Highly diastereoselective synthesis of substituted pyrrolidines using a sequence of azomethine ylide cycloaddition and nucleophilic cyclization

Article abstract of DOI:10.1021/ol902767b

(Figure Presented) Although cycloadditions of azomethine ylides usually give mixtures of endolexo adducts, we successfully tuned the mechanistic path of a new reaction cascade to afford substituted pyrrolidines In high yields and diastereomeric purity. This was achieved by forcing the demetalation of tin- or silicon-substituted iminium Ions, followed by azomethine ylide cycloaddition and nucleophilic cyclization. Structural complexity is thus built rapidly In a fully controlled one-pot reaction cascade.

Full text of DOI:10.1021/ol902767b

ORGANIC
LETTERS
2010
Vol. 12, No. 7
1396-1399
Highly Diastereoselective Synthesis of
Substituted Pyrrolidines Using a
Sequence of Azomethine Ylide
Cycloaddition and Nucleophilic Cyclization
Guillaume Be´langer,* Ve´ronique Darsigny, Michae¨l Dore´, and Franc¸ois Le´vesque
De´partement de Chimie, UniVersite´ de Sherbrooke, 2500 BouleVard UniVersite´,
Sherbrooke, Que´bec J1K 2R1, Canada
guillaume.belanger@usherbrooke.ca
Received December 3, 2009
ABSTRACT
Although cycloadditions of azomethine ylides usually give mixtures of endo/exo adducts, we successfully tuned the mechanistic path of a
new reaction cascade to afford substituted pyrrolidines in high yields and diastereomeric purity. This was achieved by forcing the demetalation
of tin- or silicon-substituted iminium ions, followed by azomethine ylide cycloaddition and nucleophilic cyclization. Structural complexity is
thus built rapidly in a fully controlled one-pot reaction cascade.
1,3-Dipolar cycloaddition of azomethine ylides is recognized
as one of the best ways to prepare pyrrolidines.1,2 The reaction
has been extensively used for their generation, and its
exploitation has been documented in very nice and efficient
syntheses of alkaloids.3 Unfortunately, for substituted pyr-
rolidines, this method usually gives mixtures of endo and
exo cycloadducts, especially in intermolecular cycloaddi-
tions.4 To get around this serious limitation, many reported
cases fell back to the use of functional groups or conforma-
tional restriction to advantageously bias the diastereoselec-
tivity of the cycloaddition; these biased systems often restrain
applications to precise substitution patterns in the resulting
pyrrolidines.
(1) For a selection of reviews on the methods for the preparation of
azomethine ylides and their 1,3-dipolar cycloadditions, see: (a) Coldham,
I.; Hufton, R. Chem. ReV. 2005, 105, 2765. (b) Harwood, L. M.; Vickers,
R. J. In Synthetic Applications of 1.3-Dipolar Cyaloaddition Chemistry
Toward Heterocycles and Natural Products; Padwa, A., Pearson, W. H.,
Eds.; Wiley: New York, 2003; pp 169-252. (c) Vedejs, E.; West, F. G.
(3) For a selection of recent examples, see: (a) Pandey, G.; Gupta, N. R.;
Pimpalpalle, T. M. Org. Lett. 2009, 11, 2547. (b) Burrell, A. J. M.; Coldham,
I.; Oram, N. Org. Lett. 2009, 11, 1515. (c) Burrell, A. J. M.; Coldham, I.;
Watson, L.; Oram, N.; Pilgram, C. D.; Martin, N. G. J. Org. Chem. 2009,
74, 2290. (d) Carra, R. J.; Epperson, M. T.; Gin, D. Y. Tetrahedron 2008,
64, 3629. (e) Coldham, I.; Burrell, A. J. M.; White, L. E.; Adams, H.; Oram,
N. Angew. Chem., Int. Ed. 2007, 46, 6159. (f) Pearson, W. H.; Kropf, J. E.;
Choy, A. L.; Lee, Y.; Kampf, J. W. J. Org. Chem. 2007, 72, 4135. (g)
Epperson, M. T.; Gin, D. Y. Angew. Chem., Int. Ed. 2002, 41, 1778.
(4) For example, see ref 2f,g and: (a) Coldham, I.; Jana, S.; Watson,
L.; Pilgram, C. D. Tetrahedron Lett. 2008, 49, 5408. (b) Alker, D.; Harwood,
L. M.; Williams, C. E. Tetrahedron 1997, 53, 12671. (c) Roussi, G.; Zhang,
J. Tetrahedron 1991, 47, 5161. (d) Grigg, R.; Heaney, F. J. Chem. Soc.,
Perkin Trans. 1 1989, 198. (e) Grigg, R.; Surendrakumar, S.; Thianpatana-
gul, S.; Vipond, D. J. Chem. Soc., Chem. Commun. 1987, 47.
Chem. ReV. 1986, 86, 941
.
(2) For the gerenation of unstabilized azomethine ylides from demeta-
lation of (trimethylsilyl)methyl- or (tributylstannyl)methyliminium ions, see
ref 1 and: (a) Vedejs, E.; Martinez, G. R. J. Am. Chem. Soc. 1979, 101,
6452. (b) Achiwa, K.; Sekiya, M. Chem. Lett. 1981, 1213. (c) Achiwa, K.;
Motoyama, T.; Sekiya, M. Chem. Pharm. Bull. 1983, 31, 3939. (d) Padwa,
A.; Chen, Y. Tetrahedron Lett. 1983, 24, 3447. (e) Padwa, A.; Chen, Y.;
Dent, W.; Nimmesgern, H. J. Org. Chem. 1985, 50, 4006. (f) Pearson,
W. H.; Mi, Y. Tetrahedron Lett. 1997, 38, 5441. (g) Pearson, W. H.; Stoy,
P.; Mi, Y. J. Org. Chem. 2004, 69, 1919
.
10.1021/ol902767b  2010 American Chemical Society
Published on Web 03/01/2010

Herein, we describe how we successfully brought a
solution to the poor endo/exo selectivity associated with the
unbiased cycloadditions of azomethine ylides, even for the
challenging intermolecular version. This work is derived
from our previously developed reaction cascade that follows
path I of Scheme 1.5 Upon demetalation of silicon- or tin-
normally seen with standard azomethine ylides.1c,8 After a
1,3-dipolar cycloaddition followed by an elimination, the
resulting vinylogous urea 8 would then become activated in
situ. The ensuing nucleophilic cyclization should occur
predominantly onto the convex face of bicyclic compound
9, and upon quenching of 10, the thermodynamically
preferred cis ring junction (C3-C4) would furnish 5 as the
major diastereomer. Hence, by choosing the right metal and
demetalation conditions for azomethine ylide preparation, the
reaction path could potentially be controlled, along with the
relative stereochemistry in the resulting pyrrolidine ring, as
the following results will show. Such pyrrolidines (and
indolizidines) are ubiquitous moieties found in numerous
natural products like in the two important families of
alkaloids depicted in Figure 1.
Scheme 1. Tunable Reaction Paths for the Diastereoselective
Formation of Pyrrolidines and Indolizidines (5) from 1
Figure 1. Examples of alkaloid families containing substituted
pyrrolidines.
The planned reaction cascade was initially tested with two
model compounds, 11a and 11b,9 bearing a 3,4-dimethoxy-
benzene as the tethered nucleophile, and either a tributylstannyl
or trimethylsilyl as the metal fragment for ylide preparation,
respectively. When compound 11a was treated with POCl310
in acetonitrile11 at room temperature to activate the amide, in
the presence of N,N-diisopropyl-N-ethylamine12 and N-phenyl-
maleimide13 as the dipolarophile, only exo-indolizidine 12 was
obtained in 79% yield (Table 1, entry 1). The stereochemistry
of the latter was unequivocally established by X-ray crystal-
lographic analysis.14 When dimethyl maleate was used instead
of N-phenylmaleimide, we were able to intercept pyrroline
intermediate 14 bearing an uncyclized aromatic ring in high
substituted iminium ions,2 ylides such as 4 typically lead to
an almost equimolar mixture of cycloadducts 5 and 6, as
supported by many examples from the literature.4 Therefore,
we predicted that forcing a reversal of the 1,3-dipolar
cycloaddition and the nucleophilic cyclization steps in the
reaction cascade would be the key to diastereocontrol in the
pyrrolidine formation (path II).6
To do so, we planned to induce the demetalation of 2
subsequent to the activation of amide 1,7 leading to an ylide
7 bearing a carbon of a higher oxidation state than those
(8) For the preparation of azomethine ylides bearing a carbon of a higher
oxidation state, see also: (a) Ohno, M.; Komatsu, M.; Miyata, H.; Ohshiro,
Y. Tetrahedron Lett. 1991, 32, 5813. (b) Washizuka, K.; Minakata, S.; Ryu,
I.; Komatsu, M. Tetrahedron 1999, 55, 12969. (c) Padwa, A.; Haffmanns,
G.; Tomas, M. Tetrahedron Lett. 1983, 24, 4303. (d) Fishwick, C. W. G.;
Foster, R. J.; Carr, R. E. Tetrahedron Lett. 1996, 37, 3915. (e) Tsuge, O.;
Hatta, T.; Kakura, Y.; Tashiro, H.; Maeda, H.; Kakehi, A. Chem. Lett. 1997,
945.
(9) Synthesis described in the Supporting Information.
(10) Activation with triflic anhydride was problematic due to competitive
irreversible activation of N-phenylmaleimide.
(11) Less polar solvent, such as toluene, caused the precipitation of polar
intermediate(s), and the yield was dramatically affected.
(12) 2,6-Di-tert-butyl-4-methylpyridine could be employed as well,
resulting in similar yields. The use of other bases, or no base at all, resulted
in a significant decrease in the overall yield.
(13) Azomethine ylide intermediates should be trapped as soon as they
are formed in the reaction medium. The use of 3 equiv of dipolarophile
was optimal; below this level, extensive decomposition and low yields were
obtained.
(5) Le´vesque, F.; Be´langer, G. Org. Lett. 2008, 10, 4939.
(6) Gin et al. recently reported a somewhat related method, although
their system does not allow for an additional nucleophilic cyclization. See
ref 3d,g.
(7) Pearson et al. already proposed a fast destannylation of N-
(tributylstannyl)methyliminium ions. See ref 2g.
(14) See the Supporting Information.
Org. Lett., Vol. 12, No. 7, 2010
1397

The results in Table 1 are all consistent with a path II
mechanism when stannylated substrate 11a was employed.
Interestingly, switching to silylated substrate 11b gave com-
pletely different results, now consistent with a path I mechanism.
For example, when substrate 11b was activated with an excess
of POCl3 and treated with N-phenylmaleimide, diastereomers
12 and 13 were obtained in an equimolar amount (Table 2, entry
Table 1. Reaction Cascade with the Stannylated
Formamido-3,4-Dimethoxybenzene Derivative 11a
Table 2. Reaction Cascade with the Silylated
Formamido-3,4-dimethoxybenzene Derivative 11b
a Conditions A: substrate, dipolarophile (3 equiv), POCl3 (8 equiv), i-Pr2NEt
(1.1 equiv), MeCN, 25 °C. Conditions B: substrate, dipolarophile (3 equiv),
Tf2O (1.1 equiv), 2,6-di-tert-butyl-4-methylpyridine (1.1 equiv), n-Bu4NI (1.1
equiv), 1,2-dichloroethane, 25 °C. b Reaction run at 80 °C. c Only one
diastereomer detected by 1H NMR. d Ratio determined by GCMS analysis.
yield (entry 2).15 Isolation of a pyrroline intermediate (cf. 8)
provided evidence for a path II mechanism.
Pyrroline 14 could also be obtained using milder conditions
we developed for the activation of amide substrates in the
presence of tethered nucleophiles.16 The use of only an
equimolar amount of triflic anhydride as the activating agent
(instead of an excess of POCl3), in the presence of an external
source of halogen (n-Bu4NI) to promote desilylation,17 furnished
pyrroline 14 in about the same yield (entry 3). As anticipated
from path II, the use of dimethyl fumarate as the dipolarophile
led to the same pyrroline 14, in 91% yield (entry 4).
According to the proposed path II mechanism, pyrroline
intermediate 14 needs to be activated in the reaction mixture
to form indolizidine 15/16. Hence, upon treatment of 14 with
an excess of POCl3 in refluxing acetonitrile (entry 5), adduct
15 was obtained as the major diastereomer in 71% yield. In
the same conditions, amide 11a afforded adduct 15 directly with
no isolation of intermediate 14 (entry 6), and the observed yield
and diastereomeric ratio were better than those obtained for the
two-step sequence (entries 4 and 5 combined).
a Conditions A: 10b, dipolarophile, POCl3 (1.5 equiv), MeCN, 80 °C.
Conditions B: (ii) 10b, Tf2O (1.1 equiv), i-Pr2NEt (1.1 equiv), 1,2-
dichloroethane; (ii) n-Bu4NI (1.0 equiv), dipolarophile, 80 °C. b 8 equiv of
POCl3 were used at 25 °C. c Ratio determined by GCMS analysis. d Ratio
1
determined by H NMR analysis.
1), clearly contrasting with the enhanced diastereocontrol
observed when stannylated substrate 11a was used under the
same reaction conditions (Table 1, entry 1).
Another remarkable difference arose from the use of
dimethyl fumarate as the dipolarophile. With stannylated
substrate 11a, we isolated pyrroline intermediate 14 (Table
1, entry 4), whereas with silylated substrate 11b, we
generated the fully cyclized indolizidine 17 (Table 2, entry
2). It should also be noted that, using either triflic anhydride
and n-Bu4NI (entry 2) or POCl3 (entries 3 and 4), the relative
stereochemistry of the esters in the dipolarophiles is trans-
posed to the cycloadducts, which could only be explained
by the path I mechanism.
From all the results of the key reaction cascade detailed
above, we demonstrated that we could tune the reaction path
by selecting the right metal: stannylated derivative 11a followed
path II, whereas silylated derivative 11b followed path I. Most
importantly, to avoid the use of organotin due to its toxicity,18
we successfully forced the desired reaction (path II) even with
(15) Pyrroline 14 was isolated because of the lower nucleophilicity
of the vinylogous carbamate moiety of 14 (compared to a vinylogous
urea moiety when N-phenylmaleimide is used) that hampered its in situ
activation.
(16) (a) Be´langer, G.; Larouche-Gauthier, R.; Me´nard, F.; Nantel, M.;
Barabe´, F. Org. Lett. 2005, 7, 4431. (b) Be´langer, G.; Larouche-Gauthier,
R.; Me´nard, F.; Nantel, M.; Barabe´, F. J. Org. Chem. 2006, 71, 704.
(17) Chloride anion (n-Bu4NCl) promoted the destannylation as well,
albeit in lower yield.
1398
Org. Lett., Vol. 12, No. 7, 2010

silylated substrate 11b by adding n-Bu4NCl to rapidly cleave
the C-Si bond as soon as the formamide was activated (entry
5). Adduct 12 was obtained as the major diastereomer in an
excellent yield. Hence, higher yields and diastereoselectivities
were obtained when path II was operating.
To further expand the scope of this reaction cascade, we
also tested other tethered nucleophiles.19 When stannylated
allylsilane substrate 18a9 was activated in the presence of
N-phenylmaleimide, only cycloadduct 19 was formed (Table
3, entry 1).20 Confirmation of the reaction path II arose, as
It should be noted that only 1.5 equiv of POCl3 (with no
base) was necessary to effect amide activation, demetalation,
and activation of pyrroline intermediate, presumably because
of traces of acid present in POCl3, that promoted the latter
step. Consequently, when we activated substrate 18a with
triflic anhydride, it was possible to isolate pyrroline 21 (entry
3), and the latter was successfully converted to the desired
adduct 20 in acidic conditions (entry 4). Both steps could
even be performed in one pot (entry 5).
In the case of silylated substrate 18b,9 addition of
n-Bu4NCl was needed in order to accelerate the demetalation
(entry 6). Path I was thus entirely suppressed, leading to 19
as a single diastereomer. Again, tin could be avoided and
path II could be selected at will.
Table 3. Reaction Cascade with a Methyl Enol Ether or an
Allylsilane as the Internal Nucleophile
When the allylsilane was traded for an enol ether, the
reaction path could no longer be selected. Indeed, treatment
of stannylated or silylated substrates 22a,b9 with triflic
anhydride21 in the presence of dimethyl maleate and a source
of fluoride (n-Bu4NPh2SiF3) afforded cycloadduct 23 in
moderate yields and diastereoselectivity (entries 7 and 8).
The nucleophilic cyclization in 22a,b presumably occurred
prior to demetalation due to the higher nucleophilicity of
the enol ethers compared to the 3,4-dimethoxybenzene or
the allylsilane, and only path I operates.22
In conclusion, we have successfully developed a new
synthetic strategy that combines 1,3-dipolar cycloadditions
of azomethine ylides and cyclizations of various nucleophiles
onto iminium ions. The distinctive advantage and novelty
of this approach resides in the method we devised for the
preparation of the azomethine ylide, leading to an ylide
bearing a carbon of high oxidation state that allowed for an
additional carbon-carbon bond formation and a greater
increase in structural complexity from a single transforma-
tion. We also showed that indolizidine adducts could be
obtained in high yields and diastereomeric purity by forcing
the desired reaction path II. Synthetic applications of this
new reaction cascade will be published in due course.
Acknowledgment. This research was supported by the
Natural Sciences and Engineering Research Council of
Canada (NSERC) and the Canadian Fund for Innovation. A
Universite´ de Sherbrooke M.Sc. scholarship to M.D. and a
Ph.D. scholarship to F.L. as well as a NSERC PGS-M
scholarship to V.D. are also gratefully acknowledged.
a Conditions A: substrate, dipolarophile (10 equiv), POCl3 (1.5 equiv),
MeCN, 25 °C. Conditions B: (i) substrate, Tf2O (1.1 equiv), i-Pr2NEt
(1.1-1.2 equiv), 1,2-dichloroethane; (ii) n-Bu4NI or n-Bu4NPh2SiF3,
dipolarophile (10 equiv), 25 °C. Conditions C: substrate, TFA (1.5 equiv),
MeCN, 25 °C. b Reaction run at 80 °C. c Only one diastereomer detected
by 1H NMR. d Only one diastereomer detected by GCMS. e Ratio determined
by GCMS analysis.
Supporting Information Available: Experimental details
and spectra for all new compounds and X-ray diffractions
structures (CIF). This material is available free of charge
via the Internet at http://pubs.acs.org.
in the 3,4-dimethoxybenzene series, from the treatment of
18a with dimethyl fumarate leading to adduct 20 bearing a
cis diester (entry 2).
OL902767B
(20) Relative stereochemistry on 19 was proven by comparison with
the X-ray diffraction stucture of the analogous p-bromo-substituted phenyl
derivative (see the Supporting Information).
(21) The enol ether was not compatible with POCl3 activation, resulting
in partial hydrolysis.
(18) Thomas, L. D.; Shah, E.; Bankhurst, A. D.; Whalen, M. M. Arch.
Toxicol. 2005, 79, 711.
(19) Formation of 5-membered rings by via Vilsmeier-Haack cycliza-
tions gives poor yields, as previously demonstrated (see ref 16).
(22) Mayr, H.; Kempf, B.; Ofial, A. R. Acc. Chem. Res. 2003, 36, 66.
Org. Lett., Vol. 12, No. 7, 2010
1399