Rapid assembly of quinolizidines via consecutive nucleophilic cyclizations onto activated amides

Article abstract of DOI:10.1021/ol201616k

A new approach to the synthesis of quinolizidines involving a cascade of nucleophilic cyclizations triggered by chemoselective amide activation is reported. Particular attention was given to the effect of the nature of the tethered nucleophiles on the cascade of cyclizations. As a result, simple acyclic amides gave rapid access to functionalized quinolizidines bearing either a tertiary or quaternary center at the ring junction. Such a fused bicyclic motif is found in several alkaloids.

Full text of DOI:10.1021/ol201616k

ORGANIC
LETTERS
2011
Vol. 13, No. 16
4268–4271
Rapid Assembly of Quinolizidines via
Consecutive Nucleophilic Cyclizations
onto Activated Amides
ꢀ
Guillaume Belanger,* Gary O’Brien, and Robin Larouche-Gauthier
ꢀ
Departement de Chimie, Universite de Sherbrooke, 2500 Boulevard de l’Universite,
Sherbrooke, Quebec J1K 2R1, Canada
ꢀ
ꢀ
ꢀ
guillaume.belanger@usherbrooke.ca
Received June 16, 2011
ABSTRACT
A new approach to the synthesis of quinolizidines involving a cascade of nucleophilic cyclizations triggered by chemoselective amide activation
is reported. Particular attention was given to the effect of the nature of the tethered nucleophiles on the cascade of cyclizations. As a result, simple
acyclic amides gave rapid access to functionalized quinolizidines bearing either a tertiary or quaternary center at the ring junction. Such a fused
bicyclic motif is found in several alkaloids.
Quinolizidines are included within the framework of
numerous biologically active and structurally complex
alkaloids (Figure 1). Such a ubiquitous framework has
attracted the attention of organic synthetic chemists for
years in a quest to discover new and efficient ways that
would allow for a rapid access to functionalized quinolizi-
dines. Although several monocyclization approaches of
appropriately substituted piperidines have been reported,1
only a few examples describe the preparation of both cycles
Figure 1. Selected alkaloids bearing a quinolizidine framework
(highlighted).
in a single operation. The latter strategy offers an impor-
tant advantageof building molecular complexityina single
sequence from easily accessible acyclic starting materials.
One-pot preparation of both cycles of quinolizidines was
double NꢀC bond formation by reductive aminations,3
ditions,6 double ring-closing metathesis reactions,7 and a
condensation followed by a Mannich (or a Pictetꢀ
N-alkylations4 or hydroaminations,5 double Michael ad-
reported to be the result of either a single reaction, such as
the intramolecular iminoacetonitrile [4 þ 2] cycloaddition,2
or the result of sequential or tandem processes, such as
Spengler) reaction8 (Scheme 1).
(1) For recent reviews on the synthesis of piperidines, see: (a) Buffat,
M. Tetrahedron 2004, 60, 1701. (b) Remuson, R.; Gelas-Mialhe, Y.
Mini-Rev. Org. Chem. 2008, 5, 193. (c) Whitten, J. P.; Muench, D.; Cube,
ꢀ
R. V.; Nyce, P. L.; Baron, B. M.; McDonald, I. A. Bioorg. Med. Chem.
Lett. 1991, 1, 441. (d) Harrity, J. P. A.; Provoost, O. Org. Biomol. Chem.
2005, 3, 1349. (e) Hsung, R. P.; Kurdyumov, A. V.; Sydorenko, N. Eur.
J. Org. Chem. 2005, 23.
(2) Amos, D. T.; Renslo, A. R.; Danheiser, R. L. J. Am. Chem. Soc.
2003, 125, 4970.
(3) Airiau, E.; Spangenberg, T.; Girard, N.; Breit, B.; Mann, A. Org.
Lett. 2010, 12, 528.
(4) (a) Walther, G. Chem. Ber. 1951, 84, 304. (b) Garcıa, D.; Foubelo,
F.; Yus, M. Eur. J. Org. Chem. 2010, 2893.
(5) Quinet, C.; Sampoux, L.; Marko, I. E. Eur. J. Org. Chem. 2009,
1806.
(6) Zou, W.; Sandbhor, M.; Bhasin, M. J. Org. Chem. 2007, 72, 1226.
(7) Ma, S.; Ni, B.; Liang, Z. J. Org. Chem. 2004, 69, 6305.
(8) (a) Auruaym, E.; Spangenberg, T.; Girard, N.; Schoenfelder, A.;
Salvadori, J.; Taddei, M.; Mann, A. Chem.;Eur. J. 2008, 14, 10938. (b)
Pearson, W. H.; Hembre, E. J. J. Am. Chem. Soc. 2009, 131, 17548. (c)
Amorde, S. M.; Jewett, I.; Martin, S. F. Tetrahedron 2009, 65, 3222. (d)
Fang, H.; Wu, X.; Nie, L.; Dai, X.; Chen, J.; Cao, W.; Zhao, G. Org.
Lett. 2010, 12, 5366.
r
10.1021/ol201616k
Published on Web 07/14/2011
2011 American Chemical Society

Scheme 1. General Synthetic Approaches to Quinolizidines
Scheme 2. Representative Example of Tandem Nucleophilic
Cyclizations onto Activated Amides
Our group already reported the monocyclization of
activated amides that were trapped by various tethered
carbon nucleophiles, including enol ethers, enamines,9 and
allylsilanes.10 More recently, we demonstrated that the
iminium ions resulting of this first cyclization could be
reacted with another tethered nucleophile in a biscyclization
sequence.11,12 Herein, we describe our expanded synthetic
study toward the preparation of quinolizidines. As depicted
in Scheme 2, the activation of the amide group in cyclization
precursor 1leads to a highly reactive iminium intermediate 2
that readily reacts with a nucleophilic tether in a Vilsmeierꢀ
Haack-type cyclization13 to generate an iminium ion inter-
mediate 3 or 5, the identity of which is dependent on the
nature of X.
Iminium ion 3 can tautomerize to R,β-unsaturated
iminium ion 4, likely throughthe corresponding dienamine
intermediate. Intermediates 3ꢀ5 are then trapped in a
Mannich-type cyclization to afford quinolizidines 6ꢀ8,
respectively. It should be noted that the first iminium ion
(2) that bears a carbon of higher oxidation state is more
electrophilic than the second iminium ion of lower oxida-
tion state (3ꢀ5) in the reaction cascade. Additionally, in
such a system, the more reactive nucleophile always cy-
clizesfirst, leavingthe lessnucleophilicbranchtoreact with
the remaining iminium ion of lower electrophilicity. As a
consequence, reactivity issues were expected for the second
cyclization.
To eliminate the possibility of generating different
iminium intermediates, we first tested the tandem nu-
cleophilic cyclizations with a symmetrical substrate
(i.e., 1, X, Y = CH2) bearing identical nucleophilic
branches. When formamide 1a was treated with triflic
anhydride (Tf2O) and 2,6-di-tert-butyl-4-methylpyri-
dine (DTBMP) in dichloroethane, quinolizidines 6a
and 7a were obtained in good yield (entry 1, Table 1).
To our delight, this method even allowed for the de-
manding formation of a quaternary center at the ring
junction (entries 2 and 3). As seen in Scheme 2, the first
cyclization of activated formamide 1a generated the
iminium ion 3 (R = H, Y = CH2) for which the second
nucleophilic cyclization was in competition with con-
jugation to the R,β-unsaturated iminium ion 4 (R = H,
Y = CH2), leading to a 1:1.6 mixture of quinolizidines
6a and 7a, respectively. When acetamide 1b was acti-
vated, the more hindered iminium ion 3 (R = Me, Y =
CH2) was less reactive and the proportion of conjugated
iminium ion 4 (R = Me, Y = CH2) increased, leading to
quinolizidine 7b bearing an endocyclic alkene as the
major product.
Another symmetrical substrate (9) was treated in the
activation conditions, but unfortunately, no formation of
quinolizidine was observed, even after extended reaction
time at elevated temperature (entry 4).14 Nonetheless, the
R,β-unsaturated iminium ion intermediate 10 was cleanly
formed.15,16 The latter arose from a rapid conjugation of
ꢀ
ꢀ
(9) Belanger, G.; Dore, M.; Menard, F.; Darsigny, V. J. Org. Chem.
ꢀ
2006, 71, 7481.
ꢀ
ꢀ
(10) (a) Belanger, G.; Larouche-Gauthier, R.; Menard, F.; Nantel,
ꢀ
ꢀ
M.; Barabe, F. Org. Lett. 2005, 7, 4431. (b) Belanger, G.; Larouche-
ꢀ
ꢀ
Gauthier, R.; Menard, F.; Nantel, M.; Barabe, F. J. Org. Chem. 2006,
71, 704.
ꢀ
(11) Larouche-Gauthier, R.; Belanger, G. Org. Lett. 2008, 10, 4501.
(12) The first example of double intramolecular nucleophilic trap-
ping of activated amides was reported by our group (ref 11). The single
closest related example is the activation of a carbamate and subsequent
nucleophilic trapping. See: Magnus, P.; Gazzard, L.; Hobson, L.; Payne,
A. H.; Rainey, T. J.; Westlund, N.; Lynch, V. Tetrahedron 2002, 58,
3423.
(13) When the nucleophile is an indole or an aryle, the reaction is
rather called a BischlerꢀNapieralski cyclization: Bischler, A.; Napier-
alski, B. Chem. Ber. 1893, 26, 903.
(14) The use of DCM, DCE, and o-dichlorobenzene, or the use of
cosolvent (benzene, MeCN, or MeNO2) to modify the reaction medium
polarity affected neither the yields nor the selectivities in any of the
entries. In cases where the reaction stopped after the first cyclization
(entries 4, 5, and 9), prolonged reaction time and/or higher temperatures
resulted in protodesilylation of the remaining nucleophilic branch, along
with degradation.
(15) Observed by 1H NMR analysis of an aliquot taken from the
reaction mixture. See the Supporting Information for more details.
Org. Lett., Vol. 13, No. 16, 2011
4269

at C3 R to iminium carbon in 10 and/or (2) the steric
hindrance generated by substitution on the remaining
0
nucleophilic carbon (C5 ). The answer to this came from
Table 1. Quinolizidines via Tandem Nucleophilic Cyclizations
onto Activated Amides
treatment of substrate 11 containing mixed nucleophiles
(entry 5): upon activation, we observed15 the formation
of iminium ion intermediate 12 that readily aromatized
at 70 °C to yield the pyridinium ion 1317 with no trace of
quinolizidine. Because the iminium ion of 12 is identical to
the iminium ion intermediate obtained from the activation
of 1a, we can conclude that the steric hindrance generated
by substitution on the remaining nucleophilic carbon has
the largest effect on the second cyclization.18
Although nonsubstituted allylsilanes are more reactive
than their substituted analogues, we could reverse the
order of addition by switching to an allylstannane.19
Indeed, this subtle change between silane 11 and stannane
14 was sufficient to satisfyingly produce quinolizidines 15
and 16, in 65% yield at 0 °C (entry 6).
For the first time throughout this study, we observed the
presence of a TMS group in the fully cyclized product
(cf. 16),20 suggesting that the more nucleophilic allyl-
stannane19 branch of 14 cyclized first.21 Trapping of the
resulting iminium ion 25 generated a cationic intermediate
26 that could be either desilylated to 15 or deprotonated to
16 (Scheme 3). The use of a weak base (DTBMP) led to a
1.9:1 ratio of 15:16 (entry 6, Table 1), whereas the use of a
more basic reagent (DIPEA) resulted in the formation of
quinolizidine 16 with an endocyclic alkene as the only
detectableproduct (entry7).22 The relative stereochemistry
(C2, C3) in 15and 16is definedby addition of the allylsilane
(17) Since thorough degassing of the reaction mixture to remove
oxygen prior to addition of triflic anhydride did not prevent the
formation of pyridinium 13 the oxidation is believed to occur via
deprotonation of eniminium 12 and oxidation of the resulting 1,2-
dihydropyridine by hydride transfer to reduce iminium species in solu-
tion and to generate pyridinium 13. For an example of oxidation of 1,2-
dihydropyridine to pyridinium ion via hydride transfer, see: Cattoe, X.;
Bourissoua, D.; Bertrand, G. Tetrahedron 2006, 47, 531.
(18) Although iminium ion 12 did not undergo the Mannich cycliza-
tion, the corresponding iminium ion 28 (from activation of lactam 27)
proved to be much more reactive, thus confirming that triflyl iminium
ions (28) are more electrophilic than iminium ions of lower oxidation
state (12) in our reaction cascade.
a Highly convergent syntheses of all substrates are described in the
Supporting Information. b Conditions A: Tf2O (1.1 equiv), DTBMP (3.0
equiv). Conditions B: Tf2O (1.1 equiv), DIPEA (5 equiv). Conditions C:
(n-Bu)4NI (5 equiv). Conditions D: (n-Bu)4PBr (2.0 equiv). c Isolated
yields. d Ratio determined by relative integration of characteristic 1H
NMR signals of the crude mixture. e 63% based on recovered starting
material. f Only 1.0 equiv of DTBMP was used.
(19) We observed the following decreasing order of nucleophilicity:
allylstannane > allylsilane with non-substituted nucleophilic carbon
(cf. 1a) > allylsilane with substituted nucleophilic carbon (cf. 9) >
vinylsilane. This order is in accord with Mayr’s calculated nucleophili-
city of carbon nucleophiles. See: Mayr, H.; Kempf, B.; Ofial, A. R. Acc.
Chem. Res. 2003, 36, 66.
(20) We do not have a clear explanation to why, in this case, we
observed a competitive deprotonation to 16. We could presume that the
vinyl group induces a conformation in which the indicated proton in 26
(Scheme 3) is acidified due to a better alignment with the p orbital of the
carbocation. For another recent example of competition between desi-
lylation and deprotonation, see: Lee, M. H.; Nieto-Oberhuber, C.;
Shair, D. M. J. Am. Chem. Soc. 2008, 130, 16864.
(21) Among the two branches, the hypothetical initial cyclization of
the allylsilane branch would have generated an iminium intermediate
almost identical to the ones produced for the cyclization of 1a and 11.
Because no TMS group was observed in the products for the cyclization
of 1a and 11 and because the allylstannane is calculated to be more
nucleophilic than the allylsilane (ref 19), we could presume that the
reaction path for the cyclization of 14 is as described in Scheme 3.
the β,γ-unsaturated iminium ion generated after the first
cyclization (cf. 3f4, Scheme 2) under these conditions.
The failure to generate the desired quinolizidine from
substrate 9 contrasts with the smooth sequential cycliza-
tions of substrates 1aꢀc in entries 1ꢀ3. Two factors
could have contributed to impede the second cyclization
from 9: (1) the steric hindrance generated by substitution
(16) Although the stereochemistry of the trisubstituted double bond
of 10 was not determined for this specific substrate, it was assumed to be
cis, as determined for the analogous iminium 20.
4270
Org. Lett., Vol. 13, No. 16, 2011

onto the face of the iminium ion opposed to the vinyl group
(C3) in 25 (Scheme 3).23
unmasking of the latent nucleophile. This way, we could
circumvent reactivity issues such as those encountered
with iminium intermediates 10, 12, and 20 that bear
a second nucleophile that is not reactive enough to
undergo the Mannich cyclization (entries 4, 5, and 9,
respectively). To test this idea, we used an alkyne as a
masked nucleophile26 that cyclized upon addition of a
halide source such as tetrabutylammonium iodide (entry
10) or tetrabutylphosphonium bromide (entry 11). In the
latter entry, we deliberately chose a TBS enol ether for
the VilsmeierꢀHaack cyclization: this type of nucleo-
phile is generally not well suited for tandem nucleophilic
cyclizations onto activated amides because, after initial
addition, the resulting iminium ion intermediate is de-
activated due to electrodonation of the silyloxy group
(cf. 5, Scheme 2). Interestingly, in the present case, the
bicyclo[2.2.2]octane unit that is formed upon the Vilsmeierꢀ
Haack cyclization prevents conjugation of the silyl enol
ether with the iminium ion. The latter was then success-
fully trapped, after spontaneous cleavage of the TBS
group,15 with the pendent alkyne upon addition of a
bromide salt.11 The topology of the bicycle allowed for
the use of otherwise incompatible nucleophile combina-
tions for the tandem cyclizations. The use of an alkyne as
latent nucleophile furnished quinolizidines 2227 and 24
bearing vinylic halides that could potentially be used for
further functionalization.
In conclusion, we successfully developed a novel
cascade of cyclizations for the rapid and convergent
synthesis of substituted quinolizidines. Tertiary and
quaternary centers at the ring junction were efficiently
prepared. The reaction tolerates a broad range of teth-
ered carbon nucleophiles. Alkynes used as latent nucleo-
philes allowed for a perfect control of the sequence of
cyclizations and generated vinylic halides, which are
useful coupling partners. Further applications of this
tandem reaction to the syntheses of complex alkaloids
are currently under investigation.
Scheme 3. Deprotonation versus Desilylation Observed for
Formamide 14
The successful cyclization of 14 to 15 and 16 indicated
that the second cyclization could tolerate steric hindrance
generated by substitution at C3 on the iminium ion inter-
mediate (cf. 25, Scheme 3), as long as the remaining branch
0
is not substituted at the nucleophilic carbon C5 . This is
also true when an indole is used as the nucleophile
(substrate 17, entry 8, Table 1), considering that the indole
undergoes the initial cyclization due to its higher nucleo-
philicity compared to allylsilanes.19 Product 18 has the
framework of natural product ajmaline (Figure 1).
Activation of substrate 19 afforded the iminium inter-
mediate 20 after conjugation (entry 9, Table 1).24 We
hoped that the latter would undergo a [3.3] rearrangement
and a subsequent trapping to afford the expected
quinolizidine.25 Unfortunately, no further reaction with
iminium ion 20 was observed after the first cyclization.
Finally, in order to increase the level of control on the
order of nucleophilic cyclizations, we tested the possibi-
lity of using a latent nucleophile. Hence, less reactive
substituted allylsilanes such as in 9 could be used in the
first VilsmeierꢀHaack-type cyclization, and the second
cyclization (Mannich) might still be performed upon
Acknowledgment. This research was supported by the
Natural Science and Engineering Research Council
ꢀ ꢀ
(NSERC) of Canada, the Fonds Quebecois pour la
Recherche sur la Nature et les Technologies (FQRNT),
(22) When the reaction was run at ꢀ45 °C, the 15:16 ratio increased to
1:9, but the global yield was lower (31%); in the absence of DTBMP, the
global yield decreased to 48% and the 15:16 ratio was lowered to 1:0.5.
(23) Signals at positions C2 and C3 in 15 were identified from the
HMQC spectrum, and relative stereochemistry was determined from
coupling constants in the 1H NMR spectrum. See the Supporting
Information.
ꢀ
the Canadian Fund for Innovation, and the Universite
de Sherbrooke. Boehringer Ingelheim, NSERC, and
FQRNT doctoral fellowships to R.L.-G. are also grate-
fully acknowledged.
(24) The stereochemistry of the trisubstituted double bond in 20 was
determined to be cis, as established by NOESY experiments conducted
on piperidine 31 obtained after reduction of the iminium ion.
Supporting Information Available. Experimental proce-
dures including detailed synthesis of cyclization precursors,
spectral data, and characterization for all relevant com-
pounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
(26) (a) Overman, L. E.; Sharp, M. J. J. Am. Chem. Soc. 1988, 110,
612. (b) Overman, L. E.; Rodriguez-Campos, I. M. Synlett 1992, 995.
(27) The low yield in 22 (entry 10) could be due to the lower
electrophilicity of the conjugated iminium intermediate (compared to
the iminium intermediate in the formation of 24, entry 11). Other sources
of halogen were tested, but yields were not improved.
(25) Attempted rearrangement with iminium 20:
Org. Lett., Vol. 13, No. 16, 2011
4271