could generate an acyclic product such as dipeptide 8
(Scheme 2), which would be similar to 6 and poised to
undergo phenolic oxidative coupling, thus mimiking the
proposed alkaloid biogenesis process.
In order to test our hypothesis, we initially selected p-
hydroxybenzaldehyde (9a), benzylamine (10a), fumaric acid
monoethyl ester (11a), and tert-butyl isocyanide (12a) as the
coupling reagents, and subjected the mixture to microwave
(MW) irradiation[18] in methanol at 300 W, 2008C, 18 bar for
20 min. The reaction gave rise to 5,5,6-fused azaspiro tricycle
14a in approximately 10% yield (Scheme 3), along with
undesired side products. As we initially expected to observe a
2-azaspiro[4.5]deca-6,9-diene-3,8-dione (such as 7), we were
isocyanide group, such as adamantyl (12c; Table 1, entry 4),
resulted in the single isomer 14e. Replacement of fumaric
acid monoethyl ester (11a) with trifunctional N-methylma-
leamic acid (11d; Table 1, entry 11) resulted in the corre-
sponding 5,6,6-fused azaspiro tricycle 15a (type B) in good
yield. Remarkably, use of 4-hydroxy-1-napthaldehyde (9h)
resulted in 5,5,6,6-fused azaspiro tetracycles 16a and 16b
(type C) in greater than 98:1 diastereoselectivity (Table 1,
entry 12 (product confirmed by X-ray crystal structure) and
entry 13). Importantly, the reaction did not occur when other
heating sources were used, although our attempts were
limited in scope. Remarkably, however, the reaction afforded
a quantitative yield of spirocycle 18 when run under stage 1
conditions. This result could be attributed to the fact that the
5-exo-trig Michael addition retains complete aromaticity of an
intact benzene ring.[20] In all successful instances, the com-
plexity of the products that result from this cascade UMAM
reaction illustrates the remarkable chemo-, regio-, and
stereoselectivity achieved by using simple and readily avail-
able materials without protecting-group manipulation.
We rationalized that the acyclic Ugi product could exist
either in trans-amide (13’) or cis-amide (13’’) conformations at
room temperature. In the trans-amide conformation, an
electron-donating p-hydroxy group on 9 (R1) would lead to
the formation of 2-azaspiro[4.5]deca-6,9-diene-3,8-dione (18)
through a 5-exo-trig Michael addition of 17 (Scheme 4) under
the influence of microwave irradiation. We noted that the
formation of 18 was controlled by the substituents at R4.[17]
Bulky groups favor the formation of 18 because of steric
effects, and disfavor the formation of DKP 21 as the
competing pathway. Since bulky R4 substituents hinder the
progression of a 6-exo-trig aza-Michael addition, we propose
that the unprecedented 5-exo-trig aza-Michael addition on
intermediate 18 occurs as a result of the proximity effect,[23]
and ultimately gives rise to (ꢁ)-type A and (ꢁ)-type C
products. Both zwitterionic intermediates 17 and 19 are
believed to be stabilized by hydrogen bonding and help to
absorb microwave energy efficiently.[24] To the best of our
knowledge, this is a rare example in which the same
substituent directs a specific reaction outcome and then
simultaneously participates in a bond-forming reaction solely
because of steric effects. On the other hand, the less bulky
group at R2 (Table 1, entry 11) led to the formation of (ꢁ)-
type B through a 6-exo-trig aza-Michael pathway.[25] Impor-
tantly, when X = OMe, the 5-exo-trig aza-Michael addition
occurred regioselectively on the carbon atom bearing Z of the
more reactive Michael acceptor. Structural characterization
and relative stereochemistry of the fused azaspiro tricycles
Scheme 3. Formation of 5,5,6-fused azaspiro tricycle 14a.
enlightened to fully characterize 14a. Recognizing that water
would accelerate the overall process,[19] we screened the
reaction in methanol/water (10–100% water) by using various
temperature, pressure, and time[20] combinations. An
improved yield of 14a (ca. 70%) was obtained when the
microwave was equilibrated to 300 W, 1908C, 19 bar for
30 min. However, in order to avoid any unwanted Passerini
side product[21] or N-formylamide formation,[22] we elected to
carry out the reaction in water alone by utilizing a two-stage
protocol (stage 1: 708C, 10 bar, 1 h; stage 2: 300 W, 1908C,
19 bar, 30 min), which further improved the yield to 85% and
avoided the formation of unwanted by-products.
In an attempt to pinpoint atom connectivity in the cascade
process and deduce mechanistic insights, the reaction was first
repeated with preformed acyclic Ugi product 13a,[20] which
ultimately provided 14a in similar yield. The result implied
that an initial Ugi reaction had occurred as the first reaction in
the Ugi/Michael/aza-Michael (UMAM) sequence of reaction
transformations.[17]
1
and azaspiro tetracycles were established by NOE, H–1H
We next examined the scope of the reaction by using
various substrates. As noted in Table 1, we observed that
electron-donating substituents on the trifunctional p-hydrox-
ybenzaldehyde (e.g., OMe; Table 1, entries 1, 2, 4–6, 8–10)
favor the formation of fused azaspiro tricycles (14, type A),
while electron withdrawing NO2 and Br substituents lead to
severe decomposition.
In contrast, 3-fluoro-p-hydroxybenzaldehyde (9 f) pro-
vided the desired fused azaspiro tricycle 14h in appreciable
yield (Table 1, entry 7). Use of very bulky substituents on the
gDQFCOSY, gHMQC, and gHMBC experiments, and
unequivocally confirmed by X-ray crystal structure analy-
sis.[20]
Although Baldwinꢀs rules favor both 6-exo-trig and 7-exo-
trig ring cyclization events,[23] it is noteworthy that 2,5-DKPs
(21), 5,6,7-fused azaspiro tricycle 23, or a 5,6,6,7-fused
azaspiro tetracycle 24 were not obtained when R4 was a
bulky tert-butyl group (Schemes 4 and 5). However, a 6-exo-
trig aza-Michael pathway was favored when substrates 9g,
10c, 11d, and 12a were used (Table 1, entry 11).[25] Our
Angew. Chem. Int. Ed. 2011, 50, 9418 –9422
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim