4684
E. W. Tate, S. Z. Zard / Tetrahedron Letters 43 (2002) 4683–4686
thate, acetone, rt, 2 h, 97% yield), is then reacted with
a 2.5-fold excess of 3 and a small amount of the radical
initiator dilauroyl peroxide (DLP) at high concentra-
tion (1.5 M in 4) in refluxing dichloroethane (DCE), to
give the addition product 5 in 86% yield. Transforma-
tion to the desired condensation precursor,
aminoketodiester 6, is performed in a single step by
removal of the xanthate with tributyltin hydride and
azoisobisbutyronitrile (AIBN), in 81% yield.7 Under
acidic conditions (trifluoroacetic acid (TFA)/
dichloromethane 1:1) 6 rapidly undergoes a deprotec-
tion–cyclisation–condensation cascade to give mono-
cyclic compound 7 in 99% yield, which constitutes a
formal synthesis of ( )-lupinine.8
Inspired by the success of this approach, we turned our
attention to the more complex skeleton of the indolic
Eburna alkaloids. We supposed that an indole ring
bearing a protected aminoaldehyde could be condensed
in an acid-mediated Mannich-type cyclisation, to give
an iminium ion, which would subsequently be trapped
by the indole ring system to give four of the five rings
present in the deethyleburnamine carbon skeleton; fur-
ther elaboration, as reported in the literature,3e would
then lead to deethyleburnamine itself. The key indole
aminoaldehyde could in turn be prepared by the cou-
pling of a highly functionalised xanthate with a
tryptamine-derived alkene.
The required xanthate is formed from commercially
available methoxy methacrylate by bromination with
N-bromosuccinimide (NBS) in methanol to give bro-
moacetal ester 9 (99% yield), followed by displacement
with the potassium salt of ethyl dithiocarbonate in
acetonitrile at reflux (46% yield) (Scheme 2). This last
reaction is atypical of such bromine displacements,
which usually proceed in much higher yield; in the
present case, elimination of methanol from product or
starting material competes with the relatively slow dis-
placement of a secondary somewhat hindered bromide.
The alkene coupling partner 10 is available starting
from tryptamine by a two-step allylation and protection
sequence in 77% overall yield.
Scheme 2. Attempted formation of 11.
16% yield, possibly due to more efficient trapping of the
transient secondary radical.
Whilst 12 itself represents an unusual synthesis of a
6/5/7 indolic tricycle, a method was sought for enabling
selective formation of xanthate 11 by blocking access of
the incipient secondary radical to the indole nucleus.
Protection of the nitrogen in both indoles and pyrroles
with a bulky group is a well-established method of
directing attack in ionic reactions from the 2-position
to the 3-position, and we wondered if analogous protec-
tion at the indole nitrogen might block the undesired
radical cyclisation at the 2-position by steric hindrance.
Di-BOC protected aminoindole 13 was duly prepared
from 10 and treated under the conditions described
above (200 mol% 9, 10 mol% DLP, reflux DCE), but to
our dismay we obtained di-BOC xanthate 15 in only
25% yield, with concurrent cyclisation also leading to
tricycle 16 in 25% yield (Scheme 3). Supposing that the
electron-withdrawing effect of the carbamate group was
possibly counteracting its steric influence by accelerat-
ing cyclisation, we instead prepared the tert-
butyldimethylsilyl (TBS) protected indole 14 in 69%
yield from 10 (1.2 equiv. n-BuLi, 2 equiv. TBSCl),10
and addition of this alkene to xanthate 9 under the
usual conditions gave exclusively xanthate 17 in 66%
yield, with the alkene returned in 23% yield. To our
In contrast to the radical addition described above,
reaction of xanthate 9 and 2 equiv. of alkene 10 with 10
mol% of DLP in refluxing DCE results in very low
conversion to the desired addition product 11 (15%
yield). Along with a substantial quantity of starting
xanthate 9 (ca. 40%), a tricyclic product 12 was isolated
in 24% yield, evidently resulting from radical cyclisation
onto the 2-position of the indole, followed by rearoma-
tisation. Cyclisations following xanthate addition on an
alkenic chain bearing an appropriately placed aromatic
substituent are well precedented, and indeed we have
deliberately exploited them in synthesis.9 As might be
expected, addition of further initiator resulted in
increased conversion to tricycle 12 (up to 55% with 2
equiv. of DLP), but changes in initiator, solvent and
temperature failed to improve the yield of the desired
intermediate 11. However, use of a two-fold excess of
xanthate over alkene provided
a
slightly more
favourable 23% yield of 11, with 12 being obtained in