Catalytic enantioselective reissert-type reaction: Development and application to the synthesis of a potent NMDA receptor antagonist (-)-L-689,560 using a solid-supported catalyst

Article abstract of DOI:10.1021/ja010654n

Full details of the first catalytic enantioselective Reissert-type reaction are described. Utilizing the Lewis acid-Lewis base bifunctional catalyst 5 or 6 (9 mol %), the Reissert products were obtained in 57 to 99% yield with 54 to 96% ee. Electron-rich

Full text of DOI:10.1021/ja010654n

J. Am. Chem. Soc. 2001, 123, 6801-6808
6801
Catalytic Enantioselective Reissert-Type Reaction: Development and
Application to the Synthesis of a Potent NMDA Receptor Antagonist
(-)-L-689,560 Using a Solid-Supported Catalyst
Masahiro Takamura, Ken Funabashi, Motomu Kanai, and Masakatsu Shibasaki*
Contribution from the Graduate School of Pharmaceutical Sciences, The UniVersity of Tokyo,
Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
ReceiVed March 12, 2001
Abstract: Full details of the first catalytic enantioselective Reissert-type reaction are described. Utilizing the
Lewis acid-Lewis base bifunctional catalyst 5 or 6 (9 mol %), the Reissert products were obtained in 57 to
99% yield with 54 to 96% ee. Electron-rich quinolines produced better yields and enantioselectivities than
electron-deficient substrates. Kinetic studies indicated that the reaction should proceed via the rate-determining
acyl quinolinium formation, followed by the attack of a cyanide. The catalyst does not facilitate the first
rate-determining step; however, it strongly facilitates the second cyanation step. The reaction was successfully
applied to an efficient catalytic asymmetric synthesis of a potent NMDA receptor antagonist (-)-L-689,560.
A key step is the one-pot process using the Reissert-type reaction from quinoline 1f, followed by stereoselective
reduction of the resulting enamine 2f. This step gave the key intermediate 20 in 91% yield with 93% ee, using
1 mol % of 6. The enantiomerically pure target compound was obtained through 10 operations (including
recrystallization) in total yield of 47%. Furthermore, 6 was immobilized to JandaJEL, and the resulting solid-
supported catalyst 11 afforded 20 in a comparable yield to the homogeneous 6, but with slightly lower
enantioselectivity.
Introduction
reasons. First, strong electrophiles such as an acid halide or
TMSX (generated during the reaction) could decompose the
catalyst by acylating and/or silylating the ligand. Second, the
conformation (the s-trans/s-cis isomers of the amide bond, see
17 and 16, Figure 2) of the reactive acyl quinolinium intermedi-
ate is rather dynamic. The two conformers produce opposing
enantiomers even if a nucleophile attacks the reactive intermedi-
ate from a defined side. These two conformers should be strictly
differentiated by the catalyst to realize high enantioselectivity.
We reported that the bifunctional catalyst 5 promotes the
cyanosilylation of aldehydes and imines with high enantiose-
lectivity and substrate generality.7 The origin of the catalysis
stems from the dual activation of the substrate and TMSCN at
defined positions by the Lewis acid (aluminum metal) and the
Lewis base (oxygen atom of the phosphine oxide) of the catalyst,
respectively. Bifunctional catalysts should be advantageous for
developing a catalytic asymmetric Reissert-type reaction of
TMSCN. The catalyst was stable even in the presence of an
electrophile (the acidic proton of alcohols and phenol).8 In
addition, the Lewis acid and the Lewis base of the catalyst could
preorganize the acyl quinolinium and TMSCN. Thus, the
reaction should proceed from a specific amide conformer. Herein
we report a full account of the development of the first catalytic
In 1905, Reissert reported the addition of KCN to quinoline
in the presence of benzoyl chloride (Reissert reaction).1 Since
then, there have been many modifications and improvements
of the Reissert reaction, such as employing other nucleophiles
(Grignard reagents, allyltin reagents, enol trimethylsilyl ethers,
and trimethylsilyl cyanide) and catalytic promotion by a Lewis
acid.2 To date, the Reissert-type reaction has been extremely
useful for synthesis of a variety of pharmaceuticals and
biologically active alkaloids.3,4 The reaction is considered to
proceed via two steps: the generation of an acyl quinolinium
intermediate by a nucleophilic attack of quinoline to an acid
chloride, followed by the addition of cyanide.5
A common approach for the enantiocontrol of the Reissert-
type reaction is to use a stoichiometric amount of chiral acylating
reagents.6 Promotion and control of the reaction by a chiral
catalyst should provide a synthetically significant advantage.
Developing a catalytic asymmetric Reissert-type reaction,
however, is a great challenge, due mainly to the following
(1) Reissert, A. Chem. Ber. 1905, 38, 1603.
(2) Ruchirawat, S.; Phadungkul, N.; Chuankamnerdkarn, M.; Thebtara-
nonth, C. Heterocycles 1977, 6, 43-46.
(3) (a) Popp, F. D. Heterocycles 1973, 1, 165-180. (b) McEwen, W.
E.; Cobb, R. L. Chem. ReV. 1955, 55, 511-549.
(4) For selected recent examples, see: (a) Comins, D. L.; Huang, S.;
McArdle, C. L.; Ingalls, C. L. Org. Lett. 2001, 3, 467-471. (b) Shair, D.
M.; Yoon, T.; Danishefsky, J. S. J. Am. Chem. Soc. 1996, 118, 9509-
9525. (c) Itoh, T.; Nagata, K.; Miyazaki, M.; Osawa, A. Synlett 1999, 7,
1154-1156. (d) Tyrell, J. A., III; McEwen, W. E. J. Org. Chem. 1981, 46,
2476-2479.
(5) (a) Duarte, F. F.; Popp, F. D. J. Heterocycl. Chem. 1991, 28, 1801-
1804. (b) Akiba, K.; Negishi, Y.; Inamoto, N. Synthesis 1979, 55-57. (c)
Abushanab, E.; Lee, D.-Y. J. Org. Chem. 1975, 40, 3376-3378.
(6) See refs 4a and 4c for representative examples.
(7) (a) Hamashima, Y.; Sawada, D.; Kanai, M.; Shibasaki, M. J. Am.
Chem. Soc. 1999, 121, 2641-2642. (b) Hamashima Y.; Sawada D.; Nogami
H.; Kanai M.; Shibasaki M. Tetrahedron 2001, 57, 805-814. (c) Takamura,
M.; Hamashima, Y.; Usuda, H.; Kanai, M.; Shibasaki, M. Angew. Chem.,
Int. Ed. 2000, 39, 1650-1652. (d) Takamura, M.; Hamashima, Y.; Usuda,
H.; Kanai, M.; Shibasaki, M. Chem. Pharm. Bull. 2000, 48, 1586-1592.
(8) The relative pKa difference between the naphthol of the chiral ligand
and the additive phenol or alcohol should be a reason for the stability of
the catalyst. In addition, one of the two phosphine oxides of the ligand
might perform a dative coordination to the aluminum metal, which should
further stabilize the complex.
10.1021/ja010654n CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/19/2001

6802 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Takamura et al.
enantiomeric excesses than the aliphatic or substituted aromatic
acid chlorides. 2-Furoyl chloride, which is more electron-rich
and therefore less reactive than benzoyl chloride, gave slightly
better enantioselectivity. No acylation or silylation of the chiral
ligand was observed by TLC analysis. Thus, the catalyst seemed
to be compatible with the strong electrophiles, although partial
inactivation of the catalyst by TMSCl might compromise yield
and enantiomeric excess of the product (vide infra).11 Next, the
solvent effect was investigated by using the more reactive
benzoyl chloride as an acylating reagent. The ee increased from
71% to 78% when the solvent polarity was decreased, such as
with toluene or benzene (entry 7 or 8), although the yield was
much lower. In contrast, the ee was decreased to 37% when
the more polar acetonitrile was used as solvent (entry 9). These
results suggested the presence of a spontaneous reaction pathway
independent of the catalyst that gives the racemic product. The
acyl quinolinium intermediate derived from 2-furoyl chloride
or an electron-rich aromatic acid chloride should be less reactive
than the one derived from an aliphatic or reactive aromatic acid
chloride. Therefore, the reaction would be mediated by the
catalyst, using 2-furoyl chloride or an electron-rich aromatic
acid chloride.12 Using a reactive acid chloride, however, the
acyl quinolinium intermediate should have a higher reactivity,
and the reaction would proceed to some extent via the racemic
pathway independent of the catalyst. The Lewis acid activation
by the catalyst might be slightly lower in a polar solvent and
therefore the racemic reaction should increase.13 As expected,
the spontaneous Reissert-type reaction of quinoline under these
conditions proceeded in the absence of catalyst, giving the
corresponding products in low yield. The corresponding Reissert
compound was obtained in 5% yield in CH2Cl2 at -40 °C for
24 h when the more reactive acetyl chloride was used as an
acylating reagent (reaction conditions corresponding to Table
1, entry 4). When 2-furoyl chloride was used as an acylating
reagent in CH2Cl2-toluene (1:1) mixed solvent in the absence
of the catalyst (optimized reaction conditions corresponding to
Table 3, entry 1), however, only 1% yield of the corresponding
Reissert compound was obtained (-40 °C for 64 h). Therefore,
using the optimized reaction conditions (an electron-rich
aromatic acid chloride and a less polar solvent), the spontaneous
racemic reaction pathway could be almost completely sup-
pressed. Thus, we have been able to significantly differentiate
the reaction rate mediated by the catalyst from that of the
spontaneous racemic reaction.
Figure 1. General scheme of the catalytic enantioselective Reissert-
type reaction.
Table 1. Optimization of Catalytic Asymmetric Reissert-Type
Reaction of 1a with Various Acylating Reagents Using 9 mol % of
5a
entry R (RCOCl)
solvent
time/h yield/%b ee/%c
1
2
3
4
5
6
7
8
9
Ph
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
24
48
24
24
24
24
24
24
24
70
58
68
42
52
63
27
25
67
71
73
64
58
54
46
78
78
37
2-furyl
o-MeOPh
CH3
PhCHdCH CH2Cl2
1-naphthyl CH2Cl2
Ph
Ph
Ph
CH2Cl2/toulene (1:1)
CH2Cl2/benzene (1:1)
CH3CN
a The reaction was conducted at -40 °C using 2 equiv of TMSCN
and 1.1 equiv of RCOCl. b Isolated yield. c Determined by chiral HPLC
analysis.
enantioselective Reissert-type reaction (Figure 1).9 The reaction
was further extended to an efficient catalytic asymmetric
synthesis of a potent N-methyl-D-aspartate (NMDA) receptor
antagonist, L-689,560. Moreover, a new solid-supported bi-
functional catalyst was developed, and the easy separation of
the catalyst facilitated the synthesis. The targeted compound is
a promising drug candidate for Alzheimer’s disease and for
reducing ischemic brain damage.10
To improve the reaction by developing a more reactive
catalyst, and avoid the contribution of the racemic pathway as
much as possible, we designed a new catalyst that would
facilitate the attack of TMSCN on the acyl quinolinium ion by
favoring the catalyst conformer in which the Lewis acid and
the Lewis base moieties are optimally positioned.14 Molecular
Results and Discussion
A. Optimization, Scope, and Limitations of Catalytic
Enantioselective Reissert-Type Reaction. First, the effect of
different acid chlorides (RCOCl) on the reaction was investi-
gated using 5 (9 mol %) as a catalyst and quinoline (1a) as a
substrate at -40 °C. As shown in Table 1 (entries 1-6), benzoyl
chloride (entry 1) and 2-furoyl chloride (entry 2) afforded higher
(11) Longer reaction time did not improve the chemical yield. As
discussed in the Mechanistic Studies section, the inactivation of the catalyst
was not caused by a silylation of the naphthol oxygen by TMSCl.
(12) Consistently, 4-methoxybenzoyl chloride gave the product with high
ee (71%) under similar conditions of Table 1, entry 1, however, in only
28% yield. The low yield was attributed to the stability of the product,
which was prone to retro-Reissert-type reaction.
(13) In a preliminary communication, we attributed the lower enanti-
oselectivity using the reactive acid chloride and polar solvent to the
generation of a larger amount of the acyl quinolinium intermediate. Because
the acyl quinolinium intermediate was below the detection limit of 1H NMR
under any conditions of Table 1, however, the alternative explanation
described in this full account seems to be more convincing.
(9) For the preliminary results, see: Takamura, M.; Funabashi, K.; Kanai,
M.; Shibasaki, M. J. Am. Chem. Soc. 2000, 122, 6327-6328.
(10) (a) Leeson, P. D.; Carling, R. W.; Moore, K. W.; Moseley, A. M.;
Smith, J. D.; Stevenson, G.; Chan, T.; Baker, R.; Foster, A. C.; Grimwood,
S.; Kemp, J. A.; Marshall, G. R.; Hoogsteen, K. J. Med. Chem. 1992, 35,
1954-1968. (b) Carling, R. W.; Leeson, P. D.; Moseley, A. M.; Smith, J.
D.; Saywell, K.; Tricklebank, M. D.; Kemp, J. A.; Marshall, G. R.; Foster,
A. C.; Grimwood, S. Bioorg. Med. Chem. Lett. 1993, 3, 65-70.
(14) For the strategy to restrict the conformation of the ligand for the
dual activation pathway, see: Kanai, M.; Hamashima, Y.; Shibasaki, M.
Tetrahedron Lett. 2000, 41, 2405-2409.

Catalytic EnantioselectiVe Reissert-Type Reaction
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6803
Table 2. Optimization of Catalyst Structure on the Reaction of 1aa
methylation, gave methyl tetrahydroquinoline-2-carboxylate (13)
without any epimerization. The absolute configuration of 13
was determined from the optical rotation.15,16 Optically active
13 has been used as a key intermediate for the synthesis of a
potent NMDA receptor antagonist.17,18 It was synthesized,
however, via resolution of a racemic mixture.
R
time/ yield/ ee/
b
c
entry catalyst (RCOCl)
solvent
CH2Cl2
h
%
%
S/R
1
2
3
4
5e
6
7
8
9
5
5
6
6
6
7
8
9
10
Ph
Ph
Ph
Ph
24
70
27
82
49
91
44
33
61
78
71
78
75
83
85
50
2
R
R
R
R
R
R
S
CH2Cl2/toluened 48
CH2Cl2
24
CH2Cl2/toluened 24
2-furyl CH2Cl2/toluened 64
When this reaction was applied to the isomeric substrate,
isoquinoline, however, the product was obtained in 75% yield
with only 3% ee at -40 °C for 7.5 h, using benzoyl chloride
and 5 (9 mol %). We attributed this discrepancy to the difficulty
in controlling the amide conformer. From molecular modeling
studies, there seems to be no significant preference between
the two transition states involving either the s-cis or s-trans acyl
isoquinolinium intermediate, at least sterically. Consequently,
the reaction could proceed via both of these conformers. To
overcome this, 3-methylisoquinoline (3) was used as the
substrate (Figure 1). The methyl group at the 3-position should
destabilize the transition state in which the 3-methyl group is
oriented to the catalyst (s-cis, corresponding to 16), thus favoring
the s-trans conformer corresponding to 17. As expected, 3
afforded the product 4 (R ) Ph) in 97% yield with 68% ee at
-40 °C for 15 h, using 5 as a catalyst and benzoyl chloride as
an acylating reagent.19 The enantioselectivity was further
improved when the reaction was performed at -78 °C using
acetyl chloride, giving 4 (R ) CH3) in 73% yield with 75% ee
(Table 3, entry 10).20 Although the selectivity and the generality
of the reaction with isoquinoline derivatives were still unsat-
isfactory, these results indicated that differentiation between the
catalyzed and uncatalyzed reaction rates, as well as between
the two reactive conformers (s-trans against s-cis), is important
for high enantioselection.
Ph
Ph
Ph
Ph
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
24
24
24
24
38
24
S
S
a The reaction was conducted at -40 °C using 9 mol % of a catalyst,
2 equiv of TMSCN, and 1.1 equiv of RCOCl, unless otherwise noted.
b Isolated yield. c Determined by chiral HPLC analysis. d 1:1. e 4 equiv
of TMSCN and RCOCl were used.
modeling studies suggested that if the steric bulkiness of the
phosphine oxide’s aryl groups is increased, the Lewis-basic
oxygen atom would be positioned close to the acyl quinolinium
ion activated by the Lewis acid. Thus, we synthesized new
catalysts 6 through 9 containing more bulky phosphine oxides.
Because these catalysts should also have higher Lewis basicity
than the original catalyst 5, higher reactivity as well as higher
enantioselectivity should be realized. The results using these
catalysts on the reaction of quinoline are summarized in Table
2. Although catalysts 7 through 9 gave disappointing results
(entries 6-8), catalyst 6 containing di-o-methylphenylphosphine
oxide afforded improved chemical yield and ee (entries 3 and
4 vs entries 1 and 2). Furthermore, using 4 equiv of TMSCN
and 2-furoyl chloride, the Reissert product was obtained in 91%
yield with 85% ee (entry 5). The inferior results by 7 and 8
might be explained as follows. One of the phosphine oxides of
7 and 8 could coordinate more strongly to the internal aluminum
than in the case of 5, due to the higher electron density (tolyl
and xylyl vs phenyl), thereby diminishing the Lewis acidity of
the catalyst, which should decrease the activity of 7 and 8
relative to 5. Because of the presence of a spontaneous reaction
pathway independent of the chiral catalyst, the reaction mediated
by the less reactive catalyst should result in the lower enanti-
oselectivity. Although catalyst 6 contains a more electron rich
tolyl phosphine oxide, ortho substitution could hinder the
internal coordination to the aluminum. Therefore, the Lewis
acidity of 6 was maintained high enough, relative to 7 or 8. On
the other hand, catalyst 9 containing di-o,o′-dimethylphe-
nylphosphine oxide afforded the opposite enantiomer (S) (entry
8). The reversal of the direction of enantioselectivity might be
important from a mechanistic point of view. In the case of 9,
the Lewis basic phosphine oxide may be too sterically hindered
and should not activate TMSCN, and only act to produce steric
hindrance. Consistent with this, catalyst 10 containing the
diphenylmethyl group gave the S-enantiomer (Table 2, entry
9). It is possible that TMSCN attacks the activated acyl
quinolinium intermediate coordinating the Lewis acid from the
less hindered side opposite the di-o,o′-dimethylphenylphosphine
oxide or diphenylmethyl group. If so, in the case of 5 and 6,
TMSCN attacks the acyl quinolinium from the side of the phos-
phine oxide, supporting the bifunctional catalysis of 5 and 6.
Using the optimized catalyst 6, Reissert compounds were
obtained in up to 96% ee in the case of the reactive, electron-
rich quinolines 1a through 1f (Table 3, entries 1-6). The less
reactive substrates 1h and 1i gave less satisfactory results,
although they still yielded the products in 67% ee (entries 8
and 9). The absolute configuration of 2a was determined as
follows (Scheme 1). Hydrogenation of 2a (R ) Ph) catalyzed
by rhodium on carbon, followed by successive hydrolysis and
(15) Paradisi, M. P.; Romeo, A. J. Chem. Soc., Perkin Trans. 1 1976,
596-600.
(16) The absolute configuration of 2i (R ) 2-furyl) was determined
converting to 12 (R ) 2-furyl) via the following two steps: (1) saturation
of the olefin by hydrogenation catalyzed by rhodium on carbon in MeOH
and (2) debromination by hydrogenolysis catalyzed by palladium on carbon
in MeOH. Unfortunately, partial epimerization (starting from 2i (R )
2-furyl) with 67% ee to obtain 12 (R ) 2-furyl) with 26% ee) occurred
during these steps.
(17) (a) Nagata, R.; Tanno, N.; Kodo, T.; Ae, N.; Yamaguchi, H.;
Nishimura, T.; Antoku, F.; Tatsuno, T.; Kato, T.; Tanaka, Y.; Nakamura,
M. J. Med. Chem. 1994, 37, 3956-3968. (b) Nagata, R.; Tanno, N. Bioorg.
Med. Chem. Lett. 1995, 5, 1527-1532. (c) Katayama, S.; Ae, N.; Nagata,
R. Tetrahedron: Asymmetry 1998, 9, 4295-4299.
(18) Other interesting stereoselective functionalizations of the Reissert
compound were conducted using optically active 2a (85% ee) without any
epimerization, following the reported procedures (Hiessbo¨ck, R.; Kratzel,
M. Heterocycles 1996, 43, 873-882).
(19) Catalyst 6 gave less satisfactory results (32% ee, 99% yield).
Molecular modeling studies indicated that the steric repulsion between the
isoquinoline ring and the bulkier phosphine oxide destabilized the desired
transition state corresponding to 17.
(20) In the case of 3, acetyl chloride gave the better enantioselectivity
because the reaction could be performed at lower temperature. Furthermore,
the desired transition state (corresponding to 17) should be more stable
using a smaller R group in RCOCl that positions cis to the 3-methyl group.

6804 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Table 3. Scope and Limitationsa
Takamura et al.
a The reaction was conducted at -40 °C using 9 mol % of 6, 2 equiv of TMSCN, and 2 equiv of 2-furoyl chloride, unless otherwise noted.
b Isolated yield. c Determined by chiral HPLC analysis. d 4 equiv of TMSCN and 4 equiv of 2-furoyl chloride were used. e TMSCN was added
slowly over 12 h. f The reaction was conducted at -78 °C using 9 mol % of 5, 2 equiv of TMSCN, and 1.1 equiv of acetyl chloride. g The product
was isolated after reduction of the enamine. See the text for details. h Absolute configurations were determined to be R. Absolute configurations of
other products were temporarily assigned.
Scheme 1. Conversion to Methyl
Tetrahydroquinoline-2-carboxylate
Scheme 2. Catalytic Enantioselective Reissert-Type
Reaction with the Isolated Acyl Quinolinium
B. Mechanistic Studies. To confirm that the Reissert-type
reaction proceeds via the acyl quinolinium intermediate, we
performed the following studies. First, formation of the acyl
quinolinium ion was observed using 1H NMR, when quinoline
(1a) was mixed with acetyl bromide in CD3CN or CD2Cl2. This
step was an equilibrium process and the ratio of quinoline and
acyl quinolinium ion was dependent on the temperature (1:1 at
-40 °C and 1:4 at 0 °C in CD3CN). When catalyst 6 and
TMSCN were added to the acyl quinolinium in CD2Cl2 at 0
°C, peaks corresponding to the acyl quinolinium disappeared
and formation of the Reissert compound 2a was confirmed.21
Moreover, we performed the reaction using isolated N-(2-
furoyl)-6-chloroquinolinium triflate (14), prepared following the
reported procedure (Scheme 2).22 As a result, the Reissert
product 2h was obtained from 14 in 11% yield with 38% ee
(vs 57% yield and 67% ee under the best reaction conditions,
see Table 3, entry 8). Although the yield and ee using isolated
14 were not fully comparable with those under the best reaction
conditions, these results provide support that the catalytic
enantioselective Reissert-type reaction proceeds via the acyl
quinolinium intermediate. The insolubility of 14 in CH2Cl2 and/
or the different counteranion (triflate) might compromise the
yield and ee in the case of 14. Finally, the alternative reaction
mechanism involving the direct attack of a cyanide to activated
quinoline could be excluded because we did not observe any
1H NMR peaks corresponding to the adduct of TMSCN to
quinoline in the absence of an acid chloride, even in the presence
of stoichiometric 6. These results, together with the reported
(21) Even if the reaction was conducted in the presence of catalyst 6,
only racemic 2a was obtained using acetyl bromide as an acylating reagent
and CH3CN or CH2Cl2 as a solvent.
(22) Pabel, J.; Ho¨sl, C. E.; Maurus, M.; Ege, M.; Wanner, K. T. J. Org.
Chem. 2000, 65, 9272-9275.

Catalytic EnantioselectiVe Reissert-Type Reaction
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6805
reaction mechanism of Reissert-type reactions,5 led to the
conclusion that the reaction proceeds via the formation of the
acyl quinolinium intermediate as the first step, followed by the
addition of cyanide as the second step.
When the Reissert-type reaction of quinoline using acetyl
bromide was traced by 1H NMR in CD2Cl2, interesting
phenomena were observed. After adding TMSCN to the solution
of the acetyl quinolinium intermediate in the presence of catalyst
6, the peaks corresponding to the intermediate disappeared in
10 min at -40 °C and the product peaks started to emerge.
After the disappearance of the intermediate, peaks corresponding
only to quinoline and the Reissert product 2a were observed.
These observations indicated that cyanation is not the rate-
determining step in the presence of the catalyst. In the absence
of the catalyst, however, all the peaks corresponding to
quinoline, acetyl quinolinium intermediate, and product 2a were
observed during the reaction. Therefore, in the absence of the
catalyst, the cyanation is slow. Thus, the catalyst facilitates the
cyanation step, and as a result, the first step (acyl quinolinium
formation) becomes the rate-determining step.
Figure 2. Postulated reaction mechanism.
Further insight into the reaction mechanism was obtained
from kinetic studies. Utilizing concentration ranges of the
reagents and the catalyst relevant to the actual reaction condi-
tions, the initial reaction rate of 2a formation in CH2Cl2-toluene
(1:1) at -40 °C was 1.2, 0.15, and 0 order with respect to
2-furoyl chloride, TMSCN, and catalyst 6, respectively.23 These
data indicate that the first step (acyl quinolinium formation) is
the major rate-determining step, because the first step involves
2-furoyl chloride, which affected the reaction rate with higher
order. Furthermore, 0 order dependence of the reaction rate on
the catalyst indicates that the catalyst is not involved in the rate-
determining step. From these studies, we postulate the reaction
mechanism shown in Figure 2. The first step is a spontaneous
and reversible addition of quinoline to the acid chloride, which
is the rate-determining step and independent of the catalyst.
When a small amount of the acyl quinolinium intermediate 15
is formed, the addition of TMSCN occurs immediately with
the assistance of the catalyst in the second step. The enanti-
oselective second step should proceed via dual activation:
TMSCN activated by the phosphine oxide of the catalyst attacks
an activated acyl quinolinium ion by the Lewis acid (aluminum
Figure 3. Reaction profiles of 1h at ambient temperature in CH2Cl2
in the presence of TMSCN (4 equiv) and 2-furoyl chloride (4 equiv):
9, standard conditions (9 mol % of 6); b, 9 mol % of 6 premixed with
Reissert product 2h; 2, 9 mol % of 6 premixed with TMSCl just before
use; [, 9 mol % of 6 premixed with TMSCl for 1.5 h; and 0, 9 mol
% of Et2AlCl.
(23) To get further information on the cyanation step, we conducted
kinetic studies using TBDMSCN, expecting the cyanation step to become
the rate-determining step. In this case, the reaction was found to be first
order in both TBDMSCN and catalyst 6 (left graph: initial reaction profiles
varying catalyst concentrations; right graph: log-log plot to determine the
order with respect to the catalyst), which is consistent with the proposed
transition state of the cyanation containing one TMSCN and one catalyst,
as shown in Figure 2. The Reissert-type reaction of 1a using TBDMSCN
(4 equiv) and 2-furoyl chloride (4 equiv) catalyzed by 6 (9 mol %) at -20
°C for 48 h afforded R-2a in 91% yield with 78% ee, which indicates a
similar transition state to that with TMSCN. See Supporting Information
for more details.
metal). The two conformers of the amide bond 16 and 17 would
exist in equilibrium. In the case of 16, the distance between the
activated TMSCN and the electrophilic carbon would be too
remote for catalysis, in addition to a steric repulsion between
the phosphine oxide moiety of the catalyst and the substrate.
The transition state model 17 presents the R configuration of
the major product. Based on this mechanism, if the TMSCN
attack on the acyl quinolinium intermediate proceeds completely
mediated by the catalyst, the order of TMSCN would be the
same as that of the catalyst. The different rate orders of TMSCN
and 6 indicate a very minor contribution of the spontaneous
catalyst-independent reaction. This racemic reaction pathway
should possess a higher order with respect to TMSCN, because
the contribution of the second step to the total reaction rate
would be more significant in the racemic pathway.
We also investigated the reason for moderate yields with
electron deficient quinolines such as 1h and 1i. Tracing the
reaction of 1h using 1H NMR at ambient temperature, the
reaction proceeded during the initial 90 min to an approximately
70% yield of 2h. The reaction slowed considerably, however,
after that point (Figure 3, 9). We speculated that this feature
of the reaction was caused by interference of the reaction by

6806 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Takamura et al.
theproducts generated during the reaction (2h or TMSCl). On
the basis of this assumption, the reaction was performed in the
presence of 1 equiv of 2h. The time course of the reaction,
however, did not change much (Figure 3, b). On the other hand,
when the catalyst (9 mol %) was first premixed with 1 equiv of
TMSCl, the time course of the reaction changed significantly.
The degree of the interference seemed dependent on the
premixing period (Figure 3, premixing period ) 0 min (2) and
1.5 h ([)). Therefore, we concluded that TMSCl inhibited the
reaction.24 Because this interference did not occur when Et2-
AlCl was used as the catalyst (Figure 3, 0), we assume that the
inactivated species is an ate-complex, in which the phosphine
oxide is silylated and two chlorides are bound to the aluminum
binaphthoxide. Deactivation of the catalyst by silylation of the
naphthoxide oxygen could be excluded, because we could not
detect any O-silylated ligand using TLC and 1H NMR.
Preliminary attempts to reduce the concentration of TMSCl by
converting TMSCl to TMSCN in situ using KCN or Bu4NCN
have not been successful.
C. Catalytic Enantioselective Synthesis of L-689,560.
Overactivation of the NMDA subtype of excitatory amino acid
receptors is implicated in several neurodegenerative disorders
including epilepsy, stroke, and Alzheimer’s disease.25 As a
result, NMDA antagonists might have therapeutic benefit,
because these compounds are neuroprotective and anticonvulsant
in a variety of animal models. L-689,560 (Scheme 3, 18) is a
strong NMDA receptor antagonist identified by the Merck
group.10 This tetrahydroquinoline-2-carboxylic acid derivative
contains two chiral centers in a trans relationship. The (-)-
enantiomer (2R,4S) is 540 times more potent than the (+)-
enantiomer (2S,4R).10a Chiral tetrahydroquinoline-2-carboxylic
acid, which should be useful for the synthesis of L-689,540,
has only been available by resolution of the racemic compound.10a
Our synthetic plan (Scheme 3) is to construct the chirality at
the 2-position by the catalytic enantioselective Reissert-type
reaction of 1f,26 followed by a stereoselective reduction of the
resulting enamine to the amine with the desired configuration.
reaction-reduction process. After the Reissert-type reaction was
completed, NaBH3CN, AcOH, and MeOH were directly added
to the reaction mixture. The reduction occurred with very high
selectivity (>20:1) to give the desired trans isomer as the major
product.28 This stereoselectivity can be explained by the
equatorial attack of the hydride to the iminium in the most stable
chairlike conformation. The 0 order dependence of the Reissert-
type reaction on the catalyst indicated that the overall reaction
rate should not change very much even if the catalyst amount
was reduced. Consistent with this expectation, the reaction
proceeded smoothly to produce an 83% yield with 93% ee after
reduction, even when using 1 mol % of the catalyst 6. When
0.5 mol % of catalyst was used, however, ee was lower (87%
ee, 74% yield). Thus, at present, the recommended catalyst
loading of the Reissert-type reaction is 1 mol %. The reaction
is possible on at least the 1-g scale.
Total synthesis of L-689,560 was completed from the key
intermediate 20 as follows (Scheme 4). 20 was converted to
the methyl ester 21 under acidic conditions. The furoyl group
was hydrolyzed, and the resulting carboxylic acid was esterified
to give 22 in 94% yield. At this stage, enantiomerically pure
22 was obtained by recrystallization from hexane (81% yield
after recrystallization). The allyl groups were removed and the
product was isolated as the hydrochloride salt 23. The urea
formation, followed by the hydrolysis of the ester, gave
enantiomerically pure L-689,560 in 47% overall yield through
10 operations (including recrystallization) from quinoline 1f.
This clearly demonstrates that the catalytic enantioselective
Reissert-type reaction offers an efficient and practical synthetic
route for producing the target compound.
The development of an immobilized asymmetric catalysts is
very important for easy separation of the product and potential
to reuse the catalyst. Solid-supported catalysts have many
advantages over homogeneous catalysts, especially in high
throughput organic chemistry.29 On the basis of the studies of
the catalytic enantioselective Strecker-type reaction by im-
mobilized 5,30 we designed a solid-supported catalyst 11. The
chiral ligand 6 was connected to JandaJEL31 through the C-8
linker at the 6-position of the binaphthol core, following a
similar synthetic route to immobilized 5. The loading of the
ligand on the polymer was determined to be 0.32-0.44 mmol/
g, based on the loading of Cl on the commercially available
JandaJEL.32
Scheme 3. Synthetic Plan for L-689,560
Using 3 mol % of 11, 4 equiv of 2-furoyl chloride, and 4
equiv of TMSCN, the Reissert-type reaction of 1f proceeded at
a comparable rate as the homogeneous catalyst 6, giving 20 in
92% yield with 86% ee (Table 4). The second cycle gave the
product in comparable yield and ee. After the third cycle,
however, the ee of the product was lower. These results indicated
that the amount of the effective enantioselective catalyst was
significantly reduced by repeated use under the reaction condi-
tions. This might be caused by either a physical modification
First, the optimized reaction conditions (2 equiv of TMSCN
and furoyl chloride in CH2Cl2 at -40 °C for 40 h) were applied
to the quinoline derivative 1f.27 The Reissert product 2f was
obtained in 29% yield with 91% ee, together with the hydrolyzed
enamine 19 (13%, 90% ee) after the usual workup. When the
reaction was quenched with AcOH to hydrolyze the enamine
completely, 19 was obtained in 80% yield with 90% ee.
Encouraged by these results, we tried a one-pot Reissert-type
(28) The relative configuration was defined by NOE observations, and
also by converting 20 to L-689,560. To explain this excellent diastereose-
lectivity, we searched the most stable conformation of the iminium by the
Monte Carlo method and found the chairlike conformation with equatorial
cyanide to be the energy minimum. The peri-repulsion between the
diallylimino group and the 5-Cl group might make the chairlike conformer
more favored.
(24) The inhibition by TMSCl should increase in the case of less reactive
electron-deficient substrates. Because inactivation of the catalyst by TMSCl
is a time-dependent process, the slower reaction should be influenced more.
To exclude this deactivation process, we tried to use benzoyl fluoride or
acetic anhydride. Reactions using these acylating reagents produce TMSF
or TMSOAc, which should not react with aluminum. However, the reaction
did not proceed at all with these reagents.
(25) Parsons, C. G.; Danysz, W.; Quack, G. Drug News & Perspect.
1998, 11, 523-569 and references cited in ref 10a.
(26) For the synthesis of 1f, see Supporting Information.
(27) No reaction took place from the electron-deficient 5,7-dichloro-
quinoline.
(29) Review: Shuttleworth, S. J.; Allin, S. M.; Wilson, R. D.; Nasturica,
D. Synthesis 2000, 1035-1074.
(30) Nogami, H.; Matsunaga, S.; Kanai, M.; Shibasaki, M. Tetrahedron
Lett. 2001, 42, 279-283.
(31) (a) Toy, P. H.; Janda, K. D. Tetrahedron Lett. 1999, 40, 6329-
6332. (b) Reger, T. S.; Janda, K. D. J. Am. Chem. Soc. 2000, 122, 6929-
6934.
(32) For the synthesis of 11, see Supporting Information.

Catalytic EnantioselectiVe Reissert-Type Reaction
J. Am. Chem. Soc., Vol. 123, No. 28, 2001 6807
Scheme 4. Catalytic Asymmetric Synthesis of L-689,560
Table 4. Catalytic Enantioselective Reissert-Type Reaction of 1f
(30 µL, 0.029 mmol in hexane) was added at ambient temperature to
a solution of the ligand 6-L (22 mg, 0.029 mmol) in CH2Cl2 (2.5 mL)
and the resulting solution was stirred for 1 h. This catalyst solution of
6 was cooled to -40 °C, and a solution of 1c (60.5 mg, 0.32 mmol) in
CH2Cl2 (0.5 mL) was added, followed by the addition of 2-furoyl
chloride (63 µL, 0.64 mmol). After adding toluene (2.5 mL), TMSCN
(85 µL, 0.64 mmol) in toluene (0.5 mL) was added slowly over 24 h
at -40 °C. A saturated aqueous solution of NaHCO3 was added after
40 h and the aqueous layer was extracted with AcOEt. The combined
organic layer was washed with saturated NaCl and dried over NaSO4.
Evaporation of the solvent and purification of the resulting crude product
by silica gel column chromatography (eluent: AcOEt/hexane ) 1/4)
gave the pure product.
Using the Solid-Supported Catalysta
cycle
yield/%
ee/%
1
2
3
4
92
93
93
92
86
84
79
64
a The reaction was performed using 3 mol % of 11, 4 equiv of
2-furoyl chloride, and 4 equiv of TMSCN at -40 °C for 40 h in CH2Cl2.
of the polymer structure under the reaction conditions or a
generation of a nonenantioselective catalyst. Although improve-
ment is needed in terms of reusability, the polymer-supported
catalyst 11 has promising utility. It should be applicable in a
combinatorial synthesis of chiral quinoline derivatives.
trans-N-(2-Furoyl)-2-cyano-4-diallylamino-5,7-dichloro-1,2,3,4-
tetrahydroquinoline (20) (Procedure for One-Pot Synthesis of 20
Using 1 mol % Catalyst). Et2AlCl (34 µL, 0.032 mmol in hexane)
was added at ambient temperature to a solution of the ligand 6-L (24.7
mg, 0.032 mmol) in CH2Cl2 (3 mL), and the resulting solution was
stirred for 1 h. A solution of 1f (940 mg, 3.20 mmol) in CH2Cl2 (2
mL), 2-furoyl chloride (630 µL, 6.40 mmol), and TMSCN (850 µL,
6.40 mmol) was added at -40 °C to the resulting catalyst solution of
6. The mixure was diluted with MeOH (10 mL) after 40 h followed by
the addition of NaBH3CN (402 mg, 6.4 mmol) and acetic acid (200
µL, 3.4 mmol). The whole mixture was allowed to warm to room
temperature, and after 15 h was poured into saturated aqueous NaHCO3
and extracted with Et2O (50 mL, ×2). The combined organic layer
was washed with brine (50 mL ×2) and dried over MgSO4. The solvent
was removed in vacuo and the residue was purified by flash column
chromatography on SiO2 (5∼20% ethyl acetate/hexane) to afford 20
(1.2 g, 91%, 93% ee).
Conclusion
We developed the first catalytic enantioselective Reissert-
type reaction using bifunctional catalysts 5 and 6. Although there
was a competing racemic reaction pathway, the reaction
proceeded mainly via the pathway mediated by the catalyst after
optimization of the reaction conditions (acid chlorides, solvents,
and ligands). Kinetic studies indicated that the reaction should
proceed via two steps: the rate-determining acyl quinolinium
formation followed by the attack of a cyanide. The catalyst
strongly facilitates the second cyanation step. The reaction was
successfully applied to the efficient catalytic asymmetric
synthesis of a potent NMDA receptor antagonist (-)-L-689,-
560. A key step was the one-pot process using the Reissert-
type reaction from quinoline 1f, followed by a stereoselective
reduction of the resulting enamine 2f. This step gave the key
intermediate 20 in 91% yield with 93% ee, using 1 mol % of 6.
The enantiomerically pure target compound was obtained after
10 operations (including recrystallization) in a total yield of 47%.
Furthermore, we immobilized 6 to JandaJEL, and the resulting
solid-supported catalyst 11 afforded 20 in a comparable yield
to the homogeneous 6, but with slightly lower enantioselectivity.
Procedure for the Reissert-Type Reaction Using Polymer-
Supported Catalyst 11. Et2AlCl (31.0 µL of a 0.93 M hexane solution,
28.8 µmol) was added to the immobilized ligand (87 mg, loading )
0.32 mmol/g resin, 38 µmol) swelled in CH2Cl2 (4.0 mL), and the
mixture was stirred for 1 h at room temperature. After the mixture
was cooled to -40 °C, substrate 1f (281 mg, 0.96 mmol) in CH2Cl2
(2.0 mL) and TMSCN (513 µL, 3.84 mmol) were added successively,
and after further addition of 2-furoyl chloride (378 µL, 3.84 mmol),
the mixture was stirred for 40 h at the same temperature. Dry Et2O (3
times the volume of CH2Cl2) was added, and then the reaction mixture
Experimental Section
A Representative Procedure for the Catalytic Asymmetric
Reissert-Type Reaction of Quinolines (Table 3, entry 3). Et2AlCl

6808 J. Am. Chem. Soc., Vol. 123, No. 28, 2001
Takamura et al.
was maintained for 1 h at -78 °C. The supernatant containing the
enamine 2f was removed by a syringe and the resin was washed with
dry Et2O twice under an inert atmosphere. The combined organic layer
was concentrated in vacuo, diluted with MeOH (4.0 mL), and allowed
to undergo reduction with NaBH3CN (60 mg, 0.96 mmol) and acetic
acid (20 µL) at room temperature for 15 h. After the catalyst was dried
under reduced pressure for 1 h, CH2Cl2 solvent, 1f, TMSCN, and
2-furoyl chloride were added at -40 °C to start a new cycle.
Acknowledgment. Financial support was provided by
CREST, The Japan Science and Technology Corporation (JST),
and RFTF of Japan Society for the Promotion of Science.
Supporting Information Available: Experimental proce-
dures and characterization of the products (PDF). This material
is available free of charge via the Internet at http://pubs.acs.org.
JA010654N