Synthetic Enantiopure Aziridinomitosenes: Preparation, Reactivity, and DNA Alkylation Studies

Article abstract of DOI:10.1021/ja030452m

An enantiocontrolled route to aziridinomitosenes had been developed from L-serine methyl ester hydrochloride. The tetracyclic target ring system was assembled by an internal azomethine ylide cycloaddition reaction based on silver ion-assisted intramolecul

Full text of DOI:10.1021/ja030452m

Published on Web 12/02/2003
Synthetic Enantiopure Aziridinomitosenes: Preparation,
Reactivity, and DNA Alkylation Studies
Edwin Vedejs,*,† B. N. Naidu,† Artis Klapars,† Don L. Warner,† Ven-shun Li,‡
Younghwa Na,‡ and Harold Kohn*,‡,§
Contribution from the Departments of Chemistry, UniVersity of Michigan,
Ann Arbor, Michigan 48109, and UniVersity of Houston, Houston, Texas 77204
Received July 25, 2003; E-mail: edved@umich.edu
Abstract: An enantiocontrolled route to aziridinomitosenes had been developed from L-serine methyl ester
hydrochloride. The tetracyclic target ring system was assembled by an internal azomethine ylide cycloaddition
reaction based on silver ion-assisted intramolecular oxazole alkylation and cyanide-induced ylide generation
via a labile oxazoline intermediate (62 to 66). Other key steps include reductive detritylation of 26, methylation
of the N-H aziridine 56, oxidation of the sensitive cyclohexenedione 68 to quinone 70, and carbamoylation
using Fmoc-NCO. Although the aziridinomitosene tetracycle is sensitive, a range of protecting group
manipulations and redox chemistry can be performed if suitable precautions are taken. A study of DNA
alkylation by the first C-6,C-7-unsubstituted aziridinomitosene 11a has been carried out, and evidence for
DNA cross-link formation involving nucleophilic addition to the quinone subunit is described.
Mitomycins A (1), B (2), and C (3)1-3 have been the focus
of intensive study due to their fascinating structures, unusual
metabolic activation pathways, and clinical antitumor activity.4
Thus, 1, 3, and mitomycin K (4) have served as targets for total
synthesis,5-8 while 1 and 3 have been investigated in detail to
clarify the mechanisms of DNA alkylation (Scheme 1).1,9-12
Reductive activation of mitomycins generates leucoaziridi-
nomitosenes such as 5a, the intermediates responsible for DNA
cross-link formation.11 Leucoaziridinomitosenes are too reactive
for isolation, but 5a can be observed in solution using NMR
techniques,12 and the bis-silyl derivative 5b can be purified by
chromatography.13 Aziridinomitosenes 6, 7, 8, and 9 are
relatively stable, although they are more sensitive than the parent
mitomycins due to the activating effect of the indole nitrogen
on heterolysis of the aziridine C-N bond.1c,11,14,15 Aziridinomi-
tosenes do not require reductive activation to alkylate DNA, in
contrast to the mitomycins. Thus, 7 has been shown to
monoalkylate DNA,15 while 9 has in vivo activity similar to
that of mitomycin C.16
Several total syntheses of racemic mitomycins have been
reported,5-8 but there has been no enantioselective synthesis.17,18
Only one synthesis of a fully functional aziridinomitosene
(racemic 6) has been reported to date (Jimenez and Dong).19
The highly sensitive aziridine ring was installed in the final step,
a strategy that minimizes problems with reactivity but that
somewhat limits options for enantioselective synthesis.
Efforts in our laboratory have been focused on previously
unknown aziridinomitosenes such as 10, 11a, and 11b. These
structures contain potentially electrophilic sites at C-6 or C-7
† University of Michigan.
‡ University of Houston.
§ Current address: Division of Medicinal Chemistry and Natural Products,
School of Pharmacy, University of North Carolina, Chapel Hill, NC 27599.
(1) Reviews: (a) Wolkenberg, S. E.; Boger, D. L. Chem. ReV. 2002, 102, 2477.
(b) Rajski, S. R.; Williams, R. M. Chem. ReV. 1998, 98, 2723. (c) Tomasz,
M.; Palom, Y. Pharmacol. Ther. 1997, 76, 73. (d) Danishefsky, S. J.;
Schkeryantz, J. M. Synlett 1995, 475. (e) Tomasz, M. Chem. Biol. 1995,
2, 575. (f) Kasai, M.; Kono, M. Synlett 1992, 778. (g) Franck, R. W.;
Tomasz, M. In The Chemistry of Antitumor Agents; Wilman, D. E. V.,
Ed.; Blackie and Sons: Scotland, 1989. (h) Remers, W. A. In The Chemistry
of Antitumor Antibiotics, Vol. 1; John Wiley & Sons: New York, 1979.
(2) (a) Hata, T.; Sano, Y.; Sugawara, R.; Matsumae, A.; Kanamori, K.; Shima,
T.; Hoshi, T. J. Antibiot. 1956, 9, 141. (b) Wakaki, S.; Marumo, H.;
Tomioka, K.; Shimizu, G.; Kato, E.; Kamada, H.; Kudo, S.; Fujimoto, Y.
Antibiot. Chemother. 1958, 8, 228.
(3) (a) Webb, J. S.; Cosulich, D. B.; Mowat, J. H.; Patrick, J. B.; Broschard,
R. W.; Meyer, W. E.; Williams, R. P.; Wolf, C. F.; Fulmor, W.; Pidacks,
C. J. Am. Chem. Soc. 1962, 84, 3187. (b) Tulinsky, A. J. Am. Chem. Soc.
1962, 84, 3190. (c) Shirahata, K.; Hirayama, N. J. Am. Chem. Soc. 1983,
105, 7199.
(13) Egbertson, M.; Danishefsky, S. J. J. Am. Chem. Soc. 1987, 109, 2204.
(14) Patrick, J. B.; Williams, R. P.; Meyer, W. E.; Fulmor, W.; Cosulich, D.
B.; Broschard, R. W.; Webb, J. S. J. Am. Chem. Soc. 1964, 86, 1889.
(15) (a) Han, I.; Kohn, H. J. Org. Chem. 1991, 56, 4648. (b) Li, V.-S.; Choi,
D.; Tang, M.-S.; Kohn, H. J. Am. Chem. Soc. 1996, 118, 3765.
(16) Iyengar, B. S.; Remers, W. A.; Bradner, W. T. J. Med. Chem. 1986, 29,
1864.
(17) (a) Papaioannou, N.; Evans, C. A.; Blank, J. T.; Miller, S. J. Org. Lett.
2001, 3, 2879. (b) Ziegler, F. E.; Berlin, M. Y. Tetrahedron Lett. 1998,
39, 2455.
(4) Iyer, V. N.; Szybalski, W. Science 1964, 145, 55.
(5) (a) Nakatsubo, F.; Fukuyama, T.; Cocuzza, A. J.; Kishi, Y. J. Am. Chem.
Soc. 1977, 99, 8115. (b) Fukuyama, T.; Nakatsubo, F.; Cocuzza, A. J.;
Kishi, Y. Tetrahedron Lett. 1977, 4295. (c) Kishi, Y. J. Nat. Prod. 1979,
42, 549.
(6) (a) Fukuyama, T.; Yang, L. J. Am. Chem. Soc. 1989, 111, 8303. (b)
Fukuyama, T.; Yang, L. J. Am. Chem. Soc. 1987, 109, 7881.
(7) Benbow, J. W.; McClure, K. F.; Danishefsky, S. J. J. Am. Chem. Soc. 1993,
115, 12305.
(8) Wang, Z.; Jimenez, L. S. Tetrahedron Lett. 1996, 37, 6049.
(9) Iyer, V. N.; Szybalski, W. Proc. Natl. Acad. Sci. U.S.A. 1963, 50, 335.
(10) Tomasz, M.; Lipman, R.; Chowdary, D.; Pawlek, J.; Verdine, G. L.;
Nakanishi, K. Science 1987, 235, 1204.
(18) For enantioselective approaches, see: (a) Vedejs, E.; Little, J. J. Am. Chem.
Soc. 2002, 124, 748. (b) Michael, J. P.; de Koning, C. B.; Petersen, R. L.;
Stanbury, T. V. Tetrahedron Lett. 2001 42, 7513. (c) Lee, S.; Lee, W.-M.;
Sulikowski, G. A. J. Org. Chem. 1999, 64, 4224. (d) Utsunomiya, I.; Fuji,
M.; Sato, T.; Natsume, M. Chem. Pharm. Bull. 1993, 41, 854. (e) Shaw,
K. J.; Luly, J. R.; Rapoport, H. J. Org. Chem. 1985, 50, 4515.
(19) Dong, W.; Jimenez, L. S. J. Org. Chem. 1999, 64, 2520.
(11) Cummings, J.; Spanswick, V. J.; Tomasz, M.; Smyth, J. F. Biochem.
Pharmacol. 1998, 56, 405.
(12) Danishefsky, S. J.; Egbertson, M. J. Am. Chem. Soc. 1986, 108, 4648.
9
15796
J. AM. CHEM. SOC. 2003, 125, 15796-15806
10.1021/ja030452m CCC: $25.00 © 2003 American Chemical Society

Synthetic Enantiopure Aziridinomitosenes
A R T I C L E S
Scheme 1
that aromatizes by loss of HCN to generate the pyrrole 17. All
of these events occur at room temperature or below. Under such
mild conditions, it seemed possible that an aziridine subunit
might survive the steps from bicyclic oxazolium salt 19 to the
internal cycloadduct 20. The potential for participation by
aziridine nitrogen during the internal alkylation from 18 to 19
was a concern in view of prior reports describing generation of
bicyclic azetidinium intermediates from related substrates.22
However, model studies indicated that the oxazole nitrogen is
more nucleophilic than the N-tritylaziridine in reactions involv-
ing intermolecular N-alkylation.21a Thus, intramolecular oxazo-
lium salt formation might be faster than participation by aziridine
nitrogen.
The overall strategy goes against the conventional wisdom
that the most labile functionality (the solvolytically reactive
aziridine) should be introduced at the end of the synthesis. To
the contrary, our approach incorporates an aziridine subunit early
on. The challenge of handling sensitive aziridines at several
stages was expected to stimulate development of methodology
and to add to our understanding of aziridinomitosene chemistry.
Another advantage of using an intact aziridine early in the
sequence is that enantioselective synthesis is easily achieved
starting from simple amino acid precursors, as discussed below
in the context of target structures 11 (a series, R) H; b series,
R) methyl).23
Results and Discussion
Scheme 2
The above strategy requires the enantiocontrolled synthesis
of a 2,5-disubstituted aziridinyloxazole related to 18. Our plan
was to assemble the aziridine subunit from amino alcohol 25
(Scheme 3). This structure should be accessible by the coupling
of aldehyde 21 (prepared by O-allylation and DIBAL reduction
of methyl N-tritylserinate24) with the lithiated oxazole 24 if
control for the desired relative stereochemistry can be achieved.
Earlier studies in our laboratory had shown that electrophiles
react cleanly with 24a at oxazole C-2, without interference by
electrocyclic ring opening.25,26 We had also shown that the
precursor oxazole 23a27 can be prepared by the reaction of
lithiated methyl isocyanide28,29 with butyrolactone. The corre-
sponding reaction with 2-methylbutyrolactone proved difficult
to control, but 23b could be made using the analogous
Scho¨llkopf oxazole synthesis28 from the ester 22.
Oxazole 23a was converted into the lithiated borane complex
24a by treatment with THF-borane followed by deprotonation
with n-BuLi, and reaction with the aldehyde 21 in THF gave a
6:1 diastereomer mixture of amino alcohols 25a (93%). Similar
diastereomer ratios were obtained in toluene or ether, although
the reactions were not as clean. The stereochemistry of the major
product could not be established at this stage, but treatment of
the mixture with diethyl azodicarboxylate and triphenylphos-
phine under Mitsunobu conditions30 afforded the single aziridine
(mitomycin numbering) in addition to the electrophilic C-1 and
C-10 carbons and may have new options for DNA alkylation
as well as additional pathways for activation. Our approach is
based on azomethine ylide generation from oxazolium salts,20
a methodology developed in our laboratory for rapid access to
indoloquinones (Scheme 2).21 The key sequence is triggered
by the nucleophilic addition of cyanide ion to oxazolium salt
13 to afford the transient 4-oxazoline 14, followed by electro-
cyclic ring opening to the azomethine ylide 15. Internal
cycloaddition to a tethered alkyne then produces pyrroline 16
(22) Deyrup, J. A.; Moyer, C. L. Tetrahedron Lett. 1968, 9, 6879. Gaertner, V.
R. J. Org. Chem. 1970, 35, 3952.
(23) Preliminary communication: Vedejs, E.; Klapars, A.; Naidu, B. N.;
Piotrowski, D. W.; Tucci, F. C. J. Am. Chem. Soc. 2000, 122, 5401.
(24) Baldwin, J. E.; Spivey, A. C.; Schofield, C. J.; Sweeney, J. B. Tetrahedron
1993, 49, 6309.
(25) Vedejs, E.; Monahan, S. D. J. Org. Chem. 1996, 61, 5192.
(26) Review: Iddon, B. Heterocycles 1994, 37, 1321.
(27) Vedejs, E.; Luchetta, L. J. Org. Chem. 1999, 64, 1011.
(28) Scho¨llkopf, U.; Schro¨der, R. Angew. Chem., Int. Ed. 1971, 10, 333.
(29) (a) Jacobi, P. A.; Walker, D. G.; Odeh, I. M. A. J. Org. Chem. 1981, 46,
2065. (b) Schuster, R. E.; Scott, J. E. Org. Synth. 1966, 46, 75.
(30) Pfister, J. R. Synthesis 1984, 969.
(20) Vedejs, E.; Grissom, J. W. J. Am. Chem. Soc. 1988, 110, 3238. Vedejs,
E.; Grissom, J. W. J. Org. Chem. 1988, 53, 1876.
(21) (a) Vedejs, E.; Piotrowski, D. W.; Tucci, F. C. J. Org. Chem. 2000, 65,
5498. (b) Vedejs, E.; Monahan, S. D. J. Org. Chem. 1997, 62, 4763. (c)
Vedejs, E.; Dax, S. L. Tetrahedron Lett. 1989, 30, 2627.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15797

A R T I C L E S
Vedejs et al.
Scheme 3
Scheme 4
derived from aziridine rearrangement and cleavage was also
isolated.36 Several other phosphorus-based reagents proved
ineffective,37 but Mitsunobu conditions gave promising results.38
The combination of triphenylphosphine, diethyl azodicarboxy-
late, and methyl iodide in THF at 70 °C effected ca. 50%
conversion to the desired iodide. Attempts to improve the
conversion by varying reaction time or stoichiometry were not
successful, but use of toluene in place of THF had a remarkable
effect on rate as well as yield. Starting material was completely
consumed in 2 h, and 27a was obtained in >95% yield. The
dramatic solvent effect may be the result of tight ion pairing in
the nonpolar toluene between the alkoxyphosphonium salt and
the iodide, thus ensuring proximity of the reactive ions.39 Once
formed, the iodides 27a and 27b were relatively stable despite
the presence of neighboring nitrogen nucleophiles at the aziridine
and oxazole subunits. Thus, deprotection with Bu4NF (TBAF)
followed by Dess-Martin oxidation proceeded smoothly to the
aldehydes 28 (87-89% from 27).
26a after chromatographic purification (71% isolated). The
desired cis stereochemistry was confirmed by NMR, based on
the characteristic 6.1 Hz vicinal coupling constant for the
aziridine protons.31,32 Furthermore, the product was shown to
have 98.1% ee by hplc assay comparison with a sample prepared
from racemic 21. Thus, addition of 24a to 21 occurs without
significant racemization, and the desired stereoisomer 25a is
the major product.
The same methods were used to convert 23b into 26b (65%
overall). As with 26a, the minor diastereomers from the
aldehyde coupling step were removed during purification of 26b,
but the presence of the remote methyl substituent resulted in a
1:1 diastereomer mixture of cis-aziridines (5.9 Hz coupling
between the aziridine protons). Because the added C-methyl
stereocenter disappears later in the synthesis, the diastereomers
were taken through the next steps without separation.33
Removal of the allyl protecting group from 26 using an in
situ generated zirconium(II) reagent34 provided the correspond-
ing alcohol in excellent yield, but conversion to the iodide 27
required for intramolecular oxazole N-alkylation was challeng-
ing. The triphenylphosphine-iodine adduct was initially used,35
and product 27a was obtained in 50% yield. A byproduct
In preparation for the crucial intramolecular 2 + 3 cycload-
dition, it was now necessary to incorporate a suitable dipolaro-
phile (Scheme 4). Model studies had shown that ester-activated
tethered alkynes are effective for this purpose,21a so the initial
experiments were conducted with an ynoate 29 as the dipo-
larophile. Following the earlier study, methyl propiolate was
treated with LiHMDS in the presence of cerium(III) chloride
(36) The allylic amine structure i is supported by 1H NMR, 13C NMR, and high-
resolution mass spectrometry, possibly resulting from a bicyclic azetidinium
cation.
(37) Other conditions include the following: (a) Ph3P-I2, Et3N or DBU, with
or without propylene oxide. (b) (PhO)3P, MeI.; Landauer, S. R.; Rydon,
H. N. J. Chem. Soc. 1953, 2224. (c) (PhO)3P-MeI Verheyden, J. P. H.;
Moffatt, J. G. J. Org. Chem. 1970, 35, 2319.
(38) Kunz, H.; Schmidt, P. Liebigs Ann. Chem. 1982, 1245.
(39) See Hughes, D. L. Org. React. 1992, 42, 335.
(31) Vedejs, E.; Kendall, J. T. J. Am. Chem. Soc. 1997, 119, 6941.
(32) Brois, S. J.; Beardsley, G. P. Tetrahedron Lett. 1966, 7, 5113.
(33) Amino alcohol 25b is formed as a 6:6:1:1 mixture of four diastereomers.
(34) Ito, H.; Taguchi, T.; Hanzawa, Y. J. Org. Chem. 1993, 58, 774.
(35) Lange, G. L.; Gottardo, C. Synth. Commun. 1990, 20, 1473.
9
15798 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Synthetic Enantiopure Aziridinomitosenes
A R T I C L E S
Scheme 5
and was reacted with 28a. The resulting alcohol was protected
with TBSOTf/iPr2NEt to provide the silyl ether 29 in 73%
overall yield.
Attempts to induce the intramolecular alkylation from 29 to
oxazolium salt 30 by heating in various solvents were not
successful. However, when the experiment was conducted using
commercial silver triflate to activate the iodide in acetonitrile,
downfield shifts were observed in the NMR spectrum consistent
with the formation of 30. When 30 was added dropwise to
BnMe3N+CN- in acetonitrile,21b each drop produced a transient
yellow color, tentatively attributed to an azomethine ylide
intermediate. The color faded within a second as each drop of
30 was added, and the desired tetracyclic pyrrole 31 was isolated
in a remarkable 91% yield. Similar experiments conducted in
the course of our model studies had rarely exceeded 60% yield.
Evidently, the N-protected aziridine does not interfere with either
the intramolecular oxazole alkylation step, nor with the multi-
stage 2 + 3 cycloaddition-aromatization sequence.
To demonstrate the viability of the approach, removal of the
N-trityl protecting group was investigated using variations of
our recently optimized reductive detritylation methodology.40
Attempted deprotection of 31 with the TFA-triethylsilane
reagent effected trityl cleavage but also opened the aziridine
ring to afford an amide 32, the result of aziridine trifluoroac-
etolysis followed by O- to N-migration of the trifluoroacetyl
group. The same complication has been encountered with other
solvolytically sensitive aziridines under these conditions.40
In a second experiment, 31 was treated with methanesulfonic
acid-triethylsilane (Scheme 4), followed by quenching with
iPr2NEt. The aziridine 33, derived from cleavage of O-silyl as
well as N-trityl groups, was isolated in 35% yield along with
unknown decomposition products. The NMR spectrum of 33
was complicated by slow inversion at the aziridine nitrogen
resulting in signal broadening and two sets of signals due to a
ca.1:1.5 mixture of the N-H invertomers. A similar phenom-
enon has been reported by Kohn and Han in their studies of
the naturally derived aziridinomitosenes.15a,41 The line shapes
could be improved dramatically by addition of activated
powdered molecular sieves to the NMR sample, suggesting that
broadening is due to proton exchange catalyzed by traces of
water in the sample.
Before proceeding further, control experiments were per-
formed to learn whether aziridine nitrogen protection is required
in the ylide generation/2 + 3 cycloaddition step. Thus, 29 was
treated with TFA-triethylsilane to remove the trityl group. This
procedure worked well and afforded 34 in 84% yield, a result
that can be contrasted with the conversion from 31 to 32 under
similar conditions. The electron-withdrawing oxazole substituent
in 34 probably retards aziridine ring cleavage, thereby improving
the yield in the detritylation step. However, treatment of 34 with
AgOTf in CD3CN resulted in a complex mixture of decomposi-
tion products, and conversion to the oxazolium salt could not
be achieved in the presence of the unprotected N-H aziridine.
The next challenge was to determine whether the ester
activating group is essential for the intramolecular cycloaddition,
or whether the reaction can also be performed at the correct
oxidation state and natural substitution pattern corresponding
to the carbamate ester of an unactivated propargyl alcohol
derivative (Scheme 5). Addition of lithiated propargyl carbamate
to the aldehyde 28a in the presence of CeCl3 followed by
silylation gave a disappointing 15% yield of 36, but sufficient
material was obtained to test the cycloaddition. The intramo-
lecular alkylation of the oxazole 36 induced by silver triflate
cleanly provided the oxazolium salt according to NMR assay,
and addition of the crude salt solution to BnMe3N+CN- in
acetonitrile afforded cycloadducts within 5 min at room tem-
perature. However, none of the desired product 38 was detected.
Instead, the nitrile assigned structure 40 was obtained in 67%
yield. Only two of the four possible diastereomers of 40 were
present (ca.1:1 ratio) and were readily separated by chroma-
tography (relative configuration of diastereomers not assigned).
The structure of 40 was deduced from the spectroscopic data.
The 1H NMR spectra of both diastereomers displayed two
characteristic signals for the exocyclic methylene protons, 5.6
(d, J ) 1.4 Hz) ppm and 5.4 (d, J ) 1.4 Hz) ppm. The presence
of a nitrile was indicated by the characteristic 13C chemical shift
(an additional signal between δ 111-120 ppm), as well as an
infrared absorption at 2243 cm-1. The 13C signals usually
observed for the pyrrole ring in structures similar to 38 (ca. δ
142, 138, 124, 118 ppm) were replaced by modified olefinic
signals in the range of δ 148-111 ppm. Finally, the UV
spectrum revealed a dramatic change in the chromophore, λmax
) 353 nm, compared to pyrrole analogues of 38 (typically λmax
) 300 nm).
(40) Vedejs, E.; Klapars, A.; Warner, D. L.; Weiss, A. H. J. Org. Chem. 2001,
66, 7542.
(41) Solvent-dependent aziridine invertomer ratios of 1.3-1.7:1 were observed
for the aziridinomitosene 7 in ref 15a, while 8 was reported as a single
isomer.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15799

A R T I C L E S
Vedejs et al.
Scheme 6
The formation of 40 indicates that the azomethine ylide was
generated normally and that intramolecular [3 + 2]-cycloaddi-
tion had probably occurred to give the expected intermediate
37. At this stage, enolization of 37 probably generates 39, an
intermediate having two potential leaving groups (C-9a cyanide;
C-10 carbamate). Carbamate is the better leaving group, and
its departure apparently is favored relative to aromatization via
loss of the cyanide. Thus, 40 is formed instead of the desired
product 38.
Nitrile 40 was quite stable after isolation, but further
transformations took place if the crude reaction mixture was
stirred an additional hour in the presence of excess BnMe3N+CN-.
In this case, isomeric, aromatized nitrile 42 was isolated in 55%
yield as a ca. 1:1.5 mixture of two diastereomers. The most
likely sequence of events leading to 42 involves conjugate
addition of cyanide anion to the exocyclic double bond in 40
followed by elimination of the C-9a cyanide. The structure of
42 was readily deduced because the spectroscopic data correlate
well with other pyrrole-containing cycloaddition products while
also containing the signals corresponding to the nitrile (13C NMR
δ 118 ppm; IR 2250 cm-1).
Although cycloadduct 38 could not be prepared in the above
experiments, the isolation of 40 and 42 suggested that the
unactivated dipolarophile had intercepted the azomethine ylide
with reasonable efficiency. The approach was therefore modified
by using a tethered propargyl ether as the dipolarophile in the
hope that the undesired elimination could be avoided (Scheme
6). Thus, lithiated tert-butyldimethylsilyl propargyl ether was
added to the aldehyde 28a at -40 °C, and the resulting alcohol
was protected as the acetate 43a (87%) or the TBS ether 44a
(93%). Either derivative proved suitable for intramolecular
oxazolium salt formation promoted by silver triflate. When the
resulting oxazolium salts were treated with BnMe3N+CN- at
room temperature as before, the desired cycloadducts 45a or
47a were obtained in 59% or 70% yield, respectively. The same
procedures were also effective starting from 28b to 43b (78%
overall) and on to 45b (66%). These exciting results confirmed
the feasibility of the internal cycloaddition strategy employing
an unactivated dipolarophile, although the yields were signifi-
cantly lower than with the activated ynoate dipolarophile.
Qualitative rate differences in the cycloaddition step were also
noted, as indicated by characteristic color changes. Thus, the
9
15800 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Synthetic Enantiopure Aziridinomitosenes
A R T I C L E S
transient yellow color attributed to the azomethine ylide
precursor of 47a faded over several seconds at room temper-
ature, while the color due to the ylide precursor of ester 31
(Scheme 4) disappeared in <1 s as each drop of the oxazolium
salt was added to a stirred solution of BnMe3N+CN-.
an additional electron-withdrawing carbonyl group. However,
the final oxidation from diketone 54 to the quinone 55 proved
very difficult to control. The best result was obtained using
triethylamine as the base in C6D6 at 70 °C under an oxygen
atmosphere; conditions that afforded 55 in 27% yield.
Selective saponification of the acetate ester 45b was easily
achieved using 1% NaOH without interference by the sensitive
aziridine. This provided an opportunity to explore the oxidative
conversion of 46b to the quinone substitution pattern that is
characteristic of the aziridinomitosenes. One added reason for
pursuing this route was that 45b and its precursors had been
prepared as mixtures of diastereomers, while the quinone 49
should be a single isomer that would be easier to characterize.
When alcohol 46b was oxidized under Dess-Martin conditions,
a complex mixture was obtained containing quinone 49 as well
as the diketone 48 and unknown side products. Bis-enolization
of 48 may have occurred to generate a hydroquinone intermedi-
ate having properties similar to those expected for a labile
leucoaziridinomitosene such as 5a (Scheme 1). Better results
were obtained using the mildly basic tetrapropylammonium
perruthenate in the presence of N-methylmorpholine N-oxide
(TPAP/NMO), a reagent combination that cleanly provided
diketone 48 in 89% yield.42
The final conversion of 48 to the quinone 49 was performed
on the basis of a double enolization-oxidation approach. The
strong base version of this transformation is unprecedented for
aziridinomitosenes, but potassium (bistrimethylsilyl)amide (KH-
MDS) is known to convert simpler cyclohexenediones to the
corresponding hydroquinone anions at low temperatures.43 Thus,
treatment of a THF solution of diketone 48 at -78 °C with
KHMDS instantaneously produced a color change to deep, dark
green. Passing oxygen into the solution rapidly discharged the
green color and produced the yellow-orange quinone 49 in 64%
yield. Alternatively, addition of N-chlorosuccinimide to the
dienolate at -78 °C also gave 49 (68%). As expected, 49 was
formed as a single isomer.
Selective deprotection of the primary TBS ether in 47a was
attempted, but the addition of 1 equiv of TBAF to the bis-silyl
ether 47a gave a 1:1 mixture of the diol 50 and the starting
material 47a. No intermediates could be detected by TLC,
suggesting that deprotection of the first silyl group facilitates
the deprotection of the second. The use of excess TBAF
provided the crude diol 50 in ca. 80% yield after quick filtration
chromatography over buffered silica gel (triethylamine), but 50
was too sensitive for more extensive purification and was used
directly in the next step. Selective carbamoylation of the primary
alcohol with trichloroacetyl isocyanate44 gave the unstable imide
52. Attempted purification resulted in decomposition, so the
crude material was stirred with aqueous K2CO3 to remove the
trichloroacetyl group, followed by oxidation of crude 53 with
pyridine-buffered Dess-Martin periodinane to afford the sensi-
tive, but isolable diketone 54 in 26% yield over four steps.
Despite the imminent threats of aziridine solvolysis and
aromatization by double enolization, the diketone 54 proved to
be reasonably stable at neutral pH and survived chromatography
even on unbuffered silica gel. Aziridine solvolysis is probably
retarded in 54 compared to 50, 52, or 53 due to the presence of
The oxidation to 55 was not optimized due to another
complication that was encountered once sufficient material had
been prepared. Our most effective reductive detritylation condi-
tions with MsOH/triethylsilane proved to be too harsh for the
fully intact aziridinomitosene skeleton of 55, and several
attempts resulted in complex product mixtures. Detritylation
occurred, as evidenced by the formation of triphenylmethane,
but disappearance of the characteristic quinone color suggested
that reduction of the quinone had also taken place. This
presumably leads to a leucoaziridinomitosene and to rapid
destruction via facile aziridine ring opening in the electron-
rich environment. Attempted deprotection of ketol 53 or diketone
54, substrates that lack the quinone moiety, also failed to
produce any of the desired deprotected aziridines. It became
clear that the trityl protecting group would have to be removed
earlier in the synthetic sequence. This was a disappointing
outcome, but it was not unexpected in view of the difficulties
already encountered in the case of tetracyclic keto ester 31.
Furthermore, the successful deprotection of an advanced
intermediate 29 (Scheme 4) suggested alternative approaches
that eventually were successful, as described in the next section.
The need to remove N-trityl at an early stage focused attention
on the oxazole aziridines 26 (Scheme 7). The same triethylsi-
lane-trifluoroacetic acid procedure (2 h at 0 °C; NEtiPr2
quench) was attempted that had worked well for deprotection
of 29. Unexpectedly, this gave an inseparable mixture of the
desired product 56a and the triethylsilyl ether 57a, resulting
from partial cleavage of the tert-butyldimethylsilyl (TBS) group
under the acidic reaction conditions. Fortunately, the reaction
of 26a with trimethylamine-borane and trifluoroacetic acid
cleanly produced the deprotected aziridine 56a in 82% yield.
Monitoring the reaction by TLC indicated that deprotection was
complete within 5 min at 0 °C, while the triethylsilane conditions
required 2 h for complete conversion.
The N-methylation of the deprotected aziridine 56a was
performed next. Competition between the aziridine and the
oxazole nitrogen was observed in model exeriments using
methyl iodide or methyl triflate, but the selectivity was
dramatically improved if the aziridine was lithiated. Thus, 56a
was treated with nBuLi followed by methyl iodide to produce
the desired N-methylaziridine 58a in 91% yield. Following the
precedents of Scheme 3, 58a was then converted into the iodide
59a (86%). Removal of the silyl protecting group in 59a using
TBAF encountered purification difficulties due to the highly
polar nature of the resulting alcohol 60a, but good results were
obtained with HF-pyridine, and 60a was produced in 94%
yield. Oxidation of 60a with the Dess-Martin periodinane
followed by addition of lithiated tert-butyldimethylsilyl prop-
argyl ether to the resulting aldehyde at -40 °C then gave the
alcohol 61a as a ca. 1:1 mixture of diastereomers. The same
optimized procedures were applied to the conversion of 26b
into 61b, and generally similar results were obtained (26%
overall from 26b).
(42) Ley, S. V.; Norman, J.; Griffith, W. P.; Mardsen, S. P. Synthesis 1994,
639.
(43) Lokan, N. R.; Craig, D. C.; Paddon-Row: M. N. Synlett 1999, 397.
(44) Kocovsky, P. Tetrahedron Lett. 1986, 27, 5521.
The key [3 + 2]-cycloaddition reactions could now be
explored. Attempts to carry out the AgOTf-BnMe3N+CN-
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15801

A R T I C L E S
Vedejs et al.
(70 °C) followed by ylide generation with BnMe3N+CN-.
Although this result was similar to that from 62a, the erosion
in yield prompted additional attempts to re-optimize the reaction.
Promising room-temperature conditions for the internal N-
alkylation from 62b to 65b were found using purified AgOTf‚
C6H645 as the activating agent, but the efficiency (30% yield of
66b, 43% based on recovered 62b) at partial conversion was
not sufficiently improved to warrant the effort needed to prepare
the pure silver complex or to recover and recycle 62b.
Scheme 7
Another experiment using AgOTf‚C6H6 was conducted start-
ing with 62b in dichloromethane at room temperature. Although
no precipitated AgI was observed, the starting halide 62b
disappeared according to TLC analysis, so the solution was
treated with BnMe3N+CN- in acetonitrile in an attempt to
induce internal cycloaddition. The resulting mixture proved to
be complex and contained little if any (<5%) of the cycloadduct
66b. After chromatography, one zone could be separated
sufficiently to allow a tentative assignment of structure 63 (19%
isolated). Although the material was obtained as a mixture of
diastereomers, incorporation of cyanide at the primary carbon
was easily deduced from replacement of the downfield (3.29-
3.48 ppm) signals of 62b by new signals at 2.65-2.90 ppm
and from characteristic ESMS (m/z ) M + Na) and IR (2250
cm-1) data. The nitrile product 63 was never observed using
the standard acetonitrile conditions for oxazolium salt formation
and ylide generation with BnMe3N+CN-. Indeed, the unreacted
starting material 62b could be recovered from incomplete
reactions in acetonitrile even though excess BnMe3N+CN- was
always used. These observations indicate that substrate com-
plexation by silver ion is different in dichloromethane compared
to acetonitrile and that the resulting activation of iodine leads
to a distinct pathway from 62b to 63.
In any event, both 66a and 66b were available in reasonable
yield using the ylide cycloaddition strategy, and further steps
would now have to contend with the solvolytically labile
teracyclic ring system. Cleavage of the acetate protecting group
in the cycloadduct 66a with NaOH in methanol provided an
alcohol 67a that was unstable on unbuffered silica gel.
Fortunately, purification of 67a was possible if the silica gel
was pretreated with triethylamine. As expected from the
experience with 53 (Scheme 6), oxidation of the sensitive
alcohol 67a to the corresponding ketone proved to be exception-
ally troublesome. Performing the oxidation with pyridine-
buffered Dess-Martin reagent46 provided the impure diketone
68a in a less than 30% yield. Because the result was not
reproducible, the reaction was investigated in more detail. Higher
purity samples of the Dess-Martin periodinane47,48 gave only
a small improvement (to 36% of 68a) after much optimization
(2 equiv of the reagent, 7 equiv of pyridine, 2 equiv of water,
CH2Cl2, room temperature).49 Given the acid sensitivity of the
alcohol 67a, the nonacidic tetrapropylammonium perruthenate
(TPAP) reagent was tried.42 Indeed, catalytic TPAP and
N-methylmorpholine N-oxide as the stoichiometric oxidant
(molecular sieves added) provided the diketone 68a in an
improved 73% yield. Once again, chromatographic purification
of 68a was difficult due to decomposition on silica gel. In
sequence with the unprotected alcohol 61a failed, so the acetate
62a was prepared. The usual conditions for AgOTf-promoted
intramolecular alkylation (MeCN; 70 °C) to form the crucial
oxazolium salt 65a were not very effective according to NMR
assay, and significant decomposition was apparent. Nevertheless,
addition of crude 65a to a solution of BnMe3N+CN- in
acetonitrile gave the desired cycloadduct 66a in a respectable
40% yield. Interestingly, the 1H NMR spectrum of 66a and all
of the subsequent tetracyclic intermediates displayed an ad-
ditional minor set of signals. Both sets of NMR signals exhibited
similar features except for a marked difference (0.5-1 ppm) in
the chemical shifts of the N-methylaziridine signals, thereby
suggesting invertomers at the aziridine nitrogen. The invertomer
ratio varied from 1:10 to 1:30 depending on the substitution
pattern of the aziridinomitosene ring system. Notably, no
invertomers could be detected in the NMR spectrum of the
N-methylaziridine 62a or other intermediates preceding the
cycloaddition step.
(45) Dines, M. B. J. Organomet. Chem. 1974, 67, C55.
(46) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1991, 113, 7277.
(47) Ireland, R. E.; Liu, L. J. Org. Chem. 1993, 58, 2899.
(48) Frigerio, M.; Santagostino, M.; Sputore, S. J. Org. Chem. 1999, 64, 4537.
(49) Meyer, S. D.; Schreiber, S. L. J. Org. Chem. 1994, 59, 7549.
When the internal alkylation-cycloaddition sequence was
applied to 62b, the cycloadduct 66b was obtained in 37% yield
from the AgOTf-induced oxazolium salt formation in MeCN
9
15802 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Synthetic Enantiopure Aziridinomitosenes
A R T I C L E S
Scheme 8
deprotonation and NCS oxidation gave 70b in a modest 56%
yield, the reaction was relatively clean and repeatable.
Attempted deprotection of the silyl group in 70a with TBAF
gave a complex mixture. The alternative of using the HF-
pyridine reagent was suspect because the reagent is acidic
(commercial HF-pyridine contains ca. 90 mol % of HF and
10 mol % of pyridine) and would cleave the aziridine. However,
when the HF-pyridine reagent was buffered with triethy-
lamine,51 deprotection of 70a provided the desired alcohol 72a
in 74% yield (88% based on recovered 70a).
Finally, introduction of the carbamate group in the alcohol
72a was attempted using the trichloroacetyl isocyanate reagent
51.5c,6a,44 However, the desired carbamoylated product 74a was
not obtained. New doublets were observed in the NMR spectrum
at δ 6.37, 6.28, 6.00, and 5.90 ppm, some of which may be due
to the C-1 proton of tentative structure 73a, derived from
aziridine ring opening. Apparently, the strongly electrophilic
isocyanate regent activates the N-methylaziridine moiety for
solvolytic ring cleavage.
The exceptional sensitivity of the aziridinomitosene alcohol
72a dictated the use of a milder carbamoylating agent. Replacing
the trichloroacetyl group of 51 with an alkoxycarbonyl group
was expected to moderate reagent electrophilicity. The previ-
ously unreported Fmoc-NdCdO (Fmoc ) 9-fluorenylmethoxy-
carbonyl) was attractive because Fmoc is removed under mildly
basic conditions52 that might be compatible with the sensitive
aziridinomitosene. Gratifyingly, the reaction of alcohol 72a with
Fmoc-NdCdO cleanly provided the desired product 75a in an
excellent 89% yield.52c Furthermore, treatment of 75a with
triethylamine at room temperature in acetonitrile provided the
aziridinomitosene 11a (81%). The same techiques were then
used to convert 70b to 11b to establish generality. These studies
were limited to small scale experiments, resulting in a relatively
low overall yield (ca. 15%). However, the formation of 11b
was confirmed by comparison of spectroscopic data with those
of 11a.
After several synthetic stages involving sensitive synthetic
intermediates, the target aziridinomitosene 11a was now in hand.
Its stability proved comparable to that of the precursor quinones.
Furthermore, 11a was highly crystalline and the pure material
could be stored with relative ease. On the other hand, solutions
of the substance in protic solvents were sensitive to decomposi-
tion as expected from the precedents with naturally derived
aziridinomitosenes.15 Unlike any of these precedents, 11a has
no substituents at the quinone carbons C-6 and C-7. Therefore,
the stability profile of this unusual aziridinomitosene was
investigated to learn about the limits for its survival.
contrast to the other tetracyclic intermediates mentioned earlier,
the decomposition of 68a was accelerated by buffering the silica
gel with triethylamine, presumably via base-catalyzed enoliza-
tion to the highly reactive hydroquinone intermediate 69a
(Scheme 8). We therefore explored ways to conduct the
conversion of crude 68a directly to quinone 70a in the hope
that the latter would be more stable.
Initially, the oxidation of 68a was performed using diazabi-
cyclo[2.2.2]octane (DABCO) in acetonitrile under oxygen
(Scheme 8). The reaction required heating at 70 °C, and only
ca. 40% yield of the desired product 70a was obtained. In an
attempt to lower the reaction temperature, a catalytic amount
of cobalt(II) salen complex was added.50 This improved the
oxidation from 68a significantly and provided the quinone 70a
in a satisfactory 80% yield after 2 h at room temperature.
Although 70a is a sensitive molecule that must be handled with
considerable care, it proved to be significantly more stable than
68a and could be purified as well as fully characterized.
The solvolytic stability of 11a was evaluated in methanolic
solutions buffered with the Tris and Bis-Tris amine hydrochlo-
rides. Solvolysis was conveniently monitored by UV spectros-
copy at “pH” 6.0 (“pH” refers to methanol conditions), and a
well-defined isosbestic point at 424 nm was observed. A similar
UV spectrum was obtained if the solvolysis of 11a was
performed at “pH” 7.0 although no isosbestic points were seen,
suggesting interference from unidentified minor decomposition
The optimized procedures were then applied to the b series.
Thus, 68b was obtained in 68% yield after saponification, TPAP
oxidation, and carefully controlled purification and workup.
Initial attempts to conduct the cobalt-catalyzed oxidation to
quinone 70b proved difficult on small scale, so the alternative
of oxidation via the dienolate was tested, based on the precedent
of 48 to 49 (Scheme 6). Although the sequence of KHMDS
(51) Nystro¨m, J.-E.; McCanna, T. D.; Helquist, P.; Iyer, R. S. Tetrahedron Lett.
1985, 26, 5393.
(52) (a) Carpino, L. A.; Han, G. Y. J. Am. Chem. Soc. 1970, 92, 5748. (b)
Carpino, L. A. Acc. Chem. Res. 1987, 20, 401. (c) For commentary on the
surprisingly facile cleavage of 75a, see: West, C. W.; Estiarte, M. A.;
Rich, D. H. Org. Lett. 2001, 3, 1205.
(50) van Dort, H. M.; Geursen, H. J. Rec. TraV. Chim. 1967, 86, 520. Coord.
Chem. ReV. 1987, 79, 321.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15803

A R T I C L E S
Vedejs et al.
Table 1. Observed First-Order Rate Constants and Half-Lives for
Methanolysis of Aziridinomitosenes 11a, 7, and 8 at 20 °C
11a
815a
715a
“pH”
k1 (s-1
)
t1/2 (min)
k1 (s-1
)
t1/2 (min)
k1 (s-1
)
t1/2 (min)
6.0 3 × 10-5 400
8.5 decomp. decomp. 3.6 × 10-5
7.0 5 × 10-6 2000
8 × 10-4
14
324
4.1 × 10-3
5.1 × 10-5
3
228
Scheme 9
Figure 1. (A) Autoradiogram of UvrABC nuclease cutting of N-methyl-
7-methoxyaziridinomitosene (9; MS-NMA)-modified and synthetic N-
methylaziridinomitosene (11a; MS-NM)-modified 3′ end 32P-labeled BstNI-
EcoRI 129-bp fragment from pBR322 plasmid. Lanes 1-3, Maxam-Gilbert
chemical sequencing reactions of AG, CT, and G, respectively; lane 4, DNA
treated with 11a (MS-NM) alone (control); lane 5, unmodified DNA treated
with UvrABC (control); lane 6, DNA modified with 1.5 mM 9 (MS-NMA)
after incubation at 37 °C (2 h); lanes 7-10, DNA modified with 1.5 mM
11a (MS-NM) after incubation at 37 °C (1 h), 37 °C (2 h), 37 °C (3 h),
and 37 °C (4 h). The band corresponding to 3′-end labeled BstNI-EcoRI
129-base fragment is labeled O. Bands above O have been attributed to
higher molecular weight DNA products and those below corresponded to
UvrABC nuclease incision adducts. (B) Same as Figure 1A except only
the central portion of the 129-base fragment is shown. The drug modification
induced UvrABC nuclease incision bands (U1-U17) are labeled on the
right side of the panel, and the numbers (1-17) corresponding to the guanine
modification sites are provided on the left side of the panel.
pathways. The reaction rate was followed by measuring the
time-dependent decrease in UV absorption at 480 nm, and first-
order rate constants were obtained (Table 1). For comparison,
Table 1 also includes the first-order rate constants reported by
Kohn and Han15a for methanolysis of the aziridinomitosenes 7
and 8 under similar conditions. Interestingly, methanolysis of
the synthetic aziridinomitosene 11a at “pH” 7.0 was ca. 160
times slower compared to that of 8. The reactivity difference
may be a result of electron donation from the C(7) amino group
of 8 into the quinone π-system. If this delocalization effect
increases electron density in the carbocyclic subunit of 8
compared to 11a, then the indole nitrogen in 79 would be better
able to stabilize the carbocation intermediate compared to the
situation in 78 (Scheme 9). The result would be to facilitate
the SN1 aziridine ring opening in 77 compared to 76.
When the methanolysis of 11a was performed on a prepara-
tive scale at “pH” 5.8, product 80 was isolated in 41% yield as
a ca. 1:1.2 mixture of the cis/trans isomers. The low yield
reflects the instability of the initially formed 80. Although
methanolysis proceeds cleanly at “pH” <7 according to TLC
analysis, significant decomposition occurs as the sample is
concentrated.
In contrast to the aziridinomitosenes 7 and 8, synthetic 11a
is sensitive to decomposition at “pH” 8.5 in a methanolic buffer
solution over ca. 10 h at room temperature. No individual
component could be isolated from the complex mixture of polar
products. A strong UV absorption band at λ 291 nm emerged
in the course of the reaction, reminiscent of the λ 286 nm
maximum reported for 7-methoxyaziridinomitosene 6.53 If this
analogy holds, then base-catalyzed decomposition of the aziri-
dinomitosene 11a may be initiated by 1,4-addition of an oxygen
nucleophile at C-6 or C-7 of the quinone. This decomposition
pathway is blocked by substituents in the naturally derived
aziridinomitosenes, but not in the synthetic analogue.
The ability of 11a (MS-NM in Figures 1A, 1B, and 2) to
modify DNA was also explored. The key experiment was
conducted under nonreductiVe conditions using the 3′ end 32P-
labeled BstNI-EcoRI 129-bp fragment from pBR322 plasmid.54
In these initial studies, we employed 9 (N-methyl-7-meth-
oxyaziridinomitosene; MS-NMA in Figures 1A, 1B, and 2)
as a control substrate because 9 efficiently and selectively
modifies DNA to give only mitosene-DNA monoadducts under
these conditions.15 The site of DNA alkylation was determined
using the UvrABC nuclease assay.54b,55
The aziridinomitosenes were incubated (37 °C, 1-4 h) with
the radiolabeled DNA, and then the DNA was separated from
the reaction mixture and treated with UvrABC nuclease. Figure
(54) (a) Li, V.-S., Kohn, H., J. Am. Chem. Soc. 1991, 113, 275. (b) Kohn, H.;
Li, V.-S.; Tang, M.-s. J. Am. Chem. Soc. 1992, 114, 5501.
(55) Friedberg, E. C.; Walker, G. C.; Siede, W. ASM Press: Washington, DC,
1955. Sancar, A.; Tang, M.-s. Photochem Photobiol. 1993, 57, 905.
(53) Iyengar, B. S.; Remers, W. A.; Bradner, W. T. J. Med. Chem. 1986, 29,
1864.
9
15804 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003

Synthetic Enantiopure Aziridinomitosenes
A R T I C L E S
Figure 3. Synthetic aziridinomitosene 82.
nuclease treatment (Supporting Information, Figure 1, lanes
8-16). We have tentatively attributed the slower moving bands
to mitosene-DNA cross-linked and/or mitosene-DNA-protein
adducts. The histogram for the UvrABC nuclease incision
products for 11a (MS-NM; Figure 1B, lanes 7, 8) is provided
in Figure 2 (panels b and c). We observed that DNA adduction
occurred only at guanine (G*) and that the reaction proceeded
at both 5′-CG* and 5′-TG* sites to a greater extent than 5′-
GG* and 5′-AG* loci. Moreover, upon comparing the DNA
profiles of 9, 11a, and 82 (Figure 3), a mitosene lacking the
C(10) substituent,56 we found that both 9 and 11a showed
increased DNA selectivity compared with 82. This finding
supports earlier results demonstrating that the mitosene C(10)
oxygen substituent facilitates preferential mitomycin-DNA
bonding.57
Our preliminary studies document that 11a readily modifies
DNA under nonreductive conditions. Similar to mitomycin C
(3), and other mitomycins and mitosenes, 11a-DNA adduction
occurred at guanine (G*) sites. Unlike previous mitomycins and
mitosenes, higher molecular weight adducts were observed under
nonreductive conditions. These products appeared both with and
without UvrABC nuclease treatment, and one of these bands
increased upon incubation with the nuclease. Higher molecular
weight products have been associated with drug-DNA cross-
linked adducts and drug-DNA-protein conjugates.58 This
finding is consistent with the structure of 11a because drug
activation can proceed by a solvolytic pathway, and adduction
may proceed at multiple sites (C(1), C(6), C(7), C(10)).
Figure 2. Relative intensities (RI) of UvrABC nuclease incision sites of
the 81-base region within 3′ end 32P-labeled BstNI-EcoRI 129-bp sequence
from pBR322 plasmid spanning G3-G17. Panels a (9; MS-NMA) and b,
c (11a; MS-NM) correspond to Figure 1B lanes 6, 7, and 8, respectively.
The intensity of UvrABC nuclease incision bands in Figure 1, lanes 6, 7,
and 8 were scanned in a Bio-Image Analyzer. The intensities were
normalized to 100% for the most intense band (100) within each experiment.
Summary
1A shows the autoradiogram for the full-length gel and Figure
1B provides a blow-up of the central region of the 129-bp
fragment. The drug-DNA bonding-induced UvrABC nuclease
incision bands are labeled U1-U17, which corresponds to
modification at guanines 1-17, respectively.
An enantiocontrolled route to aziridinomitosenes has been
developed. The longest linear sequence is 20 steps (2.9% overall
yield) starting from L-serine methyl ester hydrochloride. Several
unusual techniques were developed in the course of this project;
most importantly, the internal azomethine ylide cycloaddition
reaction based on silver ion-assisted intramolecular oxazole
alkylation. Other important developments include the N-
methylation of N-H aziridines in the aziridinomitosene environ-
ment,59 deprotection of N-trityl aziridines under reductive
conditions, oxidation of sensitive cyclohexenediones to quino-
Mitosene-DNA adduction was observed for both compounds
at 37 °C within 2 h. For the naturally derived 9, we detected no
appreciable amounts of DNA products that correspond to adduct
molecular weights higher than the starting 129-base DNA
radiolabeled fragment (Figure 1A, lane 6). UvrABC nuclease
treatment of the 9-DNA modified sample provided a DNA
bonding profile in agreement with earlier findings (Figure 2,
panel a). Only the guanine (G*) bases were modified and DNA
adduction had occurred preferentially at 5′CG* sites. By
comparison, the UvrABC-treated samples of synthesis-derived
11a showed extensive amounts of high molecular weight DNA
products along with bands associated with UvrABC-incised
adducts (Figure 1A, lanes 7-10). Increased incubation times
(1-4 h) for 11a gave a greater percentage of the radiolabeled
DNA adhered to the siliconized Eppendorf tubes (1 h, ∼ 13%;
2 h, ∼ 20%; 3 h, ∼48%; 4 h, ∼ 55%). Reduction of the reaction
temperature from 37 to 22 °C led to lower amounts of the higher
molecular weight products. Furthermore, we observed that one
of the higher molecular weight bands increased with UvrABC
(56) Wang, Z.; Jimenez, L. S. J. Org. Chem. 1996, 61, 816.
(57) Li, V.-S.; Choi, D.; Wang, Z.; Jimenez, L. S.; Tang, M.-s.; Kohn, H. J.
Am. Chem. Soc. 1996, 118, 2326.
(58) (a) Van Houten, B.; Illenye, S.; Qu, Y.; Farrell, N. Biochemistry 1993, 32,
11794. (b) Kane, S. A.; Lippard, S. J. Biochemistry 1996, 35, 2180.
(59) A similar procedure has been used to prepare N-methylaziridinomito-
sanes: Danishefsky, S. M.; Berman, E. M.; Ciufolini, M.; Etheredge, S.
J.; Segmuller, B. E. J. Am. Chem. Soc. 1985, 107, 3891.
(60) Solladie´, G.; Almario, A. Tetrahedron: Asymmetry 1995, 6, 559.
(61) Imamoto, T.; Takiyama, N.; Nakamura, K.; Hatajima, T.; Kamiya, Y. J.
Am. Chem. Soc. 1989, 111, 4392. Evans, W. J.; Feldman, J. D.; Ziller, J.
W. J. Am. Chem. Soc. 1996, 118, 4581. Dimitrov, V.; Kostova, K.; Genov,
M. Tetrahedron Lett. 1996, 37, 6787.
(62) Loev, B.; Kormendy, M. F. J. Org. Chem. 1963, 28, 3421.
(63) Araki, Y.; Konoike, T. J. Org. Chem. 1997, 62, 5299.
(64) Carpino, L. A.; Mansour, E. M. E.; Knapczyk, J. J. Org. Chem. 1983, 48,
666. Loev, B.; Kormendy, M. F. J. Org. Chem. 1963, 28, 3421.
(65) Thomas, D. C.; Levy, M.; Sancer, A. J. Biol. Chem. 1985, 260, 9875.
9
J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003 15805

A R T I C L E S
Vedejs et al.
nes, and a mild procedure for carbamoylation using Fmoc-NCO.
Diverse experimental conditions, including protecting group
manipulations and redox chemistry, have been developed that
tolerate the presence of the sensitive aziridinomitosene tetra-
cycle. Finally, a preliminary exploration of DNA alkylation by
the first C-6,C-7-unsubstituted aziridinomitosene 11a has been
carried out. This study suggests a new pathway for DNA cross-
link formation involving nucleophilic addition to the quinone
subunit.
Acknowledgment. This work was supported by NIH
(CA17918).
Supporting Information Available: Experimental details and
spectra of compounds studied (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
(66) Tang, M.-s. In Technologies for Detection of DNA Damage and Mutations;
Pfeifer, G. P., Ed.; Plenum Press: New York, 1996; pp 137-151.
(67) (a) Maxam, A. M.; Gilbert, W. Method Enzymol. 1980, 65, 499. (b) Maxam,
A. M.; Gilbert, W. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 560.
JA030452M
9
15806 J. AM. CHEM. SOC. VOL. 125, NO. 51, 2003