Reinvestigation of the fluoride-triggered condensation of allyltrimethylsilane with cinnamonitrile: 'Abnormal Sakurai' and sequential 'abnormal Sakurai'-Michael-Thorpe Ziegler side reactions

Article abstract of DOI:10.1016/S0040-4039(02)00456-2

Together with the expected 1,4-adduct, three by-products are generated in the fluoride-triggered condensation of allyltrimethylsilane with cinnamonitrile: the 1,3 adduct, an acyclic dinitrile and a cyclopentenaminonitrile issuing, respectively, from 'abnormal Sakurai', 'abnormal Sakurai'-Michael and 'abnormal Sakurai'-Michael-Thorpe Ziegler side reactions.

Full text of DOI:10.1016/S0040-4039(02)00456-2

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 43 (2002) 3225–3228
Reinvestigation of the fluoride-triggered condensation of
allyltrimethylsilane with cinnamonitrile: ‘abnormal Sakurai’ and
sequential ‘abnormal Sakurai’–Michael–Thorpe Ziegler
†
side reactions
a
a,
a
a,
b
Laurent Keller, Fran c¸ oise Dumas, * Mathieu Pizzonero, Jean d’Angelo, * Georges Morgant and
b
Dung Nguyen-Huy
a
Unit e´ de Chimie Organique Associ e´ e au CNRS, Universit e´ Paris-Sud, Facult e´ de Pharmacie, 5, rue J.-B. Cl e´ ment,
9
2296 Ch aˆ tenay-Malabry, France
Laboratoire de Cristallochimie Bioinorganique, Universit e´ Paris-Sud, Facult e´ de Pharmacie, 5, rue J.-B. Cl e´ ment,
2296 Ch aˆ tenay-Malabry, France
b
9
Received 29 January 2002; accepted 5 March 2002
Abstract—Together with the expected 1,4-adduct, three by-products are generated in the fluoride-triggered condensation of
allyltrimethylsilane with cinnamonitrile: the 1,3 adduct, an acyclic dinitrile and a cyclopentenaminonitrile issuing, respectively,
from ‘abnormal Sakurai’, ‘abnormal Sakurai’–Michael and ‘abnormal Sakurai’–Michael–Thorpe Ziegler side reactions. © 2002
Elsevier Science Ltd. All rights reserved.
2
The conjugate addition of allylsilanes to electrophilic
alkenes, referred to as Sakurai–Hosomi reaction has
been recognized as a particularly efficient method of
CꢀC bond formation, and has played an important role
Majetich (dropwise addition at 20°C of a 1 M solution
of 3 equiv. of 2 and 3 equiv. of HMPA in DMF to a
0.6 M solution of 1 equiv. of 1 and 0.1 equiv. of TBAF
in DMF, followed by 15 min stirring at 20°C and
hydrolytic workup with MeOH and HCl), the main
product characterized was invariably the cyclopentene
derivative 5 (20–30% yield), accompanied by varying
quantities of acyclic by-product 6 (0–10% yield) and
only a minute amount of adduct 3 (ca. 5% yield).
1
in the area of synthetic organic chemistry. In 1986,
Majetich et al. reported that the condensation of cinna-
monitrile (1) with allyltrimethylsilane (2), promoted by
a catalytic amount of n-Bu NF (TBAF) furnished the
4
2
1
,4-adduct 3 with a 65% yield. In connection with our
3
4
sustained synthetic efforts in the cephalotaxine series,
we recently reinvestigated this condensation to assess its
utility in the construction of this alkaloid. In this paper,
we show that together with adduct 3, three unexpected
by-products 4, 5 and 6 are in fact generated in the
fluoride-triggered addition of 1 to 2, the ratio of which
depended essentially on the time of addition and the
concentration of the reagents (Scheme 1).
Compound 5, a low melting point (58–59°C) solid, was
CN
CN
TMS
CN
Ph
Ph
Ph
2
1
cat. TBAF
3
Surprisingly, when the condensation of 1 with 2 was
conducted under the reaction conditions reported by
NC
H
NC
NH2
CN
H
2
Ph
5
3
4
Ph
Ph
3
Keywords: mechanisms; Michael reactions; rearrangements; Sakurai
reactions; Thorpe and Thorpe–Ziegler reactions.
4
H
Ph
*
Corresponding
authors.
Fax:
+33(0)146835752;
e-mail:
4
5
6
francoise.dumas@cep.u-psud.fr; jean.dangelo@cep.u-psud.fr
This paper is dedicated to Professor Gilbert Stork on the occasion
of his 80th birthday.
†
Scheme 1.
0
040-4039/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(02)00456-2

3
226
L. Keller et al. / Tetrahedron Letters 43 (2002) 3225–3228
assigned a molecular formula of C H N established
Structure of the second by-product was elucidated to be
21
20
2
5
by elemental analysis and mass spectroscopy, corre-
sponding to the formal condensation of one allyl unit
with two molecules of cinnamonitrile (1). The gross
structure of 5 was deduced from detailed analysis of the
(2R*,3S*,4R*)-2-allyl-3,4-diphenyl-hexanedinitrile (6)
on the basis of molecular ion peak at m/z 300 in mass
1
13
spectroscopy, of IR spectrum and of H and C NMR
data including 2D-experiments. The depicted relative
configuration of the three stereogenic centers, although
not definitively established, rests on mechanistic consid-
erations assuming that dinitrile 6 and enaminonitrile 5
both derived from the parent carbanion 15 (vide infra,
Scheme 3).
1
13
H and C NMR data, aided with 2D NMR experi-
1
1
ments ( H– H COSY, HMQC and HMBC) and of the
IR spectrum which revealed the presence of a b-enam-
inonitrile moiety (w(NH) at 3480 and 3363 cm , w(CN)
at 2185 cm and w(CꢁC) at 1639 cm ). Finally, the full
structure of 5, including the relative configuration of
the three stereogenic centers was defined as
−
1
−
1
−1
Both by-products 5 and 6 arising from a further inter-
molecular nucleophilic addition (see Schemes 2 and 3),
we reasoned that this undesirable side reaction may be
minimized by simple lowering of the concentration of
reactants 1 and 2. To support this hypothesis, the above
(
1
3R*,4S*,5R*)-3-allyl-2-amino-4,5-diphenyl-cyclopent-
-enecarbonitrile by means of a single crystal X-ray
diffraction analysis (Fig. 1).
2
Majetich protocol was repeated, having diluted 10
times the reagents in DMF and having added all at once
the solution of 2 and HMPA to the solution of 1 and
TBAF. In such conditions, the main products isolated
2
were found to be the expected 1,4-adduct 3 and its
regioisomer 4 as ca. 2:1 mixture in 90% combined yield,
along with a small amount of by-product 5 (ca. 5%
yield). All attempts at separating the components of
6
mixture (3+4), employing standard chromatographic
methods (TLC, GC, HPLC) proved to be unsuccessful.
However, the minor component of this mixture was
authenticated as the ‘abnormal Sakurai adduct’ (2-ben-
zyl-4-pentenenitrile, 4, resulting from the formal 1,3-
addition of allylsilane 2 to nitrile 1), by comparison of
1
13
its H and C NMR data with those reported in the
7
literature.
Formation of products 3, 4, 5 and 6 can be rational-
ized, invoking the fluoride-triggered autocatalytic mech-
anism depicted in Schemes 2 and 3. As previously
2
Figure 1. ORTEP view of 5 with labeled heteroatoms and
stereogenic atoms. Thermal ellipsoids are scaled to 30% prob-
ability level. Hydrogen atoms shown are drawn to an arbi-
trary scale.
postulated, the process is probably initiated by the
nucleophilic pentacoordinate silicon complex 7 issuing
from the condensation of TBAF with allylsilane 2.
1,4-Addition of 7 to nitrile 1 next delivers anion 8, the
Scheme 2.

L. Keller et al. / Tetrahedron Letters 43 (2002) 3225–3228
3227
Scheme 3.
trapping of which by TMSF furnishing the neutral
N-silylated species 9. Anion 8 is also capable of equili-
brating with regioisomer 12, through the benzylic anion
the one hand the trans stereochemistry of starting
nitrile 1, and on the other hand the anti relationship
between the migrating allyl appendage and the bridged
silicon ion in putative intermediate 10. The stereochem-
ical course of the subsequent Michael addition of ben-
zylic anion 11 to a,b-ethylenic nitrile 1, furnishing the
fully stereocontrolled pivotal ion 15 can in turn be
rationalized, invoking approach 14. The reason for the
orientation of the reactants in this approach pre-
sumably lies in a conjunction of two factors: the posi-
tioning of the nitrile ends close to each other, a
prerequisite of the subsequent Thorpe–Ziegler ring clo-
sure, and the preferential attack of anion 11 on the Re
p-face of electrophilic partner 1, thus minimizing the
Ph–Ph repulsive interaction.
11. This rearrangement first involves a 1,2-anion shift
and concomitant sigmatropic transposition of the allyl
group, possibly via the nonclassical silicon complex 10,
leading to transient benzylic ion 11. The latter next
stabilizes by 1,2-[H] transfer, giving rise to translocated
anion 12 which is trapped with TMSF to afford the
N-silylated derivative 13 (Scheme 2).
The two-step anionic rearrangement [8“12] deserves
comment. The first step [8“11] being clearly thermody-
namically disfavored (compare the pK values of aceto-
a
nitrile and toluene, 31 and ca. 43, respectively), the
process requires a high energy of activation. However,
since carbanions 8 and 12 are both stabilized by an
a-nitrile group, one may infer that they are close in
energy. Therefore, provided that a stationary state is
attained, a close population of regioisomeric anions 8 and
To conclude, we have demonstrated that under stan-
dard operating conditions the fluoride-mediated con-
densation of cinnamonitrile (1) with allyltrimethylsilane
(2) furnished essentially a mixture of by-products 5 and
6, resulting from sequential ‘abnormal Sakurai’–
Michael–Thorpe Ziegler and ‘abnormal Sakurai’–
Michael side reactions. Although these undesirable
processes could be almost completely suppressed by an
appropriate dilution of the reaction medium, the syn-
thesis of the desired Sakurai adduct 3 was found now to
be thwarted by the concomitant formation of a signifi-
cant amount of the ‘abnormal regioisomer’ 4.
12 is expected. On the other hand, it should be noted
that the benzylic anion 11, being less stabilized (and less
populated) than its congeners 8 and 12 is correlatively
much more reactive, and therefore can play the role of
favorite nucleophilic partner in a subsequent Michael
addition. In fact, when the reactants are present at a
relatively high concentration level
a
competing
Michael-type addition of benzylic anion 11 to a second
molecule of cinnamonitrile (1) now occurs, leading to
carbanion 15 precursor of silylated compound 16. It is
noteworthy that, although primarily governed by the
concentration of reagents, this Michael addition is ther-
modynamically favored, the developing carbanion 15
being much more stabilized than its progenitor 11.
Alternatively, anion 15 can undergo a Thorpe–Ziegler
Acknowledgements
M.P. is grateful for the fellowship funding from CNRS
and Oncopharm SA. We kindly acknowledge Dr. J.
Mahuteau and Dr. M. Ourevitch for the valuable assis-
tance with NMR studies and Mrs. S. Mairesse-Lebrun
for performing elemental analyses (Centre d’Etudes
Pharmaceutiques, Ch aˆ tenay-Malabry).
8
annulation, furnishing 17 precursor of 18. Further
protiodesilylation during workup of the N-silylated
compounds 9, 13, 18 and 16 finally delivers products 3,
4, 5 and 6, respectively (Scheme 3).
The remarkable complete stereocontrol of the three
stereogenic centers present in by-products 5 and 6 can
be interpreted, assuming first that the transient benzylic
anion retains the configuration shown in formula 11
References
1. For a recent reference, see: Lee, P. H.; Lee, K.; Sung, S.;
(
one enantiomer is arbitrarily represented), reflecting on
Chang, S. J. Org. Chem. 2001, 66, 8646–8649.

3
228
L. Keller et al. / Tetrahedron Letters 43 (2002) 3225–3228
1
2
3
4
. Majetich, G.; Casares, A.; Chapman, D.; Behnke, M. J.
Org. Chem. 1986, 51, 1745–1753.
. de Oliveira, E. R.; Dumas, F.; d’Angelo, J. Tetrahedron
Lett. 1997, 38, 3723–3726.
H NMR (CDCl , 200 MHz) l: 1.98 (t, J=7.7 Hz, 2H),
3
2.36 (d, J=5.6 Hz, 1H), 2.39 (d, J=2.7 Hz, 1H), 2.56 (dt,
J=7.7, 4.2 Hz, 1H), 3.18 (dd, J=12.0, 4.2 Hz, 1H), 3.58
(ddd, J=7.2, 4.6, 4.2 Hz, 1H), 4.98 (dq, J=17.0, 1.5 Hz,
1H), 5.08 (dq, J=10.3, 1.5 Hz, 1H), 5.54 (ddt, J=17.0,
10.3, 6.9 Hz, 1H), 7.20–7.60 (m, 10H); C NMR (CDCl3,
50 MHz) l: 139.0 (C), 136.4 (C), 132.8 (CH), 129.7 (4CH),
129.5 (CH), 128.7 (2CH), 128.6 (2CH), 127.6 (CH), 119.0
. Enaminonitrile 5: White solid; mp 58–59°C (AcOEt/pen-
13
tane); R =0.21 (silica gel, 20% AcOEt in cyclohexane); IR
f
−
1
1
(
neat, cm ) w: 3480, 3363, 2185, 1639, 1597; H NMR
(
CDCl , 400 MHz) l: 2.27–2.34 (m, 2H), 2.84 (t, J=8.3
3
Hz, 1H), 3.06 (m, 1H), 4.01 (d, J=8.3 Hz, 1H), 4.85 (bs,
(2C), 117.7 (CH
(CH), 23.8 (CH
2
), 49.8 (CH), 45.4 (CH), 34.9 (CH
); MS (ESI, +32.0 V, MeOH+HCOONH )
4
2
), 34.6
2
H), 5.14 (d, J=10.2 Hz, 1H), 5.19 (d, J=13.7 Hz, 1H),
.72–5.86 (m, 1H), 7.03–7.15 (m, 4H), 7.19–7.32 (m, 6H);
2
+
+
5
m/z (rel. intensity): 323 (M+Na , 67), 318 (M+NH
4
, 100),
13
+
C NMR (CDCl , 50 MHz) l: 163.3 (C), 142.2 (C), 141.4
301 (M+H , 32).
3
(
1
(
(
C), 135.1 (CH), 128.4 (2CH), 128.2 (2CH), 127.8 (2CH),
27.1 (2CH), 126.8 (CH), 126.7 (CH), 118.2 (C), 118.1
6. Mixture of regioisomers (3+4): colorless oil; R
ica gel, 20% AcOEt in cyclohexane); selected pertinent
NMR data (CDCl , 50 MHz) l: 3: 23.8 (CH ), 39.1 (CH
41.6 (CH); 4: 33.4 (CH), 35.6 (CH ), 37.6 (CH ).
f
=0.62 (sil-
13
C
CH ), 77.1 (C), 59.2 (CH), 56.4 (CH), 51.0 (CH), 35.3
3
2
),
2
2
CH ); MS (EI, −70.0 V) m/z (rel. intensity): 301 (M+1,
2
2
2
1
4), 300 (M, 51), 271 (30), 259 (100), 256 (63). Anal. calcd
C, 83.96; H, 6.71; N, 9.33. Found: C, 83.81; H, 6.70; N,
.30%.
. Dinitrile 6: White solid; mp 159–160°C; R =0.47 (silica
7. Walters, M. A.; Hoem, A. B.; McDonough, C. S. J. Org.
Chem. 1996, 61, 56–62.
9
8. For an example of Michael–Thorpe Ziegler tandem reac-
tion, see: Murugan, P.; Raghukumar, V.; Ramakrishnan,
V. T. Synth. Commun. 1999, 29, 3881–3887.
5
f
−
1
gel, 20% AcOEt in cyclohexane); IR (neat, cm ) w: 2243;