Titanium isopropoxide as efficient catalyst for the aza-Baylis-Hillman reaction. Selective formation of α-methylene-β-amino acid derivatives

Article abstract of DOI:10.1021/jo0163952

The direct formation of α-methylene-β-amino acid derivatives is achieved using the aza version of the Baylis-Hillman protocol. The products are readily formed in a three-component one-pot reaction between arylaldehydes, sulfonamides, and α,β-unsaturated c

Full text of DOI:10.1021/jo0163952

J . Org. Chem. 2002, 67, 2329-2334
2329
Tita n iu m Isop r op oxid e a s Efficien t Ca ta lyst for th e
Aza -Ba ylis-Hillm a n Rea ction . Selective F or m a tion of
r-Meth ylen e-â-a m in o Acid Der iva tives
Daniela Balan and Hans Adolfsson*
Department of Organic Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden
hansa@organ.su.se
Received December 20, 2001
The direct formation of R-methylene-â-amino acid derivatives is achieved using the aza version of
the Baylis-Hillman protocol. The products are readily formed in a three-component one-pot reaction
between arylaldehydes, sulfonamides, and R,â-unsaturated carbonyl compounds. The reaction is
efficiently catalyzed by titanium isopropoxide and 2-hydroxyquinuclidine in the presence of
molecular sieves. The protocol allows for structural variation of the substrates, tolerating electron-
poor and electron-rich arylaldehydes and various Michael acceptors.
Sch em e 1
In tr od u ction
Mild and selective carbon-carbon bond formations
represent one of the major challenges in organic synthe-
sis, and the frequent use of such transformations in, for
instance, the total synthesis of complex natural products
emphasizes the importance of developing useful and
general protocols for this class of reactions. From the
same perspective, atom-economic reactions become more
and more a need and a requirement.1
Sch em e 2. Th r ee-Com p on en t Aza -Ba ylis-Hillm a n
Rea ction Illu str a ted w ith Meth yl Acr yla te a s
Mich a el Accep tor
A carbon-carbon bond forming reaction which fulfills
the above criteria is the Baylis-Hillman reaction.2,3 In this
transformation, Michael acceptors are coupled with al-
dehydes to form highly functionalized R-methylene-â-
hydroxy carbonyl compounds (Scheme 1).
regular Baylis-Hillman system,2a,c these reactions tend
to be very substrate dependent and a number of different
reaction conditions have been reported.6
We have recently reported on an efficient and selective
one-pot three-component procedure for the formation of
R-methylene-â-amino acid derivatives using the aza-
Baylis-Hillman protocol (Scheme 2).7
In accordance with the general observations on the
original version of the reaction,2,3 we found that nucleo-
philic, nonsterically hindered tertiary amines showed
best activity as catalysts. Furthermore, we observed that
the addition of La(OTf)3 and molecular sieves dramati-
cally improved chemical yields and reaction rates. These
conditions turned out to be quite general, and a number
of R-methylene-â-amino acid derivatives were selectively
prepared employing this protocol.
The related R-methylene-â-amino carbonyls are typi-
cally prepared via amine substitution of the alcohol
functionality in Baylis-Hillman adducts.4 This reaction
require a two-step procedure and is often accompanied
by deleterious side-reactions, e.g. SN2′-substitution or
Michael addition, leading to the formation of regioiso-
mers.4b,c,5 An attractive alternative route toward the
formation of R-methylene-â-amino acid derivatives goes
via the aza version of the Baylis-Hillman reaction, i.e.,
employing an imine as electrophile instead of an alde-
hyde.6 This protocol produces the desired compounds in
a single reaction step, although the aldimine usually
needs to be preformed and isolated prior to the coupling
reaction. In agreement with observations made on the
The mechanism for the Baylis-Hillman reaction was
already proposed by Morita3a in the first report of the
reaction and later sustained by other studies.2b,8,9 We
have no reason to believe there are any major changes
going to the aza version of the reaction. Hence, we
* To whom correspondence should be addressed.
(1) Trost, B. M. Science 1991, 254, 1471.
(2) (a) Basavaiah, D.; Rao, P. D.; Hyma, R. S. Tetrahedron 1996,
52, 8001. (b) Drewes, S. E.; Roos, G. H. P. Tetrahedron 1988, 44, 4653.
(c) Ciganek, E. Organic Reactions, 1997, 51, 201.
(3) (a) Morita, K.; Suzuki, Z.; Hirose, H. Bull. Chem. Soc. J pn. 1968,
41, 2815. (b) Baylis, A. B.; Hillman, M. E. D. German patent 2155113,
1972 (Chem. Abstr. 1972, 77, 34174q).
(4) (a) Kim, J . N.; Lee, H. J .; Lee, K. Y.; Gong, J . H. Synlett 2002,
1, 173. (b) Drewes, S. E.; Horn, M. M.; Ramesar, N. Synth. Commun.
2000, 30, 1045. (c) Takagi, M.; Yamamoto, K. Chem. Lett. 1989, 2123.
(d) Brown, J . M.; J ames, A. P.; Prior, L. M. Tetrahedron Lett. 1987,
28, 2179.
(5) (a) Kulkarni, B. A.; Ganesan, A. J . Comb. Chem. 1999, 1, 373.
(b) Kundu, M. K.; Bhat, S. V. Synth. Commun. 1999, 29, 93. (c) Nayak,
S. K.; Thijs, L.; Zwanenburg, B. Tetrahedron Lett. 1999, 40, 981. (d)
Foucaud, A.; El Guemmout, F. Bull. Soc. Chim. Fr. 1989, 403.
(6) (a) Perlmutter, P.; Teo, C. C. Tetrahedron Lett. 1984, 25, 5951.
(b) Bertenshaw, S.; Kahn, M. Tetrahedron Lett. 1989, 30, 2731. (c)
Cyrener, J .; Burger, K. Monatsh. Chem. 1994, 125, 1279. (d) Campi,
E. M.; Holmes, A.; Perlmutter, P.; Teo, C. C. Aust. J . Chem. 1995, 48,
1535. (e) Ge´nisson, Y.; Massardier, C.; Gautier-Luneau, I.; Greene, A.
E. J . Chem. Soc., Perkin Trans. 1, 1996, 2869. (f) Richter, H.; J ung, G.
Tetrahedron Lett. 1998, 39, 2729. (g) Shi, M.; Xu, Y.-M. Chem.
Commun. 2001, 1876.
(7) Balan, D.; Adolfsson, H. J . Org. Chem 2001, 66 (19), 6498.
10.1021/jo0163952 CCC: $22.00 © 2002 American Chemical Society
Published on Web 03/13/2002

2330 J . Org. Chem., Vol. 67, No. 7, 2002
Balan and Adolfsson
Sch em e 3. P ostu la ted Mech a n ism for th e
tertiary amine catalyst on the acrylate, can attack either
the in situ formed aldimine (path A) or the aldehyde
(path B). Since the equilibrium in the initial process
(aldimine formation) is not completely in favor of the
aldimine, there will always be aldehyde present in the
reaction mixture (vide infra). The aldehyde will, thus,
compete with the aldimine in the Baylis-Hillman step
and this can lead to low selectivity of the process. The
slower the imine formation occurs, or the more active the
aldehyde is, the more alcohol adduct 2 will form in
parallel with the desired amine product 1. Therefore, one
of the important aspects of this protocol, next to the
achievement of high chemical yield, was to control the
chemoselectivity of the reaction.
Aza -Ba ylis-Hillm a n Rea ction
We have previously reported that diazabicyclo[2.2.2]-
octane (DABCO) catalyzed the formation of the aza
adduct, although with poor chemoselectivity and rather
long reaction time.7 Aggarwal et al have recently shown
that a significant rate improvement could be achieved
in the classical Baylis-Hillman reaction by the introduc-
tion of a Lewis acid catalyst together with a stoichiomet-
ric amount of DABCO.10,11 We employed this strategy in
the aza version of the reaction, performing the reaction
in THF, and observed an increase in conversion to adduct
1 by the addition of 2 mol % of La(OTf)3 and only 15 mol
% of the base. The poor chemoselectivity obtained was
attributed to slow formation of the imine. To increase the
rate of the latter reaction, and thereby push the equilib-
rium toward the imine side, molecular sieves (4 Å) were
added to trap water formed during this step. This had a
dramatic influence on both conversion and chemoselec-
tivity, resulting in almost exclusive formation of the
desired amine-adduct 1 in 79% yield after 24 h. A further
reduction of the amount of base (10 mol %) resulted in
lower chemical yield (69%), although the chemoselectivity
was unaffected and remained high. The tertiary amine
catalyst is known to coordinate to the Lewis acidic metal
center, which in terms results in a lower amount of free
DABCO available for the Baylis-Hillman step.10 Thus,
an excess of the base catalyst over the Lewis acid was
found to be a necessary prerequisite for a good turnover
in the reaction.
We then examined the influence of the solvent on the
reaction. Previous studies on the classical Baylis-Hillman
reaction have shown that a catalytic amount of alcohol,
either methanol12 or 2-propanol,6b as well as diol or triol
ligands when using Lewis acids,10 had a positive influence
on the reaction rate. This was believed to arise from a
beneficial hydrogen bond formation between the alcohol
moiety and the enolate formed by the Michael attack of
the tertiary amine on the acrylate. Stabilization via hy-
drogen bonding was suggested to result in a higher
concentration of the enolate, which directly was trans-
lated into better chemical yields. With this in mind, we
changed the solvent from THF to 2-propanol, but no
significant improvement was observed. Only small changes
in conversion were obtained upon heating the reaction
mixture to 40 °C, or when the reaction was performed in
refluxing 2-propanol. In fact, using the latter conditions
resulted in a lower yield of adduct 1. These results
Sch em e 4. Mod el System for th e Stu d y of th e
Th r ee-Com p on en t Aza -Ba ylis-Hillm a n Rea ction a
a
Path A represents the desired process. Path B illustrates the
competing side reaction.
presume the mechanism for the aza approach to be
similar to the one originally proposed, but translated for
an aldimine instead of an aldehyde (Scheme 3).
One of the major drawbacks with the aza-Baylis-
Hillman reaction, and the parent reaction, is the long
reaction times typically required to obtain synthetically
useful yields of the desired adducts. The modifications
previously reported by us somewhat improved the sys-
tem, but reaction times on the order of one to 3 days were
still necessary for high conversion to the aza adducts. In
this context, we focused our investigations toward a fast,
selective and general protocol for the formation of R-
methylene-â-amino carbonyls using a three-component
system. Herein we report on the important factors
leading to high reaction rates and good chemoselectivity
in the three-component aza-Baylis-Hillman reaction.
Resu lts a n d Discu ssion
In our previous studies on the one-pot three-component
aza-Balis-Hillman reaction we observed that reaction
rates were affected by various catalysts (Lewis acids and
bases) and additives (molecular sieves). This led us to
investigate what factors could further accelerate the
system. As a model for the optimization we chose to study
the reaction between p-toluenesulfonamide, benzalde-
hyde and methyl acrylate, in the presence of a base
catalyst (Scheme 4). As illustrated in Scheme 4, there
are two competing reaction paths in this three-component
system. The enolate, formed by Michael addition of the
(10) Aggarwal, V. K.; Mereu, A.; Tarver, G. J .; McCague, R. J . Org.
Chem. 1998, 63, 7183.
(11) Aggarwal, V. K.; Mereu, A. Chem. Commun. 1999, 2311.
(12) (a) Ameer, F.; Drewes, S. E.; Freese, S.; Kaye, P. T. Synth.
Commun. 1988, 18, 495. (b) Grob; C. A. Helv. Chim. Acta 1985, 68,
882.
(8) Hoffmann, H. M. R.; Rabe, J . Angew. Chem., Int. Ed. Engl. 1983,
22, 795.
(9) Hill, J . S.; Isaacs, N. S. Tetrahedron Lett, 1986, 27, 5007.

Formation of R-Methylene-â-amino Acid Derivatives
J . Org. Chem., Vol. 67, No. 7, 2002 2331
Ta ble 1. Lew is Acid Scr een in g for th e Aza Ver sion of
together with the high volatility of the acrylate and its
tendency to polymerize at elevated temperatures sug-
gested that further investigations were to be performed
at room temperature. Regarding the choice of nucleophilic
base, tertiary amines such as DABCO and 3-hydroxyqui-
nuclidine (3-HQD) have proven to be efficient catalysts
for the classical Baylis-Hillman reaction.13 In fact, 3-HQD
was regarded to be superior to DABCO as a catalyst due
to a favorable stabilization of the zwitterionic enolate
intermediate by formation of intramolecular hydrogen
bonds.
th e Ba ylis-Hillm a n Rea ction a
entry
Lewis acid
1b (%)
2b (%)
1
2
3
4
5
6
7
8
9
La(OTf)3
Sc(OTf)3
Yb(OTf)3
BF3‚Et2O
Cu(OTf)2
Y(OTf)3
Ti(OiPr)4
AlCl3
65
72
73
55
63
65
83
60
52
65
65
traces
traces
6
5
-
4
-
6
4
4
5
CeCl3
FeCl3
Zr(O-t-Bu)4
10
11
a
Reaction conditions: Benzaldehyde, p-toluenesulfonamide, and
methyl acrylate (1:1:1.1), 3-hydroxyquinuclidine (0.15 equiv), Lewis
acid (0.02 equiv), and molecular sieves (4 Å, 200 mg/mmol
substrate) in 2-propanol (substrate concentration 2 M) at ambient
temperature. Reaction time: 11 h. b Yields determined by 1H NMR
with benzyl alcohol as internal standard.
In our hands, the use of 3-HQD resulted in a higher
reaction rate, although the final conversion was not
affected (86% of amino-adduct 1 formed in 24 h instead
of 48 h using DABCO). A minor decrease in chemoselec-
tivity was also observed. Some of the other catalysts
previously employed in the Baylis-Hillman reaction, e.g.
DBU,11 DMAP11 or PPh3,6b gave poor results when ap-
plied in the aza-Baylis-Hillman protocol. Using cinchoni-
dine as the base catalyst in the reaction resulted 35% of
1 in 6 days.14 The poor reactivity of the alkaloid base can
be attributed to the steric bulkiness present in the vicin-
ity of the nucleophilic quinuclidine part of the molecule,
which severely hindered the attack on the acrylate.15
We then focused on determining the influence of
various Lewis acids in the aza-Baylis-Hillman reaction.16
To the model system used above, i.e., methyl acrylate,
benzaldehyde and tosylamide, were added 15 mol % of
3-HQD and 2 mol % of a Lewis acid. The reactions were
performed in 2-propanol, at ambient temperature and the
results of using a number of standard Lewis acids are
presented in Table 1.
For better comparison on the activity of different Lewis
acids, the reactions were monitored after 11 h. At this
time we anticipated the reactions not to be fully com-
pleted, and a more accurate evaluation of the results
would thus be possible.
Although small differences were observed using vari-
ous Lewis acids, Sc(OTf)3, Yb(OTf)3 and Ti(OiPr)4 proved
to be slightly more efficient in catalyzing the reaction,
both in terms of yield and selectivity. In a closer study
to gain further insight into what influences these Lewis
acids impose on the system, we monitored the reaction
profiles over time with respect to the yield of the amine-
adduct 1 (Figure 1). In addition to the above Lewis acids,
Cu(OTf)2 was included in the study since very high
chemoselectivity was obtained when using this catalyst
in the reaction (Table 1, entry 5). The four Lewis acid
catalysts were compared to La(OTf)3 and to the blank
reaction (i.e. the reaction not containing any Lewis acid).
The first observation from the plots in Figure 1 shows
upon a rather substantial rate difference between reac-
tions performed using either 3-HQD (Figure 1.a) or
DABCO (Figure 1.b) as the nucleophilic base. The reac-
tion profiles in the latter case - reactions over 24 h -
were similar to the ones obtained during the first 8-10
h using 3-HQD as the base catalyst. This is an incontest-
able evidence that 3-HQD is a much more efficient
catalyst for the reaction. Furthermore, the influence of
different Lewis acids was more evident when 3-HQD was
used. The observed yields range over more than 25% after
10 h depending on which Lewis acid was employed,
whereas in the case of DABCO, the different Lewis acids
appeared to have much less influence on the reaction.
Independently with regard to which base was employed,
Ti(OiPr)4 proved to be the superior Lewis acid, showing
high activity in catalyzing the in situ imine formation,
and the following Baylis-Hillman step. Combined with
3-HQD, Ti(OiPr)4 gave very good yield and selectivity of
the amine-adduct in significantly shorter reaction time
than what was observed using the other Lewis acids.
When using Cu(OTf)2 as Lewis acid, the shape of the
reaction profile obtained differed quite substantially
compared to what was observed using the other Lewis
acids. This effect was most evident in the reaction
performed using 3-HQD as base. In the beginning of the
reaction, a delay time was observed where the product
formation was slower than in the reaction performed
without any Lewis acid. This can find an explanation in
the high affinity of copper for nitrogen donor ligands,
which will give rise to a strong coordination of the base
catalyst to the metal-ion. The low active concentration
of base thereby obtained would effectively reduce the rate
of the Baylis-Hillman step. All the other Lewis acids used
in the study are typically oxophilic, which explains why
no such phenomena were observed in those reactions. A
further explanation can be found when considering the
rate of the in situ imine formation (Figure 2).
(13) (a) Drewes, S. E.; Freese, S. D.; Emslie, N. D.; Roos, H. P. Synth.
Commun. 1988, 18, 1565. (b) Bailey, M.; Marko´, I. E.; Ollis, W. D.;
Rasmussen, P. R. Tetrahedron Lett. 1990, 31 (31), 4509. (c) Bode, M.
L.; Kaye, P. T. Tetrahedron Lett. 1991, 32 (40), 5611.
(14) Adduct 1 was formed in 40% enantiomeric excess. For diaste-
reoselective Baylis Hillman reactions on aldimines, see; (a) Aggarwal.
V. K.; Martin Castro, A. M.; Mereu, A.; Adams, H. Tetrahedron Lett.
2002, 43, 1577. (b) Ku¨ndig, E. P.; Xu, L. H.; Schnell, B. Synlett 1994,
413. (c) Ku¨ndig, E. P.; Xu, L. H.; Romanens, P.; Bernardinelli, G.
Tetrahedron Lett. 1993, 34, 7049.
(15) Improvements in catalytic activity and stereoselectivity in the
classic Baylis-Hillman reaction were obtained using a modified cin-
chona alkaloid. For details, see; Iwabuchi, Y.; Nakatani, M.; Yokohama,
N.; Hatakeyama, S. J . Am. Chem. Soc. 1999, 121, 10219.
(16) For a recent classification of Lewis acid activity and selectivity
in enolate additions to aldehydes and aldimines, see; Kobayashi, S.;
Busujima, T.; Nagayama, S. Chem. Eur. J . 2000, 6, 3491.
The plots in Figure 2 show upon an interesting effect
regarding the three-component system. In most of the
cases, benzaldehyde was found to be in excess over the
in situ formed imine throughout the entire reaction. The
use of copper triflate and titanium isopropoxide respec-

2332 J . Org. Chem., Vol. 67, No. 7, 2002
Balan and Adolfsson
F igu r e 1. Reaction profiles for the aza-Baylis-Hillman reaction in the presence of different Lewis acids, using (a) 3-HQD or (b)
DABCO as base.
F igu r e 2. Imine-aldehyde ratio during the aza-Baylis-Hillman reaction in the presence of different Lewis acids, using (a)
3-HQD and (b) DABCO as base.
Sch em e 5
tively in combination with 3-HQD resulted, however, in
an efficient formation of the imine already from the
beginning of the reaction (Figure 2a). The excellent
chemoselectivities obtained in these cases are most
certainly reflected in the efficiency with which these
Lewis acids catalyze the imine formation. In the case of
copper triflate, the slow initial rate observed for the
Baylis-Hillman step (Figure 1a) was proposed to arise
reaction proved to be highly chemoselective even though
considerable amounts of benzaldehyde was present in the
reaction mixture next to the in situ formed imine (ratio
1:2 at low conversions, 1:1 and 2:1 during the progress
of the reaction).
From the above results, we concluded that the opti-
mized conditions for the aza-Baylis-Hillman reaction
were as follows: 15 mol % of 3-hydroxyquinuclidine and
2 mol % of titanium isopropoxide as catalysts, performing
the reaction at ambient temperature in 2-propanol in the
presence of molecular sieves. These conditions were
from a low active concentration of base. The fact that the
reaction still proceedes can then be traced to the overall
high imine concentration. It is important to notice, that
no such effects were observed using these Lewis acids in
combination with DABCO (Figure 2.b). This further
stresses the importance of choosing the proper combina-
tion of Lewis acid- and base catalysts. While the combi-
nation of copper triflate and 3-HQD most efficiently
catalyzed the imine formation, titanium isopropoxide in
the presence of the same base gave the best overall rate
of the reaction. In the latter case, the aza-Baylis-Hillman

Formation of R-Methylene-â-amino Acid Derivatives
J . Org. Chem., Vol. 67, No. 7, 2002 2333
Ta ble 2. Scop e of Aza -Ba ylis-Hillm a n Rea ction u n d er Op tim ized Con d ition sa
a
Reaction conditions: aldehyde, sulfonamide, and Michael acceptor (1:1:1.1), 3-hydroxyquinuclidine (0.15 equiv), Ti(OiPr)4 (0.02 equiv)
b
and molecular sieves (4 Å, 200 mg/mmol substrate) in 2-propanol (substrate concentration 2 M) at ambient temperature. Isolated yields.
c Chemoselectivity based on NMR yields, measured with benzyl alcohol as internal standard and expressed as the ratio between amine
d
3 and the combined two products of the reaction (amine 3 and the alcohol adduct). The imine was preformed during 2 h before the
acrylate was added in order to improve selectivity (otherwise 80%). e The imine was preformed during 12 h before the Michael acceptor
was added to improve selectivity (otherwise 50%, see text).
applied in the formation of a number of R-methylene-â-
amino acid derivatives employing the aza version of the
Baylis-Hillman protocol (Scheme 5 and Table 2). As found
for the model reaction, the optimized conditions gave in
general considerable shorter reaction times and better
chemoselectivities, as compared to our previously re-
ported results.7
â-amino acid derivatives in high yields and good chemose-
lectivities (entries 3-6). The high selectivities obtained
were very gratifying considering the fact that the side-
reaction leading to the alcohol adduct also would be
accelerated by electron-poor aldehydes. Arylaldehydes
bearing electron-releasing groups required slightly longer
reaction times to reach high conversion, although the
selectivities were found to be excellent (entries 7-8). tert-
Butyl acrylate was converted to the aza adduct in high
Arylaldehydes substituted with electron-withdrawing
groups reacted very fast giving the desired R-methylene-

2334 J . Org. Chem., Vol. 67, No. 7, 2002
Balan and Adolfsson
Sch em e 6. Sid e Rea ction Obser ved betw een
P h en yl Vin yl Su lfon e a n d p-Tolu en esu lfon a m id e
derivatives using a modified Baylis-Hillman protocol. In
the reaction, arylaldehydes, sulfonamines and Michael
acceptors were combined to form the title compounds in
high yields and with good to excellent chemoselectivities.
A systematic study, varying the reaction components led
to the optimized conditions for the aza-Baylis-Hillman
reaction. We found that the choice of Lewis acid combined
with the proper base was crucial for high yield and
selectivity. Although DABCO efficiently catalyzed the
formation of the adducts, 3-hydroxyquinuclidine, in the
presence of titanium isopropoxide as Lewis acid, was
found to be the superior base catalyst. Furthermore, the
addition of molecular sieves to the reaction mixture
dramatically improved the yield of the aza adduct, and
hence the chemoselectivity.
selectivity, but only with a modest chemical yield (entry
9). The increased reaction time and lower yield obtained
using this substrate indicates that the reaction is sensi-
tive to the steric hindrance imposed by the bulky tert-
butyl ester. Acrylonitrile, a substrate normally reacting
very fast under Baylis-Hillman conditions, gave a sur-
prisingly poor yield employing the aza-Baylis-Hillman
protocol (entry 10).
Exp er im en ta l Section
A plausible explanation can be found when considering
the role of the Lewis acid. Since esters were reacting
rather fast under these conditions, we believe that the
titanium isopropoxide catalyst which is essential for the
in situ formation of the imine, also plays an important
role in activating the Michael acceptor. The oxophilic
nature of the titanium catalyst would therefore activate
the ester more than the nitrile. Phenyl vinyl sulfone
proved to be an excellent Michael-acceptor, although not
in the fashion we had anticipated. Instead of reacting
with the basic 3-HQD producing the intermediate eno-
late, the vinyl sulfone underwent a 1,4-addition by the
rather poor nucleophile p-toluenesulfonamide. The formed
adduct reacted further with a second equivalent of the
vinyl sulfone producing compound 4 (Scheme 6). This
side-reaction could, however, be suppressed by adding the
Michael acceptor after 12 h, which allowed for the
preformation of the imine in the reaction mixture. The
active concentration of the sulfonamide was thereby
efficiently decreased and we succeeded in increasing the
chemoselectivity from 50% to 78%, although the chemical
yield remained rather poor.
Furthermore, the electron-poor sulfonamide, 4-nitroben-
zenesulfonamide, was efficiently used as amine-source
in the aza-Baylis-Hillman reaction (entry 12). The mild
cleavage protocol available for this particular sulfon-
amide suggests a straightforward approach toward non-
protected R-methylene-â-amino acid derivatives.17
As we previously observed, and in contrast to other
reports,6h,18 â-substituted Michael acceptors (e.g. cyclo-
hex-2-en-1-one) did not react to give the desired aza-
Baylis-Hillman adducts. Not even under these optimized
conditions did we observe anything but imine formation,
thus, there was neither reaction with the cyclohex-2-en-
1-one, nor alcohol adduct formation from reaction with
the aldehyde.
General experimental procedure19 for the aza version
of the Baylis-Hillman reaction, exemplified for the forma-
tion of m et h yl R-m et h ylen e-â-[(p -t olu en esu lfon yl)-
a m in o]-3-p h en ylp r op ion a te6d (1)
In a dried flask, p-toluenesulfonamide (855 mg, 5
mmol) and 3-hydroxyquinuclidine (15 mol %, 0.75 mmol,
95 mg) were measured together with molecular sieves
(4 Å, 900 mg). 2-Propanol (2.5 mL),20 benzaldehyde (505
µL, 5 mmol), methyl acrylate (495 µL, 5.5 mmol) and
Ti(OiPr)4 (2 mol %, 0.1 mmol, 30 µL) were added and the
reaction mixture was stirred for 8 h at ambient temper-
ature. The mixture was filtered through a thin layer of
Celite, which was rinsed three times with 2-propanol (10
mL). The solvent was evaporated and to the crude were
added methanol (25 mL) and aqueous sulfuric acid (10
mL, 1M).21 The solution was stirred for 1 h, then
methanol was evaporated. The remaining acidic solution
was diluted with water and extracted with dichlo-
romethane (3 × 30 mL). The organic phase was then
successively washed with NaHCO3 (sat.), NaOH (1M),
water and brine, and dried over Na2SO4. Evaporation of
the solvent gave 1.47 g (85%) of the pure product as a
1
white crystalline material: Mp 76-77 °C. H NMR (400
MHz, CDCl3) δ 2.41 (s, 3H), 3.61 (s, 3H), 5.31 (d, J ) 8.9
Hz, 1H), 5.61 (d, J ) 8.9 Hz, 1H), 5.83 (s, 1H), 6.22 (d, J
) 0.7 Hz, 1H); 7.13-7.25 (m, 7H), 7.68 (d, J ) 8.4 Hz,
2H); 13C NMR (100 MHz, CDCl3) δ 21.72, 52.20, 59.31,
126.64, 127.45, 127.98, 128.10, 128.80, 129.70, 137.84,
138.74, 138.81, 143.61, 165.98; MS (MALDI-TOF) (m/z)
384.071(MK+) 368.094 (MNa+) 346.103 (MH+).
Ack n ow led gm en t. We thank Dr. I. Pastor and P.
Va¨stila¨ for valuable comments and suggestions. The
Swedish Natural Science Research Council is gratefully
acknowledged for financial support.
Su p p or tin g In for m a tion Ava ila ble: Characterization
data for all compounds in Table 2. This material is available
free of charge via the Internet at http://pubs.acs.org.
Alifatic aldehydes did not yield any aza adducts using
the above protocol. The main reason for the low reactivity
observed for this class of aldehydes stem from the very
slow formation of the intermediate imines.
J O0163952
(19) This procedure was applied for all reactions, except in the case
of 2-pyridinecarboxaldehyde, where the Michael acceptor was added
2 h later to suppress the formation of the alcohol adduct and in the
case of phenyl vinyl sulfone which was added after 12 h to suppress
the formation of the byproduct 4.
(20) For solid aldehydes 2 mL of dichloromethane was added as
cosolvent.
Con clu sion s
Herein we have reported on an efficient, combined
Lewis acid- and base-catalyzed, three component one-pot
procedure for the formation of R-methylene-â-amino acid
(21) The acidic workup facilitates cleavage of remaining sulfo-
nylimine and efficiently removes the aldehyde. For reactions with low
conversion, an additional stirring with sulfuric acid (1M) was repeated
after the basic workup.
(17) Fukuyama, T.; J ow, C.-K.; Cheung, M. Tetrahedron Lett. 1995,
36, 6373.
(18) Rezgui, F.; El Gaied, M. M. Tetrahedron Lett. 1998, 39, 5965.