Mitomycin synthetic studies: Stereocontrolled and convergent synthesis of a fully elaborated aziridinomitosane

Article abstract of DOI:10.1021/jo048924i

Full details of a stereocontrolled and convergent synthetic route to 9a-desmethoxymitomycin A (1) are reported. The target molecule possesses the parent tetrahydropyrrolo[1,2-a]indole ring system characteristic of the mitomycin family of antitumor agents.

Full text of DOI:10.1021/jo048924i

Mitom ycin Syn th etic Stu d ies: Ster eocon tr olled a n d Con ver gen t
Syn th esis of a F u lly Ela bor a ted Azir id in om itosa n e
Robert S. Coleman,* Franc¸ois-Xavier Felpin, and Wei Chen
Department of Chemistry, The Ohio State University, 100 West 18th Avenue, Columbus, Ohio 43210
coleman@chemistry.ohio-state.edu
Received J une 25, 2004
Full details of a stereocontrolled and convergent synthetic route to 9a-desmethoxymitomycin A (1)
are reported. The target molecule possesses the parent tetrahydropyrrolo[1,2-a]indole ring system
characteristic of the mitomycin family of antitumor agents. The synthesis was based on the
diastereocontrolled addition of a fully elaborated cinnamylstannane to a pyrrolidine-based
N-acyliminium ion as the key convergent step, which resulted in the installation of the C9 and
C9a stereogenic centers.
In tr od u ction
Numerous efforts at the total synthesis of mitomycins
and related compounds have been reported, and novel
synthetic approaches are still being reported after nearly
half a century of extensive research.8 Starting with the
first total synthesis of mitomycins A and C reported in
1977 from the Kishi group,9 there have been several
subsequent successful total syntheses10 and an innumer-
able variety of synthetic approaches to the mitomycins
and the structurally11 and functionally12 related natural
products FR900482 and FR66979. The sustained and
vigorous interest in this family of natural products is
testament to their relevance as synthetic targets. The
densely and diversely functionalized ring system of these
natural products presents an attractive and challenging
objective for the practitioner of targeted organic synthe-
sis. In this article, we report the full details of a novel
synthetic route to the highly elaborated aziridinomito-
sane 113 that provides a strategically unique, syntheti-
cally efficient, and convergent entry to the tricyclic
tetrahydropyrrolo[1,2-a]indole ring system of the mito-
mycins.
Mitomycins A and B were isolated from Streptomyces
caespitosus by Hata and co-workers and were found to
possess potent antitumor and antibiotic activity.1 The
isolation of mitomycin C from the same fermentation
broth was reported subsequently. These antitumor agents
have been studied in detail with respect to mechanism
of action and clinical utility,2 and the synthetic literature
describing approaches to the total synthesis of the
mitomycins is extensive and well-appreciated.3
Mitomycin C is used clinically as a cancer chemothera-
peutic agent against a variety of solid tumors. This agent
forms covalent cross-links with duplex DNA4 by alkylat-
ing deoxyguanosine at the C2-amino group within the
DNA minor groove in 5′-d(CG)-3′ sequences.5,6 These
agents require bioreduction from the quinone to hydro-
quinone oxidation state to generate the active form of the
drug.7
The central premise of our synthetic approach to the
ring system of the mitomycins (2) was based on a
(1) Hata, T.; Hoshi, T.; Kanamori, K.; Matsumae, A.; Sano, Y.;
Shima, T.; Sugawara, R. J . Antibiot. Ser. A 1956, 9, 141.
(2) (a) Bradner, W. T. Cancer Treat. Rev. 2001, 27, 35. (b) Tomasz,
M.; Palom, Y. Pharmacol. Ther. 1997, 76, 73.
(3) Remers, W. A. Mitomycins and porfiromycin. In The Chemistry
of Antitumor Antibiotics; Wiley-Interscience: New York, 1979; pp 221-
276.
(4) Iyer, V. N.; Szybalski, W. Proc. Natl. Acad. Sci. U.S.A. 1963,
50, 355.
(5) (a) Tomasz, M.; Lipman, R.; Chowdary, D.; Pawlak, J .; Verdine,
G. L.; Nakanishi, K. Science 1987, 235, 1204. (b) Tomasz, M.; Lipman,
R.; McGuinness, B. F.; Nakanashi, K. J . Am. Chem. Soc. 1988, 110,
5892.
(8) (a) Coleman, R. S. Curr. Opin. Drug Discovery Dev. 2001, 4, 435.
(b) Danishefsky, S. J .; Schkeryantz, J . M. Synlett 1995, 475. (c)
Fukuyama, T.; Yang, L. Total Synthesis of Mitomycins. In Studies in
Natural Products Chemistry; Atta-ur-Rahman, Ed.; Elsevier: New
York, 1993; pp 433-471.
(9) (a) Fukuyama, T.; Nakatsubo, F.; Cocuzza, A. J .; Kishi, Y.
Tetrahedron Lett. 1977, 4295. (b) For a review of work from the Kishi
group, see: Kishi, Y. J . Nat. Prod. 1979, 42, 9388.
(10) (a) Fukuyama, T.; Yang, L. J . Am. Chem. Soc. 1987, 109, 7881.
(b) Fukuyama, T.; Yang, L. J . Am. Chem. Soc. 1989, 111, 8303.
(11) (a) Uchida, I.; Takase, S.; Kayakiri, H.; Kiyoto, S.; Hashimoto,
M.; Tada, T.; Koda, S.; Morimoto, Y. J . Am. Chem. Soc. 1987, 109,
4108. (b) Shimomura, K.; Hirai, O.; Mizota, T.; Matsumoto, S.; Mori,
J .; Shibayama, F.; Kikuchi, H. J . Antibiot. 1987, 40, 600.
(6) Mallard, J . T.; Weidner, M. F.; Raucher, S.; Hopkins, P. B. J .
Am. Chem. Soc. 1990, 112, 3637.
(7) Iyer, V. N.; Szybalski, W. Science 1964, 145, 55.
10.1021/jo048924i CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/23/2004
J . Org. Chem. 2004, 69, 7309-7316
7309

Coleman et al.
proposed diastereofacial selective addition of cinnamyl-
stannane 4 to a pyrrolidine-derived N-acyliminium ion
5 for introduction of the C9 and C9a stereogenic centers
in the addition product 3. In our preliminary studies on
the development of this conceptually simple strategy,14
we found that the silyl enol ether 6 was more effective
than stannane 4 as a nucleophilic partner in the addition
to iminium ion 5. The key bond could be formed by the
addition of 6 to 5 to provide 3 in near quantitative yield
with a favorably disposed 6:1 ratio of diastereomers at
the newly formed C9 and C9a stereogenic centers. Final
cyclization of the B-ring was achieved by a Buchwald-
Hartwig coupling15 of the pyrrolidine nitrogen with the
aryl bromide. Determination of relative stereochemistry
of 2 was made by detailed nuclear Overhauser effect
studies.
spite validating the feasibility and stereochemical out-
come of this convergent bond construction step in our
preliminary studies, there remained several important
issues with respect to successfully implementing the
proposed synthetic strategy for construction of fully
elaborated systems:
1. At what stage of the synthesis would the aziridine
be introduced and by which array of functional groups X
and Y in 7?
2. Could we formulate a viable and orthogonal protect-
ing group scheme for the various oxygen- and nitrogen-
based functional groups in both 8 and 9?
3. What would be the effect of substituents X and Y in
9 on the diastereofacial selectivity of the allylstannane
addition reaction?
Retrosynthetically, mapping the mitomycin aziridino-
[2,3-c]pyrrolidine C-ring of target 1 onto an appropriate
iminium ion precursor led us to three potential pyrroli-
dine systems bearing appropriate functionality for late-
stage aziridine introduction. In the case of 10, the cis-
1,2-diol could serve in a bidirectional manner as an
aziridine precursor depending upon which hydroxyl group
is inverted upon displacement with a nitrogen nucleo-
phile. Analysis of the diastereofacial selectivity in the
iminium ion addition led us to the prediction that the
R-hydroxyl groups of 10 would direct stannane addition
from the desired â-face. The two amine-containing pyr-
rolidines 11 and 12 have the advantage of earlier
introduction of what will ultimately become the aziridine
nitrogen, thereby leading to greater convergency; how-
ever, 11 bears a protected amino group adjacent to the
nascent iminium ion, and â-face diastereoselectivity in
the addition of allylstannane 8 was potentially problem-
atic. With pyrrolidine 12, the protected amino group could
still provide a steric impediment to â-face approach of
the allylstannane, but being distal to the iminium ion
may make this substituent less of a factor in control of
diastereoselectivity.
We now report the full details of our successful efforts
to implement this strategy using more highly elaborated
cinnamylstannanes and pyrrolidine constructs and de-
scribe the stereochemical details of the key N-acylimin-
ium ion addition reaction as a function of substitution of
the pyrrolidine ring. These studies have resulted in a
stereocontrolled and convergent synthesis of aziridinomi-
tosane 1 starting from the simple aromatic and carbo-
hydrate precursors 2,6-dimethoxytoluene and D-ribose.
Introduction of the central B-ring of 1 would rely on
the well-established chemistry of quinones using an
intramolecular Michael addition by the pyrrolidine ni-
trogen of 7 followed by air oxidation.16 The quinone
precursor, protected hydroquinone 7, would need to
possess functionalities X and Y that could serve as
precursors for introduction of the aziridine ring. The
identities of X and Y and the relative configuration of
the carbon atoms that bear them provided us with
considerable flexibility in the timing and strategy for
aziridine introduction. The key C9-C9a bond in 7 is an
ideal connection point, maximizing convergence in the
coupling of a fully elaborated γ-aryl allylstannane 8 and
appropriately functionalized N-acyliminium ion 9. De-
(12) Paz, M. M.; Hopkins, P. B. J . Am. Chem. Soc. 1997, 119, 5999.
(13) (a) Ziegler, F. E.; Berling, M. Y. Tetrahedron Lett. 1998, 39,
2455. (b) Danishefsky, S.; Berman, E. M.; Ciufolini, M.; Etheredge, S.
J .; Segmuller, B. E. J . Am. Chem. Soc. 1985, 107, 3891. (c) Kinoshita,
S.; Uzu, K.; Nakano, K.; Shimizu, M.; Takahashi, T. J . Med. Chem.
1971, 14, 103.
(14) Coleman, R. S.; Chen, W. Org. Lett. 2001, 3, 1141.
(15) (a) Yin, J .; Buchwald, S. L. Org. Lett. 2000, 2, 1101. (b) Wolfe,
J . P.; Tomori, H.; Sadighi, J . P.; Yin, J .; Buchwald, S. L. J . Org. Chem.
2000, 65, 1158. (c) Wolfe, J . P.; Buchwald, S. L. J . Org. Chem. 1997,
62, 1264. (d) A° hman, J .; Buchwald, S. L. Tetrahedron Lett. 1997, 38,
6363.
(16) Finley, K. T. The addition and substitution chemistry of
quinones. In Chemistry of the Quinonoid Compounds; Patai, S., Ed.,
Wiley: New York, 1974; Vol. 2, pp 877-1144.
P r ep a r a tion of Mitosa n e C-Rin g P r ecu r sor s. Our
initial work in developing this synthetic strategy started
with the preparation of an N-acyliminium ion precursors
based on the pyrrolidine diol construct 10, where the cis-
1,2-diol of 10 was mapped onto the C2/C3 diol of D-ribose.
One-pot acetonide formation and glycosylation of D-ribose
7310 J . Org. Chem., Vol. 69, No. 21, 2004

Synthesis of a Fully Elaborated Aziridinomitosane
with allyl alcohol provided the corresponding allyl gly-
coside/2,3-acetonide 13, which was converted to the
5-azido-5-deoxyribose 14 in high yield using zinc azide
under Mitsunobu conditions.17 Conversion of the azide
of 14 to the corresponding tert-butyl carbamate 15
proceeded in 88% yield for the two-step sequence of
reduction and acylation. Allyl deprotection of 15 was
affected using nickel chloride and triethylaluminum18 to
afford 16. At this stage, the furanose ring was fragmented
by treatment with iodosobenzene and iodine, which
afforded pyrrolidine 17 in good yields for this efficient
rearrangement process.19
be the ideal method for pyrrolidine ring formation. This
reaction presumably proceeds by two sequential one-
electron oxidations from 16, first to afford the glycosyl
radical 19, which fragments to afford carbon-centered
radical 20, which is further oxidized to oxonium ion 21,
which in turn is trapped by the carbamate nitrogen to
afford 17 after loss of a proton.
The pyrrolidine system bearing nitrogen-containing
functionality at C4 was prepared by the same route as
for 19. Alcohol 18 was converted to the corresponding
azide 22 in excellent yield by reaction with diphenylphos-
phoryl azide.20 The azide of 22 was reduced, and the
resulting amine was acylated to afford benzyl carbamate
23. Azide 22 and carbamate 23 are seven and eight steps,
respectively, from D-ribose.
Two methods proved effective for removing the formate
ester of 17: (1) reduction with lithium triethylboro-
hydride and (2) methanolysis under basic reaction condi-
tions. The resulting alcohol 18 was protected as the
corresponding benzyl ether to afford 19 in essentially
quantitative yield. We intended that the acetonide would
serve as a source of the corresponding iminium ion and
the Lewis acid complex of the ring-opened acetal would
serve as a steric block of the R-face, directing approach
of the allyl stannane. This system was now at the point
of convergence after seven synthetic operations from
D-ribose (27% overall).
The pyrrolidine system bearing nitrogen-containing
functionality at C3 was synthesized by a different route
using a Sharpless asymmetric aminohydroxylation.21
p-Methoxybenzylamine was alkylated with methyl 4-bro-
mocrotonate (24), and the resulting secondary amine was
acylated to afford bis-protected amine 25. Double protec-
tion of the amine group was essential for subsequent
success in the Sharpless aminohydroxylation. Using
literature reaction conditions with benzyl urethane and
(DHQD)2AQN, 25 afforded the protected amino alcohol
26 in good yields and with the desired regioselectivity.
Removal of the p-methoxybenzyl (PMB) group and reduc-
(19) (a) Armas, P.; Francisco, C. G.; Suarez, E. Angew. Chem., Int.
Ed. Engl. 1992, 31, 772. (b) Armas, P.; Francisco, C. G.; Suarez, E.
Tetrahedron Lett. 1992, 33, 6687. (c) Armas, P.; Francisco, C. G.;
Suarez, E. J . Am. Chem. Soc. 1993, 115, 8865. (d) Francisco, C. G.;
Freire, R.; Gonzalez, C. C.; Suarez, E. Tetrahedron: Asymmetry 1997,
8, 1971.
(20) (a) Lal, B.; Pramanik, B.; Manhas, M. S.; Bose, A. K. Tetrahe-
dron Lett. 1977, 1977. (b) Shioiri, T.; Ninomiya, K.; Yamada, S. J . Am.
Chem. Soc. 1972, 94, 6203.
(21) (a) Schlingloff, G.; Sharpless, K. B. Asym. Oxid. React. 2001,
104. (b) Andersson, M. A.; Epple, R.; Fokin, V. V.; Sharpless, K. B.
Angew. Chem., Int. Ed. 2002, 41, 472. (c) Fokin, V. V.; Sharpless, K.
B. Angew. Chem., Int. Ed. 2001, 40, 3455. (d) Rubin, A. E.; Sharpless,
K. B. Angew. Chem., Int. Ed. Engl. 1997, 36, 2637. (e) Li, G.; Angert,
H. H.; Sharpless, K. B. Angew. Chem., Int. Ed. Engl. 1996, 35, 2813.
This iodosobenzene-induced furanose fragmentation
reaction developed by Suarez and co-workers proved to
(17) Viaud, M. C.; Rollin, P. Synthesis 1990, 130.
(18) Taniguchi, T.; Ogasawara, K. Angew. Chem., Int. Ed. 1998, 37,
1136.
J . Org. Chem, Vol. 69, No. 21, 2004 7311

Coleman et al.
tion of the ester to the aldehyde was accompanied by
spontaneous cyclization to pyrrolidine 27. The enantio-
meric excess of the Sharpless product was determined
by chiral HPLC to be 92%, after removal of the p-
methoxybenzyl group of 26. Pyrrolidine 27 was synthe-
sized in five steps from methyl 4-bromocrotonate (28%
overall).
lithiation meta to the methoxymethyl ether was noted.
Acetate ester formation afforded allylic acetate 33, and
reductive stannylation24 occurred to afford trans-cin-
namylstannane 34 with complete regio- and stereocontrol
in good chemical yield. [The originally reported method-
ology was developed for allylic halides and occurred
without thermodynamically driven double bond migra-
tion; our application of this protocol has expanded the
scope of the method.] Cinnamylstannane 34 was synthe-
sized from 28 in six steps and in 43% overall yield.
Allylsta n n a n e/Im in iu m Ion Ad d ition s. Gen er a -
tion On e Ap p r oa ch . The convergent coupling of stan-
nane 34 with pyrrolidine 19 proceeded smoothly with
allylic transposition in the presence of boron trifluoride
etherate, presumably by way of iminium ion 35. The
coupled product 36 could be isolated in good yield and
as a single diastereomer. Stoichiometric Lewis acid was
critical to prevent deprotection of the phenolic meth-
oxymethyl ether of 34. A single diastereoisomer was
clearly present in the product 36, but line broadening in
1
the H NMR caused by slow interconversion of tert-butyl
carbamate conformers prevented determination of the
relative or absolute configuration at the C9 and C9a
stereogenic centers at this point in the synthesis. This
problem plagued us throughout the work, and high-
resolution mass spectrometry proved to be exceptionally
P r ep a r a t ion of A-R in g Cin n a m ylst a n n a n e. In a
manner similar to a number of previously reported
approaches to the mitomycins, we constructed an A-ring
precursor from 2,6-dimethoxytoluene (28). Friedel-
Crafts alkylation22 of 28 with dichloromethyl methyl
ether in the presence of tin(IV) chloride afforded aryl
aldehyde 29 in excellent yield. Baeyer-Villiger oxida-
tion23 with m-CPBA and subsequent KOH hydrolysis of
the resulting formate ester afforded phenol 30, which was
protected as the corresponding methoxymethyl ether to
afford 31. (This sequence of reactions was exceptionally
convenient to perform on a large scale, as benzaldehyde
29 could be recrystallized from hexane and both phenol
30 and methoxymethyl ether 31 could be distilled.)
Directed ortho-lithiation of 31 with n-butyllithium and
reaction of the resulting aryllithium intermediate with
acrolein afforded allylbenzene 32 in good yield. No
1
useful in the absence of useful high-resolution H NMR
data.25
Absolute and relative stereochemistry of the addition
product 36 was demonstrated by X-ray crystallography
of the corresponding bis-dinitrobenzoate. Hydrogenolysis
of the benzyl ether and hydrogenation of the vinyl group
(22) Fleming, I. Chemtracts 2001, 14, 405 and references therein.
(23) Renz, M.; Meunier, B. Eur. J . Org. Chem. 1999, 737.
(24) Tabuchi, T.; Inanaga, J .; Yamaguchi, M. Tetrahedron Lett. 1987,
28, 215.
7312 J . Org. Chem., Vol. 69, No. 21, 2004

Synthesis of a Fully Elaborated Aziridinomitosane
optimization of ozonolysis conditions using a slow flow
of ozone at -78 °C, stopping the reaction by immediate
nitrogen flush when olefin 36 was consumed (30 min),
and in situ reduction of the ozonide with sodium boro-
hydride at 0 °C afforded the alcohol 39 in good overall
yield, without detectable epimerization at the benzylic
stereogenic center. The primary alcohol of 39 was pro-
tected with the robust tert-butyldiphenylsilyl ether to
afford 40. At this point there were two remaining
objectives that needed to be achieved: aziridine introduc-
tion and oxidation of the aromatic ring to the p-quinone
with accompanying cyclization of the B-ring. The tert-
butyl carbamate and methoxymethyl ether of 40 were
removed simultaneously in the presence of trifluoroacetic
acid to afford the corresponding pyrrolidine/phenol spe-
cies 41. Oxidation of the electron-rich aromatic system
to the p-quinone with ceric ammonium nitrate was
accompanied by B-ring cyclization to afford mitosane 42,
albeit in low yields even after extensive optimization,
primarily due to competing oxidation of the product 42
and/or the pyrrolidine nitrogen. Other oxidants such as
O2, CuCl2/O2, AgO, Ag2O, or DDQ were ineffective in this
reaction sequence. In the end, this poor yielding reaction
sequence was irrelevant as we were unable to form the
methanesulfonate ester of the secondary alcohol of mi-
tosane 42. Instead we unexpectedly observed the forma-
tion of mitosene 43, presumably from base-promoted
tautomerization initiated by deprotonation of C9-H to
form the semiquinone, followed by tautomerization to the
hydroquinone by removal of C9a-H; air oxidation pre-
sumably provided the means for reoxidation to quinone
43.
F IGURE 1. Representation of the structure of 38 (X-ray), with
dinitrobenzoate esters and nonrelevant hydrogens removed for
clarity. The newly created stereogenic centers, C9 and C9a,
are labeled.
of 36 afforded diol 37. Acylation of diol 37 with 3,5-
dinitrobenzoyl chloride afforded the crystalline diester
38 (DNB ) dinitrobenzoyl).
Introduction of a trans-aminodiol prior to cyclization
was similarly unsuccessful. Direct displacement of the
secondary alcohol of 40 under modified Mitsunobu condi-
tions using zinc azide bis-pyridine complex led exclusively
to elimination and olefin migration product 44. At lower
reaction temperatures, alcohol 40 was unreactive. In the
case of 40, acylation of the secondary alcohol with
methanesulfonyl chloride was successful in forming me-
sylate 45 in good yield, but displacement with sodium
azide similarly led to the elimination/migration product
44. This particular synthetic route dead-ended with
compound 40.
Single-crystal X-ray analysis of 38 confirmed the
absolute configuration at the newly formed stereogenic
centers to be 9R,9aS (Figure 1) and demonstrated that
addition of stannane 34 to pyrrolidine 19 proceeded with
the same sense of diastereoselection that we had observed
in preliminary studies using a silyl enol ether. Presum-
ably, allylstannane 34 approaches iminium ion 35 by a
synclinal transition state like that shown (Figure 1), from
the face opposite the bulky alkoxy substituents.
The alkene present in adduct 36 tended to migrate into
conjugation with the aromatic system under acidic or
basic reaction conditions, and the electron-rich aromatic
system was sensitive to oxidation. However, careful
(25) This somewhat unorthodox characterization method involved
purification of reaction products to homogeneity by flash chromatog-
raphy (silica) or preparative TLC (silica) followed by analysis by both
1H NMR and HRMS (ESI). The MS data confirmed the molecular
formula and allowed at least some NMR assignments to be made. We
found it essentially impossible to confirm the identity of reaction
products using NMR as the sole method of characterization.
J . Org. Chem, Vol. 69, No. 21, 2004 7313

Coleman et al.
Two alternatives to this unsuccessful approach toward
aziridine introduction appeared reasonable: (1) introduc-
tion of the requisite nitrogen by displacement at the
benzyloxy carbon of 45, which would require a significant
investment in protecting group manipulations, or (2)
introduction of the emergent aziridine nitrogen at an
earlier juncture of the synthesis, prior to coupling of the
two fragments. Because we wished to avoid a potentially
troublesome series of protecting group manipulations on
a late intermeidate, we chose the option of using a
nitrogen-functionalized pyrrolidine in the iminium ion
addition reaction.
22 from D-ribose, plus the coupling step). The remainder
of the synthesis required two cyclization reactions, an
unveiling of protecting groups, and adjustment of the
oxidation state of the aromatic A-ring.
Allylsta n n a n e/Im in iu m Ion Ad d ition s. Gen er a -
tion Tw o Ap p r oa ch : 3-Am in op yr r olid in e. With a
bulky benzyl carbamate now proximal to the iminium ion
formed by treatment of 27 with boron trifluoride, addition
of cinnamylstannane 34 occurred with poor diastereo-
selectivity, affording a 1:1 mixture of stereoisomers 46
of unidentified configuration at C9/C9a. Although aziri-
dine 47 could readily be formed in good yield by Mit-
sunobu cyclization of the amino alcohol of 46, this entire
route proved unworkable as a result of the lack of
stereocontrol in the key addition reaction between 27 and
34.26
Reduction of the azide of 49 occurred smoothly and in
good yield upon treatment with stannous chloride and
thiophenol in the presence of triethylamine, which af-
forded the corresponding amine. Acylation with benzyl
chloroformate afforded carbamate 50. The use of tri-
phenylphosphine as a reductant effected only 25% yield
in the two-step sequence. Cyclization to the aziridine
occurred with essentially quantitative conversion under
Mitsunobu reaction conditions, and aziridine 51 was
isolated in 95% yield from 50. Using previously estab-
lished conditions for ozonolysis with reductive workup,
the alkene of 51 was transformed to the desired alcohol
52 by treatment with ozone at -78 °C followed by NaBH4
at 0 °C. This reaction routinely proved difficult to achieve
in reasonable yields, and other protocols (e.g., OsO4/
NaIO4) were unsuccessful in the presence of the electron-
rich aromatic system. The endgame of the synthesis from
52 proved straightforward, although it required an
unfortunate change of nitrogen protecting groups prior
to quinone formation.27 Cleavage of the tert-butyl car-
bamate of 52 with trimethylsilyl triflate and reprotection
Allylsta n n a n e/Im in iu m Ion Ad d ition s. Gen er a -
tion Th r ee Ap p r oa ch : 4-Am in op yr r olid in e. Addition
of cinnamylstannane 34 to 4-azidopyrrolidine 22 now
afforded a favorably disposed 7:3 diastereomeric ratio of
addition products 49 in 73% yield, where the major
diastereomer corresponded to that shown. [A similar
result was obtained in the addition of cinnamylstannane
to the protected 4-aminopyrrolidine 23, where the cor-
responding addition product was formed, albeit in only
54% yield but as a 3:1 ratio of diastereomers favoring
the desired isomer.] At the stage of azide 49, we had
introduced essentially all of the molecular complexity of
the target molecule in 14 total synthetic steps (six steps
to 34 from 2,6-dimethoxytoluene plus seven steps to azide
(26) Aziridine 47 could be carried forward as a stereoisomeric
mixture without complications [(1) O3, MeOH, -78 °C, then NaBH4, 0
°C, 55%; (2) t-BuPh2SiCl, Et3N, DMAP, 96%; (3) (NH4)3Ce(NO3)6, CH3-
CN/H2O, 25 °C, 87%] to afford the quinone 48, thereby validating the
feasibility of maintaining the integrity of the aziridine ring system
through this reaction sequence.
(27) It proved impossible to remove a pyrrolidine tert-butyl carbam-
ate once the C10 carbamate and/or the quinone were installed;
intermediates possessing a benzyl carbamate protecting the pyrrolidine
nitrogen could be converted to a quinone bearing system.
7314 J . Org. Chem., Vol. 69, No. 21, 2004

Synthesis of a Fully Elaborated Aziridinomitosane
TABLE 1. Ca lcu la ted P r oton -P r oton Dista n ces a n d
Nu clea r Over h a u ser En h a n cem en ts for Ster eoisom er ic
P r od u cts
as the benzyl carbamate afforded 53 in 63% unoptimized
yield.
The side chain carbamate was installed onto 53 using
trichloroacetyl isocyanate to afford 54 in 86% yield.
Deprotection and oxidation of the aromatic ring occurred
directly upon treatment of 54 with ceric ammonium
nitrate, which afforded p-quinone 55 in essentially quan-
tititative yield (95% isolated yield). Removal of the benzyl
carbamates protecting the aziridine and pyrrolidine
nitrogens occurred under standard hydrogenolysis condi-
tions from 55 to afford the corresponding aziridino[1,2-
c]pyrrolidine with the aromatic ring in the hydroquinone
oxidation state. Direct exposure of this compound to
atmospheric oxygen, following removal of traces of pal-
ladium by filtration through a 45-µm filter to prevent
oxidation of the pyrrolidine ring, affected a slow reoxi-
dation to the quinone accompanied by cyclization of the
B-ring, afforded mitosane 1 (9a-desmethoxy mitomycin
A) in good yield (74% for two steps), and completed the
synthesis of the target molecule.
Difference NOE spectroscopy with 1 demonstrated the
correct stereochemistry had been obtained at the syn-
thetically installed C9 and C9a stereocenters (Figure 2).
Relevant nuclear Overhauser effect enhancements are
listed in Table 1 and are consistent with the stereoisomer
shown in Figure 2, clearly demonstrating that the
absolute configuration at the C9 and C9a stereogenic
centers is the same as that found in mitomycin A and C.
Although NOE measurements appear to be consistent
with the third structure in Table 1 where C9-H and
C9a-H are cis and R, this structure is highly unlikely
because it would require allylstannane 34 to approach
the iminium ion generated from 22 from the R-face that
bears the bulky alkoxy groups. Precedent suggests that
this is disfavored (see addition of 34 to 19). The remain-
ing two structures, where C9-H and C9a-H are trans, are
inconsistent with observed NOE enhancements.
One synthetic problem that has yet to be seriously
addressed is the issue of the C9a angular methoxy group.
It is widely appreciated by those chemists who have
chosen the mitomycins as synthetic targets that an
alkoxy or hydroxy group at this position is highly labile.
It is exceptionally difficult to manipulate compounds
bearing an oxygen at C9a when the intact tetracyclic ring
system is in the hydroquinone oxidation state. This is,
a
Calculated from energy minimized (MM3) structures gener-
b
ated in CHCl3 using MacroModel (v 6.0). Key: minus ) not
observed; asterisk ) impossible to observe because of spectral
overlap.
in fact, what happens in vivo as the quinone is reduced
to the hydroquinone, and the B-ring undergoes elimina-
tion and aromatization to set up the system for DNA
cross-linking. Successful synthetic routes to the mito-
mycins have sidestepped this issue by avoiding hydro-
quinone (indoline) systems altogether or by placement
of a carbonyl group next to the bridgehead nitrogen,
where amide resonance prevents elimination. In the
context of the present work, early introduction of the
angular methoxy group is not feasible, and so a chemose-
lective oxidation at this position will be necessary for
completion of the total synthesis of the mitomycin C.
J . Org. Chem, Vol. 69, No. 21, 2004 7315

Coleman et al.
sequence of reactions was 16 steps,29 making this syn-
thesis of the mitomycin system among the more efficient
reported to date. The key step in this reaction sequence,
coupling of cinnamylstannane 34 with the iminium ion
derived from pyrrolidine 22, occurred in good chemical
yield with useful diastereoselectivity.
Ack n ow led gm en t. This work was supported by a
grant from the National Cancer Institute (NIH CA91904).
We thank Dr. J udith Gallucci for determining the
structure of 38.
Su p p or tin g In for m a tion Ava ila ble: Full details on the
preparation, characterization of synthetic intermediates and
products, and crystallographic data in CIF format. This
material is available free of charge via the Internet at
http://pubs.acs.org.
F IGURE 2. Representation of the energy minimized (MM2)
structure of 1 with relevant protons numbered (mitomycin
numbering).
J O048924I
Such oxidations are precedented in the context of natural
products synthesis.28
(28) He, F.; Bo, Y.; Altom, J . D.; Corey, E. J . J . Am. Chem. Soc. 1999,
121, 6771.
The convergent and diastereoselective synthesis of 1
was achieved in 23 total synthetic operations from
D-ribose and 2,6-dimethoxytoluene. The longest linear
(29) Comparable syntheses of 1 or a closely related compound have
required 32 total steps (20 linear)13a or 27 steps (27 linear).13b
7316 J . Org. Chem., Vol. 69, No. 21, 2004