Synthesis of naturally occurring antitumor agents: Stereocontrolled synthesis of the azabicyclic ring system of the azinomycins

Article abstract of DOI:10.1021/ja992274w

Full details of the synthesis of the fully elaborated aziridino[1,2-a]pyrrolidine substructure 2 of the antitumor agents azinomycins A and B are reported. Stereoselective bromination of dehydroamino acid 4 provided control of olefin configuration in the f

Full text of DOI:10.1021/ja992274w

9088
J. Am. Chem. Soc. 1999, 121, 9088-9095
Synthesis of Naturally Occurring Antitumor Agents: Stereocontrolled
Synthesis of the Azabicyclic Ring System of the Azinomycins
Robert S. Coleman,* Jian-She Kong, and Thomas E. Richardson
Contribution from the Department of Chemistry and ComprehensiVe Cancer Center,
The Ohio State UniVersity, 100 West 18th AVenue, Columbus, Ohio 43210-1185
ReceiVed July 1, 1999
Abstract: Full details of the synthesis of the fully elaborated aziridino[1,2-a]pyrrolidine substructure 2 of the
antitumor agents azinomycins A and B are reported. Stereoselective bromination of dehydroamino acid 4
provided control of olefin configuration in the final product, as a consequence of a stereospecific cyclization
of aziridine 3 onto the proximal â-bromoacrylate, which effected pyrrolidine ring introduction. Dehydroamino
acid 4 was constructed by olefination of aldehyde 5 with a glycine-based phosphonate. Two complementary
synthetic routes to 2 are presented. In the first route, the selectively protected C12/C13 diol system of the
targets was introduced into starting structure 6 in a stereocontrolled manner using Brown’s (γ-alkoxyallyl)-
diisopinocampheylborane reagent system. Transient protection of the C12 hydroxyl group of 48 as the
trimethylsilyl ether was used to prevent C13 acetate migration prior to cyclization. Deprotection of the C12
hydroxyl following cyclization to the azabicyclic system afforded the extremely unstable core substructure 44,
which could not be isolated, but was characterized in situ. In the second route, the racemic γ-alkoxystannane
8 was added in a chelation-controlled manner to serinal derivative 9 under conditions of kinetic resolution for
introduction of the C12 and C13 stereogenic centers of the target. Phenylacetate and methoxyacetate esters
were used for C12 hydroxyl protection. This work represents the first synthesis of the intact core substructure
44 of this novel class of natural products.
Introduction
azinomycin B (ne´e carzinophilin)5 covalently cross-links native
DNA without prior activation. Studies on azinomycin/oligo-
nucleotide interactions by Armstrong and co-workers6 were
interpreted to show cross-link formation between the agent and
N7 of G and N7 of G or A within the major groove of DNA,
and these results were confirmed recently by Saito and co-
workers.7 To date, the mechanism of action of these agents
remains incompletely defined.8
The intricate and unusual structure, complex molecular
mechanism of action, and effective antitumor activity make the
azinomycins particularly attractive targets for synthetic efforts.
An additional compelling rationale lies in the construction of
structurally and functionally related agents for elucidation of
the details of covalent interaction of these agents with oligo-
nucleotides.9 This is made particularly urgent because of the
poor availability and instability of the agents.
The antitumor agents azinomycins A (1a) and B (1b) were
isolated in 1986 from cultures of Streptomyces griseofuscus
S42227.1 These agents possess an intricately functionalized
structure containing the unusual aziridino[1,2-a]pyrrolidine sys-
tem as part of a dehydroamino acid. This heterocyclic system
presents the most significant synthetic challenge of these natural
products since it possesses a significant proportion of the
stereogenic elements and contains highly reactive functional
groups that require careful protection and delicate timing of
introduction.
(2) Ishizeki, S.; Ohtsuka, M.; Irinoda, K.; Kukita, K.; Nagaoka, K.;
Nakashima, T. J. Antibiot. 1987, 40, 60. In vitro cytotoxicity: IC50 ) 0.07
µg/mL (1a) and 0.11 µg/mL (1b) against L5178Y cells. In vivo antitumor
activity: 193% ILS at 16 µg/kg 1b (3/7 survivors) against P388 leukemia;
161% ILS at 32 µg/kg 1b (5/8 survivors) against Erlich carcinoma. In the
same system, mitomycin C exhibited a 204% ILS at 1 mg/kg against P388
leukemia.
(3) Tomasz, M.; Lipman, R.; McGuinness, B. F.; Nakanishi, K. J. Am.
Chem. Soc. 1988, 110, 5892 and references therein. For a recent review,
see: Tomasz, M. Chem. Biol. 1995, 2, 575.
(4) Lown, J. W.; Majumdar, K. C. Can. J. Biochem. 1977, 55, 630.
(5) Azinomycin B is apparently identical to carzinophilin A, an antitumor
agent isolated in 1954 from S. sahachiroi: Hata, T.; Koga, F.; Sano, Y.;
Kanamori, K.; Matsumae, A.; Sugawara, R.; Hoshi, T.; Shimi, T.; Ito, S.;
Tomizawa, S. J. Antibiot. Ser. A. 1954, 7, 107.
(6) Armstrong, R. W.; Salvati, M. E.; Nguyen, M. J. Am. Chem. Soc.
1992, 114, 3144.
The azinomycins exhibit potent in vitro cytotoxic activity and
significant in vivo antitumor activity against P388 leukemia in
mice.2 However, detailed biological evaluation of these agents
has been hampered by chemical instability and poor availability
from natural sources. The presence of electrophilic epoxide and
aziridine rings suggests that the azinomycins act by covalent
alkylation and cross-linking of DNA, in a manner similar to
that of mitomycin C.3 Lown and Majumdar4 demonstrated that
(1) Nagaoka, K.; Matsumoto, M.; Oono, J.; Yokoi, K.; Ishizeki, S.;
Nakashima, T. J. Antibiot. 1986, 39, 1527. Yokoi, K.; Nagaoka, K.;
Nakashima, T. Chem. Pharm. Bull. 1986, 34, 4554.
(7) Fujiwara, T.; Saito, I.; Sugiyama, H. Tetrahedron Lett. 1999, 40, 315.
(8) For a recent review of DNA cross-linking agents, see: Rajski, S. R.;
Williams, R. M. Chem. ReV. 1998, 98, 2723.
10.1021/ja992274w CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/15/1999

Synthesis of Antitumor Agents: Azinomycins
J. Am. Chem. Soc., Vol. 121, No. 39, 1999 9089
While there has been a significant amount of synthetic activity
in the area,10-15 to date no total synthesis of these agents has
been reported, largely due to difficulties surrounding the
selectively acylated C12/C13 diol system. With the exception
of our work,11 there are no reports of azabicyclic ring systems
containing a differentiated C12/C13 diol system, nor are there
reports of systems containing a free C12 hydroxyl group.
Recently, Terashima et al.8j reported the synthesis of the C12/
C13 bis-benzyl ether of the natural products, although these
workers were unsuccessful in effecting either differentiation or
deprotection of the diol.
a removable C12 hydroxyl-protecting group using this route.
Both tert-butyldimethylsilyl and benzyl ethers proved recalci-
trant in attempts to remove them from late intermediates. This
alcohol-protecting group must be carefully chosen to allow
selective removal at a late stage of the synthesis of the aziridino-
[1,2-a]pyrrolidine ring system, in the presence of a diverse and
congested assemblage of other functional groups.
We recently reported the successful development of two
conceptually different routes to 2 that were based on the crotyl-
stannylation of L-serinal12 and on the use of a (γ-alkoxyallyl)-
diisopinocampheylborane reagent system13 for introduction of
the selectively acylated diol at the C12 and C13 stereogenic
centers. Herein, we present full details of our syntheses of the
fully elaborated aziridino[1,2-a]pyrrolidine substructure of the
azinomycins that deals successfully with all structural features
of this system, including the first reported introduction of the
selectively protected 1,2-diol of the agents.12,13 In the course
of our studies on this substructure, we have uncovered a potential
origin of the instability associated with the natural agents.
In common with our previous work, a key synthetic inter-
mediate en route to azabicyclic system 2 is dehydroamino acid
4, which serves as a direct precursor to vinyl bromide 3. In
studies on the diastereoselective bromination of dehydroamino
acids related to 4, we demonstrated effective stereocontrol in
the transformation of 4 to the desired E-vinyl bromide 3.16 This
proved to be a critical transformation for achieving introduction
of the C7-C8 tetrasubstituted E-olefin of the target molecules
since the cyclization of 3 f 2 was found to be stereospecific
and to occur with complete stereoselectivity. Aldehyde 5 serves
as a fully elaborated precursor to the dehydroamino acid 4 via
Wadsworth-Horner-Emmons olefination. Aldehyde 5 pos-
sesses the three stereogenic centers and selectively acetylated
diol of the target.
Synthetic Strategy
Synthetic challenges presented by these apparently simple
natural products and specifically by substructure 2 include: (1)
diastereocontrol in the introduction of the tetrasubstituted C7-
C8 E-double bond, (2) incorporation of the differentially
acetylated C12-C13 Vic-diol from a suitably protected precur-
sor, and (3) general difficulties surrounding the highly electro-
philic aziridine ring, particularly as part of the larger, densely
functionalized system.
Our strategy for the synthesis of the 1-azabicyclo[3.1.0]hexane
substructure is based on the cyclization of the aziridine of 3
onto a proximal (E)-â-bromoacrylate to form the pyrrolidine
ring of target 2.14,15 Our original synthesis14 used D-glucosamine
as a chiral starting material for introduction of the three
stereogenic centers of 2, but we were unsuccessful at installing
The basis of the first synthetic plan described herein was the
recog-
nition that an alkene could serve as a precursor to both the
aldehyde and aziridine of 5. Retrosynthetically, this gives rise
to pseudosymmetrical diene 6, wherein differentiation of the
syn-diol serves to permit the introduction of the appropriate
acylation pattern of the natural products and to provide a means
for differentiation of the two double bonds. 1,5-Hexadien-3,4-
diol 6 is available in enantiomerically pure form using Brown’s
(γ-alkoxyallyl)diisopinocampheylborane system.17
(9) For molecular modeling work on the azinomycins relevant to their
mechanism of DNA binding, see: Alcaro, S.; Coleman, R. S. J. Org. Chem.
1998, 63, 4620.
(10) (a) Bryant, H. J.; Dardonville, C. H.; Hodgkinson, T. J.; Shipman,
M.; Slawin, A. M. Synlett 1996, 10, 973. (b) Bryant, H. J.; Dardonville, C.
Y.; Hodgkinson, T. J.; Hursthouse, M. B.; Malik, K. M. A.; Shipman, M.
J. Chem. Soc., Perkin Trans. 1 1998, 1249. (c) Armstrong, R. W.; Tellew,
J. E.; Moran, E. J. Tetrahedron Lett. 1996, 37, 447. (d) Moran, E. J.; Tellew,
J. E.; Zhao, Z.; Armstrong, R. W. J. Org. Chem. 1993, 58, 7848. (e)
Armstrong, R. W.; Moran, E. J. J. Am. Chem. Soc. 1992, 114, 371. (f)
Combs, A. P.; Armstrong, R. W. Tetrahedron Lett. 1992, 33, 6419. (g)
Armstrong, R. W.; Tellew, J. E.; Moran, E. J. J. Org. Chem. 1992, 57,
2208. (h) Moran, E. J.; Armstrong, R. W. Tetrahedron Lett. 1991, 32, 3807.
(i) England, P.; Chun, K. H.; Moran, E. J.; Armstrong, R. W. Tetrahedron
Lett. 1990, 31, 2669. (j) Hashimoto, M.; Terashima, S. Heterocycles 1998,
47, 59. (k) Hashimoto, M.; Terashima, S. Tetrahedron Lett. 1994, 35, 9409.
(l) Hashimoto, M.; Terashima, S. Chem. Lett. 1994, 6, 1001. (m) Hashimoto,
M.; Matsumoto, M.; Yamada, K.; Terashima, S. Tetrahedron Lett. 1994,
35, 2207. (n) Hashimoto, M.; Yamada, K.; Terashima, S. Chem. Lett. 1992,
6, 975. (o) Konda, Y.; Machida, T.; Sasaki, T.; Takeda, K.; Takayanagi,
H.; Harigaya, Y. Chem. Pharm. Bull. 1994, 42, 285. (p) Ando, K.; Yamada,
T.; Shibuya, M. Heterocycles 1989, 29, 2209. (q) Shishido, K.; Omodani,
T.; Shibuya, M. J. Chem. Soc., Perkin Trans. 1 1992, 2053. (r) Shibuya,
M.; Terauchi, H. Tetrahedron Lett. 1987, 28, 2619. (s) Shibuya, M.
Tetrahedron Lett. 1983, 24, 1175.
Since Brown’s methodology produces 6 with an alkyl ether
on the C3 hydroxyl group (R ) CH2OCH3) and with the C4
hydroxyl group unprotected, it was ideally suited for our
purposes. By virtue of its selectivity for allylic alcohols, the
Sharpless asymmetric epoxidation reaction was the perfect
accompaniment to the Brown chemistry and was used for
differentiation of the two double bonds of 6. The resulting 5,6-
epoxide would then serve indirectly as a precursor to the
aziridine of 5. This meant that the C3 ether of 6 would be
(11) Coleman, R. S. Synlett 1998, 1031.
(12) Coleman, R. S.; Richardson, T. E.; Carpenter, A. J. J. Org. Chem.
1998, 63, 5738.
(13) Coleman, R. S.; Kong, J.-S. J. Am. Chem. Soc. 1998, 120, 3538.
(14) Coleman, R. S.; Carpenter, A. J. J. Org. Chem. 1992, 57, 5813.
(15) Coleman, R. S.; Carpenter, A. J. Tetrahedron 1997, 53, 16313.
(16) Coleman, R. S.; Carpenter, A. J. J. Org. Chem. 1993, 58, 4452.
(17) Brown, H. C.; Jadhav, P. K.; Bhat, K. S. J. Am. Chem. Soc. 1988,
110, 1535.

9090 J. Am. Chem. Soc., Vol. 121, No. 39, 1999
Coleman et al.
transformed to the C13 acetate ester of 5 and that a suitable
protecting group would have to be installed at the free C4
hydroxyl group of 6, which becomes C12 of 5. We found that
the a p-methoxybenzyl ether was an effective solution for
protection of the C12 hydroxyl group of 2, a seemingly trivial
problem whose solution had escaped us to this point. In the
end, we uncovered a significant degree of instability associated
with the free C12 hydroxyl group that may explain prior
difficulties we had in deprotecting the hydroxyl group at this
position.
The second synthetic approach to the core aziridino[1,2-a]-
pyrrolidine system was based on chelation-controlled addition
of a γ-alkoxycrotylstannane to serinal for the key C12-C13
bond construction. Marshall and co-workers18 demonstrated that
γ-alkoxycrotylstannanes (8, R′ ) SiMe2t-Bu, CH2OCH3) un-
dergo Lewis-acid promoted addition to R-aminoaldehydes with
syn-stereoselectivity at the newly formed bond.19 In the context
of the azinomycins, this strategy would produce 7 with the C12
position unprotected and would thereby permit the divergent
introduction of C12 hydroxyl-protecting groups (azinomycin
numbering throughout). In practice, we used phenylacetyl and
methoxyacetyl esters as C12 hydroxyl-protecting groups. Ap-
propriate stereoselection for the C11-C12 syn/C12-C13 syn-
diastereomer 7 in the addition of stannane 8 to aldehyde 9 is a
consequence of a chelated aldehyde (C11-C12 bond) and the
anti-SE′ transition state for crotylstannane addition (C12-C13
bond).
methylsilyl and benzyl ethers at this position, and a triethylsilyl
group was found to be too labile.15 After considering depro-
tection conditions that would be orthogonal with the reactivity
patterns of various late synthetic intermediates and anticipating
compatibility with pending transformations, we opted for the
p-methoxybenzyl (PMB) ether, which can be removed under
neutral, mildly oxidizing conditions. In addition, an ether-
protecting group was complementary with the concurrent
crotylstannane based synthesis (vida infra).
Alkylation of the remaining alcohol of 10 with sodium
hydride and p-methoxybenzyl bromide, afforded 11 (84%) and
occurred without rearrangement of the epoxide. Addition of
azide22 to the terminal carbon of the epoxide 11 provided a 74%
yield of primary azide 12. On larger scales this reaction was
difficult to force to completion and we would typically isolate
unreacted starting epoxide, which was resubjected to the reaction
conditions. Transformation of the azide to the amine by reduc-
tion with triphenylphosphine in a toluene/water mixture23 and
N-acylation of the resulting primary amine 13 with benzyl chlor-
oformate and triethylamine afforded carbamate 14 in quantitative
yield over the two-step procedure. Manipulation of the hydroxyl-
protecting and activating groups proceeded by acylation of the
free secondary alcohol of 14 with methanesulfonyl chloride in
the presence of triethylamine to afford mesylate 15 (96%), cleav-
age of the methoxymethyl acetal with methanolic HCl (74%),
and introduction of the azinomycin C13 acetate by standard
acylation with acetic anhydride and pyridine (99%) to afford
16. Acid-catalyzed cleavage of the acetal of 15 was accompanied
by a sometimes significant amount of p-methoxybenzyl ether
cleavage, which could be minimized by carefully monitoring
the reaction as it progressed. Cyclization of 16 to the aziridine
17 occurred upon low-temperature deprotonation of the car-
bamate of 16 with potassium tert-butoxide and effectively
provided the pivotal intermediate 17 (100%) in an overall yield
of >35% from 6. This compound possesses all of the function-
ality and protecting groups for elaboration to the azinomycin
core, including the essential C13 acetate ester and a readily
removable p-methoxybenzyl ether at the emergent C12 position.
Crotylstannane Route. Aldehyde (S)-924 was complexed
with magnesium bromide etherate at -20 °C followed by the
slow addition of crotylstannane (S)-8.25 Upon warming the
reaction mixture to 25 °C, stannane addition to the aldehyde
occurred to afford the selectively protected diol 7 in near quan-
Results
Hexadienediol Route. The sequence of transformations from
diol 6, prepared as shown in 66% yield (>95% ee) following
Brown et al.,17 to the key aldehyde 5, proceeded in greater than
35% yield for the eight-step conversion. The two olefins of 6
were easily differentiated by virtue of the allylically disposed
hydroxyl group using a Sharpless asymmetric epoxidation.20
Under standard conditions with the L-(+)-diisopropyl tartrate
catalyst, epoxide 10 was obtained uneventfully in 90% yield
and g98% enantiomeric excess. Preventing this reaction from
going totally to completion served to increase the enantiomeric
purity of the system by virtue of the kinetic resolution that can
occur during asymmetric epoxidations.21
At this juncture we were faced with the choice of a C12
hydroxyl-protecting group. We had gained considerable experi-
ence in this matter from our earlier studies, albeit without
discovery of a viable protection scheme for the 1,2-diol. We
had unsuccessfully examined the overly stable tert-butyldi-
(18) Marshall, J. A.; Seletsky, B. M.; Coan, P. S. J. Org. Chem. 1994,
59, 5139.
(19) For the syn-selective addition of a vinylzinc reagent to an R-ami-
noaldehyde, see: Coleman, R. S.; Carpenter, A. J. Tetrahedron Lett. 1992,
33, 1697.
(20) Johnson, R. A.; Sharpless, K. B. In ComprehensiVe Organic
Synthesis; Trost, B. M., Ley, S. V., Eds.; Pergamon Press: Elsmford, NY,
1991; Vol. 7, p 389.
(22) Behrens, C. H.; Ko, S. Y.; Sharpless, K. B.; Walker, F. J. J. Org.
Chem. 1985, 50, 5687.
(23) Knouzi, N.; Vaultier, M.; Carrie, R. J. Bull. Soc. Chim. Fr. 1985,
815.
(21) Martin, V. S.; Woodard, S. S.; Katsuki, T.; Yamada, Y.; Ikeda, M.;
Sharpless, K. B. J. Am. Chem. Soc. 1981, 103, 6237.
(24) Preparation of aldehyde 6 proceeded smoothly, following the
published protocol: Garner, P.; Park, J. M. Org. Synth. 1992, 70, 18.

Synthesis of Antitumor Agents: Azinomycins
J. Am. Chem. Soc., Vol. 121, No. 39, 1999 9091
protecting group that could be removed at a later stage of the
synthesis in the presence of a considerable number and variety
of other functional groups, particularly in the presence of the
C13 acetate. For this duty we chose an ester group for protection,
for the following reasons: (1) the ester linkage was compatible
with all projected transformations en route to the final product;
(2) there is a great deal of flexibility possible within the ester
family, particularly with respect to mechanism of and reagents
for cleavage; (3) alkyl ethers were impossible to install at C12
of 7 and later intermediates due to a competing cyclization of
the alkoxide onto the proximal benzyl carbamate to form the
corresponding oxazolidinone.
For C12 hydroxyl protection we selected an ester that could
be removed selectively as a consequence of a differential in
the rate of hydrolysis compared to an acetate ester, either
enzymatically or chemically. The phenylacetyl group was
selected from among several alternatives (e.g., hydrocinnamyl,
valeryl), because the benzyl group of a phenylacetyl group is
recognized by penicillin G acylase.27 A methoxyacetate was
selected as the alternate in anticipation of its heightened lability
relative to a simple acetate ester.28
The sterically encumbered C12 hydroxyl group of 7 was
unreactive toward forcing acylation conditions with acid chlo-
rides or anhydrides, so the N,O-acetonide was removed prior
to acylation. An additional motivation for early cleavage of the
N-acyl oxazolidine ring of 7 comes from the fact that these
systems exist as a mixture of slowly interconverting rotamers,
which seriously complicated analysis by NMR spectroscopy.
titative yield with >10:1 selectivity for the syn-diastereomer.
The only minor isomer that could be detected was the corre-
sponding Z-olefin, which was inconsequential to our purposes,
as we planned to cleave the double bond by ozonolysis.
Cleavage of the oxazolidine ring of 7 occurred upon warming
in THF in the presence of ethylene glycol and camphorsulfonic
acid. The diol 18 was acylated at the primary hydroxyl group
with methanesulfonyl chloride and triethylamine to afford 19.
The secondary hydroxyl group was protected as the ester (R2
) CH2Ph or CH2OCH3) by treatment with the corresponding
carboxylic acid in the presence of dicyclohexylcarbodiimide
(DCC) and (dimethylamino)pyridine (DMAP) to afford 20 and
21. Interchange of C13 hydroxyl-protecting groups was ac-
complished by removal of the silyl group from 21 or 22 with
hydrofluoric acid and acylation of the resulting alcohol with
acetic anhydride and triethylamine to afford 22 and 23. In six
high-yielding steps (57% overall yield), the advanced intermedi-
ates 22 and 23 were generated with complete control of absolute
stereochemistry and introduction of suitable protecting groups.
Final aziridine installation was achieved by treatment of
methanesulfonate 22 or iodide 24 with potassium tert-butoxide
at -43 °C afforded 25 and 26, respectively, in high yields.
Iodide 24 was obtained from 23 with sodium iodide; methane-
sulfonate 23 could not be induced to cyclize effectively.
Dehydroamino Acids. With the important aldehyde precur-
sors in hand and readily available in multigram quantities, we
proceeded along the established plan for elaboration to the
azinomycin core substructure. Installation of the dehydroamino
acid system was preceded by oxidative cleavage of the ter-
minal olefin of 17 using ozone with dimethyl sulfide workup
to afford in high yield the aldehyde 27. While not particularly
labile, this aldehyde was carried into the subsequent Wadsworth-
Horner-Emmons olefination without purification29 using the
Performing the addition to (S)-9 with 2.3 equiv of the racemic
stannane rac-8 effected useful levels of kinetic resolution (>10:1
S/R) and this obviated the tedious and expensive preparation of
the enantiomerically pure γ-alkoxystannane (S)-8.26 Separation
of the addition product from unreacted stannane 8 could be
achieved in subsequent chromatographic purifications. The
atom-economy based on consumed tin was a concern because
we were throwing away half of rac-8 in the kinetic resolution.
In the conversion of 8 + 9 f 7, comparing the racemic versus
enantiomerically pure γ-alkoxycrotylstannane 8, the overall atom
economy based on total n-Bu3SnH consumed is 6.3 mol of Sn
per mol of 7 for the racemic stannane compared to 14.4 mol of
Sn per mol of 7 for the enantiomerically pure stannane. Thus,
the kinetic resolution is more than twice as efficient, based on
total tin consumed. In addition, the expense of the reagents
necessary for the synthesis of enantiomerically pure (S)-8 further
increased the economic advantage of performing this reaction
in the kinetic reaction mode, particularly considering that this
was the first synthetic step in the total synthesis effort.
Following installation of the two stereogenic centers at the
emergent C12 and C13 positions, we needed to introduce a C12-
(27) Waldmann, H.; Heuser, A.; Reidel: A. Synlett 1994, 65. Waldmann,
H.; Sebastian, D. Chem. ReV. 1994, 94, 911. Fuganti, C.; Grasselli, P.; Servi,
S.; Lazzarini, A.; Casati, P. Tetrahedron 1988, 44, 2575. Fuganti, C.; Rosell,
C. M.; Servi, S.; Tagliani, A.; Terreni, M. Tetrahedron: Asymmetry 1992,
3, 383.
(28) Reese, C. B.; Stewart, J. C. M. Tetrahedron Lett. 1968, 4273. Reese,
C. B.; Stewart, J. C. M.; van Boom, J. H.; de Leeuw, H. P. M.; Nagel, J.;
de Rooy, J. F. M. J. Chem. Soc., Perkin Trans. 1 1975, 934.
(25) Marshall, J. A.; Welmaker, G. S.; Gung, B. W. J. Am. Chem. Soc.
1991, 113, 647.
(26) The preparation of enantiomerically pure γ-alkoxycrotylstannanes
requires the reduction of an unstable intermediate acylstannane with BINAL-
H, and the optimal reagent for preparation of the acylstannane from the
R-hydroxystannane was expensive azocarbonyl dipiperidine (ADD.; $13/
mmol).

9092 J. Am. Chem. Soc., Vol. 121, No. 39, 1999
Coleman et al.
tion of a C8-vinyl bromide into dehydroamino acids.16 We
had observed that systems of general structure 34 undergo
reaction with N-bromosuccinimide (NBS) at room tempera-
ture to produce the intermediate R-bromoimines 35, which
undergo tautomerization upon treatment with base to the
diastereomeric (E)- and (Z)-â-bromo-R,â-dehydroamino acids
(E)-36 and (Z)-36, respectively. Both the Z- and E-isomers of
dehydroamino acids 34 afforded the same ratio of isomeric vinyl
bromides.
N-acetyl glycine phosphonate30 and potassium tert-butoxide, to
afford olefin 28 as a >4:1 mixture of Z/E isomers in 60-70%
yield.
Similarly, ozonolysis of the double bond of 25 and 26
afforded aldehydes 29 and 30, respectively. Olefination with
the N-methoxycarbonyl or N-acetyl glycine phosphonates31
afforded dehydroamino acids 32 and 33, respectively, as a
mixture of Z/E isomers (typically 2.5:1) in modest yields under
carefully optimized reaction conditions with diisopropylethyl-
amine and lithium chloride.32 The ratio of olefin stereoisomers
is irrelevant, since both isomers converge to the same mixture
of stereoisomeric vinyl bromides in the subsequent bromination
reaction sequence. The use of potassium tert-butoxide in THF
or CH2Cl2 was less successful, and provided the olefin 32 in
yields typically below 25%.
These olefination reactions were particularly difficult because
of the presence of the proximal acetoxy group at C13. Under
carefully controlled reaction conditions the isolated yields were
more than satisfactory, and were justified given the delicacy of
C13-protecting group manipulations on more highly elaborate
systems. We had established quite definitively in earlier studies
the undesirability of manipulating diol-protecting groups once
the dehydroamino acid double bond had been installed.
Dehydroamino Acid Bromination. Defined reaction condi-
tions had been developed earlier for stereocontrolled installa-
Stereocontrol in this transformation was important since our
previous studies had demonstrated that the cyclization used for
pyrrolidine introduction was stereospecific.12-15 By setting olefin
stereochemistry at the stage of (E)-36, we were assured the
formation of the correct E-isomer of the final bicyclic product.
The undesired Z-vinyl bromides (Z)-36 are the thermodynami-
cally more stable isomers, and are typically formed from the
kinetic E-isomers (E)-36 by nucleophile-induced isomerization,
for example with 1,8-diazabicyclo[2.2.2]octane (DABCO). High
levels of stereoselectivity for the kinetic products (E)-36 were
obtained by treatment of the intermediate R-bromoimines 35
with sterically hindered bases such as 2,2,6,6-tetramethylpip-
eridine (TMP) or potassium tert-butoxide. In our earlier studies,
we demonstrated reaction conditions for the stereodiVergent
preparation of both E- and Z-vinyl bromide products 36 from
either diastereomeric E- or Z-olefin starting materials 34 with
acceptably high levels of diastereoselectivity.
When olefin 28 was treated with 1 equiv of N-bromosuccin-
imide (NBS) in CHCl3 at room temperature, a mixture of the
stereoisomeric R-bromoimines was obtained. Treatment of the
R-bromoimines with 2,2,6,6-tetramethylpiperidine effected tau-
tomerization to the desired vinyl bromide (E)-37 with >5:1 E/Z
(29) Lieberknecht, A.; Schmidt, J.; Stezowski, J. J. Tetrahedron Lett.
1991, 32, 2113.
(30) Schmidt, U.; Lieberknecht, A.; Wild, J. Synthesis 1984, 53.
(31) Zoller, U.; Ben-Ishai, D. Tetrahedron 1975, 31, 863.
(32) Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.;
Masamune, S.; Roush, W. R.; Sakai, T. Tetrahedron Lett. 1984, 25, 2183.
1
stereoselectivity as measured by NMR analysis of the crude
reaction mixture. Demonstration of stereochemistry was made
by nuclear Overhauser enhancement between the NH and
C13-H protons of (E)-37, and the lack of a similar enhancement

Synthesis of Antitumor Agents: Azinomycins
J. Am. Chem. Soc., Vol. 121, No. 39, 1999 9093
with the corresponding Z-isomer, which was isolated in minor
quantities from the bromination/tautomerization reaction of 28.
In addition, we made use of an extensive set of chemical shift
correlations generated during our previous work. In this cor-
relation, the allylic (C13) proton of the E-bromides consistently
resonated upfield of the same proton in the Z-isomer.15
notable divergence from the other system, wherein we obtained
g10:1 E-selectivity when sterically bulky bases were used. The
origin of this divergence is not known, but the C12 ester is
obvious to implicate. In practice, the undesired isomer (Z)-43
and (Z)-44 was used to develop subsequent synthetic steps, and
so was not entirely useless.
Azabicyclic Systems and C12 Hydroxyl Group Deprotec-
tions. The 9-fluorenylmethoxy carbamate was removed easily
from 39 by treatment with piperidine to afford the intermediate
aziridine 40, which cyclized at room temperature by displace-
ment of the vinylic bromide to afford aziridino[1,2-a]pyrrolidine
45 in good yield. We had previously demonstrated this cycliza-
tion reaction to be stereospecific,11 and the E-stereochemistry
of olefin 45 was confirmed by 1H NMR spectroscopy by
observation of a strong reciprocal NOE between the C13-H
and proximal NH, and by the characteristic chemical shift of
the C13-H and NH protons.11 A strong five-bond W-coupling
(J5 ) 1.1 Hz) was observed across the bicyclic system between
the C13-H and C10-Hendo of the aziridine. Unfortunately, we
could not remove the C12-PMB ether from 45 using 2,3-
dichloro-5,6-dicyano-1,4-benzoquinone (DDQ)37 without caus-
ing concomitant destruction of the aziridine ring system.
Normally, aziridine N-CO2Bn-protecting groups can be
easily removed using triethylsilane and a palladium chloride
catalyst,33,34 but for nonobvious reasons the benzyl carbamate
of 37 was resistant to these reaction conditions, and under
forcing conditions consumption of starting material was ob-
served without the production of isolable product(s). At best,
free aziridine 40 was produced as a minor product under these
conditions.
For this reason, we opted to proceed in the synthesis using a
different aziridine-protecting group. The 9-fluorenylmethoxy-
carbonyl (FMOC)-protecting group35 seemed potentially ideal,
being removed rapidly by treatment with a secondary amine
base. Retreating one step to intermediate 28, we could inter-
change the CO2Bn for an FMOC group by hydrogenolysis
followed by acylation of the aziridine nitrogen with FMOC-
OBt36 (Bt ) benzotriazolyl) to afford 38 in good yields.
Bromination of 38 proceeded without problem to afford (E)-39
with 12:1 diastereoselection.
In contrast to the p-methoxybenzyl-protected system, in the
bromination of phenylacetyl and methoxyacetyl-protected 32
and 33, we found that the intermediate R-bromoimines 41 and
42 produced by treatment of either (Z)- or (E)-32 or 33 with
N-bromosuccinimide underwent only a modestly stereoselective
base-promoted tautomerization using potassium tert-butoxide.
The desired bromides (E)-43 and (E)-44 was obtained in at best
a 3.5:1 ratio to the Z-isomer, but more typically 1:1. This is a
The C12 p-methoxybenzyl-protecting group could be removed
at the stage of 39 to afford the corresponding free alcohol 47 in
good yield, using standard literature conditions with DDQ in a
mixed aqueous/CHCl3 solvent system. Alcohol 47 was surpris-
ingly unstable to silica gel chromatography even over triethyl-
amine-deactivated silica, but it could be partially purified by
chromatography over Sephadex. Deprotection of the FMOC
(33) Birkofer, L.; Bierwirth, E.; Ritter, A. Chem. Ber. 1961, 94, 821.
(34) Coleman, R. S.; Shah, J. A. Synthesis 1999, 1399.
(35) Carpino, L. A.; Han, G. Y. J. Org. Chem. 1972, 37, 3404.
(36) Paquet, A. Can. J. Chem. 1982, 60, 976.
(37) Nakajima, N.; Abe, R.; Yonemitsu, O. Chem. Pharm. Bull. 1988,
36, 4244.

9094 J. Am. Chem. Soc., Vol. 121, No. 39, 1999
Coleman et al.
carbamate of 45 with piperidine afforded an intermediate free
aziridine, which cyclized under the deprotection reaction condi-
tions to afford an aziridino[1,2-a]pyrrolidine system, the
structure of which was assigned as 48 where the C13 acetate
had migrated to the free C12 alcohol. This was apparent in the
1H NMR spectrum of the product, which showed the diagnostic
C13-H upfield from its expected chemical shift. Presumably
this migration occurred prior to cyclization, under the basic
conditions used to remove the FMOC carbamate, as it seems
unlikely that acetate migration could occur across the trans-
diol system of the rigid five-membered ring. It was surprising
to obtain only a single regioisomeric acetate from this reaction,
but considering the instability of systems containing the free
C12 hydroxyl group (vida infra), it may be that 48 is the only
stable product of the mixture produced in the reaction.
From these studies, we learned that a free C12 hydroxyl group
would participate as the receiving partner in a thermodynami-
cally driven acetate migration reaction from the proximal C13
ester prior to cyclization, and that C12 ester deprotection must
await final closure of the pyrrolidine ring. In the ester-protected
system, triethylsilane mediated removal of the aziridine N-CO2-
Bn group of 43 cleanly afforded the corresponding free aziridine
53, which underwent cyclization upon warming in the presence
of N-methylmorpholine to afford the pyrrolidine 54. This product
was accompanied by a significant amount of byproduct tenta-
tively identified as the product formed from opening of the
aziridine ring by bromide. This byproduct could be suppressed
by performing the reaction in the presence of Dowex anion-
exchange resin in the carbonate form, which served both as a
base and bromide ion scavenger. The cyclization was found to
be stereospecific.
This migration problem was solved by a transient protection
of the C12 alcohol of 47 as the labile trimethylsilyl ether 49,
which was formed by treatment of 47 with neat N,O-bis-
trimethylsilylacetamide at 90 °C. Similarly, the corresponding
N-CO2Bn-protected system 50 (prepared from 37 by DDQ
oxidation) could be silylated to afford 51. Removal of the FMOC
carbamate from 49 with piperidine, or the benzyl carbamate
from 51 with triethylsilane and palladium chloride (this reaction
now worked well in the presence of a C12-trimethylsilyl ether),
and cyclization afforded the aziridino[1,2-a]pyrrolidine system
52, as a moderately stable, although isolable intermediate.
Deprotection of the C12 phenylacetyl ester of 54 could be
achieved by treatment with polymer-supported penicillin G
acylase in a mixed solvent system of acetonitrile/aqueous buffer.
Since systems such as 55 where the C12 hydroxyl group is
unprotected are not sufficiently stable to permit isolation, we
attempted to characterize 55 formed in situ using 1H NMR (500
MHz, 9:1 D2O/CD3CN). Under carefully optimized conditions,
the phenylacetate ester of 54 could be removed with an
approximate half-life of 2 h, and 55 could be observed as one
of several products of this reaction. Since neither isolation nor
further characterization of 55 could be achieved, we concluded
from the data that 55 was undergoing further reaction at rates
similar to its enzymatic production from 54.
We were unsuccessful at removing the C12 phenylacetate
ester prior to cyclization, which may be the result of the
sterically crowded environment around the ester carbonyl group.
A number of systems related to 43 were completely unreactive
even upon extended exposure to penicillin G acylase, whereas
the very early intermediate 24 was smoothly hydrolyzed to the
corresponding alcohol under these conditions (data not shown).
In contrast, the labile methoxyacetate ester of uncyclized (E)-
44 could be removed selectively by virtue of its enhanced
susceptibility to saponification using K2CO3 in methanol at 0
°C. Subsequent silylation of the (previously characterized) C12
alcohol with N,O-bis-trimethylsilylacetamide afforded 51, thereby
junctioning the two synthetic routes.
Mild deprotection of the C12 silyl ether of 52 with com-
mercial 70% HF‚pyridine in THF afforded the target substruc-
ture 46, which could not be isolated, but was characterized by
1H NMR and high resolution mass spectroscopy. It was
surprising that the presence of a free hydroxyl group introduced
such a significant degree of instability into the system, especially
considering the stable nature of preceding intermediates.
Repeated attempts to isolate 46 under extremely mild conditions
were unsuccessful, leading completely to unidentifiable products.
Even removing the reaction media in vacuo was sufficient to
cause significant decomposition. In the end, we were able to
obtain sufficient data from in situ 1H NMR in THF-d8 to provide
characterization of 46. The chemical shifts of the protons on
the azabicyclic ring agreed well with those reported for the
azinomycins.1
With the aim of removing a C12 methoxyacetate as the final
synthetic operation, cyclization of the methoxyacetate-protected
system (Z)- and (E)-44 was achieved uneventfully by depro-
tection with triethylsilane and palladium chloride to afford the

Synthesis of Antitumor Agents: Azinomycins
J. Am. Chem. Soc., Vol. 121, No. 39, 1999 9095
of the unstable character of the natural products. While we
would not have conjectured a priori that the C12 hydroxyl group
would have played a major role in stability issues, it seems clear
from the studies described herein that this functional group is
to a significant degree the cause. However, without clear
evidence on the reaction pathway by which the C12 hydroxyl
reacts, which we have been unable to obtain, we can offer no
rationale at this time as to why this group effects the stability
of the agents. The stability of more advanced systems remains
to be demonstrated.
corresponding free aziridines. Cyclization as before to the
bicyclic system (E)- and (Z)-56 occurred with complete ste-
reospecificity in modest yields.
To our disappointment, the behavior of the two isomers of
56 was remarkably different toward hydrolytic conditions. We
could achieve selective saponification of the more labile C12
methoxyacetate ester on the wrong Z-isomer using K2CO3 in
methanol at 0 °C, to afford (Z)-46, isomeric with the bicyclic
system of the natural products. Upon more extended exposure
to these reaction conditions, the C13 acetate ester was hydro-
lyzed to afford unstable diol (Z)-58. In contrast, with the correct
E-isomer of 56, the C13 acetate ester was slightly more labile
toward saponification, and the initial product formed was the
C13 alcohol (E)-57, which underwent a less rapid saponification
of the C12 methoxyacetate ester to afford diol (E)-58 over the
course of 1 h. In neither case was there a useful difference in
the rates of saponification of the two esters. It seemed that the
electronic factors favoring selective hydrolysis of the methoxy-
acetate were overriden by biases resident in the densely
functionalized bicyclic system.
With respect to the rather extensive amount in effort directed
toward protecting group issues for the 1,2-diol of the natural
products, we have been unable to devise a protecting group
scheme that permitted either the chemo- or regioselective
deprotection of the C12 hydroxyl group on an azabicyclic
system. In recourse, we turned to the labile trimethylsilyl ether
that was installed immediately prior to cyclization, and which
could be removed selectively as the final synthetic step to afford
the native azabicyclic system of the azinomycins.
We have described the first synthesis of the fully elaborated
core substructures 46 and 55 of the azinomycins, including a
description of a protecting group strategy for the selectively
acylated C12/C13 diol of the natural products. This synthesis
of 46 proceeds in 15 steps from diene 6 or 11-12 steps from
crotylstannane 8 and aldehyde 9 in a modest overall yield that
is reflective of the high degree of complexity and instability of
the target system. We have constructed, for the first time, a
system containing a free C12 hydroxyl group, and we have
placed this functional group in a key position in defining the
origin of the instability of the natural agents.
Acknowledgment. This work was supported by a grant from
the National Institutes of Health (CA-65875). We thank Dr.
Andrew J. Carpenter for his important contributions to the early
stages of the work on the stannane-based route12 and to earlier
work as cited. During the course of this work, R.S.C. was the
recipient of a Dreyfus Foundation Distinguished New Faculty
Award (1989-94), an American Cancer Society Junior Faculty
Research Award (1990-93), the American Cyanamid Young
Faculty Award (1993-96), and an Alfred P. Sloan Foundation
Research Fellowship (1995-98).
Supporting Information Available: Detailed experimental
protocols and characterization of synthetic intermediates (PDF).
This information is available free of charge via the Internet at
http://pubs.acs.org.
Conclusions
Our observations on the instability of the azinomycin core
substructure 46 and 55 may provide at least a partial explanation
JA992274W