrrolidine ring. In
considering the antiperiplanar transition structures, cyclization
of major isomer (E)-110 to form aldehyde epimer 108 appears
quite favorable. In contrast, it is unlikely that 108 could arise
from minor isomer (Z)-110 because of the highly disfavorable
interaction that would exist between the bulky TIPS group and
pyrrolidine ring. According to this analysis, the maximum yield
of desired product 108 that should be obtained in a cyclization
reaction should be approximately 60%, corresponding to the
amount of E enoxysilane present in the starting material.
As the observed yields of desired aldehyde 108 were higher
than 60%, and we believed the reversibility hypothesis to be
untrue, we speculated that stereoisomers (E)- and (Z)-110 could
potentially interconvert under the reaction conditions, likely via
neutral precursors such as 107.62 This interconversion would
funnel the unproductive Z enoxysilane stereoisomer through the
productive cyclization pathway (Scheme 24, (Z)-110 f (E)-
110 f 108). To probe this isomerization hypothesis, a simple
model system was studied. (Z)-Enoxysilane 114, prepared in
one step from commercially available 2-methylpentanal,43 was
exposed to typical buffered-BCl3 reaction conditions at -78 °C.
Having gained an understanding of the factors that govern
selectivity under the optimal buffered-BCl3
enoxysilane-N-
sulfonyliminium ion cyclization conditions, we questioned
what was happening under the reaction conditions (i.e., BBr3,
-78 °C to room-temperature) employed in our early model
studies (see Scheme 9, 48 f 49 + 50). To probe this issue,
(Z)-enol ether 114 was treated with excess BBr3 at -78 °C for
3 min (Scheme 25). After quenching with Et3N, then MeOH,
approximately 80% of the material had undergone cleavage of
(62) For the isomerization of enoxysilanes, see: (a) Deyine, A.; Dujardin, G.;
Mammeri, M.; Poirier, J.-M. Synth. Commun. 1998, 28, 1817-1821. (b)
Ishihara, K.; Nakamura, H.; Nakamura, S.; Yamamoto, H. J. Org. Chem.
1998, 63, 6444-6445. (c) Denmark, S. E.; Pham, S. M. J. Org. Chem.
2003, 68, 5045-5055.
(63) It is plausible that enol ether isomerization occurs by protonation of the
enoxysilane, followed by rapid deprotonation. The proton source is
presumed to be a pyridinium salt of the 2,6-di-tert-butyl-4-methylpyridine
buffer, which would form in the presence of adventitious HCl.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11997

A R T I C L E S
Scheme 26
Becker et al.
1
the TIPS group to afford aldehyde 115, as determined by H
NMR analysis of the crude products. Thus, it is likely that when
enoxysilane-N-sulfonyliminium ion cyclizations are caried out
with BBr3 in the absence of an amine buffer, desilylation takes
place at a rate competitive to cyclization. The distribution of
products in this case would be the result of iminium ion
cyclization occurring by pathways that involve potentially three
different nucleophiles: enoxysilanes, boron enolates, and enols.
According to the analysis summarized in Scheme 23, the latter
two could well react with lower stereoselectivity. The fact that
the BBr3-promoted cyclization (BBr3 used in excess) is initially
fast, but ultimately requires long reaction times for complete
conversion (approximately 24 h at room-temperature), further
supports this notion.
Assembly of the 13-Membered Macrocycle by a Ring-
Closing Metathesis and Subsequent Elaboration Toward
Sarain A. With a reliable, high-yielding route to the diazatri-
cycloundecane core and C3 quaternary carbon stereocenter of
sarain A developed, attention was focused on assembling the
saturated macrocycle by an optimized ring-closing metathesis
strategy. The sequence ultimately developed for elaborating
aldehyde 105 to RCM precursor 118 is highlighted in Scheme
26. Following TBS-protection of the primary alcohol, the
aldehyde was reduced, and the resulting alcohol was silylated
to provide TIPS ether 116 in 75% yield over three steps. Next,
the tosyl group of intermediate 116 was removed with sodium
naphthalenide to furnish amine 117.64 To arrive at the desired
RCM substrate, several options were considered for function-
alizing the secondary amine, such as amidation, alkylation, and
reductive amination. As amidation would require later manipu-
lations, and efforts to alkylate the secondary amine with 1-iodo-
6-heptene65 were unsuccessful, we settled on a reductive
amination procedure. Under optimal conditions, reaction of
amine 117 with 6-hepten-1-al,66 4 Å molecular sieves, and
AcOH in MeCN, followed by addition of NaBH3CN, afforded
tertiary amine 118 in 94% yield.
Figure 7. Ring-closing metathesis of diene 118.
especially considering the overall complexity of our synthetic
intermediates. As shown in Figure 7, we initially explored the
use of Grubbs’ second generation catalyst 81, because this
catalyst had been used in our previous macrocyclization
experiments (see Scheme 15). Although it was possible to obtain
macrocycle 119 using these conditions, multiple byproducts
were observed, with the bulk of material corresponding to
dimeric products 120. The use of Grubbs’ first generation
catalyst 121 to form the desired macrocycle was also examined.
This catalyst was generally considered advantageous since it
probably would not require an acidic additive (i.e., AcOH),67f,68
and would be less likely to lead to thermodynamic mixtures of
products.69 In fact, the use of catalyst 121 improved the yields
of desired monomer 119. After substantial optimization, which
involved lowering the catalyst loading to 5 mol %, carrying
out the transformation at high dilution, and rigorously excluding
O2, dimer 120 could be avoided and desired product 119 was
isolated in 75-85% yield.
Macrocycle 119 was further elaborated toward sarain A as
shown in Scheme 27. Hydrogenation with Pd/C completed
installation of the saturated macrocycle, which upon exposure
to HCl afforded alcohol 122. Reaction of alcohol 122 with
NaHMDS and para-methoxybenzyl chloride (PMBCl) led to
rearrangement of the 5-hydroxymethyl oxazolidinone to the
corresponding 5-hydroxy-[1,3]oxazinan-2-one,70 with concurrent
(67) (a) Badorrey, R.; Cativiela, C.; D´ıaz-de-Villegas, M. D.; D´ıez, R.; Ga´lvez,
J. A. Tetrahedron Lett. 2004, 45, 719-722. (b) Dhavale, D. D.; Jachak, S.
M.; Karche, N. P.; Trombini, C. Synlett 2004, 1549-1552. (c) Bernarki,
L.; Bonini, B. F.; Capito`, E.; Dessole, G.; Fochi, M.; Comes-Franchini,
M.; Ricci, A. Synlett 2003, 1778-1782. (d) Wipf, P.; Rector, S. R.;
Takahasi, H. J. Am. Chem. Soc. 2002, 124, 14848-14849. (e) Davies, S.
G.; Iwamoto, K.; Smethurst, C. A. P.; Smith, A. D.; Rodriguez-Solla, H.
Synlett 2002, 1146-1148. (f) Martin, S. F.; Humphrey, J. M.; Ali, A.; Hiller,
M. C. J. Am. Chem. Soc. 1999, 121, 866-867.
(68) AcOH is required for catalyst turnover; it is thought that AcOH protonates
the tertiary amine, thereby preventing catalyst deactivation that would occur
otherwise.
(69) The mixture of monomeric and dimeric products obtained from use of
Grubbs’ 2nd generation catalyst is believed to reflect a thermodynamic
distribution of products, because individually subjecting either monomer
119 or dimer 120 to the identical reaction conditions led to similar 1:3-4
ratio of the monomeric and dimeric adducts.
(70) For a related rearrangement that proceeds through an isocyanate, see: (a)
Tadanier, J.; Martin, J. R.; Hallas, R.; Rasmussen, R.; Grampovnik, D.;
Rosenbrook, W., Jr.; Arnold, W.; Schuber, E. Carbohydr. Res. 1981, 98,
11-23. Oxazinanone could be thermodynamically preferred to the corre-
sponding oxazolidinone; see: (b) Sadybakasov, B. K.; Ashirmatov, M. A.;
Afanas’ev, V. A.; Yu. Struchkov, T. Zh. Strukt. Khim. 1989, 30, 645-
650.
Having established a route to diene 118, we turned to execute
a ring-closing metathesis reaction to construct the 13-membered
macrocycle. Although not unprecedented,67 the ability to perform
ring-closing metathesis reactions in the presence of a basic
tertiary amine was expected to be a significant challenge,48,67
(64) Sungchul, J.; Gortler, L. B.; Waring, A.; Battisti, A.; Bank, S.; Closson,
W. D.; Wriede, P. J. Am. Chem. Soc. 1967, 89, 5311-5312.
(65) Cossy, J.; Aclinou, P. Tetrahedron Lett. 1990, 31, 7615-7618.
(66) Lin, Y-. T.; Houk, K. N. Tetrahedron Lett. 1985, 26, 2517-2520.
9
11998 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 27
protection of the secondary hydroxyl, to provide PMB ether
123 in 89% yield. Next, considerable experimentation was
carried out to optimize removal of the TIPS protective group.
Although conditions employing TBAF or HF were unsuccessful,
reaction of silyl ether 123 with TAS-F71 in DMA at 100 °C led
to clean desilylation. The tetrahydrooxazinone fragment of the
resulting intermediate was cleaved with KOH in EtOH at
90 °C to provide diamine diol 124 in 65% yield over the two
steps.
Figure 8. Strategies to fashion the skipped-triene containing macrocycle
of sarain A.
Scheme 28
Toward the Unsaturated Macrocycle of Sarain A: An
Unanticipated Skeletal Rearrangement. With diamine diol
124 in hand, we considered a number of strategies to address
the next roadblock en route to sarain A, namely, installation of
the skipped-triene containing macrocycle (124 f 125). Although
attempts to prepare the unsaturated macrocycle in simple model
systems had been described by Heathcock,16c no related studies
involving advanced synthetic intermediates bearing the sarain
A core skeleton or saturated macrocycle had been reported. Of
the many methods considered, the most attractive involved
Nozaki-Hiyama-Kishi,72 Stille,73 or Sonogashira74 macrocy-
clizations. Several potential substrates for these strategies are
depicted in Figure 8 (126, 127, and 128, respectively). As an
sp2-sp2 cross-coupling approach to the unsaturated macrocycle
of sarain A would be quite direct, and the use of intramolecular
Stille couplings to form macrocycles is well precedented,73 we
opted to pursue the synthesis of stannane 127 as the primary
objective.
Our initial efforts toward a Stille macrocyclization precursor
are shown in Scheme 28. Diamine diol 124 was allowed to react
with aldehyde 12943 in refluxing PhH to afford N,O-acetal 130
as a single stereoisomer.75 Although we had anticipated the
formation of a 6-membered oxazinane product, the structure of
oxazocane 130 was confirmed by 2-dimensional NMR studies.43
This result was thought to be inconsequential, since in the next
step, we hoped to reduce the N,O-acetal to deliver an N-alkylated
product.
In our first attempt to carry out the proposed reduction, N,O-
acetal 130 was allowed to react with NaCNBH3 and AcOH at
room-temperature. Although a reduction product was obtained,
detailed NMR investigation14c,d established that extensive
skeletal rearrangement had taken place to afford tetracycle 132.
Fortunately, the desired reduction could be accomplished with
i-Bu2AlH, providing the N-alkylated product 131 in 71% yield
from secondary amine 124.
To probe the mechanism of this intricate skeletal rearrange-
ment, diol 131 was treated with AcOH in a mixture of CH2Cl2
and MeCN at ambient temperatures. Under these relatively mild
conditions, designed to mimic those that caused rearrangement
but in the absence of a reducing agent, the identical tetracyclic
product 132 formed quantitatively.76 Cognizant that the rear-
(71) Scheidt, K. A.; Chen, H.; Follows, B. C.; Chemler, S. R.; Coffey, D. S.;
Roush, W. R. J. Org. Chem. 1998, 63, 6436-6437.
(72) For reviews of Nozaki-Hiyama-Kishi reactions, see: (a) Avalos, M.;
Babiano, R.; Cintas, P.; Jime´nez, J. L.; Palacios, J. C. Chem. Soc. ReV.
1999, 28, 169-177. (b) Fu¨rstner, A. Chem. ReV. 1999, 99, 991-1045. (c)
Cintas, P. Synthesis 1992, 238-257.
(73) For reviews of the intramolecular Stille coupling, see: (a) Duncton, M. A.
J.; Pattenden, G. J. Chem. Soc., Perkin Trans. 1 1999, 1235-1246. (b)
Pattenden, G.; Sinclair, D. J. J. Organomet. Chem. 2002, 653, 261-268.
(74) For a recent review, see: Chinchilla, R.; Najera, C. Chem. ReV. 2007, 107,
874-922.
1
(75) Relative configuration at the aminal stereocenter was determined by H-
(76) In the presence of AcOH, sarain A (1) does not undergo the analogous
1H NOESY experiments.
skeletal rearrangement.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11999

A R T I C L E S
Scheme 29
Becker et al.
substantial difficulty. However, it was found that oxidation of
alcohol 130 with 2-iodoxybenzoic acid (IBX)79a in DMSO
delivered the desired aldehyde in 67% yield. Subsequent reaction
of the aldehyde with â-stannylvinylmagnesium bromide 13780
under a variety of conditions led to mixtures of products. After
inspection, it became apparent that although stereocontrolled
addition of the Grignard reagent had occurred to generate the
syn â-alkoxy alcohol (dr ∼3:1), the major product of this
reaction (i.e., 136) suffered from loss of the vinyl iodide
functional group. In an effort to minimize the halogen-
magnesium exchange that likely gave rise to des-iodide 136,
substantial experimentation was carried out. However, after
exploring various solvents, nucleophiles (e.g., Li, Ce, and Zn
reagents), and substrates (e.g., vinyl bromide instead of vinyl
iodide), no substantial improvements were observed.
Given the difficulties encountered in our somewhat ambitious
approaches to a Stille macrocyclization precursor,81 we con-
sidered an alternative stepwise strategy that would ultimately
provide access to the atomic skeleton of sarain A (Scheme 30).
To initiate efforts, diamine diol 124 was condensed with butanal
13882 to afford oxazocane 139.75 With a simplified N,O-acetal
side chain in place, it was now possible to introduce the vinyl
stannane functional group by IBX oxidation,79a followed by
addition of Grignard reagent 137 to provide 140 (dr 3-4:1).
Although 140 formed as an inseparable mixture of epimers, it
was presumed that the major adduct had the desired S config-
uration at C8′, which would arise from chelation-controlled
addition of the Grignard reagent.83 Reaction of the diastereo-
meric mixture with TBAF facilitated desilylation to provide diol
141. Despite finding that selective oxidation of the primary
alcohol of 141 was not possible, we could advance material by
resorting to less attractive protective group manipulations. Thus,
double TES-protection of diol 141 furnished silyl ether 142.
Fortuitously, at this stage, it was possible to separate the mixture
of C8′ epimers by conventional column chromatography.
Isomerically pure 142 was then allowed to react with K2CO3
in MeOH to selectively cleave the primary TES protective
group. Dess-Martin oxidation of alcohol 143,84 followed by
Wittig reaction with phosphonium salt 144,85 then delivered the
rangement is acid-catalyzed, we postulate that the process occurs
by protonation of the pyrrolidine nitrogen (131 f 133), followed
by formation of intermediate aziridinium ion 134. Subsequent
attack of the primary alcohol would lead to tetrahydrofuran ring
formation, giving the observed rearranged product, 132. Al-
though aziridinium ion rearrangements are known,77 to the best
of our knowledge, this is the first example of an aziridinium
ion rearrangement in which the leaving group is a tertiary
amine.78 The facility with which aziridium ion 134 is generated
from diamine 133 stands in marked contrast to the rate at which
half-neutralized N,N′-dimethylpropanediamine exchanges methyl
groups in H2O by displacement of an ammonium salt by an
amine (t1/2 ) 1.6 × 106 days, at room-temperature).78b The near
perfect alignment of the nonbonded electron pair of tertiary
amine N1′ with the σ* orbital of the C3′-N1 bond in 133, is
undoubtedly responsible for the ease of the observed rearrange-
ment.
Successful Construction of the Unsaturated Macrocycle.
Aware that our late-stage intermediates could be prone to
skeletal rearrangement, particularly under acidic conditions, we
returned to the difficult task of assembling the skipped triene-
containing macrocycle of sarain A. Preliminary efforts to
elaborate N-alkylated diol 131 toward Stille macrocyclization
precursor 135 by selective alcohol oxidation methods or various
protecting group strategies were unfruitful (Scheme 29). As an
alternative, we reasoned that it could be advantageous to carry
N,O-acetal 130 forward in the synthesis, then defer its reduction
to a later stage. This strategy was attractive because N1 of N,O-
acetal 130 would be less basic, and thus less prone to skeletal
rearrangement under acidic conditions. Moreover, on an inter-
mediate such as 130, the two alcohols of precursor 124 would
be differentiated. Attempts to oxidize alcohol 130 to the
corresponding aldehyde, using oxidants thought to be compatible
with the basic tertiary amines,79 were initially met with
(79) Oxidation in the presence of basic amines is not straightforward and often
requires substantial experimentation before suitable conditions are discov-
ered. Several oxidation procedures commonly employed in our studies are
as follows. For a review of IBX oxidations, see: (a) Tohma, H.; Kita, Y.
AdV. Synth. Catal. 2004, 346, 111-124. For a relevant example involving
Swern oxidation, see: (b) Ashley, E. R.; Cruz, E. G.; Stoltz, B. M. J. Am.
Chem. Soc. 2003, 125, 15000-15001. For the Narasaka-Mukaiyama
oxidation, see: (c) Narasaka, K.; Morikawa, A.; Saigo, K.; Mukaiyama,
T. Bull. Chem. Soc. Jpn. 1977, 50, 2773-2776. For Parikh-Doering
oxidation, see: (d) Parikh, J. R.; Doering, W. E. J. Am. Chem. Soc. 1967,
89, 5505-5507. See also references 39 and 51a for Dess-Martin and
TPAP/NMO oxidations, respectively.
(80) Grignard reagent 137 was prepared from the corresponding lithium reagent
and MgBr2. For the preparation of this lithium reagent, see: D. Seyferth,
S. C. Vick, J. Organomet. Chem. 1978, 144, 1-12.
(81) In addition to skeletal rearrangement and loss of iodide, we have also
observed fragmentation of the pyrrolidine ring (e.g., forming i), presumably
through a â-elimination pathway.
(77) Aziridinium ions are occasionally formed under SN1-type conditions. (a)
For an aziridinium ion that was generated by trapping of a secondary
carbocation formed by Prins cyclization, see: Graham, M. A.; Wadsworth,
A. H.; Thorton-Pett, M.; Rayner, C. M. Chem. Commun. 2001, 966-967.
(b) For an aziridium ion that was formed by trapping of a quinone methide
intermediate, see: Shamma, M.; Nugent, J. F. Tetrahedron 1973, 29, 1265-
1272.
(78) (a) Fragmentation of protonated 1,2-ethylenediamine to an aziridinium ion
has been observed in the gas phase; see: Bouchoux, G.; Choret, N.; Penaud-
Berruyer, P.; Flammang, R. J. Phys. Chem. A 2001, 105, 9166-9177. (b)
N,N′-Dimethyl-1,3-propanediamine exchanges methyl groups slowly in
water in the presence of acid; see: Callahan, B. P.; Wolfenden, R. J. Am.
Chem. Soc. 2003, 125, 310-311.
(82) Taillier, C.; Gille, B.; Bellosta, V.; Cossy, J. J. Org. Chem. 2005, 70, 2097-
2108.
(83) For a pertinent review, see: Reetz, M. T. Acc. Chem. Res. 1993, 26, 462-
468.
9
12000 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Total Synthesis of (−)-Sarain A
A R T I C L E S
Scheme 30
Scheme 31
and (c) the desired zwitterionic product would likely be difficult
to handle in the laboratory, as was the case for sarain A during
its isolation.2 Several mild oxidation procedures were exam-
ined,79 which ultimately led to the identification of bicarbonate-
buffered Dess-Martin periodinane39,89 as the most effective
oxidation conditions. In this way, diamine alcohol 147 was
transformed to aldehyde 148 in 70-80% yield (Scheme 31).
The crude oxidized product, which was stable to both aqueous
workup and filtration through SiO2, was used in subsequent
transformations without scrupulous purification.
To complete the total synthesis of sarain A, we hypothesized
that a global deprotection would be optimal, since it would
minimize the handling and purification of sensitive late-stage
compounds. Inspired by Danishefsky and co-workers who were
able to remove three secondary PMB protecting groups simul-
taneously without disturbing a skipped triene,90 we questioned
if sarain A would be stable to excess TMSI.91 Notably, when a
sample of natural sarain A (1) was allowed to react with 50
equiv of TMSI in CH2Cl2 from 0 °C f room-temperature for
short reaction times, it was possible to recover pure sarain A
after SiO2 chromatography (Scheme 32). However, when key
synthetic intermediate 148 was exposed to the identical reaction
conditions, protective group cleavage was sluggish. At longer
reaction times, complex product mixtures were obtained, with
decomposition being partially attributed to the sensitive nature
of the skipped-triene, which was likely unstable to adventitious
HI.92
We conceded that global deprotection to arrive at sarain A
(1) could be exceptionally difficult, and in our first preparation
of the natural product, stepwise removal of the TES and PMB
protective groups would be acceptable. With the hope of
removing the TES protecting group, late-stage intermediate 148
was allowed to react under mildly acidic conditions involving
either AcOH, dilute HCl, or HF•pyr. Whereas the former two
conditions led to recovery of starting material, HF•pyr readily
promoted the desired deprotection in reaction times as short as
coveted Stille coupling substrate 145 in 76% yield over two
steps. Gratifyingly, upon reaction of stannyl iodide 145 with
catalytic Pd(PPh3)4 and excess LiCl86 at ambient temperatures
in THF, cyclization took place to assemble the 14-membered
triene ring of product 146.87 Reduction of the N,O-acetal of 146
with i-Bu2AlH delivered pentacyclic alcohol 147, a compound
that possesses the full skeleton of sarain A, in 64% yield from
stannyl iodide 145. It should be noted that alternative Stille
coupling conditions employing copper additives73 were exam-
ined, but found to be inferior in assembling the unsaturated
macrocycle.
Installation of the Tertiary Amine-Aldehyde Proximity
Interaction and Completion of the Total Synthesis of (-)-
Sarain A. To complete the total synthesis of sarain A, two major
tasks remained: installation of the tertiary amine-aldehyde
functional group by oxidation of the C2 alcohol and cleavage
of the two remaining protective groups to unmask the C7′,C8′-
diol. We suspected that the former of these tasks would be
exceptionally challenging considering: (a) the C2 alcohol is
neopentylic and rests in a sterically encumbered environment,
(b) the tertiary amine lies in close proximity to the C2 alcohol,88
(87) Notably, the C8′ epimer of substrate 145 fails to cyclize under the identical
reaction conditions.
(88) The proximal tertiary amine could intercept an activated alcohol derivative
prior to oxidation to afford quaternary ammonium salt ii. We postulate
that this ammonium salt formed during attempted Swern oxidation of
alcohol 147.
(84) Attempts to achieve sequential deprotection of the primary triethylsilyl ether
and oxidation by use of Swern oxidation conditions were unsuccessful;
for this procedure, see: Rodriguez, A.; Nomen, M.; Spur, B. W.; Godfroid,
J. J. Tetrahedron Lett. 1999, 40, 5161-5164.
(85) Nicolaou, K. C.; Ramphal, J. Y.; Abe, Y. Synthesis 1989, 898-901.
(86) For the use of LiCl in Stille couplings, see: Amator, C.; Jutand, A.; Suarez,
A. J. Am. Chem. Soc. 1993, 115, 9531-9541.
(89) Dess-Martin periodinane was freshly prepared using Schreiber’s procedure;
see: Meyer, S. D.; Schreiber, S. L. J. Org. Chem. 1994, 59, 7549-7552.
(90) Gordon, D. M.; Danishefsky, S. J. J. Am. Chem. Soc. 1992, 114, 659-
663.
(91) For the use of TMSI to remove benzylic protecting groups, see: Jung, M.
E.; Lyster, M. A. J. Org. Chem. 1977, 42, 3761-3764.
(92) HI addition adducts were detected by ESI-MS.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 12001

A R T I C L E S
Scheme 32
Becker et al.
able from an authentic sample of the natural product by NMR,
mass spectrometric, and chromatographic comparisons.93 Fur-
thermore, CD comparisons confirm that the absolute configu-
ration of (-)-sarain A (1) is as depicted in Scheme 33, consistent
with the proposal made by Cimino and co-workers.2d
Conclusion
In summary, the first total synthesis of the structurally unique
alkaloid (-)-sarain A (1) has been achieved. The enantioselec-
tive route reported herein features a number of key steps,
including: (a) an intermolecular Michael addition to build the
C4′-C3′-C7′ stereotriad, (b) an enoxysilane-N-sulfonyliminium
ion cyclization to assemble the diazatricycloundecance core and
set the C3 quaternary carbon center simultaneously, (c) a high-
yielding ring-closing metathesis reaction to construct the 13-
membered macrocycle, (d) a Stille macrocyclization to forge
the skipped triene-containing 14-membered ring, and (e) a late-
stage installation of the tertiary amine-aldehyde proximity
interaction. We hope this account of the various strategies
conceived, setbacks encountered, and solutions developed en
route to sarain A will showcase the challenges and excitement
of complex molecule synthesis.
Scheme 33
Acknowledgment. This manuscript is dedicated to Professor
Nelson J. Leonard. We thank the NIH Heart, Lung, and Blood
Institute (HL-25854) for generously supporting this research.
Graduate and postdoctoral fellowship support is gratefully
acknowledged as follows: to M.B. from Studienstiftung des
deutschen Volkes (German National Merit Foundation), to P.C.
from the Fonds Que´be´cois de la Recherche sur la Nature et les
Technologies, to C.J.D. from Bristol-Myers Squibb and the
Achievement Rewards for College Scientists Foundation, to
N.K.G. from the NIH (HL-80901 and GM-79922), to S.H. from
the Natural Science and Engineering Research Council of
Canada, to S.J. from the Humboldt Foundation, and to F.W.N
from the NIH (CA-76739). We thank Dr. John Greaves and
Dr. John Huckins for experimental assistance. Professor Jin Cha
is acknowledged for providing a sample of natural sarain A
obtained originally from Professor Cimino.
10 min to furnish alcohol 149 (Scheme 33). Direct exposure of
crude alcohol 149 to TMSI at 0 °C delivered the first synthetic
sample of sarain A. Unfortunately, yields were low and the
synthetic sarain A was contaminated with various byproducts,
which proved inseparable by benchtop chromatography, HPLC,
and SFC; indeed, purification of sarain A was not a trivial task
as the isolation chemists had forewarned.2 However, in a
unanticipated discovery, it was noticed that extended reaction
of intermediate 148 with HF•pyr led to the clean formation of
a new product, immediately recognized as sarain A (1). After
aqueous workup and purification on SiO2, synthetic sarain A
(1) was isolated in 50-60% yield and found to be indistinguish-
Supporting Information Available: Experimental details and
characterization data for select compounds, in addition to
spectral comparison for synthetic and natural (-)-sarain A. This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA074300T
(93) Comparisons were made after adding CD3CO2D; see reference 2c.
9
12002 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007