Synthesis of the Aziridinomitosene Skeleton by Intramolecular Michael Addition of α-Lithioaziridines: An Aromatic Route Featuring Deuterium as a Removable Blocking Group

Article abstract of DOI:10.1021/jo030223i

A convergent synthetic route to the 1,2-aziridinopyrrolo(1,2-a)indole 34 has been developed. Key features of this route include the deuterium kinetic isotope effect to block undesired indole lithiation during tin-lithium exchange from 27a to 30a, the intr

Full text of DOI:10.1021/jo030223i

Syn th esis of th e Azir id in om itosen e Sk eleton by In tr a m olecu la r
Mich a el Ad d ition of r-Lith ioa zir id in es: An Ar om a tic Rou te
F ea tu r in g Deu ter iu m a s a Rem ova ble Block in g Gr ou p
Edwin Vedejs* and J eremy D. Little
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
edved@umich.edu
Received J uly 14, 2003
A convergent synthetic route to the 1,2-aziridinopyrrolo(1,2-a)indole 34 has been developed. Key
features of this route include the deuterium kinetic isotope effect to block undesired indole lithiation
during tin-lithium exchange from 27a to 30a , the intramolecular Michael addition to generate
the enolate 31a , and conversion into 34 by trapping with phenylselenenyl chloride. Reductive
cleavage of the N-trityl group in 34 allows access to tetracyclic aziridinomitosenes containing the
aziridine N-H subunit. Reduction of the C(9) ester in 34 with LAH gives the primary alcohol 35
with the correct C(9), C(9a), C(10) oxidation state corresponding to the aziridinomitosenes, and
deprotection of 34 affords 37.
The unusual oxygen-bridged benzoazocine antitumor
agents FR66979 (1) and FR900482 (2) form interstrand
DNA cross-links upon reductive activation.1,2 Solvolyti-
cally labile intermediates 5 and 6 are believed to explain
the activity,3 by analogy to the mode of action of mito-
mycin C.4 The semisynthetic 3 (FK973) is no longer
regarded as a promising drug candidate due to vascular
leak syndrome (VLS),5 but renewed interest in related
compounds has been stimulated by reports that the
closely related FK317 (4) retains activity, but does not
have the VLS side effect.6-8
thetic approaches to the tetracyclic skeleton of aziridi-
nomitosenes have been reported,10-26 including several
(8) (a) Naoe, Y.; Inami, M.; Matsumoto, S.; Takagaki, S.; Fujiwara,
T.; Yamazaki, S.; Kawamura, I.; Nishigaki, F.; Tsujimoto, S.; Manda,
T.; Shimomura, K. J pn. Cancer Res. 1998, 89, 1306. (b) Naoe, Y.; Inami,
M.; Nishigaki, F.; Tsujimoto, S.; Matsumoto, S.; Manda, T.; Shimo-
mura, K. J pn. Cancer Res. 1998, 89, 666.
(9) 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.
(10) Danishefsky, S. J .; Schkeryantz, J . M. Synlett 1995, 475.
(11) (a) Shaw, K. J .; Luly, J . R.; Rapoport, H. J . Org. Chem. 1985,
50, 4515. (b) Wang, Z.; J imenez, L. S. J . Org. Chem. 1996, 61, 816. (c)
Edstrom, E. D.; Yu, T. Tetrahedron 1997, 53, 4549. (d) Dong, W.;
J imenez, L. S. J . Org. Chem. 1999, 64, 2520.
(12) Hirata, T.; Yamada, Y.; Matsui, M. Tetrahedron Lett. 1969, 10,
19. Hirata, T.; Yamada, Y.; Matsui, M. Tetrahedron Lett. 1969, 10,
4107.
Both 5 and 6 contain the aziridinopyrrolo(1,2-a)indole
ring system, a tetracyclic nucleus that was originally
encountered in the mitomycin antibiotics.9,10 Many syn-
(1) (a) Kiyoto, S.; Shibata, T.; Yamashita, M.; Komori, T.; Okuhara,
M.; Terano, H.; Kohsaka, M.; Aoki, H.; Imanaka, H. J . Antibiot. 1987,
40, 594. (b) Kiyoto, S.; Shibata, T.; Yamashita, M.; Komori, T.;
Okuhara, M.; Terano, H.; Kohsaka, M.; Aoki, H.; Imanaka, H. J .
Antibiot. 1989, 42, 145.
(13) Siuta, G. J .; Franck, R. W.; Kempton, R. J . J . Org. Chem. 1974,
39, 3739 and references therein.
(14) Cory, R. M.; Ritchie, B. M. J . Chem. Soc., Chem. Commun. 1983,
1244.
(2) Shimomura, K.; Hirai, O.; Mizota, T.; Matsumoto, S.; Mori, J .;
Shibayama, F.; Kikuchi, H. J . Antibiot. 1987, 40, 600.
(15) Nakatsuka, S.; Asano, O.; Goto, T. Heterocycles 1987, 26, 2603.
(16) Egbertson, M.; Danishefsky, S. J . J . Am. Chem. Soc. 1987, 109,
2204.
(17) Feigelson, G. B.; Danishefsky, S. J . J . Org. Chem. 1988, 53,
3391.
(18) (a) J ones, G. B.; Moody, C. J . J . Chem. Soc., Chem. Commun.
1988, 1009. (b) J ones, G. B.; Moody, C. J .; Padwa, A.; Kassir, J . M. J .
Chem. Soc., Perkin Trans. 1 1991, 1721.
(19) (a) Fuji, M.; Muratake, H.; Natsume, M. Chem. Pharm. Bull.
1992, 40, 2344. (b) Utsunomiya, I.; Fuji, M.; Sato, T.; Natsume, M.
Chem. Pharm. Bull. 1993, 41, 854.
(3) (a) Fukuyama, T.; Goto, S. Tetrahedron Lett. 1989, 30, 6491. (b)
Huang, H.; Rajski, S. R.; Williams, R. M.; Hopkins, P. B. Tetrahedron
Lett. 1994, 35, 9669. (c) Paz, M.; Hopkins, P. B. Tetrahedron Lett. 1997,
38, 343. (d) Paz, M. M.; Hopkins, P. B. J . Am. Chem. Soc. 1997, 119,
5999. (e) Huang, H.; Pratum, T. K.; Hopkins, P. B. J . Am. Chem. Soc.
1994, 116, 2703. (f) Williams, R. M.; Rajski, S. R.; Rollins, S. B. Chem.
Biol. 1997, 4, 127. (g) J udd, T. C.; Williams, R. M. Org. Lett. 2002, 4,
3711.
(4) Shimomura, K.; Manda, T.; Mukumoto, S.; Masuda, K.; Naka-
mura, T.; Mizota, T.; Matsumoto, S.; Nishigaki, F.; Oku, T.; Mori, J .;
Shibayama, F. Cancer Res. 1988, 48, 1166.
(20) Utsunomiya, I.; Muratake, H.; Natsume, M. Chem. Pharm. Bull.
1995, 43, 37.
(5) (a) Majima, H.; Hasegawa, K.; Fukuoka, M.; Furuse, K.; Wakui,
A.; Furue, H.; Masaoka, T.; Hattori, T.; Taguchi, T.; Ogura, T.; Niitani,
H. Proc. Am. Soc. Clin. Oncol. 1990, 9, 78. (b) Pazdur, R.; Ho, D. H.;
Daugherty, K.; Bradner, W. T.; Krakoff, I. H.; Raber, M. N. Invest.
New Drugs 1991, 9, 337.
(6) Naoe, Y.; Inami, M.; Matsumoto, S.; Nishigaki, F.; Tsujimoto,
S.; Kawamura, I.; Miyayasu, K.; Manda, T.; Shimomura, K. Cancer
Chemother. Pharmacol. 1998, 42, 31.
(7) (a) Yoshimura, A.; Furuse, K.; Hasegawa, K.; Kudo, S.; Niitaru,
H. Proc. Annu. Meet. Am. Soc. Clin. Oncol. 1999, 18, A689. (b) Ravandi,
F.; Wenske, C.; Royce, M.; Hoff, P.; Brito, R.; Zukowski, T.; Mekki, Q.;
Pazdur, R. Proc. Annu. Meet. Am. Soc. Clin. Oncol. 1999, 18, A867.
(21) Lee, S.; Lee, W.-M.; Sulikowski, G. A. J . Org. Chem. 1999, 64,
4224.
(22) Michael, J . P.; de Koning, C. B.; Petersen, R. L.; Stanbury, T.
V. Tetrahedron Lett. 2001, 42, 7513.
(23) (a) Preliminary report: Vedejs, E.; Little, J . J . Am. Chem. Soc.
2002, 124, 748. (b) Vedejs, E.; Little, J . A.; Seaney, L. M. J . Org. Chem.
2003, 68, 0000.
(24) (a) Vedejs, E.; Piotrowski, D. W. J . Org. Chem. 1993, 58, 1341.
(b) Vedejs, E.; Monahan, S. D. J . Org. Chem. 1997, 62, 4763. (c) Vedejs,
E.; Piotrowski, D. W.; Tucci, F. C. J . Org. Chem. 2000, 65, 5498. (d)
Vedejs, E.; Klapars, A.; Naidu, B. N.; Piotrowski, D. W.; Tucci, F. C.
J . Am. Chem. Soc. 2000, 122, 5401.
10.1021/jo030223i CCC: $27.50 © 2004 American Chemical Society
Published on Web 01/15/2004
1794
J . Org. Chem. 2004, 69, 1794-1799

Synthesis of the Aziridinomitosene Skeleton
indole structures 7-12 that resemble 5 or 6.12-22 How-
ever, highly reactive indoles containing the natural
substitution pattern of 5 or 6 have rarely been observed,
and none have been prepared by total synthesis. The only
isolable substance having the correct oxidation state at
C(3), C(5), C(8), and (especially) C(9a)-C(9)-C(10) is the
semisynthetic 13a reported by Danishefsky and Egbert-
son.16 Among the fully synthetic structures 7-12, dif-
ficulties in removing the protecting groups at aziridine
nitrogen may be inferred from the absence of reports
where deprotection has been accomplished at the indole
stage. Most of the synthetic structures also contain
stabilizing carbonyl groups, as in the vinylogous carbam-
ates 7,12 9,22 or 12,21 the lactam 8,13 and the vinylogous
formamides 13b17 or 11.19,20 These compounds have lower
solvolytic reactivity at the “benzylic” aziridine C(1)-N
bond because of the electron-withdrawing effect of the
conjugated carbonyl group.
SCHEME 1
Stability is significantly improved in the absence of the
activating C(9),C(9a) double bond, as in the indoline
(aziridinomitosane) structures prepared by Ziegler and
Belema,25 or in the natural mitomycins and related
molecules encountered in total synthesis efforts.10,26
A
number of aziridinomitosene quinones of the general
structure 14 are also known.11,24 Because of the strong
electron-withdrawing effect of the quinone carbonyl
groups and vinylogous amide delocalization, the quinones
14 are relatively stable compared to 13. Nevertheless,
all of the aziridines having an adjacent pyrrolic double
bond are easily cleaved in acidic hydroxylic media. This
is the reason synthetic analogues of 5 or 6 have been so
difficult to prepare. Accordingly, we have initiated a
program designed to access structures 5 and 6. Progress
toward this goal is described below in the context of a
synthesis of tetracyclic indole derivatives having the
correct oxidation pattern at C(9a)-C(9)-C(10).
SCHEME 2
Resu lts a n d Discu ssion
Initial studies focused on determining the feasibility
of the intramolecular Michael addition of an R-lithio-
aziridine 18 to a suitably activated indole subunit
(Scheme 2). A parallel approach with nonaromatic pre-
cursors has been evaluated in the preceding article, and
has encountered partial success as well as major difficul-
ties in the cyclization step.23 In the current approach, a
fully aromatic (indole) precursor 17 was to be prepared
from the same aziridinyl mesylate 15a or nosylate 15b
alkylating agents used in the nonaromatic approach.23b
The key issue was whether the internal Michael addition
from 18 would generate an enolate 19 that might be
captured by external electrophiles en route to potential
targets 20. Success was not assured in view of the
competing lithiation adjacent to the pyrrolic nitrogen that
had been observed in the nonaromatic series.23b Other
important issues to be resolved include removal of the
N-trityl protecting group and reduction of the C(10) ester
to the correct alcohol oxidation state. Neither conversion
has been reported in aziridinomitosenes containing an
indole subunit.
The indole ester 16a was conveniently prepared from
the commercially available acid 21 by O-alkylation of the
cesium salt with ethyl iodide (73% isolated; Scheme 3).27
Subsequent coupling with nosylate 15b followed the
precedent developed in the nonaromatic series.23b Thus,
a THF/DMPU solution of 16a was deprotonated at -78
°C with NaN(TMS)2 and addition of 15b in THF gave a
deep purple solution. After warming to room temperature
and heating at 60 °C, 17a was obtained in 40% yield,
together with ca. 20% of the alcohol corresponding to
nosylate S-O bond cleavage in 15b. Marginally higher
yields of 17a (47%) were achieved when the mesylate 15a
was used in place of 15b, but the coupling was slower
and heating for 3 days at 60 °C was required. The same
procedures were also used to prepare the nitrile and
sulfone analogues 17b and 17c with similar results,
starting from the commercially available nitrile 16b and
(25) (a) Ziegler, F.; Belema, M. J . Org. Chem. 1994, 59, 7962. (b)
Ziegler, F. E.; Belema, M. J . Org. Chem. 1997, 62, 1083.
(26) (a) Papaioannou, N.; Blank, J . T.; Miller, S. J . J . Org. Chem.
2003, 68, 2728. (b) Colandrea, V. J .; Rajaraman, S.; J imenez, L. S.
Org. Lett. 2003, 5, 785.
(27) Pfeffer, P.; Silbert, L. J . Org. Chem. 1976, 41, 1373.
J . Org. Chem, Vol. 69, No. 6, 2004 1795

Vedejs and Little
SCHEME 3
SCHEME 4
the known sulfone 16c, prepared from 22 by oxidation
with oxone.28,29
With a route to cyclization substrates 17 in place,
studies of the tin-lithium exchange and internal Michael
addition were initiated in the ester series, Scheme 4, expt
1. Treatment of a cold solution of 17a in THF with excess
methyllithium (4 equiv, 15 min) followed by an ethanol
quench resulted in a mixture of starting material (17a ),
destannylated 23a , and the desired tetracycle 24a , in a
47:39:14 ratio, respectively. The two diastereomers 24
could not be separated from 23 and the NMR spectrum
of the mixture proved difficult to interpret. Eventually,
the NMR signals for one diastereomer of 24 were deduced
from a combination of decoupling studies and insights
gained from deuterium labeling as discussed below.
Specific signals were identified corresponding to He (δ
3.85 ppm, d, J ) 11 Hz) and Hf (δ 3.25 ppm, d, J ) 11,
4 Hz), the aziridine protons Hc (δ 2.12 ppm, d, J ) 5 Hz)
and Hd (δ 2.35 ppm, dd, J ) 5, 4 Hz), the C-9 proton Ha
(δ 4.05 ppm, d, J ) 7 Hz), and the C-9a proton Hb (δ 4.78
ppm, d, J ) 7 Hz), and confirmed the tetracyclic skeleton
as well as retention of configuration at the aziridine.
The formation of 23a as the major product in expt 1
was initially taken to mean that some form of competing
proton transfer had compromised the experiment. When
conversion to 24 could not be improved by minimizing
hydroxylic contaminants, nor by optimizing time or
temperature variables, an experiment was performed
with EtOD to quench the reaction and to test for the
survival of organolithium intermediates, expt 2. This
procedure revealed that lithiated species 25a and 26a
are formed in competition with the desired tin-lithium
exchange from 17a to 18a . A mixture of deuterium-
labeled products 27a , 28a , and 29a was obtained in a
ratio of 55:36:9 according to 1H NMR assay. Significantly,
the destannylated product 28a was obtained with the
indole proton completely replaced by deuterium, while
the tetracyclic 29a was formed with the original indole
proton intact. By implication, cyclization to 29a , or 24a
in expt 1, is possible only to the extent that tin-lithium
exchange to 18a precedes competing events such as the
formation of 25a . The latter process dominates, as
evidenced by the isolation of 27a as the major product
in expt 2. The desired intramolecular Michael addition
from 18a to 19a may well be an efficient process, but
the overall conversion from 17a would be practical only
if the competing lithiation to 25a can be suppressed.
Optimization of the organolithium reagent was inves-
tigated with the aim of improving the ratio of tin-lithium
exchange to 18a vs indole lithiation to 25a . After
considerable effort, no better reagent for the intended
purpose was found. Instead, an efficient procedure for
the indole lithiation was encountered. Thus, treatment
of 17a with phenyllithium (PhLi) at -65 °C for 15 min,
followed by subsequent trapping with EtOD, resulted in
the deuterated product 27a in 89% yield (Scheme 5). No
evidence for tin-lithium exchange was seen in this
reaction, even though PhLi has been used in analogous
exchange reactions.30 Similarly, tin-lithium exchange did
not occur with n-butyllithium, the reagent most com-
monly used for such transformations,31 nor with sec-
(28) Broggini, G.; Diliddo, D.; Zecchi, G. J . Heterocycl. Chem. 1991,
28, 89.
(29) Marino, J . P.; Kim, M.-W.; Lawrence, R. J . Org. Chem. 1989,
54, 1782.
(30) Reich, H.; Phillips, N. Pure Appl. Chem. 1987, 59, 1021.
(31) Wardell, J . L. In Comprehensive Organometallic Chemistry;
Wilkinson, G., Ed.; Pergamon: Oxford, UK, 1982; Chapter 2.
1796 J . Org. Chem., Vol. 69, No. 6, 2004

Synthesis of the Aziridinomitosene Skeleton
SCHEME 5
sis of the starting material zone recovered by chroma-
tography revealed that the deuterium blocking group had
been lost (as in 17a ). The destannylated 23a had also
lost the deuterium label, as expected if lithiation of the
indole ring prior to tin-lithium exchange had occurred
to some extent, and had prevented cyclization of ca. 20%
of the material. However, this process was now the minor
pathway, in striking contrast to the results of experi-
ments 1 or 2, Scheme 4. If the tricyclic byproducts 27a
and 28a in expt 2 and the analogous products 17a and
23a in expt 3 are formed only via the lithiated indole 25a ,
then the kinetic deuterium isotope effect for indole
lithiation can be estimated as kH/kD ) ca. 35. This
kinetic effect is responsible for the inversion in tetra-
cyclic:tricyclic product ratios from 1:10 in expt 2 to ca.
4:1 in expt 3.
Another cyclization experiment was performed to
confirm that the internal Michael reaction affords a long-
lived enolate intermediate as we have assumed in the
above discussion. Treatment of 27a with MeLi at -65
°C, followed by addition of methoxymethyl chloride and
warming to room temperature gave an alkylation product
33 in 58% yield (2:1 diastereomer mixture). Although this
compound was not elaborated further, its formation
demonstrates that the deuterium-containing enolate 31a
is the intermediate in the cyclization, and can be trapped
by typical electrophiles as expected.
With the feasibility of the Michael addition strategy
demonstrated for ester 17a , studies were extended to
evaluate other electron-withdrawing groups (CN, SO2-
Ph) in the analogous substrates 17b and 17c that had
been prepared as described earlier in connection with
Scheme 3. A preliminary attempt to achieve cyclization
from 17b with methyllithium was not promising, so the
deuterium blocking strategy was employed. However,
unexpected complications were encountered in the lithia-
tion of 17b with PhLi (Scheme 6). Quenching the orga-
nolithium intermediates with EtOD gave the expected
product 27b (46%), but a substantial amount of de-
stannylated 28b (27%) was also formed via premature
tin-lithium exchange to generate 26 as well as 25.
Although a clean indole ring deuteration was not
achieved, sufficient 27b was obtained after chromato-
graphic separation to explore tin-lithium exchange and
conversion to tetracyclic products. Reaction of 27b with
2 equiv of MeLi at -65 °C followed by ethanol quenching
produced a mixture of 23b and 32b in a 1:3 ratio in favor
of the cyclized product (67% combined yield). Under the
conditions optimized for the ester analogue 27a (4 equiv
of MeLi), nitrile 27b gave relatively more of the de-
stannylated product 23b. This result reflects facile
conversion from 25b to 26b in the nitrile series, as also
observed in the phenyllithium experiment starting from
17b. In view of this disadvantage compared to the ester
sequence from 17a or 27a , the nitrile series was not
pursued further.
butyllithium or tert-butyllithium. Evidently, the stannyl-
aziridine environment is too hindered for facile tin-
lithium exchange with reagents other than MeLi, a trend
that was also observed in the nonaromatic series.23b
Several blocking strategies were contemplated that
might prevent the competing formation of 25a and 26a .
By far the simplest method emerged when we realized
that a potential “blocking group” had already been
introduced in the sequence where 17a was treated with
phenyllithium, followed by quenching with EtOD to give
27a . The stronger C-D bond in 27a should resist
lithiation due to the well-known kinetic isotope effect.32
In the key experiment (3), Scheme 5, 27a was reacted
with excess MeLi at -65 °C, followed by quenching with
EtOH. This gave a mixture of 17a , 23a , and 32a in a
5:17:78 ratio (Scheme 5), corresponding to a dramatically
improved 71% yield of the desired tetracycle 32a . Analy-
(32) (a) Knaus, G.; Meyers, A. I. J . Org. Chem. 1974, 39, 1192.
Warmus, J . S.; Rodkin, M. A.; Barkley, R.; Meyers, A. I. J . Chem. Soc.,
Chem. Commun. 1993, 1357. Kopach, M. E.; Meyers, A. I. J . Org.
Chem. 1996, 61, 6764. (b) Clive, D. L. J .; Khodabocus, A.; Cantin, M.;
Tao, Y. J . Chem. Soc., Chem. Commun. 1991, 1755. Clive, D. L. J .;
Cantin, M.; Khodabocus, A.; Kong, X.; Tao, Y. Tetrahedron 1993, 49,
7917. Clive, D. L. J .; Tao, Y.; Khodabocus, A.; Wu, Y.-J .; Angoh, A. G.;
Bennett, S. M.; Boddy, C. N.; Bordeleau, L.; Kellner, D.; Kleiner, G.;
Middleton, D. S.; Nichols, C. J .; Richardson, S. R.; Vernon, P. G. J .
Am. Chem. Soc. 1994, 116, 11275. (c) Hoppe, D.; Paetow, M.; Hintze,
F. Angew. Chem., Int. Ed. Engl. 1993, 32, 394. Hoppe, D.; Kaiser, B.
Angew. Chem., Int. Ed. Engl. 1995, 34, 323. (d) Laplaza, C. E.; Davis,
W. M.; Cummins, C. C. Organometallics 1995, 14, 577. (e) Ahmed, A.;
Clayden, J .; Rowley, M. Tetrahedron Lett. 1998, 39, 6103. (f) Mathieu,
J .; Gros, P.; Fort, Y. Chem. Commun. 2000, 951. (g) Dudley, G. B.;
Danishefsky, S. J .; Sukenick, G. Terahedron Lett. 2002, 43, 5605.
Attention was turned to the sulfone 17c as a potential
substrate for cyclization (Scheme 6). Given the large
change in steric and electronic variables for the phenyl-
sulfonyl group of 17c compared to the carboethoxy (17a )
or nitrile (17b) analogues, the cyclization of 17c was
attempted without prior deuteration of the indole ring.
However, the usual conditions with excess MeLi followed
by quenching with EtOD gave no cyclized product.
J . Org. Chem, Vol. 69, No. 6, 2004 1797

Vedejs and Little
SCHEME 6
SCHEME 7
and separation of the diastereomers of tetracyclic 32 from
the destannylated 23 was not feasible in any of the
cyclization reactions. No further experiments were at-
tempted with 27b or 27c, and attention was turned to
the redox and deprotection steps that would be needed
to convert 27a into tetracyclic indole alcohols having the
substitution pattern of structure 20, mentioned earlier
(Scheme 3).
The immediate problem was to prepare an indole from
the deuterium-containing enolate 31a resulting from the
intramolecular Michael addition (Scheme 7). To this end,
the cyclization sequence from 27a was repeated, and the
enolate 31a was reacted with phenylselenenyl chloride.
No intermediates were detected. Instead, the indole
product 34 was isolated in 80% yield based on 27a , or
71% for the two steps from 17a (including deuteration
and cyclization), 19% overall (7 steps from N-tritylseri-
nal). Side products were detected in the cyclization
experiment, presumably derived from selenenylation of
anionic intermediates such as 26a and 30a , but these
minor products could not be separated or purified.
In contrast to the ethanol quench procedure used in
the initial cyclization experiments from 27a , the PhSeCl
experiment produced a single dominant product 34 that
was easy to separate from the side products. The 80%
yield of 34 is consistent with the ratio of tetracyclic 24a
or 29a compared to tricyclic products that had been
deduced for the ethanol quenching experiments with
NMR methods to analyze product mixtures, and confirms
the estimated kinetic isotope effect responsible for the
greatly improved efficiency of cyclization in the deuter-
ated series from 27a .
Contrary to expectations based on the lack of literature
precedents in the aziridinomitosene series, reduction of
34 to 35 proved to be relatively easy with lithium
aluminum hydride in diethyl ether at 0 °C (Scheme 7).
Alcohol 35 was isolated in 72% yield with use of silica
gel chromatography, although it was necessary to pre-
wash the silica plates with hexane/NEt3. Without this
precaution, facile destruction of the aziridine occurred,
presumably by heterolysis of the labile C(1)-N bond. By
comparison, the starting ester 34 was considerably more
stable, a consequence of vinylogous carbamate delocal-
ization and the electron-withdrawing effect of the ester.
For this reason, removal of the N-trityl group with use
of our recently optimized method34 was explored at the
ester stage. When ester 34 was treated with 3 equiv of
Instead, 28c was isolated in 59% yield. Clearly, the
undesired indole lithiation was even more dominant than
in the ester series (17a ) and it was necessary to use the
deuterium blocking effect. Accordingly, 17c was treated
with PhLi (2 equiv) at -78 °C, followed by EtOD to afford
27c in 90% yield. The use of more than 2 equiv of PhLi
for deuterium incorporation in this system gave lower
yields of 27c, but destannylation products were not
detected.
Tin-lithium exchange experiments with 27c proved
to be difficult and required re-optimization. The best
result was obtained at -65 °C with 1.9 equiv of MeLi
(15 min), followed by EtOH quench. This afforded a
mixture of tetracyclic 32c, recovered 17c, and destanny-
lated 23c (40%, 34%, and 8%, respectively). Thus, the
sulfonyl series proved to be the most susceptible to indole
lithiation (to 25c), although the deuterium blocking effect
was still sufficient to favor tin-lithium exchange to 30c
and eventual cyclization.33
The nitrile and sulfone cyclization experiments dem-
onstrated no preparative advantage over the ester sub-
strate 27a in the critical tin-lithium exchange process,
(33) The sulfide series corresponding to the sulfone was also briefly
investigated. No cyclization products were detected upon tin-lithium
exchange, and destannylation was the exclusive pathway.
(34) Vedejs, E.; Klapars, A.; Warner, D. L.; Weiss, A. H. J . Org.
Chem. 2001, 66, 7542.
1798 J . Org. Chem., Vol. 69, No. 6, 2004

Synthesis of the Aziridinomitosene Skeleton
Et3SiH and 3 equiv of MsOH, quenching the reaction
with excess i-Pr2NEt (1.5 h) resulted in the deprotected
aziridine 37 (65%).35 Presumably, this reaction involves
a labile ring-opened intermediate 36 that undergoes re-
cyclization upon addition of the i-Pr2NEt, a sequence of
events that was observed in a model study of deprotection
with a solvolytically sensitive aziridine substrate.34 The
conversion from 34 to 37 is the first example to our
knowledge of successful aziridine nitrogen deprotection
in the aziridinomitosene indole series.
In summary, a convergent synthetic route to 1,2-
aziridinopyrrolo(1,2-a)indole derivatives has been devel-
oped. Key features of this route include the deuterium
kinetic isotope effect to block undesired indole lithiation,
as well as the intramolecular Michael addition of a
configurationally stable C-lithioaziridine in the cycliza-
tion step. The deuterium blocking effect is precedented
in prior work where similar or even larger isotope effects
had been observed in lithiations,32 but the current
example from 27a to 34 is unusual because the deute-
rium disappears in the next step after having served its
blocking role.
Reductive cleavage of the N-trityl group allows access
to tetracyclic structures containing the aziridine N-H
subunit via a late-stage deprotection. Furthermore,
reduction of the C(9) ester moiety with LAH demon-
strates access to 35 containing the C(9) side chain at the
oxidation state found in the DNA cross-linking agents 5
and 6. Synthesis of these more highly substituted struc-
tures will require modified timing in the sequence of
reduction and deprotection steps so that the C(10)
carbamate group can be attached without interference
by the aziridine. Experiments toward this goal are under
way, and will be reported in due course.
Ack n ow led gm en t. This work was supported by the
National Institutes of Health (CA17918).
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures, characterization data, and NMR spectra of new
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
(35) According to NMR comparisons, structure 37 is identical with
the major component obtained by Michael et al. using a different
approach (ref 22). We thank Dr. Michael for providing copies of the
NMR spectra.
(36) Still, W. C.; Kahn, M.; Mitra, A. J . Org. Chem. 1978, 43, 2923.
(37) Kasˇpa´rek, S.; Heacock, R. Can. J . Chem. 1966, 44, 2805.
J O030223I
J . Org. Chem, Vol. 69, No. 6, 2004 1799