Enantioselective Cyanosilylation of Ketones with Lithium(I) Dicyanotrimethylsilicate(IV) Catalyzed by a Chiral Lithium(I) Phosphoryl Phenoxide

Article abstract of DOI:10.1002/anie.201510682

A highly enantioselective cyanosilylation of ketones was developed by using a chiral lithium(I) phosphoryl phenoxide aqua complex as an acid/base cooperative catalyst. The pentacoordinate silicate generated in situ from Me₃SiCN/LiCN acts as an extremely reactive cyano reagent. Described is a 30 gram scale reaction and the synthesis of the key precursor to (+)-13-hydroxyisocyclocelabenzine.

Full text of DOI:10.1002/anie.201510682

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201510682
German Edition: DOI: 10.1002/ange.201510682
Homogeneous Catalysis
Enantioselective Cyanosilylation of Ketones with Lithium(I)
Dicyanotrimethylsilicate(IV) Catalyzed by a Chiral Lithium(I)
Phosphoryl Phenoxide
Manabu Hatano, Katsuya Yamakawa, Tomoaki Kawai, Takahiro Horibe, and Kazuaki Ishihara*
Abstract: A highly enantioselective cyanosilylation of ketones
was developed by using a chiral lithium(I) phosphoryl
phenoxide aqua complex as an acid/base cooperative catalyst.
The pentacoordinate silicate generated in situ from Me3SiCN/
LiCN acts as an extremely reactive cyano reagent. Described is
a 30 gram scale reaction and the synthesis of the key precursor
to (+)-13-hydroxyisocyclocelabenzine.
O
ptically active cyanohydrins are important compounds
since they can readily provide a-hydroxy carboxylic acids, b-
hydroxy amines, etc., which are used in many pharmaceut-
icals.[1] However, the catalytic enantioselective cyanosilyla-
tion of ketones is still challenging, since ketones are
inherently much less reactive than aldehydes as a result of
steric and electronic constraints.[2] Progress with such reac-
tions using ketones are reported, however, there is room for
improvement with regard to the substrate scope, reaction
time (typically 24–48 h), and reaction scale.[3] To overcome
the difficulties of the enantioselective cyanosilylation of
unreactive simple ketones, we anticipated that the activation
of the trimethylsilyl cyanide reagent by achiral additives could
be effective. This strategy is highly promising as it may not
depend on the use of strong, chiral Lewis acid catalysts to the
activate substrates. In fact, even weak chiral Lewis acid
catalysts might be appropriate. In this regard, we envisioned
that the active lithium(I) dicyanotrimethylsilicate(IV)(3),[4]
formed in situ, might be suitable for use with the chiral
lithium(I) phosphoryl phenoxide aqua complex 2, a newly
designed and mild Lewis acid/Lewis base cooperative catalyst
for the cyanosilylation of ketones (Scheme 1).[5–7] An advant-
age of this catalytic system is that we can simply use the chiral
(R)-BINOL (1,1’-bi-2-naphthol)-derived ligand 1, Me3SiCN,
water, and either nBuLi or LiOH as the same lithium(I)
source to prepare both the active lithium(I) catalysts and
lithium(I) silicates(IV) in situ.
Scheme 1. Acid-base combined catalytic system with reactive lithium(I)
dicyanotrimethylsilicate(IV) 3.
and H2O (0–130 mol%) in toluene at À788C for 5 hours
(Table 1, entries 1–6). The reaction was sluggish in the
absence of water (entry 1). In contrast, the reactions were
promoted in the presence of water, and the highest enantio-
selectivity (92% ee) of 5a was observed with the use of
60 mol% of H2O (entry 4). The yield was improved with the
use of 250 mol% of Me3SiCN and 120 mol% of H2O
(entry 7). Finally, the use of 15 mol% of nBuLi provided 5a
Table 1: Optimization of the reaction conditions.[a]
Entry nBuLi
Me3SiCN H2O
Me3SiOH (Me3Si)2O Yield ee
[%] [%]
(mol%) (mol%) (mol%) (mol%) (mol%)
We initially examined the reaction of acetophenone (4a)
with 1 (10 mol%), nBuLi (10 mol%), Me3SiCN (130 mol%),
1
2
3
4
5
6
7
8
9
10
10
10
10
10
10
10
15
15[b]
15
15
130
130
130
130
130
130
250
250
250
250
250
0
20
40
60
80
130
120
120
120
120
120
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
23
44
51
35
25
0
68
94
98
90
91
44
69
86
92
91
–
90
91
90
91
89
[*] Dr. M. Hatano, K. Yamakawa, T. Kawai, Dr. T. Horibe,
Prof. Dr. K. Ishihara
Graduate School of Engineering, Nagoya University
Furo-cho, Chikusa, Nagoya 464-8603 (Japan)
E-mail: ishihara@cc.nagoya-u.ac.jp
10
11
50
0
0
50
Homepage: http://www.ishihara-lab.net/
Prof. Dr. K. Ishihara
Japan Science and Technology Agency (JST), CREST
Furo-cho, Chikusa, Nagoya 464-8603 (Japan)
[a] The reaction was carried out with 4a (0.5 mmol), Me3SiCN (130 or
250 mol%), 1 (10 mol%), nBuLi (10 or 15 mol%), and H2O
(0–130 mol%) in toluene at À788C for 5 h. The cyanohydrin 6a was not
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201510682.
obtained in any of the cases. [b] LiOH was used in place of nBuLi.
Angew. Chem. Int. Ed. 2016, 55, 4021 –4025
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4021

Angewandte
Communications
Chemie
in 94% yield with 91% ee (entry 8). LiOH reacted well in
place of nBuLi (entry 9), although we have conventionally
used nBuLi as a common catalytic lithium(I) source in
cyanosilyations.[7] Water might be essential for the generation
of active a monomeric species,[7,8] and we observed a signal,
corresponding to 2, as a major peak in ESI-MS analysis
[Eq. (1); see the Supporting Information]. Most of the
remaining water might be used to generate HCN (and
Me3SiOH) in situ as the reactions were run under the
homogeneous reaction conditions. Notably, the active cyanide
reagent in this reaction might not be HCN (entry 6), and the
cyanohydrin 6a was not obtained in any of the cases
(entries 1–11). Moreover, Me3SiOH and (Me3Si)2O, gener-
ated in situ, had almost no influence on the results (entries 10
and 11).
We show a possible catalytic cycle in Scheme 2. The
transition state (TS) 7[12] might involve the 2 + 3 complex, as
shown in Equation (2), and 4. After cyanation,[13] the product
5 is delivered via the lithium(I) cyanohydrin 8. We could not
detect 6 directly, even by in situ IR analysis (see the
For the mechanistic aspect, our catalytic system includes
multiple cyano sources, such as Me3SiCN, LiCN, HCN, and
their relevant combinations [Eq. (3)]. In particular, based on
the screening of the reaction conditions for 4a in Table 1,
a catalytic amount of LiCN, formed in situ, may play a key
role (entry 8 versus entry 7).[9,10] To identify the active reagent
Scheme 2. Possible catalytic cycles.
in the system, we performed
[D8]THF solution containing
a
13C NMR analysis of
a 1:1 molar ratio of
a
Me3SiCN (d = À2.00 ppm) and LiCN [Eq. (4); see the Sup-
porting Information]. The pentacoordinate silicate [Li]+-
[Me3Si(CN)2]À (3)[4] was observed as the only peak at d =
2.00 ppm. Ionic LiCN is essential for generation of the silicate,
since a 1:1 molar ratio of Me3SiCN and HCN gave [H]+-
[Me3Si(CN)2]À (3’) in 8% conversion, even after 5 hours
[Eq. (5); see the Supporting Information]. Based on these
results, a catalytic amount (5 mol%) of 3, formed in situ,
might serve as the active reagent [Eq. (3)].[10,11] Overall, 2
[Eq. (1)] and 3 [Eq. (4)] could associate through a cooperative
acid–base interaction. In support of this association, we
observed the 2 + 3 complex in the ESI-MS analysis [Eq. (2),
see the Supporting Information].
Supporting Information), but we cannot completely rule out
the transient generation of 6 by protonation[14,15] in an HCN
buffer, and subsequent silylation[16] of 6. Instead, it is likely
that HCN is acting as a coordinating ligand(s) for the
lithium(I) center of 2 and/or 3, and the corresponding
lithium(I)/Ln (Ln = H2O and HCN) solvates might promote
the release of 5 and the active species (e.g., Ln···2···LiCN) to
regenerate 7. In support of this assumption, we detected
2·(HCN)2 in the ESI-MS analysis when a large amount of
HCN (10 ꢀ 20 equiv)[17] was used in the presence of 2 [Eq. (6),
see the Supporting Information]. Accordingly, in the presence
of an excess amount of HCN (total of 205 mol% in situ),
a large-scale reaction of 4a was achieved (18.0 g, 150 mmol).
4022 www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 4021 –4025

Angewandte
Communications
Chemie
Additionally, by using only 5 mol% of the catalyst for
a shorter reaction time (3 h), about 30 grams of 5a were
obtained with 90% ee after extraction, concentration, and
filtration of the crude reaction mixture (no silica gel column
chromatography) [Eq. (7)]. Moreover, 1 was easily recovered
in 98% (> 99% purity) after precipitation by washing with n-
hexane and 1m aqueous HCl.
With the optimized reaction conditions in hand (Table 1,
entry 8), we examined the scope with respect to the ketones 4
(Scheme 3). The ortho-, meta-, and para-substituted aceto-
phenones were used successfully, and the corresponding
products 5b–m were obtained in high yields with high
enantioselectivities (87–98% ee).[18] In particular, electron-
withdrawing groups generally promoted the reaction, while
an electron-donating group, such as
a methyl group,
Scheme 3. Substrate scope in the catalytic enantioselective cyanosilyla-
decreased the reactivity, with recovery of the starting
ketone 4j.[19] Whereas sterically hindered ortho-substituted
ketones are often problematic for use in conventional
catalysis,[1–3] they were tolerated by our catalytic system
(5b–f). Moreover, our catalyst was used for the unprece-
dented cyanosilylation of carbonyl- and cyano-substituted
ketones (5 f, 5h, and 5l), groups which may deactivate
relatively strong Lewis acid catalysts.[1–3] The coordination
of 2-furyl and 3-thienyl ketones, 4o and 4p, respectively, to
the catalyst also proceeded smoothly. Cyclohexyl methyl
ketone (4q), as a simple aliphatic ketone, showed moderate
enantioselectivity (5q, 78% ee). In sharp contrast, acyclic and
cyclic a,b-unsaturated aliphatic ketones (4r–t) delivered the
corresponding products 5r–t in high yields with good to high
enantioselectivities (83–90% ee). Remarkably, our catalyst
could be used for other nonmethyl ketones, such as 1-
indanone, propiophenone, butyrophenone, and valerophe-
none derivatives, and the desired products 5u–x were
obtained with high enantioselectivities (90–94% ee). Overall,
our catalytic system features a much shorter reaction time (2–
9 h) than conventional systems.[1À3] It is noted again that
cyanohydrins (6) were not obtained in any of the reactions.
To demonstrate the synthetic utility of our catalytic
system, we synthesized the key intermediate 11 which is
used in the synthesis of (+)-13-hydroxyisocyclocelabenzine
tion of ketones 4. [a] Unless otherwise noted, the reaction was carried
out with 4 (0.5 mmol), Me3SiCN (250 mol%), 1 (10 mol%), nBuLi
(15 mol%), and H2O (120 mol%) in toluene at À788C. [b] 270 mol%
of Me3SiCN was used. [c] The reaction was carried out with 4
(1 mmol), Me3SiCN (250 mol%), 1 (5 mol%), nBuLi (7.5 mol%), and
H2O (120 mol%) in toluene at À788C for 6 h. [d] 300 mol% of
Me3SiCN was used.
(12),[20] a spermidine alkaloid with antibacterial and antitu-
mor activities (Scheme 4). The bulky allyl 2-bromophenyl
ketone (4y) was selected as the starting ketone. Fortunately,
by using LiOH and 1 (7.5 mol%) in the enantioselective
cyanosilylation of 4y on a 1.13 gram scale, 5y was obtained in
92% yield (1.49 g) with 90% ee. The compound 5y was then
transformed into 9 by reduction with LiAlH4 and subsequent
Boc protection (Boc = tert-butoxycarbonyl). A single recrys-
tallization of 9 increased the enantiopurity to 99% without
a serious loss of yield. Finally, the optically pure key
intermediate 11 was obtained after a copper(I)-promoted
lactonization to 10 and removal of the Boc group.
In summary, we have developed a highly enantioselective
cyanosilylation of ketones with the use of a chiral lithium(I)
phosphoryl phenoxide complex as an acid/base cooperative
catalyst. An extremely reactive pentacoordinate silicate
generated in situ from Me3SiCN and LiCN acts as the key
Angew. Chem. Int. Ed. 2016, 55, 4021 –4025
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4023

Angewandte
Communications
Chemie
Int. Ed. 2002, 41, 1009; Angew. Chem. 2002, 114, 1051; d) S.-K.
Tian, R. Hong, L. Deng, J. Am. Chem. Soc. 2003, 125, 9900;
e) D. E. Fuerst, E. N. Jacobsen, J. Am. Chem. Soc. 2005, 127,
8964; f) D. H. Ryu, E. J. Corey, J. Am. Chem. Soc. 2005, 127,
5384; g) X. Liu, B. Qin, X. Zhou, B. He, X. Feng, J. Am. Chem.
Soc. 2005, 127, 12224; h) N. Kurono, M. Uemura, T. Ohkuma,
Eur. J. Org. Chem. 2010, 1455; i) S. Kong, W. Fan, G. Wu, Z.
Miao, Angew. Chem. Int. Ed. 2012, 51, 8864; Angew. Chem. 2012,
124, 8994.
[4] Dicyanotrimethylsilicates(IV) are known as highly reactive CN
reagents. a) D. A. Dixon, W. R. Hertler, D. B. Chase, W. B.
Farnham, F. Davidson, Inorg. Chem. 1988, 27, 4012; b) M. B.
Sassaman, G. K. S. Prakash, G. A. Olah, J. Org. Chem. 1990, 55,
2016; c) H. S. Wilkinson, P. T. Grover, C. P. Vandenbossche, R. P.
Bakale, N. N. Bhongle, S. A. Wald, C. H. Senanayake, Org. Lett.
2001, 3, 553; d) N. Kurono, K. Arai, M. Uemura, T. Ohkuma,
Angew. Chem. Int. Ed. 2008, 47, 6643; Angew. Chem. 2008, 120,
6745.
[5] For reviews on acid-base combination chemistry, see: a) M.
Kanai, N. Kato, E. Ichikawa, M. Shibasaki, Synlett 2005, 1491;
b) K. Ishihara, A. Sakakura, M. Hatano, Synlett 2007, 0686; c) K.
Ishihara, Proc. Jpn. Acad. Ser. B 2009, 85, 290; d) M. Shibasaki,
M. Kanai, S. Matsunaga, N. Kumagai, Acc. Chem. Res. 2009, 42,
1117.
Scheme 4. Synthesis of the key intermediate 11 which is used in the
synthesis of (+)-13-hydroxyisocyclocelabenzine (12). DMF=N,N-dime-
thylformamide.
[6] Shibasaki and co-workers developed chiral aluminum(III)
phosphoryl phenoxides as acid – base cooperative catalysts for
cyanosilylation: Y. Hamashima, D. Sawada, M. Kanai, M.
Shibasaki, J. Am. Chem. Soc. 1999, 121, 2641.
cyano reagent. In particular, our robust catalytic system was
employed in a large-scale (30 gram) reaction. Also, a key
intermediate in the synthesis of (+)-13-hydroxyisocyclocela-
benzine was successfully synthesized in an optically pure
form.
[7] We previously developed a chiral lithium(I) binaphtholate aqua
complex for aldehydes and a chiral lithium(I) phosphate for
a very limited numbers of ketones. The reactions scarcely
proceed with these catalysts (< 15% yield with < 10% ee) under
the reaction conditions given in entry 8 of Table 1 (see the
Supporting Information). Therefore, this catalytic system with
the silicate 3 turned out to be different from our previous
catalytic systems with Me3SiCN. a) M. Hatano, T. Ikeno, T.
Miyamoto, K. Ishihara, J. Am. Chem. Soc. 2005, 127, 10776;
b) M. Hatano, T. Ikeno, T. Matsumura, S. Torii, K. Ishihara, Adv.
Synth. Catal. 2008, 350, 1776.
Acknowledgments
Financial support was partially provided by JSPS KAKENHI
(Grant Numbers 15H05755, 26288046, and 26105723), Pro-
gram for Leading Graduate Schools “IGER program in
Green Natural Sciences”, and MEXT, Japan. KY thanks the
JSPS Research Fellowships for Young Scientists.
[8] A positive nonlinear effect, which supports the notion that the
catalysts have oligomeric structures, was observed between 1 and
5a (see the Supporting Information).
[9] Shibasaki and co-workers reported the role of a catalytic amount
of LiCN in the cyano ethoxycarbonylation and cyano phosphor-
Keywords: asymmetric synthesis · homogeneous catalysis ·
ketones · lithium · silicates
ylation
of
aldehydes
with
the
use
of
chiral
YLi3tris(binaphthoxide) complexes: a) N. Yamagiwa, J. Tian, S.
Matsunaga, M. Shibasaki, J. Am. Chem. Soc. 2005, 127, 3413;
b) N. Yamagiwa, Y. Abiko, M. Sugita, J. Tian, S. Matsunaga, M.
Shibasaki, Tetrahedron: Asymmetry 2006, 17, 566.
How to cite: Angew. Chem. Int. Ed. 2016, 55, 4021–4025
Angew. Chem. 2016, 128, 4089–4093
[10] We used a catalytic amount of isolated LiCN in a model reaction
of 4a with Me3SiCN in toluene at À788C (see the Supporting
Information). As a result, a positive effect of LiCN on both yield
and enantioselectivity was observed when compared to the
control experiment in which LiCN was absent. However,
isolated LiCN was not very soluble and caused, overall, lower
efficiency under heterogeneous conditions, particularly in the
yield, than soluble LiCN which was prepared in situ from
Me3SiCN, nBuLi, and H2O under homogeneous conditions.
[11] Investigation of the reactions of aldehydes also supports that 3
might be more active than other reagents in situ. See the
Supporting Information.
[12] Possible transition states, which might explain the absolute
stereochemistry of 5, are shown in the Supporting Information.
[13] The retro reaction of the tertiary cyanohydrin 8 to the TS 7 might
be possible. Therefore, the reaction of 8 with a HCN buffer
would also provide a slight bias on the equilibrium between 7
and 8.
[1] Recent reviews on enantioselective synthesis cyanohydrin:
a) R. J. H. Gregory, Chem. Rev. 1999, 99, 3649; b) M. North,
Tetrahedron: Asymmetry 2003, 14, 147; c) J.-M. Brunel, I. P.
Holmes, Angew. Chem. Int. Ed. 2004, 43, 2752; Angew. Chem.
2004, 116, 2810; d) F.-X. Chen, X. Feng, Curr. Org. Synth. 2006,
3, 77; e) M. North, D. L. Usanov, C. Young, Chem. Rev. 2008,
108, 5146; f) S. E. Denmark, G. L. Beutner, Angew. Chem. Int.
Ed. 2008, 47, 1560; Angew. Chem. 2008, 120, 1584; g) W. Wang,
X. Liu, L. Lin, X. Feng, Eur. J. Org. Chem. 2010, 4751; h) D.
Murtinho, M. E. S. Serra, Curr. Organocatal. 2014, 1, 87.
[2] a) O. Riant, J. Hannedouche, Org. Biomol. Chem. 2007, 5, 873;
b) M. Shibasaki, M. Kanai, Org. Biomol. Chem. 2007, 5, 2027;
c) M. Hatano, K. Ishihara, Synthesis 2008, 1647.
[3] Selected reports for ketones: a) Y. Hamashima, M. Kanai, M.
Shibasaki, J. Am. Chem. Soc. 2000, 122, 7412; b) K. Yabu, S.
Masumoto, S. Yamasaki, Y. Hamashima, M. Kanai, W. Du, D. P.
Curran, M. Shibasaki, J. Am. Chem. Soc. 2001, 123, 9908; c) H.
Deng, M. P. Isler, M. L. Snapper, A. H. Hoveyda, Angew. Chem.
4024 www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 4021 –4025

Angewandte
Communications
Chemie
[14] On the basis of pKa values in water (HCN: 9.1, Me3SiOH: 12.7,
control experiment with 205 mol% of HCN showed fast
H2O: 15.8), HCN might act as a proton source. a) F. G. Bordwell,
Acc. Chem. Res. 1988, 21, 456; b) K. M. Delak, N. Sahai, Chem.
Mater. 2005, 17, 3221.
conversion (2 h, 95% yield with 91% ee). See the Supporting
Information.
[18] For some ketones, the use of either 270 or 300 mol% of
Me3SiCN gave better yields than the use of 250 mol% of
Me3SiCN.
[19] During the reaction, the inactive silylated ligand 13 was partially
detected (see the Supporting Information), particularly when
unreactive ketones such as 4j were used. A routine workup/
purification procedure always resulted in recovered 1 (90–95%),
and it could be reused after it was washed with 1m aqueous HCl.
[20] Y. Li, A. Linden, M. Hesse, Helv. Chim. Acta 2003, 86, 579.
[15] Oguni and co-workers reported a chiral titanium(IV)-catalyzed
cyanosilylation of aldehydes, in which Me3SiCN and HCN
coexist. In their report, a silyl-protected cyanohydrin was
obtained from titanium(IV)-cyanohydrins by HCN, followed
by silylation with Me3SiCN: M. Hayashi, T. Matsuda, N. Oguni,
J. Chem. Soc. Perkin Trans. 1 1992, 3135.
[16] In a control experiment, the isolated cyanohydrin 6a was quickly
trimethylsilylated within 1 minute under the standard reaction
conditions. See the Supporting Information.
[17] A large amount of HCN (i.e., 105 mol% under optimized
conditions) is necessary to promote the reaction. A control
experiment with 45 mol% of HCN showed significantly slow
conversion (5 h, 41% yield with 89% ee). In contrast, another
Received: November 18, 2015
Revised: December 23, 2015
Published online: February 2, 2016
Angew. Chem. Int. Ed. 2016, 55, 4021 –4025
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4025