Toward a rational design of the assembly structure of polymetallic asymmetric catalysts: design, synthesis, and evaluation of new chiral ligands for catalytic asymmetric cyanation reactions

Article abstract of DOI:10.1016/j.tet.2007.02.081

New chiral ligands (4 and 5) for polymetallic asymmetric catalysts were designed based on the hypothesis that the assembled structure should be stable when made from a stable module 8. A metal-ligand=5:6+μ-oxo+OH complex was generated from Gd(OⁱPr)₃ and 4 or 5, and this complex was an improved asymmetric catalyst for the desymmetrization of meso-aziridines with TMSCN and conjugate addition of TMSCN to α,β-unsaturated N-acylpyrroles, compared to the previously reported catalysts derived from 1-3. These two groups of catalysts produced opposing enantioselectivity even though the ligands had the same chirality. The functional difference in the asymmetric catalysts is derived from differences in the higher-order structure of the polymetallic catalysts.

Full text of DOI:10.1016/j.tet.2007.02.081

Tetrahedron 63 (2007) 5820–5831
Toward a rational design of the assembly structure of polymetallic
asymmetric catalysts: design, synthesis, and evaluation of new
chiral ligands for catalytic asymmetric cyanation reactions
Ikuo Fujimori,a Tsuyoshi Mita,a Keisuke Maki,a Motoo Shiro,b Akihiro Sato,c Sanae Furusho,c
a,
Motomu Kanaia, and Masakatsu Shibasaki
*
*
aGraduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bukyo-ku, Tokyo 113-0033, Japan
bRigaku Corporation, 3-9-12 Matsubara, Akishima, Tokyo 196-8666, Japan
cJASCO International Co., Ltd, 1-11-10 Myojin-Cho, Hachioji, Tokyo 192-0046, Japan
Received 22 January 2007; revised 16 February 2007; accepted 20 February 2007
Available online 23 February 2007
Abstract—New chiral ligands (4 and 5) for polymetallic asymmetric catalysts were designed based on the hypothesis that the assembled
structure should be stable when made from a stable module 8. A metal–ligand¼5:6+m-oxo+OH complex was generated from Gd(OiPr)3
and 4 or 5, and this complex was an improved asymmetric catalyst for the desymmetrization of meso-aziridines with TMSCN and conjugate
addition of TMSCN to a,b-unsaturated N-acylpyrroles, compared to the previously reported catalysts derived from 1–3. These two groups of
catalysts produced opposing enantioselectivity even though the ligands had the same chirality. The functional difference in the asymmetric
catalysts is derived from differences in the higher-order structure of the polymetallic catalysts.
Ó 2007 Elsevier Ltd. All rights reserved.
1. Introduction
products were converted to chiral b-amino acids through
simple acid hydrolysis. There remains room for improve-
ment in this reaction because of its relatively high catalyst
loading (10–20 mol %) requirement, and unsatisfactory
enantioselectivity (80–93% ee). On the other hand, Jacob-
sen’s group reported the first highly enantioselective
catalytic conjugate addition of cyanide to b-aliphatic-
substituted a,b-unsaturated imides using the chiral salen–
Al complex as the catalyst.8 The resulting b-cyano adducts
were converted to chiral g-amino acids under reducing con-
ditions. Specifically, the anticonvulsant drug pregabalin9
was synthesized using this methodology. b-Aromatic or b-
alkenyl-substituted substrate, however, was unreactive under
these conditions. More recently, we reported a more general
catalytic enatioselective conjugate addition of cyanide to
a,b-unsaturated N-acylpyrroles10 using the chiral gadoli-
nium complex derived from ligand 1 (Scheme 1, Eq. 2).11
In addition to substrates containing b-aliphatic substituents,
substrates containing b-aryl, b-alkenyl, and a,b-disubsti-
tuted derivatives produced excellent enantioselectivity. The
long-reaction time (generally, 42–139 h), however, remains
a drawback to our reaction.
Chiral amino acids are important building blocks for natural
products and pharmaceuticals. Although there are several
catalytic asymmetric methods for the synthesis of chiral
a-amino acids,1 few methods exist that can produce chiral
b-amino acids2 and g-amino acids.3 Currently, these amino
acids are of great interest; e.g., b-amino acids form a
well-defined secondary structure by oligomerization,4 and
g-amino acids are unique templates for pharmaceuticals tar-
geting the nervous system. Thus, the development of new
catalytic enantioselective methods for the synthesis of b-
amino acids or g-amino acids is in high demand. The
catalytic asymmetric ring-opening of meso-aziridines with
cyanide and the conjugate addition of cyanide to a,b-unsat-
urated carboxylic acid derivatives are two basic reactions
that can produce direct precursors of such chiral building
blocks.
We reported the first catalytic asymmetric ring-opening
of meso-aziridines with TMSCN in 2005 (Scheme 1, Eq.
1).5,6 A gadolinium complex prepared from Gd(OiPr)3,
ligand 1, and trifluoroacetic acid (TFA)7 in a 1:2:0.5 ratio
was used as the asymmetric catalyst. The ring-opening
To improve the utility and practicality of these two important
reactions, we designed new ligands based on the mechanism
and structure of the Gd catalyst derived from ligands 1–3.
The asymmetric catalysts derived from ligands 4 and 5
* Corresponding authors. Fax: +81 3 5684 5206; e-mail: mshibasa@mol.f.
u-tokyo.ac.jp
0040–4020/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2007.02.081

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5821
i
Gd(OPr)3 (10 mol %)
NO2
(eq. 1)
O
R1
R1
1 (20 mol %)
H
N
R1
R1
TFA (5 mol %)
N
O
TMSCN (3 equiv)
2,6-dimethylphenol (1 equiv)
CH3CH2CN, 0-60 °C
O
NO2
CN
80-93% ee
i
Gd(OPr)3 (5-20 mol %)
CN
O
1 (10-40 mol %)
TMSCN (0.5-1 equiv), HCN (2 equiv)
R
N
R
N
(eq. 2)
CH3CH2CN, -20 °C
R'
R'
83-98% ee
Ph
O
Ph
Ph
Ph
O
P
P
O
O
H
O
H
O
X
X
O
X
X
HO
HO
1: X = F
2: X = H
3: X = Cl
4: X = F
5: X = H
Scheme 1. Catalytic asymmetric ring-opening of meso-aziridines with TMSCN (Eq. 1) and conjugate addition of TMSCN (Eq. 2) using Gd catalysts derived
from 1–3 (previous results), and structures of the chiral ligands.
produced a significantly improved function (enantioselectiv-
ity and catalyst activity) in the two reactions. In this paper,
we describe the full account of the design and utility of the
new catalysts.12
yield¼ca. 80%).14 Interestingly, the higher-order assembly
structure of polymetallic catalysts has predominant effects
on reactivity and enantioselectivity.15 The catalytic species
that mainly contributes to the outcome of each asymmetric
reaction appeared to be different depending on the relative
kinetics of the catalytic species in the reaction mixture.
Therefore, stabilization of one specific assembled catalytic
species should simplify the analysis of the results and further
optimization of the reaction.
2. Results and discussion
2.1. Design and synthesis of new chiral ligands
Previous ESIMS studies indicated that the active asymmetric
catalysts derived from 1–3 are polymetallic complexes. A
catalyst solution prepared from Gd(OiPr)3 and a ligand (1,
2 or 3) in a 1:2 ratio (optimized procedure) contained two
main species—a Gd–ligand¼2:3 complex and a 4:5+m-oxo
complex.13 These two species can be selectively obtained
in pure forms either through a reaction of Gd(HMDS)3
(HMDS¼N(SiMe3)2) and the ligand in a 2:3 ratio for the
2:3 complex or through crystallization from a catalyst solu-
tion in propionitrile–hexane prepared from Gd(OiPr)3 and 3
in a 2:3 ratio (crystal structure¼6 in Fig. 1, crystallization
The three-dimensional crystal structure 614 provided a clue
for this approach. The crystal structure revealed a general
module for construction of the self-assembled catalyst as de-
picted in Figure 1, 7. In this module, the chiral ligand acts as
a tetradentate ligand with each ligand bridging two metals,
and a 7-, 5-, and 5-membered fused chelation ring system
was observed.16 This module structure prompted us to
design a new asymmetric self-assembled polymetallic com-
plex with a more stable higher-order structure. Our hypo-
thesis was that the higher-order structure would be more
stable when assembled from more stable modules. Thus, we
7
Ph
Ph
Ph
O
5
P
Ln
Ph
X
X
P
O
Ln
5
7
6
O
5
O
O
X
5
O
O
X
M
M
5
5
M
M
O
O
n
n
6
7
8
Figure 1. Crystal structure of the polymetallic 4:5+m-oxo complex derived from ligand 3 (6: parts of the complex are deleted for clarity), observed modular unit
of the polymetallic complex (7: a representative module in the polymetallic complex was colorized in blue in 6), and an analogous module 8 derived from new
ligands 4 and 5 designed to produce a more stable polymetallic complex.

5822
I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
designed new ligands 4 and 5 with a truncated linker be-
tween the scaffolding cyclohexane ring and the Lewis base
phosphine oxide. These ligands should form a 6-, 5-, and
5-membered fused chelation ring system in a module
(Fig. 1, 8), which is presumably more stable than the previ-
ous 7.
through complete inversion, and epoxy ether 14 was
obtained in 89% yield.18 Next, we investigated the site-
selective epoxide opening reaction of 14 with a phosphide.
First, commercially available potassium diphenylphosphide
was used as a nucleophile. This reaction, however, provided
many side products. One of the main by-products was de-
rived through b-elimination of the phenoxy group, due to
a strong basicity of potassium diphenylphosphide. To sup-
press such side reactions, lithium diphenylphosphide was
used as a nucleophile.19 The reaction then proceeded cleanly
without any by-products. Moreover, cleavage of the phenolic
methyl ether occurred simultaneously in this step, and 4 was
produced from 14 in one-pot. Enantiomerically and diaste-
reomerically pure 4 was obtained after one recrystallization
from iPrOH. The related catechol-containing 5 was synthe-
sized through the same route using 2-methoxyphenol instead
of 13.
We expected the designed ligands to be synthesized from
a known chiral allylic alcohol 1117 via hydrogen bonding-
assisted epoxidation, introduction of the catechol via
Mitsunobu reaction, and regioselective epoxide opening with
a phosphide as the key steps. Based on this plan, ligand syn-
thesis began with deracemization of racemic allylic alcohol
9 using a palladium(0)-catalyzed allylic substitution
(Scheme 2).17 After conversion of racemic 9 to the corre-
sponding methyl carbonate, allylic substitution using a Trost
ligand (10)-modified palladium(0) catalyst afforded allylic
alcohol 11 with 97% ee in 85% yield. The subsequent epoxi-
dation of 11 with mCPBA proceeded with high diastereo-
selectivity through hydrogen bonding (cis–trans¼24:1).
Slow addition of mCPBA in CH2Cl2 to a solution of 11 for
1 h at 0 ꢀC was necessary to produce high diastereoselectiv-
ity. Because epoxy alcohol 12 is water-soluble, the reaction
mixture was directly filtered through a short pad of alumina
to remove m-chlorobenzoic acid without an aqueous
workup. The Mitsunobu reaction of 12 with 13 proceeded
To evaluate the effects of the oxygen atom of the core six-
membered ring (pyran in 1–3 vs cyclohexane in 4 and 5)
on the catalytic enantioselective reactions, we synthesized
control ligand 15, which contains one methylene linker
between the cyclohexane core and the phosphine oxide,
as shown in Scheme 3. From the same synthetic intermedi-
ate 14 (95% ee), regioselective epoxide opening with
Et2AlCN,20 followed by acidic hydrolysis of the cyanide
mCPBA
NaHCO3
1) MeOCOCl
py., 98%
O
2) Pd2(dba)3 CHCl3 (2 mol %)
OH
97%
cis:trans = 24:1
OH
11
OH
10
(8 mol %)
9
12
O
O
97% ee
NH HN
PPh2 Ph2P
H2O, 85%
Ph
O
Ph
HO
F
F
P
;
1) LiPPh2
O
H2O2
88%
MeO
O
H
13
O
F
F
O
F
F
PPh3, DIAD
89%
2) recryst.
71%
MeO
14
HO
4: >99% ee
Scheme 2. Synthetic scheme of ligand 4.
HO2C
HO
HO
1) Et2AlCN
71%
2) conc. HCl
93%
HO
BH3·THF; recryst.
94%
14
O
F
F
O
F
F
(95% ee)
MeO
MeO
16
17
(99% ee)
Ph
Ph
P
O
1) TsCl, Et3N, DMAP, 100%
2) KPPh2; H2O
O
2
H
O
F
F
3) LiI, 42% (2 steps)
HO
15
Scheme 3. Synthesis of control ligand 15.

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5823
TFA. The enantioselectivity and reactivity were dramati-
cally increased (94% ee, >99% yield for 0.25 h). Screening
of the solvent led to further improvement in enantioselectiv-
ity to 98% ee in THF.
i
catalyst = Gd(OPr)3 + 4
Interestingly, there was a sharp contrast in the dependency of
enantioselectivity on the Gd–ligand ratio between the cata-
lysts prepared from ligands 1 and 4. The enantioselectivity
was dependent on the Gd–ligand ratio in the case of the cat-
alyst prepared from 1, and gradually increased according to
the ratio to the maximum point at a 1:2 ratio (Fig. 2, dotted
line). The enantioselectivity was constantly high (>95% ee)
regardless of the Gd–ligand ratio, however, in the case of the
catalyst prepared from 4 (Fig. 2, solid line). This tendency
suggests that the active catalyst derived from 4 is stable irre-
spective of the mixed ratio of metal and ligand. This depen-
dency is important information about the composition and
stability of the active asymmetric catalyst. ESIMS studies
suggest that the active catalyst derived from 4 is a Gd–
4¼5:6+m-oxo+OH complex (see below). Therefore, we de-
termined Gd–4¼1:1.5 as the optimized ratio for further
studies on substrate generality.
i
catalyst = Gd(OPr)3 + 1
ligand:Gd ratio
Figure 2. Dependency of enantioselectivity on ligand–Gd ratio in catalyst
preparation.
produced carboxylic acid 16. Reduction of the carboxylic
acid with BH3$THF complex furnished diol 17, which was
crystallized from CH2Cl2–hexane to afford enantiomerically
pure 17 in 94% yield. The absolute and relative confi-
guration of 17 was unequivocally determined by X-ray
crystallography (Fig. 4 in Section 4). From 17, selective
tosylation, introduction of diphenylphosphine oxide, and
deprotection of the phenol produced 15. In this case, one-pot
phosphine oxide introduction and deprotection with lithium
phosphide did not proceed cleanly.
Another interesting phenomenon is the sharp temperature
dependency of enantioselectivity, especially in the case of
less reactive cis-stilbene-derived aziridine 18i (Table 1).21
A long-reaction time (136 h) was required for completion
at 40 ꢀC, and product 19i (mixture of anti and syn) was ob-
tained with low to moderate enantioselectivity (entry 1).22
Enantioselectivity and reactivity dramatically improved
when the reaction was conducted at higher temperatures
(entries 3 and 4). When 5 mol % of catalyst was used under
THF reflux (67 ꢀC) conditions, the enantioselectivity and
reactivity were at satisfactory levels (entry 5). This unusual
temperature dependency might be due to the generation of a
product-incorporated asymmetric catalyst of low enantio-
selectivity and activity during the course of the reaction. In
fact, the initial reaction rate of 18i was reasonably fast, even
at 40 ꢀC. The enantioselectivity of this initial phase was rela-
tively high; product 19i was obtained in 7% yield with 80% ee
and 1% yield with 85% ee if the reaction at 40 ꢀC was stopped
in 1 h (entry 2). The reaction became very sluggish after ca.
40% conversion (on TLC analysis). Liberation of the product
from the catalyst via protonolysis (or silylation) might be
retarded in the case of 18i due to both the relatively large
size of the substrate23 and the presumable compact reaction
2.2. Catalytic asymmetric ring-opening reaction of
meso-aziridines with TMSCN
We first examined the catalytic asymmetric aziridine-
opening reaction of 18a with TMSCN using new ligand 4
under the previously optimized conditions (10 mol % of
Gd(OiPr)3, 20 mol % of ligand, 5 mol % TFA, 3 equiv of
TMSCN, and 1 equiv of 2,6-dimethylphenol in propionitrile
at room temperature).5 Product 19a was obtained with only
moderate enantioselectivity (52% ee, 99% yield for 15 h).
The absolute configuration of the product was opposite to
that obtained using previous ligand 1. Speculating that the
highly acidic TFA might have a detrimental effect in the
present case, we investigated the reaction in the absence of
Table 1. Catalytic asymmetric ring-opening of 18i with TMSCN: temperature effects
i
Gd(OPr)3 (x mol %)
O
H
N
Ph
Ph
4 (1.5x mol %)
Ph
Ph
Ar
TMSCN (3 equiv)
2,6-dimethylphenol (1 equiv)
N
O
NO2
CN
19i
THF, temp.
18i
Entry
Loading (x mol %)
Temp (ꢀC)
Time (h)
Yielda (%)
Ratiob (anti–syn)
eec (%) (anti–syn)
1
2
3
4
5
2
2
2
2
5
40
40
60
Reflux
Reflux
136
1
18
6
95
8
98
91
95
1.9:1
6.7:1
0.9:1
0.9:1
0.9:1
11:52
80:85
80:82
85:85
93:93
4
a
Isolated yield.
Determined by 1H NMR.
Determined by chiral HPLC.
b
c

5824
I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
space generated by ligand 4. Higher reaction temperature
(>60 ꢀC) should be required to regenerate a highly enantio-
selective catalyst that is free from the product.
It was previously demonstrated that the enantiomerically
pure products can be obtained via one recrystallization
with high efficiency.5 The products were converted to enan-
tiomerically pure b-amino acids in high yield through hydro-
lysis and purification by ion exchange chromatography
(Scheme 4).
Under the thus optimized conditions, substrate generality
was studied (Table 2). Previous data using ligand 1 are
shown in the same table.5 In all the reactions, catalyst turn-
over number, turnover frequency, and enantioselectivity all
markedly improved using the new ligand 4 compared to
the previous ligand 1. Simple catechol-containing ligand 5
demonstrated higher performance in the specific substrate
18a (entry 2). Importantly, catalysts prepared from ligands
1 and 4 produced opposite enantioselectivity in all cases, de-
spite the same chirality of the two ligands. This reversal in
enantioselectivity was not due to the difference in the core
six-membered ring structure (pyran vs cyclohexane). Thus,
a catalyst prepared from Gd(OiPr)3 and control ligand 15
in a 1:2 ratio (10 mol %) produced the same enantiomer
(with 44% ee) as was obtained using the catalyst prepared
from 1. Therefore, the reversed enantioselectivity be-
tween the catalysts generated from 1 and 4 is due to the
one-carbon difference in the position of the phosphine
oxide relative to Gd (for more detailed discussion, see
Section 2.4).
NH2
1) 12 M HCl, 90 °C
19a
2) Dowex 50Wx8-100
CO2H
92%
20a
1) H2O2, K2CO3
MeOH-H2O, r.t.
Me
Me
NH2
2) 6 M HCl, 90 °C
19h
2) Dowex 50Wx8-100
88%
CO2H
20h
Scheme 4. Conversion to b-amino acids.
2.3. Catalytic enantioselective conjugate addition of
TMSCN to a,b-unsaturated N-acylpyrroles
Initial studies to optimize the reaction conditions of a conju-
gate addition of TMSCN to a,b-unsaturated N-acylpyrroles
Table 2. Catalytic asymmetric aziridine-opening reaction with TMSCN
i
O
R1
Gd(OPr)3 (x mol %)-ligand (1.5x mol %)
H
N
R1
R1
Ar
TMSCN (3 equiv)
N
2,6-dimethylphenol (1 equiv)
O
R1
CN
NO2
THF, temp.
18
(R,R)-19
Entry
Substrate
Ligand
Loading (x mol %)
Temp (ꢀC)
Time (h)
Yielda (%)
eeb (%)
1
4
5
1
2
1
10
rt
rt
0
13
3
20
98
99
94
98f
99f
87g
2c
3d
18a
NR2
4
4
1
2
10
40
rt
12
95
83
85
96
82
NR2
NR2
NR2
18b
18c
18d
5d
6
4
1
2
10
40
rt
14
42
98
91
95
83
7d
8
4
1
2
20
40
40
14
14
98
98
98f
9d
91g
10
4
5
1
2
2
20
40
rt
60
22
74
23
99
99
89
96
96
84
11c
12d
18e
18f
18g
NR2
NR2
CbzN
O
13
4
1
2
20
60
60
28
96
84
92
96
84
14d
15
4
1
5
10
67
60
15
64
99
92
95
80
NR2
16d
Me
17
4
1
2
10
40
rt
14
39
98
93
98f
NR2
NR2
18h
18i
18d
85g
Me
Ph
19
20d
4
1
5
10
67
rt
4
96
95 (47:53)e
81 (54:46)e
93:93
90:80
Ph
a
Isolated yield.
Determined by chiral HPLC.
In the presence of 10 mol % of 2,6-dimethylphenol.
b
c
d
Using a catalyst generated from x mol % of Gd(OiPr)3 and 2x mol % of 1 (x¼10 or 20) in the presence of 5 (or 2.5) mol % of TFA and 1 equiv of 2,6-
dimethylphenol. See Ref. 5.
Ratio of diastereomers determined by 1H NMR.
Absolute configuration was determined as (R,R).
Absolute configuration was determined as (S,S). See Ref. 5.
e
f
g

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5825
focused on the choice of solvent. Propionitrile was the opti-
mum solvent for both the aziridine-opening reaction and the
conjugate addition reaction using ligands 1–3.11 In the case
of the aziridine-opening reaction using 4 and 5, however,
THF was a better solvent than propionitrile with regard to
enantioselectivity and reactivity (see Section 2.2). To define
the optimum solvent in the conjugate addition reaction using
the new catalyst, reactions of N-acylpyrrole 21a in THF and
propionitrile were performed using 5 mol % of catalyst
prepared from 4, 1.5 equiv of TMSCN, and 1 equiv of 2,6-
dimethylphenol as a protic additive at ꢁ20 ꢀC. A much
higher reactivity and enantioselectivity were produced in
propionitrile than in THF (99% yield with 60% ee for 18 h
in THF vs 99% yield with 93% ee for 5.5 h in propionitrile).
Next, we investigated the effects of a protic additive.24 In
the previous system, the addition of hydrogen cyanide pro-
duced markedly higher enantioselectivity than 2,6-dimethyl-
phenol. In the present case, however, 2,6-dimethylphenol
produced slightly better results than hydrogen cyanide.
Therefore, we determined the optimized conditions using
2,6-dimethylphenol as an additive in propionitrile solvent.
from both b-aryl and b-aliphatic-substituted substrates.
The catalyst activity was significantly higher in the present
case using ligand 4 than when using the previous catalyst
prepared from 1. In addition, enantioselectivity was again
reversed. With regard to b-alkenyl and a-substituted
substrates, the enantioselectivity was less satisfactory (entry
14). Thus, the new catalyst was more active and comparably
enantioselective, but demonstrated slightly narrower sub-
strate generality than the previous catalyst in asymmetric
conjugate addition of TMSCN.
The products obtained in those reactions are useful precur-
sors for chiral g-amino acids (Scheme 5), e.g., a b-phenyl-
substituted g-amino butyric acid (GABA) analog, which
has inhibitory activity in the nervous system,25 can be syn-
thesized in three steps from 22e (Scheme 5, Eq. 1). In addi-
tion, known intermediate of pregabalin (24) was synthesized
from 22a.
2.4. Structural studies of the asymmetric catalyst
There are two main contrasting characteristics between the
catalysts prepared from ligands 1–3 and new ligands 4 and
5; (1) the opposing enantioselectivity (Tables 2 and 3) and
Substrate generality was investigated under the optimized
conditions (Table 3). High enantioselectivity was produced
Table 3. Catalytic asymmetric conjugate addition of TMSCN to a,b-unsaturated N-acylpyrroles
i
Gd(OPr)3 (x mol %)
Ligand 4 (1.5x mol %)
CN
O
O
TMSCN (1.5 equiv)
2,6-Dimethylphenol (1 equiv)
EtCN, -20 °C
R
N
R
N
21
22
Entry
Substrate
Ligand
x (mol %)
Time (h)
Yielda (%)
eeb (%)
93d
93d
97e
O
N
1
4
4
1
5
2
5
5.5
14
42
99
99
89
2
21a
3c
O
N
4
4
1
5
5
2.5
42
93
91
93d
98e
21b
21c
21d
5c
O
6
4
1
5
5
2.5
88
95
87
96
90
7c
N
O
8
4
1
5
5
24
43
96
92
86
96
9c
Ph
N
O
10
4
1
5
10
38
98
91
90
88d
91e
21e
21f
11c
N
O
12
4
1
10
10
20
88
89
91
95
89
N
13c
O
14
4
1
5
5
1
8
92 (5.7:1)
99 (1.1:1)
76:12
88:83
N
15c
21h
a
Isolated yield.
Determined by chiral HPLC.
b
c
d
e
Using a catalyst generated from Gd(OiPr)3 and 1 in a 1:2 ratio with 0.5 equiv of TMSCN and 2 equiv of HCN. See Ref. 11.
Absolute configuration was determined as shown in the scheme.
Absolute configuration was opposite to the one shown in the scheme.

5826
I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
1) H2, Pd/C, AcOH
2) Et3N
H2N
Ph
NH
6 M HCl
(eq. 1)
(eq. 2)
22e
CO2H
84% (2 steps)
O
Ph
23
β-phenyl-GABA
NH2
PtO2 (cat.)
H2 (500 psi)
CN
1 M NaOH
94%
22a
CO2H
CO2H
pregabalin
HCl, H2O
(ref. 8a)
24
Scheme 5.
(2) the contrasting dependency of enantioselectivity on the
Gd–ligand ratio in catalyst preparation (Fig. 2). These differ-
ences might be due to the dramatic switching of the higher-
order catalyst structure depending on the position of the
phosphine oxide. To support this hypothesis, we conducted
structural studies on the new catalyst.
ESIMS. The 3:4+2OH complex is not the actual catalyst in
the asymmetric cyanation reaction; using the crystal as a cat-
alyst (Gd¼12.5 mol %), aziridine-opening product 18a was
obtained with only 62% ee (24 h, >99% yield). The crystal,
however, can function as a pre-catalyst of the highly enantio-
selective catalyst (5:6+m-oxo+OH complex). Thus, when
Gd(OiPr)3 was added to the crystal to adjust the Gd–4 ratio
to 5:6 and the solution was heated at 50 ꢀC for 1 h, the enan-
tioselectivity recovered to 94% ee (Gd¼12.5 mol %, 12 h,
>99% yield). Therefore, the assembly state conversion
from the 5:6+m-oxo+OH complex to the 3:4+2OH complex
in the crystallization process was reversible under appropri-
ate conditions. This result again supports the hypothesis that
the 5:6+m-oxo+OH structure of the new catalyst is stable.
Moreover, it is noteworthy that the crystal structure was in
accord with our initial design concept that the polymetallic
complex module would contain a tricyclic 6-, 5-, and
5-membered fused chelation ring system (Fig. 3c).
First, ESIQFTMS studies on the catalyst indicated that the
Gd–4¼5:6+m-oxo+OH complex is the sole species in the
catalyst solution (Fig. 3a). The isotope distribution pattern
precisely matched the calculated pattern. This composition
was constant with Gd–4 ratios of 1:1.2–1:4, which was con-
sistent with the observed constant enantioselectivity inde-
pendent of the Gd–4 ratio in the catalyst preparation.
Therefore, the active catalyst composition is different be-
tween the catalysts prepared from 1–3 (2:3 or 4:5+m-oxo)
and 4, and this difference in the higher-order assembly struc-
ture should be the origin of the improved, opposing enantio-
selectivity, and the enhanced activity of the new catalyst.
The fact that the catalyst composition and enantioselectivity
were constant regardless of the Gd–ligand ratio supports the
idea that the active catalyst structure is more stable in the
new catalyst prepared from 4 than the previous catalyst
derived from 1–3.
3. Conclusion
We developed new asymmetric polymetallic catalysts that
are useful for a ring-opening reaction of meso-aziridines
with TMSCN and conjugate addition of TMSCN to a,b-
unsaturated N-acylpyrroles. In most cases, the new catalysts
derived from 4 and 5, whose phosphine oxide is directly at-
tached to the core six-membered cyclohexane ring, produced
significantly improved activity and enantioselectivity com-
pared to the previous catalysts derived from 1–3. Although
these two groups of chiral ligands contain the same chirality,
asymmetric catalysts derived from these ligands demon-
strated opposite enantioselectivity. The origin of the func-
tional difference of the two asymmetric catalysts is the
difference in their higher-order assembled structure: the
To obtain three-dimensional information of the polymetallic
structure, we next attempted crystallization of the catalyst.
After intensive effort, we succeeded in obtaining colorless
prisms from a solution of Gd(OiPr)3 and 4 mixed in a 1:2
ratio in propionitrile (77% yield). Crystallographic analysis
revealed that the crystal was a Gd–ligand¼3:4+2OH com-
plex (Fig. 3b),26 which is different from the observed species
using ESIMS (Fig. 3a). The composition of the crystal was
stable after dissolving in a solvent, and the peak correspond-
ing to the 3:4+2OH complex was the sole observed peak in
F
(a)
(b)
(c)
Gd:4 = 5:6+ -oxo+OH+H+
3474.3881
F
Ph
Ph
O
P
Ph
F
Ph
O
O
P
OH
O
O
F
Ph
O
Ph
P
O
O
Gd
Gd
H
6
O
Ph
O H O
O
P
O
O
5
Ph
Gd
O
O
5
O
F
F
F
F
Figure 3. (a) ESI-QFT-MS chart of the catalyst prepared from Gd(OiPr)3 and 4 (the optimized catalyst). (b) X-ray structure of Gd–4¼3:4+2OH. (c) Chemical
depiction of Gd–4¼3:4+2OH.

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5827
active catalysts derived from 1–3 were Gd–ligand¼2:3 com-
plexes or 4:5+m-oxo complexes; the active catalysts derived
from 4 and 5, however, were Gd–ligand¼5:6+m-oxo+OH
complexes. Thus, a minor structural (one carbon) difference
in the module was amplified in the higher-order structure
of a polymetallic catalyst, and caused a dramatic difference
in the asymmetric catalyst function. The results of this
study clearly demonstrate the importance of the higher-order
structure of the asymmetric polymetallic catalyst. Elucida-
tion of the enantio-differentiation mechanism and extension
of the utility of the new catalyst is currently ongoing in our
group.
4.2.2. (1R,2R,6R)-2-(40,50-Difluoro-20-methoxyphenoxy)-
7-oxabicyclo[4.1.0]heptane (14). To a solution of PPh3
(1.12 g, 4.27 mmol, 1.5 equiv) and 4,5-difluoro-2-methoxy-
phenol (13: 684 mg, 4.27 mmol, 1.5 equiv), DIAD (diiso-
propyl azodicarboxylate, 840 mL, 4.27 mmol, 1.5 equiv)
was added dropwise with cooling in an ice bath. Epoxy alco-
hol 12 in THF was added dropwise to the mixture, and the
mixture was warmed to room temperature. After stirring
for 22 h, water was added, and the organic layer was sepa-
rated. The aqueous layer was extracted with EtOAc twice,
and the combined organic layers were washed with brine,
and dried over Na2SO4. The solvent was removed under re-
duced pressure, and the residue was purified by flash column
chromatography (silica gel, hexane–EtOAc, 5:1) to afford 14
(649 mg, 2.53 mmol) as a colorless solid in 89% yield with
4. Experimental
4.1. General
93% ee. IR (KBr): 2846, 1519, 1221 cmꢁ1 1H NMR
;
(CDCl3): d¼6.86 (dd, J¼12.0, 7.7 Hz, 1H), 6.73 (dd,
J¼12.0, 7.7 Hz, 1H), 4.36 (dd, J¼9.0, 5.5 Hz, 1H), 3.82 (s,
3H), 3.28–3.26 (m, 1H), 3.22 (d, J¼3.5 Hz, 1H), 2.10–
2.04 (m, 1H), 1.95–1.89 (m, 1H), 1.85–1.78 (m, 1H),
1.57–1.51 (m, 1H), 1.48–1.40 (m, 1H), 1.33–1.24 (m, 1H);
13C NMR (CDCl3): d¼145.2–141.0 (multiple peaks due to
coupling with 19F), 104.8 (d, J¼21.7 Hz), 100.5 (d,
J¼21.7 Hz), 72.8, 55.0, 52.3, 51.7, 25.0, 22.6, 13.1; MS
(ESI): m/z 279 [M+Na+]; HRMS (EI): m/z calcd for
C13H14F2O3 [M+]: 256.0906. Found: 256.0931; [a]2D5
+4.26 (c 1.00, CHCl3) (93% ee), HPLC (Chiralpak AD-H,
2-propanol–hexane 1:99, flow 1.0 mL minꢁ1, detection at
254 nm): tR 12.6 min (minor) and 20.4 min (major).
Infrared (IR) spectra were recorded on a JASCO FT/IR 410
Fourier transform infrared spectrophotometer. NMR spectra
were recorded on a JEOL JNM-LA500 spectrometer, operat-
ing at 500 MHz for 1H NMR and 126.65 MHz for 13C NMR.
Chemical shifts in CDCl3 were reported in the scale relative
to CHCl3 (7.24 ppm) for 1H NMR, and to CDCl3 (77.0 ppm)
for 13C NMR, as internal references. Optical rotations were
measured on a JASCO P-1010 polarimeter. ESI-Q mass
spectra were measured on Water-ZQ4000. EI Mass spectra
were measured on JEOL JMS-BU20 GCmate. ESI-QFT
mass spectra were measured on IonSpec QFT-7 (JASCO In-
ternational Co., Ltd.). X-ray data were collected on a Rigaku
RAXIS RAPID imaging plate area detector with graphite
monochromated Mo Ka radiation. The enantiomeric ex-
cesses (ee’s) were determined by HPLC. HPLC analysis
was performed on JASCO HPLC systems containing of
following: pump, PU-980; detector, UV-970, measured at
254 nm; column, Daicel Chiralpak AD-H, AS-H, or Daicel
Chiralcel OD-H; mobile phase, 2-propanol–hexane. In gen-
eral, reactions were carried out in dry solvents under an
argon atmosphere, unless noted otherwise. Dry solvents of
tetrahydrofuran (THF) and dichloromethane (CH2Cl2)
were purchased from Kanto Chemical Co., Inc. Propionitrile
was distilled from calcium hydride. Other reagents were
purified by usual methods. Gd(OiPr)3 was purchased from
Kojundo Chemical Laboratory Co., Ltd. (fax: +81 492 84
1351, sales@kojundo.co.jp). Caution! TMSCN is very
toxic, and experiments should be carried out in a well-
ventilated hood.
4.2.3. (1S,2S,6R)-2-Diphenylphosphinyl-6-(40,50-difluoro-
20-phenoxy)cyclohexanol (4). To a solution of 14 (1.50 g,
5.85 mmol) and Ph2PH (3 mL, 17.6 mmol, 3.0 equiv) in
THF (20 mL), BuLi (1.6 M in hexane, 11 mL, 17.6 mmol,
3.0 equiv) was added dropwise at ꢁ78 ꢀC, and the mixture
was stirred at room temperature for 6 h. Saturated aqueous
NH4Cl was added. Then, 30% H2O2 was added dropwise
to the mixture with cooling in an ice bath, and the mixture
was stirred at room temperature for 12 h. Saturated aqueous
Na2S2O3 was added at 0 ꢀC. The mixture was diluted with
EtOAc, and the organic layer was washed with water twice.
The aqueous layer was extracted with EtOAc twice, and the
combined organic layers were washed with brine before
being dried over Na2SO4. After filtration, the solvent was
removed under reduced pressure, and the residue was
purified by column chromatography (silica gel, hexane–
EtOAc, 1:1–1:2) to afford 4 (2.29 g, 5.16 mmol) as a color-
less solid in 88% yield. The compound 4 was recrystallized
i
from PrOH to afford enantiomerically and diastereomeri-
cally pure 4 in 71% yield. IR (KBr): 3334, 1515, 1173,
4.2. Syntheses of 4 and 5
1
1150 cmꢁ1; H NMR (CDCl3): d¼9.10 (br s, 1H), 7.78–
4.2.1. (1S,2S,3R)-cis-2,3-Epoxycyclohexan-1-ol (12).27 To
a suspension of 1117 (2.86 g, 29.1 mmol) and NaHCO3
(2.50 g, 29.1 mmol, 1.0 equiv) in CH2Cl2 (100 mL),
mCPBA (60% purity, 8.4 g, 29.1 mmol, 1.0 equiv) in
CH2Cl2 (50 mL) was added dropwise over 1.5 h with cool-
ing in an ice bath. After stirring for 5 h at 0 ꢀC, about
a half of the solvent was removed under reduced pressure.
The mixture was purified by short pad column chromato-
graphy (alumina, CH2Cl2–EtOAc) and flash column chro-
matography (silica gel, hexane–EtOAc, 3:1–3:2) to afford
12 (3.21 g, 28.1 mmol) as a pale yellow oil in 97% yield
(cis–trans¼96:4).
7.71 (m, 4H), 7.66–7.58 (m, 2H), 7.57–7.50 (m, 4H), 6.88
(br s, 1H), 6.79–6.73 (m, 2H), 4.10 (dd, J¼18.0, 8.6 Hz,
1H), 3.71 (ddd, J¼9.8, 9.8, 4.6 Hz, 1H), 2.67–2.61 (m,
1H), 2.22 (d, J¼10.4 Hz, 1H), 1.81–1.72 (m, 2H), 1.46–
1.41 (m, 1H), 1.38–1.30 (m, 1H), 1.30–1.26 (m, 1H) (chemi-
cal shift depends on the substrate concentration); 13C NMR
(CDCl3): d¼147.5–142.0 (multiplet, coupled with fluoride),
132.5–128.9 (multiplet, coupled with phosphorus), 108.1 (d,
J¼20.7 Hz), 105.7 (d, J¼21.7 Hz), 86.6 (d, J¼14.5 Hz),
72.5 (d, J¼5.2 Hz), 41.8, 41.2, 30.0, 25.5, 23.8, 23.7 (d, J¼
15.5 Hz); MS (ESI): m/z 467 [M+Na+]; HRMS (EI): m/z
calcd for C24H23F2O4P[M+]: 444.1297. Found: 444.1283;

5828
I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
[a]2D5 +58.2 (c 1.00, CHCl3) (>99% ee), HPLC (Chiralcel
OD-H, 2-propanol–hexane 1:9, flow 1.0 mL minꢁ1, detec-
tion at 254 nm): tR 6.9 min (minor) and 10.2 min (major).
acidify, the aqueous layer was extracted with EtOAc twice.
The combined organic layers were washed with brine and
dried over Na2SO4. The solvent was removed under reduced
pressure to afford 16 (303 mg, 1.00 mmol) in 81% yield. 1H
NMR (CDCl3): d¼6.89 (dd, J¼10.7, 7.9 Hz, 1H), 6.76 (dd,
J¼11.3, 7.7 Hz, 1H), 3.93 (dd, J¼10.6, 8.8 Hz, 1H), 3.85 (s,
3H), 3.68 (ddd, J¼11.3, 8.8, 4.9 Hz, 1H), 2.43 (ddd, J¼12.5,
10.6, 4.3 Hz, 1H), 2.21–2.15 (m, 1H), 2.14–2.08 (m, 1H),
1.90–1.84 (m, 1H), 1.60–1.46 (m, 3H), 1.37–1.28 (m, 1H);
MS (ESI): m/z 325 [M+Na+]; HRMS (EI): m/z calcd for
C14H17F2O5: 303.1044. Found: 303.1049.
4.2.4. (1S,2S,6R)-2-Diphenylphosphinyl-6-(20-phenoxy)-
cyclohexanol (5). Prepared via the same procedure as de-
scribed above using guaiacol instead of 13. IR (KBr):
1
3312, 1499, 1179, 1154 cmꢁ1; H NMR (CDCl3): d¼8.88
(br s, 1H), 7.74–7.71 (m, 4H), 7.53–7.45 (m, 4H), 6.96–
6.88 (m, 3H), 6.73–6.70 (m, 1H), 6.61 (br s, 1H), 4.14 (dd,
J¼18.0, 8.3 Hz, 1H), 3.81–3.76 (m, 1H), 2.65 (ddd,
J¼22.0, 10.0, 3.4 Hz, 1H), 2.27 (d, J¼10.0 Hz, 1H), 1.79–
1.75 (m, 1H), 1.62 (d, J¼12.8 Hz, 1H), 1.49–1.41 (m, 1H),
1.38–1.32 (m, 1H), 1.26–1.16 (m, 1H); 13C NMR (CDCl3):
d¼149.4, 146.4, 132.4–128.8 (multiple peaks due to cou-
pling with phosphorus), 124.1, 119.4, 119.3, 117.1, 85.8
(d, J¼11.2 Hz), 72.6 (d, J¼5.2 Hz), 41.9, 41.4, 30.1,
25.6, 23.7 (d, J¼15.5 Hz); MS (ESI): m/z 431 [M+Na+];
HRMS (EI): m/z calcd for C24H25O4P[M+]: 408.1485.
Found: 408.1461; [a]D26 +58.4 (c 1.00, CHCl3) (>99% ee),
HPLC (Chiralcel OD-H, 2-propanol–hexane 1:9, flow
1.0 mL minꢁ1, detection at 254 nm): tR 11.8 min (major)
and 19.0 min (minor).
4.3.3. (1R,2R,6R)-2-(40,50-Difluoro-20-methoxyphenoxy)-
6-(hydroxymethyl)cyclohexanol (17). To a solution of 16
(119 mg, 0.394 mmol) in THF (4.0 mL), BH3$THF
(1.35 mL, 1.575 mmol, 4.0 equiv) was added, and the mix-
ture was stirred at room temperature for 1.5 h. After adding
1 N HCl, the mixture was diluted with EtOAc. The organic
layer was separated and washed twice with saturated aque-
ous NaHCO3 and water. The aqueous layer was extracted
with EtOAc, and the combined organic layers were washed
with brine before being dried over Na2SO4. The solvent was
removed under reduced pressure to afford diol 17 (114 mg,
0.394 mmol) as a colorless solid in quantitative yield. The
compound 17 was recrystallized from CH2Cl2–hexane to
afford optically enriched 17 in 55% yield with 99% ee.
1H NMR (CDCl3): d¼6.87 (dd, J¼11.0, 8.0 Hz, 1H), 6.74
(dd, J¼11.9, 7.6 Hz, 1H), 3.91 (br s, 1H), 3.84 (s, 3H),
3.75–3.62 (m, 4H), 2.17–2.11 (m, 1H), 1.82–1.77 (m, 1H),
1.74–1.60 (m, 2H), 1.51–1.42 (m, 1H), 1.38–1.28 (m, 1H),
1.10–1.01 (m, 1H); MS (ESI): m/z 311 [M+Na+]; HRMS
(EI): m/z calcd for C14H19F2O4: 289.1251. Found:
289.1246; [a]2D2 ꢁ42.9 (c 0.87, CHCl3) (99% ee). HPLC
(Chiralpak AD-H, 2-propanol–hexane 1:9, flow
1.0 mL minꢁ1, detection at 254 nm): tR 12.2 min (major)
and 14.1 min (minor). The absolute and relative configura-
tions were unequivocally determined at this stage by X-ray
crystallography (Fig. 4, R2 value¼7%).26 The absolute con-
figuration was determined based on the Flack parameter
(0.01(13)). This X-ray structure also confirmed the absolute
configuration of ligands 4 and 5.
4.3. Synthesis of 15
4.3.1. (1R,2R,3R)-3-(40,50-Difluoro-20-methoxyphenoxy)-
2-hydroxycyclohexanecarbonitrile. To a solution of 14
(100 mg, 0.390 mmol) in Et2O (2 mL), Et2AlCN (470 mL,
0.468 mmol, 1.2 equiv) was added dropwise with cooling
in an ice bath. After stirring for 15 min, saturated aqueous
Rochelle salt was added to the mixture and stirred for 1 h.
The organic layer was separated and washed with water.
The aqueous layer was extracted with EtOAc. The combined
organic layers were washed with brine, and dried over
Na2SO4. The solvent was removed under reduced pressure,
and the residue was purified by column chromatography (sil-
ica gel, hexane–EtOAc, 5:2) to afford the product (80 mg,
1
0.281 mmol) as a pale yellow oil in 72% yield. H NMR
(CDCl3): d¼6.85 (dd, J¼12.0, 7.6 Hz, 1H), 6.75 (dd,
J¼12.0, 7.6 Hz, 1H), 4.02 (br s, 1H), 3.83 (s, 3H), 3.82–
3.78 (m, 1H), 3.60 (ddd, J¼12.5, 8.5, 4.6 Hz, 1H), 2.50
(ddd, J¼12.5, 10.0, 3.7 Hz, 1H), 2.18–2.07 (m, 2H), 1.88–
1.82 (m, 1H), 1.68–1.59 (m, 1H), 1.56–1.48 (m, 1H),
1.34–1.24 (m, 1H); 13C NMR (CDCl3): d¼147.6–142.7
(multiple peaks due to coupling with 19F), 120.2, 110.4 (d,
J¼19.7 Hz), 102.1 (d, J¼22.7 Hz), 86.2, 73.9, 56.8, 35.3,
30.0, 28.3, 22.8; MS (ESI): m/z 306 [M+Na+]; HRMS
(EI): m/z calcd for C14H16F2NO3: 284.1098. Found:
284.1096.
4.3.4. ((1R,2R,3R)-3-(40,50-Difluoro-20-methoxyphenoxy)-
2-hydroxycyclohexyl)methyl-400-benzene sulfonate. To
a solution of 17 (38 mg, 0.132 mmol), Et3N (37 mL,
0.264 mmol, 2.0 equiv), and DMAP (1.6 mg, 13.3 mmol,
10 mol %) in CH2Cl2 (1.3 mL), TsCl (51 mg, 0.265 mmol,
2.0 equiv) was added, and the mixture was stirred at room
temperature for 6 h. The mixture was diluted with EtOAc,
and washed twice with 1 N HCl. The aqueous layer was
extracted with EtOAc, and the combined organic layers
were washed with brine before being dried over Na2SO4.
The solvent was removed under reduced pressure, and the
residue was purified by column chromatography (silica
gel, hexane–EtOAc, 3:1) to afford the tosylate (59.6 mg,
4.3.2. (1S,2R,3R)-3-(40,50-Difluoro-20-methoxyphenoxy)-
2-hydroxycyclohexanecarboxylic acid (16). To a solution
of the nitrile (350 mg, 1.236 mmol) synthesized in Section
4.3.1 in DME (15 mL), 12 N HCl was added and the mixture
was stirred at 90 ꢀC for 24 h. The mixture was diluted with
EtOAc, and the organic layer was separated and washed
with water twice. The aqueous layer was extracted with
EtOAc twice, and the combined organic layers were washed
with brine before being dried over Na2SO4. The solvent was
removed under reduced pressure. The residue was dissolved
in 3 N NaOH, and diluted with water. The aqueous layer was
washed with Et2O four times. After 1 N HCl was added to
1
0.135 mmol) as a pale yellow oil in quantitative yield. H
NMR (CDCl3): d¼7.81 (d, J¼8.1 Hz, 2H), 7.34 (d,
J¼8.1 Hz, 2H), 6.84 (dd, J¼11.0, 7.9 Hz, 1H), 6.73 (dd,
J¼11.0, 7.3 Hz, 1H), 4.22 (dd, J¼9.5, 3.1 Hz, 1H), 4.16
(dd, J¼9.5, 6.1 Hz, 1H), 3.81 (s, 3H), 3.66–3.61 (m, 1H),
3.50 (t, J¼9.5 Hz, 1H), 3.40 (s, 1H), 2.45 (s, 3H), 2.13–
2.07 (m, 1H), 1.80–1.70 (m, 3H), 1.48–1.41 (m, 1H),
1.26–1.21 (m, 2H); MS (ESI): m/z 465 [M+Na+]; HRMS

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5829
was added at room temperature. The mixture was stirred at
54 ꢀC for 1 h, and then the solvent was evaporated. After
drying the resulting pre-catalyst under reduced pressure
(<5 mmHg) for 2 h, 18a (246 mg, 1.0 mmol), 2,6-dimethyl-
phenol (122 mg, 1.0 mmol, 1.0 equiv), and THF (5.0 mL)
were added at room temperature. TMSCN (40 mL,
0.30 mmol, 3.0 equiv) was added to start the reaction. After
13 h, water and EtOAc were added. The organic layer was
separated, and the aqueous layer was extracted twice with
EtOAc. The combined organic layers were dried over
Na2SO4. The solvent was removed under reduced pressure,
and the residue was purified by flash column chromato-
graphy (silica gel, hexane–AcOEt, 3:1–3:2) to afford 19a
(269 mg, 0.99 mmol) in 99% yield as a colorless solid.
The enantiomeric excess of the product was determined by
HPLC analysis to be 98% ee. Characterization of the prod-
ucts and ee determination methods were described in Ref. 5.
Figure 4. X-ray crystal structure of 17.
4.5. Conversion of the aziridine-opening products to
b-amino acids
(EI): m/z calcd for C21H25F2O6S [M+H+]: 443.1334. Found:
443.1335; [a]1D9 ꢁ0.47 (c 0.88, CHCl3) (99% ee).
The following conversions were conducted using the corre-
sponding enantiomeric products.
4.3.5. (1R,2R,6S)-6-Diphenylphosphinylmethyl-2-(40,50-
difluoro-2-phenoxy)cyclohexanol (15). The tosylate
(86 mg, 0.195 mmol) was dissolved in THF (1 mL), and
KPPh2 (0.5 M solution in THF, 858 mL, 0.429 mmol,
2.2 equiv) was added dropwise to the solution with cooling
in an ice bath. After stirring for 15 min, 30% H2O2 was
added and stirred for 6 h. The saturated aqueous Na2S2O3
was added, and the solution was diluted with EtOAc. The or-
ganic layer was separated, and washed with water and brine
before being dried over Na2SO4. The solvent was removed
under reduced pressure. Without purification, the resulting
crude oil was dissolved in DMF (1 mL). LiI (157 mg,
1.17 mmol, 6.0 equiv) was added, and the mixture was
stirred at 180 ꢀC for 19 h. Water and EtOAc were added.
The aqueous layer was extracted with EtOAc, and the com-
bined organic layers were washed with brine before being
dried over Na2SO4. The solvent was removed under reduced
pressure, and the residue was purified by column chromato-
graphy (silica gel, hexane–EtOAc, 1:2) to afford 15 (74 mg,
0.162 mmol) as a colorless solid in 83% yield. IR (KBr):
4.5.1. (1S,2S)-2-Amino-cyclohexanecarboxylic acid (ent-
20a).28 Amido nitrile ent-19a (>99% ee; recrystallized
from 1-butanol) (201.0 mg, 0.735 mmol) was treated with
12 M HCl (20 mL) at 90 ꢀC for 5 days. The mixture was
concentrated under reduced pressure and the residue was
dissolved in 1 M HCl. The solution was washed with
CH2Cl2 three times and the aqueous phase was concentrated.
The residue was purified through ion exchange chromato-
graphy (Dowex 50Wꢂ8-100 (acidic) (10 g)) to afford ent-
20a (97.1 mg) as a white solid in 92% yield. The absolute
configuration was determined based on the comparison of
the optical rotation with the reported value. [a]2D3 +78.8 (c
0.246, MeOH).
4.5.2. (2S,3S)-3-Amino-2-methyl butyric acid (ent-20h).29
To a solution of amido nitrile ent-19h (80% ee) (14.6 mg,
0.059 mmol) in MeOH–H2O (2 mL), H2O2 (1 mL), and sat-
urated aqueous Na2CO3 (1 mL) were added and the mixture
was stirred at room temperature for 3 h. Na2S2O6 and
CH2Cl2 were added and the organic layer was separated.
The aqueous layer was extracted with CH2Cl2 twice and
the combined organic layers were dried over Na2SO4. The
solvent was removed under reduced pressure and the residue
was treated with 6 M HCl (2 mL) at 90 ꢀC. After 62 h, the
mixture was diluted with water and washed twice with
CH2Cl2. The aqueous phase was concentrated and the resi-
due was purified through ion exchange chromatography
(Dowex 50Wꢂ8-100 (acidic) (2.2 g)) to afford ent-20h
(6.1 mg) as a white solid in 88% yield. The absolute config-
uration was determined based on the comparison of the
optical rotation with the reported value. [a]2D3 +6.5 (c
0.270, H2O).
3216, 1512, 1157 cmꢁ1
;
1H NMR (CDCl3): d¼9.59 (s,
1H), 7.78–7.69 (m, 5H), 7.60–7.46 (m, 6H), 6.78 (dd,
J¼10.7, 8.3 Hz, 1H), 6.72 (dd, J¼11.6, 8.0 Hz, 1H), 3.56
(t, J¼9.0 Hz, 1H), 3.34 (ddd, J¼11.3, 9.0, 4.9 Hz, 1H),
2.44–2.39 (m, 2H), 2.18–2.13 (m, 1H), 1.80–1.67 (m, 2H),
1.57–1.48 (m, 1H), 1.35–1.26 (m, 2H), 1.19–1.10 (m, 1H);
13C NMR (CDCl3): d¼148.2–141.4 (multiple peaks due to
coupling with fluoride), 132.5–128.8 (multiple peaks due
to coupling with phosphorus), 111.1 (d, J¼18.6 Hz), 105.2
(d, J¼21.7 Hz), 88.3, 76.7, 38.9 (d, J¼4.1 Hz), 36.8, 36.2,
34.2 (d, J¼13.4 Hz), 30.5, 22.9; MS (ESI): m/z 481
[M+Na+]; HRMS (EI): m/z calcd for C24H25O4P[M+]:
458.1453. Found: 458.1472; [a]2D5 ꢁ35.4 (c 0.70, CHCl3)
(99% ee).
4.4. General procedure for the catalytic asymmetric
ring-opening reaction of meso-aziridines with TMSCN
(Table 2, entry 1)
4.6. General procedure for the catalytic asymmetric
conjugate addition of TMSCN to a,b-unsaturated
N-acylpyrroles (Table 3, entry 1)
To a solution of ligand 4 (13.3 mg, 0.03 mmol) in THF
(0.6 mL), Gd(OiPr)3 (0.2 M in THF, 100 mL, 0.02 mmol)
To a solution of ligand 4 (6.7 mg, 0.015 mmol) in THF
(0.3 mL), Gd(OiPr)3 (0.2 M in THF, 50 mL, 0.01 mmol)

5830
I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
was added at room temperature. The mixture was stirred at
54 ꢀC for 1 h, and then the solvent was evaporated. After
drying the resulting pre-catalyst under reduced pressure
(<5 mmHg) for 2 h, 2,6-dimethylphenol (24.4 mg,
0.2 mmol, 1.0 equiv) was added, followed by addition of
21a (35.4 mg, 0.2 mmol) in THF (0.2 mL). The mixture
was cooled down to ꢁ20 ꢀC, and then TMSCN (40 mL,
0.30 mmol, 1.5 equiv) was added to start the reaction. After
5.5 h, silica gel was added. The slurry was directly loaded on
silica gel column, and purified by flash column chromato-
graphy (silica gel, hexane–AcOEt, 10:1–4:1) to afford 22a
(40.5 mg) in 99% yield as a colorless solid. The enantiomer-
ic excess of the product was determined by HPLC analysis to
be 93% ee. Characterization of the products and ee determi-
nation methods were described in Ref. 11.
were dried over Na2SO4 and concentrated to afford ent-24
(31.8 mg) in 94% yield as colorless oil. IR (neat): 2961,
1
2244, 1714, 1469, 1414, 1371, 1175, 924, 619 cmꢁ1; H
NMR (CDCl3): d¼9.29 (br s, 1H), 3.12–2.96 (m, 1H), 2.76
(dd, J¼7.5, 17.0 Hz, 1H), 2.62 (dd, J¼6.1, 17.0 Hz, 1H),
1.95–1.78 (m, 1H), 1.74–1.57 (m, 1H), 1.43–1.30 (m, 1H),
0.98 (d, J¼6.7 Hz, 3H), 0.96 (d, J¼6.7 Hz, 3H); 13C NMR
(CDCl3): d¼175.5, 120.8, 40.6, 36.8, 26.1, 25.5, 22.8,
21.2; MS (ESI): m/z 155 [M+]; HRMS (EI): m/z calcd for
C8H14NO2 [M+H+]: 156.1025. Found: 156.1026; [a]2D1
+15.0 (c 0.590, CHCl3). Compound 24 was converted to
pregabalin via PtO2-catalyzed hydrogenation.8a
4.8. Preparation of crystal from catalyst solution
To a solution of ligand 4 (89.0 mg, 0.20 mmol) in THF
(3.0 mL), Gd(OiPr)3 (0.2 M in THF, 500 mL, 0.10 mmol)
was added at room temperature. The mixture was stirred
at 54 ꢀC for 1 h, and then the solvent was evaporated at
0 ꢀC. After drying the residue under reduced pressure
(<5 mmHg) for 2 h, propionitrile (350 mL) was added to dis-
solve the residue. After 1 day, colorless prisms appeared.
The solvent was removed with a syringe, and the crystals
were washed twice with propionitrile. The resulting color-
less prisms were dried under vacuum to afford the
3:4+2OH crystal (68 mg) in 77% yield.
4.7. Conversion of the conjugate adducts to g-amino
acids
The following conversions were conducted using the corre-
sponding enantiomeric products.
4.7.1. (4S)-4-Phenyl-2-pyrrolidinone (ent-23).30 To a solu-
tion of ent-22e (>99% ee; recrystallized from 2-propanol)
(9.7 mg, 0.0067 mmol) in MeOH (0.5 mL) and AcOEt
(0.5 mL), AcOH (25 mL) and 10% Pd–C (7.1 mg,
0.0067 mmol) were added, and the mixture was stirred under
20 atm pressure of hydrogen at room temperature. After
20 h, catalyst was filtered off through Celite, and the solvent
was removed under reduced pressure. To this residue, tolu-
ene (2 mL) and Et3N (500 mL) were added. The mixture
was heated at 100 ꢀC for 2 h, and 1 M HCl and CH2Cl2
were added and the organic layer was separated. The water
layer was extracted twice with CH2Cl2, and the combined
organic layers were dried over Na2SO4. The solvent was re-
moved under reduced pressure, and the residue was purified
by silica gel column chromatography (CH2Cl2–MeOH,
20:1) to afford ent-23 (5.9 mg) in 84% yield as white solid.
IR (KBr): 3244, 2905, 2796, 1684, 1495, 1444, 1259, 1052,
Acknowledgements
Financial support was provided by a Grant-in-Aid for Spe-
cially Promoted Research of MEXT.
References and notes
1. (a) Williams, R. M.; Hendrix, J. A. Chem. Rev. 1992, 92, 889;
(b) Duthaler, R. O. Tetrahedron 1994, 50, 1539; (c) Ma, J.-A.
€
Angew. Chem., Int. Ed. 2003, 42, 4290; (d) Groger, H. Chem.
Rev. 2003, 103, 2793.
1
746, 697 cmꢁ1; H NMR (CDCl3): d¼7.38–7.31 (m, 2H),
7.29–7.21 (m, 3H), 6.65 (br s, 1H), 3.80–3.77 (m, 1H),
3.72–3.66 (m, 1H), 3.42 (dd, J¼7.4, 9.7 Hz, 1H), 2.73 (dd,
J¼9.1, 17.0 Hz, 1H), 2.51 (dd, J¼8.5, 17.0 Hz, 1H); 13C
NMR (CDCl3): d¼177.9, 142.1, 128.8, 127.0, 126.7, 49.5,
40.2, 38.0; MS (ESI): m/z 161 [M+]; HRMS (EI): m/z calcd
for C10H11NO [M+]: 161.0841. Found: 161.0836; [a]2D1
+35.2 (c 0.590, MeOH). The absolute configuration was de-
termined based on the comparison of the optical rotation
with the reported value.30 To determine the enatiomeric
excess by HPLC, the lactam was converted to its N-Boc
derivative (90% yield).31 The ee of the carbamate was deter-
mined to be 98%. HPLC (Chiralpak AS-H, 2-propanol–
hexane 1:20, flow 1.0 mL minꢁ1, detection at 210 nm): tR
24.8 min (major) and 28.3 min (minor). Compound 23 can
be converted to b-phenyl-GABA via acid hydrolysis.32
2. For reviews, see: (a) Xu, L. W.; Xia, C. G. Eur. J. Org. Chem.
2005, 633; (b) Liu, M.; Sibi, M. P. Tetrahedron 2002, 58, 7991;
(c) Davis, F. A.; Chen, B. C. Chem. Soc. Rev. 1998, 27, 13; (d)
Kobayashi, S.; Ishitani, H. Chem. Rev. 1999, 99, 1069.
3. Asymmetric synthesis of g-amino acids mainly relied on dia-
stereoselective reactions before Jacbosen’s contribution
(Ref. 8). For selected recent examples, see: (a) Capriati, V.;
Degennaro, L.; Florio, S.; Luisi, R.; Punzi, P. Org. Lett. 2006,
8, 4803; (b) Ueda, M.; Miyabe, H.; Sugino, H.; Miyata, O.;
Naito, T. Angew. Chem., Int. Ed. 2005, 44, 6190; (c) Friestad,
G. K.; Marie, J. C.; Deveau, A. M. Org. Lett. 2004, 6, 3249.
4. (a) Abele, S.; Seebach, D. Eur. J. Org. Chem. 2000, 1; (b)
Appella, D. H.; Christianson, L. A.; Karle, I. L.; Powell,
D. R.; Gellman, S. H. J. Am. Chem. Soc. 1999, 121, 6206; (c)
Gellman, S. H. Acc. Chem. Res. 1998, 31, 173.
5. Mita, T.; Fujimori, I.; Wada, R.; Wen, J.; Kanai, M.; Shibasaki,
M. J. Am. Chem. Soc. 2005, 127, 11252.
6. For a review about the utility of chiral ligands 1 and 2, see:
Kanai, M.; Kato, N.; Ichikawa, E.; Shibasaki, M. Synlett
2005, 1491.
7. Addition of TFA slightly, but constantly, improved the enantio-
selectivity. ESIMS studies demonstrated that TFA is incorpo-
rated in the polymetallic complex. TFA likely bridges two
4.7.2. (3R)-3-Cyano-5-methylhexanoic acid (ent-24).8a To
a solution of ent-22a (90% ee) (44.8 mg, 0.219 mmol) in
THF (440 mL), 1 M NaOH (440 mL, 0.440 mmol) was added
at room temperature. After 1 h, THF was removed, and sat-
urated NaHCO3 was added. The aqueous phase was washed
with CH2Cl2 three times, acidified to pH 1, and extracted
with CH2Cl2 three times. The combined organic layers

I. Fujimori et al. / Tetrahedron 63 (2007) 5820–5831
5831
Gd metals in the complex, and thus finely tunes the relative po-
sition of the two metals. In addition, TFA enhances the Lewis
acidity of Gd in the catalyst.
17. (a) Trost, B. M.; Organ, M. G. J. Am. Chem. Soc. 1994, 116,
€
10320; (b) Lussem, B. J.; Gais, H.-J. J. Am. Chem. Soc.
2003, 125, 6066.
8. (a) Sammis, G. M.; Jacobsen, E. N. J. Am. Chem. Soc. 2003,
125, 4442; (b) Sammis, G. M.; Danjo, H.; Jacobsen, E. N.
J. Am. Chem. Soc. 2004, 126, 9928.
18. When allylic alcohol 11 was subjected to the Mitsunobu reac-
tion, enantioselectivity decreased from 97 to 66% probably
through a competitive SN20-type reaction.
9. Silverman, R. B.; Andruszkiewicz, R.; Yuen, P. W.; Sobieray, D.
M.;Franklin, L.C.;Schwindt,M.A.U.S. Patent5,563,175, 1996;
For recent catalytic enantioselective synthesis of pregabalin,
see: Hoge, G. J. Am. Chem. Soc. 2003, 125, 10219.
10. (a) Matsunaga, S.; Kinoshita, T.; Okada, S.; Harada, S.;
Shibasaki, M. J. Am. Chem. Soc. 2004, 126, 7559; (b) Evans,
D. A.; Borg, G.; Scheidt, K. A. Angew. Chem., Int. Ed. 2002,
41, 3188.
11. Mita, T.; Sasaki, K.; Kanai, M.; Shibasaki, M. J. Am. Chem.
Soc. 2005, 127, 514.
12. Part of this work was reported in the previous communication,
see: Fujimori, I.; Mita, T.; Maki, K.; Shiro, M.; Sato, A.;
Furusho, S.; Kanai, M.; Shibasaki, M. J. Am. Chem. Soc.
2006, 128, 16438.
13. (a) Suzuki, M.; Kato, N.; Kanai, M.; Shibasaki, M. Org. Lett.
2005, 7, 2527; (b) Yabu, K.; Masumoto, S.; Yamasaki, S.;
Hamashima, Y.; Kanai, M.; Du, W.; Curran, D. P.; Shibasaki,
M. J. Am. Chem. Soc. 2001, 123, 9908.
14. Kato, N.; Mita, T.; Kanai, M.; Therrien, B.; Kawano, M.;
Yamaguchi, K.; Danjo, H.; Sei, Y.; Sato, A.; Furusho, S.;
Shibasaki, M. J. Am. Chem. Soc. 2006, 128, 6768.
15. An extreme example in which the 2:3 complex and the 4:5+m-
oxo crystal demonstrated contrasting catalyst function was the
asymmetric Strecker reaction of ketoimines. These two asym-
metric catalysts produced reversed enantioselectivity in an
excellent level. See Ref. 14.
19. Muller, G.; Sainz, D. J. Organomet. Chem. 1995, 495, 103.
20. Imi, K.; Yanagihara, N.; Utimoto, K. J. Org. Chem. 1987,52, 1013.
21. Similar temperature dependency was also observed when using
18f and 18g as substrates.
22. The syn isomer should be generated from the kinetically pro-
duced anti isomer through catalyst-mediated epimerization of
the benzylnitrile moiety.
23. In the reaction of 18f, which has a relatively small size com-
pared to 18g and 18i, the liberation of the product might be
hindered due to the high coordinating ability of the product
to the metal.
24. The protic additive activated the Gd–1¼2:3 catalyst by forming
a proton-containing complex, see: Kato, N.; Suzuki, M.; Kanai,
M.; Shibasaki, M. Tetrahedron Lett. 2004, 45, 3147.
25. Sytinsky, I. A.; Soldatenkov, A. T. Prog. Neurobiol. 1978, 10, 89.
26. Crystallographic data for the structural analysis have been
deposited with the Cambridge Crystallographic Data Centre,
CCDC No. 637203 for the Gd–4¼3:4+2OH complex
(Fig. 3b) and CCDC No. 637202 for 17 (Fig. 4).
27. Sato, T.; Gotoh, Y.; Watanabe, M.; Fujisawa, T. Chem. Lett.
1983, 1533.
28. Cannizzaro, C. E.; Ashley, J. A.; Janda, K. D.; Houk, K. N.
J. Am. Chem. Soc. 2003, 125, 2489.
29. Sibi, M. P.; Prabagaran, N.; Ghorpade, S. G.; Jasperse, C. P.
J. Am. Chem. Soc. 2003, 125, 11796.
30. Zelle, R. E. Synthesis 1991, 1023.
16. Crystal structure of another polymetallic complex (La–1¼
6:8+2OH complex) described in Ref. 14 also contained this
general module.
31. Meyers, A. I.; Snyder, L. J. Org. Chem. 1993, 58, 36.
32. Sobocinska, M.; Zobacheva, M. M.; Perekalin, V. V.;
Kupryszewski, G. Pol. J. Chem. 1979, 53, 435.