An amine-catalyzed enantioselective [3+2] cycloaddition of azomethine ylides and α,β-unsaturated aldehydes: Applications and mechanistic implications

Article abstract of DOI:10.1002/chem.201103015

The catalytic enantioselective [3+2] cycloaddition between azomethine ylides and α,β-unsaturated aldehydes catalyzed by α,α-diphenylprolinol has been studied in detail. In particular, the reaction has been extended to the use of 2-alkenylidene aminomalonates generated in situ as azomethine ylide precursors. These reactions lead to the formation of pyrrolidines containing a 5-alkenyl side chain with potential for chemical manipulation. Moreover, a detailed and concise computational study has been carried out to understand the exact nature of the mechanism of this reaction and especially the consequences derived from the incorporation of the chiral secondary amine catalyst on the reaction pathway.

Full text of DOI:10.1002/chem.201103015

FULL PAPER
DOI: 10.1002/chem.201103015
An Amine-Catalyzed Enantioselective [3+2] Cycloaddition of Azomethine
Ylides and a,b-Unsaturated Aldehydes: Applications and Mechanistic
Implications
Silvia Reboredo,[a] Efraꢀm Reyes,[a] Jose L. Vicario,*[a] Dolores Badꢀa,[a] Luisa Carrillo,[a]
Abel de Cꢁzar,[b] and Fernando P. Cossꢀo*[b]
Abstract: The catalytic enantioselective
[3+2] cycloaddition between azometh-
ACHTUNGTRENNUNGine ylides and a,b-unsaturated alde-
hydes catalyzed by a,a-diphenylproli-
nol has been studied in detail. In par-
ticular, the reaction has been extended
to the use of 2-alkenylidene aminomal-
onates generated in situ as azomethine
ylide precursors. These reactions lead
to the formation of pyrrolidines con-
taining a 5-alkenyl side chain with po-
tential for chemical manipulation.
Moreover, a detailed and concise com-
putational study has been carried out
to understand the exact nature of the
mechanism of this reaction and espe-
cially the consequences derived from
AHCTUNGTRENNUNG
the incorporation of the chiral sec
ary amine catalyst on the reaction
pathway.
ACHTUNGTRENNUNGond-
AHCTUNGTRENNUNG
Keywords: asymmetric catalysis
cycloaddition organocatalysis
pyrrolidines · reaction mechanisms
·
·
·
Introduction
In particular, catalytic enantioselective [3+2] cycloaddi-
tion that uses azomethine ylides as 1,3-dipoles comprises an
especially interesting transformation that has attracted the
attention of many different research groups worldwide be-
cause it is an extremely easy and direct method for the ste-
reoselective preparation of polysubstituted pyrrolidine com-
pounds.[2] A great deal of the relevance gained by this reac-
tion lies in the fact that the pyrrolidine skeleton constitutes
an ubiquitous structural motif, with numerous applications
seen throughout the pharmaceutical industry and among
natural products. In this context, several catalytic enantiose-
lective versions of this important reaction have been report-
ed, most of them focused on the use of chiral transition-
metal complexes as catalysts, which typically participate in
the reaction by generating the azomethine ylide through
metalation of the corresponding a-imino ester precursor.[2a]
On the other hand, asymmetric organocatalysis has also con-
tributed significantly to the field with several important con-
tributions.[3] Related to this topic, a few years ago we devel-
oped the first amine-catalyzed enantioselective [3+2] cyclo-
addition reaction by using azomethine ylides as 1,3-dipoles,
in which imines derived from diethyl aminomalonate were
treated with a,b-unsaturated aldehydes and the cheap and
commercially available a,a-diphenyl prolinol was used as
the catalyst.[4a]
1,3-Dipolar cycloaddition reactions constitute highly power-
ful methodologies for the fast assembly of five-membered
heterocyclic structures from simple and readily available
starting materials.[1] The rich reactivity profile of this trans-
formation and the availability of many structurally different
1,3-dipoles and dipolarophiles amenable to participation in
this reaction open the way for the application of this meth-
odology to the preparation of a wide number of families of
different heterocyclic architectures. Moreover, the stereo-
specificity associated with pericyclic processes makes 1,3-di-
polar cycloaddition reactions to be very appropriate candi-
dates for developing stereocontrolled variants that would
eventually lead to the formation of the final heterocyclic
compounds as single stereoisomers. In this context, enantio-
selective catalysis is a highly efficient approach to achieve
the desired stereocontrol.
[a] S. Reboredo, Prof. E. Reyes, Prof. J. L. Vicario, Prof. Dr. D. Badꢀa,
Prof. L. Carrillo
Departamento de Quꢀmica Orgꢁnica II
Facultad de Ciencia y Tecnologꢀa
Universidad del Paꢀs Vasco/Euskal Herriko Unibertsitatea
PO Box 644, 48080 Bilbao (Spain)
A commonly reported feature that was shared by all these
methodologies, which either used metal catalysis or organo-
catalysis, is that the scope of the reaction, although very
wide in regard to substitution at the dipolarophile, is rather
limited with respect to the substituents that can be placed at
the dipole. In particular, with the exception of a couple of
recent examples,[3b,5] only a-imino esters derived from non-
enolizable aldehydes can be normally used as suitable pre-
E-mail: joseluis.vicario@ehu.es
[b] Dr. A. de Cꢂzar, Prof. Dr. F. P. Cossꢀo
Departamento de Quꢀmica Orgꢁnica I
Facultad de Quꢀmica
Universidad del Paꢀs Vasco/Euskal Herriko Unibertsitatea
PO Box 1072, 20018 San Sebastiꢁn-Donostia (Spain)
E-mail: fp.cossio@ehu.es
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201103015.
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

cursors of the azomethine ylide, which restricts the substitu-
ent at this position of the dipole to aryl or alkyl groups in
which access to the a-hydrogen atom is sterically hindered
(e.g., cyclohexyl or isopropyl groups). Among these previous
reports, only a couple of examples in which a chemically
modifiable substituent that has the potential for further ma-
nipulation has been introduced at this position, thus ena-
bling a wider application of the methodology.[3b,5,6]
On the other hand, from a mechanistic point of view,
[3+2] cycloaddition reactions in general, and especially
those that involve azomethine ylides as 1,3-dipoles, have
very often been the subject of controversy, in particular re-
garding the degree of concertedness of the process or the
possibility that the reaction is more likely to proceed
through a stepwise pathway rather than being strictly pericy-
clic in some cases.[7] In this context, previous computational
studies that focused on the mechanism of [3+2] cycloaddi-
tion reactions under Lewis acid catalysis with N-unsubstitut-
ed azomethine ylides closely related to those employed in
this study demonstrated that the stepwise or concerted
nature of the reaction depends strongly on the structure and
the nature of the substituents on the two reagents involved
in the reaction.[8] However, with respect to the correspond-
ing organocatalytic versions, the exact mechanistic nature of
the reaction and the influence that the incorporation of the
catalyst can exert on the mechanistic profile of the transfor-
mation have not yet been covered.
Results and Discussion
We started by evaluating the use of the imine derived from
acrolein and diethyl aminomalonate as a suitable azo
AHCTUNGTREGiNNNU ne ylide precursor. The use of this imine would lead to the
ACHTUNGTRENNUNGmeth-
formation of a pyrrolidine cycloadduct incorporating a 5-
vinyl substituent and would have the potential to be con-
verted into a formyl moiety by means of an oxidative cleav-
age procedure, for example, through ozonolysis. Disappoint-
ingly, all our attempts to prepare this imine were unsuccess-
ful, probably due to its high instability and tendency toward
self-condensation. As an alternative, we subsequently evalu-
ated imine 1a (Scheme 2), which reacted in a satisfactory
Herein, we present a detailed study directed on one hand
toward the expansion of the scope of our previously report-
ed amine-catalyzed [3+2] cycloaddition reaction between
azomethine
ylides
and
a,b-unsaturated
aldehydes
(Scheme 1). In particular, we have studied access to pyrroli-
Scheme 2. Attempts to prepare 5-alkenyl-substituted pyrrolidines by
a [3+2] cycloaddition reaction from imines 1a and 1b.
way with crotonaldehyde (2a) in the presence of a,a-di-
ACHUTNGERNpNUG henAHCTUNGTRNENyUGN lprolinol as a catalyst under our previously reported
Scheme 1. Expanded scope of the amine-catalyzed enantioselective [3+2]
conditions, furnishing the corresponding cycloadduct 3a in
moderate yield, albeit with excellent diastereo- and enantio-
selectivity. However, when we tried to extend this reaction
to cinnamaldehyde (2b), a sluggish reaction took place, and
the desired cycloaddition product was obtained in low yield
accompanied by a significant amount of pyrrolidine 3a. This
outcome indicates that a dynamic process operates in which
imine 1a undergoes hydrolysis in the reaction medium, thus
releasing crotonaldehyde and diethyl aminomalonate and
the former participates as a more reactive dipolarophile in
a competitive [3+2] cycloaddition reaction with remaining
imine 1a. In fact, the participation of the catalyst in this side
process was also confirmed from the fact that the obtained
pyrrolidine 3a was isolated as a single endo diastereoisomer
with very high enantiomeric excess. We also tested the reac-
cycloaddition with the azomethine ylides presented herein.
dines that incorporate a 5-alkenyl substituent that can be
easily converted into highly functionalized enantioenriched
5-formyl-substituted pyrrolidines susceptible to further
modification. It is important to remark that this type of
compound cannot be directly prepared by any of the catalyt-
ic enantioselective [3+2] cycloaddition reactions previously
reported. In addition, we also present a concise mechanistic
study based on DFT calculations that explains the reaction
process and also clarifies the crucial role played by the orga-
nocatalyst.
&
2
&
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Organocatalytic Enantioselective [3+2] Cycloaddition
FULL PAPER
tion between 2a and cinnamaldehyde imine 1b, with the
latter expected to be more stable toward hydrolysis than 1a,
but a complex mixture of products was formed in which
again minor amounts of pyrrolidine 3a were detected in the
reaction mixture, whereas the desired cycloaddition product
that should arise from the participation of the azomethine
ylide derived from 1b was not observed.
With these results in hand and considering that the incor-
poration of the 5-alkylidene side chain in the final cycload-
ducts had been devised as an indirect strategy for the intro-
duction of a formyl group by means of a subsequent oxida-
tive cleavage process, we reasoned that general access to the
4,5-diformyl-substituted pyrrolidines incorporating different
substituents at the 3-position could also be possible by carry-
ing out the [3+2] cycloaddition reaction between the start-
ing a,b-unsaturated aldehyde 2 and imine 1 derived from
the same aldehyde 2. In fact, when we mixed two equiva-
lents of crotonaldehyde with one equivalent of diethyl ami-
nomalonate in the presence of 20 mol% of a,a-diphenylpro-
linol under the optimized reaction conditions, pyrrolidine 3a
was isolated in good yield and as a single endo isomer with
very high enantiomeric purity (Table 1). Under these condi-
extend the optimized reaction conditions to a series of dif-
ferent a,b-unsaturated aldehydes (Table 1) and observed
that the reaction behaved well in all cases, thus furnishing
the target 2-alkenyl-substituted pyrrolidines as highly dia-
stereoenriched endo isomers with very high ee values.
Alternatively, we also surveyed the crossed reaction be-
tween diethyl aminomalonate and two different a,b-unsatu-
rated aldehydes, but without success. In particular, when we
mixed one equivalent of 2a, another equivalent of 2b, and
diethyl aminomalonate with the catalyst under the typical
reaction conditions, a complex mixture of products was
formed in which the presence of homo-cycloaddition prod-
ucts 3a and 3b were observed by NMR spectroscopic analy-
sis of the crude reaction mixture, but without any evidence
of the presence of the desired crossed cycloadduct. Other
reactions tested that used different combinations of enals
also showed similar behavior.
It should be mentioned at this point that cycloadducts 3a–
h were somewhat unstable, and therefore these compounds
were subsequently reduced under standard conditions
(Scheme 3), which would also allow better characterization.
Thus, the corresponding alcohols 4a–h were isolated, con-
veniently characterized, and could be stored for weeks with-
out decomposition.
Table 1. Stereoselective synthesis of 5-alkenyl-substituted pyrrolidines.[a]
Entry
R
Product Yield endo/exo[c] ee
[%][b]
[%][d]
Scheme 3. Reduction of cycloadducts 3a–h.
1
3
2
4
5
6
7
8
Me
Ph
Et
nBu
n-C5H11
n-C6H13
nPr
3a
3b
3c
3d
3e
3 f
3g
66
90
59
63
59
72
56
50
>95:5
90:10
>95:5
96:4
92:8
92:8
97
90
>99
92
98
97
We subsequently faced the oxidative cleavage of the al-
kenyl moiety by ozonolysis to reach to the target 5-formyl-
substituted pyrrolidines. However, when we treated 4a with
ozone under standard conditions, a complex mixture of
products was formed, which points toward the instability of
the starting material under these conditions. For this reason,
we protected the primary alcohol and the amine moieties of
these derivatives by sequential O-silylation and N-benzoyla-
tion, thus cleanly forming pyrrolidines 6a–f (Table 2). Next,
we surveyed the ozonolysis reaction of protected derivative
6a and isolated the corresponding 5-formyl pyrrolidine in
good yield. However, this compound had a pronounced ten-
dency to undergo epimerization at the C5 stereocenter,
probably due to the configurational instability typically asso-
ciated with a-amino aldehyde derivatives. As a result, we
decided to deprotect the primary alcohol in situ after the ox-
idative cleavage process to stabilize the 5-formyl-substituted
pyrrolidine as the corresponding hemiacetal derivative.
Therefore, after ozonolysis of derivatives 6a–f under the
standard conditions, tetrabutylammonium fluoride was
added to the reaction mixture, thus leading to the isolation
of the corresponding hemiacetals 7a–f in good yield in all
the cases tested and without observing epimerization at the
94:6
92:8
93
97
(E)-C2H5CH=CH2CH2 3h
[a] All the reactions were performed on a scale of 1.00 mmol in THF
(4 mL). [b] Yield of the isolated products after purification by column
chromatography. [c] Determined by NMR spectroscopic analysis of the
crude reaction mixture. [d] Determined by HPLC after reduction to the
corresponding alcohol (see the Supporting Information for details).
tions, one equivalent of the enal had to be involved in the
formation of an imine by condensation with diethyl amino-
malonate, which participated as the azomethine ylide pre-
cursor by undergoing a subsequent cycloaddition reaction
with the remaining equivalent of the same a,b-unsaturated
aldehyde reagent, with the latter activated by the catalyst
through the formation of an iminium ion. We also tried to
carry out the reaction with diethyl aminomalonate as the
limiting reagent together with an excess (4 equiv) of a,b-un-
saturated aldehyde 2a, but the reaction proceeded more
sluggishly and the isolated yield of 3a remained essentially
the same. After this experiment, we next proceeded to
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

J. L. Vicario, F. P. Cossꢀo et al.
Table 2. Stereoselective synthesis of 5-alkenyl-substituted pyrrolidines.[a]
Finally, we also prepared tetrahydro-1H-furo
ACHTUNGTREN[NGNU 3,4-b]pyrrol-
6AHCTUNGTERG(NNUN 6aH)-one (10) by starting from cycloadduct 4b and using
a sequence analogous to that shown before (see Table 2 and
Scheme 4), but changing the benzoyl protecting group to the
related para-chlorobenzoyl group, which incorporates
a heavy chlorine atom that should allow the determination
of the absolute configuration by X-ray analysis based on the
Flack parameters. With compound 10 in hand, we grew
a
crystal suitable for X-ray analysis, which showed
a 3S,3aR,6aS absolute configuration for the three stereogen-
ic centers present in the structure (Figure 1).[9] This configu-
ration was extended to all cycloadducts 3 prepared in this
study, the stereostructure of which had been previously as-
signed on the basis of our preliminary report, in which the
absolute configuration of the products obtained in the [3+2]
cycloaddition reaction between azomethine ylides and a,b-
unsaturated aldehydes had been established by chemical
correlation.[4a]
Entry
R
Product Yield Product Yield Product Yield
[%][a]
[%][a]
[%][a]
1
3
2
4
5
6
Me
Ph
Et
nBu
n-C5H11 5e
n-C6H13 5 f
5a
5b
5c
5d
83
92
66
69
72
58
6a
6b
6c
6d
6e
6 f
87
79
87
97
96
91
7a
7b
7c
7d
7e
7 f
68
88
79
77
80
88
[a] Yield of the isolated products after purification by column chromatog-
raphy. TBPDS=tert-butyldiphenylsilyl, TBAF=tetrabutylammonium
fluoride.
C5
stereocenter
(Table 2).
These compounds were isolated
as mixtures of a and b anomers.
We also surveyed the possi-
bility of the selective chemical
modification of the functionali-
ties contained in compounds 7
(Scheme 4).
access to the tetrahydro-1H-
furo[3,4-b]pyrrol-6(6aH)-one
For
example,
A
ACHTUNGTRENNUNG
structure could be easily ach-
ieved by PCC-mediated oxida-
tion of 7a–f, thus leading to the
formation of derivatives 8a–f in
good yields in all cases. Alter-
Figure 1. X-ray structure of 10.
natively, a series of different bicyclic hexahydro-1H-furo-
ACHTUNGTRENNUNG[3,4-b]pyrroles could also be cleanly obtained from the same
type of precursor 7 by a Et3SiH/BF3·OEt2-promoted reduc-
tion of the hemiacetal moiety, which was tested on adducts
7b and 7d as representative substrates.
Mechanistic considerations: Two different mechanistic sce-
narios can be proposed for the transformation being studied
(see the outline given in Scheme 5). The reaction might pro-
ceed through 1) a concerted mechanism or 2) a stepwise
process that involves an initial Michael addition followed by
an intramolecular Mannich reaction. In fact, previous re-
ports on the [3+2] cycloaddition of the azomethine ylide
generated from diethyl aminomalonate benzaldehyde imine
with nitroalkenes under hydrogen-bonding catalysis indicate
that this reaction can proceed in a stepwise manner,[10] thus
detecting, in this case, the formation of intermediates that
arise after the initial Michael addition step and that could
be isolated and fully characterized. Other reports also pro-
pose a pericyclic mechanism that operates under similar
conditions,[3b,e] which was supported by computational
data.[3b]
In our case, we tried to gather some experimental data
that would allow us to make a mechanistic proposal by iden-
tifying or isolating one of these possible intermediates, al-
though with no success, which would eventually point
Scheme 4. Survey of chemical modifications carried out on adducts 7.
PCC=pyridinium chlorochromate.
&
4
&
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Organocatalytic Enantioselective [3+2] Cycloaddition
FULL PAPER
Figure 2. Main geometrical features of the iminium ion I and dipole II-
a obtained at the B3LYP/6-31G* level of theory.
Next, we studied the [3+2] cycloaddition reaction be-
tween azomethine ylide II-a with iminium ion I that yields
the corresponding cycloadducts IV-a (Figure 3). In all cases,
we considered that the approach of the nucleophile to the
Michael acceptor should occur from its less-hindered face,
which is consistent with the absolute configuration of the
stereogenic center generated at this position in the final cy-
cloadducts obtained and is also in agreement with previous
computational studies on other examples of conjugate addi-
tion reactions that involve the participation of iminium ions
related to I.[12,13] It should also be pointed out that the par-
ticipation of the free OH group of the a,a-diphenylprolinol
Scheme 5. Two possible mechanistic pathways for the catalytic formal
[3+2] cycloaddition of azomethine ylides to a,b-unsaturated aldehydes.
toward the operation of concerted-type mechanism in our
case.[11] However, this lack of evidence is not concluding
proof that allows us to definitively exclude the participation
of a stepwise mechanism because the inability to observe
the formation of a Michael addition intermediate can be at-
tributed to the fact that the subsequent intramolecular Man-
nich reaction could take place very quickly, thus leading to
the instantaneous formation of the final cycloadducts after
the initial Michael reaction had occurred. For this reason,
we decided to carry out a detailed computational study di-
rected to shed light on the exact nature of the mechanism of
this reaction.
According to the two possible reaction mechanisms out-
lined in Scheme 5 and also in accord with previous propos-
als,[8] the 1,3-dipole generated in the reaction media was
identified as the reactive species that participates in the cy-
cloaddition reaction because it interacts with the a,b-unsatu-
rated aldehyde, the latter being activated by the catalyst as
the corresponding iminium ion.[12] We started by performing
a conformational analysis of these two reagents (Figure 2),
and our calculations indicated that azomethine ylide II-a re-
mains in a preferred arrangement in which an internal hy-
drogen bond between the oxygen atom from one of the
ethoxycarbonyl groups and the NH moiety would form.
With respect to the preferred arrangement of iminium ion I,
this intermediate would be formed preferentially as the
more stable E isomer, thus adopting a preferred s-trans con-
formation in which the repulsion between the bulky hy-
ACHTUNGTRENNUNGdroxyACHTUNGTRENNUNGdiphenymethyl substituent and the alkenyl side chain
Figure 3. Optimized transition states for the Michael addition between
II-a and I and DGreact values for optimized structures endo-III-a and
exo-III-a.
are minimized, which is also in agreement with previously
presented calculations.[13]
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

J. L. Vicario, F. P. Cossꢀo et al.
catalyst on a possible stereodirecting effect through hydro-
gen-bonding interactions is ruled out by a previous related
study[4b] that showed that O-trimethylsilyl-protected a,a-di-
phenylprolinol can catalyze the same reaction, thus giving
rise to the corresponding cycloadducts with identical config-
urations.
We initially considered a possible orientation for the ap-
proach of II-a to I in which the benzylidene substituent of
the azomethine ylide remains close to the enamine moiety,
which should be produced during the reaction and would
also favor a subsequent intramolecular Mannich reaction in
the following step. Assuming this conformational predisposi-
tion during the approach of these reagents, we calculated
the transition states TS-1-a-endo and TS-1-a-exo that result
from the approach of the azomethine ylide II-a through its
Si or Re faces, respectively, which are named the endo and
exo approaches and would lead to the formation of Michael
addition intermediates endo-III-a and exo-III-a (DGreact =2.3
and 6.7 kcalmolꢀ1, respectively). The activation barrier asso-
ciated with the formation of exo-III-a (TS-1-a-exo) is
2.2 kcalmolꢀ1 higher than the barrier associated with the for-
mation of endo-III-a (TS-1-a-endo), which indicates that the
formation of endo-III-a is favored under kinetic control con-
ditions. We evaluated an alternative possible approach of
II-a to I (TS-1’-a-endo) in which the orientation of the azo-
Figure 4. Optimized transition states for the intramolecular Mannich re-
action step and DGreact values for optimized structures endo-IV-a and
exo-IV-a.
ing Mannich step should occur very quickly, and therefore
the initial geometry adopted in the first Michael addition
step would be maintained during the subsequent cyclization
that results in the formation of the stereocenters at C4 and
C5. This hypothesis is also consistent with the absolute con-
figuration observed for the C5 stereocenter formed in the
final pyrrolidine cycloadducts. Our calculations also indicate
that transition state TS-2-a-endo for the reaction of enamine
intermediate endo-III-a, which leads to the formation of the
endo-IV-a product by means of a Si–Si interaction between
the enamine unit and the iminium ion electrophile, respec-
tively, is 3.0 kcalmolꢀ1 more stable than the TS-2-a-exo spe-
cies, which would produce the exo diastereoisomer from the
intermediate enamine exo-III-a through a Si–Re interaction.
We also computed the final step of the catalytic cycle,
which involves the conversion of intermediate endo-IV-a
into pyrrolidine cycloadduct endo-V-a after release of the
catalyst, with concomitant generation of activated iminium
ion I (Scheme 6), to complete the energy diagram. This final
ACHTUNGTRENNUNGmethACHTUNGTRENNUNGine ylide is rotated with respect to the previously pre-
ꢀ
sented TS-1-a-endo species (dihedral angles for C=C···C N
are f=42.3 and 161.88, respectively). In this case, this tran-
sition state would also lead to the formation of endo-III-a,
but the benzylidene moiety would be placed far away from
the enamine reactive center. Nevertheless, the computed
energy for this transition state was higher than that obtained
for TS-1-a-endo, although lower than the computed energy
for TS-1-a-exo. These data confirmed that the preferential
approach of the two reagents, that is, azomethine ylide II-
a and iminium ion I, had to occur in a geometrical arrange-
ment that closely resembles that of the concerted mecha-
nism, although the calculated C4···C5 distance (i.e., 3.08 ꢄ)
in the lowest-energy transition state TS-1-a-endo indicates
that no appreciable bonding interaction occurs between
these two atoms.
We next evaluated the subsequent intramolecular Man-
nich reaction step, which should generate the pyrrolidine
ꢀ
ring by the formation of the C4 C5 bond (Figure 4). As this
bond formation had been carried out in the preceding step,
we considered that the approach of the iminium electrophile
should occur through the less-hindered Si face of the enam-
ine moiety, which is formed as the E diastereoisomer and re-
mains in the most-stable anti conformation in which the
bulky hydroxydiphenylmethyl substituent is located as far as
possible from the alkenyl moiety, as demonstrated in previ-
ously reported calculations that involved the reaction of en-
amines derived from diarylprolinol derivatives with different
electrophiles.[12] At this point and as a consequence of the
nature of the intermediate in which coulombic interactions
are present, no rotation could possibly occur through the
Scheme 6. Formation of endo-V-a from iminium intermediate endo-IV-a.
catalyst-turnover step involves a positive reaction energy
(Scheme 6), and therefore the whole process continues with
the formation of the final cycloaddition products and pro-
vides the required driving force for the reaction to proceed
to completion. This finding also indicates that a possible
equilibration between the iminium cation intermediates
ꢀ
C N bond in the enamine motif. For this reason, the follow-
&
6
&
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Organocatalytic Enantioselective [3+2] Cycloaddition
FULL PAPER
endo-IV-a and exo-IV-a through a hypothetic retro-Man-
nich/Mannich reaction sequence is very unlikely to occur.
The final energy profile of the reaction as it proceeds
through a stepwise Michael/Mannich mechanism indicates
that the first Michael addition step constitutes the rate-limit-
ing step of the process; therefore, it is at this point that the
kinetic control of this transformation takes place, with an
overall activation barrier of 11.2 kcalmolꢀ1. Once this Mi-
chael addition occurs, the subsequent intramolecular Man-
nich reaction takes place very quickly, thus leading to the
formation of the final cycloadducts after release of the cata-
lyst. This process might also explain the difficulties encoun-
tered during our attempts to isolate or detect the formation
of a Michael addition intermediate. It is important to note
that even though only the stereocenter at C2 in the first
transition states TS-1-a-endo and TS-1-a-exo (both have
identical configurations) is generated the coulombic stabiliz-
ing interactions in this intermediate keep the geometry from
being altered when going through the subsequent cycliza-
tion. Therefore, this initial Michael addition step determines
all the stereochemical issues that will define the absolute
configuration of the four stereocenters formed in the final
cycloadducts. On the other hand, it should also be noted
that there is a global kinetic preference for the formation of
the endo intermediates and transition states during all the
reaction coordinates. By taking into account that the differ-
ence in energy between transition states TS-2-a-endo and
TS-2-a-exo, which lead to the endo and exo adducts, respec-
tively, was calculated to be DDGꢁ =2.2 kcalmolꢀ1, an endo/
exo relationship of 97:3 would be expected, which is in good
agreement with the experimental results for the same reac-
tion in which a diastereomeric ratio (d.r.) of 95:5 was ob-
served. Finally, we tried to evaluate the energy profile asso-
ciated with the concerted pathway, but all our attempts to
find TS-3-a were unsuccessful.
Scheme 7. Calculated energy profile for the reaction of azomethine ylide
II-b with iminium ion I.
steroisomer over the exo adduct, with a difference in energy
between the transition states TS-1-b-endo and TS-1-b-exo of
DDGꢁ =1.8 kcalmolꢀ1, which is in good agreement with the
experimental results.
Alternatively, and for comparative purposes, we evaluated
the uncatalyzed direct background reaction between cro
ACHTUNGTRENNUNGton-
In addition, we computed the energy profile that corre-
sponds to the reaction between azomethine ylide II-b and
iminium ion I (Scheme 7) in a reaction that leads to the for-
mation of 5-alkenyl-substituted pyrrolidines of type 3 and
proceeds in comparable yields and stereoselectivities both
under the conditions employed in this study (which involve
the formation of 1a in situ) and our initially developed reac-
tion conditions[4a] (which involve the previous preparation
and isolation of 1a).
In line with the data obtained up to this moment, the cal-
culations indicate that the reaction also proceeds through
a stepwise Michael/Mannich cascade process, with the initial
conjugate addition reaction of the azomethine ylide II-b,
with iminium ion I delivering the enamine intermediate
endo-III-b in which the first stereocenter is generated, as the
rate-determining step. Next, the subsequent intramolecular
Mannich reaction takes place very quickly, thus delivering
the iminium intermediate endo-IV-b concomitant with the
formation of the other two stereocenters and finally releas-
ing the catalysts to generate the final cycloadduct endo-
V-b and complete the catalytic cycle. The calculations also
show a clear preference for the formation of the endo dia-
ACHTUNGTRENaNUGN ldehyde and azomethine ylide II-a to assess the influence
of the catalyst in the mechanistic profile of the reaction and
the ability of the catalyst to activate the a,b-unsaturated al-
dehyde through the cycloaddition process. In this case, and
contrary our previous attempts, we could locate a transition
state for the concerted pathway, thus leading to the forma-
tion of the endo cycloadduct (TS-4 in Figure 5).
It is important to note that the calculated energy for the
activation barrier associated with transition state TS-4-a,
which participates in this uncatalyzed process, was higher
than the calculated energy for the corresponding a,a-di-
ACHUTNGERNpNUG henAHCTUNGTRNENyUGN lprolinol-catalyzed reaction that proceeded via the
iminium ion I (TS-1-a-endo; compare Figures 3 and 5). This
finding is an indication of the ability of the catalyst to acti-
vate the a,b-unsaturated aldehyde toward the cycloaddition
process through the reversible formation of an iminium ion
intermediate and also shows that the participation of these
iminium ion intermediates leads to a change in the mecha-
nistic pathway of the reaction from being pericyclic in the
uncatalyzed version to stepwise in the amine-catalyzed pro-
cedure.
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

J. L. Vicario, F. P. Cossꢀo et al.
Figure 5. Optimized transition state structure TS-4-a for the reaction of
crotonaldehyde with azomethine ylide II-a. Note that the DGreact value
refers to the reaction of II-a with crotonaldehyde and not to the reaction
with the iminium ion as in Figure 3.
We analyzed the behavior of the parent reaction of cro-
ACHTUNGTRENNUNGtonACHTUNGTRENNUNGaldehyde and azomethine ylide II-a, both in presence
and absence of a catalyst, in greater detail to understand the
reasons for this mechanistic change from a concerted to
stepwise cycloaddition due to the use of a,a-diphenylproli-
nol as the catalyst. We considered NHMe2 as a simpler ana-
logue of the prolinol catalyst (see above). For the stepwise
mechanism, the first step would involve nucleophilic attack
of C3 on C4. In the case of the concerted [3+2] cycloaddi-
tion, there is a [p4s+p2s] process, and an interaction between
both C3/C4 and C1/C5 pairs is required that involves the
corresponding molecular orbitals (MOs). These MOs in turn
must have the appropriate symmetries to ensure a cyclic
electronic delocalization associated with an aromatic transi-
tion structure.
Figure 6. Potential energy surfaces (B3LYP/6-31G* level) projected on
the C1/C5 and C3/C4 distances associated with the [3+2] cycloaddition of
azomethine ylide II-a and crotonaldehyde without a catalyst (A) and
with NHMe2 as a catalyst (B). The stationary points of both reaction
paths are shown as spheres.
Initially, we performed a relaxed potential energy surface
(PES) scan of the possible reaction paths of both reactions
(Figure 6). For the reaction of crotonaldehyde and azo
ACHTUNGTNERmNUNG eth-
A
HOMO orbital of the dipolarophile does not participate in
the pericyclic process due to its s symmetry).
a species was found as a saddle point that connects the reac-
tants and products through a concerted mechanism (Fig-
ure 6A). In the case of the NHMe2-catalyzed reaction, the
potential-energy surface was qualitatively different because
the reaction does not follow the concerted mechanism ob-
tained for the uncatalyzed reaction (Figure 6B). Instead,
a stepwise mechanism occurred in which the TS-5-a species
connected the reactants with a zwiterionic intermediate,
thus leading in turn to the final products through a second
transition state TS-6-a.
We initially studied the MOs of the reactants to unveil
the origins of these different profiles (Figure 7). Because the
magnitude of the expansion coefficients in split-valence
basis sets is difficult to assess, the respective orbital energies
and expansion coefficients were recalculated using the AM1
semiempirical Hamiltonian.[14] The most relevant two-elec-
tron interactions arise from MOs f1 and f2 associated with
the dipolarophile and f1 and f2 associated with the azo-
Previously, we developed a simple model to understand
the mechanistic change in [4+3] intermolecular cycloaddi-
tion reactions of different allyl cations[15] in terms of the
second-order perturbation theory.[16] By applying this model
to the present case, the contribution to the increase in
energy on going from the reactants to the concerted [3+2]
transition state can be reduced to two terms [Eq. (1)]: One
ð4Þ
associated with the four-electron repulsion E
between
CON
the interacting carbon atoms and the other associated with
the stabilizing two-electron interactions between these cen-
ters. This latter interaction can in turn be reduced to the
contributions of electron transfer from the azomethine ylide
ð2Þ
(A) to the dipolarophile (B), which is denoted as E
, and
A!B
ð2Þ
the reverse electron transfer from B to A (E
), thus com-
B!A
pleting the cyclic electronic circulation required for a pericy-
clic mechanism:[15]
ð4Þ
ð2Þ
ð2Þ
ð1Þ
ACHTUNGTRENNUNGmethACHTUNGTRENNUNGine ylide. The main interaction patterns compatible
ECON ¼ E
þ E
þ E
CON
A!B
B!A
with a concerted, symmetry-allowed [3+2] cycloaddition are
f1!f2 and f1!f2, which correspond to the HOMO/LUMO
and HOMO-1/LUMO two-electron interactions (the
If the stepwise mechanism is considered, the first step as-
sociated with the nucleophilic attack of the azomethine
&
8
&
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Organocatalytic Enantioselective [3+2] Cycloaddition
FULL PAPER
Figure 8. Plots of the interaction energies associated with the initial
stages of the concerted ECON and stepwise mechanism ESTP of the cyclo-
addition of II-a and crotonaldehyde without a catalyst (A) and with
NHMe2 as a catalyst (B). In both diagrams, spheres represent the C1/C5
and C3/C4 critical distances of the calculated transition state structures.
sults for the early stages of the corresponding Born–Oppen-
heimer surfaces are shown in Figure 8.
In the case of the noncatalyzed reaction (X=O; Fig-
ure 8A), the concerted mechanism is the less energetic one
according to our model for low values of C1/C5 and C3/C4
distances. This behaviðo2Þr could be assigned to an important
contribution of the EB!A term (associated with the f1!f2
process), which ensures the cyclic electronic circulation re-
quired to complete the concerted mechanism. Activation of
the electrophile by the NHMe2 catalyst promotes an impor-
tant decrease in the energy of f2 (Figure 8B). Therefore, in
Figure 7. Significant MOs associated with the [3+2] cycloaddition of
II-a and crotonaldehyde computed at the RHF/AM1 level of theory with-
out a catalyst (A) and with NHMe2 as a catalyst (B).
ylide subunit on the conjugate iminium ion moiety is given
by Equation (2):
ð4Þ
ð2Þ
ESTP ¼ E þ E þ ECoul
ð2Þ
this case the f1!f2 gap is decreased relative to the noncata-
STP
Nu
AB
ð2Þ
lyzed reaction, and now E is the most important stabiliz-
Nu
ð4Þ
In this equation, E corresponds to the four-electron in-
ing contribution. In addition, the low contribution of the
STP
ð2Þ
teraction between the enolate-type part of A and the Mi-
E
term results in a disruption of the cyclic electronic cir-
B!A
ð2Þ
chael acceptor moiety of dipolarophile B, E represents the
culation; therefore, the stepwise mechanism is preferred.
Nu
stabilizing two-electron interaction between these centers
Coul
AB
and E
is the electrostatic (coulombic) interaction be-
tween A and B. We evaluated ECON and ESTP for the differ-
ent values of Rij, which correspond to the early stages of the
concerted and stepwise processes, respectively. The values of
the expansion coefficients and orbital energies are gathered
in Figure 7. The point charges were determined from the
natural-bonding analysis of each structure, including the at-
tached hydrogen atoms when necessary. The obtained re-
Conclusion
We have demonstrated that our initially developed proce-
dure to carry out the enantioselective [3+2] cycloaddition of
azomethine ylides with a,b-unsaturated aldehydes is a pro-
cess with a very wide scope that allows the preparation of
a variety of different pyrrolidine cycloadducts containing
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

J. L. Vicario, F. P. Cossꢀo et al.
L. Carrillo, E. Reyes, Adv. Synth. Catal. DOI: 10.1002/
adsc.201100432.
[5] Y. Yamashita, T. Imaizumi, S. Kobayashi, Angew. Chem. 2011, 123,
4995; Angew. Chem. Int. Ed. 2011, 50, 4893.
[6] a) Z.-Y. Xue, X. Fang, C.-J. Wang, Org. Biomol. Chem. 2011, 9,
3622; b) R. Robles-Machꢀn, I. Alonso, J. Adrio, J. C. Carretero,
Chem. Eur. J. 2010, 16, 5286.
[7] For a review, see: D. H. Ess, G. O. Jones, K. N. Houk, Adv. Synth.
Catal. 2006, 348, 2337.
[8] S. Vivanco, B. Lecea, A. Arrieta, P. Prieto, I. Morao, A. Linden, F. P.
Cossio, J. Am. Chem. Soc. 2000, 122, 6078.
[9] CCDC-853126 contains the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/
data_request/cif
several substituents and several stereocenters, with excep-
tional levels of chemical efficiency and stereoselectivity in
all cases. In particular, 2-alkenylideneaminomalonates gen-
erated in situ can be used as suitable azomethine ylide pre-
cursors that lead to the formation of a new family of pyrroli-
dine cycloadducts containing an alkenyl side chain at C5,
which has the potential for being chemically manipulated to
increase the synthetic potential of this methodology. On the
other hand, computational studies carried out on this reac-
tion demonstrate that the reaction proceeds through a step-
wise Michael/Mannich mechanism, with the initial Michael
addition as the rate-determining step of the process and also
the point at which all the stereochemical information is
placed. Moreover, our calculations have also shown that the
formation of the iminium ion leads to the activation of the
enal toward the formal cycloaddition process, which also
leads to a change in the mechanistic profile of the reaction
from being pericyclic in the uncatalyzed version to stepwise
in the reaction proceeding through iminium activation. A
simple model has also been developed to understand these
processes, which can also be extended to other similar and
related transformations.
[10] J. Xie, K. Yoshida, K. Takasu, Y. Takemoto, Tetrahedron Lett. 2008,
49, 6910.
[11] NMR spectroscopic analysis of aliquots from the crude reaction
mixture was carried out, but we could not identify any resonance
from a possible Michael addition intermediate and only observed
resonances from the starting materials and the final cycloaddition
product; when we tested the reaction with methacrolein, the starting
materials were recovered after prolonged reaction times; finally,
when we tested the reaction with the benzophenone imine derived
from diethyl aminomalonate as the azomethine ylide source, we iso-
lated cycloaddition product 11 in a low yield with the starting mate-
rials.
Acknowledgements
This study was supported by the University of the Basque Country UPV/
EHU (grants to S.R. and UFI11/22), the Spanish MICINN (CTQ2008-
00136/BQU, Ingenio-Consolider CSD2007-00006, and CTQ2010-16959),
and the Basque Government (Grupos IT-328-10 and IT-324-07). The au-
thors also thank the SGI/IZO-SGIker UPV/EHU for the generous allo-
cation of computational resources and for performing the X-ray analysis.
[12] For reviews on the mechanism and intermediates that participate in
other reactions that proceed through enamine or iminium ion for-
mation by using diarylprolinol derivatives as catalysts, see: a) A.
Mielgo, C. Palomo, Chem. Asian J. 2008, 3, 922; b) C. Palomo, A.
Mielgo, Angew. Chem. 2006, 118, 8042; Angew. Chem. Int. Ed. 2006,
45, 7876. See also: c) M. Nielsen, D. Worgull, T. Zweifel, B.
Gschwend, S. Bertelsen, K. A. Jørgensen, Chem. Commun. 2011, 47,
632.
[13] a) U. Groselj, D. Seebach, D. M. Badine, W. B. Schweizer, A. K.
Beck, I. Krossing, P. Klose, Y. Hayashi, T. Uchimaru, Helv. Chim.
Acta 2009, 92, 1225; b) S. Lakhdar, T. Tokuyasu, H. Mayr, Angew.
Chem. 2008, 120, 8851; Angew. Chem. Int. Ed. 2008, 47, 8723; c) I.
Ibrahem, P. Hammar, J. Vesely, R. Rios, L. Eriksson, A. Cordova,
Adv. Synth. Catal. 2008, 350, 1875; d) S. Cabrera, E. Reyes, J.
Aleman, A. Milelli, S. Kobbelgaard, K. A. Jorgensen, J. Am. Chem.
Soc. 2008, 130, 12031; e) E. Maerten, S. Cabrera, A. Kjaesgaard,
K. A. Jorgensen, J. Org. Chem. 2007, 72, 8893; f) P. Dinꢅr, M. Niel-
sen, M. Marigo, K. A. Jorgensen, Angew. Chem. 2007, 119, 2029;
Angew. Chem. Int. Ed. 2007, 46, 1983.
[1] For general reviews on [3+2] cycloaddition reactions, see: a) M. Kis-
sane, A. R. Maguire, Chem. Soc. Rev. 2010, 39, 845; b) B. Engels, M.
Christl, Angew. Chem. 2009, 121, 8110; Angew. Chem. Int. Ed. 2009,
48, 7968; c) U. Chiacchio, A. Padwa, G. Romeo, Curr. Org. Chem.
2009, 13, 422; d) L. M. Stanley, M. P. Sibi, Chem. Rev. 2008, 108,
2887; e) H. Pellissier, Tetrahedron 2007, 63, 3235; f) V. Nair, T. D.
Suja, Tetrahedron 2007, 63, 12247; g) S. Kanemasa, Synlett 2002,
1371; h) K. V. Gothelf, K. A. Jorgensen, Chem. Rev. 1998, 98, 863.
[2] For some very recent reviews on catalytic enantioselective [3+2] cy-
cloaddition reactions focusing on the use of azomethine ylides, see:
a) J. Adrio, J. C. Carretero, Chem. Commun. 2011, 47, 6784; b) P.
Garner, H. U. Kaniskan, Curr. Org. Synth. 2010, 7, 348; c) C. Nꢁjera,
J. M. Sansano, M. Yus, J. Braz. Chem. Soc. 2010, 21, 377.
[3] For a recent review, see: a) A. Moyano, R. Rios, Chem. Rev. 2011,
111, 4703; for some selected contributions, see: b) X.-H. Chen, Q.
Wei, S.-W. Luo, H. Xiao, L.-Z. Gong, J. Am. Chem. Soc. 2009, 131,
13819; c) M. Nakano, M. Terada, Synlett 2009, 1670; d) X.-H. Chen,
W.-Q. Zhang, L.-Z. Gong, J. Am. Chem. Soc. 2008, 130, 5652; e) Y.-
K. Liu, H. Liu, W. Du, L. Yue, Y.-C. Chen, Chem. Eur. J. 2008, 14,
9873; f) M.-X. Xue, X.-M. Zhang, L.-Z. Gong, Synlett 2008, 691.
[4] See: a) J. L. Vicario, S. Reboredo, D. Badia, L. Carrillo, Angew.
Chem. 2007, 119, 5260; Angew. Chem. Int. Ed. 2007, 46, 5168; for
a later report, see: b) I. Ibrahem, R. Rios, J. Vesely, A. Cꢂrdova, Tet-
rahedron Lett. 2007, 48, 6252. See also c) A. Iza, L. Carrillo, J. L. Vi-
cario, D. Badꢀa, E. Reyes, Org. Biomol. Chem. 2010, 8, 2238; d) N.
Fernꢁndez, L. Carrillo, J. L. Vicario, D. Badꢀa, E. Reyes, Chem.
Commun. 2011, 47, 12313; e) S. Reboredo, J. L. Vicario, D. Badia,
[14] M. J. S. Dewar, E. G. Zoebisch, E. F. Healy, J. J. P. Stewart, J. Am.
Chem. Soc. 1985, 107, 3902.
[15] I. Fernꢁndez, F. P. Cossꢀo, A. de Cꢂzar, A. Lledꢆs, J. L. MascareÇas,
Chem. Eur. J. 2010, 16, 12147.
[16] a) L. Salem, J. Am. Chem. Soc. 1968, 90, 543; b) G. Klopman, J. Am.
Chem. Soc. 1968, 90, 223.
Received: September 26, 2011
Revised: January 25, 2012
Published online: && &&, 0000
&
10
&
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Organocatalytic Enantioselective [3+2] Cycloaddition
FULL PAPER
Organocatalysis
S. Reboredo, E. Reyes, J. L. Vicario,*
D. Badꢀa, L. Carrillo, A. de Cꢁzar,
F. P. Cossꢀo* ................... &&&& — &&&&
An Amine-Catalyzed Enantioselective
[3+2] Cycloaddition of Azomethine
Ylides and a,b-Unsaturated Alde-
hydes: Applications and Mechanistic
Implications
Wide ranging potential: The organo-
catalytic enantioselective [3+2] cyclo-
chain with potential for further chemi-
N
G
cal manipulation (see scheme). More-
over, a detailed computational study
reveals the nature of the reaction
mechanism and the consequences
derived from the incorporation of the
catalyst on the reaction.
addition between azomethine ylides
and enals has been extended to include
2-alkenylideneaminomalonates as azo-
methine ylide precursors. The resulting
pyrrolidines contain a 5-alkenyl side
Chem. Eur. J. 2012, 00, 0 – 0
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
11
&
ÞÞ
These are not the final page numbers!