Total synthesis of thiostrepton. Retrosynthetic analysis and construction of key building blocks

Article abstract of DOI:10.1021/ja0529337

The first phase of the total synthesis of thiostrepton (1), a highly complex thiopeptide antibiotic, is described. After a brief introduction to the target molecule and its structural motifs, it is shown that retrosynthetic analysis of thiostrepton reveals compounds 23, 24, 26, 28, and 29 as potential key building blocks for the projected total synthesis. Concise and stereoselective constructions of all these intermediates are then described. The synthesis of the dehydropiperidine core 28 was based on a biosynthetically inspired aza-Diels-Alder dimerization of an appropriate azadiene system, an approach that was initially plagued with several problems which were, however, resolved satisfactorily by systematic investigations. The quinaldic acid fragment 24 and the thiazoline-thiazole segment 26 were synthesized by a series of reactions that included asymmetric and other stereoselective processes. The dehydroalanine tail precursor 23 and the alanine equivalent 29 were also prepared from the appropriate amino acids. Finally, a method was developed for the direct coupling of the labile dehydropiperidine key building block 28 to the more advanced and stable peptide intermediate 27 through capture with the highly reactive alanine equivalent 67 under conditions that avoided the initially encountered destructive ring contraction process.

Full text of DOI:10.1021/ja0529337

Published on Web 07/19/2005
Total Synthesis of Thiostrepton. Retrosynthetic Analysis and
Construction of Key Building Blocks
K. C. Nicolaou,* Brian S. Safina, Mark Zak, Sang Hyup Lee, Marta Nevalainen,
Marco Bella, Anthony A. Estrada, Christian Funke, Fre´de´ric J. Ze´cri, and
Stephan Bulat
Contribution from the Department of Chemistry and The Skaggs Institute for Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and
Department of Chemistry and Biochemistry, UniVersity of California, San Diego,
9500 Gilman DriVe, La Jolla, California 92093
Received May 4, 2005; E-mail: kcn@scripps.edu
Abstract: The first phase of the total synthesis of thiostrepton (1), a highly complex thiopeptide antibiotic,
is described. After a brief introduction to the target molecule and its structural motifs, it is shown that
retrosynthetic analysis of thiostrepton reveals compounds 23, 24, 26, 28, and 29 as potential key building
blocks for the projected total synthesis. Concise and stereoselective constructions of all these intermediates
are then described. The synthesis of the dehydropiperidine core 28 was based on a biosynthetically inspired
aza-Diels-Alder dimerization of an appropriate azadiene system, an approach that was initially plagued
with several problems which were, however, resolved satisfactorily by systematic investigations. The
quinaldic acid fragment 24 and the thiazoline-thiazole segment 26 were synthesized by a series of reactions
that included asymmetric and other stereoselective processes. The dehydroalanine tail precursor 23 and
the alanine equivalent 29 were also prepared from the appropriate amino acids. Finally, a method was
developed for the direct coupling of the labile dehydropiperidine key building block 28 to the more advanced
and stable peptide intermediate 27 through capture with the highly reactive alanine equivalent 67 under
conditions that avoided the initially encountered destructive ring contraction process.
Introduction
by the proliferating bacteria. The antibiotic properties of 1
against Gram-positive bacteria have been traced to its binding
The thiopeptide antibiotics are a growing class of naturally
occurring substances with novel molecular architectures and
powerful antibacterial properties. Their mode of action almost
invariably involves inhibition of protein biosynthesis in bacteria.
Known also as thiazolyl peptides, these antibiotics share a
number of structural motifs, including several heterocycles (e.g.
a dehydropiperidine, a pyridine, oxazoles, thiazoles, indoles),
at least one macrocycle, and dehydroamino acid residues. Figure
1 displays a number of representative examples from this class
of compounds, whose chemistry, biology, and use in medicine
have been recently reviewed.1
to the 23S region of ribosomal RNA and protein L11, an
occurrence that inhibits the GTPase-dependent activities of the
50S ribosomal subunit.4 In addition to its antimicrobial activity,
1 exhibits significant activity against Plasmodium falciparum,5
the parasite responsible for the majority of human malaria, and
selective cytotoxicity against certain cancer cells.6 Furthermore,
a recent report attributed to 1, and its thiopeptide sibling
siomycin, potent immunosuppressive properties.7
Studies aimed at elucidation of the biosynthetic origins of 1
shined considerable light on the way nature assembles this
(2) Isolation: (a) Pagano, J. F.; Weinstein, M. J.; Stout, H. A.; Donovick, R.
Antibiot. Ann. 1955-1956, 554-559. (b) Vandeputte, J.; Dutcher, J. D.
Antibiot. Ann. 1955-1956, 560-561. (c) Steinberg, B. A.; Jambor, W. P.;
Suydam, L. O. Antibiot. Ann. 1955-1956, 562-565. Structure determi-
nation: (d) Bond, C. S.; Shaw, M. P.; Alphey, M. S.; Hunter, W. N. Acta
Crystallogr. 2001, D57, 755-758. (e) Anderson, B.; Hodgkin, D. C.;
Viswamitra, M. A. Nature 1970, 225, 233-235. (f) Hensens, O. D.; Albers-
Scho¨nberg, G. J. Antibiot. 1983, 36, 799-813. (g) Hensens, O. D.; Albers-
Scho¨nberg, G. J. Antibiot. 1983, 36, 814-831. (h) Hensens, O. D.; Albers-
Scho¨nberg, G. J. Antibiot. 1983, 36, 732-845.
Among members of the thiopeptide family of antibiotics,
thiostrepton (1) occupies a special place not only due to its
historical position and imposing structure, but also because of
its biology and medicinal value. First isolated in 1954 from
Streptomyces azureus, it was subsequently found in fermentation
extracts of Streptomyces hawaiiensis and Streptomyces lauren-
tii.2
(3) Dennis, S. M.; Nagaraja, T. G.; Dayton, A. D. Res. Vet. Sci. 1986, 41,
251-256.
Recognized as the flagship of the thiopeptide class of
antibiotics, 1 is used in animal health care as a topical antibiotic,3
its use in humans being limited by its low solubility and
bioavailability which lead to development of drug resistance
(4) Xing, Y.; Draper, D. E. Biochemistry 1996, 35, 1581-1588.
(5) (a) McConkey, G. A.; Rogers, M. J.; McCutchan, T. F. J. Biol. Chem.
1997, 272, 2046-2049. (b) Hardman, J. G.; Limbird, L. E.; Molinoff, P.
B.; Ruddon, R. W. A.; Gilman, G. Goodman & Gilman’s The Pharma-
cological Basis of Therapeutics, 9th ed.; The McGraw-Hill Co.: New York,
1996; pp 965-985.
(6) Jonghee, K. PCT Int. Appl. WO 2002066046, 2002.
(1) Bagley, M. C.; Dale, J. W.; Merritt, E. A.; Xiong, X. Chem. ReV. 2005,
105, 685-714.
(7) Ueno, M.; Furukawa, S.; Abe, F.; Ushioda, M.; Fujine, K.; Johki, S.; Hatori,
H.; Ueda, J. J. Antibiot. 2004, 57, 590-596.
9
10.1021/ja0529337 CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 11159-11175
11159

A R T I C L E S
Nicolaou et al.
Figure 1. Molecular structure of thiostrepton (1) and other selected members of the thiopeptide class of antibiotics.
Figure 2. Biogenetic origins of thiostrepton’s dehydropiperidine core.
natural product from readily available building blocks.8 Thus,
feeding experiments using isotopically labeled precursors
including cysteine, isoleucine, methionine, serine, threonine, and
tryptophan to nurture S. azureus ATCC 14921 or S. laurentii
ATCC 31255 demonstrated the amino acid origin of all
structural motifs of the molecule. Most intriguingly, these studies
pointed to the Diels-Alder origin of the dehydropiperidine
nucleus, as shown in Figure 2. Equally fascinating was the
pathway suggested for the biogenesis of the quinaldic acid
residue, which apparently begins with tryptophan and involves
methylation and oxidation followed by imine ring rupture and
expansion, epoxidation, and epoxide opening (see Figure 3).
Thiostrepton’s unique molecular structure is both stunningly
complex and highly sensitive. As shown in Figure 4, at its heart
lies a dehydropiperidine core serving as a lynchpin on which
stand the bis-dehydroalanine tail and the molecule’s two
macrocyclic domains, the 26-membered thiazoline-containing
ring and the 27-membered quinaldic acid-containing system.
This acid- and base-sensitive molecule includes within its
structure 10 rings, 17 stereogenic centers, 11 peptide bonds, an
imine functionality, a secondary amine group, and several sites
(8) (a) Mocek, U.; Zeng, Z.; O’Hagan, D.; Zhou, P.; Fan, L.-D. G.; Beale, J.
M.; Floss, H. G. J. Am. Chem. Soc. 1993, 115, 7992-8001. (b) Priestley,
N. D.; Smith, T. M.; Shipley, P. R.; Floss, H. G. Bioorg. Med. Chem. 1996,
4, 1135-1147.
9
11160 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Figure 3. Biogenetic origins of thiostrepton’s quinaldic acid domain.
formation. The sterically encumbered tertiary hydroxyl group
was left unprotected, for it was not expected to cause any
problems along the synthetic path. Another strategic decision
was made at this stage, namely the introduction of a triethylsilyl
(TES)-protected hydroxyl group at a position â to the thiazoline
ring with the anticipation that this group may give rise, at the
right moment, to the desired Z-double bond in conjugation to
the thiazoline ring. These retrosynthetic manipulations led us
from 1 to the advanced intermediate 22, whose conversion to
the target molecule (1) should require only two operational steps.
Rupture of the two amide bonds and one ester bond indicated
on structure 22 then led to the detachment of the masked bis-
dehydroalanine tail and disassembly of the quinaldic acid-
containing 27-membered ring of the molecule, leading, upon
suitable protections, to the key building blocks 23 and 24 and
(9) Synthetic studies from these laboratories: (a) Nicolaou, K. C.; Zak, M.;
Safina, B. S.; Lee, S. H.; Estrada, A. A. Angew. Chem., Int. Ed. 2004, 43,
5092-5097. (b) Nicolaou, K. C.; Safina, B. S.; Zak, M.; Estrada, A. A.;
Lee, S. H. Angew. Chem., Int. Ed. 2004, 43, 5087-5092. (c) Nicolaou, K.
C.; Nevalainen, M.; Zak, M.; Bulat, S.; Bella, M.; Safina, B. S. Angew.
Chem., Int. Ed. 2003, 42, 3418-3424. (d) Nicolaou, K. C.; Nevalainen,
M.; Safina, B. S.; Zak, M.; Bulat, S. Angew. Chem., Int. Ed. 2002, 41,
1941-1945. (e) Nicolaou, K. C.; Safina, B. S.; Funke, C.; Zak, M.; Ze´cri,
F. J. Angew. Chem., Int. Ed. 2002, 41, 1937-1940. Synthetic studies from
other laboratories: (f) Higashibayashi, S.; Kohno, M.; Goto, T.; Suzuki,
K.; Mori, T.; Hashimoto, K.; Nakata, M. Tetrahedron Lett. 2004, 45, 3707-
3712. (g) Higashibayashi, S.; Hashimoto, K.; Nakata, M. Tetrahedron Lett.
2002, 43, 105-110. For the total syntheses of simpler members of the
thiopeptide class of antibiotics, see: (h) Hughes, R. A.; Thompson, S. P.;
Alcaraz, L.; Moody, C. J. Chem. Commun. 2004, 946-948. (i) Higash-
ibayashi, S.; Mori, T.; Shinko, K.; Hashimoto, K.; Nakata, M. Heterocycles
2002, 57, 111-122. (j) Bagley, M. C.; Bashford, K. E.; Hesketh, C. L.;
Moody, C. J. J. Am. Chem. Soc. 2000, 122, 3301-3313. (k) Ciufolini, M.
A.; Shen, Y.-C. Org. Lett. 1999, 1, 1843-1846. (l) Okumura, K.; Nakamura,
Y.; Shin, C.-G. Bull. Chem. Soc. Jpn. 1999, 72, 1561-1569. (m) Okumura,
K.; Saito, H.; Shin, C.-G.; Yoshimura, J.; Umemura, K. Bull. Chem. Soc.
Jpn. 1998, 71, 1863-1870. (n) Okumura, K.; Ito, A.; Yoshioka, D.; Shin,
C.-G. Heterocycles 1998, 48, 1319-1324. (o) Moody, C. J.; Bagley, M.
C. Chem. Commun. 1998, 2049-2050. (p) Shin, C.-G.; Okumura, K.;
Shigekuni, M.; Nakamura, Y. Chem. Lett. 1998, 139-140. (q) Ciufolini,
M. A.; Shen, Y. C. J. Org. Chem. 1997, 62, 3804-3805. (r) Okumura, K.;
Shigekuni, M.; Nakamura, Y.; Shin, C.-G. Chem. Lett. 1996, 1025-1026.
(s) Nakamura, Y.; Shin, C.-G.; Umemura, K.; Yoshimura, J. Chem. Lett.
1992, 1005-1008. (t) Kelly T. R.; Jagoe, C. T.; Gu, Z. Tetrahedron Lett.
1991, 32, 4263-4266.
Figure 4. Special structural motifs of 1.
of unsaturation. It is the impressive totality and puzzling
connectivity of these features that make the total synthesis of 1
a most demanding and daunting task.9 In this and the following
article,10 we describe the details of the endeavor that led to the
accomplishment of this task.
Results and Discussion
1. Retrosynthetic Analysis. Given the above-described
structural motifs of 1, and recognizing its four sensitive and
challenging sites (see Figure 4), we chose the retrosynthetic
blueprint shown in Figure 5 as one most worthy of pursuit. Thus,
because of the labile nature of the dehydroalanine moieties, it
was decided to protect all three of them as phenylselenium
derivatives, masking devices that could unleash the required
olefinic bonds at the end of the synthesis upon simple (and
innocuous) oxidation/syn elimination. Equally important at this
early stage was deemed the protection of all four secondary
hydroxyl groups of 1 by tert-butyldimethylsilyl (TBS) ether
(10) Nicolaou, K. C.; Zak, M.; Safina, B. S.; Estrada, A. A.; Lee, S. H.;
Nevalainen, M. J. Am. Chem. Soc. 2005, 127, http://dx.doi.org/10.1021/
ja052934z.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11161

A R T I C L E S
Nicolaou et al.
Figure 5. Retrosynthetic analysis of 1 and definition of key building blocks 23, 24, 26, 28, and 29. TES, triethylsilyl; Boc, tert-butoxycarbonyl; TBS,
tert-butyldimethylsilyl; Fm, 9-fluorenylmethyl; Alloc, allyloxycarbonyl.
the advanced intermediate 25. The latter substrate (25) was then
retrosynthetically doubly cut further at the indicated amide bonds
to unravel the key intermediates 26 (appropriately modified to
an azide moiety representing the required amino group) and 27
(appropriately protected as shown). Finally, excision of the
alanine residue by rupturing the indicated amide bond on
structure 27 traced the origins of this intermediate to the key
building blocks dehydropiperidine 28 and the alanine equivalent
29. The exchange of methyl for ethyl esters in the last
retrosynthetic transform reflects only the ready access to the
ethyl ester variant of one of the starting materials, as shown in
Figure 6, in which the final and most daring retrosynthetic step
is revealed.
Inspired by the biomimetic proposal regarding the origin of
the dehydropiperidine domain of thiostrepton,8 we decided to
incorporate an aza-Diels-Alder dimerization strategy in our
plans for the laboratory construction of this core structure. Figure
6 outlines, in retrosynthetic format, this proposal. Thus, scission
of the key building block 28 through a retro-Diels-Alder
reaction/dimerization as indicated led to the monomeric unit
30, an azadiene system that was expected to arise, by H2S
abstraction, from thiazolidine 31. The latter compound (31) was
then easily traced to its components, aldehyde 32 and amino
thiol derivative 33. These starting materials were, in turn,
traceable to cysteine and threonine, as will become apparent in
the following section. In addition to the biosynthetic consider-
ations, supporting this hypothetical plan was a previous report11
involving generation and dimerization of a relevant azadiene
system, although these results, shown in Figure 7, by no means
assured the desired outcome in our present situation. Neverthe-
less, the aesthetic appeal and potential practical benefits of the
approach were enough to override any doubts we may have
had at the outset, and, therefore, we embarked on the daring
adventure with optimism.
2. Construction of Building Blocks. Having decided upon
the road map toward 1 as charted by the retrosynthetic analysis
depicted in Figures 5 and 6, we embarked on the construction
of the so-defined key building blocks required for the total
synthesis. Mindful of the previous report11 alluded to above
regarding the dimerization of azadiene 34 (Figure 7) to
dehydropiperidine system 35, whose existence proved only
transient on the way to enamine 36 and subsequently to aza-
Mannich product 37 and enamine 38, we entered this venture
with considerable trepidation, but confident that we could alter
(11) (a) Wulff, G.; Klinken, H. T. Tetrahedron 1992, 48, 5985-5990. (b) Wulff,
G.; Lindner, H. G.; Bo¨hnke, H.; Steigel, A.; Klinken, H. T. Liebigs Ann.
Chem. 1989, 527-531. (c) Wulff, G.; Bo¨hnke, H. Angew. Chem., Int. Ed.
Engl. 1986, 25, 90-92.
9
11162 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Figure 6. Retrosynthetic analysis of dehydropiperidine core 28 via a
biomimetically inspired aza-Diels-Alder reaction and definition of amino
thiol 33 and aldehyde 32 as starting materials.
the outcome to fit our purposes. Scheme 1 summarizes the
syntheses of amino thiol TFA salt 33 (for abbreviations of all
reagents and protecting groups used herein, see the legends in
the schemes) and aldehyde 32 from the known L-cysteine and
L-threonine derivatives 399l,m and 43,12 respectively, and their
condensation to afford the desired thiazolidine 31 (+31′). Thus,
L-cysteine derivative 39 was transformed to thioamide 41
through amide 40 by first coupling with N-hydroxysuccinimide
(HOSu) as facilitated by DCC, followed by reaction with
aqueous ammonia (39f40, 65% yield) and exposure to Lawes-
son’s reagent (40f41, 85% yield). The latter thioamide (41)
was then engaged with ethyl bromopyruvate (BrCH2COCO2-
Et) under basic conditions (KHCO3) in a reaction followed by
TFAA-assisted dehydration in the presence of pyridine (Hantz-
sch reaction)13 to afford, in 90% yield, thiazole 42. Exposure
of this thiazole (42) to TFA cleaved both its N-Boc and
acetonide groups, producing the TFA salt (33) of the desired
amino thiol. The synthesis of aldehyde 32 commenced with
L-threonine derivative 43, which was first converted to amide
44 (ClCO2Et, Et3N; then NH4OH, 78% yield) and then to
thioamide 45 (Lawesson’s reagent, 83% yield). The conversion
of thioamide 45 to thiazole 46 proceeded along similar lines as
for the transformation of 41 to 42 described above (BrCH2-
COCO2Et, NaHCO3; then TFAA, py) in 83% yield. Subsequent
reduction of thiazole ester 46 with DIBAL-H in toluene at -78
°C resulted in the formation of the required aldehyde 32 in 87%
yield. Finally, condensation of the two fragments, amino thiol
TFA salt 33 and aldehyde 32, in aqueous EtOH in the presence
of KHCO3 produced thiazolidine 31 (+31′) as a 1:1 mixture of
two diastereomers in 90% combined yield over two steps from
42.
Figure 7. (a) Precedent11 for the proposed biomimetically inspired aza-
Diels-Alder dimerization reaction to construct the dehydropiperine core
(28) of thiostrepton. (b) Comparison of product 38 from the literature-known
reaction11 and the desired dehydropiperidine core (28) of 1.
which was expected to be only a transient species (Scheme 2).
Indeed, upon exposure of thiazolidine 31 (+31′) to Ag2CO3 and
DBU in pyridine at -12 °C, the fleeting azadiene 30 was
apparently generated and dimerized according to the expected
Diels-Alder mode to afford, upon workup, three products:
bridged polycyclic imine system 49 (+49′) (63% yield, ca. 1:1
mixture of 5R,6S and 5S,6R diastereomers), dehydropiperidine
28 (+28′) (22% yield, ca. 1:1 mixture of 5R,6S and 5S,6R
diastereomers), and aldehyde 32 (20% yield). While the
structures of the dimeric products were evident from their NMR
spectra, their stereochemical assignments required X-ray crystal-
lographic analysis, a fortunate forthcoming event (vide infra).
These initial observations were delightful in that, unlike the
previous results of such reactions mentioned above,11 the
stereochemistry of the thiazole substituents on the dehydropi-
peridine ring was trans (see Figure 7), as desired for 1, but
obviously unsatisfactory with regard to the ratio of the products.
The major dimeric compound 49 (+49′) was apparently the
result of an initial imine-to-enamine rearrangement (path B)
within the Diels-Alder adduct 47 (+47′), followed by a
stereoselective aza-Mannich cyclization as shown on the result-
With thiazolidine 31 (+31′) in hand, we then proceeded to
explore its much anticipated conversion to azadiene 30 and the
outcome of the Diels-Alder dimerization of this intermediate,
(12) Kemp, D. S.; Carey, R. I. J. Org. Chem. 1989, 54, 3640-3646.
(13) (a) Kelly, R. C.; Ebhard, I.; Wicniensky, N. J. Org. Chem. 1986, 51, 4590-
4594. (b) Holzapfel, C.; Pettit, G. J. J. Org. Chem. 1985, 50, 2323-2327.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11163

A R T I C L E S
Nicolaou et al.
Scheme 1. Synthesis of Thiazolidine 31 (+31′)a
ment product 49 (+49′), which was now obtained only in trace
amounts. Furthermore, the thiazole aldehyde 32 was isolated
in 68% yield and could be recycled through incorporation into
the starting thiazolidine 31 (+31′).
At this juncture, a method was sought to determine the
stereochemistry of the two diastereomers obtained from the
Diels-Alder dimerization of azadiene 30 (Scheme 2). Since
the two pairs of compounds received at the end of this cascade
[i.e. the dehydropiperidine 28 (+28′) and the aza-Mannich
cyclization products 49 (+49′)] were interrelated, assignment
of the structure to one of the four compounds would reveal the
structures of all. The task was not trivial, however, due to the
remoteness of the starting stereocenters from the newly gener-
ated ones (i.e. separated by the full length of the thiazole ring).
Thus, compound 49 and its diastereomer 49′ (chromatographi-
cally inseparable mixture) were chosen for further manipulation
in the hope for a critical clue. Pleasantly, reduction of the aza-
Mannich mixture of products 49 + 49′ with NaCNBH3 in EtOH
in the presence of AcOH proceeded to afford exclusively and
stereoselectively two diastereomeric diamines, 51 and 51′, which
were separable by silica gel column chromatography (Scheme
3). Much to our pleasant surprise, the less polar of these isomers
(51) crystallized beautifully (EtOAc), and X-ray crystallographic
analysis of one of these crystals (see ORTEP drawing, Figure
8) revealed its absolute stereochemistry and, by extension, that
of all other compounds in the series. Before being able to
advance dehydropiperidine 28 (+28′) (chromatographically
inseparable ca. 1:1 mixture of 5R,6S and 5S,6R diastereomers)
toward 1, it was first necessary to complete the deceptively
standard task of capping its labile primary amine.
Despite the fact that our dehydropiperidine fragment 28
(+28′) was an inseparable mixture of 5R,6S and 5S,6R diaster-
eomers, we proceeded to the next stage, which called for
establishment of the required peptide chain onto the primary
amino group of the molecule. To this end, 28 (+28′) was reacted
with N-Alloc-L-alanine derivative 52 in the presence of HOAt
and EDC in the hope of obtaining amide 53 (+53′). Indeed,
not only was an amide formed under these conditions, but
luckily, the two expected diastereomers (84% combined yield,
ca. 1:1 ratio) were now chromatographically separable and,
therefore, obtained in pure form. Furthermore, removal of the
N-Boc-acetonide protecting device (TFA, 100%) from the more
polar isomer resulted in the formation of an amino alcohol
whose 1:1 L-tartaric acid salt crystallized upon standing in
EtOAc for a prolonged period of time (Figure 9). X-ray
crystallographic analysis of this crystalline compound (58) led
to a surprise, in that it revealed (see ORTEP drawing, Figure
9) the presence of a dehydropyrrolidine ring instead of the
assumed dehydropiperidine core with which we started (see
structures 57 and 57′, Scheme 4, and 58, Figure 9). A
mechanistic explanation for this ring contraction is shown in
Scheme 4. Thus, intramolecular attack of the primary amine
within 28 (+28′) onto the imine (upon activation of the latter
under the reaction conditions) could lead to the bicyclic core
shown in structure 54 (+54′), which may reverse back to the
starting material or proceed further to form the amino dehy-
dropyrrolidine 55 (+55′), the overall process representing a ring
contraction. The five-membered imine compound 55 (+55′) then
couples to the activated alanine derivative 56 to form the
observed amide product 57 (+57′).
a Reagents and conditions: (a) DCC (1.2 equiv), HOSu (1.2 equiv), THF,
25 °C, 3 h; then 28% NH4OHaq, EtOAc, 0 °C, 30 min, 65%; (b) Lawesson’s
reagent (0.5 equiv), DME, 25 °C, 12 h, 85%; (c) BrCH2COCO2Et (3.1
equiv), KHCO3 (8.4 equiv), DME, 0 °C, 5 min; then 25 °C, 24 h; then
TFAA (1.5 equiv), py (3.0 equiv), 0 °C, 2 h, 90%; (d) TFA:CH2Cl2 (1:1),
0 °C, 4 h; then EtOH:H2O (1:1), 25 °C, concentration in vacuo; (e) ClCO2Et
(1.1 equiv), Et3N (1.1 equiv), THF, 0 °C, 2.5 h; then 28% NH4OHaq, 0 °C,
2 h; then 25 °C, 16 h, 78%; (f) Lawesson’s reagent (0.6 equiv), benzene,
80 °C, 1 h, 83%; (g) BrCH2COCO2Et (3.0 equiv), NaHCO3 (8.0 equiv),
DME, 25 °C, 24 h; then TFAA (4.0 equiv), py (9.0 equiv), DME, 0 °C, 2
h, 83%; (h) DIBAL-H (2.4 equiv), PhMe, -78 °C, 2 h, 87%; (i) 32 (1.0
equiv), 33 (1.0 equiv), KHCO3 (3.0 equiv), EtOH:H2O (1:1), 0-25 °C, 16
h, 90% (over two steps from 42). Abbreviations: DCC, 1,3-dicyclohex-
ylcarbodiimide; DME, ethylene glycol dimethyl ether; HOSu, N-hy-
droxysuccinimide; TFA, trifluoroacetic acid; TFAA, trifluoroacetic an-
hydride; py, pyridine; p-TsOH, p-toluenesulfonic acid; DIBAL-H, diiso-
butylaluminum hydride.
ing structure 48 (+48′) in Scheme 2. In an effort to suppress
the formation of this undesired product (rare as it was) and
increase the yield of the desired dehydropiperidine compound
28 (+28′), which is the result of hydrolysis of the initially
formed Diels-Alder product 47 (+47′), we added benzylamine
to the reaction mixture right from the outset. The expectation
was that, as previously demonstrated,11a the thiazole aldehyde
component of the imine 47 (+47′) would rapidly engage the
added benzylamine and release the desired primary amine 28
(+28′) prior to the damaging aza-Mannich rearrangement.
Indeed, this turned out to be the case, for when benzylamine
(1.2 equiv) was added to the initial reaction mixture (conditions
B), the yield of the dehydropiperidine fragment 28 (+28′)
increased to 60% at the expense of the aza-Mannich rearrange-
9
11164 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Scheme 2. Hetero-Diels-Alder Dimerization Reaction: Tuning of Reaction Conditions To Produce Dehydropiperidine Core 28 (+28′) or
Aza-Mannich Adduct 49 (+49′) as the Major Product of the Reactiona
a Reagents and conditions: (a) conditions A involved Ag2CO3 (1.0 equiv), DBU (0.25 equiv, 0.047 M pyridine solution), -12 °C, 1 h, then H2O:EtOAc
(1:1), -12 to 25 °C, 1 h, 49 (+49′), 63%; 28 (+28′) 22%; 32, 20%; (b) conditions B involved Ag2CO3 (1.0 equiv), DBU (0.25 equiv, 0.047 M pyridine
solution), BnNH2 (1.2 equiv), -12 °C, 45 min, then H2O:EtOAc (1:1), -12 to 25 °C, 1 h, 49 (+49′), trace; 28 (+28′), 60%; 32, 68%. Abbreviations:
BnNH2, benzylamine; DBU, 1,8-diazabicyclo[5.4.0]undec-7-ene.
Faced with this new hurdle of the imine core contraction,
we set out to investigate conditions that would avoid it during
the crucial amide bond formation, capping the labile primary
amine within dehydropiperidine structure 28 (+28′). Our first
foray envisioned a reduction-oxidation sequence9g as a safe
expedition to the desired L-alanine coupled product 27 (+27′),
as shown in Scheme 5. Thus, dehydropiperidine 28 (+28′) was
stereoselectively14 reduced with NaCNBH3 in the presence of
AcOH, leading to piperidine intermediate 59 (+59′) in 63%
yield. The latter compound [59 (+59′)] entered into selective
(reacting only with its primary amino group) coupling with
N-Alloc-L-Ala (52) as facilitated by HATU, HOAt, and i-Pr2-
NEt to produce peptides 60 (+60′) in 88% yield, whose ethyl
ester groups were conveniently exchanged to methyl esters by
the action of n-Bu2SnO in MeOH at 70 °C (95% yield).15
Finally, the targeted imino peptides 27 (+27′) were generated
from their secondary amine counterparts by oxidation with
t-BuOCl at -78 °C, a reaction that proceeded, in 69% yield, to
afford the chromatographically separable products 27 (+27′).
Despite its success, however, this sequence failed to satisfy our
need for large amounts of the desired dehydropiperidine peptide
due to nonreproducibility on a large scale and its rather round-
about nature. We, therefore, embarked on a systematic search
for a direct approach to capturing the desired six-membered
imino amine with a suitably reactive L-alanine equivalent.
As part of our search, a variety of electrophiles were explored
as partners in the coupling reaction of dehydropiperidine
fragment 28 (+28′) under appropriate conditions. The results
listed in Table 1 encapsulate the essence of our observations,
which was that relatively large and less reactive electrophiles
lead to complete rearrangement of the dehydropiperidines to
the undesired five-membered ring compounds (entries 1 and
(14) The configuration of the newly formed stereocenter (i.e. C2) was assigned
(15) Baumhof, P.; Mazitschek, R.; Giannis, A. Angew. Chem., Int. Ed. 2001,
40, 3672-3674.
by NOE studies of a peptide-coupled derivative (see ref 9d).
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11165

A R T I C L E S
Nicolaou et al.
Scheme 3. Reduction of Imino Amine 49 (+49′) to Crystalline
a
Diamine 51 and Its Diastereomer 51′
a Reagents and conditions: (a) NaCNBH3 (2.0 equiv), AcOH:EtOH
(1:1), 25 °C, 2 h; (b) column chromatography, silica gel, CH2Cl2:EtOAc,
1:1, 51 (5R,6S) 40%, 51′ (5S,6R) 37%.
Figure 9. Conversion of L-alanine-coupled dehydropyrrolidine 57′ to amino
alcohol 58 and ORTEP drawing of 58 (5S,6R) obtained by X-ray
crystallographic analysis of its 1:1 salt with L-tartaric acid (L-tartaric acid
not shown). Reagents and conditions: (a) TFA:CH2Cl2 (1:1), 0-25 °C, 2
h, 100%; (b) L-tartaric acid (1.0 equiv), EtOAc.
process. After much experimentation, the first promising
coupling results were obtained during an attempt to engage the
free amine of dehydropiperidine core 28 (+28′) with allyl
chloroformate 64 in the presence of 4-DMAP and i-Pr2NEt in
CH2Cl2 (entry 3, Table 1). This reaction, which was presumed
to proceed via the allyl 4-N,N-dimethylpyridinium carbonate
adduct (see Table 1, entry 3), produced a mixture of six- and
five-membered N-Alloc-coupled imines (84% combined yield),
with the desired six-membered dehydropiperidine 65 (+65′)
predominating in a ca. 3:1 ratio. In an effort to improve the
selectivity of this reaction for the six-membered imine, the
coupling was attempted with allyl carbonate-HOAt reagent 6216
in acetonitrile (entry 2, Table 1). In this case, the selectivity for
the desired six-membered ring product was completely eroded,
leading exclusively, and in 87% yield, to the undesired five-
membered N-Alloc derivative 63 (+63′). The coupling reaction
was then carried out with allyl chloroformate (64) and i-Pr2-
NEt in CH2Cl2 (or THF) in the absence of 4-DMAP (entry 5,
Table 1), this time resulting in the exclusive formation of the
desired six-membered N-Alloc amide 65 (+65′) in 83% yield.
These coupling reactions (i.e. those depicted in entries 2, 3, and
5, Table 1) gave two important clues as to how to secure the
capture of the desired six-membered imine with an alanine
derivative. It was reasoned that chloride-containing electrophiles
were favoring capture of the six-membered imine, as did
conducting the reaction in less polar solvents such as CH2Cl2
Figure 8. X-ray crystallographic ORTEP drawing of analysis-derived
diamine 51 (5R,6S).
2), whereas smaller and more potent electrophiles react with
retention of the desired six-membered ring core structure (entries
5 and 6). Somewhere between lay the medium-sized electro-
philes, which give mixtures of the two types of products, as
seen in entries 3 and 4. These conclusions were not arrived at
randomly, but rather through a logical sequence of experiments.
It should first be noted that, similar to the results of entry 1,
Table 1 (and Scheme 4), reaction of dehydropiperidine core 28
(+28′) with N-Alloc and other carbamate-protected L-alanine
substrates (i.e. Boc, Fmoc derivatives), in the presence of a
variety of standard coupling reagents (i.e. HATU, TBTU,
PyBOP, CIP), led to exclusive formation of the undesired five-
membered imino peptide II (+II′). Frustrated by the inability
to capture the six-membered imine with an alanine derivative,
we then attempted to rescue it with other electrophiles in an
effort to gain a better understanding of the imine contraction
(16) Hayakawa, Y.; Wakabayshi, S.; Kato, H.; Noyori, R. J. Am. Chem. Soc.
1990, 112, 1691-1696.
9
11166 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Scheme 4. Unexpected Six- to Five-Membered Imine Ring
Contraction during Attempted Coupling of N-Alloc-L-Ala (52) with
Dehydropiperidine Core 28 (+28′)a
Scheme 5. Reduction-Oxidation Route to Dehydropiperidine
Peptide Fragment 27 (+27′)a
a Reagents and conditions: (a) NaCNBH3 (2.0 equiv), EtOH:AcOH
(4:1), 0 °C, 1 h, 63%; (b) N-Alloc-L-Ala-OH (52) (5.0 equiv), HOAt (5.0
equiv), HATU (5.0 equiv), i-Pr2NEt (10.0 equiv), DMF, 0-25 °C, 48 h,
88%; (c) n-Bu2SnO (3.0 equiv), MeOH, 70 °C, 6 h, 95%; (d) t-BuOCl (2.0
equiv), THF, -78 °C, 20 min; then 4-DMAP (0.2 equiv), Et3N (5.0 equiv),
-78 to 25 °C, 80 min, 38% (27) +31% (27′). Abbreviations: HATU, O-(7-
azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate;
4-DMAP, 4-dimethylaminopyridine.
diazo-transfer reaction of L-alanine17 and (ii) exposure of the
resulting N-azido-L-alanine derivative to oxalyl chloride] in the
presence of Et3N in THF, exclusive formation of the coveted
six-membered alanine coupled imine product 68 (+68′) was
observed, in 70% yield. Thus, success was finally achieved in
securing direct and exclusive access to the desired dehydropi-
peridine peptide 68, albeit still as a 1:1 mixture with its 5S,6R-
diastereomer 68′. Separation of 68 from its undesired sibling,
68′, and definitive identification as the desired isomer (5R,6S)
were still lingering problems, however.
a Reagents and conditions: (a) i. N-Alloc-L-Ala-OH (52) (1.2 equiv),
HOAt (1.3 equiv), EDC (1.3 equiv), DMF, 0-25 °C, 24 h, 84% combined
yield; ii. separation of diastereomers (silica gel, Et2O:toluene, 6:4-8:2).
Abbreviations: HOAt, 1-hydroxy-7-azabenzotriazole; EDC, 1-ethyl-(3-
dimethylaminopropyl)carbodiimide hydrochloride; DMF, N,N-dimethyl-
formamide; R*, activating group.
Figure 10 outlines a postulated mechanistic rationale for the
dependence of the coupling reaction on electrophile size and
reactivity. When the dehydropiperidine core 28 (+28′) is
depicted in its (presumed) preferred half-chair conformation, it
becomes apparent that its free amine is rather sterically
encumbered. Not only is it attached to a quaternary center, but
it is also shielded by the two thiazole rings that surround it. A
large electrophile (e.g. activated N-Alloc-L-ala 56) is apparently
unable to link with the free amine directly (path A), thus
allowing the six-membered imine core 28 (+28′) to undergo
contraction to its five-membered counterpart (path B), and
thereby producing the coupled five-membered imine dehydro-
pyrrolidine II (+II′). Conversely, a small and more aggressive
electrophile is capable of approaching close enough to the free
or THF. Using this intelligence information as a guide,
dehydropiperidine core 28 (+28′) was exposed to N-Alloc
alanine acid chloride 66 and i-Pr2NEt in THF (entry 4, Table
1), providing a mixture of six- and five-membered N-Alloc
alanine-coupled imines (76% combined yield) in which the
undesired five-membered dehydropyrrolidine 57 (+57′) was
predominating (ca. 2:1 ratio). With these results in hand, and
mindful of the smaller size of allyl chloroformate (64) as
compared to that of N-Alloc alanine acid chloride 66, we set
out to explore the reactivity of a smaller alanine equivalent that
could, potentially, mirror more closely the six-membered imine
coupling preference of allyl chloroformate. Indeed, when
dehydropiperidine core 28 (+28′) was combined with azido
alanine acid chloride 67 [prepared by (i) copper(II)-catalyzed
(17) (a) Nyffeler, P. T.; Liang, C.-H.; Koeller, K. M.; Wong, C.-H. J. Am. Chem.
Soc. 2002, 124, 10773-10778. (b) Alper, P. B.; Hung, S.-C.; Wong, C.-
H. Tetrahedron Lett. 1996, 37, 6029-6032.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11167

A R T I C L E S
Nicolaou et al.
Table 1. Selective Capture of the Five- or Six-Membered Imine: Electrophile Effects
amine of the dehydropiperidine core 28 (+28′) (path A) so as
to allow the desired amide bond formation, producing the desired
six-membered dehydropiperidine system I (+I′).
With the coupling of dehydropiperidine core 28 (+28′) and
L-ala equivalent 67 reliably affording the intact dehydropiperi-
dine peptide 68 (+68′), as shown in Scheme 6, the latter
compound was subjected to transesterification conditions (n-
Bu2SnO, MeOH, 75 °C),15 furnishing the bis-methyl ester
mixture of diastereomeric compounds 69 (+69′) in 80% yield.
Reduction of this mixture with SnCl2‚2H2O then produced the
9
11168 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Figure 10. Proposed mechanistic rationale for the imine contraction within 28 (+28′).
Scheme 6. Separation of Dehydropiperidine Diastereomers 70 and 70′ and Elucidation of Their C5/C6 Stereochemistry by Conversion to
Known Thiostrepton-Derived Compound 74a
a Reagents and conditions: (a) 67 (2.4 equiv), Et3N (8.0 equiv), THF, 0 °C, 1 h, 70%; (b) n-Bu2SnO (3.0 equiv), MeOH, 75 °C, 7.5 h, 80%; (c) i.
SnCl2‚2H2O (3.0 equiv), MeOH, H2O, 25 °C, 2 h; ii. silica gel, CH2Cl2:MeOH 98:2-96:4, 70 (5R,6S) 44%, 70′ (5S,6R) 38%; (d) Boc2O (5.0 equiv),
i-Pr2NEt (10.0 equiv), 4-DMAP (0.1 equiv), THF, 25 °C, 1 h, 61%; (e) n-Bu2SnO (5.0 equiv), EtOH, 65 °C, 5 h, 79%; (f) TFA:MeOH (1:1), 0 °C, 1 h, 54%
(+40% recovered 72); (g) (imid)2CO (3.0 equiv), 4-DMAP (4.0 equiv), DMF, 25 °C, 48 h, 81%. Boc2O, di-tert-butyl dicarbonate.
corresponding amines which, much to our delight, exhibited
sufficiently different chromatographic mobilities on silica gel
so as to allow their separation into pure samples, 70 (5R,6S,
44% yield) and 70′ (5S,6R, 38% yield). The stereochemical
identities of the two isomers were determined by conversion of
the less polar diastereomer (70, silica gel, CH2Cl2:MeOH 96:4,
Rf ) 0.34) to the known compound 74 (Scheme 6) derived from
a thiostrepton degradation product.9g The sequence from 70 to
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11169

A R T I C L E S
Nicolaou et al.
Scheme 7. Synthesis of Thiazole Fragment 89a
a Reagents and conditions: (a) LiOH (1.1 equiv), MeOH:H2O (1:1), reflux, 4 h, 90%; (b) KOH (1.0 equiv), (COCl)2 (5.0 equiv), Et2O, DMF (cat.), 0 °C,
4 h; (c) (-)-menthol (0.7 equiv), Et2O, 25 °C, 48 h, 70% (two steps); (d) AD-mix-â (1.5 equiv), MeSO2NH2 (1.0 equiv), t-BuOH:H2O (1:1), 0 °C, 24 h,
90%, 90:10 dr; (e) Me2C(OMe)2 (22 equiv), p-TsOH (0.05 equiv), 25 °C, 1 h, 100%; (f) DIBAL-H (2.2 equiv), CH2Cl2, -78 °C, 2 h, 90%; (g) DMP (1.1
equiv), NaHCO3 (2.0 equiv), CH2Cl2, 0 °C, 1 h; then 25 °C, 4 h; (h) BnNH2 (2.2 equiv), Yb(OTf)3 (0.2 equiv), 4 Å MS, CH2Cl2, 25 °C, 3 h; (i) TMSCN
(2.5 equiv), CH2Cl2, 25 °C, 2 h, 90% (three steps), >95:5 dr; (j) (Boc)2O (1.1 equiv), EtOAc, 20% Pd(OH)2 (0.2 equiv), H2, 25 °C, 8 h, 75%; (k) H2S,
Et3N:EtOH:py (1.7:1.7:1), sealed tube, 25 °C, 24 h, 91%; (l) BrCH2COCO2Et (3.0 equiv), KHCO3 (8.0 equiv), DME, 0 °C, 4 h; then 25 °C, 16 h; (m) TFAA
(4.0 equiv), py (8.0 equiv), 0 °C, 2 h; then 25 °C, 1 h, 82% (two steps); (n) NaOMe (3.2 equiv), MeOH, 0 °C, 5 h, 91%; (o) TFA:CH2Cl2:MeOH
(1.1:1.0:0.1), 0 °C, 1 h; then 25 °C, 2 h; (p) TBSCl (2.2 equiv), Et3N (3.3 equiv), CH2Cl2, 0 °C, 3 h, 87% (over two steps). Abbreviations: DMP, Dess-
Martin periodinane; TMSCN, trimethylsilyl cyanide; p-TsOH, p-toluenesulfonic acid; TBSCl, tert-butyldimethylsilyl chloride; TFAc, trifluoroacetyl.
74 proceeded smoothly as follows. N-Boc protection (Boc2O,
i-Pr2NEt, 4-DMAP) of the newly generated amine in 70
furnished derivative 71 in 61% yield. Exposure of this derivative
to n-Bu2SnO in EtOH at 65 °C15 caused transesterification of
both ester groups, furnishing bis-ethyl ester 72 (79% yield),
whose N-Boc-acetonide structural motif was broken down to
afford the corresponding N-Boc hydroxy compound 73 by
stirring in TFA:MeOH at 0 °C (54% yield, plus 40% recovered
starting material 72). Finally, exposure of compound 73 to
carbonyl diimidazole and 4-DMAP led to the targeted carbamate
derivative 74 (81% yield), whose spectral data were in complete
agreement with those reported9g for the naturally derived
substance, thus confirming the 5R,6S stereochemistry for
intermediate 70 and, therefore, its relatives.
the minor stereoisomer was removed by silica gel chromatog-
raphy to afford isomerically pure compound 80. This compound
was then treated with DIBAL-H, conditions which cleaved the
chiral auxiliary group and generated the required primary alcohol
(81, 90% yield), setting the stage for the pending three-step,
one-pot sequence to install the final stereogenic center of the
targeted thiazole unit. Thus, alcohol 81 was first oxidized with
DMP to afford the corresponding aldehyde, which was con-
densed with benzylamine in the presence of 4 Å molecular
sieves to give benzylimine 82, whose substrate-controlled
asymmetric Strecker-type reaction with TMSCN in the presence
of catalytic amounts of Yb(OTf)3 (Scheme 7) resulted in the
formation of nitrile 83 in 90% overall yield and >95:5
diastereoselectivity. The high stereoselectivity observed in this
process can be rationalized by the rendition of 82 in Scheme 7.
Hydrogenolysis of the benzyl group within nitrile 83 in the
presence of Boc2O produced N-Boc derivative 84 (75% yield),
which was then reacted with excess H2S and Et3N in a sealed
tube to afford thioamide 85 in 91% yield. Hantzsch reaction
(ethyl bromopyruvate, KHCO3) of the latter intermediate (85),
followed by dehydration of the incipient hydroxythiazoline with
TFAA and pyridine, led to thiazole trifluoroacetate 86 in 82%
overall yield. Exposure of the latter compound (86) to NaOMe
in MeOH served both to cleave the superfluous trifluoroacetate
group and to transesterify the ethyl ester, furnishing methyl ester
87 in 91% yield. Finally, upon TFA-induced removal of both
the N-Boc and acetonide from 87 to afford amino diol 88, the
secondary hydroxyl group so generated was selectively blocked
with TBS over the tertiary hydroxyl moiety, forming the desired
thiazole intermediate 89 in 87% overall yield. That this
intermediate possessed the correct stereochemistry as shown was
proven beyond doubt by an X-ray crystallographic analysis (see
ORTEP drawing, Figure 11) of the related N-Boc ethyl ester
derivative 90, prepared from compound 86 upon reaction with
NaOEt in EtOH (90% yield) and crystallization from EtOAc.
Having solved all the problems associated with the dehydro-
piperidine core fragment and secured an expedient route to the
key building block 70, we then turned our attention to the other
required fragments for our synthesis. We will first describe the
assembly of the thiazoline-thiazole fragment 26 (see Scheme
9), beginning with the stereoselective construction of the thiazole
subunit 89, as shown in Scheme 7. Thus, commercially available
angelic acid methyl ester (75) was first hydrolyzed with LiOH
in MeOH:H2O (1:1) to afford angelic acid18 (76, 90% yield),
which was then converted to its acid chloride (77, oxalyl
chloride) and whose reaction with (-)-menthol furnished angelic
acid menthyl ester 78 (70% over two steps). Exposure of the
latter compound (78) to Sharpless asymmetric dihydroxylation
conditions (AD-mix-â, MeSO2NH2) provided diol 79 in 90%
yield and ca. 90:10 diastereomeric ratio.19 This 1,2-diol (79)
was then converted to the corresponding acetonide (2,2-
dimethoxypropane, p-TsOH cat., 100% yield), at which point
(18) Buckles, R. E.; Mock, G. V. J. Org. Chem. 1950, 15, 680-684.
(19) (a) Torres-Valencia, J. M.; Cerda-Garc´ıa-Rojas, C. M.; Joseph-Nathan, P.
Tetrahedron: Asymmetry 1998, 9, 757-764. (b) Miyashita, M.; Shiina, I.;
Miyoshi, S.; Mukaiyama, T. Bull. Chem. Soc. Jpn. 1993, 66, 1516-1527.
9
11170 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Scheme 8. Synthesis of Thiazoline Fragment 104a
Figure 11. Synthesis of thiazole amino diol derivative 90 and proof of
relative stereochemistry by X-ray crystallographic analysis. Reagents and
conditions: (a) NaOEt (1.0 equiv), EtOH, 0 °C, 1 h, 90%. TFAc,
trifluoroacetyl.
a Reagents and conditions: (a) D-Ser-OMe‚HCl (1.0 equiv), i-Pr2NEt
(2.0 equiv), HOBt (1.2 equiv), EDC (1.2 equiv), CH2Cl2, 0 °C, 1 h; then
25 °C, 2 h, 93%; (b) TESCl (2.2 equiv), imidazole (3.0 equiv), DMF, 0
°C, 30 min; then 25 °C, 12 h, 81%; (c) Lawesson’s reagent (0.55 equiv),
benzene, reflux, 3 h, 83%; (d) Et2NH (6.5 equiv), DMF, 0 °C, 30 min;
then 25 °C, 30 min, 83%; (e) TFA:CH2Cl2 (1:1), 0 °C, 1.5 h; (f) TfN3 (3.0
equiv), Et3N (4.0 equiv), CuSO4‚5H2O (0.05 equiv), MeOH:H2O:CH2Cl2
(3.3:1:1), 25 °C, 1.5 h, 84% (two steps); (g) Me3SnOH (3.0 equiv), 1,2-
DCE, 80 °C, 3 h, 100%; (h) HATU (1.1 equiv), HOAt (1.1 equiv), i-Pr2NEt
(2.0 equiv), DMF, -20 °C, 20 min; then 0 °C, 20 min, 78%; (i) THF:
AcOH:H2O (10:3.3:1), 25 °C, 18 h, 60% (+17% recovered 101); (j) DAST
(1.2 equiv), CH2Cl2, -78 °C, 30 min, 88%; (k) Me3SnOH (3.0 equiv), 1,2-
dichloroethane, 80 °C, 1.5 h, 100%. Abbreviations: HOBt, 1-hydroxy-
benzotriazole hydrate; TESCl, triethylsilyl chloride; DAST, diethylamino-
sulfur trifluoride; TfN3, trifluoromethanesulfonyl azide; AcOH, acetic acid;
Fmoc, fluorenylmethoxycarbonyl; TES, triethylsilyl.
The second required fragment for the thiazoline-thiazole
domain, thiazoline subunit 104, was synthesized as shown in
Scheme 8. N-Fmoc-L-threonine 91 was coupled with D-serine
methyl ester hydrochloride 92 in the presence of HOBt, EDC,
and i-Pr2NEt to furnish dipeptide 93 in 93% yield. Protection
of the free hydroxyls of the latter compound (93) as TES ethers
(TESCl, imid., 81% yield) was followed by exposure of the
resulting amide derivative 94 to Lawesson’s reagent in refluxing
benzene, affording, selectively, thioamide 95 (83% yield).
Removal of the N-Fmoc group (Et2NH, 83% yield) from this
intermediate then revealed free amine 96. In parallel, L-threonine
derivative 9720 was converted to the free amine 98 by the action
of TFA, and then to azide 99, via a copper-catalyzed (CuSO4‚
5H2O) diazo-transfer reaction with TfN3,17 in 84% overall yield
for the two steps. The resulting azido ester 99 (the azide group
serving as a masking device for the eventually required amino
group) was then saponified, in quantitative yield, to its car-
boxylic acid counterpart 100 by the mild action of Me3SnOH,
conditions that did not cause any epimerization at the azide
group-bearing center. The two readily available building blocks
96 and 100 were then coupled together through the action of
HATU, HOAt, and i-Pr2NEt, furnishing tripeptide 101 (78%
yield), from which the primary alcohol-bound TES group
was selectively removed by exposure to AcOH:THF:H2O
(10:3.3:1) at ambient temperature (60% yield, plus 17%
recovered starting material 101). Activation of the resulting
hydroxy thioamide 102 with diethylaminosulfur trifluoride
(DAST) in CH2Cl2 at -78 °C led to thiazoline 103 in 88%
yield, which was then treated with Me3SnOH in 1,2-dichloro-
ethane at 80 °C to afford, in quantitative yield, carboxylic acid
104, impressively suffering no significant epimerization at any
of its vulnerable centers. The mildness and selectivity of this
Me3SnOH-based method for hydrolyzing esters is indeed
remarkable, for thiazoline 103 is highly sensitive and prone to
epimerization at no less than three of its stereogenic sites. The
scope and generality of this method have been investigated and
reported elsewhere.21
Scheme 9A depicts the coupling of thiazole amine 89 with
carboxylic acid 104, which proceeded smoothly under the
influence of HOBt and EDC to afford, in 73% yield, the
thiazoline-thiazole-coupled product 105. The methyl ester
group of the latter compound (105) was then gently, and
quantitatively, hydrolyzed with Me3SnOH21 to the corresponding
carboxylic acid, leading to the key building block 26.
Furthermore, and in order to ensure the feasibility of the
anticipated desilylation of its protected hydroxyl groups, frag-
ment 105 was treated with HF‚py in THF at ambient temperature
(Scheme 9B). Much to our delight, not only did all three silyl
(21) (a) Nicolaou, K. C.; Estrada, A. A.; Zak, M.; Lee, S. H.; Safina, B. S.
Angew. Chem., Int. Ed. 2005, 44, 1378-1382. (b) Furlan, R. L. E.; Mata,
E. G.; Mascaretti, O. A. J. Chem. Soc., Perkin Trans. 1 1998, 355-358.
(c) Furlan, R. L. E.; Mata, E. G.; Mascaretti, O. A.; Pena, C.; Coba, M. P.
Tetrahedron 1998, 54, 13023-13034. (d) Furlan, R. L. E.; Mascaretti, O.
A. Aldrichim. Acta 1997, 30, 55-69. (e) Furlan, R. L. E.; Mata, E. G.;
Mascaretti, O. A. Tetrahedron Lett. 1996, 37, 5229-5232.
(20) Muir, J. C.; Pattenden, G.; Thomas, R. M. Synthesis 1998, 613-618.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11171

A R T I C L E S
Nicolaou et al.
Scheme 9. (A) Synthesis of Thiazoline-Thiazole Subunit 26; (B)
Determination of Configuration of Thiazoline Olefin 107a
Scheme 10. Synthesis of the Bis-phenylselenium Tail Fragment
23a
a Reagents and conditions: (a) ethyl chloroformate (1.1 equiv), i-Pr2NEt
(1.1 equiv), THF, 0-25 °C, 1.5 h; then 28% NH4OHaq, 0-25 °C, 13.5 h,
89%; (b) TFA:CH2Cl2 (1:1), 0-25 °C, 1 h; (c) 108 (1.2 equiv), HOAt (1.1
equiv), EDC (1.1 equiv), i-Pr2NEt (1.5 equiv), DMF, 0 °C, 25 min; then
25 °C, 5 h, 82% (two steps); (d) TFA:CH2Cl2 (1:1), 0-25 °C, 1.5 h, 100%.
Ph, phenyl.
Scheme 11. Construction of Phenylselenyl Dipeptide 116a
a Reagents and conditions: (a) AllylOH (2.0 equiv), EDC (1.1 equiv),
4-DMAP (0.1 equiv), CH2Cl2, 0 °C, 5 min; then 25 °C, 3 h, 70%; (b) TFA:
CH2Cl2 (1:1), 25 °C, 2 h; (c) N-Boc-L-Ala-OH (114) (1.1 equiv), HOAt
(1.1 equiv), EDC (1.1 equiv), DMF, 25 °C, 2 h, 60% (two steps); (d) TFA:
CH2Cl2 (1:1), 25 °C, 30 min. AllylOH, allyl alcohol.
converted to its primary amide counterpart 109 (89% yield) by
exposure to NH4OH before removal of its N-Boc protecting
group with TFA to afford amine 110. The latter compound (110)
was then coupled with acid 108 under the influence of HOAt,
EDC, and i-Pr2NEt to furnish dipeptide 111 (82% yield for two
steps), from which the N-Boc group was cleaved by treatment
with TFA in CH2Cl2 (1:1), an operation that yielded the desired
building block 23 in quantitative yield.
The phenylselenyl dipeptide 116 required for incorporation
into the quinaldic acid domain of thiostrepton was synthesized
as shown in Scheme 11. Thus, carboxylic acid 10822 was
converted to its allyl ester derivative (112, allyl alcohol, EDC,
4-DMAP, 70% yield), from which the N-Boc was cleaved (TFA)
to reveal primary amine 113. The required coupling of the latter
compound (113) with N-Boc-L-alanine 114 was then success-
fully accomplished by the action of HOAt and EDC, leading to
dipeptide 115 in 60% yield over two steps. Finally, the N-Boc
group was dismantled from 115 with TFA, furnishing the desired
amine 116.
The synthesis of the quinaldic acid fragment 24 began with
quinoline-2-carboxylic acid 117, and proceeded as shown in
Scheme 12. Thus, the methyl ester 118 was produced in 99%
yield from 117 by reaction with SOCl2 in MeOH, and was
selectively reduced to pyridine system 119, in 60% yield, by
the action of H2 (55 psi) in the presence of PtO2 and TFA.23 In
the subsequent step, an acetyl radical (•COCH3), generated from
acetaldehyde, FeSO4‚7H2O, and H2O2,24 was attached to the
position para to the protonated (TFA) pyridine nitrogen atom
of 119, leading in 99% yield to methyl ketone 120. Asymmetric
a Reagents and conditions: (a) EDC (1.1 equiv), HOBt (1.1 equiv), DMF,
0 °C, 1 h, 73%; (b) Me3SnOH (6.0 equiv), 1,2-dichloroethane, 4 h, 80 °C,
100%; (c) HF‚py (excess), THF, 0-25 °C, 18 h, 78%.
groups of 105 come off in a nondestructive manner, but also
the expulsion of the TES group was accompanied by elimina-
tion, resulting in the formation of the Z-trisubstituted olefin 107.
The Z-geometry of the olefinic bond within 107 was established
by the observation of an NOE between the protons on its vinylic
methyl group and the nearby NH proton (as indicated on the
structure in Scheme 9B).
The synthesis of the dehydropiperidine tail equivalent, bis-
phenylselenyl dipeptide 23, was carried out as summarized in
Scheme 10. Thus, the phenylselenoalanine derivative 10822 was
activated by mixed anhydride formation (EtOCOCl) and, thence,
(22) Okeley, N. M.; Zhu, Y.; van der Donk, W. A. Org. Lett. 2000, 2, 3603-
3606.
(23) Vierhapper, F. W.; Eliel, E. L. J. Org. Chem. 1975, 40, 2729-2734.
(24) Priestley, N. G.; Smith, T. M.; Shipley, P. R.; Floss, H. G. Bioorg. Med.
Chem. 1996, 4, 1135-1147.
9
11172 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
Scheme 12. Synthesis of Quinaldic Acid Fragment 24 and Deprotected Compounds 138, 139, and 140a
a Reagents and conditions: (a) SOCl2 (3.0 equiv), MeOH, 60 °C, 6 h, 99%; (b) PtO2, H2, TFA (excess), 25 °C, 5 h, 60%; (c) MeCHO (85 equiv), H2O2
(10.2 equiv), FeSO4‚7H2O (0.13 equiv), TFA (1.2 equiv), 0 °C, 2 h, 99%; (d) 121 (0.04 equiv), B(OMe)3 (0.12 equiv), BH3‚SMe2 (1.9 equiv), THF, 0 °C,
30 min, 95%, 90% ee; (e) TBSOTf (1.1 equiv), Et3N (3.3 equiv), CH2Cl2, 0-25 °C, 3 h, 95%; (f) m-CPBA (2.5 equiv), CH2Cl2, 0-25 °C, 12 h; (g) TFAA
(2.5 equiv), CH2Cl2, 0-25 °C, 14 h; (h) K2CO3 (2 M aqueous solution):CH2Cl2 (3:1), 25 °C, 6 h, 75% (three steps); (i) Burgess reagent (1.2 equiv), THF,
25 °C, 20 min, concentrated in vacuo; then benzene, reflux, 1 h, 60%; (j) 126 (0.01 equiv), 4-Ph-py-N-oxide (0.1 equiv), NaOCl [0.79 M solution in
phosphate buffer adjusted to pH 11.5 with NaOH (2 M aqueous solution)]:CH2Cl2 (1:1)], 25 °C, 1 h, 82%, 87:13 dr; (k) NBS (1.1 equiv), AIBN (0.1 equiv),
CCl4, 80 °C, 30 min, 44% (+29% recovered starting material 128); (l) DBU (1.1 equiv), THF, 25 °C, 2 h, 96%; (m) H-L-Ile-OAllyl 131 (3.0 equiv), LiClO4
(5.0 equiv), MeCN, 60 °C, 22 h, 69%; (n) TBSOTf (3.0 equiv), i-Pr2NEt (5.0 equiv), THF, 0-25 °C, 3 h, 94%; (o) NaOH (3.2 equiv), H2O:MeOH:THF
(1:1:1.2), 0 °C, 30 min; then 25 °C, 5 h, 89%; (p) 2,4,6-trichlorobenzoyl chloride (2.0 equiv), Et3N (6.0 equiv), toluene, 25 °C, 4 h; then FmOH (3.0 equiv),
4-DMAP (0.1 equiv), 25 °C, 14 h, 64%; (q) PdCl2(PPh3)2 (0.1 equiv), n-Bu3SnH (1.1 equiv), CH2Cl2, 0 °C, 1 h, 100%; (r) 116 (1.5 equiv), HATU (1.1
equiv), HOAt (1.1 equiv), DMF, 25-40 °C, 1 h, sonication; then 25 °C, 2 h, 85%; (s) PdCl2(PPh3)2 (0.1 equiv), n-Bu3SnH (25 equiv), CH2Cl2, -45 °C, 5
min; then 0 °C, 30 min, 100%; (t) t-BuOOH (5-6 M in decane):CH2Cl2 (0.9:1), 0 °C, 5 min; then 25 °C, 2 h, 82%; (u) Et2NH:CH2Cl2 (1:10), 0 °C, 5 min;
then 25 °C, 3 h, 60%; (v) PdCl2(PPh3)2 (0.1 equiv), n-Bu3SnH (50 equiv), CH2Cl2, 0 °C, 5 min; then 25 °C, 30 min, 100%. Abbreviations: TBSOTf,
tert-butyldimethylsilyl trifluoromethanesulfonate; m-CPBA, m-chloroperoxybenzoic acid; AIBN, 2,2′-azabisisobutyronitrile; NBS, N-bromosuccinimide; FmOH,
9-fluorenemethanol; Fm, 9-fluorenylmethyl.
reduction of this ketone by a modified CBS procedure,25
employing chiral ligand 121, B(OMe)3, and BH3, afforded
enantiomerically enriched (90% ee) secondary alcohol 122 in
95% yield, which was then protected as a TBS ether (TBSOTf,
Et3N, 95% yield), leading to derivative 123. The latter compound
(123) was then converted to its hydroxylated derivative 124 by
a three-step Boekelheide-type sequence26 involving (a) N-
oxidation with m-CPBA, (b) trifluoroacetylation of the resulting
N-oxide with TFAA, and (c) K2CO3-induced trifluoroacetate
migration and hydrolysis. The newly installed hydroxyl group
in 124 was then eliminated in a dehydration process effected
by Burgess reagent,27 furnishing olefinic compound 125 (60%
yield), which was subsequently utilized in an asymmetric
epoxidation reaction mediated by the Katsuki catalyst 12628 to
produce epoxide 128 in 82% yield and ca. 87:13 diastereomeric
(26) Fontenas, C.; Bejan, E.; Haddou, H. A.; Galavoine, G. A. Synth. Commun.
(25) (a) For a pertinent review in this area, see: Corey, E. J.; Helal, C. J. Angew.
Chem., Int. Ed. 1998, 37, 1986-2012. (b) Masui, M.; Shioiri, T. Synlett
1997, 273-274.
1995, 25, 629-633.
(27) Burgess, E. M.; Penton, H. R.; Taylor, E. A. J. Org. Chem. 1973, 38, 26-
31.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11173

A R T I C L E S
Nicolaou et al.
Conclusion
ratio. It is noteworthy that attempted asymmetric epoxidation
of olefin 125 with the commercially available Jacobsen-Katsuki
catalyst 12729 led to low epoxide yields and poor diastereose-
lectivity in a reaction that was accompanied by significant
aromatization to the quinoline ring system. NBS-induced
bromination of epoxide 128 in the presence of AIBN (cat.)
allowed the formation of bromide 129 (diastereomeric mixture,
44% combined yield, plus 29% recovered starting material 128),
exposure of which to DBU at ambient temperature led to allylic
epoxide 130 in 96% yield. Epoxide 130 was then opened
regioselectively and, as expected, stereospecifically30 by the
amino group of L-isoleucine allyl ester 131 in the presence of
LiClO4 in acetonitrile at 60 °C to afford amino alcohol 132 in
69% yield. Treatment of this substance (132) with TBSOTf in
the presence of i-Pr2NEt then furnished bis-TBS derivative 133
in 94% yield. The methyl ester of 133 was then selectively
hydrolyzed by the action of NaOH in H2O:MeOH:THF (1:1:
1.2), leading to carboxylic acid 134 in 89% yield. The
9-fluorenylmethyl (Fm) group was then installed onto the newly
generated carboxyl moiety through a Yamaguchi-type31 esteri-
fication reaction (2,4,6-trichlorobenzoyl chloride, Et3N; then
FmOH, 4-DMAP, 64% yield), leading to diester 135, from
which the allyl group was cleaved using the n-Bu3SnH-PdCl2-
(PPh3)2 (cat.) method32 to give carboxylic acid 136 in quantita-
tive yield. Finally, coupling of carboxylic acid 136 with
dipeptide 116, as facilitated by HATU and HOAt, led to
conjugate 137 in 85% yield. The allyl ester of 137 was then
selectively cleaved employing the n-Bu3SnH-PdCl2(PPh3)2
(cat.) protocol,32 furnishing the targeted quinaldic acid key
building block 24 in quantitative yield.
Described herein is the chemistry that provided the foundation
for the eventual total synthesis of thiostrepton (1). Thus, efficient
synthetic routes to all required key building blocks (23, 24, 26,
28, and 29) are detailed. Of particular interest was the evolution
of the synthetic scheme toward the dehydropiperidine core of
the molecule. Thus, inspired by the proposed biosynthesis of
1, the described campaign to synthesize the dehydropiperidine
core 28 began with a hetero-Diels-Alder dimerization reaction
of azadiene system 30, but was met immediately with a
problem: the formed adduct suffered rapid aza-Mannich rear-
rangement to a bridged bicyclic system (49 + 49′). This problem
was solved by adding benzylamine to the reaction mixture,
which caused imine exchange, liberating the free amine before
the destructive rearrangement, and led predominantly to the
desired dehydropiperidine product 28 (+28′). This product (28,
5R,6S) was, however, formed as an inseparable 1:1 mixture with
its diastereomer (28′, 5S,6R). The trans relationship of the two
thiazoles on the imine ring was proven by NMR spectroscopy
and X-ray crystallographic analysis of a derivative (51) of the
aza-Mannich rearrangement product, but the relative stereo-
chemistry between the core centers and those across from the
thiazole ring could not be defined within each isomer due to
the lack of single compounds at this stage. In an effort to grow
the peptide chain upon the primary amino group, and in the
hope that the new derivatives would be chromatographically
separable, the Diels-Alder mixture of products (28 + 28′) was
then coupled with N-Alloc alanine. Indeed, not only were the
two formed diastereomers 57 and 57′ separable by silica gel
chromatography, but also a derivative of one of them crystallized
with one molecule of L-tataric acid. The X-ray crystallographic
analysis of this salt (58), while solving the relative stereochem-
istry problem, revealed another: a rearrangement of the amino-
imine from the desired six-membered ring to a dead-end, five-
membered amino-imine system. This newly arisen obstacle was
overcome upon a systematic exploration of a series of alanine
equivalents and coupling conditions, eventually leading to a
coupling procedure that suffered no rearrangement and which
furnished the desired alanine-bound dehydropiperidine core (70)
of thiostrepton. The problem of determining the relative
stereochemistry within this core, however, resurfaced since no
crystalline derivatives were in sight. This thorny problem was
solved by converting one of the coupled products (70) to a
degradation product (74) obtained from natural thiostrepton. It
is worth noting that, besides overcoming the initial hurdles
toward the dehydropiperidine core of thiostrepton, these inves-
tigations provided a wealth of novel skeleta for molecular
diversity construction relevant to chemical biology and medici-
nal chemistry studies.
Attempted selective deprotection of the Fm-protected carboxyl
end of 137 under basic conditions (e.g. Et2NH) led to partial
elimination of its phenylseleno group, resulting in a mixture of
products. However, treatment of this compound (137) with
excess t-BuOOH effected selective selenium oxidation and
spontaneous elimination of the resulting selenoxide to afford
dehydroalanine derivative 138 in 82% yield. The Fm group
could then be removed from the latter intermediate (138) by
the action of Et2NH, generating carboxylic acid 139 in 60%
yield. Alternatively, the same diester derivative 138 could be
converted to the regioisomeric carboxylic acid 140 by the action
of n-Bu3SnH in the presence of catalytic amounts of PdCl2-
(PPh3)2 in quantitative yield. These results proved useful to us
as we contemplated how to proceed over the treacherous ground
that surely would lie ahead.
(28) (a) Ito, K.; Yoshitake, M.; Katsuki, T. Tetrahedron 1996, 52, 3905-3920.
(b) Sasaki, H.; Irie, R.; Hamada, T.; Suzuki, K.; Katsuki, T. Tetrahedron
1994, 50, 11827-11838. (c) For a pertinent review in this area, see:
Katsuki, T. Curr. Org. Chem. 2001, 5, 663-678.
(29) (a) Jacobsen, E.; Zhang, W.; Muci, A. R.; Ecker, J. R.; Deng, L. J. Am.
Chem. Soc. 1991, 113, 7063-7064. (b) Irie, R.; Noda, K.; Ito, Y.;
Matsumoto, N.; Katsuki, T. Tetrahedron Lett. 1990, 31, 7345-7348.
(30) (a) Boyd, D. R.; Davies, R. J. H.; Hamilton, L.; McCullough, J. J.; Malone,
J. F.; Porter, H. P.; Smith, A. J. Org. Chem. 1994, 59, 984-990. (b)
Bushman, D. R.; Sayer, J. M.; Boyd, D. R.; Jerina, D. M. J. Am. Chem.
Soc. 1989, 111, 2688-2691.
(31) (a) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M. Bull.
Chem. Soc. Jpn. 1979, 52, 1989-1993. (b) Nicolaou, K. C.; Patron, A. P.;
Ajito, K.; Richter, P. K.; Khatuya, H.; Bertinato, P.; Miller, R. A.;
Tomaszewski, M. J. Chem. Eur. J. 1996, 2, 847-868.
(32) (a) de la Torre, B. G.; Torres, J. L.; Bardaj´ı, E.; Clape´s, P.; Xaus, N.; Jorba,
X.; Calvet, S.; Albericio, F.; Valencia, G. J. Chem. Soc., Chem. Commun.
1990, 965-967. (b) For a review on the use and removal of allylic
protecting groups, see: Guibe´, F. Tetrahedron 1998, 54, 2967-3042.
With the successful completion of the extendable dehydro-
piperidine core and the remaining fragments also at hand, we
were then poised for the next phase of the campaign: the
assembly of the synthesized key building blocks and elaboration
of the resulting advanced intermediates to thiostrepton. These
endeavors, accompanied as they were with their own complica-
tions, are described in the following article.10
Acknowledgment. We thank Drs. D. H. Huang, G. Siuzdak,
and R. Chadha for NMR spectroscopic, mass spectrometric, and
X-ray crystallographic assistance, respectively. Financial support
9
11174 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Total Synthesis of Thiostrepton
A R T I C L E S
for this work was provided by The Skaggs Institute for Chemical
Biology, the National Institutes of Health (U.S.A.), fellowships
from The Academy of Finland (to M.N.), The Skaggs Institute
for Research (to M.Z. and F.J.Z.), Bayer AG (to S.B.), and the
National Institutes of Health (to A.A.E.), and grants from
Abbott, Amgen, ArrayBiopharma, Boehringer-Ingelheim, Glaxo,
Hoffmann-LaRoche, DuPont, Merck, Novartis, Pfizer, and
Schering Plough.
Supporting Information Available: Experimental procedures
and compound characterization (PDF, CIF). This material is
available free of charge via the Internet at http://pubs.acs.org.
JA0529337
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11175